Thrombomodulin-expressing monocytes are associated with low-risk features in myelodysplastic syndromes and dampen excessive immune activation

Abstract

The bone marrow of patients with low-risk myelodysplastic syndromes (MDS) is often an inflammatory environment and associated with an active cellular immune response. An active immune response generally contributes to antitumor responses and may prevent disease progression. However, chronic immune stimulation can also induce cell stress, DNA damage and contribute to the pathogenesis of MDS. The protective mechanisms against excessive immune activation are therefore an important aspect of the pathophysiology of MDS and characterizing them may help us to better understand the fine balance between protective and destabilizing inflammation in lower-risk disease. In this study we investigated the role of thrombomodulin (CD141/BDCA-3) expression, a molecule with anti-inflammatory properties, on monocytes in the bone marrow and peripheral blood of MDS patients in different risk groups. Patient-derived classical monocytes showed high expression levels of thrombomodulin, whereas monocytes from healthy donors hardly expressed any thrombomodulin. The presence of thrombomodulin on monocytes from MDS patients correlated with lower-risk disease groups and better overall and leukemia-free survival. Using multidimensional mass cytometry, in an in-vitro setting, we showed that thrombomodulin-positive monocytes could polarize naïve T cells toward cell clusters which are closer to T helper type 2 and T regulatory cell phenotypes and less likely to contribute to effective immune surveillance. In conclusion, the expression of thrombomodulin on classical monocytes is a favorable and early prognostic marker in patients with low-risk MDS and may represent a new mechanism in the protection against disproportionate immune activation.

Figure 1.

Thrombomodulin-expressing monocyte subsets in normal and myelodysplastic syndrome bone marrow and peripheral blood samples. (A) Identification of different monocyte subsets in normal bone marrow (NBM) and myelodysplastic syndrome (MDS) bone marrow (BM). Three markers, CD14, CD16 and M-DC8, were used to identify classical (CD14⁺⁺CD16⁻M-DC8⁻, in orange), intermediate (CD14⁺CD16⁺M-DC8⁻, in purple) and non-classical monocytes (CD14^+/−CD16⁺M-DC8⁺, in green). The levels of expression of thrombomodulin (TM) and HLA-DR were assessed on all separate monocyte subsets. The median fluorescence intensity (MFI) value for each subset is shown for a representative sample. (B) Frequencies of monocyte subsets in the BM of ten healthy individuals and 24 MDS patients. Percentages were calculated from the total CD45⁺ mononuclear cell fraction. Mean frequencies ± standard error of mean (SEM) are given (NBM vs. MDS BM: classical monocytes 11.47% ± 1.86 vs. 11.11% ± 2.20, intermediate monocytes 0.45% ± 0.12 vs. 0.85% ± 0.17, non-classical monocytes 0.46% ± 0.11 vs. 0.38% ± 0.05). Furthermore, the percentage of monocytes that express TM is displayed (NBM vs. MDS BM: classical monocytes 9.94% ± 2.82 vs. 37.27% ± 4.00, intermediate monocytes 42.60% ± 7.27 vs. 54.90% ± 3.48, non-classical monocytes 45.11% ± 3.72 vs. 49.46% ± 3.34). Expression levels of TM and HLA-DR on NBM and MDS BM monocyte subsets are also shown. Mean MFI values ± SEM are shown for ten NBM and 24 MDS BM samples (TM in NBM vs. MDS BM: classical monocytes 382 ± 65 vs. 1425 ± 367, intermediate monocytes 1414 ± 245 vs. 2208 ± 259, non-classical monocytes 1279 ± 169 vs. 1676 ± 144. HLA-DR in NBM vs. MDS BM: classical monocytes 5426 ± 715 vs. 11010 ± 1056, intermediate monocytes 20062 ± 2529 vs. 35639 ± 3989, non-classical monocytes 10117 ± 856 vs. 15255 ± 1915). (C) Percentages of TM-expressing classical monocytes in peripheral blood (PB) and BM. Bars indicate mean frequencies [normal PB (NPB, n=31) vs. MDS PB (n=29): 17.8% vs. 33.6%. normal BM (NBM, n=25) vs. MDS BM (n=154): 8.6% vs. 37.0%]. TM expression was correlated in PB- and BM-derived classical monocytes. In total, 25 paired MDS samples were included. *P<0.05, **P<0.01, ****P<0.0001.

Figure 2.

