Therapeutic Effect of Human Amniotic Membrane–Derived Cells on Experimental Arthritis and Other Inflammatory Disorders

Authors


Abstract

Objective

Rheumatoid arthritis (RA) is an autoimmune disease caused by loss of immunologic self tolerance and characterized by chronic joint inflammation. Cells isolated from human amniotic membrane (HAMCs) were recently found to display immunosuppressive properties. The aim of this study was to characterize the effect of HAMCs on antigen-specific T cell responses in RA patients and to evaluate their therapeutic potential in a preclinical experimental model of RA.

Methods

We investigated the effects of HAMCs on collagen-reactive T cell proliferation and cytokine production, on the production of mediators of inflammation by synoviocytes, and on the generation of Treg cells in peripheral blood mononuclear cells and synovial membrane cells isolated from RA patients. Mice with collagen-induced arthritis (CIA) were treated with HAMCs after disease onset, and clinical scores and joint levels of mediators of inflammation were evaluated. We determined Th1/Th17-mediated autoreactive responses in the mice by measuring the proliferation and the cytokine profile of lymph node cells restimulated with collagen.

Results

Treatment with HAMCs suppressed synovial inflammatory responses and antigen-specific Th1/Th17 activation in cells isolated from RA patients. Moreover, HAMCs stimulated the generation of human CD4+CD25+FoxP3+ Treg cells with a capacity to suppress collagen-specific T cell responses. Systemic infusion of HAMCs significantly reduced the incidence and severity of CIA by down-regulating the 2 deleterious components of disease: Th1-driven autoimmunity and inflammation. In mice with CIA, HAMC treatment decreased the production of various inflammatory cytokines and chemokines in the joints, impaired antigen-specific Th1/Th17 cell expansion in the lymph nodes, and generated peripheral antigen-specific Treg cells. HAMCs also protected the mice from experimental sepsis, inflammatory bowel disease, and autoimmune encephalomyelitis.

Conclusion

HAMCs have emerged as attractive candidates for a cell-based therapy for RA.

Mesenchymal stem cells (MSCs) are resident mesoderm-derived stromal cells that function as precursors of nonhematopoietic connective tissues ([1]). Besides their capacity to differentiate into various cell lineages and their clinical involvement in tissue repair, MSCs from different sources, mainly bone marrow and fat, have recently emerged as potent regulators of the immune response. Thus, MSCs inhibit the activation of T cells, B cells, and dendritic cells while they induce the generation of Treg cells and regulatory macrophages ([2-5]). Experimental evidence indicates that MSCs constitute a promising tool for the treatment of allograft rejection and autoimmunity ([2-5]). Clinical trials using MSCs in the treatment of rheumatoid arthritis (RA) and other immune disorders are already under way ([2, 5-7]).

We previously demonstrated that cells with characteristics of MSCs can be successfully isolated from human amniotic membranes ([8]). Interestingly, cells derived from either the epithelial or mesenchymal layers of human amniotic membrane–derived cells (HAMCs) impair lymphocyte activation, dendritic cell maturation, and inflammatory cytokine production in vitro ([9-13]). Moreover, treatment with total amniotic membrane, amniotic-derived cells, or their conditioned medium was found to be protective in different models of inflammatory fibrotic conditions ([9, 12, 14, 15]). On these bases, the use of HAMCs in the treatment of autoimmune diseases becomes an attractive possibility still to be explored. The aim of the present study was to investigate the effect of expanded HAMCs on collagen-reactive T cells and synovial cells from RA patients and to evaluate their therapeutic potential in a preclinical model of RA and other inflammatory/autoimmune disorders.

MATERIALS AND METHODS

Cell isolation

Cells were isolated from the amniotic membranes of 4 full-term human placentas obtained immediately after birth. The amnion was manually separated from the chorion and washed extensively in phosphate buffered saline (PBS) containing 100 mg/ml of penicillin/streptomycin and 2.5 mg/ml of amphotericin B. Cells from the amniotic membrane (from both the epithelial and mesenchymal layers) were obtained by stepwise enzymatic digestion. First, sterilized amnion fragments (9 cm2) were incubated for 9 minutes at 37°C with Dispase (250 μg/ml) and then for 3 hours with collagenase (0.9 mg/ml) and DNase (20 μg/ml). The remaining undigested amnion fragments were incubated for 1 minute with 2.5 mg/ml of trypsin.

The mixture of cells derived from the entire amniotic membrane was filtered through a 100-μm cell strainer, collected by centrifugation (300g for 10 minutes) and characterized after isolation according to the expression of CD73 (mean ± SD 90 ± 5%), CD166 (85 ± 10%), CD90 (10 ± 5%), CD13 (15 ± 5%), CD45 (1 ± 0.5%), and CD14 (1 ± 0.5%). Cells were then plated (5 × 104/cm2) in complete medium (Dulbecco's modified Eagle's medium [DMEM] containing 10% fetal bovine serum, 2 mMl-glutamine, and 100 mg/ml penicillin/streptomycin) at 37°C in an atmosphere of 5% CO2. Once the cells reached confluence (usually after 7–9 days of culture), adherent cells were then trypsinized and subcultured at a density of 2 × 104/cm2 until passage 3. After 2–3 passages, these cells (referred to herein as HAMCs) showed a high proportion of fibroblast-like cells expressing mesenchymal phenotype markers (data available upon request from the corresponding author).

