Heparanase is overexpressed in many solid tumor cells and is capable of specifically cleaving heparan sulfate, and this activity is associated with the metastatic potential of tumor cells; however, the activation mechanism of heparanase has remained unknown. In this study, we investigated the link between disulfide bond formation and the activation of heparanase in human tumor cells. Mass spectrometry analysis of heparanase purified from a conditioned medium of human fibrosarcoma cells revealed two disulfide bonds, Cys127-Cys179 and Cys437-Cys542, and one S-cysteinylation at the Cys211 residue. It was shown that, although the formation of the Cys127-Cys179 bond and S-cysteinylation at Cys211 have little effect on heparanase function, the disulfide bond between Cys437 and Cys542 is necessary for the secretion and activation of heparanase. Thus, the present findings will provide a basis for the further refinement of heparanase structural studies and for the development of novel heparanase inhibitors. [Cancer Res 2007;67(16):7841–9]
Heparan sulfate and heparan sulfate proteoglycans, which are located in the extracellular matrix and on the external surface of cell membranes, play a major role in cell-cell and cell-extracellular interactions (1–4). Because heparan sulfate chains tightly bind to a diverse repertoire of proteins under physiologic conditions, the enzymatic cleavage of heparan sulfate by heparanase is likely to be involved in several biological phenomena, including cancer metastasis and angiogenesis (5–8).
Human heparanase cDNA, which encodes endo-β-d-glucuronidase, contains 543 amino acids with six N-glycosylation sites and five Cys residues (9–12). Heparanase is synthesized as a 65-kDa precursor protein, followed by its secretion into the extracellular space. After secretion, heparanase is internalized and undergoes proteolytic processing, thus yielding an 8-kDa polypeptide (Gln36-Glu109) at the NH2 terminus and a 50-kDa polypeptide (Lys158-Ile543) at the COOH terminus, which heterodimerize to form the active heparanase enzyme with an intervening 6-kDa peptide (Ser110-Gln157), which is excised by proteolysis (13–15) by lysosomal/endosomal proteins (16). Interestingly, it has been reported that heparanase is translocated into the nucleus, degrades nuclear heparan sulfate, and thereby affects nuclear functions that are thought to be regulated by heparan sulfate (17, 18). Overexpression of heparanase has been observed in many human tumor types, such as those in the head and neck (19), pancreatic tumors (20), hepatocellular carcinoma (21), esophageal carcinoma (22), and cultured human tumor cell lines (19, 23); such associations are thought to indicate the involvement of heparanase in tumor progression. Although we previously showed that N-glycosylation is required for the secretion of heparanase protein in cultured cells (12), other posttranslational modifications responsible for the regulation of heparanase function have yet to be elucidated. To develop heparanase inhibitors, it will be necessary to understand the activation mechanisms of heparanase in human tumor cells.
The complex process of disulfide bond formation in the endoplasmic reticulum of eukaryotic cells was one of the first mechanisms of catalyzed protein folding to be discovered. Disulfide bonds influence the thermodynamics of protein folding and maintain protein integrity. Most proteins that function in the extracellular milieu contain disulfide bonds, which are covalent links between pairs of appropriate Cys residues (24–26). Many functions have been characterized in terms of disulfide bond formation, including stabilization (e.g., troponin C), secretion (e.g., antithrombin), and the induction of conformational changes that lead to high-affinity binding to the target protein (e.g., CCR2; refs. 27–29). Previously, the disulfide bonds present in mature proteins were considered to be inert; that is, once formed, they were thought to remain unchanged for the life of the protein. However, it now seems that this is not necessarily the case; it has been shown that disulfide bonds can in some cases be cleaved in mature proteins, and, when this happens, there are significant consequences with respect to protein function (30).
In this study, we show that two disulfide bonds, Cys127-Cys179 and Cys437-Cys542, are formed in heparanase, which also has an S-cysteinylation site at Cys211. It was found that the disulfide bond at Cys127-Cys179 and the S-cysteinylation at Cys211 exert only negligible effects on heparanase function, whereas the disulfide bond formation between Cys437 and Cys542 is necessary for the secretion and activation of heparanase. Therefore, our findings suggest that in the case of heparanase, inhibitors of disulfide bond formation may provide effective therapeutic drugs for human cancers that have been shown to express heparanase.
