Alternative approaches to myeloid suppressor cell therapy in transplantation: comparing regulatory macrophages to tolerogenic DCs and MDSCs

Mouse CD11b⁺ Ly6C⁺ bone marrow monocytes cultured under conditions analogous to those used in the generation of human M regs give rise to a population of suppressive macrophages which are highly similar to human M regs in morphology, cell-surface phenotype and in vitro function [16]. Mouse M regs express a selection of typical macrophage markers, including CD11b, CD11c, CD68, F4/80 and CD14, and exhibit a partially-matured phenotype with intermediate levels of MHC Class II and CD80, and no expression of CD40 or CD86. Mouse M regs express sialoadhesin (CD169), macrophage scavenger receptor (CD204) and Dectin-1, but lack other markers of notable tissue macrophage subsets, such as Dectin-2, MARCO, CD4, CD206 and CD209. M regs do not express Ly6C or Ly6G, which together constitute the Gr-1 antigen that is expressed by all mouse MDSCs. CD11c is homogenously expressed by mouse M regs, but they do not express other DC subset-defining markers, including 33D1, OX40L (CD252), CD103, CD205 and CD207. Importantly, like human M regs, mouse M regs do not express TLR2 or TLR4. Given its mode of derivation, morphology and cell-surface phenotype, it seems most appropriate to classify the M reg as a macrophage; however, mouse M regs do not express markers typical of either M1-polarised macrophages (eg. TNF-α, IL-6 and IL-12b) or M2-polarised macrophages (eg. CD206, Ym1 or Fizz1). To better understand the phenotypic relationship between mouse M regs and previously-described macrophage polarisation states, we performed whole-genome gene expression studies to compare M regs to monocytes, monocyte-derived DCs, resting macrophages, IFN-γ-stimulated macrophages and M1-, M2a-, M2b- and M2c-polarised macrophages: These experiments showed that mouse M regs represent a novel and unique state of macrophage activation. Mouse M regs inhibit T cell responses in vitro by several mechanisms. Mitogen-stimulated T cell proliferation is non-specifically inhibited in co-cultures with M regs through the action of inducible nitric oxide synthase (iNOS). M regs delete co-cultured allogeneic T cells (but not isogeneic T cells) through an unknown mechanism that ultimately leads to phagocytosis of the T cells. Any T cells which survive direct co-culture with M regs are impaired in their ability to secrete IL-2 and IFN-γ upon specific and non-specific restimulation [16].

Preclinical experiments using a heterotopic mouse heart transplant model demonstrate the potential of mouse M regs to prolong allograft survival [16]. A single intravenous administration of 5×10⁶ donor-strain M regs at 8 days prior to transplantation significantly prolongs allograft survival in unconditioned, non-immunosuppressed recipients using both the stringent C3H-to-BALB/c (32.6 ± 4.5 vs. 8.7 ± 0.2 days) and B6-to-BALB/c (31.1 ± 12 vs 9.7 ± 0.4 days) strain combinations. This graft-protective effect is specific to donor cells, as recipient cells do not prolong graft survival compared to untreated controls and third party-derived M regs provide only a marginal benefit (11.0 ± 0.6 days). Improved graft survival is observed irrespective of whether M regs are administered 8 or 35 days prior to transplantation. Co-treatment with M regs and 1 mg/kg/day rapamycin for 10 days post-transplant enhances the graft-protective effect of M regs (64.1 ± 8.6 days) compared to treatment with M regs alone or rapamycin alone, and some recipients co-treated with M regs and rapamycin accept their allografts indefinitely. The mechanism of M reg-mediated allograft protection is iNOS-dependent because M regs derived from Nos2-deficient mice only marginally prolong graft survival (12.0 ± 1.8 days). Very importantly, the iNOS-dependence of M reg treatment in vivo proves that the graft-protective effect of M regs is not simply due to alloantigen exposure, but must be mediated by living, metabolically-competent cells.

