High-throughput amino acid-level characterization of the interactions of plasminogen activator inhibitor-1 with variably divergent proteases

1 While members of large paralogous protein families share structural features, their functional 2 niches often diverge significantly. Serine protease inhibitors (SERPINs), whose members 3 typically function as covalent inhibitors of serine proteases, are one such family. Plasminogen 4 activator inhibitor-1 (PAI-1) is a prototypic SERPIN, which canonically inhibits tissue- and 5 urokinase-type plasminogen activators (tPA and uPA) to regulate fibrinolysis. PAI-1 has been 6 shown to also inhibit other serine proteases, including coagulation factor XIIa (FXIIa) and 7 transmembrane serine protease 2 (TMPRSS2). The structural determinants of PAI-1 inhibitory 8 function toward these non-canonical protease targets, and the biological significance of these 9 functions, are unknown. We applied deep mutational scanning (DMS) to assess the effects of 10 ~80% of all possible single amino acid substitutions in PAI-1 on its ability to inhibit three putative 11 serine protease targets (uPA, FXIIa, and TMPRSS2). Selection with each target protease 12 generated a unique PAI-1 mutational landscape, with the determinants of protease specificity 13 distributed throughout PAI-1’s primary sequence. Next, we conducted a comparative analysis of 14 extant orthologous sequences, demonstrating that key residues modulating PAI-1 inhibition of 15 uPA and FXIIa, but not TMPRSS2, are maintained by purifying selection. PAI-1’s activity toward 16 FXIIa may reflect how protease evolutionary relationships predict SERPIN functional 17 divergence, which we support via a cophylogenetic analysis of all secreted SERPINs and their 18 cognate serine proteases. This work provides insight into the functional diversification of 19 SERPINs and lays the framework for extending these studies to other proteases and their 20 regulators. 21 22 23 (which was not certified by peer review) is the author/funder.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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 3

1 Large paralogous protein families in which members share a common protein structure are 2 ubiquitous across all clades of life. Diversity often results from gene duplication events that 3 occurred during either whole genome duplication or as segmental duplications of one or more 4 genes (Hahn 2009). To acquire and maintain distinct functions despite similarities in sequence 5 and structure, members of these paralogous protein families must acquire unique functional 6 niches (McClune, et al. 2019; McClune and Laub 2020; Nocedal and Laub 2022). Revealing the 7 shape and complexity of these niches is crucial to understanding the emergence and insulation 8 of protein-protein interactions. 9 10 Serine protease inhibitors, or SERPINs, are a protein superfamily with high functional variability 11 and specificity. Inhibitory SERPINs are defined by their ability to efficiently and specifically 12 inhibit one or more members of the highly paralogous serine protease protein family. Both 13 protein families are found across all domains of life, where SERPINs covalently bind serine 14 proteases, irreversibly inhibiting them (Spence, et al. 2021). In humans and other vertebrates, 15 SERPINs have been shown to regulate a range of processes, including immune/inflammatory 16 responses, coagulation/fibrinolysis, and extracellular matrix remodeling (Sanrattana, et al. 17 2019). While all SERPINs share a common protein fold, most SERPINs perform their biological 18 function as specific and irreversible inhibitors of serine proteases via a “mousetrap” mechanism 19 (Fig. 1) (Law, et al. 2006). 20 21 In their active state, SERPINs exist in a metastable conformation in which the reactive center 22 loop (RCL), which contains an amino acid sequence that mimics that of their target serine 23 protease’s preferred cleavage site, extends from the globular core of the protein. Upon 24 SERPIN/protease engagement, irreversible inhibition of the serine protease occurs when the 25 RCL inserts into the SERPIN’s central -sheet A prior to the resolution of the acyl intermediate; 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 4 however, if hydrolysis of the acyl intermediate occurs prior to RCL insertion, the SERPIN serves 1 as a substrate rather than an inhibitor of the serine protease–both of these conformations 2 occupy a lower energy space (Fig. 1) (Lawrence, et al. 1995; Gettins 2002; Law, et al. 2006; 3 Huntington 2011). Although a SERPIN’s protease specificity is derived in part from the RCL 4 amino acid sequence, structural features located distal to the RCL also contribute significantly 5 (Gettins and Olson 2009; Marijanovic, et al. 2019). Some SERPINs appear to be highly specific 6 inhibitors of a single protease, while others exhibit activity toward additional protease targets, 7 often with decreased efficiency compared to their canonical protease target (Gettins 2002). 8 Exosite interactions, as well as a limited number of mutations, are also known to alter SERPIN 9 inhibitory profiles (Gettins 2002; Huntington 2011). The specificity of SERPINs for their target 10 serine protease(s) therefore provides not only a novel system for studying the development of 11 functional niches in the context of paralog evolution, but also one where the determinants of 12 success or failure of the protein-protein interaction are complex. 13 14 Plasminogen activator inhibitor-1 (PAI-1, encoded by the SERPINE1 gene) is a 379 amino acid 15 SERPIN canonically considered to be a regulator of fibrinolysis, the enzymatic degradation of 16 fibrin clots, by inhibiting two closely related proteases, urokinase and tissue-type plasminogen 17 activators (uPA (encoded by the PLAU gene) and tPA (encoded by the PLAT gene), 18 respectively) with high efficiency (second-order rate constants of ~106-107 M-1 s-1, (Sherman, et 19 al. 1992)) and a stoichiometry of inhibition (SI) of approximately one (Lawrence, et al. 2000). 20 PAI-1 has also been identified as a possible inhibitor of other hemostatic proteases, including 21 the procoagulant factor (F) XIIa (encoded by the F12 gene), which stands in stark juxtaposition 22 to PAI-1’s primary function as an inhibitor of the clot dissolving proteases tPA and uPA 23 (Berrettini, et al. 1989; Keijer, et al. 1991; Rezaie 2001; Tanaka, et al. 2009; Sen, et al. 2011; 24 Puy, et al. 2019). PAI-1 also inhibits transmembrane serine protease 2 (TMPRSS2, SI Fig. 1), 25 through which it mediates the course of viral infections such as influenza and SARS-CoV-2 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 5 (Dittmann, et al. 2015; Shen, et al. 2017), with increased PAI-1 levels reported in response to 1 influenza (Keller, et al. 2006; Bouwman, et al. 2009) and SARS-CoV-2 infections (Al-Samkari, et 2 al. 2020; Zuo, et al. 2021). 