Bispecific antibodies enhance tumor‐infiltrating T cell cytotoxicity against autologous HER‐2‐expressing high‐grade ovarian tumors

Abstract Epithelial ovarian cancer displays the highest mortality of all gynecological tumors. A relapse of the disease even after successful surgical treatment is a significant problem. Resistance against the current platinum‐based chemotherapeutic standard regime requires a detailed ex vivo immune profiling of tumor‐infiltrating cells and the development of new therapeutic strategies. In this study, we phenotypically and functionally characterize tumor cells and autologous tumor‐derived αβ and γδ T lymphocyte subsets. Tumor‐infiltrating (TIL) and tumor‐ascites lymphocytes (TAL) were ex vivo isolated out of tumor tissue and ascites, respectively, from high‐grade ovarian carcinoma patients (FIGO‐stage IIIa‐IV). We observed an increased γδ T cell percentage in ascites compared to tumor‐tissue and blood of these patients, whereas CD8+ αβ T cells were increased within TAL and TIL. The number of Vδ1 and non‐Vδ1/Vδ2‐expressing γδ T cells was increased in the ascites and in the tumor tissue compared to the blood of the same donors. Commonly in PBL, the Vγ9 chain of the γδ T cell receptor is usually associated exclusively with the Vδ2 chain. Interestingly, we detected Vδ1 and non‐Vδ1/Vδ2 T cells co‐expressing Vγ9, which is so far not described for TAL and TIL. Importantly, our data demonstrated an expression of human epidermal growth factor receptor (HER)‐2 on high‐grade ovarian tumors, which can serve as an efficient tumor antigen to target CD3 TIL or selectively Vγ9‐expressing γδ T cells by bispecific antibodies (bsAbs) to ovarian cancer cells. Our bsAbs efficiently enhance cytotoxicity of TIL and TAL against autologous HER‐2‐expressing ovarian cells.


INTRODUCTION
Epithelial ovarian cancer causes around 140,000 deaths annually in women worldwide. With a 5-yr-survival rate of only <40% after first-line therapy ovarian cancer remains a highly lethal tumor entity to this day. 1 Due to the absence of initial symptoms the disease is often diagnosed at an advanced stage accompanied with an accumulation of ascites fluid in the peritoneal cavity, which facilitates the dissemination of tumor cells in the surrounding microenvironment. 2 The standard therapy is a surgical resection followed by platinum-based therapy.
A significant problem for an effective treatment is the recurrence of the disease even after successful surgical treatment-caused in many cases by a resistance against the current platinum-based chemotherapeutic standard regimen. 3 Many attempts to improve ovarian cancer patients' outcome by adding additional drugs failed.
Trastuzumab treatment in ovarian cancer is not established partially due to the high tumor heterogeneity, which leads to lower response rates. 4 Combining chemotherapy with the humanized human epidermal growth factor receptor (HER)-2 mAb trastuzumab has been described in a few clinical trials to improve overall patients' survival compared to chemotherapy alone in HER-2-expressing uterine serous and high-grad endometrioid tumors. [4][5][6] Commonly, HER-2 expression is detected in 22-66% of ovarian cancer patients. 7 New therapeutic strategies such as the development of bispecific antibodies (bsAbs) with the aim to achieve more effective and durable response, are of major clinical interest. [8][9][10] The bsAbs enhance cytotoxic activity by selectively targeting immune cells to tumor-associated antigens such as HER-2. [11][12][13] Recently, we demonstrated an enhanced lysis of ex vivo isolated ovarian tumor cells as well as of cisplatin-resistant SK-OV-3 cells by activating T cells of peripheral blood lymphocytes (PBL) or tumor tissue via bsAbs. 14 Human T lymphocytes are attractive effector cells for T cell-based immunotherapy due to their strong cytotoxic activity, antigen recognition in HLA-independent manner, their reduced induction of a severe graft-versus-host disease, the capacity of V 2 T cell subset to present antigens to T cells, as well as their production of Th1 cytokines, which could compensate for the reduced number of Th1-producing CD4 T cells at the tumor site. 15,16 T cells can be divided into two main subsets, the T cell receptor (TCR) V 2 T cells, which predominate the blood and the TCR V 1 T cells representing the main subset in the tissue. Both subsets infiltrate in solid tumors and expand to an 800-to 1000-fold increase after their activation. 13 cells. 20,21 In our study, we phenotypically characterized the TCR -repertoire of T cell subsets by immunophenotyping with in-house anti-TCRV mAb in ex vivo isolated tumor tissue compared to autologous blood and ascites of ovarian cancer. Importantly, the cytotoxic capacity of different -and T cell subsets isolated out of PBL, tumor-ascites lymphocytes (TAL), and tumor-infiltrating lymphocytes (TIL) from the same donor against autologous tumor cells were analyzed. Recent developed bsAbs were analyzed whether they target tumor-derived T cell subsets to autologous ovarian tumor cells and enhance the cytotoxicity of these T cell subsets against ovarian cancer cells.

