Regenerative stromal cell therapy in allogeneic hematopoietic stem cell transplantation: Current impact and future directions

Abstract

Regenerative stromal cell therapy (RSCT) has the potential to become a novel therapy for preventing and treating acute graft-versus-host disease (GVHD) in the allogeneic hematopoietic stem cell transplant (HSCT) recipient. However, enthusiasm for using RSCT in allogeneic HSCT has been tempered by limited clinical data and poorly-defined in vivo mechanisms of action. As a result, the full clinical potential of RSCT in supporting hematopoietic reconstitution and as treatment for GVHD remains to be determined. This manuscript reviews the immunomodulatory activity of regenerative stromal cells in pre-clinical models of allogeneic HSCT and emphasizes an emerging literature suggesting that microenvironment influences RSC activation and function. Understanding this key finding may ultimately define the proper niche for RSCT in allogeneic HSCT. In particular, mechanistic studies are needed to delineate the in vivo effects of RSCT in response to inflammation and injury associated with allogeneic HSCT and to define the relevant sites of RSC interaction with immune cells in the transplant recipient. Furthermore, development of in vivo imaging technology to correlate biodistribution patterns, desired RSC effect, and clinical outcome will be crucial to establishing dose-response effects and minimal biologic-dose thresholds needed to advance translational treatment strategies for complications like GVHD.

Introduction

Culture adherent stem cells have been isolated from many adult and post-natal tissue sources, including bone marrow, cord blood, adipose and others. These stem cells share the common property of low immunogenicity and active immunomodulation, and depending on isolation and expansion conditions, can participate in the regeneration of injured tissue. The multipotent mesenchymal stem cell (MSC) is the most widely studied in both pre-clinical and clinical studies. We propose to classify these adherent stem cell cultures as regenerative stromal cells (RSCs), for the purpose of reviewing their role in tissue repair and immunomodulation in allogeneic hematopoietic stem cell transplantation.

Allogeneic hematopoietic stem cell transplantation (HSCT) results in graft-versus-tumor (GVT) effects, which eradicate residual malignant cells via immunologic mechanisms. However, beneficial GVT activity shares similar immune pathways with deleterious acute graft-versus-host disease (GVHD). Therefore, separating these disparate immune responses within the allogeneic HSCT recipient remains a major challenge (1). Consequently, GVHD and malignant relapse are primary causes of death following allogeneic HSCT (2). Current immunosuppressive therapy used to prevent and/or treat GVHD is suboptimal and can be detrimental by promoting infectious sequelae and relapse of malignant disease (2, 3). Regenerative stromal cell therapy (RSCT) offers the unique potential to facilitate hematopoiesis and engraftment, to modulate alloimmunity without compromising GVT effects, and to promote immune reconstitution and tissue repair (4). In this regard, RSCT is emerging as a novel form of therapy for acute GVHD. This review will examine the immunobiology of RSCT as it relates to the pathophysiology of acute GVHD and focus on the role that relevant inflammatory microenvironments have on RSC activation and function.

Acute graft-versus-host disease

The underlying pathophysiology of acute GVHD involves donor T-cell activation by host alloantigens and secretion of donor-derived cytotoxic cytokines that impact host tissues (5–7). The resultant “cytokine storm” not only targets endothelium primarily within gut, liver and skin (8), but also damages the thymus (9) and lungs (10) of the transplant recipient. Endothelial cell (EC) injury is caused by both the conditioning regimen and the cytotoxic T-cells induced by acute GVHD (11). Whether EC activation is a consequence of antigen presentation by ECs themselves or by immune effector cells via cross-presentation is controversial (12). Yet injured endothelium is a significant source of chemokines and growth factors, which recruit additional immune cells to sites of injury thereby propagating tissue damage in the host.

Murine models of allogeneic HSCT have been instrumental in defining the immune responses underlying acute GVHD (13). Specifically, mouse models have identified cellular and soluble factors that mediate and regulate the GVHD response. Consequently, these mediators have become targets for second-line therapies used to treat steroid-resistant acute GVHD (14–16). However, immunomodulatory therapy is often suboptimal in treating acute GVHD and can increase the risk for opportunistic infections (17) and potentially disease relapse (18). Therefore, novel strategies are needed to reduce GVHD, to preserve GVT effects, and to facilitate engraftment and immune reconstitution.

Potential roles for regenerative stromal cell therapy in allogeneic hematopoietic stem cell transplant

In addition to hematopoietic progenitor cells, the bone marrow microenvironment contains non-hematopoietic progenitor cells that give rise to the stroma of the bone marrow and have the potential to differentiate into cells from other connective tissue lineages such as bone, cartilage and fat. Such stromal cells are not a homogeneous population of cells and have different regenerative potential (19, 20). This review will focus on two specific regenerative stromal cell subpopulations existing within the bone marrow compartment, namely mesenchymal stem cells (MSCs) and multipotent adult progenitor cells (MAPCs) (Figure 1). We refer to their common properties and therapeutic strategies using the term RSCT as described above and provide a focused review on their relevant properties for use in allogeneic HSCT.

Figure 1. Mesenchymal stromal cells: mesenchymal stem cells and multipotent adult progenitor cells.

General comparisons between two specific types of mesenchymal stromal cells are provided. Refer to text for details. Photographs of MSCs (162) and MAPCs (36) reprinted with permission. Prochymal^™ is a mesenchymal stem cell (MSC)-based product GMP-manufactured by Osiris Therapeutics, Inc. (Baltimore, MD). MultiStem^® is a multipotent adult progenitor cell (MAPC)-based product GMP-manufactured by Athersys, Inc. (Cleveland, OH).

