Agonist Anti-Human CD27 Monoclonal Antibody Induces T Cell Activation and Tumor Immunity in Human CD27–Transgenic Mice

The use of Abs targeting immunoregulatory molecules for the treatment of cancer and other diseases has generated significant interest in recent years (1, 2). In particular, the clinical data from melanoma patients treated with ipilimumab, an Ab that blocks CTLA-4 function, have brought validation to this approach (3). More recent results with Abs that block the checkpoint inhibitory molecule PD1 are adding to the promise of effective and safe treatments for cancers using immunomodulatory approaches (4, 5). There are limited clinical data available with Abs that provide immunoactivation through T cell costimulation. Targeting members of the TNFR superfamily, including CD40, CD134 (OX40), and CD137 (4-1BB), with agonist Abs showed evidence of clinical activity, but it is accompanied by limitations related to toxicity (6–9).

Until recently, CD27, a costimulatory molecule on T cells and a member of the TNFR superfamily, largely has been overlooked as a therapeutic target for immunotherapy. CD27 is a disulphide-linked homodimer of 120 kDa and a type I transmembrane molecule that is constitutively expressed on naive and activated T cells, memory B cells, and a subset of NK cells (10). In this regard, CD27 differs from other members of the TNFR family, which are expressed primarily upon cellular activation (11–13). CD27 is known to play key roles in diverse immunological processes, such as T cell activation, effector function, maturation, and survival, as well as NK cell proliferation and cytotoxicity (14–18). Physiologically, CD27 costimulatory signaling is controlled by the restricted expression of its unique ligand, CD70, which is virtually absent on resting cells but is transiently upregulated on lymphocytes and dendritic cells upon their activation (17, 19, 20).

Several lines of evidence suggest that CD27 costimulation could provide effective T cell–mediated antitumor or antiviral immunity. Early studies showed that engineered expression of CD70 on tumor cells increased their immunogenicity and antitumor activity when transplanted into mice (21, 22). Arens et al. (23) demonstrated that constitutive expression of CD70 on B cells in CD70-transgenic mice increased the Ag-specific T cell response and protected mice from lethal tumor challenge. Studies using CD70–Ig fusion protein demonstrated that CD27 stimulation, in conjunction with protein vaccination, significantly enhanced the magnitude and quality of the CD8⁺ T cell response (24). In related work, Keller et al. (25) demonstrated that the engineered expression of CD70 on steady-state dendritic cells was sufficient to overcome tolerance and induced potent Ag-specific immunity, and more recent studies (26) showed that CD27 stimulation can lower the threshold of CD8⁺ T cell activation to low-affinity Ags and provide a broader repertoire of Ag-reactive T cells.

The therapeutic effect of an anti-CD27 agonistic mAb was first shown in studies (27) investigating the mechanisms leading to antitumor responses following treatment with an agonist anti-CD40 mAb. Surprisingly, a CD70 mAb that blocked the CD70/CD27 costimulatory pathway completely abolished the therapeutic effect of anti-CD40 mAb treatment in a BCL1 lymphoma model, demonstrating that CD27 signaling was necessary for the efficacy of anti-CD40 in this tumor model. This led to the development and testing of a rat anti-mouse CD27 mAb, AT124-1, which indeed provided similar antilymphoma efficacy as did the anti-CD40 mAb (27). The antitumor efficacy of CD27 targeting was later reproduced in the EG7 thymoma model by Sakanishi and Yagita (28) using novel anti-mouse CD27 mAbs. These studies were extended by Roberts et al. (29), who demonstrated the T cell–dependent mechanism of anti-CD27 therapy using RAG2- or IFN-γ–knockout mice, which showed minimal or no antitumor activity of the anti-CD27 mAbs in a B16cOVA model.

We developed a fully human Ab, 1F5, which recognizes human CD27 (hCD27) with high affinity, blocks soluble CD70 binding, and inhibits the growth of CD27⁺ human lymphoma cells in SCID mice (30). In this study, we generated a human CD27-transgenic (hCD27-Tg) mouse model to investigate the agonistic activities of 1F5 and its antitumor efficacy. 1F5 recognized and stimulated hCD27-expressing T cells from the transgenic mice in vitro, enhanced the level of Ag-specific CD8⁺ T cells in vivo, and mediated potent antitumor immunity. Based partly on these results, the 1F5 mAb recently entered clinical development under the name CDX-1127.

