MiR21 sensitized B-lymphoma cells to ABT-199 via ICOS/ICOSL-mediated interaction of Treg cells with endothelial cells

Background MicroRNAs (miRs) are involved in tumor progression by regulating tumor cells and tumor microenvironment. MiR21 is overexpressed in diffuse large B-cell lymphoma (DLBCL) and its biological impact on tumor microenvironment remains unclear. Methods MiR21 was assessed by quantitative RT-PCR in patients with newly diagnosed DLBCL. The mechanism of action of miR21 on lymphoma progression and tumor angiogenesis was examined in vitro in B-lymphoma cell lines and in vivo in a murine xenograft model. Results Serum miR21 was significantly elevated in patients and associated with advanced disease stage, International Prognostic Index indicating intermediate-high and high-risk, and increased tumor angiogenesis. When co-cultured with immune cells and endothelial cells, miR21-overexpressing B-lymphoma cells were resistant to chemotherapeutic agents, but sensitive to Bcl-2 inhibitor ABT-199, irrespective of Bcl-2 expression on lymphoma cells. In both co-culture systems of Bcl-2positive and Bcl-2negative B-lymphoma cells, miR21 induced inducible co-stimulator (ICOS) expression on regulatory T (Treg) cells. Through crosstalking with Treg cells by ICOS ligand (ICOSL), endothelial cells were activated, resulting in stimulation of Bcl-2 expression and vessel formation. ABT-199 directly targeted Bcl-2 on endothelial cells, induced endothelial cell apoptosis and inhibited tumor angiogenesis. In a murine xenograft model established with subcutaneous injection of B-lymphoma cells, ABT-199 particularly retarded the growth of miR21-overexpressing tumors, consistent with the induction of endothelial cell apoptosis and inhibition of tumor angiogenesis. Conclusions As a serum oncogenic biomarker of B-cell lymphoma, miR21 indicated B-lymphoma cell sensitivity to ABT-199 via ICOS/ICOSL-mediated interaction of Treg cells with endothelial cells. Electronic supplementary material The online version of this article (doi:10.1186/s13046-017-0551-z) contains supplementary material, which is available to authorized users.


Background
Diffuse large B-cell lymphoma (DLBCL) represents the most common neoplastic disorder of B-lymphocytes. Although addition of anti-CD20 antibody Rituximab to conventional chemotherapy has significantly improved the clinical outcome of the patients, due to disease heterogeneity, relapse and refractory to immunochemotherapy remains of great concern [1]. As a newly developed biotherapeutic agent, ABT-199 is a specific Bcl-2 inhibitor targeting the BH3-binding domains of Bcl-2 and has demonstrated encouraging preclinical results in treating B-cell lymphoma [2,3]. However, the underlying mechanism of ABT-199 in treating DLBCL warrants further investigation.
Besides genetic abnormalities of malignant cells themselves, the aberrant status of tumor microenvironment plays a pivotal role on disease progression [4]. As the main components of microenvironment, tumor vessels prevent the attack of chemotherapy on tumor cells [5]. Moreover, tumor angiogenesis mediated by immune cells is involved in the development of drug resistance in lymphoma [6,7]. Direct interaction of immunosuppressive regulatory T (Treg) cells with vascular endothelial cells contributes to the regulation of anti-lymphoma responses [8,9]. Inducible co-stimulator (ICOS) and its ligand ICOSL function as an important crosstalk between immune cells and endothelial cells [10]. Since it has recently been reported that the cytotoxic effect of ABT-199 on tumor cells rely on survival signals from tumor microenvironment [11], microenvironmentrelated biomarkers may become potential predictors of ABT-199 efficacy on B-cell lymphoma.
MicroRNAs (miRs), a class of 19-to 23-nucleotide non-coding RNA molecules, regulate gene expression by targeting mRNA at the 3'-untranslated region [12]. In addition to their oncogenic action on tumor cells mediated by oncogenes and/or tumor suppressor genes, miRs are emerging as key regulators of tumor microenvironment [13,14]. Among them, miR21 can target Bcl-2 expression by binding to Bcl-2 3'-untranslated region (610-617 bp) and stably expressed in peripheral blood and closely related to disease outcome in B-cell lymphoma as previously reported [15][16][17]. In the present study, we assessed serum miR21 expression in a large cohort of DLBCL patients and revealed the biological function of miR21 on tumor microenvironment both in vitro and in vivo. Although potentially oncogenic, miR21 enhanced the sensitivity of B-lymphoma cells to ABT-199, through an alternative mechanism involving tumor angiogenesis.

