A T-cell–engaging B7-H4/CD3-bispecific Fab-scFv Antibody Targets Human Breast Cancer Free

Because of the recent success of immune checkpoint–blocking antibodies, clinical trials are underway to evaluate their efficacy in various cancers (1–4). However, a majority of patients with cancer are unlikely to benefit from anti- programmed death-1 (PD-1)/PD-ligand 1 (PD-L1) antibody treatment because the response rate is approximately 20%–40% even in PD-L1⁺ cancers.

In addition to these immunomodulatory receptor blockade therapies, other modulating technologies have been developed (5, 6). MHC and T-cell receptor (TCR)-bypassed T-cell cytotoxicity was first reported in 1985 (7), and over the past 3 decades, CD3 bispecific antibodies (BsAb) and chimeric antigen receptor (CAR) T cells have been developed (8–10).

Recently, two BsAbs catumaxomab (11) and blinatumomab (12, 13) were approved by the FDA, and more BsAbs directly engaging immune cells against tumor cells are now in clinical studies. Improvements in protein engineering technology have enabled the creation of various types of artificial antibodies with greater flexibility in design, size, specificity, half-life, and distribution, and dozens of BsAb formats have been proposed (6, 14).

B7 homolog 4 (B7-H4, VTCN1) is considered to be a negative regulator of immune responses and is overexpressed in many human cancers, which indicates that B7-H4 might be a potential target for cancer therapy. B7-H4 expression is reported to be affected by the tumor microenvironment (15, 16). Previously, we generated an anti-human B7-H4 mAb that induced T-cell cytotoxicity to a B7-H4⁺ breast cancer cell line in an in vitro indirect ADCC system, but the antibody did not suppress the tumor growth in a mouse model (17).

In this study, we manufactured B7-H4/CD3 BsAbs based on the antibody-binding fragment (Fab) and single-chain variable fragment (scFv) structure of anti-B7-H4 and anti-CD3 mAbs using a human cell–based protein expression system. We found that the anti-B7-H4/CD3 Fab-scFv antibody had potent cytotoxic activity against a B7-H4⁺ breast cancer cell line in vitro, and in vivo using a MHC-double knockout (dKO) NOG mouse model (18). Here, we revealed that the anti-B7-H4/CD3 BsAb might contribute to the development of novel therapeutic antibodies against solid tumors.

A comprehensive gene expression analysis was performed as previously described in the HOPE (High-tech Omics-based Patient Evaluation) project at the Shizuoka Cancer Center (Shizuoka, Japan; ref. 19). Briefly, the tumor tissue samples were dissected from fresh surgical specimens, and the RNA samples with an RNA integrity number ≥6.0 were used for the microarray analysis. RNA was amplified, labeled, and hybridized to the Sure Print G3 Human Gene Expression 8 × 60 K v2 Microarray (Agilent Technologies) and Microsoft Excel software. Data analysis was performed using GeneSpring GX Software (Agilent Technologies). Ethical approval for the study was obtained from the institutional review board at the Shizuoka Cancer Center (Shizuoka, Japan). Written informed consent was obtained from all the enrolled patients. All the experiments using clinical samples were carried out in accordance with the approved guidelines.

Human breast cancer cell lines (MDA-MB-468, MDA-MB-231, ZR75, SKBR3, HCC-1954, and HCC-1569) and the lung adenocarcinoma cell line (NCI-H2170) were purchased from the ATCC, and were maintained in RPMI1640 (Sigma) supplemented with 10% FBS (Thermo Fisher Scientific). Human mammary epithelial cells (HMEC) were purchased from Lonza Ltd., and were cultured in the growth medium MEBM-CC3150 (Lonza Ltd.).

