A mAb against surface-expressed FSHR engineered to engage adaptive immunity for ovarian cancer immunotherapy

Generation and flow cytometry screening of anti–human FSHR antibodies. FSHR is a tumor-associated antigen present in OC (7), prostate cancer (21), and the neovessels of 80% of cancers (22). It is a G protein coupled with 7 transmembrane domains (23). This complex structure presents challenges for classical antigen protein approaches. We generated a codon-optimized sequence of the human FSHR (Figure 1A) for direct in vivo immunization allowing for the generation of responses against a putative native antigen structure on the surface. We subcloned the FSHR cDNA into a characterized expression vector (Figure 1B) and inoculated mice for the generation of antibody responses using direct plasmid injection followed by in vivo electroporation. Animals were immunized biweekly, and sera were collected a week after each immunization for analysis of antibody levels (Figure 1C).

Generation of anti-human FSHR antibodies.Figure 1

Generation of anti-human FSHR antibodies. (A) Depiction of FSHR structure. (B) Cloning strategy into pBMN-I-GFP expression vector. (C) Mouse immunization scheme. (D) cAMP response to different doses of FSH of K562 and K562-FSHR. (E) Western blot of phospho-p44/42 (Erk1/2) and p44/42 (Erk1/2) 20 minutes after stimulation of K562 and K562-FSHR cells using FSH. (F) Partial block of cAMP production in K562-FSHR cells by D2AP11 anti-FSHR antibody upon FSH stimulation of FSHR. Error bars represent mean ± SEM; all the experiments were done in duplicate or triplicate. ANOVA. ***P < 0.001.

To detect anti-FSHR antibodies that would bind to native FSHR expressed in the cell membrane, we stably transduced K562 cells to overexpress human FSHR (K562-FSHR). To validate the correct folding and functionality of the recombinant FSHR, we tested the response of K562-FSHR cells to follicle-stimulating hormone (FSH). As expected, K562-FSHR cells increased their production of cyclic AMP (cAMP) and ERK phosphorylation upon FSH stimulation, but no response was observed in the parental K562 (Figure 1, D and E). In a preliminary experiment, we observed that the anti-FSHR antibody D2AP11 partially blocked cAMP production by FSH stimulation of FSHR. We did not see any difference in cAMP production in K562-FSHR cells only in the presence of D2AP11 antibody without addition of FSH (Figure 1F), suggesting that additional studies in this area are important to understand the mechanism and to further confirm this observation (24, 25).

To monitor the ability of immune sera to bind FSHR, we combined K562 (GFP–) and K562-FSHR (GFP+) cells at equal ratios and added sera diluted up to 1:1,000, followed by anti-mouse IgG APC-conjugated secondary antibody (Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.162553DS1) (26, 27), and determined the fold mean fluorescence intensity (MFI) of K562-FSHR compared with wild-type K562. When a 1:1,000 serum dilution exceeded 20-fold MFI, the last immunization was performed 3 weeks from the previous immunization by boosting with FSHR-overexpressing A20 cells. Boosted animals were sacrificed 4 days later for hybridoma generation as described (28, 29). Two weeks after the fusion, we screened supernatants from fifteen 96-well plates using flow cytometry to analyze the potential hybridomas (Supplemental Figure 1C). The top 20 clones based on fold MFI were expanded for further analysis (Supplemental Figure 1D). We downselected a highly potent clone, D2AP11 (fold MFI 42.2), based on high binding specificity.

We next compared the binding potential of this potent downselected antibody with those of 4 different commercially available mouse anti–human FSHR antibodies at different concentrations ranging from 2,500 ng/mL to 9.77 ng/mL. As shown in Figure 2A, D2AP11 exhibited high specific binding (K562-FSHR cells) and no nonspecific binding (K562 cells). Commercial Ab#1 showed high binding to K562-FSHR cells; however, at concentrations of 2,500, 1,250, and 625 ng/mL, it showed nonspecific binding to non-FSHR-expressing wild-type K562 cells. Commercial Ab#3 showed modest binding at 2,500 and 1,250 ng/mL; however, there was high nonspecific binding to K562 cells. We did not observe FSHR binding in the case of commercial Ab#2 and commercial Ab#4 when evaluated in different concentrations up to 2,500 ng/mL. Notably, D2AP11 mAb showed potent binding at a concentration as low as 9.77 ng/mL. The dose-dependent binding of this potent mAb clone is shown in Figure 2B and represents a new potent tool for human FSHR study.

