Tumor necrosis factor (TNF)-α is a potent anticancer agent that seems to selectively target tumor-associated vasculature resulting in hemorrhagic necrosis of tumors without injury to surrounding tissues. The major limitation in the clinical use of TNF has been severe dose-limiting toxicity when administered systemically. However, when administered in isolated organ perfusion it results in regression of advanced bulky tumors. A better understanding of the mechanisms of TNF-induced antitumor effects may provide valuable information into how its clinical use in cancer treatment may be expanded. We describe here that the release of a novel tumor-derived cytokine endothelial-monocyte-activating polypeptide II (EMAPII) renders the tumor-associated vasculature sensitive to TNF. EMAPII has the unique ability to induce tissue factor production by tumor vascular endothelial cells that initiates thrombogenic cascades, which may play a role in determining tumor sensitivity to TNF. We demonstrate here that constituitive overexpression of EMAPII in a TNF-resistant human melanoma line by retroviral-mediated transfer of EMAPII cDNA renders the tumor sensitive to the effects of systemic TNF in vivo, but not in vitro. This interaction between tumors and their associated neovasculature provides an explanation for the focal effects of TNF on tumors and possibly for the variable sensitivity of tumors to bioactive agents.
TNF-α2induces procoagulant effects within tumor neovasculature, resulting in endothelial fibrin deposition, localized thrombosis, and ischemic necrosis of responsive tumors. The mechanism responsible for this effect as well as the apparent variable in vivo sensitivity of tumors to TNF is not well understood. O’Malley et al.(1) first demonstrated that the serum of animals treated with LPS contained an endogenous factor that could induce hemorrhagic necrosis of tumors in animals not exposed to LPS. Carswell et al. (2) later isolated a circulating protein from mice, pretreated with Bacillus Calmette-Guerin and challenged with LPS, that could induce significant hemorrhagic necrosis of methylcholanthrene A-induced fibrosarcomas (Meth A) that was termed “TNF.” Systemic administration of TNF to mice bearing s.c. MethA tumors resulted in marked hemorrhagic necrosis of the tumors without observable effects on nontumor tissue. This observation formed the basis for several clinical trials using TNF once the recombinant protein was made available.
Systemic TNF was used in multiple clinical trials against a variety of different tumor histologies (3). However, the results were disappointing because TNF resulted in significant systemic toxicity and no significant antitumor effects at the maximally tolerated doses. The clinical use of TNF was largely abandoned until Lienard et al. (4) reported their initial results of isolated limb perfusion as a means of delivering high concentrations to the extremity in patients with in transit melanoma or unresectable sarcoma, while minimizing systemic exposure. We and others have used isolated organ perfusion of the limb or liver using TNF plus chemotherapeutic agents to treat unresectable tumors with dramatic responses in the majority of patients (5, 6, 7, 8, 9). Despite these encouraging results, much remains unknown regarding the mechanism by which TNF exerts its effects on tumors. A better understanding of this mechanism may improve the therapeutic efficacy of TNF and help minimize systemic toxicity which limits wider application of this potent agent.
There are several postulated theories to explain the mechanism of action of TNF. These include the stimulation of T cell-mediated responses resulting in the generation of CD8+tumor-specific CTLs (10, 11), TNF-induced apoptosis (12, 13, 14, 15, 16),macrophage/granulocyte-mediated injury (17), activation of cellular adhesion molecules (18), and induction of fibrin deposition on endothelial surfaces and thrombus formation (19, 20, 21). Most of the evidence, however, supports an indirect mechanism via the tumor vasculature rather than direct cytotoxic effects of TNF. The clinical experience with TNF administered via isolated organ perfusion in patients with melanomas also supports this hypothesis (22). Renard et al. (18) demonstrated in limb perfusions using TNF that coagulative and/or hemorrhagic necrosis of tumors was specific for TNF because isolated limb perfusion with melphalan alone failed to show this type of necrosis. Fig. 1 demonstrates the complete obliteration of tumor neovasculature after a hyperthermic isolated limb perfusion with TNF and melphalan, while leaving the normal host vessels apparently unaffected.
