Distinct roles and actions of PDI family enzymes in catalysis of nascent-chain disulfide formation 1 1 Distinct roles and actions of PDI family enzymes in catalysis of nascent-chain 2 disulfide formation 3 4 Chihiro Hirayama 1 , Kodai Machida 2# , Kentaro Noi 3# , Tadayoshi Murakawa 4 , Masaki 5 Okumura 1,5 , Teru Ogura 6,7 , Hiroaki Imataka 2 , and Kenji Inaba 1* 6 7 1 Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, 8 Sendai, Miyagi 980-8577, Japan 9 2 Graduate School of Engineering, University of Hyogo, Himeji, Hyogo 671-2280, Japan 10 3 Institute for NanoScience Design, Osaka University, Toyonaka, Osaka 560-8531, Japan 11 4 Graduate School of Life Science and Technology, Tokyo Institute of Technology, 12 Yokohama, Kanagawa, 226-8503, Japan 13 5 Frontier Research Institute for Interdisciplinary Sciences, Tohoku University, Sendai, 14 Miyagi 980-8578, Japan 15 6 Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto, 16 Kumamoto 860-0811, Japan 17 7 Faculty of Life Sciences, Kumamoto University, Kumamoto 862-0973, Japan 18 19 # These authors contributed equally to this work 20 21 *Correspondence & Lead contact: 22 Kenji Inaba, Institute of Multidisciplinary Research for Advanced Materials, Tohoku 23 University, Katahira 2-1-1, Aoba-ku, Sendai, Miyagi 980-8577, Japan 24 E-mail: kenji.inaba.a1@tohoku.ac.jp 25 Tel: +81-22-217-5604 26 Fax: +81-22-217-5605 27 ORCID: 0000-0001-8229-0467 28 Running title: Nascent-chain disulfide bond formation 29 30 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 2 Abstract 31 The mammalian endoplasmic reticulum (ER) harbors more than 20 members of 32 the protein disulfide isomerase (PDI) family that act to maintain proteostasis. 33 Herein, we developed an in vitro system for directly monitoring PDI- or 34 ERp46-catalyzed disulfide bond formation in ribosome-associated nascent chains 35 (RNC) of human serum albumin. The results indicated that ERp46 more efficiently 36 introduced disulfide bonds into nascent chains with short segments exposed outside 37 the ribosome exit site than PDI. Single-molecule analysis by high-speed atomic 38 force microscopy further revealed that PDI binds nascent chains persistently, 39 forming a stable face-to-face homodimer, whereas ERp46 binds for a shorter time 40 in monomeric form, indicating their different mechanisms for substrate 41 recognition and disulfide bond introduction. Similarly to ERp46, a PDI mutant 42 with an occluded substrate-binding pocket displayed shorter-time RNC binding 43 and higher efficiency in disulfide introduction than wild-type PDI. Altogether, 44 ERp46 serves as a more potent disulfide introducer especially during the early 45 stages of translation, whereas PDI can catalyze disulfide formation in RNC when 46 longer nascent chains emerge out from ribosome. 47 48 Keywords 49 nascent chain, protein disulfide isomerase, ERp46, disulfide bond, co-translational 50 folding, high-speed atomic force microscopy, ER proteostasis 51 52 53 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 3 Introduction 54 Over billions of years of evolution, living organisms have developed ingenious 55 mechanisms to promote protein folding (Hartl et al, 2011). The oxidative network 56 catalyzing protein disulfide bond formation in the endoplasmic reticulum (ER) is a 57 prime example. While canonical protein disulfide isomerase (PDI) and ER 58 oxidoreductin-1 (Ero1) were previously postulated to constitute a primary disulfide 59 bond formation pathway (Araki & Inaba, 2012; Mezghrani et al, 2001; Tavender & 60 Bulleid, 2010), more than 20 different PDI family enzymes and multiple PDI oxidases 61 besides Ero1 have recently been identified in the mammalian ER, suggesting the 62 development of highly diverse oxidative networks in higher eukaryotes (Nguyen et al, 63 2011; Schulman et al, 2010; Tavender et al, 2010). Each PDI family enzyme is likely to 64 play a distinct role in catalyzing the oxidative folding of different substrates, 65 concomitant with some functional redundancy, leading to the efficient production of a 66 wide variety of secretory proteins with multiple disulfide bonds (Bulleid & Ellgaard, 67 2011; Okumura et al, 2015; Sato & Inaba, 2012). 68 Our previous in vitro studies using model substrates such as reduced and 69 denatured bovine pancreatic trypsin inhibitor (BPTI) and ribonuclease A (RNase A) 70 demonstrated that different PDI family enzymes participate in different stages of 71 oxidative protein folding, resulting in the accelerated folding of native enzymes (Kojima 72 et al, 2014; Sato et al, 2013). Multiple PDI family enzymes cooperate to synergistically 73 increase the speed and fidelity of disulfide bond formation in substrate proteins. 74 However, whether mechanistic insights gained by in vitro experiments using full-length 75 substrates are applicable to real events of oxidative folding in the ER remains an 76 important question. Indeed, some previous works demonstrated that newly synthesized 77 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 4 polypeptide chains undergo disulfide bond formation and isomerization 78 co-translationally, presumably via catalysis by specific PDI family members (Kadokura 79 et al, 2020; Molinari & Helenius, 1999; Robinson & Bulleid, 2020; Robinson et al, 80 2020; Robinson et al, 2017). Furthermore, nascent chains play important roles in their 81 own quality control by modulating the translation speed to increase the yield of native 82 folding; if a nascent chain fails to fold or complete translation, then the resultant 83 aberrant ribosome-nascent chain complexes are degraded or destabilized (Buhr et al, 84 2016; Chadani et al, 2017; Matsuo et al, 2017). These observations suggest that 85 understanding real events of oxidative protein folding in cells requires systematic 86 analysis of how PDI family enzymes act on nascent polypeptide chains during synthesis 87 by ribosomes. 88 To this end, we herein developed an experimental system for directly 89 monitoring disulfide bond formation in ribosome-associated human serum albumin 90 (HSA) nascent chains of different lengths from the N-terminus. The resultant 91 ribosome-nascent chain complexes (RNCs) were reacted with two ubiquitously 92 expressed PDI family members, ER-resident protein 46 (ERp46) and canonical PDI. 93 These two enzymes were previously shown to have distinct roles in catalyzing oxidative 94 protein folding: ERp46 engages in rapid but promiscuous disulfide bond introduction 95 during the early stages of folding, while PDI serves as an effective proofreader of 96 non-native disulfides during the later stages (Kojima et al., 2014; Sato et al., 2013). The 97 subsequent maleimidyl polyethylene glycol (mal-PEG) modification of free cysteines 98 and Bis-Tris (pH7.0) PAGE analysis enabled us to detect the oxidation status of the 99 HSA nascent chains conjugated with transfer RNA (tRNA). Using high-speed atomic 100 force microscopy (HS-AFM), we further visualized PDI and ERp46 acting on the RNCs 101 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 5 at the single-molecule level. Collectively, the results indicated that although both ERp46 102 and PDI could introduce a disulfide bond into the ribosome-associated HSA nascent 103 chains, they demanded different lengths of the HSA segment exposed outside the 104 ribosome exit site, and displayed different mechanisms of action against the RNC. The 105 present systematic in vitro study using RNC containing different lengths of HSA 106 nascent chains mimics co-translational disulfide bond formation in the ER, and the 107 results provide a framework for understanding the mechanistic basis of oxidative 108 nascent-chain folding catalyzed by PDI family enzymes. 109 110 Results 111 The efficiency of disulfide bond introduction into HSA nascent chains by 112 PDI/ERp46 113 To investigate whether PDI family enzymes can introduce disulfide bonds into a 114 substrate during translation, we first prepared RNCs in vitro. For this purpose, we made 115 use of a cell-free protein translation system reconstituted with eukaryotic elongation 116 factors 1 and 2, eukaryotic release factors 1 and 3 (eRF1 and eRF3), aminoacyl-tRNA 117 synthetases, tRNAs, and ribosome subunits, developed previously by Imataka and 118 colleagues (Machida et al, 2014). HSA was chosen as a model substrate for the 119 following reasons. Firstly, the three-dimensional structure of HSA has been solved at 120 high resolution (Sugio et al, 1999), providing information on the exact location of 17 121 disulfide bonds in its native structure. Secondly, native-state HSA contains an unpaired 122 cysteine, Cys34, near the N-terminal region, which has potential to form a non-native 123 disulfide bond with one of the subsequent cysteines, serving as a good indicator of 124 whether a non-native disulfide is introduced by ERp46 or PDI during the early stage of 125 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 6 translation. Thirdly, overall conformation and kinetics of disulfide bond regeneration 126 were characterized for reduced full-length HSA (Lee & Hirose, 1992), which is 127 beneficial for discussing similarities and differences in post- and co-translational 128 oxidative folding. Forth, no N-glycosylation sites are contained in the first 95 amino 129 acids of HSA, implying that HSA nascent chains synthesized by the cell-free system are 130 equivalent to those synthesized in the ER in regard to N-glycosylation. Finally, the 131 involvement of PDI family enzymes in intracellular HSA folding has been demonstrated 132 (Koritzinsky et al, 2013; Rutkevich et al, 2010; Rutkevich & Williams, 2012), ensuring 133 the physiological relevance of the present study. 