Factors that drive the increasing use of FFPE tissue in basic and translational cancer research

The decision to use 10% neutral buffered formalin fixed, paraffin embedded (FFPE) archival pathology material may be dictated by the cancer research question or analytical technique, or may be governed by national ethical, legal and social implications (ELSI), biobank, and sample availability and access policy. Biobanked samples of common tumors are likely to be available, but not all samples will be annotated with treatment and outcomes data and this may limit their application. Tumors that are rare or very small exist mostly in FFPE pathology archives. Pathology departments worldwide contain millions of FFPE archival samples, but there are challenges to availability. Pathology departments lack resources for retrieving materials for research or for having pathologists select precise areas in paraffin blocks, a critical quality control step. When samples must be sourced from several pathology departments, different fixation and tissue processing approaches create variability in quality. Researchers must decide what sample quality and quality tolerance fit their specific purpose and whether sample enrichment is required. Recent publications report variable success with techniques modified to examine all common species of molecular targets in FFPE samples. Rigorous quality management may be particularly important in sample preparation for next generation sequencing and for optimizing the quality of extracted proteins for proteomics studies. Unpredictable failures, including unpublished ones, likely are related to pre-analytical factors, unstable molecular targets, biological and clinical sampling factors associated with specific tissue types or suboptimal quality management of pathology archives. Reproducible results depend on adherence to pre-analytical phase standards for molecular in vitro diagnostic analyses for DNA, RNA and in particular, extracted proteins. With continuing adaptations of techniques for application to FFPE, the potential to acquire much larger numbers of FFPE samples and the greater convenience of using FFPE in assays for precision medicine, the choice of material in the future will become increasingly biased toward FFPE samples from pathology archives. Recognition that FFPE samples may harbor greater variation in quality than frozen samples for several reasons, including variations in fixation and tissue processing, requires that FFPE results be validated provided a cohort of frozen tissue samples is available.

OECI TuBaFrost Tumor Biobanking

OECI TuBaFrost harbors a complete infrastructure for the exchange of frozen tumor samples between European countries. OECI TuBaFrost consists of: • A code of conduct on how to exchange human residual samples in Europe • A central database application accessible over the Internet ( www.tubafrost.org ) where data can be uploaded and searched from samples that can be selected and ordered • Access rules with incentives for collectors • Standardization needed to enable the analysis of high quality samples derived from different centers • Virtual Microscopy to support sample selection with difficult pathology

The entire infrastructure was, after completion, which was entirely financed by the European Commission, implemented in the OECI. But so far it has not been used to its capacity. A recent survey held amongst the OECI members shed light on the causes. The main conclusion is that all responders see OECI TuBaFrost as a good platform for exchange of samples, however, the biggest bottleneck found was that potential users are too unfamiliar with the communication between their own biobank tracking system and the TuBaFrost central database application. Therefore, new future plans are drawn. In addition, new infrastructure plans have been developed and the first preparatory steps have been set. For biobanks the BBMRI project has started aiming for Pan-European Biobanking and Biomolecular Resources Research Infrastructure.

Involvement of cancer-activating genes on chromosomes 7 and 8 in esophageal (Barrett's) and gastric cardia adenocarcinoma

The incidence of adenocarcinomas of the distal esophagus (Barrett's esophagus) and proximal stomach (gastric cardia) has increased rapidly over the past decades. In contrast to this dramatic increase, genetic knowledge is sparse.

MATERIALS AND METHODS

We investigated genomic amplification on chromosomes 7 and 8 by comparative genomic hybridization (CGH) and protein expression of relevant oncogenes (EGFR, HGF, MET, CTSB, MYC) by immunohistochemistry (IHC) in 22 esophageal and 22 gastric cardia carcinomas.

RESULTS

The CGH and IHC patterns were very similar for the two cancer locations. IHC showed positive immunostaining in 93% of the adenocarcinomas for at least one of the investigated genes, whereas CGH disclosed genomic gains on chromosome 7 and/or 8 in 80%.

CONCLUSION

Cancer-activating genes on chromosomes 7 and 8 are frequently involved in gastro-esophageal junction adenocarcinomas. Moreover, the similarities in chromosomal changes and protein expression patterns strongly suggest that esophageal and gastric cardia adenocarcinomas have a shared etiology. This is in agreement with studies addressing gastroesophageal reflux disease and intestinal metaplasia at these locations.