A fully human anti-human CD27 Ab 1F5 was generated and characterized as described elsewhere (30). The N297S mutation was introduced into 1F5 by site-directed mutagenesis (Stratagene) and confirmed by DNA sequencing. FreeStyle CHO-S cells (Invitrogen) were transiently transfected to express the 1F5_N297S mutant Ab, which was subsequently purified from culture supernatants with a MabSelect SuRe Pro A column (GE Healthcare). The rat anti–murine CD27 (mCD27) mAb AT124-1 was described previously (27). The endotoxin level for all Abs used in functional assays or in vivo studies was <1.0 EU/mg, as measured by the Gel Clot LAL method (Lonza).

Binding affinity and binding kinetics of the 1F5 and 1F5_N297S mutant Abs to mouse FcγRs were examined by surface plasmon resonance analysis using a Biacore 2000 SPR instrument (Biacore, Uppsala, Sweden), according to the manufacturer’s guidelines. The Abs 1F5 and 1F5_N297S were covalently linked to a Biacore CM5 sensor chip using standard amine coupling chemistry with an Amine Coupling Kit (Biacore). The Mouse FcγRs (R&D Systems) were allowed to flow over the sensor chip at concentrations ranging from 12.5 to 400 nM and at a flow rate of 30 μl/min for 300 s. A blank flow cell with no protein immobilized was used for background subtraction. The affinity and kinetic parameters were derived from the time-dependent association and dissociation curves using BIAevaluation software (Biacore).

A human bacterial artificial chromosome (BAC) clone was identified and shown to contain the entire CD27 gene and promoter by PCR and sequencing analysis. A 175-kb insert DNA was prepared from the BAC and microinjected into fertilized oocytes, which were transplanted into the C57BL/6-SJL hybrid strain. The transgene transcript was detected by RT-PCR with hCD27-specific primers in total RNA extracted from peripheral blood (PB) leukocytes of three founder mice. Two transgenic lines with very similar expression profiles were established, and one line was backcrossed to C57BL/6, BALB/c, and C3H genetic backgrounds for more than five generations using selected breeders based on congenic analysis (IDEXX RADIL). All mice were housed under specific pathogen–free conditions in the Celldex animal facilities and were used in accordance with the guidelines established by the Institutional Animal Care and Use Committees at our institution.

Immunohistochemistry (IHC) and flow cytometry were performed to examine the expression profile of the hCD27 transgene. Sixteen organs (skin, lymph nodes [LNs], thymus, spleen, liver, brain, pituitary, heart, lung, kidney, jejunum, ileum, colon, testis/ovary, and muscle) were collected from C57BL/6 hCD27-Tg homozygous and heterozygous mice and wild-type (WT) littermates. Frozen sections were cut and stained with one of our well-characterized human anti-hCD27 mAbs, human IgG1 (hIgG1) clone 3H8; with murine IgG1 clone L128 (BD Biosciences) for hCD27; or with either hamster IgG1 clone LG3A10 (BD Biosciences) or rat IgG2a clone 137910 (R&D Systems) for mCD27. All of the primary Abs and their isotype controls were conjugated to FITC. Sections were then stained with rabbit anti-FITC Ab (Invitrogen), detected using the Dako EnVision Kit, and counterstained with hematoxylin.

Flow cytometry was performed on single-cell suspensions prepared from PB, LNs, spleen, and thymus of hCD27-Tg homozygous and heterozygous mice and their WT littermates by manual mechanical dissociation, followed by lysis of RBCs. After blocking non-Ag–specific staining with anti-CD16/CD32 mAb clone 2.4G2, cells were stained with anti-CD27 Abs and for lymphocyte surface markers, including CD3ε, CD4, CD8a, CD19, and NK1.1 (BD Biosciences). Isotype controls for all Abs were included. Analyses were done on an LSR or FACSCanto II flow cytometer (BD Immunocytometry Systems).

We used ELISA to compare the binding of hCD27 and mCD27 with recombinant murine CD70. Briefly, CD70 (R&D Systems) was coated on a plate at 2 μg/ml, and then the plates were blocked with 5% albumin in PBS solution. Dilutions of hCD27-Fc, mCD27-Fc, and negative control human TIM-1–Fc (all from R&D Systems) were added at various concentrations in duplicate, and binding was detected with HRP-labeled goat anti-human IgG Fc. The absorbance at 450 nm was determined using a microplate reader.

