Mapping the Mutational Landscape for Streptokinase Binding to Plasminogen

streptokinase, Group A Streptococcus, host-pathogen interaction, deep mutational 17 scanning 18 19 20 21 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 2

22 23 Group A Streptococcus (GAS) expresses streptokinase (SK), a critical virulence factor that non-24 enzymatically activates the host’s plasminogen (PLG), to an active form, PLGSK, resulting in the 25 degradation of fibrin clots and subsequent bacterial dissemination. PLGSK formation contrasts 26 with the physiologic activation of PLG to the serine protease plasmin via a proteolytic 27 mechanism. As a potent thrombolytic, SK has been used as a therapeutic to treat heart attacks 28 and strokes. GAS SK is highly specific for human PLG, and sequence variation between SK 29 from different GAS strains has been linked to differences in PLG binding and disease severity. 30 We now report the application of deep mutational scanning (DMS) to map the effects of ~71% of 31 single amino acid substitutions within SK from Group C Streptococcus, which shares high 32 sequence homology with GAS on its ability to bind human PLG. We first demonstrate that SK 33 expressed as a fusion protein to the p3 coat protein of M13 filamentous phage retains its 34 capacity to bind human PLG. Our subsequent DMS analysis using this phage system identifies 35 regions of SK in which amino acid substitutions are likely to increase or decrease its affinity for 36 PLG. Our findings suggest a complex protein-protein interaction in which long-range protein 37 dynamics influence the conformational activation of PLG to PLGSK. These data lay the 38 foundation for linking SK variation between GAS strains to differences in virulence, mapping the 39 determinants of GAS SK’s human specificity, and potentially contributing to the development of 40 improved therapeutics for heart attack and stroke. 41 42 43 44 45 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 3

46 47 Streptokinase (SK) is one of several critical bacterial virulence factor proteins that promote the 48 pathogenesis of Streptococcus species, including Group A (GAS) and Group C (GCS) 49 Streptococcus, by hijacking fibrinolysis (the enzymatic breakdown of fibrin blood clots) and 50 ultimately thwarting the body’s defense mechanism of immobilizing pathogens within thrombi (1-51 3). GAS is highly specific for humans and causes infections ranging from mild pharyngitis to 52 severe infections, including necrotizing fasciitis (4-6). GCS is an emerging pathogen that can 53 occasionally cross zoological barriers leading to potential human infections (7, 8). SK, which is 54 produced by many pathogenic strains of GAS and GCS, non-proteolytically activates the 55 zymogen plasminogen (PLG) to the potent fibrinolytic enzyme PLGSK. Although the general 56 mechanism by which SK activates PLG to PLGSK has been addressed by over three decades of 57 detailed biochemical studies, new tools now enable high-throughput analysis of the biochemical 58 functions for individual amino acids within SK. These approaches should also provide an 59 improved understanding of sequence diversity among SK from different GAS and GCS isolates 60 (9). 61 62 Under physiologic conditions within the vascular system, PLG is proteolytically activated to the 63 serine protease plasmin by either tissue- or urokinase-type plasminogen activator (tPA or uPA, 64 respectively) via cleavage of the Arg561-Val562 peptide bond (10). In contrast, SK non-65 proteolytically activates PLG to PLGSK by first binding to PLG and then inducing a 66 conformational change by inserting SK Ile1 into the activation pocket within PLG (11). PLGSK 67 also has a unique activity profile relative to plasmin, including an enhanced substrate specificity 68 for PLG to generate plasmin (12). Furthermore, SK has a higher affinity for plasmin than for 69 PLG, with SK from some strains of GAS protecting plasmin from inhibition by its primary 70 inhibitor a2-antiplasmin(9). 71 72 SK from the GCS, Streptococcus dysgalactiae subsp. equisimilis (SDSE) H46A strain (SKH46A) 73 shares high sequence homology with SK from select GAS strains (~90%) and is one of the best 74 characterized SK variants due to its use as a clinical thrombolytic agent to resolve thrombi 75 associated with acute strokes and myocardial infarctions (11, 13-15). While tPA is currently the 76 preferred thrombolytic treatment in high-income countries, SKH46A is still a first-line therapeutic in 77 middle- and lower-income countries due to its lower production costs (16), despite its reduced 78 efficacy compared to tPA (17). 79 80 The SK gene consists of 1398 nucleotides, corresponding to 414 amino acids. The SK protein is 81 composed of three distinct structural domains that each appear to be required for PLG 82 activation (18, 19), with each containing unique functionalities (18, 20-22). The α domain 83 (residues 1-150) is the major domain facilitating substrate recognition and the conversion of 84 PLG to an active conformation; the β domain (residues 151-287) enhances binding affinity and 85 substrate processing during binding; and lastly, the γ domain (residues 288-414) stabilizes the 86 SK-PLG complex and enhances proteolytic activity (23, 24). While the general functions of each 87 of these domains have been identified, we do not yet know the role of individual amino acids in 88 supporting these functions and ultimately the contribution of each amino acid residue with 89 respect to SK’s affinity for and activation of PLG. 90 91 We now report deep mutational scanning (DMS) using a high-throughput phage display system 92 (25-28) to simultaneously analyze the role of approximately 70% of all possible single amino 93 acid substitutions in SK with respect to its ability to bind human PLG. Our results provide high-94 resolution insight into which regions/domains of SK are tolerant of amino acid substitutions with 95 respect to PLG binding, building a platform for the characterization of SK variants from 96 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 4 emerging strains of GAS and GCS, as well as the development of potential novel SK 97 therapeutics. 98 99

