Effectors of the Future: Universal Chimeric Antigen Receptor Open Access

Redirecting immune effector cells to treat cancer has been exploited with antibodies, such as rituximab and trastuzumab since the 1990s (1997 and 1998 US-FDA approval, respectively) [1] and the milestone of the 100th monoclonal antibody product being FDA approved was reached in 2021. During the same time, the use of antibody variable chains in artificial immune receptors was introduced in the 1980s [2, 3] and optimized for 30 years, before the first CD19-chimeric antigen receptor (CAR) T-cell product tisagenlecleucel was FDA approved in 2017 after achieving sustained remissions in a series of pediatric patients with B-cell precursor acute lymphoblastic leukemia [4‒6]. Shortly thereafter, the FDA approved three more CD19-targeted CAR T-cell products, along with two CAR T-cell products targeted to BCMA [5].

The concept of enhancing the effector function by combining CAR-expressing effector cells in T and NK cells with antibodies and their fragments, thereby creating a potentiated artificial version of antibody-dependent cellular cytotoxicity (ADCC), has matured significantly over the past decade. This evolution and synthesis of ideas have been thoroughly reviewed by Arndt et al. [7] in their article entitled “Adaptor CAR Platforms – Next Generation of T Cell-Based Cancer Immunotherapy.”

Artificial immune receptors derived from the T-cell receptor (TcR) are depicted in Figure 1. The prevailing CAR architecture illustrated in Figure 1b is widely acknowledged as a sturdy foundation for the present phase of CAR-based cellular product development. Despite the diversity in available intracellular signaling domains, the fundamental components, such as the costimulatory signaling units sourced from CD28 and 4-1BB, along with the primary signaling driven by the CD3ζ chain [8] – have consistently demonstrated reliable performances across various cancer types in both preclinical models and clinical trials [6, 9‒13].

Alternative signaling domains, such as DAP10 and DAP12, are interestingly recognized for their unique advantages over CD3ζ signaling due to their distinct downstream signaling with reduced exhaustion (PD-1), reduced secretion of inhibitory cytokines (IL10) and enhanced function of CAR⁺ tumor-infiltrating lymphocytes [14, 15]. To date, they are viewed as complementary to CD3ζ, rather than replacements for CD3ζ, or they can be used in combination with CD3ζ [15]. Genetic modifications in the CD3ζ chain leading to the inactivation of immunoreceptor tyrosine-based activation motifs (ITAMs) counterbalance T-cell exhaustion but may also reduce the antigen-density dependent activation threshold of these distinct CARs [16]. However, the high sensitivity of CARs to recognize antigen low expressing target cells (>200 molecules per cell sufficient to induce cytolysis) [17] is a prerequisite for profound cancer cell eradication and sustained remission [6, 18].

Efforts to incorporate the artificial immune receptor into the native TcR structure and harness the TcR-signaling machinery for CAR function have emerged and been revived [2] as an intriguing and alternative strategy compared to the conventional CAR architecture [19‒21]. Native TcRs (Fig. 1a) are comprised of eight protein chains assembled in a defined configuration. The TcRα and TcRβ chains provide the TcR with its specificity through their variable alpha- and variable beta fragment. Additionally, the TcR is comprised of two CD3ε chains, one CD3γ and CD3δ as well as two intracellular CD3ζ chains. Signal transduction by TcRs is initiated by the phosphorylation of potentially ten available ITAMs of which 6 ITAMs are incorporated in the two CD3ζ chains [22]. Conventional CARs (Fig. 2b) are based on an extracellular recognition domain, e.g., single chain fragment variable (scFv), a hinge region, a transmembrane domain and intracellular costimulatory and signaling domains [8]. CAR products approved by the US-FDA are second-generation CAR T-cell therapies featuring a CD28 or 4-1BB costimulatory domain coupled with CD3ζ chain signaling. They are comprised of two identical protein monomers that become fully functional after homodimerization [5]. They contain two CD3ζ chains and thus potentially signal via 6 ITAMs which is 4 ITAMs short of native TcRs or TcR-like immune receptors.