134 To stall the translation of HSA at specified sites, a uORF2 arrest sequence 135 (Alderete et al, 1999) was inserted into appropriate sites of the expression plasmid (Fig 136 1A). We first prepared two versions of the RNC containing different lengths of HSA 137 nascent chains: RNC 69-aa and RNC 82-aa. Since the ribosome exit tunnel 138 accommodates a polypeptide chain of ~30 amino acid (aa) residues (Zhang et al, 2013), 139 the N-terminal 57 residues of HSA (excluding the N-terminal 6-aa pro-sequence) are 140 predicted to be exposed outside the ribosome exit tunnel in RNC 69-aa, including 141 Cys34 and Cys53 (Fig 1A). In the RNC 82-aa construct, the N-terminal 70 residues of 142 HSA, including Cys62 as well as Cys34/Cys53, are predicted to emerge from the 143 ribosome (Fig 1A). Notably, Cys53 and Cys62 form a native disulfide bond, whereas 144 Cys34 is unpaired in the native structure of HSA domain I. 145 When RNC 69-aa was employed as a substrate, neither PDI nor ERp46 could 146 efficiently introduce a disulfide bond into the nascent chain (Fig 1C and 1D). However, 147 both enzymes introduced a disulfide bond into RNC 82-aa with higher efficiency than 148 into RNC 69-aa (Fig 1E and 1F), suggesting that the length of the exposed HSA 149 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 7 segment or the distance of a pair of cysteines from the ribosome exit site is critical for 150 disulfide bond introduction by PDI and ERp46. For either construct, a faint band was 151 seen between the bands of ‘no SS’ and ‘1 SS’, and this band was even fainter without 152 GSH/GSSG (the second lane from the left) and had a tendency to get stronger at late 153 time points. Presumably, this band represents a species in which one of free cysteines is 154 glutathionylated, and the species increased gradually in the course of the reaction. 155 Of note, ERp46 introduced a disulfide bond into RNC 82-aa at a much higher 156 rate than PDI, indicating that ERp46 serves as a more competent disulfide bond 157 introducer to RNCs than PDI (Fig 1F). The remarkable difference in disulfide bond 158 introduction efficiency by these two enzymes seems unlikely to be explained simply by 159 the different number of redox-active Trx-like domains in PDI (two) and ERp46 (three) 160 (Fig 1B). Also, the redox states in the presence of 1 mM GSH and 0.2 mM GSSG are 161 similar between these two enzymes (Fig EV1A and EV1B), suggesting their comparable 162 redox potentials. Thus, the different ability of ERp46 and PDI to introduce a disulfide 163 into 82-aa is likely caused by other factors such as different structural features and 164 different mechanism of substrate recognition, as discussed below. 165 Next, to identify which cysteine pair forms a disulfide bond in RNC 82-aa, we 166 constructed three cysteine mutants in which either Cys34, Cys53, or Cys62 was mutated 167 to alanine (Fig 2A). The assays using the mutants showed that whereas PDI was unable 168 to introduce a disulfide bond into RNC 82-aa C34A and C53A (Fig 2B, top and middle), 169 the enzyme introduced a Cys34-Cys53 non-native disulfide bond into RNC 82-aa C62A 170 (Fig 2B, bottom), at almost the same rate as the generation of the ‘1 SS’ species in 82-aa 171 (Fig 1E and 1F). PDI could not introduce a Cys53-Cys62 native disulfide bond, 172 presumably because this cysteine pair is located too close to the ribosome exit site (see 173 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 8 also Fig 3B and 3C). Conversely, the slow but possible formation of a Cys34-Cys53 174 non-native disulfide in 82 aa by PDI suggests that the distance between a cysteine pair 175 of interest and the ribosome exit site is key to allowing the enzyme to catalyze disulfide 176 bond introduction into RNCs. Considering the different locations of the Cys34-Cys53 177 and Cys53-Cys62 pairs on RNC 82-aa, a distance of ~18 residues from the ribosome 178 exit site appears to be necessary for the PDI-catalyzed reaction (see also the 179 Discussion). 180 In contrast to PDI, ERp46 could introduce a native disulfide bond into RNC 181 82-aa C34A (Fig 2C, top). Like PDI, ERp46 also introduced a non-native disulfide bond 182 between Cys34 and Cys53 into RNC 82-aa C62A, but its efficiency was lower than that 183 of a Cys53-Cys62 native disulfide (Fig 2C, bottom). No disulfide bond was formed 184 between Cys34 and Cys62 by either ERp46 or PDI (Fig 2C, middle), presumably due to 185 the considerable spatial separation of these two cysteines. Based on these results, we 186 concluded that for efficient disulfide bond introduction into RNCs, ERp46 requires an 187 intermediary polypeptide segment with a shorter distance between a cysteine pair of 188 interest and the ribosome exit site than PDI. We here note that ERp46-catalyzed 189 generation of the ‘1 SS’ species was faster in 82-aa than in 82-aa C34A (Fig 1F and 2C). 190 This observation may suggest the occurrence of Cys34-mediated disulfide bond 191 formation in 82-aa, namely, the formation of a Cys34-Cys53 non-native disulfide and, 192 possibly, its rapid isomerization to a Cys53-Cys62 native disulfide. 193 194 Accessibility of PDI/ERp46 to cysteines on the ribosome-HSA nascent chain 195 complex 196 To examine the accessibility of PDI and ERp46 to Cys residues on RNC 82-aa, we 197 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 9 constructed three RNC 82-aa mono-Cys mutants in which either Cys34, Cys53, or 198 Cys62 on the HSA nascent chain was retained, and investigated whether a mixed 199 disulfide could be formed between the RNC 82-aa mutant and a trapping mutant of PDI 200 or ERp46 in which all CXXC redox-active sites were mutated to CXXA. Both PDI and 201 ERp46 formed a mixed disulfide bond with Cys34 and Cys53 on RNC 82-aa with high 202 probability, but covalent linkages to Cys62 were marginal (Fig 2D and 2E). The results 203 suggest that the redox-active sites of PDI and ERp46 could gain access to Cys34 and 204 Cys53, but to a much lesser extent, to Cys62, probably due to steric collision with the 205 ribosome. Nevertheless, ERp46 efficiently introduced a native disulfide bond between 206 Cys53 and Cys62 (Fig 2C, top), presumably because ERp46 first attacked Cys53 on the 207 HSA nascent chain, and the resultant mixed disulfide was subjected to nucleophilic 208 attack by Cys62 (Fig 2F, right). By contrast, the mixed disulfide between PDI and 209 Cys53 on the HSA nascent chain seems unlikely to be attacked by Cys62, probably due 210 to steric collision between PDI and the ribosome (Fig 2F, left). In line with this idea, 211 PDI adopts a U-like overall conformation with restricted movements of four thioredoxin 212 (Trx)-like domains (Tian et al, 2006; Wang et al, 2012), whereas ERp46 forms a highly 213 flexible V-shape conformation composed of three Trx-like domains and two long (~20 214 aa) interdomain linkers (Kojima et al., 2014). 215 216 Correlations between cysteine accessibility and the efficiency of disulfide bond 217 introduction by PDI/ERp46 218 Based on the results presented above, we believe that the distance between cysteines of 219 interest and the ribosome exit site is critical for efficient disulfide introduction by PDI 220 and ERp46. To test this hypothesis, we increased the distance of the Cys53-Cys62 pair 221 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 10 from the ribosome exit site by inserting an extended polypeptide segment composed of 222 [SG]5 or [SG]10 repeat immediately after Cys62 on RNC 82-aa C34A (Fig 3A), and 223 investigated the effects of the insertions on the efficiency of disulfide bond formation. 224 While PDI was unable to introduce a Cys53-Cys62 native disulfide into RNC 82-aa 225 C34A (Fig 2B, top), insertion of a [SG]5 repeat allowed this reaction, and nearly 70% of 226 82-aa C34A was disulfide-bonded within a reaction time of 360 s (Fig 3B, upper and 227 3C). The insertion of a longer repeat [SG]10 further promoted disulfide bond formation 228 (Fig 3B, lower and 3C). 229 A similar enhancement following [SG] repeat insertion was observed for 230 ERp46-catalyzed reactions. However, ERp46 exhibited a striking difference from PDI: 231 insertion of a [SG]5 repeat was long enough to introduce a Cys53-Cys62 native disulfide 232 into RNC 82-aa C34A within 15 s, and insertion of a [SG]10 repeat gave only a small 233 additional enhancement (Fig 3D and 3E). Thus, the presence of a disordered or 234 extended segment of ~18 aa (Asp63Phe70 + [SG]5 repeat) between a cysteine pair of 235 interest and the ribosome exit site was necessary and sufficient for ERp46 to generate a 236 Cys53-Cys62 disulfide rapidly, whereas PDI required a longer segment of ~28 aa 237 (Asp63Phe70 + [SG]10 repeat) in this intermediary region for efficient introduction of 238 a Cys53-Cys62 disulfide. Thus, ERp46 seems to be more capable of introducing a 239 disulfide bond near the ribosome exit site than PDI. In other words, ERp46 likely has 240 the higher potential to introduce a disulfide bond into the HSA nascent chain during the 241 earlier stages of translation than PDI. 242 To verify that Cys53-Cys62 disulfide formation facilitated by [SG]10 repeat 243 insertion was ascribed to higher accessibility of PDI/ERp46 to Cys62, we again 244 investigated mixed disulfide bond formation between trapping mutants of PDI/ERp46 245 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 11 and each cysteine on RNC 82-aa following [SG]10 repeat insertion. Both PDI and 246 ERp46 formed a mixed disulfide with all cysteines including Cys62 (Fig 3F and 3G), 247 indicating that there is a correlation between the accessibility of PDI/ERp46 to a target 248 pair of cysteines and the efficiency of disulfide bond introduction by the enzymes. 