C. elegans THSC/TREX-2 deficiency causes replication stress and genome instability

Transcription is an essential process of DNA metabolism, yet it makes DNA more susceptible to DNA damage. THSC/TREX-2 is a conserved eukaryotic protein complex with a key role in mRNP biogenesis and maturation that prevents genome instability. One source of such instability is linked to transcription as shown in yeast and human cells, but the underlying mechanism and whether is universal is still unclear. To get further insight in the putative role of THSC/TREX-2 in genome integrity we have used Caenorhabditis elegans mutants of the THP-1 and DSS-1 members of THSC/TREX-2. These mutants show similar defective meiosis, DNA damage accumulation and activation of the DNA damage checkpoint. However, they differ regarding replication defects as determined by dUTP incorporation in the germline. Interestingly, this specific thp-1 phenotype can be partially rescued by overexpression of RNase H. Furthermore, both mutants show a mild increase in the H3S10P mark previously shown to be linked to DNA-RNA hybrid-mediated genome instability. These data support the view that both THSC/TREX-2 factors prevent Jo ur na l o f C el l S ci en ce • A cc ep te d m an us cr ip t transcription-associated DNA damage derived from DNA-RNA hybrid accumulation by separate means.

The THSC/TREX-2 complex is conserved from yeast to humans and is involved in mRNP biogenesis (Garcia-Oliver et al, 2012;Rondón et al, 2010). It interacts with the nuclear pore complex (NPC) and is constituted by the stable association of the multi-domain protein Sac3/GANP with the Thp1/PCID2, Sus1/ENY2, Cdc31/CEN2 and Sem1/DSS1 subunits (Fischer et al, 2002;Gallardo et al, 2003;Lei et al, 2003), among which Cdc31 and Dss1 have also been characterized as a centrin and a 19S proteosome subunit, respectively (Faza et al, 2009;Fischer et al, 2004;Wilmes et al, 2008). In addition to these functions, the mammalian proteins PCID2 and DSS1 have been linked to BRCA2 (breast cancer susceptibility gene 2 product), a component of the homologous recombination machinery (Bhatia et al, 2014;Marston et al, 1999;Yang et al, 2002). Similarly to THO, it is believed that the genetic instability and transcriptional defects observed in yeast THSC/TREX-2 mutants (Gallardo et al, 2003) are due to R-loop accumulation as co-transcriptional cleavage of the nascent RNA reduced both defects to some extent (González-Aguilera et al, 2008). However, our laboratory has shown that depletion of PCID2, despite increasing genomic instability, does not lead to detectable accumulation of R-loops (Bhatia et al, 2014). To shed light into this conundrum we have used the nematode C. elegans to characterize two putative members of the C. elegans THSC/TREX-2 complex. We have observed that lack of the CeTHP-1 component of THSC/TREX-2 leads to sterility, DNA damage accumulation and replication defects at the germline. Importantly, microinjection of RNase H partially restores the Cy3-dUTP incorporation in the nematode germline consistent with R-loops interfering with replication progression. Interestingly, there is only a mild increase in the H3S10P chromatin mark in the THSC/TREX-2 mutants analyzed, strengthening the idea that R-loops interfere with replication in different ways. Our data provides evidence that THSC/TREX-2 contributes to the maintenance of genome integrity in part by preventing R loop accumulation. Hence this study highlights that cells have developed alternative ways to avert DNA-RNA hybrids and minimize their incidence as a source of DNA damage and genome instability.

