The α-arrestin SUP-13/ARRD-15 promotes isoform turnover of actin-interacting protein 1 in Caenorhabditis elegans striated muscle

Abstract Precise arrangement of actin, myosin, and other regulatory components in a sarcomeric pattern is critical for producing contractile forces in striated muscles. Actin-interacting protein 1 (AIP1), also known as WD-repeat protein 1 (WDR1), is one of essential factors that regulate sarcomeric assembly of actin filaments. In the nematode Caenorhabditis elegans, mutation in unc-78, encoding one of the two AIP1 isoforms, causes severe disorganization of sarcomeric actin filaments and near paralysis, but mutation in sup-13 suppresses the unc-78-mutant phenotypes to restore nearly normal sarcomeric actin organization and worm motility. Here, we identified that sup-13 is a nonsense allele of arrd-15 encoding an α-arrestin. The sup-13/arrd-15 mutation suppressed the phenotypes of unc-78 null mutant but required aipl-1 that encodes a second AIP1 isoform. aipl-1 was normally expressed highly in embryos and downregulated in mature muscle. However, in the sup-13/arrd-15 mutant, the AIPL-1 protein was maintained at high levels in adult muscle to compensate for the absence of the UNC-78 protein. The sup-13/arrd-15 mutation caused accumulation of ubiquitinated AIPL-1 protein, suggesting that a normal function of sup-13/arrd-15 is to enhance degradation of ubiquitinated AIPL-1, thereby promoting transition of AIP1 isoforms from AIPL-1 to UNC-78 in developing muscle. These results suggest that α-arrestin is a novel factor to promote isoform turnover by enhancing protein degradation.


Introduction
Dynamic assembly and disassembly of actin filaments are critical for the function of the actin cytoskeleton that plays central roles in many cell biological processes including cell migration, cell division, and morphogenesis.Because actin filaments are relatively stable, severing and depolymerization of actin filaments are often promoted for cytoskeletal reorganization and recycling of actin monomers for new rounds of polymerization (1)(2)(3).Actin depolymerizing factor (ADF)/cofilin is a key regulator of actin filament disassembly by accelerating both depolymerization and severing of actin filaments.However, ADF/cofilin by itself can only modestly sever and depolymerize actin filaments.Therefore, ADF/cofilin cooperates with other factors including coronin, actin-interacting protein 1 (AIP1), and cyclase-associated protein to promote rapid actin filament disassembly.