One elegant technology utilizing the natural TcR-signaling machinery is referred to as TcR fusion constructs (TRuCs) (Fig. 1c), in which the extracellular recognition domain (scFv) is fused to the TcRα, TcRβ, CD3γ, CD3δ, or CD3ε chain and thus integrated into the native TcR configuration and therefore can theoretically signal from 10 ITAMs like the native TcR. Since the CD3ε chain is represented twice in the TcR, three different TcRs can be paired in CD3ε-TRuC positive T cells. The TcR in TRuC⁺ cells may remain native (without a CD3ε-scFv), contain one or two CD3ε-scFv for MHC-independent target-antigen recognition. The CD3ε-based TRuC emerged as the most promising candidate, yielding robust activity and response in different in vivo cancer models including B-lineage ALL and lymphoma, multiple myeloma, and glioblastoma multiforme. Surprisingly, the tumor control of the TRuCs was superior to conventional 2nd generation CAR T with incorporated CD28 or 4-1BB costimulatory domain [19]. Another approach integrates an anchored (CD4 or CD8α) bispecific T-cell engager (BiTE) to the cell membrane (Fig. 1d), referred to as T-cell antigen coupler (TAC) connecting the CD3ε chain with the targeted antigen can also signal with 10 ITAMs [21]. Both, the TRuC and the TAC share a similar CD3ε-dependent downstream activation pattern like blinatumomab [23].

To avoid mispairing and enhance correct pairing of the synthetic TcR-like receptor chains, murinized TcR-like CAR T constructs have been shown feasible in preclinical proof-of-concept studies [24]. By far the most advanced idea and technically challenging is to substitute the TcRα and TcRβ variable fragment with the variable light and heavy fragment of antibodies, realized in the HLA-independent TcR (HIT). This concept has been studied in different variations with the hope that TcR-like receptors may be more sensitive than conventional CAR designs since they can employ 10 ITAMs like a native TcR for the downstream signaling. The site-specific integration of the TcR-like CAR construct combines different aspects of T-cell physiology and concepts (Fig. 1e). Thus, the HIT is used to kill two birds with one stone. First, the TRAC integration inactivates the TcR alpha chain and thus reduces the likelihood of mispairing, and second it leads to a more physiological synthetic receptor regulation, comparable to the native TcR and results in a tightly regulated and activation-dependent cell surface expression of the HIT which has been proven to be beneficial for the CAR function [20, 25].

Over the last decade conventional CAR-based therapeutics have created the notion that immunotherapy with genetically modified cells can make the difference in advanced hematological malignancies [26], have the potential to restructure hematology, and sparked high hopes and expectations also in immune oncology. Contemporary CAR architectures have proven to reliably exert anticancer activity in various blood cancers. These homodimeric receptors are comprised of an antigen recognition domain, a hinge region, a transmembrane domain and a combination of intracellular signaling domains [8].

FDA-approved products consist of murine antibody-derived single chain variable fragments (scFv) [6] or camelid heavy chain variable (VHH) [10] based recognition domains. Alternative strategies to overcome immunogenicity, such as fully human VHH (12 kDa) [27], fully human scFv (25–30 kDa) or natural human ligand-based CARs [28], but also artificial small dimerization domains with a molecular weight below 10 kDa [29] have been utilized with success to equip CARs with targeting specificity.

