HSPG2 could promote normal haematopoiesis in acute myeloid leukaemia patients after complete remission by repairing bone marrow endothelial progenitor cells

2.1 Sample collection

For the prospective case–control study, BM samples of AML-CR patients received IA regimen (CR-IA patients, n = 20) or HAA regimen (CR-HAA patients, n = 20) as induction chemotherapy and their HCs (n = 20) were included and no significant differences existed in age or sex amongst three groups (Table S1). Samples were collected after consent was obtained, these processes were in compliance with the Declaration of Helsinki, which were approved by the Ethics Committee of Peking University People's Hospital.

2.2 Isolation, cultivation and characterisation of bone marrow endothelial progenitor cells

BM mononuclear cells and cell cultivation were isolated as previously reported.^{14-16, 21, 32, 33} For identification, BM EPCs were gated by CD45⁺CD34⁺CD309⁺ cells. The data were analysed by BD FACSDiva v8.0 Software (BD Biosciences). All the antibodies are listed in Table S2.

After cultured in 6-well plates for 7 days as previously reported,^14-16 BM EPCs were stained with DiI-acetylated low-density lipoprotein (LDL) (Life Technologies, Gaithersburg, USA) for 4 h at 37°C. After fixed with prechilled paraformaldehyde and stained with 10 µg/mL Ulex europaeus agglutinin-I (UEA I; Sigma, St. Louis, USA), the cells counted via fluorescence microscope (Olympus, Tokyo, Japan) in three random visual fields.

2.3 Tube formation and migration assays

4 × 10⁴ BM EPCs were plated in 24-well Matrigel-precoated plates as previously reported.^{16, 21, 33} Cell migration assays performed in Transwell chambers. In brief, 4 × 10⁴ BM EPCs were plated in the upper chambers, and 500 µL of the medium was added in the lower chamber, then cultured for 1 day and fixed for 15 min. 1% crystal violet was used to stain the cells on the underside of upper chambers. All assays were observed in three random fields via microscope (Olympus, Tokyo, Japan).

2.4  BM EPCs–HSCs, leukaemia stem cells, KG-1 cells or HL-60 cells in vitro coculture assay

To assess normal or malignant haematopoiesis-supporting ability of BM EPCs, the BM EPCs in each group were cocultured with CD34⁺ cells from HCs and AML patients, KG-1 cells or HL-60 cells. After 1 × 10⁵ BM EPCs/well were pre-plated on 24-well plates for 1 day, HSCs, leukaemia stem cells (LSCs), KG-1 cells or HL-60 cells (1 × 10⁵ cells/well) were added and cocultured for 5 days. Apoptotic cell ratio and reactive oxygen species (ROS) level were conducted as previously reported.^14-16

2.5 Colony-forming unit and leukaemia colony-forming unit assays

Colony-forming unit (CFUs) assays were conducted via the Metho-Cult H4434 method (Stem Cell Technologies, Vancouver, Canada). 2 × 10³ cells were cultured in 24-well plates after coculture for 2 weeks. The method for counting the numbers of CFUs and leukaemia colony-forming unit (CFU-Ls) were as previously reported.^{16, 21, 33}

2.6 5-Ethynyl-20-deoxyuridine assay

The cells were stained with 50 µM ethynyl-deoxyuridine (EdU) at 37°C for 1 h. The percentage of EdU⁺ cells was measured by BD FACSDiva v8.0 software (BD Biosciences).

2.7 An in vitro model of chemotherapy-induced damage model of bone marrow endothelial progenitor cells

BM EPCs from HCs were cultivated at 37°C for 1 day in multiple concentrations medium of cytosine arabinoside (Ara-C). For evaluation of the damage of Ara-C to BM EPCs, tube forming experiment of BM EPCs was conducted. A 10 µM Ara-C concentration has caused severe damage to BM EPCs and prevented the formation of tubes, so 10 µM Ara-C was selected.

2.8 Transfection of bone marrow endothelial progenitor cells from healthy controls

Small interfering RNA targeting HSPG2 (sequence: 5′-CGATGCGTCTCTACCAACA-3′) (RiboBio, China) was ersatz. The transfection process was conducted in 2×10⁶ BM EPCs of HCs with Lipofectamine 3000 reagent (Thermo Fisher Scientific, Waltham, USA) in 6-well plates.

2.9 Enzyme-linked immunosorbent assay

After centrifuging the BM samples for 5 min at 2000 rpm, plasma on the upper side was harvested. The HSPG2 levels were measured via human HSPG2 enzyme-linked immunosorbent assay (ELISAs) (Abcam, Cambridge, UK).