The mouse anti-human B7-H4 antibody (clone #25) was established in-house as described previously (17). Briefly, the human B7-H4 isoform 1 extracellular domain was constructed and produced in the Expi293 Expression System (Life Technologies) and was immunized in BALB/cA mice. An antibody secreting hybridoma was generated and was screened by a common method using the mouse myeloma cell line P3 × 63ag8.653 (ATCC). The flow cytometric analyses were carried out on FACS Canto (BD Biosciences). HMECs and cancer cell lines were incubated with anti-B7-H4 mAb (clone #25) and were later incubated with a PE-labeled Polyclonal anti-mouse Ig Ab (BD Biosciences) on ice. The following antibodies were used for flow cytometric analysis of the in vivo experiments using humanized mice. For human cell labeling, anti-CD3-PerCP (HIT3a), anti-CD4-PE or anti-CD4-PE-Cy5 (RPA-T4), anti-CD8-PE-Cy5 (HIT8a), anti-CD11b-PE-Cy7 (ICRF44), anti-CD14-PerCP (MφP9), anti-CD19-APC (HIB19), anti-CD25-FITC (M-A251), anti-CD33-PE (WM53), anti-CD45-FITC (2D1), anti-CD45RA-FITC (HI100), anti-CD45RO-APC (UCHL1), anti-CD56-PE (B159), and anti-CD127-PE-Cy7 (A019D5) were purchased from BD Pharmingen. The anti-FoxP3-PE (hFOXY) antibody was purchased from eBioscience, Inc. The anti-TIM3-PE (F38-2E2) and anti-LAG3-FITC (17B4) antibodies were purchased from Miltenyi Biotech and AdipoGen Life Sciences. The anti-mouse CD45 antibody used to label the mouse cells was purchased from BD Pharmingen. The anti-PD-1-APC (EH12.2H7) and anti-Ki67-PE-Cy7 (Ki-67) antibodies were purchased from BioLegend. Splenocyte and peripheral blood cells were isolated using ACK lysis buffer. Tumor-infiltrating lymphocytes (TIL) were also separated from the control or antibody-treated tumors by anti-human CD45-microbeads (Miltenyi Biotec) using autoMACS System (Miltenyi Biotec). The staining method was described previously (16). Human cells were identified by gating human CD45⁺ fractions.

The mouse anti-human B7H4 mAb clone #25–derived VH and VL genes were cloned, and construct containing Fab (B7-H4)-scFv (CD3) and scFv (B7-H4)-scFv (CD3) linked by a (Gly4Ser)3 linker and 6 × histidine-tag was designed (Fig. 2A). These cDNAs were chemically synthesized and cloned into the expression vector pcDNA3.3. The B7-H4/CD3 BsAbs were produced using the Expi293 Expression System (Gibco, Thermo Fisher Scientific) at a ratio of 3:7 (VH-containing long fragment:VL-containing short fragment) in the Fab-scFv format, purified with a histidine tag affinity column, and used for experiments.

hPBMCs were isolated from the peripheral blood of healthy volunteers or patients with glioma as effector cells using Ficoll-Paque PLUS (GE Healthcare UK Ltd). Effector cells were incubated with cancer cell lines or HMEC at an effector/target (E/T) ratio ranging from 1.25 to 40 in the presence of various concentrations of Fab-scFv or scFv-scFv B7H4/CD3 BsAb at 37°C for 16 or 24 hours in 5% CO₂ atmosphere. The supernatant from the cultures were collected and measured using a Lactate Dehydrogenase Cytotoxicity Assay Kit (Takara Bio Inc.). The percentage of specific lysis was determined by the following formula: percentage of specific lysis = [(effector cells and target cells and agent release – effector cells release) – spontaneous target cell release]/(maximal target cells release − spontaneous target cells release) × 100. EC₅₀ was calculated using a 4-parameter logistic curve fitting using ImageJ (ver. 1.51J8, NIH, Bethesda, MD). The effector T-cell subsets were isolated or depleted from the healthy volunteer PBMCs by an autoMACS Magnetic Cell Isolator (Miltenyi Biotec), and secreted Granzyme B and IFNγ in culture supernatants were measured by an ELISA Kit (Mabtech AB, BioLegend).

