A major limitation of adoptive immunotherapy is the availability of T cells specific for both terminally differentiated tumor cells and their clonogenic precursors. We show here that marrow-infiltrating lymphocytes (MILs) recognize myeloma cells after activation with anti-CD3/CD28 beads with higher frequency than activated peripheral blood lymphocytes from the same patients. Furthermore, activated MILs target both the terminally differentiated CD138+ plasma cells and the myeloma precursor as shown by profound inhibition in a tumor clonogenic assay. The presence of antigen in the marrow microenvironment seems to be important for the maintenance of tumor specificity. Taken together, these results highlight the intrinsic tumor specificity of MILs and describe a novel approach for the generation of tumor-specific T-cell populations suitable for adoptive immunotherapy of multiple myeloma.
Adoptive immunotherapy has made significant inroads in the treatment of many malignancies over the past decades. In the allogeneic bone marrow transplant setting, the infusion of donor-derived lymphocytes imparts a significant graft versus tumor benefit in diseases such as chronic myeloid leukemia and multiple myeloma (1). However, the combination of the lack of tumor specificity, graft versus host disease, and the lack of suitable donors has limited the broader applicability of such an approach. In contrast, autologous lymphocytes specific for defined tumor antigens have been used in patients with melanoma and EBV-related lymphomas (2–5). Tumor regression and immune reactivity have been documented in certain individuals. However, relapse and the appearance of tumors with antigen loss variants highlight a major limitation of immune-based strategies targeting single antigens. Ideally, strategies using a polyclonal T-cell population of both CD4+ and CD8+ T cells targeting multiple tumor-associated antigens offer the possibility of inducing a more comprehensive and effective antitumor response.
Tumor-infiltrating lymphocytes represent such an approach to enhance tumor specificity. In melanoma tumor-infiltrating lymphocytes have shown a 10% to 20% durable response rate (6). Moreover, a recent report examining the correlation between the lymphocytic infiltration of ovarian carcinomas and improved clinical outcome suggests the presence of endogenous antitumor immunity. These results underscore the importance of the immune response in overall survival and highlight the uniqueness of the tumor microenvironment (7). Such observations clearly suggest a role for T cells in mediating antitumor immunity. Many barriers still exist to use the immune system as an effective cancer therapy. They include the ability to overcome tumor-specific T-cell tolerance, to efficiently isolate and expand tumor-specific T cells that maintain antigenic specificity, and to grow reasonable numbers of these cells to enable the generation of a clinically meaningful response when infused into the properly prepared host environment.
Tumor-specific, memory/effector lymphocytes could potentially be more effective than naïve lymphocytes in producing a measurable antitumor effect to a broad range of antigens (8). Lower levels of antigen for activation (9), more rapid activation kinetics (10), broader cytokine production (11), and specific trafficking patterns mediated by the up-regulation of surface markers such as VLA-4 (12) and CXCR-4 (13) are all hallmarks of memory T cells that make them ideal candidates for such therapeutic approaches.
Marrow-infiltrating lymphocytes (MILs) possess many of these unique features. Recent reports describe the ability of the bone marrow to serve as a site capable of priming naïve T-cell immune responses as well as enabling the persistence of antigen-specific T cells (14, 15). Human studies done in patients with breast cancer showed enrichment of memory (CD40RO+) CD4 and CD8 T cells (16). The persistence of marrow-infiltrating T cells with a memory phenotype suggests that this T-cell population may be more sensitive to reactivation and subsequent expansion as compared with peripheral blood lymphocytes (PBL). These cells, when used in adoptive immunotherapy, could improve the therapeutic outcome. In hematopoietic malignancies such as multiple myeloma, the bone marrow is the tumor microenvironment and thus represents a potentially unique site for the isolation and expansion of tumor-specific T cells. Moreover, the easy accessibility of marrow makes this an attractive source for tumor-specific T cells. Utilizing a T-cell activation platform consisting of anti-CD3/CD28 antibodies bound to magnetic beads that can polyclonally expand lymphocytes in vitro (17), we show that stimulated MILs display an activated phenotype and possess enhanced myeloma specificity as compared with activated peripheral blood-derived lymphocytes derived from the same patients. Furthermore, the ability of stimulated MILs to inhibit myeloma clonogenic precursors suggests that T cells generated by this approach target a broad range of tumor antigens present on both mature plasma cells as well as their precursors that underscore their potential efficacy in the treatment of multiple myeloma.