Molecular cytogenetic evaluation of gastric cardia adenocarcinoma and precursor lesions

Analyses of cancer incidence data in the United States and Western Europe revealed steadily rising rates over the past decades of adenocarcinomas of the esophagus and gastric cardia. Genetic information on gastric cardia adenocarcinoma and its preneoplasias is sparse. We have used comparative genomic hybridization to obtain a genome-wide overview of 20 archival gastric cardia adenocarcinomas and 10 adjacent preneoplastic lesions (4 metaplasias, 1 low-grade dysplasia, 5 high-grade dysplasias). Multiple genetic alterations were discriminated in all adenocarcinomas. Frequent loss (> or =25% of all tumors) was detected, in decreasing order of frequency, on 5q, 18q, 4q, 3p, 9p, 2q, 11q, 14q, 21q, 4p, 9q, 16q, 1p, and 8p. Frequent gain (> or =25% of all tumors) was disclosed, in decreasing order of frequency, on 20q, 7p, 8q, 1q, 7q, 20p, 17q, 13q, Xp, 6q, 8p, 19q, 5p, 6p, and Xq. Loss of the Y chromosome was found in 60% of male cases. High level amplification was frequently (>10% of all tumors) detected on 7q21, 8p22, 12p11.2, 17q12-q21, and 19q13.1-q13.2. The precursor lesions showed multiple aberrations in all high-grade dysplasias, whereas few genetic changes were discerned in LGD and metaplasias. High level amplifications were also found in high-grade dysplasias, ie, on 7q21, 8p22, and 17q12-q21. Moreover, the percentage of aberrations was not significantly different for invasive carcinomas or high-grade dysplasias. Approximately 70% of the precursor aberrations were also present in the adjacent carcinoma. Minimal overlapping regions in the preneoplasias included loss on 18q12-q21 and gains on 8q23 and 17q12-q21, suggesting involvement of genes residing in these regions. In conclusion, we have (i) created a map of genetic alterations in gastric cardia adenocarcinomas and (ii) provided evidence for the presence of a metaplasia-dysplasia-carcinoma sequence in this poorly understood type of cancer.

Genomic alterations in malignant transformation of Barrett's esophagus

The incidence of adenocarcinoma in Barrett's esophagus has been increasing rapidly over the past decades. Neoplastic progression is characterized by three well-defined premalignant stages: metaplasia, low-grade dysplasia, and high-grade dysplasia. A genome-wide overview, based on comparative genomic hybridization, was performed, evaluating 30 Barrett's adenocarcinomas and 25 adjacent precursors, i.e., 6 metaplasias, 9 low-grade dysplasias, and 10 high-grade dysplasias. The frequency of losses and gains significantly increased in the subsequent stages of malignant transformation. Losses of 5q21-q23, 9p21, 17p12-13.1, 18q21, and Y were revealed in low-grade dysplasias. This was followed by loss of 7q33-q35 and gains of 7p12-p15, 7q21-q22, and 17q21 in high-grade dysplasias along with high-level amplification (HLA) of 7q21 and 17q21. In the invasive cancers, additional losses of 3p14-p21, 4p, 4q, 8p21, 13q14-q31, 14q24.3-q31, 16q21-q22, and 22q as well as gains of 3q25-q27, 8q23-24.1, 12p11.2-12, 15q22-q24, and 20q11.2-q13.1 were distinguished along with HLAs of 8p12-p22 and 20q11.2-q13.1. Approximately one-third of the alterations in the dysplasias were also found in the adjacent adenocarcinomas, illustrating that multiple clonal lineages can be present in Barrett's esophagus. Novel findings include loss on 7q, gain on 12p, and the observation of several HLAs in high-grade dysplasias. Furthermore, loss of 7q33-q35 was found to represent a significant distinction between low-grade and high-grade dysplasia (P = 0.01), whereas loss of 16q21-q22 and gain of 20q11.2-q13.1 were disclosed to significantly discriminate between high-grade dysplasia and adenocarcinoma (P = 0.02 and P = 0.03, respectively). This inventory of genetic aberrations increases our understanding of malignant transformation in Barrett's esophagus and might provide useful biomarkers for disease progression.

