Mitochondrial DNA Mutation Stimulates Prostate Cancer Growth in Bone Stromal Environment

doi:10.1002/pros.20854

Prostate. Author manuscript; available in PMC 2010 January 1.

Published in final edited form as:

Prostate. 2009 January 1; 69(1): 1–11.

doi: 10.1002/pros.20854

PMCID: PMC2753601

NIHMSID: NIHMS75315

Mitochondrial DNA Mutation Stimulates Prostate Cancer Growth in Bone Stromal Environment

Rebecca S. Arnold,^1,² Carrie Q. Sun,¹ Jendai C. Richards,¹ Galina Grigoriev,¹ Ilsa M. Coleman,³ Peter S. Nelson,³ Chia-Ling Hsieh,^4,⁵ Jae K. Lee,⁶ Zhiheng Xu,⁷ Andre Rogatko,⁷ Adeboye O. Osunkoya,^1,^8,¹⁰ Majd Zayzafoon,⁹ Leland Chung,^1,² and John A. Petros^1,^2,^8,¹⁰

¹Department of Urology Emory University School of Medicine, Atlanta GA 30322

²Winship Cancer Institute Emory University School of Medicine, Atlanta GA 30322

³Fred Hutchinson Cancer Research Center, Seattle, Seattle, WA

⁴Graduate Institute for Cancer Biology, China Medical University, Taichung, Taiwan

⁵Center for Molecular Medicine, China Medical University Hospital, Taichung, Taiwan

⁶Department of Public Health Sciences, University of Virginia, Charlottesville, VA 22908

⁷Department of Biostatistics, Emory University School of Medicine

⁸Department of Pathology and Laboratory Medicine, Emory University School of Medicine

⁹Department of Pathology, University of Alabama, Birmingham, AL, 35294

¹⁰Atlanta VA Medical Center, Decatur 30033

Corresponding Author: John A. Petros. 1365 Clifton Rd. Building B Department of Urology Emory University School of Medicine Atlanta, GA 30322 Phone: 404-778-4847 Fax: 404-778-5016 ; Email: jpetros/at/emoryhealthcare.org

The publisher's final edited version of this article is available at Prostate

See other articles in PMC that cite the published article.

Abstract

Background and Objectives

Mitochondrial DNA (mtDNA) mutations, inherited and somatically acquired, are common in clinical prostate cancer. We have developed model systems designed to study specific mtDNA mutations in controlled experiments. Because prostate cancer frequently metastasizes to bone we tested the hypothesis that mtDNA mutations enhance prostate cancer growth and survival in the bone microenvironment.

Methods

The pathogenic nucleotide position (np) 8993 mDNA mutation was introduced into PC3 prostate cancer cells by cybrid formation. Wildtype and mutant cybrids were grown as nude mouse subcutaneous xenografts with or without bone stromal cell co-inoculation. Cybrids were also grown in the intratibial space. Tumor growth was assayed by direct tumor measurement and luciferase chemiluminescence. Gene expression was assayed using cDNA microarrays confirmed by real time PCR, western blot analysis and immunohistochemistry.

Results

Cybrids with the 8993 mtDNA mutation grew faster than wildtype cybrids. Further growth acceleration was demonstrated in the bone microenvironment. A thirty-seven gene molecular signature characterized the growth advantage conferred by the mtDNA mutation and bone microenvironment. Two genes of known importance in clinical prostate cancer, FGF1 and FAK, were found to be substantially upregulated only when both mtDNA mutation and bone stromal cell were present.

Conclusions

The ATP6 np 8993 mtDNA mutation confers a growth advantage to human prostate cancer that is most fully manifest in the bone microenvironment. The identification of specific molecular alterations associated with mtDNA mutation and growth in bone may allow new understanding of prostate cancer bone metastasis.

INTRODUCTION

The most common cancer occurring in men in the United States is prostate cancer. Metastatic prostate cancer, and in particular metastasis to the bone, is an important clinical problem in terms the number of men affected and the negative impact it has on the quality of life and survival. The propensity of prostate cancer to metastasize to the bone is not well understood but specific interactions between the prostate cancer cells and cells present in the bone microenvironment are thought to play an important role.

Evidence has been accumulating over the past decade demonstrating a role for mtDNA mutations in various types of cancers. A variety of mtDNA mutations, both inherited and somatic, have been reported in renal and colon cancer [1,2], prostate cancer [3–5], breast cancer [6] and an array of other solid tumors [6,7]. The induction of lesion-associated mutations in the hypervariable regions of mtDNA in prostate cancer and preinvasive lesions is indicative of active mtDNA mutation in these regions [4]. Recently, a study of prostate cancer, comparing tumor, adjacent benign, and distant benign tissue, suggested that mtDNA mutations could be an early indicator of malignant transformation. Frequent somatic mutations were demonstrated in prostate cancer with both tumor and normal appearing benign glands harboring somatic mtDNA mutations. Parr and co-workers concluded that mtDNA mutations appeared to demonstrate a progressive pattern of malignant disease [8]. Additionally, analysis of clinical prostate cancer comparing mtDNA from prostate tissue and peripheral blood demonstrated an increased frequency of mutations in the functionally significant cytochrome oxidase I (COI) gene. Inhibition of CO1 inhibits oxidative phosphorylation and increase reactive oxygen production [3].