C.elegans THSC/TREX-2 complex is essential for fertility
To analyze the role of THSC/TREX-2 in C. elegans, first we performed a search of THSC/TREX-2 orthologs based on the sequence comparison (Garcia-Oliver et al, 2012).
The putative components of the C. elegans THSC/TREX-2 complex are summarized in Table S1. In yeast, Thp1, Sac3, Sus1 and Cdc13 are stable associated subunits of the THSC/TREX-2 complex (Gallardo et al. 2003) while Sem1 shown to also be an integrating Journal of Cell Science • Accepted manuscript component of TREX-2 (Wilmes et al., 2008;Faza et al., 2009), is a subunit of the 19S proteasome that interact with other factors, as BRCA2 in mammals (Marston et al, 1999;Yang et al, 2002). To investigate the similar or different effects of these two THSC/TREX-2 complex members we decided to work with these two candidates for further analysis: the C27F2.10 gene (from now on thp-1) homolog to yeast THP1 and human PCID2, and the dss-1 homolog to yeast SEM1 and human DSS1. The CeTHP-1 protein displays a 45% and 27% amino-acid identity with human PCID2 and yeast Thp1, respectively (Fischer et al, 2002;Jani et al, 2012) (Table S1). Similarly to human PCID2 (Umlauf et al, 2013), we show here that CeTHP-1 is a nuclear protein and accumulates at the nuclear periphery, possibly at nuclear pore complexes (Fig. S1A,B). CeDSS-1 displays a 49% and 40% amino-acid identity with human DSS1 and yeast Sem1, respectively (Pispa et al, 2008) (Table 1).
We investigated the two THSC/TREX-2 C. elegans orthologs by characterizing two deletion mutants, thp-1(tm3507) and dss-1(tm370), which are predicted to encode nonfunctional truncated proteins. The thp-1 gene spans 1.9 kb, including 9 exons, and is predicted to encode an mRNA that produces a 413 amino-acid protein (Fig. S1C). The thp-1(tm3507) allele carries a deletion of 255 bp that can be detected by single worm PCR (Fig.   S1D), and is predicted to encode a non-functional truncated protein of 64 amino acids corresponding to the first three exons and part of the fourth one. The dss-1 gene was previously characterized as crucial for embryogenesis, larval growth and oogenesis (Pispa et al, 2008). Lethality assays revealed that the thp-1(tm3507) allele conferred complete sterility, since the mutant homozygous strain laid no eggs (Table 1), while N2(wt) and heterozygous thp-1(tm3507) mutants show the expected lethal percentage 0.05% and 62.86% respectively. The lethality observed in the heterozygous mutant is due to the nature of the balancer system used (Edgley et al, 2006). The data demonstrated that lack of the THSC/TREX-2 component THP-1 leads to sterility similarly to that previously described for the dss-1 mutant (Pispa et al, 2008). Therefore, the C. elegans thp-1 is an essential gene, as is dss-1. Yet, homozygous mutants produced by self-fertilization of heterozygous hermaphrodite parents are rescued beyond L4 stage by maternal contribution of thp-1 and dss-1 mRNA, allowing the analysis of mutants defective in either gene within adult tissues.
To determine whether the sterility phenotype was caused by a defect during meiosis, we analyzed thp-1(tm3507) and dss-1(tm370) germlines. Each germline is spatially polarized in a distal to proximal manner with respect to mitotic proliferation and progression through meiotic prophase I. In the distal region nuclei proliferate by mitotic divisions acting as stem cells until they reach the transition zone where meiosis starts (Garcia-Muse and Boulton, 2007). 4,6-Diamidino-2-phenylindole (DAPI) staining of thp-1(tm3507) and dss-1(tm370) germlines showed a mitotic region slightly reduced with respect to the N2(wt) regarding the number (Fig. 1).

Journal of Cell Science • Accepted manuscript
At the proximal region of the germline, wild type oocytes always have nuclei with six discernible DAPI-stained bivalent chromosomes, indicative of successful meiosis I. In C. elegans any variation on this number is considered an aneuploidy due to defective meiosis.
Indeed, as can be seen in Fig. 2A, N2 (wild type) oocytes show 6 bivalents (average 6±0.0, n=51), whereas in the case of the thp-1 mutant oocytes with six normally condensing chromosomes as well as oocytes with more than six bivalents (average 7,6±1,67, n=45) were observed. This is an observation shared with the dss-1 mutant, where oocytes also show a deviation from the expected number of six condensed chromosomes (average 7,32±1,59, n=65), that it has been show to increase with the age of the animal ( Fig. 2A, (Pispa et al, 2008)). This implies that both thp-1 and dss-1 mutants have abnormal diakinesis. In 24h post-L4 animals we were not able to see any difference in the number of oocytes respect to the N2(wt) that shows 4-5, however older thp-1 animals would show alterations in the organization of the proximal region, being difficult to differentiate diplotene and diakinesis (Fig. 2B).