Materials and Methods
T-Cell Expansion. Bone marrow and peripheral blood samples were obtained from myeloma patients after having obtained informed consent using an Institutional Review Board-approved protocol. T-cell stimulation was done by adding anti-CD3/CD28 antibody-coated beads (Xcyte Dynabeads, Xcyte Therapies, Seattle, WA) to Ficolled bone marrow or peripheral blood suspended in serum-free conditions at 1 × 106 cells/mL in AIM-V (Invitrogen, Gibco, Carlsbad, CA), 200 μL/well, at a 3:1 bead to T cell ratio. The cells were cultured for 5 days in a 96-well round-bottomed plate at 37°C with 5% CO2. The beads were removed from the culture using a magnet. The cells were then replated at 200 μL/well in a 96-well plate for 2 days at 37°C, 5% CO2. Prior to phenotypic and functional analysis, fluorescence-activated cell sorting (FACS Calibur, BD Biosciences, San Diego, CA) analysis was done on bone marrow and peripheral blood by staining for CD3, CD4, CD8, CD25, CD45RA, CD45RO, and CD56 (BD Biosciences, PharMingen, San Diego CA). Cell Quest software was used to analyze the results.
CD33 and CD138 Selection. Frozen Ficolled bone marrow was thawed and washed thrice in HBSS. The cells were then incubated with either anti-CD138 or anti-CD33 microbeads (Miltenyi Biotec, Auburn, CA) for 15 minutes at 4°C to 6°C. The VarioMACS (Miltenyi) was used to isolate the cells as per protocol of the manufacturer.
Proliferation Assays. Media alone, CD138+ plasma cells, or CD33+ myeloid cells were incubated with either activated or unactivated MILs or PBLs at a 1:1 ratio. Cells were plated at 1 × 105 CD3+ cells/well (+/− stimulus) and incubated for 72 hours in a 96-well plate. They were then pulsed with 1 μCi of [3H]thymidine. Cells were harvested 18 hours later with a Packard Micromate cell harvester. [3H]Thymidine incorporation was measured as counts per minute (cpm) on a Packard Matrix 96 direct β-counter. Values are displayed as the mean ± SE cpm.
T-Cell Receptor Spectratyping. RNA was extracted from ∼5 million cells using Trizol (Invitrogen, Carlsbad, CA) and cDNA was prepared using the GeneAmp Gold RNA PCR Reagent Kit (Perkin-Elmer, Wellesley, MA). Spectratyping samples were prepared for each Vβ family using a Vβ-specific primer and a common HEX-labeled Cβ primer. The reaction components were 1× AmpliTaq Gold PCR buffer, 1.5 mmol/L MgCl2, 0.2 mmol/L deoxynucleotide triphosphates, 0.5 μmol/L Cβ-Hex primer, 0.5 μmol/L Vβ-specific primer, 1.25 units AmpliTaq Gold, and 1 μL cDNA template. The thermocycler conditions consisted of a 10-minute 95°C hotstart followed by 40 cycles of 25 seconds at 94°C, 45 seconds at 59°C, 45 seconds at 72°C, and completed with a 10-minute hold at 72°C. For each reaction, a mixture of 1 μL PCR product, 0.5 μL 400HD ROX size standard, and 12 μL Hi-Di Formamide was separated on an ABI 3100 Sequencer and analyzed in GeneScan 2.1 (18).
Transwell Migration Assays. Activated PBLs or activated MILs (1.5 × 105) were placed in the top well of a 4-μm 96-transwell plate (Millipore, Billerica, MA) and co-incubated at 37°C with various concentrations of stromal cell-derived factor-1 (SDF-1; Peprotech, Rocky Hill, NJ) in the bottom wells for 4 hours. Chemotaxis-mediated transmigration was determined by fluorescence-activated cell sorting analysis of CD3+ T cells in the bottom well (19).
