DPM1, the Catalytic Subunit of Dolichol-phosphate Mannose Synthase, Is Tethered to and Stabilized on the Endoplasmic Reticulum Membrane by DPM3*

Dolichol-phosphate mannose (DPM) synthase is required for synthesis of the glycosylphosphatidylinositol (GPI) anchor, N-glycan precursor, protein O-mannose, and C-mannose. We previously identified DPM3, the third component of this enzyme, which was co-purified with DPM1 and DPM2. Here, we have established mutant Chinese hamster ovary (CHO) 2.38 cells that were defective in DPM3. CHO2.38 cells were negative for GPI-anchored proteins, and microsomes from these cells showed no detectable DPM synthase activity, indicating that DPM3 is an essential component of this enzyme. A coiled-coil domain near the C terminus of DPM3 was important for tethering DPM1, the catalytic subunit of the enzyme, to the endoplasmic reticulum membrane and, therefore, was critical for enzyme activity. On the other hand, two transmembrane regions in the N-terminal portion of DPM3 showed no specific functions. DPM1 was rapidly degraded by the proteasome in the absence of DPM3. Free DPM1 was strongly associated with the C terminus of Hsc70-interacting protein (CHIP), a chaperone-dependent E3 ubiquitin ligase, suggesting that DPM1 is ubiquitinated, at least in part, by CHIP.

Dolichol phosphate mannose (DPM) 2 is the sole donor substrate for mannosyltransferases working on the luminal side of the endoplasmic reticulum (ER) that transfer a mannose (Man) residue onto lipid-linked oligosaccharides of both N-glycan and glycosylphosphatidylinositol (GPI)-anchor precursors. Three mannosyltransferases for N-glycan, ALG3, ALG9, and ALG12 (1), and four mannosyltransferases for the GPI anchor, PIG-M/PIG-X (2, 3), PIG-V (4), PIG-B (5), and SMP3 (6), are known to use DPM. All of these enzymes, except for PIG-X (an essential subcomponent of GPI-mannosyltransferase-I), have multitransmembrane topologies distinct from Golgi-resident glycosyltransferases, which are type II transmembrane proteins, and form a superfamily of glycosyltransferases (7,8). In addition, the protein O-mannosylation and C-mannosylation that occur in the ER lumen have also been shown to be DPM-dependent (9,10).
DPM is synthesized from dolichol phosphate and GDP-Man on the cytosolic surface of the ER membrane by DPM synthase (GDP-␣-Man: dolichol-phosphate ␤-mannosyltransferase; EC 2.4.1.83) and then is flipped onto the luminal side and used as a donor substrate. In lower eukaryotes, such as Saccharomyces cerevisiae and Trypanosoma brucei, DPM synthase consists of a single component (Dpm1p and TbDpm1, respectively) that possesses one predicted transmembrane region near the C terminus for anchoring to the ER membrane (11,12). In contrast, the Dpm1 homologues of higher eukaryotes, namely fission yeast, fungi, and animals, have no transmembrane region (13), suggesting the existence of adapter molecules for membrane anchoring. In fact, we previously demonstrated that S. cerevisiae Dpm1 was able to complement two different DPM synthase-defective mutants, namely a Thy-1-negative class E mutant of mouse T-cell lymphoma and Chinese hamster ovary (CHO) Lec15 cells, whereas human DPM1 was only able to complement the class E mutant (14). We cloned a second component of DPM synthase, DPM2, which is the gene responsible for CHO Lec15 cells (15), and then further identified a third component, DPM3, in the biochemically purified active enzyme complex (16). Mammalian DPM1 was predicted to be a soluble protein and found to be associated with two small membrane proteins, DPM2 and DPM3. Our previous analyses using mammalian cells revealed that: (i) DPM2 stabilizes the expression of DPM3; (ii) DPM3 binds both DPM2 and DPM1; (iii) the N-terminal transmembrane regions of DPM3 are important for binding to DPM2; and (iv) the C-terminal hydrophilic region of DPM3 is important for binding to DPM1 (15,16). Thus, DPM3 is most likely to be the anchoring molecule that tethers DPM1 to the ER membrane, although direct evidence was missing due to the lack of DPM3 mutant or knockout cells. In the present study, we established a DPM3 mutant CHO cell line for the first time and addressed the biological functions of DPM3 using this mutant. Here we describe the essential role of the C-terminal coiled-coil domain of DPM3 and the fate of DPM1 in DPM3-defective mutant cells.
