Upregulation of Poly (ADP-Ribose) Polymerase-1 (PARP1) in Triple-Negative Breast Cancer and Other Primary Human Tumor Types

Abstract

Poly (ADP-ribose) polymerase-1 (PARP1) is a key facilitator of DNA repair and is implicated in pathways of tumorigenesis. PARP inhibitors have gained recent attention as rationally designed therapeutics for the treatment of several malignancies, particularly those associated with dysfunctional DNA repair pathways, including triple-negative breast cancer (TNBC). We investigated the PARP1 gene expression profile in surgical samples from more than 8,000 primary malignant and normal human tissues. PARP1 expression was found to be significantly increased in several malignant tissues, including those isolated from patients with breast, uterine, lung, ovarian, and skin cancers, and non-Hodgkin’s lymphoma. Within breast infiltrating ductal carcinoma (IDC) samples tested, mean PARP1 expression was significantly higher relative to normal breast tissue, with over 30% of IDC samples demonstrating upregulation of PARP1, compared with 2.9% of normal tissues. Because of known DNA repair defects, including BRCA1 dysfunction, associated with TNBC, exploration of PARP1 expression in breast cancers related to expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) led to the observation that negative expression of any of the 3 receptors was associated with upregulation of PARP1 expression, compared with receptor-positive tissues. To validate these observations, an independent set of breast adenocarcinomas was evaluated and demonstrated >2-fold upregulation of PARP1 in approximately 70% of primary breast adenocarcinomas, including TNBC, compared with syngeneic nonmalignant breast tissues. Immunohistochemistry (IHC) showed that upregulation of the PARP1 gene was consistent with increased protein expression in TNBC. These analyses suggest a potential biological role for PARP1 in several distinct malignancies, including TNBC. Further investigation of PARP1 as a biomarker for the therapeutic activity of PARP inhibitor-based therapy is warranted.

Keywords

PARP cancer targeted therapies triple-negative breast cancer BRCA

Introduction

Poly (ADP-ribose) polymerase-1 (PARP1) is a chromatin-associated enzyme with key functions in the regulation of transcription, cell cycle, tumorigenesis, and cellular response to DNA damage.¹ PARP1 is activated by DNA damage and has important roles in DNA base excision repair (BER), functioning as a nick sensor, recruiter, and modulator of key DNA repair molecules.² Upon activation, PARP1 synthesizes poly (ADP-ribose) (PAR) using nicotinamide adenine dinucleotide (NAD⁺) as a substrate and covalently transfers PAR to nuclear proteins, including nucleosomal core histones, topoisomerases I and II, high mobility group (HMG) proteins, and p53.³

Loss of PARP1 activity can lead to enhanced cancer cell death, following treatment with PARP inhibitors, both as single agents and in combination with DNA-damaging agents. Impairing PARP1-dependent BER can elicit DNA double-strand breaks (DSBs) following collapse of the replication fork, particularly in cells whose homologous recombination (HR)–dependent DSB repair is already defective due to mutations in breast cancer 1 and 2 genes (BRCA1 and BRCA2); these defects are found frequently in familial breast and ovarian cancers and can elicit profound sensitization to PARP inhibitors, resulting in cytotoxic effects.^4,5

Over 80% of BRCA-associated breast cancers are negative for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor (HER2).^6,7 These “triple-negative” breast cancers, which comprise 15% to 20% of all breast cancers,^8,9 are among the most aggressive breast cancer subtypes. Importantly, over 60% of sporadic cases of triple-negative breast cancer (TNBC) that are independent of BRCA1/2 germline mutations have been characterized by BRCA1 dysfunction due to promoter methylation or deregulation of other genes involved in transcriptional regulation.^10,11 Because low or negative expression of ER, PR, and HER2 precludes the use of currently available targeted therapies, development of novel targeted agents for TNBC represents a high priority for this patient population.

PARP1 is upregulated at levels of mRNA, protein, and enzyme activity in a number of cancers, including ovarian cancer, hepatocellular cancer, colorectal cancer, and leukemia.^12-14 Despite significant advances in understanding the role of PARP1 in DNA repair, it is still unclear whether PARP1 is upregulated across all tumor types or if its overexpression is specific to particular cancers or individual subtypes. Importantly, evaluation of global PARP1 expression across tumor types can aid in elucidating a role for PARP1 in tumorigenesis and could potentially guide the selective clinical targeting of specific cancers that have increased sensitivity to PARP inhibition.

In order to provide a more extensive profile of PARP1 mRNA expression in human primary cancers and normal tissues, we evaluated PARP1 expression in more than 8,000 primary malignant and normal human tissues. We found that while PARP1 expression was relatively low and uniform across the majority of normal human specimens, it was significantly elevated in several primary malignancies, including cancers of the breast, uterine, ovarian, and lung, as well as lymphomas. Further quantitative and statistical analyses were applied to define the degree of PARP1 upregulation within tumor subclasses, and these findings were validated using quantitative real-time multiplex polymerase chain reaction (QRT-PCR) in an independent set of syngeneic normal and malignant breast tissue samples. Importantly, we found that breast tumors negative for ER, PR, and/or HER2, including the triple-negative subtype, had higher degrees of PARP1 upregulation than receptor-positive subtypes. These findings have potential clinical importance for defining strategies for selecting the biomarkers of sensitivity to PARP1 inhibitors.