Binding potency of D2AP11 and commercial mouse anti–human FSHR antibodies.Figure 2

Binding potency of D2AP11 and commercial mouse anti–human FSHR antibodies. (A) Binding of D2AP11 anti-FSHR antibody and 4 different commercial antibodies (commercial Ab#1, #2, #3, and #4) in FSHR-overexpressing K562 cells analyzed by flow cytometry. Only secondary antibody control and irrelevant antibody control are shown at the top, where no binding to K562 and K562-FSHR cells was observed. The binding with different anti-FSHR antibodies was evaluated at different concentrations: 2,500 ng/mL, 1,250 ng/mL, 625 ng/mL, 312.5 ng/mL, 156.25 ng/mL, and 78.125 ng/mL. (B) Dose-dependent binding of D2AP11 anti-FSHR antibody to K562-FSHR cells analyzed by flow cytometry. D2AP11 binding was observed at a concentration as low as 9.77 ng/mL, indicating its potency.

Anti-FSHR antibody binds FSHR with high specificity. We characterized D2AP11 anti-FSHR antibody in detail for specificity by staining a panel of different healthy human tissues, including pancreas, lung, heart, small intestine, colon, uterus, ovary, and fallopian tube endothelium. For most of the healthy tissues, no significant binding of D2AP11 was observed. We observed D2AP11 binding in ovary and fallopian tube endothelium and high binding in different OC tissues (high-grade serous carcinoma, low-grade serous carcinoma, clear cell carcinoma, dysgerminoma, mucinous carcinoma, endodermal sinus carcinoma, and metastatic adenocarcinoma) (Figure 3, A and B). Further studies on additional healthy and OC tissues are important. Based on data from The Human Protein Atlas, we compared the differential RNA expression of FSHR versus ERBB2/Her2; mAbs and small-molecule inhibitors against ERBB2/Her2 have been developed, and currently clinical trials focusing on Her2-targeted bispecific immune cell engagers are ongoing (30, 31). These comparisons on 55 different human tissue types (31) likely support that FSHR deserves additional study for targeted OC immunotherapy (Figure 3C and Supplemental Figure 2). We then further characterized D2AP11 in detail for binding to different FSHR-expressing cells. D2AP11 binds to OC cell lines spontaneously expressing FSHR (7, 32). Cell lines (CAOV3, OVCAR3, and TOV21G) all showed the expected expression of FSHR by D2AP11 staining (Figure 4A). To further confirm that the signal elicited by D2AP11 corresponded to FSHR, we performed CRISPR-mediated deletion of FSHR in the TOV21G cell line. Flow cytometric staining with D2AP11 antibody showed absence of binding to the TOV21G cell line after CRISPR-mediated FSHR knockout (Figure 4B). This clone was highly specific for FSHR, as K562 cells transfected with LHCGR, the homologous protein to FSHR (sequence homology ~46% in the ECD and ~72% in the 7TMD) (33), did not show cross-reactivity by flow cytometry analysis (Figure 4C). We further tested whether D2AP11 was also able to bind murine FSHR, as mouse models of disease are important to study the in vivo efficacy and safety of new therapies. We expressed murine FSHR in mouse tumor lines A20 and ID8-Defb29/Vegf-a and again tested binding of D2AP11 to transfected and untransfected cells by flow cytometry. We found that D2AP11 successfully bound murine FSHR as it did human FSHR (Figure 4D).