For T cell activation, microtiter wells were precoated with anti-CD3ε (145-2C11) at 5 μg/ml or received soluble murine IL-2 (both from R&D Systems) at 100 ng/ml or the mitogen PHA at 5 μg/ml final concentration. Splenocytes from hCD27-Tg homozygous and WT mice were incubated in the plate for 3 d and then stained for hCD27 or mCD27 and the surface markers CD3, CD4, and CD8. The mean fluorescence intensity (MFI) for CD27 was compared for CD4⁺- or CD8⁺-gated cells. The culture supernatants were collected to measure soluble hCD27 and mCD27 levels by sandwich ELISA using anti-human or anti-murine Ab pairs and standards from Sanquin and R&D Systems.

T cells were isolated by depletion of non-T cells from splenocytes of homozygous C57BL/6 hCD27-Tg mice using the Pan T Cell Isolation Kit II (Miltenyi Biotec) and labeled with 2.5 μM CFSE (Invitrogen). T cells (0.5 × 10⁶ /well) were transferred to 96-well U-bottom plates that had been precoated or not with 2 μg/ml CD3ε mAb 145-2C11. 1F5 or an irrelevant hIgG1 was added to the cultures at a final concentration of 0.2 μg/ml in the presence or absence of 10-fold excess of murine anti-human IgG (Jackson ImmunoResearch Lab) for cross-linking. Immobilized CD3ε mAb, with or without 10 μg/ml coated recombinant mouse CD70, was also included for reference. The cultures were kept at 37°C in an incubator for 3 d, and brefeldin A (Sigma-Aldrich) was added to a final concentration of 10 μg/ml for the last 4 h. Cells were stained with blue fluorescent reactive dye (Live/Dead Fixable Dead Cell Stain Kit; Invitrogen), surface CD8, CD4, and human CD27 (clone MT271; BD Biosciences, noncompetitive with 1F5 binding), and, subsequently for intracellular IFN-γ and TNF-α after fixing and permeabilizing with Cytofix/Cytoperm (BD Biosciences). Samples were analyzed using a FACSCanto II flow cytometer.

C57BL/6 hCD27-Tg and WT littermates were dosed i.v. with 5 mg chicken OVA (Sigma-Aldrich) and 250 μg 1F5, rat anti-murine CD27 (clone AT124-1), or isotype control on day 0, with an additional 250 μg Ab the next day. In subsequent studies, we used i.p. dosing of OVA (5 mg) and a single dose of 1F5 or 1F5_N297S (50 μg) on day 0. Splenocytes were collected on day 7 for tetramer staining, IFN-γ intracellular cytokine staining (ICS), and ELISPOT assays following RBC lysis. Splenocytes (1 × 10⁶) were stained with 10 μl H-2K^b-SIINFEKL tetramer (Beckman Coulter) in conjunction with surface CD8 and IFN-γ ICS, as described above. A total of 100,000 events was acquired using an LSR flow cytometer, and the fraction of tetramer⁺ and IFN-γ⁺ cells in the CD8⁺-gated population was determined. For ELISPOT, assays were carried out by incubating 0.25 × 10⁶ or 1.0 × 10⁶ splenocytes/well in triplicate in IFN-γ Ab–coated 96-well filtration plates overnight in the presence or absence of 2 μg/ml SIINFEKL peptide. Spots were developed using an IFN-γ Ab reagent set with AEC substrate chromagen (BD Biosciences) and counted by ZellNet Consulting.

The antitumor activity of 1F5 was assessed in four syngeneic mouse tumor models. B-lymphoma BCL1 was maintained and expanded in WT BALB/c mice. For the BCL1 tumor model, ∼10⁷ splenocytes from the in vivo expansion were administered i.v. to the BALB/c hCD27-Tg mice. The colon carcinoma CT26 (American Type Culture Collection) was cultured in RPMI 1640 with 10% FBS. For this tumor model, 10⁴ CT26 cells were administered s.c. to BALB/c hCD27-Tg mice. The thymoma EL4 and its OVA-expressing derivative EG7 (both from American Type Culture Collection) were grown in RPMI 1640/10% FBS and in the presence of 0.5 mg/ml G418 for EG7 cells. For the tumor model, 0.5 × 10⁶ EL4 cells or 1.0 × 10⁶ EG7 cells were administered s.c. to C57BL/6 hCD27-Tg mice. 1F5 Ab was administered over a dose range of 2 to 600 μg as multiple i.p. injections, as indicated in the figure legends. Control cohorts included hCD27-Tg mice injected with saline or hIgG1 and WT mice injected with 1F5. Mice were observed daily, and tumor size was measured at least twice a week with calipers. Tumor volume was calculated using a modified ellipsoid formula (length × width² ÷ 2). Mice were euthanized when they met predetermined end points approved by the Institutional Animal Care and Use Committee. Secondary tumor challenge was performed in mice free of measurable tumors ∼3 mo postprimary tumor inoculation and paired with the same number of control naive mice. No further treatment was administered.