Patients
A total of 203 patients with newly diagnosed DLBCL were enrolled in this study. The main clinical characteristics of the patients were listed in Table 1. Paraffin tumor samples of 50 DLBCL patients were used for immunohistochemistry study. One hundred healthy volunteers were referred as normal control. The study was approved by the Shanghai Rui Jin Hospital Review Board with informed consent obtained from all subjects in accordance with the Declaration of Helsinki.

Cells and reagents
Human B-lymphoma cell lines SU-DHL-4, SU-DHL-8, human umbilical vein endothelial cell (HUVEC), and murine B-lymphoma cell line A20 were obtained from American Type Culture Collection (Manassas, VA, USA). Cells were cultured in humidified atmosphere of 95% air and 5% CO 2 at 37°C. ABT-199 was purchased from Selleck-Biotool (Houston, TX, USA). Anti-Human ICOS functional grade purified antibody was from Affymetrics Ebioscience (San Diego, CA, USA).

Serum and tissue miR21 detection
Total serum miRNA was extracted using miRNeasy Serum/Plasma Kit (Qiagen, Valencia, CA, USA). MiR21 was measured by real-time quantitative RT-PCR using miScript reverse transcription kit, hsa-miR21 primer and miScript SYBR Green PCR kit (Qiagen). MiR39 was used as endogenous control and DB cells for calibration. Total tissue miRNA was extracted using Trizol agent (Invitrogen, Carlsbad, CA, USA). RNU6 was used as endogenous control and DB cells for calibration. The

Cell proliferation assay
Cell proliferation assay was performed as previously described [18].
In vitro co-culture system Transwell cell culture chambers (1 μM, Millipore Corporation, Billerica, MA, USA) were used for coculture assay. In the co-culture system, lymphoma cells were plated on the upper chamber, with immune cells and HUVEC monolayer on the lower chamber, allowing direct contact of HUVEC with immune cells. Immune cells were mononuclear cells isolated from peripheral blood of healthy volunteers using Ficoll by density gradient centrifugation.

Lentivirus packaging and transduction
To overexpress miR21 in A20 cells, purified plasmids pGMLV-miR21 or control vector were transfected into HEK-293 T cells with package vectors using lipofectamine 2000. The supernatant of HEK-293 T cell culture was then condensed to a viral concentration of approximately 3 × 10 8 transducing units/ml. The lentiviral particles were incubated with A20 cells for 8 h. The stably transduced cells were selected by green fluorescence protein.

SU-DHL-4 cells were sorted by EasySep™ Human CD20+ Cell Isolation Kit, Treg cells by EasySep™
Human CD4 + CD127lowCD25+ Regulatory T Cell Isolation Kit (STEMCELL, Vancouver, BC, Canada), and HUVEC by CD31 microbeads kit (Miltenyi Biotec. Shanghai, China). Purity of the sorted populations was greater than 98%. ICOS expression on Treg cells was assessed using anti-ICOS antibody (Abcam, Cambridge, UK) as the primary antibody and goat anti-mouse IgG H&L (Abcam) as the secondary antibody. ICOSL expression on HUVEC was assessed using anti-ICOSL antibody (Abcam) as the primary antibody and donkey anti-rabbit IgG H&L (Abcam) as the secondary antibody. The median fluorescent intensity (MFI) was measured by flow cytometry. Cell cycle and apoptosis analysis were conducted as previously described [19].

