The implantation of small pieces of human primary lung tumor biopsy tissue into SCID mice results in a viable s.c. xenograft in which the tissue architecture, including tumor-associated leukocytes, tumor cells, and stromal cells, is preserved in a functional state. By monitoring changes in tumor volume, gene expression patterns, cell depletion analysis, and the use of function-blocking Abs, we previously established in this xenograft model that exogenous IL-12 mobilizes human tumor-associated leukocytes to kill tumor cells in situ by indirect mechanisms that are dependent upon IFN-γ. In this study immunohistochemistry and FACS characterize the early cellular events in the tumor microenvironment induced by IL-12. By 5 days post-IL-12 treatment, the constitutively present human CD45+ leukocytes have expanded and infiltrated into tumor-rich areas of the xenograft. Two weeks post-treatment, there is expansion of the human leukocytes and complete effacement of the tumor compared with tumor progression and gradual loss of most human leukocytes in control-treated xenografts. Immunohistochemical analyses reveal that the responding human leukocytes are primarily activated or memory T cells, with smaller populations of B cells, macrophages, plasma cells, and plasmacytoid dendritic cells capable of producing IFN-α. The predominant cell population was also characterized by FACS and was shown to have a phenotype consistent with a CD4+ effector memory T cell. We conclude that quiescent CD4+ effector memory T cells are present within the tumor microenvironment of human lung tumors and can be reactivated by the local and sustained release of IL-12 to proliferate and secrete IFN-γ, leading to tumor cell eradication.
The constitutive presence of tumor-associated inflammatory cells within many types of tumor, including ovarian, colorectal, cervical, and lung (reviewed in Ref. 1) is something of an enigma, because they appear to be unresponsive to the progressing tumor. Although these inflammatory cells have been defined phenotypically and studied in vitro for their functional capabilities when separated and studied independently from the tumor microenvironment (2, 3, 4), their actual functional significance in situ, whether beneficial or detrimental to the host (reviewed in Ref. 5) remains largely unexplored. It has been impossible to distinguish by morphology alone the inflammatory networks that cause tumor arrest from those that either promote tumor growth or have no effect. Therefore, inflammatory networks in tumors remain a histopathological variable that may well be important and potentially exploitable therapeutically, but are currently poorly understood (6). Valuable insights would be expected if one were able to define the cellular and molecular events that occur within the intact tumor microenvironment in response to stimuli and to correlate these events with tumor arrest. A better understanding of tumor-associated leukocyte function in situ should thus help with the design of novel therapeutic strategies that could direct and promote the antitumor activities of these inflammatory cells while suppressing the tumor-promoting effects of the leukocytes.
A human/SCID mouse chimeric model has been developed in which intact pieces of primary human non-small cell lung cancer biopsy tissue are surgically implanted into the subcutis of SCID mice (7, 8). This model has resulted in a system that permits evaluation of the cellular response to IL-12 in a nondisrupted tissue environment. The histological structure of the original tumor is preserved, and the tumor microenvironment, consisting of inflammatory leukocytes, stromal cells, and blood vessels, in addition to tumor cells, is maintained as a viable xenograft. In a previous study this model was successfully used to demonstrate that IL-12 treatment leads to the eradication of tumor cells in the lung cancer microenvironment, the persistence of CD45+ cells, and shrinkage of the xenograft over an extended observation period of 6 wk (8). Although the previous study focused on long term changes in tumor volume as an end point, in this study we monitored early changes in the tumor microenvironment in situ by immunohistochemistry and characterized the phenotype of the cytokine-responsive cells by four-color flow cytometry. In the present study we establish that IL-12 provokes the reactivation of quiescent CD4+ memory T cells that are constitutively present in the tumor microenvironment to proliferate, secrete IFN-γ, and kill tumor cells.
