Dysfunctional bone marrow endothelial progenitor cells are involved in patients with myelodysplastic syndromes

Patients with MDS (N = 30) and de novo AML (N = 15) were enrolled. Newly diagnosed MDS patients (N = 30) were divided into lower-risk MDS (N = 15) and higher-risk MDS (N = 15) groups (Additional file 1: Table S1) according to the dichotomization of the Revised International Prognostic Scoring System (IPSS-R) [37]. AML patients were diagnosed with de novo M2 (N = 8), M4 (N = 2) or M5 (N = 5) disease [38]. Bone marrow cells from allogenetic transplantation donors (N = 15) were used as normal controls. The age of lower-risk MDS (51.3 years, range 29–69), higher-risk MDS (48.4 years, range 24–68), AML (51.2 years, range 26–64) patients and HDs (49.5 years, range 29–64) showed no significant differences.

As previously reported [28‐30, 39, 40], BM mononuclear cells (BMMNCs) were isolated by density gradient centrifugation using lymphocyte separation medium (GE Healthcare, Milwaukee, USA). Precultivated BM EPCs were characterized by staining with mouse anti-human CD45 (BD Biosciences, San Jose, USA), CD34 (BioLegend, San Diego, USA), CD133 (Miltenyi Biotec, Bergisch Gladbach, Germany), and vascular endothelial growth factor receptor 2 (VEGFR2, CD309) monoclonal antibodies (BD Biosciences, San Jose, USA). Data were analysed with BD FACSDIVA v8.0 Software (BD Biosciences).

BMMNCs (5 × 10⁶ per well) were cultured in fibronectin-precoated (Sigma, St. Louis, USA) 6-well culture plates with EGM-2-MV-SingleQuots (Lonza, Walkersville, USA) and 10% foetal bovine serum (FBS; Gibco, Australia) at 37 °C in a humidified incubator with 5% CO₂ for 7 days until testing.

BM adherent cells at day 7 of cultivation were further functionally characterized as BM EPCs for their capacity to uptake Acetylated low-density lipoprotein (Ac-LDL) and to bind Ulex europaeus agglutinin I (UEA I) [28‐30, 40, 41]. The adherent cells were incubated with DiI-AcLDL (Life Technologies, Gaithersburg, USA) at 37 °C. After 4 h, the cells were fixed with 4% prechilled paraformaldehyde and incubated with 10 μg/ml fluorescein isothiocyanate-labeled-labelled UEA I (FITC-UEA I; Sigma, St. Louis, USA) for 1 h. To evaluate the numbers of double-positive-stained EPCs, three power fields were randomly counted using a fluorescence microscope (Olympus, Tokyo, Japan).

After 7 days of cultivation, EPC identity was confirmed by real-time quantitative polymerase chain reaction (qRT-PCR) for endothelial specific marker genes (VEGFR2, VE-cadherin and vWF) [41, 42].

A total of 4 × 10⁴ EPCs at day 7 of cultivation were transferred to matrigel-coated (Corning, New York, USA) plates and incubated for 48 h at 37 °C in 5% CO₂. Tube formation [28, 29, 39, 40] was measured by determining the relative tube length per field of view using an inverted light microscope. All cells were counted in three random fields.

Cell migration [28‐30, 39, 40, 43] was assayed using a transwell chamber (Corning, New York, USA). The EPCs after 7 days of culture were seeded in the upper chambers at a density of 4 × 10⁴ cells per well, while 500 μl medium was added to the lower chamber. The cells were cultured for 24 h, and migrated cells were fixed with paraformaldehyde for 30 min. Then, cells on the bottom surface of the membrane were stained with crystal violet for 20 min and counted manually in three random fields/sample. Cell images were obtained on a phase-contrast microscope (Olympus, Tokyo, Japan).