249 250 Disulfide bond introduction into a longer HSA nascent chain by PDI/ERp46 251 In addition to the [SG]-repeat insertion, we examined the effect of natural HSA 252 sequence extension on PDI- or ERp46-mediated disulfide formation. For this purpose, 253 we prepared RNC 95-aa in which the N-terminal 83 amino acids of HSA (excluding the 254 N-terminal 6-aa pro-sequence), including Cys34, Cys53, Cys62, and Cys75, are 255 predicted to emerge from ribosome (Fig 4A). With this construct, however, we had a 256 technical problem with detection of the reduced species, because mal-PEG modification 257 of four cysteines greatly diminished the gel-to-membrane transfer efficiency. We 258 overcame this problem by using photo-cleavable mal-PEG (PEG-PCMal) and 259 irradiating UV light to the SDS gel after the gel electrophoresis and before the 260 membrane transfer. 261 Consequently, we observed both PDI and ERp46 introduced a disulfide bond 262 into 95-aa (Fig 4B), but the efficiency was slower than that into 82-aa (Fig 1E and 1F), 263 although a longer polypeptide chain is exposed outside the ribosome exit site in RNC 264 95-aa. Thus, the effect of natural sequence extension was opposite to that of [SG]-repeat 265 insertion. Formation of some higher-order structure or exposure of another cysteine may 266 somehow prevent PDI and ERp46 from introducing a disulfide bond into RNC 95-aa. 267 Thus, a longer polypeptide chain exposed outside ribosome does not always lead to a 268 higher disulfide formation rate. Rather, it is suggested that PDI and ERp46 can 269 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 12 introduce a disulfide bond into a nascent chain with higher efficiency when the 270 necessary and minimum length emerges out. 271 Given that four cysteines are exposed outside the ribosome in RNC 95-aa, we 272 next investigated whether PDI and ERp46 can catalyze nascent-chain disulfide 273 formation additionally or synergistically. The mixture of PDI and ERp46 generated a ‘1 274 SS’ species, but not a ‘2 SS’ species, like PDI or ERp46 alone (Fig 4B and 4C). Notably, 275 the presence of PDI inhibited ERp46-mediated disulfide formation, possibly due to its 276 competition with ERp46 for binding to RNC 95-aa. Thus, neither additional nor 277 synergistic effect was observed (Fig 4B and 4C). In this regard, our previous 278 observation for the synergistic cooperation of PDI and ERp46 in RNase A oxidative 279 folding (Sato et al., 2013) was not true for the ribosome-associated HSA nascent chain. 280 281 Single-molecule analysis of ERp46 by high-speed atomic force microscopy 282 To explore the mechanisms by which PDI and ERp46 recognize and act on RNCs at the 283 molecular level, we employed HS-AFM (Kodera et al, 2010; Noi et al, 2013; Okumura 284 et al, 2019; Uchihashi et al, 2018). While our previous HS-AFM analysis revealed that 285 PDI molecules form homodimers in the presence of unfolded substrates (Okumura et al., 286 2019), the structure and dynamics of ERp46 have not been analyzed using this 287 experimental approach. Therefore, we first observed ERp46 molecules alone by 288 immobilizing the N-terminal His-tag on a Co 2+ -coated mica surface. AFM images 289 revealed various overall shapes of ERp46 (Fig 5A), and some particle images clearly 290 demonstrated the presence of three thioredoxin (Trx)-like domains in ERp46 (Fig 5A, 291 left). To assess the overall structures of ERp46, we calculated the circularity of each 292 molecule and performed statistical analysis (Uchihashi et al., 2018). Circularity is a 293 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 13 measure of how circular the outline of an observed molecule is, defined by the equation 294 4S/L 2 , where L and S are the contour length of the outline and the area surrounded by 295 the outline, respectively. Thus, a circularity of 1.0 indicates a perfect circle, and values 296 <1 indicate a more extended conformation. 297 Statistical analysis based on circularity classified randomly chosen ERp46 298 particles into two major groups: opened V-shape and round/compact O-shape (Fig 5A). 299 Histograms with Gaussian fitting curves indicated that ~80% of ERp46 molecules 300 adopted V-shape conformations while ~20% adopted O-shape conformations (Fig 5B). 301 There was no large difference in height between these two conformations, suggesting 302 that the three Trx-like domains of ERp46 are arranged within the same plane in either 303 conformation. Successive AFM images acquired every 100 ms revealed that ERp46 304 adopted an open V-shape conformation during nearly 75% of the observation time, 305 while the protein also adopted an O-shape conformation occasionally (Fig 5C, 5D, 5E 306 and Movie EV1). The histogram calculated from the time-course snapshots was similar 307 to that calculated from images of 200 molecules at a certain timepoint (Fig 5B and 5E). 308 Importantly, structural insights gained by HS-AFM analysis are in good agreement with 309 those from small-angle X-ray scattering (SAXS) analysis: both analyses consistently 310 indicate the coexistence of a major population of molecules with an open V-shape and a 311 minor population with a compact O-shape (Kojima et al., 2014). 312 313 Single-molecule analysis of PDI/ERp46 acting on 82-aa RNC by HS-AFM 314 PDI and ERp46 are predicted to bind RNCs transiently during disulfide bond 315 introduction, but transient interactions would make it harder to observe and analyze the 316 mode of PDI/ERp46 binding to RNCs. More practically, at least 5 mins are required to 317 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 14 prepare for starting HS-AFM measurements after adding PDI or ERp46 to RNCs 318 immobilized onto a mica surface. If we employed RNCs containing natural HSA 319 sequences, PDI or ERp46 would complete nascent-chain disulfide formation during this 320 setup time. We therefore constructed HSA 82-aa RNC with Cys34, Cys53, and Cys62 321 mutated to Ala (hereafter referred to as 82-aa CA RNC), with the intension of trapping 322 RNC molecules bound to PDI/ERp46. After testing several RNC immobilization 323 methods, we chose to immobilize RNC on a Ni 2+ -coated mica surface. As a result, most 324 RNC molecules were observed to lie sideways on the mica surface, while nascent chains 325 were difficult to visualize, probably due to their flexible and extended structural nature 326 (Fig 6A). 327 When oxidized PDI or ERp46 were added to onto the RNC-immobilized mica 328 surface, PDI/ERp46-like particles were observed in the peripheral region of ribosomes. 329 When no-chain RNC (NC-RNC), comprising only the N-terminal FLAG tag and the 330 subsequent uORF2 but no segment from HSA, was immobilized on the mica surface, 331 far fewer particles were observed near RNCs (within 25 Å from the outline of 332 ribosomes) by HS-AFM despite the presence of PDI/ERp46 (Fig EV2A and EV2B). 333 These results confirm that we successfully observed PDI/ERp46 molecules acting on 334 HSA nascent chains associated with ribosomes. 335 Notably, the HS-AFM analysis revealed that PDI bound RNCs in both 336 monomeric and dimeric forms at an approximate ratio of 7:3 (Fig 6B), as reported 337 previously for reduced and denatured BPTI and RNase A as substrates (Okumura et al., 338 2019). Thus, PDI likely recognizes HSA nascent chains in a similar manner to 339 full-length substrates. Statistical analysis of RNC binding rates revealed that whereas 340 most monomeric PDI molecules (52/55 molecules) bound RNC for 10 s or shorter (Fig 341 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 15 6D, Fig EV3A and Movie EV2), most homodimeric PDI molecules (17/19 molecules) 342 bound RNC for 60 s or longer (Fig 6D, Fig EV3B and Movie EV3). By contrast, ERp46 343 molecules in the periphery of RNCs were only present in monomeric form (Fig 6C). 344 Importantly, nearly 20% (12/59 molecules) of ERp46 molecules bound RNC for 10 to 345 20 s (Fig 6D, Fig EV3C and Movie EV4), while a smaller portion (8/59 molecules) 346 bound RNC for ~60 s (Fig 6D). It is also notable that significant portion of PDI and 347 ERp46 molecules bound ribosomes for <5 s. This may indicate that PDI/ERp46 binds or 348 approaches RNCs only transiently possibly via diffusion, without tight interactions. 349 The histogram of the distance between the edge of ribosomes and the center of 350 ribosome-neighboring PDI/ERp46 molecules indicated that both PDI and ERp46 bound 351 RNCs at positions ~16 nm distant from ribosomes with a single-Gaussian distribution 352 with a half width of ~11 nm (Fig 6E), suggesting that both enzymes recognize similar 353 sites of the HSA nascent chain. Given that the distance between adjacent amino acids is 354 approximately 3.5 Å along an extended strand, Cys34, Cys53, and Cys62 are calculated 355 to be 130 Å, 63 Å, and 35 Å distant from the ribosome exit site, respectively. The 356 distributions of PDI and ERp46 molecules bound to RNC 82-aa seem consistent with 357 their accessibility to Cys34 and Cys53, but not to Cys62, as revealed by their mixed 358 disulfide formation with RNC 82-aa (Fig 2D and E). 