|Home | About | Journals | Submit | Contact Us | Français|
Successful cancer therapies aim to induce selective apoptosis in neoplastic cells. The current suboptimal efficiency and selectivity drugs have therapeutic limitations and induce concomitant side effects. Recently, novel cancer therapies based on the use of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) have emerged. TRAIL, a key component of the natural antitumor immune response, selectively kills many tumor cell types. Earlier studies with recombinant TRAIL, however, revealed its many shortcomings including a short half-life, off-target toxicity, and existence of TRAIL-resistant tumor cells. We improved the efficacy of recombinant TRAIL redesigning its structure and the expression and purification procedures. The result is a highly stable leucine zipper (LZ)-TRAIL chimera that is simple to produce and purify This chimera functions as a trimer in a manner that similar to natural TRAIL. The formulation of the recombinant LZ-TRAIL we have developed has displayed high specific activity in both cell-based assays in vitro and animal tests in vivo. Our results have shown that the half-of LZ-TRAIL is improved and now exceeds 1 h in mice compared with a half-life of only minutes reported earlier for recombinant TRAIL. We have concluded that our LZ TRAIL construct will serve as a foundation for a new generation of fully human LZ-TRAIL proteins suitable use in preclinical and clinical studies and for effective combination therapies to overcome tumor resistance TRAIL.
Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) is a unique member of the tumor necrosis factor family. TRAIL induces apoptosis in a wide variety of neoplastic cells but not in normal cells (1, 2). TRAIL is a type II membrane protein and its membrane form is cleaved to produce a soluble, biologically active cytokine (1). Natural TRAIL is a homotrimer. Its single Zn atom is chelated by the cysteine residue of each monomer. The presence of Zn is essential for the biological activity of TRAIL (3).
TRAIL is a promising agent for cancer therapy because its effect is independent of the functional status of p53 (4–7) and because of its ability to induce apoptosis in malignant cells via both extrinsic and intrinsic pathways, thus increasing the probability of the apoptotic outcome (8). In both pathways, TRAIL induces apoptosis by binding to TRAIL receptor 1 [death receptor 4 (DR4)] and TRAIL receptor 2 [death receptor 5 (DR5)], which are widely expressed in cancer cells (9–11). These interactions lead to trimerization of DR4/DR5 and their intracellular death domains and to the formation of the death-inducing signaling complex. In the extrinsic pathway, death-inducing signaling complex recruits a Fas-associated protein with death domain adaptor with the subsequent activation of apical caspase-8 and -10 (12), which then activate the “executioner” caspase-3 followed by the cleavage of the death substrates and cell death.
In the intrinsic pathway, TRAIL stimulates the cleavage of the proapoptotic Bcl-2 family member Bid by caspase-8. Truncated Bid translocates to the mitochondria, interacts with Bax and Bak, and promotes the release of cytochrome c and SMAC/DIABLO into the cytosol (13). Binding of cytochrome c to the adaptor protein APAF-1 induces the formation of “apoptosome” that activates caspase-9, which then activates the “executioner” caspase-3, -6, and -7, leading to cell death. Antiapoptotic Bcl-2 family members Bcl-2 and Bcl-xL proteins block the release of cytochrome c and suppress the intrinsic pathway (14).
The existing formulations of recombinant TRAIL are not uniformly effective because of their instability and low activity. These deficiencies are further aggravated by a short half-life in the blood and also because of both the original and the acquired resistance of certain cancers to TRAIL. Here, we describe a reengineered leucine zipper (LZ)-TRAIL and novel preparation techniques, the combination of which generates a therapeutic agent prototype capable of efficiently causing malignant cell death. Our reengineered TRAIL is a stable trimer, and when compared with the published results by others, it displays an improved bioavailability and antitumor activity over the known recombinant preparations.
All reagents unless otherwise indicated were from Sigma. TRAIL isolated from Escherichia coli and a rabbit antibody against TRAIL were from Peprotech. Rabbit antibodies against DR4 (AB16955), DR5 (AB16942), DcR1 (AB16509), and DcR2 (AB16943), a TMB/M substrate, and the enzyme-free cell dissociation solution were from Chemicon. Rabbit anti-mouse asialo-GM-1 antibody was from Cedarline. Pichia pastoris X-33 strain and the expression vector pGAPZα were from Invitrogen.