Materials and Methods
Primary non-small cell lung cancer tissue was obtained from the Tissue Procurement Facility of Roswell Park Cancer Institute, Veterans Administration Medical Center Pathology Laboratory, and Sisters of Charity Hospital. All specimens were obtained under sterile conditions and using institutional review board-approved protocols. Tissue was transported in DMEM/Ham’s F-12 medium for preservation until implantation. A histological diagnosis for each tumor specimen received was obtained anonymously.
Implantation of nondisrupted tissue into SCID mice
Surgical specimens of primary non-small cell lung cancer were examined grossly, and all normal and/or necrotic tissue were removed. Tumor tissue was cut into approximately cubic pieces, 0.3–0.5 mm on a side. C.B-17 scid/scid mice were obtained from the breeding colony and housed under specific pathogen-free conditions at the University at Buffalo. All mice used were between 8 and 10 wk of age and were sex-matched within each experiment. SCID mice were anesthetized with Avertin (0.4–0.6 mg/g; Sigma-Aldrich) and treated with one i.p. injection of TMβ-1 (mAb to the murine IL-2R β-chain) for depletion of NK cells (9). A small midline incision was made on the abdomen and was extended to create a s.c. pocket. One piece of tumor tissue was implanted into the pocket, and the incision was sealed with Nexaband Liquid (Burns Veterinary Supply). Tissue was allowed to engraft for 7–10 days before microsphere treatment.
Microsphere preparation and treatment of mice
Recombinant human IL-12 (Wyeth Research) was incorporated into biodegradable microspheres with BSA and polylactic acid at a concentration of 0.5 mg of recombinant human (rh)3 IL-12/200 mg of polymer as previously described (10). BSA control microspheres contain no cytokine. Mice bearing tumor xenografts were randomly divided into control and treatment groups of at least five mice 7–10 days after surgical implantation of fresh tumor tissue. All mice were treated with a single injection of BSA control or IL-12/BSA-loaded microspheres (8.0 mg/100 μl/mouse) into the implanted tumor.
Treatment of mice with oligodeoxynucleotides
Type A CpG oligodeoxynucleotide 2336 and control CpG oligodeoxynucleotide 2243 were purchased from Coley Pharmaceutical. Human lung tumor xenograft-bearing mice were treated with a single intratumoral injection containing 0.1 mg of CpG or control oligodeoxynucleotide/100 μl/mouse.
Serum collection and ELISA
On day 5 post-treatment with microspheres or 24 h post-treatment with oligodeoxynucleotides, mice were bled via the tail vein to obtain 150 μl of blood. Sera were collected by microcentrifugation of the blood sample at 14,000 rpm and were stored at −20°C until ELISA. Sandwich ELISA for detection of human IFN-γ in murine serum was performed as previously described (11, 12). In short, an anti-human IFN-γ mAb (M-700-A; Endogen) was used to coat 96-well plates. Mouse sera were added to the plate with biotinylated anti-IFN-γ (M-701-B; Endogen). Bound Ab was detected with streptavidin-conjugated HRP (Sigma-Aldrich) and 3,3′,5,5′-tetra-methylbenzidine (Kirkegaard & Perry Laboratories). Sandwich ELISA for detection of human IFN-α in murine serum was performed according to the manufacturer’s instructions using the BioSource kit. Emissions were measured on a Bio-Rad absorbance microplate reader (OD450), and results were analyzed against an rhIFN-γ standard using SigmaPlot 2001 software.
Histology and immunohistochemistry
Fresh surgical specimens and surgically excised, microsphere-treated xenografts were fixed in 10% neutral-buffered formalin overnight and processed by the University at Buffalo Histology Service Laboratory, where H&E staining was also performed. Anti-human Abs to the following markers were used for immunohistochemical staining of microsphere-treated xenografts: anti-CD45, anti-CD45RO, anti-CD45RA, anti-CD3, anti-CD20, anti-CD68, and anti-CD138. Abs used for immunohistochemistry were not cross-reactive with cells in the nonengrafted SCID spleen (data not shown). The Ab to Ig-like transcript receptor 3 (ILT3) was provided by M. Colonna (Washington University, St. Louis, MO) (10, 11, 12, 13). Images were taken with a Sony color video camera CCD SSC-S20 mounted on an Olympus BX40 light microscope using Snappy Video Snapshot 4.0 software (Play).
