D-PCa-2: A novel transcript highly overexpressed in human prostate and prostate cancer

Abstract

Identification of genes selectively expressed in tumors or individual tissues is a crucial prerequisite for molecular diagnosis and treatment of cancer by addressing molecular targets. By screening an expression database, we identified the novel gene D-PCa-2 (Dresden prostate carcinoma 2), which is highly overexpressed in normal prostate tissue and prostate carcinoma (PCa). The corresponding transcript contained an open reading frame of 453 nucleotides encoding a putative protein of 150 amino acids. A large part of exon 8 of the D-PCa-2 gene shows strong similarity to the high-mobility-group nucleosomal binding protein 2 (HMGN2) cDNA. The highly specific transcription of the D-PCa-2 gene in normal and malignant prostate tissues and in a few additional tumors was demonstrated by using multiple tissue dot blot, cancer profiling dot blot and real-time PCR analyses. Examination of 18 pairs of tumorous and nontumorous prostate tissues from PCa patients by quantitative RT-PCR revealed D-PCa-2 transcripts in all specimens. The potential usefulness of D-PCa-2 as a sensitive marker for metastatic prostate carcinoma cells in lymph nodes was demonstrated by the detection of one LNCaP cell in 1 × 10⁵ normal lymph node cells using real-time RT-PCR. Examination of 22 lymph nodes from PCa patients either containing metastatic prostate cancer cells or diagnosed as cancer-free was in full concordance with histopathologic diagnoses. These results validate D-PCa-2 as a transcript with high tissue specificity and with a potential application in the diagnosis of PCa. © 2004 Wiley-Liss, Inc.

The use of molecular targets in novel diagnostic approaches and strategies of tumor treatment largely depends on the identification of genes with a tumor- or tissue-restricted expression. Crucial issues in the evaluation of target genes are the degree of tissue specificity, the expression in a high percentage of certain tumor entities, the presence in a large number of malignant cells within an individual tumor and the expression in metastatic cells.

A number of prostate-associated transcripts have been identified by searching for therapeutic targets and diagnostic markers in prostate carcinoma (PCa) in recent years. Different screening techniques were used for their identification, such as expressed sequence tag (EST) database mining,1 serial analysis of gene expression,2 differential display PCR,3, 4 yeast 2 hybrid screen,5 representational difference analysis,6 subtractive cDNA library screening,7, 8 exon trapping,9 DNA chip technology10, 11 and combinations thereof.12

The lack of expression in essential human tissues is a crucial issue when molecular markers or targets are evaluated. Expression of genes in tissues other than prostate might limit or even preclude their use as diagnostic tools or molecular targets because of low signal-to-noise ratios or unwanted side effects. In addition, low-level background expression in lymph node cells prevents the specific and sensitive detection of disseminated tumor cells in draining lymph nodes. Additional features found for some of the recently identified genes might preclude their use as targets in diagnosis or therapy of PCa. For example, expression of prostate acid phosphatase (PAP) and NKX3.1 has been described to be downregulated or even lost during tumor progression and in metastases.13, 14 Prostate-specific antigen (PSA) and prostase showed high homologies with other members of the kallikrein family that are expressed in various normal tissue types.7 Considering the limited use of some of the newly identified genes and the heterogeneity of the cells within an individual tumor, the search for new targets to be used as additional tools in diagnosis and/or treatment of PCa is warranted.

Here, by screening the GeneExpress® (Gene Logic, Inc; Gaithersburg, MD) transcriptome database,15 we identified the highly prostate-specific transcript D-PCa-2. The tissue-restricted expression was shown by hybridization of a radioactively labeled probe to a multiple tissue expression (MTE) array displaying mRNA from a great variety of pooled normal tissues. This result was confirmed by using a sensitive quantitative RT-PCR assay. D-PCa-2 expression was quantified in normal human tissue specimens, cancer cell lines, paired malignant and nonmalignant prostate tissues, as well as in lymph node metastases of PCa. The novel gene D-PCa-2 excels by its highly specific expression in prostate tissue and PCa, which allows the sensitive molecular diagnosis of lymph node PCa metastases.

CDS, coding sequence; D-PCa-2, Dresden prostate carcinoma 2; EST, expressed sequence tag; HMG17, high-mobility-group protein 17; HMGN2, high-mobility-group nucleosomal binding protein 2; HPRT, hypoxanthine phosphoribosyltransferase; IRES, internal ribosome entry site; LC, LightCycler; MTE, multiple tissue expression; NLS, nuclear localization signal; ORF, open reading frame; PAP, prostate acid phosphatase; PCa, prostate carcinoma; PSA, prostate-specific antigen; RACE, rapid amplification of cDNA ends; RPC, retroposed copy; UTR, untranslated region.