The hinge region connects the recognition domain with the CAR-expressing cell in a certain distance to allow the formation of the cytolytic synapse. The ideal proximity between the CAR-expressing cell and the target cell typically falls within the distance range of approximately 15 nm, akin to the TcR-pMHC complex. However, the optimal spacer length, and thus the ideal distance is influenced and defined by the location, accessibility, and configuration of the epitope on the targeted antigen [30, 31]. Despite the abundance of available proteins, the selection of suitable spacers is somewhat constrained, as it primarily relies on proteins capable of homodimerization, such as CD28, CD8α, or the IgG-derived hinge and constant regions used in contemporary CAR architecture [8]. This limitation also applies to the most frequently utilized transmembrane domains, namely CD28 and CD8α. The costimulatory signaling domains are derived from the immunoglobulin-like coreceptor CD28 family or the tumor necrosis factor receptor superfamily. The individual costimulatory domains are distinct in activation kinetics, induction of cytotoxicity, secretion of cytokines, induction of proliferation [32] but also determine the metabolic signature [33]. The CD3ζ signaling domain stands out as the most commonly and effectively utilized signaling component in CAR receptors, often considered the gold standard [5, 6, 10]. The intentional fusion of functional protein units is diverse and enables tailored and finely tuned functionality to meet specific requirements. Nonetheless, while numerous variations of CAR signaling have proven to be effective in preclinical models, the conventional CAR design was solidified in clinical trials and serves as the cornerstone for any CAR product seeking to leverage clinical data.

Universal CAR technologies or adapter CAR technologies can deliberately redirect CAR-expressing cells to any accessible surface-expressed antigen of interest [7]. The mode of action and its terminology of “redirecting” is key to its function. The primary specificity of the adapter CAR should be designed to be immunologically irrelevant because the targeted epitope is supposed to not be present in the human body under physiological circumstances. Only the administration of adapter molecules introduces the targeted moiety into the system (human body) since the highly defined structure recognized by the adapter CAR recognition domain is only carried by adapter molecules. Therefore, the adapter molecule redirects the adapter CAR highly specifically to the selected surface-expressed target of interest. Conclusively, the targeted epitope of adapter CAR recognition domains is exclusively accessible on the adapter molecules. Otherwise, the CAR may interact with its naturally present epitope and could cause significant toxicities. Therefore, the moiety (connecting domain) targeted by adapter CARs must be nonhuman or nonexistent and non-cross-reactive with structures present in the human body [34, 35], or inaccessible for the CAR on the cell surface of cells [36], and exclusively conjugated to or integrated into the adapter molecules [7].

At the same time, one major limiting factor of adapter CAR function is the technology’s dependency on the availability of the adapter molecule everywhere in the patient’s body. Therefore, the adapter molecule concentration needs to reach CAR-activating concentrations at the tumor site, and the primary antigen density on the target cells must be above the CAR-activating expression threshold [32] and optimally the targeted antigen should not be subject to shedding [37], rapid antigen turnover, and/or internalization of bound adapter molecules [38]. Strategic adapter molecule selection of the molecule format may contribute to realize optimized targeting. Adapter molecules need to possess the capability to permeate all pertinent body compartments and peripheral areas, including low-perfused tissues such as bone or tumor niches and metastatic sites [39]. The penetration of antibodies into the central nervous system for instance is limited, which makes it harder but not impossible to reach sufficient adapter molecule concentrations in all body compartments. Especially, the blood-brain barrier can be overcome by adapter molecule design with CNS penetration capabilities [40] or the direct administration of the adapter molecules into the cerebrospinal fluid, e.g., intrathecal injection or the application through artificial CNS access, such as the Ommaya reservoir an intraventricular catheter system that can be used for the aspiration of cerebrospinal fluid or drug administration for diagnostic and therapeutic procedures [41, 42].

Under ideal conditions, the target cells are recognized by surface-bound adapter molecules on the target cells and facilitate the formation of the cytolytic synapse via the engagement with the adapter CAR-expressing effector cells. In vivo, the pharmacokinetics and -dynamics of both the CAR-expressing cells and the adapter molecules determine the adapter CAR recruitment, engagement, and effector functions. The underlying mechanism of action and the induced basic effector functions of adapter CARs are illustrated in Figure 2.

In conventional CARs (Fig. 2a), the antigen recognition is dependent only on the CAR-expressing cell and the target cells that express the targeted antigen. They are targeted to one single antigen and cannot recognize antigen-negative targets which is one of the major reasons of CAR treatment failure – antigen or epitope loss and antigen downregulation [6]. To address antigen heterogeneity and antigen loss, bispecific (CD19/CD22 or CD19/CD20) [43, 44] and trispecific (CD19/CD20/CD22) [45] CAR T cells have been created to target B-lineage-specific antigens.

