|Home | About | Journals | Submit | Contact Us | Français|
The ubiquitin proteasome system is involved in degradation of old or damaged sarcomeric proteins. Most E3 ubiquitin ligases are associated with cullins, which function as scaffolds for assembly of the protein degradation machinery. Cullin 3 uses an adaptor to link to substrates; in Caenorhabditis elegans, one of these adaptors is the BTB-domain protein MEL-26 (maternal effect lethal). Here we show that MEL-26 interacts with the giant sarcomeric protein UNC-89 (obscurin). MEL-26 and UNC-89 partially colocalize at sarcomeric M-lines. Loss of function or gain of function of mel-26 results in disorganization of myosin thick filaments similar to that found in unc-89 mutants. It had been reported that in early C. elegans embryos, a target of the CUL-3/MEL-26 ubiquitylation complex is the microtubule-severing enzyme katanin (MEI-1). Loss of function or gain of function of mei-1 also results in disorganization of thick filaments similar to unc-89 mutants. Genetic data indicate that at least some of the mel-26 loss-of-function phenotype in muscle can be attributed to increased microtubule-severing activity of MEI-1. The level of MEI-1 protein is reduced in an unc-89 mutant, suggesting that the normal role of UNC-89 is to inhibit the CUL-3/MEL-26 complex toward MEI-1.
In muscle, the overall mass and number of sarcomeres are maintained by a fine balance between removal and degradation of old or damaged proteins and replacement by newly synthesized proteins. During muscle atrophy in humans, this balance is disrupted, with a net loss of sarcomeres and muscle mass. Muscle atrophy is attributable to muscle disuse (immobilization, microgravity of space travel) or starvation and occurs in many chronic diseases (renal failure, diabetes, cancer cachexia, sepsis, and burn injury) and also in elderly patients without underlying disease (in which case it is termed sarcopenia; Murton and Greenhaff, 2009 ). During muscle atrophy, both the degradation of myofibrillar proteins is up-regulated and the synthesis of myofibrillar proteins is down-regulated.
The major system mediating protein degradation in muscle is the ubiquitin proteasome system (Ventadour and Attaix, 2006 ; Murton et al., 2008 ). The 76-residue ubiquitin is covalently attached to a substrate, and this directs it to the proteasome for degradation. Ubiquitination involves multiple enzymes, including a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin protein ligase (E3). The majority of E3s are associated with cullin “really interesting new gene” (RING) ligases (CRLs), in which a cullin acts as a scaffold for assembly of an E3 (contains a RING Zn finger domain), a protein that recognizes a substrate (a substrate recognition subunit [SRS], usually containing an F box domain [first identified in cyclin F]), and, with the exception of Cul3-type CRLs, an adaptor subunit that links the SRS to the complex (Petroski and Deshaies, 2005 ; Bosu and Kipreos, 2008 ). There are five types of cullins: Cul1–5. For Cul3 CRLs the functions of the F box protein and the adaptor subunit are served by a single protein that contains a BTB domain (a dimerization domain; named for BR-C, ttk, and bab—the first three proteins found to have this domain). The BTB domain binds directly to Cul3, and other domains in the protein (e.g., a protein–protein interaction meprin and TRAF homology [MATH] domain) bind to specific substrates. During muscle atrophy of rodents and humans, two components of muscle-specific E3s are up-regulated—muscle ring finger protein-1 (RING finger E3) and atrogin-1 (F box protein) (Bodine et al., 2001 ; Gomes et al., 2001 ). Mice knocked out for either protein are partially resistant to muscle atrophy.
The nematode Caenorhabditis elegans is a proven platform for discovery of new information about conserved sarcomere components and for discovery of new and conserved sarcomere components (Waterston, 1988 ; Moerman and Fire, 1997 ; Moerman and Williams, 2006 ; Qadota and Benian, 2010 ; Benian and Epstein, 2011 ). The C. elegans gene unc-89 (uncoordinated) encodes a set of giant polypeptides of up to 900,000 Da that are located at the sarcomeric M-line (Benian et al., 1996 ; Small et al., 2004 ; Ferrara et al., 2005 ). The human homologue of UNC-89 is called obscurin (Bang et al., 2001 ; Young et al., 2001 ; Fukuzawa et al., 2005; Kontrogianni-Konstantopoulos et al., 2009 ). Loss of function of unc-89 results in adult worms with disorganization of the myofilament lattice, usually a lack of M-lines, and with decreased locomotion (Waterston et al., 1980 ; Benian et al., 1999 ). The largest UNC-89 isoform consists of 53 Ig domains, two Fn3 domains, two protein kinase domains at its C-terminus, and SH3, DH, and PH domains at its N-terminus. The DH domain of UNC-89 has exchange activity for RHO-1 (the RhoA orthologue in C. elegans). Partial knockdown of rho-1 in adult worms results in a pattern of disorganization of myosin thick filaments similar to loss of function of unc-89 (Qadota et al., 2008a ).
