Conjugation design
Recognition of death receptors by death ligands including Fas ligand has been considered as important targets for therapy of many serious human diseases such as cancers (Ashkenazi 2008; Russo et al. 2010), and the specific interaction between the extracellular domains of them is a fundamental to the recognition. The extracellular domains of the proteins belonging to death ligands and death receptors, also called as tumor necrosis factor (TNF) ligand superfamily and TNF receptor superfamily, share each common structural features in three-dimensions (Bodmer et al. 2002). To date, several detailed three-dimensional structures of the extracellular domain complexes, including human TNF receptor–human TNFβ complex (Banner et al. 1993) and human death receptor 5–human TNF related apoptosis inducing ligand complex (Hymowitz et al. 1999), have been elucidated by mean of crystallography. A typical structure of the extracellular domain complexes is composed of three monomers of the receptor’s extracellular domains and one trimer of the ligand’s extracellular domains. Although no three-dimensional structure of the extracellular domains complex between human Fas ligand and human Fas receptor itself has been experimentally disclosed yet, a detailed three-dimensional structure of the extracellular domains complex between human Fas ligand and human decoy receptor 3 (hDcR3), which was revealed by X-ray crystallography, is already available (Liu et al. 2013). This X-ray structural model provided important information for the design of the conjugation site in this study. Although the primary sequence identity between hFasR and hDcR3 is moderate (17 %, Pitti et al. 1998), the three-dimensional interaction modes with hFasLECD, forming the binding interfaces of the ligand-receptor complexes, are considered essentially conserved to each other (Whalen and Hymowitz 2014).
Figure 1 presents the three-dimensional structure of the hFasLECD–hDcR3 complex (Liu et al. 2013). It was aimed to create a functional hFasLECD derivative containing Fluorescein residue in this study. Probably, the most popular strategy for introducing Fluorescein moieties into protein molecules, such as antibodies, is to modify the lysine residues present on the surface of the proteins by Fluorescein isothiocyanate (Thermo Fisher Scientific Inc. 2014). However, each NFG1CG4-hFasLECD monomer contains fourteen lysine residues including the two residues within the N-terminal FLAG tag sequence, and most of them locate on the molecular surface of the protein (Additional file 1). These lysine residues are often conserved among other animals’ Fas ligand extracellular domains (Motegi-Ishiyama et al. 2001). So far, two of them are suggested to be critical for the specific binding toward hFasRECD due to the existence of the residues at the binding interface. The side chain group of Lys 228 was predicted to form a salt-bridge with that of Asp 92 in hFasRECD from a molecular modeling study (Bajorath 1999). Lys 217 also locates in the contact site and is neighboring to Tyr 218, which was suggested to play an important role in the binding interactions from a mutagenesis study (Schneider et al. 1997). Therefore, a random modification of the lysine residues with fluorochrome moieties can significantly damage the receptor binding activity.
In contrast, only one unpaired cysteine residue in the N-terminal tag region and a single pair of disulfide-bridged cysteine residues within the folded region exist in NFG1CG4-hFasLECD. Judging from the position of the N-terminal residue (Leu143) in the three-dimensional structure of the hFasLECD–hDcR3 complex (Fig. 1) as well as from the fairly hydrophilic property of the N-terminal FLAG tag plus aa 139-142 region (Asp–Tyr–Lys–Asp–Asp–Asp–Asp–Lys–Gly–Cys–Gly–Gly–Gly–Gly–Glu–Lys–Lys–Glu), the unpaired Cys was expected to locate not proximal to the binding interface, but exposed to the solvent in an aqueous solution of neutral pH. Although the pre-activation by a treatment with some reducing agents was necessary for an efficient conjugation because of the propensity to form a disulfide-bridge between two NFG1CG4-hFasLECD monomer subunits due to non-specific oxidation of the unpaired cysteine residue, it was demonstrated that selective activation of the conjugation site alone was possible by choosing the mild reducing condition using TCEP for the reaction in the previous study (Muraki 2014b). Moreover, the resulting conjugates with several kinds of single or double maleimide group(s) containing compounds maintained the receptor binding activity, and also one conjugate was proved to exhibit significant cytotoxic activity against a cancer cell line after cross-linking by an anti-FLAG tag antibody (Muraki 2014b). Consequently, FL-5Mal was chosen as the modification reagent for the conjugation in this study.
Preparation and characterization of FL-5Mal conjugated NFG1CG4-hFasLECD
The conjugation procedures used in this study are summarized in Fig. 2a. In Fig. 2b, SDS-PAGE analysis of the reaction mixtures at each step (lanes 1–3) and the early six fractions in the first purification step using size-exclusion chromatography (lanes 4–9) are presented. The purified sample after a complete removal of low-molecular weight contaminants by the second size-exclusion chromatography still contained impurities shown as some minor bands less than 20.1 kDa and a sharp band between the size markers of 66.3 and 97.4 kDa arrowed in Fig. 2b. Judging from the band position, the latter component was considered to be the host-derived alcohol oxidase 1 (AOX-1) (molecular weight under the denaturing conditions: 75 kDa) (Zhang et al. 2009), which was co-purified from the culture medium used for the secretory production of NFG1CG4-hFasLECD in P. pastoris (Muraki 2014b).
