Interplay between mesenchymal stromal cells and the immune system after transplantation: implications for advanced cell therapy in the retina

Abstract

Advanced mesenchymal stromal cell-based therapies for neurodegenerative diseases are widely investigated in preclinical models. Mesenchymal stromal cells are well positioned as therapeutics because they address the underlying mechanisms of neurodegeneration, namely trophic factor deprivation and neuroinflammation. Most studies have focused on the beneficial effects of mesenchymal stromal cell transplantation on neuronal survival or functional improvement. However, little attention has been paid to the interaction between mesenchymal stromal cells and the host immune system due to the immunomodulatory properties of mesenchymal stromal cells and the long-held belief of the immunoprivileged status of the central nervous system. Here, we review the crosstalk between mesenchymal stromal cells and the immune system in general and in the context of the central nervous system, focusing on recent work in the retina and the importance of the type of transplantation.

Introduction

Mesenchymal stromal cells (MSCs) are a heterogeneous type of stem/progenitor cells nested in different tissues from the bone marrow (BM) to all vascularized tissues such as adipose tissue and perinatal derivatives, including the amniotic membrane, placenta and umbilical cord (Pittenger et al., 1999). MSCs were first described in the BM 40 years ago by Friedenstein and colleagues who observed that non-hematopoietic cells were able to migrate to injured areas and differentiate into mesenchymal tissues (Friedenstein et al., 1968). However, the differentiation potential of MSCs in vivo is minimal and therefore, MSCs are not considered to be true “stem cells”. The International Society for Cell Therapy established the minimal criteria to define human MSCs: i) the ability to adhere to plastic surfaces under standardized culture conditions; ii) > 95% of the cells must express the surface markers CD105, CD73 and CD90, and < 2% of them may express hematopoietic markers such as CD45, CD34, CD14 or CD11b, CD79 or CD19 and human leukocyte antigen (HLA)-DR; and iii) MSCs must be able to differentiate in vitro into at least three different lineages such as osteoblasts, adipocytes, and chondrocytes (Dominici et al., 2006). Despite these criteria, differences in phenotypes, mechanisms of action and secretome have been reported due to the heterogeneity of MSCs within individual niches and between species (Keating, 2012; García-Bernal et al., 2021; Norte-Muñoz et al., 2022a). To these intrinsic differences, we must add the impact of isolation and culture conditions (Ho et al., 2008; Ren et al., 2015; García-Bernal et al., 2021).

In the central nervous system (CNS), neural stem cells (NSCs) were the best choice as therapeutics because of their potential to replace the neurons lost in response to disease or trauma, and thus to regenerate the CNS (De Gioia et al., 2020). NSCs do not survive long and do not integrate into host tissues, precluding neuronal replacement, and their beneficial effects are linked to immunomodulatory/paracrine effects, properties also observed with MSCs. However, NSCs have shown teratogenic and tumourigenic potential in the host under some conditions, which has not been reported for MSCs. These findings directly affect the safety of NSC-based therapies, and MSC therapy has been considered as an alternative due to its advantages and proven beneficial properties. MSCs are relatively easy to isolate from various sources in adults, their immunogenicity is low, and they have potent immunomodulatory properties (Doorn et al., 2012; Moll et al., 2020), characteristics that facilitate their allogeneic or autologous/syngeneic transplantation. They have been successfully used in inflammatory pathologies such as graft-versus-host disease or Crohn’s disease (Boland et al., 2022; Ringdén et al., 2022), but also in the CNS, where the intended use of MSCs is to delay or halt neuronal death, i.e., to achieve neuroprotection, through their secretion of trophic factors and modulation of inflammation (Millán-Rivero et al., 2018). Indeed, preclinical and clinical studies have demonstrated beneficial effects of MSC treatment in complex CNS diseases including stroke, Alzheimer’s and Parkinson’s diseases, multiple sclerosis, spinal cord injury and retinal neurodegeneration (Andrzejewska et al., 2021; Norte-Muñoz et al., 2021; Di Pierdomenico et al., 2022; Van den Bos et al., 2022).

