Both tolDC and Mreg can be generated in vitro starting from CD14⁺ monocytes and share some phenotypic and functional characteristics. Indeed, both tolAPC types express low to intermediate levels of T-cell costimulatory molecules, and secrete low amounts of pro-inflammatory cytokines, indicative of a partially matured APC. Similarly, immature DC (iDC) display minimal expression of costimulatory molecules and little secretion of inflammatory cytokines, demonstrating potential optimal requirements for tolerance induction in vivo (23, 24). However, iDC are unstable and may differentiate into immunogenic DC under inflammatory conditions (25, 26). This invalidates their putative use as therapeutic products for tolerance induction. Therefore, different strategies to generate stable tolAPC have been explored, including treatment with pharmacological agents or cocktails of immunomodulatory cytokines, genetic engineering, and exposure to apoptotic cells (9, 27, 28). Most of these in vitro conditioning regimens aim at stabilizing a semi-mature state of tolDC, maintaining the capacity to induce immune hyporesponsiveness of T cells, even in presence of powerful pro-inflammatory signals.

Importantly, tolAPC inhibit T cell proliferation, albeit through different immunosuppressive mechanisms depending on the approach used to generate tolAPC in vitro. Induction of peripheral T cell anergy and apoptosis (29), attenuation of effector and memory T cell responses and the generation and activation of Treg populations (30, 31) result in part from presentation of low levels of antigen in the absence of costimulation; these are typical mechanisms attributed to a variety of tolerogenic subtypes (10, 32, 33). Additionally, tolerogenic DC may express various inhibitory receptors such as programmed death-ligand (PD-L)1, PD-L2 (34), immunoglobulin-like transcripts (ILT) (35), FasL (36, 37), and TRAIL (38). Secretion of anti-inflammatory cytokines such as IL-10 (39, 40) and TGF-β (41, 42), as well as reduced expression of pro-inflammatory cytokines, also may contribute to tolerance induction. A study comparing tolDC generated in presence of dexamethasone and rapamycin demonstrated that while both tolDC subsets were able to impair T cell proliferation, rapamycin-treated tolDC have a mature phenotype and are not able to produce IL-10 upon stimulation with LPS, as opposed to vitamin D₃- or dexamethasone-treated tolDC (27). Whereas, it was demonstrated that rapamycin-treated tolDC induce Treg (27), vitD₃-treated tolDC induce T-cell hyporesponsiveness and antigen-specific Treg (7, 43). Moreover, DC-10 induce Tr1 cells (44), while autologous tolDC have a weak capacity to stimulate allogeneic T cells and suppress T-cell proliferation and IFN-γ production (45). Mreg have been shown to convert allogeneic CD4⁺ T cells to IL-10-producing, TIGIT⁺, FoxP3⁺-induced Treg (46). Variations in the process to generate tolAPC may initiate regulation through distinctive mechanisms, making it difficult to compare these different types of tolAPC. Therefore, efforts have been made to find common features unique for tolerance-inducing cells (47). For example, since tolDC conditioned using vitamin D₃ and dexamethasone exhibit high cell surface expression of TLR2 (48) or CD52 (49), such markers might be considered to assess the quality and stability of tolDC in future cell-based clinical trial protocols. In addition, it was demonstrated that the expression of single immunoglobulin IL-1-related receptor has a role in maintaining low levels of costimulatory molecules and in regulation of Treg expansion (50). Others demonstrated that C-lectin receptor CLEC-2 upregulation by DC is associated with Treg induction (51). So far, however, gene expression studies comparing different tolDC and Mreg protocols have not been able to identify common biomarkers of tolerance induction [reviewed by (52, 53) and (54)].

The difficulty in comparing characteristics of different clinical tolAPC suggests the need for a uniform set of metrics for their description, including full characterization of (at least) the immune phenotype and their functional activity (potency). Hence, a better identification of the characteristics that identify the tolerance-inducing properties of tolAPC, irrespective of the conditioning regimen, would be valuable for safe cell therapy delivery into patients. Joint efforts in translating tolAPC into the clinic by harmonizing protocols and defining functional quality parameters have been initiated (1, 55). A Minimum Information Model on Antigen-presenting cells (MITAP) has been defined to harmonize reporting on tolAPC therapy to ultimately allow the uncovering of commonalities between tolAPC and to define common quality control biomarkers and potency assays for the various tolAPC products for clinical use (55). Likewise, using similar immunomonitoring protocols in different clinical trials could help to better understand the in vivo mechanism of action of these cells (56).

