Terminal web and vesicle trafficking proteins mediate nematode single-cell tubulogenesis

The interactions of cell cytoskeleton with vesicle trafficking are crucial to provide cells with the shape needed to perform specialized functions. For example, the microtubule and actin filament structure mediates transport of cargo to form and maintain the nerve terminal at the end of an axon (Hakanen et al., 2019; Kiral et al., 2018), Rab proteins regulate the apical specialized functions of multiple cell types in intestinal villi (Zhang and Gao, 2016), and defects affecting endosomal trafficking (in neurons and glia) underlie several neural diseases such as Charcot-Marie-Tooth disease and Niemann-Pick disease (Neefjes and van der Kant, 2014). Advances in understanding these interactions have arisen from studies in a wide range of tissues and organisms, from yeast (Chiou et al., 2017; Pires and Boxem, 2018) to Caenorhabditis elegans (Sato et al., 2014), Drosophila (Mack and Georgiou, 2014; Nemetschke and Knust, 2016), and mammalian cells such as MDCK cells in culture (Román-Fernández et al., 2018). An interesting specialized case is the formation and extension of narrow tubular structures, which require both the cytoskeletal extension to reach out to target tissues and the curvature of the apical surface to create a patent tubule (Barry et al., 2015; Sigurbjörnsdóttir et al., 2014; Sundaram and Cohen, 2017).

The excretory canal cell of C. elegans offers a tractable genetic model for studying single-celled tubular morphogenesis in a “seamless cell” without intracellular junctions to hold the tube together along its length (Fig. 1 a; Sundaram and Buechner, 2016; Sundaram and Cohen, 2017) This cell sends out long hollow projections that reach the length of the animal to collect and eliminate excess liquid (Falin et al., 2009; Igual Gil et al., 2017; Nelson et al., 1983; Sundaram and Buechner, 2016). While the excretory system as a whole is required for organismal survival, a large number of viable mutants have been found that impair the ability of the excretory canals to form narrow tubular extensions (Al-Hashimi et al., 2019; Buechner et al., 1999; Igual Gil et al., 2017); cloning of the underlying genes has revealed a wide variety of cytoskeletal proteins involved, including ezrin-radixin-moesin (ERM) and β_H-spectrin anchors of the actin cytoskeleton (Khan et al., 2013; Praitis et al., 2005), a diaphanous-related formin that links actin to the microtubule cytoskeleton (Shaye and Greenwald, 2015), and three intermediate filament proteins that wrap around the apical (luminal) surface to form a terminal web similar to that surrounding intestinal tubes (Al-Hashimi et al., 2018; Geisler et al., 2020; Karabinos, 2019; Khan et al., 2019; Kolotuev et al., 2013).

In addition, vesicle trafficking is important for canal morphogenesis. Mutants impairing function of the exocyst (Armenti et al., 2014) show profound defects in canal formation, as do mutations affecting the CCM-3/STRIPAK machinery that maintain normal mammalian brain capillary morphology (Lant et al., 2015). Finally, three proteins affecting vesicle transport act in a genetic pathway to regulate canal morphology: EXC-9/CRIP1, EXC-1/IRG, and EXC-5/FGD (Grussendorf et al., 2016; Mattingly and Buechner, 2011; Tong and Buechner, 2008). All of the three encoded proteins have close human homologues: CRIP1 increases metastatic transformation and invasiveness in cell culture (Cousins and Lanningham-Foster, 2000; He et al., 2017; Li et al., 2017; Zhang et al., 2018); IRGM is involved with autophagy and pathogenic responses (Howard et al., 2011; Kumar et al., 2018; Pilla-Moffett et al., 2016); and FGD genes are loci of the developmental diseases Aarskog-Scott syndrome and Charcot-Marie-Tooth disease type 4H (Delague et al., 2007; Gao et al., 2001; Stendel et al., 2007). Mutants in these nematode genes show similar cystic defects in the structure of the canal apical surface as well as defects in vesicle transport: buildup of RAB-5–labeled vesicles and loss of RME-1–labeled vesicles in areas of cystic dilation (Grussendorf et al., 2016; Mattingly and Buechner, 2011). In addition, overexpression of any of these three genes results in a normal-diameter lumen (apical surface) folded up within a severely shortened canal. We interpreted these results as supporting a model in which EXC-9, EXC-1, and EXC-5 act in sequence to promote vesicle trafficking to the apical surface of the excretory cell, allowing that surface to be remodeled as the canals extend and widen during animal growth. It remains unknown, however, how EXC-9, EXC-1, and EXC-5 affect vesicle trafficking and whether cytoskeletal elements are involved with this process.