Several mutually redundant mechanisms might be invoked to explain the effects of M regs in vivo and, in our opinion, more than one mechanism is likely to be in operation (Figure 2). It is well-known that pre-transplant exposure to donor alloantigen promotes allograft acceptance [33] and that delivery of alloantigen as apototic cell debris enhances this effect [34, 35]. Both CD8α⁺ DCs [36] and F4/80⁺ PD-L1⁺ IL-10-producing macrophages [37] of the splenic marginal zone appear to be important for the tolerogenic effects of complement-opsonised apoptotic antigens [38]. After intravenous injection into mice, isogeneic and allogeneic mouse M regs are initially trapped in the pulmonary vasculature, then rapidly redistribute to other peripheral organs, especially the liver and spleen, but not to lymph nodes [16]. Isogeneic and allogeneic M regs are relatively short-lived after transfer into immunocompetent mice, both being detectable at 2 weeks post-infusion, but not 4 weeks. These experiments suggest that M regs have an inherently limited lifespan after transfer; therefore, M regs probably serve as a source of apoptotic donor alloantigen-expressing material. Such a mechanism operates in tolerance induction protocols using donor-specific transfusion (DST) and αCD154 treatment, in which indirect presentation of donor alloantigen results in a predominantly deletional tolerance [39]. And yet, the graft-protective effect of donor alloantigen exposure in the absence of costimulatory blockade or lymphodepletive conditioning [40] is rarely as profound as that achieved with M reg treatment, even in less stringent transplant models. Moreover, it is difficult to reconcile the requirement for iNOS expression by transferred M regs with the suggestion that M regs act merely as a passive source of alloantigen. It is possible that mouse M regs directly suppress T cell responses in vivo through iNOS activity, as they do in vitro, but the critical action of iNOS might equally be mediated by recipient APCs [41, 42]. Accordingly, one important fate of M regs might be to migrate into tissues, induce a suppressive condition in recipient APCs, before dying in a suitably self-conditioned environment.

Naturally-occuring myeloid suppressor cells fall into three ‘classes’: those present in non-inflamed tissues; those arising in inflamed tissues; and dedicated myeloid suppressor cell populations which are recruited to both inflamed and non-inflamed tissues. The ‘default’ condition of immature DCs and macrophages in non-inflamed tissues is suppressive; such cells play crucial roles in the maintenance of tissue homeostasis and self-tolerance, as well as resisting the otherwise perpetual activation of inflammatory responses [43]. Suppressive macrophages and DCs can also arise in inflamed tissues through conversion from activated, mature APCs; such cells limit the extent of inflammatory responses and promote tissue-reparative processes and the re-establishment of tissue homeostasis [44]. Committed myeloid suppressor cell populations, collectively known as MDSCs, are present in blood and can infiltrate both inflamed and non-inflamed tissues, including allografts and tumours [45]. These three general classes of naturally-occurring myeloid suppressor cell are reflected by the types of myeloid suppressor cell that can be generated in vitro.

Immature macrophages and DCs are normal stromal components of almost all peripheral tissues, where they serve many vital functions, including the removal of dead cells, microbial products and other tissue debris by phagocytosis, and the maintenance of a non-inflammatory environment through secretion of suppressive mediators. The immature state of tissue-resident macrophages and DCs is stabilised by anti-inflammatory factors present in non-inflamed tissues, such as IL-10 and glucocorticoid. Colonic macrophages exemplify the role of immature, tissue-resident APCs in preventing aberrant inflammatory responses; these cells respond to IL-10 by secreting IL-10, which suppresses the continual inflammation that would otherwise be caused by gut commensals and their products [46]. Besides their local effects in tissues, immature DCs deliver self-antigens to secondary lymphoid organs, where they are presented to T cells in a poor co-stimulatory context to propagate self-tolerance. Many in vitro-derived tolerogenic DCs fit this description of a phagocytic, maturation-resistant cell with the ability to present antigen in the absence of strong co-stimulation, including DC-10, Rapa-DC and Tol-DC [10]. A wide range of unrelated stimuli drive DCs to suppressive states, so there is no unique phenotype or mechanism of action associated with tolerogenic DC populations (Table 1). Generically, human tolerogenic DCs express cell-surface markers typical of immature monocyte-derived DCs, but there are markers of specific subtypes: Tol-DC are CD11c⁺ CD11b⁺ EBI3⁺ cells; DC-10 express high levels of HLA-G, ILT4 and secrete IL-10; Rapa-DC lack expression of CD80 and CD86, express only low-levels of HLA-DR, but secrete significant amounts of IL-12. These three tolerogenic DC subtypes exert their suppressive effects through alternative mechanisms, specifically: Tol-DC inhibit T cell responses through haemoxygenase (HO)-1 activity and T reg expansion [17]; DC-10 suppress T cell proliferation and induce antigen-specific Tr-1 cells through IL-10 production [21]; Rapa-DC induce effector T cell anergy and promote FoxP3⁺ T reg generation by presenting antigen in the absence of costimulation. Tolerogenic DC populations also differ in their migratory capacity in vivo. Studies in animals have shown that Rapa-DCs traffic to lymph nodes [25, 26], but that Tol-DCs migrate preferentially to spleen [8, 20].

