INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Basement membranes are specialized regions of extracellular matrix (ECM) that have important roles in many fundamental developmental and regenerative processes, including cell adhesion and migration, signal transduction, and even gene regulation (Martin and Timpl, 1987; Yurchenco and Schittny, 1990). Many of these processes are mediated by specific interactions between basement membrane components and transmembrane receptors such as integrin (Hynes, 1992). Basement membranes contain a large number of different components, including laminin, collagen, nidogen, and heparan sulfate proteoglycans (Yurchenco and O'Rear, 1994; Timpl and Brown, 1996). Homologues of these proteins have been identified in the nematode Caenorhabditis elegans (reviewed in Kramer, 1997), and mutations are associated with several of these components (Guo et al., 1991; Ishii et al., 1992; Rogalski et al., 1993; Sibley et al., 1993). This genetic approach is helping to reveal the function of these basement membrane proteins during morphogenesis. In this study, we focus on perlecan, the major basement membrane heparan sulfate proteoglycan, and its role in muscle development in C. elegans.

In C. elegans, a specialized basement membrane underlies the body-wall muscles and anchors the myofilament lattice through integrin-containing adhesion complexes (reviewed in Moerman and Fire, 1997). In adult animals, there are 95 body-wall muscle cells arranged in four quadrants, two dorsal and two ventral, beneath the hypodermis (reviewed in Waterston, 1988). Each quadrant runs the length of the animal and consists of a double row of spindle-shaped cells. Within each muscle cell, the thin and thick filaments of the myofilament lattice lie just beneath the plasma membrane facing the hypodermis. These filaments are anchored by a series of attachment structures to an underlying basement membrane (Francis and Waterston, 1985, 1991).

The actin-containing thin filaments are anchored by dense bodies, which are comparable to the Z-lines of vertebrate striated muscles (Waterston, 1988). Dense bodies are similar in composition to other cell-matrix adhesion complexes (Yamada and Geiger, 1997) and include vinculin (Barstead and Waterston, 1989, 1991b), talin (Moulder et al., 1996), -actinin (Barstead and Waterston, 1991b), UNC-97/PINCH (Hobert et al., 1999), and integrin (Gettner et al., 1995). The myosin-containing thick filaments, in turn, are anchored by M-lines (Waterston, 1988). The composition of M-lines in C. elegans is not well defined but includes integrin (Francis and Waterston, 1985; Gettner et al., 1995), UNC-97/PINCH (Hobert et al., 1999), and UNC-89 (Benian et al., 1996). Integrins link both dense body and M-line components to the underlying basement membrane (Francis and Waterston, 1985; Gettner et al., 1995). The interaction of integrin and basement membrane components is a key early event in the assembly of these attachment structures (reviewed in Moerman and Fire, 1997).

The unc-52 gene encodes the nematode homologue of mammalian perlecan (Rogalski et al., 1993), the major heparan sulfate proteoglycan of the ECM (Noonan et al., 1991; Kallunki and Tryggvason, 1992; Murdoch et al., 1992). Perlecan has five distinct domains, with similarity to the low-density lipoprotein receptor (domain II), laminin (domains III and V), and the neural cell adhesion molecule (NCAM) (domain IV). Perlecan has been implicated in a number of biological processes, including glomerular filtration (Farquhar, 1982), mitogenesis and angiogenesis (Aviezer et al., 1994), and cell adhesion through interactions with focal adhesion complexes (Hayashi et al., 1992; Chakravarti et al., 1995). Biochemical studies indicate that perlecan binds both itself and other ECM components, including laminin, collagen, nidogen, and fibronectin (Laurie et al., 1986; Yurchenco et al., 1987; Heremans et al., 1990; Battaglia et al., 1992; Iozzo et al., 1994). The multidomain structure of perlecan reflects the diverse functions proposed for this molecule.

In C. elegans, perlecan/UNC-52 is found in the basement membrane between the body-wall muscle cells and the hypodermis and is concentrated at muscle dense bodies and M-lines (Francis and Waterston, 1991; Rogalski et al., 1993). The absence of UNC-52 blocks myofilament lattice assembly during embryogenesis, resulting in a paralyzed, arrested elongation at twofold (Pat) terminal phenotype (Williams and Waterston, 1994). The body-wall muscles in unc-52(null) mutants lack organized A- or I-bands, and morphological studies reveal that even the earliest stages of myofilament lattice assembly are defective (Rogalski et al., 1993; Williams and Waterston, 1994).

Our previous studies have demonstrated that unc-52 pre-mRNA gives rise to a number of distinct protein isoforms through regulated alternative splicing (Rogalski et al., 1993, 1995; Lundquist et al., 1996). These studies identified two major classes of UNC-52 isoforms: a short form (domains I-III) and a medium form (domains I-IV) (see Figure 1). In this study, an additional isoform is described, a long form (domains I-V), which is very similar to mammalian perlecan. Further isoform diversity is generated by alternative splicing within the various domains. For example, alternative splicing of exons 16, 17, and 18 gives rise to isoforms that vary in the number of NCAM-like immunoglobulin repeats within domain IV (Rogalski et al., 1993, 1995). These exons each encode a single NCAM-repeat and are arranged such that single or multiple exons can be spliced from the pre-mRNA without disrupting the reading frame. Nonsense mutations in these alternatively spliced exons are not lethal, but instead result in progressive paralysis (Brenner, 1974; Gilchrist and Moerman, 1992; Rogalski et al., 1993, 1995). The relatively mild phenotype reflects the fact that not all UNC-52 isoforms are eliminated in these mutants (Rogalski et al., 1993, 1995). Removing the affected exons from the mRNA by eliminating the upstream splice acceptor site results in an almost wild-type animal (Rogalski et al., 1995). Thus, individual NCAM repeats within this region can be removed without disrupting muscle assembly.

The mec-8 gene encodes a putative RNA-binding protein that regulates some of the alternative splicing events within this region of unc-52 (Lundquist et al., 1996). mec-8 mutations exhibit a synthetic lethal interaction with viable unc-52 mutations; mec-8; unc-52 (viable) double mutants are paralyzed and arrest at the twofold stage of embryonic development (Lundquist and Herman, 1994). This synthetic lethal interaction results from the absence of mec-8-dependent splicing events. In a mec-8(+) background, splicing around affected exons permits the expression of some full-length products that are sufficient for embryogenesis to proceed normally (Lundquist et al., 1996); however, in the absence of mec-8-function, these splicing events do not occur, preventing the expression of full-length products.

