A. Recombinant protein trichosanthin expression and purification

The cDNA of full-length trichosanthin (TCS) from T. Kirilowii was synthesized by GeneWiz Co., Ltd (Suzhou, China). The DNA related to the protein sequence of mature TCS (residues D24-A270) was sub-cloned into the expression vector pET28a to form a fusion protein with His₆ tag and TEV (Tobacco Etch Virus) protease cutting site at its N termini. The constructed plasmid was transformed into expression host strain E. coli BL21 (DE3). Cell harboring this expression plasmid was grown in LB broth containing 34 μg/mL kanamycin to mid-log phase at 37 ℃. After cooling, the protein was induced by 0.1 mM IPTG at 18 ℃ for about 18 hours. The cells were harvested and re-suspended in buffer A [100 mM Tris-HCl (pH 7.5, 4 ℃), 10% (v/v) glycerol, 100 mM NaCl, 20 mM imidazole] and then were lysed by sonication. The lysate was centrifuged, and the supernatant was loaded on a 10 mL Ni²⁺-chelating Sepharose column (GE Healthcare). The column was washed with 200 mL of buffer A and eluted with buffer A with 200 mM imidazole. The eluted fraction was buffer exchanged to remove imidazole with desalting column (GE Healthcare). Self-made recombinant TEV protease was added in the pooled fraction at 20 ℃ for 12 hours to remove the His₆ tag. The reactant passed through the Ni²⁺-chelating Sepharose column again to separate the TCS protein from the His6-tagged proteins. The flow-through was applied to a Superdex 75 prep grade gel filtration column (GE Healthcare) to purify the TCS protein to >95% pure as judged by SDS/PAGE. The final protein in concentration of ∼10 mg/ml was subjected to crystallization.

B. Crystallization and preparation of xenon derivatives

Trichosanthin (TCS) crystallized in reservoir solution with 1.2 M KCl, 100 mM sodium citrate, pH 5.5 at 291 K using hanging-drop vapor-diffusion method. Xenon derivative TCS crystals were prepared at 18 ℃ by pressurizing TCS crystals using Xenon chamber (Hampton Research). For Xe gas pressurization, a crystal was mounted in a nylon loop from the mother liquor, dipped through cryoprotectant (20% glycerol and 80% reservoir solution), and placed in the pressure chamber. Following incubation (5–10 min), the chamber was depressurized in 10 seconds, and the crystal was flash-frozen in liquid nitrogen in 5 seconds. Pressures tests were conducted at 3, 2, and 1 MPa, in combination with two sets of incubation times (5 and 10 minutes), and 32 crystals were screened for diffraction. Nine complete datasets collected were used for analyses of Xe occupancy, and one was used to solve the phase of TCS by SAD method.

C. Data collection and processing

All diffraction datasets were collected on beamline BL17U1 of Shanghai Synchrotron Radiation Facility (SSRF). The wavelength was 1.54 Å, but one dataset of native TCS was collected at 0.98 Å as control. The datasets were indexed, integrated, and scaled with the HKL2000 package [20]. The data collection statistics for experimental conditions of 3 MPa pressure and 10 min incubation time are given in Table 1.

For SAD phasing solution of the TCS xenon derivative, the anomalous substructure of xenon atom were determined by SHELX C/D [21]. Xe-SAD phasing was calculated by OASIS [22] from the CCP4 suite [23] followed by density improvement in DM [24]. Initial models were built using the build-in Solve/Resolve [25], followed by model extension/rebuilding in ARP/wARP [26]. The model was refined using SAD refinement with optimization of the xenon atom occupancy in REFMAC [27]. For the native TCS dataset, a molecular replacement solution for one TSC molecule in the asymmetric unit was obtained with the reported TCS structure (pdb code: 1TCS [18]) as the search model using Phaser [28]. All final models were produced after several iterative rounds of manual rebuilding in Coot [29] and refinement in Phenix [30].

A. The preparation of xenon gas derivatives

Trichosanthin is bio-synthesized as a preproprotein consisting of 289 amino acids [14]. It is further processed to a 247-residue single-chain polypeptide mature form TCS, via removing an N-terminal 23-residue signal peptide and a C-terminal 19-residue propeptide. Not extracted from the plant T. kirilowii, the recombinant TCS protein used in this study was heterogeneously expressed and purified from bacteria E. Coli with the same amino acid sequence as the mature TCS. As listed in Table 1, high resolution native crystal diffraction data, ∼1.0 Å, were collected from crystals of this batch of protein at SSRF.

High quality TCS crystals were used to prepare Xe derivative crystals using pressurized gas chamber. To avoid the crystals drying out during the gas incubation, a stripe of filter paper soaking with reservoir solution were placed around the chamber wall. The pressurized crystals should be flash-frozen in liquid nitrogen after released from xenon chamber, because the process of xenon binding is completely reversible. Xenon atoms were trapped at the binding sites under cryogenic temperature, so the data were acquired in the same way as with native crystals.

