WAVE facilitates polarized E-cadherin transport

Abstract

Cadherin dynamics drive morphogenesis, while defects in cadherin polarity contribute to diseases, including cancers. However, the forces polarizing cadherin membrane distribution are not well understood. We previously showed that WAVE-dependent branched actin polarizes cadherin distribution and suggested that one mechanism is protein transport. While previous studies suggested that WAVE is enriched at various endocytic organelles, the role of WAVE in protein traffic is understudied. Here we test the model that WAVE regulates cadherin by polarizing its transport. In support of this model we show that 1) endogenously tagged WAVE accumulates in vivo at several endocytic organelles, including recycling endosomes and at the Golgi; 2) likewise, cadherin protein accumulates at recycling endosomes and the Golgi; 3) loss of WAVE components reduces cadherin accumulation at apically directed RAB-11–positive recycling endosomes and increases accumulation at the Golgi. In addition, live imaging illustrates that dynamics and velocity of recycling endosomes enriched for RAB-11::GFP and RFP::RME-1 are reduced in animals depleted of WAVE components and RAB-11::GFP movements are misdirected, suggesting that WAVE powers and directs their movements. This in vivo study demonstrates the importance of WAVE in promoting polarized transport in epithelia and supports a model that WAVE promotes cell–cell adhesion and polarity by promoting cadherin transport.

INTRODUCTION

Morphogenesis requires polarized movements of cells supported by polarized transport of adhesion molecules. Transport of adhesion molecules helps bring cells together to form new junctions and allows cells to migrate over other cells. A true understanding of morphogenesis will require identifying the signals and forces that initiate and maintain changes in adhesion. Branched actin is a major force generator that promotes cellular changes of morphogenesis, including changes in the location of adhesion molecules.

The WAVE complex is a nucleation promoting factor (NPF) that creates branched F-actin by activating the Arp2/3 complex. During Caenorhabditis elegans and Drosophila embryogenic morphogenesis, WAVE is the major NPF that turns on branch actin through Arp2/3 (Zallen et al., 2002; Patel et al., 2008), though other Arp2/3 NPFs act in C. elegans embryos, including WASP (Withee et al., 2004). WAVE is under tight regulation and must be recruited to membranes to act. Signals at the membrane activate the GTPase Rac1/CED-10, which in turn binds and activates WAVE (Bernadskaya et al., 2012).

Removal of the WAVE complex during specific stages of embryonic morphogenesis and during adult growth revealed that WAVE supports the formation and maintenance of junctions (Bernadskaya et al., 2011; Sasidharan et al., 2018; Cordova-Burgos et al., 2021). The C. elegans apical adherens junction is composed of at least two distinct complexes (reviewed in Pásti and Labouesse, 2014), the more apical cadherin/catenin complex (CCC) (Costa et al., 1998) and the more basal DLG-1/AJM-1 complex (DAC) (Bossinger et al., 2001; Firestein and Rongo, 2001; McMahon et al., 2001). Both complexes are affected by loss of the WAVE complex, as determined by live imaging with fluorescently tagged components, in embryos and adults (Bernadskaya et al., 2011). To avoid confusion we will use “apical” to refer to the lumen of the intestine and “apicolateral” to refer to the apical-most lateral region, where the adherens junctions are found (Figure 1A). During postembryonic development, transmission electron microscopy (EM) demonstrated that depletion of WAVE results in decreased apicolateral adherens junctions, suggesting that WAVE is required for adherens junction maintenance (Sasidharan et al., 2018). In postembryonic intestines, polarized apicolateral distribution of cadherin depends on WAVE. Loss of WAVE not only decreased cadherin membrane polarity; it led to lateral membrane abnormalities including membrane gaps and reduced apical PIP2 (Cordova-Burgos et al., 2021). Thus, there is a polarized and dynamic pool of cadherin throughout the life of the worm that depends on branched actin. Here we investigate cellular mechanisms that use branched actin to polarize membrane enrichment of the C. elegans cadherin protein, HMR-1 (Costa et al., 1998).

FIGURE 1:

Cadherin localization depends on endosomal regulators and the WAVE complex. (A) Cartoon of C. elegans adult, focused on the intestine, a tube of 20 cells, linked anteriorly to the two-bulbed pharynx (gray), and posteriorly to the excretory apparatus. Imaging in all figures focuses on intestinal rings 2 and 3 and the lateral junction between them. The cross-section shows that most rings have only two cells. Because worms are imaged on their side, imaging at the surface shows the basal intestine, while focusing deeper, at the lumen, shows the apical region, the lateral junctions between cells, and, at the top and bottom, the basal regions of the cells. This “apical” focus is used for most of the images shown. (B) The levels of E-cadherin/HMR-1::GFP, endogenously tagged using CRISPR (Marston et al., 2016), were measured at apicolateral (red arrows) and basolateral (blue arrows) regions in L4 (larval stage 4) controls and animals depleted with null mutations or RNAi of gex-3 and endosomal regulators. Left image: dotted yellow lines of regions measured, 35 pixels long (3 μm). All images are shown at the same exposure, except that rme-1 (b1045) shows reduced exposure to better show localization differences. Graphs report the mean intensity and the ratio of apicolateral to basolateral. The numbers of animals measured were as follows: Controls (24), gex-3 (6), rab-10 (3), rab-7(10), rab-5 (12), rme-1 (5), and rab-11 (19). Each animal was measured once at each side. Statistical analysis used one-way ANOVA, with Holm–Sidak’s multiple comparisons tests, unless stated otherwise. Asterisks here and in all figures mark statistical significance: *p < 0.05, **p < 0.001, ***p < 0.0001, ****p < 0.00001. Error bars show 95% confidence intervals in all figures. (C) Apical F-actin (green signal at top, large white arrow) was measured in intestinal ring 3 (see cartoon). Dotted white line shows lateral and basolateral regions. Graph of mean intensity, apical GFP::ACT-5. At least five animals were measured per genotype, with at most two measurements (left and right signal) per animal. Statistical analysis used one-way ANOVA, with Dunnett’s multiple comparisons test. (D) HMR-1::GFP strain from B in controls and animals with reduced gex-3, cup-5, and the gex-3(RNAi) cup-5(ar465) double, measured as in B. ns = not significant. Statistical tests for multiple comparisons as in C, but pairwise comparisons used an unpaired t test with Welch’s correction.