Binding of D2AP11 to healthy and OC tissues.Figure 3

Binding of D2AP11 to healthy and OC tissues. (A) Representative images showing the binding of D2AP11 anti-FSHR antibody to healthy tissues of different human organs (pancreas, lung, heart, small intestine, colon, uterus, ovary, and fallopian tube endothelium), analyzed by immunohistochemistry. (B) Representative images showing the binding of D2AP11 anti-FSHR antibody to OC tissues of different pathological conditions (high- and low-grade serous carcinoma, mucinous adenocarcinoma, clear cell carcinoma, dysgerminoma, endodermal sinus carcinoma, and metastatic adenocarcinoma), analyzed by immunohistochemistry staining of OC TMA (US Biomax). The tissues (A and B) were viewed and imaged using Nikon NIS-Element Imaging system (×20, scale: 500 μm). Images were subjected to post-acquisition adjustments to optimize brightness, contrast, and image visibility. (C) Comparison of normalized RNA expression of FSHR and ERBB2/Her2 on 55 human tissue types, based on The Human Protein Atlas data.

D2AP11 binds human and murine FSHR.Figure 4

D2AP11 binds human and murine FSHR. (A) Flow cytometry plot of CAOV3, OVCAR3, and TOV21G cells stained with D2AP11 or no primary antibody followed by secondary APC-labeled antibody. (B) Flow cytometry plot of TOV21G cells, parental or after CRISPR-mediated deletion of FSHR, stained with D2AP11. (C) Flow cytometry plot of K562, K562-FSHR, and K562-LHCGR stained with D2AP11 or no primary antibody followed by secondary APC-labeled antibody. (D) Flow cytometry plot of A20 (GFP–)/A20-Fshr (GFP+) and ID8-Defb29/Vegf-a versus ID8-Defb29/Vegf-a-Fshr cells stained with D2AP11 (both cell lines were transfected with murine FSHR).

Anti-FSHR antibody for detection of FSHR+ tumor cells. Immunohistochemical detection of proteins from biological samples is a common way of determining protein expression from tumors or other specimens to better classify them for prognostic or therapeutic purposes. To further explore whether D2AP11 detects FSHR+ tumor cells in immunohistochemistry, we generated solid tumors in NSG immunodeficient mice. To generate tumors, 5 million K562, K562-FSHR, OVCAR3, or TOV21G cells were injected in 50% PBS/Matrigel (Corning) into the axillary flank of NSG mice. D2AP11 detected FHSR from frozen tumor sections (Figure 5A). We stained human FSHR–transduced 293T cells with D2AP11. D2AP11 was able to bind human FSHR similarly to polyclonal anti-human, but not to untransfected 293T cells, confirming this activity (Figure 5, B and C).

D2AP11 binds to FSHR in immunohistochemistry and immunocytochemistry and inFigure 5

D2AP11 binds to FSHR in immunohistochemistry and immunocytochemistry and induces ADCC. (A) Immunohistochemistry images from frozen sections of tumors derived from K562, K562-FSHR, OVCAR3, and TOV21G cell lines stained with D2AP11. Original magnification, ×40; scale bars: 50 μm. (B) Immunofluorescence images of 293T cells transfected with human FSHR and stained with either mouse anti–human FSHR or D2AP11 antibodies followed by secondary anti-mouse IgG. (C) Immunofluorescence images of untransfected 293T cells stained with D2AP11 antibodies followed by secondary anti-mouse IgG. Scale bars: 10 μm (B and C). (D) Absorbance values of isotype ELISA performed on D2AP11 antibody. (E) Cytotoxicity mediated by antibody-dependent cell-mediated cytotoxicity (ADCC) of D2AP11 or irrelevant mouse IgG2a (C1.18.4) against K562-FSHR. (F) Cytotoxicity mediated by ADCC of D2AP11 or irrelevant mouse IgG2a (C1.18.4) against K562. (G) Cytotoxicity mediated by ADCC of D2AP11 or irrelevant mouse IgG2a (C1.18.4) against OVCAR3 cells. Error bars represent mean ± SEM; all the experiments were done in triplicate. t test and ANOVA. ***P < 0.001.