Human leukocytes were defined phenotypically through four-color flow cytometry of lung cancer xenograft cell suspensions. Single-cell suspensions were derived from surgically excised, treated xenografts by mechanical disruption of the tissue using a Teflon policeman to gently force cells through size 50 mesh. Cells were washed twice in DMEM/Ham’s F-12 medium supplemented with 10% FCS and counted by trypan blue exclusion.
The resultant cell suspensions were stained with multiple Ab panels, each panel containing four Abs bearing a different fluorochrome. The anti-human Abs included in the panels were anti-CD3, anti-CD4, anti-CD11a, anti-CD27, anti-CD28, anti-CD44, anti-CD45, anti-CD45RA, anti-CD45RO, anti-CD62L, anti-CD161, anti-CXCR3, anti-Vα24, and anti-Vβ11. α-Galactosylceramide (α-GalCer)-loaded CD1d tetramers, DMSO control CD1d tetramers, and an NKT cell line, used as a positive control, were provided by J. Gumperz (University of Wisconsin, Madison, WI) (14). At least one Ab in each panel, either CD45 or CD4, was shown to be human-specific and non-cross-reactive with murine cells and, therefore, was used for gating during analysis. Positive staining for CD45 (or for CD4 for studies evaluating the CD45 isoforms) was used to eliminate any cells of murine origin that may have contaminated the cell preparations and to reaffirm that analysis focused on the human lymphocyte population.
For nontetramer panels, cell suspensions were stained by pelleting 5 × 105 cells/panel in PBS. Mouse IgG was used to block FcRs before addition of the panel. Stained cells were washed in PBS and fixed in 2% formaldehyde. Tetramer staining was performed as previously described (14). Briefly, cell suspensions were blocked with nonfluorescent UPC10 tetramer, MOPC21, IgG1, OVA, and NaN3. The tetramer suspension and three-Ab panel, to select CD4+ human T cells, were added. Stained cell suspensions were washed in FACS buffer and fixed in 2% formaldehyde. All data were collected on a FACSCalibur flow cytometer (BD Biosciences; University at Buffalo and Roswell Park Cancer Institute) and analyzed in our laboratory using WinList software. Data were analyzed by first gating on CD45+ human cells, and the lymphocyte population was identified by forward and side scatter. For the nontetramer panels, at least 20,000 events were collected. For the tetramer panels, a minimum of 50,000 lymphocyte events were collected.
