Polymorphisms in cis-regulatory regions include copy number variations in potential transcription factor binding sites

Genetic variants at the multigene switch locus would be strong candidates to control mouth-form plasticity. We found a total of 41 single nucleotide polymorphisms (SNPs) within the 30-kb region spanning the multigene locus between RSA076 and RSC011. However, within the coding region of the four mouth-form associated genes, only a single nonsynonymous SNP was identified. This is found within nag-2 and causes a Phe415Ile change. Using CRISPR/Cas-9 engineering, we introduced the RSA076 parental nucleotide into the RSC011 genetic background. Two independent lines carrying this substitution (tu1489, tu1490) did not show any change in the highly St phenotype, dismissing a role for this substitution in controlling mouth-form variation (S2 Fig and S1 Table). All other SNPs between the parental strains either are in intergenic or intronic regions or represent synonymous changes in genes of the multigene locus. Therefore, we focused on potential cis-regulatory variation as numerous studies have shown the involvement of cis-regulatory elements in adaptive divergence, particularly in promoter and enhancer regions of developmental control genes [11–16].

The highest number of SNPs between RSA076 and RSC011 are in the upstream region and the first intron of eud-1 (S2 Table). Specifically, 5 SNPs in the upstream region and 1 SNP in intron 1 of eud-1 are shared between related strains of RSA076 and RSC011 (Figs 2B and S3). In addition, we detected a 32-bp element that contains sequence similarity to a potential Forkhead transcription factor binding site (hereafter, Forkhead binding site (FBS)) in the upstream region of eud-1 (Fig 2B). Interestingly, we observed copy number variation (CNV) of this 32-bp element between strains. RSA076 has 2 copies of this element, whereas RSC011 has only a single copy (Fig 2B). These SNPs and the CNV might be involved in the regulation of eud-1, as eud-1 expression in RSA076 is 40% higher, consistent with its role in the specification of the Eu morph (Fig 2C).

Systematic swapping experiments through CRISPR engineering identify cis-regulatory and intronic variants at the eud-1 locus to control plasticity

To determine a potential role for the identified SNPs, we performed systematic swapping experiments using CRISPR/Cas-9 engineering. Specifically, we introduced substitutions in the Eu parental background RSA076 with the sequence variants of the RSC011 strain. However, replacing any of the SNPs did not change the mouth-form ratio in the resulting lines (Figs 2D, S4, and S5 and S3 Table). In contrast, manipulating CNV of the FBS revealed strong changes in mouth-form ratios. When we deleted one of the 2 copies of the FBS in RSA076, both resulting lines (tu1590, tu1621) showed a reduction of the predatory morph to 75% to 90% Eu animals (Fig 2D). Elimination of the second copy (tu1866, tu1867) resulted in a further reduction of the Eu form (Fig 2D). Strikingly, however, if we swapped the A-G SNP in intron 1 in the presence of deletions of the FBS, we observed even more drastic changes in mouth-form ratios. First, the RSA076(tu1590) allele that harbors only 1 copy of the FBS showed a strong reduction of the Eu mouth form (40% to 55% Eu) after introducing the A-G swap (tu1868, tu1869 in Fig 2D). Second, when we introduced the A-G swap in a line that has both FBS copies deleted (tu1870, tu1871), the mouth-form ratio is below 20% Eu, similar to RSC011 animals (Fig 2D). Thus, CNV of the FBS in the upstream region of eud-1 regulates mouth-form plasticity synergistically with a single SNP in the first intron, and different combinations of these regulatory elements can alter the mouth-form ratio between 10% and 100% Eu (Fig 2D and S4 Table). Importantly, the mutant lines tu1868 and tu1869, while closely mimicking the RSC011 mouth-form ratio, do not completely phenocopy this strain. As such, this finding may indicate the involvement of other strain-specific background effects in the manifestation of the phenotype.

Sequence comparisons of the eud-1 cis-regulatory region in a broader diversity of P. pacificus strains provided further support for the rapid evolution of both identified elements (S6 Fig). Notably, strains of other P. pacificus clades, including the “wild-type” PS312, have 3 FBS copies and are also preferentially Eu (S6 Fig). Additionally, further sequence alignment of the FBS element revealed the existence of a similar sequence in the first intron of eud-1 (Fig 3A). This additional element is located in the middle of intron 1 but is identical in the RSA076 and RSC011 strains. We used CRISPR/Cas-9 technology to manipulate this element. While small deletions have little to no effect on mouth-form plasticity, a 4-bp insertion already shifts the mouth form substantially towards the St morph (Fig 3B). Subsequently, we were able to generate 2 independent 31-bp deletions that completely eliminate the sequence homologous to FBS. Both of these lines show 0% Eu animals, and all worms develop the St morph (tu1905, tu1906) (Fig 3B and S5 Table). Thus, the first intron of eud-1 contains an additional regulatory element that when eliminated results in an all-St phenotype similar to eud-1 knockouts [5]. This intronic element shows no sequence variation between wild isolates of P. pacificus, but strong sequence divergence in the sister species P. exspectatus, which is strongly St (S7 Fig) [17,18].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Functional analysis through CRISPR/Cas-9 engineering and evolutionary divergence of eud-1 cis-regulatory elements.

