TDP-1 and FUST-1 co-inhibit exon inclusion and control fertility together with transcriptional regulation

Abstract Gene expression is a multistep process and crosstalk among regulatory layers plays an important role in coordinating gene expression. To identify functionally relevant gene expression coordination, we performed a systematic reverse-genetic interaction screen in C. elegans, combining RNA binding protein (RBP) and transcription factor (TF) mutants to generate over 100 RBP;TF double mutants. We identified many unexpected double mutant phenotypes, including two strong genetic interactions between the ALS-related RBPs, fust-1 and tdp-1, and the homeodomain TF ceh-14. Losing any one of these genes alone has no effect on the health of the organism. However, fust-1;ceh-14 and tdp-1;ceh-14 double mutants both exhibit strong temperature-sensitive fertility defects. Both double mutants exhibit defects in gonad morphology, sperm function, and oocyte function. RNA-Seq analysis of double mutants identifies ceh-14 as the main controller of transcript levels, while fust-1 and tdp-1 control splicing through a shared role in exon inhibition. A skipped exon in the polyglutamine-repeat protein pqn-41 is aberrantly included in tdp-1 mutants, and genetically forcing this exon to be skipped in tdp-1;ceh-14 double mutants rescues their fertility. Together our findings identify a novel shared physiological role for fust-1 and tdp-1 in promoting C. elegans fertility and a shared molecular role in exon inhibition.


INTRODUCTION
Eukaryotic gene expression requires coordination across multiple layers of regulatory control, including transcription, RNA processing, and translation.Two major classes of proteins responsible for this gene expr ession r egulation are transcription factors (TFs) and RNA binding proteins (RBPs).Regulatory activities for individual TFs and RBPs have been well described, but less is known about how TFs and RBPs might coordinately control gene expression across multiple regulatory layers.Ne v ertheless, a growing body of recent evidence demonstrates e xtensi v e crosstalk between transcriptional and post-transcriptional factors occurs to regulate gene expression (1)(2)(3)(4)(5)(6).
RBPs regulate many aspects of RNA processing including pre-mRN A splicing, mRN A export and localization, and transla tion ( 7 ).RBP d ysfunction is especially notable in the nervous system, and RBP mutations have been implicated in multiple neurodegenerati v e diseases (8)(9)(10)(11).For example, the RBPs TDP-43 and FUS are involved in se v eral RNA-related functions, including splicing and RNA transport ( 12 , 13 ).Mutations in either RBP are directly linked to Amyotrophic Lateral Sclerosis (ALS) cause their mislocaliza tion and aggrega tion in the cytoplasm, leading to progressi v e degeneration of neurons (14)(15)(16)(17).The diseaseassociated roles of RBPs such as TDP-43 and FUS have been e xtensi v ely studied, but in many cases the physiological functions for these RBPs outside of disease context remain unresolved.Understanding how RBPs play a role in essential cellular functions and in the context of global gene e xpression coor dination will be key for understanding and treating such diseases.
To identify functionally important TF-RBP gene expression coordination, we set out to systematically test for genetic interactions between TF and RBP mutants.Screens for genetic interactions, in which a phenotype occurs in a double mutant that is not predicted based on the single mutant phenotypes, have a rich history of identifying genes with related activities and / or redundant functions ( 18 , 19 ).In single-celled organisms such as bacteria and yeast, genetic interaction analysis has been carried out at genomewide scale, re v ealing hundreds of thousands of synthetic interactions in which a double mutant has a fitness greater than expected (positive interaction) or less than expected (negati v e interaction) based on the single mutant fitness phenotypes ( 20 , 21 ).
In the nematode Caenorhabditis elegans , we recently employed CRISPR / Cas9 to systematically delete e volutionarily-conserv ed, neuronally-e xpressed RBPs using homology-guided replacement to insert heterologous GFP fluorescent markers in place of the deleted gene ( 22 , 23 ).These CRISPR / Cas9-generated RBP mutants enabled us to conduct a systematic pairwise genetic interaction screen across neuronal RBPs in C. elegans .We identified multiple novel synthetic interactions and re v ealed pre viously-une xplored physiological functions for se v eral RBPs.Here, we aimed to a ppl y this technolo gy to investiga te coordina tion of gene expression across regulatory layers, by conducting a pairwise genetic interaction screen between neuronally-enriched RBPs and TFs.
To do so, we generated all possible double-mutant TF-RBP combinations of 10 RBP and 11 TF gene deletion mutants, creating a total of 110 double mutants.We identified unexpected phenotypes in several double mutants, rev ealing e xtensi v e functional interactions between TFs and RBPs.One such synthetic phenotype was reduced fertility in tdp-1; ceh-14 and fust-1; ceh-14 double mutants.tdp-1 and fust-1 are the C. elegans homologs of TDP-43 and FUS, and mutations in both RBPs have been implicated in ALS ( 9 , 17 , 24 ).Both tdp-1; ceh-14 and fust-1; ceh-14 double mutants exhibit reduced egg production, decreased sperm efficacy, and gonad migration defects.As tdp-1, fust-1, and ceh-14 single mutants do not exhibit the same striking fertility phenotype as these double mutants, our findings identify a potential coregulatory role for these genes in gonad and sperm de v elopment.We find a shared role of fust-1 and tdp-1 in inhibiting exon inclusion, and identify a cassette exon in pqn-41, inhibited by tdp-1 , that contributes to the fertility defects in tdp-1; ceh-14 double mutants.Our findings thus unco ver no vel physiological functions for fust-1 and tdp-1 , in the specific context of a ceh-14 mutant background, and shed light on their shared molecular roles.

C . eleg ans strains and maintenance
All C. elegans strains were cultured on Nematode Growth Media (NGM) plates seeded with Esc heric hia coli.Strains were maintained at 20 • C unless otherwise stated.Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440), and strains PHX6090 and PHX3345 were generated by SunyBiotech.Double mutant strains were created and confirmed by visualization of GFP markers and by PCR.See Table 1 .

Competitive fitness assays
Pairwise competiti v e fitness assays were performed as previously described to establish the relati v e fitness for each single mutant ( 22 ).Briefly, four L4 larvae of each genotype were placed together on a seeded NGM plate and incubated for 5 days at 25 • C. The fraction of each mutant on the plate after 5 days was calculated to generate a fitness value relati v e to wild-type fitness for each mutant using the formula F = (# mutant / # total) / 50%.
To determine double mutant fitness each double mutant was assayed in a pairwise competition assay with the transcription factor mutant used in the cross.Expected fitness was equivalent to the fitness of the RNA binding protein mutant in the cross.To identify double mutants with unexpected fitness, the expected fitness value for each double mutant was subtracted from its observed fitness.Synthetic fitness effects ( ε ) in the double mutants were calculated by F obs − F exp .Our threshold for significance was | ε | > 0.4, and all assays were completed in triplicate.

