The loss of immune tolerance in RA related to the impaired clearance and excessive presence neutrophil extracellular traps (NETs) has been previously reviewed (47–49). When NET removal is compromised, they accumulate at inflammatory sites, thereby prolonging inflammation and producing new autoantigens (47). Elevated NET formation has been observed in the sputum of both individuals at risk for developing RA (being first degree relatives of RA patients) and in RA patients themselves (50, 51). This local NET buildup correlates with the generation of mucosal autoantibodies such as IgA and IgG ACPAs, suggesting that the airway may serve as an initiation site for systemic autoimmunity. In fact, high levels of both NETs and ACPAs have been detected in the sputum of at-risk patients, supporting a direct association between NET formation and autoantibody production (50–53). Environmental factors such as cigarette smoking exacerbate this process by inducing NET formation via protein arginine deaminase (PAD) 4-dependent pathways, which in turn increases the production of citrullinated antigens in the lung (Figure 1) (54). Smoking not only elevates the risk of ACPA development but also intensifies the inflammatory response by triggering spontaneous NETosis in neutrophils (54–56).

In addition to environmental triggers, infectious agents have been noticed for their role in breaking immune tolerance (Figure 1) (57–59). Multiple pathogens -including Epstein-Barr virus (EBV), Mycobacterium tuberculosis (MTB), Porphyromonas gingivalis (Pg) and others- have been implicated as potential instigators of RA (57–59). Antibodies towards these infections and dysbiosis of mucosae’s microbiota have been found in higher titters on RA and early RA patients, compared to healthy controls (60–64). These microorganisms may trigger autoimmunity through mechanisms such as molecular mimicry, where the structural similarities between microbial antigens and self-proteins provoke a cross-reactive immune response to self-antigens; epitope spreading, which broadens the autoimmune response to additional self-antigens; and bystander activation, where infection-induced inflammation and cytokine release non-specifically activate T cells (65). Together with NET formation, these mechanisms expand the pool of autoreactive T and B cells, lowering the threshold for autoimmunity (48, 58). As an example, high sequence homology between these microorganism’s antigens and key host molecules like interferon regulatory factor 5 (IRF5), involved in macrophage and dendritic cells (DCs) inflammatory response as well as B cell antibody production, has been found (66). This similarity results in cross-reactivity towards antigens from EBV, MAP and self-IRF5 (66, 67).

The mucosal endotype hypothesis further explains RA pathogenesis by emphasizing the role of different mucosal sites: lungs, gut and oral cavity (58). Each of these sites exhibits unique inflammatory responses that may contribute to the systemic generation of autoantibodies. As it has already been covered, early inflammatory changes and local antibody production in lungs have been linked to both smoking and chronic respiratory infections. Similarly, in the oral cavity, periodontal disease driven by pathogens such as Pg not only damages the local mucosal barrier but also promotes NETosis and subsequent citrullination of bacterial as well as host proteins (68–70). This process can initiate a B cell response that eventually leads to the production of autoantibodies, setting the stage for joint inflammation (69). Moreover, the gut microbiome in patients with RA often shows distinct patterns of dysbiosis that are associated with metabolic changes and immune activation (71–73). These gut bacteria alterations can further contribute to the systemic inflammatory setting that underlies RA. Autoreactive B cells, generated autoantibodies or new self-epitopes resulting from these mucosal inflammatory processes can migrate systemically towards the joint (52, 74–79). Collectively, these observations underscore a multifaceted interplay between genetic predisposition and environmental attacks - including smoking, microbial infections and possibly even exposure to inorganic particles like silica (80, 81) - that collectively disrupts immune tolerance, culminating in the onset and progression of RA. Given the multitude of factors and mechanisms capable of breaking tolerance, RA is inherently a heterogeneous and complex disease.

Breach of tolerance is characterized by the aberrant presentation of citrullinated proteins, which primes both innate and adaptive immune responses, ultimately leading to the chronic production of inflammatory cytokines, autoantibodies and perpetuation of tissue damage. Consequently, restoring immune tolerance, particularly to citrullinated proteins, represents a promising therapeutic avenue for achieving remission in RA.

