Loss of endu-2 causes a temperature-dependent Mrt phenotype

Freshly outcrossed endu-2(lf) mutants were phenotypically similar to wild type animals in the first few generations, except displaying slightly reduced germline proliferation, an egg-laying defect (Egl) due to abnormal development of the vulva, and a moderate reduction in adult lifespan at 20 °C (Supplementary Fig. 1). Long-term strain maintenance at 20 °C was difficult, since the endu-2(lf) mutants showed gradually increased sterility that, however, could be reset by additional outcrosses with wild type animals. We examined the fertility of two endu-2 alleles tm4977 and by188 over generations after four additional outcrosses. tm4977 allele has a 620 bp deletion starting from the 5′UTR to the end of the third intron, while the 20 bp deletion in by188 causes an early stop codon in the first exon. Therefore, these two alleles are probably null mutants. Both tm4977 and by188 alleles became sterile at 20 °C after about 15–20 generations (Fig. 1a, Supplementary Fig. 2a). Since we did not observe a significant reduction of the number of germ nuclei in the mitotic zone across generations, sterility was not the consequence of declining germline proliferation (Supplementary Fig. 2b). This result also indicates that the Mrt phenotype is probably not caused by altered activity of CTPS-1, an ENDU-2 controlled cytidine triphosphate synthase playing an essential role in germline proliferation¹¹. Instead, endu-2 day 1 adult animals in the generations with highly penetrant sterile phenotype showed pleiotropic defects in the germline. The most prominent defects were abnormal cell death (38%, n = 105) and increased number of apoptotic corpses (31% with ≥2 corpses per gonad arm, n = 105) (Fig. 1b). We also observed prolonged spermatogenesis 24 h after mid-L4 stage (11%, n = 105), whereas wild type animals had completely switched spermatogenesis to oogenesis. Furthermore, endomitosis occurred in the mitotic region (14%, n = 25) (Fig. 1b). In the generations displaying strong sterile phenotype, endu-2(lf) additionally showed high incidence of male progeny (Him, 14%, n = 218) and increased occurrence of other phenotypes (Rol, Dpy, Sma phenotypes, in total 10%, n = 218). The latter were probably caused by elevated somatic mutation rate, since these phenotypes were not heritable. The sterile phenotype was 100% penetrant at 25 °C already at generation 6–10, but could not be detected at 15 °C (Supplementary Fig. 2c and d). Notably, the highly penetrate sterility could be fully reversed within several generations by transferring the animals to 15 °C (Fig. 1a). Taken together, these results suggest an essential role of ENDU-2 in preserving germline immortality at elevated temperature and the existence of additional mechanisms to compensate for the loss of endu-2 at 15 °C.

a Upper graph: Brood size of endu-2(tm4977), after 4× initial backcrosses with wild type (G0), decreases over generations at 20 °C. Lower graph: The reproduction defect of endu-2(tm4977) animals is reversible by shifting the animals to 15 °C. To obtain this graph, adult animals from G6 that had been maintained at 25 °C were shifted back to 15 °C and counted as G0. As the numbers of generation to reach 100% sterility at 20 °C varied from 10 to 20 among the four biological replicates, only one represented replicate with n = 15 animals for each generation is shown. The data are mean ± SEM. b The sterile endu-2(tm4977) animals display multiple defects in germline mitosis and meiosis. (I–IV) DIC images of wild type and sterile endu-2(tm4977) animals 24 h after mid-L4 stage. The white arrows point to (II) increased number of germline apoptotic corpses; (III) empty gonad due to germ cell death; (IV) prolonged spermatogenesis; (V) DAPI staining of germline proliferating zone of day-one wild type and (VI) sterile endu-2(tm4977) adults, respectively. White arrows in VI point to abnormally large chromosomes due to endomitosis. Scale bar 10 µm. N = 3 biological replicates.