Research Report
ADF/cofilin to actin filaments and further accelerates filament disassembly by AIP1 (16,18,19).In contrast, tropomyosin prevents binding of ADF/cofilin to actin filaments and inhibits filament disassembly by AIP1 (20).However, other than the regulation at the level of ADF/cofilin, how AIP1-mediated actin filament disassembly is regulated remains unknown.Since coronin, ADF/cofilin, and AIP1 are present in most eukaryotic cells (2), cells need to be equipped with a mechanism to control these strong actin filament disassembly factors in order to maintain functional actin cytoskeleton.
Mutation or knockdown of AIP1 causes abnormal actin filament dynamics and organization in vivo in most tested organisms (5,6).Notably, point mutations in the human AIP1 gene, WDR1, cause severe blood disorders, including reduction and impaired motility of neutrophils (21)(22)(23).Therefore, WDR1 is considered a gene responsible for lazy leukocyte syndrome (24,25).Similar WDR1 mutant phenotypes are reported in mice (26) and Zebrafish (27).AIP1 is also an essential regulator of sarcomere assembly in striated muscle.In the nematode Caenorhabditis elegans, mutations in the AIP1 gene, unc-78, cause severe disorganization of sarcomeric actin filaments and reduced muscle contractility (28)(29)(30).A cardiomyocyte-specific WDR1 knockout in mice causes similar actin disorganization and results in postnatal lethality with hypertrophic cardiomyopathy (31).Thus, genetic studies on AIP1 should provide insight in the basic in vivo function of AIP1 as well as pathogenesis of AIP1/WDR1-linked diseases.
In this study, we took advantage of C. elegans genetics to analyze an extragenic suppressor of unc-78 encoding AIP1.The sup-13 mutation was previously shown to suppress the motility defects of unc-78 mutants, but the molecular identity of sup-13 was uncharacterized (32,33).We identified that sup-13 is a nonsense allele of arrd-15 encoding a member of the α-arrestins that are generally known as adaptors to mediate protein trafficking and degradation (34)(35)(36).We found that sup-13/arrd-15 is a regulator of a second AIP1 gene, aipl-1, that is partially redundant with unc-78 (37,38).In wild-type muscle, UNC-78 is the predominant AIP1 isoform in adults.However, in sup-13/arrd-15 mutant muscle, AIPL-1 is maintained at high levels, without significant effects on UNC-78 levels, in adult muscle to compensate for the absence of UNC-78.Ubiquitinated AIPL-1 is accumulated in sup-13/arrd-15 mutants, suggesting that sup-13/arrd-15 mediates a process for degradation of ubiquitinated AIPL-1.Thus, our study shows that an AIP1 isoform is targeted for degradation in an α-arrestin-dependent manner in vivo and suggests a novel mechanism of isoform turnover by degradation of a selective isoform(s) in differentiated cells.
The sup-13 mutation also strongly suppressed disorganization of sarcomeric actin filaments in the body wall muscle of the unc-78 mutants (Fig. 1B).The body wall muscle is responsible for body movement of the worms, and actin filaments were organized in a striated pattern in wild type (Fig. 1B).In the unc-78(gk27) mutants, actin filaments were highly disorganized and accumulated in large aggregates (Fig. 1B).However, in the sup-13;unc-78(gk27) double mutants, striated organization of actin filaments were nearly indistinguishable from that in wild type (Fig. 1B).The sup-13 single mutants did not exhibit detectable defects in the actin filament organization (Fig. 1B).During development, the actin disorganization phenotype became detectable in unc-78(gk27) at the L3 larval stage and was nearly 100% penetrant after the L4 larval stage (Fig. 1C).The sup-13 mutation almost completely suppressed the actin disorganization phenotype of unc-78(gk27) at all developmental stages (Fig. 1C), and the sup-13 mutation alone did not cause actin disorganization at any stage (Fig. 1C).These cytological observations indicate that the sup-13 mutation suppresses the muscle defects of the unc-78 mutants by suppressing disorganization of sarcomeric actin filaments to restore assembly of functional contractile apparatuses.
Expression of ARRD-15 in the body wall muscle was detected by observing localization of endogenous sup-13/arrd-15 that was C-terminally tagged with GFP (ARRD-15::GFP) by CRISPR/ Cas9-mediated gene editing (Fig. 2D and E).The fluorescence levels of ARRD-15::GFP were strongest in the nerve ring (Fig. 2D) and also detected at low levels in the body wall muscle and localized in the diffuse cytoplasm (Fig. 2E).This expression pattern is consistent with the mRNA distribution data in the GExplore database (41) (Online Supplementary Fig. S1).Together, these results of muscle-specific transgenic rescue and endogenous expression of sup-13/arrd-15 in the body wall muscle indicate that sup-13/ arrd-15 acts autonomously in the body wall muscle to influence sarcomeric actin organization.