Larval growth assay
Worms were synchronized by standard bleaching procedure to obtain a population of L1 worms for each genotype ( 25 ).Synchronized L1 larvae were then plated onto seeded NGM plates and cultured for 48 h at 20 • C. De v elopmental stages of the worms on each plate were then assessed.

Fluor escence microscop y
Images were obtained on a Zeiss Axio Imager.Z1 microscope with a 20 × objecti v e and DIC settings for white light.Excitation for fluorescence was provided by an X-Cite series 120 Q lamp.Wavelengths for RFP = excitation 592 nm, emission 614 nm, GFP = excitation 488 nm, emission 509 nm.Exposure times were 650 ms (RFP) and 500 ms (GFP).Images were processed using ImageJ.

DAPI gonad imaging
Worms were collected into microcentrifuge tubes in batches at each larval stage and at day 1 of adulthood.Whole worms were then fixed and stained with 4 ,6-diamidino-2-phenylindole (DAPI).Briefly, worms were washed three times with 0.01% Tween-PBS, then frozen in 1 mL of methanol at −20 • C for 5 min.Then, methanol was removed and worms rinsed with 1 ml of 0.1% Tween-PBS. 1 l of clec-190 (syb3345) SunyBiotech deletion PBS, and incubated in the dark at room temperature for 5 min.Tubes were then rinsed with 0.1% Tween-PBS once mor e. Tubes wer e centrifuged and all solution r emoved from worms, then 10 l of 75% glycerol added.Worms in glycerol were then placed on agar pads on microscope slides for imaging.At least 70 total animals were scored for each genotype, across fiv e independent biological replicates in total.Both posterior and anterior gonad arms were scored, and defects in either resulted in the animal being scored as defecti v e. Howe v er, most of the time in double mutants ( > 90%) either both gonad arms were normal or both were defecti v e.

Uterine egg retention
Egg retention was measured in hermaphrodites on day 1 of adulthood.Worms were placed in 4 • C refrigerator for 5-10 min to slow movement down, then examined under the microscope.Total eggs present in the uterus were counted.

Lifetime egg-laying and brood size assays
Six L4 worms were placed on a seeded NGM plate and incuba ted a t either 20 • C or 25 • C. The f ollowing da y, and every subsequent day until egg production stopped, the six adults wer e transferr ed to new seeded plates.The number of progeny on each plate was counted and divided by the total number of adults to determine average egg production per day per worm.Total brood size was quantified as the sum total of eggs produced over lifespan per worm.

Male mating efficiency
To assay male mating efficiency, four young adult males wer e pair ed with four L4 hermaphrodites, and the worms were kept at 20 • C for 24 h to allow time for mating to occur.After 24 h, the adults were removed from the plate, and progeny were gi v en another 48 hours to de v elop.Progeny were then counted, and percent of male-produced cross progeny out of the total progeny were scored.Mutant males with CRISPR deletions express GFP in either the body wall muscle or the pharynx, so this fluorescence was used to identify cross progeny.For wild-type assays, males carrying myo-2::RFP which expresses bright RFP in the pharynx were used to allow scoring of cross progeny.
To measure male mating in the absence of wild-type her maphrodite sper m, f our y oung adult males were instead paired with four L4 fog-2 (q71) hermaphrodites with feminized germlines and a complete lack of sperm.For these assays, mutant males were paired with fog-2 hermaphrodites until egg production stopped, and adults wer e transferr ed to new plates to prevent overcro w ding and starvation as breeding continued.After hermaphrodites no longer continued to produce eggs, adults wer e r emoved and the total number of progeny was counted.This total was divided by the number of adult hermaphrodites present on the assay ( 4 ) to gi v e the average brood size produced per worm.As a control, these assays were sim ultaneousl y conducted with wild-type males.

P air ed brood size assay
Four L4 wild-type males were paired with four L4 mutant hermaphrodites on a single plate, and pairs were kept paired for multiple days until egg-laying was complete.Every day, adults were transferred to new plates, and progeny left on each plate was counted to measure the total brood produced.As a control, unpaired brood assays were carried out at the same time.

RNA sequencing and analysis
Total RNA was extracted from synchronized whole animals at the fourth larval stage (L4, just prior to adulthood) using Tri reagent according to manufacturer's protocol (Sigma Aldrich).Three biological replicates were extracted per genotype.mRNA was purified from each sample using NEBNext ® Pol y(A) mRN A Magnetic Isolation Module, and cDNA libraries were prepared using NEB-Next ® Ultra ™ II RNA Library Prep Kit f or Illumina, f ollowing kit protocols.Libraries were sequenced on Illumina HiSeq 2000, paired-end 150 bp reads.At least 87% of reads from each individual sample had Q scores of ≥30.Reads were then mapped to the worm genome (versionWBcel235 using STAR version 2.5.3a) ( 26 ).Each biological replicate had at least 26 million reads per sample, and the mapping rate to the worm genome was ≥72% for each sample.Genespecific counts were tabulated for each sample using HT-Seq (0.9.1) ( 27 ) and sta tistically-significant dif ferentially expr essed transcripts wer e identified with DESeq2 (1.36.0) ( 28 ).The Junction Usage Model (2.0.2) was used to identify differentially spliced isoforms in experimental samples compared to wild type controls and quantify their expression le v els by computing the PSI (difference of Percent Spliced Isoform) ( 29 ).Alternati v e 3 splice site, alternati v e 5 splice site, skipped 'cassette' exon, and intron retention splicing e v ents were then analyzed to compare PSI in fust-1; ceh-14 and tdp-1; ceh-14 with their single mutants to identify splicing d ysregula tion.Raw fastq files are available at the NCBI SRA (PRJNA862903) and additional data is available via GEO (GSE230025).

