Rubelase was isolated from crude venom by HW-50 gel filtration and CM-cellulose cation-exchange column chromatography (see Figure 1). The molecular weight was determined by SDS-polyacrylamide gel electrophoresis and nano ESI-MS using a Micromass Q-TOF mass spectrometer and MassLynx data acquisition. The mass spectrometer was operated in the positive-ion mode. Purified sample was extensively dialyzed against water, lyophilized and then dissolved at a protein concentration of 10 μM in a solution containing equal amounts of acetonitrile and 0.2% formic acid. Two microliters of the sample solution was loaded into a nanoflow tip. A flow rate of about 50 nL/min into the analyzer was produced as a result of a potential of 1.2 kV applied to the nanoflow tip in the ion source. The cone voltage was set to 50 V. The spectra were averaged, smoothed, centroided and deconvoluted.

Elastase activity was assayed by the method of Hasegawa et al. [27] using STANA as the substrate. Proteolytic activity was measured by the method of Murata et al. [28] using casein as the substrate, and fibrinogenolytic activity was determined by the method of Ouyang and Teng [29]. Other enzyme assays were carried out as follows: arginine ester hydrolytic activity by the method of Roberts [30]; hide powder azure hydrolytic activity by the method of Rinderknecht [31]; azoalbumin hydrolytic activity by the method of Charney and Tomarelli [32]; azocollagen hydrolytic activity by the method of Moore [33]. One unit of STANA hydrolase activity was defined as the amount of enzyme that hydrolyzed 1 µM of substrate per min. The units of other enzyme activities were obtained by means of the following formula: units/mg = (absorbance change per min) × 2/mg of protein. Hemorrhagic activity was determined by the method of Bjarnason and Tu [6], and serum creatine phosphokinase activity was measured by the method based on the procedure of Tanzer and Gilvarg [34].

Rubelase was enzymatically digested with clostripain and lysyl endopeptidase. The digested fragments were also obtained by autoproteolysis, which occurs when rubelase is incubated with EGTA at a molar ratio of 1:10. The fragments were analyzed by the combination of Edman degradation method and MALDI/TOF mass spectrometry. Applied Biosystems 491 protein sequencer and Model 610A PTH analyzer were used for the sequence analysis in accordance with the manufacturer’s instructions. The molecular mass of digested fragments was also confirmed by using Voyager^TM (Applied Biosystems). Lyophilized fragments were dissolved at a concentration of 10 µM in 0.1% trifluoroacetic acid (TFA) and mixed with matrix (10 mg/mL of α-cyano-4 hydroxy cinnamic acid in 70% acetonitrile cont. 0.1% TFA). The sample mixture was then applied onto a sample plate and analyzed. MOE^TM (a molecular simulation and modeling software; purchased from Chemical Computing Group Inc.) was used for construction of protein models.

Frozen human endothelial cells (HUVEC and HPAEC) and aortic smooth muscle cells (HASMC) were cultured and maintained in the appropriate medium according to the method of the suppliers’ instructions. For bioassays, cells were seeded at a density of 1.5 × 10⁴ cells/well in 0.1 mL of medium in 96-multiwell plates. Samples were diluted in saline and then added to the cells. After 18 h, cell densities were determined by the colorimetric method using a cell counting kit-8 that was based on the tetrazolium salt/formazan system [35]. Cell staining was performed against the adherent cells on glass-plate dishes. Fluorescent probe CMFDA (0.5 µM in serum-free medium) was used according to the manufacturer’s manual, and the viable cells were observed under a fluorescence microscope.

Crude venom was fractionated using HW-50 column and elastase activity was found in fraction 6 (Figure 1A). The fraction was further purified using CM-cellulose column chromatography and a homologous preparation was found in fraction 2 (Figure 1B). The purified enzyme was named rubelase and the yield of rubelase from 69 mg crude venom was found to be approximately 5.3 mg.

When STANA, GAAPLNA and SAPANA were used as elastase substrate, rubelase preferentially hydrolyzed STANA (4.22 units/mg). Rubelase also possessed hydrolytic activities on hide powder azure (1.68 units/mg), azocollagen (1.18 units/mg) and azoalbumin (0.37 units/mg), and fibrinogenolytic activity (Aα and Bβ chains were hydrolyzed without forming fibrin clot). However, casein and arginine ester hydrolytic activities were not observed. Hemorrhagic activity was not detected when the purified enzyme (12 µg: approximately 40-times higher dose of minimum hemorrhagic dose of rubelysin) was injected s.c. into mice. Smooth muscle-damaging activity on mice, which was demonstrated as creatine phosphokinase activity in serum, was not detected, either.

Rubelase was thermolabile and the elastase activity was completely lost after incubation for 10 min at 60 °C. The chelating reagents such as EDTA and EGTA also affected elastase activity at final concentrations of 0.1–0.5 mM.

