Expression, Clinical Significance, and Receptor Identification of the Newest B7 Family Member HHLA2 Protein

The past decade has witnessed an important change in the understanding of T-cell biology and tumor immunology with the recognition of immune checkpoints through the B7–CD28 pathways (1–3). B7-1/B7-2/CD28/CTLA-4 is the prototypic B7–CD28 pathway. This loop of costimulation and coinhibition is critical for regulating the early stages of T-cell responses in lymphoid organs. Several additional B7 family members are believed to play important roles in peripheral tolerance and tumor immune evasion—PD-L1 [B7-H1 (4, 5)], B7-H3 (6), and B7x [B7-H4/B7S1 (7–9)]. The ligands PD-L1, B7-H3, and B7x function by inhibiting effector T cells in peripheral tissues (10, 11). These ligands are expressed in various human cancers and their expression can lead to immune tolerance in the tumor microenvironment by inhibiting T-cell proliferation and function (1, 12–14). In addition, B7x can interact with myeloid-derived suppressor cells (15, 16), which may also promote tumor growth. Clinically, higher expression of these ligands is associated with a poor prognosis in various malignancies. On the basis of these functional and clinical observations, blocking some of the B7–CD28 pathways has yielded some therapeutic success in human malignancies. The anti-CTLA-4 antibody achieved clinical responses in melanoma (17), whereas anti-PD-L1 or anti-PD-1 antibodies have shown responses in melanoma, renal cell cancer, and non–small cell lung cancer (18–20). The therapeutic responses seen in these patients are durable and they are longer than chemotherapy or other targeted agents.

HERV–H LTR-associating 2 (HHLA2, also called B7H7/B7-H5/B7y) has been recently discovered as the newest member of the B7 family (21–23) and has 23% to 33% similarity in amino acid sequence with the other B7 molecules (21). This ligand is the only B7 family member that is found in humans but not in mice. It is constitutively expressed on the surface of human monocytes and is induced on B cells. HHLA2 binds to its putative receptor(s) on a variety of immune cells, including CD4 and CD8 T cells and antigen-presenting cells (21). Similarly to B7-H3, both a T-cell coinhibitory role and a costimulatory role have been reported for this ligand (21, 22). There have been no reports published on the protein expression of HHLA2 in normal human tissues or cancers, and the clinical significance of this ligand is not currently understood. Here, we present the first study on the expression of this protein in peripheral tissues as well as on numerous human cancers. Furthermore, we have demonstrated that the overexpression of this protein in tumors is associated with worse clinical outcomes in patients with breast cancer. One of the mechanisms whereby HHLA2 is overexpressed in human cancer seems to be gene copy-number variation (CNV). Finally, we discovered that Transmembrane and Immunoglobulin Domain Containing 2 (TMIGD2) is one of the receptors for HHLA2.

Normal and cancerous tissue microarrays were purchased from Imgenex Corp for the analysis of HHLA2 expression in human tissues. For the TNBC cohort, 50 cases were selected from our tumor registry between 2002 and 2011 who were diagnosed with local or locally advanced breast cancer diagnosed at our institution between 2002 and 2011. All of these patients underwent surgery as the primary treatment followed by chemotherapy or radiotherapy or both. Tissue blocks from the mastectomy or lumpectomy specimen were located and slides were made from the same section. This was done so as to enable us to determine whether HHLA2 expression in the primary tissue is associated with prognostic features. Using retrospective chart review, the relevant clinical data were collected from the files. Hormone receptor status and Her2 status were also obtained by retrospective review. All protocols were reviewed and approved by the Institutional Review Board.