MATERIAL AND METHODS

Identification and isolation of D-PCa-2 cDNA sequences

The EST AA503330 was identified by screening the GeneExpress® transcriptome database15 for ESTs with prostate-specific expression pattern and an association with PCa. The data were analyzed using the GeneExpress Software System and tools available on public Web sites (www.ncbi.nlm.nih.gov).

The EST was amplified from a prostate cDNA library (BD Clontech, Heidelberg, Germany) using the primers D-PCa-2_N1 (5′-TCCAAAGGTGAATCTTTTGGTTGGTGA-3′) and D-PCa-2_C1 (5′-GATGGGGGCTGTTCATACAGGGCAAG-3′) with the following cycling profile: 95°C for 3 min; 30 cycles of 95°C for 30 sec, 70°C for 30 sec, 68°C for 30 sec; followed by one round at 68°C for 5 min. The product was ligated into pCR2.1 (Invitrogen, Karlsruhe, Germany) and sequenced.

D-PCa-2 cDNA containing the original EST and the identified open reading frame (ORF) was amplified by the primers D-PCa-2_N1 and D-PCa-2A_C1 (5′-TATGGCCCTTCTCCCCAAAACAACA-3′) using the thermal profile as follows: 95°C for 3 min; 30 cycles of 95°C for 30 sec, 70°C for 30 sec, 68°C for 90 sec; followed by one round at 68°C for 5 min. The PCR products were cloned into pCR2.1 and sequenced.

Rapid amplification of cDNA ends (RACE)

5′ RACE was performed on Marathon Ready cDNA (BD Clontech) from normal prostate using the gene-specific primer D-PCa-2_C1 and the thermal cycling profile of 94°C for 30 sec; 5 cycles of 94°C for 5 sec, 72°C for 4 min; 5 cycles of 94°C for 5 sec, 70°C for 4 min; and 25 cycles of 94°C for 5 sec, 68°C for 4 min. The PCR product was cloned into pCR2.1 and sequenced.

Dot blot analyses

The MTE Array 2 and the Cancer Profiling Array (both from BD Clontech) were hybridized with the ³²P-labeled 270 bp cDNA fragment corresponding to nucleotides 14–283 of EST AA503330 amplified with primers D-PCa-2_N1/D-PCa-2_C1 (Megaprime DNA labeling system; Amersham Biosciences, Freiburg, Germany). The MTE Array 2 was stripped according to the manufacturer's protocol and hybridized with a radioactively labeled ubiquitin probe (control) provided by the manufacturer. Hybridizations were performed according to the provider's instructions using 2 × 10⁶ cpm/ml of the radioactively labeled probe and signals were visualized by phosphoimaging after exposure overnight (Amersham Biosciences).

Prostate cancer patients, tissue samples and cell lines

All primary tumor samples were obtained from prostatectomized PCa patients with informed consent (Table I). We analyzed pairs of tissue samples (primary PCa specimens and autologous nonmalignant prostate tissue) from 18 patients. Histopathologic examination of the tumors was performed according to the UICC classification system from 1997.16 Lymph nodes (n = 22) from bilateral obturatory and internal iliac regions were obtained during prostatectomy of PCa patients and intraoperatively frozen in liquid nitrogen. The lymph nodes were dissected bilaterally; one half was subjected to a conventional histologic examination by hematoxylin-eosin staining and the other half was used for RNA preparation and determination of D-PCa-2 transcripts by quantitative RT-PCR. Lymph nodes from patients without any history and clinical evidence of PCa were obtained with informed consent. The PCa cell lines LNCaP, DU 145 and PC-3 (all from American Type Culture Collection, Manassas, VA) were cultured according to the provider's instructions.

Table I. Pathological and Clinical Parameters16 and Percentage of Tumor Cells in Tissue Samples Analyzed by Real-Time PCR

Patient	Agea	TMN classification	Gleason Score	Tumor grade	Percentage of tumor cells (%)b malignant/nonmalignant sample
N, lymph node metastases; p, pathological examination; T, tumor stage. a Age at the time of radical prostatectomy. b Percentage of tumor cells within the epithelial cells estimated by a pathologist on a representative specimen of the investigated sample. c Grading could not be determined as these patients were hormonally pretreated.
1	69	pT2b pN0 cM0	10	IIIb	80/0
2	75	pT2a pN0 cM0	6	IIb	90/0
3	78	pT2b pN0 cM0	5	IIa	90/0
4	57	PT4 pN0 cM0	6	IIb	90/0
5	62	pT2b pN0 cM0	6	IIb	90/0
6	57	pT2b pN0 cM0	6	IIb	90/0
7	69	pT2a pN0 cM0	6	IIb	80/0
8	72	pT2a pN0 cM0	8	c	75/0
9	65	pT2b pN0 cM0	6	IIb	80/0
10	72	pT2a pN0 cM0	6	IIa	80/0
11	60	pT2a pN0 cM0	6	IIa	80/0
12	54	pT2b pN0 cM0	6	IIb	90/0
13	63	pT2b pN0 cM0	6	IIa	90/0
14	67	pT2b pN0 cM0	6	IIa	90/0
15	68	pT2b pN0 cM0	6	c	60/0
16	64	pT2b pN0 cM0	6	IIb	90/0
17	73	pT2b pN0 cM0	6	IIb	70/0
18	61	pT2b pN0 cM0	7	IIb	90/0