ENDU-2 is a secreted protein

A previous report has suggested a wide-spread expression of endu-2 in somatic tissues¹⁰, but did not report ENDU-2 localization in the germline. To retest the expression pattern of ENDU-2, we generated several independent transcriptional and translational endu-2 reporters with EGFP as well as a CRISPR/Cas9 EGFP knock-in strain at the endogenous endu-2 locus (expression constructs are shown in Supplementary Fig. 3a). A transcriptional fusion reporter gene, harboring 4 kb of endu-2 upstream sequences, was expressed only in the intestine (Supplementary Fig. 3b). A translational reporter harboring the same upstream sequences and the entire genomic region of endu-2, fused to EGFP, rescued Egl, reduced germline proliferation, and short lifespan phenotypes (Supplementary Fig. 1a, c and d). The CRISPR/Cas9 EGFP knock-in strain did not display the Mrt phenotype (Supplementary Fig. 1e). These results suggest that ENDU-2::EGFP fusion proteins in the transgenic and the EGFP knock-in strains are functional. We observed ENDU-2::EGFP protein in the cytoplasm of intestine, somatic gonad, and coelomocytes in both endu-2::EGFP transgenic and the CRISPR/Cas9 EGFP knock-in strains (Fig. 2a and Supplementary Fig. 3c). Unlike suggested in the study of Ujisawa et al., we did not detect ENDU-2::EGFP in the musculature or in the neurons. Instead, we noticed extracellularly localized ENDU-2::EGFP, such as in the interspace between uterine wall and embryos (Fig. 2a and Supplementary Fig. 3c), suggesting that ENDU-2 may be a secreted protein. ENDU-2 indeed contains a predicted N-terminal (1–19 amino acid) secretion signal peptide for endoplasmic reticulum (ER) targeting, indicating that the protein is destined toward the secretory pathway. Fusion of this secretion signal peptide to the N-terminus of a EGFP reporter gene expressed exclusively in the neurons led to only weak expression level of EGFP protein in the neurons, but an EGFP signal in the extracellular space and the coelomocytes, cells that are known to endocytose secreted proteins from the pseudocoelic fluid (Fig. 2b). In addition, removing the N-terminal secretion signal peptide from ENDU-2 (\(\Delta\)_ssENDU-2::EGFP) resulted in strong localization of \(\Delta\)_ssENDU-2::EGFP only in the intestine but not in the coelomocytes (Supplementary Figs. 3e and 4a). These results together suggest that the secretion signal peptide composed of the first 19 amino acids of ENDU-2 is necessary and sufficient to trigger secretion of a protein. Strikingly, we noticed that another transgenic reporter expressing 3xFlag::ENDU-2::EGFP was expressed strongly in the intestinal and also weakly in some head neurons, muscle cells in the head region, and anal depressor muscle cells, corroborating the report from Ujisawa et al. (Supplementary Fig. 3d). We speculate that the N-terminal 3xFlag fusion may prevent secretion by impairing binding of the secretion signal peptide by signal recognition particle (SRP), thus allowing detection of the weak expression in the muscles and neurons. In addition, expression in the head neurons and muscle cells might possibly be controlled by promoter signals localized in the first intron since this was the only sequence absent in the transgene expressing \(\Delta\)_ssendu-2::EGFP (Supplementary Fig. 3a), in which we failed to observe muscular or neuronal localization. Moreover, when we expressed endu-2::EGFP selectively either in the neurons (unc-119 promoter), muscles (myo-3 promoter) or intestine (vha-6 promoter), ENDU-2::EGFP was always detected in the coelomocytes (Supplementary Fig. 4a), indicative for its secretion from these tissues. Notably, expressing endu-2::EGFP specifically in either neurons or muscles of heat-stressed animals resulted in ENDU-2::EGFP localization in the pharynx (Supplementary Fig. 4b), suggesting that either secretion or uptake of ENDU-2 in these tissues could be modulated by temperature.