A mouse monoclonal antibody against HHLA2 (Clone 566.1, IgG1) and 3T3 cell lines expressing HHLA2 or CTLA-4 were recently generated (21). Cell blocks of the 3T3 cells expressing either HHLA2 or CTLA-4 were prepared for IHC controls by fixing the cells with 4% paraformaldehyde and then embedding them in HistoGel (Thermo Scientific). These samples were then embedded in paraffin and cut onto slides and used as positive and negative controls, respectively. These controls were then stained for HHLA2 using our mAb. Briefly, 4 to 5-μm thick formalin-fixed paraffin-embedded specimen slides were used. The tissue sections were deparaffinized in xylene and dehydrated through graded alcohols to water. The samples were boiled in citrate buffer (pH, 6.0) using a microwave for 2 minutes and then incubated in an antigen retrieval steam chamber between 80°C and 100°C for 30 minutes. To block endogenous peroxidase activity, all of the sections were treated with 3% hydrogen peroxide for 10 minutes. Nonspecific binding of IgG was blocked by using serum-free protein block solution (Dako). The sections were incubated with anti-HHLA2 mAbs for 30 minutes. They were then incubated with the Dako envision+ HRP-labeled anti-mouse polymer. Signals were generated by incubation with 3,3′-diaminobenzidine. Finally, the sections were counterstained with hematoxylin and observed under the microscope. The slides were reviewed in tandem by a breast pathologist and by an oncology physician trained in breast pathology.

Cell lines were cultured in either complete RPMI or DMEM. Cells were stained with anti-HHLA2 mAb 566.1 and then with a secondary anti-mouse IgG-APC–conjugated antibody. For receptor binding, 3T3 cells expressing HHLA2-YFP or CTLA-4-YFP were incubated with TMIGD2-Ig or control Ig for 45 minutes on ice and then stained with anti-human IgG-APC. Similarly, 3T3 cells expressing TMIGD2-YFP or CTLA-4-YFP were incubated with HHLA2-Ig or control Ig for 45 minutes on ice and then stained with anti-human IgG-APC. Samples were analyzed by a BD FACSCalibur flow cytometer and FlowJo software.

The NCBI database was queried for proteins of the immunoglobulin family with homologs in humans and monkeys, but not in mice or rats. The sequences of the resulting list of proteins were analyzed by various domain-prediction programs to determine if these proteins contained Ig, IgC, IgC-like, IgV, or IgV-like domains. The list was further refined by excluding proteins that did not contain a transmembrane domain. MacVector 10.6. was used for sequence alignment and homology comparison. The phylogenetic tree was generated by PAUP (4.0b10) after removal of significant inserts and trimming C- and N-terminal extensions from sequence alignments (24). Motifs and domains were analyzed with EMBL-EBI tools, SMART, and CBS Prediction. For gene CNVs, the cBioPortal for Cancer Genomics database and TCGA were analyzed.

TMIGD2-Ig fusion protein was prepared by PCR-amplifying the extracellular domain of the protein without the signal peptide. The amplified product was inserted into a human IgG1 Fc tag of plasmid pMT/BiP as described previously (7). Drosophila Schneider 2 cells were cotransfected with this construct and a plasmid inducing hygromycin resistance. The fusion protein was expressed in Express Five serum-free medium (Life Technologies) and purified using Protein G Plus Agarose columns (Pierce). The purity of the fusion protein was confirmed by SDS-PAGE.

For continuous variables, the t test was used when the normality assumption was met and the rank-sum test and median was calculated for those, which violated normality. The χ²test was used to analyze the association between categorical variables and if the expected frequency was less than 5 in more than 20% of the cells, the Fisher exact test was used. All P values < 0.05 were considered statistically significant.

The expression of HHLA2 protein in human tissues is unknown at present. To examine the protein expression, we used our recently generated HHLA2-specific monoclonal antibody clone 566.1 to develop an immunohistochemistry protocol to detect HHLA2 in formalin-fixed paraffin-embedded specimens in which 3T3 cells expressing HHLA2 or CTLA-4 served as positive and negative controls, respectively (Supplementary Fig. S1). We used this technique to stain tissue microarrays of normal human organs (Table 1). Our results demonstrated that the majority of normal organs did not express HHLA2 at the protein level; however, trophoblastic cells of the placenta and epithelial cells of the gut, kidney, gallbladder, and breast expressed this ligand (Fig. 1 and Table 1). Although primary and secondary lymphoid organs were largely negative, a few scattered cells appeared to stain positively in these samples. These results reveal that endogenous HHLA2 protein is absent in most normal tissues, but mainly expressed on epithelial cells of a few tissues.