Apogossypol and BI-21E11, which target antiapoptotic Bcl-2 family proteins, and BI-75D2, a X-linked inhibitor of apoptosis protein (XIAP) antagonist, targeting its Bir3 domain, were synthesized and purified as described earlier (15–19). MLS0092727 (compound ID 3380841) was identified by high-throughput screening of the NIH Molecular Libraries Small Molecule Repository,1 which consisted of >200,000 compounds.
The human prostate carcinoma PPC-1 and PC-3, breast carcinoma MCF7, MDA-MB-435, and MDA-MB-231, leukemia THP-1, glioma U251, and mouse breast carcinoma 4T1 cell lines were obtained from the American Type Culture Collection. Normal human mammary epithelial 184B5 cells and primary human hepatocytes were from Lonza. Cancer cells were cultured in DMEM supplemented with 10% fetal bovine serum and 10 μg/mL gentamicin. 184B5 cells and hepatocytes were cultured in mammary epithelial cell growth medium and hepatocyte maintenance medium, respectively (Lonza).
The cDNA encoding the fragment 120-281 of human TRAIL was synthesized by Integrated DNA Technologies using the preferred P. pastoris codons (20). The synthesized fragment was linked to the modified yeast GCN4-pII LZ motif (RMKQIEDKIEEILSKIYHIENEIARIKKLIGER; ref. 21) and cloned into the pGAPZα plasmid (Invitrogen). The pGAPZα plasmid was modified to replace the original Lys-Arg-Glu-Ala-Glu-Ala sequence, which included the Kex2 and Ste13 cleavage sites, with the Ser-Arg-Lys-Lys-Arg-Ser sequence that represented the modified Kex2 cleavage site. Additional construct (named intermediate) included the Lys-Arg-Asn-Ser Kex2 cleavage sequence. P. pastoris X-33 cells were electroporated with the resulting pGAPZα-LZ-TRAIL plasmid. The medium aliquots were analyzed by Western blotting with the TRAIL antibody. The most efficient yeast clones were used for purifying the TRAIL constructs.
For the scale-up purification of LZ-TRAIL, yeast cells were grown for 2 days at 30°C in YPD medium (1 L) containing 1% casamino acids, 1 mmol/L Tris-(2-carboxyethyl) phosphine, and 100 mmol/L potassium phosphate buffer (pH 7.4) supplemented with 0.3% glycerol and 0.25 mol/L (NH4)2SO4. Next, the cells were removed by centrifugation. The medium was 50-fold concentrated using the Pellicon XL filtration device (Millipore). After buffer exchange for 20 mmol/L sodium phosphate buffer (pH 7.4) supplemented with 0.5 mol/L NaCl, LZ-TRAIL was purified by Co2+-metal chelating chromatography and eluted with a 0 to 25 mmol/L imidazole gradient.
The cDNA coding for the Thr95-Gly281 fragment of human TRAIL was cloned from the cDNA library (Clontech). The fragment was linked to the GCN4-pII LZ sequence. The resulting LZ-TRAIL-95 construct was cloned into the pGAPZα plasmid as described above.
Cells were surface biotinylated for 60 min on ice using sulfo-NHS-LC-biotin (Pierce) and lysed in TBS (pH 7.4) supplemented with 50 mmol/L N-octyl-β-D-glucopyranoside (Amresco), 1 mmol/L CaCl2, 1 mmol/L MgCl2, and a protease inhibitor cocktail containing 1 mmol/L phenylmethylsulfonyl fluoride and 1 μg/mL each of aprotinin, pepstatin, and leupeptin. The biotin-labeled proteins were precipitated from the cell lysate aliquots (1.0 mg total protein each) using streptavidin-agarose beads. The precipitates were analyzed by Western blotting with the DR4, DR5, DcR1, and DcR2 antibodies (1 μg/mL) followed by a horseradish peroxidase-conjugated goat anti-rabbit antibody and a TMB/M substrate.
Subconfluent cells grown in DMEM supplemented with 10% fetal bovine serum in wells of a 96-well plate were co-incubated for 24 h with increasing concentrations of TRAIL. The level of induced apoptosis was determined using an ATP-Lite kit (Perkin-Elmer).
The Caspase-Glo 3/7 luminescent assay (Promega) was used to determine caspase-3/7 activity. The resulting luminescence was measured using a plate reader (Tecan).
