Mesenchymal stem cells, more accurately known as mesenchymal stromal cells (MSCs), are one of the most widely investigated therapeutic cell types, owing to their ease of accessibility and expansion from tissues free of ethical concerns, as well as their immunomodulatory capacity in various preclinical disease models (1–3) . MSCs can be sourced from the stroma of almost all tissues, but most commonly from bone marrow (BM), adipose tissue (AD) and umbilical cord (UC) (4, 5). MSCs are contained within a heterogeneous population expressing CD105, CD73 and CD90 while lacking CD45, CD11b, CD19 and HLA-DR. They are characterized by their adherence to plastic and ability to differentiate into osteoblasts, adipocytes and chondrocytes in vitro (6).

With increasing clinical translation, however, this minimal set of criteria is now recognized as inadequate for defining MSCs. It does not reflect MSC potency, which is largely based on broad immunomodulatory properties (7), rather than their self-renewal or multipotential capacity (8, 9). It does not address changes in marker expression levels or biological properties due to culture expansion and cell manufacturing processes (10), including cryopreservation and freeze-thawing (11). Additionally, it does not identify the risk profile of the MSC product, particularly with regard to hemocompatibility of intravenously (IV) delivered cells and their potential to trigger adverse thromboembolic events (12). Thus, proposed amendments to the criteria have included functional potency assays based on the immunomodulatory activity of MSCs (13), and profiling of the hemocompatibility of the diverse array of MSC products now in clinical use (14).

The preclinical efficacy of MSCs in various unrelated conditions such as graft-versus-host disease (GvHD), Crohn’s disease, kidney transplantation, myocardial infarction, stroke, diabetes, acute respiratory distress syndrome (ARDS), multiple sclerosis, and brain and spinal cord injury, mainly relates to immune regulation (15–20). In vitro studies have shown that MSCs can modulate adaptive and innate immune responses. MSCs suppress T cell proliferation, cytokine response (e.g. IFN-γ production) and cytotoxic activity in response to antigen-specific stimuli (21–23) whist promoting regulatory T cells (Tregs), via their production of soluble factors ( Figure 1 ) such as nitric oxide (NO), indolamine 2,3 dioxygenase (IDO) and transforming growth factor-beta (TGF-β) (24, 30–32). MSCs can also downregulate the cytokine responses of innate immune cells, including dendritic cells (DCs) and monocytes, via the expression of prostaglandin E2 (PGE₂) (33). How these in vitro findings relate to their mode of action remains to be clarified, given the complexity of immune responses in vivo and the short persistence of MSCs following infusion.

To date, studies have investigated different routes of MSC administration in order to maximise therapeutic efficacy. The different administration routes result in variations in MSC survival. IV infusion has been the most commonly used and studied method for MSC delivery because it is convenient, minimally invasive and reproducible (4). However, MSCs administered via the IV route are entrapped in the lung and rapidly cleared, with few traces detected in other tissues (34, 35). Studies using various MSC detection techniques, including MSCs constitutively expressing fluorescent proteins or luciferase, or labelled with fluorescent dyes or radioactive tracers, showed that viable MSCs are detected in the lung within 24 hours of IV administration but not at 72 hours (35–37). The decrease in detectable viable cells is directly proportional to the increase in dead cells in the lungs, indicating that IV-infused MSCs undergo cell death (37). Activation of caspase 3, a hallmark of apoptosis, in MSCs further indicates that MSCs undergo programmed cell death following their entrapment in the lung (38, 39).

The survival of IV-delivered MSCs can also be compromised if cells trigger the instant blood-mediated inflammatory reaction (IBMIR) due to incompatibility with blood (40). Expression of secreted and cell surface immunogenic factors (e.g. tissue factor (TF)/CD142) vary across MSC tissue sources and cell manufacturing conditions, including freeze-thawing and culture passaging (12, 14). These factors can activate the innate immune system and trigger the coagulation and complement cascades, which limit MSC engraftment and efficacy, but also increase the potential for adverse thromboembolic events (41).