Materials and Methods
Establishment of heparanase-overexpressing stable cell lines. To obtain heparanase-overexpressing cell lines, we carried out gene transfection according to a previously described method (31–33). The human heparanase gene was cloned into pcDNA3.1/Myc-His(+) vectors (Invitrogen). Permanent cell lines expressing heparanase were established by transfecting pcDNA3.1/Myc-His(+)-heparanase into HT1080 cells, followed by selection with 400 μg/mL G418. Cells transfected with the pcDNA3.1/Myc-His(+) vector were designated HT1080-Neo, and the clone cells that expressed high levels of heparanase were designated HT1080-HP-MH cells.
Purification of the heparanase protein. Exponentially growing cells were washed with serum-free DMEM thrice, and were then cultured in serum-free DMEM. After 24 h, the conditioned medium (150 mL) was collected, and Ni-NTA agarose (Qiagen GmbH) was added to the conditioned medium for 1 h at 4°C. The Ni-NTA agarose was washed thrice with PBS, and Ni-NTA agarose-bound heparanase was eluted with PBS containing 300 mmol/L imidazole. The obtained samples were electrophoresed on SDS-polyacrylamide gels without any reducing reagents. The protein bands were visualized by a Brilliant Blue R staining procedure (Sigma).
Mass spectrometry. To identify the disulfide bond arrangements within heparanase, we did a matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass analysis with a slight modification (34–36). To determine the disulfide bond between Cys437 and Cys542, after electrophoresis under nonreducing condition, the protein-containing area was excised, washed, and then dried under reduced pressure. Reduction of the protein in the dried gel or peptide mixture of heparanase was carried out in a reductant solution [100 mmol/L DTT and 100 mmol/L NH4HCO3 or 10 mmol/L Tris-HCl (pH 7.5)] for 30 min at 57°C. Alkylation of the protein in the dried gel or peptide mixture of heparanase was carried out in an alkylating aqueous solution [100 mmol/L iodoacetamide or iodoacetate and 100 mmol/L NH4HCO3 or 10 mmol/L Tris-HCl (pH 7.5)] at 37°C for 30 min. Proteins were digested with sequencing grade modified trypsin (Promega) for 18 h at 37°C in 10 mmol/L Tris-HCl buffer (pH 7.5) and the resulting peptides were extracted. Peptides were digested with endoproteinase Asp-N (Roche Diagnostics) for 18 h at 37°C in 10 mmol/L Tris-HCl buffer (pH 7.5). Peptides were digested with peptide N-glycosidase F (PNGase F; Roche Diagnostics) for 48 h at 37°C in a buffer [500 mmol/L NaCl and 10 mmol/L sodium phosphate (pH 7.0)]. The digestion mixture was subjected to MALDI-TOF mass spectrometry.
To identify the modification of Cys211, heparanase was reduced and carboxymethylated followed by digestion with trypsin, as described above. Then, the digestion mixture was subjected to MALDI-TOF mass spectrometry.
To identify the disulfide bond between Cys127 and Cys179, purified heparanase was dissolved in denaturing buffer [8 mol/L urea and 1 mol/L Tris-HCl (pH 8.5)]. Then, the protein solution was subjected to either reduction with 10 mmol/L DTT for 2 h at 37°C followed by alkylation with 20 mmol/L iodoacetate for 30 min at 25°C in the denaturing buffer (reducing condition), or alkylation with 1 mmol/L iodoacetate for 30 min at 25°C in the denaturing buffer (nonreducing condition). Each protein solution was diluted and digested with sequencing grade modified trypsin for 18 h at 37°C. The resulting peptide mixture was loaded onto a concanavalin A agarose column (2.2 × 10 mm; Honen) equilibrated with an elution buffer [500 mmol/L NaCl and 10 mmol/L sodium phosphate (pH 7.0)], and concanavalin A–unbound peptides were eluted with the elution buffer. Concanavalin A–bound peptides were eluted with 250 mmol/L α-methyl-d-mannoside in the elution buffer and digested with PNGase F for 48 h at 37°C. All digests were subjected to MALDI-TOF mass spectrometry.
PCR mutagenesis. We substituted certain Cys residues in heparanase with Ser residues by PCR site-directed mutagenesis using the overlap extension technique (37, 38). The established cell lines were designated as HT1080-HP-MH/C127S, HT1080-HP-MH/C179S, HT1080-HP-MH/C211S, HT1080-HP-MH/C437S, and HT1080-HP-MH/C542S, which produced mutant heparanase proteins designated as HP-MH/C127S, HP-MH/C179S, HP-MH/C211S, HP-MH/C437S, and HP-MH/C542S, respectively.