and Discussion 100 101 Phage displayed SKH46A selectively binds, but does not activate human PLG 102 To demonstrate that SKH46A is capable of specifically binding to PLG when displayed on M13 103 filamentous phage, we cloned wild-type SKH46A with an N-terminal His6-tag and TEV protease 104 (TEVp) cleavage site (Figure 1A) into the modified pAY-FE plasmid vector (Genebank 105 #MW464120) (27, 29). We then expressed this phage displayed construct along with a control 106 phage expressing the A3 domain of von Willebrand factor (VWF-A3), as previously described 107 (25, 26). A mixture of SKH46A and excess VWF-A3 phage selected for binding to biotinylated Glu-108 PLG; (Figure 1B), demonstrated strong enrichment for SKH46A. – with ~3% of the unselected 109 phage identified by sequencing as SKH46A, compared to ~90% of the selected phage. These 110

confirm that phage displayed SKH46A can selectively bind to biotinylated PLG. 111 112 The N-terminal Ile residue (Ile1) in SK is known to be critical for its activation of PLG to PLGSK 113 (21); however, our construct design masks this function of Ile1 by the presence of the N-terminal 114 His6 tag followed by a TEVp cleavage site (see Materials and Methods). Previous studies 115 suggest that SK with a blocked or absent Ile1 may still exhibit residual PLG activation 116 activity((30)). To detect any potential residual SK activity for our phage displayed fusion protein, 117 we measured PLG activation in an amidolytic assay (Supplementary Figure 1) with the 118 fluorogenic substrate (z-Gly-Gly-Arg-7-amido-4-methylcoumarin (AMC)). No residual PLGSK 119 activity was detected under the tested experimental conditions. 120 121 Characterization of the SKH46A variant library 122 Our error-prone PCR-generated SKH46A variant library exhibited a depth of ~107 independent 123 clones, with DNA sequence analysis of 24 randomly selected phage demonstrating an average 124 of ~2 amino acid substitutions per clone as determined by Sanger sequencing (Supplementary 125 Figure 2). High-throughput sequencing (HTS) of the variant library revealed that it contained 126 5,603 of 7,866 possible single amino acid substitutions (71%). As expected, the coverage 127 frequency for individual amino acid substitutions correlates with their number of codons in the 128 genetic code, with significant underrepresentation of amino acid residues encoded by a single 129 codon (Supplementary Figure 3). 130 131 DMS of SKH46A binding to PLG 132 We next assessed the ability of variants in the SKH46A library to bind PLG, using the approach 133 depicted schematically in Figure 2A. Briefly, the SKH46A variant library was incubated with 100 134 nM biotinylated PLG, and phage displayed SKH46A variants in complex with biotinylated PLG 135 were selected with streptavidin resin. The segment of selected phage DNA encoding SKH46A was 136 amplified and sequenced in twelve 150 bp amplicons (Figure 3A) (27). Principal component 137 analysis (PCA) demonstrates reproducibility in our selection assays (Figure 2B) with clear 138 clustering of replicates for the input phage library, biotinylated PLG selected phage, and non-139 specific binding conditions (no PLG control). To identify regions of SKH46A in which amino acid 140 substitutions exhibited the largest effects on PLG binding, we also performed individual PCAs 141 for each amplicon (Figure 3B) (27). The non-specific binding control (without added PLG, 142 designated as SK-PLG) was consistently separated from the reaction mixture (SK+PLG), and 143 Input (-PLG) along the PC1 in each PCA. However, the explanatory power of PC1 varied 144 considerably among the twelve amplicons—most markedly (>80% variation explained by PC1) 145 for amplicons 1 (residues 1-23), 2 (residues 24-56), 4 (residues 94-130), 5 (residues 131-164), 146 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 5 and 8 (residues 234-268), suggesting that amino acid substitutions in these regions are more 147 frequent effectors of PLG binding. 148 149 Of the amino acid substitutions present in the variant library, 1,670 (~30%) exhibited significant 150 enrichment (binding score > 0; n = 793; base mean ≥ 100, FDR-adjusted p < 0.1) or depletion 151 (binding score < 0; n = 877; base mean ≥ 100, FDR-adjusted P < 0.1) as shown in Figure 4A 152 (and Supplementary Figure 4). We interpret amino acid substitutions in SKH46A with binding 153 scores greater than zero as retaining or increasing affinity for biotinylated PLG relative to wild 154 type (WT) SKH46A. Those with binding scores less than zero, we interpret as having decreased 155 affinity for biotinylated PLG. Consistent with their propensity to disrupt secondary structures, Pro 156 (74%) and Gly (58%) substitutions were particularly likely to be associated with negative binding 157 scores. 158 159 To further characterize segments of SKH46A in which amino acid substitutions are particularly 160 likely or unlikely to disrupt SK binding to PLG, we calculated a mutational acceptance score (27, 161 31). We then fit the positional binding score to a LOWESS regression to determine regions in 162 the SKH46A primary sequence in which amino acid substitutions were likely to maintain, increase, 163 or decrease the affinity of SKH46A for biotinylated PLG (Figure 4 and Supplementary Figure 5). 164 This analysis identifies five regions in SKH46A (75th percentile or greater) in which amino acid 165 substitutions are particularly likely to maintain or increase SKH46A’s affinity for PLG (Figure 4A; 166 residues 55-85, 171-186, 288-301, 304-313, and 337-374) and four regions (25th percentile or 167 less) in which amino acid substitutions decrease SKH46A’s affinity for PLG (Figure 4A; residues 168 9-39, 99-148, 261-267, and 389-407). These regions correspond to segments of SK in which 169 amino acid changes are known to result in a decreased ability for SK to bind and activate PLG 170 to PLGSK, including the N-terminal alpha-domain (12, 32, 33), the 250-loop (20, 34, 35), and the 171 C-terminal Lys414 (36). 