The large size of UNC-89 precludes many biochemical methods for analyzing its role in sarcomere assembly and maintenance. Thus we are systematically identifying the binding partners of UNC-89, beginning with yeast two-hybrid screens, and then determining the intracellular localization and mutant phenotypes of putative interactors. We previously reported that the protein kinase domain–containing region of UNC-89 interacts with SCPL-1, a CTD-type protein phosphatase, and with LIM-9, a homologue of human FHL (Qadota et al., 2008b ; Xiong et al., 2009 ). Here we report that two portions of UNC-89 interact with MEL-26 (maternal effect lethal), a BTB-domain protein previously shown to interact with CUL-3 (cullin 3) and to be essential in early embryonic development (Dow and Mains, 1998 ; Furukawa et al., 2003 ; Pintard et al., 2003 ; Xu et al., 2003 ). By immunolocalization, we find that MEL-26 colocalizes with UNC-89 at M-lines. Of importance, we demonstrate that mutations in mel-26 and cul-3 result in disorganization of sarcomeric thick filaments similar to mutations in unc-89. MEI-1 (meiosis defective), the catalytic subunit of the nematode orthologue of the microtubule-severing complex katanin, is a known substrate for ubiquitylation by the CUL-3/MEL-26 complex in early embryos. We show that mutation in mei-1 also disrupts thick filament organization similar to unc-89 and mel-26. We provide genetic evidence that at least some of the mel-26 loss-of-function phenotype can be attributed to increased microtubule-severing activity of MEI-1. We demonstrate that the level of MEI-1 protein is reduced in an unc-89 mutant, indicating that the normal role of UNC-89 is to inhibit the CUL-3/MEL-26 complex toward MEI-1 in striated muscle. Our results suggest a novel mechanism for regulating protein degradation in muscle, and this has a role in thick-filament assembly and/or maintenance.
As part of our systematic search for binding partners for the giant modular protein UNC-89, we used a five-domain segment of UNC-89, Ig1–Ig5, as bait to screen a library of C. elegans cDNA using the yeast two-hybrid system (Xiong, Qadota, and Benian, unpublished data). One of the preys was found to encode MEL-26. We wanted to determine whether MEL-26 interacted with any other regions of UNC-89. Thus we examined a “bookshelf” containing 16 overlapping yeast two-hybrid bait plasmids covering the whole coding sequence of UNC-89. In addition to Ig1–Ig5, MEL-26 also interacted with a region between the two kinase domains of UNC-89, including approximately one third of the interkinase region (IK), Ig53, and Fn2 domains (Figure 1A). Using yeast-two hybrid assays, we identified the minimal portions of each UNC-89 fragment that are necessary and sufficient to interact with MEL-26. As shown in Figure 1, B and andC,C, Ig2–Ig3 and Ig53–Fn2, are minimally required to interact with MEL-26. It is worth noting that MEL-26 interacts with a different minimal region of UNC-89 than does CPNA-1, the other protein identified in the original screen (Xiong et al., unpublished data), which interacts minimally with UNC-89 Ig1–Ig2–Ig3. Despite the multitude of other Ig domains present in UNC-89, MEL-26 only interacts with UNC-89 Ig2–Ig3 and Ig53–Fn2, indicating specificity.
MEL-26 is part of a CRL complex and acts as an adaptor to bring together the substrate and CUL-3 (Cul3) for ubiquitination (Furukawa et al., 2003 ; Pintard et al., 2003 ; Xu et al., 2003 ). MEL-26 contains an N-terminal MATH domain and a C-terminal BTB domain. It has been reported that the BTB domain of MEL-26 interacts with CUL-3, and that the MATH domain of MEL-26 interacts with a substrate, MEI-1 (Luke-Glaser et al., 2007b ). Using two-hybrid assays, we verified that the MATH domain interacts with MEI-1 and that the BTB domain interacts with CUL-3 (Figure 1D). In addition, we confirmed the yeast two-hybrid interactions between UNC-89 and MEL-26 using far Western assays with purified proteins (Machida and Mayer, 2009 ). UNC-89 Ig2–Ig3 and UNC-89 Ig53–Fn2 each interacts with MEL-26 (Figure 2A). We also found that both UNC-89 fragments interact with the MATH domain but not the BTB domain of MEL-26 (Figure 2B). Thus, portions of UNC-89 interact with the same domain of MEL-26 as MEI-1 (Figure 2D).
We next reasoned that if UNC-89 and MEL-26 interact in vivo, MEL-26 should colocalize with UNC-89 in the sarcomere. Antibodies to MEL-26 (kindly provided by Lionel Pintard, University of Paris, Paris, France) were used to immunostain whole fixed adult nematodes together with marker antibodies for sarcomeric M-lines (UNC-89) and dense bodies (α-actinin). As shown in Figure 3, anti–MEL-26 localizes in a striated pattern to both M-lines and I-bands (and probably dense bodies). Colocalization of MEL-26 with UNC-89 at M-lines is consistent with MEL-26 and UNC-89 interacting in vivo. Of interest, sarcomeric localization of MEL-26 in adult muscle is in contrast to the diffuse localization of MEL-26 reported in early-stage embryos (Luke-Glaser et al., 2007a ). To our knowledge, this is the first time that a component of a cullin complex has been localized in the sarcomere.