3 4 In the present work, we use deep mutational scanning (DMS) to analyze the mutational 5 landscape of PAI-1 inhibition of several proteases, including uPA and two non-canonical targets 6 of PAI-1 (FXIIa and soluble TMPRSS2), characterizing the overall degree and potential 7 regionalization of marginal specificity in these interactions. We follow prior work that defines 8 protein sequence space as the set of all possible amino acids in a protein coupled with the 9 possible evolutionary trajectories to a new sequence (McClune and Laub 2020). Within a 10 sequence space, robust and marginal specificity refer to members of paralogous protein 11 superfamilies for which multiple mutations are required to alter their specificity and cases where 12 a single amino acid substitution can alter a paralog’s specificity, respectively (Ghose, et al. 13 2023). We then use comparative evolutionary analyses to contextualize the results of our DMS 14 screen to explore the relationships between protease conservation, paralogous protein 15 relationships, and the sequence space in which these proteins interact. 16 17

and Discussion 18 Construction and characterization of a PAI-1 variant library on the I91L background. 19 We constructed a novel phage display library as previously described (Huttinger, et al. 2021; 20 Haynes, et al. 2022), though with the addition of an I91L PAI-1 variant backbone to increase the 21 half-life of PAI-1 in its metastable active conformation from 1-2 h to ~19 h (Fig. 1) (Berkenpas, 22 et al. 1995; Haynes, et al. 2022). The I91L background was chosen to isolate the effects of 23 mutants on PAI-1 serine protease target specificity from those impacting functional stability. 24 25 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 6 The I91L PAI-1 variant library exhibited a depth of 6.6x106 unique phage clones with an average 1 of three missense or nonsense mutations per clone (SI Fig. 2A). Of the 7,201 possible single 2 amino acid substitutions in PAI-1 (379x19), 5,688 (79%) are represented in the I91L variant 3 library as determined by high-throughput sequencing (HTS). The abundance of each variant is 4 highly correlated (slope = 0.95, R2 = 0.85; SI Fig. 2B) with our previous variant library on the 5 wild-type (WT) PAI-1 background (Huttinger, et al. 2021). Log2-fold enrichment scores for uPA 6 functional inhibitory variants in the I91L PAI-1 library are also highly correlated with those 7 previously determined (Huttinger, et al. 2021) on the WT PAI-1 background (SI Fig. 2C; R2 = 8 0.66, slope = 0.45). 9 10 Single amino acid substitutions in PAI-1 differentially affect its ability to inhibit target 11 proteases. We next screened our I91L PAI-1 variant library to identify the effects of single 12 amino acid substitutions in PAI-1 on its ability to inhibit FXIIa and TMPRSS2 in addition to uPA 13 (Fig. 2A). Selection with each protease generated a unique PAI-1 variant map (Fig. 2A). 14 15 The statistically significant functional log2-fold enrichment scores are most similar between uPA 16 and FXIIa, although there is a distinct set of mutations that maintain inhibitory activity toward all 17 three proteases (Fig. 2B). Similarly, principal component analyses (PCA) of the differential 18 variant enrichment data for PAI-1 specificity towards each of the target proteases (Fig. 2C) are 19 most similar for uPA and FXIIa, with TMPRSS2 yielding a relatively distinct pattern of 20 enrichment. The large separation of the principal components required for TMPRSS2 inhibition 21 from that of uPA and FXIIa is consistent with the closer phylogenetic relationship between the 22 latter two proteases (Yousef, et al. 2003) and indicative of more distinct amino acid substitutions 23 required to render PAI-1 a specific/efficient TMPRSS2 inhibitor. We further examined the unique 24 set of amino acid substitutions that render PAI-1 capable (Fig. 2D) or incapable (Fig 2E) of 25 inhibiting each of these three target proteases using the statistical thresholds described above. 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 7 Again, the set of amino acid substitutions in PAI-1 that promote/maintain or result in a loss of 1 uPA or FXIIa inhibition are more similar to each other than either is to those that impact 2 TMPRSS2 inhibition. More amino acid substitutions pass the statistical threshold (padj < 0.1 and 3 BaseMean score ≥ 50) for uPA inhibition (n = 1369) than for either FXIIa (n = 972) or TMPRSS2 4 (n = 704) (Table 1), suggesting that the more distantly related a protease is to PAI-1’s canonical 5 target uPA, the fewer possible “pathways” to inhibition exist. 6 7 The results of our DMS screens of PAI-1 inhibition of three distinct serine protease targets 8 suggest that multiple single amino acid substitutions may result in PAI-1 being a TMPRSS2 9 inhibitor with impaired or reduced inhibition of its canonical target uPA (Fig. 2B). In other words, 10 our results suggest that PAI-1 may exhibit marginal specificity with respect to TMPRSS2 11 inhibition as multiple single amino acid substitutions appear to alter its protease specificity 12 (McClune and Laub 2020; Ghose, et al. 2023). In contrast, the log2-fold enrichment scores of 13 the PAI-1 variant library with respect to uPA and FXIIa inhibition closely mirror each other (Fig. 14 2). We hypothesize that this pattern is the result of PAI-1 exhibiting robust specificity for the 15 closely related serine proteases uPA and FXIIa in which single amino acid substitutions do not 16 dramatically alter its protease inhibition profile. However, it remains unclear if PAI-1's inhibition 17 of FXIIa results from cross-reactivity between two closely related proteins or if PAI-1 is being 18 actively selected to be an inhibitor of both uPA and FXIIa (Yousef, et al. 2003; Conant and 19 Wolfe 2008). In the latter scenario, there is an active selection for PAI-1 to inhibit FXIIa, while in 20 the former, PAI-1 cross-reactivity with FXIIa may be the result of an inherent difficulty in 21 SERPINs evolving insular functional niches with respect to closely related serine proteases 22 (McClune, et al. 2019; McClune and Laub 2020). 23 24 Specificity determinants span PAI-1’s primary structure. To interrogate the role of multiple 25 regions of PAI1 in protease specificity, we next performed separate PCAs on each of the twelve 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 8 150 base pair (bp) sequencing amplicons (Fig. 3, SI Table 1) (Huttinger, et al. 2021; Haynes, et 1 al. 2022). Across all twelve amplicons, principal components (PC) 1 and 2 discriminate between 2 the log2-fold enrichment scores with respect to inhibition of uPA, FXIIa, and TMPRSS2, 3 demonstrating that SERPIN regions beyond the RCL (residues 331-350 of human PAI-1 located 4 within amplicon 12) (Gettins and Olson 2009) contribute to target protease specificity. Previous 5 reports of chimeric PAI-1 variants with RCLs from SERPINs targeting alternative proteases 6 retaining the ability to inhibit uPA/tPA are consistent with this finding (Lawrence, et al. 1990). 