Here, we show that one of the three canal intermediate filament proteins, EXC-2, retains EXC-9/CRIP to the canal apical surface, mediated by the C-terminal domain (CTD) of EXC-2. The results also show that the EXC-1/IRG protein is retained at the apical surface of the canals independent of its binding partner EXC-9/CRIP and of intermediate filament EXC-2. Finally, overexpression of the vesicle trafficking protein RAB-8 partially suppresses the effects of mutations in exc-9 and exc-1 (but not mutations in exc-5) on the development of narrow canal tubules. These results delineate a novel pathway from the intermediate filament matrix to vesicle trafficking proteins that regulate the formation of the luminal diameter in this highly regulated polarized tubule cell.

EXC-9 (Human CRIP1 orthologue) is a small protein required for development and maintenance of the excretory canals; mutants develop shortened canals with large fluid-filled cysts (Tong and Buechner, 2008). The 85 amino acids of EXC-9 encode a single conserved LIM domain of the cysteine-rich protein class (Davis et al., 1998; Smith et al., 2010) at the N-terminus (60 amino acids; Fig. 1 b) followed by a short tail of unknown function conserved from nematodes to humans. In the exc-9(n2669) mutant, cystic canals always form a medium-sized (1/4–1/2 the diameter of the animal) cyst at the distal ends of the lumen, though cyst size and canal length vary depending on the allele used (Fig. 1, c and d). Allele n2669 is a strong allele that was previously studied (Tong and Buechner, 2008). Allele qp130, a putative null CRISPR/Cas9–induced deletion of the first eight nucleotides of the coding region (the next ATG is not in-frame), produces significantly shorter canals than seen in n2669 homozygotes, while insertion of GFP at the C-terminus of the protein (see below) creates a milder mutant phenotype, with smaller cysts and longer canals.

To determine the subcellular location of EXC-9, the CRISPR/Cas9 system was used to insert the gfp coding region at either the 5′- or -3′ terminus of exc-9 (Dickinson et al., 2015). Expression of N-terminal GFP::EXC-9 (BK583) allowed the protein to remain fully functional and form long, thin, hollow canals, with the protein located predominantly at the apical surface (Fig. 1 e). Expression was also seen at lower levels within the uterine seam cell, both distal tip cells, and the lumbar ganglion (Fig. S1). Within the canals, expression was largely coincident with the intermediate filament protein marker mKate2::IFA-4, which labels the terminal web surrounding the canal lumen (Al-Hashimi et al., 2018), though some EXC-9 was also found within the cell cytoplasm. As noted above, when GFP was inserted at the C-terminus (BK589), the function of EXC-9 was compromised (Fig. 1, d and f). However, EXC-9::GFP was still enriched at the apical surface. Blockage of the C-terminal tail by GFP impaired EXC-9 function without affecting apical localization.

As EXC-9 has no transmembrane domain but is enriched at the apical membrane, we examined how EXC-9 is retained at the membrane by crossing our fully functional GFP::EXC-9 strain to mutants defective in known canal apical lumen cytoskeletal proteins (Fig. 1, g–j). In the absence of the large intermediate filament protein EXC-2 (Al-Hashimi et al., 2018), GFP::EXC-9 lost apical localization and became evenly distributed throughout the canal cytoplasm (Fig. 1 g). In contrast, loss of another canal-specific intermediate filament, IFA-4 (Flibotte et al., 2010), which likely forms heterodimers with EXC-2, did not affect EXC-9’s ability to be retained at the apical surface (Fig. 1 h). Similarly, loss of β_H spectrin SMA-1 causes formation of large irregular cysts throughout the length of the canals (Buechner et al., 1999) but retains strong expression of EXC-9 predominantly at the apical surface (Fig. 1 i). Finally, loss of the IRG GTPase protein EXC-1, a known EXC-9 binding partner (Grussendorf et al., 2016), also had no effect on the apical retention of EXC-9 (Fig. 1 j). The results indicate that EXC-2 retains the majority of EXC-9 protein at the apical surface; this was confirmed by coexpression of functional labeled EXC-2 and EXC-9, which were completely collocated at the canal apical surface (Fig. 1 k).