The isoelectric point was determined to be 9.2 by electrophoresis. The molecular mass of rubelase determined by SDS-PAGE was found to be 24,000 Da, while the ESI/Q-TOF mass method gave a molecular weight of 23,266 Da (Figure 2). The peak of 23,394 Da found in the macromolecule area indicates that there may be a macromolecule corresponding to rubelase with one residue, such as Lys and Glu, attached to its C-terminus. Since this point could be clarified by determining a complete gene sequence, analysis will be attempted if the venom glands of Crotalus r. ruber are obtained.

The enzymatically cleaved fragments of rubelase by lysyl endopeptidase and clostripain were subjected to Edman sequencing analysis and MALDI/TOF mass spectrometry. The fragments produced by autoproteolysis of rubelase were also analyzed. Although the N-terminal amino acid of rubelase could not be detected by a direct sequence analysis of the intact enzyme, analysis of mass spectra of fragments revealed that it was L-pyroglutamic acid. By combining all the data, the complete amino acid sequence of rubelase was determined (Figure 3); rubelase is composed of 202 amino acids. The molecular mass of rubelase based on the amino acid sequence (23267.73 Da) was almost identical to the molecular mass (23,266 Da) obtained by ESI/Q-TOF mass spectra.

The amino acid sequence of rubelase is extremely homologous to rubelysin, a hemorrhagic metalloproteinase previously isolated from Crotalus r. ruber venom and named HT-2 [8,13]. The amino acid residues of rubelysin (HT-2) at the position of 81 (Trp), 87 (Leu), 114 (Asn), and 151 (Glu) are substituted for Val, Lys, Lys, and Lys in rubelase, respectively. Rubelysin possesses caseinolytic activity and its minimum hemorrhagic dose (M.H.D) has been determined to be 0.27 µg, whereas rubelase is a non-hemorrhagic protein and does not hydrolyze casein. Since the putative zinc binding domain HEXXHXXGXXH in both enzymes is completely conserved, substitutions of four amino acid residues might cause the difference of substrate preference and biological activity.

Various low molecular weight metalloproteinases were purified from snake venoms [6,13,36,37,38]. The sequence alignment shown in Figure 3 indicates that rubelase is a member of family of snake venom metalloproteinases (SVMPs). Of these enzymes, adamalysin II (EC 3.4.24.46) isolated from the venom of Crotalus adamanteus and H₂-proteinase from the venom of Protobothrops flavoviridis are basic enzymes which have been shown not to exhibit significant hemorrhagic activity [36,38]. Since these properties are similar to rubelase, a significant difference in the caseinolytic activity exists between rubelase and these non-hemorrhagic metalloproteinases.

The effect of rubelase on cultured cells was investigated and compared with that of rubelysin. Figure 4 shows the changes in cell numbers after incubation with samples. Compared with control cells (white bars), viable HUVEC, HPAEC and HASMC numbers clearly decreased after addition of rubelysin, a hemorrhagic metalloproteinase (light blue bars). When 0.5 µg of rubelysin was added to HUVEC and HPAEC (Figure 4A,B), the number of viable cells decreased to similar levels as seen in the positive control (a cytotoxic control measured by using crude venom: green bars). On the other hand, rubelase possessed cytotoxicity only on HUVEC and HPAEC (Figure 4A,B: light yellow bars), and the effects were relatively weak as compared with rubelysin. The cytotoxic effects of rubelase and rubelysin on endothelial cells (HPAEC) were visualized as the cell density observed under a fluorescence microscope (Figure 5). The viability of HASMC was not affected by rubelase (Figure 4C). These results indicate that rubelase does not inflict an injury on smooth muscle cells which compose blood vessels and tissues. Therefore, rubelase may cause neither hemorrhage nor muscular damage. The muscle damaging activity measured as serum CPK levels also supports this speculation. However, although rubelase is non-hemorrhagic in nature, it clearly possesses cytotoxicity on endothelial cells (Figure 5C) when directly in contact with cells. Rubelase might exhibit its toxicity only when it enters directly into the bloodstream by envenomation.

For molecular modeling of rubelase and rubelysin, the information on adamalysin II obtained from the Protein Data Bank (PDB) was referred to, since this protein is the most homologous to these enzymes. Figure 6A shows the protein models of rubelase and rubelysin with the positions of substituted amino acid residues and zinc binding domain. Comparing with rubelysin, the environment surrounding the zinc binding domain (His-Glu-Xxx-Xxx-His-Xxx-Xxx-Gly-Xxx-Xxx-His) of rubelase seem to be basic because of the substitutions of Asn (114) to Lys and Glu (151) to Lys (Figure 6B). These changes may cause the distortion of the zinc binding site in the active pocket, resulting in the decrease in the substrate recognition ability of rubelase against large substrates (e.g., casein).