1. Introduction
Human high mobility group box B (HMGB) proteins HMGB1, 2, and 3 are differentially expressed in many different tissues and cell types, whereas HMGB4 expression is restricted to the testis [
1]. HMGB2 has 82.3% sequence similarity with HMGB1, and both proteins have common or redundant functions in inflammation [
2], chromosome remodeling activity [
3], V (D) J recombination [
4], and embryonic development [
5].
HMGB1 has been related to the onset and progression of cancer, being involved in events such as replenishing telomeric DNA and maintaining cell immortality [
6], autophagic increase, evasion of apoptosis [
7,
8], as well as cell proliferation and invasion [
9,
10]. HMGB1 is also involved in dedifferentiation during epithelial to mesenchymal transition (EMT) [
11] via the receptor for advanced glycation endproducts RAGE/ nuclear factor kappaB NF-κB signaling pathways [
12] and in angiogenesis [
13]. The role of HMGB2 in these processes, although less well studied, has also been related to cell viability and invasion [
14], EMT [
10], and angiogenesis [
15].
The majority of the prostate cancers (PCa) are adenocarcinomas characterized by glandular formation and the expression of androgen receptor (AR) and prostate-specific antigen (PSA). Hormonal inhibition of AR signaling is the therapeutic choice for patients with adenocarcinomas, but unfortunately, the disease usually progresses as it becomes independent of exogenous AR induction, leading to castration-resistant prostate cancer (CRPC) with a worse prognosis. In prostatic small cell neuroendocrine carcinoma (SCNC), the tumor cells are negative for AR and PSA expression and do not respond to hormonal therapy [
16]. Among the most frequently used PCa cell lines, PC-3 characteristics are considered closer to a SCCN PCa model and those of DU145 (ATCC
® HTB-81™) or LNCaP (lymph node carcinoma of the prostate) are considered closer to adenocarcinoma models [
16]. PC-3 and DU145 are AR-independent, and LNCaP is AR-dependent [
16,
17]. Interestingly, upregulation of HMGB1 mRNA and protein have been detected in PCa tumors [
12,
18] and PCa cell lines (including PC-3 and DU145 or LNCaP) compared to the non-transformed immortalized prostate cell line RWPE-1 (prostate epithelial transformed by HPV)) [
18]. Silencing of HMGB1 in LNCaP cells inhibits cell growth [
19]. HMGB1 expression is notably high in PCa metastasis [
12] and is positively correlated with some clinical-pathological parameters, such as Gleason score or preoperative PSA concentration, being associated with a worse prognosis [
18].
Proteomic studies in relation to PCa have been reported [
20,
21,
22], with interactome strategies being outstanding in recent developments [
23,
24,
25]. The purpose of our study was to analyze proteins interacting with HMGB1 and HMGB2 by the yeast 2-hybrid approach (Y2H), using HMGB1 and HMGB2 baits. Results from the screening of libraries constructed from the PC-3 line, as a model of metastatic AR-independent PCa, and of libraries obtained from PCa adenocarcinoma primary tumor are presented. Analyses of copy number alterations (CNA) and mRNA levels of detected targets in public PCa databases are discussed showing that dysregulation of some HMGB1/2 targets is associated with clinical prognosis. Considering that HMGB proteins are known regulators of gene expression, we also tested whether HMGB1 and HMGB2 silencing affects the expression of their Y2H detected partners and found that this regulatory mechanism is functional in PC-3 cells.
3. Discussion
High mobility group box B (HMGB) proteins are pivotal in the development of cancer [
6,
8,
10], and HMGB1 overexpression has been related to principal cancer hallmarks [
7]. Interactome targets of HMGB1 or HMGB2 that have been identified in our Y2H study were previously found to be related to cancer hallmarks (
Table S1 and
Figure 1), and are also dysregulated in PCa, as confirmed by detection of changes in mRNA or protein levels. DNAAF2 [
98], U2AF1 [
43], C1QBP [
40], Snapin, or HDLBP [
99] are upregulated in prostate tumors or PCa cell lines. Others increase their expression after androgen-deprivation therapy, such as KRT7 or NOP53 [
100]. Functional studies interfering the expression of several of the proteins revealed by our study also directly associated them to PCa. In this sense, selective knockdown of C1QBP through iRNA decreased cyclin D1, increased p21 expression, led to cell cycle arrest (G1/S transition) in PCa cells, and had no effect on a noncancerous cell line [
40]. NOP53 acts as a tumor suppressor, and knockdown of the gene in the PCa LNCaP cell line increased the invasiveness of these cells as measured in a xenograft animal model [
101].