The distinction between tumor and host vasculature centers on the phenotypic difference of the vascular endothelium. In general, tumor vessel endothelium is more prone to thrombosis and capillary leakage(20, 23). Cytokines present in the microenvironment of solid tumors may alter the endothelium of tumor vessels influencing characteristics of tumor growth and metastatic spread (24, 25, 26, 27). Specifically,tumor-derived cytokines may play fundamental roles in determining tumor phenotypes relating to tumor angiogenesis, tumor progression,metastatic potential, and sensitivity to therapeutic agents.
EMAPII is a cytokine that may influence interactions between tumor cells and associated neovasculature. EMAPII was first described by Kao et al. (28, 29), using the MethA tumor as a model for studying host-tumor responses and the effects of tumor-derived factors. The protein was isolated from MethA supernatant by its ability to activate endothelial cells inducing tissue factor procoagulant activity and up-regulating leukocyte adhesion molecules P-selectin and E-selectin. EMAPII is a polypeptide synthesized as a Mr 34,000 precursor and cleaved to produce an active Mr 22,000 mature protein. Because EMAPII was isolated from a TNF-sensitive tumor and induced tissue factor production by endothelial cells, it seemed to be a putative TNF-sensitizing agent. Marvin et al. (30) treated mice bearing TNF-resistant tumors with intratumor injection of recombinant EMAPII and rendered the tumors sensitive to systemic TNF. The effect was confined to the tumor vasculature resulting in thrombohemorrhage and tumor regression.
The identification of tumor cytokines, such as EMAPII, that specifically activate endothelial cells suggests a novel mechanism of TNF sensitivity. If tumors express variable levels of EMAPII,overexpression of EMAPII in certain tumors may predispose the tumor vasculature to the procoagulant effects of TNF and increase their sensitivity to this agent when administered via systemic or regional routes. To determine whether EMAPII production by tumors, in fact,confers TNF sensitivity, we evaluated EMAPII levels and TNF sensitivity in various human melanoma lines. In addition, we transduced a TNF-resistant and low EMAPII-expressing human melanoma line with a retroviral vector encoding the EMAPII cytokine, measured EMAPII overexpression by immunodetection and functional assays, and characterized the in vitro and in vivosensitivity of the tumor to TNF.
MATERIALS AND METHODS
Tumor Cell Lines and Animals.
Pmel, 883, Smel, and 1286 are primary human melanoma lines derived from patients treated at the National Cancer Institute and cultured in DMEM supplemented with 10% FCS, 2 mm l-glutamine,and 1% penicillin/streptomycin (Biofluids, Rockville, MD) at 37°C in a 5% CO2 incubator. Each of the melanoma lines was passaged no more than 10 generations and cryopreserved at regular intervals. Female athymic nude mice were obtained from the NIH small animal facility and housed maximum five animals/cage in a barrier care room. Human tumors were inoculated in the hind limb-flank region by s.c. injection of 1 × 106 tumor cells suspended in 100 μl of sterile HBSS. All animal protocols were approved by the Animal Care and Use Committee and conducted with strict compliance to guidelines established by the NIH Animal Research Advisory Committee.
Detection of EMAPII mRNA by Northern Blot.
Total RNA was extracted from cultured tumor lines using Rneasy Total RNA Kits (Qiagen, Chatsworth, CA), isolated on 1% agarose/5%formaldehyde gels, and hybridized using 32P-labeled cDNA probes for EMAPII and a β-actin loading standard. EMAPII transcript was quantified using a STORM phosphoimager and ImageQuaNT software analysis package (Molecular Dynamics, Sunnyvale, CA).
Generation of the Retroviral pWU-EII Vector Construct.
A cDNA clone of human pro-EMAPII was generated using primers derived from the GenBank sequence and RNA isolated from human monocytes using reverse transcription-PCR amplification with Pfu polymerase(Perkin-Elmer Corp., Norwalk, CT). The PCR-derived fragment was cloned into PCR2.1 using the TA Cloning Kit (Invitrogen, Carlsbad, CA) and confirmed by cycle sequencing using dye-labeled terminators(Perkin-Elmer Corp.). The hEMAPII cDNA was ligated into the multiple cloning site of the retroviral vector pSAMEN under the expression of an LTR promoter/enhancer with neomycin-resistance gene selection. Presence and directionality of the EMAPII insert in the pWU-EII construct were confirmed by PCR amplification and restriction digest.
Stable Transduction of EMAPII Into a Low-expressing Melanoma Line.