Evidence for a cytoskeleton attachment domain at the N-terminus of the NF2 protein

Neurofibromatosis type 2 is a hereditary cancer syndrome characterized by the development of bilateral vestibular schwannomas. Underlying the disease are inactivating mutations of the NF2 tumor suppressor gene, located on chromosome 22, encoding a 595-amino-acid protein. The NF2 protein, also known as merlin or schwannomin, is reported to act as a membrane-cytoskeleton linking protein. This assumption is based on the homology of the NF2 protein to a group of band 4.1-related proteins, ezrin, radixin, and moesin. The cytoskeletal association of the NF2 protein has in part been confirmed by its ability to resist extraction from cells by nonionic detergents. We performed detergent extraction on COS cells transfected with NF2 cDNA constructs. The extracts were analyzed by Western blotting and immunofluorescent staining with monoclonal anti-NF2 antibodies. The results provide evidence for a high-affinity cytoskeleton attachment domain at amino acids 29-131 and a putative lower affinity domain between amino acids 321 and 470.

A G-->A transition creates a branch point sequence and activation of a cryptic exon, resulting in the hereditary disorder neurofibromatosis 2

We describe a G-->A transition within intron 5 of the NF2 gene. This mutation creates a consensus splice branch point sequence. To our knowledge this is the first report of a mutation that creates a functional branch point sequence in a human hereditary disorder. The new branch point sequence is located 18 bp upstream of a consensus splice acceptor site. A consensus splice donor site is found 106 bp 3' of the acceptor site. Asa consequence the G-->A transition results in an alternatively spliced mRNA containing an additional exon 5a of 106 bp derived from intron sequences. We cloned the mutant cDNA and show that due to an in-frame stop codon the cDNA codes for a truncated NF2 protein. The mutation was observed in three affected members of an NF2 family. In a tumour of one of the family members both alternatively spliced and wild-type mRNA were found, although the wild-type allele of the gene is absent due to an interstitial deletion on chromosome 22. We also show that immunoprecipitations reveal the presence of full-length wild-type NF2 protein in the tumour lysate. These data support the hypothesis that some degree of normal splicing of the mutant precursor RNA is taking place. It is therefore likely that this residual activity of the mutant allele explains the relatively mild phenotype in the family. These data also indicate that complete inactivation of the gene is not required for tumour formation.

Identification and characterization of NF1-related loci on human chromosomes 22, 14 and 2

Neurofibromatosis type 1 (NF1) is a frequent hereditary disorder. The disease is characterized by a very high mutation rate (up to 1/10000 gametes per generation). NF1-related loci in the human genome have been implicated in the high mutation rate by hypothesizing that these carry disease-causing mutations, which can be transferred to the functional NF1 gene on chromosome arm 17q by interchromosomal gene conversion. To test this hypothesis, we want to identify and characterize the NF1-related loci in the human genome. In this study, we have localized an NF1-related locus in the most centromeric region of the long arm of chromosome 22. We demonstrate that this locus contains sequences homologous to cDNAs that include the GAP-related domain of the functional NF1 gene. However, the GAP-related domain itself is not represented in this locus. In addition, cosmids specific to this locus reveal, by in situ hybridization, NF1-related loci in the pericentromeric region of chromosome arm 14q and in chromosomal band 2q21. These cosmids will enable us to determine whether identified disease-causing mutations are present at the chromosome 22-associated NF1-related locus.

Neurofibromatosis type 2 protein co-localizes with elements of the cytoskeleton

The product of the neurofibromatosis type 2 (NF2) tumor suppressor gene is a 595-amino-acid protein bearing resemblance to a family of band-4.1-related proteins. These proteins, including ezrin, radixin, and moesin, probably function as molecular linking proteins, connecting the cytoskeleton to the cell membrane. On the grounds of the homology to the ezrin, radixin, and moesin proteins and on the basis of its predicted secondary structure, the NF2 protein is also thought to act as a cytoskeleton-cell membrane linking protein. Using monoclonal antibodies to amino- and carboxyl-terminal synthetic NF2 peptides we demonstrate the co-localization of the NF2 protein with elements of the cytoskeleton in a COS cell model system and in cultured human cells. Furthermore, the presence of the NF2 protein in tissue sections is shown. The monoclonal antibodies specifically stain smooth muscle cells and the stratum granulosum of the human epidermis. In cultured smooth muscle cells the NF2 protein co-localizes with actin stress fibers. Immunoelectron microscopy demonstrates the presence of the NF2 protein associated with keratohyalin granules and to a lesser extent with intermediate filaments in the human epidermis. We conclude that the NF2 protein is indeed associated with multiple elements of the cytoskeleton.