Other injection routes, including intraperitoneal (IP), subcutaneous (SC), and intramuscular (IM), which bypass the lung, have also been examined. Prolonged detection of MSCs has been observed following IP and SC injection (42, 43). Following IM injection, a dwell time of up to 5-month was observed (44). Other studies have administered MSCs directly into the diseased tissue, for example, intratracheal administration in models of lung inflammation or intrathecal administration in models of spinal cord injury or neuroinflammatory disease (45–48). In these studies, MSCs are directly exposed to an inflammatory environment, which can influence MSCs survival and therapeutic efficacy (26, 49). Pre-conditioning of MSCs via hypoxia or serum deprivation for instance, can promote cell survival when subsequently exposed to an ischemic environment (50–52). Although the quantity, duration and the type of stress insult matter, overall, excessive stress will likely predispose MSCs to cell death (26).

In some disease settings, dead or dying MSCs and their associated ‘by-products’ can contribute to therapeutic efficacy (38, 39, 53–55) raising additional questions about their mode of action. How are MSCs killed in different microenvironments? How does the dying process contribute to therapeutic efficacy in different disease settings? Importantly, how do IV administered MSCs dying in the lung exert anti-inflammatory effects in distal organs in disease settings that seemingly do not involve the lung?

The molecular pathway that induces the death of MSCs in vivo is unclear and complicated by pre-existing disease, inflammatory cell infiltrate and the presence of different pathogens (26). The stress signals from the inflammatory microenvironment can trigger the apoptosis of MSCs. In a mouse model of GvHD, the presence of elevated numbers of cytotoxic CD8⁺ T cells and CD56⁺ natural killer (NK) cells in the lung caused MSC apoptosis (38). Settings that do not involve cytotoxic cell infiltrate likely involve other mechanisms. We recently showed that disabling the intrinsic (mitochondrial) pathway of apoptosis in MSCs prevented caspase 3 activation in the lung shortly after IV injection, indicating that MSCs were predominantly killed via the intrinsic pathway in non-GvHD settings (39). Activation of coagulation and complement by infused MSCs has also been shown to damage and reduce viability of MSCs (40, 56). Stronger activation of these proteolytic cascades is demonstrated by freeze-thawed (without culture recovery) and high-passage MSCs compared with fresh culture-derived, minimally expanded cells. This is due to variations in expression of immunogenic triggers, which may subsequently impact their in vivo therapeutic function (11).

In general, apoptosis is an ordered event that creates a transient immunosuppressive microenvironment via the release of anti-inflammatory mediators, including IL-10, TGF-β, CCR5, annexin-A1 and thrombospondin-1 ( Figure 1 ) (28). Besides these secreted factors, cells can undergo a pre-apoptotic stage known as autophagy when they sense danger or stress signals from the microenvironment (57). By culturing MSCs in a stressed environment, autophagic MSCs can be pre-engineered to secrete immunomodulatory factors, such as TGF-β, to regulate T cell proliferation (58). Interestingly, MSCs have been shown to communicate with damaged cells via bidirectional transfer of mitochondria to increase mitochondrial biogenesis and rescue the cellular function of damaged cells (27). In preclinical models of myocardial infarction and respiratory disorders, mitochondrial transfer has been shown to contribute to the therapeutic effects of MSCs ( Figure 1 ) (27, 59–61). MSCs can transfer mitochondria to macrophages via tunnelling nanotubules, cell fusion or extracellular vesicles (EV), which influence the macrophage function to modulate inflammation (59, 62). Whilst mitochondrial transfer has yet to be demonstrated in apoptotic MSCs, apoptotic bodies, which are a distinct type of extracellular vesicles formed when apoptotic cells disassemble into fragments (63), may also contribute to therapeutic efficacy upon engulfment by macrophages (64). Thus, MSCs are susceptible to cell death post-administration, which contributes to the anti-inflammatory effects of MSC therapy (10, 39).

