Targeted knockdown of Tim3 by short hairpin RNAs improves the function of anti-mesothelin CAR T cells

Leila Jafarzadeh Tehran University of Medical Sciences Elham Masoumi Ilam University of Medical Sciences Hamid Reza Mirzaei Tehran University of Medical Sciences Khadijeh Alishah University of Tehran Keyvan Fallah-Mehrjardi Tehran University of Medical Sciences Mohammad Khakpoor-Koosheh Tehran University of Medical Sciences Hossein Rostamian Tehran University of Medical Sciences Farshid Noorbakhsh Tehran University of Medical Sciences Jamshid Hadjati (  hajatij@tums.ac.ir ) Tehran University of Medical Sciences

Although CAR-expressing T cells have made great strides in the treatment of patients with advanced hematological malignancies, their e cacy in solid tumors has been limited in part due to the presence of immunosuppressive cells and molecules within tumor microenvironment (TME) (6)(7)(8). The latter includes immunosuppressive cytokines (e.g. TGF-β) and immune checkpoint molecules such as PD-1, CTLA-4 and Tim3. Among the inhibitory molecules, Tim3 is preferentially upregulated in exhausted tumor-in ltrating lymphocytes (13,14). Upon engagement with its cognate ligands, Tim3 plays a crucial role in tumorinduced immunosuppression likely through induction of intracellular calcium ux and T cell death (15).
Various studies have demonstrated that Tim-3 blockade reverses immunosuppression through reducing regulatory T cells and increased production of IFN-γ by T cells (16,17). In the current study, in order to diminish the negative effects of Tim signaling on antitumor function of CAR T cells, fully human secondgeneration anti-MSLN-CAR T cells were generated that co-expressed short hairpin RNA (shRNA) sequences against Tim-3 (Tim3.sh.MSLN-CAR T cell). Tim3 expression was evaluated in these cells, and the cells were analyzed for their proliferative response, cytokine production and antitumor cytotoxic effects.

Cell lines
HeLa (human cervical cancer cell line), HEK293T cell (human embryonic kidney T 293), Skov3 (human ovarian cancer cell line), Ovcar3 (human ovarian cancer cell line) and NALM-6 (pre-B cell leukemia cell line) cells were purchased from the Iranian Biological Resource Center (IBRC), Iran., Skov3, Ovcar3 cell lines were used as mesothelin-expressing cell lines and NALM-6 was used as a mesothelin negative control. Before the experiments, expression of mesothelin in these cells was analyzed by ow cytometry using PE-conjugated anti-human mesothelin antibody (R&D Systems, Minneapolis, MN, USA). HEK293T cells were used as the packaging cell line for the production of lentiviral particles. Mycoplasma contamination of all cell lines was routinely examined by polymerase chain reaction (PCR). KEK293T, and Skov3 cells were cultured in DMEM (Gibco Laboratories, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS) (Sigma Chemical Co., St. Louis, MO, USA) and 100 µg/ml penicillin-streptomycin (PAN Biotech, Aidenbach, Germany), and incubated at 37 °C in 5% CO2, 95% humidity atmosphere.

Genetic Modi cation Of T Cells
To generate a fully human anti-MSLN CAR construct, a Kozak consensus ribosome-binding sequence, a human CD8a signal peptide (SP) and a fully human anti-MSLN scFv were linked to the CD8a hinge, 4-1BB transmembrane region (TM) and intracellular domains containing the 4-1BB and CD3ζ domains.
Mesothelin-speci c scFv fragment originated from P4-scFv (18). Other fragments of CAR construct have been described previously (3). The CAR expression cassette was under the control of CMV promoter. To knockdown Tim3 expression, three shRNA-encoding sequences against three different regions of Tim3 gene were designed and inserted after the CD3ξ domain sequence in MSLN-CAR constructs (hereafter referred to Tim3.sh1.MSLN-CAR, Tim3.sh2.MSLN-CAR and Tim3.sh3.MSLN-CAR). The expression of shRNA-encoding sequences was under the control of a CMV promoter. ShRNA sequences are shown in Table 1. MSLN-CAR gene cassettes and different shRNA-coding sequences containing 5′-Flank-Sense-Loop-Antisense-3′Flank segments were cloned into a pCDH-CMV-MCS-EF1α-cGFP-2A-Puro lentivector backbone for production of third-generation lentiviral particles. In vitro Cytotoxicity assay

Flow Cytometric Analyses
Data acquisition was performed on a ow cytometer (BD FACSCalibur, BD Biosciences, San Jose, California). Analysis of all samples was done using FlowJo software (v7.6.1). All experiments were done in triplicate and repeated at least three times.

