Background
Rheumatoid arthritis (RA) is an autoimmune disorder characterized by chronic inflammation of the synovial joints leading to the destruction of cartilage, bone, and ligaments [
1]. Conventional treatment of RA with disease-modifying anti-rheumatic drugs (DMARD) aims to limit disease symptoms, delay or prevent future joint destruction, and target low disease activity or remission. Low-dose methotrexate (MTX) is the traditional DMARD administered weekly either alone or in combination therapy. MTX has been proven safe and efficient [
2]. However, nearly a quarter of patients treated with MTX have to discontinue treatment because of poor responses, adverse effects (e.g., hepatic, gastrointestinal, hematological, renal, or pulmonary toxicity), or both [
3,
4]. Biological agents, such as anti-TNF therapy, combined with MTX have significantly improved the treatment of RA. However, again, some RA patients are refractory or contraindicated to these agents [
4,
5], and thus, new therapeutic strategies are needed.
Apoptotic cell administration has been shown to control chronic inflammatory disorders by diminishing the pro-inflammatory state and to induce or restore tolerance to auto-antigens by inhibiting pathogenic T or B cell responses and by inducing pro-tolerogenic/regulatory cells [
6‐
8]. Prevention of arthritis by apoptotic cell injection has been reported in mouse and rat models [
9‐
12]. Prevention means that apoptotic cells are infused at the time of arthritic disease induction (i.e., at time of immunization with auto-antigens), which does not mimic the clinical situation. However, intravenous (i.v.) apoptotic cell infusion can be used for experimental treatment of disease, such as in sepsis [
13,
14]. These data are interesting, because apoptotic cell administration during the disease (i.e., as treatment) protects mice from sepsis-induced death [
13,
14], while infusion 5 days before sepsis (as prevention) worsens mice survival, possibly by decreasing the capacity to secrete interferon (IFN)-γ [
15]. As in arthritis models [
9‐
12], sepsis is controlled independently of the apoptotic cell origin [
13,
14]. Recently, a phase 1/2a clinical study was conducted in 13 patients who received i.v. donor apoptotic cell infusion the day before allogeneic hematopoietic cell transplantation in order to alleviate the occurrence of acute graft-versus-host disease (GvHD) [
16]. The apoptotic cell number infused in patients was transposed from animal models [
17]. There was no specific toxicity associated with i.v. apoptotic cell infusion. Historical data on acute GvHD and the available literature suggest promising potential for GvHD prophylaxis [
16]. This clinical study opens the way to apoptotic cell-based therapy in other clinical settings already assessed in experimental models, such as RA. Here, we propose to assess whether i.v. apoptotic cell infusion may control ongoing collagen-induced arthritis (CIA) and determine the mechanisms involved by focusing on antigen presenting cells (APC) and regulatory CD4
+ T cells (Treg).
A major concern with novel therapeutic approaches, such as apoptotic-cell-based therapy, is the interaction with other treatments received simultaneously by the patients. For instance, MTX, the gold standard treatment for RA, may be given alongside biologic agents, including anti-TNF therapy. We have already studied the interactions of i.v. apoptotic cell infusion with immunosuppressive drugs routinely used in the context of allogeneic hematopoietic cell transplantation. Rapamycin (sirolimus) has been shown to exert a synergic effect, while cyclosporine A neutralizes apoptotic-cell-induced allogeneic hematopoietic cell engraftment [
18]. This kind of study has to be extended to other conventional drugs in the treatment of RA, such as MTX and anti-TNF agents. We also addressed interactions between i.v. apoptotic cell infusion and MTX or anti-TNF therapy in the CIA model.
Methods
Mice
Female DBA/1, (Janvier, Le Genest-Saint-Isle, France) and C57Bl/6 (Charles River Laboratories, L’Arbresle, France) mice, 8–10 week old, were housed in filter-top cages and fed a standard diet with freely available food and sterile water (Plexx, Elst, Netherlands), at the UMR1098 animal facility (agreement number D25-056-7). All experimental studies were approved (number 02831) by the local ethics committee (Comité d’éthique Bisontin en Expérimentation animale, number 58) and the French Ministry of Higher Education and Research (Ministère de l’Enseignement Supérieur et de la Recherche), and was conducted in accordance with the European Union Directive 2010/63.
