Improved anti-leukemia activities of adoptively transferred T cells expressing bispecific T-cell engager in mice

Despite the impressive clinical efficacy of T cells engineered to express chimeric antigen receptors (CAR-Ts), the current applications of CAR-T cell therapy are limited by major treatment-related toxicity. Thus, safer yet effective alternative approaches must be developed. In this study, we compared CD19 bispecific T-cell engager (BiTE)-transferred T cells that had been transfected by RNA electroporation with CD19 CAR RNA-transferred T cells both in vitro and in an aggressive Nalm6 leukemia mouse model. BiTEs were secreted from the transferred T cells and enabled both the transferred and bystander T cells to specifically recognize CD19+ cell lines, with increased tumor killing ability, prolonged functional persistence, increased cytokine production and potent proliferation compared with the CAR-T cells. More interestingly, in comparison with CD3/CD28 bead-stimulated T cells, T cells that were expanded by a rapid T-cell expansion protocol (REP) showed enhanced anti-tumor activities for both CAR and BiTE RNA-electroporated T cells both in vitro and in a Nalm6 mouse model (P<0.01). Furthermore, the REP T cells with BiTE RNAs showed greater efficacy in the Nalm6 leukemia model compared with REP T cells with CAR RNA (P<0.05) and resulted in complete leukemia remission.


INTRODUCTION
Adoptive cell transfer (ACT) using chimeric antigen receptor (CAR-T) cells has demonstrated an unprecedented anti-leukemic response leading to the sustained remission in recent clinical trials. [1][2][3] However, the current ACT approach using lentivirally or retrovirally transduced CAR-T cells has limitations associated with the lack of control over their activation and expansion in vivo, 4 which has resulted in acute cases of tumor lysis syndrome and fatal cytokine release syndrome, as well as complications caused by the persistent on-target activity of CAR-Ts, such as long-term B-cell aplasia. RNA CAR-Ts, which are generated by electroporation of a messenger RNA encoding the CAR into the T cells, are being tested in clinical trials for the treatment of metastatic mesothelioma and pancreatic carcinoma. 5 Potent anti-leukemia activities have also been demonstrated in preclinical mouse models using RNA CAR-Ts. 6,7 Because of the self-decay of the anti-tumor activity of RNA CAR-Ts, multiple infusions of a controlled dose of RNA CAR-Ts may limit tumor lysis syndrome and cytokine release syndrome. Moreover, the use of RNA CAR T cells may avoid causing long-term B-cell aplasia. However, the transient transgene expression in the RNA CAR-Ts leads to the suboptimal treatment efficacy compared with lentivirally transduced CAR-Ts. 6,8,9 Efforts to further improve the anti-tumor activities of the RNA CAR-Ts will break this bottleneck and advance this RNA-based ACT platform, rendering it as potent as viral vector-based therapy, but potentially more flexible and safer.
Bispecific T-cell engagers (BiTEs), which are constructed by fusing an anti-CD3 scFv to an anti-tumor antigen scFv, have been tested in clinical trials for cancer treatment. [10][11][12] However, the development of BiTEs to treat cancer patients still faces enormous challenges. BiTEs need to penetrate deeply into the tumor tissues, where the tumor resident T cells may have already been tolerized or anergized in the tumor microenviroment. In addition, the therapeutic potential of exogenously administered BiTEs may be limited by their short half-lives and their dissociation from triggering receptors within a relatively short period of time. Thus, integrating BiTEs with adoptive T-cell transfer in situ may enhance the cancer treatment efficiency of both BiTEs and adoptive T-cell transfer. 13,14 In this study, we tested the anti-leukemia activities of CD19 BiTE (blinatumomab) RNA-electroporated T cells that were generated through CD3/CD28 Dynal Bead stimulation or a rapid T-cell expansion protocol (REP) and found that the REP T cells transferred with a CD19 BiTE nearly completely eradicated the leukemia cells in the mice and resulted in sustained survival. Therefore, a combination of T cells generated by REP and the RNA electroporation of a CD19 BiTE has the potential to cure CD19 + malignancies with controlled toxicities and without B-cell aplasia.