The MHC-dKO NOG mice were kindly supplied by Dr. Mamoru Ito at the Central Institute for Experimental Animals (Kawasaki, Japan). All the animals were cared for and treated according to the Guidelines for the welfare and use of animals in cancer research, and the experimental procedures were approved by the Animal Care and Use Committee of Shizuoka Cancer Center Research Institute (Shizuoka, Japan). Clinical experiments using the PBMCs derived from patients with glioma and healthy volunteers were approved by the Institutional Review Board of Shizuoka Cancer Center (Shizuoka, Japan). All the patients provided written informed consent.

For in vivo imaging, all the tumor-transplanted MHC-dKO NOG mice were supplied with a low fluorescence feed for more than 1 week. Cy5.5 labeling of the Fab anti-B7-H4/CD3 antibody was performed using a Cy5.5 Labeling Kit (GE Healthcare UK Ltd). Cy5.5-labeled Fab-scFv B7-H4/CD3 BsAb localization was performed using the Optix MX2 Laser Scanner System (Advanced Research and Technologies) with excitation at 670 nm and emission at ≥700 nm. The Cy5.5-labeled B7-H4/CD3 antibody was injected intravenously and imaging was performed at sequential timepoints ranging from 24 hours to 28 days.

In the pharmacokinetic study of anti-B7-H4/CD3 BsAb, 5 9-week-old BALB/cA mice were injected with 100 μg of Cy5.5-labeled Fab-scFv anti-B7-H4/CD3 BsAb via tail vein, and then blood was drawn at timepoints ranging from 2 minutes to 48 hours after the antibody injection. Serum samples were stored at −80°C until the BsAb concentrations were measured by a sandwich ELISA or fluorescence intensity. Sandwich ELISA was performed using the recombinant B7-H4 extracellular region and horseradish peroxide–labeled polyclonal anti-human Ig antibody (GE Healthcare). Serum BsAb concentration was also determined by fluorescence intensity levels using the Optix MX2 imager and was performed using the ART Optix Optiview Software (Advanced Research and Technologies).

Humanized MHC-dKO NOG mice production method was reported previously (20). Briefly, 8-week-old MHC-dKO NOG mice were irradiated with X-rays and 1 × 10⁷ hPBMCs from patients with glioma were intravenously administered to each mouse via tail vein. The study design for the experiment evaluating the mice treated with Fab-scFv B7H4/CD3 BsAb is shown in Figs. 5 and 6 and Supplementary Fig. S5. Four in vivo experiments were performed (dose-response, short- and long-term antitumor effect evaluation, and T-cell subset depletion). Specifically, we set the starting day of the antibody injection as day 0. As shown in Fig. 5A (the short-term antitumor effect experiment), on day -14, 1 × 10⁶ MDA-MB-468 human breast cancer cells (B7-H4⁺) or NCI-H2170 lung adenocarcinoma cells (B7-H4⁻) were subcutaneously injected into the fat pad or flank region of the mice. Starting on day 0, each antibody was administered intravenously. The antitumor activity was evaluated by measuring the tumor volume. Tumor volume was calculated based on the NCI formula as follows: tumor volume (mm³) = length (mm) × [width (mm)]² × 1/2.

One week after the antibody injection, tumors, spleens, and blood were harvested from the groups. The tumors from one set of 3 mice were used for TIL flow cytometry, IHC analysis, and qRT-PCR of the immune response–associated genes. The schema for the long-term antitumor effect evaluation and T-cell subset depletion experiment are shown in Fig. 5B and C. For T-cell subset depletion in vivo, anti-CD4 and anti-CD8 mAbs were purchased from Bio X Cell.

