Title: Strain variation in Clostridioides difficile toxin activity associated with genomic variation

15 Clinical disease from Clostridioides difficile infection can be mediated by two toxins and their 16 neighboring regulatory genes encoded within the five-gene pathogenicity locus (PaLoc). We 17 provide several lines of evidence that the toxin activity of C. difficile may be modulated by 18 genomic variants outside of the PaLoc. We used a phylogenetic tree-based approach to 19 demonstrate discordance between toxin activity and PaLoc evolutionary history, an elastic net 20 method to show the insufficiency of PaLoc variants alone to model toxin activity, and a 21 convergence-based bacterial genome-wide association study (GWAS) to identify correlations 22 between non-PaLoc loci with changes in toxin activity. Combined, these data support a model of 23 C. difficile disease wherein toxin activity may be strongly affected by many non-PaLoc loci. 24 Additionally, we characterize multiple other in vitro phenotypes relevant to human infections 25 including germination and sporulation. These phenotypes vary greatly in their clonality, 26 variability, convergence, and concordance with genomic variation. Lastly, we highlight the 27 intersection of loci identified by GWAS for different phenotypes and clinical severity. This 28 strategy to identify the overlapping loci can facilitate the identification of genetic variation 29 linking phenotypic variation to clinical outcomes. 30 31 IMPORTANCE 32 Clostridioides difficile has two major disease mediating toxins, A and B, encoded within the 33 pathogenicity locus (PaLoc). In this study we demonstrate via multiple approaches that genomic 34 variants outside of the PaLoc are associated with changes in toxin activity. These genomic 35 variants may provide new avenues of exploration in the hunt for novel disease modifying 36 interventions. Additionally, we provide insight into the evolution of several additional 37 phenotypes also critical to clinical infection such as sporulation, germination, and growth rate. 38 These in vitro phenotypes display a range of responses to evolutionary pressures and as such 39 vary in their appropriateness for certain bacterial genome wide association study approaches. We 40 used a convergence-based association method to identify the genomic variants most correlated 41 with both changes in these phenotypes and disease severity. These overlapping loci may be 42 important to both bacterial function and human clinical disease. 43 44