Loss of function for unc-89 results in adult worms having disorganized sarcomeres, especially in the A-band, and usually they have no M-lines (Waterston et al., 1980 ; Benian et al., 1999 ). We showed previously that one of the most severe alleles of unc-89, su75, lacks expression of all the giant isoforms of UNC-89 (Small et al., 2004 ), including one region, Ig2–Ig3, that we showed interacts with MEL-26. In addition, unc-89(su75), when immunostained with antibodies to myosin A (MHC A) or visualized with green fluorescent protein (GFP)::MHC A, displays disorganization of thick filaments, including what appear to be abnormal aggregates of myosin, and V-shaped crossing of A-bands (Qadota et al., 2008a ). Thus, we wondered whether loss of function of either mel-26 or cul-3 might show a similar disorganization of thick filaments. We used RNA interference (RNAi) to knock down either mel-26 or cul-3, beginning at the L1 larval stage, to avoid embryonic lethality in a strain expressing GFP::MHC A. As shown in Figure 4A, knockdown of either mel-26 or cul-3 resulted in adults having body-wall muscle with disorganized thick filaments, in a pattern similar to that of unc-89(su75). In fact, cul-3(RNAi) consistently resulted in thick-filament disorganization more severe than that found in unc-89(su75) or mel-26(RNAi). These results suggest that in striated muscle, CUL-3 may act through MEL-26 and additional BTB-containing adaptor proteins. Alternatively, RNAi might be more effective for cul-3 than it is for mel-26.
To confirm the RNAi results for mel-26, we investigated the adult muscle phenotype of an allele of mel-26, ct61sb4. Although null, this allele is temperature sensitive (Lu and Mains, 2007 ), and so embryonic lethality can be bypassed at 15oC. We stained these mel-26 mutant adult worms with anti-MHC A. As shown in Figure 4B, mel-26(ct61sb4) unc-29(e1072), but not unc-29(e1072) (a linked marker used for strain construction; see Materials and Methods), shows disorganization of thick filaments similar to that observed by RNAi for mel-26.
For most genes, loss of function and gain of function will yield different phenotypes. Thus we wondered how the gain of function of mel-26 might compare to the loss of function of mel-26. To examine this, we expressed hemagglutinin (HA)-tagged MEL-26 from a heat shock promoter in adult worms. As shown in Figure 4C, HA-MEL-26 is expressed in nematodes only after heat shock. Unexpectedly, in muscle cells that overexpress MEL-26 (Figure 4D), thick filaments are disorganized in a way that is very similar to loss of function of mel-26 by RNAi or by mutation. Heat shock itself had no effect on the organization of thick filaments (Figure 4E).
MEI-1 is the C. elegans orthologue of the microtubule-severing enzyme katanin (Srayko et al., 2000 ; Roll-Mecak and McNally, 2010 ) and has been identified as a target of the MEL-26/CUL-3 ubiquitylation complex in C. elegans embryos (Furukawa et al., 2003 ; Pintard et al., 2003 ; Xu et al., 2003 ). MEI-1 has a well-established function in meiotic spindle formation of newly fertilized C. elegans embryos (Srayko et al., 2000 , 2006 ; McNally et al., 2006 ; McNally and McNally, 2011 ). Low levels of mei-1 mRNA have been reported at other stages (Clark-Maguire and Mains, 1994a ). We sought evidence that MEI-1 is expressed in adult somatic cells, which include muscle. For this, we used a temperature-sensitive mutant in glp-4 (germline proliferation defective). glp-4(bn2ts) mutants raised at the restrictive temperature of 25oC have only a few germ cell precursors, all arrested at the prophase of the mitotic cell cycle, with none entering meiosis (Beanan and Strome, 1992 ). As shown in Figure 5A, by immunoblot MEI-1 can be detected from these adults that lack a germline. Consistent with this somatic expression, Serial Analysis of Gene Expression (SAGE data), available at WormBase (www.wormbase.org), indicate that in wild-type nematodes, mei-1 mRNA is expressed in multiple somatic cell types, including body-wall muscle cells, hypodermal cells, pharyngeal marginal cells, gut, and several types of neurons. Moreover, SAGE shows that mei-1 mRNA is also expressed in glp-4 mutant adults.
To determine whether MEI-1 is the relevant target of MEL-26 in muscle, we examined the status of thick filaments in the striated muscle of adults in the mei-1 gain-of-function allele, ct46, which encodes a protein that does not bind MEL-26 and so is resistant to MEL-26 mediated degradation (Clark-Maguire and Mains, 1994b ; Pintard et al., 2003 ; Xu et al., 2003 ). This should correspond to the loss of mel-26. As shown in Figure 5B, mei-1(ct46) also shows disorganization of thick filaments, similar to loss of function of mel-26. This indicates that MEI-1 is a relevant target for the mel-26 muscle phenotype.
To further explore the function of MEI-1 in striated muscle, we knocked down mei-1 using RNAi feeding beginning from the L1 larval stage to bypass embryonic lethality. The resulting adults were immunostained with anti–MHC A, and the results are shown in Figure 5, C and andD.D. Although there was some variation among animals in the severity of the phenotype, nearly all showed body-wall muscle cells in which thick filaments were disorganized. Cells showing the strongest effect resembled unc-89 and mel-26 mutants (compare bottom three panels of Figure 5C with Figure 4, A and B). To confirm the RNAi results for mei-1, we investigated the adult muscle phenotype of a null allele of mei-1, ct46ct101 (Clark-Maguire and Mains, 1994a ). We stained mei-1-mutant adult worms with anti–MHC A. As shown in Figure 5E, mei-1(ct46ct101) unc-29(e1072), but not unc-29(e1072), shows disorganization of thick filaments similar to that observed by RNAi for mei-1.
Therefore, mei-1 gain of function and loss of function show similar phenotypes, as did gain and loss of function of mel-26, suggesting that tight regulation of the level of MEI-1 and MEL-26 is required for normal organization of thick filaments. An additional aspect of the phenotype can be noted: even in muscle cells showing a mild effect on thick-filament organization, the muscle cells appear shorter and broader than the narrower, spindle-shaped cells in wild-type muscle (Figure 5D). This effect on cell shape might be attributable to a disruption in the organization of microtubules that might result from the loss of MEI-1 or katanin activity.
