Improvement of Pharmacokinetic Profile of TRAIL via Trimer-Tag Enhances its Antitumor Activity in vivo

Abstract

TNF-related apoptosis-inducing ligand (TRAIL/Apo2L) has long been considered a tantalizing target for cancer therapy because it mediates activation of the extrinsic apoptosis pathway in a tumor-specific manner by binding to and trimerizing its functional receptors DR4 or DR5. Despite initial promise, both recombinant human TRAIL (native TRAIL) and dimeric DR4/DR5 agonist monoclonal antibodies (mAbs) failed in multiple human clinical trials. Here we show that in-frame fusion of human C-propeptide of α1(I) collagen (Trimer-Tag) to the C-terminus of mature human TRAIL leads to a disulfide bond-linked homotrimer which can be expressed at high levels as a secreted protein from CHO cells. The resulting TRAIL-Trimer not only retains similar bioactivity and receptor binding kinetics as native TRAIL in vitro which are 4-5 orders of magnitude superior to that of dimeric TRAIL-Fc, but also manifests more favorable pharmacokinetic and antitumor pharmacodynamic profiles in vivo than that of native TRAIL. Taken together, this work provides direct evidence for the in vivo antitumor efficacy of TRAIL being proportional to systemic drug exposure and suggests that the previous clinical failures may have been due to rapid systemic clearance of native TRAIL and poor apoptosis-inducing potency of dimeric agonist mAbs despite their long serum half-lives.

Figure 1

Structures of different TRAIL polypeptides. (a) Three recombinant soluble forms of human TRAIL have been used in this study: a native TRAIL comprised of the extracellular domain of TRAIL, a dimeric TRAIL-Fc comprised of extracellular TRAIL domain fused to human IgG1 Fc domain, and a TRAIL-Trimer comprised of extracellular TRAIL domain fused to Trimer-Tag. Amino acid sequence lengths are shown for each protein and domain respectively. (b) Theoretical molecular weights (kDa) of both monomeric and multimeric forms of native TRAIL, TRAIL-Fc and TRAIL-Trimer, respectively. Native TRAIL associates into a non-covalently-linked homotrimer, TRAIL-Fc forms a covalently-linked homodimer, and TRAIL-Trimer forms a covalently-linked homotrimer.

Figure 2

High-level expression and purification of TRAIL-Trimer fusion protein. (a) 10% SDS-PAGE analysis of TRAIL-Trimer expression from a fed-batch serum-free cell culture in the bioreactor. 10 µL of cell-free conditioned medium from Day 6 to Day 13 were analyzed under non-reducing condition followed by Coomassie Blue staining. (b) Bioassay analysis of TRAIL-Trimer production in conditioned medium from Day 5 to Day 13. (c) SDS-PAGE and western blot analysis of purified TRAIL-Trimer under either non-reducing or reducing conditions. 2 µg of purified protein was analyzed by a 10% SDS-PAGE and stained with Coomassie Blue. 0.2 µg of purified protein was analyzed by western blot using monoclonal antibody against TRAIL-domain and Trimer-domain, respectively.

Figure 3

Purity evaluation of TRAIL-Trimer, TRAIL-Fc and native TRAIL. SDS-PAGE analysis of purified TRAIL-Trimer, TRAIL-Fc and native TRAIL under either non-reducing or reducing conditions. 2 µg of purified protein was analyzed by 10% or 15% SDS-PAGE, respectively, and stained with Coomassie Blue.

Figure 4

Bioactivity and affinity with DR5-Fc comparison of TRAIL-Trimer, TRAIL-Fc and native TRAIL. (a) Bioactivity detection of TRAIL-Trimer, TRAIL-Fc and native TRAIL. The IC₅₀ value of TRAIL-Trimer, TRAIL-Fc and native TRAIL were assessed using a TRAIL sensitive cell line COLO205 on a Real-Time Cell Analysis (RTCA) system. The IC₅₀ value was obtained according to the dose-response cell index (CI) curve. The IC₅₀ value of TRAIL-Trimer, TRAIL-Fc and native TRAIL is 23.2 ng/mL, 260.8 µg/mL, and 6.7 ng/mL, respectively. The bottom graph displays IC₅₀ values when molar ratio-adjusted based on theoretical molecular weights for each protein. (b) Kinetic parameters of TRAIL-Trimer, TRAIL-Fc and native TRAIL binding to the soluble DR5-Fc fusion protein was assessed by biolayer interferometry measurements. The Super Streptavidin biosensor tips of the ForteBio Octet RED 96 were coated with biotinylated DR5-Fc. The biosensor tips were dipped in increasing concentrations gradient of the same molarity (15.4 nM–123.5 nM) to measure binding of TRAIL-Trimer, TRAIL-Fc and native TRAIL to DR5-Fc and subsequently moved to wells containing buffer (PBS) to measure dissociation rates. Detailed results are shown in Table 1.

Figure 5

Pharmacokinetic profile of TRAIL-Trimer and native TRAIL determined by western blot. Mice were injected intravenously with equimolar amounts of TRAIL-Trimer (40 mg/kg) or native TRAIL (15 mg/kg) (n = 3 for each protein). (a,c) Relative serum concentrations of both TRAIL-Trimer (a) and native TRAIL (c) were evaluated at periodic intervals by western blotting. TRAIL proteins were immunoblotted with an antibody against human TRAIL, and visualized by chemiluminescence. The images were acquired using the ChemiDoc Touch Imaging System (Bio-Rad) and the quantification of the amount of TRAIL proteins present in each sample was done using Image Lab software (Bio-Rad). (b,d) Estimated half-lives of TRAIL-Trimer (b) and native TRAIL (d) were calculated using GraphPad Prism v5.04. Error bars represent ± SEM.

Figure 6

In vivo antitumor activities of TRAIL-Trimer and native TRAIL. (a) Nude mice with established COLO205 xenografts were given TRAIL-Trimer (30, 50, or 80 mg/kg/day) or vehicle as an i.v. bolus for 5 consecutive days (n = 6/group). Results shown are group mean (±S.D.). (b) Nude mice with established COLO205 xenografts were given the same molar concentration of TRAIL-Trimer and native TRAIL. Results shown are group mean (±S.D.).