Intermediate filaments usually consist of a short N-terminal domain (NTD) and a longer CTD flanking a conserved filament domain used for dimerization (Herrmann and Aebi, 2016). While IFA-4 and IFB-1 follow this plan (including a conserved lamin tail domain [LTD] within the CTD; Carberry et al., 2009; Zuela and Gruenbaum, 2016), EXC-2 has a much longer NTD and no lamin tail homology in its CTD (Fig. 2 a). There is also a gene termed ifc-2 (Hüsken et al., 2008) that begins from a promoter in an exon of exc-2 and that contains only the filament domain and CTD, but isoforms of IFC-2 are found predominantly in the intestine and are not expressed in the canal cell (Geisler et al., 2020).

To determine which section of EXC-2 is required for binding EXC-9 to the luminal surface, the long NTD (amino acids 1–767) and short CTD (amino acids 1112–1339) were each cloned into a fluorescence translational construct for overexpression in wild-type and in the exc-2(qp110) null mutant background. In a previous study, overexpression of full-length exc-2 PCR product greatly shortened the canals to less than half-length and caused small dilations of the lumen (lumen diameter <1/4 body width; Al-Hashimi et al., 2018). Here, overexpression of each of these partial exc-2 constructs from the strong canal-specific vha-1 promoter caused similarly shortened canals with small dilations (Fig. 2, b and c). In the wild-type background, overexpressed EXC-2NTD::GFP remained apically localized (Fig. 2 b), but labeled EXC-9 did not. The presence of excess EXC-2NTD prevented EXC-9 from being retained at the luminal membrane. This implies that EXC-2NTD and EXC-9 compete for the same locations on the luminal membrane.

Overexpression of EXC-2CTD::GFP exhibited a unique polarized fluorescence phenotype (Fig. 2 c): canals were generally wild-type in diameter along their length, but did not grow to full length and terminated in medium-sized cysts at the distal ends. In addition, fluorescently labeled EXC-2CTD and labeled EXC-9 were fully collocated but strongly enriched toward the distal end, with very bright fluorescence at the distal tip of the canal lumen. We inferred that the CTD of EXC-2 recruits EXC-9 and that they are able to remain at high concentration at the canal apical surface only at the luminal tip of the canals.

In the exc-2(qp110) null allele mutant background, EXC-2NTD::GFP was still localized at the apical surface (Fig. 2 d), and its expression pattern resembled the localization and meshwork pattern of wild-type GFP::EXC-2 expressed in animals lacking one of the other intermediate filament proteins, IFB-1 or IFA-4 (Al-Hashimi et al., 2018). This indicated that the N-terminus of EXC-2NTD is capable of binding either to the other intermediate filament proteins IFB-1 and IFA-4 or to other proteins anchored or bound to the luminal surface.

GFP attachment to either end of the small EXC-9 protein did not affect its expression at the apical surface of the excretory canals, although C-terminal GFP impaired EXC-9 function (Fig. 1 d). We investigated the role of the LIM domain at the N-terminus of EXC-9 by using CRISPR/Cas9 to make endogenous mutations in key conserved cysteines (Fig. 2 e). LIM domains are composed of two contiguous zinc fingers containing three conserved cysteine residues and one histidine in the first zinc finger and four conserved cysteines in the second zinc finger (Weiskirchen and Günther, 2003). We used CRISPR/Cas9 to replace the last cysteine in the first zinc finger and first two cysteines in the second zinc finger with alanine (Fig. 2 e). With these changes, about half of the animals (28 of 50) were mildly affected in regard to canal length, with canals reaching longer than half the length of the organism, while the other animals had full-length canals (81% of full-length average; Fig. 2 e’’’). In the cysteine-substituted mutant canals, apical retention of labeled EXC-9 was variably impaired, with some sections showing wild-type levels of EXC-9 at the apical surface compared with the cytoplasm, while in others, almost no EXC-9 was retained at the apical surface. (Fig. 2 e’) Loss of apical EXC-9 was coincident with areas of wider lumen, possibly nascent cysts. These results suggest that LIM domain function plays a role in apical localization of EXC-9 to the canal apical surface.