Peripheral blood mononuclear cells (PBMCs) were isolated from 9 RA patients (2 men and 7 women) by density sedimentation on Ficoll-Hypaque gradients. All patients met the American College of Rheumatology/European League Against Rheumatism 2010 classification criteria for RA ([16]), had clinically active synovitis, and were treated only with nonsteroidal antiinflammatory drugs. Five of the PBMC samples responded positively to type II collagen (CII), as determined by a ratio of >2 for 3H-thymidine (3H-TdR) incorporation in the presence versus the absence of CII (see below). The PBMC samples that did not respond to CII were used as controls.

Fresh synovial membrane cells were isolated by collagenase digestion of synovial tissue obtained from 6 RA patients at the time of knee replacement surgery.

Cell cultures

PBMCs (105) were cultured in duplicate in DMEM complete medium in the presence or absence of chicken CII (400 μg/ml; Sigma), with or without HAMCs (2 × 104), in flat-bottom 96-well plates (at 37°C in an atmosphere of 5% CO2), as previously described ([17, 18]). After 72–96 hours, proliferation was evaluated by the addition of 2.5 μCi/ml of 3H-TdR during the last 8 hours of culture and the determination of incorporation (in counts per minute). After 48 hours, the cytokine content in the culture supernatants was determined by enzyme-linked immunosorbent assay (ELISA) (BD PharMingen).

To identify whether HAMCs selectively affected the production of cytokines on T cells, the intracellular cytokine contents were determined in CD4+ T cells from RA patients. HAMCs and RA PBMCs were stimulated with CII (400 μg/ml) for 14 hours and, during the last 6 hours, in the presence of monensin (1.33 μM; Sigma). Cells were then stained for 8 hours at 4°C with allophycocyanin (APC)–labeled anti-CD4, anti-CD3, or anti-CD14 monoclonal antibodies (5 μg/ml; BD Bioscience), washed, fixed/permeabilized with Cytofix/Cytoperm solution (BD Bioscience), stained with phycoerythrin (PE)–conjugated anticytokine-specific monoclonal antibodies (BD PharMingen), and analyzed in a FACSCalibur flow cytometer (BD Bioscience).

The cell-contact dependence of the suppressive response was evaluated by placing CII-stimulated PBMCs (105) in the lower chamber of a Transwell system (0.4-μm pore; Millipore) and HAMCs (2 × 104) in the upper chamber. After 96 hours, the proliferation of PBMCs in the lower compartment was determined.

To determine the number of Treg cells, cells were incubated for 8 hours at 4°C with fluorescein isothiocyanate (FITC)–labeled anti-CD25 and APC-labeled anti-CD4 monoclonal antibodies (5 μg/ml; BD Bioscience). After extensive washing, cells were fixed/permeabilized with FoxP3/Transcription Factor Staining buffer set (eBioscience), stained for 30 minutes at 4°C with PE-labeled anti-FoxP3 monoclonal antibody (4 μg/ml; eBioscience), and analyzed in a FACSCalibur flow cytometer. The suppressive activity of Treg cells is described in the Supplementary Methods (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38206/abstract).

Synovial membrane cells (2 × 105) were stimulated with lipopolysaccharide (LPS; 1 μg/ml) or tumor necrosis factor α (TNFα; 20 ng/ml) in the presence or absence of HAMCs (105). After 30 hours, culture supernatants were examined for cytokine contents by ELISA, as well as for collagenase and gelatinase activities (EnzChek kit; Molecular Probes).

When indicated, CII-activated PBMCs or LPS/TNFα-activated synovial membrane cells were incubated with conditioned medium (1:1 dilution with complete medium) collected from HAMCs cultures (see Supplementary Methods at the Arthritis & Rheumatology web site). In some experiments, synovial membrane cells, PBMCs, and HAMCs were coincubated with neutralizing anti-human interleukin-10 (IL-10) antibodies, isotype control antibodies (10 μg/ml, BD Bioscience), indomethacin (10 μM), and/or NS398 (10 μM).

Induction and treatment of collagen-induced arthritis (CIA).

DBA/1 mice (8 weeks old; The Jackson Laboratories) were injected subcutaneously with chicken CII (200 μg) emulsified in Freund's complete adjuvant containing Mycobacterium tuberculosis H37Ra (200 μg) on days 0 and 21. Treatment was initiated after disease onset (arthritis score >2) and consisted of 2 intraperitoneal injections of PBS (control) or 106 HAMCs on days 24 and 29. Alternatively, mice with CIA (arthritis score >2) were treated intraperitoneally on days 24 and 29 with 106 allogeneic cells isolated from the placenta of BALB/c mice on days 16–18 of gestation through sequential enzymatic digestion ([19]), similar to the enzymes applied for the isolation of HAMCs. Mice were monitored for signs of arthritis, as determined by hind paw swelling (mean thickness measured with calipers) and clinical scoring (0 = no swelling, 1 = slight swelling and erythema, 2 = pronounced edema, and 3 = joint rigidity) of each limb ([20]).

For histologic analysis, the paws were randomly collected by 2 independent experimenters (FO and EG-R) at the end of experiment (day 50), fixed in 4% buffered formaldehyde, decalcified, paraffin-embedded, sectioned, and stained with Masson-Goldner trichrome stain. Histopathologic changes were scored in a blinded manner, based on inflammatory cell infiltration, synovial proliferation/enlargement, articular cartilage destruction, and bone erosion as described previously ([20]).

In a separate set of experiments, mice were killed on day 35 after CII immunization, blood was obtained by cardiac puncture, and protein extracts were isolated by homogenization of paws (50 mg of tissue per ml of 50 mM Tris HCl, pH 7.4, 0.5 mM dithiothreitol, and 10 μg/ml of proteinase inhibitors). Mouse cytokine and chemokine levels in sera and protein extracts were determined by ELISA (BD PharMingen). Serum levels of IgG, IgG1, and IgG2a anti-CII antibodies were determined by ELISA (see Supplementary Methods, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38206/abstract). Synovial neutrophil infiltration was monitored by measuring myeloperoxidase (MPO) activity ([21]).