Characterization of four novel CAG repeat-containing cDNAs

Stretches of CAG nucleotides coding for the amino acid glutamine are an important feature of many transcription factors and genes that are involved in neurodegenerative disorders. In an attempt to isolate CAG repeat-containing cDNAs expressed in nervous tissue, we screened a human fetal brain cDNA library with a probe containing a CAG repeat. Five different clones were characterized and found to contain CAG repeats. Sequence data revealed that four of these cDNAs were derived from novel genes. These cDNAs were designated CAG6, CAG12, CAG24, and CAG40 and were found to correspond to transcripts of 5.0, 7.5, 4.4, and 15 kb, respectively. The genes encoding CAG6, CAG12, CAG24, and CAG40 were assigned to chromosomes 12, 16, X, and 12, respectively. For the 5th gene, CAG26/pRHpA, a localization on two different chromosomes was established: 16 and X. None of the repeats showed any length polymorphisms in human DNA.

Positional mapping of loci in the DiGeorge critical region at chromosome 22q11 using a new marker (D22S183)

The majority of patients with DiGeorge syndrome (DGS) and velo-cardio-facial syndrome (VCFS) and a minority of patients with non-syndromic conotruncal heart defects are hemizygous for a region of chromosome 22q11. The chromosomal region that is commonly deleted is larger than 2 Mb. It has not been possible to narrow the smallest region of overlap (SRO) of the deletions to less than ca 500 kb, which suggests that DGS/VCFS might be a contiguous gene syndrome. The saturation cloning of the SRO is being carried out, and one gene (TUPLE1) has been identified. By using a cosmid probe (M51) and fluorescence in situ hybridization, we show here that the anonymous DNA marker locus D22S183 is within the SRO, between TUPLE1 and D22S75 (probe N25). A second locus with weak homology to D22S183, recognized by cosmid M56, lies immediately outside the common SRO of the DGS and VCFS deletions, but inside the SRO of the DGS deletions. D22S183 sequences are strongly conserved in primates and weaker hybridizing signals are found in DNA of other mammalian species; no transcripts are however detected in polyA+ RNA from various adult human organs. Probe M51 allows fast reliable screening for 22q11 deletions using fluorescence in situ hybridization. A deletion was found in 11 out of 12 DGS patients and in 3 out of 7 VCFS patients. Two patients inherited the deletion from a parent with mild (atypical) symptoms.

Cloning and characterization of MN1, a gene from chromosome 22q11, which is disrupted by a balanced translocation in a meningioma

We have isolated a gene, called MN1, which resides on chromosome 22 and which was found to be disrupted by a balanced translocation (4;22) in meningioma 32. The MN1 gene spans about 70 kb and consists of at least two large exons of approximately 4.7 kb and 2.8 kb. The MN1 cDNA codes for a protein of 1319 amino acids when the first methionine in the open reading frame is used. The MN1 cDNA contains two CAG repeats, one of which codes for a string of 28 glutamines. The t(4;22) disrupts the 5'-exon within the open reading frame. In meningioma 32 no expression of the MN1 mRNA is observed. These results suggest that inactivation of the MN1 gene in this tumour may contribute to its pathogenesis.

Cytogenetic, molecular genetic and pathological analyses in 126 meningiomas

In a series of 126 meningiomas, tumor and patient characteristics were investigated and statistically analyzed. A combined cytogenetic and molecular genetic approach was used to study chromosomal abnormalities and loss of markers on chromosome 22q. This approach was successfully applied to 93 meningiomas. In 66 cases, complete or partial loss of chromosome 22 was observed and in at least 12 of them this chromosome was involved in structural aberrations. In addition to chromosome 22 changes, chromosomes 1, 6, 11, 13, 14, 18, 19, X, and Y were also frequently involved in structural and numerical aberrations. Statistical analysis revealed a significant association between the number of chromosomal abnormalities and tumor grade. Complex karyotypes predominated in the group of grade II/III meningiomas. Furthermore, other variables showed statistically (or marginally statistically) significant differences. Meningiomas from the convexity were more often grade II/III, displayed predominantly (partial) loss of chromosome 22 and had complex karyotypes more often. These features were frequently found in meningiomas from males. Base meningiomas, on the other hand, occurred more often in females; they were usually grade I, showed loss of (parts of) chromosome 22 less often and displayed fewer additional chromosomal abnormalities.