Results

PARP1 expression is low in nondiseased human tissues

Microarray analysis showed that PARP1 mRNA expression in histologically normal tissues was low and uniform across a wide variety of evaluated tissues (Fig. 1A and Table 1). The exceptions were lymphoid tissues, such as lymph node, B lymphocytes, adenoids, thymus, and spleen, wherein PARP1 expression was increased approximately 2-fold, compared with other normal tissues (Table 1).

Figure 1.

PARP1 mRNA expression is elevated in human primary cancers of the breast, endometrium, lung, ovary, and skin. Microarray analysis demonstrating (A) relative PARP1 and (B) PARP2 expression in primary cancers compared with normal tissues. Levels of mRNA from a tissue database, consisting of representative samples of primary normal tissues and metastatic and benign tumors, were quantified using Affymetrix U133A and B GeneChips. Solid lines represent median mRNA levels, bars indicate range, and boxes indicate interquartile range.

Table 1.

PARP1 mRNA Expression in Primary Human Tumor Specimens

Sample	Cancer Subclass	n	Mean	Fold Change	P ValueP^a
Breast	Normal	68	201.78
	Infiltrating ductal carcinoma	169	328.49	1.63	<0.001
	Infiltrating lobular carcinoma	17	317.43	1.57	0.002
Colon	Normal, no history of inflammatory bowel disease	56	199.20
	Adenocarcinoma (excluding mucinous type)	26	244.35	1.23	0.001
Endometrium	Normal	23	201.21
	Adenocarcinoma, endometrioid type	50	297.42	1.48	<0.001
	Mullerian mixed tumor	7	517.86	2.57	0.003
Esophagus	Normal	22	191.78
	Esophagus, adenocarcinoma	3	290.09	1.51	<0.001
Kidney	Normal	81	165.78
	Wilms’ tumor	8	385.07	2.32	0.001
Lung	Normal	126	170.58
	Adenocarcinoma	46	284.99	1.67	<0.001
	Squamous cell carcinoma	39	309.53	1.81	<0.001
Lymph node	Normal	10	325.59
	Non-Hodgkin’s lymphoma, all types	91	756.79	2.32	<0.001
	Non-Hodgkin’s lymphoma, diffuse large B cell type	32	958.33	2.02	<0.001
	Non-Hodgkin’s lymphoma, follicular type	30	752.45	2.31	<0.001
	Non-Hodgkin’s lymphoma, mantle cell type	8	528.09	1.62	0.018
Ovary	Normal	89	163.31
	Adenocarcinoma, endometrioid type	22	331.23	2.03	<0.001
	Adenocarcinoma, papillary serous type	36	361.56	2.21	<0.001
	Mullerian mixed tumor	5	371.40	2.27	0.032
	Serous cystadenocarcinoma	8	371.23	2.27	0.001
Rectum	Normal	44	206.94
	Adenocarcinoma (excluding mucinous type)	29	262.78	1.27	<0.001
Skin	Normal	61	154.29
	Basal cell carcinoma	4	255.43	1.66	0.043
Spleen	Normal	34	302.23
	Non-Hodgkin’s lymphoma	9	689.94	2.28	<0.001
Stomach	Normal	52	221.41
	Adenocarcinoma	27	267.48	1.21	0.04
Testis	Normal	7	333.35
	Seminoma	8	622.56	1.87	0.001
Thyroid gland	Normal	24	173.82
	Papillary carcinoma, follicular variant	29	225.17	1.30	0.017

Note: PARP1 expression was measured using Affymetrix microarrays, as described in Materials and Methods.

P values were calculated using the Student t test comparing tumor and normal tissues.

PARP1 mRNA is elevated in many human primary malignancies

Microarray analysis of PARP1 gene expression in more than 8,000 samples revealed that PARP1 was more highly expressed in several cancer types, compared with corresponding normal tissues (Fig. 1A and Table 2). Overall, the most striking differences in PARP1 expression were in breast, ovarian, endometrial, lung, and skin cancers, and non-Hodgkin’s lymphoma. However, the increase in PARP1 expression was not a universal characteristic of all cancer types since the majority of tissues from adrenal, bone, colon, and prostate carcinomas did not demonstrate significant increases in PARP1 expression (Fig. 1A).

Table 2.

Upregulation of PARP1 mRNA Expression in Human Breast IDC Tissues, Grouped according to Expression of ER, PR, and HER2

Breast IDC Sample	n	Samples with PARP1 >95% UCL of Normal Tissue, %	Median PARP1 Expression
Normal	68	2.9	178.83
IDC	169	30.2	312.34
ER+	36	22.9	307.63
ER−	18	55.6	370.61
PR+	26	23.1	300.17
PR−	20	45.0	354.76
HER2+	24	29.2	332.16
HER2−	10	70.0	419.27

Note: PARP1 expression was measured using Affymetrix microarrays, as described in Materials and Methods. UCL = upper confidence limit.