Western blot
Cells were collected and lysed in 200 μL lysis buffer (Sigma Aldrich, Shanghai, China). Protein lysates (20 μg) were electrophoresed on 10% sodium dodecylsulfate polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were blocked with 5% non-fat dried milk and incubated overnight at 4°C with appropriate primary antibody, followed by horseradish peroxidase-linked secondary antibody. The immunocomplexes were visualized using chemiluminescence phototope-horseradish peroxidase Kits. Antibodies against Bcl-2, PDCD4, STAT3, p-STAT3 were from Cell Signaling Technology (Danvers, MA, USA). β-actin was used to ensure equivalent loading of cell protein.

TUNEL assay
In situ cell apoptosis was determined with detection of fragmented DNA, using in situ cell death detection kit (Roche, Shanghai, China) according to the manufacturer's instructions.
Tube formation assay 96-well plates were coated with a thin layer of the Matrigel (BD Biosciences, CA, USA) and left to polymerize at 37°C for 0.5 h. HUVEC (2 × 10 4 cells/well) were starved overnight before being resuspended in 100 μL endothelial cell medium and added to the polymerized Matrigel. After 8 h incubation, tube formation ability was assessed by calculating total tube length in three random fields using Wimasis Image Analysis Platform.

Transmission electron microscopy
Ultrastructural studies focused on the morphology of tumor microvessels as previously described [19].

Statistical analysis
Difference of miR21 expression among groups was calculated using Mann-Whitney U test. In vitro experimental results were expressed as mean ± S.D. of data obtained from three separate experiments and determined by t-test to compare variance. Statistical significance was defined as p < 0.05.

Serum miR21 was elevated in B-cell lymphoma and indicated lymphoma progression
Comparing with healthy volunteers, serum miR21 was significantly increased in patients with DLBCL (P = 0.009, Fig. 1 Serum miR21 was elevated in B-cell lymphoma and indicated lymphoma progression. a Serum miR21 was significantly higher in DLBCL patients than in health volunteers. The relative expression level of each patient was calculated based on the lowest expression value. b A significant correlation was observed between serum and tumor miR21 expression level. Correlation coefficient was determined by Pearson correlation coefficient analysis. c Patients with high serum miR21 expression displayed significantly increased tumor CD31 positivity. The patients with miR21 expression level over and equal to the median value were regarded as high miR21 expression, while those below to the median value were included into the low miR21 expression   Fig. 1a). A significant correlation between serum and tumor miR21 expression was observed by Pearson correlation coefficient analysis (r = 0.675, Fig. 1b). Elevated miR21 levels were associated with advanced Ann Arbor stage and International Prognostic Index indicating intermediate-high and high-risk (P = 0.027 and P = 0.013, Table 1). The median expression of miR21 was 0.318 in DLBCL. The patients with miR21 expression level over and equal to the median value were regarded as high miR21 expression, while those below the median value were included into low miR21 expression. As revealed by immunohistochemistry in tumor samples of DLBCL (25 each with high or low miR21 expression), CD31-positive microvessels were more frequently observed in high miR21 group than in low miR21 group (P = 0.002, Fig. 1c). These data suggested that serum miR21 was related to tumor progression and tumor angiogenesis in B-cell lymphoma.
In the monoculture condition, compared with doxorubicin and cisplatin, the IC50 of ABT-199 in SU-DHL-8 was unachievable, confirming that the cytotoxic effect of ABT-199 relied on Bcl-2 expression in tumor cells. Accordingly, cell cycle arrest and cell apoptosis were observed in SU-DHL-4 cells, but not in SU-DHL-8 cells (Additional file 1: Figure S1A and B). To determine the biological function of miR21, SU-DHL-4 and SU-DHL-8 cells were transfected with miR21 mimics. Western blot showed that Bcl-2 protein level was increased in SU-DHL-4, while that of SU-DHL-8 was not inducible as previously reported [21] (Fig. 2b). While mimicking lymphoma microenvironment, cell proliferation assays were further performed in the co-culture system of lymphoma cells with immune cells and HUVEC cells. Different from the monoculture condition, miR21 overexpression resulted in lymphoma cell resistance to chemotherapeutic agents, but its sensitivity to ABT-199, is found not only in the co-culture system of Bcl-2 postive SU-DHL-4 cells, but also in the co-culture system of Bcl-2 negative SU-DHL-8 cells (Fig. 2c). Therefore, Fig. 4 ABT-199 counteracted miR21-mediated angiogenesis. a HUVEC cells were sorted from the co-culture system for further study, HUVEC growth was enhanced by miR21 overexpression and reduced by ABT-199 treatment in the co-culture systems of SU-DHL-4 and SU-DHL-8 cells. b-c ABT-199-mediated inhibition of HUVEC growth was associated with decreased tube formation (b) and increased HUVEC apoptosis, as revealed by TUNEL assay (c) irrespective of lymphoma cell Bcl-2 status, miR21 sensitized B-lymphoma cells to  in the presence of tumor microenvironment.

