Introduction

Chinese scorpion Buthus martensii Karsch (BmK) venom is rich source of neurotoxins which bind to various ion channels with high affinity and specificity and are thus widely used as compounds to modulate signal transduction and channel gating (Chen et al. 2005). The first chlorotoxin (Cltx)-like peptide, BmK CT, isolated from the venom gland of B. martensii Karsch (Zeng et al. 2000), is a 36 mer peptide which is cross-linked by four disulfide bridges. Its sequence identity is 68% similar to Cltx isolated the scorpion Leiurus quinquestriatus (DeBin et al. 1993). Cltx binds specifically to glioma cell surface as a specific chloride channel blocker (Olsen et al. 2003). Preclinical studies suggest that 131I-TM-601 (synthetic Cltx) might be an effective targeted therapy for the treatment of glioma (Mamelak et al. 2006). Our recent data indicate that BmK CT also shows specific toxicity against glioma cells but not astrocytes (Fu et al. 2007).

Gliomas can diffuse into the normal brain and this invasion of glioma cells involves modification of receptor-mediated adhesive properties of tumor cells, degradation, and remodeling of extracellular matrix by tumor-secreted metalloproteinases (MMPs) such as MMP-2, consequently creating an intercellular space for invasion of glioma cells. MMP-2 activity is essential for invasion of glioma cells and Cltx inhibits invasion of glioma cells by modulating the surface expression of enzymatically active MMP-2 (Deshane et al. 2003).

Electrostatic effects play a pivotal role in the interaction between the ion channels and the small basic peptides of the scorpion toxins (Wu et al. 2004). In the present study, influence of basic residues in BmK CT on the inhibitory activity to MMP-2 was studied using computational methods and we now propose a model of the structural mechanism that may explain the inhibitory effect of this Cltx-like neurotoxin on glioma migration.

Materials and methods

Materials

DMEM was purchased from Gibco Life Technologies (NY, USA). Fetal calf serum (FCS) was purchased from the Institute of Hematology (Hang Zhou, China). Site-directed mutagenesis kit (TaKaRa MutanBEST Kit) was purchased from TaKaRa Biotechnology Co., Ltd (Dalian, China). Ni2+-NTA agarose was obtained from Qiagen (Germany). Gelatin was purchased from Sigma. BmK CT encoding the Cltx-like peptide from the scorpion B. martensii Karsch was cloned according to the sequence optimized for codon usage in Escherichia coli and the expression plasmid pRSET C-BmK CT was constructed in our laboratory.

Prediction of three dimensional structure of protein by Swiss-Model workspace (http://swissmodel.expasy.org/) was performed as reported previously (Peitsch 1995; Guex and Peitsch 1997; Schwede et al. 2003; Kopp and Schwede 2004; Arnold et al. 2006). Protein–protein interactions by docking methods using ZDOCK (a fourier transform-based protein docking program) (http://zdock.bu.edu/) is preferred when high-resolution three dimensional structure of protein complex is not available (Wu et al. 2004). The quality of generated binding model is then further optimized by energy minimization and molecular dynamics (MD) simulation using Insight II 2000 software package developed by Accelrys.

Site-directed mutagenesis, expression, purification, and western blot analysis of BmK CT

Because BmK CT and Cltx are similar in primary structure, it is predicted that BmK CT may be functionally related to Cltx (Zeng et al. 2000). To analyze the location and function relationship of strong basic amino acid residues, K and R were mutated to a hydrophobic amino acid A (Table 1). Four different mutants were produced by PCR method using pRSET C-BmK CT as template.

