Reduction of UDP- N -acetylglucosamine 2-Epimerase/ N -Acetylmannosamine Kinase Activity and Sialylation in Distal Myopathy with Rimmed Vacuoles*

Distal myopathy with rimmed vacuoles is an autosomal recessive muscle disease with preferential involvement of the tibialis anterior that spares the quadriceps muscles in young adulthood. In a Japanese patient with distal myopathy with rimmed vacuoles, we identified pathogenic mutations in the gene encoding the bifunc-tional enzyme UDP-GlcNAc 2-epimerase/ManNAc kinase, which catalyzes the initial two steps in the biosynthesis of sialic acid. In this study, we demonstrated the relationship between the genetic mutations and enzymatic activities using an in vitro expression assay sys-tem. Furthermore, we also showed that the levels of sialic acid in muscle and primary cultured cells from DMRV patients were reduced to 60–75%

Distal myopathy with rimmed vacuoles (DMRV) 1 is an autosomal recessive disorder characterized clinically by the preferential involvement of the tibialis anterior muscle, sparing the quadriceps muscles as originally described in 1981 (1,2). The age at onset is relatively late with a mean of 26 years. Muscle biopsy of this disorder is characterized by many rimmed vacuoles, which are particularly abundant in atrophic fibers. Necrotic and regenerating fibers are rarely seen (3). The nucleus occasionally contains tubulofilamentous inclusions of 15-20 nm in diameter.
Hereditary inclusion body myopathy (HIBM) is an autosomal recessive disorder that presents with adult-onset slowly progressive distal and proximal weakness and has characteristic pathological features in muscle tissue, including rimmed vacuoles and filamentous inclusions, that are similar to those seen in DMRV (4). Gene loci of both diseases have been mapped to chromosome 9 (5,6). HIBM is caused by mutations in the UDP-GlcNAc 2-epimerase/ManNAc kinase gene (GNE gene) (7). Previously we identified homozygous and compound heterozygous mutations in the GNE gene in 27 DMRV patients (8), demonstrating that the two diseases are allelic.
UDP-GlcNAc 2-epimerase/ManNAc kinase is a dual functional enzyme catalyzing two initial steps in the biosynthesis of sialic acid (9). This enzyme catalyzes the conversions of UDP-GlcNAc to ManNAc and ManNAc to ManNAc 6-phosphate. Despite the identification of the GNE gene mutations, we still do not fully understand how these mutations contribute to the pathophysiology in DMRV/HIBM. Several questions remain unanswered. 1) What is the status of sialylation activity in the patients with GNE mutations? One would expect sialylation to be impaired but not completely absent in DMRV/HIBM patients, because sialic acid is essential for embryonic development (10). In fact, homozygous null mutations have never been identified in patients (8,11). 2) Why are symptoms restricted to the skeletal muscles? GNE transcripts are expressed in various tissues and are especially predominant in the liver (12). 3) Why do mutant proteins not complement each other in patients who have heterozygous mutations in each of the two domains? The two domains in GNE protein have been reported to catalyze the enzymatic reactions separately and independently (13). To address these questions, we studied the relationships between mutations and enzymatic activity using in vitro expression and enzymatic assay systems. We also determined the levels of sialylation in sera, muscles, and primary cultured cells from DMRV patients and normal individuals.

EXPERIMENTAL PROCEDURES
Mutation and Sialic Acid Analyses of Patients-All of the patients were Japanese. The patients were diagnosed as having DMRV based on both clinical features and muscle pathology. Numbering of the patients followed the protocol presented in our previous report (8). Gene analyses of DMRV patients were performed as described previously (8). Primary fibroblasts were obtained from patient 18 and patient 19, whose mutations were reported previously. Primary skeletal myocytes were obtained from patient 5 as reported previously. Sialic acid contents in sera were measured with a SIALIZYME-550 kit (Fujirevio, Tokyo, Japan), and those in muscle and cells were determined by the thiobarbituric acid method. Informed consents were obtained from all subjects using a form approved by the Ethical Review Board at the National Center of Neurology and Psychiatry (Tokyo, Japan).
