Attenuation of the Glucocorticoid Response during Ad5IL-12 Adenovirus Vector Treatment Enhances Natural Killer Cell–Mediated Killing of MHC Class I–Negative LNCaP Prostate Tumors

Natural killer (NK) cells can proliferate and be activated for killing by interleukin-12 (IL-12) (1, 2) and have been shown to be an important effector arm of the innate immune response in antitumor defenses (3). Recent studies have shown that NK cell activity is controlled through a balance between inhibitory and stimulatory receptor signals (4). One of the activating NK cell receptors, the NKG2D receptor, may have the capacity to specifically target tumor cells for killing through recognition of its ligands that include the MHC class I chain–related A and B molecules (MICA and MICB; ref. 5). These ligands are not typically displayed on the surface of normal cells, but can be up-regulated during conditions of cellular stress and consequently activate the NK cell cytolytic machinery. With regard to tumor specificity, MICA and MICB expression has been shown on several different types of carcinomas (including that of prostate cancer), but not from corresponding normal tissues, indicating that an NK cell–directed response can be targeted to such tumors (6). MICA/B expression on a subset of hepatocellular carcinomas has been shown to be the determining factor in the susceptibility of some of these tumors to NK cytolysis (7). The MICA and MICB promoters are regulated by heat shock promoter elements similar to that of heat shock protein 70, indicating that proinflammatory microenvironments may enhance NK cell killing by regulation of MICA and MICB expression on tumor cells (8).

IL-12 is a potent immunostimulatory cytokine that has promising use as an anticancer agent. In addition to its ability to activate NK cells, IL-12 has been shown to activate cytotoxic T cells, to promote Th1 immune responses, and to inhibit tumor angiogenesis (9–11). IL-12 administered as a recombinant protein has been tested in preclinical trials but can produce dosage-dependent toxicity in patients (12, 13). Containment of IL-12 production to tumor tissue is preferable for this potentially toxic cytokine, and IL-12 administered in adenovirus vector form showed significantly reduced toxicity in a phase I clinical trial (14, 15). It was toward this goal that we decided to test the recombinant adenovirus vector expressing IL-12 (Ad5IL-12) in this study for its effect on s.c. established tumors formed by the human prostate tumor LNCaP cell line in NOD.scid mice. The recombinant Ad5IL-12 vector tested in this study was previously shown to induce protection against leishmaniasis in susceptible BALB/c mice by enhancing Th1 response (16, 17). This vector was also shown to have potent antitumor effects causing complete regression of established breast cancers (18). The effective cure rate of the IL-12 vector could be increased when used in combination with an IL-2–expressing vector, supporting the contention that combinatory therapies will be more productive and lead to more consistent curative clinical outcomes (19). We were the first group to produce a recombinant adenovirus vector expressing a cytokine (20) and one of the first to test its use in the therapy of experimental cancers (21). Of note, preexisting immunity to adenovirus does not prevent intratumoral IL-12 expression by the Ad5IL-12 vector but does prevent vector dissemination (22) and could represent an additional beneficial feature that limits cytokine toxicity because the majority of the human population has had previous exposure to adenovirus. Given these considerations, it is likely that the application of adenovirus vectors for the therapy of cancer in humans will increase.

In this study, we show the effect of the Ad5IL-12 vector treatment in a combinatory therapy using mitotane to interrupt cortisol production in experimental animals by inducing pharmacologic adrenalectomy (23). The administration of mitotane is currently used in the treatment of adrenocortical cancers in humans to disrupt tumor cell growth (24). Here, we examined mitotane's effect in transiently reducing circulating cortisol levels to enhance the proinflammatory response generated by the Ad5IL-12 vector in tumor treatments. Mitotane is metabolically activated by an adrenal-specific cytochrome P450 that results in the net reduction of cortisol by the gland. Our results show that the Ad5IL-12 vector can be used to induce NK cells to mediate human tumor killing in vivo following localized administration of the vector. In addition, we show that chemical adrenalectomy with mitotane results in both enhanced NK cell cytolytic activity and tumor therapeutic efficacy upon Ad5IL-12 vector treatment.