EXC-9 binds to the N-terminal half of EXC-1 and acts genetically upstream of EXC-1 to mediate endosomal vesicle transport within the canals (Grussendorf et al., 2016). EXC-1 possesses two RAS-like GTPase domains and is the sole nematode homologue of mammalian IRG proteins (Fig. 3 a; Grussendorf et al., 2016). Since EXC-9 is largely collocated with EXC-2, we examined whether the EXC-9–binding partner EXC-1 is also retained near EXC-2 within the excretory cell. Through use of CRISPR/Cas9, GFP was inserted at the C-terminus of EXC-1. This labeled protein was found at the apical surface, and surprisingly, unlike EXC-9, little EXC-1 protein was apparent in the cytoplasm (Fig. 3 b). In addition to the excretory canal cell, EXC-1 was expressed at lower levels in the amphid sheath cells (Fig. S2). Localization of EXC-1 did not require activity of EXC-9, since in the deletion mutant exc-9(qp130), EXC-1 fully retained its localization to the apical luminal membrane (Fig. 3 c). Similarly, crossing the labeled EXC-1 strain into a null exc-2(qp110) mutant strain caused no change to the apical localization of EXC-1 (Fig. 3 d). EXC-1 appears to be anchored to the membrane independently of the EXC proteins that act genetically upstream of EXC-1.

The β_H–spectrin SMA-1 is another important cytoskeletal element regulating canal shape (Buechner et al., 1999), but it does not affect the placement of EXC-9 (Fig. 1 i). β_H-spectrin and the ERM homologue ERM-1 anchor actin filaments to the apical surface (Khan et al., 2019, 2013; Praitis et al., 2005). In erm-1 knockdown animals, labeled IFB-1 showed the intermediate filament network to be greatly disrupted (Khan et al., 2019). To investigate the ability of SMA-1 to affect placement of intermediate filaments, we examined the other two canal intermediate filaments via separately crossing the fluorescent mKate2::ifa-4 and gfp::exc-2 strains into animals carrying a null mutation either in sma-1 or in exc-4 (encoding a chloride intercellular channel ion channel, as a control mutant with large cysts), both of which also severely disrupt canal morphology (Berry et al., 2003; Praitis et al., 2005). In the homozygous sma-1(ru18) mutant background, both mKate2::IFA-4 and GFP::EXC-2 fluorescence showed disruption to the spacing of filaments surrounding the apical surface (Fig. 4, b and d). The “weave” of the intermediate filament meshwork was looser, allowing the appearance of large gaps of open membrane between intermediate filaments, as seen by the repetitive distance between peaks of fluorescence (Fig. 4, b’ and d’). In contrast, in the exc-4 background (Fig. 4, c and e), the meshwork of both labeled filaments generally showed a tighter weave, with filaments aligned closely to each other, comparable to the meshwork of wild-type animals (Fig. 4, a and a’’), even though exc-4 animals exhibit very large canal cysts comparable in size to those of sma-1 mutants (Buechner et al., 1999). Occasional filament breakdowns in very large cysts could be seen in exc-4 mutant animals, but the background filament weave was generally as tight as in wild-type animals. The disruption of filament weave in sma-1 mutants is similar to that caused by knockdown of ERM-1 (Khan et al., 2019). We conclude that both ERM-1 and SMA-1 are needed for proper layout of the intermediate filament network.

Examination of sma-1 and exc-4 mutants by means of electron microscopy (Fig. 5) showed further differences between these animals. In the exc-4(rh133) mutant, the luminal surface electron-dense terminal web was intact and associated tightly with the membrane at the luminal surface (Fig. 5 c), as is seen in wild-type animals (Fig. 5 a). The membrane itself was in contact with long chains of small canalicular vesicles (Fig. 5, a and c, marked by asterisks), rich in vacuolar ATPase and aquaporin, and which pump water into the canal lumen for expulsion from the animal (Sundaram and Buechner, 2016). In contrast, the sma-1(e30) mutant (Fig. 5 d) exhibited an electron-dense terminal web layer that was no longer juxtaposed to the apical membrane. In addition, the terminal web was broken in places, with large gaps between electron-dense areas (arrowheads in Fig. 5 d’). Although canalicular vesicles were numerous in this mutant, the canalicular membrane appeared disorganized in areas where the terminal web was distant from the membrane. Finally, in the exc-2(rh90) mutant missing the largest intermediate filament protein, the terminal web was largely missing from some areas of the apical membrane, and very few canalicular vesicles were visible, as seen in Fig. 5 b, although some areas of the canal in exc-2(rh90) and exc-2(rh209) animals retained some terminal web together with canaliculi attached, as reported previously (Al-Hashimi et al., 2018; Buechner et al., 1999). These results confirm that intermediate filaments make up the terminal web of the excretory canal, similar to terminal webs in the gut (Coch and Leube, 2016), and that β_H-spectrin is required for proper arrangement of these intermediate filaments. In addition, the terminal web appears to be necessary for stabilization of the network of canalicular vesicles that are presumed to carry out osmoregulation of the animal in hypotonic environments.

