To better understand the regulatory mechanisms of polarized secretion in epithelia, we deployed apically localized ERM-1-GFP (a membrane-cytoskeleton linker) and basolateral SLCF-1-GFP (a monocarboxylate transporter) as cargos for a genome-wide RNAi screen. We found that loss of W09D10.1 led to defective secretion in the C. elegans intestine. W09D10.1 encodes the sole C. elegans homolog of mammalian SMAPs (Funaki et al., 2013; Sato et al., 1998; Tanabe et al., 2005), which we referred to as SMAP-1. Sequence alignment indicated that the N-terminal region (aa 15-119) of SMAP-1 shares 68% identity with SMAP2, while their C-terminal regions lack significant homology. SMAP-1 contains an N-terminal Arf GAP domain (aa 20-128) (Figure 1B). Unlike the C-terminus of SMAP2, which harbors a clathrin-interacting domain and a CALM-interacting domain (Natsume et al., 2006), the C-terminal part of SMAP-1 has not been characterized yet.

In accordance with the predicted expression profile (Spencer et al., 2011), SMAP-1 is broadly expressed in tissues such as the intestine, neuron, and pharynx (Supplementary Figure S1). Whole-animal knockout of SMAP-1 causes larval arrest (Gonczy et al., 2000). Hence, we prepared the transgenic allele smap-1(ycxEx1639), a heat-shock-inducible CRISPR/Cas9 conditional mutant (Supplementary Figure S2A-A′). In smap-1(ycxEx1639) intestinal cells (Figures 1A,B), ERM-1-GFP consistently accumulated on the basal side, while SLCF-1 abnormally appeared on the apical membrane (Figure 1C-C′). Similarly, the localization of apical cargo protein NHX-2 (Na+/H+ exchanger) and basolateral recycling regulator LET-413/Erbin were affected (Figure 1C-C′). Previous studies suggested that SMAP2 could act as an Arf1GAP (Arf1 GTPase-activating protein) and regulate the formation of the clathrin coat on the trans-Golgi network (TGN) (Thomas et al., 2021), leading us to examine the distribution of GFP-ARF-1.2. In smap-1(ycxEx1639) cells, GFP-ARF-1.2 accumulated in punctate structures (Supplementary Figure S3A-A′), suggesting that SMAP-1 acts as a GAP of ARF-1.2 in intestinal cells. Consistently, overexpression of the SMAP-1 (R60A) variant that lost GAP activity was not sufficient to alleviate cytosolic overaccumulation of ARF-1.2-GFP in smap-1(ycxEx1639) cells (Supplementary Figure S3A-A′). The membrane-to-cytosol ratio of ARF-1.2-GFP was increased by ∼70% in smap-1(RNAi) animals (Supplementary Figure S3B-B′), and overexpressed SMAP-1-mCherry reduced the puncta labeling of ARF-1.2-GFP (Supplementary Figure S3A-A′).

To verify the role of ARF-1.2 in SMAP-mediated cargo sorting, we examined the localization of ERM-1 and SLCF-1. Notably, the distribution of ERM-1-GFP and SLCF-1-GFP was affected in arf-1.2(RNAi) animals (Supplementary Figure S3C-C′). However, ARF-1.2 knockdown failed to alleviate the distributional defects of ERM-1 and SLCF-1 in SMAP-1-deficient cells (Supplementary Figure S3C-C′). Conversely, overexpression of SMAP-1 (R60A)-mCherry rescued the distributional defects of ERM-1 and SLCF-1 (Supplementary Figure S3D-D′). It is noteworthy that Arf1 GTPase has been shown to recruit AP-1 to facilitate the formation of the clathrin coat on the trans-Golgi network (TGN) (Ren et al., 2013; Beacham et al., 2019). Altogether, our results suggested that in addition to acting as a GAP of ARF-1.2 during polarized cargo sorting, SMAP-1 plays an additional role in facilitating sorting in C. elegans intestinal epithelia.