The second AIP1 isoform, aipl-1, is required for sup-13/arrd-15-dependent phenotypic suppression of the unc-78 mutant phenotypes
Caenorhabditis elegans has a second AIP1 gene, aipl-1, that is partially redundant with unc-78 in actin regulation (37,38).We found that aipl-1 is essential for the phenotypic suppression by the sup-13/arrd-15 mutation in the unc-78 mutants (Fig. 3).Depletion of aipl-1 in wild type by RNA interference (RNAi) did not cause a detectable abnormality (Fig. 3A, left).However, in the unc-78(gk27) mutants, aipl-1(RNAi) caused embryonic arrest (Fig. 3A, middle), as reported previously (37), and this lethal phenotype was not suppressed by the sup-13 mutation (Fig. 3A, right).In order to observe adult phenotypes, the RNAi treatments were started from the L1 larval stage after normal embryogenesis was completed, and phenotypes analyzed in adult worms (Fig. 3B and C).Under these conditions, aipl-1(RNAi) did not cause detectable alterations in the phenotypes in wild type (Fig. 3B, left; Fig. 3C) and unc-78(gk27) (Fig. 3B, middle; Fig. 3C) but caused disorganization of actin filaments with aggregate formation in sup-13;unc-78(gk27) (Fig. 3B, right; Fig. 3C).These results and our previous reports demonstrate that unc-78 and aipl-1 have functional impacts at different developmental stages: unc-78 mutations cause stronger phenotypes in the late larval and adult stages than in the embryonic and early larval stages (28) (Fig. 1C), whereas aipl-1(RNAi) (in an unc-78 background) causes stronger phenotypes in the embryonic stages than in the adult stage (37).However, the strong adult phenotypes of  78(gk27).This phenotype was suppressed in the sup-13(st210);unc-78(gk27) double mutant to nearly an appearance of wild type.The sup-13(st210) single mutant did not show a detectable phenotype in the F-actin organization.Bar, 10 μm.C) Worms at different developmental stages were stained for F-actin by tetramethylrhodamine-phalloidin, and percentages of worms with F-actin aggregates (counted as positive when >25% of total muscle cells contained aggregates) were analyzed.For each experiment, > 500 worms were examined for each stage, and average ± standard deviation of three independent experiments are shown.***, P < 0.001.

sup-13/arrd-15 controls the protein level of AIPL-1, but not UNC-78, possibly in a Ub-dependent manner
Examination of AIPL-1 protein levels revealed that the sup-13/ arrd-15 mutation altered the AIPL-1 protein level in the adult body wall muscle (Fig. 4).Since AIPL-1 is expressed in many tissues (37), the endogenous aipl-1 gene was C-terminally tagged with GFP by CRISPR/Cas9-mediated gene editing, and its expression in the body wall muscle was examined by fluorescence microscopy.The AIPL-1::GFP protein levels were determined as fluorescence intensity.In wild type, AIPL-1::GFP was very low in the body wall muscle (Fig. 4A and B) and slightly increased in unc-78(gk27) (Fig. 4A and B).However, in sup-13;unc-78(gk27) and sup-13, AIPL-1::GFP was significantly increased in the body wall muscle as compared with wild type and unc-78(gk27) (Fig. 4A and B), indicating that the sup-13/arrd-15 mutation was responsible for the high levels of AIPL-1::GFP in the body wall muscle.In contrast, the UNC-78 protein levels were not altered by the sup-13/arrd-15 mutation (Fig. 4C and D).The UNC-78 protein was examined by Western blot using whole adult worm lysates because UNC-78 is predominantly expressed in the body wall muscle in adults (28,30).Indistinguishable levels of UNC-78 were detected in wild type and sup-13, and the absence of UNC-78 was confirmed in unc-78(gk27) and sup-13;unc-78(gk27) (Fig. 4C and D).These results indicate that the sup-13/arrd-15 mutation specifically increases the level of AIPL-1, but not UNC-78, in the adult body wall muscle.Therefore, a normal function of sup-13/arrd-15 is to decrease the level of AIPL-1 in the adult muscle.
To determine the regulatory mechanism of AIPL-1 protein levels by sup-13/arrd-15, we examined the ubiquitination levels of AIPL-1 and found that ubiquitinated AIPL-1 was accumulated in the sup-13/arrd-15 mutants (Fig. 5A and B).AIPL-1::GFP was immunoprecipitated, and probed with anti-Ub antibody in Western blot (Fig. 5A).Ubiquitinated AIPL-1::GFP was nearly undetectable in wild type and unc-78(gk27) but detected at high levels in sup-13 and sup-13;unc-78(gk27) (Fig. 5A and B).Therefore, in the sup-13/ arrd-15 mutants, AIPL-1 protein might be maintained at high levels because ubiquitinated AIPL-1 was accumulated without being processed by the downstream protein degradation pathway.However, levels of ubiquitinated proteins in total worm lysates were not significantly different among the four strains (Fig. 5C  and D), suggesting that sup-13/arrd-15 regulates degradation of only specific target proteins rather than systemic degradation of ubiquitinated proteins.These results suggest that sup-13/arrd-15 acts after the ubiquitination process and promotes degradation of ubiquitinated AIPL-1.