Induction of collagen-induced arthritis
The induction of arthritis has been described previously in detail [
19]. Briefly, DBA/1 susceptible mice were immunized by subcutaneous injection at the tail base with 100 μL of bovine type II collagen (hereafter called collagen) dissolved in 0.05 M acetic acid (4 mg/mL; MD Bioproducts, Zurich, Switzerland) emulsified in an equal volume of complete Freund’s adjuvant (CFA, 10 mg of
Mycobacterium tuberculosis (MBT) strain H37Ra (Difco, Detroit, USA) per milliliter of incomplete Freund adjuvant (Sigma Aldrich, St Louis, MO, USA)).
Arthritis developed in all mice 20 to 25 days after collagen immunization. Arthritis severity was determined by daily blinded visual examination of the paws as follows: 0 = no change; 1 = redness or swelling of one toe; 2 = redness or swelling of two toes; 3 = redness or severe swelling of two toes or more digits, erythema, or swelling involving the entire paw; and 4 = redness or swelling of the entire paw, spreading to the ankle. The clinical score for each mouse was the result of the sum of scores for the four limbs (maximum score = 16). When arthritis reached a clinical score of 7–8, mice received apoptotic cells i.v. (5.10e6 cells/300 μL/mouse, or as indicated), or vehicle (300 μL of PBS).
MTX (provided by the University Hospital, Besançon, France), was given intraperitoneally (i.p.) once a week (15 mg/kg), starting on the day of apoptotic cell injection. Anti-TGF-β antibody (clone 1D11 (R&D Systems, Lille, France) or 2G7 clone provided by Prof. L. Chatenoud (Necker Hospital, Paris, France)) was given i.p. on the day of apoptotic cell injection (150 μg/mouse) and 48 h later (100 μg/mouse). Anti-TNF antibody (clone TN3-19.12 (BD Biosciences; Le Pont de Claix, France)) was given i.p. on the day of apoptotic cell injection (300 μg/mouse) and at 72 h, 6 days, and 9 days later (300 μg/mouse/injection). CIA mice that did or did not receive apoptotic cells were killed on days 10 to 12 post treatment and were harvested for blood, lymphoid organs, and ankles for further analysis.
Induction of apoptotic cells
Cells were issued from the thymus of naïve DBA/1 mice and submitted to a 35-Gy dose of x-ray irradiation (Raycell blood irradiator; Best Theratronic, Ottawa, ON, Canada) followed by 6-h culture in complete DMEM Glutamax-I (Life Technologies, Gaithersburg, MD, USA) supplemented with 10 % heat-inactivated FCS (Life Technologies), 1 % penicillin/streptomycin, 10 mM HEPES buffer (Sigma Aldrich), 10 mM nonessential amino acids (Invitrogen, Cergy Pontoise, France), and 0.05 mM 2-mercaptoethanol (Sigma Aldrich), to allow apoptotic changes to occur before injection [
20]. Apoptotic cells were then washed twice in PBS before injection into the tail vein in 300 μL of PBS. Non-treated arthritic mice received PBS as control. Apoptosis was confirmed using fluorescein isothiocyanate (FITC) or phycoerythrin (PE)-conjugated annexin V staining and 7-aminoactinomycin D2 (7-AAD) exclusion (BD Biosciences) and flow cytometry analysis. At the time of injection, apoptotic cells were mainly early-stage apoptotic cells (70–85 % of cells were annexin V
+ and 7-AAD
–; less than 10 % of cells were 7-AAD
+) [
20].
Ex vivo analysis of T cells and antigen presenting cells
Lymphoid organs, spleen, and inguinal and axillary lymph nodes were harvested and dissociated, and erythrocytes were removed by osmotic shock. After washing with PBS, lymphoid cells were directly stained for CD4 (clone RM4-5, BD Biosciences), CD25 (clone PC-61, BD Biosciences) and Foxp3 (clone FJK-16 s, eBioscience) expression following manufacturer’s instructions. Stained cells were analyzed using a FACS Canto II cytometer with Diva software (BD Biosciences).
