Neuronal RNAi and oxygen-sensing circuit shape germline resilience to heat stress

Summary

Thermal pollution, whether local or driven by global warming, threatens biodiversity in part through its detrimental effects on reproduction. Non-coding small RNAs (sRNAs) are crucial for maintaining germline developmental robustness under heat stress. Remarkably, we uncovered that neuronal sRNAs regulate germ cells’ thermotolerance, affecting both spermatogenic and oogenic germlines in a cell-non-autonomous manner. Furthermore, we demonstrate that, in RNAi mutants, an oxygen-sensing neural circuit, modulated by neuropeptide signaling, antagonizes germline maintenance, likely reflecting the nematode’s innate association of reduced oxygen levels with food availability and reproductive permissive environments. Finally, we provide evidence that laboratory-domesticated alleles of oxygen-response genes encoding neuropeptide receptor NPR-1 and hexacoordinated globin GLB-5 compromise germline thermotolerance. Hence, our findings raise the possibility that sensory perception, independent of direct environmental change, modulates germline integrity, highlighting a novel mechanism by which neural circuits integrate environmental information to safeguard reproductive fitness in fluctuating environments.

Graphical abstract

Highlights

• •dsRNA-binding protein RDE-4 is required for germline thermotolerance
• •Neuronal RDE-4 promotes germline thermotolerance
• •Blocking oxygen sensing promotes germline development in rde-4-deficient animals

Ewe et al. demonstrate that neuronal small RNAs can modulate germline development under heat stress in Caenorhabditis elegans. They further show that disrupting the oxygen-sensing neural circuit promotes germline thermotolerance, linking environmental perception to developmental outcomes and highlighting a role for RNAi in integrating environmental information to regulate reproductive robustness.

Introduction

Reproductive health is strongly influenced by environmental factors. Mass extinctions are happening on an unprecedented scale, driven, in large part, by rising global temperatures that cause widespread fertility loss. Climate change is linked to declining male and female fertility from plants to mammals, posing a significant challenge for future generations.^1,2 How do organisms endure and adapt in such difficult times?

Non-coding small RNAs (sRNAs), including siRNAs, miRNAs, and piRNAs, together with Argonautes (AGOs), play conserved roles in germline development and fertility.³ In flies and mammals, loss of PIWI proteins leads to germ cell loss and transposon derepression.^4,5,6 In mice, endo-siRNAs are critical for oogenesis, while miRNAs and piRNAs are essential for multiple aspects of spermatogenesis.^7,8,9,10,11 Similarly, in the nematode C. elegans, sRNAs play key roles in germ cell development.^{12,13,14,15,16,17} Given the potential of sRNA/AGO pathways to drive plastic gene regulatory programs, it is of great interest to understand how they regulate fecundity in changing environments.

Growing evidence pointed to the roles of somatic tissues—especially reproductive support cells—in transmitting environmental signals to the germline, in part by regulating RNA payload in germ cells.^18,19,20,21 For example, in flies, exosomes derived from the secondary cells in the accessory gland fuse with mature sperm and facilitate reproductive success.¹⁹ In mammals, the epididymal epithelium may load RNAs onto maturing sperm via exosomes and microvesicles. Evidence suggests that paternal diet may affect RNA production in the epididymis and, in turn, reprogram sperm epigenome, enabling intergenerational inheritance of environmental signals.^20,21,22,23

Results

Endo-siRNAs protect sperm thermotolerance

TRBP2 is a dsRNA-binding protein essential for Dicer-mediated miRNA production in vertebrates.²⁴ In flies, its homolog R2D2 and Loqs-PD, together with Dicer, facilitate siRNA loading onto RNA-induced silencing complex (RISC).^25,26 In C. elegans, the TRBP2 homolog RDE-4 is critical for exogenous RNAi and antiviral defense by binding long dsRNAs and interacting with DCR-1/Dicer, DRH-1/RIC-I, and the AGO RDE-1.^27,28,29,30 RDE-4 is also required for the biogenesis of endogenous siRNAs and ensures their proper loading onto AGOs.^31,32 Central to this study, previous work uncovered an important role of RDE-4 in embryogenesis during heat stress, with the loss-of-function mutants exhibiting temperature-sensitive developmental defects and laying arrested embryos at 25°C.³³

Here, we similarly found that rde-4(−) mutants are fertile at 15°C and 20°C; however, at 25°C, they lay unfertilized oocytes—large, dark, rounded cells with prominent nuclei—that are readily distinguishable from unhatched embryos (unfertilized oocytes: 80.6%; n = 500) (Figures 1A–1C). Unless stated otherwise, we scored eggs and unfertilized oocytes laid by day-2 adults at 25°C (see STAR Methods). Arrested unfertilized oocytes usually build up in the proximal gonad (stacked appearance) and may undergo meiotic maturation in the absence of sperm and become endomitotic, resulting in polyploidy and nuclear hypertrophy.³⁴ We observed this phenotype across three different alleles of rde-4: ne299 (A123→STOP), uu53 (925 bp deletion), and pig51 (637 bp deletion), but not in wild-type animals (unfertilized oocytes: 0.06%; n = 236) (Figures 1A–1D and S1). We found that the fertility defects in rde-4(−) under heat stress stem from sperm failure, as crossing rde-4(−) to wild-type males rescued the phenotype (Figure 1E).

Figure 1.

Loss of RDE-4 compromises sperm thermotolerance

(A–D) rde-4(ne299) shows temperature-sensitive fertility defects.

(A) L4 larvae were transferred from 20°C to 15°C or 25°C. The fertility of day-2 adults was scored. ^∗q ≤ 0.05 by Kruskal-Wallis test followed by pairwise Mann-Whitney tests.

(B) At 25°C, rde-4(ne299) animals lay unfertilized oocytes, which are morphologically distinct from fertilized embryos.

(C) Unfertilized oocytes tend to accumulate and stack within the gonads of rde-4(ne299) at 25°C. The presence of stacked unfertilized oocytes in one (“one arm”) or both (“two arms”) gonadal arms was quantified. D1A: day-1 adults; D2A: day-2 adults.

(D) rde-4(pig51) shows similar phenotypes to rde-4(ne299). Scale bars, 100 μm. D2A: day-2 adults.

(E) The fertility defects of rde-4(ne299) at 25°C are rescued by crossing with wild-type males. In these experiments, L4 hermaphrodites were incubated with males at 25°C for ∼24 h. Mated hermaphrodites were then transferred to fresh plates, and the numbers of eggs and unfertilized oocytes laid were scored. ^∗∗∗∗p < 0.0001 by Mann-Whitney test.

(F) MSP expression and localization are severely disrupted in the male gonad in rde-4(ne299);him-5(e1490) at 25°C. Scale bars, 50 μm.

(G) Sperm from rde-4(ne299);him-5(e1490) exhibit chromosomal abnormalities. Arrows mark chromosomal bridges; the asterisk indicates decondensation of DNA. DAPI and antibody staining were performed on day-1 adults. Scale bars, 10 μm.

(H) rde-4(ne299);him-5(e1490) males fail to mate with fog-2(q71) female at 25°C. Mating efficiency was estimated by scoring the presence of fertilized eggs in fog-2(p71) hermaphrodites after ∼24 h of incubation with males.

For all relevant panels, error represents mean ± SD.

In (C) and (H), numbers indicate the total number of animals scored. ^∗∗∗∗p < 0.0001 by Fisher’s exact tests.

