CAR T cells: building on the CD19 paradigm

Abstract

Spearheaded by the therapeutic use of chimeric antigen receptors (CARs) targeting CD19, synthetic immunology has entered the clinical arena. CARs are recombinant receptors for antigen that engage cell surface molecules through the variable region of an antibody and signal through arrayed T cell activating and costimulatory domains. CARs allow redirection of T cell cytotoxicity against any antigen of choice, independent of MHC expression. Patient T cells engineered to express CARs specific for CD19 have yielded remarkable outcomes in subjects with relapsed/refractory B cell malignancies, setting off unprecedented interest in T cell engineering and cell-based cancer immunotherapy. In this review, we present the challenges to extend the use of CAR T cells to solid tumors and other pathologies. We further highlight progress in CAR design, cell manufacturing and genome editing, which in aggregate hold the promise of generating safer and more effective genetically instructed immunity. Novel engineered cell types, including innate T cell types, natural killer (NK) cells, macrophages and induced pluripotent stem (iPS) cell-derived immune cells, are on the horizon, as are applications of CAR T cells to treat autoimmunity, severe infections and senescence-associated pathologies.

Introduction

CD19-targeted chimeric antigen receptors (CD19 CARs) have opened a path for novel immune therapies in oncology and other fields of medicine. This therapeutic innovation is predicated on advances in genetic engineering, tumor immunology and cell manufacturing sciences [1–4]. Its foundational principles consist in genetically instructing T cells to recognize any chosen antigen, thus by-passing the restrictions of repertoire limitation and immune tolerance, and accelerating the establishment of immunity by providing potent manufactured immune cells to the patient, rather than inducing an uncertain endogenous response through active immunization. By using synthetic receptors for antigen that directly engage cell surface molecules, in contrast to the physiological T cell receptor (TCR), T cells can circumvent HLA restriction and the limitations of antigen processing [5, 6]. These principles, validated by targeting CD19 in B cell malignancies and T cell manufacturing implemented in the academic setting, have established CD19 CAR therapy as a prototype for synthetic immunity and a paradigm-shifting form of personalized medicine.

CARs are synthetic receptors that may engage proteins, carbohydrates or glycolipids independently of antigen processing and HLA expression. Similar to the natural TCR/CD3 complex, CARs require an activating domain to couple antigen recognition to the initiation of T cell activation and cytolysis [7]. However, studies enabled by retroviral-mediated T cell engineering [8] demonstrated that the CD3-ζ chain activation domain is not sufficient to sustain T cell function in primary T cells [9]. Engineered costimulation [10] eventually provided a solution when it was integrated into a single dual-signaling receptor that enabled T cells to not only kill their targets but expand upon repeated exposure to antigen [11]. The path to “living drugs” was opened. We eventually renamed CD3-ζ chain fusions as first generation CARs and dual-signaling receptors as second generation CARs [12]. Second generation CARs are the backbone of present day CAR therapies.

CD19 is a cell-surface molecule that is found in most B cell malignancies, which we initially chose as a CAR target over other B cell surface molecules such as CD20 and CD22 because of its relatively higher expression [13, 14]. Building on the demonstration that CD19-specific CARs could eliminate established leukemia and lymphoma in mice [14], Memorial Sloan Kettering, the National Cancer Institute and the University of Pennsylvania undertook clinical trials in patients with refractory/relapsed CD19+ malignancies. Early anecdotal results obtained in NHL [15], CLL [16] and ALL [17, 18] were soon confirmed in larger single- and multi-center studies [19–22]. The first CARs to be approved by the FDA in 2017 and the EMA in 2018, were second generation CARs specific for CD19 [4]. Those CAR molecules encompass either a CD28 [12] or 4-1BB [23] costimulatory domain (Figure 1A). Four CAR products are approved to date (Tisagenlecleucel, Axicabtagene ciloleucel, Brexucabtagene autoleucel, Lisocabtagene maraleucel), for relapsed or refractory (r/r) pediatric and young-adult B-cell acute lymphoblastic leukemia, diffuse large B-cell lymphoma, high-grade B-cell lymphoma, primary mediastinal large B-cell lymphoma, mantle cell lymphoma and certain follicular lymphomas. Several excellent clinical reviews addressing the efficacy of CD19 CAR T cells in different disease settings as well as their toxicity, which include B cell aplasia, cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS), are available [24–29].

Figure 1.