Citrullination of proteins is a posttranslational modification consisting of the deamination of arginine by PAD enzymes. Under physiological conditions, this modification serves as a regulatory mechanism for protein function and is well tolerated by the immune system (84). However, when citrullination overcomes physiological regulation, changes in conformation and charge distribution of peptides leads to disrupted protein interactions, converting citrullinated epitopes into self-antigens (85).

Excessive citrullination can promote protein autophagy and subsequent presentation by DCs, macrophages and thymic DCs, driving CD4⁺ T cell activation (86). It can also enhance peptide binding affinity to MHC-II (87), leading as well to CD4⁺ T cell activation and contributing to tolerance breakdown.

Normally, dominant factors of self-antigens interact with the MHC-II of antigen presenting cells (APCs), while “cryptic” epitopes remain unrecognized. As a result, dominant epitopes become available for recognition during thymic T-cell tolerance, while a population of CD4⁺ T cells remain capable of recognizing cryptic epitopes. As it has been mentioned, citrullination can unmask these epitopes by increasing their MHC-II binding affinity, enabling their presentation and recognition by autoreactive T cells (85).

Interestingly, overcitrullination does not only disrupt tolerance when it happens on a self-antigen; PAD2 enzyme has been reported to citrullinate transcription factors responsible for CD4⁺ T cell differentiation into Th1, Th2 and Th17, altering the differentiation itself and the populations of the resulting type helper T cells (85, 88). Citrullination of cytokines CXCL10 and CXCL11 reduce their interaction with T cells, hindering their chemotaxis to inflammation site (85, 89).

Overall, there are subpopulations of both B and T cells reacting towards citrullinated epitopes in the synovial of RA patients, being the last ones commonly found mainly as Th1 and Th17 phenotypes (90–93). Several efforts have been carried out to identify pathogenic B cell subsets in ACPA/+ RA patients. By single cell RNA-sequencing of CD45⁺ hematopoietic cells, Wu et al. found differences between the synovial immune cell subsets of ACPA/+ and ACPA/- RA patients, pointing out different immunopathological mechanisms related to these autoantibodies (94). Aiming to find pathogenic B cell subsets, Thorarinsdottir et al. found that in ACPA/+ RA patients most of the B cells in SF belonged to a CD21^-/low subset. Under IL-6 stimulation, these cells expressed CXCR3 and RANKL, leading to osteoclast differentiation and bone destruction (95). Among this subset, ACPA/+ patients displayed CD21^-/lowCD27^-IgG^- class significantly increased in peripheral blood and comprising 40% of the CD21^-/low cells in SF (95), matching posterior studies in which CD27^-IgD^- and memory CD27⁺ IgD^- B cells were found in higher ratios in the SF compared to peripheral blood, suggesting these subgroups are key players in RA synovium inflammation (96). Consistent with previous results, Floudas et al. further proved the reduced presence of CD27⁺ IgD⁺ B cells along with the accumulation of PD-1⁺ B cells in SF and synovial tissue of RA patients, compared to healthy controls (97). Other subtypes found in higher percentages in ACPA/+ patients were CD19⁺ B cells (91); and for patients with early RA, human leukocyte antigen (HLA)-DR⁺-peripheral type helper T cells, PD-1^hi CD8⁺ T cells, CXCR5⁻ CD11c⁻ CD38⁺ naive B cells (98) and CD19⁺ CD24^hi CD38^hi regulatory B cells (99).