RNA isolation and cDNA synthesis from tissue samples and cell lines

Total RNA was extracted by standard procedures (TRIzol LS Reagent; Invitrogen) and quality was analyzed by agarose gel electrophoresis. After DNA digestion (DNase I; Amersham Biosciences), cDNA synthesis was performed using 4 μg of total RNA and random hexamer primers in a standard 32 μl reaction (Ready to Go You Prime First Strand Kit; Amersham Biosciences).

Quantitative RT-PCR

Tissue specificity of mRNA expression was analyzed by a quantitative LC-based PCR assay in a panel of normalized cDNAs derived from 16 human tissues (Human MTC Panels I and II; BD Clontech). The relative mRNA quantity was determined applying a real-time PCR protocol based on SYBR Green I detection (LC-FastStart DNA Master SYBR Green I; Roche Diagnostics, Mannheim, Germany) using the primer pairs D-PCa-2_N1/D-PCa-2_C1 for amplification of AA503330 and D-PCa-2B_N1 (5′-CCAGTGCCTATGTCCCACCACTGTC-3′)/D-PCa-2B_C1 (5′-TGCCAGCATCAGCTTTTCCCTTCTT-3′) for amplification of a 224 bp fragment within the ORF. The PCR protocol for the D-PCa-2 LC-assays consisted of a predenaturation step (10 min at 95°C) and 50 amplification cycles (15 sec at 95°C, 5 sec at 70°C, 14 sec at 72°C). PSA transcript levels were determined by amplification of a 156 bp product with the primers PSA_N1 (5′-TCTGCGGCGGTGTTCTGGTG-3′) and PSA_C1 (5′-GCGGGTGTGGGAAGCTGTGG-3′) using the same thermal profile.

To quantify the transcript levels in matched malignant and nonmalignant samples and in lymph node specimens from PCa patients as well as in the PCa cell lines DU-145, LNCaP and PC3, 2 μl of the 1:5 diluted cDNA products were used for amplification with primer pair D-PCa-2_N1/D-PCa-2_C1.

The D-PCa-2 transcript number was normalized to the quantity of HPRT transcripts. The SYBR Green I-based quantification of HPRT was performed using the primers HPRT_N1 (5′-CCCTGGCGTCGTGATTAGTGATGAT-3′) and HPRT_C1 (5′-TGCTTTGATGTAATCCAGCAGGTCAGC-3′) applying the same PCR protocol as used for D-PCa-2.

Serial dilutions of plasmid DNA containing the D-PCa-2 and HPRT fragments over 6 log scales (10¹–10⁶ molecules per capillary) were used as internal template standards (calculation via LC quantification software version 3.5; Roche). Each determination was carried out twice for each cDNA sample as independent PCR runs. The molecule ratios of D-PCa-2 to HPRT transcripts were calculated from the mean values.

Determination of specificity and detection limit of quantitative RT-PCR assay for analysis of lymph node specimens

Single cell suspensions were obtained from lymph nodes of patients without evidence for PCa by enzymatic treatment of mechanically crushed tissue with 1 mg/ml Collagenase Type II (Invitrogen) and 0.1 mg/ml DNase I (Sigma-Aldrich, St. Louis, MO). Total numbers of 2,500, 250, or 25 LNCaP tumor cells were added to 2.5 × 10⁶ lymph node cells (1 LNCaP cell per 10,³ 10,⁴ or 10⁵ lymph node cells). Total RNA was extracted according to a standard protocol (Invisorb Spin Cell RNA Mini kit; Invitek, Berlin, Germany). The synthesis of cDNA was performed using up to 1 μg of total RNA and the Reverse Transcriptase MMLV Rh kit (Promega, Mannheim, Germany). Three independent samples were prepared for the 1:10⁴ ratio and 2 for the 1:10⁵ ratio. As a control, RNA was isolated from 2.5 × 10⁶ lymph node cells. The transcript level of D-PCa-2 was determined by quantitative PCR using 2 μl of the 1:2 diluted cDNA samples and primer pair D-PCa-2_N1/D-PCa-2_C1.

RESULTS

D-PCa-2, a transcript selectively expressed in prostate

By screening the GeneExpress® database15 for prostate-specific gene products, a 286 bp EST (Genbank accession AA503330) was identified together with other transcripts that already have been described to be prostate-specific or upregulated in PCa, such as PSA,17 PAP18 and fatty acid synthase.19 The expression profile for AA503330 indicated expression in the majority (> 60%) of normal prostate tissue samples and in almost all (> 90%) PCa specimens.