See also Figure S1.

Given the sperm defects observed in rde-4(−) hermaphrodites (Figures 1A–1E), we wondered whether rde-4(−) also plays a role in male sperm development. To enrich for males, we introduced a him-5 mutation, which induces a high frequency of X chromosome nondisjunction without compromising sperm morphology and functions.³⁵ In him-5(−) males, major sperm protein (MSP), which is essential for sperm functions in nematodes, is packed into fibrous bodies and localized to membranous organelles; on the other hand, its organization is severely disrupted, and MSP remains diffused in rde-4(−);him-5(−) males at 25°C (Figure 1F). In addition, we observed pronounced chromosomal abnormalities in spermatocytes from rde-4(−);him-5(−) males (Figure 1G). Because fog-2(−) mutants lack self-sperm and rely on mating for reproduction, fog-2(−) females crossed with him-5 males restored fertility. However, rde-4(−);him-5(−) males failed to sire progeny at 25°C (Figure 1H). Together, these results indicate that RDE-4, functioning in the endo-siRNA pathway, is required for the thermotolerance of sperm in both hermaphrodites and males.

Neuron-to-germline communication affects sperm development in animals lacking rde-4

C. elegans contains a highly elaborate sRNA/AGO machinery, consisting of at least 19 AGOs that act sequentially to mediate potent gene silencing responses in diverse biological contexts.^36,37,38 In the endogenous siRNA pathway, the enhanced RNAi (ERI) complex produces 26G sRNAs using mRNAs as templates. The 26G sRNAs are then loaded onto primary AGOs―ERGO-1 in oocytes and embryos, and ALG-3/4 in sperm―which recruit RNA-dependent RNA polymerases (RdRPs) RRF-1 and EGO-1, leading to the production of abundant secondary 22G sRNAs.^12,39,40 In sperm, WAGO-10 and CSR-1a bind 22G sRNA and function downstream of ALG-3/4 to regulate spermatogenesis.^12,13

RDE-4 is required for the full production of 26G endo-siRNAs, although it is dispensable in certain cases.^32,41,42 In this study, we performed sRNA-seq using a 5′-phosphate-independent protocol (see STAR Methods) and found that siRNAs (antisense reads without applying filters on 5′ nucleotide composition) targeting spermatogenic genes are upregulated in rde-4(−) mutants grown at 25°C compared with 20°C (Figure 2A). Notably, at 25°C, upregulated siRNAs in rde-4(−) relative to wild type are enriched for those targeting sperm genes, male-enriched genes, and certain known ALG-3/4-class sRNAs (Figures 2B and S2A).

Figure 2.

Neuronal RDE-4 promotes sperm thermotolerance

(A) siRNA targeting spermatogenic genes are upregulated in rde-4(ne299) grown at 25°C compared with 20°C. All sequencing experiments were performed on day-1 adults.

(B–D) siRNAs targeting male-enriched genes, spermatogenic genes, and some ALG-3/4-class sRNAs (i.e., sRNAs mapped to ALG-3/4 target genes)³⁷ tend to be upregulated (DESeq2 q < 0.05) in rde-4(ne299) versus wild type at 25°C, which is rescued by pigSi3[Psng-1::rde-4]. The numbers in (C) and (D) indicate the differences in average log₂FC.

(E) Spermatogenic transcripts are upregulated in rde-4(ne299) (average log₂FC = 0.75) and restored by neuronal rde-4(+) (average log₂FC = −0.25).

(F) Expression of AGO and RdRP genes is dysregulated in rde-4(ne299), which is rescued by neuronal rde-4(+) in some cases. Relative log expression (RLE) is shown. ^∗q ≤ 0.05; ^∗∗q < 0.01; ^∗∗∗∗q < 0.0001 by DESeq2.

(G–I) Neuronal rde-4(+) partially rescues fertility defects in rde-4(ne299) at 25°C. The numbers in (H) indicate the total number of animals scored. D1A: day-1 adults; D2A: day-2 adults. (I) rde-4(ne299) day-1 adult carries endomitotic nuclei and shows disrupted MSP expression at 25°C. These phenotypes are rescued by neuronal rde-4(+). Scale bars, 20 μm.

(J) The effects of neuronal rde-4(+) are not affected by the loss of sid-1.

(K) Schematic of mosaic uniparental inheritance induced by GPR-1 overexpression.

(L) Wild-type rde-4 in the neurons of mosaic animals partially rescues fertility defects.

For all relevant panels, error represents mean ± SD. Not significant, p > 0.05; ^∗p ≤ 0.05; ^∗∗p < 0.01; ^∗∗∗∗p < 0.0001 by Mann-Whitney tests. Multiple comparison corrections were applied where appropriate.

See also Figure S2 and Data S1, S2, and S3.

We recently demonstrated that neuronal RDE-4 may alter germline gene expression and trigger transgenerational behavioral changes.⁴³ Intriguingly, here, we found that expressing rde-4 in neurons as a single-copy MosSCI transgene under the control of a pan-neuronal promoter (pigSi3[Psng-1::rde-4::SL2:yfp]) could largely restore the expression of sperm siRNAs, many of which are associated with ALG-3/4, in rde-4(−) mutants (Figures 2B–2D). ALG-3/4 may mediate either negative or positive gene regulation of their targets⁴⁴; consistent with this, we observed aberrant upregulation of many spermatogenic transcripts in rde-4(−) (Figures 2E and S2B), while genes that are known to be negatively regulated by ALG-3/4 are downregulated in rde-4(−) animals (Figure S2B). In addition, we detected misregulation of several other germline AGOs, including csr-1 and wago-10, and RdRP genes (Figures 2F and S2C). These expression patterns were at least partially rescued by neuronal rde-4(+) (Figures 2E, 2F, S2B, and S2C). Together, our findings suggest that neuronal RDE-4 may regulate germline sRNA pathways cell-non-autonomously.

Strikingly, expression of neuronal rde-4(+) partially rescued the heat-induced fertility defects observed in rde-4(−) mutants (unfertilized oocytes: 89.5% in rde-4(ne299) versus 50.6% in rde-4(ne299) + neuronal rde-4(+)) (Figures 2G–2I and S2D). In contrast, hypodermal or muscle-specific rescue of rde-4 driven by nas-9 or myo-3 promoter did not show an effect (Figure S2E). By performing smFISH and RNA-seq on isolated gonads, we previously confirmed that the neuronal rde-4(+) transgene is not mis-expressed in the germline.⁴³ Importantly, similar to rde-4(−) mutants and in contrast to wild-type animals, rde-4(−) mutants carrying the neuronal rde-4(+) transgene are not responsive to exogenous dsRNA targeting gfp expressed in sperm (Figure S2F). Hence, these findings indicate that the observed effects are not artifacts caused by transgene misexpression in the germline but reflect a bona fide cell-non-autonomous role of RDE-4 in regulating sperm heat tolerance. This neuron-to-sperm communication appears to be independent of SID-1, a conserved dsRNA-selective importer required for systemic RNAi (Figure 2J).⁴⁵

ALG-3/4 are required for sperm development at elevated temperature: single alg-3 or alg-4 mutants show a severely reduced brood size at 25°C, whereas double mutants are completely sterile.¹² Our mRNA and sRNA sequencing experiments provided evidence consistent with elevated ALG-3/4 activity in rde-4(−) mutants (see above), and this is restored by neuronal rde-4(+), raising the possibility that sperm development is sensitive to ALG-3/4 dosage such that either too little or too much is detrimental. To test this, we removed alg-3 in the rde-4(ne299) mutants; however, this did not rescue fertility (Figure S2G). These results suggest that RDE-4 controls sperm development through additional AGOs that we did not test, or that neuronal rde-4(+) regulates fertility via an sRNA-independent mechanism, and the changes in sRNA profiles reflect the consequence, rather than the cause of restored fertility.