In this study, antibodies specific to domains III, IV, and V of UNC-52 were used to study the three major groups of protein products. We found both temporal and spatial differences in the localization of UNC-52 isoforms. In embryos, short (domains I-III) isoforms are associated with the pharyngeal and anal muscles, whereas domain IV-containing isoforms are associated with the body-wall muscles. In adults, domain IV-containing isoforms become more widely distributed and are detected in basement membranes adjacent to most contractile tissues. We identified Pat alleles that specifically eliminate domain IV-containing isoforms, indicating that these isoforms are essential for myofilament lattice assembly in the body-wall muscles. In addition, we show that it is possible to remove up to four consecutive NCAM repeats without adverse effects on muscle development. Our results suggest that the number of NCAM repeats within domain IV is not of primary importance, nor are specific repeats within this region essential. Using a similar approach, we examined the newly identified domain V of UNC-52. Surprisingly, a deletion that specifically removes this region has no adverse effects on myofilament assembly or stability.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Nematode Strains and Culture Conditions

Nematodes were grown on NGM plates as described by Brenner (1974). Some strains were obtained from B. D. Williams (University of Illinois at Urbana-Champaign, Urbana, IL) and R. H. Waterston (Washington University, St. Louis, MO). Additional strains were provided by the Caenorhabditis Genetics Center. C. elegans strains used in this work include the wild-type strains N2; CB444, unc-52(e444); CB669, unc-52(e669); CB998, unc-52(e998); CB1012, unc-52 (e1012); CB1421, unc-52(e1421); DM4001, unc-52(st196::Tc1); RW6010, unc-52(st549)/unc-52(st549)/mnDp34; RW6011, unc-52(st546)/unc-52(st546)/mnDp34; and RW6013, unc-52(st560)/unc-52(st560)/mnDp34. The deletion allele unc-52(gk3) was provided by the C. elegans Reverse Genetics Core facility at the University of British Columbia.

PCR Amplification of Genomic DNA

Standard PCR reactions were performed essentially as described by Barstead et al. (1991). For long-range PCR, the standard PCR buffer and Taq polymerase were replaced with low-salt buffer and TaqPlus Polymerase (Stratagene, La Jolla, CA). All PCR mixtures were amplified in a Perkin Elmer-Cetus 480 thermocycler (Perkin-Elmer, Norwalk, CT) for 30 cycles consisting of 30 s at 95°C, 60 s at 53-57°C, and 1-5 min at 72°C, followed by a 5 min incubation at 72°C.

RT-PCR

C. elegans total RNA was generously provided by Eleanor Mathews (University of British Columbia). RT-PCR was performed essentially as described by Rogalski et al. (1993), except that 1 µg of RNA was used in each RT reaction.

Sequencing of cDNAs and PCR products

The yk48h9 cDNA was obtained as a ZapII clone from the C. elegans cDNA Project. The pBluescript plasmid containing the cDNA insert was excised following the protocol provided by Stratagene and transformed into Escherichia coli (XL1-Blue strain). This construct was designated DM#201. Plasmid DNA was prepared for sequencing using an alkaline lysis/polyethylene glycol precipitation procedure, and the clone was sequenced by the Nucleic Acid/Protein Service unit at the University of British Columbia. PCR and RT-PCR products were directly sequenced using the BRL dsDNA Cycle Sequencing System (Life Technologies, Gaithersburg, MD) as described by Rogalski et al. (1993). The gk3 deletion was sequenced by the Nucleic Acid/Protein Service unit.

Generation of Polyclonal Antisera

The DH5 strain of E. coli was used for subcloning and protein expression. Fusion proteins from the DM#184 and DM#183 clones were used to generate the GM1 (domain III) and GM3 (domain IV) antisera, respectively. Construction of these clones was described previously (Rogalski et al., 1995; Moerman et al., 1996). We also generated a polyclonal serum, GM9, that recognizes domain V of UNC-52. This antiserum was raised against the DM#199 fusion protein (Ser 3250 to Gly 3332) and recognizes this fusion on Western blots; however, we were not able to obtain reproducible staining results in embryos or adult worms using this serum.

GST fusion proteins were purified as described by Smith and Johnson (1988). To generate polyclonal antisera, New Zealand White rabbits were injected subcutaneously with fusion protein emulsified in Freund's complete adjuvant (~0.5 mg protein per rabbit). Rabbits were boosted at ~4-wk intervals with fusion protein emulsified in Freund's incomplete adjuvant (~0.25 mg protein per rabbit), and blood samples were taken 12 d after injection. Immune response was monitored by Western blotting of fusion proteins and immunofluorescence staining as previously described (Moerman et al., 1996). No staining was observed with preimmune sera or with the secondary antibodies by themselves. Staining was eliminated by preincubating the antisera with the specific fusion protein but not with control fusion proteins. In addition, no staining was observed in unc-52(null) mutant embryos.

Immunofluorescence Staining

Embryos were prepared and stained as previously described (Rogalski et al., 1993). Larvae and adults were stained as described by Finney and Ruvkun (1990). For immunofluorescence staining, rabbit polyclonal sera were diluted as follows: GM1, 1:400; GM3, 1:1500-1:6000. The mouse monoclonal antibodies DM5.6 (Miller et al., 1983) and MH3 (Francis and Waterston, 1991) were diluted 1:40 and 1:100, respectively. The secondary antibodies, FITC-labeled donkey anti-rabbit IgG F(ab')₂ and TRSC-labeled donkey anti-mouse IgG F(ab')₂ (Jackson ImmunoResearch Laboratories, West Grove, PA), were diluted 1:100-1:200. In some experiments, FITC-labeled phalloidin was used to visualize actin.

Microscopy

Confocal images were collected using the MRC 600 system (Bio-Rad Microsciences Division, Hercules, CA) attached to a Nikon Optiphot-2 compound microscope. Optical sections were collected at 0.2-µm intervals and combined using the "maximum projection" function. For publication, confocal images were arranged and annotated using Adobe Photoshop and printed on a Codonics NP-1600 printer. For polarized light microscopy, worms were viewed as described by Waterston et al. (1980).