Patient cohort
Heparinized blood, ascites, and tumor tissue from patients were obtained from the Department of Gynecology   To determine cytokine expression in different  T cell subsets   within PBL, TAL, or TIL, surface stainings of T cells with anti-CD3   clone SK7, anti-pan TCR clone 11F2, anti-CD4 clone OKT4, and   anti-CD8 clone SK1 mAbs were combined with intracellular IFN-,   IL-4, IL-9, IL-10, IL- All samples were analyzed on a LSR-Fortessa flow cytometer (BD Biosciences) using CellQuestPro, Diva or FlowJo software.

Characterization of tumor cells by in situ hybridization (FISH technique) and treatment with cisplatin
Tumor-specific chromosome diversifications of primary tumor cells were analyzed by using the tricolor probe TERC (3q26)/MYC (8q24)/SE7 TC (#KBI-10704, Kreatech/Leica, Wetzlar, Germany) for fluorescence in situ hybridization (FISH) as described. 25   were added every 2 d over a culture period of 14 d. After 2 wk, most T cell lines had a purity between 60% and 99% T cells.
T cells with a purity <98% were labeled with anti-TCR mAb clone IP26 (BioLegend) and subjected to magnetic separation to deplete remaining T cells. T cells were labeled with either anti-CD4 mAb clone OKT4 (BioLegend) or anti-CD8 mAb clone SK1 (BioLegend) and subjected to magnetic separation to deplete CD4 or CD8 T cells.
To analyze the functional capacity of the different V 9-expressing T cell subsets, zoledronic acid expanded V 9-positive TILs were stained with anti-pan TCR clone 11F2 (BD Biosciences), anti-V 9 clone 7A5, 22 anti-V 2 clone 123R3 (Miltenyi Biotec), and anti-V 1 clone REA173 (Miltenyi Biotec) and sorted with a FACSAria (BD Biosciences). a heavy chain derivative and tagged with a C-terminal hexa-histidine motif were purified from supernatant by two successive steps of affinity chromatography using CaptureSelect Fab kappa affinity matrix (BAC B.V., Naarden, Netherlands) and nickel-nitrilotriacetic acid (Ni-NTA) agarose beads (Qiagen) as described earlier. [12][13][14] Purity and integrity of bsAbs were verified by capillary electrophoresis using an Experion automated electrophoresis system (Bio-Rad, Kabelsketal, Germany) and size exclusion chromatography.

Real time cell analyzer (RTCA)
Cytotoxicity against adherent autologous cancer cells was analyzed by an RTCA (X-Celligence, ACEA Biosciences, Inc., San Diego, CA, USA) in triplicates as described elsewhere. 13

STATISTICAL ANALYSIS
The statistical analysis was assessed by using Graph Pad Prism  In this study, we demonstrated a slight increase of the CD3

Immune profiling of tumor-derived T cells in comparison to autologous peripheral blood T cells of ovarian cancer patients
T cell percentage within TIL compared to the PBL of the same donors.
Interestingly, the T cell percentage within TAL was more drastically increased in comparison to T cells of the blood or tumor-tissue of these donors (Fig. 1A). Regarding the mainly described T cell subsets, we observed an increase of V 1 T cells within the TAL and TIL compared to the PBL, which is comparable to the increase of CD8

F I G U R E 1 Analysis of T lymphocyte subsets within blood, ascites, and tumor tissue of ovarian cancer patients. (A, B)
To determine the relative percentage of the different T cell subsets cells within peripheral blood (PBL), tumor-ascites lymphocytes (TAL) and tumor-infiltrating lymphocytes (TIL), ex vivo isolated cells of ovarian cancer patients (n = 31) were stained with anti-CD45, anti-CD3, anti-pan TCR, anti-V 1, anti-V 2, anti-V 9, anti-V 2,3,4, anti-V 5, anti-CD4, and anti-CD8 mAbs as indicated and analyzed by LSR-Fortessa. A gate was set on lymphocytes based on the side scatter properties and CD45-positive leukocytes to analyze T cells within leukocytes. To distinguish between CD4 and CD8 T cells a gate was set on CD3-positive cells excluding CD3/pan TCR-positive cells. For discrimination between V 1 and V 2 and the different V -expressing subsets within the CD3/pan TCR-positive cells, a gate was set on CD3/pan TCR-expressing T cells. The gating strategy is shown with PBL of one patient. (A) Each symbol presents the data of one donor, and the lines in the boxes represent the median value of different independent experiments. Normal distribution assumption of all T cell subsets was analyzed by Shapiro-Wilk normality test. Based on normal distribution assumption of nonmatched V 2, CD4, and CD8 T cells, statistical comparison was done parametrically by using 1-way ANOVA followed by Tukey's multiple comparison test. Nonparametric, nonmatched datasets of V 1 and T cells were analyzed by Kruskal-Wallis test followed by Dunn's multiple comparison test. Significances are shown as P value; * = P < 0.05, ** = P < 0.01, *** = P < 0.001, and **** = P < 0.0001 (continued on the next page) T cells of the same donors (Fig. 1A). The distribution of CD8-and

Characterization of tumor cells and their cisplatin sensitivity
Immune profiling of tumor-infiltrating cytotoxic T cells revealed the expression of co-inhibitory receptors and an exhausted phenotype as described by others. 27,28 Here, we were interested in the cytotoxic capacity of ex vivo isolated TIL against autologous ovarian cancer cells and the modulation of cytotoxicity by bsAb. Therefore, we isolated tumor cells and

Enhancement of cytotoxicity of peripheral blood and tumor-derived T cells against autologous ovarian tumor cells
A very promising strategy to overcome certain limitations of mAb and occasionally described HER-2 resistance mechanisms or chemotherapy resistance can be the use of bsAb targeting all CD3 T cells to tumor cells.