In order to evaluate the effect of supporting ability of BM EPCs to normal HSCs, T cells or HL-60 cells, the cultivated BM EPCs from HDs or MDS or AML patients were cocultured with normal CD34⁺ cells, or normal CD3⁺ cells, or HL-60 cells [28‐30, 40, 44]. HSCs or T cells were isolated from BMMNCs of HDs using CD34 or CD3 MicroBead kits (Miltenyi Biotec, Bergisch Gladbach, Germany). Cultivated EPCs (described before) were plated onto gelatinized 24-well culture plates at 1 × 10⁵ cells/well and cultured overnight to achieve confluence. Then, HSCs (1 × 10⁵ per well) or T cells (1 × 10⁵ per well) or HL-60 cells (Manassas, Virginia, USA; 5 × 10⁴ per well) were added in direct contact with confluent and adherent EPCs. EPC-HSC cocultures were maintained in StemSpan™ SFEM (Stem Cell Technologies, Vancouver, Canada) for 5 days. EPC-T cells or HL-60 cells cocultures were maintained in RPMI 1640 medium supplemented with 10% FBS for 3 days or 5 days. Appropriate controls of T cell culture alone were also included.

To detect ROS levels of HSCs and precultivated EPCs [28, 29, 40], HSCs were stained with CD34, and BMMNCs were stained with the aforementioned EPC markers and incubated with 10 μM 2ʹ,7ʹ-dichlorofluorescein diacetate (DCFH-DA, Byotime, Shanghai, China) at 37 °C for 15 min. All data were analysed on BD FACSDIVA v8.0 Software (BD Biosciences).

To detect ROS levels on day 7 of cultivated BM EPCs, adherent cells were incubated with 1 μg/ml DCFH-DA at 37 °C for 20 min. Images were obtained in three random fields/sample using a fluorescence microscope (Olympus, Tokyo, Japan). The fluorescence intensity of ROS was analysed via the mean grey value using ImageJ 1.52v (National Institutes of Health, Bethesda, USA).

To detect the apoptosis ratio, HSCs, precultivated EPCs or HL-60 cells were incubated with Annexin-V (BioLegend, San Diego, USA) and 7-amino-actinomycin D (7-AAD; BD Biosciences, San Jose, USA) for 10 min at room temperature and then analysed on BD FACSDIVA v8.0 Software (BD Biosciences).

CFUs were assayed using MethoCult™ H4434 Classic (Stem Cell Technologies, Vancouver, Canada). After 5 days of coculture, 2 × 10³ CD34⁺ cells or 2 × 10³ HL-60 cells were plated in 24-well plates and cultured for 14 days. Colony-forming unit erythroid (CFU-E), burst-forming unit erythroid (BFU-E), colony-forming unit-granulocyte/macrophages (CFU-GM), and colony-forming unit-granulocyte, erythroid, macrophage and megakaryocyte (CFU-GEMM) measurements for CD34⁺ cells [28, 30, 44] and CFU-L measurements for HL-60 cells [45] were scored.

After 3 days of coculture, CD3⁺ T cells were stimulated with a cell stimulation cocktail (eBioscience, San Diego, USA) to induce CD3⁺ T cell activation and cytokine secretion. Lymphocyte subpopulations were quantified via flow cytometry as previously described [46‐49]. Th1, Th2, Th17 cells, and Tregs were identified as CD3⁺CD8⁻IFN-γ⁺, CD3⁺CD8⁻IL-4⁺, CD3⁺CD8⁻IL-17A⁺, and CD3⁺CD8⁻CD25⁺Foxp3⁺ populations, respectively. The details of antibodies were in Additional file 1: Table S2.

HL-60 cells were harvested after coculture with EPCs and then incubated with 50 μM EdU (RiboBio, Guangzhou, China) in 48-well plates for 1 h at 37 °C. Then, according to the manufacturer’s instructions, the nuclear fluorescence intensity was analysed on BD FACSDIVA v8.0 Software (BD Biosciences).

RNA-seq analyses were performed to analyse the 7-day cultivated BM EPCs from HDs, lower-risk MDS, higher-risk MDS or AML patients. The accession number of whole transcriptome RNA-seq data is GSE197907. Differential gene expression (DEGs), principal component analysis (PCA), hierarchical clustering analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment plot were executed by the DESeq2, clusterProfiler, pheatmap, and ggplot2 packages in R (1.16.1). The top 20 up and down regulated genes in higher-risk MDS BM EPCs than lower-risk MDS BM EPCs were list in Additional file 1: Table S3 and in AML BM EPCs than higher-risk MDS BM EPCs were list in Additional file 1: Table S4. Gene set enrichment analysis (GSEA) analysis was performed with respect to MSigDB (version 5.1) genesets C5 GO biological processes [50].

