COLGALT2 is overexpressed in ovarian cancer and interacts with PLOD3

To the Editor: We here report collagen beta(1-O) galactosyltransferase 2 (COLGALT2) is overexpressed in ovarian cancer (OvCa) and interacts with procollagen-lysine,2-oxoglutarate 5dioxygenase 3 (PLOD3). OvCa is a major threat to women’s health. We exploited The Cancer Genome Atlas dataset and identified focal gain of 1q25.3 that occurred in over 50% of OvCa cases (Figure 1A). 1q25.3 Harbored protumorigenic genes in breast cancer1 on which we identified COLGALT2 whose mRNAexpressionwas higher in copy number gained cases (Figure 1B). We then performed immunohistochemistry (IHC) in a tissue microarray (TMA) containing 80 samples of primary ovarian cancer, 10 normal ovary tissues, and 10 metastatic lesions (Table S1). We found significantly higher COLGALT2 expression in both high-grade (HG) and low-grade (LG) serous ovarian cancer (SOC), the most common subtype of OvCa, compared with in healthy ovary tissue (Figure 2A). Whereas COLGALT2 expression in primary tumors did not differ by stage, metastasis, nodal status (Figure 1C), its expression was significantly higher in metastatic lesion (Figure 2B). We next measured constitutive expression of COLGALT2 in four OvCa cell lines and the ones with higher expression (SKOV3 and HEY-T30) and the one with both COLGALT2 amplification and overexpression (CAOV4) were chosen for further study (Figure 2C). We first identified satisfactory knockdown (KD) of the gene in all three cell lines (Figure 2D). Endogenous COLGALT2 was difficult to be captured directly on Western blotting and a FLAG tag was used (Figure 1D). COLGALT2-KD significantly decreased proliferation (Figure 2E) and colony formation in all three cell lines (Figures 1E and 2F). COLGALT2-KD induced significantly less cell population in G1 and G2 phases (Figure 1F). COLGALT2-KD significantly increased cell apoptosis with significant increase of caspase 3/7 activity (Figure 2G,H). COLGALT2 harbors procollagen galactosyltransferase activity and partic-


Proliferation assay
A dynamic proliferation monitoring system Celigo was used. After viral infection, approximately 2500 cells/well were seeded in 96-well plate. From 24 h to 120 h, cells were examined per day. Readings were normalized to control.

Flow cytometry
The FASCanto flow cytometry system was used to measure cell cycle and apoptotic profiles. For cell cycle analysis, cells were fixed using cold ethanol and later treated with cell cycle staining buffer. For apoptosis, cells were applied with Annexin V and apoptotic cells were defined as sum of early and late apoptotic cells.

Caspase assay
Established protocol was followed as per Promega Caspase-Glo kit. Cells were seeded in 96-well plate and cultured for 3 days. Cells were resuspended at 1×10 4 cells/ well and 100 μl of pre-mixed Caspase-Glo reaction fluid was added. After gentle shaking, cells were subject to a plate reader.

Colony formation
72 h after viral infection, approximately 400-1000 cells were seeded in each well of a 6-well plate. Medium was changed every 3 days. Cells were fixed with 4% methanol on day 11 and subsequently stained by crystal violet.

Co-immunoprecipitation and proteomics
SK-OV-3 cells with stable COALGALT2-OE or control were prepared and examine for protein level by western blotting of Flag (Sigma, F1804, mouse, at 1:1000). Cells were rinsed with PBS twice and lysed pre-chilled. Cells were fragmented by ultrasound and protein concentration was determined by BCA method. Load EP tube with Flag beads and add protein lystes to a total of 1200 μl/tube. After incubation overnight at 4 °C, samples were centrifuged. Candidate genes were pre-selected by shotgun proteomics using high performance liquid chromatography combined with mass spectrometry (MS) using Q Executive for differentially translated proteins of interest. We designated unique peptide of 1 or above as credible proteins. Genes of interest were subject to western blotting in the IP assay and western blotting was performed (antibodies listed in Suppl. Table 2).

Pulse-chase analysis
The pulse-chase analysis was used to profile the impact of COLGALT2 on intraand extra-cellular collagen. An established protocol was followed [5]. Briefly, cells were incubated overnight and room temperature and medium was supplemented with 50µg/ml ascorbate and 50 µg/ml catalase for another overnight culture. After one wash, cells were pulsed with 1 ml of medium containing 50 µg/ml ascorbate, 20µCi/ml L -[ 14 C(U)]-proline for 4 h. Chase started with medium change to complete medium containing 50 µg/ml ascorbate and 30 µg/ml L-proline. Collagens were digested in medium using 25 µg/ml pepsin in 1 mol HCL for 2 at 4 °C. Precipitation of collagen was finalized using 50% ethanol and resuspension was performed in Laemmli buffer. Separated collagen on SDS-PAGE gel was transferred to nitrocellulose membrane and a densitometry was used to quantify intensity of bands.

Transwell assays
Both invasion and migration were measured by Transwell assay. Cells were seeded in the upper chamber of the Transwell plate at the density of 1×10 6 /ml, either coated (for invasion) and uncoated (for migration) with Matrigel. Upper chamber was supplemented with serum-free media whilst the lower chamber was filled with complete medium. Cells that penetrated were stained with crystal violet and counted for number.

Xenograft mouse model
Subcutaneous tumor implantation mouse model was used to profile therapeutic effect in vivo. Approximately 10 7 SK-OV-3 cells with shCOLGALT2 or control were injected s.c. at axillary region of 10 female mice at 4 weeks of age per group. Tumors were calibrated every 4 days (checkpoint) and mice were euthanized on the fifth checkpoint unless tumors reached 2000mm 3 of size calibrated using the formula Length * Width 2 * 0.523. The overexpression modeling was performed using a similar method. All tumors harvested were paraffinized and processed with IHC for COLGALT2 and PLOD3 with methodology similar to TMA.

Statistical analysis
Statistical analysis for in silico studies were automatically performed with the platforms used, as aforementioned. Statistical analysis for in vitro assays and in vivo experiments were performed using the Prism Graphpad 7.0 for Mac. All assays were performed in triplicates with a sample size of 6. Comparisons between two groups were studied using the Mann-Whitney test for non-parametric variants (IHC scores) and using the Student's t test for parametric variants (RNA-seq reads) and using the Mann-Whitney test for non-parametric variances. The survival data was presented using the Kaplan-Meier curve and compared using the Log-rank test. The P value of < .05 was accepted as significant.