Two already known regulatory factors have been found among the HMGB1 interactome targets, YY1 and HOXA10, and both are associated with PCa. YY1 is upregulated in human prostate cancer cell lines and tissues [
66]. Inhibition of YY1 reduces expression of genes related to the Krebs cycle and electron transport chain in PCa cell lines [
67], and YY1 depletion correlates with delayed progression of PCa [
68]. Overexpression of YY1 can promote epithelial-mesenchymal transition by reducing hnRNPM expression [
69]. YY1 can also silence tumor suppressor genes, such as XAF1 in PCa [
70]. In summary, YY1 is a recognized prostate cancer driver [
66] and different complexes in which YY1 takes part can induce activation or repression of gene expression, including also AR-YY1-mediated PSA transcription [
102], which we found is also regulated by HMGB1 and HMGB2 silencing. HOXA10 is upregulated in PCa [
31], and inverse correlations between HOXA10 expression and Gleason pattern, Gleason score, and pathological stage are found [
32], although downregulation of HOXA10 gene expression may enhance lipogenesis to promote PCa cell growth and tumor progression to the castration-resistant stage [
103]. Silencing of HOXA10 expression in PC-3 cells by iRNA decreased proliferation rates, whereas HOXA10 overexpression had the opposite effect [
31]. Physical interaction between these PCa-associated proteins and HMGB proteins has not previously been described, and our results therefore show that there is a connection between HMGB1 and HMGB2 functions and those of their binding partners in PCa.
Considering that HMGB1, HMGB2, and a subset of their interactome partners are upregulated in PCa, we silenced HMGB1 and HMGB2 and analyzed the mRNA levels of a group of randomly selected partners in PC-3 cells (
Figure 6). The data show that HMGB1 and HMGB2 control the expression of them, which might contribute to the orchestrated action of all these proteins in PCa. HMGB2 activates many of the tested targets, but unexpectedly, HMGB1 has the opposite effect. One can propose several reasons to explain upregulation of targets in these circumstances. Data from the genotype-tissue expression (GTEx) project [
104] indicates that, although both HMGB1 and HMGB2 are upregulated in PCa versus noncancerous cells, the relative increase is higher for HMGB2 (×1.5) than HMGB1 (×1.3); this could explain the increased expression of several of their targets, assuming that positive regulation caused by HMGB2 predominates over negative regulation caused by HMGB1 during the onset of PCa. Alternatively, differential interaction of HMGB1 or HMGB2 with their different nuclear partners, the transcript factors detected in our Y2H analysis being among them, might condition their positive or negative regulatory roles on the expression of specific genes.
Clinically, a high frequency of CNA of the genes encoding the identified proteins is associated with the most aggressive forms of PCa: small cell neuroendocrine carcinoma (SCNC) or castration-resistant PCa (
Figure 3). Their gain or amplification in the genome of the cancerous cells are positively correlated to a lesser disease-free period for PCa patients (
Figure 4). The mRNA levels of a subset of these proteins are also higher in metastases than primary tumors (
Figure 5). In conclusion, the set of proteins detected though our HMGB1-HMGB2 Y2H analysis are associated with the most aggressive cases of PCa. Although the PSA-based test is routinely employed for screening of PCa, it has resulted in overdiagnosis and overtreatment of nonaggressive cancers, thus reducing the quality of life of patients.
