The C. elegans homolog of TMEM132D, a human panic-disorder and anxiety risk gene, modulates neuronal morphogenesis through the WAVE-regulatory complex

TMEM132D is a human gene identified with multiple risk alleles for panic disorders, anxiety and major depressive disorders. Belonging to a conserved family of transmembrane proteins, TMEM132D and its homologs are still of unknown molecular functions. By generating loss-of-function mutants of the sole TMEM132 ortholog in C. elegans, we identify abnormal morphologic phenotypes in the dopaminergic PDE neurons. Using a yeast two-hybrid screen, we find that NAP1 directly interacts with the cytoplasmic domain of human TMEM132D, and mutations in C. elegans tmem-132 that disrupt the interaction with NAP1 cause similar morphologic defects in the PDE neurons. NAP1 is a component of the WAVE regulatory complex (WRC) that controls F-actin cytoskeletal dynamics. Decreasing activity of WRC rescues the PDE defects in tmem-132 mutants, whereas gain-of-function of TMEM132D in mammalian cells inhibits WRC, leading to decreased abundance of selective WRC components, impaired actin nucleation and cell motility. We propose that metazoan TMEM132 family proteins play evolutionarily conserved roles in regulating NAP1 protein homologs to restrict inappropriate WRC activity, cytoskeletal and morphologic changes in the cell.


INTRODUCTION 19
Despite decades of genetic and molecular analyses, the genome of the common model HSPC300, ABI1/2, WAVE1/2/3 and NAP1 (also known as NCKAP1) proteins. NAP1 was initially 44 identified as a gene with strongly decreased expression in the brain of patients with sporadic 45 Alzheimer's disease (Suzuki et al., 2000). Deleterious NAP1 variants were also identified in actin cytoskeletal changes has not been reported. It has also been unclear how abundance of 52 WRC components is regulated in cell compartments where actin nucleation needs to be limited 53 in morphologically complex cells, including neurons. 54 To elucidate biological functions and mechanisms of action of TMEM132 family proteins, 55 we generated and characterized C. elegans loss-of-function (LOF) mutations in Y71H2AM.10, 56 the sole ortholog of the TMEM132 gene family. tmem-132 mutants exhibit striking morphological 57 defects in the dopaminergic PDE neurons. We further identified human NAP1 as a TMEM132D 58 interactor and show that the C. elegans homologs also interact with each other. Genetic 59 interactions between tmem-132 and WRC-encoding genes in regulating PDE morphology, the 60 LOF phenotype of tmem-132 mutants and gain-of-function (GOF) phenotype of TMEM132D in 61 mammalian cells collectively suggest that TMEM132 family proteins regulate NAP1 levels in 62 WRC to finely modulate actin nucleation, cellular cytoskeletal and morphological changes. 63