Table 1 Amino acid sequences of BmK CT and mutants
Full size table

Recombinant E. coli BL21 (DE3) cells were grown in LB medium supplemented with ampicillin (60 μg/ml) and incubated with shaking at 37°C. When the OD600 reached 0.6–0.7, the culture was induced with IPTG at 0.5 mM and shaken at 16°C for 8 h. Cells were harvested at 6,000×g for 30 min at 4°C, and then ultrasonicated. Cell debris was removed at 15,000×g for 30 min at 4°C. The supernatant was loaded onto a Ni2+-NTA affinity column pre-equilibrated with buffer I (20 mM Tris/HCl buffer containing 10 mM imidazole and 500 mM NaCl, pH 9.0). The recombinant fusion protein of the bound His6-tagged at the N-terminal was eluted with buffer II (20 mM Tris/HCl buffer containing 300 mM imidazole and 500 mM NaCl, pH 9.0). The eluent fraction containing BmK CT was collected, concentrated and desalted with 20 mM phosphate buffer (pH 7.4) at 4°C.

Not all of the mutants could not be attached to the Ni2+-NTA affinity column during the purification putatively due to the change of protein structure. Therefore, denaturation and refolding of mutants were performed. Briefly, IPTG-induced cells were harvested, followed by ultrasonication, and urea was added slowly to the total proteins to give 6 M. Debris was removed by centrifugation at 15,000×g for 25 min at 4°C and supernatant was loaded onto the Ni2+-NTA affinity column and thoroughly washed with buffer I. His6-tagged fusion mutant was eluted with buffer II. During the refolding process, urea was diluted gradually to 4, 2, 1, and 0.5 M. The resulting mutants were concentrated and desalted, respectively.

For western blot analysis, membrane was probed with the anti-His6 tag mouse monoclonal antibody as the first antibody, followed by peroxidase-conjugated goat anti-mouse IgG (H+L) was used as the second antibody. Bound antibody was detected using an ECL Enhanced Chemiluminescence Detection Kit.

Circular dichroism (CD) analysis

The CD spectra of BmK CT and mutants were collected using Bio-Logic MOS-450 spectropolarimeter (France), with protein concentration of 0.4 mg/ml suspended in PBS buffer. Spectra were measured over the wavelength range from 200 to 240 nm at room temperature. CD spectra data obtained were average of three scans.

In vitro wound healing assay

The effect of BmK CT and mutants on cell migration was examined using in vitro wound healing assay. C6 cells were grown to a confluent monolayer in 96 multiwell microtiter plates. The “scratch” was introduced by scraping the C6 cells monolayer with a p200 pipette tip and was then washed three times with cell culture medium to remove cell debris. BmK CT was initially diluted in PBS and then serially diluted (0.15, 0.30, 0.6, 1.2, and 2.4 μM) in cell culture medium. BmK CT and mutants were diluted in cell culture medium to 0.15 μM which was identified as an appropriate concentration for detection of cell migration inhibition. Percent inhibition rate (IR%) was calculated according to the following equation: IR% = (1 − T/C) × 100%, where T and C represent the mean migration distance of the treated group and the control group, respectively.

Gelatin zymography assay

Gelatinolytic activity of MMP-2 was analyzed by gelatin zymography. The cell culture media were changed to serum-free cell culture media when C6 cells were grown to a confluent monolayer. Then, BmK CT and mutants were diluted in these serum-free cell culture media to 0.15 μM. Serum-free media conditioned for 48 h were subjected to SDS-PAGE using 10% (v/v) gels containing 0.1% gelatin. The gel was incubated twice (for 30 min at room temperature) in 2.5% Triton X-100. The gel was incubated in activation buffer (5 mM CaCl2, 1 μM ZnCl2, and 0.005% Brij, 50 mM Tris/HCl, pH 8.0) overnight at 37°C. Gel was then stained with Coomassie Brilliant Blue R-250 and briefly destained in 10% (v/v) acetic acid and 40% (v.v) methanol (Deshane et al. 2003; Lu et al. 2004). Gelatinolytic activity of MMP-2 was detected as a transparent band on the blue background.

Three dimensional structure simulation

To analyze the location and function relationship of basic amino acid residues, computational methods were used to simulate BmK CT (wild type and mutants) and catalytic domain of MMP-2 complexes. The atomic coordinates of Cltx (Protein Data Bank code: 1CHL), catalytic domain of MMP-2 or gelatinase A (R88-P250, PDB code: 1QIB) were obtained from the Protein Data Bank (Berman et al. 2000).