Previous data showed that EXC-9, EXC-1, and EXC-5 act in sequence to regulate endosomal vesicle transport in the excretory canals (Grussendorf et al., 2016; Mattingly and Buechner, 2011; Tong and Buechner, 2008). To investigate the potential cargo or cytoskeletal protein being regulated, we used a strain containing two integrated constructs that both use a canal-specific promoter to express either a cytoplasmic CFP or the actin dye LifeAct (Shaye and Greenwald, 2015). To better visualize the full length of actin filaments along the curved apical surface of the canals, a series of Z-axis pictures were taken and merged into a single micrograph (Fig. 6 a). In the wild-type canal background (Fig. 6 b), the apical membrane actin filaments appear as a uniform field of short actin filaments throughout the length of the canal apical surface. This structure is maintained in the sma-1 mutant (Fig. 6 c), where the canal lumen is often enlarged to a similar degree throughout the canal without obvious septations (Buechner, 1999); this result suggests that the labeled actin filaments are anchored to the membrane by another means. In the exc-9 mutant background, some aggregation of actin filaments into thick bundles (Fig. 6 d, arrow) became evident in areas with larger septate cysts, corresponding to septate cysts that thin the amount of cytoplasm present. This effect was progressively exacerbated in exc-1 mutants (Fig. 6 e) and most dramatically in the large cysts of exc-5 mutants, where the actin filaments contained little actin except in thick bundles (Fig. 6 f). In addition, in the exc-5 mutant, these thick bundles were collocated with accumulations of cytoplasmic material around the swollen luminal cyst. We conclude that these three exc mutants therefore are needed to stabilize a uniform actin meshwork at the apical surface.

In exc-1, exc-5, and exc-9 mutants, endosomal expression of the basal trafficking marker RME-1 was significantly decreased within the cytoplasm of canal cysts, while that of other vesicular markers, especially the early endosome antigen EEA-1 and (to a lesser extent) the apical trafficking marker RAB-11.1, were increased (Grussendorf et al., 2016; Mattingly and Buechner, 2011). To better understand the relationship of these EXC proteins on traffic movement within the excretory canal, we first examined competition between apical and basal traffic markers (Fig. 7, a–c). A strain carrying a stable integrant of mCherry::rme-1 driven by a canal-specific promoter (Mattingly and Buechner, 2011) exhibited punctate endosomes and strong cytoplasmic expression (Fig. 7 a). We crossed this marker line to a strain harboring an integration of a construct that strongly expresses canal-specific gfp::rab-11.1 (Fig. 7 b). With the expression of rab-11.1, expression of mCherry::RME-1 in the canal was greatly reduced to near-background levels (Fig. 7 c). We conclude that apical and basal transport within the excretory canal cell depends on the ratio of apical-directed and basal-directed transport machinery present.

Next, since expression of Rab proteins within the canal appears to affect transport ratios, as in the intestine (Sato et al., 2014), and exc-9, exc-1, and exc-5 mutants impair vesicle trafficking and maintenance of the apical surface (Grussendorf et al., 2016; Mattingly and Buechner, 2011; Tong and Buechner, 2008), we examined whether overexpression of an apical-directed Rab protein could rescue mutation of the exc-1, exc-5, and exc-9 genes (Fig. 7, d and e). A construct overexpressing canal-specific rab-8 was injected into exc-1 (rh26), exc-5 (rh232), and exc-9(qp130) mutants (Fig. 7, d–f). Cyst formation in these mutants causes significant shortening of the canals, but in the presence of excess RAB-8, the canals of exc-1 and exc-9 mutants were significantly longer (Fig. 7 d). Similarly, overexpression of any of these exc genes caused the canals to have a “convoluted” phenotype, where the basal surface was shortened, while the normal-diameter luminal surface was still extended and ended up coiled within the canal (Tong and Buechner, 2008; Fig. 7 e). Canal overexpression of rab-8 also caused a significant increase in convoluted canals in exc-1 and exc-9 mutants (though not as strong as overexpression of the exc proteins themselves; Grussendorf et al., 2016; Fig. 7 f). Surprisingly, overexpression of rab-8 had no significant effect on the exc-5 mutant, either in canal length or formation of cystic canals (Fig. 7, d and f). These genetic epistasis experiments suggest that RAB-8 acts downstream of EXC-9 and EXC-1 in directing cargo to the apical surface.