Although MEI-1 has established microtubule-severing activity in C. elegans embryos, we sought genetic evidence for this function in body-wall muscle. To do this, we used nematode strains in which β-tubulin (TBB-2) contains a missense mutation that makes microtubules partially resistant to MEI-1 severing (Lu et al., 2004 ). As shown in Figure 6 (bottom), this mutation, tbb-2(sb26), by itself has no effect on the organization of thick filaments. However, the double mutant mel-26(ct61sb4); tbb-2(sb26) (Figure 6, middle) is partially suppressed as compared with mel-26(ct61sb4). This indicates that at least some of the mel-26 loss-of-function muscle phenotype can be attributed to increased microtubule-severing activity of MEI-1.
The fact that loss of function of cul-3 or mel-26 yields a similar phenotype to a gain-of-function allele of mei-1 is consistent with the known biochemical function of the MEL-26/CUL-3 complex to degrade MEI-1 in early embryos (Furukawa et al., 2003 ; Pintard et al., 2003 ; Xu et al., 2003 ). Therefore we hypothesize that the function of the UNC-89–MEL-26 interaction in adult muscle is to influence the activity of the MEL-26/CUL-3 complex in promoting the degradation of MEI-1. As a test of this hypothesis, we compared the level of MEI-1 protein in wild type versus unc-89(su75). As noted earlier, unc-89(su75) lacks all UNC-89 isoforms that contain one of two MEL-26 binding sites. As shown in Figure 7A and quantitatively displayed in Figure 7C, unc-89(su75) shows a moderate reduction in the level of MEI-1 proteins. This is in contrast to the levels of PAT-6 (Figure 7A) or MHC A (Figure 7B), which show no difference between wild type and unc-89(su75). PAT-6 is the nematode orthologue of actopaxin or α-parvin (Lin et al., 2003 ) and is located at the base of muscle focal adhesions (M-lines and dense bodies). That we observed only a 20% reduction in the level of MEI-1 in unc-89(su75) might be explained by the facts that the animals analyzed were gravid adults and MEI-1 is expressed in early embryos and in multiple adult somatic tissues, whereas UNC-89 is almost exclusively expressed in body-wall and pharyngeal muscle. In addition, in su75 animals the small UNC-89 isoforms C and D, which contain the second MEL-26–binding site, Ig53 and Fn2, are still expressed and, indeed, overexpressed as compared with wild type (Small et al., 2004 ). However, RNAi of the interkinase region in the unc-89(su75) strain did not show a greater decrease of MEI-1 than unc-89(su75) alone, despite the fact that this RNAi of wild type resulted in thick-filament disorganization similar to that found in unc-89(tm752) (unpublished data), which shows a lack of expression of UNC-89-C and -D (Ferrara et al., 2005 ). Nevertheless, a reduced level of MEI-1 in unc-89(su75) is consistent with a model in which the interaction of UNC-89 with MEL-26 normally functions to inhibit the activity of the CUL-3/MEL-26 complex in degrading muscle MEI-1 (see Figure 9 later in the paper).
Although MEI-1 has been shown to be ubiquitinated by CUL-3 in vitro (Furukawa et al., 2003 ), it has not been reported whether this results in proteasomal degradation. RNAi against the proteasomal subunit RPT-2 (26S proteasome regulatory subunit 4) has been used to inhibit C. elegans proteasome function (Landsverk et al., 2007 ). We performed RNAi against rpt-2, beginning at the L1 stage to avoid embryonic lethality, and prepared a protein extract. As compared with nematodes fed the empty RNAi vector, rpt-2(RNAi) led to an increased level of MEI-1 by Western blot (Figure 7, D and E). This result suggests that ubiquitination of MEI-1 leads to proteasomal degradation of MEI-1 in vivo.
Because loss of function of unc-89 results in a lower level of MEI-1 (Figure 7), this suggests that normally UNC-89 inhibits MEL-26 function. Given that UNC-89 and MEI-1 bind to the same domain of MEL-26 (the MATH domain; Figures 1D and and2B),2B), one hypothesis is that UNC-89 and MEI-1 compete for binding to MEL-26. To test this possibility, we conducted the following experiment. Anti-HA beads were used to immunoprecipitate HA-MEL-26 (driven by a heat shock promoter) from lysates and incubated with no additional protein, maltose-binding protein (MBP), MBP-UNC-89 Ig2–Ig3, or MBP-UNC-89 Ig53–Fn2. As shown in Figure 8, although HA-MEL-26 coprecipitated native MEI-1, the presence of these additional proteins had no effect on this MEL-26–MEI-1 interaction. In fact, the HA-MEL-26 also interacted with the added MBP-Ig2-Ig3 and MBP-Ig53-Fn2 but not MBP. (There was some binding of MBP-Ig2-Ig3 and MBP-Ig53-Fn2 to the anti-HA beads, but this was considerably less than binding to HA-MEL-26–coated anti-HA beads.) These results indicate that either MEI-1 and UNC-89 bind nonoverlapping regions of MEL-26 MATH domain or MEI-1 binding is much stronger than that of UNC-89. Therefore, although we could find no evidence for competition in binding for MEL-26, MEL-26 activity may be influenced by interaction with UNC-89 in other ways (Figure 9; see Discussion).