Mice and cell lines. NOD.scid and NOD.Cg-Rag1^tm1MomPrf1^tm1Sdz (NOD.Cg) congenic mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and were bred in the animal facilities at Torrey Pines Institute for Molecular Studies. The human prostate cancer cell lines (LNCaP and PC3) were grown in DMEM, supplemented with 10% fetal bovine serum, 100 μg/mL streptomycin, and 100 IU/mL of penicillin. For tumor inoculation, exponentially growing cells were harvested by trypsinization, washed twice with PBS, and their viability determined by trypan blue exclusion before use in s.c. inoculations. All work was done according to institutional guidelines for animal use and care.

Animal models and vector treatment. Human prostate cancer cell lines LNCaP and PC3 were used to generate tumors in immunodeficient NOD.scid and NK cell–deficient NOD.Cg mice. Two million tumor cells in 50 μL of PBS were mixed with 50 μL of matrigel and injected s.c. in the hind flank of animals. Mice typically formed palpable tumors (∼100 mm³ volume) within 14 days and were then injected intratumorally using a 30-gauge needle with the appropriate concentration of Ad5IL-12 or control vector in a volume of 50 μL of PBS. Tumors were measured in two dimensions using calipers, and tumor volume was calculated assuming a prolate spheroid, as previously described (21).

Adenovirus vectors. The construction and characterization of the Ad5IL-12 vector and the DL70-3 control vector used in these experiments have been previously described (18). Briefly, the Ad5IL-12 vector contains the cDNA of the p35 subunit of IL-12 in the E1 region and the p40 subunit in the E3 region and uses an SV40 polyadenylation signal sequence. The DL70-3 vector is a control adenovirus vector deleted for the E1 sequences and has a partial deletion/insertion mutation within the E3 region. All viruses were propagated in 293 cells that contain sequences of the wild-type Ad5 genome that express the E1 proteins required for replication of E1-deleted vectors. Viruses for animal experimentation were purified on cesium chloride gradients as previously described (25).

Mitotane treatment and cortisol measurements. Mitotane (Sigma, St. Louis, MO) solution was prepared freshly on the day of injection by sonication for 20 min, as previously described (23). A single 225-μg/g dose of mitotane in 200 μL of peanut oil was given i.p. To control for the normal daily variation in the circadian rhythm of cortisol levels, blood samples were collected between 12 noon and 2:00 p.m. for each time point and sera stored at −20°C until ELISA analysis. Cortisol measurements were done using the Correlate-EIA cortisol antibody detection kit from Assay Designs, Inc. (Ann Arbor, MI). Samples were analyzed using an Emax microplate reader (Molecular Devices, Sunnyvale, CA) measuring absorbance (OD) at 405 nm.

NK cell cytolytic activity. NK cell killing activity was assayed using splenocytes isolated from Ad5IL-12 vector–treated and DL70-3 control-treated mice. Cytotoxicity was determined against YAC-1 cells in a standard chromium release assay. Target cells were incubated for 1 h at 37°C with ⁵¹Cr, washed thrice, and plated at a concentration of 1 × 10⁴ cells per well in 96-well U-bottomed plates. Effectors and targets were coincubated at various E/T ratios for 4 h at 37°C. Percent ⁵¹Cr release was calculated as follows: % specific lysis = 100 × [(specific release − spontaneous release)/(maximum release − spontaneous release)]. Each experiment was done in triplicate.

Collection, processing, and IL-12 measurement of tissues. At the designated time points, tumor tissue was surgically excised and frozen at −80°C. To produce tissue extracts, frozen tumor tissues were homogenized in PBS (3 mL per tumor) containing 100 mmol/L phenylmethylsufonyl fluoride and 10 μg/mL aprotinin. Homogenate was freeze-thawed thrice and cleared by centrifugation. Aliquots were stored at −20°C until analysis for cytokine content. Total IL-12 levels in tumor homogenates were measured by sandwich ELISA, by coating the Nunc-Immuno MaxiSorp F96 plates (Nunc, Roskilde, Denmark) with anti-IL-12 (p35/p70; clone Red-T/G297-289; BD Biosciences, San Diego, CA). ELISA was done as recommended by the manufacturer of the antibody, and values were determined for cytokine content by comparison against a recombinant murine IL-12 protein standard.