2.10 RNA-seq

RNA-seq was conducted to explore the changes in BM EPCs of HCs before and after HSPG2 gene knockdown. Differential gene expressions (DGEs), Gene Ontology biological process (GOBP) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment plots were generated via DESeq2, cluster profiler and ggplot2 packages in R (1.16.1).

2.11 qRT-PCR

The mRNA levels of HSPG2, ADRA2, ECM1, ANGPT2, IL34, DCSTAMP, VSIR, INHBA and AXL were determined via a qRT-PCR kit (Thermo Fisher Scientific, Waltham, USA).³⁴ The sequences of primer are shown in Table S3.

2.12 Statistical analysis

Data analyses were conducted by GraphPad Prism 8.0. The paired data were analysed by paired t-tests and the continuous variables were analysed by Mann‒Whitney U tests. Results are showed by means ± standard error of the mean (SEM), P-values < .05 were considered to have statistical significance.

3.1 HSPG2 levels and functions of BM EPCs decreased in AML patients after CR

The prospective case–control study was carried out to contrast HSPG2 level and BM EPC's functions between CR patients and HCs. Compared with HCs, the HSPG2 expression level in BM EPCs was significantly lower in CR-IA patients (Figure 1A,B; 2751.0 ± 127.3 vs. 4059.0 ± 161.7, P < .0001) and CR-HAA patients (Figure 1A,B; 2632.0 ± 105.0 vs. 4059.0 ± 161.7, P < .0001). Meanwhile, the HSPG2 level in BM plasma was significantly lower in CR-IA patients (Figure 1C; 11560.0 ± 493.2 vs. 18016.0 ± 853.7, P < .0001) and CR-HAA patients (Figure 1C; 11682.0 ± 483.3 vs. 18016.0 ± 853.7, P < .0001). However, there was no difference in HSPG2 levels of BM EPCs and BM plasma between the two groups of patients who received IA or HAA induction. Furthermore, there were significantly fewer double-positive stained BM EPCs in CR-IA patients (Figure 1D,E; 26.2 ± 4.4 vs. 59.0 ± 8.4, P = .004) and CR-HAA patients (Figure 1D,E; 25.1 ± 2.8 vs. 59.0 ± 8.4, P = .002) than those in HCs. Compared with HCs, the tube formation ability of BM EPCs was decreased in CR-IA patients (Figure 1F,G; 3844.0 ± 298.7 vs. 9582.0 ± 540.3, P = .002) and CR-HAA patients (Figure 1F,G; 3432.0 ± 147.9 vs. 9582.0 ± 540.3, P = .002). Meanwhile, the migration ability of BM EPCs was decreased in CR-IA patients (Figure 1H,I; 92.7 ± 6.1 vs. 181.3 ± 15.3, P = .002) and CR-HAA patients (Figure 1H,I; 105.0 ± 3.8 vs. 181.3 ± 15.3, P = .002). However, there was no difference in BM EPC's functions between the two groups of patients. Thus, our findings indicate that the HSPG2 level was significantly decreased in the BM EPCs of CR patients, which may be related to the decreased quality and functions of BM EPCs.

3.2 HSPG2 knockdown impaired the functions of BM EPCs from HCs, especially their haematopoiesis-supporting ability

To clarify the impact of chemotherapy on the HSPG2 levels and the effects of HSPG2 levels on BM EPC functions, different concentrations of Ara-C intervention and HSPG2 gene knockdown assay were conducted, respectively. The pattern diagram shows the study design (Figure 2A). The relative HSPG2 mRNA expression in BM EPCs revealed that the siHSPG2 2^# sequence had the highest knockdown efficacy (Figure S1A, 0.2 ± 0.01-fold, P < .0001). After knockdown the HSPG2 gene by 2^# sequence, the HSPG2 level in BM EPCs was significantly lower than the siNC group via flow cytometry (Figure S1B; 4001.0 ± 110.1 vs. 5192.0 ± 71.9, P = .02). Thus, the 2^# sequence of siHSPG2 was used for subsequent experiments. The relative HSPG2 mRNA expression level in BM EPCs was decreased as the concentration of Ara-C increased (Figure S1C). In terms of BM EPC's quantity, the number of double-positive-stained cells was notably lower in the siHSPG2 group (Figure 2B,C; 30.3 ± 3.9 vs. 47.2 ± 5.2, P = .01). Moreover, BM EPCs in the siHSPG2 group presented a decreased tube formation ability (Figure 2D,E; 4911.0 ± 817.5 vs. 9077.0 ± 604.2, P < .0001), a worse migration ability (Figure 2F,G; 104.8 ± 8.8 vs. 167.0 ± 17.2, P = .004) and a significantly increased apoptosis rate (Figure 2H; 1.3 ± 0.1-fold, P = .031). Moreover, the siHSPG2 group showed a substantial increase in the apoptosis rate (Figure 2I; 1.4 ± 0.1%-fold, P = .035) and ROS level (Figure 2J; 1.2 ± 0.03-fold, P = .001) of cocultured CD34⁺ cells. Together, these data further verify that Ara-C led to the decrease in the HSPG2 level of BM EPCs, which may be related to their impaired functions.

