Deregulation of innate immune and inflammatory signaling in myelodysplastic syndromes

Abstract

Myelodysplastic syndromes (MDSs) are a group of heterogeneous clonal hematologic malignancies that are characterized by defective bone marrow (BM) hematopoiesis and by the occurrence of intramedullary apoptosis. During the past decade, the identification of key genetic and epigenetic alterations in patients has improved our understanding of the pathophysiology of this disease. However, the specific molecular mechanisms leading to the pathogenesis of MDS have largely remained obscure. Recently, essential evidence supporting the direct role of innate immune abnormalities in MDS has been obtained, including the identification of multiple key regulators that are overexpressed or constitutively activated in BM hematopoietic stem and progenitor cells. Mounting experimental results indicate that the dysregulation of these molecules leads to abnormal hematopoiesis, unbalanced cell death and proliferation in patients' BM, and has an important role in the pathogenesis of MDS. Furthermore, there is compelling evidence that the deregulation of innate immune and inflammatory signaling also affects other cells from the immune system and the BM microenvironment, which establish aberrant associations with hematopoietic precursors and contribute to the MDS phenotype. Therefore, the deregulation of innate immune and inflammatory signaling should be considered as one of the driving forces in the pathogenesis of MDS. In this article, we review and update the advances in this field, summarizing the results from the most recent studies and discussing their clinical implications.

INTRODUCTION

Myelodysplastic syndromes (MDS) are a heterogeneous group of hematologic stem cell malignancies clinically characterized by cytopenias associated with defective hematopoiesis, myeloid dysplasia and an increased risk of transformation to acute myelogenous leukemia (AML). One distinctive feature of this malignancy is the presence of increased apoptosis in bone marrow (BM), which is, in contrast, generally hypercellular, although it can also be normal or hypocellular. The heterogeneity of the disease characteristics complicates the diagnosis and management.¹ Although significant efforts to understand the pathophysiology of MDS made during the past decade have led to the identification of key genetic and epigenetic alterations in patients,² the definite pathogenetic mechanisms of MDS are still not fully understood. Except for a small subset that is eligible for stem cell transplantation, most MDS patients have a poor prognosis because frontline pharmacological therapies are not curative owing to the lack of well-defined molecular targets.^1,2 Thus, there is an urgent need to characterize the molecular mechanisms involved in the pathogenesis of MDS to allow the establishment of good diagnostic protocols and specific and effective targeted therapies.

Recent clinical and molecular studies of MDS have yielded accumulating evidence suggesting that abnormal activation of innate immune signals and associated inflammation contribute to the pathogenesis of this disease. These new findings have improved our understanding of the molecular mechanisms triggering MDS and have led to the development of promising therapeutic strategies. This article will review this topic and present the most recent and significant studies, summarizing their results and discussing their clinical implications and therapeutic applications.

THE INFLAMMATORY AND AUTOIMMUNE NATURE OF MDS

Association with inflammatory and autoimmune disorders

Case-report studies have traditionally associated MDS with the coexistence of other inflammatory disorders, which have an incidence between 10 and 30% in MDS patients and even higher in chronic myelomonocytic leukemia.³

Very early in the study of MDS, clinicians noticed its frequent association with rheumatic manifestations, especially rheumatoid arthritis. Although this is a common disease in the elderly, the observations suggested that those associations were not fortuitous.⁴ Interestingly, a recent literature review reported that arthritis preceded MDS in 55% of cases, and that both the pathologies were concomitantly diagnosed in 27% of cases,⁴ which suggests that the existence of inflammation precedes the appearance of MDS. Similarly, MDS and inflammatory bowel disease (IBD) are frequently diagnosed simultaneously. However, in this case, many patients diagnosed with IBD frequently presented with clinical manifestations of MDS before diagnosis.^5–7 Other acute and chronic autoimmune disorders associated with MDS are diverse types of vasculitis, autoimmune anemias, several rheumatic and skin disorders and certain thyroid diseases.^3,8,9

Larger-scale epidemiologic studies have confirmed that patients with autoimmune disorders have an increased risk of developing MDS when compared with matched controls. A 5-year case–control study including 84 MDS patients was the first to report a statistically significant association between preexisting autoimmune disorders and MDS.¹⁰ Later, a large retrospective analysis of the population-based case–controlled Surveillance Epidemiology and End Results-Medicare database, which included 2471 MDS patients, confirmed that the risk of developing AML and MDS is associated with the preexistence of an autoimmune condition.¹¹ Particularly, the association between MDS and rheumatoid arthritis was further supported by a long-term follow-up study that used a cohort of 91 291 patients with rheumatoid arthritis or osteoarthritis who had received a knee arthroplasty.¹² This study revealed an especially high incidence of MDS in those patients, pointing to arthritis rather than the surgical procedure as the responsible factor. In parallel, a population study of the central registries in Sweden validated the significant associations between MDS and most of the other disorders mentioned before.¹³

Interestingly, IBD is the only of the above mentioned conditions for which an association with the risk of MDS has not been confirmed by large-scale population studies.^14,15 Indeed, patients frequently present with clinical manifestations of MDS before diagnosis of IBD.^5–7 This has led to the speculation that MDS and IBD could have a common pathogenesis. For instance, one case study reported that the same abnormal BM karyotype is associated with the development of both MDS and IBD.⁷

Besides autoimmune diseases, more recent epidemiologic studies support the increased risk of MDS in patients affected by acute and chronic infections. Two similar population studies carried out with data from the Swedish registries¹³ and the American Surveillance Epidemiology and End Results-Medicare database¹⁶ analyzed the occurrence of infectious diseases in 1662 and 3072 patients with MDS, respectively. Both studies showed that history of infection was significantly associated with a higher risk of MDS, particularly with several infections of the respiratory tract. Interestingly, risk of MDS was consistent and, in some cases, even higher when longer periods of latency were considered. This may suggest that chronic infections make patients more susceptible of developing MDS.

Taken together, these data strongly indicate that some inflammatory and autoimmune disorders favor the development of MDS. This causal association may be a consequence of the pharmacological treatment of such disorders, but the associations are not specific to treated conditions. Other factors that deserve to be studied in depth are the common genetic predisposition to autoimmune alterations and MDS and the possibility that the underlying inflammatory/autoimmune conditions can directly damage BM precursors and drive malignant transformation.^11,16 Provided that the acquisition of certain infections is also related to a high MDS risk, the latter option seems more compelling, although the two situations are not mutually exclusive.