CAR structure, function and design. (A) Prototypic CD19 CARs. (B) Novel 28z CAR designs. Left: 1928z CAR T cells bearing a single functional ITAM in membrane-proximal position (1XX CAR) display increased persistence while retaining strong effector functions [41]. Right: CARs comprising a truncated cytoplasmic IL2Rb and a YXXQ STAT3 binding motif (28-ΔIL2RB-z(YXXQ) CAR) enable antigen-induced JAK-STAT activation [43]. (C) Novel BBz CAR designs. Left: Increasing ITAMs from 3 to 6 in a 4-1BB-based car (BBzz CAR) enhances antigen sensitivity and IL2 secretion [42]. Right: Altering the length of extracellular and intracellular domains in a 4-1BB-based CAR (BBz(86) CAR) reduced cytokine secretion and CRS severity [44]. Panel A has been partially published in [37].

The success of CD19 CAR therapy has spawned broad interest in this novel approach to immunotherapy. There are presently over 700 clinical trials listed at clinicaltrials.gov, 41% of which target CD19 and the rest either hematological malignancies (32%) or solid tumors (27%) (Figure 2A). CD22 and CD20 have proven to be effective CAR targets in B cell malignancies, albeit to a lesser degree than CD19 [30, 31], while BCMA shows great promise for the treatment of multiple myeloma, with BCMA CAR T cells under consideration for approval by the FDA [32–34]. This review addresses the next frontiers and challenges facing the implementation of CAR therapy in oncology and beyond.

Figure 2.

Summary of clinical trials targeting CAR T cells.

A. Clinical trials targeting CD19, other hematologic and solid malignancies. Based on clinicaltrial.gov (March 2021): 41% of worldwide clinical trials aim at CD19 CAR T cells. This includes dual targeting of CD19 and CD22 CAR T cell treatments. 32% of all clinical trials target hematological cancers but not CD19. 27% of all clinical trials target solid tumors.

B. CAR targeted antigens under clinical investigation for hematologic and solid tumors. Based on clinicaltrial.gov (March 2021): Analyzing the distribution of antigens of interest used for directing CAR T cells in clinical trials, reveals that anti-CD19 scFv is the leading CAR T used (37%) followed by the anti-BCMA scFv (9%). Other CAR T cells directed to additional hematology or non-hematology antigens are lacking behind with less than 4% out of all CAR T clinical trials.

C. Cell types engineered for CAR therapy. Based on clinicaltrial.gov (March 2021): The majority of CAR T cells are transduced using the αβ T cell (>96%). Other cell types are starting to be investigated, accounting for 4% of listed clinical trials. These cells include: γδ T cells, NK, and NK/T cells, pluripotent stem cell-derived immune cells and monocytes.

Evolution of CAR design

Present day CARs comprise a single-chain Fv (scFv) for antigen recognition and a dual-signaling tail, typically comprising the CD3-ζ and either CD28 or 4-1BB cytoplasmic domains [35–38]. Different scaffolds, adapted to the epitope’s position on the targeted molecule, can further improve overall CAR function [39]. The binding domain is in some cases derived from a receptor-ligand pair or may consist in a VHH element rather than an scFv [40]. This architecture is modular and can be adapted to a wide range of targets (Figure 1A).

CAR designs are evolving, utilizing affinity-optimized human binding domains and recruiting different signaling pathways (reviewed in [37]). One example of such evolution is the reduction of strength of activation in the potent 28z CAR design [41] and another its augmentation in the less potent BBz [42] (Figure 1BC). Other examples are the additional recruitment of jak/stat signaling in 28z CARs (Figure 1B) or adding spacer elements to attenuate cytokine secretion by BBz CARs (Figure 1C) [43, 44].

In contrast to the single specificity of natural T cells, CAR T cells may engage multiple antigens, through the design of multi-specific CARs or CAR co-expression. Targeting two independent antigens may reduce tumor escape and augment selective pressure on the tumor cell. The first dual-targeted CAR T cells have entered the clinic, combining two BBz CARs specific for CD19 and CD22 [45]. Combining a 28z CAR with a BBz CAR provides yet superior antigen sensitivity and potency, which is further enhanced by matching signaling features to antigen density [46]. A variant approach requires the successful recognition of two upregulated tumor antigens by two different CARs to initiate full immune cell functions, as demonstrated by Eshhar’s group using dual-targeted CAR T cells against myeloma cells. The results not only show efficient and potent CAR T but also a safe strategy sparing normal healthy tissues [47].

CAR T cell therapy for solid tumors

The only CAR T cells to be approved by the US Food and Drug Administration (FDA) and the European Medicine Agency (EMA) at this time are directed to CD19 for the treatment of some refractory leukemias and lymphomas. Adapting CAR therapy to the treatment of solid tumors is a major goal and challenge [48–51]. The first results obtained in solid tumor trials, modeled on the successful strategies implemented for B cell malignancies, have been lackluster, pointing to the need to further improve and specifically adapt CAR therapy for solid tumors. Overcoming the mechanisms of resistance to CAR therapy already encountered in the setting of B cell malignancies is likely to help reach this goal.