Sedimentation equilibrium experiments were done using a ProteomeLab XL-I analytical ultracentrifuge (Beckman-Coulter). TRAIL (0.5, 0.17, and 0.06 mg/mL aliquots in PBS) were loaded in the six-channel equilibrium cells and spun at 20°C for 24 h in an An-50 Ti 8 rotor at 18,000 rpm. Data were analyzed using HeteroAnalysis software (provided by J.L. Cole and J.W. Lary, University of Connecticut).
Differential scanning calorimetry of LZ-TRAIL (0.5 mg/mL in PBS) was done at a scanning rate of 1 K/min under 3.0 atm of pressure using a N-DSC II differential scanning calorimeter (Calorimetry Sciences).
Cells were incubated with LZ-TRAIL for 2 h. Phosphatidylserine exposed on the cells was measured by an Annexin V-FITC kit. Cells (5 × 105-1 × 106) were washed with PBS, pelleted, and resuspended in 500 μL Annexin V-FITC diluted 1:100 in the binding buffer (10 mmol/L HEPES, 100 mmol/L NaCl, 10 mmol/L KCl, 1 mmol/L MgCl2, and 1.8 mmol/L CaCl2) containing propidium iodide (1:50). Cells were incubated for 10 to 15 min in dark at an ambient temperature and then subjected to FACScan analysis using a FACScan (Becton Dickinson) with a 488 nm laser line and the data were analyzed using Cell Quest software. Annexin V-FITC fluorescence was detected in FL-1, and propidium iodide was detected in FL-3.
For the detection of DcR1, DcR2, DR4, and DR5 TRAIL receptors, MCF7, U251, and PPC-1 tumor cells and normal human hepatocytes were grown to reach confluence. Cells (5 × 105-1 × 106) were collected, washed with PBS, fixed, and permeabilized for 5 min using 70% ethanol and then incubated for 1 h with the DcR1, DcR2, DR4, and DR5 antibodies (1:100 dilution) in 10 mmol/L HEPES, 100 mmol/L NaCl, 10 mmol/L KCl, 1 mmol/L MgCl2, and 1.8 mmol/L CaCl2 followed by a 1 h incubation with the secondary FITC-labeled goat anti-rabbit IgG (Molecular Probes; 1:300 dilution). The samples were evaluated using a FACScan (Becton Dickinson) with a 488 nm laser line and the resulting data were analyzed using Cell Quest software. FITC fluorescence was detected in FL-1.
To measure the antitumor activity of LZ-TRAIL, we used the orthotopic mammary tumor xenografts in immunodeficient mice. To inactivate natural killer cells, four groups of athymic female CB.17-SCID mice (5–7 mice per group) received an i.v. injection of the asialo-GM-1 antibody (0.1 mg/animal). Immediately after this injection, MDA-MB-231 cells (1 × 106 in 0. 1 mL PBS) were injected in the mammary gland. The first three groups received 0.5, 1.5, or 5 mg/kg i.p. injections of purified LZ-TRAIL daily for 10 days, whereas the fourth group (control) received PBS. On day 4, the i.v. injection of the asialo-GM-1 antibody was repeated. After termination of TRAIL injections, tumor growth was monitored for additional 18 days. Mice were then sacrificed. Tumors were excised, cleaned from the connective tissue, measured, weighed, and photographed.
Because natural TRAIL is a homotrimer (3), we reengineered TRAIL to improve its stability and its activity and to increase its rate of production by the host. According to the cocrystal structure, the extracellular domain commencing with Glu120 is sufficient for a TRAIL complex with the DR5 receptor (22, 23). In agreement, the soluble 119-281 TRAIL fragment induced apoptosis in tumor cells (24). In contrast, recombinant TRAIL commencing from either Thr95 or Val114 was used in the earlier studies (9, 10, 25, 26). The formation of a stable recombinant TRAIL trimer was induced by linking the yeast GCN4-pII LZ sequence to the NH2 terminus of TRAIL (10). Similarly, to prepare the TRAIL construct, we synthesized, using the P. pastoris preferred nucleotide codons, the chimera coding for the modified three-stranded coiled coils GCN4-pII LZ (21) followed by a short Gly-Ser-Gly linker and the TRAIL 120-281 fragment. These synthetic constructs were recloned in the pGAPZα plasmid and electroporated into P. pastoris. For comparison, we also prepared the 95-281 TRAIL construct, which was described in the earlier publications (9, 10, 26). Based on these publications, we expected that the insertion of this trimer-forming LZ motif would promote the level of the homotrimerization of our TRAIL 120-281 and that the LZ-TRAIL chimera would mimic the performance of the natural TRAIL homotrimer.