Cells issued from lymph nodes were also plated and activated using CD3-specific antibodies (Ab) (0.5 μg/mL; clone 145-2C11; Biolegend) or collagen protein as indicated and cultured for 5 days in complete medium. T cell proliferation was then evaluated using 5-Bromo-2’-deoxyuridine (BrdU) incorporation and counting (Perkin Elmer, Waltham, MA, USA). Spleen CD4+CD25+ T cells were enriched using immuno-magnetic cell sorting (MACS; CD4+CD25+ Regulatory T Cell Isolation Kit; Miltenyi Biotec, Paris, France) according to manufacturer’s instructions and used in collagen-specific or MBT-specific proliferation assays or in CD3/CD28-stimulated T cell cultures at different concentrations. For collagen-specific or MBT-specific co-cultures, naïve CD4+CD25– T cells were isolated from CIA mice by MACS (Miltenyi Biotec) and cultured (100.10e3 cells) with CD11c+ dendritic cells (DC) (50.10e3 cells) also isolated from CIA mice by MACS (CD11c MicroBeads (Miltenyi Biotec)), in the presence of 50 μg/mL of collagen (MD Bioproducts, Zürich, Switzerland) or MBT (Difco) protein. For CD3/CD28-stimulated T cell cultures, CD4+CD25– naïve T cells were isolated from naïve DBA1 mice and cultured in the presence of coated anti-CD3 (clone 145-2C11) and soluble anti-CD28 (clone 37.51) antibodies (2.0 and 0.5 μg/mL, respectively). After adding sorted Treg and after 4 days of culture, T cell proliferation was assessed by BrdU incorporation and counting.
Conventional DC (cDC (CD11c+)), plasmacytoid DC (mPDCA+B220+) and macrophages (CD11b+) were analyzed in the spleen for ex vivo expression of costimulatory molecules CD40 (clone 3/23, Biolegend) and major histocompatibility class (MHC) II IA/IE (clone M5/114.15.2, Biolegend) molecules by FACS (with the corresponding labeled antibodies). In addition, cDC, pDC and macrophages were isolated by MACS and cultured in the presence of CpG-ODN2216 (for pDC, 12 μg/mL (Invivogen, Toulouse, France)) or lipopolysaccharide (LPS) (for cDC and macrophages, 1 μg/mL (Sigma Aldrich, St Louis, MO, USA)) for 24 h in complete medium and the same marker expression was assessed by FACS. Isolated APC were also cultured with allogeneic naïve CD4+CD25– T cells (at a ratio of one APC to two T cells) for 5 to 7 days in complete medium and Foxp3 expression was evaluated by FACS in CD4+ T cells according to manufacturer’s instructions.
Assessment of circulating anti-collagen antibody titers
Concentrations of anti-bovine type II collagen antibodies (IgG2a isotypes) were determined in plasma by enzyme-linked immunosorbent assay (ELISA) [
21]. Briefly, 96-well plates were coated with 0.5 μg/mL of bovine type II collagen. Nonspecific binding sites were blocked with a 2 % solution of BSA. Serial dilutions of mouse sera were added followed by incubation with IgG2a-specific goat anti-mouse antibody (horseradish peroxidase (HRP)-labeled) and 3,3'-5,5' tetramethylbenzidin (TMB) (MOSS Inc., Pasadena, MD, USA) as substrate. Absorbance was measured at 450 nm.
Histological analysis
Whole hind legs were harvested, cut through the sagittal axis, fixed with 10 % neutral formalin, decalcified using rapid decalcifier solution (Eurobio, Les Ulis, France), and embedded in paraffin. Serial tissue sections (5 μm) were stained with hematoxylin, eosin and saffron (Merck, Darmstadt, Germany). Joint inflammation was evaluated blinded by a research pathologist and determined in the lower ankle of the two hind legs with a cumulative score at high magnification, looking at cartilage (0 = no change, 1 = erosion, 2 = absence) and synovial inflammation (0 = no infiltrate, 1 = middle mononuclear cell infiltrate, 2 = dense mononuclear cell infiltrate), and a global score for the ankle at low magnification from 0 to 5, with a maximum score of 18 per mouse.
Statistical analysis
All data were analyzed with Prism 5 (GraphPad; La Jolla, CA, USA) or SigmaStat 3.5 (Systat Software Inc; London, UK) software. Statistical significance was determined by the indicated adequate tests.
