Deregulation of lysophosphatidic acid metabolism in oral cancer promotes cell migration via the up-regulation of COX-2

Materials

18:1 LPA was purchased from Avanti Polar Lipids (Alabaster, AL, USA) and was dissolved in ethanol:water (1:1, v/v) to create a 10 mM stock solution. Stock solutions were stored at −20 °C. NS 398 and Ki 16425 was purchased from Tocris Bioscience (Bristol, UK) and R&D Systems (Minneapolis, USA), respectively.

Cell lines

Details of the OSCC cell lines have been described previously (Edington et al., 1995; Munro et al., 2001; Prime et al., 1990). Cells were obtained from The European Collection of Authenticated Cell Cultures (ECACC) and were maintained in DMEM/F12 supplemented with 10% (v/v) foetal bovine serum. All cell culture reagents and media were obtained from Gibco (Thermo Fisher Scientific Inc, MA, USA).

Analysis of ENPP2 and PPAP2A-C expression in The Cancer Genome Atlas HNSCC data set

HNSCC data from The Cancer Genome Atlas (TCGA) were obtained from the TCGA data portal (The Cancer Genome Atlas Network, 2015). Level 3 RNA-sequencing and clinical data were downloaded in the “Biotab” format”. OSCC samples were identified using the “anatomic neoplasm subdivision” field and classified as as either human papillomavirus (HPV)-negative or HPV-positive. In total RNA-seq data were available for 265 HPV-negative OSCC samples, and for 26 of these data were also available for matched “normal” samples.

The edgeR package (Robinson, McCarthy & Smyth, 2010) in R (R Core Team, 2018) was used to normalize read counts between samples and to convert reads for each gene to counts-per-million (cpm). Differential expression analysis was also performed using edgeR and survival analyses were performed using the “survival” package in R.

Scatter assays

Cell scattering in response to LPA was measured exactly as described previously (Patmanathan et al., 2016). 5 µM epidermal growth factor (EGF) was used as a positive control.

Transwell migration and invasion assays

For cell migration, polycarbonate filters (8 µM pore size; Transwell, Corning, USA) were coated with fibronectin (10 µg/mL) in 24-well plates (Costar, Corning, USA) for two hours at 37 °C. Cells were cultured in reduced serum (DMEM/F12/1% FBS) overnight and treated with 10 µg/mL mitomycin C for two hours to inhibit cell proliferation. 5 × 10⁵ cells in media with reduced serum containing 0.25 mg/mL fatty acid-free human albumin (FAFA) were seeded in the upper chamber. LPA and inhibitors were prepared in the same migration buffer and placed in the lower chamber. Cells were allowed to migrate for 18 h. Transwell invasion assays were carried out using Corning Biocoat Matrigel Invasion chambers (BD Biosciences, CA, USA) and the cells allowed to invade for 48 h. Non-migrated or non-invaded cells were removed using cotton buds and migrated/invaded cells were rinsed with phosphate buffered saline (PBS) and stained with 0.1% crystal violet (in 20% methanol) for 2.5 h and counted in five random fields for each insert under 20x magnification.

