Loss of the E3 ubiquitin ligases UBR-5 or HECD-1 restores Caenorhabditis elegans development in the absence of SWI/SNF function

Significance Mutations in subunits of the SWI/SNF chromatin remodeling complex are found in numerous different neurodevelopmental disorders and 25% of all human cancers. Here, we use a model SWI/SNF mutation in C. elegans to determine whether SWI/SNF complex abundance is specifically regulated by two highly conserved E3 ubiquitin ligases, and inhibition of these enzymes can restore development to adulthood in SWI/SNF mutants that otherwise die early in development. These substantially advance our understanding of how SWI/SNF abundance is regulated and point toward potential therapeutic targets for the many different pathologies caused by mutations in SWI/SNF subunits.

lethality, studies have focused on studying SWI/SNF function in tissue-specific mouse knockout models (16) or in cell culture where individual gene knockouts are viable (17). However, cell culture models cannot recapitulate the role of SWI/SNF in animal development across tissues, and viable cell culture models are still unlikely to represent complete loss of SWI/SNF function (17).
In contrast to mammalian models, the Caenorhabditis elegans genome encodes only a single gene for each of the core SWI/SNF subunits. Deletions of the core subunits swsn-4 (human SMARCA2/4) or swsn-1 (human SMARCC1/2) result in embryonic or larval lethality, respectively, and RNAi-mediated knockdown of snfc-5 (human SMARCB1) similarly results in embryonic lethality (18). However, previous work in C. elegans identified a temperature-sensitive swsn-1 mutation. swsn-1 temperature-sensitive mutants can develop to adulthood at the permissive temperature of 15 °C but arrest in embryonic development with 100% penetrance at the restrictive temperature of 22.5 °C (19). Mutations in swsn-1 (in C. elegans) and in SMARCC1/2 (in humans) both cause normal animal development to be disrupted, which suggests that SWI/SNF regulation of development might be a conserved process between nematodes and humans and indicates that the temperature-sensitive swsn-1 allele in C. elegans is a useful model to study SWI/SNF function in development.
Here, we report that a specific mutation in snfc-5 (human SMARCB1) can prevent embryonic lethality and early developmental arrest of swsn-1 (human SMARCC1/2) mutants. In addition, we report that the loss-of-function mutations in either of the genes encoding the E3 ubiquitin ligases UBR-5 or HECD-1 could further restore wild-type development in the swsn-1 mutant model. Specifically, around 70% of hatched swsn-1; snfc-5; ubr-5 triple mutants developed to adulthood under conditions, where 100% of swsn-1 single mutants died as embryos. Using our mutant models, we established a set of 335 genes that were specifically dysregulated in swsn-1 mutants but exhibited near wild-type expression levels in swsn-1; snfc-5 double and swsn-1; snfc-5; ubr-5 triple mutants across three independent RNA-sequencing experiments, suggesting that the dysregulation of some subset of these genes drives the developmental defects observed in swsn-1 mutants. In addition, using multiple independent approaches, we demonstrated that UBR-5 likely regulates the levels of SWI/SNF subunits to mediate its effects on SWI/SNF function. Our findings provide insights into how defects in SWI/SNF function cause developmental defects and provide evidence suggesting that UBR5 or HECTD1, the human orthologs of UBR-5 and HECD-1, are potential therapeutic targets for developmental defects caused by missense mutations in SWI/SNF subunits.

Mutations in snfc-5, ubr-5, and hecd-1 Can Prevent Embryonic Lethality and Developmental Arrest in a Mutant Model of Loss
of SWI/SNF Function. To identify mutations that could compensate for loss of SWI/SNF function, we utilized a temperature-sensitive (ts) loss-of-function allele of the core SWI/SNF subunit swsn-1 in the model animal C. elegans (20). Specifically, we utilized a swsn-1 temperature-sensitive allele that encodes a P86L substitution mutation in the SWIRM protein domain of SWSN-1 (SI Appendix, Fig. S1A), which is the exact same mutation as identified by Sawa et al. (19). Of note, 100% of animals homozygous for the swsn-1 P86L mutation die as embryos when grown at 22.5 °C or arrest their development at early larval stages when exposed to 25 °C after completing embryonic development (19). We mutagenized swsn-1 mutants with ethyl methanesulfonate (EMS) at a permissive temperature (20 °C) and subjected their F3 offspring as synchronized embryos to the restrictive temperature (25 °C) for 72 h (Fig. 1A). We identified five mutant isolates that did not arrest development at early larval stages. By performing whole genome sequencing of these five isolates, we found that one of the recovered isolates carries an additional swsn-1 substitution mutation (V62I) nearby the original P86L mutation. This mutation is likely an internal suppressor and was not further validated. The remaining four isolates from this screen, which were all from independent pools, all carry an identical A258V substitution mutation in the gene encoding SNFC-5 (SI Appendix, Fig. S1B), another core subunit of the SWI/SNF complex and homolog of human SMARCB1. We recreated the A258V snfc-5 mutation by CRISPR-Cas9 gene editing (21) and confirmed that this mutation prevents early larval arrest in swsn-1 mutants (SI Appendix, Fig. S1 C-F). We conclude that the A258V mutation in snfc-5 can suppress some of the developmental defects observed in swsn-1 mutants.