Despite these promising data, clinical trials of advanced cell therapy in the CNS are scarce and of limited success compared to the number of preclinical studies reporting beneficial outcomes in animal models of disease. The discrepancy between clinical and preclinical results may be a consequence of several reasons that may play a role in the therapeutic effect alone or in combination (Baranovskii et al., 2022). One reason may be the complex aetiology and physiopathology of human neurodegenerative diseases which may not be well modelled in animals. However, underlying all neurodegenerative diseases there are overlapping pathological processes that cause and amplify the damage, namely neuroinflammation, trophic factor withdrawal, oxidative stress, and bioenergetic and metabolic deficits (Wareham et al., 2022). Since MSCs possess potent anti-inflammatory and immunomodulatory properties and the capacity to secrete a variety of trophic factors (Burrows et al., 2013; Millán-Rivero et al., 2018; García-Bernal et al., 2022), it is logical to think that MSC treatment in animals and humans should render similar results.

Another reason may lay in the different approaches between experimental and clinical settings. In preclinical models, most researchers use xenotransplantation to test human MSCs, which in addition, are generally isolated from healthy young donors. In humans, transplants are mainly allogeneic and if possible autologous. In the latter, MSCs are isolated from the patient, and it has been shown that both the age and health status of donors have a high impact on the therapeutic properties of MSCs mainly due to cell senescence, reduced immunosuppressive capacity and regenerative potential, and a pro-inflammatory profile (Kizilay Mancini et al., 2017; Fafián-Labora et al., 2019). Of high importance is the type of transplant, which, as mentioned above, differs between clinical and preclinical studies. Recent works from our group have shown that each type of transplant (xenogeneic, allogeneic and syngeneic) has different neuroprotective properties and induces specific inflammatory responses in the CNS (Norte-Muñoz et al., 2021, 2022b; Di Pierdomenico et al., 2022; Garcia-Ayuso et al., 2022). These differential responses may explain the discordance between human and preclinical studies although more research is needed to have the full picture in each specific situation.

Finally, there are several weaknesses in preclinical studies with MSCs, which are mostly focused on the regenerative properties of MSCs without studying all the factors involved in the graft-host crosstalk, such as the host immune activation and MSC survival, integration or migration into the host tissue after transplantation (Wu et al., 2020; García-Bernal et al., 2022). As recent review articles point out, greater consensus protocols are needed to assess the effects of local and systemic administration (Carrancio et al., 2013), single and repeated injections (Plock et al., 2017), MSCs doses and tracking to evaluate their survival (Moonshi et al., 2022; Norte-Muñoz et al., 2022a).

In the first part of this review we will explore the interaction between the MSCs, the immune system and the host tissue. In the second part, we will discuss how the type of transplantation modulates the immune recognition of MSCs in the CNS and their therapeutic effect. In the second part, we will focus on adult BM-derived MSCs assayed in models of retinal neurodegeneration. Works reviewing the neuroprotective properties of MSCs derived from BM and/or from other niches in several models of neurodegeneration, including the retina, are numerous (Ortiz Nolasco et al., 1987; Volkman and Offen, 2017; Sugaya and Vaidya, 2018; Holan et al., 2021; Hu and Wang, 2021; Limoli et al., 2021; Rahbaran et al., 2022), and will therefore not be discussed here.

Search Strategy and Selection Criteria

We have conducted a MEDLINE/PubMed search of articles published between 2006 and 2022 using the following search terms: mesenchymal stromal cells OR mesenchymal stem cells AND immune system OR immune crosstalk AND central nervous system OR retina OR neurodegeneration. Only preclinical studies were included. Results were manually curated by reading titles and abstracts and, where appropriate, the methodology to determine the type of MSC used.

Interaction between Mesenchymal Stromal Cells and the Innate Immune Response

The immune response is balanced between two effector phases, the innate or primary response and the adaptive or secondary response. During the innate response, many humoral and cellular actors play a role, such as the complement system, cytokines, neutrophils, macrophages, dendritic cells (DCs), and natural killer (NK) cells. A complete understanding of how MSCs interact with these could be crucial in modulating their effects on host inflammation.