Targeted regulation of antigen-specific T cell responses would avoid generalized immunosuppression as observed with immunosuppressive drugs and monoclonal antibodies currently in use in the clinics and may thus overcome occurrence of impaired immune-surveillance leading to infections or development of malignancies. Ex vivo generated tolAPC have the potential to therapeutically induce, enhance, or restore antigen-specific tolerance. Indeed, after loading these cells with exogenous or endogenous antigens, one major advantage is their capability to act in an antigen-specific manner.

A number of in vivo studies demonstrate that antigen loading of tolAPC is indispensable to reach efficient clinical responsiveness following tolAPC therapy. For instance, a beneficial effect of vitamin D₃-tolDC loaded with MOG₄₀₋₅₅ peptide was demonstrated in experimental autoimmune encephalomyelitis (EAE), whereas no clear beneficial effect on the clinical score of EAE mice was found when mice were treated with vitamin D₃- tolDC not loaded with myelin peptides (57, 58). Similar findings have been demonstrated in other animal models of autoimmune diseases, including collagen-induced arthritis and autoimmune thyroiditis (59–61). Altogether, these findings suggest that selection of the target self-antigen is critical for disease-specific tolerance induction in vivo. Suitable disease-associated self-antigens responsible for T cell priming have been identified for T1D and multiple sclerosis (MS). However, this is not the case for other autoimmune diseases, such as rheumatoid arthritis or Crohn's disease, for which specific disease-associated antigens are unknown or not tissue specific. Moreover, not all patients uniformly display the same set of self-antigens for a given disease. MS, for example, is associated with a range of self-antigens and auto-antibodies that are differentially expressed among patients and at different points during the disease (62). While the targetable autoreactive T cell populations may be limited to a select number of antigens at the onset of clinical disease, other “late antigen” or spread epitope autoreactive T cell populations may drive autoimmunity during progression of the disease. To overcome this hurdle, some groups have chosen to load the tolerance-inducing therapeutic cells with a broad pool of distinct, candidate disease-related peptides (63–65) (NCT02283671, NCT02618902, and NCT02903537).

In contrast to autoimmunity, transplant rejection is mediated by an undesired immune response against epitopes that differ between the transplanted donor graft and the recipient host, so-called allorecognition (66). Specifically, recipient T cells may initiate a strong immune response leading to transplant rejection in the absence of adequate immunosuppression. To avoid transplant rejection, the induction of tolerance to donor-specific antigens has been coined as a therapeutic target for decades. For this, both donor tolAPC and recipient tolAPC loaded with donor-specific antigens are being considered for development of cell-based immunotherapeutic protocols in the transplantation setting (67). However, whereas the clinical use of donor-derived tolAPC is only feasible in the context of living donor transplantation and entails a risk of sensitization (including development of allo-antibodies), the use of autologous, i.e., recipient-derived, tolAPC is a less risky approach to begin with. Indeed, the use of recipient autologous tolAPC is likely to be more feasible than that of donor-derived tolAPC, since cell products can be prepared from the peripheral blood of the recipient before transplantation, stored while the patient is on the waiting list, and loaded with donor-derived antigens (such as HLA peptides or donor cell lysates) at the time of transplantation. In the context of the ONE study, two trials using tolDC and Mreg are being performed in living-donor kidney transplant recipients (Table 1).

Although it is generally accepted that the route of administration is important for optimal tolAPC effector function, the best route of tolAPC administration is not known. To date, a variety of routes of administration have been used (see Table 1), including intradermal, intraperitoneal (19), intravenous and intra-articular (17). Different routes of administration may be required to allow tolDC to reach the relevant draining lymphoid tissue for T cell encounter or to end up at the site of inflammation. Especially since tolDC demonstrate a reduced expression of CCR7 and consequently a reduced (but not absent) ability to migrate in response to the CCR7 ligand CCL19 (68), the capacity of tolDC to reach the lymph nodes is a critical concern. While the migration of DC toward the lymph nodes increases following intradermal as compared to subcutaneous administration, only 2–4% of DC migrate to the draining lymph nodes after intradermal administration, but the situation may be different in patients with autoimmune diseases where monocyte-derived DC from MS patients have shown a significantly higher proportion of CCR7-expressing cells compared to healthy controls (69). Given these observations, and that in the setting of cancer vaccine development, DC injected into a lymphatic vessel showed a prolonged half-life as compared to DC injected intravenously (70, 71), direct intranodal injection of tolDC is being evaluated in a clinical setting (see Table 1).