The earliest described mtDNA mutation that was linked to a disease was a mutation at position 8993 in the mtDNA gene that codes for the ATPsynthase subunit 6. The mutation termed T8993G resulted in an amino acid alteration at a highly conserved leucine at position 156 in the protein to an arginine. While the mutation did not appear to alter ATP hydrolysis, it did decrease ATP synthesis in cells [9]. In addition, in a study utilizing cybrid cell lines generated by fusion of wild type or T8993G with 143B osteosarcoma cell line that lacked a functional mitochondria, Wallace and coworkers found that mitochondria isolated from the T8993G cells had reduced state III respiration [10]. The mutation also results in an increase in mitochondrial ROS production [3].

The ability to study the biological consequences of mutations in the mtDNA is hampered by the lack of appropriate cell lines in which to study single point mutations. Ideally, two cell lines are needed that vary only in the mutation of interest in the mtDNA. The T8993G mutation was used to model a mutation found in a prostate cancer specimen, C8932T, which is 20 amino acids away from the 8993G mutation. Cybrid cell lines with the wildtype and 8993G mutations were produced through fusion of mitochondrial donor cells with the prostate cancer cell line PC3 that had been depleted of its mitochondrial DNA. This fusion resulted in the production of two different cell lines containing identical nuclear backgrounds and differing only the base substitution T8993G [3]. Analysis of these cell lines demonstrated a dramatic increase in growth of the 8993G harboring cell line when compared to wildtype when cells were grown as mouse xenografts. This increased tumor growth directly correlated with increased ROS production of the cell lines [3].

The overall rationale for this study is to understand how mtDNA mutations occurring in prostate cancer affect bone metastasis. We previously have established unique cell models of prostate cancer that allow us to compare pairs of prostate cancer cells (cybrids) that differ by only one base in the mitochondrial genome. This allows us to ascribe the differences we see in gene expression, cell biologic parameters and the essential markers of the malignant phenotype (such as proliferation and tumorigenesis) to the mtDNA mutation under study. Using this approach we discovered that mtDNA mutations in prostate cancer mediate a unique interaction between prostate cancer cells and the bone microenvironment. The growth advantage conferred by mtDNA mutations is most fully realized in the bone stromal microenvironment and this effect has been demonstrated in two model systems: co-inoculation of stromal and epithelial cells subcutaneously in nude mice and the direct intratibial injection of mutant or wild type cancer cells. A unique gene expression signature of this interaction has been identified with direct relevance to prostate cancer bone metastasis including cell invasion, lysis of the extracellular matrix, cancer cell motility, migration and spreading and intracellular signaling pathways that regulate cell cycle, survival and apoptosis.

MATERIALS AND METHODS

Cell lines

The following cell lines were used in this study. Cybrid cells lines with a PC3 nuclear background and containing either mitochondria that were wild type (8993WT) or contained a mutation at position 8993 in the mitochondrial genome (8993Mut) [3]. Cybrid cell lines were maintained in RPMI containing 10 % heat inactivated FBS and supplemented with 2.4 mg/mL glucose (Sigma, St. Louis MO), 60 ug/mL uridine (Sigma, St. Louis, MO), and 1 mM sodium pyruvate (Invitrogen, Carlsbad, CA). HS27a and HS5 cells obtained from ATCC are bone marrow stromal cell lines developed from the same patient. The HS27a and HS5 cells were maintained in RPMI containing 10 % heat inactivated FBS or DMEM containing 10 % heat inactivated FBS respectively.

Luciferase transfection of cybrid cell lines

Cybrids were stably transfected with luciferase reporter for use in intratibial injection experiments. The luciferase-tagged 8993Mut and 8993WT cybrids expressing luciferase were generated by transfecting the cells with a retroviral vector containing a luciferase gene driven by a LTR promoter [11]. The integration of the transgene into cells was stabilized by G418 selection. All experiments were performed with pooled populations with high luciferase expression.

Mice

Male nu/nu mice, 6–8 weeks old, were purchased from Charles River Laboratories (Wilmington, MA) and housed in ventilated cages under sterile conditions. All subsequent mouse experiments were performed using IUCAC approved protocols.

In Vivo Tumor Study – Subcutaneous Injection

Mice were injected subcutaneously in the neck with 1) 8993WT; 2) 8993Mut; 3) 8993WT + HS27a; 4) 8993WT + HS5; 5) 8993Mut + HS27a; 6) 8993Mut + HS5 for 4, 5, 6, 5, 5 and 5 mice respectively. 10⁶ cells, resuspended in PBS, were injected per mouse. Mice were checked daily for tumor growth. When tumors were visually observed they were measured twice a week and the tumor volumes calculated using the formula [length X (width)²]/2 [12]. A second experiment was performed with 1) 8993Mut; 2) 8993Mut + HS27a; 3) 8993Mut + HS5 for 8, 10, and 10 mice respectively and tumors were monitored and measured in the same way as the first experiment. Mice were sacrificed when the tumors reached 10% of body weight. Tumor tissues were dissected for further study.