Establishment of GPI-anchored Protein-defective CHO Mutant Cells-The parental CHO cell line F21 was established by stable transfection of the wild-type CHO K1 cells with cDNAs of GPI-anchored protein markers (human CD59 and DAF) and previously known GPI biosynthetic enzymes (DPM1, DPM2, SL15, PIG-A, PIG-L, PIG-V, PIG-B, PIG-N, PIG-F, GAA1, and PIG-U) as described previously (3). CHO F21 cells (3 ϫ 10 7 ) were mutagenized with 1.2 g/ml N-methyl-NЈ-nitro-N-nitrosoguanidine (Nacalai Tesque) for 2 days, cultured in fresh medium for 1 day, transferred to 12-well plates, and cultured for 6 days. Next, the cells were treated with 2.0 nM proaerolysin from Aeromonas hydrophila (Protox Biotech) for 3 days and 2.0 nM ␣-toxin from Clostridium septicum (a gift from Dr. N. Sugimoto, Osaka University) for 3 days. Surviving cells were isolated by limiting dilution.
Transfection of Cells-Cells were transfected by either electroporation or lipofection. For electroporation, CHO cells (4 ϫ 10 6 ) suspended in 0.4 ml of culture medium were electroporated at 260 V and 1,000 microfarads with 10 g of DNA using a Gene Pulser (Bio-Rad). For lipofection, CHO cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.
Labeling of GPI Precursors and Dolichol-linked Oligosaccharides-For labeling with Man, CHO cells (1 ϫ 10 6 ) were preincubated in medium containing 100 g/ml glucose and 10 g/ml tunicamycin (Sigma) for 1 h and then incubated in the same medium containing 40 Ci/ml D-[2-3 H]Man (American Radiolabeled Chemicals) for 45 min. Lipids were extracted from the cells with chloroform:methanol (2:1, v/v) and then partitioned into water-saturated 1-butanol. The extracts were applied to a silica gel TLC plate (Merck), and developed with chloroform:methanol:water (10:10:3, v/v/v) to analyze the GPI anchor precursors. The radiolabeled lipids were detected using a Fuji BAS1500 image analyzer (Fuji Film).
For analysis of dolichol-linked N-glycan precursors, cells were metabolically labeled with [ 3 H]Man in the absence of tunicamycin. Dolichollinked oligosaccharides were extracted and hydrolyzed with 0.1 M HCl, 80% tetrahydrofuran at 65°C for 30 min to release the oligosaccharides. After neutralization, the oligosaccharides were analyzed by TLC using 1-propanol:acetic acid:water (3:3: Cloning of Chinese Hamster DPM3 cDNA-The cDNA of Chinese hamster (Cricetulus griseus) DPM3 was cloned by degenerate reverse transcription-PCR using mRNA prepared from CHO K1 cells as a template and the following primers: forward, 5Ј-GACGAARTTARCRCA-GTGGCTTTGGGGACT; reverse, 5Ј-TCGGGCCTCCWSKATCTG-GCTCTGCAGCTC. The 5Ј and 3Ј sequences were obtained by 5Ј and 3Ј rapid amplification of cDNA ends (RACE) PCR. The GenBank TM accession number of Chinese hamster DPM3 cDNA is AB219149.