Poly (ADP-ribose) polymerase-2 (PARP2) is a PARP family member that localizes to active centromeres and interacts with PARP1.¹⁵ PARP2 is also involved in the response to DNA damage and required for efficient DNA base excision repair.^16,17 Using the same set and population of human primary tumors in which elevated PARP1 expression was observed, analysis of PARP2 expression did not reveal any significant differences between cancer and normal tissues, suggesting that the observed upregulation of PARP1 was gene specific (Fig. 1B).

To define whether PARP1 gene expression was upregulated in human primary malignancies, additional statistical analyses of breast, ovarian, uterine, endometrial, and lung tissues were used to standardize levels of expression in normal tissues. Mean mRNA expression and 90%, 95%, 99%, and 99.9% upper confidence limits (UCLs) were calculated for cancer subtypes. Tissues with PARP1 upregulation were statistically defined as samples whose expression levels exceeded the 95% UCL of respective normal tissues (see Materials and Methods).

PARP1 expression in breast cancers

Results from our gene expression profiling analysis showed that PARP1 was elevated in both infiltrating ductal and lobular carcinomas. Because infiltrating ductal carcinoma (IDC) is the most common histological type of primary breast cancers, accounting for 70% to 80% of all breast malignancies,^18,19 additional analyses were applied to this subclass of breast cancers. PARP1 expression had significantly higher expression in IDC samples, compared with normal nonmalignant breast tissues (P < 0.001). Further statistical analysis of these samples revealed that approximately 30% of IDC samples demonstrated upregulation of PARP1, falling above the 95% UCL of PARP1 expression in normal tissues (Fig. 2 and Table 2).

Figure 2.

Statistical representation of PARP1 upregulation in ovarian, uterine, lung, and prostate cancers. PARP1 upregulation was defined based on levels of PARP1 mRNA, quantified by microarray analysis, which exceeded the 90%, 95%, 99%, and 99.9% UCLs of normal tissue distribution. Tumor types were grouped according to pathological subtype and subcategorized according to clinical characteristics. Data are represented by the percentage of samples that reach these criteria. UCL = upper confidence limit.

Studies have demonstrated an interdependence of BRCA and PARP1 for cancer cell survival,^4,5 shared molecular and pathological features between BRCA-deficient breast cancers and TNBC,²⁰ and high rates of BRCA1 dysfunction in TNBC.^10,11 Due to observed associations between defects in DNA repair pathways and TNBC,^21,22 we hypothesized that these underlying defects could be linked and compensated by parallel DNA repair pathways. To investigate whether such a role existed for PARP1, we examined its expression levels within subpopulations of IDC, grouped according to expression of ER, PR, or HER2. We found that the incidence of PARP1 upregulation in IDC samples negative for ER, PR, or HER2 was higher than corresponding receptor-positive samples (Table 2). Analysis of these data revealed the presence of 8 outliers, whose genomic profile did not correspond with receptor status previously defined by histopathology. Upregulation of PARP1 within the curated sample set, wherein outliers were omitted, can be found in Supplementary Table S1.

PARP1 expression in female reproductive cancers

Ovarian cancer represents a heterogeneous population of cancers defined by morphological characteristics and cell derivation. Malignant ovarian samples were classified by histopathological subclasses, and levels of PARP1 expression in these tissues were compared with histologically normal ovarian tissue samples using the 95% UCL cut-off to define upregulation. The highest PARP1 expression levels within ovarian cancers were found in papillary serous adenocarcinoma, serous cystadenocarcinoma, Mullerian mixed tumor, and granulosa cell tumor samples (Table 1). Approximately 75% of ovarian cancers expressed PARP1 above the 95% UCL of normal nonmalignant ovarian samples. These findings are consistent with previous reports that have demonstrated elevated levels of poly (ADP-ribosyl)ation in primary ovarian cancers.¹³ Of uterine cancers, Mullerian mixed tumors had the highest incidence (85.7%) of samples above the 95% UCL. Most endometrioid cancers expressed PARP1 above the 90% UCL (Fig. 2).

PARP1 expression in lung cancers

Analysis of PARP1 expression in lung cancers showed that all subtypes evaluated had higher PARP1 expression than normal lung tissue (Table 1). Most pathological subclasses of lung cancer showed a majority of samples above the 90% and 95% UCL (Fig. 2). PARP1 was more highly expressed in both lung adenocarcinomas and squamous cell carcinomas (P < 0.001), with squamous cell carcinomas having a higher percentage of samples above both the 95% and 99.9% UCLs (Fig. 2).

PARP1 expression in prostate cancers

In contrast to breast, ovarian, endometrioid, and lung cancers, which demonstrated higher degrees of PARP1 upregulation relative to normal tissues, the difference in PARP1 expression between prostate adenocarcinomas and corresponding normal control prostate tissue was only 1.14-fold (P < 0.001) (Fig. 2), as was the percentage of tumors exceeding the 95% UCL (~10%).