Discussion
In this study, we identified the molecular nature of the sup-13/ arrd-15 gene and characterized how the sup-13/arrd-15 mutation suppresses the unc-78 mutant phenotypes.Based on the current and previous data, we propose that the isoform turnover of AIP1 occurs at two levels in C. elegans muscle.First, expression of unc-78 and aipl-1 are transcriptionally regulated in an opposite manner as determined by the promoter analyses: the unc-78 promoter is more active in adult muscle than embryonic muscle (30), whereas the aipl-1 promoter is more active in embryonic muscle than in adult muscle (37).Therefore, during muscle development, expression of unc-78 is increased, but that of aipl-1 decreased.Second, the AIPL-1 protein is specifically cleared in adult muscle by sup-13/arrd-15 in a Ub-dependent manner.In the sup-13/ arrd-15 mutants, AIPL-1 protein is maintained at high levels in adult muscle and is capable of compensating for the absence of UNC-78 in the unc-78-null mutants (Fig. 6).This is supported by our previous observations that forced expression of GFP::AIPL-1 in adult muscle of the unc-78-null mutants rescues the sarcomeric actin organization (37), indicating that AIPL-1 is sufficient for suppressing the unc-78-mutant phenotype.In addition, knockdown of aipl-1 in sup-13;unc-78(gk27) suppressed the Sup-13 suppressor phenotype (Fig. 3), indicating that aipl-1 is necessary for suppressing the unc-78-mutant phenotype in the sup-13/arrd-15 mutant background.Thus, the transition of AIP1 isoforms in C. elegans muscle appears to be a highly regulated process.Multiple AIP1 isoforms are present only in a limited number of species, and protein levels of AIP1 may be regulated in a species-specific manner.In Arabidopsis thaliana, two AIP1 isoforms are differentially expressed in reproductive and vegetative tissues and regulate actin organization in respective tissues (42)(43)(44).However, most species have only one gene encoding AIP1 (5,6,45).Mammals have a single AIP1 gene, known as WDR1 (46).In the mouse heart, the WDR1 protein level is high in embryos and then gradually decreased by unknown mechanism during postembryonic development (31).AIP1 can strongly sever ADF/cofilinbound actin filaments at very low concentrations in vitro (12)(13)(14)(15)(16). Therefore, high AIP1/WDR1 levels may be maintained when actin cytoskeletal reorganization is highly active in embryonic heart, but AIP1/WDR1 may be decreased when mature myofibrils are assembled and stabilized in postembryonic heart.However, it is still not clear why turnover of the AIP1 isoforms occurs in C. elegans muscle.Biochemically, UNC-78 and AIPL-1 are qualitatively very similar in promoting disassembly of ADF/cofilin-bound actin filaments (37), preference for ADF/cofilin isoforms (29,37), and pH sensitivity (45).Genetically, unc-78 and aipl-1 are partially redundant in assembly of contractile apparatuses in striated and nonstriated muscles (37,38).Therefore, there may be unknown differences in the function and/or regulation of the two AIP1 isoforms, which might be beneficial for C. elegans.
Most of the previously characterized α-arrestins have multiple PY motifs (PP × Y or LP × Y) in their C-termini as binding sites for the WW domains of Rsp5/Nedd4-like E3 Ub ligases (34)(35)(36).However, SUP-13/ARRD-15 lacks this motif (Fig. 2A), and how it is involved in the degradation of a target protein is currently unknown.In mammals, arrestin-domain-containing 5 (ARRDC5) is the only α-arrestin lacking a PY motif, and its biological function is poorly understood.One recent study has shown that ARRDC5 is enriched in testis and required for normal spermatogenesis in mice (47), although whether ARRDC5 is involved in protein degradation is unknown.C. elegans has one β-arrestin gene, arr-1 (48), and 28 α-arrestin genes.The arr-1 β-arrestin gene is involved in G-protein-coupled receptor signaling (49).However, only two α-arrestin genes, cnp-1/arrd-17 and ttm-2, have been previously characterized in C. elegans.cnp-1/arrd-17 is required for chemosensation (50).ttm-2 is required for defense against bacterial toxins (51).Either cnp-1/arrd-17 or ttm-2 does not have a consensus PY motif and has not been implicated in protein degradation.Therefore, metazoan α-arrestins lacking a PY motif may recruit a Ub ligase using an alternative mechanism such as phosphor-Ser/Thr dependent binding to a WW domain (52) or be involved in unknown biological processes.
Our study suggests that an α-arrestin is involved in isoform switching of muscle proteins during development.Many sarcomeric proteins undergo isoform transition during muscle differentiation, development, and under pathological conditions (53).Although isoform switching at levels of transcription and alternative splicing has been extensively studied, the clearance mechanism for old isoforms is still poorly understood.Since misexpression of inappropriate isoforms can cause pathological conditions in muscle, clearing old or misexpressed isoforms will be important to complete normal isoform transition.For example, abnormal expression of troponin T isoforms due to misregulated alternative splicing is implicated in the pathogenesis of myotonic dystrophy (54).Expression of a nonmuscle tropomyosin isoform in skeletal muscle can cause dystrophic phenotypes (55,56).In addition to these structural proteins, cofilin isoforms also transition from a nonmuscle isoform (cofilin-1) to a muscle isoform (cofilin-2) during muscle differentiation (57,58).Cofilin-1 is normally cleared by the Ub-proteasomal system after muscle differentiation, but failure to clear cofilin-1 from mature muscle causes sarcomere disorganization and impaired force generation thereby enhancing phenotypes in Emery-Dreifuss muscular dystrophy (59).The Ub-proteasomal system has been shown to be important for normal sarcomere assembly (60)(61)(62), and muscle atrophy, and hypertrophy (63)(64)(65).To date, Ub ligases, MuRF-1, MAFbx-1/atrogin-1, and Cbl-b have been associated with muscle atrophy (66)(67)(68), and UBR4 with hypertrophy (64), but function of an α-arrestin in a Ub-dependent pathway has not been determined in vertebrate muscle.Therefore, our study suggests a novel role of an α-arrestin in protein turnover in normal muscle development or under pathological conditions.

