Does Valproic Acid Induce Neuroendocrine Differentiation in Prostate Cancer?

Valproic Acid (VPA) is a histone deacetylase inhibitor that holds promise for cancer therapy. Here, we investigate whether VPA treatment induces neuroendocrine differentiation of Prostate Cancer (PCa). A tissue microarray of VPA-treated and untreated tumor xenografts and cell lines of human PCa (LNCaP, C4-2, DU145, and PC-3) were generated and were analyzed by immunohistochemical analysis (IHC) for NE markers chromogranin A (CgA), synaptophysin, and NCAM (neural cell adhesion molecule). Western blot analysis for CgA was performed to confirm the results of the TMA. IHC analysis did not reveal any induction of CgA, synaptophysin, or NCAM in any xenograft after VPA treatment in vivo. In vitro, VPA treatment induced little synaptophysin expression in C4-2 and PC-3 cells and NCAM expression in LNCaP and PC-3 cells. In the case of CgA, VPA treatment decreased its expression in vitro in a dose-dependent manner, as determined by western blot analysis. Thus our data demonstrates that VPA does not induce NE differentiation of PCa cells in the physiologically relevant in vivo setting.


Introduction
Histone acetylation and deacetylation by histone acetyl transferases and histone deacetylases is involved in the epigenetic regulation in human cells [1,2]. Recently, this posttranslational modification has become a popular molecular target for cancer therapy. HDAC inhibitors (HDACIs) have demonstrated significant antitumor activity by hyperacetylation of nucleosomal histones resulting in reexpression of repressed genes that produce growth arrest, terminal differentiation, and/or apoptosis in carcinoma cells [3]. Valproic Acid (VPA), an HDACI and an antiepileptic agent, causes marked decrease in proliferation of Prostate Cancer (PCa) cells in vitro and significant reduction in tumor volume in vivo [4,5]. Multiple pathways including cell cycle arrest, apoptosis, angiogenesis, and senescence contribute to the antitumor effects of VPA.
Neuroendocrine (NE) cells are the third and minor epithelial cell type in prostate, in addition to the more abundant luminal secretory cells and basal cells [6]. NE cells have dual properties of neurons and endocrine cells and are believed to be involved in the regulation, secretion and differentiation of other prostatic cells [7]. Conventional adenocarcinoma with focal NE cells represents the most common type of PCa. Small cell PCa and prostatic carcinoid are relatively rare and are considered pure NE tumors with a poor prognosis [8]. Neuroendocrine differentiation thus has been suggested as a poor prognostic sign by some authors, but the exact role of NE differentiation of the prostate remains unclear, and its prognostic importance in prostate cancer still remains controversial [7,9]. The characteristics of NE differentiation in PCa are very much similar to those seen in patients who develop this histologic phenotype in non-small-cell lung cancer [10]. NE cells in prostate express NE markers such as Chromogranin A (CgA), synaptophysin, B-tubulin, neural cell adhesion molecule (NCAM or CD56), neuron specific enolase (NSE), and so forth. NE cells can be generally identified by electron microscopy or immunohistochemical (IHC) staining with antibodies for NE markers [11].
Recently, some studies have documented increased neuroendocrine markers after in vitro treatment of prostate cancer cell lines with HDACIs [9,12] indicating neuroendocrine transdifferentiation. In contrast, studies done in neuroendocrine tumors such as carcinoid, pheochromocytoma, and small cell lung cancers have shown VPA and other HDACIs to exert antitumor effects [13][14][15]. VPA has been shown to promote apoptosis, reduce NE phenotype and expression of NE markers, and is suggested as a promising therapy for these tumors [16,17]. Thus the role of HDACI's in neuroendocrine differentiation still remains unclear and has thus warranted further investigation.
The goal of this study is to carefully determine whether VPA induces NE differentiation in the PCa cell lines, in vivo and in vitro, by studying a variety of markers associated with NE differentiation in numerous PCa cell and tumor models. Markers including CgA, synaptophysin, and NCAM were quantified by IHC in a tissue microarray (TMA) format from several VPA-treated human PCa cells grown in vitro and in vivo as tumor xenografts in nude mice.

Establishment of Tumor Xenografts.
Cell lines were grown to 80% to 90% confluent and harvested. Cells were resuspended in 1x phosphate-buffered saline (pH 7.4; BioSource, Rockville, MD), mixed 1x with Matrigel (BD Biosciences, Palo Alto, California), and injected (1 × 10 6 per injection) subcutaneously into the lateral flanks of male athymic nu/nu mice. Once palpable tumors were established, animals were randomized into control and treatment arms.  [4] comparable to FDA approved levels in humans [18]. Animals in treatment arm were treated for 35 days before excision of tumors. In in vivo experiments, chronic treatment implies to long-term treatment with regards to life span. We considered 35 days of treatment in nude mice (with life span of 1 year approx. in our experience) as long-term treatment. This period would correspond to years of treatment in humans.