In response to tissue injury or detection of pathogens, tissue-resident macrophages and DCs initiate a localised inflammatory response, involving the recruitment of other immunological effector cells, resulting in secondary tissue damage. However, this destructive behaviour is usually only transient, because activated macrophages and DCs switch to an anti-inflammatory and tissue-reparative mode [44]. The ability of macrophages to both exacerbate and attenuate inflammatory reactions is neatly illustrated by their contribution to ischaemia-reperfusion injury and its resolution: Experimental depletion of macrophages prior to renal ischaemia-reperfusion injury reduces the number of infiltrating leucocytes and helps to preserve short-term kidney function [47]; however, the cost of preventing early macrophage-mediated injury is a worse tissue repair response and impaired long-term function [48]. We regard human M regs, which are activated during cell culture by adherence to plastic, serum components and IFN-γ, as belonging to the class of activation-induced (or ‘deactivated’) myeloid suppressor cells.

MDSCs represent a third class of myeloid suppressor cell, which is characterised by the expression of markers associated with myeloid progenitors and commitment to a suppressor phenotype prior to entering tissues from blood. Although MDSC populations in mice and humans are highly heterogenous, any standard definition of mouse MDSCs includes expression of CD11b and Gr-1, whereas human MDSCs universally express CD11b, CD33, CD34 and VEGFR1 [45]. It is well-established that MDSCs exert a local immunosuppressive effect within solid tumours [49, 50] and it appears that they have a similar function in transplanted organs, because induction of tolerance to kidney, skin and cardiac allografts is associated with infiltration of grafts by MDSCs [28, 51].

In principle, myeloid suppressor cells could favour allograft acceptance in several ways. Firstly, myeloid suppressor cells could exert transient, general immunosuppressive effects by secreting anti-inflammatory mediators or releasing of apoptotic debris. Secondly, myeloid suppressor cells could suppress inflammation and promote tissue repair processes within allografts during the immediate postoperative period. Thirdly, myeloid suppressor cells could anergize or delete recipient effector T cells, or induce alloantigen-specific regulatory T cells. Intuitively, the relative contribution of these allograft-protective mechanisms to transplant survival will depend on the route and timing of therapeutic cell administration, and the type of myeloid suppressor cell being used.

Arguably, the state of the immune system prior to transplantation is more conducive to the establishment of tolerance than afterwards. Under steady-state, physiological conditions, immature DCs capture and present innocuous antigens leading to the anergy or deletion of antigen-reactive effector T cells and the expansion of specific T regs [9]. Organ transplantation disrupts this homeostatic condition by causing massive inflammation and the abrupt activation of vast numbers of alloreactive recipient T cells. Hence, the rationale for using myeloid suppressor cells prior to transplantation is that enrichment of alloantigen-specific T regs and deletion of effector T cells should be easier in an immunological environment naturally predisposed to the maintenance of tolerance. In addition, conventional immunosuppressive therapy might antagonise the regulatory action of myeloid suppressor cells given after transplantation.

