Renal cell carcinoma (RCC) accounts for 2% of all cancers, leading to 20,000 annual deaths in Europe and 12,000 in the United States.1, 2 Although advances in understanding the biology of RCC led to novel approaches for the treatment of metastatic disease3 with subsequent increase of progression-free and overall-survival rates, the prognosis for these patients is poor with a 5-year survival rate of less than 10%.4 The predominant histological type of RCC is clear-cell carcinoma, comprising more than 85% of metastatic disease. Both, sporadic and inherited forms of clear-cell RCC are associated with mutations in the von Hippel-Lindau (VHL) tumor suppressor gene.5 Elucidation of its role in up-regulating growth factors associated with angiogenesis as well as the hypoxia-induced carbonic anhydrase IX (CA-IX) defined a series of potential targets for novel treatment strategies. Among them, targeting of a surface expressed epitope of CA-IX (called G250) using chimeric G250 monoclonal antibody (cG250) is a promising immunotherapeutic approach.6 Safe administration with excellent tumor targeting properties of the radio-labeled 131I- and 124I-cG250 antibody has been demonstrated in phase I clinical trials in patients with metastatic clear cell RCC.7–10 Additional studies showed that multiple doses of cG250 were well tolerated and combination with low dose IL-2 resulted in disease stabilization.11, 12 These encouraging clinical features prompted us to optimize effector functions of cG250 to further improve its antitumor properties.
We have previously reported on a strategy to promote the therapeutic efficacy of tumor specific antibodies by genetic fusion to tumor necrosis factor (TNF).13 The soluble form of TNF occurs mainly as a trimer of 3 identical subunits. TNF was first identified as a mediator of hemorrhagic tumor necrosis mediating regression of murine14 and xenotransplanted human tumors.15 Subsequent research revealed that Interferon gamma (IFNγ) plays a substantial role in TNF-mediated tumor rejection processes.16, 17 Many IFNγ-mediated mechanisms have been proposed to promote antitumor responses including antiproliferative and proapoptotic activity on tumor cells,18 inhibition of angiogenesis within tumors19, 20 or activation of innate21, 22 and adapted immune responses.23, 24 Thus, both TNF and IFNγ exert pleiotropic mechanisms on a variety of cell types and coadministration of both cytokines even results in synergistic antitumor activities.25–27 However, because of its life-threatening systemic toxicity28 caused by affection of normal endothelial cells in peripheral blood vessels,29, 30 the clinical use of TNF in combination with IFNγ as an anticancer drug is restricted to loco-regional treatment (e.g. isolated limb perfusion).31, 32 However, we established a 2-step approach to enlarge the therapeutic window of the cytokine: TNF molecules were genetically fused to a tumor-specific antibody to focus TNF at the tumor site. In addition, TNF subunits were forced to form a dimer to reduce their activity in peripheral blood vessels outside the tumor compartment.13 Abandoning the natural homotrimeric symmetry of TNF resulted in significantly reduced toxicity as seen both in immunodeficient and immunocompetent mouse strains.33 The dimeric IgG-TNF molecules displayed significantly stronger antitumor activity in-vivo than wild type TNF or trimeric TNF-antibody conjugates. Dose escalation increased the therapeutic index of IgG-TNF and repeated administration additionally delayed tumor growth with tolerable side effects.33
Here, we describe the construction, expression and purification of a cG250-TNF fusion protein and its preclinical evaluation in a RCC mouse model. Biodistribution studies in G250/CA-IX-positive RCC xenografted BALB/c nu/nu mice showed a significant increase of tumor-to-blood ratio over time with specific accumulation and retention of cG250-TNF in the tumor resulting in growth control of established RCC. Furthermore, a combination regimen with clinically well-tolerated doses of nontargeted IFNγ-induced synergistic activation of different tumoricidal pathways and increased significantly the antitumor response in vivo. Considering the poor outcome of patients with advanced renal cell cancer, cG250-TNF-based immunotherapeutic approaches could be a valuable therapeutic option.