Statistical analysis
Analysis of variance (ANOVA) followed by Tukey's post-hoc was used to reveal any signi cant differences among treatment groups. P values below 0.05 were considered signi cant. Statistical analyses were performed with Prism 8 software (GraphPad Software, Inc., San Diego, USA).

E cient generation of MSLN-CAR T cells and Tim3-targeted MSLN-CAR T cells using lentiviral gene transfer
To examine the effect of Tim3 targeting on anti-MSLN-CAR expression, primary T cells were exposed to third-generation lentiviral vectors at an MOI of ~ 7 to generate four types of anti-MSLN CAR T cells; a fully human second-generation anti-MSLN CAR T cell as well as three groups of CAR T cells encoding three different anti-Tim3 shRNAs [referred to as Tim3.sh1, sh2, and sh3.MSLN-CAR T cells (Fig. 1a)]. To evaluate the effect of single and combinatorial targeting of Tim3 on CAR expression, primary T cells were transduced with one, two or three lentiviral vectors to generate different MSLN-CAR T cells (Fig. 1a). Untransduced T cells (Un-T) and Mock T cells were used as control. Flow cytometric analysis revealed that approximately 50 percent of cells in each type of CAR T cells were CAR positive (Fig. 1b).

E cient Targeting Of Tim3 In Msln-car T Cells
To explore the e ciency of Tim3 targeting by shRNAs, cell surface expression of Tim3 was evaluated in 5 different donors. Flow cytometric analysis revealed that Tim3-targeted shRNAs alone (i.e. sh1, sh2 and  (Fig. 2c). Moreover, the expression levels of Tim-3 in Un-T cells and Mock T cells before and after activation with anti-CD3/CD28 antibodies were measured by ow cytometry (Fig. 2a). The results showed that anti-CD3/CD28 activation signi cantly augmented Tim3 expression in all donors. Altogether, our data revealed that expression levels of Tim3 were signi cantly To investigate the effects of Tim3 knockdown on the cytotoxic function of MSLN-CAR T cells, a PKH26/7AAD-based cytotoxic assay was used. All cytotoxic assays were performed at an effector to target ratios of 1:1, 5:1, 10:1, and 20:1 (Figs. 3 and 3b). Unlike Mock T cells and Un-T cells, MSLN-CAR T cells and two Tim3-targeted MSLN-CAR T cells showed signi cant cytotoxic activity against target cells (Fig. 3c). Our data showed the highest level of cytotoxicity in sh2.sh3.MSLN-CAR T cells at target to effector ratio of 20:1. MSLN-CAR T cells and Tim3-targeted MSLN-CAR T cells were not signi cantly different in their cytotoxic function at effector to target ratios of 1:1, 5:1 and 10:1 (Fig. 3c). Finally, our results revealed that Tim3-knockdown had no negative impact on the cytotoxic function of Tim3-targeted MSLN-CAR T cells. We did not observe any signi cant cytotoxicity against MSLN-negative NALM-6 cells (data not shown).

Tim3 Knockdown Enhances Antigen-dependent Proliferation Of Msln-car T Cells
To explore the effects of Tim3 knockdown on the proliferative response of MSLN-CAR T cells, MSLNdependent proliferation of these cells was evaluated following co-culture with target cells. To do this experiment, PKH26-labelled T cell/MSLN-CAR T cells were co-incubated with mitomycin C-treated target cells for 72hr at a 1:1 ratio. Our data revealed that both MSLN-CAR T cells and Tim3-targeted MSLN-CAR T cells had higher proliferative capacity compared with Mock T cells and Un-T cells (Fig. 4a). Consistent with the results of cytotoxicity assays, Tim3.sh2.sh3.MSLN-CART cells showed signi cantly higher proliferative responses compared with Tim3.sh1.MSLN-CAR T cells and MSLN-CAR T cells (Fig. 4b).
However, none of the T cell groups showed any signi cant proliferative response when they were cultured in the absence of target cells (Fig. 4b) or in the presence of MSLN-negative NALM-6 cells (data not shown).
Tim3 Knockdown Improves Antigen-dependent-cytokine Production Of Msln-car T Cells We next analyzed cytokine production by different groups of MSLN-CAR T cells. To do this, supernatants of Mock T cells, MSLN-CAR T cells and Tim3-targeted MSLN-CAR T cells in the presence and absence of target cells (1:1 ratio) were harvested. Concentrations of cytokines TNF-α, IL-2 and IFN-γ were measured by ELISA. Our results showed that culturing of T and/or CAR T cells in the absence of target cells led to the minimal cytokine production (Fig. 5a-c). However, co-culturing all types of MSLN-CAR T cells with target cells did signi cantly induce TNF-α, IL-2 and IFN-γ production (Fig. 5a-c). Nonetheless, minimal levels of TNF-α, IL-2 and IFN-γ were detected in cultures of Mock T cells in the presence or absence of target cells. TNF-α, IL-2 and IFN-γ levels were signi cantly higher in Tim3.sh2.sh3.MSLN-CAR T cells compared with Mock T cells, MSLN-CAR T cells and Tim3.sh1.MSLN-CAR T cells (Fig. 5a-c).