Even though the idea of autoimmune disorders being a direct cause of MDS is very intriguing, the existing literature reviews are not conclusive about the prognostic meaning of these pathologies in MDS or about their association with the MDS subtypes or cytogenetics; therefore, epidemiologic data need to be carefully interpreted and studied coordinately with mechanistic data.

Abnormal levels of cytokines and chemokines

The presence of abnormal levels of cytokines, chemokines and growth factors in the peripheral blood and BM of MDS patients has been extensively documented (Table 1). In general, levels of tumor necrosis factor alpha (TNF-α), interferon gamma (IFN-γ), transforming growth factor beta (TGF-β), interleukin (IL)-6, IL-8 and the myeloid growth factors granulocyte colony-stimulating factor and macrophage colony-stimulating factor, among others, are increased in MDS patients,^17–27 which reflects a profound dysregulation of both inflammatory signaling and myeloid differentiation. Moreover, increased levels of some of these cytokines can affect the clinical outcomes of patients. Higher levels of serum TNF-α are a potential adverse prognostic factor in AML and high-risk MDS and are associated with higher leukocyte counts and higher levels of β2-microglobulin, creatinine, uric acid and alkaline phosphatase.²⁸ Similarly, TNF-α, IL-6 and IL-1 receptor (IL-1R) levels have been related to ratings of fatigue in MDS.²⁹

Table 1.

Cytokines and growth factors differentially expressed in MDS

Cytokine/factor	Levels	Tissue	Plasma/cell type	Associations	References
GM-CSF	Normal/low	PB, BM	Plasma		Shetty et al.17
EGF	Low	PB	Plasma		Feng et al.31
CXCL5	Low	PB	Plasma		Feng et al.31
IL-10	Low	PB	Plasma		Kornblau et al.30
CCL5	Low	PB	Plasma		Pardanani et al.32
M-CSF	High	BM	Mononuclears		Allampallam et al.18
G-CSF	High	PB	Plasma	Direct: BM cellularity	Kornblau et al.,30 Feng et al.,31 Pardanani et al.32
TNF-α	High	BM, PB	Whole BM aspirates, BM plasma, BMMC cultures, fibroblast cultures, macrophage cultures, PB plasma	Direct: macrophage count, rate of BM apoptosis, FAB subtypesa, WHO stratification, fatigue; reverse: Hb levels, patient survival	Shetty et al.,17 Allampallam et al.,18 Kitagawa et al.,19 Deeg et al.,20 Flores-Figueroa et al.,21 Gersuk et al.,22 Sawanbori et al.,23 Tsimberidou et al.,28 Meyers et al.,29 Feng et al.,31 Pardanani et al.32
IFN-γ	Higha	BM, PB	Whole BM aspirates, mainly in myeloid cells		Kitagawa et al.,19 Pardanani et al.32
TGF-β	Higha	BM	Whole BM aspirates		Shetty et al.,17 Qadir et al.25
IL-1α/1βR	Higha	BM, PB	Mononuclears, parenchyma, PB plasma	Direct: fatigue	Allampallam et al.,18 Meyers et al.,29 Kornblau et al.,30 Feng et al.,31 Pardanani et al.32
IL-1RAP	High	BM	Stem and progenitor cells	IPSS risk (high)	Barreyro et al.60
IL-4	Higha	PB	Plasma		Pardanani et al.32
IL-6	Higha	BM, PB	Plasma, BMMC cultures, fibroblast cultures	Direct: fatigue; reverse: survival	Flores-Figueroa et al.,21 Hsu et al.,24 Meyers et al.,29 Feng et al.,31 Pardanani et al.32
IL-7	High	PB	Plasma	Reverse: survival	Pardanani et al.32
IL-8	Higha	BM, PB	Plasma	Direct: WHO stratification, IPSS risk	Hsu et al.,24 Kornblau et al.,30 Pardanani et al.32
IL-12	High	PB	Plasma		Kornblau et al.30
IL-13	High	PB	Plasma		Pardanani et al.32
IL-15	High	PB	Plasma		Kornblau et al.,30 Pardanani et al.32
IL-17	High	PB	Plasma		Kordasti et al.33
VEGF	High	BM, PB	PB serum, whole BM aspirates		Brunner et al.,26 Verstovsek et al.,27 Pardanani et al.32
ANG	High	PB	Plasma		Brunner et al.26
CXCL10	Higha	PB	Plasma	Direct: circulating blasts, thrombocytopenia; reverse: survival	Kornblau et al.,30 Pardanani et al.32
CCL3	High	PB	Plasma		Feng et al.31
CCL4	High	PB	Plasma		Feng et al.31
HGF	High	PB	Plasma		Feng et al.,31 Pardanani et al.32
MIP-1β	High	PB	Plasma		Pardanani et al.32
MIG	High	PB	Plasma		Pardanani et al.32
Eotaxin	High	PB	Plasma		Pardanani et al.32
MCP-1	High	PB	Plasma		Pardanani et al.32

Abbreviations: ANG, angiogenin; BM, bone marrow; CCL, C-C motif ligand; CXCL, C-X-C motif ligand; EGF, epidermal growth factor; FAB, French-American-British classification; HGF, hepatocyte growth factor; PB, perpheral blood; G-CSF, granulocyte colony-stimulating factor; Hb, hemoglobin; IFN-γ, interferon gamma; IL-1RAP, interleukin-1R accessory protein; MCP, monocyte chemotactic protein 1; MDS, myelodysplastic syndrome; MIG, monokine induced by interferon γ; MIP-1β, macrophage inflammatory protein; M-CSF, macrophage colony-stimulating factor; TGF-β, transforming growth factor beta; VEGF, vascular endothelium growth factor; WHO, World Health Organization.

