B7-H3: An Attractive Target for Antibody-based Immunotherapy

Monoclonal antibodies (mAbs), alone or as part of novel engineered structures, have had impressive clinical successes in cancer treatment (1). Furthermore, mAbs have facilitated the identification and molecular characterization of tumor antigens (TA) that can serve as antibody-based immunotherapy targets. Many TAs with a selective, although not specific, expression on malignant cells have been identified. Among them, B7-H3, also known as CD276 or B7RP-2, a member of the B7 ligand family, appears to be the “right” TA for antibody-based immunotherapy (2), given that it is (i) highly expressed with limited heterogeneity on differentiated tumor cells, (ii) expressed on cancer-initiating cells (CIC); the latter play a major role in metastasis and recurrence, and must be eradicated for a therapy to be effective (Fig. 1; ref. 3), and (iii) expressed on tumor-associated vasculature (TAV) and stroma; therefore, B7-H3 immunotargeting is expected to disrupt the tumor microenvironment (TME) and to inhibit neoangiogenesis. In contrast, B7-H3 has a restricted distribution in normal tissues. These characteristics have prompted significant interest in evaluating B7-H3 as a target of mAb-based immunotherapy.

B7-H3 is a 316 amino-acid long type I transmembrane protein encoded by a gene in chromosome 9 in mice and 15q24 in humans. It exists in two isoforms determined by its extracellular domain. In mice, the extracellular domain comprises a single pair of immunoglobulin variable (IgV)-like and immunoglobulin constant (IgC)-like domains, whereas in humans it comprises one or two identical pairs due to exon duplication (4–6). Its intracellular domain is short without any known signaling motif (Fig. 2).

B7-H3 shares 20%–27% amino acid identity with other B7 family ligands (7). B7-H3 has maintained its amino acid sequence through phylogenetic evolution, as it has an 88% amino acid sequence homology with its murine counterpart (5). Despite this similarity, no human B7-H3–specific antibody cross-reacts with the endogenous murine B7-H3 expressed on murine cells. However, human B7-H3–specific mAbs cross-react with murine B7-H3 expressed by transfected human cells (8), likely because of different B7-H3 glycosylation in the two species and the role of carbohydrates in the expression of epitopes recognized by human B7-H3–specific mAbs (9). Thus, caution should be exercised in interpreting toxicity of strategies that test human B7-H3–specific mAbs in murine models, unless the mAb used has been shown to cross-react with the endogenous B7-H3 expressed by normal murine tissues.

The molecular weight of B7-H3 protein moiety is approximately 45–66 kDa, while that of the glycosylated 4Ig-hB7-H3 isoform is approximately 100 kDa (10). A B7-H3 “soluble” isoform has been detected in plasma, a questionable statement because there is insufficient evidence about its solubility and B7-H3 is also expressed on exosomes (11). The correlation of B7-H3 level in serum of patients with cancer with clinicopathologic variables (12) suggests its use as a noninvasive biomarker for diagnosis, prognosis, and/or treatment response.

B7-H3 mRNA is expressed in most normal tissues. In contrast, B7-H3 protein has a very limited expression in normal tissues (13, 14) because of its posttranscriptional regulation by miRNAs. Specifically, miRNA-29 suppresses B7-H3 expression in normal tissues by targeting the B7-H3 3′-untranslated mRNA region. Similarly, miRNA-29, miRNA-124, miRNA-199a, and miRNA-1253 regulate B7-H3 expression in neuroblastoma (15), osteosarcoma (16), cervical cancer (17), and medulloblastoma (18), respectively. Furthermore, high miRNA-187, miRNA-1301-3p, miRNA-335-5p, and miRNA-28-5p expression downregulates B7-H3 expression in colorectal cancer; this change leads to reduced aggressiveness, lower pathologic stage, and decreased metastatic potential (19, 20).

B7-H3 may inhibit natural killer (NK)-cell activation and have a proinflammatory role leading to cytokine release from monocytes and/or macrophages, in a Toll-like receptor (TLR)-2 and TLR-4–dependent manner (21). Nevertheless, B7-H3 involvement in innate immunity remains questionable.