7 8 In contrast, PAI-1 variants with amino acid substitutions in the RCL have been reported to alter 9 PAI-1’s specificity to be either a neutrophil elastase or cathepsin G inhibitor (Stefansson, et al. 10 2004), similar to the naturally occurring Pittsburgh variant of the SERPIN 1-anti-trypsin (A1AT, 11 SERPINA1), in which a single amino acid substitution at the P1 position in the RCL converts it 12 from an elastase to a thrombin inhibitor (Lewis, et al. 1978; Owen, et al. 1983), or engineering 13 efforts on other SERPIN backbones to generate SERPINs with novel inhibitory profiles (Scott, et 14 al. 2014; Polderdijk, et al. 2017; Bhakta, et al. 2021; Sanrattana, et al. 2021; Singh, et al. 2022). 15 Therefore, although single amino acid substitutions in a SERPIN’s RCL may function in a 16 manner consistent with marginal specificity for its target protease, our data suggest that multiple 17 amino acid substitutions throughout the SERPIN’s primary sequence may be required to 18 achieve robust specificity for a SERPIN as an efficient and specific inhibitor of a given protease. 19 Furthermore, these variants may contribute to the coevolution of specific SERPIN::serine 20 protease inhibitory reactions through as of yet not fully understood epistatic mechanisms 21 (Spence, et al. 2021; Ding, et al. 2022; Park, et al. 2022) that may improve a SERPIN’s 22 tolerance for efficiently inhibiting one serine protease over another. 23 24 Evolutionary conservation of PAI-1 inhibitory functions. Next, to better understand the 25 evolution of PAI-1’s inhibitory niche, we compared the results of our DMS screens to PAI-1 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 9 sequences of extant mammalian species (n = 94). We previously demonstrated that key 1 residues in PAI-1 for inhibition of uPA are conserved in nature by comparing the results of our 2 DMS screen to natural sequence variation across mammals (Huttinger, et al. 2021). We 3 extended these analyses to the present screen to assess whether sites mediating PAI-1 4 inhibition of FXIIa and TMPRSS2 exhibit similar selective pressures. An evolutionary 5 conservation score for a given site in the PAI-1 amino acid sequence was determined 6 (Ashkenazy, et al. 2016; Huttinger, et al. 2021) and correlated with the tendency of a site to 7 accept mutations that allow PAI-1 to inhibit a target protease (denoted henceforth as the 8 normalized functional score). If there is purifying selective pressure for PAI-1 to maintain the 9 inhibition of a given target protease, then we expect that sites in PAI-1 at which amino acid 10 substitutions do not negatively impact the ability to inhibit the target protease will be less 11 conserved across extant species. Likewise, sites at which amino acid substitutions render PAI-1 12 unable to inhibit the target protease will be more conserved. 13 14 The results of these analyses corroborate our earlier findings that PAI-1 inhibition of uPA is 15 under purifying selection (Huttinger, et al. 2021) with a modest, statistically significant 16 correlation between the normalized functional score and evolutionary conservation score across 17 PAI-1 specificity space (slope = 0.58, R2 = 0.1427, p = 5.6x10-14, Fig. 4A). With respect to FXIIa 18 inhibition by PAI-1, the functional and evolutionary conservation scores associate with a smaller 19 slope and weaker, yet statistically significant, correlation (slope = 0.33, R2 = 0.066, p = 7.5x10-7, 20 Fig. 4B). However, no statistically significant association was observed between functional and 21 evolutionary conservation score with respect to PAI-1 inhibition of TMPRSS2 (slope = -0.01, R2 22 = 8.7x10-5, p = 0.87, Fig. 4C). First, these results suggest that, as expected, PAI-1 inhibition of 23 uPA is under purifying selection. Second, either through FXIIa’s close evolutionary relationship 24 with uPA or via direct effects of selection, sites impacting FXIIa inhibition are also under 25 purifying selection. Meanwhile sites impacting inhibition of TMPRSS2 are not, on average, 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 10 under differential selection compared to other sites in the molecule. These results suggest that 1 PAI-1 inhibition of TMPRSS2 is potentially an example of cross-functional SERPIN reactivity 2 resulting in inhibition of a non-canonical target and that inhibition of TMPRSS2 is not a 3 biologically significant function of PAI-1 that generates detectable signatures of selection across 4 species. 5 6 Given that a greater proportion of PAI-1 sequence variation across extant species is explained 7 by a drive to maintain PAI-1’s ability to inhibit uPA than FXIIa, we speculate that FXIIa inhibition 8 by PAI-1 may be the result of cross-reactivity between uPA and FXIIa, given their close 9 phylogenetic relationship, while TMPRSS2 represents a distinct outgroup to the uPA/FXIIa 10 clade (Yousef, et al. 2003). The F12 gene appeared in stem tetrapods after their divergence 11 from fish, and therefore interactions between PAI-1 and uPA predate the existence of FXIIa 12 (Ponczek, et al. 2008; Ponczek, et al. 2020). Insulation of PAI-1 from crosstalk to FXIIa may 13 have been prevented by the evolutionary inertia of millions of years of preceding coevolution 14 between PAI-1 and plasminogen activators. 15 16 Comparing the effects of single amino acid substitution in PAI-1 on the inhibition of uPA 17 and FXIIa. Although PAI-1 inhibits both uPA and FXIIa, PAI-1 inhibition of uPA is anti-fibrinolytic 18 (stabilizing the thrombus) while inhibition of FXIIa is anti-thrombotic. One potential explanation 19 for these opposite effects, is that due to the recent divergence of uPA and FXIIa, PAI-1’s ability 20 to inhibit uPA has been “carried over” into its ability to inhibit FXIIa, as evidenced by the 21 similarity in PAI-1’s specificity fingerprints for both uPA and FXIIa (Figs. 2 and 5). To further 22 address the similar effects of amino acid substitution in PAI-1 on its ability to inhibit uPA and 23 FXIIa, we next tested the effects of the most enriched amino acid substitutions in PAI-1 with 24 respect to the inhibition of both of these proteases. We identified the most enriched single 25 amino acid substitutions in our DMS screens for PAI-1 inhibition of both uPA and FXIIa on the 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 11 I91L background as being F98Y, T251I, and S331C (Fig. 5A and 5B). The WT human amino 1 acids at these positions are highly conserved in extant mammalian species, possibly due to the 2 requirement of the I91L substitution or a general extension of PAI-1’s functional half-life to open 3 up the functional accessibility of these substitutions. 