MiR21 induced ICOS expression on Treg cells through p-STAT3 upregulation
To clarify the underlying mechanism behind miR21mediated sensitization of ABT-199 on B-cell lymphoma, Treg cells were sorted from the co-culture system and the effect of tumor miR21 on Treg cells was studied. Ectopic expression of miR21 of lymphoma cells significantly increased ICOS expression on Treg cells in both co-culture systems of SU-DHL-4 and SU-DHL-8 cells (P = 0.036 and P = 0.015, Fig. 3a). Previous study reported that miR21 regulates tumor progression through the miR21-PDCD4-STAT3 pathway [22]. As shown by luciferase reporter assay, miR21 repressed the transcriptional activity of the PDCD4 promoter (228-249 bp) in HEK-293 T cells (Fig. 3b). In Treg cells, PDCD4 was downregulated and phosphorylation of STAT3 was upregulated (Fig. 3c). Meanwhile, pharmacological inhibition of STAT3 by the STAT3 inhibitor abrogated the increased expression of ICOS on Treg cells induced by miR21 overexpression (Fig. 3d). Thus, miR21-mediated p-STAT3 phosphorylation was necessary for the induction of ICOS expression on Treg cells, independent of Bcl-2 expression of lymphoma cells.
Of note, Bcl-2 expression was increased on HUVEC co-cultured with miR21-overexpressing SU-DHL-4 cells or SU-DHL-8 cells, which was downregulated upon ABT-199 treatment (Fig. 5a). Bcl-2 was further knocked down by siRNA in HUVEC (Fig. 5b). As detected by tube formation assay and TUNEL assay, the effects of ABT-199 on HUVEC was abrogated by molecular silencing of Bcl-2 ( Fig. 5c and d). Meanwhile, Bcl-2 expression remained constant when ICOS antibody was added to block the interaction between Treg cells and HUVEC (Fig. 5e). Similar results were obtained on tube formation and HUVEC apoptosis ( Fig. 5f and g).
Together, miR21 increased interaction of Treg cells with endothelial cells via ICOS/ICOSL axis and stimulated tumor angiogenesis, which was interrupted by ABT-199 through targeting Bcl-2 expression on endothelial cells.

ABT-199 exhibited in vivo activity on miR21-overexpressing lymphoma
Murine xenograft model was established with subcutaneous injection of A20 cells either stably transfected with pGMLV-miR21 or control vector pGMLV-ct. The tumor size of pGMLV-miR21 group was significantly larger than that of pGMLV-ct group (P = 0.043 at Day 6 and P = 0.005 at Day 7, Fig. 6a). ABT-199 treatment particularly exhibited anti-tumor activity on pGMLV-miR21 tumors, as compared to pGMLV-ct tumors (P = 0.045 at Day 6 and P = 0.031 at Day 7, Fig. 6a). Consistent with in vitro study, p-STAT3-positive Treg cells, as well as ICOS/ICOSL-mediated interaction of Treg cells with endothelial cells, were increased in untreated pGMLV-miR21 tumors (Fig. 6b and c). Instead, Bcl-2-positive microvessels were reduced in pGMLV-miR21 tumors upon ABT-199 treatment (Fig. 6d), in parallel with the induction of apoptotic bodies in endothelial cells, as revealed by ultrastructural study (Fig. 6e).