As an alternative to lymph node targeting of tolAPC, tolAPC may also be introduced directly into the site of inflammation. Indeed, injection into the affected disease site (i.e., an inflamed joint) where the tolAPC could suppress auto-reactive effector T cell responses is logical. In this context, intra-articular injection of tolDC differentiated using dexamethasone and vitamin D₃ and loaded with autologous synovial fluid in patients with rheumatoid arthritis was demonstrated to be safe and feasible. Hence, despite the fact that tolDC were directly injected at the site of inflammation, no adverse events were observed in most patients and hypertrophy, vascularity and synovitis were stable in all treated cohorts. Moreover, two patients receiving 10 million cells showed a decrease in synovitis score (17). Similarly, a phase I randomized clinical study currently evaluates the safety and efficacy of tolDC injected into the intestinal lesions in patients with refractory Crohn's disease (Table 1). In some conditions such as T1D, direct injection of cells in the inflammatory site, e.g., pancreas, might not be possible and require tolDC administration adjacent to the inflammation site. For the treatment of inflammatory diseases of the brain, the blood–brain barrier (BBB) may represent a major hurdle. Considering this potential problem, it was demonstrated that enhancing CCR5 expression in tolDC using mRNA electroporation endowed these cells with CCR5-driven migratory capacity. This enabled the cells to migrate to inflammatory sites, even when it required crossing of functional barriers such as the BBB (72). Similarly, introducing CCR7 expression in tolDC using the proposed approach of chemokine receptor mRNA electroporation could overcome the limited lymphoid homing capacity of tolDC. Indeed, DC transduced with lentiviral vectors coding for CCR7 and IL-10 genes were able to migrate to the lymph nodes and spleen, prolonging cardiac allograft survival in mice (73). However, there are still many unknowns and there is a clear clinical need to characterize the pharmacodynamics of tolDC in humans and relate this to clinical efficacy. Advances in cell imaging techniques, for example magnetic resonance imaging of 19F-labeled cells, have made it possible to address this question in future studies.

Several protocols to generate and administer tolAPC have been tested in phase I clinical trials with highly encouraging results from a safety point of view and in terms of adverse effects (Table 1). Further phase I/II studies are under way in Crohn's disease, T1D, rheumatoid arthritis, MS and kidney transplantation (Table 1). However, for the success of future tolAPC trials, there is great need to define the optimal vaccination protocol; to ensure optimal in vivo-acting of the tolAPC, future trials may require changes in administration route, dose or could demand repeated tolAPC administration. Furthermore, the identification of a common set of tolerogenic markers would enable optimized comparison of tolAPC products and their tolerance-inducing potential and provide an improved understanding of how these cells modulate the T cell response both locally and systemically. It would be of great help to analyze critical pathways contributing to programming and function of tolAPC. Ultimately, this may set the stage for new approaches improving the therapeutic potential of tolAPC for the future.

While the application of Treg for the treatment of autoimmune diseases is currently under intense investigation, Treg were first used in the clinic to treat patients with graft vs. host disease (GvHD) after hematopoietic stem cell transplantation (HSCT) (88) (Table 2). Results from the clinical trials in GvHD with polyclonal expanded Treg have suggested that altogether these cells are safe, but there is some concern about the occurrence of mild to moderate infections, and it still is unclear whether Treg treatment could promote cancer (92, 94). The latter problem has been reported in only one trial to date, however it was concluded that the tumor was present before the therapy with Treg was applied (94). The safety and feasibility of adoptive transfer of ex vivo expanded Treg was further confirmed in T1D patients (2), which has driven the application of Treg therapy to clinical trials in other autoimmune conditions such as MS, autoimmune hepatitis, systemic lupus erythematosus, Crohn's disease, and autoimmune uveitis (102) (Table 2). Another clinical trial was recently published where polyclonal Treg were injected into T1D patients; results from this trial confirm the safety of this type of therapy and also show for the first time, by deuterium labeling of the Treg, that some of the injected Treg can be detected for up to 1 year after infusion (103).