4 5 These variants were expressed as recombinant proteins, and their SIs with respect to uPA and 6 FXIIa were determined (Fig. 5C). For a SERPIN to function as a specific and efficient inhibitor 7 of a given serine protease, it must have a sufficiently rapid second-order rate constant and a SI 8 close to “1” such that virtually all the reactions proceed along the inhibitory pathway with little to 9 no substrate pathway observed (Olson and Gettins 2011). The SI for FXIIa inhibition was 10 consistently 2-3 fold higher than that of uPA inhibition—indicating that when interacting with 11 FXIIa, in contrast to uPA, PAI-1 is more likely to utilize the substrate versus the inhibitory 12 pathway (Fig. 1). Of note, these variants were not the most enriched variants in our original 13 DMS screen on the WT PAI-1 background (Huttinger, et al. 2021), with the F98Y and S331C 14 variants actually classified as non-functional against WT PAI-1 (SI Fig. 2C). This finding 15 suggests that these mutations are epistatic with the I91L substitution with respect to PAI-1 16 function (Berkenpas, et al. 1995) or that these variants decrease the functional stability of PAI-1 17 such that their half-lives are too short to be detected on the WT background in our assay, 18 despite being otherwise inhibitory variants. One potential implication of epistasis between the 19 I91L and T98Y and/or S331C substitutions, with respect to PAI-1 inhibitory function, is that 20 although the T98Y and S331C substitutions are functional inhibitors on the I91L background, 21 this functionality of these substitutions is not available to PAI-1 in the absence of the I91L 22 substitution. 23 24 As the three most highly enriched PAI-1 variants identified in our DMS screens for both uPA and 25 FXIIa inhibition were the same (Figs. 5A and 5B), we next compared the evolutionary 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 12 conservation scores for PAI-1 with respect to both uPA and FXIIa inhibition (Fig. 5D). Not only 1 are the normalized functional scores for uPA and FXIIa highly correlated with each other (slope 2 = 0.78, R2 = 0.43, p = 2.2x10-16) as expected (Fig. 2B); but amino acid sites that are conserved 3 across extant mammalian species are less likely to accept mutations in both DMS screens 4 (quadrant I, QI), while sites that are more evolutionary labile are more likely to accept mutations 5 (quadrant III, QIII). The mean evolutionary conservation scores (± standard deviation) between 6 Q1 (0.45 ± 1.21) and QIII (-0.36 ± 0.70) are highly statistically significantly different (padj < 7 0.0001, Tukey’s HSD test), suggesting that amino acid substitutions in PAI-1 that would impact 8 its protease specificity for either uPA or FXIIa are likely to affect inhibition of both proteases. 9 10 It has been speculated that the high SI for PAI-1 inhibition of FXIIa effectively results in FXIIa 11 inactivation of PAI-1, further reducing its potential as an antifibrinolytic (Tanaka, et al. 2009). As 12 the inhibition of both uPA and FXIIa occupy similar PAI-1 sequence spaces, the amino acid 13 substitutions that were identified as the best FXIIa inhibitors in our DMS screen were still more 14 efficient uPA inhibitors (Fig. 5)—consistent with previous studies (Berrettini, et al. 1989; 15 Sherman, et al. 1992). Overall, our data indicate that PAI-1 maintains its function as an inhibitor 16 of these two proteases (although there is a “preference” to inhibit uPA over FXIIa), with few 17 single amino acid substitutions (n = 16) conferring a preference for inhibition of FXIIa over uPA 18 (Figs. 2B and 5A-B). This is in contrast to PAI-1 inhibition of TMPRSS2, in which multiple single 19 amino acid substitutions (n = 184) appear to improve specificity toward TMPRSS2 over 20 FXIIa/uPA(Fig. 2B). We, therefore, hypothesized that PAI-1 has maintained inhibitory activity 21 toward FXIIa in addition to uPA as a function of the close evolutionary relationship between 22 these serine proteases(Yousef, et al. 2003; McClune, et al. 2019). In essence, our results 23 suggest closely related serine proteases form a crowded sequence space with regard to 24 SERPIN inhibition. An expectation arising from these findings is that, beyond PAI-1, a given 25 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 13 SERPIN will inhibit groups of closely related serine proteases and patterns of SERPIN and 1 protease diversification may mirror one another. 2 3 Coevolution of serpins and their cognate serine proteases. To test the role of phylogenetic 4 relatedness and codiversification in driving interactions between SERPINs and their cognate 5 serine proteases, we next performed a co-phylogenetic analysis of human secreted SERPINs 6 and their cognate serine proteases. Cophylogenetic analyses are typically used to study 7 coevolution in host-parasite interactions, where the null hypothesis is that the shapes of the two 8 phylogenies of each putatively coevolving group are not mutually informative of one another 9 (Legendre, et al. 2002). We reasoned that a similar statistic of mutual information in 10 phylogenetic topologies would quantify SERPIN:serine protease codiversification. We therefore 11 curated a list of human secreted SERPINs (n = 13) with well-described cognate serine 12 proteases (n = 16) (Olson and Gettins 2011; Heiker, et al. 2013; Godinez, et al. 2022; 13 Janciauskiene, et al. 2024). 14 15 Our cophylogenetic analysis (Fig. 6) supports the hypothesis that the secreted SERPINs and 16 serine proteases have co-diversified (p = 0.039). The bulk of this mutual signal appears to 17 reside deep in the evolution of these groups, reflected in the SERPINA clade largely inhibiting 18 the KLK/elastase clade, whereas the other SERPINs, such as the clade formed by group I/E/G 19 SERPINs, largely inhibit the related complement system and coagulation proteases. Notably for 20 the current study, the closely related SERPINE1, SERPINE2, and SERPINI1 all inhibit uPA 21 (PLAU) and tPA (PLAT) and most likely coevolved from an ancestral SERPIN, with the 22 emergence of the two plasminogen activators from a common serine protease ancestor that 23 also gave rise to FXIIa (F12) (Jendroszek, et al. 2019). Similarly, the closely related serine 24 proteases thrombin (encoded by the F2 gene) and FXa (encoded by the F10 gene) are both 25 inhibited by a common SERPIN, antithrombin (encoded by the SERPINC1 gene), with which 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 14 they also coevolved (Fig. 7) (Yousef, et al. 2003). Therefore, we anticipate that a DMS screen 1 of antithrombin’s inhibition of thrombin and FXa would also reveal overlapping effects of single 2 amino acid substitutions. 3 4