In the dose–response experiment, as shown in Fig. 6, humanized MDA-MB-468 tumor-transplanted mice were administered with BsAb sequentially in a dose-escalating manner (0.2–200 μg). Two weeks after the start of antibody injection, the tumors were resected and used for IHC analysis.

Additional in vivo experiment targeting B7-H4⁺ breast cancer cell lines, such as HCC-1954 and HCC-1569, using the humanized MHC-dKO NOG mice was performed in the same method as shown in Supplementary Fig. S5C.

The xenografts were harvested 1 or 2 weeks after the injection of anti-B7-H4/CD3 BsAb. Formalin-fixed, paraffin-embedded tissue blocks and sections were made. anti-B7-H4 antibody (clone #25 in-house made; ref. 17), anti-CD8 (C8/144B), and anti-CD4 (4B12) antibodies (Thermo Fisher Scientific), anti-granzyme B antibody (GrB-7, Dako), anti-FoxP3 antibody (236A/E7, Abcam), anti-CD204 antibody (SRA-C6, TransGenic Inc.), and anti-PD-L1 antibody (28–8, Abcam) were purchased and used for IHC analysis. Positively stained cell frequency was counted using the image-analyzing software, ImageJ (NIH, Bethesda, MD) in randomly selected 1/3 areas of a tumor section whole image at 200 × magnification. The necrotic area was excluded.

Human breast cancer paraffin-embedded tissue arrays were purchased from US Biomax, Inc., (catalog no BR1503f) and human normal and tumor paraffin tissue arrays were purchased from BioChain Institute Inc., and were used for IHC study using anti-B7-H4 mAb #25.

For in vivo studies, the intergroup differences were assessed by two-way ANOVA with Shirley–Williams test. Significant difference in the positive cell frequency by IHC was assessed using two-tailed unpaired Student t test. The correlation between different gene expression levels was analyzed using a Spearman coefficiency test. P = <0.05 was considered significant.

High expression of B7-H4 was frequently observed in breast and ovarian cancer and was partially observed in lung cancers from the HOPE project using 2,527 surgically resected tumor tissues (Fig. 1A; Supplementary Fig. S1A). The positive rate of B7-H4 mRNA expression in breast cancers (more than 2-fold upregulation in tumors compared with normal tissues) was 56%. In contrast, PD-L1 expression was low in breast and ovarian cancer tissues. B7-H4 mRNA expression was detected in various types of cell lines, especially in breast cancer cell lines (Supplementary Fig. S2A) by qPCR, and B7-H4 protein expression was detected in 43.5% of breast cancer tissues by IHC using breast tumor tissue array (Supplementary Fig. S7). B7-H4 cell surface expression was observed on breast cancer cells and not on HMEC (Fig. 1B; Supplementary Fig. S2C and S2D).

Unstable B7-H4 surface expression was suggested in some reports (21, 22). For example, breast cancer cell line SKBR3 changed B7-H4 surface expression under confluent culture conditions (Supplementary Fig. S2B).

Two types of recombinant BsAbs were constructed from the novel anti-human B7-H4 mAb clone #25 (17) and classical anti-CD3 antibody (clone: OKT3). The anti-B7-H4/CD3 Fab-scFv format that connected the antigen-binding fragment (Fab), including the human IgG4 consensus region 1, with the scFv was constructed by a single-short chain and a single-long chain (Fig. 2A and B). Therefore, 90 kDa molecular size was larger than albumin, which contributes to the prevention of renal leakage and to the stabilization of anti-B7-H4 affinity. The anti-B7-H4/CD3 scFv-scFv single-chain bispecific format was constructed in a manner similar to BiTE (23). The Fab-scFv anti-B7-H4/CD3 BsAb labeled with Cy5.5 showed positive staining for B7-H4 in positive breast cancer cell lines by flow cytometry (Fig. 2C).