Rapid clearance of apoptotic debris by phagocytes is essential in maintaining body homeostasis. Known as efferocytosis, this physiological process ‘silently’ removes apoptotic cells, inducing peripheral tolerance and avoiding inflammation. Phagocytic cells, including macrophages and monocytes, play a critical role in MSC therapy, as demonstrated in disease models where depletion of macrophages with clodronate liposomes renders MSC therapy ineffective (33, 36, 38). Phagocytes that have engulfed apoptotic cells, including MSCs, display a regulatory phenotype characterized by upregulation of PD-L1, IDO, COX2 and CD206, increased production of IL-6, IL-10, TGF-β and PGE₂, and decreased production of pro-inflammatory cytokines such as TNF-α and IL-12 (29, 37, 65). Phagocytes that have engulfed apoptotic MSCs can suppress T cell proliferation, downregulate CD4⁺ T cell activation and promote Foxp3⁺ Treg generation (29, 37). The inhibition of COX2 abrogated the immunosuppressive capacity of efferocytic monocytes that had engulfed apoptotic MSCs, highlighting the importance of the PGE₂/COX2 axis in this immunosuppressive function (29).

Since MSC apoptosis and the subsequent response of immune cells to this process contributes to their immunomodulatory effects, an outstanding question is whether viable MSCs are still required for MSC therapy, or can pre-inactivated MSCs be used as an alternative?

Several studies have investigated the efficacy of in vitro induced apoptotic, dead or inactivated MSCs. Treatment outcomes can be influenced by the type and duration of stimulation used to inactivate the cells, and also the disease setting (26). Apoptotic MSCs (39, 53, 54), but not dead MSCs (36, 66) could inhibit lung inflammation. In certain disease settings, such as LPS-induced sepsis, inactivated or dead MSCs could replicate the immunomodulatory effects, as pre-inactivation enhanced the phagocytosis of MSCs (55, 67). In other settings, non-viable MSCs were ineffective or did not fully replicate the effects of viable MSCs (33, 42, 67). For instance, in GvHD, apoptotic MSCs had to be administered at a much higher dose than viable MSCs to achieve comparable therapeutic benefits (38). It is plausible that inflammatory diseases driven by innate immune or phagocytic cells are more likely to benefit from treatment with inactivated MSCs, while suppression of T cell responses require MSCs with an active secretome (67). Importantly, the type and stage of cell death at the time of MSC administration are key considerations, as apoptosis, but not necrosis or lytic cell death, induces anti-inflammatory responses (68).

Although MSC apoptosis and the subsequent host phagocytic response contribute to immunomodulation, there remains much to learn about the full mechanisms of MSC therapy. As MSC apoptosis and efferocytosis occurs in the lung post-infusion, how does this dampen inflammation in other organs or tissues? In vitro studies have shown that MSCs can regulate immune cell function via their paracrine activity, but such molecules must have a relatively long half-life and broad biodistribution for this to be a plausible mechanism in vivo. Immune responses are initiated and maintained in the SLOs, which play a key role in immune regulation during health and disease. Thus, our knowledge of how MSC therapy modulates immune responses would be improved by considering the known function of SLOs, their role in clearance of apoptotic cells and gaining a better understanding of the effects of MSCs in these organs.

Primary lymphoid organs, also known as central lymphoid organs, are the sites for the development and maturation of leukocytes. Primary lymphoid organs include the bone marrow and thymus (69–71). Lymphocytes (a class of leukocytes, e.g. T and B cells) generated in primary lymphoid organs then seed the SLOs where they initiate adaptive immune responses. SLOs include lymph nodes (LNs), spleen, tonsils, adenoids, Peyer’s patches, and mucosal-associated lymphoid tissues (MALT) (71). The localisation of lymphocytes in SLOs maximizes their interaction with foreign antigens that drain to the SLOs via blood or lymphatics. Antigen presenting cells, such as DCs, can also transport antigens to SLOs and activate lymphocytes there. Once activated, these lymphocytes undergo expansion and differentiate into effector or memory cells to provide antigen-specific responses. This review will focus on the spleen and LNs as the major SLOs.

Essentially all tissues undergo programmed cell death, known as apoptosis, evident through the constant turnover of cells. Under homeostatic conditions, apoptotic cells rarely accumulate in tissues due to the efficient efferocytosis by tissue phagocytes. The clearance of apoptotic cells is linked to an anti-inflammatory response that also induces immunological tolerance (108, 109). The precise apoptotic cell machinery and phagocytic components involved in apoptotic cell-induced immunosuppression are reviewed extensively elsewhere (109–112). However, it is clear that the efficient clearance of apoptotic cells can be attributed to the redundancy in the mechanisms of apoptotic cell recognition.

