Are stem cells a potential therapeutic tool in coeliac disease?

Abstract

Despite the growing understanding of its pathogenesis, the treatment of coeliac disease is still based on a lifelong gluten-free diet that, although efficacious, is troublesome for affected patients, and a definitive cure is still an unmet need. In this regard, the development of new chemical- and biological-derived agents has often resulted in unsatisfactory effects when tested in vivo, probably because of their ability to target only a single pathway, whilst the immunological cascade responsible for tissue injury is complex and redundant. The advent of cellular therapies, mainly based on the use of stem cells, is an emerging area of interest since it has the advantage of a multi-target strategy. Both haematopoietic and mesenchymal stem cells have been employed in the treatment of refractory patients suffering from autoimmune diseases, with promising results. However, the lack of immunogenicity makes mesenchymal stem cells more suitable than their haematopoietic counterpart, since their transplantation may be performed in the absence of a myeloablative conditioning regimen. In addition, mesenchymal stem cells have been shown to harbour strong modulatory effects on almost all cells involved in immune response, together with a potent regenerative action. It is therefore conceivable that over the next few years their therapeutic use will increase as their biological interactions with injured tissues become clearer.

Introduction

In the gastrointestinal tract, fast renewal of epithelial cells [1] and rapid trafficking of immune cells in the lamina propria [2] continue throughout an individual’s lifetime. It is therefore believed that gut mucosa has a remarkable capacity for tissue repair when damage occurs. However, there are some pathological conditions in which this process is severely impaired, such as in coeliac disease (CD). This is caused by a dysregulated immune response towards both dietary, the gluten components [3], and self, the enzyme tissue transglutaminase [4], antigens having the small intestine as the target organ, which develops in genetically susceptible individuals [5, 6]. Its clinical picture ranges from an asymptomatic or oligosymptomatic condition in most patients, to a severe malabsorption syndrome [5]. In spite of a better understanding of its pathogenesis, treatment is still based on a gluten-free diet (GFD), as originally proposed by the Dutch paediatrician Doctor Willem Dicke [7]. Although this treatment guarantees recovery from both clinical symptoms and intestinal damage in almost all patients, it seriously affects the quality of life, since its stringency and lifelong duration cause chronic distress while segregating patients in a sort of social apartheid [8]. Moreover, GFD does not fully protect patients from developing complications, i.e., refractory CD and enteropathy-associated T-cell lymphoma (EATL) [9]. These, although very rare, represent a clinical challenge since no standardized treatments are available and the outcome is poor [9]. A further reason why the development of an alternative and effective therapy is still an unmet need is the lack of an animal model [10]. On the other hand, any proposed drug carries a number of potential adverse effects that rule out their use as a substitute for the safe GFD regimen. Among the different approaches under investigation [11], none have displayed certain efficacy when tested in vivo, probably because they have a single target, whilst the immunological cascade responsible for tissue injury is complex and redundant. Therefore, there is a high need for alternative strategies, and stem cells have always fascinated both clinicians and researchers for their possible therapeutic applications [12, 13]. These cells are characterized by the unique ability to renew themselves indefinitely while differentiating into specialized cell types. However, considering ethical concerns about the use of embryonic stem cells, only the haematopoietic stem cells (HSCs)—CD34⁺—and mesenchymal stem cells (MSCs)—CD34⁻—are regarded as the best candidate for clinical application. Below, we summarize their state-of-the-art use in CD, and highlight some related mechanisms that may have clinical implications in the near future.

Haematopoietic stem cells

In the last decade, HSC transplantation has emerged as an effective treatment to offer to patients with chronic, severe immune-mediated diseases refractory to conventional therapies [14]. In the gastrointestinal tract, HSCs have been shown to play a prominent role in mucosal healing [1] thanks to their ability to give rise to all the resident cell populations, including pericryptal myofibroblasts [15, 16], vascular cells [16, 17], and epithelial cells [18]. In CD, an increased traffic of CD34⁺ HSCs with respect to healthy controls has been reported in peripheral blood of patients with active disease [19]. Although it did not correlate with the levels of anti-transglutaminase antibodies, or the severity of the histologic damage, it is conceivable that mobilization of hematopoietic precursors from the bone marrow might supply the increased cell death observed in coeliac mucosa [20]. In addition, the evidence of patients receiving allogeneic or autologous HSC transplantation for haematological malignancies who experienced relief of the concomitant autoimmune disease [21] clearly highlighted the possibility that HSC transplantation also induces immune tolerance [22]. It is now believed that the achievement of a maximum immune ablation may be of benefit by clearing the body of committed lymphocyte clones, while HSC transplantation takes months or years for a complete immune reconstitution, thus pathogenic clones may, in theory, never re-emerge [21]. In other words, HSC transplantation is considered an intensive treatment aimed not only at regenerating intestinal mucosa, but also at ‘resetting’ the immune system by inducing specific unresponsiveness to normally tolerated antigens [22]. However, the occurrence of relapse, mainly in the autologous setting, suggests that this strategy is far from being definitively curative, but rescues those refractory patients with severe disease and poor quality of life [21]. Moreover, the relatively common and potentially serious side effects, together with the non-negligible risk of death, make this option unacceptable as first-line therapy in non-life-threatening enteropathies [23].

