Immunotherapy for Human Renal Cell Carcinoma by Adoptive Transfer of Autologous Transforming Growth Factor β–Insensitive CD8⁺ T Cells

Renal cell carcinoma (RCC) is the most common solid tumor of the kidney in adults. It has previously been reported that the overproduction of TGF-β by RCC cells may lead to tumor evasion from the host immune surveillance and subsequent tumor progression (1–4). Similarly, inhibition of transforming growth factor-βTGF-β signaling using a dominant-negative TGF-β type II receptor construct (TβRIIDN) generates an immune response capable of eradicating tumors in mice challenged with live tumor cells (5).

We have previously shown that murine CD8⁺ T cells that are rendered insensitive to TGF-β could activate the antitumor immune response cycle in prostate cancer and subsequently eradicate lung metastases (6, 7). Moreover, inhibition of TGF-β signaling in DC also enhanced the efficacy of DC-based vaccines (8, 9). Although such TGF-β inhibition strategies are promising therapeutic approaches, little is known about its effects on human malignancies in vivo. Currently, the safety, efficacy, and complexity of immunomodulation humans are major concerns that inhibit the clinical application of TGF-β blocking strategies. Therefore, a preclinical model that uses and analyzes human tumor–reactive TGF-β–insensitive CD8⁺ T cells and their effect on human RCC is needed.

The severe combined immunodeficient (SCID) mouse has been shown to be an ideal host for studying patient or healthy donor immune system responses. Since Moiser et al. (10) first reported reconstitution of SCID mice with human peripheral blood mononuclear cells (Hu-PBMC), this approach has been widely applied in the fields of cancer biology, autoimmunity, allergy, infections, and transplantation biology. Bonnet et al. (11) reported that CD8⁺ CTL clones specific for minor histocompatibility antigen could inhibit the engraftment of human acute myeloid leukemia cells in immunodeficient nonobese/SCID (NOD/SCID) mice. In another study, researchers showed that i.v. administration of in vitro generated donor-derived CTLs into NOD/SCID mice could eradicate human acute lymphoblastic leukemia (12). Recently, researchers have shown that infusion of PBMCs obtained from patients with chronic lymphocytic leukemia into NOD/SCID mice could mimic many clinical characteristics of the disease (13). Given the success of these adoptive immunotherapy models for studying hematologic malignancies, it is possible that similar techniques may be developed to study solid tumors such as RCC.

In the present study, we isolated tumor-reactive CD8⁺ T cells from RCC patients, activated them with autologous immature dendritic cells (DC) and tumor lysate, rendered them insensitive to TGF-β by infection with a TβRIIDN retrovirus, and subsequently tested their antitumor responses both in vitro and in vivo. By using the SCID mouse model humanized with PBMC from RCC patients (hu-PBMC-SCID), we were able to transfer the autologous TGF-β–insensitive CD8⁺ T cells into RCC tumor–bearing animals without inducing severe graft versus host disease (GVHD). Our results show that these autologous TGF-β–insensitive CD8⁺ T cells exhibit antitumor activity, decrease tumor burden, and suppress pulmonary metastases.

Specimens were obtained from 125 patients who were diagnosed with RCC and underwent radical nephrectomy between September 2005 and December 2008 in Xijing Hospital. The study protocol was approved by the Ethics Committee of the Xijing Hospital, Fourth Military Medical University. Informed consent was obtained from all participants. A total of 125 fresh RCC patient tissues were obtained at the time of surgery and were subsequently implanted s.c. into nude mice as described previously (14). Ten different RCC specimens remained engrafted successfully 1 mo after transplantation. Histologic sections from the original patient tumors as well as from xenograft specimens were obtained and subjected to H&E analysis. Cells from the 10 different RCC xenografts were subsequently isolated, grown in culture, and passed over 100 times. These RCC cell lines were maintained in complete medium containing RPMI 1640 (HyClone) supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies), 2 mmol/L l-glutamine, 50 μmmol/L 2-mercaptoethanol, 100 U/mL penicillin, and 100 μg/mL streptomycin (Sigma).