TMAs: Construction and IHC Staining.
For in vitro models, cells were harvested and washed in PBS. Resulting cell pellets were incubated for 1-2 hr in Bouin's fixative (75% saturated picric acid, 20% formalin (40%), 5% acetic acid, rinsed with 70% ethanol, and dehydrated according to standard procedures with ethanol and xylene. Cell pellets > 5 mm were split in order to achieve sufficient dehydration. Cells were embedded in paraffin following 90 min of incubation in liquid paraffin at 60 • C. For in vivo models, tumors were excised and portioned on day 35. Portions were either immediately frozen in liquid nitrogen and stored at −80 • C or fixed in buffered formalin and subsequently embedded in paraffin.
A tissue microarray (TMA) of the paraffin embedded materials was generated as described previously [19]. Each array block also contained control normal human prostate tissues and animal xenograft tissues such as bladder, kidney, lung and spleen. Immunohistochemical stains for were performed separately on sections cut from the TMA. Stained TMA slides were scanned (at 20x magnification setting) using the APERIO imaging system and the images were uploaded and viewed using TMAJ [20,21]. Each array spot was then formed into a composite image for viewing and scoring on a personal computer monitor.
2.6. Scoring of IHC Staining. IHC specimens were provided to pathologist for scoring. The identity of sample and treatment was blinded to the scorer. The specimens showed a varying degree of staining intensity and percentage of cells staining. Therefore, a combined intensity and percentage positive scoring method was used [22]. Strong intensity staining was scored as 3, moderate as 2, weak as 1, and negative as 0. For each intensity score, the percentage of cells with that score was estimated visually. A combined weighted score consisting of the sum of the percentage of cells staining at each intensity level was calculated for each sample, for example, a case with 50% strong staining, 20% moderate staining, and 10% weak staining would receive a score as follows: (50 × 3 + 20 × 2 + 10 × 1) = 200. The maximum score is 300.

Statistical Analysis.
Analysis was done using GraphPad Prism 4.0. Data is plotted as means(SE). One way ANOVA with post-hoc testing was done to evalauate differences in mean staining score between different groups.

In Vitro.
The most reliable method to assess NE differentiation in PCa is the detection of CgA in tumor cells. IHC staining for CgA and two other neuroendocrine markers synaptophysin and NCAM were done.

Chronic VPA Treatment Reduces CgA Expression.
The IHC staining of sections constructed of cell lines treated for 14 days with VPA was not able to detect any expression of CgA (Figure 1(a)). In order to verify the absence of CgA by the IHC in all the cell lines tested, we performed Western blot analysis (Figure 1(b)). Results revealed CgA expression in these cell lines; however, CGA protein levels were reduced in a dose-dependent manner after VPA treatment. Histone 3 acetylation confirmed VPA activity.

Chronic VPA Treatment In Vivo Does Not
Induce the Expression of CgA, Synaptophysin, or NCAM. Unilateral tumor xenografts were established in 20 animals each for every cell line. Half of the animals were randomized to receive 0.4% VPA in drinking water. We have shown previously that administration of 0.4% VPA in mouse drinking water can achieve plasma VPA levels similar to the levels obtained in human patients [4]. VPA treatment at these levels was shown to induce acetyl H3K9, p21, and reduce tumor volume, thus confirming the pharmacologic activity of VPA [4]. Animals were sacrificed, and tumors were harvested on day 35. To investigate the effects of VPA on NE markers of PCa tumors in vivo, we evaluated expression of NE markers by IHC on the excised tumors. IHC did not reveal any CgA  The tissue specimens of human pancreas and human prostate were positive internal controls for CGA staining, while another human prostate section served as a negative control. (b) Synaptophysin staining in prostate xenografts animals treated with or without VPA. A: Representative images of synaptophysin staining for xenograft sections from control and VPA-treated groups: LNCaP, C4-2, PC-3, and DU-145 (Scanned at 20X magnification using the APERIO imaging system). The tissue specimens of human pancreas and mouse colon were positive internal controls for synaptophysin staining, while mouse bladder as negative internal controls. (c) NCAM staining in prostate xenografts animals treated with or without VPA. (a): Representative images of CD56 staining for xenograft sections from control and VPA-treated groups: LNCaP, C4-2, PC-3, and DU-145 (scanned at 20X magnification using the APERIO imaging system). The tissue specimens of human colon carcinoid and normal human prostate were positive internal controls for NCAM staining, while mouse bladder was a negative control. staining in either treatment or control arms in all cell lines (Figure 4(a)). TMAs from C4-2 tumors revealed decreased expression of both synaptophysin (mean weighted score 47(±10) versus 15(±5), P < .001) and NCAM (44(9) versus 5(6), P = .002) in treatment arms (Figure 4(b)). None of the other arms revealed any significant staining (weighted scores less than 30) for NCAM or synaptophysin (Figures 4(b) and 4(c)). Thus VPA does not induce any NE markers in the physiologically relevant in vivo setting.