Discussion
Negative immune regulator Tim3 is highly expressed on exhausted or impaired T cells in chronic viral infections and tumor-bearing hosts (14,(19)(20)(21). Consistent with its role as an inhibitory molecule, it has been shown that the blockade of Tim3-Tim3L signaling with blocking antibodies or by genetic targeting not only improves the secretion of IFN-γ by activated T cells but can also restore the function of exhausted T cells. This underscores the importance of targeting Tim3 signaling for improving immunity against chronic viral infections and cancers (22)(23)(24). So far, several clinical trials using Tim3 blocking monoclonal antibodies have been started aiming to evaluate the therapeutic e cacy, safety, tolerability and dose-limiting toxicities of Tim3 blockers (ClinicalTrials.gov, NCT03652077and NCT03489343). In the current study, the effects of Tim3 gene knockdown on MSLN-speci c CAR T cells were investigated. Compared with co-administration of CAR T cells with anti-Tim3 monoclonal antibodies, genetic targeting of Tim3 in CAR T cells may not only overcome the need for repeated administration of monoclonal antibodies but also minimizes the risk of systemic toxicity (25). Moreover, stable expression of a shRNA sequence leads to continuous repression of Tim3 expression after in ltration of CAR T cells into the tumor microenvironment, while the concentrations of a systemically-administered anti-Tim3 antibodies might not reach su cient levels inside tumors. It has also been shown that the extracellular domain of Tim3 contains multiple ligand-binding sites. Anti-Tim3 monoclonal antibodies might not completely block all of these binding sites, making genetic targeting of Tim3 a superior approach for hindering its activity (26). In the current study, Tim3-targeting in MSLN-CAR T cells abrogated the inhibitory function of Tim3. ShRNA-mediated stable knockdown of Tim3 could signi cantly augment proliferation, cytokine production and cytotoxic functions of MSLN-CAR T cells.
Various studies have shown that Tim3 signaling confers its inhibitory effects on T cells through mechanisms including eliciting intracellular calcium ux, inducing apoptosis and reducing IFN-γ production (13,15,27). Although the downstream signaling events remain poorly characterized, different studies have shown that ligand-dependent Tim3 signaling leads to repression of TCR signaling and thereby suppression of T cell proliferation and survival (28)(29)(30).
It is not well known whether ex vivo manipulation of T cells for cancer immunotherapy has any negative effects on their function through up-regulation of immunoinhibitory receptors such as Tim3. Our study revealed that the activation of T cells with anti CD3/CD28 antibodies (as the rst stimulation) can signi cantly increase the expression levels of Tim3. Primary activation of T cells with anti-CD3/CD28 antibodies is essential for e cient transduction of CAR transgenes (31). More importantly, Tim3 overexpression was also detected in T cells after transduction with CAR-carrying lentiviral vectors (i.e. MSLN-CAR T cells) and also after their co-culture with MSLN-positive target cells. Moreover, these ndings are in agreement with previous reports indicating that Tim3 was up-regulated on T and CAR T cells after co-incubation with target cells or administration to tumor-bearing mice (32)(33)(34)(35). Considering the inhibitory effects of Tim3, it is clear that overexpression of Tim3, even prior to their exposure to tumor cells, might have a negative impact on (CAR) T-intrinsic qualities such as longevity and functionality which play key roles in determining the e cacy of immunotherapy.
In the present study, we also showed signi cant improvement in MSLN-CAR T cells cytotoxic function following shRNA-mediated blockade of Tim3. This nding is consistent with Kenderian et al study which has demonstrated that combination therapy with suboptimal doses of anti-CD123 CAR T cells plus anti-Tim3 antibody could signi cantly improve complete response (CR) rate (CR = 100%) in a relapsed acute myeloid leukemia (AML) xenograft model compared with suboptimal doses of anti-CD123 CAR T cells alone (CR = 45%) or suboptimal dose of anti-CD123 CAR T cells plus anti-PD-1 antibody (CR = 80%) (35).
In line with the above study, Koyoma et al. have shown that Tim3 is up-regulated in patients who were treated with anti-PD-1. Their data also shows that anti-Tim3 therapy could confer survival advantage following treatment failure with anti-PD-1 (36). Altogether, these data indicate that therapeutic regimens with Tim3-knockdown MSLN-CAR T cells might either prevent tumor relapse or have superior antitumor activity in relapsed MSLN-positive tumors. Moreover, various studies have reported that concomitant blockade of Tim-3 and PD-1 could increase cytotoxic activity of T cells (37,38). In this regard, it seems that dual or multiple targeting of inhibitory checkpoint receptors (either by gene-editing techniques or shRNAs) might further improve the therapeutic outcome T cell-based cancer immunotherapies.
To analyze the effects of Tim3 downregulation on cell proliferation, we investigated proliferation rates of Tim3-knocked down MSLN-CAR T cells. Our data revealed that genetic targeting of Tim3 could signi cantly improve the proliferative response of Tim3.sh2.sh3.MSLN-CAR T cells in an antigendependent manner. This nding is in accordance with previous reports where it has been shown that, unlike Tim3 negative T cells, Tim3 positive T cells had a lower cytokine production and proliferation rate which could be reversed by Tim3 targeting (14,16,(39)(40)(41).
In this study, cytokine analysis revealed that stable knockdown of Tim3 could signi cantly improve production of cytokines TNF-α, IFN-γ and IL-2 in an antigen-dependent manner. The pattern of produced cytokines is an indicative of T cell polarization towards a favorable antitumor phenotype and highlights the key role of these cytokines in T cell cytotoxic function and proliferation (42)(43)(44).
Another facet of Tim3 signaling which is important in cancer immunotherapy is the role of Tim3 signaling in the differentiation and regulation of Tregs, the cells that are thought to be an obstacle in antitumor immunity. Tim3 is thought to exert these effects through in uencing the expression of CD80/CD86 or TGF-β and expression of CTLA4 and TIGIT inhibitory molecules (17,45). We did not analyze Treg markers in this study, but it is likely that improved function of Tim3-targeted MSLN-CAR T cells might have been related to repressed Treg-promoting signals.
Considering different shRNA sequences and their effectiveness, current study could not detect any cellular toxicity (in terms of potency and relative e cacy) in shRNA-targeted MSLN-CAR T cells. Hence, it can be speculated that our strategy for designing and screening of shRNA sequences, before and during knockdown experiments, could empirically identify a combinatorial approach (i.e. co-transduction of two shRNA2 and shRNA3) as the most effective strategy for targeting Tim3 in the MSLN-CAR T cells.