Isolation of Deletion Revertants

To isolate Tc1 excision events from unc-52(st196::Tc1), we first established a mut-4(st700) I; unc-52(st196::Tc1) II strain. Wild-type (N2) males were crossed to dpy-5(e61) I; unc-52(st196::Tc1) II hermaphrodites. Male progeny (dpy-5/+; unc-52/+) were then mated to mut-4(st700) I hermaphrodites, and F1 animals with the genotype dpy-5/mut-4; unc-52/+ were identified by progeny testing. Unc non-Dpy progeny from these animals were picked singly onto new plates and allowed to self-cross. Unc animals that failed to segregate Dpy Unc progeny were expected to have the genotype mut-4(st700); unc-52(st196::Tc1). Several independent lines that reverted at a high frequency (>1 × 10³) were established and maintained by picking single Unc animals to new plates. Revertants were identified on the basis of their improved movement and larger body size compared with their Unc siblings. Revertants were maintained for several generations until a homozygous strain was established and then tested by PCR to determine whether they carried a detectable polymorphism in the region of interest.

Isolation of Lethal Tc1 Excision Events

To isolate lethal Tc1 excision events, N2 males were crossed to unc-52(st196::Tc1); mut-4(st700) hermaphrodites, and individual outcross progeny were transferred and brooded. These plates were scored for the presence of Unc progeny. If the Tc1 element did not excise, we expected to observe 25% Unc progeny; however, if the element excised and the excision site was repaired precisely, we expected to observe only wild-type progeny. Similarly, if the element excised and interrupted repair of the excision site resulted in a large deletion, we would not observe Unc progeny. Animals that failed to segregate Unc offspring were brooded, and plates were scored for the presence of Pat embryos. A single lethal allele, ra112, was obtained in this manner and was subsequently balanced with the free duplication mnDp34 (Herman et al., 1979).

Construction of mec-8; unc-52 Double Mutants

Previous work had established that mec-8; unc-52(viable) double mutants exhibit a synthetic lethal (Pat) phenotype (Lundquist and Herman, 1994). To construct mec-8; unc-52(viable) double mutants, N2 males were crossed to unc-52(viable) II hermaphrodites. Outcross males (unc-52/+) were then crossed to mec-8(u74) I hermaphrodites, and mec-8/+; unc-52/+ animals were identified by progeny testing. From plates segregating Pat embryos, embryos were prepared and stained. To construct mec-8; unc-52(revertant) double mutants, spontaneous males from revertant strains were crossed to mec-8(u74) hermaphrodites, and mec-8/+; unc-52/+ animals were identified by progeny testing. From plates that segregated Mec animals, single Mec hermaphrodites were transferred and brooded. The parental hermaphrodite was then tested by PCR for the deletion allele of unc-52. Strains carrying a deletion allele were retested for the wild-type unc-52 allele to determine whether they were homozygous for the deletion. To confirm that these animals were homozygous for mec-8, they were stained with DiO (Molecular Probes, Eugene, OR) to evaluate the dye-filling defects (Herman and Hedgecock, 1990).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The unc-52 Gene Encodes the Nematode Orthologue of Mammalian Perlecan

The unc-52 gene in C. elegans produces several large proteins that are homologous to perlecan, the mammalian basement membrane heparan sulfate proteoglycan. Our previous analysis of this gene identified 26 exons covering almost 15 kb of genomic DNA and revealed the presence of several alternatively spliced transcripts (Rogalski et al., 1993, 1995). Two different polyadenylation sites located 8.5 kb apart are used to generate a number of large polypeptides containing domains I-IV or smaller polypeptides containing only domains I-III.

Recently, we discovered that the unc-52 gene actually spans >20 kb of genomic DNA and consists of 37 exons (Figure 1). Eleven additional exons were identified downstream of exon 26 when the C. elegans Genome Consortium provided the sequence and annotation of cosmid C38C6. We confirmed the intron-exon boundaries in this region by sequencing a cDNA clone obtained from the C. elegans cDNA Project that extends from exon 27 to ~200 bp downstream of the putative stop codon in exon 37. In addition, cDNA fragments were generated by RT-PCR and sequenced to confirm that the newly identified exons are part of unc-52 and to identify the splice sites used to join exons 26 and 27. 

View larger version (33K): [in this window] [in a new window]   Figure 1.   Structure of the unc-52 gene and protein products. The unc-52 gene consists of 37 exons and spans over 20 kb. Exons (boxes), introns (lines), and the three classes of protein products are shown. Mutant alleles used in this study are also indicated. The longest ORF encodes a protein of 3375 amino acids that is homologous to the mammalian heparan sulfate basement membrane proteoglycan perlecan. Like mammalian perlecan, this polypeptide can be divided into five domains (I-V). The first domain is unique, whereas the remaining four domains show similarity to the low-density lipoprotein receptor (domain II), -laminin (domains III and V), and NCAM (domain IV). Additional isoforms are generated through alternative splicing of exons encoding alternative C termini, indicated on the gene as shaded regions. The various protein modules are indicated with shaded boxes or circles.

The longest potential ORF of the unc-52 gene now encodes a 3375 amino acid protein consisting of a putative signal sequence and five distinct domains with a molecular weight of ~370 kDa (Figure 2). The newly identified exons encode sequences that are very similar to domain V of mouse and human perlecan, confirming our earlier conclusion that the UNC-52 proteins are the nematode orthologues of these mammalian proteoglycans (Rogalski et al., 1993). Domain V of perlecan/UNC-52 consists of three globular regions interrupted by cysteine-rich repeats and is similar to the G domain of 1 laminin. Curiously, the nematode protein contains a region of ~180 amino acids that is not found in the mammalian proteins. This region is extremely rich in threonine (45/180) and serine (19/180) residues and also contains 12 repeats of the sequence EEP.

View larger version (80K): [in this window] [in a new window]   Figure 2.   Complete amino acid sequence of UNC-52/perlecan and isoforms. Domains are indicated, and the signal peptide is italicized and underlined. The alternative C termini of the short and medium length isoforms are indicted (bold), and the threonine-rich region in domain V is also shown (bold and underlined).