65
The C. elegans TMEM-132 localizes to neurons and regulates the dopaminergic PDE 66 neuron morphology 67 Protein homology and motif analysis identified the Pfam16070 domain characteristic of 68 both human TMEM132D and C. elegans TMEM-132, classifying both to the evolutionarily 69 conserved TMEM132 protein superfamily ( Figure 1A; Figure S1). Interestingly, a translational 70 reporter that fuses tmem-132 with GFP revealed enriched expression of C. elegans tmem-132 71 in neurons ( Figure 1B-D), implicating a specific role of TMEM-132 in the nervous system. Lack 72 of mammalian loss-of-function models and potential genetic redundancy among the 73 TMEM132A-E family members precluded us from analyzing the physiological function of 74 TMEM132D in vivo. Thus, we sought to address this issue in C. elegans, which encodes tmem-75 132 as the sole ortholog of the gene family and has served as excellent model to study neuronal 76 cell biology (Inberg et al., 2019; Richardson and Shen, 2019; Tang and Jin, 2018). We used 77 CRISPR-Cas9 techniques to generate a series of C. elegans mutants, including multiple 78 independently-derived deletions, an early stop-codon mutant and a genetic P784T knock-in 79 mutant, in which the highly conserved proline residue became threonine, corresponding to the 80 human disease risk allele for anxiety and panic disorders ( Figures 1E and 1F). We generated 81 such multiple independent mutations to seek convergent phenotype and outcrossed all mutants 82 to eliminate potential interference of phenotype by background mutations. 83 Given the exclusive localization of TMEM-132 in neurons, we subjected tmem-132 84 mutants to a variety of neuronal phenotypic analyses. We did not observe gross behavioral 85 defects under normal conditions. We next crossed the mutants to various established GFP 86 reporters to examine neurons of stereotyped morphology, including ciliated sensory AWC 87 neurons, hermaphrodyte-specific HSN neurons, mechanosensory PVD, ADE and PDE neurons 88 ( Figure S2). Among the neurons examined, the dopaminergic neuron PDE exhibited the most 89 severe defect, thus in this study we focused on PDE, which is marked by the osm-6p::GFP 90 reporter in addition to other ciliated and dopaminergic neurons. Although dense GFP signals 91 prevented us from analyzing the anterior group of ciliated and dopaminergic neurons, close 92 confocal microscopic analysis of the posterior, anatomically isolated PDE neurons revealed 93 striking abnormal morphologies in a large fraction of tmem-132 mutants (Figures 2A-D). We 94 categorized the mutant phenotype into several classes, including those with irregular soma 95 outlines, ectopic dendrites, ectopic axon branches, and axon misguidance as similarly described 96 previously (Shakir et al., 2008;Sulston et al., 1975). Although the phenotypic defects of PDE 97 are diverse, all mutants show similar profiles in distribution of different categories of phenotypic 98 defects (Figures 2Eand 2F). tmem-132 LOF mutants also exhibited morphological defects in the 99 ADE and PVD but not morphologically less complex AWC neurons ( Figure S2). Neuronal interactors from a human-cDNA prey library. We also constructed a bait vector that contains the 113 homologous C-terminus of C. elegans tmem-132 and focused on identified screen hits whose 114 homologs can interact with human and C. elegans baits, respectively. From 117 independent 115 cDNA clones isolated and identified by Sanger sequencing (Table S1), we found that the protein 116  . We found that GEX-3, the C. elegans 120 ortholog of NAP1, also interacts with the C-terminus of C. elegans TMEM-132 ( Figure 3A). 121 In addition to NAP1, the WRC also contains three other major proteins SRA1, ABI1/2 122 and WAVE1/2/3, with corresponding orthologs gex-2, abi-1 and wve-1 in C. elegans ( Figure 3B induced pluripotent stem cells (iPSC), GFP-tagged TMEM132D markedly co-localized with 132 mCherry-tagged NAP1 ( Figure 3D). Collectively, these genetic, biochemical and cellular 133 imaging results identify NAP1 as a protein interactor of TMEM132D and indicate that such 134 interaction is evolutionarily conserved also for C. elegans counterparts. 135