We found that exposure of mutant L4-staged or young adult animals to 25 °C for 16 h prior to collecting the embryos resulted in 100% embryonic lethality for swsn-1 single mutants and 95% of swsn-1; snfc-5 double mutants (SI Appendix, Fig. S1G). Of the few surviving swsn-1; snfc-5 double mutants, 98% arrested development between L1 and L3 stages ( Fig. 1 B-D and SI Appendix, Fig. S1H). These findings indicate that this specific mutation in snfc-5 can suppress the embryonic lethality in a proportion of swsn-1 mutants but is not sufficient to restore development to adulthood in the animals that do not die as embryos. Therefore, we asked whether we could suppress the developmental defects of swsn-1; snfc-5 double mutants further by introducing additional mutations and whether this would allow us to identify genetic interactors of the SWI/SNF complex. To test this hypothesis, we performed a second EMS mutagenesis screen with swsn-1; snfc-5 double mutants using this longer high-temperature period and screened for mutants that could develop to adulthood (Fig. 1E). From this screen, we identified and sequenced the genomes of 12 independently isolated mutants that developed to adulthood. We found that two of the mutant isolates have additional mutations in swsn-1 or snfc-5 (SI Appendix, Fig. S1I). These mutations are likely internal suppressor mutations and were not characterized further. Seven of the remaining isolates each carry different predicted loss-of-function mutations in the gene encoding the HECT-type E3 ubiquitin ligase UBR-5 and three have mutations in the HECT-type E3 ubiquitin ligase HECD-1, a paralog of UBR-5 ( Fig. 1F and SI Appendix, Fig. S1I). We recreated the UBR-5 Q150* allele, as it was the earliest premature stop mutation we identified (Fig. 1F), by CRISPR-Cas9 gene editing (21) (mj638). We then crossed swsn-1; snfc-5 double mutants to ubr-5(mj638) mutants and confirmed that this mutation in ubr-5 restored development to at least the L4 larval stage in approximately 70% of viable animals in the SWI/SNF mutant model ( Fig. 1 B-D and SI Appendix, Fig. S1H). Moreover, we found that approximately 29% of swsn-1; snfc-5; ubr-5 triple mutants did not die as embryos, which is a more than fivefold decrease of embryonic lethality compared to swsn-1; snfc-5 double mutants (SI Appendix, Fig. S1G). Similarly, we crossed swsn-1; snfc-5 double mutants with mutants harboring a deletion in hecd-1 (ok1437) and again confirmed that the loss of HECD-1 (introduction of the predicted hecd-1 null allele) restored development to the L4 larval stage in 46% of animals in our SWI/SNF mutant model ( Fig. 1 B-D and SI Appendix, Fig. S1H). We conclude that mutations in ubr-5 and hecd-1 can restore development to adulthood in swsn-1; snfc-5 double-mutant animals.
To test whether the loss of UBR-5 (introduction of the predicted ubr-5 null allele) is sufficient to restore developmental defects in swsn-1 single mutants even in the absence of the snfc-5 mutation, we generated swsn-1; ubr-5 double mutants and assayed their development under different conditions. We found that loss of ubr-5 alone increased the number of swsn-1 mutants that developed into gravid adults from 14 to 45% when embryos were grown at room temperature for 48 h, and then shifted to 25 °C for 24 h (Fig. 1 G-I and SI Appendix, Fig. S1J) but was not sufficient to restore developmental defects under conditions in which swsn-1 single mutants died as embryos ( Fig. 1 B-D and SI Appendix, Fig. S1H). These findings indicate that the mutation in snfc-5 is required to suppress the embryonic lethality caused by mutation in swsn-1, but that the loss of UBR-5 is sufficient to improve development from larvae to gravid adults in swsn-1 mutants when exposing them to high temperatures for a shorter period at later developmental stages.
Mutations in SWI/SNF subunits cause numerous defects in animal development and physiology. For example, the swsn-1 mutation also causes defective anchor cell (AC) invasion, a process required for establishing the uterine-vulval connection during larval development critical for adult egg-laying (22) (see the schematic of AC invasion in Fig. 1J). To test whether the suppressor mutations we identified specifically suppress the developmental arrest in swsn-1 mutants or whether they might generally suppress the loss of SWI/SNF function, we assayed AC invasion in wild-type, swsn-1 mutants, and various combinations of double-and triple-mutant animals. We found that swsn-1 single mutants exhibited defective AC invasion phenotype in 20% (10/50 animals) of animals ( Fig. 1K and SI Appendix, Fig. S1 K-L), consistent with previously published findings (22). This phenotype was partially rescued in swsn-1; snfc-5 double (5.2% invasion defects, 4/77 animals) and swsn-1; snfc-5; ubr-5 triple mutants (6.78% invasion defects, 5/74 animals), but not in swsn-1; ubr-5 double mutants (23% invasion defects, 7/30 animals) ( Fig. 1K and SI Appendix, Fig. S1 K-L). These data suggest that the mutation in snfc-5 can suppress multiple defects observed in swsn-1 single mutants, but that the loss of UBR-5 specifically suppresses the developmental arrest caused by loss of SWI/SNF function.