(A) Sequence comparison of the transcription factor binding site in the eud-1 promoter reveals a related sequence in intron 1. This sequence element is identical between RSA076 and RSC011. (B) Mouth-form ratios of various CRISPR-induced mutants introducing deletions of this intronic element in the RSA076 genetic background. A complete deletion results in a 100% St (0% Eu) phenotype. Parental phenotypes are indicated in purple (RSA076) and green (RSC011), respectively. (C) Quantitative PCR experiments of selected mutant strains exhibit lower eud-1 expression correlating with the preferential St mouth form. (D) Nucleotide diversity and F_st data of the strains from CK and NB based on population-scale whole-genome sequencing. CK strains with the RSC011 haplotype show low diversity at the QTL peak on the X chromosome. Nucleotide diversity of the other chromosomes are shown in S8 Fig. For detailed information, see S2 Data. CK, Coteau Kerveguen; Eu, eurystomatous; NB, Nez du Boeuf; QTL, quantitative-trait-locus; St, stenostomatous.

https://doi.org/10.1371/journal.pbio.3002270.g003

Cis-regulatory variants affect eud-1 expression

Next, we used quantitative reverse transcription PCR (qRT-PCR) experiments to provide direct evidence that the engineered CRISPR lines shifting the mouth-form ratio from Eu to St are indeed affecting eud-1 expression. For this, we measured eud-1 expression relative to the normalized eud-1 expression of the parental RSA076 strain. We found that lines with the deletion of both FBS and the intronic swap (tu1870, tu1871) show strongly reduced eud-1 expression, similar to RSC011 (Fig 3C). This was further validated in the line with the 31-bp deletion in intron 1, which also resulted in a strong reduction of eud-1 expression (Fig 3C). These findings indicate that the cis-regulatory variants in natural isolates of P. pacificus affect eud-1 expression and thereby influence mouth-form execution.

Population-scale whole-genome sequencing suggests the Eu pattern to be ancestral

Finally, we tested if the QTL shows any evidence for selection and wanted to determine the direction of evolutionary change in mouth-form ratio. For that, we employed available population-scale whole-genome sequencing data for the strains used in this study [8] to compare their genotype with mouth-form ratios (S8 Fig). Specifically, we compared all 10 available strains from CK, where RSC011 was isolated, with representative strains from NB, the origin of RSA076 (S8 Fig). Strikingly, we identified 4 of the 10 strains from CK that had a preferentially St mouth form and shared the same haplotype at the eud-1 locus with RSC011 (S8A Fig). In contrast, the remaining isolates exhibited either the RSA076-type variants at the eud-1 locus or a mixed pattern (RSC010, RSC173) and were all preferentially Eu (S8B Fig). No such variation was seen in the strains from NB with all strains being preferentially Eu and having the RSA076 haplotype (S8B Fig). These results indicate that the RSA076 Eu pattern is ancestral in P. pacificus clade B and that the RSC011 St phenotype has recently evolved. Moreover, the CK strains with the RSC011 haplotype at the QTL peak showed a remarkably low diversity suggesting that this haplotype was introduced very recently into the population (Figs 3D and S8C). To distinguish if these patterns are shaped by neutrality or natural selection will require future sampling efforts with a higher temporal resolution.

Parallel evolution shapes natural variation of mouth-form plasticity in P. pacificus

Given the central role of the eud-1 and its neighboring genes in the mouth-form decision [5,6], we wondered if this large locus represents a hotspot for natural variation in phenotypic plasticity. Thus, we tested for evidence of parallel evolution at this locus by repeating RIL and QTL analysis using 2 more distantly related P. pacificus strains. Specifically, we performed an RIL experiment between the highly Eu PS312 strain from California, from which the reference genome is derived, and the highly St strain RSB020 that belongs to a different clade [8]. Indeed, our analysis again identified a single major QTL on the X chromosome that covers the eud-1 locus (S9 Fig). Given the large phylogenetic distance between PS312 and RSB020, many more SNPs and other variants are found in the eud-1 regulatory region. Taken together, independent RIL and QTL analyses of genetically diverse strains indicate a major role for the multigene locus and eud-1 in natural variation of mouth-form plasticity.

With these studies, we have combined RIL- and QTL-based population genetic analysis with CRISPR-mediated experimentation to study mouth-form plasticity evolution. Our observations strongly support a role for distinct combinations of cis-regulatory elements at a single developmental switch gene controlling mouth-form plasticity. Through these studies, we (i) demonstrate the existence of natural variation associated with phenotypic plasticity; (ii) identify its molecular architecture; and (iii) establish population differentiation coupled to this regulatory mechanism.