172 173 One limitation of our current assay is that it cannot distinguish between decreased affinity due to 174 protein misfolding or to more subtle structural changes that decrease the binding affinity of a 175 well-folded protein. To distinguish between these two possibilities, we calculated DDG values for 176 amino acid substitution in SKH46A using two computational protein stability predictors, EvoEF2 177 (37) and FoldX (38) whichare moderately correlated with each other (R2 = 0.2930, p < <0.0001; 178 Supplementary Figure 6). Amino acid substitutions resulting in decreased thermodynamic 179 stability and protein misfolding should exhibit more positive DDG values, while those that 180 increase thermodynamic stability should exhibit more negative DDG values, suggesting the 181 presence of properly folded proteins, as most proteins canonically fold to their lowest energy 182 state. As anticipated, both EvoEF and FoldX (Figure 5), predict that SKH46A variants that bind 183 PLG (binding score > 0) have more negative DDG values than those that exhibit a reduction in 184 binding affinity (binding score < 0). Of note, there is a population of SKH46A variants with binding 185 scores less than zero but with DDG values similar to those variants with binding scores greater 186 than zero, potentially representing a subpopulation of variants that are properly folded yet still 187 exhibit loss of binding. 188 189 Characterization of amino acid substitutions that maintain or increase SK-affinity for PLG 190 Our DMS analysis also identified amino acid substitutions in SKH46A that appear to facilitate SK 191 binding to PLG (Figure 4). The amino acid substitutions with the highest binding scores include 192 M369F (11.1), K76L (8.4), G50L (6.1), and Y252T (5.3). (Figure 4B; Supplementary Figure 5). 193 To validate these findings and explore the effect of these substitutions on SK activation of PLG 194 to PLGSK, we recombinantly expressed each of the above individual variants in an E. coli 195 expression system, with the same N-terminal His6-tag and TEV protease cleavage sequence as 196 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 6 in our phage displayed construct. The SKH46A variants were purified, and the His6-tag removed 197 by treatment with TEV protease (see Materials and Methods). The recombinant SKH46A variants 198 were then used in a modified plasma-based halo/HoFF assay to simultaneously measure PLG 199 binding and activation to PLGSK (39, 40). No significant differences were observed in the 200 maximum amounts of plasmin generated (Figure 6A) nor the maximum rates of fibrinolysis 201 (Figure 6B) between the TEV protease treated SKH46A WT and any of the variants, suggesting 202 that amino acid substitutions in SKH46A that increase its affinity for PLG do not necessarily 203 improve nor impair its ability to induce PLG fibrinolytic activity. Furthermore, none of these 204 variants are located at the predicted SKH46A-PLG protein-protein interface (41), suggesting that 205 these amino acid substitutions may impact binding affinity via long-range interactions. 206 207 To further examine the role that long-range interactions, play in the binding of SKH46A to PLG, we 208 evaluated the propensity of an amino acid position to accept amino acid substitutions as a 209 function of its relative solvent accessibility (RSA; Figure 7). When comparing orthologous 210 sequences among different species, amino acid variation is most frequent at surface exposed 211 residues (42). One might thus anticipate that amino acid positions in SKH46A that are most 212 accepting of substitutions will generally be those that are surface exposed. However, our DMS 213 screen data suggest that single amino acid substitutions that most improve SKH46A’s affinity for 214 PLG are not located at the protein-protein interface. Indeed, both the Met369 and Lys76 variants 215 characterized above are buried residues, with RSA scores of zero. However, we observe a 216 weak yet significant positive correlation between the positional binding scores and RSA (slope = 217 1.1, R2 = 0.1, p = 1.12x10-11), with core residues (RSA < 0.2) generally less accepting of amino 218 acid substitutions, indicating their structural importance in SK, though some core residues in SK 219 are accepting of amino acid substitutions. These deviations from expectations align with recent 220 models suggesting that amino acid substitutions at a protein’s core that do not cause gross 221 protein misfolding can modify a protein’s binding interactions (43, 44). 222 223

224 225 Our study complements a growing number of reports using DMS to characterize mediators of 226 infectious disease (45), while laying the foundation for studying bacterial evolution and 227 pathogenesis. Our results suggest that amino acids beyond the site of direct interaction with 228 PLG affect SK-PLG binding through long-range interactions. Finally, this work establishes a 229 valuable foundation for future research into bacterial pathogenicity, particularly at the host-230 pathogen interface. By mapping the effects of specific mutations within SK on its interaction with 231 human PLG, we provide a resource that will enable deeper investigations into the molecular 232 mechanisms underlying bacterial virulence and adaptation. Furthermore, the insights gained 233 from this study have potential translational applications, as they may guide the development of 234 new and improved thrombolytic agents that leverage engineered SK variants with tailored 235 activity or specificity. 236 237

238 239 SKH46A cloning and variant library generation 240 The coding sequence for SKH46A from SDSE (Genbank ID: CAA51351.1) was synthesized by 241 Twist Bioscience (South San Francisco, CA) and was cloned into a modified pAY-FE plasmid 242 (GenBank ID: MW464120) that carries an ampicillin resistance marker, in which the amber stop 243 codon (TAG) immediately preceding the SK-gIIII fusion protein was mutated to a glutamine 244 residue (CAG, Gln), between the AscI (Product# R0558S, New England Biolabs, Ipswich, 245 MA,USA) and NotI (Product # R3189S, New England Biolabs) restriction sites (Supplementary 246 table 1) using T4 ligase (Product #M1794, Promega, Fitchburg, WI, USA) (26, 27). At the N-247 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 7 terminus of SK, a His6-tag followed by a TEV protease cleavage site (12, 29) was introduced, 248 while the C-terminus harbored a FLAG-tag prior to the p3 coat protein. The final construct 249 (Figure 1A) was transformed into chemically competent XL-1 Blue cells (Agilent Technologies, 250 Santa Clara, CA, USA). 