Engineering better chimeric antigen receptor T cells

The extracellular structure conventionally includes single-chain variable fragment (scFv) and spacer. ScFv is the antigen-binding domain of CAR structure, which is composed of a single heavy and light chain of monoclonal antibody connected by a linker. A murine derived scFv has been reported to induce HLA-restricted T cell-mediated cellular immune response or humoral immune response [13], while humanized scFv is expected to reduce the immunogenicity of CAR, thereby avoiding immune mediated rejection and improving the persistence and therapeutic efficacy of CAR T cells [14, 15]. The different antigen-targeted scFv on the T cells membrane has different impact on CAR T cells fates. Sustained tonic CD3ζ phosphorylation in GD2 targeting CAR, which is triggered by antigen-independent clustering of CAR scFv, can induce early exhaustion of CAR T cells, and such tonic signaling is present to varying degrees in all CARs studied [16]. Moreover, the density of CAR surface expression which can be controlled by using different eukaryotic promoters has a substantial impact on antitumor efficacy and persistence of CAR T cells [17]. Low level expression of CARs driven by weaker rather than by stronger promoters results in significant higher numbers of circulating and tumor infiltrating CAR T cells in mice [18]. As is known that the affinity of CAR binding to antigen is the premise of CAR T cells function, but lower rather than higher affinity CAR-incorporated CAR T cells showed enhanced expansion, better overall and event-free survival and lower toxicity compared with FMC63 CAR T cells [19].

The extracellular spacer domain of CAR has no signaling function, but the length and composition of the extracellular spacer domain influence the spatial configurations and Fc receptor binding, and further the in vivo persistence and antitumor effects of CART cells [20, 21]. Therefore, optimizing the design of scFv and spacer can be effective approaches to improve the therapeutic efficacy of CAR T cells.

The transmembrane domain which is usually derived from CD8 or CD28 connects the extracellular antigen binding domain with the intracellular signaling domain. Compared with T cells expressing CARs with CD28 hinge and transmembrane domains, T cells expressing CARs with CD8α counterparts produce lower levels of cytokines and exhibit lower levels of activation-induced cell death (AICD) [22].

The intracellular structure domain involving CD3ζ and co-stimulatory molecules is essential to full T cell activation. However, CAR incorporating all three CD3ζ immunoreceptor tyrosine-based activation motifs (ITAMs) may foster counterproductive T cell differentiation and exhaustion. CAR encoding a single ITAM balanced the replicative capacity of long-lived memory cells and the acquisition of effective antitumor function, favoring persistence of CAR T cells with enhanced therapeutic potency [23].

First-generation CAR T cells utilizing CD3ζ signaling alone showed limited efficacy in clinical trials probably owing to the AICD or the lack of long-term T cells expansion in vivo. The co-stimulatory molecules are critical for CAR T cells expansion and persistence following adoptive therapy. Second-generation CAR using combination of CD3ζ and an additional co‑stimulatory signaling domain such as CD28 or 4-1BB demonstrated remarkable efficacy with enhanced in vivo expansion and persistence [24‐28]. Subsequently, CARs incorporating other co‑stimulatory signaling moieties such as CD27, ICOS and OX40 [29‐31], or multiple co‑stimulatory molecules in tandem emerged and demonstrated different characteristics on their biological functions (Table 1). CD19-targeted CAR T cells incorporating the 4-1BB co-stimulatory domain show more persistent than those incorporating CD28 in clinical trials [16, 32, 33]. Compared to CAR T cells with CD28, 4-1BB is associated with more antiapoptotic proteins and reduced T cells exhaustion [16, 34]. Changing a single amino acid residue asparagine to phenylalanine in CD28 domain (CD28-YMFM) promotes durable antitumor effects with reduced CAR T cells differentiation and exhaustion as well as increased Th17 skewing [35]. Engineered CD28 lacking lck binding moiety (Delta-CD28) may also be a promising co‑stimulator for better CAR T cells persistence [36]. After antigen stimulation, CD27-bearing CAR T cells proliferate with upregulated Bcl-X_L expression and resisted apoptosis [29]. CAR containing inducible co-stimulator (ICOS) generates IL-17-producing effector cells with the characteristic of TH17 cells showing enhanced persistence in vivo [30, 37]. OX40 co‑stimulatory signaling effectively represses IL-10 secretion, and contributes to counteract self-repression, thus facilitates a prolonged CAR T cells response [31, 38]. Toll-like receptor 2 (TLR2)-incorporated CAR T cells demonstrate improved expansion and persistence due to the capacity of generating memory T cells, expressing pro-survival proteins and abolishing the suppression of regulatory T cells [39‐41]. CD40 signaling contributes to memory formation and rescuing T cells from exhaustion [42, 43]. Combination of CD40 and TLR adaptor molecule MyD88 signaling in CAR T cells results in improved persistence and enhanced efficacy in hematological malignancy and solid tumor models [44, 45].

