Add files using upload-large-folder tool

.gitattributes

@@ -569,3 +569,259 @@ preprint/preprint__297c6ce810062b98be868eae7ff1f01ee8b99eed7e2454a49f55e4e048f81
 preprint/preprint__29d03685944943092d997a84a336546d4b445d8c11637f8beeca61e49570666b/preprint__29d03685944943092d997a84a336546d4b445d8c11637f8beeca61e49570666b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
 preprint/preprint__2a24d24c09fa4c83de31bef1f81a86b08174fded498af03e843fea1cacde74d2/preprint__2a24d24c09fa4c83de31bef1f81a86b08174fded498af03e843fea1cacde74d2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
 preprint/preprint__29df6babff945bb4384b8d2f4a537b04bd087cf481aa1dab40230f3b611d39e6/preprint__29df6babff945bb4384b8d2f4a537b04bd087cf481aa1dab40230f3b611d39e6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a1af7c3b5e38adf25c0817b2a8f664f58b095022bcae65c73a63646179907e9/preprint__2a1af7c3b5e38adf25c0817b2a8f664f58b095022bcae65c73a63646179907e9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a38080b88cad246cfb6882bed30daa113b2e35994e93904552705066cd07de4/preprint__2a38080b88cad246cfb6882bed30daa113b2e35994e93904552705066cd07de4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__29f63fcc205ffec396e73e777145934f596d4ee5d02b11b20083004d7d4f5b21/preprint__29f63fcc205ffec396e73e777145934f596d4ee5d02b11b20083004d7d4f5b21_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__295b1f7da170a8f138bc9b4995b1a433d85bbf54c5fd930011909ed76867beb2/preprint__295b1f7da170a8f138bc9b4995b1a433d85bbf54c5fd930011909ed76867beb2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26140f50ed4206e87acbd697e7bdf048f5981ece463cfe114a59bc74219638cb/preprint__26140f50ed4206e87acbd697e7bdf048f5981ece463cfe114a59bc74219638cb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26762904897bc70e6fa28634c31a3942ad67800b603ba2f8b91b04bdda658fae/preprint__26762904897bc70e6fa28634c31a3942ad67800b603ba2f8b91b04bdda658fae_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__268c1a0a3c1bfa106e169a7a83471838b57e8ccd13543d34a6a125d40ed77988/preprint__268c1a0a3c1bfa106e169a7a83471838b57e8ccd13543d34a6a125d40ed77988_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2622b9014fa1412f130b9fa61d13863398aabfc717598fed487afbfcf756d2d5/preprint__2622b9014fa1412f130b9fa61d13863398aabfc717598fed487afbfcf756d2d5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26d52d823145ea21a1a607520d91858a4edffd9242234a28dbea7be9bba9acc3/preprint__26d52d823145ea21a1a607520d91858a4edffd9242234a28dbea7be9bba9acc3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__274e95da440ff6392b39f82cfd8c0d9037b52e5365e600c5936d7b07e8956ae1/preprint__274e95da440ff6392b39f82cfd8c0d9037b52e5365e600c5936d7b07e8956ae1_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__269a8f426a3bb3c3109586bd78567f8f19460f76ea9639067a9d35084da55f76/preprint__269a8f426a3bb3c3109586bd78567f8f19460f76ea9639067a9d35084da55f76_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26a1b7bc682084530d4c6c44344a45b19dd9f5c133e73c1f35b70f93956a586c/preprint__26a1b7bc682084530d4c6c44344a45b19dd9f5c133e73c1f35b70f93956a586c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26b180044ddc388dbec6235c17ee60158c468aaf9d575333db559f8784679559/preprint__26b180044ddc388dbec6235c17ee60158c468aaf9d575333db559f8784679559_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27206ab5fd911ca9f9434dc07bf89032bd42e49f4bcd5d5e91e58a0a9bda02f7/preprint__27206ab5fd911ca9f9434dc07bf89032bd42e49f4bcd5d5e91e58a0a9bda02f7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26f7c5dac8d891bdf1c6f4d2a03dfa6663bdf562aea9984920d5a6957eae9024/preprint__26f7c5dac8d891bdf1c6f4d2a03dfa6663bdf562aea9984920d5a6957eae9024_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26fdd214e0d60e72945390debb389ceecf0241a6cdf1ad6722b98fbaef5d332b/preprint__26fdd214e0d60e72945390debb389ceecf0241a6cdf1ad6722b98fbaef5d332b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__276cd8caa182d3888ddfc403c726916a7ab186b3f243bc50d8d2fd2f87ee2aeb/preprint__276cd8caa182d3888ddfc403c726916a7ab186b3f243bc50d8d2fd2f87ee2aeb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27dfbcfd8217d57898a8d9a4f95f60a339613621611ac37ff52baed37bd97f6b/preprint__27dfbcfd8217d57898a8d9a4f95f60a339613621611ac37ff52baed37bd97f6b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27924350800667e5e31983b869d13d00f23872a9c57e8f98e1999c224d2e7641/preprint__27924350800667e5e31983b869d13d00f23872a9c57e8f98e1999c224d2e7641_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27a7c75dbfb5489c029fde5177ead71fb9cb5f446fd2e12ffa423b498a331ec3/preprint__27a7c75dbfb5489c029fde5177ead71fb9cb5f446fd2e12ffa423b498a331ec3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27b19f6fb6ee01ac3ba4aed01a7b0d13ae1bc9535efec5a93bceb59cc50013e7/preprint__27b19f6fb6ee01ac3ba4aed01a7b0d13ae1bc9535efec5a93bceb59cc50013e7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27893c7b47914ec5c7d3a3c6f3de6e7f5330c33cafe35f9db68b81d5628a6905/preprint__27893c7b47914ec5c7d3a3c6f3de6e7f5330c33cafe35f9db68b81d5628a6905_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__26e49cf12393ce9275f60df74a5a50949bdb5bd7b302e6fb2435a89123ab22da/preprint__26e49cf12393ce9275f60df74a5a50949bdb5bd7b302e6fb2435a89123ab22da_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__268877759a8b80db8ba70b7d098a918d721cd468e84d71016c3ad029521085f3/preprint__268877759a8b80db8ba70b7d098a918d721cd468e84d71016c3ad029521085f3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__280e5fb71fb848d0639b8ad5d00655700dd93984af5dc9d7444dffbe97a890e4/preprint__280e5fb71fb848d0639b8ad5d00655700dd93984af5dc9d7444dffbe97a890e4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27def387562abe0fc9c943e5a1ff9f06e5e8e062efb95fca06a2d9923393017d/preprint__27def387562abe0fc9c943e5a1ff9f06e5e8e062efb95fca06a2d9923393017d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__273ea07f52661add7a9163238541f99090c9e99a761d078c0a13a26634e32077/preprint__273ea07f52661add7a9163238541f99090c9e99a761d078c0a13a26634e32077_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2748614eb30e88e35b11e5754b3f66422e43b2dc7493ebd2a92d22f021a75490/preprint__2748614eb30e88e35b11e5754b3f66422e43b2dc7493ebd2a92d22f021a75490_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__27a47e0830461c683023e8eb02bf7839792fa42de7070c14e95921a990f9162f/preprint__27a47e0830461c683023e8eb02bf7839792fa42de7070c14e95921a990f9162f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2858463fc7b2cde1b2503cf458aeed786457d957a7fd2b2720aee8617abb0ced/preprint__2858463fc7b2cde1b2503cf458aeed786457d957a7fd2b2720aee8617abb0ced_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2835dc7e5da56b58cf8caafb0d01ab338ebeb8fdf39a30ba19800f36cf8fd382/preprint__2835dc7e5da56b58cf8caafb0d01ab338ebeb8fdf39a30ba19800f36cf8fd382_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2bfa03fcde5a6f56d970eaff256dd3cd9c1c9c89d57bf2a5a2bbe5bfb585924b/preprint__2bfa03fcde5a6f56d970eaff256dd3cd9c1c9c89d57bf2a5a2bbe5bfb585924b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2cb3057231f50adebbfed4a007a4cc78181d0ff487534b21ff2a78ed399591c2/preprint__2cb3057231f50adebbfed4a007a4cc78181d0ff487534b21ff2a78ed399591c2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c5f7e14b2427211384d15702f3b2bfff9c79ab4a8c49958e44e9081da366217/preprint__2c5f7e14b2427211384d15702f3b2bfff9c79ab4a8c49958e44e9081da366217_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c60587a10ac94558772c662d025637260a36fc471370858b71911a94e3ea5e2/preprint__2c60587a10ac94558772c662d025637260a36fc471370858b71911a94e3ea5e2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b80d2125fb4e34dd3cd9f801c4387ddb7d0c5c4f3aaa3b81839750ee75cc826/preprint__2b80d2125fb4e34dd3cd9f801c4387ddb7d0c5c4f3aaa3b81839750ee75cc826_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c672d75ce06b21426605a89be85ffccc634f09ac63b4dbe71f7f7d60b62bcd1/preprint__2c672d75ce06b21426605a89be85ffccc634f09ac63b4dbe71f7f7d60b62bcd1_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c7b538babba4da963668cddb579e6574422469881295aa38695f25b230d5fb2/preprint__2c7b538babba4da963668cddb579e6574422469881295aa38695f25b230d5fb2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2cdc1f3d48bc297d8d0792cf3e80236c7bca425680de7064173edc0b81851ba8/preprint__2cdc1f3d48bc297d8d0792cf3e80236c7bca425680de7064173edc0b81851ba8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c02bd7a35300bb2bb4b7d686d36107bc23ef52abb35fb1ce7f36e5cf7b43450/preprint__2c02bd7a35300bb2bb4b7d686d36107bc23ef52abb35fb1ce7f36e5cf7b43450_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d3357edbad38d3349cd4c262fca2a99c37284e73c8f6e1e87535db2259562fa/preprint__2d3357edbad38d3349cd4c262fca2a99c37284e73c8f6e1e87535db2259562fa_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c8b8c44f2145d626c12ff525228450bb690f4856e21e95a7b66eda0aec1fa0f/preprint__2c8b8c44f2145d626c12ff525228450bb690f4856e21e95a7b66eda0aec1fa0f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c45b1e1edfefc499c496dd28c12f77e6541555c49953cdf2370bcb581d610ed/preprint__2c45b1e1edfefc499c496dd28c12f77e6541555c49953cdf2370bcb581d610ed_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2c50a5500770d6fb4c199863480f4a2741869fb9082f12209917f01172aad851/preprint__2c50a5500770d6fb4c199863480f4a2741869fb9082f12209917f01172aad851_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ce052fe1d680afbe1af94fbea6482eee6a7adc9740ab863565b687cb664832d/preprint__2ce052fe1d680afbe1af94fbea6482eee6a7adc9740ab863565b687cb664832d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d24d8f5a90404ee2a20c686bfdd88ed45aa3f13ab7fda1970d4bbdaff726673/preprint__2d24d8f5a90404ee2a20c686bfdd88ed45aa3f13ab7fda1970d4bbdaff726673_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2cedf09ced014f0b272145d828208da3a4a8a8e971c35a68fab6d1505f7057ba/preprint__2cedf09ced014f0b272145d828208da3a4a8a8e971c35a68fab6d1505f7057ba_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d75e933e3df0ba871e960581e0f926003fb13f7b7a51178e7131dd0a823a588/preprint__2d75e933e3df0ba871e960581e0f926003fb13f7b7a51178e7131dd0a823a588_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2da02e3f73431489904596075f20db6134915c6428aad654145edf6890777e8c/preprint__2da02e3f73431489904596075f20db6134915c6428aad654145edf6890777e8c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d9779a8ac5d64cac20c7cf4956a80378d6ff818b7f90cb6f98984ef6f9bf457/preprint__2d9779a8ac5d64cac20c7cf4956a80378d6ff818b7f90cb6f98984ef6f9bf457_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2db02b2220855cbd950a3ab1521af6afc5aeb3d2fbeb9956df11fcb4c6593a19/preprint__2db02b2220855cbd950a3ab1521af6afc5aeb3d2fbeb9956df11fcb4c6593a19_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2dfe75a547874ef385de5bcd845623d4e14adc77a5fcdd40529b9cc0ab69bc3c/preprint__2dfe75a547874ef385de5bcd845623d4e14adc77a5fcdd40529b9cc0ab69bc3c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d911e90ffd009a400717e5cb5cf6281afd43425321abee0fba623b5216acec7/preprint__2d911e90ffd009a400717e5cb5cf6281afd43425321abee0fba623b5216acec7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d9ad3ea7c6846674f64b8af378556415293a67507dabb6245cf751b1748534f/preprint__2d9ad3ea7c6846674f64b8af378556415293a67507dabb6245cf751b1748534f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e0e8ce87cb826d5b5cbc0cecb67fd274f572a0c5da2daea01e83def6b4ada3c/preprint__2e0e8ce87cb826d5b5cbc0cecb67fd274f572a0c5da2daea01e83def6b4ada3c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d3e2fd528a6583bb9ce39de20110a8b70a9330e2559f55e38d2febed2247714/preprint__2d3e2fd528a6583bb9ce39de20110a8b70a9330e2559f55e38d2febed2247714_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2dda839372219d83a14ae0199c9b19e93bdf2d04c61829726a2424d318620e0b/preprint__2dda839372219d83a14ae0199c9b19e93bdf2d04c61829726a2424d318620e0b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e034204349e08df1d05f24a07e25dea6baec0bd99113d48fa7db820dcad9c02/preprint__2e034204349e08df1d05f24a07e25dea6baec0bd99113d48fa7db820dcad9c02_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2de3fe26738bdca550cebdff17e627ff7f528939b67b4477e7b7e40bfc2ac8e2/preprint__2de3fe26738bdca550cebdff17e627ff7f528939b67b4477e7b7e40bfc2ac8e2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e182958f4b739b8e48748bc5bd835650effa2ce246c93945aa987b37f838880/preprint__2e182958f4b739b8e48748bc5bd835650effa2ce246c93945aa987b37f838880_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__246f63e06ca8d16871df3ea78075f808ff62ae028bcb0e3fc07dd01786b107b8/preprint__246f63e06ca8d16871df3ea78075f808ff62ae028bcb0e3fc07dd01786b107b8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24f1e3882ae57eb789c0920799be6f3886baab58206b37ed9ba7873e23adb9a9/preprint__24f1e3882ae57eb789c0920799be6f3886baab58206b37ed9ba7873e23adb9a9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24a008a37e7789b34262d8215ba6756a7305ebd33a8193fc1d1321c287f92d0b/preprint__24a008a37e7789b34262d8215ba6756a7305ebd33a8193fc1d1321c287f92d0b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__246b65c5200b9f484ccdc29c1318580790edfbfc992566ac8ed5d0524e8ff954/preprint__246b65c5200b9f484ccdc29c1318580790edfbfc992566ac8ed5d0524e8ff954_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2467c298eee7316a94dc0b589ca5886953919e4e97fcfecfb8450427cc8131e4/preprint__2467c298eee7316a94dc0b589ca5886953919e4e97fcfecfb8450427cc8131e4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2529954b86a96a05157531c788820ff771a093e99605d52d8b4161f17bf8611b/preprint__2529954b86a96a05157531c788820ff771a093e99605d52d8b4161f17bf8611b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24b330c0a0578dba1bd87490312f550f691900ee53d3f5705df6bf74206a8d77/preprint__24b330c0a0578dba1bd87490312f550f691900ee53d3f5705df6bf74206a8d77_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__25265899073cc9aa8bb1a3ab270d131932a71aa4ae1b2ed0c2f2dc2089f081b9/preprint__25265899073cc9aa8bb1a3ab270d131932a71aa4ae1b2ed0c2f2dc2089f081b9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24c1fdcd5d33eb2c6dcab73c4b3841cc014d2b979cb84b0164585a4b26025315/preprint__24c1fdcd5d33eb2c6dcab73c4b3841cc014d2b979cb84b0164585a4b26025315_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__255335177d0f2d1a1a2bf9b5725b87645fc5d556eed88c4fc30a89bf1a65655a/preprint__255335177d0f2d1a1a2bf9b5725b87645fc5d556eed88c4fc30a89bf1a65655a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24e5d34bdded255b886991f04b9b5656e19d20701335748a1b344eabe110205d/preprint__24e5d34bdded255b886991f04b9b5656e19d20701335748a1b344eabe110205d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2547409a402342dc0508dd3d0b635823cd5eefafdca9177e3a60f33301105af1/preprint__2547409a402342dc0508dd3d0b635823cd5eefafdca9177e3a60f33301105af1_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__25a1db52868bd6993ac74e75dd961147bc73fd48a03630df813924b331c71afa/preprint__25a1db52868bd6993ac74e75dd961147bc73fd48a03630df813924b331c71afa_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24c60571af15012393da8ef2f5de83120e71d6efc863787fdb9352b894873949/preprint__24c60571af15012393da8ef2f5de83120e71d6efc863787fdb9352b894873949_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2492f6f288d727a159e794891f1e45651aa47c7b22e9847cc9b84e80832abd28/preprint__2492f6f288d727a159e794891f1e45651aa47c7b22e9847cc9b84e80832abd28_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__245c381d651a5fa70daf66ff28313bc067ab7e62c8a916f1420b98d055c1cd09/preprint__245c381d651a5fa70daf66ff28313bc067ab7e62c8a916f1420b98d055c1cd09_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__25c408184b362d88e4b325539cba6266ed5a074284a0c4b361fcfb1504c30517/preprint__25c408184b362d88e4b325539cba6266ed5a074284a0c4b361fcfb1504c30517_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2632223d5470d6ac00768a9f6c2f5bbcfb7c5629203a5107f56596f8930ce3b5/preprint__2632223d5470d6ac00768a9f6c2f5bbcfb7c5629203a5107f56596f8930ce3b5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__257c548e09ecc51312a80db01c0ff093dc4baa43eea52540ea8785f176278dc2/preprint__257c548e09ecc51312a80db01c0ff093dc4baa43eea52540ea8785f176278dc2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__25554c2b0d763ee7eebd26c935001c9c898413df18de139f1170cb32c3b44d02/preprint__25554c2b0d763ee7eebd26c935001c9c898413df18de139f1170cb32c3b44d02_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__260429c99ce5084104d52cd9c018527d3936c44282318b70cd525c563fe65941/preprint__260429c99ce5084104d52cd9c018527d3936c44282318b70cd525c563fe65941_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2627044d91b253b3dc31175f80f207191bad6237d3998a1f9077d2a906f51ca3/preprint__2627044d91b253b3dc31175f80f207191bad6237d3998a1f9077d2a906f51ca3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__266d2bdaff4671eda29f45925e187839557438136a8b463f7af726b39e8ebce5/preprint__266d2bdaff4671eda29f45925e187839557438136a8b463f7af726b39e8ebce5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__24cfe37c31a2f5417aa65381c72e6cd49ce153b4ccfab3d7a76c372c04bbfc0f/preprint__24cfe37c31a2f5417aa65381c72e6cd49ce153b4ccfab3d7a76c372c04bbfc0f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2669d87695ed810730ad4b1dbb8f435931321c0605b14080576f6ae386246072/preprint__2669d87695ed810730ad4b1dbb8f435931321c0605b14080576f6ae386246072_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2d5602b748e28373dfb643917ba2dfb3979664627ec3a7d67bbe00aa954dee2e/preprint__2d5602b748e28373dfb643917ba2dfb3979664627ec3a7d67bbe00aa954dee2e_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e36f128c9205bee454d3e284bb52b013e69624f6b0c4b9ee87cd0e1a08f4e8e/preprint__2e36f128c9205bee454d3e284bb52b013e69624f6b0c4b9ee87cd0e1a08f4e8e_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e2ef3ae8f2b35f608fdddf3f7bc99085f156439f4828cebef6f5e9e9d343458/preprint__2e2ef3ae8f2b35f608fdddf3f7bc99085f156439f4828cebef6f5e9e9d343458_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2de9a4fc82ff994dd8672bfaa8e66e4ab1f28f221d38c871996f5d667af37fd8/preprint__2de9a4fc82ff994dd8672bfaa8e66e4ab1f28f221d38c871996f5d667af37fd8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e189c25d0c55356b4727508f1b8113351cacb30d93b3abed8aa7dc6cc857542/preprint__2e189c25d0c55356b4727508f1b8113351cacb30d93b3abed8aa7dc6cc857542_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e7743d5d7a2ba764184db67babcb6dbbf4d005f4e6f8980840d44a9e18af6f8/preprint__2e7743d5d7a2ba764184db67babcb6dbbf4d005f4e6f8980840d44a9e18af6f8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e71dd7abdb38ced42b0a09aba42551092b2ce3fb4bad814054685be616d5d44/preprint__2e71dd7abdb38ced42b0a09aba42551092b2ce3fb4bad814054685be616d5d44_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e9ad4ed25e7e2fda7d33f071f6c54c42637453022d051fbeaf52819f5ffa24b/preprint__2e9ad4ed25e7e2fda7d33f071f6c54c42637453022d051fbeaf52819f5ffa24b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ebb16444233a71111fb3c62270dcea3ed3eb95c11ccf9c2c2a44ac7be7e127a/preprint__2ebb16444233a71111fb3c62270dcea3ed3eb95c11ccf9c2c2a44ac7be7e127a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ea1140f63d62f1fe75d9ef35811d317ae35088342ee27f066b549722c3804a2/preprint__2ea1140f63d62f1fe75d9ef35811d317ae35088342ee27f066b549722c3804a2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2eae8df367675917e8407cdb0a1ceeb42f57d86326d5b3e8aa36d1fb848ce826/preprint__2eae8df367675917e8407cdb0a1ceeb42f57d86326d5b3e8aa36d1fb848ce826_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e4298f10a7b265dc4b10ce13c201816f8a481ad667e66496cfd9e516a41cdd0/preprint__2e4298f10a7b265dc4b10ce13c201816f8a481ad667e66496cfd9e516a41cdd0_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f27db7c5e80138fc14a0e3c624c2f56de63ff1ec634093813bca9a32c6f2c8e/preprint__2f27db7c5e80138fc14a0e3c624c2f56de63ff1ec634093813bca9a32c6f2c8e_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ef69b0c566c1f2cc63708408f3dccf10816f0c7735ceb258611b137dca6f18b/preprint__2ef69b0c566c1f2cc63708408f3dccf10816f0c7735ceb258611b137dca6f18b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ea3719c59831b70f03826f86889300eaeb8fdabeeb89e66cafe3f5f4d6b12ae/preprint__2ea3719c59831b70f03826f86889300eaeb8fdabeeb89e66cafe3f5f4d6b12ae_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2e81ad31a5b77d0c6eda8ab6103ce141edac8e4788801f40793988482358fab2/preprint__2e81ad31a5b77d0c6eda8ab6103ce141edac8e4788801f40793988482358fab2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f238fc4de049b31c3ee1ae09e051f0ab53712d97e4910ca3de302bbe0df3052/preprint__2f238fc4de049b31c3ee1ae09e051f0ab53712d97e4910ca3de302bbe0df3052_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f4f7ea2e0459f556d4cac6390657ec8d7d8826ec7f873d1526cac29b9601c54/preprint__2f4f7ea2e0459f556d4cac6390657ec8d7d8826ec7f873d1526cac29b9601c54_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ee6ff203672bcd010e3a6395b2e232a633cd36cf5f9fa2d1d924ab3a0a0d7b6/preprint__2ee6ff203672bcd010e3a6395b2e232a633cd36cf5f9fa2d1d924ab3a0a0d7b6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ee8f81403bfe23223cbd386a1f7d4e6811d3c0bd6cf856cb65126f5f1ab8bf7/preprint__2ee8f81403bfe23223cbd386a1f7d4e6811d3c0bd6cf856cb65126f5f1ab8bf7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f8bf672905a0289a94a5056c950297588158cf2fa9b884b5bb145c7d425a632/preprint__2f8bf672905a0289a94a5056c950297588158cf2fa9b884b5bb145c7d425a632_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f83e4bfae70e913b2f2cb36cb7cf5a75ccf4888f335fa6b1af3ad8414dff762/preprint__2f83e4bfae70e913b2f2cb36cb7cf5a75ccf4888f335fa6b1af3ad8414dff762_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2eec5aaca567f7c7b429faee2786f28e062c1d59586903696039949a2eebd4b1/preprint__2eec5aaca567f7c7b429faee2786f28e062c1d59586903696039949a2eebd4b1_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30fb8a871511ba072ebc62896d6e39002b6284404624badb14553c91bf68f5b5/preprint__30fb8a871511ba072ebc62896d6e39002b6284404624badb14553c91bf68f5b5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__313140db08c4f8cfa6b6c4789226129265ad7d3df364cc8852eb4ba183201b05/preprint__313140db08c4f8cfa6b6c4789226129265ad7d3df364cc8852eb4ba183201b05_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31dc3a1ef6ef06c5863d27c44672af66aa0cafa16e606a36e13c59ea9055c593/preprint__31dc3a1ef6ef06c5863d27c44672af66aa0cafa16e606a36e13c59ea9055c593_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__315a9d30a4cd5b2a37f750b48f7bdeb8bc634ca7fe034dd61211c3ae968d5036/preprint__315a9d30a4cd5b2a37f750b48f7bdeb8bc634ca7fe034dd61211c3ae968d5036_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31a6403df2120123d5c29e362ed881e4649ba8cce19b87daa7e5b5c52c8e079f/preprint__31a6403df2120123d5c29e362ed881e4649ba8cce19b87daa7e5b5c52c8e079f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3214d332380793801e11229df96135b47fb44b61f6985948a4a8bcd0fe38715b/preprint__3214d332380793801e11229df96135b47fb44b61f6985948a4a8bcd0fe38715b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__32241aa02e0808f22f5f9ce80d2db6adf2265ba188e5ba4eb10f320c65895854/preprint__32241aa02e0808f22f5f9ce80d2db6adf2265ba188e5ba4eb10f320c65895854_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31cf8d3db7d6bf52376af0e47bb1fdf4738d709005ed98f0dfb2a7461947e87b/preprint__31cf8d3db7d6bf52376af0e47bb1fdf4738d709005ed98f0dfb2a7461947e87b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31abd5f9056f44c3bc07b9423674f38ef0f1df4f691156160d21f0b9fb21d6a0/preprint__31abd5f9056f44c3bc07b9423674f38ef0f1df4f691156160d21f0b9fb21d6a0_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__318eec874a1f79b2861c4962769fb74c2392dbf3765be8faf2a59155d55111fb/preprint__318eec874a1f79b2861c4962769fb74c2392dbf3765be8faf2a59155d55111fb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__33c7c7a79a4ea6e6525114cc68b2c23aff72d2a2fd773375a1dfb819558f1217/preprint__33c7c7a79a4ea6e6525114cc68b2c23aff72d2a2fd773375a1dfb819558f1217_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31d9b47917dd0166fdb30105a94aa912483c4f1c2046f2fb25faf0d74f9f0b89/preprint__31d9b47917dd0166fdb30105a94aa912483c4f1c2046f2fb25faf0d74f9f0b89_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__33ae254b401e2ece1f639a50b3750eda55bd243c6ae1ca1e2ff64b3757635924/preprint__33ae254b401e2ece1f639a50b3750eda55bd243c6ae1ca1e2ff64b3757635924_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__32bd9c665f2b795ae3089961474735a437de5b3e9fd200d28fc75d3f8bcd0289/preprint__32bd9c665f2b795ae3089961474735a437de5b3e9fd200d28fc75d3f8bcd0289_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__32ace780350ede0e71dbb5f19a5d38ed0113d460f7f9c162f2955552d59e16e4/preprint__32ace780350ede0e71dbb5f19a5d38ed0113d460f7f9c162f2955552d59e16e4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__32c379bc583225652929f6dbd86afe289084c66935a92e09bc78ed875d01f835/preprint__32c379bc583225652929f6dbd86afe289084c66935a92e09bc78ed875d01f835_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__31fc06c16f451e6170f6dfa06368d71961e055959fc9a0a6e5b57ed96b14b273/preprint__31fc06c16f451e6170f6dfa06368d71961e055959fc9a0a6e5b57ed96b14b273_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3403160146f5236294307f5fa60a93e363638cf31b1b23f379cad1e90ad890a5/preprint__3403160146f5236294307f5fa60a93e363638cf31b1b23f379cad1e90ad890a5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__33e22f0224bdc6273d0a4c1458943434dddb70e182ed284d14f599dc0e41a424/preprint__33e22f0224bdc6273d0a4c1458943434dddb70e182ed284d14f599dc0e41a424_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3177644c5b7883d30c5033048c8e97da78802e9fe216679931ba7934e0afab6c/preprint__3177644c5b7883d30c5033048c8e97da78802e9fe216679931ba7934e0afab6c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__337e571503b28d15abdfc4ba75ff294d568592c2aadf53eb1abf668f2ed15f8a/preprint__337e571503b28d15abdfc4ba75ff294d568592c2aadf53eb1abf668f2ed15f8a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3445cb083767b32e7f13fbb196e16f883f117647191863a7e0925c3252840e9f/preprint__3445cb083767b32e7f13fbb196e16f883f117647191863a7e0925c3252840e9f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__328d5610bc05c2851e0bb9dafbe97f004c094f890bac97cf0f2987dd784643f8/preprint__328d5610bc05c2851e0bb9dafbe97f004c094f890bac97cf0f2987dd784643f8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__335f0963cb80c7128945ab73d09895a86d23ca1644dcb07bab2c97a17dfd8457/preprint__335f0963cb80c7128945ab73d09895a86d23ca1644dcb07bab2c97a17dfd8457_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3426434fcf7b3c2d7e63abe62d5af003a8db54a12a63f059de9af0e894b45a72/preprint__3426434fcf7b3c2d7e63abe62d5af003a8db54a12a63f059de9af0e894b45a72_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__349512665ea9513600c96895a09e62cfa732ef5bdbea00e42fcb3dffe2616ba4/preprint__349512665ea9513600c96895a09e62cfa732ef5bdbea00e42fcb3dffe2616ba4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a3b32522cbc0d0451d8b8ffdb0e8ed7f0a41bce4120f1a0b3a3cd5998dbc851/preprint__2a3b32522cbc0d0451d8b8ffdb0e8ed7f0a41bce4120f1a0b3a3cd5998dbc851_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__295676132315fbb744a24dcf1773d3ba715b0568fe8a8dcb13573b59f2c91ff3/preprint__295676132315fbb744a24dcf1773d3ba715b0568fe8a8dcb13573b59f2c91ff3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__29dad16850741056ed6eb25ca9df9573fc7efdde31cfd4796dd32be9720fc3b8/preprint__29dad16850741056ed6eb25ca9df9573fc7efdde31cfd4796dd32be9720fc3b8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a669417a53fc85ebf06ff4010ca26aa926fcfbe4167676f6ab94e29b345fb00/preprint__2a669417a53fc85ebf06ff4010ca26aa926fcfbe4167676f6ab94e29b345fb00_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a46ca497ddc02b1031b11efc7d80f7e28b4bf1a1b39bdf291cf15464d8a300b/preprint__2a46ca497ddc02b1031b11efc7d80f7e28b4bf1a1b39bdf291cf15464d8a300b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a70f6fca42a0ab1796c07cdeb938f8e1fc2a923251322318ec8581d8ba6cdbf/preprint__2a70f6fca42a0ab1796c07cdeb938f8e1fc2a923251322318ec8581d8ba6cdbf_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2af794e8bdc6d7671634e08f6af1da53069c1fb5bf84792aa87006addcb20726/preprint__2af794e8bdc6d7671634e08f6af1da53069c1fb5bf84792aa87006addcb20726_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a9a4b44f42ee12abc0632b48fbbeb22729b5e08e4fac14f9605dcd0750b73d0/preprint__2a9a4b44f42ee12abc0632b48fbbeb22729b5e08e4fac14f9605dcd0750b73d0_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2a7545a262e748ebc0ab9d89acde6b8030a0183dd22cadb6d94fc17e681cfcc7/preprint__2a7545a262e748ebc0ab9d89acde6b8030a0183dd22cadb6d94fc17e681cfcc7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2af9ac54537a84d7e65fbe5670f2af99cc7c43c1b79b4d127a1a2be44b23db02/preprint__2af9ac54537a84d7e65fbe5670f2af99cc7c43c1b79b4d127a1a2be44b23db02_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ae36470298dc21812c2daf83a1347cd8d7520177c40d85a6ff8aa616b82b269/preprint__2ae36470298dc21812c2daf83a1347cd8d7520177c40d85a6ff8aa616b82b269_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2aba0cad8c9e21f18a62e920b616ed2495a0f2c186dc941277f572951cc0528c/preprint__2aba0cad8c9e21f18a62e920b616ed2495a0f2c186dc941277f572951cc0528c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2aa87061110a002049ddc4429a39b3bcb5a6c40391532d938213001e3831948b/preprint__2aa87061110a002049ddc4429a39b3bcb5a6c40391532d938213001e3831948b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ad30f94c3b90c96475bc36c1aff17d563b878206dfefe71283989f76bd09092/preprint__2ad30f94c3b90c96475bc36c1aff17d563b878206dfefe71283989f76bd09092_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b43013de3bb82ae9a113872a06ac68ff47a1fe3e657a67c2b2adc2ace467e88/preprint__2b43013de3bb82ae9a113872a06ac68ff47a1fe3e657a67c2b2adc2ace467e88_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2bbfafaeea0e549a06406df3ec0de2092667f753880d5d03f95c7b3aace942cb/preprint__2bbfafaeea0e549a06406df3ec0de2092667f753880d5d03f95c7b3aace942cb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b5449656c9a8636b77273dc85ce96cbe00a6238ddd32058f9565878862bd45c/preprint__2b5449656c9a8636b77273dc85ce96cbe00a6238ddd32058f9565878862bd45c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b8c98fb229297a1262aa712f9c884323c7fbe1c918252cec11c83d0130a8dc7/preprint__2b8c98fb229297a1262aa712f9c884323c7fbe1c918252cec11c83d0130a8dc7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b4640cbbd24fef518b88dcfeb69d338b82c53d043af3c97eb48d85e522c80ce/preprint__2b4640cbbd24fef518b88dcfeb69d338b82c53d043af3c97eb48d85e522c80ce_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b38d2667b0ed62d397a8a20a8d846f75734d0824b7f39e0811c05ef40cb6eba/preprint__2b38d2667b0ed62d397a8a20a8d846f75734d0824b7f39e0811c05ef40cb6eba_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2bad4797b5cc24f4647b5de99a0eb4e23c0e77ca443e03e3c2737e1043560bec/preprint__2bad4797b5cc24f4647b5de99a0eb4e23c0e77ca443e03e3c2737e1043560bec_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2be28d29fac1837acb16f3a122ee991594a1cc474dcaae29e51328cc54c6d461/preprint__2be28d29fac1837acb16f3a122ee991594a1cc474dcaae29e51328cc54c6d461_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b0f6ea9e001b8d5fc2be38173edd65f51a0d7285b0182711c321e987e96610a/preprint__2b0f6ea9e001b8d5fc2be38173edd65f51a0d7285b0182711c321e987e96610a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2b06af1324d99d554e163785b31fdb051bec082f852a83042e26b77edd7d2edd/preprint__2b06af1324d99d554e163785b31fdb051bec082f852a83042e26b77edd7d2edd_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__342f779650189c20090b8a3b8892c697eb458158c9e9145d2c4ecb459e16b586/preprint__342f779650189c20090b8a3b8892c697eb458158c9e9145d2c4ecb459e16b586_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__342f3fd21d75282da5770a148cc3cd02d2be3b92f6778de7b311bf25ca22ab32/preprint__342f3fd21d75282da5770a148cc3cd02d2be3b92f6778de7b311bf25ca22ab32_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3336aba4117893c109b0f4991fea84ae9b3245765bc747b80f5a2a0196585fde/preprint__3336aba4117893c109b0f4991fea84ae9b3245765bc747b80f5a2a0196585fde_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__342b4c91b52b15a02db8d58ce166847173245bc30ad0b421e81f0f9dbf3a118f/preprint__342b4c91b52b15a02db8d58ce166847173245bc30ad0b421e81f0f9dbf3a118f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34fa333eb87af58697ed7152ac3e636f46795c0afd2388548e2bd581c65dbba9/preprint__34fa333eb87af58697ed7152ac3e636f46795c0afd2388548e2bd581c65dbba9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34c395e47cc6d4a8f46c4c25723b4754edcce440bdf0db6c9dd8333b1f2933f6/preprint__34c395e47cc6d4a8f46c4c25723b4754edcce440bdf0db6c9dd8333b1f2933f6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34bec9b81f07cda8f1add97db6f6f763b4b0ae16a8a4dd14262b539266b6b9be/preprint__34bec9b81f07cda8f1add97db6f6f763b4b0ae16a8a4dd14262b539266b6b9be_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__354df0d06839fa73c5f667e92fab85e60c6cb7600bee7d1519d47db22338a3d3/preprint__354df0d06839fa73c5f667e92fab85e60c6cb7600bee7d1519d47db22338a3d3_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__347f89f117f89bce457a8943a2bf082613969fd76baae7ba6aa0eb94b6222128/preprint__347f89f117f89bce457a8943a2bf082613969fd76baae7ba6aa0eb94b6222128_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34a06a85a426767e87cb3ef1826a9a91b94ab12100349007e92f54a0b900d1dd/preprint__34a06a85a426767e87cb3ef1826a9a91b94ab12100349007e92f54a0b900d1dd_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34ba1e997f44de23c566cb3c7a75ea02ec48f6aa8a786b3b6397e59c8f1450bb/preprint__34ba1e997f44de23c566cb3c7a75ea02ec48f6aa8a786b3b6397e59c8f1450bb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__34f8cc2494678eccba92c21680c985ce8330218c158efce968c984d6bba37972/preprint__34f8cc2494678eccba92c21680c985ce8330218c158efce968c984d6bba37972_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3588b3fb58e82875241f8f1e0e8adcf5dd2fd7f827453efe4d120462583fc319/preprint__3588b3fb58e82875241f8f1e0e8adcf5dd2fd7f827453efe4d120462583fc319_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35a40e26f10afead2ffabbe35404dc5029b4e0010ab4f0d49541da174f2ee902/preprint__35a40e26f10afead2ffabbe35404dc5029b4e0010ab4f0d49541da174f2ee902_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__359ed2c0a5f6012ff9d7a2466f230a6c6232f1a26600a131d0512ff136512d94/preprint__359ed2c0a5f6012ff9d7a2466f230a6c6232f1a26600a131d0512ff136512d94_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35c1d37e8b184d22d46ec9f00215b335326ba248ec2fe2b03eecec894ee76d67/preprint__35c1d37e8b184d22d46ec9f00215b335326ba248ec2fe2b03eecec894ee76d67_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35a1b0ba3d413a0f096a2e4ad20495966c78d8e92fc8663b4b249c9167d3ae31/preprint__35a1b0ba3d413a0f096a2e4ad20495966c78d8e92fc8663b4b249c9167d3ae31_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__363ba892998c3b6f04ff0ae755cc929c65a53de8b5218df0e55c4cca048c2192/preprint__363ba892998c3b6f04ff0ae755cc929c65a53de8b5218df0e55c4cca048c2192_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3525642c741f5f11ebaa926122ff30a8be3309b244a21f06a895dff02ef385d5/preprint__3525642c741f5f11ebaa926122ff30a8be3309b244a21f06a895dff02ef385d5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3620547d977be9c8dee6973b663ad77c0a755f7305b9fa90f34f01e57bab715b/preprint__3620547d977be9c8dee6973b663ad77c0a755f7305b9fa90f34f01e57bab715b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__361ec125351838cc5022b3f2159039d6a75885d96a4e50cd1c2ee5b7fdd9e01a/preprint__361ec125351838cc5022b3f2159039d6a75885d96a4e50cd1c2ee5b7fdd9e01a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__360eeea25f37e0b2e93e83be4bc0c51a401f3c86becafce9f989d9329234a302/preprint__360eeea25f37e0b2e93e83be4bc0c51a401f3c86becafce9f989d9329234a302_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35a5d9cf4004e4fe2423ae2d95572c2639d56568d24c77e5374c212be7d4d2fb/preprint__35a5d9cf4004e4fe2423ae2d95572c2639d56568d24c77e5374c212be7d4d2fb_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35a1fdd71ee6f027730302bd725b19360b1f84d5dd262925df2b1a86e9a88ae6/preprint__35a1fdd71ee6f027730302bd725b19360b1f84d5dd262925df2b1a86e9a88ae6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35e23d15b5cd6a6464eae0671a445daff29d5876d370babb087465ffd8c10d42/preprint__35e23d15b5cd6a6464eae0671a445daff29d5876d370babb087465ffd8c10d42_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__35fa0977a677b99e455ea546acc80f7d943249439f5e400425dc6fedf877ea5e/preprint__35fa0977a677b99e455ea546acc80f7d943249439f5e400425dc6fedf877ea5e_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2fd76270c44ced0d81d886d21d25410d474aa0327f8385d69939db264a378f00/preprint__2fd76270c44ced0d81d886d21d25410d474aa0327f8385d69939db264a378f00_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f5d374dbba83252a29c3cd08a9b82564f64f4afc6418192b9ef2b3ea3a54a50/preprint__2f5d374dbba83252a29c3cd08a9b82564f64f4afc6418192b9ef2b3ea3a54a50_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3049aa5011ee07429f7ea579876dae9c3e5d13683d42216644e50bd87113f033/preprint__3049aa5011ee07429f7ea579876dae9c3e5d13683d42216644e50bd87113f033_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2f942f917337d281584df998d55b7a08758b7fa03ba381aa5690b17fbf8675f1/preprint__2f942f917337d281584df998d55b7a08758b7fa03ba381aa5690b17fbf8675f1_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2ff76be2fbc4195640d845c1c0b393aca9625981edff7c0a97d5c098a18dada9/preprint__2ff76be2fbc4195640d845c1c0b393aca9625981edff7c0a97d5c098a18dada9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__2fb73edd476fde9061cb7efcf9d2dc5eae7abec436a8a6c788d80d796a61ff6d/preprint__2fb73edd476fde9061cb7efcf9d2dc5eae7abec436a8a6c788d80d796a61ff6d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__306727481ec4eb1992bd7bc51cf013ec6c115d2bdfca6129aab01512b4a02d9a/preprint__306727481ec4eb1992bd7bc51cf013ec6c115d2bdfca6129aab01512b4a02d9a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3058e16b0334a8bff28c01ae0e0fed1769305b973de6b8b3b4b969dcd84ab099/preprint__3058e16b0334a8bff28c01ae0e0fed1769305b973de6b8b3b4b969dcd84ab099_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__306610d42159947b7e1643b27d4277932776a9eeb3e8b286d821607475e994ae/preprint__306610d42159947b7e1643b27d4277932776a9eeb3e8b286d821607475e994ae_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30b50e57b10e7092afc86d2ca00499aff1c3ee8947c1a204e1cd13bd62a1b70d/preprint__30b50e57b10e7092afc86d2ca00499aff1c3ee8947c1a204e1cd13bd62a1b70d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30f361db806113189ab62b43bc8cf988a227722f081ae346abee125eada8b282/preprint__30f361db806113189ab62b43bc8cf988a227722f081ae346abee125eada8b282_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3010135e41781228d850fe66e1f638b48c5fe2b5c5130a79f5c9adf90c56d09f/preprint__3010135e41781228d850fe66e1f638b48c5fe2b5c5130a79f5c9adf90c56d09f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30cbc9f70271c3e4226fab514cde64fc2fbc4ad41b62033575fd6d32f8da4528/preprint__30cbc9f70271c3e4226fab514cde64fc2fbc4ad41b62033575fd6d32f8da4528_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30434345bd5ab72c94ba3277bb3d90c07f08a9f83fe78011ea5ae5101e656b96/preprint__30434345bd5ab72c94ba3277bb3d90c07f08a9f83fe78011ea5ae5101e656b96_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30df85193f44f960e1c50ea9c84923e9924bb902a7849ac28597bc69a31a1499/preprint__30df85193f44f960e1c50ea9c84923e9924bb902a7849ac28597bc69a31a1499_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30a0b46e3dd03f029f85286ab556631ced9914a88f5f11a62ef1ebfeb44d94d7/preprint__30a0b46e3dd03f029f85286ab556631ced9914a88f5f11a62ef1ebfeb44d94d7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__316d6bc40a3fc195bb78544249cea1536e233328a22d455e767e2dc69c59527b/preprint__316d6bc40a3fc195bb78544249cea1536e233328a22d455e767e2dc69c59527b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__303ac869c88f0f74f9d9463d009c34f3fb268db38120e6b8641bcf9306e62a50/preprint__303ac869c88f0f74f9d9463d009c34f3fb268db38120e6b8641bcf9306e62a50_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30094d8d4bccb3adbd175c32d81a7b94ee7bfe47d192e8f91240c2e180184a6c/preprint__30094d8d4bccb3adbd175c32d81a7b94ee7bfe47d192e8f91240c2e180184a6c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__307521a685de29b8c568c09a03351bf3cee1e7b71522dee0161b3b7b81e423f9/preprint__307521a685de29b8c568c09a03351bf3cee1e7b71522dee0161b3b7b81e423f9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3172d81991821eb7d48768bebee796b51490d351282429b0e80f15c4cba1abdd/preprint__3172d81991821eb7d48768bebee796b51490d351282429b0e80f15c4cba1abdd_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3178cef45b8f6224ac0bd57318c086e2d578a6c84b0379c7f84eeeb3862d93e2/preprint__3178cef45b8f6224ac0bd57318c086e2d578a6c84b0379c7f84eeeb3862d93e2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__30fb8fd87ffed41ebbba9b56ec520db77d9ea19ec5793debe43f8a22e4ead43f/preprint__30fb8fd87ffed41ebbba9b56ec520db77d9ea19ec5793debe43f8a22e4ead43f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__381fa22523ac5802875ee3d05af7e4fd3171237975d12c920bc665a5251700f6/preprint__381fa22523ac5802875ee3d05af7e4fd3171237975d12c920bc665a5251700f6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__39491cd20423273d3bb4618300d071b55cbb17d955b09e6a2585a55f87aa77d4/preprint__39491cd20423273d3bb4618300d071b55cbb17d955b09e6a2585a55f87aa77d4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3917aff68ae7c5e3c114bd690c9ddbf36b9f09a045882efa59e84cf6e32fc33a/preprint__3917aff68ae7c5e3c114bd690c9ddbf36b9f09a045882efa59e84cf6e32fc33a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__38e389ccc21de9ed707455dc5515d104a1e8c9438fd86bbebea52953e5b4679b/preprint__38e389ccc21de9ed707455dc5515d104a1e8c9438fd86bbebea52953e5b4679b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__396d3817bed3bdb747636b7bfdaf1a277372fa910b838358ccaae341697a8ee0/preprint__396d3817bed3bdb747636b7bfdaf1a277372fa910b838358ccaae341697a8ee0_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__397452e1ee14ed5f22812b8e1da6824d4b9b33f94318ceb1dbb6a09dbe89e4c8/preprint__397452e1ee14ed5f22812b8e1da6824d4b9b33f94318ceb1dbb6a09dbe89e4c8_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__39aec96c9c63c7028c9283ada82112a6fbe387c7169d29354535e45bd5ee8a7f/preprint__39aec96c9c63c7028c9283ada82112a6fbe387c7169d29354535e45bd5ee8a7f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3950811418a8a6567b6f2d5fc38280f60c31ccc85d6fc6b06137525ccea23fd7/preprint__3950811418a8a6567b6f2d5fc38280f60c31ccc85d6fc6b06137525ccea23fd7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__391721f6c94448ed3f6f15b0b459c48fa1d13adb48352d5ac73137982e708ced/preprint__391721f6c94448ed3f6f15b0b459c48fa1d13adb48352d5ac73137982e708ced_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__397710f2ea6a135e06eedf9b7d0a99e9059021d68fbdbca77d970fe9388e7832/preprint__397710f2ea6a135e06eedf9b7d0a99e9059021d68fbdbca77d970fe9388e7832_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3a3f03c05ae9d6c9bd69a27705191f0f89a9706538ce50a826ac6e08a04c46bd/preprint__3a3f03c05ae9d6c9bd69a27705191f0f89a9706538ce50a826ac6e08a04c46bd_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__39dfe7c56d9955b321a818aa5e8436bc0a4efef587e5f710f80d3cd7d2660fe0/preprint__39dfe7c56d9955b321a818aa5e8436bc0a4efef587e5f710f80d3cd7d2660fe0_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__38990c3d036933012cfa7883edc7cbeb5dee2d91efcdf19fc817ef0b32a53d39/preprint__38990c3d036933012cfa7883edc7cbeb5dee2d91efcdf19fc817ef0b32a53d39_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__393dd3295a12d84e8a560663c4447b77a2ec7da5b09bc11c1b3b7b9479228989/preprint__393dd3295a12d84e8a560663c4447b77a2ec7da5b09bc11c1b3b7b9479228989_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__397f65ea1e26d1048862cedab157d09e19199454717daef406fdccae114acfc6/preprint__397f65ea1e26d1048862cedab157d09e19199454717daef406fdccae114acfc6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__39cf5ba3ed9440b5cd2967a858ff1f7e6b1b94d56b71111096b4f8de475816d4/preprint__39cf5ba3ed9440b5cd2967a858ff1f7e6b1b94d56b71111096b4f8de475816d4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3ab80f57f86199a145a5b3342311655b26b6e155140b1beaec05562a5b92c1ba/preprint__3ab80f57f86199a145a5b3342311655b26b6e155140b1beaec05562a5b92c1ba_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3aae022c78bf5f72c01c1f6a4d8b1e50e2ebdf13bf2ad8a366a1d59a175c09d5/preprint__3aae022c78bf5f72c01c1f6a4d8b1e50e2ebdf13bf2ad8a366a1d59a175c09d5_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3a407f2d43487bffff66bf302a738d9a576bf26b2a9ab8966b8f0c595afb1ae9/preprint__3a407f2d43487bffff66bf302a738d9a576bf26b2a9ab8966b8f0c595afb1ae9_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3a6a060846223d688b5214f75628ea004649aa4301bee02fdda2a285b84d9b74/preprint__3a6a060846223d688b5214f75628ea004649aa4301bee02fdda2a285b84d9b74_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3af699911459909cdea64c1e3f042ed5e741f606bef7128c7a388be1117b1490/preprint__3af699911459909cdea64c1e3f042ed5e741f606bef7128c7a388be1117b1490_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3a8779812493791762eaae2fa18b44a05cbe747996aeb29ef061c778922d6331/preprint__3a8779812493791762eaae2fa18b44a05cbe747996aeb29ef061c778922d6331_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3ae5d4ed5962acb8d102fc10f98220ffcdaf7bdfe8c898e91cdc554b3555797d/preprint__3ae5d4ed5962acb8d102fc10f98220ffcdaf7bdfe8c898e91cdc554b3555797d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3a66655d37e36bfe7dda338793f235e4e95053956515eb09ae148574738ab612/preprint__3a66655d37e36bfe7dda338793f235e4e95053956515eb09ae148574738ab612_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3b14464004927d09480859894ef11091611af14cec537766261e3213fa1e0c7f/preprint__3b14464004927d09480859894ef11091611af14cec537766261e3213fa1e0c7f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3ab18e53c6db2ccdfe6037a67baa7f47c89d1b411c49515c1379d5baeb99be30/preprint__3ab18e53c6db2ccdfe6037a67baa7f47c89d1b411c49515c1379d5baeb99be30_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3b219dab4dc1f94ac14f00fa0ad67749164903a4f7ceb1ade2ba1e4117df28a2/preprint__3b219dab4dc1f94ac14f00fa0ad67749164903a4f7ceb1ade2ba1e4117df28a2_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__36f08dbb613d10431da0e2397ba2d9a744b2e4b2a9439c0bfb4f32bc3b952c52/preprint__36f08dbb613d10431da0e2397ba2d9a744b2e4b2a9439c0bfb4f32bc3b952c52_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__365ab6394b56346ddcb13d9b2c1abc2800a90c51dac6c626e75b21362960b70f/preprint__365ab6394b56346ddcb13d9b2c1abc2800a90c51dac6c626e75b21362960b70f_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3666774aaa9006a7fa601b37f1ab0a814895e59d243272164e6bd436e31523a6/preprint__3666774aaa9006a7fa601b37f1ab0a814895e59d243272164e6bd436e31523a6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__369090e359c8f7445e1475af7ecee92fbdfdb6633572d242b08c3a3a916e68aa/preprint__369090e359c8f7445e1475af7ecee92fbdfdb6633572d242b08c3a3a916e68aa_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__370ba59c7e2e4b5891a18183ee926a07d9637b6fe4ce807bcab1bd9422beda6b/preprint__370ba59c7e2e4b5891a18183ee926a07d9637b6fe4ce807bcab1bd9422beda6b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3726e0dcf759787b3f69fbad2cd0be78535293e2ea369f30137c7c123fc6ed42/preprint__3726e0dcf759787b3f69fbad2cd0be78535293e2ea369f30137c7c123fc6ed42_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__37410ae9e646ec6cfa85be4679bc2bb151b7e390b311e5f5caaa29e9b865c4c4/preprint__37410ae9e646ec6cfa85be4679bc2bb151b7e390b311e5f5caaa29e9b865c4c4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__37c1840535b26d442aa26b39601af406e56061c87f14efc7bc20d042e76956e7/preprint__37c1840535b26d442aa26b39601af406e56061c87f14efc7bc20d042e76956e7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__36ad814f9f5fd2cb730e5fb3f6278a895451d617f852cc13575b753eebb7389a/preprint__36ad814f9f5fd2cb730e5fb3f6278a895451d617f852cc13575b753eebb7389a_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__373fda8ee712e0f778dbd0a713c97239dc5e0289cf8ebe825bdeac468700297d/preprint__373fda8ee712e0f778dbd0a713c97239dc5e0289cf8ebe825bdeac468700297d_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__374c80a1127a3167a1259abdbf7433b94b859983f3b989ac957bcc284d13860b/preprint__374c80a1127a3167a1259abdbf7433b94b859983f3b989ac957bcc284d13860b_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__382100abf113b49bd6bbe27f7b92a52d0a0e7fe0b444d4e98ffff1077053df95/preprint__382100abf113b49bd6bbe27f7b92a52d0a0e7fe0b444d4e98ffff1077053df95_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__37c6e3af27309dfa66c62c7c8fc72f0820d0184ce377d9b3edbcdb29225658d4/preprint__37c6e3af27309dfa66c62c7c8fc72f0820d0184ce377d9b3edbcdb29225658d4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__37c9af41725d3672b64f3149f120fdb0efda630fc3d482f405338685f766165c/preprint__37c9af41725d3672b64f3149f120fdb0efda630fc3d482f405338685f766165c_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__389254bdc5a987ab8d07c8e19ed033629f890d922c9cbac54afdd1dc650176ba/preprint__389254bdc5a987ab8d07c8e19ed033629f890d922c9cbac54afdd1dc650176ba_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3802cd821026448100b65883f97e0246f1e45ede7cb453f52f51d096309b6fbe/preprint__3802cd821026448100b65883f97e0246f1e45ede7cb453f52f51d096309b6fbe_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3830f46e694f8dd69a4ec2d3792b56cd7b85a6b61ecf7e038f002e669e92bfe4/preprint__3830f46e694f8dd69a4ec2d3792b56cd7b85a6b61ecf7e038f002e669e92bfe4_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__38c26dbbdb8a0893582e3d008a4b6167c44aff42a31a2ba29c754d75fc356f09/preprint__38c26dbbdb8a0893582e3d008a4b6167c44aff42a31a2ba29c754d75fc356f09_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__385eba12651ca14ca40446725b2cdb543d0ea1431d5e9b1326fccd6ab5c05e94/preprint__385eba12651ca14ca40446725b2cdb543d0ea1431d5e9b1326fccd6ab5c05e94_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__3804f4d067d29fa6c47e21129b1f36471a862895cea382b50ff004a7b0975fb7/preprint__3804f4d067d29fa6c47e21129b1f36471a862895cea382b50ff004a7b0975fb7_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__39174209d0233158e33c46ef361d060c6a2a584d09d7e6ec7d58818d7e86fec6/preprint__39174209d0233158e33c46ef361d060c6a2a584d09d7e6ec7d58818d7e86fec6_layouts.pdf filter=lfs diff=lfs merge=lfs -text
+preprint/preprint__382891b67562d7dea62f7b2296235e31ef97c54879b0cd210b88fe2bc6b0bb0e/preprint__382891b67562d7dea62f7b2296235e31ef97c54879b0cd210b88fe2bc6b0bb0e_layouts.pdf filter=lfs diff=lfs merge=lfs -text

peer_reviews/supplementary_0_Peer Review File__1d499c0ca1281bcec022d024917ac5bad140f18fbf5fe272e3fb5cbe41be03f0/images_list.json

@@ -0,0 +1,87 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_1.jpg",
+    "caption": "Ancillary Figure 1",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 0
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_1.jpg",
+    "caption": "Ancillary Figure 1. SKP2 is regulated by a MYOD-bound enhancer.",
+    "footnote": [],
+    "bbox": [
+      [
+        50,
+        241,
+        811,
+        425
+      ]
+    ],
+    "page_idx": 20
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_5.jpg",
+    "caption": "Ancillary Figure 5. E2F1 does not regulates SKP2 expression in rhabdomyosarcoma cells.",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 20
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_6.jpg",
+    "caption": "Aancillary Figure 6. MYOD1 is a strong and selective dependency in rhabdomyosarcoma cells.",
+    "footnote": [],
+    "bbox": [
+      [
+        50,
+        68,
+        571,
+        350
+      ]
+    ],
+    "page_idx": 21
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_0.jpg",
+    "caption": "DAPI/MyHC",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 22
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_8.jpg",
+    "caption": "Ancillary Figure 8. MYOD overexpression does not affect rhabdomyosarcoma cell growth.",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 23
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_10.jpg",
+    "caption": "Ancillary Figure 10. Sensitivity to MLN4924 correlates with SKP2 expression but not with MYOD1 expression",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 24
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_11.jpg",
+    "caption": "Ancillary Figure 11. MYOD binding at SKP2 locus after SKP2 silencing",
+    "footnote": [],
+    "bbox": [
+      [
+        28,
+        60,
+        955,
+        227
+      ]
+    ],
+    "page_idx": 25
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1d499c0ca1281bcec022d024917ac5bad140f18fbf5fe272e3fb5cbe41be03f0/supplementary_0_Peer Review File__1d499c0ca1281bcec022d024917ac5bad140f18fbf5fe272e3fb5cbe41be03f0.mmd

@@ -0,0 +1,690 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+MYOD- SKP2 axis boosts tumorigenesis in fusion negative rhabdomyosarcoma by preventing differentiation through p57Kip2 targeting  
+
+![](images/Figure_1.jpg)
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+The authors seek to understand the molecular mechanisms underlying the biological functions of the MYOD oncoprotein to promote rhabdomyosarcoma by transcriptionally activating Skp2 E3 ligase to target p27 and p57 for ubiquitination mediated destruction. It is clearly written, and the authors have utilized various genetic and cell biological approaches to gather strong experimental evidence. However, additional in- depth investigation should be carried out to validate the detailed mechanisms, and the following concerns should be addressed.  
+
+1). Figure 1e: it will be nice for the authors to include IB MYOD.  
+2). Figure 1c, 1e, it will be nice to include more than two independent siMYOD to avoid off-target issues.  
+3). Figure 3a: it will be nice to include more than two independent siSKP2, and will be nice to include other known Skp2 substrates such as FOXO1.  
+4). Figure 3g: it will be nice for the authors to include IB p21 and IB p57.  
+5). Figure 4a: will ectopic expression of p57 phenocopy Skp2 depletion? Have the authors also examined the effects of Skp2 inhibitors?  
+6). Figure 4f: as Skp2 depletion did not affect MyoD mRNA but dramatically elevate MYOD protein levels, have the authors examined the possibility of MYOD being a ubiquitin substrate of Skp2? Does Skp2 binds MYOD and promote MYOD ubiquitination and if this process depends on pSer200- MYOD event?  
+7). Figure 6d: it will be nice for the authors to examine if Skp2 inhibitors can also reduce cellular transformation abilities of the cell lines they tested.  
+8). Figure 7a: it will be nice for the authors to include IB MYOD.  
+9). Other transcriptional factors such as E2F1 have been reported to activate Skp2, will E2F1 cooperates with MYOD to promote Skp2 transcription in rhabdomyosarcoma, or MYOD is the major driver of Skp2 overexpression in rhabdomyosarcoma?  
+
+## Reviewer #2 (Remarks to the Author):  
+
+Review to the manuscript:  
+
+"A MYOD- SKP2 axis boosts tumorigenesis in fusion negative rhabdomyosarcoma by preventing differentiation through p57Kip2 targeting" By Pomella et al.  
+
+In this elegant and thorough study Pomella et al reveal a novel regulatory circuit that is imperative for the progression and survival of Fusion- negative rhabdomyosarcoma (FN- RMS), the most common soft tissue sarcoma in pediatric patients. These cancerous cells overexpress the oncogenic E3- ubiquitin ligase, SKP2, at the highest levels among a diverse group of pediatric and adult malignancies. The accelerated induction of the gene is demonstrated to be driven by MYOD binding to the intronic enhancer of SKP2, and the expressed enzyme directly targets P27kip1 and p57kip2 for degradation leading to
+
+<--- Page Split --->
+
+progression of cell cycle and stemness maintenance. By employing several orthogonal approaches, the authors explored the reciprocal dynamics and established that depletion of SKP2 expedites accumulation of p21Cip1, p27Kip1, and p57Kip2, resulting in cell cycle arrest, concomitantly with MYOD stabilization that induces Myogenin- dependent differentiation. Interestingly, Pevonedistat, an indirect inhibitor of SKP2, was shown to lower MYOG levels required for myogenic differentiation, yet it recapitulated some of the outcomes of SKP2 depletion, and triggered tumor growth suppression and apoptotic cell death.  
+
+This study nicely exemplifies the valuable potential of basic research combined with translational studies. The presented study is innovative and significant to the field of pediatric oncology. It well communicates with the established RMS literature and logically adheres to the already established findings of previous reports. The study is robust, utilizes multiple orthogonal scientific approaches, and was performed on a large cohort of tumor cell lines yielding overall consistent results that put together into a coherent story. The work supports the authors' conclusions and rational. I did not find major experimental flaws that necessitate further experimental work. In addition, data analysis seemed to be adequate and suitable for the authors' assertions and summary.  
+
+I did however spot a few issues related to the sometimes- cumbersome writing style, and have a few suggestions that may better the figure presentation. I recommend publishing this manuscript upon fixing the minor issues I raised.  
+
+1. The following sentence needs to be altered as the discussed concept is only hypothetical (suggestive) and wasn't tested in practice, thus should be slightly tuned down. In addition, the grammar is a bit problematic:  
+
+"This suggests the presence of a core regulatory (CR) Transcription Factors (TF)s complex on the SKP2 locus in FP-RMS containing both PAX3-FOXO1 and MYOD explaining the more elevated expression of SKP2 in FP-RMS compared to FN-RMS, and was in line with data showing that PAX3-FOXO1 regulates SKP2 expression in FP-RMS".  
+
+A suggestive moderate version may be:  
+
+"This suggests the presence of a core regulatory (CR) Transcription Factors (TF)s complex on the SKP2 locus in FP-RMS containing both PAX3-FOXO1 and MYOD could propel the more elevated expression of SKP2 in FP-RMS compared to FN-RMS, in agreement with data showing that PAX3-FOXO1 regulates SKP2 expression in FP-RMS".  
+
+2. The following sentence is too long and should be separated into two, in addition, the word "on" should be altered to "of":  
+
+"Consistently, the analysis of publicly available RNA-seq data on human myoblasts induced to differentiate in vitro showed that the transcript levels of MYOD1 and SKP2 increase within the first 24 hours (24h) and then decrease in parallel, returning to their starting levels and even lower in later phases of differentiation, unlike those of MYOG, the master MYOD gene target and crucial inducer of differentiation (Fig. S2e)."  
+
+A suggestive version:  
+
+"Consistently, the analysis of publicly available RNA-seq data on human myoblasts induced to differentiate in vitro showed that the transcript levels of MYOD1 and SKP2 increase within the first 24 hours (24h) and then concomitantly decrease, returning to their starting levels
+
+<--- Page Split --->
+
+and further lower during later phases of differentiation. This, in contrast to MYOG, the chief gene target of MYOD and a crucial inducer of myogenic differentiation, whose expression levels remain elevated also by day 3 of differentiation (Fig. S2e)".  
+
+3. The following sentence also needs to be split into two:  
+
+"Moreover, as shown in Figs. S2f-h, retrovirus-mediated expression of exogenous MyoD in murine C3H/10T1/2 fibroblasts, a well described model of myogenic-like differentiation, resulted in increased SKP2 protein and mRNA levels 24h post-infection in growth/proliferation medium (GM, supplemented with \(10\%\) serum) that was maintained 24h after the shift to differentiation medium (DM, serum-free medium), returning to steady state levels after an additional 24h in DM, when the cells were fused into multi-nucleated structures that resembled muscle fibers."  
+
+A suggestive version:  
+
+"Moreover, as shown in Figs. S2f-h, retrovirus-mediated expression of exogenous MyoD in murine C3H/10T1/2 fibroblasts, a well described model of myogenic-like differentiation, resulted in increased mRNA and protein expression of SKP2 24h post-infection in growth/proliferation medium (GM, supplemented with \(10\%\) serum) that was maintained 24h upon shifting to differentiation medium (DM, serum-free medium). SKP2 expression levels revert to lower steady state levels by 48h in DM, however, when the cells were fused into multi-nucleated structures that resembled muscle fibers."  
+
+4. Figure S1d: The anti-correlative nature of CERES score (lower value associated with increased sensitivity, whereas higher values associated with decreased sensitivity) makes the figure a bit confusing. For clarity, it is recommended that the authors will add to the top of the graph the following designation:  
+
+At the top left side please add: \(\downarrow\) Increased sensitivity to SKP2  
+
+At the top right side add: Reduced sensitivity to SKP2 \(\diamond\)  
+
+5. Figure 1f & Figure 2a:  
+
+In both figures, please add to the gene's schematics an arrow designating the TSS location and direction of SKP2's transcription (similarly to the manner in which it is presented in Figure 8a).  
+
+6. The following sentence should be re-written:  
+
+"To this end, we depleted SKP2 expression for 48h in the high-risk FN-RMS cell lines RD and JR1, derived from recurrent and metastatic tumor samples respectively, both with mutated p53".  
+
+A suggestive version:  
+
+"To this end, we depleted SKP2 expression for 48h in the high-risk FN-RMS cell lines RD and JR1 (both with mutated p53), derived from recurrent and metastatic tumor samples, respectively".  
+
+7. Figure 2h(right):  
+
+The color of the graph's key for CH2O \(1\%\) is wrong. The current color is whitish, whereas the color of the respective bars is bluish. Please fix the key's color to become blue.
+
+<--- Page Split --->
+
+8. Please alter the following sentence:  
+
+"The defective ability to differentiate of FN-RMS cells highly contributes to tumorigenesis". A suggestive version:  
+
+"The defective ability of FN-RMS to differentiate cells highly contributes to tumorigenesis"  
+
+9. Please amend the following sentence:  
+
+"The percentage of MyHC positive cells increased by approximately 7-fold, 11-fold, 21-fold and 13-fold in RD, JR1, RD18 and RH36 SKP2 siRNA cells compared to scrambled siRNA cells"  
+
+A suggestive version:  
+
+"A comparison of SKP2 siRNA treated cell to their scrambled siRNA treated counterparts indicated that percentage of MyHC positive cells increased by approximately 7-fold, 11-fold, 21-fold and 13-fold in RD, JR1, RD18 and RH36, respectively".  
+
+10. Please alter the following sentences accordingly:  
+
+"Similar effects were observed in RD and JR1 cells after SKP2 silencing using two lentiviral vectors expressing individual short hairpin (sh)RNAs against SKP2 (shSKP2.1 and shSKP2.2), which also increased p21Cip1 and p27Kip1 expression, compared to non-targeting control shRNA (shSCR) (Figs. 4f,g). The increase of MyHC positive cells was around 12-fold and 17-fold for shSKP2.1 and 12-fold and 14-fold for shSKP2.2 in RD and JR1 cells vs shSCR cells (Fig. 4h)".  
+
+A suggestive version:  
+
+"Similar outcomes were observed in RD and JR1 cells by employing an orthogonal method for silencing SKP2 using two lentiviral vectors expressing individual short hairpin (sh)RNAs against SKP2 (shSKP2.1 and shSKP2.2). Here as well, SKP2 shutdown promoted increased expression of p21Cip1 and p27Kip1, compared to non-targeting control shRNA (shSCR) (Figs. 4f,g), and expanded the ratio of MyHC positive cells in both, RD and JR1 cells by at least 12-fold for both shRNAs (Fig. 4h)".  
+
+11. Figs. S5g,h. Please refer to text:  
+
+A significant enrichment in myogenesis and muscle contraction pathways was noticed among the [HOW MANY UP- REGULATED GENES?] genes up-regulated 48h after SKP2 siRNA transduction, with MYOG showing the highest induction (Figs. S5g,h).  
+
+Please designate in text how many up-regulated genes ( \(>1.5 \log 2\) fold change) were found in total and provide the entire gene list of Up-regulated gene as a supplementary table.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+In this study, the authors observed SKP2 is overexpressed in RMS at the highest levels among several cancers and hypothesized its expression is maintained by MYOD1. The authors showed SKP2 directly targets p27Kip1 and p57Kip2 promoting their degradation in RMS cells and SKP2 knockdown causes cell cycle arrest by enhancing p27Kip1 and promotes differentiation by increasing p57Kip2, which in turn stabilizes MYOD resulting in muscle differentiation and cell fusion. The authors further suggested that investigational NEDDylation inhibitor MLN4924 hampers SKP2 functions restraining fusion- negative RMS cell survival and tumor growth. This study suggested a MYOD- SKP2 axis crucial for the
+
+<--- Page Split --->
+
+crosstalk between transcriptional and post- translational mechanisms that contribute to RMS tumorigenesis and broaden the understanding of MYOD function. However, there are major issues needed to be addressed in the present manuscript. Especially, this study is designed based on the hypothesis that MYOD1 is an oncogene in RMS, which is unacceptable without any reasonable data.  
+
+(Major points)  
+
+1. The authors suggest the oncogenic function of MYOD1 to transcriptionally regulate the expression of SKP2. It is known that MYOD1 L122R mutation blocks wild-type MYOD1 function and bind to MYC consensus sequences acting as an oncogene to inhibit differentiation and promote proliferation. However, wild-type MYOD1 involvement in tumorigenesis is not clear. This hypothesis should be carefully confirmed. Does knock down of MYOD1 by siRNA inhibit the tumor growth and increased myogenic differentiation? Does MYOD1 overexpression, vice versa, increase the tumor growth?  
+
+2. The authors show ChIP-seq data to suggest MYOD1 involvement in SKP2 transcription. However, it doesn't reveal how strongly MYOD1 regulates SKP2. Although si MYOD1 decrease SKP2 expression, it may be caused indirectly as a result of feedback to the dedifferentiation caused by MYOD1 depletion. Reporter gene assay may be required to convince the existence of MYOD1-SKP2 axis.  
+
+3. It seems that SKP2 is a cell cycle regulator in general irrespective of MYOD1. Therefore, inhibiting SKP2 cause growth arrest not only sarcoma cell but also normal cells suggesting that SKP2 is not a reasonable target gene. It is needed to demonstrate the growth arrest by SKP2 knockdown is correlate with the MYOD1 expression.  
+
+4. The mode of action of MLN4924 is not relevant to MYOD1 function. Is MYOD1 expression and the drug sensitivity correlate?  
+
+## Reviewer #4 (Remarks to the Author):  
+
+Review of Pomella et al.  
+
+The authors here have identified SKP2 as a critical driver of tumorigenesis in fusion negative rhabdomyosarcoma (FN- RMS) and acting downstream of MYOD. The authors have utilized publicy available genome- wide datasets both from their previous work and from others to establish the transcriptional regulation of SKP2 through MYOD. The study design and experiments are good and done in a systematic manner to show the MYOD- SKP2 axis that contributes to the tumorigenic phenotype in FN- RMS. Overall, the work performed here is satisfactory and novel in dissecting the MYOD- SKP2 signaling axis in FN- RMS. I have few concerns that should be addressed to make the findings more robust and clearer from the mechanistic point of view and outreach of this research.  
+
+The major points of concern are:  
+
+1. The authors mention that SKP2 directly interacts with p27Kip1 and targets it for proteasomal degradation, and no interaction was detected for p21Cip1. Although both p21Cip1 and p27Kip1 protein levels increased post-treatment with proteasome inhibitor MG132. How do the authors explain this discrepancy?  
+2. SKP2 depletion in FN-RMS cells resulted in increased MYOD and and CDK1 p57Kip2.
+
+<--- Page Split --->
+
+What is the consequence of MYOD binding to chromatin at the SKP2 regulatory elements such as the intronic enhancer? Can the authors perform ChIP- seq or ChIP- qPCR to show the effect of MYOD binding after SKP2 depletion? Although ChIP- qPCR could be more direct for that particular SKP2 regulatory region but the genome- wide ChIP- seq after SKP2 silencing could also pinpoint the target genes for reduced stemness and tumorigenicity in RD and JR1 cells.  
+
+3. The authors used \(\beta\) -gal staining assay to show induction of senescence upon SKP2 depletion. Could the authors validate the same on the molecular levels showing the levels of any senescent marker genes for example SASP protein either by quantitative real time PCR from the available cDNA or by western blot.  
+
+4. The authors have shown effect on cell proliferation upon SKP2 depletion in cell lines and the same effect is also seen in the reduced tumor volumes after SKP2 is silenced by shRNA or using the NAE inhibitor LN4924. Since the IHC sections have been made and stained for the relevant targets but it was surprising that the authors haven't stained for the proliferation marker Ki67 in these sections. The Ki67 staining for the IHC sections shown if Fig 6k and 7c would make it clear at the molecular level.
+
+<--- Page Split --->
+
+## Response to Reviewers  
+
+Reviewer #1 (Remarks to the Author):  
+
+The authors seek to understand the molecular mechanisms underlying the biological functions of the MYOD oncoprotein to promote rhabdomyosarcoma by transcriptionally activating Skp2 E3 ligase to target p27 and p57 for ubiquitination mediated destruction. It is clearly written, and the authors have utilized various genetic and cell biological approaches to gather strong experimental evidence. However, additional in- depth investigation should be carried out to validate the detailed mechanisms, and the following concerns should be addressed.  
+
+We thank the Reviewer for the positive evaluation of our work and suggestions for improvement.  
+
+1). Figure 1e: it will be nice for the authors to include IB MYOD.  
+
+We have performed immunoblot for MYOD on the panel of RMS cell lines as requested. However, since in Figure 1, and in the corresponding paragraph, MYOD has not yet been mentioned, we included this result as Fig. S2a.  
+
+2). Figure 1c, 1e, it will be nice to include more than two independent siMYOD to avoid off- target issues.  
+
+We have performed MYOD silencing using 2 additional independent MYOD siRNAs in all the four RMS cell lines. The results for RD and RH4 cell lines are shown in Fig. S2b,c while those for JR1 and RH30 are included in the Ancillary Fig. 1 for the Reviewers. MYOD was down- regulated together with SKP2 at the protein and transcript levels starting from 24h post- transfection, supporting a direct regulation by MYOD. In the original version we also have used a CRISPR/Cas9 approach to knockout MYOD and showed the same effects on SKP2 expression (now reported in Fig. S2d,e). Altogether, these results suggest a lack of off- target effects.  
+
+3). Figure 3a: it will be nice to include more than two independent siSKP2 and will be nice to include other known Skp2 substrates such as FOXO1.  
+
+We completely agree with the Reviewer that it is mandatory to demonstrate that our approach is not the result of off- target effects. For this reason, in our initial submission, we used a validated siRNA sequence against SKP2 (by Sigma- Aldrich) in Fig. 3a and we also validated the results using a siRNA pool containing 4 different individual SKP2 siRNA sequences (by Dharmacon) (Original Fig. S3a and now Fig. S4a).  
+
+In addition, we infected our cells with two lentiviral vectors each expressing an individual SKP2 shRNA sequence (by Sigma- Aldrich) different from all the others used as siRNAs (Fig. 4f). In all these cases, we detected SKP2 down- regulation associated with both p21 and p27 protein levels increase. Therefore, we are confident our results are not due to off- target effects. We have revised the main text in the Results section to better clarify this point.  
+
+As suggested by the Reviewer, we did immunoblot for FOXO1 and added it to Fig. 3a. We found increased levels of the protein after SKP2 silencing compared to scrambled siRNA cells validating the SKP2 down- regulation effects.  
+
+4). Figure 3g: it will be nice for the authors to include IB p21 and IB p57.  
+
+We have performed immunoblot for p21 and included the results in Fig. 3g. The results show that p27 knockdown reduced p21 increase due to SKP2 silencing. In addition, it also slightly decreased p21 levels in basal conditions compared to scrambled siRNA. These results suggest that p27 could indirectly regulate p21 expression in FN- RMS with a mechanism that deserves further investigations.
+
+<--- Page Split --->
+
+Moreover, we also investigated the expression levels of p57 by immunoblot. We have included the results in the Ancillary Fig. 2 for the Reviewers since, at this point of the manuscript, we have not yet presented data on p57 and it would be difficult to explain why this detection was made. In fact, the study of p57 expression was done related to its role in cells differentiation starting from Fig. 4. Moreover, no significant effects of p27 depletion can be seen on p57 levels.  
+
+## 5). Figure 4a: will ectopic expression of p57 phenocopy Skp2 depletion?  
+
+To address the point referring to Fig. 4, we have overexpressed an exogenous p57/CDKN1C gene in RD and JR1 FN- RMS cell lines and evaluated whether higher p57 levels would be sufficient to induce differentiation of FN- RMS cell lines mirroring SKP2 down- regulation. The results have been reported in Fig. S6. Notably, the protein and mRNA levels of MYOD and MYOG increased in p57- vector compared to empty vector cells. In agreement, the p57- overexpressing cells showed a tendency to differentiate compared to empty vector cells, albeit more mildly than observed after SKP2 depletion. These results suggest that p57 positively modulates differentiation in FN- RMS cells and this function is even more evident in a molecular setting where the brakes for differentiation have been relieved, such as when SKP2 is depleted. These findings have been reported and discussed in the revised manuscript.  
+
+Have the authors also examined the effects of Skp2 inhibitors?  
+
+As an indirect SKP2 inhibitor with translational potentiality we used MLN4924/Pevonedistat, which is being investigated in several clinical trials of different tumor types (see Results section and Figs. 7 and S10). However, we have performed new experiments using the in vitro validated SKP2 inhibitor SMIP004 (Ref. 28 Revised Manuscript).  
+
+SMIP004 treatment decreased SKP2 protein levels and increased p21 and p27 protein and mRNA levels in RD and JR1 cells 48h post- treatment (Fig. 3i- n). This was associated with slowdown of cell proliferation and G1 cell cycle arrest compared to vehicle. Moreover, it promoted MYOD, MYOG and p57 protein and mRNA levels enhancement and the expression of the late differentiation marker MyHC (Fig. 4i- l).  
+
+Altogether, these results show that pharmacological inhibition of SKP2 with SMIP004 mirrors the anti- proliferative and pro- differentiation effects of genetic depletion.  
+
+In parallel, since in the original submission we investigated the in vitro effects of MLN4924 on cell proliferation, we performed new experiments to assess cell cycle distribution of the RD and JR1 cell lines also under MLN4924. The results in Ancillary Fig. 3 for Reviewers show that, conversely to SKP2 silencing and SMIP004, treatment with MLN4924 resulted in the accumulation of cells in G2 phase as compared to vehicle. This is in agreement with the evidence that the drug leads to down- regulation of MYOG and apoptosis rather than differentiation, as reported and discussed in the original version of the manuscript.  
+
+6). Figure 4f: as Skp2 depletion did not affect MyoD mRNA but dramatically elevate MYOD protein levels, have the authors examined the possibility of MYOD being a ubiquitin substrate of Skp2? Does Skp2 binds MYOD and promote MYOD ubiquitination and if this process depends on pSer200- MYOD event?  
+
+As requested, we have performed Co- IP for SKP2 and MYOD in both RD and JR1 FN- RMS cells and found no interaction between the two proteins (Ancillary Fig. 4 for Reviewers). These experiments demonstrate that MYOD is not a SKP2 substrate in FN- RMS cells. This is in line with data showing that MyoD does not interact with Skp2 in murine myoblasts (Reference for Rebuttal 1). Of note, in the original and revised manuscript we showed that, in addition to the increase in MYOD protein levels, SKP2 depletion resulted in the rise of MYOD1 mRNA levels at least in RD and JR1 cells (now Figs. 4a,b).
+
+<--- Page Split --->
+
+Thus, it is conceivable that the enhancement of MYOD levels after SKP2 depletion could be the result of, at least, two concurrent phenomena: a post- transcriptional protein stabilization by p57 and an increment of transcriptional auto- induction by MYOD its- self (Ref. n. 45). However, since Myod is regulated by the proteasome in myoblasts, we cannot exclude that SKP2 knockdown could indirectly or directly modulate other E3 ubiquitin ligases that, in turn, participate to the post- transcriptional regulation of MYOD levels 1.  
+
+7). Figure 6d: it will be nice for the authors to examine if Skp2 inhibitors can also reduce cellular transformation abilities of the cell lines they tested.  
+
+To evaluate this aspect, we have performed new experiments treating RD and JR1 cells with the two SKP2i, SMIP004 and MLN4924. The results are reported in Fig. S9a- d and Fig. S10d- g, respectively. The ability to grow as colonies in an anchorage- independent assay and to form rhabdospheres when cultured in a serum- free stem cell medium were both compromised by the two agents compared to vehicle treatment, mirroring the effects of SKP2 genetic depletion.  
+
+8). Figure 7a: it will be nice for the authors to include IB MYOD.  
+
+The protein levels of MYOD after treatment with MLN4924 for both RD and JR1 FN- RMS cell lines were already reported in our original version in Fig. S7b (now Fig. 7c).  
+
+9). Other transcriptional factors such as E2F1 have been reported to activate Skp2, will E2F1 cooperates with MYOD to promote Skp2 transcription in rhabdomyosarcoma, or MYOD is the major driver of Skp2 overexpression in rhabdomyosarcoma?  
+
+To clarify this point, we have performed E2F1 silencing in RD and JR1 FN- RMS as well as in RH4 and RH30 P3F- RMS cell lines using two individual validated siRNAs (Ancillary Fig. 5 for Reviewers). The two independent experiments show no transcript and/or protein levels modulation of SKP2 after E2F1 silencing in our tumor cell context. However, we agree that SKP2 expression has been shown to be modulated by several TFs in different cell contexts. In addition to E2F1, other TFs are GA- binding protein (GABP) and NUCKS1, which induce SKP2 expression 2,3, and STAT1 that, conversely, represses the gene 4.  
+
+Thus, we cannot rule out that other TFs relevant to the FN- RMS context could participate with MYOD in the transcriptional regulation of SKP2. However, being MYOD a lineage- specific TF highly expressed in FN- RMS, also based on our data it could be considered as one of the major drivers.  
+
+Reviewer #2 (Remarks to the Author):  
+
+Review to the manuscript:  
+
+"A MYOD- SKP2 axis boosts tumorigenesis in fusion negative rhabdomyosarcoma by preventing differentiation through p57Kip2 targeting" By Pomella et al.  
+
+In this elegant and thorough study Pomella et al reveal a novel regulatory circuit that is imperative for the progression and survival of Fusion- negative rhabdomyosarcoma (FN- RMS), the most common soft tissue sarcoma in pediatric patients. These cancerous cells overexpress the oncogenic E3- ubiquitin ligase, SKP2, at the highest levels among a diverse group of pediatric and adult malignancies. The accelerated induction of the gene is demonstrated to be driven by MYOD binding to the intronic enhancer of SKP2, and the expressed enzyme directly targets P27kip1 and p57kip2 for degradation leading to progression of cell cycle and stemness maintenance. By employing several orthogonal approaches, the authors explored the reciprocal dynamics and
+
+<--- Page Split --->
+
+established that depletion of SKP2 expedites accumulation of p21Cip1, p27Kip1, and p57Kip2, resulting in cell cycle arrest, concomitantly with MYOD stabilization that induces Myogenin- dependent differentiation. Interestingly, Pevonedistat, an indirect inhibitor of SKP2, was shown to lower MYOG levels required for myogenic differentiation, yet it recapitulated some of the outcomes of SKP2 depletion, and triggered tumor growth suppression and apoptotic cell death.  
+
+This study nicely exemplifies the valuable potential of basic research combined with translational studies. The presented study is innovative and significant to the field of pediatric oncology. It well communicates with the established RMS literature and logically adheres to the already established findings of previous reports. The study is robust, utilizes multiple orthogonal scientific approaches, and was performed on a large cohort of tumor cell lines yielding overall consistent results that put together into a coherent story. The work supports the authors' conclusions and rational. I did not find major experimental flaws that necessitate further experimental work. In addition, data analysis seemed to be adequate and suitable for the authors' assertions and summary.  
+
+I did however spot a few issues related to the sometimes- cumbersome writing style, and have a few suggestions that may better the figure presentation.  
+
+I recommend publishing this manuscript upon fixing the minor issues I raised.  
+
+We thank the Reviewer for the positive evaluation of the manuscript and for the suggestions to improve the writing which have been very valuable.  
+
+1. The following sentence needs to be altered as the discussed concept is only hypothetical (suggestive) and wasn't tested in practice, thus should be slightly tuned down. In addition, the grammar is a bit problematic:  
+
+"This suggests the presence of a core regulatory (CR) Transcription Factors (TF)s complex on the SKP2 locus in FP-RMS containing both PAX3-FOXO1 and MYOD explaining the more elevated expression of SKP2 in FP-RMS compared to FN-RMS, and was in line with data showing that PAX3-FOXO1 regulates SKP2 expression in FP-RMS".  
+
+A suggestive moderate version may be:  
+
+"This suggests the presence of a core regulatory (CR) Transcription Factors (TF)s complex on the SKP2 locus in FP-RMS containing both PAX3-FOXO1 and MYOD could propel the more elevated expression of SKP2 in FP-RMS compared to FN-RMS, in agreement with data showing that PAX3- FOXO1 regulates SKP2 expression in FP-RMS".  
+
+We have revised the manuscript as suggested.  
+
+2. The following sentence is too long and should be separated into two, in addition, the word "on" should be altered to "of":  
+
+"Consistently, the analysis of publicly available RNA-seq data on human myoblasts induced to differentiate in vitro showed that the transcript levels of MYOD1 and SKP2 increase within the first 24 hours (24h) and then decrease in parallel, returning to their starting levels and even lower in later phases of differentiation, unlike those of MYOG, the master MYOD gene target and crucial inducer of differentiation (Fig. S2e)."  
+
+A suggestive version:  
+
+"Consistently, the analysis of publicly available RNA-seq data on human myoblasts induced to differentiate in vitro showed that the transcript levels of MYOD1 and SKP2 increase within the first 24 hours (24h) and then concomitantly decrease, returning to their starting levels and further lower during later phases of differentiation. This, in contrast to MYOG, the chief gene target of MYOD and a crucial inducer of myogenic differentiation, whose expression levels remain elevated also by day 3 of differentiation (Fig. S2e)".
+
+<--- Page Split --->
+
+We have revised the manuscript as suggested.  
+
+3. The following sentence also needs to be split into two:  
+
+3. The following sentence also needs to be split into two: "Moreover, as shown in Figs. S2f-h, retrovirus-mediated expression of exogenous MyoD in murine C3H/10T1/2 fibroblasts, a well described model of myogenic-like differentiation, resulted in increased SKP2 protein and mRNA levels 24h post-infection in growth/proliferation medium (GM, supplemented with 10% serum) that was maintained 24h after the shift to differentiation medium (DM, serum-free medium), returning to steady state levels after an additional 24h in DM, when the cells were fused into multi-nucleated structures that resembled muscle fibers."  
+
+A suggestive version:  
+
+A suggestive version:"Moreover, as shown in Figs. S2f-h, retrovirus-mediated expression of exogenous MyoD in murine C3H/10T1/2 fibroblasts, a well described model of myogenic-like differentiation, resulted in increased mRNA and protein expression of SKP2 24h post-infection in growth/proliferation medium (GM, supplemented with 10% serum) that was maintained 24h upon shifting to differentiation medium (DM, serum-free medium). SKP2 expression levels revert to lower steady state levels by 48h in DM, however, when the cells were fused into multi-nucleated structures that resembled muscle fibers."  
+
+We have revised the manuscript as suggested.  
+
+4. Figure S1d: The anti-correlative nature of CERES score (lower value associated with increased sensitivity, whereas higher values associated with decreased sensitivity) makes the figure a bit confusing. For clarity, it is recommended that the authors will add to the top of the graph the following designation:  
+
+At the top left side please add: \(\downarrow\) Increased sensitivity to SKP2  
+
+At the top right side add: Reduced sensitivity to SKP2  
+
+We thank the Reviewer for his/her suggestion. We have modified the Figure accordingly.  
+
+5. Figure 1f & Figure 2a:  
+
+In both figures, please add to the gene's schematics an arrow designating the TSS location and direction of SKP2's transcription (similarly to the manner in which it is presented in Figure 8a).  
+
+The Figures have been modified as suggested.  
+
+6. The following sentence should be re-written:  
+
+"To this end, we depleted SKP2 expression for 48h in the high-risk FN-RMS cell lines RD and JR1, derived from recurrent and metastatic tumor samples respectively, both with mutated p53".  
+
+A suggestive version:  
+
+"A suggestive version:"To this end, we depleted SKP2 expression for 48h in the high-risk FN-RMS cell lines RD and JR1 (both with mutated p53), derived from recurrent and metastatic tumor samples, respectively".  
+
+We have revised the manuscript as suggested.  
+
+7. Figure 2h(right):
+
+<--- Page Split --->
+
+The color of the graph's key for CH2O \(1\%\) is wrong. The current color is whitish, whereas the color of the respective bars is bluish. Please fix the key's color to become blue.  
+
+We thank the Reviewer for noticing this error that has been, now, fixed.  
+
+8. Please alter the following sentence: "The defective ability to differentiate of FN-RMS cells highly contributes to tumorigenesis".  
+
+A suggestive version: "The defective ability of FN-RMS to differentiate cells highly contributes to tumorigenesis"  
+
+We have revised the manuscript as suggested.  
+
+9. Please amend the following sentence:  
+
+"The percentage of MyHC positive cells increased by approximately 7-fold, 11-fold, 21-fold and 13-fold in RD, JR1, RD18 and RH36 SKP2 siRNA cells compared to scrambled siRNA cells"  
+
+A suggestive version:  
+
+"A comparison of SKP2 siRNA treated cell to their scrambled siRNA treated counterparts indicated that percentage of MyHC positive cells increased by approximately 7-fold, 11-fold, 21-fold and 13-fold in RD, JR1, RD18 and RH36, respectively".  
+
+We have revised the manuscript as suggested.  
+
+10. Please alter the following sentences accordingly:  
+
+"Similar effects were observed in RD and JR1 cells after SKP2 silencing using two lentiviral vectors expressing individual short hairpin (sh)RNAs against SKP2 (shSKP2.1 and shSKP2.2), which also increased p21Cip1 and p27Kip1 expression, compared to non-targeting control shRNA (shSCR) (Figs. 4f,g). The increase of MyHC positive cells was around 12-fold and 17-fold for shSKP2.1 and 12-fold and 14-fold for shSKP2.2 in RD and JR1 cells vs shSCR cells (Fig. 4h)".  
+
+A suggestive version:  
+
+"Similar outcomes were observed in RD and JR1 cells by employing an orthogonal method for silencing SKP2 using two lentiviral vectors expressing individual short hairpin (sh)RNAs against SKP2 (shSKP2.1 and shSKP2.2). Here as well, SKP2 shutdown promoted increased expression of p21Cip1 and p27Kip1, compared to non-targeting control shRNA (shSCR) (Figs. 4f,g), and expanded the ratio of MyHC positive cells in both, RD and JR1 cells by at least 12-fold for both shRNAs (Fig. 4h)".  
+
+We have revised the manuscript as suggested.  
+
+11. Figs. S5g,h. Please refer to text:  
+
+A significant enrichment in myogenesis and muscle contraction pathways was noticed among the [HOW MANY UP- REGULATED GENES?] genes up-regulated 48h after SKP2 siRNA transduction, with MYOG showing the highest induction (Figs. S5g,h).  
+
+Please designate in text how many up-regulated genes ( \(>1.5 \log 2\) fold change) were found in total and provide the entire gene list of Up-regulated gene as a supplementary table.  
+
+In the original submission we presented data of RNA- seq after SKP2 siRNA in RD cells. In the revised version we have done a new RNA- seq assay after SKP2 silencing using shSKP2.2 and
+
+<--- Page Split --->
+
+shSCR in both FN- RMS cell lines RD and JR1 in order to identify commonly modulated genes. Based on these new experiments, we have revised the paragraph enclosing the results in Fig. S8 and provided a new gene list of total up- regulated \((>1.3 \log 2\) fold change) and down- regulated \((< 0.7 \log 2\) fold change) genes in Supplementary Table 1. Moreover, we have also provided the lists of differentially modulated genes (same criteria as above) for myogenesis, stemness and senescence (Fig. S8) in Supplementary Table 2.  
+
+Reviewer #3 (Remarks to the Author):  
+
+In this study, the authors observed SKP2 is overexpressed in RMS at the highest levels among several cancers and hypothesized its expression is maintained by MYOD1. The authors showed SKP2 directly targets p27Kip1 and p57Kip2 promoting their degradation in RMS cells and SKP2 knockdown causes cell cycle arrest by enhancing p27Kip1 and promotes differentiation by increasing p57Kip2, which in turn stabilizes MYOD resulting in muscle differentiation and cell fusion. The authors further suggested that investigational NEDDylation inhibitor MLN4924 hampers SKP2 functions restraining fusion- negative RMS cell survival and tumor growth. This study suggested a MYOD- SKP2 axis crucial for the crosstalk between transcriptional and posttranslational mechanisms that contribute to RMS tumorigenesis and broaden the understanding of MYOD function. However, there are major issues needed to be addressed in the present manuscript. Especially, this study is designed based on the hypothesis that MYOD1 is an oncogene in RMS, which is unacceptable without any reasonable data.  
+
+We thank the Reviewer very much for allowing us to clarify this crucial point. We absolutely agree with the Reviewer that MYOD1 cannot be considered a "primary" oncogene since it is not a "driver" of RMS.  
+
+However, MYOD1 is among the most RMS- selective gene dependency, as demonstrated analyzing a genome- wide CRISPR/Cas9 somatic knockout (KO) screen on the survival of tumor cell lines: MYOD1 is crucial in supporting RMS cell survival (Achilles project, https://depmap.org/portal/achilles) (Ancillary Fig. 6 for Reviewers).  
+
+This data has also been reported in a recent study of the group of Kimberly Stegmaier investigating "genetic vulnerabilities unique of a specific cancer type" in pediatric cancers (5 Fig. 1a, right panel). Pediatric cancers have a low mutational burden and it is now recognized that, in addition to the driver oncogenes on which the tumor cells depend (oncogene addiction), tumor survival mechanisms are often pre- programmed in the cells of origin. Specifically, these deregulated epigenetic mechanisms hijack wild- type lineage- specific Transcription Factors (TFs) into tumor- specific core regulatory circuits (CRCs) needed for tumorigenesis, resulting in a "lineage dependency" (reviewed in 6,7).  
+
+Therefore, we apologize for having utilized the inappropriate term "oncogenic" in two sentences about MYOD. We used the term "pro- tumorigenic" regarding MYOD to indicate a TF that reinforces an oncogenic driver program clarifying this point throughout the revised manuscript.  
+
+In addition, we would like to point out that our work was specifically focused on the study of the role of SKP2 in FN- RMS, which remained unknown until now, by dissecting the molecular pathways implicated.  
+
+In this scenario, we unexpectedly discovered that MYOD was involved in SKP2 functions as both an upstream and downstream molecular player. Our data demonstrating that MYOD regulates the expression of an oncogenic E3 ubiquitin ligase like SKP2 add a piece of evidence on the intricate molecular networks sustaining RMS tumorigenesis.  
+
+(Major points)
+
+<--- Page Split --->
+
+1. The authors suggest the oncogenic function of MYOD1 to transcriptionally regulate the expression of SKP2. It is known that MYOD1 L122R mutation blocks wild-type MYOD1 function and bind to MYC consensus sequences acting as an oncogene to inhibit differentiation and promote proliferation. However, wild-type MYOD1 involvement in tumorigenesis is not clear. This hypothesis should be carefully confirmed. Does knock down of MYOD1 by siRNA inhibit the tumor growth and increased myogenic differentiation?  
+
+This aspect has already been investigated by David Langenau's group showing that MYOD1 is needed for "sustained tumor growth" of RD and SMS- CTR FN- RMS cells in vitro and in vivo since its depletion results in tumor cell death (Ref. 7 Revised Manuscript).  
+
+However, as requested by the Reviewer we have performed new functional experiments by silencing MYOD1 in RD and JR1 FN- RMS cell lines. In line with the dependency of RMS cells on this TF (see the response to comment above), MYOD- depleted cells showed reduced proliferation and accumulated in the G1 cell cycle phase reducing S phase compared to scrambled siRNA cells (Ancillary Fig. 7a,b for Reviewers).  
+
+These results confirm those obtained by the group of Langenau about MYOD depletion (Ref. 7 Revised Manuscript) as well as the data on MYOD knockout in RMS cell lines of the DepMap portal (as reported in the previous comment to the same Reviewer).  
+
+Both mRNA and protein levels of SKP2 decreased in MYOD siRNA cells, as already shown in Figs. 2 and S2 and in Ancillary Fig. 1 for Reviewers. P21 and p27 protein levels increased in MYOD1- silenced cells independently of their mRNA levels, which is in agreement with the downregulation of SKP2 (Ancillary Fig. 7c,d for Reviewers).  
+
+However, MYOD- silenced FN- RMS cells displayed downregulation of the MYOD chief direct target MYOG, one of the master muscle regulatory factors and, thus, are unable to differentiate (Ancillary Fig. 7e for Reviewers) (Ref. 7 Revised Manuscript; reviewed in 8).  
+
+Overall, our results confirm data from other labs (see above) showing that MYOD is needed for continued FN- RMS cells growth/survival, which is also in agreement with the function of SKP2.  
+
+Does MYOD1 overexpression, vice versa, increase the tumor growth?  
+
+To answer to this question, we performed MYOD overexpression in both RD and JR1 FN- RMS cell lines using an inducible viral vector expressing an exogenous MYOD1 (Ancillary Fig. 8 for Reviewers). MYOD levels were markedly increased after induction with doxycycline compared to control cells. In parallel, SKP2 was modestly augmented maybe due to the already high basal levels. No increase in cell proliferation was evident in MYOD- overexpressing cells. This result seems to be in line with a pro- survival rather than pro- proliferative role of MYOD in FN- RMS and/or could also be related to the high basal expression of the TF suggesting that a further increase is not sufficient to enhance cell proliferation.  
+
+2. The authors show ChIP-seq data to suggest MYOD1 involvement in SKP2 transcription. However, it doesn't reveal how strongly MYOD1 regulates SKP2. Although si MYOD1 decrease SKP2 expression, it may be caused indirectly as a result of feedback to the dedifferentiation caused by MYOD1 depletion. Reporter gene assay may be required to convince the existence of MYOD1-SKP2 axis.  
+
+We thank the Reviewer for his/her suggestion that allow us to clarify our data and strengthen the manuscript. To this end, a reporter gene assay has been performed in HEK 293T cells, not expressing MYOD1, co- transfecting either (i) a plasmid vector expressing human MYOD or an empty vector and (ii) a pGL3- promoter vector expressing a luciferase gene under the control of the identified MYOD- bound SKP2 intronic enhancer region or an empty pGL3- promoter vector (Fig. S2g). Results show luciferase induction only in cells co- transfected with MYOD- expressing vector
+
+<--- Page Split --->
+
+and pGL3 plasmid harboring the MYOD- bound SKP2 intronic enhancer suggesting MYOD- transactivation activity.  
+
+Moreover, we applied an additional approach interrogating chromatin looping mechanisms in the RD FN- RMS cell line using HiChIP technique in order to identify chromatin contacts around the SKP2 promoter at 1kb resolution. RH4 FP- RMS cell line was de novo assayed as well since the sequencing used for the same cell line in the original version of the manuscript was at 5 kb resolution. As shown in Fig. S2f, the data analysis confirmed an interaction between the identified MYOD- bound intronic region and the SKP2 promoter in both cell lines.  
+
+These data, together with those obtained with the chromosome conformation capture assay (3C) in the original version of the manuscript, which showed that the MYOD- bound intronic enhancer was able to bind SKP2 promoter (Fig. 2g,h), support a direct transcriptional regulation of SKP2 expression by MYOD.  
+
+3. It seems that SKP2 is a cell cycle regulator in general irrespective of MYOD1. Therefore, inhibiting SKP2 cause growth arrest not only sarcoma cell but also normal cells suggesting that SKP2 is not a reasonable target gene. It is needed to demonstrate the growth arrest by SKP2 knockdown is correlate with the MYOD1 expression.  
+
+To answer to this concern, we firstly investigated whether FN- RMS cells had increased vulnerability to SKP2 inhibition compared to a panel of normal cells. To this end, we performed new experiments: human normal myoblasts, which are the normal lineage counterpart of RMS cells and express MYOD, lung and dermal fibroblasts, which do not express MYOD, were treated in dose response experiments with the two SKP2i used in the study, SMIP004 and MLN4924, and their survival ability assessed.  
+
+Results reported in Figs. S4k and S10b show that all normal cells are considerably less sensitive to high doses of both treatments compared to FN- RMS cells suggesting a tumor- related vulnerability (see also the revised Results section).  
+
+In addition, the response to SKP2 inhibition clearly appeared not correlated to MYOD since myoblasts and fibroblasts showed a similar response.  
+
+Our data are concordant with those of the group of Singer showing that SKP2 depletion has negligible effects on normal pre- adipocyte stem cells vs myxofibrosarcoma cells (Ref. 17 Revised Manuscript). Calandrini et al. also demonstrated selective sensitivity of Malignant Rhabdoid Tumors organoids to MLN4924 with respect to healthy kidney and liver tissue organoids (Ref. 64 Revised Manuscript). They, also, reported that only the healthy small intestine organoids were affected by MLN4924 possibly due to high rate of proliferation in vitro but they also highlighted that no side effects related to the intestines have been identified in the first clinical studies on MLN4924.  
+
+However, we agree with the Reviewer that, under a translational point of view, it is very important to identify those SKP2 targeting compounds that can be more effective in that specific tumor context and to consider the pathological mechanisms of the targeted tumor type (Refs. 53, 67 Revised Manuscript; and reviewed in Ref. 68 Revised Manuscript). In this view, preclinical studies on selected tumors could be of help.  
+
+We, therefore, have added a comment on these aspects in the revised Discussion.  
+
+In addition, to answer to the reviewer's comment more exhaustively, we decided to elucidate the interconnection between MYOD and SKP2 in FN- RMS by silencing the two factors individually or together.  
+
+The Ancillary Fig. 9a,b for Reviewers, which partly refers to data on MYOD- silenced cells (see the response to point 1 of the same Reviewer and Ancillary Fig. 7 for Reviewers), shows that silencing
+
+<--- Page Split --->
+
+of MYOD alone vs siSCR reduced cell numbers and promoted G1 phase increment and S phase diminution more markedly than SKP2 siRNA alone in RD and JR1 cells.  
+
+This finding is in line with the crucial pro- survival role of MYOD in FN- RMS cells (see also the response to point 1 raised by the same Reviewer). SKP2 transcriptional and protein downregulation following MYOD silencing was confirmed in Ancillary Fig. 9c,d for Reviewers. MYOD and SKP2 co- depleted cells showed significant decrease of cell proliferation paralleled by enhancement of G1 phase and reduction of S phase cells accumulation compared to SKP2 silenced cells, while the effects were not significant vs MYOD- siRNA alone. These findings clearly demonstrate that the effects of SKP2 on cell cycle are independent on MYOD since MYOD knockdown in a SKP2- depleted context does not reduce, but in contrast enhances, cell cycle arrest.  
+
+MYOD siRNA alone lowered MYOG protein and transcript levels, as already shown, and overcomes MYOG induction when transfected in combination with SKP2 siRNA Ancillary Fig. 9c,d for Reviewers. Consequently, reducing MYOD levels not only does not induce differentiation but also counteracts SKP2 depletion- dependent cell differentiation (Ancillary Fig. 9e,d for Reviewers). These findings clearly demonstrate that MYOD appears to be necessary for the pro- differentiation effects of SKP2 inhibition.  
+
+This is in line with the evidence that, although p57 levels were higher in MYOD/SKP2 co- depleted cells, they were not sufficient to support differentiation in the absence (or reduction) of MYOD.  
+
+Moreover, being the percentage of MYOD depleted cells arrested in the G1 phase of the cell cycle significantly higher compared to that of SKP2 siRNA cells, our results also suggest that MYOD regulates FN- RMS cell survival also modulating other pathways in addition to SKP2, as we have recently demonstrated (Ref. 6 Revised Manuscript).  
+
+In conclusion, our new results validate SKP2 as a tumor survival vulnerability in FN- RMS.  
+
+4. The mode of action of MLN4924 is not relevant to MYOD1 function. Is MYOD1 expression and the drug sensitivity correlate?  
+
+Although this point has been clarified in the above response to point 3, we performed Pearson correlation analysis taking advantage of the DepMap portal (https://depmap.org/portal/) using the 23Q2 public dataset for gene expression and Drug Sensitivity Area Under the Curve (AUC) dataset for drug sensitivity data, which includes 721 cancer lines (CTD2, Broad Institute) (Ancillary Figures 10a,b for Reviewers). The analysis revealed no significant correlation between MYOD1 expression and the response to MLN4924 while a significant correlation between SKP2 expression and the drug AUC was observed among all the cancer cell lines analyzed.  
+
+Reviewer #4 (Remarks to the Author):  
+
+Review of Pomella et al.  
+
+The authors here have identified SKP2 as a critical driver of tumorigenesis in fusion negative rhabdomyosarcoma (FN- RMS) and acting downstream of MYOD. The authors have utilized publicly available genome- wide datasets both from their previous work and from others to establish the transcriptional regulation of SKP2 through MYOD. The study design and experiments are good and done in a systematic manner to show the MYOD- SKP2 axis that contributes to the tumorigenic phenotype in FN- RMS. Overall, the work performed here is satisfactory and novel in dissecting the MYOD- SKP2 signaling axis in FN- RMS. I have few concerns that should be addressed to make the findings more robust and clearer from the mechanistic point of view and outreach of this research.
+
+<--- Page Split --->
+
+We thank the Reviewer for the positive overall evaluation of the manuscript and the suggestions for improvement.  
+
+The major points of concern are:  
+
+1. The authors mention that SKP2 directly interacts with p27Kip1 and targets it for proteasomal degradation, and no interaction was detected for p21Cip1. Although both p21Cip1 and p27Kip1 protein levels increased post-treatment with proteasome inhibitor MG132. How do the authors explain this discrepancy?  
+
+As reported in our original submission, we were unable to detect an interaction between SKP2 and p21. However, we cannot exclude that the basal levels of p21 bound to SKP2 are under the threshold of antibody detection in our cell context.  
+
+It could be also possible that the enhancement of the p21 protein levels before the increase of its transcripts levels was due to protein stabilization through the modulation of other factors. P21, indeed, is subjected to context- dependent intense post- transcriptional modulation by a number of E3 ligases \(^{9 - 12}\) .  
+
+2. SKP2 depletion in FN-RMS cells resulted in increased MYOD and and CDK1 p57Kip2. What is the consequence of MYOD binding to chromatin at the SKP2 regulatory elements such as the intronic enhancer? Can the authors perform ChIP-seq or ChIP-qPCR to show the effect of MYOD binding after SKP2 depletion? Although ChIP-qPCR could be more direct for that particular SKP2 regulatory region but the genome-wide ChIP-seq after SKP2 silencing could also pinpoint the target genes for reduced stemness and tumorigenicity in RD and JR1 cells.  
+
+As requested by the Reviewer, we performed ChIP- seq for MYOD and the H3K27ac enhancer mark in shSKP2.2 and shSCR RD cells.  
+
+Consequently, we performed a new RNA- seq experiment analyzing RD and JR1 shSKP2.2 and shSCR cells to identify common transcriptional changes at the same time point.  
+
+The results are reported in Fig. S8.  
+
+Analysis of MYOD peaks on the SKP2 region shows that MYOD deposition does not seem to be significantly modulated (Ancillary Figure 11 for Reviewers). This could be due to the specific time point but it is also in line with the modest modulation in term of global increase of MYOD peaks after SKP2 depletion (Fig. S8e and Results section).  
+
+It is, indeed, evident by our analysis that MYOD deposition increased after SKP2 knockdown at peaks of MYOD- target myogenic genes that are commonly up- regulated in the two cell lines, such as CDKN1A, MYL1, MYOG and MYBPH. Therefore, MYOD binding/transcriptional activity was induced in a SKP2- depleted context particularly on selected genes such as myogenic genes (Fig. S8g).  
+
+Overall, the induction of differentiation suggests that the stemness signature should be compromised. In accordance, analyzing the RNA- seq we discovered that several genes that are included in a stemness gene set (STEMNESS- UP) were down- regulated (see the Results section, Figs. S8a,b and Supplementary Table 2). Then, we analyzed MYOD binding on putative MYOD- bound gene regions previously identified by Tenente et al. (Ref. 7 Revised Manuscript) and included in the STEMNESS- UP gene set. However, we were unable to find any modulation of MYOD binding even on those stemness genes that were down- regulated after SKP2 knockdown suggesting that MYOD functions on these genes are not relevant in the response to SKP2 depletion.
+
+<--- Page Split --->
+
+## 3. The authors used \(\beta\) -gal staining assay to show induction of senescence upon SKP2 depletion. Could the authors validate the same on the molecular levels showing the levels of any senescent marker genes for example SASP protein either by quantitative real time PCR from the available cDNA or by western blot.  
+
+When we analyzed the new RNA- seq data on RD and JR1 cells we noticed a down- regulation of a subset of SASP genes (88 out of 112 total genes included in the gene set https://www.gsea- msigdb.org/gsea/msigdb/cards/REACTOME_SENESCENCE_ASSOCIATED_SECRETORY_PHE NOTYPE_SASP), which have been reported in the Ancillary Table 1 for Reviewers.  
+
+Therefore, it appears conceivable that in our contest, senescence could be mainly related to the up- regulation of protein levels of p21, p27 and p57, which are pro- senescent factors (Refs. 45,50 Revised Manuscript and 13- 15).  
+
+Moreover, it has been demonstrated that p21- induced senescence is often not associated to SASP response (Ref. 51 Revised Manuscript).  
+
+We added a couple of lines on p21 in the Discussion.  
+
+4. The authors have shown effect on cell proliferation upon SKP2 depletion in cell lines and the same effect is also seen in the reduced tumor volumes after SKP2 is silenced by shRNA or using the NAE inhibitor LN4924. Since the IHC sections have been made and stained for the relevant targets but it was surprising that the authors haven't stained for the proliferation marker Ki67 in these sections. The Ki67 staining for the IHC sections shown if Fig 6k and 7c would make it clear at the molecular level.  
+
+The staining for Ki67 has been done and added to the revised manuscript as requested.  
+
+## To All Reviewers:  
+
+As reported in the Cover letter, we have replaced all the radiograms for Fig. S4d (previously Fig. S3d) of RD cells. Indeed, doing the uncropped figures, we noticed that the radiogram of p27Kip1 was erroneously duplicated from that of JR1 cells. Thus, we replaced all the radiograms for the investigated proteins in RD cells. We apologize for this inaccuracy.  
+
+## References  
+
+1. Tintignac, L. A. et al. Degradation of MyoD Mediated by the SCF (MAFbx) Ubiquitin Ligase. J. Biol. Chem. 280, 2847-2856 (2005).  
+2. Imaki, H. et al. Cell cycle-dependent regulation of the Skp2 promoter by GA-binding protein. Cancer Res. 63, 4607-13 (2003).  
+3. Hume, S. et al. The NUCKS1-SKP2-p21/p27 axis controls S phase entry. Nat. Commun. 12, 6959 (2021).  
+4. Wang, S., Raven, J. F. & Koromilas, A. E. STAT1 represses Skp2 gene transcription to promote p27Kip1 stabilization in Ras-transformed cells. Mol. Cancer Res. 8, 798-805 (2010).  
+5. Lu, D. Y. et al. The ETS transcription factor ETV6 constrains the transcriptional activity of EWS-FLI to promote Ewing sarcoma. Nat. Cell Biol. (2023) doi:10.1038/s41556-022-01059-8.  
+6. Garraway, L. A. & Sellers, W. R. Lineage dependency and lineage-survival oncogenes in human cancer. Nat. Rev. Cancer 6, 593-602 (2006).  
+7. Pomella, S. et al. Genomic and Epigenetic Changes Drive Aberrant Skeletal Muscle Differentiation in Rhabdomyosarcoma. Cancers (Basel). 15, 2823 (2023).  
+8. Singh, K. & Dilworth, F. J. Differential modulation of cell cycle progression distinguishes members of the myogenic regulatory factor family of transcription factors. FEBS J. 280, 3991-4003 (2013).  
+9. Zhang, L. et al. FBXO22 promotes the development of hepatocellular carcinoma by regulating the ubiquitination and degradation of p21. J. Exp. Clin. Cancer Res. 38, 101
+
+<--- Page Split --->
+
+(2019).10. Abbas, T. et al. PCNA-dependent regulation of p21 ubiquitilation and degradation via the CRL4 Cdt2 ubiquitin ligase complex. Genes Dev. 22, 2496–2506 (2008).11. Amador, V., Ge, S., Santamaría, P. G., Guardavaccaro, D. & Pagano, M. APC/CCdc20 Controls the Ubiquitin-Mediated Degradation of p21 in Prometaphase. Mol. Cell 27, 462–473 (2007).12. Wang, F. et al. Ubiquitination of p21 by E3 Ligase TRIM21 Promotes the Proliferation of Human Neuroblastoma Cells. NeuroMolecular Med. 23, 549–560 (2021).13. Majumder, P. K. et al. A Prostatic Intraepithelial Neoplasia-Dependent p27Kip1 Checkpoint Induces Senescence and Inhibits Cell Proliferation and Cancer Progression. Cancer Cell 14, 146–155 (2008).14. Kovach, A. R. et al. Identification and targeting of a <scp>HES1-YAP1-CDKN1C</scp> functional interaction in fusion-negative rhabdomyosarcoma. Mol. Oncol. 16, 3587–3605 (2022).15. Hernandez-Segura, A. et al. Unmasking Transcriptional Heterogeneity in Senescent Cells. Curr. Biol. 27, 2652-2660.e4 (2017).
+
+<--- Page Split --->
+![](images/Figure_1.jpg)
+
+<center>Ancillary Figure 1 </center>  
+
+![](images/Figure_5.jpg)
+
+<center>Ancillary Figure 1. SKP2 is regulated by a MYOD-bound enhancer. </center>  
+
+a, Representative western blot ( \(n = 3\) independent experiments) of the indicated proteins on JR1 and RH30 cells transfected with either Scrambled (siSCR) or two different MYOD siRNA sequences (siMYOD.2 and siMYOD.3) at 24 hours (h) and 48h post- transfection. Vinculin is the loading control. b, mRNA levels (RT- qPCR) of MYOD1 and SKP2 on cells treated as in (a) were normalized to GAPDH levels and expressed as fold increase over siSCR. \(n = 3\) independent experiments, data presented as mean values \(\pm \mathrm{SD}\) , two- way ANOVA.
+
+<--- Page Split --->
+
+## Ancillary Figure 2  
+
+![](images/Figure_6.jpg)
+  
+
+Ancillary Figure 2. SKP2 depletion induces p27<sup>Kip1</sup>- dependent cell cycle arrest reducing growth.  
+
+a, Representative western blot (n = 3 independent experiments) of p57<sup>Kip2</sup> on RD and JR1 cells transfected with either SCR, SKP2, p27<sup>Kip1</sup> or SKP2 + p27<sup>Kip1</sup> siRNA at 72h post-transfection. Vinculin is the loading control.
+
+<--- Page Split --->
+
+## Ancillary Figure 3  
+
+![](images/Figure_unknown_0.jpg)
+  
+
+## Ancillary Figure 3. NEDDylation inhibition prevents SKP2 functions and induces G2/M cell cycle arrest.  
+
+a, Histogram depicts fold changes of the percentage of MLN4924 treated cells (60 nM for 24 hours) in G1, S and G2/M cell cycle phases over DMSO cells (1 arbitrary unit, not reported). \(n = 3\) independent experiments, data presented as mean values \(\pm\) SD, Student's two- tailed t- test.
+
+<--- Page Split --->
+
+## Ancillary Figure 4  
+
+![](images/Figure_8.jpg)
+  
+
+Ancillary Figure 4. MYOD is not a direct substrate of SKP2.  
+
+a, Representative western blot (n = 3 independent experiments) of co- Immunoprecipitation of either endogenous SKP2 (left) or MYOD (right) in RD and JR1 cells showing SKP2 and MYOD.
+
+<--- Page Split --->
+![](images/Figure_10.jpg)
+
+<center>Ancillary Figure 5. E2F1 does not regulates SKP2 expression in rhabdomyosarcoma cells. </center>  
+
+a, Representative western blot ( \(n = 3\) independent experiments) of the indicated proteins on RD, JR1, RH4 and RH30 cells transfected with either Scrambled (siSCR) or two different E2F1 siRNA sequences (E2F1.1: UAACUGCACUUUCGGCCUUU and E2F1.2: CUACUCAGCCUGGAGCA- AGAA, Sigma- Aldrich, St Louis, MO, USA) at 24 hours (h) post- transfection. Vinculin is the loading control. b, mRNA levels (RT- qPCR) of E2F1 and SKP2 on cells treated as in (a) were normalized to GAPDH levels and expressed as fold increase over siSCR. \(n = 3\) independent experiments, data presented as mean values \(\pm\) SD, two- way ANOVA.
+
+<--- Page Split --->
+![](images/Figure_11.jpg)
+
+<center>Aancillary Figure 6. MYOD1 is a strong and selective dependency in rhabdomyosarcoma cells. </center>  
+
+a, Volcano plot identifying selective gene dependencies in rhabdomyosarcoma cells (effect size) among 17451 genes analyzed. b, Violin plot depicting MYOD1 gene effect comparison between RMS cells and all other cell types present in DepMap public 23Q2. Data were obtained from DepMap (https://depmap.org/portal/).
+
+<--- Page Split --->
+
+# Ancillary Figure 7 
+
+a 
+
+![PLACEHOLDER_26_0]
+
+ 
+
+c 
+
+![PLACEHOLDER_26_1]
+
+ 
+
+e 
+
+![PLACEHOLDER_26_2]
+
+ 
+
+<center>DAPI/MyHC</center> 
+
+![PLACEHOLDER_26_3]
+
+ 
+
+d 
+
+![PLACEHOLDER_26_4]
+
+
+<--- Page Split --->
+
+## Ancillary Figure 7. MYOD depletion induces growth arrest and hampers myogenic differentiation.  
+
+a, Growth curve analysis of RD and JR1 cells transfected with either Scrambled (siSCR) or MYOD siRNA (siMYOD.1). \(\bar{\mathbf{n}} = 3\) independent experiments, data presented as mean values \(\pm \mathrm{SD}\) Student's two- tailed t- test. b, Histogram depicts fold changes of the percentage of cells treated as in (a) in G1, S and G2/M cell cycle phases over siSCR cells (1 arbitrary unit, not reported). \(\bar{\mathbf{n}} = 3\) independent experiments, data presented as mean values \(\pm \mathrm{SD}\) , Student's two- tailed t- test. P- value \(^*\leq 0.05\) , \(^{**} \leq 0.01\) , \(^{****} \leq 0.0001\) . c, Representative western blot ( \(\bar{\mathbf{n}} = 2\) independent experiments) of the indicated proteins on RD and JR1 cells transfected with either Scrambled (siSCR) or MYOD siRNA (siMYOD.1) at 48 hours (h) post- transfection. Vinculin is the loading control. d, mRNA levels (RT- qPCR) of the reported genes on cells treated as in (c) were normalized to GAPDH levels and expressed as fold increase over siSCR. \(\bar{\mathbf{n}} = 2\) independent experiments, data presented as mean values \(\pm \mathrm{SD}\) , Student's two- tailed t- test. e, Representative immunofluorescence of RD and JR1 cells treated as in (c) stained for Myosin Heavy Chain (MyHC) (green). Nuclei were stained with DAPI (blue). Scale Bar \(= 50 \mu \mathrm{m}\) .
+
+<--- Page Split --->
+![PLACEHOLDER_28_0]
+
+<center>Ancillary Figure 8. MYOD overexpression does not affect rhabdomyosarcoma cell growth. </center>  
+
+a, Growth curve analysis of RD and JR1 infected with a Doxycycline- inducible lentiviral vector overexpressing human MYOD1 (pINDUCER20- FLAG- hMYOD (#800), Addgene) and treated or not with \(1\mu \mathrm{g / mL}\) of Doxycycline. \(n = 3\) independent experiments, data presented as mean values \(\pm\) SD, Student's two- tailed t- test. b, mRNA levels (RT- qPCR) of the reported genes on cells treated as in (a) for 72h were normalized to GAPDH levels and expressed as fold increase over Doxycycline- untreated cells. \(n = 3\) independent experiments, data presented as mean values \(\pm\) SD, Student's two- tailed t- test. c, Representative western blot ( \(n = 3\) independent experiments) of the indicated proteins on RD and JR1 cells treated as in (b). Vinculin is the loading control.
+
+<--- Page Split --->
+
+# Ancillary Figure 9 
+
+## a 
+
+![PLACEHOLDER_29_0]
+
+ 
+
+## C 
+
+![PLACEHOLDER_29_1]
+
+ 
+
+## e 
+
+![PLACEHOLDER_29_2]
+
+
+<--- Page Split --->
+
+## Ancillary Figure 9. MYOD-SKP2 co-silencing enhances cell growth arrest but impedes myogenic differentiation.  
+
+a, Growth curve analysis of RD and JR1 cells transfected with either Scrambled (SCR), SKP2, MYOD or SKP2 + MYOD siRNAs. \(\mathrm{n} = 3\) independent experiments, data presented as mean values \(\pm\) SD, Student's two- tailed t- test. b, Histogram depicts fold changes of the percentage of cells treated as in (a) in G1, S and G2/M cell cycle phases over siSCR cells (1 arbitrary unit, not reported). \(\mathrm{n} = 3\) independent experiments, data presented as mean values \(\pm\) SD, two- way ANOVA. P- value vs siSCR \(^* \leq 0.05\) , \(^{**} \leq 0.01\) , \(^{***} \leq 0.001\) , \(^{****} \leq 0.0001\) ; P- value vs siSKP2 \(\$ \leq 0.05\) , \(\$ \leq 0.01\) , \(\$ \leq 0.001\) . c, Representative western blot ( \(\mathrm{n} = 3\) independent experiments) of the indicated proteins on RD and JR1 cells treated as in (a). Vinculin is the loading control. d, mRNA levels (RT- qPCR) of the reported genes on cells treated as in (a) for 72h were normalized to GAPDH levels and expressed as fold increase over siSCR. \(\mathrm{n} = 2\) independent experiments, data presented as mean values \(\pm\) SD. e, Representative immunofluorescence of RD and JR1 cells treated as in (a) stained for Myosin Heavy Chain (MyHC) (green). Nuclei were stained with DAPI (blue). Scale Bar \(= 50 \mu \mathrm{m}\) .
+
+<--- Page Split --->
+![PLACEHOLDER_31_0]
+
+<center>Ancillary Figure 10. Sensitivity to MLN4924 correlates with SKP2 expression but not with MYOD1 expression </center>  
+
+Pearson correlation analysis between MLN4924 AUC and (a) MYOD1 expression or (b) SKP2 expression performed on 721 cancer cell lines. Data were obtained from DepMap portal (https://depmap.org/portal/) using Expression Public 23Q2 dataset for gene expression data and Drug Sensitivity AUC (CTD^2) dataset for MLN4924 AUC. TPM, Transcripts per Million; AUC, area under the curve.
+
+<--- Page Split --->
+![PLACEHOLDER_32_0]
+
+<center>Ancillary Figure 11. MYOD binding at SKP2 locus after SKP2 silencing </center>  
+
+a, Profile of ChIP-seq read densities of MYOD (green), and RNA-seq (grey) at SKP2 locus on RD cells.
+
+<--- Page Split --->
+
+
+Ancillary Table 1. RD and JR1 Commonly downregulated genes in REACTOME_SENESCENCE_ASSOCIATED_SECRETORY_PHENOTYPE_SASP gene set   
+
+<table><tr><td></td><td colspan="4">RD</td><td colspan="4">JR1</td></tr><tr><td>GeneID</td><td>shSCR</td><td>shSKP2</td><td>FC</td><td>LOG2(FC)</td><td>shSCR</td><td>shSKP2</td><td>FC</td><td>LOG2(FC)&lt;fci&gt;</td></tr><tr><td>ANAPC1</td><td>16.7</td><td>8.62</td><td>0.5435</td><td>0.6262</td><td>11.82</td><td>7.27</td><td>0.6451</td><td>0.7182</td></tr><tr><td>ANAPC11</td><td>52.05</td><td>33.03</td><td>0.6415</td><td>0.7150</td><td>37.54</td><td>31.66</td><td>0.8474</td><td>0.8855</td></tr><tr><td>ANAPC15</td><td>16.59</td><td>6.9</td><td>0.4491</td><td>0.5352</td><td>9.43</td><td>4.14</td><td>0.4928</td><td>0.5780</td></tr><tr><td>ANAPC2</td><td>29.5</td><td>23.37</td><td>0.7990</td><td>0.8472</td><td>23.87</td><td>18.72</td><td>0.7929</td><td>0.8423</td></tr><tr><td>ANAPC5</td><td>52.79</td><td>40.32</td><td>0.7682</td><td>0.8223</td><td>51.69</td><td>41.38</td><td>0.8043</td><td>0.8515</td></tr><tr><td>ANAPC7</td><td>37.21</td><td>22.77</td><td>0.6221</td><td>0.6979</td><td>29.23</td><td>19.6</td><td>0.6814</td><td>0.7497</td></tr><tr><td>CCNA2</td><td>27.32</td><td>17.65</td><td>0.6585</td><td>0.7299</td><td>10.1</td><td>6.1</td><td>0.6396</td><td>0.7134</td></tr><tr><td>CDC23</td><td>13.67</td><td>11.01</td><td>0.8187</td><td>0.8629</td><td>13.62</td><td>12.82</td><td>0.9453</td><td>0.9600</td></tr><tr><td>CDC26</td><td>17.85</td><td>12.28</td><td>0.7045</td><td>0.7694</td><td>23.25</td><td>12.79</td><td>0.5687</td><td>0.6495</td></tr><tr><td>CDC27</td><td>22.07</td><td>17.23</td><td>0.7902</td><td>0.8401</td><td>15.29</td><td>13.52</td><td>0.8913</td><td>0.9194</td></tr><tr><td>CDK2</td><td>39.09</td><td>12.99</td><td>0.3490</td><td>0.4319</td><td>20.74</td><td>5.86</td><td>0.3155</td><td>0.3957</td></tr><tr><td>CDK4</td><td>133.91</td><td>102.68</td><td>0.7685</td><td>0.8225</td><td>121.52</td><td>116.45</td><td>0.9586</td><td>0.9698</td></tr><tr><td>CDKN1B</td><td>15.57</td><td>11.06</td><td>0.7278</td><td>0.7890</td><td>11.15</td><td>9.53</td><td>0.8667</td><td>0.9005</td></tr><tr><td>CDKN2A</td><td>34.1</td><td>28.79</td><td>0.8487</td><td>0.8865</td><td>96.8</td><td>71.93</td><td>0.7457</td><td>0.8038</td></tr><tr><td>CDKN2B</td><td>2.79</td><td>2.38</td><td>0.8918</td><td>0.9198</td><td>10.13</td><td>7.97</td><td>0.8059</td><td>0.8527</td></tr><tr><td>CDKN2C</td><td>49.08</td><td>23.69</td><td>0.4930</td><td>0.5782</td><td>15.86</td><td>9.79</td><td>0.6400</td><td>0.7137</td></tr><tr><td>CXCL8</td><td>2.17</td><td>1.33</td><td>0.7350</td><td>0.7949</td><td>4.89</td><td>1.03</td><td>0.3447</td><td>0.4272</td></tr><tr><td>EHMT1</td><td>42.27</td><td>36.68</td><td>0.8708</td><td>0.9037</td><td>32.85</td><td>30.86</td><td>0.9412</td><td>0.9570</td></tr><tr><td>EHMT2</td><td>58.25</td><td>32.07</td><td>0.5581</td><td>0.6398</td><td>16.86</td><td>14.13</td><td>0.8471</td><td>0.8853</td></tr><tr><td>FOS</td><td>93.54</td><td>24.38</td><td>0.2685</td><td>0.3431</td><td>109.8</td><td>24.11</td><td>0.2266</td><td>0.2947</td></tr><tr><td>FZR1</td><td>29.2</td><td>18.93</td><td>0.6599</td><td>0.7311</td><td>20</td><td>12.44</td><td>0.6400</td><td>0.7137</td></tr><tr><td>H2AC14</td><td>411.76</td><td>100.48</td><td>0.2459</td><td>0.3171</td><td>112.08</td><td>26.52</td><td>0.2434</td><td>0.3143</td></tr><tr><td>H2AC18</td><td>2469.16</td><td>1019.57</td><td>0.4132</td><td>0.4989</td><td>892.71</td><td>359.27</td><td>0.4031</td><td>0.4886</td></tr><tr><td>H2AC20</td><td>647.78</td><td>226.18</td><td>0.3502</td><td>0.4331</td><td>234.35</td><td>113.3</td><td>0.4857</td><td>0.5711</td></tr><tr><td>H2AC4</td><td>120.82</td><td>37.23</td><td>0.3138</td><td>0.3938</td><td>41.97</td><td>11.99</td><td>0.3023</td><td>0.3811</td></tr><tr><td>H2AC6</td><td>318.32</td><td>154.69</td><td>0.4876</td><td>0.5730</td><td>127.23</td><td>84.43</td><td>0.6662</td><td>0.7366</td></tr><tr><td>H2AC7</td><td>91.2</td><td>40.08</td><td>0.4456</td><td>0.5316</td><td>31.65</td><td>12.73</td><td>0.4205</td><td>0.5064</td></tr><tr><td>H2AC8</td><td>260.48</td><td>90.85</td><td>0.3513</td><td>0.4343</td><td>94.02</td><td>35.98</td><td>0.3892</td><td>0.4742</td></tr><tr><td>H2AJ</td><td>13.31</td><td>11.82</td><td>0.8959</td><td>0.9229</td><td>21.37</td><td>9.31</td><td>0.4609</td><td>0.5468</td></tr><tr><td>H2AX</td><td>132.82</td><td>103.32</td><td>0.7796</td><td>0.8315</td><td>70.37</td><td>42.58</td><td>0.6106</td><td>0.6876</td></tr><tr><td>H2AZ2</td><td>48</td><td>38.4</td><td>0.8041</td><td>0.8513</td><td>27.5</td><td>25.64</td><td>0.9347</td><td>0.9521</td></tr><tr><td>H2BC1</td><td>257.59</td><td>65.05</td><td>0.2554</td><td>0.3282</td><td>64.67</td><td>26.64</td><td>0.4209</td><td>0.5068</td></tr><tr><td>H2BC10</td><td>295.63</td><td>81.15</td><td>0.2769</td><td>0.3527</td><td>91.21</td><td>18.33</td><td>0.2096</td><td>0.2746</td></tr><tr><td>H2BC11</td><td>242.87</td><td>84.88</td><td>0.3522</td><td>0.4353</td><td>76.4</td><td>30.61</td><td>0.4084</td><td>0.4941</td></tr><tr><td>H2BC12</td><td>894.47</td><td>351.33</td><td>0.3935</td><td>0.4787</td><td>381.36</td><td>167.23</td><td>0.4400</td><td>0.5260</td></tr><tr><td>H2BC13</td><td>189.63</td><td>39.15</td><td>0.2106</td><td>0.2757</td><td>78.21</td><td>15.95</td><td>0.2140</td><td>0.2798</td></tr><tr><td>H2BC14</td><td>231.47</td><td>51.01</td><td>0.2237</td><td>0.2913</td><td>58.26</td><td>12.53</td><td>0.2283</td><td>0.2967</td></tr><tr><td>H2BC15</td><td>216.79</td><td>111.62</td><td>0.5171</td><td>0.6013</td><td>93.26</td><td>47.01</td><td>0.5093</td><td>0.5939</td></tr><tr><td>H2BC17</td><td>363.76</td><td>132.77</td><td>0.3667</td><td>0.4507</td><td>125.05</td><td>45.8</td><td>0.3713</td><td>0.4555</td></tr><tr><td>H2BC21</td><td>133.8</td><td>78.91</td><td>0.5928</td><td>0.6716</td><td>67.33</td><td>62.75</td><td>0.9330</td><td>0.9508</td></tr><tr><td>H2BC3</td><td>114.28</td><td>47.77</td><td>0.4231</td><td>0.5090</td><td>38.04</td><td>17.11</td><td>0.4639</td><td>0.5498</td></tr><tr><td>H2BC4</td><td>647.34</td><td>279.74</td><td>0.4330</td><td>0.5191</td><td>215.06</td><td>175.05</td><td>0.8148</td><td>0.8598</td></tr><tr><td>H2BC5</td><td>415.74</td><td>164.12</td><td>0.3962</td><td>0.4815</td><td>103.84</td><td>79.16</td><td>0.7646</td><td>0.8193</td></tr><tr><td>H2BC6</td><td>213.22</td><td>47.88</td><td>0.2282</td><td>0.2965</td><td>103.93</td><td>20.41</td><td>0.2040</td><td>0.2679</td></tr><tr><td>H2BC7</td><td>234.27</td><td>69.24</td><td>0.2986</td><td>0.3769</td><td>75.52</td><td>24.29</td><td>0.3305</td><td>0.4120</td></tr></table>
+
+<--- Page Split --->
+
+<table><tr><td>H2BC8</td><td>219.58</td><td>98.25</td><td>0.4500</td><td>0.5360</td><td>90.2</td><td>41.73</td><td>0.4685</td><td>0.5544</td></tr><tr><td>H2BC9</td><td>346.12</td><td>111.08</td><td>0.3229</td><td>0.4037</td><td>92.94</td><td>28.43</td><td>0.3133</td><td>0.3932</td></tr><tr><td>H3-3A</td><td>96.12</td><td>78.88</td><td>0.8225</td><td>0.8659</td><td>68.95</td><td>66.48</td><td>0.9647</td><td>0.9743</td></tr><tr><td>H3-3B</td><td>111.46</td><td>79.71</td><td>0.7177</td><td>0.7805</td><td>98.81</td><td>74.54</td><td>0.7568</td><td>0.8130</td></tr><tr><td>H3C1</td><td>183.04</td><td>52.31</td><td>0.2897</td><td>0.3670</td><td>61.72</td><td>13.99</td><td>0.2390</td><td>0.3092</td></tr><tr><td>H3C10</td><td>528.48</td><td>111.75</td><td>0.2129</td><td>0.2785</td><td>156.37</td><td>31.33</td><td>0.2054</td><td>0.2696</td></tr><tr><td>H3C11</td><td>157.88</td><td>41.4</td><td>0.2669</td><td>0.3413</td><td>47.48</td><td>11.98</td><td>0.2677</td><td>0.3423</td></tr><tr><td>H3C12</td><td>212.49</td><td>68.85</td><td>0.3272</td><td>0.4084</td><td>69.73</td><td>15.61</td><td>0.2348</td><td>0.3043</td></tr><tr><td>H3C13</td><td>490.78</td><td>130.97</td><td>0.2684</td><td>0.3430</td><td>120.42</td><td>20.37</td><td>0.1760</td><td>0.2339</td></tr><tr><td>H3C14</td><td>11.37</td><td>8.19</td><td>0.7429</td><td>0.8015</td><td>16.64</td><td>15.95</td><td>0.9609</td><td>0.9715</td></tr><tr><td>H3C15</td><td>1746.3</td><td>641.32</td><td>0.3676</td><td>0.4517</td><td>402.3</td><td>116.32</td><td>0.2909</td><td>0.3684</td></tr><tr><td>H3C2</td><td>1039.78</td><td>316.65</td><td>0.3052</td><td>0.3843</td><td>264.5</td><td>92.55</td><td>0.3524</td><td>0.4355</td></tr><tr><td>H3C3</td><td>496.2</td><td>187.23</td><td>0.3786</td><td>0.4632</td><td>162.92</td><td>54.07</td><td>0.3360</td><td>0.4179</td></tr><tr><td>H3C4</td><td>472.54</td><td>124.46</td><td>0.2649</td><td>0.3391</td><td>120.96</td><td>35.84</td><td>0.3021</td><td>0.3808</td></tr><tr><td>H3C6</td><td>447.63</td><td>157.8</td><td>0.3540</td><td>0.4372</td><td>125.47</td><td>36.76</td><td>0.2986</td><td>0.3769</td></tr><tr><td>H3C7</td><td>495.91</td><td>161.94</td><td>0.3279</td><td>0.4092</td><td>153.47</td><td>43.37</td><td>0.2872</td><td>0.3643</td></tr><tr><td>H3C8</td><td>257.05</td><td>115.94</td><td>0.4532</td><td>0.5392</td><td>104.02</td><td>35.8</td><td>0.3504</td><td>0.4334</td></tr><tr><td>H4C1</td><td>261.09</td><td>67.46</td><td>0.2612</td><td>0.3348</td><td>71.9</td><td>14.45</td><td>0.2119</td><td>0.2773</td></tr><tr><td>H4C11</td><td>430.57</td><td>182</td><td>0.4240</td><td>0.5100</td><td>152.39</td><td>71.86</td><td>0.4750</td><td>0.5607</td></tr><tr><td>H4C12</td><td>485.91</td><td>207.27</td><td>0.4277</td><td>0.5137</td><td>271.17</td><td>88.44</td><td>0.3286</td><td>0.4099</td></tr><tr><td>H4C13</td><td>71.04</td><td>20.07</td><td>0.2925</td><td>0.3701</td><td>14.27</td><td>2.98</td><td>0.2606</td><td>0.3342</td></tr><tr><td>H4C14</td><td>1163.49</td><td>295.87</td><td>0.2549</td><td>0.3276</td><td>335.38</td><td>53.94</td><td>0.1633</td><td>0.2183</td></tr><tr><td>H4C15</td><td>1173.55</td><td>316.89</td><td>0.2706</td><td>0.3456</td><td>351.01</td><td>75.46</td><td>0.2172</td><td>0.2836</td></tr><tr><td>H4C2</td><td>470.85</td><td>154.03</td><td>0.3286</td><td>0.4099</td><td>145.8</td><td>57.2</td><td>0.3965</td><td>0.4818</td></tr><tr><td>H4C3</td><td>456.98</td><td>168.46</td><td>0.3700</td><td>0.4542</td><td>115.4</td><td>59.05</td><td>0.5159</td><td>0.6002</td></tr><tr><td>H4C4</td><td>549.27</td><td>101.84</td><td>0.1869</td><td>0.2472</td><td>149.7</td><td>28.12</td><td>0.1932</td><td>0.2549</td></tr><tr><td>H4C5</td><td>570.74</td><td>221.96</td><td>0.3900</td><td>0.4751</td><td>202.62</td><td>97.44</td><td>0.4834</td><td>0.5690</td></tr><tr><td>H4C8</td><td>596.03</td><td>260.4</td><td>0.4378</td><td>0.5239</td><td>188.98</td><td>98</td><td>0.5211</td><td>0.6051</td></tr><tr><td>H4C9</td><td>223.01</td><td>62.11</td><td>0.2817</td><td>0.3581</td><td>57.43</td><td>13.28</td><td>0.2444</td><td>0.3154</td></tr><tr><td>IL6</td><td>7.28</td><td>0.36</td><td>0.1643</td><td>0.2194</td><td>8.18</td><td>0.55</td><td>0.1688</td><td>0.2251</td></tr><tr><td>MAPK3</td><td>22.73</td><td>11.34</td><td>0.5200</td><td>0.6041</td><td>18.85</td><td>11.75</td><td>0.6423</td><td>0.7157</td></tr><tr><td>MAPK7</td><td>6.86</td><td>5.74</td><td>0.8575</td><td>0.8934</td><td>3.75</td><td>3.31</td><td>0.9074</td><td>0.9316</td></tr><tr><td>NFKB1</td><td>14.56</td><td>10.5</td><td>0.7391</td><td>0.7983</td><td>12.4</td><td>9.88</td><td>0.8119</td><td>0.8575</td></tr><tr><td>RELA</td><td>25.96</td><td>21.98</td><td>0.8524</td><td>0.8894</td><td>23.1</td><td>19.3</td><td>0.8423</td><td>0.8815</td></tr><tr><td>RPS27A</td><td>292.54</td><td>240.57</td><td>0.8230</td><td>0.8663</td><td>297.21</td><td>232.06</td><td>0.7815</td><td>0.8331</td></tr><tr><td>RPS6KA1</td><td>4.37</td><td>2.52</td><td>0.6555</td><td>0.7273</td><td>3.07</td><td>0.61</td><td>0.3956</td><td>0.4809</td></tr><tr><td>RPS6KA2</td><td>26.39</td><td>17.16</td><td>0.6630</td><td>0.7338</td><td>4.48</td><td>4.25</td><td>0.9580</td><td>0.9694</td></tr><tr><td>STAT3</td><td>43.14</td><td>34.14</td><td>0.7961</td><td>0.8449</td><td>51.76</td><td>35.24</td><td>0.6869</td><td>0.7544</td></tr><tr><td>UBB</td><td>774.12</td><td>567.77</td><td>0.7338</td><td>0.7939</td><td>691.64</td><td>477.24</td><td>0.6905</td><td>0.7574</td></tr><tr><td>UBC</td><td>456.28</td><td>377.65</td><td>0.8280</td><td>0.8703</td><td>475.52</td><td>457.55</td><td>0.9623</td><td>0.9725</td></tr><tr><td>UBE2C</td><td>208.12</td><td>149.44</td><td>0.7194</td><td>0.7819</td><td>78.7</td><td>44.51</td><td>0.5710</td><td>0.6517</td></tr><tr><td>UBE2E1</td><td>26.7</td><td>17.61</td><td>0.6718</td><td>0.7414</td><td>20.92</td><td>13.45</td><td>0.6592</td><td>0.7305</td></tr><tr><td>UBE2S</td><td>108.15</td><td>94.12</td><td>0.8715</td><td>0.9042</td><td>59.7</td><td>51.58</td><td>0.8662</td><td>0.9001</td></tr></table>
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+The authors have addressed most of the raised concerns.  
+
+Reviewer #2 (Remarks to the Author):  
+
+I have completed the review of the revised manuscript, and I am pleased to report that the authors have successfully addressed all the concerns and suggestions raised during the initial review process. I find the paper to be of high quality and believe it is now suitable for publication.  
+
+I have no further issues or recommendations to add at this time.  
+
+Thank you for considering my assessment, and I look forward to seeing this valuable contribution in print.  
+
+Reviewer #4 (Remarks to the Author):  
+
+I am satisfied with the authors explanations and recommend this for publication.  
+
+Reviewer #5 (Remarks to the Author):  
+
+I was asked to assess whether the authors had sufficiently addressed the concerns raised by reviewer #3. To address these, the authors provided additional experimental data as well as elegant data mining approaches. They furthermore adequately rephrased the passages concerning MYOD being pro- tumorigenic. Overall, I can confidently say that the authors have addressed the comments by reviewer #3 thoroughly.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1da3cc07656e2b2ad084cb5e938a2f126f69f88db62135e01903242c34fe8e44/images_list.json

