Hsp90β promoted endothelial cell-dependent tumor angiogenesis in hepatocellular carcinoma

HUVECs and PLC-PRF-5 cells were obtained from the Cell Bank of Shanghai Institute for Biological Sciences (Shanghai, China). HUVEC cells were cultured in medium 199 (M-199) containing endothelial cell growth supplement, 10% fetal bovine serum (FBS), and 1% antibiotics (100 U/mL penicillin and 10 μg/mL streptomycin). PLC-PRF-5 cells were cultured in Roswell Park Memorial Institute 1640 medium supplemented with 10% FBS and penicillin–streptomycin solution. All cells were maintained at 37 °C in a 5% CO₂ incubator. All the plasmids were transfected into cells by Lipofectamine 2000 (Invitrogen, USA).

HUVECs were seeded in 96-well plates. After 24 h, the plasmids, namely, pEZX–PG04–VEGFR1 and pEZX–PG04–VEGFR2 (FulenGene, China), were either transfected separately into cells, co-transfected with pcDNA3–Hsp90β or pRNAT-U6-siHsp90β, or incubated with NVP-BEP800 (Selleck, USA) for 48 h. Gaussia and secreted alkaline phosphatase (SEAP) luciferase activities were consecutively measured by a dual-luciferase reporter assay system (FulenGene, China) after 48 h of transfection. Gaussia luciferase activity was normalized to SEAP activity.

HUVECs in serum-free medium were seeded onto a chamber coated with Matrigel (BD, USA) and inserted into the wells of a 24-well plate. FBS was added to a 24-well plate located below the chamber to serve as a chemoattractant. After 24 h, invasive cells located on the lower surface of the chamber were stained with 0.1% crystal violet and counted.

HUVECs were seeded at 80% to 90% confluence in 24-well plates. An area of cells was removed with a pipette tip drawn across the center of each well. Cells were washed thrice followed by treatment with FBS-free medium containing DMSO or 2 μM NVP-BEP800. The speed of wound closure was monitored after 16, 32, and 48 h by comparing the wound distance ratio at 0 h. This experiment was independently repeated thrice.

HUVECs were seeded onto a 48-well plate coated with Matrigel (BD, USA) followed by treatment with a medium containing DMSO or 2 μM NVP-BEP800. Peak tube formation was observed at 12 h post-treatment. Four random fields were observed under a microscope. The number of tubes for each treatment was quantified. This experiment was independently repeated thrice.

Cells were washed with phosphate-buffered saline (PBS) and lysed in ice-cold lysis buffer containing protease inhibitor cocktail (Sigma) for 30 min. Lysates were separated by electrophoresis and transferred onto polyvinylidene difluoride membranes (Millipore; Bedford, MA, USA). The membranes were blocked and incubated with the primary antibodies of Hsp90β, VEGFR1, and VEGFR2 (CST, USA). The membranes were then incubated with a goat anti-rabbit IgG–HRP secondary antibody (Thermo Scientific, USA). GAPDH was used as the loading control. Protein expression was assessed with enhanced chemiluminescent substrate (Millipore, USA) and by exposure to chemiluminescent film.

HUVECs were fixed with 3.7% formaldehyde in PBS for 5 min at room temperature and blocked with 1% bovine serum albumin for 30 min. The cells were incubated overnight at 4 °C with VEGFR1 antibody (1:200; CST, USA) or VEGFR2 antibody (1:200, CST, USA), and then incubated with FITC- and TRITC-labeled secondary antibodies (Earthox LLC, USA) for 1 h at room temperature. Each step was followed by two 5-min washes in PBS. The prepared specimens were counterstained with DAPI (Southern Biotechnology Associates Inc., USA) for 2 min and observed with a confocal microscope (Nikon, Japan).