Branched actin is proposed to have two roles at mature apical junctions: to link the cadherin complex to actomyosin in the apical actin ring that supports polarized epithelia (Kovacs et al., 2002; Mège and Ishiyama, 2017; Indra et al., 2020), and to continually push together lateral membranes, including to repair small breaks (Jacinto et al., 2001; Efimova and Svitkina, 2018; Li et al., 2020). Here we investigate a third role for WAVE-dependent actin in maintaining epithelial junctions: supporting transport of cadherin to the correct endosomal organelles to ensure the correct localization of active cadherin molecules.

Cadherin, like most transmembrane proteins, must be trafficked during development and in mature epithelia (reviewed in Bruser and Bogdan, 2017). In vivo transport of cadherin was proposed to take place through several transport pathways, including 1) recycling from and delivery to basolateral domains, 2) apically directed flow along the lateral membrane, and 3) actin was proposed to participate (Woichansky et al., 2016). Actin force generators involved in cadherin transport are unknown.

WAVE’s role in transport is understudied for several reasons. First, tissue culture studies in HeLa or BSC-1 cells questioned whether any actin supports clathrin-mediated endocytosis in mature epithelia (Fujimoto et al., 2000; Saffarian et al., 2009) unless the cells experienced elevated membrane tension (Boulant et al., 2011; Kaur et al., 2014). Second, studies of branched actin in protein transport have mainly focused on two other Arp2/3 regulators, WASP and WASH, thought to mainly have transport roles (reviewed in Chakrabarti et al., 2021). N-WASP, but not WAVE, was shown to be enriched at clathrin-coated pits (Benesch et al., 2005). WAVE’s major role in cellular events like cell migration may have masked a significant contribution to transport.

WAVE supports protein transport, in vivo, in embryos and in adult animals. WAVE complex and WASP bind FBAR proteins, TOCA-1 and TOCA-2, and together regulate endosomal transport of yolk proteins (Giuliani et al., 2009). In the C. elegans adult intestine, WAVE, and not WASP, is involved in transport of two cargo proteins, Wls/MIG-14 and TGN-38, from recycling endosomes to the Golgi, a poorly studied retrograde recycling step (Bai and Grant, 2015). WAVE-dependent branched actin ensures the polarized distribution of the C. elegans transmembrane proteins AQP-4, which is normally enriched apically, and AQP-1, which is normally enriched basolaterally (Patel and Soto, 2013).

Using CRISPR-tagged endogenous proteins, we reported that cadherin in adult C. elegans intestines is polarized along the apicolateral membranes and highly dynamic. The dynamic and polarized distribution of cadherin depends on WAVE (Cordova-Burgos et al., 2021). In embryonic intestines, loss of WAVE or the recycling endosome regulator RME-1/EHD (Grant et al., 2001) altered cadherin accumulation and apical F-actin enrichment. Fluorescence recovery after photobleaching (FRAP) studies suggest that both proteins help to stabilize apical cadherin (Sasidharan et al., 2018). EHD proteins regulate endocytic transport especially through recycling endosomes (Chen et al., 2006; Grant and Caplan, 2008). In mice, knockout of EHD1 results in reduced E-cadherin in the developing ocular lens, but the mechanism was not examined (Arya et al., 2015).

To investigate how the essential morphogenesis adhesion molecule E-cadherin is regulated at apical membranes, we present an in vivo analysis of cadherin as a trafficking cargo in an adult epithelium. We use new and existing strains to follow cadherin transport through the adult C. elegans intestine. We visualize WAVE relative to the major endocytic organelles using new CRISPR-tagged WVE-1 strains and visualize cadherin/HMR-1 relative to the same endocytic organelles in controls and in animals depleted of WAVE complex components. Because we detected distinct changes at different recycling endosomes, we visualize the dynamics of vesicle movements at RAB-10–, RME-1-, and RAB-11–positive endosomes in the presence and absence of WAVE. These studies address the in vivo role of branched F-actin in endocytic transport in mature epithelia by focusing on the role of branched actin in dynamic transport of cadherin.

RESULTS AND DISCUSSION

Cadherin localization depends on endosomal regulators and the WAVE complex

We previously showed that cadherin is apicolaterally enriched in the adult intestine of C. elegans and that WAVE-dependent branched actin promotes this localization (Cordova-Burgos et al., 2021). To investigate whether cadherin localization is supported by WAVE-dependent transport, we first examined how the loss of different endosomal regulators altered cadherin distribution, as compared with the loss of WAVE components. Enrichment of cadherin at the apicolateral membrane of the adult intestine is easiest to image at intestinal rings 2 and 3, which lie behind the pharynx and before the start of the germline (Figure 1A). We depleted endosomal transport regulators by using genetic null alleles when possible (rab-10, rme-1, which promote early and later stages of recycling endosomes) and with RNA interference (RNAi) for genes that are not viable when null (rab-7, rab-5, and rab-11, which affect late, early, and recycling endosomes, respectively), while the WAVE complex was depleted via RNAi of one component, gex-3. Measuring cadherin levels at apicolateral and basolateral membrane regions showed that loss of late endosome regulator rab-7 resulted in reduced apicolateral cadherin, similar to loss of gex-3. Loss of regulators for early endosomes (rab-5) or recycling endosomes (rab-10, rme-1, or rab-11) instead resulted in elevated lateral cadherin (Figure 1B). Loss of gex-3 also reduced the ratio of apicolateral to basolateral cadherin, suggesting an effect on polarized membrane distribution (Figure 1B; Cordova-Burgos et al., 2021). This ratio was significantly reduced with loss of rab-5 or rab-11, while loss of all the others, with the exception of rme-1, showed a trend toward reduction (Figure 1B). These results suggested that multiple transport steps are involved in regulating apical enrichment of lateral cadherin in an adult epithelium.