One mechanism of resistance is the absence or insufficient level of expression of the CAR target in tumor cells. Reports from multiple trials have shown that up to 25% of patients treated with CD19-targeted CAR T cells relapsed with CD19-negative or CD19-low disease [52] This phenomenon is known as antigen escape. Another mechanism of resistance can be traced to the limited ability of CAR T cells to traffic to and infiltrate the tumor, due to physical barriers and/or an immunosuppressive tumor microenvironment. In the realm of B cell malignancies, it has been frequently noted that responses may be complete in bone marrow but not other disease sites such as lymph nodes or the retroperitoneal space [53]. Even infiltrated, it is to be expected that solid tumors will deploy a range of mechanisms to inhibit T cells and CAR T cells, which involve myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs), and regulatory T cells (Tregs) [54]. The successful tackling of solid tumors by CAR T cells will thus entail, at the least, the Identification of suitable target antigens, enabling CAR T cells access all tumor sites and overcoming the immune suppressive effects of the tumor microenvironment.

Target selection

The first attempts to tackle solid tumors replicated the approaches established in B cell malignancies, utilizing the same CAR designs (either 28z or BBz) and targeting a single antigen. While good responses have been sporadically obtained in GBM, head and neck squamous cell carcinoma and prostate cancer [55], the overall responses are not comparable to CD19 CAR therapy. One limitation owes to the paucity of potential targets that, like CD19, are expressed on the surface of nearly all tumor cells and only in dispensable normal cells. Although few if any differentiation antigens meet these two criteria, some proteins overexpressed in tumors, although not tumor-specific, show promise, such as mesothelin, PSMA, GPC3 and a few others [49]. Targeting tumor antigens that are found in normal tissues may however be possible if one were to calibrate CAR function within a particular therapeutic window in order to limit reactivity to normal cells, carefully selecting scFv’s, signaling components and transcriptional regulation [56–58]. Antigen-specific inhibitor CARs may also come of use to protect normal tissues [59]. Figure 2B illustrates the distribution of CAR target antigens currently under clinical investigation for hematologic and solid tumors based on clinicaltrial.gov.

Cell surface antigens that are specifically modified in tumor cells are an attractive alternate class of CAR targets. Glycans, including glycolipids, N- and O-linked glycoproteins, and glycosaminoglycans (GAGs), can undergo specific glycosylation in cancer cells. They include Lewis^Y in advanced epithelial cancers, sialyl-Lewis^A in pancreatic adenocarcinoma, G_D2 in neuroblastomas and some gliomas, and Tn-MUC1 in various adenocarcinomas. The involvement of multiple enzymes in glycan biosynthesis and their redundancy accounts for their stable expression and may diminish the risk of antigen escape. Tumor cells would indeed have to lose several of the >20 GalNAc polypeptide transferases that contribute to O-glycosylation to cease generating Tn antigen [60, 61]. Glycoprotein expression may however still be lost if the protein backbone were mutated (e.g., loss of MUC1 for Tn- MUC1).

Tumor infiltration

Like any T cell, CAR T cells need to extravasate into the tumor site and overcome immune suppression in the tumor microenvironment (TME) to be effective (Figure 3). Immune cell trafficking into peripheral tissues is regulated by complex signaling and physical processes that may be significantly disrupted in some tumors. The challenge posed by immunologically “cold tumors” is a general obstacle for all immunotherapies [49]. T cell engineering however allows for unique solutions. For example, rate limiting entry into the tumor may be alleviated in some cases by chemokine directed migration, for example by overexpressing the IL8 receptor, CXCR2R (Figure 3A), in CAR T cells [62]. Another approach is to direct CAR T cells to the tumor surroundings, for example, CAR T cells may be targeted to fibroblast associated protein (FAP), which is found in stromal fibroblasts in the TME. CAR T cells targeting FAP effectively infiltrate solid tumors but may also be myelotoxic (Figure 3B) [63]. Another approach makes use of CAR T cells secreting enzymes to forge a path, such as heparanase to degrade heparan sulfate proteoglycans (Figure 3C) [64].

Figure 3.