Material and methods
Cell lines and reagents
RPMI 1640 medium [supplemented with 10% (v/v) heat-inactivated fetal bovine serum, penicillin (100 units/ml), streptomycin (0.1 mg/ml) and glutamine (0.3 mg/ml)] (all Gibco, Karlsruhe, Germany) was used as standard medium for all cell lines if not indicated otherwise. Human renal carcinoma cell lines NU-12, SK-RC-52 (high G250 expression level), SK-RC-17 (G250-negative), the cervix carcinoma cell line Me-180, and mouse fibrosarcoma cell lines WEHI-164 S (TNF sensitive) and the WEHI-164 R (TNF resistant; a WEHI-164 variant, cloned under the exposure to TNF) were supplied by Ludwig Institute for Cancer Research. HUVEC-c endothelial cell supplemented growth medium was obtained from PromoCell (Heidelberg, Germany). For selective cell culture mouse myeloma NSO cells were grown in Glutamine-free DMEM supplied by SAFC Biosciences (Lennexa, KS) according to Lonza's manual of operating procedures (Lonza, NH). Human recombinant TNF, human recombinant IFNγ, murine recombinant IFNγ and soluble TNF-R1 were purchased from Genzyme (Neu-Isenburg, Germany). Anti-FAP-TNF has been described previously.13 Antibodies were obtained from the following sources: rabbit antihuman tissue factor from American Diagnostica (Pfungstadt, Germany), rabbit antihuman ICAM-1 from Santa Cruz Biotechnology (Heidelberg, Germany), rabbit antihuman actin from Sigma (St. Louis, MO) PE-conjugated goat antirabbit from Dako (Glostrup, Denmark), murine anti-idiotypic anti-G250 and the chimeric G250 mAb were previously described.
Construction of cG250-TNF fusion protein
Reverse transcription-PCR on mRNA isolated from peripheral blood mononuclear cells for amplification of the mature human TNF cDNA sequence was previously described.13 The cDNA sequence coding for the cG250 variable heavy (HC) and light (LC) chain sequence was subcloned into pEAK8 mammalian cell expression vector (Edge BioSystems, Gaithersburg, MA) with the HC-expression vector containing the previously described human CH2/CH3-truncated IgG1 TNF13 C-terminally fused to His6 tag. cG250 LC was amplified by PCR with the forward primer (GAACC CGGGG CCGCC ACCAT GGGCA AGATG GAGTT TCATA CT) containing Kozak's sequence and a Sma I restriction side and the reverse primers containing the stop codon and a EcoR I restriction enzyme sequence. Amplification of the cG250 IgG1 CH1 TNF H6 sequence was done by sequential use of two reverse primers (CAAAG ATCTC AGGGC AATGA TCCCA AAGTA GAC and CAAGA ATTCT CAGTG ATGGT GATGG TGATG CAGGG CAATG ATCCC) and the forward primer introducing either the Kozak's and the Sma I restriction side sequence (GAACC CGGGG CCGCC ACCAT GAACT TCGGG CTCAG ATTG). Variable cG250 LC sequence was cloned into mammalian expression vector pEE12.4 by Sma I/EcoR I and, respectively, the H6-tagged HC-TNF sequence by Sal I/Not I in pEE6.4 (Lonza, Portsmouth, USA). The double-gene vector encoding the cG250-TNF fusion protein was created by digestion of both vectors with Not I and Sal I restriction enzymes and ligation of the pEE6.4 fragment into the larger pEE12.4 fragment according to Lonza's manual of operating procedures.