Plasmids-Site-directed mutagenesis of pME/FLAG-DPM3 was carried out using a PCR-based method. To construct pME/3HSV-DPM3tail and pME/GST-DPM3tail, a DNA fragment encoding the C-terminal 35 amino acids (Gly 58 -Phe 92 ) of DPM3 was amplified with a specific forward primer containing a SalI site (5Ј-AAGTCGACGGTG-CAGGCTATCGTGTGGCCACTTTTCATG; the SalI site is underlined) and a reverse vector primer (5Ј-GTTAACAACAACAATTG-CATTCAT) using pME/FLAG-DPM3 as a template and then was ligated into the pME/3HSV and pME/GST vectors, respectively. To prepare the chimeric construct pME/PIG-Ltm-DPM3tail, a DNA fragment encoding the N-terminal 30 amino acids (Met 1 -Ser 30 ) of PIG-L was amplified with a forward SR␣ primer (5Ј-TGACCCTGCTTGCT-CAACTCTACG) and a specific reverse primer containing a SalI site (5Ј-AAGTCGACACTCTTCATTCGTTCTGAGGAG; the SalI site is underlined) using pME/py/PIG-L as a template, and the PCR product was used to replace the 3HSV sequence of pME/3HSV-DPM3tail. Human CHIP (C terminus of Hsc70-interacting protein) cDNA was amplified by PCR using a cDNA library prepared from Hep3B cells as a template.
Affinity Precipitation and Western Blotting-Cells were lysed in TEN buffer (1.0% Nonidet P-40, 50 mM Tris-HCl, pH 7.7, 5.0 mM EDTA, 150 mM NaCl, Complete protease inhibitor mixture (Roche Applied Science)), and cell debris was removed by ultracentrifugation (100,000 ϫ g for 1 h). For immunoprecipitation of FLAG-tagged proteins, anti-FLAG M2-agarose (Sigma) was added to the supernatant and rotated at 4°C for 2 h. For immunoprecipitation of HSV-tagged or GST-tagged proteins, anti-HSV (Novagen) or anti-GST (Amersham Biosciences) antibodies plus protein G-Sepharose (Amersham Biosciences) were added. The beads were collected and washed twice with TEN buffer. Next, the absorbed proteins were eluted with reducing SDS sample buffer. Aliquots of the samples were separated in a 10 -20% gradient SDS-polyacrylamide gel and electroblotted onto a polyvinylidene difluoride membrane. The membrane was treated with a primary antibody and then with horseradish peroxidase-conjugated protein G (Bio-Rad). Detection was carried out using Western Lightning chemiluminescence reagents (PerkinElmer Life Sciences).

RESULTS
DPM3 Is Essential for DPM Synthase Activity-We previously identified DPM3 in the DPM synthase complex that was affinity-purified using DPM1 double-tagged with GST and FLAG (16). Because neither mutant nor knock-out cells of DPM3 have been established to date, it is unknown whether DPM3 is essential for the function of DPM synthase. To obtain DPM3 mutant cells, we screened GPI-anchored protein-defective mutant cells, which were generated from the parental CHO F21 cells by chemical mutagenesis, followed by selection with A. hydrophila proaerolysin and/or C. septicum ␣-toxin, both of which are GPI anchorrecognizing pore-forming cytolytic toxins (3). We found one mutant cell line, CHO2.38, in which surface expressions of the GPI-anchored proteins CD59, DAF, and uPAR were markedly decreased compared with the parental cells (Fig. 1A). The surface expressions of these GPIanchored proteins were restored by transient transfection of DPM3 cDNA but not by transfection of DPM1 or DPM2 cDNA (Fig. 1B). To confirm the presence of a mutation in the DPM3 gene, we sequenced and compared the DPM3 mRNAs from CHO K1 and CHO2.38 cells. Chinese hamster DPM3 was found to encode 92 amino