Selected novel strategies to enhance the efficacy of CAR-T cell therapy for solid tumors are illustrated: A. CAR-modified T cells express a chemokine receptor for increasing their ability to traffic/home to the tumor site (e.g., CXCR2). B. FAP specific CAR-T cells can direct CAR-T cells to the tumor surroundings. C. Heparanase expression enhances CAR-T cell forge a path in the TME. D. Intra-tumoral administration of CAR T cells in solid tumors. E. Combination therapy with monoclonal antibodies targeting immune-checkpoint inhibitory receptors to relieve immunosuppression. F. SiRNA to reduce receptor function or exhaustion markers (e.g., silencing the adenosine receptor). G. Engineered CAR T cells that secrete pro-inflammatory or CAR-T supporting cytokines such as IIL-2, IL-7, IL-12 or IL-15. H. A variety of metabolic strategies (e.g., L-Arginine) proposed to be integrated to optimize the manufacturing process and maximize the therapeutic efficacy of CAR T cells.

Local/regional CAR T cell delivery in solid tumors may not only maximize the accumulation of CAR T cells at the tumor site but also improve their safety profile by limiting their systemic distribution (Figure 3D). Regional CAR therapy targeting mesothelin in mesothelioma has shown potent and long-lasting CD4-dependent tumor immunity [63]. providing greater anti-tumor efficacy compared to intravenous administration [64].

Microenvironmental immune suppression

Tumors and their stroma attract and support a number of immunosuppressive cell types, including a heterogeneous population of myeloid-derived suppressor cells (MDSCs), macrophages, Foxp3+ Tregs, fibroblasts and platelets. A broad range of inhibitors acting on TME-associated immune suppression is under investigation [65]. Overcoming the TME by combining such agents with CAR T cells may prove to be beneficial (Figure 3). Blockade of inhibitory immune checkpoints that enforce immune suppression is one promising option. Checkpoint inhibition may be achieved by combining CAR T cells with the systemic administration of immune checkpoint inhibitory antibodies [63, 66, 67] or in cell intrinsic fashion, by ablating for example PD-1 [68] or reducing its function with a dominant-negative receptor [63][59]. Interfering RNA provides another means to reduce but not fully ablate receptor function, for example silencing the adenosine receptor and other exhaustion markers such as the TIM-3, LAG-3, TIGIT, and KLGR-1 (Figure 3E) [58, 69, 70]. Adenosine mediated T cell suppression can be counteracted in other ways as genetic targeting of A_2AR using shRNA [71] (Figure 3F), or by adenosine-responding CAR by fusing the extracellular domain of adenosine receptor to intracellular costimulatory domains, thus converting switching a normally suppressive signal to an activating signal [72] .

CAR T cells may be harnessed to reprogram the TME. For example, CAR T cells may provide local cytokine secretion, in order to enhance their infiltration, proliferation and persistence. CAR-expressing T cells can be engineered to produce a wide range of cytokines (such as IL-4, IL-2, IL-7, IL-21, and IL-15) (Figure 3G) or their cognate receptors, aiming to improve antitumor activity and persistence of these cells. These CAR T cells are sometimes referred to as armored CARs, fourth generation CARs or TRUCKs (T cells redirected for antigen- unrestricted cytokine initiated killing) [73]. For example, TRUCKs engineered to inducibly or constitutively express IL-12 have been shown to activate an innate immune anti-tumor response and modify tumor immunosuppression [74].

The heightened glycolytic metabolism of cancer cells may limit nutrients available to CAR T cells in the tumor bed, impeding their function or survival thus affecting the microenvironment and suppressing the immune cells [75]. Supplying the CAR T with nutrients using pre-treatment in vitro feeding or inhibiting specific metabolic pathways of the tumor has been shown to enhance CAR T survival, providing for example L-arginine to CAR T cells prior to their adoptive transfer (Figure 3H) [76]].

CAR toxicities

CAR T cells are powerful immune effectors that may in some instances precipitate serious toxicities that require expert and urgent medical management [24–27]. The occurrence of these acute toxicities is still difficult to predict for any given patient and hinders the broad implementation of CAR therapies.

The first toxicity was anticipated and owes to the on-target/off-tumor of CAR T cells, resulting in B cell aplasia in the case of CD19 CARs [14, 77, 78]. Different approaches have been proposed to restrain toxicity to normal cells. One makes use of trans second- generation CARs with split costimulatory configuration combining conventional activating constructs with inhibitory CARs, could potentially enable greater target precision, by ensuring that the effector cells only reach full activation in the presence of a defined array of tumor-associated antigens [79]. Another approach incorporates inhibitory CARs to retrain T cell function [80–83]. The unknown toxicity of novel CAR T cells can also be limited by infusing T cells transiently expressing the CAR construct following RNA transfection, thus minimizing the duration of unforeseen toxicity [84]. In the context of solid tumors, local delivery of genetically modified T cells could also mitigate toxicities while increasing efficacy [85]. Another approach to lower side effects is to find the minimum effective dose. Modeling the amount of CAR T cells needed for a therapeutic effect while reducing the amount of injected CAR T cells to receive a lesser side effect was performed by Globerson Levin et al. and showed the possibility to use far less CAR-T lymphocytes than currently used [79].

