Expression of cancer-associated genes in prostate tumors at mRNA and protein levels

G. V. Gerashchenko, O. V. Grygoruk, E. E. Rosenberg © 2019 G. V. Gerashchenko et al.; Published by the Institute of Molecular Biology and Genetics, NAS of Ukraine on behalf of Biopolymers and Cell. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited Genomics, Transcriptomics and Proteomics ISSN 0233-7657 Biopolymers and Cell. 2019. Vol. 35. N 1. P 39–53 doi: http://dx.doi.org/10.7124/bc.000995


Introduction
Earlier, we have demonstrated that relative expression of seven cancer-associated genes, namely the TGFB1, IL1B, FOS, EFNA5, TAGLN, PLAU and EPDR1genes, is altered in prostate cancer cell lines [1,2] and prostate cancer tissues [3]. The proteins, encoded by these genes, play an important role in carcinogenesis and are involved in a number of cellular processes and pathways. For example, TGFB1 is implicated in the control on EMT and angiogenesis [4]. Importantly, there is an interplay between TGFB1 and androgen receptor signaling pathways, that is crucial for the development and progression of prostate cancer [5]. Usually, TGFB1 is expressed in reactive tumor stromal cells, i.e. cancer-associated fibroblasts (CAFs) [6]. Another important player is IL1B, a pro-inflammatory cytokine, expressing in immune cells and activating the NF-kappa B pathway [7]. High levels of IL1B promote the skeletal colonization and progression of metastatic prostate cancer [8]. FOS is a transcription factor that takes part in many cellular processes, cell proliferation and apoptosis are among those [9,10]. FOS is involved in the development of castration-resistant prostate cancer and also in metastasizing [15], as well as in the] development of other tumor types [11][12][13][14].
EFNA5, TAGLN and EPDR1 encode proteins of the adhesion machinery, thus controlling the tumor progression [16]. PLAU may regulate migration and invasion upon the development of endometrial [17] and prostate [18] tumors.
Besides alteration of the expression pattern of the described seven genes, we found that the prostate-specific genes [19] and the tumor stromal elements [20] show differential expression in the tissues samples of prostate cancer, compared with the benign tumors. Also, the expression of genes, involved in EMT was altered [21], and in a proportion of prostate tumors the presence of the TMPRSS2:ERG fusion was detected [22].
In the present work we assessed the expression of seven genes (EFNA5, EPDR1, FOS, IL1B, PLAU, TAGLN and TGFB1) at the mRNA and protein levels and analyzed the putative correlation between the expression of these genes and the prostate-specific genes, tumor stromal elements and genes, controlling EMT.

Materials and Methods
A collection of prostate tissues. Samples of cancer tissue and CNT (at an opposite side of tumor) were frozen in liquid nitrogen immediately after surgical resection at the National Cancer Institute of National Academy of Medical Sciences of Ukraine (NAMU) (Kyiv, Ukraine). Benign prostate tumors (prostate adenoma samples) were collected at the Institute of Urology of NAMU (Kyiv, Ukraine) after radical prostatectomy and were frozen, as described above. All protocols were in accordance with the Declaration of Helsinki and the guidelines, issued by the Ethic Committees of the Institute of Urology of NASU, the National Cancer Institute of MHC and the Institute of molecular biology and genetics of NASU. Experiments were conducted on 29 prostate adenocarcinoma samples of different GS and tumor stages, 29 paired CNT samples and 14 samples of benign prostate tumors (adenomas). Tumor samples were characterized, according to the International System of Classification of Tumors, based on the tumornode-metastasis (TNM) and the World Health Organization (WHO) criteria.
[The] Clinicopathological characteristics (CPC) of adenocarcinomas and the presence and/or absence of the TMPRSS2/ERG fusion that was reported by us earlier [1,21] are presented in Table 1.
Total RNA isolation and cDNA synthesis. 50-70 mg of frozen prostate tissues were mashed to a powder in liquid nitrogen. Total RNA was extracted by TRI-reagent (SIGMA), according to a manufacturer's protocol. Total RNA concentration was analyzed by a spectrophotometer (NanoDrop Technologies Inc. USA). The quality of the total RNA was determined in a 1 % agarose gel by band intensity of 28S and 18S rRNA (28S/18S ratio). cDNA was synthesized from 1 µg of the total RNA, that was treated with the RNase free DNase I (Thermo Fisher Scientific, USA), using RevertAid H-Minus M-MuLV Reverse Transcriptase (Thermo Fisher Scientific, USA), according to the manufacturer's protocol.
Analysis of a protein expression pattern by IHC in prostate tissues. Fresh prostate tissues were fixed in a neutral buffered 4 % formaldehyde solution. After fixation, dehydration, and embedding in paraffin, serial sections were cut at a thickness of 5 μm and stained with hematoxylin/eosin for histological diagnosis.
Expression of the TGFB1, PLAU, FOS, IL1b and TAGLN proteins was assessed, using the specific antibodies by immunohistochemistry. After heating at 56ºC, paraffin was dissolved in xylol and the tissue was rehydrated by stepwise washing with ethanol in phosphate-buffered saline (PBS) (99 %, 90 %, 70 %, and 30 % ethanol). Tissues were then treated with a 2 % solution of hydrogen peroxide in methanol at room temperature for 30 min, to reduce background staining. Epitopes were exposed in a hot citrate buffer in 92°C water bath for 15 min. Antibodies were diluted (1:100 mouse antibodies and 1:100 -rabbit) in the blocking buffer (2 % bovine serum albumin, 0.2 % Tween-20, 10 % glycerol, and 0.05 % NaN3 in PBS). Protein signals were visualized by an EnVision™ Detection Peroxidase/DAB system (Dako, Glostrup, Denmark). Nuclei were stained with Mayer's hematoxylin (Dako).
Statistical analysis. The Kolmogorov-Smirnov test was used to analyze the normality of distribution. The Wilcoxon Matched Pairs test was performed to compare RE in prostate adenocarcinoma and paired CNT, using the 2 -ΔCt model. The Benjamini-Hochberg procedure with false discovery rate (FDR) 0.10-0.25 was used for multiple comparisons [24]. Differences in RE more, than two-folds were considered as significant, for the 2 -ΔΔCt model (i.e. > 2.01 and < 0.49). The Fisher exact test was performed to analyze differences between these sample groups [19,20]. The Kruskal-Wallis test was used to determine differences between groups of T, CNT and A in 2 -ΔCt model. The Dunn-Bonferoni post hoc test for multiple comparisons was performed to analyze RE differences between pairs of investigated groups. The Spearman's rank correlation test was used to find the putative correlations between RE and CPC of prostate tumors and also between RE levels of the studied genes. The K-Mean clustering was applied for prostate cancer subtyping and also for the specific gene expression profiles, following by the Kruskal-Wallis and Dunn-Bonferoni post hoc tests for detection of RE differences between clusters.