Generation of stable transfected cell lines, expression, purification and characterization of cG250-TNF fusion protein
Stable transfected NSO cell lines were established by electroporation of Pvu I linearized DNA using a GENPULSER (BioRad, Munich, Germany) according to manufacturer's instructions. Transfected cells are selected for their ability to grow in glutamine-free medium. Stable cell lines producing high level of antibodies were expanded, weaned off FCS and transferred into a Technomouse system (Integra Biosciences, Fernwald, Germany) for large scale production. Antibody supernatant from the Technomouse system was dialyzed against PBS (phosphate buffered saline, pH 7.2) overnight (4°C). The dialyzed sample was passed over Protein A and G sepharose columns (Pharmacia, Freiburg, Germany) in a 2-step procedure. Fusion proteins bound to Protein G were stepwise eluted by adding 0.1 M glycine/HCl (pH 3.5) and samples were rapidly neutralized by the addition of 1 M Tris buffer (pH 8.0). Endotoxin contamination of the final product was excluded by Limulus amebocyte assay (QCL 1000, BioWhittacker, Walkersville, MD). The size of the fusion protein was analyzed in reducing condition on SDS–PAGE and in native condition by FPLC-gel filtration on a Superdex S-200 chromatography column (Amersham Pharmacia, Freiburg, Germany).
Integrity of cG250-TNF in vitro
Structural integrity of cG250-TNF was assessed by sandwich-ELISA as described.13 In brief, 96-well flat-bottom microtiter plates were coated overnight (4°C) with Infliximab (antihuman TNF antibody), plates were blocked with 1.5% gelatine in PBS and the indicated reagents solved in PBS were added in serial dilutions (60 min, RT) and the bound cG250-TNF using a murine anti-cG250 anti-idiotypic antibody (1 μg/ml, 1 hr, room temperature) as described previously.13
TNF-receptor 1 mediated apoptosis was investigated on TNF-sensitive “WEHI-164 S” and on partially TNF-resistant34 “WEHI-164 R” as described.13 5 × 105 WEHI cells were cultured in 96-well plates in the presence of the indicated reagents. Apoptotic cells were identified by annexin V staining by flow staining as described.13
Synergistic bio-activity of cG250-TNF and IFNγ in vitro
Synergistic cytotoxic interaction of cG250-TNF and IFNγ was demonstrated by reduction of Me-180 cervical carcinoma cell viability as described.25, 35 Measurement of H2O2 release by stimulated human polymorphonuclear leukocytes (PMN) prepared from venous blood of healthy donors was done as described.13 Samples were taken at selected time points after incubation with triggering agents (37°C over 180 min in air). Activation of HUVEC endothelial cells cultured at second or third passage was done by incubation (6–24 hr at 37°C) with effectors at indicated dilutions and harvested as described.13 IP-10 expression levels were analyzed by RT-PCR (forward: AATCA AACTG CGATT CTGAT TTGC; reverse: AGGAG ATCTT TTAGA CATTT CCTT) and GADPH (forward: GTGAA GGTCG GAGTC AACGG ATTT; reverse: CTCCT TGGAG GCCAT GTGGG CCAT). Expression levels of tissue factor and ICAM-1 were detected by Western blot analysis.
Xenograft models, biodistribution and treatment protocols
Biodistribution and efficacy studies were done on xenografted 6–8-week-old athymic BALB/c nu/nu mice (Charles River Laboratories, Central Animal Facility, University Medical Center Nijmegen, The Netherlands). Renal cell tumor engraftment was achieved by subcutaneous injection of 5 × 106 SK-RC-17, SK-RC-52, or transplantation of NU-12 xenografts. The studies were approved by the local Animal Welfare Committee and performed in accordance with their guidelines.
Protein was radioiodinated with 131I or 125Iodine (MDS Nordion, Fleurus, Belgium) according to the IodoGen method and the immune-reactive fraction of the radiolabeled preparations was determined by Lindmo analysis with minor modifications as described7 and was 80–95%. Mice bearing subcutaneously established tumors were intravenously injected with 5, 25, or 50 μg of radioiodinated constructs. Cohorts of animals were sacrificed at predetermined timepoints after injection and tissues harvested, weighted and the uptake of 125I-cG250-TNF or 131I-cG250, respectively, was determined in a gamma-counter as described.36 Therapeutic efficacy of cG250-TNF alone or in combination with IFNγ was investigated in mice xenografted with G250-positive or -negative or both renal cell carcinomas and analyzed by daily monitoring of health and body weight of mice and measurement of tumor size on treatment days. Differences between treatment groups were tested for statistical significance by Student's t test.