The cytoskeleton forms and reinforces the shape of cells. Projections of actin and microtubules underlie cell protrusions, while intermediate filaments act as ropes or stretchable bungee cords to allow limited stretch to the plasma membrane and nuclei (Delacour et al., 2016; Gerace and Burke, 1988; Gruenbaum and Foisner, 2015). Both of these elements are needed for formation of long hollow tubules, which must be guided to grow out along the length of the body wall, but whose luminal diameter must be restrained as excess water is brought into the tubule to be expelled from the organism (Sundaram and Buechner, 2016; Sundaram and Cohen, 2017). The results in this paper show that EXC-1/IRG GTPase and EXC-9/CRIP1 are both held predominantly near the apical surface, either in the plasma membrane itself or together with the EXC-2/IFA-4/IFB-1 intermediate filaments of the terminal web surrounding the lumen. EXC-1 is located at the apical surface even in exc-2 mutants where there is little or no terminal web (Fig. 3). Upon reexamining the amino acid sequence of EXC-1 with the help of predictive lipidation tools, we found that its N-terminus contains a cotranslational myristoylation sequence MGHKTS (Johnson et al., 1994; Udenwobele et al., 2017; Xie et al., 2016). Our attempts to engineer a CRISPR/Cas9–mediated insertion of GFP at the N-terminus of this protein were unsuccessful, which may indicate that this sequence is essential for organismal viability. Future biochemical studies of possible lipidation of this protein should be fruitful for further understanding its interactions with cytoskeletal elements.

EXC-9 is a small protein containing only a single LIM domain and a short tail and was expected to be fully cytoplasmic. It was highly surprising to find that this protein was instead found to be predominantly concentrated at the terminal web at the apical surface, though with a visible fraction in the cytoplasm (Fig. 1). Also surprising was the finding that, of the three intermediate filament proteins within the excretory canal, EXC-2 is specifically needed to retain EXC-9 at the terminal web (Fig. 1). These results suggest that EXC-9 is either present in the cytoplasm before being bound to the terminal web, or it can be released into the cytoplasm from the terminal web. Impairment of the EXC-9 LIM domain (Fig. 2) reduced the ability of EXC-9 to bind to the terminal web, with regions of lower terminal web fluorescence associated with lumen widening, possibly indicative of formation of incipient cysts. These observations reinforce the idea that EXC-9 binding to the canal terminal web is important for maintaining the narrow diameter of the excretory canals. The effect of mutation of the LIM domain was surprisingly small, however, compared with that of null mutants of exc-9. It is possible that the C-terminal domain of EXC-9 could be most critical for the function of this protein or instead that the mutated LIM domain retained some residual binding ability.

Overexpressed labeled EXC-2 (intermediate filament) NTD appeared along the entire canal length and largely prevented EXC-9 from binding to the terminal web, resulting in EXC-9 being essentially all cytoplasmic. In contrast, overexpression of the EXC-2 CTD lowered the amount of free EXC-9 visible in the cytoplasm and concentrated this protein at the terminal web toward the cystic posterior end and especially at the canal tip (Fig. 2). The canal luminal surface extends the length of the animal, closely following the growing tip of the basal surface during embryogenesis and the first larval stage (Fujita et al., 2003). For overexpressed EXC-2 CTD and EXC-9 to label only the canal luminal tip and cystic areas, CTD must accumulate there until a step occurs to restrain dilation of the membrane, presumably via vesicle trafficking and construction of the actin cytoskeleton and terminal web.

There are 11 cytosolic intermediate filaments in C. elegans (Carberry et al., 2009; Dodemont et al., 1994; Karabinos et al., 2001; Zuela and Gruenbaum, 2016). Several studies have focused on the intestinal terminal web, where intermediate filaments are also necessary to maintain narrow luminal diameter (Coch and Leube, 2016; Geisler et al., 2020; Karabinos, 2019; Karabinos et al., 2017; Khan et al., 2013, 2019). While exc-9 and exc-1 are not expressed within the intestine, other LIM-domain proteins could be interacting with intermediate filaments in this tissue. EXC-1, EXC-9, and its homologue VALV-1 are expressed in a range of tissues that undergo stretching and bending movements, including the intestinal-pharyngeal valve, spermathecal valves, and amphid glia (Grussendorf et al., 2016; Tong and Buechner, 2008).

