Allelic diversity uncovers protein domains contributing to the emergence of antimicrobial resistance

Antimicrobial resistance (AMR) remains a major threat to global health. To date, tractable approaches that decipher how AMR emerges within a bacterial population remain limited. Here, we developed a framework that exploits genetic diversity from environmental bacterial populations to decode emergent phenotypes such as AMR. OmpU is a porin that can make up to 60% of the outer membrane of Vibrio cholerae, the cholera pathogen. This porin is directly associated with the emergence of toxigenic clades and confers resistance to numerous host antimicrobials. In this study, we examined naturally occurring allelic variants of OmpU in environmental V. cholerae and established associations that connected genotypic variation with phenotypic outcome. We covered the landscape of gene variability and found that the porin forms two major phylogenetic clusters with striking genetic diversity. We generated 14 isogenic mutant strains, each encoding a unique ompU allele, and found that divergent genotypes lead to convergent antimicrobial resistance profiles. We identified and characterized functional domains in OmpU unique to variants conferring AMR-associated phenotypes. Specifically, we identified four conserved domains that are linked with resistance to bile and host-derived antimicrobial peptides. Mutant strains for these domains exhibit differential susceptibility patterns to these and other antimicrobials. Interestingly, a mutant strain in which we exchanged the four domains of the clinical allele for those of a sensitive strain exhibits a resistance profile closer to a porin deletion mutant. Finally, using phenotypic microarrays, we uncovered novel functions of OmpU and their connection with allelic variability. Our findings highlight the suitability of our approach towards dissecting the specific protein domains associated with the emergence of AMR and can be naturally extended to other bacterial pathogens and biological processes.

diversity from environmental bacterial populations to decode emergent 23 phenotypes such as AMR. OmpU, is a porin that makes up to 60% of the outer 24 membrane of Vibrio cholerae, the cholera pathogen. This porin is directly 25 associated with the emergence of the bacterium and confers resistance to 26 numerous host antimicrobials. In this study, we examined naturally occurring allelic 27 variants of OmpU in environmental V. cholerae and established associations that 28 connected genotypic variation with phenotypic outcome. We covered the 29 landscape of gene variability and found that the porin forms two major phylogenetic 30 clusters with striking genetic diversity. We generated 14 isogenic mutant strains, 31 each encoding a unique ompU allele, and found that divergent genotypes lead to 32 convergent antimicrobial resistance profiles. We identified and characterized 33 functional domains in OmpU unique to variants conferring AMR-associated 34 phenotypes. Specifically, we identified four conserved domains that are linked with 35 resistance to bile and host-derived antimicrobial peptides. Mutant strains for these 36 domains exhibit differential susceptibility patterns to these and other  Antimicrobial resistance (AMR) is a complex phenomenon fueling the arms race 59 to combat emerging and re-emerging pathogens. By 2050, it is estimated that 10 60 million individuals will die annually due to infections caused by pathogens that 61 exhibit AMR [1]. Resistance to many currently used antimicrobials such as 62 ciprofloxacin or ampicillin has emerged in up to 80% of Klebsiella pneumoniae and the emergence of AMR, first, we examined the natural variability of ompU 110 uncovering 41 alleles of the porin out of over 1600 sequences analyzed. These 111 alleles form two phylogenetic clusters with distinct evolutionary paths and striking 112 diversity. We selected 7 representative alleles from each cluster encompassing 113 the diversity of ompU and constructed 14 isogenic mutants where we exchanged 114 the toxigenic allele for an environmental one. We exposed the ompU mutant tolerance, antimicrobial peptide resistance, and intestinal colonization [30,31,33]. 135 We recently determined that only some alleles of ompU confer these observed 136 phenotypes [27] these functional residues, we first performed a sequence alignment of the N16961 144 (resistant) and the GBE1114 (sensitive) proteins (Fig 1A) obtaining a total of 1620 sequences. After a previous clustering of the sequences 154 (100% identity), to eliminate the clonality bias of V. cholerae, we generated a phylogenetic tree to examine the evolutionary history of OmpU (Fig 1B). The 156 analysis reveals two distinct major clusters composed of 28 clades (Cluster 1; C1) 157 and 13 clades (Cluster 2; C2), respectively (Fig 1B) the resistance of these strains in the presence of bile (Fig 1C). We observed that 186 the overall resistance pattern in the presence of bile broadly correlates with tree 187 topology, with several strains from C1 exhibiting a resistance phenotype like WT 188 whereas the ones from C2 (closer to GBE1114) show a sensitive one (Fig 1C). 189 Specifically, strains ompU 1 , ompU 2 , ompU 3 , and ompU 5 from C1 display similar 190 resistance to bile as the WT strain. Interestingly, ompU 4 , ompU 6 and ompU 7 191 warrant some further investigations as they show a decrease in survival in the 192 presence of bile, with ompU 6 exhibiting a phenotype similar to some strains from 193 C2 (ompU 8 and ompU 10 ) (Fig 1C). Strains encoding alleles from C2 exhibit on 194 average ~1-log decrease in survival compared to the WT strain (Fig 1C). Those  that are so variable that no pattern could be identified even for resistant alleles 240 limiting their relevance as potential domains associated with the emergence 241 antimicrobial resistance in OmpU. Thus, they were excluded from further analysis.

