Construction of mouse cochlin mutants with different GAG-binding specificities and their use for immunohistochemistry.

Glycosaminoglycan (GAG) is a polysaccharide present on the cell surface as an extracellular matrix component, and is composed of repeating disaccharide units consisting of an amino sugar and uronic acid except in the case of the keratan sulfate. Sulfated GAGs, such as heparan sulfate, heparin, and chondroitin sulfate mediate signal transduction of growth factors, and their functions vary with the type and degree of sulfated modification. We have previously identified human and mouse cochlins as proteins that bind to sulfated GAGs. Here, we prepared a recombinant cochlin fused to human IgG-Fc or Protein A at the C-terminus as a detection and purification tag and investigated the ligand specificity of cochlin. We found that cochlin can be used as a specific probe for highly sulfated heparan sulfate and chondroitin sulfate E. We then used mutant analysis to identify the mechanism by which cochlin recognizes GAGs and developed a GAG detection system using cochlin. Interestingly, a mutant lacking the vWA2 domain bound to various types of GAGs. The N-terminal amino acid residues of cochlin contributed to its binding to heparin. Pathological specimens from human myocarditis patients were stained with a cochlin-Fc mutant. The results showed that both tryptase-positive and tryptase-negative mast cells were stained with this mutant. The identification of detailed modification patterns of GAGs is an important method to elucidate the molecular mechanisms of various diseases. The method developed for evaluating the expression of highly sulfated GAGs will help understand the biological and pathological importance of sulfated GAGs in the future.

Glucuronylated core 1 glycans are required for precise localization of neuromuscular junctions and normal formation of basement membranes on Drosophila muscles.

T antigen (Galß1-3GalNAcα1-Ser/Thr) is an evolutionary-conserved mucin-type core 1 glycan structure in animals synthesized by core 1 ß1,3-galactosyltransferase 1 (C1GalT1). Previous studies showed that T antigen produced by Drosophila C1GalT1 (dC1GalT1) was expressed in various tissues and dC1GalT1 loss in larvae led to various defects, including decreased number of circulating hemocytes, hyper-differentiation of hematopoietic stem cells in lymph glands, malformation of the central nervous system, mislocalization of neuromuscular junction (NMJ) boutons, and ultrastructural abnormalities in NMJs and muscle cells. Although glucuronylated T antigen (GlcAß1-3Galß1-3GalNAcα1-Ser/Thr) has been identified in Drosophila, the physiological function of this structure has not yet been clarified. In this study, for the first time, we unraveled biological roles of glucuronylated T antigen. Our data show that in Drosophila, glucuronylation of T antigen is predominantly carried out by Drosophila ß1,3-glucuronyltransferase-P (dGlcAT-P). We created dGlcAT-P null mutants and found that mutant larvae showed lower expression of glucuronylated T antigen on the muscles and at NMJs. Furthermore, mislocalization of NMJ boutons and a partial loss of the basement membrane components collagen IV (Col IV) and nidogen (Ndg) at the muscle 6/7 boundary were observed. Those two phenotypes were correlated and identical to previously described phenotypes in dC1GalT1 mutant larvae. In addition, dGlcAT-P null mutants exhibited fewer NMJ branches on muscles 6/7. Moreover, ultrastructural analysis revealed that basement membranes that lacked Col IV and Ndg were significantly deformed. We also found that the loss of dGlcAT-P expression caused ultrastructural defects in NMJ boutons. Finally, we showed a genetic interaction between dGlcAT-P and dC1GalT1. Therefore, these results demonstrate that glucuronylated core 1 glycans synthesized by dGlcAT-P are key modulators of NMJ bouton localization, basement membrane formation, and NMJ arborization on larval muscles.

Phenotype-based clustering of glycosylation-related genes by RNAi-mediated gene silencing.