Two other toxicities, which were not anticipated when the first CD19 CAR trials opened, have posed a greater challenge to date: cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) [4, 86]. Both toxicities occur early on after CAR T cell administration and may be significant in up to a third of all CAR T cell recipients. The mechanisms of CRS are starting to be well understood [87, 88]. Pathological cytokine secretion owes largely to interactions between CAR T cells and myeloid cells in the TME. While activated T cells produce chemokines and cytokines including IL-2, soluble IL-2R-α, IFN-γ, IL-6, soluble IL-6R, and granulocyte-macrophage colony-stimulating factor, the main source of IL6 and IL1 are macrophages in the vicinity of tumor and CAR T cells. The cornerstone of CRS management is IL6R blockade and corticosteroids [89, 90], but novel interventions including IL1Ra [87, 88], dasatinib [91] and others are emerging [86, 92, 93]. Neurotoxicity is less well understood and is mainly managed with corticosteroids [86, 94, 95].

CAR T cell activity may be terminated by reversibly shutting off CAR signaling or irreversibly eliminating CAR T cells through the action of a suicide gene or other elimination switches [96]. Suicide genes can be included in the transfection construct, which trigger apoptosis upon induction with a specific drug [92, 93][87, 88].

Thus, while acute CAR T cell toxicity presently limits the widespread use of CAR T cells, one can anticipate this limitation being eventually removed, given the rapid progress in CAR design and our understanding of CRS pathophysiology.

CAR T cell manufacturing

T cell transduction

T cell engineering was initiated with the use of γ-retroviral vectors [6, 8] and later extended to lentiviral vectors and DNA transposons, all of which mediate stable long-term transgene expression. Messenger RNA transfection may also be used for the purpose of transient CAR expression. Most γ-retroviral vectors are derived from the Moloney murine leukemia virus and the myeloproliferative sarcoma virus, and have been optimized to promote high level transgene expression [96]. Three of the four FDA- approved CAR-T cell products, Breyanzi, Tecartus and Yescarta utilize γ-retroviral vectors to deliver CD19 specific CARs into T cells [97, 98]. Most lentiviral vectors are derived from HIV-1 and are pseudotyped with the VSV-G envelope [99]. Both γ-retroviral and lentiviral vectors mediate high level and stable CAR expression. The constitutive expression of the fusogenic VSV-G glycoprotein has so far precluded the establishment of stable packaging cell lines, which is a significant burden for vector production, but lentiviral vectors are effective. Kymriah, the first 4-1BB CAR approved by the FDA and EMA, is a T cell product manufactured with vector produced by transient transfection [100]. DNA transposon-based systems, such as Sleeping Beauty (SB), have been developed to produce CAR T cells for clinical investigation and initially demonstrated modest efficacy [101]. When compared to viral vectors, the limitation of SB includes the low genetic transposition efficiency hence the need to propagate CAR T cells by stimulation with artificial antigen-presenting cells and cytokines resulting in extended ex vivo culture duration. More recently, site-specific CAR integration using targeted nucleases in conjunction with adeno-associated viral vectors as donor DNA for CAR transcription units, has enabled to tightly regulate CAR expression from the T-cell receptor α chain (TRAC) locus [102]. The site-specific double-stranded break enabled by gene editing tools such as CRISPR/cas9 and guide RNAs are subsequently repaired by homology-directed repair (HDR) [103].

Autologous CAR T cells

Autologous CAR T cell manufacturing begins with the collection of patient or donor blood or leukapheresis. CD3+ T cells are used as starting cells for CAR T cell manufacturing. Removal of tumor [104][99] cells and monocytes [105–107] can be mediated by positive selection of CD4 and CD8 T cells. CD4/CD8 T-cell selection has been shown to improve CAR T-cell manufacturing feasibility and resulted in heightened inflammatory toxicities when compared to products derived from whole apheresis, leading to dose de-escalation [108] in a phase I clinical trial of patients with relapsed/refractory CD22+ malignancies [109]. The generation of CAR T cells from a T cell population with defined properties has the potential to provide a more consistent source of material from one patient to another. It has been shown in animal models that CAR T cells derived from either TN naïve [110], TCM central memory [111, 112] or TSCM memory stem cell subsets [113] possess enhanced anti-tumor activities when compared to CAR T cells derived from bulk T cells. CD8 TCM subsets have also been demonstrated to be active in pre-clinical models [111, 112, 114]. CAR T cells derived from the CD62L+ TCM subset have been used in the clinical setting in patients with glioblastoma and promoted tumor regression in one patient [115]. CAR T cells with defined ratios of CD4+CAR+ and CD8+CAR+TCM, or CD4+CAR+ and CD8+CAR+ potentially mitigate toxicity and have promoted disease-free survival upon infusion in patients with B-ALL and NHL [112]. The possible therapeutic benefit of T cell subsets needs to be evaluated upon long-term patient follow up and in larger patient cohorts to determine if its benefits will justify the increased complexity of the manufacturing scheme and additional costs related to the selection procedure.

