`
`review
`
`Advances in understanding the pathogenesis of graft-versus-
`host disease
`
`John Magenau, Lyndsey Runaas and Pavan Reddy
`
`Blood and Marrow Transplant Program, Division of Hematology/Oncology, Department of Internal Medicine, University of Michigan,
`Ann Arbor, MI, USA
`
`Summary
`
`Allogeneic haematopoietic stem cell transplantation (HCT) is
`a potent immunotherapy with curative potential for several
`haematological disorders. Overcoming the immunological
`barrier of acute graft-versus-host disease (GVHD) remains a
`fundamental impediment to expanding the efficacy of HCT.
`GVHD reflects a complex pathological interaction between
`the innate and adaptive immune systems of the host and
`donor. Over the past decade there has been a tremendous
`advancement in our understanding of the cellular and molec-
`ular underpinnings of this devastating disease. In this review,
`we cover several recently appreciated facets of GVHD patho-
`genesis including novel extracellular mediators of inflamma-
`tion, immune subsets, intracellular signal transduction, post-
`translation modifications and epigenetic regulation. We begin
`to develop general themes regarding the immunological path-
`ways in GVHD pathogenesis, discuss critical outstanding
`questions, and explore new avenues for GVHD treatment
`and prevention.
`
`Keywords: GVHD, allogeneic, transplantation.
`
`receive allogeneic
`While increasing numbers of patients
`haematopoietic stem cell transplantation (HCT) for aggressive
`haematological disorders, graft-versus-host disease (GVHD)
`remains a formidable barrier to maximizing its therapeutic
`efficacy. Despite the routine administration of immunosup-
`pressive prophylaxis, GVHD is the principle cause of trans-
`plant-related mortality (TRM). Clinically significant acute
`GVHD will occur in approximately 40% of patients undergo-
`ing human leucocyte antigen (HLA)-matched related HCT
`and upwards of 50–70% of patients receiving HLA-matched
`or -mismatched unrelated donor HCT (Jagasia et al, 2012).
`
`Correspondence: Pavan Reddy, M.D., Professor of Medicine Blood
`
`and Marrow Transplant Program, Division of Hematology/Oncology,
`
`Department of Internal Medicine, University of Michigan, 3312
`
`Cancer Center, 1500 E. Medical Center Drive, SPC 5932, Ann Arbor,
`
`MI 48109-5932, USA.
`
`E-mail: reddypr@umich.edu
`
`First published online 27 March 2016
`doi: 10.1111/bjh.13959
`
`b1 h BRITISH JOURNAL
`
`OF HAEMATOLOGY
`
`Unfortunately, less than half of the patients who develop acute
`GVHD experience a demonstrable response, partly due to the
`rapid pace of the disease as well as the marginal efficacy of pri-
`mary treatment with high dose corticosteroids (MacMillan
`et al, 2010; Choi & Reddy, 2014; Magenau & Reddy, 2014). In
`cases where organ damage is advanced or where high dose
`corticosteroid therapy is unsuccessful, mortality can exceed
`90% (Pasquini, 2008). Many patients who do ultimately
`respond to intensified immunosuppression must endure sig-
`nificant morbidity from infection, functional impairment and
`subsequent chronic GVHD. Finally, given that the efficacy of
`HCT relies heavily upon graft-versus-leukaemia
`(GVL)
`responses that are tightly linked to GVHD, intensive immune
`suppression may increase the risk of relapse and subsequent
`mortality (Rubio et al, 2015).
`
`Overview of GVHD biology
`
`The clinical features of acute GVHD manifest as an intense
`inflammatory injury primarily involving the skin,
`intestine
`and liver. Underlying
`this
`clinical presentation is
`an
`immunologically-mediated process of allogeneic donor cells
`responding to host tissues expressing polymorphic human
`leucocyte antigens (Blazar et al, 2012). The essential function
`of allogeneic donor T cells was theorized following an almost
`complete absence of GVHD in syngeneic and T cell-depleted
`HCT. However, it is now widely recognized that allogenic
`GVHD and GVL responses are fundamentally driven by ini-
`tial interactions between host and donor antigen presenting
`cells (APCs) that encounter mature T lymphocytes from the
`donor inoculum (Shlomchik et al, 1999; Matte et al, 2004;
`Reddy et al, 2005; Koyama et al, 2012; Toubai et al, 2012).