For qRT-PCR, RNA was extracted using the RNeasy Mini kit (QIAGEN, Dusseldorf, Germany). One microgram of RNA was reverse transcribed into cDNA by the RT reagent Kit with gDNA Eraser (TaKaRa, Otsu, Japan). The mRNA levels of VEGFR2, VE-cadherin, VWF, CASP2, CASP3, BAX, CCNE1, MCL1, TP53, CDKN1A, CXCL12, KITLG, NFKB1, HAVCR2, CIITA, and LGALS9 were detected by the SYBR-Green qRT-PCR kit (Thermo Fisher Scientific, Waltham, USA). The levels of the aforementioned genes were evaluated after normalization to the 18S mRNA level [30]. All sequences of primers were list in Additional file 1: Table S5.

The representative BM EPC phenotype was characterized by CD34⁺CD309⁺CD133⁺ by flow cytometry (Fig. 1A). The percentage of BM EPCs from higher-risk MDS patients (higher-risk MDS BM EPCs) (Fig. 1A, 0.19% ± 0.03% vs. 0.11% ± 0.02%, P = 0.03) was significantly higher than BM EPCs from lower-risk MDS patients (lower-risk MDS BM EPCs), whereas the percentage of primary BM EPCs from AML patients (AML BM EPCs) was higher than higher-risk MDS BM EPCs (Fig. 1A, 0.31% ± 0.03% vs. 0.19% ± 0.03%, P = 0.002).

After 7 days of cultivation, spindle-shaped and elongated BM adherent cells were further functionally characterized as EPCs, which were capable of DiI-AcLDL uptake and FITC-UEA I binding (the typical functional EPC markers) [28‐30, 40, 41]. Moreover, EPC identity was confirmed by qRT-PCR for endothelial specific marker genes (VEGFR2, VE-cadherin and VWF) in Additional file 1: Fig. S1. The number of double-positive stained AML BM EPCs was significantly higher than higher-risk MDS BM EPCs (Fig. 1B and C, 124.5 ± 20.3 vs. 41.7 ± 6.3, P = 0.003). Together, increased numbers of BM EPCs were found in MDS patients, especially in more severe types of MDS.

To evaluate the angiogenic potential and migration of BM EPCs from different types of MDS patients, tube formation and migration were analysed on day 7 of culture [28‐30, 40]. Higher-risk MDS BM EPCs showed a significantly increased tube formation ability compared with lower-risk MDS BM EPCs (Fig. 1D and E, 4375 ± 321 vs.1874.7 ± 118.9, P < 0.0001) and a markedly decreased tube formation ability compared with AML patients (Fig. 1D and E, 4375 ± 321 vs. 6403 ± 370.4, P = 0.001). The AML BM EPCs on day 7 of culture showed increased migrated cells (Fig. 1F and G, 172.4 ± 29.8 vs. 82.6 ± 7.9, P = 0.01) compared with lower-risk MDS BM EPCs. These results demonstrated that the decreased angiogenic potential but increased migration capability in BM EPCs from MDS patients.

To uncover the underlying mechanism of the variant dysfunctions in different types MDS BM EPCs, the lower-risk MDS BM EPCs (N = 3), higher-risk MDS BM EPCs (N = 3), AML patients BM EPCs (N = 3) and primary BM EPCs from HDs (HD BM EPCs) (N = 3) were analysed via RNA-seq (Fig. 2A). The results of PCA (Fig. 2B) and hierarchical clustering analysis (Fig. 2C) showed that the entire population was clearly separated into two distinct subpopulations, HD and disease BM EPCs. More importantly, the heatmap (Fig. 2C) of different groups showed the progressive change in the total RNA expression profile of HD BM EPCs, lower-risk MDS BM EPCs, and higher-risk MDS BM EPCs to AML BM EPCs. There were 886 different genes between lower-risk MDS and higher-risk MDS BM EPCs (Fig. 2D) and 825 different genes between higher-risk MDS and AML BM EPCs (Fig. 2E). KEGG pathway enrichment analysis of Organismal Systems class (Fig. 2F and G) indicated that haematopoiesis- and immune-related pathways were enriched in MDS BM EPCs. For example, haematopoietic cell lineage was enriched in both lower-risk MDS BM EPCs vs. higher-risk MDS BM EPCs and higher-risk MDS BM EPCs vs. AML BM EPCs. Th17 and Th2 cell differentiation and Th17 cell differentiation were enriched in higher-risk MDS BM EPCs vs. AML BM EPCs.