Considering the dependence on certain cytokines and interactions with endogenous immune cells, CAR T cells with accessory genetic modifications are designed to secrete cytokines or express ligands in order to benefit from autocrine or survive in the tumor microenvironment (Table 2). IL-7 is known to enhance the proliferation and survival of T cells, and co-expression of IL-7 improves NKG2D-based CAR T cells therapy on prostate cancer by enhancing the expansion, inhibiting cell apoptosis and exhaustion [46]. The strategy of co-expressing IL-7 receptor which triggers the IL-7 signaling axis but is unresponsive to extracellular cytokine increased the persistence and antitumor activity of CAR T cells [47]. CCL12 and CCL19 is chemoattractant for T cells and DCs. Combined expression of IL-7 and CCL21 significantly improved survival and infiltration of CAR T cells and DCs in tumor [48]. IL-7 and CCL19 expression in CAR T cells improves immune cell infiltration in the tumor, where both recipient conventional T cells and administered CAR T cells generated memory responses against tumors [49]. CAR T cells with transgenic expression of IL-15 and IL-21 show superior in vivo expansion, persistence and antitumor activity against hepatocellular carcinoma with a higher percentage of stem cell memory and central memory populations after manufacturing [50]. CAR encoding a truncated cytoplasmic domain from IL-2 receptor β-chain (IL-2Rβ) and a STAT3-binding tyrosine-x-x-glutamine (YXXQ) motif activates the JAK/STAT pathway and triggers gene expression profiles analogous to those triggered by IL-21, facilitating a better CAR T cells persistence [51]. Introduction of IL-12 into the tumor microenvironment holds the promise of improving CAR T cells cytotoxic function, mitigating Treg-suppression and reprogramming tumor-associated macrophages and dendritic cells [52, 53]. IL-23, consisting of IL-23α p19 and IL-12β p40 subunits, is known to promote proliferation of memory T cells, and engineered CAR T cells with expression of the p40 subunit show improved antitumor capacity with increased granzyme B and decreased PD-1 expression via autocrine IL-23 signaling [54]. CD40L/CD40 is important to T cells activation, proliferation and memory formation [42, 55]. The altered tumor phenotype with increased immunogenicity (upregulation of CD80, CD86, HLA and Fas) which is susceptible to apoptosis can be indued by CD40L expressing CAR T cells [56]. 4-1BBL armored CAR T cells exhibited balanced tumoricidal function and increased T cell persistence accompanied by elevated CD8/CD4 ratio and decreased exhaustion due to the induction of the IRF7/IFN-β pathway [57]. Accessory ICOSL resulted in enhanced CAR T cells activity via enhanced PI3K signaling pathway and upregulated expression of Bcl2 [58].

Variability in CAR T cells expansion is in part due to the contamination of the starting peripheral blood mononuclear cells (PBMC) concentrated with monocytes. Among patients whose CAR T cells expanded poorly, manufacturing incorporating a monocyte depleting plastic adherence step could yield an adequate dose of CAR T cells for clinical use [59]. A density-based negative selection for T cells enrichment using antibody cocktail presents the advantage of leaving T cells free of antibodies [60], thereby reducing the probability of excessive or inadequate cells activation. The most commonly used method for enrichment of T cells is magnetic adsorption using CD3/CD28 antibody coated beads, but the purity of T cells can’t be guaranteed. The CAR gene may be transduced into leukemic cells during T cell manufacturing, and its product bound in cis to the CD19 epitope on the surface of leukemic cells, masking it from recognition by and conferring resistance to CD19-targeted CAR T cells [61]. Therefore, the existence of CAR-tumor cells needs to be taken into consideration when relapse occurs, and the methods for T cells purification need further exploration.

Anti-CD3/CD28 beads stimulation provided a suitable activation for clinical use of genetic modification of T cells [62], as well as the application of soluble anti-CD3/CD28 antibody or anti-CD3/CD28 coated plate. Of note, in addition to anti-CD3 for T cell activation, anti-CD28 is needed to avoid T cell anergy. Compared to soluble anti-CD3, anti-CD3/CD28 beads stimulated greater CD4 + T cells growth, but both stimulated similar CD8 expansion [63]. The conjugation of anti-CD137 antibodies to conventional anti-CD3/CD28 beads results in a significant increase in the expansion capacity for CD8 cytotoxic T cells [64]. Besides, irradiated feeder cells from PBMC in the presence of anti-CD3 antibody have been shown to be a high effective approach for expansion of CD8 + T cells with a high ratio of central memory-like T cells [65].