242
Overall, our systematic genotype to phenotype association and comparative 243 sequence analyses led us to identify regions 1, 3, 4 and 7, which we will term N-terminal coil (NTC), L3 region (L3R), L4-loop (L4) and VAS respectively, for 245 subsequent analyses (Fig 2A, S3 Fig). OmpU GBE1114 (Fig 2B and 2D). Secondly, amino acid residues present in the L3R domain identified above, are located on the inward L3 loop and differ in charge 267 and hydrophobicity between the two alleles (Figs 2C and 2E). OmpU N16961 268 encodes charged residues (E, and K), whereas OmpU GBE1114 does not encode any 269 charged residues (Fig 2A). Lastly, domains L4R and VAS are located on loops L4 270 and L8 respectively that protrude into the extracellular space (Figs 2C and 2E). 271 Interestingly, overlay of both porins reveal an extended L4 in OmpU N16961 when 272 compared to OmpU GBE1114 . Additionally, the VAS domain is hydrophobic in 273 OmpU N16961 whereas in OmpU GBE1114 it is basic (Fig 2A). Overall, the identified 274 domains translate to several structural differences that may alter protein-protein from GBE1114. This allele was assigned the binary code '0000'. First, we verified 291 protein stability using immunoblots and subsequently examined the survival of the 292 mutants in the presence of 0.4% whole bile (Fig 3A and B). Interestingly, two 293 individual domain mutants, ompU 0111 (NTC) and ompU 1011 (L3R), show the 294 greatest susceptibility to bile with survival rates of 9.6% and 4.5% respectively (Fig   295   3B). The ompU 1101 mutant exhibits 17% survival rate, whereas the ompU 1110 296 mutant shows no survival defect, and displays similar survival as the WT. The 297 ompU 0000 "null" mutant has an average survival rate of 3.4%, closely resembling 298 the phenotype of the strain encoding GBE1114 (Fig 3B). Interestingly, the tested their bile resistance profiles after testing the stability of OmpU in those 310 mutants (Fig 3C). Upon exposure to 0.4% whole bile, all of the mutants exhibit a decrease in bile resistance with survival ranging from 6 to 12% (Fig 3B) (Fig 3C). Lastly, the strain encoding both the clinical NTC and L3R 318 domain (ompU 1100 ) shows the highest resistance to bile (12%) when compared to 319 other domain combinations (Fig 3C). Our results indicate that individually both 320 NTC and L3R domains are crucial for increased resistance to bile (Fig 3C). 321 However, the synergistic effect of all domains appears to be essential for AMR in 322 the clinical OmpU to emerge. Antimicrobial peptide P2. After exposure to P2, WT has a 91% survival rate 333 whereas the ∆ompU mutant exhibits a survival rate of 7.5%, comparable to 334 previously published studies [33]. Interestingly, the mutant ompU GBE1114 shows 335 12% survival, and the domain mutant ompU 0000 23%. (Fig 4A). All these strains 336 exhibit a statistically significant difference in survival when compared to WT 337 (p<0.001), indicating that survival in the presence of P2 is both allelic and domain-338 dependent (Fig 4A) (Fig 4B). As shown in Fig 1C, there is phenotypic diversity in the alleles tested in 347 the presence of bile, thus, we considered that it would be possible that resistance 348 to PB might be allelic-dependent yet not associated with the specific allele from 349 strain GBE1114. In order to examine this, we exposed the other 13 mutant strains 350 encoding the different ompU alleles to PB and measured their survival (Fig 4B). 351 Overall, we did not find major differences between the mutant strains and the WT, 352 certainly none of them resembling the ∆ompU mutant (Fig 4B). Strains with the 353 ompU allele from clades 4, 9, and 12 show the lowest survival in the presence of 354 PB (between 42% and 55%) but not enough to warrant strong evidence that PB 355 resistance is allelic-dependent. Furthermore, there is no clear pattern that 356 associates these strains with a given cluster.

357
Organic acid. On the other hand, even though neither ompU GBE1114 (C14) nor 358 ompU 0000 show a decrease in survival compared to WT when cells were exposed 359 to acidic pH, the role of OmpU in acid tolerance appears to be allelic-dependent 360 (Fig 4C). Specifically, strains from clades 9, 10 and 11 show between 14% and 361 20% survival in acidic pH, similar to ∆ompU (12%), whereas the WT show an 362 average survival of 50% (Fig 4C). All the strains with increased sensitivity to 363 organic acid belong to Cluster 2. Overall, our results uncover complex associations 364 between the different alleles and domains of ompU and their ability to confer 365 phenotypes associated with resistance to host antimicrobials. on this, we considered that it is possible that the porin plays other roles in V. 371 cholerae survival beyond those already known. In order to test this, we screened 372 for potential novel functions of OmpU by performing high-throughput assays using 373 phenotypic microarrays (Fig 5A). We tested for sensitivity to chemicals and

399
The WT strain shows an average survival of 44% and had a survival difference of 400~28-fold when compared to ∆ompU (1.5%) (Fig 5D). This survival difference 401 provided us with enough potential resolution to observe differences betweem the 402 alleles. As for the conditions tested above (Fig 4), next, we examined the survival 403 of the ompU GBE1114 (C14) and ompU 0000 strains in the presence of RSV. We found 404 that even though resistance to RSV appears to be allelic-dependent (ompU GBE1114 405 exhibits decreased survival), the domains that we found associated with bile and 406 P2 survival do not play a role in this phenotype (ompU 0000 exhibits WT-level of 407 survival). Our results indicate that other domains within OmpU might be 408 responsible for this phenotype (Fig 5D). Finally, we tested the survival of the other 409 ompU allele mutant strains in the presence of RSV and found that four of them 410 (ompU 5 , ompU 6 , ompU 8 , and ompU 13 ) exhibit slightly increased resistance to the 411 antibiotic than the WT (Fig 5D). Overall, resistance to RSV is OmpU-dependent 412 and no allele other than ompU GBE1114 display lower survival rate than the WT allele.

505
Future work will address this and other critical questions that will take us closer to 506 dissecting the complex AMR phenomenon.  (Table 1)