Conclusions
The results of the current study illustrate that genetic targeting of Tim3 could improve the function of MSLN-CAR T cells. Overexpression of Tim3 in T cells during different phases of CAR T cell production (before and after CAR transduction) highlights the importance of revisiting the existing CAR T cell production protocols, perhaps by development of novel protocols which do not require initial activation of cells. Dual or multiple targeting of inhibitory checkpoint receptors either pharmacologically or genetically might further improve the therapeutic outcome of CAR T cell-based cancer immunotherapies. Altogether, genetic targeting of Tim3 might be worth considering in clinical studies of CAR T cells.      Tim3 knockdown improves antigen-dependent-cytokine production of MSLN-CAR T cells. Mock, MSLN-CAR, Tim-3.sh1 and sh2/sh3.MSLN-CAR T as well as Un-T cells were co-incubated with Hela cells in a 1:1 ratio or incubated in media. The supernatants were harvested after 24 hr. and TNF-α (a), IL-2 (b) and IFN-γ (c) concentrations were measured by ELISA. The data are shown as mean ± SD. Mean comparisons were performed using one-way ANOVA followed by Tukeyʼs post hoc test. P < 0.05 was considered statistically signi cant. The results are representative of three independent experiments, each performed in duplicate.  Tim3 knockdown improves antigen-dependent-cytokine production of MSLN-CAR T cells. Mock, MSLN-CAR, Tim-3.sh1 and sh2/sh3.MSLN-CAR T as well as Un-T cells were co-incubated with Hela cells in a 1:1 ratio or incubated in media. The supernatants were harvested after 24 hr. and TNF-α (a), IL-2 (b) and IFN-γ (c) concentrations were measured by ELISA. The data are shown as mean ± SD. Mean comparisons were performed using one-way ANOVA followed by Tukeyʼs post hoc test. P < 0.05 was considered statistically signi cant. The results are representative of three independent experiments, each performed in duplicate.