FACS analysis. Characterization of cell surface expression of MICA/B and MHC class I molecules on the LNCaP and PC3 cells lines was done with monoclonal antibodies (mAb) obtained from BD Biosciences using a FACScan flow cytometer (Becton Dickinson, Mountain View, CA), following previously described protocols (7). For in vitro treatment analysis, 1 × 10⁶ tumor cells were treated for 30 min with a 10 multiplicity of infection dose of Ad5IL-12 or DL70-3 vector. Recombinant human IFN-γ was used at 100 units/mL, recombinant human IL-12 at 10 μg/mL, and dexamethasone at 10⁻⁶ mol/L concentrations for tumor cell treatments. Tumor cells were harvested by trypsinization 24 h later for fluorescence-activated cell sorting (FACS) analysis.

Statistics. Statistical analysis was done using STATVIEW 4.5 programs from Abacus Concepts (Berkeley, CA). A Student's t test was used for the final determination of significance in testing the effects of the various treatments.

Ad5IL-12 vector transduction of LNCaP cells blocks their tumorigenicity in NOD.scid mice. The Ad5IL-12 vector had been tested previously in the therapy of tumors and showed significant efficacy for the treatment of experimentally established breast adenocarcinomas (18). We tested this vector's effects on prostate tumors with a similar epithelial origin. We found that the well-defined human LNCaP prostate cancer cell line formed palpable tumors with progressive growth when injected s.c. into NOD.scid mice. This system provided a model to test the impact of diverse therapies on tumor growth. Initial characterization of the effect of Ad5IL-12 vector treatment on the tumorigenicity of the LNCaP prostate cancer cell line was then done. Despite the absence of T and B cells, Ad5IL-12 vector transduction of the LNCaP prostate cell line before inoculation abolished the capacity of these cells to form tumors in NOD.scid mice. As shown in Fig. 1A, 2 × 10⁶ LNCaP cells infected with the Ad5IL-12 vector at 100 plaque-forming unit (pfu) per cell and injected s.c. after a 24 h incubation period (a time when IL-12 transgene expression is maximal, data not shown) were not capable of forming tumors. This was in contrast to DL70-3 control vector- or PBS-treated cells that formed palpable tumors and ultimately resulted in animal morbidity requiring euthanasia. Vector treatments did not affect cell growth or viability because neither Ad5IL-12 nor DL70-3 vector transduction of the LNCaP cells interfered with cell viability or proliferation as determined in vitro by trypan blue exclusion or ³H-thymidine incorporation (data not shown). Ad5IL-12–treated NOD.scid mice formed tumors upon rechallenge with 2 × 10⁶ unmodified LNCaP cells 8 weeks after the primary inoculation with Ad5IL-12–transduced cells (data not shown), implicating a transiently induced innate NK cell immunity mediating the protective response.

Effect of the Ad5IL-12 vector on established LNCaP prostate tumors in NOD.scid mice. Given the significant effect of Ad5IL-12 transduction on LNCaP cell tumorigenicity, we next tested if intratumoral injection of the Ad5IL-12 vector could impact the growth of established LNCaP tumors in NOD.scid mice, a more clinically relevant situation. Detectable tumors formed approximately 2 weeks postinjection of 2 × 10⁶ LNCaP cells, and once tumors reached palpable size, therapy was initiated by performing a single intratumoral injection with 5 × 10⁸ pfu of Ad5IL-12 vector or controls. As shown in Fig. 1B, significant differences in tumor growth became apparent between the Ad5IL-12 vector–treated mice versus the DL70-3 vector– and PBS-treated controls by week 5 (P < 0.05). Intratumoral injection with the Ad5IL-12 vector resulted in an approximate 3-fold reduction in tumor mass compared with the DL70-3 vector or the PBS control measured at the 6-week time point. In contrast to the previous use of the Ad5IL-12 vector in immunocompetent mice for breast tumor treatment, no complete regressions were evidenced by this single vector treatment (18). However, these results do indicate that localized IL-12 expression can recruit NK cells to mediate a growth delay in prostate tumors, independent of the adaptive immune response.