Discussion
NE cells are considered to be derived from local stem cells and are an example of normal, terminally-differentiated cells without proliferative activity [6]. NE cells in tumor lesions are phenotypically similar to NE cells in normal prostate epithelium in terms of expression of neuropeptides and biogenic amines. Furthermore, dual epithelial characteristics such as prostatic acid phosphatase and/or PSA production and NE marker expression, such as CgA, are frequently coexpressed in the malignant phenotype of NE cells [23]. Studies evaluating the role of focal NE differentiation in PCa prognosis have reported varied results: some reports indicate a negative correlation with prognosis while some show little or no relationship to prognosis [10,[23][24][25][26][27][28].
Histone deacetylase inhibitors are a promising new class of cancer therapy which have antiproliferative and prodifferentiation properties. For prostate cancer, it was recently reported that HDAC gene expression is elevated in tumors. Moreover, high expression levels of HDAC2 were associated with poor prognosis [29]. Thus VPA, which is capable of inhibiting HDACs classes I and IIa, may be a good option for PCa therapy. In preclinical models, VPA treatment leads to proliferation arrest and differentiation and apoptosis of cancer cells of various tissue origin, while nominal effects were reported in normal cells [2,4,5]. However, Valentini et al. reported VPA to cause an increase in the secretion of NSE in LNCaP cells (in vitro) which may indicate an NE differentiation [12]. In order to better understand the role of VPA in possibly stimulating NE differentiation in PCa cell models, we selected the clinically recommended panel of antibodies for the IHC investigation of NE cells in multiple PCa cell lines and xenograft tumors.
Chromogranin A or parathyroid secretory protein 1 is a member of the chromogranin-secretogranin family and forms the major constituent in neurosecretory peptide containing dense core granules in NE cells. CgA is highly expressed by cells of neuroendocrine origin, both normal and tumoral, functional and nonfunctional. While Neuron Specific Enolase (NSE) has also been used as an NE marker, it is known to be expressed in a variety of non-NE cells and tumors, which has led researchers to question its specificity [30,31]. Serum CgA levels, on the other hand, have been reported to be better predictors of neuroendcocrine differentiation than NSE [32,33]. Thus, CgA now is widely regarded as an excellent and more specific marker of NE differentiation. In our study, CgA expression was not detected by IHC in either control or treated groups in human prostate cell lines of LNCaP, C4-2, PC-3, and DU145 in vitro or in vivo. Western blotting, being more sensitive, revealed low CgA expression in these cell lines; which reduced further with VPA treatment in a dosedependent manner. Histone acetylation and p21 induction (data not shown) confirmed that active VPA doses were achieved as we have previously demonstrated [5,34]. Our results are further corroborated by similar reduction in NE markers and NE morphology in NE tumors after treatment with HDACIs [13,14,16]. These studies in fact report apoptotic effects of VPA on neuroendocrine cells. Yu et al. have previously demonstrated CgA to be an important neuropeptide promoting the growth of prostate cancer cells and its suppression leading to programmed cell death in multiple prostate cell lines [35]. Gong et al. later found antiapoptotic effects of CgA to be dependent on a Protein Kinase B/Akt (an antiapoptotic protein or prosurvival factor) mediated pathway [36]. Also HDACIs have been known to downregulate Akt phosphorylation in prostate cancer cells [37] Taken together, it suggests a link between HDACI'smediated apoptosis and CgA inhibition. Further studies will be required to determine the contribution of CgA and Akt to the VPA therapeutic effect.
Synaptophysin and NCAM are other specific and fairly sensitive markers for NE differentiation [38,39]. In in vitro experiments, these markers showed varying trends (increased synaptophysin staining in C4-2 and PC-3 cells but unaltered in LnCaP and DU145; increased NCAM in LnCAP and PC-3 cells but unaltered in C4-2 and DU145), and no consistent pattern was seen. In in vivo experiments, the staining did not reveal any significant expression of these markers in any of the xenografts except in C4-2 tumors where it revealed a downward trend on treatment.

Conclusion
The findings in our study do not support any neuroendocrine differentiating role of VPA. On the contrary, CgA, a very specific marker, was reduced in all studied cell lines, following chronic VPA treatment. Synaptophysin and NCAM showed some inconsistent induction following VPA treatment in some cell lines but, in vivo, VPA treatment did not induce any significant expression of any NE markers. 0.4% VPA in mouse drinking water can achieve plasma VPA levels similar to the therapeutic levels obtained in human patients. TMAs from xenografts of different cell lines either did not stain for NE markers or had very little staining without any induction on VPA treatment. Thus, our data clearly demonstrates that VPA does not induce NE differentiation of PCa cells in the physiologically relevant in vivo setting.