Note: ⁺⁺ Population doubling (PD) limit is defined as the maximum number of PDs in which the respective stromal cell maintains telomere length, cytogenetic stability, and multi-lineage differentiation potential in ex vivo culture conditions. These limits are variable and dependent upon the expansion protocol used, the age and condition of the donor, and the frequency of stromal cells in the bone marrow.

Abbreviations: MSC = mesenchymal stem cell; MAPC = multipotent adult progenitor cell; μM = micron.

MSCs are non-hematopoietic cells (CD34⁻CD45⁻) with surface expression for CD73, CD90, and CD105 and can differentiate in vitro into osteoblasts, chondroblasts, and adipocytes (21). Human MSCs have constitutive surface expression of MHC class I and IFNγ-inducible expression of MHC class II. MSCs can be expanded ex vivo from bone marrow mononuclear cells (BM MNC) obtained from animals and humans. Cells with similar characteristics can also be expanded from adipose tissue (22), umbilical cord blood (23), and placenta (24). In humans, the frequency of MSCs ranges from 1 MSC per 10,000 BM MNCs in newborns to 1 MSC per 250,000 BM MNCs in adults (25).

In contrast to HSCs, BM-derived MSCs can be culture expanded until reaching replicative senescence up to 38 ± 4 population doublings (PD) (PD time = 50–60 h) albeit with loss in differentiation capacity (25–27). For example, extensive subcultivation can reduce BM-derived MSC differentiation potential and can induce onset of senescence (28) or even cellular apoptosis (29). Consistent immunomodulation and differentiation of BM-derived human MSCs is well-maintained for up to five passages of cells expanded medium comprised of Dulbecco’s Modified Eagles Media-Low Glucose (DMEM-LG) and 10% fetal bovine serum (FBS) (30, 31). However, isolation and expansion methods vary greatly among different laboratories, and such differences may influence MSC immunomodulation (32).

Like MSCs, multipotent adult progenitor cells (MAPCs) can be expanded from mouse and human bone marrow mononuclear cells (33, 34) and can be grown to clinical scale (MultiStem^®) (35). Expanded MAPCs possess tissue regenerative and immunomodulatory properties similar to MSCs (36, 37), consistent with their position as a developmental progenitor to the MSC. In contrast to MSCs, MAPCs have higher surface expression of CD49d and broader ex vivo pluripotency, with evidence of differentiation into cellular elements of all primitive germ line layers (mesoderm, endoderm and ectoderm) (38). Human MAPCs express telomerase and are capable of more extensive expansion (~75 PDs) than human MSCs (39), particularly when cultured under hypoxic culture conditions. Expansion of undifferentiated human MAPCs from BM MNCs requires fibronectin, platelet-derived growth factor (PDGF), epidermal growth factor (EGF), and 2% FBS (39) with the requirement that cell density and oxygen tension must be tightly controlled (38).

Hematopoiesis and engraftment

MSCs function as paracrine mediators (40), producing cytokines, chemokines and extracellular matrix proteins that support in vitro HSC survival and proliferation and in vivo HSC engraftment. In particular, MSCs constitutively express mRNA for IL-6, -7, -8, -11, -12, -14, -15, M-CSF, flt-3 ligand (FL), and stem cell factor (SCF) (41). Ex vivo stimulation with IL-1α upregulates mRNA expression of G-CSF, M-CSF, leukemia inhibitory factor (LIF), IL-1, IL-6, IL-8 and IL-11 and induces mRNA expression of GM-CSF in human BM-derived MSCs (42). Moreover, MSCs express the stem cell derived factor 1 (SDF-1, CXCL12) receptor CXCR4 (43, 44), which likely contributes to MSC homing and augmentation of HSC engraftment via SDF-1 gradients.

MSCs can support ex vivo expansion of CD34⁺ umbilical cord blood cells (45) and improve human hematopoiesis in NOD-SCID mice following co-transplantation with human HSCs. Noort and colleagues used human fetal lung (FL)-derived CD34⁺ cells to generate MSCs and co-transplanted them with a limiting number of umbilical cord blood CD34⁺ cells. They observed a 3 to 4-fold increase in the level of human hematopoietic engraftment in NOD-SCID mice given FL MSCs compared to those that did not receive MSCs (46). Angelopoulou et al. co-transplanted human BM-derived MSCs with mobilized blood CD34⁺ cells and found enhanced human myeloid and megakaryocytic engraftment in NOD-SCID mice (47). Lastly, Maitra and colleagues reported increased frequency and level of human hematopoietic engraftment in mice co-transplanted with human MSCs and a limiting number of human UCB cells (48).

The profile of proteins secreted by MAPCs also supports their ability to exert hematopoietic effects via paracrine activity (37, 49). Serafini and colleagues showed that BM-derived murine MAPCs, albeit at greater absolute numbers than HSCs, functionally reconstituted the hematopoietic system in vivo in NOD-SCID mice (50). In a myeloablative syngeneic rat HSCT model, allogeneic expanded MAPCs were infused 2 days after HSCT and weekly thereafter for a total of five administrations. MAPC infusion did not affect hematopoietic recovery in transplant recipients, nor cause allogeneic antibody production, T-cell sensitization, or ectopic tissue formation (51). Tolar et al. demonstrated that both host irradiation and NK cell depletion resulted in greater distribution and engraftment of donor MAPC infusions in mice (52). Lastly, Jiang and colleagues also showed enhanced MAPC engraftment and broader distribution profiles for MAPCs in the context of host irradiation (19), suggesting that MAPC homing and engraftment during HSCT can be increased with induction of tissue injury and release of proinflammatory cytokine and chemokines.

