Background
Prostate cancer (PCa) is the second most common cancer in men and will account for approximately 28,170 deaths in 2012 [
1]. The behavior of some cancers can be variable despite Gleason score and other clinicopathologic factors. Recent efforts have focused on developing biomarkers that provide clinicians with the improved ability to identify clinically significant cancers and aid in treatment decisions. Genes important in embryogenesis frequently play a role in cancer [
2]. One such gene, heterochromatin protein 1 gamma (HP1γ), is critically involved in chromatin packaging [
3] and demonstrates altered expression during development and cell differentiation [
4‐
6].
HP1γ, along with HP1α and HP1β, belong to a family of heterochromatin proteins with stabilizing functions. The structure of HP1 contains an evolutionarily conserved chromodomain that binds to a methylated lysine on histone H
3 (H3K9me) [
7]. This binding epigenetically marks regions of silenced or reduced gene expression [
8]. HP1 stabilizes telomeric and centromeric heterochromatin structure and facilitates DNA repair after disruption or damage [
9,
10]. This chromatin repair function is important to maintain information encoded in the epigenetic histone code [
9]. The HP1 isoforms appear to have differential functions. HP1γ has the ability to regulate both heterochromatin and euchromatin structure, while the other isoforms are localized only on heterochromatin.
Decreased expression of HP1γ occurs during cell differentiation which is unique to this isoform compared to HP1α and β [
6]. In differentiated human tissues levels of HP1γ are generally not detectable [
11]. Silencing of HP1γ has functional consequences. In selected cancer cell lines, growth is inhibited. In a recent screen of cancer tissues, HP1γ protein appeared to be overexpressed in samples of lung, breast, colon and esophageal [
5]. Although decreased levels of its isoform HP1α were found in metastatic breast cancer compared to localized [
12], HP1γ has not been previously evaluated as a marker for cancer progression.
Herein, we identify a novel overexpression of HP1γ in PCa tissues using a tissue microarray (TMA) and VECTRA™ image acquisition technology. This quantitative tool permits localization of expression to nuclear versus cytoplasmic compartments. We report that nuclear and cytoplasmic HP1γ is progressively increased in HGPIN, PCa, and metastatic PCa cores compared to benign tissue. Furthermore, increased HP1γ is correlated with Ki-67, a protein known to complex with other heterochromatin family proteins [
13]. Finally, we report a novel prognostic role for cytoplasmic HP1γ in independently predicting PSA recurrence following prostatectomy. The use of high-throughput imaging technologies such as VECTRA allows for quantitative compartmentalization of marker expression to predict cancer outcomes and lends further insight into tumor biology.
Methods
Tissue microarray
The University of Wisconsin Institutional Review Board (IRB) provides ethical insight to clinical projects and reviews all human research protocols in accordance with federal regulations, state laws, and local and University policies. FFPE-patient tissues were obtained from the University of Wisconsin Department of Pathology and Laboratory Medicine under IRB approval. A tissue microarray was constructed using tissues from 64 PCa patients (mean age, 62.8 years) collected from 1995 to 2006. The mean follow-up period for this cohort was 9.6 years. The TMA consists of 336 duplicate cores from different disease groups: 43 localized PCa (pT2), 30 aggressive PCa (pT3 and 4), 21 metastatic PCa, 23 high grade intraepithelial neoplasia (HGPIN) (tissue from HGPIN tissue blocks of some of the PCa patients of this cohort) and 48 benign prostate tissues (BPT) (from the non-tumor blocks of some of the cancer patients in this cohort).
Staining
Slide preparation and antigen retrieval were conducted as previously described [
14]. Briefly, the TMA slides were taken through routine deparaffinization and rehydration, pretreated with endogenous peroxidase block and retrieval buffer. Slides were then rinsed with dH2O, then Tris Buffered Saline (TBS), then TBS with Tween (TBST), followed by protein blocking at room temperature. E-cadherin antibodies (Cell Signaling Technology, Beverly, MA) were used to define the epithelial compartment for better tissue segmentation. Slides were then stained with antibodies against HP1γ (EMD Millipore, Bilerica, MA) [
15] and Ki-67 (Abcam, Cambridge, MA).
Image analysis
For automated image acquisition and analysis, the stained slides were loaded onto the slide scanner. Slides were scanned as previously described [
14]. Cores with <5% epithelial component or loss of tissue were excluded from the analysis. Nuance system and inForm 1.2™ software (Caliper Life Sciences, Hopkinton, MA) were used to for building spectral libraries on a per-cell basis for HP1γ and Ki-67 target signals according to manufacturer’s protocols. This system allows automated quantitation of fluorescent staining on a per-cell basis and selection of cellular subsets (nucleus versus cytoplasmic) for analysis of target signals.
Statistical analysis
Nuclear and cytoplasmic expression of individual cores of various prostate tissues (benign, HGPIN, PCa, metastatic PCa) was statistically compared using the t-test. Correlation analysis was used to assess the relationship between HP1γ and Ki-67 expression in each compartment. We then assessed the relationship between HP1γ expression and patient clinicopathologic features using t-tests and analysis of variance (ANOVA). For this analysis, multiple PCa cores obtained from the same patient were first averaged, providing a more precise estimation of core expression in each biological replicate. HP1γ expression was then compared with clinicopathologic information (PSA, Gleason, pT stage, tumor volume, SV involvement, margins, extracapsular extension, and evidence of metastasis) collected from the 64 patients included in this study. Univariate and multivariate Cox regression analyses were used to assess whether nuclear and cytoplasmic HP1γ (as continuous variables) predicted biochemical recurrence following prostatectomy. Statistically significant variables from univariate Cox analysis were then entered into a multivariate Cox regression model. Variables that were not independent predictors of recurrence were removed using backward selection (P < 0.05). SPSS Version 20.0 (IBM, Armonk, New York) was used to perform statistical analyses. All tests were two-tailed and a P value < 0.05 was considered statistically significant.