`The
`ensuing inflammatory
`cascade
`is
`self-perpetuating:
`release of pro-inflammatory cytokines results in expansion of
`alloreactive T cells with specificity to host tissues that,
`in
`turn, secrete additional inflammatory mediators.
`Over the past decade our knowledge of the intricacies that
`govern T cell and APC interactions have grown tremen-
`dously. A great number of mechanistic insights are derived
`from murine models of acute GVHD that sufficiently recapit-
`ulate acute GVHD in humans. As a consequence, a bewilder-
`ing array of cellular immune subsets, extracellular receptors,
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`
`
`Immunobiology of GVHD
`
`mediators of inflammation and molecular signalling path-
`ways are now implicated in GVHD pathogenesis. In this
`review, we focus on critical advances in the understanding of
`acute GVHD pathogenesis. Emerging mechanistic themes are
`highlighted as well as opportunities for clinical translation,
`i.e. those that shape the allogeneic immune response to sup-
`press GVHD without compromising beneficial anti-tumour
`and anti-viral immunity (Table I).
`
`Initiating events
`
`According to Billingham’s postulates made almost 50 years
`ago, immunocompetent donor cells must experience histo-
`
`compatibility differences in the host for the development of
`GVHD (Billingham, 1966). It is now well established that
`disparities in major histocompatibility complex (MHC) loci
`on chromosome 6, as occurs in HLA-mismatched HCT, are
`directly proportional
`to the
`severity of
`acute GVHD
`(Flomenberg et al, 2004; Petersdorf et al, 2015). Today, the
`majority of HCT occurs with HLA-matched unrelated
`donors,
`thus differences in other polymorphic genes, so
`called minor histocompatibility antigens (mHA), provide the
`necessary tissue disparity to produce GVHD. Furthermore,
`given that adaptive immune responses require additional sig-
`nalling events to initiate GVHD, the importance of MHC
`independent events is becoming increasingly appreciated.
`
`Table I. Select pathways of GVHD pathogenesis with potential therapeutic approaches
`
`Aberrant pathway
`
`Approach
`
`Proposed mechanism(s)
`
`Clinical trial
`(phase I-II)
`
`Reference(s)
`
`CTLA4-Ig (Abatacept)
`
`Inhibiting costimulation
`CTLA4-Ig
`(Abatacept)
`Modulate cytokine(s)/chemokine(s)
`IL6
`anti-IL6 receptor mAb
`(Tocilizumab)
`anti-IL21 mAb
`Inhibit CCR5 coreceptor
`(Maraviroc)
`
`IL21
`CCR5
`
`Decrease DAMPs/PAMPs
`↑ protease
`Exogenous Alpha-1-antitrypsin
`inhibition
`(AAT/SERPINA1)
`
`Activate Siglecs
`
`CD24-Fc
`
`Alter microbiome
`
`Antibioltics, probiotics, other
`
`Improve peripheral tolerance
`Expand/induce
`Treg graft engineering, Graft
`Tregs
`Selection, Pharmacological
`(several)
`ex vivo iNKT expansion,
`adoptive transfer
`Graft Engineering (select
`memory T cells)
`
`iNKT
`
`Decreased activation of T cells
`
`↓Th1, Th17, Tissue Damage; ↑Treg;
`
`↓ Th1, Th17; ↑ iTreg
`↓ T cell homing to Gut
`
`Suppress release of DAMPs; Limit
`cytokine production
`
`Sequester DAMPs, ↓DAMP-
`mediated DC activation
`Gut protection through modifying
`microbiome and byproducts
`
`Inactivate/delete alloreactive T cells
`
`Promotes Treg expansion
`
`Reduce alloreactive T cells
`
`Naive T cell
`depletion
`Intracellular signalling
`Tyrosine kinases
`NF-jB
`
`STATs
`Epigenetic
`DNA methylation
`
`JAK-1/2 inhibition
`c-Rel inhibition, proteasome
`inhibition (Velcade),
`neddylation inhibition
`(MLN4924)
`STAT3 inhibition
`
`Increase Treg:Teff ratios, ↓STAT
`↓transcription of multiple
`inflammatory cytokines
`
`↑ FOX P3 Expression (iTREG)
`
`Azacitidine
`
`Promotes iTreg expansion
`
`Histone acetylation Vorinostat
`
`Bromodomain
`(BET) proteins
`
`I-BET-151, JQ1
`
`Increase IDO, decrease
`inflammatory cytokines
`↓DC activation; ↓T-cell
`proliferation; NF-kB signalling
`
`DAMPs, damage-associated molecular patterns; PAMPs, pathogen-associated molecular patterns.