Motility assays
Worm motility was quantified as described previously (70).Briefly, adult worms were placed in M9 buffer.Then, one beat was counted when a worm swung its head to either right or left.The total number of beats in 30 s was recorded.Fig. 6.Model for the regulatory mechanism of AIP1 isoform turnover in the body wall muscle.The current study suggests that SUP-13/ARRD-15 specifically targets AIPL-1 for degradation in the body wall muscle.Therefore, in wild type, UNC-78 is the predominant AIP1 isoform that cooperates with UNC-60B (ADF/cofilin) to promote actin filament dynamics.In unc-78 single mutants, there is not enough AIP1 proteins to promote actin filament dynamics, causing sarcomere defects.However, in sup-13;unc-78 double mutants, AIPL-1 is not degraded and compensates for the absence of UNC-78.In sup-13 single mutants, both UNC-78 and AIPL-1 promotes actin filament dynamics and do not cause apparent sarcomere defects.

Fig. 1 .
Fig. 1.The sup-13 mutation suppresses motility defects and sarcomere disorganization of the unc-78 null mutant.A) Motility of young adult worms was quantified by measuring beating frequency (beats per 30 s).An unc-78 null allele, unc-78(gk27), was used in this study and simply shown as unc-78 (n = 30, data are average ± standard deviation).***, P < 0.001.n.s., not significant.B) Adult worms were stained for F-actin by tetramethylrhodamine-phalloidin, and regions of the body wall muscle are shown.Actin was organized in a striated manner in wild type but highly disorganized with aggregates in unc-78(gk27).This phenotype was suppressed in the sup-13(st210);unc-78(gk27) double mutant to nearly an appearance of wild type.The sup-13(st210) single mutant did not show a detectable phenotype in the F-actin organization.Bar, 10 μm.C) Worms at different developmental stages were stained for F-actin by tetramethylrhodamine-phalloidin, and percentages of worms with F-actin aggregates (counted as positive when >25% of total muscle cells contained aggregates) were analyzed.For each experiment, > 500 worms were examined for each stage, and average ± standard deviation of three independent experiments are shown.***, P < 0.001.