Expression pattern of the EFNA5, EPDR1, FOS, IL1B, PLAU, TAGLN and TGFB1 genes in prostate tissues
According to the Kolmogorov-Smirnov test, RE of investigated genes in the adenoma group did not show the Gaussian distribution (normal); therefore, nonparametric statistical tests and methods were used. We assessed RE levels of seven cancer-associated genes in the paired T/CNT samples, using the 2 -ΔCt and 2 -ΔΔCt calculations. The samples were grouped, according to the GS (GS ≤ 7, GS > 7), tumor stage (stage 1-2 and stage 3-4) and by the presence of the TMPRSS2/ERG fusion transcript. Noteworthy, the EFNA5 and EPDR1 genes show very low expression in both tumors and normal tissues.
The Wilcoxon Matched paired test in the 2 -ΔCt model showed that only two genes, namely PLAU and IL1B were differentially expressed in various groups. PLAU was de- creased significantly in T, compared with corresponding CNT (N) (p = 0.0092). The same was true when the paired adenocarcinomas and CNT with GS ≤ 7 (p = 0.0258), a group of paired T/N with stage 3-4 (p = 0.0206) and the fusion negative paired T/N (p = 0.0229) were analyzed. IL1B expressed at significantly lower levels in T with GS > 7 (p = 0.0192).
Using the 2 -ΔΔCt model and calculations of the Fisher exact test, we found that EFNA5 (p = 0.021) and PLAU (p = 0.038) were expressed at lower levels in tumors, compared with the paired CNT. Of note, only IL1B expressed at lower levels in adenocarcinomas with GS > 7 (p = 0.030) and in the adenocarcinoma group where the TMPRSS2/ERG fusion transcript was detected (p = 0.030).
Importantly, statistical calculations did not vary, regardless of whether RE fold changes were assessed, according to the 2-ΔCt or 2-ΔΔCt model.
The changes in RE levels of the investigated genes were calculated for the samples of three groups (T, N, A) ( Table 2).
We found that RE values of the majority of the investigated genes fluctuated in each group, especially in adenocarcinomas. Taking into consideration a nature of CNT, the A group was used as the control [19,21]. Significant RE differences between the groups were detected by the Dunn-Bonferroni post hoc test for multiple comparisons (p < 0.05).
We found significant differences (p < 0.05) in RE of three genes (FOS, PLAU, EPDR1) between the T, N and A groups (Figure 1). RE in adenoma samples was a normalization point. FOS was induced in T and CNT, compared with A whereas PLAU and EPDR1 were decreased. The same character of RE  changes for these three genes we have observed in the groups with different tumor stages ( Figure 2). Noteworthy, RE of five genes (FOS, EFNA5, TAGLN, PLAU and EPDR1) changed depending on GS (p < 0.05) in adenocarcinomas, compared to the A group ( Figure 3). These genes were expressed similarly in the T and CNT samples. Next, we calculated the RE pattern for the groups, where the TMPRSS2/ERG fusion was either present or absent (Figure 4). We have found the specific changes in RE of FOS and EPDR1 in the group of samples, where no fusion was detected (Figure 4).