To facilitate the efficient secretion and processing of the construct by the host and to generate the correctly processed and folded, secreted LZ-TRAIL protein, the sequence coding for the signal peptide followed by either Ser-Arg-Lys-Lys-Arg-Ser or Lys-Arg-Asn-Ser was linked to the NH2 terminus of LZ-TRAIL. The Ser-Arg-Lys-Lys-Arg-Ser and Lys-Arg-Asn-Ser sequences served as the cleavage sites for the yeast prohormone processing Kex2 proteinase. We then evaluated the performance of the multiple signal peptide sequences including the signal peptides of human pancreatic lipase-related protein 2 (27), Aspergillus niger cinnamoyl esterase (28), Pleurotus sajor-caju laccase 4 (29), Aspergillus awamori glucoamylase (30), Thermus aquaticus YT-1 aqualysin (31), and 128 kDa pGKL killer protein (32), all of which were reported to support the efficient secretion of the recombinant proteins in Pichia. Our tests showed that the LZ-TRAIL construct, which included the α-factor signal peptide (which was present in the pGAPZα plasmid) and the Ser-Arg-Lys-Lys-Arg-Ser cleavage sequence, was the most efficient (Supplementary Table S1).2
To increase the efficiency of the translation termination of the LZ-TRAIL construct, we replaced the TAA stop codon with TAAA (33). We also inserted the AOX1 5′-untranslated region upstream of the initiation ATG codon of the α-factor signal peptide (20). To increase the production of TRAIL by yeast, we reconstituted the original 3′-terminal sequence (CAAAACACA) of the glyceraldehyde-3-phosphate dehydrogenase promoter in the pGAPZα expression vector (34, 35). The constructs of LZ-TRAIL are shown in Fig. 1A.
We then optimized the medium composition for the growth of the recombinant yeast and the production of LZ-TRAIL, including the buffer composition and the pH. As a result, we determined that the YPD medium supplemented with 100 mmol/L potassium phosphate buffer (pH 7.4), 1% casamino acids, 1 mmol/L Tris-(2-carboxyethyl)phosphine, 0.3% glycerol, and 0.25 mol/L (NH4)2SO4 was optimal for generating high levels of the secretory, correctly processed LZ-TRAIL construct in the medium. Our optimization efforts increased the levels of LZ-TRAIL from a submilligram per liter level to 5 mg/L (Supplementary Fig. S1).2
Lastly, we streamlined the purification procedures that led us to isolate the homogenous LZ-TRAIL protein. The procedure we developed included the removal of yeast by centrifugation, the concentration of the medium using a Pellicon filtration device, and the isolation of LZ-TRAIL by metal-chelating chromatography (Supplementary Fig. S2).2 Overall, the reengineering we performed combined with our expression optimization allowed us to produce high quantities of the purified soluble LZ-TRAIL chimera that displayed high stability and activity.
The fusion of the LZ motif with TRAIL 120-281 resulted in a highly stable homotrimer. According to an ultracentrifugation analysis, the molecular mass of LZ-TRAIL was 73.4 kDa (the calculated molecular mass of the trimer is 69.3 kDa; Fig. 1B). The sedimentation equilibrium data were analyzed using the Ideal equilibrium and Monomer-Nmer equilibrium models. Both models suggested that purified LZ-TRAIL was a homotrimer and that the level of the monomer was below the detection limits (10 nmol/L) of the method.
The results of the differential scanning calorimetry also indicated the presence of a single oligomeric LZ-TRAIL protein in the samples. The melting temperature (Tm) and the calorimetric enthalpy (ΔH) values of LZ-TRAIL were 81.1°C and 53.3 kcal/mol, respectively, suggesting that the LZ-TRAIL 120-281 chimera we designed and isolated was properly folded and highly stable (Fig. 1C). In agreement, LZ-TRAIL did not lose its activity after 6 months of storage at 4°C or after multiple freeze-thaw cycles, but the LZ-TRAIL 95-281 construct consistently and completely lost its activity under similar conditions (data not shown). Because LZ-TRAIL 120-281 was missing a potential N-glycosylation site (Asp109-Ile-Ser), the construct was not glycosylated.