Paraffin sections of human samples and tumor tissue of mice were deparaffinized in xylene and rehydrated in a series of graded alcohols and antigen was retrieved in 0.01 M sodium citrate buffer. Samples were incubated with 3% H₂O₂ for 10 min and then blocked in 1% bovine serum albumin in PBS for 1 h. The slides were incubated overnight at 4 °C with Hsp90β, VEGFR1, VEGFR2, CD31 and Ki67 primary antibody. Using the known positive section as the positive criteria, PBS displaces primary antibody as negative criteria. The secondary antibody was followed added with using HRP-Polymer anti-Mouse/Rabbit IHC Kit (Maixin Biotech, China) for 1 h at room temperature. Then, samples were developed with diaminobezidine (Vector Laboratories) reagent and counterstained with hematoxylin and mounted with permount [21].

ImageJ software was used to quantify the MVD based on CD31 staining. The percentage of positively stained area was calculated by using a color deconvolution for separating the staining components (diaminobezidine and hematoxylin) in at least six fields per each sections [21]. The results were presented as percentage of treated group compared to control one. All sections were analyzed and evaluated independently by two pathologists (T.S. and Y.R.L.) at double blinded, and the results were reconfirmed once inconsistent.

Approximately 1 × 10⁶ PLC or PLC-Hsp90β cells were injected subcutaneously into the middorsal region of each BALB/c nude mouse. Each group (Control, Hsp90β and NVP-BEP800 group) consisted of eight mice. Tumor-bearing mice began to receive treatment at the time point when the average tumor size reached 100-200 mm³. NVP-BEP800 was formulated as a suspension in 0.5% methylcellulose and was given at the dose of 30 mg/kg per day by intragastric administration for 18 days [22]. Solvent buffer at the same volume was used in control and Hsp90β groups. Tumor sizes were measured every 2 days using a caliper and calculated according to a standard formula (length × width² × 0.52) as previously described [23].

HCC tissue microarrays containing 96 cases were purchased from US Biomax for IHC. Tissue blocks of the tissue array were collected within 5 years. Each single tissue spot on every array slide was individually examined by certified pathologists according to the WHO published standardizations for diagnosis, classification and pathological grade. Both the hospital and the individual consented for each specimen from clinical patients to be included. All tissues were collected under the highest ethical standards; the donor was completely informed and provided consent. All human tissues are collected under HIPPA-approved protocols.

Statistical analyses were performed with GraphPad Prism 6 and SPSS v. 19. Statistically significant differences were calculated by student’s t-test, Wilcoxon rank-sum test, Mann–Whitney U test, and Pearson’s correlation as appropriate. Values of p < 0.05 were considered significant.

Our previous study confirmed that Hsp90β was closely related to the metastasis and recurrence of HCC. Based on the immunohistochemical (IHC) analysis of 96 HCC cases, the expression level of Hsp90β and angiogenesis were higher in HCC with a high degree of malignancy than in HCC with a low degree of malignancy (Fig. 1a). The expression of CD31, a surface marker of neovascular endothelial cells, was detected by immunohistochemical staining of HCC tissues in different stages. Then, microvessel density (MVD) was calculated (Fig. 1b). Pearson correlation and linear regression anaysis showed that the levels of steady state expression of the mRNA of Hsp90β and CD31 was positively correlated (Additional file 1 Fig. S1a). We further analyzed the correlation between Hsp90β expression and MVD. The results showed that Hsp90β was positively correlated with MVD in HCC (P = 0.027) (Fig. 1c).

Kaplan–Meier survival analysis revealed that patients with high levels of Hsp90β expression had significantly lower overall survival (OS) compared with those with low Hsp90β expression (P = 0.000) (Fig. 2a). Cox proportional hazards model analysis was performed on the basis of clinicopathological factors, Hsp90β expression (high/low), and MVD presence (positive/negative). Multivariate analysis showed that high Hsp90β expression, MVD, advanced clinical stage, and histological differentiation were indicators of poor prognosis (P = 0.018, 0.047, 0.022, and 0.009, respectively) (Fig. 2b).