Based upon results from these pre-clinical transplant models, MSC infusions have been tested in clinical autologous (30) and allogeneic (53) HSCT settings and found to be safe and well-tolerated. Importantly, no published study has demonstrated a higher incidence of graft failure associated with MSC infusions. Instead some studies have suggested that MSC infusions have decreased the incidence of graft failure over historical controls and even enhanced neutrophil and platelet engraftment in select adult (54, 55) and pediatric HSCT recipients (56), with sustained donor MSC chimerism in pediatric transplant recipients (57). However, MSC infusions have not been reported to confer a consistent engraftment advantage during HSCT (58, 59).

Regeneration and repair

MSCs have been shown to differentiate into, to repair, and to support mesengenic tissue, particularly marrow and connective tissues. Moreover, MSCs may also mediate thymic repair, given their role in thympoiesis and positive selection of T lymphocytes (60), a potential benefit in young transplant patients (61). In contrast to MSCs, MAPCs have broader in vitro differentiation capacity and therefore greater theoretical potential for tissue repair in the setting of HSCT. For example, in the presence of VEGF, MAPCs differentiate into an endothelial progenitor cell phenotype (34) as well as hepatocyte-like cells (62), both of which ultimately assume their respective differentiated cell type morphology and function. In contrast to MSCs, MAPC differentiation into hematopoietic stem cells has been demonstrated in vivo in a NOD-SCID model of allogeneic MAPC engraftment (50).

RSCT-mediated in vivo tissue repair and regeneration has been suggested, but not clearly demonstrated, in pre-clinical (63) and clinical (64) allogeneic HSCT. Evidence for direct stromal cell-mediated tissue repair in the setting of clinical HSCT is lacking (65), as paracrine effects of RSCs likely mediate and/or influence repair (66).

Immunomodulation

RSCs as a broad class of stem cells possess an intrinsic ability to modulate innate and adaptive immune responses. Importantly, MSCs alter antigen-presenting cell (APC) development, maturation and function and inhibit alloreactive T-cell responses (see excellent reviews by Rasmusson (67) and Le Blanc and Ringden (68)). Effects of MSCs on immune cell activation and response have mostly been demonstrated in ex vivo culture conditions, limiting our knowledge of their in vivo effects. Common mechanisms through which MSCs suppress APC and/or T-cell activation and function include direct cell-cell contact, production of regulatory soluble factors, and induction of regulatory cellular phenotypes (69).

Effects on innate immunity: focus on dendritic and natural killer cells

Dendritic cells (DCs) are potent APCs for naïve T cells and have critical roles in donor T-cell activation during acute GVHD (70, 71). Toll-like receptor (TLR) activation causes maturation of peripheral DCs, increasing their surface expression of adhesion and costimulatory molecules, shifting their function from antigen-capturing to antigen-processing cells, and promoting their interaction with naïve T-cells by enhancing expression of CCR7 and migration to secondary lymph nodes (72). MSCs affect DC differentiation, activation and function and inhibit differentiation of monocytes into myeloid DCs (73, 74). MSCs upregulate IL-10 production by plasmacytoid DCs (75), decrease pro-inflammatory IL-12 and TNFα production (75, 76), and decrease co-stimulatory surface markers, such as MHC class II and CD83 (74, 75). In this context, MSCs could alter the ability of DCs to function as potent APCs during GVHD. As an example, application of MSCs in a murine allogeneic BMT model resulted in decreased CCR7 expression and reduced numbers of DCs migrating to secondary lymphoid organs (77).

MSCs also inhibit natural killer (NK) cell proliferation and cytokine (IL-2, IL-15, and IFNγ) production, but are themselves susceptible to NK cell-mediated lysis (78, 79). Interestingly, IFNγ protects MSCs from NK lysis (78), suggesting that an inflammatory microenvironment may modify MSC function towards eliminating NK cells (see section on “Mesenchymal stromal cells and microenvironment”). Furthermore, MSCs could also potentially modulate DC function through their effects on NK cell function (80, 81).

Similar to MSCs, murine MAPCs have constitutive low-level expression of MHC class I that likely underlies their susceptibility to NK cell-mediated lysis (52). In contrast, expanded rat or human MAPCs were not found to be susceptible to NK lysis (82)(van’t Hof, unpublished data). Mouse MAPC MHC class I expression can be upregulated by IFNγ in cell culture, correlating with the use of total body irradiation and subsequent enhanced MAPC engraftment and biodistribution following allogeneic HSCT (52) and suggesting that proinflammatory microenvironments can modify MAPC function. Also murine MAPCs down-modulate co-stimulatory expression on DCs (83).

Effects on adaptive immunity: focus on T-cells

In general, human MSCs inhibit in vitro T-cell activation and proliferation induced by mitogens, recall antigens and alloantigens (48, 75, 84–87). Effects of MSC-mediated T-cell suppression are independent from HLA matching between MSCs and lymphocytes, are dose-dependent, and are generalized across T-cell subtypes (naïve versus memory and CD4⁺ versus CD8⁺). Furthermore, MSC effects are reversible (i.e., do not result in T-cell apoptosis), are associated with downregulation of T-cell activation markers (CD25, CD38, CD69) and are reported to be mediated by various soluble inhibitory factors [heme oxygenase 1, HO-1 (88); hepatocyte growth factor, HGF (89); HLA-G5 (90); indoleamine 2,3-dioxygenase, IDO (91); IL-10 (75, 92); prostaglandin E2, PGE2 (75); and TGF-β (93)]. Lastly, MSCs produced in some laboratories reportedly shift T-cell function to a more regulatory phenotype (75, 94).

Like MSCs, MAPCs do not stimulate in vitro alloreactive T-cell responses (51, 52) and have reversible, dose-dependent and soluble factor-mediated immunosuppressive effects on T-cell alloreactivity (82, 83). Specifically, expanded rat MAPCs mediate IDO-dependent T-cell suppression (82), while murine MAPCs utilize PGE2-dependent mechanisms for T-cell suppression (83). Furthermore, murine MAPC T-cell suppression associates with attenuation in inflammatory cytokine production and co-stimulatory molecule expression on T-cells (83).

