Virulence phenotypes result from an interaction between pathogen ploidy and genetic background

Studying fungal virulence is often challenging and frequently depends on many contexts, including host immune status and pathogen genetic background. However, ploidy has often been overlooked when studying virulence in eukaryotic pathogens. Since fungal pathogens, including the human opportunistic pathogen Candida albicans, can display extensive ploidy variation, assessing how ploidy impacts virulence has important clinical relevance. Here, we assessed how C. albicans ploidy and genetic background impact virulence phenotypes in both healthy and immunocompromised nematode hosts. In addition to reducing overall host survival, Candida negatively impacted host reproduction, which allowed us to survey lethal and non-lethal virulence phenotypes. While we did not detect any global differences in virulence between diploid and tetraploid pathogens, there were significant interactions between ploidy and C. albicans genetic background, regardless of host immune function.


47
Virulence is measured by the reduction of host fitness resulting from a host-pathogen 48 interaction 1-3 . Therefore, virulence is not solely the property of the pathogen, but rather the 49 product of the interaction between a host and its pathogen 4,5 . While many biotic and abiotic 50 factors contribute to virulence 6,7 , the genotype-by-genotype interaction between hosts and 51 pathogens is a primary determinant of whether a host gets infected and the resulting level of 52 virulence 8,9 . One important, yet understudied, element of an organism's genotype is its ploidy

60
One potential source of these contradictory results regarding ploidy is the genetic 61 background or allelic composition of the pathogen. Allelic composition refers to not just the 62 specific alleles present in a genome, but the amount of heterozygosity throughout the genome.

63
Pathogen virulence depends on the pathogen's specific allelic combination 20 and phenotypic 64 analysis of diverse clinical isolates within pathogenic species clearly demonstrate that pathogen 65 genetic background contributes to its virulence 21,22 . Ploidy intrinsically impacts allelic 66 composition -haploids contain a single set of gene alleles, whereas diploids and polyploids can either be homozygous or heterozygous for any given locus, and dominance can mask recessive 68 alleles. Allelic composition in polyploids is further complicated by multiple allelic ratios, in which 69 one to four alleles may be present, depending on the mechanism and age of the polyploidization 70 event. Thus, a major challenge in determining the specific role ploidy has on pathogen virulence 71 is disentangling it from allelic composition.

72
The opportunistic fungal pathogen Candida albicans, while typically a highly 73 heterozygous diploid 23-25 , shows tremendous ploidy variation, ranging from haploid to

85
In this study we sought to identify how C. albicans ploidy impacts its virulence by using 86 four diploid-tetraploid pairs of strains, with each pair representing a distinct genetic background.

87
We assessed virulence by monitoring four measures of host fitness using healthy and 88 immunocompromised C. elegans hosts 39 . While we find almost no overall relationship between 89 ploidy and virulence, there are detectable differences among C. albicans genetic backgrounds 90 and clear interactions between C. albicans genetic background and its ploidy state on virulence 91 phenotypes. We also observe these interactions in immunocompromised hosts, however, in 92 some cases, the ploidy-specific pattern and/or the degree of virulence severity is different 93 between healthy and immunocompromised hosts. Taken

123
One factor that may dampen any differences in virulence among C. albicans genetic 124 backgrounds is ploidy. For each genetic background, we used a related diploid strain and a 125 tetraploid strain (Table S1). The tetraploids for both the 'laboratory' strains were produced via 126 mating in the laboratory. The 'bloodstream' pair consisted of a diploid strain isolated early in the 127 infection and its corresponding tetraploid strain was isolated mid-to late-infection following hosts. We did not observe a significant difference between these two ploidy states for any of the 135 host fitness measures tested, suggesting that ploidy may not impact virulence phenotypes.

136
However, when we account for C. albicans strain background, differences between pathogen 137 ploidy emerge but depend on strain genetic background and thus, we detect a significant 138 interaction between genetic background and ploidy for host lineage growth ('interaction' 139 p=0.0354, two-way ANOVA; Fig. S1B), brood size ('interaction' p<0.0001, two-way ANOVA; Fig.   140 S1B), and reproductive timing ('interaction' p=0.0283, two-way ANOVA; Fig. S1B). While we 141 cannot directly test for an interaction using host survival curves, we do detect differences 142 between diploids and tetraploids for each C. albicans genetic background ( Fig. S1 and Table   143 S2). Taken together, these results suggest that there is no global pattern in ploidy state and 144 virulence, but ploidy in combination with genetic background does significantly contribute to C. 145 albicans virulence phenotypes.