Glycan structures are synthesized by a series of reactions conducted by glycosylation-related (GR) proteins such as glycosyltransferases, glycan-modifying enzymes, and nucleotide-sugar transporters. For example, the common core region of glycosaminoglycans (GAGs) is sequentially synthesized by peptide-O-xylosyltransferase, ß1,4-galactosyltransferase I, ß1,3-galactosyltransferase II, and ß1,3-glucuronyltransferase. This raises the possibility that functional impairment of GR proteins involved in synthesis of the same glycan might result in the same phenotypic abnormality. To examine this possibility, comprehensive silencing of genes encoding GR and proteoglycan core proteins was conducted in Drosophila. Drosophila GR candidate genes (125) were classified into five functional groups for synthesis of GAGs, N-linked, O-linked, Notch-related, and unknown glycans. Spatiotemporally regulated silencing caused a range of malformed phenotypes that fell into three types: extra veins, thick veins, and depigmentation. The clustered phenotypes reflected the biosynthetic pathways of GAGs, Fringe-dependent glycan on Notch, and glycans placed at or near nonreducing ends (herein termed terminal domains of glycans). Based on the phenotypic clustering, CG33145 was predicted to be involved in formation of terminal domains. Our further analysis showed that CG33145 exhibited galactosyltransferase activity in synthesis of terminal N-linked glycans. Phenotypic clustering, therefore, has potential for the functional prediction of novel GR genes.

Frequent glycan structure mining of influenza virus data revealed a sulfated glycan motif that increased viral infection.

MOTIVATION: It is well known influenza viruses recognize and bind terminal sialic acid (SA) on glycans that are found on the cell surface. In this work, we used a data mining technique to analyze the glycan array data of influenza viruses to find novel glycan structures other than SA that may be involved in viral infection. RESULTS: In addition to SA structures noted previously, we noted the sulfated structures in the mining results. For verification, we overexpressed the sulfotransferase that is involved in synthesizing these structures, and we performed a viral infection experiment to assess changes in infection in these cells. In our results, we found that there is a 70-fold increase in these cells compared with the control. Thus, we have found a novel pattern in glycan structures that may be involved in viral infection. AVAILABILITY AND IMPLEMENTATION: The Glycan Miner Tool is available from the RINGS resource at http://www.rings.t.soka.ac.jp.

Identification of ß1,3-galactosyltransferases responsible for biosynthesis of insect complex-type N-glycans containing a T-antigen unit in the honeybee.

Honeybees (Apis mellifera) produce unique complex-type N-glycans bearing a Galß1-3GalNAc (T-antigen) unit, and honeybee-specific N-glycans are linked to royal jelly glycoproteins. In this study, we identified two novel honeybee ß1,3-galactosyltransferase (ß1,3-GalT) genes responsible for biosynthesis of the T-antigen in insect N-glycans. The products of the two putative ß1,3-GalT genes (ß1,3-GalT1 and ß1,3-GalT2), which were expressed in Sf21 insect cells, transferred galactose (Gal) residues to GalNAc2GlcNAc2Man3GlcNAc2-PA to form the Galß1-3GalNAc unit, indicating that the identified genes were involved in biosynthesis of the ß1-3 Gal-containing N-glycan. Therefore, using biochemistry and molecular biology techniques, we revealed a unique N-glycan biosynthesis mechanism in the cephalic region of honeybees, which has not previously been found in other animal or plant cells.

Involvement of cochlin binding to sulfated heparan sulfate/heparin in the pathophysiology of autosomal dominant late-onset hearing loss (DFNA9).

Late-onset non-syndromic autosomal dominant hearing loss 9 (DFNA9) is a hearing impairment caused by mutations in the coagulation factor C homology gene (COCH). COCH encodes for cochlin, a major component of the cochlear extracellular matrix. Though biochemical and genetic studies have characterized the properties of wild-type and mutated cochlins derived from DFNA9, little is known about the underlying pathogenic mechanism. In this study, we established a cochlin reporter cell, which allowed us to monitor the interaction of cochlin with its ligand(s) by means of a ß-galactosidase assay. We found a class of highly sulfated glycosaminoglycans (GAGs), heparin, that were selectively bound to cochlin. The interaction was distinctly abrogated by N-desulfation, but not by 2-O- or 6-O-desulfation. The binding of cochlin to GAG was diminished by all of the point mutations found in DFNA9 patients. Through GAG composition analysis and immunostaining using mouse cochlin/immunoglobulin-Fc fusion protein, we identified moderately sulfated GAGs in mouse cochlea tissue; this implies that cochlin binds to such sulfated GAGs in the cochlea. Since GAGs play an important role in cell growth and survival as co-receptors of signal transduction mechanisms, the interaction of cochlin with GAGs in the extracellular matrix could aid the pathological research of autosomal dominant late-onset hearing loss in DFNA9.

Two Golgi-resident 3'-Phosphoadenosine 5'-phosphosulfate transporters play distinct roles in heparan sulfate modifications and embryonic and larval development in Caenorhabditis elegans.