Consequently, an improvement is necessary in the initial stages to discriminate between high-risk from low risk cancers. Our data on HMGB1 and HMGB2 interactome targets, considering their correlation to high aggressiveness and bad prognosis, is a good starting point to develop new serum protein panels for improvement of PCa diagnosis. Indeed, FLNA has already been proposed in a clinical validated PCa biomarker panel in serum [
74]. PSMA7 was also proposed as a PCa biomarker [
55], and KRT7 is included in a whole blood mRNA 4-gene androgen regulated panel for PCa diagnosis [
33]. Considering the relative expression levels of our HMGB1 and HMGB2 interactome targets in noncancerous cells or in blood of health subjects differ quite notably (
Figure 7), one might anticipate that more sensitive analyses could be carried out using as biomarkers those proteins that are usually lowly expressed in noncancerous cells; thus, their levels are also low in the blood of healthy individuals. For instance, FLNA reported as a possible biomarker [
74] is one of the 50 proteins most strongly expressed in normal prostate, and high levels are also detected in the blood of healthy individuals, whereas other detected HMGB1 or HMGB2 interactome targets in our study, e.g., DNAAF2, GOLM1, or TGM3, are in the lowest rank of detection in noncancerous samples and their increase should become more discriminatory.
4. Materials and Methods
4.1. Biological Materials
PC-3 is an androgen-independent cell line derived from a bone metastasis [
106]. The human PCa PC-3 cell line, regularly validated by DNA typing, was obtained from the American Type Culture Collection ATCC and grown in Roswell Park Memorial Institute RPMI-1640 media, supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin-streptomycin (Thermo Fisher Scientific, Inc. Waltham, MA, USA). Cells were cultured at 37 °C in 5% CO
2 in air in a humidified incubator. RNA from PCa tissue, isolated after radical prostate resection of a 66-year-old man diagnosed with adenocarcinoma (Gleason score 6) and not previously treated with radiotherapy or chemotherapy, was obtained from Biobanco de Andalucía (SPAIN).
4.2. Yeast Two Hybrid Methodology
Sacchacomyces cerevisiae strains were Y187 (MATα, ura3-52, his3-200, ade2-101, trp1-901, leu2-3, 112, gal4Δ, gal80Δ, met-, and URA3::GALuas-GAL1TATA-LacZ MEL1) and Y2HGold (MATa, trp1-901, leu2-3, 112, ura3-52, his3-200, gal4Δ, gal80Δ, LYS2::GAL1uas-GAl1TATA-His3, GAL2uas-Gal2TATA-Ade2 URA3:: MEL1UAS-Mel1TATA, and AUR1-C MEL1).
Total RNA from the PC-3 cell line obtained from the supplier (Sigma-Aldrich, Saint Louis, MO, USA) and RNA from PCa tissue (Biobanco de Andalucía, Spain) were used to construct cDNA libraries. HMGB1 and HMGB2 interacting partners were identified using Matchmaker Gold Yeast 2-Hybrid System (Clontech, Fremont, CA, USA). Library construction, bait construction, and Yeast 2-Hybrid library screening were done according to the Takara Bio USA Matchmaker
® Gold Yeast 2-Hybrid System manual. In brief, the baits were cloned as fusions to the GAL4 activation domain in the plasmid pGBKT7-AD and used to transform the yeast haploid strain, Y187. cDNA libraries prepared from RNA extracted from PC-3 cells and PCa cancerous tissue were included as fusions to the GAL4 DNA-binding domain in the plasmid pGBKT7-BD and were used to transform the yeast haploid strain, Y2HGold. RNAs from human samples used to prepare the Y2H libraries were obtained from Biobanco de Andalucía (Spain). RNA was extracted from frozen tissue sections in OCT (Optimal Cutting Temperature) compound, using the Qiacube robot from QIAGEN, based on ion-exchange columns with a silica membrane. RNA was obtained with the miRNeasy mini-kit from QIAGEN that allows recovery of both total RNA and miRNAs. The samples were finally treated with RNase-free DNAase from QIAGEN. The RNA was quantified at 260 nm and 280 nm by spectrophotometry using Infinite F200 equipment of TECAN with a Nanoquant plate. Finally, the integrity of the samples was evaluated by AGILENT 2200 Tape Station apparatus, with the RIN (RNA Integrity Number) parameter being >8. Efficiency in the constructions of libraries was in the range recommended in the kit (all libraries guaranteed to have >1 × 10
6 independent clones). As a previous control, we confirmed that our baits (HMGB1 and HMGB2) do not autonomously activate the reporter genes in Y2HGold in the absence of a prey protein. Bait and prey fusion proteins are each expressed in different haploid yeast strains that can form diploids. The diploid yeast cell expresses both proteins, and when fusion proteins interact, the transcriptional activator GAL4 is reconstructed and brought into proximity to activate transcription of the reporter genes. For diploid formation, 1 mL of concentrated bait culture was combined with 1 mL of library culture and incubated overnight with slow shaking. A drop of the culture was checked under a phase-contrast microscope (40×) to confirm the existence of zygotes before plating on diploid-selective media. Diploids were tested for expression of the reporter genes in selective media. To reduce the appearance of false positives, a screening based on three different independent markers (ADE2, HIS3, and MEL1) was selected. pGBKT7-BD plasmids carrying the preys were rescued from confirmed positive diploids, and DNA was used to transform
E. coli. The inserts were sequenced with primer T7 (5′-TAATACGACTCACTATAGGG-3′). Sequences were used for homology searches with BlastN and BlastX at the National Center for Biotechnology Information NCBI (
https://blast.ncbi.nlm.nih.gov/) and proteins in the database matching the queries annotated as positives.