Anti-FSHR antibody induces antibody-dependent cell-mediated cytotoxicity. To determine the isotype of D2AP11, we performed an ELISA and found D2AP11 to be IgG2a (Figure 5D), an isotype that can elicit antibody-dependent cell-mediated cytotoxicity (ADCC) (34). We first tested the ADCC capacity with K562 with or without FSHR. D2AP11 was able to increase the cytotoxic activity of PBMCs against K562-FSHR but not against K562 (Figure 5, E and F). To determine its ability to induce ADCC against unmodified FSHR+ OC cell lines, we cocultured OVCAR3 cells with PBMCs in the presence of D2AP11 or an irrelevant IgG2a antibody. We found that the physiological expression levels of FSHR in the OC cells were sufficient to be targeted by D2AP11-mediated cytotoxicity (Figure 5G) particularly with increasing doses of antibodies.

Generation, expression, and binding of FSHR-targeted bispecific T cell engager. Bispecific T cell engagers (TCEs) represent a recent significant development in the field of monoclonal technology. As D2AP11 anti-FSHR antibody exhibited initial levels of ADCC, we sought to improve on this potential. We designed an FSHR targeting TCE (D2AP11-TCE) (3537). We genetically optimized and fused the scFv of the FSHR mAb with the scFv of an optimized sequence we developed encoding anti-CD3 (modified from UCHT1) (Figure 6, A and B). D2AP11-TCE was efficiently expressed in vitro upon transfection of the DNA in Expi293F cells (Figure 6C). This bispecific showed no nonspecific binding to K562 cells, which do not have natural FSHR expression (Figure 6D), and retained binding to K562-FSHR cells (Figure 6E). Binding to FSHR was further confirmed in CAOV3 (Figure 6F) and OVCAR3-FSHR cells (Figure 6G). CD3 binding of D2AP11-TCE bispecific was confirmed using primary human T cells (Figure 6H).

Generation, expression, and antitumor activity of D2AP11-TCE.Figure 6

Generation, expression, and antitumor activity of D2AP11-TCE. (A) Cartoon of TCE engaging FSHR and the T cell receptor. VH, heavy chain variable region; VL, light chain variable region. (B) Schematic of DNA construct encoding D2AP11-TCE. GS, glycine-serine. (C) Western blot of in vitro expression of D2AP11-TCE or pVax1 empty vector after transfection in Expi293F cells. (D) The binding specificity of D2AP11-TCE was verified using K562 cells, which lack natural expression of FSHR. In K562 cells, no binding of D2AP11-TCE was observed. (E) Binding of D2AP11-TCE to FSHR-overexpressing K562 cells. (F) Binding of D2AP11-TCE to FSHR shown using the additional FSHR-expressing cell line CAOV3. The shift in the peak in D2AP11-TCE compared with pVax1 and secondary antibody alone indicates its binding to FSHR. (G) Binding of D2AP11-TCE to FSHR shown using OVCAR3 cells transduced with FSHR-encoding pBMN-I-GFP plasmid for overexpression of FSHR. There is a remarkable shift in the peak in FSHR-overexpressing OVCAR3 cells compared with empty vector and secondary antibody alone control. (H) Flow staining of primary human T cells with D2AP11-TCE and empty vector control; shift in peak denotes the binding of D2AP11-TCE to human T cells.