B. Binding site of xenon

A fluorescence excitation scan was performed on-site to verify the presence of xenon anomalous signals in the sample. Because the residual anomalous signal Cu Ka (f’ = -0.783, f" = 7.348) is still large enough to provide useful phase information, as a compromise between resolution limit and anomalous signal intensity, we selected the wavelength of 1.54 Å to collect the data for SAD phasing, while the data collection from native TCS crystal were done at wavelength of 0.98 Å for molecular replacement phasing.

The xenon was shown to bind to the pre-existing atomic-sized cavities in the interior of globular protein molecules [4]. The interaction of xenon atoms with proteins is the result of non-covalent, weak-energy van der Waals forces. Therefore, the process of xenon binding is completely reversible. As shown in Fig. 1(a), the Cα chain of native and Xe-derived TCS structures overlap exactly with each other (RMSD = 0.183 Å). The red sphere represents the location of xenon atom trapped inside of TCS. The residues forming xenon-binding cavity is enlarged in Fig. 1(b). A difference omit electron density map (Fig. 2) was calculated with |F_{obs, Xe} - F_{calc, Xe}| using phases from the native structure. The map shows a strong peak (>16 sigma), corresponding to the xenon site.

Fig. 1.Full-screen View

(Color online) The superimposed structure models of trichonsanthin and Xe-derived trichosanthin (a). The native TCS structure is shown as yellow cartoon; the Xe-derived TCS structure is shown as green cartoon, and the Xenon atom is labeled as red ball. The RMSD of Cα chain of two structures is equal to 0.183 Å. The residues forming xenon-binding cavity is enlarged in Fig. 0.

Fig. 2.Full-screen View

(Color online) Difference omit map of Xenon atom site from Xe-derived trichosanthin. The green mesh indicates the F_obs–F_calc electron density of Xenon atom cut at 5σ; the peptide model is shown as gray cartoon; the Xenon atom is shown as red sphere; the residues forming the hydrophobic cavity for Xe-binding are shown as gray stick; the residues involving the substrate binding and catalyzing are shown as yellow stick. The dark surface marked as "Sub" indicates the substrate-binding pocket; the dark surface marked as "Xe" indicates the Xe-binding cavity.

The residues around the Xe atom are Ile 71, Met 72, Gly73, Tyr74, Tyr81, Phe82, Phe83, Leu105, and Leu151. Most of the residues possess hydrophobic side-chains. The distances between Xe atom and the atoms from the residues sit in the range of 3.2 Å to 5.0 Å (Table 2). It was reported that two putative hydrophobic sites existed inside TCS [17]. Different from them, the xenon-binding hydrophobic cavity here is relatively enclosed, with a volume in the same scale as xenon atom. Compared to the native TCS structure, no significant changes in the backbone or side chain conformation were observed upon binding Xe atoms, suggesting that there may be at least one pre-existing, inflexible hydrophobic cavity inside TCS for xenon binding. Because of its relatively small size, the cavity can accommodate only one single atom in a well-ordered position as the sphere electron density of xenon, as shown in Fig. 2.

C. The hypothetical xenon diffusion pathway inside TCS

Based on the TCS-NADHP complex structures [18], we can find that the substrate binding pocket of TCS (the dark surface marked as "sub" in Fig. 2) is adjacent to the Xe-binding cavity. Except the residues (Tyr70, Glu85, Glu160 and Arg163) involving in the substrate hydrolysis, two residues (Ile71 and Phe83) act as a set of "swing-door" to connect the protein surface to the hydrophobic core. Also, the smallest distance between the two residues is 4.0 Å, which is just enough to let the xenon atom to penetrate it, considering the van der Waals radius of Xe is 2.16 Å [4]. We propose that under the moderate pressure, xenon gas diffuses along the accessing channel to the bottom of the substrate-binding site, and squeezes into the hydrophobic cavity of Xe-binding site through the gate formed by residue Ile71 and Phe83.

The results of studies on the xenon complex with heme protein [8] showed the number and occupancies of xenon binding sites vary with the pressure applied. To estimate the possible xenon sites and the occupancies, three pressures ( 3 MPa, 2 MPa and 1 MPa) and two incubation duration ( 5 min and 10 min) were used with TCS crystals. However, from the datasets we collected (Table 3), only one xenon atom is located in Xe derivative structures, and the occupancy of Xe in the binding cavity does not gain obviously by increasing the pressure and incubation time. Perhaps further increase of the Xe gas pressure and incubation time shall be needed. Nevertheless, it might indicate from the other side that this xenon site is relatively conserved in TCS.