Abstract
The excretory canals of Caenorhabditis elegans are a model for understanding the maintenance of apical morphology in narrow single-celled tubes. Light and electron microscopy shows that mutants in exc-2 start to form canals normally, but these swell to develop large fluid-filled cysts that lack a complete terminal web at the apical surface, and accumulate filamentous material in the canal lumen. Here, whole-genome sequencing and gene rescue show that exc-2 encodes intermediate filament protein IFC-2. EXC-2/IFC-2 protein, fluorescently tagged via clustered regularly interspaced short palindromic repeats/Cas9, is located at the apical surface of the canals independently of other intermediate filament proteins. EXC-2 is also located in several other tissues, though the tagged isoforms are not seen in the larger intestinal tube. Tagged EXC-2 binds via pulldown to intermediate filament protein IFA-4, which is also shown to line the canal apical surface. Overexpression of either protein results in narrow but shortened canals. These results are consistent with a model whereby three intermediate filaments in the canals—EXC-2, IFA-4, and IFB-1—restrain swelling of narrow tubules in concert with actin filaments that guide the extension and direction of tubule outgrowth, while allowing the tube to bend as the animal moves.
POLARIZED cells form tubular structures ubiquitously in living organisms (Lubarsky and Krasnow 2003; Iruela-Arispe and Davis 2009; Maruyama and Andrew 2012). Tubes vary in their width, length, and mechanism of formation (Lubarsky and Krasnow 2003; Sigurbjörnsdóttir et al. 2014). However, the mechanism by which a narrow biological tube grows and maintains a uniform diameter throughout the life span of an organism is poorly understood. The excretory system of the nematode Caenorhabditis elegans provides a useful model of “seamless” (no intracellular adherence junctions) single-celled tubular structures (Sundaram and Buechner 2016) such as vertebrate capillaries or the tip cells of the Drosophila trachea (Lubarsky and Krasnow 2003). The core excretory system consists of a large excretory cell plus a duct and pore cell (Nelson et al. 1983). The excretory cell, located beneath the pharynx, extends four hollow canals throughout the length of the worm roughly in the shape of the letter “H” (Figure 1, A and B). The canals collect and excrete excess water from the body through the duct and pore to regulate organismal osmolarity (Nelson and Riddle 1984). Worms with defects in excretory canal function exhibit pale bloated canals and bodies, and have less tolerance to high-salt environments (Buechner et al. 1999; Wang and Chamberlin 2002; Hahn-Windgassen and Van Gilst 2009).
exc-2 mutant shows short and cystic canals. (A) Diagram showing the excretory canals in wild-type C. elegans with two tubes (red, apical surface; black, basolateral surface; and lumen in white) extended over the entire length of the worm and connected at the canal cell body. Canals extend from cell body in both directions, anteriorward and posteriorward. (B) DIC image of excretory canal of wild-type worm (N2); arrows indicate narrow canal lumen with uniform diameter. (C) DIC image of exc-2(rh247) mutant shows canal extending to only ∼30% of the wild-type length. Area outlined in red is magnified in (C’) to show the fluid-filled cysts accumulated throughout the entire canal. Bar, 50 µm.
In genetic screens, mutations in nine “exc” genes were discovered to affect canal structure to allow fluid-filled cysts to accumulate during canal extension during late embryogenesis and early first larval stage (Buechner et al. 1999). Other studies found similar mutations that affect additional tubular structures in the nematode, including the seamed single-cell excretory duct cell (Jones and Baillie 1995; Mancuso et al. 2012; Gill et al. 2016; Pu et al. 2017) and the multicellular intestine (Bossinger et al. 2004; Zhang et al. 2011; Carberry et al. 2012; Khan et al. 2013; Zhu et al. 2015; Geisler et al. 2016). In the canal cell, proteins implementing tubule structure comprise apical cytoskeletal elements (McKeown et al. 1998; Praitis et al. 2005; Khan et al. 2013; Kolotuev et al. 2013; Shaye and Greenwald 2015), vesicular trafficking and exocyst proteins (Tong and Buechner 2008; Mattingly and Buechner 2011; Armenti et al. 2014; Lant et al. 2015; Grussendorf et al. 2016), and ion and lipid transporters (Berry et al. 2003; Khan et al. 2013), among others.
Cytoskeletal components play an essential role in maintaining canal structure (Sundaram and Cohen 2017). Actin filaments are aligned over the apical surface of the canal lumen and dock with the apical membrane via the ezrin/radixin/moesin homolog ERM-1 (Göbel et al. 2004) and the apical βH-spectrin (Praitis et al. 2005), while mutations in the formin gene exc-6 compromise nucleation of microtubules along the length of the canal (Shaye and Greenwald 2015). C. elegans contains 11 cytosolic intermediate filament (IF) proteins, plus one nuclear lamin protein (Dodemont et al. 1994; Karabinos et al. 2001). Three IF proteins are highly expressed in the canal cell: IFC-2, IFA-4, and IFB-1 (Spencer et al. 2011). Knockdown of the ifb-1 gene causes cystic defects in both the canal and the multicellular intestine (Woo et al. 2004; Kolotuev et al. 2013).
While most of the original exc genes have been cloned (Grussendorf et al. 2016; Sundaram and Buechner 2016), mutations in the exc-2 gene cause particularly severe canal defects. In these mutants, the canal length is shortened by over one-half, the animals accumulate multiple cysts in the canals, and are sensitive to growth at low osmolarity (Buechner et al. 1999). Four alleles of this gene were discovered in the original screen, which suggested that it encodes a large protein. Here, we report that exc-2 encodes the IF IFC-2, and additionally found that mutations in the ifa-4 IF gene also cause cystic canal defects similar to those of exc-2 mutants. Overexpression of either exc-2 or ifa-4 results in shortened canals with small or no cysts. EXC-2 and IFA-4 proteins bind to each other and are located at the apical membrane of the canals. The position of EXC-2 at the apical membrane occurs independently of IFB-1 and IFA-4 function in the canals. These results indicate the importance of these three IFs in forming and maintaining the uniform diameter of the canals in this single-celled model of long, narrow tubular structure.