The pWU-EII retroviral vector and pSAMEN vector control were each transfected into the amphotropic packaging cell line PA317 by calcium phosphate precipitation (Invitrogen). The viral supernatants from the PA317 producer cells were used to infect the Gibbon ape ecotropic packaging cell line PG13 with polybrene at a concentration of 8μg/ml. PG13 viral supernatants were then used to infect the human wild-type Pmel tumor line, which expresses the lowest levels of EMAPII. Transduced Pmel tumor lines were expanded under neomycin selection at 800 μg/ml and cloned in limiting dilution.
Rapid Genomic DNA Extraction and PCR Amplification.
Tumor cells (1 × 106) were placed into 200 μl of DNA extraction buffer containing 0.5% Tween 20(Bio-Rad, Hercules, CA), 100 μg/ml Proteinase K (Stratagene, La Jolla, CA), and 1 × PCR Buffer (Perkin-Elmer Corp.) and heated at 56°C for 45 min, followed by 95°C for 10 min. Genomic DNA samples extracted from the wild-type and transduced tumor clones were screened by PCR amplification with primers specific for both the EMAPII insert and downstream retroviral IRES region using the following primers: IRES forward, 5′-AACGTTACTGGCCGAAGCC-3′; IRES reverse,5′-AAGGAAAACCACGTCCCCGT-3′; EII forward, 5′-AACTGAAACAAGAGCTAATT-3′;EII reverse, 5′-CAGGCTCTCCTGGGAAAGCA-3′.
PCR amplification was performed using a GeneAmp thermocycler(Perkin-Elmer Corp.) for 25 cycles consisting of a 15-s 94°C denature step, a 30-s 55°C annealing step, and 2-min 72°C extension step.
Preparation of Polyclonal EMAPII Antiserum.
Polyclonal EMAPII rabbit antiserum (kindly provided by Dr. D. Stern,Columbia University, New York, NY) was purified using the ImmunoPure IgG Purification Kit (Pierce Chemical Co., Rockford, IL) by elution from a bound Protein A column. IgG-purified polyclonal EMAPII antiserum was used directly for immunostaining and biotinylated using an EZ-Link Sulfo-NHS-LC-Biotinylation Kit (Pierce Chemical Co.) for ELISA.
Cells were plated overnight on coverslips and washed three times with PBS, fixed in 10% formalin for 20 min, 1% Triton X-100 for 5 min,washed again two times with PBS, blocked with 5% FCS/PBS for 10 min,and incubated for 45 min at 37°C with a 1:100 dilution of IgG-purified polyclonal EMAPII rabbit antisera or 1 μg/ml rabbit IgG(Sigma Chemical Co.) to define background staining. After another double wash in PBS, the cells were incubated for 45 min at 37°C with a 1:50 dilution of fluorescein-conjugated goat antirabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA), followed by another double wash in PBS and mounted with DAPI/Antifade (Oncor, Gaithersburg,MD), and visualized under fluorescent microscopy. Tumors excised during necropsy were fixed in 10% formalin, and EMAPII immunohistochemistry was performed using the same antibodies by Paragon Biotech (Baltimore,MD).
An indirect ELISA was used to detect EMAPII protein in cell lysates prepared from the cultured tumor lines. Briefly, cell lysates standardized to total protein as determined by BCA Protein Assay(Pierce Chemical Co.) were diluted in carbonate buffer (Pierce Chemical Co.) and coated onto MaxiSorp immunoplates (Nunc, Inc.) overnight at 4°C. Recombinant human EMAPII (kindly provided by Dr. D. Stern) was used as a serially diluted control standard. The plates, coated with control standards and lysate samples, were blocked for 1 h at 24°C with SuperBlock Blocking Buffer (Pierce Chemical Co.), incubated for 1 h at 4°C with 10 μg/ml biotinylated polyclonal rabbitα-EMAPII, followed by another 1-h incubation at 24°C with 1 μg/ml Neutravidin-horseradish peroxidase conjugate (Pierce Chemical Co.). The plates were then developed using Turbo TMB ELISA substrate (Pierce Chemical Co.), stopped after 30 min by the addition of 1N H2SO4, and absorbance was measured at a wavelength of 450 nm. Each incubation was preceded by a triplicate wash in 0.1% PBS, 0.04% BSA, and Tween 20.