a Fluorescence micrographs of transgenic endu-2(tm4977);byEx1814[endu-2P::endu-2::EGFP::_endu-23′UTR] animals. ENDU-2::EGFP is detected in the intestine, the somatic gonad, coelomocytes, and extracellular space between the uterus and embryos. N = 5 independent experiments. b Fusion of the 1–19 amino acids of ENDU-2 (SS_endu-2) to neuronal-specific expressed EGFP is sufficient for secretion of EGFP protein. Upper panels show localization of unc-119P::EGFP which is detected only in neuronal cells (strong) of animals and embryos. Lower panels are images showing localization of unc-119P::SS_endu-2::EGFP which is detected in the neurons (weak), coelomocytes and interspace between egg shell and embryos, indicative of efficient secretion of EGFP mediated by the secretion signal peptide of ENDU-2. Images are representative for more than 20 animals, analyzed by using a ×40 objective. c Fluorescence micrographs of GFP-antibody staining of endu-2(tm4977);byEx1375[endu-2P::endu-2::EGFP], endu-2(tm4977);byEx1449[endu-2P::Δ_ssendu-2::EGFP] and endu-2(tm4977);byEx1875[endu-2P::SS_sel-1::Δ_ssendu-2::EGFP] transgenic animals at 20 °C. ENDU-2::EGFP (n = 62, 100%) and SS_sel-1::Δ_ssENDU-2::EGFP (n = 21, 100%) but not Δ_ssENDU-2::EGFP (n = 26, 0%) is detected in the germline. N = 2 independent replicates. d smFISH staining reveals presence of endu-2 mRNA in the intestine (white arrow), but not in the germline. Blue: DAPI stained nuclear DNA. Red: smFISH probes stained mRNA. N = 4. Scale bar 10 µm for all images in this Figure.

Secretion of ENDU-2 from the soma to the gonad protects germline immortality

It is generally accepted that, in C. elegans, multi-copy extrachromosomal arrays are silenced and not expressed in the germline¹². However, we found that extrachromosomal endu-2::EGFP transgenes rescued the Mrt germline phenotype of endu-2(lf) animals (Fig. 3a), suggesting that somatic ENDU-2 might preserve germline immortality across tissue boundaries. Our hypothesis was that secreted ENDU-2 could be endocytosed by the gonad. To test this assumption, we first asked whether ENDU-2 protein could be detected in the germline. All of our endu-2::EGFP transgenic reporters displayed weak expression levels unless secretion was blocked. Therefore, we could only occasionally observe faint ENDU-2::EGFP in few oocytes (Supplementary Fig. 4c). Via GFP-antibody staining we detected ENDU-2::EGFP in punctate structures in the germline at 15, 20, and 25 °C (Fig. 2c and Supplementary Fig. 5). In contrast, Δ_ssENDU-2::EGFP was not detected in the germline despite of its high expression level (Fig. 2c and Supplementary Fig. 3F). In addition, fusing the secretion signal peptide of another secreted protein SEL-1¹³ to \(\Delta\)_ssENDU-2::EGFP reactivated secretion and its localization in the germline (Fig. 2c and Supplementary Fig. 3e). These observations indicate that secretion of ENDU-2 is indispensable for its gonadal localization. Moreover, we performed single molecular FISH (smFISH) staining to visualize endu-2 mRNA. endu-2 mRNA signals were visible predominantly in the intestine, but were clearly absent from the germline (Fig. 2d). Furthermore, sequencing of RNA extracted from isolated gonads also failed to detect endu-2 mRNA in the wild type gonad (Supplementary Data 3). Taken together, these results implicate that somatically expressed ENDU-2 protein is secreted and can be taken up by the gonad.

Next, we asked from which tissue ENDU-2 ensures germline immortality. Expressing ENDU-2 specifically in the intestine or the neurons, but not the muscle or somatic gonad, was sufficient to rescue the Mrt phenotype, indicating a non-cell-autonomous ENDU-2 signal from the neurons and intestine to the germline (Fig. 3b). In addition, expressing the secretion-deficient \(\Delta\)_ssENDU-2::EGFP failed to rescue the Mrt phenotype (Fig. 3c). Furthermore, SS_sel-1::\(\Delta\)_ssENDU-2::EGFP rescued the Mrt phenotype (Fig. 3c), suggesting that guiding ENDU-2 into ER-Golgi secretory pathway via a canonical secretion signal suffices for functions of ENDU-2 in the germline.