The complement system is a molecular signalling cascade that is triggered in three different ways (classical, lectin or alternative), leading to the formation of membrane attack complexes through the liberation of chemoattracting molecules and membrane-bound proteins that increase the recruitment of phagocytotic cells to the site of inflammation (Le Blanc and Davies, 2015). This molecular cascade rapidly reacts to microorganisms and apoptotic or infected cells to induce their elimination. However, the versatility of the complement to recognize different effectors such as pathogen-associated molecular patterns or damage-associated molecular patterns, and thus Toll-like receptor activation in phagocytes might results in a misguided contribution to inflammatory states (Pouw and Ricklin, 2021). The complement system has been linked to rejection (Hughes and Cohney, 2011), and clearance after systemic injection of the administered allogeneic MSCs (Li and Lin, 2012). Receptors for chemoattractant molecules, such as chemokines, have been identified on BM-MSCs that promote their migration into inflamed tissues (Le Blanc and Mougiakakos, 2012). MSCs constitutively produce factor H, an inhibitor of the complement cascade (Tu et al., 2010). However, in the blood, MSCs trigger the complement signalling through the three abovementioned pathways (Li and Lin, 2012), resulting in the formation of membrane attack complexes that directly damage MSCs.

Regarding cytokine production, MSCs respond to an infectious microenvironment by adopting a pro-inflammatory cytokine production, a phenotype associated with an early stage of infection and inflammation. In this state, they release macrophage inflammatory proteins (MIP-1α and -1β) and chemokines (chemokine (C-X-C motif) 9 and 10) (Waterman et al., 2010; Ulivi et al., 2014). In contrast, when exposed to inflammatory mediators, MSCs produce immunosuppressive proteins such as interleukin (IL)-10, transforming growth factor-β, indoleamine-2,3-dioxygenase and prostaglandin E2. This immunosuppressive response has been associated with the beneficial effects of MSC infusion in patients with severe inflammatory conditions such as sepsis and graft-versus-host disease (Bernardo and Fibbe, 2013; Chao et al., 2014; García-Bernal et al., 2022).

Neutrophils are the most abundant cells of the innate immune system. They are mainly found in the bloodstream and are the first immune cells to reach the inflammation site. MSCs not only drive neutrophils by producing attractant cytokines such as IL-6, IL-8, and macrophage inhibitory factor (MIF), but also regulate the expression of CD11b and leukocyte extravasation (Le Blanc and Mougiakakos, 2012; Lundahl et al., 2012). Neutrophils co-cultured with MSCs downregulate the expression of the pro-apoptotic Bax protein. Thus, MSCs prolong the neutrophil lifespan and thus the activity of these short-lived cells (Raffaghello et al., 2008).

BM-MSCs regulate stem cell homeostasis in the BM in symbiosis with macrophages, reflecting the direct interaction between both cell types (Pajarinen et al., 2019). Classically, macrophages have been divided into two groups: M1 and M2 phenotypes. For a simplified description of microglial/macrophage states and functions, the terms M1/M2 polarization will be used in this review. However, it is important to note that this dualistic nomenclature is being replaced by a more specific classification (Paolicelli et al., 2022; Wang et al., 2023). The M1 phenotype is associated with pro-inflammatory, phagocytic and bactericidal functions, whereas M2 macrophages act to prevent or resolve inflammation (Hesketh et al., 2017). The acquisition of a pro- or anti-inflammatory state depends on the molecular signals in the vicinity of the macrophages. MSC-educated macrophages have been reported in co-cultures with human BM-MSCs where a M2 or anti-inflammatory phenotype was induced with high levels of CD206 and CD163 molecules (Le Blanc and Mougiakakos, 2012). In addition, M2 polarization induced by MSCs has also been observed in resident macrophages in lungs where alveolar macrophages increased the production of anti-inflammatory IL-10 (Gupta et al., 2007), and in the brain, where human MSCs injected intracerebroventricularly reduced microglial phagocytosis, which is associated with a M2 state (Zanier et al., 2014). An important link to consider has been established between macrophage polarization and the modulation of T cell behaviour. Indeed, M2 macrophages have been associated with the induction of regulatory T cells (Tregs) and thus with the adaptive immune response (Melief et al., 2013).