Expression of Recombinant GNE Proteins-The cDNA for wild-type GNE was obtained by reverse transcribed-PCR from normal muscle RNA and cloned into pCR-blunt vector (Invitrogen). The cDNAs for GNE mutants were obtained by reverse transcribed-PCR from skeletal muscle RNA of DMRV patients or by site-directed mutagenesis from wild-type cDNA. All cloned muscle cDNAs were sequenced by ABI cycle-sequencing procedures using an ABI 3100 (Applied Biosystems, Foster City, CA). The sequenced and inserted cDNAs were cut out with EcoRI and blunted, and the purified cDNA fragments were inserted in-frame into the expression vector, pCMV-Myc (Invitrogen). The expression constructs were transiently transfected into COS-7 cells using LipofectAMINE Plus (Invitrogen) according to the manufacturer's protocol. After 24 h, the Myc-tagged wild-type GNE and the mutant proteins were extracted from transfected cells. UDP-GlcNAc 2-epimerase activity was measured as described previously. The ManNAc kinase assay was performed with slight modification according the previous report (13).
Cross-linking of GNE Mutant Proteins-To analyze the oligomer structure of wild-type and mutant GNE, cell lysates were subjected to a reaction with 10 mM MBS for 30 min at room temperature for crosslinking. The Myc-tagged cross-linked products were purified with anti-Myc-agarose (Santa Cruz Biotechnology, Santa Cruz, CA) and eluted by boiling in 2% SDS solution. The products were subjected to SDS-PAGE and Western blot using anti-Myc 9E10 antibody (Santa Cruz Biotechnology).
Cell Cultures-COS-7 cells were cultured in 10% fetal bovine serum/ Dulbecco's modified Eagle's medium in 5% CO 2 . Primary fibroblasts and myoblasts from DMRV patients and normal individuals were cultured in 10% fetal bovine serum, Dulbecco's modified Eagle's medium, and Ham's F-12 medium in 5% CO 2 . The myoblasts were induced to myogenic differentiation by switching the medium to 5% horse serum, Dulbecco's modified Eagle's medium, and Ham's F-12 medium. At 24 h before lectin staining or sialic acid determination, the medium was replaced with serum-free Dulbecco's modified Eagle's medium and Ham's F-12 medium with or without 5 mM GlcNAc, ManNAc, or NeuAc. Cells were fixed and permeabilized as described previously (15). Biotinlabeled SBA, wheat germ agglutinin (WGA) (Seikagaku Kogyo), an antibody against desmin (ICN Pharmaceuticals, Costa Mesa, CA), and Alexa Fluo 594-labeled secondary antibodies (Molecular Probes, Eugene, OR) were used for staining the cells.

Novel Mutations in GNE Gene in DMRV Patients-
In the previous study, we identified 12 different GNE mutations in either homozygous or compound heterozygous states in 27 DMRV patients (8). From these results, we concluded that DMRV is allelic to HIBM. Subsequently, we identified six patients harboring five different mutations, of which three were novel mutations: 1622C3 T, 89G3 C, and 2173G3 A (Table I). These novel mutations were absent in 100 control chromosomes from normal Japanese individuals.
Enzymatic Activities of GNE Mutants in DMRV Patients-Site-directed mutagenesis of the GNE protein has shown that the two enzymatic activities are separately and independently catalyzed by two domains, an N-terminal epimerase domain and a C-terminal kinase domain (13). Previously, we reported reductions in UDP-GlcNAc 2-epimerase activities in leukocytes from the patients with mutations in the GNE gene (8). However, the enzyme activity was too weak in leukocytes to clearly demonstrate correlations between gene mutations and the enzymatic activities. To clarify this relationship, we generated recombinant proteins with each of the 13 species of mutations that we identified. All but one of the mutant and wild-type recombinant GNE proteins were expressed in COS cells and migrated at 75 kDa in SDS-PAGE. The single exception was a recombinant protein with a deletion of amino acids 206 -256 (⌬206 -256) caused by exon 4 skipping. The abnormal protein was degraded in COS cells (Fig. 1A). We determined the specific activities of UDP-GlcNAc 2-epimerase and ManNAc kinase of the mutant proteins relative to wild type (Fig. 1B). The endogenous activities in mock transfected COS cells were determined to correct for the background enzyme activity. UDP-GlcNAc 2-epimerase activities in mutants C13S, H132Q, D176V, D177C, V331A, and D378Y were reduced to less than 20% of the control. In contrast, the I472T and G708S mutants each revealed an ϳ50% reduction, and V572L, A630T, and A631V each showed only a 20 -30% reduction in activity as compared with wild-type cells. ManNAc kinase activity was retained in the N-terminal mutants C13S, H132Q, D176V, D177C, V331A, and D378Y, whereas the C-terminal mutants I472T, V572L, A630T, A631V, and G703S showed dramatic reductions in activities. These data were essentially compatible with a prior report (13). Interestingly, the A524V mutant preferentially affected UDP-GlcNAc 2-epimerase activity, although the mutation is in the kinase domain. None of the DMRV mutants showed complete loss of UDP-GlcNAc 2-epimerase or ManNAc kinase activities.