To elaborate on the significance of the Zn-binding Cys230 residue on the antitumor activity of LZ-TRAIL, we mutated Cys230 into Gly, Ala, Val, Leu, Ile, His, and Ser. All mutations, except Cys230Ile and Cys230Leu, fully inactivated LZ-TRAIL. The Cys230Ile and Cys230Leu mutants displayed a 10- to 20-fold reduced proapoptotic activity against prostate carcinoma PPC-1 cells compared with the original LZ-TRAIL. We concluded that the presence of Cys230 is of critical importance for the activity or the stability of LZ-TRAIL or both.
We tested if LZ-TRAIL induced apoptosis in prostate carcinoma PPC-1 and PC-3, breast carcinoma MDA-MB-231, lung carcinoma SK-MES-1, glioma U251, and leukemia THP-1 cells (Fig. 2A). LZ-TRAIL was efficient in the nanogram per milliliter concentration range and this range was multifold lower compared with the previously reported levels. A direct, side-by-side comparison of our LZ-TRAIL with the commercially available TRAIL samples confirms this suggestion (Fig. 2A). In agreement, in earlier studies by others (36, 37), recombinant TRAIL in concentrations as high as 50 to 100 ng/mL induced only 40% and only 15% apoptosis in PC-3 and PPC-1 cells, respectively. These and similar studies with recombinant TRAIL led to the conclusion that PPC-1 cells are TRAIL-resistant. In contrast, our LZ-TRAIL (1 ng/mL) induced an ~90% level of apoptosis in PPC-1 and PC-3 cells. Our results using prostate carcinoma PPC-1 clearly showed that the performance of TRAIL 120-281 was significantly improved over that of the 95-281 construct (Fig. 2A). Similarly, 3 ng/mL LZ-TRAIL commencing with Thr95 and purified from the stably transfected Chinese hamster ovary cells did not affect the viability of MDA-MB-231 cells (10). In contrast, our 3 ng/mL LZ-TRAIL caused a 70% level of apoptosis of these cells. As a result, the studies involving the 95-281 TRAIL were not pursued further.
Recombinant TRAIL at 30 ng/mL was reported to induce a 50% level of apoptosis in lung carcinoma SK-MES-1 cells (25). To directly compare these data with the activity of our LZ-TRAIL, we specifically used SK-MES-1 cells. We determined that LZ-TRAIL 120-281 at only 0.3 ng/mL induced the same 50% level of apoptosis in SK-MES-1 cells, confirming the high specific activity of our LZ-TRAIL chimera (Fig. 2A).
To corroborate these results, we used FITC-conjugated Annexin V for detection of phosphatidylserine exposed in the membrane of apoptotic cells and for quantification of cells undergoing apoptosis. According to a fluorescence-activated cell sorting analysis of Annexin V-stained cells, 34% of mouse breast carcinoma 4T1 cells became apoptotic following the treatment with 30 ng/mL LZ-TRAIL 120-281 (Fig. 2B). In contrast, LZ-TRAIL in concentrations as high as 10 μg/mL did not induce apoptosis in normal mammary epithelial 184B5 cells. As shown by both Annexin V staining and cell viability tests with an ATP-Lite kit, LZ-TRAIL was noncytotoxic to primary human hepatocytes in a concentration as high as 30 μg/mL and incubation as long as 24 h.
To support our results further, we determined if the caspase-3/7 activity was elevated in MDA-MB-231 cells coincubated with increasing concentrations of LZ-TRAIL. Consistent with the induction of a significant level of apoptosis by 1 to 3 ng/mL LZ-TRAIL in MDA-MB-231 cells (Fig. 2A), these low concentrations also induced a major increase of caspase-3/7 activity, again confirming the acute antitumor cytotoxicity of our optimized chimera (Fig. 2C).
Apoptosis resistance commonly occurs in cancers, preventing activation of caspase family cell death proteases. Antiapoptotic Bcl-2 family proteins neutralize the proapoptotic effects of prodeath effectors including Bax and Bak, hence conferring a growth advantage and resistance to chemotherapy and radiation on cancer cells (38). The antiapoptotic XIAP, and specifically the XIAP baculovirus IAP repeat 3 (Bir3) domain, is also a promising anticancer target because XIAP directly binds caspase-9 and inhibits cell death. Potent small-molecule inhibitors of both antiapoptotic Bcl-2 family proteins and XIAP are known to revert cancer cell resistance to chemotherapy (38) and to sensitize cancer cells to TRAIL-induced apoptosis (39).