In adapter CARs, the adapter molecule interconnects the CAR-expressing cell and the target cells. Thus, the physiology of adapter CARs, in terms of pharmacokinetics and -dynamics, is more complex than that of conventional CAR technologies, sharing similarities with antibody-mediated immunity. The features of conventional CAR and universal CAR technologies are summarized in Table 1.

In contrast to direct CARs, the antigen recognition and the CAR signaling are decoupled from each other in adapter CARs (Fig. 2a), and the CAR-positive immune cells are inactive in the absence of adapter molecules [7]. The adapter molecule is comprised of an antigen-binding domain, a structural domain, and a connecting domain. The most relevant criterion for antibody therapies, bispecific antibody therapy, and CAR therapy is the stereometry of the receptor target interaction to initiate an immune response by interconnecting immune effector cells with target cells. Mechanistically, this entails drawing the cells closer together to enhance the interaction between the immune cells and the target cells, thereby facilitating activation and the formation of the immune synapse. The distance of the immune synapse (Fig. 2a) with close proximity between the immune cell membrane and the target cell membrane is highly conserved in the range of 15 nm (_˜TcR-pMHC interaction distance) and a prerequisite to activate immune cells efficaciously and mediate effects via ADCC [46], BiTE, and CARs [47, 48], whereas antibody-dependent cellular phagocytosis (ADCP) favors slightly greater distances [46]. In conventional CAR design, the operational distance is tailored through 1st the careful selection of the targeted epitope on the antigen [49] and 2nd the size of the spacer defining the length of the extracellular CAR domain (distance) [50]. In adapter CARs, the distance between the CAR-expressing cell and the target cell is extended by the adapter molecule [34, 35]. The additional distance of the adapter CAR cell membrane to the target cell membrane corresponds to the distance between the connecting domain to the antigen binding domain (Fig. 2b). If the connecting domain is directly linked to the N-terminus of the adapter molecule, the extension of the distance can be minimal in the subnanometer range but can also exceed 7 nm if placed at the C-terminal end of a Fab fragment or even 10 nm if placed at the C-terminal end of the CH3 domain of an antibody [51].

Adapter CARs mediate the same effector functions as conventional CAR technologies but require the correct assembly and configuration of the CAR receptor, the adapter molecule, and the targeted antigen. The effector functions are activation threshold dependent and can be categorized in activation stages and as shown in Figure 2c. Minimal activation is required to modify the receptor expression profile and to induce cytolytic activity. In direct comparison, the cytolytic activity of CARs is induced at 10-fold lower antigen density than cytokine secretion [17]. Thus, cytokine and chemokine secretion require a significantly enhanced CAR engagement than cytolysis. Cytokines provide relevant signals to facilitate a robust and sustained immune response, the recruitment of accessory cells and CAR proliferation. As such, they have been incorporated into modern CAR design and are used to enhance both cytotoxicity and the proliferative capacity [52].

Since the specificity of targeting is defined by the adapter molecule, versatile polyimmunotargeting is the greatest feature of adapter CAR technologies to overcome antigen-based immune evasion [34, 53]. Hence, adapter CARs offer the distinct advantage of being activatable through the introduction of adapter molecules with differential specificity [35]. This capability enables transient targeting, synchronized combinatorial targeting, and sequential combinatorial targeting, allowing the primary targeting mechanism to adjust in accordance with the pharmacokinetics and -dynamics of the adapter molecule, increase the overall potency of targeting and spare toxicities [34, 35, 54‒56].

In universal CAR technologies, adapter molecules enable CAR transactivation in the sense of next-generation ADCC [34, 53] and ADCP. The specified terminology for adapter CARs as of adapter molecule-dependent cellular cytotoxicity (AMDCC) and adapter molecule-dependent cellular phagocytosis (AMDCP) cover both, antibody fragments, and alternative adapter molecule formats beyond wildtype IgG [35].