Caspase-Cytoxicity Assay. This assay was done according to the protocol of the manufacturer (OncoImmunin Corp., Gaithersburg, MD). In brief, autologous CD138+ plasma cells (2 × 106/mL) were stained with the phycoerythrin-labeled Target Marker in AIM-V for 1 hour at 37°C in 5% CO2, then washed thrice with AIM-V. During this hour of incubation, activated PBL and activated MIL effector cells were harvested following a 5-day stimulation with anti-CD3/CD28 beads and subsequent 2-day rest and resuspended in AIM-V. The target and effector cells were mixed at the desired effector to target ratios. The labeled targets and effectors were added together in 5 mL round-bottomed polystyrene tubes and centrifuged at 1,250 rpm for 5 minutes and then resuspended in either 75 μL of the FITC-labeled caspase substrate or in wash buffer (as a non-substrate control), and incubated for 3 hours at 37°C in 5% CO2. Following incubation, all samples were washed, resuspended in wash buffer, and analyzed using flow cytometry. Cleavage of the fluorogenic caspase substrate identified target CD138+ cells undergoing cytolysis (20).
Myeloma Progenitor Outgrowth Assay. Bone marrow mononuclear cells from myeloma patients were depleted of CD34+, CD138+, and CD3+ cells as previously reported (21). The resulting negative fraction (5 × 105 cells/mL) was then incubated with varying T-cell populations for 24 hours and then plated in methylcellulose-containing lymphocyte conditioned media. Colonies were scored at 2 weeks and confirmed as myeloma colonies by CD138+ staining and light chain restriction based on the patients' initial plasma cell population. Data shown for each group is the total number of outgrowth colonies.
Preferential Expansion of Marrow-Infiltrating Lymphocytes Compared with Peripheral Blood Lymphocytes. In an effort to develop strategies to increase the efficacy of adoptive immunotherapy, we obtained and expanded MILs from multiple myeloma patients (Fig. 1A). MILs expanded to a greater extent than PBLs after a 5-day stimulation with the anti-CD3/CD28 beads (Fig. 1B). In the patients analyzed, the most pronounced difference was observed in CD4+ cells where the increase in the activated MILs exceeded the activated PBLs by more than 10-fold (activated MILs 27.6 ± 8-fold expansion versus activated PBLs 2.8 ± 1-fold expansion) whereas the absolute expansion of CD8+ cells was considerably less (activated MILs 15.3 ± 7-fold, activated PBLs 3.6 ± 2-fold). Enhanced proliferation of MILs over PBLs is consistent with a memory/effector phenotype. To test this, we examined the surface expression of CD40L and CD45RO at baseline and following activation. Interestingly, the baseline surface expression of both these markers was greater in MILs as compared with PBLs and the subsequent increase upon activation was more pronounced in the activated MILs (Fig. 1C). A major concern regarding the nonspecific stimulation of bone marrow from patients with multiple myeloma is the possibility of also expanding tumor cells. To address this, we stained for CD138+ plasma cells pre- and post-expansion and saw no increase in cell numbers after the 5-day stimulation (data not shown). In fact, the final cultures showed undetectable levels of CD138+ cells confirming that the in vitro stimulation of lymphocytes with anti-CD3/CD28 cultures did not expand tumor cells. Taken together, these data confirm the enhanced proliferative capacity of MILs as compared with PBLs in the same patients when the cells are cultured using the conditions described in Materials and Methods.
Activated Marrow-Infiltrating Lymphocytes Show a Skewed T-Cell Receptor Vβ Repertoire. To determine the effect of anti-CD3/CD28 bead stimulation on the T-cell repertoire of lymphocytes from the peripheral blood and marrow of patients with multiple myeloma, T-cell receptor Vβ spectratyping analysis was done on patients before and after bead activation. Unactivated PBLs and MILs both showed a similar degree of Vβ T-cell skewing with similar oligoclonal peaks observed in certain Vβ families. However, whereas activation with the anti-CD3/CD28 beads tended to normalize polyclonal Vβ T-cell repertoire in activated PBLs, oligoclonality persisted in the activated MILs (Fig. 2). These results show the preservation of the skewed T-cell repertoire in MILs following activation and expansion as compared with activated PBLs, possibly suggestive of maintenance and/or enrichment of T cells with heightened tumor specificity.