Experience in coeliac disease

In this clinical context, autologous HSC transplantation has been employed as rescue therapy in a few selected patients suffering from complicated forms (Table 1), such as refractory CD and EATL [24–26]. The former is defined by the persistence of both malabsorption syndrome and intestinal villous atrophy for more than 6–12 months despite a strict GFD [27]. Moreover, two distinct conditions have been identified on the basis of the characteristics of intraepithelial lymphocytes (IELs), named type I and type II. The hallmark of the latter consists of an accumulation (>20 %) of IELs displaying genetic abnormalities [28], clonal rearrangements of the T-cell receptor [29], and an aberrant phenotype [30]. Among the molecules involved in refractoriness, interleukin (IL)-15, a cytokine found overexpressed in coeliac mucosa, plays a pivotal role in promoting T cell cytotoxicity [31]. In mice engineered to constitutively express IL-15 in the gut mucosa, in fact, the oral administration of food antigens stimulated the expansion of interferon (IFN)-γ-producing T cells, rather than a tolerogenic subset [32]. An additional mechanism by which IL-15 may cause tissue damage is also through the induction of resistance of effector T cells to the action of CD4⁺CD25⁺ transcription factor Forkhead box (Fox) P3⁺ regulatory T cells which, although increased in coeliac mucosa, are inhibited in their suppressive activity, thus leading to loss of tolerance to gluten and, possibly, self-antigens [33–35]. Finally, gluten-driven chronic stimulation is also associated with the possible emergence of an aberrant IEL-clonal population through an IL-15-dependent mechanism [31]. Specifically, via activation of the Janus kinase (JAK)-3 and signal transducer and activator of transcription (STAT)-5, IL-15 stimulates expression of the anti-apoptotic B-cell lymphoma-extra large (Bcl-xL) protein, which rescues IELs from apoptosis and allows their massive accumulation and malignant progression [31]. In this regard, indeed, type II refractory CD has a poor prognosis because of the frequent transformation into an overt EATL, whilst type I has a more favourable outcome [27]. The management of both conditions relies on a combination of nutritional support and immunosuppressive or biological therapies, whose use is based on non-controlled studies in small cohorts of patients and personal experience [27]. However, despite some clinical improvement, the lack of a substantial histologic recovery and the dismal prognosis of type II refractory CD have resulted in the consideration of autologous HSC transplantation [23]. To date, two studies recruiting a total of 20 patients were published with promising results in terms of reduction of the percentage of aberrant T cells and recovery of body weight and mucosal lesions, thus ameliorating the long-term outcome of these patients with respect to those who were unable to undergo this intensive treatment [24, 25]. Nevertheless, a few patients still developed EATL, thus indicating that the (pre)neoplastic aberrant IEL population is highly resistant to the conditioning regimen applied. Accordingly, no improvement of either the clinical picture or the outcome of patients suffering from EATL was observed when using this therapeutic strategy [26]. Specifically, three out of four patients died from relapse within a few months after autologous HSC transplantation, and only one patient underwent prolonged and complete remission (32-month follow-up). Therefore, this treatment seems of benefit in refractory CD, mainly type II, by rescuing most patients from the risk of developing an overt lymphoma, but unsatisfactory in EATL.

Table 1.

