RNAi Screen Identifies Novel Regulators of RNP Granules in the Caenorhabditis elegans Germ Line

Abstract

Complexes of RNA and RNA binding proteins form large-scale supramolecular structures under many cellular contexts. In Caenorhabditis elegans, small germ granules are present in the germ line that share characteristics with liquid droplets that undergo phase transitions. In meiotically-arrested oocytes of middle-aged hermaphrodites, the germ granules appear to aggregate or condense into large assemblies of RNA-binding proteins and maternal mRNAs. Prior characterization of the assembly of large-scale RNP structures via candidate approaches has identified a small number of regulators of phase transitions in the C. elegans germ line; however, the assembly, function, and regulation of these large RNP assemblies remain incompletely understood. To identify genes that promote remodeling and assembly of large RNP granules in meiotically-arrested oocytes, we performed a targeted, functional RNAi screen and identified over 300 genes that regulate the assembly of the RNA-binding protein MEX-3 into large granules. Among the most common GO classes are several categories related to RNA biology, as well as novel categories such as cell cortex, ER, and chromosome segregation. We found that arrested oocytes that fail to localize MEX-3 into cortical granules display reduced oocyte quality, consistent with the idea that the larger RNP assemblies promote oocyte quality when fertilization is delayed. Interestingly, a relatively small number of genes overlap with the regulators of germ granule assembly during normal development, or with the regulators of solid RNP granules in cgh-1 oocytes, suggesting fundamental differences in the regulation of RNP granule phase transitions during meiotic arrest.

Keywords: C. elegans; RNP granule; aging; germ line; oocyte quality.

Figure 1

RNAi screen identifies positive regulators of RNP granule assembly in arrested oocytes. (A) Live imaging of fog-2;GFP::MEX-3 unmated female shows rapid assembly of GFP::MEX-3 granules begins soon after the L4 stage of development. By 7 hr post-L4, granules are prominent and enriched near the cortex and nuclear membrane, and the level of cytosolic GFP is decreased relative to 4 hr post-L4. (B) Cartoon of feeding RNAi screen. 1536 genes were screened and 319 genes were identified. (C) Representative images showing disruption of RNP granule assembly after knockdown of gene expression. Top row: GFP::MEX-3 distribution after RNAi in fog-2 background. The negative control was RNAi using an empty vector (left), and GFP::MEX-3 granules are prominent. RNP granules are disrupted, and levels of diffuse GFP::MEX-3 are increased after knockdown of five positive hits. Middle row: Anti-MEX-3 staining after RNAi in fog-2 worms reveals similar large granules in negative control (left) and similar disruption of endogenous RNP granules. Bottom row: Anti-MEX-3 staining in arrested oocytes of wild-type (left) and in mutants depleted of sperm validates the GFP::MEX-3 reporter results. GFP, Green fluorescent protein; RNAi, RNA interference; RNP, ribonucleoprotein.

Figure 2

Disruption of RNP granule assembly correlates with decreased oocyte quality. (A) Blue bars show > 90% embryos are viable after RNAi of seven positive hits in wild-type worms, suggesting these genes are reasonable targets for analysis in arrested oocytes. For five of the seven genes, the percent of viable embryos significantly decreased when arrested oocytes did not assemble GFP::MEX-3 granules (green bars), as compared to worms where granules were detected (red bars). The percent embryonic viability is plotted as mean ± SEM; * P < 0.05 by Fisher’s exact test or Chi-square test (for W03C9.5, K08F4.1, and dnj-17). A minimum of 254 oocytes from a minimum of 11 females were analyzed for each RNAi target. (B) CGH-1 and GLH-1 fail to assemble into cortical granules after RNAi of a subset of the five genes. Note CGH-1 granules (green) are detected after RNAi of K08F4.1, but not str-82 or dnj-17. Large GLH-1 granules (red) are not detected after RNAi of any of the three genes shown. Quantitative data for all five genes are summarized in Table S2. GFP, Green fluorescent protein; RNAi, RNA interference; RNP, ribonucleoprotein.

Figure 3

RNP granule components include many proteins with IDRs. (A) Graphs showing disorder tendencies of sequences along each protein [calculated using IUPRED and long disorder parameters; http://iupred.enzim.hu/ (Dosztanyi et al. 2005)]. Scores > 0.5 indicate disorder. MEX-3 and GLH-1 have large stretches of IDRs; in contrast, CGH-1 is a mostly ordered protein. (B) Overrepresentation of disorder-inducing amino acids in subset of nine RNP granule proteins. Percentages highlighted in yellow are overrepresented (Ferron et al. 2006). Seven of nine proteins contain IDRs as identified by IUPRED, and eight of nine have a low complexity region identified by the SEG server. Gly, glycine; IDRs, intrinsically disordered regions; Pro, proline; RNP, ribonucleoprotein; Ser, serine.

Figure 4

Role of IDR proteins in regulating oocyte RNP granules. (A) MEG-3 appears to promote the assembly of RNP granules in oocytes; however, LAF-1 does not appear to be required. No significant differences were observed for mbk-2 or pptr-1/-2 in terms of the percent of germ lines with RNP granules. A minimum of 22 germ lines were examined for each gene; * P < 0.05 by Fisher’s exact test. (B) Live images of GFP::MEX-3 after RNAi of IDR proteins in a fog-2 background. Images on the left are midfocal views of proximal germ lines, and images on the right are cortical views of the same germ lines. MEG-3 appears to promote GFP::MEX-3 granule assembly in arrested oocytes. The phenotype of few RNP granules and high levels of diffuse GFP after pptr-1/-2 RNAi was not penetrant; however, it was never observed in the vector control. In contrast to MEG-3, LAF-1 may function in an opposite manner and repress the assembly of large granules in arrested oocytes. RNAi of laf-1 resulted in a 20% increase in the percent of germ lines with large RNP granules (> 5 μm) enriched at the cortex; compare laf-1 cortical view to vector control cortical view. Image J was used for quantitation (data not shown). GFP, green fluorescent protein; IDRs, intrinsically disordered regions; RNP, ribonucleoprotein.