ARF-1.2 is required for the retrograde transport from Golgi to the endoplasmic reticulum (ER) (Arakel et al., 2019). To this end, we set to examine the localization of COPB-1 (COP-I complex subunit beta 1), which is expressed in the intestine (Hunt-Newbury et al., 2007; Ackema et al., 2014). As expected, COPB-1 predominantly colocalized with Golgi marker mCherry-P4M (Supplementary Figure S4B-B′). In the absence of SMAP-1, the level of colocalization between COPB-1-GFP and MC-P4M was decreased significantly (Supplementary Figure S4B-B′). Similarly, the Golgi localization of COPG-1 (COP-I complex subunit gamma 1) was reduced (Supplementary Figure S4A-A′). These results indicated that the increase in ARF-1.2 activity induced by SMAP-1 deficiency affected the Golgi recruitment of COP-I. Alternatively, SMAP-1 could directly participate in the assembly of COP-I coatomer.

Additionally, Arf1 has been reported to promote the Golgi association of gamma ear-containing Arf-binding proteins (GGAs) (Doray et al., 2002; D'Souza-Schorey and Chavrier, 2006), which cooperates with clathrin in cargo sorting. In C. elegans, APT-9 is the homolog of human GGA1. To determine whether SMAP-1 affects the localization of APT-9/GGA1, we assessed the distribution of APT-9-GFP. In the absence of SMAP-1, although APT-9-GFP accumulated in the cytosol, the localization of APT-9-GFP in mCherry-P4M-labeled Golgi apparatus was significantly reduced (Supplementary Figure S4C-D′). These results suggested that SMAP-1 also functions as a negative regulator of APT-9/GGA1 in C. elegans intestinal cells. However, the increase in ARF-1.2 activity does not seem to affect the Golgi association of APT-9/GGA1 directly.

To characterize the intracellular position of SMAP-1, we compared mCherry-tagged SMAP-1 with a set of organelle markers. In the wild-type background, SMAP-1 localized to punctate structures in intestinal cells. In agreement with its functional implication, SMAP-1 overlapped with TGN marker GOLG-4/golgin-245 (Figures 2A,B) (Munro, 2011). AMAN-2 (alpha-mannosidase II) labels cis- and medial-Golgi (Sato et al., 2011). We observed an absence of colocalization between SMAP-1 and AMAN-2 (Figures 2A,B), which were often adjacent. Phosphoinositide PI(4)P is mainly enriched in the Golgi (Dickson et al., 2016). Consistently, SMAP-1 colocalized with PI(4)P marker GFP-P4M in cytosolic punctate structures (Supplementary Figure S5A,B). SMAP-1 was also juxtaposed to exocytosis-associated endosome markers RAB-11 and RAB-8 (Figures 2A,B) (Huber et al., 1993; Ang et al., 2003; Sato et al., 2008; Winter et al., 2012).

Previous studies showed that the AP-1 complex mediates clathrin assembly and acts synergistically with clathrin to regulate sorting on the TGN (Robinson and Bonifacino, 2001). The punctate structures labeled by CHC-1 (clathrin heavy chain) were consistently reduced in the absence of AP-1 subunits (Figure 3A-A′). In addition, depleting AP-1 subunits or clathrin caused ERM-1 to accumulate around the basolateral membrane and resulted in the presence of SLCF-1 in the apical membrane of intestinal cells (Supplementary Figure S6A-A′). Remarkably, most SMAP-1-GFP-labeled structures also disappeared upon loss of AP-1 subunits (Figure 3A-A′). It is noteworthy that the C. elegans genome encodes an additional AP-1 μ1 subunit UNC-101 (Shim et al., 2000), which has been implicated in the polarized sorting of KVS-4 in DA9 motor neurons (Zhou et al., 2016). Conversely, loss of UNC-101 did not disturb the distributional of CHC-GFP (Supplementary Figure S6C-C′), supporting the distinct, tissue-specific functions of APM-1 and UNC-101 (Shim et al., 2000).