The structure of Cltx in solution has been determined using 2D 1H NMR spectroscopy as reported previously (Lippens et al. 1995). The homologous spatial structure of BmK CT was modeled by using the solution structures of Cltx and insect toxin 15A (PDB code: 1SIS) from Buthus erpeus as templates through the Swiss-Model server. The atomic coordinates simulation of BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, and BmK CTR35A was performed by Swiss-Model server.

Analysis of BmK CT–MMP-2 catalytic domain complex by ZDOCK and MD simulation

Prediction of BmK CT–MMP-2 catalytic domain complex was performed using ZDOCK which can operate on a Java platform (Pierce and Weng 2005; Wiehe et al. 2005; Pierce et al. 2007). The modeled three dimensional structures were subjected to refinement by optimization (steepest descent), 300 steps of conjugate gradient minimization within 10 Å water layer model at every step. MD simulation was performed to examine the quality of the model structure via 150 ps with the step size of 1 fs at 298 K. All simulations mentioned above were accomplished without any constraint under the consistent-valence force field (CVFF) in Discover 3 module of Insight II 2000 software package.

Results and discussion

Expression, purification and western blot analysis of BmK CT and mutants

The genes encoding BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, and BmK CTR35A were amplified by PCR using pRSET C-BmK CT as template.

BmK CT and mutants (~4 kDa) were expressed in E. coli BL21 (DE3) under IPTG induction where both showed high level of expression (Fig. 1a), amounting to approximately 20% of the total cellular proteins. All the BmK CT and mutants were expressed in the soluble form present in the supernatant. The BmK CT but not the mutant proteins could effectively bind to Ni2+-NTA affinity resin. Consequently, the denaturation and refolding treatments of mutants were performed and BmK CT was also treated in the same process as control. The purified mutant proteins showed two bands on the 16% Tricine/SDS/polyacrylamide gel (Fig. 1b). Furthermore, western blot analysis showed that protein with molecular mass of 15 kDa could be the tetramer of the recombinant mutants (Fig. 1c). The CD spectra of BmK CT and mutants showed that they belonged to α+β type protein according to the ellipticity values of 208 and 222 nm (Fig. 1d).

Fig. 1
figure 1

a Expression of the His6-taged BmK CT and mutants in E. coli BL21 (DE3) M low molecular weight markers proteins; Lane 1 The total expressed protein of BL21 (DE3)/pRSET C; Lanes 2–6 The total expressed proteins of BL21 (DE3)/pRSET C–BmK CT, BL21 (DE3)/pRSET C–BmK CTR14K15AA, BL21 (DE3)/pRSET C–BmK CTR17A, BL21 (DE3)/pRSET C–BmK CTK25A, and BL21 (DE3)/pRSET C–BmK CTR35A, respectively. The arrow shows the expression product His6-taged BmK CT, BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, and BmK CTR35A (~4 kDa). b Purification of BmK CT and mutants Lanes 1, 2 The purified target protein BmK CT washed with 300 mM imidazole without denaturation; Lanes 3–7 The purified target proteins BmK CT, BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, and BmK CTR35A with the treatments of denaturation, affinity chromatography, and refolding. The arrows indicated the monomer and tetramer of these target proteins, respectively. c Western blot analysis Lanes 1–5 Western blot analysis of target proteins BmK CT, BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, and BmK CTR35A, respectively. The tetramer of the mutants (15 kDa) is marked by arrow (upper position). d CD spectra