Haematopoietic stem cell transplantation

References	N	Disease duration (years)	Age (years)	Sex (F/M)	Mobilization	Conditioning	T-cell depletion	Maintenance	Remission response
Uncomplicated coeliac disease (allogeneic)
Kline et al. [36]	1	9	12	1/0	–	Cy + busulfan	NA	Cyclosporine + methotrexate	Yes
Hoekstra et al. [37]	1	1	11	1/0	–	Cy + ATG	NA	NA	Yes
Ciccocioppo et al. [38]	2	4–16	4–17	1/1	–	Busulfan thiotepa + fludarabine	No	Cyclosporine	Yes
Ben-Horin et al. [39]	1	28	29	0/1	–	NA	NA	NA	Yes
Complicated coeliac disease (autologous)
Al-toma et al. [24]	7	Median 2 (range, 0–8)	Median 61.5 (range, 51–69)	3/4	G-CSF	Fludarabine + melphalan	Yes	No treatment	Remission for 1; relapse for 6
Tack et al. [25]	13	Median 0.5 (range, 0–2.5)	Median 59 (range, 43–68)	7/6	G-CSF	Fludarabine + melphalan	NA	NA	Remission at >2 years in 10; relapse in 3
Al-toma [ 26 ]	4	Median 1.5 (range, 0–5)	Median 65 (range, 60–69)	2/2	G-CSF	Fludarabine + melphalan	NA	No treatment	Remission for 1; relapse in 3

NA not available, Cy cyclophosphamide, ATG antithymocyte globulin, G-CSF granulocyte colony-stimulating factor, F/M female/male, n number of cases

Conversely, resolution of non-complicated CD (Table 1) was described in two patients who underwent histocompatibility locus antigen (HLA)-identical matched-sibling HSC transplantation for concomitant haematological diseases [36, 37], although reversal of mucosal lesions was proven in only one [37]. We described two further patients who underwent successful allogeneic HSC transplantation for Thalassemia major, where the introduction of a gluten-containing diet did not cause the re-appearance of clinical, serological, and histological markers of the disease [38] with a follow-up now approaching 7 years. Moreover, the immunological study showed normalization of both the number of regulatory FoxP3⁺ T cells and the cytokine profile, while gliadin stimulation did not elicit a proliferative T-cell response when tested in vitro. These data, together with the complete donor chimerism observed in all cases, suggest that recapitulation of the immune system ontogeny occurring after allogeneic HSC transplantation may lead to induction of immune tolerance to oral antigens, and then to the definitive cure of CD. Consistent with this evidence was a recent case report of a child with CD who underwent allogeneic HSC transplantation for chronic myelogenous leukaemia, where the lack of memory response of both circulating and mucosal CD4⁺ T-cells towards tissue transglutaminase, gliadin, and deamidated gliadin was clearly shown after resumption of a gluten-containing diet for 5 years after the transplantation procedure [39]. Interestingly, the immuno-FISH of duodenal mucosa demonstrated the presence of a dichotomous lymphocyte-epithelial chimeric intestine, with the epithelial cells deriving from the host, and the lymphocytic lineage cells from the donors.

In conclusion, due to the scanty clinical experience, the mechanisms by which HSC transplantation induces immune tolerance to oral antigens and suppress aberrant cells are still not fully understood. Furthermore, a question arises about the type of stem cells responsible for the therapeutic effects, since all cases examined had received an entire population of bone marrow-derived stem cells, thus also including CD34⁻ cells.

Mesenchymal stem cells

Biological properties

After the initial identification in adult bone marrow by Friedenstein and colleagues in 1976 [40], MSCs were isolated from a variety of additional tissues, including muscle connective tissue, adipose tissue, foetal tissues, placenta, and umbilical cord blood [41]. These cells were also defined as ‘stromal’ since they substantially contribute to the creation of the HSC niche by providing both the structural support and growth factors needed for the ordered development and differentiation of the lympho-haematopoietic system [42]. Because of the lack of a specific marker, the proposal of the Mesenchymal and Tissue Stem Cell Committee of the International Society for Cellular Therapy [43] suggested the following three minimal criteria for MSC identification:

• MSCs have to be plastic-adherent under standard culture conditions;

• MSCs must express CD105, CD73, and CD90, and lack expression of CD45, CD34, CD14 or CD11b, or CD19 and HLA-DR surface molecules;

• MSCs must differentiate to osteoblasts, adipocytes, and chondroblasts in vitro.