To provide additional support for our conclusion regarding RDE-4-dependent neuron-to-sperm communication, we performed mosaic analysis by inducing uniparental isodisomy. To achieve this, we crossed rde-4(ne299);him-5(e1490) males to hermaphrodites overexpressing GPR-1, a conserved microtubule force regulator. This manipulation causes premature segregation of maternal and paternal chromosomes in the pre-cleavage embryos, resulting in non-Mendelian partitioning of genetic material between the first two blastomeres: AB and P₁. In this system, the anterior AB contains exclusively maternally derived chromosomes, while the posterior P₁ carries only paternally derived chromosomes.^46,47 As nearly all neurons are derived from AB,⁴⁸ they inherit the wild-type rde-4 from the hermaphrodite, whereas the germline, which arises from the P₁ lineage, carries the rde-4(ne299) mutation from the male (Figure 2K). We found that these mosaic animals exhibit milder fertility defects compared with rde-4(−) mutants at 25°C (unfertilized oocytes: 97.1% in rde-4(ne299) versus 74.0% in rde-4 chimeras) (Figure 2L). In our control experiments, chimeras generated by crossing him-5(e1490) males with GPR-1-overexpressing hermaphrodites produced brood sizes similar to those of wild type (Figure S2H). These results support our earlier findings that neuronal sRNAs promote sperm thermotolerance (Figures 2G–2I). Importantly, this effect cannot be attributed to maternal provision, as rde-4 homozygous mutants derived from heterozygous mothers still display severe fertility defects (Figure S2I). Finally, we inserted FRT sequences at the 5′ and 3′ ends of the rde-4 locus using CRISPR-Cas9 and excised the flanked sequence by neuronal expression of FLP recombinase (rgef-1p::FLP) (Figure S2J).⁴⁹ Loss of neuronal rde-4 in an otherwise wild-type background did not overtly affect brood size at either 20°C or 25°C (Figure S2K). These results indicate that germline development is predominantly regulated by germline sRNAs, with neuronal RNAi playing a cryptic modulatory role that becomes apparent in response to genetic perturbation, such as loss of rde-4 in the germline, and environmental challenge, such as heat stress.

Oxygen-sensing neural circuit negatively impacts germline development

Previous studies showed that RDE-4 and endo-siRNAs are required for various neuronal processes.^50,51,52,53 Importantly, we recently demonstrated that rde-4(−) mutants exhibit defective chemotaxis toward various volatile and soluble attractants at 25°C, but not at 20°C.⁴³ This defect is partially rescued by neuronal expression of rde-4(+) (Data S1).⁴³ Hence, neuronal RDE-4 is important for the detection of various environmental cues, especially in the presence of heat stress.

Given the roles of RDE-4 in sensory signaling, we tested whether blocking sensory perception affects sperm development. Interestingly, eliminating cmk-1, a gene encoding the homolog of CaMKI, which modulates sensory gene expression, partially rescues rde-4(−) sterility at 25°C (unfertilized oocytes: 96.3% in rde-4(ne299) versus 58.3% in rde-4(ne299);cmk-1(oy21)) (Figure 3A). We observed similar effects when we knocked out tax-2, which encodes a subunit of cyclic nucleotide (cGMP)-gated channel that is required for many sensory responses (unfertilized oocytes: 83.7% in rde-4(ne299) versus 38.0% in tax-2(ok3403);rde-4(ne299)) (Figures 3B and 3C). A partial loss-of-function allele, tax-2(p694),⁵⁴ produced a similar, albeit weaker, effect (unfertilized oocytes: 90.8% in rde-4(ne299) versus 73.2% in tax-2(p694);rde-4(ne299)) (Figure S3A). These results indicate that the sensory inputs normally inhibit sRNA-mediated sperm heat tolerance.

Figure 3.

Oxygen-sensing neurons inhibit reproduction

(A–C) Broad inhibition of sensory perception by knocking out cmk-1 or tax-2 rescues the fertility defects of rde-4(ne299) at 25°C. (C) Endomitotic oocytes in rde-4(ne299) are highlighted. Arrows indicate the −1 oocytes, located closest to the spermatheca. DAPI staining was done in day-1 adults. Scale bars, 50 μm.

(D) Inhibiting oxygen perception, but not the other sensory modalities, improves rde-4(ne299) fertility.

(E) Low oxygen levels (7% and 1%) partially rescue germline defects in rde-4(ne299) at 25°C. L4 animals were transferred to a glove box maintained at the desired oxygen level, and the presence of stacked unfertilized oocytes in one (one arm) or both (two arms) gonadal arms in day-1 adults was quantified.

(F) Blocking neuropeptide processing rescues rde-4(ne299) sterility. Note that egl-2(n476) has egg-laying defects independent of fertility.

(G) rde-4(ne299) and pigSi3[Psng-1::rde-4]; rde-4(ne299) mutants show increased responses to 7% and 21% oxygen at 25°C, compared with wild type. n = 7–9 assays, 20–25 animals per assay. Solid lines indicate average speed, and error indicates SEM. Black horizontal bars indicate time intervals used for statistical tests.

(H) npr-1(ad609) mutation rescues rde-4(ne299) sterility.

(I and J) Ancestral (HW) alleles of npr-1 and glb-5 partially restore rde-4(ne299) fertility. glb-5(HW);npr-1(HW) mutants do not show heat-induced loss of fertility. Error represents mean ± SD.

For (A), (B), (D), (G), and (H–J), statistical significance was determined by Mann-Whitney tests. Multiple comparison corrections were applied where appropriate. For (E) and (F), the numbers indicate the total number of animals scored, and statistical significance was determined by Fisher’s exact test. Not significant, p > 0.05; ^∗p ≤ 0.05; ^∗∗p < 0.01; p^∗∗∗ < 0.001; ^∗∗∗∗p < 0.0001.

See also Figure S3.

Next, we asked which sensory modalities might regulate sperm development. Given that sterility in rde-4(−) mutants is temperature-sensitive, we first eliminated three guanylyl cyclases (GCs)—GCY-8, CGY-18, and GCY-23—which function specifically in AFD neurons, the primary thermosensory neurons in C. elegans.⁵⁵ While we previously showed that removing gcy-8/18/23 downregulates AGO genes and disrupts heritable gene silencing in the germline,⁵⁶ we found that blocking heat sensation had no effect on the fertility of rde-4(−) mutants (Figure 3D). Similarly, impairing gustatory signaling by removing CHE-1, a C2H2-type zinc finger TF required for specification of the ASE gustatory neurons,^57,58 or disrupting general chemosensation by knocking out the G protein α subunits GPA-3 and ODR-3,⁵⁹ also did not impact fertility (Figure 3D). In addition, eliminating MEC-8, which plays a role in amphid cilia fasciculation and mechanosensory neuron development,^60,61 had no observable effect (Figure S3B).