The product of the NF2 tumour suppressor gene localizes near the plasma membrane and is highly expressed in muscle cells

Neurofibromatosis type 2 (NF2) is a disease resulting in the formation of schwannomas of the eighth cranial nerve, and other central nervous system tumours. A tumour suppressor gene has been found to be responsible for this disorder. The 595 amino acid NF2 protein shows a great deal of homology to a superfamily of membrane organizing proteins. To generate antibodies against the NF2 protein four synthetic peptides (SP) were injected in rabbits. COS cells transfected with an NF2 cDNA construct in an expression vector were used for immunocytochemical staining experiments; lysates of transfected COS cells were used for Western blotting experiments, as were lysates of E. coli cultures transformed with an NF2 cDNA construct subcloned in a prokaryotic expression vector. In western blots all sera detected a band indicating the appropriate molecular weight in lysates of transfected COS cells and E. coli. Immunocytochemical staining experiments indicate that the NF2 protein localizes in or near the cell membrane. Immunohistochemical staining of human tissue sections demonstrated the presence of the NF2 protein in muscle-, and Schwann cells. These results support the hypothesis that the NF2 protein functions as a membrane organizing element.

Characterization of the human kallikrein locus

The human kallikrein gene family is composed of three members: tissue kallikrein (KLK1), prostate-specific antigen (PA or APS), and human glandular kallikrein-1 (hGK-1 or KLK2). The three genes have previously been isolated and mapped to chromosome 19q13.2-q13.4. Further analysis of an area of 110 kb surrounding the kallikrein genes by CHEF electrophoresis and chromosome walking showed clustering of the three genes. The KLK1 gene is positioned in the opposite orientation of the APS and KLK2 genes in the order KLK1-APS-KLK2. The APS and KLK2 gene are separated by 12 kb; the distance between KLK1 and APS is 31 kb. A CpG island was detected in the region between KLK1 and APS. Preliminary data indicate that this CpG island is located directly adjacent to a gene that is unrelated to the kallikreins and seems to be ubiquitously expressed.

The promoter of the prostate-specific antigen gene contains a functional androgen responsive element

Expression of prostate-specific antigen (PA) mRNA was tested at various time periods after incubation of the human prostate tumor cell line LNCaP with the synthetic androgen R1881. Androgen-stimulated expression was observed within 6 h after addition of R1881 to the cells. Run-on experiments with nuclei isolated from LNCaP cells showed that expression of the PA gene could be regulated by R1881 on the level of transcription. DNase I footprints of the promoter region of the PA gene (-320 to +12) with nuclear protein extracts from LNCaP cells showed at least four protected regions. The protected areas include the TATA-box, a GC-box sequence, and a sequence AGAACAgcaAGTGCT at position -170 to -156, which closely resembles the reverse complement of the consensus sequence GGTACAnnnTGTTCT for binding of the glucocorticoid receptor and the progesterone receptor. Fragments of the PA promoter region were cloned in front of the chloramphenicol acetyltransferase (CAT) reporter gene and cotransfected with an androgen receptor expression plasmid into COS cells in a transient expression assay. CAT activity of COS cells grown in the presence of 1 nM R1881 was compared to untreated controls. A 110-fold induction of CAT activity was found if a -1600 to +12 PA promoter fragment was used in the construct. By further deletion mapping of the PA promoter a minimal region (-320 to -155) was identified as being essential for androgen-regulated gene expression. Mutation of the sequence AGAACAgcaAGTGCT (at -170 to -156) to AAAAAAgcaAGTGCT almost completely abolished androgen inducibility of the reporter gene constructs. One or more copies of the sequence AGAACAgcaAGTGCT cloned in front of a thymidine kinase promoter-CAT reporter gene confers androgen regulation to the reporter gene. These findings provide strong evidence for transcription regulation of the PA gene by androgens via the sequence AGAACAgcaAGTGCT. Interestingly, in addition to the AGAACAgcaAGTGCT element, an upstream region (-539 to -320) is needed for optimal androgen inducibility of the PA promoter.