The cDNA fragment was amplified from a prostate cDNA library by PCR, cloned and sequenced. Hybridization of the radioactively labeled 270 bp PCR product with an MTE array representing pooled mRNA samples from 57 adult human tissues, 8 human cell lines and 7 fetal human tissues revealed exclusive expression in the prostate (Fig. 1a). The array was also hybridized with a ubiquitin cDNA control probe representing 1 of the 8 housekeeping genes used for normalization of the dotted mRNA and fairly consistent hybridization signals were obtained (Fig. 1b).

Figure 1. Dot blot analysis of D-PCa-2 mRNA expression pattern in normal tissues. (a) A 270 bp cDNA fragment localized at the 5′ end of the D-PCa-2 transcript was radioactively labeled and hybridized to the Human MTE Array 2 that provides poly(A)⁺ RNA from 65 human tissues and 8 tumor cell lines normalized to 8 housekeeping genes. After exposure overnight, strong hybridization signals were only revealed in prostate tissue. (b) The array was stripped and hybridized with a ubiquitin cDNA control probe (provided by the manufacturer) representing 1 of the 8 housekeeping genes used for normalization of the dotted mRNA and fairly consistent hybridization signals were obtained after exposure overnight. A1, whole brain; B1, cerebral cortex; C1, frontal lobe; D1, parietal lobe; E1, occipital lobe; F1, temporal lobe; G1, paracentral gyrus of the cerebral cortex; H1, pons; A2, cerebellum left; B2, cerebellum right; C2, corpus callosum; D2, amygdala; E2, nucleus caudatus; F2, hippocampus; G2, medulla oblongata; H2, putamen; B3, nucleus accumbens; C3, thalamus; E3, spinal cord; A4, heart; B4, aorta; C4, atrium left; D4, atrium right; E4, ventricle left; F4, ventricle, right; G4, interventricular septum; H4, apex of the heart; A5, esophagus; B5, stomach; C5, duodenum; D5, jejunum; E5, ileum; F5, ilocecum; G5, appendix; H5, colon ascendens; A6, colon transversum; B6, colon descendens; C6, rectum; A7, kidney; B7, skeletal muscle; C7, spleen; D7, thymus; E7, peripheral blood leukocyte; F7, lymph node; G7, bone marrow; H7, trachea; A8, lung; B8, placenta; C8, bladder; D8, uterus; E8, prostate; F8, testis; G8, ovary; A9, liver; B9, pancreas; C9, adrenal gland; D9, thyroid gland; E9, salivary gland; A10, leukemia, HL-60; B10, HeLa S3; C10, leukemia, K-562; D10, leukemia, MOLT-4; E10, Burkitt's lymphoma, Raji; F10, Burkitt's lymphoma, Daudi; G10, colorectal adenocarcinoma, SW480; H10, lung carcinoma, A549; A11, fetal brain; B11, fetal heart; C11, fetal kidney; D11, fetal liver; E11, fetal spleen; F11, fetal thymus; G11, fetal lung.

By screening the Genbank database and by custom EST clustering using the Web-based ESTCluster tool of the Heidelberg Unix Sequence Analysis Resources (genome.dkfz-heidelberg.de), we extended the sequence information in silico. Further elongation of the sequence was achieved by RACE and was supported by the existence of a consensus TATA box (TATAAAA) on the corresponding genomic DNA at positions −34 to −28 providing a putative start point of transcription (Fig. 2a). The resulting sequence was designated D-PCa-2 (Dresden prostate cancer 2, Genbank accession AY271965) and contained an ORF of 453 nucleotide coding for a putative protein of 150 amino acids. The cDNA was mapped to chromosome 15q15.1 by genomic BLAST analysis. Conceptual translation of the ORF did not show any significant homology to known proteins, but revealed 2 potential monopartite NLS regions at positions 92–98 and 127–130 (Fig. 2a).

Figure 2. Sequences of D-PCa-2 and its splice variants. (a) Nucleotide sequences and conceptual translations of D-PCa-2 (AY271965) and its splice variants (AY271964, AY271966, AY271967 and AK096745) are displayed together with exon boundaries. The genomic region directly located upstream from the determined 5′ end of the transcript contains a consensus TATA box motif (underlined) at nucleotide position −38 to −32 (+1 is the starting point of transcription). Conceptual starting ATGs are in bold letters and are doubly underlined. The alternative 3′ splice acceptor site in exon 8 is in bold letters and underlined. Two potential monopartite NLS at amino acids 92–98 and 127–130 are shown in bold letters. A potential polyadenylation signal is at position 1896. The stretch of high homology with HMGN2 is shown in italics. (b) Schematic representation of D-PCa-2 and its splice variants. The 270 and 224 bp sequences used as hybridization probes and target sequences in quantitative RT-PCR are displayed. The putative ORFs are shown in gray. D-PCa-2, AY271966 and AK096745 contain a partially deleted exon 8 resulting from usage of an alternative 3′ splice acceptor site.