Use of regenerative stromal cell therapy for graft-versus-host disease

Due to their relative ease for ex vivo expansion, their infusion safety profile, and their immunosuppressive properties, regenerative stromal cells (RSCs) are being tested clinically to prevent and to treat GVHD. However, efficacy remains inconsistent throughout both pre-clinical models (Table 1) and clinical trials (Table 2), which have used RSCT as therapy for GVHD.

Table 1.

Application of regenerative stromal cells in animal models of allogeneic hematopoietic stem cell transplant

Publication	GVHD model	Stromal cell source	Stromal cell regimen	Results
Highfill 2009 (83)	B6 → BALB\c 850cGy 10M BM, 2M T cells	Mouse MAPC	500K MAPC on Day 1 IV or intra-splenic, prior to T cells on Day 2	PGE2 dependent GVHD protection after intrasplenic administration
Kovacsovics-Bankowski 2008 (51)	Buff → BuffxLewF1 700 cGy 20M BM, 7M T Cells	Lewis Rat MAPC	2.5M MAPC Day 1 or Day 1 + 5	Survival benefit after 1 or 2 administrations.
Tian 2008 (100)	Fischer → Wistar 850 cGy 10M BM, 20M SPL	Fischer Rat BM MSC	2M MSC on Day 1 after HSCT	Survival benefit and reduction of T cells, Th1/Th2 ratio and increase of CD4+/CD25+ cells.
Badillo 2008 (96)	B6 → CB6F1 900 cGy 10M BM, 30M SPL	Mouse stromal progenitor cells (mSPC)	150K or 1M mSPC on Day 0 with BMT or Day 2; 50K SPC at Days 0, 7, and 14; or 150K SPC at Days 10 or 21	No survival or clinical score benefit observed in any prophylaxis or treatment regimens.
Prigozhina 2008 (98)	B6 → CB6F1 650cGy 10M BM, 25M SPL	Mouse BM/placental/UCB-MSC	50K or 500K MSC on Days 0+7+14	No survival benefit observed for any MSC type tested.
Ren 2008 (101)	B6 → B6C3HF1 1300 cGy 5M BM, 5M SPL	Mouse BM MSC (WT, IFNγR1 KO, or iNOS KO)	500K MSC on Days 1 + 3	GVHD survival dependent on pre-exposure to IFNγ + either TNFα, IL1α or IL1β and iNOS.
Polchert 2008 (102)	BALB/c → B6 1000 cGy Not reported	Mouse BM MSC (WT, IFNγ-deficient)	100 or 500K MSC on Days 0, 2, 20, or 30	Survival benefit after administration on Day 2, 20, or 30, increased by IFNγ priming. IFNγ-deficient T cells insensitive to GVHD protection by MSC.
Min 2007 (103)	B6 → B6D2F1 1100 cGy 10M BM, 20M SPL	Mouse BM MSC (+/− IL-10 transduction)	1 or 2M MSC on Day 1 or Days 1, 3, and 5	IL-10 dependent benefit for survival, clinical scores and reduced IFNγ serum. No survival benefit by unmodified MSC at Day 1, or 1, 3, and 5.
Tisato 2007 (104)	hPBMC → NOD-Scid 250 cGy 20M huPBMC	UCB-MSC	3M UCB-MSC Day 0, or Days 0, 7, 14 and 21, or Week 5, 6, 7, 8	Benefit after infusion on Days 0, 7, 14 and 21, but not after infusion on Week 5, 6, 7, 8.
Yanez 2006 (22)	B6 → B6D2F1 1100 cGy 10 BM, 20M SPL	hAd-MSC, hBM-MSC, mAd-MSC	50K mAd-MSC on Days 0, 7, and 14, or Days 14, 21, and 28	Benefit after infusion on Day 0, 7 and 14, but not after administration on Day 14, 21 and 28.
Sudres 2006 (95)	B6 → BALB\c 800cGy 3M BM, 100K T cells	Mouse BM-MSC	500K, 3 or 4 M MSC at −15 min prior to BMT	No survival benefit observed. Minor GI protection seen with 4M dose.
Chung 2004 (107)	C3H/He → BALB\c 875 cGy 10M BM, 5M SPL	Mouse BM-MSC	100K MSC on Day 0 with BMT	Survival benefit when co-infused with BM, but not when co-administered with BM and SPC.

Abbreviations: Ad=adipose; BM=bone marrow; B6=C57Bl/6; Buff=Buffalo rat; h=human; K=thousand; Lew=Lewis rat; m=mouse; M=million; PBMC=peripheral blood mononuclear cell; SPC=stromal progenitor cell; SPL=splenocytes.

Table 2.

Published clinical experience of MSCs to prevent or to treat graft-versus-host disease in allogeneic hematopoietic stem cell transplant recipients