Full size image

In vitro wound healing assay

As shown in Fig. 2a, incubation of C6 cells with BmK CT inhibited cell migration in a dose-dependent manner. Inhibition of C6 cell migration by BmK CT increased significantly from 0.15 to 2.4 μM. However, microscopic examination revealed that the inhibition of C6 cell migration in glioma spheroids by BmK CT (0.6 and 1.2 μM BmK CT) was associated with retraction of the long protrusions at the front of the invading C6 cells (data not shown). When BmK CT was applied into the C6 cell medium at 2.4 μM (final concentration) and incubated for 8 h, BmK CT inhibited the proliferation of C6 cells significantly. Consequently, 0.15 μM (for both BmK CT and mutants) was selected as appropriate to compare the inhibition rates of C6 cell migration. As shown in Fig. 2b, the average inhibition rates of BmK CT, BmK CT (denaturation and refolding), BmK CTK25A and BmK CTR35A were ranging from 17 to 26%, which was two-fold higher than that of BmK CTR14K15AA and BmK CTR17A.

Fig. 2
figure 2

BmK CT and mutants decreased C6 cell migration rate. a Migration of C6 cells was routinely monitored after confluent monolayers of C6 cells were gently scratched with a plastic pipette tip in the in vitro wound healing assay. The values of inhibition rate were regressed by SigmaPlot software. BmK CT inhibited the migration of glioma cells in a dose-dependent manner at the concentration ranging from 0.15 to 2.4 μM. b Inhibition rates of C6 cell migration were detected 8 h after treatment with BmK CT and mutants (0.15 μM). BmK CT* denotes the BmK CT protein was treated by denaturation and refolding processes. c Gelatin zymography assay of MMP-2. BmK CT and mutants inhibited the catalytic activity of MMP-2. Degrees of MMP-2 activity were marked from “+ (minor)” to “++++ (significant)”

Full size image

Gelatin zymography assay

BmK CT and mutants showed inhibitory effect on MMP-2 activity as compared with the control (Fig. 2c). Gelatin zymography assay suggested that BmK CT and mutants could inhibit migration of glioma cells via MMP-2. The inhibitory capability of BmK CTR14K15AA and BmK CTR17A was weaker than that of BmK CT, BmK CTK25A, and BmK CTR35A.

Electrostatic potential and structure–function relationship assay

To investigate the possible role of electrostatics for the structure–function relationships, the electrostatic surfaces for both BmK CT (wild type and mutants) and catalytic domain of MMP-2 were calculated by using DeepView V.3.7 (Fig. 3).

Fig. 3
figure 3

Three dimensional structures and electrostatic surfaces of MMP-2 catalytic domain (a), BmK CT (b) and mutants (c–f) Electrostatic surfaces are calculated by the program DeepView V.3.7. Different color codes are −1.8 kT/e (Red) and 1.8 kT/e (Blue) where k is the Blotzmann constant, T is the temperature in Kelvin, and e is the charge of the electron. Positively charged residues are shown in blue and negatively charged in red. The basic amino acid residues of BmK CT and site-directed mutagenesis are shown in yellow

Full size image

The MMP-2 catalytic domain contains three α-helices and five β-sheets are arranged in a typical matrixin fold, and two zinc ions (PDB code: 1QIB) (Falconi et al. 2003). Electrostatic potential surface analysis showed that negative charge was mainly distributed over the MMP-2 catalytic domain (Fig. 3a).

Structural model showed that the folded structure of BmK CT consisted mainly of a small two-stranded antiparallel β-sheet packed against an α-helix, which is in agreement with those observed in the known three dimensional structures of other short toxins (Fig. 3b).

The calculated electrostatic potential surface around the BmK CT molecule was found to be mainly positive, suggesting that this clustering of positively charged residues and the strong positive field might have important implications in the interaction with the receptor. The obvious change of electrostatic potential was accompanied by the mutation of R14 K15 to A14A15 which was located in the beginning of α-helix, where it formed a neutral to negative zone (Fig. 3c). Obviously, as presented in the model of BmK CTR14K15AA, the strong negatively charged zone rendered the positive surface in (or near) the α-helix disappeared completely, whereas the positive surface near the β-sheet was maintained by the positive residues. In Fig. 3d–f, strong polar charge was not formed by the mutation of one of the positive residues. These results showed clearly that strong electrostatic interaction existed between the interfaces in the BmK CT, mutants and MMP-2 catalytic domain. Also, it revealed that R14 and K15 of BmK CT played an important role of this interaction, which could affect the docking position or degree affinity.

