Up‐regulation of SPINT2/HAI‐2 by Azacytidine in bone marrow mesenchymal stromal cells affects leukemic stem cell survival and adhesion

Abstract The role of tumour microenvironment in neoplasm initiation and malignant evolution has been increasingly recognized. However, the bone marrow mesenchymal stromal cell (BMMSC) contribution to disease progression remains poorly explored. We previously reported that the expression of serine protease inhibitor kunitz‐type2 (SPINT2/HAI‐2), an inhibitor of hepatocyte growth factor (HGF) activation, is significantly lower in BMMSC from myelodysplastic syndromes (MDS) patients compared to healthy donors (HD). Thus, to investigate whether this loss of expression was due to SPINT2/HAI‐2 methylation, BMMSC from MDS and de novo acute myeloid leukaemia (de novo AML) patients were treated with 5‐Azacitidine (Aza), a DNA methyltransferase inhibitor. In MDS‐ and de novo AML‐BMMSC, Aza treatment resulted in a pronounced SPINT2/HAI‐2 levels up‐regulation. Moreover, Aza treatment of HD‐BMMSC did not improve SPINT2/HAI‐2 levels. To understand the role of SPINT2/HAI‐2 down‐regulation in BMMSC physiology, SPINT2/HAI‐2 expression was inhibited by lentivirus. SPINT2 underexpression resulted in an increased production of HGF by HS‐5 stromal cells and improved survival of CD34+ de novo AML cells. We also observed an increased adhesion of de novo AML hematopoietic cells to SPINT2/HAI‐2 silenced cells. Interestingly, BMMSC isolated from MDS and de novo AML patients had increased expression of the integrins CD49b, CD49d, and CD49e. Thus, SPINT2/HAI‐2 may contribute to functional and morphological abnormalities of the microenvironment niche and to stem/progenitor cancer cell progression. Hence, down‐regulation in SPINT2/HAI‐2 gene expression, due to methylation in MDS‐BMMSC and de novo AML‐BMMSC, provides novel insights into the pathogenic role of the leukemic bone marrow microenvironment.


| BACKGROUND
Serine protease inhibitor kunitz-type 2 (SPINT2) codes the HAI-2 transmembrane protein. The HAI-2 protein is responsible for the inhibition of the enzyme hepatocyte growth factor activator (HGFA). When active, HGFA is responsible for proteolytic cleavages of the inactive form of the hepatocyte growth factor (pro-HGF) into HGF active form. 1 HGF is a multifunctional polypeptide that regulates diverse biological processes such as cell growth, apoptosis, cytokines secretion, adhesion, survival of hematopoietic cells, anti-inflammation, and immune-regulation. 2,3 HGF is secreted by mesenchymal stromal cells (MSC) and the action of this peptide is mediated by c-met tyrosine kinase pro-oncogene transmembrane receptor. 4 Interestingly, HGF cytokine serum levels are significantly increased in myelodysplastic syndromes (MDS) and de novo acute myeloid leukaemia (de novo AML) patients, both heterogeneous diseases, which are prevalent in the elderly, and are a prognostic marker that could predict survival. 5,6 Low levels of SPINT2/HAI-2, probably due to hypermethylation, have been highlighted in various solid cancer types and have been associated with disease progression. [7][8][9][10][11][12][13] Pereira and colleagues hypothesized that SPINT2/HAI-2 is involved in prostate cancer tumourigenesis probably by regulation of SPINT2/HAI-2. 7 We previously reported that the expression of SPINT2 mRNA is significantly lower in bone marrow mesenchymal stromal cell (BMMSC) from MDS patients compared to healthy donors (HD), which could be related with increased HGF and CXCL-12 secretion. 14 Despite being related to the pathogenesis of several neoplasms, the role of SPINT2/HAI-2 has not yet been fully elucidated in haematological malignances, such as MDS and de novo AML. Thus, in this study, we investigate whether this loss of expression was due to SPINT2/HAI-2 methylation in order to better understand the role of SPINT2/HAI-2 down-regulation in MDS and de novo AML physiopathology and its contribution to leukaemic bone marrow microenvironment.

| Mesenchymal stromal cell
The BM mononuclear cells were isolated using Ficoll-Hypaque Plus density-gradient centrifugation (GE Healthcare). The mononuclear cells were plated onto Dulbecco's modified Eagle's medium (DMEM) (Sigma) supplemented foetal bovine serum (FBS), glutamine, 100 μg/ mL penicillin, 100 μg/mL streptomycin, and amphotericin B in a humidified 5% carbon dioxide and 95% air incubator at 37°C. The supernatant with nonadherent cells was removed weekly and replaced with fresh supplemented medium. When the monolayer was established (approximately 90% confluence), cells were trypsinized and plated under the same conditions. After replating them three times, a homogeneous cell population was obtained and MSC were evaluated by flow cytometry for the absence of CD31, CD34, CD45, CD68, and HLA-DR antigens and the presence of CD73, CD90, and CD105.