Correlation analysis of PARP1 expression with tumor suppressor genes and markers

Tumor suppressor genes BRCA1 and BRCA2 have a key “caretaker” function in neoplasia development, genetic instability, and regulation of multiple DNA repair pathways.^23,24 Dysfunctions in BRCA1 and BRCA2 are thought to be particularly important risk factors associated with hereditary breast and ovarian cancers.^25-27 BRCA1 and BRCA2 are found in a holoenzyme complex in association with RAD51 (RecA homolog gene), an essential factor in HR and repair of DNA DSBs.²⁸ Interestingly, we found that ovarian samples from patients with papillary serous adenocarcinoma had the highest levels of PARP1 expression, which correlated with the lowest or undetectable levels of BRCA1 expression (data not shown). Moreover, increased PARP1 expression was found to correlate with RAD51 in primary ovarian cancers (Pearson correlation = 0.49). No significant correlation was found between PARP1 and PARP2 expression in ovarian cancer samples (Pearson correlation = 0.05).

Cancer antigen 125 (CA125) is an established marker of ovarian tumor progression. Serum concentrations of CA125 are increased in 80% to 90% of advanced stage ovarian cancers and have been used for the diagnosis of ovarian cancer.²⁹ Interestingly, our data showed that CA125 mRNA correlated with increased PARP1 expression across all sets of ovarian cancer samples (Pearson correlation = 0.4).

PARP1 expression in syngeneic sets of normal and adenocarcinoma breast tissues

To confirm our observations of PARP1 gene upregulation in malignant breast tissues, we performed additional analyses of PARP1 expression in an independent set of syngeneic normal and malignant breast tissues, distinct from the IDC samples described above. All samples were isolated from patients (n = 18) who had no chemotherapy or radiation prior to tissue removal, thus eliminating the possibility of therapy-induced alterations in gene expression. All breast cancer samples were characterized as grade T2 or T3 IDC. Expression of PARP1, PARP2, proliferation-related Ki-67 antigen, topoisomerase 2α (TOP2A), ER, PR, and HER2 in these samples was evaluated by QRT-PCR (see Materials and Methods). Our analysis showed that PARP1 expression was increased >2-fold in more than 70% of breast adenocarcinomas, compared with normal tissues (Fig. 3A), confirming findings from the microarray analysis described above. In contrast, there were no significant differences in PARP2 expression between syngeneic normal and tumor samples, precluding the possibility of nonspecific upregulation of transcription in proliferating cancer cells (Fig. 3B). Ki-67 and TOP2A, validated markers of breast cancer cell proliferation, were used as positive controls. To confirm these results, PARP1 protein expression was evaluated by immunohistochemistry (IHC) in the same set of samples used for XP-PCR transcriptional profiling. IHC staining demonstrated elevated levels of PARP1 protein in TNBC, compared with normal tissues, reflecting the elevated levels of mRNA described above (Fig. 4). These findings further support our hypothesis that PARP1 upregulation is a defining characteristic of TNBC.

Figure 3.

PARP1 gene expression is elevated in syngeneic normal and IDC breast samples. Quantitative real-time multiplex RT-PCR was used to quantify (A) mean PARP1 mRNA expression in normal and malignant tissues from individual patients with breast IDC. Number represents patient code. Results are mean ± standard deviation of pooled data from 3 independent experiments (see Materials and Methods). (B) Distribution of PARP1, PARP2, Ki-67, and TOP2A mRNA expression in syngeneic sets of malignant and normal tissue samples. RNA was isolated from FFPE sections or frozen tissues, and gene expression levels were quantified and normalized to the expression of GUSB (β-glucuronidase gene). Data were reported as the mean and standard deviation of 3 independent assessments for each sample. Solid lines represent median mRNA levels, bars indicate range, and boxes indicate interquartile range. N = normal tissue; T = tumor.

Figure 4.

Increased levels of PARP1 mRNA and protein in triple-negative breast adenocarcinoma. (A) PARP1, HER2, ER, PR, Ki-67, and TOP2A gene expression was determined by multiplex RT-PCR (see Materials and Methods). RNA was isolated from FFPE sections or frozen tissues, and gene expression levels were quantified and normalized to the expression of GUSB (β-glucuronidase gene). Data were reported as the mean and standard deviation of 3 independent assessments for each sample. (B) Immunohistochemistry of syngeneic normal and triple-negative breast carcinoma tissues. FFPE sections were stained for PARP1, ER, PR, and HER2 protein (see Materials and Methods). Images were taken at 40x magnification; scale bar represents 100 µm.