Mutants in sma-1 (encoding β_H spectrin) show disorganization of the terminal web (Fig. 4 and Fig. 5; Buechner et al., 1999; Praitis et al., 2005), which indicates that this spectrin has an organizing effect on the intermediate filament network, though the sma-1 mutation does not abrogate the terminal web’s overall formation near the apical membrane. Significantly, sma-1 mutants tend to exhibit very wide canals with few narrow septations, with numerous canaliculi attached to the apical membrane (Fig. 5 d), unlike mutants of intermediate filament genes (Fig. 5 b; Al-Hashimi et al., 2018; Khan et al., 2019; Kolotuev et al., 2013). The presence of intermediate filaments at the apical membrane therefore appears to be necessary for attachment of these canaliculi, as much as the activity of the exocyst and STRIPAK complex is (Armenti et al., 2014; Lant et al., 2015).

Loss of EXC-9, EXC-1, or EXC-5 function resulted in ACT-5 actin filaments appearing as thick cables restraining the lumen, which swelled into cysts around these cables (Fig. 6). Surprisingly, sma-1 mutants lacking β_H-spectrin still maintained a tightly knit meshwork of actin around the canal lumen, which suggests that unlike the FERM-domain protein ERM-1 (Khan et al., 2013, 2019), this spectrin is not the most critical organizer of these actin filaments. This result was surprising, since SMA-1 regulates actomyosin cytoskeletal contractility in several C. elegans tissues (Wirshing and Cram, 2018). Disorganization of the terminal web could therefore be secondary to SMA-1’s effects on cytoskeletal actin, or perhaps actomyosin contractility may not play a significant role in organizing the canal actin meshwork.

EXC-9 acts genetically upstream of and binds to EXC-1, which itself acts upstream of EXC-5. Results here show that overexpressing RAB-8 partially rescues the exc-9 and exc-1 mutants, but not the exc-5 mutants (Fig. 7). While the result is consistent with RAB-8 acting downstream of EXC-9 and EXC-1, overexpression of these two genes by themselves gives stronger rescue (Grussendorf et al., 2016; Tong and Buechner, 2008), suggesting that EXC-9 and EXC-1 may stimulate multiple trafficking proteins. Both RAB-11.1 and RAB-8 target vesicles to the apical membrane (Sato et al., 2014), so they may be acting together downstream of EXC-1.

Rab8 has been seen to direct traffic apically and basolaterally in different organisms and cell types (Ang et al., 2003; Bellec et al., 2018; Sato et al., 2007), which complicates interpretation of our overexpression results. The exc mutants show clear defects in morphogenesis and maintenance of the apical surface, and overexpression of exc genes yields normal-diameter “convoluted” lumens within canals that are severely shortened on the basal surface. While RAB-8–directed cargo in vesicles inside the excretory canals has not been identified, overexpression of rab-8 phenocopies the convoluted canals. This suggests that RAB-8 directs traffic to the apical surface in this cell type.

RAB-8 did not rescue the effects of exc-5 mutation. This could indicate that RAB-8 acts upstream of EXC-5 (as in the model posited above), which is expressed at or near the canal apical surface (Mattingly and Buechner, 2011), or that RAB-8 stimulates a different guanine exchange factor while a different protein (perhaps again RAB-11) stimulates EXC-5. Further research on the function of various Rab proteins and guanine exchange factors at the canal apical surface should provide a fruitful avenue of study.

Several alternative hypotheses can explain how these interactions form the terminal web and guide outgrowth of the canal lumen. One attractive hypothetical model is shown in Fig. 8. EXC-1/IRG is associated with the membrane along the entire length of the canal. At sites where EXC-1 is exposed to the cytoplasm (i.e., at the growing canal tip and at sites of membrane addition during growth, such as evenly spaced “beads” in the L1 stage; Kolotuev et al., 2013), soluble EXC-9/CRIP can bind to and activate EXC-1 (Grussendorf et al., 2016), as well as be retained by the CTD of EXC-2. EXC-1 then directs the movement of vesicles via RAB-8 and/or via guanine exchange factor EXC-5/FGD to facilitate actin polymerization for vesicle delivery to that area of the tubules. Polymerization of cytoskeletal elements ACT-5/actin, SMA-1/spectrin, and the three intermediate filaments EXC-2, IFA-4, and IFB-1 follows.