DNA damage and checkpoint activation in thp-1 and dss-1 mutants
Next, we analyzed whether the diakinesis defect could be a consequence of early meiotic defects. One essential feature of successful meiosis is the meiotic recombination, which is required for the generation of genetic variation and for proper chromosome segregation.
Meiotic recombination is initiated by SPO-11-dependent DSB formation (Dernburg et al, 1998), and requires RAD-51, a member of the RecA-strand exchange protein family, to catalyze the invasion of DNA single-strand overhangs into a recipient double-strand DNA to initiate formation of D loops and the later steps of meiotic recombination (West, 2003). First, to examine DSBs during meiosis in thp-1 and dss-1 mutants we analyzed RAD-51 foci (Alpi et al, 2003). In N2(wt) worms, the levels of RAD-51 foci increased upon entrance to the transition zone (zone 2), peaked at mid-pachytene (zone 4) and disappeared by the end of pachytene (zone 5; Fig. 3). Upon entrance into meiosis (zone 3), thp-1 and dss-1 mutants accumulated RAD-51 foci earlier than N2(wt) worms ( Fig. 3; Fig. S2), suggesting that the DSBs observed could originate from problems during mitotic replication or pre-meiotic replication at the entrance of the transition zone. Also, in these mutants we observed that RAD-51 foci remained until the end of the germline implying that DSBs do not repair with the efficiency of N2(wt) animals ( Fig. 3; Fig. S2). With the aim to check if the increase in DSB observed in thp-1 was due to a defect in their repair we performed a DSB repair analysis.
24h post-L4 animals were treated with 75Gy (which generates mainly DSBs), leaved for 36-48 hours and then RAD-51 foci accumulation was analyzed. At mitotic and transition zone regions N2(wt) animals the levels of RAD-51 were similar to the non-treated control worms while the thp-1 irradiated shows a decrease in the nuclei without RAD-51 foci with respect to Journal of Cell Science • Accepted manuscript the non-treated (Fig. S3). This is can also be observed in the meiotic regions, although at 36 hours the N2(wt) is not completely recovered the thp-1 mutant shows a more dramatic accumulation of DSB when comparing both irradiated and not treated (Fig. S3). These results suggest that in thp-1 DSB repair is compromised.
To determine if the increase in DSBs in thp-1 and dss-1 mutants was a readout of DNA damage caused by replication problems we performed immunofluorescence using antibodies against the single-strand binding protein RPA-1, which is involved in replication and recombination, and against the DNA damage checkpoint protein ATL-1, which is recruited to sites of DNA damage where it activates the DNA damage checkpoint (Garcia-Muse andBoulton, 2005, Stergiou et al. 2011). Unlike N2(wt) worms, thp-1 and dss-1 mutants exhibited regions of ssDNA as indicated by the presence of RPA-1 and ATL-1 foci in mitotic and meiotic nuclei ( Fig. 4; Fig. S5). This is consistent with the interpretation that DNA damage is occurring (throughout the germline) in thp-1 and dss-1 mutants, and subsequently activating the DNA damage checkpoint. This phenotype is reminiscent of that observed in thoc-2 mutants (Castellano-Pozo et al, 2012a) and mutants with defects during S-phase such as atl-1 and clk-2, in which endogenous DNA breaks arise from replication defects (Ahmed et al, 2001;Garcia-Muse and Boulton, 2005).
This result demonstrates that mitotic replication is impaired in thp-1 mutants but not in dss-1.
In yeast and C. elegans, other mRNA biogenesis mutants such as mutants from the THO complex also show replication defects. These defects have been linked to R-loop accumulation (Castellano-Pozo et al, 2012a;Huertas and Aguilera, 2003;Wellinger et al., 2006). To address directly if this was also the case in thp-1 mutant, we co-microinjected Escherichia coli RNase H1, which specifically digest the RNA moiety of DNA-RNA hybrids, into the worms in our in vivo replication assay (Castellano-Pozo et al, 2012a). Comicroinjection of RNase H1 did not affect incorporation of Cy3-dUTP in N2(wt) worms, whereas in thp-1 mutants caused a clear increase in Cy3-dUTP labeled nuclei (Fig. 5).
Although incomplete, a partial recovery of mitosis proficiency was possible upon RNase H1 microinjection. This result suggests that R-loops can form in thp-1 C. elegans mutant and Journal of Cell Science • Accepted manuscript could be responsible for the mitotic replication impairment that leads to the DNA damage observed.
It is not yet well understood the mechanism and factors responsible for genome instability mediated by R-loops. We previously found that R-loops are linked to H3S10 phosphorylation, a mark of chromatin condensation, which led us to propose a model in which R-loops trigger the formation of condensed chromatin patches that would interfere with replication and/or transcription (Castellano-Pozo et al, 2012a). This conclusion was reinforced by the identification of yeast histone mutants unable to phosphorylate histone H3S10P (due to the lack of the residue or its mutation into alanine), which were able to accumulate high levels of R loops although without compromising genome integrity (Garcia-Pichardo et al, 2017). To determine if R-loop-mediated chromatin modifications could explain the genome instability of THSC/TREX-2 C. elegans mutants, we performed immunostaining against H3S10P. Normally a N2(wt) germline shows 1-2 mitotic nuclei with H3S10P correlating with chromosome condensation during mitosis and at the compacted bivalents of the diakinesis region ( Fig. 6; Fig.S6). We observed a slight increase of nuclei with H3S10P signal in both mutants, which was clearer in thp-1 (Fig. 6). In this mutant some nuclei with H3S10P signal were observed beyond the germline mitotic region, reaching the transition zone and early pachytene regions ( Fig. 6; Fig.S6). In addition, we observed H3S10P at later meiosis stages, in N2(wt) the compacted six bivalents show a bright signal like the thp-1 and dss-1 mutants. We noticed that 36h post-L4 worms, when diakinesis is more disorganized in the thp-1 mutant, H3S10P signal could be seeing at the late pachytene region ( Fig. 2B; Fig.   S6B). Thus, at least part of the observed replication impairment could be due to chromatin compaction triggered by R-loops as it has been shown in yeast and mammals (Castellano-Pozo et al, 2013).