Cell lines and primary human T-lymphocyte cultures
The Nalm6 (DSMZ, Braunschweig, Germany), Raji (American Type Culture Collection, Manassas, VA, USA) and K562 (American Type Culture Collection) cell lines were cultured per the providers' instructions. The CD19-expressing K562 cells and click beetle green (CBG)-expressing Nalm6 cells were generated as previously described. 7 Primary lymphocytes from normal donors were provided by the University of Pennsylvania Human Immunology Core. The primary T lymphocytes were stimulated and expanded using two different methods. (1) CD3/CD28 Dynabeads (Life Technologies, Grand Island, NY, USA) were used as previously described. 6 (2) The REP approach was performed as previously described. 15 In brief, 1 × 10 6 purified CD4 and CD8 T cells in a 1:1 ratio were added to 1 × 10 8 irradiated allogeneic peripheral blood mononuclear cells in a T150 flask in a total volume of 150 ml of R/10 medium in the presence of 50 ng ml − 1 OKT3. Interleukin-2 (IL-2) was added to the culture for a final concentration of 300 IU ml − 1 at day 2. At day 5, 120 ml of the culture supernatant was replaced with fresh R/10 medium containing 300 IU ml − 1 of IL-2. The T cells were split every other day beginning 7 days after stimulation until day 11. The expanded T cells were aliquoted and frozen for further use.
Construction of the in vitro transcribed (IVT) RNA vectors and RNA in vitro transcription and electroporation The in vitro transcription vectors for the CD19-BBZ and CD19-28Z CARs were constructed as previously described. 7 The DNA encoding the blinatumomab BiTE was synthesized based on the published sequence data from patent US7575923 and subcloned into a pGEM.64A-based in vitro transcription vector. 16 The in vitro transcription vector was linearized by digestion with the proper restriction enzyme, and the mMESSAGE mMACHINE T7 Ultra kit (Life Technologies) was used to generate the IVT RNA, according to the procedure provided with the kit. The frozen stimulated T cells were thawed and cultured in R/10 medium overnight before electroporation. Before electroporation, the T cells were washed three times with OPTI-MEM (Life Technologies) and resuspended in OPTI-MEM (Life Technologies) at a final concentration of 1-3 × 10 8 cells per ml before electroporation. Subsequently, 0.1 ml of the T cells was mixed with the indicated IVT RNA and electroporated in a 2-mm cuvette (Harvard Apparatus BTX, Holliston, MA, USA) using an ECM830 Electro Square Wave Porator (Harvard Apparatus BTX). 8

Enzyme-linked immunosorbent assay
The T cells or target cells were washed and suspended in R/10 medium at 1 × 10 6 cells per ml. Approximately 0.1 ml of each cell line was added to a well of a 96-well plate (Corning) and incubated at 37°C for 18-20 h. The supernatant was collected and subjected to an enzyme-linked immunosorbent assay.

CD107a assay
The cells were plated at an effector:target (E:T) cell ratio of 1:1 (10 5 effectors:10 5 targets) in 160 μl of R/10 medium in a 96-well plate. An anti-CD107a antibody was added and incubated with the cells at 37°C for 1 h before Golgi Stop was added and incubated for an additional 2.5 h. The anti-CD8 and anti-CD3 antibodies were added and incubated at 37°C for 30 min. After incubation, the samples were washed once and subjected to flow cytometry with a BD FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA). The data were analyzed with the FlowJo software (FlowJo LLC, Ashland, OR, USA).

