Evolution of New cis-Regulatory Motifs Required for Cell-Specific Gene Expression in Caenorhabditis

Abstract

Patterning of C. elegans vulval cell fates relies on inductive signaling. In this induction event, a single cell, the gonadal anchor cell, secretes LIN-3/EGF and induces three out of six competent precursor cells to acquire a vulval fate. We previously showed that this developmental system is robust to a four-fold variation in lin-3/EGF genetic dose. Here using single-molecule FISH, we find that the mean level of expression of lin-3 in the anchor cell is remarkably conserved. No change in lin-3 expression level could be detected among C. elegans wild isolates and only a low level of change-less than 30%-in the Caenorhabditis genus and in Oscheius tipulae. In C. elegans, lin-3 expression in the anchor cell is known to require three transcription factor binding sites, specifically two E-boxes and a nuclear-hormone-receptor (NHR) binding site. Mutation of any of these three elements in C. elegans results in a dramatic decrease in lin-3 expression. Yet only a single E-box is found in the Drosophilae supergroup of Caenorhabditis species, including C. angaria, while the NHR-binding site likely only evolved at the base of the Elegans group. We find that a transgene from C. angaria bearing a single E-box is sufficient for normal expression in C. elegans. Even a short 58 bp cis-regulatory fragment from C. angaria with this single E-box is able to replace the three transcription factor binding sites at the endogenous C. elegans lin-3 locus, resulting in the wild-type expression level. Thus, regulatory evolution occurring in cis within a 58 bp lin-3 fragment, results in a strict requirement for the NHR binding site and a second E-box in C. elegans. This single-cell, single-molecule, quantitative and functional evo-devo study demonstrates that conserved expression levels can hide extensive change in cis-regulatory site requirements and highlights the evolution of new cis-regulatory elements required for cell-specific gene expression.

Fig 1. lin-3 expression level in the anchor cell is overall conserved in different Caenorhabditis species.

(A) Cartoon depicting the position of the anchor cell (AC) and Pn.p cells at the time of induction. Three Pn.p cells (P5.p –P7.p) are induced upon LIN-3 secretion. (B-E) smFISH using a lin-3 probe in C. elegans N2 (data from [2]) (B), C. briggsae AF16 (C), C. angaria RGD1 (D) and O. tipulae CEW1 (E). Red arrow marks the position of the anchor cell. (F) Quantification of the number of spots detected in the anchor cell of these species at the time of induction (n = 32* animals for N2, n = 24 for AF16, n = 26 for RGD1 and n = 22 for CEW1). *: these include 20 animals from [2] (see Fig 6 for an independent dataset with a similar result). The difference between C. elegans and C. briggsae is not statistically significant with a Tukey’s multiple comparison test (P value = 0.99), whereas the difference between C. elegans and C. angaria, or C. elegans and O. tipulae is significant (P values < 0.0002).

Fig 2. lin-3 activity in vulval induction is conserved in Caenorhabditis species.

(A) Comparative lin-3 RNAi effect on vulva induction in C. elegans, C. briggsae and C. remanei. Tables show graphically the observed defects in vulval cell fate pattern after scoring at least 100 nematodes. Every column is a distinct Pn.p cell (3 to 8) and 1° fate is depicted in blue, 2° fate is depicted in red and 3° fate in yellow. Half fates represent cases where the Pn.p daughter cells adopt different cell fates after the first cell division. The defects are ordered based on their consequence on vulval induction index, from high index to low. (B) Treatment with the MEK inhibitor U0126 decreases vulval induction in C. elegans (n = 15 for DMSO control and 10 μM U0126 treatment), C. angaria (n = 32 for control, n = 27 for 150 μM U0126 treatment) and C. afra (n = 100 for control, n = 30 for 150 μM treatment), as measured by the vulval induction index (average number of induced Pn.p cells at the population, wild-type index = 3). In all cases P<0.0001 with a Mann Whitney test.

Fig 3. Evolution in the cis-regulatory elements necessary for lin-3 expression in C. elegans.

Distribution of cis-regulatory elements (the regulatory triplet) in different species. 300 bp upstream of the ATG of the vulval isoform of lin-3 are shown (and up to 2 kb for C. sp. 1, where no upstream ATG is found). Orange depicts the E-box and purple the NHR site. “>” or “<” show orientation of the regulatory site.

Fig 4. A single E-box in the Cel-lin-3 CRM is not sufficient for lin-3 expression in the anchor cell of C. elegans.

(A) New cis-regulatory lin-3 alleles with deleted E-box_L and NHR or NHR and E-box_R. (B) Quantification of vulval induction in these new mutants. Note the complete absence of any induction in the recovered lin-3 alleles (n>30). Scorings of lin-3(1417) animals are the same as those reported in Fig 5 and are used here to indicate that this mutation leads to vulval hypo-induction rather than no induction at all. (C-D) smFISH in lin-3(mf72) (C) and N2 (D) animals. Green spots correspond to lin-3 transcripts and red spots to lag-2 that is used as an anchor cell marker. Blue is DAPI staining of nuclei. Note the absence of lin-3 expression in the anchor cell in the lin-3(mf72) mutant animal. Absence of lin-3 signal in the anchor cell was also confirmed for the other lin-3 alleles.

Fig 5. Evolution in lin-3 cis-element requirement does not result from a modified trans environment.

(A) Transcriptional lin-3-CRM::Cherry fusions from different Caenorhabditis species are all expressed in the anchor cell of C. elegans. (B) A Can-lin-3 single-copy transgene is expressed in the C. elegans anchor cell. smFISH detection of Can-lin-3 in C. elegans strain harbouring a single extra-copy of Can-lin-3. Green spots correspond to lin-3 expression, while blue is DAPI staining. (C) A single extra-copy of Can-lin-3 fully rescues vulval induction of the Cel-lin-3(e1417) mutant to wild-type levels, both when homozygous (two copies, n = 100) or when hemizygous (one copy, n = 30). See the corresponding brood size rescue results in S8 Fig.

Fig 6. A 58 bp C. angaria fragment with a single E-box is able to replace the C. elegans regulatory triplet.

(A) Expression of a Can-lin-3 fragment in C. elegans containing the Can-lin-3 CRM, coding sequences and 3’ UTR leads to Can-lin-3 expression in the anchor cell, as detected by FISH. (B) Expression of the same fragment with a mutated E-box in the CRM results in loss of anchor cell expression. (C) A chimeric CRM that is mostly C. elegans apart from a 58 bp region around the regulatory triplet that is taken from C. angaria is also expressed in the anchor cell of C. elegans. (A) and (C) are using classical transgenesis in multicopy arrays experiments. (D) Over-expression of the Can-lin-3(+) fragment in C. elegans causes an increase in vulval induction, but not if the Can-lin-3 fragment with a mutated E-box is used. (E) Summary of the compensatory changes in cis to the C. angaria lin-3 locus allowing lin-3 expression in the anchor cell. C. elegans sequences are depicted in green and C. angaria sequences in blue. Orange box corresponds to the E-box. (F) Modification at the Cel-lin-3 endogenous locus, replacing the regulatory triplet with 58 bp from C. angaria containing a single canonical E-box with its genomic context (mf91 and mf92) or with 7 bp modified to its right (mf95 and mf112). The violin plots show the number of lin-3 mRNAs spots quantified in the anchor cell in the CRISPR replacements compared to N2 and lin-3(e1417) which are used as control strains. The whisker plot is superimposed in red. The average number of induced VPCs is shown below with the number of scored animals being 130, 159, 111, 42, and 35 from left to right.