HHLA2 is able to inhibit both human CD4 and CD8 T-cell functions (21), therefore, it is possible that human cancer may exploit this pathway as an immune-evasion mechanism. As no knowledge currently exists about HHLA2 protein in human cancer tissues, we next examined the expression of HHLA2 in human cancer tissues using IHC staining of tissue microarrays of common cancers from various organs (Fig. 2). HHLA2 was expressed in 50% or more of cancer samples from the breast (7 of 10), lung (6 of 9), thyroid (6 of 9), melanoma (5 of 9), ovary (4 of 8), and pancreas (5 of 10) (Table 2). The localization of the protein was membranous and cytoplasmic in tumor cells (Fig. 2). HHLA2 protein was also expressed in cancers of the liver, bladder, colon, prostate, kidney, and esophagus (Table 2 and Fig. 2). In addition, we found that of the 20 human cell lines examined in breast cancer and hematologic malignancies of leukemia and lymphoma, 12 expressed HHLA2 on their surface by flow cytometry (Supplementary Table S1). These results demonstrate that HHLA2 protein is widely overexpressed in human cancers and has a high prevalence in certain malignancies.

We also evaluated HHLA2 expression in a cohort of 50 patients with early-stage TNBC because of the limited therapeutic options for this breast cancer subtype. The characteristics of the patient population are shown in Table 3. The mean age of the patient population was 57.6 years, average tumor size was 2.35 cm (SD, 1.49), 70% had at least one positive lymph node (n = 35), and the distribution of American Joint Committee on Cancer (AJCC) Stage was stage I in 20%, stage II in 50%, and stage III in 30%. All patients underwent primary therapy (i.e., mastectomy or lumpectomy and radiotherapy), and most of them (92%) received adjuvant chemotherapy.

Using IHC staining, we found that within a given TNBC tumor, the expression and intensity of HHLA2 protein was quite homogeneous throughout the tumor. Hence, HHLA2 expression was graded on the basis of intensity of staining between 0 and 3. Grades of 0 and 1 had no or minimal staining, respectively, whereas 2 and 3 had moderate and strong membranous/cytoplasmic staining. Zero and 1 were considered to be negative or low expression, whereas 2 and 3 were considered to be high expression (Fig. 3).

In the 50 TNBC samples stained, HHLA2 was graded as 0 in 12% (n = 6), 1 in 32% (n = 16), 2 in 40% (n = 20), and 3 in 16% (n = 8) of the tissue sections (Table 3). When classified as a binary variable, 56% (n = 28) exhibited high (grades 2 or 3) and 44% (n = 22) had low expression (grades 0 or 1) of HHLA2. In the bivariate analysis, high HHLA2 expression was associated with lymph node positivity (0 vs. ≥1, P = 0.04) and advanced stage of the disease (P = 0.03) at the time of diagnosis, features known to be associated with an increased risk of recurrence. HHLA2 expression was not related to age or to the size of the tumor. Together, these data demonstrate that more than half of TNBC tumors have HHLA2 overexpression and that patients with higher levels of HHLA2 on their tumors are significantly more likely to have the disease spread and at an advanced stage.

HHLA2 was overexpressed at the protein level in breast cancer, but the mechanism(s) upregulating the expression in cancer cells is unknown. Therefore, we sought to determine whether gene amplification was a potential mechanism of overexpression. By analyzing the cBioPortal for Cancer Genomics database (25, 26), we found that HHLA2 gene alterations were present in 18.8% and 23% of all cases with breast cancer in TCGA (27) and in the TCGA provisional database studies, respectively. Because TNBC is predominantly comprised of the basal subtype, we compared the copy-number gain of the HHLA2 locus in basal with nonbasal breast cancer using the TCGA registry. HHLA2 was altered in 32% of the basal subtype, which is almost twice the frequency observed in all breast cancers independent of their subtype (18%). The vast majority of HHLA2 CNVs in basal breast cancers were amplifications or gains (>95%). Hence, given that HHLA2 protein upregulation is present in approximately 56% of our samples, our results suggest that one mechanism of upregulation of HHLA2 protein in TNBCs is gene copy-number gain.