F-actin is highly enriched at the apical region of the adult intestine, and this apical F-actin depends on both cadherin and WAVE, which suggests that the apicolateral junction supports the boundary to ensure apical morphology (Bernadskaya et al., 2011; Sasidharan et al., 2018; Cordova-Burgos et al., 2021). We cannot separately image the most apical F-actin (much of it at the microvilli) and apicolateral F-actin bound to cadherin, yet changes in apical F-actin reflect changes in apical/basal polarity. To test whether changes in the levels and ratio of apicolateral cadherin correlated with changes in the apical F-actin, we used the intestinal-specific actin strain gfp::act-5 (MacQueen et al., 2005) measuring the same region, intestinal ring 3 (Figure 1C). Loss of gex-3 resulted in significantly reduced apical F-actin, as expected, and as we showed with other intestinal F-actin strains (Figure 1C; Sasidharan et al., 2018; Cordova-Burgos et al., 2021). All of the endosomal transport regulators significantly reduced apical F-actin (Figure 1C, large white arrow), though some had larger effects, especially rab-11, rme-1, and rab-7. In control animals, GFP::ACT-5 is highly enriched apically, with faint expression at lateral regions (Figure 1C, arrowheads). In mutants, loss of F-actin from apical regions correlated with ectopic enrichment at lateral regions (rab-7, rab-5, rab-11) and even basal regions (rab-5 RNAi). These results suggested that all transport steps examined contribute to the overall polarity of this epithelial tissue, with some playing more significant roles.

In mutants that affect cargo sorting, the cargo can be missorted to the lysosome, which leads to increased cargo degradation and reduced steady-state levels of the cargo (reviewed in Pocha et al., 2011; Zhou et al., 2011; Bai and Grant, 2015; Cullen and Steinberg, 2018; Wang et al., 2018). If loss of GEX-3 reduces levels of cadherin by altering its transport, it may lead cadherin to be missorted to the lysosome after endocytosis, thus explaining the lower levels. We found that loss of cadherin in animals depleted of GEX-3 depends on lysosome function. Loss of gex-3 alone resulted in a significant decrease in apicolateral cadherin (Figure 1, A and D). By comparison, blocking lysosome function with a mutation in cup-5 (Fares and Greenwald, 2001a,b), the orthologue of mammalian TRPML1, did not significantly change apicolateral cadherin. A strain missing both cup-5 and gex-3 resembled the cup-5 single mutant and rescued the loss of apicolateral cadherin. This result suggested that gex-3 effects on cadherin levels depended on altered protein transport.

WVE-1 localizes to endosomal organelles and Golgi

If WAVE supports protein transport, then it should localize at specific endomembrane trafficking compartments. Previous studies with overexpressed transgenes suggested that WVE-1/WAVE is enriched at recycling endosomes (Bai and Grant, 2015). To determine the subcellular distribution and localization of endogenous WVE-1, we generated functional, rescuing CRISPR-tagged SEC-GFP-WVE-1 and mKate2::WVE-1 strains (Sasidharan et al., 2018, and this study). GFP-WVE-1 and mKate2::WVE-1 in the adult intestine are enriched along the apical domain (Figure 2A, large split arrow). In addition, puncta enriched for mKate2::WVE-1 or GFP::WVE-1 are visible throughout the adult intestine and overlap with RFP::RME-1, which is enriched at recycling endosomes (Chen et al., 2006) (Figure 2, A and B, arrowheads). GFP-WVE-1 and mKate2::WVE-1 are also enriched in rings around large autofluorescent gut granules, which are lysosome-related organelles (LROs), and in rings around nonfluorsecent structures, possibly LROs at different biogenesis stages (Delahaye et al., 2014; Morris et al., 2018) (Figure 2B, small arrows).

FIGURE 2:

WVE-1 localizes to endosomes and Golgi. (A) WVE-1 endogenously tagged via CRISPR (GFP::WVE-1 or mKate2::WVE-1) is enriched apically along the lumen (large split arrow) and also in small puncta and larger rings. GFP::WVE-1 or mKate2::WVE-1 was combined with strains with endosomal markers. Colocalization was determined using line scans in ImageJ and saved using the ROI manager, as RGB images. Applying RGB Profile Plot generated graphs with the intensity per channel. Regions were eliminated from consideration if autofluorescence (blue, lysosomal) peaks overlapped other channels (marked by an asterisk, *). Regions were scored as colocalizing for GFP and RFP or mKate2 only if the peaks overlapped, well above background, and lacked autofluorescent (AUTO.) blue signal (marked in this figure and in Figure 3 by arrowheads). (B) Endogenously CRISPR-tagged WVE-1 is enriched at small puncta (yellow arrowheads) and also at large rings (white arrows in right panels). Some rings surround autofluorescent LROs (blue signal) while others surround nonfluorescent structures. (C) In each animal, a square region, 35 × 50 μm long, was analyzed, in the region around rings 2 and 3 of the intestine (cartoon). Graph shows WVE-1 enrichment at different endosomal organelles as scatterplots, where each dot indicates an individual worm. N = 3–6 worms per genotype. (D) Colocalization of WVE-1 with various recycling endosomes (GFP::RAB-11; RFP::RME-1, RFP::RAB-10), late endosomes (RFP::RAB-7), early endosomes (RFP::RAB-5), and Golgi (AMAN-2::GFP). Smaller crops show representative regions tested for colocalization by line scans. (E) Summary of vesicle analysis shows subcellular localization of endogenous WAVE/WVE-1. These data were used for the scatterplots in C.

To quantify colocalization of WVE-1 with endosomes, we did line scans avoiding autofluorescent lysosomal regions (Figure 2A, blue signal, 405 nm illumination). The rationale for using this method to analyze colocalization is expanded in the Materials and Methods section. To identify possible WAVE-1–positive endosomes, mKate2::WVE-1 or GFP::WVE-1 was crossed into strains with markers for early endosomes (tagRFP::RAB-5), late endosomes (tagRFP::RAB-7), recycling endosomes (GFP::RAB-11, tagRFP::RAB-10, and tagRFP::RME-1), and Golgi (AMAN-2::GFP) (Treusch et al., 2004; Gleason et al., 2016) (Figure 2D). Some GFP::WVE-1 associated with RME-1– or RAB-7–positive endosomes (8 or 12%, respectively), slightly more with RAB-11– or RAB-10–positive endosomes (18 or 17%, respectively), and little or no GFP::WVE-1 associated with RAB-5–positive early endosomes. In contrast, the greatest enrichment of WVE-1::mKate2 was at Golgi marked by AMAN-2::GFP, with 42% colocalization (Figure 2, C–E). These results show that WVE-1 is enriched at recycling endosomes, as previously suggested (Bai and Grant, 2015), and at the Golgi. Studies in mammalian BSC1 cells proposed an important role for WAVE complex components GEX-2/CYFIP and GEX-3/HEM-1 at the trans-Golgi network (TGN) (Anitei et al., 2010). Thus, WVE-1 localized to specific endosomal organelles and Golgi. We next examined how loss of WAVE components affected each endosomal organelle and the Golgi.