146
When we look at the specific diploid-tetraploid pairs, representing different C. albicans 147 genetic backgrounds, we see significant differences in virulence between the diploid and 148 tetraploid for at least one fitness measure, for every C. albicans genetic background (Fig. 2). For 149 two genetic backgrounds, the laboratory homozygous and clinical bloodstream strains, the 150 diploid strain was more virulent than its tetraploid counterpart. For the laboratory heterozygous 151 and clinical oral/vaginal strains, the tetraploid strain was more virulent than its diploid 152 counterpart, when differences between ploidy states were detected. Furthermore, these genetic 153 background specific ploidy patterns are generally consistent across host fitness measures. For 154 example, the laboratory heterozygous and clinical oral/vaginal tetraploids are also more virulent 155 than their diploid counterparts for host brood size (Fig 2C), and the clinical bloodstream diploid 156 was more virulent than its tetraploid counterpart for lineage growth and delayed host 157 reproduction ( Fig. 2B and D). We performed every pairwise comparison between treatments for 158 host survival (Table S2), lineage growth (Table S3), brood size (Table S4), and reproductive 159 timing (Table S5) and find significant differences in virulence between different C. albicans 160 genetic backgrounds and ploidy states for most host fitness measures, except host lineage 161 growth, were very few differences between C. albicans strains were detected. Taken together, 162 these data support that C. albicans ploidy does contribute to its virulence phenotypes, but 163 whether it attenuates or enhances virulence depends on its genetic background.

165
Host immune status and pathogen genetic background contribute to virulence phenotypes 166 We have previously established that C. albicans and other non-albicans Candida 167 species cause more severe infections in an immunocompromised C. elegans hosts compared to healthy C. elegans hosts 39 . Given that we see pathogen ploidy patterns depend on genetic 169 background for virulence phenotypes in wildtype healthy hosts, we next wanted to assess if we 170 could detect comparable patterns in immunocompromised hosts. For each C. albicans strain,

171
we assessed four measures of host fitness and compared C. albicans virulence in 172 immunocompromised sek-1 C. elegans hosts: host survival (Fig. 3A), host lineage growth ( Fig.   173 3B), host brood size (Fig. 3C), and host reproductive timing (Fig. 3D). Nearly all the C. albicans 174 genetic backgrounds significantly reduced host fitness compared to uninfected controls, the only 175 exception was the laboratory homozygous strains did not significantly delay host reproduction.

176
The laboratory homozygous background is also less virulent than the clinical oral/vaginal 177 genetic background for host. Together, these results suggest that global differences in pathogen

195
When we look at specific diploid-tetraploid pairs, representing different C. albicans 196 genetic background, we see significant differences in virulence in immunocompromised hosts 197 between the diploid and tetraploid state for at least one fitness measure, for all C. albicans 198 genetic backgrounds except for the laboratory homozygous (Fig. 4). For the laboratory 199 heterozygous genetic background, the tetraploid counterpart was more virulent than its diploid 200 counterpart when differences between ploidy states were detected (lineage growth and brood 201 size), similar to the pattern observed in healthy hosts (Fig. 2). Furthermore, the clinical bloodstream diploid was more virulent than its tetraploid counterpart for immunocompromised 203 host survival and brood size ( Fig. 4A and C), similar to the pattern observed in healthy hosts.

204
However, the oral/vaginal strain had significant differences between its diploid and tetraploid 205 counterparts for host survival and reproductive timing ( Fig. 4A and D), however the tetraploid 206 was more virulent in the former, while the diploid was more virulent in the latter. We also 207 performed every pairwise comparison between treatments for host survival (Table S2), lineage 208 growth (Table S3), brood size (Table S4), and reproductive timing (Table S5)

216
Ploidy-specific interactions between host and pathogen genotypes.

217
Finally, we were curious if host immune status impacted the virulence relationship 218 between each diploid-tetraploid pair of strains. Since there are significant differences for most 219 host fitness measures between healthy and immunocompromised hosts even when uninfected 220 (Table S6)