Activated Marrow-Infiltrating Lymphocytes Display Enhanced Antitumor Specificity Compared with Activated Peripheral Blood Lymphocytes. The propensity of MILs to retain a skewed T-cell repertoire profile might reflect the selective pressure maintained by the presence of tumor during bead activation. We thus sought to determine if functional differences existed between activated MILs and activated PBLs. Specifically, we evaluated the effect of in vitro activation with anti-CD3/CD28 beads on the tumor specificity of lymphocyte populations obtained from either compartment. PBLs and MILs were expanded and activated by adding magnetic beads conjugated to anti-CD3/CD28 antibodies at a 3:1 ratio of beads to CD3+ cells. The beads were added directly to the Ficolled bone marrow or peripheral blood mononuclear cells and removed after a 5-day stimulation. Following a 2-day rest, T-cell–mediated tumor specificity was assessed by examining the proliferative response to autologous CD138+ plasma cells. CD33+ myeloid cells or T cells alone served as controls. Proliferative responses of activated T cells were compared with unactivated T cells from both compartments (PBL and MIL). Whereas unactivated MILs showed minimal proliferation towards CD138+ cells, activated MILs showed significant tumor specificity with a 62-fold stimulation index, 121,692 (±11,916) cpm pulsed with CD138+ versus 1,918 (±75) cpm unpulsed activated MILs. In sharp contrast, activated PBLs showed only a 3.6-fold stimulation index of activated PBLs pulsed with CD138+ cells versus no antigen, 5,631 (±1,196) cpm versus 1,558 (±190) cpm), similar to the unactivated PBLs demonstrating a 3.4-fold stimulation index towards CD138+ cells versus no antigen (Fig. 3A). A major concern of directly activating MILs within the bone marrow microenvironment is the possibility of generating nonspecific T-cell responses to normal hematopoietic elements that would limit the applicability of this approach in a clinical setting. The absence of measurable T-cell reactivity towards autologous CD33+ myeloid cells over baseline shows the significant degree of tumor specificity that activated MILs possess. To determine if the antitumor effect was mediated via T cell receptor-major histocompatibility complex, engagement pan Class I and pan Class II antibodies were added to tumor specificity cultures. The addition of blocking Class I and Class II antibodies completely abrogated the proliferative response of activated MILs to autologous CD138+ cells (data not shown).
The Presence of Antigen during T-Cell Expansion May Be Critical in Maintaining the Tumor Specificity of Activated Marrow-Infiltrating Lymphocytes. One major difference between the bone marrow and peripheral blood compartments of patients with multiple myeloma is the significantly greater tumor burden in the bone marrow compared with the peripheral blood. To evaluate the contribution of tumor within the bone marrow in enhancing tumor reactivity in the activated MILs, MILs were selected and added to the T-cell–depleted peripheral blood whereas peripheral blood T cells were selected and added to the T-cell–depleted bone marrow at greater than 90% purity (data not shown). With this experimental design, MILs would be activated and expanded in the absence of any elements present within the marrow microenvironment. In contrast, if the marrow contained elements solely responsible for imparting and/or maintaining antigen specificity, we would expect to see augmentation of tumor specificity of the PBLs activated in the bone marrow environment. These groups were activated with anti-CD3/CD28 beads as previously described, rested for 2 days, and placed in a proliferation assay with autologous CD138+ cells, autologous myeloid CD33+ cells, or no antigen. The T-cell–depleted bone marrow reconstituted with PBL-derived T cells showed a similar degree of proliferation as the PBLs activated in the blood. In contrast, the group containing T-cell–depleted peripheral blood reconstituted with marrow-derived T cells (MILs) proliferated less than the activated MILs group (Fig. 3B).