Conventional CARs are more sensitive than antibody-based therapies and universal adapter CAR technologies (Fig. 3a). Most clinical experience has been gained with second generation-based CAR T-cell products in B-lineage cancer targeted to CD19 [4, 6] and BCMA [57] that have illuminated the potential of CAR-based therapies. They have evidenced that artificially enhanced signaling utilizing CARs can outperform any other available therapeutic today including combination therapies with antibodies as well as bispecific antibodies [18, 58]. Adapter CARs are more sensitive and less toxic than BiTEs (Fig. 4a, b) and are more potent than antibody therapies in preclinical models (unpublished data). This is attributed to the target-independent activation of T cells by blinatumomab. Universal CARs require stronger forces to induce target-independent activation.

Blinatumomab is clinically used strictly at serum concentrations of 0.5–0.7 ng/mL [59] due to life-threatening hyperinflammatory complications at higher dosing. In contrast, clinical data concerning the use of adapter molecules at similar low concentrations, ranging from nanograms to micrograms per milliliter, is scarce [60, 61]. The broad therapeutic spectrum of adapter molecule dosing in universal CAR therapy will support reaching clinically relevant concentrations of adapter molecules in various pharmacokinetic compartments [60, 61] as well as sequential or simultaneous combinatorial targeting which is compromised in BiTE therapy due to dose-limiting toxicities [62].

The generation and optimization of adapter molecules are comparable to antibody development including immunization strategies predominantly in mice (Fig. 3b). However, adapter CAR technologies can also build on clinically proven and US-FDA/EMA-approved antibodies [34, 35]. Immunization with proteins, protein fragments, Fc-fusion proteins, peptides, DNA, or mRNA is used to generate new antibodies [63]. The target and epitope selection may vary between conventional antibody therapy and adapter CARs due to the unique requirements for each distinct application [49, 64].

Contemporary antibody optimization includes the humanization and affinity maturation of the antibody of interest to fit their purpose. In most instances, the goal is to increase the antibody affinity during this process [63] and optionally can be grafted into an Fc-modified antibody framework (Fig. 3c), to enhance ADCC, ADCP, and CDC [65, 66]. Contrarily, in adapter CARs either antibody fragments or strategically silenced antibodies [67] are used to circumvent activation-induced cell death which is mediated by the interaction of CAR⁺ cells with FcR-expressing immune cells [68] leading to unwanted in vivo depletion of CARs.

Normally, NK cells are the major immune effector cell population to exert ADCC, and M1 macrophages to mediate ADCP predominantly via IgG1 antibodies [58, 69]. The substantial contribution of ADCP in achieving tumor control with antibody therapies, such as rituximab, cetuximab, and trastuzumab underscores the importance in future efforts to exploit macrophages or induced pluripotent stem cell (iPSC)-derived macrophage-like cells for CAR-based immunotherapies [69‒71]. Although native IgG1 antibodies may exhibit limited recruitment and anticancer function with NK cells due to their generally low affinity to FcγRIIIA in the micromolar range [72], often leading to insufficient downstream signaling and cytotoxicity [58], they robustly mediate ADCP [69]. Advanced antibody technologies enhance the affinity of wildtype IgG1 to FcRγIIIA of K_D 252 nm by greater two log-fold in IgG1_S239d/I332E to K_D 2 nm through the substitution of amino acids in the CH2 domain [66] or similarly by glycoengineering in the means of defocusylation [73‒76]. These modifications enhance the affinity of the modified IgG1 for FcRγIIIA, thereby boosting their anticancer efficacy [77]. Moreover, such changes can transform antibodies into highly potent biological agents with significant clinical effects [58]. Two prominent examples of antibody engineering are represented by the clinical success of the humanized IgG1_S239d/I332E CD19 antibody tafasitamab in the treatment of patients with diffuse large B-cell lymphoma [78] as well as the humanized glycoengineered CD20 antibody obinutuzumab [75] for the administration in patients with chronic lymphocytic leukemia and small cell lymphocytic leukemia [79], despite the need to identify the best functional affinity of therapeutic antibodies which depends on their mode of action, e.g., immune activating antibodies [80] versus antigen-targeted antibodies [81]. Through the evolution of antibody engineering, it has become apparent that both affinities, 1st the K_D of the antibody to the targeted antigen [82] and 2nd the K_D of the immune receptor, e.g., FcRγIIIA to the antibody significantly impact on the overall potency of therapeutic antibody [66, 74]. Antibodies must reach a minimum affinity threshold to exert immune functions and in most instances high affinity antibodies are superior to antibodies with lower affinity, but there certainly is a controversy around the best affinity, e.g., claiming that moderate antibody affinities might increase the penetration of antibodies into the tumor in solid cancer [83]. Absolute affinities for therapeutic antibodies seem to be less important than affinity thresholds [81] if the affinities lie in the low nanomolar or sub nanomolar range, like the EGFR-targeted antibodies panitumumab (IgG2a, K_D 0.05) and cetuximab (IgG1, K_D 0.39) [84]. In general, the best affinity for therapeutic antibodies is to be identified empirically by their biological activity which is also impacted by various factors, such as the binding epitope, the internalization kinetics, the IgG subtype as well as functional modifications [63]. For the use in adapter CARs, the IgG subclass seems irrelevant since the recruitment of effector cells is independent of the primary antibody subclass and function [34, 53]. Therefore, the development of adapter molecules can leverage clinically proven and effective antibodies, such as the CD33-targeted IgG4 antibody gemtuzumab, which features a core-hinge mutation [85]. Additionally, the conversion of the IgG isotype to IgG1 remains an option during the empirical optimization process.