ACPAs isolated from serum, plasma and SF of RA patients have been found as targeting over 100 citrullinated proteins (100–104). Notably, the affinities of these antibodies vary significantly both among patients and as disease progresses (83, 105–107). While some ACPAs only bind a single target such as citrullinated vimentin, fibrinogen or collagen (108, 109), most are highly promiscuous towards multiple citrullinated epitopes (110–112) or even other posttranslational modifications as acetylation and carbamylation (10, 112–114). Structurally, ACPAs are heavily N-glycosylated in their fragment antigen-binding (Fab) domain. Over 90% of ACPAs (compared to 15-25% of IgGs in human serum) are N-glycosylated in their variable domain (115, 116), and over 80% of receptors on ACPA-producing B cells contains N-glycosylation sites (116, 117). It is suggested that N-glycosylation provides ACPA-producing B cells with a selective advantage, enabling them to escape negative selection of the B cell receptor, thereby promoting autoimmunity (116, 118). On the other hand, Zhao et al. unveiled the complexity N-glycosylation in ACPAs by proving that upregulating sialylation of the crystallizable fragment (Fc) of ACPAs in B cells from collagen induced arthritis CIA mice attenuates disease progression (119), correlating to previous literature reporting decreased sialylation in the Fc region of serum ACPAs from RA patients and how this desialylation is related to inflammatory processes (119–122).

The presence and pathogenesis of ACPA in murine arthritic models has been debated (108, 123). Their proposed pathogenic mechanisms include direct targeting and degradation of citrullinated proteins in joint cartilage, such as type II collagen; enhancing fibroblast-like synoviocyte migration and adhesion within the synovium, where they release proinflammatory cytokines, create an erosive interphase and are involved in the citrullination of new self-antigens (124, 125); direct targeting of osteoclast precursors promoting their differentiation (108, 126); or, as it will be further discussed, interaction with several immune system components resulting in a feedback loop that enhances the production of more ACPAs and proinflammatory agents such as cytokines, reactive oxygen species (ROS) and degradative enzymes, among others (Figure 2). Interestingly, ACPAs have different mechanisms when they interact on their own or via Fc gamma receptor (FcγR) after forming immune complex (IC) with RF (Figure 2). A protective role of ACPAs has also been suggested (127), highlighting the functional diversity of these autoantibodies.

Pointing out the effect of ACPA in different immune subsets, evidence shows that T follicular Th cells responses were reported higher in ACPA/+ than in ACPA/- (96). On the other hand, percentage of disease relevant Th17 was not dependent on seropositivity (91, 92).

ACPAs interaction with citrullinated glucose-regulated protein 78 (grp78) on macrophages’ surface reported activation of extracellular signal-regulated kinases (ERK)1/2 and c-Jun N-terminal kinase (JNK) signaling pathways, as well as enhancing NF-κB activity and tumor necrosis factor-alpha (TNF-α) secretion (128–130). ACPAs can also activate macrophages via TLR4- and MyD88- dependent (131–133) or CD147-integrinb1-PI3K-Akt pathways, this last one activating NF-κB signaling and NLRP3 inflammasome cascade and pro-IL-1β release (134). Otherwise, when found as ICs with RF, monocytes were also stimulated by binding FcγRs which enhanced proinflammatory cytokine release in synovial membrane (135) as well as regulating differentiation into osteoclasts (136, 137). Breedveld et al. stimulated monocytes with SF isolated ICs, resulting in IL-6 and IL-8 release and subsequent activation of osteoclast activation (133, 138). Connection of PAD4 and macrophages in RA has been described. Enzymatically active PAD4 was found present on the monocyte surface, being a source of novel ACPA autoantigens by citrullinating both soluble and surface proteins (139). These findings correlate with the already reported role of SF and lymphoid tissue macrophages in citrullination of proteins and ACPA production (140). Interestingly, autocitrullination of PAD4, which is found in SF ACPA/- patients, exacerbated inflammatory arthritis in mice models through monocyte recruitment, suggesting an ACPA-independent role of PAD4 in RA pathogenesis (141).