We then sought to determine whether the reduced tumor specificity of bone marrow lymphocytes stimulated in the T-cell–depleted peripheral blood was due to a loss of critical marrow-derived elements or simply dependent on the absence of CD138+ cells. The anti-CD3/CD28 beads were added to the bone marrow cells and magnetically removed after an initial incubation period to isolate the bead-T cell complex. These enriched T cells were expanded for 5 days and rested for 2 days as before. MILs were analyzed for their ability to proliferate to autologous tumor (CD138+), autologous myeloid cells (CD33+), or no antigen. Maximal tumor-specific proliferation was achieved in the group expanded and activated in the presence of tumor, but tumor-specific proliferation was not completely lost in the MIL group lacking the tumor antigen (Fig. 3C). Taken together, these data show that the presence of antigen may be critical during anti-CD3/CD28 activation to maintain tumor specificity and highlight the importance of the bone marrow in myeloma in creating a permissive environment for the generation and maintenance of T cells with heightened tumor specificity.
Activated Marrow-Infiltrating Lymphocytes Show Increased Stromal Cell-Derived Factor-1–Mediated Transwell Migration. Tumor-specific T cells can possess high affinities for their cognate antigen, but the ability to generate an effective immune response requires direct contact with the tumor in the proper context. One factor that may play a role in facilitating such interactions is SDF-1. It is a known bone marrow chemoattractant (22) and its cognate receptor on T cells is CXCR-4 (23). CXCR-4 expression was much greater on activated MILs than on activated PBLs (Fig. 4A). To examine whether differences in SDF-1-mediated responsiveness existed between MILs and PBLs, we did a transwell experiment. Activated MILs or activated PBLs were placed in the upper well of a transwell plate with SDF-1 in the bottom well at the indicated concentrations and harvested after 4 hours of incubation. The transmigration of activated MILs significantly exceeded that of activated PBLs when incubated with 100 ng/mL of SDF-1 (Fig. 4B). No significant differences were seen using these conditions with the unactivated T cells (data not shown). The significant trans-migratory ability of activated MILs and their pronounced up-regulation of CXCR-4 could play a critical role in facilitating trafficking of these activated T cells to marrow/tumor microenvironment that may be critical in establishing and/or maintaining tumor specificity.
Activated Marrow-Infiltrating Lymphocytes Exhibit Greater Tumor-Specific Cytotoxicity as Compared with Activated Peripheral Blood Lymphocytes. To determine whether these cells effectively kill tumor, we sought to analyze their cytolytic capabilities. We used a cytotoxic fluorometric assay that measures the cleavage of a cell-permeable fluorogenic caspase substrate in cells undergoing apoptosis to analyze the differences in cytolytic capabilities of the activated MILs compared with the activated PBLs (20). This assay was chosen instead of the traditional cytotoxicity assay because of the inability to effectively label autologous plasma cells with chromium. In addition, the ability to specifically examine plasma cell cytotoxicity as opposed to generalized, nonspecific cell killing enables a more accurate determination of T-cell–mediated tumor cell killing. Furthermore, caspase cleavage represents an early event in cell death and thus serves as a more sensitive marker of cytotoxicity than the classic chromium release assay. Lymphocytes obtained from both compartments (peripheral blood and marrow) from patients were stimulated for 5 days with the anti-CD3/CD28 beads at a bead to T cell ratio of 3:1, and rested for 2 days. Autologous CD138+ cells (targets) were selected for each patient and fluorescently labeled with the Target Marker, then co-incubated with either activated MILs or activated PBLs (effectors) in the presence of a fluorogenic caspase substrate. Prior titration studies showed maximal cytotoxicity at an effector to target ratio of 5:1. This effector to target ratio was used in subsequent studies. Following a 3-hour incubation the cells were washed and analyzed by flow cytometry. Cleavage of the caspase substrate increases the fluorescent intensity (in the FL-1 channel) of the dying CD138+ tumor cells (Fig. 5A). Activated MILs showed almost a 100-fold greater tumor-specific cytolytic function as compared with activated PBLs in an assay normalized for the same number of T cells (Patient 1: 22.3% activated MILs versus 0.26% activated PBLs; Patient 2: 51.5% versus 0.12%; and Patient 3: 42.3% versus 0.36%, respectively)(Fig. 5B). The addition of pan-HLA-Class I and Class II antibodies completely blocked the cytotoxic ability of activated MILs (data not shown). In the unactivated state, no cytotoxic activity was observed in PBLs and minimal activity in MILs (Fig. 5A). The absence of tumor-specific cytotoxicity of unstimulated lymphocytes shows the profound unresponsiveness present in this population and underscores the requirement for in vitro T-cell activation to generate tumoricidal activity.