For adoptive transfer experiments, draining lymph node (DLN) cells obtained on day 35 from untreated or HAMC-treated arthritic mice were purified and restimulated for 72 hours in vitro with CII (20 μg/ml). After stimulation, CD4 cells were isolated (CD4+ >97%) and CD25+ cells were depleted (CD4+CD25– >99%) by immunomagnetic selection (Miltenyi Biotec), and on day 20, the cells were injected intravenously (5 × 106 cells) into mice with CIA. For in vivo CD25+ T cell depletion (>98%), DBA/1 mice were treated intravenously with 1 mg of anti-mouse CD25 antibody (clone PC61) 3 days before CII immunization.

Assessment of the T cell autoreactive response in mice with CIA

Single-cell suspensions (106 cells/ml) from DLNs were obtained 30 days postimmunization and stimulated in complete medium with heat-inactivated CII (25 μg/ml). Proliferation was determined at 72 hours by measuring 3H-TdR incorporation, and mouse cytokine contents in culture supernatants were determined at 48 hours by specific ELISAs. For intracellular cytokine analysis, DLN cells were stimulated for 8 hours with CII (25 μg/ml) and monensin (1.33 μM) and then stained with antibodies specific for mouse cytokines, as described above for human cells.

To determine the suppressive activity of HAMCs in vitro, DLN cells (105) isolated from arthritic mice at the peak of disease were stimulated with CII (25 μg/ml) in the presence of HAMCs (2 × 104). Proliferation and cytokine production were then determined. In some experiments, DLNs were cultured in conditioned medium from HAMC cultures diluted 1:1 with DMEM complete medium.

The number of Treg cells was determined in DLN cells isolated on day 30 postimmunization. Samples were incubated for 8 hours at 4°C with APC-labeled anti-CD4 monoclonal antibody (5 μg/ml; BD Bioscience), fixed/permeabilized as described above, stained for 30 minutes at 4°C with PE-labeled anti-FoxP3 (eBioscience) and Alexa Fluor 647–labeled anti-Helios (BioLegend) monoclonal antibodies (4 μg/ml), and analyzed in a FACSCalibur flow cytometer.

Analysis of the biodistribution of HAMCs

To trace the injected cells in vivo, HAMCs were labeled with 5,6-carboxyfluorescein succinimidyl ester (CFSE) before injection and were analyzed by flow cytometry in different tissue samples obtained on day 35, as described previously ([22]). In order to exclude the possibility of detecting HAMCs that had been phagocytosed by macrophages, HLA–A/B/C+ and CFSE+ cell analysis was determined in the CD11b-negative cell population. Moreover, the numbers of HAMCs in different tissues from the mice with CIA were quantified by real-time polymerase chain reaction (PCR) for human-specific Alu sequences, as previously validated for other MSCs ([23, 24]).

Models of sepsis, colitis, and encephalomyelitis

Sepsis was induced by ligation and puncture of the cecum of C57BL/6 mice ([21]). Colitis was induced by intrarrectal infusion of 2,4,6- trinitrobenzenesulfonic acid (TNBS) into BALB/c mice or by oral administration of Dextran sulfate sodium (DSS) to C57BL/6 mice ([21, 22]). Chronic experimental autoimmune encephalomyelitis (EAE) was induced by immunization of C57BL/6 mice with myelin oligodendrocyte glycoprotein fragment MOG33–55 ([25]). These models are described in detail in the Supplementary Methods (available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38206/abstract).

Statistical analysis

All results are expressed as the mean ± SD, except for the arthritis and EAE scores, which are expressed as the mean ± SEM. Statistical analysis was carried out with two-way analysis of variance, followed by Student's t-test in most experiments. The Mann-Whitney U test was used to determine differences in overall disease course scores between groups. P values less than 0.05 were considered significant.

RESULTS

HAMC-induced deactivation of CII-specific CD4+ T cells and synovial cells from RA patients

Recent studies have characterized the immunoregulatory properties of T cells and dendritic cells of amniotic-derived cells ([9]). In this study, we investigated the ability of expanded HAMCs to inhibit the activation of PBMCs isolated from RA patients. HAMCs decreased the proliferative response of PBMCs that responded positively to CII, a major component of the hyaline cartilage (Figure 1A). Moreover, HAMCs down-regulated the production of the Th1 cytokines interferon-γ (IFNγ) and TNFα, as well as that of IL-17, by CII-reactive T cells, whereas they induced the secretion of the antiinflammatory cytokine IL-10 (Figure 1A). Intracellular cytokine staining confirmed these findings in the CD4+ T cell population (Figure 1B) (additional data available upon request from the corresponding author).

Figure 1.