5 Inhibitory SERPINs have evolved to fulfill unique functional niches in which they inhibit one or a 6 few proteases with residual inhibitory activity towards additional proteases (Gettins 2002; Olson 7 and Gettins 2011). Here, we have harnessed the power of DMS and combined it with 8 comparative evolutionary approaches to assess PAI-1’s inhibitory function towards its canonical 9 target, uPA, as well as non-canonical targets that are both closely (FXIIa) and distantly 10 (TMPRSS2) related to uPA. Our data demonstrate that the determinants of PAI-1 specificity are 11 scattered throughout its amino acid sequence space, with no one region of the primary 12 sequence completely dictating the specificity of a target protease, suggesting that residues 13 within the PAI-1 structure mediate long-range effects that help dictate its specificity as a 14 protease inhibitor. Furthermore, we find that residues modulating PAI-1 inhibition of uPA and 15 FXIIa are under purifying selection, with no such selective signature for inhibition of TMPRSS2. 16 PAI-1 instead exhibits marginal specificity towards TMPRSS2, but robust specificity that 17 prevents it from becoming a FXIIa specialist. As uPA and FXIIa are closely related serine 18 proteases, the ability of PAI-1 to be a specific inhibitor of uPA plasminogen activators, but not 19 FXIIa, is restricted by a locally crowded sequence space. Coevolution of SERPINs with their 20 cognate target serine protease may explain many overlapping and non-specific SERPIN::serine 21 protease interactions, and could help guide future efforts to engineer SERPIN specificity. In 22 conclusion, we have demonstrated the power of DMS to interpret the evolution and functional 23 specificity of members of the SERPIN protein superfamily and anticipate that similar analyses 24 have the potential to improve understanding of the mechanisms driving the diversification and 25 specialization of other large paralogous protein families. 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 15 1 2 3

4 Phage displayed PAI-1 variant library preparation. A phage display PAI-1 variant library was 5 constructed on the I91L PAI-1 background using error prone PCR as previously described (7, 8) 6 with the GeneMorph II Random Mutagenesis Kit (Agilent Technologies, Santa Clara, CA) using 7 primers that preserved the AscI and NotI restriction sites, and cloned into a modified pAY-FE 8 plasmid (Genebank #MW464120) in which an amber stop codon (TAG) immediately preceding 9 the SERPINE1-gIII fusion construct was mutated to glutamine (CAG, Gln) to increase 10 expression of the PAI-1-pIII coat fusion protein (pLMH1, SI Table 1) (Huttinger, et al. 2021; 11 Haynes, et al. 2022). Following cloning of the I91L PAI-1 variant library into the pLMH1 plasmid, 12 the library was transformed into electrocompetent XL-1 Blue MRF’ E. coli (Agilent 13 Technologies). Library depth was determined by counting the number of ampicillin-resistant 14 colonies. Mutation frequency was estimated by Sanger sequencing of the PAI-1 (SERPINE1) 15 inserts from randomly selected colonies (n=24). 16 17 The I19L PAI-1 phage displayed library was produced as described previously (Huttinger, et al. 18 2021; Haynes, et al. 2022). Briefly, E. coli harboring the I91L PAI-1 library were grown in LB 19 media supplemented with 2% glucose and ampicillin (0.1 mg/mL) to mid-log phase at 37oC, 20 infected with M13KO7 helper phage (Cytiva) for 1 h, and transferred to 2xYT media 21 supplemented with ampicillin (0.1 mg/mL), kanamycin (0.03 mg/mL) and IPTG (0.4 mM) to 22 induce expression of the PAI-1 phage displayed library and grown for 2h at 37oC. E. coli were 23 removed by centrifugation (4200xg and 4500xg at 4oC for 10 min) and phage were precipitated 24 from the supernatant with PEG-8000 (2.5% w/v) and NaCl (0.5 M) overnight at 4oC. Phage were 25 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 16 pelleted by centrifugation (20,000xg for 20 min at 4oC) and resuspended in 50 mM Tris 1 containing 150 mM NaCl at pH 7.4 (TBS). 2 3 PAI-1 variant selection. uPA selection assays were performed as previously described 4 (Huttinger, et al. 2021; Haynes, et al. 2022). Human coagulation FXIIa (Innovative Research, 5 Novi, MI) was biotinylated (FXIIa-biotin) and the degree of labeling assessed using the EZ-Link 6 Sulfo-NHS-LC-Biotin labelling kit (ThermoFisher Scientific) with 1-2 molecules of biotin per 7 protein molecule. Soluble recombinant TMPRSS2 (aa 106-492) with an N-terminal 6xHis tag 8 expressed in yeast was purchased from Creative BioMart (Shirley, NY) and shown to be 9 inhibited by WT PAI-1 (SI Fig. 1). FXIIa-biotin (100 nM) or TMPRSS2 (1 uM (Shrimp, et al. 10 2020)) were incubated with a 1:10 dilution of the input I91L PAI-1 phage display library for 30 11 min at 37oC in TBS (50 mM Tris base, 150 mM NaCl, pH 7.4) containing 5% BSA (TBS-BSA). 12 All reactions were quenched with the addition of cOmplete EDTA-free protease inhibitor cocktail 13 (MilliporeSigma) for 10 min at 37oC. The biotinylated and 6xHis-tagged screens were 14 immunoprecipitated overnight with streptavidin or Ni-NTA magnetic beads (New England 15 Biolabs, Ipswich, MA), respectively. phPAI-1::enzyme complexes were further washed, eluted, 16 and quantified as described previously (Huttinger, et al. 2021; Haynes, et al. 2022). Replicas 17 represent the selection of independent cultures of the input phage display PAI1 variant library. 18 19 HTS. Phage displayed PAI-1 cDNA were sequenced in twelve 150 bp amplicons using primers 20 listed in SI Table 1. Amplicons were prepared for HTS as previously described (Huttinger, et al. 21 2021; Haynes, et al. 2022) and sequenced with 8x105-2x106 150 bp paired-end reads per 22 amplicon. HTS data were analyzed using the DESeq2 software package (Love, et al. 2014; Zhu, 23 et al. 2019). Significance thresholds for DESeq2 were set to padj < 0.1 and BaseMean score ≥ 24 50. MA plots showing the distribution of amino acid substitutions in PAI-1 with respect to these 25 thresholds are shown in SI Figs. 2-4. Notably, when amber stop codons are read through as 26 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 17 Gln, those variants are enriched in the selected libraries relative to other stop codons introduced 1 by error-prone PCR. 2 3 Evolutionary variability of PAI-1. As previously described, evolutionary conservation scores at 4 each amino acid position in PAI-1 were calculated using ConSurf (https://consurf.tau.ac.il/), 5 where higher ConSurf scores indicate more evolutionarily variable positions (Ashkenazy, et al. 6 2016; Huttinger, et al. 2021). To relate the results of our DMS screen to evolutionary 7 conservation scores, we define the normalized functional score as the sum of the mean log2-fold 8 determined at each position. Conservation scores were then analyzed as a function of the 9 normalized functional score at each position using the R software package (version 4.3.3). 