Identification and androgen-regulated expression of two major human glandular kallikrein-1 (hGK-1) mRNA species

The screening of an oligo(dT)-primed prostate cDNA library with a human glandular kallikrein-1 (hGK-1) genomic DNA fragment resulted in the isolation of two different hGK-1 cDNAs. A 1.2 kb cDNA (pGK-1) contains an open reading frame of 510 bp, encoding the major part of the previously predicted hGK-1 protein (Schedlich et al. (1987) DNA 6, 429-437). This cDNA contains a 3'-untranslated region of 677 nucleotides and terminates in a poly(A) stretch, preceded by the canonical AATAAA polyadenylation signal. A second cDNA (pGK-10A), with a size of 1.5 kb, contains an open reading frame of 669 nucleotides preceded by 16 nucleotides of the 5'-untranslated region. pGK-10A differs from pGK-1 by the presence of an additional 37 bp fragment, interrupting the protein coding region of hGK-1, which results from the use of an alternative splice donor site of intron IV of the hGK-1 gene. The mature protein (excluding presumed pre- and propeptides) as deduced from the pGK-10A cDNA sequence, has a size of 199 amino acids and differs at the COOH-terminus from the 237 amino acid hGK-1 protein. The alternatively spliced mRNA comprises approximately 20% of the hGK-1 transcripts, as deduced from analysis of mRNA from prostate cells by PCR amplification of specific fragments. The regulation of hGK-1 mRNA expression was studied in different human prostate tumors and cell lines by Northern blotting, using a hGK-1-specific oligonucleotide probe. A high level of hGK-1 expression was found in the androgen-dependent tumors PC 82 and PC EW. hGK-1 mRNA was also present in the androgen-sensitive LNCaP cell line, but undetectable in the androgen-insensitive prostate tumors PC 133, PC 135 and the PC 3 cell line. In LNCaP cells, the expression of hGK-1 mRNA was strongly induced by androgens. Regulation of expression of the closely related prostate-specific antigen (PA) gene showed a similar pattern.

An MspI RFLP detected by the human glandular kallikrein gene (hGK) on chromosome 19q

Images

The prostate-specific antigen gene and the human glandular kallikrein-1 gene are tandemly located on chromosome 19

Using a prostate-specific antigen cDNA as a hybridization probe, clones containing the kallikrein genes encoding prostate-specific antigen, human glandular kallikrein-1 and pancreas/kidney kallikrein were isolated from a human genomic library. Clones containing the prostate-specific antigen gene and the human glandular kallikrein-1 gene overlap and span a region of about 36 kb. The two genes are aligned in a head to tail orientation at a mutual distance of 12 kb. Southern blot analysis of DNA from a panel of human-hamster hybrid cells with specific probes revealed the genes to be situated on chromosome 19. Assuming that the pancreas/kidney kallikrein gene is located in the same cluster, the distance to the prostate-specific antigen gene and the human glandular kallikrein gene must be at least 15 kb.

Characterization of the prostate-specific antigen gene: a novel human kallikrein-like gene

Using Prostate-specific Antigen cDNA fragments as hybridization probes a clone containing the information for the gene encoding Prostate-specific Antigen was isolated form a human genomic DNA library. The complete gene (about 6 kb) was sequenced and shown to be composed of four introns and five exons. Two major transcription initiation sites were found. The sequence of the promoter region revealed the presence of various well known transcription regulatory elements including a TATA box. A high percentage of homology was found between the Prostate-specific Antigen gene and the hGK-1 gene (82%). This homology extended into the promoter region. Two previously described variant Prostate-specific Antigen cDNAs can now be explained by intron retention and alternative splicing of the primary transcript.

Molecular cloning and characterization of novel prostate antigen cDNA's

Three different prostate antigen cDNA's were isolated from a PC 82 prostate tumor cDNA library. PA 75 has a size of 1.4 kb and contains the almost complete information for the 35 kD prostate antigen preproprotein. The 1.6 kb PA 525 cDNA lacks about 0.2 kb of the 3'-non coding region and contains an additional internal 0.4 kb fragment as a result of alternative splicing. PA 424 represents a 0.6 kb variant of PA 75. It contains a 0.15 kb internal fragment and a poly(A) tail preceded by an AAGAAA motive at the 3'-end. The predicted protein products of PA 525 and PA 424 will be different from PA 75 at the C-terminal end. In RNA preparations of two human prostate tumors (PC 82 and PC EW) seven different prostate antigen transcripts can be detected ranging in size from 0.5 kb to 5.6 kb. PA 75 cDNA represents the major 1.5 kb mRNA. PA 424 correlates with a 0.9 kb transcript and PA 525 with a 1.9 kb mRNA species.