Receptors for HHLA2 are widely expressed on both naïve and activated T cells as well as dendritic cells, monocytes, and B cells (21). As HHLA2 is a member of the immunoglobulin superfamily and has orthologs in humans and monkeys but not in mice or rats, we hypothesized the receptors for HHLA2 may belong to the immunoglobulin superfamily and have the same phylogenetic pattern due to coevolution.

From more than 500 members of the immunoglobulin superfamily (28, 29), a list was compiled of the immunoglobulin family members expressed in humans and monkeys but not in mice or rats. This list was further refined by only including members with predicted transmembrane domains and we then stably transfected these candidates into 3T3 cells. We tested their ability to bind to the HHLA2-Ig fusion protein. Using flow cytometry, we found that HHLA2-Ig bound to cells expressing Transmembrane and Immunoglobulin Domain Containing 2 (TMIGD2, Fig. 4A and B). A TMIGD2-Ig fusion protein was then constructed in which the extracellular domain of TMIGD2 was linked to the Fc region of human IgG1 (TMIGD2-Ig). TMIGD2-Ig bound strongly to 3T3 cells expressing HHLA2 but not CTLA-4 by flow cytometry (Fig. 4B). TMIGD2 contains an N-terminal signal peptide, an extracellular IgV-like domain, three potential sites for N-linked glycosylation, a transmembrane region, and a cytoplasmic tail with four potential sites for phosphorylation, and a possible site for SH3 domain recognition (Fig. 4A). By sequence analysis, we found TMIGD2, the immunoglobulin-containing and proline-rich receptor-1 (IGPR-1; ref. 30), and CD28 homolog (CD28H; ref. 22) are the same molecule. IGPR-1 was originally identified as a adhesion molecule involved in angiogenesis (30), whereas CD28H was recently reported as a receptor by a high-throughput screen of transmembrane proteins (22). Thus, TMIGD2/IGPR-1/CD28H is one of the receptors for HHLA2.

The B7 ligand family and the CD28 receptor family are the major driving force of T-cell costimulation and coinhibition. These molecules have been intensely studied for their potential clinical impact in human malignancies, especially with regard to ectopic tumor cell expression of negative coinhibitory molecules. Here, we present the first study on the protein expression, clinical significance, and mechanism of upregulation of HHLA2 in human tissues and cancers. The results reveal that HHLA2 is a suitable target for cancer therapy.

HHLA2 appears to have limited expression in normal tissues. Most human tissues we tested were negative for HHLA2 protein. A few organs, including intestines, breast, kidney, gallbladder, and placenta, express HHLA2, particularly on epithelial cells. We previously showed that blood B cells activated by LPS and IFN expressed HHLA2 (21). IHC staining of tissue microarrays of normal human organs showed there were only scattered HHLA2-positive cells in primary and secondary lymphoid organs. Currently, it is unknown whether activated B cells in blood and in tissues have different expression patterns of HHLA2. Tissue-expressed PD-L1, another B7 family member, was recently shown to protect against gut inflammation in mouse models (31). Future investigation is warranted to determine whether HHLA2 in the intestines contributes to the immune tolerance or intestinal inflammation. We observed high expression of HHLA2 in the placental tissue, suggesting that it may play a role in fetal maternal immune tolerance. Interestingly, HHLA2 polymorphisms are associated with Autism spectrum disorders, a disease whose etiology remains poorly understood (32). Immune dysregulation was recently proposed to contribute to the rapid development of Autism spectrum disorder in genetically susceptible children (32). Further study will also be required to determine whether HHLA2 is involved in the development of Autism spectrum disorder.