10 11 Expression and characterization of recombinant PAI-1 variants. PAI-1 variant cDNA (I91L, 12 I91L/F98Y, I91L/T251I, and I91L/S331C PAI-1) in the pET-24(+) expression plasmid with a C-13 terminal Gly-Ser-Gly hinge and 6XHis-tag was purchased from Twist Bioscience (South San 14 Francisco, CA) and transformed into NiCo21(DE3) chemically competent E. coli (New England 15 Biolabs, Ipswitch, MA). PAI-1 variants were expressed as previously described (Haynes, et al. 16 2022), and concentration was determined by absorption at 280 nm (1%, 280 nm = 7.0). SI for each 17 variant was determined by incubating uPA (2.5 nM) or factor XIIa (100 nM) with PAI-1 variant 18 concentrations ranging from 0-4 nM and 0-10 nM respectively. The SIs for each variant were 19 normalized to an SI of uPA inhibition defined as 1. 20 21 PAI-1 inhibition of TMPRSS2. WT PAI-1 was expressed and purified as described above. 22 TMPRSS2 (1 M total protein, yet only fractionally active (Shrimp, et al. 2020)) was incubated 23 with WT PAI-1 (1 nM effective active concentration) or vehicle control for 30 min at room 24 temperature (~25oC). TMPRSS2 activity was determined by monitoring its ability to cleave 25 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 18 peptidyl substrate boc-QAR-AMC (ex = 370 nm, em = 440 nm; VWR International) for 10 min 1 (Shrimp, et al. 2020). Following this monitoring, samples were run on a 4-20% Tris-glycine gel 2 (Invitrogen) under non-reducing conditions, transferred to nitrocellulose, and blotted using a 3 rabbit-anti-PAI-1 primary antibody (Abcam). 4 5 Cophylogenetic analyses. A list of human secreted SERPINs (n = 13) and serine proteases (n 6 = 16) was curated from the literature (Olson and Gettins 2011; Heiker, et al. 2013; Godinez, et 7 al. 2022; Janciauskiene, et al. 2024). We defined secreted SERPINs and serine proteases as 8 those containing a signal peptide. We used the COBALT tool to produce a domain-centric 9 alignment of both the SERPINs and serine proteases in the lists, and then pruned the alignment 10 manually to include only sites with fewer than 90% gaps. We used TreeBeST (Vilella, et al. 11 2009) to produce phylogenetic trees for both SERPINs and serine proteases, using 1000 12 bootstrap replicates, and recovered the maximum likelihood tree for analyses. A cophylogenetic 13 analysis was then performed using the Parafit package in R with the null hypothesis that the 14 evolution of serine proteases and SERPINS were independent of each other (Legendre, et al. 15 2002). The p-value was derived by 1000 random permutations of the tip states of the two trees. 16 17 Funding 18 This work was supported by the National Institutes of Health grant R35--HL171421 (to D.G.) 19 20 Data Availability 21 All data is contained within the manuscript and supplementary files. Raw sequencing data and 22 bioinformatics results will be made available upon request. 23 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 19

1 Al-Samkari H, Karp Leaf RS, Dzik WH, Carlson JCT, Fogerty AE, Waheed A, Goodarzi K, 2 Bendapudi PK, Bornikova L, Gupta S, et al. 2020. COVID-19 and coagulation: bleeding 3 and thrombotic manifestations of SARS-CoV-2 infection. Blood 136:489-500. 4 Ashkenazy H, Abadi S, Martz E, Chay O, Mayrose I, Pupko T, Ben-Tal N. 2016. ConSurf 2016: 5 an improved methodology to estimate and visualize evolutionary conservation in 6 macromolecules. Nucleic Acids Res 44:W344-350. 7 Berkenpas MB, Lawrence DA, Ginsburg D. 1995. Molecular evolution of plasminogen activator 8 inhibitor-1 functional stability. EMBO J 14:2969-2977. 9 Berrettini M, Schleef RR, España F, Loskutoff DJ, Griffin JH. 1989. Interaction of type 1 10 plasminogen activator inhibitor with the enzymes of the contact activation system. 11 Journal of Biological Chemistry 264:11738-11743. 12 Bhakta V, Hamada M, Nouanesengsy A, Lapierre J, Perruzza DL, Sheffield WP. 2021. 13 Identification of an alpha-1 antitrypsin variant with enhanced specificity for factor XIa by 14 phage display, bacterial expression, and combinatorial mutagenesis. Sci Rep 11:5565. 15 Bouwman JJ, Diepersloot RJ, Visseren FL. 2009. Intracellular infections enhance interleukin-6 16 and plasminogen activator inhibitor 1 production by cocultivated human adipocytes and 17 THP-1 monocytes. Clin Vaccine Immunol 16:1222-1227. 18 Conant GC, Wolfe KH. 2008. Turning a hobby into a job: how duplicated genes find new 19 functions. Nat Rev Genet 9:938-950. 20 Ding D, Green AG, Wang B, Lite TV, Weinstein EN, Marks DS, Laub MT. 2022. Co-evolution of 21 interacting proteins through non-contacting and non-specific mutations. Nat Ecol Evol 22 6:590-603. 23 Dittmann M, Hoffmann HH, Scull MA, Gilmore RH, Bell KL, Ciancanelli M, Wilson SJ, Crotta S, 24 Yu Y, Flatley B, et al. 2015. A serpin shapes the extracellular environment to prevent 25 influenza A virus maturation. Cell 160:631-643. 26 Gettins PG. 2002. Serpin structure, mechanism, and function. Chem Rev 102:4751-4804. 27 Gettins PG, Olson ST. 2009. Exosite determinants of serpin specificity. J Biol Chem 284:20441-28 20445. 29 Ghose DA, Przydzial KE, Mahoney EM, Keating AE, Laub MT. 2023. Marginal specificity in 30 protein interactions constrains evolution of a paralogous family. Proc Natl Acad Sci U S 31 A 120:e2221163120. 32 Godinez A, Rajput R, Chitranshi N, Gupta V, Basavarajappa D, Sharma S, You Y, Pushpitha K, 33 Dhiman K, Mirzaei M, et al. 2022. Neuroserpin, a crucial regulator for axogenesis, 34 synaptic modelling and cell-cell interactions in the pathophysiology of neurological 35 disease. Cell Mol Life Sci 79:172. 36 Hahn MW. 2009. Distinguishing among evolutionary models for the maintenance of gene 37 duplicates. J Hered 100:605-617. 38 Haynes LM, Huttinger ZM, Yee A, Kretz CA, Siemieniak DR, Lawrence DA, Ginsburg D. 2022. 39 Deep mutational scanning and massively parallel kinetics of plasminogen activator 40 inhibitor-1 functional stability to probe its latency transition. J Biol Chem 298:102608. 41 Heiker JT, Klöting N, Kovacs P, Kuettner EB, Sträter N, Schultz S, Kern M, Stumvoll M, Blüher 42 M, Beck-Sickinger AG. 2013. Vaspin inhibits kallikrein 7 by serpin mechanism. Cell Mol 43 Life Sci 70:2569-2583. 44 Huntington JA. 2011. Serpin structure, function and dysfunction. J. Thromb. Haemost. 9 Suppl 45 1:26-34. 46 Huttinger ZM, Haynes LM, Yee A, Kretz CA, Holding ML, Siemieniak DR, Lawrence DA, 47 Ginsburg D. 2021. Deep mutational scanning of the plasminogen activator inhibitor-1 48 functional landscape. Sci Rep 11:18827. 49 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 20 Janciauskiene S, Lechowicz U, Pelc M, Olejnicka B, Chorostowska-Wynimko J. 2024. 1 Diagnostic and therapeutic value of human serpin family proteins. Biomed Pharmacother 2 175:116618. 3 Jendroszek A, Madsen JB, Chana-Munoz A, Dupont DM, Christensen A, Panitz F, Fuchtbauer 4 EM, Lovell SC, Jensen JK. 2019. Biochemical and structural analyses suggest that 5 plasminogen activators coevolved with their cognate protein substrates and inhibitors. J 6 Biol Chem 294:3794-3805. 