T cell–depletion studies were performed using anti-CD4 (GK1.5) or anti-CD8 (53-6.7; both from eBioscience) Abs to deplete CD4⁺ and CD8⁺ T cells, respectively. In the CT26 model, depleting Abs were administered i.p. on day 1 (500 μg) and day 10 (50 μg) posttumor inoculation. In the EG7 model, depleting Abs were administered i.p. on day 3 (500 μg) and day 12 (100 μg). In pilot studies, these regimens were shown to effectively deplete CD4⁺ or CD8⁺ splenic T cells by flow cytometry.

The Student t test was used for mean ± SD (two-tailed unequal variance) comparisons. Kaplan–Meier analysis was used for survival.

hCD27-Tg mice were generated by standard microinjection techniques in the C57BL/6-SJL hybrid strain using a 175-kb fragment of BAC DNA including the entire CD27 gene and additional upstream sequences. Therefore, expression of the hCD27 transgene was anticipated to be controlled by its own promoter. One of three founder mice was expanded and successfully back-crossed onto the C57BL/6, C3H, or BALB/c background. Characterization of the transgene expression was performed by IHC on 16 tissues collected from hCD27-Tg and WT mice using human and mouse-specific anti-CD27 Abs. Limited expression of hCD27 or mCD27 was observed in nonlymphatic tissues that appear to be scattered infiltrating lymphocytes or lymphatic structures (data not shown). As expected, there was significant staining of CD27 in lymphoid tissues (Fig. 1). IHC staining of hCD27 and mCD27 displayed a generally similar pattern of expression in most lymphoid tissues. In particular, strong staining of hCD27 and mCD27 was observed in T cell areas, such as the periarterial lymphatic sheath in the white pulp of spleen and paracortex of LN, and occasional to rare staining was noted in B cell areas, sinuses, medulla, and red pulp (Fig. 1). We observed a marked difference in the staining pattern for hCD27 compared with the endogenous mCD27 in the thymus, where hCD27⁺ cells were frequent in medulla but rarely seen in the cortex, whereas mCD27-stained cells were frequent in both medulla and cortex (Fig. 1). This difference is consistent with the previous data on thymic T cells that support a more limited expression of hCD27 compared with mCD27 (31).

Flow cytometry analysis of CD27 expression was performed on cells isolated from PB, LNs, spleen, and thymus of hCD27-Tg mice and WT mice. The frequency of hCD27 and mCD27 expression was observed in a similar percentage of CD4⁺ (82% versus 86%) and CD8⁺ (73% versus 85%) T cells from spleen (Fig. 2A). However, there appeared to be a greater variability in the expression level of hCD27, which contained both CD27^bright and CD27^low populations, compared with mCD27, which was more homogeneous with CD27^intermediate staining. hCD27 was expressed at a higher frequency than was mCD27 on CD19⁺ B cells (10% versus 2%), which likely represents the difference in CD27 expression on somatically mutated memory B cells between the two species (32). hCD27 and endogenous mCD27 were predominantly coexpressed in lymphoid tissues (Fig. 2B) with some hCD27⁺/mCD27⁻ and hCD27⁻/mCD27⁺ cells observed, which likely represent B cells and NK cells, respectively. hCD27 was expressed at much lower frequency on murine NK cells compared with mCD27 (Fig. 2C).

To evaluate the regulation of hCD27 expression, we investigated the effect of agents that activate T cells and are known to upregulate CD27 expression and shedding (11, 19). PHA or TCR stimulation (anti-CD3ε) upregulated hCD27 and mCD27 expression levels on T cells, as shown by flow cytometry analysis (Fig. 3A). IL-2, as expected, did not enhance hCD27 expression (Fig. 3A). A similar pattern was observed for the shedding of soluble CD27 into the media (Fig. 3B) of the same cultures used for flow cytometry.