In our quest to identify binding partners for the giant sarcomeric protein UNC-89 (obscurin), we found that two portions of UNC-89 interact with the MEL-26, a BTB–domain containing substrate recognition protein known to interact with CUL-3 (Cul3)-based E3 ubiquitin ligase. Furthermore, we found that interaction of UNC-89 with MEL-26 occurs through the MATH domain of MEL-26 (Figures 1 and and2),2), the same domain of MEL-26 that had been identified previously as interacting with a known substrate of the CUL-3/MEL-26 complex in C. elegans embryos, the MEI-1 microtubule-severing protein (katanin). Using antibodies, we found that muscle MEL-26 is localized to M-lines, colocalizing with UNC-89, and to I-bands (Figure 3). To our knowledge this is the first time that a cullin-associated protein or a cullin has been localized in the sarcomere. Loss of function of unc-89 results in a characteristic pattern of disorganization of muscle thick filaments, including myosin aggregates, which suggests a role for unc-89 in thick-filament assembly and/or maintenance. We found that loss or gain of function of mel-26 or loss of function or gain of function of mei-1 results in thick-filament disorganization similar to that of unc-89 loss of function (Figures 4 and and5).5). Because a mutant β-tubulin that is resistant to cleavage by MEI-1 (katanin) can partly suppress the thick-filament disorganization of a mel-26 mutant (Figure 6), at least one function of MEI-1 in muscle is to sever microtubules. We observed that loss of function of unc-89 results in a reduction in the level of MEI-1 protein (Figure 7, A–C). MEI-1, previously reported to be ubiquitinated in vitro, is likely to be degraded in the proteasome based on inhibiting proteasome function by rpt-2(RNAi) (Figure 7, D and E). However, compensating for decreased levels of MEI-1 caused by unc-89 by inhibiting the proteasome did not improve the organization of thick filaments in unc-89(su75) versus unc-89(su75); rpt-2(RNAi) (unpublished data). This is consistent with our observations that rpt-2(RNAi) in the unc-89-mutant background still resulted in a rise of MEI-1 beyond wild-type levels and that gain of function of MEI-1 (Figure 5B) causes disorganization of thick filaments, as does MEI-1 loss of function.
To account for our findings, we propose a model in which the CUL-3/MEL-26 complex is localized to the M-line through the interaction of UNC-89 with MEL-26, and the function of this interaction is to inhibit ubiquitin-mediated degradation of MEI-1 (Figure 9). Given that both UNC-89 and MEI-1 bind to the MATH domain of MEL-26, we tested the hypothesis that UNC-89 and MEI-1 compete for binding to MEL-26. However, we could not demonstrate release of MEI-1 from MEL-26 in the presence of excess UNC-89 fragments (Figure 8). We could show that both MEI-1 and UNC-89 bound to MEL-26, but we do not know whether a single MEL-26 molecule can simultaneously bind to both MEI-1 and UNC-89 (e.g., using opposite surfaces of the MATH domain). Nevertheless, although there may not be direct competition for binding, the interaction of UNC-89 influences the activity of the CUL-3/MEL-26 complex toward MEI-1 and, given our data, most likely by inhibiting its activity. At the very least, the interaction of UNC-89 with MEL-26 localizes the activity of the CUL-3/MEL-26 complex at the sarcomeric M-line.
If our model that UNC-89 inhibits the complex is correct, then normally the activity of MEL-26 is reduced and the protein level of MEI-1 is elevated. Thus the similarity in phenotypes for loss of function of unc-89, for gain of function of mel-26, and for loss of function of mei-1 is consistent with the model. The fact that we observed that both gain and loss of function of mei-1 are also similar to loss of function of unc-89 is more difficult to explain. Perhaps proper thick-filament assembly and/or maintenance depend on tightly controlled activities of mei-1, restricting it to a certain range of activity. This idea is not unprecedented: proper thick-filament assembly/maintenance is known to depend on precise levels of UNC-96 and UNC-45, as loss or gain of function of each gene results in similar effects on thick-filament organization (Barral et al., 1998 ; Mercer et al., 2006 ; Landsverk et al., 2007 ; Qadota et al., 2007 ). During meiosis, mei-1 activity must also be kept within a specific range. Increased or decreased mei-1 activity results in defective meiotic spindles, too short for increased activity or too long for decreased activity. In both cases, these abnormal spindles missegregate chromosomes, resulting in aneuploid gametes (Johnson et al., 2009 ).
We also found that loss of function of cul-3 results in a somewhat more severe disorganization of thick filaments than unc-89, mel-26 or mei-1 (Figure 4A). This perhaps suggests that in body-wall muscle, CUL-3 can associate with additional BTB adaptor proteins besides MEL-26. Indeed, our analysis reveals that the C. elegans genome contains 49 BTB domain–containing proteins. By SAGE data (on WormBase), 20 of the 49 are expressed in body-wall muscle, four of these probably at a high level of expression (including mel-26).
The ability of tbb-2(sb26) to partly suppress the thick-filament disorganization of mel-26 is consistent with β-tubulin being at least one substrate for MEI-1 in muscle. However, since the phenotype is not entirely suppressed, there is the possibility that there may be additional substrates for MEI-1 muscle. Indeed, in the embryo the preference of MEI-1 for the TBB-2 β-tubulin isotype is not absolute (Lu et al., 2004 ) and muscle may express additional tubulin isotypes. In addition, other MEI-1 functions, such as microtubule bundling (McNally and McNally, 2011 ), might be relevant in muscle. Nevertheless, we observed an additional aspect of the mei-1(RNAi) phenotype that is consistent with a role for microtubules in striated muscle: the spindle-shaped body-wall muscle cells are shorter in two dimensions. Disruption of the microtubule network in these cells may lead to a change in cell shape either during development or as a result of hypercontraction.