Another interaction that points to a regulator role for MSC in the immune response is how they modulate DC functions. DCs are professional antigen presenting cells that directly activate T-cell responses. The expression of co-stimulatory molecules in DCs is reduced in response to MSC exposure, including HLA-II, CD80 and CD86, which in turn, results in a decreased stimulation of T cell proliferation and effector functions (Zhang et al., 2009; Du Rocher et al., 2012).

It is also important to talk about the role of NK cells as effectors of the innate response against virus-infected and stressed cells. In the absence of ligands for its inhibitory receptors (i.e. containing an immunoreceptor tyrosine-based inhibitory motif), NK cells could be activated, leading to the production of inflammatory cytokines (interferon-γ and tumor necrosis factor-α) and cytolysis. MSCs have multiple ligands on their surface to activate NK cells such as UL-16-binding proteins (i.e. ULBP3), poliovirus receptor and nectin-2/PVRL2. As a result, activated NK cells can kill MSCs. Nevertheless, in vitro studies have shown that interferon-γ exposure confers greater resistance to NK cell-mediated cytotoxicity on BM-MSCs (Spaggiari et al., 2006). Overall, interactions between NK cells and MSCs appear to be highly dependent on the presence of pro-inflammatory mediators in the tissue microenvironment. However, these observations need to be confirmed in vivo.

Interaction between Mesenchymal Stromal Cells and the Adaptive Immune Responses

Adaptive immune responses are mediated by T and B cells, which produce antigen-specific memory cells and antibodies to capture and neutralize antigens. Cytokines produced by innate immune cells in the milieu drive T helper cell proliferation to different pathways (Th1, Th2, Th17, and Treg).

Similar to the effect of MSCs on DC proliferation, MSCs reduce the expression of co-stimulatory molecules (CD28) and inhibit the proliferation of naïve T cells (Kitazawa et al., 2012). In contrast, MSCs have been shown to be stimulators of Treg differentiation through the production of anti-inflammatory cytokines, such as transforming growth factor β, which promote the expression of transcription factors (e.g., Foxp3) in T cells that induce their polarization into Tregs. The main functions of Tregs are to contribute to cellular immunotolerance and to prevent excessive immune responses as immunoregulatory cells (Svobodova et al., 2012). Transforming growth factor β can also induce T-cell differentiation into the Th17 phenotype. The Th17 response has been implicated with autoimmune diseases such as multiple sclerosis (Korn et al., 2009). In models of multiple sclerosis and rheumatoid arthritis the infusion of human BM-MSCs increased the differentiation of T cells into Tregs by inhibiting T cell differentiation to Th17. This modulation resulted in significant clinical improvements (Bai et al., 2009). Furthermore, human BM-MSCs have high levels of IL-17 receptors, whose binding to IL-17 present in the milieu triggers a negative feedback mechanism to reduce Th17 polarization (Huang et al., 2006).

In line with reduced Th1/Th2/Th17 proliferation, it has been reported that MSCs suppress the proliferation and differentiation of B cells into plasma cells by inhibiting the expression of some key genes such as IL-1 receptor-associated kinase 4 (Magatti et al., 2020) and thus the production and secretion of antibodies (Tabera et al., 2008; Asari et al., 2009).

The interplay between MSCs and the innate and adaptive immune systems of the host is summarized in Figure 1.

Figure 1.

Summary of MSC and immune system crosstalk.

Graphical scheme that summarizes the interaction of MSCs with innate immune cells (yellow arrows) and adaptive immune cells (green arrows). BAX: Bcl-2 associated to X gen; CD: cluster of differentiation; CXCL: chemokine (C-X-C motif); DCs: dendritic cells; HLA: human leukocyte antigen; IDO: indoleamine-2.3-dioxygenase; IL: interleukin; IRAK-1: interleukin-1 receptor-associated kinase 1; MIF: macrophage inhibitory factor; MIP: macrophage inflammatory protein; MSC: mesenchymal stromal cell; NK: natural killer; TGF-β: transforming growth factor-β; Treg: regulatory T cell; Th: helper T cell; PGE2: prostaglandin E2; ULBP3: UL16 binding protein-3. Partly generated using Servier Medical Art, provided by Servier, licensed under a Creative Commons Attribution 3.0 unported license.