Reverse transcription PCR
Relati v e abundances of splicing isoforms of sav-1 , and pqn-41 were determined by RT-PCR to confirm RNA seq results, using qScript ® XLT One-Step RT-PCR Kit.The kit and r eagents wer e used following the kit reaction protocol.Densitometry was performed using non-sa tura ted gel ima ges.Ima geJ was used to draw a rectangular ROI around the extent of each band, and an equal-sized backgr ound ROI (fr om a lane with no PCR product) was also obtained.PSI was calculated using the intensity values as follows: 100 × (included bandbackgr ound) / ((included band-backgr ound)+(skipped band-background)).Primer sequences used were as follows: sav-1 F -GACTT CATT CAAGAT CTA CGG , sav-1 R -CACTGGGAA GA GTTTGAA GCG, pqn-41 exon 18 F -A CTA CGCCTGCAA CAA CGTCG , pqn-41 exon 21 R -AGCT GCT GTT GAACTT GTT GAGC

Genetic interaction screen identifies TF-RBP pairs causing synthetic fitness defects
To identify regulatory crosstalk with important functional consequences we performed a genetic interaction screen between TF and RBP mutants in C. elegans , focusing on e volutionarily-conserv ed RBPs and TFs expressed in the nervous system.To facilitate the generation of double mutants, we used existing deletion alleles which can be genotyped by simple PCR, as well as CRISPR / Cas9-mediated deletions in which we inserted a traceable fluorescent maker into the deletion locus, enabling in vivo monitoring of the genotype ( 30 ) (See methods for list of genotypes used).We generated all possib le doub le-mutant combinations of 10 RBPs and 11 TFs.
To test for genetic interactions, we measured relati v e fitness using a simple and quantitati v e competiti v e fitness assay ( 22 , 30 ).In this assay, equal numbers of stage-matched mutant and wild-type worms are grown together on a single growth plate.The worms are gi v en fiv e days to de v elop, eat available food, and reproduce for multiple generations (Figure 1 A).Then the relati v e proportions of mutant and wildtype worms are quantified, assigning a value to the fitness of each genotype.A mutant with an identical fitness to wildtype would grow and reproduce at the same rate as wildtype worms, resulting in a population of 50% mutants and 50% wild-type.This would yield a fitness value of 1 (Figure 1 A), and increased or decreased fitness would result in values greater than or less than 1, respecti v ely.Competiti v e fitness assays can identify mutants with a variety of underlying phenotypes, including lethality, de v elopmental defects, reproducti v e defects, and behavioral defects ( 21 , 22 ).
Each RBP and TF single mutant strain was first assayed against wild-type worms to establish their respecti v e relati v e fitness values (Figure 1 B, C).The fitness values of the 10 RBPs and 11 TFs we assayed range from strong decreases in fitness to mild increases in fitness (Figure 1 B, C).Several mutants with known behavioral defects, including unc-86 and mec-8 ( 31 , 32 ), have significantly lower fitness than wild-type.We identified a few single mutants with novel fitness phenotypes not predicted by previously-described phenotypes, for example a reduction of fitness in tab-1 mutants (Figure 1 C).We also found that a few mutants, including aptf-1 and f ox-1 , modestl y outperform wild-type worms in our assays and have fitness values greater than 1 (Figure 1 B, C).
After establishing baseline fitness values for single mutants, each RBP mutant was crossed to each TF mutant to obtain all possible RBP; TF mutant combinations, yielding a total of 110 RBP; TF double mutants.To systematically identify genetic interactions between the RBPs and TFs, we conducted competiti v e fitness assays in which each doub le mutant was competed against one of its constituent single mutants.Assuming no genetic interaction, when an RBP; TF double mutant is competed against its constituent TF mutant, the effect of the TF mutation on fitness should be equal for both the single and the double mutant.Ther efor e, the measured fitness of the RBP; TF should be equal to the fitness of the constituent RBP m utant w hen competed against wild type.For example, msi-1; unc-86 double mutants were competed against unc-86 mutants, and the expected fitness value was equal to the fitness of msi-1 single mutants competed against wild type (Figure 1 D).In this case, the observed fitness shows no difference from the expected fitness of the double mutant, and is ther efor e not considered a genetic interaction (Figure 1 D).
Any deviation in the observed fitness from the expected fitness value ( ε = observed -expected) constitutes a synthetic effect on fitness, which is to say, an effect stronger than expected based on additive effects of the two single mutations.Such a result signifies a synthetic genetic interaction.A positi v e value indicates a positi v e interaction, while a negati v e value constitutes a negati v e interaction.As an e xample, exc-7; ets-5 double mutants were competed with ets-5 single mutants.The expected outcome was for the measured exc-7; ets-5 double mutant fitness value to be similar to that of exc-7 competed against wild type, but instead exc-7; ets-5 double mutants have significantly lower fitness than expected (Figure 1 E).We set a conservati v e threshold of an absolute value of 0.4 or greater change in fitness (| ε | > 0.4) to be considered a strong genetic interaction.Most RBP; TF double mutants do not exhibit substantial deviations from their expected fitness value, indicating that most TFs and RBPs do not act synthetically in ways that result in changes in fitness.Howe v er, we identified eight RBP; TF double mutants with strong synthetic effects (Figure 1 F, Supplementary Figure S1A).

