Methodological aspects of the molecular and histological study of prostate cancer: focus on PTEN.

Prostate cancer is among the most frequent cancers in men, and despite its high rate of cure, the high number of cases results in an elevated mortality worldwide. Importantly, prostate cancer incidence is dramatically increasing in western societies in the past decades, suggesting that this type of tumor is exquisitely sensitive to lifestyle changes. Prostate cancer frequently exhibits alterations in the PTEN gene (inactivating mutations or gene deletions) or at the protein level (reduced protein expression or altered sub-cellular compartmentalization). The relevance of PTEN in this type of cancer is further supported by the fact that the sole deletion of PTEN in the murine prostate epithelium recapitulates many of the features of the human disease. In order to study the molecular alterations in prostate cancer, we need to overcome the methodological challenges that this tissue imposes. In this review we present protocols and methods, using PTEN as proof of concept, to study different molecular characteristics of prostate cancer.

Introduction

Prostate cancer (PCa) is among the deadliest forms of cancer (WHO), and represents the third cause of death by cancer in men (www.globocan.iarc.fr). The tumor suppressor PTEN is among the most mutated and lost tumor suppressors in PCa [1]. Up to 70% of PC as harbor loss of PTEN at presentation. This tumor suppressor is located at the top of a highly oncogenic signaling pathway, the PI-3 Kinase (PI-3K) cascade, which contains many other oncogenes and tumor suppressors [2]. In addition, regulatory feedback loops stem from the PTEN/PI-3K pathway to ensure cell homeostasis, which decrease the efficacy of single agent therapies [2,3].

PTEN down-regulation is not restricted to genetic events, and regulation of its transcription, translation and stability can play an important role. PTEN is frequently lost in heterozygosity, whereas mostly advanced cancers exhibit complete loss of the tumor suppressor. Interestingly, the prostate epithelium is exquisitely sensitive to the reduction in PTEN levels. This concept has been formally proven in mice through the use of genetic interference, which allows a partial reduction of the expression of the interfered allele [4,5]. While PTEN heterozygous mice present PIN lesions in the prostate with long latency [6], PTEN hypomorphic mice show progression of the prostate lesions to invasive cancer at higher penetrance [5]. Importantly, while a gradual decrease of PTEN promotes prostate cancer progression, acute and complete PTEN-loss elicits the activation of a fail-safe senescence response, which is driven by the up-regulation of the tumor suppressor p53 [7]. This novel type of senescence is genuinely distinct from the classic oncogene-induced senescence [8]. Importantly, genetic or environmental events regulating this process may be key players in the progression of prostate cancer and therefore attractive targets for anticancer therapy [9,10].

All these evidence point to the need of studying PTEN-dependent pathways in prostate cancer. However, the technical challenges related to the study of this type of tumor require special attention, and hence, in this review we aim at describing a series of methodologies to study prostate cancer biology, with a reference to the pathway aforementioned.

Methods and results

Preparation of well-diagnosed prostate cancer specimens for molecular studies

Cancerous lesions in the prostate, unlike in other tissues, are difficult to identify macroscopically. This poses a challenge when the aim is to obtain well-diagnosed frozen tissue. To overcome this limitation, we have set up together with the Basque Biobank and Basurto University Hospital (OSI-Basurto, Bilbao, Spain), in collaboration with the Dept. of Pathology at Mount Sinai, a procedure to obtain this type of specimen.

Key materials

A biopsy punch (Miltex Ref. 33–34).

Due to the characteristics of prostate cancer, we established a procedure by which fresh tissue obtained from radical prostatectomy is sliced into left and right lobe (after delimiting the margins of the surgical piece with ink and fixing the ink with acetic acid). All prostate specimens were obtained upon informed consent and with evaluation and approval from the corresponding ethics committee (CEIC code OHEUN11-12 and OHEUN14-14). From each lobe, the dermatologic punch is employed to harvest 8 tissue cylinders of 4 mm diameter. The site of the punches is selected blindly due to the lack of macroscopic alterations associated to cancerous lesions. However, we did notice that the expertise of the pathologist does influence the rate of success in harvesting cylinders with cancer. Of note, this approach prevents from damaging the capsule and a drop of eosin on the site of tissue harvest can help monitoring the histological alterations surrounding the area for diagnostic purposes. Tissue cylinders are then divided longitudinally with a scalpel and dedicated to snap-freeze (in liquid nitrogen or isopentane at −80 °C) and to paraffin embedding for diagnostic purposes (procedure in Fig. 1A–D). Due to the width of the cylinder (4 mm diameter), the diagnosed tissue fraction will closely represent the histological properties of the frozen adjacent tissue. In Fig. 1E, hematoxylin/eosin staining of whole tissue sections from cylinders with different tumor abundance are shown, together with a zoom that shows the correct preservation of the histological properties of the sections. Importantly, this protocol allows us to closely estimate the tumor abundance that we have in the frozen tissue piece, hence solving an otherwise challenge in the acquisition of frozen material. The material obtained from this approach is sufficient to carry out different molecular biology studies, including RNA preparation (described below), protein extraction and metabolite profiling (data not shown).