Tumor cell eradication is observed shortly after rhIL-12 administration
Human, primary, non-small cell lung cancer biopsy tissue can be implanted s.c. into SCID mice in small, nondisrupted pieces, resulting in the establishment of a viable tumor xenograft. The tumor microenvironment is structurally and functionally preserved and includes human tumor-associated leukocytes, stromal cells, and blood vessels in close proximity to the tumor cells. Using this xenograft model, we have previously shown that xenograft volume decreases over a 6-wk period after treatment with biodegradable, IL-12/BSA-loaded microspheres (8). In the current report we have evaluated the more immediate effects of a single intratumoral IL-12 microsphere treatment on the tumor microenvironment as an intact entity through histological analysis. As shown in Fig. 1, as early as day 5 after treatment, differences were observable between the IL-12-treated xenograft and the control-treated tissue (Fig. 1, top). The inflammatory cell population was much more numerous in the IL-12-treated xenograft than in the BSA control xenograft. Additionally, there appeared to be weakened intercellular junctions, movement of human leukocytes into the tumor parenchyma (asterisk), and more prominent apoptotic bodies in the tumor beds (arrows) after IL-12 treatment, compared with the tightly nested, viable tumor cells (with mitotic figures (arrowheads)) found in control xenografts. By 14 days, local and sustained release of rhIL-12 had radically changed the appearance of the tumor microenvironment; few tumor cells remained, and an expanding inflammatory cell population had effaced the majority of the xenograft (Fig. 1, bottom). The inflammatory cells in the IL-12-treated xenografts were shown to be human by immunohistochemical staining for HLA class I, and minimal murine cell infiltration was observed for up to 3 wk postengraftment (data not shown). In contrast, the BSA control microsphere-treated xenograft was composed primarily of tumor cells with fibrous connective tissue and few human inflammatory leukocytes remaining. Interestingly, 2–3 wk after implantation, the xenografts in the BSA control and IL-12/BSA-loaded microsphere-treated groups were of similar size on gross examination despite their radically different compositions on microscopic inspection. These data establish that the local release of IL-12 from biodegradable microspheres activates a cascade of inflammatory events within days of treatment and leads to an expansion of the human inflammatory reactive tissue within the tumor microenvironment that culminates in tumor cell eradication in a far shorter period of time than was previously appreciated.
Presence of activated T cells in the microenvironment
To identify the populations of human inflammatory cells responding to IL-12 and to establish their distribution within the nondisrupted tumor microenvironment, fixed tissues were sectioned and stained by immunohistochemistry. Fourteen days post-treatment with IL-12, there were increased numbers of CD3+ cells in the cytokine-treated xenograft compared with the control-treated tissue. In this study (Fig. 2, top) we demonstrated that the CD3+ T cells are diffusely distributed throughout the xenograft, suggesting that they are mobilized by IL-12 to infiltrate the tumor-rich regions. These infiltrating T cells are primarily CD45RO+ (Fig. 2, bottom), which is consistent with the phenotype of an activated or memory T cell. In contrast, relatively few naive T cells (CD45RA+) were identified in either the IL-12-treated or control-treated xenografts (data not shown).
Presence of other human leukocytes in the tumor microenvironment
In addition to activated T cells, other human inflammatory leukocytes were present in the tumor microenvironment 14 days after treatment with IL-12. However, in contrast to the diffuse presence of T cells across the entire xenograft, CD20+ B cells (Fig. 3, top row) and CD68+ tissue macrophages (Fig. 3, middle row) were fewer in number and located in more discrete foci. After treatment with IL-12, the macrophages identified by anti-CD68 were more numerous and larger in size compared with the control-treated xenograft (Fig. 3, middle row, arrowheads). Additionally, cells with plasma cell morphology (i.e., large lymphocyte, eccentric nucleus, prominent Golgi complex, and rough endoplasmic reticulum) were present in most IL-12-treated xenografts, and the majority of these cells stained positively for CD138, which identified these cells as Ig-producing plasma cells (Fig. 3, bottom row). Each of these cell types was absent or present in only small numbers in the BSA-treated control xenografts. We conclude that IL-12 treatment sustains B cells, tissue macrophages, and plasma cells in the tumor microenvironment.
Plasmacytoid dendritic cells in the tumor microenvironment
In light of the observation that a small number of cells are present in the tumor microenvironment that are morphologically similar to plasma cells but do not stain with the anti-CD138 Ab, we first examined pre-engraftment human lung tumors for the presence of plasmacytoid dendritic cells. Similar to plasma cells, these cells have eccentric nuclei and prominent Golgi apparatus, but upon activation they produce large quantities of IFN-α, rather than Ig, and are capable of processing and presenting Ag to T cells. As shown in Fig. 4, a small percentage (0.32% of the total cells) was identified by four-color flow cytometry to be plasmacytoid dendritic cells. These low side scatter cells were positive for BDCA-2, CD123, and ILT3 and negative for the lineage markers CD3, CD14, CD20, and CD56. This phenotype has been established to define plasmacytoid dendritic cells (13, 15, 16). From these results we conclude that plasmacytoid dendritic cells are present in non-small cell lung tumors obtained from patients postoperatively.