To further determine the genetic relationship between SMAP-1 and AP-1 or clathrin, we examined the distribution of clathrin and the AP-1 complex in smap-1 mutants. Of note, CHC-1-positive punctate structures decreased in SMAP-1-deficient cells (Figure 3B-B′). Large subunits APG-1 (γ subunit) and APB-1 (β1 subunit) are associated with the membrane and clathrin (Heldwein et al., 2004; Doray et al., 2007). Both GFP-APG-1 and GFP-APB-1 accumulated on the punctate structures in smap-1 mutants (Figure 3B-B′). Furthermore, we noticed that SMAP-1-mCherry colocalized with CHC-1, APB-1, and APM-1 (μ1 subunit) in intestinal cells (Figure 3C-C′; Supplementary Figure S6B-B′). To determine the interaction between SMAP-1 and AP-1 subunits APM-1 (μ1), APB-1 (β1), APG-1 (γ), and APS-1 (σ1), we performed GST pull-down assays. We found that SMAP-1 was bound to APG-1, while there was no significant interaction between SMAP-1 and APM-1, APB-1, and APS-1 (Figures 3D–G). In contrast, we did not observe the interaction between SMAP-1 and clathrin heavy chain (CHC-1) (Figure 3H). Together, these data suggested that the integrity of the AP-1 complex is required for SMAP-1 localization in TGN and that the clathrin assembly event likely occurs downstream of SMAP-1.

Next, we inspected the subcellular distribution of ERM-1-GFP and SLCF-1-GFP in SMAP-1 knockdown animals (Supplementary Figure S2B). As expected, overexpression of CHC-1 rescued the mislocalization phenotype of ERM-1 and SLCF-1 in the case of SMAP-1 deficiency (Figure 4A-B′). Conversely, the simultaneous overexpression of mCherry-tagged APM-1 (μ1), APB-1 (β1), APG-1 (γ), and APS-1 (σ1) failed to fully alleviate distribution defects of cargos in smap-1(RNAi) animals (Figure 4A-B′). Hence, our results suggested that SMAP-1 helps couple the AP-1 complex and clathrin in TGN-mediated sorting in the C. elegans intestine.

Thus far, our analysis revealed that clathrin assembly is likely to occur downstream of SMAP-1. To further clarify the effect of SMAP-1 on clathrin localization, we compared clathrin with the PI(4)P marker GFP-P4M in the absence of SMAP-1. Remarkably, loss of SMAP-1 reduced the overlap between residual CHC-1-GFP and mCherry-P4M (Figure 5A-A′), validating that SMAP-1/SMAP2 regulates the occurrence of clathrin in the Golgi apparatus. However, the Golgi localization of APM-1 was not affected upon loss of SMAP-1 (Figure 5B-B′). We subsequently examined the level of colocalization between CHC-1-GFP and mCherry-APG-1. In the absence of SMAP-1, we found that the remaining CHC-1-labeled structure no longer colocalized with APG-1 (Figure 5C-C′). In addition to biosynthetic sorting, the clathrin coat is known to mediate the formation of endocytic clathrin-coated vesicles (Chen and Schmid, 2020; Moulay et al., 2020). Also, clathrin has been reported to function as a component of the retrograde transport machinery on the surface of the endosome (Saint-Pol et al., 2004; Shi et al., 2009). Therefore, the punctate structures distinct from the P4M- or APG-1-positive puncta are likely clathrin-coated vesicles and sorting endosomes (Figure 5A-A′, C-C′). Taken together, our results indicated that SMAP-1 acts as an indispensable regulator that directs TGN clathrin coat assembly downstream of the AP-1 complex.

Genetic crosses of C. elegans were performed by standard methods (Brenner, 1974). A list of strains was provided in Supplemental Materials. RNAi-mediated gene expression interference was implemented by the feeding protocol (Timmons and Fire, 1998). RNAi constructs were from the Ahringer library (Kamath and Ahringer, 2003). For chc-1, apb-1, apg-1, apm-1, and aps-1 RNAi experiments, L2-L3 stage larvae were cultured for 48–60h and scored as adults.

Mouse anti-α-Tubulin monoclonal antibody (T6199, Sigma, St. Louis, MO), and mouse anti-Flag monoclonal antibody (F1804; Sigma, St. Louis, MO) were used in this study.