The most common and effective methods for gene transduction in CAR T cells production is using viral vectors carrying CAR transgene, including γ-retrovirus and lentivirus. These vectors are widely used for T cells modification and are successful in clinical trials due to the stable and durable expression of target genes by integration into the genome of host cells. However, the random integration of viral vectors may result in variegated transgene expression which ultimately influence the in vivo persistence of CAR T cells. The non-viral gene editing, especially the CRISPR/Cas9-mediated site-specific gene integration, results in uniform CAR expression, which averts tonic CAR signaling and establishes effective internalization and re-expression of the CAR following single or repeated exposure to antigen, delaying effector T cells differentiation and exhaustion [66].

In order to obtain sufficient amount of CAR T cells, a proper duration for T cell expansion ex vivo is necessary. Most protocols for T cells engineering routinely expand T cells for 9–14 days, but there is a trend of shortening recently. The potential for engraftment and persistence of CAR T cells are related to the state of T cells differentiation, and reducing the duration of ex vivo culture could limit cells differentiation and enhance the efficacy of CAR T cells therapy [67].

IL-2 is necessary for T cell culture, but it may drive T cell differentiation. Low concentration or short-term use of IL-2 favors generation of early memory T cells over effector phenotypes during CAR T cells expansion, thereby enhancing therapeutic efficacy and saving production cost [68, 69]. Addition of soluble IL-7, IL-15 and/or IL-21 in the process of cell culture can reduce the CAR T cells terminal differentiation and increase the frequency of memory stem cells, yielding improved in vivo persistence [70‐72].

T-cell activation is a necessary step in the production of CAR T cells, and the robust cytotoxicity of CAR T cells depend on the antigen recognition by chimeric antigen receptors and subsequent activation signaling. However, excessive activation drives T cells differentiation and exhaustion [73], which rationalizes the notion that appropriate activation is essential to optimize therapeutic potency and durability of CAR T cells. Modulation of multiple links in the activation signal pathway can affect the state of CAR T cells differentiation and exhaustion. With CARs encoding a single ITAM, CAR T cells show less differentiation and exhaustion [23]. Inhibition of PI3K preserves a less differentiated state of CAR T cells without affecting cell expansion [74, 75]. MAP Kinase inhibition protects tumor-infiltrating CD8 T cells from death driven by chronic TCR stimulation [76]. Akt inhibitor treated CAR T cells present an early memory phenotype and superior antitumor efficacy owing to the intranuclear localization of T cell memory associated transcriptional regulator FOXO1 [77, 78]. Different from unmodified T cells, CAR-modified T cells with sustained activation signaling induced by spontaneous aggregation of CAR molecules lead to cell differentiation and exhaustion during ex vivo expansion, limiting their therapeutic efficacy and in vivo persistence [16, 74]. Thus, in addition to direct regulating activation signaling, it might be useful to modulate the density of CAR molecules on the cell surface in order to prevent the aggregation of CARs induced sustained cell activation. This is consistent with the observations that the level of CAR surface expression correlates with the in vivo persistence of CAR T cells, and the density of CAR can be controlled by modifying the promoters.

The initiation of T cell antigen receptor (TCR) signaling is a key step that result in T cell activation and the orchestration of an adaptive immune response. Theoretically, CAR T cells retain the structure and function of TCR which can still recognize specific antigens and transduce activation signals. The dual stimulation of CAR and TCR may lead to loss of CD8 CAR T cells efficacy owing to excessive activation induced cell exhaustion and apoptosis [79]. Allogeneic CD19- specific CAR T cells- mediated anti-lymphoma activity without causing a significant increase in the incidence of graft-versus-host disease (GVHD) also suggests that both the TCR and CAR are engaged to accelerate T cells exhaustion [80]. CAR T cells with TCR knock-out exhibit high anti-leukemic effects and long-term persistence in the absence of alloreactivity [81]. Directing a CD19-targeted CAR to the alpha constant locus of TCR which undergoes structural disruption not only results in uniform CAR expression, but also enhances CAR T cells therapeutic potency with delayed effector T cells differentiation and exhaustion [66].

According to the stage of cell differentiation, T cell can be categorized as Naive T cell (T_N), stem cell memory T cell (T_SCM), central memory T cell (T_CM), memory effector T cell (T_EM), effector T cell (T_EF). It is well defined that the therapeutic efficacy and in vivo persistence of CAR T cells significantly correlate with their differentiation stage. CAR T cells products rich in T_N,T_CM and T_SCM exhibit superior antitumor responses and long-term persistence in vivo [98, 114, 115]. Both of CD8 and CD4 CAR T cells derived from TN and T_CM are more effective than those from T_EM [98]. Transcriptomic profiling revealed that CAR T cells from complete-responding patients with CLL were enriched in memory-related genes, whereas CAR T cells from non-responders upregulated programs involving effector differentiation, and sustained remission was associated with an elevated frequency of memory-like T cells before CAR T cells generation [116].