The retina

The rodent retina is a widely used model to study the response of CNS neurons to injury and neuroprotection. The retina is easily accessible (compared to other CNS areas), its anatomy is very well known (Galindo-Romero et al., 2022), there are very well established models of neuronal degeneration aimed to retinal ganglion cells (RGCs) or photoreceptors (Villegas-Pérez et al., 1996; Wang et al., 2003; Marco-Gomariz et al., 2006), treatment administration can be intravitreal, subretinal, systemic or topical (Vidal-Sanz et al., 2015, 2017), and finally, both its anatomy and function can be studied in vivo using non-invasive approaches such as optical coherence tomography and electroretinogram, respectively (Rovere et al., 2015).

Photoreceptors and RGCs, the first and last neurons of the retinal circuit, respectively, are the retinal neurons most sensitive to degeneration. Their loss is the cause of several human diseases leading to visual impairment or vision loss such as glaucoma (RGCs) or retinal dystrophies (photoreceptors).

RGC axons form the optic nerve and therefore most models of RGC degeneration involve optic nerve injury (Vidal-Sanz et al., 2017), either direct (optic nerve crush or transection) or indirect (ocular hypertension models). Photoreceptor degeneration models include induced (García-Ayuso et al., 2011; Reisenhofer et al., 2017) or hereditary (García-Ayuso et al., 2010, 2015) models.

Influence of the transplantation type on the neuroprotective properties of MSCs

RGCs

In our first work with MSCs, we tested the neuroprotective potential of human MSCs derived from the Wharton’s jelly of the umbilical cord (hWj-MSCs) in a rat model of optic nerve crush (Millán-Rivero et al., 2018). We observed that the intravitreal administration of hWj-MSCs significantly delayed the loss of axotomized RGCs. Importantly, the same degree of neuroprotection was observed with hWj-MSCs isolated from different umbilical cords, although the afforded neuroprotection was transient in all cases. The neuroprotective effect was probably due to the increased expression of trophic factors such as vascular endothelial growth factor and brain-derived neurotrophic factor in the transplanted retinas. Interestingly, the levels of these beneficial proteins differed between intact and injured retinas, suggesting that hWj-MSCs tailor their secretome depending on the status of the host tissue. Beneficial effects in this model have also been reported using MSCs from different sources (Zaverucha-do-Valle et al., 2011, 2014; Mesentier-Louro et al., 2014, 2019; da Silva-Junior et al., 2021).

However, what we observed and reported for the first time was anatomical remodeling of the retina involving the formation of retinal folds and detachments where microglial cells were highly activated throughout all the retinal layers. Inside the folds, there were Iba1⁺ cells, which could be resident microglial cells or macrophages infiltrating from the choroid.

This result led us to believe that may be assaying human cells on rats, i.e., xenotransplant, might trigger the innate immune system causing the retinal abnormalities, even in the so called immunoprivileged environment of the CNS. Furthermore, we wondered whether this immune reaction had any effect on the elicited MSC therapeutic effect. In other words: does the type of transplantation influence the neuroprotective and neuroregenerative potential of MSCs? To answer this question, we tested MSCs from the same niche in the same model of neurodegeneration in the three transplantation settings. We changed to BM-derived MSCs because isolation of hWj-MSCs from rodents is not feasible, and to mice as a model to take advantage of the EGFP transgenic strain as a donor to track the administered cells. BM-MSCs were isolated from healthy human donors (hBM-MSCs) and C57BL/6 (CAG-EGFP) transgenic mice (mBM-MSCs). Intravitreal transplants were performed immediately after optic nerve axotomy in the following modalities: syngeneic (mBM-MSCs from C57BL/6 to C57BL/6 mice), allogeneic (mBM-MSCs from C57BL/6 to BALB/c mice) and xenogeneic (hBMSCs to C57BL/6 mice) (Norte-Muñoz et al., 2021).