CFSE-based T-cell proliferation assay
The RNA electroporation, stimulation and flow cytometry analyses were performed as previously described. 17 In brief, resting CD4 T cells were washed and suspended in phosphate-buffered saline at a concentration of 1 × 10 7 cells per ml. Then, carboxyfluorescein succinimidyl ester (CFSE) was added to the T cells at a final concentration of 2 μM at 25°C for 3.5 min. The labeling reaction was stopped by adding 10 volumes of 5% fetal bovine serum (in phosphate-buffered saline), and the cells were then washed and cultured in R/10 medium. After an overnight culture, the CFSElabeled T cells were electroporated with the indicated RNA. Two to four hours after electroporation, the T cells were suspended in R/10 medium at a concentration of 1 × 10 6 cells per ml. The K562, K562-CD19 or K562-CD86 cell lines were irradiated and suspended in R/10 medium at 1 × 10 6 cells per ml. The cells were plated at an E:T of 1:1 (5 × 10 5 effectors:5 × 10 5 targets) in 1 ml of complete R/10 medium in a 48-well plate. The T cells were then counted and fed every 2 days beginning at day 3. The CFSE dilution was monitored by flow cytometry at the indicated time points.
Flow cytotoxic T-lymphocyte assay A slightly modified version of a 4-h flow cytometry cytotoxicity assay was performed as previously described. 8,18 Figure 1. Blinatumomab BiTEs were secreted from RNA-transferred T cells and bound to T cells for tumor recognition. The T cells were electroporated with an RNA encoding CAR19 (CAR RNA), blinatumomab BiTEs (Blina-RNA) or GFP (at an RNA dose of 10 μg of RNA per 0.1 ml of T cells per electroporation). Eighteen hours after electroporation, the T cells were stained with a goat anti-mouse IgG Fab (mIgG Fab) to detect the expression of the CAR or blinatumomab on the T-cell surface (gated on CD3 + T cells) (a). Eighteen hours after electroporation, the CAR RNA or BiTE RNA T cells alone or mixed with an equal amount of GFP RNA T cells (GFP) were tested for their lytic activity using a cytotoxic T-lymphocyte assay at the effector:target ratio of 5:1 (b). The supernatant from the Blina-RNA T cells (Blina-RNA Sup.) was collected 18 h after electroporation, diluted 10 (1/10) or 100 times (1/100) with culture medium, added to T cells that were not electroporated with any RNA (No RNA) and co-cultured with the CD19 + cell lines (Nalm6, K562-CD19 or Raji cells). The K562 cell line was used as a negative control. T cells that had been electroporated with the CAR RNA or Blina-RNA were used as positive controls in the CD107a assay (gated on CD8 + T cells) (c) (representative of three independent experiments). *P o0.05; **P o0.01.

Mouse xenograft studies
These studies were performed as previously described, with certain modifications. 6 In brief, 6-10-week-old NOD-SCID-γc − / − mice were bred in house under an approved Institutional Animal Care and Use Committee protocol. Approximately 1 × 10 6 Nalm6-CBG cells were injected into each mouse via the tail vein. The T cells were injected via the tail vein 5 days after the Nalm6-CBG cells were injected. Tumor growth was monitored by bioluminescence imaging as previously described. 6 Statistical tests To compare survival among the groups of treated mice, the log-rank (Mantel-Cox) test was used to determine statistical significance. The leukemia burdens, as measured by the bioluminescence imaging of the different groups, were compared with the Mann-Whitney test. Student's t-test was performed to compare differences in T-cell proliferation, lytic activity and cytokine levels. To further confirm that BiTEs were produced by the Blina-RNA T cells, the supernatant collected from the Blina-RNA-electroporated T cells was added to non-electroporated T cells in a CD107a assay. Indeed, the non-tumor-reactive T cells became highly tumor reactive when the supernatant collected from the BiTE RNA T cells was added (Figure 1c).

Functional
The BiTE RNA T cells exhibited superior sensitivity and killing ability, with prolonged tumor reactivity It has previously been shown that T cells can be re-directed by loading an extremely low amount of BiTEs. 19 Our previous work has shown that the T-cell activities of the RNA CAR-T cells are associated with the CAR RNA input. 6 In a 4-h cytotoxic T-lymphocyte assay, the lytic ability of T cells transferred with 1, 5 or 10 μg of either the Blina-RNA or CD19-BBZ CAR RNA was compared. As shown in Figure 2a (Figures 3c and d). The finding that CD45RO − naive T cells with a low dose of Blina-RNA were more sensitive to the stimulation with or without CD28 co-stimulation is consistent with a recent finding that naive T cells proliferate at a lower threshold with reduced co-stimulation requirements compared with memory T cells. 20 To exclude the possibility that this CD28 co-stimulationdependent proliferation of CAR RNA T cells was due to the use of the BBZ configuration in the CAR construct, in which there is no CD28 signaling moiety, T cells transferred with the CD19-28Z CAR were used in a new experiment to assess the division and proliferation of CAR RNA-and Blina-RNA-electroporated T cells. There was no significant difference between the CD19-28Z and CD19-BBZ RNA T cells for both the CFSE dilution ( Figure 3e) and T-cell expansion (Figure 3f) after stimulation with the K562-CD19 cells, with or without additional K562-CD86 cells. However, as shown above, the proliferation of both the CD19-28Z and CD19-BBZ RNA T cells was significantly lower than that of the Blina-RNA T cells, even in the presence of additional CD28 co-stimulation.

