Fig 1.
Schematic illustration of different bioinformatics analyses and filtering methods that applied to mine large-scale gene expression profiling datasets and to select the most interesting biomarkers for subsequent clinical and functional validation.
Fig 2.
Representation of the most important data format, plots and visualization results, used in data mining and prioritization of candidate biomarkers, using the CACNA1D gene as an example.
A) Dot plot chart illustrating the association of mRNA gene expression of CACNA1D in normal and cancer samples with a panel of the most informative clinico-pathological parameters. B) Kaplan- Meier analyses, showing the correlation of high levels of mRNA expression for CACNA1D with reduced overall survival analysis for PCa patients. C) Relative expression of CACNA1D across a panel of 60+ normal and cancer-related human tissue types and diseases.
Table 1.
Characteristics of the patient cohort of 90 men with prostate carcinoma who underwent radical prostatectomy, PSA values were measured from serum preoperatively.
Table 2.
Genomic location and physiologic function of the candidate biomarkers.
Table 3.
Frequency of detection of target mRNAs in benign prostate tissue from patients without PCa (CP-B samples) and with incidental PCa (CP-IPCa samples) and in histologically benign (RP-B samples) and cancerous (RP-PCa) tissue of patients with PCa.
Table 4.
Association of mRNA expression of selected target genes in prostate tissues, and statistical correlation with major clinical and pathological parameters of PCa biopsies, calculated with the Mann-Whitney test.
Fig 3.
Differential mRNA expression levels for eight candidate biomarker genes and KLK3 (mRNA copy number/μg total RNA) in cystoprostatectomy samples (CP), histologically benign radical prostatectomy samples (RP-Be) and cancerous radical prostatectomy samples (RP-PCa).
Boxes show the interquartile range with the line in the middle of them denoting the median value and circles represent the outliers.
Fig 4.
ROC analyses for KLK3 mRNA levels and expression levels of 8 target mRNAs in cases classified as positive or negative for PCa.
To simplify the analysis, patients with two cancerous/benign samples were represented by a single value of mRNA expression of each gene. A. RP-PCa samples were considered as positive samples and compared against all CP samples (defined as negative). High sensitivity and specificity was observed for all eight biomarkers. B. RP-Be samples were considered as positive samples and compared against all CP samples (defined as negative).
Table 5.
AUCs (Area under the Curve) calculated for each of the eight biomarkers and KLK3 for comparison.
The table presents the AUC values for each gene in the ROC analyses.
Table 6.
Knock-down efficacy as confirmed by qRT-PCR after transfection of PC3 cells with corresponding siRNAs.
Table 7.
Growth inhibition of VCaP cells by siRNA knock-down in 2D culture, as measured by the CellTiterGlo assay (2000 cells/well; 384-well microtiter plate).
Table 8.
Impact of gene silencing on spheroid size, relative degree of apoptosis, and number of dead cells in VCaP spheroids.
Fig 5.
Wound healing assay with PC-3 cells following a 72-hour siRNA transfection.
The wound healing results of PC-3 cells treated with RHOU and scrambled siRNA vs. untreated PC-3 cells (mock transfection). Representative images taken 50 h after scratching/wound healing.
Fig 6.
Wound healing curves of untreated (scrambled controls) and specific gene silencing experiments by siRNA, transfected into PC-3 cells.
A. Wound healing curves of untreated PC-3 cells and PC-3 cells after siRNA mediated knock-down of all eight candidate genes. AllStar control siRNA induces programmed cell death and was used here as a control for efficacy of siRNA transfection. B. Wound healing/cell migration for RHOU-silenced PC-3 cells, compared to scrambled control.
Fig 7.
Representative confocal images 3D organoids, formed by PC3 cells embedded in Matrigel.
A. Control cells were treated with non-functional scrambled siRNA; or with only transfection agent alone. These organoids show multiple invasive processes, typically involving chains of cells (collective invasion pattern). B. Silencing of DLX1 results in prominent growth inhibition and formation of small, round, dense and poorly proliferative organoids, but do not show any invasive properties. C. Similarly, silencing of the RHOU gene and D. of the CACNA1D calcium channel result in round organoids devoid of any invasive processes.
Fig 8.
Images of 3D organotypic cell cultures of PC3 cells, embed in Matrigel, after segmentation and subsequent image analysis using the AMIDA software package.
A. Untreated cells form large organoids with overt invasive processes. B. Silencing of PLA2G7 and C. laminin beta 1 (LMNB1) result in well-rounded, poorly invasive organoids, with higher cell density. D. In contrast, silencing of the TDRD1 gene results in significant induction of invasive properties, further loss of the structural organization or maturation of organoids, and decreased cell density.
Fig 9.
Quantitative assessment of phenotypic changes as the result of siRNA silencing in organotypic 3D cultures.