Our comparative study showed significant differences in functional, anatomical and neuroprotective outcomes among the three transplantation settings. Functionally, xenotransplants impaired retinal function more than the axotomy alone or with allo- and syngeneic transplants. Anatomically, only xenotransplants resulted in long lasting retinal detachment and folds. We observed that RGC neuroprotection depended on both, the functional subtype of RGC (Nadal-Nicolás et al., 2023) and the transplantation modality. Vision forming RGCs were only rescued by the syngeneic transplant, whereas non-vision forming RGCs were also neuroprotected by the xenotransplants. Regarding axonal regeneration, only syngeneic transplants induced a significant axonal growth beyond the injury site compared to vehicle animals. Allogeneic transplants had no effect on the course of RGC loss or their axonal regrowth.

Thus, both, the type of transplant and the type of neurons to be rescued, modulate the neuroprotective potential of MSCs.

Photoreceptors

To date, we have studied the effect of BM-derived mononuclear cells (BM-MNCs) on photoreceptor survival in two animal models of retinitis pigmentosa that carry the same mutations as those observed in human patients. This cell fraction contains MSCs in a proportion in the range of 0.01–0.001% but also hematopoietic stem cells (CD34⁺) in a proportion in the range of 0.5–5% (Garcia-Ayuso et al., 2022). Our work follows the promising previous results published by scholars (Siqueira et al., 2011, 2015a, b). Theoretically, this fraction of BM cells may be able to differentiate into neurons (Tomita et al., 2002) and would neuroprotect the retina through their paracrine trophic effects (Garcia-Ayuso et al., 2022), as MSCs do (Millán-Rivero et al., 2018).

In our first work, we analyzed the effect of the injection of human BM-MNCs on photoreceptor survival. We found that human BM-MNCs administration had an anti-gliotic effect in the dystrophic retinas, but failed to rescue photoreceptors (Di Pierdomenico et al., 2020b). As the poor neuroprotective effect might be due to the nature of the xenotransplantation (Di Pierdomenico et al., 2020b), we decided to test syngeneic transplantation in the same animal models (Di Pierdomenico et al., 2022). Interestingly, we confirmed that rat BM-MNCs also had an anti-gliotic effect, and importantly, that the syngeneic transplant induced a greater photoreceptor survival (Di Pierdomenico et al., 2022), in contrast to what was found after xenotransplant. Since we were unable to demonstrate differentiation of the injected cells or their incorporation into the retina, it seems that the most plausible theory is that the beneficial effect of BM-MNCs in these models is achieved through a paracrine trophic effect. In addition, the anti-gliotic effect is very interesting as it could counteract the retinal remodelling that occurs secondary to photoreceptor degeneration (Garcia-Ayuso et al., 2018), and in which reactive gliosis plays an important role (Di Pierdomenico et al., 2020a). It is important to note that in both works we studied two routes of cell delivery: intravitreal and subretinal, with similar results. Therefore, we suggest that intravitreal injections may be more appropriate, as they present fewer risks (Garcia-Ayuso et al., 2022), although they are not innocuous (Di Pierdomenico et al., 2016).

Therefore, in agreement with the data on RGCs, the potential neuroprotective effect of BM-MNCs on photoreceptors also depends on the type of transplantation.

The response of the intact retina to MSCs depends on the type of transplant

Most papers focus on the therapeutic effect of MSCs in a given model, but to our knowledge none study their effect on healthy tissue, which is surprising since the healthy tissue response is an important control to gauge the actual therapeutic outcome. For instance, what if the transplant itself causes neuronal loss or functional impairment?

Thus, using the same experimental design as described above, human or mouse BM-MSCs were injected into the vitreous of otherwise healthy intact retinas. Inner and outer retinal function was impaired in both xeno- and allografted retinas up to 21 days (the latest time point we studied), indicating that RGCs and photoreceptors were functionally affected by both transplants. As expected, xenotransplanted retinas were more functionally impaired than allotransplants. Interestingly, syngrafts did not have a negative effect on the retinal function.