aptf-1; fox-1 double mutants cause developmental delay
Performance in the competiti v e fitness assay depends on the ability of worms to de v elop, survi v e , reproduce , and consume f ood, competing f or r esour ces with other worms on the plate.Ther efor e, some of the underlying phenotypes that could directly impact fitness include changes in the ability to feed, move, develop and reproduce.For each double mutant tha t genera ted a significant synthetic fitness effect, where | ε | > 0.4, w e follow ed up with multiple assays to determine underlying phenotypes that contributed to a decrease or increase in fitness.
In one interesting case, we found that apft-1; fox-1 double mutants exhibit a strong negative synthetic fitness effect, wher e the measur ed fitness is much lower than expected (Figure 2 A).The RBP fox-1 is a key regulator of C. elegans sex determination, while aptf-1 is a neuronal TF important for sleep behavior ( 33 , 34 ).Both factors are highly conserved, but the loss of either fox-1 or aptf-1 in a single mutant did not result in a measurable fitness deficit (Figure 1 B, C).In the aptf-1; fox-1 double mutant, we found moderate but significant reductions in egg-laying rate and pumping rate compared to aptf-1 and fox-1 single mutants (Supplementary Figure S1).Upon further investigation, we noticed tha t apft-1; fo x-1 worms seemed to exhibit slower than normal growth.We measured de v elopment time from larval stage L1 to L4 and confirmed that apft-1; fox-1 double mutants experience a significant delay in de v elopmental timing (Figure 2 B, C).48 h after hatching, when wild-type worms have reached the L4 larval stage, the majority of apft-1; fox-1 double mutants are still L3 (Figure 2 B).These effects are  C, differences in brood size are subtle ( C ), but at 25 • C both fust-1; ceh-14 and tdp-1; ceh-14 produce significantly smaller brood sizes than wild-type, while fust-1 tdp-1 double mutants do not exhibit a reproducti v e defect ( D ).Asterisks indicate significant difference from wild-type, P < 0.05, ANOVA.(In C, P = 0.022; in D, P = 0.0031 and 0.0039 for fust-1; ceh-14 and tdp-1; ceh-14 , respecti v el y). ( E ) Pro geny were counted e v ery day, and brood sizes for tdp-1; ceh-14 and fust-1; ceh-14 are consistently lower than wild-type and single mutants.Asterisk at day 2 indicates egg production of both fust-1; ceh-14 and tdp-1; ceh-14 was significantly lower than that of wild-type worms, ANOVA P < 0.05.( F ) tdp-1 and fust-1 exhibit negative genetic interactions with ceh-14 , but not with each other, to affect C. elegans brood size.
structural and functional commonalities and the roles of both RBPs in the context of disease have been well-studied (37)(38)(39).In ALS, mutations in TDP43 or FUS cause them to mislocalize and aggregate in the cytoplasm, depleting them from the nucleus and disrupting their function (40)(41)(42)(43)(44)(45).Howe v er, their roles under non-diseased conditions are less understood.
In contrast with mammals, where loss of function of either TDP-43 or FUS is fatal ( 46 , 47 ), loss of tdp-1 or fust-1 in C. elegans does not cause a strong phenotype, and indeed our tdp-1 and fust-1 mutants have fitness values indistinguishable from wild-type (Figure 1 B).This gives us the opportunity to investigate the molecular functions of tdp-1 and fust-1 , taking advantage of the ceh-14 mutant background to uncover essential roles for tdp-1 and fust-1 in fitness.
ceh-14 mutants display a slight decrease in fitness compared to wild-type, but have no strong visible phenotype (Figures 1 C, 3 A).Only when tdp-1 or fust-1 mutations are combined with ceh-14 mutations does a strong fitness defect occur (Figure 3 A, B).One readily discernible commonality between fust-1; ceh-14 and tdp-1; ceh-14 double mutants is a reduced progeny count, based on the observation that plates of these strains appear less cro w ded and take longer to consume all the available food on a plate.We quantified progeny produced per worm at both 20 • C and at the mildly stressful tempera ture 25 • C ( 48-50 ).W hile ther e is a modest decr ease in brood size at 20 • C, the defect becomes more pronounced at 25 • C (Figure 3 C-E).We measured progeny produced per day to further characterize the rate of reproduction in double mutants.We found that tdp-1; ceh-14 and fust-1; ceh-14 have consistently lower rates of reproduction than wild-type worms, with significantly fewer pr ogeny pr oduced on day 2 of adulthood compared to wild type (Figure 3 E).None of the constituent single mutants exhibit this strong defect (Figure 3 C, D).
To confirm that the nature of this interaction between tdp-1 or fust-1 and ceh-14 is due to on-target mutations and not background effects, we generated double mutants using alternate alleles ( 22 , 51 ).These new double mutants recapitulated the fertility defect, confirming the synthetic phenotypes in tdp-1; ceh-14 and fust-1; ceh-14 are due to on-target TF and RBP mutations (Supplementary Figure S2A).
tdp-1; ceh-14 and fust-1; ceh-14 double mutants have fertility defects, but tdp-1 fust-1 double mutants do not (Figure 3 D).Furthermore, triple mutants fust-1 tdp-1; ceh-14 do not have decreased fertility compared to double mutants fust-1; ceh-14 or tdp-1; ceh-14 (Supplementary Figure S2B).Together this indicates that tdp-1 and fust-1 genetically interact with ceh-14, but not with each other, to coordinately affect reproduction in C. elegans (Figure 3 F) .This suggests that the reproducti v e defects of both double mutants might stem from a shared underlying dysfunction in which the activity of both tdp-1 and fust-1 are required in conjunction with ceh-14 .One plausible mechanistic scenario would be that both tdp-1 and fust-1 regulate a gene at the le v el of RN A processing, w hile ceh-14 regulates a second gene at the le v el of transcription, and together both gene targets are r equir ed for fertility.In sum, we find that tdp-1 and fust-1 , whose human counterparts are implicated in shared disease states, also share similar physiological roles in C. elegans .This role in promoting fertility is only re v ealed in the context of the ceh-14 mutant background.

fust-1; ceh-14 and tdp-1; ceh-14 double mutants cause gonad development defects
A reduction in progeny count could be caused by a number of underlying phenotypes, including a physical inability to push eggs out of the vulva, a defect in gonadogenesis, or deficient gametes.We tested each possibility to determine the underlying causes of the double mutant phenotype.Worms with a mechanical defect in egg-laying retain eggs in the uterus ( 52 , 53 ).We quantified uterine eggs in two-day-old adults and found no significant difference between double mutant and wild-type worms (Supplementary Figure S3A).
We ne xt e xamined the gonad to see if there were any differences in de v elopment, using DAPI (4 ,6-diamidino-2-phenylindole) nuclear stain to visualize gonads within whole worms.During normal gonad de v elopment the distal tip cell (DTC) guides migration of each of the two symmetrical C. elegans gonad arms.The DTC guides the de v eloping gonad out from the ventral midbody, then makes one dorsal turn, followed by a second turn towards the midbody ( 54 , 55 ) (Figure 4 A).This migrating tissue recei v es a signal to stop when the DTC crosses the midbody and reaches the vulva ( 56 , 57 ).The uterine cells undergo a characteristic outgrowth, expanding and setting up the uterus centered around the vulva (58)(59)(60).
In a wild-type adult, there is typically a gap between the DTCs of the anterior and posterior arm, pre v enting the gonad arms from overlapping and creating ample space for the uterus (Figure 4 A).When maintained at 25 • C, fust-1; ceh-14 and tdp-1; ceh-14 double mutants present a variety of defects in gonad formation, the most common of which is apparent overmigration of the distal gonad tip in about 30% of double mutant adults (Figure 4 A, Supplementary Figure S3C).The degree of overmigration varies from slight crossing of the anterior and posterior tips to more se v ere cases in which an overmigrated tip infiltrates the uterus or the opposite gonad arm (Figure 4 A).We found no discernable differences in gonad morphology in larval stage worms, and overgrown distal arms are not observed until adulthood (Figure 4 B, Supplementary Figure S3B).This suggests that rather than a dysfunction during larval gonad de v elopment, there may be a disruption in stop signaling as the distal tip reaches the vulva.Previous studies have identified similar overmigration defects in C. elegans in response to de v elopment under stress, including under conditions of changing temperature ( 61 ), which could be related to the temperaturesensiti v e gonad de v elopment phenotypes observ ed in our double mutants.