D2AP11-TCE induces potent killing in multiple ovarian tumor lines. To determine the ability of the FSHR-targeted bispecific TCE, D2AP11-TCE, to induce cytotoxicity through activation of T cells, in vitro cytotoxicity assay was performed based on impedance using an xCELLigence real-time cell analyzer (RTCA). The target cells (CAOV3, OVCAR3-FSHR, OVCAR4, OVISE, PEO-4, and Kuramochi-FSHR) were placed in the xCELLigence RTCA device and incubated for 18–24 hours, and subsequently human PBMCs and D2AP11-TCE were added. As a control we used HEK293T cells, an FSHR-negative cell line (38). Notably, we did not observe off-target killing against FSHR-negative HEK293T cells (Figure 7, A and B). As additional controls, we used 2 other non-FSHR-expressing cells, AGS gastric adenocarcinoma (Figure 7, C and D) and WM3743 human melanoma (Figure 7E) cells, and D2AP11-TCE did not induce off-target toxicities in these 2 cells either. Evaluation of different FSHR-expressing ovarian tumor cells demonstrated that D2AP11-TCE was highly efficient in the killing of CAOV3 (Figure 7, F and G), OVCAR3-FSHR (Figure 7, H and I, and Supplemental Figure 3A), OVCAR4 (Figure 7J and Supplemental Figure 3, B and C), OVISE (Figure 7, K and L), PEO-4 (Figure 7M), and Kuramochi-FSHR cells (Figure 7N). We observed dose-dependent killing in OVISE-FSHR and OVCAR3 cells in the presence of D2AP11-TCE and human PBMCs/T cells with EC50 value of 24.7 ng/mL and 15.9 ng/mL, respectively (Figure 7, O and P, and Supplemental Figure 4). OVCAR4, a high-grade serous ovarian adenocarcinoma cell line, is reported to have distinct positive expression of the surface receptor; FSHR (39) and D2AP11-TCE induced potent killing in this cell line in the presence of human PBMCs as well as human T cells (Figure 7J and Supplemental Figure 3, B and C). The studied ovarian tumor lines harbor different cancer driver mutations and exhibit resistance to multiple anticancer drugs (Supplemental Table 1). Notably, Kuramochi and PEO-4 bear BRCA2 mutations, and the latter also exhibits resistance to PARP inhibitors (27, 40). In our assays there is no escape from D2AP11-TCE killing in these 2 resistant ovarian tumor lines as well. As shown in Figure 7, G, I, and L, 3 days after addition of effector cells and treatment with D2AP11-TCE, no attached tumor cells were observed in the treated wells. However, in control HEK293T and AGS cells (Figure 7, B and D), all cells were found to be growing healthily 2–3 days after addition of effector cells and D2AP11-TCE. Additionally, an irrelevant TCE targeting human IL-13 receptor α2 (IL13Rα2-TCE) did not exhibit toxicity in OVCAR3-FSHR cells in the presence of human PBMCs, indicating the role of the D2AP11 arm in specific killing of FSHR-expressing tumors (Supplemental Figure 5). Notably, D2AP11-TCE was able to induce significant toxicity to FSHR-positive OVCAR4 cells when compared with D2AP11 antibody, at concentrations around 1,000-fold lower, indicating the enhanced killing efficacy of D2AP11 through its design as a bispecific engager, D2AP11-TCE (Supplemental Figure 6, A and B). EC50 values of D2AP11 and D2AP11-TCE were obtained at 30.3 μg/mL and 11.3 ng/mL, respectively, indicating approximately 1,000-fold higher potency of D2AP11-TCE compared with the anti-FSHR antibody D2AP11 (Supplemental Figure 6, C and D).

D2AP11-TCE induces specific killing of target OC cells.Figure 7

D2AP11-TCE induces specific killing of target OC cells. Assessment of the cytotoxic effect of D2AP11-TCE in FSHR-negative HEK293 cells (A and B), AGS gastric adenocarcinoma cells (C and D), and WM3743 human melanoma cells (E) as well as in the target human OC cell lines CAOV3 (F and G), OVCAR3-FSHR (H and I), OVCAR4 (J), OVISE (K and L), PEO-4 (M), and Kuramochi-FSHR (N) and dose-dependent killing of OVISE-FSHR (O) and OVCAR3 cells (P) in the presence of D2AP11-TCE and human PBMCs. In vitro cytotoxicity was measured based on impedance using xCELLigence real-time cell analyzer (RTCA) equipment (Agilent Technologies). The electrical conductivity was converted into the unitless cell index parameter by the xCELLigence device every 15 minutes, and images were captured at intervals of 1 hour. The data generated were normalized per the time point when the effector (E) cells (PBMCs) and D2AP11-TCE were added to the target (T) cells; E/T is 5:1 (A, B, F, G, M, and N) and 10:1 (CE, HL, O, and P). The data were analyzed using RTCA/RTCA Pro Software. No nonspecific killing was observed in HEK293T, AGS, and WM3743 cells, whereas potent killing was observed in CAOV3, OVCAR3-FSHR, OVCAR4, OVISE, PEO-4, and Kuramochi-FSHR target OC cells. Arrows indicate the time point at which D2AP11-TCE and effector cells were added to the target cells. Images shown display killing 2–3 days after the addition of effector cells and the TCE.