Reducing systemic cortisol levels with mitotane enhances the anti-tumor effect of Ad5IL-12 vector treatment. After showing that a single dose treatment with the Ad5IL-12 vector had a significant effect on LNCaP tumor growth, we next explored the impact that attenuation of the glucocorticoid response would have in conjunction with the vector therapy. There is a correlation between the proinflammatory immune response and the tumor therapeutic efficacy (26), although in this instance, the response measured would be devoid of the adaptive immune response (T and B cells). We used mitotane as an agent to produce pharmacologic adrenalectomy, aiming to extend the duration of the proinflammatory response generated by the Ad5IL-12 vector tumor treatment. Mitotane is currently used in the treatment of human adrenal carcinomas and has been shown to transiently reduce glucocorticoid production in animals (23, 24). No delay of tumor growth was observed by treatment with the DL70-3 vector alone or in combination with mitotane (Fig. 1B and data not shown). However, i.p. administration of a single dose of mitotane, concomitant with intratumoral Ad5IL-12 vector treatment, resulted in an approximate 2-fold reduction in tumor growth in comparison to IL-12 vector treatment alone (Fig. 1C). This result indicated that the Ad5IL-12 vector effect was impacted by the mitotane treatment, and that the combinatory therapy could provide enhanced efficacy.

Systemic cortisol concentrations were monitored over a 1-week period following adenovirus vector and mitotane treatment (Fig. 2). As expected, mice treated with mitotane alone exhibited a marked decrease in serum cortisol levels by 48 h after initiation of treatment, reaching a concentration of 9.4 ± 1.9 ng/mL. Interestingly, Ad5IL-12 vector treatment alone also strikingly reduced serum cortisol levels to lower concentrations than the mitotane treatment alone, reaching 6.9 ± 0.3 ng/mL at 48 h posttreatment. Consistent with our proposed relationship between cortisol levels and enhancement of the NK response, combined therapy of Ad5IL-12 and mitotane resulted in an even further significant reduction in cortisol levels (5.09 ± 0.3 ng/mL at 48 h) in comparison to each treatment alone (P < 0.05). For all treatment groups, cortisol levels reached their lowest levels at 48 h and then proceeded to increase and approach pretreatment “normal” baseline levels by day 7.

The Ad5IL-12 vector activates NK cytotoxicity that can be further enhanced with mitotane treatment. To directly show that the effect on tumor growth was mediated by NK cells, we tested the efficacy of Ad5IL-12 vector treatment on LNCaP tumors established in NOD.Cg mice. These mice are devoid of NK cell activity, and as shown in Fig. 3A, no protection is afforded by the Ad5IL-12 vector in the absence of NK cells. Earlier reports had shown some effect on tumor growth by IL-12 treatment alone by altering tissue neovascularization (27). However, we found no such impact for IL-12 vector on established tumors independent of NK cells, suggesting that the tumor vascular bed is already established and not susceptible to cytokine effects alone at this time point.

To determine whether Ad5IL-12 intratumoral vector therapy directly influenced systemic NK cell cytolytic activity, we did a standard ⁵¹chromium release assay using the YAC-1 cell line as the NK target. As shown in Fig. 3B, NK cytolytic activity detected from Ad5IL-12 intratumoral vector–treated animals was dramatically increased compared with controls. NK lytic activity reached ∼42% at the highest effector to target (E/T) ratio tested (50:1) following Ad5IL-12 vector treatment alone. This activity shows the potent capacity of the IL-12 vector to induce NK cell responses, taking into account the reported low detectable background NK activities reported for the NOD.scid strains (28). In comparison to the Ad5IL-12 vector treatment, negligible NK activity was detected in DL70-3 vector control-treated animals, even at the highest E/T ratio at which the cytolytic activity detected was equivalent to background levels, under 1%. We also assessed the NK activity in mice treated with the Ad5IL-12 vector and mitotane. Previous experiments studying the effect of cortisol on NK cell cytolytic activity have shown a direct suppressive effect on their function (29). Mitotane treatment, when done in combination with the Ad5IL-12 vector therapy, significantly enhanced the NK activity across all effector to target ratios. Mitotane, in combination with Ad5IL-12, resulted in an approximate 50% maximal detected lytic activity for the 50:1 E/T ratio. The enhancing effect of mitotane produced an approximate 3-fold higher increase in NK activity versus IL-12 vector treatment alone, as measured by the numbers of effectors required for equivalent lysis.

