Lymphocyte Activation Gene-3 Fusion Protein Increases the Potency of a Granulocyte Macrophage Colony-Stimulating Factor–Secreting Tumor Cell Immunotherapy

An effective antitumor response is associated with the induction of cellular and/or humoral antigen-specific immune responses (1–3). Both arms of the immune response require effective antigen presentation and T-cell priming (4). Dendritic cells are recognized as one of the most potent antigen-presenting cells and naive immature dendritic cells are characterized by high endocytic activity and low T-cell activation potential (5). Once dendritic cells have engulfed antigen, they mature and present antigen-derived peptides in the context of MHC molecules to naive T cells. In addition, mature dendritic cells up-regulate cell surface receptors such as CD40, CD80, and CD86 that are necessary for priming effective T-cell responses (6, 7). Antigen presentation by dendritic cells without costimulatory signals leads to the induction of T-cell anergy or tolerance (7, 8).

Granulocyte macrophage colony-stimulating factor (GM-CSF)-secreting tumor cell immunotherapy stimulates the recruitment of dendritic cells to the immunotherapy injection site, a location of high abundance of immunotherapy-derived antigens (9). Antigen uptake leads to the maturation and activation of dendritic cells, which in turn activate naive T cells after migrating to draining lymph nodes (10). High numbers of activated dendritic cells and T cells are found in the draining lymph nodes of animals injected with GM-CSF–secreting tumor cell immunotherapy, which correlate with potent antigen-specific T-cell responses and prolonged survival (11, 12).

The lymphocyte activation gene-3 (LAG-3 or CD223) is a protein related to CD4 at the gene and protein level (13, 14). LAG-3, which is expressed by CD4⁺ and CD8⁺ T cells, binds a nonpolymorphic region of class II with higher affinity than CD4 (15, 16) and is associated with the CD3/T-cell receptor complex at the cell surface (17, 18). Recombinant soluble LAG-3Ig fusion protein (sLAG) binds MHC class II molecules with higher avidity than CD4 (16). sLAG increases the capacity of antigen-presenting cells to induce T-cell responses in vitro and in vivo (16, 19, 20) and has been reported to act as an adjuvant, driving immune responses toward a T helper type 1 (Th1) response (21, 22). Due to these properties, sLAG was evaluated in combination with GM-CSF–secreting tumor cell immunotherapy with the objective of triggering potent antigen-presenting cell activation via the MHC II molecule and generating a more pronounced T-cell response that could potentially translate into prolonged survival of tumor-bearing animals.

In this report, data are presented showing that, in the poorly immunogenic and highly aggressive B16 murine melanoma model, tumor-bearing animals receiving GM-CSF–secreting tumor cell immunotherapy in combination with sLAG have prolonged survival compared with animals receiving either therapy alone. Importantly, the survival advantage of animals treated with the combination therapy was observed when therapy was initiated as late as 7 days after tumor challenge, a time point when GM-CSF–secreting tumor cell immunotherapy as monotherapy provides limited therapeutic benefit in this tumor model. In addition, greater numbers of IFNγ-secreting CD8⁺ T cells were detected in combination therapy–treated animals and a more rapid infiltration of activated effector CD8⁺ T cells was observed in the tumor. Most importantly, Th1 and Th2 type-specific cytokines were generated by splenocytes from animals treated with the combination therapy. An enhanced tumor-specific humoral response was also detected compared with animals treated with either monotherapy. These data suggest that the enhancement of antitumor efficacy by GM-CSF–secreting tumor cell immunotherapy plus sLAG correlates with the activation of both cellular and humoral antitumor immune responses.

Mice and cell lines. Female C57BL/6 mice (Taconic) were purchased and maintained according to American Association for Accreditation of Laboratory Animal Care guidelines. Efficacy and mechanism studies were initiated with mice between 8 and 12 wk of age. Study designs were approved and done according to the guidelines of the Cell Genesys Animal Use and Care Committee.

The B16F10 melanoma cells were purchased from the American Type Culture Collection. The generation of the retrovirally transduced GM-CSF–secreting cell line has been previously described (23). These GM-CSF–secreting cell lines were further retrovirally transduced to express the Kd haplotype (GMKd) to generate an immunotherapy that expresses an allogeneic component as is used in the majority of clinical studies. GMKd generates ∼150 ng of murine GM-CSF/10⁶ cells/24 h. Cells were maintained in DMEM (Hyclone) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone), 2 mmol/L l-glutamine, and 1× penicillin/streptomycin (JRH) in a humidified incubator with 5% CO₂ at 37°C.