Using primers corresponding to the 5′ and 3′ ends of the identified cDNA (D-PCa-2_N1 and D-PCa-2A_C1, respectively), we identified 4 additional transcript variants of D-PCa-2 that were characterized as products of differential splicing by genomic BLAST analysis (Fig. 2b). Three of them are novel cDNAs that we submitted to Genbank (Genbank accessions AY271964, AY271966 and AY271967); one corresponds to a known cDNA clone isolated from a prostate library (Genbank accession AK096745). D-PCa-2 and each of the 4 splice variants contain the originally identified EST sequence that consists of parts of exons 1 and 3 spliced together (Fig. 2b). D-PCa-2, AY271964, AY271966 and AY271967 lack exon 2 and therefore contain the EST AA503330 as consecutive sequence spanning exons 1 and 3, whereas in AK096745, the AA503330 sequence is interrupted by exon 2. In addition, the ORFs of all cDNA variants comprise the putative ORF initially identified in D-PCa-2. Due to the splicing events, the ORFs of 3 of the variants are elongated at their 5′ ends and consequently the conceptual translations for AY271964 and AY271967 have the N-terminal extension, MEPWAMRALDFADESGSVSCKD, whereas AK096745 displays the shorter N-terminal extension, MKEDESGSVSCKD (Fig. 2a). In D-PCa-2 and 2 variants, exon 8 is partially deleted at its 5′ end (Fig. 2b), obviously resulting from an alternative splicing event to an alternative 3′ acceptor site in exon 8 located 18 bp upstream from the original splice site. Insertion of 15 additional nucleotides into the mRNA by use of the proximal 3′ acceptor site did not change the open reading frame in the resulting transcript but would result in the insertion of 5 amino acids (ALDFA) into the putative protein sequence (Fig. 2a). All exon/intron boundaries in the D-PCa-2 gene follow the gt-ag rule.20 The alternative distal 3′ acceptor site in exon 8 displays the extended consensus sequence (py)₁₀ncag.

Homology searches at the nucleotide level revealed similarity of a part of exon 8 to the high-mobility-group nucleosomal binding protein 2 (HMGN2; formerly high-mobility-group 17) cDNA. Nevertheless, a putative D-PCa-2 protein did not have any homology to this nucleosome-binding protein as it originated from a different frame, nor does it show any homology to other proteins.

Expression quantification of D-PCa-2 transcript in various human tissues

The hybridization experiment using the radioactively labeled 270 bp probe (corresponding to EST AA503330) present in all variants suggested prostate-specific expression of D-PCa-2 and all its splice variants in summa. In the literature, it was shown that for some genes only certain splice variants are expressed in a tissue-specific manner.21, 22 Consequently, in our experiments, we did not distinguish between individual expression levels of single D-PCa-2 splice variants in various tissues but rather assessed the overall expression profile of all D-PCa-2 variants with highest possible sensitivity. To this end, the expression level of the D-PCa-2 transcripts was determined in a normalized panel of 16 human tissues by quantitative real-time PCR. Assays were performed with 2 different primer pairs that specifically amplify either the 270 bp fragment of the original EST (with an elongation time also sufficient for detection of variant AK09745) or a 224 bp fragment within the putative ORF (Fig. 2b) common to all D-PCa-2 splice variants. Both assays clearly demonstrated the prostate-restricted expression pattern. As shown in Figure 3(a), transcripts were absent or only minimally expressed in human tissues other than prostate. Because of the marked differences between the expression levels in prostate tissue and other normal tissues, the PCR data are also given at different scales. As one can see from Figure 3(a), the expression in normal tissues varied between undetectable (in brain, kidney and small intestine) and about 20 transcripts (in heart, liver and leukocytes) per 2 μl of normalized cDNA. It can be seen clearly that D-PCa-2 transcript levels in the nonprostatic tissue with highest expression (heart) were still about 2,000-fold lower in comparison to prostate tissue. This ratio is crucial when assessing the suitability of a tissue-specific marker. In order to estimate the degree of tissue specificity of D-PCa-2 expression, it was compared to the highly prostate-restricted marker PSA. As shown in Figure 3(b), while the absolute number of PSA transcripts found in prostate tissue was quite high (12,850,000 transcripts per 2 μl of normalized cDNA), its relative specificity was lower than that of the D-PCa-2 transcript. A marked PSA expression was also found in small intestine, at a level of about 450 times lower compared to prostate tissue (28,575 transcripts per 2 μl of normalized cDNA).

Figure 3. Real-time PCR analysis of D-PCa-2 mRNA expression in human tissues. (a) The tissue specificity of D-PCa-2 mRNA was determined by an SYBR Green I-based quantitative PCR assay in normalized cDNA samples derived from 16 different human tissues. In this PCR assay, the 270 bp cDNA fragment spanning the first 2 exons was amplified. High transcript levels were only found in pooled prostate tissue, whereas transcripts were absent or minimally expressed in other human tissues. For clarification, expression in tissues other than prostate is also shown at a different scale (see insert). (b) Expression pattern of PSA in the same normalized cDNA panel as determined by quantitative PCR with SYBR Green I. For clarification, expression in tissues other than prostate is again shown at a different scale (see insert). The results represent the means of 2 independent LC runs. Bars = SE.