3.3 HSPG2 alleviated Ara-C induced dysfunction in BM EPCs from HCs

To explore the roles of Ara-C and HSPG2 on BM EPCs, we divided 7-day cultivated HC EPCs into the following groups for Ara-C and HSPG2 intervention: the control (CTL) group, the HSPG2 intervention (HSPG2) group, the Ara-C intervention group (Ara-C) group and the HSPG2+Ara-C intervention (Ara-C+HSPG2) group. The Ara-C treated BM EPCs presented a declined quality of double-positive-stained cells (Figure 3A,B; 20.0 ± 2.5 vs. 52.0 ± 6.7, P = .002), an impaired tube formation ability (Figure 3C,D; 3298.0 ± 221.8 vs. 8728.0 ± 372.7, P = .0002), a worse migration ability (Figure 3E,F; 100.2 ± 10.4 vs. 226.3 ± 19.8, P = .0003), increased levels of cell apoptosis (Figure 3G; 11.6 ± 1.8% vs. 6.3 ± 0.7%, P = .013) and ROS level (Figure 3H,I; 1.4 ± 0.1-fold, P = .03). These data indicate that Ara-C treatment impaired both the cellular status and functions of BM EPCs. Although HSPG2 treatment led to no obvious differences in the quality or functions of BM EPCs in the CTL group (Figure 3A–I), we found that the Ara-C-impaired BM EPCs were repaired by the HSPG2 treatment, as demonstrated by the increased number of double-positive-stained cells (Figure 3A,B; 30.2 ± 3.4 vs. 20.0 ± 2.5, P = .001), restored tube formation ability (Figure 3C,D; 5621.0 ± 188.9 vs. 3298.0 ± 221.8, P = .0002), migration ability (Figure 3E,F; 161.8 ± 17.0 vs. 100.2 ± 10.4, P = .0007), decreased level of cell apoptosis (Figure 3G; 7.7 ± 1.3% vs. 11.6 ± 1.8%, P = .003) and decreased ROS level (Figure 3H,I; 1.2 ± 0.1-fold vs. 1.4 ± 0.1-fold, P = .006). These results uncovered that the HSPG2 treatment can reverse the chemotherapy-induced BM EPC damage in vitro.

3.4 HSPG2 recovered the impaired haematopoiesis-supporting ability in BM EPCs from HCs caused by Ara-C

To investigate the effects of chemotherapy on the haematopoiesis-supporting ability of BM EPCs, we performed the BM EPC-CD34⁺ cell coculture assays. After coculture for 5 days, the percentage of apoptotic cells, the level of ROS and the CFU efficiencies of CD34⁺ cells were determined. Consistent with the results of the functional assays, the Ara-C treatment severely damaged the haematopoiesis-supporting ability of BM EPCs, as measured by the significantly increased level of apoptosis rate (Figure 4B,C; 0.9% ± 0.04%-fold vs. 0.5% ± 0.09%-fold, P = .005) and ROS level (Figure 4D; 1.0 ± 0.08-fold vs. 0.8 ± 0.07-fold, P = .001), and decreased CFU efficacy of CD34⁺ cells (Figure 4E): colony forming unit-erythroid (CFU-E) (1.3 ± 0.1-fold vs. 1.8 ± 0.1-fold, P = .001), burst forming unit-erythroid (BFU-E) (1.4 ± 0.07-fold vs. 2.1 ± 0.09-fold, P = .004), and colony forming unit-erythroid, granulocyte-macrophage and megakaryocyte (CFU-GEMM) (1.0 ± 0.07-fold vs. 1.8 ± 0.11-fold, P = .003).