@@ -0,0 +1,17 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_0.jpg",
+    "caption": "Figure above (now Supplementary Fig. S7): Estimation of admixture proportions (f4-ratios) between populations.",
+    "footnote": [],
+    "bbox": [
+      [
+        123,
+        214,
+        875,
+        781
+      ]
+    ],
+    "page_idx": 9
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1da3cc07656e2b2ad084cb5e938a2f126f69f88db62135e01903242c34fe8e44/supplementary_0_Peer Review File__1da3cc07656e2b2ad084cb5e938a2f126f69f88db62135e01903242c34fe8e44.mmd

@@ -0,0 +1,322 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+African bushpigs exhibit porous species boundaries and appeared in Madagascar concurrently with human arrival  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+The manuscript on African bush pigs is a very detailed and thorough study of the two Potamochoerus species in Africa and based on the complete genome sequence of 67 individuals from across Africa, including Madagascar. The authors have used state of the art methodology to study the population genetics, speciation and admixture of these species. I have few comments on this manuscript which describes a very coherent and convincing story about these two species. In their discussion, the authors briefly tough upon the ongoing debate about the taxonomy of Potamochoerus and I agree with their comment that this very much depends on the species concept used. In fact, the species concept reflects the urge of humans to categorize, while the past decades have increasingly shown the diffuse borders between species with ongoing hybridization between many populations assumed to represent different species. I also welcomed their cautionary remark about the interpretation of PSMC curves and the cautionary remark about the assumption of total panmixia which indeed not always might be correct.  
+
+Minor comments:  
+
+Lines 201- 203. I found the comment about the lowest Dxy between Ugandan and Congolese populations because of the closest geographical distance between these populations not very convincing. The distance between several others (e.g. Gabon) as well as the actual Dxy values between Ugandan and several others were not that very much different. Consider removing this sentence.  
+
+Figure 3c. The authors do not very clearly indicate why only these 18 individuals are shown. Although the authors mention it is based on read depth, they could have been a bit clearer and also might have used a read depth of \(>10x\) as the threshold and shown the 23 individuals that are above that threshold.  
+
+Reviewer #2:  
+
+Remarks to the Author:  
+
+Balboa et al present a fascinating study on the evolutionary history of African suids. This manuscript specifically interrogates the putative suture zone of red river hogs and bushpis, including estimating levels of admixture and divergence times. In addition, the manuscript focusses on the Malagasy bushpig divergence, and how this likely relates to historic human wildlife translocations.  
+
+The study is well- presented, and appealing to a wide audience, with broader relevance to phylogenetic patterns in Africa, factors influencing speciation and human- influenced species distributions.  
+
+I have questions around some of the analyses/interpretations that I would like the authors to clarify before I can recommend publication:  
+
+## METHODS  
+
+The "Runs of homozygosity" section:  
+
+"at least two heterozygous reads to make a heterozygous call."  
+
+- What do you mean heterozygous reads? Do you mean allelic balance of \(\geq 0.2\) ? I.e. at least 2 alleles out of 10 being e.g. alternate?  
+
+"SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded."- Why? So you are filtering out regions of high heterozygosity? Presumably because you are assuming
+
+<--- Page Split --->
+
+that these correspond to repeat regions / duplications? Shouldn't this already be taken into account with your repeat masking that you have already done on the genome?  
+
+Have you considered a slightly more sophisticated program for identifying RoH's? E.g. RoHan? If you don't believe this is necessary/appropriate - why not?  
+
+TT and split time estimations  
+
+- This all sounds great, however, I have not used this method before and so it is difficult for me to judge the robustness of the analysis. The reference explains the theory of the approach, but some expansion of how the theory has been applied to this specific case would be very useful to the reader, for example in the supp methods? Or if not, at least a reference to a different paper that has used the approach and has given a more complete description.  
+
+## RESULTS  
+
+"with only the Congo individuals being closer to the bushpigs than the other red river hogs" - I initially read this as you saying that the Congo individuals were closer to bushpigs than red river hogs, but actually I think you mean "were closer to bushpigs than the other red river hogs, but overall still closer to red river hogs". Perhaps consider rewording it to clarify (as long as you can find a way that is better than my suggestion!)  
+
+In222- 229 "Given the observed FST and Dxy values..."  
+
+- In this paragraph you appear to shift back and forth when talking about between "species" ranges to within. When you then end with Uganda it is not clear which of these two you are talking about  
+
+For the Dstatistic analysis, you state "Madagascar as H1" in the Figure 2c legend and "non- Malagasy bushpigs (H2)", but the figure shows "SA" as H1. Also in the text In242, you state "Madagascar as H1". Please clarify.  
+
+Could the "strength" of the signal you detect, not just be reflecting the power of the analysis for each pairwise comparison, rather than the amount of gene- flow itself? Could it be worth calculating f4 ratios to attempt to quantify the admixture proportions for these pairwise comparisons?  
+
+"Given the results reported above, the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of introgression between the two taxa." - This seems a bit speculative to me. The PSMC results are interesting, and visually appealing, and therefore I do think that they add to the story, however they are difficult to interpret in terms of any kind of specific hypothesis. Consider re- wording or removing part of this section.  
+
+"This analysis suggested that the Malagasy population experienced a severe bottleneck, likely a result of a founder event between 1- 5 kya."  
+
+- The decline seems to start quite some time before this though - why might this be?  
+
+## DISCUSSION  
+
+"had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current 63. However, from the PCA, NGAadmix and IBS tree we did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+- To?: "...North Mozambique current 63. However, while we were unable to explicitly test this hypothesis with our dataset (due to no samples from Comoros etc), PCA NGAadmix and the IBS tree did not..."?
+
+<--- Page Split --->
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+The paper is clearly written and illustrated, though as an archaeologist I am in no position to pass judgement on the detailed methodology of the genetic study undertaken. That said, references to the debate surrounding the timing of human settlement of Madagascar are fair and the conclusion that bushpigs were likely introduced there \(\sim 1000 - 1500\) years ago does indeed fit extremely well with: a) dated faunal remains from the island, not just of bushpig but also of other exotic mammal taxa; b) the oldest archaeologically unambiguous, well- dated evidence for human presence; c) and specifically the African origins (and names) of domestic mammals and evidence for Triangular Incised Ware (Tana Ware), a form of pottery found in late first- millennium AD contexts in the southwest of the island - see Parker Pearson et al. 2010: 79 (Pastoralists, Warriors and Colonists: The Archaeology of Southern Madagascar; Oxford: Archaeopress).  
+
+For further discussion/reference of the timing of human settlement I would recommend Mitchell, Journal of Island and Coastal Archaeology 2020, arguments further elaborated in his 2022 book African Islands: A Comparative Archaeology (London: Routledge).  
+
+Provided that the genetic analysis is sound, I recommend acceptance.
+
+<--- Page Split --->
+
+## Response to reviewer comments  
+
+## A note to the Reviewers:  
+
+We thank Reviewers #1, #2 and #3 for taking time to review our manuscript and for the constructive and positive feedback that they have provided. We have detailed our point- by- point responses in red below and have provided a revised version of our manuscript. Line numbers in this response document refer to the line numbers in the original submitted manuscript.  
+
+## REVIEWER COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+The manuscript on African bush pigs is a very detailed and thorough study of the two Potamochoerus species in Africa and based on the complete genome sequence of 67 individuals from across Africa, including Madagascar. The authors have used state of the art methodology to study the population genetics, speciation and admixture of these species. I have few comments on this manuscript which describes a very coherent and convincing story about these two species. In their discussion, the authors briefly tough upon the ongoing debate about the taxonomy of Potamochoerus and I agree with their comment that this very much depends on the species concept used. In fact, the species concept reflects the urge of humans to categorize, while the past decades have increasingly shown the diffuse borders between species with ongoing hybridization between many populations assumed to represent different species. I also welcomed their cautionary remark about the interpretation of PSMC curves and the cautionary remark about the assumption of total panmixia which indeed not always might be correct.  
+
+Response: Thank you for taking the time to review our manuscript and for your positive comments.  
+
+Minor comments:  
+
+Lines 201- 203. I found the comment about the lowest Dxy between Ugandan and Congolese populations because of the closest geographical distance between these populations not very convincing. The distance between several others (e.g. Gabon) as well as the actual Dxy values between Ugandan and several others were not that very much different. Consider removing this sentence.  
+
+Response: Thank you for your comment. We have updated lines 200- 203 to be more precise in our description, where we include values and describe between- species \(D_{xy}\) relative to within- species \(D_{xy}\) between DR Congo and Ghana, as below:  
+
+"When comparing across species, \(D_{xy}\) was lowest for populations that were geographically central, with the Ugandan population exhibiting the lowest between- species \(D_{xy}\) for all bushpigs and the Congolese the lowest \(D_{xy}\) for all red river hogs. Additionally, the lowest \(D_{xy}\) between species was observed between the Ugandan and Congolese populations (0.00355), similar to within- species \(D_{xy}\) comparisons for Ghana and DR Congo (0.00352)."
+
+<--- Page Split --->
+
+Figure 3c. The authors do not very clearly indicate why only these 18 individuals are shown. Although the authors mention it is based on read depth, they could have been a bit clearer and also might have used a read depth of \(>10x\) as the threshold and shown the 23 individuals that are above that threshold.  
+
+Response: We agree that we were unclear in our description of medium- high depth samples.  
+
+We used a minimum of 12X as our threshold since PSMC curves exhibit high variance when using samples \(\leq 10x\) in combination with missing data (Nadachowska- Brzyska et al. (2016); ref. 71). As a result, we had 18 unrelated individuals that met this threshold, where the sample with the lowest coverage was 14X (RRivGab00178; Supplementary Data 1). Note that the same 18 individuals were consistently used for analyses which required medium- high depth and utilised genotype calls (D/f- branch statistics, PSMC, ROH, outgroup f3 and TT; described in Supplementary Data 1).  
+
+We have therefore added information in Methods ('Sequencing and mapping'; line 493) to reflect why these 18 individuals were chosen for specific analyses:  
+
+"As downstream analyses such as PSMC exhibit high variance when using lower- coverage samples ( \(\sim \leq 10x\) ) \(^{71}\) , we utilised a threshold of \(>12x\) depth of coverage to classify 18 unrelated individuals ( \(\geq 14x\) ) as medium- high depth samples within this study (Supplementary Data 1)."  
+
+In addition, we have added information to the figure legends in all main figures so that they include "medium- high depth ( \(\geq 14x\) )".  
+
+Reviewer #2 (Remarks to the Author):  
+
+Balboa et al present a fascinating study on the evolutionary history of African suids. This manuscript specifically interrogates the putative suture zone of red river hogs and bushpigs, including estimating levels of admixture and divergence times. In addition, the manuscript focusses on the Malagasy bushpig divergence, and how this likely relates to historic human wildlife translocations.  
+
+The study is well- presented, and appealing to a wide audience, with broader relevance to phylogenetic patterns in Africa, factors influencing speciation and human- influenced species distributions.  
+
+I have questions around some of the analyses/interpretations that I would like the authors to clarify before I can recommend publication:  
+
+Response: Thank you for taking the time to review our manuscript and for the positive and constructive comments.  
+
+## METHODS
+
+<--- Page Split --->
+
+The "Runs of homozygosity" section:  
+
+"at least two heterozygous reads to make a heterozygous call." - What do you mean heterozygous reads? Do you mean allelic balance of \(\geq 0.2?\) I.e. at least 2 alleles out of 10 being e.g. alternate?  
+
+Response: Yes, this was an error on our part; thank you for catching this. We meant at least two reads carrying each of the two alleles at heterozygous sites. We have now updated the main text in lines 644- 647 to:  
+
+"Runs of homozygosity (ROH) analyses were performed using PLINK v1.9 86. PLINK files included only filtered variable sites within medium- high depth samples \((n = 18)\) , with an additional depth filter (10 reads minimum) and at least two reads carrying each of the two alleles at heterozygous sites".  
+
+"SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded." - Why? So you are filtering out regions of high heterozygosity? Presumably because you are assuming that these correspond to repeat regions / duplications? Shouldn't this already be taken into account with your repeat masking that you have already done on the genome?  
+
+Response: Before performing downstream analyses, we removed sites where almost all individuals in our entire dataset were expected to be heterozygous \((F< - 0.9)\) as such sites were very likely error prone. As the ROH analysis is particularly sensitive to spurious heterozygous sites, where even a single heterozygous site will break up a ROH, this specific analysis uses a smaller subset of individuals (the 18 medium- high depth individuals) and a more stringent filter for the retention of sites with many heterozygous genotypes.  
+
+This is now reflected in the main text, where lines 650- 651 have been updated to:  
+
+"A few SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded since these were prone to genotype errors which can break up longer ROHS."  
+
+Have you considered a slightly more sophisticated program for identifying RoH's? E.g. RoHan? If you don't believe this is necessary/appropriate - why not?  
+
+Response: ROHan is a great software created by one of the co- authors on this paper, so we are glad that you have suggested it :- ). That being said, ROHan will most likely not give better results than PLINK for the high- depth individuals. However, it does allow us to infer ROHS for low- depth samples. So we agree that it is a useful analysis, providing a more complete picture of ROHS in the studied populations.  
+
+We have now updated the manuscript to include the analysis of all 67 individuals using ROHan, and have presented the results as part of Supplementary Figure S8. We observed similar trends to our previous ROH analyses. However, we caution that due to its window- based approach, ROHan has a hard time identifying shorter ROHS \((< 2Mb)\) and will underestimate the length of most ROHS. So compared to PLINK- based ROH analyses, the overall length of ROHS from ROHan is slightly shorter but the length of the longer ROHS
+
+<--- Page Split --->
+
+(>2Mb) is very similar. We have added illustrations of the ROHs inferred by ROHan on the Github repository for this study.  
+
+We have updated lines 290- 292 to include this information:  
+
+"Additionally, we explored ROH in all 67 samples, including low- depth samples with ROHan \(^{43}\) , which overall yielded similar results. However, due to its window- based approach, ROHan could not identify most of the shorter ROHs (<2 Mb), and therefore the overall \(F_{ROH}\) was slightly smaller using this approach (Supplementary Fig. S8). When considering both analyses, we observed heterogeneous levels of ROH within Uganda, Ethiopia, Cameroon and Equatorial Guinea likely driven by recent inbreeding events, leading to differences in longer ROHs."  
+
+TT and split time estimations  
+
+- This all sounds great, however, I have not used this method before and so it is difficult for me to judge the robustness of the analysis. The reference explains the theory of the approach, but some expansion of how the theory has been applied to this specific case would be very useful to the reader, for example in the supp methods? Or if not, at least a reference to a different paper that has used the approach and has given a more complete description.  
+
+Response: Thank you for reminding us that not all of the methods we used are widely known! TT and very similar methods estimating split times have been used in multiple studies, with the first variant of the method being introduced in Rasmussen et al. (2014) (ref. 92). The method was named TT in Schlebusch et al. (2017) (ref. 46), where it was used to estimate the divergence times for human populations, and the method paper we reference, Sjodin et al. (2021) (ref. 45), then came afterwards to describe and test the method in more detail. Subsequent studies have then further used this approach, including Larena et al. (2021) (ref. 89), Hollfelder et al. (2021) (ref. 90), Garcia- Erill et al. (2022) (ref. 4), Oliveira et al. (2023) (ref. 91) and Quinn et al. (2023) (ref. 93).  
+
+To make the interpretation easier for the reader, we have made a "tree" inside Figure 4b and have updated the figure legend to clarify what T1 and T2 describe. We have also updated the Methods section with additional detail (from line 667):  
+
+"Population split times were estimated using the Two- Two (TT) method \(^{45,46}\) . This approach estimates separate split times from a common ancestor for two different populations and has been used in various studies, including human \(^{46,89 - 91}\) , direct ancestry \(^{92}\) and animal studies \(^{4,93}\) . T1 describes the estimated time to the common ancestor for population 1 and T2 the time to the common ancestor for population 2. Further details for the method are described in Sjodin et al. \(^{45}\) and its relationship with other similar methods in Mualim et al. \(^{94}\) . For this analysis, we utilised the unfolded 2D- SFS from medium- high depth individuals polarised using several outgroups: against the common warthog, desert warthog (Supplementary Data 1) and the domestic pig (SRA: SAMN28197093). A mutation rate of \(1.49e^{-8}\) per site per generation and a generation time of six years were used to scale split times, as in our PSMC analyses \(^{4,42}\) ."  
+
+## RESULTS  
+
+"with only the Congo individuals being closer to the bushpigs than the other red river hogs"
+
+<--- Page Split --->
+
+- I initially read this as you saying that the Congo individuals were closer to bushpigs than red river hogs, but actually I think you mean "were closer to bushpigs than the other red river hogs, but overall still closer to red river hogs". Perhaps consider rewording it to clarify (as long as you can find a way that is better than my suggestion!)  
+
+Response: Yes, this is what we meant - that these individuals are closer to the bushpigs when compared to the other red river hogs. We have now reworded lines 165- 169 to the following (and hope that it is clearer!):  
+
+"Principal component analysis (PCA) revealed that the first two principal components exhibited a spatial distribution pattern reflecting the taxonomic and geographic origins of the sampled pigs (Fig. 1b). All the red river hog samples clustered together, yet the Congolese individuals were closer to the bushpigs than the other red river hogs. Malagasy samples formed a separate cluster from the other bushpigs."  
+
+In222- 229 "Given the observed FST and Dxy values..."  
+
+- In this paragraph you appear to shift back and forth when talking about between "species" ranges to within. When you then end with Uganda it is not clear which of these two you are talking about  
+
+Response: We have now clarified this by rewriting lines 222- 229 as:  
+
+"Given the observed \(F_{ST}\) and \(D_{xy}\) values, we explored spatial population structure and gene flow between and within the two species through estimating effective migration rates (Fig. 2b) 38. Between-species comparisons revealed a general barrier through the Central African rainforest and the East African Rift Valley, separating W/C and E/S populations. Within- species comparisons revealed high connectivity within both bushpig and red river hog ranges respectively, with the exception of Malagasy and non- Malagasy bushpigs which exhibited a barrier across the Mozambique Channel, particularly with the northernmost non- Malagasy populations. A decrease in effective migration in Ethiopia was also observed; this is in contrast with Uganda where we observed weak barriers, suggesting a possible corridor of gene flow connectivity within this region."  
+
+For the Distatic analysis, you state "Madagascar as H1" in the Figure 2c legend and "non- Malagasy bushpigs (H2)", but the figure shows "SA" as H1. Also in the text In242, you state "Madagascar as H1". Please clarify.  
+
+Response: Thank you for catching this - it was indeed an error on our part with respect to the main text and the Fig. 2c legend. We have corrected the main text (line 242; see below) and Fig. 2c (removed 'non- Malagasy') so that these only describe South Africa as H1.  
+
+"Similarly, to test for gene flow in the opposite direction, we performed similar tests with the southernmost bushpig population, South Africa as H1, each of the remaining red river hog populations as H2, and each of the bushpig populations as H3 (Fig. 2c; lower panel)."  
+
+Could the "strength" of the signal you detect, not just be reflecting the power of the analysis for each pairwise comparison, rather than the amount of gene- flow itself? Could it be worth
+
+<--- Page Split --->
+
+calculating f4 ratios to attempt to quantify the admixture proportions for these pairwise comparisons?  
+
+Response: We agree that the D- statistic value by itself is hard to interpret and does not translate well to determining the amount of gene flow. As suggested, we have calculated f4- ratios to quantify the amount of gene flow within Uganda and other populations (described below); these results show similar patterns to D- statistics analyses.  
+
+![](images/Figure_unknown_0.jpg)
+
+<center>Figure above (now Supplementary Fig. S7): Estimation of admixture proportions (f4-ratios) between populations. </center>  
+
+f4- ratios were calculated using the common warthog as an outgroup, constructed as f4(A,O;X,C)/f4(A,O;B,C). a) f4- ratios into Uganda from non- Ghanaian red river hog populations (B), of the form (f4(Ghana, Warthog; Uganda, South Africa)/f4(Ghana, Warthog; B, South Africa). b) f4- ratios into Ethiopia from non- Ghanaian red river hog populations (B) (f4(Ghana, Warthog; Ethiopia, South Africa)/f4(Ghana, Warthog; B, South Africa). c) f4- ratios into bushpig
+
+<--- Page Split --->
+
+populations (X), using Equatorial Guinea as a proxy. \(\alpha\) indicates estimated admixture proportions from B' into X (f4(Ghana, Warthog; X, South Africa)/f4(Ghana, Warthog; Eq Guinea, South Africa)). Error bars represent \(\pm\) three standard errors from the estimated f4- ratio. SA - South Africa; EqG - Equatorial Guinea.  
+
+In the figure above, we describe three scenarios: a) where we estimate admixture proportions into Uganda from non- Ghanaian red river hog populations, b) where we estimate admixture proportions into Ethiopia from non- Ghanaian red river hog populations, and c) where we estimate admixture proportions into each of the bushpig populations from Equatorial Guinea red river hogs.  
+
+As can be seen in the first scenario (a), we show that there is an increase in f4- ratios for the Uganda population as one moves from the westernmost part of Africa (Togo; \(10.9\%\) ) to the most central (Gabon; \(21.1\%\) ), in line with our hypothesis that Uganda is a potential zone of introgression and in agreement with the large signals observed in D- statistics comparisons (Fig. 2c). This is similarly seen for Ethiopia (b), with estimated admixture proportions of \(6.8\%\) to \(13.2\%\) , from west (Togo) to central (Gabon), respectively.  
+
+In the third scenario (c), we observe high levels of gene flow into Uganda and Ethiopia ( \(19.7\%\) and \(12.2\%\) , respectively) and much lower (albeit significant) levels in Tanzania and Zimbabwe, supporting our results in the manuscript.  
+
+Overall, these results, combined with the D- statistics, suggest significant gene flow into Uganda and Ethiopia from red river hog populations or one or more populations closely related to them that were not included in this study. We have added the plot to Supplementary Information as Supplementary Fig. S7.  
+
+Although the analysis infers \(\sim 20\%\) gene flow, it is worth noting that there are assumptions that the populations considered had evolved as a perfectly bifurcating tree and that the only gene flow event is the one that is modelled. As suggested by our \(f\) - branch ( \(f_{b}\) ) results (Supplementary Fig. S6), the relationship between the populations is likely much more complicated, and therefore one should be cautious in interpreting these results as evidence for a single point migration event. The amount of gene flow could be much higher from an unsampled population or it could be many migration events happening between several populations at different time points.  
+
+In light of these results, we have added the following to the main text (line 247):  
+
+"Furthermore, we estimated the amount of gene flow between species based on \(f_{4}\) - ratios, under assumptions that the populations considered had evolved together as a perfectly bifurcating tree and that the only gene flow event is the one that is modelled. This analysis estimated up to \(21\%\) gene flow from red river hogs into Uganda and up to \(13\%\) into Ethiopia, with increasing signals in more central populations (Supplementary Fig. S7). Given the complicated history of these populations, as suggested by our \(f\) - branch ( \(f_{b}\) ) results (Supplementary Fig. S6), these values are unlikely to accurately represent the historical gene flow events which likely occurred between multiple populations at different timepoints. Nevertheless, when considering all three analyses, these results suggest that there is or has been gene flow between the two taxa currently identified as species, and that the gradient of
+
+<--- Page Split --->
+
+allele sharing between them is consistent with isolation by distance, where genetic similarity is strongest in populations from Central Africa."  
+
+"Given the results reported above, the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of introgression between the two taxa."  
+
+- This seems a bit speculative to me. The PSMC results are interesting, and visually appealing, and therefore I do think that they add to the story, however they are difficult to interpret in terms of any kind of specific hypothesis. Consider re-wording or removing part of this section.  
+
+Response: The above statement was not meant to be purely based on PSMC; rather, this was meant to be in light of the D-statistics (and now \(f_{4}\) - ratio) results that were described before this section. We have therefore revised the text to reflect this (lines 263- 264):  
+
+"These results, in combination with the D- statistics and \(f_{4}\) - ratio results reported above, suggest that the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of possible introgression between the two taxa."  
+
+"This analysis suggested that the Malagasy population experienced a severe bottleneck, likely a result of a founder event between 1- 5 kya."  
+
+- The decline seems to start quite some time before this though - why might this be?  
+
+Response: The decline in effective population size does appear to start in the interval \(\sim 5\) - 8kya where it drops to \(\sim 20,000\) individuals, followed by the large decline to \(\sim 1,000\) individuals around 1kya. When replying to this comment, we realised how difficult it was to see when the decline happened due to the log scale, so we have subsequently updated the figure with additional tick marks. It is hard to say whether the early start of the decline is due to a decline in the population in mainland Africa prior to colonisation of Madagascar or whether it is a methodological issue; this is now reflected in the main text.  
+
+We have rewritten the relevant part of the Discussion (lines 418- 429) as:  
+
+"Our popSizeABC estimate of an effective population size of 1,000 individuals during the bottleneck 1,500 years ago is surprisingly high, assuming that the founder event was a single occurrence involving a limited number of individuals carried to Madagascar by ship. However, the estimate is supported by their observed heterozygosity level, which is similar to levels observed in southern Africa, although we cannot know to what extent southern African populations have been subjected to drift since the founding of the Malagasy population. Multiple introductions, spanning over a longer period of time, or even with animals sourced from different mainland populations (including already admixed individuals), may have inflated this estimate. In addition, an instant large drop in population size followed by subsequent regrowth will be estimated by popSizeABC as a more gradual decline that starts earlier \(^{46}\) , making it difficult to pinpoint the exact timing of when the bottleneck started."  
+
+## DISCUSSION  
+
+"had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current 63. However, from the PCA, NGAadmix and IBS tree we
+
+<--- Page Split --->
+
+did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+- To?: "...North Mozambique current 63. However, while we were unable to explicitly test this hypothesis with our dataset (due to no samples from Comoros etc), PCA NGAdmix and the IBS tree did not." ?  
+
+Response: We agree that we should have made it clear that we could not have made statements about bushpig populations in the Comoros Island, since we do not have data from that area. We have therefore incorporated your suggestion and have changed lines 446- 449 to:  
+
+"...had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current \(^{67}\) . Although we were unable to explicitly test this hypothesis without samples from the Comoros, our results from the PCA, NGSadmix and IBS tree do not suggest multiple pulses into Madagascar and did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+Reviewer #3 (Remarks to the Author):  
+
+The paper is clearly written and illustrated, though as an archaeologist I am in no position to pass judgement on the detailed methodology of the genetic study undertaken. That said, references to the debate surrounding the timing of human settlement of Madagascar are fair and the conclusion that bushpigs were likely introduced there \(\sim 1000 - 1500\) years ago does indeed fit extremely well with: a) dated faunal remains from the island, not just of bushpig but also of other exotic mammal taxa; b) the oldest archaeologically unambiguous, well- dated evidence for human presence; c) and specifically the African origins (and names) of domestic mammals and evidence for Triangular Incised Ware (Tana Ware), a form of pottery found in late first- millennium AD contexts in the southwest of the island - see Parker Pearson et al. 2010: 79 (Pastoralists, Warriors and Colonists: The Archaeology of Southern Madagascar; Oxford: Archaeopress).  
+
+For further discussion/reference of the timing of human settlement I would recommend Mitchell, Journal of Island and Coastal Archaeology 2020, arguments further elaborated in his 2022 book African Islands: A Comparative Archaeology (London: Routledge).  
+
+Provided that the genetic analysis is sound, I recommend acceptance.  
+
+Response: Thank you for spending time reviewing our manuscript and for providing us with important references to include. To further guide the discussion about human arrival, we have now incorporated both the Mitchell (2020) and Parker Pearson (2010) references in our Discussion, in line 434:  
+
+"...most likely through populations that started to arrive on Madagascar from southeastern Africa at least 1,500 years ago and possibly earlier \(^{26}\) , although the latter has been contested \(^{64}\) . Triangular Incised Ware (Tana Ware) style pottery remains were found in human settlements in southern Madagascar dated to 1,000- 1,400 years ago, suggesting either contact with or colonisation from the African Swahili region at this time \(^{65}\) ."
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+I already only had small number of minor comments on the original manuscript, and these have all been sufficiently addressed in the revised manuscript.  
+
+Reviewer #2:  
+
+Remarks to the Author:  
+
+All my comments/questions have been addressed. The authors should congratulate themselves on producing an well- written and important piece of work!
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+I already only had small number of minor comments on the original manuscript, and these have all been sufficiently addressed in the revised manuscript.  
+
+Response: We are pleased to hear that – thank you for your time and effort in reviewing our manuscript!  
+
+Reviewer #2 (Remarks to the Author):  
+
+All my comments/questions have been addressed. The authors should congratulate themselves on producing an well- written and important piece of work!  
+
+Response: We appreciate the kind comment and for taking the time to provide positive and constructive comments to help improve our manuscript!
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1da3cc07656e2b2ad084cb5e938a2f126f69f88db62135e01903242c34fe8e44/supplementary_0_Peer Review File__1da3cc07656e2b2ad084cb5e938a2f126f69f88db62135e01903242c34fe8e44_det.mmd

@@ -0,0 +1,451 @@
+<|ref|>title<|/ref|><|det|>[[99, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[106, 110, 372, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[106, 154, 828, 211]]<|/det|>
+African bushpigs exhibit porous species boundaries and appeared in Madagascar concurrently with human arrival  
+
+<|ref|>text<|/ref|><|det|>[[270, 733, 880, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 104]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[116, 119, 216, 133]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[116, 135, 290, 148]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 149, 880, 327]]<|/det|>
+The manuscript on African bush pigs is a very detailed and thorough study of the two Potamochoerus species in Africa and based on the complete genome sequence of 67 individuals from across Africa, including Madagascar. The authors have used state of the art methodology to study the population genetics, speciation and admixture of these species. I have few comments on this manuscript which describes a very coherent and convincing story about these two species. In their discussion, the authors briefly tough upon the ongoing debate about the taxonomy of Potamochoerus and I agree with their comment that this very much depends on the species concept used. In fact, the species concept reflects the urge of humans to categorize, while the past decades have increasingly shown the diffuse borders between species with ongoing hybridization between many populations assumed to represent different species. I also welcomed their cautionary remark about the interpretation of PSMC curves and the cautionary remark about the assumption of total panmixia which indeed not always might be correct.  
+
+<|ref|>text<|/ref|><|det|>[[116, 343, 245, 357]]<|/det|>
+Minor comments:  
+
+<|ref|>text<|/ref|><|det|>[[116, 358, 832, 432]]<|/det|>
+Lines 201- 203. I found the comment about the lowest Dxy between Ugandan and Congolese populations because of the closest geographical distance between these populations not very convincing. The distance between several others (e.g. Gabon) as well as the actual Dxy values between Ugandan and several others were not that very much different. Consider removing this sentence.  
+
+<|ref|>text<|/ref|><|det|>[[116, 447, 875, 507]]<|/det|>
+Figure 3c. The authors do not very clearly indicate why only these 18 individuals are shown. Although the authors mention it is based on read depth, they could have been a bit clearer and also might have used a read depth of \(>10x\) as the threshold and shown the 23 individuals that are above that threshold.  
+
+<|ref|>text<|/ref|><|det|>[[116, 551, 216, 565]]<|/det|>
+Reviewer #2:  
+
+<|ref|>text<|/ref|><|det|>[[116, 567, 290, 580]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[116, 581, 870, 641]]<|/det|>
+Balboa et al present a fascinating study on the evolutionary history of African suids. This manuscript specifically interrogates the putative suture zone of red river hogs and bushpis, including estimating levels of admixture and divergence times. In addition, the manuscript focusses on the Malagasy bushpig divergence, and how this likely relates to historic human wildlife translocations.  
+
+<|ref|>text<|/ref|><|det|>[[116, 655, 877, 686]]<|/det|>
+The study is well- presented, and appealing to a wide audience, with broader relevance to phylogenetic patterns in Africa, factors influencing speciation and human- influenced species distributions.  
+
+<|ref|>text<|/ref|><|det|>[[116, 700, 861, 730]]<|/det|>
+I have questions around some of the analyses/interpretations that I would like the authors to clarify before I can recommend publication:  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 744, 192, 758]]<|/det|>
+## METHODS  
+
+<|ref|>text<|/ref|><|det|>[[117, 774, 388, 789]]<|/det|>
+The "Runs of homozygosity" section:  
+
+<|ref|>text<|/ref|><|det|>[[116, 804, 580, 819]]<|/det|>
+"at least two heterozygous reads to make a heterozygous call."  
+
+<|ref|>text<|/ref|><|det|>[[116, 820, 870, 850]]<|/det|>
+- What do you mean heterozygous reads? Do you mean allelic balance of \(\geq 0.2\) ? I.e. at least 2 alleles out of 10 being e.g. alternate?  
+
+<|ref|>text<|/ref|><|det|>[[116, 864, 875, 895]]<|/det|>
+"SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded."- Why? So you are filtering out regions of high heterozygosity? Presumably because you are assuming
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 860, 120]]<|/det|>
+that these correspond to repeat regions / duplications? Shouldn't this already be taken into account with your repeat masking that you have already done on the genome?  
+
+<|ref|>text<|/ref|><|det|>[[115, 134, 863, 164]]<|/det|>
+Have you considered a slightly more sophisticated program for identifying RoH's? E.g. RoHan? If you don't believe this is necessary/appropriate - why not?  
+
+<|ref|>text<|/ref|><|det|>[[116, 180, 332, 194]]<|/det|>
+TT and split time estimations  
+
+<|ref|>text<|/ref|><|det|>[[115, 195, 875, 269]]<|/det|>
+- This all sounds great, however, I have not used this method before and so it is difficult for me to judge the robustness of the analysis. The reference explains the theory of the approach, but some expansion of how the theory has been applied to this specific case would be very useful to the reader, for example in the supp methods? Or if not, at least a reference to a different paper that has used the approach and has given a more complete description.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 283, 184, 297]]<|/det|>
+## RESULTS  
+
+<|ref|>text<|/ref|><|det|>[[115, 312, 876, 388]]<|/det|>
+"with only the Congo individuals being closer to the bushpigs than the other red river hogs" - I initially read this as you saying that the Congo individuals were closer to bushpigs than red river hogs, but actually I think you mean "were closer to bushpigs than the other red river hogs, but overall still closer to red river hogs". Perhaps consider rewording it to clarify (as long as you can find a way that is better than my suggestion!)  
+
+<|ref|>text<|/ref|><|det|>[[115, 402, 523, 416]]<|/det|>
+In222- 229 "Given the observed FST and Dxy values..."  
+
+<|ref|>text<|/ref|><|det|>[[115, 417, 875, 446]]<|/det|>
+- In this paragraph you appear to shift back and forth when talking about between "species" ranges to within. When you then end with Uganda it is not clear which of these two you are talking about  
+
+<|ref|>text<|/ref|><|det|>[[115, 461, 865, 506]]<|/det|>
+For the Dstatistic analysis, you state "Madagascar as H1" in the Figure 2c legend and "non- Malagasy bushpigs (H2)", but the figure shows "SA" as H1. Also in the text In242, you state "Madagascar as H1". Please clarify.  
+
+<|ref|>text<|/ref|><|det|>[[115, 521, 871, 566]]<|/det|>
+Could the "strength" of the signal you detect, not just be reflecting the power of the analysis for each pairwise comparison, rather than the amount of gene- flow itself? Could it be worth calculating f4 ratios to attempt to quantify the admixture proportions for these pairwise comparisons?  
+
+<|ref|>text<|/ref|><|det|>[[115, 581, 870, 655]]<|/det|>
+"Given the results reported above, the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of introgression between the two taxa." - This seems a bit speculative to me. The PSMC results are interesting, and visually appealing, and therefore I do think that they add to the story, however they are difficult to interpret in terms of any kind of specific hypothesis. Consider re- wording or removing part of this section.  
+
+<|ref|>text<|/ref|><|det|>[[115, 670, 872, 700]]<|/det|>
+"This analysis suggested that the Malagasy population experienced a severe bottleneck, likely a result of a founder event between 1- 5 kya."  
+
+<|ref|>text<|/ref|><|det|>[[115, 700, 750, 715]]<|/det|>
+- The decline seems to start quite some time before this though - why might this be?  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 730, 214, 744]]<|/det|>
+## DISCUSSION  
+
+<|ref|>text<|/ref|><|det|>[[115, 759, 866, 818]]<|/det|>
+"had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current 63. However, from the PCA, NGAadmix and IBS tree we did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+<|ref|>text<|/ref|><|det|>[[115, 820, 857, 864]]<|/det|>
+- To?: "...North Mozambique current 63. However, while we were unable to explicitly test this hypothesis with our dataset (due to no samples from Comoros etc), PCA NGAadmix and the IBS tree did not..."?
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 103, 216, 117]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[115, 119, 291, 132]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 133, 881, 283]]<|/det|>
+The paper is clearly written and illustrated, though as an archaeologist I am in no position to pass judgement on the detailed methodology of the genetic study undertaken. That said, references to the debate surrounding the timing of human settlement of Madagascar are fair and the conclusion that bushpigs were likely introduced there \(\sim 1000 - 1500\) years ago does indeed fit extremely well with: a) dated faunal remains from the island, not just of bushpig but also of other exotic mammal taxa; b) the oldest archaeologically unambiguous, well- dated evidence for human presence; c) and specifically the African origins (and names) of domestic mammals and evidence for Triangular Incised Ware (Tana Ware), a form of pottery found in late first- millennium AD contexts in the southwest of the island - see Parker Pearson et al. 2010: 79 (Pastoralists, Warriors and Colonists: The Archaeology of Southern Madagascar; Oxford: Archaeopress).  
+
+<|ref|>text<|/ref|><|det|>[[116, 297, 833, 342]]<|/det|>
+For further discussion/reference of the timing of human settlement I would recommend Mitchell, Journal of Island and Coastal Archaeology 2020, arguments further elaborated in his 2022 book African Islands: A Comparative Archaeology (London: Routledge).  
+
+<|ref|>text<|/ref|><|det|>[[115, 356, 635, 371]]<|/det|>
+Provided that the genetic analysis is sound, I recommend acceptance.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[118, 106, 540, 128]]<|/det|>
+## Response to reviewer comments  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 155, 335, 170]]<|/det|>
+## A note to the Reviewers:  
+
+<|ref|>text<|/ref|><|det|>[[118, 172, 880, 259]]<|/det|>
+We thank Reviewers #1, #2 and #3 for taking time to review our manuscript and for the constructive and positive feedback that they have provided. We have detailed our point- by- point responses in red below and have provided a revised version of our manuscript. Line numbers in this response document refer to the line numbers in the original submitted manuscript.  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 275, 331, 292]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[119, 310, 430, 326]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[117, 344, 881, 567]]<|/det|>
+The manuscript on African bush pigs is a very detailed and thorough study of the two Potamochoerus species in Africa and based on the complete genome sequence of 67 individuals from across Africa, including Madagascar. The authors have used state of the art methodology to study the population genetics, speciation and admixture of these species. I have few comments on this manuscript which describes a very coherent and convincing story about these two species. In their discussion, the authors briefly tough upon the ongoing debate about the taxonomy of Potamochoerus and I agree with their comment that this very much depends on the species concept used. In fact, the species concept reflects the urge of humans to categorize, while the past decades have increasingly shown the diffuse borders between species with ongoing hybridization between many populations assumed to represent different species. I also welcomed their cautionary remark about the interpretation of PSMC curves and the cautionary remark about the assumption of total panmixia which indeed not always might be correct.  
+
+<|ref|>text<|/ref|><|det|>[[118, 586, 880, 620]]<|/det|>
+Response: Thank you for taking the time to review our manuscript and for your positive comments.  
+
+<|ref|>text<|/ref|><|det|>[[118, 639, 261, 654]]<|/det|>
+Minor comments:  
+
+<|ref|>text<|/ref|><|det|>[[118, 656, 880, 740]]<|/det|>
+Lines 201- 203. I found the comment about the lowest Dxy between Ugandan and Congolese populations because of the closest geographical distance between these populations not very convincing. The distance between several others (e.g. Gabon) as well as the actual Dxy values between Ugandan and several others were not that very much different. Consider removing this sentence.  
+
+<|ref|>text<|/ref|><|det|>[[118, 758, 880, 809]]<|/det|>
+Response: Thank you for your comment. We have updated lines 200- 203 to be more precise in our description, where we include values and describe between- species \(D_{xy}\) relative to within- species \(D_{xy}\) between DR Congo and Ghana, as below:  
+
+<|ref|>text<|/ref|><|det|>[[118, 825, 880, 911]]<|/det|>
+"When comparing across species, \(D_{xy}\) was lowest for populations that were geographically central, with the Ugandan population exhibiting the lowest between- species \(D_{xy}\) for all bushpigs and the Congolese the lowest \(D_{xy}\) for all red river hogs. Additionally, the lowest \(D_{xy}\) between species was observed between the Ugandan and Congolese populations (0.00355), similar to within- species \(D_{xy}\) comparisons for Ghana and DR Congo (0.00352)."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 83, 880, 152]]<|/det|>
+Figure 3c. The authors do not very clearly indicate why only these 18 individuals are shown. Although the authors mention it is based on read depth, they could have been a bit clearer and also might have used a read depth of \(>10x\) as the threshold and shown the 23 individuals that are above that threshold.  
+
+<|ref|>text<|/ref|><|det|>[[118, 174, 880, 192]]<|/det|>
+Response: We agree that we were unclear in our description of medium- high depth samples.  
+
+<|ref|>text<|/ref|><|det|>[[118, 200, 881, 336]]<|/det|>
+We used a minimum of 12X as our threshold since PSMC curves exhibit high variance when using samples \(\leq 10x\) in combination with missing data (Nadachowska- Brzyska et al. (2016); ref. 71). As a result, we had 18 unrelated individuals that met this threshold, where the sample with the lowest coverage was 14X (RRivGab00178; Supplementary Data 1). Note that the same 18 individuals were consistently used for analyses which required medium- high depth and utilised genotype calls (D/f- branch statistics, PSMC, ROH, outgroup f3 and TT; described in Supplementary Data 1).  
+
+<|ref|>text<|/ref|><|det|>[[118, 352, 880, 387]]<|/det|>
+We have therefore added information in Methods ('Sequencing and mapping'; line 493) to reflect why these 18 individuals were chosen for specific analyses:  
+
+<|ref|>text<|/ref|><|det|>[[118, 408, 882, 479]]<|/det|>
+"As downstream analyses such as PSMC exhibit high variance when using lower- coverage samples ( \(\sim \leq 10x\) ) \(^{71}\) , we utilised a threshold of \(>12x\) depth of coverage to classify 18 unrelated individuals ( \(\geq 14x\) ) as medium- high depth samples within this study (Supplementary Data 1)."  
+
+<|ref|>text<|/ref|><|det|>[[118, 500, 880, 539]]<|/det|>
+In addition, we have added information to the figure legends in all main figures so that they include "medium- high depth ( \(\geq 14x\) )".  
+
+<|ref|>text<|/ref|><|det|>[[119, 560, 430, 577]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 594, 880, 681]]<|/det|>
+Balboa et al present a fascinating study on the evolutionary history of African suids. This manuscript specifically interrogates the putative suture zone of red river hogs and bushpigs, including estimating levels of admixture and divergence times. In addition, the manuscript focusses on the Malagasy bushpig divergence, and how this likely relates to historic human wildlife translocations.  
+
+<|ref|>text<|/ref|><|det|>[[118, 699, 880, 750]]<|/det|>
+The study is well- presented, and appealing to a wide audience, with broader relevance to phylogenetic patterns in Africa, factors influencing speciation and human- influenced species distributions.  
+
+<|ref|>text<|/ref|><|det|>[[118, 768, 880, 801]]<|/det|>
+I have questions around some of the analyses/interpretations that I would like the authors to clarify before I can recommend publication:  
+
+<|ref|>text<|/ref|><|det|>[[118, 819, 880, 852]]<|/det|>
+Response: Thank you for taking the time to review our manuscript and for the positive and constructive comments.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 871, 213, 887]]<|/det|>
+## METHODS
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[119, 84, 421, 100]]<|/det|>
+The "Runs of homozygosity" section:  
+
+<|ref|>text<|/ref|><|det|>[[117, 118, 884, 181]]<|/det|>
+"at least two heterozygous reads to make a heterozygous call." - What do you mean heterozygous reads? Do you mean allelic balance of \(\geq 0.2?\) I.e. at least 2 alleles out of 10 being e.g. alternate?  
+
+<|ref|>text<|/ref|><|det|>[[118, 204, 880, 255]]<|/det|>
+Response: Yes, this was an error on our part; thank you for catching this. We meant at least two reads carrying each of the two alleles at heterozygous sites. We have now updated the main text in lines 644- 647 to:  
+
+<|ref|>text<|/ref|><|det|>[[118, 273, 880, 342]]<|/det|>
+"Runs of homozygosity (ROH) analyses were performed using PLINK v1.9 86. PLINK files included only filtered variable sites within medium- high depth samples \((n = 18)\) , with an additional depth filter (10 reads minimum) and at least two reads carrying each of the two alleles at heterozygous sites".  
+
+<|ref|>text<|/ref|><|det|>[[118, 360, 880, 429]]<|/det|>
+"SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded." - Why? So you are filtering out regions of high heterozygosity? Presumably because you are assuming that these correspond to repeat regions / duplications? Shouldn't this already be taken into account with your repeat masking that you have already done on the genome?  
+
+<|ref|>text<|/ref|><|det|>[[118, 446, 880, 550]]<|/det|>
+Response: Before performing downstream analyses, we removed sites where almost all individuals in our entire dataset were expected to be heterozygous \((F< - 0.9)\) as such sites were very likely error prone. As the ROH analysis is particularly sensitive to spurious heterozygous sites, where even a single heterozygous site will break up a ROH, this specific analysis uses a smaller subset of individuals (the 18 medium- high depth individuals) and a more stringent filter for the retention of sites with many heterozygous genotypes.  
+
+<|ref|>text<|/ref|><|det|>[[118, 567, 784, 583]]<|/det|>
+This is now reflected in the main text, where lines 650- 651 have been updated to:  
+
+<|ref|>text<|/ref|><|det|>[[118, 601, 880, 636]]<|/det|>
+"A few SNP sites with \(>50\%\) heterozygous genotypes across individuals were also excluded since these were prone to genotype errors which can break up longer ROHS."  
+
+<|ref|>text<|/ref|><|det|>[[118, 653, 880, 688]]<|/det|>
+Have you considered a slightly more sophisticated program for identifying RoH's? E.g. RoHan? If you don't believe this is necessary/appropriate - why not?  
+
+<|ref|>text<|/ref|><|det|>[[118, 705, 880, 790]]<|/det|>
+Response: ROHan is a great software created by one of the co- authors on this paper, so we are glad that you have suggested it :- ). That being said, ROHan will most likely not give better results than PLINK for the high- depth individuals. However, it does allow us to infer ROHS for low- depth samples. So we agree that it is a useful analysis, providing a more complete picture of ROHS in the studied populations.  
+
+<|ref|>text<|/ref|><|det|>[[118, 808, 880, 911]]<|/det|>
+We have now updated the manuscript to include the analysis of all 67 individuals using ROHan, and have presented the results as part of Supplementary Figure S8. We observed similar trends to our previous ROH analyses. However, we caution that due to its window- based approach, ROHan has a hard time identifying shorter ROHS \((< 2Mb)\) and will underestimate the length of most ROHS. So compared to PLINK- based ROH analyses, the overall length of ROHS from ROHan is slightly shorter but the length of the longer ROHS
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 83, 879, 117]]<|/det|>
+(>2Mb) is very similar. We have added illustrations of the ROHs inferred by ROHan on the Github repository for this study.  
+
+<|ref|>text<|/ref|><|det|>[[118, 135, 600, 151]]<|/det|>
+We have updated lines 290- 292 to include this information:  
+
+<|ref|>text<|/ref|><|det|>[[118, 169, 881, 274]]<|/det|>
+"Additionally, we explored ROH in all 67 samples, including low- depth samples with ROHan \(^{43}\) , which overall yielded similar results. However, due to its window- based approach, ROHan could not identify most of the shorter ROHs (<2 Mb), and therefore the overall \(F_{ROH}\) was slightly smaller using this approach (Supplementary Fig. S8). When considering both analyses, we observed heterogeneous levels of ROH within Uganda, Ethiopia, Cameroon and Equatorial Guinea likely driven by recent inbreeding events, leading to differences in longer ROHs."  
+
+<|ref|>text<|/ref|><|det|>[[118, 292, 356, 307]]<|/det|>
+TT and split time estimations  
+
+<|ref|>text<|/ref|><|det|>[[118, 309, 881, 394]]<|/det|>
+- This all sounds great, however, I have not used this method before and so it is difficult for me to judge the robustness of the analysis. The reference explains the theory of the approach, but some expansion of how the theory has been applied to this specific case would be very useful to the reader, for example in the supp methods? Or if not, at least a reference to a different paper that has used the approach and has given a more complete description.  
+
+<|ref|>text<|/ref|><|det|>[[117, 412, 881, 565]]<|/det|>
+Response: Thank you for reminding us that not all of the methods we used are widely known! TT and very similar methods estimating split times have been used in multiple studies, with the first variant of the method being introduced in Rasmussen et al. (2014) (ref. 92). The method was named TT in Schlebusch et al. (2017) (ref. 46), where it was used to estimate the divergence times for human populations, and the method paper we reference, Sjodin et al. (2021) (ref. 45), then came afterwards to describe and test the method in more detail. Subsequent studies have then further used this approach, including Larena et al. (2021) (ref. 89), Hollfelder et al. (2021) (ref. 90), Garcia- Erill et al. (2022) (ref. 4), Oliveira et al. (2023) (ref. 91) and Quinn et al. (2023) (ref. 93).  
+
+<|ref|>text<|/ref|><|det|>[[118, 583, 880, 635]]<|/det|>
+To make the interpretation easier for the reader, we have made a "tree" inside Figure 4b and have updated the figure legend to clarify what T1 and T2 describe. We have also updated the Methods section with additional detail (from line 667):  
+
+<|ref|>text<|/ref|><|det|>[[117, 653, 881, 824]]<|/det|>
+"Population split times were estimated using the Two- Two (TT) method \(^{45,46}\) . This approach estimates separate split times from a common ancestor for two different populations and has been used in various studies, including human \(^{46,89 - 91}\) , direct ancestry \(^{92}\) and animal studies \(^{4,93}\) . T1 describes the estimated time to the common ancestor for population 1 and T2 the time to the common ancestor for population 2. Further details for the method are described in Sjodin et al. \(^{45}\) and its relationship with other similar methods in Mualim et al. \(^{94}\) . For this analysis, we utilised the unfolded 2D- SFS from medium- high depth individuals polarised using several outgroups: against the common warthog, desert warthog (Supplementary Data 1) and the domestic pig (SRA: SAMN28197093). A mutation rate of \(1.49e^{-8}\) per site per generation and a generation time of six years were used to scale split times, as in our PSMC analyses \(^{4,42}\) ."  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 842, 206, 858]]<|/det|>
+## RESULTS  
+
+<|ref|>text<|/ref|><|det|>[[118, 877, 860, 894]]<|/det|>
+"with only the Congo individuals being closer to the bushpigs than the other red river hogs"
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 83, 880, 153]]<|/det|>
+- I initially read this as you saying that the Congo individuals were closer to bushpigs than red river hogs, but actually I think you mean "were closer to bushpigs than the other red river hogs, but overall still closer to red river hogs". Perhaps consider rewording it to clarify (as long as you can find a way that is better than my suggestion!)  
+
+<|ref|>text<|/ref|><|det|>[[118, 169, 880, 222]]<|/det|>
+Response: Yes, this is what we meant - that these individuals are closer to the bushpigs when compared to the other red river hogs. We have now reworded lines 165- 169 to the following (and hope that it is clearer!):  
+
+<|ref|>text<|/ref|><|det|>[[118, 238, 881, 325]]<|/det|>
+"Principal component analysis (PCA) revealed that the first two principal components exhibited a spatial distribution pattern reflecting the taxonomic and geographic origins of the sampled pigs (Fig. 1b). All the red river hog samples clustered together, yet the Congolese individuals were closer to the bushpigs than the other red river hogs. Malagasy samples formed a separate cluster from the other bushpigs."  
+
+<|ref|>text<|/ref|><|det|>[[118, 341, 568, 358]]<|/det|>
+In222- 229 "Given the observed FST and Dxy values..."  
+
+<|ref|>text<|/ref|><|det|>[[118, 360, 880, 411]]<|/det|>
+- In this paragraph you appear to shift back and forth when talking about between "species" ranges to within. When you then end with Uganda it is not clear which of these two you are talking about  
+
+<|ref|>text<|/ref|><|det|>[[118, 428, 680, 445]]<|/det|>
+Response: We have now clarified this by rewriting lines 222- 229 as:  
+
+<|ref|>text<|/ref|><|det|>[[118, 462, 881, 637]]<|/det|>
+"Given the observed \(F_{ST}\) and \(D_{xy}\) values, we explored spatial population structure and gene flow between and within the two species through estimating effective migration rates (Fig. 2b) 38. Between-species comparisons revealed a general barrier through the Central African rainforest and the East African Rift Valley, separating W/C and E/S populations. Within- species comparisons revealed high connectivity within both bushpig and red river hog ranges respectively, with the exception of Malagasy and non- Malagasy bushpigs which exhibited a barrier across the Mozambique Channel, particularly with the northernmost non- Malagasy populations. A decrease in effective migration in Ethiopia was also observed; this is in contrast with Uganda where we observed weak barriers, suggesting a possible corridor of gene flow connectivity within this region."  
+
+<|ref|>text<|/ref|><|det|>[[118, 653, 880, 705]]<|/det|>
+For the Distatic analysis, you state "Madagascar as H1" in the Figure 2c legend and "non- Malagasy bushpigs (H2)", but the figure shows "SA" as H1. Also in the text In242, you state "Madagascar as H1". Please clarify.  
+
+<|ref|>text<|/ref|><|det|>[[118, 721, 880, 773]]<|/det|>
+Response: Thank you for catching this - it was indeed an error on our part with respect to the main text and the Fig. 2c legend. We have corrected the main text (line 242; see below) and Fig. 2c (removed 'non- Malagasy') so that these only describe South Africa as H1.  
+
+<|ref|>text<|/ref|><|det|>[[118, 790, 880, 842]]<|/det|>
+"Similarly, to test for gene flow in the opposite direction, we performed similar tests with the southernmost bushpig population, South Africa as H1, each of the remaining red river hog populations as H2, and each of the bushpig populations as H3 (Fig. 2c; lower panel)."  
+
+<|ref|>text<|/ref|><|det|>[[118, 860, 880, 894]]<|/det|>
+Could the "strength" of the signal you detect, not just be reflecting the power of the analysis for each pairwise comparison, rather than the amount of gene- flow itself? Could it be worth
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 84, 880, 117]]<|/det|>
+calculating f4 ratios to attempt to quantify the admixture proportions for these pairwise comparisons?  
+
+<|ref|>text<|/ref|><|det|>[[118, 135, 881, 204]]<|/det|>
+Response: We agree that the D- statistic value by itself is hard to interpret and does not translate well to determining the amount of gene flow. As suggested, we have calculated f4- ratios to quantify the amount of gene flow within Uganda and other populations (described below); these results show similar patterns to D- statistics analyses.  
+
+<|ref|>image<|/ref|><|det|>[[123, 214, 875, 781]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[118, 786, 880, 820]]<|/det|>
+<center>Figure above (now Supplementary Fig. S7): Estimation of admixture proportions (f4-ratios) between populations. </center>  
+
+<|ref|>text<|/ref|><|det|>[[118, 821, 881, 908]]<|/det|>
+f4- ratios were calculated using the common warthog as an outgroup, constructed as f4(A,O;X,C)/f4(A,O;B,C). a) f4- ratios into Uganda from non- Ghanaian red river hog populations (B), of the form (f4(Ghana, Warthog; Uganda, South Africa)/f4(Ghana, Warthog; B, South Africa). b) f4- ratios into Ethiopia from non- Ghanaian red river hog populations (B) (f4(Ghana, Warthog; Ethiopia, South Africa)/f4(Ghana, Warthog; B, South Africa). c) f4- ratios into bushpig
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 83, 880, 152]]<|/det|>
+populations (X), using Equatorial Guinea as a proxy. \(\alpha\) indicates estimated admixture proportions from B' into X (f4(Ghana, Warthog; X, South Africa)/f4(Ghana, Warthog; Eq Guinea, South Africa)). Error bars represent \(\pm\) three standard errors from the estimated f4- ratio. SA - South Africa; EqG - Equatorial Guinea.  
+
+<|ref|>text<|/ref|><|det|>[[118, 170, 880, 256]]<|/det|>
+In the figure above, we describe three scenarios: a) where we estimate admixture proportions into Uganda from non- Ghanaian red river hog populations, b) where we estimate admixture proportions into Ethiopia from non- Ghanaian red river hog populations, and c) where we estimate admixture proportions into each of the bushpig populations from Equatorial Guinea red river hogs.  
+
+<|ref|>text<|/ref|><|det|>[[118, 274, 880, 377]]<|/det|>
+As can be seen in the first scenario (a), we show that there is an increase in f4- ratios for the Uganda population as one moves from the westernmost part of Africa (Togo; \(10.9\%\) ) to the most central (Gabon; \(21.1\%\) ), in line with our hypothesis that Uganda is a potential zone of introgression and in agreement with the large signals observed in D- statistics comparisons (Fig. 2c). This is similarly seen for Ethiopia (b), with estimated admixture proportions of \(6.8\%\) to \(13.2\%\) , from west (Togo) to central (Gabon), respectively.  
+
+<|ref|>text<|/ref|><|det|>[[118, 394, 880, 445]]<|/det|>
+In the third scenario (c), we observe high levels of gene flow into Uganda and Ethiopia ( \(19.7\%\) and \(12.2\%\) , respectively) and much lower (albeit significant) levels in Tanzania and Zimbabwe, supporting our results in the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[118, 463, 880, 532]]<|/det|>
+Overall, these results, combined with the D- statistics, suggest significant gene flow into Uganda and Ethiopia from red river hog populations or one or more populations closely related to them that were not included in this study. We have added the plot to Supplementary Information as Supplementary Fig. S7.  
+
+<|ref|>text<|/ref|><|det|>[[118, 550, 880, 687]]<|/det|>
+Although the analysis infers \(\sim 20\%\) gene flow, it is worth noting that there are assumptions that the populations considered had evolved as a perfectly bifurcating tree and that the only gene flow event is the one that is modelled. As suggested by our \(f\) - branch ( \(f_{b}\) ) results (Supplementary Fig. S6), the relationship between the populations is likely much more complicated, and therefore one should be cautious in interpreting these results as evidence for a single point migration event. The amount of gene flow could be much higher from an unsampled population or it could be many migration events happening between several populations at different time points.  
+
+<|ref|>text<|/ref|><|det|>[[118, 705, 765, 721]]<|/det|>
+In light of these results, we have added the following to the main text (line 247):  
+
+<|ref|>text<|/ref|><|det|>[[118, 740, 881, 911]]<|/det|>
+"Furthermore, we estimated the amount of gene flow between species based on \(f_{4}\) - ratios, under assumptions that the populations considered had evolved together as a perfectly bifurcating tree and that the only gene flow event is the one that is modelled. This analysis estimated up to \(21\%\) gene flow from red river hogs into Uganda and up to \(13\%\) into Ethiopia, with increasing signals in more central populations (Supplementary Fig. S7). Given the complicated history of these populations, as suggested by our \(f\) - branch ( \(f_{b}\) ) results (Supplementary Fig. S6), these values are unlikely to accurately represent the historical gene flow events which likely occurred between multiple populations at different timepoints. Nevertheless, when considering all three analyses, these results suggest that there is or has been gene flow between the two taxa currently identified as species, and that the gradient of
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 84, 880, 117]]<|/det|>
+allele sharing between them is consistent with isolation by distance, where genetic similarity is strongest in populations from Central Africa."  
+
+<|ref|>text<|/ref|><|det|>[[118, 135, 880, 186]]<|/det|>
+"Given the results reported above, the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of introgression between the two taxa."  
+
+<|ref|>text<|/ref|><|det|>[[118, 188, 880, 239]]<|/det|>
+- This seems a bit speculative to me. The PSMC results are interesting, and visually appealing, and therefore I do think that they add to the story, however they are difficult to interpret in terms of any kind of specific hypothesis. Consider re-wording or removing part of this section.  
+
+<|ref|>text<|/ref|><|det|>[[118, 257, 880, 308]]<|/det|>
+Response: The above statement was not meant to be purely based on PSMC; rather, this was meant to be in light of the D-statistics (and now \(f_{4}\) - ratio) results that were described before this section. We have therefore revised the text to reflect this (lines 263- 264):  
+
+<|ref|>text<|/ref|><|det|>[[118, 326, 880, 377]]<|/det|>
+"These results, in combination with the D- statistics and \(f_{4}\) - ratio results reported above, suggest that the unique demographic histories in Uganda and Ethiopia could be influenced by their geographic location as a place of possible introgression between the two taxa."  
+
+<|ref|>text<|/ref|><|det|>[[118, 395, 880, 428]]<|/det|>
+"This analysis suggested that the Malagasy population experienced a severe bottleneck, likely a result of a founder event between 1- 5 kya."  
+
+<|ref|>text<|/ref|><|det|>[[118, 429, 808, 445]]<|/det|>
+- The decline seems to start quite some time before this though - why might this be?  
+
+<|ref|>text<|/ref|><|det|>[[118, 463, 880, 584]]<|/det|>
+Response: The decline in effective population size does appear to start in the interval \(\sim 5\) - 8kya where it drops to \(\sim 20,000\) individuals, followed by the large decline to \(\sim 1,000\) individuals around 1kya. When replying to this comment, we realised how difficult it was to see when the decline happened due to the log scale, so we have subsequently updated the figure with additional tick marks. It is hard to say whether the early start of the decline is due to a decline in the population in mainland Africa prior to colonisation of Madagascar or whether it is a methodological issue; this is now reflected in the main text.  
+
+<|ref|>text<|/ref|><|det|>[[118, 602, 711, 618]]<|/det|>
+We have rewritten the relevant part of the Discussion (lines 418- 429) as:  
+
+<|ref|>text<|/ref|><|det|>[[118, 636, 881, 826]]<|/det|>
+"Our popSizeABC estimate of an effective population size of 1,000 individuals during the bottleneck 1,500 years ago is surprisingly high, assuming that the founder event was a single occurrence involving a limited number of individuals carried to Madagascar by ship. However, the estimate is supported by their observed heterozygosity level, which is similar to levels observed in southern Africa, although we cannot know to what extent southern African populations have been subjected to drift since the founding of the Malagasy population. Multiple introductions, spanning over a longer period of time, or even with animals sourced from different mainland populations (including already admixed individuals), may have inflated this estimate. In addition, an instant large drop in population size followed by subsequent regrowth will be estimated by popSizeABC as a more gradual decline that starts earlier \(^{46}\) , making it difficult to pinpoint the exact timing of when the bottleneck started."  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 844, 235, 860]]<|/det|>
+## DISCUSSION  
+
+<|ref|>text<|/ref|><|det|>[[118, 879, 880, 912]]<|/det|>
+"had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current 63. However, from the PCA, NGAadmix and IBS tree we
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 83, 880, 117]]<|/det|>
+did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+<|ref|>text<|/ref|><|det|>[[118, 118, 880, 169]]<|/det|>
+- To?: "...North Mozambique current 63. However, while we were unable to explicitly test this hypothesis with our dataset (due to no samples from Comoros etc), PCA NGAdmix and the IBS tree did not." ?  
+
+<|ref|>text<|/ref|><|det|>[[118, 187, 880, 255]]<|/det|>
+Response: We agree that we should have made it clear that we could not have made statements about bushpig populations in the Comoros Island, since we do not have data from that area. We have therefore incorporated your suggestion and have changed lines 446- 449 to:  
+
+<|ref|>text<|/ref|><|det|>[[118, 273, 881, 377]]<|/det|>
+"...had led to suggestions of multiple, distinct introduction pulses through the Comoros Islands and the North Mozambique current \(^{67}\) . Although we were unable to explicitly test this hypothesis without samples from the Comoros, our results from the PCA, NGSadmix and IBS tree do not suggest multiple pulses into Madagascar and did not identify substantial structure within the island, which is consistent with a relatively homogeneous founder population."  
+
+<|ref|>text<|/ref|><|det|>[[120, 394, 430, 410]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 428, 881, 618]]<|/det|>
+The paper is clearly written and illustrated, though as an archaeologist I am in no position to pass judgement on the detailed methodology of the genetic study undertaken. That said, references to the debate surrounding the timing of human settlement of Madagascar are fair and the conclusion that bushpigs were likely introduced there \(\sim 1000 - 1500\) years ago does indeed fit extremely well with: a) dated faunal remains from the island, not just of bushpig but also of other exotic mammal taxa; b) the oldest archaeologically unambiguous, well- dated evidence for human presence; c) and specifically the African origins (and names) of domestic mammals and evidence for Triangular Incised Ware (Tana Ware), a form of pottery found in late first- millennium AD contexts in the southwest of the island - see Parker Pearson et al. 2010: 79 (Pastoralists, Warriors and Colonists: The Archaeology of Southern Madagascar; Oxford: Archaeopress).  
+
+<|ref|>text<|/ref|><|det|>[[119, 636, 880, 687]]<|/det|>
+For further discussion/reference of the timing of human settlement I would recommend Mitchell, Journal of Island and Coastal Archaeology 2020, arguments further elaborated in his 2022 book African Islands: A Comparative Archaeology (London: Routledge).  
+
+<|ref|>text<|/ref|><|det|>[[119, 705, 692, 721]]<|/det|>
+Provided that the genetic analysis is sound, I recommend acceptance.  
+
+<|ref|>text<|/ref|><|det|>[[119, 740, 880, 807]]<|/det|>
+Response: Thank you for spending time reviewing our manuscript and for providing us with important references to include. To further guide the discussion about human arrival, we have now incorporated both the Mitchell (2020) and Parker Pearson (2010) references in our Discussion, in line 434:  
+
+<|ref|>text<|/ref|><|det|>[[118, 826, 881, 911]]<|/det|>
+"...most likely through populations that started to arrive on Madagascar from southeastern Africa at least 1,500 years ago and possibly earlier \(^{26}\) , although the latter has been contested \(^{64}\) . Triangular Incised Ware (Tana Ware) style pottery remains were found in human settlements in southern Madagascar dated to 1,000- 1,400 years ago, suggesting either contact with or colonisation from the African Swahili region at this time \(^{65}\) ."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 285, 104]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 120, 217, 133]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[115, 136, 291, 149]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 150, 853, 180]]<|/det|>
+I already only had small number of minor comments on the original manuscript, and these have all been sufficiently addressed in the revised manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 223, 217, 237]]<|/det|>
+Reviewer #2:  
+
+<|ref|>text<|/ref|><|det|>[[115, 240, 291, 252]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 254, 848, 284]]<|/det|>
+All my comments/questions have been addressed. The authors should congratulate themselves on producing an well- written and important piece of work!
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[118, 83, 350, 100]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[118, 121, 428, 138]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 159, 863, 196]]<|/det|>
+I already only had small number of minor comments on the original manuscript, and these have all been sufficiently addressed in the revised manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[118, 216, 875, 252]]<|/det|>
+Response: We are pleased to hear that – thank you for your time and effort in reviewing our manuscript!  
+
+<|ref|>text<|/ref|><|det|>[[118, 273, 428, 290]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 311, 810, 347]]<|/det|>
+All my comments/questions have been addressed. The authors should congratulate themselves on producing an well- written and important piece of work!  
+
+<|ref|>text<|/ref|><|det|>[[118, 367, 875, 404]]<|/det|>
+Response: We appreciate the kind comment and for taking the time to provide positive and constructive comments to help improve our manuscript!
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1e09d1f0c744080040ce271de3a8b26e159b4dd20eaf5d470baf8cba1c52d41e/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1e09d1f0c744080040ce271de3a8b26e159b4dd20eaf5d470baf8cba1c52d41e/supplementary_0_Peer Review File__1e09d1f0c744080040ce271de3a8b26e159b4dd20eaf5d470baf8cba1c52d41e.mmd

@@ -0,0 +1,67 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+A cyclin- dependent kinase- mediated phosphorylation switch of disordered protein condensation  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+The authors have submitted a revised version of their MS, which was originally submitted to Nature. In the revised version, they have made several changes that strengthen their narrative and their claims. They have clarified various aspects that were confusing in the original version. I read the responses and the revisions. Overall, the revisions seem to address the concerns raised by the reviewers. As far as my comments are concerned, I have nothing major to ask as revisions. There, however, are some aspects of Fig. 5 that are puzzling. It would help to address these in a final version.  
+
+The new aspects of the theory are interesting, but also puzzling. Please note that many of the proteins listed as IDPs are not substrates of CDKs. Further, the model assumes that these proteins are IDPs. Many feature IDRs, but they also feature ordered domains. NCL and NPM1 are two classic examples of this hybrid architecture. The theory does not make this distinction, which could be seen as a weakness.  
+
+As an aside, the responses by the authors were pugilistic and offensive. Everything that the reviewers wrote is was judged as being confused, incorrect, surprising, dismissive, or inaccurate. Yes, the review process is a frustrating one because we do not get our way. Instead of plaudits, we are asked questions, our assumptions and assertions are challenged, and clarity is sought. Every author and every manuscript goes through this level of scrutiny. I sympathize with the authors. They are disappointed. However, all four reviewers judged what was submitted, not what the authors wanted us to see. I have reviewed a lot, but this is honestly the first time I have felt that the process was not worth the effort.
+
+<--- Page Split --->
+
+Reviewer #1 (Remarks to the Author):  
+
+The authors have submitted a revised version of their MS, which was originally submitted to Nature. In the revised version, they have made several changes that strengthen their narrative and their claims. They have clarified various aspects that were confusing in the original version. I read the responses and the revisions. Overall, the revisions seem to address the concerns raised by the reviewers. As far as my comments are concerned, I have nothing major to ask as revisions.  
+
+We appreciate the reviewer's recognition that the paper has improved compared to the original and that it fully addresses concerns of all reviewers.  
+
+There, however, are some aspects of Fig. 5 that are puzzling. It would help to address these in a final version.  
+
+The new aspects of the theory are interesting, but also puzzling. Please note that many of the proteins listed as IDPs are not substrates of CDKs.  
+
+All of the human IDPs whose analysis was presented in Fig. 5 have previously been documented as CDK substrates, as explained in the text detailing our compilation of human CDK substrates based on the available literature (Suppl Data 2; see also Methods for sources). We only analysed the effect of phosphorylations that are confirmed as mediated by CDKs, as well as documented phosphorylations on the minimal CDK consensus motif. The exact phosphorylation sites are indicated.  
+
+Further, the model assumes that these proteins are IDPs. Many feature IDRs, but they also feature ordered domains. NCL and NPM1 are two classic examples of this hybrid architecture. The theory does not make this distinction, which could be seen as a weakness.  
+
+Our model does not assume that the proteins are entirely disordered, as explained in the text and in the figure legend. The term 'intrinsically disordered protein' does not exclusively denominate a protein that is \(100\%\) disordered; IDPs include both proteins that contain intrinsically disordered regions (IDRs), as well as entirely disordered proteins. Since we show that CDK phosphorylation is highly skewed to the disordered regions, which are thought to participate in protein- protein interactions, we only analysed the effects of phosphorylation on the disordered regions of the selected CDK targets. The analysed IDR sequences are indicated in the figure, and described in detail in Suppl. Data 5. We do, however, present the full- length Ki- 67, where the size of the structured N- terminal FHA domain, which is not itself phosphorylated by CDKs, is negligible compared to the disordered remainder. To be clear, we do not make any claim about the effect of CDK- mediated phosphorylation on the structured regions within these proteins: it would indeed be useful to understand the effect of the phosphorylation on the entire protein entity, but no analytical theory exists yet that would allow modeling of the effect of phosphorylation of a single chain with an IDR and a folded domain.  
+
+As an aside, the responses by the authors were pugilistic and offensive.
+
+<--- Page Split --->
+
+We are very surprised that the reviewer perceived our replies in this manner, as they were not personal, but scientific.  
+
+Everything that the reviewers wrote is was judged as being confused, incorrect, surprising, dismissive, or inaccurate.  
+
+We feel that this comment is an exaggeration as in no way did we dismiss everything the reviewers wrote in our reply. It is perfectly normal to point out inaccuracies in the reviews, which are often a result of unclear explanation on our part in the version that was reviewed.  
+
+Yes, the review process is a frustrating one because we do not get our way.  
+
+It is not a question of getting one's way or not. In general, we do not find review processes frustrating, but when reviews use dismissive language (this particular reviewer used the phrase "Beyond Figure 1, the MS comes off the rails in terms of novelty", which no doubt reflects the reviewer's perception and which we did not take personally), it is surely understandable that we explain why we do not agree.  
+
+Instead of plaudits, we are asked questions, our assumptions and assertions are challenged, and clarity is sought. Every author and every manuscript goes through this level of scrutiny.  
+
+We have no problem with this whatsoever, and, no doubt like the reviewer, we have been through this process many times before, and understand it perfectly.  
+
+I sympathize with the authors. They are disappointed. However, all four reviewers judged what was submitted, not what the authors wanted us to see. I have reviewed a lot, but this is honestly the first time I have felt that the process was not worth the effort.  
+
+We are disappointed that the reviewer feels this way. We do not expect sympathy. We have also reviewed many papers, and indeed it is not an easy task to evaluate years of work condensed into a scientific paper in an insufficient space, especially when authors might be insufficiently clear. Yet one has to recognise that reviewers are not superhuman and cannot be experts on all aspects of a multidisciplinary study; they can also make mistakes; they may not have the time to read all of the relevant literature, and so on. We were of course disappointed with the editorial decision at Nature – who wouldn't be? – as we felt that we could revise the paper in accordance with the reviewers' criticisms, but in no way did we consider the reviews incompetent, and we apologise if the reviewer misunderstood our perception of his/her review.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1e09d1f0c744080040ce271de3a8b26e159b4dd20eaf5d470baf8cba1c52d41e/supplementary_0_Peer Review File__1e09d1f0c744080040ce271de3a8b26e159b4dd20eaf5d470baf8cba1c52d41e_det.mmd

@@ -0,0 +1,90 @@
+<|ref|>title<|/ref|><|det|>[[61, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[70, 111, 362, 140]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[70, 155, 900, 211]]<|/det|>
+A cyclin- dependent kinase- mediated phosphorylation switch of disordered protein condensation  
+
+<|ref|>image<|/ref|><|det|>[[57, 732, 240, 780]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 912, 785]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 844, 163]]<|/det|>
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 203, 300, 219]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 260, 393, 277]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 316, 877, 426]]<|/det|>
+The authors have submitted a revised version of their MS, which was originally submitted to Nature. In the revised version, they have made several changes that strengthen their narrative and their claims. They have clarified various aspects that were confusing in the original version. I read the responses and the revisions. Overall, the revisions seem to address the concerns raised by the reviewers. As far as my comments are concerned, I have nothing major to ask as revisions. There, however, are some aspects of Fig. 5 that are puzzling. It would help to address these in a final version.  
+
+<|ref|>text<|/ref|><|det|>[[115, 465, 881, 537]]<|/det|>
+The new aspects of the theory are interesting, but also puzzling. Please note that many of the proteins listed as IDPs are not substrates of CDKs. Further, the model assumes that these proteins are IDPs. Many feature IDRs, but they also feature ordered domains. NCL and NPM1 are two classic examples of this hybrid architecture. The theory does not make this distinction, which could be seen as a weakness.  
+
+<|ref|>text<|/ref|><|det|>[[115, 576, 879, 704]]<|/det|>
+As an aside, the responses by the authors were pugilistic and offensive. Everything that the reviewers wrote is was judged as being confused, incorrect, surprising, dismissive, or inaccurate. Yes, the review process is a frustrating one because we do not get our way. Instead of plaudits, we are asked questions, our assumptions and assertions are challenged, and clarity is sought. Every author and every manuscript goes through this level of scrutiny. I sympathize with the authors. They are disappointed. However, all four reviewers judged what was submitted, not what the authors wanted us to see. I have reviewed a lot, but this is honestly the first time I have felt that the process was not worth the effort.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[151, 85, 440, 101]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[150, 118, 844, 223]]<|/det|>
+The authors have submitted a revised version of their MS, which was originally submitted to Nature. In the revised version, they have made several changes that strengthen their narrative and their claims. They have clarified various aspects that were confusing in the original version. I read the responses and the revisions. Overall, the revisions seem to address the concerns raised by the reviewers. As far as my comments are concerned, I have nothing major to ask as revisions.  
+
+<|ref|>text<|/ref|><|det|>[[150, 241, 815, 275]]<|/det|>
+We appreciate the reviewer's recognition that the paper has improved compared to the original and that it fully addresses concerns of all reviewers.  
+
+<|ref|>text<|/ref|><|det|>[[150, 293, 842, 327]]<|/det|>
+There, however, are some aspects of Fig. 5 that are puzzling. It would help to address these in a final version.  
+
+<|ref|>text<|/ref|><|det|>[[150, 345, 835, 380]]<|/det|>
+The new aspects of the theory are interesting, but also puzzling. Please note that many of the proteins listed as IDPs are not substrates of CDKs.  
+
+<|ref|>text<|/ref|><|det|>[[150, 397, 847, 502]]<|/det|>
+All of the human IDPs whose analysis was presented in Fig. 5 have previously been documented as CDK substrates, as explained in the text detailing our compilation of human CDK substrates based on the available literature (Suppl Data 2; see also Methods for sources). We only analysed the effect of phosphorylations that are confirmed as mediated by CDKs, as well as documented phosphorylations on the minimal CDK consensus motif. The exact phosphorylation sites are indicated.  
+
+<|ref|>text<|/ref|><|det|>[[150, 519, 838, 589]]<|/det|>
+Further, the model assumes that these proteins are IDPs. Many feature IDRs, but they also feature ordered domains. NCL and NPM1 are two classic examples of this hybrid architecture. The theory does not make this distinction, which could be seen as a weakness.  
+
+<|ref|>text<|/ref|><|det|>[[149, 606, 842, 867]]<|/det|>
+Our model does not assume that the proteins are entirely disordered, as explained in the text and in the figure legend. The term 'intrinsically disordered protein' does not exclusively denominate a protein that is \(100\%\) disordered; IDPs include both proteins that contain intrinsically disordered regions (IDRs), as well as entirely disordered proteins. Since we show that CDK phosphorylation is highly skewed to the disordered regions, which are thought to participate in protein- protein interactions, we only analysed the effects of phosphorylation on the disordered regions of the selected CDK targets. The analysed IDR sequences are indicated in the figure, and described in detail in Suppl. Data 5. We do, however, present the full- length Ki- 67, where the size of the structured N- terminal FHA domain, which is not itself phosphorylated by CDKs, is negligible compared to the disordered remainder. To be clear, we do not make any claim about the effect of CDK- mediated phosphorylation on the structured regions within these proteins: it would indeed be useful to understand the effect of the phosphorylation on the entire protein entity, but no analytical theory exists yet that would allow modeling of the effect of phosphorylation of a single chain with an IDR and a folded domain.  
+
+<|ref|>text<|/ref|><|det|>[[150, 884, 689, 902]]<|/det|>
+As an aside, the responses by the authors were pugilistic and offensive.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[150, 101, 830, 136]]<|/det|>
+We are very surprised that the reviewer perceived our replies in this manner, as they were not personal, but scientific.  
+
+<|ref|>text<|/ref|><|det|>[[150, 154, 835, 189]]<|/det|>
+Everything that the reviewers wrote is was judged as being confused, incorrect, surprising, dismissive, or inaccurate.  
+
+<|ref|>text<|/ref|><|det|>[[150, 206, 840, 258]]<|/det|>
+We feel that this comment is an exaggeration as in no way did we dismiss everything the reviewers wrote in our reply. It is perfectly normal to point out inaccuracies in the reviews, which are often a result of unclear explanation on our part in the version that was reviewed.  
+
+<|ref|>text<|/ref|><|det|>[[150, 275, 718, 293]]<|/det|>
+Yes, the review process is a frustrating one because we do not get our way.  
+
+<|ref|>text<|/ref|><|det|>[[150, 310, 835, 397]]<|/det|>
+It is not a question of getting one's way or not. In general, we do not find review processes frustrating, but when reviews use dismissive language (this particular reviewer used the phrase "Beyond Figure 1, the MS comes off the rails in terms of novelty", which no doubt reflects the reviewer's perception and which we did not take personally), it is surely understandable that we explain why we do not agree.  
+
+<|ref|>text<|/ref|><|det|>[[150, 415, 840, 450]]<|/det|>
+Instead of plaudits, we are asked questions, our assumptions and assertions are challenged, and clarity is sought. Every author and every manuscript goes through this level of scrutiny.  
+
+<|ref|>text<|/ref|><|det|>[[150, 467, 830, 502]]<|/det|>
+We have no problem with this whatsoever, and, no doubt like the reviewer, we have been through this process many times before, and understand it perfectly.  
+
+<|ref|>text<|/ref|><|det|>[[150, 519, 836, 571]]<|/det|>
+I sympathize with the authors. They are disappointed. However, all four reviewers judged what was submitted, not what the authors wanted us to see. I have reviewed a lot, but this is honestly the first time I have felt that the process was not worth the effort.  
+
+<|ref|>text<|/ref|><|det|>[[150, 588, 844, 762]]<|/det|>
+We are disappointed that the reviewer feels this way. We do not expect sympathy. We have also reviewed many papers, and indeed it is not an easy task to evaluate years of work condensed into a scientific paper in an insufficient space, especially when authors might be insufficiently clear. Yet one has to recognise that reviewers are not superhuman and cannot be experts on all aspects of a multidisciplinary study; they can also make mistakes; they may not have the time to read all of the relevant literature, and so on. We were of course disappointed with the editorial decision at Nature – who wouldn't be? – as we felt that we could revise the paper in accordance with the reviewers' criticisms, but in no way did we consider the reviews incompetent, and we apologise if the reviewer misunderstood our perception of his/her review.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1e1bedff740ea0146a02a6d351b3babf234584dc6acf3acb3873b690d8852174/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1e1bedff740ea0146a02a6d351b3babf234584dc6acf3acb3873b690d8852174/supplementary_0_Peer Review File__1e1bedff740ea0146a02a6d351b3babf234584dc6acf3acb3873b690d8852174.mmd

@@ -0,0 +1,395 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+# NDUFS4 Regulates Cristae Remodeling in Diabetic Kidney Disease  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+This paper investigates the interesting hypothesis that the NDUFS4 accessory subunit of OXPHOS complex I is involved in the control/remodeling of mitochondrial cristae architecture. Conceptually, it presents extensive experimental data focusing on effects in NDUFS4 overexpression systems, as well as the potential interaction between NDUFS4 and STOML2. Although the scientific message of this paper certainly can be appealing, I still have several questions/concerns related to the used experimental strategies and proposed concepts.  
+
+Fig. 1b: it is unclear to me which CI subunits were significantly reduced  
+
+Fig. 1c: what does the Z- score indicate?  
+
+Fig. 1d (and others): why is CI activity expressed as NAD+/NADH? This ratio is dependent on many other metabolic pathways. For proper analysis of (rotenone- sensitive) CI activity more "classical" enzymatic assays in mitoplasts should be used.  
+
+Fig. 1h: Why was only NDUFS4 analysed? I guess there should also be significant correlations with other CI subunits? When such correlations are absent, this would strengthen the choice to focus on NDUFS4 for further analysis. When such correlations are present, why was NDUFS4 selected?  
+
+Fig. 1i- l: more details are needed on how NDUFS4 staining was exactly quantified  
+
+Page 6, lines 8- 10: in my view, the data suggests that CI expression is important, meaning that there is no evidence for NDUFS4- specific effects.  
+
+Fig. 3a: please compute the relevant respiratory control ratios (RCRs). These are independent of cell number and will highlight functional defects. Furthermore, there is no ECAR data presented. Why?  
+
+Fig. 3g (bargraphs): the average WT value is 1.0; how was data normalization performed for this and all relevant other figures?  
+
+Fig. 3h- n (and page 7; line 23): the presented parameters do not reflect mitochondrial "dynamics" but steady- state mitochondrial morphology. One can not talk about enhanced fission (it can also be reduced fusion) since mitochondrial fission/fusion proteins were not investigated. Moreover, it is unclear how the conclusion that "cristae morphology is altered" is reached.  
+
+Fig. 3m (page 8; line 2): MitoSOX does not specifically detect "mitochondrial ROS". The ethidium molecule (which is the ROS- detection moiety of MitoSOX) reacts extremely fast (but not exclusively with) superoxide. This means that MitoSOX can become fluorescent in the cytosol during its transit to the mitochondrial matrix (driven by the presence of the decylTPP moiety of MitoSOX).  
+
+Fig. 3n (page 8; line 2): the used assay reports on total cellular ATP content, so is not informative on ATP production.  
+
+page 8; line 6: what is meant with "mitochondrial reprogramming"?  
+
+Page 8; line 7: please rule out that the used DOX concentration and incubation regime impacts on mitochondrial function (described in the literature).  
+
+Page 8; line 10: it is not explained why HG is compared to NG and why this is relevant for the rest
+
+<--- Page Split --->
+
+of the study.  
+
+Fig. 4f: what are all the other (non- marked) bands representing. Or is this not relevant? Please explain. I guess the AB cocktail was also used here?  
+
+Page 9; line 7: why is it surprising that NDUFS4 OE restored CI in- gel activity "even under HG conditions"?  
+
+Page 9; line 11: that NDUFS4 exerts a "regulatory effect" is not demonstrated by the preceding experimental results.  
+
+Regarding the STOML2- related experiments, I wonder if the proposed binding of STOML2 to NDUFS4 protein is compatible with the currently available structural information on the CI holocomplex and CI in a respiratory supercomplex? In my opinion, this is the most potentially interesting finding of this work since all functional effects of NDUFS4 OE could be explained by the NDUFS4 increase leading to more fully assembled CI. How does the cell "know" that more other CI subunits are required in the NDUFS4- overexpressing case? This would mean that NDUFS4 might be (among) the rate- limiting subunits for CI assembly. Is this supported by the current literature on CI biogenesis?  
+
+## Reviewer #2 (Remarks to the Author):  
+
+The paper by Mise et al has explored the role of the Complex I subunit Ndufs4, in the pathogenesis of diabetic kidney disease. The aim of this study was to assess ETC abundance, mitochondrial function and morphology in kidney podocytes in a mouse model of diabetes and whether this affected DKD progression. First, ETC abundance was determined by mass spectrometry in two mouse models of diabetes, the Ins2Akita mouse and the db/db mouse model. Complex I abundance and mRNA expression was decreased in podocytes of the Ins2Akita mouse and db/db mouse.  
+
+Out of the several downregulated subunits of mitochondrial OXPHOS complexes, Ndufs4 was chosen as a target. Glomerular Ndufs4 protein was decreased in human biopsies from patients with DKD. Next, podocyte- specific Ndufs4- transgenic mice were generated and crossed onto the Ins2Akita model of type 1 diabetes. Podocyte- specific Ndufs4- transgenic mice were partially protected from the diabetic kidney disease phenotype. Similar data were shown in the db/db mouse model. Ndufs4 overexpression in podocytes rescued high glucose- induced mitochondrial cristae changes as well as supercomplex assembly. A molecular interaction was identified between STOML2 and NDUFS4. The authors conclude that the findings represent a major paradigm shift in the current management of DKD by suggesting that targeting ETC remodeling could be a promising approach for developing therapies to mitigate the progression of DKD.  
+
+## Major comments  
+
+ETC remodeling, mitochondrial function and morphology in kidney podocytes in rodent models with diabetes has been widely reported, including from the current group. Several other studies have explored the role of mitochondrial Complex I subunits, leading to ETC remodeling in the development of chronic kidney disease (e.g., PMID: 23320803), therefore the idea itself is not novel. These papers should be referenced. Many others have suggested that targeting the electron transport chain is an approach to treat CKD/DKD (PMID: 36781216), therefore the author's suggestion of a major paradigm shift should be moderated.  
+
+It is not clear which method was used to determine Complex I activity. The method referenced in the paper (ref 43), reports using NBT on tissue sections. This is not a generally accepted method. The gold standard method used by mitochondrial laboratories for determination of complex I activity is via respiratory chain enzymology using kinetic spectrophotometric assays (Frazier AE et al., Assessment of mitochondrial respiratory chain enzymes in cells and tissues, Methods Cell Biol
+
+<--- Page Split --->
+
+2020). Each assay requires specific inhibitors to ensure that the assay reflects true activity of the specific respiratory chain enzymology complex being assayed, i.e., rotenone-sensitive CI activity. In order to avoid misinterpretation, the mitochondrial marker enzyme citrate synthase is typically assayed to enable correction for mitochondrial content by expressing enzymes as citrate synthase ratios. Complex I activity should be determined using the kinetic spectrophotometric assay outlined in Frazier.  
+
+"We also evaluated NDUFS4 staining in glomeruli from diabetic subjects with a wide spectrum of DKD histology, and found that NDUFS4 staining in glomeruli was progressively reduced with worsening of DKD histology (test for trend \(P< 0.01\) ) (Fig. 1l)." Was there a significant difference from Class I CKD to Class III CKD, or between any of the DKD classes? Can you explain the Test for trend? Is it really a progressive decrease in glomerular Ndufs4 staining?  
+
+Fig1K. Statistical analysis was performed on a sample size of \(n = 2\) within the microalbuminuria group. How is this possible when using one way- analysis of variance followed by Tukey's multiple comparisons test in Graphpad Prism? Please confirm the statistical test used.  
+
+What was the mitochondrial density of the podocytes in diabetic mice? Ins2Akita and db/db mice.  
+
+Fig 3n, ATP production. What are the units of the ATP assay? Both the mitosox and ATP data are expressed as relative to control. As is the Complex I. The ATP assay used appears to be a one- step single reagent assay for detection of viable cells in culture, rather than an assay which can reliably determine ATP production over time in a sensitive manner. I would caution against using such an assay when screening for differences between groups. Was a standard curve generated and used? Did the authors account for cell number or total protein? The same questions apply to the mitosox assay and data. Representative histograms showing mean intensity of MitoSOX fluorescence should be shown for each group.  
+
+## Minor comments  
+
+Some of the individual data points on graphs are missing through the manuscript (e.g., Fig 2h, k, l & m; Figure 3g, i, j, k, l) and throughout the paper.  
+
+Fig 2a. What sample is contained in each lane in the genotyping gel? Fig 2b. The mRNA expression of Ndufs4 appears higher in the "podocyte depleted" fraction of renal cortex vs the podocyte fraction. It is important to show Ndufs4 protein in podocytes vs tubules and total renal cortex.  
+
+For the Seahorse cell culture studies, how many independent cell culture experiments are shown? It is stated that there were \(n = 6 - 8\) replicates per group. Are these data derived from one cell culture experiment?  
+
+Extended data figure 5a. The western blots are not convincing. The quantitation needs to be shown. In mouse kidney, OPA1 generally has 5 isoforms. There should be 5 bands present. Samples need to be run on a gradient gel to reveal the five isoforms as per PMID: 26822084.  
+
+Introduction. Page 4, line 13. "ETC dysfunction is recognized as an important cause of organ failure in several human pathologies including heart failure, diabetes, and neurodegeneration in a tissue- specific manner. Mitochondrial cytopathies have long been recognised to lead to kidney failure. This should be mentioned in the introduction (PMID: 33305107)
+
+<--- Page Split --->
+
+## Reviewer #3 (Remarks to the Author):  
+
+In this manuscript, Mise et al. examine the importance of the mitochondrial ETC Complex I protein Ndufs4, in the maintenance of mitochondrial structure in the context of the diabetic kidney. The Authors show that Ndufs4 plays a role in the regulation of cristae structure and supercomplex formation, while providing evidence of binding partners in the process. The studies address an interesting topic, and the experimentation appears thorough and well- executed. Specific comments are indicated below.  
+
+1) It is not entirely clear why the Authors chose to focus on Ndufs4. The heat map in Fig.1 shows many ETC proteins, including in Complex I that were also decreased as a result of diabetes. In fact, there are some that were decreased by a greater extent. The decision to choose Ndufs4 comes across as a biased decision without scientific justification. A greater case should be made for choosing this target. In the Discussion, the Authors indicate that examination of other subunits would be meaningful, and this is appreciated, but does not preclude the need for a greater justification for the Ndufs4 focus.  
+
+2) Introduction, page 4; in the last sentence, the Authors state that their "results unexpectedly reveal that ETC integrity determines the stability of RSCs." This seems disingenuous. Do the Authors really believe that the very integrity of the components of the ETC would not be an important factor in the stability of the supercomplex in which they reside? My suggestion would be to either remove or soften this statement.  
+
+3) The use of multiple experimental models of diabetes (Akita, db/db) and human patient samples is appreciated and strengthen the conclusions and relevance of the studies. With that being said, the inclusion of complete db/db data (especially proteomic profiling in Fig.1b) is glaringly absent. Inclusion of this data makes a stronger case for commonality in the models and the ultimate focus on Ndufs4.
+
+<--- Page Split --->
+
+We greatly appreciate the comments/suggestions made by the reviewers. In the revised version of our manuscript, we have made a number of additional experiments that are now presented in 15 new panels that include:  
+
+- Rotenone-sensitive complex I enzymatic activity (Figs. 1d,2d; and Extended Data Fig. 4g)- Real-time ATP production (Figs. 3p,3q),- Mitochondrial ROS assessment using mito-roGFP, a specific mitochondrial matrix-targeted redox-sensitive reporter (Fig. 3o)- Proteomic data from type 2 diabetic mice Leprdb/+ and Leprdb/db mice (Figs. 1e and Extended Data Fig. 1d),- RCR and ECAR data (Extended Data Figs. 3a,3b)- GST-NDUFS4 pulldown assay to identify more specific binding sites between NDUFS4 and STOML2 (Figs. 6g,6i).- Excluding the DOX effect in our DOX-inducible model (Extended Data Figs. 4c, 4d, 4e)  
+
+The additions to the manuscript are highlighted in yellow for ease of the reviewers.  
+
+## Reviewer #1  
+
+Comment 1: Fig. 1b: it is unclear to me which CI subunits were significantly reduced. In response to the reviewer's comment, we have since introduced an informative table (Extended Data Fig 1d) in addition to the previously shown Fig 1b. This table serves to outline the CI subunits with significantly reduced values, providing a convenient and comprehensive reference for those specific CI subunits.  
+
+Comment 2: Fig. 1c: what does the Z- score indicate?  
+
+In the original manuscript, we employed Z- scores, a widely used statistical tool for proteomic analysis, primarily to indicate the extent to which a specific data point is deviated from the mean, measured in terms of standard deviations. In the revised version of the manuscript, we adopted an alternative approach by using adjusted intensity- based absolute quantification (iBAQ) values. Adjusted iBAQ values represent protein abundance estimates that have been normalized for meaningful comparison between samples \(^{1,2}\) (Cabrera- Orefice A. et al Front Cell Dev Biol 2022; 9:796128 and Válikangas T. et al., Brief Bioinform 2018;19(1):1- 11). Both methods are commonly used in proteomic analysis, but they provide different types of information. Whereas Z- scores can help identify proteins of interest based on their deviation from the mean, adjusted iBAQ values provide information about the relative abundance of proteins across different conditions or samples. Furthermore, we have now included heatmaps generated using these adjusted iBAQ values for each complex subunit (Extended Data Figure 1b).  
+
+Comment 3: Fig. 1d (and others): why is CI activity expressed as NAD+/NADH? This ratio is dependent on many other metabolic pathways. For proper analysis of (rotenone- sensitive) CI activity, more "classical" enzymatic assays in mitoplasts should be used. We appreciate the reviewer's comment. In the revised version of the manuscript, we have assessed CI activity using the classical kinetic spectrophotometric assays \(^{3}\) (Frazier AE et al. Methods Cell Biol 2020;155:1221- 156) (Fig. 1d and Fig. 2d.). Our findings closely mirror the
+
+<--- Page Split --->
+
+results previously presented in our initial submission, as illustrated in Extended Data Fig. 1c, Extended Data Fig. 2a, and Extended Data Fig. 4d. The initial assessment of CI activity was based on a well- established commercially available assay (Abcam, ab109721) used in other publications<sup>4-6</sup> (Wang T. et al., Cell Metab 2021; 33 (3):531- 546. e9; Balsa E., et al., Mol Cell 2019; 74 (5):877- 890. e6; and Cao LL. et al., Nature 2016; 539(7630):575- 578).  
+
+Comment 4: Fig. 1h: Why was only NDUFS4 analyzed? I guess there should also be significant correlations with other CI subunits? When such correlations are absent, this would strengthen the choice to focus on NDUFS4 for further analysis. When such correlations are present, why was NDUFS4 selected?  
+
+We selected NDUFS4 among several CI subunits in our proteomic experiment for several compelling reasons: 1. Proteomic Analysis: in our comprehensive proteomic investigation, we identified NDUFS4 as a significantly downregulated subunit within CI in kidney podocytes from diabetic Akita mice, in comparison to those from a control group of non- diabetic mice. 2. Consistent Validation: subsequent validation studies, spanning various models including cellular, murine, and human samples, consistently revealed NDUFS4 as the most prominently downregulated subunit in diabetes when compared to nondiabetic controls, further substantiating its significance. In contrast, several other potential CI candidates, including NDUFA2, NDUFB3, NDUFB4, NDUFB5, NDUFB8, NDUFB11, and NDUFV3 were not consistently downregulated in the diabetic mice or in the glomeruli of patients with DKD (Extended Data Fig. 1e, f, g). 3. Early Downregulation: examination of human samples of individuals with DKD indicated that the downregulation of NDUFS4 occurs prior to the onset of albuminuria in patients, suggesting its potential role as an early marker of dysfunction (Fig. 1k, i). 4. Genetic Evidence: importantly, both human and murine genetic data underscore the critical role of NDUFS4 in mitochondrial function since the absence of functional NDUFS4 is associated with Leigh syndrome, emphasizing its indispensable function within mitochondrial processes.  
+
+Comment 5: Fig. 1i- l: more details are needed on how NDUFS4 staining was exactly quantified.  
+
+In the revised version of our manuscript, we have incorporated comprehensive information regarding the staining and quantification of NDUFS4 ("Methods" section on page 33). To quantify NDUFS4, we computed the mean intensity within the glomerular tuft area across all glomeruli obtained from healthy donor and DKD kidney samples. This quantification was carried out utilizing Image J software, following an established protocol previously outlined<sup>7</sup> (Falkevall A, et al. Cell Metab 2017;25(3):713- 726).  
+
+Comment 6: Page 6, lines 8- 10: in my view, the data suggests that CI expression is important, meaning that there is no evidence for NDUFS4- specific effects. Our initial data indicate a significant downregulation of CI subunits within the podocytes of diabetic mice, as demonstrated in Fig. 1c. However, among these CI subunits, Ndufs4 consistently displayed reduced expression levels in both diabetic mouse models and human patients. It is important to clarify that our study at that point simply highlighted the potential significance of Ndufs4 in the context of our experimental approach.  
+
+Comment 7: Fig. 3a: please compute the relevant respiratory control ratios (RCRs). These are independent of cell number and will highlight functional defects. Furthermore, there is no ECAR data presented. Why? We appreciate this concern raised by the reviewer. We have added RCRs values to the revised version of the manuscript. Consistent with our previous data, we observe that RCR is
+
+<--- Page Split --->
+
+significantly reduced in primary podocytes from diabetic Ins2Akit/+ mice compared to those in WT mice and Ins2Akit/+ - Ndufs4podTg mice. We have added these data to the Extended Data Fig. 3a. Regarding ECAR, we have also added ECAR data into Extended Data Fig. 3b.  
+
+Comment 8: Fig. 3g (bargraphs): the average WT value is 1.0; how was data normalization performed for this and all relevant other figures?  
+
+We quantified the NADH oxidase staining using Image J software. For each glomerulus (a total of 60 glomeruli per group), we determined the intensity divided by the median intensity observed in the WT (wild type) images. This standardized approach ensures that the median intensity of the WT group is set to 1.0. The same normalization process was applied in Extended Data Fig. 2a to assess the relative CI activity in glomeruli.  
+
+Comment 9: Fig. 3h- n (and page 7; line 23): the presented parameters do not reflect mitochondrial "dynamics" but steady- state mitochondrial morphology. One cannot talk about enhanced fission (it can also be reduced fusion) since mitochondrial fission/fusion proteins were not investigated. Moreover, it is unclear how the conclusion that "cristae morphology is altered" is reached.  
+
+We appreciate this insightful question raised by the reviewer and agree that mitochondrial dynamics predominately depend on fission, fusion, shape transition, and transport or tethering along the cytoskeleton. Consistent with these concepts, we have previously shown that kidney podocytes exposed to high glucose conditions exhibit a fragmented or punctate phenotype<sup>8,9</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200; and Galvan DL., et al., J Clin Invest 2019;129(7):2807- 2823). Our earlier work also has clearly demonstrated that this shift towards shorter and fragmented mitochondria in high glucose- treated podocytes is attributed in part to an enhanced mitochondrial fission process characterized by increased Drp1 activity, the primary protein governing mitochondrial fission<sup>8</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200). In a subsequent study, we developed mouse models where the phosphorylation site of DRP1 at Ser 600 was mutated and rendered inactive specifically in podocytes of diabetic mice<sup>9</sup> (Galvan DL., et al., J Clin Invest 2019;129(7):2807- 2823). Additionally, we reported an increase in actin/Drp1 interactions associated with DRP1 phosphorylation<sup>9</sup> (Galvan DL., et al., J Clin Invest 2019, 129(7):2807- 2823). Therefore, the mitochondrial morphology depicted in Fig. 3h, particularly the observed shortening of mitochondria in the kidney podocytes of diabetic Ins2Akit/+ mice, is consistent with our prior publications in various experimental models of diabetic kidney disease. We concur with the reviewer that mitochondria maintain their morphology through a dynamic balance of fission and fusion processes in cells which is regulated by a number of regulatory kinetic proteins. Regarding cristae morphology, we quantified mitochondrial cristae abundance in transmission electron microscopy (TEM) micrographs obtained from primary podocytes using Image J. Our analysis revealed a significant reduction in cristae abundance in podocytes from Ins2Akit/+ mice, which was subsequently improved in the Ins2Akit/+ - Ndufs4podTg mice. These results have been included in Fig. 3m and n.  
+
+Comment 10: Fig. 3m (page 8; line 2): MitoSOX does not specifically detect "mitochondrial ROS". The ethidium molecule (which is the ROS- detection moiety of MitoSOX) reacts extremely fast (but not exclusively with) superoxide. This means that MitoSOX can become fluorescent in the cytosol during its transit to the mitochondrial matrix (driven by the presence of the decylTPP moiety of MitoSOX).  
+
+We appreciate and acknowledge that the use of MitoSOX, a widely employed tool for assessing mitochondrial ROS (mROS) in numerous studies<sup>10,11</sup> (Sutandy F.X.R., et al., Nature 2023;618(7966):849- 854, Labuschagne CF., et al., Cell Metab 2019;30(4):720-
+
+<--- Page Split --->
+
+734. e5), may lack specificity in detecting mROS. In collaboration with Dr. Paul Schumacker, a co- author of this manuscript and an internationally known expert in the area of ROS<sup>12</sup> (Murphy et al. Guidelines for measuring reactive oxygen species and oxidative damage in cells and in vivo. Nat Metab. 2022;4(6):651- 662), we adopted a more precise approach for evaluating mROS in the revised version of our manuscript. We utilized a ratiometric mitochondrial matrix- targeted redox- sensitive reporter, mito- rGFP, which offers several advantages over traditional methods of assessing mROS, as previously documented by our group and others<sup>13,14</sup> (Galvan DL. et al., Kidney Int 2017;92(5):1282- 128; Waypa GB. et al., Circ Res 2010;106(3):526- 535). The mito- rGFP reporter is highly sensitive to changes in the redox state within the mitochondrial matrix. To this aim, we introduced mito- rGFP into primary podocytes isolated from WT, Ndufs4<sup>podTg</sup>, Ins2<sup>Akita/+</sup>, and Ins2<sup>Akita/+</sup>- Ndufs4<sup>podTg</sup> mice. We quantified the ratios of oxidized mito- rGFP to reduced mito- rGFP intensities using confocal microscopy images. As illustrated in Fig 3o, we observed that the ratio of oxidized/reduced rGFP was elevated in primary podocytes from Ins2<sup>Akita/+</sup> mice compared to WT mice while it was reduced in Ins2<sup>Akita/+</sup>- Ndufs4<sup>podTg</sup> mice.  
+
+Comment 11: Fig. 3n (page 8; line 2): the used assay reports on total cellular ATP content, so is not informative on ATP production. We appreciate the critique by the reviewer and consequently have changed the term ATP production to ATP content (Extended Data Fig. 3n and Extended Data Fig. 4h). Moreover, we performed a real- time, label- free assay to quantify cellular and mitochondrial ATP production rates in live primary podocytes (Fig. 3p,q).  
+
+Comment 12: page 8; line 6: what is meant with "mitochondrial reprogramming"? The concept of mitochondrial reprogramming underscores the dynamic nature of mitochondria and their ability to respond to signals that they receive from other cells or their microenvironment, resulting in changes to their function, morphology, number, and distribution within the cell. This degree of remodeling allows mitochondria to quickly adapt to their changing environmental cues.  
+
+Comment 13: Page 8; line 7: please rule out that the used DOX concentration and incubation regime impacts on mitochondrial function (described in the literature).  
+
+We appreciate the reviewer's comment. High doses of doxycycline have been previously shown to hinder protein translation of mitochondrially encoded genes—a phenomenon known as mitonuclear protein imbalance leading to defects in basal and maximal OCRs<sup>15</sup> (Moullan N., et al., Cell Rep 2015;10(10):1681- 1691). To determine the potential influence of DOX at a 200 nM concentration, as employed in our experiments, on mitochondrial function in our experimental models, we conducted an experiment comparing mitochondrial basal and maximal OCRs in podocytes treated with 200nM DOX compared to those treated with 1000 nM and 2000 nM. We found that using DOX at 200 nM, mitochondrial OCR remained largely unaffected. However, when incubated at a substantially higher dose (2000 nM), we observed significant changes in the basal and maximal OCR consistent with the findings in the literature. We have incorporated these new results into Extended Data Fig. 4c,d,e and provided a brief description in the Results section.  
+
+Comment 14: Page 8; line 10: it is not explained why HG is compared to NG and why this is relevant for the rest of the study.  
+
+High glucose (HG: 25mM glucose) is conventionally added to culture media to mimic diabetic conditions in cell models. In prior studies, both our lab and others have comprehensively evaluated the impact of various HG concentrations and treatment durations on eliciting
+
+<--- Page Split --->
+
+cellular responses in podocytes<sup>8,16- 19</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200, Long J. et al., J Clin Invest 2016;126(11):4205- 18, Qi W. et al., Nat Med 2017;23(6):753- 762, Fu Y. et al., Cell Metab 2020;32(6):1052- 1062. e8, and Cao A. et al., J Clin Invest 2021;131(10):e141279).  
+
+Comment 15: Fig. 4f: what are all the other (non- marked) bands representing. Or is this not relevant? Please explain. I guess the AB cocktail was also used here? We appreciate this question raised by the reviewer. The AB cocktail was used as indicated on Fig. 4f. While we have presented the identities of the unmarked bands (Extended Data Fig. 4h), it is important to clarify that these bands are not pertinent to our current focus, which centers on identifying mitochondrial supercomplexes.  
+
+Comment 16: Page 9; line 7: why is it surprising that NDUFS4 OE restored CI in- gel activity "even under HG conditions"?  
+
+We have removed "even" from the sentence in response to the reviewer's feedback. We were simply attempting to convey our surprise at the significant regulatory impact of Ndufs4 in modulating the effect of high glucose in several experimental experiments in this study.  
+
+Comment 17: Page 9; line 11: that NDUFS4 exerts a "regulatory effect" is not demonstrated by the preceding experimental results.  
+
+Based on our comprehensive in vivo and in vitro observations, it is evident that NDUFS4 exerts a regulatory influence over mitochondrial morphology, cristae remodeling, and supercomplexes formation. This assertion is supported by our findings that both the downregulation and overexpression of NDUFS4 yield distinct and opposing effects on mitochondrial and cristae/mitochondria structure, effectively modulating their morphology. However, in response to the reviewer's insights, we acknowledge the need for further clarification regarding the precise regulatory role of NDUFS4. Consequently, we are actively conducting additional experiments within our laboratory to delve deeper into this important aspect of NDUFS4 function.  
+
+Comment 18: Regarding the STOML2- related experiments, I wonder if the proposed binding of STOML2 to NDUFS4 protein is compatible with the currently available structural information on the CI holocomplex and CI in a respiratory supercomplex? In my opinion, this is the most potentially interesting finding of this work since all functional effects of NDUFS4 OE could be explained by the NDUFS4 increase leading to more fully assembled CI. How does the cell "know" that more other CI subunits are required in the NDUFS4- overexpressing case? This would mean that NDUFS4 might be (among) the rate- limiting subunits for CI assembly. Is this supported by the current literature on CI biogenesis?  
+
+We appreciate this interesting comment raised by the reviewer. Our Extended Data Fig. 4g suggests that CI is mainly incorporated into the RSC in podocytes. Therefore, we performed molecular docking analysis using the NDUFS4 structural data derived from the Cryo- EM resolved respiromase structure (PDB database 5XTB; https://www.rcsb.org/structure/5XTB) and the predicted human STOML2 structure (Contact- guided Iterative Threading ASSembly Refinement web tool (C- I- TASSER; https://zhanggroup.org/C- I- TASSER) since the crystal structure of STOML2 has not yet been resolved. We supplemented our initial findings with STOML2 deletion mutant studies which indicate that stomatin domain of the STOML2 protein plays a central role for its binding to NDUFS4. In the revised version, we provide new data to delineate further this interaction and show that within the stomatin domain, the \(\beta 2\) , \(\beta 3\) and \(\beta 4\)
+
+<--- Page Split --->
+
+strands are necessary for the binding of STOML2 with NDUFS4 (Fig. 6g,i). These results are also consistent with the molecular docking model. Taken together, our findings suggest that the interaction between NDUFS4 and STOML2 is necessary for the proper maintenance of cristae morphology, RSC integrity, and ETC function in podocytes. Interestingly, while Ndufs4 overexpression alone does not appear to trigger a cellular response to recruit additional CI subunits, the context changes in high glucose conditions where our data suggest that reduced levels of NDUFS4 result in compromised interactions between Ndufs4 and STOML2. This disruption leads to a disorganized cristae platform for RSC assembly leading to decreased RSC assembly and contributing to decreased CI function and mitochondrial dysfunction. Overexpressing Ndufs4 in diabetic podocytes restores this imbalance leading to enhanced CI stability, improved mitochondrial RSCs assembly and cristae morphology. We believe that the interaction between Ndufs4 and STOML2 could be an important mechanism adapting the CI assembly and function in response to metabolic cues.  
+
+## Reviewer #2  
+
+The paper by Mise et al has explored the role of the Complex I subunit Ndufs4, in the pathogenesis of diabetic kidney disease. The aim of this study was to assess ETC abundance, mitochondrial function and morphology in kidney podocytes in a mouse model of diabetes and whether this affected DKD progression. First, ETC abundance was determined by mass spectrometry in two mouse models of diabetes, the Ins2Akita mouse and the db/db mouse model. Complex I abundance and mRNA expression was decreased in podocytes of the Ins2Akita mouse and db/db mouse. Out of the several downregulated subunits of mitochondrial OXPHOS complexes, Ndufs4 was chosen as a target. Glomerular Ndufs4 protein was decreased in human biopsies from patients with DKD. Next, podocyte- specific Ndufs4- transgenic mice were generated and crossed onto the Ins2Akita model of type 1 diabetes. Podocyte- specific Ndufs4- transgenic mice were partially protected from the diabetic kidney disease phenotype. Similar data were shown in the db/db mouse model. Ndufs4 overexpression in podocytes rescued high glucose- induced mitochondrial cristae changes as well as supercomplex assembly. A molecular interaction was identified between STOML2 and NDUFS4. The authors conclude that the findings represent a major paradigm shift in the current management of DKD by suggesting that targeting ETC remodeling could be a promising approach for developing therapies to mitigate the progression of DKD.  
+
+## Major comments  
+
+Comment 1: ETC remodeling, mitochondrial function and morphology in kidney podocytes in rodent models with diabetes has been widely reported, including from the current group. Several other studies have explored the role of mitochondrial Complex I subunits, leading to ETC remodeling in the development of chronic kidney disease (e.g., PMID: 23320803), therefore the idea itself is not novel. These papers should be referenced. Many others have suggested that targeting the electron transport chain is an approach to treat CKD/DKD (PMID: 36781216), therefore the author's suggestion of a major paradigm shift should be moderated  
+
+We concur with the reviewer's comment on the potential impact of mitochondrial dysfunction as a prominent factor implicated in the pathogenesis of kidney diseases, including diabetic kidney disease (DKD). However, the precise nature of mitochondrial dysfunction and the molecular mechanisms responsible for ETC dysfunction in podocytes leading to the
+
+<--- Page Split --->
+
+progression of DKD remain largely unknown. Our findings provide detailed insights into the pathobiology of mitochondrial respiration in podocytes and its central role in the pathogenesis of DKD. We provide evidence that Ndufs4 ties CI integrity to high glucose metabolic cues in the cell and the progression of DKD. While several studies have indeed suggested that compromised ETC function may serve as a risk factor for CKD, no prior investigations, to the best of our knowledge, have definitively demonstrated that improving CI structure/function can effectively reverse key features of DKD. Importantly, our study has provided strong experimental data to support an important biological function of Ndufs4 in preserving the integrity of cristae morphology. This novel function, in turn, exerts a protective influence on the progression of DKD. While previous studies have explored the regulatory impact of cristae- shaping proteins on supercomplexes, our work goes a step further by demonstrating for the first time that the overexpression of Ndufs4 in the diabetic milieu yields modulatory effects on cristae- shaping proteins and cristae morphology. We firmly believe that our study has introduced a completely new area of research, shedding light on the interplay between mitochondrial complex subunits and cristae forming proteins in the context of kidney pathology.  
+
+Comment 2: It is not clear which method was used to determine Complex I activity. The method referenced in the paper (ref 43), reports using NBT on tissue sections. This is not a generally accepted method. The gold standard method used by mitochondrial laboratories for determination of complex I activity is via respiratory chain enzymology using kinetic spectrophotometric assays (Frazier AE et al., Assessment of mitochondrial respiratory chain enzymes in cells and tissues, Methods Cell Biol 2020). Each assay requires specific inhibitors to ensure that the assay reflects true activity of the specific respiratory chain enzymology complex being assayed, i.e., rotenone- sensitive CI activity. In order to avoid misinterpretation, the mitochondrial marker enzyme citrate synthase is typically assayed to enable correction for mitochondrial content by expressing enzymes as citrate synthase ratios. Complex I activity should be determined using the kinetic spectrophotometric assay outlined in Frazier.  
+
+We appreciate the reviewer's comment. To answer the concern raised by the reviewer, we also employed kinetic spectrophotometric assays following the protocol detailed by Frazier AE et al as suggested by the reviewer<sup>3</sup> (Frazier AE. et al., Methods Cell Biol 2020;155:121- 156). As with our previously shown CI enzymatic activity assays, the results from these kinetic spectrophotometric assays reveal a reduction in rotenone- sensitive CI enzymatic activity within podocyte mitochondria derived from Ins2<sup>Akita/+</sup> as well as in Lepr<sup>db/db</sup> compared to control mice (Fig 1d). However, in regards to normalization using citrate synthase activity, it is worth noting that while using citrate synthase may be suitable in other conditions, our mitochondrial proteomic analysis indicate a significant reduction in citrate synthase abundance in podocytes from Ins2<sup>Akita/+</sup> mice compared to those from WT mice (Ins2<sup>Akita/+</sup> to WT ratio: 0.63). Western blotting also showed similar results (shown below). Therefore, relying on citrate synthase to normalize measurements in the diabetic condition could potentially lead to an overestimation. Consequently, we opted for an alternative approach, normalizing our data using total mitochondrial protein input as recommended by the Frazier's protocol. Of note, our initial assessment of CI activity involved NADH oxidase staining on frozen tissue sections obtained from mouse kidneys. This approach, widely utilized in the analysis of CI in frozen tissue samples<sup>20,21</sup> (Kruse SE. et al., Cell Metab 2008;7(4):312- 20 and Salagre D. et al., Antioxidants 2023;12(8):1499), served as an initial means to assess CI activity. For a more precise measurement of CI, we employed  
+
+WT ratio: 0.63). Western blotting also showed similar results (shown below). Therefore, relying on citrate synthase to normalize measurements in the diabetic condition could potentially lead to an overestimation. Consequently, we opted for an alternative approach, normalizing our data using total mitochondrial protein input as recommended by the Frazier's protocol. Of note, our initial assessment of CI activity involved NADH oxidase staining on frozen tissue sections obtained from mouse kidneys. This approach, widely utilized in the analysis of CI in frozen tissue samples<sup>20,21</sup> (Kruse SE. et al., Cell Metab 2008;7(4):312- 20 and Salagre D. et al., Antioxidants 2023;12(8):1499), served as an initial means to assess CI activity. For a more precise measurement of CI, we employed  
+
+![PLACEHOLDER_11_0]
+
+
+<--- Page Split --->
+
+mitochondria from primary podocytes placed on a 96- well assay plate precoated with cocktail antibodies against CI subunits. This method is also a widely accepted approach<sup>4-6</sup> (Wang T et al., Cell Metab 2021;33(3)531- 546. e9; Balsa E et al., Mol Cell 2019;74(5):877- 890. e6; and Cao LL et al., Nature 2016;539(7630):575- 578). Nevertheless, we agree with the reviewer that using different established methods will add to the validity of our experimental approach.  
+
+Comment 3: "We also evaluated NDUFS4 staining in glomeruli from diabetic subjects with a wide spectrum of DKD histology, and found that NDUFS4 staining in glomeruli was progressively reduced with worsening of DKD histology (test for trend \(\mathsf{P}< 0.01\) ) (Fig. 1l)." Was there a significant difference from Class I CKD to Class III CKD, or between any of the DKD classes? Can you explain the Test for trend? Is it really a progressive decrease in glomerular Ndufs4 staining?  
+
+When specifically comparing different Classes of DKD, there is a significant difference in glomerular NDUFS4 staining between Class I and Class IV kidneys, where the difference reached  
+
+![PLACEHOLDER_12_0]
+  
+
+statistical significance \((\mathsf{P} = 0.007)\) (see figure below). To explore the trend further, we conducted a linear trend analysis. A linear trend analysis, also known as a test for linear trend<sup>22,23</sup> (Nowak N. et al., Kidney Int 2016; 89:459- 467 and Sakaguchi Y. et al., J Am Soc Nephrol 2018; 29:991- 999), is a statistical method used to examine whether there is a "systematic and linear relationship between a set of ordered groups or categories and a measured variable." It is often applied when there is a natural order or progression in the groups being analyzed. The key idea behind linear trend analysis is to assess whether there is a consistent change in the variable of interest as you move from one group to the next. This change is evaluated to determine if it follows a linear pattern, meaning that as you go from one group to the next (e.g., from low to high levels of a variable or from one category to another), there is a consistent increase or decrease in the variable. Based on the results of the trend analysis, we conclude that glomerular NDUFS4 staining exhibits a progressive decline corresponding to the increase in glomerular pathological class of DKD, providing valuable insights into the disease progression.  
+
+Comment 4: Fig1k. Statistical analysis was performed on a sample size of \(n = 2\) within the microalbuminuria group. How is this possible when using one way- analysis of variance followed by Tukey's multiple comparisons test in Graphpad Prism? Please confirm the statistical test used.  
+
+We acknowledge the limited statistical power within the microalbuminuria group, although it's worth noting that GraphPad did provide the capability to perform a one- way ANOVA followed by Tukey's- Kramer multiple comparison test (https://www.graphpad.com/support/faqid/591/). However, following consultation with our biostatistician and in response to the comment by the reviewer, we have made the decision to combine all DKD patients with albuminuria into a single group. Consequently, we have updated Figure 1k to reflect this modification.
+
+<--- Page Split --->
+
+Comment 5: What was the mitochondrial density of the podocytes in diabetic mice? Ins2Akita and dbdb mice.  
+
+Using mitochondrial DNA copy number as an indicator of mitochondrial density, we have generated additional data that demonstrate reduced mitochondrial density in podocytes from Ins2Akita/+ mice when compared to their wild- type (WT) counterparts. We have previously established a similar pattern, illustrating lower mitochondrial copy numbers in podocytes from Lepr\(^{db/db}\) mice in comparison to their Lepr\(^{db/+}\) littermates\(^{16}\) (Long J. et al., J Clin Invest 2016; 126(11):4205)  
+
+![PLACEHOLDER_13_0]
+  
+
+Comment 6: Fig 3n, ATP production. What are the units of the ATP assay? Both the mitosox and ATP data are expressed as relative to control. As is the Complex I. The ATP assay used appears to be a one- step single reagent assay for detection of viable cells in culture, rather than an assay which can reliably determine ATP production over time in a sensitive manner. I would caution against using such an assay when screening for differences between groups. Was a standard curve generated and used? Did the authors account for cell number or total protein? The same questions apply to the mitosox assay and data. Representative histograms showing mean intensity of MitoSOX fluorescence should be shown for each group.  
+
+In the ATP assay, we made a standard curve to measure ATP concentration. In addition, we generated a standard curve to obtain cell numbers based on the DNA concentration using CyQUANT Cell proliferation Assay Kit (Molecular Probes). In response to the reviewer's comments, however, we also performed a real- time label- free ATP assay using Seahorse Analyzer in live primary podocytes to assess real- time total cellular and mitochondrial ATP production rates. As shown in Fig. 3p,q, total and mitochondrial ATP production rates were significantly reduced in podocytes from Ins2Akita/+ mice, while were normalized in podocytes from Ins2Akita/+- Ndufs4podTg mice. Regarding ATP content, MitoSOX and CI enzymatic activity, please refer to our response to comments 3, 10, and 11 to Reviewer #1.  
+
+Comment 7: Some of the individual data points on graphs are missing through the manuscript (e.g., Fig 2h, k, I & m; Figure 3g, i, j, k, l) and throughout the paper.  
+
+We did not originally include the data points in some of the figures since our understanding was that data points should only be shown when \(n< 10\) based on the Journal's guidelines. However, we have added the data points in those figures as well in response to the reviewer.  
+
+Comment 8: Fig 2a. What sample is contained in each lane in the genotyping gel? Sample labels have been provided below the gel image for your reference. To clarify further, the first three lanes represent PCR results from three wild type (WT) samples, while the last three lanes correspond to PCR results from three Ndufs4podTg mice.  
+
+Comment 9: The mRNA expression of Ndufs4 appears higher in the "podocyte depleted" fraction of renal cortex vs the podocyte fraction. It is important to show Ndufs4 protein in podocytes vs tubules and total renal cortex.  
+
+The expression levels of Ndufs4 mRNA are indeed lower in podocytes compared to the podocytes- depleted fraction, which predominately consists of kidney tubules. In the revised version of this manuscript, we have included Western blot data illustrating the levels of NDUFS4 in both podocytes and tubules (Extended Data Fig. 1h). Additionally, we have mentioned in the text (page 5) that the NDUFS4 protein expression in kidney tubular cells
+
+<--- Page Split --->
+
+remains unchanged in diabetic mice. It is important to highlight that the mitochondrial density in tubules greatly exceeds that in podocytes. This difference in mitochondrial density may account for the elevated expression of Ndufs4 in tubules in comparison to podocytes.  
+
+Comment 10: For the Seahorse cell culture studies, how many independent cell culture experiments are shown? It is stated that there were \(n = 6 - 8\) replicates per group. Are these data derived from one cell culture experiment?  
+
+The original results were derived from a single cell culture experiment, in which primary podocytes were isolated and pooled from three mice in each experimental group. We have performed similar experiments with primary podocytes isolated from an additional three mice in each group. The results from these additional experiments corroborate with our initial findings, reinforcing the consistency and reliability of our results.  
+
+![PLACEHOLDER_14_0]
+  
+
+Comment 11: Extended data figure 5a. The western blots are not convincing. The quantitation needs to be shown. In mouse kidney, OPA1 generally has 5 isoforms. There should be 5 bands present. Samples need to be run on a gradient gel to reveal the five isoforms as per PMID: 26822084.  
+
+We have incorporated the quantification results obtained from Western blots and conducted an analysis of OPA1 using gradient gel electrophoresis. This analysis was performed on primary podocytes isolated from four distinct groups of mice (Extended Data Figure 5a). Of note, OPA1 does not appear to exhibit 5 distinct isoform bands, possibly due to cell or tissue- specific variations in isoform patterns.  
+
+Comment 12: Page 4, line 13. "ETC dysfunction is recognized as an important cause of organ failure in several human pathologies including heart failure, diabetes, and neurodegeneration in a tissue- specific manner". Mitochondrial cytopathies have long been recognized to lead to kidney failure. This should be mentioned in the introduction (PMID: 33305107).  
+
+We thank the reviewer and have added these important citations to the manuscript<sup>24,25</sup> (Schijvens AM et al., Kidney Int Rep 2020; 5(12):2146- 2159; and Emma F et al., Nat Rev Nephrol 2016; 12(5):267- 280).  
+
+## Reviewer #3:  
+
+In this manuscript, Mise et al. examine the importance of the mitochondrial ETC Complex I
+
+<--- Page Split --->
+
+protein Ndufs4, in the maintenance of mitochondrial structure in the context of the diabetic kidney. The Authors show that Ndufs4 plays a role in the regulation of cristae structure and supercomplex evidence of binding partners in the process. The studies address an interesting topic, and the experimentation appears thorough and well- executed. Specific comments are indicated below.  
+
+Comment 1: It is not entirely clear why the Authors chose to focus on Ndufs4. The heat map in Fig.1 shows many ETC proteins, including in Complex I that were also decreased as a result of diabetes. In fact, there are some that were decreased by a greater extent. The decision to choose Ndufs4 comes across as a biased decision without scientific justification. A greater case should be made for choosing this target. In the Discussion, the Authors indicate that examination of other subunits would be meaningful, and this is appreciated, but does not preclude the need for a greater justification for the Ndufs4 focus.  
+
+We appreciate the reviewer's comment. Please also refer to our response to Reviewer #1 (comment #4). Briefly, we chose NDUFS4 among several CI subunits in our proteomic experiment for several reasons: Subsequent validation studies in other experimental models of diabetes consistently revealed NDUFS4 as the most prominently downregulated subunit in diabetes when compared to nondiabetic controls, further substantiating its significance. In contrast, several other potential candidates, including NDUFA2, NDUFB3, NDUFB4, NDUFB5, NDUFB8, NDUFB11, and NDUFV3 were not consistently downregulated (Extended Data Fig. 1e,g). Importantly, examination of human samples of individuals with DKD indicated that the downregulation of NDUFS4 occurs prior to the onset of albuminuria in patients, suggesting its potential role as an early marker of dysfunction.  
+
+Comment 2: Introduction, page 4; In the last sentence, the Authors state that their "results unexpectedly reveal that ETC integrity determines the stability of RSCs." This seems disingenuous. Do the Authors really believe that the very integrity of the components of the ETC would not be an important factor in the stability of the supercomplex in which they reside? My suggestion would be to either remove or soften this statement.  
+
+We appreciate the reviewer's suggestion. In the revised manuscript, we have removed the term "unexpectedly" from the content. Our intention simply was to convey that we did not anticipate that Ndufs4 overexpression would impact not only CI activity, but also RSC stability.  
+
+Comment 3: The use of multiple experimental models of diabetes (Akita, db/db) and human patient samples is appreciated and strengthen the conclusions and relevance of the studies. With that being said, the inclusion of complete db/db data (especially proteomic profiling in Fig.1b) is glaringly absent. Inclusion of this data makes a stronger case for commonality in the models and the ultimate focus on Ndufs4.  
+
+We thank the reviewer's suggestion and have performed mitochondrial proteomic analysis using mitochondria isolated from podocytes of Leprdb/+ and Leprdb/db mice. The data showed that NDUFS4 was one of the most downregulated among the CI subunits. This gave us stronger support in selecting NDUFS4 in this study. We have included the new data in Fig. 1e and Extended Data Fig. 1d.
+
+<--- Page Split --->
+
+1 Cabrera- Orefice, A., Potter, A., Evers, F., Hevler, J. F. & Guerrero- Castillo, S. Complexome profiling- exploring mitochondrial protein complexes in health and disease. Front. Cell Dev. Biol. 9, 796128 (2021).2 Válikangas, T., Suomi, T. & Elo, L. L. A systematic evaluation of normalization methods in quantitative label- free proteomics. Brief Bioinform. 19, 1- 11 (2018).3 Frazier, A. E., Vincent, A. E., Turnbull, D. M., Thorburn, D. R. & Taylor, R. W. Assessment of mitochondrial respiratory chain enzymes in cells and tissues. Methods Cell Biol. 155, 121- 156 (2020).4 Wang, T. et al. C9orf72 regulates energy homeostasis by stabilizing mitochondrial complex I assembly. Cell Metab. 33, 531- 546. e539 (2021).5 Balsa, E. et al. ER and nutrient stress promote assembly of respiratory chain supercomplexes through the PERK- eIF2alpha axis. Mol. Cell 74, 877- 890. e876 (2019).6 Cao, L. L. et al. Control of mitochondrial function and cell growth by the atypical cadherin Fat1. Nature 539, 575- 578 (2016).7 Falkevall, A. et al. Reducing VEGF- B Signaling Ameliorates Renal Lipotoxicity and Protects against Diabetic Kidney Disease. Cell Metab. 25, 713- 726 (2017).8 Wang, W. et al. Mitochondrial fission triggered by hyperglycemia is mediated by ROCK1 activation in podocytes and endothelial cells. Cell Metab. 15, 186- 200 (2012).9 Galvan, D. L. et al. Drp1S600 phosphorylation regulates mitochondrial fission and progression of nephropathy in diabetic mice. J. Clin. Invest. 129, 2807- 2823 (2019).10 Sutandy, F. X. R., Gößner, I., Tascher, G. & Münch, C. A cytosolic surveillance mechanism activates the mitochondrial UPR. Nature 618, 849- 854 (2023).11 Labuschagne, C. F., Cheung, E. C., Blagih, J., Domart, M. C. & Vousden, K. H. Cell Clustering Promotes a Metabolic Switch that Supports Metastatic Colonization. Cell Metab. 30, 720- 734. e725 (2019).12 Murphy, M. P. et al. Guidelines for measuring reactive oxygen species and oxidative damage in cells and in vivo. Nat Metab 4, 651- 662 (2022).13 Galvan, D. L. et al. Real- time in vivo mitochondrial redox assessment confirms enhanced mitochondrial reactive oxygen species in diabetic nephropathy. Kidney Int. 92, 1282- 1287 (2017).14 Waypa, G. B. et al. Hypoxia triggers subcellular compartmental redox signaling in vascular smooth muscle cells. Circ. Res. 106, 526- 535 (2010).15 Moullan, N. et al. Tetracyclines Disturb Mitochondrial Function across Eukaryotic Models: A Call for Caution in Biomedical Research. Cell Rep. 10, 1681- 1691 (2015).16 Long, J. et al. Long noncoding RNA Tug1 regulates mitochondrial bioenergetics in diabetic nephropathy. J. Clin. Invest. 126, 4205- 4218 (2016).17 Qi, W. et al. Pyruvate kinase M2 activation may protect against the progression of diabetic glomerular pathology and mitochondrial dysfunction. Nat. Med. 23, 753- 762 (2017).18 Fu, Y. et al. Elevation of JAML Promotes Diabetic Kidney Disease by Modulating Podocyte Lipid Metabolism. Cell metabolism 32, 1052- 1062. e1058 (2020).19 Cao, A. et al. DACH1 protects podocytes from experimental diabetic injury and modulates PTIP- H3K4Me3 activity. The Journal of clinical investigation 131 (2021).20 Kruse, S. E. et al. Mice with mitochondrial complex I deficiency develop a fatal encephalomyopathy. Cell Metab. 7, 312- 320 (2008).21 Salagre, D., Raya Alvarez, E., Cendan, C. M., Aouichat, S. & Agil, A. Melatonin Improves Skeletal Muscle Structure and Oxidative Phenotype by Regulating Mitochondrial Dynamics and Autophagy in Zücker Diabetic Fatty Rat. Antioxidants (Basel) 12 (2023).22 Nowak, N. et al. Increased plasma kidney injury molecule- 1 suggests early progressive renal decline in non- proteinuric patients with type 1 diabetes. Kidney Int 89, 459- 467 (2016).
+
+<--- Page Split --->
+
+23 Sakaguchi, Y., Hamano, T., Wada, A., Hoshino, J. & Masakane, I. Magnesium and Risk of Hip Fracture among Patients Undergoing Hemodialysis. Journal of the American Society of Nephrology: JASN 29, 991- 999 (2018).24 Schijvens, A. M. et al. Mitochondrial Disease and the Kidney With a Special Focus on CoQ(10) Deficiency. Kidney Int. Rep. 5, 2146- 2159 (2020).25 Emma, F., Montini, G., Parikh, S. M. & Salviati, L. Mitochondrial dysfunction in inherited renal disease and acute kidney injury. Nat. Rev. Nephrol. 12, 267- 280 (2016).
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+I highly appreciate the inclusion of additional experiments, as well as the thorough rebuttal, who have taken away several of my concerns. However, I feel that some of my previous suggestions deserve further attention:  
+
+1. Regarding Fig. 1b: I appreciate the new table but still cannot easily determine if these changes are significant; please include p-values in the new table (Extended data Fig. 1d)  
+
+2. Although the authors present some arguments (e.g. their reply to comment 4, 6 and 17), I'm still not convinced that the observed effects are Ndufs4-specific. The accessory NDUFS4 protein is one of the essential CI subunits. Unless it has an additional function(perhaps the role described in this manuscript?), a drop in NDUFS4 protein levels will always induce a drop in the total protein level of assembled CI. The latter will induce a drop in protein level of virtually all CI subunits. In my opinion, the described effects all can be explained by a reduction in the level of fully assembled CI and therefore cannot be regarded as ndufs4-specific?  
+
+3. Extended data Fig. 3b: the y-axis title contains an error. Moreover, in the main text the authors state that: "...ECAR... also showed similar changes..". What does this mean? If there really is a drop in OCR, I would expect an increase in ECAR? As described in the literature, ECAR is not exclusively a measure of glycolytic rate (e.g. lactate production); please discuss the OCR and ECAR results in an integrated manner.  
+
+4. Regarding my previous comments 8, 9,12 and 14: please ensure that the info provided in the author's rebuttal is included in the manuscript.  
+
+5. I appreciate the use of roGFP (I guess it is roGFP1?). This contains an S=S bridge, which is broken or formed based upon its thiol redox environment. Please interpret the results obtained with this sensor as a measure of thiol redox state (this off course is hydrogen peroxide sensitive in most systems).  
+
+6. Regarding my previous comment 15. The Extended data Fig. 4h referred to in the rebuttal is not on supercomplexes but on ATP levels?  
+
+7. Regarding the NDFUS4-STOML2 interaction. In the rebuttal there is referred to Extended data Fig. 4g, but this panel has nothing to do with STOML2. I do not understand the choice for SXTB, which represents only part of the CI matrix arm. Please investigate whether the STOML2 binding sites on the NDUFS4 protein are accessible when NDUFS4 is incorporated in fully assembled CI and the latter is part of the RSC. This will provide insight on whether STOML2 can bind to NDUFS4 when the latter is within the ETC supercomplex.  
+
+## Reviewer #2 (Remarks to the Author):  
+
+The authors have responded to all of my queries satisfactorily. The manuscript is much improved.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+The Authors have addressed my concerns.
+
+<--- Page Split --->
+
+## Reviewer #1  
+
+Comment 1: Regarding Fig. 1b: I appreciate the new table but still cannot easily determine if these changes are significant; please include p- values in the new table (Extended data Fig. 1d)  
+
+We isolated podocytes from 8 wild type (WT) and 8 diabetic Ins2Akit/+ mice (Fig1. Legend, a). Subsequently, we pooled all the podocytes within each group for a comprehensive screening of differentially expressed mitochondrial subunits in our proteomic analysis. While this pooling strategy might have limited the statistical significance of individual subunits when comparing WT and diabetic mice, we addressed this limitation by incorporating and validating our results using additional approaches, including qRT- PCR analyses (Extended Data Fig. 1e- g) and using an alternative diabetic Leprdb/db mouse model.  
+
+Comment 2: Although the authors present some arguments (e.g., their reply to comment 4, 6 and 17), I'm still not convinced that the observed effects are Ndufs4- specific. The accessory NDUFS4 protein is one of the essential CI subunits. Unless it has an additional function (perhaps the role described in this manuscript?), a drop in NDUFS4 protein levels will always induce a drop in the total protein level of assembled CI. The latter will induce a drop in protein level of virtually all CI subunits. In my opinion, the described effects all can be explained by a reduction in the level of fully assembled CI and therefore cannot be regarded as Ndufs4- specific?  
+
+Our findings strongly suggest that the observed effects on mitochondrial and kidney function described in this study are indeed Ndufs4- specific since we specifically targeted Ndufs4 in our experimental approach. However, it is important to clarify that while our results support the critical role of Ndufs4 in the observed effects, this does not imply exclusivity. Our findings suggest that Ndufs4 overexpression through its interaction with STOML2 stabilizes cristae morphology, providing a platform for assembly of the mitochondrial respiratory chain complexes. However, we recognize that these findings do not negate the potential role of other CI subunits in sustaining CI integrity and enhancing overall mitochondrial function. Indeed, we concur with the reviewer that improving CI and mitochondrial dysfunction in the diabetic environment may not be confined solely to Ndufs4, recognizing the plausible involvement of other CI subunits in CI integrity and improved mitochondrial function. We have highlighted this point in the Discussion.  
+
+Comment 3: Extended data Fig. 3b: the y- axis title contains an error. Moreover, in the main text the authors state that: "...ECAR... also showed similar changes.". What does this mean? If there really is a drop in OCR, I would expect an increase in ECAR. As described in the literature, ECAR is not exclusively a measure of glycolytic rate (e.g., lactate production); please discuss the OCR and ECAR results in an integrated manner.  
+
+We appreciate the reviewer's comment. It seems that an error occurred during the PDF conversion, and original \(10^{5}\) was mistakenly displayed as \(10^{1}\) . We have now rectified this error in the revised version. Regarding the interpretation of ECAR results, the interplay between OCR and ECAR in diabetic kidney disease is indeed complex. Consistent with our findings, a recent publication has provided evidence that chronic hyperglycemia results in lower ECAR and OCR levels in podocytes. However, we took the reviewer's comment as an
+
+<--- Page Split --->
+
+opportunity to further clarify the relationship between aberrant OCR and ECAR in our experimental model. We have now added a new figure (Extended Data Fig. 3c), elucidating the energy maps of primary podocytes isolated from the four distinct experimental groups of mice by combining OCR and ECAR data. The overall goal was to understand how the podocytes energy metabolism adapts and changes under diabetic environment. This graphical representation suggests that podocytes derived from diabetic \(Ins^{2Akit / + }\) mice are inefficient in energy production in diabetic conditions, whereas overexpression of Ndufs4 in diabetic \(Ins^{2Akit / + }\) ; Ndufs4<sup>PodTg</sup> mice exhibit an energy map similar to the podocytes from WT mice, indicating that they use both mitochondrial respiration and glycolysis for energy production.  
+
+Comment 4: Regarding my previous comments 8, 9, 12 and 14: please ensure that the info provided in the author's rebuttal is included in the manuscript.  
+
+We have incorporated additional information based on our response to the reviewer's comments in the revised manuscript. Most comments were already present in the previous revision manuscript and thereby are no longer highlighted in the current revised version. Please refer to pages 4, 9, 25, and 38.  
+
+Comment 5: I appreciate the use of roGFP (I guess it is roGFP1?). This contains an S=S bridge, which is broken or formed based upon its thiol redox environment. Please interpret the results obtained with this sensor as a measure of thiol redox state (this of course is hydrogen peroxide sensitive in most systems).  
+
+As previously described by our group and others<sup>2,3</sup>, we used mito- roGFP2, a mitochondrial matrix- targeted ratiometric redox- sensitive green fluorescent protein. This sensor contains two adjacent cysteine residues that do not interact in reduced thiol redox state where the sensor exhibits high emission at 525 nm when excited at 488 nm, and relatively low emission at 525 nm when excited at 405 nm. Protein thiol oxidation mediated by \(H_2O_2\) leads to the formation of a disulfide linkage, resulting in a conformational change in the sensor that increases emission at 525 nm during excitation at 405 nm, and decreases emission at 525 nm when excited at 488 nm<sup>2,3</sup>. We measured mito- roGFP ratios in live mitochondria using confocal microscopy, in live cells. We found that primary podocytes from diabetic mice showed an increase in oxidized/reduced mito- roGFP ratios compared to those from WT mice (Fig. 3o), suggesting that mitochondrial thiol oxidation increases in diabetic podocytes. On the other hand, primary podocytes from diabetic mice overexpressing Ndufs4 show significantly reduced oxidized/reduced mito- roGFP ratios compared with podocytes from diabetic mice, indicating Ndufs4 overexpression prevents enhanced mROS in diabetes. We added this information in the text and Methods.  
+
+Comment 6: Regarding my previous comment 15. The Extended data Fig. 4h referred to in the rebuttal is not on supercomplexes but on ATP levels?  
+
+In the revised manuscript, the data are shown in Extended data Fig. 4i.  
+
+Comment 7: Regarding the NDFUS4- STOML2 interaction. In the rebuttal there is referred to Extended data Fig. 4g, but this panel has nothing to do with STOML2. I do not understand the choice for 5XTB, which represents only part of the CI matrix arm. Please investigate whether the STOML2 binding sites on the NDUFS4 protein are accessible when NDUFS4 is
+
+<--- Page Split --->
+
+incorporated in fully assembled CI and the latter is part of the RSC. This will provide insight on whether STOML2 can bind to NDUFS4 when the latter is within the ETC supercomplex.  
+
+In the revised version of the manuscript, the Extended Data Fig. 4j provides evidence that the CI is present almost exclusively in the form of stable RSCs in podocytes.  
+
+The structural information for macromolecules containing Ndufs4 was retrieved from the Protein Data Bank (PDB; https://www.rcsb.org/). There are four entries in PDB with macromolecules containing NDUFS4: 5XTB (matrix arm of CI), 5XTC (CI), 5XTH (RSC with I1ll2IV1 stoichiometry), and 5XTI (I2ll2IV2 stoichiometry), each with a slightly different resolution. We chose PDB- 5XTB for further study because of its highest resolution (3.40A). To address whether NDUFS4 protein is accessible to STOML2 in the context of RSC, we used the molecular structure visualization and analysis tool ChimeraX (UCSF ChimeraX, https://www.cgl.ucsf.edu/chimeraX/), a powerful molecular modeling engine, to analyze the structure of the I1ll2IV1 within RSC resolved by CryoEM (PDB- 5XTH). This program shows that CIII and CIV bind to the hydrophobic membrane arm of CI, forming the RSC. Furthermore, it reveals that the NDUFS4 protein (magenta, Extended Data Fig. 6c) is localized on the surface of the matrix arm of the CI within the RSC, exposing NDUFS4 to potential interactions with other molecules such as STOML2. Our experimental data with STOML2 knockout cells and STOML2 mutation studies provide additional support to this interaction. Overall, these findings shed light on potential interactions involving NDUFS4 and STOML2. The figure was added to the manuscript (Extended Data Fig. 6c).  
+
+structure of the I1ll2IV1 within RSC resolved by CryoEM (PDB- 5XTH). This program shows that CIII and CIV bind to the hydrophobic membrane arm of CI, forming the RSC. Furthermore, it reveals that the NDUFS4 protein (magenta, Extended Data Fig. 6c) is localized on the surface of the matrix arm of the CI within the RSC, exposing NDUFS4 to potential interactions with other molecules such as STOML2. Our experimental data with STOML2 knockout cells and STOML2 mutation studies provide additional support to this interaction. Overall, these findings shed light on potential interactions involving NDUFS4 and STOML2. The figure was added to the manuscript (Extended Data Fig. 6c).  
+
+![PLACEHOLDER_21_0]
+  
+
+## References:  
+
+1. Qi, W. et al. Pyruvate kinase M2 activation may protect against the progression of diabetic glomerular pathology and mitochondrial dysfunction. Nat. Med. 23, 753-762 (2017).  
+2. Waypa, G. B. et al. Hypoxia triggers subcellular compartmental redox signaling in vascular smooth muscle cells. Circ. Res. 106, 526-535 (2010).  
+3. Galvan, D. L. et al. Real-time in vivo mitochondrial redox assessment confirms enhanced mitochondrial reactive oxygen species in diabetic nephropathy. Kidney Int. 92, 1282-1287 (2017).
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+I thank the authors for their rebuttal, which have addressed my remaining concerns.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1e1bedff740ea0146a02a6d351b3babf234584dc6acf3acb3873b690d8852174/supplementary_0_Peer Review File__1e1bedff740ea0146a02a6d351b3babf234584dc6acf3acb3873b690d8852174_det.mmd