To generate the two ovalbumin-expressing cell lines, B16.ova (from B16F10) and GM.ova (from GM-CSF–secreting B16F10 producing 150 ng of murine GM-CSF/10⁶ cells/24 h), a CD16 ovalbumin LAMP-1 fusion protein was excised and cloned into a third-generation lentiviral transfer vector (24). Lentiviral vectors were generated by transient transfection of 293T cells. The lentiviral vector–transduced B16.ova and GM.ova cells were stained using an ovalbumin-specific antibody conjugated to biotin (U.S. Biological), detected with a streptavidin-conjugated phycoerythrin secondary antibody (BD PharMingen), and sorted for ovalbumin expression using a MoFlo (Cytomation, Inc.).

Recombinant soluble LAG-3Ig fusion protein. sLAG binds MHC class II molecules with higher avidity than CD4. mLAG-3Ig (sLAG) diluted in PBS plus 50 mmol/L arginine plus 50 mmol/L glutamine was supplied by Immutep S.A. Stock solutions of sLAG were kept at −80°C and thawed on ice before use.

Tumor model studies. In the B16 model, mice (n = 10 per group) were inoculated with 2 × 10⁵ live B16F10 cells by s.c. injection at a dorsal site and immunized with 1 × 10⁶ irradiated allogeneic GM-CSF–secreting B16 cells (GMKd) by s.c. injection at a ventral site at designated time points. sLAG was administered s.c. at the immunotherapy site. Animals were monitored for the formation of palpable tumors twice weekly and sacrificed when tumors became necrotic or estimated to exceed 1,500 mm³ by palpation.

In the B16.ova model, mice were inoculated with 2 × 10⁵ live B16.ova cells by s.c. injection at a dorsal site and immunized with 1 × 10⁶ irradiated GM.ova cells by s.c. injection at a ventral site 3 d after immunotherapy. sLAG was administered s.c. at the immunotherapy site. Animals were euthanized at indicated time points (n = 5 per group) for immune monitoring experiments.

Evaluation of tumor-infiltrating lymphocytes. At designated time points, tumors were removed from mice and digested in 1 mg/mL collagenase (Sigma) and 0.1 mg/mL DNase (Sigma) for 1 h at 37°C. Dissociated cells were filtered through a 0.3-μm filter and leukocytes were positively selected using anti-CD45 MACs beads according to the manufacturer's instructions (Miltenyi Biotec). Enriched leukocytes were directly stained with conjugated antibodies (BD PharMingen) for phenotype characterization by flow cytometry.

Cell characterization and flow cytometry. Cells from the draining lymph nodes (axillary, lateral axillary, and superficial inguinal) and spleens were collected and mechanically dissociated using glass slides. Cells were counted and stained with conjugated antibodies purchased from BD PharMingen. Flow cytometry acquisition and analysis was done on a FACScan apparatus using CellQuest Pro obtained from BD Biosciences.

Enzyme-linked immunospot assay. Antigen-specific responses were enumerated by an IFNγ enzyme-linked immunospot (ELISPOT) assay (R&D Biosystems) according to the manufacturer's instructions. Briefly, 96-well filtration ELISPOT plates (Millipore) were coated with the specified amount of capture antibody in 100 μL of reagent diluent for 2 h at 37°C. Plates were washed twice with wash buffer (PBS with 0.5% Tween 20) and blocked for 2 h at room temperature in blocking buffer containing PBS supplemented with 1% bovine serum albumin and 5% sucrose. Erythrocyte-depleted splenocytes (5 × 10⁵) were plated and incubated for 48 h at 37°C, 5% CO₂ with 1 × 10⁴ irradiated B16 whole cells or with 1 μmol/L of the K^b binding peptide derived from Trp2 (SVYDFFVWL) or gp100 (KTWGKYWQV; Genmed Synthesis). ELISPOT plates were developed according to the manufacturer's specifications. Wells containing medium and splenocytes only served as negative control. Spots were enumerated by an automated plate scanning service from Cellular Technology Ltd.