WAVE regulates levels and spatial enrichment of endosomal and Golgi markers

Reducing the WAVE complex using RNAi depletion of gex-3 altered the endosomal organelles examined and the Golgi. RAB-11::GFP is normally highly enriched apically. Depletion of WAVE resulted in significantly lower apical enrichment and reduced intensity of puncta. The Golgi, imaged with AMAN-2::GFP, are distributed throughout the C. elegans intestine. Depleting WAVE led to brighter and larger Golgi puncta. RFP::RME-1 is normally enriched basolaterally, with some subapical puncta. Depleting WAVE reduced the overall intensity of the puncta and drastically reduced the bright basolateral puncta (Figure 3, A and C, arrowheads). RFP::RAB-10 endosomes are most enriched subapically in large puncta. Depleting WAVE strongly reduced the intensity of all RAB-10 puncta, especially the subapical large puncta, and increased the overall cytoplasmic signal (Figure 3, A and C). RFP::RAB-7 is enriched at puncta throughout the intestine. As we reported previously, depleting WAVE results in increased intensity, also visible as large punctal aggregates (Figure 3; Patel and Soto, 2013). The early endosome–enriched RFP::RAB-5 is visible throughout the cytoplasm with bright subapical puncta. Depleting WAVE led to reduced cytoplasmic signal and lower intensity of the puncta, including the subapical puncta (Figure 3, A and C). Because WVE-1 localized to a subset of these endosomal organelles and to the Golgi, we next tested whether the transport of cadherin through those organelles required WVE-1/WAVE.

FIGURE 3:

WAVE regulates levels of endosomal organelle and Golgi markers and cadherin accumulation at endosomes and Golgi. (A) Colocalization of HMR-1::GFP (green) or HMR-1::mKate2 (magenta) with endosomes enriched for distinct proteins in controls (top panels), and after depletion of the WAVE component gex-3 (bottom panels). Arrowheads mark representative areas of colocalized puncta in controls and mutants. Left panels show merged two-color images; yellow dotted box shows region enlarged in the insets. Insets are shown at higher contrast to illustrate regions of overlap. Blue signal is from intestinal autofluorescence, as explained in Figure 2. Middle and right panels are the grayscale images of HMR-1 and the endosomal organelle shown: apical recycling endosomes (RAB-11); Golgi (AMAN-2); recycling endosomes (RME-1, RAB-10), late endosomes (RAB-7), and early endosomes (RAB-5). Exposures were equally applied to controls and mutants except for the insets. All colocalizations were determined with individual line scans, as in Figure 2. Only regions significantly above background and without 405 nm autofluorescent signal (blue) were counted as colocalizing. (B) Cartoon shows the orientation and region used for the crops and the quantitation and the basal and apical regions in this tubular organ. (C) Quantitation of the mean intensity of puncta in the ring 2/3 region of the L4 larval intestine, measured using the Line tool of ImageJ. Statistical analysis used an unpaired t test with Welch’s correction. (D) Comparison of HMR-1/cadherin enrichment at different endosomal organelles, in controls and in animals depleted of WAVE component gex-3, shown as scatterplots, where each dot indicates an individual worm. N = 4–6 worms for each genotype. Statistical analysis here used one-way ANOVA with Holm–Sidak’s multiple comparisons tests. For the raw data used to generate the plots, see Supplemental Table S1.

WAVE regulates cadherin accumulation at endosomes and Golgi

To detect cadherin accumulation at specific endocytic organelles, we crossed endogenously tagged Cadherin/HMR-1, mKate2::hmr-1, and gfp::hmr-1 with the same endocytic markers used to localize WAVE/WVE-1 at endosomes. We detected minimal cadherin accumulation at RAB-5–positive endosomes (5%). In contrast, cadherin accumulates at late endosomes (RAB-7; 40%), recycling endosomes enriched for RAB-10, RME-1, and RAB-11 (20, 36, and 59%, respectively), and Golgi (51%). Thus, both cadherin and WAVE appear to accumulate most at recycling endosomes and Golgi (Figure 3, A and D and Supplemental Table S1).

We tested whether the accumulation of cadherin at each of these organelles was affected by loss of the WAVE complex through RNAi depletion of gex-3. When WAVE was depleted, cadherin accumulation shifted away from RAB-11–positive recycling endosomes (RAB-11 colocalization drops from 59 to 16%) and toward the Golgi (cadherin colocalization with AMAN-2-GFP increased from 51 to 83%). There were other, milder shifts of cadherin accumulation in other endocytic populations. In gex-3 RNAi animals, early endosomes (RAB-5) accumulated slightly more cadherin (5–12%). Changes in cadherin accumulation at endosomes enriched for RAB-10 or RAB-7 were not significant. Recycling endosomes enriched for RFP::RME-1 had less cadherin when gex-3 was depleted; however, more of these RME-1–positive endosomes colocalized with mKate2::HMR-1 (36–45%) (Figure 3, A and D).

These findings demonstrated that in the absence of WAVE, cadherin became significantly overaccumulated at the Golgi at the expense of apical RAB-11–positive recycling endosomes. There was also an increase in cadherin at basal recycling endosomes enriched for RME-1. Given that apical secretion in polarized epithelial cells can require passage through apical recycling endosomes (Ang et al., 2004), these findings could indicate that apical efflux of cadherin from the Golgi requires WAVE, so that in the absence of WAVE, failed apical-directed transport out of the Golgi results in increased accumulation within the Golgi at the expense of cadherin apical transport through apical RAB-11 recycling endosomes (Figure 3, A and D). Effects on basolateral recycling endosomes could be direct or an indirect consequence of reduced apical secretion.