HHLA2 was highly expressed in most malignant tissues. This wide expression in various tumors indicates that HHLA2 expression may be a critical step in tumor evolution and that HHLA2 could confer a survival advantage to tumors via suppression of host antitumor immunity. Because breast cancer had a high expression of HHLA2, we evaluated its clinical impact in our TNBC cohort. Analysis of HHLA2 expression in the breast cancer cohort revealed that HHLA2 was highly expressed in 56% of TNBC patients; about 80% of TNBCs are basal subtype by gene expression. Tumors with high HHLA2 expression exhibited uniform expression of HHLA2 in the membrane or cytoplasm with minimal intratumoral heterogeneity. In contrast, other B7 ligands such as PD-L1 show focal expression and significant tumor heterogeneity. This uniform expression of HHLA2 in tumors suggests that it may be a primary change in tumors during tumor evolution and/or its expression could be induced by factors in the tumor microenvironment. There are at least two possible mechanisms governing the upregulation of HHLA2 expression: inflammatory stimulation and gene copy-number gain. HHLA2 expression is induced on B cells and enhanced on monocytes by stimulation with LPS and IFNγ (21). Our gene copy-number analysis revealed that the vast majority of HHLA2 CNVs in basal breast cancer were gains (>95%), suggests that gene copy-number gain is one of the mechanisms upregulating HHLA2 expression in cancer. Expression of HHLA2 was associated with two prognostic factors—advanced stage and lymph node positivity—but it was not related to the size of the tumor by bivariate analysis. The association of HHLA2 with lymph node–positive disease suggests that this may be a change that is required for early tumor invasion.

We have previously reported that receptors for HHLA2 exist on a wide variety of immune cells, including T cells, B cells, monocytes, and DCs. We utilized bioinformatics analysis and immunology approaches to determine that TMIGD2/IGPR-1/CD28H is one of the receptors for HHLA2. This strategy was used because of the unique phylogenetic profile of HHLA2 and the observation that ligand and receptor pairs tend to coevolve. It is unlikely, however, that tumor-expressed HHLA2 inhibits the immune system through the interaction with TMIGD2/IGPR-1/CD28H, as CD28H is reported to be expressed on naïve T cells but not on other immune cells and is lost rapidly after activation of naïve T cells (22), whereas tumor-infiltrating T cells are not naïve cells. Further study will be required to discover coinhibitory receptors on immune cells that tumor cell-expressed HHLA2 interacts with to inhibit anticancer immunity.

Because this is a cross-sectional study and this is a small cohort of TNBC patients analyzed and hence, limited inference could be drawn from this cross sectional study. The association of HHLA2 positivity and advanced stage could have been driven by the association of HHLA2 with lymph node positivity. Hence, larger studies are needed to confirm these associations and other important characteristics like disease-free and overall survival in triple negative and other breast cancers.

In summary, HHLA2 has restricted tissue expression, but is overexpressed in various human cancers. Importantly, its expression is associated with important prognostic factors in breast cancer. Furthermore, cancer cells may amplify the HHLA2 gene to increase its expression and to undermine host immunity. Finally, our results suggest that HHLA2 in human cancers is a suitable target for cancer immunotherapies such as checkpoint blockade and antibody–drug conjugate treatment.

No potential conflicts of interest were disclosed.

Conception and design: M. Janakiram, A. Zhao, J.A. Sparano, X. Zang

Development of methodology: M. Janakiram, J.M. Chinai, R. Zhao, A. Zhao, S.C. Almo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Janakiram, J.M. Chinai, S. Fineberg, J.A. Sparano

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Janakiram, J.M. Chinai, S. Fineberg, A. Fiser, C. Montagna, K.C. Ohaegbulam, A. Zhao, J.A. Sparano

Writing, review, and/or revision of the manuscript: M. Janakiram, J.M. Chinai, S. Fineberg, E. Castano, A. Zhao, J.A. Sparano, X. Zang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Janakiram, R. Medavarapu, H. Jeon, R. Zhao

Study supervision: S. Fineberg, J.A. Sparano, X. Zang

Other (reviewed cases, gathered information about patients): R. Medavarapu

Other (reviewed cases as a pathologist): E. Castano

Other (developed the antibody used in the article): R. Zhao

The authors thank Joseph Albanese for his help in devising an immunohistochemistry protocol.

This work was supported by NIH R01CA175495 and Department of Defense PC131008 (to X. Zang). M. Janakiram was supported through CTSA UL1TR000086; J.M. Chinai and K.C. Ohaegbulam were supported by NIH T32GM007288 and F31CA183493; R. Zhao was partially supported by the China Scholarship Council. The authors also acknowledge support from the Albert Einstein Cancer Center (P30CA013330), Diabetes Research Center (P60DK020541), Center for AIDS Research (AI51519), and Institute for Aging Research (P30AG038072).

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.