Loss of the UBR-5 Key Catalytic Residue Is Sufficient to Suppress
the Developmental Arrest of SWI/SNF Mutants. HECT-type E3 ubiquitin ligases such as UBR-5 and HECD-1 have a catalytic cysteine within their C-terminal HECT domains (23). Ligation reactions depend on this catalytic cysteine, which forms a thioester-linked intermediate with the ubiquitin before ligating it onto a substrate protein (23). We replaced the catalytic residue of UBR-5 (cysteine 2913) with an alanine or serine by CRISPR-Cas9 gene editing (21) (Fig. 2A). The C2913A substitution mutation should prevent the loading of ubiquitin onto UBR-5, whereas the C2913S substitution mutation should still enable the loading but prevent the transfer of ubiquitin onto a substrate (24). We found that substitution of the catalytic cysteine of UBR-5 to alanine or serine in swsn-1; snfc-5 double mutants resulted in similar suppression of the temperature-sensitive larval arrest as introducing the premature stop ubr-5 allele (mj638) (Fig. 2 B-D and SI Appendix, Fig. S2). These results indicate that inactivation of the catalytic function of UBR-5 is sufficient for the suppression of developmental arrest of swsn-1; snfc-5 mutants.

The Combined Loss of UBR-5 and HECD-1 Has Deleterious Effects
on C. elegans Reproduction. Since the loss of either UBR-5 or HECD-1 restored development to adulthood in some of the swsn-1; snfc-5 double mutants ( Fig. 1 B-D and SI Appendix, Fig. S1H), we wondered whether the combined loss of both E3 ubiquitin ligases would have an even greater effect. However, when generating swsn-1; snfc-5; ubr-5; hecd-1 quadruple mutants, we observed that those animals were substantially sicker and had fewer offspring than other mutant combinations. This effect appears to be a synthetic interaction between ubr-5 and hecd-1 because we found that ubr-5; hecd-1 double mutants exhibited a similar phenotype even in the absence of any SWI/SNF subunit mutations. Specifically, we found that the loss of either of the paralogous HECT-type E3 ubiquitin ligases alone did not affect animal reproduction, but the loss of both UBR-5 and HECD-1 resulted in a substantial reduction of live offspring. On average, the ubr-5; hecd-1 double mutants had about four-times fewer live offspring than either of the ubr-5 or hecd-1 single mutants (Fig. 2E). The quadruple mutants also had significantly fewer live offspring compared to the swsn-1; snfc-5; ubr-5 triple mutants and did not have significantly fewer offspring compared to the ubr-5; hecd-1 double mutants (Fig. 2E). Thus, UBR-5 and HECD-1 likely have redundant functions, and loss of one can be compensated by the other protein. To our knowledge, these findings indicate that these two ubiquitin ligases might function redundantly.

UBR-5 Regulates SWI/SNF Protein Levels.
Our findings suggest that the catalytic activity of UBR-5 is involved in suppressing the swsn-1 mutant developmental arrest ( Fig. 2  to tag them for proteasomal degradation (25). The human homologs of UBR-5 and HECD-1, UBR5 and HECTD1 respectively, have both been shown to mediate K48-linked ubiquitination (23), which is a signal for proteasomal degradation (26). A possible and direct link between the SWI/SNF complex and these two ubiquitin ligases could be that the swsn-1 mutation destabilizes the SWI/SNF complex and that these enzymes ubiquitinate unstable complexes to promote their degradation. In this case, loss of UBR-5 should result in higher levels of SWSN-1, which in turn might explain the observed developmental arrest suppression. To test this hypothesis, we measured SWSN-1 protein levels in wild-type animals, swsn-1 single, swsn-1; snfc-5 double and swsn-1; snfc-5; ubr-5 triple mutants by western blotting. We generated endogenously FLAGtagged wild-type and mutant versions of SWSN-1 (Fig. 3A). We synchronized L1-staged animals by starvation and subsequently fed them and exposed them to the restrictive temperature (25 °C) for 6 h. These conditions were chosen so that all of the mutants would be at closely matched developmental stages. We found that levels of SWSN-1 were on average approximately 40% reduced in swsn-1 single mutants when compared to wild-type animals, suggesting that the P86L mutation reduces SWSN-1 protein levels. This reduction in SWSN-1 levels was largely restored in swsn-1; snfc-5; ubr-5 triple mutants (Fig. 3B). However, these effects were highly variable, and the levels of SWSN-1 were not statistically significantly different between swsn-1; snfc-5 double mutants and swsn-1; snfc-5; ubr-5 triple mutants (Fig. 3B) even though we found that most swsn-1; snfc-5 double mutants arrested their development at early larval stages, while most swsn-1; snfc-5; ubr-5 triple mutants were able to develop to L4-stages and adulthood ( Fig. 1 B-D and SI Appendix,  Fig. S1H). These results suggest that mutation of swsn-1 leads to a reduction of SWSN-1 protein levels. Furthermore, UBR-5 potentially has either a small effect on SWSN-1 protein levels or an effect that occurs only in specific cells or at specific developmental stages that is difficult to detect when measuring SWSN-1 levels in whole animals by western blotting.