Materials and Methods
DNA constructs and double-stranded RNA synthesis
This study utilized two canal markers: pCV01 (used at 15 ng/µl) contains the gfp gene driven by the canal-specific vha-1 promoter and pBK162 (used at 25 ng/µl) contains the mCherry gene driven by the exc-9 promoter. Fosmid WRM0630A_E08 was provided by the Max Planck Institute, Dresden, Germany. Genomic DNA was used for exc-2 rescue and was prepared via PCR with LongRange enzyme (QIAGEN, Valencia, CA) to amplify the full-length ifc-2 gene, including 500 bp downstream and 1.3 kb upstream, which included all DNA between ifc-2 and the next gene, lpr-7. The translational construct of ifa-4 was made by ligating ifa-4 cDNA in-frame with the gfp gene in plasmid pCV01.
Double-stranded RNA (dsRNA) constructs were synthesized via PCR amplification of 250–350-bp regions of selected exons in genes of interest (Supplemental Material, Table S1). A MEGAscript T7 kit (ThermoFisher Scientific, Waltham, MA) was used for transcription. For ifb-1, RNA interference (RNAi) constructs were created by placing two constructs, each containing a complementary sequence corresponding to exon 4, under control of the canal-specific vha-1 promoter. Even with the canal-specific promoter, some F1 animals showed lethality and intestinal defects, so ifb-1 was knocked down solely through injections of our construct rather than attempting to form viable stable array lines.
Constructs for clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9-mediated mutation were created according to the method of the Goldstein laboratory (Dickinson et al. 2015). The single guide RNA constructs were made by amplifying plasmid pDD162 using primers containing a 20-base sequence specific for the gene of interest. The PCR product was self-ligated after treatment with T4 kinase. Goldstein group construct pDD282 was used for tagging exc-2, and pDD287 to tag ifa-4. Repair constructs were prepared through Gibson assembly of constructs containing the gene-specific tags in four overlapping amplified fragments, and ligated via NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs, Beverly, MA).
Nematode genetics and genetic mapping
Strains of C. elegans are shown in Table 1. They were maintained on lawns of bacterial strain BK16 (a streptomycin-resistant OP50 strain) on NGM agar plates as described (Sulston and Hodgkin 1988). Positions and sequences of qp110, rh90, rh105, rh209, and rh247 mutations, as well as locations of CRISPR/Cas9-induced insertions, are shown in Table S2.
By means of complementation tests and deficiency mapping, exc-2 was previously mapped to the left end of the X chromosome (Buechner et al. 1999). Strains of exc-2 to be sequenced were each outcrossed to a wild-type Hawaiian isolate (CB4856) as described (Minevich et al. 2012). For each of four mutant allele strains, 12 F2 progeny homozygous for the exc-2 mutation were selected and grown to populations that were combined for whole-genome sequencing. Sequencing was completed at the Genome Sequencing Core at the University of Kansas. Genome data analysis was carried out to identify mutations in the expected genetic area by use of the Galaxy cloud-map website (https://usegalaxy.org).
Genetic rescue assays of exc-2(rh247) mutants were performed through co-injection into the gonad of carrier DNA pCV01, plus either fosmid WRM0630A_E08 or PCR-amplified genomic exc-2 DNA. Injected animals were allowed to lay eggs, which were screened for expression of the GFP-expressing carrier DNA. F1 progeny expressing GFP were examined for canal morphology.
Rescue of the ifa-4 deletion mutant strain RB1483 was performed through gonad co-injection of an ifa-4 cDNA construct at 40 ng/µl together with marker plasmid pBK162. Injection of these exc-2 and ifa-4 constructs was also used to cause overexpression of the genes in wild-type worms.
RNAi knockdown of specific exc-2 isoforms was accomplished through co-injection of forward and reverse RNA together with carrier pCV01 into the gonads of young adult wild-type worms. Control injections of RNAi specific to the following genes from the left end of LGX, of which several are expressed in the canal cell, showed no effects on either canal length or lumen width in progeny of injected animals: C31H2.3 C31H2.4, mans-3, Y34B4A.2, vps-41, ags-3, pks-1, T23F2.3, lgx-1, and mtm-5. Injected animals were pooled in groups of 5 and F1 progeny exhibiting the GFP fluorescence of pCV01 were examined; the number of necessarily independent injections (which is the number shown in the table and used for calculating error bars) is therefore one-fifth of the actual number of injected animals; if more than one out of five injected animals actually provided transformed eggs, the error bars would be narrower than shown.
Complementation tests were carried out by mating male exc-2(rh247) to hermaphrodites of the BK530 (the CRISPR ifc-2 deletion) strain. Fluorescent hermaphrodite cross progeny all showed the strong cystic canal phenotype of exc-2 (n = 30).
CRISPR/Cas9-mediated knock-in strains were created through injection of repair constructs from modifications to pDD282 and pDD287 CRISPR reagents (AddGene.org, Cambridge, MA) to make plasmids pBK301 and pBK302, to insert a fluorescent marker and epitope tag between the 5′-UTR and coding region of exc-2 and ifa-4, respectively. Selection of strains containing the constructs was performed on NGM plates containing 250 μg/ml hygromycin (Sigma [Sigma Chemical], St. Louis, MO). Heat shock at 35° for 4–5 hr was used to activate removal of the selection cassette via self-excision. As the exc-2 and ifa-4 genes are located close together on the X chromosome, the doubly tagged exc-2; ifa-4 strain BK533 was created via CRISPR/Cas9-mediated tagging of BK531 (tagged exc-2) with pBK302.
Microscopy and canal measurement
Worms were examined through a Zeiss (Carl Zeiss, Thornwood, NY) Axioskop microscope with Nomarski (DIC) optics and epifluorescence. Animals were placed on 3% agarose pads in water and immobilized either through addition of Polybeads polystyrene beads (Polysciences, Warrington, PA) or of sodium azide (35 mM). Nonconfocal images were taken using an Optronics MagnaFire Camera. Some images of larger worms required two or three photographs that were “stitched” together to provide a picture of the entire animal. Contrast on DIC images was uniformly enhanced over the entire image to increase clarity. For protein subcellular location, worms were examined using an Olympus FluoView FV1000 laser-scanning confocal microscope (Olympus, Tokyo, Japan). Lasers were set to 488 nm excitation and 520 nm emission (GFP), or 543 nm excitation and 572 nm emission (mKate2). All images were captured via FluoView optics (Olympus) and collocation was analyzed using ImageJ software by drawing a straight line perpendicular to the length of the canal. Fluorescence plot profiles were then recorded and analyzed as shown in each image.