Oligomerization of GNE Mutants-GNE protein forms a homohexamer by oligomerization (13). We examined whether the mutations affect the oligomerization of GNE molecules, because homohexamer structure of GNE protein was reported to be essential for UDP-GlcNAc 2-epimerase activity (13). The recombinant mutant proteins were subjected to the cross-linking with MBS. Fig. 1C shows the electrophoretic tracing patterns of cross-linked products of wild-type and mutant GNE proteins. The product of wild-type GNE predominantly migrated at Ͼ400 kDa, which corresponds to homohexamer. C13S, V572L, A630T, and A631V gave cross-linked products similar to that of wild-type GNE, whereas the mutants H132Q and D176V mainly generated a 200-kDa product, and D177C, V331A, D378Y, and A524V produced 98-kDa proteins. These data indicate that the hexameric oligomerization is necessary for UDP-GlcNAc 2-epimerase activity and that N-terminal mutants generally fail to form large oligomers although the molecular region responsible for the oligomerization is not clearly The mutations were identified in patients diagnosed as having DMRV based on both clinical features and muscle pathology. Numbering of patients followed that in our previous report (8). Exon indicates the exon number where mutation was found. Protein domain is predicted ones from the sequence homology as previously (13 These results led us to hypothesize that sialylation should be affected in the tissues of DMRV patients. We measured the sialic acid content in sera and muscles from DMRV patients ( Fig. 2A). In sera, no difference was detected between patients and normal controls, whereas in skeletal muscle, a 25% reduction of sialic acid was observed in DMRV muscles. We also assessed the status of sialylation in DMRV muscles by lectin staining. We used three lectins: SSA for detection of Sia␣2-6Gal/GalNAc, MAM for Sia␣2-3Gal, and SBA for GalNAc␣1-3Gal (16 -18) (Fig. 2B). The results of lectin staining are summarized in Table II. SSA uniformly stained sarcolemma in control muscle, whereas it faintly stained sarcolemma and strongly stained interstitial tissues in DMRV muscle (Fig. 2B, panels d-f). MAM strongly stained sarcolemma and interstitial tissues in both the control and patient muscles. We did not observe any reduction in MAM staining in  patients, which may be attributed to the strong intensity in our staining condition (data not shown). In contrast, SBA strongly highlighted the rimmed vacuoles containing fibers and the surrounding atrophic fibers in the patients both in sarcolemma and cytoplasm (Fig. 2B, panels h and i; see arrows), whereas it did not stain myofibers in the control (Fig. 2B, panel g). These data suggest that sialylation, other glycosylation, or both are at least partly disturbed in some myofibers in DMRV. Furthermore, we examined the expression of glycosylated ␣-dystroglycan in DMRV muscles using an antibody (VIA4-I) that recognizes a carbohydrate epitope. The ␣-dystroglycan staining was negative in rimmed vacuoles containing fibers and the surrounding atrophic fibers in one DMRV patient (Fig. 2B, panel  l). However, positive staining in another patient demonstrated that ␣-dystroglycan expression varies among patients (Fig. 2B,  panel k). Therefore, we concluded that the loss of ␣-dystroglycan staining is an extreme down-stream phenomenon in DMRV muscles. We also analyzed muscle sialylated glycoproteins (LAMP-1, and ␣and ␤-dystroglycans) by one-or two-dimensional polyacrylamide gel electrophoresis, but when they were extracted in whole amounts, there was no significant change in the electrophoretic patterns of these proteins between control and patients (data not shown). Furthermore, we analyzed the laminin-binding property of ␣-dystroglycan from DMRV patients, and the ␣-dystroglycan showed a strong binding as the control (data not shown).