In vivo CTL assay. Mice were inoculated with B16F10 tumor cells on day 0 and immunized with irradiated GM-CSF–secreting tumor cells on day 3. On days 5 to 7, mice were injected s.c. at the immunotherapy site with 0.1 μg of sLAG. At indicated time points, mice were injected with syngeneic splenocytes labeled for 15 min at 37°C with either 5 μmol/L carboxyfluorescein diacetate succinimidyl ester (CFSE; CFSE^hi cells; Molecular Probes) or 0.5 μmol/L CFSE (CFSE^lo cells). CFSE^hi cells were pulsed with antigen-specific peptide at 1 μg/mL for 1 h at 37°C. Unpulsed CFSE^lo cells served as an internal control. Splenocytes were analyzed 18 h after injection for the detection and quantification of CFSE-labeled cells by flow cytometry. Naive animals with no CTL activity display two equal CFSE^hi and CFSE^lo peaks at a 1:1 ratio, whereas experimental animals that contain lytic T cells specific for the peptide that has been loaded onto the splenocytes display a reduction in the CFSE^hi population. The formula to calculate percent specific killing is as follows: 100 × [(naive m2/m1) / (experimental M2/M1)] = % lysis, where m1 = CFSE^hi and m2 = CFSE^lo.

Cytokine assay. Irradiated B16F10 cells were cocultured with splenocytes isolated from experimental animals for 2 d. Evaluation of cytokine secretion of cocultured supernatant was done using Th1/Th2 and inflammatory cytometric bead array kits (BD PharMingen) by flow cytometry according to the manufacturer's instructions and analyzed by BD CBA system (BD PharMingen).

Anti-ovalbumin ELISA. Ninety-six–well Nunc plates were coated with 1 mg/mL ovalbumin (Sigma) in PBS overnight. Plates were blocked for 2 h at room temperature. Serum dilutions of 1:500 and 1:50 were used for IgG1- and IgG2a-specific ELISA, respectively. Horseradish peroxidase–conjugated anti-mouse IgG1 or IgG2a was used as secondary detection antibodies (Caltag). Plates were developed and analyzed by a 96-well plate reader at 450 nm.

Statistical analysis and data presentation. Multivariable statistics for the Kaplan-Meier survival curves were done by a log-rank test using GraphPad Prism software. Relative differences between groups were also done using a Student's t test and GraphPad Prism software.

GM-CSF–secreting tumor cell immunotherapy in combination with sLAG leads to prolongation of survival of tumor-bearing animals. The therapeutic model of established s.c. B16 melanoma was used to evaluate the potency of the combination therapy consisting of a GM-CSF–secreting tumor cell immunotherapy and murine soluble LAG-3Ig (sLAG3). The B16 melanoma model was set up to achieve 20% long-term (70 days) survival when mice received GMKd on day 3 after tumor inoculation but no therapeutic benefit when such immunotherapy was commenced at any later time point to provide a model well suited for the evaluation of agents that may have synergistic and/or additive effects when combined with GM-CSF–secreting tumor cell immunotherapy. C57BL/6 mice were inoculated with a lethal dose of live B16 tumor cells and followed by a single GMKd immunization on either day 3 or day 7 after tumor inoculation. sLAG monotherapy (0.1 μg) was given as s.c. injections on days 5 to 7. In the combination therapy groups, GMKd treatment was followed by s.c. injections of sLAG at the immunotherapy injection site on days 2 to 4 after immunotherapy. Mean survival time (MST) in HBSS-injected mice was 34 days, and the use of sLAG as monotherapy provided no survival benefit (MST = 31 days). When immunotherapy was initiated on day 3, 20% of the GMKd immunotherapy–injected mice survived long-term (>70 days) with a MST of 47 days. Combination therapy led to long-term survival in 40% of animals, with an overall MST of 54 days (Fig. 1A). When the initiation of immunotherapy was delayed to day 7 after tumor challenge, an overall reduction of antitumor protection was observed in both immunotherapy-alone and combination therapy treatment groups. In this setting, GMKd immunotherapy provided only modest survival benefit (MST = 37 days), whereas the MST in the combination therapy–treated animals was 49 days, with 10% long-term survivors (Fig. 1B). Furthermore, rechallenge of surviving mice in both immunotherapy alone and combination therapy groups with 2.5 times the initial dose of live tumor cells showed the presence of a potent B16-specific memory response, suggesting that the induction of memory responses is B16.GM dependent (data not shown). In summary, combination therapy consisting of GM-CSF–secreting tumor cell immunotherapy and sLAG3 was more efficacious in prolonging the survival of tumor-bearing mice than immunotherapy alone, showing the ability of sLAG to augment the antitumor response induced by the immunotherapy.