Discussion

We surveyed PARP1 gene expression in a variety of human primary cancers with 2 primary goals: 1) to better understand the role of PARP1 in tumor biogenesis; and 2) to identify and select potentially important cancer targets for PARP1-directed therapy. A genome-wide gene expression database with histopathology and clinical details for over 22,000 clinically defined tissues was used in this study to assess PARP1 expression. We found that PARP1 expression was significantly increased in several primary malignancies, including breast, uterine, lung, and ovarian cancers, compared with their histologically normal counterparts. In contrast, increased PARP1 expression was less striking in colon and prostate cancers. In normal tissues, PARP1 expression was generally low, with the exception of lymphoid tissues, which have increased frequency of recombination and genetic events during differentiation of B cells.³⁰ Due to previously reported observations of a role for PARP1 in the survival of BRCA1/2-deficient cells,^4,5 in addition to the high rates of BRCA1 mutations and/or dysfunction in TNBC,^10,11 we were particularly interested in comparing the levels of PARP1 expression in the available samples of breast IDC tissues, grouped according to the expression of ER, PR, and HER2. Importantly, negative expression of each receptor was associated with higher degrees of PARP1 upregulation, relative to receptor-positive counterparts of breast cancers evaluated. Despite the fact that PARP2 forms a complex with PARP1 and is involved in the regulation of DNA repair and cell cycle,^15,17 no significant differences in PARP2 expression were observed between malignant and normal tissues, suggesting that the increase in PARP1 expression was specific to selective types of cancers rather than as a consequence of rapid tumor cell proliferation.

The PARP1 promoter belongs to the TATA-deficient, GC-rich class of promoters³¹ and has binding sites for several transcription factors, including Sp1, AP-2α, yin and yang 1 (YY1), and v-ets erythroblastosis virus E26 oncogene homolog 1 (ETS1),^32-35 indicating that regulation of PARP1 may occur at the level of transcription via cancer-associated transcription factors or oncogenes. Indeed, the ETS1 oncogene, which appears to play an important role in the biology of Ewing’s sarcoma, acts as a transcriptional regulator of PARP1 in Ewing’s sarcoma, where PARP1 is upregulated at levels of mRNA, protein, and enzymatic activity.^35,36 However, the specific effects of malignant transformation and PARP1 transcription on each other are unknown and remain a topic of broad investigation.

Here, we have shown that upregulation of PARP1 was characteristic of all histological types of lung cancer with little discrimination among them. In contrast, the level of PARP1 mRNA was highest in mixed Mullerian tumors of the endometrium and in receptor-negative breast IDCs. Reasons for high PARP1 expression in these particular cancer subtypes are unclear. Interestingly, TNBC shares molecular signatures with breast cancers from patients who harbor BRCA1 or BRCA2 germline mutations.^20,37

Inherent defects in DNA repair pathways represent a common feature characterizing inherited susceptibility to PARP inhibitors in several cancers, including breast and ovarian cancers that harbor defects in BRCA1 or BRCA2.^5,38 Interestingly, our observation of an inverse relationship of PARP1 and BRCA1 expression in an ovarian cancer subset supports the hypothesis that loss of one critical DNA repair pathway can be compensated by upregulation of alternative DNA repair mechanisms integral to the cellular response to genotoxic stress. Indeed, this hypothesis is corroborated in a study by Wang et al., which demonstrated that PARP1 downregulates BRCA2 gene expression through interaction with the BRCA2 promoter.³⁹ The specific driver of PARP1 expression in selected cancers is unknown, begging the question whether activation of PARP1 represents a response to defects in other DNA repair pathways or to some other mechanism.

Recent progress in the understanding of the molecular basis of cancer cell proliferation and tumor biogenesis has led to the development of “targeted” therapies for treating several types of cancer. Such therapies are developed based on the specific targeting of distinct key molecular pathways that are altered in malignant but not benign cells. These innovative therapies have transformed cancer therapeutics from empirically based cytotoxic treatments that lack specificity to rational therapeutic models based on genetic signatures of tumors and molecular characterization of distinct cancer types. The successful clinical validation of targeted therapies has been demonstrated with the FDA approval of biological agents, such as bevacizumab and trastuzumab, which target vascular endothelial growth factor (VEGF) and HER2, respectively. Targeting of DNA repair pathways, such as those requiring PARP1, represents a new generation of rationally designed cancer therapies, particularly when used in combination with DNA-alkylating agents or irradiation. Ultimately, interference with key tumor survival mechanisms that are codependent represents an extremely powerful and efficient therapeutic strategy.

Development of targeted therapies has initiated a new approach to drug discovery and may allow molecular criteria and biomarkers, such as PARP1, to serve as predictive indicators of drug activity, as well as patient selection. The increased susceptibility of BRCA-associated tumors to PARP inhibition has prompted growing attention toward the development of PARP inhibitors as novel therapeutic agents for the treatment of cancer. Our study has shown that PARP inhibitor therapy may not be limited to these cancers only, and expansion of these agents to the treatment of other cancer types, especially in combination with DNA-damaging agents, should be explored.

Screening for differential gene expression as a way to identify potentially important cancer biomarkers is a novel use of genome-wide expression databases. Establishing a method for predicting sensitivity of selected cancer types to potent PARP inhibitors will be valuable in guiding its development as a human therapeutic. The observation of PARP1 upregulation in various cancers and their subtypes has implications for the possible expansion in the use of PARP inhibitors for targeting a broad spectrum of human cancers.