Synthesis of extracellular sulfated molecules requires active 3'-phosphoadenosine 5'-phosphosulfate (PAPS). For sulfation to occur, PAPS must pass through the Golgi membrane, which is facilitated by Golgi-resident PAPS transporters. Caenorhabditis elegans PAPS transporters are encoded by two genes, pst-1 and pst-2. Using the yeast heterologous expression system, we characterized PST-1 and PST-2 as PAPS transporters. We created deletion mutants to study the importance of PAPS transporter activity. The pst-1 deletion mutant exhibited defects in cuticle formation, post-embryonic seam cell development, vulval morphogenesis, cell migration, and embryogenesis. The pst-2 mutant exhibited a wild-type phenotype. The defects observed in the pst-1 mutant could be rescued by transgenic expression of pst-1 and hPAPST1 but not pst-2 or hPAPST2. Moreover, the phenotype of a pst-1;pst-2 double mutant were similar to those of the pst-1 single mutant, except that larval cuticle formation was more severely defected. Disaccharide analysis revealed that heparan sulfate from these mutants was undersulfated. Gene expression reporter analysis revealed that these PAPS transporters exhibited different tissue distributions and subcellular localizations. These data suggest that pst-1 and pst-2 play different physiological roles in heparan sulfate modification and development.

Expression and the role of 3'-phosphoadenosine 5'-phosphosulfate transporters in human colorectal carcinoma.

Sulfation represents an essential modification for various molecules and regulates many biological processes. The sulfation of glycans requires a specific transporter for 3'-phosphoadenosine 5'-phosphosulfate (PAPS) on the Golgi apparatus. This study investigated the expression of PAPS transporter genes in colorectal carcinomas and the significance of Golgi-specific sulfation in the proliferation of colorectal carcinoma cells. The relative amount of PAPST1 transcripts was found to be higher than those of PAPST2 in colorectal cancerous tissues. Immunohistochemically, the enhanced expression of PAPST1 was observed in fibroblasts in the vicinity of invasive cancer cells, whereas the expression of PAPST2 was decreased in the epithelial cells. RNA interference of either of the two PAPS transporter genes reduced the extent of sulfation of cellular proteins and cellular proliferation of DLD-1 human colorectal carcinoma cells. Silencing the PAPS transporter genes reduced fibroblast growth factor signaling in DLD-1 cells. These findings indicate that PAPS transporters play a role in the proliferation of colorectal carcinoma cells themselves and take part in a desmoplastic reaction to support cancer growth by controlling their sulfation status.

Sulfated glycans containing NeuAcα2-3Gal facilitate the propagation of human H1N1 influenza A viruses in eggs.

When human influenza viruses are isolated and passaged in chicken embryos, variants with amino acid substitutions around the receptor binding site of hemagglutinin (HA) are selected; however, the mechanisms that underlie this phenomenon have yet to be elucidated. Here, we analyzed the receptor structures that contributed to propagation of egg-passaged human H1N1 viruses. The analysis included seasonal and 2009 pandemic strains, both of which have amino acid substitutions of HA found in strains isolated or passaged in eggs. These viruses exhibited high binding to sulfated glycans containing NeuAcα2-3Gal. In MDCK cells overexpressing the sulfotransferase that synthesize Galß1-4(SO3--6)GlcNAc, production of human H1N1 viruses was increased up to 90-fold. Furthermore, these sulfated glycans were expressed on the allantoic and amniotic membranes of chicken embryos. These results suggest that 6-sulfo sialyl Lewis X and/or NeuAcα2-3Galß1-4(SO3--6)GlcNAc are involved in efficient propagation of human H1N1 viruses in chicken embryos.

Identification of the Drosophila core 1 beta1,3-galactosyltransferase gene that synthesizes T antigen in the embryonic central nervous system and hemocytes.

T antigen (Galbeta1-3GalNAcalpha1-Ser/Thr), the well-known tumor-associated antigen, is a core 1 mucin-type O-glycan structure that is synthesized by core 1 beta1,3-galactosyltransferase (C1beta3GalT), which transfers Gal from UDP-Gal to Tn antigen (GalNAcalpha1-Ser/Thr). Three putative C1beta3GalTs have been identified in Drosophila. However, although all three are expressed in embryos, their roles during embryogenesis have not yet been clarified. In this study, we used P-element inserted mutants to show that CG9520, one of the three putative C1beta3GalTs, synthesizes T antigen expressed on the central nervous system (CNS) during embryogenesis. We also found that T antigen was expressed on a subset of the embryonic hemocytes. CG9520 mutant embryos showed the loss of T antigens on the CNS and on a subset of hemocytes. Then, the loss of T antigens was rescued by precise excision of the P-element inserted into the CG9520 gene. Our data demonstrate that T antigens expressed on the CNS and on a subset of hemocytes are synthesized by CG9520 in the Drosophila embryo. In addition, we found that the number of circulating hemocytes was reduced in third instar larvae of CG9520 mutant. We, therefore, named the CG9520 gene Drosophila core 1 beta1,3-galactosyltransferase 1 because it is responsible for the synthesis and function of T antigen in vivo.