| CD34 + cells from de novo AML patients
CD34 + cells were isolated from BM mononuclear cells by MIDI-MACS immunoaffinity columns (Miltenyi Biotec) and purity was determined by flow cytometry (minimum of 90%), using anti-CD34 antibody conjugated to allophycocyanin (APC; Becton Dickinson).

| Quantitative polymerase chain reaction (qPCR)
Total RNA was extracted from cells using the RNeasy Micro Kit (Qiagen) and cDNA was generated using RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). qRT-PCR was performed with SYBR Green Master Mix PCR (Thermo Fisher Scientific) using the ABI 7500 Sequence Detection System (Applied-Biosystem). The relative quantification gene expression values were calculated using the equation 2 −ΔΔCT15 with the housekeeping genes hypoxanthine guanine phosphoribosyltransferase 1 (HPRT1), beta actin (ACTB), and glyceraldehyde-3-phosphate desidrogenase (GAPDH).
The control was performed for each primer pair. Amplification specificity was verified using a dissociation curve at the end of each run.
Three replicas were run on the same plate for each sample.

| HGF secretion
Mesenchymal stromal feeder layers from shControl and shSPINT2 cells were seeded (1 × 10 5 cells/well) in serum-free RPMI plus BSA and incubated for 48 hours at 37°C. Culture supernatants from shControl and shSPINT2 cells were evaluated for HGF human cytokines secretion using a HGF Elisa Kit (ab100534, Abcam). The cytokine concentration was determined from the standard curve.
Moreover, shControl and shSPINT2 cells were seeded in a bioscaffold (1 × 10 5 cells/bioscaffold) produced by our laboratory. These bioscaffolds are well-preserved 3D microenvironment structures obtained from decellularized bovine bone marrow. Bioscaffolds were cultured for 7 days at 37°C. After this period, the cells on the scaffold were used for an immunohistochemistry (IC) analysis that was performed using the biotin-streptavidin immunoperoxidase method. Briefly, bioscaffolds were fixed in paraformaldehyde 1% for 5 min. Endogenous peroxidase activity was blocked with 1% H 2 O 2 and 1% BSA for 30 minutes and then incubated with antibody against HGF (sc13087, Santa Cruz Biotechnology) overnight at 4°C.
Following primary antibody incubation, cells were rinsed and incubated with biotinylated secondary antibodies using the Vectastain Elite ABC kit (Vector Laboratories) for 2 hours at room temperature, followed by incubation with diaminobenzidine for 6 minutes at room temperature. The cells were counter-stained with Harris' hematoxylin and mounted in permanent mounting medium. Stained cells were examined by light microscopy using an Eclipse i80 (Nikon).

| Adhesion molecules profile of mesenchymal stromal cells
Mesenchymal stromal feeder layers from primary cultures generated from MDS, de novo AML and HD BM aspirates were seeded onto plates (1 × 10 5 cells/well) in serum-free RPMI plus BSA and incubated for 48 hours. After this period, BMMSC were collected, washed, and resuspended in PBS. BMMSC were incubated for 30 minutes with monoclonal antibodies against the following adhesion alpha-family receptors CD49b, CD49d, and CD49e (BD Biosciences), evaluated by flow cytometer (FACSCalibur) and analysed using the FACS Diva software. A total of 10 000 events were collected per sample. The distribution histogram was used to determine the geometric mean of fluorescence intensity (MFI) for each antibody tested. The positivity degree for each tested surface adhesion receptor was expressed as a numerical MFI of the positively stained cells.

| Statistical analysis
Statistical analysis was performed using GraphPad Prism5 software.
Data were expressed as the median [minimum-maximum]. For comparisons, an appropriate Mann-Whitney or ANOVA or t Student test was used. The values of P < 0.05 were considered as statistically significant. All experiments were repeated at least three times independently.