^aDiscrepancies exist among different studies: one work reported IFN-γ, IL-4 and IL-6 to be downregulated and IL-1 to be expressed at normal levels;³⁰ the direct association of TNF-α with FAB subtypes was not confirmed by some groups;^21,30 no differences were found in IL-8 and CXCL10 levels in one study.³¹

Several recent studies have utilized comprehensive approaches, such as multiplex-based analyses, to systematically determine the association between elevated cytokine levels and the clinical characteristics and outcomes of MDS patients. A parallel profiling of 27 cytokines/chemokines in the peripheral plasma of 114 MDS patients by Kornblau et al.³⁰ revealed that the mean expression of several cytokines was significantly higher in MDS, whereas the pleiotropic cytokines IL-10 and IL-4 were expressed at lower rates and directly correlated with patient survival. Kornblau et al.³⁰ further clustered the cytokines into nine recurrent expression patterns as `cytokine signatures' and showed that eight of them had prognostic implications, including effects on remission, primary resistance, relapse rates and overall survival. However, there were some remarkable discrepancies between the results of this work and those of previous clinical studies. Indeed, some of the cytokines found to be highly produced in MDS before, such as IL-6 or IFN-γ, appeared to be downregulated in this study. In turn, two later large-scale studies in the peripheral blood of 57 MDS patients³¹ and BM of 78 patients³² analyzed the expression of 32 and 30 cytokines, respectively, and both studies confirmed previous clinical data, in disagreement with the report by Kornblau et al.³⁰

Discrepancies between studies can largely be ascribed to the fact that they analyze MDS subtypes with different rates of apoptosis. Whereas low-risk disease (or subtypes without excess blasts such as refractory anemia, RA) is characterized by an elevated apoptotic index, high-risk MDS and the subtypes with high counts of blasts (RA with excess blasts, RAEB) are associated with more aggressive clonal expansion, tolerance to self-immunity and poor response to immunosuppressive therapy (IST). The occurrence of apoptosis in MDS BM is closely associated with TNF-α levels.¹⁷ Thus, the secretion of TNF-α and other related cytokines, such as IFN-γ or IL-6, is higher in low-risk MDS, whereas these and other cytokines are more likely to be downregulated in high-risk cases. Likewise, immunosuppressive cytokines, such as IL-10, are more intensely secreted in high-risk MDS, in which the survival of the malignant clone is vital for the progression of the disease.³³ Thus, cytokine secretion profiles vary between types of MDS, and this fact might be the origin of the discrepancies among some of the cytokine profiling studies summarized above. For instance, patients with RAEB are particularly abundant in the cohort studied by Kornblau et al.,³⁰ which might explain their discrepant results.

Of note, peripheral immunosuppresor metabolites derived from the indoleamine 2,3-dioxygenase-1 (IDO/IDO1)-mediated tryptophan metabolism have also been found in high levels in MDS patients, although this increase was irrespective of the risk score.³⁴

DEREGULATION OF INNATE IMMUNE AND INFLAMMATORY SIGNALING IN MDS

Pro-inflammatory signaling and death receptor pathways

Probably the greatest difficulty for the understanding of the pathogenesis of MDS and also a source of controversy in this field is the coexistence of increased cell proliferation and cell death in BM.³⁵

Increased rates of intramedullary apoptosis are considered the main cause of the cytopenias that characterize MDS. Apoptosis is thought to be initiated by the death receptor Fas and its specific ligand (Fas-L), which is overexpressed and correlates with the rate of apoptosis in MDS^20,22,36,37 (Figure 1). Although BM CD34⁺ progenitors do not express Fas under physiologic conditions, they can do so after exposure to cytokines such as TNF-α or IFN-γ.^38,39 Accordingly, high levels of TNF-α are directly associated with apoptosis rates in MDS BM cells.¹⁷

Figure 1.

Signaling pathways frequently deregulated in MDS. The transmembrane receptors Fas (CD95), TNFR1, TNFR2, Toll-like receptors (TLRs) and IFN-γ receptor (IFNGR) and their associated signal transducers are frequently overexpressed and/or constitutively activated in MDS. Fas/CD95 is specifically engaged by Fas-L/CD95L, which induces caspase-dependent cell death by activating the initiator caspase-8 via its FAS-associated death domain (FADD). TNFR1 and TNFR2 activate the adapter protein TNFR-associated death domain (TRADD), which in turn activates TNFR-associated factors (TRAFs) to ultimately induce the phosphorylation of the mitogen-activated protein kinases (MAPKs) JUN N-terminal kinase (JNK) and/or p38 MAPK, the latter via the receptor-interacting protein (RIP). JNK induces the transcriptional activity of AP-1 and p38 MAPK, in turn, activates other transcription factors (TF) that carry out various functions. TNFR1/2 can also activate the transcription factor NF-κB via IκB kinase (IKK). In addition, TNFR1 can directly induce apoptosis through the death receptor pathway by activating FADD via TRADD, initiating caspase cleavage. TNFR2 lacks a death domain, so its functions are predominately pro-survival. After recognition of pathogen- or damage-associated molecular patterns (PAMPs and DAMPs, respectively), TLRs signal through several specific adapter molecules that ultimately lead to the activation of AP-1, p38 MAPK and NF-κB. Lastly, IFN-γ initiates a transcriptional response mediated by the activation of the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway. The transcriptional programs activated by these receptors generally lead to the expression of genes involved in the innate immune response, survival and differentiation but may also lead to the transcription of pro-apoptotic genes.

The role of TNF-α in the pathogenesis of MDS is not limited to the induction of the expression of Fas. TNF-α selectively binds two receptors, TNF receptor (TNFR) 1 and TNFR2. The primary role of TNFR1 is the induction of apoptosis through caspase-8 activation, whereas TNFR2 has anti-apoptotic functions induced by the c-Jun N-terminal kinase pathway.⁴⁰ A switch in TNFR expression associated with changes in apoptotic rates has been reported in MDS BM cells.^23,41 Whereas TNFR1 is abundantly expressed in RA, TNFR2 is more highly expressed in RAEB, indicating a correlation with the apoptotic activity in BM. Accordingly, RA patients also overexpress TRADD, FADD and RIP compared with controls and RAEB,²³ whereas levels of Fas are lower in advanced stages of the disease and negatively correlated with higher counts of BM blasts.³⁶ The blockade of TNF-α or Fas function by specific antibodies can partially restore growth of MDS hematopoietic progenitors,^22,37 which shows a reverse association between the functionality of TNF/Fas-dependent signaling and BM cellularity. Further evidence that pro-apoptotic signaling is strongly associated with lower-risk MDS is that a global expression profiling study conducted in 183 patients found `apoptosis' to be one of the most significantly upregulated functional groups of genes in CD34⁺ cells from RA versus controls and RAEB.⁴²

On the other hand, various cytokines, such as TGF-β, IFN-α and TNF-α itself, activate the p38 mitogen-activated protein kinase (MAPK) downstream signaling pathway in hematopoietic stem and progenitor cells.⁴³ p38 mitogen-activated protein kinase is known to increase apoptotic signaling in hematopoietic stem cells and to be hyperactivated in MDS BMs.⁴⁴

The deregulation of Toll-like receptor (TLR) signaling

The TLR family comprises some of the most important types of cell-associated mammalian pattern recognition receptors, which have a major role in innate immunity.⁴⁵ Remarkably, TLRs participate in the pathogenesis of several non-infectious inflammatory and autoimmune diseases that are clinically associated with the increased risk of MDS, such as chronic polyarthritis.⁴⁶

There are 10 different TLRs in humans (Table 2), all of which recognize different microbial antigens and self-components released in response to stress, tissue damage and cell death. Most TLRs are localized on cell surfaces and are especially abundant in macrophages, dendritic cells (DCs) and neutrophils, whereas others are associated with intracellular membranes from organelles such as the endoplasmic reticulum, endosomes and endolysosomes.⁴⁵

Table 2.