fust-1; ceh-14 and tdp-1; ceh-14 double mutants cause gamete defects
C. elegans hermaphrodites produce both male and female gametes and can reproduce by self-fertilization or by mating with a male.The defects we observed in self-fertilizing hermaphrodites (Figure 3 ) could thus stem from defects in male gametes, female gametes, or both.To determine the functionality of the male and female gametes, we conducted reciprocal mated brood assays (Supplementary Figure S4).First, we mated mutant males with wild-type hermaphrodites and quantified the proportion of crosspr ogeny versus self-pr ogeny to determine the efficiency of mutant male sperm.If there are no sperm defects in mutant males, the proportion of cross-progeny produced should be similar to that produced by wild-type males.At 20 • C, ∼40% of progeny from wild-type male crosses are cross progeny, and at 25 • C the proportion is ∼15% (Supplementary Figure S4A, B).In contrast, the proportion of cr oss pr ogeny fr om double mutant males is ∼20% at 20 • C and ∼0% at 25 • C.These results suggest that tdp-1; ceh-14 and fust-1; ceh-14 double mutant male sperm are partially defecti v e.
In the abov e crosses, doub le mutant male sperm had to compete with wild-type sperm harbored by the selffertile hermaphrodite.When wild-type males are crossed to hermaphrodites, male sperm is able to outcompete hermaphrodite self sperm for fertilization of oocytes ( 62 , 63 ).To test whether double-mutant male sperm is fertile in the absence of competition from hermaphrodite sperm, we mated double mutant males with feminized fog-2 (q71) hermaphrodites which are unable to generate sperm ( 64 ).Ther efor e, all pr ogeny pr oduced w hen f og-2 hermaphrodites mate with males are cross progeny.Similar to the findings from the previous male mating assays (Supplementary Figure S4), we observe a reduction in brood size for double mutant males crossed with feminized hermaphrodites, with the defect more pronounced at 25 • C (Figure 4 C).These findings indicate that double mutant male sperm is defecti v e e v en in the absence of competition from wild-type hermaphrodite sperm.
To determine oocyte viability, wild-type males were mated with double mutant hermaphrodites.In self-fertile hermaphrodites, the total number of progeny produced is limited by the number of sperm generated.Ther efor e, if hermaphrodites are mated with males, the increased availability of sperm from the male will significantly increase the total progeny produced ( 65 ).Indeed, we observe tha t ma ting with a male more than doubles wildtype brood size (Figure 4 D).Howe v er, when doub le mutant hermaphrodites are paired with wild-type males, there is no significant increase in brood size (Figure 4 D).Together these da ta indica te tha t fust-1; ceh-14 and tdp-1; ceh-14 double mutants are defective in both male (sperm) and female (oocyte) gametes.

Expression of tdp-1 , fust-1 and ceh-14 overlaps in the spermatheca
To visualize where tdp-1, fust-1 and ceh-14 are expressed in C .elegans, w e endo genousl y tagged each gene using CRISPR / Cas9.fust-1 and tdp-1, both tagged with RFP, exhibit nuclear expression in many tissues, including neurons, muscles , intestine , and the gonad.The expression of ceh-14 , which we tagged with GFP, is limited to neurons and the spermatheca (Figure 4 E).As previously described, we find that it is expressed in a handful of neurons including se v eral in the head and tail ( 66 ).Within the gonad, ceh-14 exhibits nuclear expression specifically in the membrane of the spermatheca, which houses the sperm and initiates ovulation and fertilization of oocytes ( 67 ) (Figure 4 E).fust-1 and tdp-1 are also expressed in the nuclei of these cells (Figure 4 F-G).We hypothesize that this overlapping expression in the spermatheca could be a source of their combinatorial effect on reproduction.
To investigate the development and morphology of the spermatheca in fust-1; ceh-14 and tdp-1; ceh-14 , we crossed each double mutant with a strain containing a spermatheca GFP reporter.Both spermathecae are present in fust-1; ceh-14 and tdp-1; ceh-14 adults, and they appear structurally similar to those of wild-type worms (Supplementary Figure S5).We did not see any defects in the spermathecal-uterine valve or the neck of the spermatheca, both of which are indicati v e of spermatheca dysfunction (68)(69)(70).This suggests that the defect in reproduction observed in our double mutants is not explained by obvious defects in spermatheca development or morphology.

Distinct transcriptional and post-transcriptional networks in double mutants
To investigate the gene regulatory networks controlled by the three factors, we analyzed the transcriptomes of tdp-1; ceh-14 and fust-1; ceh-14 double mutants, as well as the constituent single mutants.At the le v el of gene expression, double mutants display changes in the expression of hundreds of genes compared to wild-type animals (Figure 5 A, Supplementary Figure S6A-C, Supplementary Figure S7A).Such gene expression changes could be the result of (i) losing a single regulatory factor, (ii) additi v e effects of losing both factors or (iii) synthetic effects of losing both factors.To distinguish among these possible scenarios, we first compared gene expression changes between single mutants and double mutants.Linear r egr essions show that ceh-14 accounts for the majority of gene expression changes observed in tdp-1; ceh-14 double mutants (Figure 5 B), while tdp-1 accounts for very few gene expression changes in the double mutant (Figure 5 C, Supplementary Table S1).Likewise, ceh-14 accounts for the majority of gene expression changes observed in fust-1; ceh-14 mutants (Supplementary Figure S6B).
Since most gene expression changes in the double mutant are accounted for by ceh-14 regulation, this suggests that very few gene expression changes are regulated in an additi v e or synthetic manner by tdp-1 and ceh-14 .One notab le e xception is the gene clec-190 , whose expression is unchanged in single mutants, but strongly downregulated in tdp-1; ceh-14 and modestly downregulated in fust-1; ceh-14 mutants (Figure 5 D).clec-190 encodes a C-type lectinlike domain (CTLD) containing protein, a highly di v erse protein family that fulfills a wide variety of functions ( 71 ).Gi v en the strong synthetic regulation of clec-190 , we wondered whether loss of clec-190 expression might contribute to the synthetic double mutant phenotypes.To test this, we genera ted a c lec-190 null mutant in which the entire coding sequence is deleted, but found that the mutant results in no discernible fertility defects (Figure 5 E).Therefore, although clec-190 r epr esents an inter esting example of combinatorial regulation, it does not on its own contribute to the double mutant phenotypes.
To test whether specific functional classes of genes are d ysregula ted in tdp-1; ceh-14 double mutants, we performed Gene Ontology and Tissue Enrichment analysis.Upregulated genes have no statistically significant ( q < 0.01) enrichment categories, but downregulated genes are enriched in a few categories, including genes expressed in the spermatheca (Figure 5 F).This is notable given the co-expression of all three factors in the spermatheca (Figure 4 E-G) and the central role played by the spermatheca in fertilization.
We examined all sperma theca-annota ted genes with dysregulated expression in tdp-1; ceh-14 and found that most are downregulated in the double mutant, and that fust-1; ceh-14 doub le mutants hav e similar patterns of d ysregula ted gene expr ession (Figur e 5 G, Supplementary Figur e S7B).Mor eover, we found that ceh-14 , but not fust-1 or tdp-1 , is the main dri v er of these changes, as ceh-14 mutants display similar gene expression patterns to the double mutants (Figure 5 G).Together these da ta indica te tha t ceh-14 is necessary for stimulating the expression of a network of genes in the spermatheca, and motivate future work to determine whether these genes play a role in the double mutant fertility phenotypes.
Analysis of alternati v e splicing re v eals a contrasting regulatory landscape to that of gene expression.tdp-1 accounts for the majority of splicing changes observed in tdp-1; ceh-14 m utants, w hile ceh-14 accounts for very few splicing changes (Figure 5 H-J, Supplementary Table S2).Likewise, fust-1 is largely responsible for the splicing changes observed in fust-1; ceh-14 mutants (Supplementary Figure S6B).As with gene expression, we observe very little additi v e or synthetic regulation of alternati v e splicing.We also observ e v ery little ov erlap between genes with altered splicing regulation and genes with altered expression levels in the double mutants (Supplementary Figure S6B).Together, these da ta indica te tha t d ysregula ted genes in double mutants are either regulated transcriptionally by ceh-14 or post-transcriptionally by fust-1 and / or tdp-1.