Production and characterization of cG250-TNF
We established an antibody fusion protein named cG250-TNF that consists of human TNF and chimeric antibody cG250. The N-terminal domain of human TNF was linked to the IgG hinge region as described.13 The variable domains of cG250 were inserted through specific restriction sites into the pEE6.4 and pEE12.4 plasmids. The vector construct was confirmed by DNA sequencing (Fig. 1a). High-level expression of the recombinant fusion protein using NSO cells was achieved with about a 270 mg/l of culture supernatant and a specific productivity of 12.46 pg/cell/day (data not shown). The immunocytokine was purified using a sequential 2-step affinity chromatography with Protein A (flow-through) and Protein G (capture), respectively. Deletion of the CH2- and the CH3-domains of cG250 and their replacement with TNF-molecules reduced substantially the affinity of cG250-TNF to Protein A. During the first pass of the dialyzed sample over Protein A column, contaminating residual bovine globulins were captured. The flow through still contained the cG250-TNF fraction, and the construct was captured by a second pass over Protein G column. Fractions corresponding to dimeric species were collected during gel-filtration chromatography (Fig. 1b). Purity and identity of parental cG250 (Fig. 1b: lanes 1, 3, 5, 7) and cG250-TNF (Fig. 1b: lanes 2, 4, 6, 8) were characterized using SDS-PAGE under reducing and nonreducing conditions and Western blotting with anti-TNF. The recombinant heavy chain of cG250-TNF migrates with an apparent molecular weight of 47 kDa (Fig. 1b, lane 2) and is slightly smaller than a natural IgG heavy chain (Fig. 1b, lane 1). The human light chain is not altered with an apparent MW of 28 kDa. Western blotting of SDS–PAGE gels under both conditions and subsequent staining with anti-TNF identifies the heavy-chain of cG250-TNF (Fig. 1b, lanes 4 and 8). As expected, staining with anti-TNF failed to detect cG250 (Fig. 1b, Lanes 3 and 7). Correct folding of cG250-TNF hybrid protein was confirmed by sandwich ELISA with recognition of the TNF-part by Infliximab (antihuman TNF antibody) and by binding of the antibody part to an anti-cG250 anti-idiotypic antibody (Fig. 1c). The affinity constant for the targeted G250 antigen was identical for the cG250-TNF (Ka = 2.5 × 109 M−1) and parental cG250 (Ka = 2.2 × 109 M−1) construct as measured by ELISA and Scatchard blot analysis (data not shown). The capacity of dimerized TNF subunits in triggering TNF-receptor 1 (TNF-R1) dependent signaling was determined by WEHI-164 cytotoxicity assay system. With regard to different molarities of cG250-fused dimeric TNF and recombinant trimeric TNF, assays were performed at TNF-equivalent doses. In agreement with our previous results on dimeric TNF fusion proteins, TNF-R1 mediated cytotoxicity of cG250-TNF was approximately 10-fold lower when compared with trimerized, commercially available recombinant human TNF (rhTNF; Fig. 1d).