251 252 SK variant library preparation 253 The SK variant library was generated by error-prone PCR using the GeneMorph II Random 254 Mutagenesis Kit (Product# 200550, Agilent Technologies). Primers used for PCR mutagenesis 255 (Supplementary table 1) maintained the AscI and NotI restriction sites for ligation of the 256 restriction-digested insert into the modified pAY-FE vector described above. The PCR products 257 were further amplified with GoTaq (Product# M7123, Promega) using the same primers 258 described above and gel purified using the Qiagen Gel Purification Kit (Product # 28704, 259 Qiagen, Venlo, Netherlands). The resulting library was restriction digested with AscI and NotI 260 and ligated using T4 ligase into the modified pAY-FE plasmid. Following ligation, the library was 261 transformed into electrocompetent XL-1 Blue MRF’ E. coli per manufacturer’s instructions. The 262 depth of the library was determined by titering the number of ampicillin-resistant colonies on LB 263 agar (Product #22700025, Fisher Scientific, Waltham, MA, USA) plates supplemented with 264 ampicillin (100 mg/mL) and 2% glucose. Mutation frequency was assessed by Sanger 265 sequencing of randomly selected individual colonies (n=24), and the range of the number of 266 mutations was plotted. 267 268 Expression of phage displayed SK 269 XL-1 blue cells harboring plasmids for the SK- or VWF-A3-pIII fusion proteins were grown in LB 270 media (Invitrogen) supplemented with 2% glucose (LBG; 100 mL) containing 100 mg/mL 271 ampicillin to mid-log phase at 37 oC and then inoculated with M13K07 Helper Phage (1014 272 CFU/mL; Cytiva, Marlborough, MA, USA) expanded for 1 h at 37 oC, and quantified following 273 manufacturer’s instructions. E. coli were pelleted by centrifugation (4,500xg for 10 min at 4 oC) 274 and resuspended in 2xYT Broth (Novagen, Product # 71755-4, Millipore Sigma, Burlington, MA, 275 USA) supplemented with 100 mg/mL ampicillin, 30 mg/mL kanamycin, and 0.4 mM IPTG, and 276 grown for 2 hours at 37oC. E. coli were pelleted by centrifugation successively at 4,200xg, 277 followed by 4,500xg for 10 min at 4°C. Phage in the supernatant were precipitated with 278 PEG-8000 (2.5% w/v) and NaCl (500 mM) overnight at 4°C, pelleted by centrifugation 279 (20,000xg) and resuspended in 1X TBS (0.05M Tris Base, 0.15M NaCl; pH 7.4). 280 281 Plasminogen biotinylation 282 Human Glu-PLG (HCPG-0130; Prolytix; Essex Junction, VT, USA) was biotinylated using NHS 283 biotin (Product # 21925, ThermoFisher, Waltham, MA, USA) per manufacturer’s directions. Free 284 biotin was removed by dialysis against 1X HBS (20 mM HEPES containing 150 mM NaCl; pH 285 7.4). Protein integrity and biotinylation were confirmed by gel electrophoresis (Invitrogen Wedge 286 Well Novex 4 to 20% TrisGlycine , Product# XP04200BOX, Invitrogen) followed by Coomassie 287 staining. For western blots, proteins were transferred to a nitrocellulose membrane (iBlot™ 2 288 Transfer Stacks, nitrocellulose, mini, Product# IB23002 XP04200BOX, Invitrogen), which was 289 subsequently blocked with 5% bovine serum albumin (BSA) in TBS containing 0.1% Tween-20 290 (TBST). The blot was probed with streptavidin conjugated to HRP (1:1000) (Product#N100, 291 ThermoFisher) prepared in TBST containing 5% BSA, washed with TBST, developed with 292 SuperSignal™ West Femto Maximum Sensitivity Substrate (Product# PI34095, ThermoFisher), 293 and imaged. 294 295 Screening of SK variant affinity for PLG 296 Phage displayed SK was screened for PLG binding using an approach adapted from our 297 previous studies of plasminogen activator inhibitor-1 (25-27, 31, 46). Assay conditions were 298 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 8 optimized by confirming that phage displayed SK could be selectively enriched for their PLG 299 affinity compared to excess negative control- phage displayed VWF-A3 (1SK: 9VWF-A3 300 volume:volume mixture) (25). Once the expression was confirmed, our assay was performed 301 using the SK error-prone PCR library. The reaction was initiated by incubating 50ul of 302 concentrated SK phage (~1013) with biotinylated 100nM Glu-PLG in 1×HBS buffer 303 supplemented with 5% bovine serum albumin (BSA) and incubated at room temperature for 30 304 minutes on a rotator. Subsequently, the mixture was added to magnetic streptavidin beads 305 (Product # S1420S, New England Biolabs) and incubated overnight at 4 °C with continuous 306 rotation. The bead-bound complexes were then washed with 1×HBS buffer containing BSA. 307 Enteropeptidase (EK) (Product #P8070L, New England Biolabs), which cleaves C-terminal to 308 the FLAG tag, was then added in 50 uL of EK buffer (200 mM Tris, 50 mM NaCl, 2 mM CaCl2, 309 pH 8) and incubated overnight at 4 °C. In parallel, XL1 cells were inoculated for subsequent 310 phage titer determination. After incubation, the supernatant was separated from the magnetic 311 beads, and the beads were washed with EK buffer to achieve a final volume of 100 µL. The 312 recovered supernatant containing the phage particles was serially diluted, incubated with XL1 313 cells grown to an OD600 of 0.4–0.6 for 1h at 37 oC, and 100 µL aliquots were plated onto agar 314 plates supplemented with 2% glucose and ampicillin (100 μg/mL). Colony forming units were 315 enumerated to assess phage enrichment relative to control samples. In competition assays 316 (n=2) with 9 fold excess VWF (volume:volume), random colonies (n= 24) were genotyped by 317 PCR as previously described (25) using primers that anneal to region of the plasmid common to 318 both SK and the VWF-A3 constructs (Supplementary Table 1; SK cloning seq check FP, SK 319 cloning seq check RP) and the results were plotted as percentages on Graphpad. 320 321 High-throughput DNA sequencing 322 Libraries for sequencing were generated as previously described (25). Briefly, primers were 323 designed to generate twelve partially overlapping 150 bp amplicons from pAY-FE error-prone 324 SK via PCR (Supplementary Table 1), with each overlapping sequence assessed in only one 325 designated amplicon (Figure 3A). PCR products were purified from agarose gels and combined 326 so that the pooled DNA (100 ng) represented equal contributions from each of the twelve 327 amplicons. The mixed amplicons were end-repaired and dA-tailed using the NEBNext Ultra End 328 Repair/dA-tail kit (New England Biolabs), followed by ligation to NextFlex barcoded adapters 329 (Bioo Scientific, Austin, TX, USA) utilizing the NEBNext Ultra Ligation kit (New England Biolabs). 