On the basis of this new information, UNC-52 isoforms can be divided into three major groups (Figures 1 and 2). Short (S) isoforms contain the first three domains (I, II, and III), medium (M) isoforms contain the first four domains (I, II, III, and IV), and long (L) isoforms contain all five domains (I, II, III, IV, and V). Alternative splicing of exons 6, 16, 17, 18, 21, and 22 generates additional diversity within domains III and IV (Rogalski et al., 1993, 1995).

Spatial and Temporal Differences in the Localization of UNC-52 Isoforms

Several studies have examined the localization of perlecan/UNC-52 in C. elegans (Francis and Waterston, 1991; Hresko et al., 1994; Moerman et al., 1996). In this study, we compare the distribution of UNC-52 isoforms using domain-specific antibodies. The polyclonal serum GM1 recognizes a region of domain III present in all UNC-52 isoforms (Figure 1) (Moerman et al., 1996). The polyclonal serum GM3 recognizes a conserved region of domain IV (Figure 1) and has the same domain specificity as the mAbs MH2 and MH3 (Rogalski et al., 1993, 1995). We double-labeled wild-type animals at different stages of development with these antisera and the mAb DM5.6, which recognizes the body-wall muscle myosin MHC A (Miller et al., 1983). Our results demonstrate that there are spatial and temporal differences in the localization of UNC-52 isoforms.

Previous studies established that some UNC-52 isoforms are expressed in the body-wall muscles during embryogenesis and are localized to the underlying basement membrane (Rogalski et al., 1993; Hresko et al., 1994; Moerman et al., 1996). In this study, we found that both GM1 and GM3 stain the body-wall muscles and are identical in this respect. Staining is first observed in comma stage embryos and is primarily found at regions of contact between adjacent muscle cells (Figure 3, A and B). Some intracellular staining of body-wall muscle cells is also observed, but neither sera stain the underlying hypodermis, suggesting that muscle cells are the primary source of UNC-52. Between the comma and 1.5-fold stages, staining spreads from regions of cell-cell contact over the basal face of each muscle cell, where the basement membrane is located (Figure 3, C and D).

View larger version (59K): [in this window] [in a new window]   Figure 3.   Localization of UNC-52 isoforms during embryonic development. Embryos were visualized by confocal microscopy. Wild-type embryos were labeled with GM1 (domain III; A, C, E, and G) and GM3 (domain IV; B, D, F, and H). A and B show the dorsal view of a comma-stage embryo (~350 min); these images are projections of the dorsal Z-sections. Arrows indicate positions of two adjacent muscle cells; the arrowhead indicates intracellular staining. Subsequent images are projections of the complete Z-series. C and D show the lateral view of late comma-stage embryo (350-400 min). Arrows indicate the position of a muscle cell, whereas the arrowheads indicate the basal face of the cell. E and F show the lateral view of a 1.5-fold embryo (~420 min). The arrow in both panels indicates the basal face of the muscle cells in a dorsal quadrant. Note the pharyngeal staining (arrowhead) in E. G and H show a threefold embryo (~800 min). Note that GM1 stains the pharynx (arrowhead in G), body-wall muscles, and anal muscles (arrow in G), whereas GM3 only stains the body-wall muscles. Bar, 10 µm.

In older embryos, dramatic differences in staining are observed with GM1 and GM3. Beginning at the 1.5-fold stage, GM1 stains the posterior end of the pharynx (Figure 3E), which is derived from the MS lineage (Sulston et al., 1983). The anterior pharynx, which is derived from the AB lineage (Sulston et al., 1983), begins to stain somewhat later. By the threefold stage, GM1 staining surrounds the pharynx from the anterior margin to the pharyngeal-intestinal valve (Figure 3G). GM1 also stains the anal sphincter and depressor muscles at this stage (Figure 3G). UNC-52 is specifically associated with contractile tissues in C. elegans and is not found in the basement membranes lining the pseudocoelom or surrounding the intestine (Table 1).

In contrast, GM3 only stains the body-wall muscles and does not stain the pharynx or anal muscles at any stage of embryonic development (Figure 3). On this basis, we conclude that M and/or L isoforms are restricted to the body-wall muscles during embryogenesis. GM1 staining of the pharynx and anal muscles must therefore be due to the presence of S isoforms in these tissues.

Both GM1 and GM3 stain the body-wall muscles in embryos, larvae, and adults. Briefly, in larvae and adults, these antisera stain the dense bodies, M-lines, and muscle cell margins, in addition to the basement membrane underlying the muscle quadrants. Both antisera also stain the basement membranes associated with the pharyngeal, anal, and sex-specific muscles in adult animals (Table 1). GM1 staining of the pharynx, body-wall muscles, and reproductive muscles is shown in Figure 4. We conclude that there are developmental changes in isoform localization because domain IV-containing isoforms are more widely distributed in adults than in embryos. For example, GM3 does not stain the pharynx or anal muscles in embryos but does stain these tissues in adults. These results are similar to those reported for the mAbs MH2 and MH3 (Table 1) (Francis and Waterston, 1991; Hresko et al., 1994).

View larger version (104K): [in this window] [in a new window]   Figure 4.   Immunolocalization of UNC-52 in adult hermaphrodites. Wild-type hermaphrodites were labeled with GM1. A shows the head from a young adult. The arrow indicates the terminal bulb of the pharynx. Note the punctate pattern over this region. B shows a section of the body-wall muscles from a young adult. The large arrow indicates a dense body, the small arrow indicates an M-line, and the arrowhead indicates the margin of a body-wall muscle cell. C shows the uterine region from an older adult (dorsal view). The arrow indicates the base of the uterine muscles, whereas the arrowhead indicates the vulva. Bar, 10 µm.

Other studies have shown that UNC-52 is found in the basement membrane surrounding the pharynx in larvae and adults (Francis and Waterston, 1991; Rogalski et al., 1993). In this study, we also found that UNC-52 is concentrated over focal adhesion-like structures in the pharyngeal muscles. In animals prepared as described by Finney and Ruvkun (1990), GM1 staining surrounds the pharynx from the anterior margin to the pharyngeal-intestinal valve, and muscle cells within the terminal bulb stain with a punctate pattern (Figure 4A). This punctate staining pattern was not observed in animals prepared by freeze-fracture.