@@ -0,0 +1,533 @@
+<|ref|>title<|/ref|><|det|>[[61, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[66, 110, 361, 138]]<|/det|>
+Peer Review File  
+
+<|ref|>title<|/ref|><|det|>[[106, 214, 890, 270]]<|/det|>
+# NDUFS4 Regulates Cristae Remodeling in Diabetic Kidney Disease  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 910, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[119, 84, 357, 101]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 137, 448, 152]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 165, 861, 280]]<|/det|>
+This paper investigates the interesting hypothesis that the NDUFS4 accessory subunit of OXPHOS complex I is involved in the control/remodeling of mitochondrial cristae architecture. Conceptually, it presents extensive experimental data focusing on effects in NDUFS4 overexpression systems, as well as the potential interaction between NDUFS4 and STOML2. Although the scientific message of this paper certainly can be appealing, I still have several questions/concerns related to the used experimental strategies and proposed concepts.  
+
+<|ref|>text<|/ref|><|det|>[[118, 292, 667, 307]]<|/det|>
+Fig. 1b: it is unclear to me which CI subunits were significantly reduced  
+
+<|ref|>text<|/ref|><|det|>[[119, 320, 427, 334]]<|/det|>
+Fig. 1c: what does the Z- score indicate?  
+
+<|ref|>text<|/ref|><|det|>[[118, 348, 836, 390]]<|/det|>
+Fig. 1d (and others): why is CI activity expressed as NAD+/NADH? This ratio is dependent on many other metabolic pathways. For proper analysis of (rotenone- sensitive) CI activity more "classical" enzymatic assays in mitoplasts should be used.  
+
+<|ref|>text<|/ref|><|det|>[[118, 404, 877, 446]]<|/det|>
+Fig. 1h: Why was only NDUFS4 analysed? I guess there should also be significant correlations with other CI subunits? When such correlations are absent, this would strengthen the choice to focus on NDUFS4 for further analysis. When such correlations are present, why was NDUFS4 selected?  
+
+<|ref|>text<|/ref|><|det|>[[118, 459, 744, 474]]<|/det|>
+Fig. 1i- l: more details are needed on how NDUFS4 staining was exactly quantified  
+
+<|ref|>text<|/ref|><|det|>[[118, 487, 857, 516]]<|/det|>
+Page 6, lines 8- 10: in my view, the data suggests that CI expression is important, meaning that there is no evidence for NDUFS4- specific effects.  
+
+<|ref|>text<|/ref|><|det|>[[118, 529, 861, 572]]<|/det|>
+Fig. 3a: please compute the relevant respiratory control ratios (RCRs). These are independent of cell number and will highlight functional defects. Furthermore, there is no ECAR data presented. Why?  
+
+<|ref|>text<|/ref|><|det|>[[118, 585, 864, 614]]<|/det|>
+Fig. 3g (bargraphs): the average WT value is 1.0; how was data normalization performed for this and all relevant other figures?  
+
+<|ref|>text<|/ref|><|det|>[[118, 627, 878, 684]]<|/det|>
+Fig. 3h- n (and page 7; line 23): the presented parameters do not reflect mitochondrial "dynamics" but steady- state mitochondrial morphology. One can not talk about enhanced fission (it can also be reduced fusion) since mitochondrial fission/fusion proteins were not investigated. Moreover, it is unclear how the conclusion that "cristae morphology is altered" is reached.  
+
+<|ref|>text<|/ref|><|det|>[[118, 697, 866, 770]]<|/det|>
+Fig. 3m (page 8; line 2): MitoSOX does not specifically detect "mitochondrial ROS". The ethidium molecule (which is the ROS- detection moiety of MitoSOX) reacts extremely fast (but not exclusively with) superoxide. This means that MitoSOX can become fluorescent in the cytosol during its transit to the mitochondrial matrix (driven by the presence of the decylTPP moiety of MitoSOX).  
+
+<|ref|>text<|/ref|><|det|>[[118, 783, 870, 811]]<|/det|>
+Fig. 3n (page 8; line 2): the used assay reports on total cellular ATP content, so is not informative on ATP production.  
+
+<|ref|>text<|/ref|><|det|>[[118, 825, 638, 839]]<|/det|>
+page 8; line 6: what is meant with "mitochondrial reprogramming"?  
+
+<|ref|>text<|/ref|><|det|>[[118, 853, 870, 881]]<|/det|>
+Page 8; line 7: please rule out that the used DOX concentration and incubation regime impacts on mitochondrial function (described in the literature).  
+
+<|ref|>text<|/ref|><|det|>[[115, 894, 872, 909]]<|/det|>
+Page 8; line 10: it is not explained why HG is compared to NG and why this is relevant for the rest
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 84, 216, 97]]<|/det|>
+of the study.  
+
+<|ref|>text<|/ref|><|det|>[[118, 111, 850, 140]]<|/det|>
+Fig. 4f: what are all the other (non- marked) bands representing. Or is this not relevant? Please explain. I guess the AB cocktail was also used here?  
+
+<|ref|>text<|/ref|><|det|>[[118, 153, 840, 182]]<|/det|>
+Page 9; line 7: why is it surprising that NDUFS4 OE restored CI in- gel activity "even under HG conditions"?  
+
+<|ref|>text<|/ref|><|det|>[[118, 195, 852, 223]]<|/det|>
+Page 9; line 11: that NDUFS4 exerts a "regulatory effect" is not demonstrated by the preceding experimental results.  
+
+<|ref|>text<|/ref|><|det|>[[118, 237, 876, 351]]<|/det|>
+Regarding the STOML2- related experiments, I wonder if the proposed binding of STOML2 to NDUFS4 protein is compatible with the currently available structural information on the CI holocomplex and CI in a respiratory supercomplex? In my opinion, this is the most potentially interesting finding of this work since all functional effects of NDUFS4 OE could be explained by the NDUFS4 increase leading to more fully assembled CI. How does the cell "know" that more other CI subunits are required in the NDUFS4- overexpressing case? This would mean that NDUFS4 might be (among) the rate- limiting subunits for CI assembly. Is this supported by the current literature on CI biogenesis?  
+
+<|ref|>sub_title<|/ref|><|det|>[[120, 415, 448, 429]]<|/det|>
+## Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 442, 877, 540]]<|/det|>
+The paper by Mise et al has explored the role of the Complex I subunit Ndufs4, in the pathogenesis of diabetic kidney disease. The aim of this study was to assess ETC abundance, mitochondrial function and morphology in kidney podocytes in a mouse model of diabetes and whether this affected DKD progression. First, ETC abundance was determined by mass spectrometry in two mouse models of diabetes, the Ins2Akita mouse and the db/db mouse model. Complex I abundance and mRNA expression was decreased in podocytes of the Ins2Akita mouse and db/db mouse.  
+
+<|ref|>text<|/ref|><|det|>[[118, 541, 872, 682]]<|/det|>
+Out of the several downregulated subunits of mitochondrial OXPHOS complexes, Ndufs4 was chosen as a target. Glomerular Ndufs4 protein was decreased in human biopsies from patients with DKD. Next, podocyte- specific Ndufs4- transgenic mice were generated and crossed onto the Ins2Akita model of type 1 diabetes. Podocyte- specific Ndufs4- transgenic mice were partially protected from the diabetic kidney disease phenotype. Similar data were shown in the db/db mouse model. Ndufs4 overexpression in podocytes rescued high glucose- induced mitochondrial cristae changes as well as supercomplex assembly. A molecular interaction was identified between STOML2 and NDUFS4. The authors conclude that the findings represent a major paradigm shift in the current management of DKD by suggesting that targeting ETC remodeling could be a promising approach for developing therapies to mitigate the progression of DKD.  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 710, 248, 722]]<|/det|>
+## Major comments  
+
+<|ref|>text<|/ref|><|det|>[[118, 723, 878, 821]]<|/det|>
+ETC remodeling, mitochondrial function and morphology in kidney podocytes in rodent models with diabetes has been widely reported, including from the current group. Several other studies have explored the role of mitochondrial Complex I subunits, leading to ETC remodeling in the development of chronic kidney disease (e.g., PMID: 23320803), therefore the idea itself is not novel. These papers should be referenced. Many others have suggested that targeting the electron transport chain is an approach to treat CKD/DKD (PMID: 36781216), therefore the author's suggestion of a major paradigm shift should be moderated.  
+
+<|ref|>text<|/ref|><|det|>[[118, 835, 871, 906]]<|/det|>
+It is not clear which method was used to determine Complex I activity. The method referenced in the paper (ref 43), reports using NBT on tissue sections. This is not a generally accepted method. The gold standard method used by mitochondrial laboratories for determination of complex I activity is via respiratory chain enzymology using kinetic spectrophotometric assays (Frazier AE et al., Assessment of mitochondrial respiratory chain enzymes in cells and tissues, Methods Cell Biol
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 83, 870, 168]]<|/det|>
+2020). Each assay requires specific inhibitors to ensure that the assay reflects true activity of the specific respiratory chain enzymology complex being assayed, i.e., rotenone-sensitive CI activity. In order to avoid misinterpretation, the mitochondrial marker enzyme citrate synthase is typically assayed to enable correction for mitochondrial content by expressing enzymes as citrate synthase ratios. Complex I activity should be determined using the kinetic spectrophotometric assay outlined in Frazier.  
+
+<|ref|>text<|/ref|><|det|>[[118, 195, 870, 280]]<|/det|>
+"We also evaluated NDUFS4 staining in glomeruli from diabetic subjects with a wide spectrum of DKD histology, and found that NDUFS4 staining in glomeruli was progressively reduced with worsening of DKD histology (test for trend \(P< 0.01\) ) (Fig. 1l)." Was there a significant difference from Class I CKD to Class III CKD, or between any of the DKD classes? Can you explain the Test for trend? Is it really a progressive decrease in glomerular Ndufs4 staining?  
+
+<|ref|>text<|/ref|><|det|>[[118, 307, 866, 350]]<|/det|>
+Fig1K. Statistical analysis was performed on a sample size of \(n = 2\) within the microalbuminuria group. How is this possible when using one way- analysis of variance followed by Tukey's multiple comparisons test in Graphpad Prism? Please confirm the statistical test used.  
+
+<|ref|>text<|/ref|><|det|>[[118, 363, 864, 378]]<|/det|>
+What was the mitochondrial density of the podocytes in diabetic mice? Ins2Akita and db/db mice.  
+
+<|ref|>text<|/ref|><|det|>[[118, 391, 878, 504]]<|/det|>
+Fig 3n, ATP production. What are the units of the ATP assay? Both the mitosox and ATP data are expressed as relative to control. As is the Complex I. The ATP assay used appears to be a one- step single reagent assay for detection of viable cells in culture, rather than an assay which can reliably determine ATP production over time in a sensitive manner. I would caution against using such an assay when screening for differences between groups. Was a standard curve generated and used? Did the authors account for cell number or total protein? The same questions apply to the mitosox assay and data. Representative histograms showing mean intensity of MitoSOX fluorescence should be shown for each group.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 532, 247, 545]]<|/det|>
+## Minor comments  
+
+<|ref|>text<|/ref|><|det|>[[118, 546, 875, 574]]<|/det|>
+Some of the individual data points on graphs are missing through the manuscript (e.g., Fig 2h, k, l & m; Figure 3g, i, j, k, l) and throughout the paper.  
+
+<|ref|>text<|/ref|><|det|>[[118, 588, 877, 644]]<|/det|>
+Fig 2a. What sample is contained in each lane in the genotyping gel? Fig 2b. The mRNA expression of Ndufs4 appears higher in the "podocyte depleted" fraction of renal cortex vs the podocyte fraction. It is important to show Ndufs4 protein in podocytes vs tubules and total renal cortex.  
+
+<|ref|>text<|/ref|><|det|>[[118, 672, 868, 714]]<|/det|>
+For the Seahorse cell culture studies, how many independent cell culture experiments are shown? It is stated that there were \(n = 6 - 8\) replicates per group. Are these data derived from one cell culture experiment?  
+
+<|ref|>text<|/ref|><|det|>[[118, 728, 835, 770]]<|/det|>
+Extended data figure 5a. The western blots are not convincing. The quantitation needs to be shown. In mouse kidney, OPA1 generally has 5 isoforms. There should be 5 bands present. Samples need to be run on a gradient gel to reveal the five isoforms as per PMID: 26822084.  
+
+<|ref|>text<|/ref|><|det|>[[118, 784, 866, 840]]<|/det|>
+Introduction. Page 4, line 13. "ETC dysfunction is recognized as an important cause of organ failure in several human pathologies including heart failure, diabetes, and neurodegeneration in a tissue- specific manner. Mitochondrial cytopathies have long been recognised to lead to kidney failure. This should be mentioned in the introduction (PMID: 33305107)
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[120, 97, 448, 112]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 125, 875, 210]]<|/det|>
+In this manuscript, Mise et al. examine the importance of the mitochondrial ETC Complex I protein Ndufs4, in the maintenance of mitochondrial structure in the context of the diabetic kidney. The Authors show that Ndufs4 plays a role in the regulation of cristae structure and supercomplex formation, while providing evidence of binding partners in the process. The studies address an interesting topic, and the experimentation appears thorough and well- executed. Specific comments are indicated below.  
+
+<|ref|>text<|/ref|><|det|>[[118, 223, 875, 321]]<|/det|>
+1) It is not entirely clear why the Authors chose to focus on Ndufs4. The heat map in Fig.1 shows many ETC proteins, including in Complex I that were also decreased as a result of diabetes. In fact, there are some that were decreased by a greater extent. The decision to choose Ndufs4 comes across as a biased decision without scientific justification. A greater case should be made for choosing this target. In the Discussion, the Authors indicate that examination of other subunits would be meaningful, and this is appreciated, but does not preclude the need for a greater justification for the Ndufs4 focus.  
+
+<|ref|>text<|/ref|><|det|>[[118, 335, 875, 405]]<|/det|>
+2) Introduction, page 4; in the last sentence, the Authors state that their "results unexpectedly reveal that ETC integrity determines the stability of RSCs." This seems disingenuous. Do the Authors really believe that the very integrity of the components of the ETC would not be an important factor in the stability of the supercomplex in which they reside? My suggestion would be to either remove or soften this statement.  
+
+<|ref|>text<|/ref|><|det|>[[118, 419, 874, 490]]<|/det|>
+3) The use of multiple experimental models of diabetes (Akita, db/db) and human patient samples is appreciated and strengthen the conclusions and relevance of the studies. With that being said, the inclusion of complete db/db data (especially proteomic profiling in Fig.1b) is glaringly absent. Inclusion of this data makes a stronger case for commonality in the models and the ultimate focus on Ndufs4.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 62, 878, 116]]<|/det|>
+We greatly appreciate the comments/suggestions made by the reviewers. In the revised version of our manuscript, we have made a number of additional experiments that are now presented in 15 new panels that include:  
+
+<|ref|>text<|/ref|><|det|>[[118, 132, 876, 350]]<|/det|>
+- Rotenone-sensitive complex I enzymatic activity (Figs. 1d,2d; and Extended Data Fig. 4g)- Real-time ATP production (Figs. 3p,3q),- Mitochondrial ROS assessment using mito-roGFP, a specific mitochondrial matrix-targeted redox-sensitive reporter (Fig. 3o)- Proteomic data from type 2 diabetic mice Leprdb/+ and Leprdb/db mice (Figs. 1e and Extended Data Fig. 1d),- RCR and ECAR data (Extended Data Figs. 3a,3b)- GST-NDUFS4 pulldown assay to identify more specific binding sites between NDUFS4 and STOML2 (Figs. 6g,6i).- Excluding the DOX effect in our DOX-inducible model (Extended Data Figs. 4c, 4d, 4e)  
+
+<|ref|>text<|/ref|><|det|>[[92, 364, 806, 384]]<|/det|>
+The additions to the manuscript are highlighted in yellow for ease of the reviewers.  
+
+<|ref|>sub_title<|/ref|><|det|>[[92, 417, 207, 436]]<|/det|>
+## Reviewer #1  
+
+<|ref|>text<|/ref|><|det|>[[92, 451, 844, 540]]<|/det|>
+Comment 1: Fig. 1b: it is unclear to me which CI subunits were significantly reduced. In response to the reviewer's comment, we have since introduced an informative table (Extended Data Fig 1d) in addition to the previously shown Fig 1b. This table serves to outline the CI subunits with significantly reduced values, providing a convenient and comprehensive reference for those specific CI subunits.  
+
+<|ref|>text<|/ref|><|det|>[[92, 556, 555, 574]]<|/det|>
+Comment 2: Fig. 1c: what does the Z- score indicate?  
+
+<|ref|>text<|/ref|><|det|>[[90, 574, 903, 802]]<|/det|>
+In the original manuscript, we employed Z- scores, a widely used statistical tool for proteomic analysis, primarily to indicate the extent to which a specific data point is deviated from the mean, measured in terms of standard deviations. In the revised version of the manuscript, we adopted an alternative approach by using adjusted intensity- based absolute quantification (iBAQ) values. Adjusted iBAQ values represent protein abundance estimates that have been normalized for meaningful comparison between samples \(^{1,2}\) (Cabrera- Orefice A. et al Front Cell Dev Biol 2022; 9:796128 and Válikangas T. et al., Brief Bioinform 2018;19(1):1- 11). Both methods are commonly used in proteomic analysis, but they provide different types of information. Whereas Z- scores can help identify proteins of interest based on their deviation from the mean, adjusted iBAQ values provide information about the relative abundance of proteins across different conditions or samples. Furthermore, we have now included heatmaps generated using these adjusted iBAQ values for each complex subunit (Extended Data Figure 1b).  
+
+<|ref|>text<|/ref|><|det|>[[90, 817, 900, 923]]<|/det|>
+Comment 3: Fig. 1d (and others): why is CI activity expressed as NAD+/NADH? This ratio is dependent on many other metabolic pathways. For proper analysis of (rotenone- sensitive) CI activity, more "classical" enzymatic assays in mitoplasts should be used. We appreciate the reviewer's comment. In the revised version of the manuscript, we have assessed CI activity using the classical kinetic spectrophotometric assays \(^{3}\) (Frazier AE et al. Methods Cell Biol 2020;155:1221- 156) (Fig. 1d and Fig. 2d.). Our findings closely mirror the
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 62, 895, 150]]<|/det|>
+results previously presented in our initial submission, as illustrated in Extended Data Fig. 1c, Extended Data Fig. 2a, and Extended Data Fig. 4d. The initial assessment of CI activity was based on a well- established commercially available assay (Abcam, ab109721) used in other publications<sup>4-6</sup> (Wang T. et al., Cell Metab 2021; 33 (3):531- 546. e9; Balsa E., et al., Mol Cell 2019; 74 (5):877- 890. e6; and Cao LL. et al., Nature 2016; 539(7630):575- 578).  
+
+<|ref|>text<|/ref|><|det|>[[91, 166, 901, 237]]<|/det|>
+Comment 4: Fig. 1h: Why was only NDUFS4 analyzed? I guess there should also be significant correlations with other CI subunits? When such correlations are absent, this would strengthen the choice to focus on NDUFS4 for further analysis. When such correlations are present, why was NDUFS4 selected?  
+
+<|ref|>text<|/ref|><|det|>[[90, 237, 900, 533]]<|/det|>
+We selected NDUFS4 among several CI subunits in our proteomic experiment for several compelling reasons: 1. Proteomic Analysis: in our comprehensive proteomic investigation, we identified NDUFS4 as a significantly downregulated subunit within CI in kidney podocytes from diabetic Akita mice, in comparison to those from a control group of non- diabetic mice. 2. Consistent Validation: subsequent validation studies, spanning various models including cellular, murine, and human samples, consistently revealed NDUFS4 as the most prominently downregulated subunit in diabetes when compared to nondiabetic controls, further substantiating its significance. In contrast, several other potential CI candidates, including NDUFA2, NDUFB3, NDUFB4, NDUFB5, NDUFB8, NDUFB11, and NDUFV3 were not consistently downregulated in the diabetic mice or in the glomeruli of patients with DKD (Extended Data Fig. 1e, f, g). 3. Early Downregulation: examination of human samples of individuals with DKD indicated that the downregulation of NDUFS4 occurs prior to the onset of albuminuria in patients, suggesting its potential role as an early marker of dysfunction (Fig. 1k, i). 4. Genetic Evidence: importantly, both human and murine genetic data underscore the critical role of NDUFS4 in mitochondrial function since the absence of functional NDUFS4 is associated with Leigh syndrome, emphasizing its indispensable function within mitochondrial processes.  
+
+<|ref|>text<|/ref|><|det|>[[92, 550, 827, 585]]<|/det|>
+Comment 5: Fig. 1i- l: more details are needed on how NDUFS4 staining was exactly quantified.  
+
+<|ref|>text<|/ref|><|det|>[[91, 585, 891, 690]]<|/det|>
+In the revised version of our manuscript, we have incorporated comprehensive information regarding the staining and quantification of NDUFS4 ("Methods" section on page 33). To quantify NDUFS4, we computed the mean intensity within the glomerular tuft area across all glomeruli obtained from healthy donor and DKD kidney samples. This quantification was carried out utilizing Image J software, following an established protocol previously outlined<sup>7</sup> (Falkevall A, et al. Cell Metab 2017;25(3):713- 726).  
+
+<|ref|>text<|/ref|><|det|>[[91, 707, 888, 829]]<|/det|>
+Comment 6: Page 6, lines 8- 10: in my view, the data suggests that CI expression is important, meaning that there is no evidence for NDUFS4- specific effects. Our initial data indicate a significant downregulation of CI subunits within the podocytes of diabetic mice, as demonstrated in Fig. 1c. However, among these CI subunits, Ndufs4 consistently displayed reduced expression levels in both diabetic mouse models and human patients. It is important to clarify that our study at that point simply highlighted the potential significance of Ndufs4 in the context of our experimental approach.  
+
+<|ref|>text<|/ref|><|det|>[[91, 845, 896, 935]]<|/det|>
+Comment 7: Fig. 3a: please compute the relevant respiratory control ratios (RCRs). These are independent of cell number and will highlight functional defects. Furthermore, there is no ECAR data presented. Why? We appreciate this concern raised by the reviewer. We have added RCRs values to the revised version of the manuscript. Consistent with our previous data, we observe that RCR is
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 60, 895, 116]]<|/det|>
+significantly reduced in primary podocytes from diabetic Ins2Akit/+ mice compared to those in WT mice and Ins2Akit/+ - Ndufs4podTg mice. We have added these data to the Extended Data Fig. 3a. Regarding ECAR, we have also added ECAR data into Extended Data Fig. 3b.  
+
+<|ref|>text<|/ref|><|det|>[[91, 130, 888, 166]]<|/det|>
+Comment 8: Fig. 3g (bargraphs): the average WT value is 1.0; how was data normalization performed for this and all relevant other figures?  
+
+<|ref|>text<|/ref|><|det|>[[91, 165, 896, 255]]<|/det|>
+We quantified the NADH oxidase staining using Image J software. For each glomerulus (a total of 60 glomeruli per group), we determined the intensity divided by the median intensity observed in the WT (wild type) images. This standardized approach ensures that the median intensity of the WT group is set to 1.0. The same normalization process was applied in Extended Data Fig. 2a to assess the relative CI activity in glomeruli.  
+
+<|ref|>text<|/ref|><|det|>[[91, 270, 895, 358]]<|/det|>
+Comment 9: Fig. 3h- n (and page 7; line 23): the presented parameters do not reflect mitochondrial "dynamics" but steady- state mitochondrial morphology. One cannot talk about enhanced fission (it can also be reduced fusion) since mitochondrial fission/fusion proteins were not investigated. Moreover, it is unclear how the conclusion that "cristae morphology is altered" is reached.  
+
+<|ref|>text<|/ref|><|det|>[[90, 357, 904, 778]]<|/det|>
+We appreciate this insightful question raised by the reviewer and agree that mitochondrial dynamics predominately depend on fission, fusion, shape transition, and transport or tethering along the cytoskeleton. Consistent with these concepts, we have previously shown that kidney podocytes exposed to high glucose conditions exhibit a fragmented or punctate phenotype<sup>8,9</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200; and Galvan DL., et al., J Clin Invest 2019;129(7):2807- 2823). Our earlier work also has clearly demonstrated that this shift towards shorter and fragmented mitochondria in high glucose- treated podocytes is attributed in part to an enhanced mitochondrial fission process characterized by increased Drp1 activity, the primary protein governing mitochondrial fission<sup>8</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200). In a subsequent study, we developed mouse models where the phosphorylation site of DRP1 at Ser 600 was mutated and rendered inactive specifically in podocytes of diabetic mice<sup>9</sup> (Galvan DL., et al., J Clin Invest 2019;129(7):2807- 2823). Additionally, we reported an increase in actin/Drp1 interactions associated with DRP1 phosphorylation<sup>9</sup> (Galvan DL., et al., J Clin Invest 2019, 129(7):2807- 2823). Therefore, the mitochondrial morphology depicted in Fig. 3h, particularly the observed shortening of mitochondria in the kidney podocytes of diabetic Ins2Akit/+ mice, is consistent with our prior publications in various experimental models of diabetic kidney disease. We concur with the reviewer that mitochondria maintain their morphology through a dynamic balance of fission and fusion processes in cells which is regulated by a number of regulatory kinetic proteins. Regarding cristae morphology, we quantified mitochondrial cristae abundance in transmission electron microscopy (TEM) micrographs obtained from primary podocytes using Image J. Our analysis revealed a significant reduction in cristae abundance in podocytes from Ins2Akit/+ mice, which was subsequently improved in the Ins2Akit/+ - Ndufs4podTg mice. These results have been included in Fig. 3m and n.  
+
+<|ref|>text<|/ref|><|det|>[[91, 792, 900, 881]]<|/det|>
+Comment 10: Fig. 3m (page 8; line 2): MitoSOX does not specifically detect "mitochondrial ROS". The ethidium molecule (which is the ROS- detection moiety of MitoSOX) reacts extremely fast (but not exclusively with) superoxide. This means that MitoSOX can become fluorescent in the cytosol during its transit to the mitochondrial matrix (driven by the presence of the decylTPP moiety of MitoSOX).  
+
+<|ref|>text<|/ref|><|det|>[[91, 881, 846, 934]]<|/det|>
+We appreciate and acknowledge that the use of MitoSOX, a widely employed tool for assessing mitochondrial ROS (mROS) in numerous studies<sup>10,11</sup> (Sutandy F.X.R., et al., Nature 2023;618(7966):849- 854, Labuschagne CF., et al., Cell Metab 2019;30(4):720-
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 60, 904, 325]]<|/det|>
+734. e5), may lack specificity in detecting mROS. In collaboration with Dr. Paul Schumacker, a co- author of this manuscript and an internationally known expert in the area of ROS<sup>12</sup> (Murphy et al. Guidelines for measuring reactive oxygen species and oxidative damage in cells and in vivo. Nat Metab. 2022;4(6):651- 662), we adopted a more precise approach for evaluating mROS in the revised version of our manuscript. We utilized a ratiometric mitochondrial matrix- targeted redox- sensitive reporter, mito- rGFP, which offers several advantages over traditional methods of assessing mROS, as previously documented by our group and others<sup>13,14</sup> (Galvan DL. et al., Kidney Int 2017;92(5):1282- 128; Waypa GB. et al., Circ Res 2010;106(3):526- 535). The mito- rGFP reporter is highly sensitive to changes in the redox state within the mitochondrial matrix. To this aim, we introduced mito- rGFP into primary podocytes isolated from WT, Ndufs4<sup>podTg</sup>, Ins2<sup>Akita/+</sup>, and Ins2<sup>Akita/+</sup>- Ndufs4<sup>podTg</sup> mice. We quantified the ratios of oxidized mito- rGFP to reduced mito- rGFP intensities using confocal microscopy images. As illustrated in Fig 3o, we observed that the ratio of oxidized/reduced rGFP was elevated in primary podocytes from Ins2<sup>Akita/+</sup> mice compared to WT mice while it was reduced in Ins2<sup>Akita/+</sup>- Ndufs4<sup>podTg</sup> mice.  
+
+<|ref|>text<|/ref|><|det|>[[90, 340, 904, 450]]<|/det|>
+Comment 11: Fig. 3n (page 8; line 2): the used assay reports on total cellular ATP content, so is not informative on ATP production. We appreciate the critique by the reviewer and consequently have changed the term ATP production to ATP content (Extended Data Fig. 3n and Extended Data Fig. 4h). Moreover, we performed a real- time, label- free assay to quantify cellular and mitochondrial ATP production rates in live primary podocytes (Fig. 3p,q).  
+
+<|ref|>text<|/ref|><|det|>[[90, 462, 894, 568]]<|/det|>
+Comment 12: page 8; line 6: what is meant with "mitochondrial reprogramming"? The concept of mitochondrial reprogramming underscores the dynamic nature of mitochondria and their ability to respond to signals that they receive from other cells or their microenvironment, resulting in changes to their function, morphology, number, and distribution within the cell. This degree of remodeling allows mitochondria to quickly adapt to their changing environmental cues.  
+
+<|ref|>text<|/ref|><|det|>[[90, 584, 901, 630]]<|/det|>
+Comment 13: Page 8; line 7: please rule out that the used DOX concentration and incubation regime impacts on mitochondrial function (described in the literature).  
+
+<|ref|>text<|/ref|><|det|>[[90, 620, 904, 829]]<|/det|>
+We appreciate the reviewer's comment. High doses of doxycycline have been previously shown to hinder protein translation of mitochondrially encoded genes—a phenomenon known as mitonuclear protein imbalance leading to defects in basal and maximal OCRs<sup>15</sup> (Moullan N., et al., Cell Rep 2015;10(10):1681- 1691). To determine the potential influence of DOX at a 200 nM concentration, as employed in our experiments, on mitochondrial function in our experimental models, we conducted an experiment comparing mitochondrial basal and maximal OCRs in podocytes treated with 200nM DOX compared to those treated with 1000 nM and 2000 nM. We found that using DOX at 200 nM, mitochondrial OCR remained largely unaffected. However, when incubated at a substantially higher dose (2000 nM), we observed significant changes in the basal and maximal OCR consistent with the findings in the literature. We have incorporated these new results into Extended Data Fig. 4c,d,e and provided a brief description in the Results section.  
+
+<|ref|>text<|/ref|><|det|>[[90, 845, 896, 878]]<|/det|>
+Comment 14: Page 8; line 10: it is not explained why HG is compared to NG and why this is relevant for the rest of the study.  
+
+<|ref|>text<|/ref|><|det|>[[90, 879, 901, 932]]<|/det|>
+High glucose (HG: 25mM glucose) is conventionally added to culture media to mimic diabetic conditions in cell models. In prior studies, both our lab and others have comprehensively evaluated the impact of various HG concentrations and treatment durations on eliciting
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 60, 902, 133]]<|/det|>
+cellular responses in podocytes<sup>8,16- 19</sup> (Wang W. et al., Cell Metab 2012;15(2):186- 200, Long J. et al., J Clin Invest 2016;126(11):4205- 18, Qi W. et al., Nat Med 2017;23(6):753- 762, Fu Y. et al., Cell Metab 2020;32(6):1052- 1062. e8, and Cao A. et al., J Clin Invest 2021;131(10):e141279).  
+
+<|ref|>text<|/ref|><|det|>[[90, 148, 900, 255]]<|/det|>
+Comment 15: Fig. 4f: what are all the other (non- marked) bands representing. Or is this not relevant? Please explain. I guess the AB cocktail was also used here? We appreciate this question raised by the reviewer. The AB cocktail was used as indicated on Fig. 4f. While we have presented the identities of the unmarked bands (Extended Data Fig. 4h), it is important to clarify that these bands are not pertinent to our current focus, which centers on identifying mitochondrial supercomplexes.  
+
+<|ref|>text<|/ref|><|det|>[[90, 270, 890, 307]]<|/det|>
+Comment 16: Page 9; line 7: why is it surprising that NDUFS4 OE restored CI in- gel activity "even under HG conditions"?  
+
+<|ref|>text<|/ref|><|det|>[[90, 306, 904, 360]]<|/det|>
+We have removed "even" from the sentence in response to the reviewer's feedback. We were simply attempting to convey our surprise at the significant regulatory impact of Ndufs4 in modulating the effect of high glucose in several experimental experiments in this study.  
+
+<|ref|>text<|/ref|><|det|>[[90, 374, 894, 410]]<|/det|>
+Comment 17: Page 9; line 11: that NDUFS4 exerts a "regulatory effect" is not demonstrated by the preceding experimental results.  
+
+<|ref|>text<|/ref|><|det|>[[90, 410, 890, 567]]<|/det|>
+Based on our comprehensive in vivo and in vitro observations, it is evident that NDUFS4 exerts a regulatory influence over mitochondrial morphology, cristae remodeling, and supercomplexes formation. This assertion is supported by our findings that both the downregulation and overexpression of NDUFS4 yield distinct and opposing effects on mitochondrial and cristae/mitochondria structure, effectively modulating their morphology. However, in response to the reviewer's insights, we acknowledge the need for further clarification regarding the precise regulatory role of NDUFS4. Consequently, we are actively conducting additional experiments within our laboratory to delve deeper into this important aspect of NDUFS4 function.  
+
+<|ref|>text<|/ref|><|det|>[[90, 583, 900, 736]]<|/det|>
+Comment 18: Regarding the STOML2- related experiments, I wonder if the proposed binding of STOML2 to NDUFS4 protein is compatible with the currently available structural information on the CI holocomplex and CI in a respiratory supercomplex? In my opinion, this is the most potentially interesting finding of this work since all functional effects of NDUFS4 OE could be explained by the NDUFS4 increase leading to more fully assembled CI. How does the cell "know" that more other CI subunits are required in the NDUFS4- overexpressing case? This would mean that NDUFS4 might be (among) the rate- limiting subunits for CI assembly. Is this supported by the current literature on CI biogenesis?  
+
+<|ref|>text<|/ref|><|det|>[[90, 736, 900, 923]]<|/det|>
+We appreciate this interesting comment raised by the reviewer. Our Extended Data Fig. 4g suggests that CI is mainly incorporated into the RSC in podocytes. Therefore, we performed molecular docking analysis using the NDUFS4 structural data derived from the Cryo- EM resolved respiromase structure (PDB database 5XTB; https://www.rcsb.org/structure/5XTB) and the predicted human STOML2 structure (Contact- guided Iterative Threading ASSembly Refinement web tool (C- I- TASSER; https://zhanggroup.org/C- I- TASSER) since the crystal structure of STOML2 has not yet been resolved. We supplemented our initial findings with STOML2 deletion mutant studies which indicate that stomatin domain of the STOML2 protein plays a central role for its binding to NDUFS4. In the revised version, we provide new data to delineate further this interaction and show that within the stomatin domain, the \(\beta 2\) , \(\beta 3\) and \(\beta 4\)
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 60, 904, 308]]<|/det|>
+strands are necessary for the binding of STOML2 with NDUFS4 (Fig. 6g,i). These results are also consistent with the molecular docking model. Taken together, our findings suggest that the interaction between NDUFS4 and STOML2 is necessary for the proper maintenance of cristae morphology, RSC integrity, and ETC function in podocytes. Interestingly, while Ndufs4 overexpression alone does not appear to trigger a cellular response to recruit additional CI subunits, the context changes in high glucose conditions where our data suggest that reduced levels of NDUFS4 result in compromised interactions between Ndufs4 and STOML2. This disruption leads to a disorganized cristae platform for RSC assembly leading to decreased RSC assembly and contributing to decreased CI function and mitochondrial dysfunction. Overexpressing Ndufs4 in diabetic podocytes restores this imbalance leading to enhanced CI stability, improved mitochondrial RSCs assembly and cristae morphology. We believe that the interaction between Ndufs4 and STOML2 could be an important mechanism adapting the CI assembly and function in response to metabolic cues.  
+
+<|ref|>sub_title<|/ref|><|det|>[[93, 342, 208, 361]]<|/det|>
+## Reviewer #2  
+
+<|ref|>text<|/ref|><|det|>[[90, 375, 904, 675]]<|/det|>
+The paper by Mise et al has explored the role of the Complex I subunit Ndufs4, in the pathogenesis of diabetic kidney disease. The aim of this study was to assess ETC abundance, mitochondrial function and morphology in kidney podocytes in a mouse model of diabetes and whether this affected DKD progression. First, ETC abundance was determined by mass spectrometry in two mouse models of diabetes, the Ins2Akita mouse and the db/db mouse model. Complex I abundance and mRNA expression was decreased in podocytes of the Ins2Akita mouse and db/db mouse. Out of the several downregulated subunits of mitochondrial OXPHOS complexes, Ndufs4 was chosen as a target. Glomerular Ndufs4 protein was decreased in human biopsies from patients with DKD. Next, podocyte- specific Ndufs4- transgenic mice were generated and crossed onto the Ins2Akita model of type 1 diabetes. Podocyte- specific Ndufs4- transgenic mice were partially protected from the diabetic kidney disease phenotype. Similar data were shown in the db/db mouse model. Ndufs4 overexpression in podocytes rescued high glucose- induced mitochondrial cristae changes as well as supercomplex assembly. A molecular interaction was identified between STOML2 and NDUFS4. The authors conclude that the findings represent a major paradigm shift in the current management of DKD by suggesting that targeting ETC remodeling could be a promising approach for developing therapies to mitigate the progression of DKD.  
+
+<|ref|>sub_title<|/ref|><|det|>[[93, 690, 238, 708]]<|/det|>
+## Major comments  
+
+<|ref|>text<|/ref|><|det|>[[90, 723, 896, 864]]<|/det|>
+Comment 1: ETC remodeling, mitochondrial function and morphology in kidney podocytes in rodent models with diabetes has been widely reported, including from the current group. Several other studies have explored the role of mitochondrial Complex I subunits, leading to ETC remodeling in the development of chronic kidney disease (e.g., PMID: 23320803), therefore the idea itself is not novel. These papers should be referenced. Many others have suggested that targeting the electron transport chain is an approach to treat CKD/DKD (PMID: 36781216), therefore the author's suggestion of a major paradigm shift should be moderated  
+
+<|ref|>text<|/ref|><|det|>[[92, 864, 892, 935]]<|/det|>
+We concur with the reviewer's comment on the potential impact of mitochondrial dysfunction as a prominent factor implicated in the pathogenesis of kidney diseases, including diabetic kidney disease (DKD). However, the precise nature of mitochondrial dysfunction and the molecular mechanisms responsible for ETC dysfunction in podocytes leading to the
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 60, 900, 343]]<|/det|>
+progression of DKD remain largely unknown. Our findings provide detailed insights into the pathobiology of mitochondrial respiration in podocytes and its central role in the pathogenesis of DKD. We provide evidence that Ndufs4 ties CI integrity to high glucose metabolic cues in the cell and the progression of DKD. While several studies have indeed suggested that compromised ETC function may serve as a risk factor for CKD, no prior investigations, to the best of our knowledge, have definitively demonstrated that improving CI structure/function can effectively reverse key features of DKD. Importantly, our study has provided strong experimental data to support an important biological function of Ndufs4 in preserving the integrity of cristae morphology. This novel function, in turn, exerts a protective influence on the progression of DKD. While previous studies have explored the regulatory impact of cristae- shaping proteins on supercomplexes, our work goes a step further by demonstrating for the first time that the overexpression of Ndufs4 in the diabetic milieu yields modulatory effects on cristae- shaping proteins and cristae morphology. We firmly believe that our study has introduced a completely new area of research, shedding light on the interplay between mitochondrial complex subunits and cristae forming proteins in the context of kidney pathology.  
+
+<|ref|>text<|/ref|><|det|>[[90, 358, 905, 567]]<|/det|>
+Comment 2: It is not clear which method was used to determine Complex I activity. The method referenced in the paper (ref 43), reports using NBT on tissue sections. This is not a generally accepted method. The gold standard method used by mitochondrial laboratories for determination of complex I activity is via respiratory chain enzymology using kinetic spectrophotometric assays (Frazier AE et al., Assessment of mitochondrial respiratory chain enzymes in cells and tissues, Methods Cell Biol 2020). Each assay requires specific inhibitors to ensure that the assay reflects true activity of the specific respiratory chain enzymology complex being assayed, i.e., rotenone- sensitive CI activity. In order to avoid misinterpretation, the mitochondrial marker enzyme citrate synthase is typically assayed to enable correction for mitochondrial content by expressing enzymes as citrate synthase ratios. Complex I activity should be determined using the kinetic spectrophotometric assay outlined in Frazier.  
+
+<|ref|>text<|/ref|><|det|>[[90, 567, 895, 744]]<|/det|>
+We appreciate the reviewer's comment. To answer the concern raised by the reviewer, we also employed kinetic spectrophotometric assays following the protocol detailed by Frazier AE et al as suggested by the reviewer<sup>3</sup> (Frazier AE. et al., Methods Cell Biol 2020;155:121- 156). As with our previously shown CI enzymatic activity assays, the results from these kinetic spectrophotometric assays reveal a reduction in rotenone- sensitive CI enzymatic activity within podocyte mitochondria derived from Ins2<sup>Akita/+</sup> as well as in Lepr<sup>db/db</sup> compared to control mice (Fig 1d). However, in regards to normalization using citrate synthase activity, it is worth noting that while using citrate synthase may be suitable in other conditions, our mitochondrial proteomic analysis indicate a significant reduction in citrate synthase abundance in podocytes from Ins2<sup>Akita/+</sup> mice compared to those from WT mice (Ins2<sup>Akita/+</sup> to WT ratio: 0.63). Western blotting also showed similar results (shown below). Therefore, relying on citrate synthase to normalize measurements in the diabetic condition could potentially lead to an overestimation. Consequently, we opted for an alternative approach, normalizing our data using total mitochondrial protein input as recommended by the Frazier's protocol. Of note, our initial assessment of CI activity involved NADH oxidase staining on frozen tissue sections obtained from mouse kidneys. This approach, widely utilized in the analysis of CI in frozen tissue samples<sup>20,21</sup> (Kruse SE. et al., Cell Metab 2008;7(4):312- 20 and Salagre D. et al., Antioxidants 2023;12(8):1499), served as an initial means to assess CI activity. For a more precise measurement of CI, we employed  
+
+<|ref|>text<|/ref|><|det|>[[90, 744, 704, 933]]<|/det|>
+WT ratio: 0.63). Western blotting also showed similar results (shown below). Therefore, relying on citrate synthase to normalize measurements in the diabetic condition could potentially lead to an overestimation. Consequently, we opted for an alternative approach, normalizing our data using total mitochondrial protein input as recommended by the Frazier's protocol. Of note, our initial assessment of CI activity involved NADH oxidase staining on frozen tissue sections obtained from mouse kidneys. This approach, widely utilized in the analysis of CI in frozen tissue samples<sup>20,21</sup> (Kruse SE. et al., Cell Metab 2008;7(4):312- 20 and Salagre D. et al., Antioxidants 2023;12(8):1499), served as an initial means to assess CI activity. For a more precise measurement of CI, we employed  
+
+<|ref|>image<|/ref|><|det|>[[714, 777, 896, 933]]<|/det|>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 60, 900, 150]]<|/det|>
+mitochondria from primary podocytes placed on a 96- well assay plate precoated with cocktail antibodies against CI subunits. This method is also a widely accepted approach<sup>4-6</sup> (Wang T et al., Cell Metab 2021;33(3)531- 546. e9; Balsa E et al., Mol Cell 2019;74(5):877- 890. e6; and Cao LL et al., Nature 2016;539(7630):575- 578). Nevertheless, we agree with the reviewer that using different established methods will add to the validity of our experimental approach.  
+
+<|ref|>text<|/ref|><|det|>[[91, 184, 648, 325]]<|/det|>
+Comment 3: "We also evaluated NDUFS4 staining in glomeruli from diabetic subjects with a wide spectrum of DKD histology, and found that NDUFS4 staining in glomeruli was progressively reduced with worsening of DKD histology (test for trend \(\mathsf{P}< 0.01\) ) (Fig. 1l)." Was there a significant difference from Class I CKD to Class III CKD, or between any of the DKD classes? Can you explain the Test for trend? Is it really a progressive decrease in glomerular Ndufs4 staining?  
+
+<|ref|>text<|/ref|><|det|>[[91, 325, 652, 375]]<|/det|>
+When specifically comparing different Classes of DKD, there is a significant difference in glomerular NDUFS4 staining between Class I and Class IV kidneys, where the difference reached  
+
+<|ref|>image<|/ref|><|det|>[[677, 181, 897, 365]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[90, 375, 895, 622]]<|/det|>
+statistical significance \((\mathsf{P} = 0.007)\) (see figure below). To explore the trend further, we conducted a linear trend analysis. A linear trend analysis, also known as a test for linear trend<sup>22,23</sup> (Nowak N. et al., Kidney Int 2016; 89:459- 467 and Sakaguchi Y. et al., J Am Soc Nephrol 2018; 29:991- 999), is a statistical method used to examine whether there is a "systematic and linear relationship between a set of ordered groups or categories and a measured variable." It is often applied when there is a natural order or progression in the groups being analyzed. The key idea behind linear trend analysis is to assess whether there is a consistent change in the variable of interest as you move from one group to the next. This change is evaluated to determine if it follows a linear pattern, meaning that as you go from one group to the next (e.g., from low to high levels of a variable or from one category to another), there is a consistent increase or decrease in the variable. Based on the results of the trend analysis, we conclude that glomerular NDUFS4 staining exhibits a progressive decline corresponding to the increase in glomerular pathological class of DKD, providing valuable insights into the disease progression.  
+
+<|ref|>text<|/ref|><|det|>[[91, 654, 864, 725]]<|/det|>
+Comment 4: Fig1k. Statistical analysis was performed on a sample size of \(n = 2\) within the microalbuminuria group. How is this possible when using one way- analysis of variance followed by Tukey's multiple comparisons test in Graphpad Prism? Please confirm the statistical test used.  
+
+<|ref|>text<|/ref|><|det|>[[91, 725, 900, 830]]<|/det|>
+We acknowledge the limited statistical power within the microalbuminuria group, although it's worth noting that GraphPad did provide the capability to perform a one- way ANOVA followed by Tukey's- Kramer multiple comparison test (https://www.graphpad.com/support/faqid/591/). However, following consultation with our biostatistician and in response to the comment by the reviewer, we have made the decision to combine all DKD patients with albuminuria into a single group. Consequently, we have updated Figure 1k to reflect this modification.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 62, 901, 99]]<|/det|>
+Comment 5: What was the mitochondrial density of the podocytes in diabetic mice? Ins2Akita and dbdb mice.  
+
+<|ref|>text<|/ref|><|det|>[[91, 98, 736, 220]]<|/det|>
+Using mitochondrial DNA copy number as an indicator of mitochondrial density, we have generated additional data that demonstrate reduced mitochondrial density in podocytes from Ins2Akita/+ mice when compared to their wild- type (WT) counterparts. We have previously established a similar pattern, illustrating lower mitochondrial copy numbers in podocytes from Lepr\(^{db/db}\) mice in comparison to their Lepr\(^{db/+}\) littermates\(^{16}\) (Long J. et al., J Clin Invest 2016; 126(11):4205)  
+
+<|ref|>image<|/ref|><|det|>[[748, 85, 900, 237]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[90, 253, 901, 411]]<|/det|>
+Comment 6: Fig 3n, ATP production. What are the units of the ATP assay? Both the mitosox and ATP data are expressed as relative to control. As is the Complex I. The ATP assay used appears to be a one- step single reagent assay for detection of viable cells in culture, rather than an assay which can reliably determine ATP production over time in a sensitive manner. I would caution against using such an assay when screening for differences between groups. Was a standard curve generated and used? Did the authors account for cell number or total protein? The same questions apply to the mitosox assay and data. Representative histograms showing mean intensity of MitoSOX fluorescence should be shown for each group.  
+
+<|ref|>text<|/ref|><|det|>[[90, 410, 900, 569]]<|/det|>
+In the ATP assay, we made a standard curve to measure ATP concentration. In addition, we generated a standard curve to obtain cell numbers based on the DNA concentration using CyQUANT Cell proliferation Assay Kit (Molecular Probes). In response to the reviewer's comments, however, we also performed a real- time label- free ATP assay using Seahorse Analyzer in live primary podocytes to assess real- time total cellular and mitochondrial ATP production rates. As shown in Fig. 3p,q, total and mitochondrial ATP production rates were significantly reduced in podocytes from Ins2Akita/+ mice, while were normalized in podocytes from Ins2Akita/+- Ndufs4podTg mice. Regarding ATP content, MitoSOX and CI enzymatic activity, please refer to our response to comments 3, 10, and 11 to Reviewer #1.  
+
+<|ref|>text<|/ref|><|det|>[[90, 582, 904, 619]]<|/det|>
+Comment 7: Some of the individual data points on graphs are missing through the manuscript (e.g., Fig 2h, k, I & m; Figure 3g, i, j, k, l) and throughout the paper.  
+
+<|ref|>text<|/ref|><|det|>[[91, 631, 900, 684]]<|/det|>
+We did not originally include the data points in some of the figures since our understanding was that data points should only be shown when \(n< 10\) based on the Journal's guidelines. However, we have added the data points in those figures as well in response to the reviewer.  
+
+<|ref|>text<|/ref|><|det|>[[91, 700, 894, 771]]<|/det|>
+Comment 8: Fig 2a. What sample is contained in each lane in the genotyping gel? Sample labels have been provided below the gel image for your reference. To clarify further, the first three lanes represent PCR results from three wild type (WT) samples, while the last three lanes correspond to PCR results from three Ndufs4podTg mice.  
+
+<|ref|>text<|/ref|><|det|>[[91, 787, 866, 840]]<|/det|>
+Comment 9: The mRNA expression of Ndufs4 appears higher in the "podocyte depleted" fraction of renal cortex vs the podocyte fraction. It is important to show Ndufs4 protein in podocytes vs tubules and total renal cortex.  
+
+<|ref|>text<|/ref|><|det|>[[91, 840, 884, 928]]<|/det|>
+The expression levels of Ndufs4 mRNA are indeed lower in podocytes compared to the podocytes- depleted fraction, which predominately consists of kidney tubules. In the revised version of this manuscript, we have included Western blot data illustrating the levels of NDUFS4 in both podocytes and tubules (Extended Data Fig. 1h). Additionally, we have mentioned in the text (page 5) that the NDUFS4 protein expression in kidney tubular cells
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 62, 896, 116]]<|/det|>
+remains unchanged in diabetic mice. It is important to highlight that the mitochondrial density in tubules greatly exceeds that in podocytes. This difference in mitochondrial density may account for the elevated expression of Ndufs4 in tubules in comparison to podocytes.  
+
+<|ref|>text<|/ref|><|det|>[[91, 131, 875, 185]]<|/det|>
+Comment 10: For the Seahorse cell culture studies, how many independent cell culture experiments are shown? It is stated that there were \(n = 6 - 8\) replicates per group. Are these data derived from one cell culture experiment?  
+
+<|ref|>text<|/ref|><|det|>[[91, 185, 896, 273]]<|/det|>
+The original results were derived from a single cell culture experiment, in which primary podocytes were isolated and pooled from three mice in each experimental group. We have performed similar experiments with primary podocytes isolated from an additional three mice in each group. The results from these additional experiments corroborate with our initial findings, reinforcing the consistency and reliability of our results.  
+
+<|ref|>image<|/ref|><|det|>[[95, 288, 896, 508]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[91, 530, 866, 601]]<|/det|>
+Comment 11: Extended data figure 5a. The western blots are not convincing. The quantitation needs to be shown. In mouse kidney, OPA1 generally has 5 isoforms. There should be 5 bands present. Samples need to be run on a gradient gel to reveal the five isoforms as per PMID: 26822084.  
+
+<|ref|>text<|/ref|><|det|>[[91, 600, 896, 689]]<|/det|>
+We have incorporated the quantification results obtained from Western blots and conducted an analysis of OPA1 using gradient gel electrophoresis. This analysis was performed on primary podocytes isolated from four distinct groups of mice (Extended Data Figure 5a). Of note, OPA1 does not appear to exhibit 5 distinct isoform bands, possibly due to cell or tissue- specific variations in isoform patterns.  
+
+<|ref|>text<|/ref|><|det|>[[91, 703, 875, 792]]<|/det|>
+Comment 12: Page 4, line 13. "ETC dysfunction is recognized as an important cause of organ failure in several human pathologies including heart failure, diabetes, and neurodegeneration in a tissue- specific manner". Mitochondrial cytopathies have long been recognized to lead to kidney failure. This should be mentioned in the introduction (PMID: 33305107).  
+
+<|ref|>text<|/ref|><|det|>[[91, 792, 866, 844]]<|/det|>
+We thank the reviewer and have added these important citations to the manuscript<sup>24,25</sup> (Schijvens AM et al., Kidney Int Rep 2020; 5(12):2146- 2159; and Emma F et al., Nat Rev Nephrol 2016; 12(5):267- 280).  
+
+<|ref|>sub_title<|/ref|><|det|>[[92, 879, 213, 897]]<|/det|>
+## Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[92, 912, 875, 931]]<|/det|>
+In this manuscript, Mise et al. examine the importance of the mitochondrial ETC Complex I
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[91, 60, 884, 150]]<|/det|>
+protein Ndufs4, in the maintenance of mitochondrial structure in the context of the diabetic kidney. The Authors show that Ndufs4 plays a role in the regulation of cristae structure and supercomplex evidence of binding partners in the process. The studies address an interesting topic, and the experimentation appears thorough and well- executed. Specific comments are indicated below.  
+
+<|ref|>text<|/ref|><|det|>[[90, 166, 896, 290]]<|/det|>
+Comment 1: It is not entirely clear why the Authors chose to focus on Ndufs4. The heat map in Fig.1 shows many ETC proteins, including in Complex I that were also decreased as a result of diabetes. In fact, there are some that were decreased by a greater extent. The decision to choose Ndufs4 comes across as a biased decision without scientific justification. A greater case should be made for choosing this target. In the Discussion, the Authors indicate that examination of other subunits would be meaningful, and this is appreciated, but does not preclude the need for a greater justification for the Ndufs4 focus.  
+
+<|ref|>text<|/ref|><|det|>[[90, 289, 901, 464]]<|/det|>
+We appreciate the reviewer's comment. Please also refer to our response to Reviewer #1 (comment #4). Briefly, we chose NDUFS4 among several CI subunits in our proteomic experiment for several reasons: Subsequent validation studies in other experimental models of diabetes consistently revealed NDUFS4 as the most prominently downregulated subunit in diabetes when compared to nondiabetic controls, further substantiating its significance. In contrast, several other potential candidates, including NDUFA2, NDUFB3, NDUFB4, NDUFB5, NDUFB8, NDUFB11, and NDUFV3 were not consistently downregulated (Extended Data Fig. 1e,g). Importantly, examination of human samples of individuals with DKD indicated that the downregulation of NDUFS4 occurs prior to the onset of albuminuria in patients, suggesting its potential role as an early marker of dysfunction.  
+
+<|ref|>text<|/ref|><|det|>[[90, 480, 881, 568]]<|/det|>
+Comment 2: Introduction, page 4; In the last sentence, the Authors state that their "results unexpectedly reveal that ETC integrity determines the stability of RSCs." This seems disingenuous. Do the Authors really believe that the very integrity of the components of the ETC would not be an important factor in the stability of the supercomplex in which they reside? My suggestion would be to either remove or soften this statement.  
+
+<|ref|>text<|/ref|><|det|>[[91, 568, 880, 638]]<|/det|>
+We appreciate the reviewer's suggestion. In the revised manuscript, we have removed the term "unexpectedly" from the content. Our intention simply was to convey that we did not anticipate that Ndufs4 overexpression would impact not only CI activity, but also RSC stability.  
+
+<|ref|>text<|/ref|><|det|>[[90, 654, 892, 741]]<|/det|>
+Comment 3: The use of multiple experimental models of diabetes (Akita, db/db) and human patient samples is appreciated and strengthen the conclusions and relevance of the studies. With that being said, the inclusion of complete db/db data (especially proteomic profiling in Fig.1b) is glaringly absent. Inclusion of this data makes a stronger case for commonality in the models and the ultimate focus on Ndufs4.  
+
+<|ref|>text<|/ref|><|det|>[[90, 742, 903, 829]]<|/det|>
+We thank the reviewer's suggestion and have performed mitochondrial proteomic analysis using mitochondria isolated from podocytes of Leprdb/+ and Leprdb/db mice. The data showed that NDUFS4 was one of the most downregulated among the CI subunits. This gave us stronger support in selecting NDUFS4 in this study. We have included the new data in Fig. 1e and Extended Data Fig. 1d.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[88, 110, 904, 900]]<|/det|>
+1 Cabrera- Orefice, A., Potter, A., Evers, F., Hevler, J. F. & Guerrero- Castillo, S. Complexome profiling- exploring mitochondrial protein complexes in health and disease. Front. Cell Dev. Biol. 9, 796128 (2021).2 Válikangas, T., Suomi, T. & Elo, L. L. A systematic evaluation of normalization methods in quantitative label- free proteomics. Brief Bioinform. 19, 1- 11 (2018).3 Frazier, A. E., Vincent, A. E., Turnbull, D. M., Thorburn, D. R. & Taylor, R. W. Assessment of mitochondrial respiratory chain enzymes in cells and tissues. Methods Cell Biol. 155, 121- 156 (2020).4 Wang, T. et al. C9orf72 regulates energy homeostasis by stabilizing mitochondrial complex I assembly. Cell Metab. 33, 531- 546. e539 (2021).5 Balsa, E. et al. ER and nutrient stress promote assembly of respiratory chain supercomplexes through the PERK- eIF2alpha axis. Mol. Cell 74, 877- 890. e876 (2019).6 Cao, L. L. et al. Control of mitochondrial function and cell growth by the atypical cadherin Fat1. Nature 539, 575- 578 (2016).7 Falkevall, A. et al. Reducing VEGF- B Signaling Ameliorates Renal Lipotoxicity and Protects against Diabetic Kidney Disease. Cell Metab. 25, 713- 726 (2017).8 Wang, W. et al. Mitochondrial fission triggered by hyperglycemia is mediated by ROCK1 activation in podocytes and endothelial cells. Cell Metab. 15, 186- 200 (2012).9 Galvan, D. L. et al. Drp1S600 phosphorylation regulates mitochondrial fission and progression of nephropathy in diabetic mice. J. Clin. Invest. 129, 2807- 2823 (2019).10 Sutandy, F. X. R., Gößner, I., Tascher, G. & Münch, C. A cytosolic surveillance mechanism activates the mitochondrial UPR. Nature 618, 849- 854 (2023).11 Labuschagne, C. F., Cheung, E. C., Blagih, J., Domart, M. C. & Vousden, K. H. Cell Clustering Promotes a Metabolic Switch that Supports Metastatic Colonization. Cell Metab. 30, 720- 734. e725 (2019).12 Murphy, M. P. et al. Guidelines for measuring reactive oxygen species and oxidative damage in cells and in vivo. Nat Metab 4, 651- 662 (2022).13 Galvan, D. L. et al. Real- time in vivo mitochondrial redox assessment confirms enhanced mitochondrial reactive oxygen species in diabetic nephropathy. Kidney Int. 92, 1282- 1287 (2017).14 Waypa, G. B. et al. Hypoxia triggers subcellular compartmental redox signaling in vascular smooth muscle cells. Circ. Res. 106, 526- 535 (2010).15 Moullan, N. et al. Tetracyclines Disturb Mitochondrial Function across Eukaryotic Models: A Call for Caution in Biomedical Research. Cell Rep. 10, 1681- 1691 (2015).16 Long, J. et al. Long noncoding RNA Tug1 regulates mitochondrial bioenergetics in diabetic nephropathy. J. Clin. Invest. 126, 4205- 4218 (2016).17 Qi, W. et al. Pyruvate kinase M2 activation may protect against the progression of diabetic glomerular pathology and mitochondrial dysfunction. Nat. Med. 23, 753- 762 (2017).18 Fu, Y. et al. Elevation of JAML Promotes Diabetic Kidney Disease by Modulating Podocyte Lipid Metabolism. Cell metabolism 32, 1052- 1062. e1058 (2020).19 Cao, A. et al. DACH1 protects podocytes from experimental diabetic injury and modulates PTIP- H3K4Me3 activity. The Journal of clinical investigation 131 (2021).20 Kruse, S. E. et al. Mice with mitochondrial complex I deficiency develop a fatal encephalomyopathy. Cell Metab. 7, 312- 320 (2008).21 Salagre, D., Raya Alvarez, E., Cendan, C. M., Aouichat, S. & Agil, A. Melatonin Improves Skeletal Muscle Structure and Oxidative Phenotype by Regulating Mitochondrial Dynamics and Autophagy in Zücker Diabetic Fatty Rat. Antioxidants (Basel) 12 (2023).22 Nowak, N. et al. Increased plasma kidney injury molecule- 1 suggests early progressive renal decline in non- proteinuric patients with type 1 diabetes. Kidney Int 89, 459- 467 (2016).
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[89, 60, 904, 175]]<|/det|>
+23 Sakaguchi, Y., Hamano, T., Wada, A., Hoshino, J. & Masakane, I. Magnesium and Risk of Hip Fracture among Patients Undergoing Hemodialysis. Journal of the American Society of Nephrology: JASN 29, 991- 999 (2018).24 Schijvens, A. M. et al. Mitochondrial Disease and the Kidney With a Special Focus on CoQ(10) Deficiency. Kidney Int. Rep. 5, 2146- 2159 (2020).25 Emma, F., Montini, G., Parikh, S. M. & Salviati, L. Mitochondrial dysfunction in inherited renal disease and acute kidney injury. Nat. Rev. Nephrol. 12, 267- 280 (2016).
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[119, 85, 357, 101]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 137, 448, 152]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[119, 166, 860, 208]]<|/det|>
+I highly appreciate the inclusion of additional experiments, as well as the thorough rebuttal, who have taken away several of my concerns. However, I feel that some of my previous suggestions deserve further attention:  
+
+<|ref|>text<|/ref|><|det|>[[118, 223, 868, 250]]<|/det|>
+1. Regarding Fig. 1b: I appreciate the new table but still cannot easily determine if these changes are significant; please include p-values in the new table (Extended data Fig. 1d)  
+
+<|ref|>text<|/ref|><|det|>[[118, 264, 875, 362]]<|/det|>
+2. Although the authors present some arguments (e.g. their reply to comment 4, 6 and 17), I'm still not convinced that the observed effects are Ndufs4-specific. The accessory NDUFS4 protein is one of the essential CI subunits. Unless it has an additional function(perhaps the role described in this manuscript?), a drop in NDUFS4 protein levels will always induce a drop in the total protein level of assembled CI. The latter will induce a drop in protein level of virtually all CI subunits. In my opinion, the described effects all can be explained by a reduction in the level of fully assembled CI and therefore cannot be regarded as ndufs4-specific?  
+
+<|ref|>text<|/ref|><|det|>[[118, 375, 875, 445]]<|/det|>
+3. Extended data Fig. 3b: the y-axis title contains an error. Moreover, in the main text the authors state that: "...ECAR... also showed similar changes..". What does this mean? If there really is a drop in OCR, I would expect an increase in ECAR? As described in the literature, ECAR is not exclusively a measure of glycolytic rate (e.g. lactate production); please discuss the OCR and ECAR results in an integrated manner.  
+
+<|ref|>text<|/ref|><|det|>[[118, 459, 858, 488]]<|/det|>
+4. Regarding my previous comments 8, 9,12 and 14: please ensure that the info provided in the author's rebuttal is included in the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[118, 502, 875, 558]]<|/det|>
+5. I appreciate the use of roGFP (I guess it is roGFP1?). This contains an S=S bridge, which is broken or formed based upon its thiol redox environment. Please interpret the results obtained with this sensor as a measure of thiol redox state (this off course is hydrogen peroxide sensitive in most systems).  
+
+<|ref|>text<|/ref|><|det|>[[118, 572, 875, 600]]<|/det|>
+6. Regarding my previous comment 15. The Extended data Fig. 4h referred to in the rebuttal is not on supercomplexes but on ATP levels?  
+
+<|ref|>text<|/ref|><|det|>[[118, 614, 877, 698]]<|/det|>
+7. Regarding the NDFUS4-STOML2 interaction. In the rebuttal there is referred to Extended data Fig. 4g, but this panel has nothing to do with STOML2. I do not understand the choice for SXTB, which represents only part of the CI matrix arm. Please investigate whether the STOML2 binding sites on the NDUFS4 protein are accessible when NDUFS4 is incorporated in fully assembled CI and the latter is part of the RSC. This will provide insight on whether STOML2 can bind to NDUFS4 when the latter is within the ETC supercomplex.  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 763, 448, 777]]<|/det|>
+## Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 791, 865, 806]]<|/det|>
+The authors have responded to all of my queries satisfactorily. The manuscript is much improved.  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 857, 448, 872]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[119, 886, 444, 900]]<|/det|>
+The Authors have addressed my concerns.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[93, 63, 207, 80]]<|/det|>
+## Reviewer #1  
+
+<|ref|>text<|/ref|><|det|>[[92, 96, 896, 151]]<|/det|>
+Comment 1: Regarding Fig. 1b: I appreciate the new table but still cannot easily determine if these changes are significant; please include p- values in the new table (Extended data Fig. 1d)  
+
+<|ref|>text<|/ref|><|det|>[[91, 165, 904, 291]]<|/det|>
+We isolated podocytes from 8 wild type (WT) and 8 diabetic Ins2Akit/+ mice (Fig1. Legend, a). Subsequently, we pooled all the podocytes within each group for a comprehensive screening of differentially expressed mitochondrial subunits in our proteomic analysis. While this pooling strategy might have limited the statistical significance of individual subunits when comparing WT and diabetic mice, we addressed this limitation by incorporating and validating our results using additional approaches, including qRT- PCR analyses (Extended Data Fig. 1e- g) and using an alternative diabetic Leprdb/db mouse model.  
+
+<|ref|>text<|/ref|><|det|>[[91, 305, 904, 446]]<|/det|>
+Comment 2: Although the authors present some arguments (e.g., their reply to comment 4, 6 and 17), I'm still not convinced that the observed effects are Ndufs4- specific. The accessory NDUFS4 protein is one of the essential CI subunits. Unless it has an additional function (perhaps the role described in this manuscript?), a drop in NDUFS4 protein levels will always induce a drop in the total protein level of assembled CI. The latter will induce a drop in protein level of virtually all CI subunits. In my opinion, the described effects all can be explained by a reduction in the level of fully assembled CI and therefore cannot be regarded as Ndufs4- specific?  
+
+<|ref|>text<|/ref|><|det|>[[90, 461, 901, 673]]<|/det|>
+Our findings strongly suggest that the observed effects on mitochondrial and kidney function described in this study are indeed Ndufs4- specific since we specifically targeted Ndufs4 in our experimental approach. However, it is important to clarify that while our results support the critical role of Ndufs4 in the observed effects, this does not imply exclusivity. Our findings suggest that Ndufs4 overexpression through its interaction with STOML2 stabilizes cristae morphology, providing a platform for assembly of the mitochondrial respiratory chain complexes. However, we recognize that these findings do not negate the potential role of other CI subunits in sustaining CI integrity and enhancing overall mitochondrial function. Indeed, we concur with the reviewer that improving CI and mitochondrial dysfunction in the diabetic environment may not be confined solely to Ndufs4, recognizing the plausible involvement of other CI subunits in CI integrity and improved mitochondrial function. We have highlighted this point in the Discussion.  
+
+<|ref|>text<|/ref|><|det|>[[91, 688, 896, 781]]<|/det|>
+Comment 3: Extended data Fig. 3b: the y- axis title contains an error. Moreover, in the main text the authors state that: "...ECAR... also showed similar changes.". What does this mean? If there really is a drop in OCR, I would expect an increase in ECAR. As described in the literature, ECAR is not exclusively a measure of glycolytic rate (e.g., lactate production); please discuss the OCR and ECAR results in an integrated manner.  
+
+<|ref|>text<|/ref|><|det|>[[91, 800, 896, 916]]<|/det|>
+We appreciate the reviewer's comment. It seems that an error occurred during the PDF conversion, and original \(10^{5}\) was mistakenly displayed as \(10^{1}\) . We have now rectified this error in the revised version. Regarding the interpretation of ECAR results, the interplay between OCR and ECAR in diabetic kidney disease is indeed complex. Consistent with our findings, a recent publication has provided evidence that chronic hyperglycemia results in lower ECAR and OCR levels in podocytes. However, we took the reviewer's comment as an
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 60, 890, 250]]<|/det|>
+opportunity to further clarify the relationship between aberrant OCR and ECAR in our experimental model. We have now added a new figure (Extended Data Fig. 3c), elucidating the energy maps of primary podocytes isolated from the four distinct experimental groups of mice by combining OCR and ECAR data. The overall goal was to understand how the podocytes energy metabolism adapts and changes under diabetic environment. This graphical representation suggests that podocytes derived from diabetic \(Ins^{2Akit / + }\) mice are inefficient in energy production in diabetic conditions, whereas overexpression of Ndufs4 in diabetic \(Ins^{2Akit / + }\) ; Ndufs4<sup>PodTg</sup> mice exhibit an energy map similar to the podocytes from WT mice, indicating that they use both mitochondrial respiration and glycolysis for energy production.  
+
+<|ref|>text<|/ref|><|det|>[[90, 266, 879, 303]]<|/det|>
+Comment 4: Regarding my previous comments 8, 9, 12 and 14: please ensure that the info provided in the author's rebuttal is included in the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[91, 319, 881, 390]]<|/det|>
+We have incorporated additional information based on our response to the reviewer's comments in the revised manuscript. Most comments were already present in the previous revision manuscript and thereby are no longer highlighted in the current revised version. Please refer to pages 4, 9, 25, and 38.  
+
+<|ref|>text<|/ref|><|det|>[[91, 405, 877, 477]]<|/det|>
+Comment 5: I appreciate the use of roGFP (I guess it is roGFP1?). This contains an S=S bridge, which is broken or formed based upon its thiol redox environment. Please interpret the results obtained with this sensor as a measure of thiol redox state (this of course is hydrogen peroxide sensitive in most systems).  
+
+<|ref|>text<|/ref|><|det|>[[90, 492, 904, 756]]<|/det|>
+As previously described by our group and others<sup>2,3</sup>, we used mito- roGFP2, a mitochondrial matrix- targeted ratiometric redox- sensitive green fluorescent protein. This sensor contains two adjacent cysteine residues that do not interact in reduced thiol redox state where the sensor exhibits high emission at 525 nm when excited at 488 nm, and relatively low emission at 525 nm when excited at 405 nm. Protein thiol oxidation mediated by \(H_2O_2\) leads to the formation of a disulfide linkage, resulting in a conformational change in the sensor that increases emission at 525 nm during excitation at 405 nm, and decreases emission at 525 nm when excited at 488 nm<sup>2,3</sup>. We measured mito- roGFP ratios in live mitochondria using confocal microscopy, in live cells. We found that primary podocytes from diabetic mice showed an increase in oxidized/reduced mito- roGFP ratios compared to those from WT mice (Fig. 3o), suggesting that mitochondrial thiol oxidation increases in diabetic podocytes. On the other hand, primary podocytes from diabetic mice overexpressing Ndufs4 show significantly reduced oxidized/reduced mito- roGFP ratios compared with podocytes from diabetic mice, indicating Ndufs4 overexpression prevents enhanced mROS in diabetes. We added this information in the text and Methods.  
+
+<|ref|>text<|/ref|><|det|>[[92, 771, 880, 808]]<|/det|>
+Comment 6: Regarding my previous comment 15. The Extended data Fig. 4h referred to in the rebuttal is not on supercomplexes but on ATP levels?  
+
+<|ref|>text<|/ref|><|det|>[[92, 824, 706, 842]]<|/det|>
+In the revised manuscript, the data are shown in Extended data Fig. 4i.  
+
+<|ref|>text<|/ref|><|det|>[[91, 857, 892, 930]]<|/det|>
+Comment 7: Regarding the NDFUS4- STOML2 interaction. In the rebuttal there is referred to Extended data Fig. 4g, but this panel has nothing to do with STOML2. I do not understand the choice for 5XTB, which represents only part of the CI matrix arm. Please investigate whether the STOML2 binding sites on the NDUFS4 protein are accessible when NDUFS4 is
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 61, 885, 98]]<|/det|>
+incorporated in fully assembled CI and the latter is part of the RSC. This will provide insight on whether STOML2 can bind to NDUFS4 when the latter is within the ETC supercomplex.  
+
+<|ref|>text<|/ref|><|det|>[[90, 113, 901, 150]]<|/det|>
+In the revised version of the manuscript, the Extended Data Fig. 4j provides evidence that the CI is present almost exclusively in the form of stable RSCs in podocytes.  
+
+<|ref|>text<|/ref|><|det|>[[90, 150, 888, 293]]<|/det|>
+The structural information for macromolecules containing Ndufs4 was retrieved from the Protein Data Bank (PDB; https://www.rcsb.org/). There are four entries in PDB with macromolecules containing NDUFS4: 5XTB (matrix arm of CI), 5XTC (CI), 5XTH (RSC with I1ll2IV1 stoichiometry), and 5XTI (I2ll2IV2 stoichiometry), each with a slightly different resolution. We chose PDB- 5XTB for further study because of its highest resolution (3.40A). To address whether NDUFS4 protein is accessible to STOML2 in the context of RSC, we used the molecular structure visualization and analysis tool ChimeraX (UCSF ChimeraX, https://www.cgl.ucsf.edu/chimeraX/), a powerful molecular modeling engine, to analyze the structure of the I1ll2IV1 within RSC resolved by CryoEM (PDB- 5XTH). This program shows that CIII and CIV bind to the hydrophobic membrane arm of CI, forming the RSC. Furthermore, it reveals that the NDUFS4 protein (magenta, Extended Data Fig. 6c) is localized on the surface of the matrix arm of the CI within the RSC, exposing NDUFS4 to potential interactions with other molecules such as STOML2. Our experimental data with STOML2 knockout cells and STOML2 mutation studies provide additional support to this interaction. Overall, these findings shed light on potential interactions involving NDUFS4 and STOML2. The figure was added to the manuscript (Extended Data Fig. 6c).  
+
+<|ref|>text<|/ref|><|det|>[[90, 292, 585, 516]]<|/det|>
+structure of the I1ll2IV1 within RSC resolved by CryoEM (PDB- 5XTH). This program shows that CIII and CIV bind to the hydrophobic membrane arm of CI, forming the RSC. Furthermore, it reveals that the NDUFS4 protein (magenta, Extended Data Fig. 6c) is localized on the surface of the matrix arm of the CI within the RSC, exposing NDUFS4 to potential interactions with other molecules such as STOML2. Our experimental data with STOML2 knockout cells and STOML2 mutation studies provide additional support to this interaction. Overall, these findings shed light on potential interactions involving NDUFS4 and STOML2. The figure was added to the manuscript (Extended Data Fig. 6c).  
+
+<|ref|>image<|/ref|><|det|>[[606, 297, 878, 515]]<|/det|>  
+
+<|ref|>sub_title<|/ref|><|det|>[[91, 564, 206, 582]]<|/det|>
+## References:  
+
+<|ref|>text<|/ref|><|det|>[[118, 598, 884, 740]]<|/det|>
+1. Qi, W. et al. Pyruvate kinase M2 activation may protect against the progression of diabetic glomerular pathology and mitochondrial dysfunction. Nat. Med. 23, 753-762 (2017).  
+2. Waypa, G. B. et al. Hypoxia triggers subcellular compartmental redox signaling in vascular smooth muscle cells. Circ. Res. 106, 526-535 (2010).  
+3. Galvan, D. L. et al. Real-time in vivo mitochondrial redox assessment confirms enhanced mitochondrial reactive oxygen species in diabetic nephropathy. Kidney Int. 92, 1282-1287 (2017).
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[118, 85, 377, 101]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 138, 450, 153]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 166, 767, 181]]<|/det|>
+I thank the authors for their rebuttal, which have addressed my remaining concerns.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1e369c655e2cff1cb25b6641a4035c982b0a10c2ec5eb94ed8a1982a90990f7c/images_list.json

@@ -0,0 +1,213 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_6.jpg",
+    "caption": "Fig. 6 | Experimental setup of c. w. DFG based on another ZGP micro-waveguide and characterizations of spectra of pump, signal, and corresponding LWIR idler waves. (a) The schematic of c. w. mid-infrared laser generation through DFG in ZGP waveguide. WDM, wavelength division multiplexer; L, lens; FC, fiber collimator; LPF, long-pass filter. The setup is composed of three parts, including a home-built c. w. tunable random Raman fiber laser (RRFL) at a wavelength around 1350 nm, which is used as DFG pump source, as depicted in Supplementary Note 4, a commercial erbium-doped fiber amplifier (EDFA) seeded by",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 0
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_0.jpg",
+    "caption": "Fig. R1. The simulated output power from the ZGP \\(\\chi^{(2)}\\) waveguide as a function of the pump pulse energy, by solving the \\(\\chi^{(2)}\\) -based coupled-wave equations. The drop of output power around \\(9\\mathrm{nJ}\\) pump pulse energy is reproduced",
+    "footnote": [],
+    "bbox": [
+      [
+        333,
+        100,
+        708,
+        304
+      ]
+    ],
+    "page_idx": 8
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_1.jpg",
+    "caption": "Fig. R2. The measured signal and idler spectra and power from the ZGP \\(\\chi^{(2)}\\) birefringence waveguide. (a) The spectra of signal and idler waves. (b) The output power of signal and idler waves as a function of pump power.",
+    "footnote": [],
+    "bbox": [
+      [
+        163,
+        172,
+        833,
+        386
+      ]
+    ],
+    "page_idx": 11
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_2.jpg",
+    "caption": "Fig. R3. Characterization of \\(\\chi^{(3)}\\) effect in the 10-mm-long \\(\\chi^{(2)}\\) ZGP waveguide. (a) Comparison of the measured pump spectra before entering and after propagating through the \\(\\chi^{(2)}\\) ZGP waveguide. (b) Comparison of the simulated idler spectra with and without \\(\\chi^{(3)}\\) effects.",
+    "footnote": [],
+    "bbox": [
+      [
+        202,
+        592,
+        875,
+        783
+      ]
+    ],
+    "page_idx": 13
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_3.jpg",
+    "caption": "Fig. R4. The temporal characterization by reproducing the OPG experiment in a bulk ZGP crystal with the same experimental condition of the \\(\\chi^{(2)}\\) ZGP waveguide. (a) The measured (black) and retrieved (red) IAC traces. (b) The retrieved temporal profile showing 270 fs pulse width and certain pulse splitting.",
+    "footnote": [],
+    "bbox": [
+      [
+        191,
+        84,
+        880,
+        287
+      ]
+    ],
+    "page_idx": 14
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_4.jpg",
+    "caption": "Fig. R5. (a) The comparison of measured transmission spectra between 8-mm-thick YS-ZGP crystal and traditional ZGP crystal with the same thickness, across the wavelength range of \\(0.5 - 2.5 \\mu \\mathrm{m}\\) . (b) The measured transmission spectrum of 8-mm-thick YS-ZGP crystal across \\(0.5 - 12 \\mu \\mathrm{m}\\) (The Fresnel reflection is not subtracted). The enlargement of marked red frame is plot in Fig. R5(a).",
+    "footnote": [],
+    "bbox": [
+      [
+        238,
+        553,
+        750,
+        711
+      ]
+    ],
+    "page_idx": 15
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_5.jpg",
+    "caption": "Fig. R6. Experimental setup of c. w. DFG based on newly fabricated ZGP waveguide and",
+    "footnote": [],
+    "bbox": [
+      [
+        252,
+        592,
+        770,
+        871
+      ]
+    ],
+    "page_idx": 16
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_6.jpg",
+    "caption": "Fig. 6 | Experimental setup of c. w. DFG based on another ZGP waveguide and characterizations of spectra of pump, signal, and corresponding LWIR idler waves. (a) The schematic of c. w. mid-infrared laser generation through DFG in ZGP waveguide. WDM, wavelength division multiplexer; L, lens; FC, fiber collimator; LPF, long-pass filter. The setup is composed of three parts, including a home-built c. w. tunable random Raman fiber laser (RRFL) at a wavelength around 1350 nm, which is used as DFG pump source, as depicted in Supplementary Note 4, a commercial erbium-doped fiber amplifier (EDFA) seeded by tunable signal-frequency DFB laser (Conquer, KG-TLS-13-P-FA) emitting in the spectral range of 1527-1567 nm as DFG signal source, and another ZGP waveguide, cut at \\(\\theta = 65.4^{\\circ}\\) and \\(\\phi = 0^{\\circ}\\) . The RRFL and EDFA sources are combined by a fiber wavelength division multiplexer (WDM) as an all-fiber laser driver for DFG in the ZGP waveguide. (b) Typical pump spectra running from 1340 nm to 1370 nm. (c) Typical signal spectra running from 1550 nm to 1565 nm. (d) The measured spectra of generated LWIR lasing tunable from 10.28 μm to 10.58 μm through DFG in the ZGP waveguide.",
+    "footnote": [],
+    "bbox": [
+      [
+        230,
+        87,
+        800,
+        393
+      ]
+    ],
+    "page_idx": 17
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_6.jpg",
+    "caption": "Fig. R7. Experimental setup of c. w. DFG based on newly fabricated ZGP waveguide and characterizations of spectra of pump, signal, and corresponding LWIR idler waves. (a) The schematic of c. w. mid-infrared laser generation through DFG in ZGP waveguide. WDM, wavelength division multiplexer; L, lens; FC, fiber collimator; LPF, long-pass filter. The setup is composed of three parts, including a home-built c. w. tunable random Raman fiber laser (RRFL) at a wavelength around \\(1350\\mathrm{nm}\\) , which is used as DFG pump source, as depicted in Supplementary Note 4, a commercial erbium-doped fiber amplifier (EDFA) seeded by tunable signal-frequency DFB laser (Conquer, KG-TLS-13-P-FA) emitting in the spectral range of \\(1527 - 1567\\mathrm{nm}\\) as DFG signal source, and a newly fabricated ZGP waveguide, cut at \\(\\theta = 65.4^{\\circ}\\) and \\(\\phi = 0^{\\circ}\\) . The RRFL and EDFA sources are combined by a fiber wavelength division multiplexer (WDM) as an all-fiber laser driver for DFG in the ZGP waveguide. (b) Typical pump spectra running from \\(1340\\mathrm{nm}\\) to \\(1370\\mathrm{nm}\\) . (c) Typical signal spectra running from \\(1550\\mathrm{nm}\\) to \\(1565\\mathrm{nm}\\) . (d) The measured spectra of generated LWIR lasing tunable from \\(10.28\\mu \\mathrm{m}\\) to \\(10.58\\mu \\mathrm{m}\\) through DFG in the ZGP waveguide.",
+    "footnote": [],
+    "bbox": [
+      [
+        228,
+        216,
+        794,
+        523
+      ]
+    ],
+    "page_idx": 23
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_7.jpg",
+    "caption": "Fig. R8. The simulated OPG process in the temporal and spectral domains as a function of the ZGP waveguide length, revealing back conversion happening at a length of \\(\\sim 10 \\mathrm{mm}\\) .",
+    "footnote": [],
+    "bbox": [
+      [
+        156,
+        95,
+        844,
+        343
+      ]
+    ],
+    "page_idx": 25
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_8.jpg",
+    "caption": "Fig. R9. Comparison of the measured pump spectra before entering and after propagating through the \\(10\\mathrm{-mm - long}\\chi^{(2)}\\) ZGP waveguide.",
+    "footnote": [],
+    "bbox": [
+      [
+        305,
+        430,
+        696,
+        640
+      ]
+    ],
+    "page_idx": 27
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_9.jpg",
+    "caption": "Fig. R11. The numerical analysis based on finite-element method in calculating transmission losses of ordinary and extraordinary fundamental modes at the pump \\((2.4 \\mu \\mathrm{m})\\) , signal \\((3.4 \\mu \\mathrm{m})\\) and idler \\((8 \\mu \\mathrm{m})\\) wavelengths when propagating along different angles \\(\\theta\\) with the optical axis, in the ZGP \\(\\chi^{(2)}\\) birefringence waveguide",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 27
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_1.jpg",
+    "caption": "Fig. 1(c) Calculated propagation losses of fundamental modes in the designed \\(\\chi^{(2)}\\) waveguide in TE and TM polarizations with the angle between the propagation direction and the crystal optical axis as \\(48.3^{\\circ}\\) , in a broad spectral range of \\(2 - 11\\mu \\mathrm{m}\\) . The results indicate that the propagation loss of fundamental TE and TM mode is less than \\(1.5\\mathrm{dB / cm}\\) and \\(4.0\\mathrm{dB / cm}\\) , respectively. Notably, calculated propagation losses of fundamental modes are less than \\(0.05\\mathrm{dB / cm}\\) at \\(8\\mu \\mathrm{m}\\) , and less than \\(0.25\\mathrm{dB / cm}\\) (TE) and \\(0.5\\mathrm{dB / cm}\\) (TM) at \\(9\\mu \\mathrm{m}\\) , respectively. The raise of waveguide losses in the wavelength range of \\(9 - 11\\mu \\mathrm{m}\\) is attributed to the absorption of fused silica substrate peaked at \\(9.5\\mu \\mathrm{m}\\) , as shown in the top inset which depicts the absorption coefficient (imaginary part of complex refractive index) of silica, as a function of wavelength \\(^{22}\\) . Meanwhile, the material loss of ZGP crystal is minimal in the spectral range of \\(2 - 11\\mu \\mathrm{m}\\) as shown by the measured transmission spectrum of a \\(10\\mathrm{-nm}\\) -thick uncoated ZGP crystal in the bottom inset of Fig. 1(c).",
+    "footnote": [],
+    "bbox": [
+      [
+        267,
+        303,
+        707,
+        513
+      ]
+    ],
+    "page_idx": 28
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_10.jpg",
+    "caption": "Fig. R12. The 3-dimensional characterization of the waveguide surface and side-wall roughness. (a) The scanning-electron image of the \\(\\chi^{(2)}\\) ZGP waveguide. (b) The measurement apparatus: laser microscope (Olympus, OSL5000), and the schematic of the microscopic profile measurement of the groove sidewall by tilting the ZGP waveguide by 30 degrees. (c) The measured cross-sectional profile of the \\(\\chi^{(2)}\\) ZGP waveguide. (d) The measured three-dimensional surface profile of the \\(\\chi^{(2)}\\) ZGP waveguide. An average line roughness between points A and B on the micro-groove side wall by measuring eight grooves, revealing an averaged roughness value of \\(0.595 \\mu \\mathrm{m}\\) . The baseline of \\(0 \\mu \\mathrm{m}\\) is set based on the original surface of the ZGP crystal. The sampling length is \\(639.4 \\mu \\mathrm{m}\\) .",
+    "footnote": [],
+    "bbox": [
+      [
+        165,
+        306,
+        844,
+        619
+      ]
+    ],
+    "page_idx": 34
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_11.jpg",
+    "caption": "Fig. R13. The repeated coupling efficiency measurement for six times by using the pinhole method, and the corresponding quantum conversion efficiencies.",
+    "footnote": [],
+    "bbox": [
+      [
+        293,
+        338,
+        696,
+        572
+      ]
+    ],
+    "page_idx": 35
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1eb5a9097a98c358a3d5ddb23a47389b98cafbd877613684718e11c19c97fe39/images_list.json

@@ -0,0 +1,25 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Supplementary_Figure_8.jpg",
+    "caption": "Supplementary Figure 8: a) Top view and side view of \\(G / TaS_{2}\\) heterostructure with the \\(5\\times 5G / \\sqrt{13}\\times \\sqrt{13} TaS_{2}R13.9^{0}\\) supercell containing 89 atoms with one star used in the DFT calculation (reproduced from main text Fig. 2d). b) DFT calculated charge transfer of the graphene layer based the \\(5\\times 5G / \\sqrt{13}\\times \\sqrt{13} TaS_{2}R13.9^{0}\\) supercell shows a local modulation of doping with the periodicity of the CDW in \\(TaS_{2}\\) . Color scales range from -4 \\(10^{4}e / \\mathring{A}^{3}\\) (red) to -2 \\(10^{4}e / \\mathring{A}^{3}\\) (green). c) Out-of-plane displacement of Graphene induced by \\(TaS_{2}\\) CDW based on the \\(5\\times 5G / \\sqrt{13}\\times \\sqrt{13} TaS_{2}R13.9^{0}\\) supercell. d) Top view and side view of \\(G / TaS_{2}\\) heterostructure with the \\(8\\times 8G / \\sqrt{39}\\times \\sqrt{39} TaS_{2}R16.5^{0}\\) supercell containing 245 atoms with 3 stars used in the DFT calculation. e) DFT calculated charge transfer of the graphene layer based the \\(8\\times 8G / \\sqrt{39}\\times \\sqrt{39} TaS_{2}R16.5^{0}\\) supercell shows a local modulation of doping with the periodicity of the CDW in \\(TaS_{2}\\) . Color scales range from -5 \\(10^{4}e / \\mathring{A}^{3}\\) (red) to -3 \\(10^{4}e / \\mathring{A}^{3}\\) (light blue). f) Out-of-plane displacement of Graphene induced by \\(TaS_{2}\\) CDW based on the \\(8\\times 8G / \\sqrt{39}\\times \\sqrt{39} TaS_{2}R16.5^{0}\\) supercell. The black, green, and orange diamonds in (a) and (d) represent the \\(G / TaS_{2}\\) supercell, \\(TaS_{2}\\) unit cell, and graphene unit cell, respectively. The lattice constants indicated in (a) and (d) are determined by DFT fully relaxation calculations. The interlayer distances between \\(TaS_{2}\\) and Graphene are 4.03 \\(\\mathring{A}\\) (a) and 4.04 \\(\\mathring{A}\\) (d) in the minima \\(5\\times 5G / \\sqrt{13}\\times \\sqrt{13} TaS_{2}R13.9^{0}\\) supercell and \\(8\\times 8G / \\sqrt{39}\\times \\sqrt{39} TaS_{2}R16.5^{0}\\) supercell, respectively.",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 0
+  },
+  {
+    "type": "image",
+    "img_path": "images/Supplementary_Figure_9.jpg",
+    "caption": "Supplementary Figure 9: (a) Side view and top view of \\(G / T a S_{2}\\) monolayer discussed in the main text. (b) Band structure of \\(G / T a S_{2}\\) . Red and blue curves denote spin up and spin down bands, respectively. (c) Zoom-in plot of the Dirac bands of (b). The proximity CDW gap of \\(\\sim 1.7 \\mathrm{meV}\\) is induced at the Dirac point. (d) Band structure of CDW-distorted graphene with \\(T a S_{2}\\) layer removed. A CDW gap of size 1.6 meV can also be seen at the graphene Dirac point. (e-h) Counterparts of (a-d), respectively, for \\(G / T a S_{2}\\) bilayer with AA-stacking. (i-l) Counterparts of (a-d), respectively, for \\(G / T a S_{2} b i l a y e r\\) with AL-stacking. All the lattice structures are from DFT fully relaxation calculations. In all \\(G / T a S_{2}\\) heterostructures considered, the graphene CDW gaps at the Dirac point are in meV order ranging from 0.9 meV to 1.9 meV compatible with each other. With the \\(T a S_{2}\\) layers removed, the distorted graphene exhibits CDW gap at the Dirac point of 1.6, 4.0, and 4.1 meV compatible with other as well as with the \\(G / T a S_{2}\\) cases.",
+    "footnote": [],
+    "bbox": [
+      [
+        93,
+        222,
+        896,
+        814
+      ]
+    ],
+    "page_idx": 2
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1eb5a9097a98c358a3d5ddb23a47389b98cafbd877613684718e11c19c97fe39/supplementary_0_Peer Review File__1eb5a9097a98c358a3d5ddb23a47389b98cafbd877613684718e11c19c97fe39.mmd

@@ -0,0 +1,55 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+Proximity induced charge density wave in graphene/1T- TaS2 Heterostructure  
+
+![](images/Supplementary_Figure_8.jpg)
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+I think all my concerns have been well addressed in the point- to- point reply and the revised manuscript. I would recommend the publication of this work now.  
+
+Reviewer #2 (Remarks to the Author):  
+
+The manuscript reports on a discovery of proximity CDW in graphene on 1t- TaS2. The STM/STS images are convincing, the claim of the authors relies on the topography maps within and outside of the Mott gap of the TaS2 CDW phase. The authors perform supporting measurements and even DFT simulations, exhausting alternative explanations. The argument based on out- of- phase proximity induced CDW in graphene makes sense, I recommend publishing this work.  
+
+Certainly, the manuscript could be improved for clarity, especially regarding the details of the simulations. For example, what are the geometries (relevant distances and Gr and TaS unit cells in the supercells)? The meV resolution with which the authors discuss their band structures is rather hopeless to see on their energy scale, and probably their conclusions would strongly depend on the stacking (not clear what it is in the experiment as well), but as a supporting information it is useful, albeit not truly representative.  
+
+Minor: I am not sure that anti- lattice is a common terminology. Probably the authors meant dual lattice? "Hexagonal lattice is the dual lattice to triangular."
+
+<--- Page Split --->
+
+Reviewer #2 (Remarks to the Author):  
+
+Certainly, the manuscript could be improved for clarity, especially regarding the details of the simulations. For example, what are the geometries (relevant distances and Gr and TaS unit cells in the supercells)? The meV resolution with which the authors discuss their band structures is rather hopeless to see on their energy scale, and probably their conclusions would strongly depend on the stacking (not clear what it is in the experiment as well), but as a supporting information it is useful, albeit not truly representative.  
+
+Our Reply:  
+
+1. The geometries (relevant distances and Gr and TaS unit cells in the supercells) are included in the revised Supplementary Fig. 8 as also attached below.  
+
+![](images/Supplementary_Figure_9.jpg)
+
+<center>Supplementary Figure 8: a) Top view and side view of \(G / TaS_{2}\) heterostructure with the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell containing 89 atoms with one star used in the DFT calculation (reproduced from main text Fig. 2d). b) DFT calculated charge transfer of the graphene layer based the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell shows a local modulation of doping with the periodicity of the CDW in \(TaS_{2}\) . Color scales range from -4 \(10^{4}e / \mathring{A}^{3}\) (red) to -2 \(10^{4}e / \mathring{A}^{3}\) (green). c) Out-of-plane displacement of Graphene induced by \(TaS_{2}\) CDW based on the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell. d) Top view and side view of \(G / TaS_{2}\) heterostructure with the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell containing 245 atoms with 3 stars used in the DFT calculation. e) DFT calculated charge transfer of the graphene layer based the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell shows a local modulation of doping with the periodicity of the CDW in \(TaS_{2}\) . Color scales range from -5 \(10^{4}e / \mathring{A}^{3}\) (red) to -3 \(10^{4}e / \mathring{A}^{3}\) (light blue). f) Out-of-plane displacement of Graphene induced by \(TaS_{2}\) CDW based on the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell. The black, green, and orange diamonds in (a) and (d) represent the \(G / TaS_{2}\) supercell, \(TaS_{2}\) unit cell, and graphene unit cell, respectively. The lattice constants indicated in (a) and (d) are determined by DFT fully relaxation calculations. The interlayer distances between \(TaS_{2}\) and Graphene are 4.03 \(\mathring{A}\) (a) and 4.04 \(\mathring{A}\) (d) in the minima \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell and \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell, respectively. </center>  
+
+2. Since the accuracy of our calculations is well-below meV, the CDW gaps of graphene in meV level as shown in Supplementary Fig. S9 (also attached below) are reliable. It is reasonable, as the reviewer mentioned, that such small graphene CDW gaps are truly difficult to be observed experimentally. Future genius work may resolve this issue.
+
+<--- Page Split --->
+
+3. To examine the TaS2 stacking effect on the graphene CDW gap, we further performed calculations for G/TaS2 bilayer with AA-stacking and AL-stacking as shown in Fig. S9(e-h) and Fig. S9(i-l), respectively, as also attached below. Different stackings affect the induced graphene CDW gap slightly with the gap in meV order ranging from 0.9 meV to 1.9 meV compatible with each other. With the TaS2 layers removed, the distorted graphene exhibits CDW gap at the Dirac point of 1.6, 4.0, and 4.1 meV compatible with other as well as with the G/TaS2 cases. In all cases, the induced graphene CDW gap survives with different TaS2 stackings.  
+
+![PLACEHOLDER_3_0]
+
+<center>Supplementary Figure 9: (a) Side view and top view of \(G / T a S_{2}\) monolayer discussed in the main text. (b) Band structure of \(G / T a S_{2}\) . Red and blue curves denote spin up and spin down bands, respectively. (c) Zoom-in plot of the Dirac bands of (b). The proximity CDW gap of \(\sim 1.7 \mathrm{meV}\) is induced at the Dirac point. (d) Band structure of CDW-distorted graphene with \(T a S_{2}\) layer removed. A CDW gap of size 1.6 meV can also be seen at the graphene Dirac point. (e-h) Counterparts of (a-d), respectively, for \(G / T a S_{2}\) bilayer with AA-stacking. (i-l) Counterparts of (a-d), respectively, for \(G / T a S_{2} b i l a y e r\) with AL-stacking. All the lattice structures are from DFT fully relaxation calculations. In all \(G / T a S_{2}\) heterostructures considered, the graphene CDW gaps at the Dirac point are in meV order ranging from 0.9 meV to 1.9 meV compatible with each other. With the \(T a S_{2}\) layers removed, the distorted graphene exhibits CDW gap at the Dirac point of 1.6, 4.0, and 4.1 meV compatible with other as well as with the \(G / T a S_{2}\) cases. </center>
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1eb5a9097a98c358a3d5ddb23a47389b98cafbd877613684718e11c19c97fe39/supplementary_0_Peer Review File__1eb5a9097a98c358a3d5ddb23a47389b98cafbd877613684718e11c19c97fe39_det.mmd

@@ -0,0 +1,68 @@
+<|ref|>title<|/ref|><|det|>[[61, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[70, 111, 362, 140]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[70, 162, 901, 219]]<|/det|>
+Proximity induced charge density wave in graphene/1T- TaS2 Heterostructure  
+
+<|ref|>image<|/ref|><|det|>[[57, 732, 239, 782]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 911, 785]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 876, 143]]<|/det|>
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 183, 313, 199]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 240, 402, 256]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 268, 816, 303]]<|/det|>
+I think all my concerns have been well addressed in the point- to- point reply and the revised manuscript. I would recommend the publication of this work now.  
+
+<|ref|>text<|/ref|><|det|>[[116, 344, 402, 360]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 371, 866, 462]]<|/det|>
+The manuscript reports on a discovery of proximity CDW in graphene on 1t- TaS2. The STM/STS images are convincing, the claim of the authors relies on the topography maps within and outside of the Mott gap of the TaS2 CDW phase. The authors perform supporting measurements and even DFT simulations, exhausting alternative explanations. The argument based on out- of- phase proximity induced CDW in graphene makes sense, I recommend publishing this work.  
+
+<|ref|>text<|/ref|><|det|>[[115, 501, 880, 620]]<|/det|>
+Certainly, the manuscript could be improved for clarity, especially regarding the details of the simulations. For example, what are the geometries (relevant distances and Gr and TaS unit cells in the supercells)? The meV resolution with which the authors discuss their band structures is rather hopeless to see on their energy scale, and probably their conclusions would strongly depend on the stacking (not clear what it is in the experiment as well), but as a supporting information it is useful, albeit not truly representative.  
+
+<|ref|>text<|/ref|><|det|>[[115, 660, 855, 695]]<|/det|>
+Minor: I am not sure that anti- lattice is a common terminology. Probably the authors meant dual lattice? "Hexagonal lattice is the dual lattice to triangular."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[90, 71, 392, 88]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[88, 89, 900, 201]]<|/det|>
+Certainly, the manuscript could be improved for clarity, especially regarding the details of the simulations. For example, what are the geometries (relevant distances and Gr and TaS unit cells in the supercells)? The meV resolution with which the authors discuss their band structures is rather hopeless to see on their energy scale, and probably their conclusions would strongly depend on the stacking (not clear what it is in the experiment as well), but as a supporting information it is useful, albeit not truly representative.  
+
+<|ref|>text<|/ref|><|det|>[[90, 220, 176, 236]]<|/det|>
+Our Reply:  
+
+<|ref|>text<|/ref|><|det|>[[90, 237, 884, 273]]<|/det|>
+1. The geometries (relevant distances and Gr and TaS unit cells in the supercells) are included in the revised Supplementary Fig. 8 as also attached below.  
+
+<|ref|>image<|/ref|><|det|>[[110, 288, 896, 616]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[88, 622, 905, 820]]<|/det|>
+<center>Supplementary Figure 8: a) Top view and side view of \(G / TaS_{2}\) heterostructure with the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell containing 89 atoms with one star used in the DFT calculation (reproduced from main text Fig. 2d). b) DFT calculated charge transfer of the graphene layer based the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell shows a local modulation of doping with the periodicity of the CDW in \(TaS_{2}\) . Color scales range from -4 \(10^{4}e / \mathring{A}^{3}\) (red) to -2 \(10^{4}e / \mathring{A}^{3}\) (green). c) Out-of-plane displacement of Graphene induced by \(TaS_{2}\) CDW based on the \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell. d) Top view and side view of \(G / TaS_{2}\) heterostructure with the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell containing 245 atoms with 3 stars used in the DFT calculation. e) DFT calculated charge transfer of the graphene layer based the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell shows a local modulation of doping with the periodicity of the CDW in \(TaS_{2}\) . Color scales range from -5 \(10^{4}e / \mathring{A}^{3}\) (red) to -3 \(10^{4}e / \mathring{A}^{3}\) (light blue). f) Out-of-plane displacement of Graphene induced by \(TaS_{2}\) CDW based on the \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell. The black, green, and orange diamonds in (a) and (d) represent the \(G / TaS_{2}\) supercell, \(TaS_{2}\) unit cell, and graphene unit cell, respectively. The lattice constants indicated in (a) and (d) are determined by DFT fully relaxation calculations. The interlayer distances between \(TaS_{2}\) and Graphene are 4.03 \(\mathring{A}\) (a) and 4.04 \(\mathring{A}\) (d) in the minima \(5\times 5G / \sqrt{13}\times \sqrt{13} TaS_{2}R13.9^{0}\) supercell and \(8\times 8G / \sqrt{39}\times \sqrt{39} TaS_{2}R16.5^{0}\) supercell, respectively. </center>  
+
+<|ref|>text<|/ref|><|det|>[[88, 849, 903, 923]]<|/det|>
+2. Since the accuracy of our calculations is well-below meV, the CDW gaps of graphene in meV level as shown in Supplementary Fig. S9 (also attached below) are reliable. It is reasonable, as the reviewer mentioned, that such small graphene CDW gaps are truly difficult to be observed experimentally. Future genius work may resolve this issue.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[88, 70, 899, 201]]<|/det|>
+3. To examine the TaS2 stacking effect on the graphene CDW gap, we further performed calculations for G/TaS2 bilayer with AA-stacking and AL-stacking as shown in Fig. S9(e-h) and Fig. S9(i-l), respectively, as also attached below. Different stackings affect the induced graphene CDW gap slightly with the gap in meV order ranging from 0.9 meV to 1.9 meV compatible with each other. With the TaS2 layers removed, the distorted graphene exhibits CDW gap at the Dirac point of 1.6, 4.0, and 4.1 meV compatible with other as well as with the G/TaS2 cases. In all cases, the induced graphene CDW gap survives with different TaS2 stackings.  
+
+<|ref|>image<|/ref|><|det|>[[93, 222, 896, 814]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[88, 816, 904, 928]]<|/det|>
+<center>Supplementary Figure 9: (a) Side view and top view of \(G / T a S_{2}\) monolayer discussed in the main text. (b) Band structure of \(G / T a S_{2}\) . Red and blue curves denote spin up and spin down bands, respectively. (c) Zoom-in plot of the Dirac bands of (b). The proximity CDW gap of \(\sim 1.7 \mathrm{meV}\) is induced at the Dirac point. (d) Band structure of CDW-distorted graphene with \(T a S_{2}\) layer removed. A CDW gap of size 1.6 meV can also be seen at the graphene Dirac point. (e-h) Counterparts of (a-d), respectively, for \(G / T a S_{2}\) bilayer with AA-stacking. (i-l) Counterparts of (a-d), respectively, for \(G / T a S_{2} b i l a y e r\) with AL-stacking. All the lattice structures are from DFT fully relaxation calculations. In all \(G / T a S_{2}\) heterostructures considered, the graphene CDW gaps at the Dirac point are in meV order ranging from 0.9 meV to 1.9 meV compatible with each other. With the \(T a S_{2}\) layers removed, the distorted graphene exhibits CDW gap at the Dirac point of 1.6, 4.0, and 4.1 meV compatible with other as well as with the \(G / T a S_{2}\) cases. </center>
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1eb88b8e99ba8ca78ed896f518cb1347831354bf2216abcb3b0f8e3c72eb76f4/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1eb88b8e99ba8ca78ed896f518cb1347831354bf2216abcb3b0f8e3c72eb76f4/supplementary_0_Peer Review File__1eb88b8e99ba8ca78ed896f518cb1347831354bf2216abcb3b0f8e3c72eb76f4.mmd