As an alternative approach to assessing SWSN-1 protein levels at the single cell level in the different mutant animals, we measured SWSN-1::EGFP levels in the AC of mid-late L3-staged animals that had been exposed to the restrictive temperature (25 °C) from the L2-L3 molt/early L3 stage until the P6.p 4-cell stage by confocal fluorescence microscopy. For this purpose, we obtained an available SWSN-1::EGFP strain, in which we introduced the P86L temperature-sensitive mutation (Fig. 3C) and crossed it to snfc-5 and ubr-5 mutants to generate double and triple mutants. Consistent with the western blot quantifications, we found that SWSN-1 protein levels in the AC were significantly reduced in swsn-1 single and swsn-1; snfc-5 double mutants compared to wildtype animals (Fig. 3 D and E and SI Appendix, Fig. S3A). In this cell-specific context, we observed a statistically significant increase of SWSN-1 levels in swsn-1; snfc-5; ubr-5 triple mutants when compared to swsn-1 single and swsn-1; snfc-5 double mutants (Fig. 3 D and E and SI Appendix, Fig. S3A), indicating that UBR-5 regulates SWSN-1 levels in the AC.
To gain better insight into the role of UBR-5 in regulating the abundance of SWI/SNF subunits, we employed a mass spectrometry approach to look at alterations in protein levels in wild-type animals, swsn-1; snfc-5 double mutants and swsn-1; snfc-5; ubr-5 triple mutants. Label-free quantification (LFQ) of whole proteome samples from synchronized L1-staged animals revealed a significant decrease in SWSN-1 protein levels in swsn-1; snfc-5 double mutants compared to wild-type animals (Fig. 3F). Consistently with western blotting results, SWSN-1 levels were partially restored to wild-type levels in swsn-1; snfc-5; ubr-5 triple mutants (Fig. 3F), a trend that was consistent in all detected SWI/SNF subunits (Fig. 3G). Notably, every subunit of the complex showed this same pattern; the highest levels were in the wild-type animals, reduced levels in the swsn-1; snfc-5 double mutants, and partially recovered levels in the swsn-1; snfc-5; ubr-5 triple mutants; however, the mean protein abundance of each subunit was different. Therefore, we performed a multiple linear regression analysis to determine the effect of the mutation on the entire complex by regressing out the effect of the different mean expression levels. This analysis revealed that the SWI/SNF complex, as a whole, is significantly depleted in the swsn-1; snfc-5 double mutants when compared to wild-type animals and that this depletion is partially rescued in swsn-1; snfc-5; ubr-5 triple mutants (Fig. 3H and SI  Appendix, Fig. S3 B and C). The same analysis with 100 randomly chosen sets of twelve proteins was not significant (depicted by the intervals to the right of the boxplots in Fig. 3H). SI Appendix, Fig.  S3 D and E shows an example of a regression analysis for one set of twelve randomly selected proteins (ARX-6, EXOS-1, CYN-1, NBET-1, EMB-4, ZK1236.5, VPS-29, TFTC-3, MDT-9, TBA-1, HMT-1, and Y39G8B.1). All together, these data indicate that the loss of UBR-5 results in increased protein levels of all SWI/ SNF subunits in the swsn-1; snfc-5 mutant model. Furthermore, these results suggest that UBR-5 likely ubiquitinates the SWI/ SNF complex to tag it for proteasomal degradation and that the loss of UBR-5 likely prevents some of the developmental defects exhibited in SWI/SNF mutants by increasing SWI/SNF complex abundance. Our findings that levels of all SWI/SNF subunits of swsn-1; snfc-5 double mutants are reduced compared to wild-type animals are consistent with previously published mouse data (16).