Restoration of Sialylation in DMRV Cells by Feeding with
ManNAc and NeuAc-Previous studies reported that hyposialylation in GNEϪ/Ϫ embryonic stem cells and GNE-defective cells can be repaired by feeding with the natural sialic acid precursor ManNAc (10,19). We examined the recovery of sialylation by the addition of NeuAc or ManNAc using primary cultured cells from patients and evaluated the sialylation status by lectin staining (Fig. 3A). We used two lectins, WGA, which specifically recognizes a cluster structure of sialic acids (20), and SBA. WGA strongly stained the perinuclear region in control fibroblasts. Fibroblasts from patient 18, who harbors compound heterozygous missense mutations D176V and I472T, showed weaker staining with WGA compared with normal controls ( Fig. 3A; WGA, DMRV, and Control). SBA did not stain the fibroblasts from normal controls but strongly stained cells from DMRV patients. These results were also confirmed in myotubes from another DMRV patient (patient 5) who has compound heterozygous mutations causing exon 4 skipping and V572L (Fig. 3B) and fibroblasts from patient 19 with a homozygous D176V mutation (data not shown). The sialic acid
Arrows indicate rimmed vacuoles. SBA lectin strongly stained sarcolemma and cytoplasmic areas in the cluster of atrophic or rimmed vacuoles containing myofibers. FIG. 3. Recovery of the sialylation in DMRV cells by treatment with ManNAc and NeuAc. A, the fibroblasts from patient 18 (DMRV) and control individuals were stained with WGA and SBA lectins. The patient 18 fibroblasts cultured in the presence of GlcNAc (ϩGlcNAc), ManNAc (ϩManNAc) and NeuAc (ϩNeuAc) were also stained with both lectins. B, differentiated myotubes from patient 5 cultured without (DMRV) and with ManNAc (ϩManNAc) and NeuAc (ϩNeuAc) were stained by WGA and SBA lectins and desmin antibody. Desmin-positive cells are myotubes. C, sialic acids in myotubes (n ϭ 4 from patient 5, black) and fibroblasts (n ϭ 4 from patients 18 and 19, white) were measured by the thiobarbituric acid method. levels in skin fibroblasts and myotubes from DMRV patients were significantly decreased at ϳ60 -74% of control cells when cells were cultured in serum-free medium (Fig. 3C). By adding ManNAc or NeuAc into the culture medium, sialic acid levels in the fibroblasts and myotubes were restored to normal levels (Fig. 3C, ϩManNAc and ϩNeuAc). Furthermore, WGA staining of cells from patients also increased to normal levels, and particularly strong WGA staining was observed in the plasma membrane of myotubes. In contrast, the SBA staining in DMRV cells disappeared by the addition of either sugar (Fig. 3,  A and B; ϩManNAc and ϩNeuAc). The addition of GlcNAc into the medium had no effect on the staining pattern with either lectin (Fig. 3, A and C, ϩGlcNAc). These results suggest the potential therapeutic use of ManNAc and NeuAc. DISCUSSION In this study, we identified six additional DMRV patients with GNE mutations. Combined with prior results, the 1765G3 C mutation accounts for 55% (36 of 66) of the abnormal alleles confirming the high frequency of this mutation in Japan. Haplotype analysis suggests that this common mutation is due to a founder effect (8). All of the mutations identified in DMRV patients caused reduction (but not total loss) of enzymatic activities of either UDP-GlcNAc 2-epimerase or ManNAc kinase. These results strongly suggest that DMRV is caused by partial loss of function of the gene product. Interestingly, we previously identified the compound heterozygous mutations D378Y and A631V in a North American DMRV patient of German and Irish origin; these mutations have also been identified in an Irish HIBM patient (13). D378Y reduced UDP-GlcNAc 2-epimerase activity, and A631V decreased ManNAc kinase activity. Together with clinical and pathological similarities, these biochemical and molecular genetic results suggest that DMRV and HIBM are actually the same disease.
Through our study, we obtained information about novel molecular aspects in GNE. The two catalytic domains of the GNE molecule do not always work separately or independently in contrast to a published report (13). For example, the A524V mutation is within the predicted ManNAc kinase domain; however, it strongly inhibited UDP-GlcNAc 2-epimerase. Interestingly, this mutant did not form an oligomeric structure similar to the other N-terminal mutants. The failure of oligomerization in this A524V mutant is probably responsible for the reduced UDP-GlcNAc 2-epimerase activity as suggested previously (13).