@@ -0,0 +1,422 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+A multi- factor trafficking site on spliceosome remodeling enzyme, BRR2, recruits C9ORF78 to regulate alternative splicing  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+The manuscript by Bergfort et al., employs methods of structural and molecular biology to study the role of poorly characterised C9ORF78 protein in splicing. C9ORF78 has been previously reported as a constituent part of spliceosome whose function remained largely unknown. The manuscript reports a cryoEM structure of C9ORF78 bound to its interaction partner BRR2. The second cryoEM structure, reported in the manuscript is that of BRR2 bound to an inhibitor FBP21. These structures together with pulldown experiments allowed authors to concluded that C9ORF78 and FBP21 binding to BRR2 is mutually exclusive. Armed with that knowledge, the authors then went on to investigate the effect of C9ORF78 knockdown on mRNA splicing. siRNA knockdown of C9ORF78 had an effect on 3' splice site selection which induced exon skipping events, or altered exon length if the two 3' splice sites were in close proximity.  
+
+This structural knowledge allowed the authors to pinpoint C9ORF78 residues, critical for interaction with BRR2. The fact that these residues are conserved over hundreds millions of years of evolution and the corresponding point mutant C9ORF78- R41A fails to bind to BRR2 is a highlight of the paper. The place of C9ORF78 in splicing is further cemented by UV crosslinking experiments that show a link between C9ORF78 and U5 snRNA. These experiments however did not reveal significantly enriched mRNAs, suggesting that C9ORF78 does not affect splicing via direct binding to mRNAs, and rather exerts it's influence on the 3' splice site selection via its spliceosome interaction partners.  
+
+This body of work is novel and represents a significant interest to the field of splicing and broader readership. The structures presented in the paper are of particular value to the field as C9ORF78 has not been observed in the published spliceosomal structures. However I have 1 major and several minor comments that I hope would help the authors improve their manuscript further.  
+
+## Major point  
+
+The manuscript presents detailed structural description of how C9ORF78 interacts with BRR2 and PRPF8 proteins. However, experimental support for functional significance of these interactions is perhaps the weakest part of otherwise solid story the authors portray. Indeed, this support becomes vital, as the authors report that more than a couple of thousand proteins are enriched over background in C9ORF78 IP. This would suggest that in addition to a direct interaction with BRR2, C9ORF78 could influence splicing indirectly via it's non- spliceosomal interaction partners. The functional validation experiment, shown in figure 5b has produced mixed results. C9ORF78- WT and C9ORF78- R41A rescue si RNA mediated C9ORF78 knockdown to identical extents (figure 5b SMARCA4 and C1ORF131 panels). For the experiment in panel PTBP2, the authors compared C9ORF78- WT knockdown rescue to that of C9ORF78- R41A and reported a difference between the two. Firstly, that comparison may not be the right one as C9ORF78- WT "rescues" exon skipping to the levels better than in si- Control. Why does it do that is an unsolved question, perhaps C9ORF78 is limiting and it's overexpression produces better splicing efficiency in si- C9ORF78 cells than in unperturbed cells that do not express si resistant C9ORF78. Therefore, as far as splicing efficiency is concerned, a comparison between PTBP2 exon 10 retention in si- Control against a C9ORF78 knockdown rescued by C9ORF78- R41A is the one to make in this case. It would appear as if this comparison would not produce a statistically significant difference between the two. If that is correct, then we would have to conclude that splicing of PTBP2 exon 10 is indistinguishable in si- Control vs si- C9ORF78 rescued with C9ORF78- R41A, hence C9ORF78 interaction with BRR2 is dispensable for splicing of that particular exon.  
+
+Even if the authors disagree with this logic, I would draw their attention to absolute fold change values between si- Control and C9ORF78- R41A they report in panel PTBP2 of figure 5. The reported difference in fold change is miniscule. How would that translate into protein level and what effect would it have on the function of PTBP2? From figure 5 I concluded that the effect would negligible at best.
+
+<--- Page Split --->
+
+Therefore I would strongly urge the authors to find a clear cut example for functional significance of C9ORF78- BRR2 interaction. A good starting point would be to expand the experimental approach and analysis reported in figure 4 onto figure 5 and perform the necessary RNA- seq experiments with the samples of Figure 5.  
+
+Minor points:  
+
+1. Figure 2a C9ORF78 labelling - there are 2 alpha 1 and no alpha 3 present.  
+
+2. "We exchanged C9ORF78 F8 and R41 individually for alanine residues, and tested BRR2 binding of the C9ORF78 variants via analytical SEC. While C9ORF78F8A showed reduced binding to BRR2 in analytical SEC, BRR2 binding by C9ORF78R41A in vitro was completely abolished"  
+
+C9ORF78- F8A is not present in Figure 3b, perhaps the authors have forgotten to add the panel with WB analysis of a SEC experiment for the C9ORF78F8A mutant.  
+
+3. Figure 4c. The first 2 panel appear to be only depicting statistically significant splicing events, while the NAGNAG panel appears to show all events. Perhaps the authors could consider showing only statistically significant NAGNAG events similarly to the previous 2 panels of the figure.  
+
+4. Figure 5a. Labelling of y axis. It would be beneficial to explain in the text or figure legend what PSI is and how it is calculated.  
+
+5. Figure 5b. It would be beneficial to explain in the text or figure legend how the fold change was calculated.  
+
+6. Figure 6a. The authors could look at their FLASH data for sites of crosslink between C9ORF78 and U5 snRNA. With FLASH I would expect that an RNA-protein crosslink site would manifest itself as a tight cluster of deletions or point mutations in cDNA sequences. If it is possible to elucidate the exact crosslink site, it could significantly contribute to the structural data and perhaps could allow further speculations on the location of the part of C9ORF78 that is not engaged in interaction with BRR2.  
+
+7. Comments on the C9ORF78 interactome.  
+
+The authors show that spliceosomal proteins are enriched in in both C9ORF78- WT and C9ORF78- R41A. It would be good to see a GO term analysis, considering the manuscript reports that C9ORF78 IP results in enrichment over control for 2411 proteins. This is perhaps 1/5th of the proteome, expressed in HEK cells. In fact, the very limited GO analysis that I performed on the manuscripts data does reveal a strong enrichment for GO CC "spliceosome", which helps interpretation of the data, as it shows that the IP has been specific.  
+
+To that end, I would suggest that in addition to figure 6b the chapter "C9ORF78 interacts with additional spliceosomal proteins" could benefit from a figure that would depict the proteins, belonging to two spliceosomal complexes B and C. Those proteins should be colour coded to depict 2 key parameters, discussed in the chapter: whether a given protein was detected as enriched in the IP and if so, whether its enrichment is higher or lower in C9ORF78- WT compared to C9ORF78- R41A. I am of the opinion that this would be a great visual help for the chapter. Perhaps the authors could consider including a figure like that in the manuscript.  
+
+Reviewer #2: Remarks to the Author:
+
+<--- Page Split --->
+
+The manuscript by Bergfort et al reports an unstructured protein C9ORF78 tightly interacts with the key spliceosomal RNA helicase BRR2 through a series of evidence, including yeast two- hybrid screen, in vitro protein- protein interaction, affinity purification and mass spectrometry, and cryoEM structures. They find that C9ORF78 and another spliceosomal protein FBP21 interact with the C- terminal cassette of BRR2 in a mutually exclusive manner using both structural information and biochemical competition assay. RNAi of C9ORF78 leads to alternative splicing changes including a substantial usage of alternative 3'SSs. This manuscript provides insightful information in understanding the function of a flexibly or dynamically bound spliceosomal protein during the process of spliceosome assembly and catalysis. In general, their findings are convincing and interesting, and the manuscript is well written.  
+
+Below are several questions and concerns:  
+
+1. Typo: In Figures 1-3, several places of C9ORF78 are shown in "C9ORF8"; In Figures 2a and 3a, there are two a1s labelled for the structure of C9ORF78, of which one should be "a3"; In Figure 7, consistent with its legends, the forest green component marked as "PRP22" should be "PRPF22".  
+
+2. C9ORF78 was observed in the C complex and the binding of C9ORF78 with BRR2 was also described in S. pombe. In Figures 1a & 1b and later, overexpression of GST- or Flag-tagged C9ORF78 are used for SEC and IP experiments, indicating that the interaction between C9ORF78 and BRR2 might be not strong in HEK293 cells. Could this be done using an endogenous normally expressed C9ORF78, either through an C9ORF78-specific antibody or by a CRSIPR-Cas9 knock-in tag system?  
+
+3. In Figure 4a, this should be RT-qPCR, not qPCR, detection of mRNAs, better to present this with an additional agarose gel.  
+
+4. In Figure 4b, SE (skipped exon) is the dominant feature of AS events when the C9ORF78 is KD. I am curious, what is the feature of those exons? For those increased inclusion of exons (up-regulated), do they have weaker 3'SSs; vice versa, for those decreased inclusion of exons (down-regulated), do they have stronger 3'SSs? Therefore, I would like to suggest strength analyses (scores) of both the 3'SSs and 5'SSs of those SE events.  
+
+5. In Figure 4c, the negative value of deltaPSI are not presented, this is confusing of which sample vs which sample. In addition, RMATS should be rMATS.  
+
+6. The primes in "3' or 5'-splice site" are incorrect, should be 3' or 5'.  
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+In this manuscript the authors present a rather comprehensive analysis of the interactions of the human splicing factor C9ORF78 with the spliceosome and particularly with the helicase Brr2, revealing a previously unappreciated role for this factor during catalysis in modulating 3'SS selection. The manuscript suggests a molecular mechanism that can explain how Brr2 may act during the catalytic stage.  
+
+I generally find the authors' data compelling, of timely interest to the field, and potentially more broadly, while the proposed mechanistic models are mostly supported by the presented experiments. However, there are a few points that the authors should try to address prior to publication.  
+
+While the cryo- EM analysis appears sound and expertly performed, the authors do not provide any clear figure showing their fit of the C9ORF78 model into their determined EM map. Given the claimed high resolution, it is critical for the authors to show this data as a figure in the paper, especially for the critical parts where C9ORF78 interacts with Brr2. The mutational data in Fig. 3 does support the proposed modelling, but still it is impossible to fully judge the quality of the map and the model fit
+
+<--- Page Split --->
+
+without this data, especially as the presented local resolution in Sup.Fig. 2e shows significant variation in local resolution along the proposed C9ORF78 path. Certainly the map presented in Fig. 1c is contoured at an RMSD that makes it hard to judge the presence of high resolution features.  
+
+Although the in vitro assays show a modest effect of C9ORF78 on Brr2 helicase activity for U4/U6, I think it is premature to argue that C9ORF78 does not act through modulation of Brr2 helicase activity, as the authors do on p.8. The assays used as not the native situation in the spliceosome and C9ORF78 binds at the U2/U6 stage of splicing, while Brr2 has been implicated in spliceosome disassembly. Thus, one could easily imagine that C9ORF78 may act at the disassembly stage and that its role in 3'SS selection could be coupled to a role in modulating Brr2 activity during disassembly. I am aware that this role for Brr2 is a matter of contention in the field but I think the authors should be more cautious with their statements here, as their data cannot exclude a role for C9ORF78 in modulating Brr2 during this later stage.  
+
+The observations regarding competition between FBP21 and C9ORF78 for Brr2 binding are strongly supported by the data in Fig. 2d. Nonetheless, it is unclear why the authors chose to use Brr2HR complexes rather than complexes that also contained the Jab1 domain of Prp8, given that the Jab1 domain remains bound to Brr2 from the B complex onwards and that the FBP21 interaction is observed in the B complex structure in the context of Brr2 being bound to Jab1. I think the authors could and likely should strengthen their argument here by performing the competition experiment with Brr2HR/Jab1/C9ORF78 complexes and even better do it in the proper physiological progression by asking whether C9ORF78 can compete off FBP21, which is what they propose actually happens during the splicing pathway.  
+
+Finally, while I appreciate the authors' model for C9ORF78 function during the C\* stage as a compelling main mechanism of action, I think the authors need to be much more careful with their discussion of various proteins they claim are present at the C complexes stage, such as DDX23 or Prp6. These have generally only been detected as such in C complexes prepared by biochemical stalls that are not entirely clean and could have contaminating earlier and later complexes. A good example are studies that use a 3'SS mutation to capture C complexes and which are now known from cryo- EM studies to actually reach the C\* and even P complex stage and then revert back to the C complex stage. Simply citing proteomic analyses of such preparations as evidence of stage- specific association is sufficient evidence for such claims. None of the single particle EM studies of C or C\* complexes have identified any subpopulations containing DDX23 for example, yet the authors routinely seem to suggest this protein could bind at the this stage. The claim that their IP of DDX23 occurs at the C complex stage is particularly problematic in this sense, especially as one could imagine much more easily how complexes that have transitioned to the B complex stage may not have fully lost DDX23 until the Bact stage, making the exchange of FBP21 for C9ORF78 an alternative point at which a transient interaction between C9ORF78 and DDX23 could have occurred. Similarly, the claim that Prp22 binds at the C complex stage should be revised, as what people have reported as binding at that stage most likely results from 3'SS mutant complexes that have reached the C\* stage and then reverted back to the C complex stage without Prp22 dissociation, as was shown in a recent study on equilibrium of spliceosome conformations during catalysis in yeast. I urge the authors to be much more rigorous with their terminology when describing potential binding to various complexes.  
+
+Related to this matter, I find that the authors overlook too quickly a potential role for C9ORF78 in regulating splice site use also during the B to Bact transition. They observe significant numbers of exon skipping and mutually exclusive exon use changes in their KD experiments. It is much harder to imagine how such events could be regulated at the C or C\* stage, but much easier to imagine how transfer of the 5'SS and docking of the BP helix at the active site, which occur during the B to Bact transition, or are influenced by the relative stability of these complexes, could impact these types of alternative splicing events. Indeed, the crosslinks to U5 snRNA are consistent with such a role and much more likely to reflect interactions at the Bact stage than at the C complex stage, when U5 is buried very deeply into the active site, making an interaction with a flexible part of C9ORF78 harder to
+
+<--- Page Split --->
+
+imagine. Brr2- dependent association with Cwc27 also supports this idea, as in the Bact structure the flexible C9ORF78 residues proposed to interact with Prp22 in C\*, could easily be imagined to interact with Cwc27 in Bact. Since C9ORF78 likely regulates the Bact transition, as the authors argue with strong experimental support, they should at least discuss the possibility of an earlier role in regulating alternative splicing at this stage through some of the other factors they observe in their IP studies.
+
+<--- Page Split --->
+
+## Response to Reviewer Comments  
+
+Reviewer comments are repeated in bold italics, responses are in regular font, changed text passages are highlighted in yellow.  
+
+## Reviewer #1:  
+
+The manuscript by Bergfort et al., employs methods of structural and molecular biology to study the role of poorly characterised C9ORF78 protein in splicing. C9ORF78 has been previously reported as a constituent part of spliceosome whose function remained largely unknown. The manuscript reports a cryoEM structure of C9ORF78 bound to its interaction partner BRR2. The second cryoEM structure, reported in the manuscript is that of BRR2 bound to an inhibitor FBP21. These structures together with pulldown experiments allowed authors to concluded that C9ORF78 and FBP21 binding to BRR2 is mutually exclusive. Armed with that knowledge, the authors then went on to investigate the effect of C9ORF78 knockdown on mRNA splicing. siRNA knockdown of C9ORF78 had an effect on 3' splice site selection which induced exon skipping events, or altered exon length if the two 3' splice sites were in close proximity.  
+
+This structural knowledge allowed the authors to pinpoint C9ORF78 residues, critical for interaction with BRR2. The fact that these residues are conserved over hundreds millions of years of evolution and the corresponding point mutant C9ORF78- R41A fails to bind to BRR2 is a highlight of the paper. The place of C9ORF78 in splicing is further cemented by UV crosslinking experiments that show a link between C9ORF78 and U5 snRNA. These experiments however did not reveal significantly enriched mRNAs, suggesting that C9ORF78 does not affect splicing via direct binding to mRNAs, and rather exerts it's influence on the 3' splice site selection via its spliceosome interaction partners.  
+
+This body of work is novel and represents a significant interest to the field of splicing and broader readership. The structures presented in the paper are of particular value to the field as C9ORF78 has not been observed in the published spliceosomal structures. However I have 1 major and several minor comments that I hope would help the authors improve their manuscript further.  
+
+We thank the reviewer for this very positive overall assessment, specifically for considering our work to be of significant interest to the field and beyond, and our structures to be of particular value.  
+
+## Major point  
+
+The manuscript presents detailed structural description of how C9ORF78 interacts with BRR2 and PRPF8 proteins. However, experimental support for functional significance of these interactions is perhaps the weakest part of otherwise solid story the authors portray. Indeed, this support becomes vital, as the authors report that more than a couple of thousand proteins are enriched over background in C9ORF78 IP. This would suggest that in addition to a direct interaction with BRR2, C9ORF78 could influence splicing indirectly via it's non- spliceosomal interaction partners.  
+
+The functional validation experiment, shown in figure 5b has produced mixed results. C9ORF78- WT and C9ORF78- R41A rescue si RNA mediated C9ORF78 knockdown to identical extents (figure 5b SMARCA4 and C1ORF131 panels). For the experiment in panel PTBP2, the authors compared C9ORF78- WT knockdown rescue to that of C9ORF78- R41A and reported a difference between the two. Firstly, that comparison may not be the right one as C9ORF78- WT "rescues" exon skipping to the levels better than in si- Control. Why does it do that is an unsolved question, perhaps C9ORF78 is limiting and it's overexpression produces better splicing efficiency in si- C9ORF78 cells than in unperturbed cells that do not express si resistant
+
+<--- Page Split --->
+
+C9ORF78. Therefore, as far as splicing efficiency is concerned, a comparison between PTBP2 exon 10 retention in si- Control against a C9ORF78 knockdown rescued by C9ORF78- R41A is the one to make in this case. It would appear as if this comparison would not produce a statistically significant difference between the two. If that is correct, then we would have to conclude that splicing of PTBP2 exon 10 is indistinguishable in si- Control vs si- C9ORF78 rescued with C9ORF78- R41A, hence C9ORF78 interaction with BRR2 is dispensable for splicing of that particular exon. Even if the authors disagree with this logic, I would draw their attention to absolute fold change values between si- Control and C9ORF78- R41A they report in panel PTBP2 of figure 5. The reported difference in fold change is miniscule. How would that translate into protein level and what effect would it have on the function of PTBP2? From figure 5 I concluded that the effect would negligible at best.  
+
+Therefore I would strongly urge the authors to find a clear cut example for functional significance of C9ORF78- BRR2 interaction. A good starting point would be to expand the experimental approach and analysis reported in figure 4 onto figure 5 and perform the necessary RNA- seq experiments with the samples of Figure 5.  
+
+While we generally agree with the reviewer's assessment, we would like to point out that due to the single point mutation we introduced into the C9ORF78R41A variant (which abrogates stable interaction with BRR2 in vitro in the absence of other factors but may be partly "overridden" by additional interactions of C9ORF78 with other components of spliceosomes, as suggested by our IP experiments) and unavoidable overexpression of the siRNA- resistant C9ORF78wt or C9ORF78R41A variants for rescue experiments, full differential rescue effects by the C9ORF78wt or C9ORF78R41A variants are not to be expected.  
+
+We thank the reviewer for the suggestion of conducting KD/rescue experiments in combination with RNA- seq, and took it up. In the course of these additional experiments, we had to conduct a second, independent KD experiment, which allowed us to further validate our observations from the first KD experiment.  
+
+Importantly, the second, independent KD experiment confirmed the findings of the first KD experiment: (i) increased skipping of upstream \(3^{\prime}\) - ss in alternative \(3^{\prime}\) - ss pairs with a strong overlap in the affected \(3^{\prime}\) - ss pairs; (ii) a significant number of affected exon skipping events, albeit with a lower overlap between the two KD events. Consistent with our previous experiments, over- expression of siRNA- resistant variants of both C9ORF78wt and C9ORF78R41A globally reverted C9ORF78 KD- dependent changes in alternative \(3^{\prime}\) - ss usage, unequivocally confirming that the observed effects are specific to C9ORF78, but failing to reveal a direct role of the observed BRR2- C9ORF78 interaction in these events. However, 49 exon skipping events, which were reproducibly altered upon C9ORF78 KD in the two independent KD experiments, were significantly rescued only by over- expression of C9ORF78wt, but not by over- expression of C9ORF78R41A. PTBP2 exon 10 skipping was among the exon skipping events reproducibly affected in both independent KD experiments and was only significantly rescued by over- expression of C9ORF78wt. Furthermore, in the new, global RNA- seq based assessment of KD/rescue, C9ORF78wt over- expression did not restore PTBP2 exon 10 skipping levels "beyond" the si control situation. We now also statistically compare si control and rescue by C9ORF78R41A, as suggested (new Fig. 5g).  
+
+Quantifying the effect of C9ORF78 KD on PTBP2 exon 10 skipping based on the new KD/rescue analysis, revealed that C9ORF78 KD led to a decrease in the level of PTBP2 exon 10 inclusion from \(\sim 62\%\) spliced- in for si control to \(\sim 50\%\) spliced- in for si C9ORF78 (new Fig. 5g). While we agree that this effect is moderate, we note that the PTBP2 isoform obtained upon exon 10 skipping is an NMD target, which might diminish the observable effect.  
+
+Taken together, the additional RNA- seq- based KD/rescue experiments we performed strongly suggest that (i) the observed effects are specific to C9ORF78 and that (ii) effects on at least some exon skipping events can only be reverted by over- expression of C9ORF78wt, but not by over- expression of C9ORF78R41A, and are thus most likely dependent on our observed BRR2- C9ORF78 interaction.  
+
+Finally, we would like to point out that a direct involvement of C9ORF78 in regulating alternative splicing events and a role of the observed BRR2- C9ORF78 interaction in this regulation is not
+
+<--- Page Split --->
+
+only suggested by our KD/rescue experiments, but also by our observation that putative additional interactions of C9ORF78 with other spliceosomal proteins change upon disrupting or weakening the C9ORF78- BRR2 interaction by the C9ORF78 R41A exchange (please also see the additional GO analysis we now provide in new Fig. 7a). These observations also suggest credible mechanisms by which C9ORF78 could exert such roles (please refer to our revised Discussion and responses to Reviewer 3).  
+
+We describe the new RNA- seq- based KD/rescue experiments and results in the revised manuscript (line 262):  
+
+We then transfected HEK293 cells with siRNA- resistant genes encoding either C9ORF78wt or C9ORF78R41A, and after two hours knocked down endogenous C9ORF78 via siRNAs for 72 hours. RNA- seq analysis confirmed KD of endogenous C9ORF78 and over- expression of the siRNA- resistant variants to a similar extent (Supplementary Fig. 9). rMATS analysis confirmed the global changes in alternative splicing upon C9ORF78 KD as seen in the first C9ORF78 KD experiment. Significantly changed alternative 3'- ss strongly overlapped between the two KD experiments, with almost all of the overlapping targets being NAGNAG sites (28 of 33; Fig. 5a). Strikingly, we find C9ORF78 KD- induced alternative 3'- ss skipping globally reverted upon both C9ORF78wt and C9ORF78R41A over- expression (Fig. 5b,c), strongly arguing for a C9ORF78- specific effect.  
+
+Skipped exon events overlapped to a lower extent between the two KD experiments (Fig. 5d) and we observed only a partial rescue of C9ORF78 KD- induced changes in exon skipping events via the siRNA- resistant variants (Fig. 5e). Nonetheless, 376 exon skipping events were significantly altered in both KD datasets (ApPercent spliced- in [PSI] \(>0.1\) ; \(p < 0.05\) ), 49 of which were significantly reverted only by over- production of C9ORF78wt (Fig. 5f), including skipping of PTBP2 exon 10 (Fig. 5g), indicating a regulatory mechanism that depends on the observed BRR2- C9ORF78 interaction. Together, these findings confirm that the observed alternative splicing changes upon C9ORF78 KD are indeed specific and suggest different mechanisms of splicing regulation, as C9ORF78- regulated alternative 3'- ss appear to be less dependent on the BRR2- C9ORF78 interaction than C9ORF78- regulated cassette exons.  
+
+The new results are now presented in new Supplementary Fig. 9 and new Fig. 5: New Supplementary Fig. 9:  
+
+![PLACEHOLDER_8_0]
+
+
+<--- Page Split --->
+
+New Fig. 5:  
+
+![PLACEHOLDER_9_0]
+  
+
+We also amended the Discussion section accordingly (line 381):  
+
+C9ORF78 KD elicited changes in many exon skipping events, a significant number of which are dependent on the BRR2- C9ORF78 interaction. Exon skipping is thought to be decided before the C complex stage, providing additional indirect evidence that C9ORF78 is already present at an earlier stage. We also observed a highly reproducible effect of C9ORF78 KD on alternative usage of NAGNAG 3'- ss, with C9ORF78 strongly favoring usage of the upstream 3'- ss, for which differential recue experiments indicated C9ORF78 specificity but failed to support a dependence on the observed BRR2- C9ORF78 interaction. While these data suggest that different C9ORF78- splicing factor interactions play a predominant role for the regulation of cassette exons and of alternative 3'- ss, assay limitations may also have prevented the detection of subtle effects of the BRR2- C9ORF78 interaction on some alternative splicing events. E.g., the single residue exchange in C9ORF78R41A is sufficient to destabilize the binary interaction with BRR2 in vitro, but C9ORF78 interaction with other spliceosomal factors and
+
+<--- Page Split --->
+
+over- expression of the siRNA- resistant C9ORF78 variants may have obscured differences in rescue efficiencies between C9ORF78wt and C9ORF78R41A.  
+
+## Minor points  
+
+1. Figure 2a C9ORF78 labelling – there are 2 alpha 1 and no alpha 3 present.  
+
+Sorry for this mistake, corrected.  
+
+2. "We exchanged C9ORF78 F8 and R41 individually for alanine residues, and tested BRR2 binding of the C9ORF78 variants via analytical SEC. While C9ORF78F8A showed reduced binding to BRR2 in analytical SEC, BRR2 binding by C9ORF78R41A in vitro was completely abolished" C9ORF78-F8A is not present in Figure 3b, perhaps the authors have forgotten to add the panel with WB analysis of a SEC experiment for the C9ORF78F8A mutant.  
+
+We have now included the C9ORF78F8A SEC experiment in new Fig. 3b:  
+
+![PLACEHOLDER_10_0]
+  
+
+3. Figure 4c. The first 2 panel appear to be only depicting statistically significant splicing events, while the NAGNAG panel appears to show all events. Perhaps the authors could consider showing only statistically significant NAGNAG events similarly to the previous 2 panels of the figure.  
+
+Thank you for the suggestion. As the vast majority of regulated 3'- ss events are NAGNAG events, we decided to highlight by color all NAGNAG events in the panel showing the regulated 3'- ss events (new Fig. 4c, right):
+
+<--- Page Split --->
+![PLACEHOLDER_11_0]
+  
+
+## 4. Figure 5a. Labelling of y axis. It would be beneficial to explain in the text or figure legend what PSI is and how it is calculated.  
+
+This is now new Fig. 4d. We have included an explanation of PSI and how it was calculated in the figure legends of new Fig. 4 (for gel- based experiments) and new Fig. 5 (for sequencing- based experiments):  
+
+PSI, percent spliced- in (gel analysis), ratio of the quantified band representing exon inclusion and the sum of the quantified bands representing exon inclusion and exon skipping.  
+
+PSI, percent spliced- in (RNA- seq analysis), ratio of the quantified junction reads representing exon inclusion and the sum of the quantified junction reads representing exon inclusion and exon skipping.  
+
+## 5. Figure 5b. It would be beneficial to explain in the text or figure legend how the fold change was calculated.  
+
+This experiment has been replaced by the new RNA- seq analyses upon endogenous C9ORF78 KD and rescue with siRNA- resistant C9ORF78wt or C9ORF78R41A (shown in new Fig. 5; please see above).  
+
+6. Figure 6a. The authors could look at their FLASH data for sites of crosslink between C9ORF78 and U5 snRNA. With FLASH I would expect that an RNA-protein crosslink site would manifest itself as a tight cluster of deletions or point mutations in cDNA sequences. If it is possible to elucidate the exact crosslink site, it could significantly contribute to the structural data and perhaps could allow further speculations on the location of the part of C9ORF78 that is not engaged in interaction with BRR2.  
+
+Again, thanks for the suggestion. We indeed observe gaps in the U5 snRNA sequencing reads that suggest the C9ORF78 cross- linking site. They map to U5 snRNA positions 69- 73, i.e., to internal loop 1 of U5 snRNA. This observation is now shown in new Fig. 6b:
+
+<--- Page Split --->
+![PLACEHOLDER_12_0]
+  
+
+Furthermore, we now describe in the revised text (line 297):  
+
+A cluster of gaps in the U5 snRNA sequencing reads suggested cross- links of C9ORF78 to U5 snRNA residues 69- 73, which form internal loop 1 (IL1) at the base of the extended U5' stem- loop (Fig. 6b).  
+
+We also highlighted the identified cross- link site in new Fig. 8a, in which we now present the locations of putative C9ORF78 interactors in the spliceosomal Bact complex (please also see our replies to comments by Reviewer 3):  
+
+![PLACEHOLDER_12_1]
+  
+
+7. Comments on the C9ORF78 interactome.
+
+<--- Page Split --->
+
+The authors show that spliceosomal proteins are enriched in in both C9ORF78- WT and C9ORF78- R41A. It would be good to see a GO term analysis, considering the manuscript reports that C9ORF78 IP results in enrichment over control for 2411 proteins. This is perhaps 1/5th of the proteome, expressed in HEK cells. In fact, the very limited GO analysis that I performed on the manuscripts data does reveal a strong enrichment for GO CC "spliceosome", which helps interpretation of the data, as it shows that the IP has been specific.  
+
+We thank the reviewer for this suggestion, which we have implemented. Just to clarify - there were 2,411 proteins identified and quantified in all experiments combined, but not all of them were enriched over the control. Many of these proteins represent background that was not removed completely from the beads by washing. This is very common in MS- based IP analyses because of the high sensitivity. The control experiment is therefore essential to separate this non- specific background from specifically enriched proteins. To extract the specifically enriched proteins, it is necessary to define a minimum level of enrichment. In label- free quantitation at least a 2- fold enrichment (over control) should be used as a threshold. By filtering the protein list for "log2 fold change \(> 1\) " and a significant t- test result \((q < 0.05)\) , we ended up with 560 enriched proteins in C9ORF78wt Flag- IP vs. the control, and 809 enriched proteins in C9ORF78R41A Flag- IP vs. the control. This is now clarified in the revised text (line 309):  
+
+Filtering for a two- fold enrichment over the control (log2- fold change \(> 1\) ) and a significant t- test result \((q < 0.05)\) yielded 560 enriched proteins in the C9ORF78wt Flag- IP and 809 enriched proteins in the C9ORF78R41A Flag- IP. We subjected proteins with a log2- fold enrichment \(> 3\) in either of the Flag- IPs to a GO analysis, which indicated "U5 snRNP" as the most enriched GO term (Fig. 7a).  
+
+As suggested, we have included a GO- term analysis in new Fig. 7a:  
+
+![PLACEHOLDER_13_0]
+  
+
+To that end, I would suggest that in addition to figure 6b the chapter "C9ORF78 interacts with additional spliceosomal proteins" could benefit from a figure that would depict the proteins, belonging to two spliceosomal complexes B and C. Those proteins should be colour coded to depict 2 key parameters, discussed in the chapter: whether a given protein was detected as enriched in the IP and if so, whether its enrichment is higher or lower in C9ORF78- WT compared to C9ORF78- R41A. I am of the opinion that this would be a great visual help for the chapter. Perhaps the authors could consider including a figure like that in the manuscript.  
+
+Again, thanks for the suggestion. We now provide a new Fig. 7b that includes all spliceosomal proteins from all complexes/stages for which our experiments show any interaction with C9ORF78 (wt or R41A; these include spliceosomal B and C complex proteins as suggested).
+
+<--- Page Split --->
+
+As a reference list of proteins, we took proteins observed in cryoEM structures of the respective spliceosomal complexes/stages, as listed in Kastner et al. (2019) Cold Spring Harb Perspect Biol 11, a032417 (PMID: 30765414). The figure also lists all proteins of the various complexes/stages that were not enriched, as suggested. Enrichments via C9ORF78wt and C9ORF78R41A are shown by dark and light orange bars, respectively. We have experimented with more complex color coding, but found it to be confusing. For each complex/stage we now show enriched proteins in the following order: 
+
+1. Proteins enriched more by C9ORF78wt than by C9ORF78R41A (ordered from most to least enriched by C9ORF78wt) 
+
+2. Proteins enriched more by C9ORF78R41A than by C9ORF78wt (ordered from most to least enriched by C9ORF78R41A) 
+
+3. Proteins only enriched by C9ORF78wt (ordered from most to least enriched) 
+
+4. Proteins only enriched by C9ORF78R41A (ordered from most to least enriched) 
+
+New Fig. 7b: 
+
+![PLACEHOLDER_14_0]
+
+ 
+
+## Reviewer #2 
+
+The manuscript by Bergfort et al reports an unstructured protein C9ORF78 tightly interacts with the key spliceosomal RNA helicase BRR2 through a series of evidence, including yeast two-hybrid screen, in vitro protein-protein interaction, affinity purification and mass spectrometry, and cryoEM structures. They find that C9ORF78 and another spliceosomal protein FBP21 interact with the C-terminal cassette of BRR2 in a mutually exclusive manner using both structural information and biochemical competition assay. RNAi of C9ORF78 leads to alternative splicing changes including a
+
+<--- Page Split --->
+
+substantial usage of alternative 3'SSs. This manuscript provides insightful information in understanding the function of a flexibly or dynamically bound spliceosomal protein during the process of spliceosome assembly and catalysis. In general, their findings are convincing and interesting, and the manuscript is well written.  
+
+We thank the reviewer for the very positive general evaluation, in particular for considering our findings convincing and interesting and the manuscript to be well- written.  
+
+Below are several questions and concerns:  
+
+1. Typo: In Figures 1-3, several places of C9ORF78 are shown in "C9ORF8"; In Figures 2a and 3a, there are two α1s labelled for the structure of C9ORF78, of which one should be "α3"; In Figure 7, consistent with its legends, the forest green component marked as "PRP22" should be "PRPF22".  
+
+Sorry for these mistakes and thank you for catching them, all corrected.  
+
+2. C9ORF78 was observed in the C complex and the binding of C9ORF78 with BRR2 was also described in S. pombe. In Figures 1a & 1b and later, overexpression of GST- or Flag-tagged C9ORF78 are used for SEC and IP experiments, indicating that the interaction between C9ORF78 and BRR2 might be not strong in HEK293 cells. Could this be done using an endogenous normally expressed C9ORF78, either through an C9ORF78-specific antibody or by a CRSIPR-Cas9 knock-in tag system?  
+
+While we agree with the reviewer that pull- down experiments with over- expressed, tagged proteins do not appropriately reflect the strength of an interaction in cells, we also report SEC analyses in various configurations. Due to the extended times that complexes need to persist while being separated, SEC represents a rather stringent test for stable complexes. We also provide evidence via competitive interaction tests that C9ORF78 binds BRR2 stronger than the known BRR2 interactor, FBP21. Together with previous observations listed by the reviewer, we think that these findings corroborate our suggestion based on pull- down experiments that the interaction can also ensue in cells. We have slightly modified our corresponding statement in the revised text (line 123):  
+
+We also observed co- immunoprecipitation (co- IP) of BRR2 via Flag- C9ORF78 in HEK293 cells (Fig. 1b).  
+
+3. In Figure 4a, this should be RT-qPCR, not qPCR, detection of mRNAs, better to present this with an additional agarose gel.  
+
+We corrected the description. We report results from RT- qPCR, as this assay is more sensitive than agarose gel quantification. However, we have analyzed RT- qPCR products via agarose gels, which we are presenting below to document that we obtain single bands for C9ORF78 and GAPDH, respectively, and that the expression level of C9ORF78 is reduced upon siRNA knock- down.
+
+<--- Page Split --->
+![PLACEHOLDER_16_0]
+  
+
+4. In Figure 4b, SE (skipped exon) is the dominant feature of AS events when the C9ORF78 is KD. I am curious, what is the feature of those exons? For those increased inclusion of exons (up-regulated), do they have weaker 3'SSs; vice versa, for those decreased inclusion of exons (down-regulated), do they have stronger 3'SSs? Therefore, I would like to suggest strength analyses (scores) of both the 3'SSs and 5'SSs of those SE events.  
+
+We thank the reviewer for this suggestion and now report this information in the revised text (line 228):  
+
+Further analysis showed that C9ORF78 KD- induced exon skipping is associated with short exons, while exons included upon C9ORF78 KD exhibited an increased average length. Additionally, exons whose inclusion changed upon C9ORF78 KD showed weaker 5'- ss but average- strength 3'- ss, independent of the direction of regulation (Supplementary Fig. 6).
+
+<--- Page Split --->
+
+The analysis is now also presented in new Supplementary Fig. 6:  
+
+![PLACEHOLDER_17_0]
+  
+
+5. In Figure 4c, the negative value of deltaPSI are not presented, this is confusing of which sample vs which sample. In addition, RMATS should be rMATS.  
+
+Sorry for this, we have corrected both. Here the corrected Fig. 4c:  
+
+![PLACEHOLDER_17_1]
+
+
+<--- Page Split --->
+
+6. The primes in "3' or 5'- splice site" are incorrect, should be 3' or 5'.  
+
+Sorry again, corrected throughout.  
+
+## Reviewer #3  
+
+In this manuscript the authors present a rather comprehensive analysis of the interactions of the human splicing factor C9ORF78 with the spliceosome and particularly with the helicase Brr2, revealing a previously unappreciated role for this factor during catalysis in modulating 3'SS selection. The manuscript suggests a molecular mechanism that can explain how Brr2 may act during the catalytic stage. I generally find the authors' data compelling, of timely interest to the field, and potentially more broadly, while the proposed mechanistic models are mostly supported by the presented experiments. However, there are a few points that the authors should try to address prior to publication.  
+
+We thank the reviewer for this very positive overall evaluation, specifically for considering our data to be compelling and interesting.  
+
+While the cryo- EM analysis appears sound and expertly performed, the authors do not provide any clear figure showing their fit of the C9ORF78 model into their determined EM map. Given the claimed high resolution, it is critical for the authors to show this data as a figure in the paper, especially for the critical parts where C9ORF78 interacts with Brr2. The mutational data in Fig. 3 does support the proposed modelling, but still it is impossible to fully judge the quality of the map and the model fit without this data, especially as the presented local resolution in Sup.Fig. 2e shows significant variation in local resolution along the proposed C9ORF78 path. Certainly the map presented in Fig. 1c is contoured at an RMSD that makes it hard to judge the presence of high resolution features.  
+
+We thank the reviewer for pointing this out and now provide an additional Supplementary Figure (new Supplementary Fig. 4) documenting the quality of the cryoEM reconstructions:
+
+<--- Page Split --->
+![PLACEHOLDER_19_0]
+  
+
+Although the in vitro assays show a modest effect of C9ORF78 on Brr2 helicase activity for U4/U6, I think it is premature to argue that C9ORF78 does not act through modulation of Brr2 helicase activity, as the authors do on p.8. The assays used as not the native situation in the spliceosome and C9ORF78 binds at the U2/U6 stage of splicing, while Brr2 has been implicated in spliceosome disassembly. Thus, one could easily imagine that C9ORF78 may act at the disassembly stage and that its role in 3'SS selection could be coupled to a role in modulating Brr2 activity during disassembly. I am aware that this role for Brr2 is a matter of contention in the field but I think the authors should be more cautious with their statements here, as their data cannot exclude a role for C9ORF78 in modulating Brr2 during this later stage.  
+
+We thank the reviewer for the insightful comments. We agree and have adjusted the revised manuscript accordingly. However, we restricted the description to pointing out the possibility that C9ORF78- mediated regulation of BRR2 helicase activity may play a role, but in the absence of clear evidence refrained from speculating about a possible stage/events for which this may be important.  
+
+We changed the headline of the corresponding Results section to:  
+
+## C9ORF78 moderately down-regulates BRR2 helicase activity  
+
+We left out the statement from the Results that the findings argue "against a modulation of BRR2 helicase activity constituting a major C9ORF78 function".  
+
+We included a short section on this aspect in the revised Discussion (line 345):  
+
+The BRR2- modulatory activity of C9ORF78 we report here is weaker, and C9ORF78 seems to be associated with the spliceosome only at stages when BRR2 has already unwound U4/U6.
+
+<--- Page Split --->
+
+However, we presently cannot exclude that C9ORF78- dependent regulation of BRR2 helicase activity may play a role during other stages of splicing.  
+
+The observations regarding competition between FBP21 and C9ORF78 for Brr2 binding are strongly supported by the data in Fig. 2d. Nonetheless, it is unclear why the authors chose to use Brr2HR complexes rather than complexes that also contained the Jab1 domain of Prp8, given that the Jab1 domain remains bound to Brr2 from the B complex onwards and that the FBP21 interaction is observed in the B complex structure in the context of Brr2 being bound to Jab1. I think the authors could and likely should strengthen their argument here by performing the competition experiment with Brr2HR/Jab1/C9ORF78 complexes and even better do it in the proper physiological progression by asking whether C9ORF78 can compete off FBP21, which is what they propose actually happens during the splicing pathway.  
+
+We thank the reviewer for this suggestion and have performed the suggested experiments. We have pre- incubated BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup> with FBP21116- 376, which gives rise to a stable BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex as revealed by SEC (new Fig. 2d). Upon adding GST- C9ORF78 to the pre- formed BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex, GST- C9ORF78 displaces FBP21116- 376, as again revealed by SEC (new Fig. 2d). We used a longer FBP21 fragment for these experiments than employed for cryoEM, because PRPF8<sup>Jab1</sup> and FBP21200- 376 (the fragment used in cryoEM) cannot be distinguished on SDS- PAGE gels. We now describe these results in the revised manuscript (line 207) and display them in a new Fig. 2d:  
+
+To test mutually exclusive binding biochemically, we performed SEC analyses with a preformed BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex to which we added GST- C9ORF78 (Fig. 2d). We used a slightly longer FBP21 fragment in these experiments than in cryoEM studies to allow distinction from PRPF8<sup>Jab1</sup> in SDS- PAGE. While FBP21116- 376 formed a stable complex with BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup> (Fig. 2d, top), GST- C9ORF78 displaced FBP21116- 376 from the BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex (Fig. 2d, bottom), indicating mutually exclusive binding and a stronger affinity of GST- C9ORF78 for BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>. These findings suggest that C9ORF78 may aid in the displacement of FBP21 upon conversion of the B to the B<sup>act</sup> complex, when FBP21 and other B- specific proteins are released.  
+
+![PLACEHOLDER_20_0]
+  
+
+Finally, while I appreciate the authors' model for C9ORF78 function during the \(C^*\) stage as a compelling main mechanism of action, I think the authors need to be much more careful with their discussion of various proteins they claim are present at the C complexes stage, such as DDX23 or Prp6. These have generally only been detected as such in C complexes prepared by biochemical stalls that are not entirely clean and could have contaminating earlier and later complexes. A good example are studies that use a 3'SS mutation to capture C complexes and which are now known from cryo- EM studies to actually reach the \(C^*\) and even P complex stage and then revert back to the C complex stage. Simply citing proteomic analyses of such preparations as evidence of stage- specific association is sufficient evidence for such claims. None of the single particle EM studies of C or \(C^*\) complexes have identified any subpopulations containing
+
+<--- Page Split --->
+
+DDX23 for example, yet the authors routinely seem to suggest this protein could bind at the this stage. The claim that their IP of DDX23 occurs at the C complex stage is particularly problematic in this sense, especially as one could imagine much more easily how complexes that have transitioned to the B complex stage may not have fully lost DDX23 until the Bact stage, making the exchange of FBP21 for C9ORF78 an alternative point at which a transient interaction between C9ORF78 and DDX23 could have occurred. Similarly, the claim that Prp22 binds at the C complex stage should be revised, as what people have reported as binding at that stage most likely results from 3'SS mutant complexes that have reached the \(C^{*}\) stage and then reverted back to the C complex stage without Prp22 dissociation, as was shown in a recent study on equilibrium of spliceosome conformations during catalysis in yeast. I urge the authors to be much more rigorous with their terminology when describing potential binding to various complexes.  
+
+Again, we very much appreciate these knowledgeable comments. Taking them into careful account, we have now much more judiciously discussed our findings in the revised manuscript. For the detailed revision of the text, please refer to our reply below the next point. The only aspect we want to bring up here is that we did not want to insinuate at all that DDX23 is normally present at the C complex stage, sorry if our descriptions were confusing. Instead, we only wanted to suggest an explanation for our observation that DDX23 was more enriched in our C9ORF78<sup>R41A</sup> Flag- IP compared to the C9ORF78<sup>wt</sup> Flag- IP, which we interpret to reflect the "liberation" of a DDX23- binding site on BRR2 in case of the C9ORF78<sup>R41A</sup> variant, but not in case of C9ORF78<sup>wt</sup>. However, as the reviewer rightly points out that C9ORF78 and DDX23 most likely do not occur in the same or directly neighboring spliceosomal intermediates, and as our descriptions were obviously confusing, we now completely left out the descriptions/discussions referring to DDX23.  
+
+Related to this matter, I find that the authors overlook too quickly a potential role for C9ORF78 in regulating splice site use also during the B to Bact transition. They observe significant numbers of exon skipping and mutually exclusive exon use changes in their KD experiments. It is much harder to imagine how such events could be regulated at the C or \(C^{*}\) stage, but much easier to imagine how transfer of the 5'SS and docking of the BP helix at the active site, which occur during the B to Bact transition, or are influenced by the relative stability of these complexes, could impact these types of alternative splicing events. Indeed, the crosslinks to U5 snRNA are consistent with such a role and much more likely to reflect interactions at the Bact stage than at the C complex stage, when U5 is buried very deeply into the active site, making an interaction with a flexible part of C9ORF78 harder to imagine. Brr2- dependent association with Cwc27 also supports this idea, as in the Bact structure the flexible C9ORF78 residues proposed to interact with Prp22 in \(C^{*}\) , could easily be imagined to interact with Cwc27 in Bact. Since C9ORF78 likely regulates the Bact transition, as the authors argue with strong experimental support, they should at least discuss the possibility of an earlier role in regulating alternative splicing at this stage through some of the other factors they observe in their IP studies.  
+
+Again, many thanks for these comments, we agree that in our original discussion we took the nominal association of C9ORF78 with the C complex stage too much at face value. We have amended and thoroughly reworked the corresponding Discussion sections, taking all remarks and suggestions by the reviewer into account (lines 355 and 381):  
+
+The mutually exclusive binding of FBP21 and C9ORF78 to BRR2 suggests that C9ORF78 might first bind to the spliceosome during the B- to- B<sup>act</sup> transition, when FBP21 is released. While proteomics analyses have suggested that C9ORF78 might be associated with the C complex<sup>27</sup>, the analyzed complexes had been enriched on a modified pre- mRNA lacking a 3'- ss AG dinucleotide and a 3'- exon<sup>55</sup>. CryoEM and biochemical studies have shown that
+
+<--- Page Split --->
+
+spliceosomes assembled on such modified pre- mRNAs can progress to the \(\mathbb{C}^*\) complex state<sup>40</sup>, that neighboring states tend to converge on the C complex state when exon ligation is inhibited<sup>56</sup> and that under appropriate conditions both catalytic steps of splicing can be reversed<sup>57</sup>. Thus, factors identified via proteomics in nominal C complex preparations may to some extent represent contaminations from neighboring states. Presence of C9ORF78 already during the \(\mathsf{B}^{\mathsf{act}}\) stage is further supported by putative interactions we observe with the \(\mathsf{B}^{\mathsf{act}}\) proteins CWC22 and CWC27. However, as the C9ORF78- binding site of BRR2 remains unobstructed in C, \(\mathbb{C}^*\) and P complexes<sup>34,40,41,58,59</sup>, and as we also identified putative C9ORF78 interactions with 1<sup>st</sup> step, \(\mathbb{C}^*\) and 2<sup>nd</sup> step factors, C9ORF78 may also remain bound after the B- to- \(\mathsf{B}^{\mathsf{act}}\) transition.  
+
+C9ORF78 KD elicited changes in many exon skipping events, a significant number of which are dependent on the BRR2- C9ORF78 interaction. Exon skipping is thought to be decided before the C complex stage, providing additional indirect evidence that C9ORF78 is already present at an earlier stage. We also observed a highly reproducible effect of C9ORF78 KD on alternative usage of NAGNAG 3'- ss, with C9ORF78 strongly favoring usage of the upstream 3'- ss, for which differential recue experiments indicated C9ORF78 specificity but failed to support a dependence on the observed BRR2- C9ORF78 interaction. While these data suggest that different C9ORF78- splicing factor interactions play a predominant role for the regulation of cassette exons and of alternative 3'- ss, assay limitations may also have prevented the detection of subtle effects of the BRR2- C9ORF78 interaction on some alternative splicing events. E.g., the single residue exchange in C9ORF78<sup>41A</sup> is sufficient to destabilize the binary interaction with BRR2 in vitro, but C9ORF78 interaction with other spliceosomal factors and over- expression of the siRNA- resistant C9ORF78 variants may have obscured differences in rescue efficiencies between C9ORF78<sup>wt</sup> and C9ORF78<sup>41A</sup>.  
+
+Exon skipping might be influenced by the kinetics with which two mutually exclusive splicing scenarios transition from the B via the \(\mathsf{B}^{\mathsf{act}}\) to the \(\mathsf{B}^*\) stage, and our findings suggest that C9ORF78 could modulate these transitions. Recently, additional assembly intermediates between the B and \(\mathsf{B}^{\mathsf{act}}\) stages have been characterized biochemically and structurally<sup>60</sup>. These pre- \(\mathsf{B}^{\mathsf{act}}\) complexes contain, among others, reduced levels of the B- specific FBP21 protein, but also largely lack \(\mathsf{B}^{\mathsf{act}}\) proteins CWC22 and CWC27 and the step 1 factor GPKOW. Given our observations that C9ORF78 can displace FBP21 from BRR2 and could also contact CWC22, CWC27 and GPKOW, presence of C9ORF78 might modulate the kinetics of B- to- \(\mathsf{B}^{\mathsf{act}}\) conversion by driving displacement of FBP21 and helping recruitment of \(\mathsf{B}^{\mathsf{act}}\) proteins and GPKOW. Notably, the multi- step B- to- \(\mathsf{B}^{\mathsf{act}}\) transition is also accompanied by a stepwise repositioning of BRR2<sup>60</sup>, which might likewise be influenced by C9ORF78 that putatively links BRR2 to other components according to our data. Moreover, a large- scale cryoEM analysis has revealed that the human \(\mathsf{B}^{\mathsf{act}}\) complex can adopt at least eight major conformations, which could be arranged along a trajectory towards catalytic activation due to the degree of their similarity to a later intermediate<sup>61</sup>. Such a situation most likely also applies to other splicing stages, and it has been suggested that any additional incoming factor will alter the conformational space available to the respective spliceosomal intermediate<sup>61</sup>. Based on a structural superposition, it is easily conceivable that in \(\mathsf{B}^{\mathsf{act}}\) the intrinsically unstructured C9ORF78 could bridge between BRR2, CWC22/CWC27 and U5 IL1 (Fig. 8a) This presumed cross- strutting of several \(\mathsf{B}^{\mathsf{act}}\) elements would most likely significantly alter the conformational space available to \(\mathsf{B}^{\mathsf{act}}\). C9ORF78 might thereby again alter the kinetics of B- to- \(\mathsf{B}^{\mathsf{act}}\) conversion and/or influence the tendency of the \(\mathsf{B}^{\mathsf{act}}\) complex to adopt a conformation conducive to PRPF2 remodeling.  
+
+Alternative NAGNAG splice site choice was suggested to take place during the second step of the splicing reaction<sup>48</sup>, indirectly supporting our notion of a continued presence of C9ORF78 at post- \(\mathsf{B}^{\mathsf{act}}\) stages. Furthermore, our observed interaction of C9ORF78 with the second step factor, PRPF22, suggests that C9ORF78 remains associated also with the \(\mathbb{C}^*\) complex. Comparison of our BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- C9ORF78 structure with the structure of a human \(\mathbb{C}^*\) complex revealed that the C- terminal 231 residues of C9ORF78 could easily reach and directly contact PRPF22, which resides in immediate vicinity of BRR2 in the \(\mathbb{C}^*\) complex, as well as CWC22 that is still present at the \(\mathbb{C}^*\) stage (Fig. 8b). PRPF22 has been shown to be involved
+
+<--- Page Split --->
+
+in 3'- ss selection and exon ligation in yeast<sup>62,63</sup>. Thus, one possible mechanism for C9ORF78 to regulate alternative 3'- ss usage may be direct C9ORF78-PRPF22 interactions that affect PRPF22 motor activity, which is thought to reposition alternative 3'- ss in the spliceosome's active site from a distance<sup>4,63</sup>. This interpretation is consistent with our observation that presence of C9ORF78 leads to preferential use of upstream alternative 3'- ss.  
+
+New Fig. 8:  
+
+![PLACEHOLDER_23_0]
+
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+First of all, I wanted to say that the authors have done a remarkable job addressing the few points I raised in my original review. I believe the new data obtained by the authors has improved an already good manuscript. As with every good data, it raises more questions than it answers, which is to be expected. Furthermore the data is well presented and adequately discussed.  
+
+In addition to revealing the relatively small pool of transcripts, sensitive to the integrity of BRR2- C9ORF78 interaction, the RNA- seq results clearly hint at an exciting possibility that C9ORF78 may influence splicing outcomes via interactions with spliceosomal proteins other than BRR2.  
+
+I believe, that the manuscript in its current version contains a wealth of essential data which would be invaluable for scientists attempting to further investigate the role of C9ORF78 in splicing and beyond. I am of the opinion that the manuscript meets the expected publication standards and I raise no further issues with it.  
+
+Reviewer #2: Remarks to the Author: No further questions.  
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+In this revised manuscript the authors provide a compelling and timely analysis of the role of C9ORF78 in modulating splice site usage by the human spliceosome, through direct interactions with the core spliceosomal ATPase Brr2. Altogether, the structural, biochemical, and sequencing data provide strong support for the idea that C9ORF78 is recruited to the spliceosome during the pre- catalytic phase, may modulate spliceosome activation, and likely remains bound during the catalytic stage, when it influences selection of the 3'SS.  
+
+I think the revised manuscript has a better flow and the authors have presented a better and more nuanced discussion of their data.  
+
+Overall, I am satisfied that my previous concerns, as well as the potential concerns of the other reviewers, have been addressed in the revised manuscript, including by several new experiments, as requested.  
+
+I think the manuscript elucidates several functions for C9ORF78 and provides a significant advance in understanding the indirect roles played by Brr2 during later stages of splicing.  
+
+Thus I strongly recommend publication in its present form.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1eb88b8e99ba8ca78ed896f518cb1347831354bf2216abcb3b0f8e3c72eb76f4/supplementary_0_Peer Review File__1eb88b8e99ba8ca78ed896f518cb1347831354bf2216abcb3b0f8e3c72eb76f4_det.mmd

@@ -0,0 +1,551 @@
+<|ref|>title<|/ref|><|det|>[[100, 40, 508, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[106, 110, 373, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[106, 154, 823, 242]]<|/det|>
+A multi- factor trafficking site on spliceosome remodeling enzyme, BRR2, recruits C9ORF78 to regulate alternative splicing  
+
+<|ref|>image<|/ref|><|det|>[[95, 732, 261, 780]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[271, 732, 880, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 103]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[116, 120, 217, 133]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[116, 135, 291, 148]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 149, 880, 298]]<|/det|>
+The manuscript by Bergfort et al., employs methods of structural and molecular biology to study the role of poorly characterised C9ORF78 protein in splicing. C9ORF78 has been previously reported as a constituent part of spliceosome whose function remained largely unknown. The manuscript reports a cryoEM structure of C9ORF78 bound to its interaction partner BRR2. The second cryoEM structure, reported in the manuscript is that of BRR2 bound to an inhibitor FBP21. These structures together with pulldown experiments allowed authors to concluded that C9ORF78 and FBP21 binding to BRR2 is mutually exclusive. Armed with that knowledge, the authors then went on to investigate the effect of C9ORF78 knockdown on mRNA splicing. siRNA knockdown of C9ORF78 had an effect on 3' splice site selection which induced exon skipping events, or altered exon length if the two 3' splice sites were in close proximity.  
+
+<|ref|>text<|/ref|><|det|>[[115, 298, 872, 403]]<|/det|>
+This structural knowledge allowed the authors to pinpoint C9ORF78 residues, critical for interaction with BRR2. The fact that these residues are conserved over hundreds millions of years of evolution and the corresponding point mutant C9ORF78- R41A fails to bind to BRR2 is a highlight of the paper. The place of C9ORF78 in splicing is further cemented by UV crosslinking experiments that show a link between C9ORF78 and U5 snRNA. These experiments however did not reveal significantly enriched mRNAs, suggesting that C9ORF78 does not affect splicing via direct binding to mRNAs, and rather exerts it's influence on the 3' splice site selection via its spliceosome interaction partners.  
+
+<|ref|>text<|/ref|><|det|>[[116, 417, 868, 478]]<|/det|>
+This body of work is novel and represents a significant interest to the field of splicing and broader readership. The structures presented in the paper are of particular value to the field as C9ORF78 has not been observed in the published spliceosomal structures. However I have 1 major and several minor comments that I hope would help the authors improve their manuscript further.  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 494, 202, 507]]<|/det|>
+## Major point  
+
+<|ref|>text<|/ref|><|det|>[[115, 507, 870, 803]]<|/det|>
+The manuscript presents detailed structural description of how C9ORF78 interacts with BRR2 and PRPF8 proteins. However, experimental support for functional significance of these interactions is perhaps the weakest part of otherwise solid story the authors portray. Indeed, this support becomes vital, as the authors report that more than a couple of thousand proteins are enriched over background in C9ORF78 IP. This would suggest that in addition to a direct interaction with BRR2, C9ORF78 could influence splicing indirectly via it's non- spliceosomal interaction partners. The functional validation experiment, shown in figure 5b has produced mixed results. C9ORF78- WT and C9ORF78- R41A rescue si RNA mediated C9ORF78 knockdown to identical extents (figure 5b SMARCA4 and C1ORF131 panels). For the experiment in panel PTBP2, the authors compared C9ORF78- WT knockdown rescue to that of C9ORF78- R41A and reported a difference between the two. Firstly, that comparison may not be the right one as C9ORF78- WT "rescues" exon skipping to the levels better than in si- Control. Why does it do that is an unsolved question, perhaps C9ORF78 is limiting and it's overexpression produces better splicing efficiency in si- C9ORF78 cells than in unperturbed cells that do not express si resistant C9ORF78. Therefore, as far as splicing efficiency is concerned, a comparison between PTBP2 exon 10 retention in si- Control against a C9ORF78 knockdown rescued by C9ORF78- R41A is the one to make in this case. It would appear as if this comparison would not produce a statistically significant difference between the two. If that is correct, then we would have to conclude that splicing of PTBP2 exon 10 is indistinguishable in si- Control vs si- C9ORF78 rescued with C9ORF78- R41A, hence C9ORF78 interaction with BRR2 is dispensable for splicing of that particular exon.  
+
+<|ref|>text<|/ref|><|det|>[[116, 818, 881, 878]]<|/det|>
+Even if the authors disagree with this logic, I would draw their attention to absolute fold change values between si- Control and C9ORF78- R41A they report in panel PTBP2 of figure 5. The reported difference in fold change is miniscule. How would that translate into protein level and what effect would it have on the function of PTBP2? From figure 5 I concluded that the effect would negligible at best.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 870, 150]]<|/det|>
+Therefore I would strongly urge the authors to find a clear cut example for functional significance of C9ORF78- BRR2 interaction. A good starting point would be to expand the experimental approach and analysis reported in figure 4 onto figure 5 and perform the necessary RNA- seq experiments with the samples of Figure 5.  
+
+<|ref|>text<|/ref|><|det|>[[115, 179, 213, 193]]<|/det|>
+Minor points:  
+
+<|ref|>text<|/ref|><|det|>[[115, 208, 689, 223]]<|/det|>
+1. Figure 2a C9ORF78 labelling - there are 2 alpha 1 and no alpha 3 present.  
+
+<|ref|>text<|/ref|><|det|>[[115, 238, 869, 281]]<|/det|>
+2. "We exchanged C9ORF78 F8 and R41 individually for alanine residues, and tested BRR2 binding of the C9ORF78 variants via analytical SEC. While C9ORF78F8A showed reduced binding to BRR2 in analytical SEC, BRR2 binding by C9ORF78R41A in vitro was completely abolished"  
+
+<|ref|>text<|/ref|><|det|>[[115, 297, 860, 326]]<|/det|>
+C9ORF78- F8A is not present in Figure 3b, perhaps the authors have forgotten to add the panel with WB analysis of a SEC experiment for the C9ORF78F8A mutant.  
+
+<|ref|>text<|/ref|><|det|>[[115, 342, 875, 386]]<|/det|>
+3. Figure 4c. The first 2 panel appear to be only depicting statistically significant splicing events, while the NAGNAG panel appears to show all events. Perhaps the authors could consider showing only statistically significant NAGNAG events similarly to the previous 2 panels of the figure.  
+
+<|ref|>text<|/ref|><|det|>[[115, 401, 870, 431]]<|/det|>
+4. Figure 5a. Labelling of y axis. It would be beneficial to explain in the text or figure legend what PSI is and how it is calculated.  
+
+<|ref|>text<|/ref|><|det|>[[115, 446, 850, 475]]<|/det|>
+5. Figure 5b. It would be beneficial to explain in the text or figure legend how the fold change was calculated.  
+
+<|ref|>text<|/ref|><|det|>[[115, 491, 870, 564]]<|/det|>
+6. Figure 6a. The authors could look at their FLASH data for sites of crosslink between C9ORF78 and U5 snRNA. With FLASH I would expect that an RNA-protein crosslink site would manifest itself as a tight cluster of deletions or point mutations in cDNA sequences. If it is possible to elucidate the exact crosslink site, it could significantly contribute to the structural data and perhaps could allow further speculations on the location of the part of C9ORF78 that is not engaged in interaction with BRR2.  
+
+<|ref|>text<|/ref|><|det|>[[117, 580, 440, 594]]<|/det|>
+7. Comments on the C9ORF78 interactome.  
+
+<|ref|>text<|/ref|><|det|>[[115, 610, 876, 700]]<|/det|>
+The authors show that spliceosomal proteins are enriched in in both C9ORF78- WT and C9ORF78- R41A. It would be good to see a GO term analysis, considering the manuscript reports that C9ORF78 IP results in enrichment over control for 2411 proteins. This is perhaps 1/5th of the proteome, expressed in HEK cells. In fact, the very limited GO analysis that I performed on the manuscripts data does reveal a strong enrichment for GO CC "spliceosome", which helps interpretation of the data, as it shows that the IP has been specific.  
+
+<|ref|>text<|/ref|><|det|>[[115, 715, 875, 819]]<|/det|>
+To that end, I would suggest that in addition to figure 6b the chapter "C9ORF78 interacts with additional spliceosomal proteins" could benefit from a figure that would depict the proteins, belonging to two spliceosomal complexes B and C. Those proteins should be colour coded to depict 2 key parameters, discussed in the chapter: whether a given protein was detected as enriched in the IP and if so, whether its enrichment is higher or lower in C9ORF78- WT compared to C9ORF78- R41A. I am of the opinion that this would be a great visual help for the chapter. Perhaps the authors could consider including a figure like that in the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 864, 291, 893]]<|/det|>
+Reviewer #2: Remarks to the Author:
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 880, 223]]<|/det|>
+The manuscript by Bergfort et al reports an unstructured protein C9ORF78 tightly interacts with the key spliceosomal RNA helicase BRR2 through a series of evidence, including yeast two- hybrid screen, in vitro protein- protein interaction, affinity purification and mass spectrometry, and cryoEM structures. They find that C9ORF78 and another spliceosomal protein FBP21 interact with the C- terminal cassette of BRR2 in a mutually exclusive manner using both structural information and biochemical competition assay. RNAi of C9ORF78 leads to alternative splicing changes including a substantial usage of alternative 3'SSs. This manuscript provides insightful information in understanding the function of a flexibly or dynamically bound spliceosomal protein during the process of spliceosome assembly and catalysis. In general, their findings are convincing and interesting, and the manuscript is well written.  
+
+<|ref|>text<|/ref|><|det|>[[117, 238, 432, 252]]<|/det|>
+Below are several questions and concerns:  
+
+<|ref|>text<|/ref|><|det|>[[115, 253, 848, 297]]<|/det|>
+1. Typo: In Figures 1-3, several places of C9ORF78 are shown in "C9ORF8"; In Figures 2a and 3a, there are two a1s labelled for the structure of C9ORF78, of which one should be "a3"; In Figure 7, consistent with its legends, the forest green component marked as "PRP22" should be "PRPF22".  
+
+<|ref|>text<|/ref|><|det|>[[115, 312, 880, 388]]<|/det|>
+2. C9ORF78 was observed in the C complex and the binding of C9ORF78 with BRR2 was also described in S. pombe. In Figures 1a & 1b and later, overexpression of GST- or Flag-tagged C9ORF78 are used for SEC and IP experiments, indicating that the interaction between C9ORF78 and BRR2 might be not strong in HEK293 cells. Could this be done using an endogenous normally expressed C9ORF78, either through an C9ORF78-specific antibody or by a CRSIPR-Cas9 knock-in tag system?  
+
+<|ref|>text<|/ref|><|det|>[[115, 402, 872, 432]]<|/det|>
+3. In Figure 4a, this should be RT-qPCR, not qPCR, detection of mRNAs, better to present this with an additional agarose gel.  
+
+<|ref|>text<|/ref|><|det|>[[115, 447, 877, 521]]<|/det|>
+4. In Figure 4b, SE (skipped exon) is the dominant feature of AS events when the C9ORF78 is KD. I am curious, what is the feature of those exons? For those increased inclusion of exons (up-regulated), do they have weaker 3'SSs; vice versa, for those decreased inclusion of exons (down-regulated), do they have stronger 3'SSs? Therefore, I would like to suggest strength analyses (scores) of both the 3'SSs and 5'SSs of those SE events.  
+
+<|ref|>text<|/ref|><|det|>[[115, 536, 872, 566]]<|/det|>
+5. In Figure 4c, the negative value of deltaPSI are not presented, this is confusing of which sample vs which sample. In addition, RMATS should be rMATS.  
+
+<|ref|>text<|/ref|><|det|>[[115, 580, 630, 595]]<|/det|>
+6. The primes in "3' or 5'-splice site" are incorrect, should be 3' or 5'.  
+
+<|ref|>text<|/ref|><|det|>[[115, 640, 216, 654]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[115, 657, 291, 670]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 671, 877, 745]]<|/det|>
+In this manuscript the authors present a rather comprehensive analysis of the interactions of the human splicing factor C9ORF78 with the spliceosome and particularly with the helicase Brr2, revealing a previously unappreciated role for this factor during catalysis in modulating 3'SS selection. The manuscript suggests a molecular mechanism that can explain how Brr2 may act during the catalytic stage.  
+
+<|ref|>text<|/ref|><|det|>[[115, 760, 870, 805]]<|/det|>
+I generally find the authors' data compelling, of timely interest to the field, and potentially more broadly, while the proposed mechanistic models are mostly supported by the presented experiments. However, there are a few points that the authors should try to address prior to publication.  
+
+<|ref|>text<|/ref|><|det|>[[115, 819, 881, 894]]<|/det|>
+While the cryo- EM analysis appears sound and expertly performed, the authors do not provide any clear figure showing their fit of the C9ORF78 model into their determined EM map. Given the claimed high resolution, it is critical for the authors to show this data as a figure in the paper, especially for the critical parts where C9ORF78 interacts with Brr2. The mutational data in Fig. 3 does support the proposed modelling, but still it is impossible to fully judge the quality of the map and the model fit
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 880, 135]]<|/det|>
+without this data, especially as the presented local resolution in Sup.Fig. 2e shows significant variation in local resolution along the proposed C9ORF78 path. Certainly the map presented in Fig. 1c is contoured at an RMSD that makes it hard to judge the presence of high resolution features.  
+
+<|ref|>text<|/ref|><|det|>[[115, 149, 881, 283]]<|/det|>
+Although the in vitro assays show a modest effect of C9ORF78 on Brr2 helicase activity for U4/U6, I think it is premature to argue that C9ORF78 does not act through modulation of Brr2 helicase activity, as the authors do on p.8. The assays used as not the native situation in the spliceosome and C9ORF78 binds at the U2/U6 stage of splicing, while Brr2 has been implicated in spliceosome disassembly. Thus, one could easily imagine that C9ORF78 may act at the disassembly stage and that its role in 3'SS selection could be coupled to a role in modulating Brr2 activity during disassembly. I am aware that this role for Brr2 is a matter of contention in the field but I think the authors should be more cautious with their statements here, as their data cannot exclude a role for C9ORF78 in modulating Brr2 during this later stage.  
+
+<|ref|>text<|/ref|><|det|>[[115, 297, 881, 432]]<|/det|>
+The observations regarding competition between FBP21 and C9ORF78 for Brr2 binding are strongly supported by the data in Fig. 2d. Nonetheless, it is unclear why the authors chose to use Brr2HR complexes rather than complexes that also contained the Jab1 domain of Prp8, given that the Jab1 domain remains bound to Brr2 from the B complex onwards and that the FBP21 interaction is observed in the B complex structure in the context of Brr2 being bound to Jab1. I think the authors could and likely should strengthen their argument here by performing the competition experiment with Brr2HR/Jab1/C9ORF78 complexes and even better do it in the proper physiological progression by asking whether C9ORF78 can compete off FBP21, which is what they propose actually happens during the splicing pathway.  
+
+<|ref|>text<|/ref|><|det|>[[115, 446, 880, 746]]<|/det|>
+Finally, while I appreciate the authors' model for C9ORF78 function during the C\* stage as a compelling main mechanism of action, I think the authors need to be much more careful with their discussion of various proteins they claim are present at the C complexes stage, such as DDX23 or Prp6. These have generally only been detected as such in C complexes prepared by biochemical stalls that are not entirely clean and could have contaminating earlier and later complexes. A good example are studies that use a 3'SS mutation to capture C complexes and which are now known from cryo- EM studies to actually reach the C\* and even P complex stage and then revert back to the C complex stage. Simply citing proteomic analyses of such preparations as evidence of stage- specific association is sufficient evidence for such claims. None of the single particle EM studies of C or C\* complexes have identified any subpopulations containing DDX23 for example, yet the authors routinely seem to suggest this protein could bind at the this stage. The claim that their IP of DDX23 occurs at the C complex stage is particularly problematic in this sense, especially as one could imagine much more easily how complexes that have transitioned to the B complex stage may not have fully lost DDX23 until the Bact stage, making the exchange of FBP21 for C9ORF78 an alternative point at which a transient interaction between C9ORF78 and DDX23 could have occurred. Similarly, the claim that Prp22 binds at the C complex stage should be revised, as what people have reported as binding at that stage most likely results from 3'SS mutant complexes that have reached the C\* stage and then reverted back to the C complex stage without Prp22 dissociation, as was shown in a recent study on equilibrium of spliceosome conformations during catalysis in yeast. I urge the authors to be much more rigorous with their terminology when describing potential binding to various complexes.  
+
+<|ref|>text<|/ref|><|det|>[[115, 760, 880, 895]]<|/det|>
+Related to this matter, I find that the authors overlook too quickly a potential role for C9ORF78 in regulating splice site use also during the B to Bact transition. They observe significant numbers of exon skipping and mutually exclusive exon use changes in their KD experiments. It is much harder to imagine how such events could be regulated at the C or C\* stage, but much easier to imagine how transfer of the 5'SS and docking of the BP helix at the active site, which occur during the B to Bact transition, or are influenced by the relative stability of these complexes, could impact these types of alternative splicing events. Indeed, the crosslinks to U5 snRNA are consistent with such a role and much more likely to reflect interactions at the Bact stage than at the C complex stage, when U5 is buried very deeply into the active site, making an interaction with a flexible part of C9ORF78 harder to
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 877, 165]]<|/det|>
+imagine. Brr2- dependent association with Cwc27 also supports this idea, as in the Bact structure the flexible C9ORF78 residues proposed to interact with Prp22 in C\*, could easily be imagined to interact with Cwc27 in Bact. Since C9ORF78 likely regulates the Bact transition, as the authors argue with strong experimental support, they should at least discuss the possibility of an earlier role in regulating alternative splicing at this stage through some of the other factors they observe in their IP studies.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[117, 83, 605, 108]]<|/det|>
+## Response to Reviewer Comments  
+
+<|ref|>text<|/ref|><|det|>[[117, 121, 880, 154]]<|/det|>
+Reviewer comments are repeated in bold italics, responses are in regular font, changed text passages are highlighted in yellow.  
+
+<|ref|>sub_title<|/ref|><|det|>[[117, 166, 262, 186]]<|/det|>
+## Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[117, 200, 882, 368]]<|/det|>
+The manuscript by Bergfort et al., employs methods of structural and molecular biology to study the role of poorly characterised C9ORF78 protein in splicing. C9ORF78 has been previously reported as a constituent part of spliceosome whose function remained largely unknown. The manuscript reports a cryoEM structure of C9ORF78 bound to its interaction partner BRR2. The second cryoEM structure, reported in the manuscript is that of BRR2 bound to an inhibitor FBP21. These structures together with pulldown experiments allowed authors to concluded that C9ORF78 and FBP21 binding to BRR2 is mutually exclusive. Armed with that knowledge, the authors then went on to investigate the effect of C9ORF78 knockdown on mRNA splicing. siRNA knockdown of C9ORF78 had an effect on 3' splice site selection which induced exon skipping events, or altered exon length if the two 3' splice sites were in close proximity.  
+
+<|ref|>text<|/ref|><|det|>[[117, 368, 882, 504]]<|/det|>
+This structural knowledge allowed the authors to pinpoint C9ORF78 residues, critical for interaction with BRR2. The fact that these residues are conserved over hundreds millions of years of evolution and the corresponding point mutant C9ORF78- R41A fails to bind to BRR2 is a highlight of the paper. The place of C9ORF78 in splicing is further cemented by UV crosslinking experiments that show a link between C9ORF78 and U5 snRNA. These experiments however did not reveal significantly enriched mRNAs, suggesting that C9ORF78 does not affect splicing via direct binding to mRNAs, and rather exerts it's influence on the 3' splice site selection via its spliceosome interaction partners.  
+
+<|ref|>text<|/ref|><|det|>[[117, 502, 881, 577]]<|/det|>
+This body of work is novel and represents a significant interest to the field of splicing and broader readership. The structures presented in the paper are of particular value to the field as C9ORF78 has not been observed in the published spliceosomal structures. However I have 1 major and several minor comments that I hope would help the authors improve their manuscript further.  
+
+<|ref|>text<|/ref|><|det|>[[117, 592, 881, 637]]<|/det|>
+We thank the reviewer for this very positive overall assessment, specifically for considering our work to be of significant interest to the field and beyond, and our structures to be of particular value.  
+
+<|ref|>sub_title<|/ref|><|det|>[[117, 668, 220, 683]]<|/det|>
+## Major point  
+
+<|ref|>text<|/ref|><|det|>[[117, 682, 881, 787]]<|/det|>
+The manuscript presents detailed structural description of how C9ORF78 interacts with BRR2 and PRPF8 proteins. However, experimental support for functional significance of these interactions is perhaps the weakest part of otherwise solid story the authors portray. Indeed, this support becomes vital, as the authors report that more than a couple of thousand proteins are enriched over background in C9ORF78 IP. This would suggest that in addition to a direct interaction with BRR2, C9ORF78 could influence splicing indirectly via it's non- spliceosomal interaction partners.  
+
+<|ref|>text<|/ref|><|det|>[[117, 787, 882, 923]]<|/det|>
+The functional validation experiment, shown in figure 5b has produced mixed results. C9ORF78- WT and C9ORF78- R41A rescue si RNA mediated C9ORF78 knockdown to identical extents (figure 5b SMARCA4 and C1ORF131 panels). For the experiment in panel PTBP2, the authors compared C9ORF78- WT knockdown rescue to that of C9ORF78- R41A and reported a difference between the two. Firstly, that comparison may not be the right one as C9ORF78- WT "rescues" exon skipping to the levels better than in si- Control. Why does it do that is an unsolved question, perhaps C9ORF78 is limiting and it's overexpression produces better splicing efficiency in si- C9ORF78 cells than in unperturbed cells that do not express si resistant
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 82, 883, 255]]<|/det|>
+C9ORF78. Therefore, as far as splicing efficiency is concerned, a comparison between PTBP2 exon 10 retention in si- Control against a C9ORF78 knockdown rescued by C9ORF78- R41A is the one to make in this case. It would appear as if this comparison would not produce a statistically significant difference between the two. If that is correct, then we would have to conclude that splicing of PTBP2 exon 10 is indistinguishable in si- Control vs si- C9ORF78 rescued with C9ORF78- R41A, hence C9ORF78 interaction with BRR2 is dispensable for splicing of that particular exon. Even if the authors disagree with this logic, I would draw their attention to absolute fold change values between si- Control and C9ORF78- R41A they report in panel PTBP2 of figure 5. The reported difference in fold change is miniscule. How would that translate into protein level and what effect would it have on the function of PTBP2? From figure 5 I concluded that the effect would negligible at best.  
+
+<|ref|>text<|/ref|><|det|>[[117, 262, 881, 325]]<|/det|>
+Therefore I would strongly urge the authors to find a clear cut example for functional significance of C9ORF78- BRR2 interaction. A good starting point would be to expand the experimental approach and analysis reported in figure 4 onto figure 5 and perform the necessary RNA- seq experiments with the samples of Figure 5.  
+
+<|ref|>text<|/ref|><|det|>[[115, 337, 881, 444]]<|/det|>
+While we generally agree with the reviewer's assessment, we would like to point out that due to the single point mutation we introduced into the C9ORF78R41A variant (which abrogates stable interaction with BRR2 in vitro in the absence of other factors but may be partly "overridden" by additional interactions of C9ORF78 with other components of spliceosomes, as suggested by our IP experiments) and unavoidable overexpression of the siRNA- resistant C9ORF78wt or C9ORF78R41A variants for rescue experiments, full differential rescue effects by the C9ORF78wt or C9ORF78R41A variants are not to be expected.  
+
+<|ref|>text<|/ref|><|det|>[[116, 443, 881, 503]]<|/det|>
+We thank the reviewer for the suggestion of conducting KD/rescue experiments in combination with RNA- seq, and took it up. In the course of these additional experiments, we had to conduct a second, independent KD experiment, which allowed us to further validate our observations from the first KD experiment.  
+
+<|ref|>text<|/ref|><|det|>[[115, 503, 881, 743]]<|/det|>
+Importantly, the second, independent KD experiment confirmed the findings of the first KD experiment: (i) increased skipping of upstream \(3^{\prime}\) - ss in alternative \(3^{\prime}\) - ss pairs with a strong overlap in the affected \(3^{\prime}\) - ss pairs; (ii) a significant number of affected exon skipping events, albeit with a lower overlap between the two KD events. Consistent with our previous experiments, over- expression of siRNA- resistant variants of both C9ORF78wt and C9ORF78R41A globally reverted C9ORF78 KD- dependent changes in alternative \(3^{\prime}\) - ss usage, unequivocally confirming that the observed effects are specific to C9ORF78, but failing to reveal a direct role of the observed BRR2- C9ORF78 interaction in these events. However, 49 exon skipping events, which were reproducibly altered upon C9ORF78 KD in the two independent KD experiments, were significantly rescued only by over- expression of C9ORF78wt, but not by over- expression of C9ORF78R41A. PTBP2 exon 10 skipping was among the exon skipping events reproducibly affected in both independent KD experiments and was only significantly rescued by over- expression of C9ORF78wt. Furthermore, in the new, global RNA- seq based assessment of KD/rescue, C9ORF78wt over- expression did not restore PTBP2 exon 10 skipping levels "beyond" the si control situation. We now also statistically compare si control and rescue by C9ORF78R41A, as suggested (new Fig. 5g).  
+
+<|ref|>text<|/ref|><|det|>[[116, 743, 881, 820]]<|/det|>
+Quantifying the effect of C9ORF78 KD on PTBP2 exon 10 skipping based on the new KD/rescue analysis, revealed that C9ORF78 KD led to a decrease in the level of PTBP2 exon 10 inclusion from \(\sim 62\%\) spliced- in for si control to \(\sim 50\%\) spliced- in for si C9ORF78 (new Fig. 5g). While we agree that this effect is moderate, we note that the PTBP2 isoform obtained upon exon 10 skipping is an NMD target, which might diminish the observable effect.  
+
+<|ref|>text<|/ref|><|det|>[[116, 820, 881, 894]]<|/det|>
+Taken together, the additional RNA- seq- based KD/rescue experiments we performed strongly suggest that (i) the observed effects are specific to C9ORF78 and that (ii) effects on at least some exon skipping events can only be reverted by over- expression of C9ORF78wt, but not by over- expression of C9ORF78R41A, and are thus most likely dependent on our observed BRR2- C9ORF78 interaction.  
+
+<|ref|>text<|/ref|><|det|>[[115, 894, 881, 925]]<|/det|>
+Finally, we would like to point out that a direct involvement of C9ORF78 in regulating alternative splicing events and a role of the observed BRR2- C9ORF78 interaction in this regulation is not
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 82, 881, 174]]<|/det|>
+only suggested by our KD/rescue experiments, but also by our observation that putative additional interactions of C9ORF78 with other spliceosomal proteins change upon disrupting or weakening the C9ORF78- BRR2 interaction by the C9ORF78 R41A exchange (please also see the additional GO analysis we now provide in new Fig. 7a). These observations also suggest credible mechanisms by which C9ORF78 could exert such roles (please refer to our revised Discussion and responses to Reviewer 3).  
+
+<|ref|>text<|/ref|><|det|>[[115, 173, 880, 205]]<|/det|>
+We describe the new RNA- seq- based KD/rescue experiments and results in the revised manuscript (line 262):  
+
+<|ref|>text<|/ref|><|det|>[[115, 217, 881, 368]]<|/det|>
+We then transfected HEK293 cells with siRNA- resistant genes encoding either C9ORF78wt or C9ORF78R41A, and after two hours knocked down endogenous C9ORF78 via siRNAs for 72 hours. RNA- seq analysis confirmed KD of endogenous C9ORF78 and over- expression of the siRNA- resistant variants to a similar extent (Supplementary Fig. 9). rMATS analysis confirmed the global changes in alternative splicing upon C9ORF78 KD as seen in the first C9ORF78 KD experiment. Significantly changed alternative 3'- ss strongly overlapped between the two KD experiments, with almost all of the overlapping targets being NAGNAG sites (28 of 33; Fig. 5a). Strikingly, we find C9ORF78 KD- induced alternative 3'- ss skipping globally reverted upon both C9ORF78wt and C9ORF78R41A over- expression (Fig. 5b,c), strongly arguing for a C9ORF78- specific effect.  
+
+<|ref|>text<|/ref|><|det|>[[115, 368, 881, 519]]<|/det|>
+Skipped exon events overlapped to a lower extent between the two KD experiments (Fig. 5d) and we observed only a partial rescue of C9ORF78 KD- induced changes in exon skipping events via the siRNA- resistant variants (Fig. 5e). Nonetheless, 376 exon skipping events were significantly altered in both KD datasets (ApPercent spliced- in [PSI] \(>0.1\) ; \(p < 0.05\) ), 49 of which were significantly reverted only by over- production of C9ORF78wt (Fig. 5f), including skipping of PTBP2 exon 10 (Fig. 5g), indicating a regulatory mechanism that depends on the observed BRR2- C9ORF78 interaction. Together, these findings confirm that the observed alternative splicing changes upon C9ORF78 KD are indeed specific and suggest different mechanisms of splicing regulation, as C9ORF78- regulated alternative 3'- ss appear to be less dependent on the BRR2- C9ORF78 interaction than C9ORF78- regulated cassette exons.  
+
+<|ref|>text<|/ref|><|det|>[[115, 532, 779, 563]]<|/det|>
+The new results are now presented in new Supplementary Fig. 9 and new Fig. 5: New Supplementary Fig. 9:  
+
+<|ref|>image<|/ref|><|det|>[[336, 575, 657, 828]]<|/det|>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 83, 213, 98]]<|/det|>
+New Fig. 5:  
+
+<|ref|>image<|/ref|><|det|>[[113, 110, 884, 690]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[115, 712, 642, 730]]<|/det|>
+We also amended the Discussion section accordingly (line 381):  
+
+<|ref|>text<|/ref|><|det|>[[115, 741, 881, 923]]<|/det|>
+C9ORF78 KD elicited changes in many exon skipping events, a significant number of which are dependent on the BRR2- C9ORF78 interaction. Exon skipping is thought to be decided before the C complex stage, providing additional indirect evidence that C9ORF78 is already present at an earlier stage. We also observed a highly reproducible effect of C9ORF78 KD on alternative usage of NAGNAG 3'- ss, with C9ORF78 strongly favoring usage of the upstream 3'- ss, for which differential recue experiments indicated C9ORF78 specificity but failed to support a dependence on the observed BRR2- C9ORF78 interaction. While these data suggest that different C9ORF78- splicing factor interactions play a predominant role for the regulation of cassette exons and of alternative 3'- ss, assay limitations may also have prevented the detection of subtle effects of the BRR2- C9ORF78 interaction on some alternative splicing events. E.g., the single residue exchange in C9ORF78R41A is sufficient to destabilize the binary interaction with BRR2 in vitro, but C9ORF78 interaction with other spliceosomal factors and
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 82, 880, 113]]<|/det|>
+over- expression of the siRNA- resistant C9ORF78 variants may have obscured differences in rescue efficiencies between C9ORF78wt and C9ORF78R41A.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 143, 230, 157]]<|/det|>
+## Minor points  
+
+<|ref|>text<|/ref|><|det|>[[118, 157, 784, 174]]<|/det|>
+1. Figure 2a C9ORF78 labelling – there are 2 alpha 1 and no alpha 3 present.  
+
+<|ref|>text<|/ref|><|det|>[[118, 187, 383, 202]]<|/det|>
+Sorry for this mistake, corrected.  
+
+<|ref|>text<|/ref|><|det|>[[116, 231, 883, 327]]<|/det|>
+2. "We exchanged C9ORF78 F8 and R41 individually for alanine residues, and tested BRR2 binding of the C9ORF78 variants via analytical SEC. While C9ORF78F8A showed reduced binding to BRR2 in analytical SEC, BRR2 binding by C9ORF78R41A in vitro was completely abolished" C9ORF78-F8A is not present in Figure 3b, perhaps the authors have forgotten to add the panel with WB analysis of a SEC experiment for the C9ORF78F8A mutant.  
+
+<|ref|>text<|/ref|><|det|>[[115, 337, 715, 354]]<|/det|>
+We have now included the C9ORF78F8A SEC experiment in new Fig. 3b:  
+
+<|ref|>image<|/ref|><|det|>[[113, 366, 884, 725]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[116, 748, 883, 811]]<|/det|>
+3. Figure 4c. The first 2 panel appear to be only depicting statistically significant splicing events, while the NAGNAG panel appears to show all events. Perhaps the authors could consider showing only statistically significant NAGNAG events similarly to the previous 2 panels of the figure.  
+
+<|ref|>text<|/ref|><|det|>[[116, 824, 881, 870]]<|/det|>
+Thank you for the suggestion. As the vast majority of regulated 3'- ss events are NAGNAG events, we decided to highlight by color all NAGNAG events in the panel showing the regulated 3'- ss events (new Fig. 4c, right):
+
+<--- Page Split --->
+<|ref|>image<|/ref|><|det|>[[202, 80, 792, 299]]<|/det|>  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 326, 880, 358]]<|/det|>
+## 4. Figure 5a. Labelling of y axis. It would be beneficial to explain in the text or figure legend what PSI is and how it is calculated.  
+
+<|ref|>text<|/ref|><|det|>[[115, 370, 881, 417]]<|/det|>
+This is now new Fig. 4d. We have included an explanation of PSI and how it was calculated in the figure legends of new Fig. 4 (for gel- based experiments) and new Fig. 5 (for sequencing- based experiments):  
+
+<|ref|>text<|/ref|><|det|>[[115, 430, 880, 462]]<|/det|>
+PSI, percent spliced- in (gel analysis), ratio of the quantified band representing exon inclusion and the sum of the quantified bands representing exon inclusion and exon skipping.  
+
+<|ref|>text<|/ref|><|det|>[[115, 475, 880, 522]]<|/det|>
+PSI, percent spliced- in (RNA- seq analysis), ratio of the quantified junction reads representing exon inclusion and the sum of the quantified junction reads representing exon inclusion and exon skipping.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 550, 880, 582]]<|/det|>
+## 5. Figure 5b. It would be beneficial to explain in the text or figure legend how the fold change was calculated.  
+
+<|ref|>text<|/ref|><|det|>[[115, 596, 881, 643]]<|/det|>
+This experiment has been replaced by the new RNA- seq analyses upon endogenous C9ORF78 KD and rescue with siRNA- resistant C9ORF78wt or C9ORF78R41A (shown in new Fig. 5; please see above).  
+
+<|ref|>text<|/ref|><|det|>[[115, 671, 881, 762]]<|/det|>
+6. Figure 6a. The authors could look at their FLASH data for sites of crosslink between C9ORF78 and U5 snRNA. With FLASH I would expect that an RNA-protein crosslink site would manifest itself as a tight cluster of deletions or point mutations in cDNA sequences. If it is possible to elucidate the exact crosslink site, it could significantly contribute to the structural data and perhaps could allow further speculations on the location of the part of C9ORF78 that is not engaged in interaction with BRR2.  
+
+<|ref|>text<|/ref|><|det|>[[115, 776, 881, 822]]<|/det|>
+Again, thanks for the suggestion. We indeed observe gaps in the U5 snRNA sequencing reads that suggest the C9ORF78 cross- linking site. They map to U5 snRNA positions 69- 73, i.e., to internal loop 1 of U5 snRNA. This observation is now shown in new Fig. 6b:
+
+<--- Page Split --->
+<|ref|>image<|/ref|><|det|>[[115, 81, 884, 340]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[116, 352, 606, 369]]<|/det|>
+Furthermore, we now describe in the revised text (line 297):  
+
+<|ref|>text<|/ref|><|det|>[[116, 381, 880, 429]]<|/det|>
+A cluster of gaps in the U5 snRNA sequencing reads suggested cross- links of C9ORF78 to U5 snRNA residues 69- 73, which form internal loop 1 (IL1) at the base of the extended U5' stem- loop (Fig. 6b).  
+
+<|ref|>text<|/ref|><|det|>[[116, 441, 880, 489]]<|/det|>
+We also highlighted the identified cross- link site in new Fig. 8a, in which we now present the locations of putative C9ORF78 interactors in the spliceosomal Bact complex (please also see our replies to comments by Reviewer 3):  
+
+<|ref|>image<|/ref|><|det|>[[225, 500, 763, 856]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[117, 887, 500, 903]]<|/det|>
+7. Comments on the C9ORF78 interactome.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 81, 882, 189]]<|/det|>
+The authors show that spliceosomal proteins are enriched in in both C9ORF78- WT and C9ORF78- R41A. It would be good to see a GO term analysis, considering the manuscript reports that C9ORF78 IP results in enrichment over control for 2411 proteins. This is perhaps 1/5th of the proteome, expressed in HEK cells. In fact, the very limited GO analysis that I performed on the manuscripts data does reveal a strong enrichment for GO CC "spliceosome", which helps interpretation of the data, as it shows that the IP has been specific.  
+
+<|ref|>text<|/ref|><|det|>[[116, 203, 881, 385]]<|/det|>
+We thank the reviewer for this suggestion, which we have implemented. Just to clarify - there were 2,411 proteins identified and quantified in all experiments combined, but not all of them were enriched over the control. Many of these proteins represent background that was not removed completely from the beads by washing. This is very common in MS- based IP analyses because of the high sensitivity. The control experiment is therefore essential to separate this non- specific background from specifically enriched proteins. To extract the specifically enriched proteins, it is necessary to define a minimum level of enrichment. In label- free quantitation at least a 2- fold enrichment (over control) should be used as a threshold. By filtering the protein list for "log2 fold change \(> 1\) " and a significant t- test result \((q < 0.05)\) , we ended up with 560 enriched proteins in C9ORF78wt Flag- IP vs. the control, and 809 enriched proteins in C9ORF78R41A Flag- IP vs. the control. This is now clarified in the revised text (line 309):  
+
+<|ref|>text<|/ref|><|det|>[[116, 397, 881, 476]]<|/det|>
+Filtering for a two- fold enrichment over the control (log2- fold change \(> 1\) ) and a significant t- test result \((q < 0.05)\) yielded 560 enriched proteins in the C9ORF78wt Flag- IP and 809 enriched proteins in the C9ORF78R41A Flag- IP. We subjected proteins with a log2- fold enrichment \(> 3\) in either of the Flag- IPs to a GO analysis, which indicated "U5 snRNP" as the most enriched GO term (Fig. 7a).  
+
+<|ref|>text<|/ref|><|det|>[[116, 489, 675, 505]]<|/det|>
+As suggested, we have included a GO- term analysis in new Fig. 7a:  
+
+<|ref|>image<|/ref|><|det|>[[301, 515, 690, 715]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[116, 742, 882, 866]]<|/det|>
+To that end, I would suggest that in addition to figure 6b the chapter "C9ORF78 interacts with additional spliceosomal proteins" could benefit from a figure that would depict the proteins, belonging to two spliceosomal complexes B and C. Those proteins should be colour coded to depict 2 key parameters, discussed in the chapter: whether a given protein was detected as enriched in the IP and if so, whether its enrichment is higher or lower in C9ORF78- WT compared to C9ORF78- R41A. I am of the opinion that this would be a great visual help for the chapter. Perhaps the authors could consider including a figure like that in the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[116, 879, 880, 925]]<|/det|>
+Again, thanks for the suggestion. We now provide a new Fig. 7b that includes all spliceosomal proteins from all complexes/stages for which our experiments show any interaction with C9ORF78 (wt or R41A; these include spliceosomal B and C complex proteins as suggested).
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 82, 881, 188]]<|/det|>
+As a reference list of proteins, we took proteins observed in cryoEM structures of the respective spliceosomal complexes/stages, as listed in Kastner et al. (2019) Cold Spring Harb Perspect Biol 11, a032417 (PMID: 30765414). The figure also lists all proteins of the various complexes/stages that were not enriched, as suggested. Enrichments via C9ORF78wt and C9ORF78R41A are shown by dark and light orange bars, respectively. We have experimented with more complex color coding, but found it to be confusing. For each complex/stage we now show enriched proteins in the following order: 
+
+<|ref|>text<|/ref|><|det|>[[115, 188, 881, 217]]<|/det|>
+1. Proteins enriched more by C9ORF78wt than by C9ORF78R41A (ordered from most to least enriched by C9ORF78wt) 
+
+<|ref|>text<|/ref|><|det|>[[115, 218, 881, 247]]<|/det|>
+2. Proteins enriched more by C9ORF78R41A than by C9ORF78wt (ordered from most to least enriched by C9ORF78R41A) 
+
+<|ref|>text<|/ref|><|det|>[[115, 248, 761, 263]]<|/det|>
+3. Proteins only enriched by C9ORF78wt (ordered from most to least enriched) 
+
+<|ref|>text<|/ref|><|det|>[[115, 263, 777, 278]]<|/det|>
+4. Proteins only enriched by C9ORF78R41A (ordered from most to least enriched) 
+
+<|ref|>text<|/ref|><|det|>[[115, 293, 224, 308]]<|/det|>
+New Fig. 7b: 
+
+<|ref|>image<|/ref|><|det|>[[115, 321, 881, 757]]<|/det|>
+ 
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 788, 256, 807]]<|/det|>
+## Reviewer #2 
+
+<|ref|>text<|/ref|><|det|>[[115, 822, 881, 928]]<|/det|>
+The manuscript by Bergfort et al reports an unstructured protein C9ORF78 tightly interacts with the key spliceosomal RNA helicase BRR2 through a series of evidence, including yeast two-hybrid screen, in vitro protein-protein interaction, affinity purification and mass spectrometry, and cryoEM structures. They find that C9ORF78 and another spliceosomal protein FBP21 interact with the C-terminal cassette of BRR2 in a mutually exclusive manner using both structural information and biochemical competition assay. RNAi of C9ORF78 leads to alternative splicing changes including a
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 82, 881, 144]]<|/det|>
+substantial usage of alternative 3'SSs. This manuscript provides insightful information in understanding the function of a flexibly or dynamically bound spliceosomal protein during the process of spliceosome assembly and catalysis. In general, their findings are convincing and interesting, and the manuscript is well written.  
+
+<|ref|>text<|/ref|><|det|>[[117, 157, 880, 188]]<|/det|>
+We thank the reviewer for the very positive general evaluation, in particular for considering our findings convincing and interesting and the manuscript to be well- written.  
+
+<|ref|>text<|/ref|><|det|>[[117, 218, 501, 233]]<|/det|>
+Below are several questions and concerns:  
+
+<|ref|>text<|/ref|><|det|>[[117, 234, 881, 294]]<|/det|>
+1. Typo: In Figures 1-3, several places of C9ORF78 are shown in "C9ORF8"; In Figures 2a and 3a, there are two α1s labelled for the structure of C9ORF78, of which one should be "α3"; In Figure 7, consistent with its legends, the forest green component marked as "PRP22" should be "PRPF22".  
+
+<|ref|>text<|/ref|><|det|>[[117, 308, 700, 324]]<|/det|>
+Sorry for these mistakes and thank you for catching them, all corrected.  
+
+<|ref|>text<|/ref|><|det|>[[117, 352, 881, 444]]<|/det|>
+2. C9ORF78 was observed in the C complex and the binding of C9ORF78 with BRR2 was also described in S. pombe. In Figures 1a & 1b and later, overexpression of GST- or Flag-tagged C9ORF78 are used for SEC and IP experiments, indicating that the interaction between C9ORF78 and BRR2 might be not strong in HEK293 cells. Could this be done using an endogenous normally expressed C9ORF78, either through an C9ORF78-specific antibody or by a CRSIPR-Cas9 knock-in tag system?  
+
+<|ref|>text<|/ref|><|det|>[[116, 458, 881, 593]]<|/det|>
+While we agree with the reviewer that pull- down experiments with over- expressed, tagged proteins do not appropriately reflect the strength of an interaction in cells, we also report SEC analyses in various configurations. Due to the extended times that complexes need to persist while being separated, SEC represents a rather stringent test for stable complexes. We also provide evidence via competitive interaction tests that C9ORF78 binds BRR2 stronger than the known BRR2 interactor, FBP21. Together with previous observations listed by the reviewer, we think that these findings corroborate our suggestion based on pull- down experiments that the interaction can also ensue in cells. We have slightly modified our corresponding statement in the revised text (line 123):  
+
+<|ref|>text<|/ref|><|det|>[[116, 608, 880, 639]]<|/det|>
+We also observed co- immunoprecipitation (co- IP) of BRR2 via Flag- C9ORF78 in HEK293 cells (Fig. 1b).  
+
+<|ref|>text<|/ref|><|det|>[[116, 668, 880, 699]]<|/det|>
+3. In Figure 4a, this should be RT-qPCR, not qPCR, detection of mRNAs, better to present this with an additional agarose gel.  
+
+<|ref|>text<|/ref|><|det|>[[116, 713, 880, 788]]<|/det|>
+We corrected the description. We report results from RT- qPCR, as this assay is more sensitive than agarose gel quantification. However, we have analyzed RT- qPCR products via agarose gels, which we are presenting below to document that we obtain single bands for C9ORF78 and GAPDH, respectively, and that the expression level of C9ORF78 is reduced upon siRNA knock- down.
+
+<--- Page Split --->
+<|ref|>image<|/ref|><|det|>[[234, 95, 707, 375]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[116, 403, 883, 496]]<|/det|>
+4. In Figure 4b, SE (skipped exon) is the dominant feature of AS events when the C9ORF78 is KD. I am curious, what is the feature of those exons? For those increased inclusion of exons (up-regulated), do they have weaker 3'SSs; vice versa, for those decreased inclusion of exons (down-regulated), do they have stronger 3'SSs? Therefore, I would like to suggest strength analyses (scores) of both the 3'SSs and 5'SSs of those SE events.  
+
+<|ref|>text<|/ref|><|det|>[[116, 509, 880, 541]]<|/det|>
+We thank the reviewer for this suggestion and now report this information in the revised text (line 228):  
+
+<|ref|>text<|/ref|><|det|>[[117, 554, 881, 617]]<|/det|>
+Further analysis showed that C9ORF78 KD- induced exon skipping is associated with short exons, while exons included upon C9ORF78 KD exhibited an increased average length. Additionally, exons whose inclusion changed upon C9ORF78 KD showed weaker 5'- ss but average- strength 3'- ss, independent of the direction of regulation (Supplementary Fig. 6).
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 82, 652, 99]]<|/det|>
+The analysis is now also presented in new Supplementary Fig. 6:  
+
+<|ref|>image<|/ref|><|det|>[[113, 110, 884, 545]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[115, 572, 884, 604]]<|/det|>
+5. In Figure 4c, the negative value of deltaPSI are not presented, this is confusing of which sample vs which sample. In addition, RMATS should be rMATS.  
+
+<|ref|>text<|/ref|><|det|>[[115, 617, 656, 634]]<|/det|>
+Sorry for this, we have corrected both. Here the corrected Fig. 4c:  
+
+<|ref|>image<|/ref|><|det|>[[202, 645, 792, 863]]<|/det|>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 82, 723, 99]]<|/det|>
+6. The primes in "3' or 5'- splice site" are incorrect, should be 3' or 5'.  
+
+<|ref|>text<|/ref|><|det|>[[118, 113, 397, 129]]<|/det|>
+Sorry again, corrected throughout.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 157, 254, 177]]<|/det|>
+## Reviewer #3  
+
+<|ref|>text<|/ref|><|det|>[[115, 191, 883, 328]]<|/det|>
+In this manuscript the authors present a rather comprehensive analysis of the interactions of the human splicing factor C9ORF78 with the spliceosome and particularly with the helicase Brr2, revealing a previously unappreciated role for this factor during catalysis in modulating 3'SS selection. The manuscript suggests a molecular mechanism that can explain how Brr2 may act during the catalytic stage. I generally find the authors' data compelling, of timely interest to the field, and potentially more broadly, while the proposed mechanistic models are mostly supported by the presented experiments. However, there are a few points that the authors should try to address prior to publication.  
+
+<|ref|>text<|/ref|><|det|>[[115, 341, 880, 374]]<|/det|>
+We thank the reviewer for this very positive overall evaluation, specifically for considering our data to be compelling and interesting.  
+
+<|ref|>text<|/ref|><|det|>[[115, 400, 882, 555]]<|/det|>
+While the cryo- EM analysis appears sound and expertly performed, the authors do not provide any clear figure showing their fit of the C9ORF78 model into their determined EM map. Given the claimed high resolution, it is critical for the authors to show this data as a figure in the paper, especially for the critical parts where C9ORF78 interacts with Brr2. The mutational data in Fig. 3 does support the proposed modelling, but still it is impossible to fully judge the quality of the map and the model fit without this data, especially as the presented local resolution in Sup.Fig. 2e shows significant variation in local resolution along the proposed C9ORF78 path. Certainly the map presented in Fig. 1c is contoured at an RMSD that makes it hard to judge the presence of high resolution features.  
+
+<|ref|>text<|/ref|><|det|>[[115, 568, 880, 600]]<|/det|>
+We thank the reviewer for pointing this out and now provide an additional Supplementary Figure (new Supplementary Fig. 4) documenting the quality of the cryoEM reconstructions:
+
+<--- Page Split --->
+<|ref|>image<|/ref|><|det|>[[115, 81, 880, 480]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 883, 660]]<|/det|>
+Although the in vitro assays show a modest effect of C9ORF78 on Brr2 helicase activity for U4/U6, I think it is premature to argue that C9ORF78 does not act through modulation of Brr2 helicase activity, as the authors do on p.8. The assays used as not the native situation in the spliceosome and C9ORF78 binds at the U2/U6 stage of splicing, while Brr2 has been implicated in spliceosome disassembly. Thus, one could easily imagine that C9ORF78 may act at the disassembly stage and that its role in 3'SS selection could be coupled to a role in modulating Brr2 activity during disassembly. I am aware that this role for Brr2 is a matter of contention in the field but I think the authors should be more cautious with their statements here, as their data cannot exclude a role for C9ORF78 in modulating Brr2 during this later stage.  
+
+<|ref|>text<|/ref|><|det|>[[116, 672, 881, 750]]<|/det|>
+We thank the reviewer for the insightful comments. We agree and have adjusted the revised manuscript accordingly. However, we restricted the description to pointing out the possibility that C9ORF78- mediated regulation of BRR2 helicase activity may play a role, but in the absence of clear evidence refrained from speculating about a possible stage/events for which this may be important.  
+
+<|ref|>text<|/ref|><|det|>[[116, 763, 658, 780]]<|/det|>
+We changed the headline of the corresponding Results section to:  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 792, 650, 809]]<|/det|>
+## C9ORF78 moderately down-regulates BRR2 helicase activity  
+
+<|ref|>text<|/ref|><|det|>[[115, 822, 881, 854]]<|/det|>
+We left out the statement from the Results that the findings argue "against a modulation of BRR2 helicase activity constituting a major C9ORF78 function".  
+
+<|ref|>text<|/ref|><|det|>[[115, 867, 761, 884]]<|/det|>
+We included a short section on this aspect in the revised Discussion (line 345):  
+
+<|ref|>text<|/ref|><|det|>[[115, 898, 880, 929]]<|/det|>
+The BRR2- modulatory activity of C9ORF78 we report here is weaker, and C9ORF78 seems to be associated with the spliceosome only at stages when BRR2 has already unwound U4/U6.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 82, 880, 113]]<|/det|>
+However, we presently cannot exclude that C9ORF78- dependent regulation of BRR2 helicase activity may play a role during other stages of splicing.  
+
+<|ref|>text<|/ref|><|det|>[[117, 141, 881, 294]]<|/det|>
+The observations regarding competition between FBP21 and C9ORF78 for Brr2 binding are strongly supported by the data in Fig. 2d. Nonetheless, it is unclear why the authors chose to use Brr2HR complexes rather than complexes that also contained the Jab1 domain of Prp8, given that the Jab1 domain remains bound to Brr2 from the B complex onwards and that the FBP21 interaction is observed in the B complex structure in the context of Brr2 being bound to Jab1. I think the authors could and likely should strengthen their argument here by performing the competition experiment with Brr2HR/Jab1/C9ORF78 complexes and even better do it in the proper physiological progression by asking whether C9ORF78 can compete off FBP21, which is what they propose actually happens during the splicing pathway.  
+
+<|ref|>text<|/ref|><|det|>[[117, 308, 881, 429]]<|/det|>
+We thank the reviewer for this suggestion and have performed the suggested experiments. We have pre- incubated BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup> with FBP21116- 376, which gives rise to a stable BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex as revealed by SEC (new Fig. 2d). Upon adding GST- C9ORF78 to the pre- formed BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex, GST- C9ORF78 displaces FBP21116- 376, as again revealed by SEC (new Fig. 2d). We used a longer FBP21 fragment for these experiments than employed for cryoEM, because PRPF8<sup>Jab1</sup> and FBP21200- 376 (the fragment used in cryoEM) cannot be distinguished on SDS- PAGE gels. We now describe these results in the revised manuscript (line 207) and display them in a new Fig. 2d:  
+
+<|ref|>text<|/ref|><|det|>[[117, 442, 881, 578]]<|/det|>
+To test mutually exclusive binding biochemically, we performed SEC analyses with a preformed BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex to which we added GST- C9ORF78 (Fig. 2d). We used a slightly longer FBP21 fragment in these experiments than in cryoEM studies to allow distinction from PRPF8<sup>Jab1</sup> in SDS- PAGE. While FBP21116- 376 formed a stable complex with BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup> (Fig. 2d, top), GST- C9ORF78 displaced FBP21116- 376 from the BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- FBP21116- 376 complex (Fig. 2d, bottom), indicating mutually exclusive binding and a stronger affinity of GST- C9ORF78 for BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>. These findings suggest that C9ORF78 may aid in the displacement of FBP21 upon conversion of the B to the B<sup>act</sup> complex, when FBP21 and other B- specific proteins are released.  
+
+<|ref|>image<|/ref|><|det|>[[275, 590, 720, 734]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[117, 761, 881, 928]]<|/det|>
+Finally, while I appreciate the authors' model for C9ORF78 function during the \(C^*\) stage as a compelling main mechanism of action, I think the authors need to be much more careful with their discussion of various proteins they claim are present at the C complexes stage, such as DDX23 or Prp6. These have generally only been detected as such in C complexes prepared by biochemical stalls that are not entirely clean and could have contaminating earlier and later complexes. A good example are studies that use a 3'SS mutation to capture C complexes and which are now known from cryo- EM studies to actually reach the \(C^*\) and even P complex stage and then revert back to the C complex stage. Simply citing proteomic analyses of such preparations as evidence of stage- specific association is sufficient evidence for such claims. None of the single particle EM studies of C or \(C^*\) complexes have identified any subpopulations containing
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 81, 883, 278]]<|/det|>
+DDX23 for example, yet the authors routinely seem to suggest this protein could bind at the this stage. The claim that their IP of DDX23 occurs at the C complex stage is particularly problematic in this sense, especially as one could imagine much more easily how complexes that have transitioned to the B complex stage may not have fully lost DDX23 until the Bact stage, making the exchange of FBP21 for C9ORF78 an alternative point at which a transient interaction between C9ORF78 and DDX23 could have occurred. Similarly, the claim that Prp22 binds at the C complex stage should be revised, as what people have reported as binding at that stage most likely results from 3'SS mutant complexes that have reached the \(C^{*}\) stage and then reverted back to the C complex stage without Prp22 dissociation, as was shown in a recent study on equilibrium of spliceosome conformations during catalysis in yeast. I urge the authors to be much more rigorous with their terminology when describing potential binding to various complexes.  
+
+<|ref|>text<|/ref|><|det|>[[116, 292, 882, 473]]<|/det|>
+Again, we very much appreciate these knowledgeable comments. Taking them into careful account, we have now much more judiciously discussed our findings in the revised manuscript. For the detailed revision of the text, please refer to our reply below the next point. The only aspect we want to bring up here is that we did not want to insinuate at all that DDX23 is normally present at the C complex stage, sorry if our descriptions were confusing. Instead, we only wanted to suggest an explanation for our observation that DDX23 was more enriched in our C9ORF78<sup>R41A</sup> Flag- IP compared to the C9ORF78<sup>wt</sup> Flag- IP, which we interpret to reflect the "liberation" of a DDX23- binding site on BRR2 in case of the C9ORF78<sup>R41A</sup> variant, but not in case of C9ORF78<sup>wt</sup>. However, as the reviewer rightly points out that C9ORF78 and DDX23 most likely do not occur in the same or directly neighboring spliceosomal intermediates, and as our descriptions were obviously confusing, we now completely left out the descriptions/discussions referring to DDX23.  
+
+<|ref|>text<|/ref|><|det|>[[116, 502, 882, 758]]<|/det|>
+Related to this matter, I find that the authors overlook too quickly a potential role for C9ORF78 in regulating splice site use also during the B to Bact transition. They observe significant numbers of exon skipping and mutually exclusive exon use changes in their KD experiments. It is much harder to imagine how such events could be regulated at the C or \(C^{*}\) stage, but much easier to imagine how transfer of the 5'SS and docking of the BP helix at the active site, which occur during the B to Bact transition, or are influenced by the relative stability of these complexes, could impact these types of alternative splicing events. Indeed, the crosslinks to U5 snRNA are consistent with such a role and much more likely to reflect interactions at the Bact stage than at the C complex stage, when U5 is buried very deeply into the active site, making an interaction with a flexible part of C9ORF78 harder to imagine. Brr2- dependent association with Cwc27 also supports this idea, as in the Bact structure the flexible C9ORF78 residues proposed to interact with Prp22 in \(C^{*}\) , could easily be imagined to interact with Cwc27 in Bact. Since C9ORF78 likely regulates the Bact transition, as the authors argue with strong experimental support, they should at least discuss the possibility of an earlier role in regulating alternative splicing at this stage through some of the other factors they observe in their IP studies.  
+
+<|ref|>text<|/ref|><|det|>[[117, 772, 880, 835]]<|/det|>
+Again, many thanks for these comments, we agree that in our original discussion we took the nominal association of C9ORF78 with the C complex stage too much at face value. We have amended and thoroughly reworked the corresponding Discussion sections, taking all remarks and suggestions by the reviewer into account (lines 355 and 381):  
+
+<|ref|>text<|/ref|><|det|>[[117, 848, 880, 926]]<|/det|>
+The mutually exclusive binding of FBP21 and C9ORF78 to BRR2 suggests that C9ORF78 might first bind to the spliceosome during the B- to- B<sup>act</sup> transition, when FBP21 is released. While proteomics analyses have suggested that C9ORF78 might be associated with the C complex<sup>27</sup>, the analyzed complexes had been enriched on a modified pre- mRNA lacking a 3'- ss AG dinucleotide and a 3'- exon<sup>55</sup>. CryoEM and biochemical studies have shown that
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[117, 82, 880, 234]]<|/det|>
+spliceosomes assembled on such modified pre- mRNAs can progress to the \(\mathbb{C}^*\) complex state<sup>40</sup>, that neighboring states tend to converge on the C complex state when exon ligation is inhibited<sup>56</sup> and that under appropriate conditions both catalytic steps of splicing can be reversed<sup>57</sup>. Thus, factors identified via proteomics in nominal C complex preparations may to some extent represent contaminations from neighboring states. Presence of C9ORF78 already during the \(\mathsf{B}^{\mathsf{act}}\) stage is further supported by putative interactions we observe with the \(\mathsf{B}^{\mathsf{act}}\) proteins CWC22 and CWC27. However, as the C9ORF78- binding site of BRR2 remains unobstructed in C, \(\mathbb{C}^*\) and P complexes<sup>34,40,41,58,59</sup>, and as we also identified putative C9ORF78 interactions with 1<sup>st</sup> step, \(\mathbb{C}^*\) and 2<sup>nd</sup> step factors, C9ORF78 may also remain bound after the B- to- \(\mathsf{B}^{\mathsf{act}}\) transition.  
+
+<|ref|>text<|/ref|><|det|>[[117, 250, 880, 459]]<|/det|>
+C9ORF78 KD elicited changes in many exon skipping events, a significant number of which are dependent on the BRR2- C9ORF78 interaction. Exon skipping is thought to be decided before the C complex stage, providing additional indirect evidence that C9ORF78 is already present at an earlier stage. We also observed a highly reproducible effect of C9ORF78 KD on alternative usage of NAGNAG 3'- ss, with C9ORF78 strongly favoring usage of the upstream 3'- ss, for which differential recue experiments indicated C9ORF78 specificity but failed to support a dependence on the observed BRR2- C9ORF78 interaction. While these data suggest that different C9ORF78- splicing factor interactions play a predominant role for the regulation of cassette exons and of alternative 3'- ss, assay limitations may also have prevented the detection of subtle effects of the BRR2- C9ORF78 interaction on some alternative splicing events. E.g., the single residue exchange in C9ORF78<sup>41A</sup> is sufficient to destabilize the binary interaction with BRR2 in vitro, but C9ORF78 interaction with other spliceosomal factors and over- expression of the siRNA- resistant C9ORF78 variants may have obscured differences in rescue efficiencies between C9ORF78<sup>wt</sup> and C9ORF78<sup>41A</sup>.  
+
+<|ref|>text<|/ref|><|det|>[[117, 459, 880, 804]]<|/det|>
+Exon skipping might be influenced by the kinetics with which two mutually exclusive splicing scenarios transition from the B via the \(\mathsf{B}^{\mathsf{act}}\) to the \(\mathsf{B}^*\) stage, and our findings suggest that C9ORF78 could modulate these transitions. Recently, additional assembly intermediates between the B and \(\mathsf{B}^{\mathsf{act}}\) stages have been characterized biochemically and structurally<sup>60</sup>. These pre- \(\mathsf{B}^{\mathsf{act}}\) complexes contain, among others, reduced levels of the B- specific FBP21 protein, but also largely lack \(\mathsf{B}^{\mathsf{act}}\) proteins CWC22 and CWC27 and the step 1 factor GPKOW. Given our observations that C9ORF78 can displace FBP21 from BRR2 and could also contact CWC22, CWC27 and GPKOW, presence of C9ORF78 might modulate the kinetics of B- to- \(\mathsf{B}^{\mathsf{act}}\) conversion by driving displacement of FBP21 and helping recruitment of \(\mathsf{B}^{\mathsf{act}}\) proteins and GPKOW. Notably, the multi- step B- to- \(\mathsf{B}^{\mathsf{act}}\) transition is also accompanied by a stepwise repositioning of BRR2<sup>60</sup>, which might likewise be influenced by C9ORF78 that putatively links BRR2 to other components according to our data. Moreover, a large- scale cryoEM analysis has revealed that the human \(\mathsf{B}^{\mathsf{act}}\) complex can adopt at least eight major conformations, which could be arranged along a trajectory towards catalytic activation due to the degree of their similarity to a later intermediate<sup>61</sup>. Such a situation most likely also applies to other splicing stages, and it has been suggested that any additional incoming factor will alter the conformational space available to the respective spliceosomal intermediate<sup>61</sup>. Based on a structural superposition, it is easily conceivable that in \(\mathsf{B}^{\mathsf{act}}\) the intrinsically unstructured C9ORF78 could bridge between BRR2, CWC22/CWC27 and U5 IL1 (Fig. 8a) This presumed cross- strutting of several \(\mathsf{B}^{\mathsf{act}}\) elements would most likely significantly alter the conformational space available to \(\mathsf{B}^{\mathsf{act}}\). C9ORF78 might thereby again alter the kinetics of B- to- \(\mathsf{B}^{\mathsf{act}}\) conversion and/or influence the tendency of the \(\mathsf{B}^{\mathsf{act}}\) complex to adopt a conformation conducive to PRPF2 remodeling.  
+
+<|ref|>text<|/ref|><|det|>[[117, 804, 880, 925]]<|/det|>
+Alternative NAGNAG splice site choice was suggested to take place during the second step of the splicing reaction<sup>48</sup>, indirectly supporting our notion of a continued presence of C9ORF78 at post- \(\mathsf{B}^{\mathsf{act}}\) stages. Furthermore, our observed interaction of C9ORF78 with the second step factor, PRPF22, suggests that C9ORF78 remains associated also with the \(\mathbb{C}^*\) complex. Comparison of our BRR2<sup>HR</sup>- PRPF8<sup>Jab1</sup>- C9ORF78 structure with the structure of a human \(\mathbb{C}^*\) complex revealed that the C- terminal 231 residues of C9ORF78 could easily reach and directly contact PRPF22, which resides in immediate vicinity of BRR2 in the \(\mathbb{C}^*\) complex, as well as CWC22 that is still present at the \(\mathbb{C}^*\) stage (Fig. 8b). PRPF22 has been shown to be involved
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 81, 881, 160]]<|/det|>
+in 3'- ss selection and exon ligation in yeast<sup>62,63</sup>. Thus, one possible mechanism for C9ORF78 to regulate alternative 3'- ss usage may be direct C9ORF78-PRPF22 interactions that affect PRPF22 motor activity, which is thought to reposition alternative 3'- ss in the spliceosome's active site from a distance<sup>4,63</sup>. This interpretation is consistent with our observation that presence of C9ORF78 leads to preferential use of upstream alternative 3'- ss.  
+
+<|ref|>text<|/ref|><|det|>[[115, 173, 212, 190]]<|/det|>
+New Fig. 8:  
+
+<|ref|>image<|/ref|><|det|>[[230, 201, 763, 850]]<|/det|>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 103]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[116, 120, 217, 133]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[116, 135, 291, 148]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 149, 875, 210]]<|/det|>
+First of all, I wanted to say that the authors have done a remarkable job addressing the few points I raised in my original review. I believe the new data obtained by the authors has improved an already good manuscript. As with every good data, it raises more questions than it answers, which is to be expected. Furthermore the data is well presented and adequately discussed.  
+
+<|ref|>text<|/ref|><|det|>[[115, 210, 880, 255]]<|/det|>
+In addition to revealing the relatively small pool of transcripts, sensitive to the integrity of BRR2- C9ORF78 interaction, the RNA- seq results clearly hint at an exciting possibility that C9ORF78 may influence splicing outcomes via interactions with spliceosomal proteins other than BRR2.  
+
+<|ref|>text<|/ref|><|det|>[[115, 255, 881, 314]]<|/det|>
+I believe, that the manuscript in its current version contains a wealth of essential data which would be invaluable for scientists attempting to further investigate the role of C9ORF78 in splicing and beyond. I am of the opinion that the manuscript meets the expected publication standards and I raise no further issues with it.  
+
+<|ref|>text<|/ref|><|det|>[[115, 358, 291, 401]]<|/det|>
+Reviewer #2: Remarks to the Author: No further questions.  
+
+<|ref|>text<|/ref|><|det|>[[115, 447, 217, 460]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[116, 462, 291, 475]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 476, 881, 566]]<|/det|>
+In this revised manuscript the authors provide a compelling and timely analysis of the role of C9ORF78 in modulating splice site usage by the human spliceosome, through direct interactions with the core spliceosomal ATPase Brr2. Altogether, the structural, biochemical, and sequencing data provide strong support for the idea that C9ORF78 is recruited to the spliceosome during the pre- catalytic phase, may modulate spliceosome activation, and likely remains bound during the catalytic stage, when it influences selection of the 3'SS.  
+
+<|ref|>text<|/ref|><|det|>[[115, 580, 857, 610]]<|/det|>
+I think the revised manuscript has a better flow and the authors have presented a better and more nuanced discussion of their data.  
+
+<|ref|>text<|/ref|><|det|>[[115, 625, 869, 670]]<|/det|>
+Overall, I am satisfied that my previous concerns, as well as the potential concerns of the other reviewers, have been addressed in the revised manuscript, including by several new experiments, as requested.  
+
+<|ref|>text<|/ref|><|det|>[[115, 685, 870, 715]]<|/det|>
+I think the manuscript elucidates several functions for C9ORF78 and provides a significant advance in understanding the indirect roles played by Brr2 during later stages of splicing.  
+
+<|ref|>text<|/ref|><|det|>[[115, 730, 553, 745]]<|/det|>
+Thus I strongly recommend publication in its present form.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1ec0e20dfe1397bcc4c3b75dcfb56d1f19c1e910da1612eea66abbbed7b5cf35/images_list.json

@@ -0,0 +1,145 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_0.jpg",
+    "caption": "Figure R1. XANES of Fe foil, FePc, \\(\\mathrm{Fe}_2\\mathrm{O}_3\\) and pFeSAN.",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 0
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_1.jpg",
+    "caption": "Figure R2. Comparison of the experimental XANES curve (pFeSAN) with the calculated XANES data of Fe–N3. The insets show the optimized Fe–N3 structure. The relative pre-edge and white line peak of the simulated spectrum are consistent with the experiment.",
+    "footnote": [],
+    "bbox": [
+      [
+        348,
+        580,
+        645,
+        788
+      ]
+    ],
+    "page_idx": 5
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_2.jpg",
+    "caption": "Figure R3. EXAFS fitting results for Fe foil, FePc, Fe<sub>2</sub>O<sub>3</sub> and pFeSAN at k-space.",
+    "footnote": [],
+    "bbox": [
+      [
+        370,
+        384,
+        614,
+        555
+      ]
+    ],
+    "page_idx": 5
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_3.jpg",
+    "caption": "Figure R4. Oxidase-like activity evaluation. a. Comparison of relative oxidase-like catalytic activities of different nanozymes. b. Comparison of kinetics for pFeSAN, Pt and \\(\\mathrm{Mn_3O_4}\\) . \\(\\mathrm{K_m}\\) is the Michaelis-Menten constant. \\(\\mathrm{V_{max}}\\) is the maximal reaction velocity. c. Steady-state kinetic assay of Pt and \\(\\mathrm{Mn_3O_4}\\) with TMB as substrate.",
+    "footnote": [],
+    "bbox": [
+      [
+        207,
+        560,
+        770,
+        812
+      ]
+    ],
+    "page_idx": 6
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_4.jpg",
+    "caption": "Figure R5. a. The photo of comparison of the mixed solution \\(\\mathrm{H}_2\\mathrm{O}_2\\) in the absence and presence of pFeSAN with the concentration of \\(40~\\mu \\mathrm{g / mL}\\) . b. Comparison of CAT-like activities of pFeSAN at various pH (4-12) conditions. c. The \\(\\mathrm{H}_2\\mathrm{O}_2\\) scavenging of pFeSAN with different concentrations.",
+    "footnote": [],
+    "bbox": [
+      [
+        222,
+        696,
+        772,
+        821
+      ]
+    ],
+    "page_idx": 8
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_5.jpg",
+    "caption": "Figure R6. Cellular viability of AML12 and Hepa 1-6 cells treated with various concentrations of the pFeSAN for 3 h.",
+    "footnote": [],
+    "bbox": [
+      [
+        300,
+        376,
+        666,
+        600
+      ]
+    ],
+    "page_idx": 10
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_6.jpg",
+    "caption": "Figure R1. a. Pore size distribution curves of Fe- \\(\\mathrm{N}_4\\) and pFeSAN b. BET surface areas of Fe- \\(\\mathrm{N}_4\\) and pFeSAN.",
+    "footnote": [],
+    "bbox": [
+      [
+        208,
+        209,
+        757,
+        386
+      ]
+    ],
+    "page_idx": 13
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_7.jpg",
+    "caption": "Figure R2. Reaction-time curves of the TMB colorimetric reaction catalyzed by pFeSAN, Fe-N4 and \\(\\mathrm{Fe_3O_4}\\) .",
+    "footnote": [],
+    "bbox": [
+      [
+        330,
+        500,
+        640,
+        699
+      ]
+    ],
+    "page_idx": 27
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_8.jpg",
+    "caption": "Figure R3. EPR spectra of pFeSAN in the presence of phenyloxoiodine at 77 K.",
+    "footnote": [],
+    "bbox": [
+      [
+        348,
+        336,
+        630,
+        531
+      ]
+    ],
+    "page_idx": 28
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_9.jpg",
+    "caption": "Figure R4. Calculated electron transfer number derived from rotating ring-disk electrode and \\(\\mathrm{H}_2\\mathrm{O}_2\\) yields of the pFeSAN.",
+    "footnote": [],
+    "bbox": [
+      [
+        340,
+        163,
+        680,
+        367
+      ]
+    ],
+    "page_idx": 30
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1ec0e20dfe1397bcc4c3b75dcfb56d1f19c1e910da1612eea66abbbed7b5cf35/supplementary_0_Peer Review File__1ec0e20dfe1397bcc4c3b75dcfb56d1f19c1e910da1612eea66abbbed7b5cf35.mmd