Discussion
MiR21 is a key regulator of disease progression in B-cell lymphoma [23,24]. Experimentally, the overexpression of miR21 leads to a pre-B malignant lymphoid-like phenotype [25]. In clinical settings, increased circulating miR21 level in sera from DLBCL patients is associated with matched tumor tissue, advanced disease stage and inferior overall survival [26,27]. More recently, gene ontology and pathway analysis has suggested that cell- Fig. 6 ABT-199 exhibited in vivo activity on miR21-overexpressing lymphoma. a MiR21-overexpressing pGMLV-miR21 tumors grew more quickly than the control pGMLV-ct tumors and were more sensitive to ABT-199 treatment. b-c p-STAT3-positive Treg cells (b) and ICOS/ICOSL interaction (c) were decreased in untreated pGMLV-miR21 tumors. d-e Bcl-2 expression (d) and cell apoptosis revealed by ultrastructural analysis (e) were increased in microvessels of ABT-199-treated pGMLV-miR21 tumors. Microvessels were indicated by red broken lines and apoptotic bodies by red arrow environment interaction is another important miR21 pathogenic mechanism [28]. Here we not only confirmed miR21 as a serum oncogenic biomarker of DLBCL, but also provided a direct link of miR21 with lymphoma progression and tumor angiogenesis.
Dysregulated tumor microenvironment determines cancer cell chemosensitivity [29]. Our study showed that ectopic expression of miR21 led to chemoresistance of Blymphoma cells, more profoundly when lymphoma cells were co-cultured with immune cells and endothelial cells, main components of tumor microenvironment. This is in accordance with previous studies which showed miR21 provokes myeloma cell adhesion to bone marrow stromal cells and resistance to chemotherapeutic agents [30], in addition, miR21 initiates inflammatory signaling in HER2positive breast cancer and reduces the cytotoxic effect of neoadjuvant trastuzumab and chemotherapy [31].
Crosstalk between Treg cells and endothelial cells is essential in drug resistance [6]. ICOS, being a member of the CD28 family of co-stimulatory molecules, is implicated in maintaining durable immune reactions upon binding to ICOSL [32,33]. Particularly, ICOS/ICOSL axis plays a pivotal role in Treg cell function and promotes Treg differentiation through activating the phosphoinositide 3-kinase-Akt pathway [34,35]. MiR21 can modulate ICOS-ICOSL expression and contribute to the progression of colorectal cancer [36]. To our knowledge, we provided the first evidence that miR21 enhanced the interaction of Treg cells with endothelial cells, induced ICOS expression on Treg cells, stimulated tumor angiogenesis via ICOS/ICOSL signaling, and led to chemoresistance of B-cell lymphoma. Recent study showed that follicular lymphoma cells generate Treg cells via ICOS/ICOSL cascade and are susceptible to anti-ICOS/ICOSL therapy [37]. Therefore, our results highlighted a key role for Treg cells in DLBCL progression and suggested that targeting ICOS/ICOSL pathway may be an alternative immunotherapy for DLBCL treatment.
ABT-199 can specially target Bcl-2 on tumor cells and induces B-lymphoma cell apoptosis. However, it is unclear whether ABT-199 would affect tumor microenvironment in a Bcl-2-dependent manner. Here, both in vitro and in vivo, our study characterized a Bcl-2-dependent inhibition of ABT-199 on tumor angiogenesis, mediated by intrinsic miR21-Treg cell pathway, leading to sensitizing effect of ABT-199 on B-cell lymphoma. Therefore, in addition to the oncogenic role of miR21 on malignant B-lymphocytes [38], miR21 possessed a potential activity on tumor microenvironment, indicative an alternative mechanism of Blymphoma cell response to  The above data demonstrated that miR21 plays an oncogenic role in B-cell lymphoma by modulating tumor microenvironment and supported clinical rationale for using miR21 as a biomarker to select chemoresistant B-lymphoma patients who may benefit from ABT-199 treatment.