Rapid and enhanced tumor lymphocyte infiltration correlates with antitumor efficacy. The detection of high numbers of tumor-infiltrating lymphocytes (TILs) is associated with smaller tumor size and has been described as a prognostic marker for favorable clinical outcome (25). To analyze the association between tumor burden and TILs in this study, mice were challenged with live B16 tumor cells on day 0 and injected with GMKd immunotherapy on day 3. sLAG (0.1 μg) was given on days 5 to 7 after tumor inoculation as monotherapy and to GMKd-injected mice as combination therapy. At indicated time points, tumors were removed, collagenase digested, and analyzed by flow cytometry as described in Materials and Methods. Very low numbers of TILs were detected in HBSS-injected and sLAG monotherapy–injected animals at all time points evaluated (Fig. 2A-D). These findings suggest that sLAG may be unable to induce an antigen-specific immune response in this aggressive tumor model when used as a monotherapy. As early as 7 days after immunotherapy, the numbers of detectable CD4⁺ and CD8⁺ TILs were significantly higher in the combination therapy–injected group than in all other experimental groups (*, P < 0.01, for both CD4⁺ and CD8⁺ TILs; Fig. 2A and B). Evaluation of the kinetics of CD4⁺ TILs in the tumor showed that, in the combination therapy–injected group, CD4⁺ TILs were found at high numbers as early as day 7 and persisted until day 14 before returning to baseline levels by day 21. In contrast, the number of CD4⁺ TILs in the GMKd immunotherapy–injected group was still low on day 7, peaked on day 14, and returned to baseline levels by day 21. CD8⁺ TILs and effector CD8⁺ TILs (CD8⁺IFNγ⁺ and CD8⁺CD107a⁺) followed the same kinetics in animals receiving immunotherapy alone as their corresponding CD4⁺ TILs, whereas animals injected with the combination therapy showed high numbers of these cells on day 7 followed by a gradual decline that reached baseline levels by day 21 (Fig. 2C and D). The loss of the CD8⁺ TILs with effector phenotype in the tumor was inversely correlated with the appearance of CD4⁺FoxP3⁺ regulatory T cells as well as an increase in tumor cell number (Supplementary Fig. S1A and B). These data suggested that the control of tumor burden is at least partially T-cell mediated and that a higher ratio of effector TILs to regulatory T cells or tumor burden is essential for tumor growth control. Animals injected with the combination therapy showed an early infiltration of CD8⁺ TILs with effector phenotype associated with smaller tumor burden and enhanced overall survival.

sLAG enhances tumor-specific T-cell responses. The overall potency of GM-CSF–secreting tumor cell immunotherapies is dependent on the activation of tumor-specific effector T cells (1). For the assessment of tumor-specific T-cell responses, splenocytes from selected animals were evaluated by the IFNγ ELISPOT assay following various stimuli. Specifically, tumor-bearing mice received a single GMKd injection 3 days after tumor inoculation. sLAG (0.1 μg) was given on days 5 to 7 after tumor inoculation as monotherapy or in combination with GMKd immunotherapy. On day 10 (7 days after immunotherapy), 5 × 10⁵ splenocytes from each experimental group were cocultured with medium alone (baseline control), irradiated B16 cells, or Kb-specific peptides for the B16-derived antigens, Trp2 and gp100. The baseline level of IFNγ-secreting cells was low and comparable among all experimental groups (Fig. 3). For all tested stimulants, HBSS and sLAG monotherapy did not increase the number of IFNγ-secreting cells above background levels, suggesting that HBSS or sLAG monotherapy was insufficient to induce tumor-reactive effector T cells in this poorly immunogenic model. Mice that received GMKd alone showed a significant increase in the number of tumor-specific T cells compared with HBSS- and sLAG-injected mice (*, P = 0.02, for Trp2 and gp100 peptides; *, P = 0.007, for irradiated B16F10 cells), confirming the ability of immunotherapy to induce tumor-reactive effector T cells. Moreover, the addition of sLAG to GMKd immunotherapy showed a trend toward an increase in the number of tumor-specific T cells responding to irradiated B16F10 or Trp2 peptide stimulation that was, however, not statistically significant from that seen in the immunotherapy alone group probably because of the sample size evaluated.