Human TLRs and their ligands

TLR	Localization	Canonical ligand	Abnormalities in MDS	References
TLR1	Plasma membrane	Triacyl lipoproteins	Overexpressed	Wei et al.51
TLR2	Plasma membrane	Lipoproteins (dimerizes with TLR1 or TLR6)	Overexpressed, higher in low-risk MDS, mutated	Kuninaka et al.,50 Wei et al.51
TLR3	Endolysosomes	dsRNA	Not reported
TLR4	Plasma membrane	LPS	Overexpressed, correlated with apoptosis	Maratheftis et al.,49 Kuninaka et al.50
TLR5	Plasma membrane	Flagellin	Not reported
TLR6	Plasma membrane	Diacyl lipoproteins	Overexpressed, higher in high-risk MDS	Wei et al.51
TLR7	Endolysosomes	ssRNA and small purine analogs	Overexpressed	Ganan-Gomez et al.120
TLR8	Endolysosomes	ssRNA	Not reported
TLR9	Endolysosomes	CpG-DNA	Overexpressed, correlated with TNF-α expression, decreased with progression	Kuninaka et al.50
TLR11	Plasma membrane	Unknown	Not reported

Abbreviations: LPS, lipopolysaccharide; MDS, myelodysplastic syndrome; TLR, Toll-like receptor.

Engagement of TLRs by their specific ligands leads to the activation of transcription factors that cooperatively regulate the expression of IFNs and pro-inflammatory cytokines and chemokines (Figure 2). Moreover, in monocytes/macrophages, TLRs upregulate the expression of hundreds of other genes that might be involved in antimicrobial defense, metabolic changes, tissue repair and differentiation.^45,47 Interestingly, TLR signaling also induces the expression of several microRNAs such as miR-146a/b, miR-147, miR-155, miR-181 and miR-21, which participate in the fine-tuning of the inflammatory response and some of which are likely involved in the pathogenesis of MDS (reviewed elsewhere).⁴⁸

Figure 2.

TLR signaling and its activation in MDS. Receptors and mediators colored in different shades of red represent molecules found to be overexpressed or constitutively activated in MDS. TLRs transduce their signals through two different adapter molecules, myeloid differentiation primary response gene 88 (MyD88) and TIR domain-containing adapter inducing IFN-β (TRIF). Virtually all TLRs signal via MyD88, except for TLR3 (not depicted here), which is an intracellular receptor signaling via TRIF. In addition, TLR4 is the only TLR that can use both mediators. MyD88-driven signaling mediates a rapid and acute pro-inflammatory response through the activation of NF-κB, AP-1 and p38 MAPK-dependent transcription factors. The intracellular receptors TLR7/8 and TLR9 additionally activate interferon-regulatory factor (IRF)-7, which induces the expression of type I IFN. In contrast, TRIF triggers a delayed pro-inflammatory response mediated by NF-κB and IRF-3-dependent type I IFN expression. Herein, TLR2 and TLR4 are depicted as examples of cytoplasmic membrane-bound TLRs, and TLR9 is shown as an example of intracellular TLRs. Abbreviations not defined in the text: Toll-interleukin 1 receptor domain-containing adapter protein (TIRAP), TRIF-related adapter molecule (TRAM), TGF-β activated kinase (TAB), TANK-binding kinase 1 (TBK1), NF-κB-inducing kinase (NIK), ubiquitin (Ub).

Gene expression profiling assays have revealed that a number of TLRs, as well as many of the signal transducers in this pathway, are overexpressed in a high proportion (40–80%) of MDS patients. Maratheftis et al.⁴⁹ were the first to postulate that TLR4 is overexpressed in BM mononuclear cells (BMMCs) and CD34⁺ cells of MDS patients, and found its levels to be significantly correlated with apoptotic rates. Another study by Kuninaka et al. reported increased expression of TLR2 and TLR9 in all MDS subtypes, with levels of TLR9 also correlated with those of TNF-α, and the expression of both decreased with disease progression to AML.⁵⁰ Notably, TLR4 levels, which did not significantly differ from those of controls in that study, were also correlated with TNF-α expression. Our group further confirmed those results by gene expression profiling in a cohort of MDS CD34⁺ cells.⁵¹ Importantly, we found that not only was TLR2 highly expressed in a great majority of MDS patients, but also that TLR1 and TLR6, its heterodimerization partners, were significantly overexpressed when compared with healthy controls. Moreover, we showed that functional TLR2/TLR1 and TLR2/TLR6 dimers were associated with the inflammatory milieu observed in MDS. In addition, we identified an MDS-related somatic mutation of TLR2, TLR2-F217F, which was present in 11% of patients and associated with enhanced NF-κB activation.

In line with these findings, multiple TLR downstream signaling mediators have been shown to be also overexpressed in MDS. Velegraki et al.⁵² demonstrated increased expression of a wide panel of genes involved in TLR4 signaling in MDS BMMCs. A gene expression microarray showed that TRAF6 is overexpressed in MDS CD34⁺ cells when compared with healthy controls.⁵³ Furthermore, DNA arrays revealed the amplification of the TRAF6 locus (chromosome 11p12) and the TIRAP locus (chromosome 11q24.2) in MDS.^54,55 Our group recently reported that MyD88 is overexpressed in BM progenitors of MDS and is associated with risk stratification and patient survival.⁵⁶ Lastly, Rhyasen et al.⁵⁷ demonstrated IRAK1 upregulation in MDS BMMCs.