Materials and Methods

Tissue samples

For the microarray studies, specimens were collected as part of a normal surgical procedure and were flash frozen within 30 minutes of resection. Tissue collection was approved by institutional review boards at sample accrual sites and patient consent obtained for research use. Tissues used in the study included samples from a representative cross-section of primary, metastatic, and benign tumors, alongside samples of diseased tissues associated with increased cancer risk and samples of normal tissue from corresponding anatomical locations. All samples were subjected to stringent quality controls, applied at all levels of tissue acquisition and handling. Pathology diagnosis of each sample was confirmed by a board-certified pathologist prior to being integrated into the database. Samples were accompanied by annotation, including donor demographics, medical and medication history, and specific diagnoses, clinical values, pathology reviews, histopathology images (Gene Logic, Gaithersburg, MD; Cureline Inc., South San Francisco, CA). H&E-stained glass slides generated from adjacent tissues were used to confirm and classify diagnostic categories and to evaluate neoplastic cellularity. For breast cancer samples, expression of ER, PR, and HER2 was determined using immunohistochemistry and/or fluorescence in situ hybridization. These results, as well as attendant pathology and clinical data, were annotated within sample gene expression and analysis databases for both population-based and individual sample-based studies (Ascenta, BioExpress Systems, Gene Logic; Cureline BioPathology, Burlingame, CA).

Gene expression profiling and statistical analysis

RNA extraction and microarray hybridization to Affymetrix U133A and B GeneChips were performed as previously described⁴⁰ (http://gene-chips.net). Array data quality was evaluated using array high-throughput applications, which assess the data against multiple objective standards including 5′/3′ glyceraldehyde-3-phosphate dehydrogenase 1 (GAPDH) ratio, signal/noise ratio, and signal background as well as other additional metrics (Gene Logic; Affymetrix, Santa Clara, CA). GeneChip analysis was performed using Affymetrix Microarray Analysis Suite version 5.0, Data Mining Tool 2.0, and microarray database software (Affymetrix). All of the genes represented on the array were globally normalized and scaled to a signal intensity of 100. Sample sets were created using annotations in the Ascenta and BioExpress System databases (Gene Logic).

Histologically normal tissue samples were used to determine baseline expression of PARP1 mRNA. The mean and 90%, 95%, 99%, and 99.9% UCLs for an individual predicted value were calculated. In order to determine the likelihood that individual cancer samples external to the normal set are within the baseline distribution, the prediction interval, rather than the confidence interval for the mean, was selected to estimate the expected range for future individual measurements. The prediction interval is defined by the formula

\bar{X} \pm A S \sqrt{1 + (1 / n)}

where X̄ is the mean of the normal breast samples, S is the standard deviation, n is the sample size, and A is the 100(1−(p/2))^th percentile of the Student t distribution with n − 1 degrees of freedom.

The histologically normal samples used in follow-up studies to determine baseline expression of PARP1 included 68 breast, 23 endometrial, 122 lung, and 57 prostate samples. Cancer samples were grouped by pathological subtype where appropriate and included 169 breast IDC, 106 ovarian carcinoma, 57 endometrial, 122 lung, and 57 prostate samples. In some studies, samples were subcategorized where appropriate according to clinical characteristics, which included tumor stage, smoking status, CA125 status, or age, as well as expression of ER, PR, and HER2 receptors. Each tumor sample was evaluated according to UCL thresholds. Analysis was performed using SAS v8.2 for Windows (www.sas.com).

In the ovarian cancer subset, Pearson correlations were calculated for BRCA1, BRCA2, CA125, PARP2, and RAD51 genes, compared with PARP1. Correlations were based on the complete set of 194 samples. The Pearson product-moment correlation is defined by the formula

r_{x y} = \frac{\sum (x_{i} - \bar{x}) (y_{i} - \bar{y})}{\sqrt{\sum {(x_{i} - \bar{x})}^{2} {(y_{i} - \bar{y})}^{2}}}

where x̄ is the mean of the PARP1 probe set, and ȳ is the mean of the probe set to which PARP1 is being correlated. Statistical significance was determined by the formula

\frac{{(n - 2)}^{1 / 2} r}{{(1 - r^{2})}^{1 / 2}}

where r is the correlation, and n is the number of samples. The resultant value is assumed to have a t distribution with n − 2 degrees of freedom.

QRT-PCR–based transcriptional profiling of gene expression in formalin-fixed paraffin-embedded (FFPE) human breast tumor and normal tissues

XP-PCR (Althea Technology, AltheaDx, San Diego, CA) was performed using 25 ng of total RNA isolated from each sample as previously described.⁴¹ Real-time PCR-based transcriptional profiling and multiplex assays used for this study were designed to detect RNA from FFPE samples or from frozen tissues. RNA quality and quantity were analyzed using the Agilent 2100 bioanalyzer (Agilent Technologies, Palo Alto, CA) and the RiboGreen RNA Quantitation Kit (Invitrogen, Carlsbad, CA) as recommended by the manufacturer. Reverse transcription (RT) reactions were carried out as previously described.⁴¹