To determine whether these cells were sustained in the tumor xenograft after implantation into SCID mice, tissue sections were evaluated by immunohistochemistry using an Ab that has been shown to identify plasmacytoid dendritic cells in formalin-fixed tissues in situ when used in conjunction with cytomorphological examination (13). As illustrated in Fig. 5, a small proportion of cells with plasmacytoid morphology were positively identified.
To substantiate the presence of plasmacytoid dendritic cells in the lung tumor xenograft, we analyzed single-cell suspensions from explanted tumor xenografts by four-color flow cytometry, as we did for the original lung tumor immediately postsurgery. Between 0.2 and 3.0% of the human CD45+ cells in the lung tumor xenograft demonstrated a phenotype consistent with that of a plasmacytoid dendritic cell, i.e., BDCA-2+, CD123+, HLA-DR+, and CD3−. Furthermore, treatment of the lung tumor xenografts with CpG oligodeoxynucleotides (but not control oligodeoxynucleotides), induced IFN-α, which could be identified intracytoplasmically in the plasmacytoid cells by flow cytometry (Fig. 6). IFN-α was produced in quantities that could be detected in the sera of the human lung tumor xenograft-bearing SCID mice by ELISA (0.400–0.600 ng/ml) 24 h after stimulation with CpG by a single intratumoral injection. In contrast, no human IFN-α was detected in the sera of mice treated with control oligodeoxynucleotides. We conclude that plasmacytoid dendritic cells are present in non-small cell lung tumors and are preserved and functionally responsive in the tumor microenvironment in our chimeric model.
Enhanced inflammatory response observed in IL-12-treated xenografts correlates with increased serum levels of IFN-γ
We previously established that treatment of lung tumor xenografts with IL-12/BSA-loaded microspheres led to levels of human IFN-γ that were detectable in the sera of these mice and were necessary for tumor suppression over an extended period (8). In this study we show that the levels of human IFN-γ detected in sera after IL-12 treatment correlate with increased numbers of human leukocytes in the tumor microenvironment within 5 days of microsphere treatment. Tumor xenografts were removed from SCID mice 5 days post-treatment with IL-12/BSA-loaded or BSA control microspheres. H&E-stained sections of these xenografts clearly demonstrate many more inflammatory cells in the IL-12-treated xenograft compared with the control xenograft. This difference in the presence of inflammatory cells directly correlates with the amount of human IFN-γ detected in the sera of the mice bearing these xenografts, a >200-fold difference (data not shown). Although the absolute amount of IFN-γ varies from one tumor to the next, an increase in human IFN-γ levels is detected in SCID sera 5 days post-treatment with IL-12/BSA microspheres compared with BSA control microspheres in 17 of 19 tumors (Fig. 7). Thus, despite patient-to-patient variability in the different histologic types of non-small cell lung tumors, including squamous cell carcinoma, adenocarcinoma, and alveolar carcinoma, and various stages of differentiation from well differentiated to poorly or undifferentiated, a measurable and reproducible end point is available. We conclude that despite the variability among patient tumors in terms of histologic grade and stage of tumor development, the local and sustained release of IL-12 leads to an expansion of the human leukocytes constitutively present in the tumor microenvironment within 5 days of treatment and results in detectable levels of human IFN-γ in the sera of xenograft-bearing SCID mice that can be used as a parameter for comparison among patient tumors.
