Identification of prostate cancer mRNA markers by averaged differential expression and their detection in biopsies, blood, and urine

Abstract

The advent of serum prostate-specific antigen (PSA) as a biomarker has enabled early detection of prostate cancer and, hence, improved clinical outcome. However, a low PSA is not a guarantee of disease-free status, and an elevated PSA is frequently associated with a negative biopsy. Therefore, our goal is to identify molecular markers that can detect prostate cancer with greater specificity in body fluids such as urine or blood. We used the RT-PCR differential display method to first identify mRNA transcripts differentially expressed in tumor vs. patient-matched nontumor prostate tissue. This analysis led to the identification of 44 mRNA transcripts that were expressed differentially in some but not all tumor specimens examined. To identify mRNA transcripts that are differentially expressed in most tumor specimens, we turned to differential display of pooled tissue samples, a technique we name averaged differential expression (ADE). We performed differential display of mRNA from patient-matched nontumor vs. tumor tissue, each pooled from 10 patients with various Gleason scores. Differentially expressed mRNA transcripts identified by ADE were fewer in number, but were expressed in a greater percentage of tumors (>75%) than those identified by differential display of mRNA from individual patient samples. Differential expression of these mRNA transcripts was also detected by RT-PCR in mRNA isolated from urine and blood samples of prostate cancer patients. Our findings demonstrate the principle that specific cDNA probes of frequently differentially expressed mRNA transcripts identified by ADE can be used for the detection of prostate cancer in urine and blood samples.

Fig. 1.

DD analysis of RNA from tumor vs. patient-matched nontumor prostate tissue: RNA was isolated from prostate tumor (T) and matched nontumor (N) prostate tissue from individual patients and reverse-transcribed with anchor primer H-T11C. The resultant cDNA was amplified with primer H-T11C and arbitrary primer H-AP19 as described in Materials and Methods. The PCRs for each sample were run in duplicate. The amplified products were separated on an extended format 6% polyacrylamide gel. Differentially expressed mRNA transcripts in individual patients are indicated by arrowheads; filled arrowheads indicate overexpressed mRNA transcripts, and open arrowheads indicate down-regulated mRNA transcripts in tumor, as compared with nontumor, prostate tissue from individual patients. Tumors of patients 1 and 2 were of Gleason grade 3 + 3, and those in patients 3 and 4 were of Gleason grade 4 + 4.

Fig. 2.

Averaged differential expression (ADE) of RNA pooled from multiple patients. RNA was isolated from tumor and patient-matched nontumor prostate tissues. DD was performed on individual tumor–nontumor pairs or on pooled tumor vs. pooled nontumor, by using anchor primer H-T11C and arbitrary primer H-AP17. Two DD profiles of pooled RNA revealed one band higher in tumor in 7 of 10 individual tumor–nontumor pairs and another band lower in three of five tumor–nontumor pairs, respectively. These bands were identified as KB208E9 (A) and rp11-442e11 (B), based on their excision, cloning, sequencing, and BLAST analysis. The Gleason grades of the tumors used in this study were 3 + 3 (patients 15, 17, and 19), 3 + 4 (patients 18 and 31), 3 + 5 (patient 23), 4 + 3 (patients 25 and 30), and 4 + 4 (patients 2 and 38). N, nontumor tissue; T, tumor tissue.

Fig. 3.

RT-PCR analysis of genes identified by ADE. RT-PCR using gene-specific primers was carried out to analyze the levels of KB208E9 (A) and rp11-442e11 (B) mRNA in tumors and matched nontumor prostate tissue. GAPDH was included as a housekeeping gene. KB208E9 and GAPDH were amplified by using 25 cycles; rp11-442e11, present at lower levels, was amplified by using 30 cycles. The number of PCR cycles used for each of these transcripts was determined to be in a linear range for semiquantitative analysis. KB208E9 (A) and rp11-442e11 (B) were quantitated by densitometry, normalized to GAPDH, and expressed as a ratio in tumor vs. nontumor (numbers below). (A and B) Data from 10 tumor–nontumor pairs. (C) Summary of data from these 10 patients plus an additional 9 patients; the mean tumor to nontumor ratio of KB208E9 and rp11-442e11 was 1.96 ± 0.263 and 0.89 ± 0.09 (n = 19), respectively.

Fig. 4.

RT-PCR analysis of KB208E9 and rp11-442e11 mRNA in urine of prostate cancer patients. RNA was isolated from individual urine specimens, and RT-PCR was performed with sequence-specific primers for KB208E9, rp11-442e11, and GAPDH. PCRs were performed for 30 cycles. Numbers below represent the ratio of KB208E9 to rp11-442e11, based on densitometry. GAPDH is shown as an indicator of RNA in each sample. (A) The level of KB208E9 (probe a) and rp11-442e11 (probe b) in the urine RNA of a healthy man (HM1) and nine prostate cancer patients (A–I). (B) The level of KB208E9 (probe a) and rp11-442e11 (probe b) in urine RNA of nine healthy men (HM1-HM9). (C) The mean ratio of KB208E9 to rp11-442e11 in healthy men (0.66 ± 0.12; n = 9) vs. prostate cancer patients (4.04 ± 1.67; n = 9). ∗, Approximate value; a more reliable value could not be obtained because of low rp11-442e11 levels in the sample.

Fig. 5.

RT-PCR analysis of KB208E9 and rp11-442e11 mRNA in blood of prostate cancer patients. RNA was isolated from individual blood specimens, and RT-PCR was performed with sequence-specific primers for KB208E9, rp11-442e11, and GAPDH. PCRs were performed for 30 cycles. Numbers below represent the ratio of KB208E9 to rp11-442e11, based on densitometry. GAPDH is shown as an indicator of RNA per sample. (A) The level of KB208E9 (probe a) and rp11-442e11 (probe b) in blood RNA of one healthy man (HM1) and nine prostate cancer patients (A–I). (B) The level of KB208E9 (probe a) and rp11-442e11 (probe b) in blood RNA of nine healthy men (HM1–HM9). (C) The mean ratio of KB208E9 to rp11-442e11 in healthy men (0.74 ± 0.04; n = 9) and prostate cancer patients (2.97 ± 0.42; n = 9).