BsAb-mediated crosslinking of B7-H4 on the target cell surface with CD3 on T cell causes effector T-cell–dependent lysis of the target. The cytotoxic activity of the B7-H4/CD3 BsAbs with human PBMCs against MDA-MB-468 cells for scFv-scFv was 67 ng/mL (EC₅₀). For Fab-scFv, it was 12 ng/mL after 16 hours, and for Fab-scFv it was 0.23 ng/mL after 24 hours in vitro (Fig. 3A). The Fab-scFv anti-B7-H4/CD3 antibody showed antibody-dependent cytotoxicity on B7-H4⁺ cells, whereas no cytotoxic activity was seen against B7-H4⁻ cancer cell lines (Fig. 3B; Supplementary Fig. S5B).

There was no significant difference in cytotoxic activity (EC₅₀ value) against MDA-MB-468 cells between healthy volunteer–derived PBMCs and glioma patient–derived PBMCs (data not shown). The anti-B7-H4/CD3 BsAb and volunteer PBMCs showed no cytotoxic effect on normal HMEC (Fig. 3C). The BsAb did not kill target cells without PBMCs (Fig. 3D and E). The effector cell killing activity was elicited by BsAb at a dose above 1 ng/mL against MDA-MB-468 cells after 16 hours in vitro, and a decrease in the killing activity was observed under high concentrations of BsAb (Fig. 3C–E), which might be caused by a decrease in crosslinking of the target molecules because of BsAb saturation. Unexpectedly, positively isolated CD4⁺ as well as CD8⁺ T-cell subsets showed strong cytotoxicity against B7-H4⁺ tumor cell lines with granzyme B and IFNγ secretion by the stimulation of BsAb. Interestingly, even CD4 and CD8 double–negative T cells showed a weak cytotoxicity. However, granzyme B and IFNγ were not produced (Fig. 3F; Supplementary Fig. S3A and S3B).

BsAb accumulation at the MDA-MB-468 tumor site occurred within 24 hours after the antibody injection, and the signal was detected even 28 days after the antibody injection. In contrast, specific antibody accumulation was not recognized in B7-H4⁻ NCI-H2170 tumor (Fig. 4A and B; Supplementary Fig. S4). The BsAb concentration in the serum was determined by a recombinant B7-H4 and an anti-human-IgG-Ab sandwich ELISA, and the T1/2-beta was 8.5 hours. Twenty-four hours after the 100 μg/body BsAb injection, the serum concentration was estimated at 0.1 μg/mL (Fig. 4C), but the serum concentration by fluorescent imaging of the Cy5.5-labeled BsAb was approximately 1 μg/mL after 24 hours (Supplementary Fig. S4), which was higher than the antibody level obtained by the sandwich ELISA method. The BsAb in the sera may partially exist in complex with other proteins.

A single injection of Fab-scFv B7-H4/CD3 BsAb, triple-negative breast cancer (TNBC) MDA-MB-468 xenograft tumors decreased in size by about 70% in hPBMC-transplanted humanized NOG mouse model in a week (Fig. 5A). In the long-term experiment, the reduction in tumor size was maintained until 2 weeks after the BsAb injection (Fig. 5B). In contrast, the growth of the tumors treated with full body anti-B7-H4 antibody (clone #25) was not inhibited. In addition, no antitumor effect was observed in B7-H4⁻ NCI-H2170 tumors treated with BsAb (Fig. 5A). The BsAb also induced T-cell cytotoxicity to HER2⁺ and B7-H4⁺ breast cancer cell lines, HCC-1954 and HCC-1569 in vitro (Supplementary Fig. S5A and S5B), and BsAb injection suppressed HCC-1954 and HCC-1956 tumor growth by more than 50% in the humanized mouse model (Supplementary Fig. S5C).

A relatively high dose of anti-CD3 antibody (clone: OKT3), formerly used in the clinic as an immunosuppressive agent, resulted in no inhibitory effect and obvious weight loss. The BsAb-treated mice showed no adverse effects such as weight loss. An escalating BsAb dose (0.2–200 μg/body) was administered without harm (Fig. 6A).