C. elegans has strong behavioral responses to the gases oxygen and carbon dioxide, which are highly variable in its natural habitats such as soil, compost heaps, and rotting substrates. Environmental oxygen is primarily detected by the AQR, URX, and BAG neurons in the head, and the PQR neuron in the tail, with additional contributions from other neurons.^62,63,64 Mitochondrial activity modulates cellular oxygen consumption, affecting behavioral and physiological responses to environmental oxygen levels.^65,66 We performed differential expression analyses that include RAPToR age estimates as a covariate to control for confounding differences in developmental timing.⁶⁷ Using this approach, we found that genes downregulated in rde-4(−) mutants at 25°C compared with those at 20°C are significantly enriched for mitochondrial processes (Figure S3C). We observed a similar pattern of mitochondrial gene downregulation when comparing rde-4(−) mutants to wild-type animals at 25°C (Figure S3D).

We found that deletion of gcy-33, an oxygen receptor gene expressed in BAG neurons,^64,68 did not affect the fertility of rde-4(−) mutants (Figure 3D). In contrast, deleting gcy-35, which encodes a heme-containing oxygen receptor expressed in the URX, AQR, PQR, SDQ, ALN, and PLN neurons that promote hyperoxia avoidance, rescues sterility of rde-4(−) mutants (unfertilized oocytes: 95.7% in rde-4(ne299) versus 69.8% in gcy-35(ok769);rde-4(ne299)) (Figures 3D and S3E–S3G). To confirm these results, we generated an in-house loss-of-function allele in gcy-35 using CRISPR-Cas9 and observed similar effects (unfertilized oocytes: 95.3% in rde-4(ne299) versus 73.3% in gcy-35(db2020);rde-4(ne299)) (Figures S3G–S3J). We further showed that genetically ablating URX, AQR, and PQR neurons alone by expressing the cell-death activator gene egl-1 driven by the gcy-36 promoter (qaIs2241) is insufficient to rescue rde-4(−) sterility (Figure S3K). This is consistent with a previous study showing that animals carrying qaIs2241 transgene retain hyperoxia avoidance, unlike gcy-35(−) mutants,⁶² and points to other gcy-35-expressing neurons, including SDQ, ALN, and PLN, as potential candidates for regulating germline development in rde-4(−) animals. Given that these neurons do not appear to express TAX-2,⁵⁴ additional sensory neurons are likely to contribute to the regulation of germline development.

TRPV sensory channel OSM-9 acts in the polymodal nociceptive ASH neurons to regulate avoidance of high osmolarity, chemical repellents, and touch, and also acts in non-nociceptive cell types to mediate olfactory responses and sensory adaptation.^69,70,71 Of note, OSM-9 has been shown to function in ASH and serotonergic ADF neurons to promote hyperoxia avoidance.^62,72 We found that knocking out osm-9 modestly rescues rde-4(−) phenotype (unfertilized oocytes: 94.7% in rde-4(ne299) versus 84.3% in rde-4(ne299);osm-9(ky10)) (Figure 3D). However, removing the tryptophan hydroxylase enzyme required for serotonin biosynthesis, TPH-1, does not affect rde-4(−) fertility (Figure S3L), nor does genetic ablation of ASH neurons alone (Figure S3M). Together, these results indicate that multiple oxygen-sensing neurons function synergistically to regulate sRNA-mediated germline development.

Hypoxia exposure, for example, living at high altitude, has been known to negatively impact male fertility in animals.⁷³ Here, we show that mere perception of oxygen—without changing oxygen levels—can antagonize sperm development and fertility modulated by sRNA pathways. We propose that inhibiting oxygen-sensing neurons may mimic cues associated with active bacterial growth that create oxygen sinks, suggestive of conditions favorable for reproduction in C. elegans. Supporting this notion, exposure to low oxygen levels (7% and 1%) partially alleviates the germline defects of rde-4(−) mutants at 25°C (Figure 3E). Note that, in our standard assay, day-1 adults were transferred to fresh plates for egg laying (see STAR Methods). To avoid confounding effects of shifting animals from low oxygen to normoxia, we instead incubated worms at 25°C for 24 h without transfers and scored germline morphological defects using the germline marker mex-5::gfp.

Neuronal signaling pathways, especially the sensory system, exhibit extensive evolutionary plasticity and may be subjected to positive selection in response to changing habitats.^74,75 In the canonical C. elegans “wild-type” strain (N2), laboratory domestication led to the fixation of a gain-of-function allele in the neuropeptide Y homolog npr-1 and a duplication/insertion in hexacoordinated globin gene glb-5, both of which alter the animals’ response to oxygen and carbon dioxide.^63,76,77 These laboratory-derived alleles are absent in LSJ1, a sister strain derived from N2 at least six years before its cryopreservation in 1969.⁷⁸

In N2, the gain-of-function NPR-1(215V) downregulates the activities of oxygen-sensing AQR, PQR, and URX neurons, thereby suppressing hyperoxia avoidance. NPR-1(215V) animals do not respond to drops in oxygen levels and are solitary feeders, dispersing across the bacterial lawn. In contrast, wild isolates such as CB4856 from Hawaii (HW), which carry NPR-1(215F), or animals with an npr-1 loss-of-function allele, slow their movement in response to reduced oxygen and exhibit social feeding (Figure S3N), aggregating along the bacterial lawn border where oxygen levels are lower (effective oxygen concentration of ∼12.8% versus ∼17.1% in the center of the lawn).⁶³ Given the central role of neuropeptide signaling in C. elegans oxygen sensing, we asked whether it also influences germline development in rde-4(−) mutants. Removing carboxypeptidase E ortholog EGL-21, which is required to process endogenous neuropeptides, partially restores rde-4(−) fertility (unfertilized oocytes: 94.2% in rde-4(ne299) versus 71.0% in rde-4(ne299);egl-21(n476)) (Figure 3F). We observed a similar effect by depleting UNC-31/CAPS⁷⁹ required for neuropeptide release from dense-core vesicles, using the auxin-inducible degron (AID) system (unfertilized oocytes: 88.3% in EtOH versus 70.3% in 1 mM auxin) (Figure S3O). We subsequently screened a subset of FMRFamide-like neuropeptides previously implicated in oxygen response and found that loss of FLP-8, FLP-19, or FLP-21 failed to restore fertility in rde-4(−) mutants (Figures S3P–S3R), suggesting the involvement of other neuropeptides that we did not test here.

At 20°C in the presence of food, rde-4(ne299) mutants behave similarly to wild-type N2 in their oxygen response (Figure S3S). However, at 25°C, we found that rde-4(ne299) mutants, unlike N2, respond to oxygen, slowing their movement at 7% and speeding up at 21% oxygen (Figure 3G). This phenotype is alleviated by loss of gcy-35 (Figure S3T). Unexpectedly, pan-neuronal overexpression of rde-4 does not rescue the mutant oxygen response (Figure 3G), suggesting that RNAi may play complex, dose-sensitive roles in the nervous system. Moreover, neuronal rde-4(+) (pigSi3 transgene) may regulate germline development independent of the oxygen-sensing neural pathway, which calls for future studies.