Preparation of Tumor-conditioned Media and Neutralizing Antibody.
To produce tumor-conditioned media, 5 × 106tumor cells were plated/75 cm2 tissue culture flask in 15 ml of Medium 199 (NIH Media Services) with 1% penicillin/streptomycin and incubated 24 h at 37°C. The conditioned media was collected and filtered across a 0.45-um low protein-binding membrane (Corning Costar, Cambridge, MA) to remove any cellular debris. Tumor-conditioned media was prepared for immediate use in tissue factor induction assays avoiding freeze-thaw cycles. The polyclonal α-EMAPII rabbit IgG was used as neutralizing antibody at a concentration of 200 μg/ml and incubated with tumor-conditioned media for 30 min at 37°C immediately before tissue factor induction experiments.
Tissue Factor Induction Assay.
HUVECs (Clonetics, San Diego, CA) were passaged in complete EGM-2 media(Clonetics) for no more than four generations and plated at a concentration of 5 × 105 cells/well in 6-well multiwell plates (Corning Costar) and incubated 48 h at 37°C in a 5% CO2 chamber reaching ∼80% confluency. The cells were washed twice with sterile PBS and treated with 1 ml of tumor-conditioned media or 150 ng/ml recombinant IL-1β (R&D Systems,Minneapolis, MN) as a positive control for endothelial cell activation. The treated endothelial cells were cultured at 37°C for 16 h,washed twice with sterile PBS, and harvested in 300 μl of PBS/well using mechanical cell scrapers and stored at −70°C. Each cell suspension was quickly thawed at 37°C and centrifuged at 2000 × g for 5 min, and the resultant cell pellet was resuspended in 200 μl of 50 nm Tris, 100 nm NaCl, and 0.1% BSA. The one-stage procoagulant assay was performed by adding 100 μl of Factor VIII-deficient human plasma(George-King Biomedical, Overland Park, KA) to 100 μl of endothelial cell suspension and incubated for 3 min at 37°C. The coagulation reaction was catalyzed by the addition of 100 μl of 30 nmCaCl2, and clotting time was measured using a clinical fibrometer (Baxtor, Deerfield, IL). Standard curves were generated using recombinant human tissue-factor (American Diagnostica, Greenwich,CT), and assay sensitivity was <10 pg/ml.
In Vitro MTT Cytotoxicity Assay.
Sensitivity of the tumor lines Pmel, 1286, Pmel-SAMEN, and Pmel-EII transduced clones to TNF was assessed in the following manner. Cells were plated in flat-bottomed 96-well plates at a concentration of 3.0 × 103 cells/well in 100 μl of DMEM supplemented with 10% FCS and allowed to grow for 48 h before treatment. Recombinant TNF (Knoll Pharmaceuticals, Whippany, NJ) was reconstituted in media to 20 μg/ml, and serial dilutions were performed to desired treatment concentrations. The TNF was added to each well in 100-μl aliquots in six replicates, plates were incubated at 37°C for 16 h, and MTT (Sigma Chemical Co.) cytotoxicity assays were performed. Briefly, 100 μl of 2 mg/ml MTT was added to each sample well and incubated at 37°C for 4 h. The media was then aspirated, the formazan precipitate was solubilized in 120 μl of DMSO (dimethylsulfoxid; Fluka Chemika), and absorbance at 570 nm was measured. Cytotoxicity, expressed as percentage control survival, was determined by dividing treatment absorbance values by the mean of control values for each experiment and expressed as percentage values.
Treatment of Human Melanomas Established in Nude Mice.
Athymic nude mice were implanted with human melanoma tumors as described above. Tumor volumes were determined from caliper measurements of width, length, and height based on calculated partial spherical volume (V), V = πh(h2 + 3a2)/6, where h = tumor height and a = (length + width)/2. When tumors had reached ∼150 mm3, typically occurring within 3–4 weeks,mice were randomized into two groups. One group received systemic recombinant TNF (Knoll Pharmaceuticals) administered 6 μg/mouse via lateral tail vein injection. The other group received tail vein injections of the vehicle alone (0.9% NaCl solution and 0.5% BSA). Cytokine treatment and tumor measurements were performed in a double-blinded fashion. Each animal was calculated as percentage baseline volume before treatment. Animals were sacrificed and necropsied according to NIH animal care guidelines.