The enhancing NK activity obtained by combined mitotane-Ad5IL-12 therapy does not result from increased IL-12 expression in tumor tissue. Changes in IL-12 levels were measured in tumor tissue protein extracts following therapy (Fig. 4). No significant differences were detected in IL-12 expression when compared either with Ad5IL-12 + mitotane, or with Ad5IL-12 vector alone. At the 24 h time point, IL-12 expression reached peak levels in both Ad5IL-12–treated groups, with levels ranging between 17 and 20 ng per tumor. This expression profile was similar to the previously reported kinetics using this vector in treatments of breast tumor and leishmania vaccination with peak expression also occurring at 24 h (16, 18). IL-12 expression diminished by day 3 in all Ad5IL-12–treated animals and became undetectable by day 7. Control animals did not produce detectable levels of IL-12 cytokine at any time point. Because the IL-12 expression pattern did not differ between mitotane and the Ad5IL-12 treatment alone, both in duration or in level, it seems that vector cytokine transgene expression is not affected by lower cortisol levels induced by mitotane and does not account for the differential impact on tumor growth.

Intratumoral injection with the Ad5IL-12 vector can attenuate the tumor growth of both MICA/B-positive class I–negative LNCaP and MICA/B-negative class I–positive PC3 tumors in NOD.scid mice. Differences in γδ-T cell-mediated tumor killing have been reported, showing that γδ-T cells derived from various human donors were only capable of killing the PC3 prostate cancer cell line tumors but not LNCaP tumors (30). To assess if there were any phenotypic differences between prostate tumor types that could account for this differential response, we examined LNCaP and PC3 cells for their surface expression of MHC class I molecules and MICA/B, major ligands for NK-activating receptors. Interestingly, we found a reciprocal pattern of expression for these molecules between these two prostate tumor types (Fig. 5). FACS analysis revealed that LNCaP tumor cells bore negligible surface expression of MHC class I molecules but had high expression of the NKG2D MICA/B ligands. This finding suggested that LNCaP cells, defective in MHC class I expression, should not be effectively targeted by T cells and might explain the differential effects of γδ-T cells on these tumors.

Likewise, differential expression of the MICA/B molecules could affect the tumor killing capacity of NK cells on prostate tumor cells. Differential NK-mediated killing of different subsets of hepatoma cells based on MICA/B expression has already been shown (7). Due to the defective expression of MICA/B in PC3 cells, we examined PC3 cells to assess their resistance to NK cell–mediated killing. As shown in Fig. 6, Ad5IL-12 intratumoral injection inhibited the growth of PC3 tumors, similar to the vector's effect on LNCaP tumors. Therefore, additional NK cell receptor ligands, independent of MICA/B, must be involved in targeting PC3 tumors during IL-12 therapy. Despite differences in putative targeting mechanisms between the PC3 and LNCaP tumor types, these results support the use of Ad5IL-12 treatment to elicit an NK cell response as a generic application to induce prostate tumor–specific killing.

We also assayed for changes in MICA/B and MHC class I expression on LNCaP and PC3 tumors after 24 h of treatment with the Ad5IL-12 and DL70-3 vectors following in vitro culture. No changes in surface expression of these molecules were detected for either tumor type, nor were any changes evident following coculture with dexamethasone treatment, indicating that changes in tumor killing capacity following treatment are dependent on NK cell activation and function (data not shown). Likewise, neither treatment with recombinant human IFN-γ nor IL-12 resulted in changes in MICA/B or class I surface expression, suggesting that vector transduction was not an interfering factor, but that activation of these abnormal polyploid tumors is highly askew (data not shown).