4.3. Expression Analysis by Quantitative Polymerase Chain Reaction (RT-qPCR)
Individual analyses of gene expression were carried out as follows. RNA samples were retro-transcribed into cDNA and labeled with the
KAPPA SYBR FAST universal one-step qRT-PCR kit (Kappa Biosystems, Inc, Woburn, Massachusetts, USA). The primers for qPCR are shown in
Table S3. Reaction conditions for thermal cycling were 42 °C for 5 min, 95 °C for 5 s, 40 cycles of 95 °C for 3 s, and finally 60 °C for 20 s. ECO Real-Time PCR System was used for the experiments (Illumina, Inc., San Diego, California, USA), and calculations were made by the 2
−ΔΔCt method [
107]. Student’s test was used to check the statistical significance of differences between samples (
p < 0.05). The relative mRNA levels of the experimentally selected genes (target genes) were calculated by referring to the mRNA levels of the housekeeping gene, encoding glyceraldehyde phosphate dehydrogenase (GAPDH), which had been verified as being expressed constitutively under the assay conditions. For valid quantification using the 2
−ΔΔCt method, it is crucial that target and housekeeping PCR amplification efficiencies are approximately equal: we therefore verified that the efficiencies of the 2 PCR reactions differed by <10%. At least 2 independent biological replicas and 3 technical replicas of each of them were made for all the experiments.
4.4. Immunoprecipitation
One hundred µl of Protein G Plus-Agarose immunoprecipitation-reagent (Santa Cruz Biotechnology, Dallas, TX, USA) were coupled with 4 µg of anti-HMGB1 antibody (sc-74085; Santa Cruz Biotechnology) or anti-mouse antibody (Molecular Probes, A10534) in phosphate buffered saline (PBS) for 1 h at 4 °C with rotation. PC-3 cells were lysed in 20 mM Tris/HCl, 150 mM, 1% Triton X-100, 1× phenylmethylsulfonyl fluoride (PMSF), and protease inhibitor cocktail (Sigma-Aldrich, Saint Louis, MO, USA) and incubated for 30 min at 4 °C with rotation. Total protein (500 µg) was incubated with the antibody agarose beads overnight and eluted by incubation in 1× lithium dodecyl sulfate LDS loading buffer containing 350 mM β-mercaptoethanol at 95 °C for 10 min. Mass spectrometry and data analysis were done as previously described [
26].
4.5. Western Blot Analysis
Protein samples were run on 10% SDS-PAGE gels at 80 V for 20 min followed by 200 V for 45–60 min. Proteins were transferred onto a polyvinylidene fluoride (PVDF) membrane at 0.2 A for 1 h. Membranes were blocked by incubating with 5% non-fat dry milk for 1 h at room temperature (RT) and then incubated with primary antibodies, anti-HMGB1 (sc-74085; Santa Cruz Biotechnology) or anti-Cytokeratin 7 (ab181598; Abcam, Cambridge, UK) in phosphate-buffered saline with 0.1% Tween 20® detergent PBST overnight at 4 °C. After incubation with the corresponding horseradish peroxidase-conjugated secondary antibody, enhanced chemiluminescence for high sensitivity and long-lasting signal (ECL) Anti-mouse IgG (NXA931 from GE Healthcare Sciences, Chicago, IL, USA) or ECL Anti-rabbit IgG (NA934 from GE Healthcare Sciences, Chicago, IL, USA), protein bands were detected using LuminataTMCrescendo Western HRP Substrate (Millipore Corporation, Burlington, MA, USA) and a ChemiDocTM imager (Bio-Rad laboratories Hercules, CA, USA).