Discussion
Despite significant improvement in the management of patients with RA through the development of innovative drugs (e.g., biologic agents) [
25,
26], new therapeutic strategies are constantly needed. Indeed, some patients are still non-responsive to the current drugs and toxicities, or contraindications have been reported [
4,
26]. Most of the biologic agents target inflammatory cytokines or B cells, while limited biologic agents (i.e., abatacept) neutralize T cell activation [
25,
26]. There is no approved drug that targets APC, such as dendritic cells, while APC orchestrate the immuno-pathogenic response.
Apoptotic cell-based therapy can be an interesting approach as apoptotic cell infusion has been shown to prevent experimental RA [
9‐
12] and to target T-cell-mediated diseases associated with dysregulated inflammatory cytokine secretion and APC dysfunction [
6]. A dysregulated intricate interplay between T cells, macrophages and inflammatory cytokines is implicated in RA pathophysiology. Here, we used a mouse model of arthritis that recapitulates RA [
27]. We report in this model that i.v. administration of apoptotic cells is able to treat ongoing arthritis and is associated with the reprogramming of APC, in particular pDC and macrophages, and induction of auto-antigen-specific Treg. Moreover, this new therapeutic approach can be used alongside either MTX or anti-TNF therapy (i.e., two standard treatments of RA) with similar efficacy. This allows us to propose this approach in clinical settings, as already performed for the prevention of GvHD after allogeneic hematopoietic cell transplantation [
16]. As observed in mice, patients would have to receive such a treatment when presenting with a moderate 20 % improvement in the American College of Rheumatology (ACR20) clinical score for a higher therapeutic effect.
The novelty of this study resides in the use of apoptotic cell infusion to treat ongoing arthritis, and not, as previously reported by us [
11] and others [
9,
10,
12], to prevent the induction of arthritic disease. The analysis of the mechanisms involved in this therapeutic effect of i.v. apoptotic cell infusion identifies the modulation of APC functions and reprogramming, induction of auto-antigen-specific Treg, and reduction of circulating auto-antibody levels. Adoptive transfer of polyclonal Treg has been already shown to inhibit ongoing arthritis in the CIA model, but this approach only ameliorated the severity of CIA without affecting collagen-specific T and B cell responses [
28]. This is in contrast with our present data showing that apoptotic cell infusion through the direct targeting of APC function affects collagen-specific immune response. The CIA model, and more precisely the immunization of mice with bovine type II collagen in CFA containing MBT, allowed us to appreciate the antigenic specificity of apoptotic cell-induced Treg with a selective induction of collagen-specific Treg, but not MBT-specific Treg. The precise mechanisms allowing this specific increase in auto-antigen Treg (here collagen as the model antigen), but not infectious antigen (MBT), remains to be determined. One may believe that distinct APC subsets or the antigen-presenting pathways are in charge of auto-antigen or infectious antigen and that this subset or pathway is differentially targeted by apoptotic cell infusion. This remains to be determined.
We previously observed that pDC were sensitive to factors issued from macrophages eliminating apoptotic cells but not to apoptotic cells directly [
20], in contrast to DC or macrophages [
23]. Here, macrophages will be the first ones to interact and eliminate apoptotic cells injected i.v., in contrast to cDC. Then, as previously reported [
20], this microenvironment provided by macrophages eliminating apoptotic cells will directly favor pDC to acquire a tolerogenic profile, notably with the
de novo production of TGF-β, favoring pDC to allow generation of antigen-specific Treg [
19]. However, this microenvironment only allowed cDC to favor less Th1 polarization ex vivo (not shown). It seems that apoptotic cells differentially affect cDC, as when engulfed by cDC they favored cDC pro-tolerogenic reprogramming, but when engulfed by macrophages, the issued factors only limited their pro-inflammatory properties.
Similar observations have been performed in another apoptotic cell-based therapy implicating in vivo induction of apoptosis and phagocyte administration together with auto-antigen-derived peptides [
29]. Despite differences in the process and the targeted diseases (EAE and type 1 diabetes), this latter approach described similar mechanisms to our therapy, including: apoptotic cell phagocytosis, TGF-β-dependent Treg generation [
29].