In addition, the previously reported haematopoiesis- and immunomodulation-related genes [51‐56] in BM EPCs were detected by RNA-seq (Fig. 2H) and verified by qRT-PCR. As shown in Fig. 2H–J, mRNA levels of haematopoiesis-related genes CXCL12, KITLG, NFKB1 and immune-related genes HAVCR2, LGALS9, CIITA were increased in higher-risk MDS and AML BM EPCs compared with lower-risk MDS BM EPCs. These results suggested markedly enriched haematopoiesis- and immunomodulatory-related pathways and upregulation of related genes in higher-risk MDS BM EPCs. However, normal or malignant haematopoiesis-related genes could not be distinguished via RNA-seq analysis. Therefore, we further carried out in vitro experiments to analyse the haematopoiesis and immunomodulatory function of BM EPCs from MDS patients.

To investigate the effects of BM EPCs on HSCs in vitro, we sorted CD34⁺ cells from BMMNCs of HDs with magnetic beads and cocultured them with BM EPCs. After 5 days, we analysed the apoptosis rates, intracellular ROS levels and CFU plating efficiencies of HSCs (Fig. 3A). Compared with the lower-risk MDS group, the higher-risk MDS group exhibited a marked increase in the apoptosis rates of HSCs after coculture (Fig. 3B and C, 21.09 ± 1.21% vs. 16.79 ± 0.97%, P = 0.01). The ROS level (Fig. 3D, 6039 ± 654.4 vs. 4308 ± 394.7, P = 0.03) of CD34⁺ cells after coculture with AML BM EPCs was significantly higher than that of lower-risk MDS BM EPCs. HSCs cocultured with higher-risk MDS BM EPCs had lower CFU plating efficiencies than those cocultured with lower-risk MDS BM EPCs (Fig. 3E), as determined by CFU-E, BFU-E (12.9 ± 1.67 vs. 26.9 ± 3.62, P = 0.003), CFU-GM, and CFU-GEMM. The AML group had lower CFU-E, BFU-E, CFU-GM and CFU-GEMM plating efficiencies than the higher-risk MDS group. These data suggest that of MDS BM EPCs were less able to support HSCs in more severe type of MDS patients.

To further investigate the immunoregulatory effects of BM EPCs in vitro, we sorted CD3⁺ cells from BMMNCs of HDs and cocultured them without or with HD BM EPCs, lower-risk MDS BM EPCs, higher-risk MDS BM EPCs and AML BM EPCs. After 3 days, we analysed differences in T cell subsets after coculture or not (control of T cell alone cultures) (Fig. 4A). The subtypes of Th1 cells, Th2 cells, Th17 cells and Tregs were characterized by positive expression of CD4⁺IFN-γ⁺, CD4⁺IL-4⁺, CD4⁺IL-17⁺ and CD4⁺CD25⁺Foxp3⁺, respectively (Fig. 4B). There was a markedly lower percentage of Th1 cells in the AML group than in the higher-risk MDS group (Fig. 4C, 4.86% ± 0.54% vs. 8.07% ± 0.51%, P = 0.004) and in the higher-risk MDS group than in the lower-risk MDS group (Fig. 4C, 8.07% ± 0.51% vs. 9.95% ± 0.6%, P = 0.04). A significantly lower percentage of Th17 cells was observed in the higher-risk MDS group than in the lower-risk MDS group (Fig. 4D, 6.54% ± 0.43% vs. 9.35% ± 0.92%, P = 0.004) and in the AML group than in the higher-risk MDS group (Fig. 4D, 4.48% ± 0.58% vs. 6.54% ± 0.43%, P = 0.04). However, higher percentages of Th2 cells (Fig. 4E, 8.77% ± 1.0% vs. 5.67% ± 0.31%, P = 0.02) in the higher-risk MDS group than in the lower-risk MDS group. In addition, higher percentages of Tregs (Fig. 4F, 4.74% ± 0.59% vs. 3.14% ± 0.31%, P = 0.01) were observed in the AML group than in the higher-risk MDS group. As a result, the ratio of Th1/Th2 cells was markedly lower in the higher-risk MDS group than in the lower-risk MDS group (Fig. 4G, 1.0 ± 0.1 vs. 1.81 ± 0.17, P = 0.004), whereas that was significantly lower in the AML group than in the higher-risk MDS group (Fig. 4G, 0.51 ± 0.06 vs. 1.0 ± 0.1, P = 0.02). These data showed that lower-risk MDS BM EPCs regulated T cell differentiation into Th17 cells, whereas higher-risk MDS and AML BM EPCs regulated T cell differentiation into Th2 cells and Tregs. Our data suggested that BM EPCs might be inclined to induce T cell differentiation towards more immune-tolerant cells in higher-risk MDS patients.