To directly test whether loss of somatic endu-2 expression is sufficient to induce a mortal germline, we performed endu-2 RNAi knock-down in a ppw-1 mutant background in which germline RNAi does not function¹⁴. Although neither wild type nor ppw-1 mutants upon endu-2 RNAi displayed a fully penetrant Mrt phenotype within 15 generation, the soma-specific endu-2 RNAi resulted in gradually reduced brood sizes over generations in both ppw-1 and wild type background, indicating that somatically expressed endu-2 mRNA is required for normal reproduction (Supplementary Fig. 6a). The failure to obtain a fully penetrant Mrt phenotype in this experiment may be due to the relative inefficacy of RNAi in the nervous system, a tissue that expresses endu-2 to ensure germline immortality.

ENDU-2 affects the oocytes to preserve germline immortality independent of the nuclear RNAi pathway

To know whether ENDU-2 affects oocytes or sperm to ensure germline immortality, we tested the parental contribution of ENDU-2 during sexual reproduction. For this purpose, we crossed endu-2(+/+) wild type parents with either males or hermaphrodites of endu-2(−/−) that had been grown at 25 °C for five to seven generations and displayed strongly reduced brood size and high percentage of sterility (Fig. 3d). Whereas the heterozygous progeny of endu-2(+/+) mothers had brood sizes with typically more than 100 F2 animals, the brood size of F1 cross-progeny derived from endu-2(−/−) mothers was as low as that of their mothers. We conclude that ENDU-2 function is required predominantly in the oocytes to preserve germline immortality.

The temperature-dependent Mrt phenotype of endu-2 mutants resembles that of hrde-1 mutants in the nuclear RNAi pathway^5,15. In addition to loss of germline immortality, hrde-1 is also defective in multigenerational inheritance of germline RNAi. To test whether ENDU-2 acts in the nuclear RNAi pathway, we examined oma-1 RNAi inheritance that suppresses embryonic lethality of an oma-1 gain-of-function mutant for several generations¹⁶. Unlike the hrde-1 mutants that lost inheritance of oma-1 RNAi within two generations, effect of oma-1 RNAi knock-down persisted for 5–6 generations both in endu-2(lf) and wild type animals (Supplementary Fig. 6b), suggesting that ENDU-2 does not act in the nuclear RNAi pathway to control transgenerational inheritance of germline RNAi. Therefore, maintenance of germline immortality by ENDU-2 probably functions via a mechanism distinct from that of the nuclear RNAi pathway.

mRNA binding by ENDU-2, but not its mRNA-cleavage, is essential for maintaining germline immortality