However, this cell population is endowed with the capability to differentiate into several cell types not only of the mesoderm lineage but also of the endoderm, such as hepatocytes, and ectoderm, such as neural precursors [44, 45]. Such immense plasticity coupled with a great expansive potential under optimal conditions ex vivo, the ability to home to inflamed sites and to repair injured tissues, as well as the absence of ethical controversies, make them very attractive for cell-based therapy [46]. As far as the intestine is concerned, MSCs have been shown to populate the injured regions and improve lesions to the same extent as HSCs [47], and to favour epithelial barrier integrity [48] in experimental models of colitis. Moreover, thanks to the lack of expression of both HLA class II and costimulatory molecules, such as CD40, CD80, and CD86, and the low level of HLA class I antigens [49, 50], MSCs enjoy a degree of immune privilege [51] allowing their transplantation across HLA barriers without rejection [52]. Finally, MSCs have been described affecting the multiple arms of the immune system through their anti-inflammatory and immunomodulatory properties [53], thus contributing to the homoeostatic mechanisms of immune tolerance [54]. In a mouse model of colitis, MSCs were found to preferentially localize in mesenteric lymph nodes after systemic infusion, and favour the expansion of a regulatory subset of T-cells, which display suppressive capacity on T-helper 1 effector response [55]. It is therefore conceivable that MSCs may prime naive immune cells towards a tolerogenic profile by creating an appropriate microenvironment, called a ‘quasi-niche’, through the secretion of a great array of molecules [56]. In fact, MSCs seem to exert their protective effects in the absence of differentiation into specific end-organ cells, but rather via the paracrine release of protective substances, such as growth factors, antioxidants, and immune-modulatory molecules. Among them, those playing a prominent role in inducing immune tolerance were indoleamine 2,3-dioxygenase (IDO), prostaglandin E₂ (PGE₂), nitric oxide (NO), and insulin-like growth factor binding proteins [57]. A further molecule shown to possess potent immunomodulatory functions was the soluble HLA-G, through several mechanisms which include the induction of apoptosis in CD8⁺ T cells [58], suppression of NK cell lytic activity [59, 60], inhibition of T-cell proliferation [61], and dendritic cell maturation [62], while inducing regulatory T-cell expansion [62].

In humans, MSCs have been employed in a number of pathological conditions [46], with striking results obtained in terms of resolution of steroid-refractory acute intestinal graft-versus-host disease [63]. Following these findings, together with the demonstration of efficacy of MSC transplantation in a mouse model of autoimmune disease [64], we rescued a 61-year-old woman with a life-threatening malabsorption syndrome due to steroid-refractory adult autoimmune enteropathy with two systemic infusions of autologous bone marrow-derived MSCs [65]. The treatment was feasible, safe, and effective with disappearance of both the serologic and histological hallmarks of the disease within a month, thus successfully overcoming the critical stage. Moreover, an increase in mucosal FoxP3⁺ Tregs and secretory immunoglobulins was observed. Altogether, these evidences provide a strong rationale to consider the use of MSCs as a new therapeutic strategy in CD also, since these two conditions share several pathogenic mechanisms other than the characteristic mucosal lesions. What sets CD apart is the precise determination of both the foreign antigens [3] and the auto-antigen [4], as well as the most relevant genetic susceptibility [5, 6]. However, the lack of a well-established animal model [10] and the absence of notable side effects of the GFD make it hard to find alternative and competitive treatments. Below, we try to consider all the potential mechanisms by which MSCs may interact and overcome the inflammatory cascade responsible for gluten-dependent enteropathy.

Immunological basis for the use of MSC in coeliac disease

The mucosal damage in CD is driven by a complex network of cells and molecules which synergise in sustaining chronic inflammation [66]. Specifically, both innate and adaptive immune responses are involved in causing intestinal lesions in active CD and the lymphomagenesis process in complicated forms. We analysed these components (see Figs. 1, 2, respectively) separately in order to set the scenario for a putative MSC action.

Fig. 1.

Possible MSC effects on innate immunity in coeliac disease. Mesenchymal stem cells (MSC) are known to possess immunoregulatory and regenerative properties on both the epithelial and immune cells involved in the abnormal innate immunity in coeliac disease. These include the potential to suppress both activated intra-epithelial lymphocytes and natural killer cells; to inhibit interleukin (IL)-15 secretion and function; to preserve the epithelial barrier; to rescue enterocyte from apoptosis while protecting crypt stem cells; and to inhibit dendritic cell maturation and activation, thus avoiding an efficient antigen presentation to T lymphocytes (for comments see the text). AJ adherens junctions, DC dendritic cell, HGF hepatocyte growth factor, HLA histocompatibility locus antigen, IL interleukin, IFN interferon, MSC mesenchymal stem cell, M-CSF monocyte-colony stimulating factor, NK natural killer cell, PGE ₂ prostaglandin E₂, TJ tight junctions, VEGF vascular endothelial growth factor

Fig. 2.