In the T-cell subset depletion in vivo study, CD8⁺ T-cell depletion blocked the growth inhibition induced by B7-H4 BsAb, and CD4⁺ T-cell depletion showed only a weak blocking effect on growth inhibition (Fig. 5C). These results demonstrated that the antitumor effect of anti-B7-H4 BsAb was mediated partially by CD4⁺ T cells and mostly by CD8⁺ T cells.

From the flow cytometry analysis of TIL and splenocytes from the antibody-treated mice, the total cell number and CD3⁺CD45⁺ T-cell number from spleens showed a tendency to increase in the BsAb-treated mice compared with the controls (Supplementary Table S1), but there were no significant differences in the TIL T-cell subset (data not shown).

In the Fab-scFv anti-B7-H4/CD3 BsAb-administered group, hematoxylin and eosin (H&E)- stained tumor specimens showed remarkable infiltration of lymphoid cells inside the tumor cores and resulted in almost no viable cancer cells at the tumor site (Fig. 6B). Immunostaining revealed that CD8⁺ and granzyme B⁺ lymphocytes were more frequently observed in the BsAb-treated group (Fig. 6B and C), but CD4⁺ and FoxP3⁺ lymphocytes and CD204⁺ immune cells did not significantly differ (data not shown).

In 28 human normal and tumor tissues array analysis, seven B7-H4⁺ tumors (pharynx, esophagus, stomach, lung, kidney, fallopian tube, and kidney) were identified (Supplementary Fig. S6). In the normal tissues, only the tonsil (epithelial cell) was positively stained with the anti-B7-H4 antibody clone #25. The B7-H4 IHC study of human breast cancer tissue array showed that a B7H4⁺ rate of 69 breast cancer tissues was 43.5%, and there was a tendency of high B7H4 stain in triple-negative tumor compared with PD-L1 (Supplementary Fig. S7). However, it was not definite because of small number of cases.

Because of the recent success of immune checkpoint blocking antibodies, such as ipilimumab and nivolumab, in patients with metastatic melanoma and other malignancies, clinical trials are underway to evaluate their efficacy in various solid cancers (1–4). These studies show that the modulation of the suppressed immune system is an effective way to combat cancers and that immunotherapy can be used to treat cancer.

In addition to immune checkpoint antibodies, BsAbs that directly engage immune cells (24) are becoming another promising strategy in cancer antibody therapy. Advances in recombinant protein technology allow for the construction of BsAbs in a variety of formats with great flexibility. A number of formats have been proposed including bispecific T-cell engager (BiTE; refs. 25, 26), tandem diabody (27), dual affinity retargeting (DART; ref. 28), and antibody-TCR format (ImmTac; ref. 29). Catumaxomab-targeting EpCAM and blinatumomab-targeting CD19 are approved for clinical use and others are in various stages of clinical development (30–33).

New T-cell–engaging BsAbs, including BiTE and DART, work at a much lower dose in xenograft models and clinical use (23, 28) compared with conventional antibody therapies. Our BsAb, which connects the anti-B7-H4 antibody and anti-CD3 antibody, immediately elicited a strong B7-H4⁺ target tumor lysis at very low dose by CD8⁺ effector T cells or other T-cell subsets, bypassing peptide-MHC/TCR recognition (Figs. 3 and 5).