ENDU-2 harbors two XendoU domains, of which the C-terminally localized domain is more similar to both human EndoU and Xenopus XendoU (Supplementary Fig. 7). It was suggested recently that ENDU-2 might also function as RNA-binding protein¹⁰. To identify the RNAs bound by ENDU-2, we precipitated ENDU-2::EGFP and analyzed co-immunoprecipitated RNAs by deep-sequencing (RIP-Seq). Since wild type ENDU-2 would potentially cleave its RNA targets, we thought that this might prevent enrichment of intact RNA targets and their subsequent identification. Although we had no direct proof of an RNA cleaving activity of ENDU-2 yet, we reasoned that generation of an ENDU-2 variant that maintains RNA binding, but has lost RNA-cleavage activity, should facilitate RNA target detection. Xenopus XendoU mutants with E to Q exchanges at positions 161 or 167 of the EndoU domain lose RNA cleavage without reducing their RNA-binding activity¹⁷. The second glutamic acid (167E) is conserved in both XendoU domains of C. elegans ENDU-2 (175E and 460E), whereas the first (161E) is only found in the second XendoU domain (454E) (Supplementary Fig. 7b). We expressed both ENDU-2(E454Q)::EGFP and ENDU-2(E460Q)::EGFP variants in endu-2(lf) background and performed RIP-Seq with these two strains in additional to wild type ENDU-2::EGFP. By plotting normalized reads (RPM) of each transcript we were able to investigate RNA-binding affinity of ENDU-2 under different conditions. In general, all three ENDU-2 variants tested showed stronger RNA-binding affinity at 15 °C than at 25 °C (Fig. 4a). E460Q showed weaker RNA-binding already at 15 °C than wild type ENDU-2 and almost completely lost RNA-binding capacity at 25 °C. In addition, ENDU-2(E454Q) displayed stronger RNA-binding activity than ENDU-2(wt) at 15 °C. Therefore, we used RIP-Seq data of ENDU-2(E454Q) at 15 °C for detecting RNAs bound by ENDU-2. 5920 transcripts were co-immunoprecipitated with ENDU-2(E454Q) (Supplementary Data 1). Most of them (>99%) were protein-coding transcripts, the remaining were non-coding RNAs (5 snoRNAs, 27 pseudogenes, 5 ncRNAs, and 1 lincRNA). In addition, reads distribution analysis showed >99% of sequenced reads were mapped to exons, suggesting that ENDU-2(E454Q) primarily binds to processed mRNAs (Fig. 4b). We performed in vitro mRNA-binding assays with recombinant ENDU-2 proteins and two selected mRNA targets from the RIP-Seq data and confirmed that both ENDU-2 and ENDU-2(E454Q) directly interact with these mRNAs (Fig. 4c and Supplementary Fig. 11b).

a Comparison of RNA binding activities of ENDU-2(wt)::EGFP, ENDU-2(E454Q)::EGFP, and ENDU-2(E460Q)::EGFP at 15 and 25 °C. Shown are plotted normalized reads (RPM) from RIP-Seq of each identified transcript under different conditions. b Mapping of the RIP-Seq reads of two representative binding targets of ENDU-2. Only fragments of mRNA exons (red boxes) but not introns (black lines) were co-immunoprecipitated with ENDU-2. c ENDU-2::EGFP variants bind to selected mRNAs in vitro. Shown are western blots to detect proteins binding to fat-7 and trcs-1 mRNA, respectively. EGFP is a negative control. N = 3 biological replicates. The uncropped blots are included in Supplementary Fig. 11b. d Wild type ENDU-2, but not ENDU-2(E454Q), leads to RNA decay (smear) in a Ca²⁺ and Mn²⁺ dependent manner, N = 3. Shown is fused images of two gels from one experiment. The results of additional two biological replicates are shown in Supplementary Fig. 8a. e endu-2(E454Q)::EGFP but not endu-2(E460Q)::EGFP transgene rescues the Mrt phenotype. Data are mean ± SD, N = 3. Both endu-2(tm4977) and endu-2(tm4977) carrying endu-2(E454Q)::EGFP or endu-2(E460Q)::EGFP transgenes were decedents of one single P0 animal carrying the respective transgenes.

Studies of Xenopus XendoU had shown that that the RNA hydrolysis activity of XendoU requires Mn²⁺ or Ca^2+ 18. ENDU-2 seems to require similar conditions, since the addition of 5 mM Mn²⁺ or Ca²⁺ in the buffer medium, but not 5 mM Mg²⁺, led to degradation of bulk RNA in the wild type worm lysates (Fig. 4d and Supplementary Fig. 8). Bulk RNA degradation was strongly reduced in endu-2(lf) animals. In addition, expression of the wild type endu-2::EGFP transgene, but not of endu-2(E454Q)::EGFP, restored RNA decay in endu-2(lf) mutants, supporting our hypothesis that ENDU-2(E454Q) lost RNA-cleavage capacity despite its increased RNA-binding affinity. Surprisingly, extrachromosomal expression of ENDU-2(E454Q), but not of ENDU-2(E460Q), fully rescued the Mrt phenotype of endu-2(lf) at 25 °C (Fig. 4e). We conclude that RNA binding, rather than RNA-cleavage activity of ENDU-2, is essential in the germline to maintain stem cell immortality at elevated temperature.