Whilst exploiting the pro-tolerogenic condition of the pre-transplant immune system seems a sensible therapeutic strategy, a few arguments have been levelled against this approach. Firstly, and most importantly, in order to induce antigen-specific immunological regulation prior to transplantation, it is necessary to deliberately expose the recipient to donor alloantigen, which carries an attendant risk of allo-sensitisation. Secondly, patients with end-stage organ failure awaiting transplantation may not be in a immunologically quiescent state, either because of their underlying disease or concurrent subclinical infections. Thirdly, pre-treatment with donor-derived myeloid suppressor cells is not possible in the case of transplantation from deceased donors.

From an immunological standpoint, the principal reason for using myeloid suppressor cells of donor origin is to expose the recipient to intact donor alloantigen via the direct and semi-direct pathways. Alloantigen released by donor-derived cells could also be captured and presented in a pro-tolerogenic context by immature recipient DCs via the indirect pathway [52]. Consequently, donor-derived myeloid suppressor cells find their main application in pre-transplant conditioning therapies [3, 53]. Donor-derived cells have the advantage that they can be reliably obtained from healthy, living donors. Recipient-derived myeloid suppressor cells are less likely to be eliminated by recipient T cells and NK cells, so have a greater capacity to migrate and engraft, and are less likely to sensitise the recipient against donor alloantigen. Production of recipient-derived myeloid suppressor cells prior to deceased-donor transplantation is feasible; however, since recipient-derived myeloid suppressor cells must capture and present graft-derived alloantigen in the indirect pathway, postoperative administration is the generally favoured approach of groups working with recipient-derived cells. Recipient-derived myeloid suppressor cells can be loaded with donor-alloantigen prior to infusion and this approach has proven to be a very effective therapy in animal models. Notably, antigen-pulsed tolerogenic DC were found to induce transplantation tolerance by expanding T regs which recognised alloantigen in the indirect pathway [25]. It has also been suggested that using third-party myeloid suppressor cells could eliminate some of the risks inherent to using donor or recipients cells, although it is difficult to envisage how third-party cells could induce allo-specific regulation, especially considering that the use of third-party tolerogenic DCs shows little or no effect in many animal models [20, 25, 54].

Selecting a route of administration for a tolerogenic cell therapy involves a trade-off between the most efficient means of delivering cells to their site of action and issues of clinical practicality and safety. Most animal experiments with myeloid suppressor cells have evaluated their therapeutic potential after intravenous administration [16, 20, 25], although some groups have investigated subcutaneous [55] and intramuscular injection [26]. The intravenous route seems quite suitable for M regs, since they normally traffic to liver, spleen and bone marrow. Tol-DCs and Rapa-DCs prolong allograft survival after intravenous injection; however, because recipient-derived tolerogenic DCs must capture graft antigens and suppress T cell activation in graft-draining lymph nodes, direct application of cells into lymph nodes may prove a superior route of administration. Since MDSCs exert important immunosuppressive actions within allografts, the possibility of injecting them directly into the arterial supply of the transplanted organ deserves further investigation.

As we have seen, when applying myeloid suppressor cell therapy in solid organ transplantation, the choice of myeloid suppressor cell type, whether it is given pre- or post-transplantation, whether it is of donor or recipient origin, and its route of administration are interdependent considerations (Table 2).

On the present evidence, we cannot say which cell type or clinical approach represents an optimal therapy; however, based on our preclinical animal studies and the outcomes of the TAIC-I and TAIC-II clinical trials, our research group favours the preoperative administration of donor-derived M regs. A handful of clinical trials have been conducted (or are presently underway) using tolerogenic DCs in the treatment of Type I diabetes [5] and rheumatoid arthritis [4]; however, these cells have not yet been applied in solid organ transplantation. As we describe below, M reg-containing cell preparations have now been trialled in a total of 21 renal transplant recipients (Table 3).

The TAIC-I trial was a single-centre, open-label, single-arm study with the objective of obtaining information on the safety and tolerability of administering M reg-containing cell preparations to renal transplant recipients (http://​www.​clinicaltrials.​gov, NCT00223093) [1]. A total of 12 patients receiving their first transplant from a deceased donor were enrolled in the study. Initially immunosuppression comprised tacrolimus, sirolimus and corticosteroids. From week 4 posttransplantation, patients were aggressively weaned from immunosuppressive therapy with the intention of achieving tacrolimus monotherapy (8–10 ng/ml trough level) by week 12 and further reduction to ≤4 ng/ml within 24 weeks. Patients were treated with 0.9–5.0 × 10⁸ donor-derived cells by central venous infusion at 5 days after transplantation (Figure 3). No acute complications or later adverse reactions relating to the cell infusion were observed. Thus, the TAIC-I trial demonstrated the clinical feasibility of producing and administering M reg-containing cell preparations to kidney transplant recipients.