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`+
`
`+
`
`
`+
`
`+
`
`
`
`
`
`+
`
`
`
`+
`
`+
`+
`
`
`
`+
`
`+
`
`
`
`Miller et al (2010)
`
`Drobyski et al (2011);Kennedy
`et al (2014)
`Hippen et al (2012)
`Reshef et al (2012)
`
`Marcondes et al (2011); Tawara
`et al (2012); Brennan et al
`(2012)
`Chen et al (2009a); Toubai et al
`(2014)
`Atarashi et al (2013); Jenq et al
`(2015)
`
`Di Ianni et al (2011); Brunstein
`et al (2011)
`
`Schneidawind et al (2015)
`
`Bleakley et al (2015)
`
`Spoerl et al (2014)
`Shono et al (2014); Mathewson
`et al (2013)
`
`Laurence et al (2012)
`
`Choi et al (2010); Goodyear
`et al (2012)
`Reddy et al (2008); Choi et al
`(2014)
`Sun et al (2015a)
`
`191
`
`
`
`J. Magenau et al
`
`Costimulation
`
`While the primary interaction occurs between the MHC/al-
`lopeptide on APCs and the T cell receptor (TCR) of donor
`T cells, this signal alone is insufficient to induce T cell acti-
`vation. Several studies in HCT and related haematological
`fields have focused on the effect of the required second or
`co-stimulatory signal on T cell activation and GVHD
`(McDonald-Hyman et al, 2015). Second signals between
`APCs and T cells may positively or negatively influence the
`immune response (Fig 1A). CTLA4 is one of several co-sti-
`mulatory molecules
`expressed on differentiated T cells
`which functions as a key negative regulator of T cell activa-
`tion. CD28, a structural homolog of CTLA4, provides a
`positive co-stimulatory signal to T cells. By competing with
`greater affinity than CD28, CTLA4 preferentially binds
`ligands on APCs (B7-1/2 or CD80/86) thereby constraining
`T cell proliferation (Mueller, 2010; Fig 1B). CTLA4-immu-
`noglobulin or Abatacept is a US Food and Drug Adminis-
`tration (FDA) approved therapy for rheumatoid arthritis
`that fuses a modified Fc portion of human immunoglobu-
`lin-G1with CTLA4. Abatabcept increases the availability of
`CTLA4 which probably minimizes activating interactions
`with CD28. Based on encouraging pre-clinical evidence of
`GVHD suppression and pilot studies in humans (Miller
`et al, 2010), abatacept is currently being tested in a phase
`II multi-centre randomized study. This mechanism is nota-
`bly distinct from the alternate strategy of directly inhibiting
`CTLA4, which activates T cells (e.g.
`ipilimumab; Fig 1C).
`
`Interestingly, toxicity related to ipilimumab (without allo-
`geneic HCT) may resemble GVHD.
`Programmed cell death-1 (PD-1, also termed PCDC1) is
`another distinct inhibitory signal that has been implicated in
`maintaining peripheral
`tolerance (Mueller, 2010). Because
`PD-1 and PD-1 ligand (also termed CD274) interactions
`promote T cell exhaustion, this pathway may be physiologi-
`cally important
`for eliminating chronically self-reactive T
`cells. While pre-clinical studies after HCT suggest interrupt-
`ing these interactions may aggravate acute GVHD, delaying
`PD-1 blockade or selective interruption of PD ligands with
`differing
`tissue distribution may promote
`graft-versus-
`tumour (GVT) responses without aggravating GVHD (Koest-
`ner et al, 2011; Saha et al, 2013).
`Therapeutic
`interventions
`surrounding the TCR:MHC
`immune synapse must account for orchestrating co-stimula-
`tory and co-inhibitory signals, but also time-dependent effects
`specific to immune cells. As an example, the B7-H3 (also ter-
`med CD276) membrane protein, widely expressed on T cells,
`natural killer cells, dendritic cells (DCs) and macrophages, is
`an activating co-stimulatory signal that mediates transplant
`rejection (Wang et al, 2005). However, B7-H3 can also effec-
`tively curtail alloreactive T cells and GVHD in the early HCT
`period (Veenstra et al, 2015) and donor lymphocyte infusion
`with T cells lacking B7-H3 promotes GVL responses without
`GVHD. These studies illustrate the time dependence of co-sti-
`mulatory pathways, thus agonism or antagonism both may
`present viable therapeutic strategies after HCT.