Tumor volumes in the in vivo experiments were compared using ANOVA between groups. Ps given were determined using the Tukey-Kramer Test for multiple comparisons.
EMAPII Expression in National Cancer Institute Human Melanoma Lines.
Total RNA was extracted from each of the four human melanoma lines,separated by Northern blot, and hybridized with cDNA probes for EMAPII and β-actin as shown in Fig. 2. Quantitative analysis of relative EMAPII mRNA identified the 1286 melanoma line with the highest level of EMAPII expression, an average 2.2-fold higher than the lowest expressing Pmel melanoma line. Both lines were analyzed for EMAPII protein content by ELISA, whereas functional EMAPII activity in tumor-conditioned media was quantitated by tissue factor induction from HUVECs. The 1286 melanoma had 2.5-fold higher EMAPII protein expression compared with Pmel melanoma by ELISA of cell lysates standardized to total protein (data not shown). Functional assay of EMAPII protein showed a 5.8-fold higher induction of endothelial cell tissue factor production by 1286 melanoma-conditioned media compared with Pmel-conditioned media (Fig. 3). The reliability and specificity of the tissue factor induction assay to detect functional EMAPII protein was confirmed by demonstrating that neutralizing EMAPII antibody completely abrogated endothelial cell activation by 1286 melanoma-conditioned media.
EMAPII Expression Is Related to in Vivo TNF Sensitivity.
To determine whether the level of EMAPII expression in human melanoma lines correlated with in vivo sensitivity to systemic recombinant TNF, Pmel and 1286 melanoma lines were s.c. inoculated into the hind limb-flank regions of athymic nude mice and grown to uniform size. When the tumors reached ∼150 mm3, animals were treated with either 6 μg of i.v. recombinant TNF or NaCl. Blinded tumor measurements were analyzed at regular intervals after treatment to calculate tumor volume response after exposure to a single dose of TNF. Fig. 4, a and b, demonstrates the in vivosensitivity of the wild-type tumors Pmel and 1286 to systemic TNF. Treatment of the 1286 tumor-bearing mice with systemic TNF consistently demonstrated significant tumor regression compared with the steady tumor progression observed in controls. Tumor necrosis occurred in the TNF-treated 1286 melanoma-bearing animals within 48 h of TNF administration evidenced by reduction in tumor volumes and eschar formation. When similar sized Pmel tumors were treated with TNF or NaCl, no difference was seen. The lack of significant tumor response and no observable evidence of ischemic necrosis among the TNF-treated tumor-bearing animals demonstrates the overall resistance of Pmel melanoma to systemic TNF. Thus, the level of EMAPII expression in these human melanoma lines was associated with in vivo TNF sensitivity of the cell lines. The level of EMAPII expression in three other primary melanoma lines was also found to correlate with relative in vivo sensitivity to systemic TNF (data not shown).
Standard MTT cytotoxicity assays on 1286 and Pmel tumor lines showed that neither melanoma line was sensitive to TNF in vitro(Fig. 4 c). The contrast of TNF exerting no direct cytotoxic effect on tumor cells to the observed antitumor effects when administered in vivo, strongly suggested an indirect effect of TNF, presumably on tumor neovasculature. The characterization of Pmel melanoma as a relatively low EMAPII-expressing tumor line and TNF-resistant tumor identified it as a putative target for constituitive retroviral-mediated overexpression of EMAPII.
Constituitive Overexpression of EMAPII in Pmel Melanoma Using a Retroviral Vector.
To test whether the lack of TNF sensitivity of the Pmel tumor might be due to the lower levels of EMAPII expression, we hypothesized that the retroviral-mediated transduction of the EMAPII cDNA under the transcriptional control of an LTR promoter/enhancer element into the Pmel tumor wild-type genome might result in the overexpression of EMAPII, which would lead to a phenotypic change from a previously TNF-resistant tumor into a TNF-sensitive tumor. The sequence for EMAPII was cloned into the pSAMEN retroviral vector under the transcriptional control of the LTR promoter/enhancer element derived from the Moloney murine leukemia virus (Fig. 5) and used to establish an amphotropic packaging cell line in PA317, as described previously (31, 32). To increase the efficiency of transducing the Pmel human melanoma line, the viral supernatants harvested from the pWU-EII bulk-transfected PA317 producer cells were used to infect the packaging cell line PG13 expressing the Gibbon ape viral envelope. Stable transformants were screened by successive passages under neomycin selection and were cloned in limiting dilution.