WAVE and CDC-42 have similar effects on cadherin recycling to the Golgi

Transmembrane proteins and lipids often require endosome-to-Golgi transport for proper function, which occurs through a protein complex called retromer (Seaman et al., 1998; Hierro et al., 2007). However, different cargoes require transport to different endosomal organelles (Mallet and Maxfield, 1999). Previous studies found similarities in the roles of WVE-1 and CDC-42 in the transport of yolk proteins, MIG-14/Wls, and TGN-38::GFP (Giuliani et al., 2009; Bai and Grant, 2015). TGN-38::GFP was not recycled through the well-studied early endosome-to-Golgi retrograde pathway, instead traveling another retrograde route from recycling endosomes to Golgi. WAVE, together with CDC-42, was proposed to support this poorly studied retrograde transport from recycling endosomes to the Golgi (Bai and Grant, 2015). However, those studies did not examine effects on the Golgi or transport of cadherin. Therefore, we investigated whether CDC-42 has a role similar to that of WVE-1 in the transport of cadherin. Depleting cdc-42 by RNAi resulted in increased cadherin localization at RFP::RME-1 endosomes and decreased cadherin localization at GFP::RAB-11 endosomes (Figure 4, A and B), similar to WAVE depletion. Cadherin accumulation at the Golgi did not change apically but increased basally with loss of cdc-42 (Figure 4, B and C), suggesting that WAVE and CDC-42 collaborate in the sorting of cadherin from recycling endosomes to the Golgi.

FIGURE 4:

Proposed retrograde pathway uses WAVE and CDC-42 to transport cadherin from recycling endosomes to Golgi. (A) Region of intestine and orientation shown is the same as in the Figure 3B cartoon. The endosomal organelles most affected by WAVE loss were tested for effects by CDC-42 loss. Cadherin accumulation at RME-1, RAB-11, and the Golgi were compared in controls and in animals depleted of cdc-42 by RNAi. (B) Summary of results, as in Figure 3D and Supplemental Table S1. All statistical analysis in this figure used unpaired t test with Welch’s correction. (C) AMAN-2 levels at apical and basal regions in controls and in animals depleted of cdc-42 by RNAi. (D) Effect of blocking retromer, with snx-3(tm1595) mutation, on cadherin accumulation at the Golgi. Arrowheads indicate lateral membrane. Panel on the right is overexposed for GFP to show altered AMAN-2::GFP in the snx-3 mutant. (E) Effects of depleting the WAVE complex on transport of other cargoes, TGN-38:GFP and MIG-14/Wls::GFP, measured by Line Scans in ImageJ. (F) Overlap of MIG-14::GFP with the Golgi in controls and in animals depleted of gex-3.

Cadherin may utilize retromer to promote transport from early endosomes to recycling endosomes, in addition to utilizing transport from recycling endosomes to the Golgi, as is true for MIG-14/Wls (Bai and Grant, 2015). If this is the case, blocking retromer would shift accumulation to early endosomes at the expense of the Golgi. To address this possibility, we crossed a mutation in a retromer component, snx-3(tm1595) (Harterink et al., 2011), into animals that carry hmr-1::mKate2 and aman-2::gfp. Loss of snx-3 reduced cadherin at the Golgi and increased cadherin at or near the plasma membrane (Figure 4D). This result could support that cadherin utilizes retromer for early endosome to recycling endosome retrograde transport.

These results support a role for WAVE and CDC-42 in the proposed retrograde route from recycling endosome to the Golgi, provide evidence that cadherin is a cargo in this proposed route, and suggest that retromer supports cadherin transport from early endosomes to recycling endosomes (Figure 4, A–D).

The changes in cadherin seen with loss of WAVE could be due to changes in exit from the Golgi or to changes in recycling though RME-1–positive endosomes, or both. To address these possibilities, we built strains to image the overlap of RAB-11– or RME-1–positive endosomes with the Golgi, using AMAN-2::GFP and a newly integrated AMAN-2::CemOrange2 strain built from strain VK2620 (Thomas et al., 2019). If WAVE mainly regulates exit from the Golgi, then loss of gex-3 might result in increased colocalization of AMAN-2 and RAB-11. If WAVE mainly regulates recycling through RME-1 endosomes, then loss of gex-3 might result in decreased colocalization of AMAN-2 and RME-1. Instead, neither showed a significant change (unpublished data). While additional experiments are needed, these results are consistent with WAVE promoting both exit from the Golgi, through RAB-11 endosomes, and recycling of cadherin, through RME-1–positive recycling endosomes.

Other cargo requires WAVE for retrograde transport through recycling endosomes and Golgi

To test whether other cargo besides cadherin required WAVE for transport, we tested cargoes known to depend on the recycling endosome-to-Golgi pathway, TGN-38 and MIG-14 (Bai and Grant, 2015). Similar to what was previously shown, depleting gex-3 resulted in altered accumulation of TGN-38 and MIG-14, which showed significantly elevated levels (Figure 4E). Crossing MIG-14::GFP into AMAN-2::CemOrange2 suggested that elevated MIG-14 localized to the Golgi at higher levels (55–65%) although the difference was not significant (Figure 4F). Therefore, WAVE supports the transport of other cargo, besides cadherin, that transit through recycling endosomes to reach the Golgi.

Dynamics and directionality of RME-1 and RAB-11 vesicles depend on WAVE

If WAVE-dependent branched actin is promoting vesicle movements, then loss of WAVE should alter vesicle dynamics, and this change could lead to changes in the polarized distribution of endosomes. We used live imaging to follow individual endosomal tracks to measure their velocity and to track the directions of movement. We first imaged at the apical region of the intestinal epithelia in a single Z plane every 0.3, 0.5, or 2 s for 30 time points and compared control animals to animals depleted of WAVE via gex-3 RNAi. Using the ImageJ MTrackJ plug-in, we tracked the velocity and direction of vesicles until they exited the Z plane.The dramatic increase of cadherin accumulation at Golgi when we depleted the WAVE complex led us to ask whether there are dynamics changes at the Golgi. The Golgi do not show much movement in our movies. This is likely due to structural molecules that hold the Golgi in place, including perhaps microtubules (Thyberg and Moskalewski, 1999; Mascanzoni et al., 2022). When we depleted gex-3, the Golgi showed no significant difference in movement; therefore WAVE is likely not part of the mechanism that stabilizes the Golgi (Figure 5, A and B). Vesicles or tubules exiting the TGN may not be visible in our spinning-disk movies. The sizes reported for endosomes and tubules that exit the Golgi in mammalian cells, which are larger than C. elegans cells, were at or near the diffraction limit of light (reviewed in Chakrabarti et al., 2021).