Crystal structure of the catalytic fragment of human brain 2',3'-cyclic-nucleotide 3'-phosphodiesterase.

2',3'-Cyclic-nucleotide 3'-phosphodiesterase (CNP), a member of the 2H phosphoesterase superfamily, is firmly bound to brain white matter and found mainly in the central nervous system of vertebrates, and it catalyzes the hydrolysis of 2',3'-cyclic nucleotide to produce 2'-nucleotide. Recent studies on CNP-knockout mice have revealed that the absence of CNP causes axonal swelling and neuronal degeneration. Here, the crystal structure of the catalytic fragment (CF) of human CNP (hCNP-CF) is solved at 1.8A resolution. It is an alpha+beta type structure consisting of three alpha-helices and nine beta-strands. The structural core of the molecule is comprised of two topologically equivalent three-stranded antiparallel beta-sheets that are related by a pseudo 2-fold symmetry. Each beta-sheet contains an H-X-T-X motif, which is strictly conserved among members of the 2H phosphoesterase superfamily. The phosphate ion is bound to the side-chains of His and Thr from each of the two motifs. Structural comparison of hCNP-CF with plant 1'',2''-cyclic nucleotide phosphodiesterase (CPDase) and bacterial 2'-5' RNA ligase reveals that the H-X-T-X motifs are structurally conserved among these enzymes, but the surface properties of the active site are quite different among the enzymes, reflecting the differences in their substrates. On the basis of the present crystal structure of the hCNP-CF/phosphate complex, the available structure of the CPDase/cyclic-nucleotide analogue complex, and the recent functional studies of rat CNP-CF, we propose a possible substrate-binding mode and catalytic mechanism of CNP, which employs the nucleophilic water molecule activated by His310. The proposed mechanism is basically equivalent to the second step of the well-accepted reaction mechanism of RNase A. Since the overall structure of hCNP-CF differs considerably from that of RNase A, it is likely that the similar active sites with two catalytic histidine residues in these enzymes arose through convergent evolution.

Sulfation of keratan sulfate proteoglycan reduces radiation-induced apoptosis in human Burkitt's lymphoma cell lines.

This study focuses on clarifying the contribution of sulfation to radiation-induced apoptosis in human Burkitt's lymphoma cell lines, using 3'-phosphoadenosine 5'-phosphosulfate transporters (PAPSTs). Overexpression of PAPST1 or PAPST2 reduced radiation-induced apoptosis in Namalwa cells, whereas the repression of PAPST1 expression enhanced apoptosis. Inhibition of PAPST slightly decreased keratan sulfate (KS) expression, so that depletion of KS significantly increased radiation-induced apoptosis. In addition, the repression of all three N-acetylglucosamine-6-O-sulfotransferases (CHST2, CHST6, and CHST7) increased apoptosis. In contrast, PAPST1 expression promoted the phosphorylation of p38 MAPK and Akt in irradiated Namalwa cells. These findings suggest that 6-O-sulfation of GlcNAc residues in KS reduces radiation-induced apoptosis of human Burkitt's lymphoma cells.

3-O-sulfated heparan sulfate recognized by the antibody HS4C3 contributes [corrected] to the differentiation of mouse embryonic stem cells via fas signaling.