M Isoforms of UNC-52 Are Essential for Myofilament Lattice Assembly in the Body-Wall Muscles of C. elegans

The st549 mutation introduces a stop codon into exon 7 of the unc-52 gene and eliminates all UNC-52 protein isoforms (Rogalski et al., 1995). We have shown that unc-52(st549) mutant embryos fail to stain with MH3, which recognizes M and L isoforms of UNC-52 (Rogalski et al., 1993), and GM1, which recognizes all UNC-52 isoforms (Moerman et al., 1996). In this study, we use these antibodies to examine the expression of UNC-52 isoforms in embryos homozygous for several additional lethal alleles of unc-52. These mutant embryos are phenotypically indistinguishable from unc-52(st549) embryos and fail to show body-wall muscle staining with GM1 or MH3; however, they do show pharyngeal staining with GM1, indicating expression of the S isoforms of UNC-52.

We double-labeled mutant embryos prepared from balanced stocks with GM1 and either DM5.6 or MH3. Four lethal alleles, ra112, st546, st560, and st578, lead to reduced anti-UNC-52 staining of the body-wall muscles relative to the pharynx or anal muscles. For example, in homozygous st560 embryos, staining of the body-wall muscles with GM1 is greatly reduced or absent (Figure 5E). Similarly, we observed no detectable staining of the body-wall muscles with MH3, indicating that isoforms with domain IV are greatly reduced or absent (Figure 5L). Within the body-wall muscle cells, myosin is not organized into ordered A-bands but instead forms large aggregates (Figure 5F) (Williams and Waterston, 1994). GM1 staining of the pharynx and anal muscles, however, appears to be unaffected (Figure 5E). Thus, S isoforms of UNC-52 are still expressed in these mutants. Similar results were observed with ra112, st546, and st578 (our unpublished results). The lack of body-wall muscle staining with GM1 leads us to conclude that S isoforms are not present in body-wall muscles during embryogenesis. The body-wall muscle staining observed in wild-type embryos must, therefore, be due to M and/or L isoforms.

View larger version (46K): [in this window] [in a new window]   Figure 5.   unc-52(st560) mutant embryos have tissue-specific staining defects and lack a subset of UNC-52 isoforms. Embryos were visualized by confocal microscopy. Embryos in A-F were double-labeled with GM1 (green), which recognizes all UNC-52 isoforms, and DM5.6 (red), which recognizes myosin heavy chain A (MHC A). Small arrows indicate the pharynx, and large arrows indicate a body-wall muscle quadrant. A and D show both channels simultaneously, whereas B, C, E, and F show single-channel images. A-C show a wild-type embryo, whereas D-F show an arrested unc-52(st560) mutant embryo. Note the reduced staining of body-wall muscles with GM1 and disorganization of MHC A in the mutant (E and F; compare with B and C). The pharynx and anal muscles (arrowhead in E) in the mutant, however, exhibit a wild-type staining pattern (compare B and E). Embryos in G-L were double-labeled with GM1 (green, FITC) and MH3 (red, TRSC), which recognizes an epitope in domain IV of UNC-52. G-I show a wild-type embryo; J-L show a unc-52(st560) mutant embryo. G and J show both channels simultaneously. Note the absence of MH3 staining in the mutant embryo in L. Although the unc-52(st560) mutant embryos shown are arrested at the twofold stage, they are comparable to threefold wild-type embryos. Bar, 10 µm.

On the basis of these results, we predicted that the sequence changes in these alleles would be localized to the region encoding domain IV. To test this prediction, we amplified genomic DNA from homozygous mutant embryos using PCR and began sequencing downstream from exon 11, which encodes the first NCAM repeat in domain IV. We identified the nucleotide alterations in two of the unc-52 alleles described above, and as expected, both were localized to the region encoding domain IV (Table 2). The st560 mutation is a C to T transition in exon 13, which changes a glutamine residue (CAA) to an ochre stop codon. ra112 has a 3283-bp out-of-frame deletion that extends from exon 13 to exon 19. Both mutations eliminate all UNC-52 isoforms with domain IV, demonstrating that these isoforms are essential for myofilament assembly in the body-wall muscles.

The st560 and ra112 mutations disrupt the expression of both M and L isoforms of UNC-52. We obtained a domain V-specific deletion from the C. elegans Reverse Genetics Core Facility to determine the effect of specifically eliminating the L isoforms. This deletion, gk3, is 2735 bp in length and removes most of the domain V-encoding region of unc-52 (Figure 1). Animals homozygous for the gk3 deletion are viable and essentially wild type in appearance. The only discernible phenotype is a slight difference in egg-laying behavior; unc-52(gk3) mutants tend to retain eggs within the gonad to a greater extent than wild-type animals. We emphasize that this defect is quite mild and resembles the behavior of wild-type animals under starved conditions. On the basis of this mild phenotype, we conclude that domain V is not essential, and M isoforms of UNC-52 are sufficient for myofilament lattice assembly.

UNC-52 Is Not Essential for Myofilament Lattice Assembly in the Pharyngeal Muscles of C. elegans

We have demonstrated that domain IV-containing isoforms of UNC-52 are essential for myofilament lattice assembly in the body-wall muscles. Because S isoforms of UNC-52 are restricted to the pharynx and anal muscles during embryonic development, we speculated that these isoforms might be required for myofilament assembly in these muscles. To test this prediction, we stained wild-type (N2) and unc-52(st549) embryos with fluorescently labeled phalloidin. Homozygous st549 embryos do not express detectable levels of any UNC-52 isoforms, including those associated with the pharynx. Surprisingly, pharyngeal thin filaments did not appear to be disorganized in unc-52(st549) mutants.

In wild-type threefold embryos, phalloidin stains I-bands in the body-wall, anal, and pharyngeal muscles. By this stage, actin is organized into distinct half I-bands in the pharyngeal muscles. For example, in lateral views of M3 cells (procorpus), half I-bands can be seen extending inward from the basal and luminal faces of each cell (Figure 6A). The H-zone, which appears as a gap between half I-bands, can also be observed in appropriately oriented embryos. In cross sections, thin filaments radiate from a focal attachment site at the luminal face of each pharyngeal muscle cell to more broadly distributed attachment sites on the basal face (Figure 6B).