Neutrophiles are another immune cell type targeted by ACPAs (142, 143). The already mentioned ICs have been reported to activate neutrophiles leading to cartilage and tissue destruction due to neutrophil degranulation, release of degradative enzymes, ROS, as well as activation of soluble receptors and cytokines causing general tissue inflammation (144, 145). NET formation has also been observed on SF and sera of RA patients (146–149), correlating to ACPA levels and their immune complexes, which enhances inflammatory response in synovial fibroblasts via activation of IL-6, IL-8 and adhesion molecules, among others (138, 147–149). Some forms of NETosis rely on PAD4 activity (150, 151) and results into citrullination of proteins (specially histones) in the synovial space, engaging a positive feedback loop for which either synovial autoreactive ACPA-producing B cells or direct presentation of citrullinated antigens to T cells by fibroblast-like synoviocytes leads to the production of more autoantibodies (148). Indirectly, neutrophils can also get activated through ACPA binding to osteoclasts, as this leads to secretion of CXCL8, promoting neutrophil attraction and NET release, which again increases ACPA activity through binding to citrullinated histones in the released NETs (152).

Even though the appearance of ACPAs has been linked to environmental factors as smoking and some viral infections (among others) (153, 154), ACPA/+ RA patients have shown a gene signature based on the already mentioned HLA complex, which is crucial for antigen presentation between immune cells. Both RA and ACPA development were found to be connected to HLA haplotypes expressing the shared epitope (SE), which codes for a QKRAA peptide motif on the MHC (153–157). Similarly, both humanized and non-humanized mice models expressing different RA-related haplotypes of HLA containing the SE generated ACPAs to a greater extent upon disease induction with PAD rather than those with haplotypes lacking SE (158, 159).

Existing treatments such as MTX and several costly biological therapies can slow disease progression but do not cure the disease. Depending on the cohort, range from 39% to 70% patients do not reach the preferred goal of sustained remission or low disease activity (161–163).

The effect of bDMARDs and JAK inhibitors can also be dependent on ACPA seropositivity. For example, drugs like the JAK inhibitor tofacitinib (169), B cell depletor rituximab (170, 171) and T cell modulator abatacept (172) have better efficacy on seropositive groups, compared to seronegative groups. TNF inhibitors show similar efficacy in seropositive and seronegative disease (172).

In a cross-sectional analysis of RA patients treated with various csDMARDs and/or JAK inhibitors, Neppelenbroek et al. suggested that ACPA⁺ B cells retained their activated and proliferative phenotype, despite effective control of inflammation and clinical disease. The absence of immunological remission might explain why ACPA/+ patients rarely reach sustained drug-free remission. This continued activated state of ACPA-B cells indicates chronic exposure of these cells to stimulating triggers along disease course, which in this study was 11 years (average) (173). Tocilizumab, another FDA-approved bDMARD, managed to decrease synovial T cells and disease activity on patients after 8 weeks of treatment, but did not manage neither to decrease the count of CD68⁺ macrophages or CD20⁺ B cells in synovium, maintaining unchanged local levels of RANKL and significantly increasing systemic levels of IL-6 and RANKL (174), two cytokines that as previously mentioned, are expressed by synovial macrophages and B cells and are related to joint erosion (45, 133, 138).

Despite significant advances, current RA therapies do not achieve durable, immunological remission across all patient groups. Their effectiveness often depends on ACPA status, with seropositive individuals responding more favorably, yet still without showing immunological remission despite clinical improvement. To address this disparity, emerging antigen-specific therapeutic strategies are proposed as considerable therapeutics toward sustained immunological and disease-modifying remission.

Tolerogenesis or tolerance recovery is understood as the process by which the immune system re-establishes its ability to recognize and tolerate self-antigens, thereby preventing autoimmune responses and maintaining immune homeostasis. Mechanisms of immune tolerance can be broadly divided into central and peripheral tolerance. Central tolerance occurs primarily in the bone marrow and thymus, where autoreactive T cells undergo clonal deletion before entering the circulation (175, 176). Peripheral tolerance, by contrast, regulates mature T cells in the periphery through multiple mechanisms, including (i) T cell anergy, where T cells become non-proliferative upon antigen stimuli, commonly lack co-stimulatory molecules and are functionally inactive (177, 178); (ii) T cell ignorance, being ignorant T cells unresponsive to their autoantigens yet potentially able to be activated again (179); (iii) T cell exhaustion, associated with constant antigen exposure (180, 181); (iv) clonal deletion of mature T cells in the periphery, mediated through antigen presentation (182, 183).