T cells expanded by REP showed enhanced anti-tumor activities in vitro
Broad ex vivo cell expansion is required to produce a large number of effector T cells for adoptive immunotherapy, such as RNA electroporation-based adoptive immunotherapy. The REP approach Treating leukemia with BiTE-transferred T cells X Liu et al expanded T cells up to 1000-fold in 2 weeks, which is~10 times more than produced by CD3/CD28 bead stimulation (data not shown). Compared with T cells expanded by CD3/CD28 bead stimulation, which were primarily central memory T cells, the phenotypes of the REP-expanded T cells were more heterogeneous (CD45RO + /CCR7 + 32.00 ± 9.07% (REP) versus 70 ± 6.72% (beads), Po0.01), with fewer cells expressing CD62L (65.83 ± 3.38% versus 85.02 ± 11.13%, respectively, Po0.05), whereas there was no signi ficant difference in CD27 (75.36 ± 4.08% versus 80.97 ± 18.73%, respectively, P40.05) and CD28 expression (71.74 ± 10.70% versus 78.92% ± 9.72%, respectively, P40.05; Figure 4a). Unlike the REPexpanded tumor-infiltrating lymphocytes (TILs), which are mainly CCR7-negative effector or effector memory cells, the REP T cells that were expanded from peripheral blood mononuclear cells were heterogeneous, and the majority of this population maintained an young phenotype (Figure 4a). Upon stimulating the CAR RNA-or Blina-RNA-electroporated T cells that were generated by REP or CD3/CD28 bead stimulation, we found that the REP-expanded T cells showed enhanced anti-tumor activities for both the RNA CAR and Blina-RNA T cells, as evidenced by increased CD137 expression (Figure 4b) and increased lytic ability (Figure 4c), particularly when the RNA input dose was limited. Moreover, the Blina-RNA T cells showed superior in vitro anti-tumor activities compared with the RNA CAR T cells for both the CD3/CD28 bead and REP expansion methods (Figures 4b and c).   Figure 5b). Consistently with results of the leukemia burden measured by bioluminescence imaging, a significant survival benefit was observed for the mice that had been treated with the Blina-RNA-transferred REP T cells (Figure 5c).