`
`Fig 1. Co-stimulatory interactions between
`APCs & T cells. Co-stimulatory ligands and
`receptors (A) provide positive or negative sec-
`ond signals together with the TCR that deter-
`mine the activation status of T cells. Clinically
`available pharmacological agents modulate co-
`stimulatory interactions on APCs and T cells.
`Interrupting B7:CD28 interactions (B) can sup-
`press T cells through loss of activating second
`signals, potentially impeding graft-versus-host
`disease. Conversely, targeting CTLA4 or
`PDCD1 (C) can activate T cells through blunt-
`ing inhibitory second signals. MHC, major his-
`tocompatibility complex; TCR, T-cell receptor;
`APC, antigen presenting cell.
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`192
`
`
`
`Mediators of inflammation: cytokines and chemokines. Once
`activated, T cells initiate transcriptional programmes that
`result in the massive release of pro-inflammatory mediators
`[tumour necrosis factor a (TNF-a, TNF), interleukin 1 (IL1),
`c interferon (IFN-c,
`IFNG)]
`that amplify the immune
`response and result in tissue damage (Paczesny et al, 2010).
`In addition, cytokines and chemokines influence prolifera-
`tion, differentiation and homing of effector cells to GVHD
`target
`tissues. This knowledge laid the foundation for
`attempts at blocking single cyto/chemokines to ameliorate
`GVHD. This approach has been met with modest clinical
`success, but
`is gaining renewed interest
`from pre-clinical
`studies suggesting that early interruption of novel cytokines
`can impede multiple cellular pathways in GVHD.
`Interleukin 6 (IL6): There is compelling evidence that IL6
`promotes GVHD by divergent mechanisms including increas-
`ing inflammatory T-helper cells types 1 and 17 (Th1 and
`Th17, respectively) subsets, decreasing regulatory T cells
`(Tregs) and by direct cytotoxicity (Chen et al, 2009b; Tawara
`et al, 2011). In models of HCT, levels of IL6 are elevated,
`following cytotoxicity from conditioning or GVHD.
`IL6
`blockade suppresses experimental GVHD without impairing
`the GVL response. Given the clinical availability of
`the
`humanized anti-IL6 receptor antibody,
`tocilizumab,
`for
`rheumatoid arthritis, IL6 blockade was piloted in steroid-
`refractory acute GVHD without untoward toxicity (Drobyski
`et al, 2011). Because IL6 is elevated early after HCT, a single
`centre phase I/II trial evaluated whether early intervention
`with tocilizumab might reduce GVHD in HLA-matched
`related and unrelated donors (Kennedy et al, 2014). Correla-
`tive studies revealed that downstream STAT3 was reduced in
`monocytes and T cells from patients receiving tocilizumab,
`implying successful IL6 blockade. The rate of grade II-IV
`GVHD was 12% and severe grade III-IV GVHD was 4%,
`supporting additional studies in GVHD prevention.
`Interleukin-21 (IL21):
`is capable of promoting Th1 and
`Th17 differentiation, NK cell expansion and formation of
`inducible Tregs (iTregs) in addition to impacting a wide
`range of
`lymphoid and myeloid cells
`including APCs.
`Administration of IL21 inhibitors or IL21-deficient T cells
`mitigates gastrointestinal
`(GI) GVHD without
`impairing
`GVL responses, perhaps due to its tissue-specific effects
`(Bucher
`et al, 2009).
`In humans,
`IL21 expression was
`increased in the GI tracts of patients with GVHD. In a xeno-
`graft model, prophylactic administration of anti-human IL21
`reduced GVHD lethality (Hippen et al, 2012). Exogenous
`IL21 is currently being evaluated in immunotherapy trials for
`cancer, while IL21 blocking strategies are being developed for
`autoimmune conditions.