Western blot analysis. Cells were lysed in lysis buffer [10 mmol/L HEPES, 142.5 mmol/L KCl, 5 mmol/L MgCl2, 1 mmol/L EGTA, 0.2% NP40, 0.1% aprotinin, and 1 mmol/L phenylmethylsulfonyl fluoride (pH 7.2)] at 4°C with sonication (39). The lysates were centrifuged at 15,000 rpm for 15 min, and the amount of protein in each lysate was measured by staining the proteins with Coomassie brilliant blue G-250 (Bio-Rad Laboratories). Loading buffer [42 mmol/L Tris-HCl (pH 6.8), 10% glycerol, 2.3% SDS, 5% 2-mercaptoethanol, and 0.002% bromophenol blue] was then added to each lysate, which was subsequently boiled for 3 min and electrophoresed on an SDS-polyacrylamide gel. The proteins were transferred to polyvinylidene difluoride membranes and immunoblotted with antiheparanase antibody (clone A00032; GenScript), which can react with the pro-form (65–70 kDa) and active-form (∼50 kDa) heparanase. Detection was done with enhanced chemiluminescence reagent (Pierce).
To compare the mobility shift of heparanase under reducing and nonreducing conditions using SDS-PAGE, the cells were washed and lysed in the loading buffer in the absence of 2-mercaptoethanol. The cell lysates were sonicated with or without DTT, and the results were resolved by SDS-PAGE.
Measurement of heparanase activity. The cells were washed, suspended in PBS (pH 6.0), and disrupted by three cycles of freezing at −70°C and thawing at 37°C. The lysates were centrifuged at 15,000 rpm for 15 min at 4°C. A mixture of 27 μL of cell lysate and 3 μL of heparan sulfate solution [10 mg/mL in PBS (pH 6.0)] was incubated at 37°C. Following the addition of 6 μL of sampling solution (36% glycerol, 1% bromophenol blue), 20 μL of each reaction mixture were subjected to SDS-PAGE (20%). The electrophoresed gels were soaked in water for 2 h to remove the SDS. The gels were then stained with 0.1% Alcian blue 8GX in 50% ethanol for 1 h and then destained with the 1:2:7 mixture of acetic acid-ethanol-water for 12 h (12, 40).
Fluorescence microscopy. To observe the intracellular localization of heparanase, cells grown on coverslips were fixed with 4% paraformaldehyde in PBS. After washing the cells with PBS, they were incubated in 0.1% Triton X-100 in PBS for 5 min, washed with PBS, and incubated with anti-Myc antibody (clone 9E10; Santa Cruz Biotechnology) for 1 h. Alexa 488–conjugated anti-mouse IgG (Molecular Probes) was used as the secondary antibody. The intracellular organelles were labeled with anti-GRASP65 (Golgi apparatus; Santa Cruz Biotechnology) or anti-Bip/GRP78 (endoplasmic reticulum; Santa Cruz Biotechnology) antibodies. Alexa 568–conjugated anti-rabbit IgG (Molecular Probes) was used as the secondary antibody. After being washed three additional times, the cells were incubated with 2 μg/mL Hoechst 33258 (Wako) for 5 min in the dark to stain the nuclei (12, 41, 42). Then, the cells were washed thrice with PBS and were examined using a fluorescence microscope (Olympus).
Detection of secreted proteins. To detect the secreted proteins, we used a slightly modified version of a previously described method (12, 31). Exponentially growing cells were washed and cultured with serum-free medium for 24 h. The conditioned medium was collected by centrifugation at 15,000 rpm for 15 min at 4°C. Loading buffer was added to the conditioned medium, which was subsequently boiled for 3 min and electrophoresed on SDS-polyacrylamide gels. The cell lysates were prepared as described above. The amount of heparanase protein in the cell lysates and in the culture medium was measured by Western blotting using antiheparanase antibody.