To confirm effector function of the tumor-specific T-cell responses identified by the IFNγ ELISPOT assay and to investigate whether the addition of sLAG could significantly enhance tumor-specific T-cell responses, an in vivo CTL experiment was done. B16 antigen-specific responses are undetectable in the in vivo CTL assay as used in our laboratory; therefore, ovalbumin was introduced as a surrogate antigen. Mice were inoculated with B16.ova and immunized with GM.ova 3 days after tumor challenge. sLAG was given on days 5 to 7 after tumor inoculation as monotherapy and to GM.ova-treated mice as combination therapy. On days 14 and 21 after immunotherapy, mice were injected with syngeneic CSFE-labeled nonpulsed splenocytes and splenocytes pulsed with Kb-specific ovalbumin peptide (SIINFEKL). Cytolytic activity was determined 18 h later by measuring the difference in the ratio of CSFE-labeled cells between SIINFEKL-loaded or SIINFEKL-unloaded splenocytes.

Consistent with the data obtained from the IFNγ ELISPOT assay, HBSS- and sLAG-injected animals did not show any cytolytic function but GM.ova-injected mice showed significant antigen-specific cytolytic activity when compared with HBSS- or sLAG-injected animals (*, P < 0.001, for both time points; Fig. 4). Furthermore, mice that had received the combination therapy showed a significant enhancement of cytolytic activity compared with GM.ova-treated mice (*, P = 0.03 and *, P = 0.001, for days 14 and 21, respectively). In summary, the data obtained from the IFNγ ELISPOT and the in vivo CTL experiments suggested that sLAG enhances the cellular immune responses in mice receiving GM-CSF–secreting tumor cell immunotherapy and that the tumor-specific immune response correlated with prolongation of overall survival.

GM-CSF–secreting tumor cell immunotherapy plus sLAG induce production of proinflammatory, Th1- and Th2-specific cytokines. Cytokines play an important role in immune-mediated responses and are generated predominantly by helper T cells and macrophages. It is a well-accepted concept that Th1 cytokines drive responses toward cellular immunity and inflammation, whereas Th2 cytokines stimulate antibody production by B cells. The balance between Th1 and Th2 activity may steer the immune response in the direction of cellular- or humoral-mediated immunity. Splenocytes from animals immunized with GM-CSF–secreting tumor cell immunotherapy has been shown to express both Th1 and Th2 cytokines (9). To determine if sLAG changes the overall cytokine profile of this immunotherapy, the cytokine profile of splenocytes from all experimental groups was evaluated. The same experimental procedure was followed as described for the IFNγ ELISPOT assay, and on day 10 (7 days after immunotherapy), 5 × 10⁵ splenocytes from each treatment group were cocultured with irradiated B16 cells, and 2 days later, supernatants from the cocultures were collected and analyzed by the Th1/Th2 cytokine and inflammation cytometric bead array kits. Cytokines with secretion levels below the detection limit of the assay for all experimental groups included interleukin (IL)-4 and IL-12 (data not shown). In addition, whereas baseline levels of IL-2 were observed in HBSS-injected and sLAG monotherapy–injected animals, significantly higher than baseline but comparable levels were detected in animals injected with GMKd monotherapy or the combination therapy (data not shown). For tumor necrosis factor α (TNFα), IFNγ, IL-5, IL-6, IL-10, and MCP (Fig. 5), supernatants from HBSS-treated or sLAG monotherapy–treated mice had only baseline levels of these cytokines, suggesting that the splenocytes from these groups were not tumor reactive. In contrast, supernatant obtained from GMKd monotherapy–treated mice showed enhanced cytokine levels of IFNγ, IL-5, IL-6, IL-10, and MCP-1 and supernatants from splenocytes from combination therapy–treated mice showed a further increase in the level of these cytokines. Pronounced levels of the proinflammatory cytokine TNFα were detected in the supernatant of the combination group, whereas only baseline levels were detected in all other experimental groups. These data showed that the addition of sLAG to the GMKd immunotherapy enhanced proinflammatory cytokine production that correlated with an overall enhancement of in vivo T-cell activation.