Publication	Indication	Trial specifics	MSC regimen	Endpoints and results
Kebriaei 2009 (111)	De novo acute GVHD (II–IV)	N=32 patients (21/10 M/F) Median age 52 yr (34–67) 21 grade II, 8 grade III and 3 grade IV acute GVHD	Allogeneic (unrelated, unmatched) BM-MSC 2 or 8 M/kg + steroids Dose 1 at 24–48 hr after GVHD, Dose 2 at +3 days	No infusion-related toxicities or ectopic tissue formation 94% initial response by Day 28 (77% CR and 16% PR) No difference in safety or efficacy between low- and high-MSC dose
von Bonin 2009 (112)	Steroid-refractory, severe acute GVHD	N=13 patients (7/6 M/F) Median age 58 years (21–69) 2 grade III, 11 grade IV acute GVHD	0.9 M/kg (0.6 to 1.1 M) 2 doses third-party BM-MSC expanded in platelet lysate-containing media Dose 1 at median 16 days after aGVHD onset	No infusion-related side effects All MSCs used impaired PHA-stimulated CD4+ proliferation Only 2 patients did not require additional escalation of concomitant immunosuppressive therapy 5/7 patients with initial response required additional MSC therapy 4/9 deaths attributed to GVHD
Le Blanc 2008 (110)	Steroid-refractory, severe acute GVHD	N=55 patients (34/21 M/F) Median age 22 yr (0.5–64) 5 grade II, 25 grade III and 25 grade IV acute GVHD	1.4 M/kg MSC (0.4 to 9) 1 dose (27); 2 doses (22); 3–5 MSC doses (6) 5 Sib, 18 Haplo, 69 HLA-MM MSC	No infusion-related side effects 30/55 complete response, 9 partial, CR show survival benefit over PR/NR at 2 years 53% versus 16% No relation between response and MSC HLA match
Muller 2008 (109)	Immunological complications after allo-HSCT in pediatric patients	N=7 patients (M/F not specified) Median age 14 yr (4–17) 2 acute GHVD and 3 chronic GVHD	0.4 to 3 M/kg MSC 1 (4), 2 (2), and 3 doses (1) 5 Haplo, 2 third-party parental MSC	No infusion-related side effects (at 29 months max) 2 patients with severe acute GVHD did not progress to chronic GVHD 1/3 slight improvement of chronic GVHD
Fang 2007 (161)	Steroid-refractory, severe acute GVHD	N=6 patients (2/4 M/F) Median age 39 yr (22–49) 2 grade III and 4 grade IV acute GVHD	1 M/kg Adipose MSC 1 dose (5), 2 doses (1) 2 Haplo, 4 HLA-MM MSC	No infusion-related side effects 5/6 complete response 4/6 survival (18–90 months post treatment)
Ringden 2006 (31)	Steroid-refractory, severe acute GVHD	N=9 patients (7/1 M/F), Median age 56 yr (8–61), 2 grade II (1 chronic), 5 grade III and 1 grade IV acute GVHD, 1 grade III after DLI	1 M/kg MSC (0.7 to 9) 1 dose (6); 2 doses (4) 2 Sib, 6 Haplo, 4 HLA-MM MSC	No infusion-related side effects 6/8 complete response 5/8 survival (2–36 months post treatment)
Lazarus 2005 (53)	GVHD prevention	N=46 patients (24/22 M/F), Median age 45 yr (19–61)	1–5 M/kg Sib MSC 4 hr pre-HSCT	No infusion-related side effects or ectopic tissue formation No increase in incidence or severity of GVHD

Note: Criterion for inclusion in table was published clinical trial and/or experience with greater than 5 patients.

Abbreviations: CR=complete response; Haplo=Haploidentical donor; M=million; M/F=male/female; MM=HLA mismatched donor; NR=no response; PR=partial response; Sib=HLA-identical sibling.

Pre-clinical experience

Use of RSCT for preventing and treating GVHD in pre-clinical models has shown mixed benefit (Table 1). Such inconsistency in effect likely reflects the variability in disease models with respect to how alloreactivity is induced (minor versus major MHC mismatch, T-cell dose), the conditioning regimen used (lethal versus sublethal radiation), and the types of hematopoietic and regenerative stem cell types given (stem cell source and method of expansion, stem cell dose, and timing and mechanism of delivery). In addition, pre-clinical transplant models have shown that in vivo effects of MSCs are not necessarily predicted by their in vitro effects (95, 96) and, in contrast to their intended effect, some studies have even shown that MSC infusions may cause graft rejection (97, 98).

Despite these drawbacks, pre-clinical models have helpful in revealing key aspects of RSCT in the context of allogeneic HSCT. First, the level of host immuncompetency is an important factor that can influence the use of RSCT. For example, murine MHC-mismatched MSCs are either rejected (99) or induce graft rejection (73) in the context of absent or minimal sublethal radiation, respectively. Pre-clinical studies have also revealed that intended benefit of RSCT is independent of the source or subset of RSCs used as well as the haplotype match between the stromal cells used and the host, as long as the host is sufficiently immunocompromised (Table 1). Finally, survival benefit in using RSCT closely associates with infusion of cells early post-transplant and after induction of GVHD. Specifically, administration of MSCs (100–103) and MAPCs (82, 83) after HSC administration confers a greater survival benefit than co-administration with HSCs (95, 96, 98). Furthermore, RSC infusion after 21 days of HSC infusion does not confer a survival advantage (22, 104). Collectively, these data suggest an early post-transplant “window” of time when RSCs are infused that uniquely affects RSC activation and function.

However, direct translation application of pre-clinical findings into clinical approaches remains limited for several reasons. First, most pre-clinical GVHD models are driven by CD4⁺ T cell activities (13), whereas CD8⁺ T cell subsets may be the primary mediators of clinical acute GVHD in the allogeneic transplant setting (105). Second, pre-clinical models may be slanted towards diverse immunological tendencies. For example, C57Bl/6 mice display prototypic Th1 responses, whereas BALB/c mice are prone to Th2 responses (106). Therefore, choice of donor and recipient mouse strain combinations can potentially impact efficacy of RSCT as a consequence of exposure to different immunomodulatory milieus. In support, pre-clinical GVHD models suggest trends in measured benefit based upon mouse strains (Table 1). Six studies in mice and two studies in rats report survival advantage and improved clinical GVHD scores in animals given RSCT (22, 51, 100–104, 107). In contrast, four studies show no GVHD benefit; and these studies all utilize BALB/c or BALB/c-derived F1 as transplant recipients (83, 95, 96, 98). One study using BALB/c recipients did show protective activity of MSCs, but only when they were administered at time of transplant and not infused simultaneously with T-cells (107). It is possible that variations in the immunological context encountered by the infused stromal cells may impact their benefit. Yet other factors including the degree of MHC mismatch between donor T cells and the BALB/c recipients must also be considered. However, at minimum, variation in results from pre-clinical models suggest the need for utilizing multiple models in parallel in order to permit more generalized interpretation for the benefit of RSCT.