Besides the expression level, accessibility and configuration of the targeted antigen, the proximity of the binding epitope to the target cell surface [86], the stereometry including the presentation of the CAR-targeted moiety (connecting domain), and the antibody affinity are the most relevant functionally determining factors [35]. The optimization of the stereometry (Fig. 3d) by placement (integration or site-specific conjugation) of the connecting domain on adapter molecules is a meticulous and challenging process, yet it is crucial for determining the best system’s potency. Antibodies that are used to mediate natural ADCC, ADCP, and CDC do not allow to optimize the accessibility, configuration, or distance of the highly conserved binding structure for Fc-receptors and complement factors. This is a significant advantage of AMDCC and AMDCP over ADCC and ADCP.

Furthermore, adapter CAR technologies have the potential to enhance the functionality of ADCC and ADCP across different populations of immune effector cells (Fig. 3e). Various CAR-expressing effector cells can theoretically be recruited by adapter molecules of any format with a functionally tailored downstream signaling. This immune modulatory function may be proinflammatory activating, or altered to mediate target-antigen dependent immunosuppression [87]. Deductively, introducing a CAR not only enhances the function of T cells [4, 6] but also extends its benefits to a broad spectrum of other immune cells, including NK cells [88], invariant NK-T cells (iNKT) [89], and macrophages [70]. The ADCC and ADCP conversion into AMDCC and AMDCP facilitates the technology to build on versatile adapter CAR concepts that have been fully optimized for this bespoke purpose including the integration of application- and cell type-specific signaling and allow the flexible optimization of the adapter molecules far beyond the possible modifications and optimizations of basic antibody functions, such as ADCC, ADCP, and CDC [65, 66, 74, 90].

Even though T cells have been the main cell population for CAR therapies, other immune subsets like NK cells, iNKT or macrophages display unique properties which are highly valuable for immunotherapy [70, 88, 89]. The major reason why T cells have been most efficacious in the context of CAR therapies is 1st their exponential proliferation capacity [4, 6], 2nd T cells comprise an autonomously managed immune compartment with various T cell subsets and functions, e.g., T_C1, T_C22, T_H1, and T_H2 [91, 92], and 3rd CAR T cells can be transferred into patients and retain their high capacity for self-renewal and can persist long-term [4, 6, 10], especially through naïve T cells, stem-cell memory, and central memory T cell subsets that are able to engraft permanently and become part of the patient’s integratively regulated immune system [6]. The value of alternative immune cell populations has been recognized and are increasingly being harnessed to enhance anticancer immune responses with NK cells [93], iNKT cells [94] as well as specific subsets of macrophages M1, and neutrophils N1, but on the other hand excluding immunosuppressive, cancer-promoting M2 macrophages and N2 neutrophils [95, 96].