View larger version (109K): [in this window] [in a new window]   Figure 6.   Phalloidin staining in wild-type and unc-52(null) embryos. Embryos were visualized by confocal microscopy. Wild-type (A-C) and unc-52(st549) (D-F) embryos were labeled with FITC-phalloidin. Images in A and D show staining from all focal planes. Computer-generated cross sections (B and E) and single focal plane images (C and F) are also shown. The arrowhead indicates the basal surface of the pharynx, whereas the arrow indicates the basal face of the body-wall muscles. Note the well organized thin filaments extending from the basal and apical faces of the pharynx in both wild-type and mutant embryos. Also note that actin in the body-wall muscles of the mutant is not associated with the basal cell membrane. Bar, 10 µm.

In unc-52(st549) embryos, the pharynx is compressed lengthwise and distorted in shape. The terminal bulb can be identified by its position, but the metacorpus cannot be readily distinguished from the procorpus and isthmus. These morphological defects are observed in all Pat mutants and probably result from failure to elongate; however, within the pharyngeal muscles, thin filaments are organized into well ordered half I-bands extending from both luminal and basal faces (Figure 6B). We examined the half I-bands in the five largest muscle layers (M3, M4, M5, M6, and M7 cells) and did not observe any disorganization. The apparent wild-type organization of the pharyngeal muscle thin filaments in unc-52(null) mutants is consistent with the observation of pharyngeal pumping in these animals (Williams and Waterston, 1994). We conclude that UNC-52 is not essential for myofilament lattice assembly in pharyngeal muscles.

Alternative Splicing within Domain IV Is Associated with Temporal and Spatial Differences in Isoform Expression

Alternative splicing of exons 16, 17, and 18 gives rise to isoforms that vary in the number of NCAM repeats within domain IV (Rogalski et al., 1993, 1995). Mutations in these alternatively spliced exons eliminate a subset of M and L isoforms (Rogalski et al., 1993, 1995). Five point mutations in this region have been sequenced; four are nonsense mutations in either exon 17 (e669 and e1012) or 18 (e444 and e998), whereas the fifth (e1421) alters the splice donor site of exon 16 (Figure 1) (Rogalski et al., 1995). Animals homozygous for these viable alleles develop normally as young larvae but become progressively paralyzed as they mature. This paralysis is caused by the gradual disruption of the myofilament lattice in the body-wall muscle cells posterior to the head (Mackenzie et al., 1978b; Waterston et al., 1980). We have shown that these mutations do not disrupt the accumulation of UNC-52 during early development (Rogalski et al., 1995).

In contrast, these mutations have dramatic effects on the accumulation of UNC-52 isoforms associated with the body-wall muscles in older animals (Figure 7C), suggesting that there is an "adult-specific" subset of isoforms. In homozygous e444 hermaphrodites, for example, staining of the body-wall muscles appears to be normal until the L4 stage but is greatly reduced relative to wild-type animals in adults (Figure 7C). Within the body-wall muscles, myosin becomes highly disorganized and forms large aggregates (Figure 7D) (Waterston et al., 1980). Curiously, staining of body-wall muscles in the head is not affected, even in older adults, and staining of the pharynx and uterine muscles also appears normal. We conclude that the e444 mutation affects an adult-specific subset of UNC-52 isoforms associated with most of the body-wall muscles. Other viable alleles, including e669, e998, e1012, and e1421, behave in a similar manner.

View larger version (125K): [in this window] [in a new window]   Figure 7.   Immunolocalization of UNC-52 and myosin in unc-52(viable) mutants. Wild-type (A and B) and unc-52(e444) adults (C and D) were double-labeled with GM1 (UNC-52) and DM5.6 (MHC A). Note the reduced GM1 staining in C and disorganized myosin in D (compare with A and B). unc-52(viable) mutations disrupt the accumulation of UNC-52 and organization of myosin in late larval and adult animals. Bar, 10 µm.

Alternative splicing within this region is also associated with spatial differences in isoform expression. The gene mec-8 encodes a putative RNA-binding protein that is required for a subset of these alternative splicing events (Lundquist et al., 1996). We have shown that nonsense mutations in exon 18 block the expression of domain IV-containing isoforms in a mec-8() background (Lundquist et al., 1996). We constructed additional mec-8; unc-52 double mutants and found that some combinations do not eliminate all domain IV-containing isoforms. In mec-8; unc-52(e669) and mec-8; unc-52(e1012) double mutants, most of the body-wall muscles fail to stain with MH3; however, we did observe strong staining of the anteriormost muscle cells in each of the body-wall muscle quadrants (Figure 8B). Both e669 and e1012 are point mutations that introduce translational stop codons into exon 17 (Rogalski et al., 1995). Thus, in a mec-8() background, nonsense mutations in exon 17 greatly reduce or eliminate most domain IV-containing isoforms but allow expression of certain spatially restricted splice variants. These observations imply that certain splice variants are restricted to a subset of body-wall muscles.

View larger version (48K): [in this window] [in a new window]   Figure 8.   Expression of UNC-52 isoforms in mec-8; unc-52 double mutants. A shows GM1 staining of a mec-8(u74); unc-52(e1012) double mutant. Note that pharyngeal staining appears to be normal, but body-wall muscle staining is greatly reduced, except over the anteriormost muscles. B shows MH3 staining of the same embryo. Note that body-wall muscle staining is restricted to the anteriormost muscle cells (indicated by arrows). Bar, 10 µm.

Several NCAM Repeats within Domain IV Are Dispensable for Muscle Assembly

Previous studies suggested that single exons in the region of alternative splicing in domain IV are dispensable for protein function (Gilchrist and Moerman, 1992; Rogalski et al., 1995). Here we have extended these studies and show that removing up to four exons (four NCAM repeats) does not affect viability or muscle assembly. The allele unc-52(st196::Tc1) has a Tc1 transposon insertion in exon 18, and homozygous animals exhibit the paralyzed phenotype typical of unc-52(viable) mutants. In a mutator background, the transposon excises at a high frequency (>1 × 10³) to give phenotypically wild-type animals. Excision can be precise, where the wild-type sequence is restored, or imprecise, where deletions and other rearrangements occur (Kiff et al., 1988; Moerman et al., 1991). We isolated >120 independent revertants from a mut-4; unc-52(st196::Tc1) strain and asked whether any carried deletions or insertions in the region of interest.