@@ -0,0 +1,554 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione  
+
+![](images/Figure_unknown_0.jpg)
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to  
+
+the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+This paper presents an innovative biomimetic synthetic strategy for the scalable synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using hemoglobin as a template. The authors demonstrate the superior catalytic activity of pFeSAN compared to Fe- N4 and Fe3O4 nanozymes, attributing it to the suppressed aggregation of atomically dispersed Fe, facilitated mass transfer, and maximized exposure of active sites. Additionally, pFeSAN was successfully applied in the rapid colorimetric detection of glutathione and developed as a real- time, facile, rapid, and precise visualization analysis methodology for tumors via glutathione levels. Overall, this paper is a valuable contribution to the field of nanomaterials with potential applications in diagnostics and therapeutics. Therefore, I recommend it can be accepted by Nature Communications after revision if some problems can be properly addressed:  
+
+1. Figure 5f: the EPR patterns of the high-valent iron-oxygen species formed by SACs with different coordination numbers (Fe-Nx) should be different. In this manuscript, the number of N atoms coordinated with Fe is 3. Why is the EPR signal of the high-valent Fe oxygen species formed by FeN3 the same as the signal formed by FeN5 (Refs. 21 and 23)?  
+
+2. Page 10, in the XANES results, it should be "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe2O3 and Fe foil (not FePC), and very close to that of FePC.".  
+
+3. Page 11, SACs can be regarded as a complex (coordination compound). The chemical configuration (such as the coordination number, the chemical environment of the coordination atom, etc.) of the active site in SACs will change the electronic structure of the metal by affecting the splitting energy and stability energy of the central d orbital of the metal. Why the author judged that the electronic structure around Fe in FeN3 is similar to that in FePC through the partial overlap of absorption edges? In addition, the author did not comprehensively consider the Edge front peak and white line peak.  
+
+4. Compared with R-space, the data in k-space can better reflect the quality of synchrotron radiation data. So, the real data and fitting data in k-space should be included.  
+
+5. For R-space, the data after \(4\mathrm{\AA}\) is meaningless. In order to better show the R-space data and the situation, it is recommended to shorten the Abscissa to less than \(6\mathrm{\AA}\) (Figures 3b and 3c). In addition, the Abscissa of R-space should be R \((\mathrm{\AA})\) only after the phase shift is corrected. I am wondering whether no phase shift correction was performed for the data in the figure, if it is, the Abscissa should be \(\mathrm{R} + \mathrm{a}(\mathrm{\AA})\) , please check it.  
+
+6. It is mentioned by the author that Hb is competitive in price. But the price of Hb (US \(\)3.60\(/g,\) Shanghai Yuanye, S12021-5g) is significantly higher than that of FeCl3.  
+
+7. Lack of comparison, compared with other nano-enzymes with oxidase-like activity, what is the advantage of pFeSAN in the kinetic constant?  
+
+8. For Figure 5a and 5b, the author pointed out that "no apparently detectable signals of any ROS further proved that ROSs were not the main intermediates for the oxidase-like activity of pFeSAN, which was very different from the majority of the previously reported nanozymes", please give more discussion on this phenomenon.  
+
+9. For Figure 2f, the resolution is too low.  
+
+Reviewer #2:  
+
+Remarks to the Author:  
+
+Due to the high atom utilization and remarkable catalytic activity, single- atom catalysts exhibit great potentials for applications of nanozyme. In this work, a porous Fe single- atom catalyst prepared by a biomineralization- pyrolysis strategy exhibited remarkable oxidase- like activity, which was used for GSH sensing and cancer cell identification. The porous Fe single- atom nanozyme was well characterized. And, the mechanism of oxidase- like activity was investigated as well, which opened up a new way to design and optimize nanozymes. Therefore, this manuscript can be accepted to be published by Nature Communications after proper revisions.  
+
+1. Fe single-atom nanozyme reported in the past often exert catalase-like catalytic activity. In this study, dose pFeSAN have catalase-like activity?
+
+<--- Page Split --->
+
+2. Compared with previously reported Fe single-atom nanozyme, what are the advantages of pFeSAN in structure-activity characteristics?  
+3. The label of elements and scale bar at Supplementary Fig. 11 and 15 was not clearly visible.  
+4. Please provide the quantitative analysis of EIS (Fig. 5d).  
+5. In this work, did the pFeSAN show obvious toxicity to normal cells and tumor cells, resulting in the influenced cellular GSH analysis?  
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+This paper deals with the development of a porous 3- coordinate iron- based nanozyme which can exhibit oxidase- like activity. The authors use hemoglobin as Fe source/template and incorporates it into a zeolitic imidazolate framework (ZIF- 8). The single- atom nanozyme called pFeSAN acts as an oxidase model by oxidizing the colorimetric substrate TMB to oxTMB, which helps in the detection of glutathione (GSH) in nM concentrations. The authors show a significant improvement in the oxidase- like activity; the nanozyme exhibiting 3.3- and 8791- fold higher oxidase- like activity than the Fe- N4 and Fe3O4 nanozymes reported earlier. As the GSH level is altered in many diseases, including cancer, the authors propose that the nanozyme with oxidase- like activity can be used for the visualization of tumor. The work is interesting, but does not warrant publication in Nat. Commun. due to the following reasons.  
+
+(1) Development of Fe-based single-atom nanozyme by incorporating hemoglobin into ZIF-8: The topic of Fe-based single atom nanozyme is not new. It has been shown that bovine hemoglobin can be incorporated into ZIF-8 to develop single atom nanozymes (ACS Appl. Mater. Interfaces 2016, 8, 29052-29061). This nanozyme has been used as peroxidase mimic to oxidize TMB for the colorimetric detection. Many other Fe-based single atom nanozymes with oxidase-like activity have been reported in a review (Nano Res. 2023, 16, 1992-2002). The mechanism of the structure-dependent oxidase-like activity of Fe-N/C nanozyme has also been reported (Chem. Commun. 2019, 55, 5271-5274).  
+
+(2) Single atom nanozymes for the detection of glutathione (GSH). Nanozymes and Fe-N-C-based single atom nanozymes have been used extensively for the detection of glutathione (Journal of Materiomics, 2022, 6, 1251-1259). Mn3O4 microspheres have been used as an oxidase mimic for rapid detection of glutathione (RSC Adv., 2019,9, 16509-16514). Light-responsive MOF as an oxidase mimic for cellular GSH detection (Anal. Chem. 2019, 91, 13, 8170-8175). MnO2 nanosheets as an artificial enzyme to mimic oxidase for rapid and sensitive detection of glutathione (Biosensors & Bioelectronics, 2017, 90, 69-74). Therefore, the present study lacks novelty in the detection of glutathione.  
+
+(3) The concept of using GSH as biomarker for visualization of cancer cells is not entirely new. There are many recent reports in the literature which highlight the concept. For example, a MOF has been reported to exhibit oxidase-like activity by oxidizing TMB to oxTMB, which has been used as a colorimetric probe for GSH detection. The oxidase mimic has been used to analyze the GSH level in the lysates of normal and cancer cells (Anal. Chem. 2019, 91, 8170-8175). There are many other reports which describe the use of nanozymes for tumor visualization through GSH detection.
+
+<--- Page Split --->
+
+## Point-by-point Response to Reviewer(s)' Comments  
+
+Reviewer #1 (Remarks to the Author):  
+
+This paper presents an innovative biomimetic synthetic strategy for the scalable synthesis of porous \(\mathrm{Fe - N_3}\) single atom nanozymes (pFeSAN) using hemoglobin as a template. The authors demonstrate the superior catalytic activity of pFeSAN compared to \(\mathrm{Fe - N_4}\) and \(\mathrm{Fe_3O_4}\) nanozymes, attributing it to the suppressed aggregation of atomically dispersed Fe, facilitated mass transfer, and maximized exposure of active sites. Additionally, pFeSAN was successfully applied in the rapid colorimetric detection of glutathione and developed as a real- time, facile, rapid, and precise visualization analysis methodology for tumors via glutathione levels. Overall, this paper is a valuable contribution to the field of nanomaterials with potential applications in diagnostics and therapeutics. Therefore, I recommend it can be accepted by Nature Communications after revision if some problems can be properly addressed:  
+
+We thank the reviewer for his/her constructive comments on our manuscript. We would like address his/her comments as below.  
+
+1. Figure 5f: the EPR patterns of the high-valent iron-oxygen species formed by SACs with different coordination numbers \((\mathrm{Fe - N_x})\) should be different. In this manuscript, the number of N atoms coordinated with Fe is 3. Why is the EPR signal of the high-valent Fe oxygen species formed by \(\mathrm{FeN_3}\) the same as the signal formed by \(\mathrm{FeN_5}\) (Refs. 21 and 23)?  
+
+## Response:  
+
+Thank you for this constructive comment. Fe- heme structure of natural cytochrome c oxidase (CcO) plays the core roles for the \(\mathrm{O_2}\) activation through the \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) pathway, delivering high catalytic activity for many oxidation reactions. The catalytic cycle of \(\mathrm{CcO}\) for oxygen reduction starts with the binding of \(\mathrm{O_2}\) on a ferrous heme to afford a \(\mathrm{Fe(IV) = O}\) intermediate, followed by the reduction of \(\mathrm{O_2}\) into \(\mathrm{H_2O}\) (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). The \(\mathrm{Fe(IV) = O}\) is considered as the key intermediate in the catalytic cycles of many heme iron enzymes and heme analogs for oxidation. Although
+
+<--- Page Split --->
+
+the coordination environments of Fe- \(\mathrm{N}_3\) and Fe- \(\mathrm{N}_5\) are not identical, they share the common mechanistic grounds through the \(\mathrm{Fe(IV) = O}\) intermediate (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). In this work, a typical diamond- shaped label signal at \(\mathrm{g} = 2.03\) was detected by EPR, consistent with the \(\eta^2\) - peroxy heme species, signifying the formation of \(\mathrm{Fe(IV) = O}\) intermediate. This suggests that Fe- \(\mathrm{N}_3\) active center of pFeSAN may have a similar reaction process with heme analogs. We suggest that the same oxidase- like reactivities and \(\mathrm{Fe(IV) = O}\) intermediate of Fe- \(\mathrm{N}_3\) and Fe- \(\mathrm{N}_5\) structure are mainly caused by their active site structures, but not the axial ligand. The more information has been added in the revised manuscript (Page 20, Line 3- 6).  
+
+2. Page 10, in the XANES results, it should be "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe₂O₃ and Fe foil (not FePC), and very close to that of FePC.".  
+
+## Response:  
+
+Thank you sincerely for your suggestion. In the revised manuscript, we have modified the related information "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe₂O₃ and Fe foil, and very close to that of FePC," in the main context (Page 10, Line 19- 21).  
+
+3. Page 11, SACs can be regarded as a complex (coordination compound). The chemical configuration (such as the coordination number, the chemical environment of the coordination atom, etc.) of the active site in SACs will change the electronic structure of the metal by affecting the splitting energy and stability energy of the central d orbital of the metal. Why the author judged that the electronic structure around Fe in FeN₃ is similar to that in FePC through the partial overlap of absorption edges? In addition, the author did not comprehensively consider the Edge front peak and white line peak.  
+
+## Response:  
+
+Thank you very much for pointing this out. XANES reveals slight differences between the XANES spectra of pFeSAN and FePc. The latter shows a pre- edge peak at 7118 eV
+
+<--- Page Split --->
+
+assigned to a \(1\mathrm{s} \rightarrow 4\mathrm{p}_z\) shakedown transition characteristic for a square- planar configuration with high \(\mathrm{D_{4h}}\) symmetry (Figure R1). For characteristics of pFeSAN, the pre- edge feature is absent for pFeSAN, revealing a broken \(\mathrm{D_{4h}}\) symmetry (Ref. 38: Nat. Mater., 2015, 14, 937–942). The theoretical XANES was also simulated with FDMNES code (J. Phys.: Condens. Matter, 2009, 21, 345501). The optimized Fe–N3 structure was used as the input structure. The sphere radius to calculate the cluster adsorption was 12 Å. The result showed that the relative pre- edge and white line peaks of the simulated spectrum were consistent with the experiment, validating the Fe–N3 local structure of pFeSAN (Figure R2).  
+
+![](images/Figure_unknown_1.jpg)
+
+<center>Figure R1. XANES of Fe foil, FePc, \(\mathrm{Fe}_2\mathrm{O}_3\) and pFeSAN. </center>  
+
+![](images/Figure_unknown_2.jpg)
+
+<center>Figure R2. Comparison of the experimental XANES curve (pFeSAN) with the calculated XANES data of Fe–N3. The insets show the optimized Fe–N3 structure. The relative pre-edge and white line peak of the simulated spectrum are consistent with the experiment. </center>
+
+<--- Page Split --->
+
+The relative information has been updated in the revised manuscript (Page11, Line 1–4 and 16–19, and Supplementary Fig. 19).  
+
+4. Compared with R-space, the data in k-space can better reflect the quality of synchrotron radiation data. So, the real data and fitting data in k-space should be included.  
+
+## Response:  
+
+Thank you for your careful review. The best- fit analysis is shown in Figure R3. The dominant contribution is given by Fe–N first shell coordination. In the revised manuscript, we have added the related information in the main context (Page 11, Line 9–11 and Fig. 3d).  
+
+![](images/Figure_unknown_3.jpg)
+
+<center>Figure R3. EXAFS fitting results for Fe foil, FePc, Fe<sub>2</sub>O<sub>3</sub> and pFeSAN at k-space. </center>  
+
+5. For R-space, the data after 4 Å is meaningless. In order to better show the R-space data and the situation, it is recommended to shorten the Abscissa to less than 6 Å (Figures 3b and 3c). In addition, the Abscissa of R-space should be R (Å) only after the phase shift is corrected. I am wondering whether no phase shift correction was performed for the data in the figure, if it is, the Abscissa should be R + α (Å), please check it.  
+
+## Response:  
+
+We appreciate the reviewer's suggestion. We changed the abscissa according to the reviewer's suggestion and updated Figures 3b and 3c in the revised manuscript (Page 12, Fig. 3b and 3c). Meanwhile, the wavelet transform used the raw data of XAFS directly, no phase shift correction was performed.
+
+<--- Page Split --->
+
+6. It is mentioned by the author that Hb is competitive in price. But the price of Hb (US \(3.60 / \mathrm{g}\) , Shanghai Yuanye, S12021-5g) is significantly higher than that of FeCl₃.  
+
+## Response:  
+
+We appreciate the reviewer's valuable comments. The prices of Hb and FeCl₃ are greatly depended on their grades and suppliers. Herein, we provided the prices of two iron resources for reference. At least, the price of Hb is accepted as the cheap resource. The price of Hb is US \(42.7 per 100 g from Energy Chemical, D110156- 100g and the FeCl₃ is US\) 50.3 per 100 g from Sigma- Aldrich (157740- 100G), whereas the activity of Hb- templated pFeSAN can be up to 8791 times higher than that Fe₂O₃ with FeCl₃ as iron source. Therefore, this one- pot biomimetic synthetic strategy is efficient and low- cost. We agree the Reviewer's comments and made the change in the revised manuscript in Page 13, Line 6- 7.  
+
+Besides, blood protein (hemoglobin and albumin) as a biomass, which is normally discarded as waste and causes serious environmental pollution (Adv. Mater., 2018, 30, e1703691). This work offers a promising green route to attain high- value catalysts from biomass and demonstrates another advantage of using Hb as an iron source and template for iron single atom catalysts.  
+
+More importantly, Hb has three significant advantages over FeCl₃:  
+
+(1) Fe ions residing in the structure of Hb could be isolated by protein structure, effectively preventing the aggregation of atomically dispersed iron active sites during harsh pyrolysis process;  
+
+(2) Hb with a size of 2\~3 nm forms a large number of mesopores during pyrolysis, benefiting the formation of mesopores and the maximum exposure of atomic Fe sites and also facilitating mass transfer during catalysis;  
+
+(3) Unsymmetrically coordinated Fe-N₃ active site of pFeSAN was prepared by the Hb-pyrolysis strategy, delivering elevated oxidase-like activity.  
+
+Oppositely, FeCl₃ as a Fe source requires careful regulation to prevent metal site from agglomeration. Also, other hard templates are needed to create the mesoporous
+
+<--- Page Split --->
+
+structures, requiring the precise control of the operation. In contrast, our synthesis strategy is facile and easily scale- up.  
+
+7. Lack of comparison, compared with other nano-enzymes with oxidase-like activity, what is the advantage of pFeSAN in the kinetic constant?  
+
+## Response:  
+
+We appreciate the reviewer's valuable suggestions. Besides the control catalysts of \(\mathrm{Fe_3O_4}\) and FeSAN \(\mathrm{(Fe - N_4)}\) , we also systematically compare the performance of pFeSAN with the previously reported Pt NPs (Ref. 41: Biosens. Bioelectron. 2017, 92, 442) and \(\mathrm{Mn_3O_4}\) (Ref. 42: Sens. Actuators B Chem. 2021, 333, 129560) on their oxidase- like activity using TMB as a chromogenic substance. The relative activities of the pFeSAN were 5882 and 1176 times higher than that of Pt and \(\mathrm{Mn_3O_4}\) , respectively, indicating that the oxidase- like activity of the pFeSAN was much higher than Pt and \(\mathrm{Mn_3O_4}\) (Figure R4a). According to Figure R4b,c, pFeSAN has a higher affinity for substrate TMB than Pt and \(\mathrm{Mn_3O_4}\) , and the catalytic reaction rate is much higher than both Pt and \(\mathrm{Mn_3O_4}\) . The relative information has been updated in revised manuscript (Page 16, Line 20–22 and Supplementary Fig. 31).  
+
+![](images/Figure_unknown_4.jpg)
+
+<center>Figure R4. Oxidase-like activity evaluation. a. Comparison of relative oxidase-like catalytic activities of different nanozymes. b. Comparison of kinetics for pFeSAN, Pt and \(\mathrm{Mn_3O_4}\) . \(\mathrm{K_m}\) is the Michaelis-Menten constant. \(\mathrm{V_{max}}\) is the maximal reaction velocity. c. Steady-state kinetic assay of Pt and \(\mathrm{Mn_3O_4}\) with TMB as substrate. </center>
+
+<--- Page Split --->
+
+8. For Figure 5a and 5b, the author pointed out that “no apparently detectable signals of any ROS further proved that ROSs were not the main intermediates for the oxidase-like activity of pFeSAN, which was very different from the majority of the previously reported nanozymes”, please give more discussion on this phenomenon.  
+
+## Response:  
+
+Thank you very much for your comments. Under \(\mathrm{N}_2\) atmosphere, TMB could not be oxidized by pFeSAN. Oppositely, the oxidase-like activity of pFeSAN could be boosted up to 2.2-fold by increasing the \(\mathrm{O}_2\) concentration to saturation (Figure 5a). Therefore, the high \(\mathrm{O}_2\) - dependent activity of pFeSAN for substrate oxidation suggests the essence of \(\mathrm{O}_2\) as oxidants for catalytic reaction. Meanwhile, ROSs were undetectable by using either ROS quenchers (Supplementary Fig. 32a) and EPR spectra (Supplementary Fig. 32b). These results strongly confirmed that pFeSAN mediated the complete reduction of \(\mathrm{O}_2\) to \(\mathrm{H}_2\mathrm{O}\) without the release of free ROSs. This phenomenon verified that pFeSAN likely followed an oxygen atom transfer mechanism, similar to that of the natural \(CCO\) enzymes mediated by the \(\mathrm{Fe(IV) = O}\) intermediate, which was very different from the majority of the previously reported oxidase-like nanozymes through the free ROS pathway (Ref. 23: Anal. Chem. 2022, 94, 15270; Ref. 43: Catal. Sci. Technol. 2021, 11, 7255). The \(\mathrm{Fe(IV) = O}\) is considered a key intermediate in the oxidase-like reaction of many heme iron enzymes and heme analogs. (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). We speculated that the formation of \(\mathrm{Fe(IV) = O}\) intermediates enabled the efficient oxidation of TMB substrate into ox-TMB. In the revised manuscript, we have added the related information in the main context (Page 18, Line 6–8).  
+
+9. For Figure 2f, the resolution is too low.  
+
+## Response:  
+
+Thank you for your kind suggestions. In the revised manuscript, we have updated the related information in the Results section (Page 7, Fig. 2f).
+
+<--- Page Split --->
+
+Reviewer #2 (Remarks to the Author):  
+
+Due to the high atom utilization and remarkable catalytic activity, single- atom catalysts exhibit great potentials for applications of nanozyme. In this work, a porous Fe single- atom catalyst prepared by a biomineralization- pyrolysis strategy exhibited remarkable oxidase- like activity, which was used for GSH sensing and cancer cell identification. The porous Fe single- atom nanozyme was well characterized. And, the mechanism of oxidase- like activity was investigated as well, which opened up a new way to design and optimize nanozymes. Therefore, this manuscript can be accepted to be published by Nature Communications after proper revisions.  
+
+We thank the reviewer for his/her useful suggestions on our manuscript. The point- to- point response to the reviewers' comments is shown as below.  
+
+1. Fe single-atom nanozyme reported in the past often exert catalase-like catalytic activity. In this study, does pFeSAN have catalase-like activity?  
+
+## Response:  
+
+Thank you very much for this constructive comment. Our experimental results show that the pFeSAN possesses ideal catalase- like (CAT- like) activity under neutral conditions (pH 6.0- 7.0), similar to those previously reported studies (Chem. Commun., 2019, 55, 14534- 14537; Adv. Mater., 2022, 34, e2205324; Nat. Commun., 2022, 13, 4744. ) (Figure R5a,b). Besides, the elimination efficiency of \(\mathrm{H}_2\mathrm{O}_2\) was dependent on the concentration of pFeSAN (Figure R5c). The CAT- like activity of pFeSAN may provide opportunities for its in biomedical applications, such as the treatments of neurodegenerative diseases, inflammation, as well as cancer treatment.  
+
+![](images/Figure_unknown_5.jpg)
+
+<center>Figure R5. a. The photo of comparison of the mixed solution \(\mathrm{H}_2\mathrm{O}_2\) in the absence and presence of pFeSAN with the concentration of \(40~\mu \mathrm{g / mL}\) . b. Comparison of CAT-like activities of pFeSAN at various pH (4-12) conditions. c. The \(\mathrm{H}_2\mathrm{O}_2\) scavenging of pFeSAN with different concentrations. </center>
+
+<--- Page Split --->
+
+2. Compared with previously reported Fe single-atom nanozyme, what are the advantages of pFeSAN in structure-activity characteristics?  
+
+## Response:  
+
+We thank the Reviewer for the positive comments. Developing artificial enzymes with the excellent catalytic performance of natural enzymes has been a long- standing goal for chemists. Fe single- atom catalysts coordinated with N atoms (Fe- N- C) exhibit well- defined atomic structures and electronic coordination environments and thereby deliver various enzyme- like catalytic activity. Unfortunately, majority of them were synthesized via pyrolysis at high temperatures, leading to structural collapse and part of the buried Fe- N<sub>x</sub> units inaccessible to biomolecules. Also, the strong stacking of those N- doped graphite carbon in various Fe- N- C nanozymes generally induces the frustrated diffusion of bio- substrates to metal sites. Hence, various methods including spatial confinement, defect/vacancy engineering and coordination modulations have been developed to solve those problems. However, only part of those inadequacies could be overcome. Thus, seeking for a new synthetic strategy of single- atom metal nanozymes to simultaneously achieve the atomic metal dispersion, modulated electronic structure, elevated mass transport and tailorable coordination environment is still on high demands.  
+
+In this work, we report a straightforward Hb- mediated approach to efficiently distribute iron atoms homogeneously and in an atomically dispersed fashion across the nitrogen- doped carbon support. Meanwhile, the as- prepared three- coordinated single- atom Fe nanozyme possesses significantly higher specific surface area and mesoporosity to facilitate the mass transport and exposure of active Fe sites, and exhibits ideal oxidase- like catalytic activity. The benefits of our approach could be described as below.  
+
+(1) pFeSAN with the unsymmetrically coordinated Fe-N₃ active sites was synthesized successfully by using Hb as Fe source. This synthesis represented a simple, facile and scalable one.  
+(2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and
+
+<--- Page Split --->
+
+significantly facilitating the mass transfer of reactants/products during the catalytic process.  
+
+(3) pFeSAN delivered outstanding oxidase-like activity, which was 3.3- and 8791-times higher than those of Fe-N4 and Fe3O4 nanozymes, respectively.  
+
+(4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of \(\mathrm{O_2}\) into \(\mathrm{H_2O}\) , being identical to that of \(\mathrm{CcO}\) , which was very different from the majority of the previously reported oxidase-like nanozymes with the generation of ROS.  
+
+(5) pFeSAN as a highly-performed nanozyme exhibited a much higher upper detection limit of GSH at \(1\mathrm{mM}\) , which was 2.5 to 40-fold higher than those of the previously reported investigations (Table R4).  
+
+(6) Visual and rapid detection of tumor tissues through GSH colorimetric analysis was achieved for the first time, which was expected to help the effective resection of tumor tissues.  
+
+3. The label of elements and scale bar at Supplementary Fig. 11 and 15 was not clearly visible.  
+
+## Response:  
+
+We are grateful for the suggestion. We have added the label of elements and scale bar in the revised Supplementary Information Fig. 11 and 15.  
+
+4. Please provide the quantitative analysis of EIS (Fig. 5d).  
+
+## Response:  
+
+Thank you for your insightful comments. According to the EIS fitting results, it is known that the material resistance of pFeSAN is much smaller than that of FeSAN. Figure 5d shows the Nyquist plots of pFeSAN and FeSAN, where the diameter of the semicircle is associated with the electron-transfer resistance \((\mathrm{R}_{\mathrm{ct}})\) . The \(\mathrm{R}_{\mathrm{ct}}\) decreased substantially from 407.3 to \(298.6 \Omega\) , implying enhanced electron transfer property and better mass transfer performance, which were consistent with the above results from CV (Fig. 5c). We have revised the manuscript and added the description in Results section (Page 19, Line 13).
+
+<--- Page Split --->
+
+5. In this work, did the pFeSAN show obvious toxicity to normal cells and tumor cells, resulting in the influenced cellular GSH analysis?  
+
+## Response:  
+
+As suggested by the reviewer, we have now performed a biosafety analysis in vitro. In order to make sure the biosafety of pFeSAN, we examined the biocompatibility of the pFeSAN through in vitro cell viability. As shown in Figure R6, pFeSAN exhibited negligible cytotoxicity (cell viability \(>90\%\) ) on normal liver cells (AML12) and liver tumor cells (Hepa 1- 6) even at a high concentration of \(200 \mu \mathrm{g} / \mathrm{mL}\) , corroborating the high biocompatibility of the pFeSAN.  
+
+![](images/Figure_unknown_6.jpg)
+
+<center>Figure R6. Cellular viability of AML12 and Hepa 1-6 cells treated with various concentrations of the pFeSAN for 3 h. </center>
+
+<--- Page Split --->
+
+Reviewer #3 (Remarks to the Author):  
+
+This paper deals with the development of a porous 3- coordinate iron- based nanozyme which can exhibit oxidase- like activity. The authors use hemoglobin as Fe source/template and incorporates it into a zeolitic imidazolate framework (ZIF- 8). The single- atom nanozyme called pFeSAN acts as an oxidase model by oxidizing the colorimetric substrate TMB to oxTMB, which helps in the detection of glutathione (GSH) in nM concentrations. The authors show a significant improvement in the oxidase- like activity; the nanozyme exhibiting 3.3- and 8791- fold higher oxidase- like activity than the Fe- N4 and Fe3O4 nanozymes reported earlier. As the GSH level is altered in many diseases, including cancer, the authors propose that the nanozyme with oxidase- like activity can be used for the visualization of tumor. The work is interesting, but does not warrant publication in Nat. Commun. due to the following reasons.  
+
+We thank the Reviewer for raising the critical and constructive comments on our manuscript. According to the referees' comments, the manuscript has been carefully revised. The answers to the comments by referees are enclosed as below.  
+
+(1) Development of Fe-based single-atom nanozyme by incorporating hemoglobin into ZIF-8: The topic of Fe-based single atom nanozyme is not new. It has been shown that bovine hemoglobin can be incorporated into ZIF-8 to develop single atom nanozymes (ACS Appl. Mater. Interfaces 2016, 8, 29052–29061). This nanozyme has been used as peroxidase mimic to oxidize TMB for the colorimetric detection. Many other Fe-based single atom nanozymes with oxidase-like activity have been reported in a review (Nano Res. 2023, 16, 1992–2002). The mechanism of the structure-dependent oxidase-like activity of Fe-N/C nanozyme has also been reported (Chem. Commun. 2019, 55, 5271–5274).  
+
+## Response:  
+
+Thank you for your constructive comments. Developing artificial enzymes with the excellent catalytic performance of natural enzymes has been a long- standing goal for
+
+<--- Page Split --->
+
+chemists. Fe single- atom catalysts coordinated with N atoms (Fe- N- C) exhibit welldefined atomic structures and electronic coordination environments and thereby deliver various enzyme- like catalytic activity. Unfortunately, majority of them were synthesized via pyrolysis at high temperatures, leading to structural collapse and part of the buried Fe- N<sub>x</sub> units inaccessible to biomolecules. Also, the strong stacking of those N- doped graphite carbon in various Fe- N- C nanozymes generally induces the frustrated diffusion of bio- substrates to metal sites. Hence, various methods including spatial confinement, defect/vacancy engineering and coordination modulations have been developed to solve those problems. However, only part of those inadequacies could be overcome. Thus, seeking for a new synthetic strategy of single- atom metal nanozymes to simultaneously achieve the atomic metal dispersion, modulated electronic structure, elevated mass transport and tailorable coordination environment is still on high demands.  
+
+In this work, we report a straightforward hemoglobin- mediated approach to efficiently distribute iron atoms homogeneously and in an atomically dispersed fashion across the nitrogen- doped carbon support. Meanwhile, the as- prepared three- coordinated single- atom Fe nanozyme possesses significantly higher specific surface area and mesoporosity to facilitate the mass transport and exposure of active Fe sites, and exhibits ideal oxidase- like catalytic activity. The benefits of our approach could be described as below.  
+
+(1) pFeSAN with the unsymmetrically coordinated Fe-N<sub>3</sub> active sites was synthesized successfully by using Hb as Fe source. This synthesis represented a simple, facile and scalable one.  
+(2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants/products during the catalytic process.  
+(3) pFeSAN delivered outstanding oxidase-like activity, which was 3.3- and 8791- times higher than those of Fe-N<sub>4</sub> and Fe<sub>3</sub>O<sub>4</sub> nanozymes, respectively.  
+(4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of O<sub>2</sub> into H<sub>2</sub>O, being identical to that of C<sub>6</sub>O, which was very different from the majority of the previously reported
+
+<--- Page Split --->
+
+oxidase-like nanozymes with the generation of reactive oxygen species (ROS).  
+
+(5) pFeSAN as a highly-performed nanozyme exhibited a much higher upper detection limit of GSH at 1 mM, which was 2.5 to 40-fold higher than those of the previously reported investigations (Table R4).  
+
+(6) Visual and rapid detection of tumor tissues through GSH colorimetric analysis was achieved for the first time, which was expected to help the effective resection of tumor tissues.  
+
+Herein, we detailed comparisons the previous works listed by the Reviewer and wished to clearly illustrate the value and innovation of our work.  
+
+ACS Appl. Mater. Interfaces, 2016, 8, 29052- 29061:  
+
+In this work, hemeprotein was directly embedded in ZIF- 8 and the hemeprotein- embedded ZIF- 8 as peroxidase mimic applied for \(\mathrm{H}_2\mathrm{O}_2\) and phenol detection. The ZIF- 8@BHb hybrid composite is a MOF- based nanozyme, not a Fe single atom nanozyme. Compared with our work, the material structure, type of mimic- enzyme activity and detection substrate of the above studies are very difference (Table R1). Thus, the two studies present different research contents.  
+
+Table R1. Comparison of pFeSAN and ZIF-8@BHb hybrid composites.   
+
+<table><tr><td></td><td>Catalyst</td><td>Mimic function</td><td>Application</td></tr><tr><td>Our work</td><td>Mesoporous Fe-N3</td><td>Oxidase-like (OXD)</td><td>GSH detection</td></tr><tr><td>ACS Appl. Mater. Interfaces, 2016, 8, 29052.</td><td>ZIF-8@BHb hybrid composite</td><td>Peroxidase-like (POD)</td><td>H2O2 and phenol detection</td></tr></table>  
+
+Nano Res. 2023, 16, 1992- 2002:  
+
+Despite many Fe- based single atom nanozymes with oxidase- like activity have been reported. Most Fe single atom nanozymes are manufactured by forming an \(\mathrm{Fe - N_x}\) structure, which is achieved by using inorganic metal precursors and nitrogen- doped carbon support under the high- temperature treatments. In this process, the drawback is that Fe can aggregate to block the pores and the active site of the \(\mathrm{Fe - N_x}\) can
+
+<--- Page Split --->
+
+collapse. Moreover, this process generally requires the further acid treatment for the removal of large particles. As catalyst activity decreases of the active center and mass transfer properties, the Fe single atom nanozymes typically exhibit lower performance. To solve these issues, we demonstrated a bioinspired synthetic strategy for the massive preparation of highly active porous three- coordinated Fe single- atom nanozymes (pFeSAN) with oxidase- like activity by using biomineralized hemoglobin- containing ZIF as pyrolytic templates. Hemoglobin is an iron protein that contains nitrogen and metals that are useful as an ideal template and Fe source for bio- inspired Fe single- atom nanozyme. The iron- porphyrin contained in the hemoglobin plays an important role in forming the Fe- \(\mathrm{N_x}\) active sites during the heat treatment process. Compared to previous studies, pFeSAN with Fe- \(\mathrm{N_3}\) coordination and mesoporous structure increased the substrate transfer and thereby exhibited an outstanding oxidase- like activity.  
+
+The paper (Nano Res. 2023, 16, 1992- 2002) is a review article, which summarizes the recent progress of single- atom nanozymes in biomedicine. In this Review article, there is no general strategy to simultaneously achieve atomically dispersed, mesoporous structure and a well- regulated single- atom coordination environment (Table R2). Moreover, the mechanisms of \(\mathrm{O_2}\) activation and electrons transfer in the oxidase- like reaction of Fe- based single atom nanozymes are not particularly clear. Therefore, compared to other Fe- based single atom nanozymes, pFeSAN via one- step synthesis has the advantages of facile preparation, atomically dispersed Fe single atoms, mesoporous structure, Fe- \(\mathrm{N_3}\) coordination and excellent oxidase- like activity, making it a promising nanozyme for GSH detection and other biological applications. To further clarify the reaction mechanism of the pFeSAN, we investigated the \(\mathrm{O_2}\) activation and electron- transfer by EPR, DFT and electrochemical analysis. We believe this study will inspire other high- performance Fe- based single atom nanozymes development.
+
+<--- Page Split --->
+
+
+Table R2. Comparison of pFeSAN and other Fe single-atom nanozymes.   
+
+<table><tr><td>Fe source</td><td>Mespor <br>ous</td><td>Template</td><td>Aftertre <br>atment</td><td>Mimic <br>function</td><td>Applications</td><td>Ref</td></tr><tr><td>Hb</td><td>Yes</td><td>Hb</td><td>None</td><td>OXD</td><td>GSH <br>detection</td><td>Our work</td></tr><tr><td>FePc</td><td>None</td><td>None</td><td>HCl</td><td>OXD</td><td>Antibacterial</td><td>Sci. Adv., 2019, 5, <br>eaav5490.</td></tr><tr><td>(NH4)2Fe(SO4)2</td><td>Yes</td><td>F127</td><td>PEG</td><td>POD/CAT</td><td>Anti-tumor</td><td>Biomaterials, 2022, <br>281, 121325.</td></tr><tr><td>Fe(acac)3</td><td>None</td><td>None</td><td>H2SO4</td><td>OXD/POD</td><td>Anti-tumor</td><td>ACS Nano, 2022, <br>16, 1, 855</td></tr><tr><td>Fe(acac)3</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>Antibacterial</td><td>Small, 2019, 15, <br>e1901834.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>MgO</td><td>HNO3</td><td>POD</td><td>Glucose <br>detection</td><td>Small, 2020, <br>16, e2002343.</td></tr><tr><td>Fe(OAc)2</td><td>None</td><td>MgO</td><td>HNO3</td><td>POD</td><td>Osteosarcoma <br>treatment</td><td>Adv. Mater., 2021, <br>33, e2100150.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>SiO2</td><td>NaOH</td><td>POD</td><td>Anti-tumor</td><td>Adv. Mater., 2022, <br>34, e2107088.</td></tr><tr><td>FePc</td><td>None</td><td>None</td><td>H2SO4</td><td>CAT/SOD</td><td>Anti-oxidant</td><td>Chem. Commun., <br>2019, 55, 159.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>Butyrylcholine <br>esterase <br>detection</td><td>Biosens. <br>Bioelectron., 2019, <br>142, 111495.</td></tr><tr><td>FeCl2</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>H2O2 <br>detection</td><td>Anal. Chem., 2019, <br>91, 11994.</td></tr><tr><td>FeCl3</td><td>None</td><td>SiO2</td><td>HF</td><td>POD</td><td>Acetylcholine <br>sterase <br>detection</td><td>Small, 2019, 15, <br>e1903108.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>None</td><td>None</td><td>POD</td><td>Acetylcholine <br>sterase <br>detection</td><td>ACS Nano, 2022, <br>16, 2, 2997.</td></tr><tr><td>Fe(OAc)2</td><td>None</td><td>None</td><td>H2SO4</td><td>OXD</td><td>GSH <br>detection</td><td>Chem. Commun., <br>2019, 55, 5271.</td></tr></table>  
+
+Chem. Commun. 2019, 55, 5271- 5274:  
+
+In this work, authors suggest that the \(\cdot \mathrm{O}_2^-\) radical is the main active species in the oxidase- like reaction of Fe- N/C- CNTs. However, ROS were undetectable by using either ROS quenchers and electron paramagnetic resonance spectra in our work
+
+<--- Page Split --->
+
+(Supplementary Fig. 32). These results confirmed that pFeSAN mediated the complete reduction of \(\mathrm{O_2}\) to \(\mathrm{H_2O}\) without releasing free ROS, indicating that the reaction mechanisms of the structure-dependent oxidase-like activity of Fe- N/C- CNTs and pFeSAN were very different. Meanwhile, the mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of oxygen into \(\mathrm{H_2O}\) , identical to that of natural \(CCO\) . We also verified that pFeSAN likely followed an oxygen atom transfer mechanism similar to that of the natural \(CO\) by the \(\mathrm{Fe(IV) = O}\) intermediate, different from majority of previously reported oxidase mimics.  
+
+Table R3. Comparison of pFeSAN and Fe-N3/C.   
+
+<table><tr><td></td><td>Catalyst</td><td>Acid treatment</td><td>Free ROS</td></tr><tr><td>Our work</td><td>Mesoporous FeN3</td><td>Not required</td><td>None</td></tr><tr><td>Chem. Commun., 2019, 55, 5271.</td><td>Fe-N3/C</td><td>H2SO4 (1 M), 5 h</td><td>·O2-</td></tr></table>  
+
+Overall, our study is a new synthetic strategy by employing the biomineralized hemoglobin@ZIF- 8 as sacrificial templates and Fe sources to prepare the atomically dispersed Fe single- atom nanozyme (featured by Fe- N3 coordination) within the mesopores of carbon support. Benefiting from the simple and facile one- step Hb- templated strategy, pFeSAN shows six key advantages (Response Letter: Page R13- R14) over the conventional FeSAN in terms of catalyst structures, performance of oxidase- like, and GSH detection, as mentioned above as well as in the revised manuscript. We hope that we illustrate the novelty of our investigations as well as distinguish the present study from previous investigations.  
+
+(2) Single atom nanozymes for the detection of glutathione (GSH). Nanozymes and Fe-N-C-based single atom nanozymes have been used extensively for the detection of glutathione (Journal of Materiomics, 2022, 8, 1251-1259). \(\mathrm{Mn_3O_4}\) microspheres have been used as an oxidase mimic for rapid detection of glutathione (RSC Adv., 2019,9, 16509-16514). Light-responsive MOF as an oxidase mimic for cellular GSH detection
+
+<--- Page Split --->
+
+(Anal. Chem. 2019, 91, 13, 8170–8175). MnO₂ nanosheets as an artificial enzyme to mimic oxidase for rapid and sensitive detection of glutathione (Biosensors & Bioelectronics, 2017, 90, 69- 74). Therefore, the present study lacks novelty in the detection of glutathione.  
+
+## Response:  
+
+Thank you for this comment. Complete surgical resection is the ideal first- line treatment for most malignancies. Compared with biological normal cells, the concentration of GSH in cancer cells is considerably greater, reaching \(0.5 - 1.0 \text{mM}\) , which is about 1000 times that of normal cells, making it one of the most significant signal molecules to diagnose cancer. Therefore, the goal of complete surgical resection would be facilitated by GSH imaging that enables more precise visualization of tumor margins. However, due to its relatively high concentration, a high- performance detection system for GSH analysis is highly desired to provide a rapid localization of tumor area via GSH visualization detection. Unfortunately, the direct detection of GSH in the mM range has been recognized as a difficult challenge. The concentrations of GSH in tumor tissue are usually higher than the upper limit of GSH detection for those previously reported nanozymes (Table R4). Hence, the detection range of these above- mentioned nanozymes is too narrow to detect cell and tissue samples directly, as shown in Table R4. To address this challenge, our work focused on developing a high- performance detection system for the GSH detection using biomineralized hemoglobin@ZIF- 8 as sacrificial templates and Fe sources to prepare highly dispersed and FeN₃- coordinated single- atoms nanozyme. Based on these properties, a colorimetric method was developed to detect GSH, which presented a wide linear detection range of \(50 \text{nM} - 1.0 \text{mM}\) for GSH and successfully employed as a colorimetric probe for GSH visualization in tumor tissues.
+
+<--- Page Split --->
+
+
+Table R4. Comparison of the pFeSAN with other nanozymes for GSH detection.   
+
+<table><tr><td>Materials</td><td>Linear range (μM)</td><td>LOD (μM)</td><td>Reference</td></tr><tr><td>pFeSAN</td><td>0.05-1000</td><td>0.0024</td><td>This work</td></tr><tr><td>Fe-N-C SANs</td><td>100-400</td><td>78.3</td><td>J. Materiomics, 2022, 8, 1251.</td></tr><tr><td>MnO2</td><td>1-25</td><td>0.3</td><td>Biosens. Bioelectron., 2017, 90, 69.</td></tr><tr><td>PSMOF</td><td>0-40</td><td>0.68</td><td>Anal. Chem., 2019, 91, 8170.</td></tr><tr><td>Mn3O4</td><td>50-60</td><td>0.889</td><td>RSC Adv., 2019, 9, 16509.</td></tr></table>  
+
+In our work, the exceptionally high detection upper limit of pFeSAN is 2.5 to 40 times higher than that of these conventional nanozymes, making it suitable for detecting GSH with high levels (Table R4). The excellent detection performance of GSH by the pFeSAN originates from its synergistic advantages comprising highly dispersed metal single- atoms, the presence of mesopores, and well- regulated coordination environments of the single- atoms. We further utilize pFeSAN- DAB system as a GSH sensor to realize in vitro quantitative GSH visualization for Hep 1- 6 cells and tumor tissue with high GSH states have been accurately distinguished by visualization detection.  
+
+Besides, the comparison illustrated the dramatical difference of the pFeSAN from the previously reported nanozymes, which could be attributed to the unique structural features of pFeSAN and an enzyme- like catalytic pathway of the four- electron \(\mathrm{O}_2\) - to- \(\mathrm{H}_2\mathrm{O}\) reduction. The outstanding performance, unique structures and the natural enzyme- like catalytic pathway of pFeSAN indeed reflect the novelty of the dedicatedly designed catalysts for GSH detection, which serve as a real- time, facile, rapid ( \(\sim 6\) min) and precise visualization analysis methodology of tumors and shows its potential for diagnostic and clinic applications.
+
+<--- Page Split --->
+
+(3) The concept of using GSH as biomarker for visualization of cancer cells is not entirely new. There are many recent reports in the literature which highlight the concept. For example, a MOF has been reported to exhibit oxidase-like activity by oxidizing TMB to oxTMB, which has been used as a colorimetric probe for GSH detection. The oxidase mimic has been used to analyze the GSH level in the lysates of normal and cancer cells (Anal. Chem. 2019, 91, 8170–8175). There are many other reports which describe the use of nanozymes for tumor visualization through GSH detection.  
+
+## Response:  
+
+Thank you very much for pointing this out. The author developed a light- dependent metal–organic framework with oxidase- like activity and colorimetric detection of cancer cells through GSH detection after cell lysis treatment (Ref. 53: Anal. Chem. 2019, 91, 8170–8175). In our work, pFeSAN was applied to monitor the GSH levels in normal and cancer cells without additional requirements of light irradiation and complex pretreatment of cells. Importantly, methods for the visualization detection of tumor tissue though GSH has not been reported yet, which brought the possibility to realize specific detection in practical applications. The visualized analysis of the GSH in tumor tissue is gaining interests to improve surgical safety and to promote surgical therapeutic effects.  
+
+Table R5. Comparison of the pFeSAN with PSMOF for GSH detection.   
+
+<table><tr><td></td><td>Catalyst</td><td>Light dependent</td><td>Cell pretreatment</td><td>Intratumoral GSH detection</td></tr><tr><td>Our work</td><td>pFeSAN</td><td>Not required</td><td>Without pretreatment</td><td>Yes</td></tr><tr><td>Anal. Chem. 2019, 91, 8170.</td><td>PSMOF</td><td>300 W Xe lamp</td><td>Cell lysis</td><td>None</td></tr></table>
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+This paper presents an innovative biomimetic synthetic strategy for the synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using natural enzyme as a template, and the nanozyme exhibits highly efficient oxidase- like activity and potential applications in diagnostics. This research has value for the researchers in the related areas. A thorough, point- by- point response to each point raised by reviewers has been made. So I think this work can be accepted for publication in Nature Communications.  
+
+Reviewer #2:  
+
+Remarks to the Author:  
+
+Authors have improved the manuscript according to the comments, and thus I recommend it publication.  
+
+Reviewer #4:  
+
+Remarks to the Author:  
+
+[Note from the editor: Reviewer #4 was invited to assess the response given to Reviewer #3]  
+
+In the manuscript "Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione" rebuttal letter, the authors responded to the three questions raised by the reviewer 3 with lots of discussion. However, it might still not fully address the concerns about the novelty mentioned there.  
+
+Author response: (2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants / products during the catalytic process.  
+
+Comments: The increase in the transfer of reactants / products usually related to Km value of the enzyme. However, in the manuscript, the authors showed that the pFeSAN showed a really small increase in binding affinity compared to the Fe- N4 they mentioned. This might need further clarification or more thoughts.  
+
+Author response: (3) pFeSAN delivered outstanding oxidase- like activity, which was 3.3- and 8791- times higher than those of Fe- N4 and Fe3O4 nanozymes, respectively.  
+
+Comments: The oxidase- like activity of the nanozymes were compared using activity ratio which is a not standardized item. How that is calculated and how to use that value to cross- compare with other oxidase- like nanozyme require further clarifications.  
+
+Author response: (4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four- electron reduction of O2 into H2O, being identical to that of CcO, which was very different from the majority of the previously reported oxidase- like nanozymes with the generation of reactive oxygen species (ROS).  
+
+Comments: This finding is not novel. As the nanozymes prepared by the authors only showed OxD- like activity, it is expected that there will be no ROS generated. (Similar to Sci. Adv., 2019, 5, eaav5490). If the authors stated it is a similar catalytic pathway to CcO, is there any of the reaction intermediates or transient species the authors trapped could be compared to CcO?  
+
+Author response: (5) pFeSAN as a highly- performed nanozyme exhibited a much higher upper detection limit of GSH at 1 mM, which was 2.5 to 40- fold higher than those of the previously reported investigations (Table R4).  
+
+Comments: Previously, other systems based on nanozyme activity detected similar or even higher range. (New J. Chem., 2022, 46, 10682- 10689)
+
+<--- Page Split --->
+
+## Point-by-point Response to Reviewers' Comments  
+
+Reviewer #1 (Remarks to the Author):  
+
+This paper presents an innovative biomimetic synthetic strategy for the synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using natural enzyme as a template, and the nanozyme exhibits highly efficient oxidase- like activity and potential applications in diagnostics. This research has value for the researchers in the related areas. A thorough, point- by- point response to each point raised by reviewers has been made. So I think this work can be accepted for publication in Nature Communications.  
+
+## Response:  
+
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.  
+
+Reviewer #2 (Remarks to the Author):  
+
+Authors have improved the manuscript according to the comments, and thus I recommend it publication.  
+
+## Response:  
+
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.
+
+<--- Page Split --->
+
+Reviewer #4 (Remarks to the Author):  
+
+[Note from the editor: Reviewer #4 was invited to assess the response given to Reviewer #3]  
+
+In the manuscript "Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione" rebuttal letter, the authors responded to the three questions raised by the reviewer 3 with lots of discussion. However, it might still not fully address the concerns about the novelty mentioned there.  
+
+We really appreciate the Reviewer's useful comments and suggestions. We have carefully revised the manuscript based on his/her comments.  
+
+Author response: (2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3- 4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants / products during the catalytic process.  
+
+Comments: The increase in the transfer of reactants / products usually related to Km value of the enzyme. However, in the manuscript, the authors showed that the pFeSAN showed a really small increase in binding affinity compared to the Fe- N4 they mentioned. This might need further clarification or more thoughts.  
+
+## Response:  
+
+Thanks for the Reviewer's kind suggestions, which are valuable for improving the accuracy of the manuscript.  
+
+For analyzing the catalytic mechanism and acquiring kinetic parameters, the oxidase- like activities of pFeSAN and Fe- N4 under the same conditions were studied by enzyme kinetics theory and methods. With the Lineweaver- Burk equation, the important
+
+<--- Page Split --->
+
+enzyme kinetic parameters such as Michaelis-Menten constant \(\mathrm{(K_m)}\) , maximal velocity \(\mathrm{(V_{max})}\) , catalytic constant \(\mathrm{(K_{cat})}\) and \(\mathrm{K_{cat} / K_m}\) were presented in Table R1 (Adv. Mater., 2022, 34, e2201736; Nat. Protoc., 2018, 13, 1506- 1520. ). \(\mathrm{K_m}\) was identified as an indicator of enzyme affinity to substrates. Smaller \(\mathrm{K_m}\) values thus indicate a stronger affinity between the enzyme and the substrate. \(\mathrm{V_{max}}\) represents the reaction rate when the enzyme is saturated with substrate, and a higher \(\mathrm{V_{max}}\) value indicates a quicker reaction rate. The \(\mathrm{K_{cat}}\) value gave a direct measure of the enzymatic catalytic activity. Generally, the maximal velocity of reaction \(\mathrm{V_{max}}\) and \(\mathrm{K_{cat}}\) reveals the catalytic activity of enzyme. Increasing the transfer of chemicals not only increase the decrease of \(\mathrm{Km}\) but also increase the diffusion of chemicals, leading to the overall enhanced catalytic kinetics. Besides, \(\mathrm{K_{cat} / K_m}\) is known to be a descriptor of enzyme efficiency and a better indicator to compare two enzymes. The higher catalytic efficiency of pFeSAN signifies more substrate- to- product conversion, which is happening due to its large affinity (low \(\mathrm{K_m}\) ) for TMB and greater proportion of bound substrate conversion to product before its dissociation (large turnover \(\mathrm{K_{cat}}\) ).  
+
+Table R1. Comparison of kinetics for pFeSAN and Fe-N4.   
+
+<table><tr><td>Catalyst</td><td>Km (mM)</td><td>Vmax (μM s-1)</td><td>Kcat (s-1)</td><td>Kcat/Km (mM-1s-1)</td></tr><tr><td>pFeSAN</td><td>0.17</td><td>1.67</td><td>2.6×106</td><td>1.53×107</td></tr><tr><td>Fe-N4</td><td>0.29</td><td>0.036</td><td>9.6×104</td><td>3.31×105</td></tr></table>  
+
+From Table R1, the \(\mathrm{K_m}\) value for pFeSAN (0.17 mM) to the TMB was about \(58\%\) of that for Fe- N4 (0.29 mM), indicating that it has higher affinity with TMB and lower concentration of TMB required to reach the maximal activity of \(\mathrm{V_{max}}\) . Hence, the \(\mathrm{V_{max}}\) and \(\mathrm{K_{cat}}\) values of pFeSAN for TMB showed 46.4- fold and 27.1- fold increases relative to Fe- N4, verifying the mesoporous structure in improving the oxidase- like performance. Meanwhile, the catalytic efficiency \(\mathrm{(K_{cat} / K_m)}\) of pFeSAN (1.53×107 mM-1s-1) is 46.2- fold higher than that of Fe- N4 (3.31×105 mM-1s-1). Overall, all these kinetic parameters including Michaelis- Menten constant \(\mathrm{(K_m)}\) , maximal reaction velocity
+
+<--- Page Split --->
+
+\((\mathrm{V}_{\mathrm{max}})\) , catalytic rate constant \((\mathrm{K}_{\mathrm{cat}})\) , and \(\mathrm{K}_{\mathrm{cat}} / \mathrm{K}_{\mathrm{m}}\) of pFeSAN showed the optimum values than Fe- \(\mathrm{N}_4\) , indicating a distinct positive contribution of the mesoporous structure (3\~4 nm) and higher surface area to the oxidase- like activity of the pFeSAN (Figure R1). The larger pore size can make a great contribution to fast mass transfer.  
+
+![](images/Figure_unknown_7.jpg)
+
+<center>Figure R1. a. Pore size distribution curves of Fe- \(\mathrm{N}_4\) and pFeSAN b. BET surface areas of Fe- \(\mathrm{N}_4\) and pFeSAN. </center>  
+
+In conclusion, we demonstrated a biomimetic synthetic strategy for scalable synthesis of porous Fe single- atom nanozymes using hemoglobin as both template and Fe- source, which delivered a high oxidase- like activity. The mesoporous features of pFeSAN significantly promoted mass transport and maximumly exposed active iron sites during reaction, which could greatly enhance the oxidase- like activity of pFeSAN. The relative information has been updated in the revised manuscript (Page 16, Line 19–22; Page 17, Line 1–5 and Supplementary Table 2).  
+
+Author response: (3) pFeSAN delivered outstanding oxidase- like activity, which was 3.3- and 8791- times higher than those of Fe- \(\mathrm{N}_4\) and \(\mathrm{Fe}_3\mathrm{O}_4\) nanozymes, respectively. Comments: The oxidase- like activity of the nanozymes were compared using activity ratio which is a not standardized item. How that is calculated and how to use that value to cross- compare with other oxidase- like nanozyme require further clarifications.  
+
+## Response:  
+
+We would like to express our point for your valuable suggestions regarding our work.
+
+<--- Page Split --->
+
+In response, we want to clarify that we initially utilized a generic method to calculate the activity of nanozymes in the main text. We recognized the importance of consistency in comparing the activity of different nanozymes.  
+
+We calculated the oxidase- like activities of pFeSAN, Fe- N4 and \(\mathrm{Fe_3O_4}\) according to the standardized assay protocol (Nat. Catal. 2021, 4, 407- 417; Nat. Protoc. 2018, 13, 1506- 1520). By quantitatively determined the specific activity values (U/mg) of pFeSAN, Fe- N4 and \(\mathrm{Fe_3O_4}\) by measuring the absorption intensity of the nanozyme- catalyzed TMB colorimetric reactions (Figure R2). The specific activity of pFeSAN was determined to be \(593~\mathrm{U / mg}\) , which is 3.5 times higher than that of Fe- N4 (169 U/mg) and 8471 times higher than that of \(\mathrm{Fe_3O_4}\) (0.07 U/mg). The high specific activity of pFeSAN following factors: Firstly, the mesoporous structure (3- 4 nm) of pFeSAN exhibits larger surface area (705.8 \(\mathrm{m^2 / g}\) ) than that of Fe- N4 (561.6 \(\mathrm{m^2 / g}\) ), and the large specific surface area can help fast mass transfer. Secondly, the pFeSAN exposes more Fe active sites, greatly enhance its oxidase- like activity.  
+
+![](images/Figure_unknown_8.jpg)
+
+<center>Figure R2. Reaction-time curves of the TMB colorimetric reaction catalyzed by pFeSAN, Fe-N4 and \(\mathrm{Fe_3O_4}\) . </center>  
+
+The calculation part of the specific activity is as follows:  
+
+Calculate the nanozyme activity (units) using the following equation:  
+
+\[\mathrm{b_{nanozyme} = V / (\epsilon\times 1)\times(\Delta A / \Delta t)}\]  
+
+where \(\mathrm{b_{nanozyme}}\) is the catalytic activity of nanozyme expressed in units. One unit is
+
+<--- Page Split --->
+
+defined as the amount of nanozyme that catalytically produces \(1\mu \mathrm{mol}\) of product per min at room temperature; V is the total volume of reaction solution \((\mu \mathrm{L})\) ; \(\epsilon\) is the molar absorption coefficient of the colorimetric substrate, which is maximized at \(39,000\mathrm{M}^{- 1}\) \(\mathrm{cm}^{- 1}\) at \(652\mathrm{nm}\) for TMB; I is the path length of light traveling in the cuvette (cm); A is the absorbance after subtraction of the blank value; and \(\Delta \mathrm{A} / \Delta \mathrm{t}\) is the initial rate of change in absorbance at \(652\mathrm{nm}\mathrm{min}^{- 1}\) .  
+
+Calculate the specific activity of the nanozyme (U \(\mathrm{mg}^{- 1}\) ) by  
+
+\[\mathrm{a_{nanozyme}} = \mathrm{b_{nanozyme}} / [\mathrm{m}]\]  
+
+where \(\mathrm{a_{nanozyme}}\) is the specific activity expressed in units per milligram (U \(\mathrm{mg}^{- 1}\) ) nanozymes, and [m] is the nanozyme weight (mg) of each assay. In the revised manuscript, we have updated these data according to the Reviewer's suggestion (Page 14, Fig. 4f; Page 16, Line 1- 7; Page 33, line 3- 16).  
+
+Author response: (4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four- electron reduction of \(\mathrm{O}_2\) into \(\mathrm{H}_2\mathrm{O}\) , being identical to that of \(\mathrm{CcO}\) , which was very different from the majority of the previously reported oxidase- like nanozymes with the generation of reactive oxygen species (ROS).  
+
+Comments: This finding is not novel. As the nanozymes prepared by the authors only showed OXD- like activity, it is expected that there will be no ROS generated. (Similar to Sci. Adv., 2019, 5, eaav5490). If the authors stated it is a similar catalytic pathway to \(\mathrm{CcO}\) , is there any of the reaction intermediates or transient species the authors trapped could be compared to \(\mathrm{CcO}\) ?  
+
+## Response:  
+
+Thank you for your valuable comments. Previous studies demonstrated that the reactive intermediate of \(\mathrm{Fe(IV) = O}\) was very important for the catalytic oxidative reactions of natural \(\mathrm{CcO}\) (Chem. Rev. 2018, 118, 2491- 2553). As the common oxidant, the intermediate of \(\mathrm{Fe(IV) = O}\) , which usually presents in the catalytic cycle of natural
+
+<--- Page Split --->
+
+oxidases, is considered as the active transient state. To verify the presence of the \(\mathrm{Fe(IV) = O}\) intermediate of \(\mathrm{O_2}\) activation process by pFeSAN, the electron paramagnetic resonance spectrum of the pFeSAN- enabled oxidation with excessive phenyloxoiodine was recorded at 77K (Figure R3). A typical diamond- shaped sign signal at \(\mathrm{g} = 2.03\) , consistent with \(\eta^2\) - peroxy heme species, indicated the formation of \(\mathrm{Fe(IV) = O}\) intermediate in pFeSAN for oxidation (Fig. 5f, Sci. Adv., 2019, 5, eaav5490). Therefore, the oxidase- like activity of pFeSAN proceeds through the \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) pathway, and similar to the reaction process with \(\mathrm{CcO}\) .  
+
+![](images/Figure_unknown_9.jpg)
+
+<center>Figure R3. EPR spectra of pFeSAN in the presence of phenyloxoiodine at 77 K. </center>  
+
+Previous study (Sci. Adv., 2019, 5, eaav5490) demonstrated that the \(\mathrm{FeN_5}\) SA/CNF showed oxidase- like activity and no generation of ROS. However, the experimental verification was lacking in that study. In our work, to further explore the electron transfer path during the oxidation, the rotating ring- disk electrode (RRDE) tests were performed. Figure R4 showed that the \(\mathrm{H_2O_2}\) yield of pFeSAN remained below \(7.5\%\) over a wide potential range of \(0.1 - 0.8\mathrm{V}\) . Derived from the RRDE test, the average electron transfer number (n) of pFeSAN was 3.7, indicating the oxygen activation on the pFeSAN through a four- electron oxygen reduction reaction pathway (Nano Energy, 2021, 83, 105798). This process requires four \(\mathrm{H^{+}}\) and four electrons \((\mathrm{O_2 + 4H^{+} + 4e^{- }}\) \(\rightarrow 2\mathrm{H_2O})\) for the complete \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) reduction. Thus, the electrochemical understandings of these stepwise proton and electron transfers reveal the essence of pFeSAN for its oxidase- like performance. In our work, the exploration of oxidase- like
+
+<--- Page Split --->
+
+reaction by the RRDE tests would help us to understand the mechanism of oxidase- like reactions.  
+
+![PLACEHOLDER_31_0]
+
+<center>Figure R4. Calculated electron transfer number derived from rotating ring-disk electrode and \(\mathrm{H}_2\mathrm{O}_2\) yields of the pFeSAN. </center>  
+
+In the revised manuscript, we have highlighted the related information in the main context (Page 20, Line 1- 9).  
+
+Author response: (5) pFeSAN as a highly- performed nanozyme exhibited a much higher upper detection limit of GSH at \(1\mathrm{mM}\) , which was 2.5 to 40- fold higher than those of the previously reported investigations (Table R4).  
+
+Comments: Previously, other systems based on nanozyme activity detected similar or even higher range. (New J. Chem., 2022, 46, 10682- 10689)  
+
+## Response:  
+
+Thank you for your evaluation. In this GSH detection system (New J. Chem., 2022, 46, 10682- 10689), a novel nanozyme based on the ultrathin two- dimensional metal- organic framework nanomaterial D- ZIF- 67 was prepared and characterized. D- ZIF- 67 exhibited significant oxidase- like activity due to the large specific surface area of the two- dimensional sheet structure as well as the large number of active sites exposed compared to crystalline MOFs. By taking advantage of the excellent property of D- ZIF- 67, authors constructed an effective and sensitive colorimetric sensor for visual GSH
+
+<--- Page Split --->
+
+detection, the detected GSH concentrations ranged between 0.5–10 μM.  
+
+In our work, pFeSAN provided a wider detection range of 50 nM to 1 mM, higher than those previous literatures (Table R2). The excellent detection performance of GSH by the pFeSAN originates from its synergistic advantages comprising highly dispersed metal single- atoms, the presence of mesopores, and well- regulated coordination environments of the single- atoms. We further utilize pFeSAN- DAB system as a GSH sensor to realize in vitro quantitative GSH visualization for Hep 1- 6 cells and tumor tissue with high GSH states have been accurately distinguished by visualization detection. Notably, methods for the visualization detection of tumor tissue though GSH has not been reported yet, which brought the possibility to realize specific detection in practical applications. The visualized analysis of the GSH in tumor tissue is gaining interests to improve surgical safety and to promote surgical therapeutic effects. We have checked the literature carefully and added this literature in the revised manuscript to support this work (Ref. 57: New J. Chem., 46, 10682–10689 (2022)).  
+
+Table R2. Comparison of our approach with other colorimetric detection systems of GSH.   
+
+<table><tr><td>Materials</td><td>Linear range</td><td>LOD</td></tr><tr><td>pFeSAN</td><td>0.05-1000 μM</td><td>0.0024 μM</td></tr><tr><td>AuNPs</td><td>1-40 μM</td><td>0.013 μM</td></tr><tr><td>Au nanoclusters</td><td>2-25 μM</td><td>0.42 μM</td></tr><tr><td>PSMOF</td><td>1-20 μM</td><td>0.68 μM</td></tr><tr><td>Acre+-Mes</td><td>0.1-40 μM</td><td>0.1 μM</td></tr><tr><td>MnO2</td><td>0.3-15 μM</td><td>0.11 μM</td></tr><tr><td>TiO2/MoS2</td><td>0.05-1 μM</td><td>0.05 μM</td></tr><tr><td>Fe-N-C SANs</td><td>100-400 μM</td><td>78.3 μM</td></tr></table>
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #4:  Remarks to the Author:  The authors have addressed the comments. There is no further comment.
+
+<--- Page Split --->
+
+## Point-by-point Response to Reviewers' Comments  
+
+Reviewer #4 (Remarks to the Author):  
+
+The authors have addressed the comments. There is no further comment.  
+
+## Response:  
+
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1ec0e20dfe1397bcc4c3b75dcfb56d1f19c1e910da1612eea66abbbed7b5cf35/supplementary_0_Peer Review File__1ec0e20dfe1397bcc4c3b75dcfb56d1f19c1e910da1612eea66abbbed7b5cf35_det.mmd

@@ -0,0 +1,732 @@
+<|ref|>title<|/ref|><|det|>[[61, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[70, 110, 362, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[70, 154, 800, 241]]<|/det|>
+Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione  
+
+<|ref|>image<|/ref|><|det|>[[57, 732, 240, 780]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 911, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to  
+
+<|ref|>text<|/ref|><|det|>[[57, 785, 934, 924]]<|/det|>
+the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[119, 85, 294, 98]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[119, 112, 223, 125]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[119, 127, 300, 140]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[118, 140, 875, 280]]<|/det|>
+This paper presents an innovative biomimetic synthetic strategy for the scalable synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using hemoglobin as a template. The authors demonstrate the superior catalytic activity of pFeSAN compared to Fe- N4 and Fe3O4 nanozymes, attributing it to the suppressed aggregation of atomically dispersed Fe, facilitated mass transfer, and maximized exposure of active sites. Additionally, pFeSAN was successfully applied in the rapid colorimetric detection of glutathione and developed as a real- time, facile, rapid, and precise visualization analysis methodology for tumors via glutathione levels. Overall, this paper is a valuable contribution to the field of nanomaterials with potential applications in diagnostics and therapeutics. Therefore, I recommend it can be accepted by Nature Communications after revision if some problems can be properly addressed:  
+
+<|ref|>text<|/ref|><|det|>[[118, 293, 875, 350]]<|/det|>
+1. Figure 5f: the EPR patterns of the high-valent iron-oxygen species formed by SACs with different coordination numbers (Fe-Nx) should be different. In this manuscript, the number of N atoms coordinated with Fe is 3. Why is the EPR signal of the high-valent Fe oxygen species formed by FeN3 the same as the signal formed by FeN5 (Refs. 21 and 23)?  
+
+<|ref|>text<|/ref|><|det|>[[118, 350, 860, 392]]<|/det|>
+2. Page 10, in the XANES results, it should be "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe2O3 and Fe foil (not FePC), and very close to that of FePC.".  
+
+<|ref|>text<|/ref|><|det|>[[118, 393, 875, 490]]<|/det|>
+3. Page 11, SACs can be regarded as a complex (coordination compound). The chemical configuration (such as the coordination number, the chemical environment of the coordination atom, etc.) of the active site in SACs will change the electronic structure of the metal by affecting the splitting energy and stability energy of the central d orbital of the metal. Why the author judged that the electronic structure around Fe in FeN3 is similar to that in FePC through the partial overlap of absorption edges? In addition, the author did not comprehensively consider the Edge front peak and white line peak.  
+
+<|ref|>text<|/ref|><|det|>[[118, 490, 820, 519]]<|/det|>
+4. Compared with R-space, the data in k-space can better reflect the quality of synchrotron radiation data. So, the real data and fitting data in k-space should be included.  
+
+<|ref|>text<|/ref|><|det|>[[118, 519, 860, 590]]<|/det|>
+5. For R-space, the data after \(4\mathrm{\AA}\) is meaningless. In order to better show the R-space data and the situation, it is recommended to shorten the Abscissa to less than \(6\mathrm{\AA}\) (Figures 3b and 3c). In addition, the Abscissa of R-space should be R \((\mathrm{\AA})\) only after the phase shift is corrected. I am wondering whether no phase shift correction was performed for the data in the figure, if it is, the Abscissa should be \(\mathrm{R} + \mathrm{a}(\mathrm{\AA})\) , please check it.  
+
+<|ref|>text<|/ref|><|det|>[[118, 590, 867, 618]]<|/det|>
+6. It is mentioned by the author that Hb is competitive in price. But the price of Hb (US \(\)3.60\(/g,\) Shanghai Yuanye, S12021-5g) is significantly higher than that of FeCl3.  
+
+<|ref|>text<|/ref|><|det|>[[118, 618, 866, 646]]<|/det|>
+7. Lack of comparison, compared with other nano-enzymes with oxidase-like activity, what is the advantage of pFeSAN in the kinetic constant?  
+
+<|ref|>text<|/ref|><|det|>[[118, 646, 870, 702]]<|/det|>
+8. For Figure 5a and 5b, the author pointed out that "no apparently detectable signals of any ROS further proved that ROSs were not the main intermediates for the oxidase-like activity of pFeSAN, which was very different from the majority of the previously reported nanozymes", please give more discussion on this phenomenon.  
+
+<|ref|>text<|/ref|><|det|>[[119, 702, 437, 716]]<|/det|>
+9. For Figure 2f, the resolution is too low.  
+
+<|ref|>text<|/ref|><|det|>[[119, 757, 222, 770]]<|/det|>
+Reviewer #2:  
+
+<|ref|>text<|/ref|><|det|>[[119, 772, 300, 784]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[118, 785, 875, 885]]<|/det|>
+Due to the high atom utilization and remarkable catalytic activity, single- atom catalysts exhibit great potentials for applications of nanozyme. In this work, a porous Fe single- atom catalyst prepared by a biomineralization- pyrolysis strategy exhibited remarkable oxidase- like activity, which was used for GSH sensing and cancer cell identification. The porous Fe single- atom nanozyme was well characterized. And, the mechanism of oxidase- like activity was investigated as well, which opened up a new way to design and optimize nanozymes. Therefore, this manuscript can be accepted to be published by Nature Communications after proper revisions.  
+
+<|ref|>text<|/ref|><|det|>[[118, 885, 868, 912]]<|/det|>
+1. Fe single-atom nanozyme reported in the past often exert catalase-like catalytic activity. In this study, dose pFeSAN have catalase-like activity?
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 84, 860, 168]]<|/det|>
+2. Compared with previously reported Fe single-atom nanozyme, what are the advantages of pFeSAN in structure-activity characteristics?  
+3. The label of elements and scale bar at Supplementary Fig. 11 and 15 was not clearly visible.  
+4. Please provide the quantitative analysis of EIS (Fig. 5d).  
+5. In this work, did the pFeSAN show obvious toxicity to normal cells and tumor cells, resulting in the influenced cellular GSH analysis?  
+
+<|ref|>text<|/ref|><|det|>[[118, 210, 222, 222]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[118, 225, 300, 238]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[118, 239, 877, 377]]<|/det|>
+This paper deals with the development of a porous 3- coordinate iron- based nanozyme which can exhibit oxidase- like activity. The authors use hemoglobin as Fe source/template and incorporates it into a zeolitic imidazolate framework (ZIF- 8). The single- atom nanozyme called pFeSAN acts as an oxidase model by oxidizing the colorimetric substrate TMB to oxTMB, which helps in the detection of glutathione (GSH) in nM concentrations. The authors show a significant improvement in the oxidase- like activity; the nanozyme exhibiting 3.3- and 8791- fold higher oxidase- like activity than the Fe- N4 and Fe3O4 nanozymes reported earlier. As the GSH level is altered in many diseases, including cancer, the authors propose that the nanozyme with oxidase- like activity can be used for the visualization of tumor. The work is interesting, but does not warrant publication in Nat. Commun. due to the following reasons.  
+
+<|ref|>text<|/ref|><|det|>[[118, 391, 870, 504]]<|/det|>
+(1) Development of Fe-based single-atom nanozyme by incorporating hemoglobin into ZIF-8: The topic of Fe-based single atom nanozyme is not new. It has been shown that bovine hemoglobin can be incorporated into ZIF-8 to develop single atom nanozymes (ACS Appl. Mater. Interfaces 2016, 8, 29052-29061). This nanozyme has been used as peroxidase mimic to oxidize TMB for the colorimetric detection. Many other Fe-based single atom nanozymes with oxidase-like activity have been reported in a review (Nano Res. 2023, 16, 1992-2002). The mechanism of the structure-dependent oxidase-like activity of Fe-N/C nanozyme has also been reported (Chem. Commun. 2019, 55, 5271-5274).  
+
+<|ref|>text<|/ref|><|det|>[[118, 518, 870, 630]]<|/det|>
+(2) Single atom nanozymes for the detection of glutathione (GSH). Nanozymes and Fe-N-C-based single atom nanozymes have been used extensively for the detection of glutathione (Journal of Materiomics, 2022, 6, 1251-1259). Mn3O4 microspheres have been used as an oxidase mimic for rapid detection of glutathione (RSC Adv., 2019,9, 16509-16514). Light-responsive MOF as an oxidase mimic for cellular GSH detection (Anal. Chem. 2019, 91, 13, 8170-8175). MnO2 nanosheets as an artificial enzyme to mimic oxidase for rapid and sensitive detection of glutathione (Biosensors & Bioelectronics, 2017, 90, 69-74). Therefore, the present study lacks novelty in the detection of glutathione.  
+
+<|ref|>text<|/ref|><|det|>[[118, 644, 873, 742]]<|/det|>
+(3) The concept of using GSH as biomarker for visualization of cancer cells is not entirely new. There are many recent reports in the literature which highlight the concept. For example, a MOF has been reported to exhibit oxidase-like activity by oxidizing TMB to oxTMB, which has been used as a colorimetric probe for GSH detection. The oxidase mimic has been used to analyze the GSH level in the lysates of normal and cancer cells (Anal. Chem. 2019, 91, 8170-8175). There are many other reports which describe the use of nanozymes for tumor visualization through GSH detection.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[275, 90, 721, 108]]<|/det|>
+## Point-by-point Response to Reviewer(s)' Comments  
+
+<|ref|>text<|/ref|><|det|>[[149, 132, 460, 149]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[147, 159, 853, 456]]<|/det|>
+This paper presents an innovative biomimetic synthetic strategy for the scalable synthesis of porous \(\mathrm{Fe - N_3}\) single atom nanozymes (pFeSAN) using hemoglobin as a template. The authors demonstrate the superior catalytic activity of pFeSAN compared to \(\mathrm{Fe - N_4}\) and \(\mathrm{Fe_3O_4}\) nanozymes, attributing it to the suppressed aggregation of atomically dispersed Fe, facilitated mass transfer, and maximized exposure of active sites. Additionally, pFeSAN was successfully applied in the rapid colorimetric detection of glutathione and developed as a real- time, facile, rapid, and precise visualization analysis methodology for tumors via glutathione levels. Overall, this paper is a valuable contribution to the field of nanomaterials with potential applications in diagnostics and therapeutics. Therefore, I recommend it can be accepted by Nature Communications after revision if some problems can be properly addressed:  
+
+<|ref|>text<|/ref|><|det|>[[148, 479, 850, 525]]<|/det|>
+We thank the reviewer for his/her constructive comments on our manuscript. We would like address his/her comments as below.  
+
+<|ref|>text<|/ref|><|det|>[[147, 548, 852, 678]]<|/det|>
+1. Figure 5f: the EPR patterns of the high-valent iron-oxygen species formed by SACs with different coordination numbers \((\mathrm{Fe - N_x})\) should be different. In this manuscript, the number of N atoms coordinated with Fe is 3. Why is the EPR signal of the high-valent Fe oxygen species formed by \(\mathrm{FeN_3}\) the same as the signal formed by \(\mathrm{FeN_5}\) (Refs. 21 and 23)?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 689, 238, 705]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 715, 853, 902]]<|/det|>
+Thank you for this constructive comment. Fe- heme structure of natural cytochrome c oxidase (CcO) plays the core roles for the \(\mathrm{O_2}\) activation through the \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) pathway, delivering high catalytic activity for many oxidation reactions. The catalytic cycle of \(\mathrm{CcO}\) for oxygen reduction starts with the binding of \(\mathrm{O_2}\) on a ferrous heme to afford a \(\mathrm{Fe(IV) = O}\) intermediate, followed by the reduction of \(\mathrm{O_2}\) into \(\mathrm{H_2O}\) (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). The \(\mathrm{Fe(IV) = O}\) is considered as the key intermediate in the catalytic cycles of many heme iron enzymes and heme analogs for oxidation. Although
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 853, 330]]<|/det|>
+the coordination environments of Fe- \(\mathrm{N}_3\) and Fe- \(\mathrm{N}_5\) are not identical, they share the common mechanistic grounds through the \(\mathrm{Fe(IV) = O}\) intermediate (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). In this work, a typical diamond- shaped label signal at \(\mathrm{g} = 2.03\) was detected by EPR, consistent with the \(\eta^2\) - peroxy heme species, signifying the formation of \(\mathrm{Fe(IV) = O}\) intermediate. This suggests that Fe- \(\mathrm{N}_3\) active center of pFeSAN may have a similar reaction process with heme analogs. We suggest that the same oxidase- like reactivities and \(\mathrm{Fe(IV) = O}\) intermediate of Fe- \(\mathrm{N}_3\) and Fe- \(\mathrm{N}_5\) structure are mainly caused by their active site structures, but not the axial ligand. The more information has been added in the revised manuscript (Page 20, Line 3- 6).  
+
+<|ref|>text<|/ref|><|det|>[[147, 366, 853, 440]]<|/det|>
+2. Page 10, in the XANES results, it should be "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe₂O₃ and Fe foil (not FePC), and very close to that of FePC.".  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 451, 238, 467]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 477, 853, 580]]<|/det|>
+Thank you sincerely for your suggestion. In the revised manuscript, we have modified the related information "The rising-edge position of the X-ray absorption near-edge structure spectra (XANES) of pFeSAN was between those of Fe₂O₃ and Fe foil, and very close to that of FePC," in the main context (Page 10, Line 19- 21).  
+
+<|ref|>text<|/ref|><|det|>[[147, 616, 853, 802]]<|/det|>
+3. Page 11, SACs can be regarded as a complex (coordination compound). The chemical configuration (such as the coordination number, the chemical environment of the coordination atom, etc.) of the active site in SACs will change the electronic structure of the metal by affecting the splitting energy and stability energy of the central d orbital of the metal. Why the author judged that the electronic structure around Fe in FeN₃ is similar to that in FePC through the partial overlap of absorption edges? In addition, the author did not comprehensively consider the Edge front peak and white line peak.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 813, 238, 829]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 839, 851, 885]]<|/det|>
+Thank you very much for pointing this out. XANES reveals slight differences between the XANES spectra of pFeSAN and FePc. The latter shows a pre- edge peak at 7118 eV
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[146, 88, 852, 330]]<|/det|>
+assigned to a \(1\mathrm{s} \rightarrow 4\mathrm{p}_z\) shakedown transition characteristic for a square- planar configuration with high \(\mathrm{D_{4h}}\) symmetry (Figure R1). For characteristics of pFeSAN, the pre- edge feature is absent for pFeSAN, revealing a broken \(\mathrm{D_{4h}}\) symmetry (Ref. 38: Nat. Mater., 2015, 14, 937–942). The theoretical XANES was also simulated with FDMNES code (J. Phys.: Condens. Matter, 2009, 21, 345501). The optimized Fe–N3 structure was used as the input structure. The sphere radius to calculate the cluster adsorption was 12 Å. The result showed that the relative pre- edge and white line peaks of the simulated spectrum were consistent with the experiment, validating the Fe–N3 local structure of pFeSAN (Figure R2).  
+
+<|ref|>image<|/ref|><|det|>[[345, 337, 652, 533]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[262, 540, 731, 559]]<|/det|>
+<center>Figure R1. XANES of Fe foil, FePc, \(\mathrm{Fe}_2\mathrm{O}_3\) and pFeSAN. </center>  
+
+<|ref|>image<|/ref|><|det|>[[348, 580, 645, 788]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 808, 850, 881]]<|/det|>
+<center>Figure R2. Comparison of the experimental XANES curve (pFeSAN) with the calculated XANES data of Fe–N3. The insets show the optimized Fe–N3 structure. The relative pre-edge and white line peak of the simulated spectrum are consistent with the experiment. </center>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 84, 849, 122]]<|/det|>
+The relative information has been updated in the revised manuscript (Page11, Line 1–4 and 16–19, and Supplementary Fig. 19).  
+
+<|ref|>text<|/ref|><|det|>[[147, 159, 851, 232]]<|/det|>
+4. Compared with R-space, the data in k-space can better reflect the quality of synchrotron radiation data. So, the real data and fitting data in k-space should be included.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 244, 238, 260]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 270, 852, 371]]<|/det|>
+Thank you for your careful review. The best- fit analysis is shown in Figure R3. The dominant contribution is given by Fe–N first shell coordination. In the revised manuscript, we have added the related information in the main context (Page 11, Line 9–11 and Fig. 3d).  
+
+<|ref|>image<|/ref|><|det|>[[370, 384, 614, 555]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[163, 563, 831, 582]]<|/det|>
+<center>Figure R3. EXAFS fitting results for Fe foil, FePc, Fe<sub>2</sub>O<sub>3</sub> and pFeSAN at k-space. </center>  
+
+<|ref|>text<|/ref|><|det|>[[147, 612, 852, 770]]<|/det|>
+5. For R-space, the data after 4 Å is meaningless. In order to better show the R-space data and the situation, it is recommended to shorten the Abscissa to less than 6 Å (Figures 3b and 3c). In addition, the Abscissa of R-space should be R (Å) only after the phase shift is corrected. I am wondering whether no phase shift correction was performed for the data in the figure, if it is, the Abscissa should be R + α (Å), please check it.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 782, 238, 798]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 808, 851, 909]]<|/det|>
+We appreciate the reviewer's suggestion. We changed the abscissa according to the reviewer's suggestion and updated Figures 3b and 3c in the revised manuscript (Page 12, Fig. 3b and 3c). Meanwhile, the wavelet transform used the raw data of XAFS directly, no phase shift correction was performed.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 131, 850, 179]]<|/det|>
+6. It is mentioned by the author that Hb is competitive in price. But the price of Hb (US \(3.60 / \mathrm{g}\) , Shanghai Yuanye, S12021-5g) is significantly higher than that of FeCl₃.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 189, 238, 204]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 214, 852, 456]]<|/det|>
+We appreciate the reviewer's valuable comments. The prices of Hb and FeCl₃ are greatly depended on their grades and suppliers. Herein, we provided the prices of two iron resources for reference. At least, the price of Hb is accepted as the cheap resource. The price of Hb is US \(42.7 per 100 g from Energy Chemical, D110156- 100g and the FeCl₃ is US\) 50.3 per 100 g from Sigma- Aldrich (157740- 100G), whereas the activity of Hb- templated pFeSAN can be up to 8791 times higher than that Fe₂O₃ with FeCl₃ as iron source. Therefore, this one- pot biomimetic synthetic strategy is efficient and low- cost. We agree the Reviewer's comments and made the change in the revised manuscript in Page 13, Line 6- 7.  
+
+<|ref|>text<|/ref|><|det|>[[147, 464, 852, 595]]<|/det|>
+Besides, blood protein (hemoglobin and albumin) as a biomass, which is normally discarded as waste and causes serious environmental pollution (Adv. Mater., 2018, 30, e1703691). This work offers a promising green route to attain high- value catalysts from biomass and demonstrates another advantage of using Hb as an iron source and template for iron single atom catalysts.  
+
+<|ref|>text<|/ref|><|det|>[[147, 604, 687, 622]]<|/det|>
+More importantly, Hb has three significant advantages over FeCl₃:  
+
+<|ref|>text<|/ref|><|det|>[[147, 631, 850, 707]]<|/det|>
+(1) Fe ions residing in the structure of Hb could be isolated by protein structure, effectively preventing the aggregation of atomically dispersed iron active sites during harsh pyrolysis process;  
+
+<|ref|>text<|/ref|><|det|>[[147, 715, 850, 789]]<|/det|>
+(2) Hb with a size of 2\~3 nm forms a large number of mesopores during pyrolysis, benefiting the formation of mesopores and the maximum exposure of atomic Fe sites and also facilitating mass transfer during catalysis;  
+
+<|ref|>text<|/ref|><|det|>[[147, 798, 850, 845]]<|/det|>
+(3) Unsymmetrically coordinated Fe-N₃ active site of pFeSAN was prepared by the Hb-pyrolysis strategy, delivering elevated oxidase-like activity.  
+
+<|ref|>text<|/ref|><|det|>[[147, 854, 850, 901]]<|/det|>
+Oppositely, FeCl₃ as a Fe source requires careful regulation to prevent metal site from agglomeration. Also, other hard templates are needed to create the mesoporous
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 90, 850, 135]]<|/det|>
+structures, requiring the precise control of the operation. In contrast, our synthesis strategy is facile and easily scale- up.  
+
+<|ref|>text<|/ref|><|det|>[[148, 172, 850, 218]]<|/det|>
+7. Lack of comparison, compared with other nano-enzymes with oxidase-like activity, what is the advantage of pFeSAN in the kinetic constant?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 229, 238, 245]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 255, 853, 551]]<|/det|>
+We appreciate the reviewer's valuable suggestions. Besides the control catalysts of \(\mathrm{Fe_3O_4}\) and FeSAN \(\mathrm{(Fe - N_4)}\) , we also systematically compare the performance of pFeSAN with the previously reported Pt NPs (Ref. 41: Biosens. Bioelectron. 2017, 92, 442) and \(\mathrm{Mn_3O_4}\) (Ref. 42: Sens. Actuators B Chem. 2021, 333, 129560) on their oxidase- like activity using TMB as a chromogenic substance. The relative activities of the pFeSAN were 5882 and 1176 times higher than that of Pt and \(\mathrm{Mn_3O_4}\) , respectively, indicating that the oxidase- like activity of the pFeSAN was much higher than Pt and \(\mathrm{Mn_3O_4}\) (Figure R4a). According to Figure R4b,c, pFeSAN has a higher affinity for substrate TMB than Pt and \(\mathrm{Mn_3O_4}\) , and the catalytic reaction rate is much higher than both Pt and \(\mathrm{Mn_3O_4}\) . The relative information has been updated in revised manuscript (Page 16, Line 20–22 and Supplementary Fig. 31).  
+
+<|ref|>image<|/ref|><|det|>[[207, 560, 770, 812]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 818, 855, 890]]<|/det|>
+<center>Figure R4. Oxidase-like activity evaluation. a. Comparison of relative oxidase-like catalytic activities of different nanozymes. b. Comparison of kinetics for pFeSAN, Pt and \(\mathrm{Mn_3O_4}\) . \(\mathrm{K_m}\) is the Michaelis-Menten constant. \(\mathrm{V_{max}}\) is the maximal reaction velocity. c. Steady-state kinetic assay of Pt and \(\mathrm{Mn_3O_4}\) with TMB as substrate. </center>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 116, 852, 219]]<|/det|>
+8. For Figure 5a and 5b, the author pointed out that “no apparently detectable signals of any ROS further proved that ROSs were not the main intermediates for the oxidase-like activity of pFeSAN, which was very different from the majority of the previously reported nanozymes”, please give more discussion on this phenomenon.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 230, 238, 246]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 255, 853, 721]]<|/det|>
+Thank you very much for your comments. Under \(\mathrm{N}_2\) atmosphere, TMB could not be oxidized by pFeSAN. Oppositely, the oxidase-like activity of pFeSAN could be boosted up to 2.2-fold by increasing the \(\mathrm{O}_2\) concentration to saturation (Figure 5a). Therefore, the high \(\mathrm{O}_2\) - dependent activity of pFeSAN for substrate oxidation suggests the essence of \(\mathrm{O}_2\) as oxidants for catalytic reaction. Meanwhile, ROSs were undetectable by using either ROS quenchers (Supplementary Fig. 32a) and EPR spectra (Supplementary Fig. 32b). These results strongly confirmed that pFeSAN mediated the complete reduction of \(\mathrm{O}_2\) to \(\mathrm{H}_2\mathrm{O}\) without the release of free ROSs. This phenomenon verified that pFeSAN likely followed an oxygen atom transfer mechanism, similar to that of the natural \(CCO\) enzymes mediated by the \(\mathrm{Fe(IV) = O}\) intermediate, which was very different from the majority of the previously reported oxidase-like nanozymes through the free ROS pathway (Ref. 23: Anal. Chem. 2022, 94, 15270; Ref. 43: Catal. Sci. Technol. 2021, 11, 7255). The \(\mathrm{Fe(IV) = O}\) is considered a key intermediate in the oxidase-like reaction of many heme iron enzymes and heme analogs. (Ref. 45: Acc. Chem. Res., 2007, 40, 7, 554). We speculated that the formation of \(\mathrm{Fe(IV) = O}\) intermediates enabled the efficient oxidation of TMB substrate into ox-TMB. In the revised manuscript, we have added the related information in the main context (Page 18, Line 6–8).  
+
+<|ref|>text<|/ref|><|det|>[[148, 771, 487, 788]]<|/det|>
+9. For Figure 2f, the resolution is too low.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 800, 238, 816]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 826, 850, 872]]<|/det|>
+Thank you for your kind suggestions. In the revised manuscript, we have updated the related information in the Results section (Page 7, Fig. 2f).
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 90, 460, 107]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[147, 117, 852, 330]]<|/det|>
+Due to the high atom utilization and remarkable catalytic activity, single- atom catalysts exhibit great potentials for applications of nanozyme. In this work, a porous Fe single- atom catalyst prepared by a biomineralization- pyrolysis strategy exhibited remarkable oxidase- like activity, which was used for GSH sensing and cancer cell identification. The porous Fe single- atom nanozyme was well characterized. And, the mechanism of oxidase- like activity was investigated as well, which opened up a new way to design and optimize nanozymes. Therefore, this manuscript can be accepted to be published by Nature Communications after proper revisions.  
+
+<|ref|>text<|/ref|><|det|>[[148, 339, 850, 386]]<|/det|>
+We thank the reviewer for his/her useful suggestions on our manuscript. The point- to- point response to the reviewers' comments is shown as below.  
+
+<|ref|>text<|/ref|><|det|>[[148, 394, 850, 441]]<|/det|>
+1. Fe single-atom nanozyme reported in the past often exert catalase-like catalytic activity. In this study, does pFeSAN have catalase-like activity?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 451, 238, 467]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 477, 852, 690]]<|/det|>
+Thank you very much for this constructive comment. Our experimental results show that the pFeSAN possesses ideal catalase- like (CAT- like) activity under neutral conditions (pH 6.0- 7.0), similar to those previously reported studies (Chem. Commun., 2019, 55, 14534- 14537; Adv. Mater., 2022, 34, e2205324; Nat. Commun., 2022, 13, 4744. ) (Figure R5a,b). Besides, the elimination efficiency of \(\mathrm{H}_2\mathrm{O}_2\) was dependent on the concentration of pFeSAN (Figure R5c). The CAT- like activity of pFeSAN may provide opportunities for its in biomedical applications, such as the treatments of neurodegenerative diseases, inflammation, as well as cancer treatment.  
+
+<|ref|>image<|/ref|><|det|>[[222, 696, 772, 821]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 826, 852, 900]]<|/det|>
+<center>Figure R5. a. The photo of comparison of the mixed solution \(\mathrm{H}_2\mathrm{O}_2\) in the absence and presence of pFeSAN with the concentration of \(40~\mu \mathrm{g / mL}\) . b. Comparison of CAT-like activities of pFeSAN at various pH (4-12) conditions. c. The \(\mathrm{H}_2\mathrm{O}_2\) scavenging of pFeSAN with different concentrations. </center>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 89, 850, 136]]<|/det|>
+2. Compared with previously reported Fe single-atom nanozyme, what are the advantages of pFeSAN in structure-activity characteristics?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 147, 238, 163]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 172, 852, 581]]<|/det|>
+We thank the Reviewer for the positive comments. Developing artificial enzymes with the excellent catalytic performance of natural enzymes has been a long- standing goal for chemists. Fe single- atom catalysts coordinated with N atoms (Fe- N- C) exhibit well- defined atomic structures and electronic coordination environments and thereby deliver various enzyme- like catalytic activity. Unfortunately, majority of them were synthesized via pyrolysis at high temperatures, leading to structural collapse and part of the buried Fe- N<sub>x</sub> units inaccessible to biomolecules. Also, the strong stacking of those N- doped graphite carbon in various Fe- N- C nanozymes generally induces the frustrated diffusion of bio- substrates to metal sites. Hence, various methods including spatial confinement, defect/vacancy engineering and coordination modulations have been developed to solve those problems. However, only part of those inadequacies could be overcome. Thus, seeking for a new synthetic strategy of single- atom metal nanozymes to simultaneously achieve the atomic metal dispersion, modulated electronic structure, elevated mass transport and tailorable coordination environment is still on high demands.  
+
+<|ref|>text<|/ref|><|det|>[[147, 589, 852, 774]]<|/det|>
+In this work, we report a straightforward Hb- mediated approach to efficiently distribute iron atoms homogeneously and in an atomically dispersed fashion across the nitrogen- doped carbon support. Meanwhile, the as- prepared three- coordinated single- atom Fe nanozyme possesses significantly higher specific surface area and mesoporosity to facilitate the mass transport and exposure of active Fe sites, and exhibits ideal oxidase- like catalytic activity. The benefits of our approach could be described as below.  
+
+<|ref|>text<|/ref|><|det|>[[166, 779, 850, 890]]<|/det|>
+(1) pFeSAN with the unsymmetrically coordinated Fe-N₃ active sites was synthesized successfully by using Hb as Fe source. This synthesis represented a simple, facile and scalable one.  
+(2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[225, 84, 850, 120]]<|/det|>
+significantly facilitating the mass transfer of reactants/products during the catalytic process.  
+
+<|ref|>text<|/ref|><|det|>[[167, 122, 850, 159]]<|/det|>
+(3) pFeSAN delivered outstanding oxidase-like activity, which was 3.3- and 8791-times higher than those of Fe-N4 and Fe3O4 nanozymes, respectively.  
+
+<|ref|>text<|/ref|><|det|>[[167, 161, 850, 235]]<|/det|>
+(4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of \(\mathrm{O_2}\) into \(\mathrm{H_2O}\) , being identical to that of \(\mathrm{CcO}\) , which was very different from the majority of the previously reported oxidase-like nanozymes with the generation of ROS.  
+
+<|ref|>text<|/ref|><|det|>[[167, 236, 850, 292]]<|/det|>
+(5) pFeSAN as a highly-performed nanozyme exhibited a much higher upper detection limit of GSH at \(1\mathrm{mM}\) , which was 2.5 to 40-fold higher than those of the previously reported investigations (Table R4).  
+
+<|ref|>text<|/ref|><|det|>[[167, 293, 850, 348]]<|/det|>
+(6) Visual and rapid detection of tumor tissues through GSH colorimetric analysis was achieved for the first time, which was expected to help the effective resection of tumor tissues.  
+
+<|ref|>text<|/ref|><|det|>[[148, 408, 850, 454]]<|/det|>
+3. The label of elements and scale bar at Supplementary Fig. 11 and 15 was not clearly visible.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 466, 238, 482]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 492, 850, 538]]<|/det|>
+We are grateful for the suggestion. We have added the label of elements and scale bar in the revised Supplementary Information Fig. 11 and 15.  
+
+<|ref|>text<|/ref|><|det|>[[147, 605, 627, 622]]<|/det|>
+4. Please provide the quantitative analysis of EIS (Fig. 5d).  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 634, 238, 650]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 660, 852, 875]]<|/det|>
+Thank you for your insightful comments. According to the EIS fitting results, it is known that the material resistance of pFeSAN is much smaller than that of FeSAN. Figure 5d shows the Nyquist plots of pFeSAN and FeSAN, where the diameter of the semicircle is associated with the electron-transfer resistance \((\mathrm{R}_{\mathrm{ct}})\) . The \(\mathrm{R}_{\mathrm{ct}}\) decreased substantially from 407.3 to \(298.6 \Omega\) , implying enhanced electron transfer property and better mass transfer performance, which were consistent with the above results from CV (Fig. 5c). We have revised the manuscript and added the description in Results section (Page 19, Line 13).
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 116, 850, 163]]<|/det|>
+5. In this work, did the pFeSAN show obvious toxicity to normal cells and tumor cells, resulting in the influenced cellular GSH analysis?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 173, 238, 190]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 199, 852, 358]]<|/det|>
+As suggested by the reviewer, we have now performed a biosafety analysis in vitro. In order to make sure the biosafety of pFeSAN, we examined the biocompatibility of the pFeSAN through in vitro cell viability. As shown in Figure R6, pFeSAN exhibited negligible cytotoxicity (cell viability \(>90\%\) ) on normal liver cells (AML12) and liver tumor cells (Hepa 1- 6) even at a high concentration of \(200 \mu \mathrm{g} / \mathrm{mL}\) , corroborating the high biocompatibility of the pFeSAN.  
+
+<|ref|>image<|/ref|><|det|>[[300, 376, 666, 600]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 620, 850, 657]]<|/det|>
+<center>Figure R6. Cellular viability of AML12 and Hepa 1-6 cells treated with various concentrations of the pFeSAN for 3 h. </center>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 90, 460, 107]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[147, 144, 853, 444]]<|/det|>
+This paper deals with the development of a porous 3- coordinate iron- based nanozyme which can exhibit oxidase- like activity. The authors use hemoglobin as Fe source/template and incorporates it into a zeolitic imidazolate framework (ZIF- 8). The single- atom nanozyme called pFeSAN acts as an oxidase model by oxidizing the colorimetric substrate TMB to oxTMB, which helps in the detection of glutathione (GSH) in nM concentrations. The authors show a significant improvement in the oxidase- like activity; the nanozyme exhibiting 3.3- and 8791- fold higher oxidase- like activity than the Fe- N4 and Fe3O4 nanozymes reported earlier. As the GSH level is altered in many diseases, including cancer, the authors propose that the nanozyme with oxidase- like activity can be used for the visualization of tumor. The work is interesting, but does not warrant publication in Nat. Commun. due to the following reasons.  
+
+<|ref|>text<|/ref|><|det|>[[147, 465, 850, 540]]<|/det|>
+We thank the Reviewer for raising the critical and constructive comments on our manuscript. According to the referees' comments, the manuscript has been carefully revised. The answers to the comments by referees are enclosed as below.  
+
+<|ref|>text<|/ref|><|det|>[[147, 562, 853, 802]]<|/det|>
+(1) Development of Fe-based single-atom nanozyme by incorporating hemoglobin into ZIF-8: The topic of Fe-based single atom nanozyme is not new. It has been shown that bovine hemoglobin can be incorporated into ZIF-8 to develop single atom nanozymes (ACS Appl. Mater. Interfaces 2016, 8, 29052–29061). This nanozyme has been used as peroxidase mimic to oxidize TMB for the colorimetric detection. Many other Fe-based single atom nanozymes with oxidase-like activity have been reported in a review (Nano Res. 2023, 16, 1992–2002). The mechanism of the structure-dependent oxidase-like activity of Fe-N/C nanozyme has also been reported (Chem. Commun. 2019, 55, 5271–5274).  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 814, 238, 830]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 854, 850, 901]]<|/det|>
+Thank you for your constructive comments. Developing artificial enzymes with the excellent catalytic performance of natural enzymes has been a long- standing goal for
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 853, 441]]<|/det|>
+chemists. Fe single- atom catalysts coordinated with N atoms (Fe- N- C) exhibit welldefined atomic structures and electronic coordination environments and thereby deliver various enzyme- like catalytic activity. Unfortunately, majority of them were synthesized via pyrolysis at high temperatures, leading to structural collapse and part of the buried Fe- N<sub>x</sub> units inaccessible to biomolecules. Also, the strong stacking of those N- doped graphite carbon in various Fe- N- C nanozymes generally induces the frustrated diffusion of bio- substrates to metal sites. Hence, various methods including spatial confinement, defect/vacancy engineering and coordination modulations have been developed to solve those problems. However, only part of those inadequacies could be overcome. Thus, seeking for a new synthetic strategy of single- atom metal nanozymes to simultaneously achieve the atomic metal dispersion, modulated electronic structure, elevated mass transport and tailorable coordination environment is still on high demands.  
+
+<|ref|>text<|/ref|><|det|>[[147, 450, 852, 634]]<|/det|>
+In this work, we report a straightforward hemoglobin- mediated approach to efficiently distribute iron atoms homogeneously and in an atomically dispersed fashion across the nitrogen- doped carbon support. Meanwhile, the as- prepared three- coordinated single- atom Fe nanozyme possesses significantly higher specific surface area and mesoporosity to facilitate the mass transport and exposure of active Fe sites, and exhibits ideal oxidase- like catalytic activity. The benefits of our approach could be described as below.  
+
+<|ref|>text<|/ref|><|det|>[[166, 653, 852, 899]]<|/det|>
+(1) pFeSAN with the unsymmetrically coordinated Fe-N<sub>3</sub> active sites was synthesized successfully by using Hb as Fe source. This synthesis represented a simple, facile and scalable one.  
+(2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants/products during the catalytic process.  
+(3) pFeSAN delivered outstanding oxidase-like activity, which was 3.3- and 8791- times higher than those of Fe-N<sub>4</sub> and Fe<sub>3</sub>O<sub>4</sub> nanozymes, respectively.  
+(4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of O<sub>2</sub> into H<sub>2</sub>O, being identical to that of C<sub>6</sub>O, which was very different from the majority of the previously reported
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[226, 85, 849, 119]]<|/det|>
+oxidase-like nanozymes with the generation of reactive oxygen species (ROS).  
+
+<|ref|>text<|/ref|><|det|>[[168, 122, 850, 177]]<|/det|>
+(5) pFeSAN as a highly-performed nanozyme exhibited a much higher upper detection limit of GSH at 1 mM, which was 2.5 to 40-fold higher than those of the previously reported investigations (Table R4).  
+
+<|ref|>text<|/ref|><|det|>[[168, 180, 850, 234]]<|/det|>
+(6) Visual and rapid detection of tumor tissues through GSH colorimetric analysis was achieved for the first time, which was expected to help the effective resection of tumor tissues.  
+
+<|ref|>text<|/ref|><|det|>[[148, 255, 849, 301]]<|/det|>
+Herein, we detailed comparisons the previous works listed by the Reviewer and wished to clearly illustrate the value and innovation of our work.  
+
+<|ref|>text<|/ref|><|det|>[[148, 325, 577, 342]]<|/det|>
+ACS Appl. Mater. Interfaces, 2016, 8, 29052- 29061:  
+
+<|ref|>text<|/ref|><|det|>[[147, 366, 851, 524]]<|/det|>
+In this work, hemeprotein was directly embedded in ZIF- 8 and the hemeprotein- embedded ZIF- 8 as peroxidase mimic applied for \(\mathrm{H}_2\mathrm{O}_2\) and phenol detection. The ZIF- 8@BHb hybrid composite is a MOF- based nanozyme, not a Fe single atom nanozyme. Compared with our work, the material structure, type of mimic- enzyme activity and detection substrate of the above studies are very difference (Table R1). Thus, the two studies present different research contents.  
+
+<|ref|>table<|/ref|><|det|>[[148, 580, 849, 699]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 548, 731, 567]]<|/det|>
+Table R1. Comparison of pFeSAN and ZIF-8@BHb hybrid composites.   
+
+<table><tr><td></td><td>Catalyst</td><td>Mimic function</td><td>Application</td></tr><tr><td>Our work</td><td>Mesoporous Fe-N3</td><td>Oxidase-like (OXD)</td><td>GSH detection</td></tr><tr><td>ACS Appl. Mater. Interfaces, 2016, 8, 29052.</td><td>ZIF-8@BHb hybrid composite</td><td>Peroxidase-like (POD)</td><td>H2O2 and phenol detection</td></tr></table>  
+
+<|ref|>text<|/ref|><|det|>[[148, 730, 416, 747]]<|/det|>
+Nano Res. 2023, 16, 1992- 2002:  
+
+<|ref|>text<|/ref|><|det|>[[147, 772, 851, 902]]<|/det|>
+Despite many Fe- based single atom nanozymes with oxidase- like activity have been reported. Most Fe single atom nanozymes are manufactured by forming an \(\mathrm{Fe - N_x}\) structure, which is achieved by using inorganic metal precursors and nitrogen- doped carbon support under the high- temperature treatments. In this process, the drawback is that Fe can aggregate to block the pores and the active site of the \(\mathrm{Fe - N_x}\) can
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 853, 415]]<|/det|>
+collapse. Moreover, this process generally requires the further acid treatment for the removal of large particles. As catalyst activity decreases of the active center and mass transfer properties, the Fe single atom nanozymes typically exhibit lower performance. To solve these issues, we demonstrated a bioinspired synthetic strategy for the massive preparation of highly active porous three- coordinated Fe single- atom nanozymes (pFeSAN) with oxidase- like activity by using biomineralized hemoglobin- containing ZIF as pyrolytic templates. Hemoglobin is an iron protein that contains nitrogen and metals that are useful as an ideal template and Fe source for bio- inspired Fe single- atom nanozyme. The iron- porphyrin contained in the hemoglobin plays an important role in forming the Fe- \(\mathrm{N_x}\) active sites during the heat treatment process. Compared to previous studies, pFeSAN with Fe- \(\mathrm{N_3}\) coordination and mesoporous structure increased the substrate transfer and thereby exhibited an outstanding oxidase- like activity.  
+
+<|ref|>text<|/ref|><|det|>[[147, 421, 853, 776]]<|/det|>
+The paper (Nano Res. 2023, 16, 1992- 2002) is a review article, which summarizes the recent progress of single- atom nanozymes in biomedicine. In this Review article, there is no general strategy to simultaneously achieve atomically dispersed, mesoporous structure and a well- regulated single- atom coordination environment (Table R2). Moreover, the mechanisms of \(\mathrm{O_2}\) activation and electrons transfer in the oxidase- like reaction of Fe- based single atom nanozymes are not particularly clear. Therefore, compared to other Fe- based single atom nanozymes, pFeSAN via one- step synthesis has the advantages of facile preparation, atomically dispersed Fe single atoms, mesoporous structure, Fe- \(\mathrm{N_3}\) coordination and excellent oxidase- like activity, making it a promising nanozyme for GSH detection and other biological applications. To further clarify the reaction mechanism of the pFeSAN, we investigated the \(\mathrm{O_2}\) activation and electron- transfer by EPR, DFT and electrochemical analysis. We believe this study will inspire other high- performance Fe- based single atom nanozymes development.
+
+<--- Page Split --->
+<|ref|>table<|/ref|><|det|>[[147, 120, 831, 744]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 90, 736, 108]]<|/det|>
+Table R2. Comparison of pFeSAN and other Fe single-atom nanozymes.   
+
+<table><tr><td>Fe source</td><td>Mespor <br>ous</td><td>Template</td><td>Aftertre <br>atment</td><td>Mimic <br>function</td><td>Applications</td><td>Ref</td></tr><tr><td>Hb</td><td>Yes</td><td>Hb</td><td>None</td><td>OXD</td><td>GSH <br>detection</td><td>Our work</td></tr><tr><td>FePc</td><td>None</td><td>None</td><td>HCl</td><td>OXD</td><td>Antibacterial</td><td>Sci. Adv., 2019, 5, <br>eaav5490.</td></tr><tr><td>(NH4)2Fe(SO4)2</td><td>Yes</td><td>F127</td><td>PEG</td><td>POD/CAT</td><td>Anti-tumor</td><td>Biomaterials, 2022, <br>281, 121325.</td></tr><tr><td>Fe(acac)3</td><td>None</td><td>None</td><td>H2SO4</td><td>OXD/POD</td><td>Anti-tumor</td><td>ACS Nano, 2022, <br>16, 1, 855</td></tr><tr><td>Fe(acac)3</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>Antibacterial</td><td>Small, 2019, 15, <br>e1901834.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>MgO</td><td>HNO3</td><td>POD</td><td>Glucose <br>detection</td><td>Small, 2020, <br>16, e2002343.</td></tr><tr><td>Fe(OAc)2</td><td>None</td><td>MgO</td><td>HNO3</td><td>POD</td><td>Osteosarcoma <br>treatment</td><td>Adv. Mater., 2021, <br>33, e2100150.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>SiO2</td><td>NaOH</td><td>POD</td><td>Anti-tumor</td><td>Adv. Mater., 2022, <br>34, e2107088.</td></tr><tr><td>FePc</td><td>None</td><td>None</td><td>H2SO4</td><td>CAT/SOD</td><td>Anti-oxidant</td><td>Chem. Commun., <br>2019, 55, 159.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>Butyrylcholine <br>esterase <br>detection</td><td>Biosens. <br>Bioelectron., 2019, <br>142, 111495.</td></tr><tr><td>FeCl2</td><td>None</td><td>None</td><td>Lyophilization</td><td>POD</td><td>H2O2 <br>detection</td><td>Anal. Chem., 2019, <br>91, 11994.</td></tr><tr><td>FeCl3</td><td>None</td><td>SiO2</td><td>HF</td><td>POD</td><td>Acetylcholine <br>sterase <br>detection</td><td>Small, 2019, 15, <br>e1903108.</td></tr><tr><td>Fe(NO3)3</td><td>None</td><td>None</td><td>None</td><td>POD</td><td>Acetylcholine <br>sterase <br>detection</td><td>ACS Nano, 2022, <br>16, 2, 2997.</td></tr><tr><td>Fe(OAc)2</td><td>None</td><td>None</td><td>H2SO4</td><td>OXD</td><td>GSH <br>detection</td><td>Chem. Commun., <br>2019, 55, 5271.</td></tr></table>  
+
+<|ref|>text<|/ref|><|det|>[[148, 772, 468, 789]]<|/det|>
+Chem. Commun. 2019, 55, 5271- 5274:  
+
+<|ref|>text<|/ref|><|det|>[[147, 814, 851, 889]]<|/det|>
+In this work, authors suggest that the \(\cdot \mathrm{O}_2^-\) radical is the main active species in the oxidase- like reaction of Fe- N/C- CNTs. However, ROS were undetectable by using either ROS quenchers and electron paramagnetic resonance spectra in our work
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 853, 303]]<|/det|>
+(Supplementary Fig. 32). These results confirmed that pFeSAN mediated the complete reduction of \(\mathrm{O_2}\) to \(\mathrm{H_2O}\) without releasing free ROS, indicating that the reaction mechanisms of the structure-dependent oxidase-like activity of Fe- N/C- CNTs and pFeSAN were very different. Meanwhile, the mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four-electron reduction of oxygen into \(\mathrm{H_2O}\) , identical to that of natural \(CCO\) . We also verified that pFeSAN likely followed an oxygen atom transfer mechanism similar to that of the natural \(CO\) by the \(\mathrm{Fe(IV) = O}\) intermediate, different from majority of previously reported oxidase mimics.  
+
+<|ref|>table<|/ref|><|det|>[[147, 359, 848, 459]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[148, 325, 546, 344]]<|/det|>
+Table R3. Comparison of pFeSAN and Fe-N3/C.   
+
+<table><tr><td></td><td>Catalyst</td><td>Acid treatment</td><td>Free ROS</td></tr><tr><td>Our work</td><td>Mesoporous FeN3</td><td>Not required</td><td>None</td></tr><tr><td>Chem. Commun., 2019, 55, 5271.</td><td>Fe-N3/C</td><td>H2SO4 (1 M), 5 h</td><td>·O2-</td></tr></table>  
+
+<|ref|>text<|/ref|><|det|>[[147, 488, 853, 733]]<|/det|>
+Overall, our study is a new synthetic strategy by employing the biomineralized hemoglobin@ZIF- 8 as sacrificial templates and Fe sources to prepare the atomically dispersed Fe single- atom nanozyme (featured by Fe- N3 coordination) within the mesopores of carbon support. Benefiting from the simple and facile one- step Hb- templated strategy, pFeSAN shows six key advantages (Response Letter: Page R13- R14) over the conventional FeSAN in terms of catalyst structures, performance of oxidase- like, and GSH detection, as mentioned above as well as in the revised manuscript. We hope that we illustrate the novelty of our investigations as well as distinguish the present study from previous investigations.  
+
+<|ref|>text<|/ref|><|det|>[[147, 781, 853, 910]]<|/det|>
+(2) Single atom nanozymes for the detection of glutathione (GSH). Nanozymes and Fe-N-C-based single atom nanozymes have been used extensively for the detection of glutathione (Journal of Materiomics, 2022, 8, 1251-1259). \(\mathrm{Mn_3O_4}\) microspheres have been used as an oxidase mimic for rapid detection of glutathione (RSC Adv., 2019,9, 16509-16514). Light-responsive MOF as an oxidase mimic for cellular GSH detection
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 852, 191]]<|/det|>
+(Anal. Chem. 2019, 91, 13, 8170–8175). MnO₂ nanosheets as an artificial enzyme to mimic oxidase for rapid and sensitive detection of glutathione (Biosensors & Bioelectronics, 2017, 90, 69- 74). Therefore, the present study lacks novelty in the detection of glutathione.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 202, 238, 218]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 238, 853, 792]]<|/det|>
+Thank you for this comment. Complete surgical resection is the ideal first- line treatment for most malignancies. Compared with biological normal cells, the concentration of GSH in cancer cells is considerably greater, reaching \(0.5 - 1.0 \text{mM}\) , which is about 1000 times that of normal cells, making it one of the most significant signal molecules to diagnose cancer. Therefore, the goal of complete surgical resection would be facilitated by GSH imaging that enables more precise visualization of tumor margins. However, due to its relatively high concentration, a high- performance detection system for GSH analysis is highly desired to provide a rapid localization of tumor area via GSH visualization detection. Unfortunately, the direct detection of GSH in the mM range has been recognized as a difficult challenge. The concentrations of GSH in tumor tissue are usually higher than the upper limit of GSH detection for those previously reported nanozymes (Table R4). Hence, the detection range of these above- mentioned nanozymes is too narrow to detect cell and tissue samples directly, as shown in Table R4. To address this challenge, our work focused on developing a high- performance detection system for the GSH detection using biomineralized hemoglobin@ZIF- 8 as sacrificial templates and Fe sources to prepare highly dispersed and FeN₃- coordinated single- atoms nanozyme. Based on these properties, a colorimetric method was developed to detect GSH, which presented a wide linear detection range of \(50 \text{nM} - 1.0 \text{mM}\) for GSH and successfully employed as a colorimetric probe for GSH visualization in tumor tissues.
+
+<--- Page Split --->
+<|ref|>table<|/ref|><|det|>[[147, 115, 854, 344]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 85, 800, 103]]<|/det|>
+Table R4. Comparison of the pFeSAN with other nanozymes for GSH detection.   
+
+<table><tr><td>Materials</td><td>Linear range (μM)</td><td>LOD (μM)</td><td>Reference</td></tr><tr><td>pFeSAN</td><td>0.05-1000</td><td>0.0024</td><td>This work</td></tr><tr><td>Fe-N-C SANs</td><td>100-400</td><td>78.3</td><td>J. Materiomics, 2022, 8, 1251.</td></tr><tr><td>MnO2</td><td>1-25</td><td>0.3</td><td>Biosens. Bioelectron., 2017, 90, 69.</td></tr><tr><td>PSMOF</td><td>0-40</td><td>0.68</td><td>Anal. Chem., 2019, 91, 8170.</td></tr><tr><td>Mn3O4</td><td>50-60</td><td>0.889</td><td>RSC Adv., 2019, 9, 16509.</td></tr></table>  
+
+<|ref|>text<|/ref|><|det|>[[147, 360, 852, 601]]<|/det|>
+In our work, the exceptionally high detection upper limit of pFeSAN is 2.5 to 40 times higher than that of these conventional nanozymes, making it suitable for detecting GSH with high levels (Table R4). The excellent detection performance of GSH by the pFeSAN originates from its synergistic advantages comprising highly dispersed metal single- atoms, the presence of mesopores, and well- regulated coordination environments of the single- atoms. We further utilize pFeSAN- DAB system as a GSH sensor to realize in vitro quantitative GSH visualization for Hep 1- 6 cells and tumor tissue with high GSH states have been accurately distinguished by visualization detection.  
+
+<|ref|>text<|/ref|><|det|>[[147, 610, 852, 825]]<|/det|>
+Besides, the comparison illustrated the dramatical difference of the pFeSAN from the previously reported nanozymes, which could be attributed to the unique structural features of pFeSAN and an enzyme- like catalytic pathway of the four- electron \(\mathrm{O}_2\) - to- \(\mathrm{H}_2\mathrm{O}\) reduction. The outstanding performance, unique structures and the natural enzyme- like catalytic pathway of pFeSAN indeed reflect the novelty of the dedicatedly designed catalysts for GSH detection, which serve as a real- time, facile, rapid ( \(\sim 6\) min) and precise visualization analysis methodology of tumors and shows its potential for diagnostic and clinic applications.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 853, 275]]<|/det|>
+(3) The concept of using GSH as biomarker for visualization of cancer cells is not entirely new. There are many recent reports in the literature which highlight the concept. For example, a MOF has been reported to exhibit oxidase-like activity by oxidizing TMB to oxTMB, which has been used as a colorimetric probe for GSH detection. The oxidase mimic has been used to analyze the GSH level in the lysates of normal and cancer cells (Anal. Chem. 2019, 91, 8170–8175). There are many other reports which describe the use of nanozymes for tumor visualization through GSH detection.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 285, 238, 302]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 310, 855, 579]]<|/det|>
+Thank you very much for pointing this out. The author developed a light- dependent metal–organic framework with oxidase- like activity and colorimetric detection of cancer cells through GSH detection after cell lysis treatment (Ref. 53: Anal. Chem. 2019, 91, 8170–8175). In our work, pFeSAN was applied to monitor the GSH levels in normal and cancer cells without additional requirements of light irradiation and complex pretreatment of cells. Importantly, methods for the visualization detection of tumor tissue though GSH has not been reported yet, which brought the possibility to realize specific detection in practical applications. The visualized analysis of the GSH in tumor tissue is gaining interests to improve surgical safety and to promote surgical therapeutic effects.  
+
+<|ref|>table<|/ref|><|det|>[[147, 628, 850, 737]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 599, 730, 618]]<|/det|>
+Table R5. Comparison of the pFeSAN with PSMOF for GSH detection.   
+
+<table><tr><td></td><td>Catalyst</td><td>Light dependent</td><td>Cell pretreatment</td><td>Intratumoral GSH detection</td></tr><tr><td>Our work</td><td>pFeSAN</td><td>Not required</td><td>Without pretreatment</td><td>Yes</td></tr><tr><td>Anal. Chem. 2019, 91, 8170.</td><td>PSMOF</td><td>300 W Xe lamp</td><td>Cell lysis</td><td>None</td></tr></table>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[119, 84, 294, 97]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[119, 112, 224, 125]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[119, 127, 300, 140]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[118, 140, 875, 224]]<|/det|>
+This paper presents an innovative biomimetic synthetic strategy for the synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using natural enzyme as a template, and the nanozyme exhibits highly efficient oxidase- like activity and potential applications in diagnostics. This research has value for the researchers in the related areas. A thorough, point- by- point response to each point raised by reviewers has been made. So I think this work can be accepted for publication in Nature Communications.  
+
+<|ref|>text<|/ref|><|det|>[[118, 266, 234, 279]]<|/det|>
+Reviewer #2:  
+
+<|ref|>text<|/ref|><|det|>[[118, 280, 300, 293]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[118, 293, 833, 322]]<|/det|>
+Authors have improved the manuscript according to the comments, and thus I recommend it publication.  
+
+<|ref|>text<|/ref|><|det|>[[118, 364, 224, 377]]<|/det|>
+Reviewer #4:  
+
+<|ref|>text<|/ref|><|det|>[[118, 379, 300, 392]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[120, 392, 835, 406]]<|/det|>
+[Note from the editor: Reviewer #4 was invited to assess the response given to Reviewer #3]  
+
+<|ref|>text<|/ref|><|det|>[[118, 420, 878, 475]]<|/det|>
+In the manuscript "Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione" rebuttal letter, the authors responded to the three questions raised by the reviewer 3 with lots of discussion. However, it might still not fully address the concerns about the novelty mentioned there.  
+
+<|ref|>text<|/ref|><|det|>[[118, 475, 866, 531]]<|/det|>
+Author response: (2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3\~4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants / products during the catalytic process.  
+
+<|ref|>text<|/ref|><|det|>[[118, 531, 868, 587]]<|/det|>
+Comments: The increase in the transfer of reactants / products usually related to Km value of the enzyme. However, in the manuscript, the authors showed that the pFeSAN showed a really small increase in binding affinity compared to the Fe- N4 they mentioned. This might need further clarification or more thoughts.  
+
+<|ref|>text<|/ref|><|det|>[[118, 587, 833, 616]]<|/det|>
+Author response: (3) pFeSAN delivered outstanding oxidase- like activity, which was 3.3- and 8791- times higher than those of Fe- N4 and Fe3O4 nanozymes, respectively.  
+
+<|ref|>text<|/ref|><|det|>[[118, 616, 875, 658]]<|/det|>
+Comments: The oxidase- like activity of the nanozymes were compared using activity ratio which is a not standardized item. How that is calculated and how to use that value to cross- compare with other oxidase- like nanozyme require further clarifications.  
+
+<|ref|>text<|/ref|><|det|>[[118, 658, 860, 714]]<|/det|>
+Author response: (4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four- electron reduction of O2 into H2O, being identical to that of CcO, which was very different from the majority of the previously reported oxidase- like nanozymes with the generation of reactive oxygen species (ROS).  
+
+<|ref|>text<|/ref|><|det|>[[118, 714, 878, 771]]<|/det|>
+Comments: This finding is not novel. As the nanozymes prepared by the authors only showed OxD- like activity, it is expected that there will be no ROS generated. (Similar to Sci. Adv., 2019, 5, eaav5490). If the authors stated it is a similar catalytic pathway to CcO, is there any of the reaction intermediates or transient species the authors trapped could be compared to CcO?  
+
+<|ref|>text<|/ref|><|det|>[[118, 771, 844, 813]]<|/det|>
+Author response: (5) pFeSAN as a highly- performed nanozyme exhibited a much higher upper detection limit of GSH at 1 mM, which was 2.5 to 40- fold higher than those of the previously reported investigations (Table R4).  
+
+<|ref|>text<|/ref|><|det|>[[118, 814, 870, 842]]<|/det|>
+Comments: Previously, other systems based on nanozyme activity detected similar or even higher range. (New J. Chem., 2022, 46, 10682- 10689)
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[280, 90, 715, 108]]<|/det|>
+## Point-by-point Response to Reviewers' Comments  
+
+<|ref|>text<|/ref|><|det|>[[148, 132, 464, 150]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[147, 186, 852, 345]]<|/det|>
+This paper presents an innovative biomimetic synthetic strategy for the synthesis of porous Fe- N3 single atom nanozymes (pFeSAN) using natural enzyme as a template, and the nanozyme exhibits highly efficient oxidase- like activity and potential applications in diagnostics. This research has value for the researchers in the related areas. A thorough, point- by- point response to each point raised by reviewers has been made. So I think this work can be accepted for publication in Nature Communications.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 382, 238, 399]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 409, 850, 456]]<|/det|>
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[148, 550, 464, 567]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 604, 850, 650]]<|/det|>
+Authors have improved the manuscript according to the comments, and thus I recommend it publication.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 688, 238, 704]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 715, 850, 760]]<|/det|>
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 90, 463, 107]]<|/det|>
+Reviewer #4 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 145, 850, 191]]<|/det|>
+[Note from the editor: Reviewer #4 was invited to assess the response given to Reviewer #3]  
+
+<|ref|>text<|/ref|><|det|>[[147, 228, 852, 357]]<|/det|>
+In the manuscript "Bioinspired porous three- coordinated single- atom Fe nanozyme with oxidase- like activity for tumor visual identification via glutathione" rebuttal letter, the authors responded to the three questions raised by the reviewer 3 with lots of discussion. However, it might still not fully address the concerns about the novelty mentioned there.  
+
+<|ref|>text<|/ref|><|det|>[[148, 394, 850, 440]]<|/det|>
+We really appreciate the Reviewer's useful comments and suggestions. We have carefully revised the manuscript based on his/her comments.  
+
+<|ref|>text<|/ref|><|det|>[[147, 477, 851, 580]]<|/det|>
+Author response: (2) Evenly distributed Fe atoms in Hb effectively avoided the agglomeration of active sites during pyrolysis and created mesoporous structure (3- 4 nm) in the pFeSAN, thereby maximumly exposing the atomic Fe sites and significantly facilitating the mass transfer of reactants / products during the catalytic process.  
+
+<|ref|>text<|/ref|><|det|>[[147, 588, 851, 690]]<|/det|>
+Comments: The increase in the transfer of reactants / products usually related to Km value of the enzyme. However, in the manuscript, the authors showed that the pFeSAN showed a really small increase in binding affinity compared to the Fe- N4 they mentioned. This might need further clarification or more thoughts.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 730, 238, 745]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 755, 850, 801]]<|/det|>
+Thanks for the Reviewer's kind suggestions, which are valuable for improving the accuracy of the manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[147, 810, 850, 885]]<|/det|>
+For analyzing the catalytic mechanism and acquiring kinetic parameters, the oxidase- like activities of pFeSAN and Fe- N4 under the same conditions were studied by enzyme kinetics theory and methods. With the Lineweaver- Burk equation, the important
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[146, 88, 852, 498]]<|/det|>
+enzyme kinetic parameters such as Michaelis-Menten constant \(\mathrm{(K_m)}\) , maximal velocity \(\mathrm{(V_{max})}\) , catalytic constant \(\mathrm{(K_{cat})}\) and \(\mathrm{K_{cat} / K_m}\) were presented in Table R1 (Adv. Mater., 2022, 34, e2201736; Nat. Protoc., 2018, 13, 1506- 1520. ). \(\mathrm{K_m}\) was identified as an indicator of enzyme affinity to substrates. Smaller \(\mathrm{K_m}\) values thus indicate a stronger affinity between the enzyme and the substrate. \(\mathrm{V_{max}}\) represents the reaction rate when the enzyme is saturated with substrate, and a higher \(\mathrm{V_{max}}\) value indicates a quicker reaction rate. The \(\mathrm{K_{cat}}\) value gave a direct measure of the enzymatic catalytic activity. Generally, the maximal velocity of reaction \(\mathrm{V_{max}}\) and \(\mathrm{K_{cat}}\) reveals the catalytic activity of enzyme. Increasing the transfer of chemicals not only increase the decrease of \(\mathrm{Km}\) but also increase the diffusion of chemicals, leading to the overall enhanced catalytic kinetics. Besides, \(\mathrm{K_{cat} / K_m}\) is known to be a descriptor of enzyme efficiency and a better indicator to compare two enzymes. The higher catalytic efficiency of pFeSAN signifies more substrate- to- product conversion, which is happening due to its large affinity (low \(\mathrm{K_m}\) ) for TMB and greater proportion of bound substrate conversion to product before its dissociation (large turnover \(\mathrm{K_{cat}}\) ).  
+
+<|ref|>table<|/ref|><|det|>[[147, 545, 850, 675]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 520, 618, 539]]<|/det|>
+Table R1. Comparison of kinetics for pFeSAN and Fe-N4.   
+
+<table><tr><td>Catalyst</td><td>Km (mM)</td><td>Vmax (μM s-1)</td><td>Kcat (s-1)</td><td>Kcat/Km (mM-1s-1)</td></tr><tr><td>pFeSAN</td><td>0.17</td><td>1.67</td><td>2.6×106</td><td>1.53×107</td></tr><tr><td>Fe-N4</td><td>0.29</td><td>0.036</td><td>9.6×104</td><td>3.31×105</td></tr></table>  
+
+<|ref|>text<|/ref|><|det|>[[147, 690, 852, 905]]<|/det|>
+From Table R1, the \(\mathrm{K_m}\) value for pFeSAN (0.17 mM) to the TMB was about \(58\%\) of that for Fe- N4 (0.29 mM), indicating that it has higher affinity with TMB and lower concentration of TMB required to reach the maximal activity of \(\mathrm{V_{max}}\) . Hence, the \(\mathrm{V_{max}}\) and \(\mathrm{K_{cat}}\) values of pFeSAN for TMB showed 46.4- fold and 27.1- fold increases relative to Fe- N4, verifying the mesoporous structure in improving the oxidase- like performance. Meanwhile, the catalytic efficiency \(\mathrm{(K_{cat} / K_m)}\) of pFeSAN (1.53×107 mM-1s-1) is 46.2- fold higher than that of Fe- N4 (3.31×105 mM-1s-1). Overall, all these kinetic parameters including Michaelis- Menten constant \(\mathrm{(K_m)}\) , maximal reaction velocity
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 89, 852, 192]]<|/det|>
+\((\mathrm{V}_{\mathrm{max}})\) , catalytic rate constant \((\mathrm{K}_{\mathrm{cat}})\) , and \(\mathrm{K}_{\mathrm{cat}} / \mathrm{K}_{\mathrm{m}}\) of pFeSAN showed the optimum values than Fe- \(\mathrm{N}_4\) , indicating a distinct positive contribution of the mesoporous structure (3\~4 nm) and higher surface area to the oxidase- like activity of the pFeSAN (Figure R1). The larger pore size can make a great contribution to fast mass transfer.  
+
+<|ref|>image<|/ref|><|det|>[[208, 209, 757, 386]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 400, 850, 437]]<|/det|>
+<center>Figure R1. a. Pore size distribution curves of Fe- \(\mathrm{N}_4\) and pFeSAN b. BET surface areas of Fe- \(\mathrm{N}_4\) and pFeSAN. </center>  
+
+<|ref|>text<|/ref|><|det|>[[147, 455, 852, 641]]<|/det|>
+In conclusion, we demonstrated a biomimetic synthetic strategy for scalable synthesis of porous Fe single- atom nanozymes using hemoglobin as both template and Fe- source, which delivered a high oxidase- like activity. The mesoporous features of pFeSAN significantly promoted mass transport and maximumly exposed active iron sites during reaction, which could greatly enhance the oxidase- like activity of pFeSAN. The relative information has been updated in the revised manuscript (Page 16, Line 19–22; Page 17, Line 1–5 and Supplementary Table 2).  
+
+<|ref|>text<|/ref|><|det|>[[147, 677, 852, 808]]<|/det|>
+Author response: (3) pFeSAN delivered outstanding oxidase- like activity, which was 3.3- and 8791- times higher than those of Fe- \(\mathrm{N}_4\) and \(\mathrm{Fe}_3\mathrm{O}_4\) nanozymes, respectively. Comments: The oxidase- like activity of the nanozymes were compared using activity ratio which is a not standardized item. How that is calculated and how to use that value to cross- compare with other oxidase- like nanozyme require further clarifications.  
+
+<|ref|>sub_title<|/ref|><|det|>[[149, 845, 238, 861]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 872, 848, 890]]<|/det|>
+We would like to express our point for your valuable suggestions regarding our work.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 89, 852, 163]]<|/det|>
+In response, we want to clarify that we initially utilized a generic method to calculate the activity of nanozymes in the main text. We recognized the importance of consistency in comparing the activity of different nanozymes.  
+
+<|ref|>text<|/ref|><|det|>[[147, 172, 853, 469]]<|/det|>
+We calculated the oxidase- like activities of pFeSAN, Fe- N4 and \(\mathrm{Fe_3O_4}\) according to the standardized assay protocol (Nat. Catal. 2021, 4, 407- 417; Nat. Protoc. 2018, 13, 1506- 1520). By quantitatively determined the specific activity values (U/mg) of pFeSAN, Fe- N4 and \(\mathrm{Fe_3O_4}\) by measuring the absorption intensity of the nanozyme- catalyzed TMB colorimetric reactions (Figure R2). The specific activity of pFeSAN was determined to be \(593~\mathrm{U / mg}\) , which is 3.5 times higher than that of Fe- N4 (169 U/mg) and 8471 times higher than that of \(\mathrm{Fe_3O_4}\) (0.07 U/mg). The high specific activity of pFeSAN following factors: Firstly, the mesoporous structure (3- 4 nm) of pFeSAN exhibits larger surface area (705.8 \(\mathrm{m^2 / g}\) ) than that of Fe- N4 (561.6 \(\mathrm{m^2 / g}\) ), and the large specific surface area can help fast mass transfer. Secondly, the pFeSAN exposes more Fe active sites, greatly enhance its oxidase- like activity.  
+
+<|ref|>image<|/ref|><|det|>[[330, 500, 640, 699]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 715, 850, 752]]<|/det|>
+<center>Figure R2. Reaction-time curves of the TMB colorimetric reaction catalyzed by pFeSAN, Fe-N4 and \(\mathrm{Fe_3O_4}\) . </center>  
+
+<|ref|>text<|/ref|><|det|>[[148, 771, 610, 789]]<|/det|>
+The calculation part of the specific activity is as follows:  
+
+<|ref|>text<|/ref|><|det|>[[148, 799, 710, 818]]<|/det|>
+Calculate the nanozyme activity (units) using the following equation:  
+
+<|ref|>equation<|/ref|><|det|>[[384, 841, 612, 860]]<|/det|>
+\[\mathrm{b_{nanozyme} = V / (\epsilon\times 1)\times(\Delta A / \Delta t)}\]  
+
+<|ref|>text<|/ref|><|det|>[[147, 883, 850, 901]]<|/det|>
+where \(\mathrm{b_{nanozyme}}\) is the catalytic activity of nanozyme expressed in units. One unit is
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 88, 852, 247]]<|/det|>
+defined as the amount of nanozyme that catalytically produces \(1\mu \mathrm{mol}\) of product per min at room temperature; V is the total volume of reaction solution \((\mu \mathrm{L})\) ; \(\epsilon\) is the molar absorption coefficient of the colorimetric substrate, which is maximized at \(39,000\mathrm{M}^{- 1}\) \(\mathrm{cm}^{- 1}\) at \(652\mathrm{nm}\) for TMB; I is the path length of light traveling in the cuvette (cm); A is the absorbance after subtraction of the blank value; and \(\Delta \mathrm{A} / \Delta \mathrm{t}\) is the initial rate of change in absorbance at \(652\mathrm{nm}\mathrm{min}^{- 1}\) .  
+
+<|ref|>text<|/ref|><|det|>[[148, 255, 630, 275]]<|/det|>
+Calculate the specific activity of the nanozyme (U \(\mathrm{mg}^{- 1}\) ) by  
+
+<|ref|>equation<|/ref|><|det|>[[402, 299, 593, 318]]<|/det|>
+\[\mathrm{a_{nanozyme}} = \mathrm{b_{nanozyme}} / [\mathrm{m}]\]  
+
+<|ref|>text<|/ref|><|det|>[[147, 339, 850, 444]]<|/det|>
+where \(\mathrm{a_{nanozyme}}\) is the specific activity expressed in units per milligram (U \(\mathrm{mg}^{- 1}\) ) nanozymes, and [m] is the nanozyme weight (mg) of each assay. In the revised manuscript, we have updated these data according to the Reviewer's suggestion (Page 14, Fig. 4f; Page 16, Line 1- 7; Page 33, line 3- 16).  
+
+<|ref|>text<|/ref|><|det|>[[147, 478, 851, 582]]<|/det|>
+Author response: (4) Mechanism investigations illustrated that pFeSAN underwent a catalytic pathway of the four- electron reduction of \(\mathrm{O}_2\) into \(\mathrm{H}_2\mathrm{O}\) , being identical to that of \(\mathrm{CcO}\) , which was very different from the majority of the previously reported oxidase- like nanozymes with the generation of reactive oxygen species (ROS).  
+
+<|ref|>text<|/ref|><|det|>[[147, 590, 851, 720]]<|/det|>
+Comments: This finding is not novel. As the nanozymes prepared by the authors only showed OXD- like activity, it is expected that there will be no ROS generated. (Similar to Sci. Adv., 2019, 5, eaav5490). If the authors stated it is a similar catalytic pathway to \(\mathrm{CcO}\) , is there any of the reaction intermediates or transient species the authors trapped could be compared to \(\mathrm{CcO}\) ?  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 758, 238, 775]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 784, 851, 887]]<|/det|>
+Thank you for your valuable comments. Previous studies demonstrated that the reactive intermediate of \(\mathrm{Fe(IV) = O}\) was very important for the catalytic oxidative reactions of natural \(\mathrm{CcO}\) (Chem. Rev. 2018, 118, 2491- 2553). As the common oxidant, the intermediate of \(\mathrm{Fe(IV) = O}\) , which usually presents in the catalytic cycle of natural
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[146, 88, 852, 302]]<|/det|>
+oxidases, is considered as the active transient state. To verify the presence of the \(\mathrm{Fe(IV) = O}\) intermediate of \(\mathrm{O_2}\) activation process by pFeSAN, the electron paramagnetic resonance spectrum of the pFeSAN- enabled oxidation with excessive phenyloxoiodine was recorded at 77K (Figure R3). A typical diamond- shaped sign signal at \(\mathrm{g} = 2.03\) , consistent with \(\eta^2\) - peroxy heme species, indicated the formation of \(\mathrm{Fe(IV) = O}\) intermediate in pFeSAN for oxidation (Fig. 5f, Sci. Adv., 2019, 5, eaav5490). Therefore, the oxidase- like activity of pFeSAN proceeds through the \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) pathway, and similar to the reaction process with \(\mathrm{CcO}\) .  
+
+<|ref|>image<|/ref|><|det|>[[348, 336, 630, 531]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[171, 548, 822, 567]]<|/det|>
+<center>Figure R3. EPR spectra of pFeSAN in the presence of phenyloxoiodine at 77 K. </center>  
+
+<|ref|>text<|/ref|><|det|>[[146, 584, 852, 910]]<|/det|>
+Previous study (Sci. Adv., 2019, 5, eaav5490) demonstrated that the \(\mathrm{FeN_5}\) SA/CNF showed oxidase- like activity and no generation of ROS. However, the experimental verification was lacking in that study. In our work, to further explore the electron transfer path during the oxidation, the rotating ring- disk electrode (RRDE) tests were performed. Figure R4 showed that the \(\mathrm{H_2O_2}\) yield of pFeSAN remained below \(7.5\%\) over a wide potential range of \(0.1 - 0.8\mathrm{V}\) . Derived from the RRDE test, the average electron transfer number (n) of pFeSAN was 3.7, indicating the oxygen activation on the pFeSAN through a four- electron oxygen reduction reaction pathway (Nano Energy, 2021, 83, 105798). This process requires four \(\mathrm{H^{+}}\) and four electrons \((\mathrm{O_2 + 4H^{+} + 4e^{- }}\) \(\rightarrow 2\mathrm{H_2O})\) for the complete \(\mathrm{O_2}\) - to- \(\mathrm{H_2O}\) reduction. Thus, the electrochemical understandings of these stepwise proton and electron transfers reveal the essence of pFeSAN for its oxidase- like performance. In our work, the exploration of oxidase- like
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 89, 850, 135]]<|/det|>
+reaction by the RRDE tests would help us to understand the mechanism of oxidase- like reactions.  
+
+<|ref|>image<|/ref|><|det|>[[340, 163, 680, 367]]<|/det|>
+<|ref|>image_caption<|/ref|><|det|>[[147, 381, 850, 419]]<|/det|>
+<center>Figure R4. Calculated electron transfer number derived from rotating ring-disk electrode and \(\mathrm{H}_2\mathrm{O}_2\) yields of the pFeSAN. </center>  
+
+<|ref|>text<|/ref|><|det|>[[148, 437, 850, 483]]<|/det|>
+In the revised manuscript, we have highlighted the related information in the main context (Page 20, Line 1- 9).  
+
+<|ref|>text<|/ref|><|det|>[[148, 519, 851, 594]]<|/det|>
+Author response: (5) pFeSAN as a highly- performed nanozyme exhibited a much higher upper detection limit of GSH at \(1\mathrm{mM}\) , which was 2.5 to 40- fold higher than those of the previously reported investigations (Table R4).  
+
+<|ref|>text<|/ref|><|det|>[[148, 603, 850, 650]]<|/det|>
+Comments: Previously, other systems based on nanozyme activity detected similar or even higher range. (New J. Chem., 2022, 46, 10682- 10689)  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 688, 238, 704]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[147, 714, 852, 900]]<|/det|>
+Thank you for your evaluation. In this GSH detection system (New J. Chem., 2022, 46, 10682- 10689), a novel nanozyme based on the ultrathin two- dimensional metal- organic framework nanomaterial D- ZIF- 67 was prepared and characterized. D- ZIF- 67 exhibited significant oxidase- like activity due to the large specific surface area of the two- dimensional sheet structure as well as the large number of active sites exposed compared to crystalline MOFs. By taking advantage of the excellent property of D- ZIF- 67, authors constructed an effective and sensitive colorimetric sensor for visual GSH
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 90, 733, 108]]<|/det|>
+detection, the detected GSH concentrations ranged between 0.5–10 μM.  
+
+<|ref|>text<|/ref|><|det|>[[147, 116, 853, 469]]<|/det|>
+In our work, pFeSAN provided a wider detection range of 50 nM to 1 mM, higher than those previous literatures (Table R2). The excellent detection performance of GSH by the pFeSAN originates from its synergistic advantages comprising highly dispersed metal single- atoms, the presence of mesopores, and well- regulated coordination environments of the single- atoms. We further utilize pFeSAN- DAB system as a GSH sensor to realize in vitro quantitative GSH visualization for Hep 1- 6 cells and tumor tissue with high GSH states have been accurately distinguished by visualization detection. Notably, methods for the visualization detection of tumor tissue though GSH has not been reported yet, which brought the possibility to realize specific detection in practical applications. The visualized analysis of the GSH in tumor tissue is gaining interests to improve surgical safety and to promote surgical therapeutic effects. We have checked the literature carefully and added this literature in the revised manuscript to support this work (Ref. 57: New J. Chem., 46, 10682–10689 (2022)).  
+
+<|ref|>table<|/ref|><|det|>[[149, 536, 847, 764]]<|/det|>
+<|ref|>table_caption<|/ref|><|det|>[[147, 488, 852, 525]]<|/det|>
+Table R2. Comparison of our approach with other colorimetric detection systems of GSH.   
+
+<table><tr><td>Materials</td><td>Linear range</td><td>LOD</td></tr><tr><td>pFeSAN</td><td>0.05-1000 μM</td><td>0.0024 μM</td></tr><tr><td>AuNPs</td><td>1-40 μM</td><td>0.013 μM</td></tr><tr><td>Au nanoclusters</td><td>2-25 μM</td><td>0.42 μM</td></tr><tr><td>PSMOF</td><td>1-20 μM</td><td>0.68 μM</td></tr><tr><td>Acre+-Mes</td><td>0.1-40 μM</td><td>0.1 μM</td></tr><tr><td>MnO2</td><td>0.3-15 μM</td><td>0.11 μM</td></tr><tr><td>TiO2/MoS2</td><td>0.05-1 μM</td><td>0.05 μM</td></tr><tr><td>Fe-N-C SANs</td><td>100-400 μM</td><td>78.3 μM</td></tr></table>
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 85, 294, 98]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[118, 113, 684, 155]]<|/det|>
+Reviewer #4:  Remarks to the Author:  The authors have addressed the comments. There is no further comment.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[280, 90, 715, 108]]<|/det|>
+## Point-by-point Response to Reviewers' Comments  
+
+<|ref|>text<|/ref|><|det|>[[148, 132, 465, 150]]<|/det|>
+Reviewer #4 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 186, 736, 204]]<|/det|>
+The authors have addressed the comments. There is no further comment.  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 244, 238, 260]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[148, 271, 850, 317]]<|/det|>
+We appreciate the Reviewer's positive comment and valuable suggestions, which helped us improve our manuscript.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1f0e1a3d63db588f869f132183e57a3969bf75e622f40af39e54d489d558c4e2/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1f0e1a3d63db588f869f132183e57a3969bf75e622f40af39e54d489d558c4e2/supplementary_0_Peer Review File__1f0e1a3d63db588f869f132183e57a3969bf75e622f40af39e54d489d558c4e2.mmd