T cell activation

T cell activation with OKT3 antibody can be modulated with costimulatory molecules such as CD28. Off-the-shelf cGMP beads conjugated with anti-CD3 and anti- CD28 antibodies have been developed to optimize T-cell selection and expansion ex vivo (magnetic Dynabeads™ [116], T-cell TransAct™ biodegradable nanobeads) [117, 118]and are widely used for the activation of T cells in the context of clinical CAR T cell manufacturing [109].

Alternatively, artificial antigen presenting cells (AAPCs) such as the K562 CML cells have been customized and decorated with costimulatory molecules to activate and expand subsets of T cells such as CD8+, antigen-specific and CD19-CAR-T cells [111, 119]. Irradiated autologous PBMCs [120] alone or together with Epstein-Barr virus (EBV)-specific LCLs [121] in combination with OKT3 antibody and IL2 have also been used to generate CD19- and CD20-specific CAR T cells, respectively. Nonetheless, the number of clinical applications with AAPCs as T cell activators is rather limited due to the complexity of generating AAPCs according to current good manufacturing practices (cGMP).

T cell expansion

Several platforms such as static culture in flasks and bags, wave-mixed bioreactors and expansion in continuous bioreactors are available to enable manufacturing of clinical CAR T cells (reviewed in Wang and Riviere) [117]. Wave-mixed bioreactors are wildly used by academic centers and biotech companies for CAR T cell expansion to support early stage clinical trials [109]. Other integrated manufacturing platforms such as the CliniMACS Prodigy System are fully enclosed and designed to incorporate T cell selection, activation, transduction, expansion and formulation, enabling a higher degree of automation, supporting continuous cell culture, and largely decreasing interpersonal operator variability. Phase I clinical trials conducted with CD19 CAR T cells manufactured in the CliniMACS Prodigy have generated encouraging clinical responses which have the potential to enable the concept of point-of-care manufacturing [122–126]. Another newly available integrated end-to-end cell manufacturing solution is the Cocoon platform, which is currently being investigated in a clinical trial at the Sheba Medical Center, Israel [127](NCT02772198). The need to improve the anti-tumor efficacy of autologous patient derived T cells is prompting the development of new culture reagents, technologies, and manufacturing platforms [128]. In particular, minimally manipulated CAR T cells kept in culture for less than 24hr have demonstrated some level of anti-tumor activities in animal models and in human subjects. The optimization of this approach could simplify CAR-T manufacturing and has the potential to drastically decrease the overall cost [129, 130].

Allogeneic CAR T cells

While autologous CAR T cell therapies require a bespoke manufacturing process for every patient leukapheresis, off-the-shelf and allogeneic CAR therapies offer the prospect of eluding the variable quality and impairment of patient T cells, reducing the variability in yield and homogenizing the phenotype of infused CAR T cells. Allogeneic CAR T cells would also shorten the delay to T cell infusion once CAR T cells were prescribed. Allogeneic approaches however face two major challenges. The first is the risk of causing graft- versus-host disease (GVHD) and the second is the rapid elimination of allogeneic cells by the host immune system, limiting their anti-tumor activity. Several solutions are available to address the first challenge, using either allogeneic CAR T cells derived from a stem cell transplant donor, virus-specific memory T cells, non-αβ T cells and/or gene editing to ablate TCR expression, reviewed in [131]. The second challenge, to avoid rapid elimination, is more complex and still awaits an effective solution.

T-iPS

The generation of immune cells from induced pluripotent stem cells (iPSCs) offer an alternative platform to produce “off-the-shelf” and synthetic allogeneic T-cells [132]. Proof-of-principle studies support the feasibility of this approach [133, 134]. iPS-derived NK cells have already entered the clinic (NCT03841110) and iPS-derived T cells will be investigated shortly in subjects with relapsed/refractory B-cell Lymphoma (NCT04629729).

In vivo CAR T cell generation

CAR T cells could potentially be engineered in vivo, altogether by-passing ex vivo manufacturing. This technology is at very early stage of development and its potential remains unknown. Early studies using γ-retroviral vectors, adenoviral vectors and lipid nanoparticles, suggest the feasibility of transducing T cells in vivo [84, 135].

