Maternal Ribosomes Are Sufficient for Tissue Diversification during Embryonic Development in C. elegans

Abstract

Caenorhabditis elegans provides an amenable system to explore whether newly composed ribosomes are required to progress through development. Despite the complex pattern of tissues that are formed during embryonic development, we found that null homozygotes lacking any of the five different ribosomal proteins (RPs) can produce fully functional first-stage larvae, with similar developmental competence seen upon complete deletion of the multi-copy ribosomal RNA locus. These animals, relying on maternal but not zygotic contribution of ribosomal components, are capable of completing embryogenesis. In the absence of new ribosomal components, the resulting animals are arrested before progression from the first larval stage and fail in two assays for postembryonic plasticity of neuronal structure. Mosaic analyses of larvae that are a mixture of ribosome-competent and non-competent cells suggest a global regulatory mechanism in which ribosomal insufficiency in a subset of cells triggers organism-wide growth arrest.

Keywords: Caenorhabditis elegans; cell non-autonomous; checkpoint; embryo; maternal; rDNA locus; ribosomes.

Figure 1:. Zygotic ribosomal RNA is transcribed but functional ribosomes remain predominantly maternal A.

C. elegans strain My14 harbors a single nucleotide variant in the rrn-3.1 gene, encoding for 28S rRNA. Genomic DNA sequencing of the canonical laboratory strain N2 (Brenner, 1974) shows only the reference allele at this site, while sequencing of the My14 strain reveals only copies containing the variant allele (Thompson et al., 2013). B. RNA-seq of total RNA, monosome and polysome pools from cross progeny of either My14 males and N2 females (feminized strain: tra-2(q122)) or N2 males (mIn1) and My14 hermaphrodites (Goodwin et al., 1993). Total RNA, genomic DNA and ribosomes from cross progeny L1 or L4 larvae were extracted and uniquely mapping reads to each variant were counted. C. Number of unique reads that map to each allele is shown for total RNA at L1 and L4 stages as well as monosome and polysome fractions. Comparing My14 rRNA read counts to all rRNA provides a measure of zygotic rRNA representation for total RNA, polysomes, and monosomes. At hatching [L1²*], lower zygotic representation in monosome and polysome fractions is significant as compared to total RNA (p-values at hatching for monosome and polysome fractions p<1e-06, p = 0.0004 respectively for cross progeny of My14 males and N2 females. This significant difference persisted at 3 hours after hatching [L1¹*], (p = 0.016 and p = 0.00018 for the monosome and polysome fractions respectively for cross progeny of My14 males and N2 females, right). L1 progeny of the reciprocal cross of N2 (mIn1) sperm and My14 oocytes at hatching [L1²*] and 3 hours after hatching [L1¹*], shows significantly lower zygotic N2 variant representation in monosome and polysome fractions as compared to total RNA. (p<1e-6)

Figure 2:. Complete embryonic development and partial post-embryonic development for ribosomal protein null mutants

A. Mutated ribosomal proteins (RPs) with corresponding C. elegans gene names are shown in color on the human 80S ribosome structure (adapted from PDB ID: 4ug0) (Khatter et al., 2015). The three colors depict the extent of conservation of ribosomal proteins. Group-1 RPs have homologs in archaebacteria and eubacteria. Group-2 RPs have homologs only in archaebacteria, and Group-3 is unique to eukaryotes (Wool et al., 1995). Ribosomal RNA (rRNA) is labeled in blue. B. Representative diagram for one of the mutations: rpl-5(0) depicts how the strain is maintained. The heterozygous mutation is maintained with the balancer mIn1 [dpy-10(e128)] (II). Representative Sanger sequencing traces in the mutated region is shown. All ribosomal protein mutations were made via CRISPR-Cas9 gene editing and via insertion of tandem early stop (TAA) mutations. All mutations were verified by PCR amplification of the target region and Sanger sequencing. C. The body length of the newly hatched homozygous larvae matched the length of heterozygous and wild-type siblings in the same genetic background at hatching. Homozygous larvae failed to grow over time unlike its heterozygous or wild-type siblings. D. Comparative differential interference microscopy images of pharynx for wild-type and rpl-5(0) larvae: different z-sections (top and bottom) of the pharynx of the wild-type larvae (left) and the representative rpl-5(0) homozygous larvae are shown (right). Scale bar, 10 μm. E. Diagram depicts mesoblast (M) cell and gonadal primordium cellular divisions during L1 stage (adapted from Kasuga et al., 2013). F. Gonadal primordium images of three-day-old rpl-5(0) in two different z-sections shows at least a single cellular division from the 4 cell gonadal primordium present canonically at hatching whereas wild-type starved L1 lacks post-embryonic cell division in gonadal primordium (left images). Brightness and contrast were individually adjusted. Right graph shows the number of cells in the gonadal primordium 72 hours after hatching at 16°C. For rpl-33 and rpl-5, gonadal primordium cells were also counted after 10 days. For all homozygous mutants, gonadal cells divides (N=15 for each mutant), whereas no cell division occurs in the gonadal primordium of starved L1 (N=15; also see Fukuyama et al., 2006). Scale bar, 10 μm. G. GFP-based imaging of the mesodermal M lineage in arrested rpl-5(0) larvae. hlh-8::gfp labels the M cell and its descendants (Harfe et al., 1998). We observed that M lineage is arrested at a single-cell stage with no evidence of postembryonic divisions in larvae after 72 hours at 16°C.