Immunohistochemical analysis was done to assess TGF-β expression levels in both the surgical patient specimens as well as the 10 different RCC cell lines. To this end, paraffin-embedded sections (4 μm) obtained from either the surgical specimens or xenografts were deparaffinized and rehydrated. After quenching endogenous peroxidase and blocking with serum, tissue sections were incubated with anti–TGF-β1 monoclonal antibody (1:100 dilution; ab27969, Abcam Biotechnology). Sections were subsequently washed with PBS and then incubated with biotinylated goat-anti-mouse secondary antibody (1:500; Abcam Biotechnology). Peroxidase substrate solution containing 3,3′-diaminobenzidine was used for direct staining. Counter-staining was done with 10% hematoxylin. Nonimmune murine antibody was used as a negative control. For immunofluorescence analysis, cells were incubated with TGF-β1 monoclonal antibody for 2 h, and then immunostained with FITC-conjugated anti-mouse IgG (1:1,000, Abcam Biotechnology) for 1 h. Cellular nuclei were identified using 100 ng/mL DAPI and all cells were subsequently examined by fluorescence microscopy (Nikon Corp.).

To efficiently transfer CD8⁺ T cells into RCC tumor–bearing animals without inducing severe GVHD, SCID mice were first humanized with PBMCs obtained from human patients with RCC. To this end, male or female SCID/beige mice age 6 to 8 wk were obtained from the Laboratory Animal Research Center of the Fourth Military Medical University and housed in sterile filter-top cage placed in a laminar backflow cabinet under specific pathogen-free conditions. The SCID-beige mice were divided into 10 different groups. Each group was injected with PBMCs that were derived from 1 of the 10 different patients that were used for generation of the RCC cell lines. Autologous PBMCs were purified from each patient's blood using a Ficoll-HyPaque (Pharmacia) gradient after platelet depletion and washing as previously described (8). One day before PBMCs injection, mice were sublethally irradiated with 3.5 Gy [(60Co) source Gammatron F 80S, Simens]. Each mouse received 0.3 mL of the PBMCs (2 × 10⁷ cells) suspended in PBS through i.p. injection.

The pooled conditioned medium was collected and concentrated by using YM-3 Centriprep Centrifugal Filter Devices (Millipore). TGF-β1 ELISA was carried out using the Quantikine Human TGF-β1 Immunoassay kit from R&D Systems (Minneapolis). The total number of cells in each flask was counted using a Coulter Counter and levels of TGF-β1 were reported as pg/1 × 10⁵ cells/48 h. Human prostate cancer cell line PC-3 (American Type Culture Collection) was used as a control.

Four weeks after PBMCs injection into SCID-beige mice, 100 μL of tail vein blood were obtained and a sandwich ELISA was done to quantify serum human IgG as previously described (15). Briefly, microculture plates were coated with affinity-purified goat anti-human IgG (Abcam Biotechnology). Affinity-purified alkaline phosphatase-conjugated goat anti-human IgG (Abcam Biotechnology) was used to detect human IgG. The absorbance at 405 nm was quantified on an ELISA reader.

With the use of CD8⁺ Microbeads (Miltenyi Biotec), patient's CD8⁺ T cells were positively selected from whole blood with a purity of >98%. CD8⁺ T cells was expanded with using autologous patient's DCs pulsed with the tumor lysate in the presence of recombinant human interleukin-2 (500 U/mL; PeproTech) as previously described (8).

There were two types of CD8⁺ T cells used for experimentation: (a) tumor-reactive TGF-β–insensitive CD8⁺ T cells that were rendered insensitive to TGF-β by infection using a retrovirus containing TβRIIDN-green fluorescent protein (GFP)7 and (b) naive CD8⁺ T cells isolated from PBMC (controls). The efficiency of infection of TβRIIDN was 85% (green fluorescent protein–positive cells: total cells) under fluorescence microscopy (Nikon Corp.; Supplementary Fig. S1A). Under the treatment of TGF-β (10 ng/mL for 16 h), phosphorylation of Smad-2 was observed in naive CD8⁺ T cells but not in tumor-reactive CD8⁺ T cells infected with the TβRIIDN. This result indicates that these TβRIIDN-infected CD8⁺ T cells were insensitive to TGF-β (Supplementary Fig. S1B).