Electron microscopy was performed as described (Buechner et al. 1999). Young adult worms were cut and fixed in buffered glutaraldehyde and OsO4, encased in agar, dehydrated, and embedded in resin. Serial sections are ∼70 nm thick, and stained in uranyl acetate and lead citrate.
Canals were measured for length and cyst sizes as described (Tong and Buechner 2008). The length of each canal was scored between 0 and 4, where 4 indicates a full-length canal, 3 for canals that extended between the vulva and full-length, 2 for canals extending to the vulva, 1 for short canals that extended between the cell body and the vulva, and 0 for canals that did not extend past the cell body. Cyst size was scored by measuring cyst diameter relative to worm body width. Large cysts have a diameter similar to that of the worm body width, medium cysts have a diameter of approximately half-worm width, and small cysts are smaller than one-half the worm body width. A 3 × 2 Fisher’s exact test (http://www.vassarstats.net/fisher2x3.html) was used to compare the number of animals with short, medium-length, and full-length canals; and separately to compare number of animals with cysts, no cysts but canals with enlarged diameter, and normal-diameter canals.
Biochemistry and binding assays
Worms were collected after growth on strain BK16 cultured on twelve 100 mm plates of NGM medium until the bacterial lawn was consumed, then washed in M9 solution and frozen in liquid nitrogen until needed. Worm lysates were prepared through vortexing 500 µl thawed worms in a mixture of 300 µl dry 425–600-µm diameter glass beads (Sigma) and 700 µl lysis buffer [0.5% NP40, 150 mM NaCl, 50 mM Tris pH 8.0, 0.5 mM EDTA, 5% glycerol, 1 mM DTT, and protease-inhibitor tablet (ThermoFisher Scientific)]. The mixture was vortexed for 15 min at maximum speed at 4°, followed by 15 min on ice to cool down the samples, and again for another 15 min at 4°. The lysate was then centrifuged in an Eppendorf centrifuge at maximum speed for 5 min at 4° and protein concentration of the supernatant was measured.
Protein samples for co-immunoprecipitation were incubated with preblocked (using 5% BSA) anti-FLAG M2 magnetic beads (Sigma) at 4° for 30 min. Samples were then washed thoroughly with PBS-Tween (PBS-T) and eluted using 3× FLAG peptides (Sigma).
Samples for western blots were loaded onto Mini-PROTEAN TGX gels with a 4–20% gradient of bis-acrylamide (Bio-Rad, Hercules, CA). For immunoprecipitation samples, an equal volume was loaded in each well, while for lysate samples an equal amount of protein was loaded each well. Nitrocellulose membranes were blocked in 5% instant milk in PBS-T.
Tagged EXC-2 protein was detected via monoclonal anti-FLAG M2 antibody (Sigma) at a final concentration of 1 µg/ml in blocking buffer, and tagged IFA-4 protein was detected via HRP-bound anti-c-Myc antibody (Sigma) at a final concentration of 0.5 µg/ml in blocking buffer.
Data availability
Five supplemental figures, two supplemental tables, and two supplemental video files have been deposited at Figshare. Whole-genome sequences of Illumina reads of exc-2 alleles rh90, rh105, rh209, and rh247 are deposited in GenBank as submission identifier SUB4058160, BioProject identifier PRJNA472422. Allelic mutations of all exc-2 alleles are submitted to WormBase.org. Strains of all exc-2 mutant strains, and CRISPR/Cas9-modified strains of fluorescently and antigen-labeled exc-2 and ifa-4 are available upon request, and may be made available through the Caenorhabditis Genetics Center (CGC), University of Minnesota (cgc.umn.edu), pending acceptance to that repository. CRISPR/Cas9 repair construct plasmids pBK301 and pBK302 are available upon request. Supplemental material available at Figshare: https://doi.org/10.25386/genetics.6281537.
Results
Mutants in all alleles of exc-2 show severely cystic canal phenotypes, with multiple fluid-filled cysts evident along the entire length of the greatly shortened canals (Figure 1, C and C’ and Figure S1). Canal length is only 30–40% that of wild-type canals, and cysts have an average diameter four times as wide as that of a wild-type canal lumen. Electron microscopy images of exc-2 canals show areas where the electron-dense actin-rich terminal web of the apical membrane has been thinned or lost (Figure 2, B and B’) in comparison with that of a wild-type canal (Figure 2, A and A’), consistent with a loss in apical membrane support. The thinned areas lack canalicular vesicles relative to other areas where the terminal web is still intact. These vesicles connect transiently to the canals and are believed to regulate acidification and osmotic regulation of the animal (Kolotuev et al. 2013). The lumen of the cystic canals contains visible long filaments (Figure 2B’) that have not been detected in wild-type canals or in most other exc mutant canals, save for mutants affecting apical cytoskeleton proteins sma-1 (encoding βH-spectrin), erm-1 (ezrin-moesin-radixin homolog), and the excretory duct cell gene let-653 (mucin) (McKeown et al. 1998; Buechner et al. 1999; Khan et al. 2013; Gill et al. 2016). The exc-2 mutant defects are specific to the excretory canals; electron microscopy of the intestine shows an apparently normal terminal web surrounding well-formed and normally arranged microvilli, with no cystic effects (Figure 2, C and C’).
Electron microscopy of exc-2(rh209) excretory canal and intestine. Cross-sectional electron microscope images of wild-type (N2) and mutant [exc-2(rh209)] tissues. (A) Wild-type canal; area outlined in yellow contains lumen (connected to myriad small canaliculi) and is magnified in (A’). White arrows point to the dark actin-rich terminal web surrounding the lumen on all sides. (B) exc-2 (rh209) canal; area outlined in red is magnified in (B’). White arrows point to thick terminal web where present; red arrows point to regions of apical membrane lacking visible terminal web. Black arrowheads indicate presumed luminal scaffold material accumulating abnormally in the lumen of multiple exc-2 mutant alleles, but not visible in lumen of wild-type canals or other exc mutants. (C) Lower-magnification image of intestine in (C) wild-type and (C’) exc-2 (rh209) shows intact intestine with normal arrangement of microvilli and normal basal membrane surrounded by intact terminal web surrounding entire apical surface. Mutant also shows cystic canal. Bar, 1 µm.