Figure 3:. Ribosomal protein null mutants show normal development of embryonically born touch receptor neurons, but display impaired axonal tiling and regrowth after injury.

A. Confocal images of ALM (yellow arrowhead) and PLM (blue arrowhead) in young L1 larvae. Left panels: Normal positions and axon trajectory of ALM and PLM. PLM anterior axons overlap and extend beyond ALM soma (*), which are indistinguishable between wild-type and rpl-33(0). Middle panels: Positions of ALM and PLM in the same larva 6 h later. In wild-type, slow growth of PLM axons results in separation from ALM soma (process of tiling). In rpl-33(0) mutants, PLM axons remain overlapping with ALM soma. Right panels: 48 hours post-hatching, wild-type larvae reach L4 stage, PLM axons terminate posterior to ALM soma, but rpl-33(0) mutants remained as young larvae, and PLM axons remain overlapping with ALM soma. B. Quantitative analysis of tiling defects of rpl-33(0). Larvae were mounted for observation (defined as time 0) from <6 hours post-hatching, and re-mounted 6 hours later. C. Confocal images of AVM (green arrow) and PVM (orange arrow) in animals 48 hours post-hatching. Note that wild-type larvae reach L4 stage, whereas rpl-33(0) mutants remain as young larvae. Graph shows quantification of the phenotypes observed. N=38 for wild-type; N=65 for rpl-33(0). D. PLM axons in ribosomal mutant L1 larvae do not regenerate. Representative images show regrowth of PLM in wild-type, but not in rpl-33(0). White arrows mark site of axotomy. The red arrow indicates the distal fragment of axotomized axon in rpl-33(0) mutants. Graph shows normalized PLM regrowth 24 h post-axotomy.

Figure 4:. Maternal ribosomes are active in ribosomal protein null mutants

A. Arrested rpl-5 null larvae express heat-shock inducible hsp-16 GFP. rpl-5(0), hsp16::GFP were kept at 16°C for 48, 72 and 120 hours (top, middle and bottom image panels respectively) after being laid as embryos, then heat-shocked at 34°C for 3 hours, followed by 3 hours incubation at 16°C before imaging (top diagram). First row of each image depicts the sibling larva in the absence of heat-shock (HS–) and second row shows the heat-shocked larva (HS+). Differential interference contrast (left), and GFP (right) images were captured sequentially with identical settings. B. Metagene analysis of normalized ribosome footprint start site density in stop and start codon proximity in wild-type and rpl null larvae. 5’ end of mapped reads were plotted on their position relative to start and stop codons on protein coding transcripts in wild-type and rpl null larvae.

Figure 5:. Genetic mosaics with rpl-5 null and wild-type cells show normal embryonic development with a cell non-autonomous larval growth arrest.