In Vivo Tumor Study - Intratibial Injection

Wild type and mutant prostate cancer cybrids were transfected with luciferase. All experiments have been performed with pooled populations with high luciferase expression. Nude mice were anesthetized and a tuberculin syringe with 10⁴ cybrid cells (mutant or wild type at base of interest) was inserted into the marrow space of the tibia through the knee joint. Animals were injected in cohorts of 5 mice (8993WT) and 6 mice (8993Mut). Development of tumor growth in bone was monitored by bioluminescent imaging of luciferase-expressing cybrid cells (Living Image ver. 2.20; Xenogen Corp. Hopkinton, MA) and whole body X-ray. For detection of luciferase-expressing cells, 125 mg/kg of D-luciferin sodium solution (Xenogen Corp.) was injected intraperitoneally 10 minutes before photon recording began. Total image intensity was collected every 5 minutes until a high plateau was reached and this value was recorded and images obtained. The bioluminescent signal was quantified by measuring the amount of highlighted pixels in the area shaped around each site of photon emission with the aid of the software (Living Image ver. 2.50; Xenogen Corp.).

Statistical Analysis of In Vivo Tumor Study

Growth curve models were used to find appropriate growth functions for the data. Mixed models for repeated measurements were performed to determine whether there were significant differences between the treatment groups. Log transformation was used to attain homoscedasticity. Least-squares means were computed and Bonferroni method was employed in the a posteriori multiple comparison. All fixed effects were significant predictors (defined as p-value<0.05 of tumor size).

RNA preparation

RNA was prepared from the different mouse tumor types. Total RNA was isolated from the mouse tumor samples using TRIzol (Invitrogen) followed by GenElute RNA columns (Sigma) according to the manufacturer's protocols.

DNA Array Experiments

To provide a reference standard RNA for use on two-color cDNA microarrays, we pooled equal amounts of total RNA isolated from LNCaP, DU145, PC3, and CWR22 cell lines (American Type Culture Collection, Manassas, VA) growing at log phase in dye-free RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) (Invitrogen). Reference RNA was purified using TRIzol following the manufacturer's protocol. Reference RNA was then further purified by RNeasy maxi kit (Qiagen Inc, Valencia, CA) and treated with DNAse using the RNase-Free DNase Set (Qiagen Inc) Experimental and reference total RNA samples were amplified one round using the Ambion MessageAmp aRNA Kit (Ambion Inc, Austin, TX), and sample quality and quantity were assessed by agarose gel electrophoresis and absorbance at A260. cDNA probe pairs were prepared by amino-allyl reverse transcription using 2 μg of amplified RNA and labeling with Cy3-dCTP (experimental) or Cy5-dCTP (reference) fluorescent dyes (Amersham Bioscience, Piscataway, NJ.) Custom Prostate Expression Database (PEDB) cDNA microarrays were constructed and hybridized as previously described [13] using 6,760 clones derived from the Prostate Expression Database, a sequence repository of human prostate expressed sequence tag data available to the public (www.pedb.org.). Fluorescence array images were collected for both Cy3 and Cy5 using a GenePix 4000B fluorescent scanner (Molecular Devices, Sunnyvale, CA), and GenePix Pro 4.1 software was used to grid and extract image intensity data. Spots of poor quality, as determined by visual inspection, were removed from further analysis. Normalization of the Cy3 and Cy5 fluorescent signal on each array was done using Silicon Genetics GeneSpring 7.3 software (Agilent Technologies, Santa Clara, CA.) A print tip-specific Lowess curve was fit to the log-intensity versus log-ratio plot and 20.0% of the data was used to calculate the Lowess fit at each point. This curve was used to adjust the control value for each measurement. If the control channel was lower than 10, then 10 was used instead. Normalized log2-ratios from the replicated cDNA spots on each PEDB chip were averaged and used for subsequent analysis. Data were filtered to remove values from poorly hybridized cDNAs with average background subtracted intensity levels <300.

Statistical Analysis on cDNA Microarray Studies

Two samples in each array were highly correlated in our cDNA microarray data, a common finding in two-color cDNA microarrays. The sample size of each condition of the microarray experiment was two or three, therefore, we used the paired local pooled error (paired LPE) test [14,15], which is useful in identifying differentially-expressed genes with low-replication microarray data. In brief, intensity differences between each paired experimental condition were calculated for all genes. Then, LPE variance estimates were formed by pooling variance estimates for genes with similar expression intensities from replicated arrays at all the ranges of the above difference intensity. The paired LPE approach controls the situation where a gene with low expression may have very low variance by chance and the resulting signal-to-noise ratio is unrealistically large. Statistical significance of the paired LPE-based test was evaluated as follows. First, each gene's medians m1 and m2 under the two compared conditions were calculated to avoid artifacts from outliers. The paired LPE statistic for the median (log-intensity) difference z was then calculated as:

where s_LPEpooled is the pooled standard error from the LPE-estimated baseline variances from the two conditions.

cDNA preparation and Real Time PCR

cDNA was obtained from the RNA by reverse transcription using Advantage RT for PCR from Clontech (a Takara Bio Company, Mountain View, CA) according to protocol. Quantitative PCR was performed to determine the message levels of the various genes using primers specific for each gene or 18S RNA (see table 1). PCR reaction was as follows: 7.3 μL cDNA, 0.1 μL 10 μM forward primer, 0.1 μL 10 μM reverse primer, and 7.5 μL SYBR Green PCR Master Mix (AB, Foster City, CA). The PCR reaction included an initial cycle of 95 °C for 10 min followed by 45 cycles of 95 °C for 30 sec, 56 °C for 30 sec, 72 °C for 30 sec. All results are reported as the Mean Starting Quantities relative to 18S. A 1.5 % agarose gel was run after PCR and single band for each gene was observed.