Finally, we demonstrated that human CD27 can bind efficiently to murine CD70, which would allow for interactions with its ligand in hCD27-Tg mice (Fig. 4). Overall, the data support that the hCD27 transgene is expressed and regulated in a manner generally consistent with the known biology of hCD27, and hCD27-Tg mice would be a useful model for evaluating hCD27-targeting agents.

Proliferation and cytokine-induction experiments were performed on purified T cells from hCD27-Tg mice to demonstrate the functionality of the hCD27 transgene and to confirm the agonist activity of the anti-CD27 mAb 1F5. CD27 costimulation of T cell proliferation requires additional signals, in particular TCR activation (11, 14), and our prior studies using purified T cells from healthy donors confirmed that 1F5 activated human T cells only in the context of TCR stimulation (30). Therefore, CD27 activation was assessed on T cells derived from hCD27-Tg mouse spleens in combination with immobilized anti-CD3ε mAb. As shown in Fig. 5A, the 1F5 mAb combined with a suboptimal stimulation of anti-CD3ε mAb, significantly enhanced CD8⁺ T cell proliferation and IFN-γ production but only when the 1F5 was cross-linked with anti-human IgG. The cross-linked 1F5 also induced the proliferation and cytokine production of CD4⁺ T cells when combined with anti-CD3ε but to a significantly lower magnitude than that of the CD8⁺ T cells. We observed the same results for TNF-α production (data not shown). Demonstrating relevance to the activity of the native ligand, we observed a similar effect on CD8⁺ T cell proliferation and cytokine production with plate-bound murine CD70 combined with anti-CD3ε (Fig. 5B).

Costimulation of CD27 was shown to effectively enhance CD8⁺ T cell responses to vaccination using a sCD70–Ig fusion protein (24), which suggests that agonist anti-CD27 mAbs may have similar effects. We compared the ability of 1F5 and rat anti-mCD27 mAb (clone AT124-1) to enhance CD8⁺ T cell responses to OVA (Fig. 6). The hCD27-Tg mice and WT littermates were immunized with OVA protein and administered two doses of anti-CD27 mAbs or isotype control prior to analysis for OVA-specific CD8⁺ T cells with H-2K^b/SIINFEKL tetramer and IFN-γ ICS (Fig. 6A–C) or by IFN-γ ELISPOT after stimulation with the SIINFEKL peptide (Fig. 6D). The 1F5 mAb and AT124-1 mAb showed a significant enhancement in the percentage and absolute amount of Ag-specific CD8⁺ T cell responses in hCD27-Tg mice and WT mice, respectively. The activation of the tetramer-stained cells was confirmed by intracellular staining for INF-γ. The specificity of the enhanced response was demonstrated by the weak CD8⁺ T cell response in WT mice treated with 1F5 or the hCD27-Tg mice treated with an irrelevant hIgG1.

To assess whether the immunomodulatory activities of 1F5 can be translated to effective antitumor therapy, we investigated the effect of varying doses of 1F5 in the BCL1 tumor model using syngeneic BALB/c hCD27-Tg mice. We observed significant prolongation of survival in 1F5-treated animals compared with isotype-treated controls when doses ≥ 10 μg were delivered on days 3, 5, 7, 9, and 11 after tumor challenge. The lowest dose of 1F5 tested, five doses of 2 μg, had no survival benefit compared with controls (Fig. 7). There was a trend toward longer median survival times with higher doses of 1F5, although doses of 50, 150, or 600 μg were not statistically different from each other. The 600-μg dose group was statistically significantly different from the 10-μg dose group (median survival of 78 d versus 59 d, respectively, p = 0.032). As an additional control, we administered the BCL1 tumor to six WT BALB/c mice, followed by treatment with 1F5 (five doses of 600 μg each using the same schedule as above), but it had no benefit on median survival (23 d), and no mice survived past day 30 (data not shown).