To determine how UBR-5 and HECD-1 regulate protein levels in animals, we similarly obtained mass spectrometry proteomics data of wild-type animals and ubr-5 single and hecd-1 single mutants. When comparing protein levels of SWI/SNF subunits, we found that SWSN-7 and PHF-10 are significantly up-regulated (FDR < 0.05) by 11 and 23%, respectively, in ubr-5 single mutants compared to wild-type animals (SI Appendix, Table S1) and that SWSN-3 is significantly up-regulated (FDR < 0.05) by 10% in hecd-1 single mutants compared to wild-type animals (SI Appendix, Table S2). These data suggest that the steady-state levels of specific individual SWI/SNF subunits could also be regulated by UBR-5 and HECD-1 ubiquitination but that the loss of UBR-5 or HECD-1 does not broadly increase the abundance of all SWI/ SNF subunits in wild-type animals.
It remains possible that in addition to regulating SWI/SNF protein levels, UBR-5 also ubiquitinates other proteins and that restoring the levels of these proteins also helps SWI/SNF mutants develop to adulthood. Our proteomics analysis identified twelve significantly up-regulated proteins (fold-change ≥ 1.  , Fig. S3F). The collective up-regulation of those 12 proteins in swsn-1; snfc-5 mutants might contribute to the suppression of observed developmental defects. Furthermore, using the proteomics data of ubr-5 and hecd-1 single mutants, we could identify 21 proteins significantly up-regulated in both mutant conditions (SI Appendix, Fig. S3G), suggesting that the regulation of those proteins might be a redundant function of the two ubiquitin ligases.

Mutations That Suppress the Embryonic Lethality of swsn-1 Mutants Restore Wild-type Expression of 335 SWI/SNF-Regulated
Genes. As chromatin remodelers, the SWI/SNF complexes are thought to control the expression of various genes by enabling or preventing DNA access for the transcription machinery. This involves positioning of nucleosomes to expose promoter and enhancer sequences, thereby enabling the binding of transcription factors and RNA polymerase II (RNA pol II). Similarly, SWI/ SNF complexes can facilitate the binding of repressors and disable access to transcription start sites (TSS) (1,13). Previous studies analyzing the transcriptomes of young adult-staged swsn-1 mutants reported that between 7.5% (27) and approximately one third (28) of C. elegans genes are regulated by the SWI/SNF complex. To identify changes in gene expression that might lead to the developmental defects of swsn-1 mutants, we conducted three independent RNA-sequencing experiments (sets 1 to 3) using synchronized L1-staged animals (same conditions as for the quantitative western blotting and proteomics experiments). We analyzed gene expression profiles of wild-type animals, ubr-5 mutants, swsn-1 mutants, and various combinations of doubleand triple-mutant animals to investigate how these different mutations affect SWI/SNF-regulated gene expression. We found that approximately 8% of C. elegans protein-coding genes (1803)   were differentially expressed in swsn-1 single mutants compared to wild-type animals (Fig. 4A). The expression changes of most of these genes were restored to wild-type levels in the swsn-1; snfc-5 double, swsn-1; snfc-5; ubr-5 triple, and swsn-1; snfc-5; hecd-1 triple suppressor mutants but not restored in swsn-1; ubr-5 double mutants (Fig. 4A). The gene expression profiles of swsn-1; ubr-5 double mutants were similar to those of swsn-1 single mutants (SI Appendix, Fig. S4A) which is consistent with our previous findings that both of these mutant backgrounds exhibit embryonic lethality under heat-shock conditions where parent animals are exposed to 25 °C for 16 h prior to isolating embryos and exposing them to 25 °C (Fig. 1 B-D). These data are consistent with our hypothesis that the snfc-5 A258V mutation generally suppresses the defects caused by the swsn-1 mutation and that the ubr-5 mutation alone is not sufficient to suppress defects when animals are exposed to the restrictive temperature at early larval stages. ubr-5 single mutants had only 62 differentially expressed genes compared to wild-type animals (Fig. 4A).
We hypothesized that the suppressor mutants we identified restore developmental defects of swsn-1 mutants by restoring the expression of developmentally critical genes. However, identifying the precise SWI/SNF-regulated genes that promote animal development is complicated by the asynchronous developmental rate of SWI/SNF mutant animals (see developmental progression estimation for the sequenced animals using an available RNA-seq timecourse data set in Fig. 4B), which can be slower than that of wild-type animals. To not confound developmentally important genes with genes that are differentially expressed due to developmental staging differences, we compared differentially expressed genes between mutant backgrounds that exhibited early developmental arrest (swsn-1 single mutants and swsn-1; ubr-5 double mutants, see SI Appendix, Fig. S1G) and genetic backgrounds that did not exhibit early developmental arrest (wild-type animals, swsn-1; snfc-5 double mutants, swsn-1; snfc-5; ubr-5 triple mutants, swsn-1; snfc-5; hecd-1 triple mutants, and ubr-5 single mutants) at the developmentally synchronized L1 larval stage (i.e., all genetic backgrounds were synchronized as L1 stage animals). Furthermore, we performed these RNA-seq experiments in multiple different "sets" (each with four or five biological replicates) and focused only on genes exhibiting consistent expression changes in all experiments. From this filtered analysis, we identified 335 genes (out of the 1,083 genes differentially expressed in swsn-1 single mutants when compared to wild-type animals) that were consistently differentially expressed in the strongly developmentally arrested swsn-1 single and swsn-1; ubr-5 double mutants ( Fig. 1 B-D and SI Appendix, Fig. S1 C-H) but displayed near wild-type gene expression in all mutant backgrounds that did not exhibit the early larval arrest (Fig. 4C). We conclude that the dysregulation of some subset of these 335 genes is likely at least partially responsible for driving the early developmental arrest that occurs in swsn-1 single mutants.