Relations of changes in RE patterns of the investigated genes with CPC and expression of genes, encoding hormone receptors, stromal markers and controlling EMT
The Spearman Rank Order Correlations (r s ) analysis did not show any correlation between RE of the investigated genes and CPC, as we found earlier [3].
We calculated many significant gene-togene correlations between RE of the investigated genes in adenocarcinomas (Table 3A). The biggest number of correlations (5 out of 6 calculated) was found for TGFB1 and Earlier, we demonstrated that the expression pattern of the genes, controlling EMT [21], encoding receptors, metabolic enzymes [19] and tumor microenvironment markers [20] dramatically altered in prostate tumors, compared with adenomas. Now we report that expression of the seven presently investigated genes follows many correlations to RE of these genes (Table 3B). We have found that RE of 23 genes (out of 56) correlated significantly with RE of seven presently investigated genes. The most interesting among all the genes is TAGLN in this sense.

K-means clustering
Next, we wanted to group the samples of prostate adenocarcinoma, considering RE of

Table 3. The Spearman Rank Order Correlations (r s ) of RE patterns of the investigated genes (A) in relation to expression of genes, encoding hormone receptors, stromal elements and controlling EMT (B)
A.  the seven investigated genes and CPC, i.e. GS and tumor stage. The K-means clustering was performed and as a result, two specific clusters were formed, that included all the samples of prostate adenocarcinoma ( Figure 5, Table 4). In these clusters, the expression of FOS, PLAU and EDPR1 varied significantly. The first cluster contained mainly the tumors with median GS = 6, and the second cluster -with GS = 9.

Gene/Gene
In other words, the second cluster (Cluster 2 in Figure 5) consisted of more aggressive prostate adenocarcinomas.   probably, in blood cells in tumor (green arrows). More infiltrating lymphocytes were found in low differentiated prostate carcinoma, than in hyperplasia. In epithelial prostate cells IL1B was hardly detectable. Noteworthy, the PLAU protein showed expression pattern, opposite to FOS -the weak PLAU signal in hyperplasia vanished in highly advanced carcinomas ( Figure 6C). Notice a decrease of the brown signal in the epithelial prostate cells (violet arrows). The right panel shows the magnified field, indicated by a red square on the left panel.
The TAGLN protein was not detected in prostate cells in hyperplasia (red arrows, Figure 6D, the top panel). Of note, it was highly expressed in stromal fibroblasts (black

Expression pattern of FOS, IL1B, PLAU, TAGLN and TGFB1 proteins in prostate tissues
Using the IHC, we found that the expression of FOS protein was different in hyperplasia and tumors: the FOS signal was more intensive in low differentiated tumors, compared to prostate hyperplasia ( Figure 6A). Notice an increase of the brown signal in the epithelial prostate cells (red arrows). The right panel shows the magnified field, indicated by a red square on the left panel.
The IL1B signal was detected, most probably, in blood cells ( Figure 6B). Notice the absence of the brown signal in hyperplasia (red arrows). The brown signal is detected, most arrows, Figure 6D, the top panel). Upon cancer development, the prostate cells remained negative for TAGLN ( Figure 6D, the middle and bottom panels). Notice the absence of the brown signal in the epithelial prostate cells (red arrows). The right panel shows the magnified field, indicated by a black square on the left panel. Of note, the stroma cells express TAGLN (black arrows). Due to the fact, that less fibroblasts were present in the Stage IV tumors, the TAGLN expression at the mRNA levels was reduced as shown, using the q-PCR.
The TGFB1 signal was quite strong in hyperplasia (red arrows in Figure 6E, the top panel). In moderately differentiated cancers TGFB1 was decreased (black arrows in Figure  6E, the middle panel). In low differentiated tumors the TGFB1 protein was hardly detected ( Figure 6E, the bottom panel). Notice the strong brown signal in the epithelial prostate cells in hyperplasia (red arrows). Of note, the TGFB1 signal is decreased in stage I tumor (black arrows) and is absent in stage IV tumor. The right panel shows the magnified field, indicated by a black square on the left panel. In other words, upon the development of pros-tate cancer the levels of TGFB1 gradually decreased in the prostate tissue cells.

Discussion
In the present paper, we investigated whether the expression of seven genes, namely EFNA5, EPDR1, FOS, IL1B, PLAU, TAGLN and TGFB1, follows the pattern of the EMT-related genes, the prostate cancer-associated genes and several tumor stromal markers. In addition, we wanted to understand, whether the presence and/or absence of the TMPRSS2/ERG fusion can influence the expression of the abovementioned genes. Moreover, the expression assessment of these seven genes at the mRNA levels was supplemented by the analysis of the encoded proteins ( Table 5).
The TGFB1 protein signal decreased upon the tumor progression. The q-PCR analysis demonstrated high levels of dispersion of RE values. The means at the minimum and maximum were scattered for more than 100 fold. Of course, TGFB1 is expressed in various cells. Therefore, we did not demonstrate significant changes in TGFB1 RE between the groups of prostate tumors.