The signaling in CAR-expressing immune cells is expanded and enhanced compared to native immune cells. To further enhance the potency of adapter CARs, additional effector functions can be incorporated into the basic CAR construct and are constitutively [89] or inducibly expressed in an antigen-dependent manner, e.g., under NFAT, AP-1, or NF_KB and others [97, 98] referred to as T cells redirected for antigen-unrestricted cytokine-initiated killing or 4th generation CAR constructs [5, 97]. The transgene expression of additional effector functions in T cells redirected for antigen-unrestricted cytokine-initiated killings is tightly regulated and antigen-dependent via the CAR activation [99] and due to the decoupled signaling in adapter CARs, it is directly related to the presence or absence of the adapter molecule in a dose-dependent manner [34].

To overcome immunosuppressive signals in the tumor microenvironment [100], proinflammatory cytokines have proven to potentiate the CAR performance. Among these are the common γ chain cytokines IL7 [101], IL15 [102], and IL21 [103], as well as IL12 [104] and IL18 [105] triggering an increase of cytotoxicity in immune cells and interferon-γ production [106, 107]. IL12 represents one of the most potent but also toxic cytokines and thus requires reliable safety measures [108].

Other strategies overcome inhibitory signaling with functionally inert ligand or cytokine receptors also referred to as dominant-negative immune receptors. This strategy has increased the potency of CARs through competitive ligand/cytokine binding for PD-1 and TGF-β also in a clinical setting with acceptable toxicity profiles [109, 110] but also PD-1 and CTLA-4 gene disruption has been used to overcome immune checkpoint inhibition [111, 112]. Furthermore, immune checkpoint inhibition of PD-1 and switch receptor signaling via PD-1 to CD28 and TGF-β to IL7 signaling conversion has enhanced the potency and persistence of CAR T cells [113, 114].

According to the safety concerns, the pharmacokinetic features of the adapter molecule can be adjusted by selecting the most suitable adapter molecule format with the intended half-life, e.g., Fab fragment for rapid clearance within hours to days [115], Fc-fusion for medium-term half-life within days and full-size antibody within several days up to weeks [116]. It is worth mentioning that the on-switch mechanism and passive elimination is also dependent on the application route and obviously on the molecular weight of the adapter molecule [35].

Monotargeted CAR therapy is insufficient to control most cancers and antigen loss is a major cause of relapse in CD19- and BCMA-targeted CAR therapy [6, 10]. Besides multitargeting [44, 45], additional effector functions are necessary to overcome roadblocks, such as the hazardous immunosuppressive milieu in and around the cancer, leading to reluctant chemotaxis and CAR trafficking to the tumor site and lack of penetration into the tumor [117]. Moreover, tumors impair antigen-dependent proliferation, induce hypoxia and metabolic deficiency, immune cell exhaustion, and apoptosis [100] not only in solid cancers but also impact on the performance in CD19CAR T therapy [118].

Consequently, the success in CAR therapy relies on multitargeting and the integration of additional effector functions. IL2, IL7, and IL15 are key homeostatic T-cell cytokines [119]. The co-expression of IL15 and the inducible expression of IL7 under NFAT responsive elements have been shown to be safe in clinical trials (ClinicalTrials.gov identifier: NCT03056339, NCT03258047) and to improve the engraftment, proliferation, anticancer function, and persistence of CARs [88, 120]. Due to the existing clinical experience on safety, IL15 and IL7 co-expression should be prioritized in adapter CAR clinical trial efforts. The same accounts for 3rd generation adapter CAR architecture CD28-BB-CD3ζ which was superior to 2nd generation CARs with CD28-CD3ζ and BB-CD3ζ signaling in anti-LLE CAR T (Fig. 5a) and also demonstrated convincing preclinical activity in anti-FITC CARs in various cancer models [34, 121].