To evaluate the potential direct targets of altered SWI/SNF function on nucleosome positioning at gene promoters, we performed MNase-sequencing of synchronized L1-staged (same conditions as for RNA-seq) wild-type animals, swsn-1 single, swsn-1; snfc-5 double and swsn-1; snfc-5; ubr-5 triple mutants to obtain genome-wide nucleosome coverage profiles. MNase digests in nucleosome-free regions resulting in a higher signal corresponding to regions protected by nucleosomes, which we refer to as higher nucleosome coverage. This can result from increased nucleosome density (i.e., increased probability of a nucleosome being present at all) or a more precisely positioned nucleosome (i.e., the nucleosome is less "fuzzy") (30). Since only promoters of ubiquitous and germline-specific genes have clear and well-positioned nucleosomes (31) and we sequenced C. elegans L1 larvae that do not have extensive germline tissue (32), the MNase-seq analysis was restricted to ubiquitous genes. The clearest difference we observed between wild-type animals and swsn-1 single mutants was a strongly reduced +1 nucleosome coverage (the first nucleosome downstream of the promoter). The +1 nucleosome coverage in both the swsn-1; snfc-5 double and swsn-1; snfc-5; ubr-5 triple suppressor mutants was higher than in swsn-1 single mutants ( Fig. 4D and SI Appendix, Fig. S4B). These findings indicate that SWI/SNF regulates the nucleosomes at the +1 position of ubiquitously expressed genes and that our suppressor mutations can partially restore the defects observed in swsn-1 single mutants. Notably, the reduction of the +1 nucleosome coverage in swsn-1 mutants relative to wild-type nucleosome coverage can also be seen comparing across all ubiquitous promoters separately, indicating a consistent difference that affects most ubiquitous promoters (Fig. 4E and SI  Appendix, Fig. S4C). The reduced coverage could reflect less well-positioned +1 nucleosomes or reduced nucleosome density at the +1 region. These data suggest that mutation of swsn-1 affects the chromatin remodeling function of the SWI/SNF complex, which might in turn explain the observed changes in gene expression.
We next asked whether our set of 335 SWI/SNF-regulated genes (defined in Fig. 4C) has a distinct nucleosome positioning profile. To do so, we first asked whether genes with altered expression showed larger changes in nucleosome coverage than genes that did not show changes in expression. We compared the +1 nucleosome coverage of all ubiquitous genes to ubiquitously up-(green) and down-regulated (purple) SWI/SNF-regulated genes, which are subsets of the 335 genes. This analysis revealed that the promoters of ubiquitous genes altered in swsn-1 mutants had a similar reduction in +1 nucleosome coverage relative to wild-type animals (Fig. 4F). These results suggest that the coverage of the +1 nucleosome at gene protomers changes for many genes in swsn-1 single mutants, but this does not always result in changes in their expression.
Interestingly, we found that the subsets of ubiquitously up-(green) and down-regulated (purple) SWI/SNF-dependent genes have a higher nucleosome coverage in wild-type animals when compared to all ubiquitous genes in wild-type animals (Fig. 4G). Fig. 4F indicates that swsn-1 mutants also have a higher nucleosome coverage in this subset of genes. These findings confirm that the swsn-1 mutation studied here does alter proper nucleosome positioning. These findings also show that the 335 SWI/ SNF-regulated genes tend to have more well-positioned +1 nucleosomes or a higher nucleosome density than (ubiquitous) genes on average. This could suggest that SWI/SNF-regulated genes depend on a well-positioned +1 nucleosome or on a high nucleosome density for their regulation and that other genes might be more robust to changes in nucleosome coverage.