Interestingly, we found that strong loss-of-function mutations in npr-1 rescue the fertility defects of rde-4(−) mutants (unfertilized oocytes: 96.5% in rde-4(ne299) versus 80.0% in rde-4(ne299);npr-1(ad609)) (Figure 3H). We confirmed these results using two independent EMS-induced alleles (ok1447 and ad609) (unfertilized oocytes: 95.6% in rde-4(ne299) versus 78.3% in rde-4(ne299);npr-1(ok1447)) (Figures S3U and S3V). To test whether the NPR-1 regulation of germline thermotolerance in rde-4(−) mutants is linked to social behaviors, we repeated the experiments on uniform bacterial lawns, which prevent oxygen gradients arising from uneven bacterial thickness on standard lawns. We similarly observed rescue under these conditions (Figure S4W), indicating that NPR-1 exerts pleiotropic effects on life-history traits independent of foraging behaviors, perhaps by altering metabolism, as previously shown.⁸⁰ Although npr-1(HW) alone does not affect rde-4(−) fertility, glb-5(HW);npr-1(HW) double mutant weakly but significantly rescues rde-4(−) phenotype (unfertilized oocytes: 98.4% in rde-4(ne299) versus 92.6% in rde-4(ne299);glb-5(HW);npr-1(HW)) (Figure 3I). Consistent with previous findings that these two laboratory-derived alleles confer fitness advantages in a standard laboratory environment, glb-5(HW);npr-1(HW) mutants show a lower brood size compared with wild type at 20°C (Figure 3J).^80,81 At 25°C, the brood size of wild-type animals drops significantly owing to sperm dysfunction⁸²; we found that this decline is not evident in glb-5(HW);npr-1(HW) (Figure 3J), suggesting that the ancestral alleles of npr-1 and glb-5 promote reproductive resilience under heat stress—a major challenge in their natural habitat.

Hence, our results indicate that laboratory domestication alters the oxygen-sensing circuit, which subsequently impacts germline development under heat stress. However, we note that knocking out rde-4 in LSJ1 and CB4856 causes severe loss of fertility, as observed in N2 (Figure S3X), suggesting other natural genetic variants in these strains may act to affect germline development and/or sRNA pathways.^83,84,85,86 Indeed, AGO genes and sRNA pathways are evolving rapidly, exhibiting extensive intraspecies variation that may have broad impacts on germline gene regulation.^84,87 Additionally, other laboratory-derived alleles have been shown to affect sperm development,^85,86 including a deletion in nurf-1, which encodes the ortholog of the BPTF subunit of the NURF chromatin remodeling complex, that arose in the LSJ1 lineage.⁸⁶

Neuronal signaling influences genome stability in the germline

We note that, although mating largely rescues the fertility defect in rde-4(−) mutants, it does not fully restore brood size to wild-type levels, suggesting additional, non-sperm-origin defects (Figure 4A). To investigate further, we examined RAD-51/RecA, which localizes to the double-strand break (DSB) repair loci, in the oogenic germline of day-1 hermaphrodites.^88,89 In wild-type gonads, RAD-51 appears as distinct foci mostly in the leptotene/zygotene and pachytene stages, which coincide with the onset of meiosis (Figures 4B and 4C). In contrast, at 25°C, but not at 20°C, rde-4(−) mutants exhibited an elevated frequency of DSBs throughout the germline, including the mitotic proliferative stem cells, indicating compromised genome integrity under heat stress (Figures 4D and S4A). Remarkably, consistent with our earlier findings, this defect was rescued by neuronal expression of rde-4(+) (Figures 4E–4G). Moreover, eliminating cmk-1 or tax-2 reduces the number of RAD-51 foci in rde-4(−) mutants (Figures 4H, 4J, 4K, and S4B). Finally, loss of gcy-35 or egl-21 rescued the DSBs in rde-4(−) (Figures 4I–4M). Our data do not preclude the possibility that DNA damage machinery in these mutants responds to altered sperm signals.⁹⁰ Nevertheless, these results support the conclusion that neuronal activity—particularly the oxygen-sensing circuit—plays an important role in regulating reproductive robustness. Perception of oxygen levels may modulate germline maintenance as low oxygen levels correlate with bacteria-rich environments and a reproductive permissive niche.

Figure 4.

Neuronal RDE-4 and impaired oxygen-sensing neural pathways promote germline genome stability

(A) Crossing rde-4(ne299) hermaphrodites to him-5(e1490) males does not completely restore brood size at 25°C. D1A: day-1 adults (overnight mating initiated at the L4 stage); D2A: day-2 adults (overnight mating initiated in day-1 adults).

(B–G) Neuronal rde-4(+) rescues DSBs in rde-4(ne299) at 25°C. L4 animals were transferred from 20°C to 25°C for ∼24 h. Antibody staining was performed on day-1 adults.

(H–M) Loss of oxygen sensation (tax-2(ok3403) or gcy-35(ok769)) or neuropeptide signaling (egl-21(n476)) rescues DSBs in rde-4(ne299).

Images in (G), (K), and (M) show whole gonads stained with RAD-51 (red) and DAPI (blue). Scale bars, 50 μm. The right panels show magnification of mitotic nuclei. Scale bars, 10 μm.

For all relevant panels, error represents mean ± SD. ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001 by Mann-Whitney tests. Multiple comparison corrections were applied where appropriate.

See also Figure S4.

Discussion

One of the key features of sRNAs/AGOs is their capacity to help animals mitigate environmental changes by rapidly reprogramming gene expression. This is evident in their conserved roles in conferring stress tolerance to gametes from plants to mammals.^3,91,92 In C. elegans, previous studies have shown that ERI-1/3/5, RRF-3, and DCR-1, together with downstream ALG-3/4 in the endo-siRNA pathway, are important for sperm development under heat stress^12,16; however, the role of RDE-4 in this pathway has not been explicitly addressed. Here, we found that RDE-4 is crucial for both spermatogenic and oogenic germline thermotolerance. Loss of RDE-4 at elevated temperature leads to misregulation of spermatogenic genes, chromosomal defects in the sperm, and an increased DSB in oogenic germ cells, highlighting the crucial role of endo-siRNAs in maintaining reproductive robustness in stressful conditions.

Regulatory sRNAs can exert cell non-autonomous effects via microvesicles or exosomes, influencing diverse biological processes in both health and disease.^93,94 The intercellular transfer of RNAs and other molecular cargoes mediates soma-germline interactions from plants to mammals, regulating germline gene expression.^18,20,21 In this study, we showed that neuronal RNAi modulates germline development in response to genetic and environmental perturbations, buffering developmental systems under stress. This may allow distributed robustness—a common feature of metabolic and developmental programs.^95,96 Although more work is required to define the detailed mechanism underlying neuron-to-germline communication during heat stress, our data showed that neuronal sRNAs can regulate reproduction through a SID-1-independent mechanism. Other candidate mediators of RNA transfer from neurons to the germline include RSD (RNA spreading defective) proteins, which have been implicated in systemic RNAi in C. elegans,⁹⁷ as well as gap junctions,^98,99 warranting future investigation. It also remains possible that other tissues—for example, the intestine, which can influence germline epigenome via yolk provisioning^100,101—contribute to amplifying and passing on epigenetic information from neurons to germline.