The CRISPR/Cas9 vectors were assembled by swapping the eft-3 promoter in pDD162 (Addgene, #47549) with the heat-shock promoter Phsp-16.2 (Shen et al., 2014; Li et al., 2015). CRISPR design tool (https://chopchop.cbu.uib.no/) was used to identify the knockout targets. Three smap-1 target sequences were selected, including AGG​TAA​AGT​GGA​TCC​GAA​GAa​gg, AAG​ACT​GCA​AGG​CTT​CCT​GTt​gg, AAA​AGT​GCG​CTC​AGT​GAA​TCt​gg. The CRISPR/Cas9 plasmids were validated by sequencing. CRISPR/Cas9 conditional knockout strains were created by microinjection of plasmids at 50 ng/μl and Podr-1:rfp (50 ng/μl) into wild-type hermaphrodites germline (Zhou et al., 2016). Heat-shock was executed at 0 h, 8 h, 16 h, 24 h, and 32 h after egg-hatching. The apical membrane cargo SLCF-1-GFP showed significant defects after heat shock at 0 h.

For the SMAP-1 (R60A) rescue assay, a guide RNA (sgRNA) resistant plasmid was prepared by introducing silent mutations into each target sequence (5′-AGG​CAA​GGT​TGA​CCC​AAA​AAa​ag-3′, 5′-GAG​ACT​ACA​GGG​ATT​TCT​ATt​ag-3′, 5′-GAA​GGT​ACG​TTC​TGT​TAA​CCt​ag-3′). We also introduced a single missense mutation in the SMAP-1 GAP domain. To construct transgenes expressed explicitly in C. elegans intestine, the intestine-specific promoter vha-6 driven vectors modified with a Gateway cassette were deployed. The cDNA sequences of smap-1(w09d10.1), erm-1, slcf-1, nhx-2, chc-1, apb-1, apg-1, apm-1, aps-1, golg-4, P4M, and arf-1.2 lacking a stop codon or a start codon were cloned into intestinal vectors by LR reaction (Chen et al., 2006). Transgenic strains were generated by standard microinjection; plasmids were co-injected with selection markers Podr-1:gfp or Podr-1:rfp into wild-type or smap-1(ycxEx1639) hermaphrodites germ lines.

Around 100 wild-type or smap-1(RNAi) young adults (24 h after L4 stage) were picked into 20 μl lysis buffer [100 mM Tris pH 6.8, 8% SDS, 20 mM β-mercaptoethanol], then mixed with 20 μl 2xSDS-PAGE loading buffer and boiled at 100 °C for 10min. Lysates were resolved on SDS-PAGE [12% (wt/vol) polyacrylamide], blotted to nitrocellulose. After 5% milk blocking and washing, the membrane was blotted with anti-Flag and anti-Tubulin antibodies.

Live animals were mounted on 2% agarose pads (100 mM levamisole). Fluorescence images were obtained with a Nikon C2 laser scanning confocal microscope (Nikon, Tokyo, Japan) equipped with a 100×N.A. 1.2 oil-immersion objective. Images were collected with NIS-Elements AR 4.40.00 software. Z-series of optical sections were acquired using 0.8–1 μm step size. Fluorescence data were evaluated with Metamorph software version 7.10.3.279 (Universal Imaging, West Chester, PA). The “Integrated Morphometry Analysis” component was utilized to assess the fluorescence intensity (total intensity), puncta number (structure count), and fluorescence area (total area) within unit regions. For each genotype, a total of 6 animals were analyzed in three unit regions of each intestine defined by a 100 × 100 (pixl²) box located randomly (n = 18 each). In this case, “total area” is a comprehensive parameter indicating the number and size of the fluorescent structures. Colocalization images were analyzed by Fiji (Image J) software (Schindelin et al., 2012). Pearson’s correlation coefficients were calculated with 6 animals for all genotypes.

Prism software version 8.02 (GraphPad Software, La Jolla, CA) was deployed to perform statistical analyses.