DISCUSSION
Failures of DNA metabolism processes such as DNA replication and repair are a major source of DNA damage accumulation in cells (Aguilera and Garcia-Muse, 2013). Importantly, it has been shown that DNA damage occurs at higher levels at regions with high transcriptional activity. Indeed, the loss of several RNA processing factors is associated with genome instability which is explained, in part, by the formation of co-transcriptional DNA-RNA hybrids (Aguilera and Garcia-Muse, 2012; Garcia-Muse and Aguilera, 2019). To further understand this process in pluricellular organisms we have analyzed mitosis and meiosis in C. elegans depleted of the THSC/TREX-2 mRNP biogenesis complex. Analysis of deletion mutants of two components of THSC/TREX-2, thp-1(tm3507) and dss-1(tm370), revealed Journal of Cell Science • Accepted manuscript that this complex is required for fertility since the mutants do not lay eggs. The absence of eggs is not due to compromised gonad development since a two-arm gonad with developing germ cells was present in all mutant worms examined. Germline analysis of both TREX-2 mutants reveled defects in diakinesis since oocyctes with an abnormal number of bivalents were observed. Nevertheless, the eggless phenotype cannot be explained neither by the diakinesis defects observed in the thp-1 and dss-1 mutants since other mutants with defective diakinesis do form and lay eggs, and the lethality is observed at the embryonic stage. This is the case of mutants of brc-2, the C. elegans homologue of DNA repair gene BRCA-2 (Martin et al, 2005;Pispa et al, 2008). BRCA2 interacts with DSS1 (Marston et al, 1999;Yang et al, 2002) and loss of either BRCA2 or DSS1 results in defects in homologous recombination (West, 2003). However, in brc-2 mutant oocytes chromosomes aggregate instead of condensing normally and produce eggs unlike thp-1 and dss-1 mutants. One In agreement with this, replication impairment can be partially suppressed by the addition of RNase H that removes DNA-RNA hybrids. This observation supports that the defective replication due to the loss of a core component of the THSC/TREX-2 is in part caused by DNA:RNA hybrids, as it is also the case for the absence of a functional THO complex in all organisms analyzed (Castellano-Pozo et al, 2012a;Huertas and Aguilera, 2003; Journal of Cell Science • Accepted manuscript Domínguez-Sánchez et al, 2011a). In contrast, loss of dss-1 does not hamper replication, suggesting that the genomic instability observed in this mutant is more likely to be related with BRCA2 (Pispa et al., 2008;Bhatia et al., 2014). In addition to cause the generation of DSBs, R-loops have recently been shown to accumulate at DSBs, especially those induced in transcriptionally active loci. However, is still unclear the significance of this, since data supporting both that R-loops actively participate or that interfere in DSB repair has been shown (AguileraandGomez-Gonzalez, 2017; Marnef and Legube, 2020). We have observed that thp-1 mutants show a delay in repair, which is in harmony with a detrimental effect of R loops in DSB repair. In other hand, it has been shown that spontaneous R loops must be removed to avoid genomic instability and that repair factors such as BRCA2 play a role in this (Bhatia et al., 2014;D'Alessandro et al., 2018), and, as we mentioned above, this could be behind the differences observed between the thp-1 and dss-1 mutants. Therefore, our results show how two components of the C. elegans THSC/TREX-2 impact differently in genome stability. Lastly, there are important differences between the THO and the THSC/TREX-2 complexes, and likely the mechanism by which R loop accumulation leads to genetic instability. During mRNA biogenesis the two complexes act at different steps. The THO complex is recruited to chromatin and is required early during transcription elongation and mRNA export. Additional proteins would be recruited to the nascent mRNP via THO to later on facilitate the recruitment of THSC/TREX-2 to bind the mRNP to the nuclear pore complex (NPC). Importantly, it has been shown that physical proximity of transcribed chromatin to NPCs restrains the formation of pathological R loops during transcription (Garcia-Benitez et al, 2017). This observation agrees with the gene-gating hypothesis, which proposes that localization of transcribed DNA at the NPC facilitates the formation of an export-competent mRNP (Blobel, 1985). While this mechanism is well characterized in yeast it is less defined in mammalian cells, for which further studies would be required for cells depleted of different THSC/TREX-2-factors. In this regard C. elegans has been shown to have developmental and stress-induced gene gating (Rohner et al, 2013) and we provide evidence that THSC/TREX-2 accumulates at the nuclear periphery. This grants the hypothesis that R loopdependent phenotypes observed in the THSC/TREX-2 mutants could be originated by defects in the association with the NPC in line with data provided for Mlp1,2 yeast mutants.