Remarkably, we provided evidence that sensory signaling may affect germline development during stress. Specifically, disrupting GCY-35, a soluble guanylate cyclase involved in neuronal oxygen sensing and neuropeptide signaling, promotes germline maintenance and reproduction in RNAi mutants at high temperature, even without altering ambient oxygen levels. Serendipitously, we uncovered that RNAi may suppress oxygen response at high temperature in N2, either by modulating oxygen sensory perception directly or by more broadly affecting neuronal development and functions. This can be further studied by conditionally depleting rde-4 in a defined window to avoid potential developmental defects in constitutive mutants and examining neuronal differentiation states using the NeuroPAL system.¹⁰² Although we cannot exclude the possibility that impairing the oxygen-sensing circuit elicits a systemic stress response that indirectly affects germline physiology, or that GCY-35 may have non-canonical roles outside of direct oxygen sensing, our study strongly supports a model in which signaling initiated in sensory neurons can exert non-cell-autonomous effects on the germline developmental program. This neuron-to-germline interaction may, in some cases, trigger a lasting transgenerational epigenetic memory.⁵⁶ Hence, we propose that sRNAs integrate sensory information, promote developmental plasticity and reproductive adaptation, and potentially facilitate epigenetic buffering in fluctuating environments.^{103,104,105,106}

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Oded Rechavi (odedrechavi@gmail.com).

Materials availability

C. elegans strains generated in this study are available upon request.

Data and code availability

• •All NGS data are available through GEO under accession number GSE331410.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank Itai Reiger for their assistance with experiments. We thank Cori Bargmann (Rockefeller University) for providing introgressed strains carrying HW alleles of npr-1 and glb-5. Some graphics were created with Biorender.com. We are grateful to WormBase for providing valuable data and resources. Some strains were provided by the Caenorhabditis Genetics Center (CGC), which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). O.R. is grateful for the support of the Morris Kahn Foundation. C.K.E. was supported by an EMBO fellowship ALTF 6-2022. This work is funded by Eric and Wendy Schmidt Fund for Strategic Innovation Polymath Award 0140001000 (O.R.); European Research Council grant 335624 (O.R.); Israel Science Foundation 979/21 (Y.B.T.); and the US-Israel Binational Science Foundation 2023036 (Y.B.T.).

Author contributions

Conceptualization, C.K.E. and O.R.; methodology, C.K.E. and H.A.; investigation, C.K.E., H.A., S.W., A.M., H.S., L.K., M.V., G.T., S.A., H.G., R.P., O.A., and R.B.; visualization, C.K.E., H.A., and H.S.; funding acquisition, Y.B.T. and O.R.; supervision, C.K.E., M.d.B., Y.B.T., and O.R.; writing—original draft, C.K.E.; writing—review & editing, C.K.E., H.A., H.S., Y.B.T., and O.R.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse anti-MSP	DSHB	RRID: AB_760348
Rabbit anti-RAD-51	Gift from Nicolas Silva	N/A
Cy™3 AffiniPure® Donkey Anti-Mouse IgG	Jackson ImmunoResearch	RRID: AB_2315777
Cy™3 AffiniPure® Donkey Anti-Rabbit IgG (H+L)	Jackson ImmunoResearch	RRID: AB_2307443
Bacterial and virus strains
OP50	Caenorhabditis Genetics Center	N/A
HT115	Caenorhabditis Genetics Center	N/A
Chemicals, peptides, and recombinant proteins
Alt-R™ S.p. Cas9 Nuclease V3	IDT	Cat#1081058
Critical commercial assays
NEBNext® Small RNA Library Prep Set for Illumina®	New England Biolabs	Cat#E7330L
NEBNext® Ultra™ II Directional RNA Library Prep Kit for Illumina®	New England Biolabs	Cat#E7760, #E7765
Deposited data
Raw and analyzed sequencing data	This paper	GEO: GSE331410
AGO IP sequencing data	Seroussi et al.37	GEO: GSE208702
rde-4 RNA sequencing data	Posner et al.43	GEO: GSE124049
Experimental models: Organisms/strains
C. elegans strain: WM48 rde-4(ne299) III	Caenorhabditis Genetics Center	WM48
C. elegans strain: BFF337 rde-4(ne299) III (3X BC to N2)	This study	BFF337
C. elegans strain: N2	Caenorhabditis Genetics Center	WB Strain: N2
C. elegans strain: BFF496 rde-4(pig51) III	This study	BFF496
C. elegans strain: BFF497 rde-4(pig52) III	This study	BFF497
C. elegans strain: BFF499 LSJ1; rde-4(pig53) III	This study	BFF499
C. elegans strain: BFF509 CB4856; rde-4(pig54) III	This study	BFF509
C. elegans strain: BFF504 rde-4(ne299) III; kyIR1(V,CB4856>N2) V; qqIR1(X,CB4856>N2) X	This study	BFF504
C. elegans strain: BFF505 rde-4(ne299) III; qqIR1(X,CB4856>N2) X	This study	BFF505
C. elegans strain: DR466 him-5(e1490) V	Caenorhabditis Genetics Center	WB Strain: DR466
C. elegans strain: BFF401 rde-4(ne299) III; him-5(e1490) V	This study	BFF401
C. elegans strain: JK574 fog-2(q71) V	Caenorhabditis Genetics Center	WB Strain: JK547
C. elegans strain: BFF16 pigSi3[Psng-1::rde-4:SL2::yfp+cb-unc-119(+)] II; rde-4(ne299)	Posner et al.43	BFF16
C. elegans strain: BFF397 rde-4(ne299) III; sid-1(qt78) V	This study	BFF397
C. elegans strain: BFF431 pigSi3[Psng-1::rde-4::SL2:yfp+cb-unc-119(+)] II; rde-4(ne299) III; sid-1(qt78) V	This study	BFF431
C. elegans strain: BFF430 tph-1(mg280) II; rde-4(ne299) III	This study	BFF430
C. elegans strain: BFF438 mec-8(ok2043) I; rde-4(ne299) III	This study	BFF438
C. elegans strain: BFF468 rde-4(ne299) III; osm-9(ky10) IV	This study	BFF468
C. elegans strain: BFF476 rde-4(ne299) III; gpa-3(pk35) odr-3(n1605) V.	This study	BFF476
C. elegans strain: BFF477 che-1(luc174) I; rde-4(ne299) III	This study	BFF477
C. elegans strain: BFF487 rde-4(ne299) III; npr-1(ad609)X	This study	BFF487
C. elegans strain: BFF485 pigSi3[Psng-1::rde-4:SL2::yfp+cb-unc-119(+)] II; rde-4(ne299)III; qIs147 [sur-5::GFP] IV	This study	BFF485
C. elegans strain: BFF486 rde-4(ne299) III; qIs147 [sur-5::GFP] IV	This study	BFF486
C. elegans strain: BFF489 gcy-35(ok769) I; rde-4(ne299) III	This study	BFF489
C. elegans strain: BFF491 rde-4(ne299) III; gcy-33(ok232) V	This study	BFF491
C. elegans strain: BFF531 rde-4(ne299) III; peIs1713 [sra-6p::mCasp-1 + unc-122p::mCherry]	This study	BFF531
C. elegans strain: BFF362 rde-4(ne299) III; cmk-1(oy21) IV	This study	BFF362
C. elegans strain: BFF419 tax-2(ok3403) I; rde-4(ne299) III	This study	BFF419
C. elegans strain: BFF423 rde-4(uu53) III; him-5(e1490) V	This study	BFF423
C. elegans strain: BFF463 rde-4(ne299) III; egl-21(n476) IV; him-5(e1490) V	This study	BFF463
C. elegans strain: JK4846 qIs147 [sur-5::GFP] IV	Caenorhabditis Genetics Center	JK4846
C. elegans strain: PD2227 oxIs322 [myo-2p::mCherry::H2B + myo-3p::mCherry::H2B + Cbr-unc-119(+)] II. ccTi1594 [mex-5p::GFP::gpr-1::smu-1 3′UTR + Cbr-unc-119(+), III: 680195] III	Caenorhabditis Genetics Center	WB Strain: PD2227
C. elegans strain: BFF429 mjIs134 [mex-5p::gfp::h2b::tbb-2] II; rde-4(ne299) III	This study	BFF429
C. elegans strain: SX1263 mjIs134 [mex-5p::gfp::h2b::tbb-2] II	Caenorhabditis Genetics Center	SX1263
C. elegans strain: BFF467 rde-4(ne299) III; alg-3(tm1155) IV; him-5(e1490)V	This study	BFF467
C. elegans strain: AMJ912 jamSi28[Pmyo-3::rde-4(+)::rde-4 3′UTR] II; rde-4(ne301) III	Caenorhabditis Genetics Center	WB Strain: AMJ912
C. elegans strain: AMJ565 jamSi6 [Pnas-9::rde-4(+)::rde-4 3′UTR] II; rde-4(ne301) III; unc-119(ed3) III (?)	Caenorhabditis Genetics Center	AMJ565
C. elegans strain: WM49 rde-4(ne301)	Caenorhabditis Genetics Center	WB Strain: WM49
C. elegans strain: BFF555 reSi7 I [rgef-1p::TIR1::F2A::mTagBFP2::AID∗::NLS::tbb-2 3′UTR] (I: 5.32); rde-4(ne299) III; unc-31(rp166[GFP::TEV::AID∗::FLAG::unc-31]) IV	This study	BFF555
C. elegans strain: BFF554 tax-2(p694) I; rde-4(ne299) III	This study	BFF554
C. elegans strain: BFF572 gcy-35(db2020(gcy-35[framshift+stop codon])) I; rde-4 (ne299) III (line 1) (2X BC)	This study	BFF572
C. elegans strain: BFF573 gcy-35(db2020(gcy-35[framshift+stop codon])) I; rde-4 (ne299) III (line 2) (2X BC)	This study	BFF573
C. elegans strain: BFF563 gcy-35(db2020(gcy-35[framshift+stop codon])) I; npr-1(ok1447) X	This study	BFF563
C. elegans strain: RB1330 npr-1(ok1447) X	Caenorhabditis Genetics Center	WB Strain: RB1330
C. elegans strain: BFF523 npr-1(ok1447) X (2X BC of RB1330)	This study	BFF523
C. elegans strain: BFF542 rde-4(ne299) III; npr-1(ok1447) X	This study	BFF542
C. elegans strain: BFF547 rde-4(pig51) III; flp-19(ok2460) X	This study	BFF547
C. elegans strain: BFF550 rde-4(pig51) III; flp-21(ok889) V	This study	BFF550
C. elegans strain: BFF556 rde-4(pig51) III; flp-8(pk360) X	This study	BFF556
C. elegans strain: PHX11501 rde-4(syb11410 syb11501)	This study/SunyBiotech	PHX11501
C. elegans strain: BFF574 bmdSi348 [loxN::rgef-1p::FLP::P2A::H2B::2xmTurq2] I; rde-4(syb11410 syb11501) III	This study	BFF572
Oligonucleotides
rde-4 Alt-R gRNA #1:ACTTGGGGCACTGTCGAACT	IDT	N/A
rde-4 Alt-R gRNA #2: ATCTCTGGAATCATATGATA	IDT	N/A
rde-4 Alt-R HDR donor block: AGGCTGCTAAGGCT GTCTATCAAAAGACGCCAACTATATGGGTATGCCT CCAAATAATTGTAGTTAATAT	IDT	N/A
gcy-35 Alt-R gRNA: GCATGATTCTCACAACGCTC	IDT	N/A
gcy-35 Alt-R HDR donor block: GTGATGCCTATATGATTGTGG GCGGAGTTCCGGAGGGGAAG TTTGTCCAGAGCAGAGGTGAC TAAGTGATAAGCTAGCCGTTGT GAGAATCATGCAGAGCGAGTTCTCAATAT	IDT	N/A
Recombinant DNA
Plasmid: pL4440-RNAi control (HT115)	Vidal RNAi library	N/A
Plasmid: pL4440-gfp (HT115)	Vidal RNAi library	N/A
Software and algorithms
RNAlysis	Teichman et al.107	https://github.com/GuyTeichman/RNAlysis
ImageJ	Schindelin et al.108	https://imagej.net/ij/
GraphPad Prism 10	GraphPad	https://www.graphpad.com/
R/RStudio	R Core Team	https://www.r-project.org/
BioRender	BioRender	https://www.biorender.com/
DoubleBlind	Teichman et al.56	https://github.com/GuyTeichman/DoubleBlind
Inkscape v1.4	Inkscape	https://inkscape.org/
Zentracker	de Bono lab	https://github.com/wormtracker/zentracker; RRID SCR_022006