acid residues, as is the case for humans and mice, and showed 92% (85/92) and 97% (89/92) identities with these species, respectively. By comparison of the nucleotide sequences, we found an 8-base deletion at nucleotides 108 -115 in the DPM3 cDNA from CHO2.38 cells (Fig. 1C). This mutation, which causes a frameshift, may completely abolish the function of DPM3. CHO2.38 cells stably transfected with pME/puro/DPM3 (CHO2.38/DPM3 cells) showed the same expression levels of surface GPI-anchored proteins as the parental CHO F21 cells (Fig. 1D). In vitro enzyme assays using microsomes revealed that CHO2.38 cells did not synthesize either DPM or Man-containing GPI anchor precursors (Fig.  1E, lane 3), whereas CHO2.38/DPM3 cells produced more DPM than CHO F21 cells (Fig. 1E, lane 2 versus lane 1). Furthermore, CHO2.38 cells did not synthesize the mature N-glycan precursor Dol-P-P-GlcNAc 2 Man 9 Glc 3 and, instead, accumulated two immature precursors (Fig. 1F, lane 2). The upper large spot (Fig. 1F, lane 2) was GlcNAc 2 Man 5 , which was identical to the spot accumulated in CHO  (17)). The lower weak spot (Fig. 1F, lane 2) was probably GlcNAc 2 Man 5 Glc 3 . Taken together, these results indicate that the CHO2.38 cell line is a DPM3-defective mutant and that DPM3 is an essential component for DPM synthase activity.
A Cytosolic Coiled-coil Domain Near the C Terminus of DPM3 Is Critical for DPM Synthase Activity-DPM3 is a small protein that consists of 92 amino acid residues. In silico analyses of human DPM3 pre-dicted the presence of two transmembrane regions (Trp 9 -Pro 31 and Glu 36 -Gly 58 ) in the N-terminal portion and a coiled-coil domain (Glu 68 -Arg 88 ) in the C-terminal hydrophilic portion ( Fig. 2A). Because a FLAG tag attached to the N terminus of DPM3 faced the cytosolic surface of the ER membrane, as revealed by immunomicroscopic analysis (data not shown), the C-terminal coiled-coil domain must also be oriented toward the cytosol. To form such a coiled-coil structure, regularly aligned hydrophobic amino acid residues need to be on the same side of an ␣-helix.   JANUARY 13, 2006 • VOLUME 281 • NUMBER 2

JOURNAL OF BIOLOGICAL CHEMISTRY 899
homologues from various eukaryotes revealed four hydrophobic amino acid residues, Leu 74 , Ile 78 , Ala 81 , and Leu 85 , that were conserved (Fig. 2B) and present on the same side of the ␣-helix (Fig. 2C). Thus, we introduced point mutations into three of these aliphatic residues, namely Leu 74 , Ile 78 , and Leu 85 , and analyzed the biological activities of the mutants. The constructs were FLAG-tagged at the N terminus, and their expressions were confirmed to be similar by Western blotting (data not shown). We transiently transfected CHO2.38 cells with the mutant constructs and analyzed their surface expressions of CD59 by FACS (Fig. 2D). The single-point mutants I78T and L85S and the double-point mutant I78T/L85S showed nearly the same activities as wildtype DPM3, whereas the single-point mutant L74S showed much lower restoration and the two double-point mutants L74S/I78T and L74S/ L85S showed only slight restoration. The triple-point mutant L74S/ I78T/L85S, designated DPM3(triple), completely lost its activity and was similar to DPM3⌬C, which lacked the C-terminal 27 amino acids (16). These results indicate that hydrophobic residues in the coiled-coil domain of DPM3 are very important for DPM synthase activity.