Disulfide bond formation in heparanase. The primary amino acid sequence of human heparanase contains five Cys residues (Fig. 1A). To clarify the role of these Cys residues with respect to enzymatic function, we first attempted to determine whether disulfide bond formation occurs in heparanase. Electrophoresis of heparanase from HT1080-HP-MH cells under nonreducing conditions (i.e., without DTT) revealed a more rapidly migrating form than that observed under reducing conditions (i.e., with DTT), as detected by Western blotting (Fig. 1B); these findings suggested that heparanase contains at least one disulfide bond. The pro and active form heparanase proteins migrated predominantly as a 65 to 70 kDa and ∼50 kDa bands, respectively, in the presence or absence of DTT; thus, it seemed unlikely that heparanase forms a disulfide bond either with other proteins (heteromeric) or with itself (homomeric).
Determination of cysteine modification within heparanase. To identify the linkage site(s) of disulfide bond formation within heparanase, we purified heparanase protein from a conditioned medium of cultured cells (Fig. 2A). Purified heparanase was cleaved with several peptidases, and the resulting mixture of peptides was analyzed by MALDI-TOF mass spectrometry. The peak of a fragment was observed at m/z 1,693.8, which corresponds to the disulfide peptide with a linkage between Cys437 and Cys542 (Fig. 2B , i). Upon reduction with DTT and carbamidomethylation with iodoacetamide, the fragment was resolved into two smaller peptides with masses of 1,007.4 and 803.4, corresponding to the carbamidomethylated Cys437-containing peptide and the carbamidomethylated Cys542-containing peptide. Thus, it was suggested that Cys437 and Cys542 form disulfide bond.
Heparanase is N-glycosylated, and trypsin-digested peptides (amino acids 116–128 and 194–214) containing Cys179 and Cys211 are glycosylated at Asn178 and Asn200, respectively (12). To confirm another disulfide linkage, glycosylated heparanase peptides were digested with trypsin and separated by concanavalin A agarose column chromatography. Concanavalin A–bound peptides were deglycosylated with PNGase F, which converts glycosylated Asn to Asp, and analyzed by MALDI-TOF mass spectrometry. A peak was observed at m/z 4,398.9, which corresponds to the disulfide peptide with a linkage between Cys127 and Cys179 (Fig. 2C,, i). After treatment with DTT and iodoacetate, heparanase was digested with trypsin and the resulting peptides were separated by concanavalin A agarose column chromatography. Concanavalin A–bound peptides were deglycosylated with PNGase F and subjected to MALDI-TOF mass spectrometry. A peak was observed at m/z 2,860.6, which corresponds to the carboxymethylated/deglycosylated peptide containing Cys179 (Fig. 2C,, ii). Under the same condition, in the mass spectra of concanavalin A–unbound peptides, a carboxymethylated peptide containing Cys127 was found with mass of 1,656.9 (Fig. 2C , iii). These findings indicated that Cys127 and Cys179 form a disulfide bond.
To determine the modification of the Cys211 residue, the glycol-peptide–containing Cys211 was analyzed. Two peaks of fragments were observed at m/z 2,477.1 and 2,358.1, which correspond to the S-cysteinylation at the Cys211-containing peptide and the reduced Cys211-containing peptide, respectively (Fig. 2D,, i). The sample was carboxymethylated without reduction; however, the reduced peptide containing Cys211 with mass of 2,358.1 was still observed and the carboxymethylated peptide was not observed (Fig. 2D,, i), suggesting that the cysteinyl modification was broken from the peptide chain in the MALDI ion source, as reported before (36). Moreover, the peaks at 2,477.1 and 2,358.1 were eliminated by reduction with DTT, and the carboxymethylated peptide with a mass of 2,416.3 was found upon carboxymethylation with iodoacetate (Fig. 2D , ii). These results suggest that the Cys211 residue of heparanase was not reduced but rather modified with cysteine. No ion peaks derived from other modifications of any of the five Cys residue were observed (data not shown). Taking these results together, we concluded that heparanase contains two disulfide bonds linking Cys127-Cys179 and Cys437-Cys542, and that the modification of the Cys211 residue is S-cysteinylation.