New clinical trials with adapter CARs are recommended to enter the clinical stage with a multitargeted approach utilizing well-characterized target antigens, such as CD19, CD20, and CD22 to prove their capabilities for combinatorial targeting. This would represent a risk-averse strategy because adapter CARs are theoretically safer than conventional CARs. The objective would built on the facts that adapter CARs are currently being trialed in patients with relapsed/refractory CD19 positive malignancies (NCT04450069) and conventional CD19- [6], CD20- [122], and CD22-targeted [43] CAR T cells are used in the clinic as well as multitargeted CD19/CD20- and CD19/CD22-targeted CAR T products are used in clinical trials [44, 123]. The concept of “lineage-specific toxicities” in CAR T-cell therapy (Fig. 5b) reduces the risk for unexpected adverse events that can significantly compromise, slow down, or discontinue clinical trials [124]. The antigens CD19, CD20, and CD22 are “clean” B-lineage expressed proteins and have been studied carefully with various therapeutics including antibodies [78], BiTEs [125], antibody drug conjugates [126], and CAR T cells [6, 43, 122]. Once multitargeted adapter CAR therapy has been shown to be safe, the combination with immune checkpoint blockade may additionally improve the clinical outcome like it has in selected patients treated with conventional CD19CAR T [127]. The intermediate step in universal CAR therapy will be the use of patient-individualized adapter molecule combinations to increase the potency and spare unnecessary toxicities.

The future aspiration of transgenic immune cell therapy has no limits. Novel immune receptors will be developed that contain tailored immune functions, with complex bespoke signaling and logic gating, cytokine, and chemokine delivery. The immunomodulatory function of future transgenic immune cells will go beyond oncology and the plain depletion of B cells like it has rescued patients with systemic lupus erythematosus [128].

The exploitation transgenic immune cells in different naturally occurring cell populations and distinct subsets of T cells [6], NK cells [88], iNKT [89], and macrophages [70] will expand the arsenal and therapeutic options. Most likely this will transition into the phase of generating a versatile library of universal allogeneic iPSC-derived effector cells for various indications and applications [129]. There is a chance these artificial immune cells will grow in large tanks, will be cryopreserved (Fig. 5c) and at our disposal at any time in conjunction with off-the-shelf adapter molecule combinations (Fig. 5d). These artificial iPSC-derived effectors will have little in common with their natural counterparts but rather serve as their role model to create the most efficacious next-generation therapeutics. It is a question of scientific art and creativity to combine all possible modifications per individual cancer indication but technically it will be possible to generate these cells. Modifications will be comprised of TcR-like receptors [20], cytokines (e.g., IL12, IL18) [97, 105, 130], and chemokines (e.g., CCL19) [117] as well as their corresponding receptors [131], switch receptors [113], calibrated signaling [16], reverse signaling receptors, logic gating (e.g., SynNotch) [132], site-specific integration of transgenes [25], serial gene disruption of immune checkpoint receptors, and gene regulators [133].

Indirect CAR technologies may become the future of CAR therapies. These include universal adapter CARs [35, 53] and remote switch-on CARs [134‒136]. Indirect CARs are complementary to CD19 and BCMA CARs; however beyond B-lineage cancers, universal adapter CARs are the straightforward solution to the complex challenge of antigen heterogeneity in CAR therapy and the antigen question after all. There is a bright outlook for adapter CAR technologies today and in the future and thus the challenges of adapter molecule development and most importantly to create new flexible concepts for clinical trial design in close collaboration with regulatory institutions need to be taken on. In the era of patient-individualized therapies, this will be the most important achievement. Otherwise, time is against individualized immunotherapy and the dream of versatile futuristic therapies is unlikely to become reality.

During the preparation of this work the authors used ChatGPT 3.5 (OpenAI) to find suitable substitute words (thesaurus function) and overcome non-native language barriers. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

P.S. is an inventor of patents in cancer immunology in the areas of antibody and CAR therapy. L.S. has nothing to disclose. The authors have no ethical conflicts to disclose.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

P.S. conceptualized the manuscript. L.S. and P.S. wrote the manuscript. Both authors reviewed and approved the final version of the manuscript prior to submission.