@@ -0,0 +1,158 @@
+
+## REVIEWER COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+I recommend the paper entitled "Superconductivity beyond Pauli's limit in bulk NbS2: Evidence for the Fulde- Ferrell- Larkin Ovchinnikov state" Chang- woo Cho et al. for publication.  
+
+However, I think the authors should consider the following revisions:  
+
+1)  
+
+The main evidence for an FFLO state comes from an upturn in the experimental Hc2 curve shown in Figure 5. This is then compared to a prediction for a model for Ising superconductivity (solid green curve). In my opinion, the paper does not give enough detail on how the Ising limit was calculated. Ref. 38 only explains tight binding band structure calculations for dichalcogenides. I suggest adding more details on how to go from Ref. 38 to the solid green line of Fig. 5.  
+
+2)  
+
+I fail to see the relevance of the mauve crosses and teal solid circles. I can see that these curves level off at the Pauli limit. However, is there a particular reason this should happen?  
+
+I suggest to give some explanation.  
+
+3)  
+
+Open circles in Fig. 1, and the orange stars in Fig. 4b for magnetostriction. I believe some discussion on how the position of the phase transition was determined from these measurements is appropriate, as the transitions are rather subtle.  
+
+4)  
+
+Abstract "We present magnetic torque, specific heat and thermal expansion measurements combined with a piezo rotary positioner of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields applied strictly parallel to its layer structure. The upper critical field of superconducting TMDs in the 2D form is known to be dramatically enhanced by a special form of Ising spin orbit coupling. This Ising superconductivity is very robust against the Pauli limit for superconductivity."  
+
+The phrase with the piezo rotary positioner requires an explanation on why such a device is needed. The last phrase needs to be rewritten.  
+
+5)  
+
+"Among them are intrinsic superconductors with 2H- NbSe2 and 2H- NbS2 as representatives of the highest critical temperatures."  
+
+With respect to what are the Tc compared?  
+
+6)  
+
+"In their 2D form, they have aroused great interest due to the discovery of Ising superconductivity, which allows them to withstand the Pauli limit of superconductivity [4,5]."  
+
+Withstand should probably be replaced with exceed.  
+
+7)  
+
+"In reality bilayers have Hc2 values, which still exceed the Pauli limit by a factor of \(\sim 4\) [5]." From the way this sentence is written it is not clear if it is good or bad that Hc2 exceeds the Pauli limit of ISOI superconductors.
+
+<--- Page Split --->
+
+Reviewer #2 (Remarks to the Author):  
+
+The present manuscript describes magnetic torque, specific heat and thermal expansion measurements on single crystalline NbS2. The results suggest an enhancement of the upper critical field Hc2 above the so- called Pauli limit which the authors attribute to the presence of the Fulde- Ferrell- Larkin- Ovchinnikov (FFLO) state.  
+
+The FFLO state is an exotic form of superconductivity and as such, evidence for such a state has been must sought after by experimentalists. The number of cases for which this evidence is considered robust, however, is rather small and as such, one should tread carefully when considering candidate materials. Provided that evidence is robust, however, the more experimental realizations of the FFLO state are made, the more opportunities there will be to study this exotic state.  
+
+In my opinion, the evidence presented in the current manuscript is reasonably convincing and therefore, I am tempted to recommend its publication. Prior to that, however, I suggest the authors seriously consider the following additions to their manuscript:  
+
+(i) Comparative measurements for H//c, in particular of the T-dependence of Hc2, to confirm whether or not the upturn is observed in this field orientation. As stated by the authors, NbS2 is a multi-pocket conductor. As such, there is always the possibility of that the upturn in Hc2 is a signature of multiband superconductivity. Measurements in the orthogonal direction should help address whether or not the upturn reported in this manuscript is solely induced by in-plane magnetic fields.  
+
+(ii) One of the criterion for the observation of FFLO state is a that the superconductor is in the ultraclean limit, meaning that the mean free path is significantly longer than the corresponding coherence length. Discussion of this point should also be included in the present manuscript.  
+
+Provided these two aspects are addressed, I would be happy to consider its publication in Nature Communications.  
+
+I note, in passing, that the most recent evidence for the FFLO state was reported in FeSe by Kasahara et al. (PRL vol. 124, 107001 (2020)) not in KFe2As2.
+
+<--- Page Split --->
+
+## Responses to Reviewer #1  
+
+Comment: I recommend the paper entitled "Superconductivity beyond Pauli's limit in bulk NbS2: Evidence for the Fulde- Ferrell- Larkin Ovchinnikov state" Chang- woo Cho et al. for publication.  
+
+Our reply: We thank the reviewer for the positive evaluation!  
+
+Comment: The main evidence for an FFLO state comes from an upturn in the experimental Hc2 curve shown in Figure 5. This is then compared to a prediction for a model for Ising superconductivity (solid green curve). In my opinion, the paper does not give enough detail on how the Ising limit was calculated. Ref. 38 only explains tight binding band structure calculations for dichalcogenides. I suggest adding more details on how to go from Ref. 38 to the solid green line of Fig. 5.  
+
+Our reply: The green curve of \(\mathsf{H}_{c2}\) in Fig. 5 is obtained by numerically solving the linearized gap equation, which can be written as  
+
+\[\frac{1}{g} = T\int_{BZ}d^2\pmb {k}\sum_{n,l,m}\frac{|\langle l,k|\delta H_{\Delta}|m, - \pmb {k}\rangle|^2}{[i\omega_n - \xi_l(\pmb {k})][i\omega_n + \xi_m(-\pmb {k})]} \quad (R1)\]  
+
+where \(\omega_{n} = \pi T(2n + 1)\) is the Matsubara frequency, \(\xi_{m}(\pmb {k})\) is the \(m\) - th eigenenergy of the normal Hamiltonian (including the Zeeman field) of wave vector \(\pmb{k}\) with \(|m,\pmb {k}\rangle\) being the corresponding eigenstate, and \(H_{\Delta}\) is the matrix form of the pairing terms in the Hamiltonian  
+
+\[\delta \hat{H}_{\Delta} = \sum_{j}\psi_{j\uparrow}^{\dagger}(\pmb {k})\psi_{j\downarrow}^{\dagger}(-\pmb {k}).\]  
+
+The index \(j\) includes the layer index and the band index. With the number of layers being ten and the number of bands in each layer being 3, the dimension of the matrix \(\delta H_{\Delta}\) is 60 by 60 (with a factor of two from the spin). For a certain interaction strength \(g\) , Eq. (R1) gives the relation between the temperature \(T\) and the critical field \(H_{c2}\) . The actual \(g\) is determined by the measured critical temperature \(T_{c}\) at zero field.  
+
+We have added these details to the Methods section (page 7, from line 10).  
+
+Comment: I fail to see the relevance of the mauve crosses and teal solid circles. I can see that these curves level off at the Pauli limit. However, is there a particular reason this should happen? I suggest to give some explanation.  
+
+Our reply: The hallmark of the FFLO state is not only the Hc2 upturn at low temperatures, in addition a phase transition should occur that separates the ordinary low- field superconducting phase from the high- field FFLO state. This transition is expected to occur not far from the line representing the Pauli limit, although the exact position may vary depending on the material parameters. The data mentioned highlight features that indicate this phase transition within the superconducting state. It indeed occurs near the Pauli limit. To highlight this more clearly, we have marked the regime of the FFLO state in the phase diagram (revised Fig. 5. ) with a magenta shaded background. We have also simplified the diagram by focusing on the exact parallel orientation, while data in tilted fields have been moved to an inset.
+
+<--- Page Split --->
+
+The mauve star together with the magenta hexagon are taken from field sweep data of the magnetostriction and specific heat showing a small phase transition anomaly at about \(10 \text{T}\) at \(\text{T} = 0.3 \text{K}\) . The corresponding data are shown in Fig. 4a and we have inserted the same symbols into these revised graphs to highlight the transition field in Fig. 4 (see also page 3, line 39- 42). The teal- colored solid circles are from anomalies in the magnetic torque, and a selection of corresponding data is shown in Fig. 4b. Again, we have used the same symbols as in Fig. 5 in the revised plots to mark the anomalies in Fig. 4b. Another curve (magenta crosses) in the plot results from the torque data measured in a field tilted by 1 degree, as shown in Fig. 1, which also flattens out near the Pauli limit. This implies that the FFLO state is no longer formed in such a titled field and the data remain Pauli limited. In the revised manuscript we explain this more clearly (page 4, line 30- 34 and caption of Fig. 5).  
+
+Comment: Open circles in Fig. 1, and the orange stars in Fig. 4b for magnetostriction. I believe some discussion on how the position of the phase transition was determined from these measurements is appropriate, as the transitions are rather subtle.  
+
+Our reply: Concerning Fig. 1: For this tilted field data, there is in principle only one phase transition occurring at \(\text{Hc2}\) , and the steep step- like transition in the torque likely indicates a first- order Pauli- limited nature. We therefore decided to use the points where the steepest slope occurs near \(\text{Hc2}\) , as mentioned in the caption, which should best represent the first- order transition. Note that other criteria (e.g. the very continuous small transition onset) only lead to an almost parallel shift of the phase transition line. In the revised manuscript we explain more clearly why we use this criterion (page 3, line 1- 4). This is different from the exactly parallel alignment, where the much sharper kink- like onset (see Fig. 2) is the only sharp feature that occurs.  
+
+Regarding Fig. 4b, there is no orange star in Fig. 4b and the data in this graph represents torque data and not magnetostriction. The reviewer is probably referring to the orange star in Fig. 5, which was determined from Fig. 3. It represents thermal expansion data measured during a temperature sweep. We believe this transition is fairly sharply defined, and we used the midpoint of the jump to determine the transition temperature. We have added the orange star in Fig. 3 (as well as the magenta hexagon for Cp) to mark the transition temperatures, hopefully clarifying the origin of this symbol in Fig. 5.  
+
+Comment: Abstract "We present magnetic torque, specific heat and thermal expansion measurements combined with a piezo rotary positioner of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields applied strictly parallel to its layer structure. The upper critical field of superconducting TMDs in the 2D form is known to be dramatically enhanced by a special form of Ising spin orbit coupling. This Ising superconductivity is very robust against the Pauli limit for superconductivity."  
+
+The phrase with the piezo rotary positioner requires an explanation on why such a device is needed. The last phrase needs to be rewritten.  
+
+Our reply: We thank the reviewer for pointing out our unclear wording in the abstract. Observing the FFLO state in layered compounds often requires strictly parallel fields, and, as can be seen from our data in Fig. 1, even 1 degree of misalignment causes superconductivity to no longer persist beyond the Pauli limit. The piezo rotator is therefore essential to align the layered structure of the sample with the field prior to the measurements. We have rephrased the abstract as follows: "We present measurements of
+
+<--- Page Split --->
+
+the magnetic torque, specific heat and thermal expansion of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields, with its layer structure aligned strictly parallel to the field using a piezo rotary positioner. ... superconductivity beyond the Pauli limit still exists in bulk single crystals of NbS2 for a precisely parallel field alignment."  
+
+The other sentence mentioned by the reviewer has been rewritten as: "This Ising superconductivity is very robust to the Pauli paramagnetic effect and can therefore exist beyond the Pauli limit for superconductivity."  
+
+Comment: "Among them are intrinsic superconductors with 2H- NbSe2 and 2H- NbS2 as representatives of the highest critical temperatures." With respect to what are the Tc compared?  
+
+Our reply: The intrinsic transition metal superconductors are mainly NbSe2, NbS2 and TaS2, with TaS2 having a much lower Tc. We have removed the second half of the sentence, which now simply reads like: "Among them are intrinsic superconductors including 2H- NbSe2 and 2H- NbS2" (page 1, line 25).  
+
+Comment: "In their 2D form, they have aroused great interest due to the discovery of Ising superconductivity, which allows them to withstand the Pauli limit of superconductivity [4,5]." Withstand should probably be replaced with exceed.  
+
+Our reply: We thank the reviewer for this suggestion, we have replaced it accordingly.  
+
+## Comment:  
+
+"In reality bilayers have Hc2 values, which still exceed the Pauli limit by a factor of \(\sim 4\) [5]." From the way this sentence is written it is not clear if it is good or bad that Hc2 exceeds the Pauli limit of ISOI superconductors.  
+
+Our reply: It was not our intention to make a statement about whether it is good or bad. What we wanted to express is that Ising superconductivity is not limited to monolayers but still occurs in multiple layers, and it remains unclear at what thickness it is suppressed. In this sense, it is not clear whether Ising superconductivity could still exist to some extent in the limit of bulk samples. We have rewritten this sentence to express this more clearly (page 5, line 10 - 15).  
+
+## Responses to Reviewer #2  
+
+Comment: In my opinion, the evidence presented in the current manuscript is reasonably convincing and therefore, I am tempted to recommend its publication. Prior to that, however, I suggest the authors seriously consider the following additions to their manuscript:  
+
+Our reply: We thank the reviewer for the positive review!
+
+<--- Page Split --->
+
+Comment: Comparative measurements for \(H / / c\) , in particular of the \(T\) - dependence of \(Hc2\) , to confirm whether or not the upturn is observed in this field orientation. As stated by the authors, \(NbS2\) is a multipocket conductor. As such, there is always the possibility of that the upturn in \(Hc2\) is a signature of multiband superconductivity. Measurements in the orthogonal direction should help address whether or not the upturn reported in this manuscript is solely induced by in- plane magnetic fields.  
+
+Our reply: We did not focus on perpendicular fields, as such data are available in the literature (e.g. PRB 82, 014518 (2010)), but we quickly performed such measurements with both SQUID and torque magnetometry. The data are included in the inset of revised Fig. 5 of the manuscript. These data show no upturn at low temperatures, which we comment on in the revised manuscript (page 4, line 30- 34).  
+
+Comment: One of the criterion for the observation of FFLO state is a that the superconductor is in the ultra- clean limit, meaning that the mean free path is significantly longer than the corresponding coherence length. Discussion of this point should also be included in the present manuscript.  
+
+Our reply: We have included this statement, along with a reference, in the revised manuscript (page2, line 9 - 12).  
+
+Comment: I note, in passing, that the most recent evidence for the FFLO state was reported in FeSe by Kasahara et al. (PRL vol. 124, 107001 (2020)) not in KFe2As2.  
+
+Our reply: We thank the reviewer, yes of course this work should be cited and we have added the reference (page 2, line 9).
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+I find the authors have addressed all the points of my previous review and the manuscript should be published.  
+
+Reviewer #2 (Remarks to the Author):  
+
+I have looked at the response of the authors to the previous referee reports and the revised manuscript and I am satisfied that they have addressed the main concerns of the referees and have improved the clarity of the manuscript accordingly. I am therefore happy to recommend publication of the paper in its present format. It is remarkable to me how the original predictions of Fulde, Ferrell, Larkin and Ovchinnikov have come to be realized in such diverse families of superconductors. This manuscript adds the dichalcogenides to that list.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1f0e1a3d63db588f869f132183e57a3969bf75e622f40af39e54d489d558c4e2/supplementary_0_Peer Review File__1f0e1a3d63db588f869f132183e57a3969bf75e622f40af39e54d489d558c4e2_det.mmd

@@ -0,0 +1,223 @@
+<|ref|>sub_title<|/ref|><|det|>[[116, 90, 287, 104]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 121, 404, 137]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 153, 876, 185]]<|/det|>
+I recommend the paper entitled "Superconductivity beyond Pauli's limit in bulk NbS2: Evidence for the Fulde- Ferrell- Larkin Ovchinnikov state" Chang- woo Cho et al. for publication.  
+
+<|ref|>text<|/ref|><|det|>[[116, 201, 627, 216]]<|/det|>
+However, I think the authors should consider the following revisions:  
+
+<|ref|>text<|/ref|><|det|>[[116, 234, 133, 247]]<|/det|>
+1)  
+
+<|ref|>text<|/ref|><|det|>[[115, 249, 860, 326]]<|/det|>
+The main evidence for an FFLO state comes from an upturn in the experimental Hc2 curve shown in Figure 5. This is then compared to a prediction for a model for Ising superconductivity (solid green curve). In my opinion, the paper does not give enough detail on how the Ising limit was calculated. Ref. 38 only explains tight binding band structure calculations for dichalcogenides. I suggest adding more details on how to go from Ref. 38 to the solid green line of Fig. 5.  
+
+<|ref|>text<|/ref|><|det|>[[115, 344, 133, 357]]<|/det|>
+2)  
+
+<|ref|>text<|/ref|><|det|>[[115, 360, 875, 390]]<|/det|>
+I fail to see the relevance of the mauve crosses and teal solid circles. I can see that these curves level off at the Pauli limit. However, is there a particular reason this should happen?  
+
+<|ref|>text<|/ref|><|det|>[[116, 392, 382, 406]]<|/det|>
+I suggest to give some explanation.  
+
+<|ref|>text<|/ref|><|det|>[[115, 424, 133, 437]]<|/det|>
+3)  
+
+<|ref|>text<|/ref|><|det|>[[115, 439, 874, 485]]<|/det|>
+Open circles in Fig. 1, and the orange stars in Fig. 4b for magnetostriction. I believe some discussion on how the position of the phase transition was determined from these measurements is appropriate, as the transitions are rather subtle.  
+
+<|ref|>text<|/ref|><|det|>[[115, 503, 133, 516]]<|/det|>
+4)  
+
+<|ref|>text<|/ref|><|det|>[[115, 518, 875, 612]]<|/det|>
+Abstract "We present magnetic torque, specific heat and thermal expansion measurements combined with a piezo rotary positioner of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields applied strictly parallel to its layer structure. The upper critical field of superconducting TMDs in the 2D form is known to be dramatically enhanced by a special form of Ising spin orbit coupling. This Ising superconductivity is very robust against the Pauli limit for superconductivity."  
+
+<|ref|>text<|/ref|><|det|>[[115, 614, 863, 646]]<|/det|>
+The phrase with the piezo rotary positioner requires an explanation on why such a device is needed. The last phrase needs to be rewritten.  
+
+<|ref|>text<|/ref|><|det|>[[115, 664, 133, 677]]<|/det|>
+5)  
+
+<|ref|>text<|/ref|><|det|>[[115, 679, 852, 710]]<|/det|>
+"Among them are intrinsic superconductors with 2H- NbSe2 and 2H- NbS2 as representatives of the highest critical temperatures."  
+
+<|ref|>text<|/ref|><|det|>[[123, 711, 445, 725]]<|/det|>
+With respect to what are the Tc compared?  
+
+<|ref|>text<|/ref|><|det|>[[115, 744, 133, 757]]<|/det|>
+6)  
+
+<|ref|>text<|/ref|><|det|>[[115, 760, 857, 789]]<|/det|>
+"In their 2D form, they have aroused great interest due to the discovery of Ising superconductivity, which allows them to withstand the Pauli limit of superconductivity [4,5]."  
+
+<|ref|>text<|/ref|><|det|>[[120, 791, 510, 805]]<|/det|>
+Withstand should probably be replaced with exceed.  
+
+<|ref|>text<|/ref|><|det|>[[115, 824, 133, 837]]<|/det|>
+7)  
+
+<|ref|>text<|/ref|><|det|>[[115, 840, 875, 887]]<|/det|>
+"In reality bilayers have Hc2 values, which still exceed the Pauli limit by a factor of \(\sim 4\) [5]." From the way this sentence is written it is not clear if it is good or bad that Hc2 exceeds the Pauli limit of ISOI superconductors.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 121, 404, 136]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[116, 152, 860, 216]]<|/det|>
+The present manuscript describes magnetic torque, specific heat and thermal expansion measurements on single crystalline NbS2. The results suggest an enhancement of the upper critical field Hc2 above the so- called Pauli limit which the authors attribute to the presence of the Fulde- Ferrell- Larkin- Ovchinnikov (FFLO) state.  
+
+<|ref|>text<|/ref|><|det|>[[115, 231, 874, 311]]<|/det|>
+The FFLO state is an exotic form of superconductivity and as such, evidence for such a state has been must sought after by experimentalists. The number of cases for which this evidence is considered robust, however, is rather small and as such, one should tread carefully when considering candidate materials. Provided that evidence is robust, however, the more experimental realizations of the FFLO state are made, the more opportunities there will be to study this exotic state.  
+
+<|ref|>text<|/ref|><|det|>[[115, 326, 864, 374]]<|/det|>
+In my opinion, the evidence presented in the current manuscript is reasonably convincing and therefore, I am tempted to recommend its publication. Prior to that, however, I suggest the authors seriously consider the following additions to their manuscript:  
+
+<|ref|>text<|/ref|><|det|>[[115, 389, 878, 469]]<|/det|>
+(i) Comparative measurements for H//c, in particular of the T-dependence of Hc2, to confirm whether or not the upturn is observed in this field orientation. As stated by the authors, NbS2 is a multi-pocket conductor. As such, there is always the possibility of that the upturn in Hc2 is a signature of multiband superconductivity. Measurements in the orthogonal direction should help address whether or not the upturn reported in this manuscript is solely induced by in-plane magnetic fields.  
+
+<|ref|>text<|/ref|><|det|>[[115, 485, 870, 532]]<|/det|>
+(ii) One of the criterion for the observation of FFLO state is a that the superconductor is in the ultraclean limit, meaning that the mean free path is significantly longer than the corresponding coherence length. Discussion of this point should also be included in the present manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 548, 839, 580]]<|/det|>
+Provided these two aspects are addressed, I would be happy to consider its publication in Nature Communications.  
+
+<|ref|>text<|/ref|><|det|>[[115, 596, 875, 627]]<|/det|>
+I note, in passing, that the most recent evidence for the FFLO state was reported in FeSe by Kasahara et al. (PRL vol. 124, 107001 (2020)) not in KFe2As2.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[115, 91, 312, 107]]<|/det|>
+## Responses to Reviewer #1  
+
+<|ref|>text<|/ref|><|det|>[[115, 125, 883, 163]]<|/det|>
+Comment: I recommend the paper entitled "Superconductivity beyond Pauli's limit in bulk NbS2: Evidence for the Fulde- Ferrell- Larkin Ovchinnikov state" Chang- woo Cho et al. for publication.  
+
+<|ref|>text<|/ref|><|det|>[[115, 181, 566, 199]]<|/det|>
+Our reply: We thank the reviewer for the positive evaluation!  
+
+<|ref|>text<|/ref|><|det|>[[114, 217, 883, 310]]<|/det|>
+Comment: The main evidence for an FFLO state comes from an upturn in the experimental Hc2 curve shown in Figure 5. This is then compared to a prediction for a model for Ising superconductivity (solid green curve). In my opinion, the paper does not give enough detail on how the Ising limit was calculated. Ref. 38 only explains tight binding band structure calculations for dichalcogenides. I suggest adding more details on how to go from Ref. 38 to the solid green line of Fig. 5.  
+
+<|ref|>text<|/ref|><|det|>[[115, 327, 886, 362]]<|/det|>
+Our reply: The green curve of \(\mathsf{H}_{c2}\) in Fig. 5 is obtained by numerically solving the linearized gap equation, which can be written as  
+
+<|ref|>equation<|/ref|><|det|>[[313, 359, 880, 407]]<|/det|>
+\[\frac{1}{g} = T\int_{BZ}d^2\pmb {k}\sum_{n,l,m}\frac{|\langle l,k|\delta H_{\Delta}|m, - \pmb {k}\rangle|^2}{[i\omega_n - \xi_l(\pmb {k})][i\omega_n + \xi_m(-\pmb {k})]} \quad (R1)\]  
+
+<|ref|>text<|/ref|><|det|>[[115, 414, 833, 470]]<|/det|>
+where \(\omega_{n} = \pi T(2n + 1)\) is the Matsubara frequency, \(\xi_{m}(\pmb {k})\) is the \(m\) - th eigenenergy of the normal Hamiltonian (including the Zeeman field) of wave vector \(\pmb{k}\) with \(|m,\pmb {k}\rangle\) being the corresponding eigenstate, and \(H_{\Delta}\) is the matrix form of the pairing terms in the Hamiltonian  
+
+<|ref|>equation<|/ref|><|det|>[[395, 476, 601, 522]]<|/det|>
+\[\delta \hat{H}_{\Delta} = \sum_{j}\psi_{j\uparrow}^{\dagger}(\pmb {k})\psi_{j\downarrow}^{\dagger}(-\pmb {k}).\]  
+
+<|ref|>text<|/ref|><|det|>[[115, 528, 880, 601]]<|/det|>
+The index \(j\) includes the layer index and the band index. With the number of layers being ten and the number of bands in each layer being 3, the dimension of the matrix \(\delta H_{\Delta}\) is 60 by 60 (with a factor of two from the spin). For a certain interaction strength \(g\) , Eq. (R1) gives the relation between the temperature \(T\) and the critical field \(H_{c2}\) . The actual \(g\) is determined by the measured critical temperature \(T_{c}\) at zero field.  
+
+<|ref|>text<|/ref|><|det|>[[115, 610, 667, 628]]<|/det|>
+We have added these details to the Methods section (page 7, from line 10).  
+
+<|ref|>text<|/ref|><|det|>[[115, 674, 883, 730]]<|/det|>
+Comment: I fail to see the relevance of the mauve crosses and teal solid circles. I can see that these curves level off at the Pauli limit. However, is there a particular reason this should happen? I suggest to give some explanation.  
+
+<|ref|>text<|/ref|><|det|>[[114, 747, 883, 895]]<|/det|>
+Our reply: The hallmark of the FFLO state is not only the Hc2 upturn at low temperatures, in addition a phase transition should occur that separates the ordinary low- field superconducting phase from the high- field FFLO state. This transition is expected to occur not far from the line representing the Pauli limit, although the exact position may vary depending on the material parameters. The data mentioned highlight features that indicate this phase transition within the superconducting state. It indeed occurs near the Pauli limit. To highlight this more clearly, we have marked the regime of the FFLO state in the phase diagram (revised Fig. 5. ) with a magenta shaded background. We have also simplified the diagram by focusing on the exact parallel orientation, while data in tilted fields have been moved to an inset.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[114, 89, 883, 272]]<|/det|>
+The mauve star together with the magenta hexagon are taken from field sweep data of the magnetostriction and specific heat showing a small phase transition anomaly at about \(10 \text{T}\) at \(\text{T} = 0.3 \text{K}\) . The corresponding data are shown in Fig. 4a and we have inserted the same symbols into these revised graphs to highlight the transition field in Fig. 4 (see also page 3, line 39- 42). The teal- colored solid circles are from anomalies in the magnetic torque, and a selection of corresponding data is shown in Fig. 4b. Again, we have used the same symbols as in Fig. 5 in the revised plots to mark the anomalies in Fig. 4b. Another curve (magenta crosses) in the plot results from the torque data measured in a field tilted by 1 degree, as shown in Fig. 1, which also flattens out near the Pauli limit. This implies that the FFLO state is no longer formed in such a titled field and the data remain Pauli limited. In the revised manuscript we explain this more clearly (page 4, line 30- 34 and caption of Fig. 5).  
+
+<|ref|>text<|/ref|><|det|>[[115, 291, 882, 344]]<|/det|>
+Comment: Open circles in Fig. 1, and the orange stars in Fig. 4b for magnetostriction. I believe some discussion on how the position of the phase transition was determined from these measurements is appropriate, as the transitions are rather subtle.  
+
+<|ref|>text<|/ref|><|det|>[[114, 364, 883, 509]]<|/det|>
+Our reply: Concerning Fig. 1: For this tilted field data, there is in principle only one phase transition occurring at \(\text{Hc2}\) , and the steep step- like transition in the torque likely indicates a first- order Pauli- limited nature. We therefore decided to use the points where the steepest slope occurs near \(\text{Hc2}\) , as mentioned in the caption, which should best represent the first- order transition. Note that other criteria (e.g. the very continuous small transition onset) only lead to an almost parallel shift of the phase transition line. In the revised manuscript we explain more clearly why we use this criterion (page 3, line 1- 4). This is different from the exactly parallel alignment, where the much sharper kink- like onset (see Fig. 2) is the only sharp feature that occurs.  
+
+<|ref|>text<|/ref|><|det|>[[114, 518, 883, 626]]<|/det|>
+Regarding Fig. 4b, there is no orange star in Fig. 4b and the data in this graph represents torque data and not magnetostriction. The reviewer is probably referring to the orange star in Fig. 5, which was determined from Fig. 3. It represents thermal expansion data measured during a temperature sweep. We believe this transition is fairly sharply defined, and we used the midpoint of the jump to determine the transition temperature. We have added the orange star in Fig. 3 (as well as the magenta hexagon for Cp) to mark the transition temperatures, hopefully clarifying the origin of this symbol in Fig. 5.  
+
+<|ref|>text<|/ref|><|det|>[[114, 646, 883, 754]]<|/det|>
+Comment: Abstract "We present magnetic torque, specific heat and thermal expansion measurements combined with a piezo rotary positioner of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields applied strictly parallel to its layer structure. The upper critical field of superconducting TMDs in the 2D form is known to be dramatically enhanced by a special form of Ising spin orbit coupling. This Ising superconductivity is very robust against the Pauli limit for superconductivity."  
+
+<|ref|>text<|/ref|><|det|>[[115, 755, 880, 790]]<|/det|>
+The phrase with the piezo rotary positioner requires an explanation on why such a device is needed. The last phrase needs to be rewritten.  
+
+<|ref|>text<|/ref|><|det|>[[114, 810, 883, 900]]<|/det|>
+Our reply: We thank the reviewer for pointing out our unclear wording in the abstract. Observing the FFLO state in layered compounds often requires strictly parallel fields, and, as can be seen from our data in Fig. 1, even 1 degree of misalignment causes superconductivity to no longer persist beyond the Pauli limit. The piezo rotator is therefore essential to align the layered structure of the sample with the field prior to the measurements. We have rephrased the abstract as follows: "We present measurements of
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 882, 162]]<|/det|>
+the magnetic torque, specific heat and thermal expansion of the bulk transition metal dichalcogenide (TMD) superconductor NbS2 in high magnetic fields, with its layer structure aligned strictly parallel to the field using a piezo rotary positioner. ... superconductivity beyond the Pauli limit still exists in bulk single crystals of NbS2 for a precisely parallel field alignment."  
+
+<|ref|>text<|/ref|><|det|>[[115, 171, 882, 225]]<|/det|>
+The other sentence mentioned by the reviewer has been rewritten as: "This Ising superconductivity is very robust to the Pauli paramagnetic effect and can therefore exist beyond the Pauli limit for superconductivity."  
+
+<|ref|>text<|/ref|><|det|>[[115, 262, 882, 298]]<|/det|>
+Comment: "Among them are intrinsic superconductors with 2H- NbSe2 and 2H- NbS2 as representatives of the highest critical temperatures." With respect to what are the Tc compared?  
+
+<|ref|>text<|/ref|><|det|>[[115, 317, 882, 371]]<|/det|>
+Our reply: The intrinsic transition metal superconductors are mainly NbSe2, NbS2 and TaS2, with TaS2 having a much lower Tc. We have removed the second half of the sentence, which now simply reads like: "Among them are intrinsic superconductors including 2H- NbSe2 and 2H- NbS2" (page 1, line 25).  
+
+<|ref|>text<|/ref|><|det|>[[115, 390, 882, 444]]<|/det|>
+Comment: "In their 2D form, they have aroused great interest due to the discovery of Ising superconductivity, which allows them to withstand the Pauli limit of superconductivity [4,5]." Withstand should probably be replaced with exceed.  
+
+<|ref|>text<|/ref|><|det|>[[115, 464, 744, 482]]<|/det|>
+Our reply: We thank the reviewer for this suggestion, we have replaced it accordingly.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 502, 196, 517]]<|/det|>
+## Comment:  
+
+<|ref|>text<|/ref|><|det|>[[115, 519, 882, 572]]<|/det|>
+"In reality bilayers have Hc2 values, which still exceed the Pauli limit by a factor of \(\sim 4\) [5]." From the way this sentence is written it is not clear if it is good or bad that Hc2 exceeds the Pauli limit of ISOI superconductors.  
+
+<|ref|>text<|/ref|><|det|>[[115, 593, 882, 683]]<|/det|>
+Our reply: It was not our intention to make a statement about whether it is good or bad. What we wanted to express is that Ising superconductivity is not limited to monolayers but still occurs in multiple layers, and it remains unclear at what thickness it is suppressed. In this sense, it is not clear whether Ising superconductivity could still exist to some extent in the limit of bulk samples. We have rewritten this sentence to express this more clearly (page 5, line 10 - 15).  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 739, 312, 755]]<|/det|>
+## Responses to Reviewer #2  
+
+<|ref|>text<|/ref|><|det|>[[115, 775, 882, 829]]<|/det|>
+Comment: In my opinion, the evidence presented in the current manuscript is reasonably convincing and therefore, I am tempted to recommend its publication. Prior to that, however, I suggest the authors seriously consider the following additions to their manuscript:  
+
+<|ref|>text<|/ref|><|det|>[[115, 849, 538, 865]]<|/det|>
+Our reply: We thank the reviewer for the positive review!
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 883, 180]]<|/det|>
+Comment: Comparative measurements for \(H / / c\) , in particular of the \(T\) - dependence of \(Hc2\) , to confirm whether or not the upturn is observed in this field orientation. As stated by the authors, \(NbS2\) is a multipocket conductor. As such, there is always the possibility of that the upturn in \(Hc2\) is a signature of multiband superconductivity. Measurements in the orthogonal direction should help address whether or not the upturn reported in this manuscript is solely induced by in- plane magnetic fields.  
+
+<|ref|>text<|/ref|><|det|>[[115, 199, 883, 272]]<|/det|>
+Our reply: We did not focus on perpendicular fields, as such data are available in the literature (e.g. PRB 82, 014518 (2010)), but we quickly performed such measurements with both SQUID and torque magnetometry. The data are included in the inset of revised Fig. 5 of the manuscript. These data show no upturn at low temperatures, which we comment on in the revised manuscript (page 4, line 30- 34).  
+
+<|ref|>text<|/ref|><|det|>[[115, 291, 883, 345]]<|/det|>
+Comment: One of the criterion for the observation of FFLO state is a that the superconductor is in the ultra- clean limit, meaning that the mean free path is significantly longer than the corresponding coherence length. Discussion of this point should also be included in the present manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 364, 883, 400]]<|/det|>
+Our reply: We have included this statement, along with a reference, in the revised manuscript (page2, line 9 - 12).  
+
+<|ref|>text<|/ref|><|det|>[[115, 437, 883, 472]]<|/det|>
+Comment: I note, in passing, that the most recent evidence for the FFLO state was reported in FeSe by Kasahara et al. (PRL vol. 124, 107001 (2020)) not in KFe2As2.  
+
+<|ref|>text<|/ref|><|det|>[[115, 491, 883, 527]]<|/det|>
+Our reply: We thank the reviewer, yes of course this work should be cited and we have added the reference (page 2, line 9).
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[116, 88, 301, 103]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 120, 404, 135]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[112, 152, 866, 184]]<|/det|>
+I find the authors have addressed all the points of my previous review and the manuscript should be published.  
+
+<|ref|>text<|/ref|><|det|>[[116, 215, 404, 230]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 247, 880, 343]]<|/det|>
+I have looked at the response of the authors to the previous referee reports and the revised manuscript and I am satisfied that they have addressed the main concerns of the referees and have improved the clarity of the manuscript accordingly. I am therefore happy to recommend publication of the paper in its present format. It is remarkable to me how the original predictions of Fulde, Ferrell, Larkin and Ovchinnikov have come to be realized in such diverse families of superconductors. This manuscript adds the dichalcogenides to that list.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1f6c8e672b6887aeacdaec3615e7a8c9aed8a2fd462da50a662918cb10006724/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1f6c8e672b6887aeacdaec3615e7a8c9aed8a2fd462da50a662918cb10006724/supplementary_0_Peer Review File__1f6c8e672b6887aeacdaec3615e7a8c9aed8a2fd462da50a662918cb10006724.mmd

@@ -0,0 +1,229 @@
+
+# nature portfolio  
+
+# Peer Review File  
+
+Mi- 2β promotes immune evasion in melanoma by activating EZH2 methylation  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+## REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+Reviewer #1 (Remarks to the Author):Li et al. identify the chromatin remodeling enzyme Mi- 2B as a key melanoma- intrinsic chromatin regulatory factor whose expression is inversely correlated with anti- tumor T cell responses. Using a subcutaneous as well as a genetically engineered cancer model, they demonstrate that loss of Mi- 2B sensitizes melanoma cancer cells to anti- PD- 1 therapy, resulting in an effective anti- tumor immune response. The original manuscript identified a role for Mi- 2B in the regulation of the IFN- responsive genes and that loss of Mi- 2B enhances the expression of, amongst others, the T cell chemoattractants CXCL9 and CXCL10, leading to increased CD8 T cell infiltration into tumors. In the revised manuscript the role of Mi- 2B is explored in more detail by using ATAC- sequencing, ChIP- sequencing, and mass spectrometry experiments. Now the authors show that Mi- 2B interacts with EZH2, promoting EZH2 auto- methylation, which is required to enhance its H3K27 trimethylation activity. Furthermore, Li and colleagues develop an inhibitor of Mi- 2B, Z36- MP5, which induces a response to anti- PD- 1 therapy in otherwise resistant melanomas. Taken together, this is a well- performed study that demonstrates a role for an epigenetic regulator in the regulation of the immunotherapy response against melanoma.  
+
+## Minor comments/suggestions:  
+
+Minor comments/suggestions:1. The authors convincingly demonstrate an interaction between Mi- 2B and EZH2 and that Mi- 2B promotes the activity of EZH2, leading to increased global H3K27me3 levels (Supplemental Figure 4d- e). Furthermore, they demonstrate that depletion of Mi- 2B increases the chromatin accessibility of the Cxcl9 and Cxcl10 promoter. This suggests that chromatin accessibility is increased in shMi- 2B cells due to reduced H3K27me3 in these promoter regions, but this is not experimentally addressed. Can the authors look at H3K27me3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2B
+
+<--- Page Split --->
+
+cells (for example by ChIP- qPCR)?  
+
+2. Fig 2b: The text states: "There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of Mi-2β status (Fig. 2b)." However, Fig 2b demonstrates a significant difference between BRafV600E/Ptennull and BRafV600E/PtennullMi-2βnull mice.  
+
+3. Supplemental Figs 2a and 4c: Quantification of IHC data would be ideal here (as done for Supplemental Fig 2e CD8 T cells).  
+
+4. Supplemental Fig 6j: It does not appear to me that expression of the Ezh2 K to A mutants actually exhibit reduced H3K27me3 as the text states. Can this be quantified?  
+
+5. Fig 4i-j: I found the stated impact in text of the K510A mutation in Ezh2 impacting its own monomethylation hard to follow from the data presented here. While it is clear to me that K735A does not impact Kme1, Kme1 looks prevalent in the K510A mutant experiments. Notably, the Kme1 is high in the negative control lanes too, which I suppose is why the lack of increase in Kme1 is interpreted to mean that K510-methylation is critical here. A clear description of the interpretation of the results would be helpful here.  
+
+6. TIMER analysis demonstrates a negative correlation between Mi-2B mRNA levels and CCL5, CD74 and CD40 mRNA levels in melanoma patients. What about CXCL9 and CXCL10 levels? Why aren't these reported?  
+
+7. In the text, "GSK126" is mis-spelled "SGK-126" although spelled correctly in Supplemental Fig 10.  
+
+## Reviewer #2 (Remarks to the Author):  
+
+Authors provided more details on the manuscript, particularly with more mechanistic insights. However, the characterization of the compound on target effect is still not enough to convince the effect is on target. Especially with the ATP competitive data in figure 5e. with the normally cellular ATP concentration been around 1- 10 mM, the compound will be have very low inhibitory activity, and the inhibitory IC50 to the mi2- b would be 10 uM or even higher, which are not supporting the on- target effect observed with compound treatment at 5 uM or even 25 uM. When the selectivity profiling performed in supplement table 3, mi- 2b (or CHD4) is not included for some reason. it should not be compared with
+
+<--- Page Split --->
+
+the IC50 from other binding assay to show as selectivity.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+Overall, the authors have significantly revised their manuscript and answered most of the reviewer comments. But the clarity with regards to the mechanism behind immune sensitization of Mi- 2B loss is still not sufficient and there are still three major points to be addressed before this paper is acceptable for publication:  
+
+1. While they show how Mi-2B loss activates the ISGs via EZH2 repression, they don't establish that this is the mechanism for the anti-tumor/ CD8 activation response. Akin to the figure S8f, they need to demonstrate that cells lacking EZH2 do not show a combinatorial effect when treated with anti-PD1 and Mi-2B inhibition. 
+2. All their flow data is still in percentage of CD4/CD8s and since the tumor sizes of the combination treatment in particular is much smaller, this is not an accurate representation. They need to show counts of these cells normalized by tumor weight, their counts from IHC alone is not sufficient to answer this critique. 
+3. The results with Z36-MP5 differ from their results using Mi-2B loss (shRNA or genetic loss)- the combinatorial effect with anti-PD1 is lower probably because there is no effect on CD4/T regs. The latter effect is consistently seen with both forms of genetic ablation of Mi-2B. They need to provide an explanation for this difference. For example, by comparing degree of EZH2 inhibition or activation of ISGs between the shRNA/genetic ablation of Mi-2B and the drug.  
+
+## Minor points:  
+
+- They refer to B16 implantation in B6 mice incorrectly as a xenograft model in a few places 
+- Line 335 is confusing as it suggests that Mi-2B loss promotes T cell mediated cytotoxicity in vitro but their data doesn't support it and they reiterate in the rebuttals that Mi-2B loss doesn't alter efficacy of T cell killing.  
+
+"Together, all these data suggest that Z36-MP5 is a potent and effective inhibitor 335 for Mi-2B and stimulates T cell mediated cytotoxicity in vitro, which warranted further in vivo studies."
+
+<--- Page Split --->
+
+- Line 378 is only true with respect to B16 cells. In PM cells, even Z36-MP5 does increase H3K27ac levels, although the degree of increase may be less than that of GSK126. This requires a quantification and rewording on the statement to reflect the actual data. "We confirmed these results, and found that Z36-MP5 repressed the level of H3K27me3, but did not activate H3K27ac"
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+Reviewer #1 (Remarks to the Author):Li et al. identify the chromatin remodeling enzyme Mi- 2B as a key melanoma- intrinsic chromatin regulatory factor whose expression is inversely correlated with anti- tumor T cell responses. Using a subcutaneous as well as a genetically engineered cancer model, they demonstrate that loss of Mi- 2B sensitizes melanoma cancer cells to anti- PD- 1 therapy, resulting in an effective anti- tumor immune response. The original manuscript identified a role for Mi- 2B in the regulation of the IFN- responsive genes and that loss of Mi- 2B enhances the expression of, amongst others, the T cell chemoattractants CXCL9 and CXCL10, leading to increased CD8 T cell infiltration into tumors. In the revised manuscript the role of Mi- 2B is explored in more detail by using ATAC- sequencing, ChIP- sequencing, and mass spectrometry experiments. Now the authors show that Mi- 2B interacts with EZH2, promoting EZH2 auto- methylation, which is required to enhance its H3K27 trimethylation activity. Furthermore, Li and colleagues develop an inhibitor of Mi- 2B, Z36- MP5, which induces a response to anti- PD- 1 therapy in otherwise resistant melanomas. Taken together, this is a well- performed study that demonstrates a role for an epigenetic regulator in the regulation of the immunotherapy response against melanoma.  
+
+Response: We thank the reviewer for recognizing a well- performed study of our work.  
+
+Minor comments/suggestions:  
+
+1. The authors convincingly demonstrate an interaction between Mi-2B and EZH2 and that Mi-2B promotes the activity of EZH2, leading to increased global H3K27me3 levels (Supplemental Figure 4d-e). Furthermore, they demonstrate that depletion of Mi-2B increases the chromatin accessibility of the Cxcl9 and Cxcl10 promoter. This suggests that chromatin accessibility is increased in shMi-2B cells due to reduced H3K27me3 in these promoter regions, but this is not experimentally addressed. Can the authors look at H3K27me3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi-2B cells (for example by ChIP-qPCR)?  
+
+Response: We thank the reviewer for the constructive suggestion. We fully agree with the reviewer that it is necessary to identify the H3K27m3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2 \(\beta\) cells. To this end, we performed a ChIP-qPCR to identify the H3K27m3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2 \(\beta\) cells. We found the H3K27me3 levels at the Cxcl9 and Cxcl10 promoter region was indeed repressed in cells with Mi- 2 \(\beta\) silencing (please see Fig 4e in the resubmitted manuscript). This result further confirmed that the chromatin accessibility of the Cxcl9 and Cxcl10 promoter was activated in cells with Mi- 2 \(\beta\) silencing.  
+
+2. Fig 2b: The text states: "There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of Mi-2 \(\beta\) status (Fig. 2b)."
+
+<--- Page Split --->
+
+However, Fig 2b demonstrates a significant difference between BRafV600E/Ptennull and BRafV600E/PtennullMi- 2βnull mice.  
+
+Response: We thank the reviewer for pointing out this typo error and apologize for this. The correct text should be: There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of anti- PD- 1 treatment (Fig. 2b). This typo error has been corrected.  
+
+3. Supplemental Figs 2a and 4c: Quantification of IHC data would be ideal here (as done for Supplemental Fig 2e CD8 T cells).  
+
+Response: We thank the reviewer for the suggestion. We have added the quantification of IHC data. (please see Fig S4c in the resubmitted manuscript)  
+
+4. Supplemental Fig 6j: It does not appear to me that expression of the Ezh2 K to A mutants actually exhibit reduced H3K27me3 as the text states. Can this be quantified? Response: We thank the reviewer for pointing out this error. It's a label error. We apologize for this. It has been fixed in the resubmitted manuscript.  
+
+5. Fig 4i-j: I found the stated impact in text of the K510A mutation in Ezh2 impacting its own monomethylation hard to follow from the data presented here. While it is clear to me that K735A does not impact Kme1, Kme1 looks prevalent in the K510A mutant experiments. Notably, the Kme1 is high in the negative control lanes too, which I suppose is why the lack of increase in Kme1 is interpreted to mean that K510-methylation is critical here. A clear description of the interpretation of the results would be helpful here.  
+
+Response: We thank the reviewer for the comments. We have repeated the experiments presented in Fig 4i-j. Specifically, in the repeated experiments, we used the same exposure time in our WB experiment and added a control group. As demonstrated in Fig. 4i-j, we found that the K510 EZH2 methylation was Mi- 2β dependent. This result is consistent with our original results. In our previous result, the WB exposure time was different, which cannot compare the expression in different panels of WB experiment.  
+
+6. TIMER analysis demonstrates a negative correlation between Mi-2B mRNA levels and CCL5, CD74 and CD40 mRNA levels in melanoma patients. What about CXCL9 and CXCL10 levels? Why aren't these reported?  
+
+Response: We thank the reviewer for the helpful suggestion. TIMER analysis indicated that there was a negative correlation tendency between Mi- 2β and Cxcl9 and Cxcl10 at the mRNA level in melanomas in TCGA. However, the correlation did not reach the statistical significance. Please see the analytic results, below. These results were mentioned in the revised manuscript (lines 210- 211).
+
+<--- Page Split --->
+![PLACEHOLDER_7_0]
+  
+
+7. In the text, "GSK126" is mis-spelled "SGK-126" although spelled correctly in Supplemental Fig 10. Response: We thank the reviewer for pointing out this typo. It has been fixed.  
+
+Reviewer #2 (Remarks to the Author):  
+
+Authors provided more details on the manuscript, particularly with more mechanistic insights. However, the characterization of the compound on target effect is still not enough to convince the effect is on target. Especially with the ATP competitive data in figure 5e. with the normally cellular ATP concentration been around 1- 10 mM, the compound will be have very low inhibitory activity, and the inhibitory IC50 to the mi2- b would be 10 uM or even higher, which are not supporting the on- target effect observed with compound treatment at 5 uM or even 25 uM. When the selectivity profiling performed in supplement table 3, mi- 2b (or CHD4) is not included for some reason. it should not be compared with the IC50 from other binding assay to show as selectivity.  
+
+Response: We thank the reviewer for raising this concern. As the ATP- competitive biochemical assay may fail to fully replicate the physiological complexity of the full- length kinase protein in the presence of the cellular milieu and regulatory circuits that modulate kinase function and cellular microenvironment, it may not be sufficient to estimate the cellular on- target inhibitory activity of a compound merely basing on the readout of biochemical assay. There are multiple examples of compounds which showed unexpected higher on- target cellular activity than the readouts from competitive biochemical assay. For instance, structure- guided drug designs for FLT3- WT or FLT3- ITD mutant kinase inhibitors have reported significantly higher cellular potency than that obtained for kinase biochemical assay with ATP concentrations lower than 10 μM (Tomáš Gucký et al., Discovery of N2-(4- amino- cyclohexyl)-9- cyclopentyl- N6-(4- morpholin- 4- ylmethyl- phenyl)- 9H- purine- 2,6- diamine as a potent
+
+<--- Page Split --->
+
+FLT3 kinase inhibitor for acute myeloid leukemia with FLT3 mutations. J Med Chem, 61, 3855- 3869 (2018); Lexian Tong et al., Identification of 2- aminopyrimidine derivatives as FLT3 kinase inhibitors with high selectivity over c- KIT. J Med Chem, 65, 3229- 3248 (2022)). In these cases, the cellular form of the FLT3- WT or FLT3- ITD mutant was preferentially bound. Another example is paclitaxel, a microtubule- stabilizing agent that is widely used in cancer chemotherapy. Biochemical competitive binding assay suggests that paclitaxel bind with GMPcPP- stabilized microtubules with binding affinity at micromole concentrations, whereas paclitaxel shows drastically higher cytotoxicity against PC3 prostate cancer cells with IC50 value 1.41 nM (Shubhada Sharma, et al., Dissecting paclitaxel- microtubule association: quantitative assessment of the 2'- OH group. Biochemistry, 52, 2328- 2336 (2013)). Paclitaxel sensitizes BG- 1 ovarian cancer cells to radiation at a dose that was not cytotoxic and did not cause cell cycle perturbations (Albert Steren, et al., Taxol as a radiation sensitizer: a flow cytometric study. Gynecol Oncol, 50, 89- 93 (1993)).  
+
+Like the Mi- 2β silencing by shMi- 2β (Supplementary Fig. 3C), we observed substantial increases in expressions of Mi- 2β target genes, Cxcl9 and Cxcl10, to similar levels after treatment of B16F10 cells with Z36- MP5 at concentrations \(5\mu \mathrm{M}\) to \(100\mu \mathrm{M}\) (Fig. 5F). Meanwhile, the increases in expression of both Cxcl9 and Cxcl10 were observed with an apparently dose- dependent fashion (Fig. 5F), indicating that \(5\mu \mathrm{M}\) to \(100\mu \mathrm{M}\) Z36- MP5 treatment would be sufficient for inhibiting Mi- 2β in cells. Moreover, we observed a maintenance of blood concentration of \(4\mu \mathrm{M}\) to \(10\mu \mathrm{M}\) Z36- MP5 in rat for around 1 hour after the intraperitoneal injection at a dose of Z36- MP5 as low as \(1\mathrm{mg}\) per kg body weight (Supplementary Fig. 8C), suggesting the sufficient blood concentrations of Z36- MP5 in mice after intraperitoneal injection at a dose as high as \(30\mathrm{mg}\) per kg body weight (estimated highest blood concentration in mice: around \(0.97\mathrm{mM}\) ) for Mi- 2β inhibition in vivo. Indeed, consistent with the cell assay data, treatment of B16F10- graft mice model with Z36- MP5 alone or in combination with anti- PD- 1 perfectly recreated the in vivo data by shMi- 2β (Figs. 1 and 6), suggesting the Mi- 2β on- target inhibition in vivo.  
+
+We thank the review for the comment and understand the concern of the reviewer about the selectivity profiling of compound. To profile the selectivity of Z36- MP5 on ATPases in native cell lysates, we did our best to contact with all the commercially available services and chose the ActivX Biosciences Inc, as it provided the inhibitory profiling assay of compounds against the largest panel of 233 ATPases when preparing this manuscript. Unfortunately, the ATPases panel by ActivX Biosciences Inc does not include Mi- 2β protein. In this ATPase profiling assay, we showed that, against the panel of 233 diverse ATPases, Z36- MP5 possessed less than \(35\%\) inhibition at \(1\mu \mathrm{M}\) (Supplementary Table 3), a concentration that completely inhibits Mi- 2β ATPase. These results suggest that Z36- MP5 has a high Mi- 2β ATPase selectivity and specificity. Additionally, we observed the extremely high kinase selectivity of Z36- MP5 by KINOMEscan (Supplementary Fig. 7e), in which the inhibitory activity of Z36- MP5 against a panel of 468 diverse kinases were examined using an in vitro ATP- competitive binding assay.  
+
+To further identify the targeted specificity of Z36- MP5 in Mi- 2β, we generated a Mi
+
+<--- Page Split --->
+
+\(2\beta\) construct mutated at H727A and then tested the effect of Z36- MP5 by the FRET- based nucleosome repositioning assays. We found that Mi- 2β H727A mutant was no longer sensitive to Z36- MP5 (Supplementary Fig. 7f). This result further confirmed the specificity of Z36- MP5 to Mi- 2β protein.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+Overall, the authors have significantly revised their manuscript and answered most of the reviewer comments. But the clarity with regards to the mechanism behind immune sensitization of Mi- 2B loss is still not sufficient and there are still three major points to be addressed before this paper is acceptable for publication:  
+
+1. While they show how Mi-2B loss activates the ISGs via EZH2 repression, they don't establish that this is the mechanism for the anti-tumor/ CD8 activation response. Akin to the figure S8f, they need to demonstrate that cells lacking EZH2 do not show a combinatorial effect when treated with anti-PD1 and Mi-2B inhibition.  
+
+Response: We thank the reviewer for the suggestion. We fully agree with the reviewer and performed the experimental therapeutics as suggested. Specifically, we implanted B16F10 cells into C57BL6 mice, treated the mice with MS177 to get the EZH2-targeted inhibition, and then gave anti-PD-1 in combination with Z36-MP5 or vehicles treatment. No treatment effects were detected in mice treated with MS177 to inhibit EZH2. This result further indicates the treatment effect of Z36-MP5 is EZH2 dependent (Fig. S8g in the revised manuscript).  
+
+2. All their flow data is still in percentage of CD4/CD8s and since the tumor sizes of the combination treatment in particular is much smaller, this is not an accurate representation. They need to show counts of these cells normalized by tumor weight, their counts from IHC alone is not sufficient to answer this critique.  
+
+Response: We thank the review for the comment and understand the concern of the reviewer. We have quantified the CD4/CD8 cells normalized by tumor weight as suggested. The cell numbers based on the result of flow cytometry analysis has been presented in Fig. 1e and Fig. 1c in the resubmitted manuscript).  
+
+3. The results with Z36-MP5 differ from their results using Mi-2B loss (shRNA or genetic loss)- the combinatorial effect with anti-PD1 is lower probably because there is no effect on CD4/T regs. The latter effect is consistently seen with both forms of genetic ablation of Mi-2B. They need to provide an explanation for this difference. For example, by comparing degree of EZH2 inhibition or activation of ISGs between the shRNA/genetic ablation of Mi-2B and the drug.  
+
+Response: We thank the reviewer for the great suggestion and has performed the experiment as suggested. Specifically, B16F10 cells with stable Mi-2β silencing were
+
+<--- Page Split --->
+
+grafted into C57BL6 mice and then treated with Z36- MP5. The expression of H3K27me3, Cxcl9 and Cxcl10 were detected in mouse tumors. We found that the expression of H3K27me3, Cxcl9 and Cxcl10 were significantly repressed after Mi- 2β silencing or Z36- MP5 treatment. However, the Mi- 2β silencing- induced repression was more sufficient than Z36- MP5 treatment- induced repression (Please see the figure below the passage). Z36- MP5 is our first generation of Mi- 2β- targeted inhibitor. We will further improve its specificity and sensitivity by the structure modification.  
+
+![PLACEHOLDER_10_0]
+  
+
+## Minor points:  
+
+- They refer to B16 implantation in B6 mice incorrectly as a xenograft model in a few places  
+
+Response: We thank the reviewer for pointing out these errors. We corrected the errors.  
+
+- Line 335 is confusing as it suggests that Mi-2B loss promotes T cell mediated cytotoxicity in vitro but their data doesn't support it and they reiterate in the rebuttals that Mi-2B loss doesn't alter efficacy of T cell killing.  
+
+Response: We thank the reviewer for pointing out this error. We apologize for this. We have corrected the description.  
+
+- Line 378 is only true with respect to B16 cells. In PM cells, even Z36-MP5 does increase H3K27ac levels, although the degree of increase may be less than that of GSK126. This requires a quantification and rewording on the statement to reflect the actual data.  
+
+"We confirmed these results, and found that Z36-MP5 repressed the level of H3K27me3, but did not activate H3K27ac"  
+
+Response: We thank the reviewer for the suggestion. We now correct it to: "We found that Z36- MP5 repressed the level of H3K27me3 in B16 and PM cells. However, compared to GSK126 treatment, the H3K27ac level only slightly increased in PM cells after treatment with Z36- MP5."
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+The authors have fully and satisfactorily responded to all of my comments. Most importantly, they performed additional experiments to show enrichment of H3K27me3 by ChIP- qPCR at the Cxcl9 and Cxcl10 loci and performing TIMER analysis based on Cxcl9 and Cxcl10 expression levels from TCGA melanomas. They also corrected all editorial mistakes. This is a well- done study.  
+
+## Reviewer #2 (Remarks to the Author):  
+
+Authors addressed all my concerns.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+I appreciate the efforts that authors have made to answer my comments and the ones of other reviewers. I have 2 more minor comments that should be addressed before publication  
+
+Comment to response 2: The y axis title of Figure 1e can be improved as it's slightly confusing (typically should read number of cells/mg or g tumor) and neither the legends nor the methods mentioned how the cell numbers of CD4/CD8 were normalized (to weight of tumor and volume of resuspension etc....).  
+
+This information should be added before before publication.  
+
+Comment on response 3: This comparison showing that the Z36- MP5 is not entirely as effective as complete genetic ablation of Mi- 2B need to feature in some part of the manuscript as well, even if just as a sentence at the end of the results section or discussion (without the data as such if there is lack of figure space). Because as it stands the manuscript heavily emphasizes that Z36- MP5 is an "effective" inhibitor of Mi- 2B without acknowledging that there is room for improvement to fully capture the immune advantages of complete Mi- 2B silencing.
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+The authors have fully and satisfactorily responded to all of my comments. Most importantly, they performed additional experiments to show enrichment of H3K27me3 by ChIP- qPCR at the Cxcl9 and CxCl10 loci and performing TIMER analysis based on Cxcl9 and Cxcl10 expression levels from TCGA melanomas. They also corrected all editorial mistakes. This is a well- done study.  
+
+Response: We sincerely appreciate your efforts of reviewing and the recognition of our improvement of our manuscript.  
+
+Reviewer #2 (Remarks to the Author):  
+
+Authors addressed all my concerns.  
+
+Response: We sincerely appreciate your efforts of reviewing and the recognition of our improvement of our manuscript.  
+
+Reviewer #3 (Remarks to the Author):  
+
+I appreciate the efforts that authors have made to answer my comments and the ones of other reviewers. I have 2 more minor comments that should be addressed before publication  
+
+Response: We sincerely appreciate your carefully review and your very helpful suggestions.  
+
+Comment to response 2: The y axis title of Figure 1e can be improved as it's slightly confusing (typically should read number of cells/mg or g tumor) and neither the legends nor the methods mentioned how the cell numbers of CD4/CD8 were normalized (to weight of tumor and volume of resuspension etc....).  
+
+This information should be added before before publication.  
+
+Response: We thank the reviewer for the constructive suggestion. The y axis title of Figure 1e has been replaced as suggested. Specifically, the y axis title of Figure 1e has been replaced: cell number / g tumor. We have also added the necessary information that how the cell numbers of CD4/CD8 were normalized (please see line 642- 644 in the resubmitted manuscript).  
+
+Comment on response 3: This comparison showing that the Z36- MP5 is not entirely as effective as complete genetic ablation of Mi- 2B need to feature in some part of the manuscript as well, even if just as a sentence at the end of the results section or discussion (without the data as such if there is lack of figure space). Because as it stands the manuscript heavily emphasizes that Z36- MP5 is an "effective" inhibitor of Mi- 2B without acknowledging that there is room for improvement to fully capture the immune advantages of complete Mi- 2B silencing.
+
+<--- Page Split --->
+
+Response: We thank the reviewer for the suggestion. According to suggestion, we have added the information in discussion. (please see line 445- 447 in the resubmitted manuscript).
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1f6c8e672b6887aeacdaec3615e7a8c9aed8a2fd462da50a662918cb10006724/supplementary_0_Peer Review File__1f6c8e672b6887aeacdaec3615e7a8c9aed8a2fd462da50a662918cb10006724_det.mmd

@@ -0,0 +1,308 @@
+<|ref|>title<|/ref|><|det|>[[61, 40, 506, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>title<|/ref|><|det|>[[66, 110, 361, 139]]<|/det|>
+# Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[85, 154, 914, 210]]<|/det|>
+Mi- 2β promotes immune evasion in melanoma by activating EZH2 methylation  
+
+<|ref|>image<|/ref|><|det|>[[57, 732, 240, 781]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 912, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 84, 868, 155]]<|/det|>
+Editorial Note: This manuscript has been previously reviewed at another journal that is not operating a transparent peer review scheme. This document only contains reviewer comments and rebuttal letters for versions considered at Nature Communications.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 208, 315, 225]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 260, 437, 277]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 285, 880, 670]]<|/det|>
+Reviewer #1 (Remarks to the Author):Li et al. identify the chromatin remodeling enzyme Mi- 2B as a key melanoma- intrinsic chromatin regulatory factor whose expression is inversely correlated with anti- tumor T cell responses. Using a subcutaneous as well as a genetically engineered cancer model, they demonstrate that loss of Mi- 2B sensitizes melanoma cancer cells to anti- PD- 1 therapy, resulting in an effective anti- tumor immune response. The original manuscript identified a role for Mi- 2B in the regulation of the IFN- responsive genes and that loss of Mi- 2B enhances the expression of, amongst others, the T cell chemoattractants CXCL9 and CXCL10, leading to increased CD8 T cell infiltration into tumors. In the revised manuscript the role of Mi- 2B is explored in more detail by using ATAC- sequencing, ChIP- sequencing, and mass spectrometry experiments. Now the authors show that Mi- 2B interacts with EZH2, promoting EZH2 auto- methylation, which is required to enhance its H3K27 trimethylation activity. Furthermore, Li and colleagues develop an inhibitor of Mi- 2B, Z36- MP5, which induces a response to anti- PD- 1 therapy in otherwise resistant melanomas. Taken together, this is a well- performed study that demonstrates a role for an epigenetic regulator in the regulation of the immunotherapy response against melanoma.  
+
+<|ref|>sub_title<|/ref|><|det|>[[119, 705, 371, 721]]<|/det|>
+## Minor comments/suggestions:  
+
+<|ref|>text<|/ref|><|det|>[[116, 729, 860, 905]]<|/det|>
+Minor comments/suggestions:1. The authors convincingly demonstrate an interaction between Mi- 2B and EZH2 and that Mi- 2B promotes the activity of EZH2, leading to increased global H3K27me3 levels (Supplemental Figure 4d- e). Furthermore, they demonstrate that depletion of Mi- 2B increases the chromatin accessibility of the Cxcl9 and Cxcl10 promoter. This suggests that chromatin accessibility is increased in shMi- 2B cells due to reduced H3K27me3 in these promoter regions, but this is not experimentally addressed. Can the authors look at H3K27me3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2B
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 85, 395, 101]]<|/det|>
+cells (for example by ChIP- qPCR)?  
+
+<|ref|>text<|/ref|><|det|>[[116, 110, 880, 205]]<|/det|>
+2. Fig 2b: The text states: "There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of Mi-2β status (Fig. 2b)." However, Fig 2b demonstrates a significant difference between BRafV600E/Ptennull and BRafV600E/PtennullMi-2βnull mice.  
+
+<|ref|>text<|/ref|><|det|>[[116, 214, 870, 258]]<|/det|>
+3. Supplemental Figs 2a and 4c: Quantification of IHC data would be ideal here (as done for Supplemental Fig 2e CD8 T cells).  
+
+<|ref|>text<|/ref|><|det|>[[116, 266, 870, 310]]<|/det|>
+4. Supplemental Fig 6j: It does not appear to me that expression of the Ezh2 K to A mutants actually exhibit reduced H3K27me3 as the text states. Can this be quantified?  
+
+<|ref|>text<|/ref|><|det|>[[116, 318, 872, 468]]<|/det|>
+5. Fig 4i-j: I found the stated impact in text of the K510A mutation in Ezh2 impacting its own monomethylation hard to follow from the data presented here. While it is clear to me that K735A does not impact Kme1, Kme1 looks prevalent in the K510A mutant experiments. Notably, the Kme1 is high in the negative control lanes too, which I suppose is why the lack of increase in Kme1 is interpreted to mean that K510-methylation is critical here. A clear description of the interpretation of the results would be helpful here.  
+
+<|ref|>text<|/ref|><|det|>[[117, 476, 856, 546]]<|/det|>
+6. TIMER analysis demonstrates a negative correlation between Mi-2B mRNA levels and CCL5, CD74 and CD40 mRNA levels in melanoma patients. What about CXCL9 and CXCL10 levels? Why aren't these reported?  
+
+<|ref|>text<|/ref|><|det|>[[117, 580, 876, 624]]<|/det|>
+7. In the text, "GSK126" is mis-spelled "SGK-126" although spelled correctly in Supplemental Fig 10.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 685, 438, 702]]<|/det|>
+## Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[116, 711, 870, 911]]<|/det|>
+Authors provided more details on the manuscript, particularly with more mechanistic insights. However, the characterization of the compound on target effect is still not enough to convince the effect is on target. Especially with the ATP competitive data in figure 5e. with the normally cellular ATP concentration been around 1- 10 mM, the compound will be have very low inhibitory activity, and the inhibitory IC50 to the mi2- b would be 10 uM or even higher, which are not supporting the on- target effect observed with compound treatment at 5 uM or even 25 uM. When the selectivity profiling performed in supplement table 3, mi- 2b (or CHD4) is not included for some reason. it should not be compared with
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[118, 85, 581, 101]]<|/det|>
+the IC50 from other binding assay to show as selectivity.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 163, 438, 180]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 188, 856, 285]]<|/det|>
+Overall, the authors have significantly revised their manuscript and answered most of the reviewer comments. But the clarity with regards to the mechanism behind immune sensitization of Mi- 2B loss is still not sufficient and there are still three major points to be addressed before this paper is acceptable for publication:  
+
+<|ref|>text<|/ref|><|det|>[[115, 317, 880, 680]]<|/det|>
+1. While they show how Mi-2B loss activates the ISGs via EZH2 repression, they don't establish that this is the mechanism for the anti-tumor/ CD8 activation response. Akin to the figure S8f, they need to demonstrate that cells lacking EZH2 do not show a combinatorial effect when treated with anti-PD1 and Mi-2B inhibition. 
+2. All their flow data is still in percentage of CD4/CD8s and since the tumor sizes of the combination treatment in particular is much smaller, this is not an accurate representation. They need to show counts of these cells normalized by tumor weight, their counts from IHC alone is not sufficient to answer this critique. 
+3. The results with Z36-MP5 differ from their results using Mi-2B loss (shRNA or genetic loss)- the combinatorial effect with anti-PD1 is lower probably because there is no effect on CD4/T regs. The latter effect is consistently seen with both forms of genetic ablation of Mi-2B. They need to provide an explanation for this difference. For example, by comparing degree of EZH2 inhibition or activation of ISGs between the shRNA/genetic ablation of Mi-2B and the drug.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 712, 231, 728]]<|/det|>
+## Minor points:  
+
+<|ref|>text<|/ref|><|det|>[[117, 736, 872, 832]]<|/det|>
+- They refer to B16 implantation in B6 mice incorrectly as a xenograft model in a few places 
+- Line 335 is confusing as it suggests that Mi-2B loss promotes T cell mediated cytotoxicity in vitro but their data doesn't support it and they reiterate in the rebuttals that Mi-2B loss doesn't alter efficacy of T cell killing.  
+
+<|ref|>text<|/ref|><|det|>[[117, 840, 857, 909]]<|/det|>
+"Together, all these data suggest that Z36-MP5 is a potent and effective inhibitor 335 for Mi-2B and stimulates T cell mediated cytotoxicity in vitro, which warranted further in vivo studies."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 83, 850, 207]]<|/det|>
+- Line 378 is only true with respect to B16 cells. In PM cells, even Z36-MP5 does increase H3K27ac levels, although the degree of increase may be less than that of GSK126. This requires a quantification and rewording on the statement to reflect the actual data. "We confirmed these results, and found that Z36-MP5 repressed the level of H3K27me3, but did not activate H3K27ac"
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[149, 85, 388, 102]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[149, 122, 480, 140]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 159, 850, 455]]<|/det|>
+Reviewer #1 (Remarks to the Author):Li et al. identify the chromatin remodeling enzyme Mi- 2B as a key melanoma- intrinsic chromatin regulatory factor whose expression is inversely correlated with anti- tumor T cell responses. Using a subcutaneous as well as a genetically engineered cancer model, they demonstrate that loss of Mi- 2B sensitizes melanoma cancer cells to anti- PD- 1 therapy, resulting in an effective anti- tumor immune response. The original manuscript identified a role for Mi- 2B in the regulation of the IFN- responsive genes and that loss of Mi- 2B enhances the expression of, amongst others, the T cell chemoattractants CXCL9 and CXCL10, leading to increased CD8 T cell infiltration into tumors. In the revised manuscript the role of Mi- 2B is explored in more detail by using ATAC- sequencing, ChIP- sequencing, and mass spectrometry experiments. Now the authors show that Mi- 2B interacts with EZH2, promoting EZH2 auto- methylation, which is required to enhance its H3K27 trimethylation activity. Furthermore, Li and colleagues develop an inhibitor of Mi- 2B, Z36- MP5, which induces a response to anti- PD- 1 therapy in otherwise resistant melanomas. Taken together, this is a well- performed study that demonstrates a role for an epigenetic regulator in the regulation of the immunotherapy response against melanoma.  
+
+<|ref|>text<|/ref|><|det|>[[148, 457, 847, 475]]<|/det|>
+Response: We thank the reviewer for recognizing a well- performed study of our work.  
+
+<|ref|>text<|/ref|><|det|>[[150, 501, 393, 517]]<|/det|>
+Minor comments/suggestions:  
+
+<|ref|>text<|/ref|><|det|>[[148, 520, 850, 666]]<|/det|>
+1. The authors convincingly demonstrate an interaction between Mi-2B and EZH2 and that Mi-2B promotes the activity of EZH2, leading to increased global H3K27me3 levels (Supplemental Figure 4d-e). Furthermore, they demonstrate that depletion of Mi-2B increases the chromatin accessibility of the Cxcl9 and Cxcl10 promoter. This suggests that chromatin accessibility is increased in shMi-2B cells due to reduced H3K27me3 in these promoter regions, but this is not experimentally addressed. Can the authors look at H3K27me3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi-2B cells (for example by ChIP-qPCR)?  
+
+<|ref|>text<|/ref|><|det|>[[148, 667, 850, 833]]<|/det|>
+Response: We thank the reviewer for the constructive suggestion. We fully agree with the reviewer that it is necessary to identify the H3K27m3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2 \(\beta\) cells. To this end, we performed a ChIP-qPCR to identify the H3K27m3 levels specifically in the Cxcl9 and Cxcl10 promoter region in WT and shMi- 2 \(\beta\) cells. We found the H3K27me3 levels at the Cxcl9 and Cxcl10 promoter region was indeed repressed in cells with Mi- 2 \(\beta\) silencing (please see Fig 4e in the resubmitted manuscript). This result further confirmed that the chromatin accessibility of the Cxcl9 and Cxcl10 promoter was activated in cells with Mi- 2 \(\beta\) silencing.  
+
+<|ref|>text<|/ref|><|det|>[[148, 870, 849, 907]]<|/det|>
+2. Fig 2b: The text states: "There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of Mi-2 \(\beta\) status (Fig. 2b)."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 85, 848, 120]]<|/det|>
+However, Fig 2b demonstrates a significant difference between BRafV600E/Ptennull and BRafV600E/PtennullMi- 2βnull mice.  
+
+<|ref|>text<|/ref|><|det|>[[148, 123, 849, 195]]<|/det|>
+Response: We thank the reviewer for pointing out this typo error and apologize for this. The correct text should be: There was no significant difference in mouse survival observed on a BRafV600E/Ptennull background irrespective of anti- PD- 1 treatment (Fig. 2b). This typo error has been corrected.  
+
+<|ref|>text<|/ref|><|det|>[[148, 215, 848, 250]]<|/det|>
+3. Supplemental Figs 2a and 4c: Quantification of IHC data would be ideal here (as done for Supplemental Fig 2e CD8 T cells).  
+
+<|ref|>text<|/ref|><|det|>[[148, 252, 848, 288]]<|/det|>
+Response: We thank the reviewer for the suggestion. We have added the quantification of IHC data. (please see Fig S4c in the resubmitted manuscript)  
+
+<|ref|>text<|/ref|><|det|>[[148, 325, 849, 399]]<|/det|>
+4. Supplemental Fig 6j: It does not appear to me that expression of the Ezh2 K to A mutants actually exhibit reduced H3K27me3 as the text states. Can this be quantified? Response: We thank the reviewer for pointing out this error. It's a label error. We apologize for this. It has been fixed in the resubmitted manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[148, 418, 850, 547]]<|/det|>
+5. Fig 4i-j: I found the stated impact in text of the K510A mutation in Ezh2 impacting its own monomethylation hard to follow from the data presented here. While it is clear to me that K735A does not impact Kme1, Kme1 looks prevalent in the K510A mutant experiments. Notably, the Kme1 is high in the negative control lanes too, which I suppose is why the lack of increase in Kme1 is interpreted to mean that K510-methylation is critical here. A clear description of the interpretation of the results would be helpful here.  
+
+<|ref|>text<|/ref|><|det|>[[148, 549, 850, 658]]<|/det|>
+Response: We thank the reviewer for the comments. We have repeated the experiments presented in Fig 4i-j. Specifically, in the repeated experiments, we used the same exposure time in our WB experiment and added a control group. As demonstrated in Fig. 4i-j, we found that the K510 EZH2 methylation was Mi- 2β dependent. This result is consistent with our original results. In our previous result, the WB exposure time was different, which cannot compare the expression in different panels of WB experiment.  
+
+<|ref|>text<|/ref|><|det|>[[148, 696, 850, 750]]<|/det|>
+6. TIMER analysis demonstrates a negative correlation between Mi-2B mRNA levels and CCL5, CD74 and CD40 mRNA levels in melanoma patients. What about CXCL9 and CXCL10 levels? Why aren't these reported?  
+
+<|ref|>text<|/ref|><|det|>[[148, 752, 850, 842]]<|/det|>
+Response: We thank the reviewer for the helpful suggestion. TIMER analysis indicated that there was a negative correlation tendency between Mi- 2β and Cxcl9 and Cxcl10 at the mRNA level in melanomas in TCGA. However, the correlation did not reach the statistical significance. Please see the analytic results, below. These results were mentioned in the revised manuscript (lines 210- 211).
+
+<--- Page Split --->
+<|ref|>image<|/ref|><|det|>[[151, 95, 765, 295]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[147, 325, 850, 381]]<|/det|>
+7. In the text, "GSK126" is mis-spelled "SGK-126" although spelled correctly in Supplemental Fig 10. Response: We thank the reviewer for pointing out this typo. It has been fixed.  
+
+<|ref|>text<|/ref|><|det|>[[149, 437, 445, 452]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[147, 474, 851, 658]]<|/det|>
+Authors provided more details on the manuscript, particularly with more mechanistic insights. However, the characterization of the compound on target effect is still not enough to convince the effect is on target. Especially with the ATP competitive data in figure 5e. with the normally cellular ATP concentration been around 1- 10 mM, the compound will be have very low inhibitory activity, and the inhibitory IC50 to the mi2- b would be 10 uM or even higher, which are not supporting the on- target effect observed with compound treatment at 5 uM or even 25 uM. When the selectivity profiling performed in supplement table 3, mi- 2b (or CHD4) is not included for some reason. it should not be compared with the IC50 from other binding assay to show as selectivity.  
+
+<|ref|>text<|/ref|><|det|>[[147, 678, 851, 899]]<|/det|>
+Response: We thank the reviewer for raising this concern. As the ATP- competitive biochemical assay may fail to fully replicate the physiological complexity of the full- length kinase protein in the presence of the cellular milieu and regulatory circuits that modulate kinase function and cellular microenvironment, it may not be sufficient to estimate the cellular on- target inhibitory activity of a compound merely basing on the readout of biochemical assay. There are multiple examples of compounds which showed unexpected higher on- target cellular activity than the readouts from competitive biochemical assay. For instance, structure- guided drug designs for FLT3- WT or FLT3- ITD mutant kinase inhibitors have reported significantly higher cellular potency than that obtained for kinase biochemical assay with ATP concentrations lower than 10 μM (Tomáš Gucký et al., Discovery of N2-(4- amino- cyclohexyl)-9- cyclopentyl- N6-(4- morpholin- 4- ylmethyl- phenyl)- 9H- purine- 2,6- diamine as a potent
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 84, 850, 345]]<|/det|>
+FLT3 kinase inhibitor for acute myeloid leukemia with FLT3 mutations. J Med Chem, 61, 3855- 3869 (2018); Lexian Tong et al., Identification of 2- aminopyrimidine derivatives as FLT3 kinase inhibitors with high selectivity over c- KIT. J Med Chem, 65, 3229- 3248 (2022)). In these cases, the cellular form of the FLT3- WT or FLT3- ITD mutant was preferentially bound. Another example is paclitaxel, a microtubule- stabilizing agent that is widely used in cancer chemotherapy. Biochemical competitive binding assay suggests that paclitaxel bind with GMPcPP- stabilized microtubules with binding affinity at micromole concentrations, whereas paclitaxel shows drastically higher cytotoxicity against PC3 prostate cancer cells with IC50 value 1.41 nM (Shubhada Sharma, et al., Dissecting paclitaxel- microtubule association: quantitative assessment of the 2'- OH group. Biochemistry, 52, 2328- 2336 (2013)). Paclitaxel sensitizes BG- 1 ovarian cancer cells to radiation at a dose that was not cytotoxic and did not cause cell cycle perturbations (Albert Steren, et al., Taxol as a radiation sensitizer: a flow cytometric study. Gynecol Oncol, 50, 89- 93 (1993)).  
+
+<|ref|>text<|/ref|><|det|>[[147, 345, 850, 621]]<|/det|>
+Like the Mi- 2β silencing by shMi- 2β (Supplementary Fig. 3C), we observed substantial increases in expressions of Mi- 2β target genes, Cxcl9 and Cxcl10, to similar levels after treatment of B16F10 cells with Z36- MP5 at concentrations \(5\mu \mathrm{M}\) to \(100\mu \mathrm{M}\) (Fig. 5F). Meanwhile, the increases in expression of both Cxcl9 and Cxcl10 were observed with an apparently dose- dependent fashion (Fig. 5F), indicating that \(5\mu \mathrm{M}\) to \(100\mu \mathrm{M}\) Z36- MP5 treatment would be sufficient for inhibiting Mi- 2β in cells. Moreover, we observed a maintenance of blood concentration of \(4\mu \mathrm{M}\) to \(10\mu \mathrm{M}\) Z36- MP5 in rat for around 1 hour after the intraperitoneal injection at a dose of Z36- MP5 as low as \(1\mathrm{mg}\) per kg body weight (Supplementary Fig. 8C), suggesting the sufficient blood concentrations of Z36- MP5 in mice after intraperitoneal injection at a dose as high as \(30\mathrm{mg}\) per kg body weight (estimated highest blood concentration in mice: around \(0.97\mathrm{mM}\) ) for Mi- 2β inhibition in vivo. Indeed, consistent with the cell assay data, treatment of B16F10- graft mice model with Z36- MP5 alone or in combination with anti- PD- 1 perfectly recreated the in vivo data by shMi- 2β (Figs. 1 and 6), suggesting the Mi- 2β on- target inhibition in vivo.  
+
+<|ref|>text<|/ref|><|det|>[[148, 622, 850, 880]]<|/det|>
+We thank the review for the comment and understand the concern of the reviewer about the selectivity profiling of compound. To profile the selectivity of Z36- MP5 on ATPases in native cell lysates, we did our best to contact with all the commercially available services and chose the ActivX Biosciences Inc, as it provided the inhibitory profiling assay of compounds against the largest panel of 233 ATPases when preparing this manuscript. Unfortunately, the ATPases panel by ActivX Biosciences Inc does not include Mi- 2β protein. In this ATPase profiling assay, we showed that, against the panel of 233 diverse ATPases, Z36- MP5 possessed less than \(35\%\) inhibition at \(1\mu \mathrm{M}\) (Supplementary Table 3), a concentration that completely inhibits Mi- 2β ATPase. These results suggest that Z36- MP5 has a high Mi- 2β ATPase selectivity and specificity. Additionally, we observed the extremely high kinase selectivity of Z36- MP5 by KINOMEscan (Supplementary Fig. 7e), in which the inhibitory activity of Z36- MP5 against a panel of 468 diverse kinases were examined using an in vitro ATP- competitive binding assay.  
+
+<|ref|>text<|/ref|><|det|>[[147, 881, 848, 899]]<|/det|>
+To further identify the targeted specificity of Z36- MP5 in Mi- 2β, we generated a Mi
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 85, 850, 159]]<|/det|>
+\(2\beta\) construct mutated at H727A and then tested the effect of Z36- MP5 by the FRET- based nucleosome repositioning assays. We found that Mi- 2β H727A mutant was no longer sensitive to Z36- MP5 (Supplementary Fig. 7f). This result further confirmed the specificity of Z36- MP5 to Mi- 2β protein.  
+
+<|ref|>sub_title<|/ref|><|det|>[[149, 196, 480, 214]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 233, 850, 307]]<|/det|>
+Overall, the authors have significantly revised their manuscript and answered most of the reviewer comments. But the clarity with regards to the mechanism behind immune sensitization of Mi- 2B loss is still not sufficient and there are still three major points to be addressed before this paper is acceptable for publication:  
+
+<|ref|>text<|/ref|><|det|>[[148, 326, 850, 400]]<|/det|>
+1. While they show how Mi-2B loss activates the ISGs via EZH2 repression, they don't establish that this is the mechanism for the anti-tumor/ CD8 activation response. Akin to the figure S8f, they need to demonstrate that cells lacking EZH2 do not show a combinatorial effect when treated with anti-PD1 and Mi-2B inhibition.  
+
+<|ref|>text<|/ref|><|det|>[[147, 401, 852, 533]]<|/det|>
+Response: We thank the reviewer for the suggestion. We fully agree with the reviewer and performed the experimental therapeutics as suggested. Specifically, we implanted B16F10 cells into C57BL6 mice, treated the mice with MS177 to get the EZH2-targeted inhibition, and then gave anti-PD-1 in combination with Z36-MP5 or vehicles treatment. No treatment effects were detected in mice treated with MS177 to inhibit EZH2. This result further indicates the treatment effect of Z36-MP5 is EZH2 dependent (Fig. S8g in the revised manuscript).  
+
+<|ref|>text<|/ref|><|det|>[[148, 558, 850, 632]]<|/det|>
+2. All their flow data is still in percentage of CD4/CD8s and since the tumor sizes of the combination treatment in particular is much smaller, this is not an accurate representation. They need to show counts of these cells normalized by tumor weight, their counts from IHC alone is not sufficient to answer this critique.  
+
+<|ref|>text<|/ref|><|det|>[[148, 635, 850, 715]]<|/det|>
+Response: We thank the review for the comment and understand the concern of the reviewer. We have quantified the CD4/CD8 cells normalized by tumor weight as suggested. The cell numbers based on the result of flow cytometry analysis has been presented in Fig. 1e and Fig. 1c in the resubmitted manuscript).  
+
+<|ref|>text<|/ref|><|det|>[[148, 754, 852, 865]]<|/det|>
+3. The results with Z36-MP5 differ from their results using Mi-2B loss (shRNA or genetic loss)- the combinatorial effect with anti-PD1 is lower probably because there is no effect on CD4/T regs. The latter effect is consistently seen with both forms of genetic ablation of Mi-2B. They need to provide an explanation for this difference. For example, by comparing degree of EZH2 inhibition or activation of ISGs between the shRNA/genetic ablation of Mi-2B and the drug.  
+
+<|ref|>text<|/ref|><|det|>[[148, 867, 850, 902]]<|/det|>
+Response: We thank the reviewer for the great suggestion and has performed the experiment as suggested. Specifically, B16F10 cells with stable Mi-2β silencing were
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[147, 84, 851, 214]]<|/det|>
+grafted into C57BL6 mice and then treated with Z36- MP5. The expression of H3K27me3, Cxcl9 and Cxcl10 were detected in mouse tumors. We found that the expression of H3K27me3, Cxcl9 and Cxcl10 were significantly repressed after Mi- 2β silencing or Z36- MP5 treatment. However, the Mi- 2β silencing- induced repression was more sufficient than Z36- MP5 treatment- induced repression (Please see the figure below the passage). Z36- MP5 is our first generation of Mi- 2β- targeted inhibitor. We will further improve its specificity and sensitivity by the structure modification.  
+
+<|ref|>image<|/ref|><|det|>[[152, 243, 849, 420]]<|/det|>  
+
+<|ref|>sub_title<|/ref|><|det|>[[148, 474, 269, 490]]<|/det|>
+## Minor points:  
+
+<|ref|>text<|/ref|><|det|>[[148, 493, 848, 528]]<|/det|>
+- They refer to B16 implantation in B6 mice incorrectly as a xenograft model in a few places  
+
+<|ref|>text<|/ref|><|det|>[[148, 529, 848, 547]]<|/det|>
+Response: We thank the reviewer for pointing out these errors. We corrected the errors.  
+
+<|ref|>text<|/ref|><|det|>[[148, 566, 850, 620]]<|/det|>
+- Line 335 is confusing as it suggests that Mi-2B loss promotes T cell mediated cytotoxicity in vitro but their data doesn't support it and they reiterate in the rebuttals that Mi-2B loss doesn't alter efficacy of T cell killing.  
+
+<|ref|>text<|/ref|><|det|>[[148, 622, 848, 657]]<|/det|>
+Response: We thank the reviewer for pointing out this error. We apologize for this. We have corrected the description.  
+
+<|ref|>text<|/ref|><|det|>[[148, 696, 850, 768]]<|/det|>
+- Line 378 is only true with respect to B16 cells. In PM cells, even Z36-MP5 does increase H3K27ac levels, although the degree of increase may be less than that of GSK126. This requires a quantification and rewording on the statement to reflect the actual data.  
+
+<|ref|>text<|/ref|><|det|>[[148, 770, 850, 806]]<|/det|>
+"We confirmed these results, and found that Z36-MP5 repressed the level of H3K27me3, but did not activate H3K27ac"  
+
+<|ref|>text<|/ref|><|det|>[[148, 808, 850, 881]]<|/det|>
+Response: We thank the reviewer for the suggestion. We now correct it to: "We found that Z36- MP5 repressed the level of H3K27me3 in B16 and PM cells. However, compared to GSK126 treatment, the H3K27ac level only slightly increased in PM cells after treatment with Z36- MP5."
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[118, 85, 330, 102]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 137, 437, 154]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[117, 162, 864, 285]]<|/det|>
+The authors have fully and satisfactorily responded to all of my comments. Most importantly, they performed additional experiments to show enrichment of H3K27me3 by ChIP- qPCR at the Cxcl9 and Cxcl10 loci and performing TIMER analysis based on Cxcl9 and Cxcl10 expression levels from TCGA melanomas. They also corrected all editorial mistakes. This is a well- done study.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 345, 437, 362]]<|/det|>
+## Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 371, 412, 388]]<|/det|>
+Authors addressed all my concerns.  
+
+<|ref|>sub_title<|/ref|><|det|>[[118, 450, 437, 467]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[117, 475, 854, 546]]<|/det|>
+I appreciate the efforts that authors have made to answer my comments and the ones of other reviewers. I have 2 more minor comments that should be addressed before publication  
+
+<|ref|>text<|/ref|><|det|>[[117, 579, 875, 676]]<|/det|>
+Comment to response 2: The y axis title of Figure 1e can be improved as it's slightly confusing (typically should read number of cells/mg or g tumor) and neither the legends nor the methods mentioned how the cell numbers of CD4/CD8 were normalized (to weight of tumor and volume of resuspension etc....).  
+
+<|ref|>text<|/ref|><|det|>[[118, 683, 616, 701]]<|/det|>
+This information should be added before before publication.  
+
+<|ref|>text<|/ref|><|det|>[[116, 735, 875, 911]]<|/det|>
+Comment on response 3: This comparison showing that the Z36- MP5 is not entirely as effective as complete genetic ablation of Mi- 2B need to feature in some part of the manuscript as well, even if just as a sentence at the end of the results section or discussion (without the data as such if there is lack of figure space). Because as it stands the manuscript heavily emphasizes that Z36- MP5 is an "effective" inhibitor of Mi- 2B without acknowledging that there is room for improvement to fully capture the immune advantages of complete Mi- 2B silencing.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[149, 86, 360, 101]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[149, 123, 446, 138]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 159, 850, 250]]<|/det|>
+The authors have fully and satisfactorily responded to all of my comments. Most importantly, they performed additional experiments to show enrichment of H3K27me3 by ChIP- qPCR at the Cxcl9 and CxCl10 loci and performing TIMER analysis based on Cxcl9 and Cxcl10 expression levels from TCGA melanomas. They also corrected all editorial mistakes. This is a well- done study.  
+
+<|ref|>text<|/ref|><|det|>[[149, 252, 848, 287]]<|/det|>
+Response: We sincerely appreciate your efforts of reviewing and the recognition of our improvement of our manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[149, 308, 446, 323]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[149, 345, 421, 360]]<|/det|>
+Authors addressed all my concerns.  
+
+<|ref|>text<|/ref|><|det|>[[149, 363, 848, 398]]<|/det|>
+Response: We sincerely appreciate your efforts of reviewing and the recognition of our improvement of our manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[149, 419, 446, 434]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[148, 456, 850, 490]]<|/det|>
+I appreciate the efforts that authors have made to answer my comments and the ones of other reviewers. I have 2 more minor comments that should be addressed before publication  
+
+<|ref|>text<|/ref|><|det|>[[149, 493, 848, 528]]<|/det|>
+Response: We sincerely appreciate your carefully review and your very helpful suggestions.  
+
+<|ref|>text<|/ref|><|det|>[[148, 549, 850, 620]]<|/det|>
+Comment to response 2: The y axis title of Figure 1e can be improved as it's slightly confusing (typically should read number of cells/mg or g tumor) and neither the legends nor the methods mentioned how the cell numbers of CD4/CD8 were normalized (to weight of tumor and volume of resuspension etc....).  
+
+<|ref|>text<|/ref|><|det|>[[149, 623, 605, 639]]<|/det|>
+This information should be added before before publication.  
+
+<|ref|>text<|/ref|><|det|>[[148, 641, 850, 732]]<|/det|>
+Response: We thank the reviewer for the constructive suggestion. The y axis title of Figure 1e has been replaced as suggested. Specifically, the y axis title of Figure 1e has been replaced: cell number / g tumor. We have also added the necessary information that how the cell numbers of CD4/CD8 were normalized (please see line 642- 644 in the resubmitted manuscript).  
+
+<|ref|>text<|/ref|><|det|>[[148, 752, 850, 880]]<|/det|>
+Comment on response 3: This comparison showing that the Z36- MP5 is not entirely as effective as complete genetic ablation of Mi- 2B need to feature in some part of the manuscript as well, even if just as a sentence at the end of the results section or discussion (without the data as such if there is lack of figure space). Because as it stands the manuscript heavily emphasizes that Z36- MP5 is an "effective" inhibitor of Mi- 2B without acknowledging that there is room for improvement to fully capture the immune advantages of complete Mi- 2B silencing.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[148, 85, 850, 139]]<|/det|>
+Response: We thank the reviewer for the suggestion. According to suggestion, we have added the information in discussion. (please see line 445- 447 in the resubmitted manuscript).
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1fb3cea523d354b60366327047668709eea5b0faa24eec8cd860a6f697ad9193/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1fd1dd3b5dac179f03616a5320ecad38d64396b51fab45d2d356fb2ae6986e2f/images_list.json

@@ -0,0 +1,40 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_1.jpg",
+    "caption": "Figure 1. Base effect on the decarboxylation process.",
+    "footnote": [],
+    "bbox": [
+      [
+        360,
+        70,
+        630,
+        284
+      ]
+    ],
+    "page_idx": 10
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_2.jpg",
+    "caption": "Figure 2. Energetic profile for the C-C coupling by considering the (a) proline molecule and the (b) Boc-protected proline.",
+    "footnote": [],
+    "bbox": [
+      [
+        92,
+        108,
+        904,
+        365
+      ]
+    ],
+    "page_idx": 14
+  },
+  {
+    "type": "image",
+    "img_path": "images/Figure_1.jpg",
+    "caption": "Figure 1. Effect of the decarboxylation process in the presence of sodium (a) and potassium (b) cations from inorganic bases.",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 17
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1fd1dd3b5dac179f03616a5320ecad38d64396b51fab45d2d356fb2ae6986e2f/supplementary_0_Peer Review File__1fd1dd3b5dac179f03616a5320ecad38d64396b51fab45d2d356fb2ae6986e2f.mmd

@@ -0,0 +1,597 @@
+
+# nature portfolio  
+
+# Peer Review File  
+
+## Photocatalytic C(sp3)-C(sp3) cross-coupling of carboxylic acids and alkyl halides using a nickel complex and carbon nitride  
+
+Corresponding Author: Professor Gianvito Vile  
+
+This file contains all reviewer reports in order by version, followed by all author rebuttals in order by version.  
+
+Version 0:  
+
+Reviewer comments:  
+
+Reviewer #1  
+
+(Remarks to the Author)  
+
+I evaluated the manuscript by Gianvito Vile and co. entitled "Photocatalytic C(sp3)- C(sp3) cross- coupling of carboxylic acids and alkyl halides using a nickel complex and carbon nitride" with great interest. In the report, the author developed an approach merging nickel catalysis and heterogeneous photoredox catalysis, through the use carbon nitride nanosheets (nC3N4), to form a C(sp3)- C(sp3). The reaction proceeds via the use of carboxylic acid containing compounds (mostly cyclic 4, 5 and 6- membered rings) that undergo decarboxylation to form the corresponding alkyl radical the undergoes reaction with a halogenated substrate. Overall, the study is nicely executed but unfortunately, as stated by the authors, and as analyzed by this reviewer, the main advancement of the present study is the substitution of iridium photosensitizers for more abundant carbon nitride photosensitizers. Whereas this is a noble goal, since the reaction on some of these derivatives are already known and reported using Ir(III) photosensitizer, this study only focuses on repeating this transformation using another abundant photosensitizer. It is this reviewer's opinion that such iterative study should not warrant publication in Nature Communications as they solely rely on changing the photosensitizer. To emphasize this point, the mechanism as proposed relies on the oxidation of the proline derivative. As such, other organic photosensitizers such as acridinium and pyrylium derivatives could also be as efficient for this transformation and it is not conceivable to publish in Nature Communication each time a new photosensitizer is screened, unless additional mechanistic information are obtained, which is not the case here. I would therefore recommend publication in a more specialized journal such as Org. Lett. for example. Here are also specific points to consider, in no particular order:  
+
+- Figure 1: "High temperature" should be "High temperature"  
+
+- K-PHI is not defined in the text.  
+
+- The authors mention purple LED at \(420 \text{nm}\) and then Blue LED at \(420 \text{nm}\) (on page 7 and 8). This should be corrected.  
+
+- Table 1: The structure of the proline derivative is wrong (it should be COOH and not CH2OOH)  
+
+- What is the recovery and recyclability of C3N4? The authors have a short paragraph on it that is immediately re-engaged in the same reaction but I wonder how much would be recovered at the end of the reaction and if it would operate as efficiently over 5 new reactions.  
+
+- I could not judge the DFT calculations as it is beyond my area of expertise but I would recommend that the author carry out some excited-state quenching experiments to corroborate their mechanism (although it is identical to the one proposed by MacMillan). Stern-Volmer experiments using carbon nitride and the different partners (proline, Nickel, Proline and nickel together) should be investigated to investigate possible quenching routes.  
+
+## Reviewer #2  
+
+(Remarks to the Author)  
+
+sp3 C- sp3 C cross- coupling of carboxylic acids and alkyl halides presents a strong tool for the synthesis of a diverse array of functionalized drug molecules. This work modified the reported Ni- Ir process by replacing Ni with PCN, and the results are important and interesting in some degree. However, the following issues would be seriously addressed before the possible
+
+<--- Page Split --->
+
+acceptance: 1) the roles of CNx, Ni, ligand, base can be further explored. 2) The yield of 3a increases, and that of 4a decreases as the Na2CO3 increases from 1 to 2 equiv; while the 3a yield decreases but the 4a yield increases as the Na2CO3 increases from 2 to 3; and then 3a increases but 4a decreases as the alkaline increases from 3 to 4. Please explain. 3) Compare Table 8 and Table 1, there are too many same control results are repeatedly presented in the two tables. 4) Many related reports regarding of sp3 C- sp3 C cross coupling are missed, ie. ACS Med. Chem. Lett. 2018, 9, 7, 773- 777. 5) The description regarding of Figure 1 is different from the Figure. Figure 1a is not for a photocatalytic homogeneous decarboxylative \(\mathsf{C}(\mathsf{sp}^2) - \mathsf{C}(\mathsf{sp}^3)\) coupling using Ni catalyst and Ir photocatalyst; that for Figure 1b is also wrong. 6) Where is the result for recyclability? 7) some experimental proof for the change of Ni chemical state is better to be presented? 8) Please explain how to obtain the compared CO2 emission in Figure 5.  
+
+## Reviewer #3  
+
+(Remarks to the Author)  
+
+This manuscript describes dual Ni/carbon nitride catalyzed C(sp3)- C(sp3) coupling of carboxylic acids and alkyl halides. Although replacing noble Ir by carbon nitride has been demonstrated previously, this study offers a more practical protocol for dicarboxylic C(sp3)- C(sp3) coupling, in terms of cost, catalyst recycle and low CO2 emission.  
+
+(1) In an earlier publication, the authors reported carbon-oxygen coupling from the same coupling partners employing single atom Ni anchored on carbon nitride. Carbon-oxygen coupling also proceed with different yield, depending on conditions. It is worth to find the underlying reasons for different coupling manners, which will increase the scientific impact and guide catalyst development.  
+
+(2) The statement "molecular insights into the role of nickel single atoms in facilitating photodecarboxylation and subsequent C-C bond formation", where single atom is used for Ni in the by coordinated complex might not be suitable.  
+
+(3) The manuscript requires substantial reorganization. The characterizations and condition optimizations should be  
+
+(4) The photoreaction conditions should be provided in detail (solution volume, incident light intensity at different wavelengths); MeCN (x M) is confusing, whose concentration in the bracket indicate? MeCN or alkyl bromide?  
+
+(5) Please provide the data for catalytic performance of the recovered catalyst, together with the structural characterizations (XRD, TEM, and the possible Ni aggregates)  
+
+(6) Ni(O) is proposed to interact with alkyl halide or Intermediate IV. The initial state of Ni is Ni(II). And there are controversies on the active forms of Ni (Ni(I) or Ni(O)). The authors may want to track the evolution of Ni on CN experimentally.  
+
+(7) Beside free energy changes, the transition states and the corresponding barriers in the two proposed pathways should be provided to support the assignment of the dominant one.  
+
+(8) Calculation and references for GHG emissions should be provided.  
+
+## Reviewer #4  
+
+(Remarks to the Author)  
+
+In this manuscript, Gianvito Vile et al. present a study on C(sp3)- C(sp3) cross- coupling in a semi- heterogeneous system using an nCNx photocatalyst. While the exploration of such catalysts and their application in cross- coupling reactions is undoubtedly significant, and the concept of utilizing nCNx as a heterogeneous photocatalyst is intriguing, the manuscript falls short in distinguishing this approach from prior research. For instance, in a previous study (Nat. Synth. 2023, 2, 1092- 1103), Gianvito Vile already employed a Ni single- atom catalyst for C- O coupling but did not adequately elucidate the rationale behind the occurrence of decarboxylation in this study. A more in- depth investigation into the decarboxylation mechanism is needed, particularly in light of earlier reports (Angew. Chem. Int. Ed. 2024, 63, e202405902; Adv. Synth. Catal. 2020, 362, 3898- 3904). Additionally, it is well- established that homogeneous nickel catalysts are prone to deactivation in dual photoredox/nickel- catalyzed processes, as highlighted in a previous study (Nat. Catal. 2020, 3, 611- 620). This manuscript does not sufficiently address this critical issue. I do not recommend its publication in Nature Communications. The manuscript should include more comprehensive studies, incorporating extensive experimental work and computational analysis, to elucidate a mechanism and demonstrate the clear advantages of semi- heterogeneous decarboxylative cross- coupling.  
+
+Comment #1: The previous studies referenced on page 2 do not correspond accurately with the information presented in Figure 1. For instance, the study shown in Figure 1c is from 2016, not 2014 as stated in the text. Please ensure that all studies mentioned in the text are correctly aligned with those depicted in the figure, and address any typographical errors.  
+
+Comment #2: The authors optimized several photocatalysts, including nCNx, gCNx, mpg- CNx, K- PHI, and recovered nCN. However, the manuscript lacks sufficient characterization data to substantiate the comparison and validation of nCNx against the other photocatalysts. Although nCNx was prepared through a thermal exfoliation process from gCNx, no data are provided to describe the morphology, thickness, or surface area of nCNx relative to gCNx. Therefore, it is essential to include comprehensive characterization data for all the photocatalysts in Figure 2, such as BET, XRD, AFM, and UV analyses. Additionally, TEM, SEM, XPS, and FT- IR data should be provided for each photocatalyst. Furthermore, the method used to calculate the reported C/N ratios ('0.61- 0.67') should be clearly explained.  
+
+Comment #3: In this study, the authors present an optimization of a photocatalyst, identifying nCNx as the optimal choice. However, several concerns arise regarding the rationale behind selecting nCNx over other catalysts. The authors suggest that mpg- CNx is less effective due to its higher solution viscosity, which they claim reduces the availability of active sites. This assertion, however, lacks sufficient experimental evidence or comprehensive characterization of the photocatalysts in question. While it is acknowledged that mpg- CNx may exhibit increased viscosity, this factor alone does not provide a strong
+
+<--- Page Split --->
+
+basis for dismissing its catalytic potential. In fact, mpg- CNx typically has a larger surface area, which is generally associated with an increased number of active sites, as documented in the literature (Chem. Eur. J. 2015, 21, 526- 530; J. Phys. Chem. C 2012, 116, 19644- 19652; Angew. Chem. Int. Ed. 2015, 54, 12868- 12884). This contradicts the authors' claim that mpgCNx is less suitable due to fewer active sites. The manuscript lacks detailed characterization data, such as BET surface measurements, pore size distribution, or SEM/TEM imaging, which are essential for substantiating the claim that mpg- CNx has a reduced number of active sites or is significantly impacted by viscosity- related issues. Moreover, the authors report comparable reaction yields between CNx (3a 41% and 4a 23%) and mpg- CNx (3a 34% and 4a 18%), indicating that mpgCNx could be a viable alternative. The rejection of its efficacy based on unproven assumptions about solution viscosity and active site availability is unconvincing.  
+
+Comment #4: In addition to the concerns mentioned above, there is an inconsistency in the manuscript regarding the mention of "boron- doped mpg- CNx." This material is introduced without prior context or discussion in the manuscript, leading to confusion. The authors should ensure that all materials discussed are properly introduced and contextualized.  
+
+Comment #5: The manuscript includes product yield data that have been confirmed using HPLC calibration. However, the authors have not provided the detailed evidence of the HPLC calibration curves that were used. To ensure scientific reproducibility, it is imperative that the authors provide comprehensive details regarding the calibration process. Furthermore, there are notable inconsistencies between the product yield values reported in the main text and those presented in the Supporting Information (SI). These inconsistencies call into question the credibility of the data and must be addressed. The authors should conduct a thorough review and make the necessary corrections to ensure consistency across the manuscript. Additionally, the SI lacks NMR spectra for several compounds discussed, despite the fact that these are typically required to confirm the synthesized products. The authors must provide the NMR data for all relevant compounds to support the claims made in the manuscript.  
+
+Comment #6: This paper presents DFT calculation data to explain the Ni- catalyzed reaction pathway. However, the DFT data presented, which uses proline, does not offer a sufficiently reliable explanation for the mechanistic pathway. Based on the substrate scope study in this paper, the reactivity and selectivity are influenced by the substituents, with even methyl- substituted proline showing no reactivity. Therefore, it would be more appropriate to present DFT calculation results using Boc- protected proline rather than proline. Additionally, the Ni- catalytic cycle has already been extensively addressed in previous studies (Nature 2016, 536, 322- 325; J. Org. Chem. 2024, 89, 11136- 11147; Science 2014, 345, 437- 440). It would be beneficial to develop data that provides deeper mechanistic insights into decarboxylation beyond esterification.  
+
+Comment #7: The authors present GHG emission data derived from the equations described in the 'Energy Calculations' section of the Supplementary Information (SI) to support the sustainability of carbon nitride compared to Ir- based complexes. However, the mentioned 'Energy Calculation' section is absent from the SI. To substantiate the sustainability advantages of carbon nitride, it is recommended that the authors provide a detailed calculation process along with reliable references.  
+
+## Reviewer #5  
+
+(Remarks to the Author)  
+
+The authors described the use of graphitic carbon nitride as a photocatalytic system to drive cross- coupling between alkyl halides and carboxylic acids. Mechanistic studies are also reported. In my view, the manuscript has been well written and the results are sound. Thus, I recommend its publication after having addressed the points below.  
+
+1. Additional references regarding the use of carbon nitride-based photocatalytic systems for the functionalization of organic compounds should be included in the introductory section, for instance: Science 365, 360-366 (2019), Angew. Chem. Int. Ed., 2023, e202313540, ACS Catal. 2023, 13, 13414-13422, Sci. Adv., 2020, 6, eabc9923, ACS Catal. 2024, 14, 11308-11317, ACS Nano, 2021, 15, 3621-3630, Chem. Sci., 2022, 13, 9927, Nature Catalysis, 3, 611-620 (2020), Adv. Sci. 2023, 10, 2303781, among others.  
+
+2. The reaction scope should be expanded. Is it possible to use fatty acids, alkyl chlorides and alkyl triflates as starting materials? Moreover, the generality of the photocatalytic system with respect to more synthetically useful organic substrates, namely natural products or active drugs, should be addressed. For instance, is it possible to use other natural products or bio-active molecules containing carboxylic moieties as substrates?  
+
+3. Is it possible to perform the model reaction in a gram scale? 
+4. I suggest to include a general procedure for the photocatalytic experiments along with a picture of the reaction set-up and its description within the SI. 
+5. The authors should better characterize the photocatalyst after a series of catalytic cycles. 
+6. I suggest an improvement on the structure of the manuscript due to the presence of typos (e.g., structure compound 1 within table 1)  
+
+Version 1:  
+
+Reviewer comments:  
+
+Reviewer #1  
+
+(Remarks to the Author) I commend the authors for the quality and seriousness of their revisions. The manuscript is much improved and I felt like the
+
+<--- Page Split --->
+
+author responded professionally to all the comments. I have a very few minor comments:  
+
+In the novel figure 4, panel B, the "micro" symbol does not display properly, and only "u" is readable. In Figure 4, panel C, the authors mislabeled the Y- axis, and I should be 10/1 and not 1/10. In addition, compound 2, 1 and 1+ Nicat + dMeObpy (probably mislabeled as well), seem to have a negative slope that the author did not comment on. A negative slope would imply that the photoluminescence intensity increases as the quencher is added. Still in figure 4, I would recommend forcing the intercept of all lines at (0;1), as this is a "true" datapoint according to the Stern- Volmer equation, \(10 / 1 = 1 + \text{Ksv}[Q]\) , so when \([Q] = 0\) , \(10 / 1\) should be 1.  
+
+## Reviewer #2  
+
+(Remarks to the Author)  
+
+I read through the revised draft and the response to comments carefully, and find that the issues were basically addressed, and now can be acceptable.  
+
+## Reviewer #3  
+
+(Remarks to the Author)  
+
+In the revised manuscript, the authors provided more insights into the C- C coupling (other than the previously reported C- O) by systematically excited state quenching, photon energy/base/solvent dependence and calculation. Other issues are also addressed.  
+
+## Reviewer #4  
+
+(Remarks to the Author)  
+
+In the revised manuscript, Gianvito Vile and colleagues have addressed some of the issues raised in previous comments. While the updated version demonstrates improvements, several critical concerns remain, rendering the manuscript unsuitable for publication in Nature Communications. The manuscript still lacks a thorough characterization of the materials used, which is essential for validating the reported material properties. Moreover, the choice of a semi- heterogeneous system requires a stronger justification to establish the significance of the findings. Additional concerns stem from experimental aspects, particularly the "numbering- up" scale- up process, which raises doubts about the reliability and scalability of the results. Consequently, despite the authors' notable efforts, I regret to recommend against publication in Nature Communications. Detailed comments are provided below:  
+
+Comment #1: The manuscript does not sufficiently highlight the advantages of the semi- heterogeneous system through a detailed comparison with other catalytic systems, such as homogeneous and single- atom heterogeneous systems. While the authors mention that the semi- heterogeneous system mitigates catalyst deactivation via stable interactions between the photocatalyst and nickel species, the manuscript lacks a clear and rational explanation to support this claim.  
+
+Comment #2: Although the authors stated that they included additional characterizations, critical analyses such as TEM, SEM, XPS, and AFM remain missing. The explanation of catalyst screening results is also insufficient to justify the selection of nCNx as the optimal material. Characterization data for the recovered nCNx are particularly lacking. For instance, TEM images and EDS mapping are necessary to determine whether Ni aggregates have formed. Despite claims that such data were included, they are absent from both the manuscript and the supplementary information.  
+
+Comment #3: For gram- scale synthesis, the authors employed a numbering- up strategy instead of scaling up the system size. This approach is unconvincing for demonstrating the scalability of heterogeneous semiconductors. Moreover, the numbering- up experiments were conducted on an even smaller scale (0.2 mmol) than the optimized conditions (0.3 mmol). As a result, the manuscript does not sufficiently support the claim that nCNx is a sustainable, efficient, and cost- effective alternative to traditional iridium- based photocatalysts, especially from an industrial perspective.  
+
+Comment #4: The authors propose that energy differences resulting from interactions between 1 and the base explain variations in product yields. This rationale is plausible in cases without a base (1.48 eV) or with DBU (2.31 eV). However, the manuscript fails to address the significant discrepancies between Na2CO3 and K2CO3 results. Despite their similar intermediate energies (1.24 eV for Na2CO3 and 1.25 eV for K2CO3), their yields differ substantially, with Na2CO3 yielding 33% of 3a and 17% of 4a, while K2CO3 yields 34% and 44%, respectively. Additional justification or experimental data is needed to clarify these differences.  
+
+Comment #5: Previously, it was suggested to provide detailed descriptions of the "Energy Calculations" for GHG emission data. However, the revised manuscript still lacks a clear explanation of the calculation procedures and data references. To emphasize the environmental advantages of the semi- heterogeneous system over homogeneous systems, a detailed explanation of how the GHG emission data were derived is essential.  
+
+Comment #6: Several inconsistencies in the manuscript undermine its reliability. These include: (1) Comparison of substrate yield with a prior study (MacMillan, 2016), despite differing experimental conditions (lr system: room temperature, 48 h), which makes the comparison invalid. (2) Unexplained changes in nCNx surface area from 12 to 23 m²/g, raising questions about synthesis reproducibility. (3) Conflicting details about the washing procedure for the recovered catalyst. The supporting information states the catalyst was washed with ethyl acetate and water, while the manuscript and prior responses mention acetonitrile.  
+
+Comment #7: Several minor typos and formatting errors persist in the manuscript. For instance, the author's name is misspelled as "MCMillan" instead of "MacMillan" on page 3. Additionally, discrepancies between figure and data references within the manuscript need to be corrected.
+
+<--- Page Split --->
+
+(Remarks to the Author)  
+
+The revised version of the Manuscript has been well organized and the results are very interesting. I personally found this version of the manuscript more detailed and clear. In addition, the authors addressed most of the reviewers' comments satisfactory. Thus, I think that the new version of the manuscript is now suitable for publication in Nature Communications.  
+
+Version 2:  
+
+Reviewer comments:  
+
+Reviewer #1  
+
+(Remarks to the Author)  
+
+I was personally satisfied by the changes carried out last time by the authors. I have however to stress that I agree with reviewer 4, that a "numbering- up" approach is, in my opinion, not a proper scale up, but rather represents a reproducibility experiment from which average yields and standard errors can be obtained. I understand the rational from the authors about light penetration, Beer- Lambert law etc, but a true scale up also implies accounting for these issues and showing applicability and practicability for industrial developments. Light penetration is often industrially tackled by multiplying irradiation sources, increasing fluence, inserting irradiation systems (light tubes) inside the reaction reactor etc. I am not sure if a scale up is desperately needed, but I feel like the author should acknowledge the limitation of the "numbering- up" approach that they propose.  
+
+Reviewer #4  
+
+(Remarks to the Author)  
+
+Some of the previous comments have been addressed, and the revised manuscript shows improvement. However, several important issues remain insufficiently explained. I recommend the authors provide further clarification based on the following comments. I will reconsider the publication of the manuscript once these concerns are fully addressed.  
+
+Comment 1:  
+
+It remains unclear whether byproduct 4a was formed when the \(\mathrm{Ni_3@nCNx}\) catalyst was used. According to the authors' own cited reference (Nat. Synth. 2023, 2, 1092- 1103), \(\mathrm{Ni_3@nCNx}\) is expected to favor C- O coupling, which in this case would yield byproduct 4a. The authors must clearly state this and provide a mechanistic rationale for the observed difference in coupling selectivity (C- C vs. C- O) between the current semi- heterogeneous system and the previously reported heterogeneous system. Without such clarification, the mechanistic underpinnings of the semi- heterogeneous system- the core focus of the manuscript- remain ambiguous and must be addressed in the revised manuscript.  
+
+Comment 2:  
+
+While I acknowledge the practical limitations associated with photochemical scale- up, I remain unconvinced that the use of a numbering- up strategy in a batch system sufficiently demonstrates the scalability of the catalytic process. As discussed in reference 43 (Chem. Rev. 2022, 122, 2752- 2906), numbering- up is a strategy more appropriately applied to continuous flow systems. In contrast, the current study employs a batch process, where this approach is less relevant. To substantiate claims of scalability, the authors should either demonstrate a genuine scale- up within a batch system or apply a numbering- up strategy in a flow system.  
+
+Comment 3:  
+
+The issue regarding yield comparison remains unresolved. The standard conditions in the present study require 72 hours at elevated temperature, whereas MacMillan's protocol achieves C- C coupling in 48 hours at room temperature. Under these markedly different conditions, similar yields (e.g., 71% in this study vs. 85% in MacMillan's work for compound 3a) do not constitute a valid comparison. Furthermore, the article cited by the authors (Chem. Sci. 2019, 10, 5837- 5842) compares mechanochemical and solution- phase conditions within a single study, carefully controlling all variables aside from the activation method. In contrast, the current manuscript compares yields from two independent reports without such control, which invalidates the justification provided. To enable a sound comparison, the authors must present data obtained under matched experimental conditions. Without these corrections, I respectfully maintain that the justification for the manuscript remains unconvincing.  
+
+Version 3:  
+
+Reviewer comments:  
+
+Reviewer #4  
+
+(Remarks to the Author)  
+
+I have reviewed the revised manuscript and the responses to the reviewers' comments. The major issues have been adequately addressed, and the manuscript is now suitable for acceptance.
+
+<--- Page Split --->
+
+Open Access This Peer Review File is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.  
+
+In cases where reviewers are anonymous, credit should be given to 'Anonymous Referee' and the source.  
+
+The images or other third party material in this Peer Review File are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.  
+
+To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/
+
+<--- Page Split --->
+
+## Point-by-point response to the Reviewers' comments  
+
+(original comments in blue, replies in black, actions in bold)  
+
+## Reviewer #1  
+
+I evaluated the manuscript by Gianvito Vile and co. entitled "Photocatalytic C(sp3)- C(sp3) cross- coupling of carboxylic acids and alkyl halides using a nickel complex and carbon nitride" with great interest. In the report, the author developed an approach merging nickel catalysis and heterogeneous photoredox catalysis, through the use carbon nitride nanosheets (nC3N4), to form a C(sp3)- C(sp3). The reaction proceeds via the use of carboxylic acid containing compounds (mostly cyclic 4, 5 and 6- membered rings) that undergo decarboxylation to form the corresponding alkyl radical the undergoes reaction with a halogenated substrate. Overall, the study is nicely executed but unfortunately, as stated by the authors, and as analyzed by this reviewer, the main advancement of the present study is the substitution of iridium photosensitizers for more abundant carbon nitride photosensitizers. Whereas this is a noble goal, since the reaction on some of these derivatives are already known and reported using Ir(III) photosensitizer, this study only focuses on repeating this transformation using another abundant photosensitizer. It is this reviewer's opinion that such iterative study should not warrant publication in Nature Communications as they solely rely on changing the photosensitizer. To emphasize this point, the mechanism as proposed relies on the oxidation of the proline derivative. As such, other organic photosensitizers such as acridinium and pyrrylium derivatives could also be as efficient for this transformation and it is not conceivable to publish in Nature Communication each time a new photosensitizer is screened, unless additional mechanistic information are obtained, which is not the case here.  
+
+We sincerely thank the Referee for the opportunity to openly discuss this matter. While it may appear that our work simply substitutes a more abundant photosensitizer (carbon nitride) in place of the traditionally used Ir(III) photosensitizer, our study goes beyond a mere replacement and reflects a significant advancement in catalyst design and reaction engineering. The use of carbon nitride nanosheets (nCNx) leverages, in fact, a novel catalytic interface where nickel and nCNx act cooperatively to achieve selective decarboxylative alkylation. This transformation required a fine- tuning of reaction conditions and a careful engineering of the catalyst interface to facilitate the desired coupling while avoiding side reactions. Through this design, we demonstrate that carbon nitride is not simply an alternative photosensitizer but a crucial component enabling controlled reactivity that was previously challenging to achieve with traditional photosensitizers. Moreover, modification of carbon nitride alters the selectivity patterns.  
+
+This aspect, perhaps not fully elaborated in the original submission, has now been addressed more comprehensively. We have added DFT calculations and spectroscopic studies, which enhance the mechanistic understanding of the reaction and provide valuable insights into the adaptive role of carbon nitride. These findings clarify how the carbon nitride interface facilitates the formation of decarboxylative C(sp3)- C(sp3) coupling, while its modification leads to ester formation under specific conditions, revealing the dynamic influence on reaction pathways.  
+
+We would like to add a final note: the field of chemistry requires not only the discovery of new reactions but also the development of greener catalysts—two complementary and equally important research directions. In this work, we address the latter by replacing a homogeneous, precious metal- based photosensitizer with an earth- abundant, heterogeneous alternative, aligning directly with green chemistry principles. By advancing the use of sustainable, metal- free photosensitizers, this research meets the pressing need for environmentally friendly and cost- effective catalytic systems that do not rely on scarce, expensive metals. The application of carbon nitride as a photosensitizer in our study not only cuts costs but also opens the door to broader applications in photoredox catalysis; and its scalability and environmental benefits provide a promising pathway for C- C bond formation, thus making a meaningful contribution to the need of more sustainable chemical production. Therefore, we are convinced that the work fits the scope of Nature Communications.
+
+<--- Page Split --->
+
+1) Figure 1: "High temperature" should be "High temperature". The authors mention purple LED at 420 nm and then Blue LED at 420 nm (on page 7 and 8). This should be corrected. Table 1: The structure of the proline derivative is wrong (it should be COOH and not CH2OOH).  
+
+We have corrected these typographical errors. We extend our thanks to the Reviewer for their careful reading.  
+
+2) K-PHI is not defined in the text.  
+
+In the revised version of the manuscript, all photocatalysts (including K-PHI) have been defined. We thank the Reviewer for the comment.  
+
+3) The authors should emphasize also or compare the yields with those reported by MacMillan in 2016, which for \(25\%\) of them are roughly identical.  
+
+We agree with the Reviewer that this can be important. For the substrates that have been precedingly reported by MacMillan in 2016, we now include a vis-à-vis comparison yield comparison.  
+
+4) What is the recovery and recyclability of C3N4? The authors have a short paragraph on it that is immediately re-engaged in the same reaction but I wonder how much would be recovered at the end of the reaction and if it would operate as efficiently over 5 new reactions.  
+
+We thank the Reviewer for their suggestion. We have now conducted 5 consecutive reactions, consistently obtaining the alkylated product with yields consistently between \(77\%\) to \(81\%\) . After each reaction, the nCNx catalyst was recovered by centrifugation, washed with acetonitrile and water, and dried at \(65^{\circ}C\) overnight. We monitored the recovery rate of the catalyst, obtaining quantitative yields in each instance. Although not requested, we also carried out post-catalysis characterizations (i.e., N2 physisorption, XRD, CHN elemental analysis, and ICP-OES) to prove the structural robustness of the catalyst and the absence of any potential Ni doping of the carbon nitride carrier (ICP). These results validate the stability of the catalyst and its potential for long-term use in repeated catalytic cycles without degradation or contamination.  
+
+5) I could not judge the DFT calculations as it is beyond my area of expertise but I would recommend that the author carry out some excited-state quenching experiments to corroborate their mechanism (although it is identical to the one proposed by MacMillan). Stern-Volmer experiments using carbon nitride and the different partners (proline, Nickel, Proline and nickel together) should be investigated to investigate possible quenching routes.  
+
+We appreciate the Reviewer's suggestion. We would like to emphasize that the mechanism we propose differs from that of MacMillan's reported in Nature 2016, 536- 325. In MacMillan's mechanism, it was reported that Ni(0) first captures the decarboxylated radical to form Ni(I), followed by the oxidative addition of the alkyl bromide to generate Ni(III). In contrast, our proposed mechanism involves Ni(0) undergoing oxidative addition of the alkyl halide first, forming Ni(II), which is then followed by a radical oxidative trap, resulting in Ni(III). To gain insights into the SET step and validate the mechanism, we have conducted Stern- Volmer experiments to study the interaction of nCNx with various substrates. Our results show significant quenching of nCNx fluorescence upon interaction with deprotonated proline. These new findings have been incorporated into the amended manuscript to provide additional evidence supporting our proposed mechanism. We are grateful to the Reviewer for prompting this investigation.  
+
+## Reviewer #2  
+
+sp3 C- sp3 C corss- coupling of carboxylic acids and alkyl halides presents a strong tool for the synthesis of a diverse array of functionalized drug molecules. This work modified the reported Ni- Ir process by replacing Ni with PCN, and the results are important and interesting in some degree.
+
+<--- Page Split --->
+
+We appreciate the Reviewer's recognition of the significance of these results and the broader impact they may have on the field of synthesis and process chemistry.  
+
+1) the roles of CNx, Ni, ligand, base can be further explored.  
+
+We greatly appreciate the Reviewer's suggestion. In response, we have conducted further experiments and DFT calculations to clarify the role of these compounds in the revised manuscript. Specifically, Stern- Volmer quenching experiments have been performed and have shown a quenching interaction between the deprotonated carboxylic acid substrate and the photoexcited nCNx, leading to a reduction in the fluorescence of the nCNx photocatalyst. Furthermore, to better understand the base's selectivity, DFT calculations have been added to demonstrate the favorable interaction between the base counter anion and the deprotonated carboxylic acids, facilitating the decarboxylation step. Finally, electron paramagnetic resonance (EPR) experiments were performed to observe the presence of different oxidation states of Ni during the homogeneous catalytic cycle. These findings strengthen our mechanistic hypothesis and provide a comprehensive picture of the factors that contribute to the catalytic activity, selectivity, and efficiency of the system. We have incorporated these new insights and experimental data into the revised manuscript and hope that these additions address the Reviewer's suggestions satisfactorily. Thank you once again for your constructive feedback, which has greatly improved the clarity and depth of our study.  
+
+2) The yield of 3a increases, and that of 4a decreases as the Na2CO3 increases from 1 to 2 equiv; while the 3a yield decreases but the 4a yield increases as the Na2CO3 increases from 2 to 3; and then 3a increases but 4a decreases as the alkaline increases from 3 to 4. Please explain.  
+
+We thank the Reviewer for this observation regarding the yield variations of 3a and 4a with different Na2CO3 concentrations. While it is correct that there are small fluctuations in yield as the base concentration changes, these variations are minor, in the range of 5- 6%. Given this limited change, we do not believe that strong conclusions can be drawn regarding the effect of Na2CO3 concentration on product distribution. Within the range of base concentrations and types investigated, we observed no significant influence on the reaction outcome.  
+
+3) Compare Table 8 and Table 1, there are too many same control results are repeatedly presented in the two tables.  
+
+Our intention was to provide a detailed description of the optimization procedures and control experiments in the Supporting Information (Tables S1- S7), while highlighting the key results in the main manuscript (Table 1). In the revised manuscript, we have ensured clarity on this point.  
+
+4) Many related reports regarding of sp3 C- sp3 C cross coupling are missed, ie. ACS Med. Chem. Lett. 2018, 9, 7, 773-777.  
+
+We appreciate the Reviewer's suggestion and have included the referenced report (ACS Med. Chem. Lett. 2018, 9, 7, 773- 777) for C(sp2)- C(sp3) coupling in the revised manuscript. Additionally, we have incorporated other relevant photocatalytic reports regarding C(sp3)- C(sp3) bond formation, including: Adv. Synth. Catal. 2020, 362, 2367- 2373; J. Am. Chem. Soc. 2018, 140, 50, 17433- 17438; and J. Am. Chem. Soc. 2023, 145, 14, 7736- 7742.  
+
+5) The description regarding of Figure 1 is different from the Figure. Figure 1a is not for a photocatalytic homogeneous decarboxylative C(sp2)-C(sp3) coupling using Ni catalyst and Ir photocatalyst; that for Figure 1b is also wrong.  
+
+We have corrected the typo. We extend our thanks to the Reviewer for their comment.
+
+<--- Page Split --->
+
+## 6) Where is the result for recyclability?  
+
+6) Where is the result for recyclability?We thank the Reviewer for their question. In our revised manuscript, we have repeated the reaction five times using nCNₓ photocatalyst, consistently achieving the alkylated product in the stable range of 77% to 81% yield. After each reaction, the nCNₓ catalyst was recovered through centrifugation, washed with MeCN and water, and then dried in the oven overnight, resulting in a quantitative recovery of the catalyst in each instance. Additionally, post-catalysis characterization analysis was performed to demonstrate the stability and structure endurance of nCNₓ in the five reaction cycles.  
+
+7) some experiomntal proof for the change of Ni chemical state is better to be presented?  
+
+7) some experiomntal proof for the change of Ni chemical state is better to be presented?We perceive this comment as related to the preceding discussion (point 1 above). Electron paramagnetic resonance (EPR) experiments were performed to observe the presence of different oxidation states of Ni during the homogeneous catalytic cycle. These findings align with and support the proposed mechanism.  
+
+8) Please explain how to obtain the compared CO2 emission in Figure 5.  
+
+8) Please explain how to obtain the compared CO2 emission in Figure 5.Many thanks for this valuable feedback. We have added further explanation on how the CO₂ emissions in Figure 5 were calculated to the revised manuscript. We hope this addition meets the expectations of the Reviewer.  
+
+## Reviewer #3  
+
+This manuscript describes dual Ni/carbon nitride catalyzed C(sp3)- C(sp3) coupling of carboxylic acids and alkyl halides. Although replacing noble Ir by carbon nitride has been demonstrated previously, this study offers a more practical protocol for dicarboxylic C(sp3)- C(sp3) coupling, in terms of cost, catalyst recycle and low CO2 emission.  
+
+We sincerely thank the Reviewer for their positive feedback.  
+
+(1) In an earlier publication, the authors reported carbon-oxygen coupling from the same coupling partners employing single atom Ni anchored on carbon nitride. Carbon-oxygen coupling also proceed with different yield, depending on conditions. It is worth to find the underlying reasons for different coupling manners, which will increase the scientific impact and guide catalyst development.  
+
+The formation of compounds 3a and 4a proceeds via distinct reaction pathways. Therefore, when the pathway leading to 3a is slowed or inhibited, the formation of 4a becomes more favorable. We propose that the selectivity for 3a is linked to a preferred single electron transfer (SET) step, which ultimately leads to the decarboxylation process. Among the factors promoting SET, we emphasize the use of 420 nm light over 460 nm, which corresponds to the maximal absorbance of carbon nitride. This may help stabilize its triplet state. Furthermore, in acetonitrile, a polar solvent known for facilitating SET, the electron transfer from photoexcited nCNₓ to the deprotonated carboxylic acid becomes the dominant pathway. Finally, a selectivity trend was observed between the formation of 3a and the type of base utilized. To stress this point, Stern-Vomer experiments indicate a stronger interaction of the photoexcited nCNₓ between 1 in presence of Na₂CO₃ than in presence of Cs₂CO₃. Besides, we modeled the interactions between Na, K, and DBU with the deprotonated carboxylic acid by considering the Boc-protected L-proline. The results suggest that Na may better promote the decarboxylation process while the DBU will block it (see Figure 1). Collectively, these results shed light on the factors that govern the interplay between carbon-oxygen and carbon-carbon coupling, and expand the versatility of single-atom catalysts for diverse coupling reactions. We have included this discussion in the text.
+
+<--- Page Split --->
+![](images/Figure_1.jpg)
+
+<center>Figure 1. Base effect on the decarboxylation process. </center>  
+
+(2) The statement "molecular insights into the role of nickel single atoms in facilitating photodecarboxylation and subsequent C–C bond formation", where single atom is used for Ni in the bpy coordinated complex might not be suitable.  
+
+We thank to the Reviewer's for their comment. In the revised manuscript, we have corrected that statement.  
+
+(3) The manuscript requires substantial reorganization. The characterizations and condition optimizations should be condensed, and many details are better suited for inclusion in the Supplementary Materials.  
+
+We are not in line with the Reviewer's suggestion to condense the characterization and condition optimization sections, as these elements are integral to our study's focus on catalyst design. Detailed descriptions of these aspects provide critical insights into how specific conditions and structural features influence catalytic performance. Since we aim to offer a comprehensive understanding that will be valuable for researchers working on similar catalyst development, we have not taken any action on this specific matter.  
+
+(4) The photoreaction conditions should be provided in detail (solution volume, incident light intensity at different wavelengths); MeCN (x M) is confusing, whose concentration in the bracket indicate? MeCN or alkyl bromide?  
+
+We appreciate the Reviewer's suggestion. In the revised version of Supplementary Information, we now better explain the reaction conditions, indicating the reaction volume and the light intensity.  
+
+We apologize if the concentration of MeCN was confusing; this was related to the limiting reagent, which is the alkyl halide (0.2 mmol, 1.0 equiv.). therefore, \(0.05M = 0.2V; V = 4\) mL. In the revised version of the manuscript we express the amount of MeCN in mL instead of M.  
+
+(5) Please provide the data for catalytic performance of the recovered catalyst, together with the structural characterizations (XRD, TEM, and the possible Ni aggregates)  
+
+We thank the Reviewer for their suggestion. In our revised manuscript, we have carried out post-catalysis characterizations (BET, XRD, elemental analysis) to prove the structural robustness of the catalyst.  
+
+(6) Ni(0) is proposed to interact with alkyl halide or Intermediate IV. The initial state of Ni is Ni(II). And there are controversies on the active forms of Ni (Ni(I) or Ni(0)). The authors may want to track the evolution of Ni on CN experimentally.
+
+<--- Page Split --->
+
+We agree that tracking the evolution of nickel during the catalytic cycle is essential. While the initial state of nickel in our system is indeed Ni(II), electron paramagnetic resonance (EPR) experiments confirm the presence of distinct oxidation states of nickel throughout the catalytic cycle. These EPR findings align well with the proposed mechanism.  
+
+(7) Beside free energy changes, the transition states and the corresponding barriers in the two proposed pathways should be provided to support the assignment of the dominant one.  
+
+We have modified the text to better explain our choice. We considered the analysis of the reaction barriers using the thermodynamics approach developed by Norskov et al. (see: J. Catal. 2002, 209, 275-278; J. Catal. 2001, 197, 229-231; Nat. Chem. 2009, 1, 37-46; J. Catal. 2004, 224, 206-217; Nat. Mater. 2006, 5, 909-913; Adv. Catal. 2000, 45, 71-129) and assuming the existence of relations correlating the activation and free energies (see: J. Catal. 2002, 209, 275-278; J. Catal. 2001, 197, 229-231; Nat. Chem. 2009, 1, 37-46; J. Catal. 2004, 224, 206-217; Nat. Mater. 2006, 5, 909-913; Adv. Catal. 2000, 45, 71-129).  
+
+According to the Bronsted- Evans- Polanyi relationship (BEP), the more exergonic a reaction is, the lower the barrier will be, and the transition state energies should scale with the free energies (see: Chem. Rev. 1928, 5, 231-338; Trans. Faraday Soc. 1938, 34, 11-24; J. Catal. 2002, 209, 275-278). However, calculating the transition states would introduce errors due to the inherent difficulty of accounting for multiple reaction pathways and the solvation effects of leaving groups like Br, which to date are hard to capture thermodynamically in transition states. In light of the potential inaccuracies, we have chosen to avoid an extensive discussion on this aspect to preserve the scientific consistency and reliability of our DFT profile and energy levels.  
+
+### (8) Calculation and references for GHG emissions should be provided  
+
+We have included a more thorough explanation in the revised manuscript regarding the GHG emissions. We trust that this addition enhances the clarity of our work.  
+
+## Reviewer #4  
+
+In this manuscript, Gianvito Vile et al. present a study on C(sp3)- C(sp3) cross- coupling in a semiheterogeneous system using an nCNx photocatalyst. While the exploration of such catalysts and their application in cross- coupling reactions is undoubtedly significant, and the concept of utilizing nCNx as a heterogeneous photocatalyst is intriguing, the manuscript falls short in distinguishing this approach from prior research. For instance, in a previous study (Nat. Synth. 2023, 2, 1092- 1103), Gianvito Vile already employed a Ni single- atom catalyst for C- O coupling but did not adequately elucidate the rationale behind the occurrence of decarboxylation in this study. A more in- depth investigation into the decarboxylation mechanism is needed, particularly in light of earlier reports (Angew. Chem. Int. Ed. 2024, 63, e202405902; Adv. Synth. Catal. 2020, 362, 3898- 3904). Additionally, it is well- established that homogeneous nickel catalysts are prone to deactivation in dual photoredox/nickel- catalyzed processes, as highlighted in a previous study (Nat. Catal. 2020, 3, 611- 620). This manuscript does not sufficiently address this critical issue.  
+
+We sincerely appreciate the Reviewer's insightful comment and the recognition of the significance of our work on C(sp3)- C(sp3) cross- coupling using an nCNx photocatalyst. However, we understand the Reviewer's concern regarding the need to better differentiate this approach from prior research. To address this, the revised manuscript emphasizes the unique aspects of the semi- heterogeneous system, where the nCNx photocatalyst operates in a distinct manner compared to conventional homogeneous systems. The manuscript highlights the advantages of this dual- phase approach, including improved catalyst stability and recyclability, which were not explored in the context of cross- coupling reactions in prior studies. In response to the Reviewer's suggestion for a more detailed investigation of the decarboxylation mechanism, we have now included additional experiments and computational studies. These include Stern- Volmer quenching experiments, DFT calculations, and EPR measurements, which provide a clearer understanding of the reaction pathway and strengthen the explanation of the decarboxylation step.
+
+<--- Page Split --->
+
+Regarding the Reviewer's comment on the potential deactivation of homogeneous nickel catalysts in dual photoredox/nickel- catalyzed processes, we acknowledge the importance of this issue. The manuscript now explicitly addresses how the semi- heterogeneous system helps mitigate catalyst deactivation through the stable interaction between the photocatalyst and the nickel species, providing enhanced stability and overcoming the typical deactivation challenges observed in homogeneous systems.  
+
+We trust that these additions and clarifications will meet the Reviewer's expectations and significantly improve the manuscript.  
+
+The previous studies referenced on page 2 do not correspond accurately with the information presented in Figure 1. For instance, the study shown in Figure 1c is from 2016, not 2014 as stated in the text. Please ensure that all studies mentioned in the text are correctly aligned with those depicted in the figure, and address any typographical errors.  
+
+We extend our thanks to the Reviewer for noting these errors. We have corrected all typographical mistakes in the amended text.  
+
+The authors optimized several photocatalysts, including nCNx, gCNx, mpg- CNx, K- PHI, and recovered nCN. However, the manuscript lacks sufficient characterization data to substantiate the comparison and validation of nCNx against the other photocatalysts. Although nCNx was prepared through a thermal exfoliation process from gCNx, no data are provided to describe the morphology, thickness, or surface area of nCNx relative to gCNx. Therefore, it is essential to include comprehensive characterization data for all the photocatalysts in Figure 2, such as BET, XRD, AFM, and UV analyses. Additionally, TEM, SEM, XPS, and FT- IR data should be provided for each photocatalyst. Furthermore, the method used to calculate the reported C/N ratios ('0.61- 0.67') should be clearly explained.  
+
+We thank the Reviewer for their suggestion. In our revised manuscript, all the photocatalysts tested have been defined and characterized. We have also added missing details in the text.  
+
+In this study, the authors present an optimization of a photocatalyst, identifying nCNx as the optimal choice. However, several concerns arise regarding the rationale behind selecting nCNx over other catalysts. The authors suggest that mpg- CNx is less effective due to its higher solution viscosity, which they claim reduces the availability of active sites. This assertion, however, lacks sufficient experimental evidence or comprehensive characterization of the photocatalysts in question. While it is acknowledged that mpg- CNx may exhibit increased viscosity, this factor alone does not provide a strong basis for dismissing its catalytic potential. In fact, mpg- CNx typically has a larger surface area, which is generally associated with an increased number of active sites, as documented in the literature (Chem. Eur. J. 2015, 21, 526- 530; J. Phys. Chem. C 2012, 116, 19644- 19652; Angew. Chem. Int. Ed. 2015, 54, 12868- 12884). This contradicts the authors' claim that mpg- CNx is less suitable due to fewer active sites. The manuscript lacks detailed characterization data, such as BET surface measurements, pore size distribution, or SEM/TEM imaging, which are essential for substantiating the claim that mpg- CNx has a reduced number of active sites or is significantly impacted by viscosity- related issues. Moreover, the authors report comparable reaction yields between CNx (3a 41% and 4a 23%) and mpg- CNx (3a 34% and 4a 18%), indicating that mpg- CNx could be a viable alternative. The rejection of its efficacy based on unproven assumptions about solution viscosity and active site availability is unconvincing.  
+
+We sincerely appreciate the Reviewer's insightful suggestion and completely agree that these are important considerations. We acknowledge that mpgCNx typically exhibits a larger surface area, which, as the Reviewer correctly points out, is generally associated with an increased number of active sites. However, during the course of our experimental campaign, we noticed that mpgCNx formed a more viscous reaction environment and, moreover, it adhered to the walls, which potentially led to less efficient light irradiation. While we acknowledge that the reactivity of both compounds is comparable, we chose to proceed with optimization using nCNx due to its easier reaction. However, in the revised paper, we removed any claims that mpgCNx is less effective and we have revised the manuscript to present a more balanced discussion, emphasizing
+
+<--- Page Split --->
+
+that while nCNx was found to be optimal for our specific conditions, mpgCNx remains a viable candidate for future studies. We hope these revisions clarify our approach and address the Reviewer's concerns.  
+
+In addition to the concerns mentioned above, there is an inconsistency in the manuscript regarding the mention of "boron- doped mpg- CNx." This material is introduced without prior context or discussion in the manuscript, leading to confusion. The authors should ensure that all materials discussed are properly introduced and contextualized.  
+
+We thank to the Reviewer's for their valuable comment. We fully agree that the statement needed correction. In the revised manuscript, we have removed any mention of using a boron- doped catalyst.  
+
+The manuscript includes product yield data that have been confirmed using HPLC calibration. However, the authors have not provided the detailed evidence of the HPLC calibration curves that were used. To ensure scientific reproducibility, it is imperative that the authors provide comprehensive details regarding the calibration process. Furthermore, there are notable inconsistencies between the product yield values reported in the main text and those presented in the Supporting Information (SI). These inconsistencies call into question the credibility of the data and must be addressed. The authors should conduct a thorough review and make the necessary corrections to ensure consistency across the manuscript. Additionally, the SI lacks NMR spectra for several compounds discussed, despite the fact that these are typically required to confirm the synthesized products. The authors must provide the NMR data for all relevant compounds to support the claims made in the manuscript.  
+
+We thank the Reviewer's for their valuable suggestions. We concur with the Reviewer's comments. In the revised version of the manuscript we provided details of the HPLC calibration curve. Upon purifying the product 3a, several aliquots of 100 microliters were taken, resembling different reaction yields (0, 12,5%, 25%, 50%, 75% and 100%), The different absorbances were measured, obtaining a robust linear relationship between concentration and AUC ( \(r^2 = 0.997\) ).  
+
+Additionally, we agree that the NMR data was missing. In the revised manuscript we incorporated \(^1\) H- NMR, \(^{13}\) C- NMR and \(^{19}\) F- NMR (if needed) of every purified product.  
+
+This paper presents DFT calculation data to explain the Ni- catalyzed reaction pathway. However, the DFT data presented, which uses proline, does not offer a sufficiently reliable explanation for the mechanistic pathway. Based on the substrate scope study in this paper, the reactivity and selectivity are influenced by the substituents, with even methyl- substituted proline showing no reactivity. Therefore, it would be more appropriate to present DFT calculation results using Boc- protected proline rather than proline. Additionally, the Ni- catalytic cycle has already been extensively addressed in previous studies (Nature 2016, 536, 322- 325; J. Org. Chem. 2024, 89, 11136- 11147; Science 2014, 345, 437- 440). It would be beneficial to develop data that provides deeper mechanistic insights into decarboxylation beyond esterification.  
+
+We appreciate the Reviewer's comment on the DFT part. The suggested papers (Nature 2016, 536, 322- 325; J. Org. Chem. 2024, 89, 11136- 11147; Science 2014, 345, 437- 440) have been included in the manuscript.  
+
+We also agree with the Reviewer's point on the Boc- protected proline; we have thus addressed the point by calculating the full reaction profile in the presence of the Boc group. As shown in the amended paper, the reaction energy is equivalent with and without Boc, although more stable intermediates are formed when comparing Boc- protected and free proline. This result is not surprising. In fact, while the Boc group serves as a protective group in proline to prevent its amine functionality from unwanted reactions, influencing the local environment of the reactant and intermediate states, the fact that DFT calculations did not show any significant change in the reaction profile is due to the inherent nature of DFT as a local optimization method. DFT tends to capture the electronic structure accurately but may not fully account for long- range effects or subtle changes that could arise from the conformational flexibility or dynamic interactions of the Boc group in a real reaction context. Therefore, the overall reaction mechanism remains energetically equivalent in both cases, which is consistent with the predictions from DFT. As a result, all our mechanistic conclusions are
+
+<--- Page Split --->
+
+unchanged, and, for example, the Ni(II) intermediate remains still more stable than Ni(I), as shown below in Figure 2 where the energetic profile with proline (a) and the Boc-protected proline (b) are shown.  
+
+![](images/Figure_2.jpg)
+
+<center>Figure 2. Energetic profile for the C-C coupling by considering the (a) proline molecule and the (b) Boc-protected proline. </center>  
+
+The authors present GHG emission data derived from the equations described in the 'Energy Calculations' section of the Supplementary Information (SI) to support the sustainability of carbon nitride compared to Ir- based complexes. However, the mentioned 'Energy Calculation' section is absent from the SI. To substantiate the sustainability advantages of carbon nitride, it is recommended that the authors provide a detailed calculation process along with reliable references.  
+
+Thank you and we sincerely apologize that this section was missing in the original submission. In response, we have added a detailed explanation on how the \(\mathrm{CO_2}\) emissions in Figure 5 were calculated. We hope this addition aligns well with the Reviewer's expectations and provides further clarity.  
+
+## Reviewer #5  
+
+The authors described the use of graphitic carbon nitride as a photocatalytic system to drive cross- coupling between alkyl halides and carboxylic acids. Mechanistic studies are also reported. In my view, the manuscript has been well written and the results are sound. Thus, I recommend its publication after having addressed the points below.  
+
+We thank the Reviewer for recognizing the clarity and robustness of our results. We will carefully address the points they raised to ensure the manuscript meets the highest standards for publication.  
+
+1. Additional references regarding the use of carbon nitride-based photocatalytic systems for the functionalization of organic compounds should be included in the introductory section, for instance: Science 365, 360-366 (2019), Angew. Chem. Int. Ed., 2023, e202313540, ACS Catal. 2023, 13, 13414-13422, Sci. Adv., 2020, 6, eabc9923, ACS Catal. 2024, 14, 11308-11317, ACS Nano, 2021, 15, 3621-3630, Chem. Sci., 2022, 13, 9927, Nature Catalysis, 3, 611-620 (2020), Adv. Sci. 2023, 10, 2303781, among others.  
+
+We thank the Reviewer for their helpful suggestion regarding the inclusion of additional references on carbon nitride- based photocatalytic systems. We agree that these references would provide a more balanced and comprehensive view of the field. We have thus cited these works in the introductory section of the manuscript.
+
+<--- Page Split --->
+
+2. The reaction scope should be expanded. Is it possible to use fatty acids, alkyl chlorides and alkyl triflates as starting materials? Moreover, the generality of the photocatalytic system with respect to more synthetically useful organic substrates, namely natural products or active drugs, should be addressed. For instance, is it possible to use other natural products or bio-active molecules containing carboxylic moieties as substrates?  
+
+We thank the Reviewer for this valuable suggestion. In line with their recommendation, we have expanded the reaction scope demonstrating the functionalization of bioactive compounds, including lysine and levodopa precursors. These new results validate the versatility and applicability of our photocatalytic system to a wide range of synthetically relevant substrates, supporting its potential for broader utility in organic synthesis.  
+
+## 3. Is it possible to perform the model reaction in a gram scale?  
+
+We thank the Reviewer for this insightful question. Scalability can be approached through either sizing up (increasing the volume of a single reaction vessel) or numbering up (using multiple reaction vessels in parallel). However, sizing up is limited in photocatalytic systems due to the constraints imposed by the Lambert- Beer Law, which describes the exponential attenuation of light as it penetrates through an absorbing medium. According to this, light intensity decreases with path length and concentration of the photocatalyst, resulting in uneven photon distribution and suboptimal reaction rates in larger reaction volumes. Consequently, as vessel size increases, light penetration becomes insufficient to drive the reaction uniformly, leading to decreased efficiency and yield.  
+
+To overcome this limitation and address the Reviewer's concern, we adopted the numbering up strategy, where multiple smaller reaction vessels are used in parallel, each receiving adequate and consistent light exposure to maintain reaction efficiency. This approach aligns well with the guidelines developed by several scientists who have pioneered scale up/scale down approaches for process intensification (Chem. Eng. Process. 2007, 46, 781- 789; Ind. Eng. Chem. Res. 2019, 58, 5349- 5357; Chem. Eng. Sci. X 2021, 10, 100097). Applying this approach, we successfully conducted the model reaction in a single pot and simultaneously with 15 parallel reactions. This enabled us to scale up the reaction from 0.2mmol of reagent (71% isolated yield of desired product 3a) to 3.0 mmol of reagent (77% isolated yield of desired product 3a). This demonstrates the practicality and effectiveness of numbering up for scaling photocatalytic batch reactions, allowing for higher throughput without compromising yield.  
+
+4. I suggest to include a general procedure for the photocatalytic experiments along with a picture of the reaction set-up and its description within the SI.  
+
+We have included a picture of the photoreactor in the Supporting Information.  
+
+5. The authors should better characterize the photocatalyst after a series of catalytic cycles.  
+
+We thank the Reviewer for their suggestion. In response, we have included a recyclability test for the nCNx photocatalyst in the revised manuscript. Specifically, we conducted five consecutive reactions, consistently obtaining the alkylated product with yields consistently between 77% to 81%. After each reaction, the nCNx catalyst was recovered by centrifugation, washed with acetonitrile and water, and dried at 65°C overnight. We monitored the recovery rate of the catalyst, obtaining quantitative (95-99%) yields in each instance. We carried out post-catalysis characterizations (i.e., N2 physisorption, XRD, CHN elemental analysis, and ICP-OES) to prove the structural robustness of the catalyst and the absence of any potential Ni doping of the carbon nitride carrier (ICP).  
+
+6. I suggest an improvement on the structure of the manuscript due to the presence of typos (e.g., structure compound 1 within table 1.  
+
+We have corrected all the typographical errors. We extend our thanks to the Reviewer for their careful reading.
+
+<--- Page Split --->
+
+## Point-by-point response to the Reviewers' comments  
+
+(original comments in blue, replies in black, actions in bold)  
+
+## Reviewer #1  
+
+I commend the authors for the quality and seriousness of their revisions. The manuscript is much improved and I felt like the author responded professionally to all the comments. I have a very few minor comments:  
+
+We appreciate the Reviewer's positive feedback. We are also pleased that our responses were perceived as professional. The Reviewer's comments have been instrumental in refining the manuscript, and we have now addressed the remaining minor points herein.  
+
+In the novel figure 4, panel B, the "micro" symbol does not display properly, and only "u" is readable.  
+
+We thank the Reviewer for pointing out the typo. We have corrected the symbol to ensure that \(\mu\) is clearly readable.  
+
+In Figure 4, panel C, the authors mislabeled the Y- axis, and I should be I0/I and not I/10. In addition, compound 2, 1 and \(1+\) Nicat \(^+\) dMeObpy (probably mislabeled as well), seem to have a negative slope that the author did not comment on. A negative slope would imply that the photoluminescence intensity increases as the quencher is added.  
+
+We are grateful for the Reviewer's insight. We have adjusted the label and the slope accordingly.  
+
+Still in figure 4, I would recommend forcing the intercept of all lines at (0;1), as this is a "true" datapoint according to the Stern- Volmer equation, \(10 / 1 = 1 + \text{Ksv} [\text{Q}]\) , so when \([\text{Q}] = 0\) , \(10 / 1\) should be 1.  
+
+Thank you. We have corrected Figure 4 in the revised manuscript.  
+
+## Reviewer #2  
+
+I read through the revised draft and the response to comments carefully, and find that the issues were basically addressed, and now can be acceptable.  
+
+We thank the Reviewer for their positive feedback and are pleased that the revised version of the manuscript is considered suitable for publication in Nature Communications.  
+
+## Reviewer #3  
+
+In the revised manuscript, the authors provided more insights into the C- C coupling (other than the previously reported C- O) by systematically excited state quenching, photon energy/base/solvent dependence and calculation. Other issues are also addressed.  
+
+We appreciate the Reviewer's recognition of our efforts to provide deeper insights into C- C coupling, and we are grateful for the constructive feedback received.  
+
+## Reviewer #4  
+
+In the revised manuscript, Gianvito Vile and colleagues have addressed some of the issues raised in previous comments. While the updated version demonstrates improvements, several critical concerns remain, rendering the manuscript unsuitable for publication in Nature Communications. The manuscript still lacks a thorough characterization of the materials used, which is essential for validating the reported material properties. Moreover, the choice of a semi- heterogeneous system requires a stronger justification to establish the significance of the findings. Additional concerns stem from experimental aspects, particularly the
+
+<--- Page Split --->
+
+"numbering- up" scale- up process, which raises doubts about the reliability and scalability of the results. Consequently, despite the authors' notable efforts, I regret to recommend against publication in Nature Communications. Detailed comments are provided below:  
+
+We sincerely appreciate the Reviewer's time and effort in highlighting areas for improvement. However, we respectfully disagree with the Reviewer's assessment and we are convinced that this study is well- suited for Nature Communications. Nonetheless, we remain committed to improving the quality and clarity of our work and have made further changes to refine our approach and ensure a more rigorous presentation of our findings.  
+
+Comment #1: The manuscript does not sufficiently highlight the advantages of the semi- heterogeneous system through a detailed comparison with other catalytic systems, such as homogeneous and single- atom heterogeneous systems. While the authors mention that the semi- heterogeneous system mitigates catalyst deactivation via stable interactions between the photocatalyst and nickel species, the manuscript lacks a clear and rational explanation to support this claim.  
+
+While we appreciate the Reviewer's feedback, we respectfully disagree with them. The advantages of the semi- heterogeneous system have been clearly demonstrated through recycling tests, economic comparisons, and an in- depth analysis of GHG emissions. To address this point once again, we have now performed a direct comparison of reaction outcomes between the reported Ir complex homogeneous photocatalyst, our conditions, and a fully heterogeneous \(\mathrm{Ni_{1}@nCN_{x}}\) system. This \(\mathrm{Ni_{1}@nCN_{x}}\) single- atom catalyst was recently synthesized and characterized by our group (see Nat. Synth. 2023, 2, 1092- 1103). The results highlight the advantages of the semi- heterogeneous approach we have developed.  
+
+![](images/Figure_1.jpg)
+  
+
+Specifically, the table shows that, under optimized conditions, the semi- heterogenous system surpasses not only the homogeneous catalysts, but also the heterogeneous single- atom catalysis. We attribute this to the stringent environmental requirements of Ni for catalytic activity, which are hindered by the absence of suitable ligands necessary for oxidative addition over alkyl bromides. Furthermore, since Ni in the single- atom catalyst is not present in the same phase as the substrate, the interactions between the catalyst and substrates are significantly reduced. However, we want to point out that we cannot exclude that, through catalysis engineering, a fully heterogeneous system may be developed in the near future. Upon discussing this with our co- authors, we decided to exclude the single- atom catalyst results from the manuscript, as they fall outside the scope of our study.  
+
+Furthermore, we want to remark that, in our studies, we propose a synergistic catalytic mechanism involving two distinct catalytic cycles operating simultaneously, and we never state that the stable interaction between \(\mathrm{nCN_{x}}\) and nickel mitigates catalyst deactivation.  
+
+Comment #2: Although the authors stated that they included additional characterizations, critical analyses such as TEM, SEM, XPS, and AFM remain missing. The explanation of catalyst screening results is also insufficient to justify the selection of nCNx as the optimal material. Characterization data for the recovered nCNx are particularly lacking. For instance, TEM images and EDS mapping are necessary to determine
+
+<--- Page Split --->
+
+whether Ni aggregates have formed. Despite claims that such data were included, they are absent from both the manuscript and the supplementary information.  
+
+We thank the Reviewer for the feedback. In the revised manuscript, we have further characterized the material, including, SEM, XPS and TEM for the fresh and recycled \(\mathsf{nCN_x}\) (Figure S3- 5). The results are in line with the literature, showing that the structural and electronic properties of the material remain consistent with previously reported findings. Given that carbon nitride is a well- studied material with extensive characterization available in the literature (see ACS Sustainable Chem. Eng. 2023, 11, 5284- 5292; ACS Appl. Nano Mater. 2022, 5, 14520- 14528; J. Colloid. Interface Sci. 2024, 673, 943- 957; ACS Catal. 2021, 11, 1593- 1603; ACS Omega 2019, 4, 12544- 12554), we do not consider it necessary to include redundant analyses such as AFM, which would not provide significantly novel insights.  
+
+Moreover, AFM microscopes are currently not available at our institution, and conducting these analyses would require involving additional authors from outside our institution who, in our view, do not meet the criteria for authorship, as the analyses are not fundamental to the study.  
+
+In terms of stability, we have provided sufficient evidence demonstrating the robustness of the material, as shown in Figure S3 and Figure S5. After five catalytic cycles, the presence of Ni adhered to \(\mathsf{nCN_x}\) photocatalyst was consistent with the findings of König and Noël (see ACS Catal. 2021, 11, 3, 1593- 1603; Angew. Chem. Int. Ed. 2024, 63, e202405902). The recycled photocatalyst was found to be inactive without the addition of the Ni homogeneous catalyst (Table S9, entry 6), and washing away the adhered Ni traces from \(\mathsf{nCN_x}\) with 1 M HCl did not impact its intrinsic photocatalytic activity (Table S9, entry 7). In this context, we are convinced that no further characterization is considered necessary, as it can be found in Angew. Chem. Int. Ed. 2024, 63, e202405902.  
+
+Comment #3: For gram- scale synthesis, the authors employed a numbering- up strategy instead of scaling up the system size. This approach is unconvincing for demonstrating the scalability of heterogeneous semiconductors. Moreover, the numbering- up experiments were conducted on an even smaller scale (0.2 mmol) than the optimized conditions (0.3 mmol). As a result, the manuscript does not sufficiently support the claim that \(\mathsf{nCN_x}\) is a sustainable, efficient, and cost- effective alternative to traditional iridium- based photocatalysts, especially from an industrial perspective.  
+
+We encourage the Reviewer to carefully revisit the manuscript. The reaction scale for optimization, scope, and scale- up studies consistently employed 0.2 mmol of alkyl bromide (2), and 0.3 mmol of carboxylic acid (1), and this can be easily verified for the scale- up experiment as well. In fact, the scale- up was performed using 15 vials and a total of 3 mmol of alkyl bromide (our limiting reagent); therefore, each vial contained 0.2 mmol of alkyl bromide (3 mmol / 15 vials = 0.2 mmol).  
+
+Besides, as the Reviewer knows, one major limitation of photocatalysis is the challenge of scalability due to light absorption constraints. This issue arises from the attenuation effect of photon transport, as described by the Lambert- Beer law, which prevents the conventional scale- up of batch reactions by simply increasing reactor dimensions. In large reactors, the distribution of heterogeneous catalyst can be inefficient, leading to underutilization in certain regions. A numbering- up strategy addresses these challenges by optimizing the surface area and ensuring a more effective contact between the light and the photocatalyst. This approach enhances photon penetration, improving overall reaction efficiency while reducing energy costs.  
+
+We also respectfully disagree with the Reviewer's final statement. Iridium- based homogeneous photocatalysts often present challenges for large- scale applications due to their high cost, limited availability, and difficulties in recyclability. Our study demonstrates that solid \(\mathsf{nCN_x}\) exhibits catalytic activity comparable to its homogeneous counterpart while providing key advantages, including lower cost, greater abundance, and enhanced recyclability. These characteristics make \(\mathsf{nCN_x}\) a promising alternative for large scale decarboxylative cross- coupling reactions.  
+
+Comment #4: The authors propose that energy differences resulting from interactions between 1 and the base explain variations in product yields. This rationale is plausible in cases without a base (1.48 eV) or with DBU (2.31 eV). However, the manuscript fails to address the significant discrepancies between Na2CO3 and K2CO3 results. Despite their similar intermediate energies (1.24 eV for Na2CO3 and 1.25 eV for K2CO3),
+
+<--- Page Split --->
+
+their yields differ substantially, with Na2CO3 yielding \(33\%\) of 3a and \(17\%\) of 4a, while K2CO3 yields \(34\%\) and \(44\%\) , respectively. Additional justification or experimental data is needed to clarify these differences.  
+
+We thank the Referee for his comment. The interaction of sodium and potassium cations from sodium and potassium carbonates with deprotonated Boc- L- proline leads to the release of radical Boc- proline and \(\mathrm{CO_2}\) with similar energy barriers for each cation: \(1.24 \mathrm{eV}\) for \(\mathrm{Na_2CO_3}\) and \(1.25 \mathrm{eV}\) for \(\mathrm{K_2CO_3}\) . However, the sodium cation forms an intermediate that has a stronger and more localized basic environment compared to that of the potassium- based intermediate. This effect is primarily attributed to the smaller size of the sodium cation and its electronic interactions with the deprotonated Boc- L- proline. The sodium cation enhances the stabilization of the intermediate by forming two bonds with oxygen atoms, each measuring approximately \(2.21 \mathrm{\AA}\) in length (Figure 1a). This strong interaction favors the selective formation of product 3a over byproduct 4a. In contrast, the potassium cation forms an intermediate with three bonds to oxygen atoms: two at \(2.59 \mathrm{\AA}\) and one at \(2.70 \mathrm{\AA}\) (Figure 1b). These weaker interactions do not provide the same stabilizing effect, resulting in the formation of both product 3a and byproduct 4a at comparable rates. This discussion has now been included in the revised manuscript.  
+
+![PLACEHOLDER_19_0]
+
+<center>Figure 1. Effect of the decarboxylation process in the presence of sodium (a) and potassium (b) cations from inorganic bases. </center>  
+
+Comment #5: Previously, it was suggested to provide detailed descriptions of the "Energy Calculations" for GHG emission data. However, the revised manuscript still lacks a clear explanation of the calculation procedures and data references. To emphasize the environmental advantages of the semi- heterogeneous system over homogeneous systems, a detailed explanation of how the GHG emission data were derived is essential.  
+
+We understand that LCA assessment is not a common practice for organic and inorganic chemists, but we want to assure the Reviewer that all necessary details are provided in the Supporting Information. The study followed a cradle- to- gate approach, meaning that it considers emissions from raw material extraction, acquisition, and production, while excluding usage and disposal stages. The functional unit was defined as \(1 \mathrm{kg}\) of the target compound. The life cycle inventory was performed collecting all relevant data on material and energy inputs, emissions, and waste flows associated with the catalytic reaction, and calculating the emissions accordingly. To do this, the reaction was inserted LCA software, allowing for the quantification of energy requirements and mass balances with the experimental information required. The background life cycle inventories for raw materials and energy sources were sourced in Ecoinvent 3.9.1 cut- off with Carbon Minds databases. The life cycle impact assessment phase was conducted translating raw emissions data into meaningful environmental indicators. In this study, the Global Warming Potential metric was used, which measures greenhouse gas emissions in \(\mathrm{kg} \mathrm{CO_2}\) equivalent over a 100- year time horizon, following the methodology outlined in the 6th IPCC assessment report. These steps are common practice in the chemical engineering field, and our group has already published similar LCA studies that have received recognition from our peers (see ACS Sustainable Chem. Eng. 2025, 13, 7, 2864- 2874; Cell Rep. 2025, 2, 1, 100286).
+
+<--- Page Split --->
+
+Thus, we are confident that the LCA calculations are robust and sufficiently detailed to highlight the environmental advantages of the semi- heterogeneous system over homogeneous alternatives.  
+
+Comment #6: Several inconsistencies in the manuscript undermine its reliability. These include: (1) Comparison of substrate yield with a prior study (MacMillan, 2016), despite differing experimental conditions (Ir system: room temperature, \(48h\) ), which makes the comparison invalid. (2) Unexplained changes in nCNx surface area from 12 to \(23m^2 /g\) , raising questions about synthesis reproducibility. (3) Conflicting details about the washing procedure for the recovered catalyst. The supporting information states the catalyst was washed with ethyl acetate and water, while the manuscript and prior responses mention acetonitrile.  
+
+We appreciate the Reviewer's perspective and insights. The comparison was included at the request of the Editor and another Reviewer. Comparing substrate yield is a valid approach, as demonstrated by numerous examples in the literature where reactions under different conditions have been compared (e.g., see Chem. Sci. 2019, 10, 5837- 5842, where mechanochemistry vs. solution- based reactions have been compared).  
+
+Regarding the surface area of nCNx, we apologize for the oversight, and we confirm the value of \(23m^2 /g\) . To further substantiate this, we prepared three additional batches, consistently obtaining values between 21 and \(24m^2 /g\) . Thus, we can now determine that, by using our synthetic recipe, the reader will obtain a surface area of \(22.5\pm 1.5m^2 /g\) . The value previously- included in the very first version of this manuscript was an error, as we erroneously utilized the surface area of a gCN, batch synthesized with a heating ramp of \(5^{\circ}C\) min \(^{- 1}\) . We apologize for the oversight. We also thank the Reviewer for the (correct) observation regarding the solvent choice for washing the photocatalyst. Both acetonitrile (MeCN) and ethyl acetate (EtOAc) are indeed suitable solvents for washing the nCNx photocatalyst, as they both effectively remove residual organic reagents and byproducts. However, we opted for EtOAc in our studies as it is considered a greener solvent and has broader industrial acceptance compared to MeCN (see Green Chem. 2016, 18, 3879- 3890). We have corrected the typographical error in the revised manuscript.  
+
+Comment #7: Several minor typos and formatting errors persist in the manuscript. For instance, the author's name is misspelled as "MCMillan" instead of "MacMillan" on page 3. Additionally, discrepancies between figure and data references within the manuscript need to be corrected.  
+
+We thank the Reviewer for pointing out these typos. Additionally, we have performed thorough proofreading to ensure consistency and accuracy throughout the text, and corrected all remaining errors.  
+
+## Reviewer #5  
+
+The revised version of the Manuscript has been well organized and the results are very interesting. I personally found this version of the manuscript more detailed and clear. In addition, the authors addressed most of the reviewers' comments satisfactory. Thus, I think that the new version of the manuscript is now suitable for publication in Nature Communications.  
+
+We thank the Reviewer for their encouraging feedback and for recognizing the improvements in the clarity and organization of the manuscript.
+
+<--- Page Split --->
+
+## Point-by-point response to the Reviewers' comments  
+
+(original comments in blue, replies in black, actions in bold)  
+
+## Reviewer #1  
+
+I was personally satisfied by the changes carried out last time by the authors. I have however to stress that I agree with reviewer 4, that a "numbering- up" approach is, in my opinion, not a proper scale up, but rather represents a reproducibility experiment from which average yields and standard errors can be obtained. I understand the rational from the authors about light penetration, Beer- Lambert law etc, but a true scale up also implies accounting for these issues and showing applicability and practicability for industrial developments. Light penetration is often industrially tackled by multiplying irradiation sources, increasing fluence, inserting irradiation systems (light tubes) inside the reaction reactor etc. I am not sure if a scale up is desperately needed, but I feel like the author should acknowledge the limitation of the "numbering- up" approach that they propose.  
+
+We thank the Reviewer for their valuable insights and for recognizing the improvements made in our previous revision. We also appreciate the remarks regarding the distinction between "numbering- up" and a scalable sizing- up strategy.  
+
+We agree that numbering- up does not address all challenges typically associated with industrial- scale photochemical reactions, particularly those related to photon flux, light penetration, and reactor engineering. Our approach was aimed at demonstrating reproducibility and robustness across multiple microreactors, as a preliminary validation step toward process intensification. However, we have now clarified that our numbering- up strategy should not be interpreted as a comprehensive scale- up, but rather as a practical, modular approach to address throughput enhancement while maintaining reaction performance. We have also included a short discussion of potential scale- up routes, including the integration of internal irradiation systems, multi- source light fields, or intensified flow- through designs, as commonly pursued in industrial photochemistry.  
+
+## Reviewer #4  
+
+Some of the previous comments have been addressed, and the revised manuscript shows improvement. However, several important issues remain insufficiently explained. I recommend the authors provide further clarification based on the following comments. I will reconsider the publication of the manuscript once these concerns are fully addressed.  
+
+We thank the Reviewer for their engagement with our work and for recognizing the improvements made in the revised version. In response, we have made additional revisions to further refine our manuscript and ensure a more rigorous and transparent presentation of our work.  
+
+Comment #1: It remains unclear whether byproduct 4a was formed when the \(\mathrm{Ni_1@nCN_x}\) catalyst was used. According to the authors' own cited reference (Nat. Synth. 2023, 2, 1092- 1103), \(\mathrm{Ni_1@nCN_x}\) is expected to favor C- O coupling, which in this case would yield byproduct 4a. The authors must clearly state this and provide a mechanistic rationale for the observed difference in coupling selectivity (C- C vs. C- O) between the current semi- heterogeneous system and the previously reported heterogeneous system. Without such clarification, the mechanistic underpinnings of the semi- heterogeneous system - the core focus of the manuscript - remain ambiguous and must be addressed in the revised manuscript.  
+
+While we appreciate the Reviewer's feedback, we respectfully believe that the rationale for the selectivity switch has been clearly demonstrated. The central goal and challenge of this project was to shift the selectivity toward C- C bond formation. This observed difference in selectivity is attributed to a key mechanistic distinction in the decarboxylation step between our system and previously reported conditions. Under our optimized conditions, the decarboxylation step is significantly favored, enabling C- C bond formation. This conclusion is supported by both experimental and computational evidence. Stern- Volmer experiments revealed a markedly stronger interaction between the photocatalyst (nCNx) and the deprotonated substrate 1 in the presence of \(\mathrm{Na_2CO_3}\) compared to \(\mathrm{Cs_2CO_3}\) . This enhanced interaction
+
+<--- Page Split --->
+
+facilitates the single electron transfer (SET) step that results in the decarboxylation. Furthermore, our computational analysis indicates that \(\mathrm{Na_2CO_3}\) promotes decarboxylation more effectively due to a stronger electron- withdrawing interaction between the \(\mathrm{Na^{+}}\) counterion and the carboxylate group.  
+
+To further support our hypothesis, we applied the conditions reported in Nat. Synth. 2023, 2, 1092- 1103, using DMF as solvent, \(\mathrm{Cs_2CO_3}\) or \(\mathrm{K_2CO_3}\) as base, and white light irradiation to our semi- heterogeneous system. Under these conditions, we observed exclusive formation of the C- O bond product, with no C- C bond formation. This result is consistent with the findings of MacMillan and co- workers, where also was reported a preference for C- O coupling using DMF as solvent and \(\mathrm{Cs_2CO_3}\) as base (see Supporting Information of Nature 2016, 536, 322- 325).  
+
+In conclusion, the selectivity for C- C vs. C- O bond formation is closely tied to the system's ability to promote the decarboxylation step. In this study, we believe that we have provided both qualitative and quantitative evidence to support the origin of the observed selectivity and to clarify the mechanistic divergence between the two systems. In the revised version, we have added a few additional sentences to further emphasize this point and to highlight the key factors governing the divergent reactivity.  
+
+Comment #2: While I acknowledge the practical limitations associated with photochemical scale- up, I remain unconvinced that the use of a numbering- up strategy in a batch system sufficiently demonstrates the scalability of the catalytic process. As discussed in reference 43 (Chem. Rev. 2022, 122, 2752- 2906), numbering- up is a strategy more appropriately applied to continuous flow systems. In contrast, the current study employs a batch process, where this approach is less relevant. To substantiate claims of scalability, the authors should either demonstrate a genuine scale- up within a batch system or apply a numbering- up strategy in a flow system.  
+
+We thank the Reviewer for this thoughtful and important comment regarding scalability. We fully acknowledge that numbering- up is most commonly and effectively applied in continuous flow systems, as noted in reference 43 (Chem. Rev. 2022, 122, 2752- 2906). However, in the context of our study, the numbering- up strategy was used as a proof- of- concept to emphasize the critical role of maximizing light- exposed surface area, particularly relevant in heterogeneous systems, where light penetration is further hindered by solid components.  
+
+In response to the Reviewer's suggestion, we have now included a scale- up experiment using a sizing- up approach within the batch system. The reaction was scaled up by a factor of 10, from 0.2 mmol to 2 mmol, in a single round- bottom flask illuminated with four Kessil lamps (427 nm), yielding product 3a in 63% isolated yield. This yield is lower than the 77% yield obtained using the numbering- up method. We attribute this 14% drop primarily to differences in the geometry of the reaction vessels, which affected the effective irradiated surface area: in the numbering- up setup, small vials provided an irradiated surface area of approximately \(2.16 \mathrm{cm^2 mL^{- 1}}\) , whereas the round- bottom flask used in the scale- up experiment had a slightly lower irradiated surface area of about \(1.57 \mathrm{cm^2 mL^{- 1}}\) . These results, now detailed in the revised manuscript, help clarify the relationship between reactor geometry, light exposure, and reaction efficiency in batch photochemistry, and demonstrate a ten- fold scale up is feasible with only a modest loss in yield.  
+
+Comment #3: The issue regarding yield comparison remains unresolved. The standard conditions in the present study require 72 hours at elevated temperature, whereas MacMillan's protocol achieves C- C coupling in 48 hours at room temperature. Under these markedly different conditions, similar yields (e.g., 71% in this study vs. 85% in MacMillan's work for compound 3a) do not constitute a valid comparison. Furthermore, the article cited by the authors (Chem. Sci. 2019, 10, 5837- 5842) compares mechanochemical and solution- phase conditions within a single study, carefully controlling all variables aside from the activation method. In contrast, the current manuscript compares yields from two independent reports without such control, which invalidates the justification provided. To enable a sound comparison, the authors must present data obtained under matched experimental conditions. Without these corrections, I respectfully maintain that the justification for the manuscript remains unconvincing.  
+
+Our intention was not to claim strict equivalence between the two protocols, but rather to highlight that, despite the differing conditions, our method delivers synthetically useful yields using a more accessible setup.
+
+<--- Page Split --->
+
+That said, we acknowledge the limitations of comparing results from independent studies and have revised the text to clarify this point, explicitly stating that the comparison is qualitative and does not account for all variables. A direct comparison under identical conditions is, in fact, not feasible due to fundamental differences between the two catalytic systems: MacMillan's protocol employs a homogeneous catalyst, whereas our approach uses a heterogeneous system. These differ not only in terms of active site location and availability, but also in mass and heat transfer properties, dispersion, and reaction dynamics. Attempting to apply identical conditions to both systems would not result in a meaningful comparison, as each catalyst requires a distinct optimization strategy tailored to its physicochemical characteristics.  
+
+In response to the Reviewer's suggestion, we have already a control experiment in which the nCNx photocatalyst is replaced with a homogeneous Ir- based photocatalyst (Ir[dF(CF3)ppy]2(dtbbpy)PF6) under the otherwise identical conditions we have developed. This substitution resulted in a \(61\%\) yield of product 3a with MacMillan's catalyst in Nature 2016, 536, 322- 325, supporting the relevance and effectiveness of our system (Table 1, Entry 7).
+
+<--- Page Split --->
+
+## Point-by-point response to the Reviewers' comments  
+
+(original comments in blue, replies in black, actions in bold)  
+
+## Reviewer #4  
+
+I have reviewed the revised manuscript and the responses to the reviewers' comments. The major issues have been adequately addressed, and the manuscript is now suitable for acceptance.  
+
+We thank the Reviewer for the positive feedback.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1fd77ffe2a84b4cd5aa0c17b1d5a10ad75b7f3e14cf2668ed265017f6fda9b10/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1fd77ffe2a84b4cd5aa0c17b1d5a10ad75b7f3e14cf2668ed265017f6fda9b10/supplementary_0_Peer Review File__1fd77ffe2a84b4cd5aa0c17b1d5a10ad75b7f3e14cf2668ed265017f6fda9b10.mmd