Trastuzumab and other HER2-targeted agents improved patient outcomes in breast cancer therapy, but initially the responsive tumors develop resistance. HER2-targeted BsAb showed an antitumor potency against trastuzumab-resistant tumor cells in in vitro and in vivo models (32). Its EC₅₀ was inversely correlated with surface HER2 expression and could be effective for tumors that express a lower level of HER2. Given that B7-H4 is frequently expressed on breast cancer cells irrespective of HER2 expression (Supplementary Fig. S8) and that B7-H4 expression has a tendency to be higher in TNBC cells (34), B7-H4 is potentially a novel substitute target for primary and recurrent breast cancers, including triple-negative cancers that are nonresponsive to HER2-targeting therapies. Interestingly, our anti-B7-H4 BsAb showed a potent antitumor effect in vivo against B7-H4⁺ breast cancer cell lines, HCC-1954 and HCC-1569, which was demonstrated to be basal type HER2⁺ and trastuzumab-resistant by some researchers (35, 36). This observation might suggest that this type of breast cancers with positive B7-H4 expression could be the target for B7-H4-targeting therapy.

In addition, based on B7-H4 gene expression and the clinical data from the TCGA tumors samples, high B7-H4 gene expression can be a poor prognostic factor in patients with breast cancer (Supplementary Fig. S1B).

B7-H4 is widely detected in cancers and normal tissues at the mRNA level, but its cell surface expression is limited and tends to be unstable (21, 22). Our IHC study showed a limited expression in normal tissues and positive expressions in several tumor tissues other than breast cancer (Supplementary Fig. S6).

The BsAb eliminated breast cancer cells in a short amount of time (Fig. 5A, 6B), and the immediate antitumor response may help overcome cancers with continuous genomic evolution and immune evasions in contrast to the immune checkpoint blockade antibody therapy, which takes several weeks to induce an adequate amount of tumor antigen–specific effector T cells at the tumor site.

Tumor-specific antigens, such as carcinoembryonic antigens, cancer/testis antigens, or differentiation antigens, are also expressed in normal tissues to varying degrees, except for mutation neoantigens. Therefore, the therapeutic approaches targeting these antigens require antibody dose control for harnessing cytotoxic effector cells. B7-H4-targeted CAR-T–cell treatment showed lethal toxicity including a delayed GVHD in a mouse model (37), and a clinical trial using trastuzumab-based CAR-T cells resulted in patient death and was aborted (38). These reported adverse events suggest that the strict regulation of effector function is important for clinical use. A shorter in vivo half-life of the smaller size BsAb is described as a negative property, but this characteristic enables easier control of effector cells for clinical use compared with CAR-T–cell therapy.

Immune check point modulators are not effective for the majority of patients with cancer, and positive outcome is likely to require a high mutation burden and effector T-cell accumulation against mutation-derived neoantigens (39, 40). BsAb therapy may be an alternative anticancer immunotherapeutic strategy to bypass neoantigen-MHC/TCR recognition.

In this study, we demonstrated that the B7-H4-targeted BsAb had potent antitumor activity in vitro and in vivo, which was specific for B7-H4⁺ tumors and was not cytotoxic on a normal mammary duct epithelial cell line. More importantly, in vivo imaging showed a long-lasting retention of the injected antibody at the tumor site. These results suggest that the Fab-scFv anti-B7-H4/CD3 BsAb might be a therapeutic agent for PD-L1⁻ B7-H4-expressing tumors or anti-HER2 antibody nonresponsive breast tumors.

Y. Nakasu reports receiving speakers bureau honoraria from Eisai Co., Chugai Co., and Dai-ichi Pharmaceutical Co., and is a consultant/advisory board member for Ono Pharmaceutical Co. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Iizuka, Y. Akiyama

Development of methodology: A. Iizuka

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Iizuka, T. Ashizawa, K. Mitsuya, N. Hayashi, Y. Nakasu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Iizuka, C. Nonomura, R. Kondou, K. Ohshima, T. Sugino

Writing, review, and/or revision of the manuscript: A. Iizuka, K. Yamaguchi, Y. Akiyama

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Nonomura, K. Maruyama, Y. Akiyama

Study supervision: A. Iizuka

We thank Koji Takahashi for his excellent assistance in maintaining the NOG-dKO mouse in the animal facility. This study was supported by JSPS KAKENHI, Japan (grant no. 17K07235; to A. Iizuka).

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.