Putative MSC effects on adaptive immunity in coeliac disease. Mesenchymal stem cells (MSC) have been shown to modulate the adaptive immune response through their interaction with all the immune cells involved in coeliac disease pathogenesis, including T cells (a), regulatory T cells (b), B cells (c), and endothelium (d). The inhibitory effect of MSC depends on both cell–cell contact and soluble factors (for comments see the text). CXCR chemokines, COX cyclooxygenase, Fox transcription factor Forkhead box, G gliadin peptides, HGF hepatocyte growth factor, HLA human leukocyte antigen, Ig immunoglobulin, IDO indoleamine 2,3-dioxygenase, IFN interferon, IL interleukin, IEL intraepithelial lymphocyte, MSC mesenchymal stem cell, NK natural killer cell, NO nitric oxide, NOS nitric oxide synthase, PGE ₂ prostaglandin E₂, Th T helper, Treg regulatory T cell, STAT signal transducer and activator of transcription, TG2 tissue transglutaminase, TGF transforming growth factor, TNF tumour necrosing factor

Innate immunity

Epithelial barrier

In the intestine, the epithelial barrier guarantees the correct nutrient absorption while protecting the body from penetration of dangerous foods and infective agents [67]. This is governed by a complex interplay among cells and molecules whose anatomical and functional integrity allows maintenance of a selective permeability [68]. Both tight and adherens junctions take place in this system by sealing up the adjacent enterocytes. A disruption of both complexes was found in active CD as a consequence of an altered phosphorylation mediated by pro-inflammatory cytokines [69]. Moreover, an impaired epithelial barrier function is thought to play a crucial role in the early step of CD pathogenesis [70] through a gliadin-dependent activation of the zonulin pathway in the enterocytes leading to cytoskeleton reorganisation, tight junction opening and, in turn, increased intestinal permeability [71]. Recently, larazotide acetate, a molecule that should prevent opening of the intestinal epithelial tight junctions by antagonizing zonulin action, has been proposed as a new therapeutic tool in CD [72]. However, in two clinical randomized, double-blind, placebo-controlled trials recruiting a total of 270 patients on a GFD, the gluten-dependent increase of intestinal permeability was not affected by this treatment, even though an improvement of gastrointestinal symptoms was observed [73, 74]. Remarkably, MSC systemic infusion was shown to contribute in maintaining epithelial barrier function in a mouse model of colitis by reassembling claudins, the apical-most proteins of the tight junctions [48]. In addition, the intravenous injection of human bone marrow-derived MSCs proved of benefit in limiting the radiation-induced intestinal damage in NOD/SCID mice, probably by preserving the integrity of the stem cells at crypt levels, and controlling the exaggerated apoptotic rate of the enterocyte [75]. Finally, in both in vitro [76] and in vivo [77] models of necrotising enterocolitis, MSCs were able to rescue intestinal epithelium from hypoxia–reperfusion injury, via the secretion of IL-6, hepatocyte growth factor (HGF), and vascular endothelial growth factor (VEGF), and the inhibition of the cognate interaction between the receptor Fas with its ligand, thus avoiding the activation of the intracellular mediators caspase-3 and -8, which ultimately leads to enterocyte apoptosis [77]. It is amazing to point out that the same pathway has been found to trigger the exaggerated enterocyte apoptosis responsible for villous atrophy in active CD [20].