`Interleukin-2 (IL2): Interrupting IL2 secretion by donor T
`cells, via the calcineurin inhibitors (CNIs), ciclosporin and
`tacrolimus, is an established standard of care in GVHD pre-
`vention. However, subsequent randomized trials directly tar-
`geting
`the
`IL2 receptor
`(i.e. daclizumab) have been
`unsuccessful in acute GVHD treatment, partly due to high
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`Immunobiology of GVHD
`
`rates of relapse (Lee et al, 2004). While CNIs reduce effector
`T cell function, an unintended consequence of prolonged IL2
`inhibition may be that Treg generation is constrained. In
`support of
`this hypothesis, murine data and a recent
`prospective trial demonstrated that administration of
`low
`dose IL2 in chronic GVHD results in regression of disease
`(Koreth et al, 2011). The mechanism of IL2 may relate to
`differential effects on STAT5 proteins in dividing lympho-
`cytes that restore a more favourable balance between Tregs
`and conventional T cells (Matsuoka et al, 2013).
`Chemokine (C-C motif) receptor-5 (CCR5): CCR5 is upreg-
`ulated in T lymphocytes upon allogeneic stimulation and
`directs homing to target tissues such as the GI tract. HCT
`recipients with a missense mutation of CCR5 (Delta32)
`appear less susceptible to developing clinical GVHD (Bogu-
`nia-Kubik et al, 2006). A phase I/II clinical trial using a small
`molecule antagonist of CCR5 (maraviroc) demonstrated
`promising results for GVHD prevention (Reshef et al, 2012)
`and is now being prospectively evaluated in a larger clinical
`trials network (CTN) study (NCT02208037).
`Interleukin-22 (IL22): Our understanding of cytokines has
`expanded to include identification of molecules capable of
`tissue repair. IL22 acts upon intestinal stem cells (ISCs) that
`are critically involved in epithelial repair. Because the GI
`tract is exquisitely sensitive to the cytotoxic effects of condi-
`tioning, its damage responses are instrumental for propagat-
`ing GVHD. Deficiency of IL22 promotes loss of ISCs, greater
`intestinal damage and more severe experimental GVHD
`(Hanash et al, 2012).
`Suppression of Tumourigenicity 2 (ST2): ST2 functions as
`both a soluble (sST2) and membrane bound receptor (ST2L)
`for IL33, a cytokine in the IL1 receptor family with diverse
`immunological
`roles depending on disease, cell
`type or
`model system. Using proteomic methods to compare plasma
`levels prior to and during GVHD, elevations in sST2 have
`been identified as an independent biomarker of treatment
`resistance and mortality (Vander Lugt et al, 2013). While
`high ST2 levels predict GVHD mortality, non-specific tissue
`damage or the genetic background of the host could also
`influence plasma concentrations (Ito & Barrett, 2015). High
`levels of ST2 and IL33 are produced by non-haematopoietic
`cells in the GI tracts of animals with GVHD (Reichenbach
`et al, 2015; Zhang et al, 2015). IL33 binding to ST2 aggra-
`vates GVHD, which can be reversed in IL33( / ) deficient
`hosts or in T cells lacking ST2 expression. In pre-clinical
`models, minimizing IL33/ST2 interactions by administering
`an ST2-Fc fusion protein reduced GVHD.
`FTY720 (FTY): FTY is an immunomodulator derived from
`a metabolite of the fungus Isaria sinclairii. It binds with high
`affinity to sphingosine 1-phosphate receptors found on all
`cell types and is thought to exert its immunomodulatory
`effects through sequestering of lymphocytes within secondary
`lymphoid organs. Given this, there was interest in evaluating
`FTY to prevent or treat GVHD as an oral FTY (fingolimod)
`was approved by the FDA in 2010 to treat relapsing multiple
`
`193
`
`
`
`J. Magenau et al
`
`sclerosis. Taylor et al (2007) were able to demonstrate that
`FTY significantly inhibited but did not completely prevent
`GVHD in a model of GVHD. Similarly, administration of
`FTY slightly mitigated but did not eliminate GVL effect
`(Taylor et al, 2007).
`Retinoic acids: Retinoic acid (RA) is produced by intestinal
`cells and is known to have a role in intestinal
`immune
`homeostasis; however, its exact role in the pathophysiology
`of GVHD remains uncertain. Nishimori et al (2012) demon-
`strated that administration of a RA analogue ameliorated
`cGVHD. The effects on acute GVHD appear more nuanced
`with administration of a vitamin-A deficient diet reducing
`gut acute GVHD but worsening liver GVHD (Koenecke et al,
`2012). Similarly, Aoyama et al (2013) found that inhibiting
`donor T-cell RA receptor signalling reduced GVHD while
`preserving graft-versus-lymphoma effects.
`
`Modulating innate immune responses
`
`As DCs are primary sensors of initial inflammatory signals,
`their regulation can significantly shape GVHD. Currently,
`several efforts are underway to understand if modulating DC
`function can be utilized for treating GVHD.
