Boosting In Vivo CAR-Redirected Virus-Specific CTLs With Universal-Artificial Antigen Presenting Cells

Virus-specific cytotoxic T lymphocytes (CTLs) targeting EBV and/or CMV and engrafted with chimeric antigen receptors (CARs) can receive appropriate costimulation from professional antigen presenting cells (APCs) that process latent viral antigens and can then target tumor cells through their CAR. In a clinical trial in neuroblastoma patients, CAR-redirected virus specific CTLs persisted for over 6 weeks after infusion and produced complete tumor responses in 3/11 patients. To improve the in vivo expansion and persistence of CAR-redirected virus specific CTLs, we determined whether K562 cells engineered to express viral antigens and either CD40L or OX40L can act as universal artificial APCs (aAPCs) to boost CAR-modified virus-specific CTLs in vivo through a cross-presentation mechanism. Using CMV as the viral antigen model, we found that aAPCs/pp65 produce superior in vitro activation of CMV-specific CTLs from seropositive healthy donors (292±56 IFNγ spot forming cells (SFC)/10⁵ cells) than control non modified K562 (83±25 IFNγ SFC/10⁵ cells). The frequency of CMV-CTLs was further enhanced if the aAPC coexpressed CD40L (aAPCs/pp65/CD40L) (502±104 SFC/10⁵ cells; p<0.001) but not by incorporation of OX40L (aAPCs/pp65/OX40L) (357±40 SFC/10⁵ cells) or the combination CD40L/OX40L (477±91 SFC/10⁵ cells). We then assessed whether these aAPCs can boost in vivo CMV-specific CTL responses in NSG mice engrafted with human cells from CMV-seropositive healthy donors. Spleens isolated from mice that had been vaccinated with either aAPCs/pp65 or aAPCs/pp65/CD40L or aAPCs/pp65/OX40L showed an increased frequency of CMV-specific precursors (57±24 vs. 41±14 vs. 35±17 IFNγ-SFC/10⁵ cells) as compared to mice infused with control aAPCs (27±6) (p=0.005). In contrast to the in vitro experiments, the combination in vivo of aAPCs/CD40L/pp65 and aAPCs/OX40L/pp65 elicited the highest number of CMV-specific precursors (101±21 SFC/10⁵ cells). We then tested if CMV-CTLs retained this response to the aAPC vaccine when the CTLs were grafted with a CAR. As a model we used a CAR targeting the neuroblastoma associated antigen GD2. We found that in vitro stimulation with aAPCs/CD40L/OX40L/pp65 promoted the greatest increase of CMV-specific (925±201 IFNγ SFC/10⁵ cells) and CAR-GD2 responses (2725±585 IFNγ SFC/10⁵ cells). The effector function of CAR⁺ CMV-CTLs was maintained through both their native TCR and their CAR as indicated by cytotoxicity against both GD2⁺ target cells (CHLA-255=63±14% at a 20:1 E:T ratio) and pp65-infected cells (59±3%); there was no killing of GD2- targets (Raji = 18%±8%) or of CMV uninfected cells (3%±1%). Sustained persistence of CAR⁺ CMV-CTLs in response to the aAPCs was also maintained in vivo. Thus, vaccination of NSG mice with aAPCs/CD40L/OX40L/pp65 enhanced the frequency of CAR⁺ CTLs (range 2.4% - 6.25%) as compared to mice vaccinated with control aAPCs (0.98% - 1.08%). IFNγ ELISpot assays confirmed increased functional frequency of both CAR and pp65 mediated recognition of T cells isolated from mice vaccinated with aAPCs/CD40L/OX40L/pp65 against GD2⁺ targets (71±24 SFC/10⁵ cells) and pp65-infected targets (85±16) as compared to control mice (23±8 and 41±11, respectively) (p=0.048 and p=0.035). Importantly, vaccination increased the antitumor activity of CAR-CTLs in a xenograft model of neuroblastoma since 48% of the mice vaccinated with aAPCs/CD40L/OX40L/pp65 were tumor free by day 40, while mice vaccinated with control aAPCs uniformly succumbed to the tumor (p≤0.037). In conclusion, we have identified a broadly applicable strategy to stimulate CAR redirected virus-specific CTLs in vivo through their native TCR that lengthens their persistence and thereby increases their antitumor activity.