Little is known about the potential role of microtubules in muscle cell shape or sarcomeric organization in C. elegans. To our knowledge, the localization of microtubules in nematode body-wall muscle has not been reported. Nevertheless, the C. elegans echinoderm microtubule-associated protein–like protein ELP-1, which binds to microtubules in vitro, is localized in a criss-crossing network resembling microtubules, and the network is sensitive to the microtubule disruptor nocodazole (Hueston et al., 2008 ). Of interest, RNAi-mediated knockdown of elp-1 in a dystrophin (dys-1)-null animal results in adult worms that have hypercontracted muscle cells (Hueston and Suprenant, 2009 ), similar to what we observed for mei-1(RNAi). In mammals, electron microscopy of cardiac muscle revealed a fairly extensive network of microtubules; microtubules surround myofibrils and sarcoplasmic reticulum in a helical arrangement (Goldstein and Entman, 1979 ). In myotubes induced to regenerate their myofibrils, in the presence of the microtubule-stabilizing drug Taxol, A-bands are formed, but thin filaments and Z-bands are not formed; in the presence of the microtubule-depolymerizing drug Colcemid, complete sarcomeres are formed, but they are not laterally aligned (Toyama et al., 1982 ). Generation of a stable, postranslationally modified microtubule array is an early event in the differentiation of myotubes (Gundersen et al., 1989 ). During cardiac hypertrophy, there is an increase in microtubule network density associated with sarcomere dysfunction (Tsutsui et al., 1993 ).
The results presented here are not the first to implicate the ubiquitin proteasome system in controlled degradation of sarcomeric proteins in C. elegans. The RING finger protein RNF-5, not known to be associated with cullins, is localized to dense bodies (Z-disks) and regulates the levels of the LIM domain protein UNC-95 (Broday et al., 2004 ). A report from our laboratory demonstrated a connection to CRLs. We showed that CSN-5 interacts with two sarcomeric M-line proteins, UNC-98 and UNC-96 (Miller et al., 2009 ). CSN-5 is one of eight highly conserved subunits of the COP9 signalosome complex, which regulates CRLs in a complicated manner, primarily through deneddylation (Cope and Deshaies, 2003 ; Schwechheimer, 2004 ). Antibodies to CSN-5 localize the protein to A-bands in wild type and colocalize with abnormal accumulations of paramyosin found in unc-98, unc-96, and unc-15 (paramyosin) mutants. Knockdown of csn-5 results in an increase in the level of UNC-98 protein and a slight reduction in the level of UNC-96 protein, suggesting that normally CSN-5 promotes the degradation of UNC-98 and stabilizes UNC-96.
Cullins and BTB-domain proteins are associated with human diseases. Litterman et al. (2011) reported that in humans, obscurin-like 1 (OBSL1) directly interacts with Cul7, and both proteins are crucial for Golgi morphogenesis and dendrite growth of neurons. Cul7 and OBSL1 also cause human 3M syndrome, an autosomal recessive disorder characterized by growth retardation, characteristic facial features, and skeletal anomalies. The majority of patients have mutations in Cul7, whereas others have null mutations in OBSL1 (Hanson et al., 2009 ). Cirak et al. (2010) reported a family in which early-onset autosomal-dominant distal myopathy is associated with a heterozygous missense mutation in Kelch-like homologue 9 (KLHL9), which is a substrate recognition protein that interacts with Cul3. Of interest, the mutation, L95F, is located in the conserved BTB domain of KLHL9, which mediates interaction with Cul3.
Finally, in Lange et al. (2012) , our colleagues studying obscurin in the mammalian heart show that the turnover of the small ankyrin protein sAnk1.5, previously known to interact with obscurin, is regulated by ubiquitylation mediated by the BTB-domain protein KCTD6. KCTD6, like MEL-26, is a substrate recognition protein for Cul3. Furthermore, these authors demonstrate that in the absence of obscurin, degradation of sAnk1.5 is increased, and conversely, RNAi-mediated knockdown of KCTD6 results in increased levels of sAnk1.5. Thus, although the ultimate substrates identified by our study and that of Lange et al. are different—MEI-1 and sAnk1.5, respectively—the mechanism is similar: apparent inhibition of a Cul3 complex in striated muscle by UNC-89 (obscurin).
To construct plasmids for yeast two-hybrid assays and other purposes, we amplified the various portions of UNC-89, MEL-26, CUL-3, and MEI-1, as depicted in Figure 1, as well as RPT-2 (boundaries of constructs, primer sequences, and enzymes used are shown in Supplemental Tables S1 and S2). After cloning into pBluescript and verifying error-free sequences, we subcloned the fragments into the bait vector pGBDU-C1 and/or the prey vector pGAD-C1 (James et al., 1996 ). In addition, the DNA fragments were subcloned into pGEX-KK1 and/or pMAL-KK1 (kindly provided by Kozo Kaibuchi, Nagoya University, Nagoya, Japan) for the purification of glutathione S-transferase (GST)- or MBP-tagged proteins, respectively. Similarly, sequences were subcloned into pPD129.36 for RNAi or into pKS-HA8(NheX2) to add HA tag to the N-terminus and then into nematode heat shock promoter vectors pPD49.78 and pPD49.83 (gifts from Andrew Fire, Stanford University, Stanford, CA).