Activated Marrow-Infiltrating Lymphocytes Are Potent Inhibitors of Plasma Cell Outgrowth in a Clonogenic Assay. Multiple myeloma is a B-cell malignancy marked by an increase of terminally differentiated plasma cells with minimal self-renewing capabilities. We recently showed that clonogenic cells in multiple myeloma resemble post-germinal center B-cells rather than the terminally differentiated plasma cells (21). For therapies to induce durable remissions they must significantly reduce both the terminally differentiated plasma cells as well as their self-renewing precursors. Due to activated MILs showing significant antitumor activity against plasma cells, we investigated whether they also inhibited the clonogenic outgrowth of myeloma precursors. Autologous bone marrow cells were co-incubated with PBLs, activated PBLs, MILs, or activated MILs, plated in methylcellulose, and evaluated for CD138+ light chain-restricted plasma cell colony formation. Activated MILs inhibited plasma cell colony formation by 86% (Fig. 6), whereas PBLs, MILs, and activated PBLs had significantly less activity against clonogenic myeloma outgrowth (27%, 38%, and 47%, respectively). Similar inhibition of clonogenic outgrowth was also observed using the two myeloma cell lines, H929 and RPMI 8226, where activated MILs inhibited plasma cell colony outgrowth by 100% in the H929 cell line and by 84% in the RPMI 8226 line (data not shown). Considering the potent antitumor response on plasma cell clonogenic precursors, we sought to analyze whether this inhibitory response would also extend to normal hematopoietic precursors owing to such a finding precluding the ability to use this approach therapeutically. Normal hematopoietic granulocyte-macrophage colony-forming unit outgrowth was unaltered with the addition of activated PBLs or activated MILs (data not shown), thus confirming the tumor specificity of activated MILs. Taken together, these data also show the greater antitumor activity of activated MILs over activated PBLs towards the plasma cell progenitors.
This is the first description of the activation and expansion of MILs and the demonstration of their greater antitumor specificity towards both mature plasma cells as well as their clonogenic precursors as compared with their peripheral blood counterparts. Importantly, several attributes of MILs make them suitable candidates for immunotherapy. Specifically, under the conditions described in this manuscript, they expand more rapidly upon stimulation than PBLs and often maintain a skewed T-cell repertoire upon activation, possibly suggesting augmented tumor specificity. Whereas the unactivated MILs show profound hyporesponsiveness toward autologous tumor, the ability to activate and expand T cells and markedly enhance their tumor reactivity argues against deletional tolerance as a presumptive mechanism mediating T-cell unresponsiveness in this setting. Furthermore, activated MILs show tumor specificity with little cross-reactivity towards nonmalignant hematopoietic elements, have a higher expression of CXCR-4, and possess a greater responsiveness to SDF-1, suggesting an increased migratory ability of MILs to the bone marrow. Taken together, these findings show the ability to activate and expand marrow-infiltrating T cells with a memory/effector phenotype that seem to target the broad range of tumor antigens present on both mature terminally differentiated plasma cells as well as their precursors and possess chemokine receptors that would seem to facilitate trafficking to the bone marrow compartment—features that would be necessary for maximizing antitumor immunity of adoptive immunotherapy.