Synergistic biological activity of cG250-TNF and IFNγ in vitro
We next determined the capacity of dimeric cG250-TNF to induce tumoricidal activities with IFNγ in synergistic manner, as it is known for IFNγ and homotrimeric physiological TNF.26 As expected, incubation of the human cervical cancer cell line ME-180 with either IFNγ or cG250-TNF as single substances was rather ineffective even at high concentrations,25, 35 but combined incubation with IFNγ and cG250-TNF resulted in dose-dependent cell death (Fig. 2a). However, the efficacy of the combined administration of TNF and IFNγ as antitumor strategy depends rather on its effects on tumor vessels27 than on lysis of the malignant cells themselves. We, therefore, investigated the effect of cG250-TNF and IFNγ on key regulators of early and late events in vessel destruction. Activated polymorphonuclear leukocytes (PMN) participate in endothelial cell injury through generation of reactive oxygen species. On stimulation with hydrogen peroxide (H2O2) and under direct influence of TNF and IFNγ, endothelial cells up-regulate expression of adhesion molecules such as ICAM-1 (intercellular adhesion molecule 1) on their surface, thus rendering endothelial cells more susceptible to neutrophil-mediated endothelial cell injury. Combined incubation of PMN with cG250-TNF and IFNγ-induced persistent generation of H2O2 to a significant higher extend, when compared with single TNF activity and H2O2 release induced by single IFNγ treatment was negligible (Fig. 2b). Addition of IFNγ to HUVEC endothelial cells incubated with TNF or cG250-TNF also enhanced up-regulation of ICAM-1 expression (Fig. 2c, Lanes 7 and 8). Staining of ICAM-1 was more intensive, after stimulation of HUVEC cells using trimeric TNF (Lanes 1, 3, 5, 7) when compared with cG250-fused dimeric TNF (Lanes 2, 4, 6, 8). This effect was evident following treatment with single substances and using the combination regimen. Different levels of TNF-R1-mediated up-regulation of key regulators in response to receptor triggering by dimeric or trimeric TNF confirm our results obtained from the WEHI assay system. In addition to the activation loop between PMN and endothelial cells, we investigated the effects of the combined regimen on endothelial expression of IP-10 and tissue factor. IP-10 contributes to IFNγ-dependent tumor angiostasis promoting tumor rejection by inhibition of neo-vascularization, while endothelial tissue factor represents the key event for the activation of the extrinsic coagulation cascade initiating thrombotic infarction of established tumor vessels. Treatment with cG250-TNF or with rhuTNF in combination with IFNγ resulted in synergistic upregulation of IP-10 (Fig. 2c) and tissue factor (Fig. 2d). In addition to this procoagulant activity, endothelial cells undergo striking morphological changes37 together with changes in vascular permeability following exposure to TNF and IFNγ.27 Combined treatment of HUVEC cells with cG250-TNF and IFNγ changed cell morphology from a confluent cobblestone mono-layer to a spindle endothelial cell morphology with interrupted confluence (data not shown). In summary, the combination regimen of dimeric TNF fused to cG250 and IFNγ induced important cellular response programs displaying antitumor activities in a synergistic manner.
Biodistribution of cG250-TNF in vivo
To investigate the in-vivo behavior of cG250-TNF, biodistribution experiments were performed in BALB/c nu/nu mice grafted with NU-12 solid renal cell carcinomas as described.38 On the basis of our experience with cG250 IgG, each animal received 5 μg of radiolabeled antibody. Because the half-life of IgG is mainly determined by the CH2/CH3 constant regions, exchange of these by TNF was expected to lead to a dramatic alteration of the clearance rate. Mice were sacrificed within 24 hr postinjection, tissues were harvested and uptake of 125I-cG250-TNF was measured. cG250-TNF tissue levels were no greater than those in blood, with a rapid decrease of percent of injected dose per gram tissue (%ID/g) and similar tissue-to-blood ratios over time (Fig. 3a). In contrast, tumor-to-blood ratios increased over time, demonstrating specific accumulation and retention of both cG250-TNF (Fig. 3b) and the parental cG250 antibody (Fig. 3c) at the tumor site. To determine whether similar tumor-to-blood ratios could be reached for cG250-TNF and cG250 antibody, dual labeling experiments were performed with 131I-labeled cG250 IgG and 125I-cG250-TNF. Indeed, tumor-to-blood ratios of the parental cG250 IgG and the cG250-TNF fusion protein were similar (Fig. 4a), though %ID/g for cG250-TNF was much lower (Fig. 4b). These effects are due to the shorter half-life and the accelerated clearance of cG250-TNF leading to a short time period where cG250-TNF can accumulate at the tumor environment. In view of the rapid blood clearance, we performed a dose-escalation study, because higher cG250-TNF doses might result in higher tumor uptake levels. Cohorts of animals received 5, 25 or 50 μg radioiodinated cG250-TNF. Tissue-to-blood ratios remained stable at all dose levels and for almost all tissues, with exception of stomach uptake, attributed to free iodine in gastric mucosa (Fig. 5a). In contrast, relative as well as absolute amounts of cG250-TNF more than doubled in the tumors with increasing protein doses, reaching absolute levels comparable with the parental cG250 IgG (Fig. 3c). To ensure G250-antigen specific targeting, we performed a complementary biodistribution study in BALB/c nu/nu mice simultaneously grafted with solid G250-negative renal cell carcinoma. Minimal but not significant uptake of cG250-TNF was seen in antigen-negative tumors. Biodistribution in organs did not differ between animals carrying G250-positive or G250-negative tumors. Most notably, tumor-to-blood ratios were significantly lower for G250-negative than for antigen-positive tumors, demonstrating G250-antigen specific accumulation (Fig. 5c).