Fig. 1

Preparation of well-diagnosed fresh frozen biopsies. (A–D) Preparation of the punch biopsy (A) and excision with scalpel (B), identification of the harvest point in surgical piece with eosin (C) and longitudinal separation of the punch with scalpel (D). (E) Histological features of punch biopsies with different abundance of tumoral tissue, whole section hematoxylin/eosin staining is shown together with a zoom to show the histological features of the piece.

Molecular biology analysis from frozen tissue: tips for good quality RNA preparation

Preparation of RNA of high quality from prostate cancer specimens remains a challenge, primarily due to the abundance of RNAses and proteases in the prostate and prostatic fluid. A variety of protocols have been proposed to maximize the quality and yield from biopsies of different origin [11-14] (see also protocols from Prostate Cancer Biorepository Network; SOP N:006 http://www.prostatebiorepository.org). While real time PCR is a low-demanding approach in terms of RNA integrity, the latest OMIC technologies, including RNA sequencing, require material in optimal conditions.

To define the technical needs of an appropriate RNA extraction strategy, we have tested one main technical implementation (the use of phenolic extraction agents) and one variable (the presence of ink and acetic acid in the preparation).

Trizol (Life Technologies/Invitrogen Ref. 15596-018).

Total RNA extraction kit (NucleoSpin® miRNA Ref. 740971.10/50/250).

The protocol is the following, where the alternative procedure with and without Trizol is underlined (the Trizol-based implementation is described in the user manual of the NucleoSpin® miRNA kit):

RNAse inhibition and tissue thawing (a minimal amount of tissue of 10 mg is sufficient for the procedure). RNA later ICE (Life Technologies Ref. AM7030) is used to ensure the maximal inhibition of RNAses and the optimization of tissue homogenization afterwards. The protocol is based on transferring frozen tissue (stored dry at −80 °C) to RNA later ICE (also at −80 °C) and thawing the tissue at −20 °C overnight.

Regular lysis buffer. Tissue is transferred to the recommended volume of NucleoSpin® miRNA lysis buffer.

Trizol-based lysis. Tissue is transferred to 400 μL volume of Trizol. Additional 400 μL are added after homogenization.

Homogenization. 5–6 beads/tube (Ceramic Bead Tubes 2.8 mm, Cat.: 13114-50; MO BIO Laboratories). Homogenization is carried out in Precellys in two cycles of 6000 rpm and 30 s.

RNA extraction. Following the manufactureŕs instructions.

RNA extraction. Following homogenization, we add 160 μL of Chloroform, mix by vortex, incubate 3 minand centrifuge 15 minat 12,000g in tabletop centrifuge. The supernatant (350–400 μL) is transferred to a new tube and mixed with 1 mL of MX buffer. After vortex, the product is loaded in the column and the same process indicated in point 4 is followed.

The results obtained from frozen tissues with a stabilizing agent (RNA later ICE), a total RNA extraction kit, and with or without Trizol implementation are shown in Fig. 2. RNA stabilizing agents and the standard non-phenol based lysis buffer is not sufficient to prevent the RNA from degrading (Fig. 2A), while Trizol implementation results in total RNA of optimal quality for transcriptomic studies (Fig. 2B, RNA Integrity Number – RIN – values in Fig. 2C). Of note, although small RNAs have not been monitored in this procedure, the kit presented herein would allow for their isolation.

Fig. 2

Evaluation of the impact on phenol-based lysis and ink/acetic acid contaminants in RNA quality. (A–C) Bioanalyzer analysis of RNA preparations performed in the absence (A) or presence (B) of Trizol lysis (average RIN values for the samples analyzed are presented in C; ∗∗, significance p < 0.01). (D and E) Representative images of the lysis of samples with increasing amount of ink (the intensity of the dark color reflects the increasing concentration of ink in the sample of origin, which has been separated in three groups as indicated) (D), and RIN values obtained from the RNA preparation (E). (F) Real time quantitative PCR of PTEN (two Taqman probes) and GAPDH shows average Ct amplification values in all samples (left panel) and the lack of correlation between Ct values and the increase in ink (right panel).

