The first approach involves the induction of Tregs in vitro. Indeed, under certain circumstances and factors such as TGF-β and IL-2, naïve CD4⁺ T cells can express Foxp3 and develop a regulatory phenotype (Zheng et al., 2002; Chen et al., 2003). This approach was first tested in HCT recipients in a phase I safety study (MacMillan et al., 2021). Human CD4⁺CD25⁻ conventional T cells were cultured with IL-2, rapamycin and TGF- β with anti-CD3, and artificial antigen-presenting cells to generate induced Tregs, which were then infused into HCT from HLA-identical sibling donors. However, a major concern with induced Tregs is the potential for instability, particularly under inflammatory conditions (Koenecke et al., 2009; Beres et al., 2011). Although patients from this first study did not show an increase in GVHD after infusion, alternative approaches are being developed.

Tregs do not produce IL-2 themselves, but express the high affinity IL-2 receptor CD25. Therefore, the administration of low doses of IL-2 can selectively expand Tregs without conventional T cell activation and expansion (reviewed in (Ye et al., 2018)). However, IL-2 can also activate conventional T cells and NK cells. Consequently, numerous strategies have been developed to selectively increase Tregs and prevent adverse effects, such as Il-2 muteins with greater Treg specificity (Khoryati et al., 2020), anti-IL-2 monoclonal antibodies, or IL-2/anti-IL-2 complexes to modify binding to the IL-2 receptor (Boyman et al., 2006; Polhill et al., 2012; Trotta et al., 2018). Other strategies aim to combine IL-2 with cytokine pathway blockade, such as an anti-IL-6 receptor (Ye et al., 2018), or to modify IL-2 to prolong its half-life, allowing for the use of an ultra-low doses (Bell et al., 2015).

The first approach to generate Tregs from conventional cells is to overexpress Foxp3 (Figure 1). Indeed, Foxp3 is a master transcription factor for Treg development and function. Sakaguchi et al. pioneered this approach, inducing Foxp3 expression in CD25⁻, CD45RO⁻ CD4⁺ lymphocytes by retroviral transduction. The transduced cells exhibited a regulatory phenotype similar to CD25⁺ CD4⁺ Tregs, demonstrated in vitro suppressive capabilities, decreased proliferation, cytokine production reduction, and heightened expression of Treg-associated molecules, such as CD25 and CTLA4 (Yagi et al., 2004). While this initial study offers valuable and promising insights into the involvement of Foxp3 in Treg development, it solely showed that the transduced cells displayed a Treg-like phenotype. However, they also exhibited unconventional Treg features, including reduced production of IL-10 (Yagi et al., 2004). Subsequently, Allan et al. developed a third-generation lentivirus-based strategy to ectopically express high levels of Foxp3 under the control of the human elongation factor 1α promoter, which does not fluctuate with T cell activation status (Allan et al., 2008). This method facilitated the development of suppressor cells from both naïve and memory CD4⁺ T cells. Nevertheless, despite their demonstrated in vitro suppressive capacity, the expected significant production of IL-10 and TGF-β was not observed, suggesting that Foxp3 alone might not be sufficient to induce Treg production (Allan et al., 2008).

Bacchetta et al. also tested a third-generation lentivirus-mediated FOXP3 gene transfer in conventional CD4⁺ T cells, confirming its in vitro suppressive capacity. Notably, they were the first to test the suppressive ability in vivo, demonstrating its effectiveness in a xenogeneic graft-versus-host disease model (xGvHD) (Passerini et al., 2013). However, they did not evaluate IL-10 and TGF-β production in their cells, a notable omission given previous observations. While IL-10 and TGF-β production may not be essential for mediating in vitro suppressive function (Allan et al., 2008), their role in the xGvHD model remains unexplored. Finally, Honaker et al. used gene editing based on homology-directed repair (HDR) to induce Foxp3 expression (Honaker et al., 2020). The targeted insertion of a robust enhancer-promoter proximal to the first coding exon bypassed epigenetic silencing and ensured stable and strong expression of endogenous Foxp3. The HDR-edited T cells expressed canonical Treg markers and exhibited suppressive capabilities both in vitro and in vivo (Honaker et al., 2020). However, these cells exhibited lower suppressive potential than normal Tregs and displayed transcriptomic differences from thymic Tregs, such as IKZF2 (Honaker et al., 2020). These observations collectively suggest that Foxp3 expression alone is insufficient to fully recapitulate the Treg function. Given that the transcription factor Helios is implicated in Treg differentiation, stability and suppression, researchers explored the possibility of co-expressing Foxp3 and Helios in conventional cells. However, this co-expression led to significantly reduced cell proliferation and survival, limiting its clinical application (Seng et al., 2020).

