Cathepsin L-induced galectin-1 may act as a proangiogenic factor in the metastasis of high-grade serous carcinoma

Non-malignant omental tissue samples were collected from patients undergoing surgery at the Royal Devon and Exeter NHS Foundation Trust (Exeter, United Kingdom) with ethical approval and informed written consent. HOMECs were isolated using collagen-digestion and anti-CD31 magnetic beads, characterised and cultured as primary cells as previously described [24]. Briefly, HOMECs were cultured in endothelial cell (EC) growth media (MV2, PromoCell, Heidelberg, Germany) supplemented with supplied growth factors, 5% (v/v) foetal calf serum (FCS) and 0.1% (v/v) gentamycin (Sigma, Poole, UK). Cells were maintained at 37 °C in a humidified atmosphere supplemented with 5% CO₂.

Phosphorylation levels of NFκB, ERK1/2 and AKT were measured using specific cell-based ELISA kits (Bio-Techne Ltd., Abingdon, UK) according to the manufacturer’s instructions. Cells were starved, and subsequently treated ± recombinant human VEGF165 (20 ng/ml, positive control; Peprotech, London, UK), CathL (50 ng/ml; from human liver; Sigma-Aldrich, Poole, UK) and recombinant human Gal1 (50 ng/ml; Sigma-Aldrich, Poole, UK) ± NFκB, ERK1/2 and AKT inhibitors at their given concentrations (Table 1) for 4, 10 min or 4 h. To confirm the effect of these inhibitors, HOMECs were pre-incubated with the inhibitors and then co-treated ± proangiogenic factors. Fluorescence intensity was measured, and the results were expressed as fold change in phospho-NFκB (p65), -ERK1/2 or -AKT relative to total NFκB, ERK or AKT levels (compared to control).

Assessment of cellular migration was carried out using a Cultrex Cell 96 transwell migration assay (Bio-Techne Ltd., Abingdon, UK) as previously described (5). Briefly, cells were incubated in growth factor-deprived media supplemented with 0.5% FCS overnight. Next, cells were seeded at a density of 5 × 10⁴ in the upper assay chamber and treated ± Gal1 (50 ng/ml) and in the presence or absence of ERK1/2 and/or AKT inhibitors at their given concentrations (Table 1). Negative controls received carrier alone. After 6 h incubation at 37 °C, the bottom chambers were washed, followed by addition of cell dissociation solution/calcein AM for a further hour to label and detach migrated cells. Fluorescence in the bottom wells was read at Ex/Em: 485/520 nm.

Total RNA from cultured HOMECs treated with CathL ± sulfasalazine was isolated with TRI reagent as per the manufacturer’s protocol (Sigma-Aldrich, Gillingham, UK). Extracted RNA was DNase-treated using a DNAfree kit (Ambion, Northumberland, UK). Complementary DNA (cDNA) was synthesised from 1 µg of RNA per sample using the qScript supermix kit (Quanta Biosciences, USA). Real-time RT-PCR was conducted using the TaqMan PrecisionPLUS master mix (Primer Design, Southampton, UK) with cycling conditions as follows: 1 cycle of 60 °C for 2 min and 95 °C for 2 min; 40 cycles of 95 °C for 10 s, 60 °C for 30 s; and 1 cooling cycle of 37 °C for 30 s, on a LightCycler 96 Real-time Detection System (Roche, Welwyn Garden City, UK). The specificity of the PCR product was confirmed by melting-curve analysis. The Gal1, and two house-keeping genes GAPDH and β-2-microtubulin primers were purchased from Thermo Fisher Scientific (Northumberland, UK). For each gene, crossing point (Cp) values were determined from the linear region of the amplification plot and normalized by subtraction of the geometric mean of the Cp values for two housekeeping genes. Relative expression was subsequently calculated using the 2-ΔCp approach, and the data were presented as fold change in LGALS1 gene expression (normalised to control).