Treg therapy is now being applied as a “tolerogenic” therapy to reduce dependency on immunosuppressant drugs in patients receiving solid organ transplants. The idea behind this strategy is very similar to the application of Treg in autoimmune diseases, namely to tilt the balance toward Treg dominance over rejection-causing Teff cells. The first reports using adoptive transfer of Treg in kidney transplant patients have been recently published demonstrating the safety of this strategy in the context of solid organ transplantation (103, 104). Recently, clinical trials are being completed using different variations of Treg products (The ONE Study and ThRIL) (Table 2). The ONE Study includes Phase I clinical trials comparing the safety of different types of regulatory cells, including polyclonal and donor-reactive Treg in patients receiving kidney transplants (www.onestudy.org) (20). The ThRIL trial is a Phase I/IIa dose-escalation clinical trial in the setting of liver transplantation. Results from the ThRIL and the various ONE Study trials are currently being prepared for publication. The impact of Treg on the recipient immune system will be revealed only when the very detailed immunomonitoring is completed, which is a major objective of the described clinical trials (105–107).

Altogether, from the outcomes of the completed clinical trials so far, it can be concluded that adoptive transfer of Treg is safe and technically feasible (Table 2). Therefore, increasing efforts are currently focusing on clinical trials to test their therapeutic efficacy. Importantly, several lessons have been learned from recent experiences with Treg to improve future trial designs. For example, the clinical state of the patients has been shown to influence the function and properties of Treg, and therefore can condition the success of ex vivo cell product expansion (108). Furthermore, the specific expansion protocol can affect Treg function and specificity, and can improve tissue targeting and suppression capacity (108). Finally, the immune modulatory therapies received by patients at the time of Treg adoptive transfer can positively or negatively impact therapeutic outcome (109, 110).

Treg therapies are currently undergoing intense testing. An interest in therapeutic Treg preparations has resulted in several ongoing clinical trials in the transplantation setting, in autoimmune diseases, but now also in conditions such as beta thalassemia major and amyotrophic lateral sclerosis (101) (Table 2). We await new testing in the setting of hypersensitivity and cardiovascular disease (102), and anticipate that Treg therapy will be considered for any condition where there is evidence for an immune regulation imbalance.

Since cell-based therapies are becoming more common, it is important to reliably monitor the immune system for both desired and undesired immunological effects. Rigorous immunomonitoring will therefore provide information about the safety of these treatments, ideally at early time points after CTT administration. In addition it will give insight in the therapy-related mechanisms of tolerance-induction and maintenance and may aid in patient-tailoring of therapy. To accurately measure the effects, especially across different trials, it will require introduction of harmonized and validated immunomonitoring assays.

There are a number of possible assays that can determine cell therapy effects in humans. For instance, measurement of circulating cytokines, C-reactive protein or changes in antibody titer can determine immune status. Similarly, measuring CD4⁺ T cell responses after viral-antigen stimulation is possible by flow cytometry via CD40L expression or cytokine production in a functional assay; these assays could potentially identify non-specific immunosuppressive effects (133). Although the completed clinical trials have shown so far that cell-based tolerogenic therapies are safe and do not cause serious undesirable immune responses (96, 99, 103, 104), the extent to which these treatments achieve therapeutic efficacy remains largely undetermined. Current cell-based tolerogenic trials in autoimmunity and transplantation include clinical outcome measures such as C-peptide response, insulin consumption or reduction of immunosuppressive doses. However, clinical endpoints may not necessary reflect the efficacy of cell-based therapies, since the tolerogenic effect of the transferred cells may not directly lead to immediate changes in systemic parameters such as inflammation, potentially underestimating a longer term effect. It is therefore important that future clinical trials incorporate suitable monitoring methods to assess the immunomodulatory effects of cellular therapies.