In adaptive immunity, B7-H3 plays a role in T-cell regulation (22). Surprisingly, the molecular structure of its receptor remains unidentified, although soluble B7-H3 has been shown to bind to CD4⁺ T, CD8⁺ T, NK, and NKT cells (23) and the extent of its binding is increased upon T-cell activation (2). In mice, B7-H3 binds to the CD8⁺ T-cell trem-like transcript 2 (TLT-2) receptor leading to their activation (24), however, such interaction has not been found with human B7-H3 (25). At any rate, the B7-H3 receptor differs from CD28, cytotoxic T lymphocyte-associated protein 4 (CTLA-4), programmed cell death protein 1 (PD-1), and inducible T-cell costimulatory (ICOS), receptors known to bind other B7 family ligands (2, 5).

B7-H3 was initially considered necessary for T-cell costimulation (2). Its activation was associated with worse outcomes in autoimmune and infectious diseases (26). The costimulatory role of B7-H3 is further supported by detection of B7-H3–positive mononuclear cells in acute renal rejection biopsies (27). Finally, B7-H3 expression on mononuclear cells promoted cardiac allograft rejection in mice through CD4⁺ and CD8⁺ T-cell activation, while B7-H3 deletion in combination with immunosuppression decreased allograft rejection (27).

Nevertheless, B7-H3 likely has a predominantly inhibitory role in adaptive immunity, suppressing T-cell activation and proliferation. Indeed, B7-H3 expression on antigen-presenting cells (APC) reduces CD4⁺ and CD8⁺ T-cell activation, as well as effector cytokine release (28). Furthermore, B7-H3 decreases T-cell activation and function by inhibiting the expression of major transcriptional factors, such as NF-κB. The B7-H3 inhibitory function is further supported by the capacity of APCs to stimulate T-cells in B7-H3–deficient mice (7). In addition, B7-H3 silencing accelerates autoimmune encephalomyelitis via T-helper (T_h)-type 1 immune response upregulation (7). Similarly, in graft-versus-host-disease, cardiac allograft transplantation, and hypersensitivity reaction models, B7-H3 blockade with anti-B7-H3 mAb, MIH35, or gene deletion resulted in uncontrolled T_h-type 1 responses and worse outcomes (26, 29, 30).

B7-H3 has also a nonimmunologic function in normal tissues, it is highly expressed on osteoblasts during embryogenesis, and is crucial for osteoblastic differentiation and bone mineralization (31).

B7-H3 was first demonstrated to have antitumor activity in syngeneic murine lymphoma and mastocytoma where it induced CD8⁺ T-cell and NK-cell activation–mediated tumor regression (32, 33). Furthermore, B7-H3 transfection to colorectal cancer cells grafted to mice reduced their metastatic potential and prolonged the survival of tumor-bearing mice (34), indicating that B7-H3 upregulation in tumor cells can lead to immune cell recruitment and activation. Indeed, in mice, B7-H3 has been suggested to bind to the TLT-2 receptor on CD8⁺ T-cells leading to their activation, and IFNγ and IL2 secretion (24).

Such evidence has been rarely found in humans, because an association between B7-H3 expression and favorable clinical course of the disease has been described in only three of the 61 analyzed studies (gastric; ref. 35 and pancreatic cancer; ref. 36), and acute myeloid leukemia; ref. 37). Whether this difference between mouse and man reflects the lack of B7-H3 binding to T-cell TLT-2 receptors (25) in humans and/or the differential expression of the 4 and 2Ig B7-H3 isoforms in the two species, remains to be determined.

B7-H3 appears to promote tumorigenesis mainly via immunologic mechanisms. Preclinical and clinical evidence indicates that B7-H3 inhibits TA-specific immune responses, leading to a protumorigenic effect. Indeed, B7-H3 expression in tumors correlates with aggressive biology, low tumor-infiltrating T-lymphocyte density, and poor prognosis (38, 39). Nevertheless, B7-H3–targeting strategies to date do not focus on blocking this inhibition, but rather on eliminating B7-H3–expressing cells. This tumorigenic effect may be mediated by immunoglobulin-like transcript-4 (ILT-4) overexpression on cancer cells. The latter increases B7-H3 expression via PI3K/AKT/mTOR pathway activation (40). B7-H3 is also correlated with activated regulatory T-cell infiltration in non–small cell lung cancer (NSCLC; ref. 41) and suppression of NK-cell–mediated glioma cell lysis (42).