359 360 Role of the PDI hydrophobic pocket in oxidation of the HSA nascent chain 361 It is widely known that the PDI b’ domain contains a hydrophobic pocket that acts as a 362 primary substrate-binding site (Klappa et al, 1998). To examine the involvement of the 363 hydrophobic pocket in PDI-catalyzed disulfide bond formation in the HSA nascent 364 chain, we mutated I289, one of the central residues that constitute the hydrophobic 365 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 16 pocket, to Ala, and compared the efficiency of disulfide bond introduction into RNC 366 82-aa between wild-type (WT) and mutant I289A proteins. In this mutant, the x-linker 367 flanked by b’ and a’ domains tightly binds the hydrophobic pocket, unlike in WT, 368 thereby preventing PDI from tightly binding an unfolded substrate (Bekendam et al, 369 2016; Nguyen et al, 2008). ERp57, another primary member of the PDI family, has a 370 U-shape domain arrangement similar to PDI, but does not contain the hydrophobic 371 pocket in the b’ domain. For comparison, we also monitored ERp57-catalyzed disulfide 372 introduction into RNC 82-aa. 373 Despite the occlusion or lack of the hydrophobic substrate-binding pocket, both 374 PDI I289A and ERp57 were found to introduce a disulfide bond into RNC 82-aa at a 375 higher rate than PDI WT (Fig 7A and B). This result suggests that the hydrophobic 376 pocket is involved in binding the HSA nascent chain, but this binding appears to rather 377 slow down disulfide introduction into a nascent chain. 378 To further explore the mechanism by which PDI I289A introduced a disulfide 379 bond at a faster rate than PDI WT, we analyzed its binding to RNC using HS-AFM. The 380 analysis revealed that, while nearly one-third of PDI I289A molecules formed dimers in 381 the presence of RNC 82-aa like PDI WT, the mutant dimers bound RNC for a shorter 382 time than the WT dimers (Fig 7C and Movie EV6). Thus, the RNC-binding time of PDI 383 I289A showed similar distribution to that of ERp46 (Fig 7D and Movies EV5 and EV6), 384 which seems consistent with the higher disulfide introduction efficiency of PDI I289A 385 than that of PDI WT. PDI I289A also bound RNCs at positions ~16 nm distant from 386 ribosome with a single-Gaussian distribution (Fig 7E), suggesting that PDI I289A 387 recognizes similar sites of the HSA nascent chain as PDI and ERp46. 388 389 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 17 Discussion 390 A number of studies have recently investigated co-translational oxidative 391 folding in the ER (Kadokura et al., 2020; Robinson et al., 2020; Robinson et al., 2017). 392 The present study showed that while both PDI and ERp46 can introduce a disulfide 393 bond into a nascent chain co-translationally, ERp46 catalyzes this reaction more 394 efficiently than PDI and requires a shorter nascent chain segment exposed outside the 395 ribosome exit. Thus, ERp46 appears to be capable of introducing a disulfide bond into a 396 nascent chain during the earlier stages of translation than PDI. The efficient introduction 397 of a Cys53-Cys62 native disulfide on RNC 82-aa by ERp46 (Fig 2) suggests that a 398 separation of ~8 aa residues between a C-terminal cysteine on a nascent chain and the 399 ribosome exit site (i.e., residues 63-70) is sufficient for ERp46 to catalyze this reaction 400 (Fig 8). When a nascent chain was elongated by the insertion of [SG]-repeat sequences, 401 PDI could also introduce the native disulfide bond into RNCs to some extent (Fig 3B 402 and 3C). Thus, PDI appears to act on a nascent chain to introduce a disulfide bond when 403 the distance between a C-terminal cysteine on a nascent chain and the ribosome exit site 404 reaches ~18 aa residues (i.e., residues 63-70 + [SG]5 repeat; Fig 8). 405 Disulfide bond formation in partially ER-exposed nascent chains was indeed 406 observed with the ADAM10 disintegrin domain, which has a dense disulfide bonding 407 pattern and little defined structure (Robinson et al., 2020). Thus, disulfide bond 408 formation seems to be allowed before the higher order structure is defined in a nascent 409 chain. This could be the case with a Cys34-Cys53 nonnative disulfide and a 410 Cys53-Cys62 native disulfide on RNC 82-aa, since the N-terminal 82-residue HSA 411 fragment alone is unlikely to fold to a globular native-like structure though the fragment 412 of residue 35 to 56 is predicted to form an -helix according to the HSA native structure. 413 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 18 In contrast, some proteins including 2-microglobulin (2M) and prolactin are shown to 414 form disulfide bonds only after a folding domain is fully exposed to the ER or a 415 polypeptide chain is released from ribosome, suggesting their folding-driven disulfide 416 bond formation. Notably, PDI binds 2M when the N-terminal ~80 residues of 2M are 417 exposed to the ER, and completes disulfide bond introduction at the even later stages of 418 translation (Robinson et al., 2017). Thus, PDI has been demonstrated to engage in 419 disulfide bond formation during late stages of translation or after translation in the ER. 420 Regarding mechanistic insight, the present HS-AFM analysis visualized PDI 421 and ERp46 acting on nascent chains at the single-molecule level. We found that PDI 422 forms a face-to-face homodimer that binds a nascent chain, as is the case with reduced 423 and denatured full-length substrates (Okumura et al., 2019). On the other hand, ERp46 424 maintains a monomeric form while binding a nascent chain. Interestingly, the PDI dimer 425 binds a nascent chain much more persistently than the PDI monomer and ERp46, 426 suggesting that the PDI dimer holds a nascent chain tightly inside its central 427 hydrophobic cavity. In agreement with this observation, a hydrophobic-pocket mutant 428 (I289A) of PDI bound a nascent chain for shorter time and introduced a disulfide bond 429 into a nascent chain more rapidly than the WT enzyme, as was the case with ERp46. In 430 this context, PDI competed with ERp46 for acting on RNC 95-aa, and thereby inhibited 431 ERp46-mediated disulfide introduction (Fig 4 and Fig 8). Thus, PDI family enzymes do 432 not always work synergistically to accelerate oxidative protein folding, but may 433 possibly inhibit each other during co-translational disulfide bond formation. 434 How the ER membrane translocon channel is involved in co-translational 435 oxidative folding catalyzed by PDI family enzymes remains an important question. It is 436 possible that PDI and ERp46 form a supramolecular complex with ribosomes and the 437 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 19 Sec61 translocon channel via a nascent chain. Indeed, PDI was previously identified as 438 a luminal protein that was in close contact with translocating nascent chains (Klappa et 439 al, 1995). Additionally, the oligosaccharyltransferase complex (Harada et al, 2009) and 440 an ER chaperone calnexin (Farmery et al, 2000) have been reported to interact with the 441 ribosome-associated Sec61 channel to catalyze N-glycosylation and folding of nascent 442 chains in the ER, respectively. In this regard, it will be interesting to examine the close 443 co-localization of PDI/ERp46 with the Sec61 channel in the presence or absence of 444 nascent chains in transit into the ER lumen by super-resolution microscopy or other 445 tools. Systematic studies with a wider range of substrates of different lengths from the 446 ribosome exit site and different numbers of cysteine pairs, and with other PDI family 447 members potentially having different functional roles, will provide further mechanistic 448 and physiological insights into co-translational oxidative folding and protein quality 449 control in the ER. 450 451 Materials & Methods 452 Construction of HSA plasmids 453 DNA fragments encoding specific regions (69-aa, N-terminal pro-sequence 6-aa + the 454 subsequent 63-aa; 82-aa, N-terminal pro-sequence 6-aa + the subsequent 76-aa; 95-aa, 455 N-terminal pro-sequence 6-aa + the subsequent 89-aa) of HSA were amplified by PCR 456 with appropriate primers and inserted into the pUC-T7-HCV-FLAG-2A-uORF 457 expression plasmid, as described in Machida et al. (2014). The amplified fragments 458 were replaced with the 2A region to generate pUC-T7-HCV-FLAG-HSA (69-aa or 459 82-aa)-uORF2. RNC 82-aa C34A/C53A/C62A and mono-Cys mutants were constructed 460 using the QuikChange method with appropriate primers (Table 1). RNC 82-aa C34A 461 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 20 with [SG]5 or [SG]10 repeats were constructed by the Prime STAR MAX (Takara Bio 462 Inc., Japan) method using appropriate primers (Table 1). 463 464 Expression and purification of PDI and ERp46 465 Overexpression and purification of human PDI and ERp46, and their mutants, were 466 performed as described previously (Kojima et al., 2014; Sato et al., 2013). An ERp46 467 trapping mutant with a CXXA sequence in all Trx-like domains was constructed by the 468 QuikChange method using appropriate sets of primers. 469 470 Preparation of RNCs using a translation system reconstituted with human factors 471 A cell-free translation system was reconstituted with eEF1 (50 M), eEF2 (1 M), 472 eRF1/3 (0.5 M), aminoacyl-tRNA synthetases (0.15 g/l), tRNAs (1 g/l), 40S 473 ribosomal subunit (0.5 M), 60S ribosomal subunit (0.5 M), PPA1 (0.0125 M), 474 amino acids mixture (0.1 mM) and T7 RNA polymerase (0.015 g/l) (Machida et al., 475 2014). We added 1.0 µL template plasmid (0.5 mg/mL) into 19 µL of this cell-free 476 system, and the mixture was incubated for at least 34.5 h at 32C. After HKMS buffer 477 (comprising 25 mM HEPES-KOH (pH 7.0), 150 mM KCl, 5 mM Mg(OAc)2, and 1.0 M 478 sucrose) was added, samples were ultra-centrifuged at 100,000 g overnight at 4 C to 479 recover the RNC as a pellet. After removing the supernatant, pellets were resuspended 480 in HKM buffer comprising 25 mM HEPES-KOH (pH 7.0), 150 mM KCl, and 5 mM 481 Mg(OAc)2. 