We further tested the efficacy of the 1F5 mAb in the s.c. implanted colon carcinoma CT26 tumor model, which allowed for tumor growth to be monitored. Groups of BALB/c hCD27-Tg mice were inoculated with CT26 cells and then treated with saline or 1F5 (5 doses of 600 μg on days 3, 5, 7, 9 and 11 posttumor challenge). Initially, tumors grew similarly in both groups of mice; however, in most animals treated with 1F5 this growth was followed by a slowing and subsequent regression of the tumor, which was not observed in the saline-treated mice (Fig. 8A). Mice treated with 1F5 that underwent tumor regression remained tumor free for 3 mo and were rechallenged with CT26 cells (Fig. 8B). All rechallenged mice were resistant to CT26 tumor growth, whereas a concurrent group of similarly challenged naive mice showed rapid tumor progression from the CT26 inoculation. The pattern of tumor regression and immunity to rechallenge indicates the induction of T cell–mediated antitumor and memory responses. When 1F5 administration was delayed until tumors were of significant size (0.15–0.2 cm³), the treatment effect was diminished, but it still resulted in a measurable survival benefit (Fig. 8C).

The aggressively growing EL4 thymoma was resistant to 1F5 treatment, even when administered early after tumor challenge (Fig. 9A). However, 1F5 induced marked tumor regression in ∼40% of mice given EG7 tumors, an OVA-expressing variant of EL4 (Fig. 9B), which suggested the OVA may be the dominant tumor-rejection Ag. Subsequently, we investigated whether mice that rejected EG7 tumors after 1F5 treatment were resistant to either EL4 or EG7 rechallenge (Fig. 9C). The EG7 survivors were protected against EG7 rechallenge, with 90% of the mice never developing tumors. The protection of EG7 survivors against EL4 tumor was substantially lower, but it did result in a clear survival benefit, suggesting that some of the immune response was directed to shared Ags and not only to OVA.

The contribution of CD4⁺ and CD8⁺ T cells to the antitumor activity of 1F5 was investigated by specific depletion of these populations in the CT26 and EG7 tumor models. Elimination of either CD4⁺ or CD8⁺ T cells abolished the enhanced survival benefit from 1F5 in both tumor models (Fig. 10). These results support the expectation that 1F5 treatment leads to an effective CD8 antitumor response, but they also demonstrate the critical role for CD4⁺ T cells in this process.

The in vitro studies demonstrated that T cell activation by 1F5 required cross-linking of the Ab. We presumed that this was also required in vivo and likely occurred through interaction with FcγR. To confirm this hypothesis, we introduced a mutation into the 1F5 Fc domain (1F5_N297S) that resulted in the loss of binding to FcRs (Fig. 11B), as expected (33), but retained the same binding to hCD27 (Fig. 11A). The 1F5_N297S mutant lost the capacity to enhance the SIINFEKL-specific T cell response when coinjected with OVA protein (Fig. 11C) and was unable to mediate antitumor activity (Fig. 11D). Therefore, we conclude that FcγR engagement is required for the in vivo costimulatory and antitumor activity of 1F5.

Targeting the CD27-costimulation pathway with specific Abs showed considerable promise in a variety of murine model systems (27–29) and motivated the development of human anti-hCD27 Abs with the potential for human clinical use. We recently reported on the characterization of the anti-hCD27 mAb 1F5, which was selected based on its effective costimulation of human T cells and potent antitumor activity against CD27-expressing lymphoblastic tumors in immunodeficient (SCID) mice (30). The objective of this study was to establish the costimulatory and antitumor activity of the 1F5 mAb in immunocompetent mice, as well as to investigate the mechanism of 1F5-induced CD27 signaling.

For our in vivo model we generated and characterized hCD27-Tg mice. The expression pattern for the transgene was examined by IHC and flow cytometry on relevant tissues and showed a profile consistent with the endogenous mCD27 and with the expected distribution for hCD27 (11–13). Some discrepancies between murine and human CD27 expression were noted (e.g., broader expression of mCD27 in thymus, higher expression of hCD27 on B cells), which are consistent with the reported literature for CD27 expression in the two species (11, 19). We demonstrated in vitro that the cell surface levels and shedding of hCD27 were appropriately upregulated by PHA and TCR stimulation of spleen-derived T cells. Last, we showed that cross-linking hCD27 with 1F5 in combination with TCR stimulation led to the proliferation and activation (Th1 cytokine production) of hCD27-Tg T cells. The 1F5 mAb does not cross-react with mCD27 and is ineffective in WT mice; thus, the T cell activation occurred via signaling through hCD27.