Efficacy of combined immunotherapy in vivo
To compare the bioactivity of cG250-TNF with rhTNF at molar levels, we performed experiments at TNF-equivalent doses (30 μg of recombinant human TNF vs. 100 μg of cG250-TNF). Initial studies revealed that the toxicity profile of cG250-TNF were analogous to our previously defined LD50 of dimeric-TNF-fusion proteins such as anti-FAP-TNF for BALB/c nu/nu mice.33 Tumor-bearing mice treated with 30 μg of rhTNF all died within 24 hr and could not be challenged with a second injection. Mice treated with 100 μg of cG250-TNF showed some signs of stress but all completed treatment without any lethal side effects. Further dose escalation of cG250-TNF led to significant weight loss and alteration of behavior when applied in combination with IFNγ without improvement of therapeutic efficacy (data not shown). Animals were simultaneously xenografted with solid growing G250-positive (SK-RC-52) and G250-negative (SK-RC-17) renal cell carcinomas on the opposite flanks and therapeutic agents were administered three times per week. Repeated i.v. injections of 100 μg of cG250-TNF resulted in a significant growth retardation of G250-expressing RCC when compared with treatment with parental cG250 IgG at the 100 μg dose or to treatment with 100 μg of control construct anti-FAP-TNF (p = 0.0001). Tumor size was controlled and remained stable during antigen-specific treatment without achieving complete remissions and tumor regrowth was only measurable after termination of cG250-TNF administration. In accordance to G250 antigen-restricted enrichment of radiolabeled cG250-TNF (Fig. 5c), the therapeutic activity of the unlabeled construct was also dependent on the presence of the targeted antigen as no therapeutic effect was seen on G250-negative tumors (Fig. 6a). A control fusion protein (anti-FAP-TNF) which does not recognize a relevant antigen in this model as well as the parental cG250 IgG failed to induce significant antitumor responses (Fig. 6a). Combined application of cG250-TNF i.v. and low-dose IFNγ s.c. was significantly superior to monotherapy (p = 0.01) as additional down-sizing of established xenografts could be induced. The groups receiving cG250-TNF or cG250-TNF in combination with IFNγ showed only progressive disease after termination of the treatment (Fig. 6b). Treatment at these dose levels could be completed without severe side effects. This clearly demonstrates that low dose IFNγ has a synergistic effect on cG250-TNF therapy in vivo and is well tolerated in combination with therapeutic cG250-TNF concentrations.