To investigate the effect of BM EPCs on leukaemia cells in vitro, we assessed the proliferation, apoptosis and CFU-L plating efficiency of HL-60 cells after coculture with BM EPCs (Fig. 5A). In addition, the relative mRNA expression levels of apoptosis- and cell cycle-related genes were detected in HL-60 cells after coculture. The EdU -positive rate of HL-60 cells was significantly higher in the higher-risk MDS group than in the lower-risk MDS group (Fig. 5B and C, 45.79% ± 1.12% vs. 36.52% ± 2.20%, P = 0.003). More importantly, CFU-L plating efficiencies were markedly increased in AML group compared with higher-risk MDS group (Fig. 5D and E, 374.1 ± 37.2 vs. 257.6 ± 33.2, P = 0.04). However, the apoptotic rate of HL-60 cells was notably decreased in the AML group compared with the lower-risk MDS group (Fig. 5F, 2.6% ± 0.52% vs. 4.38% ± 0.46%, P = 0.01). The CASP2, CASP3, BAX, TP53 and CDKN1A mRNA levels in HL-60 cells were downregulated, whereas the CCNE1 and MCL1 mRNA levels were upregulated in the higher-risk MDS and AML groups than the lower-risk MDS group (Fig. 5G). These data suggest that BM EPCs may become more supportive of leukaemia cells in higher-risk MDS patients.

To further explore the internal changes in dysfunctional EPCs and identify underlying therapeutic targets, we investigated damage related pathways from the RNA-seq data. GSEA highlighted abnormalities of mitochondrial signaling pathway (Fig. 6A, B), so we verified them by detecting ROS levels of precultivated and cultivated BM EPCs. The ROS level of precultivated AML BM EPCs was significantly higher than that in higher-risk MDS BM EPCs (Fig. 6C, 5369 ± 426.0 vs. 3039 ± 335.3, P = 0.0002). The ROS level in the cultivated AML BM EPCs was higher than that in the cultivated higher-risk MDS BM EPCs, whereas the ROS level of the cultivated higher-risk MDS BM EPCs was higher than that in the lower-risk MDS BM EPCs (Fig. 6D, E). Moreover, the pathway of apoptotic nuclear changes was identified in GSEA (Fig. 6F). The apoptosis rate (Fig. 6G, 43.79% ± 2.77% vs. 20.63% ± 3.01%, P < 0.0001) was significantly higher in AML BM EPCs than in lower-risk MDS EPCs. The relative mRNA expression levels of apoptosis-related genes in BM EPCs were further analysed by qRT-PCR. The CASP2 mRNA level was significantly upregulated in AML BM EPCs (Fig. 6H, 6.43 ± 0.89-fold vs. 2.28 ± 0.56-fold, P = 0.03) compared to higher-risk MDS BM EPCs. The CASP3 (Fig. 6H, 2.09 ± 0.26-fold vs. 0.88 ± 0.3-fold, P = 0.03) and BAX (Fig. 6H, 2.0 ± 0.35-fold vs. 1.08 ± 0.14-fold, P = 0.03) mRNA levels were significantly upregulated in higher-risk MDS BM EPCs compared to lower-risk MDS BM EPCs. Together, elevated levels of ROS and apoptosis of BM EPCs were found in MDS patients, especially in higher-risk MDS patients, which may be the underlying repair targets for MDS patients (Fig. 7).