Several mechanisms may explain the dual anti-/protumorigenic role of B7-H3. First, B7-H3, like other checkpoint inhibitor ligands, might interact with both stimulatory and inhibitory receptors (43). Second, the differential B7-H3 function may reflect its different binding affinity for the interacting receptors. Finally, tumor cells may express aberrant B7-H3 isoforms or splice variants with different immune functions.

B7-H3 promotes cancer progression and invasion also through nonimmunologic mechanisms. Indeed, siRNA-induced B7-H3 downregulation on melanoma cells, as well as on breast and prostate cancer cells, decreases their in vitro migration and invasiveness (10, 44). In contrast, high B7-H3 expression on colorectal cancer cells is associated with increased matrix metalloproteinase (MMP)-9 expression, thereby promoting tumor cell migration and invasion. Inhibition of the JAK2/STAT3 pathway, which regulates MMP-9 expression, decreases colorectal cancer cells' in vitro aggressiveness (45). Furthermore, B7-H3 affects the metastatic potential of human melanoma cells in mice most likely by modulating the metastasis-associated proteins, MMP-2 and STAT3 (46). Finally, B7-H3 promotes angiogenesis in colorectal cancer, by upregulating VEGFA expression through NF-κB pathway activation (47).

B7-H3 enhances chemoresistance, as shown by the decreased apoptosis of breast cancer and colorectal cancer cells treated with paclitaxel and oxaliplatin/5-fluorouracil, respectively, via JAK/STAT3/survivin pathway activation (48, 49), or specifically in colorectal cancer by reducing the G₂–M-phase arrest (50). Similarly, B7-H3 silencing increases pancreatic ductal adenocarcinoma (PDAC) cells' gemcitabine sensitivity (51). Therefore, B7-H3 is a potential chemoresistance target.

B7-H3 promotes epithelial-to-mesenchymal transition (EMT) in glioma and hepatoma cells through JAK2/STAT3/Slug pathway activation (52, 53). Finally, B7-H3 affects the metabolism of triple-negative breast cancer (TNBC) cells, decreased B7-H3 expression reduces tumor cell glycolytic capacity and increases their AKT/mTOR inhibitor sensitivity (54). Consistent with these activities, B7-H3 protein expression correlated to significant positive enrichment in EMT, UV response, protein secretion, WNT/β-catenin signaling, and Notch signaling proteins in 378 cancer cell lines included in the Cancer Cell Line Encyclopedia (ref. 55; Supplementary Tables S1–S3).

Immunohistochemical (IHC) staining with mAb, 8H9, of a large number of multiple normal tissues has detected only heterogeneous nonspecific cytoplasmic staining in stomach, liver, pancreas, and adrenal cortex (56). Similarly, we have also reported only weak cytoplasmic IHC staining of salivary gland acinar cells, gastric epithelial cells, and adrenal gland cells (8).

In contrast, B7-H3 was highly expressed with limited heterogeneity in all cancer types tested. Representative results obtained by IHC and flow cytometric analysis of malignant and normal cells stained with the B7-H3–specific mAb, 376.96, are shown in Fig. 3. In a comprehensive literature review, comprising 94 studies on 21 cancer types including 26,703 patients, we found that the cumulative frequency of B7-H3 positivity among all tumors is 60% (Fig. 4). B7-H3 displays high expression on stromal fibroblasts and TAV even in tumors with low B7-H3 expression. Indeed, B7-H3 TAV expression frequency ranged between 86% and 98% in hepatoma, colorectal cancer, renal cell carcinoma, and melanoma, with frequency in ovarian cancer outlying at 44% (unpublished).

In addition, of the 61 studies that associated B7-H3 expression in tumors with prognosis, 67% demonstrated an association of positive/high expression of B7-H3 with poor prognosis, 28% found no correlation, and only three (5%; ref. 35–37) demonstrated an association of positive/high B7-H3 expression with better prognosis (unpublished).

The recent advances in molecular biology and antibody engineering have allowed targeting B7-H3 utilizing multiple effector mechanisms. Most of these strategies have been tested in vitro and in mice, generating encouraging safety and/or antitumor activity data, paving the way for B7-H3–targeting clinical trials (Table 1).