In recent years, adoptive immunotherapy has shown measurable clinical activity yet many obstacles still exist to maximize its efficacy (3, 24, 25). For adoptive immunotherapy to be effective, T cells must overcome the intrinsic tolerogenic mechanisms that often limit immune responsiveness (6). Specifically they must be tumor-specific, capable of being expanded to clinically meaningful numbers, traffic to the tumor microenvironment upon infusion into the host, and kill the tumor upon its encounter. To date, multiple strategies have been attempted to increase the tumor specificity of adoptive immunotherapy (26–29). Whereas antigen-specific lines or clones have been used, they require the need to identify and isolate peptides or tumor antigens thereby limiting this approach to a finite number of known tumor antigens and increasing the likelihood of recurrent disease associated with antigen-escape variants.
Effective adoptive immunotherapy requires activated T cells with broad tumor specificity present in sufficient numbers to achieve a clinically measurable antitumor response. Such qualitative and quantitative requirements have been difficult to realize with many techniques employed to date. The development of the bead-based platform consisting of anti-CD3/anti-CD28 antibodies coupled to magnetic beads (17) has enabled activation and 300- to 500-fold expansion of T cells in a polyclonal manner. Whereas such technology addresses two major requirements for effective immunotherapy, (a) the ability to obtain adequate cell numbers and (b) the activation of lymphocytes, to overcome the unresponsiveness associated with tumor-induced tolerance, its clinical use has thus far been limited to PBLs (30, 31). In an effort to increase the tumor specificity of adoptive immunotherapy, we obtained and expanded lymphocytes residing within the bone marrow of multiple myeloma patients.
Activation and expansion of MILs was based on two previously reported phenomena: the enhanced tumor specificity of tumor-infiltrating lymphocytes (32) and the demonstration of tumor-reactive T cells in the bone marrow of patients with melanoma (33), breast cancer (34) as well as multiple myeloma—a disease in which the bone marrow also represents the tumor microenvironment (35). These previous studies describe the presence of tumor-specific cells in the bone marrow and possibly address their enhanced tumor reactivity as compared with peripheral lymphocytes. This is the first study demonstrating the ability to activate and expand MILs as a means of overcoming their unresponsiveness and significantly increasing their tumor specificity compared with activated PBLs. Our experiments show that the presence of tumor in the bone marrow may play a critical role in preserving the antigen specificity of activated MILs. Other putative components in the marrow microenvironment, which may contribute to maintenance of tumor reactivity, remain to be elucidated. Several hypotheses may explain the increased reactivity of activated MILs over activated PBLs. First, the persistence of antigen in the bone marrow may be essential for the maintenance of a memory response (36). CD3/CD28 bead activation may be reversing tolerance in the bone marrow T-cell population. Furthermore, in an experimental murine model using a lacZ-expressing tumor, the generation and persistence of a T-cell response was directly dependent on the presence of tumor (37). The observations in the murine model were confirmed in our system where the tumor specificity of activated MILs was clearly dependent on the presence of antigen during T-cell activation. Second, the bone marrow is a functional lymphoid organ capable of mounting both a primary immune response (14, 38) as well as secondary responses via reactive lymphoid follicles in the presence of danger signals (infection, inflammation, autoimmunity, and cancer; ref. 39).
T cells in myeloma patients show considerable skewing of the Vβ T-cell receptor repertoire. Such skewing suggests either the selective outgrowth of T cells with marked tumor specificity or results from the profound underlying T-cell defects characteristic of patients with a significant tumor burden (40). In the latter case, a benefit of polyclonal stimulation of PBLs with the anti-CD3/CD28 beads is the ability to restore a normal T-cell repertoire and thus correct any underlying T-cell defects (41). In contrast, if the oligoclonal expression of specific Vβ families reflects the presence of T cells with tumor specificity, activation and expansion of this pool of T cells with maintained antitumor activity and T-cell receptor repertoire skewing may be preferable. We show that PBLs normalized their Vβ T-cell repertoire upon activation and expansion with anti-CD3/CD28 beads whereas MILs maintained the Vβ restriction. Considering the enhanced tumor-specific response of activated MILs, their skewed T-cell repertoire may be suggestive of greater tumor recognition. As such, it may be important to conserve and possibly increase the degree of Vβ skewing during T-cell expansion.