Treg cells can suppress local immune responses elicited by Th cells upon receptor activation of disease-causing antigen (184, 185). Additionally, Treg expansion has been proved to reinduce tolerance (177, 186, 187). Highlighting the pivotal role of antigen-specific Treg expansion in tolerance recovery, imbalances in Th1/Treg and Th17/Treg (as well as Th1/Th2 ratios) are often associated with loss of tolerance in RA (17–20).

The bDMARD abatacept targets CD80 and CD86 on the surface APCs including B cells. CD80 and CD86 are key co-stimulatory molecules for antigen presentation and T cell activation. In a study by Lorenzetti et al, in vitro abatacept treatment was shown to decrease CD80-CD86 expression on B cells in a dose-dependent manner. In contrast, clinical assessment revealed only a moderate reduction in ACPA levels but a significant decrease in the ACPA-specific B cell population, suggesting a restoration of tolerance (Figure 3) (188). The bDMARD rituximab targets CD20, leading to a depletion of B cell populations for 6-9 months (189), yet without elevating the likelihood of infections in patients relative to other forms of bDMARD treatment (190, 191). Tolerance recovery is suggested by the posterior regeneration of the B cell subpopulations, finding different subset composition than found before treatment (Figure 3) (192). Naïve B cell population increased, while CD27⁺ memory cells stayed significantly reduced (0.5-fold) for up to 2 years (192).

In a small trial consisting of three patients with treatment resistant RA, CD19-directed Chimeric antigen receptor (CAR)-T cell treatment caused B cell depletion and reduced the pathogenic interleukins IL-6 and TNF-α as well as RF and ACPA levels (194). Along with lowering joint inflammation and the absence of relapse, the progressive repopulation of B cells non-associated to an increase of pathogenic antibodies 9 months after treatment makes CAR-T therapy a promising tool to restore tolerance in difficult to treat cases (Table 1 for detailed information) (194). A reported case with this outcome described one RA patient treated with CD20/CD19-directed CAR-T cell therapy following a diagnosis of diffuse large B cell lymphoma (205). However, most available data for this type of treatment in RA come from very small cohorts, sometimes down to individual case reports (206, 207). In vitro data of similar CAR-T cell therapies backs up the previous results by eliminating autoreactive B cell populations from RA patients’ serum (208). However, longer and bigger trials need to be carried out to dismiss the serious toxic effects that this type of therapy is suspected to have in rheumatic autoimmune disease treatment (206, 209).

Ex vivo-generated autologous tolDCs introduced to a specific antigen have been explored due to their capacity to present antigens to T cells (210). TolDCs are not only able to cause T cell anergy or the expansion of Treg cells by providing constant exposition to the specific antigen in CIA murine model (211, 212), but they also express PD-1 and anti-inflammatory cytokines IL-10 and IL-35 (213). Thus, it is understandable that peptide loaded tolDCs have been successfully utilized in multiple clinical trials aimed at restoring tolerance in autoimmune diseases such as multiple sclerosis (214) and type I diabetes (215, 216). In the case of RA treatment, it is important to mention the few candidates that showed promising results in small cohort clinical trials phase I (9 to 18 patients), currently recruiting for further studies or ongoing clinical trials: Rheumavax (ACTRN12610000373077), AuToDeCRA (ISRCTN14999554) and CreaVax-RA (KCT0000894). AutoDeCRA tolDCs have been exposed to the antigens of the patients’ SF, while CreaVax-RA tolDCs have been exposed to PAD4, RA33 (heterogeneous nuclear ribonucleoprotein A2/B1 (hnRNP A2/B1)), citrullinated filaggrin and vimentin (217). AutoDeCRA trial failed to show efficacy on clinical inflammation parameters, changes in serum cytokine levels or in peripheral T cell phenotype (218). In the case of Rheumavax, consisting on tolDCs with the NF-κB pathway inhibited and exposed to citrullinated peptides derived from vimentin, collagen type II, aggrecan and fibrinogen, better results were observed. 1 month after Rheumavax treatment, T effector (Teff) cells were reduced compared to untreated controls, while the ratio of Treg/Teff increased, pointing to a shift in the immune balance (195). Cytokine IL-15, IL-29, CX3CL1 and CXCL11 levels as well as T cell mediated IL-6 response towards the citrullinated vimentin peptide found in Rheumavax were reduced (195). It is also worth mentioning the TOLERANT clinical study, which is in recruiting stage (phase I, NCT05251870). As well as the previously mentioned therapies, in this trial HSP70 peptide loaded DCs will be employed in order to induce and/or expand Treg populations (219).