Yeast two-hybrid assays were performed as previously described (Mackinnon et al., 2002 ). Briefly, PJ69-4A strains carrying pGBDU bait constructs were transformed with prey clones (using either pACT or pGAD vectors). Assays were scored for growth on media lacking either histidine or adenine. MEL-26 was used as bait to screen a collection (“bookshelf”) of 16 overlapping sequences that cover all of the coding sequence of the largest isoform of UNC-89 (Xiong et al., unpublished data). Deletion derivatives of UNC-89 Ig1–5, IK–Fn2, and MEL-26, as depicted in Figure 1 and Supplemental Table S2, were tested by two-hybrid assay in a similar manner.
Bacterially expressed GST- and MBP-tagged proteins were purified as previously described (Mercer et al., 2006 ). Far Western assays were used to verify interactions using purified proteins as previously described (Qadota et al., 2007 ). Briefly, 2 μg of GST or GST-tagged proteins were resolved by SDS–PAGE and electroblotted onto nitrocellulose. Blots were then reacted with 5 µg/ml MBP or MBP-tagged proteins. MBP binding was detected by a 1:5000 dilution of anti-MBP (mouse monoclonal)–horseradish peroxidase (HRP; E8038S; New England BioLabs, Ipswich, MA) and visualized using enhanced chemiluminescence (ECL; 32106; Pierce, Thermo Fisher Scientific, Rockford, IL).
The wild-type strain, Bristol N2, strain RW1596 (see later discussion), HE75 (unc-89(su75), originally isolated and provided by Henry Epstein, University of Texas Medical Branch at Galveston, Galveston, TX; Small et al., 2004 ), CB1072 (unc-29(e1072); Lewis et al., 1980 ), HR1060 (tbb-2(sb26); Lu et al., 2004 ), and HR311 (mei-1(ct46ct101) unc-29(e1072)/hT2 [bli-4(e937) let(h661)] (I); +/hT2 III; Johnson et al., 2009 ) were grown at 20°C on nematode growth media (NGM) plates with Escherichia coli strain OP50 as food source (Brenner, 1974 ). The temperature-sensitive strains HR1329 (mel-26(ct61sb4) unc-29(e1072); Lu and Mains, 2007 ), BW729 (mei-1(ct46) unc-29(e1072); Clark-Maguire and Mains, 1994b ), SS104 (glp-4(bn2ts); Beanan and Strome, 1992 ), and HR1065 (mel-26 (ct61sb4) unc-29(e1072); tbb-2(sb26); Lu et al., 2004 ) were grown at 15°C. RW1596 is a myo-3 mutant rescued by a transgenic array containing copies of the wild-type myo-3 gene (encodes MHC A) translationally fused to GFP (Campagnola et al., 2002 ) and provided by Pamela Hoppe (Western Michigan University, Kalamazoo, MI).
Because mel-26 and mei-1 do not have easily scorable adult phenotypes (other than by progeny testing), the closely linked marker gene on chromosome I, unc-29, which has a recognizable adult motility defect, was used to follow the presence of the mutant genes through genetic crosses (unc-29 shows <1% recombination with either gene). unc-29 encodes a non-α subunit of the nicotinic acetylcholine receptor superfamily and is expressed in body-wall muscle (Fleming et al., 1997 ), but, as shown in Figures 4B and 5, B and E), loss of function does not affect sarcomeric structure.
Worm fixation by the Nonet method and worm immunostaining were as previously described (Wilson et al., 2012 ). In the case of mei-1(ct46ct101), to collect homozygous adult worms, we used strain HR311 (in which mei-1 is genetically marked by the closely linked unc-29(e1072)); we picked 200 Unc-29 worms (identified by reduced forward movement) for fixation. Primary antibodies were used at the following dilutions: 1:100 rabbit anti–MEL-26 (affinity purified; a generous gift from Lionel Pintard, University of Paris; Luke-Glaser et al., 2007a ), 1:200 mouse anti–UNC-89 (MH42; monoclonal, ascites) from Benian et al., 1996 ), 1:200 mouse anti–α-actinin (MH35; monoclonal, ascites) from Francis and Waterston (1985) ), 1:200 mouse anti–myosin heavy chain A (MHC A) (5-6; monoclonal, tissue culture supernatant) from Miller et al. (1983 ), and 1:200 mouse anti-HA monoclonal antibodies (H3663; Sigma-Aldrich, St. Louis, MO). Secondary antibodies and confocal microscopy were as described (Qadota et al., 2007 ). The color balances of the images were adjusted with Photoshop (Adobe, San Jose, CA).
RNAi of mel-26, cul-3, or mei-1 was performed by feeding the specified worm strain bacteria expressing double-stranded RNA beginning at the L1 larval stage and continuing until the animals reached adulthood (Miller et al., 2006 ). L1 larvae were isolated by performing a bleach/NaOH digestion of an unsynchronized worm culture resulting in only viable embryos. Embryos were plated on unseeded NGM plates overnight, allowing the L1 larvae to hatch but not grow. These synchronized L1 larvae were collected and distributed onto plates containing bacteria expressing double-stranded RNA. Plates were incubated for 2 d, and then adults were either examined for localization of GFP-MHC A by confocal microscopy or were fixed, immunostained with anti-MHC A, and examined by confocal microscopy. To examine the effect on the level of MEI-1 by knocking down a component of the proteasome, RPT-2, we conducted RNAi for rpt-2 beginning at the L1 stage, as described earlier. In this case, embryos were harvested from a heavy growth of N2 from 10 10-cm plates and allowed to hatch into L1, and these L1 were grown into adults on 50 10-cm plates before harvesting for Western blot analysis.