Deactivation of type II collagen (CII)–reactive peripheral blood mononuclear cells (PBMCs) from rheumatoid arthritis (RA) patients following treatment with human amniotic membrane cells (HAMCs [hAMCs]). A, PBMCs isolated from 5 RA patients were stimulated with CII in the presence or absence of allogeneic HAMCs. Proliferation and cytokine levels in the supernatants (interferon-γ [IFNγ], tumor necrosis factor α [TNFα], interleukin-4 [IL-4], IL-17, and IL-10) were determined 96 hours and 72 hours later, respectively. CII-activated PBMCs were separated from HAMCs by a semipermeable membrane in Transwells (basket drawn at hAMC) or were activated with CII in conditioned medium (CM) collected from HAMC cultures. CII-treated PBMCs from 3 healthy donors and from 4 CII-nonresponder (non-resp) RA patients were used as controls. B, PBMCs isolated from 5 RA patients were stimulated with CII in the presence or absence of HAMCs. After 14 hours, CD4 and intracellular cytokine levels were analyzed by flow cytometry. C, PBMCs isolated from 5 RA patients were activated with CII in the presence or absence of HAMCs, as well as in the presence or absence (none) of human IL-10–neutralizing antibody (Ab), isotype control antibody, indomethacin (indo), and/or NS398. Proliferation and IFNγ levels were determined 96 hours and 48 hours later, respectively. D, PBMCs isolated from 5 RA patients were stimulated with CII in the presence or absence of HAMCs. The expression of CD25 and FoxP3 in the CD4+ cell fraction was determined 96 hours later by flow cytometry. Results are expressed as the percentage of positive cells of total PBMCs. E, PBMCs isolated from 3 RA patients were stimulated with CII in the presence or absence of HAMCs. After 96 hours, T cells (control [Tcontrol] and HAMC-treated [TAMC]) were isolated from cocultures, rested, added to PBMCs that had been isolated from the same patient, and then stimulated with CII or anti-CD3 antibody. Proliferation and IFNγ levels were determined in duplicate cultures. Values are the mean ± SD. ∗ = P < 0.001 versus CII-activated CII responder PBMCs in A, B, and C, and for the indicated comparisons in D and E; # = P < 0.001 for the indicated comparisons.

Experiments in Transwells suggested a marginal dependence on cell-to-cell contact of the HAMC-mediated inhibition of CII-induced proliferation and IFNγ, IL-17, and TNFα production, while the induction of IL-10 seemed highly dependent on contact between HAMCs and PMBCs (Figure 1A). Flow cytometry findings indicated that upon contact, both HAMCs and PBMCs (CD3+ T cells and CD14+ monocytes) increased the production of IL-10 (data available upon request from the corresponding author). The major involvement of soluble factors produced by HAMCs was supported by the fact that conditioned media collected from HAMC cultures suppressed CII-induced PBMC activation (Figure 1A). We previously reported similar results with polyclonally activated PBMCs and identified products of cyclooxygenase 2 (COX-2) activity as major mediators ([26]). Consistent with this, indomethacin and NS398 (nonspecific and selective inhibitors of COX-2, respectively) impaired the immunosuppression induced by HAMCs, and blockade of both COX and IL-10 activities almost completely reversed it (Figure 1C).

Published evidence indicates that Treg cells play critical roles in self immune tolerance in RA ([27, 28]). The observed down-regulation of the CII-reactive Th1/Th17 responses, together with the elevation in the regulatory cytokine IL-10, prompted us to investigate the capacity of HAMCs to generate Treg cells. HAMCs increased the percentage of the CD4+CD25+FoxP3+ T cell population in CII-activated PBMCs (Figure 1D), but decreased the numbers of CD4+CD25–FoxP3– cells (data available upon request from the corresponding author). Noteworthy, T cells generated in the presence of HAMCs suppressed CII-specific T cell responses, but not T cell polyclonal activation (Figure 1E).

We next investigated the capacity of HAMCs to regulate the inflammatory response of resident cells of the synovial membrane of RA patients. Incubation with HAMCs decreased TNFα production and both collagenase and gelatinase activities in activated RA synovial membrane cells (Figure 2A). Again, the inhibitory effects were mediated mainly by soluble factors, especially COX-1/2–derived mediators and IL-10 (Figure 2B).

Figure 2.

Reduction of the inflammatory response in synovial cells from RA patients following treatment with HAMCs. A, Synovial membrane cells (SMCs) isolated from 6 RA patients were left unstimulated or were stimulated with lipopolysaccharide (LPS) (for TNFα determination) or TNFα (for collagenase/gelatinase determination) in the presence or absence of HAMCs. After 30 hours, TNFα levels and gelatinase and collagenase activities were determined in culture supernatants. B, TNFα- or LPS-activated synovial membrane cells and HAMCs were cocultured in the presence of neutralizing anti–human IL-10 antibody (Ab), isotype control antibody, and/or indomethacin (indo). Alternatively, synovial membrane cells were activated in the presence of conditioned medium (CM) collected from HAMC cultures. TNFα levels and collagenase activity were determined in culture supernatants. Results are expressed as the inhibitory activity of HAMCs on activated synovial membrane cells. Values are the mean ± SD. ∗ = P < 0.001 versus synovial membrane cells; # = P < 0.001 versus synovial membrane cells plus HAMCs. See Figure 1 for other definitions.

HAMC-induced amelioration of experimental arthritis

Having demonstrated the immunosuppressive activity of HAMCs on RA cells, we investigated their potential in the treatment of CIA, an experimental model of arthritis in mice that shares a number of histopathologic and immunologic features of RA ([29]). Two separate intraperitoneal injections of HAMCs to mice with signs of established clinical arthritis progressively attenuated the severity and incidence of the disease (Figure 3A). This therapeutic effect was also observed with allogeneic murine placenta-derived cells (Figure 3A), but not with apoptotic HAMCs or human fibroblasts (data not shown). Improvement in clinical settings correlated with a significant amelioration of the histopathologic signs and neutrophil infiltration in the joints of HAMC-treated mice (Figure 3B).

Figure 3.