Generation of antigen-specific IgG1- and IgG2a-specific antibodies in animals injected with GM-CSF–secreting tumor cell immunotherapy plus sLAG. IgG2a and IgG1 immunoglobulin isotypes are markers for Th1 and Th2 responses. IgG1 is the most abundant subclass of IgG and has the capability to activate complement, form immune complexes, and facilitate phagocytosis by binding to macrophages. In contrast, IgG2a up-regulates antigen-specific antibody responses. Previous experiments revealed that antibodies specific for endogenous B16 antigens are undetectable by ELISA when B16 cell lysate was used to coat the plates (data not shown) and recombinant B16 antigens are not commercially available; therefore, ovalbumin was used as a surrogate antigen, which allows tracking of ovalbumin-specific antibody responses in vivo. Animals receiving GM-CSF–secreting immunotherapy elicit an early IgG1 and a delayed IgG2a ovalbumin-specific antibody response in this system (Fig. 6A and B). To determine if either IgG1 or IgG2a antibody responses are augmented by sLAG, serum from all treatment groups was evaluated for the presence of antigen-specific IgG1- and IgG2a-specific responses. On day 0, mice were inoculated with B16 tumor cells modified to express ovalbumin (B16.ova), and on day 3, mice were immunized with GM-CSF–secreting B16.ova cells (GM.ova). sLAG (0.1 μg) was given on days 5 to 7 after tumor inoculation as monotherapy and to GM.ova-treated mice as combination therapy. Blood was collected from mice weekly and serum samples were analyzed by an ovalbumin-specific ELISA using ovalbumin protein to coat the plates as described in Materials and Methods. Serum from sLAG-injected animals showed only baseline levels of ovalbumin-specific IgG1- and IgG2a-specific humoral responses at all time points evaluated compared with serum from HBSS-injected mice, indicating the absence of ovalbumin-specific humoral immunity. However, although the combination of sLAG and GM.ova immunotherapy significantly augmented the ovalbumin-specific IgG1 response (*, P = 0.02 and *, P = 0.04, for D14 and D21, respectively; Fig. 6A), it did not further increase the ovalbumin-specific IgG2a response in mice treated with GM.ova immunotherapy or the combination therapy (Fig. 6B). These data suggest that sLAG specifically augments the Th2 antibody response in immunotherapy-injected animals but has no effect on the Th1 antibody response.

GM-CSF–secreting tumor cell immunotherapy has been shown to stimulate immune responses against tumors by the recruitment of dendritic cells to the immunotherapy injection site (10–12) and has shown preliminary evidence that it may provide therapeutic benefit in patients with cancer (10, 23, 26, 27). Although antitumor activity is observed when the immunotherapy is used as a monotherapy, combination therapies are currently being evaluated preclinically with the goal of enhancing overall antitumor activity, which could allow treatment of patients with large tumor burden. In this study, a combination therapy consisting of a GM-CSF–secreting tumor cell immunotherapy and sLAG3 was evaluated. LAG-3 associates with MHC class II molecules expressed on monocytes/dendritic cells, and numerous studies have shown that sLAG affects antigen-presenting cell maturation and activation by the up-regulation of costimulatory molecules and the production of IL-12 and TNFα (13, 15, 21, 22) and can therefore potentially induce stronger antigen-specific T-cell responses. The objective of the studies reported here was to evaluate whether sLAG could augment antitumor activity induced by GM-CSF–secreting tumor cell immunotherapy (GMKd). The results in animals treated with the combination therapy showed that enhanced Th1 and Th2 cellular and Th2 humoral antigen-specific responses are elicited by the combination of GMKd and sLAG and that these responses correlated with improved overall survival.

The activation of potent tumor-specific T cells with high effector activity is one of the goals of an effective whole-cell immunotherapy, and the data presented in this report showed that antitumor responses induced by GM-CSF–secreting tumor cell immunotherapy are augmented by sLAG. First, the detection of a greater number of tumor-specific IFNγ-secreting T cells and improved tumor-specific cytolytic activity showed that enhanced numbers of tumor-reactive T cells are induced in animals treated with the combination therapy compared with animals injected with either treatment as monotherapy. In addition, a more rapid and pronounced infiltration of effector CD8 TILs into the tumor microenvironment in animals receiving the combination therapy that translated into prolonged survival clearly showed a correlation between the presence of these effector T cells in the tumor and the control of progressive disease. Moreover, enhanced production of proinflammatory cytokines by the effector T cells together with an increase in the percentage of CD8 memory T cells (Ly6C⁺ and CD69⁻; Supplementary Fig. S2A and B) was detected in animals receiving the combination therapy. IL-6 has previously been reported to contribute to the survival of activated CD8 T cells during the contraction phase of the immune response (28) and was detected at higher levels in animals that received the combination therapy than in animals treated with GMKd immunotherapy only, suggesting that increased IL-6 levels may contribute to a greater memory pool in the former group. Memory T cells have been reported to respond more quickly on restimulation and are less prone to apoptosis during the contraction phase of the immune response (29). Thus, the addition of sLAG to a GM-CSF–secreting tumor cell immunotherapy not only induces a more potent primary immune response but also increases the pool of memory T cells that is expected to rapidly expand with subsequent antigen exposure from multiple immunizations or recurrent disease.