Clinical experience

Clinical experience using RSCs is also limited and results are equally mixed as the published pre-clinical experience (Table 2). Initial success using maternal haploidentical MSCs in a pediatric HSCT patient with steroid and second-line immunomodulatory therapy-resistant acute GVHD (108) has driven the broader application of BM-derived MSC therapy during allogeneic HSCT (31, 53, 109–111). The two largest published series of patients treated with BM-derived MSCs have shown complete and/or partial responses to MSC therapy in 70% (n = 39/55) and 90% (n = 29/32) of patients receiving MSCs for steroid-resistant (110) or MSCs in combination with steroids for de novo acute GVHD (111), respectively. Furthermore, steroid-resistant GVHD patients having a complete response to MSC therapy also had improved overall survival and decreased transplant-related mortality compared to patients with partial or no response (110). Although variable doses of MSCs were administered, the median single dose of MSCs in the EBMT Development Group experience (1.4 × 10⁶ MSCs/kg)(110) was similar in therapeutic efficacy to the single dose used in the randomized, open-label phase II Osiris trial (2 × 10⁶ MSCs/kg)(111). In the Osiris trial efficacy using the clinical-grade MSC product Prochymal^® was similar between high- (2 doses, each 8 × 10⁶ MSCs/kg) and low-dosed study groups (2 doses, each 2 × 10⁶ MSCs/kg)(111). Results from the phase III Osiris clinical trials have recently been reported via press release from the company (September 8, 2009 http://investor.osiris.com/releasedetail.cfm?ReleaseID=408763). In both studies weekly or bi-weekly MSC administrations were given to patients for four weeks with individual dosing at 2 × 10⁶ MSCs/kg, the lower-end of MSC dosing compared to previous published experiences (110, 111). Neither the steroid-refractory (Protocol 280, NCT00366145, n=260) nor the newly-diagnosed (Protocol 265, NCT00562497, n=192) acute GVHD trials reached the primary endpoint of durable complete response. However, select patients with either steroid-refractory liver or gastrointestinal GVHD who received Prochymal^® were reported to have significantly improved response rates (76% vs. 47%, p=0.026, n=61 and 88% vs. 64%, p=0.018, n=71, respectively). Lower MSC dosing and/or differences in treatment regimens across institutions may have affected these results. Other smaller-sized trials in pediatrics (109) and in adults (112) have also shown either modest or transient response, respectively, following MSC administration.

Together, these clinical studies highlight the need for continued pre-clinical evaluations to optimize dosing strategies, to define biodistribution profiles, and to assess efficacy. Moreover, the inconsistent effects of RSCT shown in pre-clinical models as well as clinical trials suggest the fundamental need to understand mechanisms for RSC activation and function, specifically in relationship to the microenvironment into which RSCs are infused and to where these cells ultimately home.

Regenerative stromal cells and their microenvironment

How RSCs respond to and interact with their in vivo microenvironment remains largely undefined, but is critical to understanding how these cells function in GVHD and to improving their efficacy in preventing and treating GVHD. This section will focus on how RSCs may functionally respond to the inflammatory microenvironment they encounter and will also review an emerging literature that defines how the microenvironment can modulate RSC activation and function.

Toll-like receptors

Within the microenvironment, constitutive and inducible cytokines, chemokines, growth factors and other soluble factors are produced by resident and migrant hematopoietic and non-hematopoietic cells. Pathogen-associated molecular patterns (PAMPs) and other danger signals for inflammation and injury are recognized by Toll-like receptors (TLRs) on the surface of APCs (113). Once activated, TLRs signal through MyD88-dependent and MyD88-independent signaling cascades resulting in NF-κB dependent gene transcription of proinflammatory cytokines and chemokines (114). Activation of TLR2 and TLR4 on hematopoietic progenitor cells has recently been shown to influence progenitor cell differentiation (115, 116). Murine BM-derived MSCs (117) and human BM and adipose-derived MSCs (118) also express TLR2 and TLR4. TLR2 and TLR4 are important in sensing and initiating host defense responses to injury and to inflammation (119) and are upregulated by disease states like sepsis (120, 121) and by soluble factors including cytokines (122, 123) and other TLR ligands (124, 125). Hypoxic culture conditions can increase TLR2 mRNA expression on human adipose-derived MSCs (118), while LPS can upregulate levels of TLR4 mRNA and TNFα production from ex vivo stimulated rat BM-derived MSCs (126).

LPS further promotes inflammation by inducing production of reactive oxygen species, an effect balanced by anti-oxidant enzymes like heme-oxygenase 1 (HO-1) (127, 128). HO-1 modulates innate (129) and adaptive (130) immune cell function by attenuating inflammation (131) and potentially preventing GVHD (132). HO-1 produced by rat and human MSCs suppress T-cell proliferation (88). Thus, LPS could potentially activate both TLR and HO-1 signaling (133) cascades within MSCs to modulate LPS-induced inflammation associated with acute GVHD. Furthermore, combining agents like the synthetic triterpenoids, which are potent inducers of HO-1 (134), with MSCs might also augment the beneficial effects of HO-1 induction for the treatment of GVHD (135).