Immunophenotypical characterization of TGF-β–insensitive CD8⁺ T cell and naive CD8⁺ T cell were done using FITC-conjugated anti-CD27 monoclonal antibody (Cell Signaling) and anti-CD45RA monoclonal antibody (Cell Signaling) before administration to the mice. Cells were stained with monoclonal antibody in PBS, 0.2% bovine serum albumin, and 50 μmol/L EDTA for 20 min at 4°C and either directly analyzed or sorted into defined populations on a FACSVantae SE, using CellQuest software (BD Bioscience; ref. 16).

The two types of CD8⁺ T cells (TGF-β–insensitive CD8⁺ and naive CD8⁺ T cells) were subjected to a standard ⁵¹Chromium release (⁵¹Cr) assay as previously described (6). The 10 different RCC cell lines and, as a negative control, the human PC-3 prostate carcinoma cell line were used as targets. Target cells were labeled with 100 μ Ci ⁵¹Cr/10⁵ cells. Different groups of CD8⁺ T cells were added to U-bottomed plates containing 5,000 cells per well with various E/T ratios ranging from 1:1 to 100:1. Equal volumes of RPMI 1640 and 1 mol/L HCl were added to other wells as the negative and positive controls, respectively. After a 4-h incubation, 100 μL of supernatants was harvested from each well and the ⁵¹Cr released was measured using a γ counter. The percent cell lysis was calculated according to the formula as previously described (6): percent-specific ⁵¹Cr release = (experimental release − spontaneous release) × 100/(maximum release − spontaneous release)].

There are 30 Hu-PBMC-SCID mice that received an injection in the right flank using 5 × 10⁶ of patient's autologous RCC tumor cell line (day 0). Tumors were found in all these mice at approximately 2 to 3 mm in diameter and were palpable 14 d later. On day 14, adoptive transfer with patient's autologous CD8⁺ T cells was done in the tumor-bearing Hu-PBMC-SCID mice. Three groups (10 mice per group) received i.p. injection with different types of adoptive transfer composed of either (a) TGF-β–insensitive CD8⁺ T cells (1 × 10⁷), (b) naive CD8⁺ T cells (1 × 10⁷), or (c) PBS (0.5 mL). The adoptive transfer methodology was repeated on day 21. Tumor growth and animal survival was monitored daily after vaccination.

To evaluate the effects of TGF-β on cellular metastasis, the pulmonary metastasis model was done as previously described (6). Briefly, a single injection of 5 × 10⁶ RCC cells were injected through the tail vein. On day 7, the tumor-bearing mice (5 mice per group) were inoculated with either TGF-β–insensitive or naive CD8⁺ T cells (1 × 10⁷ cells) or PBS through i.p. injection. Forty days after adoptive transfer, all mice were sacrificed and the tumors were isolated for evaluation of the volume (volume = length × width2 × π/6), weight, and histologic analysis. Lung tissues were also harvested.

The sera of the mice were obtained using tail vein blood. Because IFN-γ is commonly used as a measure of T-cell activation, serum levels of INF-γ were determined using an ELISA kit (R&D Systems) according to an established protocol. Serum samples were stored at −70°C until the assay.

Paraffin-embedded tumor sections (obtained from mice that underwent adoptive transfer) were used for apoptosis assays. The nuclear and terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL; R & D System) apoptosis assays were done as described previously. The intensity of the fluorescent signal was standardized by the standard fluorescent index (positive lymphocytes or signal/100 tumor cells/1,000 μm²) as follows: −, <5; ±, 6 to 10; +, 11 to 30; ++, 31 to 50; +++, 51 to 70; ++++, >70 as described previously (7).

Numerical data were expressed as mean ± SD. ANOVA and χ² tests were done to determine the differences in the means among the various treatment groups. P < 0.05 was considered statistically significant. The SPSS 10.0.2 software package (SPSS, Inc.) was used for analysis. The Kaplan-Meier survival curve was analyzed by the log-rank test with the Graphpad Prism 5.0 software (Graphpad Software, Inc.).