Experimental model and study participant details

C. elegans strains were maintained on standard Nematode Growth Medium (NGM) plates seeded with E. coli OP50 at 20^oC, unless stated otherwise. A complete list of strains used in this study is provided in the key resources table. Note that ne301 and ne299 alleles contain an identical lesion (see WBVar00090971)^27,28; we used the original designations to preserve a clear record.

Method details

Fertility assay

L4 animals, identified by the characteristic white crescent surrounding the developing vulva, were transferred to 25^oC. After ∼24 hours, individual day 1 adults were moved to fresh NGM plates seeded with a small (∼30μL) drop of OP50 that had been allowed to dry for 1–2 days. About 12 hours later, the number of laid eggs and unfertilized oocytes was quantified and the percent of unfertilized oocytes was calculated. Samples with fewer than 20 total eggs and unfertilized oocytes were excluded from analysis. For egl-21(n476) mutants, which display egg-laying defects, counts from three biological replicates were pooled. In all cases, the investigators were blinded to the genotypes.

Fertility assay on uniform bacterial lawn

To create uniform lawns, 40μL of OP50 bacteria was added onto 30mm NGM plates and spread evenly across the entire plate surface with a bacterial spreader. On the same day, an equal volume was seeded as a small drop on separate plates to serve as the standard/small lawn control. Plates were allowed to dry for 1–2 days and fertility assays were performed as described above.

RNAi

RNAi feeding clones were obtained from the Vidal library.¹⁰⁹ The bacteria were inoculated overnight at 37^oC in Lysogeny broth (LB) containing 50μg/ml carbenicillin. 1mM of IPTG was then added to the bacterial culture and 50–100μL was seeded onto 60mm NGM agar plates containing 1mM IPTG and 25μg/mL carbenicillin. After the seeded plates were dried for ∼24 hours, 5–10 L4 animals were placed on the RNAi plates. The progeny was then collected for analyses.

AID-Mediated degradation of UNC-31

NGM plates were supplemented with 1mM auxin (indole-3-acetic acid) prepared from a stock of 400mM dissolved in ethanol (EtOH). NGM plates supplemented with EtOH were used as a control. Auxin solution and auxin-containing plates were protected from light at all time. Mid-L4 animals (BFF555: reSi7 I [rgef-1p::TIR1::F2A::mTagBFP2::AID^∗::NLS::tbb-2 3′UTR] (I: 5.32); rde-4(ne299) III; unc-31(rp166[GFP::TEV::AID^∗::FLAG::unc-31]) IV) were transfer to seeded NGM plates containing auxin or EtOH and shifted to 25^oC. Fertility was assessed as described above.