Whole-genome sequencing of four exc-2 alleles revealed that the IF gene ifc-2 is mutated in all four strains (Figure 3A, Figure S2, and Table S2). Two of these alleles, rh209 and rh247, encode nonsense mutations in exon 10 and 12, respectively. Alleles rh90 and rh105 include deletions of multiple coding regions that cause frameshift mutations that lead to early stop codons (Figure 3A). To confirm the identity of the exc-2 gene, a null allele (qp110) in ifc-2 was generated via CRISPR/Cas9-induced deletion; this allele harbors a deletion in the 5′ region of the gene, including part of the fourth exon encompassing part of the 5′-UTR along with the first two bases of the coding sequence of the gene (Figure 3A and Table S2). The qp110 strain showed similar canal length and cyst number and size as for exc-2(rh247) (Figure 3, B and B’), and these two alleles failed to complement each other. To further confirm the identity of the exc-2 gene, dsRNA targeting the 12th exon of the ifc-2 coding region (dsRNA#1, Figure 4A) was injected into wild-type worms. These worms exhibited canal defects equivalent to those in exc-2 mutant animals (Figure 4, B and B’).
exc-2 encodes the intermediate filament (IF) protein IFC-2. (A) Structure of exc-2 gene (from WormBase, release WS262). Conserved region homologous to IF domain is shown in dark red. Alleles rh209 and rh247 contain nonsense mutations in exon 10 and 12, respectively. Alleles rh90 and rh105 include deletions in multiple coding regions that cause frameshift mutations, while clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9-generated allele qp110 deletes part of exon 4, including the promoter and two bases of the start codon of isoforms A, B, and C. CRISPR/Cas9-generated allele qp111 inserts gfp linked to three copies of FLAG-tag sequence at the start codon of isoforms A, B, and C. Bar, 1 kb. (B) exc-2(qp110) mutants (CRISPR/Cas9-generated deletion at the start codon) exhibit similar canal defects to those of other exc-2 mutants, as measured by canal length (B) and cyst size (B’); n = 50. (C) Phenotypic rescue of progeny animal of exc-2(rh247) animal injected with PCR-amplified ifc-2 gene, including 1.3 kb upstream and 500 bp downstream at 15 ng/ml. While 19% of animals were completely rescued, 15% were partially rescued by this concentration of injected gene (Table 2). (C’ and C’’) show magnification of boxed areas. Red arrows indicate terminus of canal. Bar, 50 µm.
EXC-2 isoform A (and likely isoform B) maintain canal structure. (A) Schematic diagram of isoforms of exc-2 (from WormBase release 262) showing positions of RNA interference (RNAi)-targeted regions (blue dashed lines). Below the diagram, exon numbers are shown for reference. Bar, 1 kb. (B) DIC and (B’) fluorescence images for progeny of injected worms with double-stranded RNA (dsRNA) #1 targeting the 12th exon of ifc-2. Animals presented canals as short and cystic as those of uninjected exc-2(rh247) animals. (C) Measurements of canal length (C) and cyst number (C’) in different isoform-specific RNAi knockdown animals (n > 50 each). To right of bars in (C) is minimum number of independent injections (one-fifth of total number of animals injected; see Materials and Methods). Controls were uninjected N2 (wild-type) and exc-2(rh247) animals. Wild-type animals injected with RNAi to genes located nearby to exc-2 on LGX showed canals as long as those of wild-type (see Materials and Methods). (D) Protein expression of N-terminal FLAG-tagged EXC-2. Western blot shows two predominant bands corresponding to predicted sizes of isoforms A and B.
Finally, a rescue assay was conducted via injection, either of the ∼51-kb fosmid WRM0630A_E08 containing the ifc-2 gene (as well as lpr-7, M6.11, pha-2, M6.4, and a part of rund-1) or of the ifc-2 gene and regulatory element (1.3 kb upstream and 500 bp downstream, total length 15.3 kb; the short upstream sequence included all DNA between ifc-2 and the next gene, lpr-7) PCR amplified from the WRM0630A_E08 fosmid, together with a GFP marker construct, into the gonad of exc-2(rh247) worms. We did not obtain stable lines that rescued the mutant; progeny of rescued animals tended to revert to the mutant phenotype in later generations, which suggests a possible gene-dosage effect of exc-2, as seen for other exc genes (Mattingly and Buechner 2011), as well as the occurrence either of gene silencing or loss of copies in the array over multiple generations (Praitis and Maduro 2011). Therefore, we examined only progeny of animals injected at a variety of concentrations of fosmid or amplified DNA (Table 2). Of the surviving progeny labeled in the canals, up to 16% showed complete rescue via fosmid injection at 12.5 ng/μl, with a full-length canal and complete absence of cysts (Figure 3C). An additional 9% were partially rescued by fosmid, exhibiting canals shorter than wild-type but much longer than those of exc-2(rh247) animals, with no cysts visible along the canal length. For the amplified gene at a concentration of 15 ng/μl, 19% were fully rescued plus 15% were partially rescued. The relatively low rate of rescue was significantly higher than that which occurred through control injection of a different IF gene, ifa-4 (0% Table 2). The exc-2(qp110) strain showed a similar low but significant rate of recovery (9%) when injected with the rescuing amplified ifc-2 gene product (Table 2). We conclude that the IF protein IFC-2 is encoded by the exc-2 gene, and will refer here to the gene and protein by the prior name, EXC-2.