Discussion
Here, we identified that the combination of a specific missense mutation in snfc-5 and the loss of either of two E3 ubiquitin ligases (UBR-5 or HECD-1) can suppress some of the developmental defects caused by a missense mutation in the core SWI/SNF subunit swsn-1. UBR-5 and HECD-1 are genetic interactors of the SWI/ SNF complex, and our studies revealed a previously unknown functional redundancy between these two ubiquitin ligases in regulating animal development and fertility. In addition, we established that swsn-1; snfc-5 double mutants have reduced SWI/SNF protein levels and that the loss of UBR-5 can partially restore these levels. Together, these results are consistent with a model in which UBR-5 and HECD-1 promote the degradation of the SWI/SNF complex and that the loss of UBR-5 or HECD-1 can prevent developmental defects from manifesting by increasing SWI/SNF levels (Fig. 5). SWI/SNF levels have been shown to be critical for cell proliferation versus differentiation decisions. Lower levels of SWI/SNF proteins or activity have been associated with cell proliferation both in the M cell lineage (33) and the AC (22) in C. elegans. Therefore, by regulating SWI/SNF protein levels, UBR-5 and HECD-1 may be critical for dose-dependent roles of the SWI/SNF complex. In humans, missense mutations in SWI/SNF complex subunits cause a multitude of developmental disorders and cancer and are associated with alcohol use disorders. These include ClinVar annotated missense mutations in the SWIRM domain of human SMARCC1 (His526Pro) and SMARCC2 (Tyr465His, Asn454Asp), analogous to the SWIRM domain mutation in swsn-1 studied here. Most of these disease-associated mutations in SWI/SNF subunits are heterozygous. Heterozygous mutations identified in developmental disorders are generally dominant, implying that dosage-sensitive processes underlie the roles of SWI/SNF complexes in development (6). Our findings suggest that the inhibition of UBR5 and HECTD1 could be a potential therapeutic strategy to treat developmental disorders caused by dosage-sensitive missense mutations in SWI/ SNF complex subunits. SWI/SNF Complex Stability. The human SMARCC1/2 dimer is important in early SWI/SNF complex assembly where it serves as a scaffold to which other subunits bind (7,8). Studies of SWI/ SNF complex assembly found that loss of SMARCC1/2 in mice leads to the dissociation and degradation of the entire complex, presumably because unassembled subunits are less stable (16). In the SWI/SNF mutant C. elegans model we used here, a P86L mutation in the SWIRM domain of the SMARCC1/2 homolog swsn-1 resulted in embryonic lethality at 25 °C. We rescued the embryonic lethality by introducing a A258V mutation in the RPT2 domain of snfc-5, homolog of SMARCB1. The SWIRM domains of human SMARCC1/2 and yeast SWI3 dimers directly bind the RPT1 and RPT2 domains of SMARCB1 and SNF5, respectively (8,34). Based on these data, we propose a model in which the P86L SWSN-1 mutation results in a structural change which prevents stable binding to SNFC-5 and complex assembly. This is supported by the fact that the allele is temperature sensitive, as we expect destabilization to be exacerbated by increased temperature. The A258V SNFC-5 mutation in turn introduces another structural change that allows SNFC-5 to bind mutant SWSN-1 more stably (Fig. 5). The resulting swsn-1; snfc-5 mutant complex is partially stable, but not as stable as the wild-type assembly, since we find that the levels of all SWI/SNF subunits in swsn-1; snfc-5 double mutants are reduced compared to wild-type animals (Fig. 3  G and H). Consistent with this, it has previously been proposed that loss of one SWI/SNF subunit could alter the abundance of the other subunits (10).
We also showed that the levels of all SWI/SNF subunits are partially restored to wild-type levels in swsn-1; snfc-5; ubr-5 triple mutants (Fig. 3 G and H). Moreover, even though the loss of either UBR-5 or HECD-1 did not substantially affect SWI/SNF levels in wild-type animals, we found that some individual SWI/ SNF subunits were increased by approximately 10 to 20% in ubr-5 and hecd-1 single mutants (SI Appendix, Tables S1 and S2). This suggests that UBR-5 and HECD-1 can also affect the steady-state levels of wild-type SWI/SNF subunits. Overall, our data are consistent with a model in which UBR-5 (and presumably also HECD-1) directly ubiquitinates either SWSN-1 or another subunit of the SWI/SNF complex to regulate SWI/SNF protein levels. Specifically, our data suggest that swsn-1 single mutants form an unstable SWI/SNF complex that consequently becomes degraded by the proteasome, resulting in a modest reduction of SWI/SNF protein levels and embryonic lethality. By contrast, swsn-1; snfc-5 double mutants form a partially stable complex that is less prone to degradation resulting in increased SWI/SNF complex abundance, when compared to swsn-1 single mutants, and the prevention of embryonic lethality. Finally, the loss of UBR-5 or HECD-1 further increases SWI/SNF complex abundance in swsn-1; snfc-5 double mutants by preventing some of the turnover of the complex by the proteasome. This ultimately promotes SWI/SNF function and enables swsn-1; snfc-5; ubr-5 triple mutants to develop to adulthood (Fig. 5).