Classical monocytes are clonally involved and the presence of thrombomodulin correlates with disease states. (A) Sorted cells, including B cells and CD34⁺ blast cells, from patients with a known cytogenetic aberrancy were subjected to fluorescence in situ hybridization (FISH) analysis. Furthermore, whole bone marrow (BM) samples were used for degree of cytogenetic load. Representative interphase cells hybridized with the chromosome 5q probe, showing loss of 5q in CD34⁺ blasts and monocytes, and no loss of 5q in B cells. In three tested cases (monosomy 7, deletion 5q and trisomy 8), classical monocytes and CD34⁺ blast cells were highly involved in the dysplastic clone, whereas B cells were not involved. Interphase FISH on whole BM samples showed both an aberrant and a normal cell line. (B) Thrombomodulin-positive (TM⁺) classical monocytes in different myelodysplastic syndrome (MDS) risk groups defined according to the Revised International Prognostic Scoring System and the World Health Organization (WHO) 2016 classification. The percentage of classical monocytes that express TM was highest in the very low/low-risk groups: the percentage of TM expression was significantly reduced in higher-risk groups compared to the lower-risk group [mean ± standard error of mean (SEM): normal BM (n=25) 11.3% ± 3.0% vs. very low/low-risk MDS BM (n=54) 40.1% ± 2.9% vs. intermediate-risk MDS BM (n=22) 22.7% ± 3.7% vs. high/very high-risk MDS BM (n=16) 28.3% ± 4.9%). Patients with low-risk MDS according to the WHO 2016 classification (single/multiple lineage dysplasia with or without ring sideroblasts, n=87) had higher percentages of TM expression on monocytes (40.9% ± 2.3%) compared to patients in higher-risk groups (excess blasts-1 and -2, n=42) (24.2% ± 2.7%) and normal BM (n=25) (11.3% ± 3.0%). (C) Percentages of classical monocytes that are positive for TM in patients with low and higher blast counts. The percentage of monocytes expressing TM was significantly higher in the group of patients with blast counts below 5% (n=97) than in the group of patients with blast counts ≥5% (n=44) (blast count <5%, 41.3% ± 2.2% vs. blast count ≥5%, 25.7% ± 2.8%). Patients with ring sideroblasts (RS) (n=40) had higher percentages of TM⁺ monocytes than patients who did not have RS (n=106) (RS yes, 45.4% ± 3.7% vs. RS no, 33.2% ± 2.0%). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. IPSS-R: Revised International Prognostic Scoring System; NBM: normal bone marrow; SLD: single lineage dysplasia; MLD: multilineage dysplasia;.RS: ring sideroblasts; EB, excess blasts.

Figure 3.

Deep phenotyping of T cells after co-culture with thrombomodulin-negative or –positive monocytes from patients with myelodysplastic syndromes. Healthy donor-derived CD4⁺ T cells were co-cultured with thrombomodulin (TM)-negative or -positive monocytes from two patients with myelodysplastic syndromes (MDS), or in the stimulated control condition with plate-bound anti-CD3 antibody only. After culture for 5 days, T cells were stained with an extensive panel of surface and intracellular markers as well as stains for transcription factors and cytokines and analyzed using time of flight mass cytometry. (A) Viable T cells were identified and visualized using stochastic neighbor embedding (viSNE). Different T-cell subsets were identified based on interferon (IFN)- γ, GATA3, interleukin (IL)-17, IL-4, FoxP3 and CD127 (Th1 were considered to be IFN-γ⁺, IL-17 and GATA3⁻; Th2 were GATA3⁺IL-4⁺; Th17 were IL-17⁺; Treg were CD127⁻ and FoxP3⁺CD25⁺ and IFN-γ⁻) (Online Supplementary Figure S6). Overlays were created from condition-specific biaxial viSNE contour plots and each individual T-cell subset. T-cell populations at the start of the experiment (D=0) and after 5 days of culture with anti-CD3 only, TM⁻ monocytes or TM⁺ monocytes are shown for a representative sample. (B) Cells were then further clustered with spanning-tree progression analysis of density-normalized events (SPADE) into 50 nodes using the clustering channels tSNE1 and tSNE2. Different clusters of nodes representing various T-cell subsets could be identified in the stimulated control condition by using selected markers. The color intensity in each node reflects the expression level of the indicated marker and the size of the node reveals the number of cells involved. (C) Frequencies of all clusters were compared between the TM⁻ and TM⁺ conditions. T-cell clusters that were most prevalent in the TM⁻ condition are highlighted in red. Black circles represent clusters that show higher percentages in the TM⁺ condition. The top five of highest frequencies for both conditions are shown. Using marker enrichment modeling (MEM), profiles of T-cell clusters with highest frequencies in TM⁻ or TM⁺ cells were characterized. Two subgroups of clusters were generated: (i) five nodes with highest frequency in the TM⁺ condition (called “up”) and (ii) five nodes with highest frequency in TM-condition (called “down”). Expression levels of given markers are shown for the TM⁺ as well as the TM⁻ conditions, showing that global expression in the identified set of clusters for the different markers is nearly similar in both conditions. MEM scores were calculated for the markers and results are presented in a heat map. The group of nodes that are present in a higher percentage in cultures with TM⁺ monocytes compared to cultures with TM⁻ monocytes reflect an anti-inflammatory profile. T cells in this group (all within “non-Treg nodes”) express higher levels of FoxP3, GATA3, CD279 (PD-1), IL-4 and IL-10, a phenotype which suggests they are polarizing toward a Treg phenotype. tSNE, t-distributed stochastic neighbor embedding.

Figure 4.

Overall and leukemia-free survival is related to the presence of thrombomodulin-positive monocytes. As a cut-off percentage for the presence of thrombomodulin (TM) on myelodysplastic syndrome (MDS) monocytes, the expression rate in the healthy donor cohort was used. The mean percentage + two standard deviations was calculated, resulting in a cut-off of 25.53%. Statistical differences were calculated using the log-rank test. (A) Overall survival data for 122 MDS patients and leukemia-free survival (LFS) data for 102 patients. A significant difference in overall survival was found between MDS patients with or without TM on BM monocytes (P=0.006). The median overall survival for patients with TM⁺ monocytes was 58 months while that for patients without TM was 30 months. The time to development of leukemia was significantly longer for patients with TM expression than for patients without TM expression (P=0.029). (B) Patients were further selected based on their low-risk status. Survival curves are shown for low-risk patients (according to the International Prognostic Scoring System and its revision) with or without TM⁺ monocytes. (C) LFS of low-risk patients in the TM⁺ and TM⁻ groups. IPSS: International Prognostic Scoring System; IPSS-R: Revised International Prognostic Scoring System.