330 Ampure beads (Beckman Coulter, Indianapolis, IN, USA) were used for purification in 331 accordance with the manufacturer’s protocol. HTS (2x150bp paired-end reads) was performed 332 on the Illumina platform (Illumina, San Diego, CA, USA) by MedGenome (Foster City, California, 333 USA). Sequence data analysis was performed with DESeq2 (47) (10% FDR cut off), evaluating 334 mutation frequency at each nucleotide position independently of other mutations present in the 335 SK coding sequence. 336 337 Amidolytic assays 338 To check for background PLG activation during our SK phage binding protocol, we measured 339 PLG activation at the incubation points during the course of the binding experiment (n=2). We 340 added 100 µL of the reaction mix to the substrate after 30 minutes of incubation, the first 341 overnight, or after the second overnight step. We also tested samples taken after both overnight 342 steps by mixing them with 100 µL of (z-Gly-Gly-Arg-7-amido-4-methylcoumarin (AMC).HCl 343 substrate (Bachem 4002155.0025, Product #50-260-284, Fisher Scientific)). Fluorescence 344 intensity was measured in technical duplicates (excitation at 370 nm, emission at 440 nm) every 345 minute for 4 hours at 25°C. Commercial Streptokinase C (Product #S0577, Millipore Sigma) 346 served as a positive control, and glu-PLG (Product #HCPG-0130, Prolytix) was used as a 347 negative control. Fluorescence was measured on a SpectraMax M3, (Molecular Devices, San 348 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 9 Jose, CA, USA), baseline corrected and means of the technical replicates for each experiment 349 were plotted using Graphpad Prism. 350 351 Protein stability predictors and surface accessibility calculation 352 The impact of individual amino acid substitutions was assessed using EvoEF2 (37) and FoldX 353 (38), as described previously (31) using the structural information from the AlphaFold2-354 generated structure of SKH46A (48). We calculated the relative solvent accessibility of each 355 residue using the same structure and the DSSP v2.3 software (49). The raw solvent 356 accessibility values were transformed by dividing each by the residue-specific maximum allowed 357 solvent accessibilities in the empirical matrix provided in (50). The resulting relative solvent 358 accessibilities were used for downstream analyses. 359 360 Recombinant SKH46A expression and purification 361 DNA coding for wildtype, M369F, K76L, G50L, and Y252T SKH46A with a C-terminal His6-tag and 362 TEV protease cleavage site were purchased from Twist Bioscience in a pET-24(+) expression 363 vector carrying a kanamycin resistance gene. Plasmids were transformed into chemically 364 competent NiCo21 DE3 E. coli (Product #C2529H, New England Biolabs). Protein expression 365 and purification were performed as described in (51) .The cultures were grown in LB media at 366 30°C in log phase with shaking (OD600 0.4-0.6). IPTG (0.4 mM) was added for 2 hours followed 367 by centrifugation of samples at 6000g x 20 mins at 4°C. The pellet was resuspended with 0.85% 368 NaCl and the samples were spun again at 6000g x 20 mins at 4°C.The pellets were 369 resuspended with buffer (25mM Hepes and 150Mm NaCl; (1X HBS pH 7.4)) and treated with 370 6.25 μM MgCl2, 0.6uM CaCl2, DNase (2mg/ml) and incubated with rotation at room temperature. 371 The samples were freeze thawed in liquid nitrogen (1 min) and 37°C (5 min) three times before 372 additional DNase (2mg/ml) was added. Finally, the volume was brought up to 5 mL to centrifuge 373 at 16,000g x 20 mins. The resulting supernatant and pellet were checked for the presence of 374 protein on a Coomassie gel, and the supernatant was purified by Fast Protein Liquid 375 Chromatography (FPLC, AKTA Pure, Cytiva Corp, MA, USA). The SK protein was purified with a 376 HiTrap TALON crude 1ml column (Cytiva) and equilibrated with 25mM HEPES containing 377 150mM NaCl and 5mM imidazole. The protein was eluted in 25mM HEPES containing 150mM 378 NaCl and150mM imidazole and subsequently buffer exchanged into 25mM HEPES containing 379 150mM NaCl using a HiPrep 26/10 desalting column (Cytiva). The eluted protein fractions were 380 pooled and stored at -80 oC and the purity of the protein assessed by SDS-PAGE followed by 381 Coomassie staining. 382 383 The protein was then treated with TEV protease (Product # P8112S, New England Biolabs) 384 according to the manufacturer’s instructions. To perform our assays, recombinant SK (15 μg) 385 was treated with 1X TEV Protease buffer with or without TEV protease enzyme (reaction and 386 control, respectively) at 4°C overnight. The samples were buffer exchanged with 1XHBS in 387 10KDa ultracentrifugal filters (Product# UFC901008, Millipore Sigma) to remove DTT from the 388 TEV protease buffer and quantified spectrophotometrically (assuming a 1% extinction coefficient 389 of 10). 390 391 Mutational acceptance score calculations 392 Mutational acceptance scores for each position in SK were determined by averaging the log2-393 fold functional stability scores from our binding screen (27, 31). Using GraphPad Prism, we then 394 applied a LOWESS regression with a 20-point smoothing window to the data. Areas with higher 395 binding were defined as regions where the LOWESS curve was above the 75th percentile value, 396 while regions of lower binding were identified where the curve was below the 25th percentile. 397 These data were further used to generate heatmaps. 398 399 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 10 Modified Halo/HoFF test assay 400 The Halo assays were performed based on previous reports (39, 52). Human α-thrombin 401 (Product #HCATIII-0120, Prolytix, Essex Junction, VT, USA) was diluted to 240 nM in halo 402 assay buffer (25 mM HEPES, 137 mM NaCl, 400 mM CaCl₂, pH 7.4). Five microliters of this 403 solution were added to each well of a white-walled, tissue culture-treated 96-well plate 404 (Product#3610, Corning, Corning, NY), positioning the droplet against the wall. Human pooled 405 plasma (25 µL; Innovative Research, Novi, MI), spiked with Alexa Fluor™ 488 Conjugate 406 Fibrinogen (19.2 µg/mL; Product # F13191, ThermoFisher) , and was dispensed into the 407 thrombin droplet by circling the pipette tip along the bottom outer edge of the well, forming a 408 halo-shape. The plasma mixtures were incubated for 1 hour at 37 °C to allow clot formation. 