Fig. 3 .
Fig.3.The second AIP1 gene aipl-1 is required for sup-13/arrd-15-dependent suppression of the unc-78 mutant phenotypes.A) Wild type, unc-78, or sup-13; unc-78 worms were treated with control RNAi or aipl-1(RNAi) from the L4 larval stage, and the phenotypes were examined on agar plates in the next generation.In control RNAi, all strains grew to adult worms.However, in aipl-1(RNAi), only wild type grew to adult worms, but unc-78 and sup-13;unc-78 were arrested as embryos.Bar, 1.0 mm.B) Wild type, unc-78, or sup-13;unc-78 worms were treated with control RNAi or aipl-1(RNAi) from the L1 larval stage, and stained for F-actin with tetramethylrhodamine-phalloidin when they became adults, and regions of the body wall muscle are shown.In adult muscle, aipl-1(RNAi) did not affect the F-actin organization in wild type and unc-78 but caused severe F-actin disorganization in sup-13;unc-78.Bar, 20 μm.C) Percentages of worms with F-actin aggregates (counted as positive when >25% of total muscle cells contained aggregates) were analyzed.For each experiment, >200 worms were examined for each sample, and average ± standard deviation of three independent experiments are shown (error bars are too small to be visible).***, P < 0.001.n.s., not significant.

Fig. 4 .
Fig. 4. The sup-13/arrd-15 mutation increases the protein levels of AIPL-1 in the body wall muscle.A, B) The AIPL-1::GFP protein levels in the body wall muscle were determined in the head region where the autofluorescence of the gut granules is absent.Fluorescence images of AIPL-1::GFP were captured using the same exposure settings for the four strains (A; bar, 20 μm), and the AIPL-1::GFP protein levels were determined as fluorescence intensity (arbitrary units) within the body wall muscle (BWM in A) in B. Data are average ± standard deviation (n = 10).*, 0.01 < P < 0.05.***, P < 0.001.C, D) The UNC-78 protein levels in whole worm lysates were determined by Western blot using anti-UNC-78 antibody.Whole worm lysates of an equal number of young adults (30 worms) were loaded for each strain and probed with anti-UNC-78 antibody (C, top).Total proteins were stained with Coomassie Brilliant Blue (C, bottom) and used as a loading control to normalize the UNC-78 signals (arbitrary units) in D. Positions of molecular weight markers in kDa are shown on the left.Data are average ± standard deviation (n = 6).*, 0.01 < P < 0.05.**, 0.001 < P < 0.01.n.s., not significant.

Fig. 5 .
Fig. 5. Ubiquitinated AIPL-1 is accumulated in the sup-13/arrd-15 mutants.A, B) AIPL-1::GFP was immunoprecipitated with anti-GFP magnetic beads and probed with anti-Ub (Ub) antibody (A, top) and anti-DYKDDDDK (FLAG epitopes in AIPL-1::GFP) antibody (A, bottom).The anti-Ub signals (arbitrary units) were normalized to anti-DYKDDDDK signals and shown in the graph (B).Data are average ± standard deviation (n = 3).*, 0.01 < P < 0.05.**, 0.001 < P < 0.01.***, P < 0.001.n.s., not significant.C, D) Whole lysates of adult worms were examined by Western blot using anti-Ub antibody (C, left).Total proteins on the blot were stained with Coomassie Brilliant Blue (CBB) (C, right) and used as a loading control to normalize the anti-Ub signals (arbitrary units) in D. Positions of molecular weight markers in kDa are shown on the left in A and C. Data are average ± standard deviation (n = 3).There were no significant differences among the data.