Plasmids capable of expressing HA-tagged MEL-26 under the control of a heat shock promoter (described earlier) were coinjected with the SUR-5::NLS::GFP transformation marker plasmid pTG96 (Yochem et al., 1998 ). F1 worms with GFP expression were cloned, and two lines were established (both lines behaved similarly; unpublished data). HA-MEL-26 expression was examined using worms with and without a 2-h heat shock at 30°C, preparing a worm lysate (Hannak et al., 2002 ), and assaying by Western blot using rabbit anti-HA (H6908; Sigma-Aldrich). Worms were also treated with and without a 2-h heat shock at 30°C, allowed to recover at 20°C overnight, and then fixed and immunostained with anti-MHC A, as described previously.
We used the procedure of Hannak et al. (2002) to prepare total protein lysates from wild-type, glp-4(bn2ts), and unc-89(su75) strains. When comparing wild-type and mutant strains, we loaded approximately equal amounts of protein extract estimated by finding volumes of extracts that would give equal intensity of banding after Coomassie staining. We used quantities of extracts and dilutions of antibodies that would place us into the linear range of detection by ECL and exposure to film. The following antibodies and dilutions were used: rabbit anti–MEI-1 (affinity purified; Clark-Maguire and Mains, 1994b ) at 1:400; rat anti–PAT-6 (affinity purified; Xiong et al., unpublished data) at 1:200; monoclonal MH4 (tissue culture supernatant; Francis and Waterston, 1985 ), which detects intermediate filaments expressed in hypodermal cells at 1:1000; and monoclonal 5-6 (Miller et al., 1983 ) for MHC A at 1:1000. The specificities of all antibodies used in this study were established previously (see the respective references cited earlier). The quantitation of steady-state levels of MEI-1, PAT-6, and MHC A shown in Figure 7C was performed as described in Miller et al. (2009) . The relative amount of each of these muscle proteins in each lane was normalized to the amount of MH4 antigen.
An extrachromosomal array containing pPD49.78/83-HA-MEL-26 and pTG96 was integrated into the genome by UV irradiation (Mitani, 1995 ) with some modifications (P. Barrett, personal communication). The resulting integrated nematode line is called sfIs8 and was grown on 20 150-mm high-peptone plates seeded with NA22 E. coli and heat shocked for 3 h at 30oC. The worms were harvested and washed in M9 buffer, frozen at −80oC, and used to prepare a powder by extensive grinding with a mortar and pestle sitting on dry ice and the worms immersed in liquid nitrogen. A lysate was prepared by vortexing for 2 min a 10% (vol/vol) mixture of worm powder in lysis buffer (20 mM Tris, pH 8.0, 10% glycerol, 0.5% NP-40, 2 mM EDTA, 150 mM NaCl, complete Mini Protease Inhibitors [Roche, Indianapolis, IN]) spinning at top speed in a microcentrifuge for 10 min at 4oC and saving the supernatant. Four separate Eppendorf tubes contained 500 μl of worm lysate, and 50 μl of a 1:1 (vol/vol) slurry of monoclonal anti–HA-agarose beads (A2095; Sigma-Aldrich) was incubated with mixing at 4oC for 1.5 h. After washing the beads three times with lysis buffer, 20 μg of the following proteins were added separately: MBP, MBP-UNC-89 Ig2–Ig3, and MBP-UNC-89 Ig53–Fn2. The solutions were incubated with mixing at 4oC for 1.5 h. After washing the beads three times with lysis buffer, the beads were transferred to fresh tubes, pelleted, and eluted with Laemmli buffer at 95oC for 5 min and pelleted, and portions of supernatants were run on SDS–PAGE and blotted. One blot was used to detect MEI-1 (using affinity-purified rabbit anti–MEI-1 at 1:400 dilution) and MBP or MBP fusion proteins (using anti-MBP conjugated to HRP at 1:5000 dilution). Another blot was used to detect HA-MEL-26 (using anti-HA [mouse monoclonal] at 1:200 dilution). A control experiment was set up in which anti–HA-agarose beads were incubated with 20 μg of MBP, MBP-UNC-89 Ig2–Ig3, and MBP-UNC-89 Ig53–Fn2, washed, and eluted, and the MBP and MBP fusion proteins were detected by Western blot.
We thank Stephan Lange (University of California, San Diego, La Jolla, CA) for helpful discussions and for communicating his results before their publication. We are grateful to Lionel Pintard (University of Paris) for supplying the antibody to MEL-26, and we thank Andrew Fire (Stanford University) for RNAi and heat shock promoter vectors. G.M.B. thanks the Department of Pathology and School of Medicine of Emory University for bridge funds. We also acknowledge a Fellowship in Research and Science Teaching Postdoctoral Fellowship (K12GM000680) from the National Institutes of Health (for K.J.W.) and support from National Institutes of Health Grant R01AR051466. P.E.M. was supported by a grant from the Canadian Institutes of Health Research. Some of the strains used in this work were provided by the Caenorhabditis Genetics Center (funded by the National Institutes of Health, Center for Research Resources).
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E12-01-0055) on May 23, 2012.