Figure 3a presents the high-performance size-exclusion chromatography profiles for the partially purified samples of FL-5Mal conjugated NFG1CG4-hFasLECD and hFasRECD-Fc, and a mixture of them. The conjugated NFG1CG4-hFasLECD sample after the second size-exclusion chromatography using a gravity flow column showed one major peak eluted at 28.61 min (peak 2) and a small preceding peak at 24.70 min (peak 1) (Fig. 3a, top panel) with respect to the absorbance at both 280 and 495 nm. The hFasRECD-Fc sample after the anion-exchange chromatography purification showed a main peak at 23.32 min presenting only the absorbance at 280 nm (Fig. 3a, middle panel). The mixture sample showed a main peak at an earlier position (22.36 min) than the receptor alone sample with regard to the absorbance at both 280 and 495 nm (Fig. 3a, bottom panel), indicating a stable complex formation derived from the strong interactions between them. The two independent peak fractions regarding the FL-5Mal conjugated NFG1CG4-hFasLECD sample, the main peak fraction concerning the hFasRECD-Fc sample and that of the mixture sample were collected and the components were analyzed by SDS-PAGE (Fig. 3b, lanes 2–5). In addition, the two peak fractions of the FL-5Mal conjugated NFG1CG4-hFasLECD sample were also examined by a receptor-mediated co-immunoprecipitaion experiment (Fig. 3b, lanes 6–9). The peak 2 sample showed an almost uniform molecular weight slightly larger than NFG5-hFasLECD (Fig. 3b, lane 1) due to the addition of FL-5Mal moiety and no longer contained the AOX-1 impurity (Fig. 3b, lanes 2 and 6), which was still observed in the sample after the second gravity flow size-exclusion chromatography purification. This sample precipitated with Protein A conjugated magnetic beads in the presence of hFasRECD-Fc, confirming the specific binding activity of the hFasLECD conjugate to the hFasRECD-Fc (Fig. 3b, lane 7). The peak 1 sample was a mixture of mainly three components, showing discrete molecular-weights (Fig. 3b, lanes 3 and 8). Interestingly, only the asterisked component showing the highest molecular-weight did not co-precipitated with the hFasRECD-Fc (Fig. 3b, lane 9). The main peak fraction obtained from the mixture sample contained both the ligand and the receptor components (Fig. 3b, lane 5). Under the same resolving conditions, molecular-weight standard samples of Thyrogloblin (669 kDa), Aldolase (158 kDa) and Ovalbumin (43 kDa) showed the peak retention time of 17.56, 24.28 and 29.20 min, respectively (Additional file 2). Purification of the FL-5Mal conjugated NFG1CG4-hFasLECD sample using a cation-exchange chromatography was not effective because of the broadening of the peak concomitant with a significant retardation probably due to the attached hydrophobic Fluorescein moieties (Additional file 3).
The main peak sample of FL-5Mal conjugated NFG1CG4-hFasLECD purified by high-performance size-exclusion chromatography was then characterized using a couple of spectroscopic analyses. As shown in Fig. 4a, the sample presented two major peaks at 280 and 495 nm, which can be attributed to mainly from the protein part and the conjugated Fluorescein group, in the UV–Vis spectrum, respectively. From the ratio of the absorbance value at 280 nm to that at 495 nm, the conjugation number of FL-5Mal per a single NFG1CG4-hFasLECD trimer was estimated to be 2.5.
This result indicated that approximately 83 % of the unpaired cysteine residues in the NFG1CG4-hFasLECD sample were modified by the FL-5Mal molecules, since each single monomer constituting the trimer has one conjugation site within its N-terminal tag sequence. On the other hand, the measurement using a spectrofluorometer excited at 495 nm presented the characteristic emission spectrum derived from a Fluorescein moiety (BD Bioscience 2016) showing an intense yellow–green fluorescence with the maximum wavelength at 520 nm (Fig. 4b).
Finally, the complex formation of FL-5Mal conjugated NFG1CG4-hFasLECD with hFasRECD-Fc was further examined more in detail by altering the relative mixing ratio of the main peak samples purified by the high-performance size-exclusion chromatography (Fig. 5, panels a–d). The 495 nm absorbance observed only with FL-5Mal conjugated NFG1CG4-hFasLECD was helpful in monitoring the elution position of this component. By gradually increasing the amount of FL-5Mal conjugated NFG1CG4-hFasLECD under the fixed amount of hFasRECD-Fc in the sample mixture, the 495 nm peak showing constant elution time (22.74 min) corresponding to the formed complex first became higher (panels a and b) and then the peak of the left over ligand component after saturation appeared at 28.02 min or 28.50 min (panels c and d). This phenomenon suggested that the peak fraction eluted at 22.74 min was consisted of the complex formed from a limited number of the ligand component and the receptor component.