409 Reaction mixes containing 25 mM HEPES, 137 mM NaCl, 2.5 mM Ca²⁺, 0.8 mM S-2251 410 (plasmin chromogenic substrate; DiaPharma, West Chester, OH, USA), and 12.8 nM of a given 411 SK variant were prepared for triplicate wells (final well concentrations: 8 nM SK, 0.5 mM S-412 2251). The plate reader was preheated to 37 °C. After warming, 50 µL of reaction mixture was 413 added to each well, and absorbance at 405 nm (S-2251 cleavage) and fluorescence (excitation 414 488 nm, emission 530 nm; Alexa488-fibrin cleavage) were measured every 90 seconds for 1 415 hour and 20 minutes at 37 °C. The absorbance and the fluorescence were measured 416 simultaneously on a BioTek CYTATION5 imaging reader (BioTek, Winooski, VT, USA). 417 Absorbance data were baseline-corrected, and the means of the three replicates were plotted 418 as a function of time for each variant. 419 Data analysis and Code 420 All code for data analysis has been deposited on GitHub 421 (https://github.com/SrishtiBaid/SK_PLG_binding.git), and HTS sequencing results are available 422 at https://doi.org/10.7302/d3aw-z910. 423 424

425 Funding was provided by AHA Postdoctoral Fellowship: 25POST1372745 (SB), and NIH 426 R35HL171421 (DG). UM-GPT was used to refine the language and the manuscript flow. The 427 authors thank Khushali Kataria for assistance with the schematic figures of this manuscript. 428 429 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 11

430 431 1. Floden AM, Watt JA, Brissette CA. 2011. Borrelia burgdorferi enolase is a surface-432 exposed plasminogen binding protein. PLoS One 6:e27502. 433 2. Parry MA, Zhang XC, Bode I. 2000. Molecular mechanisms of plasminogen 434 activation: bacterial cofactors provide clues. Trends Biochem Sci 25:53-9. 435 3. Sun H, Ringdahl U, Homeister JW, Fay WP , Engleberg NC, Yang AY , Rozek LS, Wang X, 436 Sjöbring U, Ginsburg D. 2004. Plasminogen is a critical host pathogenicity factor for 437 group A streptococcal infection. Science 305:1283-6. 438 4. Porcellato D, Smistad M, Skeie SB, Jørgensen HJ, Austbø L, Oppegaard O. 2021. 439 Whole genome sequencing reveals possible host species adaptation of 440 Streptococcus dysgalactiae. Scientific Reports 11:17350. 441 5. Brouwer S, Rivera-Hernandez T, Curren BF , Harbison-Price N, De Oliveira DMP , 442 Jespersen MG, Davies MR, Walker MJ. 2023. Pathogenesis, epidemiology and 443 control of Group A Streptococcus infection. Nat Rev Microbiol 21:431-447. 444 6. Thacharodi A, Hassan S, Vithlani A, Ahmed T, Kavish S, Geli Blacknell NM, Alqahtani 445 A, Pugazhendhi A. 2025. The burden of group A Streptococcus (GAS) infections: The 446 challenge continues in the twenty-first century. iScience 28:111677. 447 7. Klos M, Wojkowska-Mach J. 2017. Pathogenicity of Virulent Species of Group C 448 Streptococci in Human. Can J Infect Dis Med Microbiol 2017:9509604. 449 8. Eraso JM, Olsen RJ, Long SW, Gadd R, Boukthir S, Faili A, Kayal S, Musser JM. 2024. 450 Integrative genomic, virulence, and transcriptomic analysis of emergent 451 Streptococcus dysgalactiae subspecies equisimilis (SDSE) emm type stG62647 452 isolates causing human infections. mBio 15:e0257824. 453 9. Cook SM, Skora A, Gillen CM, Walker MJ, McArthur JD. 2012. Streptokinase variants 454 from Streptococcus pyogenes isolates display altered plasminogen activation 455 characteristics - implications for pathogenesis. Mol Microbiol 86:1052-62. 456 10. del Zoppo GJ. 2016. 2 - Mechanisms of Thrombosis and Thrombolysis, p 13-27. In 457 Grotta JC, Albers GW, Broderick JP , Kasner SE, Lo EH, Mendelow AD, Sacco RL, 458 Wong LKS (ed), Stroke (Sixth Edition) doi:https://doi.org/10.1016/B978-0-323-459 29544-4.00002-5. Elsevier, London. 460 11. Cook Simon M, Skora A, Walker Mark J, Sanderson-Smith Martina L, McArthur 461 Jason D. 2014. Site-restricted plasminogen activation mediated by group A 462 streptococcal streptokinase variants. Biochemical Journal 458:23-31. 463 12. Boxrud PD, Fay WP , Bock PE. 2000. Streptokinase binds to human plasmin with high 464 aiinity, perturbs the plasmin active site, and induces expression of a substrate 465 recognition exosite for plasminogen. J Biol Chem 275:14579-89. 466 13. Huang TT, Malke H, Ferretti JJ. 1989. The streptokinase gene of group A streptococci: 467 cloning, expression in Escherichia coli, and sequence analysis. Mol Microbiol 3:197-468 205. 469 14. Kunamneni A, Abdelghani TT, Ellaiah P . 2007. Streptokinase--the drug of choice for 470 thrombolytic therapy. J Thromb Thrombolysis 23:9-23. 471 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 12 15. Keramati M, Mianroodi RA, Memarnejadian A, Mirzaie A, Sazvari S, Aslani MM, 472 Roohvand F . 2013. Towards a superior streptokinase for fibrinolytic therapy of 473 vascular thrombosis. Cardiovasc Hematol Agents Med Chem 11:218-29. 474 16. Muoghalu CG, Ekong N, Wyns W, Ofoegbu CC, Newell M, Ebirim DA, Alex-Ojei ST. 475 2023. A Systematic Review of the Eiicacy and Safety of Tenecteplase Versus 476 Streptokinase in the Management of Myocardial Infarction in Developing Countries. 477 Cureus 15:e44125. 478 17. White HD. 1989. Comparison of tissue plasminogen activator and streptokinase in 479 the management of acute myocardial infarction. Chest 95:265S-269S. 480 18. Conejero-Lara F , Parrado J, Azuaga AI, Dobson CM, Ponting CP . 1998. Analysis of the 481 interactions between streptokinase domains and human plasminogen. Protein Sci 482 7:2190-9. 483 19. Wang X, Lin X, Loy JA, Tang J, Zhang XC. 1998. Crystal structure of the catalytic 484 domain of human plasmin complexed with streptokinase. Science 281:1662-5. 485 20. Verhamme IM, Panizzi PR, Bock PE. 2015. Pathogen activators of plasminogen. J 486 Thromb Haemost 13 Suppl 1:S106-14. 487 21. Boxrud PD, Verhamme IM, Fay WP , Bock PE. 2001. Streptokinase triggers 488 conformational activation of plasminogen through specific interactions of the 489 amino-terminal sequence and stabilizes the active zymogen conformation. J Biol 490 Chem 276:26084-9. 491 22. Loy JA, Lin X, Schenone M, Castellino FJ, Zhang XC, Tang J. 2001. Domain 492 interactions between streptokinase and human plasminogen. Biochemistry 493 40:14686-95. 494 23. Ayinuola YA, Brito-Robinson T, Ayinuola O, Beck JE, Cruz-Topete D, Lee SW, Ploplis 495 VA, Castellino FJ. 2021. Streptococcus co-opts a conformational lock in human 496 plasminogen to facilitate streptokinase cleavage and bacterial virulence. J Biol 497 Chem 296:100099. 