Peresolimab, an IgG monoclonal antibody that stimulates PD-1 pathway, showed a positive primary outcome by reducing DAS28 score compared to the placebo group 12 weeks after treatment in a phase II clinical trial. However, in secondary outcome measures, peresolimab was only significantly better than placebo with respect to ACR20 responses, but not with respect to ACR50 or ACR70 responses (220). Aiming for T cell activation suppression, peresolimab is intended to reset the immune response to restore immune tolerance (221). Differently, inhibiting B cell activation and plasma cell differentiation by means of CD40L binding protein targeting, dazodalibep was tested in a phase I trial (196). CD40 is expressed on many APCs (incl. DCs, macrophages and B cells) and non-hematopoietic cells. Effective humoral response to T cell-depending antigens rely heavily on CD40/CD40L interactions between B cells and T cells. Not only DAS28-CRP went down to -2.3 compared to baseline, but a significant reduction in B cell proliferation and T cell-dependent antibody production were reported (196). Further clinical trial (phase II, NCT04163991) confirmed the reduction of DAS28-CRP score over 309 days on a bigger cohort of 62 treated RA patients and 16 disease controls (197).

Part of HSP70 family, binding immunoglobulin protein (BiP) is involved in the peripheral blood monocytes differentiation into DCs and osteoclasts. Treatment of maturing monocytes with BiP results in reduced antigen presentation capacities of DCs due to lower expression of HLA-DR and CD86. Recombinant human BiP administration has been reported to prevent and ameliorate disease in murine CIA models (222, 223). BiP analogue IRL201805 was administrated to RA patients and its effects were monitored for 12 weeks in a phase I/IIa trial. DAS28 score was consistently reduced in the fraction of patients that responded to treatment (43%) (198) without any serious adverse drug reactions reported (199). When the 4 week-after treatment PBMCs of responder RA fraction was incubated with their own PBMCs before treatment, these ones produced significantly less IFN-γ than RA patients treated with placebo. As a part of the inflammatory response regulation, serum levels of pro-inflammatory cytokines IL-1β, TNF-α and IFN-γ were reduced while sCTLA-4 increased. Related to the pro- to anti-inflammatory shift observed in serum cytokine levels, Treg stability and potency related CD39 was found overexpressed in the Tregs of the patients responding to the treatment (198).

Peptides emerge as highly specific and versatile drug candidates for tolerance recovery in RA and other autoimmune diseases (224–231). Their high specificity minimizes potential drug–drug interactions, making them suitable for combination with other RA therapeutics. Preliminary and exploratory trials in RA patients have employed them in combination with different csDMARDs reporting treatment efficacy and no concerning adverse effects (200, 203, 204). Compared to antibodies, their small size, enhanced stability, scalable production and customizable structure make peptides particularly attractive for achieving precise interactions with immune targets while maintaining relatively low immunogenicity and lower production cost characteristic of small molecules (232, 233).