D-PCa-2 expression in matched samples of malignant and nonmalignant prostate tissues and tumor cell lines

Since gene expression is often differentially regulated in normal and the corresponding malignant tissue, expression of D-PCa-2 was further characterized by hybridization of the 270 bp probe to a Cancer Profiling Array. This membrane array provides normalized cDNA samples of matched specimens (normal/malignant) from a variety of tissues. D-PCa-2 transcripts were detected in all 4 normal prostate tissues and in the matched malignant samples (Fig. 4). The transcript quantity was upregulated in the malignant prostate specimens of 3 prostate tissue pairs (sample pairs 2, 3 and 4). In addition, high transcript quantities were found in 1 of 19 renal tumors (a renal oncocytoma characterized as a benign tumor of the kidney), but in none of the normal kidney samples. Weak hybridization signals were also detected in one uterus adenocarcinoma, one adenoma of the colon and one adenocarcinoma of the lung, but in none of the corresponding normal tissues.

Figure 4. Dot blot analysis of D-PCa-2 mRNA expression in matched samples of malignant and nonmalignant tissues. The 270 bp cDNA fragment of D-PCa-2 was radioactively labeled and hybridized to the Cancer Profiling Array that provides normalized cDNA samples from cancerous tissues and corresponding normal tissues derived from the same patients. Strong signals were detected in all malignant and nonmalignant prostate specimens and in a renal oncocytoma. N, nontumorous samples; T, tumorous samples. The gray fields do not contain cDNA samples. The fields right to them represent metastatic tissue from the patient of the upper line.

In order to validate expression of D-PCa-2 in a larger group of PCa patients and to evaluate a possible correlation of clinical or pathologic parameters of tumor progression with D-PCa-2 expression, transcripts were quantified in paired malignant and nonmalignant prostate tissue specimens from 18 PCa patients using a real-time RT-PCR assay. As documented in Figure 5, D-PCa-2 transcripts could be detected in all normal and all malignant prostate tissue samples. Staging and grading of the analyzed tumor specimens are summarized in Table I. The relative expression level was found to be highly variable among patients' samples and the relation of transcript quantity in the tumorous tissues to that in the nonmalignant samples did not correlate with pathologic and clinical parameters of cancer progression (Fig. 5).

Figure 5. Real-time PCR analysis of D-PCa-2 mRNA expression in matched samples of malignant and nonmalignant prostate tissues and in the PCa cell line LNCaP. To quantify the D-PCa-2 mRNA expression in 18 paired cDNA samples of tumorous (black columns) and nontumorous (white columns) prostate tissue, the 270 bp fragment was amplified in an SYBR Green I-based LC assay. The transcript quantity was normalized to the expression level of HPRT. For demonstration of relatively low D-PCa-2 expression in LNCaP cells, transcript quantities in LNCaP cells and in one tissue pair are also shown at a different scale (see insert).The results represent the means of 2 independent LC runs. Bars = SE.

Quantitative analysis of D-PCa-2 transcripts in 3 PCa cell lines revealed expression in LNCaP and the lack of D-PCa-2 transcripts in DU145 and PC-3. Remarkably, the relative expression in LNCaP is rather low when compared to the paired normal and malignant prostate samples (Fig. 5).

Analysis of D-PCa-2 expression in lymph node metastases of PCa and sensitivity of tumor cell detection in a high background of lymph node cells

To assess the potential suitability of D-PCa-2 transcripts as a molecular marker for the detection of lymph node metastases, we performed quantitative RT-PCR on normal lymph nodes from patients without evidence of PCa and on histologically negative or positive lymph nodes from PCa patients. The cDNA samples prepared from 4 normal lymph nodes and a cDNA library of normal lymph node tissue pooled from 12 individuals were completely negative for D-PCa-2 transcripts. Fifteen lymph nodes with pathologically confirmed metastases were clearly positive for D-PCa-2 mRNA (Fig. 6a), whereas 7 lymph nodes characterized as negative according to histologic examination did not express D-PCa-2 (data not shown).

Figure 6. Suitability of the LC-based RT-PCR assay for D-PCa-2 for a sensitive detection of lymph node metastases of PCa. (a) Fifteen lymph nodes with histologically confirmed PCa metastases were analyzed for D-PCa-2 expression by an SYBR Green I-based LC assay amplifying the 270 bp fragment. The transcript quantity was normalized to the expression level of HPRT. The results represent the means of 2 independent LC runs. Bars = SE. (b) Determination of specificity and sensitivity of the D-PCa-2-based RT-PCR assay. Lymph node cells from a patient without evidence of a PCa were mixed with LNCaP cells at a ratio of 1 LNCaP cell per 10³, 10⁴, or 10⁵ lymph node cells and RNA was prepared. The RT-PCR assay was sensitive to detect 1 LNCaP cell in a background of 10⁵ lymph node cells.