Briefly, 25 ng of RNA was converted into single-stranded cDNA, according to the XP Start kit (Beckman Coulter) protocol. The reactions were incubated for 1 minute at 48°C, 5 minutes at 37°C, 60 minutes at 42°C, and then 5 minutes at 95°C. The 20 µL reactions were performed in a Thermo-Fast 96-well PCR Detection Plate (ABgene, Epsom, Surrey, UK). QRT-PCR–based transcriptional profiling was then performed on each sample as previously described.⁴¹ Samples were analyzed with a GeXP Genetic Analysis System (Althea Technology, AltheaDx) and eXpress Analysis module of the GeXP Genetic Analysis System.⁴¹ The expression of each target gene was analyzed relative to the expression of β-glucuronidase (GUSB) within the same reaction. Data were reported as the mean and standard deviation of 3 independent assessments for each sample.

Immunohistochemistry

FFPE sections of syngeneic pairs of human normal and tumor tissues (Cureline Inc.) were deparaffinized, rehydrated, and subjected to antigen retrieval as it was previously described.⁴⁰ PARP1 was detected with goat anti-PARP1 primary antibody (SC-156, Santa Cruz Biotechnology Inc., Santa Cruz, CA) as per the manufacturer’s instructions, using FFPE normal human spleen samples (Cureline BioPathology) as a positive control. ER status in human breast carcinomas was detected with monoclonal mouse antihuman estrogen receptor α antibody, clone 1D5 (1:35; Dako, Glostrup, Denmark), and PR status was detected with mouse monoclonal antihuman PR antibody, clone PgR 636 (Dako), as per the manufacturer’s instructions, using FFPE breast carcinoma (Cureline BioPathology) as a positive control. HER2 was detected with HercepTest (Dako) according to the manufacturer’s instructions.

Footnotes

Acknowledgements

The authors thank their collaborators at Gene Logic, Cureline, Cureline BioPathology, and Althea Technologies and AltheaDx for their technical assistance and helpful discussions.

This work was supported by BiPar Sciences Inc.

Valeria Ossovskaya, Ingrid Koo, and Barry Sherman are employees of BiPar Sciences Inc. Eric Kaldjian and Christopher Alvares are former employees of Gene Logic Inc.

References

Kim

Zhang

Kraus

. Poly(ADP-ribosyl)ation by PARP-1: ‘PAR-laying’ NAD+ into a nuclear signal. Genes Dev. 2005;19:1951-67.

Burkle

. Poly(APD-ribosyl)ation, a DNA damage-driven protein modification and regulator of genomic instability. Cancer Lett. 2001;163:1-5.

Schreiber

Dantzer

Ame

de Murcia

. Poly(ADP-ribose): novel functions for an old molecule. Nat Rev Mol Cell Biol. 2006;7:517-28.

Bryant

Schultz

Thomas

. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature. 2005;434:913-7.

Farmer

McCabe

Lord

. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature. 2005;434:917-21.

Foulkes

Stefansson

Chappuis

. Germline BRCA1 mutations and a basal epithelial phenotype in breast cancer. J Natl Cancer Inst. 2003;95:1482-5.

Lakhani

Reis-Filho

Fulford

. Prediction of BRCA1 status in patients with breast cancer using estrogen receptor and basal phenotype. Clin Cancer Res. 2005;11:5175-80.

Nielsen

Hsu

Jensen

. Immunohistochemical and clinical characterization of the basal-like subtype of invasive breast carcinoma. Clin Cancer Res. 2004;10:5367-74.

Carey

Perou

Livasy

. Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study. JAMA. 2006;295:2492-502.

10.

Miyoshi

Murase

. Basal-like subtype and BRCA1 dysfunction in breast cancers. Int J Clin Oncol. 2008;13:395-400.

11.

Turner

Reis-Filho

Russell

. BRCA1 dysfunction in sporadic basal-like breast cancer. Oncogene. 2007;26:2126-32.

12.

Nomura

Yaguchi

Togawa

. Enhancement of poly-adenosine diphosphate-ribosylation in human hepatocellular carcinoma. J Gastroenterol Hepatol. 2000;15:529-35.

13.

Singh

. Enhanced poly ADP-ribosylation in human leukemia lymphocytes and ovarian cancers. Cancer Lett. 1991;58:131-5.

14.

Yalcintepe

Turker-Sener

Sener

Yetkin

Tiryaki

Bermek

. Changes in NAD/ADP-ribose metabolism in rectal cancer. Braz J Med Biol Res. 2005;38:361-5.

15.

Saxena

Wong

Kalitsis

Earle

Shaffer

Choo

. Poly(ADP-ribose) polymerase 2 localizes to mammalian active centromeres and interacts with PARP-1, Cenpa, Cenpb and Bub3, but not Cenpc. Hum Mol Genet. 2002;11:2319-29.

16.

Ame

Rolli

Schreiber

. PARP-2, a novel mammalian DNA damage-dependent poly(ADP-ribose) polymerase. J Biol Chem. 1999;274:17860-8.

17.