Decreased severity of collagen-induced arthritis following treatment with amniotic membrane cells. A, Mice with established collagen-induced arthritis (CIA) were injected twice intraperitoneally (IP) with either phosphate buffered saline (CIA group [control]), HAMCs, or mouse placenta-derived cells (MAMCs [mAMCs]). Arthritis severity was assessed by clinical scoring and measurement of hind paw thickness on the indicated days. Numbers in parentheses are the arthritis frequency (percentage of mice with arthritis score >2 on day 45). Arrows indicate injection times. B, Histopathologic analysis was performed on hind limbs obtained on day 50 from naive mice, mice with CIA, and mice with CIA treated with HAMCs (left). Asterisks indicate inflammatory cell infiltrates; arrows indicate cartilage damage and bone destruction. Bars = 150 μm. Histologic scores for inflammation, cartilage damage, and bone erosions (right) in these tissues were also determined. Myeloperoxidase (MPO) activity, measuring neutrophil infiltration in the joints, was determined on day 35. C, HAMCs were labeled with 5,6-carboxyfluorescein succinimidyl ester (CFSE), injected IP into mice with CIA, and their presence in draining lymph nodes (DLNs), spleen, joints, lungs, and kidneys collected on day 30 postimmunization was determined by flow cytometry (left). Results are expressed as the number of CFSE+HLA–A/B/C+CD11b– cells per 100 mg of tissue. Human-specific Alu sequences in organs collected on day 35 postimmunization were also determined (right). Results are expressed as the percentage of HAMCs (of the total injected HAMCs). Similar results were obtained with samples collected on day 50. D, Levels of TNFα, IL-6, IL-12, IFNγ, IFNγ-inducible 10-kd protein (IP-10), RANTES, IL-1β, IL-10, and IL-17 in protein extracts of hind limbs obtained on day 35 were determined in the indicated groups of mice. Values are the mean ± SEM of 8–10 mice per group in A and the mean ± SD of 5 mice per group in B, C (right), and D, and 4 mice per group in C (left). ∗ = P < 0.01 versus control in A and P < 0.001 versus control in B and D; # = P < 0.01 for the indicated comparisons in A. See Figure 1 for other definitions.

To better understand the half-life and trafficking of the infused HAMCs, we injected CFSE-labeled HAMCs into mice with CIA. We detected the inoculated cells in the DLNs, spleen, and joints of the recipient mice at the peak of the disease (Figure 3C), indicating that the cells are immunologically tolerated by the recipient and target lymphoid and inflamed organs. We confirmed these results by quantifying human-specific Alu sequences in these organs (Figure 3C).

HAMC-induced down-regulation of inflammatory and Th1/Th17-mediated autoreactive responses in mice with CIA

We next investigated the mechanisms underlying the decrease in severity of CIA following HAMC administration. In patients with RA and in mice with CIA, progression of the autoimmune response involves the development of autoreactive Th17 and Th1 cells, their entry into the joints, and subsequent recruitment of inflammatory cells ([30]). The reduction in inflammatory infiltration observed in the limbs of HAMC-treated arthritic mice (Figure 3B) correlated with a significant decrease in the amounts of various cytokines and chemokines that are mechanistically linked to joint inflammation, but no appreciable decrease in the amount of IL-10 (Figure 3D).

As expected, DLN cells from mice with CIA showed marked CII-specific proliferation and effector T cells that produced high levels of IL-17 and Th1-type cytokines (IFNγ, IL-2, and TNFα) (Figure 4A). However, DLN cells from HAMC-treated mice proliferated much less, produced low levels of Th1 and Th17 cytokines, and produced high levels of IL-10 (Figure 4A). Determination of intracellular cytokines showed that treatment with HAMCs decreased the number of IFNγ-producing Th1 cells and increased the number of IL-10–producing CD4 cells in DLNs (Figure 4B). Interestingly, the suppressive effect of HAMCs was antigen-specific, because HAMC treatment did not affect proliferation or cytokine production by polyclonal stimulation (Figure 4A). Moreover, T cell priming was not significantly affected by HAMCs in mice with CIA, because the delayed-type hypersensitivity response was positively recalled by administration of CII in both untreated and HAMC-treated mice with CIA (data available upon request from the corresponding author). This suggests that HAMC treatment deactivated tissue-specific self-reactive Th1 cell clones. Consistent with this, HAMC treatment reduced serum levels of CII-specific IgG, particularly IgG2a, autoantibodies (Figure 4C).

Figure 4.

Impaired Th1- and Th17-mediated responses in mice with CIA following treatment with HAMCs. Mice with established CIA were injected IP with phosphate buffered saline (CIA group [control]) or HAMCs on days 24 and 29. A, Proliferation and cytokine production (IFNγ, IL-2, TNFα, IL-4, IL-17, and IL-10) by DLN cells isolated on day 30 and cultured in medium or stimulated with CII were determined. DLN cells cultured with 1 μg/ml of anti-CD3 plus 2.5 μg/ml of anti-CD28 antibodies (αCD3) were used to assess nonspecific stimulation (similar results were obtained with 2.5 μg/ml of concanavalin A). B, Numbers of cytokine-producing CD4 T cells in DLNs from control or HAMC-treated mice with CIA restimulated in vitro with CII were determined. C, Levels of CII-specific IgG, IgG1, and IgG2a in sera collected from control or HAMC-treated mice with CIA were determined on day 35. D, DLN cells isolated from mice with CIA on day 30 were cultured in the presence or absence of CII and in the presence or absence of HAMCs, conditioned medium collected from HAMC cultures, neutralizing anti-human or anti-mouse IL-10 antibodies (10 μg/ml), isotype control antibody, and/or indomethacin. Proliferation and IFNγ levels were determined 72 hours and 48 hours later, respectively. Values are the mean ± SD of 5 mice per group in duplicate cultures in A, 5 mice per group in B and C, and 5–6 mice per group in duplicate cultures in D. ∗ = P < 0.01 versus control in A, P < 0.001 versus control in B and C, and P < 0.001 for the indicated comparisons in D; # = P < 0.001 for the indicated comparisons in D. ND = not determined (see Figures 1 and 3 for other definitions).