To determine the sensitivity of the D-PCa-2-based RT-PCR assay for the detection of a low number of PCa cells in a high background of lymph node cells, LNCaP cells were mixed with a preparation of 2.5 × 10⁶ normal lymph node cells at ratios of 1 LNCaP cell per 10,³ 10,⁴ or 10⁵ lymph node cells. Although D-PCa-2 was expressed at relatively low levels in LNCaP, transcripts could still be detected at a ratio of 1:10⁵ but were not found in a cDNA preparation from the same number of lymph node cells (Fig. 6b).

DISCUSSION

High-density microarrays with a large collection of cDNAs or gene-specific oligonucleotides have been used for the assessment of PCa gene expression profiles, for the development of gene signatures to distinguish between benign prostate hyperplasia and PCa,23, 24, 25 as well as for the delineation of prognostic biomarkers in PCa26, 27 during the last years. In addition, gene chips have proven to be a powerful tool for the identification of prostate-associated transcripts as therapeutic targets and diagnostic markers in PCa.10, 11 Here, we screened the GeneExpress® transcriptome database for transcripts showing high specificity for prostate tissue and expression in prostate tumors. The GeneExpress® contains expression data of a variety of tissue samples profiled on Affymetrix gene chips.15 Using the GeneExpress Software System providing different algorithms for expression data analysis, we identified the EST AA503330, which was indicative for the novel gene D-PCa-2. Further experimental evaluation of the D-PCa-2 expression pattern in 57 different human tissues was done by mRNA dot blot analysis and demonstrated almost exclusive transcription of D-PCa-2 in prostate tissue.

Using primers corresponding to the 5′ and 3′ ends of the identified cDNA, we identified 4 additional splice variants of D-PCa-2 that all comprised the sequence of the originally identified EST AA503330 and arose from differential splicing of 8 exons as identified by genomic BLAST analysis. All exon/intron boundaries in the D-PCa-2 gene follow the gt-ag rule.20 Interestingly, D-PCa-2 and 2 of the splice variants (AY271966 and AK096745) displayed a partial deletion of 15 nucleotides at the 5′ end of exon 8, probably arising from the usage of an alternative 3′ splice acceptor site. The distal 3′ acceptor site used alternatively in exon 8 contained the extended consensus sequence (py)₁₀ncag. Similar alternative splicing events arising from alternative 3′ acceptor sites are described for an increasing number of mRNAs, for example, for transforming growth factor alpha and RNA-specific adenosine deaminase.28, 29

Sequence analysis of the 5 identified splice variants of D-PCa-2 revealed a common ORF of 453 nucleotides encoding a putative protein of 150 amino acids that lacks significant similarities to known proteins. Due to the splicing events (which did not change the putative reading frame in either variant), the ORF is elongated by 66 nucleotides at the 5′ ends of splice variants AY271964 and AY271967 and by 39 nucleotides at the 5′ end of splice variant AK096745.

Although no homology was detected for a putative D-PCa-2 protein, at the nucleotide level the last two-thirds of exon 8 (comprising the potential ORF) displayed striking homology (93% identity) to the cDNA of the nonhistone chromosomal protein HMGN2. HMGN2 belongs to one of the largest known retropseudogene families, and the human genome contains multiple copies of sequence homologues. Some of these mobile elements may be transposed next to functional promoters and might have acquired important roles during the evolution of the human genome.30, 31 Recently, it was shown that 65 retroposed copies (RPCs) of HMGN2 with identities ranging from 82% to 98% are dispersed throughout the human genome and that chromosome arm 15q represents a region of especially high density of HMGN2 RPC integration.32 Most RPCs, if expressed at all, give rise to novel mutated HMG proteins or to modular combinations of HMG domains with domains from other genes. However, the occurrence of ORFs unrelated to the HMG ORF resulting from different frames or even a different orientation was also described. In addition, some of the RPCs were shown to be included as part of the ORF of a different gene, thereby contributing to the coding sequence (CDS).32 In fact, in their bioinformatic analysis, Strichman-Almashanu et al.32 predicted a HMGN2 RPC located at 15q15.2a to be potentially expressed with an ORF different from the HMGN2 ORF.