482 483 Monitoring PDI- and ERp46-mediated disulfide bond introduction into RNCs 484 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 21 The RNC suspension prepared as described above was mixed with PDI or ERp46 (0.1 485 M each) and glutathione/oxidized glutathione (GSH/GSSG; 1.0 mM:0.2 mM; 486 NACALAI TESQUE, INC., Japan). Aliquots were collected after incubation at 30C for 487 the indicated times, and reactions were quenched with mal-PEG 5K (2 mM; NOF 488 CORPORATION, Japan) for RNC 69-aa and RNC 82-aa. After cysteine alkylation at 489 room temperature for 20 min, samples were separated by 12% Bis-Tris (pH7.0) PAGE 490 (Thermo Fisher Scientific K.K., Japan) in the presence of the reducing reagent 491 -mercaptoethanol -ME; 10% v/v; NACALAI TESQUE, INC., Japan). After 492 transferring onto a polyvinylidene fluoride (PVDF) membrane (Merck KGaA, 493 Darmstadt, Germany), bands on the membrane were visualized using Chemi-Lumi One 494 Ultra (NACALAI TESQUE, INC., Japan) and a ChemiDocTM Imaging System 495 (Bio-Rad Laboratories, Inc., CA, USA). Signal intensity was quantified using ImageLab 496 software (Bio-Rad Laboratories, Inc., CA, USA). 497 For RNC 95-aa, reactions were quenched with PEG-PCMal (Dojindo, Japan). 498 After cysteine alkylation at room temperature for 20 min, samples were separated by 499 10% Bis-Tris (pH7.0) PAGE (Thermo Fisher Scientific K.K., Japan) in the presence of 500 the reducing reagent -ME10% v/v;). After gel electrophoresis, the gel was subjected 501 to UV irradiation (302 nm, 8 W) for 30 min. The subsequent procedures were the same 502 as described above. 503 504 Monitoring intermolecular disulfide bond linkage between PDI/ERp46 and 505 ribosome-HSA nascent chain complexes 506 To detect the intermolecular disulfide bond linkage between PDI/ERp46 and the 507 ribosome-HSA nascent chain complex, we employed RNC 82-aa mono-Cys mutants 508 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 22 retaining one of Cys34, Cys53, or Cys62. The RNC suspension prepared as described 509 above was mixed with a PDI or ERp46 trapping mutant (1 M each) and diamide (100 510 µM). Aliquots were collected after incubation at 30C for 10 min, and reactions were 511 quenched with N-ethylmaleimide (2 mM; NACALAI TESQUE, INC., Japan). Samples 512 were analyzed by Nu-PAGE and western blotting as described above. 513 514 High-speed atomic force microscopy imaging 515 The structural dynamics of PDI and ERp46 were probed using a high-speed AFM 516 instrument developed by Toshio Ando’s group (Kanazawa University). Data acquisition 517 for ERp46 was performed as described previously (Okumura et al., 2019). Briefly, 518 His6-tagged ERp46 was immobilized on a Co 2+ -coated mica surface through the 519 N-terminal His-tag. To this end, a droplet (10 L) containing 1 nM ERp46 was loaded 520 onto the mica surface. After a 3 min incubation, the surface was washed with TRIS 521 buffer (50 mM TRIS-HCl pH7.4, 300 mM NaCl). Single-molecule imaging was 522 performed in tapping mode (spring constant, ~0.1 N/m; resonant frequency, 0.8–1 MHz; 523 quality factor in water, ~2) and analyzed using Kodec4.4.7.39 software developed by 524 Toshio Ando’s group (Kanazawa University). AFM observations were made in fixed 525 imaging areas (400 × 400 Å 2 ) at a scan rate of 0.1 s/frame. Each molecule was observed 526 separately on a single frame with the highest pixel setting (60 × 60 pixels). Cantilevers 527 (Olympus, Tokyo, Japan) were 6–7 m long, 2 m wide, and 90 nm thick. For AFM 528 imaging, the free oscillation amplitude was set to ~1 nm, and the set-point amplitude 529 was around 80% of the free oscillation amplitude. The estimated tapping force was <30 530 pN. A low-pass filter was used to remove noise from acquired images. The area of a 531 single ERp46 molecule in each frame was calculated using LabView 2013 (National 532 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 23 Instruments, Austin, TX, USA) with custom-made programs. 533 To observe the binding of PDI/ERp46 to RNCs by HS-AFM, RNCs were 534 immobilized on a Ni 2+ -coated mica surface via electrostatic interactions. To this end, a 535 droplet (10 L) containing RNCs was loaded onto the mica surface. After a 10 min 536 incubation, the surface was washed with HSA buffer comprising 25 mM HEPES-KOH 537 pH 7.0, 150 mM KCl, and 5 mM Mg(OAc)2. PDI/ERp46 lacking the N-terminal 538 His6-tag was added to the RNC-immobilized mica surface at a final concentration of 1 539 nM. Measurements were performed under the same conditions described above. 540 541 Acknowledgments 542 This work was supported by Grants-in-Aid for Scientific Research from MEXT to KI 543 (26116005 and 18H03978), the NAGASE Science Technology Foundation (K.I.) and 544 the MITSUBISHI Foundation (K.I.). This work was also supported by Grant-in-Aid for 545 JSPS Fellows (Grant Number 20J11932 to C.H.) and a Grant-in-Aid of Tohoku 546 University, Division for Interdisciplinary Advanced Research and Education (to C.H.). 547 548 Author contributions 549 C.H. and T.M. developed an experimental system for directly monitoring 550 co-translational disulfide bond formation. K.M. and H.I. developed and prepared 551 cell-free protein translation system reconstituted with human factors. C.H. prepared 552 various plasmids. C.H. and M.O. purified PDI and ERp46, and their mutants. C.H. and 553 K.N. performed HS-AFM measurements and analyses. C.H., K.N., M.O. and T.O. 554 discussed the results of HS-AFM. K.I. supervised the work. C.H. and K.N. prepared the 555 Figures. C.H. and K.I. wrote the manuscript. All of the authors discussed the results and 556 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 24 approved the manuscript. 557 558 Conflict of interests 559 We declare that there are no competing interests related to this work. 560 561 562 References 563 Alderete JP, Jarrahian S, Geballe AP (1999) Translational effects of mutations and 564 polymorphisms in a repressive upstream open reading frame of the human 565 cytomegalovirus UL4 gene. J Virol 73: 8330-8337 566 567 Araki K, Inaba K (2012) Structure, mechanism, and evolution of Ero1 family enzymes. 568 Antioxidants & redox signaling 16: 790-799 569 570 Bekendam RH, Bendapudi PK, Lin L, Nag PP, Pu J, Kennedy DR, Feldenzer A, Chiu J, 571 Cook KM, Furie B et al (2016) A substrate-driven allosteric switch that enhances PDI 572 catalytic activity. Nature communications 7: 12579 573 574 Buhr F, Jha S, Thommen M, Mittelstaet J, Kutz F, Schwalbe H, Rodnina MV, Komar 575 AA (2016) Synonymous Codons Direct Cotranslational Folding toward Different 576 Protein Conformations. Molecular cell 61: 341-351 577 578 Bulleid NJ, Ellgaard L (2011) Multiple ways to make disulfides. Trends in biochemical 579 sciences 36: 485-492 580 581 Chadani Y, Niwa T, Izumi T, Sugata N, Nagao A, Suzuki T, Chiba S, Ito K, Taguchi H 582 (2017) Intrinsic Ribosome Destabilization Underlies Translation and Provides an 583 Organism with a Strategy of Environmental Sensing. Molecular cell 68: 528-539.e525 584 585 Farmery MR, Allen S, Allen AJ, Bulleid NJ (2000) The role of ERp57 in disulfide bond 586 formation during the assembly of major histocompatibility complex class I in a 587 synchronized semipermeabilized cell translation system. The Journal of biological 588 chemistry 275: 14933-14938 589 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 25 590 Harada Y, Li H, Li H, Lennarz WJ (2009) Oligosaccharyltransferase directly binds to 591 ribosome at a location near the translocon-binding site. Proceedings of the National 592 Academy of Sciences of the United States of America 106: 6945-6949 593 594 Hartl FU, Bracher A, Hayer-Hartl M (2011) Molecular chaperones in protein folding 595 and proteostasis. Nature 475: 324-332 596 597 Kadokura H, Dazai Y, Fukuda Y, Hirai N, Nakamura O, Inaba K (2020) Observing the 598 nonvectorial yet cotranslational folding of a multidomain protein, LDL receptor, in the 599 ER of mammalian cells. Proceedings of the National Academy of Sciences of the United 600 States of America 117: 16401-16408 601 602 Klappa P, Freedman RB, Zimmermann R (1995) Protein disulphide isomerase and a 603 lumenal cyclophilin-type peptidyl prolyl cis-trans isomerase are in transient contact 604 with secretory proteins during late stages of translocation. Eur J Biochem 232: 755-764 605 606 Klappa P, Ruddock LW, Darby NJ, Freedman RB (1998) The b' domain provides the 607 principal peptide-binding site of protein disulfide isomerase but all domains contribute 608 to binding of misfolded proteins. The EMBO journal 17: 927-935 609 610 Kodera N, Yamamoto D, Ishikawa R, Ando T (2010) Video imaging of walking myosin 611 V by high-speed atomic force microscopy. Nature 468: 72-76 612 613 Kojima R, Okumura M, Masui S, Kanemura S, Inoue M, Saiki M, Yamaguchi H, 614 Hikima T, Suzuki M, Akiyama S et al (2014) Radically different thioredoxin domain 615 arrangement of ERp46, an efficient disulfide bond introducer of the mammalian PDI 616 family. Structure (London, England : 1993) 22: 431-443 617 618 Koritzinsky M, Levitin F, van den Beucken T, Rumantir RA, Harding NJ, Chu KC, 619 Boutros PC, Braakman I, Wouters BG (2013) Two phases of disulfide bond formation 620 have differing requirements for oxygen. The Journal of cell biology 203: 615-627 621 622 Lee