Page 120 - Haematologica April 2020

N. van Leeuwen-Kerkhoff et al.
ways, e.g. Toll-like receptor (TLR) signaling, can induce malignant transformation and disease progression by causing genotoxic cell stress. Indeed, in low-risk MDS ele- vated levels of several stress-inducing molecules, such as the damage-associated molecular pattern molecules S100A8/A9, are actively secreted from mesenchymal niche cells in the bone marrow (BM) microenvironment, thereby causing niche-induced DNA damage in hematopoietic stem and progenitor cells.13 High S100A9 levels in MDS BM also result in inflammasome assembly and subsequent initiation of pyroptosis, an immunogenic form of cell death, which could potentially explain the high rate of cell death in low-risk MDS BM.14–16 These sol- uble inflammatory molecules are able to bind to TLR on the surface of hematopoietic stem and progenitor cells and immune cells. Constitutively activated TLR-signaling and downstream mitogen-activated protein kinase (MAPK) and nuclear factor kappa B (NF-κB) activation are evident and have been implicated in the pathogenesis of MDS.17–24 Besides active secretion of stress-inducing molecules, pas- sive release from cells undergoing immunogenic cell death has also been described in MDS. Levels of high mobility group box 1 (HMGB1), a mediator strongly involved in inflammatory processes and a ligand for TLR4, were found to be increased in the BM of MDS patients due to impaired clearance of apoptotic cells causing secondary necrosis and leakage of this molecule into the BM environ- ment.25
As a result of this vicious circle of inflammation and cell death, immune-inhibitory mechanisms that interfere with this excessive inflammatory process kick in. While these immune-inhibitory pathways may control the inflamma- tory response to some extent, they also facilitate the expansion of immunosuppressive cells, such as Treg and myeloid-derived suppressor cells, which further suppress the already weakened immune surveillance against the malignant clone. A delicate balance between immune acti- vation and inhibition is, therefore, required to maintain effective immunosurveillance. Thrombomodulin (TM) is known for its anticoagulant function by serving as a co- factor for thrombin. Notably, the lectin-like domain of the TM molecule has marked anti-inflammatory activities and interferes with the complement pathway.26–28 Several stud- ies have shown strong correlations between disease sever- ity and TM levels in, for instance, autoimmune and infec- tious diseases as well as in cancer.29–31 In the immune sys- tem, TM, also known as CD141 or BDCA-3, is mainly expressed on dendritic cells.32–34 We have previously described elevated expression of TM/BDCA-3 on tumor- conditioned and immunosuppressive monocyte-derived dendritic cells that acquire a M2-like macrophage pheno- type.35,36 The anti-inflammatory potential of TM has also been assigned to the fact that TM is able to bind HMGB1, thereby inhibiting the strong pro-inflammatory effect of this molecule. Since high levels of this molecule were found in low-risk MDS BM, this interactive mechanism may be relevant in keeping excessive immune activation to a minimum. The aim of this study was to evaluate the possible role and prognostic value of TM in regulating the inflammatory immune response in MDS. The expression of TM was evaluated on different monocyte subsets (clas- sical, intermediate and non-classical) in the peripheral blood (PB) and BM within different MDS risk groups. Multidimensional mass cytometry was used to investigate the putative impact of TM+ monocytes on the T-cell phe-
notype. The cell surface expression of TM was higher on classical monocytes in both the PB and BM of MDS patients than on healthy donor-derived monocytes. The expression of TM was related to a more favorable progno- sis and functional skewing of the T-cell response toward a more tolerized state.
Methods
Patient and control samples
Twenty-nine PB and 154 BM samples from newly diagnosed MDS patients were collected in this study. Patients were assigned to different risk categories using the Revised International Prognostic Scoring System (IPSS-R) and the 2016 World Health Organization (WHO) classification (details are given in the Online Supplementary Methods file and Table 1). A set of 25 age-matched control BM samples was obtained after written informed consent from hematologically healthy patients who were undergoing car- diac surgery at Amsterdam University Medical Center (the Netherlands). For the PB analysis, 31 control samples were collect- ed. The study was approved by the local ethical committee and was conducted in accordance with the declaration of Helsinki.
Flow cytometry and fluorescence in situ hybridization PB and BM cells were analyzed on a flow cytometer (FACSCantoTM, BD Biosciences, San Jose, CA, USA) after incuba- tion with a panel of monoclonal antibodies (see Online Supplementary Methods for details). Data were analyzed using FlowJo software (Tree Star, Ashland, OR, USA). Monocyte subsets were identified based on the differential expression of CD14, CD16 and M-DC8 [anti-6-Sulfo LacNAc (Slan)], using recent rec- ommendations (Online Supplementary Figure S1).37,38 Classical monocytes were characterized by high CD14 expression, and CD16 and M-DC8 negativity. Intermediate and non-classical monocytes were both defined as CD16+. However, only interme- diate monocytes expressed CD14. We used M-DC8 as a marker to discriminate between intermediate and non-classical monocytes
as suggested by Hofer et al. (Figure 1A).38
Three samples containing monocytes with high TM expression
and a known cytogenetic aberrancy were used for the isolation of classical monocytes and subsequent interphase fluorescence in situ hybridization (FISH) analysis (details are given in the Online Supplementary Methods).
T-cell cultures and multidimensional mass cytometry
A multiparameter deep-phenotyping strategy, known as cytom- etry by time-of-flight (CyTOF), was used for T cells cultured in the presence of MDS-derived TM- or TM+ monocytes (culture details are provided in the Online Supplementary Methods). Data were ana- lyzed using a combination of automated dimension reduction and clustering methods including t-distributed stochastic neighbor embedding (t-SNE)39 to visually (viSNE) identify cell populations.40 This was followed by spanning-tree progression analysis of densi- ty-normalized events (SPADE)41 for the clustering of T cells as pub- lished before.42,43 The deep immunophenotyping of T-cell clusters was performed using our in-house pipeline (publicly available here: https://github.com/kordastilab/cytoClustR) followed by marker enrichment modeling (MEM) to calculate MEM scores of the iden- tified subpopulations.44
Statistical analysis
Significant differences for two-group comparisons were ana- lyzed by applying a non-parametric Mann-Whitney U test, whereas for multi-group comparisons a Kruskal-Wallis test with
962
haematologica | 2020; 105(4)