Splenic and LN macrophages are known to participate in the efferocytosis of apoptotic cells and maintenance of immune tolerance. TZMs and TBMs in the LN are known to engulf apoptotic debris silently without stimulating T cells, in order to maintain local immune homeostasis (100). The splenic MZ is also involved in the clearance of apoptotic cells from the circulation. IV-injected apoptotic cells drain to the splenic MZ, where they are rapidly engulfed by macrophages in the marginal zone (83). MZMs are known to be critical for particulate trapping in the spleen (72). Depletion studies (in which both MZMs and MMMs are depleted) have also indicated a crucial role for MZ macrophages in apoptotic cell-driven immunomodulation. The initial engulfment of apoptotic cells by macrophages in the MZ is vital in the generation of tolerance to self, whereby delayed clearance of apoptotic cells results in reduced immune tolerance to apoptotic cell-associated antigens (113). For example, depletion of MZ macrophages led to development of systemic tolerance breakdown in mouse models of systemic lupus erythematosus (SLE) and induced inflammatory responses towards apoptotic cell antigens (113).

The infusion of apoptotic cells has been reported to induce immunosuppression in experimental inflammatory diseases, autoimmunity and transplantation. In animal models of autoimmune arthritis, when apoptotic cells were infused not at the joint, but via the IV (114–116) or IP route (114, 117), the resolution of arthritic inflammation was conserved at the joint. In mouse models of transplantation, IV infusion of donor apoptotic splenocytes was shown to promote donor-specific immunosuppression, prolonging the survival of heart allografts (118, 119). Clinical studies have also shown that infusion of leukocytes rendered non-viable, either via a chemical cross-linker, 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (ECDI) or extracorporeal photochemotherapy, is safe and potentially beneficial in multiple sclerosis (120) and cutaneous T cell lymphoma (121). In apoptotic cell-based therapies, the spleen plays a pivotal role as splenic macrophages and DCs are involved in the phagocytosis of apoptotic leukocytes administrated via the IV route (122, 123). Together, these findings point to a critical role for the SLOs in the clearance of apoptotic cells and establishment of a tolerogenic state.

SLOs have a crucial role in host immune defense. In SLOs, antigen priming and immune cell activation occurs, followed by clonal expansion of antigen-specific effector lymphocytes, and the formation of immunological memory and tolerance. The importance of the spleen in host immunity has been demonstrated in various disease settings. In a long-term follow-up study, patients whose spleen had been surgically removed had an increased risk of developing bacterial infections (124). In malaria, the spleen is important in controlling the blood stage infection, clearance of parasitized red blood cells, induction of memory lymphocytes and replenishment of healthy red blood cells (125, 126). Splenectomized patients infected with malaria experienced enhanced parasitic burden, severe disease symptoms, and higher mortality rate (126).

Although SLOs are protective during infection, they may also promote the establishment and progression of some inflammatory diseases. LNs are the key sites for the initiation of GvHD and acute colitis (127, 128). In GvHD, donor cells migrate to the recipient LNs via their expression of lymphoid homing molecules, CD62L (L-selectin) and CCR7 (C-C chemokine receptor 7) (129). Upon recognition of alloantigen on the donor cells, T cells in the recipient LNs are activated and migrate to tissues where they cause damage (commonly in skin, gut and liver) (127). In acute colitis, intestinal migratory DCs drain to the mesenteric LNs where they present antigen and induce Th1 or Th17 responses (128, 130, 131). Pathogenic Th17 cells can also migrate from the mesenteric LNs to the gut and cause inflammatory bowel disease (131). Some studies have shown that lymphadenectomy (surgical removal of a group of lymph nodes and surrounding lymphatic tissues) protected rats from GvHD (132, 133). Lymphadenectomy is rarely investigated in clinical studies for the purpose of reducing inflammation, although it is performed on cancer patients to stop the spread of tumor metastases via the lymphatic system, or to remove the tumor cells in the lymphatic tissues (134, 135).