Table 1

Primers used for selected candidate gene quantitative real-time PCR.

cDNA was also used to determine if HS5 or HS27a cells were present in tumor tissue from mouse xenografts. According to ATCC, HS27a and HS5 cells were immortalized through transformation of long term bone marrow cells with the amphotropic retrovirus vector LXSN16E6E7 which contains the neomycin resistance gene. Using the Amplitaq Gold Kit (AB), reactions were as follows: cDNA, 1XBuffer, 1.5 mM MgCl₂, 0.2 mM each dNTPs, 150 nM each forward and reverse primers for neomycin resistance gene (NRG) (see table 1) and 5 units of Taq for a final volume of 100 uL. In a separate reaction, 18S primers were used. The PCR reaction included an initial cycle of 95 °C for 10 min followed by 35 cycles of 95 °C for 30 sec, 56 °C for 30 sec, 72 °C for 30 sec with a final cycle at 72 °C for 7 min. A 1.5 % agarose gel was run after PCR. The presence of the neomycin resistance gene was indicated by the presence of a 240 nt band.

Western Blot Analysis

Whole-cell extracts were obtained by lysing cells with lysis buffer containing 50 mmol/L, Tris Base, 5 mmol/L EGTA, 150 mmol/L NaCl, 1 % Triton X-100 (pH 7.4). One tablet of protease inhibitor (Roche Applied Science, Indianapolis, IN) was dissolved in 7 ml of lysis buffer. Total protein 30 μg/well was loaded in 4–12 % gradient NuPAGE MES SDS gel (Invitrogen) and transferred into Immun-Blot PVDF Membrane (BIO-RAD, Hercules, CA). The membrane was immuno-blotted with anti-FGF1 (Santa Cruz, CA) at 1:1000 dilution, anti-FAK (Biosource, Invitrogen) at 1:1000 dilution and anti-β-actin (Sigma) at 1:2000 dilution. Immunodetection was completed by using the corresponding secondary horseradish peroxide-conjugated antibodies (Amersham, Piscataway, NJ). Horseradish peroxide activity was detected using enhanced chemiluminescence from ECL Western Blotting Analysis System (Amersham).

Immunohistochemistry of Mouse Tumor Tissue

Wild type or mutant prostate cancer cybrids were grown as xenografts in nude mice with or without the co-inoculation of bone stromal cells. Samples were stained for FGF1 using a polyclonal FGF1 antibody (Santa Cruz Biotechnologies, Santa Cruz, CA). In brief, slides were deparaffinized in xylene and ethanol and rehydrated in water prior to antigen retrieval by heating in Low pH Target Retrieval Solution (DAKO, Carpinteria, CA). After antigen blocking, rabbit anti-human FGF1 was diluted 1:100 in SuperBlock (Scy Tek Laboratories, Logan, UT) and incubated 2hrs at 20 °C. After washing, slides were incubated with goat anti-rabbit secondary antibody conjugated to horseradish peroxidase (Jackson Immunoresearch, West Grove, PA) at 1:500 for 1 hour at 20 °C, washed and the peroxidase developed for 2 minutes after application of DAB solution (ImmunoVision Technologies, Brisbane, CA).

RESULTS

Xenograft Mouse Model

Recent studies by our laboratory demonstrated that mitochondrial DNA mutations, in particular a mutation in the ATP6 gene (T8993G), play an important role in the development of prostate cancer. To test the hypothesis that prostate cancer cell lines harboring the ATP6 mutation at position 8993 would grow more aggressively in a bone environment, we injected nude mice subcutaneously with PC3 cybrid cells containing either wild-type (T8993T) or mutant (T8993G) mitochondrial DNA in the presence or absence of either HS27a or HS5 bone marrow stromal cells. Figure 1A demonstrates that the 8993Mut cells had increased tumor volume when co-injected with either of the bone marrow stromal cell lines when compared to 8993Mut alone or 8993WT alone or in combination with HS27a or HS5 cells. As early as day 20, the average tumor volume of mice that were co-injected with bone marrow stromal cells was 3-fold (HS27a) or 5-fold (HS5) larger than mice injected with 8993Mut alone. The presence of HS27a cells with 8993WT cells had a modest effect on tumor growth. By the end of the experiment, the fold increase in the tumor volumes of the different experimental models compared to the 8993WT control were: 8993Mut, 9-fold; 8993WT/HS27a, 5-fold; 8993WT/HS5, 9-fold; 8993Mut/HS27a, 17-fold; and 8993Mut/HS5, 27-fold. Mice injected with HS27a or HS5 cells produced no tumors.