Female patients signed an informed consent form for the study which was reviewed by the Institutional Review Board (Gomel State Medical University), Gomel, Belarus. Sixty formalin-fixed and paraffin embedded archival tissue specimens including 20 cases of each of the following: normal omenta (from organ transplant donors), omenta of serous ovarian carcinoma patients without omental metastasis (SC wo MTS) and, with metastasis (SC w MTS), were used for immunohistochemistry (IHC). This was assessed by histopathologists D.A.Z. and S.L.A using the following criteria: a presence of papillary and solid pattern of tumour moderate to marked nuclear atypia, nuclear pleomorphism, stromal invasion, more than 12 mitotic figures in 10 high power fields [25]. Clinical information was obtained from the patients’ medical records. None of the patients within the control group had tumours or infections, or previous peritoneal surgical operations and traumas. The group of HGSC (SC wo MTS) patients were selected based on the following criteria: absence of secondary tumours, absence of previous chemotherapy treatment (intraoperative diagnostic of cancer), absence of previous peritoneal surgical operations and traumas, stages II–III (FIGO, 2009). The clinic-pathologic features of the ovarian serous carcinoma cases used in the present study are shown in Table 2. In stage II, cancer cells are found in other organs within the pelvis but not the omentum, and thus omenta from stage II patients were not included in the study [26].

Data are expressed as mean ± standard deviation (SD) and analysed using Mann–Whitney U test and one-way ANOVA test. A Spearman (r) test with Chadock scale were used for correlation analysis. The univariate and multivariate hazard ratios (HR) were estimated using Cox regression analysis. A p value of less than 0.05 was considered statistically significant. For all data, n represents the number of patients, wells or dishes tested under each condition and also the results from at least two primary cell populations. Statistical analysis was performed using SPSS Statistical Software Package (SPSS Inc, Chicago, USA) and GraphPad Prism (GraphPad Software, California).

Our recent data suggests a mitogen ligand-like activity for CathL in HOMECs [11], and our unpublished, preliminary data indicates that CathL induces a differential expression of galectin1 (Gal1) mRNA (LGALS1) in these cells. Therefore, we hypothesised that CathL induces production and secretion of Gal1. Figure 1a confirms that extracellular levels of Gal1 were, indeed, significantly increased in the presence of CathL between 30 min (twofold) and 8 h (5.2-fold) but returned to control levels after 24 h treatment (Fig. 1a). The latter may reflect Gal1 degradation since the protein has a reported serum half-life of 1.07 h [30] or re-uptake by the cells [31]. Interestingly, Gal1 release was not induced in HOMECs treated with the other HGSC-secreted factors insulin-like growth factor binding protein 7 and CathD (data not shown), suggesting that Gal1 secretion may be CathL-specific.

NFκB is a transcription factor involved in the production of many physiological and pathological proteins including Gal1 [22, 33] and thus, we examined the potential involvement of NFκB in CathL-induced Gal1 secretion in HOMECs. CathL-induced secretion of Gal1 was significantly reduced to control levels by sulfasalazine (inhibitor of NFκB) after 8 h treatment (0.9-fold vs 4.8-fold (CathL-treatment), both normalised to control; Fig. 1c). These data not only suggest that CathL induces NFkB activation (confirmed in Fig. 1d), but that CathL induces transcriptional regulation of Gal1 via NFκB activation. This was further supported by the qRT-PCR data demonstrating that sulfasalazine significantly reduced LGALS1 mRNA expression in CathL-treated cells after 6 h treatment (Fig. 1e). Interestingly, sulfasalazine alone increased LGALS1 mRNA levels in HOMECs (Fig. 1e). This could be due to a compensatory effect whereby other transcription factor(s), due to an inhibition of NFκB activity, may induce LGALS1 expression, which is also reflected in the protein level. These data suggest that CathL induces Gal1 secretion in HOMECs via activation of NFκB.

We then examined whether the extracellular Gal1 released from HOMECs could have autocrine pro-angiogenic cellular effects i.e. on proliferation and migration. HOMEC proliferation was initially investigated over a range of Gal1 concentrations based on the concentrations of Gal1 (20–80 ng/ml) released into the supernatant of CathL-treated ECs over 8 h (data not shown), and the median range concentration (30–163 ng/ml) of Gal1 in the sera of HGSC stage I patients [19]. Since, we are investigating induction of angiogenesis in ECs from non-cancerous/non-metastatic omenta, we selected concentrations that fell within the lower spectrum of the in vivo range. Gal1 significantly induced cell proliferation at all concentrations tested vs control (100%, WST-1 assay) (Fig. 2a, Table 3). Additional studies with BrdU and CyQUANT proliferation assay kits confirmed that 50 ng/ml Gal1 significantly enhanced proliferation vs control (100%) (Fig. 2b, c). On the basis of these results and the levels of secreted Gal1, 50 ng/ml of Gal1 was selected for further experiments.