ENDU-2 negatively regulates somatic mRNA abundance via its endoribonuclease activity

To test if ENDU-2 affects mRNA levels, we performed microarray experiments at 25 °C to determine alterations in the transcriptome mediated by ENDU-2 (procedure of sample preparation is illustrated in Supplementary Fig. 9a). A comparison of differential gene expression in endu-2(−) and endu-2(+) backgrounds suggested a similar number of transcripts being up- and downregulated by endu-2 (fold-change >2) (Fig. 5a). Among the 258 transcripts downregulated in endu-2(lf) animals, germline expressed genes (n = 206) were over-represented (Supplementary Data 2). qPCR quantifications of two selected genes cav-1 and trsc-1 confirmed our microarray results and revealed that their expression levels strongly decreased in endu-2 mutants compared to wild type animals at 25 °C (Supplementary Fig. 10c and 10d). However, neither a germline expressed cav-1::GFP reporter nor smFISH staining of trcs-1 confirmed significantly altered germline expression in situ (Supplementary Fig. 10a, 10b and 10d). As the gonads of endu-2(lf) mutants at 25 °C became significantly smaller (Supplementary Fig. 9b), we speculate that apparent downregulation of germline mRNA within the transcriptome data set may be caused by reduced sample size of the gonad in endu-2 mutants.

a Volcano blot of microarray results comparing endu-2(tm4977) mutant and endu-2(tm4977) animals carrying the rescuing endu-2::EGFP transgene at 25 °C. N = 3 biological replicates. The dots in the bright red or blue areas are differentially regulated transcripts (FC > 2 with two-tailed P-value <0.05). The blue dots mark transcripts bound by ENDU-2. Transcripts with blue and black label were ENDU-2 associated according to the RIP-Seq result and those with black labels were verified with qPCR. b ENDU-2 negatively regulates lipid content at 25 °C. Shown are mean ± SD of quantification of relative Oil Red O stained signal in wild type, endu-2(tm4977) as well as endu-2(tm4977);byEx1375[endu-2::EGFP] day-one adult animals. Numbers of examined samples over three independent experiments: n = 80 for wild type and endu-2(tm4977) at 15 °C, wild type at 25 °C, n = 89 for endu-2(tm4977) at 25 °C, n = 86 for endu-2(tm4977);byEx1375[endu-2::EGFP] at 25 °C. Statistical test with one-way ANOVA. P-values were calculated with Tukey’s multiple comparison test. P-value for WT vs. endu-2(tm4977) at 15 °C: 0.0562. ***P-value <0.0001. c qPCR to quantitate selected ENDU-2 target mRNAs in the soma at 20 and 25 °C. Data are mean ± SD, N = three biological replicates. d Fluorescent images of smFISH staining of the ENDU-2 target fat-7 mRNA and quantification of the smFISH stained mRNA foci per examined area in wild type and endu-2(tm4977) animals. Scale bar 10 µm. Data are mean ± SD. Numbers of examined samples over three independent experiments: wild type at 20 °C n = 42, endu-2(tm4977) at 20 °C n = 31, wild type at 25 °C n = 49, endu-2(tm4977) at 25 °C n = 41. Statistical test with one-way ANOVA. P-values were calculated with Tukey’s multiple comparison test. ***P-value <0.0001. e Fluorescent images and quantification of fluorescence intensity of FAT-7::GFP in wild type and endu-2(tm4977) day-one adult animals at different temperatures. Data are mean ± SD. Numbers of examined samples over three independent experiments: wild type at 15 °C n = 92, endu-2(tm4977) at 15 °C n = 80, wild type at 25 °C n = 102, endu-2(tm4977) at 25 °C n = 89. Statistical test with one-way ANOVA. P-values were calculated with Tukey’s multiple comparison test. ***P-value <0.0001. Scale bar 500 µm. f Fluorescent images and quantification of fluorescence intensity of FAT-7::GFP in wild type, endu-2(tm4977), endu-2(tm4977);Ex[endu-2(wt)] and endu-2(tm4977);Ex[endu-2(E454Q)] day-one adult animals at 25 °C. Data are mean ± SD. Numbers of examined samples over three independent experiments: For endu-2(wt) rescue experiment, wild type n = 55, endu-2(tm4977) n = 53, endu-2(tm4977);Ex[endu-2(wt)] n = 29. For endu-2(E454Q) rescue experiment, wild type n = 38, endu-2(tm4977) n = 38, endu-2(tm4977);Ex[endu-2(E454Q)] n = 31. Statistical test with one-way ANOVA. P-values were calculated with Tukey’s multiple comparison test. P-value for WT vs. endu-2(tm4977);Ex[endu-2(wt)]: 0.9182, for endu-2(tm4977) vs. endu-2(tm4977);Ex[endu-2(E454Q)]: 0.7984. ***P-value <0.0001. Scale bar 500 µm.