The significant changes induced in cancer cells by silencing B7-H3 (38, 39) and the impressive clinical effect of mAbs blocking checkpoint molecules has provided a strong rationale to test B7-H3–specific inhibitory mAbs in solid tumors. B7-H3 blocking with mAbs has been shown to increase CD8⁺ T-cell and NK-cell tumor infiltration and reduce tumor growth, and/or prolong survival in mouse models of hematopoietic cancers, melanoma (57), colorectal cancer (58), and more recently, ovarian cancer (23). However, translation to a clinical setting has been hampered by lack of human B7-H3–specific blocking mAbs.

B7-H3–specific mAbs have been used as carriers to selectively target radioisotopes to tumors. In an early single-arm imaging study in patients with B7-H3–positive tumors, ¹³¹I conjugated to murine mAb 8H9 (¹³¹I-8H9, omburtamab, Y-mAbs) displayed moderate hepatic uptake (NCT00582608). Although no hepatotoxicity was reported, to avoid potential toxicity and to increase the therapeutic index, in subsequent studies, systemic administration was replaced with compartmental administration. Specifically, in phase I trials, intrathecal omburtamab in metastatic central nervous system neuroblastoma (NCT00089245) and intraperitoneal ¹³¹I-8H9 in desmoplastic small round cell tumors (NCT01099644) was well-tolerated (59). Convection-enhanced brainstem delivery of ¹²⁴I-8H9 to diffuse pontine glioma resulted in negligible systemic exposure and no toxicity (NCT01502917). To facilitate rapid radioactivity clearance, mAb 8H9 was conjugated with the chelator, diethylenetriamine pentaacetate, and radiolabeled with lutetium-177 (60).

Furthermore, the B7-H3–specific mAb 376.96, has been conjugated with ²¹²Pb, a source of α-particles. Ovarian tumor–bearing mice treated with this conjugate, alone or with carboplatin, survived 2–3 times longer than controls (61). This conjugate also inhibited human PDAC cell growth in vitro and patient-derived xenograft growth in mice (62).

The antibody–drug conjugate approach has been tested utilizing MGC018 (humanized B7-H3 mAb with a cleavable linker-duocarmycin payload; MacroGenics), which delivers duocarmycin to tumors. A phase I/II trial is assessing its safety alone or in combination with an anti-PD-1 mAb in B7-H3–expressing solid tumors (NCT03729596). DS-7300a (Daiichi Sankyo), a B7-H3–specific mAb conjugated to four topoisomerase I inhibitor particles is being tested in a phase I/II trial (NCT04145622).

The fully humanized mAb enoblituzumab (MGA271; MacroGenics), bearing an Fc domain engineered to enhance its antitumor function by increasing binding to the activating receptor, CD16A, and reducing that to the inhibitory receptor, CD32B (63), was the first mAb tested against B7-H3–expressing tumors. Enoblituzumab was effective in multiple cancer types through antibody-dependent cellular cytotoxicity (63). Interim results from a phase I trial (NCT01391143) including patients with B7-H3–expressing tumor or TAV cells, demonstrated that enoblituzumab was well-tolerated with no dose-limiting toxicity (64). A phase II trial is assessing neoadjuvant enoblituzumab in prostate cancer (NCT02923180). Furthermore, a phase I trial (NCT02475213) demonstrated that enoblituzumab with pembrolizumab in head and neck squamous cell cancer (HNSCC), NSCLC, urothelial cancer, and melanoma was safe; the frequency of immune-related adverse events was comparable with pembrolizumab alone (65). Enoblituzumab with ipilimumab was also tested in a phase I trial in NSCLC and melanoma (NCT02381314, results pending). More recently, a novel Fc-enhanced bispecific anti-B7-H3 mAb/PD-1 fusion protein showed promising results in mice (66).

Bispecific antibodies (BsAb) have been tested utilizing a B7-H3 mAb single-chain variable fragment (scFv) linked to an anti-CD3 mAb scFv to recruit and activate T-cells against tumor cells (67). Only obrindatamab (MGD009, MacroGenics), a humanized CD3xB7-H3 dual affinity protein, has been tested in humans in advanced B7-H3–expressing tumors (NCT02628535). In 2018, the FDA imposed a partial hold on this study due to hepatic adverse events likely due to cytokine release syndrome, secondary to increased T-cell activation. These events were uncomplicated and short-lived, and the hold was lifted in 2019. However, the trial was terminated in November 2019 by MacroGenics due to administrative reasons (E. Bonvini, personal communication). Another B7-H3xCD3 BsAb created by coupling an anti-human B7-H3 mAb with an anti-CD3 mAb showed potent cytotoxicity toward hematologic malignant cells in vitro (68).