RNA isolation

Total RNA was isolated from day-1 adults (∼72 hours at 20°C following L1 synchronization) using standard phenol-chloroform method with TRIzol™ (Invitrogen). RNA quality was accessed using Agilent 4150 BioAnalyzer instrument and High Sensitivity RNA ScreenTapes.

RNA-seq

To sequence the mRNA, RNA libraries were prepared using NEBNext® Ultra II Directional RNA Library Prep Kit for Illumina® coupled with NEBNext® poly(A) mRNA Magnetic Isolation Module. The cDNA libraries were pooled and paired-end sequencing was performed on the NextSeq 2000 platform.

For sRNA sequencing, we treated the RNA samples with RNA 5′ polyphosphatase (epicentre) and the libraries were prepared using NEBNext® Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs) or TruSeq Small RNA Library Prep Kit (Illumina) according to the manufacturer’s protocol. RNA ranging from ∼140 to 160nt was size-selected by gel extraction using 4% agarose E-Gel (Life Technologies). The pooled samples were the sequenced using the Illumina HiSeq 2500 instrument.

Bioinformatics

All bioinformatic analyses were performed using RNAlysis.¹⁰⁷ For sRNA-seq data, sRNA reads were aligned to PRJNA13758 CE11 genome assembly using ShortStack¹¹⁰ and aligned anti-sense reads were quantified using FeatureCounts (Data S2).¹¹¹ We included all reads with > 5 RPM without imposing any constraint on the length and 5′ nucleotide. For mRNA, we pseudo-aligned reads using Kallisto (Data S3).¹¹² We then performed differential expression analysis using DESeq2.¹¹³ For gene set enrichment analysis, log₂ (fold enrichment) scores were computed and the FDR for enrichment was calculated using 10,000 random gene sets identical in size to the tested group. For Gene Ontology (GO) and KEGG pathways analyses, Fisher’s exact tests were performed.

CRISPR/Cas9-mediated knockout of rde-4

CRISPR/Cas9 was performed as previously described.¹¹⁴ To generate rde-4 deletion, we used two crRNAs: ACTTGGGGCACTGTCGAACT and ATCTCTGGAATCATATGATA. The following homology-directed repair (HDR) donor was used: AGGCTGCTAAGGCTGTCTATCAAAAGACGCCAACTATATGGGTATGCCTCCAAATAATTGTAGTTAATAT. This generates 637bp deletion in rde-4. The results strains were backcrossed at least two times to remove potential background mutations.

CRISPR/Cas9-mediated knockout of gcy-35

A 43bp STOP-IN cassette, containing stop codons in all three reading frames and cause frameshift,¹¹⁵ was inserted into gcy-35 in rde-4(ne299) (BFF337). The sequences of the crRNA and HDR donor were GCATGATTCTCACAACGCTC and GTGATGCCTATATGATTGTGGGCGGAGTTCCGGAGGGGAAGTTTGTCCAGAGCAGAGGTGACTAAGTGATAAGCTAGCCGTTGTGAGAATCATGCAGAGCGAGTTCTCAATAT, respectively. The results strains were backcrossed twice to remove potential background mutations.

Low oxygen exposure experiment

Wild-type and rde-4(ne299) animals carrying germline-specific transgene mjIs134 [mex-5p::gfp::h2b::tbb-2] were maintained under normoxia at 20^oC before the experiments. L4 worms were picked onto several NGM plates seeded with OP50. Half the plates were placed in a glove box (Coy Laboratory Products) maintained at either 7% or 1% oxygen (balanced with nitrogen) at 25°C for 24 hours. Control plates were placed in a 25^oC incubator under normoxia (∼21% oxygen) for the same duration. Imaging was then performed on day-1 adults using Nikon Ti2E Imaging System with a Plan Apo λ 20x/0.75 DIC 1 air PFS objective. Their germline morphology – the accumulation of unfertilized oocytes (“stacked”) accompanied by the absence of fertilized embryos in the proximal gonad – was scored blind using Fiji¹⁰⁸ and DoubleBlind (https://github.com/GuyTeichman/DoubleBlind). Note that this assay avoids the plate transfers used in our standard protocol (see above) and therefore minimizes fluctuations in oxygen levels.

Locomotion assays

60μl of OP50 was seeded on low peptone NGM plates (0.13% wt/vol bactopeptone) 2 days before the assay. L4 animals were picked 24 hours before the experiment and incubated at 25 or 20^oC. A rubber stamp was used to remove the edge of the bacterial lawn on the day of the assay. 20–25 animals were picked onto the lawn and left for 10mins before starting the assay. A 10 x 10 x 0.4 mm PDMS chamber was placed on top of the bacterial lawn and defined gas mixtures were delivered to the chamber at 1.25ml/min using a syringe pump (PHD 2000, Harvard Apparatus). Before recording behavior, we acclimated animals to 7 % oxygen for 2 mins. For standard assays, animals were exposed to 2 mins of 7%, 2 mins of 21%, and 2 mins of 7% oxygen. Video recordings were acquired at 2 frames per second (fps) using a Grasshopper camera (Point Grey) mounted on a stereomicroscope (Leica MZ6 or MZ7.5). Videos were analyzed and animal speed was calculated using custom-written MATLAB software (Zentracker: https://github.com/wormtracker/zentracker; RRID SCR_022006). Average speed values were extracted using Metaverage, a custom-written MATLAB software. Average speed during 30 second intervals at the end of the first 2 mins at 7% and at the end of the 21% oxygen period were used for statistical comparisons.

Antibody and DAPI staining

All experiments were performed on day-1 adults. For experiments at 25^oC, L4 animals were shifted to 25^oC as described above. Dissected gonads were fixed with 3.5% formaldehyde for 10 mins and washed at least twice with PBS-T (0.1% Tween-20 in PBS buffer). Tissues were then permeabilized in 0.25% PBS-Triton for 20 mins, followed by blocking in 0.5% BSA in PBS-T for 1 hour with gentle shaking. Next, primary antibodies – mouse anti-MSP (Developmental Studies Hybridoma Bank) and rabbit anti-RAD-51 (gift from Nicolas Silva) – were diluted 1:1000 in PBS-T and applied for 1 hour with gentle agitation. After two washes with PBS-T, secondary antibodies (Jackson ImmunoResearch anti-rabbit and anti-mouse; 1:200 in PBS-T) were added and incubated for 1 hour. Samples were then washed for two more times and were incubated with DAPI (0.02% in PBS-T) for 10 mins, followed by two more washes with PBS-T. Finally, samples were mounted on glass slides in VECTASHIELD® Antifade Mounting Medium (Vector Laboratories, Cat# H-1000) prior to imaging.

Quantification and statistical analysis

Statistical analyses were conducted using R software v4.2.3 and GraphPad Prism 10. For comparisons between two groups, non-parametric unpaired Wilcoxon rank sum tests were used, unless stated otherwise. For experiments involving multiple groups, Kruskal-Wallis test followed by pairwise Wilcoxon rank-sum tests were performed. Multiple comparison corrections using the Benjamini-Hochberg method were applied where appropriate.

Published: July 6, 2026