Therefore, we put our focus on genes that were upregulated in endu-2(lf) background. Thirty-two percent (76 out of 237) of these transcripts were candidates for direct ENDU-2 targets, since they had been co-immunoprecipitated with ENDU-2 (Supplementary Data 1). In addition, the vast majority of them (62 out of 76) did not have increased expression in the gonad of endu-2(lf) animals (see next chapter and Supplementary Data 1). This suggests that, at minimum, these 62 transcripts are downregulated by ENDU-2 in the somatic tissues. GO term analyses implicated these somatic ENDU-2 targets in regulation of various metabolic processes (Fig. 5a). Consistently, we found that loss of endu-2 resulted in increased lipid content at 25 °C but not at 15 °C (Fig. 5b). Taken together, these results indicate a putative regulatory role of ENDU-2 in limiting abundance of the mRNAs that are primarily involved in metabolic functions in the soma.

As some endu-2(lf) phenotypes were temperature-dependent, we asked whether ENDU-2 regulates mRNA abundance in response to temperature alterations. To test this, we performed qPCR to quantitate ten selected mRNAs that were bound and downregulated by ENDU-2. We found that mRNA levels of these targets were upregulated in endu-2 mutant background, compared to wild type, already at standard growth temperature (20 °C), and were even more abundant at 25 °C (Fig. 5c). Increased temperature, on the other hand, did not strongly affect the abundance of most of these mRNAs in wild type background. These results suggest a temperature-dependent negative influence of ENDU-2 on the levels of these mRNA targets. We also performed smFISH to inspect the influence of ENDU-2 on the mRNA level of yet another target fat-7. fat-7 mRNA was only detected in the intestine and endu-2(−) animals had higher fat-7 transcript levels than wild type at both 20 and 25 °C (Fig. 5d). Moreover, we used a fat-7::GFP translational fusion reporter to monitor FAT-7::GFP protein level. Wild type animals showed reduced FAT-7::GFP levels at 25 °C vs. 15 °C (Fig. 5e). endu-2(lf) displayed stronger FAT-7::GFP expression levels at both temperatures. Furthermore, transgenic expression of endu-2(wt) but not the endu-2(E454Q) transgene (that has lost RNA-cleavage activity) restored decreased FAT-7::GFP level (Fig. 5f). We conclude that ENDU-2 mediated mRNA-cleavage is required for decreasing expression of at least some of its somatic target genes, such as fat-7.