Intraepithelial lymphocytes and natural killer cells

Intraepithelial lymphocytes form a large population of oligoclonal T lymphocytes sparsely distributed along the small intestinal mucosa, where they exert a pivotal role in intestinal barrier function [78]. In CD, however, their chronic activation and expansion lead to epithelial injury and, in a small subset of patients, to refractoriness and lymphomagenesis [31, 79]. The main subset responsible for tissue damage is the CD8⁺TCRαβ⁺ cells, which trigger increased enterocyte apoptosis through the production of IFN-γ [80], perforin, and granzymes [81]. Moreover, these cells are also activated by the large amounts of IL-15 released by the enterocyte upon gluten stimulus, which, in turn, upregulate the expression of NK receptors, i.e., CD94 and NKG2D, on their surface, and that of their ligands, i.e., HLA-E and MICA, on enterocytes [82]. As a result, the enterocytes become the target of a double cytolytic effect by IELs. At this point, it should be underlined that MSCs are endowed with the ability to inhibit NK-cell cytotoxicity by downregulating the expression of the NKp30, NKp44 and NKG2D receptors and by suppressing the IFN-γ production [83, 84]. Moreover, although MSCs have been shown to be susceptible to recognition and lysis by IL-2 activated NK cells [84, 85], they may skip this attack when stimulated by IFN-γ [86], the cytokine found overexpressed in CD mucosa [80], thus suggesting that MSCs might find a favourable microenvironment for their action in coeliac mucosa. In addition, MSCs have also been demonstrated to be capable of inhibiting the proliferation of IL-15-stimulated NK cells [84], a useful property in view of a possible therapeutic application of MSCs in CD. In this regard, by considering that IL-15 favours the emergence of aberrant IELs by reprogramming a T-cell precursor into NK cells [30], and that it plays a pivotal role in orchestrating either the gluten-dependent mucosal damage in active CD or the gluten-independent attack in refractory CD, while promoting the lymphomagenesis process [31], our recent finding of an inhibitory action of MSCs on IL-15 production by gliadin-specific T cells paves the biological basis for the clinical use of this cell population in both complicated and uncomplicated conditions [87]. Finally, MSCs were also shown to inhibit the proliferation of invariant NKT (iNK T, Vα24⁺Vβ11⁺) and γδT (Vδ2⁺) cells [88], which have recently been recognized as taking part in CD pathogenesis [89].

Antigen-presenting cells

Following the recognition of the HLA-DQ2 and DQ8 haplotypes as the main genetic predisposing factors to CD [90], it subsequently became evident how these molecules expressed on the surface of dendritic cells have a high physicochemical affinity for deamidated gliadin peptides, thus favouring their presentation to CD4⁺ T cells [91]. Antigen-presenting cells, therefore, are recognized as playing a crucial role in triggering the inflammatory cascade leading to mucosal injury [92]. Recently, several studies have discovered the ability of MSCs to affect both the phenotype and function of dendritic cells, by inhibiting the differentiation of CD34⁺ precursors [93] and CD14⁺ monocytes [94], while interfering with the activity of mature DC [95]. In greater depth, the impairment of monocyte differentiation into dendritic cells was attributed to either a block in the G₀ to G₁ phase of the cell cycle [96] or the production of inhibitory soluble factors, including PGE2 [97], IL-6 and monocyte-colony stimulating factor [98]. Moreover, exposure of mature dendritic cells to MSCs or their soluble factors resulted in a shift towards a less mature phenotype characterized by a decrease in expression of HLA class II, CD80, CD86, CD40, and CD83 molecules, an increase in their endocytic activity, and reduced production of IL-12 [98–100]. As a consequence, the resulting cells were severely impaired in their ability to stimulate proliferation of allogeneic T cells. This abortive maturation was associated with the expression of a regulatory profile, characterized by the secretion of large quantities of IL-10, which seemed fundamental for the delivery of the immunosuppressive effect. It is conceivable, therefore, that this shift towards a more tolerogenic profile of dendritic cells may help to avoid T-cell activation, possibly contributing to subdued inflammation in CD as well.

Adaptive immunity

Continuing the dissection of the immunological mechanisms leading to gluten-dependent enteropathy, we now analyse how both T- and B-cell functions may be modulated by MSCs (Fig. 2).