JY, Hirose M (1992) Partially folded state of the disulfide-reduced form of human 623 serum albumin as an intermediate for reversible denaturation. The Journal of biological 624 chemistry 267: 14753-14758 625 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 26 626 Machida K, Mikami S, Masutani M, Mishima K, Kobayashi T, Imataka H (2014) A 627 translation system reconstituted with human factors proves that processing of 628 encephalomyocarditis virus proteins 2A and 2B occurs in the elongation phase of 629 translation without eukaryotic release factors. The Journal of biological chemistry 289: 630 31960-31971 631 632 Matsuo Y, Ikeuchi K, Saeki Y, Iwasaki S, Schmidt C, Udagawa T, Sato F, Tsuchiya H, 633 Becker T, Tanaka K et al (2017) Ubiquitination of stalled ribosome triggers 634 ribosome-associated quality control. Nature communications 8: 159 635 636 Mezghrani A, Fassio A, Benham A, Simmen T, Braakman I, Sitia R (2001) 637 Manipulation of oxidative protein folding and PDI redox state in mammalian cells. The 638 EMBO journal 20: 6288-6296 639 640 Molinari M, Helenius A (1999) Glycoproteins form mixed disulphides with 641 oxidoreductases during folding in living cells. Nature 402: 90-93 642 643 Nguyen VD, Saaranen MJ, Karala AR, Lappi AK, Wang L, Raykhel IB, Alanen HI, Salo 644 KE, Wang CC, Ruddock LW (2011) Two endoplasmic reticulum PDI peroxidases 645 increase the efficiency of the use of peroxide during disulfide bond formation. Journal 646 of molecular biology 406: 503-515 647 648 Nguyen VD, Wallis K, Howard MJ, Haapalainen AM, Salo KE, Saaranen MJ, Sidhu A, 649 Wierenga RK, Freedman RB, Ruddock LW et al (2008) Alternative conformations of 650 the x region of human protein disulphide-isomerase modulate exposure of the substrate 651 binding b' domain. Journal of molecular biology 383: 1144-1155 652 653 Noi K, Yamamoto D, Nishikori S, Arita-Morioka K, Kato T, Ando T, Ogura T (2013) 654 High-speed atomic force microscopic observation of ATP-dependent rotation of the 655 AAA+ chaperone p97. Structure (London, England : 1993) 21: 1992-2002 656 657 Okumura M, Kadokura H, Inaba K (2015) Structures and functions of protein disulfide 658 isomerase family members involved in proteostasis in the endoplasmic reticulum. Free 659 radical biology & medicine 83: 314-322 660 661 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 27 Okumura M, Noi K, Kanemura S, Kinoshita M, Saio T, Inoue Y, Hikima T, Akiyama S, 662 Ogura T, Inaba K (2019) Dynamic assembly of protein disulfide isomerase in catalysis 663 of oxidative folding. Nature chemical biology 15: 499-509 664 665 Robinson PJ, Bulleid NJ (2020) Mechanisms of Disulfide Bond Formation in Nascent 666 Polypeptides Entering the Secretory Pathway. Cells 9 667 668 Robinson PJ, Kanemura S, Cao X, Bulleid NJ (2020) Protein secondary structure 669 determines the temporal relationship between folding and disulfide formation. The 670 Journal of biological chemistry 295: 2438-2448 671 672 Robinson PJ, Pringle MA, Woolhead CA, Bulleid NJ (2017) Folding of a single domain 673 protein entering the endoplasmic reticulum precedes disulfide formation. The Journal of 674 biological chemistry 292: 6978-6986 675 676 Rutkevich LA, Cohen-Doyle MF, Brockmeier U, Williams DB (2010) Functional 677 relationship between protein disulfide isomerase family members during the oxidative 678 folding of human secretory proteins. Molecular biology of the cell 21: 3093-3105 679 680 Rutkevich LA, Williams DB (2012) Vitamin K epoxide reductase contributes to protein 681 disulfide formation and redox homeostasis within the endoplasmic reticulum. Molecular 682 biology of the cell 23: 2017-2027 683 684 Sato Y, Inaba K (2012) Disulfide bond formation network in the three biological 685 kingdoms, bacteria, fungi and mammals. The FEBS journal 279: 2262-2271 686 687 Sato Y, Kojima R, Okumura M, Hagiwara M, Masui S, Maegawa K, Saiki M, Horibe T, 688 Suzuki M, Inaba K (2013) Synergistic cooperation of PDI family members in 689 peroxiredoxin 4-driven oxidative protein folding. Scientific reports 3: 2456 690 691 Schulman S, Wang B, Li W, Rapoport TA (2010) Vitamin K epoxide reductase prefers 692 ER membrane-anchored thioredoxin-like redox partners. Proceedings of the National 693 Academy of Sciences of the United States of America 107: 15027-15032 694 695 Sugio S, Kashima A, Mochizuki S, Noda M, Kobayashi K (1999) Crystal structure of 696 human serum albumin at 2.5 A resolution. Protein engineering 12: 439-446 697 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 28 698 Tavender TJ, Bulleid NJ (2010) Molecular mechanisms regulating oxidative activity of 699 the Ero1 family in the endoplasmic reticulum. Antioxidants & redox signaling 13: 700 1177-1187 701 702 Tavender TJ, Springate JJ, Bulleid NJ (2010) Recycling of peroxiredoxin IV provides a 703 novel pathway for disulphide formation in the endoplasmic reticulum. The EMBO 704 journal 29: 4185-4197 705 706 Tian G, Xiang S, Noiva R, Lennarz WJ, Schindelin H (2006) The crystal structure of 707 yeast protein disulfide isomerase suggests cooperativity between its active sites. Cell 708 124: 61-73 709 710 Uchihashi T, Watanabe YH, Nakazaki Y, Yamasaki T, Watanabe H, Maruno T, Ishii K, 711 Uchiyama S, Song C, Murata K et al (2018) Dynamic structural states of ClpB involved 712 in its disaggregation function. Nature communications 9: 2147 713 714 Wang C, Yu J, Huo L, Wang L, Feng W, Wang CC (2012) Human protein-disulfide 715 isomerase is a redox-regulated chaperone activated by oxidation of domain a'. The 716 Journal of biological chemistry 287: 1139-1149 717 718 Zhang Y, Wölfle T, Rospert S (2013) Interaction of nascent chains with the ribosomal 719 tunnel proteins Rpl4, Rpl17, and Rpl39 of Saccharomyces cerevisiae. The Journal of 720 biological chemistry 288: 33697-33707 721 722 723 724 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 29 Figure 1 - Disulfide bond introduction into RNC 69-aa and 82-aa by PDI and 725 ERp46 726 A Schematic structure of plasmids constructed in this study. ‘uORF2’ is an arrest 727 sequence that serves to stall translation of the upstream protein and thereby prepare 728 stable ribosome-nascent chain complexes (RNCs). The bottom cartoon represents the 729 location of cysteines and disulfide bonds in HSA domain I. HSA domain I consists of 730 195 amino acids and contains five disulfide bonds and one free cysteine at residue 34. A 731 green box indicates the pro-sequence. Orange circles and red lines indicate cysteines 732 and native disulfide bonds, respectively. The region predicted to be buried in the 733 ribosome exit tunnel is shown by a cyan box. 734 B Domain organization of PDI and ERp46. Redox-active Trx-like domains with a 735 CGHC motif are indicated by cyan boxes, while redox-inactive ones in PDI are by 736 light-green boxes. Note that the PDI b’ domain contains a substrate-binding 737 hydrophobic pocket. 738 C, E Time course of PDI-, ERp46-, and glutathione (no enzyme)-catalyzed disulfide 739 bond introduction into RNC 69-aa (C) and 82-aa (E). ‘noSS’ and ‘1SS’ denote reduced 740 and single-disulfide-bonded species of HSA nascent chains, respectively. Note that faint 741 bands observed between “no SS” and “1SS” likely represent a species in which one of 742 cysteines is not subjected to mal-PEG modification due to glutathionylation. In support 743 of this, these minor bands are even fainter under the conditions of no GSH/GSSG. 744 D, F Quantification of disulfide-bonded species for RNC 69-aa (D) and 82-aa (F) based 745 on the results shown in (C) and (E), respectively (n = 3). 746 747 Figure 2 - Disulfide bond introduction into RNC 82-aa Cys mutants by PDI and 748 ERp46 749 A Cartoon of RNC constructs used in this study. In each construct, a cysteine 750 (represented by a black circle) was mutated to alanine. Note that RNC 82-aa C34A 751 retains a native cysteine pairing (i.e., Cys53 and Cys62), while RNC 82-aa C53A and 752 C62A retain a non-native pairing. 753 B and C Time course of PDI- and ERp46-catalyzed disulfide bond introduction into 754 RNC 82-aa C34A (top), C53A (middle), and C62A (bottom) mutants. Note that faint 755 bands observed between “no SS” and “1SS” likely represent a species in which one of 756 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 30 cysteines is not subjected to mal-PEG modification due to glutathionylation. 757 Quantification of disulfide-bonded species of RNC 82-aa Cys mutants is based on the 758 results shown for the upper raw data (n = 3). 759 D Formation of a mixed disulfide bond between RNC 82-aa mono-Cys mutants and PDI 760 (upper)/ERp46 (lower). ‘Mixed’ and ‘No SS’ denote a mixed disulfide complex between 761 PDI/ERp46 and RNC mono-Cys mutants and isolated RNC 82-aa, respectively. Note 762 that faint bands observed between ‘Mixed’ and ‘no SS’ are likely non-specific bands, as 763 they were seen at the same position regardless of which 82-aa mono-Cys mutant was 764 tested or whether an RNC was reacted with PDI or ERp46. 765 E Quantification of mixed disulfide species based on the results shown in (D). n = 3. 766 F The cartoon on the left shows possible steric collisions between ribosomes and PDI 767 when Cys62 attacks the mixed disulfide between Cys53 on RNC 82-aa and PDI (left). 768 The cartoon on the right shows that ERp46 can avoid this steric collision due to its 769 higher flexibility and domain arrangement. 770 771 Figure 3 - Correlation of the distance between Cys residues and the ribosome exit 772 site with the efficiency of disulfide bond introduction by PDI/ERp46 773 A Cartoons of RNC constructs with [SG]-repeat insertions. A [SG]5 or [SG]10 repeat 774 sequence was inserted into RNC-82 aa C34A immediately after Cys62. 