ENDU-2 prevents misexpression of soma-specific genes in the germline

To investigate germline transcriptomes regulated by ENDU-2, we isolated gonads from wild type and endu-2(lf) animals which had been grown at 25 °C from L1 stage, and sequenced two biological replicates of polyadenylated RNA from each strain. A gene with normalized reads of RPKM > 1 was scored as being expressed. We identified 8356 expressed genes in the gonad of wild type animals at 25 °C, among which 88% (7365) have been reported as expressed in the wild type gonad at 20 °C¹⁹. The following DESeq2 analysis showed that upregulation of mRNA levels was the prevalent change in the gonad of endu-2 mutants (422 upregulated, 35 downregulated, FC > 2, P-value <0.05; Fig. 6a, Supplementary Data 3). Only genes upregulated in endu-2(lf) background were enriched in ENDU-2 binding, as 218 out of 422 upregulated genes (P-value < 0.0001, Chi-square test) and 11 out of 35 downregulated genes (P-value = 0.7675, Chi-square test) were co-immunoprecipitated with ENDU-2. We considered them as direct targets of ENDU-2 and grouped them in two classes, depending on whether they are up- (class I) or downregulated (class II) in endu-2(lf) background (Fig. 6b). Expression of a class I targets is repressed, while class II genes are activated by wild type ENDU-2 activity. Consistently, we noticed that most of the class I, but not the class II target genes, were not, or very lowly, expressed in the gonad of wild type animals (Fig. 6c). Additional tissue enrichment analysis suggested that the class I targets are enriched in genes expressed in the somatic tissues (intestine, neurons, muscles, somatic gonad, and pharynx). GO term enrichment analysis revealed that the major function of class I targets are involved in immune and defense responses (Fig. 6b). We wondered whether the class I target genes were preferentially soma-specific genes that are misexpressed in the germline in the absence of ENDU-2. To validate this assumption, we combined our wild type gonad transcriptome data with a previous published whole animal’s transcriptome analysis at 25 °C²⁰ and calculated a soma enrichment factor (SEF) for each transcript to estimate the expression ratio between the soma and the gonad. Class I, but not class II, targets had on average significantly higher SEF values than the total mRNAs (Fig. 6d), suggesting that the class I genes are predominantly expressed in the soma, but only expressed at low levels in the germline of wild type animals. As maintenance of germline immortality by ENDU-2 does not require its RNA-cleavage activity, we additionally sequenced mRNA from isolated gonads of endu-2(lf) mutant carrying the endu-2(E454Q)::EGFP transgene. Expressing endu-2(E454Q)::EGFP was sufficient to repress germline expression of most class I targets (Fig. 6e and f). In summary, our data indicate that the inhibitory role of ENDU-2 to prevent germline expression of soma-specific genes is probably independent of its RNA-cleavage activity.

a Volcano plot of gonadal expressed genes in endu-2(tm4977) relative to wild type day-one adult animals grown up at 25 °C. Two biological replicates were obtained for each strain. Significantly misregulated genes (FC > 2, FDR < 0.05) were calculated with DESeq2 and are shown in red or blue. n refers to the total numbers of up- or downregulated genes. b Comparison of the total transcripts co-immunoprecipitated with ENDU-2 and ENDU-2 regulated genes in the gonad. c Comparison of expression levels (RPKM) of total gonad-expressed genes (n = 8356), ENDU-2 class I (n = 218) and class II targets (n = 11) in wild type animals. Shown are boxes extending from the 25th to the 75th percentile, with the median indicated by the horizontal line and Min/Max whiskers. Statistical test with one-way ANOVA. P-values were calculated with Dunnett’s multiple comparison test. P-value for total vs. class II targets: 0.5283. ***P-value <0.0001. d Comparison of relative expression between the soma and the gonad (SEF) of the total transcripts (n = 12371), ENDU-2 class I (n = 218), and class II (n = 11) targets in wild type animals. Shown are boxes extending from the 25th to the 75th percentile, with the median indicated by the horizontal line and Min/Max whiskers. Statistical test with one-way ANOVA. P-values were calculated with Dunnett’s multiple comparison test. P-value for total vs. class II targets: 0.1477. ***P-value <0.0001. e The majority of the alerted expression of the germline class I targets is rescued by endu-2(E454Q)::EGFP transgene. f Heat map of ENDU-2 repressed germline genes in wild type, endu-2(tm4977) and endu-2;Ex[endu-2(E454Q)::EGFP] animals.