FIGURE 5:

WAVE regulates vesicle dynamics and polarized movements. (A) The movements of endosomes were measured for velocity and directionality in apical and basal regions of the intestine using the ImageJ plug-in MTrackJ. Using the same strains as in Figure 3, controls and animals depleted of gex-3 via RNAi were imaged in one Z plane (either apical focus or basal focus) for 31 time points, as explained further in Materials and Methods. Numbered, colored marks indicate tracks of individual endosomes. Small white arrows show direction of the tracks over time. Large yellow arrow in bottom right panel indicates basal aggregates of GFP:RAB-11. All images shown are apical focus views of the intestine, except the bottom two of basal GFP::RAB-11, as explained in Figure 1A. (B) Graphs of velocity comparisons and tables of average velocity and mobile fraction. Movies of at least four animals were used for each genotype, with 100 to more than 500 individual movements tracked, depending on the strain. Statistical analysis used an unpaired t test. (C) Measurements of directionality toward apical or basal intestine. Graphs and polar plots show percentage of total movements toward apical or basal regions from the RFP::RME-1 and RAB-11::GFP movies, made with the apical focus view of the intestine. See Materials and Methods for details. See Supplemental Movies S2, S3, and S4 for representative movies for RFP::RAB-10, RFP::RME-1, and GFP::RAB-11. The apically oriented movements (red in the cartoon) and basally directed movements (blue in the cartoon) were plotted and statistically analyzed with a one-way ANOVA with Sidak’s multiple comparisons test.

We compared the distribution, velocity, and direction of RAB-10– and RME-1–positive endosomes in controls and in animals depleted of the WAVE component gex-3. In mammals and Drosophila, RAB-10 is considered an exocytosis protein (Sano et al., 2007; Lerner et al., 2013). In the C. elegans intestine, RAB-10 and RME-1 are proposed to regulate recycling, but they play distinct roles. While RAB-10 is required for transport for some cargoes from early endosomes to recycling endosomes, RME-1 appears to act at a later step, supporting transport from recycling endosomes back to the plasma membrane and also to the Golgi (Chen et al., 2006). While RAB-10–positive endosomes are distributed throughout the cytoplasm, with large puncta at subapical regions, RME-1–positive endosomes are most highly enriched basolaterally. RAB-10–positive vesicles move quickly and were imaged at 0.5 s time points. The movies showed increased dynamics of RAB-10–positive endosomes when gex-3 was depleted, from 0.55 to 0.88 μm/s, which was also visible as longer tracks (Figure 5, A and B; Supplemental Movie S2). Plotting the average direction of the tracks showed relatively random orientation for control RAB-10–positive vesicles, while depleting WAVE component gex-3 resulted in similarly randomly oriented movements, with no statistical differences (Figure 5A, small white arrows, and unpublished data).

Movie S2.

Download video file ^{(1.8MB, mov)}

rfp::rab-10, Control and gex-3 RNAi. Vesicles move faster in gex-3 RNAi, with no clear change in direction.

RME-1–positive endosomes moved more slowly and were imaged at 2 s time points. In contrast to RAB-10, RME-1–positive endosomes depleted of gex-3 displayed reduced velocity (from 0.31 to 0.19 μm/s) and reduced track length (Figure 5, A and B; Supplemental Movie S3). In mutant RME-1 movies, instead of a large percent of tracks moving along or toward basal regions, as happened in controls, we detected more movements of puncta distributed at apical regions (Figure 5A, white arrows). Controls showed more basally directed RME-1 tracks (54% basal, 46% apical), while animals depleted of gex-3 showed more apically directed tracks (47% basal, 53% apical), and the change was significant (Figure 5C, graphs and polar plots). Thus, changes in the dynamics (Figure 5) correlated with changes in location of endosomes (Figure 3) and were different for endosomes involved at different steps of endosome recycling, as seen in the opposite effects on RAB-10 versus RME-1 velocity.

Movie S3.

Download video file ^{(2.5MB, mov)}

rfp::rme-1, Control and gex-3 RNAi. Vesicles move slower in gex-3 RNAi, and lose basal direction.

RAB-11–positive recycling endosomes are highly dynamic and apically enriched and were imaged every 0.3 s for 10 s in a single apical Z plane. In apical controls, imaged with the apical lumen in focus, endosomes enriched for GFP::RAB-11 had an average velocity of 0.7 μm/s, while animals depleted of gex-3 showed reduced velocity, 0.54 μm/s. The direction of the vesicle tracks in controls was toward apical regions or laterally along the apical region (Figure 5A, small white arrows). In contrast, loss of gex-3 led to reduced apical GFP::RAB-11 that correlated with more vesicle movements away from the apical region (Supplemental Movie S4). Graphs and polar plots clearly demonstrate a loss of apically directed movements from 75 to 59%, a highly significant change (Figure 5C). To detect whether the mislocalized vesicles might be accumulating basally, we made movies at the basal surface. On the basal surface, the velocity of endosomes enriched for GFP::RAB-11 was higher than at the apical regions, 1.07 μm/s for controls and slower for gex-3 depleted animals (0.75 μm/s). No clear directionality was evident in movements in controls or mutants at the basal surface. However, animals depleted of gex-3 accumulated large aggregates of GFP::RAB-11 along the basal junction between two cells, something never seen in controls (Figure 5A, bottom panels, large yellow arrow). Therefore WAVE regulates the velocity and directionality of GFP::RAB-11 endosomes, ensuring the normal apical enrichment. Because RAB-11 is normally enriched apically and RME-1 is normally enriched basally (Figure 3), these results suggested that WAVE-dependent movements help keep these recycling endosomes properly distributed.

Movie S4.

Download video file ^{(2.5MB, mov)}

gfp::rab-11, Control and gex-3 RNAi. Vesicles move slower in gex-3 RNAi, and lose apical direction.

CONCLUSION

Loss of WAVE complex components contributes to cancer progression, which may be due in part to loss of the apical membrane enrichment of cadherin (Silva et al., 2009; Nath et al., 2019). However, cadherin’s apical enrichment has not been examined in terms of what forces polarize its transport. The polarized distribution of cadherin requires not just localization to the apicolateral membrane, but also mechanisms to ensure that a robust and properly assembled supply of cadherin molecules is generated and maintained through endocytic recycling and destruction of damaged molecules. In this study we establish that the adult C. elegans intestine is an excellent epithelial system in which to examine the transport of cadherin. We show that a significant population of cadherin molecules accumulates at recycling endosomes and Golgi. In addition, we demonstrate that the branched actin force generator, WAVE/Scar, plays an essential role in maintaining polarized transport of cadherin (summarized in Figure 6).