To confirm that the inhibitors of Bcl-2 family proteins (apogossypol and BI-21E11) and XIAP (BI-75D2) sensitized cancer cells to our LZ-TRAIL chimera, we used TRAIL-resistant breast carcinoma MDA-MB-435 cells (Fig. 3A). The individual compounds, at either 2 or 10 μmol/L, and 10 ng/mL LZ-TRAIL, when tested alone, were inefficient in inducing apoptosis in these cells. In turn, if used jointly with LZ-TRAIL, the compounds, especially apogossypol, enhanced the cytotoxic properties of TRAIL in MDA-MB-435 cells.
The isobologram analysis of apogossypol, MLS0092727, and LZ-TRAIL showed a significant potentiation of LZ-TRAIL cytotoxicity by MLS0092727 in breast carcinoma MDA-MB-435 (Fig. 3B). Isobologram analysis was done according to ref. 40.
We next determined if the levels of the DR4/DR5 receptors correlate with the sensitivity of cells to LZ-TRAIL. For these purposes, prostate cancer PPC-1 (highly sensitive to TRAIL), glioma U251 (sensitive to TRAIL), and breast carcinoma MCF7 cells (resistant to TRAIL) and normal hepatocytes (highly resistant to TRAIL) were labeled with membrane-impermeable biotin.
Biotin-labeled cell proteins were captured using streptavidin-agarose beads. The precipitates were analyzed by Western blotting with the DR4, DR5, DcR1, and DcR2 antibodies (Fig. 4). Our results suggested that both sensitive and resistant cell lines expressed comparable levels of the DR5 receptor. The levels of DR4 were low in hepatocytes and especially in U251 cells. The level of DcR2 and especially DcR1 were significantly higher in hepatocytes relative to tumor cells. Fluorescence-activated cell sorting analysis of U251, PPC-1, and MCF7 cells and hepatocytes also showed the high levels of DcR1/DcR2 in the latter and confirmed the data of our immunoblotting studies (Fig. 4). Our data are consistent with the earlier observations by others that normal cells express high levels of the TRAIL decoy receptors DcR1/DcR2 and, as a result, are resistant to TRAIL-induced apoptosis (41). The permeabilization procedures we used in our fluorescence-activated cell sorting analyses allowed us to detect the intracellular DR4 in U251 cells. Because only the cell surface-associated DR4 was measured in biotin-labeled U251 cells, the intracellular DR4 was not observed in our immunoblotting studies. In agreement with the studies by others (42), we concluded that DR4 is retained intracellularly in U251 cells.
Earlier, the antitumor efficacy of LZ-TRAIL 95-281 had been determined using the MDA-MB-231 s.c. xenograft model in mice (10). To determine the antitumor efficacy of our chimera and to allow a direct comparison of the results, we also used MDA-MB-231 cell xenografts.
MDA-MB-231 cells were orthotopically xenografted in immunodeficient mice. In the earlier studies by others, animals normally received high dosages of TRAIL. Thus, MDA-MB-231 xenograft mice received ~5 mg/kg LZ-TRAIL 95-281 twice daily for 10 days (10). Because our in vitro studies suggested a high activity of LZ-TRAIL 120-281, we reduced the dosages in vivo. Animals were randomized into the four groups. Three groups received the 0.5, 1.5, or 5 mg/kg i.p. injections of LZ-TRAIL daily for 10 days, whereas the fourth group (control) received PBS alone. The 0.5 and 1.5 mg/kg dosages of LZ-TRAIL decreased tumor incidence but did not affect the size of the resulting tumors compared with the control (Fig. 5A). In turn, 5 mg/kg LZ-TRAIL caused a 2-fold reduction in tumor incidence and a 4-fold reduction in the tumor size, thus confirming the high potency of our LZ-TRAIL formulation. Administration of LZ-TRAIL was not toxic to the normal liver tissues of mice. Our examination of H&E-stained sections including liver sections did not reveal any changes to cell morphology or destruction of hepatocytes.
We suspected that the high efficacy of our LZ-TRAIL 120-281 in part results from its improved half-life and pharmacokinetics in animals. To test this suspicion, we determined the pharmacokinetics of our LZ-TRAIL in mice. LZ-TRAIL 120-281 (100 μg/mouse or 5 mg/kg) was injected i.v. in mice. Blood samples (50 μL) were obtained from the tail vein followed by plasma recovery. The plasma samples were diluted and the cytotoxicity of the diluted samples was measured using prostate carcinoma PPC-1 cells as a target (Fig. 5B). Our measurements showed the presence of LZ-TRAIL in the bloodstream even 24 h post-injection and suggested that a half-life of LZ-TRAIL 120-281 exceeded 1 h in mice. This figure is quite favorable when compared with a half-life of only minutes reported earlier for recombinant TRAIL (26).