A. gpr-1 overexpression (Artiles et al. 2018 co-submitted) led to a mosaic embryo with (i) rpl-5(0) in AB lineage and wild-type cells in P1 lineage and (i) wild-type cells in AB lineage and rpl-5(0) cells in P1 lineage because of non-canonical mitosis. B. Pharyngeal musculature contains both AB and P1 lineage cells with anterior pharynx being predominantly AB-derived and posterior being predominantly P1-derived. GFP+ wild-type pharyngeal cells are P1-derived in the top image and AB-derived in the bottom image. GFP–pharynx cells are rpl-5(0). Both of these mosaic larvae were imaged 72 hours post egg laying at 16°C. Both have the typical anatomy and size of L1 arrested larvae. Scale bar, 50 μm. C. Differential interference contrast image of the pharynx of a mosaic larva where the P1 lineage is wild-type and AB lineage is rpl-5(0)). D. Left diagram: C. elegans body muscle tissue is predominantly P1-derived with the exception of a single AB-derived cell (adapted from (Sulston and Horvitz, 1977, Sulston et al., 1983, Moerman & Fire. C. elegans II, Chapter 16).). Right section, top image: Confocal GFP (unc-54::GFP, present in both lineages) and mCherry (myo2:mCherry:let858[3’UTR], only present in the AB lineage) image of a 48 hour old arrested mosaic larva whose P1 lineage is rpl-5(0) and AB lineage is wild-type. GFP in this animal labels muscle structure, with the heavy chain myosin gene unc-54 endogenously tagged with GFP at the C terminus. mCherry labels pharyngeal and body muscle cells from the AB lineage. Bottom left: In this image GFP and nuclear mCherry coincide for the single wild-type AB-derived body muscle cell. This cell doesn’t grow significantly over 3 days in the (AB: wild-type, P1: rpl-5(0)) mosaic larva; in contrast to a wild-type animal wherein the cell would have grown > 5 fold in length in a wild-type animal. Bottom right: A more detailed image of the muscle arrangement surrounding the nuclear mCherry labeled wild-type AB lineage cell. No defects in muscle structure were observed.

Figure 6:. A shared gene expression response in rpl-5 and rpl-33 null homozygous larvae

A. Comparison of RNA abundance in wild type and RP mutant larvae. Read counts per gene in wild-type and rpl null conditions are shown. Each point on each graph indicates counts for a single gene in two conditions. Points along a single diagonal would indicate comparable read counts in the two samples; below-diagonal and above-diagonal points are indicative of genes which are outliers in their relative expression between the two conditions. Top two graphs: RNA-seq counts per million reads (RPM) compare rpl-5(0) on the y axis to wild-type L1 larvae on the x axis. Bottom two graphs highlight the same for rpl-33(0) on the y axis compared to wild-type L1 larvae on the x axis. Genes that were at least 6 fold up (red) or down (blue) in a mutant RP (rpl-5(0) on the left, rpl-33(0) on the right) compared to wild-type larvae are labeled (total minimum counts= 5 counts per million reads). B. Gene Ontology analysis of differentially expressed transcripts. rpl null mutant larvae expression from two different datasets (rpl-5(0) and rpl-33(0)) were compared to wild-type larvae data using R limma package. RNA-seq libraries of each mutant and wild-type larval population were prepared and sequenced at separate times to ensure full biological replication. Genes that were significantly over or under expressed (>6-fold or <1/6-fold; (adjusted p-value cut-off = 0.002) were analyzed for significant gene ontology (GO) term enrichment (adjusted p-value cut-off = 0.05) using FuncAssociate 3.0 (Berriz et al., 2009). Selected categories are shown with a violin plot where the distribution of log2 fold change of each GO category is plotted. Points in each distribution represent the log2 fold change of each differentially expressed gene in the category. Complete list is in Table S2.

Figure 7:. Embryonic development can be completed in the absence of all functional copies of 18S, 5.8S, and 28S rDNA in the zygote.

A. The chromosomal locus containing all functional copies of 18S, 5.8S and 28S rDNA was deleted in two independent strains. The diagrammed Cas9-mediated break is repaired using a homologous recombination template that juxtaposes the desired 5’-deletion breakpoint to a set of putative subtelomeric sequences (annotated sequence from end of chromosome I). B. Whole genome sequencing data from homozygous rDNA-deletion larvae and wild-type worms were generated and plotted using the R-Sushi Package (Phanstiel et al., 2014). The end of chromosome I where the rDNA locus is located is marked with dashed lines. No copies of the 18S/28S/5.8 S locus remain in homozygotes for both ccDf2620 and ccDf2621. C. Differential interference contrast microscopy images of rRNA deletion homozygotes in whole larva.