T-cell response

T cells were the first immunologic cell type shown to be influenced by MSCs. In this regard, biological studies have indicated that MSC-driven immunomodulation is associated with suppression of pro-inflammatory T-helper 1 response while rebalancing the T-helper 1/2 ratio towards the T-helper 2 profile [101]. Moreover, MSCs have been shown to strongly inhibit the proliferation of T cells upon a wide array of non-specific stimuli [102–104]. Remarkably, in a recent work [87], we demonstrated a robust inhibitory effect of MSCs on gliadin-specific T-cells, in terms of suppression of the antigen-specific proliferative response while increasing their apoptotic rate, inhibition of the production of those pro-inflammatory cytokines, i.e., IFN-γ and IL-21, directly involved in tissue injury [105, 106], and expansion of a regulatory subset. The effects of MSCs on T cells appeared largely dependent on the activity of IDO, an inducible enzyme that causes deprivation of the amino acid tryptophan needed for T-cell growth and activation [107]. However, the role of additional mediators cannot be excluded: in fact, Aggarwal and Pittenger [108] observed that MSC-derived PGE2 is involved in skewing an IFN-γ polarized, inflammatory T helper-1 into an IL-4-secreting T helper-2 response. Another candidate molecule is the NO produced in large amounts by the inducible isoform of its synthase upon the contact of MSCs with activated CD4⁺ or CD8⁺ lymphocytes [104]. Interestingly, although B cells are not capable of inducing the NO synthase in MSCs, they are just as sensitive to inhibition by the NO as T cells, an effect dependent on the blocking of phosphorylation of STAT5 [109], which results in cell cycle arrest at the G₀–G₁ phase [110, 111]. Finally, the non-classical HLA class I molecule, HLA-G, has recently been shown to be involved in immunomodulation of MSCs through its interaction with inhibitory receptors on dendritic cells, NK, and T cells [112]. Remarkably, full expression of its soluble isoform needs MSC contact with alloreactive T cells and IL-10 stimulation [60]. HLA-G, in turn, inhibits the cytolytic activity of NK and CD8⁺ T cells, shifts the allogeneic T-cell response to a T helper-2 cytokine profile, and induces the expansion of CD4⁺CD25^highFoxP3⁺ regulatory T cells [60]. Additional molecules that have been reported to be significantly upregulated in licensed MSCs and whose neutralization at least partially reverses their inhibitory activity include transforming growth factor (TGF)-β [57, 113, 114], HGF [114], cyclooxygenase-2 [115], IL-10 [116], and heme-oxygenase-1 [117]. Last but not least, MSCs have been found to downregulate the production of tumour necrosing factor (TNF)-α [118], a harmful cytokine that, at variance with uncomplicated CD, has been found to be overexpressed in the mucosa of refractory CD patients [119]. Altogether, these findings support the hypothesis that MSCs might be of some efficacy in both complicated and uncomplicated forms of CD by exerting a thorough action on almost all the abnormal immunological mechanisms involved in inducing and sustaining the mucosal damage. Moreover, they may be specifically licensed to release their ultimate effects directly in the gut mucosa by the peculiar microenvironment [120, 121], since MSCs have been recognized to express various chemokine receptors that make them prone to homing to sites of inflammation and injury [122, 123].

Finally, by considering the genetic and immunological similarities between CD and type 1 diabetes [124], it is worth noting that MSC infusion was proved to be of benefit in preventing and ameliorating the autoimmune attack against the insulin-producing β cells of the pancreas, in both the streptozotocin [125] and non-obese [126] diabetic mouse models. In addition, MSC infusion conferred protection against the transfer of diabetes by T cells isolated from treated mice, suggesting that regulatory T cells are involved in the MSC-mediated suppression of the disease [127], while in in vitro experiments, MSCs were shown to suppress diabetes-specific T cell proliferation and IFN-γ production [128], exactly as they do on gliadin-specific T cells [87].

Regulatory T cells

A pivotal mechanism needed to achieve and maintain peripheral tolerance to dietary, microbial and self-antigens relies on the action of CD4⁺CD25⁺FoxP3⁺ T regulatory cells [129]. In this regard, there is widespread agreement indicating the ability of MSCs in favouring the expansion of regulatory T cells either in vivo [130] or in vitro [108, 131] from both naive and memory T cells [132, 133]. In our clinical experience, an increase of regulatory FoxP3⁺ T cells in both peripheral blood and intestinal mucosa was invariably found in patients suffering from either Crohn’s disease [134] or autoimmune enteropathy [65] who underwent successful treatment with autologous MSCs. Interestingly, MSCs also express the FoxP3 molecule, although at a variable extension depending on culture conditions, cell growth, and maturity [135]. Particularly, MSCs expressing high levels of FoxP3 display a greater suppressive activity in mixed lymphocyte reaction than those with low expression [135]. Furthermore, MSCs also share with regulatory T cells the ability to secrete the pleiotropic cytokine TGF-β, which has relevant modulating effects [103]. However, it has recently been found that regulatory T cells are increased in active CD, but less effective since IL-15 made lymphocytes resistant to their suppressive function [34]. Our recent finding of an upregulation of TGF-β in the co-cultures of gliadin-specific T cells with MSCs, while inhibiting IL-15 [87] reinforces the hypothesis that MSC treatment may be able to orchestrate a shift from a prevalently pro-inflammatory response towards an anti-inflammatory and tolerogenic profile in this specific clinical setting.