775 B, D PDI- (B) and ERp46 (D)-mediated disulfide bond introduction into RNC 82-aa 776 C34A with insertion of [SG]5 (upper) or [SG]10 (lower) repeats after Cys62. 777 C, E Quantification of disulfide-bonded species (1SS) based on the results shown in (B) 778 and (D). n = 3 for PDI and 2 for ERp46. 779 F Formation of a mixed disulfide bond between the 82-aa mono-Cys mutant with a 780 [SG]10 repeat and PDI (upper)/ERp46 (lower). Note that bands observed between 781 ‘Mixed’ and ‘no SS’ are likely non-specific bands, as they were seen at the same 782 position regardless of which 82-aa mono-Cys [SG]10 mutant was tested or whether an 783 RNC was reacted with PDI or ERp46. 784 G Quantification of mixed disulfide species based on the results shown in (F). n = 3. 785 786 Figure 4 - Disulfide bond introduction into RNC 95-aa by PDI and ERp46 787 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 31 A Schematic structure of RNC-95-aa. Orange circles and red lines in the bottom cartoon 788 indicate cysteines and native disulfides, respectively. The region predicted to be buried 789 in the ribosome exit tunnel is shown by a cyan box. 790 B Time course of PDI (0.1 M)-, ERp46 (0.1 M)-, and their mixture (0.1 M 791 each)-catalyzed disulfide bond introduction into RNC 95-aa. ‘noSS’ and ‘1SS’ denote 792 reduced and single-disulfide-bonded species of the HSA nascent chain, respectively. 793 C Quantification of the single-disulfide-bonded (1 SS) species based on the result 794 shown in (B) (n = 3). 795 796 Figure 5 - High-speed AFM analysis of ERp46 797 A AFM images (scan area, 200  200 Å; scale bar, 30 Å) for ERp46 V-shape (left) and 798 O-shape (right) conformations. 799 B Left upper: Histograms of circularity calculated from AFM images of ERp46. Values 800 represent the average circularity (mean ± s.d.) calculated from curve fitting with a 801 single- (middle and right) or two- (left) Gaussian model. Left lower: Histograms of 802 height calculated from AFM images of ERp46. Values represent the average height 803 (mean ± s.d.) calculated from curve fitting with a single-Gaussian model. Right: 804 Two-dimensional scatterplots of the height versus circularity for ERp46 molecules 805 observed by HS-AFM. 806 C Time-course snapshots of oxidized ERp46 captured by HS-AFM. The images were 807 traced for 10 s. See also Movie EV1. 808 D Time trace of the circularity of an ERp46 molecule. 809 E Histogram of the circularity of ERp46 calculated from the time-course snapshots 810 shown in (D). 811 812 Figure 6 - Single-molecule observation of PDI/ERp46 acting on 82-aa CA RNC by 813 high-speed atomic force microscopy 814 A The AFM images (scan area, 500 Å  500 Å; scale bar, 100 Å) displaying 82-aa CA 815 RNC in the absence of PDI family enzymes on a Ni 2+ -coated mica surface. The surface 816 model on the right side of each AFM image illustrates ribosome whose view angle is 817 approximately adjusted to the observed RNC particle. 40S and 60S ribosomal subunits 818 are shown in red and blue, respectively. 819 B Upper AFM images (scan area, 500 Å  500 Å; scale bar, 100 Å) displaying 82-aa CA 820 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 32 RNC in the presence of oxidized PDI (1 nM). PDI molecules that appear to bind 82-aa 821 CA RNC are marked by red squares. Lower images (scan area, 250 Å  250 Å; scale bar, 822 50 Å) highlight the regions surrounded by red squares in the upper images. 823 C Upper AFM images (scan area, 500 Å  500 Å; scale bar, 100 Å) displaying 82-aa 824 CA RNC in the presence of oxidized ERp46 (1 nM). ERp46 molecules that appear to 825 bind 82-aa CA RNC are marked by blue squares. Lower images (scan area, 250 Å  250 826 Å; scale bar, 50 Å) highlight the regions surrounded by blue squares in the upper 827 images. 828 D Histograms of the RNC binding time of the PDI monomer (left), the PDI dimer 829 (middle), and ERp46 (right), calculated from the observed AFM images. 830 E Histograms of the distance between the edge of the ribosome and the centers of 831 RNC-neighboring PDI (left) and ERp46 (right) molecules, calculated from the observed 832 AFM images. Values represent the average distance (mean ± s.d.) calculated from curve 833 fitting with a single-Gaussian model. 834 835 Figure 7 - Role of the PDI hydrophobic pocket in PDI-mediated disulfide bond 836 introduction into RNC 82-aa 837 A Disulfide bond introduction into RNC 82-aa by PDI I289A (upper) and ERp57 838 (lower). Note that faint bands observed between “no SS” and “1SS” likely represent a 839 species in which one of cysteines is not subjected to mal-PEG modification due to 840 glutathionylation. In support of this, these minor bands are even fainter under the 841 conditions of no GSH/GSSG. 842 B Quantification of disulfide-bonded species based on the results shown in (A). 843 Quantifications for ERp46 and PDI are based on the results shown in Fig 1E and 1F. n = 844 3. 845 C HS-AFM analyses for binding of PDI I289A to RNC CA 82-aa. Upper AFM images 846 (scan area, 500 Å  500 Å; scale bar, 100 Å) display the PDI I289A molecules that bind 847 82-aa CA RNC, as marked by red squares. Lower images (scan area, 250 Å  250 Å; 848 scale bar, 50 Å) highlight the regions surrounded by red squares in the upper images. 849 D Histograms show the distribution of the RNC binding time of the PDI I289A 850 monomers (left) and dimers (right). 851 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 33 E Histogram shows the distribution of the distance between the edge of the ribosome 852 and the centers of RNC-neighboring PDI I289A molecules, calculated from the 853 observed AFM images. Values represents the average distance (mean ± s.d.) calculated 854 from curve fitting with a single-Gaussian model. 855 856 Figure 8 - Proposed model of co-translational disulfide bond introduction into 857 nascent chains by ERp46 and PDI 858 During the early stages of translation, ERp46 introduces disulfide bonds through 859 transient binding to a nascent chain. For efficient disulfide introduction by ERp46, a 860 pair of cysteines must be exposed by at least ~8 amino acids from the ribosome exit site. 861 By contrast, PDI introduces disulfide bonds by holding a nascent chain inside the 862 central cavity of the PDI homodimer during the later stages of translation, where a pair 863 of cysteines must be exposed by at least ~18 amino acids from the ribosome exit site. 864 However, when a longer polypeptide is exposed outside the ribosome, ERp46- or 865 PDI-mediated disulfide bond formation can be slower, possibly due to formation of 866 higher-order conformation in the nascent chain. Longer nascent chains may allow PDI 867 family enzymes to compete with each other for binding and acting on RNC. 868 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 34 Table 1 – Primers used in this study 869 870 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 Fig.1 A N CHSA domainⅠ (X aa) uORF2 (22 aa)FLAG (8 aa) Arrest sequence Ribosome exit tunnel ~30 aa Nascent chain 82aa (pro 6 aa + HSA 76 aa) 69 aa (pro 6 aa + HSA 63 aa) 34 53 62 75 90 91 101 168124 169 177 C C C C CC C C C C CproN C Phe70Glu57 D no SS 1 SS C no SS 1 SS 75 50 37 75 50 37 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI GSH/GSSG mal-PEG 5K 69 aa Time(s) 0 15 30 60 180 360 － － － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ GSH/GSSG mal-PEG 5K 69 aa Time(s) E PDI ERp46 no SS 1 SS 75 50 37 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp46 GSH/GSSG mal-PEG 5K 69 aa Time(s) Glutathione no SS 1 SS 75 50 37 100 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI GSH/GSSG mal-PEG 5K 82 aa Time(s) F PDI ERp46 Glutathione no SS 1 SS 75 50 37 100 IB : FLAG 0 15 30 60 180 360 － － － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ GSH/GSSG mal-PEG 5K 82 aa Time(s) no SS 1 SS 75 50 37 100 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp46 GSH/GSSG mal-PEG 5K 82 aa Time(s) CGHC CGHC CGHC -S-S- -S-S- -S-S- Trx1 Trx2 Trx3 ERp46CGHC CGHC -S-S- -S-S- a b a’b’ PDI B Hydrophobic pocket 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 0 20 40 60 80 100 Cys34 Cys53 Cys62 ** *** p=0.06 Fig.2 native 82 aa C34A A C C 34 53 62 non-native 82 aa C62A C C A 34 53 62 82 aa C53A C A C 34 53 62 non-native A B C no SS 1 SS no SS 1 SS 75 50 37 75 50 37 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp46 GSH/GSSG mal-PEG 5K 82 aa C34A (native) Time(s) 82 aa C62A (non-native) 82 aa C53A (non-native) 75 50 37 no SS 1 SS no SS 1 SS no SS 1 SS 75 50 37 75 50 37 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI GSH/GSSG mal-PEG 5K 82 aa C34A (native) Time(s) 82 aa C62A (non-native) 82 aa C53A (non-native) 75 50 37 no SS 1 SS D E Non-reducing Reducing Remaining Cys residue Non-reducing Reducing Remaining Cys residue 82 aa mono-Cys mutant + PDI Mixed No SS 75 50 37 100 82 aa mono-Cys mutant + ERp46 Mixed no SS 75 50 37 100 M ix e d d is u lf id e b o n d f o rm e d ( % ) PDI ERp46 62 53 PDI Low flexibility 62 53 ERp46 High flexibility F 82 aa C34A 82 aa C62A 82 aa C53A82 aa C53A 82 aa C62A 82 aa C34A 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) * * (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 no SS 1 SS no SS 1 SS 75 50 37 75 50 37 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp46 GSH/GSSG mal-PEG 5K 82 aa C34A [SG]5 Time(s) 82 aa C34A [SG]10 D B no SS 1 SS no SS 1 SS 75 50 37 75 50 37 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI GSH/GSSG mal-PEG 5K 