@@ -0,0 +1,284 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+Impacts of shared mobility on vehicle lifetimes and the carbon footprint of electric vehicles  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1: Remarks to the Author: Dear authors,  
+
+This is a very interesting study investigating the relationship between vehicle use intensity, lifetime, the implications of shared mobility, and the carbon footprint. I think there is potential for this piece of study to be accepted for publication. But intensive revisions are needed.  
+
+- Generally, this study seems very like a combination of three pieces of small studies, which are (1) the relationship between vehicle use intensity, lifetime and lifetime driving distance, which is obtained by using conventional ICEV as the example. (2) a carbon footprint calculation for BEV with the previous relationships in (1) incorporated. (3) a calculation on the empty travel that can be tolerated. I would say these three parts of the study are not very closely integrated with each other. It is more like a loose combination of researches in three different research areas, and trying to find some stories behind the combination. I recommend that the authors might put far more focus on the first part of the study, because currently there lacks understanding in the relationship between driving intensity and vehicle lifetime. And the first part could be not necessarily connected to the BEV carbon footprint and shared AV empty mileage contexts.  
+
+- Another problem when the authors try to incorporate the relationship found in conventional vehicles to shared AVs is that as people are expecting longer driving range for AVs, so the lifetime design for AVs could be very different from conventional vehicles. In that circumstance, I am afraid that the analysis could incorporate significant bias.  
+
+- Figure 1, I would recommend the authors to expand the explanations on this figure. More information should be provided, for example, the relationship between lifetime and lifetime driving distance; driving distance and lifetime driving distance.  
+
+- To use the relationship obtained on ICEVs on BEVs could incorporate biases, so that I would recommend the authors to conducted some sensitivity analysis.  
+
+- If the authors have already determine the value of e to be -0.59, does it make a lot of senses to analyze the cases of e equaling to 0 and 1?  
+
+- the empty travel section, all the discussions are very hard to follow due to the complication to the analysis. I recommend the authors to simply the discussion and highlight the most important findings in an easy-to-follow way.  
+
+- Line 241, yes higher driving intensity leads to shorter vehicle lifetime, which makes the use-phase GHG emissions because more emissions occur in the near future than in the long future. However, this analysis is only from one-vehicle perspective. But when considering that after the retirement of this vehicle, the next vehicle will get the benefit of lower emissions of electricity in the long future, than the near-future loss could be filled.  
+
+- line 243, the analysis on the manufacturing-phase GHG emissions actually need also to incorporate the impact from electricity emission factor, as much of energy consumption for battery production is electricity. This will lead to some differences in the manufacturing phase emissions, like the impact from near-term emission factor and long-term emissions factor analyzed in the use-phase emissions.  
+
+- line 257, the authors should not only identify the research weakness, actually this can be addressed by a sensitivity analysis, which I recommend that the authors should add.
+
+<--- Page Split --->
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+Summary of contribution:  
+
+This paper contributes meaningfully to the literature on vehicle lifetime carbon footprints by using retirement statistics to model relationships between driving intensity and lifetime, and by using those relationships to characterize whether high driving intensities tend to improve or harm carbon footprints. Sensitivity to an appropriate range of models is considered. The value of these findings largely comes from its implications for "future mobility", e.g., ridesourcing, car sharing, and autonomous vehicles. The dataset used is generally appropriate to investigate those implications and provides a much- needed look at intensity- lifetime relationships.  
+
+That said, it isn't clear to me whether the dataset can yield findings that cleanly extrapolate to those "future mobility" options - - this may be something that simply warrants an additional caveat in the discussion section but seems important to acknowledge. It also isn't clear to me whether there are specific insights that could translate these vehicle- level findings to fleet- level findings (are fleets of very few high- use cars better or worse than fleets of very many low- use cars?) - - maybe nothing can be said on that front, but maybe there are at least some specific directions the authors can suggest as future work based on their model findings.  
+
+Comments on data and methods:  
+
+1. The study depends on using historical vehicle retirements to infer trends for future (shared) mobility. It seems like one missing piece of the discussion section is whether this subsetted excerpt of all passenger cars may or may not cleanly extrapolate to uses such as car sharing and ride sourcing. Should we expect that the vehicle failure causes and reasons for retirement will be pretty much the same for shared mobility fleets as for all private vehicles? Or is there anything about their driving cycles (mostly long-duration urban driving shifts), ongoing maintenance regimes, or scrappage decisions (made by professional fleet managers instead of individual owners) that might make the data not extrapolate as cleanly such that results are biased in a direction we can characterize? This may be worth some brief comment.  
+
+2. I am glad to see multiple statistical model types tested for the semi-empirical model. The specific tradeoffs between model types (Table 1) are outside of my expertise, but the explanations provided were clear and concise. Including elasticities of 0 and -1 in the main figures helped me understand how the model is working and also serves as a useful bound.  
+
+3. This analysis is conducted and presented on a per-vehicle basis. I might hesitate to draw fleetwide conclusions from this analysis, and I think it should be mentioned in the problem framing or in discussion that the interactions between average vehicle use level and fleetwide impacts are outside of the paper's scope. For example, if demand is fixed and served 100% via AVs and ridesourcing, each unit increase in vehicle driving intensity seems to imply fewer total cars are needed -- but I don't think this analysis can tell us whether the net carbon impacts of that increase in intensity would be positive or negative.  
+
+Additional comments:  
+
+1. Please comment on the choice to use average electricity grid emissions factors instead of marginal and how this choice may affect your results and findings. In particular, as renewables increase, average electricity emissions factors will fall, but marginal emissions factors may or may not change (in nearly all regions the marginal generator is some form of fossil fuel generation). Might using marginal factors alter any findings? If so, it is worth some justification in the text of this choice. 
+2. It seems that AVs, ridesourcing, and/or car sharing may cause not only empty vehicle travel, but also additional induced demand (ie, the reduced costs or inconveniences of travel due to these additional options may lead to new trip generation). An interesting complement to the sensitivity analysis of breakeven points for deadheading would be a similar look at breakevens for percent of increased vehicle-distance traveled due to new travel demand. This may be out of scope for this
+
+<--- Page Split --->
+
+manuscript, but it may not require any new analysis, but instead a re- interpretation of the existing breakeven results. If nothing else, it is worth mentioning in the discussion section that induced demand is a similar issue of concern (conceivably on the same order of magnitude as deadheading in the long term, e.g., if AVs lead to land use changes and drastic shifts in behavior).  
+
+3. Enough observations were removed ( \(>50\%\) ) that I would hesitate to characterize the exercise as data cleaning - - a more generic term such as filtering, subsetting, or preparation may be more appropriate. This may be pedantic, but given the "cleaning" term, I thought that only anomalous observations were being removed, but in this case, the goal is not simply to remove anomalies but also to create a dataset more representative and applicable to the analysis.  
+
+4. Removal of 360,000 vehicles sitting unused for \(>14\) months prior to scrappage: I am not expert in scrappage issues, but removing this many observations (nearly as many as the size of the final dataset) seems to warrant a brief discussion of how we expect it to de-bias the dataset and how including it could have altered findings. (If sitting unused in a garage is how most cars ultimately transition towards scrappage, does excluding it alter findings? Similarly, if many Swedish cars find a second life in other countries but are not included in the data excerpt from the government, would that dataset bias alter findings?)  
+
+5. The methods section states that: "The annual driving distance, d, for year t, is assumed to decrease by b = 4.4% per year". How was this value chosen and are results very sensitive to it? 
+6. I don't have a specific suggestion, but on plots where Normal distribution and Weibull distribution results are plotted as separate line types (ex: Fig. 4), I found it difficult to make out some of the dotted lines that were close together.
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+We would like to thank both reviewers for taking the time to thoroughly review our study. Their comments have greatly improved the manuscript. Please find our point- by- point response to all the comments below. Changes have been made in the corresponding areas in the manuscript. All line references are in regards to the revised manuscript. Text highlighted in yellow in citations show what part of the text that was revised.  
+
+Reviewer #1 (Remarks to the Author):  
+
+Dear authors,  
+
+This is a very interesting study investigating the relationship between vehicle use intensity, lifetime, the implications of shared mobility, and the carbon footprint. I think there is potential for this piece of study to be accepted for publication. But intensive revisions are needed.  
+
+- Generally, this study seems very like a combination of three pieces of small studies, which are (1) the relationship between vehicle use intensity, lifetime and lifetime driving distance, which is obtained by using conventional ICEV as the example. (2) a carbon footprint calculation for BEV with the previous relationships in (1) incorporated. (3) a calculation on the empty travel that can be tolerated. I would say these three parts of the study are not very closely integrated with each other. It is more like a loose combination of researches in three different research areas, and trying to find some stories behind the combination. I recommend that the authors might put far more focus on the first part of the study, because currently there lacks understanding in the relationship between driving intensity and vehicle lifetime. And the first part could be not necessarily connected to the BEV carbon footprint and shared AV empty mileage contexts.  
+
+Thank you for pointing out that the three parts do not seem closely integrated. We have thoroughly revised the manuscript to improve the coherence of the study and, also added a more thorough analysis of the relationship between driving intensity and vehicle lifetime.  
+
+We agree that there is a lack of understanding of the relationship between driving intensity and vehicle lifetime. At the same time, there is a strong trend towards electrification both by many car manufacturers and through the ICEV phase- out policies proposed by several governments worldwide. Furthermore, car sharing and ride sharing are to some extent already implemented (e.g., ride hailing) and thoroughly discussed in academia as well as industry, including in the latest assessment from WG3 of the IPCC. Hence, a study trying to understand the relationship between driving intensity and vehicle lifetime to inform studies of future scenarios for transportation and policy analyses should relate them to the potential issues with electrification and potential future shared AVs. That is the reasoning behind our study's objective and why the three different parts are all vital to the study.  
+
+- Another problem when the authors try to incorporate the relationship found in conventional vehicles to shared AVs is that as people are expecting longer driving range for AVs, so the lifetime design for AVs could be very different from conventional vehicles. In that circumstance, I am afraid that the analysis could incorporate significant bias.  
+
+Thank you for highlighting this problem. We acknowledge the risk of biases when using data for ICEVs as the basis for discussing the relationship between driving intensity and vehicle lifetime. Although we agree that there is a risk with using data based on ICEVs to analyze the impact of an emerging technology, we argue that both the design of future regular/autonomous BEVs and to what extent longer driving ranges would affect the relationship are highly uncertain. Even if enough data for a statistical analysis of EVs were available, the risk of bias would still persist. To highlight this, the following discussions has been added in the first section of the manuscript (lines 134- 167):  
+
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs28- 30, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020.31,32 Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first33- 37. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today
+
+<--- Page Split --->
+
+through completely different battery chemistries<sup>38</sup>, changes in charging and use behavior<sup>39</sup>, and/or changed battery design<sup>40</sup> that could potentially yield a cumulative driving distance of more than three million kilometers – effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime<sup>25</sup>, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models<sup>41</sup>.  
+
+In summary, the results suggest that the annual driving intensity indeed has a strong influence on vehicle lifetimes. The relationship between driving intensity and vehicle lifetime may differ between BEVs and ICEVs, but not enough data is yet available to make such a claim. As a consequence, the remainder of this article explores how changes in annual driving intensity may influence the carbon footprint of passenger car travel, assuming that the relationship shown for ICEVs is applicable as a proxy for individually owned and shared autonomous BEVs. We capture the uncertainty in future vehicle lifetimes of (shared and autonomous) BEVs by highlighting extreme values for the relationship between annual driving intensity and vehicles lifetime as well as the empirically estimated relationship based on ICEV retirement data.<sup>7</sup>  
+
+And the following in the discussion section (lines 376- 387):  
+
+"Finally, our conclusions rely on the assumption that the relationship between driving intensity and vehicle lifetime established in the semi- empirical model will hold also for future regular and autonomous BEVs. In this article, we present preliminary evidence suggesting that cars with batteries follow similar trends as ICEVs, but the design and use of future batteries and vehicles are still highly uncertain. Hence, the intention here is to highlight potential consequences based on currently available data and discuss them in relation to extreme cases. Those extreme cases highlight a range of plausible outcomes if the lifetime characteristics of future batteries and vehicles may deviate from those of current passenger cars. In any case, the analysis shows that the carbon footprint may be substantially reduced if the relationship between average annual driving intensity and vehicle lifetime is weakened, pointing to the importance of designing future BEVs (both autonomous and regular) for durability."  
+
+- Figure 1, I would recommend the authors to expand the explanations on this figure. More information should be provided, for example, the relationship between lifetime and lifetime driving distance; driving distance and lifetime driving distance.  
+
+Thank you for this suggestion. Figure 1 has been revised to visualize not only the relationship between driving intensity and vehicle lifetime but also vehicle lifetime vs. total driving distance and total driving distance vs driving intensity. The following analysis is also added (lines 102- 133):  
+
+"The stratification is made for individual average annual driving intensity classes, varying from 0 to 100,000 km per year in steps of 10,000 km per year. For each individual driving intensity class, a close to linear relationship exists between vehicle lifetime and cumulative driving distance. The linear slope becomes steeper with each higher driving intensity class, see Figure 1a. This suggests that the calendar age of a vehicle becomes generally shorter with increasing annual driving intensity. Further, the cumulative driving distances are distributed across a wide range for higher driving intensity classes, see Figure 1c, while the distribution is narrower for lower driving intensities. Hence, the probability of a retirement decision at a specific cumulative driving distance becomes smaller as the annual driving intensity increases. A fixed cumulative driving distance is assumed in many lifecycle assessments of vehicles<sup>13,18</sup>. However, this assumption is not corroborated by the data presented here. Finally, the distribution of vehicle lifetimes becomes narrower and shifts towards lower vehicle lifetimes as the average driving intensity increases, see Figure 1b. Hence, we focus the following analysis on empirically describing the relationship between driving intensity and vehicle lifetime in order to capture the impact of vehicle use on retirement age.  
+
+The average vehicle lifetime decreases with each higher driving intensity class, from 19 years for average driving intensities of 0- 10,000 km per year to 3.9 years for average driving intensities of 90,001- 100,000 km per year, see Figure 1b. The standard deviation of the distributions also indicates that the range of probable lifetimes becomes narrower with increasing annual driving intensity (although the standard deviation increases in relative terms). The standard deviation decreases from 5.0 years for driving intensities of 0- 10,000 km per year to 1.9 years for driving intensities of 90,001- 100,000 km per year (assuming Normal- distributed data). Results for a categorization in four vehicle sizes (mini, medium, large and luxury size cars, see Supplementary Figure 5) suggest that cars with low annual driving intensity are mainly represented by small size cars, while large to luxury size cars mainly have higher annual driving intensities. Medium size cars cover the full spectrum of annual driving intensities."  
+
+- To use the relationship obtained on ICEVs on BEVs could incorporate biases, so that I would recommend the authors to conducted some sensitivity analysis.  
+- If the authors have already determine the value of e to be -0.59, does it make a lot of senses to analyze the
+
+<--- Page Split --->
+
+Thanks again for highlighting the risk of bias in the dataset. We have added a visualization (Supplementary Figure 11) of the data points available for cars with batteries in the analyzed dataset. Although the dataset is too limited for a thorough statistical analysis, the available data points follow similar distributions as the data analyzed for ICEVs. This is thoroughly discussed in Supplementary Notes 1- 3. The following discussion has also been added to the main body of the manuscript (lines 134- 157), as previously mentioned.  
+
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs<sup>28- 30</sup>, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020<sup>31,32</sup>. Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first<sup>33- 37</sup>. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today through completely different battery chemistries<sup>38</sup>, changes in charging and use behavior<sup>39</sup>, and/or changed battery design<sup>40</sup> that could potentially yield a cumulative driving distance of more than three million kilometers - effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime<sup>25</sup>, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models<sup>41,4</sup>.  
+
+Finally, we would also like to highlight the benefit of presenting results for elasticities ranging from 0 to - 1 as a way of testing the sensitivity in the carbon footprint estimations as well as the empty travel breakeven level. Including these two extreme cases serves two purposes: (i) increasing the understanding of the model design, and (ii) how sensitive the model is to the relationship between driving intensity and vehicle lifetime. Hence, if future shared autonomous BEVs are designed in a way where the driving intensity plays a less important role in the decision to retire vehicles, the results are more likely related to an elasticity close to 0. This could be the case if battery degradation is less influenced by going through many charging cycles. The opposite case, where calendar lifetime plays a less important role in the decision to retire vehicles and the elasticity is close to - 1, represents a future where batteries are largely impacted by the number of charging cycles and the total driving distance is fixed. The latter assumption is often used in LCA studies, in which a certain total driving distance over the vehicle's lifetime is assumed. However, the elasticity of - 1 case seems less realistic given the incentives for battery manufacturers to improve battery longevity and enable batteries to cope with extreme events such as fast charging and recent laboratory studies supporting that such battery chemistries are feasible (Yang et al., 2021). We highlight this benefit of the extreme cases more clearly in the manuscript (lines 182- 194):  
+
+"Carbon footprints are also estimated for two extreme cases, \(\epsilon = 0\) and \(\epsilon = - 1\) , representing no influence of driving intensity on lifetime and full influence of driving intensity, respectively. The two extreme cases show the sensitivity of the model design to the assumed elasticity. The range represents possible cases if the model was trained on different retirement data, such as future BEVs when sufficient data becomes available. \(\epsilon = 0\) is a relevant extreme case if future individually owned and/or shared autonomous BEVs are designed in a way where driving intensity has no importance in the decision to retire vehicles. This could be the case if the vehicle and battery degradation is only influenced by calendar age. \(\epsilon = - 1\) represent a case where vehicle aging, including aging of the battery, is only dependent on distance driven (i.e., battery aging only depends on the number of charging cycles). This approach is used in many lifecycle assessments<sup>13,18</sup>, where fixed cumulative vehicles distances are assumed. Note though that the elasticity affecting the distribution is based on the empirical data \((\beta \approx 0.51)\) also for the extreme cases."  
+
+We also note that Reviewer #3 consider the analysis of the extreme values for the elasticity as a strength of the study.  
+
+- the empty travel section, all the discussions are very hard to follow due to the complication to the analysis. I recommend the authors to simply the discussion and highlight the most important findings in an easy-to-follow way.
+
+<--- Page Split --->
+
+We are sorry that you found this section hard to follow and agree that it is complex. We have thoroughly reworked this section in the revised manuscript and highlighted the main outcome that now is based on the fleet- wide analysis.  
+
+- Line 241, yes higher driving intensity leads to shorter vehicle lifetime, which makes the use-phase GHG emissions because more emissions occur in the near future than in the long future. However, this analysis is only from one-vehicle perspective. But when considering that after the retirement of this vehicle, the next vehicle will get the benefit of lower emissions of electricity in the long future, than the near-future loss could be filled.  
+
+Thank you for this suggestion. We agree that modelling a fleet would be a more accurate way of determining the impact of shared autonomous BEVs on the carbon footprint. Hence, we have reworked the sections on the carbon footprint impacts (lines 168- 261) and the breakeven level for empty travel (lines 262- 339) using a simple vehicle fleet turnover simulation considering a fleet of 1000 vehicles. The carbon footprint estimation section of the Methods section was also revised to include the vehicle fleet turnover simulation (lines 498- 588). The carbon footprint estimation section was also moved to the end of the Methods section to follow the structure of the manuscript in general. Since these two sections and the related section in Methods are fully reworked, we have not included the whole text here.  
+
+We would also like to highlight that implementing our semi- empirical lifetime- intensity model in a fleet- wide analysis revealed additional aspects of the model that are important to consider when implementing the full distributions. Using the elasticity design with Normal distributions is simple and easy to understand but has a vital flaw when analyzing high driving intensities. As driving intensities increase, the distribution shifts to lower and even negative lifetimes, which of course is not realistic. This issue is overcome by instead using Weibull distributions, which is by definition always larger than zero. Hence, we have changed the elasticities used in the carbon footprint analysis to those estimated for the Weibull distribution and added the caveat of the Normal distribution to Table 1. Again, thank you for suggesting us to go beyond the one- vehicle perspective, which highlighted this important aspect of our lifetime- intensity model design.  
+
+- line 243, the analysis on the manufacturing-phase GHG emissions actually need also to incorporate the impact from electricity emission factor, as much of energy consumption for battery production is electricity. This will lead to some differences in the manufacturing phase emissions, like the impact from near-term emission factor and long-term emissions factor analyzed in the use-phase emissions.  
+
+Thank you for highlighting that this was not clear in the previous version. The model does incorporate the impact of the electricity emission factor on emissions related to battery production. All vehicles and batteries are assumed to be produced by global average manufacturing industries, using average global electricity. The electricity emission factor is based on direct emissions estimated by the IEA and adjusted to account for upstream processes, as described in the Methods section. The Methods highlights this on the following lines (546- 549 and 574- 577), included below for your convenience.  
+
+"Manufacturing- phase \(\mathrm{CO_2}\) emissions are estimated for car sales in each year based on manufacturing processes as implemented in GREET® for the Stated Policies Scenario, while new and innovative processes are phased in over time for the Sustainable Development Scenario based on a literature review".  
+
+"For the global electricity mix used in manufacturing and for charging, future direct emissions and adjustments to account for transmission and distribution losses (based on the difference between estimated supply and demand) are based on estimates by the IEA® for the two decarbonization pathways, Stated Policies and Sustainable Development."  
+
+- line 257, the authors should not only identify the research weakness, actually this can be addressed by a sensitivity analysis, which I recommend that the authors should add.  
+
+Thanks again for highlighting these issues. We hope that the analyses and discussions added, and outlined above, are sufficient to address this main weakness of our study.  
+
+Reviewer #3 (Remarks to the Author):  
+
+Summary of contribution:  
+
+This paper contributes meaningfully to the literature on vehicle lifetime carbon footprints by using retirement statistics to model relationships between driving intensity and lifetime, and by using those relationships to characterize whether high driving intensities tend to improve or harm carbon footprints. Sensitivity to an appropriate range of models is considered. The value of these findings largely comes from its implications for "future mobility", e.g., ridesourcing, car sharing, and autonomous vehicles. The dataset used is generally appropriate to investigate those implications and provides a much- needed look at intensity- lifetime relationships.
+
+<--- Page Split --->
+
+That said, it isn't clear to me whether the dataset can yield findings that cleanly extrapolate to those "future mobility" options -- this may be something that simply warrants an additional caveat in the discussion section but seems important to acknowledge. It also isn't clear to me whether there are specific insights that could translate these vehicle- level findings to fleet- level findings (are fleets of very few high- use cars better or worse than fleets of very many low- use cars?) -- maybe nothing can be said on that front, but maybe there are at least some specific directions the authors can suggest as future work based on their model findings.  
+
+Thank you for taking the time to thoroughly review our manuscript. We have thoroughly revised the manuscript in response to the comments, including replacing the previous carbon footprint analysis with one based on vehicle fleet turnover simulations as well as more thorough discussions on the applicability of the results for future mobility options. The latter also includes insights from the available but limited data on vehicle retirement of cars with batteries.  
+
+Comments on data and methods:  
+
+1. The study depends on using historical vehicle retirements to infer trends for future (shared) mobility. It seems like one missing piece of the discussion section is whether this subsetted excerpt of all passenger cars may or may not cleanly extrapolate to uses such as car sharing and ride sourcing. Should we expect that the vehicle failure causes and reasons for retirement will be pretty much the same for shared mobility fleets as for all private vehicles? Or is there anything about their driving cycles (mostly long-duration urban driving shifts), ongoing maintenance regimes, or scrappage decisions (made by professional fleet managers instead of individual owners) that might make the data not extrapolate as cleanly such that results are biased in a direction we can characterize? This may be worth some brief comment.  
+
+Thank you for highlighting this problem. We acknowledge the risk of biases when using data for ICEVs as the basis for discussing the relationship between driving intensity and vehicle lifetime. Although we agree that there is a risk with using data based on ICEVs to analyze the impact of an emerging technology, we argue that both the design of future regular/autonomous BEVs and to what extent longer driving ranges would affect the relationship are highly uncertain. Even if enough data for a statistical analysis of EVs were available, the risk of bias would still persist. To highlight this, the following discussions has been added in the first section of the manuscript (lines 134- 167):  
+
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs28- 30, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020.31,32. Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first33- 37. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today through completely different battery chemistries38, changes in charging and use behavior39, and/or changed battery design40 that could potentially yield a cumulative driving distance of more than three million kilometers - effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime25, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models41.  
+
+In summary, the results suggest that the annual driving intensity indeed has a strong influence on vehicle lifetimes. The relationship between driving intensity and vehicle lifetime may differ between BEVs and ICEVs, but not enough data is yet available to make such a claim. As a consequence, the remainder of this article explores how changes in annual driving intensity may influence the carbon footprint of passenger car travel, assuming that the relationship shown for ICEVs is applicable as a proxy for individually owned and shared autonomous BEVs. We capture the uncertainty in future vehicle lifetimes of (shared and autonomous) BEVs by highlighting extreme values for the relationship between annual driving intensity and vehicles lifetime as well as the empirically estimated relationship based on ICEV retirement data.  
+
+And the following in the discussion section (lines 376- 387):
+
+<--- Page Split --->
+
+"Finally, our conclusions rely on the assumption that the relationship between driving intensity and vehicle lifetime established in the semi- empirical model will hold also for future regular and autonomous BEVs. In this article, we present preliminary evidence suggesting that cars with batteries follow similar trends as ICEVs, but the design and use of future batteries and vehicles are still highly uncertain. Hence, the intention here is to highlight potential consequences based on currently available data and discuss them in relation to extreme cases. Those extreme cases highlight a range of plausible outcomes if the lifetime characteristics of future batteries and vehicles may deviate from those of current passenger cars. In any case, the analysis shows that the carbon footprint may be substantially reduced if the relationship between average annual driving intensity and vehicle lifetime is weakened, pointing to the importance of designing future BEVs (both autonomous and regular) for durability."  
+
+2. I am glad to see multiple statistical model types tested for the semi-empirical model. The specific tradeoffs between model types (Table 1) are outside of my expertise, but the explanations provided were clear and concise. Including elasticities of 0 and -1 in the main figures helped me understand how the model is working and also serves as a useful bound.  
+
+Thank you! We agree that highlighting the extreme values for the elasticities in the figures improves understanding of the model and how sensitive it is to the relationship between driving intensity and vehicle lifetime. Hence, if future shared AVs are designed in a way where the driving intensity plays a less important role in the decision to retire vehicles, the results are more likely related to an elasticity close to 0. This could be the case if battery degradation is less influenced by going through many charging cycles. The opposite case, where vehicle lifetime plays a less important role in the decision to retire vehicles and the elasticity is close to -1, represents a future where batteries are largely impacted by the number of charging cycles.  
+
+We have highlighted this benefit more clearly in the manuscript (lines 182- 194):  
+
+"Carbon footprints are also estimated for two extreme cases, \(\epsilon = 0\) and \(\epsilon = - 1\) , representing no influence of driving intensity on lifetime and full influence of driving intensity, respectively. The two extreme cases show the sensitivity of the model design to the assumed elasticity. The range represents possible cases if the model was trained on different retirement data, such as future BEVs when sufficient data becomes available. \(\epsilon = 0\) is a relevant extreme case if future individually owned and/or shared autonomous BEVs are designed in a way where driving intensity has no importance in the decision to retire vehicles. This could be the case if the vehicle and battery degradation is only influenced by calendar age. \(\epsilon = - 1\) represent a case where vehicle aging, including aging of the battery, is only dependent on distance driven (i.e., battery aging only depends on the number of charging cycles). This approach is used in many lifecycle assessments \(^{13,18}\) where fixed cumulative vehicles distances are assumed. Note though that the elasticity affecting the distribution is based on the empirical data \((\beta = 0.51)\) also for the extreme cases."  
+
+3. This analysis is conducted and presented on a per-vehicle basis. I might hesitate to draw fleetwide conclusions from this analysis, and I think it should be mentioned in the problem framing or in discussion that the interactions between average vehicle use level and fleetwide impacts are outside of the paper's scope. For example, if demand is fixed and served 100% via AVs and ridesourcing, each unit increase in vehicle driving intensity seems to imply fewer total cars are needed -- but I don't think this analysis can tell us whether the net carbon impacts of that increase in intensity would be positive or negative.  
+
+Thank you for this suggestion. We agree that modelling a fleet would be a more accurate way of determining the impact of shared autonomous BEVs on the carbon footprint. Hence, we have reworked the sections on the carbon footprint impacts (lines 168- 261) and the breakeven level for empty travel (lines 262- 339) using a simple vehicle fleet turnover simulation considering a fleet of 1000 vehicles. The carbon footprint estimation section of the Methods section was also revised to include the vehicle fleet turnover simulation (lines 498- 588). The carbon footprint estimation section was also moved to the end of the Methods section to follow the structure of the manuscript in general. Since these two sections and the related section in Methods are fully reworked, we have not included the whole text here.  
+
+We would also like to highlight that implementing our semi- empirical lifetime- intensity model in a fleet- wide analysis revealed additional aspects of the model that are important to consider when implementing the full distributions. Using the elasticity design with Normal distributions is simple and easy to understand but has a vital flaw when analyzing high driving intensities. As driving intensities increase, the distribution shifts to lower and even negative lifetimes, which of course is not realistic. This issue is overcome by instead using Weibull distributions, which is by definition always larger than zero. Hence, we have changed the elasticities used in the carbon footprint analysis to those estimated for the Weibull distribution and added the caveat of the Normal distribution to Table 1. Again, thank you for suggesting us to go beyond the one- vehicle perspective, which highlighted this important aspect of our lifetime- intensity model design.  
+
+## Additional comments:  
+
+1. Please comment on the choice to use average electricity grid emissions factors instead of marginal and how this choice may affect your results and findings. In particular, as renewables increase, average electricity
+
+<--- Page Split --->
+
+emissions factors will fall, but marginal emissions factors may or may not change (in nearly all regions the marginal generator is some form of fossil fuel generation). Might using marginal factors alter any findings? If so, it is worth some justification in the text of this choice.  
+
+Marginal coefficients are preferred in consequential LCAs, where the effects on the system of increased use of the analyzed product or service is assessed (Yang, 2016). A prospective LCA can be either consequential or attributional (Arvidsson et al., 2018b), where an attributional prospective LCA provides a snapshot of the future given the evolution of foreground as well as background systems in line with scenario assumptions. A typical consequential prospective LCA would have the intention to go one step further in order to evaluate what impact a specific decision, in our case introduction of shared AVs, would have on each sub- system (Jones et al., 2017), e.g., vehicle manufacturing or electricity generation. This is considered to be beyond the scope of this study. Although, as clearly stated in Arvidsson et al. (2018a): "... an attributional LCA ... [is] effectively identical to a consequential LCA where only first- order (linear) physical flow consequences are considered – or "a consequential LCA based on the attributional [LCA] framework".  
+
+Hence, given the nature of the prospective LCA applied in this study, average coefficients (incl. carbon intensity of electricity) are preferred. Nevertheless, the main concern with applying average carbon intensity of electricity should still be addressed – the risk that changes in electricity demand in response to the analyzed question is large enough to affect the carbon intensity of electricity (Harmsen and Graus, 2013). To handle this risk, care has been taken to choose scenarios for background systems (incl. electricity generation) that match foreground systems. The scenarios for electricity generation referenced for Swedish and EU averages assume increased electricity demand following strong electrification not only in transportation but also in industry. The IEA scenarios (i.e., Stated Policies and Sustainable Development) both also account for increases in electricity demand, specifically looking at global trends within industry and electric vehicle deployment among others.  
+
+The following was added to the manuscript in response to this (lines 559- 561):  
+
+"The carbon intensities used for electricity represent averages for each respective geographic area following the attributional nature of the chosen prospective lifecycle assessment framework \(^{58,59}\) ."  
+
+2. It seems that AVs, ridesourcing, and/or car sharing may cause not only empty vehicle travel, but also additional induced demand (ie, the reduced costs or inconveniences of travel due to these additional options may lead to new trip generation). An interesting complement to the sensitivity analysis of breakeven points for deadheading would be a similar look at breakevens for percent of increased vehicle-distance traveled due to new travel demand. This may be out of scope for this manuscript, but it may not require any new analysis, but instead a re-interpretation of the existing breakeven results. If nothing else, it is worth mentioning in the discussion section that induced demand is a similar issue of concern (conceivably on the same order of magnitude as deadheading in the long term, e.g., if AVs lead to land use changes and drastic shifts in behavior).  
+
+Thanks for highlighting the important aspect of induced travel demand for AVs and car sharing and/or ride sharing. We do consider this to be outside of the scope of this particular study and mention this as one of the trends that will affect pathways for decarbonizing passenger car travel, see lines 344- 346:  
+
+"These trends will affect the pathways towards decarbonization of passenger car travel, including changes in charging patterns \(^{13}\) , cost structures \(^{9}\) and the value of travel time \(^{42 - 44}\) , which may induce additional travel activity \(^{45}\) and cause modal shifts \(^{46,47}\) ."  
+
+To further highlight the impact induced travel demand could have, we have added the following sentence in the discussion section (lines 368- 375):  
+
+"Note that induced travel by autonomous BEVs (both individually owned and shared) has not been assessed in the study. However, this risk is important to consider since the use of autonomous vehicles may substantially increase the travel demand. Autonomous vehicles may effectively reduce the value of travel time and the generalized travel cost \(^{45}\) since the driver does not need to be attentive and can instead use their time in the vehicle for whatever they find convenient. This potential increases in the travel demand could further increase the total carbon footprint for the fleet as a whole."  
+
+3. Enough observations were removed (>50%) that I would hesitate to characterize the exercise as data cleaning -- a more generic term such as filtering, subsetting, or preparation may be more appropriate. This may be pedantic, but given the "cleaning" term, I thought that only anomalous observations were being removed, but in this case, the goal is not simply to remove anomalies but also to create a dataset more representative and applicable to the analysis.  
+
+Thank you for this suggestion. The wording has been changed to "filtering" throughout the manuscript.
+
+<--- Page Split --->
+
+4. Removal of 360,000 vehicles sitting unused for \(>14\) months prior to scrappage: I am not expert in scrappage issues, but removing this many observations (nearly as many as the size of the final dataset) seems to warrant a brief discussion of how we expect it to de-bias the dataset and how including it could have altered findings. (If sitting unused in a garage is how most cars ultimately transition towards scrappage, does excluding it alter findings? Similarly, if many Swedish cars find a second life in other countries but are not included in the data excerpt from the government, would that dataset bias allter findings?)  
+
+Thanks for raising this point. We have also reconsidered the other filtering categories and criteria used since we realized that we had been too conservative in the set criteria. We have decided to make the following adjustments: the time considered between first registration and the manufacturing year has been increased to two years instead of one (resulting in 107,430 additional vehicles), engine types include also ethanol and natural gas since they are internal combustion engines (15,268 additional vehicles), and the average distance traveled must not be greater than 600 km per day instead of 400 km per day (153 additional vehicles). Note that the number of additional vehicles may not sum to the total since one vehicle may fulfil more than one criterion. This enabled us to do a more even stratification of the data with equal size categories for driving intensity (between 0 and 100,000 km/year in 10,000 km/year steps). The adjustments resulted in a slightly higher elasticity of - 0.67 instead of - 0.59 (fitted to Normal distribution).  
+
+In response to your comment on the time between de- registration and last inspection, we have performed a sensitivity analysis. If the time is increased with two years (i.e., vehicles are removed if they sit unused for \(>38\) months), 88,568 vehicles are removed instead of 360,110 and the elasticity would be - 0.65 (fitted to Normal distribution). Hence, largely unaffected. The following statement was added to the manuscript (lines 408- 409):  
+
+"Criterion (ii) filters many observations but including them does not significantly impact the results."  
+
+Exports were limited 2014- 2016 - 8- 18 % of total annual de- registrations. For 2017, the share increased to 25% and from 2018 onwards it's been at levels of 35- 40%. The number of annually retired vehicles has been fairly constant over the same time period (equal to the time period for our data set - 2014- 2018). The exports have been distributed across all vehicle types, but the recent increase in export has been specifically in the age segment 0- 5 years. The Swedish governmental agency Traffic Analysis ties this to a recent policy that gives incentives for new car purchases and the leasing deals that lasts around 3 years (see https://www.trafa.se/vagtrafik/export av personbilar okade kraftigt 2018- 8201/, https://www.trafa.se/vagtrafik/fordon/export- av- personbilar- 2020- 12094/, both in Swedish). The fact that the number of retired vehicles has been fairly constant over the time period, also supports that the increase in exports are likely related to the purchasing prerequisites and should not have any major impact on the findings. It is more difficult to draw a conclusion for the exports during 2014- 2016, but the impact on the results should in any case be minor since they still are limited in number compared to the full dataset.  
+
+5. The methods section states that: "The annual driving distance, d, for year t, is assumed to decrease by \(b = 4.4\%\) per year". How was this value chosen and are results very sensitive to it?  
+
+Thank you highlighting that we omitted including the source. The value is estimated based on statistics on driving distances in Sweden from the governmental agency Transport Analysis. The source has been added to the list of references and the following is added in the Methods section (lines 532- 533):  
+
+"... the annual driving range is assumed to decrease by \(b = 4.4\%\) per year over its lifetime (estimated based on statistics on driving distances in Sweden \(^{57}\) )."  
+
+We have also included a sensitivity analysis of this assumption as part of the new fleet- wide analysis of the breakeven carbon footprint, which is included in Supplementary Figures 9- 10. The sensitivity analysis is summarized in the manuscript as follows:  
+
+Lines 238- 244:  
+
+"... Two other factors also contribute to the decreasing manufacturing- phase emissions as driving intensities increase: the assumed reduction in annual driving intensity for each individual car of 4.4% per year, and the time discretization of one year for the vehicle fleet turnover simulation. The significance of the former factor is tested in Supplementary Figure 9, showing slightly higher carbon footprint when driving intensity is assumed to be constant over each vehicle's lifetime."  
+
+Lines 330- 335:  
+
+"... The sensitivity of the assumed driving intensity decrease rate of 4.4% is also tested, showing higher breakeven levels for the additional empty travel with higher driving intensity decrease rates (i.e., when a larger share of the travel for one vehicle is concentrated to early years in the vehicle's lifetime). Nevertheless, the assumed elasticity in the lifetime- intensity model has a higher impact on the results than the assumed driving intensity decrease rate."
+
+<--- Page Split --->
+
+6. I don't have a specific suggestion, but on plots where Normal distribution and Weibull distribution results are plotted as separate line types (ex: Fig. 4), I found it difficult to make out some of the dotted lines that were close together.  
+
+Thank you for highlighting that this graph is difficult to read. We have adjusted the linetype and opacity of the points to improve readability.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1: Remarks to the Author: Dear authors,  
+
+The response to my comments, and the corresponding revisions are very satisfying. I think the paper has been substantially improved. I have no further comments for the paper.  
+
+Reviewer #3:  
+
+Remarks to the Author:  
+
+The authors have responded comprehensively to my initial feedback. In particular, I appreciate the revisions to the carbon footprint and empty travel breakeven level analyses using a fleet turnover model, which I believe strengthened those sections. I also was glad to see the additional discussion of potential biases regarding BEV implications, the relaxation of some dataset filters, and the additional sensitivity analysis of driving distance decrease rate.  
+
+My substantive concerns have been addressed, and so I believe the manuscript is appropriate for publication.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1fd77ffe2a84b4cd5aa0c17b1d5a10ad75b7f3e14cf2668ed265017f6fda9b10/supplementary_0_Peer Review File__1fd77ffe2a84b4cd5aa0c17b1d5a10ad75b7f3e14cf2668ed265017f6fda9b10_det.mmd

@@ -0,0 +1,394 @@
+<|ref|>title<|/ref|><|det|>[[99, 40, 507, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[106, 110, 373, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[107, 155, 890, 211]]<|/det|>
+Impacts of shared mobility on vehicle lifetimes and the carbon footprint of electric vehicles  
+
+<|ref|>image<|/ref|><|det|>[[95, 732, 262, 780]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[271, 732, 880, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 104]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 120, 291, 163]]<|/det|>
+Reviewer #1: Remarks to the Author: Dear authors,  
+
+<|ref|>text<|/ref|><|det|>[[115, 178, 875, 223]]<|/det|>
+This is a very interesting study investigating the relationship between vehicle use intensity, lifetime, the implications of shared mobility, and the carbon footprint. I think there is potential for this piece of study to be accepted for publication. But intensive revisions are needed.  
+
+<|ref|>text<|/ref|><|det|>[[115, 237, 874, 388]]<|/det|>
+- Generally, this study seems very like a combination of three pieces of small studies, which are (1) the relationship between vehicle use intensity, lifetime and lifetime driving distance, which is obtained by using conventional ICEV as the example. (2) a carbon footprint calculation for BEV with the previous relationships in (1) incorporated. (3) a calculation on the empty travel that can be tolerated. I would say these three parts of the study are not very closely integrated with each other. It is more like a loose combination of researches in three different research areas, and trying to find some stories behind the combination. I recommend that the authors might put far more focus on the first part of the study, because currently there lacks understanding in the relationship between driving intensity and vehicle lifetime. And the first part could be not necessarily connected to the BEV carbon footprint and shared AV empty mileage contexts.  
+
+<|ref|>text<|/ref|><|det|>[[115, 402, 872, 462]]<|/det|>
+- Another problem when the authors try to incorporate the relationship found in conventional vehicles to shared AVs is that as people are expecting longer driving range for AVs, so the lifetime design for AVs could be very different from conventional vehicles. In that circumstance, I am afraid that the analysis could incorporate significant bias.  
+
+<|ref|>text<|/ref|><|det|>[[115, 477, 850, 522]]<|/det|>
+- Figure 1, I would recommend the authors to expand the explanations on this figure. More information should be provided, for example, the relationship between lifetime and lifetime driving distance; driving distance and lifetime driving distance.  
+
+<|ref|>text<|/ref|><|det|>[[115, 536, 812, 566]]<|/det|>
+- To use the relationship obtained on ICEVs on BEVs could incorporate biases, so that I would recommend the authors to conducted some sensitivity analysis.  
+
+<|ref|>text<|/ref|><|det|>[[115, 581, 845, 611]]<|/det|>
+- If the authors have already determine the value of e to be -0.59, does it make a lot of senses to analyze the cases of e equaling to 0 and 1?  
+
+<|ref|>text<|/ref|><|det|>[[115, 626, 872, 671]]<|/det|>
+- the empty travel section, all the discussions are very hard to follow due to the complication to the analysis. I recommend the authors to simply the discussion and highlight the most important findings in an easy-to-follow way.  
+
+<|ref|>text<|/ref|><|det|>[[115, 685, 880, 760]]<|/det|>
+- Line 241, yes higher driving intensity leads to shorter vehicle lifetime, which makes the use-phase GHG emissions because more emissions occur in the near future than in the long future. However, this analysis is only from one-vehicle perspective. But when considering that after the retirement of this vehicle, the next vehicle will get the benefit of lower emissions of electricity in the long future, than the near-future loss could be filled.  
+
+<|ref|>text<|/ref|><|det|>[[115, 775, 870, 834]]<|/det|>
+- line 243, the analysis on the manufacturing-phase GHG emissions actually need also to incorporate the impact from electricity emission factor, as much of energy consumption for battery production is electricity. This will lead to some differences in the manufacturing phase emissions, like the impact from near-term emission factor and long-term emissions factor analyzed in the use-phase emissions.  
+
+<|ref|>text<|/ref|><|det|>[[112, 849, 870, 879]]<|/det|>
+- line 257, the authors should not only identify the research weakness, actually this can be addressed by a sensitivity analysis, which I recommend that the authors should add.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 120, 218, 133]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[115, 136, 291, 149]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 151, 308, 164]]<|/det|>
+Summary of contribution:  
+
+<|ref|>text<|/ref|><|det|>[[115, 165, 872, 270]]<|/det|>
+This paper contributes meaningfully to the literature on vehicle lifetime carbon footprints by using retirement statistics to model relationships between driving intensity and lifetime, and by using those relationships to characterize whether high driving intensities tend to improve or harm carbon footprints. Sensitivity to an appropriate range of models is considered. The value of these findings largely comes from its implications for "future mobility", e.g., ridesourcing, car sharing, and autonomous vehicles. The dataset used is generally appropriate to investigate those implications and provides a much- needed look at intensity- lifetime relationships.  
+
+<|ref|>text<|/ref|><|det|>[[115, 283, 877, 388]]<|/det|>
+That said, it isn't clear to me whether the dataset can yield findings that cleanly extrapolate to those "future mobility" options - - this may be something that simply warrants an additional caveat in the discussion section but seems important to acknowledge. It also isn't clear to me whether there are specific insights that could translate these vehicle- level findings to fleet- level findings (are fleets of very few high- use cars better or worse than fleets of very many low- use cars?) - - maybe nothing can be said on that front, but maybe there are at least some specific directions the authors can suggest as future work based on their model findings.  
+
+<|ref|>text<|/ref|><|det|>[[116, 403, 365, 417]]<|/det|>
+Comments on data and methods:  
+
+<|ref|>text<|/ref|><|det|>[[115, 418, 875, 550]]<|/det|>
+1. The study depends on using historical vehicle retirements to infer trends for future (shared) mobility. It seems like one missing piece of the discussion section is whether this subsetted excerpt of all passenger cars may or may not cleanly extrapolate to uses such as car sharing and ride sourcing. Should we expect that the vehicle failure causes and reasons for retirement will be pretty much the same for shared mobility fleets as for all private vehicles? Or is there anything about their driving cycles (mostly long-duration urban driving shifts), ongoing maintenance regimes, or scrappage decisions (made by professional fleet managers instead of individual owners) that might make the data not extrapolate as cleanly such that results are biased in a direction we can characterize? This may be worth some brief comment.  
+
+<|ref|>text<|/ref|><|det|>[[115, 551, 864, 610]]<|/det|>
+2. I am glad to see multiple statistical model types tested for the semi-empirical model. The specific tradeoffs between model types (Table 1) are outside of my expertise, but the explanations provided were clear and concise. Including elasticities of 0 and -1 in the main figures helped me understand how the model is working and also serves as a useful bound.  
+
+<|ref|>text<|/ref|><|det|>[[115, 611, 882, 715]]<|/det|>
+3. This analysis is conducted and presented on a per-vehicle basis. I might hesitate to draw fleetwide conclusions from this analysis, and I think it should be mentioned in the problem framing or in discussion that the interactions between average vehicle use level and fleetwide impacts are outside of the paper's scope. For example, if demand is fixed and served 100% via AVs and ridesourcing, each unit increase in vehicle driving intensity seems to imply fewer total cars are needed -- but I don't think this analysis can tell us whether the net carbon impacts of that increase in intensity would be positive or negative.  
+
+<|ref|>text<|/ref|><|det|>[[116, 731, 278, 744]]<|/det|>
+Additional comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 745, 868, 893]]<|/det|>
+1. Please comment on the choice to use average electricity grid emissions factors instead of marginal and how this choice may affect your results and findings. In particular, as renewables increase, average electricity emissions factors will fall, but marginal emissions factors may or may not change (in nearly all regions the marginal generator is some form of fossil fuel generation). Might using marginal factors alter any findings? If so, it is worth some justification in the text of this choice. 
+2. It seems that AVs, ridesourcing, and/or car sharing may cause not only empty vehicle travel, but also additional induced demand (ie, the reduced costs or inconveniences of travel due to these additional options may lead to new trip generation). An interesting complement to the sensitivity analysis of breakeven points for deadheading would be a similar look at breakevens for percent of increased vehicle-distance traveled due to new travel demand. This may be out of scope for this
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[113, 89, 872, 150]]<|/det|>
+manuscript, but it may not require any new analysis, but instead a re- interpretation of the existing breakeven results. If nothing else, it is worth mentioning in the discussion section that induced demand is a similar issue of concern (conceivably on the same order of magnitude as deadheading in the long term, e.g., if AVs lead to land use changes and drastic shifts in behavior).  
+
+<|ref|>text<|/ref|><|det|>[[113, 150, 870, 225]]<|/det|>
+3. Enough observations were removed ( \(>50\%\) ) that I would hesitate to characterize the exercise as data cleaning - - a more generic term such as filtering, subsetting, or preparation may be more appropriate. This may be pedantic, but given the "cleaning" term, I thought that only anomalous observations were being removed, but in this case, the goal is not simply to remove anomalies but also to create a dataset more representative and applicable to the analysis.  
+
+<|ref|>text<|/ref|><|det|>[[113, 225, 872, 328]]<|/det|>
+4. Removal of 360,000 vehicles sitting unused for \(>14\) months prior to scrappage: I am not expert in scrappage issues, but removing this many observations (nearly as many as the size of the final dataset) seems to warrant a brief discussion of how we expect it to de-bias the dataset and how including it could have altered findings. (If sitting unused in a garage is how most cars ultimately transition towards scrappage, does excluding it alter findings? Similarly, if many Swedish cars find a second life in other countries but are not included in the data excerpt from the government, would that dataset bias alter findings?)  
+
+<|ref|>text<|/ref|><|det|>[[113, 328, 881, 402]]<|/det|>
+5. The methods section states that: "The annual driving distance, d, for year t, is assumed to decrease by b = 4.4% per year". How was this value chosen and are results very sensitive to it? 
+6. I don't have a specific suggestion, but on plots where Normal distribution and Weibull distribution results are plotted as separate line types (ex: Fig. 4), I found it difficult to make out some of the dotted lines that were close together.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[152, 83, 311, 96]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[183, 107, 840, 159]]<|/det|>
+We would like to thank both reviewers for taking the time to thoroughly review our study. Their comments have greatly improved the manuscript. Please find our point- by- point response to all the comments below. Changes have been made in the corresponding areas in the manuscript. All line references are in regards to the revised manuscript. Text highlighted in yellow in citations show what part of the text that was revised.  
+
+<|ref|>text<|/ref|><|det|>[[153, 170, 386, 183]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[152, 195, 236, 207]]<|/det|>
+Dear authors,  
+
+<|ref|>text<|/ref|><|det|>[[152, 218, 812, 256]]<|/det|>
+This is a very interesting study investigating the relationship between vehicle use intensity, lifetime, the implications of shared mobility, and the carbon footprint. I think there is potential for this piece of study to be accepted for publication. But intensive revisions are needed.  
+
+<|ref|>text<|/ref|><|det|>[[151, 266, 847, 380]]<|/det|>
+- Generally, this study seems very like a combination of three pieces of small studies, which are (1) the relationship between vehicle use intensity, lifetime and lifetime driving distance, which is obtained by using conventional ICEV as the example. (2) a carbon footprint calculation for BEV with the previous relationships in (1) incorporated. (3) a calculation on the empty travel that can be tolerated. I would say these three parts of the study are not very closely integrated with each other. It is more like a loose combination of researches in three different research areas, and trying to find some stories behind the combination. I recommend that the authors might put far more focus on the first part of the study, because currently there lacks understanding in the relationship between driving intensity and vehicle lifetime. And the first part could be not necessarily connected to the BEV carbon footprint and shared AV empty mileage contexts.  
+
+<|ref|>text<|/ref|><|det|>[[183, 390, 838, 429]]<|/det|>
+Thank you for pointing out that the three parts do not seem closely integrated. We have thoroughly revised the manuscript to improve the coherence of the study and, also added a more thorough analysis of the relationship between driving intensity and vehicle lifetime.  
+
+<|ref|>text<|/ref|><|det|>[[183, 439, 838, 552]]<|/det|>
+We agree that there is a lack of understanding of the relationship between driving intensity and vehicle lifetime. At the same time, there is a strong trend towards electrification both by many car manufacturers and through the ICEV phase- out policies proposed by several governments worldwide. Furthermore, car sharing and ride sharing are to some extent already implemented (e.g., ride hailing) and thoroughly discussed in academia as well as industry, including in the latest assessment from WG3 of the IPCC. Hence, a study trying to understand the relationship between driving intensity and vehicle lifetime to inform studies of future scenarios for transportation and policy analyses should relate them to the potential issues with electrification and potential future shared AVs. That is the reasoning behind our study's objective and why the three different parts are all vital to the study.  
+
+<|ref|>text<|/ref|><|det|>[[152, 562, 833, 613]]<|/det|>
+- Another problem when the authors try to incorporate the relationship found in conventional vehicles to shared AVs is that as people are expecting longer driving range for AVs, so the lifetime design for AVs could be very different from conventional vehicles. In that circumstance, I am afraid that the analysis could incorporate significant bias.  
+
+<|ref|>text<|/ref|><|det|>[[183, 635, 836, 725]]<|/det|>
+Thank you for highlighting this problem. We acknowledge the risk of biases when using data for ICEVs as the basis for discussing the relationship between driving intensity and vehicle lifetime. Although we agree that there is a risk with using data based on ICEVs to analyze the impact of an emerging technology, we argue that both the design of future regular/autonomous BEVs and to what extent longer driving ranges would affect the relationship are highly uncertain. Even if enough data for a statistical analysis of EVs were available, the risk of bias would still persist. To highlight this, the following discussions has been added in the first section of the manuscript (lines 134- 167):  
+
+<|ref|>text<|/ref|><|det|>[[183, 735, 845, 870]]<|/det|>
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs28- 30, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020.31,32 Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+<|ref|>text<|/ref|><|det|>[[183, 870, 838, 908]]<|/det|>
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first33- 37. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[184, 82, 843, 170]]<|/det|>
+through completely different battery chemistries<sup>38</sup>, changes in charging and use behavior<sup>39</sup>, and/or changed battery design<sup>40</sup> that could potentially yield a cumulative driving distance of more than three million kilometers – effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime<sup>25</sup>, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models<sup>41</sup>.  
+
+<|ref|>text<|/ref|><|det|>[[184, 181, 843, 281]]<|/det|>
+In summary, the results suggest that the annual driving intensity indeed has a strong influence on vehicle lifetimes. The relationship between driving intensity and vehicle lifetime may differ between BEVs and ICEVs, but not enough data is yet available to make such a claim. As a consequence, the remainder of this article explores how changes in annual driving intensity may influence the carbon footprint of passenger car travel, assuming that the relationship shown for ICEVs is applicable as a proxy for individually owned and shared autonomous BEVs. We capture the uncertainty in future vehicle lifetimes of (shared and autonomous) BEVs by highlighting extreme values for the relationship between annual driving intensity and vehicles lifetime as well as the empirically estimated relationship based on ICEV retirement data.<sup>7</sup>  
+
+<|ref|>text<|/ref|><|det|>[[184, 292, 544, 306]]<|/det|>
+And the following in the discussion section (lines 376- 387):  
+
+<|ref|>text<|/ref|><|det|>[[183, 316, 845, 442]]<|/det|>
+"Finally, our conclusions rely on the assumption that the relationship between driving intensity and vehicle lifetime established in the semi- empirical model will hold also for future regular and autonomous BEVs. In this article, we present preliminary evidence suggesting that cars with batteries follow similar trends as ICEVs, but the design and use of future batteries and vehicles are still highly uncertain. Hence, the intention here is to highlight potential consequences based on currently available data and discuss them in relation to extreme cases. Those extreme cases highlight a range of plausible outcomes if the lifetime characteristics of future batteries and vehicles may deviate from those of current passenger cars. In any case, the analysis shows that the carbon footprint may be substantially reduced if the relationship between average annual driving intensity and vehicle lifetime is weakened, pointing to the importance of designing future BEVs (both autonomous and regular) for durability."  
+
+<|ref|>text<|/ref|><|det|>[[150, 452, 844, 491]]<|/det|>
+- Figure 1, I would recommend the authors to expand the explanations on this figure. More information should be provided, for example, the relationship between lifetime and lifetime driving distance; driving distance and lifetime driving distance.  
+
+<|ref|>text<|/ref|><|det|>[[184, 501, 812, 540]]<|/det|>
+Thank you for this suggestion. Figure 1 has been revised to visualize not only the relationship between driving intensity and vehicle lifetime but also vehicle lifetime vs. total driving distance and total driving distance vs driving intensity. The following analysis is also added (lines 102- 133):  
+
+<|ref|>text<|/ref|><|det|>[[183, 550, 842, 713]]<|/det|>
+"The stratification is made for individual average annual driving intensity classes, varying from 0 to 100,000 km per year in steps of 10,000 km per year. For each individual driving intensity class, a close to linear relationship exists between vehicle lifetime and cumulative driving distance. The linear slope becomes steeper with each higher driving intensity class, see Figure 1a. This suggests that the calendar age of a vehicle becomes generally shorter with increasing annual driving intensity. Further, the cumulative driving distances are distributed across a wide range for higher driving intensity classes, see Figure 1c, while the distribution is narrower for lower driving intensities. Hence, the probability of a retirement decision at a specific cumulative driving distance becomes smaller as the annual driving intensity increases. A fixed cumulative driving distance is assumed in many lifecycle assessments of vehicles<sup>13,18</sup>. However, this assumption is not corroborated by the data presented here. Finally, the distribution of vehicle lifetimes becomes narrower and shifts towards lower vehicle lifetimes as the average driving intensity increases, see Figure 1b. Hence, we focus the following analysis on empirically describing the relationship between driving intensity and vehicle lifetime in order to capture the impact of vehicle use on retirement age.  
+
+<|ref|>text<|/ref|><|det|>[[183, 722, 843, 847]]<|/det|>
+The average vehicle lifetime decreases with each higher driving intensity class, from 19 years for average driving intensities of 0- 10,000 km per year to 3.9 years for average driving intensities of 90,001- 100,000 km per year, see Figure 1b. The standard deviation of the distributions also indicates that the range of probable lifetimes becomes narrower with increasing annual driving intensity (although the standard deviation increases in relative terms). The standard deviation decreases from 5.0 years for driving intensities of 0- 10,000 km per year to 1.9 years for driving intensities of 90,001- 100,000 km per year (assuming Normal- distributed data). Results for a categorization in four vehicle sizes (mini, medium, large and luxury size cars, see Supplementary Figure 5) suggest that cars with low annual driving intensity are mainly represented by small size cars, while large to luxury size cars mainly have higher annual driving intensities. Medium size cars cover the full spectrum of annual driving intensities."  
+
+<|ref|>text<|/ref|><|det|>[[150, 857, 830, 896]]<|/det|>
+- To use the relationship obtained on ICEVs on BEVs could incorporate biases, so that I would recommend the authors to conducted some sensitivity analysis.  
+- If the authors have already determine the value of e to be -0.59, does it make a lot of senses to analyze the
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[183, 108, 840, 183]]<|/det|>
+Thanks again for highlighting the risk of bias in the dataset. We have added a visualization (Supplementary Figure 11) of the data points available for cars with batteries in the analyzed dataset. Although the dataset is too limited for a thorough statistical analysis, the available data points follow similar distributions as the data analyzed for ICEVs. This is thoroughly discussed in Supplementary Notes 1- 3. The following discussion has also been added to the main body of the manuscript (lines 134- 157), as previously mentioned.  
+
+<|ref|>text<|/ref|><|det|>[[183, 194, 847, 331]]<|/det|>
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs<sup>28- 30</sup>, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020<sup>31,32</sup>. Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+<|ref|>text<|/ref|><|det|>[[183, 341, 842, 464]]<|/det|>
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first<sup>33- 37</sup>. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today through completely different battery chemistries<sup>38</sup>, changes in charging and use behavior<sup>39</sup>, and/or changed battery design<sup>40</sup> that could potentially yield a cumulative driving distance of more than three million kilometers - effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime<sup>25</sup>, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models<sup>41,4</sup>.  
+
+<|ref|>text<|/ref|><|det|>[[183, 475, 847, 662]]<|/det|>
+Finally, we would also like to highlight the benefit of presenting results for elasticities ranging from 0 to - 1 as a way of testing the sensitivity in the carbon footprint estimations as well as the empty travel breakeven level. Including these two extreme cases serves two purposes: (i) increasing the understanding of the model design, and (ii) how sensitive the model is to the relationship between driving intensity and vehicle lifetime. Hence, if future shared autonomous BEVs are designed in a way where the driving intensity plays a less important role in the decision to retire vehicles, the results are more likely related to an elasticity close to 0. This could be the case if battery degradation is less influenced by going through many charging cycles. The opposite case, where calendar lifetime plays a less important role in the decision to retire vehicles and the elasticity is close to - 1, represents a future where batteries are largely impacted by the number of charging cycles and the total driving distance is fixed. The latter assumption is often used in LCA studies, in which a certain total driving distance over the vehicle's lifetime is assumed. However, the elasticity of - 1 case seems less realistic given the incentives for battery manufacturers to improve battery longevity and enable batteries to cope with extreme events such as fast charging and recent laboratory studies supporting that such battery chemistries are feasible (Yang et al., 2021). We highlight this benefit of the extreme cases more clearly in the manuscript (lines 182- 194):  
+
+<|ref|>text<|/ref|><|det|>[[183, 672, 842, 813]]<|/det|>
+"Carbon footprints are also estimated for two extreme cases, \(\epsilon = 0\) and \(\epsilon = - 1\) , representing no influence of driving intensity on lifetime and full influence of driving intensity, respectively. The two extreme cases show the sensitivity of the model design to the assumed elasticity. The range represents possible cases if the model was trained on different retirement data, such as future BEVs when sufficient data becomes available. \(\epsilon = 0\) is a relevant extreme case if future individually owned and/or shared autonomous BEVs are designed in a way where driving intensity has no importance in the decision to retire vehicles. This could be the case if the vehicle and battery degradation is only influenced by calendar age. \(\epsilon = - 1\) represent a case where vehicle aging, including aging of the battery, is only dependent on distance driven (i.e., battery aging only depends on the number of charging cycles). This approach is used in many lifecycle assessments<sup>13,18</sup>, where fixed cumulative vehicles distances are assumed. Note though that the elasticity affecting the distribution is based on the empirical data \((\beta \approx 0.51)\) also for the extreme cases."  
+
+<|ref|>text<|/ref|><|det|>[[183, 824, 843, 850]]<|/det|>
+We also note that Reviewer #3 consider the analysis of the extreme values for the elasticity as a strength of the study.  
+
+<|ref|>text<|/ref|><|det|>[[152, 860, 827, 900]]<|/det|>
+- the empty travel section, all the discussions are very hard to follow due to the complication to the analysis. I recommend the authors to simply the discussion and highlight the most important findings in an easy-to-follow way.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[182, 82, 840, 122]]<|/det|>
+We are sorry that you found this section hard to follow and agree that it is complex. We have thoroughly reworked this section in the revised manuscript and highlighted the main outcome that now is based on the fleet- wide analysis.  
+
+<|ref|>text<|/ref|><|det|>[[152, 131, 844, 184]]<|/det|>
+- Line 241, yes higher driving intensity leads to shorter vehicle lifetime, which makes the use-phase GHG emissions because more emissions occur in the near future than in the long future. However, this analysis is only from one-vehicle perspective. But when considering that after the retirement of this vehicle, the next vehicle will get the benefit of lower emissions of electricity in the long future, than the near-future loss could be filled.  
+
+<|ref|>text<|/ref|><|det|>[[182, 193, 844, 294]]<|/det|>
+Thank you for this suggestion. We agree that modelling a fleet would be a more accurate way of determining the impact of shared autonomous BEVs on the carbon footprint. Hence, we have reworked the sections on the carbon footprint impacts (lines 168- 261) and the breakeven level for empty travel (lines 262- 339) using a simple vehicle fleet turnover simulation considering a fleet of 1000 vehicles. The carbon footprint estimation section of the Methods section was also revised to include the vehicle fleet turnover simulation (lines 498- 588). The carbon footprint estimation section was also moved to the end of the Methods section to follow the structure of the manuscript in general. Since these two sections and the related section in Methods are fully reworked, we have not included the whole text here.  
+
+<|ref|>text<|/ref|><|det|>[[182, 304, 845, 416]]<|/det|>
+We would also like to highlight that implementing our semi- empirical lifetime- intensity model in a fleet- wide analysis revealed additional aspects of the model that are important to consider when implementing the full distributions. Using the elasticity design with Normal distributions is simple and easy to understand but has a vital flaw when analyzing high driving intensities. As driving intensities increase, the distribution shifts to lower and even negative lifetimes, which of course is not realistic. This issue is overcome by instead using Weibull distributions, which is by definition always larger than zero. Hence, we have changed the elasticities used in the carbon footprint analysis to those estimated for the Weibull distribution and added the caveat of the Normal distribution to Table 1. Again, thank you for suggesting us to go beyond the one- vehicle perspective, which highlighted this important aspect of our lifetime- intensity model design.  
+
+<|ref|>text<|/ref|><|det|>[[152, 426, 843, 479]]<|/det|>
+- line 243, the analysis on the manufacturing-phase GHG emissions actually need also to incorporate the impact from electricity emission factor, as much of energy consumption for battery production is electricity. This will lead to some differences in the manufacturing phase emissions, like the impact from near-term emission factor and long-term emissions factor analyzed in the use-phase emissions.  
+
+<|ref|>text<|/ref|><|det|>[[183, 488, 844, 564]]<|/det|>
+Thank you for highlighting that this was not clear in the previous version. The model does incorporate the impact of the electricity emission factor on emissions related to battery production. All vehicles and batteries are assumed to be produced by global average manufacturing industries, using average global electricity. The electricity emission factor is based on direct emissions estimated by the IEA and adjusted to account for upstream processes, as described in the Methods section. The Methods highlights this on the following lines (546- 549 and 574- 577), included below for your convenience.  
+
+<|ref|>text<|/ref|><|det|>[[183, 574, 814, 626]]<|/det|>
+"Manufacturing- phase \(\mathrm{CO_2}\) emissions are estimated for car sales in each year based on manufacturing processes as implemented in GREET® for the Stated Policies Scenario, while new and innovative processes are phased in over time for the Sustainable Development Scenario based on a literature review".  
+
+<|ref|>text<|/ref|><|det|>[[183, 636, 843, 688]]<|/det|>
+"For the global electricity mix used in manufacturing and for charging, future direct emissions and adjustments to account for transmission and distribution losses (based on the difference between estimated supply and demand) are based on estimates by the IEA® for the two decarbonization pathways, Stated Policies and Sustainable Development."  
+
+<|ref|>text<|/ref|><|det|>[[152, 697, 803, 724]]<|/det|>
+- line 257, the authors should not only identify the research weakness, actually this can be addressed by a sensitivity analysis, which I recommend that the authors should add.  
+
+<|ref|>text<|/ref|><|det|>[[180, 734, 843, 761]]<|/det|>
+Thanks again for highlighting these issues. We hope that the analyses and discussions added, and outlined above, are sufficient to address this main weakness of our study.  
+
+<|ref|>text<|/ref|><|det|>[[153, 783, 386, 797]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[152, 808, 308, 821]]<|/det|>
+Summary of contribution:  
+
+<|ref|>text<|/ref|><|det|>[[152, 821, 820, 910]]<|/det|>
+This paper contributes meaningfully to the literature on vehicle lifetime carbon footprints by using retirement statistics to model relationships between driving intensity and lifetime, and by using those relationships to characterize whether high driving intensities tend to improve or harm carbon footprints. Sensitivity to an appropriate range of models is considered. The value of these findings largely comes from its implications for "future mobility", e.g., ridesourcing, car sharing, and autonomous vehicles. The dataset used is generally appropriate to investigate those implications and provides a much- needed look at intensity- lifetime relationships.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[152, 95, 843, 170]]<|/det|>
+That said, it isn't clear to me whether the dataset can yield findings that cleanly extrapolate to those "future mobility" options -- this may be something that simply warrants an additional caveat in the discussion section but seems important to acknowledge. It also isn't clear to me whether there are specific insights that could translate these vehicle- level findings to fleet- level findings (are fleets of very few high- use cars better or worse than fleets of very many low- use cars?) -- maybe nothing can be said on that front, but maybe there are at least some specific directions the authors can suggest as future work based on their model findings.  
+
+<|ref|>text<|/ref|><|det|>[[183, 181, 845, 244]]<|/det|>
+Thank you for taking the time to thoroughly review our manuscript. We have thoroughly revised the manuscript in response to the comments, including replacing the previous carbon footprint analysis with one based on vehicle fleet turnover simulations as well as more thorough discussions on the applicability of the results for future mobility options. The latter also includes insights from the available but limited data on vehicle retirement of cars with batteries.  
+
+<|ref|>text<|/ref|><|det|>[[153, 255, 357, 268]]<|/det|>
+Comments on data and methods:  
+
+<|ref|>text<|/ref|><|det|>[[152, 267, 841, 368]]<|/det|>
+1. The study depends on using historical vehicle retirements to infer trends for future (shared) mobility. It seems like one missing piece of the discussion section is whether this subsetted excerpt of all passenger cars may or may not cleanly extrapolate to uses such as car sharing and ride sourcing. Should we expect that the vehicle failure causes and reasons for retirement will be pretty much the same for shared mobility fleets as for all private vehicles? Or is there anything about their driving cycles (mostly long-duration urban driving shifts), ongoing maintenance regimes, or scrappage decisions (made by professional fleet managers instead of individual owners) that might make the data not extrapolate as cleanly such that results are biased in a direction we can characterize? This may be worth some brief comment.  
+
+<|ref|>text<|/ref|><|det|>[[183, 378, 837, 466]]<|/det|>
+Thank you for highlighting this problem. We acknowledge the risk of biases when using data for ICEVs as the basis for discussing the relationship between driving intensity and vehicle lifetime. Although we agree that there is a risk with using data based on ICEVs to analyze the impact of an emerging technology, we argue that both the design of future regular/autonomous BEVs and to what extent longer driving ranges would affect the relationship are highly uncertain. Even if enough data for a statistical analysis of EVs were available, the risk of bias would still persist. To highlight this, the following discussions has been added in the first section of the manuscript (lines 134- 167):  
+
+<|ref|>text<|/ref|><|det|>[[183, 476, 845, 612]]<|/det|>
+"Currently, battery degradation is often raised as a constrain to the cumulative driving distance and lifetime of BEVs28- 30, but the BEV is a relatively new technology on the market and, hence, statistics on battery lifetimes from real- world driving are scarce. The number of electric vehicles on the world's roads were in the thousands in 2010 and grew rapidly to reach about 2 million by 2016 and over 10 million by 2020.31,32. Hence, if enough retirement statistics for electric vehicles were available to make thorough statistical analyses, most vehicles would be much less than 10 years old. However, the limited data currently available on cars with batteries in Swedish vehicle retirement statistics show similar distributions as the stratified data presented above, see Supplementary Notes 1- 3 and Supplementary Figures 11- 12. However, the data show shorter lifetimes on average (due to the limited historic data on electrified vehicles) and with a bias towards hybrid electric vehicles (HEVs) due to very few BEVs and plug- in hybrid electric vehicles (PHEVs) having been retired during the analyzed period.  
+
+<|ref|>text<|/ref|><|det|>[[183, 622, 842, 747]]<|/det|>
+Many BEV manufacturers already have warranties for their batteries of about seven to eight years or about 150,000 to 240,000 km, whichever comes first33- 37. Future battery chemistries may further reduce degradation. Some studies suggest that future batteries may have significantly longer lifetimes than today through completely different battery chemistries38, changes in charging and use behavior39, and/or changed battery design40 that could potentially yield a cumulative driving distance of more than three million kilometers - effectively outliving the vehicle. These improvements, if they materialize, would likely improve the cycling of the batteries. However, other factors could still limit the vehicle's lifetime25, such as accidents, aging of other vehicle parts (e.g., structural elements of chassis and body), economic reasons and consumer trends. Further, the durability of the vehicle is significantly dependent on the vehicle design, material selection and business models41.  
+
+<|ref|>text<|/ref|><|det|>[[183, 757, 842, 859]]<|/det|>
+In summary, the results suggest that the annual driving intensity indeed has a strong influence on vehicle lifetimes. The relationship between driving intensity and vehicle lifetime may differ between BEVs and ICEVs, but not enough data is yet available to make such a claim. As a consequence, the remainder of this article explores how changes in annual driving intensity may influence the carbon footprint of passenger car travel, assuming that the relationship shown for ICEVs is applicable as a proxy for individually owned and shared autonomous BEVs. We capture the uncertainty in future vehicle lifetimes of (shared and autonomous) BEVs by highlighting extreme values for the relationship between annual driving intensity and vehicles lifetime as well as the empirically estimated relationship based on ICEV retirement data.  
+
+<|ref|>text<|/ref|><|det|>[[183, 870, 545, 884]]<|/det|>
+And the following in the discussion section (lines 376- 387):
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[183, 82, 847, 208]]<|/det|>
+"Finally, our conclusions rely on the assumption that the relationship between driving intensity and vehicle lifetime established in the semi- empirical model will hold also for future regular and autonomous BEVs. In this article, we present preliminary evidence suggesting that cars with batteries follow similar trends as ICEVs, but the design and use of future batteries and vehicles are still highly uncertain. Hence, the intention here is to highlight potential consequences based on currently available data and discuss them in relation to extreme cases. Those extreme cases highlight a range of plausible outcomes if the lifetime characteristics of future batteries and vehicles may deviate from those of current passenger cars. In any case, the analysis shows that the carbon footprint may be substantially reduced if the relationship between average annual driving intensity and vehicle lifetime is weakened, pointing to the importance of designing future BEVs (both autonomous and regular) for durability."  
+
+<|ref|>text<|/ref|><|det|>[[152, 219, 848, 270]]<|/det|>
+2. I am glad to see multiple statistical model types tested for the semi-empirical model. The specific tradeoffs between model types (Table 1) are outside of my expertise, but the explanations provided were clear and concise. Including elasticities of 0 and -1 in the main figures helped me understand how the model is working and also serves as a useful bound.  
+
+<|ref|>text<|/ref|><|det|>[[183, 280, 847, 368]]<|/det|>
+Thank you! We agree that highlighting the extreme values for the elasticities in the figures improves understanding of the model and how sensitive it is to the relationship between driving intensity and vehicle lifetime. Hence, if future shared AVs are designed in a way where the driving intensity plays a less important role in the decision to retire vehicles, the results are more likely related to an elasticity close to 0. This could be the case if battery degradation is less influenced by going through many charging cycles. The opposite case, where vehicle lifetime plays a less important role in the decision to retire vehicles and the elasticity is close to -1, represents a future where batteries are largely impacted by the number of charging cycles.  
+
+<|ref|>text<|/ref|><|det|>[[183, 378, 670, 392]]<|/det|>
+We have highlighted this benefit more clearly in the manuscript (lines 182- 194):  
+
+<|ref|>text<|/ref|><|det|>[[183, 403, 841, 543]]<|/det|>
+"Carbon footprints are also estimated for two extreme cases, \(\epsilon = 0\) and \(\epsilon = - 1\) , representing no influence of driving intensity on lifetime and full influence of driving intensity, respectively. The two extreme cases show the sensitivity of the model design to the assumed elasticity. The range represents possible cases if the model was trained on different retirement data, such as future BEVs when sufficient data becomes available. \(\epsilon = 0\) is a relevant extreme case if future individually owned and/or shared autonomous BEVs are designed in a way where driving intensity has no importance in the decision to retire vehicles. This could be the case if the vehicle and battery degradation is only influenced by calendar age. \(\epsilon = - 1\) represent a case where vehicle aging, including aging of the battery, is only dependent on distance driven (i.e., battery aging only depends on the number of charging cycles). This approach is used in many lifecycle assessments \(^{13,18}\) where fixed cumulative vehicles distances are assumed. Note though that the elasticity affecting the distribution is based on the empirical data \((\beta = 0.51)\) also for the extreme cases."  
+
+<|ref|>text<|/ref|><|det|>[[152, 553, 847, 628]]<|/det|>
+3. This analysis is conducted and presented on a per-vehicle basis. I might hesitate to draw fleetwide conclusions from this analysis, and I think it should be mentioned in the problem framing or in discussion that the interactions between average vehicle use level and fleetwide impacts are outside of the paper's scope. For example, if demand is fixed and served 100% via AVs and ridesourcing, each unit increase in vehicle driving intensity seems to imply fewer total cars are needed -- but I don't think this analysis can tell us whether the net carbon impacts of that increase in intensity would be positive or negative.  
+
+<|ref|>text<|/ref|><|det|>[[183, 639, 844, 739]]<|/det|>
+Thank you for this suggestion. We agree that modelling a fleet would be a more accurate way of determining the impact of shared autonomous BEVs on the carbon footprint. Hence, we have reworked the sections on the carbon footprint impacts (lines 168- 261) and the breakeven level for empty travel (lines 262- 339) using a simple vehicle fleet turnover simulation considering a fleet of 1000 vehicles. The carbon footprint estimation section of the Methods section was also revised to include the vehicle fleet turnover simulation (lines 498- 588). The carbon footprint estimation section was also moved to the end of the Methods section to follow the structure of the manuscript in general. Since these two sections and the related section in Methods are fully reworked, we have not included the whole text here.  
+
+<|ref|>text<|/ref|><|det|>[[183, 750, 844, 862]]<|/det|>
+We would also like to highlight that implementing our semi- empirical lifetime- intensity model in a fleet- wide analysis revealed additional aspects of the model that are important to consider when implementing the full distributions. Using the elasticity design with Normal distributions is simple and easy to understand but has a vital flaw when analyzing high driving intensities. As driving intensities increase, the distribution shifts to lower and even negative lifetimes, which of course is not realistic. This issue is overcome by instead using Weibull distributions, which is by definition always larger than zero. Hence, we have changed the elasticities used in the carbon footprint analysis to those estimated for the Weibull distribution and added the caveat of the Normal distribution to Table 1. Again, thank you for suggesting us to go beyond the one- vehicle perspective, which highlighted this important aspect of our lifetime- intensity model design.  
+
+<|ref|>sub_title<|/ref|><|det|>[[152, 874, 285, 886]]<|/det|>
+## Additional comments:  
+
+<|ref|>text<|/ref|><|det|>[[150, 886, 831, 912]]<|/det|>
+1. Please comment on the choice to use average electricity grid emissions factors instead of marginal and how this choice may affect your results and findings. In particular, as renewables increase, average electricity
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[152, 82, 847, 120]]<|/det|>
+emissions factors will fall, but marginal emissions factors may or may not change (in nearly all regions the marginal generator is some form of fossil fuel generation). Might using marginal factors alter any findings? If so, it is worth some justification in the text of this choice.  
+
+<|ref|>text<|/ref|><|det|>[[205, 131, 844, 270]]<|/det|>
+Marginal coefficients are preferred in consequential LCAs, where the effects on the system of increased use of the analyzed product or service is assessed (Yang, 2016). A prospective LCA can be either consequential or attributional (Arvidsson et al., 2018b), where an attributional prospective LCA provides a snapshot of the future given the evolution of foreground as well as background systems in line with scenario assumptions. A typical consequential prospective LCA would have the intention to go one step further in order to evaluate what impact a specific decision, in our case introduction of shared AVs, would have on each sub- system (Jones et al., 2017), e.g., vehicle manufacturing or electricity generation. This is considered to be beyond the scope of this study. Although, as clearly stated in Arvidsson et al. (2018a): "... an attributional LCA ... [is] effectively identical to a consequential LCA where only first- order (linear) physical flow consequences are considered – or "a consequential LCA based on the attributional [LCA] framework".  
+
+<|ref|>text<|/ref|><|det|>[[205, 280, 841, 405]]<|/det|>
+Hence, given the nature of the prospective LCA applied in this study, average coefficients (incl. carbon intensity of electricity) are preferred. Nevertheless, the main concern with applying average carbon intensity of electricity should still be addressed – the risk that changes in electricity demand in response to the analyzed question is large enough to affect the carbon intensity of electricity (Harmsen and Graus, 2013). To handle this risk, care has been taken to choose scenarios for background systems (incl. electricity generation) that match foreground systems. The scenarios for electricity generation referenced for Swedish and EU averages assume increased electricity demand following strong electrification not only in transportation but also in industry. The IEA scenarios (i.e., Stated Policies and Sustainable Development) both also account for increases in electricity demand, specifically looking at global trends within industry and electric vehicle deployment among others.  
+
+<|ref|>text<|/ref|><|det|>[[205, 415, 686, 429]]<|/det|>
+The following was added to the manuscript in response to this (lines 559- 561):  
+
+<|ref|>text<|/ref|><|det|>[[205, 440, 812, 465]]<|/det|>
+"The carbon intensities used for electricity represent averages for each respective geographic area following the attributional nature of the chosen prospective lifecycle assessment framework \(^{58,59}\) ."  
+
+<|ref|>text<|/ref|><|det|>[[150, 476, 847, 576]]<|/det|>
+2. It seems that AVs, ridesourcing, and/or car sharing may cause not only empty vehicle travel, but also additional induced demand (ie, the reduced costs or inconveniences of travel due to these additional options may lead to new trip generation). An interesting complement to the sensitivity analysis of breakeven points for deadheading would be a similar look at breakevens for percent of increased vehicle-distance traveled due to new travel demand. This may be out of scope for this manuscript, but it may not require any new analysis, but instead a re-interpretation of the existing breakeven results. If nothing else, it is worth mentioning in the discussion section that induced demand is a similar issue of concern (conceivably on the same order of magnitude as deadheading in the long term, e.g., if AVs lead to land use changes and drastic shifts in behavior).  
+
+<|ref|>text<|/ref|><|det|>[[184, 587, 830, 626]]<|/det|>
+Thanks for highlighting the important aspect of induced travel demand for AVs and car sharing and/or ride sharing. We do consider this to be outside of the scope of this particular study and mention this as one of the trends that will affect pathways for decarbonizing passenger car travel, see lines 344- 346:  
+
+<|ref|>text<|/ref|><|det|>[[184, 637, 835, 675]]<|/det|>
+"These trends will affect the pathways towards decarbonization of passenger car travel, including changes in charging patterns \(^{13}\) , cost structures \(^{9}\) and the value of travel time \(^{42 - 44}\) , which may induce additional travel activity \(^{45}\) and cause modal shifts \(^{46,47}\) ."  
+
+<|ref|>text<|/ref|><|det|>[[184, 686, 840, 712]]<|/det|>
+To further highlight the impact induced travel demand could have, we have added the following sentence in the discussion section (lines 368- 375):  
+
+<|ref|>text<|/ref|><|det|>[[184, 722, 841, 797]]<|/det|>
+"Note that induced travel by autonomous BEVs (both individually owned and shared) has not been assessed in the study. However, this risk is important to consider since the use of autonomous vehicles may substantially increase the travel demand. Autonomous vehicles may effectively reduce the value of travel time and the generalized travel cost \(^{45}\) since the driver does not need to be attentive and can instead use their time in the vehicle for whatever they find convenient. This potential increases in the travel demand could further increase the total carbon footprint for the fleet as a whole."  
+
+<|ref|>text<|/ref|><|det|>[[150, 808, 844, 871]]<|/det|>
+3. Enough observations were removed (>50%) that I would hesitate to characterize the exercise as data cleaning -- a more generic term such as filtering, subsetting, or preparation may be more appropriate. This may be pedantic, but given the "cleaning" term, I thought that only anomalous observations were being removed, but in this case, the goal is not simply to remove anomalies but also to create a dataset more representative and applicable to the analysis.  
+
+<|ref|>text<|/ref|><|det|>[[183, 881, 808, 896]]<|/det|>
+Thank you for this suggestion. The wording has been changed to "filtering" throughout the manuscript.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[152, 95, 843, 170]]<|/det|>
+4. Removal of 360,000 vehicles sitting unused for \(>14\) months prior to scrappage: I am not expert in scrappage issues, but removing this many observations (nearly as many as the size of the final dataset) seems to warrant a brief discussion of how we expect it to de-bias the dataset and how including it could have altered findings. (If sitting unused in a garage is how most cars ultimately transition towards scrappage, does excluding it alter findings? Similarly, if many Swedish cars find a second life in other countries but are not included in the data excerpt from the government, would that dataset bias allter findings?)  
+
+<|ref|>text<|/ref|><|det|>[[183, 181, 839, 305]]<|/det|>
+Thanks for raising this point. We have also reconsidered the other filtering categories and criteria used since we realized that we had been too conservative in the set criteria. We have decided to make the following adjustments: the time considered between first registration and the manufacturing year has been increased to two years instead of one (resulting in 107,430 additional vehicles), engine types include also ethanol and natural gas since they are internal combustion engines (15,268 additional vehicles), and the average distance traveled must not be greater than 600 km per day instead of 400 km per day (153 additional vehicles). Note that the number of additional vehicles may not sum to the total since one vehicle may fulfil more than one criterion. This enabled us to do a more even stratification of the data with equal size categories for driving intensity (between 0 and 100,000 km/year in 10,000 km/year steps). The adjustments resulted in a slightly higher elasticity of - 0.67 instead of - 0.59 (fitted to Normal distribution).  
+
+<|ref|>text<|/ref|><|det|>[[184, 316, 844, 379]]<|/det|>
+In response to your comment on the time between de- registration and last inspection, we have performed a sensitivity analysis. If the time is increased with two years (i.e., vehicles are removed if they sit unused for \(>38\) months), 88,568 vehicles are removed instead of 360,110 and the elasticity would be - 0.65 (fitted to Normal distribution). Hence, largely unaffected. The following statement was added to the manuscript (lines 408- 409):  
+
+<|ref|>text<|/ref|><|det|>[[184, 390, 787, 404]]<|/det|>
+"Criterion (ii) filters many observations but including them does not significantly impact the results."  
+
+<|ref|>text<|/ref|><|det|>[[183, 414, 841, 563]]<|/det|>
+Exports were limited 2014- 2016 - 8- 18 % of total annual de- registrations. For 2017, the share increased to 25% and from 2018 onwards it's been at levels of 35- 40%. The number of annually retired vehicles has been fairly constant over the same time period (equal to the time period for our data set - 2014- 2018). The exports have been distributed across all vehicle types, but the recent increase in export has been specifically in the age segment 0- 5 years. The Swedish governmental agency Traffic Analysis ties this to a recent policy that gives incentives for new car purchases and the leasing deals that lasts around 3 years (see https://www.trafa.se/vagtrafik/export av personbilar okade kraftigt 2018- 8201/, https://www.trafa.se/vagtrafik/fordon/export- av- personbilar- 2020- 12094/, both in Swedish). The fact that the number of retired vehicles has been fairly constant over the time period, also supports that the increase in exports are likely related to the purchasing prerequisites and should not have any major impact on the findings. It is more difficult to draw a conclusion for the exports during 2014- 2016, but the impact on the results should in any case be minor since they still are limited in number compared to the full dataset.  
+
+<|ref|>text<|/ref|><|det|>[[152, 574, 820, 600]]<|/det|>
+5. The methods section states that: "The annual driving distance, d, for year t, is assumed to decrease by \(b = 4.4\%\) per year". How was this value chosen and are results very sensitive to it?  
+
+<|ref|>text<|/ref|><|det|>[[184, 612, 845, 650]]<|/det|>
+Thank you highlighting that we omitted including the source. The value is estimated based on statistics on driving distances in Sweden from the governmental agency Transport Analysis. The source has been added to the list of references and the following is added in the Methods section (lines 532- 533):  
+
+<|ref|>text<|/ref|><|det|>[[183, 660, 841, 687]]<|/det|>
+"... the annual driving range is assumed to decrease by \(b = 4.4\%\) per year over its lifetime (estimated based on statistics on driving distances in Sweden \(^{57}\) )."  
+
+<|ref|>text<|/ref|><|det|>[[184, 698, 840, 736]]<|/det|>
+We have also included a sensitivity analysis of this assumption as part of the new fleet- wide analysis of the breakeven carbon footprint, which is included in Supplementary Figures 9- 10. The sensitivity analysis is summarized in the manuscript as follows:  
+
+<|ref|>text<|/ref|><|det|>[[184, 748, 277, 760]]<|/det|>
+Lines 238- 244:  
+
+<|ref|>text<|/ref|><|det|>[[184, 760, 844, 822]]<|/det|>
+"... Two other factors also contribute to the decreasing manufacturing- phase emissions as driving intensities increase: the assumed reduction in annual driving intensity for each individual car of 4.4% per year, and the time discretization of one year for the vehicle fleet turnover simulation. The significance of the former factor is tested in Supplementary Figure 9, showing slightly higher carbon footprint when driving intensity is assumed to be constant over each vehicle's lifetime."  
+
+<|ref|>text<|/ref|><|det|>[[184, 834, 277, 846]]<|/det|>
+Lines 330- 335:  
+
+<|ref|>text<|/ref|><|det|>[[184, 847, 835, 908]]<|/det|>
+"... The sensitivity of the assumed driving intensity decrease rate of 4.4% is also tested, showing higher breakeven levels for the additional empty travel with higher driving intensity decrease rates (i.e., when a larger share of the travel for one vehicle is concentrated to early years in the vehicle's lifetime). Nevertheless, the assumed elasticity in the lifetime- intensity model has a higher impact on the results than the assumed driving intensity decrease rate."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[150, 95, 835, 135]]<|/det|>
+6. I don't have a specific suggestion, but on plots where Normal distribution and Weibull distribution results are plotted as separate line types (ex: Fig. 4), I found it difficult to make out some of the dotted lines that were close together.  
+
+<|ref|>text<|/ref|><|det|>[[180, 144, 828, 172]]<|/det|>
+Thank you for highlighting that this graph is difficult to read. We have adjusted the linetype and opacity of the points to improve readability.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 103]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 120, 292, 163]]<|/det|>
+Reviewer #1: Remarks to the Author: Dear authors,  
+
+<|ref|>text<|/ref|><|det|>[[115, 179, 872, 209]]<|/det|>
+The response to my comments, and the corresponding revisions are very satisfying. I think the paper has been substantially improved. I have no further comments for the paper.  
+
+<|ref|>text<|/ref|><|det|>[[115, 254, 218, 267]]<|/det|>
+Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[115, 270, 292, 283]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 284, 877, 357]]<|/det|>
+The authors have responded comprehensively to my initial feedback. In particular, I appreciate the revisions to the carbon footprint and empty travel breakeven level analyses using a fleet turnover model, which I believe strengthened those sections. I also was glad to see the additional discussion of potential biases regarding BEV implications, the relaxation of some dataset filters, and the additional sensitivity analysis of driving distance decrease rate.  
+
+<|ref|>text<|/ref|><|det|>[[115, 357, 840, 386]]<|/det|>
+My substantive concerns have been addressed, and so I believe the manuscript is appropriate for publication.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1fdd6716183b848b4cb6460ca8ad882add822f660a1ffea2933c3b9322d7296b/images_list.json

@@ -0,0 +1,10 @@
+[
+  {
+    "type": "image",
+    "img_path": "images/Figure_unknown_0.jpg",
+    "caption": "Experiment 1",
+    "footnote": [],
+    "bbox": [],
+    "page_idx": 0
+  }
+]

peer_reviews/supplementary_0_Peer Review File__1fe0377a7d59da2f25d6882ed6c3993e12e578637cdd1bbe96209d247121411a/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1fe0377a7d59da2f25d6882ed6c3993e12e578637cdd1bbe96209d247121411a/supplementary_0_Peer Review File__1fe0377a7d59da2f25d6882ed6c3993e12e578637cdd1bbe96209d247121411a.mmd