Several functional studies in MDS patient cells have confirmed increased TLR signaling, leading to elevated cytokine secretion. Maratheftis et al.⁴⁹ showed that increased TLR4 signaling contributes to the elevated BM levels of TNF-α in MDS BMMCs and that TNF-α, in turn, induces TLR4 expression in a positive feedback loop. Velegraki et al.⁵² later demonstrated that MDS BM plasma can induce TLR4-dependent cytokine secretion in BMMCs from both healthy and MDS subjects. Authors ascribed this effect to high mobility group box 1, which they found to be upregulated in MDS BM plasma, and further showed that this protein is released by apoptotic BMMCs. In turn, we showed that specific ligands of the two TLR2 heterodimers induce IRAK1 phosphorylation, NF-κB activation and IL-8 secretion in BM precursors.⁵¹ Notably, TLR2 stimulation also induced the expression of the histone demethylase JMJD3,⁵¹ which we reported to be significantly overexpressed in MDS and to form a positive feedback loop with NF-κB activation, leading to the expression of IL-8.⁵⁸ Taken together, these results suggest that TLR4 and TLR2 signaling is not only activated in MDS but also has the ability to self-maintain. Additional evidence was recently provided by Rhyasen et al, who showed that IRAK1 is constitutively active in MDS BM cells and that its inhibition significantly downregulates genes involved in the inflammatory response, including TLR6 and IL-8^57,59 Another hint of the existence of feedback loops is the elevated expression of IL-1RAP (IL-1R accessory protein), an IL-1-dependent alternative activator of the MyD88/IRAK1/TRAF6 signaling axis, in CD34⁺ cells of high-risk MDS.⁶⁰

Although the precise role of TLR-mediated signaling in MDS has not yet been elucidated, in vitro and in vivo assays suggest that the deregulation of this pathway might be involved in the loss of progenitor cell function and impaired differentiation in BM cells. Starczynowski et al.⁶¹ have reported the loss of miR-145 and miR-146a in 5q deletion (5q-) syndrome. These microRNAs, respectively, target the downstream TLR transducers TIRAP and TRAF6. Both the functional knockdown of miR-145 and miR-146a and the enforced overexpression of TRAF6 in mouse BM lead to multiple hematopoietic abnormalities that recapitulated features of 5q- syndrome and were associated with NF-κB activation and increased the production of IL-6. Moreover, mice transplanted with TRAF6-expressing cells presented relevant hematological phenotypes as well as progress to AML. More recently, our group showed that the blockade of TLR2-mediated signaling in primary CD34⁺ cells with a specific inhibitor of MyD88 increased the number of erythroid colonies and the expression of erythroid marker genes.⁵⁶ We obtained similar effects by inhibiting the IL-8 receptor⁵⁶ and knocking down TLR2⁵¹ Blockade of IRAK1 activity in MDS by Rhyasen et al, in turn, decreased overall cell growth and colony formation of MDS BM cells and suppressed MDS xenografts in immunodeficient mice.^57,59 On the contrary, Velegraki et al.⁵² reported a TLR4-dependent decrease in the clonogenic potential of both normal and MDS CD34⁺ cells in the presence of apoptotic BMMCs or recombinant high mobility group box 1.

Taken together, these results indicate that TLR-dependent signaling deregulates hematopoiesis and hematopoietic stem cell (HSC) growth in MDS, although its specific effects are still not clear. This notion is consistent with the thought that the main function of TLR-mediated signaling in BM hematopoietic precursors is the replenishment of the cellular components of the innate immune system.⁶² Upon stimulation, both in vivo and in vitro, these receptors initiate a transcriptional response that mediates MyD88-dependent and growth factor-independent differentiation of hematopoietic progenitors into monocytes/macrophages and DCs, at the expense of lymphopoiesis.^47,63–65 This process requires quiescent HSCs to reenter the cell cycle and is accompanied by the secretion of inflammatory cytokines/chemokines frequently overexpressed in MDS.^47,63 Interestingly, TLR1/2 stimulation seems to be more effective in inducing monocytic differentiation, whereas TLR7/8 activation is more effective at inducing the DC subset.⁶³

Therefore, constitutive TLR signaling could cause abnormalities in myeloid/lymphoid differentiation and eventually affect the outcomes of hematopoiesis. In line with this hypothesis, a study has shown that chronic TLR stimulation induces durable changes in mouse BM physiology that are very similar to the MDS phenotype, including increased cycling rates and limited self-renewal of HSCs and loss of lymphopoietic potential. Remarkably, the authors found a correlation between these changes and aging.⁶⁶

Recent evidence has also shown that the role of TLR signaling in MDS is not limited to its effects on HSCs and early progenitors. The gene DIAPH1, which encodes the protein mDia1, is located in the 5q region and is downregulated in patients with 5q-syndrome.⁶⁷ Keerthivasan et al.⁶⁸ recently reported that young mDia1 heterozygous or knockout mice have granulocytopenias originating from defects in differentiated granulocytes and that, upon aging, they acquire prominent myeloid dysplasia with neutropenia, which is characteristic of MDS. mDia1 deficiency induced the upregulation of the TLR4 adapter protein CD14, which was dramatically overexpressed in committed granulocytic progenitors and, especially, in mature granulocytes. Chronic TLR4 stimulation in CD14-overexpressing mice mimicked the hematologic phenotype of MDS, which suggests that the deletion of DIAPH1 in 5q-syndrome could contribute to the pathogenesis of MDS by inducing the overexpression of CD14. In agreement, CD14 overexpression was confirmed in granulocytes of 5q- patients.

The role of transcription factor NF-κB

The nuclear factor (NF)-κB family of transcription factors is composed of several proteins from the NF-κB (p50/p105, p52/p100) and Rel subfamilies (RelA/p65, RelB, c-Rel), which heterodimerize and bind to various target genes.⁶⁹ NF-κB transcription factors are activated in response to a variety of stimuli, such as inflammatory cytokines, pathogenic antigens, oxidative stress, DNA damage and the activation of pattern recognition receptors. NF-κB activation triggers the expression many target genes involved in the adaptive response to different types of stress and in regulating the expression of a number of inflammatory cytokines and chemokines including TNF-α, IL-6 or IL-8, inducible enzymes, adhesion molecules and proteins regulating immune responses. Importantly, these factors also regulate the expression of several anti-apoptotic proteins and proliferative factors.⁶⁹

Accordingly, NF-κB signaling has an important role in the survival of MDS progenitors. NF-κB activity is significantly elevated in MDS BM progenitors and cell lines and has been correlated with the progression of the disease, with later stages of MDS presenting with the highest activity levels.⁴¹ The blockade of NF-κB activity has been shown to induce apoptosis in normal and MDS BM precursors,^70–72 suggesting that constitutive NF-κB signaling provides malignant cells, which overpopulate BM in the late stages of MDS, with a survival advantage.

