$${\sum }_{i = 1}^{n}\left[\text{c}\text{o}\text{e}\text{f}\left(\text{m}\text{R}\text{N}\text{A}i\right)\text{*} \text{E}\text{x}\text{p}\text{r}\text{e}\text{s}\text{s}\text{i}\text{o}\text{n}\left(\text{m}\text{R}\text{N}\text{A}i\right)\right]$$
Median value of the risk score set as the cut-off value in each cohort, and for these patients with lower risk scores than the median value were assigned into the low-risk subgroup, while others belonged to the high-risk subgroup. The risk score of patients in the other four external validation cohorts was also calculated by this risk formula and then dichotomized these patients into two different risk subgroups by the median risk score in each cohort.
To validate the association between SRD5A2 and ITGA11 and the clinicopathological features, we used the IHC assay to detect the protein expression of the above two genes in prostate cancer tissue array (Outdo Biotech Co., Ltd., Shanghai, China), which contains the tumor tissue from 42 patients. The detailed steps of the IHC procedure have been previously reported 13,14. The antibodies of SRD5A2 (Cat. #: DF8416, Affinity Biosciences LTD., Ohio, USA) and ITGA11 (Cat. #: bs-13771R, Bioss Antibodies LTD., Massachusetts, USA) were applied for the IHC staining with a dilution of 1:250. We recorded the staining intensity as: 0, negative; 1, weak positive; 2, moderate positive; and 3, strong positive. Besides, the staining area was indicated as: 0, 0%; 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, > 76%. The intensity score multiplied with staining area were defined as the ultimate scores (>= 3, positive staining; < 3, negative staining) 15.
2.6 | Cell culture and knockdown of SRD5A2 and ITGA11
We cultured the C4-2 and PC-3 cell lines with RPMI 1640 medium, which also added 10% fetal bovine serum and 1% penicillin and streptomycin, cells cultured in the condition of 37°C and 5% CO2. We obtained the 1 × 108 TU/ml shSRD5A2 and shITGA11 lentivirus from Shanghai Novobio Co, Ltd (Shanghai, China). To obtain the lentivirus, the shITGA11-1#, shITGA11-2#, shSRD5A2-1#, shSRD5A2-2# were inserted into the PDS126_pL-U6-shRNA-GFP vector. The knockdown sequences showed as follow: shITGA11-1#-F: GCTCTTACTTTGGGAGTGAAA, shITGA11-1#-R: TTTCACTCCCAAAGTAAGAGC; shITGA11-2#-F: GCCATCCAAGATCAACATCTT, shITGA11-2#-R: AAGATGTTGATCTTGGATGGC; shSRD5A2-1#-F: GTGGTGTCTGCTTAGAGTTTA, shSRD5A2-1#-R: TAAACTCTAAGCAGACACCAC; shSRD5A2-2#-F: CTCAATCGAGGGAGGCCTTAT, shSRD5A2-2#-R: ATAAGGCCTCCCTCGATTGAG.
2.7 | Assay of cell proliferation, migration, and invasion
To evaluate the impact of SRD5A2 and ITGA11 on prostate cancer cells, we employed MTT assay and colony formation assay to assess the alteration of cell proliferation, while Transwell-based invasion and migration assay was used to evaluate the ability of cell migration and invasion.
For MTT assay, 5000 cells seeded in per well of 24-well plates, and then collected the results by add 50 μL prepped 5 mg/mL MTT reagent to 450 μL refreshed medium (with a concentration of 0.5 mg/mL) and incubated at 37°C for 1.5 hours. On the sixth day, add the DMSO solution to all plates (0, 2 days, 4 days, and 6 days) to dissolve the formazan crystals and then read the optical density value at 570 nm to display the cell viability. For colony formation, 800 cells were seeded per well and grew for 12 days. The cells in plates will be fixed by 4% paraformaldehyde for 20 minutes and then the 0.05% crystal violet was used to stain these fixed cells for another 20 minutes.
For migration assay, Transwell Permeable Supports (Corning Inc., Maine, USA) was used. A total of 1 × 105 cells was seeded in the upper chamber for each well, and 500 μL fresh medium with 10% FBS added into the lower chamber, and then incubated at 37°C and 5% CO2 incubator for 24 hours. The cells migrated to the bottom of the membranes were permeabilized by methanol and stained with 0.01% Crystal Violet. The steps of invasion assay are similar to migration assay, which also used Permeable Supports, but with extra Matrigel (Biocoat, Corning, New York, USA) diluted and coated at the upper chambers and incubated for 36 hours. The cell numbers were calculated by counting three random fields.
2.8 | Functional prediction
Increasing evidence indicated that highly co-expressed genes potentially have similar biological functions 16-18, and we used Pearson correlation analysis to find out the highly co-expressed genes of ITGA11 on the five cohorts, respectively. After overlapping these co-expressed genes, we performed the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis to subclassify their functions based on the “clusterProfiler” R package 19. Besides, we also used Cytoscape (v3.5.1, San Diego, La Jolla, California, USA) to visualize the functional network. Two external gene sets, HALLMARK PI3K/AKT SIGNALING and REACTOME PI3K/AKT ACTIVATION, were employed to assess the association between ITGA11 expression and the activation PI3K signaling pathway. The single sample gene set enrichment analysis (ssGSEA) 20,21, implemented in the GSVA R package, was applied to calculate the normalized enrichment score (NES) of the above 2 gene sets.
2.9 | Western blot validation
Proteins (40-50 μg) were separated on 12.5% SDS/PAGE gels then transferred onto nitrocellulose blotting membranes (GE Healthcare Life Science, Germany). Membranes were blocked by 5% Bovine Serum Albumin (Sigma-Aldrich, St. Louis, MO, USA) for 1 hour at room temperature and then incubated with appropriate dilutions of specific primary antibodies against SRD5A2 (Cat. #: DF8416, Affinity Biosciences LTD., Ohio, USA), ITGA11 (Cat. #: bs-13771R, Bioss Antibodies LTD., Massachusetts, USA), AKT1/2/3 (Cat. #: AF6216, Affinity Biosciences LTD., Ohio, USA), p-AKT1/2/3 (Ser473) (Cat. #: AF0016, Affinity Biosciences LTD., Ohio, USA), GAPDH (Cat. #: 1049-1-AP, Proteintech Group, Illinois, USA) overnight at 4°C. Next day, after incubating with HRP-conjugated secondary antibodies for one hour, the membranes were visualized using the ECL system (Pierce; Thermo Fisher Scientific, Inc., USA).
2.10 | Statistics
All the statis Comparisons of continuous data between two subtypes were performed by Student’s t-test and Mann-Whitney U test for normal and non-normal distribution data. Correlations between staining intensity subgroups and clinicopathological subgroups were evaluated by Fisher's exact test. Spearman's correlation analysis was utilized to explore the correlation between continuous variables. For all statistical analyses, a two-tailed P value less than 0.05 was considered statistically significant.