On the other hand, we have evaluated with an independent phenol-based RNA extraction kit (Absolutely RNA miRNA KIT. Cat. 400814, Agilent) whether the presence of ink and acetic acid from the margins of the non-tumoral prostate tissue could influence RNA quality. To this end, we selected biopsies containing increasing amounts of these contaminants (Fig. 2D). The presence of these agents did not impact the quality of RNA, as quantified by Agilent Bioanalyzer (Fig. 2E). We further studied if despite yielding good quality RNA, ink and acetic acid could interfere with the retrotranscription and real time quantitative PCR process. We predicted that if the ink/acetic acid interferes with the retrotranscription or real time PCR, we would observe an increase in the Ct values of the genes studied in the high ink conditions. However, evaluation of PTEN expression with two independent Taqman probes (PTEN 48: Universal Probe library [Roche] #48; primer F: ggggaagtaaggaccagagac Primer R: tccagatgattctttaacaggtagc; PTEN 60: Universal Probe library [Roche] #60; primer F: gcacaagaggccctagatttc Primer R: cgcctctgactgggaatagt) and GAPDH (REF. Life Technologies Hs02758991_g1) as housekeeping gene clearly showed a lack of correlation between the amount of ink and any alteration in gene expression (Fig. 2F). In summary, phenol-based RNA extraction coupled to column-based purification significantly improves RNA quality and the presence of ink/acetic acid in the tissue sample does not influence RNA preparation, retrotranscription, or real time PCR amplification.

Monitoring PTEN expression in prostate cancer: an immunohistochemical (IHC) procedure

Immunodetection of PTEN could become critical in the coming years to stratify patients and define the best therapeutic strategies [15,16]. Therefore, good standardized IHC procedures need to be established. Lotan et al. recently established an immunohistochemical protocol for PTEN [17]. We have employed a different clone from Cell Signaling Technology PTEN (138G6) and we have established a sensitive and specific IHC protocol for research purposes.

Key material

Rabbit monoclonal PTEN antibody, clone 138G6 (Cell Signaling Technology, Ref. 9559).

Antigen retrieval was performed with Tris–EDTA (pH 9) in microwave (4 min). H2O2 was used to block the endogenous peroxidase, followed by blocking with goat serum and primary antibody (1:100) incubation overnight at 4 °C. Goat anti-rabbit IgG antibody (1:1000) was incubated at room temperature for 30 min. IHC detection was performed with the ABC Kit from Vector Laboratories. This protocol with DAB-based development results in specific detection of PTEN, which was setup in DU145 (PTEN positive) and PC3 (PTEN negative) xenograft-derived formalin fixed, paraffin embedded (FFPE) slides. Sections were counterstained with hematoxylin.

With this protocol, tumors with known PTEN status (described above) were correctly identified (Fig. 3A and B). We also stained human biopsies consisting of benign hyperplasias and prostate cancer. We could identify PTEN positive epithelia in the hyperplasia cases as well as prostate cancer biopsies with and without detectable PTEN immunoreactivity (Fig. 3C). Of note, we observed that often the stromal component exhibited greater PTEN expression that the adjacent epithelial tissue (see asterisks in Fig. 3). In summary, we present here a protocol that is valuable for the detection of PTEN in human specimens for research purposes.

Fig. 3

An immunostaining protocol for PTEN in human prostate cancer specimens. (A and B) Representative immunohistochemical images (200×) of PTEN expressing (DU145) and PTEN deficient (PC3) human tumor xenografts. Asterisks indicate stromal cells. (C) Representative micrographs (200×) of PTEN staining in benign hyperplasia tissue (BPH) and prostate cancer (PCa) biopsies with PTEN high and low immunoreactivity, arrows indicate epithelial cells and asterisk depict stromal area.

Extracellular vesicle isolation from urine samples of prostate cancer patients

Due to the close proximity of the prostate to the urinary track, urine-mediated diagnosis of prostate cancer has remained an attractive concept. Extracellular vesicles (EVs) have been described to contain mRNA, protein and metabolites that could be selectively loaded [18]. Importantly, EVs have been identified in urine and cancerous alterations in the bladder have been shown to impact on their composition, suggesting that they could serve as a source for non-invasive biomarker identification. Since current non-invasive prostate cancer biomarkers have been proven to have limitations [19-21], urine EVs might provide a future source of novel biomarkers. Here, we describe the current protocol for urine EV isolation we are employing (a setup carried out by the group of Dr. Falcón-Pérez).

Ultracentrifuge.