In addition to the described putative CDS, all 5 transcripts have unusually long 5′ UTR regions that contain multiple splice product-dependent AUG codons in different frames that produce very short ORFs. Consequently, the putative CDS is likely to be translated inefficiently by the classical ribosome scanning mechanism. An internal ribosome entry site (IRES) could contribute an internal initiation of translation as published for c-myc,33 vascular endothelial growth factor34 and the voltage-gated potassium channel Kv1.4.35

The most striking feature of the D-PCa-2 gene is its highly tissue-specific transcription and the potential use of the transcript as a diagnostic marker. Even noncoding mRNAs can be expressed in a prostate-restricted way3, 4 and might represent useful marker molecules for diagnosis of PCa as was demonstrated for DD3.36 However, for a number of genes, it was reported that only some splice variants are expressed in a tissue-specific manner.21 For example, the 4.5 kb transcript of the ABC transporter MRP9 is expressed in breast, breast cancer and testis, whereas the 1.3 kb variant is found in additional tissues.22 The existence of splice variants that are expressed in several normal tissues was also demonstrated for the widely recognized PCa marker DD3 recently.37 Nevertheless, detection of DD3 exon 4 obviously remains to be a specific and sensitive marker for PCa.36, 38 Here, we demonstrated tissue-specific expression of D-PCa-2 for all transcript variants by using a radioactively labeled probe representing a sequence that is common to all described splice variants.

Data on the tissue specificity of many of the prostate-associated transcripts published in recent years were based on Northern blot or mRNA dot blot analysis, techniques that may be inadequate in their ability to exclude expression in tissues other than prostate.39, 40 The lack of expression in essential human tissues is a crucial issue when molecular markers or targets are evaluated. To pinpoint the tissue specificity of D-PCa-2 expression with high sensitivity, we complemented mRNA dot blot analysis with a more sensitive quantitative RT-PCR assay. Both techniques demonstrated the almost exclusive expression of all D-PCa-2 transcript variants in prostate tissue with nearly 2,000-fold higher copy numbers present in prostate tissue than in any other essential normal tissue.

The specificity of a tissue marker depends on its relative expression level when compared to the tissue with second highest expression level. As expected, the number of PSA transcripts in prostate tissue was quite high, but there was also significant expression detectable in small intestine leading to a relative expression level of 450. In terms of tissue specificity, this ratio has to be compared to the relative expression level of 1,966 determined for D-PCa-2.

Expression of D-PCa-2 mRNA was thoroughly analyzed in paired malignant and nonmalignant samples from prostate and other tissues using quantitative RT-PCR and dot blot analysis. Transcripts were detected in all tested tumorous and nontumorous prostate tissue samples and in a low percentage of other tumors but were not detected in any of the normal tissues other than prostate. Obviously, the prostate restriction of D-PCa-2 expression demonstrated here for the normal tissues is not completely retained when investigating tumorous tissue samples. Although D-PCa-2 transcripts were detected in few tumor samples apart from PCa, expression in one oncocytoma was even higher than in one of the normal prostate samples on the cancer profiling array.

Quantitative determination of expression did not reveal a general upregulation of D-PCa-2 mRNA in prostate tumors when compared to the corresponding nontumorous tissue. This reflects the fact that, during our initial screening of the expression database, we put emphasis on tissue selectivity of expressed transcripts rather than on upregulation of expression in tumorous vs. normal prostate tissue. The expression level in the malignant specimens did not correlate with the stage or grade of the tumors. This result may be due to the heterogeneity of expression of the cells within a tumor specimen. The expression of D-PCa-2 is not modulated by R1881 in the androgen-dependent cell line LNCaP (data not shown). In addition, D-PCa-2 was detected in lymph nodes containing metastatic cells originating from PCa, whereas cDNA prepared from 4 normal lymph nodes as well as a cDNA library of normal lymph node tissue pooled from 12 individuals were completely negative for D-PCa-2 transcripts.

Expression of genes with a unique tissue selectivity such as D-PCa-2 provides the opportunity to detect small numbers of tumor cells disseminated into metastasis-prone tissues such as lymph nodes draining tumorous tissues. The clear detection of one LNCaP cell in a number of 1 × 10⁵ lymph node cells by determining the D-PCa-2 transcripts using real-time PCR indicates the sensitivity and specificity of this approach. The sensitivity of PCa cell detection might be even higher since LNCaP expresses D-PCa-2 at a clearly lower level than all analyzed PCa samples.

In conclusion, based on its comprehensive expression in primary prostate tumors of any progression state and on its androgen independence, D-PCa-2 appears to be a promising novel candidate target for the molecular tracking of small numbers of tumor cells in lymph nodes.

Acknowledgements

The authors thank Sabine Heinicke, Barbara Uteß, Sandra Schwind, Karin Gunther, Katja Richter, Romy Kranz and Jorg Stade for technical assistance, Michael Behrend for preparation of lymph nodes and help with androgen stimulation experiments and Prof. Gustavo Barreton (Department of Pathology, Medical Faculty, Technical University Dresden) for providing histopathologic data of tissue specimens. Supported by the Wilhelm-Vaillant-Stiftung, München, Germany (to B.W.), and grant 1999.009.2 from the Wilhelm-Sander-Stiftung, München, Germany (to A.M., M.S., E.P.R. and M.P.W.).