Strains and maintenance
Standard methods were used for the maintenance and manipulation of C. elegans strains (Brenner, 1974). The wild type Bristol N2 and JK2739 nematode strains were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources. dss-1(tm370) was kindly provided by Dr. Jäntti (Pispa et al, 2008) and thp-1(tm3507) was generated and kindly provided by Dr. Mitani. C. elegans thp-1(tm3507) deletion was backcrossed six times with wild type Bristol N2 and then balanced with JK2739.
Embryonic lethality was scored by comparing the number of eggs that hatch to produce viable progeny versus the total number of eggs laid. Briefly L4 hermaphrodites grown at 20°C were individually plated. The animals were transferred to new plates once every 24 hours until the egg laying stopped. Every day eggs laid and hatched larvae were counted.
The total number of single hermaphrodites for each stain is indicated in Table 1.
For DNA repair analysis after irradiation, 24h. post-L4 animals were exposed to 75 Gy of γray from BioBeam8000. After 36-48h. post-irradiation worms were processed for gonad analysis by immunofluorescence. As control non-Irradiated plates were maintained in parallel.
To express GFP::THP-1, we inserted a full-length thp-1 PCR product amplified from N2 genomic DNA into plasmid pBN16, which contains a heat-inducible hsp-16.41 promoter and the 3'UTR of unc-54 (Ródenas et al, 2012). The resulting plasmid, designated pBN467, was used for Mos-mediated single copy integration into ttTi5605 on chrII of EG4322 (Frøkjaer-Jensen et al, 2008) to generate strain BN967, which next was crossed to BN903 that has mKate2 inserted into the 5'-end of the mel-28 coding sequence, resulting in strain BN974.
Tagging of MEL-28, which localises to nuclear pore complexes, with mKate2 was done similarly to the protocol for insertion of GFP into the mel-28 locus (Gomez-Saldivar et al., 2016).