7 Keijer J, Linders M, Wegman JJ, Ehrlich HJ, Mertens K, Pannekoek H. 1991. On the Target 8 Specificity of Plasminogen Activator Inhibitor 1: The Role of Heparin, Vitronectin, and the 9 Reactive Site. Blood 78:1254-1261. 10 Keller TT, Sluijs KFvd, Kruif MDd, Gerdes VEA, Meijers JCM, Florquin S, Poll Tvd, Gorp ECMv, 11 Brandjes DPM, Büller HR, et al. 2006. Effects on Coagulation and Fibrinolysis Induced 12 by Influenza in Mice With a Reduced Capacity to Generate Activated Protein C and a 13 Deficiency in Plasminogen Activator Inhibitor Type 1. Circulation Research 99:1261-14 1269. 15 Law RH, Zhang Q, McGowan S, Buckle AM, Silverman GA, Wong W, Rosado CJ, Langendorf 16 CG, Pike RN, Bird PI, et al. 2006. An overview of the serpin superfamily. Genome Biol 17 7:216. 18 Lawrence DA, Ginsburg D, Day DE, Berkenpas MB, Verhamme IM, Kvassman JO, Shore JD. 19 1995. Serpin-protease complexes are trapped as stable acyl-enzyme intermediates. J. 20 Biol. Chem. 270:25309-25312. 21 Lawrence DA, Olson ST, Muhammad S, Day DE, Kvassman JO, Ginsburg D, Shore JD. 2000. 22 Partitioning of serpin-proteinase reactions between stable inhibition and substrate 23 cleavage is regulated by the rate of serpin reactive center loop insertion into beta-sheet 24 A. J. Biol. Chem. 275:5839-5844. 25 Lawrence DA, Strandberg L, Ericson J, Ny T. 1990. Structure-function studies of the SERPIN 26 plasminogen activator inhibitor type 1. Analysis of chimeric strained loop mutants. J. 27 Biol. Chem. 265:20293-20301. 28 Legendre P, Desdevises Y, Bazin E. 2002. A statistical test for host-parasite coevolution. Syst 29 Biol 51:217-234. 30 Lewis JH, Iammarino RM, Spero JA, Hasiba U. 1978. Antithrombin Pittsburgh: an alpha1-31 antitrypsin variant causing hemorrhagic disease. Blood 51:129-137. 32 Love MI, Huber W, Anders S. 2014. Moderated estimation of fold change and dispersion for 33 RNA-seq data with DESeq2. Genome Biol 15:550. 34 Marijanovic EM, Fodor J, Riley BT, Porebski BT, Costa MGS, Kass I, Hoke DE, McGowan S, 35 Buckle AM. 2019. Reactive centre loop dynamics and serpin specificity. Sci Rep 9:3870. 36 McClune CJ, Alvarez-Buylla A, Voigt CA, Laub MT. 2019. Engineering orthogonal signalling 37 pathways reveals the sparse occupancy of sequence space. Nature 574:702-706. 38 McClune CJ, Laub MT. 2020. Constraints on the expansion of paralogous protein families. Curr 39 Biol 30:R460-r464. 40 Nocedal I, Laub MT. 2022. Ancestral reconstruction of duplicated signaling proteins reveals the 41 evolution of signaling specificity. eLife 11. 42 Olson ST, Gettins PG. 2011. Regulation of proteases by protein inhibitors of the serpin 43 superfamily. Prog Mol Biol Transl Sci 99:185-240. 44 Owen MC, Brennan SO, Lewis JH, Carrell RW. 1983. Mutation of antitrypsin to antithrombin. 45 alpha 1-antitrypsin Pittsburgh (358 Met leads to Arg), a fatal bleeding disorder. N. Engl. 46 J. Med. 309:694-698. 47 Park Y, Metzger BPH, Thornton JW. 2022. Epistatic drift causes gradual decay of predictability 48 in protein evolution. Science 376:823-830. 49 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 21 Polderdijk SG, Adams TE, Ivanciu L, Camire RM, Baglin TP, Huntington JA. 2017. Design and 1 characterization of an APC-specific serpin for the treatment of hemophilia. Blood 2 129:105-113. 3 Ponczek MB, Gailani D, Doolittle RF. 2008. Evolution of the contact phase of vertebrate blood 4 coagulation. J Thromb Haemost 6:1876-1883. 5 Ponczek MB, Shamanaev A, LaPlace A, Dickeson SK, Srivastava P, Sun MF, Gruber A, 6 Kastrup C, Emsley J, Gailani D. 2020. The evolution of factor XI and the kallikrein-kinin 7 system. Blood Adv 4:6135-6147. 8 Puy C, Ngo ATP, Pang J, Keshari RS, Hagen MW, Hinds MT, Gailani D, Gruber A, Lupu F, 9 McCarty OJT. 2019. Endothelial PAI-1 (Plasminogen Activator Inhibitor-1) Blocks the 10 Intrinsic Pathway of Coagulation, Inducing the Clearance and Degradation of FXIa 11 (Activated Factor XI). Arterioscler Thromb Vasc Biol 39:1390-1401. 12 Rezaie AR. 2001. Vitronectin Functions as a Cofactor for Rapid Inhibition of Activated Protein C 13 by Plasminogen Activator Inhibitor-1: Implications for the mechanism of profibrinolytic 14 action of activated protein C. J Biol Chem 276:15567-15570. 15 Sanrattana W, Maas C, de Maat S. 2019. SERPINs—From Trap to Treatment. Front Med 16 (Lausanne) 6:25. 17 Sanrattana W, Sefiane T, Smits S, van Kleef ND, Fens MH, Lenting PJ, Maas C, de Maat S. 18 2021. A reactive center loop-based prediction platform to enhance the design of 19 therapeutic SERPINs. Proc Natl Acad Sci U S A 118. 20 Scott BM, Matochko WL, Gierczak RF, Bhakta V, Derda R, Sheffield WP. 2014. Phage display 21 of the serpin alpha-1 proteinase inhibitor randomized at consecutive residues in the 22 reactive centre loop and biopanned with or without thrombin. PLoS One 9:e84491. 23 Sen P, Komissarov AA, Florova G, Idell S, Pendurthi UR, Vijaya Mohan Rao L. 2011. 24 Plasminogen activator inhibitor-1 inhibits factor VIIa bound to tissue factor. J Thromb 25 Haemost 9:531-539. 26 Shen LW, Mao HJ, Wu YL, Tanaka Y, Zhang W. 2017. TMPRSS2: A potential target for 27 treatment of influenza virus and coronavirus infections. Biochimie 142:1-10. 28 Sherman PM, Lawrence DA, Yang AY, Vandenberg ET, Paielli D, Olson ST, Shore JD, 29 Ginsburg D. 1992. Saturation mutagenesis of the plasminogen activator inhibitor-1 30 reactive center. J. Biol. Chem. 267:7588-7595. 31 Shrimp JH, Kales SC, Sanderson PE, Simeonov A, Shen M, Hall MD. 2020. An Enzymatic 32 TMPRSS2 Assay for Assessment of Clinical Candidates and Discovery of Inhibitors as 33 Potential Treatment of COVID-19. ACS Pharmacol Transl Sci 3:997-1007. 34 Singh S, O’Reilly S, Gewaid H, Bowie AG, Gautier V, Worrall DM. 2022. Reactive Centre Loop 35 Mutagenesis of SerpinB3 to Target TMPRSS2 and Furin: Inhibition of SARS-CoV-2 Cell 36 Entry and Replication. International Journal of Molecular Sciences 23:12522. 37 Spence MA, Mortimer MD, Buckle AM, Minh BQ, Jackson CJ. 2021. A Comprehensive 38 Phylogenetic Analysis of the Serpin Superfamily. Mol Biol Evol 38:2915-2929. 39 Stefansson S, Yepes M, Gorlatova N, Day DE, Moore EG, Zabaleta A, McMahon GA, Lawrence 40 DA. 2004. Mutants of plasminogen activator inhibitor-1 designed to inhibit neutrophil 41 elastase and cathepsin G are more effective in vivo than their endogenous inhibitors. J 42 Biol Chem 279:29981-29987. 43 Tanaka A, Suzuki Y, Sugihara K, Kanayama N, Urano T. 2009. Inactivation of plasminogen 44 activator inhibitor type 1 by activated factor XII plays a role in the enhancement of 45 fibrinolysis by contact factors in-vitro. Life Sci 85:220-225. 46 Vilella AJ, Severin J, Ureta-Vidal A, Heng L, Durbin R, Birney E. 2009. EnsemblCompara 47 GeneTrees: Complete, duplication-aware phylogenetic trees in vertebrates. Genome 48 Res 19:327-335. 