Maintenance of self-renewal and pluripotency in mouse embryonic stem cells (mESCs) is regulated by the balance between several extrinsic signaling pathways. Recently, we demonstrated that heparan sulfate (HS) chains play important roles in the maintenance and differentiation of mESCs by regulating extrinsic signaling. Sulfated HS structures are modified by various sulfotransferases during development. However, the significance of specific HS structures during development remains unclear. Here, we show that 3-O-sulfated HS structures synthesized by HS 3-O-sulfotransferases (3OSTs) and recognized by the antibody HS4C3 increase during differentiation of mESCs. Furthermore, expression of Fas on the cell surface of the differentiated cells also increased. Overexpression of the HS4C3-binding epitope in mESCs induced apoptosis and spontaneous differentiation even in the presence of LIF and serum. These data showed that the HS4C3-binding epitope was required for differentiation of mESCs. Up-regulation of the HS4C3-binding epitope resulted in the recruitment of Fas from the cytoplasm to lipid rafts on the cell surface followed by activation of Fas signaling. Indeed, the HS4C3-binding epitope interacted with a region that included the heparin-binding domain (KLRRRVH) of Fas. Reduced self-renewal capability in cells overexpressing 3OST resulted from the degradation of Nanog by activated caspase-3, which is downstream of Fas signaling, and was rescued by the inhibition of Fas signaling. We also found that knockdown of 3OST and inhibition of Fas signaling reduced the potential for differentiation into the three germ layers during embryoid body formation. This is the first demonstration that activation of Fas signaling is mediated by an increase in the HS4C3-binding epitope and indicates a novel signaling pathway for differentiation in mESCs.

O-sulfate groups of heparin are critical for inhibition of ecotropic murine leukemia virus infection by heparin.

There is increasing evidence that soluble glycosaminoglycans such as heparin can interfere with the infectivity of various viruses, including ecotropic murine leukemia viruses (MLVs). The ecotropic MLV, Friend MLV (F-MLV) and the neuropathogenic variants A8 MLV and PVC-211 MLV, were susceptible to heparin-mediated inhibition of infection of NIH 3T3 cells. To investigate the interaction between the envelope glycoprotein (Env) of MLV and heparin, we prepared vesicular stomatitis virus-based pseudotyped viruses carrying the Env of F-, A8, or PVC-211 MLVs. Surface plasmon resonance analyses indicated that the Env of A8 and PVC-211 MLVs had a higher binding activity to heparin than that of F-MLV. We examined the influence of N- or O-sulfation of heparin on binding activity to Env and on the inhibition of the infectivity of MLV and pseudotyped viruses carrying Env. This analysis indicated that the O-sulfate groups of heparin play a major role in determining Env-dependent inhibitory effects.

Increased apoptosis of myoblasts in Drosophila model for the Walker-Warburg syndrome.

Walker-Warburg syndrome, a progressive muscular dystrophy, is a severe disease with various kinds of symptoms such as muscle weakness and occasional seizures. The genes of protein O-mannosyltransferases 1 and 2 (POMT1 and POMT2), fukutin, and fukutin-related protein are responsible for this syndrome. In our previous study, we cloned Drosophila orthologs of human POMT1 and POMT2 and identified their activity. However, the mechanism of onset of this syndrome is not well understood. Furthermore, little is known about the behavioral properties of the Drosophila POMT1 and POMT2 mutants, which are called rotated abdomen (rt) and twisted (tw), respectively. First, we performed various kinds of behavioral tests and described in detail the muscle structures by using these mutants. The mutant flies exhibited abnormalities in heavy exercises such as climbing or flight but not in light movements such as locomotion. Defective motor function in mutants appeared immediately after eclosion and was exaggerated with aging. Along with motor function, muscle ultrastructure in the tw mutant was altered, as seen in human patients. We demonstrated that expression of RNA interference (RNAi) for the rt gene and the tw mutant was almost completely lethal and semi-lethal, respectively. Flies expressing RNAi had reduced lifespans. These findings clearly demonstrate that Drosophila POMT mutants are models for human muscular dystrophy. We then observed a high density of myoblasts with an enhanced degree of apoptosis in the tw mutant, which completely lost enzymatic activity. In this paper, we propose a novel mechanism for the development of muscular dystrophy: POMT mutation causes high myoblast density and position derangement, which result in apoptosis, muscle disorganization, and muscle cell defects.

The 3'-phosphoadenosine 5'-phosphosulfate transporters, PAPST1 and 2, contribute to the maintenance and differentiation of mouse embryonic stem cells.