Synergistic Functions of UBR-5 and HECD-1. We identified two paralogous HECT-type E3 ubiquitin ligases as suppressors of SWI/SNF mutation in our screen. Despite the finding that the loss of either of the two enzymes alone appears to promote animal health in swsn-1; snfc-5 mutants, we found that the combined loss of these two proteins has deleterious effects on animal health. We report a synergistic genetic interaction between these two ubiquitin ligases. These findings suggest that UBR-5 and HECD-1 likely function at least partially redundantly. Our data indicate that the SWI/SNF complex is one likely such common target of UBR-5 and HECD-1; however, our proteomics profiling of ubr-5 and hecd-1 single mutants showed only small increases of different individual SWI/SNF subunits. We suspect that due to the potentially redundant functions of UBR-5 and HECD-1, there would be substantially larger changes in SWI/SNF subunit abundance and other protein abundances in ubr-5; hecd-1 double   In wild-type animals, a stable SWI/SNF complex can assemble at 25 °C (indicated by a pink box and a thermometer). UBR-5 and HECD-1 ubiquitinate SWI/SNF complexes for proteasomal degradation to maintain their steady-state levels.
The swsn-1 mutation prevents the assembly of the SWI/SNF complex or leads to the assembly of an unstable complex at 25 °C due to structural changes that impair the interaction of SWSN-1 and SNFC-5. Mutant SWI/SNF complexes are more frequently degraded by the proteasome, which is also mediated by UBR-5 and HECD-1 ubiquitination. Additional mutation of snfc-5 enables the assembly of a partially stable mutant SWI/SNF complex that gets overall less frequently degraded by the proteasome. Loss of UBR-5 leads to an increase of SWI/SNF complex protein levels in swsn-1; snfc-5 mutants by preventing some of the turnover of the complex by the proteasome.
mutants which cannot properly regulate protein levels via either ubiquitin ligase. These dysregulated protein levels might in turn drive the strong synthetic phenotypes observed in ubr-5; hecd-1 double mutants.

UBR-5 Function at Later Developmental Time Points and in
Specific Tissues. Despite our finding that swsn-1; snfc-5; ubr-5 triple mutants could develop to adulthood, making them better developmental suppressors than the swsn-1; snfc-5 double mutants, we did not observe an improved rescue of gene expression or of nucleosome coverage in the triple mutants when compared to double mutants. We suspect that the loss of UBR-5 on SWI/SNFdependent gene expression likely becomes more apparent at later developmental stages. This is supported by our observations that ubr-5 alone, in the absence of any snfc-5 mutation, was capable of partially suppressing swsn-1 single mutants if they survived to a later (post L1) larval stage ( Fig. 1 G-I and SI Appendix, Fig.  S1J). Furthermore, the effect of loss of UBR-5 or HECD-1 could be tissue specific and not detectable in synchronized L1 animals, which is supported by the clear effect of the ubr-5 mutation on SWSN-1 levels that we observed in the AC of L3-staged animals ( Fig. 3 D and E), compared to synchronized L1s (Fig. 3 B and F). It has also been shown that certain SWI/SNF accessory subunits have an impact on specific developmental processes, which adds another layoff of regulation to SWI/SNF. For example, PBRM-1 is important for somatic gonadal precursors development, while LET-526 for the distal tip cell (18) and SWSN-2.1 and SWSN-2.2 have distinct roles in embryonic development (35). UBR-5 may have higher activity on certain accessory subunits at different stages in development. Future studies looking at different developmental time points or specific tissues might shed light on how and whether UBR-5 affects the gene expression of SWI/SNF targets during development.
Evolution of SWI/SNF and Complex Stability. Interestingly, many animals, including humans, mice, chicken, zebrafish, and fruit flies, all already have a valine at the equivalent position as the A258V substitution in the RPT2 domain of SNFC-5 that we identified as a suppressor in the first mutagenesis screen (SI Appendix, Fig. S1B). Moreover, one of the mutants isolated in the second mutagenesis screen carries the additional V129I substitution mutation in the SWIRM domain of swsn-1, and the human, mouse, and fruit fly SWSN-1 homologs also have an isoleucine at the equivalent position (SI Appendix, Fig. S1A). These findings suggest that the valine and isoleucine at these specific positions could make the SWI/SNF complex more robust, and these exact substitutions might have played a role in the evolution of SWI/SNF function. For example, these specific substitutions might help animals better adapt to higher temperatures such as those found in warm-blooded animals or even the poikilotherm model animals zebrafish and fruit flies, which develop at approximately 28 °C (36) and 25 °C (37), respectively.

Materials and Methods
C. elegans Strain Maintenance. C. elegans strains were grown and maintained on nematode growth medium (NGM) agar plates seeded with HB101 bacteria (Caenorhabditis Genetics Center, University of Minnesota, Twin Cities, MN, USA) as food source (38). Strains containing the temperature-sensitive swsn-1 mutation were routinely kept at 15 °C unless otherwise indicated, and other strains were routinely kept at 20 °C or also at 15 °C to match growth conditions of temperature-sensitive strains. The C. elegans strains that were used in this study are derived from the Bristol N2 strain and are listed in SI Appendix, Table S3.