Expansion of CD45+ human leukocytes after IL-12 treatment
To examine quantitatively how the leukocyte population was affected by the local and sustained release of IL-12, control BSA-treated and IL-12-treated human lung tumor xenografts were surgically removed from SCID mice and mechanically disrupted, and human leukocytes in the resulting single-cell suspensions were phenotyped by FACS analysis of human CD45+ cells. At 7 days after treatment with IL-12 microspheres, there was an obvious increase in the number of lymphocytes staining positively for CD45 compared with the BSA control microsphere-treated xenografts (Fig. 8). As expected, a variation in the increase in human CD45+ cells, ranging from 2- to 10-fold, was observed in 10 different patient tumors after treatment with IL-12/BSA-loaded microspheres. These variations were consistent and predictable due to the varying numbers of inflammatory cells present within the original tumor biopsy tissue pre-engraftment, as demonstrated by H&E staining of tissue sections (data not shown). However, comparison of similarly treated xenografts, i.e., IL-12 or BSA control microsphere-treated grafts, derived from the same original tumor biopsy tissue showed little variability.
Phenotypic characterization of xenograft leukocytes after IL-12 treatment
To further define the human CD45+ leukocyte population in the treated xenograft, we used four-color flow cytometry. A multitude of cell surface markers have been used by others to phenotypically define effector memory T cells (17, 18, 19, 20, 21). In addition to those traditionally used to describe activated and memory cells, such as CD45RO, the up-regulation of markers of activation such as CXCR3 and CD44 and the down-regulation of other markers, CD27 and CD62L, for example, were also examined. The xenograft-derived human leukocytes were stained for cell surface markers to CD3, CD4, CD27, CD28, CD45RO, CD45RA, CXCR3, CD62L, CD44, CD11a, and the IL-12R (β1 subunit) in panels composed of four Abs labeled with different fluorochromes for evaluation by flow cytometry. Because flow cytometric analysis of IL-12-treated primary lung tumor xenografts previously identified CD4+ T cells as the prominent producer of IFN-γ, which was shown to be a necessary mediator in tumor eradication (8), we elected to focus on the CD4+ cell population (8, 22, 23). In each experiment the cells were gated on human lymphocytes, as defined by forward and side scatter and by the human leukocyte marker, CD45. Analysis of xenografts derived from six different original patient tumors at 7 and 14 days post-treatment with IL-12/BSA-loaded microspheres revealed that the predominant CD4+CD45+ human leukocyte population was positive for CD3 (42.86–70.72%) and CXCR3 (41.82–61.81%) and negative for CD62L. In terms of CD45 isomer expression, the CD4+CD3+ T cells were positive for CD45RO (68.31–98.98%) and negative for CD45RA (Fig. 9). Additional markers were similarly examined in subsequent experiments with xenografts derived from four different primary lung tumors, demonstrating that the majority of the CD4+CD45RO+ T cell population was also positive for CD28, CD11a, CD44, and the β1 subunit of the IL-12R, and a subset of these cells was positive for CD161. In comparison with the IL-12-treated xenografts, the CD4+ T cells in the BSA control-treated xenografts were not expanded in the absence of IFN-γ and were eventually eliminated from the xenograft.
The phenotype of the CD4+ T cell is consistent with that of an effector memory T cell. Of notable interest was the expression of CD161, present on a subset of the CD4+ cell population (25.89–46.03%). The expression of CD161 is up-regulated in response to IL-12 (24) and has been observed on both memory T cells and NKT cells (25). This observation together with multiple reports identifying a role for NKT cells in the IL-12-induced eradication of tumors in murine models (25, 26, 27, 28) led us to question whether this CD4+ subpopulation observed in the treated tumor xenograft consisted of NKT cells.
NKT cells do not constitute a significant effector cell population after rhIL-12 treatment
Invariant NK-like T (NKT) cells, which are CD1d restricted and bind α-GalCer as a ligand, share some surface markers with memory T cells, including CD3, CD4, and CD161 (29, 30). The presence of NKT cells in the xenografts was evaluated by two methods: 1) simultaneous expression of both the Vα24 and Vβ11 TCR chains and 2) the ability to bind the α-GalCer-loaded CD1d tetramer.