`
`Damage-associated molecular patterns (DAMPs)
`
`Immune and myeloablative conditioning therapy preceding
`HCT provides important anti-tumour effects, but also causes
`direct immune activation by tissue injury and release of pro-
`inflammatory mediators. Large scale observational studies
`
`have confirmed a modest yet significant increase in acute
`GVHD incidence with higher conditioning intensity, espe-
`cially in regimens containing total body irradiation (TBI)
`(Nakasone et al, 2015). A growing body of evidence suggests
`that highly conserved Toll-like receptors (TLRs) and other
`pattern recognition receptors on innate immune cells acutely
`sense endogenous ‘danger’ signals from DAMPs, such as
`ATP, HMGB1, uric acid (UA) and heat-shock proteins, that
`provide a crucial
`initiating step in alloreactive T cell
`responses (Wilhelm et al, 2010; Jankovic et al, 2013; Brennan
`et al, 2015; Fig 2).
`The purine nucleoside ATP, when present in the extracel-
`lular space following tissue injury, is one of several early dan-
`ger signals or DAMPs. ATP released into peritoneal fluids of
`humans and mice following radiation interacts with P2X7
`causing expression of co-stimulatory molecules, phosphoryla-
`tion of
`signal
`transducer and activator of
`transcription
`(STAT) proteins and production of inflammatory cytokines.
`Antagonizing ATP:P2X7 interactions
`limits mortality in
`models of GVHD (Wilhelm et al, 2010). Other DAMPs are
`increasingly being recognized as initiators of the alloimmune
`response. For example, tissue injury releases the extracellular
`matrix component UA, which, in turn, activates the NLRP3
`inflammasome (Jankovic et al, 2013). Mice under conditions
`of less UA or deficient in components of the NLRP3 complex
`demonstrate less severe GVHD. Heparin sulfate (HS), also a
`DAMP, is similarly capable of activating alloreactive T cell
`responses through binding TLR4 receptors on DCs and ele-
`vated levels correlate with GVHD (Brennan et al, 2012).
`While inhibiting single DAMPs is effective in suppressing
`
`Cytokines
`
`•• • A..._ Tocilizumab (anti-lL6)
`
`Maraviroc
`
`Activation of Adaptive
`Immunity
`
`AAT
`CD24Fc
`
`Activation of Innate Immunity
`
`Fig 2. Role of DAMPs & Novel Cytokines in GVHD Pathogenesis. Tissue damage following conditioning (chemotherapy/radiation) results in the
`release of numerous sterile inflammatory mediators, termed damage-associated molecular patterns (DAMPs), that together with cytokines con-
`tribute to the initiation of acute GVHD. Inflammatory mediators activate innate immunity through interactions with toll-like receptors (TLR)
`and cytokine receptors (not depicted) on APCs. These interactions promote activation of the adaptive immune response characterized by T cell
`differentiation, proliferation, and migration that perpetuates GVHD and worsens tissue damage. Certain investigational agents (e.g. AAT,
`CD24Fc) may mitigate GVHD by either sequestering DAMPs or regulating APC-mediated responses to DAMPs. GVHD, graft-versus-host disease;
`APC, antigen presenting cell; AAT, alpha-1-antitrypsin; CD24Fc, CD24 fusion protein; HS, heparin sulfate; UA, uric acid; HS, heat shock protein.
`
`194
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`
`
`translation may
`clinical
`successful
`experimental GVHD;
`require inhibiting many DAMPs and their recognition by
`immune cells. Furthermore, it remains to be determined if
`DAMPs play as pivotal a role in settings of reduced intensity
`conditioning where tissue damage is less pronounced.
`
`Alpha-1-antitrypsin (AAT, also termed SERPINA1). AAT is
`an abundant serine protease inhibitor produced as an acute
`phase reactant by the liver which is classically described for
`its role in inhibiting neutrophil elastase. Congenital AAT
`deficiency results in emphysema, particularly under condi-
`tions of chronic noxious stimuli, such as cigarette smoke
`(Silverman & Sandhaus, 2009). Over the past several years,
`numerous additional roles of AAT have been identified, such
`as promoting tolerance in models of islet cell transplantation
`(Koulmanda et al, 2008). In allogeneic HCT models, human
`AAT administration reduces GVHD, proportionally increases
`Tregs without
`impairing T cell responses and suppresses
`multiple pro-inflammatory cytokines (Tawara et al, 2012).