The pro-proliferative role of Gal1 in HOMECs suggested possible activation of intracellular signalling pathways downstream of Gal1-cell surface interaction. We hypothesised activation of the ERK1/2 and PI3K/AKT pathways based on our previous publications [8, 11] in these cells. Gal1 treatment rapidly (< 4 min) increased phosphorylation of ERK1/2 (1.8-fold vs control, Fig. 2d), with phosphorylated levels decreasing to control levels after 10 min (Fig. 2e); suggesting a transient activation. Similar results were seen for AKT, with phosphorylation transiently increased (1.7-fold vs control; Fig. 2f, g) within 4 min of Gal1 treatment. The ELISA data were verified using specific inhibitors of ERK1/2 (U0126 and PD98059) and PI3K/AKT (LY294002 and MK2206) using previously determined concentrations (Additional file 1: Figure S1) [8, 11].

Since Gal1 induced activation of the ERK1/2 and PI3K/AKT pathways, we tested the effect of inhibition of these pathways on Gal1-induced HOMEC proliferation. Both ERK1/2 inhibitors, U0126 and PD98059, significantly reduced Gal1-induced cell proliferation (p < 0.001, Fig. 3a). Gal1-induced HOMEC proliferation was also significantly reduced by the PI3K inhibitor, LY294002 (p < 0.001) but the selective AKT inhibitor MK2206 had no effect (Fig. 3b, discussed later).

A second key element of angiogenesis is EC migration and thus, the pro-migratory effect of Gal1 and the pathways involved were examined using inhibitors of ERK1/2 and AKT as above. Interestingly, although Gal1 significantly induced HOMEC migration (163.4 ± 47.2% vs control, 100%) none of the inhibitors tested significantly reduced this (Fig. 4a, b). These data combined with the ELISA data (Additional file 1: Figure S1) suggest that Gal1-induced HOMEC migration may not solely depend on the activation of the ERK1/2 and AKT pathways, and that other kinases may be involved.

The in vitro data presented above suggest that HOMEC-secreted Gal1 may play a role in pro-angiogenic changes in the omental microvasculature. To examine this further in human disease we immunohistochemically investigated expression of Gal1 protein levels in the microvascular endothelium in patient samples; specifically, in normal omenta (control), omenta with HGSC metastasis (SC w MTS) and omenta with no metastatic tumour (SC wo MTS).

Interestingly, although the microvessels stained positively for Gal1 in all groups (Fig. 5a), the intensity of Gal1 expression (designated 1–3, see supplementary methods and [19]) differed between the groups. In the control, SC wo MTS and SC w MTS groups, Gal1 mean scores were 1.32 ± 0.29, 1.42 ± 0.37 and 2.39 ± 0.28 respectively with statistically significant differences (p < 0.001) between the control and the SC w MTS group, and the SC wo MTS and SC w MTS groups (Fig. 5b).

Additionally, we investigated microvessel density (as assessed by CD34 staining of the endothelium), a measure of vascularisation i.e. angiogenesis, in each group (Fig. 5a). There was a significant increase (p < 0.001) in vessel density in the SC w MTS group compared with both the control and SC wo MTS groups (Fig. 5c), confirming a significant increase in microvessel density in the omenta invaded by tumour. No non-specific staining was observed in our negative controls of each group (Additional file 2: Figure S2).

We also measured area of vessels (per 1 mm²) and found a significant increase in microvessel area (p < 0.001) in the SC w MTS group compared with SC w/o MTS and control groups (Fig. 5d).

An association study revealed a positive correlation between Gal1 and number of vessels (Spearman coefficient, r = 0.8702, p < 0.001), and area of vessels (r = 0.7283, p < 0.001) (Fig. 5e), suggesting a potential proangiogenic role for Gal1 in the omentum of advanced ovarian carcinoma.

Univariate Cox regression analysis of prognosis factors showed a significant association between predictive patient survival and Gal1 expression (p < 0.001) and number of vessels (p = 0.032), but not area of vessels (Fig. 5f); possibly due to atypical vessels—a phenomenon seen in tumour invasion. Interestingly, in multivariate Cox regression analysis only Gal1 expression (p < 0.001) (but not area or number of vessels) demonstrated significant prognostic value in ovarian cancer patients (Fig. 5g), suggesting that Gal1 expression could be a prognostic factor in advanced ovarian cancer with a role in angiogenesis. These data all strongly support a proangiogenic role for Gal1 in HGSC metastasis to the omentum.