T cells that coexpress Vα24 and Vβ11 comprised a very small proportion of the total human leukocyte population in the tumor xenograft (1–3%). When the xenograft-derived cell suspensions were evaluated with the α-GalCer-loaded CD1d tetramers, no binding was observed compared with DMSO-loaded control tetramers. Comparatively, >90% of the NKT-positive control cells (a human NKT cell line) bound the α-GalCer-loaded CD1d tetramers. Collectively, these results demonstrate that the NKT cell is not a predominant cell in the lung tumor xenograft and that the primary cell responsible for the IL-12-induced antitumor effects is a CD4+ effector memory T cell.
In this report we have examined the cellular composition of human lung tumor xenografts in SCID mice by immunohistochemistry, early in the response to IL-12. Evaluation of the lung tumor xenografts after IL-12 treatment revealed that the inflammatory cell population was retained and expanded, and tumor cells apoptosed and were largely eliminated within 2 wk. Although the majority of the inflammatory cells in the IL-12-treated xenografts were CD3+ T cells that were distributed throughout the tumor microenvironment, other inflammatory leukocytes present in small foci were also identified, including CD20+ B cells, CD68+ tissue macrophages, and CD138+ plasma cells. In contrast, 2 wk after the BSA control treatment, xenografts were composed of large, viable-appearing tumor cells with minimal necrosis and were nearly devoid of inflammatory cells. A portion of the tumor-associated human leukocytes present at the time of implantation migrate out of the xenograft and have been detected in the spleen and lung of xenograft-bearing mice (data not shown). It is possible that apoptosis may also contribute to the loss of leukocytes in the xenograft.
The local and sustained release of rhIL-12 in the microenvironment of human lung tumor xenografts results in the expansion primarily of a population of human T cells that are constitutively present in the microenvironment of most human lung tumors. Although we previously showed that the antitumor effect of IL-12 was dependent upon CD4+ cells (8), we now present a more comprehensive phenotypic characterization of this cell. The T cell population was found to display a phenotype consistent with that of a CD4+ effector memory T cell, i.e., positive for CD3, CD4, CD45RO, CXCR3, CD28, CD44, CD11a, and IL-12R (β1 subunit), and was negative for CD27, CD45RA, and CD62L. Of particular interest is the pattern of chemokine receptors expressed on CD4+ T cells in the IL-12-treated lung tumor xenograft. The homing receptor L-selectin (CD62L), which is present on the central memory T cell population found in lymph nodes (31), is not expressed on the responding effector memory T cells. On the contrary, we did observe expression of the chemokine receptor CXCR3, which is preferentially expressed on Th1 cells and is known to bind monokine induced by IFN-γ and IFN-inducible protein-10 (reviewed in Ref. 32), two chemokines previously shown to be up-regulated in the tumor microenvironment in response to IL-12 treatment (8). CXCR3 is not present on central memory T cells (31). These chemokines and their receptors expressed on the cells may be significant factors that contribute to the homing of the T cells to the tumor microenvironment originally and could well be responsible for the retention of these cells in the tumor xenografts of IL-12-treated mice that were not retained in control-treated xenografts. These possibilities are currently being addressed.
A significant portion (26–46%) of the human leukocytes expressing a memory CD4+ T cell phenotype found in IL-12-treated xenografts express CD161, a phenotypic marker also found on NKT cells (29). Analysis by tetramer-based staining with a fluorescent α-GalCer-loaded CD1d tetramer and examination of coexpression of the Vα24 and Vβ11 invariant TCR chains demonstrated that NKT cells were not the primary cell population responding to exogenous cytokine treatment in our SCID xenograft model. We conclude that the CD4+ T cells constitutively present in the microenvironment of human lung tumors, which are mobilized by IL-12 to proliferate and kill tumor cells in the xenograft, are memory T cells. This is an important distinction, because CD1d-restricted NKT cells that home to peripheral sites of inflammation have been described as having a phenotypic profile similar to that of effector memory T cells and have been implicated in mediating IL-12-induced tumor killing (30).