To identify the isoforms of EXC-2 that are responsible for the excretory canal phenotype, we directly injected dsRNA into wild-type worms (synthesized and transcribed in vitro) targeted to specific exc-2 isoforms (Figure 4A). The canals in progeny animals were then evaluated with regard to canal length and cyst number and size (Figure 4, C and C’ and Table 3). The first dsRNA (#1 in Figure 4A) targeted the 12th exon, which knocks down isoforms A, B, and C. These worms showed a canal phenotype similar to that of exc-2 null strains in both length and cyst size (Figure 4, B and C). dsRNA #2 targets the 16th exon to knock down isoforms A, B, and D, and this knockdown also resulted in a phenotype similar to that of exc-2 mutants. As antibodies to isoform D bind to the intestine (Karabinos et al. 2001) and long-term knockdown of isoform D has been reported to affect intestinal structure (Hüsken et al. 2008), we also examined this organ in progeny animals, but saw no intestinal effects in progeny, even in animals exhibiting strongly cystic canals (Figure S3). dsRNA #3 targeted the 3′-UTR solely in isoform A, in exon 19. These worms showed a milder phenotype, in which the canal length reached the vulva midway along the length of the animal, and displayed cysts, but much smaller than those of exc-2 knockout animals, with no obvious luminal swelling. dsRNA #4 targeted the coding sequence that is uniquely transcribed at the 3′-end of isoform C. This knockdown had no effect on canal length and no cysts were formed. Finally, dsRNA #5 targeted the 5′-UTR of isoform D. Canal length was as long as in wild-type animals, no cysts were observed, and again, no intestinal effects were observed. Although we saw no obvious effect on canal cell or intestine of knockdown of isoforms C or D from multiple dsRNAs, we cannot determine conclusively that these isoforms have no function in these tissues. It is possible that our dsRNAs were degraded or otherwise did not achieve strong knockdown, or that the knockdown had effects on tissues that we did not examine. Since knockdown of isoforms A, B, and C together, and of isoforms A, B, and D together, showed stronger effects on canal length than did knockdown of isoform A alone (Table 3), we infer (but cannot prove) that isoform B is likely to be needed, along with the demonstrated need for isoform A in canal formation.
To validate the dsRNA data, we looked at total protein levels of isoforms A, B, and C in strain BK531 (tagged exc-2 strain), in which exc-2 was modified via CRISPR/Cas9 to place gfp and three copies of a flag tag just upstream of the starting AUG codon of these three isoforms. The western blot (Figure 4D) showed only two large isoforms in the worm lysate sample, corresponding to the size of isoforms A and B. We did not see a band corresponding to isoform C and RNA-sequencing (RNA-seq) results on WormBase show low levels of this predicted isoform compared to others. It should also be noted that the anti-FLAG antibody cannot detect isoform D; however, the result is still in agreement with the knockdown results. We conclude that both isoforms A and B are needed for EXC-2 function in the excretory canals.
Examination of strain BK531 (tagged exc-2) showed the expression pattern of the exc-2 gene (Figure 5). Labeled EXC-2 is located at the lumen of the excretory canal, in the corpus, posterior bulb, and pharyngeal–intestinal valve of the pharynx, as well as in the uterine seam and intestinal–rectal valve. The subcellular location of EXC-2 in the excretory canals was compared to that of exogenously expressed cytosolic mCherry (Figure 6). Labeled EXC-2 is located apical to canal cytoplasm (Figure 6, A–A’’), as determined via cross-sectional fluorescence intensity measurements (Figure 6A’’’). This result was confirmed by evaluating the subcellular location of EXC-2 relative to a known apical membrane protein, ERM-1 (Göbel et al. 2004). The results demonstrate that EXC-2 and ERM-1 show overlapping expression at the canal apical (luminal) membrane (Figure 6, B–B’’’).
EXC-2 is expressed in multiple epithelial cells. (A) A clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 knock-in of gfp at the N-terminus of exc-2 is expressed in four tissues: (1) the excretory canals (A and C); (2) the pharyngeal corpus, posterior bulb, and pharyngeal–intestinal valve (PIV) (B); (3) the interfacial uterine cell (UTSE) (C); and (4) the intestinal–rectal valve (VIR) (D). Bar, 50 µm.
EXC-2 is expressed at the apical membrane of the excretory canals. (A) Apical position of EXC-2 (GFP) relative to (A’) canal cytosol labeled with mCherry; (A’’) merged panel. (A’’’) Graph of fluorophore intensity for each color within canal section [box in (A’’)] along length of arrow. (B) Coincident position of EXC-2 (GFP); with (B’) apical epithelial marker ERM-1 labeled with mCherry; (B’’) merged panel. (B””) Graph of fluorophore intensity measurement within canal section boxed in (B’’). Bar, 5 µm.
Besides EXC-2 and IFB-1, IFA-4 is a third IF gene that is highly enriched in the canal (Spencer et al. 2011). We investigated whether IFA-4 plays a role in canal maintenance similar to that of EXC-2 and IFB-1 (Woo et al. 2004). The ifa-4(ok1734) strain from the CGC carries a deletion of the ifa-4 coding sequence and exhibits a canal phenotype similar to that of exc-2 mutants (Figure 7, A–A’’). Injection of dsRNA specific to ifa-4 phenocopies the short cystic canals of the ifa-4 deletion strain (Figure 7, B–B’’). Strain BK532 (tagged ifa-4 strain) was created via CRISPR/Cas9 to tag ifa-4 with mKate2 and three myc-tags at the 5′-end of the coding region (Figure 7C). Of these worms, 40% exhibited wild-type morphology, while in the other 60% (which tended to have brighter expression) the canals were slightly shortened (to position ∼3.5 on the scale of canal length (Figure 3B)), but did not exhibit any cysts. IFA-4::mKate2 showed expression in several of the same locations as exc-2: the excretory canals, the pharyngeal–intestinal valve, and the intestinal–rectal valve. IFA-4 is additionally located in the spermathecal–uterine valve and cells in the vulva, including the uterine muscles and two neurons in the tail (Figure 7C). The IFA-4 protein was also found in the gut of dauer-stage but not well-fed worms, and appeared as well in the tips of the rays and neurons of the male tail (Figure S4). An overexpressing translational construct under control of a strong canal-specific promoter (Pvha-1::ifa-4::gfp) rescued the mutant canal phenotype of the ifa-4(ok1734) deletion mutant strain (Table 2), expressed the gene in the same tissues as did our CRISPR/Cas9-modified ifa-4 strain (Figure 7), and showed subcellular expression of this protein at the apical membrane of the canal, though with large inclusions protruding deeper into the cytosol (Figure 8, A–A’’’). The 60% of BK532 (tagged ifa-4) animals that exhibited shortened canals also contained these subcellular inclusions. Finally, subcellular collocation of IFA-4 and EXC-2 at the canal apical membrane was confirmed in strain BK533 carrying CRISPR/Cas9-tagged insertions both of gfp into the 5′-end of exc-2 and of mKate2 into the 5′-end of ifa-4 (Figure 8, B–B’’’).