To investigate whether HAMCs directly deactivated autoreactive Th1 cells, we cocultured HAMCs with DLNs from mice with CIA. HAMCs suppressed CII-induced T cell proliferation and IFNγ production, and conditioned medium from HAMCs mimicked this effect (Figure 4D). This suppressive effect was mainly dependent on the production of human IL-10 and mouse IL-10 and the activation of COX-1/2 (Figure 4D). Interestingly, DLN cells and HAMCs influence each other bidirectionally to produce IL-10 (data available upon request from the corresponding author).

HAMC-induced promotion of the emergence of antigen-specific Treg cells in mice with CIA

Treg cells confer significant protection against CIA by decreasing the activation and joint homing of autoreactive Th1 and Th17 cells ([31, 32]). Remarkably, HAMC-treated mice with CIA had significantly higher numbers of CD4+CD25+Foxp3+ Treg cells in DLNs than did untreated mice with CIA (Figure 5A) (additional data available upon request from the corresponding author). Moreover, T cells isolated from HAMC-treated arthritic mice function as suppressive Treg cells, because they prevented arthritis progression when transferred to mice with CIA (Figure 5B). Depletion of CD25+ cells from this population before transfer significantly abolished this therapeutic effect (Figure 5B), supporting the role of Treg cells.

Figure 5.

Induction of CD4+CD25+FoxP3+ Treg cells in mice with CIA following treatment with HAMCs. Mice with established CIA were injected IP with phosphate buffered saline (CIA group [controls]) or HAMCs on days 24 and 29. A, Numbers of CD4+CD25+FoxP3+ and CD4+CD25–FoxP3– T cells in DLNs obtained on day 30 were determined by flow cytometry. The percentages of CD4+CD25+Foxp3+ Treg cells in peripheral blood obtained on day 30 were also determined. B, DLN T cells from control (Tcontrol) and HAMC-treated (TAMC) mice with CIA were stimulated in vitro with CII and then injected into arthritic mice on day 20. HAMC-treated T cells were depleted of CD25+ cells before transfer. C, DBA/1 mice were depleted in vivo of CD25+ cells 10 days before CII immunization. HAMC treatment was initiated on day 22 postimmunization. The mean arthritis scores were determined on the indicated days (left). Numbers of CD4+CD25+FoxP3+ T cells per DLN were determined on day 30 in the same groups (right). D, Percentages of Helios+FoxP3+ and Helios–FoxP3+ cells in the CD4+ T cell population of DLNs isolated on day 30 were determined by flow cytometry. Values are the mean ± SD of 5 mice per group in A and D, 4 mice per group in C (left; mean scores), and 3 mice per group in C (right) and the mean ± SEM of 5 mice per group in B. ∗ = P < 0.01 versus control in A and D and for the indicated comparisons in C, P < 0.001 for the indicated comparison in B. See Figures 1 and 3 for other definitions.

Besides their central generation in the thymus, CD4+CD25+ Treg cells can be generated peripherally from CD4+CD25– T cells ([27]). To determine whether the HAMC-induced increase in CD4+CD25+ Treg cells during CIA was due to the expansion of the existing, naturally occurring CD4+CD25+ Treg cells or to newly generated Treg cells from the CD4+CD25– T cell population, mice with CIA were depleted of CD4+CD25+ T cells before HAMC injection. As expected, CD25+ T cell depletion prior to CIA induction resulted in an earlier onset and more severe disease than in the controls (Figure 5C). CD25+ T cell depletion did not affect the beneficial effect of HAMCs (Figure 5C), indicating its independence of the initial presence of Treg cells. Importantly, HAMC treatment recovered the number of DLN CD4+CD25+FoxP3+ T cells in CD25-depleted mice at the peak of disease (Figure 5C), suggesting that HAMCs could induce peripheral generation of CD4+CD25+ Treg cells from the CD4+CD25– compartment. This hypothesis was corroborated by the fact that the increased CD4+FoxP3+ T cell population induced by HAMC in the DLNs of mice with CIA was negative for Helios (Figure 5D) (additional data available upon request from the corresponding author), an Ikaros transcription factor family member that distinguishes thymus-derived Treg cells from inducible Treg cells ([33]). Moreover, HAMC treatment increased gene expression of FoxP3 in the DLNs of mice with CIA, while it did not affect the expression of Helios and neuropilin 1 (not shown).

HAMC-induced protection against experimental sepsis, colitis, and encephalomyelitis

Finally, we investigated whether the therapeutic action of HAMCs could be extended to other inflammatory/autoimmune disorders by using various experimental models of sepsis, colitis, and multiple sclerosis.

HAMC treatment protected against death caused by cecal ligation and puncture–induced diffuse peritonitis (Figure 6A), which mimics the clinical situation in patients with polymicrobial sepsis ([34]). Sepsis is characterized pathologically by overwhelmed inflammatory/immune responses that can lead to multiple organ failure and death. HAMCs decreased the systemic levels of inflammatory cytokines (Figure 6A) and the inflammatory infiltration into the peritoneal cavity, lung, liver, and intestine (data available upon request from the corresponding author).

Figure 6.