Using primers flanking the Tc1 insertion, we amplified DNA fragments from homozygous revertants and identified 10 strains with DNA alterations (Table 2). One revertant has a 12-bp in-frame insertion, whereas the remaining nine contain deletions, ranging in size from 982 to 1519 bp. These deletions remove between two and four NCAM repeats without altering the reading frame (Table 2). We examined the body-wall muscles in these revertants using polarized light microscopy and found that these animals have normal muscle structure, including a wild-type number of sarcomeres in each muscle cell. These results suggest that the number of NCAM repeats is not of primary importance, nor are specific repeats within this interval essential.

As described earlier, a putative RNA-binding protein encoded by the mec-8 gene regulates some of the alternative splicing events within this region of unc-52 (Lundquist et al., 1996). mec-8 mutants exhibit a number of phenes, including mechanosensory and chemosensory defects, and a low-penetrance cold-sensitive lethality associated with muscle attachment defects (Lundquist and Herman, 1994). mec-8 mutations also exhibit a synthetic lethal interaction with unc-52(viable) mutations; synthetic lethality results from the absence of mec-8-dependent splicing events. The deletions described in this study were isolated in a mec-8(+) background and remove part or all of the region in which mec-8-dependent splicing occurs. We tested two deletions, ra515 and ra516, in combination with a putative null allele of mec-8, and found that these animals do not exhibit a synthetic lethal phenotype; however, they exhibit the full range of mec-8 phenes, including the mechanosensory defects and cold-sensitive lethality.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Mammalian perlecan, a heparan sulfate proteoglycan, is an abundant component of most basement membranes. It is synthesized by a wide variety of cell types, including epithelial cells (Ohji et al., 1994; van Det et al., 1995), fibroblasts (Heremans et al., 1989; Murdoch et al., 1992), and myocytes (Murdoch et al., 1993), and has been detected in all mammalian basement membranes surveyed to date (reviewed in Noonan and Hassell, 1993). In the nematode, we found that UNC-52 is localized to basement membranes associated with contractile tissues, including the pharyngeal, body-wall, and anal muscles. We did not detect UNC-52 in the basement membranes lining the pseudocoelom or surrounding the intestine. On this basis, we conclude that UNC-52 is not a general basement membrane component in C. elegans but is specifically associated with contractile tissues.

Recently, Graham et al. (1997) examined the distribution of the 1 and 2 collagen IV chains in C. elegans. The expression and localization of UNC-52 and collagen IV differ in several respects. First, collagen IV is more widely distributed than UNC-52 and is associated with the gonad and intestine at most developmental stages (Graham et al., 1997). Second, collagen IV is expressed predominately in the body-wall muscles and is exported to basement membranes surrounding other tissues, including the pharynx and intestine (Graham et al., 1997). In contrast, UNC-52 is expressed in a wider range of cell types, including pharyngeal and anal muscles, but does not diffuse beyond the expressing cells. Laser ablation studies established that UNC-52 acts cell autonomously and does not spread beyond the site of expression (Moerman et al., 1996). We report here that UNC-52 isoforms exhibit spatial differences in localization. The inability of UNC-52 to diffuse or to be transported beyond the site of expression is probably important for establishing and maintaining these distinct spatial patterns.

A Subset of UNC-52 Isoforms Are Associated with Body-Wall Muscles during Embryogenesis and Are Required for Myofilament Lattice Assembly

In this study, we found that M isoforms of UNC-52 are essential for myofilament lattice assembly in the body-wall muscles. Myofilament lattice assembly in the nematode is remarkably similar to assembly of focal adhesions in mammalian cell culture (Burridge et al., 1988; Moerman and Fire, 1997). In both processes, integrin-ECM interactions are required to initiate assembly and stabilize existing adhesion complexes (Moerman and Fire, 1997; Yamada and Geiger, 1997). The localization of UNC-52 over the body-wall muscles and the effects of unc-52(lethal) mutations on myofilament assembly suggest that UNC-52 anchors the dense bodies and M-lines, perhaps through interactions with integrin. Whether UNC-52 plays an instructive role or simply an attachment role in assembly of integrin complexes at the muscle cell membrane is not clear; however, without a stable focal attachment structure at the muscle cell membrane, sarcomere units within muscle cells cannot be properly organized (Williams and Waterston, 1994; reviewed in Moerman and Fire, 1997).

Mammalian perlecan is widely expressed and has been detected in basement membranes of skeletal and cardiac myocytes (Murdoch et al., 1993). A number of studies have demonstrated cell adhesive properties for various domains within perlecan (Hayashi et al., 1992; Battaglia et al., 1993; Chakravarti et al., 1995), and integrin has been identified as a cell-surface mediator of this attachment (Hayashi et al., 1992; Battaglia et al., 1993). Two genes, pat-3 (Gettner et al., 1995) and ina-1 (Baum and Garriga, 1997), have been shown to encode integrin subunits in C. elegans. The -integrin encoded by the pat-3 gene (PAT-3) is expressed in various tissues, including the body-wall muscles, where it is localized at the transmembrane regions of dense bodies and M-lines (Francis and Waterston, 1985; Gettner et al., 1995). In contrast, the  integrin encoded by the ina-1 gene (INA-1) is expressed most abundantly in the pharynx where it is concentrated beneath the basement membrane (Baum and Garriga, 1997). Thus, the body-wall and pharyngeal muscles synthesize distinct integrin heterodimers. An intriguing possibility is that different UNC-52 isoforms are capable of binding distinct integrin heterodimers. In the body-wall muscles, M isoforms colocalize with PAT-3, whereas in the pharynx, S isoforms colocalize with INA-1. These observations suggest that M and S isoforms may interact with different integrin complexes and function in distinct processes.

Several observations suggest that UNC-52 interacts directly with PAT-3 integrin. First, the distribution of UNC-52 in the basement membrane overlaps with that of PAT-3 integrin at the plasma membrane. Second, unc-52 and pat-3 lethal mutants both exhibit a severe Pat phenotype and have similar defects in myofilament lattice assembly (Williams and Waterston, 1994). Third, PAT-3 integrin is highly disorganized in unc-52(lethal) mutants (Hresko et al., 1994), supporting the idea that UNC-52 is required to anchor integrin. In various extracellular and cell surface proteins, RGD (Arg-Gln-Asp) sequences mediate interactions with cell-surface integrins (Ruoslahti, 1996). Murine and nematode perlecan both contain RGD sequences in domain III, whereas nematode perlecan has an additional RGD in domain IV (Rogalski et al., 1993). In contrast, human perlecan completely lacks this sequence (Kallunki and Tryggvason, 1992; Murdoch et al., 1992). Murine perlecan has been shown to support integrin-mediated cell adhesion in an RGD-dependent manner (Hayashi et al., 1992; Chakravarti et al., 1995), although a recent study found that recombinant domain III from mammalian perlecan is not sufficient to bind integrin (Schulze et al., 1996). Our results imply that domain IV is specifically required for myofilament lattice assembly in C. elegans. One of the NCAM repeats in domain IV, IgR12, has an RGD sequence that could mediate interaction with integrin; however, no direct evidence for an interaction between this domain and integrin has been demonstrated to date.