Most of current approaches for maintaining durable memory CAR T cells in vivo are improving the capacity of memory formation in the stage of manufacture. These approaches including design refinement of CAR structure, control of activation signaling, epigenetic regulations and optimization of in vitro culture, ultimately regulate the expression or activities of T cell differentiation-associated transcription factors (TFs). Numerous TFs have been identified as critical regulators of conventional CD8 T cell differentiation, including memory-associated TFs (e.g., TCF-1, Eomes, Id3, Bcl-6 and FOXO1) and effector-associated TFs (e.g., T-bet, Blimp-1, Id2, and IRF4) [117, 118]. Considering that CAR T cells may share the similar differentiation-associated transcriptional regulations as conventional T cell, gene modifications or small molecule drugs targeting these TFs may be promising for CAR T cells memory maintenance.

Upon antigen exposure, naive T cells are activated and differentiate into multiple types of effector T cells, and the differentiation is accompanied by robust proliferation, transcriptional, epigenetic and metabolic reprogramming [117, 119]. Following the peak of effector expansion, the resolution of inflammation and the clearance of antigen, most activated T cells die, but a subset persists and transits into the memory T cells pool which maintains the ability to rapidly reactivate effector functions upon re-stimulation [120]. CAR T cells may undergo the similar process of differentiation following antigen recognition after therapeutic infusion. However, the strategies of in vivo intervention for CAR T cells memory maintenance still need further exploration.

CAR T cells present less effective always because of exhaustion which occurs both in vivo and in vitro. During in vitro expansion, spontaneous CAR molecule aggregation-induced sustained tonic signaling lead to CAR T cells exhaustion [16]. After therapeutic infusion, upon repeated antigen stimulation, part of CAR T cells enter into an exhausted state which is defined by poor effector function, sustained expression of inhibitory receptors (e.g., PD1, TIM3 and LAG3) and widespread transcriptional and epigenetic alterations distinct from that of functional effector or memory T cells [73, 120, 121].

At the level of transcription, AP-1 related bZIP-IRF families have been identified as major modulators that drive exhaustion-associated gene expression. Exhaustion-associated dysfunction resulted from increased levels of AP-1/IRF complexes leads to a functional deficiency in activating AP-1 Fos/Jun heterodimers, whereas c-Jun overexpression prevents phenotypic and functional hallmarks of exhaustion and improves antitumor efficacy [60]. NFAT/NR4A axis which controls the expression of multiple inhibitory receptors plays a crucial role in T cell exhaustion, and CAR T cells lacking all three NR4A transcription factors results in tumor regression and prolonged survival in tumor-bearing mice [122, 123]. Transcription factor TOX cooperates with NR4A to regulate T cells exhaustion [124, 125], and disruption of TOX activity could be a promising strategy for enhancing therapeutic potency of CAR T cells.

Inhibitory receptors such as PD1, TIM3 and LAG3 are recognized as the markers of T cells exhaustion, although they also increase significantly in response to cell activation [126]. Tumor cells or tissues up-regulate the inhibitory ligands (e.g., PDL1) upon immune attack, while CAR T cells up-regulate the inhibitory receptors (e.g., PD1) in response to antigen exposure, and the PDL1/PD1 pathway significantly inhibits CAR T cells activity. Thus, intervention of PDL1/PD1 pathway emerges as a promising approach for reinvigorating CAR T cells in certain clinical application. Strategies for intervention of PDL1/PD1 axis developed recently involving CAR T cells modified to secrete PD1 blocking scFv [127], disruption of PD1 by gene editing [128, 129], and the application of monoclonal antibodies targeting PDL1 or PD1. Of note, compared to 4-1BB/CAR T cells, CD28/CAR T cells are more susceptible to exhaustion that PD-1/PD-L1 blockade confers a priority [130, 131]. And even, downregulation of three inhibitory receptors (PD1, TIM3 and LAG3) simultaneously on CAR T cells show increased tumor infiltration and durable tumor control [132].

The exhaustion of CAR T cells limits their efficacy and durability in vivo, but in certain application, exhaustion may present a positive effect. Due to the exhausting alloreactivity caused by cumulative TCR and CAR signaling, allogeneic CAR T cells demonstrated reduced risk of GVHD in mice model [80, 133].