Human clear cell carcinoma is the most common and aggressive renal tumor type. The recognition of hypoxia-inducible factor (HIF) related pathways involved in clear-cell RCC was the major breakthrough for the development of novel targeted therapies.4 Activation of the HIF cascade leads to up-regulation of membrane bound G250/CA IX in more than 94% of clear-cell RCC.39 Positron emission tomography (PET) imaging of patients with renal masses using 124I-cG250 could identify aggressive clear-cell RCC with 100% specificity and positive predictive accuracy.10 According to previous clinical trials using iodinated cG250, even metastatic spread of the disease could be visualized.8, 10 These prospective trials support the potency of cG250 for tumor detection and the preclinical study presented here establishes cG250 for the targeted delivery of TNF to xenotransplanted renal cell carcinoma. The cG250-based TNF fusion protein was designed in analogy to a previously generated IgG-TNF construct allowing for efficient systemic TNF administration.13, 33 Intensity of TNF-R1 mediated signaling in vitro using the cytotoxicity assay system was comparable with the well-known range of the formerly characterized dimeric TNF-construct.13 Overall, cG250-TNF displays bioactivity in tumoricidal effector systems in vitro like induction of respiratory burst and triggering of endothelial cell injury. These effects, which play an important role in the antitumor efficacy of TNF in vivo, are further up-regulated by IFNγ leading to synergistic rather than additive stimulation of key regulators responsible for antitumor effectiveness in vivo. Biodistribution studies in xenografted nude mice demonstrated the tumor-vasculature specific uptake of cG250-TNF and its enrichment over time exclusively at the tumor site. The labeling did not affect the immuno-reactivity of the protein, i.e., the pharmacokinetic behavior as measured by counting the radioactivity in harvested organs was a representation of true cG250-TNF targeting. Of note, part of the purified labeled material consisted of aggregates and thus, some uptake of aggregated material in liver and spleen was anticipated. The radioactivity measured in stomach was likely due to free iodide accumulation in this tissue.40 We conclude that the shorter clearance time of cG250-TNF in comparison to the full-size IgG-format together with its lower TNF activity because of the dimeric TNF-structure led to the favorable low-toxicity profile. These properties allowed for the present dose dense application regimen and targeting TNF as a truncated IgG-construct induced significant antitumor responses in the xenograft RCC nude mouse model. Moreover, therapeutic potency of cG250-TNF could circumvent previously reported limitations of TNF-based immunotherapy in nude mice: Preclinical and clinical data suggest production of endogenous IFNγ as a central triggering antitumor mechanism in response to treatment with TNF.41, 42 As activated T cells seem to play the major role for triggering synergistic endogenous IFNγ release, nontargeted TNF alone, or even in combination with chemotherapy failed to induce antitumor effects in athymic nude mice bearing human xenografts.43 Similarly, no antitumor effect was observed for the combination of targeted trimeric TNF at lower doses and an alkylating agent. Only the additional application of exogenous IFNγ improved the response rate to targeted TNF in these trials.44 In this study, however, the dose dense administration schedule of cG250-TNF resulted in significant size reduction of xenotransplanted renal cell carcinomas and led to disease control during treatment. The combination of IFNγ and cG250-TNF could even improve treatment efficacy further resulting in a significant advantage of the combination regimen when compared to cG250-TNF monotherapy. Other trials reported up-regulation and enhanced shedding of TNF-receptors in response to IFNγ or TNF treatment.45, 46 Thus, the release of TNF receptors represents a counter-regulatory mechanism potentially contributing to a decrease in therapeutic TNF activity.47 Antitumor efficacy of single agent treatment with cG250-TNF was not limited by such undesirable effects. Furthermore, addition of IFNγ at clinically well-tolerated doses used in our trial was already sufficient to further improve the good results obtained from the single agent treatment and alteration of soluble TNF-R1 levels is almost negligible in response to such low doses of IFNγ.45 Furthermore, the TNF part of cG250-TNF consists of two human TNF subunits thereby circumventing murine TNF-R2 binding with concomitant inhibition.48 In summary, administration of cG250-TNF alone or in combination with IFNγ displayed impressive therapeutic efficacy. Initial reduction in tumor size and therapeutic control over established renal cell carcinomas was observed throughout continued administration of cG250-TNF. The phenomenon that withdrawal of targeted therapy provoked tumor regrowth is a common feature also observed in other therapeutic tumor vascular targeting settings.49 The schedule of treatment and dosage of combined drugs is very critical for biological effects of applied cytokines.50 The large therapeutic window and its antitumor efficacy qualify cG250-TNF for further trials with potent co-effectors. Moreover, the very limited progression-free survival of patients suffering from metastatic disease promotes treatment strategies basing on cG250-TNF.51 The encouraging targeting and antitumor properties of cG250-TNF and the strong and stable expression pattern of CAIX/G250 in sporadic and inherited forms of RCC warrant clinical evaluation of this construct.
We thank Ms. Natalie Fadle and Ms. Barbara Williamson for excellent technical assistance. We thank the Department of Experimental Surgery at the University of the Saarland, 66421 Homburg, Germany and the Central Animal Facility, University Medical Center Nijmegen, The Netherlands for housing and handling of animals.