The C-terminal Coiled-coil Domain of DPM3 Tethers DPM1-To test whether these three hydrophobic residues in the coiled-coil domain of DPM3 function by interacting with the catalytic subunit DPM1, we analyzed the binding of DPM3 to DPM1 by co-immunoprecipitation experiments. CHO K1 cells were transiently co-transfected with pME/3HSV-DPM1 and either pME/FLAG-DPM3(wt), pME/FLAG-DPM3(triple), or pME/FLAG-DPM3⌬C. The cells were then lysed in 1.0% digitonin, and the FLAG-tagged proteins were immunoprecipitated with anti-FLAG agarose beads. FLAG-DPM3 proteins and coimmunoprecipitated 3HSV-DPM1 were detected by Western blotting. Although the three constructs of FLAG-DPM3 were expressed at similar levels (Fig. 3, top panel), 3HSV-DPM1 was only co-immunoprecipitated with FLAG-DPM3(wt) (Fig. 3, middle panel, lane 1) and not with FLAG-DPM3(triple) (Fig. 3, middle panel, lane 2) or FLAG-DPM3⌬C (Fig. 3, middle panel, lane 3). The total expression level of 3HSV-DPM1 was about 3-fold higher after co-expression with FLAG-DPM3(wt) than after co-expression with FLAG-DPM3(triple) or FLAG-DPM3⌬C (Fig. 3, bottom panel), suggesting that DPM1 expression was enhanced by DPM3. These results indi-cate that the cytosolic coiled-coil domain of DPM3 is essential for tethering DPM1 to the ER membrane. When 1.0% Nonidet P-40 was used for solubilization, co-immunoprecipitation of 3HSV-DPM1 with FLAG-DPM3(wt) was lost (data not shown), indicating that the association of these two molecules is maintained by a relatively weak hydrophobic interaction.
Transmembrane Regions of DPM3 Are Not Essential for DPM Synthase-To address the function of the two transmembrane regions of DPM3, we replaced them with a transmembrane region from another membrane protein. For this, we chose PIG-L, which encodes GlcNAcphosphatidylinositol N-deacetylase, the second-step enzyme in GPI biosynthesis, which is a type I transmembrane protein with a large C-terminal portion in the cytosol and a transmembrane region (Val 3 -Ala 25 ) near the N terminus (18,19). PIG-L is widely distributed in the ER membrane, and its ER localization is partly dependent upon the transmembrane region (19). We prepared a chimeric construct of the N-terminal portion (Met 1 -Ser 30 ) of PIG-L, including the transmembrane region and the C-terminal tail (Gly 58 -Phe 92 ) of DPM3, and designated it PIG-Ltm-DPM3tail. CHO2.38 cells were transiently transfected with pME/PIG-Ltm-DPM3tail(wt) or pME/PIG-Ltm-DPM3tail(triple), and their surface expressions of CD59 were analyzed by FACS. Interestingly, PIG-Ltm-DPM3tail(wt) restored the surface expression of CD59 similar to that of the wild-type DPM3 (Fig. 4A, panels 1 and 2). PIG-Ltm-DPM3tail(triple) and PIG-L showed no restoration, as expected (Fig.  4A, panels 3 and 4). These results suggest that the two transmembrane regions of DPM3 have no specific function in terms of the DPM synthesis required for the GPI anchor.