Pmel-EMAPII-transduced clones were screened for stable retroviral integration by PCR amplification using a forward primer for EMAPII and a reverse primer for the downstream IRES sequence. This generated a 950-bp fragment from genomic DNA of tumor clones transduced by retroviral infection (data not shown). We screened EMAPII overexpression in the transduced clones in vitro by immunofluorescence. Two overexpressing clones, Pmel-EII.08 and Pmel-EII.09 compared with wild-type and vector-transduced controls are shown in Fig. 6,a. Additionally, Pmel-EII tumor clones were analyzed by tissue factor induction assay to confirm that the overexpression and processing of the EMAPII protein resulted in functional secreted cytokine. The tumor clones Pmel-EII.08 and Pmel-EII.09 showed a 7.0-and 19.9-fold increase, respectively, in tissue factor induction compared with wild-type Pmel and vector-transduced Pmel (Fig. 6 b), confirming constituitive overexpression of EMAPII in the retroviral-transduced tumor clones.
After confirming functional EMAPII overexpression in the transduced clones Pmel-EII.08 and Pmel-EII.09, both clones were assayed for in vitro TNF cytotoxicity by MTT assay. As seen in both wild-type Pmel and 1286 melanoma, neither tumor clone was sensitive to TNF in vitro (Fig. 6 c). Thus, overexpression of EMAPII in wild-type Pmel melanoma did not change the observation that TNF has no direct cytotoxic effect on these tumor cells.
In Vivo Expression of EMAPII-transduced Tumor Clones.
To confirm that the EMAPII cDNA is being expressed in vivo,it is necessary to screen EMAPII-transduced clones for retroviral-driven expression by growing each tumor clone to 150 mm3 and assessing EMAPII expression. Transduced tumor clone screening in vivo not only determines that high levels of functionally active EMAPII cytokine are expressed, but also establishes that EMAPII overexpression can be sustained through the time intervals required for tumor growth and systemic TNF treatment. From the bulk-transduced Pmel-EMAPII tumors, clones Pmel-EII.08 and Pmel-EII.09 demonstrated constituitive EMAPII overexpression in vivo after 4 weeks (Fig. 6 d), providing sufficient time for tumor growth and treatment.
EMAPII Overexpression Results in a TNF-sensitive Phenotype.
To determine whether constituitive EMAPII overexpression would change the TNF-resistant phenotype of wild-type Pmel melanoma, clones Pmel-EII.08 and Pmel-EII.09 were grown in athymic nude mice to approximately 150 mm3. Tumor-bearing animals were then treated with systemic TNF or NaCl solution injection. Fig. 7 demonstrates that both tumor clones Pmel-EII.08 and Pmel-EII.09 are rendered sensitive to the effects of TNF in vivo; the response curves resemble those of the TNF-sensitive 1286 melanoma. Both wild-type Pmel and Pmel transfected with the pSAMEN vector alone showed no response to systemic TNF. The constituitive overexpression of the cytokine EMAPII resulted in the in vivo conversion of wild-type Pmel tumor from a TNF-resistant tumor into a TNF-sensitive tumor.
The mechanism of TNF-induced tumor necrosis is not well understood. Clinical observations made from therapeutic applications of TNF have consistently yielded evidence for a tumor vascular endothelial cell-mediated response. The isolation of a tumor-derived factor that specifically activates endothelial cells by up-regulation of tissue factor expression provides a possible explanation for this observation. Contrino et al. (33) reported that functional tissue factor is expressed by the vascular endothelial cells of malignant infiltrating intraductal breast cancer, but not in benign fibrocystic breast disease, suggesting an effect of tumor cells on the associated vascular endothelium. Our studies demonstrate a similar relationship in primary human melanoma lines. The elaboration of tissue factor by tumor-associated endothelium in response to mediators, such as EMAPII,generated by tumor cells may play a fundamental role in influencing the activation of tumor neovasculature.