Protection against experimental sepsis, colitis, and encephalomyelitis following treatment with HAMCs. A, Sepsis was induced by cecal ligation and puncture (CLP). Mice were treated IP with phosphate buffered saline (PBS) (CLP group [control]) or HAMCs 4 hours later, and survival was monitored (left). Levels of mouse cytokines (TNFα, IFNγ, IL-1β, and IL-6) were determined in sera obtained 18 hours after cecal ligation and puncture (right). B, Colitis was induced by intracolonic administration of 2,4,6-trinitrobenzenesulfonic acid (TNBS). Mice were treated IP with PBS or HAMCs 12 hours after TNBS injection. Control mice received 50% ethanol (vehicle). Clinical evolution was monitored by body weight changes (left) and colitis scores (right). Macroscopic damage scores were determined on day 10. Histologic scores and colonic MPO activity were determined on day 3. HAMC treatment reduced TNBS-induced transmural inflammation, depletion of mucin-producing goblet and epithelial cells, disseminated fibrosis, focal loss of crypts, and infiltration of inflammatory cells in the lamina propria. C, Mice with chronic progressive experimental autoimmune encephalomyelitis (EAE) were treated IP with PBS (EAE group) or HAMCs beginning after disease onset (clinical score 1–2) (arrow). The mean clinical score, cumulative disease index (sum of daily scores from day 17 to day 50), and incidence of disease severity (none, mild [clinical score <2], moderate [clinical score 2–4], and severe [clinical score >4]) were determined on day 50 (left and both middle panels). Levels of cytokines and chemokines (IFNγ, IL-17, IL-12, IL-10, IP-10, and monocyte chemotactic protein 1 [MCP-1]) in protein extracts purified from brain and spinal cord at the peak of disease were determined by enzyme-linked immunosorbent assay (right). D, Mice with chronic progressive EAE were treated as in C. Proliferation (left) and cytokine production (IL-17, IFNγ, and IL-2) (right) were determined in DLN cells obtained at the peak of disease and incubated with medium or stimulated with the encephalitogenic antigen myelin oligodendrocyte glycoprotein (MOG33–55) or with a polyclonal stimulus (anti-CD3 plus anti-CD28 antibodies [αCD3]). Values are the mean ± SD of 10 mice per group in A (right), C (both middle panels), and D (left), 8 mice per group in B (left), 5 mice per group in B (right), and 5 mice per group in duplicate cultures in C (right) and D (right) and the mean ± SEM of 10 mice per group in C (left). ∗ = P < 0.001 versus control in A, versus TNBS in B, and versus EAE in C, and P < 0.01 versus EAE in D; # = P < 0.01 for the indicated comparisons in B and C. See Figures 1 and 3 for other definitions.

Moreover, HAMCs ameliorated colitis induced by TNBS or DSS. In these models, intestinal inflammation results from impairment of the intestinal epithelial cell barrier function, subsequent exposure of the submucosa to luminal antigens (bacteria and food), and activation of the Th1- and Th17-driven inflammatory responses ([35]). TNBS- and DSS-treated mice developed a severe illness, characterized by bloody diarrhea, rectal prolapse, and pancolitis and accompanied by sustained weight loss resulting in high mortality rates (Figure 6B) (additional data available upon request from the corresponding author). Mice treated with HAMCs had an increased survival rate, rapidly recovered body weight, improved colitis and histopathologic signs, and reduced levels of markers of inflammation in their colons (Figure 6B).

HAMC injection reduced the severity and incidence of chronic EAE (Figure 6C), a preclinical model of multiple sclerosis ([36]). This amelioration in clinical signs correlated with a decrease in inflammatory cytokines in spinal cord (Figure 6C) and with impairment in peripheral MOG-specific T cell responses (Figure 6D).

DISCUSSION

A desirable therapeutic approach for RA should prevent both inflammatory and autoimmune components of the disease. As examined in this study, we propose a novel cell-based therapeutic strategy for RA by using cells isolated from human amniotic membrane and then expanded in vitro. HAMCs exerted profound suppressive responses on CII-reactive T cells from RA patients by impairing CII-induced proliferation and production of Th1- and Th17-type cytokines by CD4 cells while stimulating the antiinflammatory/suppressive cytokine IL-10. Moreover, HAMC treatment inhibited the production of inflammatory cytokines and matrix-degrading enzymes by activated synovial cells involved in the destruction of the cartilage and bone in RA.

Important for their clinical translation is the fact that HAMCs maintain the immunosuppressive activity observed in vitro after their infusion in vivo. Thus, HAMCs provided highly effective therapy for CIA by strikingly reducing the 2 deleterious components of the disease, the Th1/Th17-mediated autoimmune and inflammatory responses. As a consequence, HAMCs were shown to reduce the frequency of arthritis, ameliorate arthritis symptoms, and prevent joint damage. The beneficial effect of AMCs on CIA was not restricted to a xenogeneic system since murine allogeneic placenta-derived cells were as efficient as HAMCs in reducing the clinical signs of arthritis. Moreover, by using various experimental preclinical models, we found evidence that the therapeutic action of HAMCs could be extended to other inflammatory/autoimmune disorders, such as sepsis, inflammatory bowel disease, or multiple sclerosis.