The major observation of this study is that isolation and expansion of MILs with anti-CD3/CD28 beads generates potent antitumor activity and that the persistence of antigen during this expansion may be of significant importance in maintaining (and augmenting) the tumor specificity. Dhodapkar et al. (35) have also studied the role of MILs in myeloma patients. Similar to our findings, freshly isolated MILs or PBLs showed no activity upon stimulation with autologous tumor or tumor peptides. However, whereas that study saw no significant differences between T cells obtained from the peripheral blood and the marrow compartment in the enzyme-linked immunospot assay following 12 to 16 days of incubation with tumor-pulsed dendritic cells, a 10-fold greater antitumor response of activated MILs over activated PBLs was observed in our system in all assays examined. These discrepant results may be related to potency of anti-CD3/CD28 bead stimulation as compared with dendritic cell activation of MILs. What seems to be an increase in frequency of tumor-reactive T cells in the activated and expanded MILs cultures may reflect the breaking of tolerance and restoration of function of tumor-reactive T cells. Furthermore, our stimulation of MILs within the bone marrow microenvironment is another important factor that may explain these results.
Another mechanism that may account for the increased immune responsiveness in the bone marrow could be related to the preferential trafficking of tumor-specific T cells to the bone marrow. Activated MILs exhibited far greater transwell migration in response to SDF-1 than did activated PBLs (Fig. 4). Whereas this chemokine is a known bone marrow chemoattractant in addition to being the ligand for CXCR-4 expressed on T cells (42), it has also been implicated in the migration of myeloma cells to the bone marrow (43) and the establishment of bone marrow metastases of other tumors (44, 45). It is reasonable to hypothesize that this chemokine-mediated physical colocalization of tumor and T cells to the same compartment may be yet another mechanism by which the bone marrow microenvironment is enriched in the frequency of antigen-specific T cells.
A major requirement for clinically meaningful T-cell immunotherapy is the generation of T cells recognizing a broad spectrum of tumor-specific antigens that possess a measurable effector function. Multiple myeloma represents a disease in which the malignant plasma cell likely represents a terminally differentiated phenotype and not the clonogenic “stem cell.” An increasing body of literature has focused on targeting the cancer “stem cell” as a therapeutic intervention aimed at imparting a sustainable antitumor effect by eliminating the self-renewing source of tumor cells (46). As such, any therapy that produces long-term remissions is dependent on its ability to inhibit both the terminally differentiated and self-renewing cell populations. Activated MILs show significant anti-myeloma activity against both the terminally differentiated CD138+ plasma cells and their clonogenic precursors without affecting normal hematopoietic function. This suggests that activated MILs may target a broad range of tumor-specific antigens in a tumor-specific manner, potentially using both cytotoxic CD4+ and CD8+ cells, or possibly a small population of exceptionally specific and potent CD8+ CTLs. The depth of tumor recognition coupled to the absence of non-tumor-specific activity provides a safe and effective therapeutic approach for multiple myeloma.
In summary, these findings confirm the presence of tumor-specific memory T cells within the marrow that show enhanced antitumor efficacy upon activation and seem to possess features that enhance their trafficking to the bone marrow compartment (also the tumor microenvironment). Such features would likely augment the efficacy of adoptive immunotherapy and provide the rationale for using activated MILs in a therapeutic setting. Furthermore, we describe a method for the simple and rapid expansion of T cells with specificity towards a broad range of antigens present not only on the parental tumor but also on its precursor with minimal reactivity towards normal hematopoietic elements using clinically available technology, and highlight several critical requirement for maintaining tumor specificity during T-cell expansion. The clinical implementation of this approach in the proper setting alone or in combination with approaches to further augment tumor specificity such as tumor vaccination may improve our understanding of immunotherapy and increase its therapeutic efficacy.
Grant support: NIH grant CA15396.
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 thank the A.L. James Hanford III Foundation for their generous support of this project, Jimiane Ashe and Alice Long for their excellent technical assistance, Lisa Ferraro for her assistance in the preparation of the manuscript, and Dr. Drew Pardoll for his critical review of the manuscript.