FIGURE 6:

Model for how cadherin is transported in the presence, wild type, left, or the absence, right, of WAVE-dependent branched actin. Cadherin molecules that are endocytosed from the membrane can be sorted to late endosomes and then lysosomes for degradation or returned to the Golgi for refolding and modifications and recycled back to the apical membrane. Newly synthesized cadherin exits the Golgi with the help of WVE-1–dependent branched actin. In the absence of WAVE, 1) some retrograde transport steps are disrupted, trapping more cadherin at the Golgi and at some recycling endosomes; 2) apically directed transport of cadherin is reduced; and 3) exit of newly made cadherin molecules from the Golgi is also disrupted.

Loss of WAVE disrupts apically directed transport of cadherin by significantly altering three important transport steps (Figure 3, A and D, and Figure 6). First, exit from the TGN appears to be reduced, with additional cadherin accumulating at the Golgi. This strong shift of cadherin localization supports an important role for WAVE-dependent branched actin at the Golgi. Our analysis shows that the Golgi are an important subcellular location for endogenous WAVE (Figure 2). This finding supports a previously proposed role for WAVE at the TGN (Anitei et al., 2010) and appears to be the first report of WAVE/SCAR localization to Golgi structures. Important future studies will require identifying molecules that work with WAVE at the TGN, because several pathways exist to guide cargo out of the TGN and some involve Rac1, the GTPase that activates WAVE (reviewed in Anitei and Hoflack, 2011; Anitei et al., 2017). Second, we show that RAB-11–positive recycling endosomes, which are normally enriched apically, colocalize with cadherin and depend on WAVE for their apical distribution and cadherin colocalization (Figure 3, A and D). The decrease in apical RAB-11 enrichment in WAVE mutants correlates with slower and misdirected RAB-11 endosomal movements (Figure 5; Supplemental Movie S4). To our knowledge this is the first report that WAVE-dependent branched actin supports apical enrichment of RAB-11. While we cannot image individual cadherin molecules traveling on these RAB-11 endosomes, these results support a model that WAVE-dependent branched actin promotes apical transport of cadherin on apically directed RAB-11 vesicles. In addition, these findings support a proposed recycling endosome-to-Golgi retrograde pathway that requires WAVE and CDC-42 (Figures 3 and 4; Bai and Grant, 2015) and suggest that cadherin is a cargo that is recycled through this pathway. While retromer has been shown to support transmembrane proteins required for planar polarity (Strutt et al., 2019), this is the first report that a recycling endosome-to-Golgi retrograde pathway supports cadherin transport. Owing to these three specific requirements, when WAVE is depleted, cadherin accumulation increases at the Golgi and decreases at mislocalized, slower moving RAB-11–positive and RME-1–positive recycling endosomes (Supplemental Movies S3 and S4). These roles for WAVE at the Golgi and in recycling endosomes would influence both the synthesis and the recycling of cadherin. Future studies, for example with a photoactivatable cadherin, may distinguish how much cadherin depends on WAVE for synthesis versus for recycling. However, the data that we present in Figure 4 suggests that both recycling and synthesis of cadherin are likely to depend on WAVE.

Another in vivo system has been developed for examining cadherin transport. Studies by the Riechmann lab on the Drosophila follicular epithelial cells of the female germline revealed that cadherin/DE-cad undergoes RAB-11/Rab11-dependent apical transport. Rab11 seemed to mainly transport newly synthesized cadherin and may also transport endocytosed and recycled cadherin (Woichansky et al., 2016). A more recent study by the Riechmann lab, also using the female germline, revealed a role for RAB-11, RAB-7, and MyoV in promoting apical distribution of cadherin. Thus, both in vivo systems support that RAB-11 is essential for apically directed transport, with a significant population of cadherin depending on RAB-11 for exocystosis of newly made cadherin. Tanasic and colleagues examined the distribution of F-actin relative to puncta that have DE-cad present and detected two populations of F-actin that help move DE-cad toward membranes: apical F-actin that colocalizes with DE-cad, thus likely the apical actin belt, and a basal population arranged in parallel bundles. The actin motor myosin V is associated with both of these F-actin populations, making it a likely motor to move DE-cad toward membranes (Tanasic et al., 2022). Yeast studies first showed that myosin Vs transport cargo along F-actin bundles (Pruyne et al., 1998). In the C. elegans intestine, we have not seen evidence for basal F-actin cables but expect apical actin bundles to be present (Figure 1C). C. elegans has a myosin V homologue, HUM-2 (Baker and Titus, 1997), whose expression has not been described.

If loss of WAVE reduces cadherin transport at several important steps, why is the overall effect relatively mild in this in vivo system? At least some cadherin still makes it to the apical junction, and at least some apical F-actin assembles (Figures 1, B and C, and 3A; Bernadskaya et al., 2011). We previously showed that residual apical F-actin accumulation after WAVE depletion may be due to formins, which also support cadherin localization (Bernadskaya et al., 2011). In addition, two other Arp2/3 NPFs, WASH and WASP, exist in C. elegans and may also support cadherin transport. While future studies will analyze how WASP and WASH contribute to cadherin apical enrichment and overall regulation of protein transport, the current study establishes the importance of WAVE in promoting polarized transport of cadherin. Identifying the proteins that recruit WAVE to recycling endosomes and the TGN would reveal the signals that polarize transport in epithelia by harnessing the force of WAVE-dependent branched actin.