We analyzed 96 liver cancer cases and found that the expression levels of Hsp90β, VEGFR1, and VEGFR2 in liver tumor tissues were correlated with the pathological grade and clinical stage of the tumor (Table 1). Immunohistochemical staining showed that HCC patients with high Hsp90β expression levels also highly expressed VEGFR1 and VEGFR2. Additionally, VEGFR1 and VEGFR2 expression levels were higher in the endothelial cells surrounding vessels (Fig. 3a). Correlation analysis showed that Hsp90β was associated with VEGFR1 and VEGFR2 expression (P = 0.022 and P = 0.035, respectively) (Fig. 3b). The correlation studies of Hsp90β and VEGFR1, VEGFR2 mRNA levels were consistent with those of IHC evaluation (Additional file 1 Fig. S1b, c).

VEGFRs are major receptors on endothelial cells that are involved in multiple signaling pathways, including the induction of mitogenesis, migration, invasion, and differentiation in neoangiogenesis [24]. Sanderson et al. [18] reported that VEGFR2 is a client protein of Hsp90 and they can form a complex in HUVECs to promote endothelial cell migration; they also reported that VEGFR2 deletion mutants are unable to associate with Hsp90 [19, 25]. To elucidate the underlying mechanism of Hsp90β in inducing angiogenesis, we evaluated the effects of Hsp90β on VEGFRs expression. Our results showed that as with VEGF treated HUVEC cells, VEGFR1 and VEGFR2 expression levels also increased after Hsp90β overexpression (Fig. 4a). This result was validated by immunofluorescence staining (Fig. 4b). Whereas, VEGFR1 and VEGFR2 expression levels decreased after Hsp90β knocked down. When the HUVEC cells simultaneously treated with siHsp90β and VEGF, VEGFR1 and VEGFR2 expression levels increased compared with alone Hsp90β knocked down group (Fig. 4a). Furthermore, we used a dual-luciferase report system to determine the effect of Hsp90β on the promoter activities of VEGFR1 and VEGFR2. The results showed that Hsp90β and/or VEGF increased VEGFR1 and VEGFR2 promoter activities, Hsp90β knockdown decreased VEGFR1 and VEGFR2 promoter activities and VEGF released the inhibition effect of Hsp90β knocked down on VEGFRs promoter activities (Fig. 4c). In addition, Hsp90β promoted HUVEC proliferation and Hsp90β knockdown inhibited HUVEC proliferation. VEGF treatment counteracted the inhibitory effect of knocked down Hsp90β (Fig. 4d). Then, we performed Transwell chamber (with or without Matrigel on the filters) invasion and migration assays to examine the effects of Hsp90β on HUVEC invasion and migration. The number of HUVECs that migrated through the filters showed that cell invasion and migration significantly increased after Hsp90β overexpression and/or VEGF treatment. The cell invasion and migration decreased obviously when Hsp90β knocked down and VEGF released the inhibition effect of knocked down Hsp90β (Fig. 4e and f).

In vitro, endothelial cells can form a three-dimensional tube in Matrigel culture. To examine the effects of Hsp90β on HUVEC tube formation, we seeded HUVECs that were transfected with Hsp90β, Hsp90β siRNA, and/or treated with VEGF for 48 h on Matrigel. The results showed that Hsp90β significantly enhanced HUVEC tube formation, whereas Hsp90β silencing inhibited HUVEC tube formation. VEGF increased the effect of Hsp90β on angiogenesis and knockdown of Hsp90β obviously decreased VEGF-induced tube formation (Fig. 4g). Our results showed that Hsp90β affected in vitro VEGFRs promoter activities, VEGFRs expression, and HUVEC proliferation, migration, invasion, tube formation, and angiogenesis.