Structural analysis of binding models between BmK CT (wild and mutant types) and MMP-2 catalytic domain complexes

The total energies of top ten ranked predictions are listed in Table 2. The lowest energy configurations were chosen for energy minimization and 150 ps MD simulation (Fig. 4). The electrostatic energies and interactive residues by hydrogen bond are shown in Table 3.

Table 2 Total energies of candidates of BmK CT and mutants–MMP-2 catalytic domain complexes
Full size table
Fig. 4
figure 4

Ribbon models of BmK CT–MMP-2 catalytic domain complex (a) and mutants–MMP-2 catalytic domain complexes (b–e) The whole backbones of complexes were shown. The α-helix (H) and β-sheets (S1, S2) in the BmK CT are explicitly labeled. The amino acid residues of contacts by hydrogen bond (indicated by green line) between BmK CT (or mutants) and MMP-2 catalytic domain were indicated

Full size image
Table 3 Electrostatic energy and interactive residues by hydrogen bond of each minimum energy model
Full size table

Two significant findings were presented in this work. Firstly, in structural aspect, based on the calculated electrostatic potential surface analysis, α-helices section of MMP-2 catalytic domain was the interactive interface bearing a large negative electrostatic potential for wild type and mutant BmK CT (Fig. 4). In Fig. 4a and Table 3, as for BmK CT, α-helix (R17, C19) was the only interface in the complex. As shown in Fig. 4b–e, the interfaces of BmK CTR14K15AA, BmK CTR17A, BmK CTK25A, BmK CTR35A were β-sheet 2 (Q30), C-terminal of α-helix (G21), α-helix (C19) and α-helix (R14), respectively, which suggested that α-helix was important for the location of BmK CT in the complexes. Once the mutation occurred in α-helix as demonstrated in BmK CTR14K15AA and BmK CTR17A, the β-sheet 2 and C-terminal of α-helix were necessary in interacting with MMP-2 catalytic domain. So, three important basic residues namely R14,17 and K15 were the active residues of BmK CT.

Secondly, with respect to the binding energy, electrostatic energy and the number of H2 bonds are important for the tightness of the complex and can affect the functionality of BmK CT. The average values of total energies of BmK CTR14K15AA–MMP-2 catalytic domain and BmK CTR17A–MMP-2 catalytic domain complexes were 1531 and 1516 kcal/mol, respectively. These were higher than those of BmK CT–MMP-2 catalytic domain, BmK CTK25A–MMP-2 catalytic domain and BmK CTR35A–MMP-2 catalytic domain complexes. This result demonstrated that the structures of BmK CT–MMP-2 catalytic domain, BmK CTK25A–MMP-2 catalytic domain and BmK CTR35A–MMP-2 catalytic domain complexes were more stable than BmK CTR14K15AA–MMP-2 catalytic domain and BmK CTR17A–MMP-2 catalytic domain complexes. Moreover, according to the electrostatic energies and the number of interactive residues by hydrogen bond (Table 3), it was concluded that the structure of BmK CT–MMP-2 catalytic domain, BmK CTK25A–MMP-2 catalytic domain and BmK CTR35A–MMP-2 catalytic domain complex was tighter than that of BmK CTR14K15AA–MMP-2 catalytic domain and BmK CTR17A–MMP-2 catalytic domain complex. These factors led to the decrease of functionality of BmK CTR14K15AA and BmK CTR17A.

Conclusion

In vitro experimental measurements and molecular dynamics simulation collectively showed that three important basic residues (R14,17 and K15) represented the active residues of BmK CT that interacted with the catalytic domain of MMP-2. The study on structure and function of BmK CT may allow us to develop more inhibitive anti-MMP2 reagents against glioma cells.