More importantly, there is substantial evidence that the activity of both the canonical (p50/RelA-mediated) and non-canonical (p52/Rel-B-mediated) NF-κB signaling pathways are essential for the modulation of HSC functions, including proliferation, self-renewal and differentiation.^73,74 These roles suggest that the deregulation of the transcriptional activity of NF-κB may lead to or enhance the differentiation and proliferative abnormalities characteristic of MDS. Although it has been postulated that constitutive canonical NF-κB activation is not sufficient to induce changes in CD34⁺ cell growth and differentiation,⁷⁵ there are data showing that mice deficient in the NF-κB inhibitor IκBα develop a dysregulation of hematopoiesis characterized by an increased number of cycling myeloid progenitors in BM, high counts of granulocytes in the BM and liver, myeloid dysplasia and increased number of leukocytes in the peripheral blood with neutrophil dysplasia.⁷⁶ Discrepancies among the few existing studies may be caused by the fact that the effects of NF-κB on hematopoiesis are not cell autonomous. For instance, its activation in myelopoietic cells alone is not sufficient for the induction of the MDS phenotype, but the deregulation of NF-κB in the nonhematopoietic compartment causes a myeloproliferative disorder.⁷⁶ This paradoxical effect of NF-κB activation should be analyzed in depth in the future.

Studies of some NF-κB target genes also provide indirect evidence of the potential relevance of this factor in MDS. One example is the pro-inflammatory cytokine IL-6, which stimulates B- and T-cell differentiation and is also a macrophage and granulocyte inducer. IL-6-transgenic mice develop a transplantable myeloproliferative disorder characterized by thrombocytosis, anemia and transient neutropenia with progression to leukocytosis.⁷⁷

Overall, NF-κB activation has an important role in the pathogenesis of MDS by inducing the expression of inflammatory cytokines and pro-survival factors and probably also by contributing to dysregulated hematopoiesis and is a molecule of great interest in the study of MDS.

However, although many TLR-induced cytokines are transcriptionally activated by NF-κB, the effects of TLRs on differentiation do not necessarily have to be ascribed to this factor. A global expression profiling and hierarchical clustering analysis carried out in the CD34⁺ cells of 183 patients⁴² showed that the most significantly deregulated pathway for upregulated genes in MDS was the IFN signaling pathway (Figure 1). IFN-γ, which is secreted in response to the activation of several TLRs and also secreted in high levels in MDS patients, appears to have a strong inhibitory effect on hematopoietic progenitors and stem cells that includes impairing erythropoiesis⁷⁸ and reducing the long-term repopulation potential of HSCs.⁷⁹ Thus, the effects of TLR signaling on hematopoiesis could be mediated by different effectors other than NF-κB.

Furthermore, the stimulation of TLR4 and TLR2 in vitro induces apoptosis in primary BM cells.^49,80,81 In the case of TLR2, this apoptosis is NF-κB-independent,^80,81 which is consistent with the pro-survival role of NF-κB. Notably, a p38 MAPK inhibitor significantly decreased TLR2-dependent cell death, suggesting that p38 is involved in the induction of NF-κB-independent apoptosis by TLRs.⁸⁰

INFLAMMATION AND INNATE IMMUNITY IN OTHER CELL TYPES

Involvement of the BM microenvironment in MDS

Mesenchymal stem cells (MSCs) are primitive, non-hematopoietic stem cells that give rise to all of the various types of stromal cells that form the BM niche.⁸² MSCs carry out immunosuppressive functions through the impairment of DC maturation.⁸³ Because BM HSCs and MSCs may have a common multipotent progenitor, malignant HSCs could coexist with a malignant MSC clone with altered immunosuppressive properties in MDS. Many efforts have been made to detect abnormal MDS MSC clones, but various groups have failed to show significant morphological and functional differences between the patients and healthy individuals.^84,85 Although BM MSCs from many patients have chromosomal alterations, these seem to have no correlation with the karyotypic/cytogenetic abnormalities of their HSC counterparts,^84,85 which indicates that MDS-derived MSCs and their progeny have a different clonal origin. Nevertheless, it was recently demonstrated that the immunosuppressive capacity of MSCs is decreased in MDS and that these cells fail to efficiently inhibit DC maturation. Remarkably, this effect was only observed in cells from low-risk MDS, whereas immunosuppressive functions of high-risk MDS-derived MSCs were similar to those of controls.⁸⁶ These results indicate that, despite not belonging to the malignant clone, the functionality of MSCs is altered in MDS and may favor the expansion of cytotoxic T cells in the early stages of the disease.

The notion of the active participation of MSCs in the pathogenesis of MDS has been reinforced by a recent publication showing that MDS HSCs can `reprogram' MSCs by inducing changes in their gene expression profiles. `Reprogrammed' MDS-derived MSCs showed increased ability to allow the in vivo engraftment of MDS CD34⁺ cells, which exhibited long-term renewal and myeloid skewing of differentiation.⁸⁷ Interestingly, `response to inflammation' and `cytokine-cytokine receptor interaction' were two of the functional groups of genes upregulated in MSCs, which suggests that MDS HSCs induce adaptation of their neighboring cells to the inflammatory microenvironment.

Another line of evidence of the abnormal behavior of the BM niche in MDS is the fact that Fas-L is more prominently expressed in stromal cells and macrophages than in hematopoietic cells, which in turn widely express Fas and TNFR.^23,88 This distribution suggests that non-hematopoietic cells in the BM niche could be responsible for the induction of apoptosis in hematopoietic precursors. In agreement with that hypothesis, Stirewalt et al.⁸⁹ demonstrated that the apoptotic effects of TNF-α on hematopoietic cells depend on their direct contact with stromal cells, in which TNF-α induces significant changes in gene expression, particularly in apoptosis-related genes and cytokines/chemokines such as IL-6 and IL-8.

Taken together, these results suggest that MDS-derived MSCs and BM stromal cells are determinants of the fate of hematopoietic progenitors and have an important role in the pathogenesis of MDS.