Organotypic invasion assays

Organotypic cultures with an air–tissue interface were prepared, as described previously (Nystrom et al., 2005; Yap et al., 2009). Briefly, gels comprising a 50:50 mixture of Matrigel (Becton-Dickinson, Oxford, UK) and type I collagen (Upstate) containing 1% (v/v) FBS, 1% (v/v) Hank’s Balanced Salt Solution (HBSS) and 1 × 10⁶/ml ICRF 23 fibroblasts were prepared in cell culture inserts (pore size 3 µm) placed in six-well biocoat plates (Becton Dickinson). After 1 h, sterile glass rings were placed on the gels, and 45 min later DMEM, containing 20% (v/v) foetal clone III (Hyclone), l-glutamine 0.3 µg/ml (FCM) was added to the wells to the level of the insert bases and inside the glass rings. After a further 20 h, the FCM in the glass rings was replaced by 1 ml FAD medium [DMEM: Hams F12 (3:1), 5% FBS(v/v), 5 µg/ml insulin, 10 ng/ml hEGF, 24 ng/ml adenine, 0.4 µg/ml hydrocortisone, 100 U/ml penicillin, 10 µg/ml streptomycin, 250 ng/ml amphotericin B] containing 5 × 10⁵ H357 cells. After a further 24 h, the rings and the media were removed and the media in the wells replaced with rFAD [DMEM:Hams F12 (3:1), 10% FBS, 50 µg/ml l-ascorbic acid, 100 U/ml penicillin, 10 µg/ml streptomycin, 250 ng/ml amphotericin B] supplemented with 5 µM LPA. Media were changed every 48 h, supplemented with 5µM LPA for experimental wells and ethanol:water (1:1 v/v) for controls. After 10 days, the gels were fixed in formal-saline, embedded in a solution containing 1% (v/v) agarose, 10% (v/v) formol saline and processed to paraffin. Sections (2 µm) were stained with H&E.

Radiation treatment and clonogenic cell survival assays

OSCC cells were seeded at a density of 200,000 cells per dish in 60-mm dishes and cultured overnight before pretreatment with 10 µM LPA for one hour prior to irradiation. In experiments using inhibitors, cells were pretreated with 10 µM LPA for one hour prior to the addition of 10 µM NS-398 for a further one hour before irradiation. Media were changed prior to irradiating the cells with 0-8 gray (Gy) gamma using the Gammacell 3000 (Best Theratronics, Ottawa, Canada). Irradiated cells were harvested immediately by trypsinisation. Cells were counted and 300-500 cells were plated in triplicate for clonogenic survival assays. The cells were incubated for 12 days with media changed every four days and colonies stained with 0.1% (w/v) crystal violet. Colonies with >50 cells were counted as one colony. The plating efficiency (PE) and survival fraction (SF) were calculated using the following equations: ${\displaystyle \mathrm{PE}=\frac{\text{number of colonies formed}}{\text{number of cells seeded}}\times 100\text{\%}}$ ${\displaystyle \mathrm{SF}=\frac{\text{PE after irradiation}}{\text{PE of control}}\times 100\text{\%}}$

Quantitative real-time polymerase chain reaction (RT-qPCR)

Total RNA extraction and RT-qPCR were carried out as described previously (Lai et al., 2019) The TaqMan^® Gene Expression Assays used were PTGS2(Hs00153133_m1) and GAPDH (4326317E) (Applied Biosystems, California, USA). The using the comparative threshold cycle method (ΔΔCt) was used to measure fold changes in gene expression.

Western blotting

Western blotting was performed as described previously (Lai et al., 2019) The primary antibodies used in this study were anti-COX-2 (SC-1745; 1:500; Santa Cruz Biotechnology, Dallas, USA). and anti-ß-actin (A5441; 1:5000; Sigma Aldrich, St. Loius, USA). Bound antibodies were detected with peroxidase conjugated secondary antibodies and enhanced chemiluminescence reagents (Advansta, Menlo Park, CA, USA). ImageJ was used to measure signal intensities.

Statistical analyses

All statistical analyses were performed using GraphPad Prism Version 5.01 and SPSS Version 20. Student’s T test, One-way ANOVA with post hoc Dunnett’s test or post hoc Tukey’s test were performed for all experiments.