Involvement of disulfide bond formation between Cys437 and Cys542 in the activation of heparanase. To clarify the role of disulfide bond formation and S-cysteinylation in the activation of heparanase, we established several cell lines expressing mutant forms of heparanase, in which each Cys residue is substituted with a Ser residue. As shown in Fig. 3A, HP-MH, HP-MH/C127S, HP-MH/C179S, and HP-MH/C211S assumed both a 65-kDa pro form and a 50-kDa active form; however, HP-MH/C437S or HP-MH/C542S did not exist as a 50-kDa active form. We then observed the resulting heparanase activity, based on the digestion of heparan sulfate, and assessed the activity of heparanase by fragment separation in SDS-polyacrylamide gels. As shown in Fig. 3B, concomitant with the detection of the active-form heparanase, the samples incubated with extracts from HT1080-HP-MH, HT1080-HP-MH/C127S, HT1080-HP-MH/C179S, and HT1080-HP-MH/C211S cells showed broader smears and higher levels of partially digested heparan sulfate migrating toward the bottom of the gels, compared with the results obtained with the HT1080-Neo cell sample. However, if some of the cysteine residues had a weak effect on enzymatic activity, our assay system would not detect it. On the other hand, neither the extracts from HT1080-HP-MH/C437S cells nor those from HT1080-HP-MH/C542S cells exhibited heparanase activity (Fig. 3B); these findings suggest that disulfide bond formation between Cys437 and Cys542 is required for the activation of heparanase. We subsequently examined the effects of a reducing reagent on heparanase activity in vitro. Treatment with DTT did not inhibit the activity of heparanase even in the presence of iodoacetamide in vitro (Fig. 3C). Therefore, it is likely that once heparanase has formed the disulfide bond and is activated, the disulfide bond formation is no longer required for the maintenance of enzymatic activity.
Effect of disulfide bond formation on the intracellular trafficking of heparanase. Because heparanase was secreted into the cultured medium via the endoplasmic reticulum-to-Golgi apparatus (12), we next examined the intracellular localization of point-mutated heparanase. Wild-type heparanase stained with anti-Myc antibody was detected in the perinuclear organelles and was also diffused in an intracellular manner (Fig. 4A). Use of an endoplasmic reticulum–specific (anti-Bip/GRP78) antibody (data not shown) and a Golgi apparatus–specific (anti-GRASP65) antibody (Fig. 4A) showed that heparanase tended to localize in the perinuclear organelles, whereas the heparanase that diffused within the cells was in the Golgi apparatus and endoplasmic reticulum, respectively. HP-MH/C127S, HP-MH/C179S, and HP-MH/C211S were also found to localize in the endoplasmic reticulum and Golgi apparatus (Fig. 4A). In contrast, the amount of HP-MH/C437S and HP-MH/C542S in the Golgi apparatus was substantially lower than that of wild-type heparanase (Fig. 4A). Subsequently, to examine the effects of disulfide bond formation on the secretion of heparanase, we compared the amount of heparanase protein in the cell lysates with that in the conditioned medium using wild-type and mutant heparanase-expressing cells. As shown in Fig. 4B, in the case of HP-MH– and HP-MH/C211S–expressing cells, the same levels of heparanase were observed in the culture medium. However, the levels of secretion of HP-MH/C127S and HP-MH/C179S were slightly low, and the secretion of both HP-MH/C437S and HP-MH/C542S was completely blocked. Therefore, it was revealed that disulfide bond formation between Cys437 and Cys542 is required for the intracellular trafficking and secretion of heparanase and that the disulfide bond formation between Cys127 and Cys179 also regulates the amount of heparanase in culture medium.
Cancer cells must be able to degrade the extracellular matrix to become invasive and metastatic. Many proteases (e.g., matrix metalloproteinase) and glycosidase (e.g., heparanase) are essential components in the degradation of the extracellular matrix. Because heparanase is overexpressed in many tumor types, the mechanisms regulating the activation of heparanase should be clarified. We recently reported that the secretion of heparanase is regulated by glycosylation (12); however, other postmodifications that may be required for the activation of heparanase have not yet been clarified.