We have undertaken this work to begin developing curative clinical approaches to the very widespread problem of prostate cancer, one of the cancers affecting large numbers of male citizens in the United States, approaching breast cancer in its death rate. To date, most attempts at tumor-specific immunotherapy have predominantly focused on the generation of antigen-specific T cell responses alone. However, T cell targeting may not be a sufficient approach against many tumors that bear low levels of surface MHC class I molecules. Even in those circumstances where MHC class I expression is adequate, only addressing the activation of the T cell may not be a successful strategy.

T cell therapies may fail owing to defective antigen presentation by tumor cells themselves. This has been shown by many examples in previous reports. Using a recombinant adenovirus vector vaccine targeting gp100, it was shown that resistance to tumor killing of B16F10 tumors was due to a lack of appropriate TAP-1–dependent epitope display by the tumor cell (31). A similar result was shown in patients that were unable to kill autologous melanoma tumor cells in response to an HLA-A0201–restricted telomerase-derived peptide (p540-548) vaccine, unless the antigenic peptide was added exogenously during the killing assay (32). These deficits in tumor antigen presentation can be affected by many different elements along the class I antigen processing pathway. Defective human class I–associated antigen presentation on melanoma cells can be caused by the loss of β-2-microglobulin function in the tumor cells (33). Processing of peptides by the proteosome complex of proteases has also been shown to affect CTL recognition of tumor cells (34). Clearly, many genetic changes may exist in tumor cells favoring the escape from T cell immunosurveillance.

Due to these many potential failures associated with T cell–directed therapy, we have focused in the present study on the use of NK cells to effect tumor killing. We have studied the in vivo therapeutic properties of the Ad5IL-12 vector by locally treating human LNCaP and PC3 tumors established in immunodeficient NOD.scid mice. Most importantly, we could show that these elicited NK cell responses were capable of killing tumors that are low in MHC class I expression (e.g., LNCaP cells). The loss of MHC class I expression is a frequent occurrence in cancer cells of many human tumors and is consistent with the concept of tumor immunoediting (35). During immunoediting, adaptive immune responses generated against tumor-specific antigens provide selection pressures that result in the outgrowth of escape variants resistant to adaptive immune clearance. Using the recombinant Ad5IL-12 adenovirus vector to produce IL-12 cytokine in the local tumor microenvironment, we were able to show the induction of NK cell activity in treated animals as well as avoiding any systemic toxicity related to IL-12 delivery. This technique of direct tumor injection with vector could be clinically appropriate for the treatment of prostate cancers in humans.

Using this NOD.scid animal model to monitor the effects of vector Ad5IL-12 NK-elicited killing, the inhibitory killer cell immunoglobulin-like receptors (KIR) are ligand mismatched to the three major specific class I allele groups of human cells (36). In humans, the three major NK inhibitory receptors (KIR2DL2/3, KIR2DL1, and KIR3DL1) recognize specific motifs of HLA-C and HLA-B alleles, allowing NK cells to mediate the killing of targeted cells that do not express the appropriate blocking allele (37). Thus, the tumor killing mediated by murine NK cells of human prostate tumors in the NOD.scid model most probably reflects the therapeutic approach in humans of using haploidentical, alloreactive NK cell–adoptive immunotherapy of solid tumors that has been implemented in some patients (38). The use of IL-12 as an adjuvant in this context has never been tested. We propose that the combination of Ad5IL-12 and cortisol blockade therapy can be effective, independent of NK cellular adoptive transfer regimens. However, in the near future, we also plan to expand our investigations to include NK cell transfers in therapy using the Ad5IL-12 vector. This use of IL-12 as an adjuvant in the context of alloreactivity has not been tested. We propose that the combination of Ad5IL-12 and cortisol blockade therapy should also be effective in NK cellular adoptive transfer therapy regimens, although all the effects we describe here of the Ad5IL-12 vector treatment were mediated through induced NK activity on the established prostate tumors.