498 24. Mican J, Toul M, Bednar D, Damborsky J. 2019. Structural Biology and Protein 499 Engineering of Thrombolytics. Comput Struct Biotechnol J 17:917-938. 500 25. Huttinger ZM, Haynes LM, Yee A, Kretz CA, Holding ML, Siemieniak DR, Lawrence 501 DA, Ginsburg D. 2021. Deep mutational scanning of the plasminogen activator 502 inhibitor-1 functional landscape. Sci Rep 11:18827. 503 26. Haynes LM, Huttinger ZM, Yee A, Kretz CA, Siemieniak DR, Lawrence DA, Ginsburg 504 D. 2022. Deep mutational scanning and massively parallel kinetics of plasminogen 505 activator inhibitor-1 functional stability to probe its latency transition. J Biol Chem 506 298:102608. 507 27. Haynes LM, Holding ML, DiGiovanni HL, Siemieniak D, Ginsburg D. 2025. High-508 throughput amino acid-level characterization of the interactions of plasminogen 509 activator inhibitor-1 with variably divergent proteases. Protein Sci 34:e70088. 510 28. Haynes LM, Holding ML, Woodard J, Siemieniak D, Ginsburg D. 2025. 511 Thermodynamics and selection of the plasminogen activator inhibitor-1 latency 512 transition. J Biol Chem doi:10.1016/j.jbc.2025.110896:110896. 513 29. Laha M, Panizzi P , Nahrendorf M, Bock PE. 2011. Engineering streptokinase for 514 generation of active site-labeled plasminogen analogs. Anal Biochem 415:105-15. 515 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 13 30. Thelwell C, Longstai C. 2014. Biosimilars: the process is the product. The example 516 of recombinant streptokinase. J Thromb Haemost 12:1229-33. 517 31. Haynes LM, Holding ML, Woodard J, Siemieniak D, Ginsburg D. 2025. 518 Thermodynamics and selection of the plasminogen activator inhibitor-1 latency 519 transition. J Biol Chem 301:110896. 520 32. Bean RR, Verhamme IM, Bock PE. 2005. Role of the streptokinase alpha-domain in 521 the interactions of streptokinase with plasminogen and plasmin. J Biol Chem 522 280:7504-10. 523 33. Mundada LV , Prorok M, DeFord ME, Figuera M, Castellino FJ, Fay WP . 2003. 524 Structure-function analysis of the streptokinase amino terminus (residues 1-59). J 525 Biol Chem 278:24421-7. 526 34. Dhar J, Pande AH, Sundram V , Nanda JS, Mande SC, Sahni G. 2002. Involvement of a 527 nine-residue loop of streptokinase in the generation of macromolecular substrate 528 specificity by the activator complex through interaction with substrate kringle 529 domains. J Biol Chem 277:13257-67. 530 35. Tharp AC, Laha M, Panizzi P , Thompson MW, Fuentes-Prior P , Bock PE. 2009. 531 Plasminogen substrate recognition by the streptokinase-plasminogen catalytic 532 complex is facilitated by Arg253, Lys256, and Lys257 in the streptokinase beta-533 domain and kringle 5 of the substrate. J Biol Chem 284:19511-21. 534 36. Panizzi P , Boxrud PD, Verhamme IM, Bock PE. 2006. Binding of the COOH-terminal 535 lysine residue of streptokinase to plasmin(ogen) kringles enhances formation of the 536 streptokinase.plasmin(ogen) catalytic complexes. J Biol Chem 281:26774-8. 537 37. Huang X, Pearce R, Zhang Y . 2020. EvoEF2: accurate and fast energy function for 538 computational protein design. Bioinformatics 36:1135-1142. 539 38. Schymkowitz J, Borg J, Stricher F , Nys R, Rousseau F , Serrano L. 2005. The FoldX web 540 server: an online force field. Nucleic Acids Res 33:W382-8. 541 39. Liu Z, Zaacks O, Niego B, Medcalf RL. 2025. Halo fluorescence fibrinolysis test: a 542 novel quantitative assay to evaluate fibrinolysis on established plasma clots. Res 543 Pract Thromb Haemost 9:102874. 544 40. Palazzolo JS, Medcalf RL, Hagemeyer CE, Niego B. 2022. A novel ex vivo approach 545 for measuring plasminogen activation upon established plasma clots. Res Pract 546 Thromb Haemost 6. 547 41. Takada A, Takada Y, Sugawara Y. 1988. Activation of Val442-plasminogen (mini-548 plasminogen) by urokinase, streptokinase and tissue plasminogen activator. 549 Thromb Res 49:253-63. 550 42. Franzosa EA, Xia Y . 2009. Structural Determinants of Protein Evolution Are Context-551 Sensitive at the Residue Level. Molecular Biology and Evolution 26:2387-2395. 552 43. Ghose DA, Soderstrom CBW, Mahoney EM, Laub MT. 2025. Multi-environment deep 553 mutational scanning reveals the distribution of temperature-sensitive variants in a 554 bacterial kinase. Cell Reports 44:116446. 555 44. Escobedo A, Voigt G, Faure AJ, Lehner B. 2025. Genetics, energetics, and allostery in 556 proteins with randomized cores and surfaces. Science 389:eadq3948. 557 45. Mullen LM, Nair SP , Ward JM, Rycroft AN, Henderson B. 2006. Phage display in the 558 study of infectious diseases. Trends Microbiol 14:141-7. 559 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 14 46. Haynes LM, Holding ML, Woodard J, Siemieniak D, Ginsburg D. 2025. 560 Thermodynamics and selection of the plasminogen activator inhibitor-1 latency 561 transition. bioRxiv doi:10.1101/2025.06.03.655624. 562 47. Love MI, Huber W, Anders S. 2014. Moderated estimation of fold change and 563 dispersion for RNA-seq data with DESeq2. Genome Biology 15:550. 564 48. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool 565 K, Bates R, Žídek A, Potapenko A, Bridgland A, Meyer C, Kohl SAA, Ballard AJ, Cowie 566 A, Romera-Paredes B, Nikolov S, Jain R, Adler J, Back T, Petersen S, Reiman D, 567 Clancy E, Zielinski M, Steinegger M, Pacholska M, Berghammer T, Bodenstein S, 568 Silver D, Vinyals O, Senior AW, Kavukcuoglu K, Kohli P , Hassabis D. 2021. Highly 569 accurate protein structure prediction with AlphaFold. Nature 596:583-589. 570 49. Touw WG, Baakman C, Black J, te Beek TA, Krieger E, Joosten RP , Vriend G. 2015. A 571 series of PDB-related databanks for everyday needs. Nucleic Acids Res 43:D364-8. 572 50. Tien MZ, Meyer AG, Sydykova DK, Spielman SJ, Wilke CO. 2013. Maximum allowed 573 solvent accessibilites of residues in proteins. PLoS One 8:e80635. 574 51. Ward MS, Holding ML, Haynes LM, Margres MJ, Matocq MD, Ginsburg D. 2025. 575 Tandem Duplication of Serpin Genes Yields Functional Variation and Snake Venom 576 Inhibitors. Molecular Biology and Evolution 42. 577 52. Palazzolo JS, Medcalf RL, Hagemeyer CE, Niego Be. 2022. A novel ex vivo approach 578 for measuring plasminogen activation upon established plasma clots. Research 579 and Practice in Thrombosis and Haemostasis 6. 580 53. Nolan M, Bouldin SD, Bock PE. 2013. Full time course kinetics of the streptokinase-581 plasminogen activation pathway. J Biol Chem 288:29482-93. 