Tumors obtained from human patients with RCC were injected into nude mice and used to establish 10 different RCC cell lines. All RCC cell lines were stably cultured and were able to withstand repeated cryopreservation and thawing. Histologic analysis of xenografts obtained from mice was similar to the histologic evaluation of the original patient tissue specimen (Supplementary Table S1; Fig. 1). In addition, cultured cells appeared to retain many of the morphologic properties observed in vivo.

We next sought to examine the expression patterns of TGF-β1 in RCC xenografts as well as the 10 immortalized human RCC cell lines. Immunohistochemical analysis of RCC xenografts obtained from SCID-beige mice shows strong expression of TGF-β1 throughout the cytoplasm and cellular membranes of neoplastic cells (Fig. 2A and B). Similar staining patterns were observed by immunofluorescence staining of RCC cell lines (Fig. 2C and D). This result is consistent with ELISA analysis (Supplementary Fig. S2).

To develop an adoptive transfer murine model to study the effects of TGF-β1 inhibition on RCC tumor progression, SCID mice were first humanized with PBMCs obtained from 10 different human patients with RCC. Of note, PBMCs were obtained from the same patients whose tumors were used to establish the RCC cell lines. Four weeks after PBMCs were injected into SCID mice, human immunoglobulins could be detected in 300 of 440 (68.2%) Hu-PBMC-SCID mice sera. The IgG levels of each group averaged between 0.8 and 2.2 mg/mL, which are in agreement with results of previous studies (15). There were no significant differences in the success rate of PBMC engraftment and the levels of IgG were similar between each group of mice (P > 0.05; Fig. 3A). Furthermore, we found no evidence of severe xenogenic GVHD.

To analyze the effects of TGF-β on immune surveillance of RCC, CD8⁺ T cells were isolated using the 10 patients' PBMCs. Some of these CD8⁺ cells were infected with a retroviral construct containing a dominant negative TGβRII-green fluorescent protein construct. Next, the phenotypes of both the TGF-β1–insentsitive and naïve CD8⁺ T-cell lines were characterized. To this end, flow cytometry analysis showed that the expression of the costimulatory molecules CD27 and CD45RA were different in the two CD8⁺ T-cell groups. In the TGF-β–insensitive CD8⁺ T cells, the dominant phenotype was the CD27⁺CD45RA⁻, which is phenotype of effector CD8⁺ T cells. However, most of naive CD8⁺ T cells were identified as CD27⁺CD45RA⁺, which is consistent with unprimed CD8⁺ T cells (Supplementary Table 2; Fig. 3B and C).

The specific tumor-killing ability of the autologous TGF-β–insensitive CD8⁺ T cells (described above) was first examined using an in vitro CTL assay. To this end, TGF-β–insensitive CD8⁺ T cells exhibited a 5-fold higher tumor-killing activity compared with naïve CD8⁺ T cells [75.5% versus 15.8% at an effector/target (E/T) cell ratio of 100:1; Fig. 4A]. No apparent lytic activity was observed when the assay was done using the PC-3–negative control cell line. (Fig. 4B).

In vivo assays further confirmed the antitumor abilities of TGF-β–insensitive CD8⁺ T cells. Adoptive transfer of either TGF-β–insensitive CD8⁺ T cells or naïve CD8⁺ T cells into Hu-PMBC-SCID mice inoculated with RCC tumors from the same human patients (see Materials and Methods) were used in this regard. Forty days after adoptive transfer, mice injected with TGF-β–insensitive CD8⁺ T cells exhibited average tumor volumes and tumor weights that were significantly decreased compared with mice injected with either naive CD8⁺ T cells or PBS (P < 0.05; Fig. 4C and D). Taken together, TGF-β–insensitive CD8⁺ T cells decreased tumor growth and were therefore associated with antitumor activity in vivo.

Finally, it was of interest to determine whether inhibition of TGF-β in CD8⁺ T cells could inhibit cellular metastasis of RCC. Therefore, each one of the 10 different RCC cell lines was independently injected into Hu-PMBC-SCID mice tail veins. Seven days after tail vein injection, mice were inoculated with either TGF-β–insensitive or naive CD8⁺ T cells or with PBS as a control. Interestingly, all the animals died in the PBS-treated group before day 30, 80% of mice died in the naïve CD8⁺ T–treated group before day 32 of the experiment due to poor health conditions, whereas all the mice survived in the TGF-β–insensitive CD8⁺ T cell–treated group at the end of the experiment (Fig. 5A). According to the long-rank test, there were significant differences among three groups (P < 0.05; Fig. 5B).