The spleen can also promote inflammatory responses in some disease settings. For example, splenectomy showed protective effects (e.g. reduced infarct volume) in rat models of brain, liver, kidney and intestine ischemic injury (136–140). In animal models of ischemic brain injury and also in patients with stroke, spleen atrophy (reduction in splenic weight and size) is commonly observed and thought to be a result of splenic leukocytes egressing via the blood circulation and into the injured brain (141). These circulating leukocytes are composed of monocytes, macrophages, neutrophils and lymphocytes, and their migration to the brain exacerbates inflammation and neurodegeneration (142, 143). Splenectomy performed on a rat stroke model (middle cerebral artery occlusion, MCAO), prior to injury, showed neuroprotective effects, with a reduction in brain infarct volume, peripheral immune cells and activated microglia in the brain infarct tissue (140, 142). Apart from the infiltration of splenic immune cells into the brain injury sites, the spleen also contributes to brain inflammation by producing pro-inflammatory signals, such as IFN-γ (140).

The involvement of SLOs and peripheral immune cells in the progression of central nervous system (CNS) inflammation has also been highlighted in other disease models, including spinal cord injury and experimental autoimmune encephalomyelitis (EAE, mouse model of multiple sclerosis) (144–146). In these disease settings, peripheral T cells and myeloid cells are recruited to the CNS and promote disease symptoms by enhancing inflammation. The crucial contribution of splenic T cells in promoting neuroinflammation is further demonstrated in neuropathic pain studies, in which the adoptive transfer of splenic T cells elevated neuropathic pain symptoms in the recipient (144).

Understanding how SLOs regulate the inflammatory response in such disease settings is important in defining the role of these SLOs in mediating the immunomodulatory effects of cell therapy.

Given the importance of SLOs in disease progression and tolerance induction, there is emerging evidence for their involvement in cell therapy in experimental disease models. A notable example is stroke, where IV administration of different therapeutic cell types, such as umbilical cord blood cells, haematopoietic stem cells, amnion epithelial cells, bone marrow stromal cells and multipotent adult progenitor cells, were shown to induce neuroprotective effects (147–150). IV-injected therapeutic cells were detected in the spleen and induced an anti-inflammatory environment that promoted repair in the CNS (e.g. infarct size, peripheral immune cell infiltration) (149–151). Of note, a study using neural stem cells to treat stroke showed that the direct interaction between IV-administered stem cells with splenocytes, particularly CD11b⁺ cells, was necessary for the treatment to be neuroprotective, with splenectomy abolishing the beneficial effect (152). In line with this, the improvement in functional recovery from stroke in MAPC-treated animals was also not evident in the absence of a spleen, supporting a critical role for the spleen (153).

MSCs exhibit broad spectrum immunomodulatory effects in various inflammatory diseases, but do not persist for significant periods in any part of the body following IV infusion. Instead, their lung entrapment and rapid clearance has shifted the focus to lung phagocytic cells as mediators of their therapeutic effects. Although traces of MSCs have been found in SLOs, along with changes in immune cell composition and function, limitations in labelling and imaging techniques make it difficult to establish with certainty that the small numbers detected are viable MSCs, cell debris or label redistribution following phagocytic uptake. There are several cell types with phagocytic capacity in the SLOs, including migratory cells from the circulation. Cells found in different compartments of the spleen and LNs perform different function, which depends crucially upon their spatial interactions with other cells and chemical cues. Furthermore, SLOs comprise not just immune cells, but also a variety of stromal populations that have functional roles in health and disease (180, 181), including the capacity to efferocytose apoptotic cells (182). Whether these stromal populations have a role in MSC therapy is an open area to explore.

DZ, TB, NP, and TH contributed conception and design of the manuscript. DZ wrote the first draft of the manuscript and made the figures. TB, NP, and TH wrote sections of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.

DZ and TB are recipients of the Australian Government Research Training Program (RTP) Scholarship. TH is supported by funding from the National Health and Medical Research Council of Australia (GNT1107188, GNT1162499, GNT2012290) and the Australian Research Council (IC190100026). The Australian Regenerative Medicine Institute is supported by grants from the State Government of Victoria and the Australian Government.

TSPH received funding from Regeneus Ltd outside of this work.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.