Involvement of immune cells in MDS

Myeloid-derived suppressor cells (MDSCs) are inflammatory and immunosuppressive effectors localized to the BM that express the immune-receptor CD33.⁹⁰ Chen et al.⁹¹ found that MDS patients have increased numbers of MDSCs and that they induce defects in myeloid and erythroid differentiation. Furthermore, MDSCs appear to reduce T-cell proliferation and functionality in MDS patients. These effects are mediated by CD33, for which the inflammatory signaling molecule S100A9 is a specific ligand. S100A molecules are the ligands of other innate immune receptors, such as TLR4,⁹² and are known to be overexpressed in MDS BM CD34⁺ cells.⁵⁸ Moreover, S100A9 appears to be upregulated in the hematopoietic cell compartment of telomere-dysfunctional mice, an animal model of premature aging with perturbed BM hematopoiesis.⁹³ Chen et al. found that the levels of S100A9 are also elevated in MDS BMMCs, supporting the increased counts of MDSCs and the secretion of immunosuppressive cytokines. Furthermore, S100A9-transgenic mice developed an MDS-like phenotype with multilineage cytopenias and cytological dysplasia.⁹¹ Forced maturation of MDSCs restored hematopoiesis, suggesting that these cells are deeply involved in the pathogenesis of MDS. Of interest, and potentially linked to the switch to immunosuppression during MDS progression, the development of MDSCs relies on inflammatory cytokines, and granulocyte macrophage colony-stimulating factor and IL-6 generate the most suppressive MDSCs.⁹⁴

Macrophages are also potentially involved in MDS. It was recently shown that there is a recurrent and specific loss of granulocyte/monocyte progenitor populations in the BM of low-risk MDS, which is likely due to the increased phagocytosis of granulocyte/monocyte progenitors by macrophages. This deregulated phagocytosis is thought to be controlled by the interaction between cell surface calreticulin on target cells and the low-density LRP1 (lipoprotein receptor-related protein) receptor on macrophages.⁹⁵ Macrophages also mediate angiogenesis, which is elevated in high-risk MDS.^26,27

Although MDS patients generally present with lymphopenias, cellular immunity may be upregulated in low-risk MDS. These patients have higher counts of cytotoxic (CD8⁺) and helper (Th17) T cells and NK cells and lower counts of T-regulatory lymphocytes (Treg).^33,96–98 The expansion of CD8⁺ cells is particularly detrimental in patients with chromosome 8 trisomy because CD8⁺ cells specifically target WT1, which is overexpressed in the CD34⁺ progenitors of these patients. Probably for this reason, patients with trisomy 8 are more responsive to IST.⁹⁹ Overall, these changes in cell number and functionality cooperate with the release of inflammatory cytokines and trigger an autoimmune response against hematopoietic cells that may contribute to intramedullary apoptosis.^98,100–102 In agreement with this hypothesis, the depletion of the CD8⁺ cells allows colony formation in primary BMMCs from MDS patients; however, it remains unclear if the proliferating cells belong to the normal or the malignant clone.¹⁰²

On the other hand, in high-risk cases, impaired cellular responses with lower levels of CD8⁺, Th17 and NK cell function and increased numbers of Tregs are more common.^{97,98,103,104} The number of these cells is also associated with higher levels of IL-10⁹⁸ and a poorer prognosis.¹⁰⁵ Similarly, the decreased cytolytic function of NK cells, which might be related to the fact that a subset of these cells could be derived from the MDS clone,¹⁰⁶ correlates with MDS progression.¹⁰⁴ Overall, this dysregulation leads to the acquisition of immune tolerance by the proliferating clone and enhances the risk of progression.

The events triggering the clonal expansion of CD8⁺ cells in low-risk MDS, as well as the switch in the CD8⁺:Treg ratio during progression, are poorly understood. The expansion of CD8⁺ cells could be induced to fight the malignant clone or contribute to the annihilation of normal hematopoietic progenitors. Despite numerous efforts to identify putative antigenic sequences in MDS T-cell receptors, the causal antigens eliciting the CD8⁺ cell response, other than WT1, have not yet been characterized.^96,107,108 Similarly, little is known about the decline in the proportion of CD8⁺ cells in high-risk MDS. However, data from our group suggest that this phenomenon could be related to the expression of the negative co-stimulatory T-cell receptor programmed death-1 (PD-1) and its ligand, PDL-1. These molecules are upregulated in a subset of MDS patients, with PD-1 being highly expressed in the peripheral blood mononuclear cells and PDL-1 being preferentially overexpressed in BM CD34⁺ cells.¹⁰⁹ These results suggest that MDS BM cells may actively participate in the repression of the CD8⁺ T-cell response. Indeed, we found that higher levels of PD-1/PDL-1 in BM cells are associated with resistance to therapy and with a poorer prognosis. More research in this field is necessary to shed light on the role of T-cell immunity in the pathogenesis of this disease. Other frequent alterations that are common to all stages of MDS are the deficiency of B cells¹¹⁰ and γδ T cells,¹¹¹ important regulators of T-cell responses.

PROGRESS OF ANTI-INNATE IMMUNE THERAPIES

There is wide clinical experience on the treatment of MDS with IST, which was used before molecular evidence of innate immune involvement in MDS arose. Some of the first approaches used cyclosporine, but the risk of renal failure made other treatments, such as antithymocyte globulin or the immunomodulatory drug lenalidomide, the ISTs of choice. About 30% of patients treated with IST become transfusion-independent and improve cell counts, although these do not revert to normal.¹¹²

Encouragingly, recent findings regarding innate immune and inflammatory signals in MDS have provided a strong biological rationale for the development of novel therapeutic strategies. Preclinical studies have demonstrated that this interfering strategy may lead to promising therapeutic effects. As explained above, the specific inhibition of the activity or expression of TLR2 and its downstream effectors in primary MDS BM cells significantly improved differentiation, induced apoptosis and impaired their clonal generation potential, particularly in cells from patients with lower-risk disease. Interestingly, the effects of the IRAK1 inhibitor are further improved when it is combined with a Bcl-2 specific inhibitor.^51,56–58 Furthermore, interference with TRAF6 sensitized MDS/AML cells to bortezomib-induced cytotoxicity.¹¹³ Similarly, inhibition of the p38 MAPK pathway by the specific inhibitor SCIO-469 stimulated hematopoietic activity in vitro while simultaneously decreasing the expression of TNF-α or IL-1β-induced proinflammatory chemokines in BM stromal cells.^44,114 Finally, SB-332235, a specific inhibitor of the IL-8 receptor, significantly reduced growth and colony formation in primary MDS BM CD34⁺ cells.¹¹⁵

Consistent with these preclinical findings, clinical trials of novel innate immune-targeted interventions are starting to emerge. ARRY614 is a potent dual inhibitor of p38 MAPK/Tie2, key downstream effectors of innate immune signaling. In an open-label phase I study in patients with low-risk MDS, nearly 30% of patients achieved hematological improvement with this drug, almost all of which had previously failed treatment with azanucleosides. On the basis of these results, a new formulation of ARRY614 has entered a phase I/II study in patients with low-risk MDS.¹¹⁶ Another drug that targets immune signaling and has entered clinical studies is OPN-305, a TLR2-directed antibody with promising utility in MDS,¹¹⁷ which is about to start phase II studies in MDS patients. An oral small molecule inhibitor of TGF-β receptor I kinase, LY-2157299, is also being tested in a phase II trial in low- and intermediate-risk MDS.¹¹⁸ Finally, a pharmacological inhibitor of IDO1 with effects on T and NK cell function and tumor growth is currently being studied in MDS.¹¹⁹ Ongoing preclinical and clinical trials of targeted innate immune interventions are summarized in Table 3.