Discussion
Patients and clinicians are in need of more accurate biomarkers to predict the prognosis of prostate cancer, especially for intermediate grade tumors [
18]. Few markers have been reported that reliably predict treatment failure (e.g. PSA recurrence after surgery). Altered nuclear shape and size are a hallmark of cancer and reflect changes in chromatin structure. Image analyses of heterochromatin content and nuclear shape have been reported to improve the prediction of PCa prognoses [
19]. HP1γ is a member of the HPI family of proteins involved in chromatin packaging and gene regulation. The current study represents the first in depth analysis of HP1γ expression for any cancer. We observe an increase in HP1γ expression in the majority of prostate cancers. Furthermore, by analyzing staining in distinct cellular compartments in a cohort of patients with extended follow-up, we find a role for the expression of this gene in predicting treatment failure.
A unique strength of this study lies in our use of the newly validated VECTRA™ imaging technology. The platform merges automated slide scanning, multi-spectral imaging technology, and pattern recognition software into a system for biomarker analysis. VECTRA™ has many advantages over other imaging systems for biomarker quantitation [
14]. VECTRA™ allows researchers to objectively analyze expression within separate epithelial and stromal tissue compartments. Nuclear and cytoplasmic staining within these compartments is concurrently assessed and multiple markers may be addressed simultaneously. The system is highly automated and uses pattern recognition to measure expression on a per-cell, core, or compartment basis on TMAs. The algorithm designed to segment tissue compartments by our genitourinary pathologist [WH] had a 97% rate of acceptable tissue segmentation. In the current analysis, VECTRA™ permitted a reproducible assessment of HP1γ, and its association with Ki-67, in multiple cell compartments for a large cohort of patients.
HP1γ and the other HP1 family members have been primarily studied during embryogenesis and development, including that of the prostate [
4]. Expression of HP1γ was noted in the fetal prostate at 14 and 24 weeks of gestation. This is in contrast to HP1α which is not expressed at those developmental stages [
4]. Decreased expression of HP1γ is required for normal cellular differentiation [
5,
6]. Our study demonstrates that expression of HP1γ occurs in 98% of metastatic and 76% of localized PCa tumors compared to 26% of benign tissues. This cut-off was determined by ROC analysis to highlight the marked differences between patients with benign versus cancer. Adjusting this cut point for Hp1-gamma overexpression can increase the sensitivity or specificity depending on its required use. Total expression was associated with higher Gleason score (p = 0.04).
One notable finding was that HP1γ was not only an independent predictor of PSA-recurrence for patients who underwent radical prostatectomy, but was superior to pathological Gleason score using both a multivariate Cox model and a Kaplan-Meier analysis (Table
3 and Figure
4). Few biomarkers to date are able to predict PSA recurrence more robustly than Gleason score. Recently, Cuzick
et al. reported a panel of 31 RNA markers had a more significant
P value compared to Gleason score for predicting PSA recurrence [
20]. One explanation for the robustness of HP1γ is that its prognostic value is not collinearly associated with other clinicopathologic variables (Table
1). Given the role HP1γ plays in gene repression [
3,
21] and development [
4], we postulate that aberrant expression of this protein may be central to the pathophysiology of PCa.
Increased expression of HP1γ in the nucleus compared to the cytoplasm is consistent with data demonstrating this protein is primarily localized to centromeric and telomeric heterochromatin [
3,
8] and euchromatin [
22,
23]. It was of interest to find expression of the protein in the cytoplasm and note that this carried prognostic significance. Its presence in the cytoplasm may represent altered protein processing, or may have some as of yet unknown functional significance. Ki-67 is a nuclear protein that correlates with cell proliferation and is a known marker of PCa progression [
24]. The C-terminal domain has been found to bind to all three members of the HP1 family [
13]. We found a significant association between HP1γ and Ki-67 on a per core basis (Figure
2). Given the more uniform expression of HP1γ across prostate cancer cells when compared to Ki-67, it clearly has other functions independent from Ki-67 in the regulation of heterochromatin.
Conclusion
In conclusion, we demonstrate that HP1γ expression is elevated in PCa and this independently predicts PSA-recurrence for patients more accurately than pathological Gleason score. Use of novel biomarkers to identify men at lower risk for recurrence may reduce the need for unnecessary adjuvant radiation therapy. HP1γ represents a novel marker of prostate cancer progression that will be useful to the clinician with further validation in other datasets. In addition, examination of HP1γ expression in biopsy specimens might fulfill an urgent need for more accurate pre-treatment risk stratification tools. Given the role HP1γ plays in epigenetic gene regulation through its binding to methylated lysine on histone H
3 (H3K9me) [
7], it represents a novel target for cancer treatment. Furthermore, a recent survey of tumors suggests HP1γ may be overexpressed in other epithelial cancers [
3].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JS, MT, and DJ analyzed the results and wrote the paper. MT performed statistical analysis. WH carried out the experiments and operated VECTRA™ technology. DJ designed the study and provided supervision. All authors read and approved the final manuscript.