@@ -0,0 +1,218 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+Broad Protection Against Clade 1 Sarbecoviruses After a Single Immunization with Cocktail Spike- Protein- Nanoparticle Vaccine  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #1:  
+
+Remarks to the Author:  
+
+This manuscript describes a novel multi- valent nanoparticle vaccine and shows that it confers protection in a hamster challenge. The vaccine is based off a platform of streptavidin coated VLPs that can combine with biotinylated spike proteins to assemble into spike- coated MS2 VLPs. Trivalent cocktails of these nanoparticles were examined in different challenge models. Overall the manuscript is well written and conveys the goals and successes of the research well. There are limitations inherent in the manuscript – only showing lung titers and not any other sublethal measures in the studies – but it is challenge with omicron variants in hamsters. I think it is suitable for publication in Nat. Comms. Following minor revision, including some important nanoparticle characterization experiments.  
+
+Major Comments:  
+
+1. While some VLP platforms are not amenable to mosaics, it is confusing why the authors did not consider a mosaic vaccine with this platform. Perhaps in the discussion the reasoning could be explained. If it was tested and worse, that result should be added as it would be incredibly interesting to the field.  
+
+2. I agree that the "diameter was measured to be \(90 - 100 \text{nm}\) , consistent with prior characterization," according to the previous paper:  
+
+Chiba, S., Frey, S.J., Halfmann, P.J. et al. Multivalent nanoparticle-based vaccines protect hamsters against SARS-CoV-2 after a single immunization. Commun Biol 4, 597 (2021). https://doi.org/10.1038/s42003-021-02128-8  
+
+But I would consider an additional analysis to confirm particle integrity. Perhaps TEM, as was done previously. This could help rule out that the spike component on some fraction was unfolded. The limited data – DLS and ELISA, are both crude methods, particularly the way the ELISA was done, so in my opinion this is the aspect of the manuscript that needs the most improvement.  
+
+3. The authors could do a better job citing other COVID nanoparticle vaccines, which seems like an oversight given the platform. Most of the introduction is spent talking in length about current bivalent vaccines, but if these were efficacious we wouldn't need this manuscript. I would consider shortening this section and describing, to some extent, nanoparticle COVID vaccines. Potential additional references:  
+
+a. M. Gordon Joyce et al., vA SARS-CoV-2 ferritin nanoparticle vaccine elicits protective immune responses in nonhuman primates. Sci. Transl. Med.14, eabi5735(2022).DOI:10.1126/scitranslmed.abi5735 
+b. Weidenbacher, P.AB., Sanyal, M., Friedland, N. et al. A ferritin-based COVID-19 nanoparticle vaccine that elicits robust, durable, broad-spectrum neutralizing antisera in non-human primates. Nat Commun 14, 2149 (2023). https://doi.org/10.1038/s41467-023-37417-9 
+c. Brouwer PJM, Et. al, Two-component spike nanoparticle vaccine protects macaques from SARS-CoV-2 infection. Cell. 2021 Mar 4;184(5):1188-1200. e19. doi: 10.1016/j.cell.2021.01.035. Epub 2021 Jan 26. PMID: 33577765; PMCID: PMC7834972. 
+d. Etc.  
+
+4. I find the Pfizer comparator to be not informative and confusing. Without a control indicating this vaccine retained potency, it's difficult to understand why it shows no detectable efficacy? There is language in the text hypothesizing what happened, but if this serum is available, I would test it for binding (hopefully more sensitive than neutralization) to demonstrate the animals were properly immunized – or provide a citation that a 10ug dose doesn't confer protection, neutralization, or any apparent activity in hamsters. In Fig 4c there is protection with no neutralization – so to not show any protection suggests, in my mind, the procured vaccine was ineffective.
+
+<--- Page Split --->
+
+5. The result that "The bivalent vaccine provided complete protection against the XBB.1 challenge despite eliciting significantly lower neutralizing antibody titers against XBB.1 than the trivalent vaccines." Should be further discussed. Does this data argue against the inclusion of an omicron variant? Does this suggest neutralization is a bad surrogate for protection?  
+
+6. Given how similar all the ELISAs look, negative controls should be shown for the ELISAs – ideally the VLP alone binding to Ab control.  
+
+## Minor Comments:  
+
+- Please use fewer significant digits in Fig. 3 top (278616 = 2.8\*105 or 2.8e5)- Please provide the dose of aluminum in the alhydrogel adjuvanted samples (as opposed to "equal volume").- I like the analysis in Figure 3c- "While a naive animal might respond equally to all components of a vaccine, an animal with pre-existing immunity might instead be biased towards one component to which it has been exposed to before." – please add a citation for original antigenic sin/immunological imprinting.- Methods around "HiScreen Capto Core" need elaboration. My understanding was that these columns did not bind to large nanoparticles, so the discussion around elution in just 20mM tris is surprising? Did you collect the entire flow through?- ELISA methods should include coating concentration and Ab concentration – methods as written could not be replicated. "The final concentration of ACE2-Fc and antibodies was such that there would be one molecule per S protein trimer." Is confusing, and assumes 100% bound to the plate? But we know the vast majority of protein remains unbound to the ELISA plate.  
+
+## Reviewer #3:  
+
+Remarks to the Author:  
+
+The study by Halfmann et al evaluates the use of a Spike based nanoparticle vaccine to produce broad protection against Sarbecoviruses. This model uses VLP composed of hexapro Spike proteins linked to the MS2 phage protein to produce a nanoparticle. This vaccine had previously been used in a variety of studies based on the D614G Spike sequence. In this study, the authors have tested the immunogenicity and cross neutralization capacity of a range of Sarbecovirus Spike proteins to identify the minimal combination that can be used together and still produce broadened neutralization and protection in a hamster model of coronavirus infection.  
+
+The authors show that Spike protein nanoparticles from a variety of SARS2 viruses, SARS1 and SHC014 are able to neutralize some but not all of the matched strains. They then chose combination of either D614D/SHC014 or 2 other combinations adding in a thirs Spike protein from BA.2.75.2 or XBB. In hamster vaccination models, the combination of D614G/SHC/XBB produced broadly neutralizing antibody and protection in WT hamsters challenged with XBB.1 and in hACE2 transgenic hamsters challenged with WIV1 or SHC014. No virus was detected in lungs of either of these challenged hamsters demonstrating robust protection from a single vaccine dose. As a control the Pfizer/Biontech vaccine was used at a single dose of 19ug, and it failed to produce neutralizing antibody as well as protect in hamsters. The study is well performed and the addition of the hACE2 hamster models add to the robustness of the data to show broader protection from challenge.  
+
+Questions about the study are below  
+
+1. While not needed in this manuscript, I would be interested to know if lower protein concentrations have been tested in other hamster vaccine experiments since submission. 15ug of each nanoparticle (45ug in all) is a large dose. The data suggest that much lower concentrations should be as effective. Have then been tested?
+
+<--- Page Split --->
+
+2. Mucosal immunity is a major driver of protective responses in humans, and while not as dramatic in hamster models, it would be intriguing to know if neutralizing antibody levels were tested in the vaccinated hamsters. Also were nasal washes analyzed for virus titer after challenge?  
+
+Several interesting points are broad up by the data that should be commented on in the discussion: 1. Is there a structural basis for SHC014 Spike to neutralize better against WIV1 than SARS1? Has that been seen in other vaccine models or is this unique to this nanoparticle design? 2. A separate discussion is warranted in this manuscript for explaining implications of this work across the context of broadly protective vaccines. Important points to touch on are 1) mucosal vs systemic immunity of this vaccine platform, 2) structural basis for the neutralization and protection with emphasis on other widely divergent bat coronaviruses and how they may fair against this vaccine based on other published structural data, 3) longevity of the response and how this vaccine design could be used commercially.
+
+<--- Page Split --->
+
+Response to Reviewers:  
+
+## Reviewer #1  
+
+Reviewer 1 noted that "Overall the manuscript is well written and conveys the goals and successes of the research well. There are limitations inherent in the manuscript - only showing lung titers and not any other sublethal measures in the studies - but it is challenge with omicron variants in hamsters. I think it is suitable for publication in Nat. Comms. Following minor revision, including some important nanoparticle characterization experiments."  
+
+We thank the reviewer and have responded to the other comments below. We have also added nasal turbinate virus titers as an additional sublethal measure in response to the reviewer's comment.  
+
+Comment: While some VLP platforms are not amenable to mosaics, it is confusing why the authors did not consider a mosaic vaccine with this platform. Perhaps in the discussion the reasoning could be explained. If it was tested and worse, that result should be added as it would be incredibly interesting to the field.  
+
+Response: We did not test mosaic vaccines as part of the current work and have added the following text to the discussion to explain the reasoning as suggested by the reviewer:  
+
+"We did not test mosaic vaccines as part of the current work. While the underlying concept is interesting, a recent study reported a head- to- head comparison between cocktail and mosaic approaches for spike protein- nanoparticle vaccines and did not identify a significant difference in neutralization titers between the two approaches.24 A cocktail vaccine is also aligned with the hypothesis motivating this work - that if the mixture of antigens comprising the cocktail is carefully selected by characterizing the antigenic landscape, the antibody response elicited by each component would collectively result in a broadly protective polyclonal antibody response. It would be interesting to revisit the mosaic idea in future work."  
+
+Comment: I agree that the "diameter was measured to be \(90 - 100 \text{nm}\) , consistent with prior characterization," according to the previous paper:  
+
+Chiba, S., Frey, S.J., Halfmann, P.J. et al. Multivalent nanoparticle- based vaccines protect hamsters against SARS- CoV- 2 after a single immunization. Commun Biol 4, 597 (2021). https://doi.org/10.1038/s42003- 021- 02128- 8  
+
+But I would consider an additional analysis to confirm particle integrity. Perhaps TEM, as was done previously. This could help rule out that the spike component on some fraction was unfolded. The limited data - DLS and ELISA, are both crude methods, particularly the way the ELISA was done, so in my opinion this is the aspect of the manuscript that needs the most improvement.  
+
+Response: We have added TEM characterization as requested by the reviewer as Figure 2f in the revised manuscript.  
+
+Comment: The authors could do a better job citing other COVID nanoparticle vaccines, which seems like an oversight given the platform. Most of the introduction is spent talking in length about current bivalent vaccines, but if these were efficacious we wouldn't need this manuscript. I would consider shortening this section and describing, to some extent, nanoparticle COVID vaccines. Potential additional references:
+
+<--- Page Split --->
+
+a. M. Gordon Joyce et al., A SARS-CoV-2 ferritin nanoparticle vaccine elicits protective immune responses in nonhuman primates. Sci. Transl. Med. 14, eabi5735(2022). DOI:10.1126/scitranslmed.abi5735 
+b. Weidenbacher, P.AB., Sanyal, M., Friedland, N. et al. A ferritin-based COVID-19 nanoparticle vaccine that elicits robust, durable, broad-spectrum neutralizing antisera in non-human primates. Nat Commun 14, 2149 (2023). https://doi.org/10.1038/s41467-023-37417-9 
+c. Brouwer PJM, Et. al, Two-component spike nanoparticle vaccine protects macaques from SARS-CoV-2 infection. Cell. 2021 Mar 4;184(5):1188-1200.e19. doi: 10.1016/j.cell.2021.01.035. Epub 2021 Jan 26. PMID: 33577765; PMCID: PMC7834972.  
+
+## Response:  
+
+We have cited these manuscripts and added the following text to the introduction based on the reviewer's suggestions:  
+
+"Protein nanoparticles have emerged as attractive platforms for the display of S protein antigens. Brouwer et al. generated two component protein nanoparticles displaying stabilized perfusion SARS-CoV-2 S proteins that protected vaccinated macaques from a challenge with SARS- CoV- 2.25 Joyce et al. showed that adjuvanted SARS- CoV- 2 S protein-ferritin nanoparticle vaccines protected non- human primates from a challenge with SARS- CoV- 2.26 Weidenbacher et al. reported that adjuvanted ferritin nanoparticle vaccines displaying a truncated form of the SARS- CoV- 2 S protein ectodomain elicited a broad neutralizing antibody response in non- human primates.27 Hutchinson et al. designed self-assembling protein nanoparticles displaying multiple S protein antigens that protected mice from a challenge with MERS- CoV.28 As described above, Brinkkemper et al. designed nanoparticles presenting mixtures of the SARS- CoV- 1 and SARS- CoV- 2 S proteins.24  
+
+Comment: I find the Pfizer comparator to be not informative and confusing. Without a control indicating this vaccine retained potency, it's difficult to understand why it shows no detectable efficacy? There is language in the text hypothesizing what happened, but if this serum is available, I would test it for binding (hopefully more sensitive than neutralization) to demonstrate the animals were properly immunized - or provide a citation that a 10ug dose doesn't confer protection, neutralization, or any apparent activity in hamsters. In Fig 4c there is protection with no neutralization - so to not show any protection suggests, in my mind, the procured vaccine was ineffective.  
+
+Response: We agree with the reviewer's concerns regarding the potency of a \(10\mu \mathrm{g}\) dose of the Pfizer- BioNTech vaccine and have repeated the experiment with a higher dose (30 \(\mu \mathrm{g}\) ). This was a sufficient dose to result in a significant decrease in lung titers following an XBB.1 challenge, and the higher dose did elicit neutralizing antibodies against 614G, BA.5, and in one case against XBB.1. We have updated the text to include these results.  
+
+Regarding neutralization:  
+
+"To compare, hamsters immunized with 30 \(\mu \mathrm{g}\) of the Pfizer- BioNTech bivalent vaccine did produce neutralizing antibodies against an early SARS- CoV- 2 isolate (S- 614G) and BA.5 as expected given the composition of the vaccine. Like our bivalent 614D/SHC014 vaccine, neutralization titers were significantly reduced against XBB.1 (3 out of 4 hamsters had no detectable titers), but unlike any of our bivalent or trivalent vaccines, no detectable neutralization titers were observed against the Clade 1A bat CoVs SHC014 and WIV1. These results further support the inclusion of antigens from both XBB- like and Clade 1A CoVs in a broadly protective cocktail vaccine."  
+
+Regarding protection from challenge
+
+<--- Page Split --->
+
+"Meanwhile, a single immunization with 30 \(\mu g\) of the Pfizer- BioNTech bivalent vaccine significantly decreased lung and nasal titers after challenge with XBB.1 (Fig. 4c, e)."  
+
+Comment: The result that "The bivalent vaccine provided complete protection against the XBB.1 challenge despite eliciting significantly lower neutralizing antibody titers against XBB.1 than the trivalent vaccines." Should be further discussed. Does this data argue against the inclusion of an omicron variant? Does this suggest neutralization is a bad surrogate for protection?  
+
+Response: We have added the following discussion to the revised manuscript as suggested by the reviewer:  
+
+"While neutralizing antibody titers against SARS- CoV- 2 have been identified as a correlate of protection against symptomatic disease \(^{8,39}\) , even low neutralizing titers can be sufficient for protection. High neutralization titers elicited by vaccination are still desirable, since they are likely to provide greater protection against viral escape; neutralization titers against viral variants are generally lower than those against the vaccine strain. The inclusion of an Omicron antigen in the cocktail is thus justified, particularly in light of the continuing emergence of new Omicron variants."  
+
+Comment: Given how similar all the ELISAs look, negative controls should be shown for the ELISAs – ideally the VLP alone binding to Ab control.  
+
+Response: We have added VLP- only controls to Figure 2 and Supplementary Figure 2b.  
+
+We have also made changes in response to the reviewer's other minor comments:  
+
+Comment: Please use fewer significant digits in Fig. 3 top (278616 = 2.8\*105 or 2.8e5) Response: We thank the reviewer for the recommendation and have updated Figure 3 accordingly.  
+
+Comment: Please provide the dose of aluminum in the alhydrogel adjuvanted samples (as opposed to "equal volume").  
+
+Response: We have added the dose of aluminum in the alhydrogel samples as requested.  
+
+Comment: I like the analysis in Figure 3c  
+
+Response: We thank the reviewer for this encouraging comment.  
+
+Comment: "While a naive animal might respond equally to all components of a vaccine, an animal with pre- existing immunity might instead be biased towards one component to which it has been exposed to before." – please add a citation for original antigenic sin/immunological imprinting.  
+
+Response: We have added citations for "original antigenic sin" as suggested by the reviewer.  
+
+- Francis Jr., T. On the Doctrine of Original Antigenic Sin Proc Am Phil Soc 104, 572-578 (1960)- de St. Groth, F. & Webster, R.G. Disquisitions of Original Antigenic Sin. I. Evidence in man. J Exp Med 124, 331-345 (1966)- de St. Groth, F. & Webster, R.G. Disquisitions on Original Antigenic Sin. II. Proof in lower creatures. J Exp Med 124, 347-361 (1966).  
+
+Comment: Methods around "HiScreen Capto Core" need elaboration. My understanding was that these columns did not bind to large nanoparticles, so the discussion around elution in just 20mM tris is surprising? Did you collect the entire flow through?
+
+<--- Page Split --->
+
+Response: We have updated our description of MS2 purification using the HiScreen Capto Core columns. The revised text reads:  
+
+"25 mL of the lysate was loaded onto four HiScreen Capto Core columns (Cytvia) in series using an Akta Start system. MS2 was purified by washing with \(20mM\) Tris Base for \(\sim 5\) column volumes (CVs). The entire flowthrough was collected in \(1.8mL\) fractions, and SDS- PAGE was used to determine purity and yield. Fractions 7- 12 were typically those most enriched in MS2 while smaller impurities would concentrate in later fractions due to the CaptoCore system's size- exclusion character."  
+
+Comment: ELISA methods should include coating concentration and Ab concentration - methods as written could not be replicated. "The final concentration of ACE2- Fc and antibodies was such that there would be one molecule per S protein trimer." Is confusing, and assumes \(100\%\) bound to the plate? But we know the vast majority of protein remains unbound to the ELISA plate.  
+
+Response: We have rewritten the ELISA methods as follows:  
+
+"VLP- S and S protein were diluted in PBS such that the concentration of S protein in each solution was \(1\mu g / mL\) . \(100\mu L\) (0.1 \(\mu g\) of S protein) per well of the diluted protein was then coated onto a Nunc Maxisorp 96- well plate. After a 1- hour incubation, the protein solutions were discarded, and the wells were blocked with \(200\mu L\) of \(5\%\) BSA (EMD Millipore) in PBST (0.05% Tween- 20) for 1 hour. The wells were then washed three times with PBST. Stock solutions of primary antibodies at \(0.3mg / mL\) , \(3.4mg / mL\) , \(3.8mg / mL\) , and \(1.9mg / mL\) (by BCA) for ACE2- Fc, CR3022, S309, and S2P6 respectively were diluted 9:1500, 1:1500, 1:1500, and 1:30000 respectively in PBST with \(1\%\) BSA, and \(100\mu L\) per well of these diluted primary antibodies were added. After 1 hour, the wells were washed three times with PBST, and \(100\mu L\) of horseradish peroxidase- conjugated anti- human IgG Fc goat antibody (MP Biomedical, 1:5000) diluted in PBST with \(1\%\) BSA was added to all wells. The wells were washed three times with PBST after a 1- hour incubation. \(100\mu L\) of TMB (Thermo Scientific) were added to each well and allowed to develop for 3 minutes. The reaction was stopped with \(160mM\) sulfuric acid, and the absorbance at \(450nm\) was read with a Spectramax i3x plate reader (Molecular Devices) and Gen5 2.07 software (BioTek)."  
+
+## Reviewer #2  
+
+Reviewer 2 noted that "The study is well performed and the addition of the hACE2 hamster models add to the robustness of the data to show broader protection from challenge." We thank the reviewer and have responded to the reviewer's other comments below.  
+
+Comment: While not needed in this manuscript, I would be interested to know if lower protein concentrations have been tested in other hamster vaccine experiments since submission. 15ug of each nanoparticle (45ug in all) is a large dose. The data suggest that much lower concentrations should be as effective. Have then been tested?  
+
+Response: We have not yet tested reducing the dose of our vaccine cocktails. We hypothesize that it would be possible to reduce the dose though, since our trivalent cocktails elicit robust and broad neutralizing antibody titers that are likely to be well above the threshold needed for protection.  
+
+Comment: Mucosal immunity is a major driver of protective responses in humans, and while not as dramatic in hamster models, it would be intriguing to know if neutralizing antibody levels were
+
+<--- Page Split --->
+
+tested in the vaccinated hamsters. Also were nasal washes analyzed for virus titer after challenge?  
+
+Response: We have added data for nasal turbinate titers in Figures 3c, 4d- e, and 5b. Discussion of these nasal turbinate titers has also been added.  
+
+Regarding nasal titers after monovalent vaccination and challenge with BA.5 (Figure 3c): "Similar trends were observed in viral titers in the nasal turbinates as VLP- 614D- S and all of the Omicron VLP- S vaccines significantly reduced virus levels, while hamsters vaccinated with VLP- SHC014- S and VLP- SARS- CoV- 1- S had viral levels comparable to those of the control hamsters (Fig. 3c)."  
+
+Regarding nasal titers after cocktail vaccination and challenge with BA.5 and XBB.1 (Figure 4d): "Similar trends were also seen for nasal titers. The VLP- 614D- S and VLP- SHC014- S bivalent cocktail significantly reduced nasal virus titers, while hamsters immunized with either trivalent cocktail had undetectable nasal titers (Fig 4d)."  
+
+Regarding nasal titers after Pfizer- BioNTech bivalent vaccination and challenge with XBB.1 (Figure 4e):  
+
+"Meanwhile, a single immunization with 30 μg of the Pfizer- BioNTech bivalent vaccine significantly decreased lung and nasal titers after challenge with XBB.1 (Fig. 4c,e)."  
+
+Regarding nasal titers after cocktail vaccination in hACE2 hamsters and challenge with WIV1 and SHC014 (Figure 5b):  
+
+"Nasal turbinate titers were also significantly reduced in vaccinated transgenic hamsters in both cases (Fig. 5b), though the magnitude of this reduction was weaker than for the Clade 1B viruses."  
+
+Comment: Several interesting points are brought up by the data that should be commented on in the discussion:  
+
+1. Is there a structural basis for SHC014 Spike to neutralize better against WIV1 than SARS1? Has that been seen in other vaccine models or is this unique to this nanoparticle design?  
+
+Response: We thank the reviewer for this comment; as seen in Fig. 1c, the WIV1 S is in fact more homologous to the SHC014 S than to the SARS-CoV- 1 S. We have corrected this text in the revised manuscript, which now reads  
+
+"VLP- SHC014- S elicited higher neutralizing antibody titers against WIV1 compared to VLP- SARS- CoV- 1- S, consistent with the WIV1 S being more homologous to the SHC014 S than to the SARS- CoV- 1 S (Fig. 1c)."  
+
+Comment: 2. A separate discussion is warranted in this manuscript for explaining implications of this work across the context of broadly protective vaccines. Important points to touch on are 1) mucosal vs systemic immunity of this vaccine platform, 2) structural basis for the neutralization and protection with emphasis on other widely divergent bat coronaviruses and how they may fair against this vaccine based on other published structural data, 3) longevity of the response and how this vaccine design could be used commercially.  
+
+Response: We have added the following text to the discussion based on the reviewer's comments:
+
+<--- Page Split --->
+
+"While we are encouraged by these results, there are several additional avenues that would be interesting to explore in future work. Enhancing mucosal immunity might not only enhance protection against viral infection, but also decrease viral transmission. \(^{40}\) Intranasal vaccination against SARS- CoV- 2 has been explored with several platforms, including mRNA- lipid nanoparticles \(^{41}\) , nanoparticles displaying the RBD \(^{42}\) , live attenuated influenza viruses also encoding the RBD \(^{43}\) , adenovirus- vectored vaccines \(^{44}\) , and by using an intranasal boost with the unadjuvanted spike protein \(^{40}\) . The adaptation of our platform for intranasal delivery could be a promising avenue for improving the mucosal response. Characterizing the longevity of protection would also be an interesting avenue for future research. It would be particularly interesting to determine whether stronger mucosal immunity results in more durable protection against symptomatic disease.  
+
+Secondly, while the focus of this paper was on protecting against viruses in Clades 1A and 1B, extending protection to further sarbecovirus clades would be interesting to explore. Significant differences in their receptor binding domains are the primary basis for virus classification into clades, with Clade 2 viruses being unable to use ACE2 as an entry receptor and Clade 3 & 4 viruses harboring one deletion relative to Clade 1. \(^{45}\) Nevertheless, some Clade 2 viruses (a proposed "Clade 2A") \(^{46}\) and several Clade 3 viruses \(^{47}\) may be capable of infecting human cells, albeit in some cases only with exogenous protease treatment. RBD- and NTD- focused humoral immunity is unlikely to be cross- reactive between clades, as inter- clade RBD and NTD amino acid identity percentages are approximately 65- 75% and 45- 55%, respectively. As such, the inclusion of additional antigens to cover these clades may be necessary. Using the approach described in this work, antigenic cartography should help us identify the minimal mixture of antigens required to elicit broad protection against sarbecoviruses.  
+
+Given that protein- nanoparticle vaccines have been approved for clinical use \(^{48}\) , commercialization of this vaccine platform could be possible in the future. Moreover, the optimal components for a cocktail vaccine that are suggested by antigenic cartography may be platform- agnostic and could therefore be applied not just to other protein nanoparticle platforms, but also to other modalities such as mRNA- based vaccines. Indeed, the selection of an XBB strain (XBB.1.5) for incorporation in the recently approved monovalent vaccine is consistent with our results and it would be interesting to explore the protective efficacy of trivalent mRNA vaccines based on the compositions identified in this work against clade 1 sarbecoviruses."
+
+<--- Page Split --->
+
+Reviewers' Comments:  
+
+Reviewer #3:  Remarks to the Author:  The changes to the manuscript are sufficient to answer my questions.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1fe0377a7d59da2f25d6882ed6c3993e12e578637cdd1bbe96209d247121411a/supplementary_0_Peer Review File__1fe0377a7d59da2f25d6882ed6c3993e12e578637cdd1bbe96209d247121411a_det.mmd

@@ -0,0 +1,300 @@
+<|ref|>title<|/ref|><|det|>[[100, 40, 508, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[106, 110, 373, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>text<|/ref|><|det|>[[106, 154, 890, 240]]<|/det|>
+Broad Protection Against Clade 1 Sarbecoviruses After a Single Immunization with Cocktail Spike- Protein- Nanoparticle Vaccine  
+
+<|ref|>image<|/ref|><|det|>[[95, 732, 262, 780]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[271, 732, 877, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[116, 90, 286, 103]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[116, 119, 216, 132]]<|/det|>
+Reviewer #1:  
+
+<|ref|>text<|/ref|><|det|>[[116, 135, 290, 148]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 149, 882, 270]]<|/det|>
+This manuscript describes a novel multi- valent nanoparticle vaccine and shows that it confers protection in a hamster challenge. The vaccine is based off a platform of streptavidin coated VLPs that can combine with biotinylated spike proteins to assemble into spike- coated MS2 VLPs. Trivalent cocktails of these nanoparticles were examined in different challenge models. Overall the manuscript is well written and conveys the goals and successes of the research well. There are limitations inherent in the manuscript – only showing lung titers and not any other sublethal measures in the studies – but it is challenge with omicron variants in hamsters. I think it is suitable for publication in Nat. Comms. Following minor revision, including some important nanoparticle characterization experiments.  
+
+<|ref|>text<|/ref|><|det|>[[116, 284, 248, 298]]<|/det|>
+Major Comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 313, 875, 373]]<|/det|>
+1. While some VLP platforms are not amenable to mosaics, it is confusing why the authors did not consider a mosaic vaccine with this platform. Perhaps in the discussion the reasoning could be explained. If it was tested and worse, that result should be added as it would be incredibly interesting to the field.  
+
+<|ref|>text<|/ref|><|det|>[[115, 387, 750, 416]]<|/det|>
+2. I agree that the "diameter was measured to be \(90 - 100 \text{nm}\) , consistent with prior characterization," according to the previous paper:  
+
+<|ref|>text<|/ref|><|det|>[[115, 416, 857, 460]]<|/det|>
+Chiba, S., Frey, S.J., Halfmann, P.J. et al. Multivalent nanoparticle-based vaccines protect hamsters against SARS-CoV-2 after a single immunization. Commun Biol 4, 597 (2021). https://doi.org/10.1038/s42003-021-02128-8  
+
+<|ref|>text<|/ref|><|det|>[[115, 460, 880, 520]]<|/det|>
+But I would consider an additional analysis to confirm particle integrity. Perhaps TEM, as was done previously. This could help rule out that the spike component on some fraction was unfolded. The limited data – DLS and ELISA, are both crude methods, particularly the way the ELISA was done, so in my opinion this is the aspect of the manuscript that needs the most improvement.  
+
+<|ref|>text<|/ref|><|det|>[[115, 535, 875, 609]]<|/det|>
+3. The authors could do a better job citing other COVID nanoparticle vaccines, which seems like an oversight given the platform. Most of the introduction is spent talking in length about current bivalent vaccines, but if these were efficacious we wouldn't need this manuscript. I would consider shortening this section and describing, to some extent, nanoparticle COVID vaccines. Potential additional references:  
+
+<|ref|>text<|/ref|><|det|>[[115, 610, 875, 760]]<|/det|>
+a. M. Gordon Joyce et al., vA SARS-CoV-2 ferritin nanoparticle vaccine elicits protective immune responses in nonhuman primates. Sci. Transl. Med.14, eabi5735(2022).DOI:10.1126/scitranslmed.abi5735 
+b. Weidenbacher, P.AB., Sanyal, M., Friedland, N. et al. A ferritin-based COVID-19 nanoparticle vaccine that elicits robust, durable, broad-spectrum neutralizing antisera in non-human primates. Nat Commun 14, 2149 (2023). https://doi.org/10.1038/s41467-023-37417-9 
+c. Brouwer PJM, Et. al, Two-component spike nanoparticle vaccine protects macaques from SARS-CoV-2 infection. Cell. 2021 Mar 4;184(5):1188-1200. e19. doi: 10.1016/j.cell.2021.01.035. Epub 2021 Jan 26. PMID: 33577765; PMCID: PMC7834972. 
+d. Etc.  
+
+<|ref|>text<|/ref|><|det|>[[115, 775, 876, 878]]<|/det|>
+4. I find the Pfizer comparator to be not informative and confusing. Without a control indicating this vaccine retained potency, it's difficult to understand why it shows no detectable efficacy? There is language in the text hypothesizing what happened, but if this serum is available, I would test it for binding (hopefully more sensitive than neutralization) to demonstrate the animals were properly immunized – or provide a citation that a 10ug dose doesn't confer protection, neutralization, or any apparent activity in hamsters. In Fig 4c there is protection with no neutralization – so to not show any protection suggests, in my mind, the procured vaccine was ineffective.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 844, 150]]<|/det|>
+5. The result that "The bivalent vaccine provided complete protection against the XBB.1 challenge despite eliciting significantly lower neutralizing antibody titers against XBB.1 than the trivalent vaccines." Should be further discussed. Does this data argue against the inclusion of an omicron variant? Does this suggest neutralization is a bad surrogate for protection?  
+
+<|ref|>text<|/ref|><|det|>[[115, 163, 857, 194]]<|/det|>
+6. Given how similar all the ELISAs look, negative controls should be shown for the ELISAs – ideally the VLP alone binding to Ab control.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 223, 248, 237]]<|/det|>
+## Minor Comments:  
+
+<|ref|>text<|/ref|><|det|>[[113, 252, 875, 463]]<|/det|>
+- Please use fewer significant digits in Fig. 3 top (278616 = 2.8\*105 or 2.8e5)- Please provide the dose of aluminum in the alhydrogel adjuvanted samples (as opposed to "equal volume").- I like the analysis in Figure 3c- "While a naive animal might respond equally to all components of a vaccine, an animal with pre-existing immunity might instead be biased towards one component to which it has been exposed to before." – please add a citation for original antigenic sin/immunological imprinting.- Methods around "HiScreen Capto Core" need elaboration. My understanding was that these columns did not bind to large nanoparticles, so the discussion around elution in just 20mM tris is surprising? Did you collect the entire flow through?- ELISA methods should include coating concentration and Ab concentration – methods as written could not be replicated. "The final concentration of ACE2-Fc and antibodies was such that there would be one molecule per S protein trimer." Is confusing, and assumes 100% bound to the plate? But we know the vast majority of protein remains unbound to the ELISA plate.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 506, 216, 520]]<|/det|>
+## Reviewer #3:  
+
+<|ref|>text<|/ref|><|det|>[[115, 522, 291, 536]]<|/det|>
+Remarks to the Author:  
+
+<|ref|>text<|/ref|><|det|>[[115, 536, 878, 640]]<|/det|>
+The study by Halfmann et al evaluates the use of a Spike based nanoparticle vaccine to produce broad protection against Sarbecoviruses. This model uses VLP composed of hexapro Spike proteins linked to the MS2 phage protein to produce a nanoparticle. This vaccine had previously been used in a variety of studies based on the D614G Spike sequence. In this study, the authors have tested the immunogenicity and cross neutralization capacity of a range of Sarbecovirus Spike proteins to identify the minimal combination that can be used together and still produce broadened neutralization and protection in a hamster model of coronavirus infection.  
+
+<|ref|>text<|/ref|><|det|>[[115, 655, 865, 804]]<|/det|>
+The authors show that Spike protein nanoparticles from a variety of SARS2 viruses, SARS1 and SHC014 are able to neutralize some but not all of the matched strains. They then chose combination of either D614D/SHC014 or 2 other combinations adding in a thirs Spike protein from BA.2.75.2 or XBB. In hamster vaccination models, the combination of D614G/SHC/XBB produced broadly neutralizing antibody and protection in WT hamsters challenged with XBB.1 and in hACE2 transgenic hamsters challenged with WIV1 or SHC014. No virus was detected in lungs of either of these challenged hamsters demonstrating robust protection from a single vaccine dose. As a control the Pfizer/Biontech vaccine was used at a single dose of 19ug, and it failed to produce neutralizing antibody as well as protect in hamsters. The study is well performed and the addition of the hACE2 hamster models add to the robustness of the data to show broader protection from challenge.  
+
+<|ref|>text<|/ref|><|det|>[[116, 819, 390, 833]]<|/det|>
+Questions about the study are below  
+
+<|ref|>text<|/ref|><|det|>[[115, 834, 870, 893]]<|/det|>
+1. While not needed in this manuscript, I would be interested to know if lower protein concentrations have been tested in other hamster vaccine experiments since submission. 15ug of each nanoparticle (45ug in all) is a large dose. The data suggest that much lower concentrations should be as effective. Have then been tested?
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 879, 134]]<|/det|>
+2. Mucosal immunity is a major driver of protective responses in humans, and while not as dramatic in hamster models, it would be intriguing to know if neutralizing antibody levels were tested in the vaccinated hamsters. Also were nasal washes analyzed for virus titer after challenge?  
+
+<|ref|>text<|/ref|><|det|>[[115, 148, 876, 283]]<|/det|>
+Several interesting points are broad up by the data that should be commented on in the discussion: 1. Is there a structural basis for SHC014 Spike to neutralize better against WIV1 than SARS1? Has that been seen in other vaccine models or is this unique to this nanoparticle design? 2. A separate discussion is warranted in this manuscript for explaining implications of this work across the context of broadly protective vaccines. Important points to touch on are 1) mucosal vs systemic immunity of this vaccine platform, 2) structural basis for the neutralization and protection with emphasis on other widely divergent bat coronaviruses and how they may fair against this vaccine based on other published structural data, 3) longevity of the response and how this vaccine design could be used commercially.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 312, 106]]<|/det|>
+Response to Reviewers:  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 108, 220, 123]]<|/det|>
+## Reviewer #1  
+
+<|ref|>text<|/ref|><|det|>[[115, 124, 883, 210]]<|/det|>
+Reviewer 1 noted that "Overall the manuscript is well written and conveys the goals and successes of the research well. There are limitations inherent in the manuscript - only showing lung titers and not any other sublethal measures in the studies - but it is challenge with omicron variants in hamsters. I think it is suitable for publication in Nat. Comms. Following minor revision, including some important nanoparticle characterization experiments."  
+
+<|ref|>text<|/ref|><|det|>[[115, 210, 883, 261]]<|/det|>
+We thank the reviewer and have responded to the other comments below. We have also added nasal turbinate virus titers as an additional sublethal measure in response to the reviewer's comment.  
+
+<|ref|>text<|/ref|><|det|>[[115, 279, 883, 348]]<|/det|>
+Comment: While some VLP platforms are not amenable to mosaics, it is confusing why the authors did not consider a mosaic vaccine with this platform. Perhaps in the discussion the reasoning could be explained. If it was tested and worse, that result should be added as it would be incredibly interesting to the field.  
+
+<|ref|>text<|/ref|><|det|>[[115, 348, 883, 382]]<|/det|>
+Response: We did not test mosaic vaccines as part of the current work and have added the following text to the discussion to explain the reasoning as suggested by the reviewer:  
+
+<|ref|>text<|/ref|><|det|>[[115, 399, 883, 537]]<|/det|>
+"We did not test mosaic vaccines as part of the current work. While the underlying concept is interesting, a recent study reported a head- to- head comparison between cocktail and mosaic approaches for spike protein- nanoparticle vaccines and did not identify a significant difference in neutralization titers between the two approaches.24 A cocktail vaccine is also aligned with the hypothesis motivating this work - that if the mixture of antigens comprising the cocktail is carefully selected by characterizing the antigenic landscape, the antibody response elicited by each component would collectively result in a broadly protective polyclonal antibody response. It would be interesting to revisit the mosaic idea in future work."  
+
+<|ref|>text<|/ref|><|det|>[[115, 572, 883, 606]]<|/det|>
+Comment: I agree that the "diameter was measured to be \(90 - 100 \text{nm}\) , consistent with prior characterization," according to the previous paper:  
+
+<|ref|>text<|/ref|><|det|>[[115, 607, 883, 658]]<|/det|>
+Chiba, S., Frey, S.J., Halfmann, P.J. et al. Multivalent nanoparticle- based vaccines protect hamsters against SARS- CoV- 2 after a single immunization. Commun Biol 4, 597 (2021). https://doi.org/10.1038/s42003- 021- 02128- 8  
+
+<|ref|>text<|/ref|><|det|>[[115, 659, 883, 727]]<|/det|>
+But I would consider an additional analysis to confirm particle integrity. Perhaps TEM, as was done previously. This could help rule out that the spike component on some fraction was unfolded. The limited data - DLS and ELISA, are both crude methods, particularly the way the ELISA was done, so in my opinion this is the aspect of the manuscript that needs the most improvement.  
+
+<|ref|>text<|/ref|><|det|>[[115, 728, 883, 762]]<|/det|>
+Response: We have added TEM characterization as requested by the reviewer as Figure 2f in the revised manuscript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 779, 883, 865]]<|/det|>
+Comment: The authors could do a better job citing other COVID nanoparticle vaccines, which seems like an oversight given the platform. Most of the introduction is spent talking in length about current bivalent vaccines, but if these were efficacious we wouldn't need this manuscript. I would consider shortening this section and describing, to some extent, nanoparticle COVID vaccines. Potential additional references:
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[112, 88, 881, 245]]<|/det|>
+a. M. Gordon Joyce et al., A SARS-CoV-2 ferritin nanoparticle vaccine elicits protective immune responses in nonhuman primates. Sci. Transl. Med. 14, eabi5735(2022). DOI:10.1126/scitranslmed.abi5735 
+b. Weidenbacher, P.AB., Sanyal, M., Friedland, N. et al. A ferritin-based COVID-19 nanoparticle vaccine that elicits robust, durable, broad-spectrum neutralizing antisera in non-human primates. Nat Commun 14, 2149 (2023). https://doi.org/10.1038/s41467-023-37417-9 
+c. Brouwer PJM, Et. al, Two-component spike nanoparticle vaccine protects macaques from SARS-CoV-2 infection. Cell. 2021 Mar 4;184(5):1188-1200.e19. doi: 10.1016/j.cell.2021.01.035. Epub 2021 Jan 26. PMID: 33577765; PMCID: PMC7834972.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 246, 208, 261]]<|/det|>
+## Response:  
+
+<|ref|>text<|/ref|><|det|>[[115, 262, 883, 296]]<|/det|>
+We have cited these manuscripts and added the following text to the introduction based on the reviewer's suggestions:  
+
+<|ref|>text<|/ref|><|det|>[[114, 295, 883, 486]]<|/det|>
+"Protein nanoparticles have emerged as attractive platforms for the display of S protein antigens. Brouwer et al. generated two component protein nanoparticles displaying stabilized perfusion SARS-CoV-2 S proteins that protected vaccinated macaques from a challenge with SARS- CoV- 2.25 Joyce et al. showed that adjuvanted SARS- CoV- 2 S protein-ferritin nanoparticle vaccines protected non- human primates from a challenge with SARS- CoV- 2.26 Weidenbacher et al. reported that adjuvanted ferritin nanoparticle vaccines displaying a truncated form of the SARS- CoV- 2 S protein ectodomain elicited a broad neutralizing antibody response in non- human primates.27 Hutchinson et al. designed self-assembling protein nanoparticles displaying multiple S protein antigens that protected mice from a challenge with MERS- CoV.28 As described above, Brinkkemper et al. designed nanoparticles presenting mixtures of the SARS- CoV- 1 and SARS- CoV- 2 S proteins.24  
+
+<|ref|>text<|/ref|><|det|>[[114, 502, 883, 640]]<|/det|>
+Comment: I find the Pfizer comparator to be not informative and confusing. Without a control indicating this vaccine retained potency, it's difficult to understand why it shows no detectable efficacy? There is language in the text hypothesizing what happened, but if this serum is available, I would test it for binding (hopefully more sensitive than neutralization) to demonstrate the animals were properly immunized - or provide a citation that a 10ug dose doesn't confer protection, neutralization, or any apparent activity in hamsters. In Fig 4c there is protection with no neutralization - so to not show any protection suggests, in my mind, the procured vaccine was ineffective.  
+
+<|ref|>text<|/ref|><|det|>[[114, 640, 883, 726]]<|/det|>
+Response: We agree with the reviewer's concerns regarding the potency of a \(10\mu \mathrm{g}\) dose of the Pfizer- BioNTech vaccine and have repeated the experiment with a higher dose (30 \(\mu \mathrm{g}\) ). This was a sufficient dose to result in a significant decrease in lung titers following an XBB.1 challenge, and the higher dose did elicit neutralizing antibodies against 614G, BA.5, and in one case against XBB.1. We have updated the text to include these results.  
+
+<|ref|>text<|/ref|><|det|>[[115, 727, 316, 743]]<|/det|>
+Regarding neutralization:  
+
+<|ref|>text<|/ref|><|det|>[[114, 743, 883, 880]]<|/det|>
+"To compare, hamsters immunized with 30 \(\mu \mathrm{g}\) of the Pfizer- BioNTech bivalent vaccine did produce neutralizing antibodies against an early SARS- CoV- 2 isolate (S- 614G) and BA.5 as expected given the composition of the vaccine. Like our bivalent 614D/SHC014 vaccine, neutralization titers were significantly reduced against XBB.1 (3 out of 4 hamsters had no detectable titers), but unlike any of our bivalent or trivalent vaccines, no detectable neutralization titers were observed against the Clade 1A bat CoVs SHC014 and WIV1. These results further support the inclusion of antigens from both XBB- like and Clade 1A CoVs in a broadly protective cocktail vaccine."  
+
+<|ref|>text<|/ref|><|det|>[[115, 881, 407, 898]]<|/det|>
+Regarding protection from challenge
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 883, 124]]<|/det|>
+"Meanwhile, a single immunization with 30 \(\mu g\) of the Pfizer- BioNTech bivalent vaccine significantly decreased lung and nasal titers after challenge with XBB.1 (Fig. 4c, e)."  
+
+<|ref|>text<|/ref|><|det|>[[115, 140, 883, 210]]<|/det|>
+Comment: The result that "The bivalent vaccine provided complete protection against the XBB.1 challenge despite eliciting significantly lower neutralizing antibody titers against XBB.1 than the trivalent vaccines." Should be further discussed. Does this data argue against the inclusion of an omicron variant? Does this suggest neutralization is a bad surrogate for protection?  
+
+<|ref|>text<|/ref|><|det|>[[115, 210, 883, 243]]<|/det|>
+Response: We have added the following discussion to the revised manuscript as suggested by the reviewer:  
+
+<|ref|>text<|/ref|><|det|>[[115, 244, 883, 348]]<|/det|>
+"While neutralizing antibody titers against SARS- CoV- 2 have been identified as a correlate of protection against symptomatic disease \(^{8,39}\) , even low neutralizing titers can be sufficient for protection. High neutralization titers elicited by vaccination are still desirable, since they are likely to provide greater protection against viral escape; neutralization titers against viral variants are generally lower than those against the vaccine strain. The inclusion of an Omicron antigen in the cocktail is thus justified, particularly in light of the continuing emergence of new Omicron variants."  
+
+<|ref|>text<|/ref|><|det|>[[115, 364, 883, 398]]<|/det|>
+Comment: Given how similar all the ELISAs look, negative controls should be shown for the ELISAs – ideally the VLP alone binding to Ab control.  
+
+<|ref|>text<|/ref|><|det|>[[115, 399, 828, 416]]<|/det|>
+Response: We have added VLP- only controls to Figure 2 and Supplementary Figure 2b.  
+
+<|ref|>text<|/ref|><|det|>[[115, 433, 772, 450]]<|/det|>
+We have also made changes in response to the reviewer's other minor comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 451, 884, 503]]<|/det|>
+Comment: Please use fewer significant digits in Fig. 3 top (278616 = 2.8\*105 or 2.8e5) Response: We thank the reviewer for the recommendation and have updated Figure 3 accordingly.  
+
+<|ref|>text<|/ref|><|det|>[[115, 519, 884, 553]]<|/det|>
+Comment: Please provide the dose of aluminum in the alhydrogel adjuvanted samples (as opposed to "equal volume").  
+
+<|ref|>text<|/ref|><|det|>[[115, 554, 844, 572]]<|/det|>
+Response: We have added the dose of aluminum in the alhydrogel samples as requested.  
+
+<|ref|>text<|/ref|><|det|>[[115, 588, 448, 604]]<|/det|>
+Comment: I like the analysis in Figure 3c  
+
+<|ref|>text<|/ref|><|det|>[[115, 606, 644, 623]]<|/det|>
+Response: We thank the reviewer for this encouraging comment.  
+
+<|ref|>text<|/ref|><|det|>[[115, 640, 883, 692]]<|/det|>
+Comment: "While a naive animal might respond equally to all components of a vaccine, an animal with pre- existing immunity might instead be biased towards one component to which it has been exposed to before." – please add a citation for original antigenic sin/immunological imprinting.  
+
+<|ref|>text<|/ref|><|det|>[[115, 693, 866, 710]]<|/det|>
+Response: We have added citations for "original antigenic sin" as suggested by the reviewer.  
+
+<|ref|>text<|/ref|><|det|>[[144, 711, 883, 816]]<|/det|>
+- Francis Jr., T. On the Doctrine of Original Antigenic Sin Proc Am Phil Soc 104, 572-578 (1960)- de St. Groth, F. & Webster, R.G. Disquisitions of Original Antigenic Sin. I. Evidence in man. J Exp Med 124, 331-345 (1966)- de St. Groth, F. & Webster, R.G. Disquisitions on Original Antigenic Sin. II. Proof in lower creatures. J Exp Med 124, 347-361 (1966).  
+
+<|ref|>text<|/ref|><|det|>[[115, 833, 883, 885]]<|/det|>
+Comment: Methods around "HiScreen Capto Core" need elaboration. My understanding was that these columns did not bind to large nanoparticles, so the discussion around elution in just 20mM tris is surprising? Did you collect the entire flow through?
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 882, 124]]<|/det|>
+Response: We have updated our description of MS2 purification using the HiScreen Capto Core columns. The revised text reads:  
+
+<|ref|>text<|/ref|><|det|>[[115, 124, 883, 228]]<|/det|>
+"25 mL of the lysate was loaded onto four HiScreen Capto Core columns (Cytvia) in series using an Akta Start system. MS2 was purified by washing with \(20mM\) Tris Base for \(\sim 5\) column volumes (CVs). The entire flowthrough was collected in \(1.8mL\) fractions, and SDS- PAGE was used to determine purity and yield. Fractions 7- 12 were typically those most enriched in MS2 while smaller impurities would concentrate in later fractions due to the CaptoCore system's size- exclusion character."  
+
+<|ref|>text<|/ref|><|det|>[[115, 243, 883, 313]]<|/det|>
+Comment: ELISA methods should include coating concentration and Ab concentration - methods as written could not be replicated. "The final concentration of ACE2- Fc and antibodies was such that there would be one molecule per S protein trimer." Is confusing, and assumes \(100\%\) bound to the plate? But we know the vast majority of protein remains unbound to the ELISA plate.  
+
+<|ref|>text<|/ref|><|det|>[[115, 313, 612, 330]]<|/det|>
+Response: We have rewritten the ELISA methods as follows:  
+
+<|ref|>text<|/ref|><|det|>[[115, 330, 883, 572]]<|/det|>
+"VLP- S and S protein were diluted in PBS such that the concentration of S protein in each solution was \(1\mu g / mL\) . \(100\mu L\) (0.1 \(\mu g\) of S protein) per well of the diluted protein was then coated onto a Nunc Maxisorp 96- well plate. After a 1- hour incubation, the protein solutions were discarded, and the wells were blocked with \(200\mu L\) of \(5\%\) BSA (EMD Millipore) in PBST (0.05% Tween- 20) for 1 hour. The wells were then washed three times with PBST. Stock solutions of primary antibodies at \(0.3mg / mL\) , \(3.4mg / mL\) , \(3.8mg / mL\) , and \(1.9mg / mL\) (by BCA) for ACE2- Fc, CR3022, S309, and S2P6 respectively were diluted 9:1500, 1:1500, 1:1500, and 1:30000 respectively in PBST with \(1\%\) BSA, and \(100\mu L\) per well of these diluted primary antibodies were added. After 1 hour, the wells were washed three times with PBST, and \(100\mu L\) of horseradish peroxidase- conjugated anti- human IgG Fc goat antibody (MP Biomedical, 1:5000) diluted in PBST with \(1\%\) BSA was added to all wells. The wells were washed three times with PBST after a 1- hour incubation. \(100\mu L\) of TMB (Thermo Scientific) were added to each well and allowed to develop for 3 minutes. The reaction was stopped with \(160mM\) sulfuric acid, and the absorbance at \(450nm\) was read with a Spectramax i3x plate reader (Molecular Devices) and Gen5 2.07 software (BioTek)."  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 606, 222, 622]]<|/det|>
+## Reviewer #2  
+
+<|ref|>text<|/ref|><|det|>[[115, 623, 883, 675]]<|/det|>
+Reviewer 2 noted that "The study is well performed and the addition of the hACE2 hamster models add to the robustness of the data to show broader protection from challenge." We thank the reviewer and have responded to the reviewer's other comments below.  
+
+<|ref|>text<|/ref|><|det|>[[115, 692, 883, 761]]<|/det|>
+Comment: While not needed in this manuscript, I would be interested to know if lower protein concentrations have been tested in other hamster vaccine experiments since submission. 15ug of each nanoparticle (45ug in all) is a large dose. The data suggest that much lower concentrations should be as effective. Have then been tested?  
+
+<|ref|>text<|/ref|><|det|>[[115, 761, 882, 830]]<|/det|>
+Response: We have not yet tested reducing the dose of our vaccine cocktails. We hypothesize that it would be possible to reduce the dose though, since our trivalent cocktails elicit robust and broad neutralizing antibody titers that are likely to be well above the threshold needed for protection.  
+
+<|ref|>text<|/ref|><|det|>[[115, 846, 883, 881]]<|/det|>
+Comment: Mucosal immunity is a major driver of protective responses in humans, and while not as dramatic in hamster models, it would be intriguing to know if neutralizing antibody levels were
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 883, 124]]<|/det|>
+tested in the vaccinated hamsters. Also were nasal washes analyzed for virus titer after challenge?  
+
+<|ref|>text<|/ref|><|det|>[[115, 124, 883, 159]]<|/det|>
+Response: We have added data for nasal turbinate titers in Figures 3c, 4d- e, and 5b. Discussion of these nasal turbinate titers has also been added.  
+
+<|ref|>text<|/ref|><|det|>[[115, 160, 881, 245]]<|/det|>
+Regarding nasal titers after monovalent vaccination and challenge with BA.5 (Figure 3c): "Similar trends were observed in viral titers in the nasal turbinates as VLP- 614D- S and all of the Omicron VLP- S vaccines significantly reduced virus levels, while hamsters vaccinated with VLP- SHC014- S and VLP- SARS- CoV- 1- S had viral levels comparable to those of the control hamsters (Fig. 3c)."  
+
+<|ref|>text<|/ref|><|det|>[[115, 261, 883, 331]]<|/det|>
+Regarding nasal titers after cocktail vaccination and challenge with BA.5 and XBB.1 (Figure 4d): "Similar trends were also seen for nasal titers. The VLP- 614D- S and VLP- SHC014- S bivalent cocktail significantly reduced nasal virus titers, while hamsters immunized with either trivalent cocktail had undetectable nasal titers (Fig 4d)."  
+
+<|ref|>text<|/ref|><|det|>[[115, 347, 881, 382]]<|/det|>
+Regarding nasal titers after Pfizer- BioNTech bivalent vaccination and challenge with XBB.1 (Figure 4e):  
+
+<|ref|>text<|/ref|><|det|>[[115, 382, 883, 417]]<|/det|>
+"Meanwhile, a single immunization with 30 μg of the Pfizer- BioNTech bivalent vaccine significantly decreased lung and nasal titers after challenge with XBB.1 (Fig. 4c,e)."  
+
+<|ref|>text<|/ref|><|det|>[[115, 433, 883, 468]]<|/det|>
+Regarding nasal titers after cocktail vaccination in hACE2 hamsters and challenge with WIV1 and SHC014 (Figure 5b):  
+
+<|ref|>text<|/ref|><|det|>[[115, 468, 883, 503]]<|/det|>
+"Nasal turbinate titers were also significantly reduced in vaccinated transgenic hamsters in both cases (Fig. 5b), though the magnitude of this reduction was weaker than for the Clade 1B viruses."  
+
+<|ref|>text<|/ref|><|det|>[[115, 536, 883, 570]]<|/det|>
+Comment: Several interesting points are brought up by the data that should be commented on in the discussion:  
+
+<|ref|>text<|/ref|><|det|>[[115, 571, 881, 606]]<|/det|>
+1. Is there a structural basis for SHC014 Spike to neutralize better against WIV1 than SARS1? Has that been seen in other vaccine models or is this unique to this nanoparticle design?  
+
+<|ref|>text<|/ref|><|det|>[[115, 606, 883, 658]]<|/det|>
+Response: We thank the reviewer for this comment; as seen in Fig. 1c, the WIV1 S is in fact more homologous to the SHC014 S than to the SARS-CoV- 1 S. We have corrected this text in the revised manuscript, which now reads  
+
+<|ref|>text<|/ref|><|det|>[[115, 658, 883, 710]]<|/det|>
+"VLP- SHC014- S elicited higher neutralizing antibody titers against WIV1 compared to VLP- SARS- CoV- 1- S, consistent with the WIV1 S being more homologous to the SHC014 S than to the SARS- CoV- 1 S (Fig. 1c)."  
+
+<|ref|>text<|/ref|><|det|>[[115, 727, 883, 829]]<|/det|>
+Comment: 2. A separate discussion is warranted in this manuscript for explaining implications of this work across the context of broadly protective vaccines. Important points to touch on are 1) mucosal vs systemic immunity of this vaccine platform, 2) structural basis for the neutralization and protection with emphasis on other widely divergent bat coronaviruses and how they may fair against this vaccine based on other published structural data, 3) longevity of the response and how this vaccine design could be used commercially.  
+
+<|ref|>text<|/ref|><|det|>[[115, 846, 883, 880]]<|/det|>
+Response: We have added the following text to the discussion based on the reviewer's comments:
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 88, 883, 278]]<|/det|>
+"While we are encouraged by these results, there are several additional avenues that would be interesting to explore in future work. Enhancing mucosal immunity might not only enhance protection against viral infection, but also decrease viral transmission. \(^{40}\) Intranasal vaccination against SARS- CoV- 2 has been explored with several platforms, including mRNA- lipid nanoparticles \(^{41}\) , nanoparticles displaying the RBD \(^{42}\) , live attenuated influenza viruses also encoding the RBD \(^{43}\) , adenovirus- vectored vaccines \(^{44}\) , and by using an intranasal boost with the unadjuvanted spike protein \(^{40}\) . The adaptation of our platform for intranasal delivery could be a promising avenue for improving the mucosal response. Characterizing the longevity of protection would also be an interesting avenue for future research. It would be particularly interesting to determine whether stronger mucosal immunity results in more durable protection against symptomatic disease.  
+
+<|ref|>text<|/ref|><|det|>[[115, 288, 883, 496]]<|/det|>
+Secondly, while the focus of this paper was on protecting against viruses in Clades 1A and 1B, extending protection to further sarbecovirus clades would be interesting to explore. Significant differences in their receptor binding domains are the primary basis for virus classification into clades, with Clade 2 viruses being unable to use ACE2 as an entry receptor and Clade 3 & 4 viruses harboring one deletion relative to Clade 1. \(^{45}\) Nevertheless, some Clade 2 viruses (a proposed "Clade 2A") \(^{46}\) and several Clade 3 viruses \(^{47}\) may be capable of infecting human cells, albeit in some cases only with exogenous protease treatment. RBD- and NTD- focused humoral immunity is unlikely to be cross- reactive between clades, as inter- clade RBD and NTD amino acid identity percentages are approximately 65- 75% and 45- 55%, respectively. As such, the inclusion of additional antigens to cover these clades may be necessary. Using the approach described in this work, antigenic cartography should help us identify the minimal mixture of antigens required to elicit broad protection against sarbecoviruses.  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 883, 644]]<|/det|>
+Given that protein- nanoparticle vaccines have been approved for clinical use \(^{48}\) , commercialization of this vaccine platform could be possible in the future. Moreover, the optimal components for a cocktail vaccine that are suggested by antigenic cartography may be platform- agnostic and could therefore be applied not just to other protein nanoparticle platforms, but also to other modalities such as mRNA- based vaccines. Indeed, the selection of an XBB strain (XBB.1.5) for incorporation in the recently approved monovalent vaccine is consistent with our results and it would be interesting to explore the protective efficacy of trivalent mRNA vaccines based on the compositions identified in this work against clade 1 sarbecoviruses."
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 285, 104]]<|/det|>
+Reviewers' Comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 120, 635, 165]]<|/det|>
+Reviewer #3:  Remarks to the Author:  The changes to the manuscript are sufficient to answer my questions.
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1ff816831c1878fcdce6de0a4d1b9390aa7f2f7e5804b1700b0dca50f70ee965/images_list.json

@@ -0,0 +1 @@
+[]

peer_reviews/supplementary_0_Peer Review File__1ff816831c1878fcdce6de0a4d1b9390aa7f2f7e5804b1700b0dca50f70ee965/supplementary_0_Peer Review File__1ff816831c1878fcdce6de0a4d1b9390aa7f2f7e5804b1700b0dca50f70ee965.mmd

@@ -0,0 +1,367 @@
+
+# nature portfolio  
+
+Peer Review File  
+
+![PLACEHOLDER_0_0]
+  
+
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+
+## REVIEWER COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+How the correct metals locate to the correct proteins is a puzzle. Moreover, this is an important puzzle since almost half the reactions of life are estimated to be metal catalysed and typically alternative metals will not drive catalysis. The puzzle is further complicated by the observation that most metalloproteins bind one or more wrong metals many orders of magnitude more tightly than the required metal. For cytosolic proteins the puzzle is substantially resolved by the actions of metal homeostatic systems: These systems maintain the availabilities of the tightest binding metals much lower than the availabilities of the weaker binding ones and so it becomes possible to simultaneously have proteins that require tight binding metals and those that require weak binding ones in the same cytosolic compartment. This becomes more challenging however outside the plasma- membrane, especially of Gram positive bacteria, since this location is essentially continuous with the external environment. Intriguingly, in this manuscript we learn that even manganese- proteins (available divalent manganese being a relatively weak binding metal) that are secreted in an unfolded state, still take advantage of the homeostatic mechanisms that control cytosolic metal availabilities to the inverse of the Universal affinity series (the Irving Williams series). Previously, we only knew this to be true of manganese- proteins that are secreted in a folded state via the TAT system. In this manner metal specificity becomes the product of the multiple partitioning events that control cytosolic metal availabilities for both TAT and Sec substrates. This is an significant advance which adds to a body of literature that needs to be brought to the attention of the widest possible audience.  
+
+Do we know if manganese becomes kinetically trapped in Lipoteichoic acid synthase post- folding, or if the exchange of manganese with a tighter binding metal such as zinc is slow? Could this be experimentally addressed or alternatively could the authors add further comments on how, post MeeF/Y- assisted metalation and secretion, subsequent exchange of manganese for a tighter binding metal is avoided.  
+
+Is it known what fraction of the Sec secretome is composed of LTA? Which data exclude the possibility that MeeF/Y act directly in the secretion of LTA but only indirectly exert effects on manganese because LTA is a major sink for extra plasma- membrane manganese (perhaps bound in less- exchangeable oxidised forms)?  
+
+Could the inhibitory phenotypes related to zinc be a consequence of known inhibitory effects of zinc on the mechanisms of manganese import as documented by Chris McDevitt and co- workers?  
+
+line 251. Metalation of CucA was subsequently shown to require two copper transporting ATPases and a cytosolic metallochaperone, with CucA secretion being absent in mutants lacking these proteins suggesting that metalation is assisted and co- coincident with secretion (J Biol Chem 2010 285: 32504- 32511).  
+
+The use of "suggest" and "possibly" in the closing sentence: 'These results "suggest" that TerC proteins are important in both bacteria and eukaryotes for the proper functioning of exported proteins, "possibly" by mediating the co- translocational insertion of metals into nascent proteins during transit across membranes' is more cautious than elsewhere and perhaps some statements need to be toned- down and/or caveats added.  
+
+Reviewer #2 (Remarks to the Author):  
+
+The manuscript by He and colleagues describes a novel role for TerC proteins in the delivery of manganese to proteins exported via the SEC translocase. The findings built on their own analyses which defined TerC as manganese export systems, and the authors introduce these proteins as possibly functionally conserved across multiple kingdoms of life. The work performed here is of a high standard and the manuscript is very clearly written. The experiments performed here are logical and provide the necessary information required to draw conclusions regarding the proposed novel role for TerC in B. subtilis. However, the brevity of the manuscript presents a limitation in terms of drawing
+
+<--- Page Split --->
+
+broader conclusions across bacteria and definitely when considering other kingdoms of life.  
+
+Main comments:  
+
+This work may benefit from some in silico work, including the genetic organisation and conservation across the species, genus, between other bacteria and across kingdoms. Structural models are likely to assist too. These don't need to be described in detail in the main body of the text, but may appeal to those that are not in the field of Bacillus research.  
+
+Expanding upon this, I would like to learn more about why MeeF appears to have a slightly stronger functional role than MeeY, and why more interacting proteins were identified. Other than the \(40\%\) similarity, what's different: expression, structure, distribution across the cell surface?  
+
+Why is YjbE not contributing to the processes described here, and how may it fulfill its role during sporulation? Please discuss.  
+
+How are TerC proteins and their role in metalation of secreted proteins impacted by Mn starvation, something that bacterial pathogens readily encounter during infection?  
+
+Minor comments:  
+
+Line 64: Please don't specify the name of the mutant in the introduction, leave that until the results, where it has already been clearly stated too.  
+
+Line 75: This statement is superfluous, as this can be derived from the intro.  
+
+Line 643: Is it clearer bands, or simply more bands?  
+
+Line 101: Listing the molecular weights of the proteins seems confusing and Fig 1d doesn't actually show the specifics of those proteases listed, only that of the delta7 strain. Hence, denoting specific candidates seems appropriate for Fig S2 alone.  
+
+Line 654: Please indicate total samples (were multiple biological replicates included on one day, as I assume that technical replicates are not displayed)  
+
+Line 212: Please use these references for the strains with FLAG tag, as it may cause confusion.  
+
+Line 138- 139: This should be the other way around.  
+
+Line 803: This sentence seems incomplete.  
+
+Line 145: Please define that the MeeF interactome is broader than that of the MeeY.  
+
+Line 811- 812: I would say that the pattern is similar, it's the difference in intensity that is more striking.  
+
+Lines 183- 185: Bit odd to explain what they do after already defining the result, restructure this order. Fig S8: Relative transcript to what? Scale is confusing. Why only two reps?  
+
+Line 209: Please ascertain this transcriptionally.  
+
+Lines 229- 236: This really needs bioinformatics to support the choice of those two candidates and the claim that it's conserved.  
+
+Lines 239- 246: This can be condensed.  
+
+Line 259: delete "its".  
+
+Line 266: Is Mn structural or any role in folding, as lack of metalation seems like an insignificant reason to not release the protein by the translocase? Or perhaps degradation is favourable as the protein would be non- functional, please discuss.  
+
+The legends of Fig 2D and S2 appear contradictory regarding the role of AprE.  
+
+Reviewer #3 (Remarks to the Author):  
+
+This MS addresses the important issue of the metalation of extracytoplasmic proteins. This is important because many secretory Gram- positive proteins are metalloproteins or require metal ions for folding following translocation across the membrane (via the Sec translocase) in an essential unfolded/unstructured form. Slowly folding proteins are subject to proteolysis by quality control proteases and metal ions are known to act as folding factors, speeding up the folding at least for some proteins. This MS provides a mechanism for the metalation by the lower affinity metal ion Mn, so as to potentially optimise folding and avoid mis- metalation by higher affinity metal ions. There is a suggestion that folding provides an energy force that helps "pull" for the protein from the translocase and so metalation my help in this process.
+
+<--- Page Split --->
+
+The current view is that metalation occurs in most cases via the concentration (cf the medium) of metal ions in the negatively charged cell wall. However, this does not allow for mis- metalation for lower affinity metal ions. Therefore, the hypothesis proposed in this manuscript is attractive. I do, however, have some comments on what I see as potential experimental limitations that the authors should address. These are indicated below.  
+
+## L104/Fig 1c/d  
+
+The legend states that "Supernatants were collected from ONCs with the same cell number(s)". What medium was use and did all the cultures reach stationary phase with the same numbers of cells and at the same time? I might have expected the FY mutant to have lower numbers or to have a slower in some media. I am also cautious about simply using data derived from overnight cultures (ONCs) without considering growth kinetics. Bacillus continues to produce many secretory enzymes in stationary phase and therefore cells that enter stationary phase earlier than others could generate more enzyme during their longer stationary phase. Therefore, the lower protease activity from the mutant could be due to a slower growth rate (ie reaching stationary phase later) and a resulting shorter time in stationary phase. Has this been taken into account?  
+
+## L 107  
+
+The same point as above, although I suspect the interpretation is correct. The length of time in stationary phase could significantly affect secretory protein production and a better experiment would have been to take samples at a fixed time following the transition from exponential to stationary phase. The \(>5\) - fold reduction in extracellular Mn is to be expected from previous work on these genes as Mn exporters.  
+
+## L117  
+
+As above, and AmyQ (which requires Ca ions for folding) certainly accumulates in stationary phase.  
+
+## L127  
+
+It is well known that AmyQ induces secretion stress as compared with native enzymes. As a Ca requiring enzymes it is not clear what the significance of these findings are. I think the most significant finding here is that secretion stress is not induced in the FY mutant, which I find surprising if Mn requiring enzymes are unable to fold/fold rapidly.  
+
+## L142  
+
+It could be argued that the results in Table 1 indicate that the MeeF and MeeY (membrane proteins themselves) simply interact with other membrane proteins. I don't see a clear justification, based on these data, for claiming MeeF and MeeY appear to function as part of the secretosome.  
+
+## L156  
+
+Same point as above about ONCs and secretory proteins.  
+
+## L168  
+
+Despite my comments about the way protease levels were estimated, the potential role of MeeF and MeeY in providing Mn as a folding factor helping to "pull" proteins from the translocase would be consistent with the role of FtsH clearing the translocase in the FY mutants.  
+
+## L204  
+
+The data on Mn addition to the medium in relation to LtaS seem clear. However, the medium (as indicated above) already contains some Mn that, in the wild type, would be at a higher concentration at the membrane/wall interface due to its mobile interaction with the phosphate in LTA. Could the influence of Mn addition on LtaS activity in the FY mutant be simply the result of the controlling the level of Mn added to the medium to "just enough" (as indicated above and the influence on the WT as indicated below) for the wild type but not enough for the mutant. If LtaS folding was facilitated simply by the level of Mn in the medium, and this was lower in the FY mutant due to Mn retention by the cell, then the rate of folding of LtaS at lower Mn concentrations could slower, leading to its removal by quality control proteases. Additional Mn allow this to recover LTA synthesis.
+
+<--- Page Split --->
+
+From the data presented I think it is too early to argue that MeeF and MeeY are accessory subunits of the holotranslocon
+
+<--- Page Split --->
+
+## RESPONSE TO REVIEWER COMMENTS  
+
+## Reviewer #1 (Remarks to the Author):  
+
+How the correct metals locate to the correct proteins is a puzzle. Moreover, this is an important puzzle since almost half the reactions of life are estimated to be metal catalysed and typically alternative metals will not drive catalysis. The puzzle is further complicated by the observation that most metallo- proteins bind one or more wrong metals many orders of magnitude more tightly than the required metal. For cytosolic proteins the puzzle is substantially resolved by the actions of metal homeostatic systems: These systems maintain the availabilities of the tightest binding metals much lower than the availabilities of the weaker binding ones and so it becomes possible to simultaneously have proteins that require tight binding metals and those that require weak binding ones in the same cytosolic compartment. This becomes more challenging however outside the plasma- membrane, especially of Gram positive bacteria, since this location is essentially continuous with the external environment. Intriguingly, in this manuscript we learn that even manganese- proteins (available divalent manganese being a relatively weak binding metal) that are secreted in an unfolded state, still take advantage of the homeostatic mechanisms that control cytosolic metal availabilities to the inverse of the Universal affinity series (the Irving Williams series). Previously, we only knew this to be true of manganese- proteins that are secreted in a folded state via the TAT system. In this manner metal specificity becomes the product of the multiple partitioning events that control cytosolic metal availabilities for both TAT and Sec substrates. This is an significant advance which adds to a body of literature that needs to be brought to the attention of the widest possible audience.  
+
+\*\* We appreciate the detailed summary of the referee who has judged this work to be highly significant.  
+
+Do we know if manganese becomes kinetically trapped in Lipoteichoic acid synthase post- folding, or if the exchange of manganese with a tighter binding metal such as zinc is slow? Could this be experimentally addressed or alternatively could the authors add further comments on how, post MeeF/Y- assisted metalation and secretion, subsequent exchange of manganese for a tighter binding metal is avoided.  
+
+\*\* We agree that the problem of mismetallation is important and that the Mn bound LtaS enzyme might, under some conditions, be inhibited by tighter binding metals. We address this experimentally by showing that mutants deficient in metalation of LtaS with Mn (FY mutants) are sensitive to inhibition by Zn, consistent with a mismetallation mechanism (Figure S9). Further studies of metal exchange after metal loading will require future study.  
+
+Is it known what fraction of the Sec secretome is composed of LTA?  
+
+\*\* We believe the referee meant to inquire about how abundant the LtaS protein is within the Sec secretome. The secretome of B. subtilis includes both extracellular enzymes and proteins destined for the membrane (the membrane proteome). The text has been revised to clarify that LtaS is a low abundance membrane protein (line 194), as shown previously (PMID 31424929).  
+
+Which data exclude the possibility that MeeF/Y act directly in the secretion of LTA but only indirectly exert effects on manganese because LTA is a major sink for extra plasma- membrane manganese (perhaps bound in less- exchangeable oxidised forms)?  
+
+\*\* LTA (lipoteichoic acid) is synthesized extracellularly by the LtaS enzyme (it is not a secreted product). MeeF/Y are homologous to ion transporters, are often regulated by Mn- inducible riboswitches, and have been shown to function in Mn export from cells both in B. subtilis (PMID: 31685536) and more recently in E. coli (PMID: 37214827). LTA polymers are not secreted, but are assembled on the cell surface, where they indeed function in metal buffering (lines 326- 329).  
+
+Could the inhibitory phenotypes related to zinc be a consequence of known inhibitory effects of zinc on the mechanisms of manganese import as documented by Chris McDevitt and co- workers? \*\* We believe the referee is referring to the combined effect of Zn and the LtaS inhibitor (1771) as reported in Figure S9. These results demonstrate that Zn intoxication is specific to cells that are deficient in LTA synthesis (through mutation of one or more LtaS isozymes) and chemical inhibition of LtaS by 1771. This is unrelated to mechanism of inhibition cited in Chris McDevitt and coworkers, which refers to the ability of Zn to competitively
+
+<--- Page Split --->
+
+inhibit Mn uptake through the MntABCD ABC transporter (PMID 22072971). This mechanism is not operative in B. subtilis since the major Mn importer is MntH, which is not inhibited by Zn. Instead, high Zn interferes with the electron transport chain as shown in B. subtilis (PMID: 27935957) and in E. coli (PMID: 7557331).  
+
+line 251. Metalation of CucA was subsequently shown to require two copper transporting ATPases and a cytosolic metallochaperone, with CucA secretion being absent in mutants lacking these proteins suggesting that metalation is assisted and co- coincident with secretion (J Biol Chem 2010 285: 32504- 32511).  
+
+\*\* Thank you for this clarification. We have limited our discussion of the Mn- dependent cupin (MncA).  
+
+The use of "suggest" and "possibly" in the closing sentence: 'These results "suggest" that TerC proteins are important in both bacteria and eukaryotes for the proper functioning of exported proteins, "possibly" by mediating the co- translocational insertion of metals into nascent proteins during transit across membranes' is more cautious than elsewhere and perhaps some statements need to be toned- down and/or caveats added.  
+
+\*\* The Introduction and Discussion sections have been re- organized and expanded to clarify our conclusions (see referee 2, first comment).  
+
+## Reviewer #2 (Remarks to the Author):  
+
+The manuscript by He and colleagues describes a novel role for TerC proteins in the delivery of manganese to proteins exported via the SEC translocase. The findings built on their own analyses which defined TerC as manganese export systems, and the authors introduce these proteins as possibly functionally conserved across multiple kingdoms of life. The work performed here is of a high standard and the manuscript is very clearly written. The experiments performed here are logical and provide the necessary information required to draw conclusions regarding the proposed novel role for TerC in B. subtilis. However, the brevity of the manuscript presents a limitation in terms of drawing broader conclusions across bacteria and definitely when considering other kingdoms of life.  
+
+\*\* We appreciate the referee's supportive assessment. We have amended the text to expand and clarify our Discussion and further develop those ideas that may not have been clear due to the brevity of our presentation.  
+
+## Main comments:  
+
+This work may benefit from some in silico work, including the genetic organisation and conservation across the species, genus, between other bacteria and across kingdoms. Structural models are likely to assist too. These don't need to be described in detail in the main body of the text, but may appeal to those that are not in the field of Bacillus research.  
+
+\*\* We have added a new panel (Fig. 4a) to highlight the phylogenetic relatedness of the proteins tested in Fig. 4. We agree that a much broader in silico analysis will be valuable, although considerable efforts of this type are already published. Protein sequence similarity has been used to argue that the bacterial TerC proteins are part of a larger family of ion transporters as discussed in prior in silico work and now cited (lines 283, 340): "TerC proteins (Pfam03741) are a subgroup of the lysine exporter (LysE) superfamily of transporters and have seven TM segments and a conserved metal- binding site \(^{14,70}\) and "Functional studies have linked diverse TerC and related UPF0016 proteins to the transport of Mn and Ca \(^{4,19,87,88}\) . Readers interested in the conservation across species and kingdoms can refer to these detailed analyses and the public resources that describe PFAM/UPF protein alignments and families. Further insights into functional conservation requires further study in diverse systems, and experimental determination of structures and mapping of the ion channels in model proteins.  
+
+Expanding upon this, I would like to learn more about why MeeF appears to have a slightly stronger functional role than MeeY, and why more interacting proteins were identified. Other than the \(40\%\) similarity, what's different: expression, structure, distribution across the cell surface?  
+
+\*\* Our current model for MeeF/Y function is that these proteins allow the transit of Mn from the cytosol to the cell surface, where the ion is held by predicted metal- binding sites on extracytoplasmic loops. These loops may allow for specific delivery of Mn to client proteins as they begin to fold upon exit from the secretory channel. While MeeF/Y are redundant in function for many of the phenotypes noted, they are not identical in function (as
+
+<--- Page Split --->
+
+the referee notes). Indeed, for some phenotypes we believe that there are dramatic differences in role, which provide a tool for better understanding the differentiation of roles for these two proteins in future studies. Our current thinking regarding the biological significance of having two paralogs is now further explained in the Discussion section (line 288- 300).  
+
+Why is YjbE not contributing to the processes described here, and how may it fulfill its role during sporulation? Please discuss.  
+
+\*\* The third paralog (YjbE) is selectively expressed during sporulation according to Nicolas et al. as cited. While this protein plays a minor role under the conditions studied here, we do believe that it is likely more important during sporulation. We have added new data in Fig. 1a that demonstrate that YjbE is not contributing to fitness under the conditions studied here. YjbE is now discussed explicitly in the discussion section (see comment above).  
+
+How are TerC proteins and their role in metalation of secreted proteins impacted by Mn starvation, something that bacterial pathogens readily encounter during infection?  
+
+\*\* The phenotypes described here are observed under conditions of relatively low Mn availability. LB medium has just enough Mn to support growth, and this is probably one reason why we observed the phenotypes reported. As host cells limit Mn during infection (e.g. through the action of calprotectin) the role of TerC family proteins is selectively directing Mn to surface enzymes (e.g. LtaS) will become even more critical than for cells growing in the presence of excess Mn. This is now discussed (l. 300- 303).  
+
+Minor comments:  
+
+Line 64: Please don't specify the name of the mutant in the introduction, leave that until the results, where it has already been clearly stated too.  
+
+\*\* The mutant name was deleted from the line as requested. The mutant strain is described at the beginning of results.  
+
+Line 75: This statement is superfluous, as this can be derived from the intro.  
+
+\*\* The statement was deleted.  
+
+Line 643: Is it clearer bands, or simply more bands?  
+
+\*\* The statement in question, "Higher protease activities correspond to clearer bands on the gel matrix." was deleted since it is not needed. Samples with greater protease activity may have more bands, greater clearing (activity) in each band, or some combination.  
+
+Line 101: Listing the molecular weights of the proteins seems confusing and Fig 1d doesn't actually show the specifics of those proteases listed, only that of the delta7 strain. Hence, denoting specific candidates seems appropriate for Fig S2 alone.  
+
+\*\* We agree and have removed the molecular weights as suggested for Fig. 1d, referring instead to Fig. S2 as suggested.  
+
+Line 654: Please indicate total samples (were multiple biological replicates included on one day, as I assume that technical replicates are not displayed)  
+
+\*\* Figures have been revised so that each point represents a separate biological replicate, with the values shown being the average of technical replicates. This is now explained in the figure legends and the original data are in the provided Source Data file.  
+
+Line 212: Please use these references for the strains with FLAG tag, as it may cause confusion.  
+
+\*\* We do not see strains mentioned at this point in the text. However, we have reviewed the text carefully to make sure that all references to strains are unambiguous.  
+
+Line 138- 139: This should be the other way around. \*\* We have amended the sentence for clarity.
+
+<--- Page Split --->
+
+Line 803: This sentence seems incomplete. \*\* We have re- written the figure legend for clarity.  
+
+Line 145: Please define that the MeeF interactome is broader than that of the MeeY. \*\* We have added a sentence to address this point (l. 157).  
+
+Line 811- 812: I would say that the pattern is similar, it's the difference in intensity that is more striking. \*\* We agree and have amended the text to clarify this point.  
+
+Lines 183- 185: Bit odd to explain what they do after already defining the result, restructure this order. \*\* We have rewritten for clarity.  
+
+Fig S8: Relative transcript to what? Scale is confusing. Why only two reps?  
+
+\*\* This experiment now has 4 biological replicates shown and we have clarified that all values are normalized to the gyrA transcript.  
+
+Line 209: Please ascertain this transcriptionally.  
+
+\*\* We did not mean to imply transcriptional activation. We have changed the text (l. 229) to say "due to increased activity of the YqgS synthase."  
+
+Lines 229- 236: This really needs bioinformatics to support the choice of those two candidates and the claim that it's conserved.  
+
+\*\* A phylogenetic tree is added as panel 4a to provide context for the selection of these orthologs.  
+
+Lines 239- 246: This can be condensed. \*\* We have edited this paragraph for brevity.  
+
+Line 259: delete "its". \*\* Corrected  
+
+Line 266: Is Mn structural or any role in folding, as lack of metalation seems like an insignificant reason to not release the protein by the translocase? Or perhaps degradation is favourable as the protein would be non- functional, please discuss.  
+
+\*\* The discussion section has been revised to more fully describe the proposed role of TerC proteins in exoenzyme metalation.  
+
+The legends of Fig 2D and S2 appear contradictory regarding the role of AprE.  
+
+\*\* The data in S2 suggest that AprE is required for the 17 kD band, assigned as a degradation product of Bpr. This band is still present in the FY mutant in Fig 1d, suggesting that there is at least some AprE secretion, as seen also in the immunoblot analysis of Fig. 2d. Thus, there is no contradiction.  
+
+## Reviewer #3 (Remarks to the Author):  
+
+This MS addresses the important issue of the metalation of extracytoplasmic proteins. This is important because many secretory Gram- positive proteins are metalloproteins or require metal ions for folding following translocation across the membrane (via the Sec translocase) in an essential unfolded/unstructured form. Slowly folding proteins are subject to proteolysis by quality control proteases and metal ions are known to act as folding factors, speeding up the folding at least for some proteins. This MS provides a mechanism for the metalation by the lower affinity metal ion Mn, so as to potentially optimise folding and avoid mis- metalation by higher affinity metal ions. There is a suggestion that folding provides an energy force that helps "pull" for the protein from the translocase and so metalation my help in this process. The current view is that metalation occurs in most cases via the concentration
+
+<--- Page Split --->
+
+(cf the medium) of metal ions in the negatively charged cell wall. However, this does not allow for mis- metalation for lower affinity metal ions. Therefore, the hypothesis proposed in this manuscript is attractive. I do, however, have some comments on what I see as potential experimental limitations that the authors should address. These are indicated below.  
+
+\*\* We thank referee for their comments. We agree with the referee that metal ions buffered by the anionic cell wall provide one source for metalation of nascent proteins. Further, metal ions (and often Ca, specifically) are often invoked as folding factors for extracellular enzymes. The role of LTA in metal buffering is now explicitly discussed in the text (lines 326- 329).  
+
+## L104/Fig 1c/d  
+
+The legend states that "Supernatants were collected from ONCs with the same cell number(s)". What medium was use and did all the cultures reach stationary phase with the same numbers of cells and at the same time? I might have expected the FY mutant to have lower numbers or to have a slower in some media. I am also cautious about simply using data derived from overnight cultures (ONCs) without considering growth kinetics. Bacillus continues to produce many secretory enzymes in stationary phase and therefore cells that enter stationary phase earlier than others could generate more enzyme during their longer stationary phase. Therefore, the lower protease activity from the mutant could be due to a slower growth rate (ie reaching stationary phase later) and a resulting shorter time in stationary phase. Has this been taken into account?  
+
+\*\* Yes, this has been taken into account. As noted, protein secretion is mostly a stationary phase phenomenon with proteins accumulating for many hours after cells enter stationary phase. Under our conditions (LB) all strains used grow at very similar rates in liquid cultures as seen in Fig S1B. (This is distinct from the slow growth phenotype reported on solid medium, where the diffusion of nutrients is limiting and optimal growth requires the action of secreted proteases digest the peptides present in the medium). This is clarified where relevant in the text.  
+
+## L107  
+
+The same point as above, although I suspect the interpretation is correct. The length of time in stationary phase could significantly affect secretory protein production and a better experiment would have been to take samples at a fixed time following the transition from exponential to stationary phase. The \(>5\) - fold reduction in extracellular Mn is to be expected from previous work on these genes as Mn exporters.  
+
+\*\* As noted above, Fig. S1B demonstrates identical growth kinetics so the time in stationary phase is the same. We agree that the reduction is Mn in the mutants is consistent with the role of these proteins as Mn exporters.  
+
+## L117  
+
+As above, and AmyQ (which requires Ca ions for folding) certainly accumulates in stationary phase.  
+
+\*\* As above, samples were taken at a fixed time following stationary phase since growth rates are so similar.  
+
+## L127  
+
+It is well known that AmyQ induces secretion stress as compared with native enzymes. As a Ca requiring enzymes it is not clear what the significance of these findings are. I think the most significant finding here is that secretion stress is not induced in the FY mutant, which I find surprising if Mn requiring enzymes are unable to fold/fold rapidly.  
+
+\*\* As the referee correctly notes, high levels of AmyQ induces the secretion stress response. However, in the FY mutant strain the secretion of AmyQ is compromised (Fig. 2e), an effect we attribute to translocon jamming. Since the level of secreted protein is decreased, one would not expect induction of the secretion stress response.  
+
+## L142  
+
+It could be argued that the results in Table 1 indicate that the MeeF and MeeY (membrane proteins themselves) simply interact with other membrane proteins. I don't see a clear justification, based on these data, for claiming MeeF and MeeY appear to function as part of the secretosome.  
+
+\*\* We understand the point raised but we respectfully disagree in light of the data that support a colocalization with components of the secretosome. The exact composition of the secretion complex is still under investigation,
+
+<--- Page Split --->
+
+but most papers agree that there are core components of the secretion apparatus that interact reversibly with accessory factors such as YidC (for membrane proteins), foldases, and even the F1Fo ATPase. These are precisely the proteins we find in the Co- IP studies, whereas the many other membrane complexes and proteins are not detected. We have added a footnote to the relevant data table (Table 1) to point out that the membrane proteins we detect are a selected subset (12/40) of the most abundant membrane proteins.  
+
+We have also added citation to a prior study (Turkovicova et al.; PMID 26778143) that looked at the interactome of E. coli TerC using both Co- IP and blue native PAGE. Inspection of their results (in SI datesets) reveals that the complexes they see in blue native PAGE include TerC(Alx) together with many of the same secretosome components we identified.  
+
+L156  
+
+Same point as above about ONCs and secretory proteins.  
+
+\*\* The referee raises the point that perhaps the decrease in secreted proteases in the ftsH mutant strain results from slower growth and therefore a reduced amount of time in stationary phase. Under our conditions, the WT and the ftsH mutant strain grow similarly and reach stationary phase at about the same time, as now shown in Fig. S1b  
+
+L168  
+
+Despite my comments about the way protease levels were estimated, the potential role of MeeF and MeeY in providing Mn as a folding factor helping to "pull" proteins from the translocase would be consistent with the role of FtsH clearing the translocase in the FY mutants.  
+
+\*\* We agree that the strong epistasis observed with the ftsH mutation is supportive of translocon jamming when metalation of nascent proteins is impaired.  
+
+L204  
+
+The data on Mn addition to the medium in relation to LtaS seem clear. However, the medium (as indicated above) already contains some Mn that, in the wild type, would be at a higher concentration at the membrane/wall interface due to its mobile interaction with the phosphate in LTA. Could the influence of Mn addition on LtaS activity in the FY mutant be simply the result of the controlling the level of Mn added to the medium to "just enough" (as indicated above and the influence on the WT as indicated below) for the wild type but not enough for the mutant. If LtaS folding was facilitated simply by the level of Mn in the medium, and this was lower in the FY mutant due to Mn retention by the cell, then the rate of folding of LtaS at lower Mn concentrations could slower, leading to its removal by quality control proteases. Additional Mn allow this to recover LTA synthesis.  
+
+\*\* We think this situation is complex, and that some proteins (e.g. LtaS) may be folded and released into the membrane even in the absence of metalation. The inactive LtaS can then be metalated by simply increasing Mn availability in the medium. For some other proteins, including perhaps metalloproteases, an inability to metalate the nascent protein may lead to translocon jamming. Unfortunately, there are no clearly defined assays for the phenomenon of translocon jamming, nor robust mechanisms for identification of "jammed" substrates.  
+
+Addressing these challenges will require further study. However, we have more fully described the model for the role of TerC proteins in protein metalation in the Discussion.  
+
+L264  
+
+From the data presented I think it is too early to argue that MeeF and MeeY are accessory subunits of the holotranslocon.  
+
+\*\* We understand the concern. However, the data we present based on Co- IP, defects in secretion, and epistasis with ftsH are consistent with the types of data used previously to assign other proteins as accessory subunits (part of a larger secretosome complex) that forms around the holotranslocon. This conclusion is also supported by prior work on E. coli TerC (Turkovicova et al.). Further, the Arabidopsis homolog (AtTerC) also interacts with a YidC homolog (ALB3) (Schneider, A. et al. 2014), consistent with our model.
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+To address the first question raised in the initial review, the authors should add a sentence at a prominent location in the manuscript to make readers aware that:  
+
+"Future studies will be needed to establish how a weak binding metal such as Mn(II) is retained by the active sites of enzymes located outside the plasma- membrane post metalation by MeeF/Y: Possibilities include oxidation to less exchangeable Mn(III), kinetic trapping post folding or repeated insertion by MeeF/Y."  
+
+Inclusion of such a statement is important if the paper is published in Nature Communications with its broad readership, many of whom may have limited knowledge of bioinorganic chemistry and may be left with the impression that the delivery of Mn(II) once to an active site located outside the plasma- membrane (such as in LtaS) fully resolves the challenge (set out in the introduction and opening paragraphs of the discussion) to metalate exoenzymes with lower affinity metals in such a location where buffered metal concentrations (activities) are not controlled.  
+
+Reviewer #2 (Remarks to the Author):  
+
+Excellent work team. The revised version of the manuscript adequately addresses my comments and requests from the previous round of review.  
+
+Reviewer #3 (Remarks to the Author):  
+
+This revised manuscript improves and clarifies aspects of the original. The work is novel and the experimental work is appropriate and justifies the conclusions. It represents significant contribution to an important component of the protein secretion pathway, by providing a mechanism for the metalation of proteins on the trans side of the membrane with Mn. This manuscript is of broad significance, and particularly important in relation to the essential process of secretion and the exploitation of this process for the production of industrial enzymes.
+
+<--- Page Split --->
+
+## REVIEWERS' COMMENTS  
+
+Reviewer #1 (Remarks to the Author):  
+
+To address the first question raised in the initial review, the authors should add a sentence at a prominent location in the manuscript to make readers aware that:  
+
+"Future studies will be needed to establish how a weak binding metal such as Mn(II) is retained by the active sites of enzymes located outside the plasma- membrane post metalation by MeeF/Y: Possibilities include oxidation to less exchangeable Mn(III), kinetic trapping post folding or repeated insertion by MeeF/Y."  
+
+Inclusion of such a statement is important if the paper is published in Nature Communications with its broad readership, many of whom may have limited knowledge of bioinorganic chemistry and may be left with the impression that the delivery of Mn(II) once to an active site located outside the plasma- membrane (such as in LtaS) fully resolves the challenge (set out in the introduction and opening paragraphs of the discussion) to metalate exoenzymes with lower affinity metals in such a location where buffered metal concentrations (activities) are not controlled. We added the relative sentences in the paper (line 335): "Future studies will be needed to establish how weak binding metals such as Mn are retained by exoenzymes. For some proteins, Mn may be oxidized to less exchangeable Mn(III), or the bound Mn may be kinetically trapped after protein folding. Alternatively, MeeF and MeeY may be able to repeatedly load metal into those proteins that are retained in the membrane (LtaS) or in the vicinity."  
+
+Reviewer #2 (Remarks to the Author):  
+
+Excellent work team. The revised version of the manuscript adequately addresses my comments and requests from the previous round of review. We really appreciate former comments and suggestions.  
+
+Reviewer #3 (Remarks to the Author):  
+
+This revised manuscript improves and clarifies aspects of the original. The work is novel and the experimental work is appropriate and justifies the conclusions. It represents significant contribution to an important component of the protein secretion pathway, by providing a mechanism for the metalation of proteins on the trans side of the membrane with Mn. This manuscript is of broad significance, and particularly important in relation to the essential process of secretion and the exploitation of this process for the production of industrial enzymes.  
+
+Thank you for these comments!
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__1ff816831c1878fcdce6de0a4d1b9390aa7f2f7e5804b1700b0dca50f70ee965/supplementary_0_Peer Review File__1ff816831c1878fcdce6de0a4d1b9390aa7f2f7e5804b1700b0dca50f70ee965_det.mmd

@@ -0,0 +1,522 @@
+<|ref|>title<|/ref|><|det|>[[60, 40, 505, 90]]<|/det|>
+# nature portfolio  
+
+<|ref|>text<|/ref|><|det|>[[66, 110, 362, 139]]<|/det|>
+Peer Review File  
+
+<|ref|>image<|/ref|><|det|>[[56, 732, 240, 783]]<|/det|>  
+
+<|ref|>text<|/ref|><|det|>[[250, 732, 911, 784]]<|/det|>
+Open Access This file is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. In the cases where the authors are anonymous, such as is the case for the reports of anonymous peer reviewers, author attribution should be to 'Anonymous Referee' followed by a clear attribution to the source work. The images or other third party material in this file are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[116, 88, 305, 104]]<|/det|>
+## REVIEWER COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 116, 404, 131]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 144, 880, 394]]<|/det|>
+How the correct metals locate to the correct proteins is a puzzle. Moreover, this is an important puzzle since almost half the reactions of life are estimated to be metal catalysed and typically alternative metals will not drive catalysis. The puzzle is further complicated by the observation that most metalloproteins bind one or more wrong metals many orders of magnitude more tightly than the required metal. For cytosolic proteins the puzzle is substantially resolved by the actions of metal homeostatic systems: These systems maintain the availabilities of the tightest binding metals much lower than the availabilities of the weaker binding ones and so it becomes possible to simultaneously have proteins that require tight binding metals and those that require weak binding ones in the same cytosolic compartment. This becomes more challenging however outside the plasma- membrane, especially of Gram positive bacteria, since this location is essentially continuous with the external environment. Intriguingly, in this manuscript we learn that even manganese- proteins (available divalent manganese being a relatively weak binding metal) that are secreted in an unfolded state, still take advantage of the homeostatic mechanisms that control cytosolic metal availabilities to the inverse of the Universal affinity series (the Irving Williams series). Previously, we only knew this to be true of manganese- proteins that are secreted in a folded state via the TAT system. In this manner metal specificity becomes the product of the multiple partitioning events that control cytosolic metal availabilities for both TAT and Sec substrates. This is an significant advance which adds to a body of literature that needs to be brought to the attention of the widest possible audience.  
+
+<|ref|>text<|/ref|><|det|>[[115, 406, 876, 476]]<|/det|>
+Do we know if manganese becomes kinetically trapped in Lipoteichoic acid synthase post- folding, or if the exchange of manganese with a tighter binding metal such as zinc is slow? Could this be experimentally addressed or alternatively could the authors add further comments on how, post MeeF/Y- assisted metalation and secretion, subsequent exchange of manganese for a tighter binding metal is avoided.  
+
+<|ref|>text<|/ref|><|det|>[[115, 488, 880, 545]]<|/det|>
+Is it known what fraction of the Sec secretome is composed of LTA? Which data exclude the possibility that MeeF/Y act directly in the secretion of LTA but only indirectly exert effects on manganese because LTA is a major sink for extra plasma- membrane manganese (perhaps bound in less- exchangeable oxidised forms)?  
+
+<|ref|>text<|/ref|><|det|>[[115, 558, 880, 587]]<|/det|>
+Could the inhibitory phenotypes related to zinc be a consequence of known inhibitory effects of zinc on the mechanisms of manganese import as documented by Chris McDevitt and co- workers?  
+
+<|ref|>text<|/ref|><|det|>[[115, 599, 877, 655]]<|/det|>
+line 251. Metalation of CucA was subsequently shown to require two copper transporting ATPases and a cytosolic metallochaperone, with CucA secretion being absent in mutants lacking these proteins suggesting that metalation is assisted and co- coincident with secretion (J Biol Chem 2010 285: 32504- 32511).  
+
+<|ref|>text<|/ref|><|det|>[[115, 668, 880, 739]]<|/det|>
+The use of "suggest" and "possibly" in the closing sentence: 'These results "suggest" that TerC proteins are important in both bacteria and eukaryotes for the proper functioning of exported proteins, "possibly" by mediating the co- translocational insertion of metals into nascent proteins during transit across membranes' is more cautious than elsewhere and perhaps some statements need to be toned- down and/or caveats added.  
+
+<|ref|>text<|/ref|><|det|>[[116, 779, 404, 794]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 807, 874, 904]]<|/det|>
+The manuscript by He and colleagues describes a novel role for TerC proteins in the delivery of manganese to proteins exported via the SEC translocase. The findings built on their own analyses which defined TerC as manganese export systems, and the authors introduce these proteins as possibly functionally conserved across multiple kingdoms of life. The work performed here is of a high standard and the manuscript is very clearly written. The experiments performed here are logical and provide the necessary information required to draw conclusions regarding the proposed novel role for TerC in B. subtilis. However, the brevity of the manuscript presents a limitation in terms of drawing
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 790, 105]]<|/det|>
+broader conclusions across bacteria and definitely when considering other kingdoms of life.  
+
+<|ref|>text<|/ref|><|det|>[[115, 118, 239, 131]]<|/det|>
+Main comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 132, 867, 188]]<|/det|>
+This work may benefit from some in silico work, including the genetic organisation and conservation across the species, genus, between other bacteria and across kingdoms. Structural models are likely to assist too. These don't need to be described in detail in the main body of the text, but may appeal to those that are not in the field of Bacillus research.  
+
+<|ref|>text<|/ref|><|det|>[[115, 200, 861, 243]]<|/det|>
+Expanding upon this, I would like to learn more about why MeeF appears to have a slightly stronger functional role than MeeY, and why more interacting proteins were identified. Other than the \(40\%\) similarity, what's different: expression, structure, distribution across the cell surface?  
+
+<|ref|>text<|/ref|><|det|>[[115, 255, 848, 283]]<|/det|>
+Why is YjbE not contributing to the processes described here, and how may it fulfill its role during sporulation? Please discuss.  
+
+<|ref|>text<|/ref|><|det|>[[115, 283, 850, 312]]<|/det|>
+How are TerC proteins and their role in metalation of secreted proteins impacted by Mn starvation, something that bacterial pathogens readily encounter during infection?  
+
+<|ref|>text<|/ref|><|det|>[[115, 339, 245, 352]]<|/det|>
+Minor comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 353, 864, 381]]<|/det|>
+Line 64: Please don't specify the name of the mutant in the introduction, leave that until the results, where it has already been clearly stated too.  
+
+<|ref|>text<|/ref|><|det|>[[115, 381, 686, 395]]<|/det|>
+Line 75: This statement is superfluous, as this can be derived from the intro.  
+
+<|ref|>text<|/ref|><|det|>[[115, 395, 507, 408]]<|/det|>
+Line 643: Is it clearer bands, or simply more bands?  
+
+<|ref|>text<|/ref|><|det|>[[115, 409, 857, 450]]<|/det|>
+Line 101: Listing the molecular weights of the proteins seems confusing and Fig 1d doesn't actually show the specifics of those proteases listed, only that of the delta7 strain. Hence, denoting specific candidates seems appropriate for Fig S2 alone.  
+
+<|ref|>text<|/ref|><|det|>[[115, 450, 861, 478]]<|/det|>
+Line 654: Please indicate total samples (were multiple biological replicates included on one day, as I assume that technical replicates are not displayed)  
+
+<|ref|>text<|/ref|><|det|>[[115, 478, 825, 492]]<|/det|>
+Line 212: Please use these references for the strains with FLAG tag, as it may cause confusion.  
+
+<|ref|>text<|/ref|><|det|>[[115, 492, 504, 505]]<|/det|>
+Line 138- 139: This should be the other way around.  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 432, 519]]<|/det|>
+Line 803: This sentence seems incomplete.  
+
+<|ref|>text<|/ref|><|det|>[[115, 519, 740, 533]]<|/det|>
+Line 145: Please define that the MeeF interactome is broader than that of the MeeY.  
+
+<|ref|>text<|/ref|><|det|>[[115, 533, 832, 561]]<|/det|>
+Line 811- 812: I would say that the pattern is similar, it's the difference in intensity that is more striking.  
+
+<|ref|>text<|/ref|><|det|>[[115, 561, 880, 589]]<|/det|>
+Lines 183- 185: Bit odd to explain what they do after already defining the result, restructure this order. Fig S8: Relative transcript to what? Scale is confusing. Why only two reps?  
+
+<|ref|>text<|/ref|><|det|>[[115, 589, 470, 602]]<|/det|>
+Line 209: Please ascertain this transcriptionally.  
+
+<|ref|>text<|/ref|><|det|>[[115, 603, 872, 630]]<|/det|>
+Lines 229- 236: This really needs bioinformatics to support the choice of those two candidates and the claim that it's conserved.  
+
+<|ref|>text<|/ref|><|det|>[[115, 631, 407, 644]]<|/det|>
+Lines 239- 246: This can be condensed.  
+
+<|ref|>text<|/ref|><|det|>[[115, 645, 275, 657]]<|/det|>
+Line 259: delete "its".  
+
+<|ref|>text<|/ref|><|det|>[[115, 658, 840, 700]]<|/det|>
+Line 266: Is Mn structural or any role in folding, as lack of metalation seems like an insignificant reason to not release the protein by the translocase? Or perhaps degradation is favourable as the protein would be non- functional, please discuss.  
+
+<|ref|>text<|/ref|><|det|>[[115, 700, 696, 714]]<|/det|>
+The legends of Fig 2D and S2 appear contradictory regarding the role of AprE.  
+
+<|ref|>text<|/ref|><|det|>[[115, 752, 404, 766]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 780, 879, 905]]<|/det|>
+This MS addresses the important issue of the metalation of extracytoplasmic proteins. This is important because many secretory Gram- positive proteins are metalloproteins or require metal ions for folding following translocation across the membrane (via the Sec translocase) in an essential unfolded/unstructured form. Slowly folding proteins are subject to proteolysis by quality control proteases and metal ions are known to act as folding factors, speeding up the folding at least for some proteins. This MS provides a mechanism for the metalation by the lower affinity metal ion Mn, so as to potentially optimise folding and avoid mis- metalation by higher affinity metal ions. There is a suggestion that folding provides an energy force that helps "pull" for the protein from the translocase and so metalation my help in this process.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 861, 160]]<|/det|>
+The current view is that metalation occurs in most cases via the concentration (cf the medium) of metal ions in the negatively charged cell wall. However, this does not allow for mis- metalation for lower affinity metal ions. Therefore, the hypothesis proposed in this manuscript is attractive. I do, however, have some comments on what I see as potential experimental limitations that the authors should address. These are indicated below.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 173, 221, 187]]<|/det|>
+## L104/Fig 1c/d  
+
+<|ref|>text<|/ref|><|det|>[[115, 187, 880, 312]]<|/det|>
+The legend states that "Supernatants were collected from ONCs with the same cell number(s)". What medium was use and did all the cultures reach stationary phase with the same numbers of cells and at the same time? I might have expected the FY mutant to have lower numbers or to have a slower in some media. I am also cautious about simply using data derived from overnight cultures (ONCs) without considering growth kinetics. Bacillus continues to produce many secretory enzymes in stationary phase and therefore cells that enter stationary phase earlier than others could generate more enzyme during their longer stationary phase. Therefore, the lower protease activity from the mutant could be due to a slower growth rate (ie reaching stationary phase later) and a resulting shorter time in stationary phase. Has this been taken into account?  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 325, 159, 338]]<|/det|>
+## L 107  
+
+<|ref|>text<|/ref|><|det|>[[115, 339, 875, 409]]<|/det|>
+The same point as above, although I suspect the interpretation is correct. The length of time in stationary phase could significantly affect secretory protein production and a better experiment would have been to take samples at a fixed time following the transition from exponential to stationary phase. The \(>5\) - fold reduction in extracellular Mn is to be expected from previous work on these genes as Mn exporters.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 421, 155, 434]]<|/det|>
+## L117  
+
+<|ref|>text<|/ref|><|det|>[[115, 435, 856, 450]]<|/det|>
+As above, and AmyQ (which requires Ca ions for folding) certainly accumulates in stationary phase.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 463, 155, 475]]<|/det|>
+## L127  
+
+<|ref|>text<|/ref|><|det|>[[115, 476, 872, 533]]<|/det|>
+It is well known that AmyQ induces secretion stress as compared with native enzymes. As a Ca requiring enzymes it is not clear what the significance of these findings are. I think the most significant finding here is that secretion stress is not induced in the FY mutant, which I find surprising if Mn requiring enzymes are unable to fold/fold rapidly.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 546, 155, 558]]<|/det|>
+## L142  
+
+<|ref|>text<|/ref|><|det|>[[115, 559, 866, 601]]<|/det|>
+It could be argued that the results in Table 1 indicate that the MeeF and MeeY (membrane proteins themselves) simply interact with other membrane proteins. I don't see a clear justification, based on these data, for claiming MeeF and MeeY appear to function as part of the secretosome.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 615, 155, 627]]<|/det|>
+## L156  
+
+<|ref|>text<|/ref|><|det|>[[115, 628, 541, 642]]<|/det|>
+Same point as above about ONCs and secretory proteins.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 656, 155, 668]]<|/det|>
+## L168  
+
+<|ref|>text<|/ref|><|det|>[[115, 669, 864, 711]]<|/det|>
+Despite my comments about the way protease levels were estimated, the potential role of MeeF and MeeY in providing Mn as a folding factor helping to "pull" proteins from the translocase would be consistent with the role of FtsH clearing the translocase in the FY mutants.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 725, 155, 737]]<|/det|>
+## L204  
+
+<|ref|>text<|/ref|><|det|>[[115, 738, 876, 862]]<|/det|>
+The data on Mn addition to the medium in relation to LtaS seem clear. However, the medium (as indicated above) already contains some Mn that, in the wild type, would be at a higher concentration at the membrane/wall interface due to its mobile interaction with the phosphate in LTA. Could the influence of Mn addition on LtaS activity in the FY mutant be simply the result of the controlling the level of Mn added to the medium to "just enough" (as indicated above and the influence on the WT as indicated below) for the wild type but not enough for the mutant. If LtaS folding was facilitated simply by the level of Mn in the medium, and this was lower in the FY mutant due to Mn retention by the cell, then the rate of folding of LtaS at lower Mn concentrations could slower, leading to its removal by quality control proteases. Additional Mn allow this to recover LTA synthesis.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[113, 90, 872, 119]]<|/det|>
+From the data presented I think it is too early to argue that MeeF and MeeY are accessory subunits of the holotranslocon
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[116, 90, 371, 104]]<|/det|>
+## RESPONSE TO REVIEWER COMMENTS  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 121, 372, 136]]<|/det|>
+## Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[113, 152, 877, 400]]<|/det|>
+How the correct metals locate to the correct proteins is a puzzle. Moreover, this is an important puzzle since almost half the reactions of life are estimated to be metal catalysed and typically alternative metals will not drive catalysis. The puzzle is further complicated by the observation that most metallo- proteins bind one or more wrong metals many orders of magnitude more tightly than the required metal. For cytosolic proteins the puzzle is substantially resolved by the actions of metal homeostatic systems: These systems maintain the availabilities of the tightest binding metals much lower than the availabilities of the weaker binding ones and so it becomes possible to simultaneously have proteins that require tight binding metals and those that require weak binding ones in the same cytosolic compartment. This becomes more challenging however outside the plasma- membrane, especially of Gram positive bacteria, since this location is essentially continuous with the external environment. Intriguingly, in this manuscript we learn that even manganese- proteins (available divalent manganese being a relatively weak binding metal) that are secreted in an unfolded state, still take advantage of the homeostatic mechanisms that control cytosolic metal availabilities to the inverse of the Universal affinity series (the Irving Williams series). Previously, we only knew this to be true of manganese- proteins that are secreted in a folded state via the TAT system. In this manner metal specificity becomes the product of the multiple partitioning events that control cytosolic metal availabilities for both TAT and Sec substrates. This is an significant advance which adds to a body of literature that needs to be brought to the attention of the widest possible audience.  
+
+<|ref|>text<|/ref|><|det|>[[116, 397, 821, 414]]<|/det|>
+\*\* We appreciate the detailed summary of the referee who has judged this work to be highly significant.  
+
+<|ref|>text<|/ref|><|det|>[[115, 428, 870, 490]]<|/det|>
+Do we know if manganese becomes kinetically trapped in Lipoteichoic acid synthase post- folding, or if the exchange of manganese with a tighter binding metal such as zinc is slow? Could this be experimentally addressed or alternatively could the authors add further comments on how, post MeeF/Y- assisted metalation and secretion, subsequent exchange of manganese for a tighter binding metal is avoided.  
+
+<|ref|>text<|/ref|><|det|>[[115, 490, 872, 568]]<|/det|>
+\*\* We agree that the problem of mismetallation is important and that the Mn bound LtaS enzyme might, under some conditions, be inhibited by tighter binding metals. We address this experimentally by showing that mutants deficient in metalation of LtaS with Mn (FY mutants) are sensitive to inhibition by Zn, consistent with a mismetallation mechanism (Figure S9). Further studies of metal exchange after metal loading will require future study.  
+
+<|ref|>text<|/ref|><|det|>[[115, 583, 561, 598]]<|/det|>
+Is it known what fraction of the Sec secretome is composed of LTA?  
+
+<|ref|>text<|/ref|><|det|>[[115, 599, 872, 660]]<|/det|>
+\*\* We believe the referee meant to inquire about how abundant the LtaS protein is within the Sec secretome. The secretome of B. subtilis includes both extracellular enzymes and proteins destined for the membrane (the membrane proteome). The text has been revised to clarify that LtaS is a low abundance membrane protein (line 194), as shown previously (PMID 31424929).  
+
+<|ref|>text<|/ref|><|det|>[[115, 675, 870, 721]]<|/det|>
+Which data exclude the possibility that MeeF/Y act directly in the secretion of LTA but only indirectly exert effects on manganese because LTA is a major sink for extra plasma- membrane manganese (perhaps bound in less- exchangeable oxidised forms)?  
+
+<|ref|>text<|/ref|><|det|>[[115, 722, 874, 799]]<|/det|>
+\*\* LTA (lipoteichoic acid) is synthesized extracellularly by the LtaS enzyme (it is not a secreted product). MeeF/Y are homologous to ion transporters, are often regulated by Mn- inducible riboswitches, and have been shown to function in Mn export from cells both in B. subtilis (PMID: 31685536) and more recently in E. coli (PMID: 37214827). LTA polymers are not secreted, but are assembled on the cell surface, where they indeed function in metal buffering (lines 326- 329).  
+
+<|ref|>text<|/ref|><|det|>[[115, 813, 872, 905]]<|/det|>
+Could the inhibitory phenotypes related to zinc be a consequence of known inhibitory effects of zinc on the mechanisms of manganese import as documented by Chris McDevitt and co- workers? \*\* We believe the referee is referring to the combined effect of Zn and the LtaS inhibitor (1771) as reported in Figure S9. These results demonstrate that Zn intoxication is specific to cells that are deficient in LTA synthesis (through mutation of one or more LtaS isozymes) and chemical inhibition of LtaS by 1771. This is unrelated to mechanism of inhibition cited in Chris McDevitt and coworkers, which refers to the ability of Zn to competitively
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 90, 878, 137]]<|/det|>
+inhibit Mn uptake through the MntABCD ABC transporter (PMID 22072971). This mechanism is not operative in B. subtilis since the major Mn importer is MntH, which is not inhibited by Zn. Instead, high Zn interferes with the electron transport chain as shown in B. subtilis (PMID: 27935957) and in E. coli (PMID: 7557331).  
+
+<|ref|>text<|/ref|><|det|>[[115, 152, 872, 197]]<|/det|>
+line 251. Metalation of CucA was subsequently shown to require two copper transporting ATPases and a cytosolic metallochaperone, with CucA secretion being absent in mutants lacking these proteins suggesting that metalation is assisted and co- coincident with secretion (J Biol Chem 2010 285: 32504- 32511).  
+
+<|ref|>text<|/ref|><|det|>[[115, 199, 810, 214]]<|/det|>
+\*\* Thank you for this clarification. We have limited our discussion of the Mn- dependent cupin (MncA).  
+
+<|ref|>text<|/ref|><|det|>[[115, 229, 880, 288]]<|/det|>
+The use of "suggest" and "possibly" in the closing sentence: 'These results "suggest" that TerC proteins are important in both bacteria and eukaryotes for the proper functioning of exported proteins, "possibly" by mediating the co- translocational insertion of metals into nascent proteins during transit across membranes' is more cautious than elsewhere and perhaps some statements need to be toned- down and/or caveats added.  
+
+<|ref|>text<|/ref|><|det|>[[115, 289, 880, 321]]<|/det|>
+\*\* The Introduction and Discussion sections have been re- organized and expanded to clarify our conclusions (see referee 2, first comment).  
+
+<|ref|>sub_title<|/ref|><|det|>[[116, 351, 372, 366]]<|/det|>
+## Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 382, 880, 505]]<|/det|>
+The manuscript by He and colleagues describes a novel role for TerC proteins in the delivery of manganese to proteins exported via the SEC translocase. The findings built on their own analyses which defined TerC as manganese export systems, and the authors introduce these proteins as possibly functionally conserved across multiple kingdoms of life. The work performed here is of a high standard and the manuscript is very clearly written. The experiments performed here are logical and provide the necessary information required to draw conclusions regarding the proposed novel role for TerC in B. subtilis. However, the brevity of the manuscript presents a limitation in terms of drawing broader conclusions across bacteria and definitely when considering other kingdoms of life.  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 875, 537]]<|/det|>
+\*\* We appreciate the referee's supportive assessment. We have amended the text to expand and clarify our Discussion and further develop those ideas that may not have been clear due to the brevity of our presentation.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 553, 228, 565]]<|/det|>
+## Main comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 568, 878, 628]]<|/det|>
+This work may benefit from some in silico work, including the genetic organisation and conservation across the species, genus, between other bacteria and across kingdoms. Structural models are likely to assist too. These don't need to be described in detail in the main body of the text, but may appeal to those that are not in the field of Bacillus research.  
+
+<|ref|>text<|/ref|><|det|>[[115, 630, 880, 783]]<|/det|>
+\*\* We have added a new panel (Fig. 4a) to highlight the phylogenetic relatedness of the proteins tested in Fig. 4. We agree that a much broader in silico analysis will be valuable, although considerable efforts of this type are already published. Protein sequence similarity has been used to argue that the bacterial TerC proteins are part of a larger family of ion transporters as discussed in prior in silico work and now cited (lines 283, 340): "TerC proteins (Pfam03741) are a subgroup of the lysine exporter (LysE) superfamily of transporters and have seven TM segments and a conserved metal- binding site \(^{14,70}\) and "Functional studies have linked diverse TerC and related UPF0016 proteins to the transport of Mn and Ca \(^{4,19,87,88}\) . Readers interested in the conservation across species and kingdoms can refer to these detailed analyses and the public resources that describe PFAM/UPF protein alignments and families. Further insights into functional conservation requires further study in diverse systems, and experimental determination of structures and mapping of the ion channels in model proteins.  
+
+<|ref|>text<|/ref|><|det|>[[115, 799, 878, 845]]<|/det|>
+Expanding upon this, I would like to learn more about why MeeF appears to have a slightly stronger functional role than MeeY, and why more interacting proteins were identified. Other than the \(40\%\) similarity, what's different: expression, structure, distribution across the cell surface?  
+
+<|ref|>text<|/ref|><|det|>[[115, 846, 880, 907]]<|/det|>
+\*\* Our current model for MeeF/Y function is that these proteins allow the transit of Mn from the cytosol to the cell surface, where the ion is held by predicted metal- binding sites on extracytoplasmic loops. These loops may allow for specific delivery of Mn to client proteins as they begin to fold upon exit from the secretory channel. While MeeF/Y are redundant in function for many of the phenotypes noted, they are not identical in function (as
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 868, 152]]<|/det|>
+the referee notes). Indeed, for some phenotypes we believe that there are dramatic differences in role, which provide a tool for better understanding the differentiation of roles for these two proteins in future studies. Our current thinking regarding the biological significance of having two paralogs is now further explained in the Discussion section (line 288- 300).  
+
+<|ref|>text<|/ref|><|det|>[[115, 167, 850, 198]]<|/det|>
+Why is YjbE not contributing to the processes described here, and how may it fulfill its role during sporulation? Please discuss.  
+
+<|ref|>text<|/ref|><|det|>[[115, 198, 870, 276]]<|/det|>
+\*\* The third paralog (YjbE) is selectively expressed during sporulation according to Nicolas et al. as cited. While this protein plays a minor role under the conditions studied here, we do believe that it is likely more important during sporulation. We have added new data in Fig. 1a that demonstrate that YjbE is not contributing to fitness under the conditions studied here. YjbE is now discussed explicitly in the discussion section (see comment above).  
+
+<|ref|>text<|/ref|><|det|>[[115, 290, 875, 322]]<|/det|>
+How are TerC proteins and their role in metalation of secreted proteins impacted by Mn starvation, something that bacterial pathogens readily encounter during infection?  
+
+<|ref|>text<|/ref|><|det|>[[115, 322, 867, 399]]<|/det|>
+\*\* The phenotypes described here are observed under conditions of relatively low Mn availability. LB medium has just enough Mn to support growth, and this is probably one reason why we observed the phenotypes reported. As host cells limit Mn during infection (e.g. through the action of calprotectin) the role of TerC family proteins is selectively directing Mn to surface enzymes (e.g. LtaS) will become even more critical than for cells growing in the presence of excess Mn. This is now discussed (l. 300- 303).  
+
+<|ref|>text<|/ref|><|det|>[[115, 414, 235, 428]]<|/det|>
+Minor comments:  
+
+<|ref|>text<|/ref|><|det|>[[115, 429, 870, 460]]<|/det|>
+Line 64: Please don't specify the name of the mutant in the introduction, leave that until the results, where it has already been clearly stated too.  
+
+<|ref|>text<|/ref|><|det|>[[115, 460, 870, 491]]<|/det|>
+\*\* The mutant name was deleted from the line as requested. The mutant strain is described at the beginning of results.  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 618, 521]]<|/det|>
+Line 75: This statement is superfluous, as this can be derived from the intro.  
+
+<|ref|>text<|/ref|><|det|>[[117, 522, 325, 536]]<|/det|>
+\*\* The statement was deleted.  
+
+<|ref|>text<|/ref|><|det|>[[115, 552, 457, 566]]<|/det|>
+Line 643: Is it clearer bands, or simply more bands?  
+
+<|ref|>text<|/ref|><|det|>[[115, 567, 858, 614]]<|/det|>
+\*\* The statement in question, "Higher protease activities correspond to clearer bands on the gel matrix." was deleted since it is not needed. Samples with greater protease activity may have more bands, greater clearing (activity) in each band, or some combination.  
+
+<|ref|>text<|/ref|><|det|>[[115, 629, 839, 675]]<|/det|>
+Line 101: Listing the molecular weights of the proteins seems confusing and Fig 1d doesn't actually show the specifics of those proteases listed, only that of the delta7 strain. Hence, denoting specific candidates seems appropriate for Fig S2 alone.  
+
+<|ref|>text<|/ref|><|det|>[[115, 676, 852, 707]]<|/det|>
+\*\* We agree and have removed the molecular weights as suggested for Fig. 1d, referring instead to Fig. S2 as suggested.  
+
+<|ref|>text<|/ref|><|det|>[[115, 721, 864, 752]]<|/det|>
+Line 654: Please indicate total samples (were multiple biological replicates included on one day, as I assume that technical replicates are not displayed)  
+
+<|ref|>text<|/ref|><|det|>[[115, 753, 878, 799]]<|/det|>
+\*\* Figures have been revised so that each point represents a separate biological replicate, with the values shown being the average of technical replicates. This is now explained in the figure legends and the original data are in the provided Source Data file.  
+
+<|ref|>text<|/ref|><|det|>[[115, 814, 737, 829]]<|/det|>
+Line 212: Please use these references for the strains with FLAG tag, as it may cause confusion.  
+
+<|ref|>text<|/ref|><|det|>[[115, 830, 857, 860]]<|/det|>
+\*\* We do not see strains mentioned at this point in the text. However, we have reviewed the text carefully to make sure that all references to strains are unambiguous.  
+
+<|ref|>text<|/ref|><|det|>[[115, 876, 457, 905]]<|/det|>
+Line 138- 139: This should be the other way around. \*\* We have amended the sentence for clarity.
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 106, 460, 135]]<|/det|>
+Line 803: This sentence seems incomplete. \*\* We have re- written the figure legend for clarity.  
+
+<|ref|>text<|/ref|><|det|>[[115, 152, 675, 182]]<|/det|>
+Line 145: Please define that the MeeF interactome is broader than that of the MeeY. \*\* We have added a sentence to address this point (l. 157).  
+
+<|ref|>text<|/ref|><|det|>[[115, 198, 795, 228]]<|/det|>
+Line 811- 812: I would say that the pattern is similar, it's the difference in intensity that is more striking. \*\* We agree and have amended the text to clarify this point.  
+
+<|ref|>text<|/ref|><|det|>[[115, 244, 790, 274]]<|/det|>
+Lines 183- 185: Bit odd to explain what they do after already defining the result, restructure this order. \*\* We have rewritten for clarity.  
+
+<|ref|>text<|/ref|><|det|>[[115, 290, 604, 305]]<|/det|>
+Fig S8: Relative transcript to what? Scale is confusing. Why only two reps?  
+
+<|ref|>text<|/ref|><|det|>[[115, 306, 875, 336]]<|/det|>
+\*\* This experiment now has 4 biological replicates shown and we have clarified that all values are normalized to the gyrA transcript.  
+
+<|ref|>text<|/ref|><|det|>[[115, 352, 430, 366]]<|/det|>
+Line 209: Please ascertain this transcriptionally.  
+
+<|ref|>text<|/ref|><|det|>[[115, 368, 821, 397]]<|/det|>
+\*\* We did not mean to imply transcriptional activation. We have changed the text (l. 229) to say "due to increased activity of the YqgS synthase."  
+
+<|ref|>text<|/ref|><|det|>[[115, 413, 861, 443]]<|/det|>
+Lines 229- 236: This really needs bioinformatics to support the choice of those two candidates and the claim that it's conserved.  
+
+<|ref|>text<|/ref|><|det|>[[115, 444, 783, 459]]<|/det|>
+\*\* A phylogenetic tree is added as panel 4a to provide context for the selection of these orthologs.  
+
+<|ref|>text<|/ref|><|det|>[[115, 475, 421, 504]]<|/det|>
+Lines 239- 246: This can be condensed. \*\* We have edited this paragraph for brevity.  
+
+<|ref|>text<|/ref|><|det|>[[115, 521, 258, 550]]<|/det|>
+Line 259: delete "its". \*\* Corrected  
+
+<|ref|>text<|/ref|><|det|>[[115, 567, 840, 612]]<|/det|>
+Line 266: Is Mn structural or any role in folding, as lack of metalation seems like an insignificant reason to not release the protein by the translocase? Or perhaps degradation is favourable as the protein would be non- functional, please discuss.  
+
+<|ref|>text<|/ref|><|det|>[[115, 614, 879, 643]]<|/det|>
+\*\* The discussion section has been revised to more fully describe the proposed role of TerC proteins in exoenzyme metalation.  
+
+<|ref|>text<|/ref|><|det|>[[115, 660, 628, 675]]<|/det|>
+The legends of Fig 2D and S2 appear contradictory regarding the role of AprE.  
+
+<|ref|>text<|/ref|><|det|>[[115, 676, 879, 722]]<|/det|>
+\*\* The data in S2 suggest that AprE is required for the 17 kD band, assigned as a degradation product of Bpr. This band is still present in the FY mutant in Fig 1d, suggesting that there is at least some AprE secretion, as seen also in the immunoblot analysis of Fig. 2d. Thus, there is no contradiction.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 752, 372, 767]]<|/det|>
+## Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[115, 783, 882, 905]]<|/det|>
+This MS addresses the important issue of the metalation of extracytoplasmic proteins. This is important because many secretory Gram- positive proteins are metalloproteins or require metal ions for folding following translocation across the membrane (via the Sec translocase) in an essential unfolded/unstructured form. Slowly folding proteins are subject to proteolysis by quality control proteases and metal ions are known to act as folding factors, speeding up the folding at least for some proteins. This MS provides a mechanism for the metalation by the lower affinity metal ion Mn, so as to potentially optimise folding and avoid mis- metalation by higher affinity metal ions. There is a suggestion that folding provides an energy force that helps "pull" for the protein from the translocase and so metalation my help in this process. The current view is that metalation occurs in most cases via the concentration
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[115, 89, 869, 151]]<|/det|>
+(cf the medium) of metal ions in the negatively charged cell wall. However, this does not allow for mis- metalation for lower affinity metal ions. Therefore, the hypothesis proposed in this manuscript is attractive. I do, however, have some comments on what I see as potential experimental limitations that the authors should address. These are indicated below.  
+
+<|ref|>text<|/ref|><|det|>[[115, 152, 870, 214]]<|/det|>
+\*\* We thank referee for their comments. We agree with the referee that metal ions buffered by the anionic cell wall provide one source for metalation of nascent proteins. Further, metal ions (and often Ca, specifically) are often invoked as folding factors for extracellular enzymes. The role of LTA in metal buffering is now explicitly discussed in the text (lines 326- 329).  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 245, 209, 259]]<|/det|>
+## L104/Fig 1c/d  
+
+<|ref|>text<|/ref|><|det|>[[115, 260, 881, 383]]<|/det|>
+The legend states that "Supernatants were collected from ONCs with the same cell number(s)". What medium was use and did all the cultures reach stationary phase with the same numbers of cells and at the same time? I might have expected the FY mutant to have lower numbers or to have a slower in some media. I am also cautious about simply using data derived from overnight cultures (ONCs) without considering growth kinetics. Bacillus continues to produce many secretory enzymes in stationary phase and therefore cells that enter stationary phase earlier than others could generate more enzyme during their longer stationary phase. Therefore, the lower protease activity from the mutant could be due to a slower growth rate (ie reaching stationary phase later) and a resulting shorter time in stationary phase. Has this been taken into account?  
+
+<|ref|>text<|/ref|><|det|>[[115, 384, 872, 476]]<|/det|>
+\*\* Yes, this has been taken into account. As noted, protein secretion is mostly a stationary phase phenomenon with proteins accumulating for many hours after cells enter stationary phase. Under our conditions (LB) all strains used grow at very similar rates in liquid cultures as seen in Fig S1B. (This is distinct from the slow growth phenotype reported on solid medium, where the diffusion of nutrients is limiting and optimal growth requires the action of secreted proteases digest the peptides present in the medium). This is clarified where relevant in the text.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 492, 152, 504]]<|/det|>
+## L107  
+
+<|ref|>text<|/ref|><|det|>[[115, 506, 880, 567]]<|/det|>
+The same point as above, although I suspect the interpretation is correct. The length of time in stationary phase could significantly affect secretory protein production and a better experiment would have been to take samples at a fixed time following the transition from exponential to stationary phase. The \(>5\) - fold reduction in extracellular Mn is to be expected from previous work on these genes as Mn exporters.  
+
+<|ref|>text<|/ref|><|det|>[[115, 568, 869, 599]]<|/det|>
+\*\* As noted above, Fig. S1B demonstrates identical growth kinetics so the time in stationary phase is the same. We agree that the reduction is Mn in the mutants is consistent with the role of these proteins as Mn exporters.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 615, 149, 627]]<|/det|>
+## L117  
+
+<|ref|>text<|/ref|><|det|>[[115, 629, 770, 644]]<|/det|>
+As above, and AmyQ (which requires Ca ions for folding) certainly accumulates in stationary phase.  
+
+<|ref|>text<|/ref|><|det|>[[115, 645, 850, 660]]<|/det|>
+\*\* As above, samples were taken at a fixed time following stationary phase since growth rates are so similar.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 677, 149, 689]]<|/det|>
+## L127  
+
+<|ref|>text<|/ref|><|det|>[[115, 691, 880, 752]]<|/det|>
+It is well known that AmyQ induces secretion stress as compared with native enzymes. As a Ca requiring enzymes it is not clear what the significance of these findings are. I think the most significant finding here is that secretion stress is not induced in the FY mutant, which I find surprising if Mn requiring enzymes are unable to fold/fold rapidly.  
+
+<|ref|>text<|/ref|><|det|>[[115, 753, 880, 799]]<|/det|>
+\*\* As the referee correctly notes, high levels of AmyQ induces the secretion stress response. However, in the FY mutant strain the secretion of AmyQ is compromised (Fig. 2e), an effect we attribute to translocon jamming. Since the level of secreted protein is decreased, one would not expect induction of the secretion stress response.  
+
+<|ref|>sub_title<|/ref|><|det|>[[115, 815, 149, 827]]<|/det|>
+## L142  
+
+<|ref|>text<|/ref|><|det|>[[115, 830, 876, 875]]<|/det|>
+It could be argued that the results in Table 1 indicate that the MeeF and MeeY (membrane proteins themselves) simply interact with other membrane proteins. I don't see a clear justification, based on these data, for claiming MeeF and MeeY appear to function as part of the secretosome.  
+
+<|ref|>text<|/ref|><|det|>[[115, 876, 880, 907]]<|/det|>
+\*\* We understand the point raised but we respectfully disagree in light of the data that support a colocalization with components of the secretosome. The exact composition of the secretion complex is still under investigation,
+
+<--- Page Split --->
+<|ref|>text<|/ref|><|det|>[[114, 90, 875, 168]]<|/det|>
+but most papers agree that there are core components of the secretion apparatus that interact reversibly with accessory factors such as YidC (for membrane proteins), foldases, and even the F1Fo ATPase. These are precisely the proteins we find in the Co- IP studies, whereas the many other membrane complexes and proteins are not detected. We have added a footnote to the relevant data table (Table 1) to point out that the membrane proteins we detect are a selected subset (12/40) of the most abundant membrane proteins.  
+
+<|ref|>text<|/ref|><|det|>[[115, 168, 850, 229]]<|/det|>
+We have also added citation to a prior study (Turkovicova et al.; PMID 26778143) that looked at the interactome of E. coli TerC using both Co- IP and blue native PAGE. Inspection of their results (in SI datesets) reveals that the complexes they see in blue native PAGE include TerC(Alx) together with many of the same secretosome components we identified.  
+
+<|ref|>text<|/ref|><|det|>[[115, 245, 149, 258]]<|/det|>
+L156  
+
+<|ref|>text<|/ref|><|det|>[[115, 261, 494, 275]]<|/det|>
+Same point as above about ONCs and secretory proteins.  
+
+<|ref|>text<|/ref|><|det|>[[115, 275, 872, 336]]<|/det|>
+\*\* The referee raises the point that perhaps the decrease in secreted proteases in the ftsH mutant strain results from slower growth and therefore a reduced amount of time in stationary phase. Under our conditions, the WT and the ftsH mutant strain grow similarly and reach stationary phase at about the same time, as now shown in Fig. S1b  
+
+<|ref|>text<|/ref|><|det|>[[115, 353, 149, 365]]<|/det|>
+L168  
+
+<|ref|>text<|/ref|><|det|>[[115, 367, 882, 412]]<|/det|>
+Despite my comments about the way protease levels were estimated, the potential role of MeeF and MeeY in providing Mn as a folding factor helping to "pull" proteins from the translocase would be consistent with the role of FtsH clearing the translocase in the FY mutants.  
+
+<|ref|>text<|/ref|><|det|>[[115, 413, 881, 444]]<|/det|>
+\*\* We agree that the strong epistasis observed with the ftsH mutation is supportive of translocon jamming when metalation of nascent proteins is impaired.  
+
+<|ref|>text<|/ref|><|det|>[[115, 460, 149, 472]]<|/det|>
+L204  
+
+<|ref|>text<|/ref|><|det|>[[115, 475, 872, 598]]<|/det|>
+The data on Mn addition to the medium in relation to LtaS seem clear. However, the medium (as indicated above) already contains some Mn that, in the wild type, would be at a higher concentration at the membrane/wall interface due to its mobile interaction with the phosphate in LTA. Could the influence of Mn addition on LtaS activity in the FY mutant be simply the result of the controlling the level of Mn added to the medium to "just enough" (as indicated above and the influence on the WT as indicated below) for the wild type but not enough for the mutant. If LtaS folding was facilitated simply by the level of Mn in the medium, and this was lower in the FY mutant due to Mn retention by the cell, then the rate of folding of LtaS at lower Mn concentrations could slower, leading to its removal by quality control proteases. Additional Mn allow this to recover LTA synthesis.  
+
+<|ref|>text<|/ref|><|det|>[[115, 599, 880, 675]]<|/det|>
+\*\* We think this situation is complex, and that some proteins (e.g. LtaS) may be folded and released into the membrane even in the absence of metalation. The inactive LtaS can then be metalated by simply increasing Mn availability in the medium. For some other proteins, including perhaps metalloproteases, an inability to metalate the nascent protein may lead to translocon jamming. Unfortunately, there are no clearly defined assays for the phenomenon of translocon jamming, nor robust mechanisms for identification of "jammed" substrates.  
+
+<|ref|>text<|/ref|><|det|>[[115, 675, 880, 704]]<|/det|>
+Addressing these challenges will require further study. However, we have more fully described the model for the role of TerC proteins in protein metalation in the Discussion.  
+
+<|ref|>text<|/ref|><|det|>[[115, 722, 149, 734]]<|/det|>
+L264  
+
+<|ref|>text<|/ref|><|det|>[[115, 737, 816, 765]]<|/det|>
+From the data presented I think it is too early to argue that MeeF and MeeY are accessory subunits of the holotranslocon.  
+
+<|ref|>text<|/ref|><|det|>[[115, 767, 875, 844]]<|/det|>
+\*\* We understand the concern. However, the data we present based on Co- IP, defects in secretion, and epistasis with ftsH are consistent with the types of data used previously to assign other proteins as accessory subunits (part of a larger secretosome complex) that forms around the holotranslocon. This conclusion is also supported by prior work on E. coli TerC (Turkovicova et al.). Further, the Arabidopsis homolog (AtTerC) also interacts with a YidC homolog (ALB3) (Schneider, A. et al. 2014), consistent with our model.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[119, 83, 330, 98]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[119, 111, 415, 125]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[119, 139, 850, 167]]<|/det|>
+To address the first question raised in the initial review, the authors should add a sentence at a prominent location in the manuscript to make readers aware that:  
+
+<|ref|>text<|/ref|><|det|>[[119, 181, 874, 237]]<|/det|>
+"Future studies will be needed to establish how a weak binding metal such as Mn(II) is retained by the active sites of enzymes located outside the plasma- membrane post metalation by MeeF/Y: Possibilities include oxidation to less exchangeable Mn(III), kinetic trapping post folding or repeated insertion by MeeF/Y."  
+
+<|ref|>text<|/ref|><|det|>[[118, 251, 875, 336]]<|/det|>
+Inclusion of such a statement is important if the paper is published in Nature Communications with its broad readership, many of whom may have limited knowledge of bioinorganic chemistry and may be left with the impression that the delivery of Mn(II) once to an active site located outside the plasma- membrane (such as in LtaS) fully resolves the challenge (set out in the introduction and opening paragraphs of the discussion) to metalate exoenzymes with lower affinity metals in such a location where buffered metal concentrations (activities) are not controlled.  
+
+<|ref|>text<|/ref|><|det|>[[119, 377, 415, 391]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 405, 857, 433]]<|/det|>
+Excellent work team. The revised version of the manuscript adequately addresses my comments and requests from the previous round of review.  
+
+<|ref|>text<|/ref|><|det|>[[119, 475, 415, 490]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[118, 503, 861, 589]]<|/det|>
+This revised manuscript improves and clarifies aspects of the original. The work is novel and the experimental work is appropriate and justifies the conclusions. It represents significant contribution to an important component of the protein secretion pathway, by providing a mechanism for the metalation of proteins on the trans side of the membrane with Mn. This manuscript is of broad significance, and particularly important in relation to the essential process of secretion and the exploitation of this process for the production of industrial enzymes.
+
+<--- Page Split --->
+<|ref|>sub_title<|/ref|><|det|>[[116, 88, 360, 108]]<|/det|>
+## REVIEWERS' COMMENTS  
+
+<|ref|>text<|/ref|><|det|>[[116, 123, 446, 141]]<|/det|>
+Reviewer #1 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[116, 158, 819, 194]]<|/det|>
+To address the first question raised in the initial review, the authors should add a sentence at a prominent location in the manuscript to make readers aware that:  
+
+<|ref|>text<|/ref|><|det|>[[116, 210, 867, 282]]<|/det|>
+"Future studies will be needed to establish how a weak binding metal such as Mn(II) is retained by the active sites of enzymes located outside the plasma- membrane post metalation by MeeF/Y: Possibilities include oxidation to less exchangeable Mn(III), kinetic trapping post folding or repeated insertion by MeeF/Y."  
+
+<|ref|>text<|/ref|><|det|>[[113, 297, 874, 525]]<|/det|>
+Inclusion of such a statement is important if the paper is published in Nature Communications with its broad readership, many of whom may have limited knowledge of bioinorganic chemistry and may be left with the impression that the delivery of Mn(II) once to an active site located outside the plasma- membrane (such as in LtaS) fully resolves the challenge (set out in the introduction and opening paragraphs of the discussion) to metalate exoenzymes with lower affinity metals in such a location where buffered metal concentrations (activities) are not controlled. We added the relative sentences in the paper (line 335): "Future studies will be needed to establish how weak binding metals such as Mn are retained by exoenzymes. For some proteins, Mn may be oxidized to less exchangeable Mn(III), or the bound Mn may be kinetically trapped after protein folding. Alternatively, MeeF and MeeY may be able to repeatedly load metal into those proteins that are retained in the membrane (LtaS) or in the vicinity."  
+
+<|ref|>text<|/ref|><|det|>[[116, 557, 446, 575]]<|/det|>
+Reviewer #2 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[116, 592, 861, 647]]<|/det|>
+Excellent work team. The revised version of the manuscript adequately addresses my comments and requests from the previous round of review. We really appreciate former comments and suggestions.  
+
+<|ref|>text<|/ref|><|det|>[[116, 680, 446, 699]]<|/det|>
+Reviewer #3 (Remarks to the Author):  
+
+<|ref|>text<|/ref|><|det|>[[116, 715, 880, 840]]<|/det|>
+This revised manuscript improves and clarifies aspects of the original. The work is novel and the experimental work is appropriate and justifies the conclusions. It represents significant contribution to an important component of the protein secretion pathway, by providing a mechanism for the metalation of proteins on the trans side of the membrane with Mn. This manuscript is of broad significance, and particularly important in relation to the essential process of secretion and the exploitation of this process for the production of industrial enzymes.  
+
+<|ref|>text<|/ref|><|det|>[[117, 840, 390, 856]]<|/det|>
+Thank you for these comments!
+
+<--- Page Split --->

peer_reviews/supplementary_0_Peer Review File__200f5993a10ee2edaa3e86b32ad847d78ef724a6d969c0f7467386d4412eeb78/images_list.json

@@ -0,0 +1 @@
+[]