The ability of Hsp90 inhibitors, such as geldanamycin and its derivatives, to suppress endothelial cell proliferation has been previously described [26]. However, the effect of NVP-BEP800, a new Hsp90 inhibitor, on endothelial cells has not been reported. To determine the direct cytotoxic effects of NVP-BEP800 on endothelial cells, we conducted an in vitro cytotoxicity experiment. The different concentrations of NVP-BEP800 were cytotoxic to HUVECs when given as a treatment for 48 h (Fig. 5a). In addition, the toxicity of NVP-BEP800 in normal human liver cells (HHL-5) and various HCC cell lines (SMMC-7721, PLC-PRF-5, BEL-7402) were also measured. The results showed that NVP-BEP800 has no significant toxicity in HHL-5, and inhibited the proliferation of HCCs (Additional file 2 Fig. S2). HUVECs in log-phase growth were measured over 72 h and showed that NVP-BEP800 (2 μM) inhibited the proliferation of human endothelial cells (Fig. 5b). There was no obvious effect on death and apoptosis of HUVECs under 2 μM NVP-BEP800 treatment for 48 h (Additional file 3 Fig. S3). Western blot analysis showed that VEGFR1 and VEGFR2 expressions decreased under 48 h of 2 μM NVP-BEP800 treatment (Fig. 5c). Furthermore, we analyzed the effect of NVP-BEP800 on VEGFRs promoter activities. The results of the dual-luciferase reporter assay showed that NVP-BEP800 significantly decreased VEGFR1 and VEGFR2 promoter activities (Fig. 5d). In addition, a scratch wound assay showed that Hsp90 inhibitors prevented HUVEC haptotaxis. Control cells completed wound closure after 48 h. By contrast, cells treated with 2 μM NVP-BEP800 failed to complete wound closure within 48 h (Fig. 5e). To test the effect of NVP-BEP800 inhibition on HUVEC tube formation, HUVECs were treated with 2 μM NVP-BEP800 for 24 h and then seeded on Matrigel for 12 h. The results showed that NVP-BEP800 significantly decreased tube formation compared with control cells (Fig. 5f). We evaluated the tube formation in HUVECs that knocked down Hsp90α or Hsp90β under NVP-BEP800 treatment. The results showed that compared to Hsp90α knocked down group, the tube formation in siHsp90α+NVP-BEP800 group were significantly reduced. While there was no statistically significant difference in Hsp90β knocked down group and siHsp90β+NVP-BEP800 group (Additonal file 4 Fig. S4). It is indicated that the inhibition of endothelial cell-dependent tumor angiogenesis of NVP-BEP800 was primarily by inhibiting the activity of Hsp90β.  In addition, 17-Allylamino-17-demethoxygeldanamycin (17-AAG) were compared with NVP BEP800 on the proliferation and tube formation of HUVECs (Additonal file 5 Fig. S5). The results showed that 17-AAG inhibited cell proliferation, while the inhibition effect of NVP-BEP800 was stronger than 17-AAG on tube formation.

We demonstrated that Hsp90β promoted endothelial cell proliferation, migration, and tube formation and that the Hsp90 inhibitor attenuated VEGFR levels and inhibited tube formation in vitro. Then, we constructed a nude mice subskin model that was transplanted with human liver cancer. Control group mice subcutaneously received PLC cell suspensions. Hsp90β group mice subcutaneously received a suspension of PLC cells transfected with pcDNA-Hsp90β plasmids. NVP-BEP800 group mice subcutaneously received a suspension of PLC cells. When the tumors volume reached 100-200 mm³, NVP-BEP800 group mice were treated with NVP-BEP800 (30 mg/kg/daily) by intragastric administration for 18 days and solvent buffer at the same volume was used in control and Hsp90β groups. The results showed that compared with the control group, Hsp90β promoted the liver tumor growth, and NVP-BEP800 inhibited tumor growth in vivo (Fig. 6a, b). To further check the effect of Hsp90β and NVP-BEP800 on cell proliferation, immunohistochemical analysis of Ki67 was performed with the tumor issues (Fig. c, d). Also, we demonstrated that VGFR1 and VEGFR2 expression levels increased in cells overexpressing Hsp90β, but decreased after NVP-BEP800 treatment (Fig. 6e, f). Immunohistochemical detection revealed that CD31 expression and MVD in the Hsp90β-overexpressing group were significantly higher than those in the control group, but decreased in the presence of NVP-BEP800 (Fig. 6g, h). There was no significant difference about blood vessel tissue between control and NVP-BEP800 group by analysis mice liver and lung pathological section using hematoxylin and eosin staining (Additonal file 6 Fig. S6). Our data suggested that Hsp90β contributed to endothelial cell-dependent angiogenesis.