Expression of ENPP2 in normal oral and OSCC tissues

To investigate ENPP2 and LPP expression in OSCC tissues, we used expression data from The Cancer Genome Atlas (TCGA). The vast majority of OSCCs are not associated with HPV, so we confined our analyses to HPV-negative tumours only, for which data was available for 265 cases. There was no difference in the expression of ENPP2 in OSCCs relative to normal samples, although it was evident from the box-plots that a small number of cases expressed very high levels of ENPP2 (Figs. 1A and 1B). By contrast, both LPP1 (PPAP2A) and LPP3 (PPAP2B) were significantly down-regulated in OSCC tissues, but the levels of LPP2 (PPAP2C) were unaltered (Figs. 1C– 1H). We also used the TCGA data to test whether there were any potential associations between expression levels and overall survival in the HPV-negative OSCC cases. For ENPP2 and PPAP2A we observed no such association, but for PPAP2B there was some evidence that lower expression may be associated with reduced overall survival, although this was not quite statistically significant. A Kaplan–Meier plot with cases split by median PPAP2B expression (log-rank p = 0.069) is shown in Fig. 1I. We also performed a Cox proportional hazards analysis using PPAP2B expression as the explanatory variable, and this produced a consistent result (Wald p = 0.059).

Taken together, these data indicate that a subset of OSCCs expressed elevated levels of ENPP2 compared to normal oral mucosa, which would lead to the increased production of LPA; the low levels of LPP1 and LPP3 in tumors would augment the effects of LPA (Tang & Brindley, 2020).

LPA promotes the migration and invasion of OSCC cells

As we showed that LPA metabolism was deregulated in OSCCs, we next examined the phenotypic effects of LPA on OSCC cell behaviour in vitro. Scatter assays were performed to assess the effect of LPA on the motility of two OSCC cell lines, H357 and H413. Without treatment, OSCC cells grew to form tightly clustered colonies (Fig. 2A). LPA induced the dispersal of single cells away from the pre-existing colony (Fig. 2A). Quantification of these effects revealed a highly significant increase in motility in response LPA (P < 0.001, Figs. 2B and 2C).

We next examined the effect of LPA on OSCC cell migration. LPA significantly enhanced the migration of H357 and H376 cells in transwell assays (Figs. 2D and 2E). H357 cells are able to invade through matrigel in vitro (Patmanathan et al., 2016), so we used this cell line to examine LPA-induced invasion and showed that LPA significantly promoted the invasion of H357 cells through matrigel in transwells (Fig. 2F). Further, in 3D organotypic cultures that are biochemically and physiologically more similar to in vivo tissues than transwells, LPA greatly stimulated the invasion of H357 cells with large clusters of cells spreading deep into the matrix (Fig. 2G).

LPA protected OSCC cells from the effects of radiation

LPA has been reported to suppress apoptosis in non-transformed intestinal crypt-derived epithelial cell lines after high dose radiotherapy (>10Gy) (Deng et al., 2002). Therefore, we investigated the protective effect of LPA on OSCC cells following exposure to radiation using clonogenic cell survival assays. Pre-treatment of cells with LPA increased the survival fractions of both H357 and BICR31 significantly following exposure to 1 Gy irradiation (Figs. 3A and 3B).

LPA mediates its effects via induction of COX-2 expression

COX-2 is a known downstream target of LPA signalling and has been shown to regulate cell migration (Jeong et al., 2008) and response to radiation (Terakado et al., 2004). Therefore, we next examined whether LPA could upregulate COX-2 expression in OSCC cells and investigated whether COX-2 mediated some of the phenotypic effects of LPA. Treatment of three OSCC cell lines (H357, H376 and BICR31) with 10 µM LPA resulted in a clear increase in COX-2 mRNA (Figs. 4A– 4C) and protein (Figs. 4D– 4F). To investigate whether the induction of COX-2 expression mediated the effects of LPA on OSCC cell migration and response to radiation, we used a widely used COX-2 inhibitor, NS398, to suppress COX-2 activity. LPA-induced migration of H357 and H376 cells was significantly attenuated following COX-2 inhibition with NS398 in transwell migration assays (Figs. 4G and 4H). NS398 did not affect the survival of LPA-treated H357 and BICR 31 cells exposed to 1 Gy irradiation (Fig. S1).