In this study, we showed that heparanase contains two disulfide bonds, Cys127-Cys179 and Cys437-Cys542, and one S-cysteinylation at the Cys211 residue (Fig. 5A). The disulfide bond formation between Cys437 and Cys542 is necessary for the activation of heparanase; however, treatment with a reducing reagent was not found to exert any effect on heparanase activity in vitro (Fig. 3). Thus, once heparanase has formed the disulfide bond between Cys437 and Cys542, it is secreted and activated; after that, the disulfide bond is no longer necessary for the enzyme activity. Thus, it has been suggested that once heparanase has been activated under nonreducing conditions, it can remain active under both reducing and nonreducing conditions. Furthermore, two mutants of heparanase, HP-MH/C437S and HP-MH/C542S, were found to be unable to localize in the Golgi apparatus, and therefore secretion into cultured medium was suppressed (Fig. 4); these results suggest that the disulfide bond between Cys437 and Cys542 is also related to intracellular trafficking, followed by secretion. Thus, disruption of disulfide bond formation affects the structure of heparanase in a manner that determines its movement through the secretory pathway, which may, in turn, be a crucial factor in the ultimate secretion of the enzyme. Because cell surface expression and the secretion of heparanase have both been shown to markedly promote tumor angiogenesis and metastasis (43, 44), the disulfide bond formation at Cys437-Cys542 may represent a critical determinant of tumor invasiveness.
Heparanase and heparanase-like genes have been cloned from a variety of organisms. Alignment of the human amino acid sequence with those of the rat, mouse, cow, dog, chicken, zebrafish, thale-cress, rice plant, and silkworm has shown that the Cys437 and Cys542 residues of human heparanase are well conserved among all forms; however, other Cys residues (Cys127, Cys179, and Cys211) are not conserved in plants and worms (Fig. 5B). Therefore, the disulfide bond between residues Cys437 and Cys542 in heparanase may be a common and pivotal linkage that determines the secretion and activation of this enzyme in various organisms.
However, it should be noted in this context that the formation of a disulfide bond between Cys127 and Cys179 regulated the amount of heparanase in culture medium, but was not involved in the activation of heparanase (Fig. 4). Heparanase is processed, thus yielding an 8-kDa polypeptide (Gln36-Glu109) at the NH2 terminus and a 50-kDa polypeptide (Lys158-Ile543) at the COOH terminus, which heterodimerize to form the active heparanase enzyme, with an intervening 6-kDa peptide (Ser110-Gln157), which is subsequently excised by proteolysis (13–15). Cys127 is the only disulfide linkage site located within the intervening 6-kDa peptide. It has been reported that overexpression of the 8- and 50-kDa subunits individually had no effect on heparanase activity, and that cotransfection with both the 8- and 50-kDa protein subunits resulted in a pronounced increase in heparanase activity (14, 15). Thus, it is possible that the 6-kDa peptide is not required for the activation of heparanase. In support of this notion was the finding that a defect in the formation of the disulfide bond between Cys127 and Cys179 failed to inhibit the activation of heparanase. Moreover, we constructed a Myc tag within the intervening 6-kDa peptide of heparanase (designated as an HP-6M)–expressing vector, and established stable clones that overexpressed HP-6M (Supplementary Fig. S1). Using these cells, however, we could not detect any interaction between the intervening 6-kDa peptide and the 50-kDa active form in the cells (Supplementary Fig. S1). Thus, it is suggested that the disulfide bond formation between Cys127 and Cys179 plays a role in the amount of heparanase in culture medium; once heparanase is secreted and activated, it may be broken.
Moreover, S-cysteinylation occurs at the Cys211 residue (Fig. 2D), although at present, we have not found the role of this modification for heparanase function (e.g., intracellular trafficking, secretion, or activation). Despite several reports on S-cysteinylated proteins, such as plasma albumin (45), amyloidogenic κ1 light-chain protein (36), and protein kinase C (46), few functions have been characterized for protein S-cysteinylation. The identification of additional S-cysteinylated proteins will enhance our general understanding of the important roles played by S-cysteinylation.
In summary, we show here that disulfide bond formation between residues Cys437 and Cys542 is an important step in the activation of heparanase. The antimetastatic effects of heparanase inhibitors have been reported in vitro and in vivo (8, 40, 42, 47, 48). In addition, our findings suggest that inhibitors of the formation of the disulfide bond may be developed into effective therapeutic agents for the treatment of heparanase-overexpressing cancers.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Grant support: A grant-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT), the Chemical Biology Project (RIKEN), the CREST Research Project (Japan Science and Technology Agency; T. Suzuki and N. Dohmae), and a MEXT scholarship (N.S. Lai).
We thank Y. Asami, N. Kato, T. Teruya, and K. Sato for their valuable suggestions; K. Ito for technical assistance with the MALDI-TOF mass spectrometry analysis; and Y. Ichikawa and R. Nakazawa (Bioarchitect Research Group, RIKEN) for the DNA sequencing.