Table 3.

Summary of recent anti-innate immune preclinical and clinical trials in MDS

Targeted molecule	Agent	Status	Potential therapeutic effect	References
TLR2/JMJD3	shRNA	Preclinical	Improvement of erythroid differentiation of MDS BM CD34+ cells	De Luca et al.,64 Perkins et al.69
MyD88	Inhibitory peptide	Preclinical	Improvement of erythroid differentiation of MDS BM CD34+ cells	Dimicoli et al.56
IL-8	Neutralizing antibody	Preclinical	Improvement of erythroid differentiation of MDS BM CD34+ cells	Dimicoli et al.56
IL-8 receptor	SB-332235	Preclinical	Reduction of growth and colony formation in MDS BM CD34+ cells	Giricz et al.115
IRAK1	RNAi and specific inhibitor molecule	Preclinical	Induction of apoptosis and impairment of clonal generation in MDS BM cells	Dimicoli et al.,56 Rhyasen et al.57
p38 MAPK	SCIO-469	Preclinical	Enhancement of hematopoiesis and reduction of apoptosis in MDS BM CD34+ cells, anti-inflammatory effects in BM stromal cells	Navas et al.,44 Navas et al.114
p38 MAPK	ARRY614	Phase I	Hematological improvement in patients who previously failed azanucleoside treatment	Sekeres et al.116
TLR2	OPN-305	Entering phase II	Improvement of erythroid differentiation of MDS BM CD34+ cells	Reilly et al.117
TGF-β receptor I kinase	LY-2157299	Recruiting for phases II/III	Improvement of MDS BM progenitor colony formation in vitro and in vivo, stimulation of hematopoiesis	Zhou et al.118
IDO1	INCB024360	Phase II	Enhancement of Tand NK cell-mediated immune response against the malignant clone; expansion of granulocytic/erythroid progenitors	Berthon et al.,34 Liu et al.119

Abbreviations: BM, bone marrow; IL-8, interleukin-8; MDS, myelodysplastic syndrome; TGF-β, transforming growth factor beta; TLR, Toll-like receptor.

CONCLUDING REMARKS

With the recent development of new technologies and the consequent experimental optimization, understanding of the role of innate immune deregulation in the MDS pathogenesis has been greatly improved. It is now commonly recognized that constitutively activated innate immune and inflammatory pathways can directly affect hematopoiesis, lead to altered cytokine secretion and impact T-cell immunity. All these biological effects contribute to the development and progression of MDS (Figure 3). Furthermore, innate immune deregulation seems to be chronic rather than transient and affects all stages of the disease. This deregulation could arise from cellular stresses associated with senescent changes, genomic instability and other genetic and epigenetic abnormalities that occur in hematopoietic cells with aging, but it could also be initiated by abnormal cellular interactions in the BM microenvironment. To better evaluate the biological and clinical implications of innate immune signaling in MDS, deep investigations of innate immune alterations are necessary, especially in purified specific BM hematopoietic populations. Special attention should be drawn to the key aspects that remain unknown, such as the signaling pathways activated by innate immunity that determine disease evolution and/or define distinct MDS subtypes. Significant efforts are also needed to identify the endogenous ligands responsible for TLR activation and the conditions that contribute to their release or make MDS patients more vulnerable to the deleterious effects of TLR signaling. This information could eventually be applied to develop effective therapeutic regimens. Because ~ 50% of deaths in MDS patients are related to cytopenias rather than to progression to AML, the development of immunomodulatory therapies potentially improving hematopoiesis is of great interest for the management of MDS.

Figure 3.

Proposed model for the central role of inflammation/innate immunity in the pathogenesis of MDS. (1) The malignant clone or MDS HSC originates in the BM of patients with the characteristic phenotype of aging. The main characteristics of the BM in old individuals are summarized in the box. In this context, the MDS HSC might originate from the genetic/epigenetic changes occurring in susceptible individuals during aging; be generated by exposure to various types of stress, including DNA damage; or could develop after a sustained exposure to inflammatory molecules derived from an existing or past inflammatory condition. (2) Either the changes in gene expression or the pre-exposure to inflammatory molecules trigger the activation of innate immune signaling pathways and the subsequent secretion of cytokines, chemokines and growth factors, which create an inflammatory microenvironment. (3) As a consequence, BM HSCs increase their cycling rates. Also driven by the release of cytokines, HSCs express Fas and other immune receptors on their surface, and CD8⁺, Th17 and NK cells are recruited. (4) The expression of death receptors and the continuous inflammatory signaling induce apoptosis in some HSCs in addition to the T-cell-mediated cytotoxicity. However, it is not clear if the dying HSCs belong to the normal or MDS clone, or to both. (5) Regardless of HSC origin, intramedullary apoptosis decreases the number of functional BM progenitors, which results in a reduced number of fully differentiated cells. In addition, intrinsic defects on the differentiation potential of the MDS clone, and the sustained inflammatory signaling, cause differentiation to be dysregulated and skewed toward the myeloid lineage. (6) The released cytokines and chemokines, and probably also certain cell-to-cell contact proteins, eventually trigger the recruitment of MDSCs to the tumor site and induce profound gene expression changes in the surrounding MSCs. MDSCs exacerbate the defects of differentiation by inducing myeloid skewing and killing erythroid precursors, and they suppress the autoimmune response by CD8⁺ T cells as well as probably participating in the switch to an immunotolerant microenvironment. Likewise, `reprogrammed' MSCs express genes involved in the adaptation to inflammation. (7) The high proliferation rates make MDS HSCs more prone to the accumulation of additional genetic/epigenetic aberrations. In addition, unknown mechanisms lead to a switch in the expression of TNFRs and probably also in the expression of other molecules, which makes malignant cells resistant to apoptosis. (8) Altogether, these alterations confer the MDS clone a survival advantage and contribute to the aberrant proliferation of the clone, which at this point overpopulates the BM. (9) This switch in the cellular processes that prevail in the BM is accompanied by the recruitment of immunomodulatory cells, which are probably triggered by changes in the cytokine/chemokine milieu. Treg cells confer immune resistance to the MDS clone and allow abnormally proliferating cells to escape the immune system. Along with step 8, this event increases the risk of progression to AML.