Sialylation was decreased in muscle and in cultured cells from patients but was not completely lost, because all of the mutant proteins with missense mutations partially retained both enzymatic activities. Sialic acid levels in sera from DMRV patients were normal. Sialic acids are predominantly produced in the liver and transferred to synthesized glycoproteins. The sialylated proteins are released into the blood plasma, and free sialic acid in the plasma is derived from desialylation of these glycoproteins. GNE is expressed in the liver in large amounts; therefore, the reduction in enzymatic activities by mutations may not significantly affect the synthesis of sialic acid in the livers of DMRV patients, and sialic acids are present at concentrations comparable with normal blood levels. In contrast, in DMRV skeletal muscles, the sialic acid contents are reduced. The reduced enzymatic activities along with weak expression of GNE protein are probably responsible for the more serious reduction in sialic acid synthesis in muscle tissue compared with plasma. Lectin staining showed abnormal staining only in some fibers, indicating that a restricted number of myofibers has glycosylation abnormalities. This selective involvement may be due to muscle uptake of sialic acid, which can compensate for the defect of sialic acid synthesis in most fibers and explains why patients are normal at birth and develop late onset myopathies.
By feeding DMRV myotubes and fibroblasts with NeuAc as well as ManNAc, sialic acid concentrations in the cells increased to normal levels. As reported previously (21), treatment with NeuAc resulted in more rapid and potent effects on the restoration of sialylation than treatment with ManNAc. This strongly suggests that pharmacological therapy may be effective against DMRV/HIBM. Interestingly, even in myotubes harboring mutations that severely decrease ManNAc kinase activity, sialylation was restored by treatment with ManNAc. Schwarzkopf et al. (10) also reported similar observations in the embryonic stem cell culture in which the GNE gene was disrupted. They suggested that another sugar kinase may convert ManNAc to ManNAc-6-phosphate in those cells; therefore, ManNAc kinase activity of GNE may not be essential for sialic acid synthesis. If so, then why do the GNE mutations retaining UDP-GlcNAc 2-epimerase activity cause loss of sialylation and disease? One possible explanation is that these mutations may destabilize the GNE molecule resulting in decreased amounts of mutated proteins. However, we did not detect any reductions in the expressed amounts or defects in oligomerization of ManNAc-mutated recombinant proteins. Further analysis is necessary to clarify the mechanisms for the rescue resulting from the addition of ManNAc.
Enhanced staining with SBA lectin was observed in the sarcolemma and within the cytoplasmic area of some myofibers. These fibers were clustered and tended to be atrophic or have rimmed vacuoles. There is a report describing negative staining with SBA in Duchenne and Becker muscular dystrophies (22), suggesting that it is probably not because of dystrophic changes of myofibers but rather because of the lack of sialic acids. This abnormal glycosylation apparently preceded the formation of rimmed vacuoles, which is a pathological hallmark of DMRV. These rimmed vacuoles were electron-microscopically recognized as focal accumulations of autophagic vacuoles, which sometimes surround degenerated myofibrils and amyloid deposits. However, it is unknown whether the focal accumulations of autophagic vacuoles are the cause or result of the degeneration of myofibrils and amyloid deposits. Hypo-sialylation and abnormal glycosylation could cause the misfolding of some glycoproteins, and thus these misfolded glycoproteins may be targets of autophagic degradation and also behave as cores for formation of amyloid deposits. In our study, dystroglycan and SSA lectin staining was variable among patients. One possible explanation is that sialylation may not be the direct cause of the disease. For example, the loss of GNE enzymatic activity may induce the accumulation of the substrate, UDP-GlcNAc, leading to the abnormal O-GlcNAc modification of various proteins in the cells (23). Nevertheless, this possibility may also be unlikely because the overexpression of O-GlcNAc transferase did not cause any morphological abnormality in skeletal muscles in mice (24). In the next step, further analyses using animal models, as well as further testing of therapy with ManNAc or NeuAc, will be necessary to clarify the pathomechanism of DMRV and HIBM and the pathway from hyposialylation to rimmed vacuole formation and muscle atrophy.