The TAIC-II study was a phase I/II clinical trial (http://​www.​clinicaltrials.​gov, NCT00223067) designed to assess the safety and efficacy of administering donor-derived M reg-containing cell preparations to recipients of living-donor renal transplants [2]. Five days prior to surgery, five living-related kidney transplant recipients were treated with 1.4-5.9 × 10⁸ cells (Figure 4). No clinical complications of the cell infusion were observed. All patients received induction therapy with anti-thymocyte globulin (ATG) on days 0, 1 and 2. From the time of transplantation onwards, patients received a dual immunosuppressive regime comprising conventional steroid treatment and tacrolimus therapy, aiming for trough levels of 8–12 ng/ml. Steroids were weaned by week 8 post-transplantation and tacrolimus was reduced to 5–8 ng/ml over several weeks. Four patients were successfully minimised to low-dose tacrolimus monotherapy. No rejection occurred in two of five patients. Following the reduction of tacrolimus treatment to <2 ng/ml for 6 weeks, one patient underwent a rejection episode at 36 weeks. The two remaining patients experienced acute rejection episodes only after complete cessation of immunosuppression for 2 and 34 weeks. All patients in TAIC-II were monitored for indices of graft acceptance and rejection through the RISET network. None of the patients developed anti-donor HLA antibodies as a consequence of M reg administration and no accelerated graft loss occurred. Anti-donor T cell reactivity was serially assayed by MLR and was found to be consistently reduced.

Since the TAIC-I and TAIC-II clinical trials, we have arrived at a detailed understanding of the derivation, phenotype and T cell-suppressive functions of in vitro-derived human regulatory macrophages. This knowledge has inspired methodological advances in regulatory macrophage manufacture, leading to a far purer and more homogeneous cell product, which has now been applied to two further living-donor renal transplant recipients with encouraging results [3].

The first of these patients, MM, a 23 year-old female with renal failure owing to IgA nephropathy, received a living-donor kidney transplant from her 58 year-old mother. Mother and daughter had only single HLA-B and- DR mismatches (Figure 5). Six days prior to transplantation, patient MM received 8.0 × 10⁶ donor-derived M regs/kg by slow central venous infusion under cover of 2 mg/kg/d azathioprine. Conventional treatment with steroids and tacrolimus was started at the time of transplantation. Azathioprine was stopped at 8 weeks post-transplant and steroids were weaned by 14 weeks. Thereafter, MM was maintained on tacrolimus monotherapy with trough levels of less than 6 ng/ml. Protocol biopsies at 8 and 24 weeks showed no signs of rejection. At 3 years, MM was in a stable clinical condition, receiving tacrolimus 2 mg BD with trough levels of 4–5 ng/ml as her sole maintenance immunosuppression.

The second patient, CA, a 47 year-old man, received a fully-mismatched kidney from a 40 year-old living, unrelated male donor. CA was treated with 7.1 × 10⁶ donor-derived M regs/kg seven days prior to transplantation under cover of 2 mg/kg/d azathioprine. Treatment with tacrolimus and steroids began the time of transplantation. Protocol biopsies at 8, 24 and 52 weeks showed no signs of rejection. At 3 years post-transplantation, CA had stable renal function and was being maintained with sustained-release tacrolimus 5 mg OD with a trough tacrolimus level of 2.7 ng/ml.