B-cell response

High levels of serum class A immunoglobulins specific for both the foreign antigen/s, the gluten peptides [3], and the auto-antigen, the enzyme tissue transglutaminase [4], are the hallmarks of active CD, and are produced by intestinal plasma cells [5]. To date, only a few in vitro studies have focused on the effects of MSCs on B-cell function. It has been shown that B-cell development is partly dependent on the close interaction of B-cell progenitors with MSCs, and that soluble factors produced by MSCs can interfere with B-cell activation, proliferation, and differentiation into plasma cells [136]. Moreover, similarly to the effect on T-cells, that the effect (it should be eliminated ‘that’ or ‘the effect’) on B-cells also consists of a block in the G₀/G₁ phases of the cell cycle, thus interfering with immunoglobulin production [137, 138]. However, contrasting evidence was obtained in two in vitro studies where a stimulatory effect of MSCs on immunoglobulin production was found [139, 140]. In this regard, it should be mentioned that MSCs display their inhibitory effect on B cells only following cell-to-cell contact and in the presence of IFN-γ [141], which stimulates IDO activity, thus causing deprivation of the amino acid tryptophan also needed for B-cell proliferation [110]. Finally, an in vivo study on a mice model of autoimmune myasthenia gravis clearly showed that repeated MSC infusions were of benefit in both critically reducing the levels of specific anti-acetylcholine receptor antibodies and significantly improving the functional deficits [142]. The specific ability of suppression plasma cell immunoglobulin production appeared dependent on the action of MSC-derived chemokine ligands CCL2 and CCL7, thus leading to inhibition of the STAT3 [143]. In human intestinal diseases, the only evidence comes from a patient suffering from adult autoimmune enteropathy who was rescued with intravenous infusions of MSCs, where a transient disappearance of the specific anti-enterocyte antibodies was observed [65]. No data on the effects of MSCs on antigen-specific humoral immune response in CD have been available until now.

Open questions

Considering the ethical concerns about the use of embryonic stem cells and the invasiveness and the risks of HSC transplantation, MSCs seem to be the best candidate for clinical application mainly in non-life-threatening conditions such as non-complicated CD. As reviewed, MSCs exhibit a wide range of immunomodulating and regenerative properties that target virtually any mechanisms involved in CD pathogenesis. However, MSC-based cellular therapy cannot be considered an immunological panacea, and several fundamental issues should be further addressed. One such question is the influence of the mucosal microenvironment on their effects and the hierarchy by which MSCs display their action in this specific pathological setting. An additional important question is for how long the administered MSCs can maintain their function within the host. Nonetheless, their persistence in the diseased organ does not seem to be a crucial prerequisite for their efficacy, since via cell–cell interactions and the release of a broad range of bioactive molecules, MSCs form a ‘quasi-niche’ where they display their action, without the need for significant engraftment and differentiation [144]. Another caveat is the potential for infused cells to lose their functions, or simply the inability to travel directly to the site where they are expected to operate. To date, the most common route of MSC therapy has been intravenous, and various evidence has shown that MSCs injected in this way become trapped in the lungs because of their large size [145]. Furthermore, as regards which source of MSCs seems better suitable for clinical application, apparently no gross biological differences have been found with respect to the tissue of origin [146]. However, the potential limitation of donors and the invasiveness of the procedures applied when using bone marrow and adipose tissue as sources of MSCs have given rise to the need to obtain MSCs from different sources with unlimited donors and to ease access, such as birth-associated tissues, including human umbilical cord blood [147], Wharton’s Jelly [148], amniotic fluid [149], amnion [150], and placenta [151]. These last may also display the best proliferative and regenerative potential [152], together with a high tolerogenic activity, since they guarantee the foeto-maternal immune-privileged status to ensure a successful pregnancy [113, 153]. Finally, further points, such as the standardization of production, the route, the dose/s and the interval of administration, as well as the exact therapeutic targets, should be fine-tuned through a translational approach, before the therapeutic use of MSC in CD becomes a real prospect.

Abbreviations

CD

Coeliac disease

EATL

Enteropathy-associated T cell lymphoma

Fox

Transcription factor Forkhead box

GFD

Gluten-free diet

HSC

Haematopoietic stem cell

HGF

Hepatocyte growth factor

HLA

Histocompatibility locus antigen

IDO

Indoleamine 2,3-dioxygenase

IFN

Interferon

IL

Interleukin

IEL

Intraepithelial lymphocyte

JAK

Janus kinase

MSC

Mesenchymal stem cells

NO

Nitric oxide

PGE₂

Prostaglandin E₂

STAT

Signal transducer and activator of transcription

TGF

Transforming growth factor

TNF

Tumour necrosing factor

VEGF

Vascular endothelial growth factor