Genetic engineering and genome editing

Genome editing offers new prospects for T cell engineering [136]. The precision afforded by CRISPR, zinc-finger nucleases (ZFNs), and transcription activator-like effector nucleases (TALENs) is useful to not only knock-out endogenous genes but also to knock-in transgenes [137]. T cells edited to remove their endogenous TCR or inhibitory receptors such as PD-1 have already been tested in pre-clinical and clinical trials [68, 138–140]. The abrogation of HLA class I expression via the ablation of ß₂-microglobulin is being actively pursued to build allogeneic universal CAR T cells [141]. These mechanisms allow efficient and relatively simple gene knock-out and are pursued for the generation of universal T cells, exhaustion- resistant T cells, and to reduce the off-target toxicity of redirected T cells by fratricide[142].

Clinical trials using either TALENs or CRISPR/Cas9 for gene ablation are summarized in [136]. Currently 4 clinical trials make use of TALEN technology and a dozen utilize CRISPR/Cas9 to address, exhaustion, fratricide killing or allo-rejection. The preclinical results and data from early clinical trials are encouraging [68], but more clinical experience with these technologies, in particular CRISPR-Cas9, is required to confirm safety and efficacy.

Precision engineering may also be used to insert genes at selected genomic locations to achieve regulated transgene expression. Thus, CARs expressed from the TRAC locus, which encodes the T cell receptor alpha chain, are expressed at an optimal level that is sufficient for effective expression but minimizes tonic signaling and premature T cell exhaustion [102]. TRAC targeting, which simultaneously ablates endogenous TCR expression, has since been adopted by many groups developing allogeneic approaches, while we pursue this approach to demonstrate that the generation of optimally engineered CAR T cells will reduce T cell dosing and toxicities in the autologous setting. We are also actively searching for genomic safe harbors [143] to engineer CARs and other transgenes to reliably regulate expression of cytokines, chemokines, secreted scFv’s, bites and other molecules in CAR T cells.

Novel cell types and technologies using CAR therapy

Virus-specific T cells (VST), natural killer (NK), invariant NKT cells.

One of the major evolutions of CAR therapy is the diversification of cell source [132] (Figure 2C). CAR constructs were initially investigated in conventional αβ-T cells, but are now investigated using other cell types including virus-specific T cells (VST), natural killer (NK), invariant NKT cells, γδ-T cells and myeloid cells. At first, memory cells with specificity towards human cytomegalovirus (CMV) or Epstein-Barr virus (EBV) were evaluated. EBV and CMV infections induce a CMV-specific CD8⁺ T-cell pools with an effector-memory phenotype. CMV T cells isolated and used for transduction in an attempt to increase T cell persistence were evaluated [144, 145]. CARs are also being investigated in γδ-T cells[133, 146], invariant NKT cells [147] and NK cells [45]. NK cells have become highly attractive for their potent effector functions and safety based on their cytokine secretion profile and lack of induction of graft-versus-host disease (GVHD) in allogeneic settings [148]. Similar characteristics can be attributed to NKT cells. G_D2 CAR NKT cells have been administered to patients with neuroblastoma[149]. Still, over 96% of all CAR clinical trials use conventional αβ-T cells for CAR T cells manufacturing (Figure 2C).

γδ T cells

A small subset of peripheral blood cytotoxic T cells are the γδ-T cell. Similar to CAR T cells, they do not need antigenic presentation by MHC molecules for recognition and function. γδ-T cells function across MHC-barriers, and do not cause graft-vs-host disease. Their anti-tumor activity has been demonstrated using Vγ9Vδ2 T cells in preclinical studies and early phase CD20 CAR γδ T cells for r/r NHL (NCT04735471) [146].

Macrophages

The ability of macrophages to infiltrate solid tumors and be reprogrammed, as well as the antitumor effects associated with a switch to the M1 phenotype, have been exploited in cancer immunotherapy and make these cells a desirable carrier for CARs. CAR macrophages (CAR-M) have been shown to reduce the tumor burden and prolonged overall survival in mice [150]. The potential of engineered CAR macrophages in cancer immunotherapy was recently illustrated in an in vivo mouse model of HER2-expressing ovarian cancer cells [151].

CAR T exosomes

Albeit acellular, this approach derives from cell-based approaches, aiming to enhance tumor infiltration by using exosomes that carry CARs on their surface [152]. CAR-containing exosomes express a high level of cytotoxic molecules and can inhibit tumor growth without CRS [153]. Their potential toxicity of T cell-derived exosomes remains to be investigated.