Figure 1

Growth of 8993 Cybrid Cells in Mice

8993WT and 8993Mut cells were also grown in culture in the absence or presence of bone stromal cells either using tissue culture inserts or through cell contact. There was a modest increase in growth rate when 8993Mut was compared to 8993WT (doubling time of 1.11 days compared to 1.83 days). The presence of HS27a or HS5 cells either in an insert above the 8993Mut or 8993WT or in contact with the cells had no significant effect on cell growth (data not shown).

Intratibial Mouse Model

To further test the hypothesis that prostate cancer cells containing the 8993 mtDNA mutation grow faster in a bone environment, we injected 8993WT and 8993Mut cells stably transfected with the luciferase gene directly into the tibial bone marrow cavity of nude mice. Figure 1B indicates the growth of the 8993WT (black bars) and the 8993Mut (gray bars) at 4, 8 and 11 weeks after injection. At week 4, both cell lines had similar growth but by week 8 the 8993Mut cells had formed tumors that were much larger than the tumors formed by the 899WT cells. The increased osteolytic lesion due to the faster growing 8993Mut tumor was also observed by x-ray analysis of the mouse bone. In addition, hematoxylin and eosin staining of the bone samples demonstrates a significant increase in the size of the tumors from mice injected with 8993Mut cells (Figure 2). These data support the hypothesis that the cells harboring the mutation at position 8993 in the mitochondria genome have a growth advantage in the bone environment.

Figure 2

8993Mut Cells form Larger Tumors in Bone

DNA Array Analysis

We analyzed the tumors formed through subcutaneous injection for differences in gene expression to determine what genes may play a role in the growth advantage seen in the 8993Mut cells in the presence of bone stroma. The three circle diagram (Figure 3) represents the method we chose to identify those genes whose expression represented the unique molecular alterations associated both mtDNA mutations in the cancer epithelium and interaction with bone stromal cells. The top circle represents those genes that were significantly over- or under-expressed when comparing 8993 mutant cybrids alone (without bone stromal cells) to 8993 wild type cybrids alone. The lower left is the comparison of genes differentially expressed when mutant or wild type cybrids were grown with HS27a bone stromal cells, and the bottom right circle is those genes differentially expressed when mutant or wild type cybrids were grown with HS5 bone stromal cells. This diagram shows that there were 37 genes that were differentially expressed when mutant cybrid xenografts were compared to wild type cybrids in the context of bone stromal co-inoculation but not differentially expressed when mutant cybrids were compared to wild type cybrids without stromal cells in the tumor. In other words, they are similarly expressed when one compares mutant to wild type without the bone stromal microenvironment but become differentially expressed between mutant and wild type in the bone stromal microenvironment. These 37 genes (Table 2) represent the molecular signature that depends both upon the mitochondrial component of the cancer cell and the presence of the bone stromal environment. As the tumor growth curves indicate (Figure 1A), this is the condition of maximal tumor induction (mutant cancer epithelium + bone stromal co-inoculation). The complete list of 37 genes is given in the table. A positive value indicates that the gene is more highly expressed in the mutant condition and a negative value indicates that the gene is more highly expressed in the wild type condition.

Figure 3

Schematic representation of selection of genes that define the unique interaction of mutant prostate cancer epithelial cells with bone stroma.

Table 2

37 genes up or down regulated when 8993Mut cybrid cells are grown with HS27a or HS5 cells. DNA array analysis is described in Methods. Check marks indicate the genes that to date have been verified by real time PCR from tumor tissue of xenografts from (more ...)

Analysis of Tumor Tissues for Mutation and Absence of Bone Stromal Cells

DNA was prepared from tumor tissues and an analysis was performed to verify that the tumor tissues from the 8993Mut cells contained the 8993 mutation. Double stranded PCR amplification of the region of mtDNA containing the 8993 mutation followed by digestion of the DNA products with HpaII reveals the absence or presence of the mutation at position 8993. The T to G mutation results in the introduction of an HpaII cleavage site. All tumor tissues resulting from tumor formation in mice injected with the 8993Mut cells contained the mitochondria DNA mutation present in the PC3 cybrid cell line (data not shown). This would not have been the case had the tumors arisen from the wildtype cybrids or from other mouse cells. In addition, tumor tissues were analyzed for the presence of HS27a or HS5 cells. According to ATCC, these cells were immortalized through transformation of long term bone marrow cells with the amphotropic retrovirus vector LXSN16E6E7 which contains the neomycin resistance gene. Based on this premise, the HS27a and HS5 cells would contain the neomycin resistance gene and the PC3 cybrid cells would not. To verify that the bone stromal cells were not growing as part of the tumors, cDNA prepared from the RNA of the tumor tissues was analyzed for the presence of the neomycin resistance gene which is present in both the HS27a and HS5 cells but not in the 8993WT or 8993Mut cells. Figure 4 demonstrates that the neomycin gene is present in the HS27a cells used as a positive control and not in the 8993WTcells, used as a negative control. The neomycin resistance gene is absent in the tumor tissues tested confirming that the bone stromal cells were not part of the tumor tissue analyzed. The neomycin resistance gene was also present in HS5 cells (data not shown).