Foxp3 overexpression in conventional CD4⁺ T cells has also been combined with CAR expression (Fransson et al., 2012; Martin et al., 2021). This combination enabled the expression of a Treg phenotype, resulting in a reduced IL-2 and TNF-ɑ secretion and increased in IL-10 secretion. Additionally, these cells were shown to be effective in an xGvHD model (Martin et al., 2021) and an autoimmune encephalomyelitis mouse model (Fransson et al., 2012). While this approach is promising, further validation is required before clinical application can be considered.

Strategies aiming at increasing Foxp3 expression may not only serve to convert conventional T cells into Tregs, but also to increase Treg stability. Indeed, Henschel et al. developed a specialized CAR vector incorporating Foxp3 into CAR-Tregs (Henschel et al., 2023). They observed significant improvements in both the safety and efficacy of the resulting CAR-Treg product. Notably, in a challenging microenvironment characterized by pro-inflammatory conditions and limited IL-2 availability, Foxp3-CAR-Tregs exhibited sustained Foxp3 expression, surpassing the expression levels observed in control CAR-Tregs. Furthermore, the introduction of additional exogenous Foxp3 expression did not induce any discernible phenotypic alterations or functional impairments, such as cellular exhaustion, loss of Treg characteristics, or aberrant cytokine secretion (Henschel et al., 2023). Overall, these advancements offer potential pathways for the development of more accessible and safe immunotherapies targeting limited Treg numbers. However, further optimization is needed to generate Tregs from conventional CD4⁺ T cells.

Another method to induce Tregs from conventional CD4⁺ T cells involves the use of epigenetic regulators (Figure 1). Histone/protein deacetylase (HDAC)s can regulate chromatin remodeling, gene expression, and transcription factor expression by removing acetyl groups from histones. This deacetylation process by HDACs typically results in chromatin condensation, restricting access by the transcription machinery and consequently repressing gene expression. Additionally, HDACs function as post-transcriptional modifiers that regulate protein acetylation (reviewed in (Kim and Bae, 2011)). Indeed, HDAC inhibition therapy can induce Foxp3 expression. This therapy increases histone acetylation, which leads to the relaxation of the chromatin structure surrounding the FOXP3 gene. This chromatin decondensation facilitates easier access for transcription factors to interact with the FOXP3 gene, thereby enhancing its transcription and subsequent protein synthesis. Consequently, cells become more proficient at suppressing immune responses and promoting transplant tolerance (Tao et al., 2007; Lucas et al., 2009). There are 18 known mammalian HDACs divided into four classes sometimes with subclasses (I, IIa, IIb, III, IV) and the use of pan-HDAC inhibitors may have various side effects. Therefore, not all HDAC inhibitors promote Treg functions (Wang et al., 2009). Pharmacological targeting and deletion experiments have shown a positive impact on Tregs for HDAC2, HDAC7, HDAC9, HDAC6, HDAC10, HDAC11 and Sirtuin-1. HDAC6, in particular, stands out as the most promising candidate due to the availability of a specific inhibitor and the fact that HDAC6 knockout mice exhibit normal health and behaviour. Hence, the HDAC6 isoform is the primary target for enhancing the immunosuppressive activity of Tregs (de Zoeten et al., 2011; Kalin et al., 2012; Kurohara et al., 2021; Lee et al., 2022). However, HDAC10 could also be of interest, as Wayne Hancock’s team demonstrated an increased suppressive ability of HDAC10−/− mouse Tregs, both in vitro and in vivo (Dahiya et al., 2020). There is a lot of potential for HDAC inhibitors to induce Tregs in vivo (reviewed in (Wang et al., 2018)). However, their use in adoptive immunotherapies and Treg engineering has not yet been explored.