IFA-4 effects canal maintenance. (A) DIC and (A’ and A’’) fluorescence micrographs of excretory canal of strain BK535 (ifa-4 knockout; canal marker). Boxed area of (A’) magnified in (A’’). (B) DIC and (B’) fluorescence micrographs for progeny of animal injected with double-stranded RNA targeting the seventh exon of ifa-4. Animals exhibited short, cystic canals. Boxed area of (B’) magnified in (B’’). (C) Diagram of clustered regularly interspaced short palindromic repeat/Cas9 knock-in of mKate2 at the N-terminus of ifa-4. ifa-4 is expressed in five tissues: (1) the excretory canals (red arrows); (2) the pharyngeal–intestinal valve (blue box); (3) the spermathecal–uterine valve (yellow box); (4) the uterine muscles (red box); and (5) the intestinal–rectal valve and neurons at the tip of the tail (green box). Boxed areas are below main image, outlined in boxes of the corresponding color. All bars, 50 µm.
IFA-4 is colocalized with EXC-2 to the apical membrane. (A) Fluorescence micrographs of injected (40 ng/µl) translational construct (Pvha-1::ifa-4::gfp) ectopically expressed in ifa-4(ok1734) deletion mutant, which fully rescued canal morphology. (A’) Cytosolic mCherry marker coexpressed in the canal; (A’’) merged image. (A’’’) Graph of fluorophore intensity for each color within canal section [box in (A’’)] along length of arrow. (B) Fluorescence micrographs of clustered regularly interspaced short palindromic repeat/Cas9-modified strains labeling exc-2 and ifa-4. (B) shows GFP::EXC-2 expression, (B’) mKate2::IFA-4 expression, (B’’) merged image. (B’’’) Graph of fluorophore intensity for each color within canal section. Bar, 5 µm.
Several exc genes (exc-1, exc-5, and exc-9) with knockout canal phenotypes of large cysts also exhibit characteristic overexpression phenotypes that rescue the canal lumen diameter, while shortening canal length, and show epistatic genetic interactions (Tong and Buechner 2008; Mattingly and Buechner 2011; Grussendorf et al. 2016). We therefore looked at overexpression phenotypes of exc-2 and ifa-4. PCR-amplified exc-2 that rescued exc-2 mutants (Figure 3C) was injected (together with a fluorescent canal marker) into N2 wild-type worms (Figure 9A and Table 4). All progeny showing the injection marker also exhibited substantially shortened canals extending close to the vulva [average canal length of 1.5 (n = 42)], but with small unseptate swellings or cysts that did not enlarge the overall canal diameter. After injection of the ifa-4 rescuing construct into wild-type worms, progeny expressing GFP (Figure 9A’) showed shortened canals extending just beyond the vulva (length 2.6, n = 59). The similarity of overexpression phenotype between the two IF genes is consistent with these proteins performing similar functions.
EXC-2 and IFA-4 interact directly. (A) Representative N2 worms injected with (A) exc-2 genomic construct (including 1.3 kb upstream and 500 bp downstream of coding region) or (A’) ifa-4 cDNA construct under control of strong canal promoter vha-1 to overexpress these proteins. Both lines exhibited shortened canal lumen of wild-type diameter with no cysts. For exc-2 overexpression, length = 1.5 (n = 40); for ifa-4 overexpression, length = 2.6 (n = 59). Bar, 50 µm. (B) Western blot of whole-worm lysates probed with antibody to MYC. (B’) Western blot of lysates purified via anti-FLAG magnetic beads, which bind to FLAG-tagged EXC-2. Lysates are from wild-type animals (N2, left), animals expressing Myc-tagged IFA-4 (middle), and animals expressing both Myc-tagged IFA-4 and FLAG-tagged EXC-2 (right); lanes on blots are flanked by size marker lanes. Red arrow in blot on right indicates size of IFA-4 band at ∼100 kDa.
A co-immunoprecipitation assay was conducted to test whether EXC-2 and IFA-4 bind each other directly (Figure 9, B and B’). Protein lysates were prepared from three worm strains: wild-type (N2); BK532 (tagged ifa-4), and BK533 (tagged exc-2 & ifa-4). Tagged IFA-4 could be detected in blots of whole-worm lysates (Figure 9B), but was only detectable in αFLAG pulldowns when tagged EXC-2 was present (Figure 9B’), which indicates that the two proteins bind to each other.
The interaction of EXC-2 with IFA-4 led us to investigate whether the apical localization of EXC-2 in the canals depends on the function of the other two IFs. Strain RB1483, carrying the ifa-4(ok1734) deletion, was crossed to strain BK531 (tagged exc-2) (Figure 10, A–A’’’’) and then injected with a cytosolic mCherry marker construct. The subcellular location of EXC-2 at the apical membrane of the canal was unchanged in these ifa-4 knockout mutants.
EXC-2 apical location is independent of IFB-1 and IFA-4 function. Fluorescence micrographs of GFP-tagged EXC-2 in: (A) ifa-4 knockout strain (ok1734), (B) animal exhibiting canal-specific RNA interference (RNAi) knockdown of ifb-1, and (C) RNAi knockdown of ifb-1 in ifa-4 deletion strain. All animals express cytosolic mCherry marker. (A, B, and C) GFP::EXC-2 fluorescence. (A’, B’, and C’) Cytosolic mCherry fluorescence. (A’’, B’’, and C’’) Merged images. (A’’’, B’’’, and C’’’) Plane of focus higher in Z-plane, to observe merged fluorescence of GFP::EXC-2 at apical surface of swollen cystic areas of canals flattened against hypodermis. Boxed areas are magnified in (A’’’’, B’’’’, and C’’’’). (D) Fluorescence of GFP::EXC-2 at the narrower surface next to hypoderm in wild-type canals. All bars, 5 µm.
The apical location of EXC-2 was similarly tested in the absence of IFB-1. Due to embryonic lethality of ifb-1 null mutants [due to hypodermal defects (Woo et al. 2004)], constructs expressing dsRNA to ifb-1 under a canal-specific promoter were injected together with cytosolic mCherry marker construct into strain BK531 (tagged exc-2) (Figure 10, B–B’’’’). These animals exhibited cystic canals consistent with knockout of IFB-1, and EXC-2 retained its apical location in the presence of this ifb-1 knockdown. Finally, we investigated whether the apical location of EXC-2 depends on the presence of at least one of the other two IFs, IFA-4 and IFB-1 (Figure 10, C–C’’’’). Injection of the same canal-specific dsRNA constructs against ifb-1 along with cytosolic mCherry marker construct into the ifa-4(ok1734) mutant strain and crossed to BK531 (tagged exc-2) resulted in the same location for EXC-2. The IFA-4 knockout/IFB-1 knockdown was sickly, with lethality higher than for either ifb-1 knockdown or ifa-4 mutation alone (n = 97, Figure S5A), consistent with an additive function of IFA-4 and IFB-1. Death in these animals occurred primarily throughout midlarval stages (Figure S5, B and C), much later than the twofold embryonic muscle failure and death of ifb-1 null mutants (Woo et al. 2004), which suggests a general inability of canals to function effectively in the combined absence of these two IF proteins. (Buechner et al. 1999; Hahn-Windgassen and Van Gilst 2009).