229
We next examined whether the relationship between ploidy-specific virulence differences 230 changed depending on host immune status. To do this, we plotted the relative impact of the 231 diploid (solid lines) and tetraploid (dotted lines) in healthy (N2) and immunocompromised (sek-232 1) host backgrounds for all four C. albicans genetic backgrounds (Fig 5E-H). In general, there 233 was a high degree of similarity in the relationship between diploid and tetraploids across for both 234 host genotypes, as indicated by a non-significant interaction term measured by two-way pair of strains. In healthy hosts, we observed the diploid displaying more severe virulence 237 phenotypes than its tetraploid counterpart for lineage growth and reproductive timing, yet these 238 differences are diminished in immunocompromised hosts. Thus, we detect a significant 239 interactions between host immune status and C. albicans ploidy for the bloodstream genetic 240 background. Strikingly, we detect the reverse pattern for host survival with the C. albicans 241 bloodstream diploid and tetraploid strains, in which no detectable differences are observed in 242 healthy hosts, but the diploid is significantly more virulent than the tetraploid in 243 immunocompromised hosts, yet regardless of the host context or specific fitness measure, the 244 diploid is more virulent than its tetraploid counterpart. From these results, we conclude that   284 albicans ploidy and genetic backgrounds. While we did not find any overall pattern in virulence 285 between diploids and tetraploids, we did identify ploidy-specific differences depending on 286 genetic background (Figs. 2 and 4). Importantly, these specific patterns were consistent across 287 multiple host fitness measures and host immune status (Fig. 5). Furthermore, we found that 288 immunocompromised hosts had significantly more severe infections than immune competent 289 hosts (Fig. 5), similar to what is observed clinically.

290
It is important to note that the clinical C. albicans genetic backgrounds, bloodstream and 291 oral/vaginal, are not completely isogenic between the diploid and tetraploid strains, and that 292 some allelic variation exists in addition to their differences in ploidy 37 . It is feasible that some of 293 the differences in virulence that we observe between the diploid and tetraploids isolates in these 294 two clinical C. albicans backgrounds (i.e. the bloodstream diploid is more virulent than its 295 tetraploid, whereas the oral/vaginal tetraploid is more virulent than its diploid) is due to these 296 allelic differences. By also measuring virulence phenotypes in laboratory-derived genomes that 297 only differ in the number of chromosome sets they contain, we can directly assess the impact 298 ploidy has on virulence. Here, we still observe different patterns of ploidy-specific virulence, 299 depending on genetic background. In the laboratory heterozygous genome, we consistently 300 observe the tetraploid as more virulent than its diploid in both healthy (Fig. 2) and 301 immunocompromised (Fig. 4) host contexts. However, in the laboratory homozygous genome,

302
we frequently failed to find any significant differences between diploid and tetraploid for any of 303 the host fitness measures, the only exception being healthy host survival, in which the tetraploid was avirulent and the diploid was virulent ( Fig. 2A). This result is consistent with previous 305 findings that C. albicans homozygous genomes do not show significant growth differences in 306 vitro or in vivo between ploidy states 27,46 .