| Increased SPINT2 gene and HAI-2 protein expression after MDS-BMMSC and AML-BMMSC treatment with AZA
As we previously observed a significantly decreased expression of SPINT2 mRNA in MDS-BMMSC compared to HD-BMMSC, 14

| Down-regulation of SPINT2 resulted in an increased HGF secretion, de novo AML CD34 + cell adhesion to HS-5 stromal cells and CD34 + de novo AML cells survival
To better understand the role of SPINT2/HAI-2 down-regulation in MDS-BMMSC and de novo AML-BMMSC physiology, we investigated the role of SPINT2 inhibition by transducing human HS-5 stromal cells with lentivirus-mediated shRNA targeting SPINT2 or an appropriate control. After puromicin selection, SPINT2 mRNA and HAI-2 protein levels were determined by qPCR and Western blot, respectively (Figure 2A,B). Significant decreases in SPINT2 mRNA (by 65 ± 3%, P < 0.001) and HAI-2 protein (by 68 ± 4%, P < 0.0001) levels were observed in shSPINT2 cells when compared with shControl cells (Figure 2A Moreover, transduced stromal cells were infused into a bioscaffold and were cultured for 7 days. We observed an improvement in cell growth into the bioscaffold containing cells with SPINT2/HAI-2 depletion ( Figure 3A,B). Moreover, our immunohistochemistry analysis showed that the SPINT2/HAI-2 silenced cell-seeded scaffold expressed more HGF than the control silenced cells (Figure 3C,D).
As HGF is capable of maintaining the stromal microenvironment niche, promoting hematopoiesis by inducing constitutive production of CXCL12, we also analysed CXCL12 expression. Interestingly, the SPINT2/HAI-2 silenced cell-seeded scaffold expressed more CXCL12 than the control silenced cells (Figure 3E,F). Since the matrix is ROVERSI ET AL.

| MDS-BMMSC and de novo AML-BMMSC has alteration in the alpha integrin expression profile
The increased CD34 + cell adhesion to MSC after SPINT2/HAI-2 knockdown, prompted us to evaluate the expression of alpha integrins, CD49b (α 2 ), CD49d (α 4 ), and CD49e (α 5 ), in BMMSC from 10 MDS patients, six de novo AML patients, and three healthy donors, as these molecules are known to mediate cell adhesion via HGF.

| DISCUSSION
In recent years, the role of tumour microenvironment niche in neoplasm initiation and malignant evolution/progression has been increasingly recognized. Leukemic stem cells are capable of modifying the interaction between hematopoietic stem cells and the bone marrow microenvironment to induce leukemogenesis. 16 However, the contribution of BMMSC to disease progression remains poorly explored. We had previously performed a microarray analysis of MDS patient-derived BMMSC (MDS-BMMSC) 17 and, after primary MDS and de novo AML BMMSC validation, 14 found an underexpression of the SPINT2 gene. This gene encodes a HAI-2 protein, an endogenous inhibitor of the HGF activator (HFGA), which is responsible for the activation of HGF, a polypeptide secreted by MSC that acts as a multifunctional cytokine, regulating cell adhesion, cell  18 In MDS and de novo AML patients, the levels of HGF cytokine in the serum are significantly increased and represent a predictor of survival. 5, 6 Kentsis and colleagues showed that up-regulation of HGF expression was responsible for the maintenance of leukemogenic signalling. 19 In diverse tumours, the down-regulation of SPINT2 expression has been associated with gene promoter hypermethylation. 7-13 AZA is a DNA methyltransferase inhibitor 20 that has been successfully used in epigenetic neoplastic therapy to reactivate epigenetically silenced tumour suppressor genes. 21 Furthermore, AZA is the first-line treatment for patients with high-risk myelodysplastic syndrome and has been successful in prolonging survival and delayed AML evolution. 22 These findings prompted us to examine whether methylation-mediated epigenetic silencing SPINT2/HAI-2 was involved in Myelodysplastic Syndrome and de novo AML pathophysiology.
Therefore, we obtained primary BMMSCs from healthy donor (HD), MDS, and de novo AML patients, isolated from bone marrow mononuclear cells. HD-BMMSC, MDS-BMMSC, and de novo AML-BMMSC were then treated with AZA. SPINT2 gene and HAI-2 protein expressions levels were evaluated by qPCR and Western blot.
We observed a significant increase in SPINT2/HAI-2 levels after AZA Mesenchymal stromal cells production of HGF has protective effects on hematopoietic cells due to anti-inflammatory, anti-fibrotic, anti-apoptotic, and cell-to-cell adhesion mechanisms. 2,3 Crosstalk between MSC and hematopoietic stem cells, in the bone marrow microenvironment, are essential to support stem cell behaviour and contribute to abnormal hematopoietic cell growth, survival, and maturation. 24,25 Moreover, the cell-to-cell contact between BMMSC and myelodysplastic and leukemic stem cells is important for cell selfrenewal, proliferation, and survival as well as for leukemogenesis. 26,27 In this concern, we evaluated the contribution of SPINT2/HAI-2