It is now clear that many factors can control the tumor microenvironment and the adaptive immune responses against tumors. Tumor escape from the innate immune response is also subject to conditions such as hormonal control within the tumor microenvironment that could have a significant impact on the NK cell response. Restoration mechanisms of homeostasis mediated by systemic release of cortisol, as a downstream consequence of local inflammatory signals, could affect host immune responses by controlling the activation status of various tumor-infiltrating cells. With regard to tumor immunity, cortisol has been shown to directly inhibit NK cell cytolytic functions as well as to suppress both adaptive T cell responses and dendritic cell (DC) immune stimulatory activity (29, 39, 40, 41, 42). The nuclear factor protein, glucocorticoid-induced leucine zipper (GILZ), is up-regulated in T cells and DCs in response to cortisol and may exert a powerful general suppressive effect by controlling expression of many cytokine and immune response genes that mediate or contribute to tumor-killing responses (43). GILZ has been shown to prevent downstream transcriptional activation of nuclear factor (NF) κB–mediated genes by direct protein-protein interaction with NFκB (44). Thus, the activation status of NFκB in tumor-infiltrating cells of both the adaptive and innate immune response is subject to control by cortisol levels in the tumor microenvironment.

Supporting this contention, we were able to increase the NK activity generated by the Ad5IL-12 vector through combination therapy with mitotane. This drug targets the zones of the adrenal cortex that produce cortisol and, dependent upon dosage, can be used to control the level of hormone produced. We found that mitotane treatment decreased circulating cortisol levels in the sera of treated animals in comparison with controls. We also found that Ad5IL-12–treated mice experience a pronounced down-regulation of systemic cortisol that could be further decreased by mitotane cotreatment. This combined therapy resulted in the enhanced killing of the MHC class I–negative LNCaP tumors and could prove to be beneficial in the treatment of other class I–negative tumor types. Given our results with mitotane, it will be of interest to test the impact of cortisol receptor blockade using treatments of RU486 in conjunction with Ad5IL-12 vector therapy. RU486 (mifepristone) is a potent glucocorticoid receptor antagonist. This combination of therapies offers an interesting potential dual benefit of enhancing the NK response while protecting against one of the more serious side effects of cancer, because repeated administrations of RU486 have been shown to prevent cachexia with limited toxicity in animals (45).

Cortisol blockade could also act to prevent matrix metalloproteinase (MMP)–mediated shedding of soluble MICA (sMICA) from tumors that has been shown to impair NK cell and CD8 T cell effector functions (46). Gene profiling experiments to categorize the actions of cortisol using dexamethasone treatment showed that with the exception of MMP-9, metalloproteinases are up-regulated in response to glucocorticoids (47). Thus, it is possible that the blocking of cortisol actions in the tumor microenvironment could prevent or attenuate the shedding of sMICA. This effect could be involved in the NK-enhancing activity caused by mitotane in the presence of Ad5IL-12.

These results represent a preliminary examination to learn the effects of altering the tumor microenvironment in humanized mouse models of prostate cancer. We now plan to expand these studies and to test the coadministration of tumor antigen–specific CD4 and CD8 T cell lines in conjunction with the IL-12 vector/cortisol therapy to build a stronger combinatorial therapy. The recent revelation that interactions between DC and NK cells can bypass the T helper arm in CTL induction suggests that NK-targeted responses may be highly beneficial as an adjunct to lymphocyte-mediated immunotherapies as well (48, 49). We believe that this approach of NK cell induction through Ad5IL-12 vector treatment offers the promise of creating a generic tumor therapy, and that cortisol control could represent the flip side to the coin that the historical treatment with cortisone has represented for immune system suppression in the therapy of inflammatory- and autoimmune-mediated diseases. Additionally, cortisol attenuation could possibly counteract the immunosuppressed and cachexic status often seen in patients with advanced stages of cancer that may be associated with elevated cortisol levels.

Grant support: Department of Defense research award DAMD-17-02-1-0080 and a grant from the Alzheimer's and Aging Research Center (San Diego, CA).

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.

We would like to thank Drs. Ramesh Halder and Yang Dai for their assistance with FACS analysis and Lindsey Harvey for administrative help.