TLR ligation also influences MSC proliferation, differentiation, and immune function. Cho and colleagues have demonstrated that TLR2 (Staphylococcus aureus peptidoglycan) and TLR4 (LPS) ligands promote osteogenic differentiation of human adipose-derived MSCs (118), effects dependent upon TLR signaling cascades (136). Using murine MSCs, Pevsner-Fisher and colleagues have shown that TLR2 agonist, Pam₃Cys, augments MSC proliferation and inhibits MSC differentiation into osteoblasts, adipocytes and chondrocytes (117). In addition, TLR2-stimulated murine MSCs retain their ability to inhibit T-cell proliferation. In contrast, ligation of TLR3 and TLR4 on human BM-derived MSCs reversed their ability to inhibit T-cell proliferation without affecting MSC phenotype or differentiation, a process shown to be dependent upon Notch signaling (137). In human BM-derived MSCs, ex vivo stimulation with TLR3 ligand, poly(I:C), induced NF-κB-dependent cytokines and chemokines as well as enhanced MSC transwell migration, suggesting that TLR3 ligation is critically involved in mediating human MSC migration (138). Opitz and colleagues have also demonstrated the importance of TLR3 ligation in enhancing MSC-mediated immunosuppression via IDO induction requiring type I IFN signaling (139).

Cytokines and chemokines

As important as the effect of MSCs on immune effector cell function is the cross-talk effect that immune effector cells and their inducible factors may have on MSC activation and function. For example, MSCs require activation via soluble factors like IL-1β and IFNγ to mediate T-cell inhibition. IL-1β produced by CD14⁺ cells activates human MSC-mediated in vitro T-cell suppression via TGF-β1 (93). Similarly, IFNγ added to human MSC cultures induces MSC MHC class II expression and enhances their inhibitory activity in mixed lymphocyte cultures (84).

Levels of IFNγ may also affect MSC function as immunosuppressive cells or immunoactivating APCs. In one report, low concentrations of IFNγ upregulate MHC class II expression on MSCs and result in MSCs functioning as APCs, whereas increasing concentrations of IFNγ decrease MHC class II expression and shift MSC function into alloinhibitory cells (140). IFNγ activation of MSCs and subsequent alloinhibition has been shown to be mediated by IDO (141), an effect due to the accumulation of tryptophan metabolites (142). IFNγ pretreatment also increases murine MSC cross presentation of exogenous antigen via MHC class I to induce CD8⁺ cytotoxicity, further suggesting that cytokine exposure modulates MSC APC function (143). Finally, IFNγ and TLR priming induces MSC proinflammatory cytokine production, which by inference may increase local recruitment of innate effector cells that may subsequently be modulated by MSCs themselves (144). Together, these results suggest an immunomodulatory role for MSCs in the LPS-induced inflammatory microenvironments present during sepsis (145) and after GVHD induction.

MSCs express chemokine receptors and adhesion molecules involved in migration to sites of tissue injury (146, 147). MSC migration is broad in the absence of injury, but preferential in response to injury and inflammation (36, 148), an effect mediated by chemokine receptors and gradients. For example, low-dose radiation upregulates CCR2 expression on MSCs, enhancing their migration into tumor beds (149). Once homed to sites of injury or inflammation, MSCs have been shown to act like neutrophils in their ability to extravasate from blood vessels through coordinated rolling and adhesion (150). Furthermore, matrix metalloproteinases released by MSCs degrade the endothelial vessel basement membrane to allow extravasation into damaged tissue (151). Chemokine receptor expression on MSCs may further be influenced by the inflammatory microenvironment and even the very soluble factors produced by MSCs themselves (152, 153). Thus, paracrine and autocrine induction of chemokines and cytokines likely culminate to modulate MSC function within and migration to a particular microenvironment.

Dynamic immunomodulation between mesenchymal stromal cells and their microenvironment

The allure for using RSCT to treat immune-mediated diseases such as GVHD (110) is tempered by studies suggesting that MSCs may not be as immune-privileged (99) or alloinhibitory (97) as generally perceived. MSC immune function seemingly is a delicate balance between immune privilege and suppression (154). Plasticity in MSC function likely reflects the microenvironment in which these cells reside and to which they ultimately migrate. Consequently, “net” MSC immune function may be determined by the microenvironment in which indigenous and migratory hematopoietic and non-hematopoietic cells and their associated constitutive and inducible soluble factors combine to modify MSC activation and function. In turn, “activated” MSCs modulate the activation, function, and maturation of immune cells. Therefore, a dynamic interplay ensues between mesenchymal stromal cells and immune cells within this immunomodulatory milieu (see Figure 2).

Figure 2. Emerging model of mesenchymal stromal cell activation and function in the context of tissue injury and/or inflammation.

Migrating hematopoietic and non-hematopoietic cells home to sites of tissue injury and inflammation along gradients created by inflammatory chemokines and danger signals. Resident hematopoietic (macrophages/monocytes, dendritic cells) and non-hematopoietic (stromal cells) cells are also activated by this inflammatory milieu. Activated mesenchymal stromal cells (MSCs) produce immunomodulatory soluble factors (e.g., HGF, HO-1, IDO, IL-10, PGE2, TGF-β) that inhibit antigen-presenting cell (APC) maturation and development, inhibit T-cell activation and function and induce regulatory dendritic cell (DC) and T-cell phenotypes. Thus, “net” MSC immunomodulation likely reflects the summation of modulatory cellular and soluble factors within the microenvironment that can influence MSC activation and function. Other potential desirable effects mediated by MSCs during allogeneic stem cell transplant area also shown.