Identification of the IL-12-responsive cell as a memory T cell raises the question of the mechanism leading to their retention in the tumor microenvironment. IFN-α has been described as having a role in the peripheral retention, low level proliferation, and inhibition of apoptosis of activated and memory T cells (33). Because plasmacytoid dendritic cells are responsible for large quantities of IFN-α (13, 34), we speculate that their presence in the tumor microenvironment may contribute to prolong the survival of memory T cells. Studies are currently underway to determine whether IFN-α produced by the tumor-associated plasmacytoid dendritic cells is required for CD4+ memory T cell retention and activation.
The persistence of T cells in peripheral nonlymphoid tissues, including lung, has been described after the resolution of viral and bacterial infections. These T cells have been identified as memory T cells (25, 35). It has recently been established that the contraction of the CD8+ effector T cell population responding to Listeria monocytogenes infection occurs even before the pathogen is completely cleared (36). It is suggested from such studies that a programmed contraction of effector T cells and the generation of memory T cells occur that are independent of the persistence or clearance of Ag. This may be important to prevent damage to host cells in the face of an infection that may not be resolvable even with prolonged encounter with more immune cells. Our working hypothesis is that a similar mechanism exists in the human lung tumor microenvironment where there is a persistence of the tumor, loss of effector T cells, and generation and persistence of effector memory T cells. The reason for the failure of these tumor-associated memory T cells to control tumor progression is widely debated. Recently, it was established that a costimulatory molecule, B7-H1, is expressed on human lung carcinomas, and that it has a negative regulatory effect on Ag-specific T cells (37). This negative regulatory pathway has also been implicated in the down-regulation of CD4+ memory T cells found in the synovial fluids of rheumatoid arthritis patients (38). Another possible explanation that has been offered to explain the nonresponsiveness of these cells is that there is a loss of signaling molecules that are required for activation via the TCR (39, 40, 41). However, it has recently been postulated that the loss of these molecules occurs as a normal consequence of the transition from naive or resting memory to effector cells as part of the maturation process (42). One possibility is that memory T cell development may be initiated, but in the face of continued exposure to Ag these cells fail to reset their TCR-mediated signaling potential, thereby persisting in a quiescent state until Ag is cleared from the microenvironment or until the cells are activated by an alternative pathway, such as that initiated by IL-12. It will be important to test these assumptions and to determine how the expression of signaling molecules may be altered after IL-12 microsphere treatment. Such findings would have important implications for immunotherapy against solid tumors, in that it may be possible to eradicate the tumor locally and induce an antitumor immunity systemically, either directly, by activating specific antitumor cells in the tumor microenvironment, or indirectly, by targeting stromal cells to initiate the release of chemokines and cytokines that would mobilize tumor-specific lymphocytes to kill tumor cells (43) and perhaps alter the expression of coregulatory molecules, resulting in a prolongation of the T cell response to the tumor (44).
We thank Dr. Jenny Gumperz for providing the α-galactosylceramide-loaded CD1d tetramers, the control tetramers, and the NKT cell line. We also thank Dr. Marco Colonna for providing the Ab to human ILT3, Dr. Michael McLean for assistance with tumor implantation, Robert Parsons and Jenni Loyall for technical assistance, and Drs. Hiroshi Takita, Todd Demmy and Harry Slocum and the Tissue Procurement Facility at Roswell Park Cancer Institute for providing us with lung tumor biopsy tissues.
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.
This work was supported by U.S. Public Health Service Grants CA96528, and CA108970/ CA79879, (to R.B.B.), and The John R. Oishei Foundation and Mary Kay Ash Foundation.
Abbreviations used in this paper: rh, recombinant human; α-GalCer, α-galactosylceramide; ILT3, Ig-like transcript receptor 3.