Activated CD8⁺ T cells release INF-γ, and therefore, measurement of INF-γ levels can be used to determine the activity of T lymphocytes in vivo. Therefore, serum levels of INF-γ were determined in Hu-PMBC-SCID mice that were inoculated with RCC tumors and underwent adoptive transfer with either TGF-β–insensitive or naive CD8⁺ T cells. The serum levels of INF-γ exhibited significant (P < 0.01) increases in mice with TGF-β–insensitive CD8⁺ T cells, but not in mice treated with either naive CD8⁺ T cells or PBS. (Fig. 5C). The increased levels of serum INF-γ observed in the TGF-β–insensitive CD8⁺ T cell–treated group indicated that immune cells were more strongly activated in these hosts.

The above experiments suggest that TGF-β–insensitive CD8⁺ T cells are associated with increased immunomodulatory functions in vivo, which result in tumor shrinkage and decreased metastases. To further confirm this increased activity, TUNEL assays were used to show that autologous TGF-β–insensitive CD8⁺ T cells could induce tumor cell apoptosis in the TβRIIDN group (+++, 52 apoptosis signal/100 tumor cells/1,000 μm²). However, almost no apoptotic cells was observed in the naive CD8⁺ T cell (−, 2 apoptosis signal/100 tumor cells/1,000 μm²) or PBS groups (−, 0 apoptosis signal/100 tumor cells/1,000 μm²; Fig. 6).

In this study, we used a one-to-one adoptive immunotherapy strategy to treat human RCC with patients' own CD8⁺ T cells in a SCID mouse model. This preclinical study show that adoptive transfer of autologous TGF-β–insensitive CD8⁺ T cells can induce tumor apoptosis that subsequently decreases tumor burden and suppress pulmonary metastases. These tumor-reactive, TGF-β–insensitive CD8⁺ T cells show a strong antitumor ability both in vitro and in vivo.

Nude mice engrafted with RCC primary xenograft present advantages to study tumor biological behaviors. However, the success rate in established xenograft animal model is not always satisfactory due to the difference on histoincompatibility between human and animal (17). Here, we reported that the establishment of 10 RCC cell lines derived from human patient tumors. Only 10 of 125 patient tumors were able to be used in this regard (8.0% success rate), which is lower than that of other reports (18, 19). Because of this reason, we used the nude mice to establish the RCC cell lines, and then used the SCID/beige mice to assess tumor burden and immune cell reconstitution. We believe that this provides improved immunohistocompatibility compared with other methods that directly implant human tumor tissues directly into SCID mice.

Since the first report of a successful transfer of normal human immune cells to SCID mice, the reconstitution of a human immune system in mice for studying human immune reaction in vivo has been extensively studied (10). The SCID/beige mouse model has a combined defect of the T/B cell system and the innate immune system, and is therefore advantageous for studying human immune cell reconstitution (20). Berney et al. (21) reported that after i.p. injection, the engraftment of human PBMC in SCID/beige mouse is largely improved, and although human B cells are virtually absent from the circulation, they are present in lymphoid organs and significant levels of human IgG are found in the blood. Our experiments confirmed this finding without inducing severe GVHD.

The effectiveness of adoptive therapy as a treatment strategy may depend on generating and administering antigen specific T cells in ways that mimic physiologic conditions to preserve normal function. For example, adoptive T-cell therapy using antigen-specific CD8⁺ T-cell clones for the treatment of patients with metastatic melanoma showed that the transferred T cells could persist in vivo, and show specific migration and antitumor effect (22). In this study, antigen-specific CD8⁺ T cells were generated in vitro by cyclical stimulation with autologous DCs plus tumor lysate and low-dose interleukin-2. Using this cocktail approach, we obtained large number of activated CD8⁺ T cells for further experiments. By infecting with a retrovirus containing TβRIIDN, this type of CD8⁺ T cells was rendered TGF-β insensitive. Phenotypic analysis showed that this type of CD8⁺ T cells were CD27⁺CD45RA⁻, which indicated the effector-type CTLs (16, 23–25). The CTL assay also confirmed its potent ability to kill target tumor cells in vitro.