F I G U R E 2 Characterization of tumor cells from tumor tissue and ascites at d0 and d13-20 after culture. (A) A representative gating strategy
for flow cytometric analysis of pan-cytokeratin + ( -pan CK) and CD45-negative ( -CD45) tumor cells out of tumor tissue and ascites from ovarian cancer patients (dot plots) is shown. Histograms are showing HER-2 ( -HER-2) or EpCAM ( -EpCAM) expression (dark gray) compared to isotypecontrol (bright gray) at day 0 (d0, immediately after ex vivo isolation) or after 13 d of culture (d13). (B) Flow cytometric analysis of HER-2 + (upper two panels) and EpCAM + (lower two panels) tumor cells from ovarian cancer cells directly after ex vivo isolation (d0) and after 13-20 d of culture are presented. In the second panel HER-2-and EpCAM expression are shown on cultured tumor cells derived from tumor tissue (tumor) compared to ascites derived tumor cells (ascites). The median fluorescence intensity was calculated as an x-fold increase in relation to the staining with isotype control. As the majority of the samples did not follow a normal distribution (Shapiro-Wilk normality test), Mann-Whitney test was applied for nonparametric, nonmatched datasets. Significances are presented as P value *** = P < 0.001. Because these values do not follow a standard deviation the quartiles for each dot plot have been added.  Fig. 4A (a)), which could be further enhanced by the addition of bsscFv [HER2xCD3] (Fig. 4A (b)). Similar results were obtained by using CD8 T cells or V 2 T cells of donor OC1 against KI-OC-1 tumor cells ( Fig. 4A (c)). CD4 T cells isolated out of PBL or TAL did only slightly induce cytotoxic activity against autologous tumor cells (data not shown, Fig. 4A (d)). In contrast to cytotoxic T cells (CD8, V 1, and V 2) isolated out of PBL of cancer patients OC12 and OC1, T cells isolated out of the ascites (TAL) from the same donors showed an reduced capacity to lyse autologous tumor cells (Fig. 4A (d), (f)), which can be drastically enhanced in the presence of bsscFv [HER2xCD3] (Fig. 4A (e) and (f)). Interestingly, the capacity of cytotoxic T cells isolated from the tumor tissue (TIL) against autologous tumor cells was impressive compared to the TAL subsets ( Fig. 4A (g) and (i)); however, TIL cytotoxicity can, similar to TAL cytotoxicity, be further enhanced by bsscFv [HER2xCD3] (Fig. 4A (h) and (i)). BsscFv [HER2xCD3] alone (Fig. 4A (c)) and control constructs such as bsAb [HER2xCD89] did not trigger target cell lysis (Fig. 4A (g)).
In additional experiments with further pairs of T cell subsets and autologous tumor cells isolated out of tumor tissue (n = 8), we confirmed our results (Fig. 4B). We observed a specific lysis of tumor cells by autologous V 1 and V 2 T cells that is similar or slightly enhanced compared to cytotoxic CD8 T cells after 4 or 10 hr of coculture. As expected, the cytotoxic activity of CD4 T cells was less pronounced. Importantly, bsscFv [HER2xCD3] efficiently enhances cytotoxicity of TIL subsets against HER-2-expressing primary ovarian cancer cells (Fig. 4B).
In addition, our results revealed that HER-2 expressed on highgrade ovarian tumors can be an efficient tumor antigen for bsAb tar-

Autologous TIL showed a higher cytotoxic activity capacity compared to PBL
Recently, others reported that an inflammatory tumor microenvironment contributes to exhaustion and the expression of co-inhibitory receptors on cytotoxic T cells. 27,28 We, therefore, investigated whether the application of bsAb can overcome a suggested exhaustion of cytotoxic T cells within whole TIL population (which can also contain suppressive immune cells). We applied tumor cells initially generated out of the tumor from patient OC11, which revealed an intermediate cisplatin response (Fig. 3) and a high expression of HER-2. 14

Impaired cytotoxicity of whole TIL population can be recovered by selective activation of T cells by bsAb
Considering that the polyclonal activation of all T cell subsets by bsscFv [HER2xCD3] could have fatal consequences for instance due to a cytokine storm, we investigated whether it could be beneficial

DISCUSSION
The high recurrence rate in ovarian cancer disease-even after suc- In our study, we expand the phenotypic characterization of the T cell subsets within PBL, TAL, and TIL to the different T cell subsets.
We revealed an inversion of the V 1/V 2 ratio in TAL and TIL compared to PBL of the same patients and an enhancement of V