Recently, we have identified two 3'-phosphoadenosine 5'-phosphosulfate (PAPS) transporters (PAPST1 and PAPST2), which contribute to PAPS transport into the Golgi, in both human and Drosophila. Mutation and RNA interference (RNAi) of the Drosophila PAPST have shown the importance of PAPST-dependent sulfation of carbohydrates and proteins during development. However, the functional roles of PAPST in mammals are largely unknown. Here, we investigated whether PAPST-dependent sulfation is involved in regulating signaling pathways required for the maintenance of mouse embryonic stem cells (mESCs), differentiation into the three germ layers, and neurogenesis. By using a yeast expression system, mouse PAPST1 and PAPST2 proteins were shown to have PAPS transport activity with an apparent K(m) value of 1.54 microM or 1.49 microM, respectively. RNAi-mediated knockdown of each PAPST induced the reduction of chondroitin sulfate (CS) chain sulfation as well as heparan sulfate (HS) chain sulfation, and inhibited mESC self-renewal due to defects in several signaling pathways. However, we suggest that these effects were due to reduced HS, not CS, chain sulfation, because knockdown of mouse N-deacetylase/N-sulfotransferase, which catalyzes the first step of HS sulfation, in mESCs gave similar results to those observed in PAPST-knockdown mESCs, but depletion of CS chains did not. On the other hand, during embryoid body formation, PAPST-knockdown mESCs exhibited abnormal differentiation, in particular neurogenesis was promoted, presumably due to the observed defects in BMP, FGF and Wnt signaling. The latter were reduced as a result of the reduction in both HS and CS chain sulfation. We propose that PAPST-dependent sulfation of HS or CS chains, which is regulated developmentally, regulates the extrinsic signaling required for the maintenance and normal differentiation of mESCs.

Structural comparison analysis of 2H phosphodiesterase family proteins.

2',3'-Cyclic-nucleotide 3'-phosphodiesterase (CNP) is found mainly in the central nervous system of vertebrates and catalyzes the hydrolysis of 2',3'-cyclic nucleotides to produce 2'-nucleotides in vitro. Recently, Several 2H phosphodiesterase super family protein structures have been determined by X-ray crystallography and NMR spectroscopy. Here we report the structure-function relationship studies of two hydrophobic residues in CNP family proteins.

Crystallization and preliminary X-ray crystallographic studies of human 2',3'-cyclic nucleotide 3'-phosphodiesterase.

The catalytic fragment of human 2',3'-cyclic nucleotide 3'-phosphodiesterase (hCNP-CF) has been crystallized by the hanging-drop vapour-diffusion method using polyethylene glycol 300 as the precipitating agent. The crystals belong to the orthorhombic space group P2(1)2(1)2(1), with unit-cell parameters a = 44.39, b = 55.35, c = 78.76 A. There is one molecule per asymmetric unit. The crystals diffract to at least 1.8 A resolution using synchrotron radiation and are suitable for X-ray structure analysis at high resolution.

Three-dimensional structure of a cyclic-nucleotide phosphodiesterase from human brain.

2',3'-Cyclic-nucleotide 3'-phosphodiesterase (CNP) is found mainly in the central nervous system of vertebrates and catalyzes the hydrolysis of 2',3'-cyclic nucleotides to produce 2'-nucleotides in vitro. Recently, CNP has been identified as a member of the 2H phosphoesterase superfamily. Here we have determined the crystal structure of the catalytic fragment of human CNP (hCNP-CF) at 1.3 A resolution.

The twisted abdomen phenotype of Drosophila POMT1 and POMT2 mutants coincides with their heterophilic protein O-mannosyltransferase activity.

Walker-Warburg syndrome, caused by mutations in protein O-mannosyltransferase-1 (POMT1), is an autosomal recessive disorder characterized by severe brain malformation, muscular dystrophy, and structural eye abnormalities. As humans have a second POMT, POMT2, we cloned each Drosophila ortholog of the human POMT genes and carried out RNA interference (RNAi) knock-down to investigate the function of these proteins in vivo. Drosophila POMT2 (dPOMT2) RNAi mutant flies showed a "twisted abdomen phenotype," in which the abdomen is twisted 30-60 degrees , similar to the dPOMT1 mutant. Moreover, dPOMT2 interacted genetically with dPOMT1, suggesting that the dPOMTs function in collaboration with each other in vivo. We expressed dPOMTs in Sf21 cells and measured POMT activity. dPOMT2 transferred a mannose to the dystroglycan protein only when it was coexpressed with dPOMT1. Likewise, dPOMT1 showed POMT activity only when coexpressed with dPOMT2, and neither dPOMT showed any activity by itself. Each dPOMT RNAi fly totally reduced POMT activity, despite the specific reduction in the level of each dPOMT mRNA. The expression pattern of dPOMT2 mRNA was found to be similar to that of dPOMT1 mRNA using whole mount in situ hybridization. These results demonstrate that the two dPOMTs function as a protein O-mannosyltransferase in association with each other, in vitro and in vivo, to generate and maintain normal muscle development.