Recent publications provide evidence for and insight into a “dynamic immunomodulation model” between MSCs and their microenvironment, especially in the context of murine allogeneic HSCT and GVHD. Ren and colleagues demonstrated IFNγ-activated, MSC-mediated suppression of T-cells via nitric oxide (NO) production (101). Specifically, IFNγ in combination with proinflammatory cytokines (IL-6, IL-1α, and TNFα) within the microenvironment was necessary to induce MSC production of chemokines (CXCL-9, MIG and CXCL-10, IP-10). These MSC-derived chemokines are hypothesized to bring alloreactive T-cells in close proximity to MSCs, wherein NO produced by MSCs then suppresses T-cell activity. To further demonstrate the critical roles of IFNγ and NO, MSCs derived from mice deficient in IFNγR1 or iNOS were used and these cells failed to prevent GVHD.

Polchert and colleagues also showed that IFNγ is necessary to activate MSC-mediated prevention of GVHD and that timing of MSC infusion was critical to GVHD prevention (102). Specifically, MSC infusion at the time of BM transplantation (day 0, D0) did not prevent GVHD. However, MSC infusions given at D2 or D20 significantly improved overall survival and histological evidence for GVHD, suggesting the need for MSCs to be primed by their microenvironment in order to be effective. Moreover, MSCs pretreated with increasing concentrations of IFNγ were infused into transplant recipients and shown to prevent GVHD in a dose-dependent manner. Taken together, these pre-clinical results seem to recapitulate the clinical experience using BM-derived MSCs in that MSCs given at the time of transplantation do not significantly improve the incidence of severe GVHD (53), while MSCs given at the time of GVHD onset have therapeutic benefit (110), likely reflecting critical differences in the microenvironments encountered by the RSCs at these different times following allogeneic HSCT.

Future directions

The emerging concept of microenvironment influence on RSC activation and function should be applied to future mechanistic study using pre-clinical models to evaluate therapeutic approaches with direct clinical application to allogeneic HSCT. For example, defining the influence of RSCT on inflammatory tissue as a site of initiating naïve T-cell allo-sensitization may reveal that RSCT influences disease regionally in addition to its local immunomodulatory effects within secondary lymphoid organs. Specifically, administering RSCT early after inflammatory tissue damage versus later after T-effector amplification in the secondary lymphoid organs may critically influence RSC homing and effect. For example, Highfill et al. have reported GVHD benefit, but only with direct intra-splenic administration; as MAPCs intravenously injected at time of transplant did not show GVHD and survival benefit (83). Another important question is whether ex vivo manipulation of RSCs can preferentially alter their biodistribution and function in different microenvironments. In this regard, Min et al. have shown that infusion of MSCs over-expressing IL-10 enhance protection from GVHD (103); and Polchert et al. have reported increased MSC efficacy after IFNγ priming (102).

Intravenous injection remains the most convenient and widely-used route of RSCT administration. Yet few published animal models address where infused RSCs home and/or reside (52, 83, 95, 96), and only one study addresses effects of alternative routes of administration on biodistribution (52). Defining homing patterns that impact RSC interactions with target cells is critical to establishing dose dependency and timing of administration to optimize RSCT effects. Following the observation by Highfill and colleagues (83), ex vivo manipulation of RSCT or the use of alternative dosing regimens could guide homing of sufficient numbers of RSCs to localized sites of GVHD in order to reduce exposure to higher numbers of third-party, immunomodulatory cells that may adversely mediate tolerogenic effects and increase the risk of malignant disease relapse or metastasis (59, 155). In addition to encountering hematopoietic cells within immunomodulatory milieus, systemically administered RSCs may also interact with endothelium and epithelium (150, 156). Damaged endothelium serves as a significant source of soluble factors that recruit additional immune cells to sites of injury, thereby contributing to propagation of host tissue damage in the context of GVHD (157). Preliminary in vitro findings show that MAPCs can inhibit activation of endothelial cells by inflammatory triggers such as TNFα, IL-1β, and LPS and diminish endothelial expression of cytokines and adhesion molecules involved lymphocyte extravasation (158). In addition, human MAPCs inhibit fucosyltransferase 7 (Fut7) and surface expression of Lewis antigen (CD15s) on activated lymphocytes (159, 160), resulting in reduced binding of T-cells to activated endothelium (van’t Hof unpublished observations). Therefore, RSCT may prevent GVHD-mediated tissue damage by down-regulating cytotoxic lymphocyte extravasation and/or cytokine production from damaged endothelium.

Conclusions

Regenerative stromal cellular therapy has the potential to become a novel treatment for graft-versus-host disease following allogeneic hematopoietic stem cell transplantation. However, key issues that likely impact efficacy of RSCT require further investigation.

• Standardization in expansion protocols and administration of RSC products is needed to ensure product homogeneity and consistency within and across clinical studies.

• Refined and sensitive analysis of RSC biodistribution at the level of single cell detection is needed to identify anatomical sites where RSCs interact with immune effector cells. Development of these imaging technologies would also permit correlation of biodistribution patterns with RSC activity and outcome measures.

• Establishing relevant potency assays for RSC products would define dose and delivery strategies and may reveal additional ways for exploiting RSC immunomodulation and tissue repair capacity, including ex vivo manipulation to direct RSC immunomodulatory activity by enhancing delivery to target areas.

• Sufficient biomarkers have yet to be identified to define optimal timing of RSC administration as well as to follow clinical response in order to determine the need for repeat RSC dosing. Similarly, delineating dose-response effects and a consistent and minimal biologic-dose threshold is critical for advancing clinical treatment strategies.

• Finally, defining whether immunomodulatory effects of RSCs at the site of inflammatory tissue injury and within secondary lymphoid organs differ, as these two sites may utilize different anti-inflammatory and immunomodulatory pathways.

In closing, the proper niche for RSCT in allogeneic HSCT is still undefined. Yet the clinical potential and promise of RSCs may ultimately be achieved through more advanced mechanistic study into their in vivo effects and through well-designed clinical trials demonstrating their clear benefit.