307
In this work we found that ploidy and genetic background interact to contribute to C. 308 albicans virulence (Figs. S1 and S2). We also observe significant interactions between C. 309 albicans strains and host immune status (Fig. 5). These results indicate that virulence is not 310 simply a binary 'avirulent/virulent' classification, but rather a complex trait and we need to start                                     A) Survival curves for C. elegans populations that are either uninfected (exposed to just an E. coli food source, grey) or when infected with different C. albicans strains (indicated in legend). The number of worms analyzed (n) for each treatment is indicated in Table S1. Statistical significance was tested using pairwise comparisons of survival curves with Log-rank (Mantel-Cox) test. Astrisks denote statistical differences compared to the uninfected control (* indicates p <0.05, **** indicates p < 0.0001). C. albicans treatments that share letters are not significantly different, whereas treatments with differing letters are stastically different. B) Box and whiskers plot of host lineage growth which represents the total population size (representing the number of F1 and F2 progeny) produced within 7 days from a single founder C. elegans host infected with C. albicans. Boxes indicate the 25-75th quartiles with median indicated. Error bars are the normalized range of the data and circles indicate outliers. The mean and 95% CI of the uninfected control treatment are indicated by the grey dashed line and shaded grey box. Statistical significance was tested using one-way ANOVA. Astrisks denote statistical differences compared to the uninfected control (* indicates p <0.05, *** indicates p < 0.001). C. albicans treatments that share letters are not significantly different, whereas treatments with differing letters are stastically different, post-hoc Dunn's multiple comparison test. C) Total brood size and D) Average percentage of host progeny produced during days 1-3 of adulthood (normal reproductive timing) of C. elegans infected with C. albicans. Data and statistical analysis are the same as (B). E) % host survival on Day 7 for diploid (dip) and tetraploid (tet) C. albicans strains (colored symbols indicate specific C. albicans genetic background). Statistical significance was tested using Wilcoxon matched-pairs signed rank test and p values are indicated.    A) Survival curves for sek-1 C. elegans populations that are either uninfected (exposed to just an E. coli food source, grey) or when infected with different C. albicans strains (indicated in legend). Statistical significance was tested using pairwise comparisons of survival curves with Log-rank (Mantel-Cox) test. Astrisks denote statistical differences compared to the uninfected control (* indicates p <0.05, **** indicates p < 0.0001). C. albicans treatments that share letters are not significantly different, whereas treatments with differing letters are stastically different. B) Box and whiskers plot of host lineage growth which represents the total population size (representing the number of F1 and F2 progeny) produced within 7 days from a single founder sek-1 C. elegans host infected with C. albicans. Boxes indicate the 25-75th quartiles with median indicated. Error bars are the normalized range of the data and circles indicate outliers. The mean and 95% CI of the uninfected control treatment are indicated by the grey dashed line and shaded grey box. Statistical significance was tested using one-way ANOVA. Astrisks denote statistical differences compared to the uninfected control (* indicates p <0.05, *** indicates p < 0.001). C. albicans treatments that share letters are not significantly different, whereas treatments with differing letters are stastically different, post-hoc Dunn's multiple comparison test. C) Total brood size and D) Average percentage of host progeny produced during days 1-3 of adulthood (normal reproductive timing) of sek-1 C. elegans infected with C. albicans. Data and statistical analysis are the same as (B). E) % sek-1 host survival on Day 7 for diploid (dip) and tetraploid (tet) C. albicans strains (colored symbols indicate specific C. albicans genetic background). Statistical significance between diploid and tetraploids was tested using Wilcoxon matched-pairs signed rank test and p values are indicated. F) Lineage growth, G) Brood size, and H) Reproductive timing of sek-1 hosts infected with C. albicans diploid and tetraploid strains. Data and statistical analysis are the same as for (E).  Figure 4: Ploidy-specific differences across C. albicans genetic backgrounds in immunocompromised hosts. A) Survival curves for sek-1 C. elegans populations that are either uninfected (exposed to just an E. coli food source, grey) or when infected with diploid or tetraploid C. albicans strains from laboratory homozygous (pink), laboratory heterozygous (green), bloodstream (orange), or oral/vaginal (blue) genetic backgrounds. Statistical significance was tested using pairwise comparisons of diploid and tetraploid survival curves with Log-rank (Mantel-Cox) test and p-values are indicated and significant differences are highlighted in bold text. B) Box and whiskers plot of host lineage growth which represents the total population size (representing the number of F1 and F2 progeny) produced within 7 days from a single founder sek-1 C. elegans host infected with C. albicans. Boxes indicate the 25-75th quartiles with median indicated. Error bars are the normalized range of the data and circles indicate outliers. The mean and 95% CI of the uninfected control treatment are indicated by the grey dashed line and shaded grey box. Statistical significance between diploid and tetraploid strains was determined using Mann Whitney test and p-values are indicated and significant differences are highlighted in bold text. C) Total brood size and D) Average percentage of host progeny produced during days 1-3 of adulthood (normal reproductive timing) of sek-1 C. elegans infected with diploid and tetraploid C. albicans. Data and statistical analysis are the same as (B).  Figure 5: Ploidy-specific interactions between healthy and immunocompromised hosts. A) Relative impact on host survival from Day 7 survival data (C. albicans D7 survival/uninfected D7 survival for each host type) for all C. albicans strains (colored symbols indicate specific C. albicans genetic background) in healhty (N2) and immunocompromised hosts (sek-1). The mean and 95% CI of the uninfected control treatment are indicated by the grey dashed line and shaded box. Statistical significance between host genotypes was tested using Wilcoxon matched-pairs signed rank test and p values are indicated. B) Box and whiskers plot of relative host lineage growth, C) Brood size, and D) Reproductive timing between healthy. Data and statistical analysis are the same as (A). E) Relative impact of C. albicans ploidy on host survival. Relative virulence of diploid (solid lines) and tetraploid (dotted lines) C. albicans lab homozygous (pink), lab heterozygous (green), bloodstream (orange), and oral/vaginal (blue) genetic backgrounds in healthy (N2) and immunocompromised hosts (sek-1). Y-axis scale bar is the same as in (A). F) Relative host lineage growth, G) brood size, and H) reproductive timing between healthy and immunocompromised hosts across C. albicans genetic backgrounds. Symbols represent the mean value and error bars +/-SD. Y-axis scale bar is the same as in (B, C and D) respectively. Stastistical significance was tested by two-way ANOVA and 'interaction' p value indicated.