Finally, the subcellular expression pattern of EXC-2 formed a mesh-like network on the apical surface of the canals, both in wild-type canals (Figure 10D) and in animals lacking IFA-4, IFB-1, or both (Figure 10, A’’’’–C’’’’, File S1, and File S2). This meshwork is more easily observed on cysts, where a larger surface area is in the focal plane, than in the narrow canals of wild-type animals, but the size and general arrangement of filaments appears similar in all cases, which indicates that EXC-2 filaments can be localized to the apical membrane even without the function of the other two canal IF proteins.
Discussion
exc-2 encodes for the IF IFC-2
This study has used multiple methods to confirm the identity of the exc-2 gene: (1) whole-genome sequencing of four alleles; (2) injection of dsRNA to knock down multiple isoforms of the gene; (3) rescue of a null allele by both fosmid and PCR-amplified DNA; (4) generation of a new allele of the gene via CRISPR/Cas9-induced deletion, which caused the same phenotype; (5) noncomplementation between the CRISPR-induced deletion allele and a previous allele; and (6) CRISPR/Cas9-mediated tagging of the exc-2 gene, showing the tissue and subcellular location of the encoded protein at the apical surface of the tissue affected by mutation; this subcellular excretory canal location is confirmed through pulldown of tagged EXC-2 protein to another IF expressed at the same apical location, which has the same genetic effects on the excretory canals.
This cloning result is surprising, since it contradicts multiple previous studies that found IFC-2 not to be in the excretory canals, but primarily in desmosomes of the C. elegans intestine (Karabinos et al. 2002, 2004; Hüsken et al. 2008; Carberry et al. 2012; Coch and Leube 2016; Geisler et al. 2016), though expression of one fluorescent construct showed expression at the intestinal apical surface (Coch and Leube 2016). These previous studies assumed isoform D to be the full-length IFC-2, since this isoform is about the same size as other C. elegans IF proteins, and some confusion of RNA-seq results presented on earlier versions of WormBase (e.g., wormbase.org, release WS180) listed up to eight splice forms and suggested the possibility that the first 13 exons (i.e., isoform C) comprised a separate gene. More extensive RNA-seq since that time (summarized on current WormBase release WS262) shows that although isoform C and isoform D do not overlap, both are transcripts from different parts of the same gene, and that the large isoform B in fact comprises the entire 24 exons of the gene. Western blot of FLAG-labeled EXC-2 protein in the present study (Figure 4D) matches the current predicted protein sizes on WormBase. We also note that the pioneering comprehensive study of C. elegans IF genes (Dodemont et al. 1994) reported eight IF genes, including ifc-2, and presents a northern blot (Dodemont et al. ’94, Figure 4) showing two much larger transcripts that correspond in size to the isoform A and B mRNAs. Subsequent work on IFC-2 from the Weber, Leube, and Karabinos laboratories (Karabinos et al. 2001, 2002, 2003, 2004) created and used a polyclonal antibody to a fragment of the conserved IF domain of IFC-2; this antibody bound a single ∼55-kDa protein on blots (Karabinos et al. 2003) and further studies showed this antibody binding specifically to proteins in the intestine (Hüsken et al. 2008), as long-term treatment with RNAi to ifc-2 removed intestinal staining but not staining of other tissues. Our CRISPR-tagged GFP was inserted at the first codon of isoforms A, B, and C, and so cannot show the expression of isoform D. It is therefore possible that isoform D is expressed in the intestine, while the other isoforms are expressed in the excretory canals and other tissues shown here. However, since isoforms A, B, and D all include the conserved IF domain, we cannot explain why previous studies did not find expression, or a phenotype, within the excretory canals.
Previous studies (Hüsken et al. 2008) found that in worms fed RNAi specific to the IF domain over the course of three generations, the animal intestines slowly acquired gross morphological effects, although they saw that at the ultrastructural level, the microvilli and terminal web appeared intact. We directly microinjected dsRNAs specific to multiple isoforms into the gonad and found that, in all cases, progeny intestines were unaffected, while the excretory canals uniformly were strongly cystic, matching the phenotype of all five of our exc-2 mutant alleles. In particular, this result was obtained for dsRNA #2 (Figure 4A), a knockdown of the conserved IF domain of isoforms A, B, and D. Finally, the rh105 allele has deletions in the conserved domain of these three isoforms (Figure 3A) and exhibits large canal cysts with no intestinal defects (Figure S1). In summary, our results strongly suggest that, while isoform D may be expressed in the intestine, the major locus of function of the EXC-2/IFC-2 IF isoforms is in the excretory canals.
EXC-2, IFB-1, and IFA-4 maintain tubular morphology at the apical membrane
At 12.7 kb, the exc-2 gene is easily the largest IF gene in C. elegans, which helps explain the relatively high frequency of alleles of this gene identified in genetic screens for cystic canal mutants (Buechner et al. 1999). Knockdown of the large isoform A (and likely isoform B) of this protein had clear effects within the canals, as seen by mutations at multiple sites and effects of dsRNA to all areas of the gene. No effects were noticed from knockdown solely of isoform C or isoform D.