The HSPG2 treatment reversed the impaired haematopoiesis-supporting ability caused by Ara-C, with a notably decreased level of the apoptosis rate (Figure 4B,C; 0.6% ± 0.08%-fold vs. 0.9% ± 0.04%-fold, P = .004), the decreased ROS level and significantly increased CFU efficacy of CD34⁺ cells (Figure 4E): CFU-E (1.7 ± 0.1-fold vs. 1.3 ± 0.1-fold, P = .01), BFU-E (1.8 ± 0.11-fold vs. 1.4 ± 0.07-fold, P = .02), CFU-GEMM (1.6 ± 0.09-fold vs. 1.0 ± 0.07-fold, P = .007). These results indicate that the HSPG2 treatment can restore the haematopoiesis-supporting ability of BM EPCs damaged by Ara-C in vitro.

3.5 HSPG2 repaired BM EPC function and improved the haematopoiesis-supporting ability of BM EPCs from AML-CR patients

In vitro treatment with HSPG2 improved the number of double-positive stained cells (Figure 5A,B; 32.4 ± 2.1 vs. 19.6 ± 0.7, P = .001), the tube formation ability (Figure 5C,D; 6615.0 ± 529.1 vs. 4426.0 ± 413.5, P = .0008) and the migration capacity (Figure 5E,F; 200.3 ± 8.7 vs. 140.2 ± 6.2, P = .0003) of CR EPCs. Moreover, the HSPG2 treatment decreased the cell apoptosis rate (Figure 5G; 5.6% ± 1.1% vs. 9.9% ± 0.9%, P = .002) and the ROS level (Figure 5H; 0.9 ± 0.04-fold, P = .037) of CR EPCs. The HSPG2 treatment promoted the haematopoiesis-supporting ability of CR EPCs, as measured by notably decreased level of the apoptosis rate (Figure 5I; 15.4% ± 3.03% vs. 19.6% ± 2.57%, P = .035), decreased ROS level (Figure 5J; 0.8 ± 0.04-fold vs. 0.9 ± 0.03-fold, P = .040) of CD34⁺cells enhanced CFU efficacy of CD34⁺cells(Figure 5K), including CFU-E (1.5 ± 0.12-fold vs. 1.2 ± 0.04-fold, P = .009), BFU-E (1.7 ± 0.11-fold vs. 1.3 ± 0.09-fold, P = .036), colony forming unit-granulocyte and macrophage (CFU-GM) (1.6 ± 0.06-fold vs. 1.3 ± 0.07-fold, P = .007) and CFU-GEMM (1.6 ± 0.16-fold vs. 1.3 ± 0.10-fold, P = .018). The results above proved that the reduced haematopoiesis-supporting ability of CR EPCs could be restored by the HSPG2 treatment in vitro.

3.6 HSPG2 did not increase the leukaemia cell-supporting ability of BM EPCs from AML-CR patients

To explore the role of HSPG2 on malignant haematopoiesis, we used a series of in vitro coculture assays, and a schematic diagram of the experiment plan is provided (Figure 6A). Importantly, the HSPG2 treatment did not increase the leukaemia cell-supporting ability of CR EPCs, which specifically manifested as no significant increase in cell proliferation ability (Figure 6B,C), no significant decrease in cell apoptosis (Figure 6D) and no significant changes in the CFU-L formation ability (Figure 6E,F). These findings indicate that HSPG2 promotes the normal haematopoiesis-supporting ability of CR EPCs (Figure 5K), without affecting their leukaemia cell-supporting ability (Figure 6F).

3.7 RNA-seq analysis revealed the internal changes in HSPG2 knockdown BM EPCs from HCs

To explore the possible mechanism, RNA-seq was conducted and a schematic diagram of our experimental plan is provided (Figure 7A). Amongst the 17761 genes, 351 were upregulated whilst 461 were downregulated in the siHSPG2 group (Figure 7B). Moreover, DGEs were enriched in GOBP terms, as shown in the bubble diagram: myeloid cell differentiation, regulation of angiogenesis and regulation of haemopoiesis were decreased, whereas negative regulation of cell migration, locomotion and adhesion were increased in the BM EPCs of the siHSPG2 group (Figure 7C).

To further verify, we validated the downregulated genes associated with these biological behaviours through qPCR. The mRNA expression levels of endothelial cell functional genes, including ADRA2, ECM1 and haematopoietic regulation related genes, including ANGPT2, IL-34, DCSTAMP, VSIR, INHBA and AXL, were decreased in the EPCs of the siHSPG2 group (Figure 7D), which are in accordance with our in vitro experiments. The above results confirmed that HSPG2 is a key molecule in the regulation of EPC functions.