Tri-specific killer engagers (TriKEs) form an antigen-specific immunologic synapse between NK and tumor cells, thereby triggering NK-cell–mediated tumor cell lysis (69). TriKEs are composed of either three scFvs of antibodies with different specificity or two scFvs (CD16 specific and TA specific) and a cytokine, most frequently IL15. Vallera and colleagues have generated a B7-H3/IL15 TriKE using the scFv of the B7-H3–specific mAb 376.96. This strategy demonstrated significant tumor burden decrease in vitro and in mice against HNSCC, PDAC, and ovarian cancer (70, 71). A second-generation TriKE has been bioengineered by the same group, with human IL15 as a modified cross-linker between an anti-B7-H3 scFv and a humanized camelid anti-CD16 single-domain antibody. The latter allows improved function of the IL15 moiety, improving NK-cell activation and proliferation, and augmenting killing of ovarian cancer cells in vitro and in mice (72).

Chimeric antigen receptor (CAR) T-cells permit rapid generation of polyclonal T cells with TA specificity and potent cytotoxicity, and can recognize tumor cells independently of HLA class I antigen expression (73). B7-H3 CAR T-cells engineered with various B7-H3–specific scFvs have demonstrated potent in vitro antitumor activity against multiple cancer types (8, 74–77). However, their efficacy remains limited in mice. This limitation, which is common with CAR T-cells targeting different TAs in mice and humans, likely reflects the negative impact of TME escape mechanisms on interactions between tumor cells and CAR T-cells (78). To counteract this limitation, combinatorial strategies with agents which can enhance the antitumor activity of CAR T-cells and/or tumor cell susceptibility to the effector mechanism are being tested. Preclinical studies suggest an acceptable safety profile. B7-H3 CAR T-cells are being evaluated in trials targeting glioblastoma (NCT04077866) and pediatric glioma (NCT04185038). More recently, NK cells have been used to generate CAR NK-cells, which have controlled growth of human NSCLC cells grafted in mice and prolonged their survival (79).

B7-H3 has a complex, still not entirely clear, role in the TME with several immunologic and nonimmunologic functions. Endeavors to identify its receptor are of outmost importance; this information will elucidate its role in the TME, but, most importantly, will facilitate the development of B7-H3–blocking reagents. To date, this area remains unexploited, because most B7-H3–targeting strategies focus on eliminating B7-H3–expressing tumor cells. Blocking the B7-H3–mediated inhibition on T-cells is expected to greatly improve TA-specific immune responses and, in turn, significantly ameliorate the disease course. As observed with other checkpoint inhibition strategies, B7-H3–blocking mAbs may prove to be a significant addition to the antitumor strategy armamentarium.

B7-H3 is highly expressed, with limited heterogeneity, on cancer cells in primary tumors. Additional data on B7-H3 expression in metastases, crucial for designing B7-H3–targeting therapies, are needed. Other areas needing additional investigation include (i) the assessment of the value of B7-H3 serum level as a diagnostic, prognostic, and monitoring marker and (ii) the evaluation of B7-H3–targeting ability to eliminate CICs, given their postulated role in metastasis and recurrence.

The attractiveness of B7-H3 as a target for cancer antibody-based therapy has stimulated the development of B7-H3–targeting immunotherapeutic strategies. Preclinical results are encouraging. In addition, data from clinical trials demonstrate a favorable safety profile with limited toxicity. Multiple ongoing clinical trials are evaluating the therapeutic efficacy of B7-H3–targeting strategies, alone or in combination with other checkpoint inhibitors. This information will assess its value as a target of antibody-based immunotherapy.

No disclosures were reported.

We would like to thank Dr. Ezio Bonvini (Senior Vice President, Research and Chief Scientific Officer, MacroGenics) for reviewing the article, for his constructive suggestions, and for the update he provided regarding the status of the clinical trials conducted by MacroGenics. This work was supported by the NIH (grant RO1DE028172 to S. Ferrone), by the NCI (grants RO3CA223886 to S. Ferrone and RO3CA231766 to S. Ferrone and C.R. Ferrone), and by the Department of Defense (grants W81XWH-16-1-0500 and W81XWH-20-1-0315 all to S. Ferrone).

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.