Three conserved non-coding DNA sequence (CNS) elements located at the FOXP3 locus have significant implications for the fate of Tregs. Specifically, CNS1 serves as a TGF-β-sensitive enhancer element crucial for induced Treg generation, CNS2 plays a critical role in stabilizing Foxp3 expression within dividing Treg cells, and CSN3 can modulate Treg frequency in the thymus and periphery but becomes dispensable once Foxp3 is expressed (Zheng et al., 2010). Notably, CSN2, also known as Treg-specific demethylation region (TSDR), exhibits demethylation in stable Treg populations (Feng et al., 2014), in contrast to T cells with transient Foxp3 expression (Floess et al., 2007; Polansky et al., 2008). Recent scientific advances have enabled the selective de-methylation of the TSDR within the endogenous chromatin environment. This epigenetic editing resulted in an increased Foxp3 expression in Jurkat cells (Wilk et al., 2022) and human T cells (Kressler et al., 2020). However, this increase was not associated with phenotypic Treg induction (Kressler et al., 2020). Selective demethylation of the TSDR represents a significant scientific breakthrough. However, it requires further investigation and may need to be combined with other strategies to achieve the desired outcomes.

One strategy to expand the diversity of recognized antigens involves the use of bispecific TanCARs, capable of targeting two different antigens (Figure 2). Grada et al. developed this novel receptor and demonstrated that TanCARs retain the cytolytic capacity of T cells, even when one of the target molecules is lost. This approach also results in a more controlled response against tumors by recognizing both targets in an animal disease model (Grada et al., 2013). Other studies have adopted a similar strategy, yielding comparable results (Jensen and Riddell, 2014; Zah et al., 2016; Jia et al., 2019). However, this approach has not yet been applied to Tregs.

The application of CAR-Tregs and CAR-T cells is limited by wide range of target antigens dependent on the patient’s condition and by the currently limited catalog of available CAR receptors. To address this challenge, universal CAR platforms (UCAR) were developed (Figure 2). In summary, T cells are engineered with a signaling module that binds to a specific epitope on a switching molecule. The switching module includes an antigen-binding domain and a switching epitope that is recognized by the signaling module (Figure 2). Crosslinking with target cells is thus mediated by these independent antigen-specific switching modules that contain the peptide epitope recognized by the UCAR. These adapters can be easily exchanged to target different tissue antigens, Many UCARs have been engineered and tested in conventional T cells and are reviewed in (Liu et al., 2019). To the best of our knowledge, two groups have tested them in Tregs. Meyer et al. tested a CAR that binds antibodies conjugated in the Fc region with FITC. They showed that these CAR-Tregs could prolong mice pancreatic islet allografts and promote tolerance to secondary skin grafts (Batista et al., 2017). Bachmann et al. tested a UCAR that recognizes a sequence of 10 amino acids (5B9 tag) of the nuclear protein LA/SS-B (Koristka et al., 2014; Koristka et al., 2018; Kegler et al., 2019), and the transduction procedure itself did not affect the phenotype and activity of the Tregs (Koristka et al., 2014; Koristka et al., 2018). The advantages of UCARs are their high versatility and easy reprogrammability. They could also be safer, as they do not have an effect without the infusion of the switching module. However, as they rely on the infusion of exogenous sequences or epitopes, they require more regulatory examination for safety and efficacy.

The BAR strategy involves replacing the scFv of the CAR with the antigen itself to regulate B-cells and plasma cells that are antigen-specific (Figure 2). It was initially used by Ellebrecht et al., in cytotoxic T cells, to destroy autoantigen-specific hybridoma in a pemphigus vulgaris model (Ellebrecht et al., 2016). This concept was further tested using immunodominant FVIII domains (A2 and C2) to destroy anti-FVIII hybridomas, which are problematic in hemophilia patients (Baaten et al., 2018). The FVIII A2 and C2 domains of BAR-CD8 T cells were assessed in vitro (human) and in vivo (mouse). The same team then tested their FVIII BAR in Tregs and demonstrated their ability to suppress the memory antibody response in spleen cultures from FVIII-immunized mice in vitro and completely prevented the development of anti-FVIII antibodies following FVIII immunization (Zhang et al., 2018). The use of BAR-Tregs also demonstrated promising potential in providing clinical protection against severe allergic reactions. In their study, Abdeladhim et al. developed a BAR using ovalbumin (OVA) as the targeted protein (Abdeladhim et al., 2019) and observed a reduction in hypothermia (an indicator of anaphylaxis) following OVA exposure. However, in the in vivo setting, no discernible reduction in the circulating levels of OVA-specific IgE was detected (Abdeladhim et al., 2019). Nonetheless, these results have opened the door to the use of BAR-T cells and BAR-Tregs in autoimmune diseases and transplantation.