49 Yousef GM, Kopolovic AD, Elliott MB, Diamandis EP. 2003. Genomic overview of serine 50 proteases. Biochemical and Biophysical Research Communications 305:28-36. 51 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 22 Zhu A, Ibrahim JG, Love MI. 2019. Heavy-tailed prior distributions for sequence count data: 1 removing the noise and preserving large differences. Bioinformatics 35:2084-2092. 2 Zuo Y, Warnock M, Harbaugh A, Yalavarthi S, Gockman K, Zuo M, Madison JA, Knight JS, 3 Kanthi Y, Lawrence DA. 2021. Plasma tissue plasminogen activator and plasminogen 4 activator inhibitor-1 in hospitalized COVID-19 patients. Sci Rep 11:1580. 5 6 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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 23 1 2 Figure 1. SERPINs inhibit serine proteases via a “mousetrap” mechanism to inhibit their 3 primary and non-canonical protease targets. In its metastable active state, the SERPIN’s 4 reactive center loop (RCL, yellow) extends from its central globular core and contains a 5 preferred cleavage site for proteases targeted for inhibition (green). Under conditions favoring 6 the inhibitory pathway (left pathway), after the protease cleaves the scissile bond in the RCL, 7 the acyl intermediate is trapped when the RCL inserts into -sheet A forming a fifth -strand. 8 Alternatively, the protease cleaves the RCL and the acyl intermediate is hydrolyzed resulting in 9 the insertion of the RCL into -sheet A and a non-functional SERPIN, while the protease 10 remains active (center pathway). Unique among SERPINs, PAI-1 can also undergo a 11 spontaneous latency transition in which the RCL inserts into -sheet in the absence of a 12 proteolytic event, rendering the SERPIN inactive (right pathway). 13 14 15 16 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 24 1 2 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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 25 Figure 2. PAI-1 DMS and differential patterns of serine protease inhibition. (A) PAI-1 1 specificity fingerprints are shown for uPA, FXIIa, and TMPRSS2. The PAI-1 amino acid 2 sequence is shown along the x-axis with potential amino acid substitutions shown along the y-3 axis. Tolerated mutations are shown in blue (log2-fold enrichment score > 0), while loss-of-4 function mutations are shown in red (log2-fold enrichment score ≤ 0). The amino acids of the 5 canonical human PAI-1 sequence are shown in dark grey. Amino acids that were either not 6 present or not present at sufficient levels to quantify (BaseMean ≥ 50 as determined by DESeq2 7 (Love, et al. 2014)) in the phage library are shown in white. (B) Heatmap showing the common 8 amino acid substitutions across PAI-1’s specificity spaces for uPA, FXIIa, and TMPRSS2 with 9 positive (blue) and negative (red) functional enrichment scores. (C) PCA plot comparing the 10 specificity spaces of PAI-1 for uPA (pink), FXIIa (yellow), TMPRSS2 (blue), and the starting I91L 11 PAI-1 DMS library (black). Groups are annotated with the minimum enclosing ellipse. (D and E) 12 Venn diagrams depicting the number of significantly (D) enriched and (E) depleted amino acid 13 substitutions shared and unique to each specificity fingerprint. 14 15 16 17 18 19 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 26 1 2 Figure 3. Determinants of PAI-1 specificity are found throughout its sequence space. (A) 3 A cartoon of the PAI-1 primary amino acid sequence showing which amino acids were included 4 in each amplicon. Amplicon 12 contains PAI-1’s RCL. (B) PCA plots comparing the specificity 5 spaces of PAI-1 for uPA (pink), FXIIa (yellow), TMPRSS2 (blue), and the I91L PAI-1 variant 6 library (black) across PAI-1’s twelve amplicons. Groups are annotated with the minimum 7 enclosing ellipse. 8 9 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 27 1 2 Figure 4. PAI-1 is under purifying selection to inhibit uPA and FXIIa but not TMPRSS2. 3 Conservation scores (Ashkenazy, et al. 2016) for each amino acid position in PAI-1 (grey dots) 4 are shown as a function of the normalized functional scores determined in our DMS screens of 5 PAI-1 inhibition of (A) uPA (R2 = 0.14, p = 4.6x10-14), (B) FXIIa (R2 = 0.06, p = 7.5x10-7), and (C) 6 TMPRSS2 (R2 = -0.003, p = 0.87). Dashed lines indicated the best fit linear regression with the 7 95% confidence interval shown in grey shading. 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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 28 1 2 Figure 5. PAI-1 variants that are most enriched for FXIIa inhibition do not exhibit 3 improved SIs. Volcano plots showing the I91L PAI-1 variants activity toward (A) FXIIa and (B) 4 uPA. Dark blue circles indicate statistically significant enrichment scores (p < 0.005). The three 5 most enriched PAI-1 variants (I91L/F98Y, I91L/T251I, and I91L/S331C) are shown in pink. The 6 Log2-fold change is plotted on the x-axis and the statistical significance (-Log10(padj)) is plotted 7 on the y-axis. (C) SIs were determined for each of the variants for both uPA (green) and FXIIa 8 (pink) inhibition. For each variant, the SI for uPA was set to one to account for inactive PAI-1 9 present in the protein preparation. t-tests were used to determine statistical significance (*, p 10 <0.05; **, p<0.01; ***, p < 0.001). (D) FXIIa normalized functional mutation scores are compared 11 to the uPA normalized functional mutation scores for each variant. The conservation score is 12 indicated by the color of the points. Residues 98, 251, and 331 are labelled and outlined in 13 black. 14 15 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 29 1 2 Figure 6. Cophylogenetic analysis reveals that human secreted SERPINs coevolved with 3 secreted serine proteases. The phylogenic tree of human secreted SERPINs (left) is mutually 4 informative with that of human secreted serine proteases (right) as determined with the parafit R 5 package (p = 0.039) (Legendre, et al. 2002). SERPINs and proteases are denoted by their 6 human gene names. For SERPINs, colors denote closely related proteins (SERPINC clade, red; 7 SERPIN E and I clades, blue; SERPIN A clade, orange; SERPIN G and F clades, green; 8 SERPIN D clade, purple), while for serine proteases, colors denote the group of SERPINs most 9 commonly inhibiting a given protease. Proteases equally likely to be inhibited by SERPINs from 10 more than one group are shown in black. 11 12 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint 30 Table 1. PAI-1 variants* that lead to maintenance/decrease of inhibitory capacity 1 Target Protease Number of variants that maintain/ inhibitory capacity Number of variants resulting in loss of inhibitory capacity uPA 541 828 FXIIa 349 623 TMPRSS2 513 191 *Variants identified that pass significance thresholds (padj < 0.1, Base Mean score ≥ 50). 2 (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 September 20, 2024. ; https://doi.org/10.1101/2024.09.16.612699doi: bioRxiv preprint