`Similar results have been observed by other groups (Marcon-
`des et al, 2011; Brennan et al, 2012). The mechanisms by
`which AAT attenuates DC responses remain uncertain, but
`may involve less tissue release of DAMPs, such as HS, (Bren-
`nan et al, 2012) or by limiting the glycolytic capacity of
`metabolically active T cells (Marcondes et al, 2014).
`In
`humans, stool levels of AAT are increased in patients with
`severe GI GVHD (Rodriguez-Otero et al, 2012), suggesting
`patients may have a relative deficiency or be less able main-
`tain tissue concentrations that exert a protective effect. Given
`these observations, proof of concept clinical trials of exoge-
`nous AAT supplementation in steroid-refractory GVHD are
`now underway.
`
`Sialic-acid–binding
`lectin-G (Siglec-
`immunoglobulin-like
`G). Siglec-G residues, broadly expressed on B-cells, DCs and
`macrophages, serve as key negative regulators of DAMP dri-
`ven immune activation (Crocker et al, 2007). Recent pre-
`clinical data suggest Siglec-G expression is important
`for
`maintaining protective effects against GVHD (Toubai et al,
`2014). Following HLA-matched and -mismatched HCT,
` /
`Siglec-G
`deficient animals have heightened release of mul-
`tiple pro-inflammatory cytokines, stimulate greater numbers
`of alloreactive T cells and ultimately experience inferior sur-
`vival due to increased GVHD severity. Host tissue injuries
`caused by myeloablative HCT regimens, including high-dose
`chemotherapy and/or TBI not only promote release of
`DAMPs, but also reduce Siglec-G expression (Toubai et al,
`2014). Thus, conditioning therapy,
`through its effect on
`siglecs, may limit the capacity of immune cells to attenuate
`the damage response. The membrane glycoprotein CD24,
`expressed on haematopoietic cells including T lymphocytes,
`directly binds DAMPs and Siglec-G down-regulating immune
`stimulation (Liu et al, 2009).
`It is postulated that the CD24-Siglec-G pathway plays a
`critical role in preventing exaggerated responses to pathologi-
`
`ª 2016 John Wiley & Sons Ltd
`British Journal of Haematology, 2016, 173, 190–205
`
`Immunobiology of GVHD
`
`cal cell death, and most importantly, discriminating between
`exaggerated damage responses (i.e. DAMPs) while retaining
`immunity against pathogens (i.e. pathogen-associated molec-
`ular patterns, PAMPs) (Chen et al, 2009a). As HCT is associ-
`ated with tissue injury that additionally places hosts at risk
`for infection, the CD24-Siglec-G pathway is a compelling tar-
`get for GVHD prevention after myeloablative conditioning.
`For example, CD24Fc (CD24Fc immunoglobulin) is a fully
`humanized fusion protein consisting of
`the extracellular
`domain of mature CD24 linked to the human immunoglob-
`ulin G1 (IgG1) Fc domain. Similar to native CD24, in vitro
`studies demonstrate that CD24Fc binds murine Siglec-G
`(and its human orthologue, Siglec 10) (Chen et al, 2009a).
`In pre-clinical models, administration of exogenous CD24Fc
`promoted Siglec G activation and prevented GVHD (Toubai
`et al, 2014). Clinical studies of CD24Fc for GVHD preven-
`tion are currently underway in the settings of heightened tis-
`sue injury, such as myeloablative HCT. Nonetheless, whether
`enhanced signalling through other negative immune regula-
`tory pathways besides Siglecs can also mitigate GVHD
`remains to be explored.
`
`Role of microbiome in GVHD
`
`The interplay of the intestinal microbiome and GVHD has
`been a topic of investigation since the 1970s when van Bek-
`kum (1977) observed delayed GVHD after gut decontamina-
`tion. Rigorous examination of
`the relationship between
`GVHD and microbiota has only been possible with recent
`advances
`in technology allowing for culture-independent
`rRNA gene sequencing. With this new tool,
`investigators
`have noted a profound loss of bacterial diversity in murine
`models of GVHD (Eriguchi et al, 2012; Jenq et al, 2012;
`Tawara et al, 2013). Holler et al (2014) demonstrated that
`eliminating certain bacterial species prior to HCT, such as
`Lactobacillus, correlates with increased GVHD pathology.