Since the initial observation that mice bearing MethA fibrosarcoma exposed to systemic TNF develop ischemic necrosis of the tumor,application to a variety of tumor types has shown that TNF sensitivity varies markedly among different tumors. Even among primary human melanoma lines, variable in vivo TNF sensitivity was observed. Indeed, there were no reliable predictors of in vivo TNF response and the mechanism of TNF sensitivity remained elusive. The identification of the tumor-derived cytokine EMAPII, with its ability to activate endothelial cells and induce coagulation,offers a plausible explanation for the variable TNF response seen in different tumors. The level of EMAPII expression in various human melanoma lines correlates with in vivo TNF sensitivity in nude mice.
EMAPII is expressed ubiquitously in eukaryotic cells. Tas et al. (34) have shown by RT-PCR that mRNA for EMAPII is detected in nearly all tumor cells, established cell lines, and primary cultures. We have also detected EMAPII mRNA from total RNA extracted from organ tissues in both tumor-bearing and normal mice (data not shown),confirming the ubiquitous transcription of EMAPII in eukaryotic cells. However, immunohistochemistry of organ tissues from a tumor-bearing mouse detects EMAPII protein within tumor and not in any other tissues(data not shown), supporting previous reports that EMAPII cytokine is produced only within tumor tissue (30). Furthermore,immunohistochemistry supports the variable level of EMAPII protein expression between different tumor types.
A direct causal effect of EMAPII in determining sensitivity to TNF is shown by our retroviral transduction studies. The constituitive overexpression of EMAPII in a TNF-resistant melanoma line (Pmel) by retroviral-mediated transfer of the EMAPII cDNA renders the wild-type tumor sensitive to the effects of systemic TNF. The demonstration of this effect in more than one transduced tumor clone, while not absolutely ruling out, does argue against the possibility of selecting a TNF-sensitive wild-type clone by chance from the tumor line.
Our results demonstrate that EMAPII production by tumors can influence their sensitivity to systemic TNF. The mechanism by which EMAPII renders tumor vasculature sensitive to the proinflammatory effects of TNF is not clearly understood. Previous work has demonstrated that recombinant EMAPII can up-regulate endothelial cell TNFR1 expression in a dose-dependent fashion (35). Because TNFR1 expression has been associated with the induction of endothelial cell apoptosis, it follows that EMAPII produced by tumors may determine in vivosensitivity to TNF by up-regulating TNFR1 expression on endothelial cells triggering cell death in the presence of TNF, leading to eventual ischemic necrosis of the tumor.
An alternative mechanism to explain EMAPII sensitization of tumor vascular endothelium to TNF would be an additive effect on endothelial cell tissue factor expression. TNF has been found to induce tissue factor expression on endothelial surfaces (36). In fact, the use of neutralizing TNF receptor antibodies inhibits endothelial tissue factor production induced by TNF (37). There may exist a critical threshold of tissue factor expression by tumor associated vasculature that provokes TNF-induced thrombohemorrhage and EMAPII may function by promoting this process of tumor vascular activation.
The essential role of EMAPII, a tumor-derived cytokine, in determining TNF sensitivity provides further insight into the interactions between tumor cells and tumor neovasculature. The procoagulant effect on the vascular endothelial cells induced by tissue factor production in response to high levels of EMAPII elaborated by tumor cells may play an important role in tumor neovascularization, primary tumor growth,metastatic potential, and sensitivity to therapeutic agents. Clinical studies are needed to determine whether EMAPII expression in tumors can predict clinical response to TNF therapy and potentially identify patients with cytokine-responsive tumors that may respond to lower TNF doses, thus, reducing dose-related toxicity. Furthermore, novel methods of delivery of bioactive agents such as EMAPII may result in promising therapeutic approaches against human malignancies.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: TNF, tumor necrosis factor; LPS, lipopolysaccharide; EMAPII,endothelial-monocyte-activating polypeptide II; LTR, long terminal repeat; MTT, 3-(4,5-dimethyl-2-thiazoyl)-2, 5-diphenyl-2H tetrazolium bromide; TNFR1, TNF p55 receptor; MethA, methylcholanthrene A-induced fibrosarcoma; HUVEC, human umbilical vein endothelial cell; MRA,magnetic resonance angiogram; IRES, internal ribosomal entry site.
We thank Drs. David Stern (New York, NY), Michael Nishimura(Bethesda, MD), and Steven Rosenberg (Bethesda, MD) for helpful comments and suggestions.