The T cell-activating peptide based on the immunogenic HSP60 Jusvinza, approved in Cuba for cases of COVID-19 with hyperinflammation, is currently under clinical trials (phase III, RPCEC00000433) to treat RA patients. As it has been mentioned previously, NETosis is related to inflammation in RA as well as citrullination of new antigens and production of new ACPAs. Protein expression of neutrophiles from patients treated with Jusvinza was found to be differently modulated, including differences on the already mentioned NF-κB pathway (234). Overall, RA patients treated with Jusvinza displayed PAD4-mediated inhibition of NETosis, which was further confirmed with in vitro experiments (234). Phase I clinical trials in RA patients treated with Jusvinza showed a reduction on blood ACPA levels DAS28 score and achievement of ACR50 and ACR70 in 6 months (200, 201). Complementary, ex vivo assays showed a 1.5-fold increase of Treg cells when PBMCs of RA patients were exposed to Jusvinza (202), suggesting that tolerance recovery towards citrullinated antigens could be mediated by NETosis inhibition. Also derived from an HSP, T cell proinflammatory epitope DNAJP1 peptide was used to treat RA patients with PBMCs reactive towards the candidate (75% of the tested participants) in a clinical trial phase II (203). Upon treatment, TNF-α expressed by T cells decreased significantly while IL-10 expressed by T cell increased, along with PD-1 and CTLA-4 (203). These results match previous phase I outcomes for the same peptide, where DNAJP1-specific T cell number did not change over treatment (hence, there was no clonal deletion) but changes on clusters of differentiation on them pointed that immune reactivity towards the self-peptide did (235).

DEN-181, a subcutaneous formulation consisting of RA-joint HLA-DRB1*04:01- and *01:01-haplotypes specific collagen type II_259-273 peptide and NF-κB pathway inhibitor calcitriol in liposome formulation, reduced the population of citrullinated vimentin-specific T cells in MTX-treated patients under a phase I clinical trial (204). The improvement in disease activity observed in RA patients was associated with the tolerogenic effects of the peptide-based therapy DEN-181, including an early expansion of PD1⁺ collagen type II– and citrullinated vimentin–specific T cells, followed by a reduction in ACPAs, an increase in CCR7⁺ naïve T cells and a decrease in memory B cells (204). CCR7⁺ expression in T cells is related to T cell migration from peripheral tissue to lymph node (236). Disruptions in this migratory process lead to peripheral tissue Teff cell accumulation in inflammation and autoimmunity, incl. RA (237). Used in DEN-181, calcitriol is a metabolite of vitamin D. Vitamin D has been reported to elevate the percentage of Tregs and lower the DAS-28 score just after 3 months of supplementation along MTX and hydroxychloroquine in RA patients, compared to a group that were just treated with the csDMARDs (238).

Citrullinated antigens have also demonstrated potential in promoting tolerance recovery. This process is typically achieved through the persistent exposure to the antigen via repeated administration of low doses, aiming to modulate T cell population by depleting or causing T cell exhaustion on pathologic Th1 and Th17 cells, reducing the expression of proinflammatory cytokines that mediate these, or increasing the population of Tregs (239). Gertel et al. utilized a multiepitope citrullinated peptide, containing motifs from key citrullinated proteins in RA such as filaggrin, fibrinogen, vimentin and collagen. Their approach successfully improved the clinical status of adjuvant-induced RA rats (Table 2). The increase of Treg and reduction in Th17 cells, previously associated with reactivity towards citrullinated epitopes (90–93), indicated tolerance induction (239).