Immunization of hCD27-Tg mice with OVA with the concurrent administration of 1F5 significantly enhanced Ag-specific CD8 T cell expansion and effector function (IFN-γ response), demonstrating that 1F5 was able to trigger the costimulatory signals by binding to hCD27 in vivo. This “adjuvant” effect of 1F5 was dependent on Fc–FcγR interactions, as shown by mutating the 1F5 mAb (IgG1) to a form with greatly reduced binding to FcγRs (1F5_N297S). This finding emphasizes that CD27 stimulation by specific Abs in vivo may be altered by modifying the isotype or affinity of the Fc domain for FcγR family members, as was shown for anti-CD40 agonist Abs (34, 35).

We demonstrated the antitumor activity of 1F5 monotherapy using several syngeneic models commonly investigated for tumor immunotherapy. Importantly, these tumors express little or no CD27, and no direct binding is observed with the 1F5 Ab (Supplemental Fig. 1). 1F5 was effective at dose levels as low as 10 μg (50 μg total dose) in extending the survival of mice challenged systemically with BCL1 lymphoma cells; however, most mice eventually succumbed to the tumor. We only tested the highest dose of 600 μg in the s.c. colon tumor model CT26 and found that the majority of animals survived the tumor challenge if treated early (starting day 3 after tumor challenge), which was characterized by slow tumor growth and eventual complete regression. However, the efficacy was significantly diminished when treatment was delayed (e.g., when 1F5 therapy was started on ∼day 15), suggesting that, under these conditions, there was insufficient time to develop immunity (median survival without treatment was 31 d) and/or the immunosuppressive mechanisms have overcome the antitumor effect. Importantly, the mice that underwent tumor regression from the initial challenge were resistant to a subsequent administration of CT26 tumor cells given 3 mo after initial treatment with 1F5, indicating that a memory response likely had been established. Consistent with this observation we found that eliminating either CD4⁺ or CD8⁺ T cells abrogated the antitumor efficacy of 1F5. The observation that CD4⁺ cells were also critical to 1F5 activity was not necessarily anticipated, but it may well reflect the importance of promoting a strong helper response to potentiate the cytolytic T cell response. In addition, 1F5 may be impacting regulatory T cells, but additional studies will be required to investigate this point.

The 1F5 mAb also induced protective immunity during the treatment of mice bearing EG7 (EL4-OVA) tumors, and this immunity was partially effective against a subsequent challenge with EL4 cells, suggesting that the tumor immunity was induced against shared Ags and not just against OVA. The antitumor response in the EG7 model was not observed when mice were treated with the mutant 1F5_N297S, indicating the requirement for FcγR engagement for in vivo efficacy of 1F5.

Collectively, these data demonstrate that targeting hCD27 with 1F5 in hCD27-Tg mice can induce T cell activation leading to enhanced Ag-specific immunity and durable antitumor efficacy. The requirement for FcγR engagement for the agonist activity of 1F5 in vivo suggests that the activity of 1F5, and presumably other anti-CD27 mAbs, can be manipulated through selection of the Ab isotype or through specific mutations in the Fc domain that influence the binding to specific FcγRs. In a fashion similar to other immunomodulating Abs, 1F5 exhibited potent antitumor activity when administered soon after tumor challenge, as well as when using tumor models that are considered immunogenic; however, the activity was diminished in the case of delayed treatment. It is likely that for the more challenging tumor models, combinations of 1F5 with vaccines or other interventions that help to induce immunity and/or reduce tumor burden will provide greater efficacy; these studies are underway. These data support 1F5 as a promising candidate for clinical investigation. Phase 1 studies with 1F5, under the name CDX-1127, in patients with advanced solid tumors or hematological malignancies are ongoing.

We thank Dr. Yan Zhang for contributions to the IHC studies and Jennifer Widger, Crystal Sisson, Kathleen Borrelli, James Boyer, Eric Forsberg, Lauren Gergel, and Sarah Naylor for expert technical assistance.

L.-Z.H., L.J.T., L.V., J.W., A.C., C.D.P., S.M.R., H.M., and T.K. are employees of Celldex Therapeutics and may hold stock or equity interests in the company. A.T. and M.J.G. are coinventors of U.S. Patent 8,481,029 on the development of agonist anti-CD27 Abs. N.P. has no financial conflicts of interest.