Next, we expressed soluble forms of the DPM3tail constructs. CHO2.38 cells were transiently transfected with the wild-type and triple-mutant forms of 3HSV-DPM3tail and GST-DPM3tail. All of the soluble constructs were strongly expressed (data not shown). Surprisingly, both the 3HSV-DPM3tail(wt) and GST-DPM3tail(wt) constructs showed some activity (Fig. 4B, panels 2 and 4), whereas the triple-mutant constructs did not (Fig. 4B, panels 3 and 5). These results indicate that DPM1 is partially active in the presence of the coiled-coil domain of DPM3, even in a soluble form, and that the transmembrane regions of DPM3 are not essential. DPM1 Is Degraded by the Proteasome in DPM3-defective Cells-The expression of DPM1 was enhanced by co-transfection of the wild-type DPM3 into CHO K1 cells (Fig. 3), suggesting that DPM3 stabilizes DPM1 on the ER membrane. To examine whether DPM1 is expressed in the absence of DPM3, we transiently transfected CHO2.38 cells with pME/3HSV-DPM1 in the presence or absence of pME/3HSV-DPM3. Rat microsomal aldehyde dehydrogenase (ALDH) was used as a control for transfection efficiency. After 2 days of culture, the cells were lysed, and the protein expression was analyzed by Western blotting. Unexpectedly, 3HSV-DPM1 was hardly detected in the absence of DPM3 (Fig. 5A, lane 2), whereas it was strongly expressed in the presence of DPM3 (Fig. 5A, lane 1). Similar experiments were carried out for DPM2. Unlike DPM1, DPM2 was stably expressed in the presence or absence of DPM3 (Fig. 5A, lanes 3 and 4).
Next, we speculated that DPM1 expressed in CHO2.38 cells may be dissociated from the ER membrane and degraded by the proteasome, because DPM1 is predicted to be located on the cytosolic surface of the ER membrane. To test this hypothesis, we cultured CHO K1/3HSV-DPM1 and CHO2.38/3HSV-DPM1 cells in the presence of protease inhibitors. In CHO K1/3HSV-DPM1 cells, the expression levels of 3HSV-DPM1 remained unchanged following the addition of lactacystin, an inhibitor of the proteasome, or leupeptin, an inhibitor of lysosomal proteases (Fig. 5C, lanes 1-3). In contrast, in CHO2.38/3HSV-DPM1 cells, the expression of 3HSV-DPM1 was restored after the addition of lactacystin (Fig. 5C, lane 5) but not leupeptin (Fig. 5C, lane  6). These results strongly suggest that DPM1 is highly unstable and degraded by the proteasome in the cytosol when DPM3 is absent. It is worth noting that CD59 expression was not restored at all in CHO2.38/ 3HSV-DPM1 cells following the treatment with lactacystin (Fig. 5D,  panels 2 and 4), although the level of DPM1 protein was similar to that in CHO2.38/3HSV-DPM1 cells transiently transfected with DPM3 cDNA (Fig. 5C, lane 5, versus Fig. 5B, lane 3). These results suggest that the coiled-coil domain of DPM3 is necessary not only for the stabilization of DPM1 but also for its enzyme activity.
Free DPM1 Strongly Interacts with CHIP, a Chaperone-dependent E3 Ubiquitin Ligase-There are several hundred known candidates for E3 ubiquitin ligases in mammalian genomes, and these define the specificity of the target protein degradation by the proteasome. To explore which ubiquitin ligase is responsible for the degradation of DPM1, we focused on CHIP, which was originally discovered as a co-chaperone containing a tetratricopeptide repeat domain (20) and later identified to be a chaperone-dependent U-box-type E3 ubiquitin ligase (21,22). We transiently transfected CHO2.38 cells with GST-CHIP and either 3HSV-ALDH as a control or 3HSV-DPM1. GST-CHIP was strongly expressed in all transfectants (Fig. 6, bottom panel). 3HSV-ALDH was also strongly expressed in the presence or absence of lactacystin (Fig. 6, top panel, lanes 1-4; lanes 3 and 4 contain 10% of the loading of lanes 1 and 2, respectively). However, very little GST-CHIP co-immunoprecipitated with 3HSV-ALDH (Fig. 6, middle panel, lanes 1 and 2), and it was hardly detected in the lanes with the 10% loadings (Fig. 6, middle  panel, lanes 3 and 4). In contrast to ALDH, the expression of 3HSV-DPM1 was very weak in the absence of lactacystin and increased after the addition of lactacystin (Fig. 6, top panel, lanes 5 and 6, respectively). Despite its weak expression, GST-CHIP was much more strongly coimmunoprecipitated with 3HSV-DPM1 than with ALDH (Fig. 6, middle panel, lanes 5 and 6 versus lanes 3 and 4). These results strongly suggest that free DPM1 is ubiquitinated by CHIP.