Induction of collagen-specific Treg by apoptotic cell injection is inhibited by TGF-β neutralization, as previously reported in other apoptotic cell-based therapies [
17,
29]. Most of the studied effects observed here after apoptotic cell treatment are dependent on TGF-β and neutralization of this cytokine prevents improvement in arthritis. Anti-collagen IgG2a antibody levels reached the levels found in untreated arthritic mice when neutralizing anti-TGF-β antibody is infused together with apoptotic cells. This confirms our previous data with anti-donor allo-antibodies in the settings of bone marrow graft rejection [
30]. In fact, only APC reprogramming persisted after neutralization of TGF-β in the setting of apoptotic cell infusion in arthritic mice. This appears logical as TGF-β is secreted by APC after apoptotic cell removal [
23,
31] but elimination of apoptotic cells (called efferocytosis) per se is sufficient to modify several APC functions, including the refractory response to TLR ligands [
32]. The effects on APC functions induced after apoptotic cell treatment are conserved when apoptotic cells are infused together with anti-TNF therapy. This can be an encouraging argument to combine both treatments.
Association of new therapeutic approaches with drugs used routinely is infrequently assessed in experimental models. To the best of our knowledge, this has only been tested in the context of apoptotic cell infusion in one study by our group in the settings of allogeneic hematopoietic cell transplantation [
18]. Here, we reported that the therapeutic effects of apoptotic cell infusion are conserved in the presence of MTX or anti-TNF therapy. However, there are differences in apoptotic cell-induced mechanisms. For instance, MTX abolishes the diminution of anti-collagen auto-antibodies induced by apoptotic cell injection. MTX, a traditional DMARD with folate antagonist and anti-inflammatory activity, remains the gold standard treatment for RA. Several potential mechanisms of MTX activity have been proposed, but the precise mechanism(s) of action in RA are still not well-understood [
33]. In animal models, the effects of MTX have been attributed to MTX-induced anti-inflammatory adenosine production [
33]. This production is mediated by CD39 and CD73 ectoenzymes [
33]. Both ecto-enzymes are known to be expressed by Treg and represent one of the suppressive mechanisms of Treg [
34].
MTX treatment demonstrated a limited effect on arthritis severity when started in mice presenting with an advanced arthritic clinical score of 8. In this context, after the injection of apoptotic cells, collagen-specific Treg are still generated. We can propose that apoptotic cell therapy in CIA favors the generation of Treg from existing antigen-specific T cells, as no cell proliferation is needed and MTX can exert an anti-proliferative effect even if the cytostatic effect of MTX is still a matter of debate in RA [
33]. However, the origin of antigen-specific Treg in our settings needs to be further investigated. Thus, through the induction of an anti-inflammatory microenvironment, apoptotic cells can favor the establishment of collagen-specific Treg-associated long-term tolerance. Anti-TNF therapy is synergic with apoptotic cell infusion on clinical outcomes, but we did not identify any supplementary effects on the immune parameters studied when anti-TNF antibody is combined with apoptotic cells compared with apoptotic cells alone. This suggests that additional mechanism(s) such as the complete neutralization of circulating TNF, may participate in the synergic clinical outcomes. The relationship between TNF, anti-TNF therapy and Treg plasticity and suppressive functions is complex and remains a matter of debate [
35], notably in humans [
36,
37]. Here, anti-TNF therapy does not affect the number of splenic Treg or the collagen-specific Treg suppressive function.
Abbreviations
7-AAD, 7-aminoactinomycin D2; Ab, antibodies; APC, antigen presenting cells; BrdU, 5-Bromo-2’-deoxyuridine; BSA, bovine serum albumin; CFA, complete Freund adjuvant; CIA, collagen-induced arthritis; cDC, conventional dendritic cells; DC, dendritic cells; DMARD, disease-modifying anti-rheumatic drug; DMEM, Dulbecco’s modified Eagle’s medium; FACS, fluorescence-activated cell sorting; FCS, fetal calf serum; GvHD, graft-versus-host disease; IFN, interferon; i.p., intraperitoneally; i.v., intravenous; LPS, lipopolysaccharide; MBT, mycobacterium tuberculosis; MTX, methotrexate; PBS, phosphate-buffered saline; pDC, plasmacytoid dendritic cells; RA, Rheumatoid Arthritis; TGF-β, transforming growth factor-β; TLR, toll-like receptor; TNF, tumor necrosis factor; Treg, regulatory T cells
Acknowledgements
We thank Dominique Paris and Francois Coulon for animal care.