To address the issue of hyperactivity associated with CAR-T cells and the subsequent toxicity concerns, researchers have developed SynNotch receptors incorporating a two-step positive feedback circuit (Figure 2). This circuit comprises a priming signaling receptor devoid of activating-signaling capacity. Upon antigen recognition, these receptors trigger a trans-membrane cleavage, akin to a notch receptor, releasing a transcription factor. This transcription factor then initiates the expression of a CAR targeting another antigen (Figure 2). Both antigens must be present to activate the cell. This technology has been extensively tested in various cancer models using conventional T cells (reviewed in (Stock et al., 2023)) and it holds potential to enhance the effectiveness and safety of CAR-T cell therapies. Thus far it has not been explored in Tregs, but it could prove valuable for targeting combinations of antigens when one or both are widely distributed throughout the body.

Typically, CARs consist of four domains: an extracellular antigen-binding domain responsible for recognizing the target antigen, a spacer or hinge region that provides the necessary receptor length, a transmembrane domain anchoring the receptor to the plasma membrane, and an intracellular signaling domain that transmits both the first and second co-stimulatory signals (Bridgeman et al., 2010) (Figure 2). Interestingly, different intracellular co-stimulatory domains in second-generation CAR-T cells exert distinct effects on cell differentiation, metabolism, and function. For Tregs, the CD28 co-stimulation domain appears to be the most effective in maintaining Treg function. Initially, Maus et al. compared the CD28 co-stimulation domain to 4-1BB, and only CD28 CAR-Tregs demonstrated efficacy in vivo (Boroughs et al., 2019). Subsequently, Dawson et al. designed a total of 10 anti-HLA-A2 CAR receptors differing in their co-receptor signaling domains. In an xGvHD model, CARs encoding TNFR family co-receptors (OX40, GITR, 4-1BB, and TNFR2) were ineffective in promoting CAR-Tregs function, while CD28-CAR-Tregs exhibited highest effectiveness. PD-1-CAR Tregs displayed ineffective activation, and TNFR2 CARs resulted in a marked increase in TSDR methylation (Dawson et al., 2020). These results were further corroborated by Zuber et al., confirming the negative impact of 4-1BB signaling in Tregs (Lamarthée et al., 2021). These findings offer valuable insights for the future design and optimization of CAR-Treg therapies.

The genetic engineering of CAR-T cells has paved the way for a wide range of innovative approaches aimed at enhancing their efficacy and anti-tumoral function. One particularly intriguing strategy involves the incorporation of “armour” proteins through modifications in CAR genes. These proteins, whether secreted or expressed on the cell surface, possess the potential to augment T cell activity and positively influence the tumor microenvironment (Figure 2). By finely tuning each component of CAR genetic constructs, precise control can be exerted over the functionality, capacity, and safety of the resulting CAR-T cells.

An example of such advancement can be found in Cherkassky et al.‘s study, which developed anti-mesothelin CAR-T cells transduced with a truncated dominant-negative PD-1 receptor. This receptor, devoid of intracellular signaling domains, binds to PD-L1 but fails to transmit inhibitory signals. Consequently, it enables the CAR-T cells to resist PD-L1-induced exhaustion, thereby prolonging survival in xenograft models of pleural mesotheliomas (Cherkassky et al., 2016). Another notable study by Yeku et al. focused on the design of CAR-T cells capable of constitutive secretion of IL-12. In a model of ovarian peritoneal carcinomatosis, these CAR-T cells exhibited enhanced expansion and greater anti-tumoral efficacy (Yeku et al., 2017).

In the field of adoptive immunotherapy, the use of armour proteins holds great potential for enhancing the suppressive and survival capabilities of CAR-Tregs. Employing secreted armour proteins such as IL-2 could potentially improve proliferation and survival, while TGF-β or IL-10 (Mohseni et al., 2021) could foster a more immunosuppressive environment. However, it should be noted that IL-10 overexpression alone may not suffice to maintain tolerance if cells lose other immunosuppressive abilities (Rana et al., 2021). The co-expression of membrane molecules like Pd-1 and Ctla-4 with the CAR could enhance contact-mediated inhibitory capacities. Furthermore, the expression of transgenic receptors can also be combined with small-molecule therapeutics to promote efficacy and/or survival. For example, the incorporation of an orthogonal IL-2 receptor, selectively responsive to orthogonal IL-2 injection yet retaining retains native IL-2 signalling through STAT5, can facilitate the selective expansion of the transduced Tregs (Sasaki et al., 2014; Hirai et al., 2021). Given these significant advancements, CAR-Tregs designed with armour proteins emerge as a promising immunotherapy modality applicable to various disorders.