Figure 4

Verification of the absence of HS27a and HS5 cell growth in mouse tumors

Validation of Array Data

Several genes were selected for validation of the array data using real time PCR. Figure 5 A and B verifies that the elevated levels of the messages for FGF1 and FAK are observed in tumors that resulted from the injection of 8993Mut/HS27a cell combinations when compared to the 8993WT/HS27a and 8893Mut or 8993WT cells alone. All PCR reactions were normalized to 18S RNA levels.

Figure 5

RT-PCR confirmation that both FGF-1 and FAK are most strongly induced in mutant cancer cells grown with bone stromal cells

Gene expression confirmed at protein level by western and immunohistochemistry

As shown in Figure 5, FGF-1 and FAK are significantly overexpressed at the mRNA level in prostate cancer xenografts when both the mtDNA mutation is present in the cancer cells and human bone stromal cells are co-inoculated in the xenograft. In order to confirm this at the protein level we performed immunohistochemical analysis of the xenografts that resulted from the co-inoculation of prostate cancer cybrids with bone stromal cells and also performed western analysis of the protein lysates of the cybrids when grown in vitro in co-culture with bone stromal cells (Figure 6). These data show that in fact FGF-1 and FAK are upregulated in the prostate cancer cells when exposed to the bone stromal cells.

Figure 6

Confirmation of gene expression at protein level by western and immunohistochemistry

DISCUSSION

These data support the hypothesis that prostate cancer cells containing a mutation in the mitochondrial DNA may be relevant to bone metastasis. The data demonstrate that the growth advantage conferred by mtDNA mutations is most fully realized in the bone stromal microenvironment. This effect has been demonstrated in two model systems: co-inoculation of stromal and epithelial cells subcutaneously in nude mice and the direct intratibial injection of mutant or wild type cancer cells. In addition, we present the findings of differential regulation of genes important in the progression of prostate cancer by the presence of bone stromal cells. Gene expression analysis reveals a 37 gene signature of this highly growth-inductive interaction between prostate cancer cells with mtDNA mutation and bone stromal cells. Amongst these 37 genes, two stand out as significantly overexpressed only when both mtDNA mutations and bone stroma are present and as highly relevant to prostate cancer: FGF-1 and FAK, and others may ultimately be shown to be important.

Both FGF-1 (acidic or aFGF) and FGF-2 (basic or bFGF) are expressed in clinical prostate cancer, along with other FGF family members [16]. In a series of 72 patients with prostate cancer, aFGF was overexpressed in 62/72 (86.1%) cases while bFGF was only overexpressed in 65.5% of cases. In this detailed analysis of cellular site of aFGF immunoreactivity in clinical prostate specimens, only 6.9% of cases showed aFGF being produced in the stroma, and 7.4% of cases showed aFGF production in the surrounding benign tissue from malignant prostate specimens. In benign prostatic hyperplasia (BPH) specimens, none showed significant aFGF protein. This increase in aFGF in clinical prostate cancer, compared to either benign hypertrophy or surrounding stroma, is consistent with findings in other malignancies such as colorectal and pancreatic carcinomas [17,18]. Thus, aFGF appears to be the major FGF isoform in clinical prostate cancer and is secreted predominantly by the malignant epithelium, suggesting autocrine stimulation.

In prostate cancer, FGF-1 interacts (primarily) with FGF-receptor 1 (FGFR-1) to activate it by phosphorylation. Activated FGFR-1 also phosphorylates other intracellular proteins including phospholipase C and extracellular signal-regulated kinases (ERKs), members of the MAP kinase family. FGF-1 stimulation of this pathway results in phosphorylation (and activation) of STAT3 and expression of matrix metalloproteinases MMP 14 and MMP 7 [19]. MMP 14 is also known as membrane-type 1 matrix metalloproteinase (MT1-MMP) and is a key enzyme in the initiation of extracellular matrix (ECM) protein breakdown. MMP7 (matrilysin) is overexpressed in prostate cancer and increases prostate cancer cell invasion. Thus, in the bone, prostate cancer cells with mtDNA mutations may increase metastatic potential and the ability to breakdown the extracellular matrix via the induction of the FGF-1 signaling pathway.

Focal adhesion kinase (FAK) also known as protein tyrosine kinase 2 (PTK2) is important in cell spreading, cell migration and motility, cell proliferation and suppression of apoptosis. FAK is a non-receptor tyrosine kinase that is ubiquitously expressed and becomes activated upon phosphorylation of tyrosine 397.

FAK has been examined by immunohistochemistry in clinical prostate cancer specimens. In the normal (benign) prostate glands FAK is absent from the luminal epithelium but present at a high level in the basal cells. It is also present in PIN, PIA and prostate cancer. Strong immunostaining was seen in 70% of primary prostate cancers, but the pattern was heterogeneous while metastatic foci had uniform and strong staining, suggesting a selective advantage for the development of metastatic disease [20]. These findings corroborate an earlier study using RT-PCR that demonstrated increased FAK message levels in metastatic patients compared to patients with localized disease [21]. FAK can also be upregulated by interaction with the extracellular matrix [22].