`This process can be reversed by re-introduction of these
`organisms, suggesting that loss of certain commensal bacteria
`may promote GVHD severity. Similarly, the human micro-
`biome shifts towards potentially pathogenic organisms, such
`as Enterococci, after HCT. This shift is pronounced after
`antimicrobial usage and most notable after developing
`intestinal GVHD (Holler et al, 2014). Recently,
`increased
`bacterial diversity was associated with reduced mortality
`(Jenq et al, 2015). Specifically, higher proportions of bacteria
`belonging to the genus Blautia was linked to decreased
`GVHD-specific mortality and improved survival. Of particu-
`lar clinical interest, loss of Blautia is associated with use of
`antibiotics targeting anaerobes as well as parenteral nutrition;
`suggesting two potential
`interventions
`to mitigate acute
`GVHD. However, murine experiments suggest that the donor
`microbiome does not seem to affect GVHD severity in the
`hosts (Tawara et al, 2013).
`Given these findings, work is underway to actively manip-
`ulate the intestinal microbiome in a favourable manner.
`
`195
`
`
`
`J. Magenau et al
`
`Atarashi et al (2013) identified a cocktail of 17 species of
`Clostridia that potently enhance host Treg cell abundance
`and can reverse autoimmune colitis. In a murine model of
`GVHD, administration of this same cocktail was shown to
`reduce GVHD and improve survival
`(Mathewson et al,
`2014). Despite advances in technology that have allowed a
`better understanding into the microbiome, many questions
`remain, including the effect of prophylactic antimicrobials,
`optimal strategies for protection of regenerative cell popula-
`tions and advisability of the administration of exogenous
`commensal flora. These remain areas of active research.
`
`Cellular mediators of peripheral tolerance
`
`The actions of inflammatory signals play a large role in shap-
`ing the cellular milieu that follows HCT. As adults have
`undergone thymic involution, and mature lymphocytes from
`the donor inoculum drive GVHD, post-HCT immunity lacks
`normative central tolerance mechanisms, such as negative
`selection, that delete reactive T cells and B cells. These condi-
`tions imply a greater role for peripheral immune tolerance
`mechanisms in regulating the allogeneic responses. Induction
`of peripheral tolerance to reduce GVHD are being actively
`explored.
`
`Expansion of Tregs
`
`Tregs serve a crucial role in maintaining peripheral self-toler-
`ance. Natural Tregs are characterized by intracellular expres-
`sion of the transcription factor forkhead box P3 (FOXP3). In
`HCT, murine
`studies
`show that
`adoptive
`transfer of
`CD4+CD25+ Tregs protects against lethal GVHD (Edinger
`et al, 2003). A notable observation is that the suppressive
`effects of Tregs appear to have a nominal impact on immune
`recovery and GVL (Blazar et al, 2012). However, evaluating
`this in patients will require methods for generating high
`ratios of Tregs to T effectors that are sustainable in vivo. One
`approach is ex vivo expansion with adoptive transfer. In the
`setting of HLA-mismatched haploidentical HCT, infusion of
`donor Tregs prior to CD34+ selected stem cells resulted in
`very low rates of acute GVHD, without standard prophylactic
`immunosuppression (Di Ianni et al, 2011). In double cord
`unit HCT, escalating doses of third party expanded Tregs
`limited GVHD compared to historical controls (Brunstein
`et al, 2011). As further refinements occur, production of
`more stable populations of Tregs is anticipated, which will
`enable more definitive clinical studies of GVHD prevention.
`Nevertheless, these bench to clinic translations provide strik-
`ing examples of the potential of cellular therapy to mitigate
`human disease.
`
`Inducible Tregs
`
`Another approach to improving the availability of Tregs is
`the conversion of conventional T cells into iTregs. In fact,
`
`196
`
`some pharmacological agents described in this review exert
`their suppressive actions through iTreg expansion. The pro-
`cess of conferring antigen specificity to iTregs, as a means of
`improving the efficiency and selectively of their suppressive
`function is also being explored. For example, iTregs specific
`for the naturally occurring male Y chromosome minor histo-
`compatibility antigen (miHAg), termed HY, can be generated
`to prevent experimental GVHD (Li et al, 2015). As antici-
`pated, HY-specific iTregs exhibit enhanced expansion and
`stability in male hosts, making it conceivable that these tech-
`niques could expand Tregs against other relevant mHA.
`Whether iTregs and conventional T cells require simil