tdp-1 and fust-1 co-inhibit e x on inclusion
Loss of tdp-1 or fust-1 results in many types of dysregulated splicing, including 5 and 3 splice site selection, intron retention, and alternati v e e xon skipping v ersus inclusion ('cassette exons') (Figure 6 A).One notable pattern we observed is that the effect of tdp-1 or fust-1 mutation on cassette exons is almost exclusively an increase in exon inclusion (Figure 6 B).This is in contrast with many other RN A binding proteins, w hich both stim ulate and inhibit exon inclusion, in a context-specific manner ( 72 ).Therefore, we conclude that tdp-1 and fust-1 function specifically to inhibit exon inclusion.We next asked whether tdp-1 and fust-1 inhibit expression of overlapping or distinct alternati v e e xons.Strikingly, we found that half of fust-1 -regulated cassette exons are also regulated by tdp-1 (Figure 6 C), and that the direction of splicing change is always concordant, with increased exon inclusion in both mutants.An example of an exon r epr essed by both factors is in the gene sav-1 , which harbors an unannotated cassette exon.In wildtype or ceh-14 mutants, this exon is predominantly skipped, but in either fust-1 or tdp-1 mutants, the exon becomes predominantly included (Figure 6 D, E).In fust-1 tdp-1 double mutants, the exon is included at levels similar to either single mutant (Figure 6 E), suggesting that tdp-1 and fust-1 do not act synthetically, but rather are both sim ultaneousl y required to repress sav-1 exon inclusion.
Gi v en the striking concordance of inhibition of exon inclusion by fust-1 and tdp-1 , we next asked whether such activity is an evolutionarily-conserved attribute of the two RBPs.This would be of particular interest gi v en both factors' prominent links to the human neuronal disorders ALS and FTD ( 73 ).To this end we re-analyzed data in which either of the mouse homologues (FUS or TDP-43) was knocked down in mouse brains and splicing analyzed by microarray ( 45 ).Focusing on cassette exons, we found a substantial overlap between the regulatory activity of FUS and TDP-43 (Figure 6 F).As in our C. elegans experiments, exons co-regulated by both FUS and TDP-43 in mouse brain tend to be inhibited by both factors (68.3% have increased inclusion in both knockdowns, Figure 6 G).Thus, tdp-1 / TDP43 and fust-1 / FUS have a propensity to coordinately inhibit exon inclusion both in C. elegans and in mouse brain.