One result of 8993 mtDNA mutations is an increase in ROS in subcutaneous tumors developing from the mutant cybrids [3]. This mutation has been studied in cybrid cell lines and shown to cause the hyperpolarization of the mitochondrial inner membrane and increase mitochondrial ROS production [26]. Similarly, mutations in respiratory complex I have been shown to increase metastatic potential of cancers cells in vivo, an effect that is mediated by ROS [27]. We have also demonstrated that respiratory complex IV mutations increase ROS (data not shown). Thus, one common result of mtDNA mutations in cancer is to increase cellular ROS. There is an intimate relationship between ROS and FAK. Overexpression of FAK protects cells against oxidative-stress induced apoptosis and inhibits the activation of caspase-3 protease by hydrogen peroxide. In order for FAK to be in the active form, it must remain phosphorylated, a process that is enhanced in the presence of ROS and the oxidative inhibition of FAK tyrosine phosphatase is required for cell adhesion. Multiple studies document that ROS trigger FAK phosphorylation, including in pulmonary artery endothelium [23] and human umbilical vein endothelial cells [24]. ROS from mitochondria trigger FAK phosphorylation in endothelial cells, a response that arises in response to mechanical stretch and involves protein kinase C (PKC) [25].

While there is currently ample documentation of mtDNA mutations in prostate cancer there remains a great need to determine their functionality. Because of our previous work showing that mtDNA mutations confer only minimal growth advantage in vitro but are responsible for a large, significant and reproducible growth advantage in vivo [3], we sought to further explore the impact of mtDNA on stromal-epithelial interactions in prostate cancer.

Clinically, prostate cancer metastases show a consistent tropism for bone. The specific causes of this site-specific metastatic pattern remain an area of active investigation. Our data show that mtDNA mutations in the potentially bone-metastatic cell may preferentially enhance growth in the bone and bone stromal microenvironment, an effect likely mediated by upregulation of FGF-1 and FAK. We therefore propose that mtDNA mutations in prostate cancer are functional, enhancing bone metastasis. These findings are corroborated by the recent discovery of the importance of mtDNA mutations in metastatic capacity of murine cancer cell lines [27].

Acknowledgements

This work was supported by a grant from the NIH: PO1, CA98912.

REFERENCES

1. Horton TM, Petros JA, Heddi A, Shoffner J, Kaufman AE, Graham SD, Jr, Gramlich T, Wallace DC. Novel mitochondrial DNA deletion found in a renal cell carcinoma. Genes Chromosomes Cancer. 1996 Feb;15(2):95–101. [PubMed]

2. Polyak K, Li Y, Zhu H, Lengauer C, Willson JK, Markowitz SD, Trush MA, Kinzler KW, Vogelstein B. Somatic mutations of the mitochondrial genome in human colorectal tumours. Nat Genet. 1998 Nov;20(3):291–3. [PubMed]

3. Petros JA, Baumann AK, Ruiz-Pesini E, Amin MB, Sun CQ, Hall J, Lim S, Issa MM, Flanders WD, Hosseini SH, Marshall FF, Wallace DC. mtDNA mutations increase tumorigenicity in prostate cancer. Proc Natl Acad Sci U S A. 2005 Jan 18;102(3):719–24. [PubMed]

4. Chen JZ, Gokden N, Greene GF, Mukunyadzi P, Kadlubar FF. Extensive somatic mitochondrial mutations in primary prostate cancer using laser capture microdissection. Cancer Res. 2002 Nov 15;62(22):6470–4. [PubMed]

5. Jerónimo C, Nomoto S, Caballero OL, Usadel H, Henrique R, Varzim G, Oliveira J, Lopes C, Fliss MS, Sidransky D. Mitochondrial mutations in early stage prostate cancer and bodily fluids. Oncogene. 2001 Aug 23;20(37):5195–8. [PubMed]

6. Brandon M, Baldi P, Wallace DC. Mitochondrial mutations in cancer. Oncogene. 2006 Aug 7;25(34):4647–62. [PubMed]

7. Copeland WC, Wachsman JT, Johnson FM, Penta JS. Mitochondrial DNA alterations in cancer. Cancer Invest. 2002;20(4):557–69. Review. [PubMed]

8. Parr RL, Dakubo GD, Crandall KA, Maki J, Reguly B, Aguirre A, Wittock R, Robinson K, Alexander JS, Birch-Machin MA, Abdel-Malak M, Froberg MK, Diamandis EP, Thayer RE. Somatic mitochondrial DNA mutations in prostate cancer and normal appearing adjacent glands in comparison to age-matched prostate samples without malignant histology. J Mol Diagn. 2006 Jul;8(3):312–9. [PubMed]

9. Tatuch Y, Robinson BH. The mitochondrial DNA mutation at 8993 associated with NARP slows the rate of ATP synthesis in isolated lymphoblast mitochondria. Biochem Biophys Res Commun. 1993 Apr 15;192(1):124–8. [PubMed]

10. Trounce I, Neill S, Wallace DC. Cytoplasmic transfer of the mtDNA nt 8993 T-->G (ATP6) point mutation associated with Leigh syndrome into mtDNA-less cells demonstrates cosegregation with a decrease in state III respiration and ADP/O ratio. Proc Natl Acad Sci U S A. 1994 Aug 30;91(18):8334–8. [PubMed]