To assess therapeutic effectiveness, different methods have been proposed to identify tolerogenic responses. The assessment of in vitro autoreactive or donor-specific T cells responses prior to and after treatment could provide a precise evaluation of therapeutic efficacy. Antigen-specific assays allow discrimination between targeted tolerance to the induction of general immune suppression and loss of responses to pathogens. In addition, these methods provide an efficacy readout for antigen-specific therapies such as tolerogenic APC loaded with antigens or the generation of donor-specific Treg. The identification of antigen-specific immune responses by ELISPOT (134) or through flow cytometry detection of CD40L upregulation (135) have shown promising results in predicting kidney and liver allograft rejection, suggesting potential applicability for the evaluation of tolerogenic therapies. Monitoring targets will be different depending on the main immune population involved in the disease (e.g., CD4, CD8 T cells or antibodies). Unfortunately, the antigens mediating the immune responses in autoimmunity are not always available or identified (as discussed in paragraph Antigen Specificity of tolAPC-Based Immunomodulation); HLA antigens in the case of transplantation are known and can be used, or stimulation with donor or donor-matched cells is possible. Though less specific, identification of phenotypic or functional changes by flow cytometry on the total pool of cells targeted (e.g., Treg) by the tolerogenic treatment may reveal therapeutic effectiveness. Indeed, the acquisition of tolerance in animal models and in the clinic is associated with an increased number of regulatory cells and decreased pro-inflammatory function of innate and adaptive immune cells (5, 136, 137). Therefore, flow cytometry analysis to delineate the distribution and activation status of different cell types, or in vitro assays to evaluate the suppressive and inflammatory function of circulating cells can provide a non-specific approach to assess the development of tolerogenic properties (138). Other non-specific strategies such as gene expression profiling of circulating immune cells or tissue biopsies can add to the functional assessment of immune responses. Furthermore, there is a growing body of work focusing on the validation of transcriptional signatures to predict transplantation tolerance in liver and kidney transplant recipients (139–141), which may prove to be a valuable tool in assessing the efficacy of tolerogenic therapies.

The evaluation of homeostatic characteristics of infused cells such as survival, stability or tissue migration, constitutes another monitoring objective to assess therapeutic efficacy. Being able to track infused cells will help to determine the best site for administration/application of tolerogenic cell products. While simple phenotypic detection (flow cytometry) of the transferred cell subset after treatment suggests the presence of infused cells, it does not distinguish transferred from endogenous cells. Therefore, current techniques for tracking infused cells depend on direct cell labeling strategies, such as indium labeling or deuterium introduction during ex vivo expansion, with subsequent isotope detection in the different tissues or compartments (99, 103, 142). Unfortunately, this method is only semi-quantitative, since individual cells are not detected. Individual cells can be labeled with rare earth metals and detected with precision in vivo using laser ablation-inductively coupled plasma-mass spectrometry, but so far this has only been tested in mouse models (143, 144). Other emerging therapeutic approaches such as CAR-Treg could take advantage of genetic modifications to adapt reported gene imaging strategies to detect the transferred cells by non-invasive methods (e.g., MRI, PET, SPECT) (145, 146). Additionally, the use of T cell receptor (TCR) engineered Treg or ex vivo expanded antigen-specific Treg create the opportunity to track infused cells by predefining the TCR clones and identifying them among the T cell compartment repertoire through TCR sequencing analysis (147).

Current tolerogenic cell-based trials are highly heterogeneous, comprising different cell types with variable manufacturing approaches, and targeting various autoimmune diseases and transplantation settings. Furthermore, by their very nature such trials tend to involve low numbers of patients, often participating in different centers or countries. It is therefore essential to establish common immunomonitoring strategies in the research community to achieve robust and reliable data which can be compared and combined between trials. First, not all the immunomonitoring methods are similarly standardisable, and second, not all the centers have the technical expertise or infrastructure to perform certain assays. Sample collection and storage of whole blood or tissue biopsies for transcriptional analysis does not involve extensive sample manipulation, making standardization of this method achievable. On the contrary, while flow cytometry analysis of circulating blood is an accessible technology, instrumentation must be calibrated appropriately to accurately define cell subsets, and analysis of the data needs to be strictly regulated. Although centralized phenotypic analysis in multicenter trials is feasible (105), this involves significant logistical challenges, including the decision to either test fresh or frozen samples; frozen samples sacrifice accuracy due to loss of certain cell populations during separation and freezing procedures. In general, flow cytometry standardization requires extensive cooperation between centers and precise planning. To further harmonize flow cytometry data implementation of automated gating approaches will be of outmost importance in the future (148–150). Finally, functional in vitro assays also represent a challenging method to standardize, involving different approaches depending of the cell type and function. Nevertheless, several efforts have been made to establish a minimal harmonization of antigen-specific functional assays (56). While these assays are likely to be the most informative in the assessment of therapeutic efficacy, it is unlikely that current assay results will be directly comparable between independent trials. Therefore, the inclusion of adequate control cohorts and reference groups in the assays, considering the specific cell therapy approaches and disease characteristics, remains an objective to achieve feasible comparisons (107).

PT Medical University of Gdańsk–owns a patent related to presented content. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.