582 54. Wickham H. 2016. ggplot2: Elegant Graphics for Data Analysis. Springer 583 International Publishing. 584 585 586 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 15 Figures and Figure legends 587 Figure 1 588 589 590 Figure 1. Phage display of functional recombinant SK. A) SK was displayed on M13 591 filamentous phage by fusion with truncated pIII protein with an N-terminal His-TEV and C-592 terminal FLAG tag. The NotI and AscI restriction sites flank the His-TEV and the FLAG 593 sequences, respectively, to facilitate cloning of the recombinant SK DNA into the phagemid pAY-594 FE plasmid vector, as previously described (25, 26). B) Phage displayed SK was incubated with 595 an excess of VWF A3 domain phage (Input), selected for binding to biotinylated plasminogen 596 (Selected). Individual colonies (36 colonies for each condition, for a total of 2 experiments) of 597 phage were genotyped by PCR with flanking primers (Supplementary Table 1) and run on 598 agarose gel and are represented as a percentage of total clones screened for each condition, 599 as described in Materials and Methods. 600 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 16 Figure 2: 601 602 603 Figure 2. Selection of variants in the phage displayed SK library by affinity for 604 biotinylated PLG exhibits distinct clustering. A) Schematic of the procedure to measure PLG 605 binding by the SK phage library. Phage were isolated and incubated with biotinylated PLG, 606 collected by immunoprecipitation with streptavidin beads, followed by High throughput DNA 607 sequencing (HTS). B) PCA plot shows the correlation of data points among the control reaction 608 (SK-PLG) (pink), the input (black) and the experimental reaction (SK+PLG) (yellow). Replicates 609 (n=4 for inputs and control and n=3 for the experimental samples) for each reaction cluster 610 together, with high variance between the experimental sets. 611 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 17 Figure 3: 612 613 614 615 616 Figure 3: Specific amplicons in SK correspond to regions critical for PLG binding. A) 617 Schematic representation of the SK sequence (1398 bp) showing each amplicon along with the 618 included amino acids. Each amplicon was approximately 150bp in length. B) PCA plots show 619 comparison (log2- fold enrichment scores with respect to binding) among the 12 different 620 amplicons across Input (black, n=4), SK-PLG (pink, n=4) control and SK+PLG (orange, n=3) 621 reaction sets to show region specific effects. 622 623 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 18 Figure 4: 624 625 626 627 Figure 4: Heat map of DMS analysis and structure alignment of SK and PLG interaction 628 show amino acids critical for binding. A) The consolidated heat map at each position of the 629 SK amino acid sequence indicating those mutations that result in either loss of PLG binding 630 (red) or maintained/enhanced binding (blue). Amino acid positions are indicated on the x- axis, 631 with the y- axis showing the amino acid substitution at each position. The charcoal grey 632 indicates residues and white indicates lack of coverage at that position. B) An AlphaFold (48) 633 model of the SK PLG complex is shown, with SK in grey and PLG in yellow. Selected domains 634 of SK are highlighted in the colors corresponding to the colored segments at the top of panel A – 635 residues 9-39 (red) ,55-85 (blue), 99-148 (maroon), 171-186 (purple), 261-267 (pink), 288-301 636 and 302-313 (purple) 337-374 (blue), 389-407 (red). The figure is rotated 180° on the left. The 637 residues that were expressed and validated for activation experiments (Figure 6) are 638 highlighted with black circles. 639 640 641 642 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 19 Figure 5 643 644 645 646 647 Figure 5: Stability predictions (ΔΔG values) for mutants, using 2 models. EvoEF and 648 FoldX predict that SKH46A variants with higher PLG binding (n=754; binding score > 0) have 649 more negative ΔΔG values than those with reduced binding (n=448; binding score < 0). Select 650 SKH46A variants exhibit low binding scores but ΔΔG values similar to strong binders, 651 suggesting these are properly folded proteins with impaired binding. Amino acid substitutions 652 that decrease thermodynamic stability and cause misfolding show higher (more positive) ΔΔG 653 values, while those that stabilize the protein have lower (more negative) ΔΔG values, consistent 654 with proper folding. 655 656 657 658 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 20 Figure 6: 659 660 661 662 663 664 Figure 6: Assessment of select mutants show similar levels of plasmin activation and 665 fibrinogen breakdown as compared to wildtype. The select SK variants (G50L, M369F, 666 Y252T and K76L) were expressed, treated with TEV protease, to free the Ile1 and checked for 667 activity by performing a modified Halo assay (described in Materials and Methods). A) The 668 absorbance (labs = 405nm) was measured for plasmin activity using plasminogen chromogenic 669 substrate S-2551 and B) Fluorescence (lex = 488 nm and lem = 530 nm) was measured for clot 670 fibrinolysis using fibrinogen simultaneously. Statistical analysis using ANOVA suggested no 671 significant difference in activity between wildtypes and mutants for both experiments. 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint 21 Figure 7 693 694 695 696 Figure 7: Amino acid substitutions in surface-exposed SK residues are more likely to 697 enhance PLG binding. The mean positional enrichment scores (log₂-fold change) for each site 698 in SK (from Figure 4) are plotted against the relative solvent-accessible surface area (RSA) as 699 predicted by the AlphaFold2 SK structure (see Methods and Figure 4D). Residues with RSA 700 values below 0.2 are typically buried within the protein core, while higher RSA values denote 701 greater surface exposure. Elevated mean enrichment scores correspond to positions more 702 tolerant of substitutions, whereas more negative values indicate lower tolerance for amino acid 703 changes. The red line depicts the best-fit linear regression, with a 95% confidence interval, 704 yielding a slope of 1.1, R² of 0.1, and p-value of 1.12 × 10⁻¹¹ 705 706 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted February 4, 2026. ; https://doi.org/10.64898/2026.02.04.703780doi: bioRxiv preprint