Aberrant e x on inclusion of pqn-41 contributes to fertility defects
One intriguing example in which C. elegans tdp-1 (but not fust-1 ) inhibits exon inclusion is found in pqn-41 , a gene encoding a polyglutamine-containing protein.This gene harbors an alternati v e e xon which in wild-type conditions is primarily skipped, but in tdp-1 mutants is primarily included (Figure 7 A-C).This pqn-41 exon is the most strongly d ysregula ted exon in tdp-1; ceh-14 mutants ( exon inclusion = 41%, Figure 7 A).pqn-41 was previously shown to be important for proper de v elopmental cell death of the linker cell in the gonad of male C. elegans ( 74 ), which prompted us to ask whether pqn-41 might contribute to the gonad development or fertility defects observed in tdp-1; ceh-14 mutant hermaphrodites.We obtained a potentially null deletion allele, pqn-41(ok3590) ( 75 ), and tested fertility.We found that brood sizes of pqn-41 mutants are significantly lower than wild-type worms, and that these fertility defects are particularly pronounced at higher temperatures (Figure 7 D).Ther efor e, loss-of-function pqn-41 mutants have similar temperature-sensiti v e fertility defects to tdp-1; ceh-14 .
We ne xt e xamined mechanisms by which TDP-1 might inhibit pqn-41 exon inclusion.Both C. elegans and mammalian TDP43 / TDP-1 bind GU repeats in vitro with an af finity tha t increases in proportion to the number of repeats ( 76 , 77 ).We examined the introns and exons flanking the pqn-41 exon for the presence of GU repeats and found one such element, located immediately downstream of the 5 splice site of the alternati v e e xon, consisting of se v en consecuti v e GU repeats followed by an additional GU-rich r egion (Figur e 7 E, Supplementary Figur e S7C).This potential cis element shares commonalities with a recentlydescribed mechanism in which mammalian TDP43 inhibits inclusion of an exon in the STMN2 gene by binding to GU repea ts immedia tely adjacent to the 3 splice site ( 78 ).To test this mechanism in vivo , we generated a two-color fluorescent splicing reporter (79)(80)(81)(82) for the pqn-41 alternati v e e xon (Supplementary Figure S7D).When expressed in the spermatheca, we observe nearly 100% exon skipping, but a splicing reporter lacking the GU r epeats r esults in increased exon inclusion (Supplementary Figure S7E-F).Similar cis elements have been implicated in the activity of human TDP43 , and we likewise find such elements in the flanking introns of the top TDP-1-regulated exon skipping e v ents in worms (Supplementary Figure S7G).To test whether tdp-1 s role in double-mutant fertility is mediated by pqn-41 e xon skipping, we ne xt generated a pqn-41 mutant using CRISPR / Cas9 in which the alternati v e exon is removed and the flanking exons are precisely fused together, ther eby for cing expr ession of the exon skipped version (Figure 7 F).We then crossed this pqn-41 exon-deletion mutant into a tdp-1; ceh-14 background, thus restoring pqn-41 to the isoform most abundant under wild-type conditions (e xon skipped).Remar kab l y, these triple m utants ( tdp-1; pqn-41[exon-skipped]; ceh-14 ) exhibit a strong rescue (increase of brood size) at 25 • C compared to tdp-1; ceh-14 double mutants (Figure 7 G).This suggests that aberrant pqn-41 exon inclusion plays a major role in the fertility defects observed in tdp-1; ceh-14 double mutants.
In contrast with tdp-1; ceh-14 double mutants, fust-1; ceh-14 double mutants do not exhibit pqn-41 splicing defects.Likewise, crossing the pqn-41[exon-skipped] into the fust-1; ceh-14 double mutant does not cause an increase in brood size (Figure 7 G), suggesting that the rescue of tdp-1; ceh-14 by pqn-41 [exon-skipped] is mechanistically linked to the mis-splicing of pqn-41 caused by tdp-1 loss of function.Together these results highlight a new role for the polyglutamine gene pqn-41 in fertility, and indica te tha t pqn-41 missplicing is a major cause of the fertility defects observed in tdp-1; ceh-14 double mutants.In sum, using a systema tic combina torial genetic interaction screen, we found that two RBPs, fust-1 and tdp-1 , ar e both r equir ed in the context of a ceh-14 mutant background to maintain fitness and fertility in C. elegans .These two RBPs have both been implicated ALS and FTD in humans, and we now identify a common physiological role for both RBPs in C. elegans .Both RBPs have overlapping roles in inhibiting exon inclusion, pointing to shared molecular activities, and a potential molecular basis for the physiological roles for the two RBPs described her e.Failur e to inhibit exon inclusion in the pqn-41 gene is a major cause of the fertility defects in tdp-1; ceh-14 double mutants, thus providing a mechanistic link between the molecular activity of the TDP-1 RBP and the fertility phenotype observed in tdp-1; ceh-14 double mutants (Figure 7 H).

Novel genetic interactions across gene regulatory la y ers
We took a systematic genetic interaction approach to identify cross-regulatory genetic interactions in which a TF and RBP ar e combinatorially r equir ed for phenotypes affecting organismal fitness.This screen re v ealed a number of TF-RBP pairs r equir ed for phenotypes including fitness, de v elopment, and fertility.The strongest of these genetic interactions involves the homeodomain TF ceh-14 and either of the ALS-associated RBPs tdp-1 or fust-1 .
Extensi v e mechanistic coupling between transcription and splicing has been observed ( 6 ), suggesting that TF-RBP genetic interactions could arise from, for example, TF-RBP coordination of specific splicing events.However, for this specific genetic interaction our transcriptome analysis re v eals largely non-ov erlapping regulatory networ ks, in which ceh-14 regulates transcription, and tdp-1 / fust-1 regulate splicing, with few genes additively or synthetically regulated, and few genes with alterations in both splicing and transcription.Ther efor e, it seems likely that the synthetic fertility phenotypes result from the combination of distinct gene d ysregula tion e v ents.
We identify one such d ysregula tion in the alternati v elyspliced exon in pqn-41 .Mis-splicing of this exon is a major contributor to the phenotype, as restoring exon skipping rescues fertility defects of tdp-1; ceh-14 double mutants.Aberrant splicing of this exon in isolation does not cause fertility defects, as tdp-1 single mutants mis-splice pqn-41 at the same le v els as tdp-1; ceh-14 mutants, but do not have fertility defects.The data ther efor e suggest that missplicing of pqn-41 , in combination with altered expression of one or more ceh-14 target genes, results in fertility defects (Figure 7 H).A number of genes with spermatheca expression dependent on ceh-14 (Figure 5 G) represent promising candidates.
These results highlight the utility of the systematic re v erse-genetic interaction approach both for understanding relationships between regulatory factors and for understanding the roles of individual factors whose regulatory roles are only apparent when redundant or compensatory pathways are sim ultaneousl y perturbed ( 18 ).In this case, a shared molecular and physiological role for tdp-1 and fust-1 is re v ealed by their shared genetic interaction profile.Future studies characterizing additional genetic interactions identified here may shed light on novel physiological roles for additional RBPs and TFs.

fust-1 and tdp-1 interact with ceh-14 to affect C. elegans fertility
The her maphrodite sper matheca is a key site of overlapping expression for tdp-1 , fust-1 , and ceh-14 , and double mutants have reduced sperm efficiency.Signaling from somatic gonad cells such as spermathecal cells is r equir ed for germline de v elopment and function, as ablation of specific spermatheca cells results in defecti v e germ cell function, and e v en sterility ( 83 ).We hypothesize that faulty signaling between spermathecal cells and germ cells might explain the defects in fertility and gonad de v elopment observ ed in our doub le m utants.The double-m utant fertility phenotype is particularly pronounced at 25 • C, which is a mildly stressful temperature for wild-type worms, causing modest defects in fertility and gonad de v elopment ( 25 , 50 , 61 ).Temperatures higher than 25 • C result in damage to sperm and strong fertility defects ( 84 , 85 ).We specula te tha t tdp-1; ceh-14 and fust-1; ceh-14 double mutants are deficient in their heat stress responses ( 48 , 49 ), and ar e ther efor e unable to maintain normal homeostasis under mild heat str ess.Ther efor e, tempera tures tha t cause mild fertility defects in wild-type animals result in strong defects in double mutants.
How might pqn-41 splicing contribute to this temperature-specific fertility defect?The PQN-41 protein is abundant with glutamines, and contains a particularly polyglutamine-rich domain at the C-terminus ( 74 ).This domain begins immediately downstream of the alternati v e exon (Supplementary Figure S7H).We hypothesize that, like other polyglutamine proteins such as the Huntington protein, PQN-41 is subject to pathogenic aggregation ( 86 ).If the exon-included isoform of PQN-41 is particularly prone to aggregation, and if stressful conditions such as higher temperatures further increase the likelihood of aggregation, this could lead to temperature-sensiti v e defects.Indeed, there is evidence that PQN-41 forms aggregates in vivo ( 74 ), but a full mechanistic test of this hypothesis awaits further investigation.It will be interesting in future studies to investigate whether there is a related prion-like pol yQ protein underl ying the similar fertility defects of fust-1; ceh-14 double mutants.