Schreiber

Ame

Dolle

. Poly(ADP-ribose) polymerase-2 (PARP-2) is required for efficient base excision DNA repair in association with PARP-1 and XRCC1. J Biol Chem. 2002;277:23028-36.

18.

Berg

Hutter

. Breast cancer. Cancer. 1995;75:257-69.

19.

Glass

Lacey

Jr Carreon

Hoover

. Breast cancer incidence, 1980-2006: combined roles of menopausal hormone therapy, screening mammography, and estrogen receptor status. J Natl Cancer Inst. 2007;99:1152-61.

20.

Cleator

Heller

Coombes

. Triple-negative breast cancer: therapeutic options. Lancet Oncol. 2007;8:235-44.

21.

Alli

Sharma

Sunderesakumar

Ford

. Defective repair of oxidative dna damage in triple-negative breast cancer confers sensitivity to inhibition of poly(ADP-ribose) polymerase. Cancer Res. 2009;69:3589-96.

22.

Jaspers

Rottenberg

Jonkers

. Therapeutic options for triple-negative breast cancers with defective homologous recombination. Biochim Biophys Acta. 2009;1796:266-80.

23.

Kinzler

Vogelstein

. Cancer-susceptibility genes: gatekeepers and caretakers. Nature. 1997;386:761-3.

24.

Gudmundsdottir

Ashworth

. The roles of BRCA1 and BRCA2 and associated proteins in the maintenance of genomic stability. Oncogene. 2006;25:5864-74.

25.

Turnbull

Rahman

. Genetic predisposition to breast cancer: past, present, and future. Annu Rev Genomics Hum Genet. 2008;9:321-45.

26.

Lynch

Casey

Shaw

Lynch

. Hereditary factors in gynecologic cancer. Oncologist. 1998;3:319-38.

27.

Weberpals

Clark-Knowles

Vanderhyden

. Sporadic epithelial ovarian cancer: clinical relevance of BRCA1 inhibition in the DNA damage and repair pathway. J Clin Oncol. 2008;26:3259-67.

28.

Dong

Hakimi

Chen

. Regulation of BRCC, a holoenzyme complex containing BRCA1 and BRCA2, by a signalosome-like subunit and its role in DNA repair. Mol Cell. 2003;12:1087-99.

29.

Fritsche

Bast

. CA 125 in ovarian cancer: advances and controversy. Clin Chem. 1998;44:1379-80.

30.

Borggrefe

Wabl

Akhmedov

Jessberger

. A B-cell-specific DNA recombination complex. J Biol Chem. 1998;273:17025-35.

31.

Ingolia

Al-Ubaidi

Yeung

Bigo

Wright

Kellems

. Molecular cloning of the murine adenosine deaminase gene from a genetically enriched source: identification and characterization of the promoter region. Mol Cell Biol. 1986;6:4458-66.

32.

Oei

Griesenbeck

Schweiger

Babich

Kropotov

Tomilin

. Interaction of the transcription factor YY1 with human poly(ADP-ribosyl) transferase. Biochem Biophys Res Commun. 1997;240:108-11.

33.

Yokoyama

Kawamoto

Mitsuuchi

. Human poly(ADP-ribose) polymerase gene: cloning of the promoter region. Eur J Biochem. 1990;194:521-6.

34.

Ogura

Nyunoya

Takahashi-Masutani

Miwa

Sugimura

Esumi

. Characterization of a putative promoter region of the human poly(ADP-ribose) polymerase gene: structural similarity to that of the DNA polymerase beta gene. Biochem Biophys Res Commun. 1990;167:701-10.

35.

Soldatenkov

Albor

Patel

Dreszer

Dritschilo

Notario

. Regulation of the human poly(ADP-ribose) polymerase promoter by the ETS transcription factor. Oncogene. 1999;18:3954-62.

36.

Prasad

Thraves

Bhatia

Smulson

Dritschilo

. Enhanced poly(adenosine diphosphate ribose) polymerase activity and gene expression in Ewing’s sarcoma cells. Cancer Res. 1990;50:38-43.

37.

Turner

Tutt

Ashworth

. Hallmarks of ‘BRCAness’ in sporadic cancers. Nat Rev Cancer. 2004;4:814-9.

38.

McCabe

Turner

Lord

. Deficiency in the repair of DNA damage by homologous recombination and sensitivity to poly(ADP-ribose) polymerase inhibition. Cancer Res. 2006;66:8109-15.

39.

Wang

Bian

Couch

Zhao

. Poly(ADP-ribose) polymerase-1 down-regulates BRCA2 expression through the BRCA2 promoter. J Biol Chem. 2008;283:36249-56.

40.

Hansel

Rahman

House

. Met proto-oncogene and insulin-like growth factor binding protein 3 overexpression correlates with metastatic ability in well-differentiated pancreatic endocrine neoplasms. Clin Cancer Res. 2004;10:6152-8.

41.

Chen

Vansant

Oades

. Diagnosis of the small round blue cell tumors using multiplex polymerase chain reaction. J Mol Diagn. 2007;9:80-8.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.12 MB