Cytokine/cytotoxic molecule secretion profile of FSHR-targeting TCE. Cytokines are involved in promoting the proliferation, survival, differentiation, and activation of lymphocytes (41). Different findings of bispecific TCEs as well as CARs suggest that cytokines secreted upon target cell ligation cause the lysis of antigen-negative tumor cells in close proximity to the antigen-specific engagement (42). We were interested in examining the cytokine/cytotoxic molecule secretion profile of D2AP11-TCE. Incubation of OVCAR3-FSHR target cells plus human PBMCs with D2AP11-TCE led to the significant induction of IFN-γ, soluble Fas (sFas), granzyme A, granzyme B, and perforin compared with empty vector control at 48 hours, and this construct drove robust tumor antigen-specific killing (Figure 8A). These effector molecules are known to possess potential to change the tumor microenvironment and to induce endogenous antitumor immunity (42).

Cytokine/cytotoxic molecule secretion profile and in vivo activity of D2AP1Figure 8

Cytokine/cytotoxic molecule secretion profile and in vivo activity of D2AP11-TCE. (A) Secretion profile of IFN-γ, sFas, granzymes A and B, and perforin in the presence of D2AP11-TCE upon coculturing of OVCAR3-FSHR and human PBMCs; E/T = 10:1. The supernatants analyzed were collected 48 hours after the addition of effector cells and TCE to target OVCAR3-FSHR cells. PBMCs from 3 different donors were used. Error bars represent mean ± SEM. t test. *P < 0.05, **P < 0.01. (B) Schematic of tumor study to evaluate the effect of D2AP11-TCE on tumor progression in K562/K562-FSHR–challenged NSG mouse model. (C) Average growth curve of K562 tumors grafted into NSG mice treated with D2AP11-TCE or empty vector (n = 5 mice per group). (D) Average growth curve of K562-FSHR tumors grafted into NSG mice treated with D2AP11-TCE or empty vector (n = 5 mice per group). (E) Schematic of tumor study to evaluate the effect of D2AP11-TCE on tumor progression in OVCAR3-FSHR–challenged NSG mouse model. (F) Average growth curve of OVCAR3-FSHR tumors grafted into NSG mice treated with D2AP11-TCE or empty vector (n = 10 mice per group). Two-way ANOVA. *P < 0.05.

FSHR-targeted TCE decreases tumor burden in vivo. To evaluate the in vivo antitumor effects of D2AP11-TCE, we used an in vivo challenge model we have developed. For this model we administered K562 cells or FSHR-overexpressing K562 cells (K562-FSHR) to NSG mice (Figure 8B). The mice were inoculated with DNA-encoded D2AP11-TCE (100 μg) and human T cells as described. Interestingly, there were tumor escapes in K562-challenged mice in both D2AP11-TCE–treated and empty vector control groups (Figure 8C), whereas there was a significant reduction in tumor volume in the D2AP11-TCE–treated group compared with empty vector control in the K562-FSHR–challenged mice (Figure 8D). After confirming the specificity and potency of D2AP11-TCE in this model, we further examined its effect in the OVCAR3-FSHR–challenged ovarian tumor mouse model. Fourteen days after tumor implantation, DNA-encoded D2AP11-TCE (100 μg) or empty vector (100 μg) was administered twice, 2 weeks apart, for in vivo antibody generation. On day 14, mice were inoculated with human T cells (10 million per mouse), and tumor volumes were measured periodically (Figure 8E). Treatment using this bispecific led to significantly decreased tumor burden in OVCAR3-FSHR tumor–bearing mice, while no similar impact was observed in the control group (Figure 8F), supporting the potential of this approach for therapeutic development against OC.

留言 (0)

沒有登入
gif