CEL-4000 consists of a proteoglycan (PG) non-citrullinated epitope derived from cartilage PG aggrecan (PG70) conjugated to a ligand specific for CD4⁺ T cells. This design allows the T cell presentation of the immunomodulatory peptide to an APC via MHC II while the CD4⁺ ligand modulates T- cell activity. CEL-4000 was tested in PG-induced arthritis (PGIA) and G1 domain-induced arthritis (GIA) mice models, switching cytokine production from Th1 and Th17 pro-inflammatory (TNF-α, IL-17 and IFN-γ) signature to an Th2 and Treg anti-inflammatory (IL-10, IL-4 and TGF-β) one, as well as an increase in Treg cells (240, 246). CEL-5000, which introduces a citrullinated PG epitope was also tested in PGIA and GIA mice models. CEL-4000 and CEL-5000 developed different immune responses, since mice did not produce high antibody titters for the citrullinated epitope conjugate while they did for CEL-4000’s. However, both treatments lowered arthritic score, reduced inflammation levels (assessed by immunohistochemistry) and achieved the same Th2-like anti-inflammatory cytokine response (240).

Fibrinogen-derived citrullinated peptides have been intensively investigated due to their high capacity to scavenge ACPA isolated from RA patients, showing that cyclized structures bind with higher affinities (247, 248). A fibrinogen-derived citrullinated cyclic peptide have also been reported to treat CIA rat, showing a significant decrease of joint swelling when compared to untreated or non-citrullinated peptide control groups, along with an increase of IL-10 (241). Data obtained from McElwee et al. suggests that citrullinated fibrinogen may have potent tolerogenic properties (242). When they immunized a transgenic HLA-DR mice model with citrullinated peptides derived from cartilage intermediate layer protein or fibrinogen, the arthritis-initiating response from CD4⁺ T cells upon presentation of citrullinated antigens was not observed. Instead, expansion of CD4⁺ T cell population binding to these HLA-DR:citrullinated peptides was observed, with lower levels of Th1 and higher levels of Treg cells. These results were not observed when same mice model was immunized with citrullinated vimentin or enolase 1 peptides (242). Restoration of Treg over Th or Teff cell populations and balance recovery was also seen in the already discussed Gertel et al. study, where the multiepitope used to treat adjuvant-induced RA rats contained citrullinated fibrinogen (239); or on the successful Rheumavax trial, which contained tolDCs introduced to citrullinated fibrinogen, among other citrullinated peptides (195).

Gomez et al. recently showed that the injection of several ACPAs isolated from RA patients ameliorated inflammation and disease severity in collagen antibody-induced arthritis (CAIA) model (127) adding up to a long list of examples where ACPAs had therapeutic or preventive effects in RA murine models (243, 244, 249). It is worth mentioning that in the experiments carried out by Gomez et al, patients derived ACPAs were grouped and dosed based on the predominant citrullinated antigen they targeted and all groups had similar effects specially when injecting in early steps of CAIA (127).

This seems to point out that in a target independent manner, ACPAs have the ability to induce tolerance (in earlier stages) or prevent break of tolerance that will exacerbate the disease in a CAIA model (Figure 4A). He et al. also injected patient derived ACPAs in healthy mice to observe neither arthritogenicity nor pain signs. One of the antibodies protected the mice from antibody-induced arthritis (CAIA model) by forming ICs with citrullinated α-enolase and other few citrullinated proteins from SF which posteriorly bound to macrophages in the joints, resulting in increased secretion of anti-inflammatory IL-10 and reduced osteoclastogenesis (243). Authors attributed the reduction of osteoclastogenesis to the interaction of the interaction of these ICs with FcγRIIB in macrophages (Figure 4A) (136, 243). Another example of ACPA usage to prevent the development of inflammation not only in CAIA RA model but also in other NET-mediated pathologies as inflammatory bowel disease, pulmonary fibrosis and sepsis, was reported by Chirivi et al. (245) They developed an ACPA that specifically targets citrullinated histones citH2A and citH4, which are generated during NET release. Their lead therapeutic ACPA prevented the inflammatory response in several autoimmune models (incl. CAIA), reduced inflammation in CIA mice RA model, inhibited NET formation in both murine models and in vitro experiments where healthy individuals neutrophiles were stimulated with different disease-related stimuli (including SF containing ACPAs from RA patients) and potentially favoured macrophage mediated clearance of NETs (Figure 4B) (245).