In contrast, domain V is not essential for integrin anchorage and myofilament lattice assembly. This domain is similar to the globular G-domain of 1 laminin and the C-terminal region of agrin (Noonan et al., 1991; Kallunki and Tryggvason, 1992; Murdoch et al., 1992; Patthy and Nikolics, 1994). The G-domain of 1 laminin is important for cell adhesion, including myoblast adhesion and neurite outgrowth, and has been shown to bind -integrin (Sonnenberg et al., 1990; Skubitz et al., 1991; Yurchenco et al., 1993). Consequently, domain V could interact directly with cell-surface components such as integrin. In the context of myofilament lattice assembly in C. elegans, such interactions are clearly dispensable; however, we do not exclude the possibility that domain V has a role in other developmental processes.

A Subset of UNC-52 Isoforms Are Associated with the Pharynx and Anal Muscles during Embryogenesis but Are Not Essential for Myofilament Assembly

In this study, we found that S isoforms of UNC-52 are associated with the pharynx and the anal sphincter and depressor muscles during embryogenesis; however, the role of these isoforms in pharyngeal or anal muscle development is not clear because we found no evidence of pharyngeal disruption in unc-52(null) mutants. Phalloidin staining established that the absence of UNC-52 has no discernible effect on the assembly of thin filaments in the pharynx. Similarly, preliminary studies suggest that the pharyngeal muscle-specific myosin, MHC C, assembles into well ordered A-bands in these mutants (Mullen, unpublished results).

Body-wall and pharyngeal muscle cells are known to express distinct sets of muscle proteins. For example, myosin heavy chains A and B are expressed in the body-wall muscles, whereas C and D are expressed in the pharyngeal muscles (MacKenzie et al., 1978a; Ardizzi and Epstein, 1987). In this study, we describe the differential expression and localization of UNC-52 isoforms in the body-wall and pharyngeal muscles. Our results also emphasize the differences between body-wall and pharyngeal myofilament assembly. UNC-52 is clearly important for myofilament assembly in body-wall muscles but is dispensable for assembly in pharyngeal muscles. Similarly, vinculin, which is abundantly expressed in both body-wall and pharyngeal muscles (Francis and Waterston, 1985), is not essential for myofilament assembly in pharyngeal muscles (Mullen, unpublished results). The implication is that body-wall and pharyngeal muscles require different proteins for myofilament lattice assembly, even in cases where the same or similar proteins are expressed in both muscle types.

Evidence for a Temporal Shift between Early and Late UNC-52 Isoforms

Domain IV of perlecan is composed of Ig repeats similar to those of the NCAM. The number of NCAM repeats in this domain varies between species; mouse perlecan has 14 repeats, nematode perlecan has 15 repeats, and human perlecan has 22 repeats (Noonan et al., 1991; Kallunki and Tryggvason, 1992; Murdoch et al., 1992; Rogalski et al., 1993). This repetitive structure is typical of ECM components and transmembrane proteins with large extracellular domains (reviewed in Vaugh and Bjorkman, 1996). In human perlecan, as in UNC-52, the NCAM repeats are encoded by multiple exons in a manner compatible with different combinatorial possibilities of expression (Kallunki and Tryggvason, 1992; Murdoch et al., 1992; Cohen et al., 1993). There are data indicating that alternative splicing occurs in domain IV of mouse perlecan, but, as yet, no data confirming splice variants of human perlecan (Noonan and Hassell, 1993; Iozzo et al., 1994).

Alternative splicing between exons 15 and 19 regulates the number of NCAM repeats within domain IV of UNC-52 (Rogalski et al., 1993, 1995) and is associated with both temporal and spatial differences in isoform expression. Our results imply that there is a transition or "shift" between early and late isoforms of UNC-52. This shift in isoform expression may be regulated through mec-8. Our observations on mec-8; unc-52 double mutants also imply that alternative splicing within domain IV is associated with spatial differences in isoform expression within the body-wall muscles. Several recent studies suggest that body-wall muscles do not uniformly express the same set of genes. The unc-129 gene, which encodes a member of the TGF- family, is expressed in motor neurons and the dorsal body-wall muscles (Colavita et al., 1998). In addition, the homeotic gene mab-5 is expressed in a subset of the posterior body wall muscles (Wang et al., 1993). These observations suggest that body-wall muscle cells in different locations express distinct sets of regulatory and structural genes (reviewed in Moerman and Fire, 1997). The spatially restricted expression of certain UNC-52 isoforms implies that they are specifically required by muscle cells in these locations. The distinct arrangement of muscle and hypodermal cells in the head of the animal may necessitate a specific UNC-52 isoform for myofilament assembly or cell adhesion.

Interestingly, alternative splicing within domain IV of UNC-52 is dispensable. Considering that these splicing events are highly regulated both temporally and spatially, this observation was surprising. A deletion identified in this study eliminates all variants except the 15-19 splice product and does not noticeably affect muscle development. This observation suggests that alternative splicing within this region is associated with fine modulation of function rather than large-scale changes in biophysical properties. In contrast, alternative splicing events that give rise to the three major groups of isoforms are likely to significantly change the properties of these proteins, including their ability to interact with transmembrane receptors such as integrin.

Of particular importance in this study of perlecan function is our demonstration that a domain IV-containing isoform, not a domain V-containing isoform, is sufficient for myofilament organization and attachment. If perlecan/UNC-52 serves as an adhesive substrate for anchoring the myofilament lattice, then these results help to identify regions of UNC-52 that are critical for integrin anchorage and myofilament lattice assembly. In summary, this study demonstrates some of the functional complexity and structural plasticity of nematode perlecan. We suggest that the redundancy observed with perlecan may be an inherent property of large ECM proteins.