A map of gene expression for specific tissues of C. elegans (Spencer et al. 2011) examined expression in the excretory canal cell and, in addition to exc-2, found two other highly expressed IF genes, ifb-1 and ifa-4, confirming earlier studies of IF expression (Karabinos et al. 2003). Knockdown of ifb-1 has previously been found to cause formation of cysts in the canal, as well as surprising breaks in canal cytoplasm during cell outgrowth (Karabinos et al. 2003; Woo et al. 2004; Kolotuev et al. 2013). The present study shows that IFA-4 is also necessary to maintain canal morphology. IFs form homo- and heteropolymers to carry out their functions (Zuela and Gruenbaum 2016). While IFA-4, IFB-1, and EXC-2 are all expressed at the apical (luminal) surface of the excretory canals, they each have varied expression (and presumably function) in other tissues, both overlapping and nonoverlapping. IFB-1, in particular, has an essential role in embryonic muscle attachment and hypodermal cell elongation (Woo et al. 2004), tissues where IFA-4 and EXC-2 are not expressed. Future studies of these tissues may find other functions for these IF proteins. For example, as stretching of dissected intestines can be measured (Jahnel et al. 2016), it may be possible to determine the role of IFA-4 in intestinal integrity during dauer formation, as compared to intestines in other stages where IFA-4 is not expressed.
Ultrastructural analysis of exc-2 mutants found visible fibrous material in the lumen (Figure 2B’), which has also been seen only in other excretory system mutants affecting apical cytoskeleton proteins: sma-1 (encoding βH-spectrin), erm-1 (ezrin-moesin-radixin homolog), and the excretory duct cell gene let-653 (mucin) (McKeown et al. 1998; Buechner et al. 1999; Khan et al. 2013; Gill et al. 2016). Future studies may show whether the fibrous material represents proteins or other material normally anchored to the membrane directly or indirectly by these cytoskeletal proteins.
Three IF proteins support the canal apical membrane
The three IF proteins EXC-2, IFA-4, and IFB-1 are collocated at the apical membrane of the canal, have similar knockdown effects, and bind to each other in pulldown assays (Figure 9B; Karabinos et al. 2003). The ratio between these proteins affects their function, as overexpression of either IFC-2 or IFA-4 allows formation of small cysts in short canals (Figure 9, A and A’).
EXC-2 forms homo- and heterodimers in in vitro studies (Karabinos et al. 2017), which may be necessary to form a strong meshwork at the canal apical surface, as seen for lamins and other IF proteins at cell membranes. Lamin, for example, is bound directly to the membrane to form such structures at the inner nuclear membrane (Fawcett 1966; Dechat et al. 2008; de Leeuw et al. 2018) through farnesylation of the CAAX domain at the lamin C-terminus (Dechat et al. 2008; Webster et al. 2009). EXC-2, IFA-4, and IFB-1 do not have a CAAX domain, so do not appear to be bound to the apical canal membrane through the same mechanism as lamin. Other IF proteins such as keratin and vimentin form such meshworks linked to mammalian cell cytoplasm through a scaffolding protein, Plastin1 (Iwatsuki et al. 2002; Grimm-Günter et al. 2009). Though C. elegans does not have an obvious plastin homolog, it will be interesting to see if a protein yet to be identified serves a similar purpose to link EXC-2 to the canal apical surface. Since EXC-2 retains its location in the canal, and meshwork appearance when ifa-4 and ifb-1 are mutated and knocked down, respectively (Figure 10), EXC-2 appears to be located to the apical membrane independently of these other two IF proteins. All three proteins are necessary to prevent cyst formation and overexpression of ifa-4 does not rescue exc-2 mutation; these results suggest that the three filament proteins provide complementary functions to regulate tubule diameter and length.
We present a working model of these three proteins at the surface of the excretory canal in Figure 11. The three IF proteins EXC-2, IFA-4, and IFB-1 are collocated at the apical surface of the canal. We do not know if they are linked as obligate heterodimers; as EXC-2 binds to IFA-4 in our pulldown assay, this heterodimer presumably makes up some of the filaments. Nonetheless, EXC-2 is properly placed even without IFA-4 function, so EXC-2 dimers, either homodimers of one isoform or heterodimers of two isoforms, are likely part of the makeup of the filaments in wild-type animals.
Proposed model of EXC-2, IFA-4, and IFB-1 in excretory canals. (A) Lateral section of the excretory canal, where intermediate filaments (green, red, and blue) surround the apical membrane (black). Actin filaments make up the thick terminal web, extending from the apical membrane into the cytoplasm of the canals. Actin filaments are cut away in bottom half of drawing to show intermediate filaments more clearly. (B) Cross-section of the canal shows intermediate filaments forming a meshwork that surrounds the lumen surrounded by actin filaments.
The actin cytoskeleton forms a thick terminal web around the canal lumen that is visible in electron micrographs, and much thicker than the narrow band of IFs visible in the fluorescently tagged confocal micrographs here. The terminal web is tethered to the luminal membrane through the action of the SMA-1 βH-spectrin (Buechner et al. 1999; Fujita et al. 2003; Praitis et al. 2005) and the ezrin-radixin-moesin homolog ERM-1 (Khan et al. 2013). We do not know if actin is closer to the membrane than are the IF proteins, but tagging of ACT-5 actin (Shaye and Greenwald 2015), as well as electron micrographs of wild-type canals (Figure 2A, Nelson et al. 1983; Buechner et al. 1999), suggests that the thick actin-based terminal web likely extends outside of the thin layer of IFs seen in this study. Future studies on the in vivo interactions between these IF isoforms and proteins, and between these IFs and the actin cytoskeleton, should provide further insights into the ability of this network of proteins to provide the rigidity to maintain a firm luminal diameter, while maintaining the flexibility to allow these narrow tubes to lengthen and bend during growth and movement.
Acknowledgments
We acknowledge the assistance of the University of Kansas (KU) Molecular Biosciences Sequencing Center. Other helpful discussions were provided by Stuart MacDonald, Meera Sundaram, members of the KU Genetics of Development Seminar, and members of the Baltimore Worm Club. We thank Verena Göbel for the fluorescently labelled erm-1 strain. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health (NIH) Office of Research Infrastructure Programs (P40 OD-010440). Fosmid WRM0630A_E08 was the gift of the Max Planck Institute, Dresden, Germany. H.I.A. was supported in part by KU Graduate Research Funds numbers 2301847 and 2144091. The Center for C. elegans Anatomy is supported by NIH grant OD-010943 to D.H.H. E.A.L. was supported by NIH grants NS-0090945, NS-0095682, NS-0076063, and GM-103638.
Footnotes
Supplemental material available at Figshare: https://doi.org/10.25386/genetics.6281537.
Communicating editor: B. Grant
- Received April 25, 2018.
- Accepted May 23, 2018.
- Copyright © 2018 by the Genetics Society of America