CARs beyond cancer

Infectious disease still represents a major threat worldwide. Despite the availability of a range of treatment options, the incidence rates and mortality rates of some infectious diseases remain high, exemplified by the COVID19 pandemic. T cells play a key role in the control of most of these diseases, supporting the potential application of T cell-based therapeutic approaches, such as CAR-T-cell therapy. Other non-malignant diseases including colitis, systemic lupus, GVHD, autoimmune diseases, fungal diseases, cardiac fibrosis and cellular senescence may also benefit from CAR T cells [154].

Infectious diseases

The gp120 envelope protein expressed in HIV-infected T cells has been targeted by engineered CD4 and CD8+ T-cells [144, 155]. Another approach is based on HIV-specific CAR-T cells lacking CCR5 expression, which exhibited antiviral effects. This approach blocks virus propagation, and importantly, precludes integration of the virus into the host genome where it could persist in a latent state. There are still limitations such as CAR T-cell expansion, persistence, and potential off-target effects. A similar approach could be used to target other viruses including CMV, and HBV [154]

Transplantation and autoimmunity

HLA-A2-specific CAR Tregs have been used to protect against graft-versus-host disease (GVHD) and skin transplant rejection in immunodeficient mice. Human HLA molecules in the context of HLA-disparate transplantation are ideal targets for CARs, as the antigen is abundant and expressed solely on the transplanted organ, thus will affect only on the grafted T cells eliminating the GVHD. Moreover, the ligation of HLA molecules by CAR Tregs is unlikely to have any negative effect on graft cell function, as these molecules have no signaling potential [154].

Treating autoimmune diseases is another new focus of CAR therapy. Several approaches are in pre-clinical development to target different autoimmune mechanisms: 1. Targeting the soluble antibodies formed by using a chimeric autoantibody receptor (CAAR T) to inhibit autoantibody reaction; 2. Using T regs to inhibit T cell autoreactivity, and 3. Eliminating autoreactive B cells to halt the accumulation of autoantibodies by directing the CAR T against the specific B cell [156, 157]. Specifically, treatment with ex vivo expanded Tregs has shown promise in inflammatory bowel disease (IBD), albeit with impaired Treg accumulation and function at inflammatory sites. Tregs can be efficiently transduced to express functional, antigen- specific chimeric receptors that enable the specific suppression of effector T cells. This approach may enable future cell-based therapeutic applications in inflammatory bowel disease, as well as other inflammatory disorders [155, 158–160].

Senolytic CARs

Senescence is a cellular program that leads to irreversible cell cycle arrest in response to stress. Defective immune clearance of senescent cells results in the latter’s accumulation and the establishment of a pro-inflammatory environment that contributes to a variety of pathologies associated with chronic tissue damage and aging, including fibrosis, atherosclerosis and some neurodegenerative conditions [161]. As shown in mouse models of liver fibrosis, CAR T cells directed to cell surface markers selectively expressed on the surface of senescent cells can function as effective senolytic agents [162], opening a path for developing engineered immune cells to remove senescent cells and abrogate inflammation in a number of pathologies.

Cardiac diseases are a leading cause for mortality worldwide. Most myocardial diseases lead to cardiac fibrosis, which diminishes organ function and contributes to heart failure. In a murine model of cardiac injury induced by angiotensin II and phenylephrine, CAR T cells targeting FAP in interstitial fibroblasts have also been shown to facilitate myocardial regeneration, without adversely affecting wound healing [163]. Perivascular fibrosis persisted, consistent with the absence of FAP expression in perivascular fibroblasts, but without myelotoxicity. Altogether, the studies support the potential of CAR therapy to address pathologies associated with senescence, fibrosis and chronic inflammation.

Conclusions and prospects

CD19 CAR T therapy has opened Pandora’s box for cell-based immunotherapy and synthetic immunology. These advances will rest on progress in receptor and circuit design, tumor immunology, genetic engineering and cell manufacturing sciences. The clinical results achieved with CD19 CAR T cells over the past 10 years have swayed the pharmaceutical industry to explore and invest in cell-based immune therapy, which no previous adoptive cell therapy had achieved. Considerable work lies ahead to develop a range of personalized therapies, which will require the identification of suitable targets, overcoming the tumor microenvironment, breaking into cold tumors, minimizing toxicities and further advancing T cell manufacturing, whether in autologous, allogeneic or in vivo settings. The potential of immunotherapy using CAR T cells is vast in cancer and beyond, including autoimmunity, senescence-associated pathologies and infectious diseases. While the challenges remain considerable, it seems that more applications will arise, and that the CAR T cell era is only at its beginning.