Journal of Cell Science • Accepted manuscript
Immunostaining Immunofluorescences for Figures 2 and 5 were performed as described (Martin et al, 2005).
One day post-L4 adult gonads were dissected in PBS with levimasole on polylisine slides, fixed for 20 m. in 4% paraformaldehyde and replaced for 10 min. in TBSBTx (TBSB + 0.4% TX100). The slides were washed twice for 10 min. and once for 30 min. with TBSB (TBS + 0.5% BSA). They were incubated overnight at 4ºC with the 1ry antibodies (Table S3). Next day gonads were washed 3 times in TBSB, each for 30 minutes at RT, and incubated for 2 hours with the secondary antibody in TBSB (Table S3). Gonads were washed three times for 30 minutes in TBSB and mounted with 10 μl Vectashield (with 1 μg/ml DAPI) per sample for further analysis.
Gonads were washed once in TBSTw+DAPI (1 μg/ml) for 10 min. and then washed 4 times for 10 min. in TBSTw and mounted with 10 μl Vectashield per sample for further analysis.

In situ detection of germline DNA synthesis
Direct incorporation of Cy3-dUTP (Amersham Bioscences) into germline nuclei was performed as described (Castellano-Pozo et al, 2012a). The injection mix consisted of 50 μM Cy3-dUTP (Amersham Bioscences, Piscataway, NJ) in PBS, pH 7.2. After ~2.5 hours of exposure to the Cy3-dUTP, gonads were dissected, fixed and DAPI-stained. The total number of cells that incorporated Cy3-dUTP was determined for each dissected germline.

Statistical Analysis
Statistical significance was determined with a Student's t test or ANOVA using PRISM software (Graphpad Software Inc.).

ACKNOWLEDGMENTS
We thank Shohei Mitani at the National Bioresource Project and Dr. Jäntti for kindly providing thp-1(tm3507) and dss-1(tm370), some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440), A. Gartner for reagents, and C. Ayuso and U. Galindo for technical assistance. This work was supported by grants from Spanish Ministry of Economy and Competitiveness (BFU2016- (C) Quantification of RAD-51 foci of germline nuclei. Zone 1 (mitosis), zone 2 (transition zone), zones 3-4-5 (pachytene), zone 6 (diplotene) of N2(wt), thp-1 and dss-1 strains (n=18, n=20 and n=20, respectively). Error bars indicate standard deviation. Statistical analysis is shown in figure S3. Statistical analysis is shown in Figure S4. Unpaired t-test two-tailed analysis. Scale bar 10µm.