Minimising the maintenance immunosuppression of renal transplant recipients to tacrolimus monotherapy is not recommended by the Kidney Disease Improving Global Outcomes (KDIGO) guidelines, which advise the combination of a calcineurin inhibitor and an antiproliferative agent, with or without corticosteroids [61]. Nevertheless, in practice, patients receiving tacrolimus monotherapy are not uncommon and use of Alemtuzumab induction as a means of establishing patients on maintenance tacrolimus monotherapy has stimulated much recent interest. So, should we be surprised by the clinical outcomes of MM and CA? Although, we must be cautious in our interpretation of these two case studies, there are intriguing aspects of MM and CA’s histories that deserve particular mention: Both patients were minimised to low-dose tacrolimus monotherapy; this reduction in tacrolimus dosing was made relatively early after transplantation; neither MM nor CA received conventional induction therapy; and, both MM and CA developed a peripheral blood biomarker profile converging upon the IOT-RISET signature of tolerance [62].

What constitutes low-dose tacrolimus therapy? Definitions of low-dose and standard-dose tacrolimus treatment are largely a matter of convention. The Symphony Study, which assessed whether a mycophenolate mofetil (MMF)-based regimen allows minimisation of adjunct immunosuppression, incorporated a low-dose tacrolimus arm in which 75% of patients had trough tacrolimus levels of 4.3-10.0 ng/mL [63]. By this standard, MM and CA were treated with very low-dose tacrolimus, since both patients registered drug levels in the lowest 12.5 percentile of this range [64]. Unlike subjects in the Symphony Study, patients MM and CA were not given Dacluzimab induction or maintenance MMF therapy. Viewed in this context, the fact that both MM and CA remain rejection-free and with stable graft function at >4 years post-transplantation is an encouraging outcome.

Is low-dose tacrolimus monotherapy difficult to achieve in renal transplant recipients? Shapiro’s 2003 study remains a benchmark trial of minimised tacrolimus monotherapy in renal transplant recipients [65]. 150 patients were treated with 5 mg/kg ATG and bolus prednisone as an induction therapy, and were subsequently maintained tacrolimus monotherapy, which was minimised in a step-wise fashion over many months (Figure 6). Under this regimen, 37% of patients underwent acute rejection prior to minimisation of tacrolimus dosing. 113 patients were then selected to undergo tacrolimus weaning: These patients were followed-up for a mean of 11 ± 5.4 months, during which time 23% of patients underwent acute rejection. Other studies with the aim of establishing renal transplant patients on tacrolimus monotherapy after ATG induction achieved similar outcomes [66]. More recently, Alemtuzumab induction with tacrolimus monotherapy has been used with some success [67]. Margreiter et al. reported a 20% 1-year biopsy-proven acute rejection rate in patients undergoing Alemtuzumab induction, followed by tacrolimus monotherapy with trough drug levels of 8–12 ng/ml for 6 months, reduced to 5–8 ng/ml thereafter [68]. More impressively, Chan and colleagues observed an 89.9% 2-year rejection-free renal allograft survival rate using Alemtuzumab and tacrolimus monotherapy with a target-range of 5–8 ng/ml [69]. Clearly, tacrolimus monotherapy can be achieved in renal transplant patients treated with a powerful induction agent, such as Alemtuzumab; nevertheless, it is still surprising that patients MM and CA, who were not treated with T cell-depleting monoclonal antibodies, were able to tolerate an early, fairly abrupt minimisation of immunosuppression.

A pattern of peripheral blood gene expression associated with a drug-free, tolerant state in renal transplant recipients has been defined by the IOT-RISET consortium [62]. To assess the immunological consequences of M reg therapy on patients MM and CA, expression of the ten most discriminatory gene markers of tolerance identified by the IOT-RISET group was quantified in serial blood samples taken from both patients [3]. Over the first year post-transplant, the pattern of expression of these markers gradually converged upon the IOT-RISET gene signature, suggesting that MM and CA might have achieved a degree of immunological regulation against their donors. To put this result into context, we have to consider the frequency of renal transplant recipients on CNI monotherapy whose gene expression profiles converge on the IOT-RISET signature: Of the 30 patients on CNI monotherapy included in the IOT-RISET cohort, all of whom were 6 to 9 years post-transplantation, only 5 displayed such a gene expression profile. Because the IOT-RISET study purposefully selected long-term, clinically stable patients, it is likely that 16% is an over-estimation of the frequency of CNI monotherapy patients in the general transplant population with the tolerance signature.