While the present results provide a plausible framework by which intermediate filaments can serve as a luminal scaffold for proteins that regulate directed tubular growth, other models are equally plausible. Future studies examining the biochemistry of EXC-1 and EXC-9 activation during canal development will be invaluable in understanding tubular growth mediated by these two proteins and by intermediate filaments. Recent manuscripts on bioRxiv (Sun et al., 2020,Preprint; Winkelman et al., 2020 Preprint; not yet peer reviewed) suggest that other LIM-domain proteins are capable of measuring tension in cytoskeletal actin filaments. The growing excretory canal could provide an example of where the LIM domain of CRIP is performing a similar function with the stretchable domains of intermediate filaments. This model of the excretory canal potentially suggests that cytoskeletal filaments as measured by CRIP/LIM–domain proteins for repair/growth via IRG proteins and vesicle transport could be a widespread mechanism for growth and repair of cytoskeletal elements linked to membrane. The homology of EXC-9 to CRIP and EXC-1 to the IRG family of proteins (including the autophagy regulator IRGM) suggests that study of the excretory canal cell provides a model for greater understanding of the regulation of these ubiquitous mammalian proteins.

C. elegans strains were grown on BK16 bacteria (a streptomycin-resistant derivative of OP50; Brenner, 1974) at 20C according to standard techniques. All strains used are listed in Table S1.

EXC-2 NTD (amino acids 1–767), EXC-2 CTD (amino acids 1112–1339), RAB-8, and GFP::RAB-11.1 constructs were built by means of Gibson assembly (NEBuilder HiFi DNA Assembly Master Mix, Cat. #E2621S). pCV01 (Oka and Futai, 2000) was used as the backbone vector. Detailed primers are listed in Table S2. CRISPR/Cas9 modifications were performed according to the protocols of the Goldstein laboratory (Dickinson et al., 2015). The sites of modifications made are shown in Table S3.

Animals were examined via a Zeiss Axioskop microscope equipped for both epifluorescence and differential interference contrast microscopy with 40× (NA 1.3) and 63× (NA 1.4) objectives, and photographs were captured by use of an Optronics MagnaFire Camera. Animals were placed on 2% agarose pads in M9 solution + 35 mM NaN₃. Contrast on DIC images was uniformly enhanced over the entire image to increase clarity. The subcellular location of fluorescent proteins at high resolution was examined through a FluoView FV1000 laser-scanning confocal microscope (Olympus) with 60× objective (NA 1.42). Lasers were set to 488-nm excitation and 520-nm emission (for GFP and CFP) or 543-nm excitation and 572-nm emission (for mKate2 and mCherry). All confocal images were captured via FluoView software (Olympus), and collocation was analyzed by use of ImageJ software. Fluorescent images from both microscopes were false-colored (green to cyan, red to magenta). Electron microscopy on thin sections of immersion-fixed animals was performed on a Philips CM10 electron microscope following standard methods (Hall, 1995).

Canal length was measured as a percentage of full-length posterior canals (reaching past the anus); full length was rated 100%; canals reaching to the vulva in the center of the animal, 50%; and complete lack of extension, 0%; lengths between these landmarks were interpolated. Comparisons of canal length and of percentage of convoluted canals were calculated via one-way ANOVA.

Brightness of expression in the canals was measured in ImageJ by measuring brightness along a line perpendicular to the animal of maximum width (300 pixels). Average brightness of the background was subtracted from all, then the ratio of average brightness in areas outside of the canal was subtracted from average brightness in the area of the canal. The ratio of canal-specific brightness to brightness of the worm overall was plotted (e.g., Fig. 7 C).

Fig. S1 shows the expression pattern of EXC-9 outside the excretory canals. Fig. S2 shows the expression pattern of EXC-1 outside the excretory canals. Table S1 lists the strains used for this study. Table S2 lists the primers used for making constructs. Table S3 lists the mutations induced via CRISPR/Cas9 injection.

We thank anonymous reviewers for substantial improvements to the manuscript and for pointing out the recent studies of LIM-domain proteins on bioRxiv.org.

Z. Yang was supported by Kansas University Graduate Research fund no. 2144091. D.H. Hall was supported by National Institutes of Health OD 010943. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health, Office of Research Infrastructure Programs (P40 OD-010440).

The authors declare no competing financial interests.

Author contributions: Z. Yang participated in conceptualization; carried out the investigation, visualization, and use of software; and wrote the original draft. B.C. Mattingly participated in the investigation to create strains. D.H. Hall participated in investigation and visualization and analysis of electron micrographs. B.D. Ackley provided resources and methodology for acquisition of confocal images. M. Buechner was responsible for overall conceptualization, supervision, resources, and visualization and reviewed and edited the manuscript.