136
The C-terminal domain and the conserved P784 are required for the interactions between 137

TMEM132 and WRC components 138
We used Y2H and CoIP assays to further define the C-terminal domain of human 139 TMEM132D or C. elegans TMEM-132 that is crucial for interacting with C-termini of NAP1 140 homologs. To examine interaction between GEX-3 (C. elegans NAP1 homolog) and TMEM-132, 141 we generated transgenic strains in which HA epitope-tagged GEX-3 and mCherry-tagged 142 TMEM-132 Ct are co-induced by heat shock promoters ( Figure 4A). Using the mCherry 143 nanobody-Trap CoIP assay, we found that mCherry-tagged TMEM-132 Ct specifically pulled 144 down HA epitope-tagged GEX-3, compared with heat shock-induced mCherry only as control 145 ( Figure 4A). We also confirmed interaction between GEX-3 Ct and TMEM-132 Ct in Y2H 146 assays, in which the most C-terminal 60 a.a. of GEX-3 was sufficient to mediate the interaction 147 ( Figure 4B). In the C-terminus homologous to the WIRS-containing TMEM132D, mutation of a 148 WIRS-like motif attenuated TMEM-132 interaction with GEX-3 ( Figure 4B). Furthermore, we 149 generated mutations to convert the conserved proline 784 to alanine or threonine (to model 150 psychiatric disorder-associated risk allele in humans, see Figure 1F) in TMEM-132 and found 151 that both mutations abolished the interaction with GEX-3 ( Figure 4C). 152 Previous studies revealed that the WIRS of diverse transmembrane proteins in 153 mammals mediates binding to an interaction surface of WRC (Chen et al., 2014). As 154 TMEM132D contains such a motif at its C-terminus, we performed mutation analysis in Co-IP 155 assays and found that deletion of the entire cytoplasmic portion, deletion of the 120 a.a. C-156 terminus or mutation of the WIRS-like motif in TMEM132D markedly attenuated its interaction 157 with Nap1 ( Figure 5A). Deletion of the C-terminal 60 a.a. of TMEM132D did not appear to affect 158 the interaction, indicating that additional sites beyond the 60 a.a. may interact with WIRS and 159 contribute to interaction with Nap1. We made similar observations for Sra1 ( Figure 5B), 160 consistent with previous structural findings that Nap1 and Sra1 form an integral heterodimeric 161 sub-complex of WRC (Chen et al., 2010). Systematic deletion mutation analysis using Co-IP 162 and Y2H assays underscored the importance of C termini of TMEM132 family proteins from 163 both C. elegans and humans in interacting with Nap homologs ( Figure 5C). Since canonical 164 WIRS binds to a composite surface formed by Sra and Abi but not Nap (Chen et al., 2014), our 165 results suggest that TMEM132 differs from certain canonical WRC ligands, such as PCDH10, in 166 specific interaction with WRC components, consistent with the idea that TMEM132 acts to 167 sequester selective components of WRC rather than to recruit or activate WRC. 168

WRC acts downstream of TMEM-132 to regulate morphology of the PDE neurons 170
Since C. elegans TMEM-132 binds to GEX-3 as human TMEM132D binds to NAP1, we 171 next addressed whether TMEM-132 regulates neuronal morphology via WRC in C. elegans. We 172 generated an integrated transgenic reporter with neuronal specific expression of the WRC 173 component ABI-1 fused to GFP. We found that ABI-1::GFP in the nerve ring, along with the 174 ganglia of the head and tail in C. elegans, is weakly fluorescent in wild type animals but strongly 175 up-regulated in tmem-132 mutants ( Figures 6A-C). Close microscopic analysis of ABI-1::GFP 176 specifically in PDE neurons revealed that ABI-1::GFP forms puncta, numbers of which decrease 177 from the larval to adult stages in wild type animals ( Figure 6D). By contrast, numbers of ABI- To test the prediction of the idea that TMEM132D inhibits WRC, we used the LifeAct 216 reporter and a wound-recovery assay to examine effects of TMEM132D expression on actin 217 nucleation and cell motility, respectively. LifeAct is a 17-amino-acid polypeptide that labels 218 filamentous actin (F-actin) structures; its fusion with GFP allows visualization and quantification 219 of actin nucleation in eukaryotic cells (Riedl et al., 2008a). We found that expression of epitope-220  Figure 7D). Furthermore, a wound-recovery assay showed that ectopic 232 TMEM132D-expressing cells exhibited strongly reduced motility during the 24 and 48 hrs 233 recovery phases after scratching-induced wounding in cultured cells (Figures 7E and 7F).  type and tmem-132 mutants, including those carrying three independent deletion alleles, the 516 P784T knock-in allele and the Q781X nonsense allele that truncates the C-terminus of TMEM-517 132. All strains carry the PDE reporter osm-6p::GFP and were outcrossed to minimize potential 518 effects of background mutations. Scale bars: 5 µm. 519 Gene Name Occurrence frequency Gene Name Occurrence frequency NCKAP1 3