To evaluate if the injections of LZ-TRAIL led to acute hepatotoxicity in mice, we measured the levels of the liver alanine aminotransferase and aspartate aminotransferase enzymes in the serum (Fig. 5C). Our measurements suggested that the levels of both enzyme were close to the normal range. Because, generally, a grade 1 hepatotoxicity is characterized by a 1.25- to 2.5-fold increase of these enzyme levels in the mouse serum, we concluded that our LZ-TRAIL samples did not cause any significant hepatotoxicity in mice (43).
To facilitate the pilot-scale production of TRAIL, we specifically selected the methylotrophic yeast P. pastoris as a host. The absence of the endotoxic lipopolysaccharides, a feature of E. coli, and of viral DNA and oncogene contamination, which is possible in mammalian cells, makes P. pastoris an attractive system for producing recombinant proteins that can serve as therapeutic agents.
A recombinant TRAIL monomer frequently displays toxicity and low specific activity, thus limiting its clinical applications (9, 44). To make recombinant TRAIL similar to natural TRAIL, which functions as an active homotrimer (3), we optimized the previously described LZ-TRAIL 95-281 construct (9, 10, 45). The presence of the NH2-terminal LZ sequence stabilizes the LZ-TRAIL homotrimer. The extracellular domain commencing with Glu120 is sufficient for TRAIL to form a complex with the receptor DR5 (22, 23). Because of these data, we specifically elected to express the TRAIL 120-281 fragment. The TRAIL 95-281 was also expressed and purified for comparison purposes. Our studies clearly revealed that the performance of TRAIL 120-281 was significantly improved over that of the 95-281 construct.
In the course of our study, we modified the secretion and processing sequences of TRAIL as well as the 3′-and 5′-sequences of the recombinant gene, which were favorable for the transcription initiation and termination in yeast. We then optimized the growth conditions, the composition of the nutrient medium, and the purification procedures. In sum, a favorable combination of our structural and technological improvements resulted in the LZ-TRAIL 120-281 stable homotrimer that exhibited high cytotoxicity in tumor cells.
Both a comparison of our data with those available in the literature and with the performance of the LZ-TRAIL 95-281 suggest that the LZ-TRAIL 120-285 homotrimer we designed and purified exceeds the antitumor activity of the previously reported TRAIL constructs (Table 1). In our cell and animal systems, we did not identify any significant level of toxicity of LZ-TRAIL 120-285 to normal cells, including human hepatocytes. It appears that the undesirable off-target effects are limited with the LZ-TRAIL 120-285 homotrimer especially when compared with those of the earlier recombinant TRAIL preparations (25, 44, 45). Our animal studies using an orthotopic xenograft model of breast cancer confirmed the antitumor activity of LZ-TRAIL. Treatment of mice with low dosages of LZ-TRAIL (5 mg/kg/d) resulted in a significant decrease in tumor incidence combined with a 75% retardation of tumor growth. Our additional studies showed that LZ-TRAIL displayed a half-life of over 1 h and retained its bioactivity in mice as long as 24 h post-injection. These favorable results are in contrast with the short (3.5 min) half-life reported for recombinant TRAIL (26). A combination of LZ-TRAIL with the proapoptotic small-molecule inhibitors including apogossypol (an antagonist of antiapoptotic Bcl-2 family proteins) provides an opportunity to overcome the resistance of tumor cells to apoptosis and suggests a potential use of the fully human derivatives of the LZ-TRAIL chimera in a combination therapy.
We believe that our results provide a rationale for transforming the yeast LZ-TRAIL construct into a fully human trimeric recombinant TRAIL and for the new generation of TRAIL proteins for preclinical and clinical studies. For these purposes, we will replace the three-stranded coiled coils yeast LZ motif with the LZ sequence derived from human matrilin 1 and coronin A1 (46). These redesigning activities are now in progress. Combination with chemotherapeutic agents may further enhance the efficacy of LZ-TRAIL and overcome tumor resistance to TRAIL therapies (47–49).
Grant support: NIH grants CA83017, CA77470, and RR020843 and Susan G. Komen Breast Cancer Foundation grant BCTR0601231 (A.Y. Strongin).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.