82 aa C34A [SG]5 Time(s) 82 aa C34A [SG]10 82 aa C34A [SG]x + PDI C 10 SG 5 SG 0 SG 82 aa C34A [SG]x + ERp46 E 10 SG 5 SG 0 SG G M ix e d d is u lf id e b o n d f o rm e d ( % ) PDI ERp46 0 20 40 60 80 100 120 Cys34 Cys53 Cys62 n.s. n.s. n.s. F Non-reducing Reducing Remaining Cys residue 82 aa mono-Cys [SG]10 mutant + PDI Mixed no SS 75 50 37 100 82 aa mono-Cys [SG]10 mutant + ERp46 Mixed no SS 75 50 37 100 Non-reducing Reducing Remaining Cys residue A 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) 82 aa C34A [SG]10 native A C C 34 53 62 [SG]10 82 aa C34A [SG]5 native A C C 34 53 62 [SG]5 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) * * Fig.3 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 A B no SS 1 SS 50 37 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI GSH/GSSG PEG-PCMal 95 aa Time(s) no SS 1 SS 50 37 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI & ERp46 GSH/GSSG PEG-PCMal 95 aa Time(s) IB : FLAG no SS 1 SS 50 37 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp46 GSH/GSSG PEG-PCMal 95 aa Time(s) C 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) PDI ERp46 ERp46+PDI N CHSA domainⅠ (X aa) uORF2 (22 aa)FLAG (8 aa) Arrest sequence Ribosome exit tunnel ~30 aa Nascent chain 95 aa (pro 6 aa + HSA 89 aa) 34 53 62 75 90 91 101 168124 169 177 C C C C CC C C C C CproN C Thr83 Fig.4 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 N u m b e r o f fr a m e s Circularity 0 1 2 3 4 5 0 20 40 60 N u m b e r o f m o le c u le s Height (nm) 0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30 40 N u m b e r o f m o le c u le s Circularity 0.0 0.2 0.4 0.6 0.8 1.0 0 1 2 3 4 5 H e ig h t (n m ) Circularity 0.53 ± 0.10 0.80 ± 0.04 Total (n=200) B 2.7 ± 0.6 nm Total (n=200) D E A (2) 1.8 sec (3) 2.5 sec (5) 7.6 sec (1) 1.4 sec (4) 4.9 sec C 30 Å 0.0 2.5 O-shape molecule Cir:0.776 Cir:0.820 30 Å 0.0 3.0 0.0 3.7 V-shape molecule Cir:0.421 Cir:0.535 30 Å 0.0 2.4 0.0 1.8 30 Å 30 Å 30 Å30 Å30 Å30 Å 0 2 4 6 8 10 0 0.2 0.4 0.6 0.8 1 C ir c u la ru ty Time (sec) (1) (2) (3) (5) (4) O-shape V-shape O-shape V-shape 0.58 ± 0.06 0.78 ± 0.05 N u m b e r o f m o le c u le s N u m b e r o f m o le c u le s circularity height (nm) circularity h e ig h t (n m ) 0 0 2 4 6 8 10 0.2 Time (s) 0.4 0.6 0.8 1 c ir c u la ri ty N u m b e r o f fr a m e s circularity Fig.5 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 0 10 20 30 40 50 60 0 5 10 15 20 N u m b e r o f m o le c u le s Binding time (sec) 0 5 10 15 20 25 30 35 40 0 5 10 15 20 N u m b e r o f m o le c u le s Distance (nm) 0 5 10 15 20 25 30 35 40 0 5 10 15 20 N u m b e r o f m o le c u le s Distance (nm) 16.9 ± 4.7 nm 15.7 ± 4.1 nm B D C PDI monomer ERp46 PDI ERp46 E PDI dimer 100 Å 50 Å 82-aa CA RNC + Oxidized PDI monomer dimer dimer c lo s e d -u p 0.0 15.2 0.0 13.0 0.0 18.0 0.0 6.0 0.0 5.8 0.0 14.5 50 Å 150 Å 82-aa CA RNC + Oxidized ERp46 monomer monomer c lo s e d -u p 0.0 14.7 0.0 7.9 0.0 18.1 0.0 10.9 0.0 19.0 0.0 20.4 0.0 17.9 A 0 10 20 30 40 50 60 0 10 20 30 40 N u m b e r o f m o le c u le s Binding time (sec) 100 Å 100 Å100 Å 100 Å100 Å 50 Å 50 Å 150 Å 50 Å 0 10 20 30 40 50 60 0 5 10 15 20 25 N u m b e r o f m o le c u le s Binding time (sec) N u m b e r o f m o le c u le s Binding time (s) N u m b e r o f m o le c u le s Binding time (s) N u m b e r o f m o le c u le s Binding time (s) N u m b e r o f m o le c u le s Distance (nm) N u m b e r o f m o le c u le s Distance (nm) Fig. 6 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 no SS 1 SS A no SS 1 SS 75 50 37 100 75 50 37 100 IB : FLAG 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ PDI I289A GSH/GSSG mal-PEG 5K 82 aa Time(s) 0 15 30 60 180 360 － － － － － ＋ － ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋ ERp57 GSH/GSSG mal-PEG 5K 82 aa Time(s) B PDI ERp46 ERp57 PDI I289A 0 20 40 60 80 100 0 60 120 180 240 300 360 D is u lf id e b o n d in tr o d u c ti o n (% ) Time (s) c lo s e d -u p monomer dimer 0.0 18.0 0.0 15.3 0.0 4.4 0.0 3.7 100 Å100 Å 50 Å 50 Å 100 Å 50 Å dimer 0.0 14.0 0.0 4.8 82-aa CA RNC + Oxidized PDI I289A 0 10 20 30 40 50 60 0 2 4 6 8 10 N u m b e r o f m o le c u le s Binding time (sec) 0 10 20 30 40 50 60 0 10 20 30 40 50 N u m b e r o f m o le c u le s Binding time (sec) PDI I289A monomer PDI I289A dimer N u m b e r o f m o le c u le s Binding time (s) N u m b e r o f m o le c u le s Binding time (s) 0 5 10 15 20 25 30 35 40 0 10 20 30 N u m b e r o f m o le c u le s Distance (nm) PDI I289A N u m b e r o f m o le c u le s Distance (nm) 15.0 ± 3.5 nm C D E Fig. 7 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 18 aa Ribosome cytosol ER lumen 8 aa ーSH ERp46 PDI PDIERp46 competition Fig. 8 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 1 Expanded View 1 2 Distinct roles and actions of PDI family enzymes in catalysis of nascent-chain 3 disulfide formation 4 5 Chihiro Hirayama1, Kodai Machida2#, Kentaro Noi3#, Tadayoshi Murakawa4, Masaki 6 Okumura1,5, Teru Ogura6,7, Hiroaki Imataka2, and Kenji Inaba1* 7 8 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 2 9 10 Figure EV1 - Redox states of PDI and ERp46 in glutathione redox buffer and 11 disulfide bond introduction into 82 aa C34A, catalyzed by PDI a domain 12 A Redox states of PDI and ERp46 in the presence of 1 mM GSH and 0.2 mM GSSG. 13 Purified PDI and ERp46 were incubated for 6 mins at 30 ºC in the above glutathione 14 redox buffer and modified with 2 mM mal-PEG 5K for separation on SDS gels. 15 B Quantification based on the results shown in (A). 16 17 18 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 3 19 20 Figure EV2 - Statistical analysis of RNC molecules observed by HS-AFM in the 21 presence or absence of PDI/ERp46 22 A Number of particles observed for NC-RNC or 82-aa CA RNC molecules present in 23 isolation or bound to PDI/ERp46 molecules. 24 B Ratio of NC-RNC or 82-aa CA RNC molecules present in isolation or bound to 25 PDI/ERp46, calculated based on the observed number of particles in (A). Note that a 26 minor portion of NC-RNC or 82-aa CA RNC molecules were bound to many ERp46/PDI 27 molecules, possibly due to serious structural damages of the RNC molecules. 28 29 30 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 4 31 Figure EV3 - Representative time-course snapshots captured by HS-AFM for 82-aa 32 CA RNC bound to the PDI monomer (A), the PDI dimer (B), and ERp46 (C). 33 A Time-course snapshots captured by HS-AFM for the PDI monomer binding to 82-aa 34 CA RNC. The AFM images (scan area, 650 Å  650 Å; scale bar, 130 Å) displaying 82-35 aa CA RNC in the presence of oxidized PDI (1 µM). White arrows indicate the 36 monomeric PDI molecules that bind to 82-aa CA RNC. See also supplementary video 2. 37 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 5 B Time-course snapshots captured by HS-AFM for the PDI dimer binding to 82-aa CA 38 RNC. The AFM images (scan area, 700 Å  700 Å; scale bar, 140 Å) displaying 82-aa 39 CA RNC in the presence of oxidized PDI (1 µM). White arrows indicate the dimeric PDI 40 molecules that bind to 82-aa CA RNC. See also supplementary video 3. 41 C Time-course snapshots captured by HS-AFM for ERp46 binding to 82-aa CA RNC. 42 The AFM images (scan area, 1,000 Å  1,000 Å; scale bar, 200 Å) displaying 82-aa CA 43 RNC in the presence of oxidized ERp46 (1 µM). White arrows indicate the ERp46 44 molecules that bind to 82-aa CA RNC. See also supplementary video 4. 45 46 47 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 6 48 Figure EV4 - Representative time-course snapshots captured by HS-AFM for 82-aa 49 CA RNC bound to the PDI I289A monomer (A), and the PDI I289A dimer (B). 50 A Time-course snapshots captured by HS-AFM for the PDI I289A monomer binding to 51 82-aa CA RNC. The AFM images (scan area, 900 Å  900 Å; scale bar, 200 Å) displaying 52 82-aa CA RNC in the presence of oxidized PDI I289A (1 µM). White arrows indicate the 53 monomeric PDI I289A molecules that bind to 82-aa CA RNC. See also supplementary 54 video 5. 55 B Time-course snapshots captured by HS-AFM for the PDI I289A dimer binding to 82-56 aa CA RNC. The AFM images (scan area, 800 Å  800 Å; scale bar, 200 Å) displaying 57 82-aa CA RNC in the presence of oxidized PDI I289A (1 µM). White arrows indicate the 58 dimeric PDI I289A molecules that bind to 82-aa CA RNC. See also supplementary video 59 6. 60 61 62 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348 7 Movie EV1 - HS-AFM movies showing structure dynamics of oxidized ERp46. This 63 movie is a source of the time-course snapshots shown in Fig 5C. 64 65 Movie EV2 - HS-AFM movies showing the binding of the PDI monomer to 82-aa CA 66 RNC. This movie is a source of the time-course snapshots shown in supplementary Fig 67 EV3A. 68 69 Movie EV3 - HS-AFM movies showing the binding of the PDI dimer to 82-aa CA 70 RNC. This movie is a source of the time-course snapshots shown in supplementary Fig 71 EV3B. 72 73 Movie EV4 - HS-AFM movies showing the binding of ERp46 to 82-aa CA RNC. This 74 movie is a source of the time-course snapshots shown in supplementary Fig EV3C. 75 76 Movie EV5 - HS-AFM movies showing the binding of the PDI I289A monomer to 77 82-aa CA RNC. This movie is a source of the time-course snapshots shown in 78 supplementary Fig EV4A. 79 80 Movie EV6 - HS-AFM movies showing the binding of the PDI I289A dimer to 82-81 aa CA RNC. This movie is a source of the time-course snapshots shown in 82 supplementary Fig EV4B. 83 84 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 5, 2021. ; https://doi.org/10.1101/2021.01.04.425348doi: bioRxiv preprint https://doi.org/10.1101/2021.01.04.425348