Studies have shown that CD4⁺ T cells are required for the cytolytic activity of CD8⁺ T cells (5, 26, 27). Our previous studies have also shown that the adoptive transfer of tumor-reactive, TGF-β–insensitive CD8⁺ T cells alone were insufficient for an antitumor response unless they are supported by other immune cells (7). In this study, we established the Hu-PBMC-SCID mouse model instead of transferring TGF-β–insensitive CD8⁺ T cells directly to SCID mouse. Results of pathologic evaluation show that tumor-reactive TGF-β–insensitive CD8⁺ T cells persist in tumor-bearing hosts and subsequently reduce tumor burden. The circulating level of INF-γ, which a critical cytokine for antitumor activity in the host, is also elevated in TGF-β–insensitive CD8⁺ T cell–treated hosts (28). Furthermore, the TUNEL assay shows that patient's autologous TGF-β–insensitive CD8⁺ T cells could induce tumor tissue apoptosis. However, in naive CD8⁺ T cell–treated hosts, there is no significant tumor apoptosis found in tumor parenchyma.

Many studies have shown that blockade of TGF-β signal in cancer offers a promising immunotherapy strategy (29, 30). Results of our previous studies have also shown that adoptive transfer of murine-derived TGF-β–insensitive CD8⁺ T cells or DCs represent an efficient immunotherapeutic strategy in the treatment of murine prostate cancer (6–9). In the present study, using this one-to-one adoptive transfer strategy, we further confirm that tumor-reactive TGF-β–insensitive CD8⁺ T cells is sufficient for tumor rejection.

We did not observe the complete resolution of RCC tumors in the TGF-β–insensitive CD8⁺ T cell–treated mice. However, these TGF-β–insensitive CD8⁺ T cells were almost 100% effective in preventing pulmonary metastases. The exact mechanisms of these antitumor effects are currently unknown and should be further explored.

It has been shown that the dominant negative TGF-β receptor type II is an effective method to inhibit TGF-β signaling (30). This technique is effective in our murine model and is associated with RCC tumor apoptosis and inhibition of metastasis. However, before its clinical application, the safety should be first considered. Although we did not observe any toxic reactions in humanized animal during the experiment period, to gain full appraisal of the safety and efficiency, more clinical trials and further experiments should be done.

The present translational project proposes the following new potential advances when compared with our previous studies and other studies in the same field. For example, this new systemic immunotherapy based on adoptive transfer of autotrophic CD8⁺ T cells for both localized and metastatic models in immunodeficient mice mimic similar human RCC disease. To the best of our knowledge, these novel approaches have yet to be reported. These findings will hopefully establish the groundwork for additional clinical research in this field. Specifically, the success of this project will establish a preclinical foundation for the treatment of RCC recurrence and metastasis after surgical treatment. For example, we envision that in the future, it may be possible to preserve naive CD8⁺ T cells after surgical operation and render them insensitive to TGF-β before adoptively transferring these specific CD8⁺ T cells back to the same patient for immunotherapy.

In summary, results of the present study showed that the adoptive transfer of patient's autologous TGF-β–insensitive CD8⁺ T cells to RCC-bearing humanized SCID mouse could decrease tumor burden, suppress pulmonary metastasis, prolong survival time, and show the superior antitumor responses in vitro and in vivo. This approach may lead to the highly effective treatment for patients with recurrence or metastasis.

No potential conflicts of interest were disclosed.

Grant Support: Grants from the National Basic Research Program of China 2009CB521705, the National High Technology Research and Development Program of China 2006AA02A245, the Nature Science Foundation of China 30672097-30300413, the Technology Plan Emphasis Item of Peking D0206011000091, the Portes Center/Institute of Medicine of Chicago (Q. Zhang) and the American Cancer Society, Illinois (#08-22).

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.