In this study, we decided to use cells isolated from whole amniotic membranes because of the ease of cell isolation and expansion ex vivo, resulting in higher amounts of cells. Although samples initially consisted of a mixture of epithelial and mesenchymal–stromal cells, enrichment of the mesenchymal population occurred after their expansion in culture. This could be due to the previously described epithelial-to-mesenchymal cell transition that amniotic epithelial cells undergo after various passages in culture ([37]). Both in vitro and in vivo studies have demonstrated that cells from either the epithelial or mesenchymal layers exert immunomodulatory properties and ameliorate disease with inflammatory bases, including lung and liver fibrosis and EAE ([15, 19, 38-41]). Interestingly, an important percentage of the HAMCs used in our study expressed CD106 (vascular cell adhesion molecule 1), which identifies a subpopulation of mesenchymal stem cells with unique immunoregulatory properties in various tissues, especially in the placenta ([42]).

There are several potential mechanisms for the effect of HAMCs on the effector phase of CIA. HAMC treatment down-regulated the proliferative response and the expression of Th1- and Th17-type cytokines in DLN cells. The inhibition of Th1/Th17 responses might be the result of a direct action on peripheral effector T cells, because DLN cells obtained from HAMC-treated animals were refractory in vitro to Th1 restimulation by antigen challenge. Similar to the characteristics of other MSCs ([21, 22]), we demonstrated the capacity of HAMCs to migrate to lymphoid organs in arthritic mice. Accordingly, HAMCs directly inhibited the in vitro activation of CII-reactive T cells from arthritic mice. The priming of the T cells, however, was not significantly affected by HAMC treatment, since the delayed-type hypersensitivity response to the immunizing antigen was preserved. These results suggest that the suppression of the Th1 response by HAMCs is not the result of a general, nonspecific immunosuppressive activity by these cells, but rather, it is induced by a robust, tissue-specific mechanism. In contrast to the effect on Th1-type cytokines, HAMC treatment increased the production of IL-10, but not IL-4, in CD4+ T cells from DLNs, which is evidence against the involvement of a predominant Th2 response.

Our findings also indicate that HAMCs migrate to inflamed joints and strongly reduce joint inflammation by down-regulating the production of a wide range of mediators involved in the pathogenic inflammatory response that causes joint damage. Reduction of chemokines could partially explain the absence of inflammatory infiltrates in the synovium of HAMC-treated mice. The decrease in mediators of inflammation could be the consequence of a diminished infiltration of inflammatory cells in the joint or the deactivation of the Th1-driven inflammatory response in the periphery. However, we present evidence of a certain capacity of HAMCs to switch off the inflammatory response of RA synovial cells, independently of its action on T cells.

The specific molecular mechanisms involved in the immunoregulatory activity of HAMCs are not yet fully understood. In samples from both RA patients and mice with CIA, the effects of HAMC seemed to depend primarily on the production of soluble factors by these cells, mainly mediators derived from COX-1/2 activation and IL-10, although cell-to-cell contacts between HAMC and immune cells are critical for the production of IL-10. This is consistent with the mechanisms reported for MSCs from other sources ([2-6]). Interestingly, COX-derived mediators have been shown to augment the immunosuppressive activity of IL-10 ([43]). Moreover, the fact that COX-derived mediators have cross-species activities and that human IL-10 is immunosuppressive of mouse cells ([44]) supports the involvement of these factors in vivo and could explain the therapeutic action of HAMCs in the xenogeneic model used in our study.

On the other hand, we demonstrated that the induction of FoxP3+CD4+CD25+ Treg cells could significantly contribute to the suppressive activity of HAMCs on T cells from RA patients and mice with CIA. Although other MSCs that induce Treg cells have previously been described ([3-6, 45-49]), our study is the first to demonstrate the generation of Treg cells by HAMCs. Importantly, HAMC-induced Treg cells specifically suppressed CII-driven responses, but not other responses, to polyclonal activation. This effect could contribute to the antigen-specific nature of the action of HAMCs on arthritis. It could also explain why delayed administration of HAMCs inhibited events in the inflammatory phase of established arthritis following the activation/differentiation of antigen-specific effector Th1 cells. Regarding the mechanism involved, we present evidence that HAMCs favor the new induction of peripheral Treg cells in vivo. However, it is still unknown whether HAMCs act directly on Treg cells or whether they act indirectly, generating tolerogenic DCs, as occurred with MSCs from other sources ([2-6, 46-50]). Indeed, cells derived from the mesenchymal layer of human amniotic membrane impair the maturation/activation of dendritic cells ([9, 11]).

We envision that once the benefits and risks associated with the injection of HAMCs are well-defined, the use of HAMCs opens new therapeutic perspectives for cell-based therapy for autoimmune/inflammatory diseases. The capacity of HAMCs to regulate various mediators of inflammation and to suppress Th1/Th17-type responses through the generation of Treg cells might offer a therapeutic advantage over existing therapies directed against a single mediator. It is noteworthy that the delayed administration of HAMCs ameliorated ongoing CIA, which fulfills an essential prerequisite for an antiarthritic treatment, and that initial treatment with HAMCs prevented the recurrence of disease within the observation period. Our findings also suggest that the immunosuppressive action of HAMCs is not major histocompatibility complex restricted and that the infused HAMCs are immunotolerated by the host, which is very convenient for a future clinical application of these cells in RA as an out-of-shield strategy. Indeed, large amounts of HAMCs can be easily obtained from human term placentas (a highly accessible/abundant tissue that is mostly discarded) of allogeneic healthy donors and rapidly expanded in vitro to generate a clinically effective doses.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Delgado had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Parolini, Gonzalez-Rey, Delgado.

Acquisition of data. Souza-Moreira, O'Valle, Magatti, Hernández-Cortés, Gonzalez-Rey, Delgado.

Analysis and interpretation of data. Parolini, Souza-Moreira, Gonzalez-Rey, Delgado.

Acknowledgments

We are grateful for the essential collaboration of the patients who participated in this study.

Ancillary