fust-1 and tdp-1 co-inhibit e x on inclusion
Identification of shared phenotypes between tdp-1; ceh-14 and fust-1; ceh-14 double mutants led to the observation that tdp-1 and fust-1 also have shared effects on the transcriptome.Most notably, they both act to inhibit inclusion of alternati v ely-spliced cassette e xons, including many inhibited by both RBPs.fust-1 and tdp-1 do not appear to act redundantly, as fust-1 tdp-1 double mutants do not result in increased exon inclusion compared to either of the single mutants.Rather, fust-1 and tdp-1 are both sim ultaneousl y r equir ed for inhibition of these exons.One plausible explanation for this finding is that both RBPs bind together to specific pre-mRN As w here they act in concert to pre v ent aberrant exon inclusion.
The activity of C. elegans tdp-1 and fust-1 in inhibiting exon inclusion is also a shared feature of mammalian TDP-43 and FUS.We find that knockdown of TDP-43 or FUS in mouse brain ( 45 ) results in aberrant exon inclusion, and that many of these exons co-inhibited by both TDP-43 and FUS.This is interesting in light of recent findings suggesting a patholo gicall y-relevant role for TDP-43 in inhibiting cryptic ex ons.Ex ons are sometimes classified as cryptic if they exhibit low inclusion le v els, lack of e volutionary conservation, and / or propensity to disrupt the function of the gene they reside in ( 87 , 88 ).TDP-43 has been identified as an inhibitor of cryptic exons ( 87 , 88 ), and recent evidence implicates aberrant inclusion of two different cryptic exons in the genes STMN2 and UNC13A as potential causati v e mechanisms underlying TDP-43 pathology in ALS (89)(90)(91)(92).
Our findings are consistent with a role for tpd-1 / TDP-43 in inhibiting aberrant exon inclusion, and we extend this observation to also include a role for fust-1 / FUS in inhibiting exon inclusion.This leads us to speculate whether FUSrela ted pa tho genesis might also be mechanisticall y linked to inappropriate inclusion of exons inhibited by FUS.Previous work on mammalian TDP-43 and FUS has concluded that the two RBPs share many common RNA targets, but also hav e considerab le non-ov erlapping regulatory functions ( 45 , 93 ).We focused here on the regulation of cassette exons, and found substantial overlap between the RBPs in inhibiting exon inclusion.It will ther efor e be inter esting to ask whether aberrant exon inclusion underlies FUSmedia ted pa thology in an analogous way to that of TDP-43-media ted pa thology.
Many of the exons identified as targets of tdp-1 and / or fust-1 in C. elegans have attributes of cryptic exons as well.For example, the alternative exons in sav-1 and pqn-41 are expressed at low levels in wild-type (Figures 6 D, 7 B), and in the case of sav-1 the exon is unannotated.In the case of pqn-41 , failure of tdp-1 to inhibit exon inclusion leads to detrimental phenotypes (fitness and fertility defects).This is an interesting parallel to the pathogenic consequences of TDP-43 failing to inhibit cryptic exon inclusion in the STMN2 or UNC13A genes, and suggests that inhibition of aberrant exon inclusion may be an e volutionarily-conserv ed feature of tdp-1 / TDP-43 and fust-1 / FUS.

DA T A A V AILABILITY
The data underlying this article are available in GEO at https://www.ncbi.nlm.nih.gov/geo/, and can be accessed with identifier GSE230025.

SUPPLEMENT ARY DA T A
Supplementary Data are available at NAR Online.

FFigure 1 .Figur e 2 .Figure 3 .
Figure 1.Identifying genetic interactions with competiti v e fitness assays.( A ) Schematic for competiti v e fitness assays.Green worms r epr esent fluor escent mutants, black r epr esents wild-type.(B-C) Relative fitness values for all RN A binding protein m utants ( B ) transcription factor mutants ( C ) used in genetic interaction screen.(D, E) Double mutant relati v e fitness data for msi-1; unc-86 ( D ) and exc-7; ets-5 ( E ), illustrating examples of interactions deemed either non-significant (| ε | < 0.4, panel D) or significant (| ε | > 0.4 plus a t -test with p < 0.05, panel E).Single mutants are assayed against wild-type worms and double mutants are assayed against transcription factor single mutants.( F ) Heat map of 110 RBP; TF double mutants created.Each square represents one double mutant.Blue squares represent synthetic negati v e effects on fitness ( ε < 0.4), green represents synthetic positi v e effect ( ε > 0.4), and gray squares did not yield strong synthetic fitness effects ( −0.4 < ε < 0.4)

Figure 4 .
Figure 4. fust-1; ceh-14 and tdp-1; ceh-14 exhibit defects in adult hermaphrodite gonad and gametes.( A ) Schematic of adult gonad in wild type compared to two examples of over-migrated gonad arms.Bottommost graphic illustrates infiltration of distal arms into both the proximal gonad and the uterus.( B ) Quantification of gonad defects.Double mutants do not differ from wild-type at younger larval sta ges, b ut exhibit increased defects at adulthood.Asterisk indicates significant difference from wild-type, t -test P < 0.05.n > 70 animals across 5 independent biological replicates.( C ) Male mating with fog-2 (q71) feminized germline mutants.Double mutant males produced smaller brood sizes than wild type, and differences are significant at 25 • C. See also Supplementary Figure S4C.( D ) Wild-type males were mated with either wild-type or double mutant hermaphrodites.Mating with a male significantly increases brood size for wild type, but does not increase brood size in either tdp-1; ceh-14 or fust-1; ceh-14.Asterisk indicates ANOVA P < 0.05 followed by post-hoc t -test P < 0.05 for individual within-graph comparisons.See also Supplementary Figure S4D.(E-G) Representati v e images show expression of CEH-14 ( E ), TDP-1 ( F ) and FUST-1 ( G ) in the anterior half of adult hermaphrodites.Solid outline indicates location of gonad, and dotted outline indica tes loca tion of sperma theca.Bright gr een visible through center of worm in (E) is gut autofluor escence.Rightmost panels show sperma theca a t higher magnification.CEH-14, TDP-1 and FUST-1 are expressed in the cells making up the bag-like structure of the spermatheca.Scale bar represents 50 m.

Table 1 .
List of strains used in this study