DISCUSSION
Four of nine Man residues in the mature precursor of N-glycan are transferred from DPM in the ER lumen. After the glycan is transferred to proteins, all four Man residues are sequentially removed by ER ␣-mannosidase and Golgi ␣-mannosidases I and II. These residues are not essential for the viability of cultured mammalian cells but are important for quality control of glycoprotein production in the ER. It has been lethality (27). This observation suggests that complete loss of DPM synthase activity may cause death in early embryogenesis. Specific defects in protein O-mannosyltransferase, which is also DPM-dependent, cause certain types of congenital muscular dystrophies with neuronal migration disorders, such as Walker-Warburg syndrome (28). Therefore, taken together, DPM synthesis is critical for multicellular functions.
DPM synthases of higher eukaryotes consist of three subunits. It is difficult to understand how the DPM synthase complex is assembled and disassembled and why this enzyme complex evolved from a singlecomponent enzyme found in some lower eukaryotes. We previously showed that DPM2 stabilized the expression of DPM3. On the other hand, however, DPM3 did not affect the expression of DPM2 (Fig. 5A). Because DPM2 is a multifunctional regulator, which is also found in the GPI-GlcNAc-transferase complex (29), it could be expressed independently. By co-immunoprecipitation experiments, DPM3 was confirmed to be the tethering molecule for DPM1, and its cytosolic coiled-coil domain was found to be important for this interaction (Fig. 3). In contrast to the importance of the DPM3 coiled-coil domain, the function of the two transmembrane regions of DPM3 remains unclear. Our previous analyses showed that DPM2 and DPM3 associated via their transmembrane regions and formed an active enzyme complex (16). However, the chimeric protein PIG-Ltm-DPM3tail functioned in CHO2.38 cells (Fig. 4A), despite its inability to associate with DPM2 (data not shown). This indicates that the stable association of DPM2 and DPM3 is not necessary for the enzyme activity in vivo. The chimera did not work well in DPM2-null mutant CHO Lec15 cells (data not shown), suggesting that DPM2 has some functions other than stabilizing DPM3. Because there is evidence that DPM2 binds to dolichol phosphate (15), it may be required for recruitment of lipid substrates.
DPM3 appeared to be a limiting factor for the formation of the DPM synthase complex, because the enzyme activity was enhanced by overexpression of DPM3 in CHO2.38 cells (Fig. 1E). When the amount of DPM3 is limited or depleted, excess DPM1 may be degraded by the proteasome, as demonstrated in CHO2.38 cells (Fig. 5). Although DPM1 does not have a transmembrane region, it does possess a hydrophobic peptide sequence near the C terminus that functions in its binding to DPM3 (15,16). Thus, DPM1 dissociated from DPM3 must bear the hydrophobic peptide on the surface of the molecule, which could be a binding target for cytosolic chaperones. We demonstrated that free DPM1 formed a complex with an E3 ubiquitin ligase, CHIP (Fig. 6), which is known to bind an E2 ligase through its U-box domain and is also able to bind the cytosolic chaperones Hsc70/Hsp70 or Hsp90 through its tetratricopeptide repeat domain. It has been reported that CHIP accelerates the degradation of several cytosolic and membrane proteins and also that CHIP prefers misfolded proteins over correctly folded proteins (30,31). The interaction between DPM3 and DPM1 must be relatively weak, because it was stable in digitonin but labile in Nonidet P-40 (Fig. 3), suggesting that some intracellular stimuli may dissociate these two molecules under physiological conditions. It has been reported that cAMP-dependent protein kinase regulates mammalian DPM synthase activity by phosphorylating a serine residue in DPM1 (32,33). The conformational changes of DPM1 caused by such post-translational modifications may trigger dissociation of the DPM synthase complex.