11. Bisanz K, Yu J, Edlund M, Spohn B, Hung MC, Chung LW, Hsieh CL. Targeting ECM-integrin interaction with liposome-encapsulated small interfering RNAs inhibits the growth of human prostate cancer in a bone xenograft imaging model. Mol Ther. 2005 Oct;12(4):634–43. [PubMed]

12. Hammond-McKibben D, Lake P, Zhang J, Tart-Risher N, Hugo R, Weetall M. A high-capacity quantitative mouse model of drug-mediated immunosuppression based on rejection of an allogeneic subcutaneous tumor. J Pharmacol Exp Ther. 2001 Jun;297(3):1144–51. [PubMed]

13. Nelson PS, Clegg N, Arnold H, Ferguson C, Bonham M, White J, Hood L, Lin B. The program of androgen-responsive genes in neoplastic prostate epithelium. Proc Natl Acad Sci U S A. 2002 Sep 3;99(18):11890–5. Epub 2002 Aug 16. [PubMed]

14. Jain N, Ley K, Thatte J, Lee JK. Local pooled error test for identifying differentially expressed genes with a small number of replicated microarrays. Bioinformatics. 2003;19(15):1945–51. [PubMed]

15. Cho HJ, Smalley D, Ross M, Ley K, Theodorescu D, Lee JK. Statistical Identification of Differentially Labeled Peptides from Liquid Chromatography Tandem Mass Spectrometry. Proteomics. 2007;7(20):3681–92. [PubMed]

16. Dorkin TJ, Robinson MC, Marsh C, Neal DE, Leung HY. aFGF immunoreactivity in prostate cancer and its co-localization with bFGF and FGF8. J Pathol. 1999 Dec;189(4):564–9. [PubMed]

17. el-Hariry I, Pignatelli M, Lemoine N. Fibroblast growth factor 1 and fibroblast growth factor 2 immunoreactivity in gastrointestinal tumours. J Pathol. 1997 Jan;181(1):39–45. [PubMed]

18. Yamanaka Y, Friess H, Buchler M, Beger HG, Uchida E, Onda M, Kobrin MS, Korc M. Overexpression of acidic and basic fibroblast growth factors in human pancreatic cancer correlates with advanced tumor stage. Cancer Res. 1993 Nov 1;53(21):5289–96. [PubMed]

19. Udayakumar TS, Stratton MS, Nagle RB, Bowden GT. Fibroblast growth factor-1 induced promatrilysin expression through the activation of extracellular-regulated kinases and STAT3. Neoplasia. 2002 Jan-Feb;4(1):60–7. [PMC free article] [PubMed]

20. Rovin JD, Frierson HF, Jr, Ledinh W, Parsons JT, Adams RB. Expression of focal adhesion kinase in normal and pathologic human prostate tissues. Prostate. 2002 Oct 1;53(2):124–32. [PubMed]

21. Tremblay L, Hauck W, Aprikian AG, Begin LR, Chapdelaine A, Chevalier S. Focal adhesion kinase (pp125FAK) expression, activation and association with paxillin and p50CSK in human metastatic prostate carcinoma. Int J Cancer. 1996 Oct 9;68(2):164–71. [PubMed]

22. Tremblay L, Hauck W, Nguyen LT, Allard P, Landry F, Chapdelaine A, Chevalier S. Regulation and activation of focal adhesion kinase and paxillin during the adhesion, proliferation, and differentiation of prostatic epithelial cells in vitro and in vivo. Mol Endocrinol. 1996 Aug;10(8):1010–20. [PubMed]

23. Vepa S, Scribner WM, Parinandi NL, English D, Garcia JG, Natarajan V. Hydrogen peroxide stimulates tyrosine phosphorylation of focal adhesion kinase in vascular endothelial cells. Am J Physiol. 1999 Jul;277(1 Pt 1):L150–8. [PubMed]

24. Ben Mahdi MH, Andrieu V, Pasquier C. Focal adhesion kinase regulation by oxidative stress in different cell types. IUBMB Life. 2000 Oct-Nov;50(4–5):291–9. Review. [PubMed]

25. Ali MH, Mungai PT, Schumacker PT. Stretch-induced phosphorylation of focal adhesion kinase in endothelial cells: role of mitochondrial oxidants. Am J Physiol Lung Cell Mol Physiol. 2006 Jul;291(1):L38–45. Epub 2006 Mar 1. [PubMed]

26. Trounce I, Neill S, Wallace DC. Cytoplasmic transfer of the mtDNA nt 8993 T-->G (ATP6) point mutation associated with Leigh syndrome into mtDNA-less cells demonstrates cosegregation with a decrease in state III respiration and ADP/O ratio. Proceedings of the National Academy of Sciences of the United States of America. 1994;91:8334–8338. [PubMed]

27. Ishikawa K, Takenaga K, Akimoto M, Koshikawa N, Yamaguchi A, Imanishi H, Nakada K, Honma Y, Hayashi J. ROS-generating mitochondrial DNA mutations can regulate tumor cell metastasis. Science. 2008 May 2;320(5876):661–4. [PubMed]