Urine EVs can be isolated through this methodology starting from 50 mL of urine. Urine is centrifuged in a tabletop centrifuge at 3000 rpm for 5 min and the supernatant is filtered (0.22 micra) at the moment of collection, and then frozen at −80 °C. At the time of processing, urine is subjected to a first centrifugation of 11,500g for 30 min, and the supernatant is subjected to a second centrifugation of 118,000g for 90 min. The pellet (containing EVs) is then collected, resuspended in 150 μL of cold PBS and frozen for later processing. The EV pellet is subjected to RNA extraction, for which purpose we employ the miRCURY RNA isolation kit (EXIQON, following manufactureŕs instructions, DNAse I – Qiagen – digestion) and we carried out the retrotranscription with SuperScript III (Invitrogen). 35 μL of total RNA is isolated, and despite the low yield of RNA in the preparation (in the range of nanograms), 60–80 μL of cDNA can be prepared for qPCR analysis (Fig. 4A and B). As proof of concept of the validity of this method, we have carried out qPCR analysis in 10 benign hyperplasias and 13 prostate cancers (paired samples to the biopsies presented in the histochemical analysis). We have used as positive control a gene known to be present in EVs, GAPDH [22] (Fig. 4C).

Fig. 4

A method to harvest RNA from urine EVs. (A) Experimental procedure of the EV isolation from urine samples. (B) Representative image by cryo-Transmission Electron Microscopy (TEM) of the isolated EVs with this approach (scale represents 100 nm). (C) Abundance of PTEN (with two probes) and GAPDH transcript in urine EVs by real time quantitative PCR.

PTEN has been recently reported to be secreted [23,24], and PTEN protein abundance in blood exosomes has been suggested to reflect status of the tumor suppressor in the prostate tumor ([25]. Hence we sought to ascertain to which extent the transcript abundance of PTEN would be altered in urine EVs from prostate cancer patients. The results revealed that both PTEN and GAPDH were present in all EV preparations analyzed at a similar abundance regardless of the benign of the tumoral status. This result was in discordance with PTEN protein expression, since the urine samples analyzed include cases that we identified as negative for PTEN immunoreactivity (displayed in Fig. 3). This lack of differences could be due to two main factors: first, the content of EVs in urine might be strongly influenced by bladder cells, perhaps more than by prostate cells. Second, PTEN is down-regulated at multiple levels, through mutations, deletions, but also through post-transcriptional regulation, which would not necessarily impact on the transcript levels.

Discussion

In this methods manuscript, we present approaches that allow us to study the biology of prostate cancer. While much work remains to be carried out in order to understand the molecular changes in this disease, we believe that the technological improvements that we present herein could serve as the basis to ensure the acquisition of (i) fresh and well diagnosed prostate cancer tissue, (ii) RNA of high quality for OMIC studies, (iii) immunostaining methodology to ascertain the expression of PTEN in human tissues and (iv) isolation of urine EVs for molecular studies.

The interaction between pathologists, uro-oncologists and basic scientists is fundamental in order to reach clinically relevant conclusions in prostate cancer research. The fresh tissue preparation procedure that we present has proven to be sustainable in a hospital with biobanking support and, importantly, to preserve the integrity of the surgical material for diagnostic purposes. Unpublished evidence also suggest that the area/volume ratio of the biopsy is directly proportional to the quality of the RNA obtained, and it is therefore plausible that the dimensions of these punch biopsies will allow molecular studies of the highest quality requirements. It is worth noting that the surgical material in our studies was obtained from robotic surgeries, where the warm ischemia period (the time the surgical piece stays excised and inside the patient) is of 60–80 min, while the cold ischemia (the time elapsed from the extraction of the piece to the snap-freeze of the punch biopsy) is at least of 30 min. These ischemic periods do not alter the RNA quality of the biopsy (which we consider a good readout of tissue integrity) and can be achieved in any urology and pathology service.

Importantly, the molecular studies described herein can greatly benefit from the analysis of public databases. In the recent years, bioinformatic platforms have been developed in order to aid in the analysis of publicly available genomic, epigenomic, transcriptomic and proteomic studies. These platforms now allow quickly browsing through tens of studies (which imply thousands of samples) looking at a gene or pathway of interest. Two outstanding examples of this effort are Oncomine (www.oncomine.org) [26] and cbioportal (www.cbioportal.org) [27,28]. These sites allow the researcher to get information about the status of a gene or genes of interest in a given cancer, the mutational landscape throughout different cancers, the epigenetic modifications regulating its expression and the clinical variables associated with its expression. Therefore, these platforms can serve both as a discovery starting point or a clinical validation end point. In summary, a good balance between experimental approaches with human cancer specimens and data mining studies can maximize the relevance of the conclusions met by the researcher.