DISCUSSION
It has been shown that BiTEs are highly cytotoxic against various cell lines, using unstimulated human peripheral blood mononuclear cells in the absence of co-signaling, and that BiTEs can mediate the serial killing of many target cells. 19,21 Although effective, BiTEs have a short half-life, necessitating continuous systemic infusions, which may be associated with toxicity and a lack of active biodistribution, and, similarly to conventional monoclonal antibodies, they do not self-amplify. 12,22 The feasibility of delivering BiTEs using lentivirally transduced human T lymphocytes, 13,14 human mesenchymal stem cells and other cells 23,24 has been reported and tested in xenograft animal models, and these results have shown that integrating BiTEs with adoptive T-cell transfer in situ may overcome some of the shortcomings of systemic BiTE infusions to enhance the cancer treatment efficiency. However, although these studies have demonstrated the feasibility of redirecting T-cell activity with BiTE gene transfer, the methods generally do not allow us to control T-cell activities to prevent unwanted side effects, such as tumor lysis syndrome and cytokine release syndrome. In clinical studies with permanent CAR-engineered T cells that target CD19, the patients remained disease-free and displayed persistent engineered T cells for 44 years post treatment, although they also showed ongoing B-cell aplasia due to the targeting of normal CD19-positive B cells. Thus, these results highlight the practical need to eventually ablate the engineered cells and reconstitute normal B cells.
Because of the self-decay of the anti-tumor activity of RNA CAR-T cells, multiple infusions of a controlled dose of RNA CAR-Ts may potentially limit tumor lysis syndrome and cytokine release syndrome, and provide the additional benefit of avoiding longterm B-cell aplasia. However, the anti-tumor efficacy is limited by the transiently expressed CAR on the RNA-electroporated T cells compared with lenti-or retrovirally CAR-transduced T cells. 7,8,25 In this study, we found that treating Nalm6 leukemia with REP T cells that were transferred with a blinatumomab RNA significantly prolonged the survival of the treated mice, results comparable to our observations in the same mouse model treated with CD19 CAR lentivirally transduced T cells. Therefore, the combination of T cells generated by REP with the RNA electroporation of BiTE RNAs has the potential to cure CD19 + malignancies with controllable toxicities and without long-term B-cell aplasia. Thus, integrating BiTEs with RNA-electroporated T cells may provide a safer and effective therapy to target CD19-positive tumors.
Currently, ACT uses viral and non-viral integration techniques that usually target only a single TAA. The selective pressure of an antigen-specific immune response may lead to the outgrowth of cancer cells lacking the target antigen, leading to incomplete treatment and treatment relapse, a phenomenon recently documented in failed treatments with CD19-specific CAR T cells. 26 Preclinical ACT studies in mice have suggested that bystander destruction of the tumor has an important role in controlling tumors, [27][28][29] and epitope spreading has been proposed as a mechanism by which T-cell-mediated killing of a limited population of tumor cells can lead to the immunologic destruction of other tumor cells expressing unrelated antigens. 30,31 T lymphocytes are considered to be antigen-presenting cells and are actively being incorporated into cancer vaccines. 32,33 A clinical trial using T cells that had been electroporated with a CAR messenger RNA targeting mesothelin to treat two cancer patients with pancreatic cancer and mesothelioma has reported that the T cells exhibited anti-tumor activity, and the repeated infusions of the messenger RNA CAR T cells elicited an anti-tumor immune response in both patients, as revealed by the development of novel anti-tumor antigen antibodies. 5 A recent study has reported that proximal contact between T cells and live target cells through BiTEs directly reactivated the patients' pre-existing T cells to react with other cancer antigens via epitope spreading, thus supporting the hypothesis that the BiTE brings the T-cell receptor close to the tumor cell and facilitates direct recognition of new peptide-major histocompatibility complex epitopes on the tumor. 34 Thus, repeated infusions of BiTE RNA T cells may not only kill the cancer cells that express the target antigen but also serve as a vaccine to mount immunologic immune responses against other tumor cells that express unrelated antigens, thus potentially preventing relapse from the outgrowth of targeted antigen-negative cancer cells.
Generally, it is believed that the T-cell anti-tumor efficiency primarily depends on the differentiation status of the adoptively transferred T cells, wherein T-cell differentiation is inversely correlated Figure 5. REP-expanded T cells further improved the in vivo anti-leukemia activities of the blinatumomab BiTE RNA T cells. NOD-SCID-γc − / − mice were injected with 1 × 10 6 Nalm6-CBG cells (intravenous), and 5 days later, the animals were treated with 3 × 10 7 CD19-BBZ CAR RNA or Blina-RNA T cells generated by CD3/CD28 Dynal beads (beads) or REP, respectively, for the first treatment, followed by 5 × 10 6 cells each, twice per week for 2 weeks, starting 8 days after the Naml6-CBG injection. The bioluminescence imaging (BLI) was conducted at the indicated time points (a), and the BLI and survival curves were plotted (b, c). Representative of three independent experiments.
with the in vivo anti-tumor effectiveness. 35,36 The use of REP to expand TILs yields T cells with a loss of expression of CD28 and CD27; 37,38 these expanded TILs are prone to apoptosis and hypo-responsive to re-stimulation with tumor antigens. 39 Instead of starting with TILs, T cells isolated from peripheral blood mononuclear cells were used for REP expansion in our current study. In contrast to T cells stimulated with soluble anti-CD3 antibody (OKT3) and IL-2, which produce more effector memory-type T cells with a significantly reduced number of CD62L + , CD28 + and CD27 + T cells that manifest inferior in vivo anti-leukemia ability compared with cells generated with CD3/CD28 beads, 40 the majority of the REP-expanded T cells included both central memory and effector memory T cells and maintained a young phenotype (CD28 + , CD27 + and CD62L + ). When transferred with a CD19 CAR RNA or the Blina-RNA, those T cells not only exhibited more potent anti-tumor activity in vitro, probably because of the presence of more effector memory T cells, but also showed improved anti-tumor activity in an aggressive leukemia model compared with the CD3/CD28 bead-expanded T cells, which exhibited a more uniform central memory phenotype. These results suggest that T-cell heterogeneity may be important in controlling tumors, at least in T cells that are transiently transferred with CAR or BiTE RNAs, for which long-term persistence is not required, and proper migration and the maximum tumor killing ability are more important.
In summary, our study showed that BiTEs can be produced and delivered by T cells transferred with messenger RNAs encoding BiTEs, which may overcome the limitations of using exogenously administered BiTEs and improve the current CAR T cell-based adoptive immunotherapies against leukemia. Moreover, these T cells may be ideal vehicles to deliver other cargo molecules to orchestrate the tumor microenvironment and further facilitate anti-tumor activities. T cells generated using different methods for adoptive immunotherapy may significantly alter the outcome of the treatment; therefore, the selection of an optimized T-cell expansion and culture system is critically important to achieve the maximum therapeutic efficacy of this form of treatment.