The Th1/Th2 balance is critically important in antitumor immune responses (30, 31). Th1 cells produce IFNγ and IL-2 and are essential for the induction of cellular immunity, whereas Th2 cells produce IL-4, IL-5, and IL-10 and play a key role in humoral immunity. Both Th1- and Th2-derived cytokines are known to possess antitumor activity (32–35). Th1 cells can induce lymphocyte infiltration into the tumor mass, resulting in the eradication of the tumor via cellular immunity. Th2 cells can induce inflammation within the tumor (31) and promote humoral immune responses by secreting antibodies that recognize specific antigens. Our studies show that sLAG potentiates immunotherapy-induced Th1 and Th2 immune responses. Most strikingly, highly elevated levels of TNFα were detected in the supernatant of splenocytes isolated from animals treated with the combination therapy compared with splenocytes from animals injected with the immunotherapy alone. TNFα is an immune-modulatory cytokine produced by activated macrophages and T cells and endothelial cells are its primary target. Some TNFα-induced biological effects include the up-regulation of leukocyte adhesion molecules (36, 37), the induction of a cascade of cytokines (38), fibrin deposition (39), and modulation of nitric oxide production (40). In addition, TNFα induces the alteration of endothelial cytoskeletal actin and the formation of intracellular gaps with increased permeability (41). High levels of TNFα could potentially be responsible for the observed rapid recruitment of high numbers of TILs into the tumor microenvironment by increasing vascular permeability and up-regulating cell adhesion molecules.

Among the many molecules shown to regulate T-cell function, LAG-3 has recently attracted significant interest. In a recent report, Grosso et al. (42) showed that LAG-3 blockade affected CD8 T cells by the reversal of CTL functional tolerance, which correlated with delay of tumor progression in a spontaneously developing tumor model. LAG-3 can affect antigen-specific immunoglobulin secretion by the interaction of LAG-3 on activated T cells with MHC II on activated B cells. It is postulated that antigen-specific B cells, through LAG-3/MHC II interactions, can be primed to proliferate and undergo class switch recombination and somatic hypermutation to result in an enhanced humoral responses. Although the experiments in this report do not directly show this point, early IgG1 and delayed IgG2a-specific antibodies are detected in serum from animals injected with GM-CSF–secreting tumor cells and the addition of sLAG significantly augmented the IgG1 but not the IgG2a antigen-specific responses. The discrepancy of the ability of sLAG3 to affect IgG1 but not IgG2a antibody responses may be the result of the treatment schedule used for sLAG3 in this report. sLAG was given 2 to 4 days after immunotherapy and IgG2a responses were not detectable until 4 weeks after immunotherapy. Because sLAG possesses a very short half-life, it might augment IgG2a responses if it were administered at later time points during therapy. In the clinic, therapeutic benefit associated with the induction of antitumor humoral responses has been reported (2, 43, 44). Because the induced antibodies in this study were antigen specific, the enhancement of humoral responses by the combination therapy could potentially contribute to overall therapeutic benefit.

In summary, this study has shown that the antitumor responses can be enhanced by the combination of a GM-CSF–secreting tumor cell immunotherapy and sLAG, leading to an improvement in overall survival in animal tumor models. Our data showed that the enhancement of the antitumor responses by the combination therapy correlated with increased induction of proinflammatory as well as Th1- and Th2-specific cytokines, increased induction of antigen-specific cellular and humoral responses, and a more rapid infiltration of effector T cells into the tumor. In summary, the addition of sLAG as an adjuvant to GM-CSF–secreting tumor cell immunotherapy results in enhanced antitumor protection that involves the coordination of T- and B-cell immunity.

No potential conflicts of interest were disclosed.

We thank P. Working and S. Liang for critical reading of the manuscript and B. Batiste, J. Ho, S. Tanciongo, and T. Langer for their technical assistance.