4.6. Immunofluorescence and Confocal Microscopy
Cells were plated in 6-well plates, each containing 4 sterile 13-mm glass coverslips. When 80% confluent, cells were fixed in 4% paraformaldehyde in PBS for 15 min at RT. Cells were washed 3 times with PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, and 2 mM KH2PO4) and finally treated with 0.1% Triton/PBS for 15 min at RT. They were then blocked in 1% bovine serum albumin (BSA) in PBS for 1 h at RT. Primary antibodies, anti-HMGB1 (sc-74085; Santa Cruz Biotechnology) or anti-Cytokeratin 7 (ab181598; Abcam, Cambridge, UK) were diluted in 1% BSA in PBS. Cells were incubated with the corresponding primary antibodies overnight at 4 °C, followed by 3 washes with PBS and staining with the secondary antibodies, modified with Alexa Fluor 488 and 568 (Invitrogen, Carlsbad, CA, USA) previously diluted in 1% BSA in PBS for 1 h at RT in the dark. For nuclear staining, after secondary antibody incubation, wells were washed 3 times and stained with Hoechst (Life Technologies, Carlsbad, CA, USA) for 5 min at RT in the dark. Cells were washed once with PBS and once with sterile distilled water. Each coverslip was mounted on a clean slide using ProLong™ Gold Antifade Mountant (Invitrogen). After drying, the slides were stored at 4 °C in the dark until they were examined by confocal microscopy (Nikon A1R). Meander’s correlation coefficient was calculated using Nis Elements software from Nikon.
4.7. HMGB1 and HMGB2 Silencing by siRNA
The PC-3 cell line was transfected with small interfering (si)RNA oligonucleotides using Lipofectamine 2000 (Invitrogen). siRNA and Lipofectamine 2000 were each diluted separately with Opti-MEM (Gibco), mixed together, and incubated for 5 min at RT. The mixture was added to cells plated in 3 mL RPMI 1610 medium (final concentration of siRNA, 50 nM). Cells were collected at 48 h post transfection for further analysis. The following siRNAs (Life Technologies) were used for the silencing of each gene: s20254 Silencer Select for HMGB1, s6650 for HMGB2, and AS02A5Z3 for the siRNA negative control.
Total RNA was extracted from different conditions (siHMGB1, siHMGB2, and siCtrl#2) of the PC-3 cell line using GeneJET RNA Purification Kit (#K0731, Thermo Scientific). The remaining DNA was removed by incubating with DNase I, RNase-free (#EN0521, Thermo Scientific). DNA-free RNA was finally purified using GeneJET RNA Cleanup and Concentration Micro Kit (#K0842, Thermo Scientific). qPCR reactions were run in triplicate in an Eco Real-Time PCR System (Illumina) using 1 ng per reaction. PC-3 lysates of each condition were extracted with lysis buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 0.1% NP40, 1 mM ethylenediaminetetraacetic acid disodium salt (EDTA), and 2 mM MgCl2), and protein concentration was quantified using Bradford Reagent (Bio-Rad). Protein samples of 25–40 µg were loaded for western blotting. PVDF membranes were incubated overnight at 4 °C with primary antibodies, anti-HMGB1 (ab18256, Abcam), anti-HMGB2 (ab67282, Abcam), or anti-α-tubulin (sc53646, Santa Cruz Biotechnology).
4.9. Statistical Analysis
Analyses were carried out using GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA). Continuous variables were expressed as mean ± SE. Relative gene expression assays were tested using independent t-tests. A 2-tailed p-value test was used with p < 0.05 considered significant.