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

C. elegans strains built for this study

OX741 pj76[mKate2::3xMyc::wve-1] generated using the SEC protocol (Marston et al., 2016), to tag wve-1 at its endogenous locus; OX886 hmr-1::gfp; rme-1(b1045); OX985 hmr-1::gfp; rab-10 (ok1494); OX823 gfp::wve-1; tagRFP::rab-5; OX830 gfp::wve-1; tagRFP::rab-10; OX824 gfp::wve-1; tagRFP::rab-7; OX818 gfp::wve-1; tagRFP::rme-1; OX870 mKate-2::wve-1; gfp::rab-11; OX999 mKate-2::wve-1; aman2::gfp; OX902 hmr-1::gfp; tagRFP::rme-1; OX810 hmr-1::gfp; tagRFP::rab-5; OX806 hmr-1::gfp; tagRFP::rab-7; OX901 hmr-1::gfp; tagRFP rab-10; OX971 hmr-1::mKate2; gfp::rab-11; OX983 hmr-1::mKate-2; aman-2::GFP; OX1017 snx-3(tm1595) hmr-1::mKate-2; aman-2::GFP; OX1019 pjIs17 [nhx-2p::aman-2::CemOrange2]; OX1023 pjIs17; vha-6p::gfp::rab-11; OX1013 rfp::rme-1; vha-6p::aman-2::gfp, OX1028 mig-14::gfp; pjIs17.

Strains used in this study

LP172 hmr-1::gfp (Marston et al., 2016); RT774 act-5::gfp (gift from Barth Grant, Rutgers Universty); OX669 pj64[gfp::3xFLAG::wve-1] (Sasidharan et al., 2018); tagRFP::rab-5; tagRFP::rab-10; tagRFP::rab-7; tagRFP::rme-1 (Gleason et al., 2016); RT311 vha-6p::gfp::rab-11 (Chen et al., 2006); GS2813 cup-5(ar465) (Fares and Greenwald, 2001a,b); snx-3(tm1595) (Harterink et al., 2011); OX970 Pvha-6::PH::gfp; hmr-1::mKate2 (Cordova-Burgos, 2021), OX974 hmr-1::mKate2; gfp::3xFLAG-gex-3 (Cordova-Burgos, 2021); VK2620 nhx-2p::aman-2:: Cem-Orange2 (Thomas et al., 2019).

RNAi experiments

All RNAi bacterial strains used in this study were administered by the feeding protocol in Sasidharan et al. (2018). RNAi feeding experiments were done at 23°C unless otherwise mentioned. Worms were synchronized and transferred onto seeded plates containing RNAi-expressing bacteria. To monitor the effectiveness of the RNAi, we used two methods. We counted the percent dead embryos, which after 2 d is expected to be >90% for gex-3. We also monitored postembryonic silencing of a gfp-tagged strain in the intestine, such as gfp::gex-3 or gfp::rab-11. All RNAi treatments were done for 2 d.

Live imaging

Imaging was done in a temperature-controlled room set to 23°C on a laser spinning-disk confocal microscope with a Yokogawa scan head or on a Zeiss AxioImager Z1 Microscope using Plan-Apo 63×/1.4NA or Plan-Apo 40×/1.3NA oil lenses. Images were captured on a Hamamatsu CMOS Camera using MetaMorph software and analyzed using ImageJ. Controls and mutants were imaged within 3 d of each other with the same imaging conditions. All measurements were performed on raw data using ImageJ. For fluorescence measurements, background intensity was subtracted by using a box or line of the same size and measuring average intensity in the same focal plane, near the animal.

Microscopy of L4s and adults

Young adult-stage animals (1 d after the L4 stage) and L4 larvae were placed on 10% agar pads in M9 solution and immobilized using 2 μl of levamisole (100 μM) salts and covered with 1.5 µm coverslips. Images were taken within 15 min of making the pads. Imaging was done on the Zeiss AxioImager Z1 with a Yokogawa CSUX1-5000 spinning disk using the Plan Apo 63×/1.4NA Oil lens.

Quantitation of immunofluorescence

Quantitation of live fluorescence was performed using the line selection and the dynamic profile function of ImageJ to measure fluorescence along lines of equal length along, for example, the apical intestine or lateral regions of the intestinal cells. For all experiments shown, the images were captured at the same exposure settings for wild type and mutants. All quantitation was done on the raw images. The figure legends indicate when images were enhanced for contrast, and the same enhancement was applied to a mosaic of the related images for that experiment. Each measurement was taken following the subtraction of background fluorescence.

Rationale for colocalization method used

Automated colocalization methods are suspect when applied in our system (and possibly other systems) because they all require setting artificial thresholds. In our system, the presence of the gut granules creates large regions of autofluorescence. We compared the results from automated systems versus the much more laborious manual line scans. Manual line scans allowed us to choose regions with the lowest autofluorescence and manually choose specific regions, the same region per worm, for line scans. We chose the more difficult manual measurements so that we can stand by our data to reflect what is actually happening in these regions. Only values significantly over the background autofluorescence were used for the analysis.

Colocalization analysis ( Figures 2 and 3)

Measurements avoiding autofluorescence from lysosomes in the blue channel were made with the line scan tool in ImageJ, and colocalization was verified using the RGB Plot Profile plug-in as illustrated in Figure 2A. Overlapping red and green peaks with a low blue background were counted as positive sites of colocalization.

Vesicle dynamics

Movies of gfp::rab-11, rfp::rme-1, rfp::rab-10, or aman-2::gfp were made at either one apical or one basal Z plane of the intestine, int 2/3 region, every 0.3, 2, 0.5, and 2 s, respectively, for a total of 31 time points, creating movies ∼10–62 s long. The ImageJ plug-in MTrackJ was used to follow the location of individual endosomes through several time points until they moved out of the Z plane. MTrackJ uses the resulting tracks to measure the velocity of tracks and points. Once the average speed is calculated, one can estimate the mobile fraction above a cutoff velocity based on the average velocity. Data including alpha values from MTrackJ were used to generate polar plots using RStudio, and the RStudio plots were redrawn in Adobe Illustrator to enable better labeling. Because these are live animals, we used only animals with minimal movements during the time to be measured.

Statistical analysis

For grouped data in Figure 1, statistical significance was established by performing a one-way analysis of variance (ANOVA), the Brown–Forysythe and Welch ANOVA, followed by a Dunnett’s multiple comparisons T3 posttest. For ungrouped data in other figures, an unpaired t test, the unequal variance (Welch) t test, was used. Error bars show 95% confidence intervals. Asterisks (*) denote p values *p < 0.05, **p < 0.001, ***p < 0.0001, ****p < 0.00001. All statistical analysis was performed using GraphPad Prism 8.

Abbreviations used:

CCC

cadherin/catenin complex

DAC

DLG-1/AJM-1 complex

EM

electron microscopy

NPF

nucleation promoting factor