Retinal folds and detachment were observed in the three transplantation settings, being always present in xenotransplanted retinas, followed by a lower incidence in allografted retinas, and scarcely observed in syngeneic transplants.

As for gliosis, Müller cells, a retina-specific radial macroglia, and microglial cells were always hypertrophied in the retinal folds, regardless of the transplantation setting. However, in xeno and allografts, but not in syngrafts, Müller cells and microglial cells showed anatomical signs of activation in the whole retina.

Microglial expression of CD45, signifying activation, increased in microglial cells located in the ganglion cell layer after allotransplants. In xenotransplants, Iba1⁺ CD45⁺ activated microglial cells were observed in all retinal layers. In fact, using flow cytometry we observed that the number of CD45⁺ cells doubled in allotransplanted and tripled in xenotransplanted retinas compared to intact or to syngrafted retinas. Furthermore, only after xenotransplants, there were Iba^– CD45⁺ cells present in the retina, indicative of lymphoid or myeloid recruitment.

This glial activation was pro-inflammatory, as the pro-inflammatory cytokine tumor necrosis factor-α was upregulated in xeno- and allotransplants, but not in syngrafts. However, in syngrafted retinas, we observed a significant upregulation of IL-6, a dual cytokine with anti and pro-inflammatory functions that has also been associated with axonal regeneration (Hirota et al., 1996; Fischer, 2017). The latter function may explain why only singeneically administered BM-MSCs induced axonal regeneration after axotomy (Norte-Muñoz et al., 2021).

Importantly, all these functional, molecular, anatomical and glial changes, did not induce the loss of RGCs, which is promising for BM-MSC-based therapies in the CNS.

MSC survival after transplantation

One important aspect to have into account in cell-based therapies is the fate of the transplanted cells. Are they rejected by innate immune cells? How long do they survive? Do they integrate into the tissue? By identifying hBM-MSCs with anti-human mitochondrial antibodies, and mBM-MSCs by their expression of green fluorescent protein, we reported that intravitreally injected MSCs form a mesh in the vitreous overlying the retina, without integrating into the retinal layers (Norte-Muñoz et al., 2021, 2022b).

BM-MSCs were present in the vitreous at 21 days but not at 90 days. We found that they were surrounded by Iba1⁺ CD45⁺ cells (macrophages and/or microglial cells) that migrated into the vitreous and surrounded them. This process is quicker after allogeneic than after xenogeneic or syngeneic transplantation (Norte-Muñoz et al., 2022b). In other words, the type of transplantation determines the survival of MSCs and therefore the duration of their therapeutic properties.

Regarding human or rat BM-MNCs, they were detected up to 15 days after transplant and were not found at later times. Their survival was not affected by the route of administration (intravitreal or subretinal) and, as observed with BM-MSCs, there were no differences between xeno- and syngeneic transplanted BM-MNCs (Di Pierdomenico et al., 2020b, 2022; Garcia-Ayuso et al., 2022).

Conclusion

MSC transplantation activates the innate immune system in the CNS, causing gliosis, microglial activation, tissue remodelling, functional impairment and an altered cytokine profile. The host-graft crosstalk varies according to the type of transplant being xenotransplants the most harmful, followed by allotransplants. Syngeneic transplants, although not completely innocuous, do not alter neuronal functionality and are, therefore, the best for neuroprotection and induction of axonal regeneration.

The mechanisms underlying the crosstalk between the retina and the immune system with the three types of cell transplantation are still not fully understood. Much research remains to be done to unravel the complex graft-host interaction in order to develop better strategies to manage both, modulate adverse and enhance beneficial responses, and thus provide patients with safe and successful MSC therapy, although at present the translation of these findings into clinical practice currently appears to be distant.

The small number of articles currently published on the relationship between the immune system and MSCs in the CNS is the main limiting factor of this review. However, this is a flourishing area of research and light will be shed on the intricacies of this dialogue.