Acquisition of biologically relevant gene expression data by Affymetrix microarray analysis of archival formalin-fixed paraffin-embedded tumours.

Robust protocols for microarray gene expression profiling of archival formalin-fixed paraffin-embedded tissue (FFPET) are needed to facilitate research when availability of fresh-frozen tissue is limited. Recent reports attest to the feasibility of this approach, but the clinical value of these data is poorly understood. We employed state-of-the-art RNA extraction and Affymetrix microarray technology to examine 34 archival FFPET primary extremity soft tissue sarcomas. Nineteen arrays met stringent QC criteria and were used to model prognostic signatures for metastatic recurrence. Arrays from two paired frozen and FFPET samples were compared: although FFPET sensitivity was low ( approximately 50%), high specificity (95%) and positive predictive value (92%) suggest that transcript detection is reliable. Good agreement between arrays and real time (RT)-PCR was confirmed, especially for abundant transcripts, and RT-PCR validated the regulation pattern for 19 of 24 candidate genes (overall R(2)=0.4662). RT-PCR and immunohistochemistry on independent cases validated prognostic significance for several genes including RECQL4, FRRS1, CFH and MET - whose combined expression carried greater prognostic value than tumour grade - and cmet and TRKB proteins. These molecules warrant further evaluation in larger series. Reliable clinically relevant data can be obtained from archival FFPET, but protocol amendments are needed to improve the sensitivity and broad application of this approach.

Proteome analysis of human liver tumor tissue by two-dimensional gel electrophoresis and matrix assisted laser desorption/ionization-mass spectrometry for identification of disease-related proteins.

Hepatocellular carcinoma (HCC) is a common malignancy worldwide and is a leading cause of death. To contribute to the development and improvement of molecular markers for diagnostics and prognostics and of therapeutic targets for the disease, we have largely expanded the currently available human liver tissue maps and studied the differential expression of proteins in normal and cancer tissues. Reference two-dimensional electrophoresis (2-DE) maps of human liver tumor tissue include labeled 2-DE images for total homogenate and soluble fraction separated on pH 3-10 gels, and also images for soluble fraction separated on pH 4-7 and pH 6-9 gels for a more detailed map. Proteins were separated in the first dimension by isoelectric focusing on immobilized pH gradient (IPG) strips, and by 7.5-17.5% gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels in the second dimension. Protein identification was done by peptide mass fingerprinting with delayed extraction-matrix assisted laser desorption/ionization-time of flight-mass spectrometry (DE-MALDI-TOF-MS). In total, 212 protein spots (117 spots in pH 4-7 map and 95 spots in pH 6-9) corresponding to 127 different polypeptide chains were identified. In the next step, we analyzed the differential protein expression of liver tumor samples, to find out candidates for liver cancer-associated proteins. Matched pairs of tissues from 11 liver cancer patients were analyzed for their 2-DE profiles. Protein expression was comparatively analyzed by use of image analysis software. Proteins whose expression levels were different by more than three-fold in at least 30% (four) of the patients were further analyzed. Numbers of protein spots overexpressed or underexpressed in tumor tissues as compared with nontumorous regions were 9 and 28, respectively. Among these 37 spots, 1 overexpressed and 15 underexpressed spots, corresponding to 11 proteins, were identified. The physiological significance of the differential expressions is discussed.

Generation of a contig comprising YACs and BACs within chromosome region 1p13.1.

Chromosome region 1p13 is known to show loss of heterozygosity (LOH) in a number of human tumor types, including breast. We have generated a contig comprising YACs and BACs spanning part of 1p13.1 which includes the smallest region of overlapping loss identified in our earlier studies. The contig is anchored to the genetic map by a number of microsatellite markers, and by the use of CEPH YACs. We have excluded a number of candidate genes from this region, and we have oriented the contig with respect to the centromere and a number of other genes and markers on 1p13. This resource will be valuable in mapping the target for LOH in breast and other tumors, and may also be useful for the genetic analysis of other genes or diseases known to map to this region.

Identification and cloning in yeast artificial chromosomes of a region of elevated loss of heterozygosity on chromosome 1p31.1 in human breast cancer.

We have mapped a region of high loss of heterozygosity in breast cancer to a 2-cM interval between the loci D1S430 and D1S465 on chromosome 1p31.1. This region shows allelic imbalance in around 60% of breast tumors. As part of a strategy to clone the target gene(s) within this interval, we have generated a yeast artificial chromosome contig spanning over 7 Mb. YACs from the CEPH and Zeneca (formerly ICI) libraries have been obtained by screening with PCR-based STSs from the region for both previously identified loci and newly isolated STSs. The YACs have been assembled into a contig by a combination of approaches, including analysis of their STS content, generation of new STSs from the ends of key YACs, and long-range restriction mapping. These YAC clones provide the basis for complete characterization of the region of high loss in breast cancer and for the ultimate identification of the target gene(s).

Allelic imbalance of chromosome 6q in ovarian tumours.

Previous work has implicated putative tumour-suppressor (ts) genes at 6q27 and a broad region at 6p12-q23. Here we report the results of a coded, randomised study of allelic imbalance at 12 loci on 6q on 40 pairs of coded tumour-blood pairs from patients with ovarian tumours. Our results provide clear evidence for the involvement of different regions of 6q in tumours of different histological subtypes. The involvement in serous tumours of a ts gene at the distal site is confirmed. However, proximal 6q presents a complex picture, with possibly three further ts genes: one at 6q21-23.3 involved at high frequency in benign and endometrioid tumours, another at 6q14-q15, also involved in endometrioid tumours, and a third suggested by a smallest region of deletion at 6q16.3-q21, between D6S275 and D6S300, that appears to be involved in early stage tumours. These observations point the way to a statistical study of the involvement of 6q in tumours of different histological type and staging performed on larger cohorts of samples.

Establishment of the marker order pter-NRAS-NGFB-D1S189-D1S252-D1S440-D1S453-D1S514-CEN-D1S442-D1S498-qte r in relation to the centromere on human chromosome 1.

Recent developments in genetic linkage mapping of the human genome have generated a large number of short tandem repeat polymorphic markers (Weissenbach et al. 1992, Gyapay et al. 1994), and eventual integration of these markers into a physical map is a logical progression. A number of Généthon microsatellite (CA repeat) markers have been provisionally localized to 1p13, but their exact position with respect to other sequences is unknown. In order to confirm the order of these markers and their position with respect to known genes within 1p13 and the centromere, we have isolated yeast artificial chromosomes (YACs) corresponding to the markers and have carried out double and triple fluorescence in situ hybridization (FISH) studies. Knowledge of both the order of microsatellite markers and their integration with a physical map of known genes can be an essential component in analysis of disease loci such as human cancer, where regions of chromosomes showing high levels of loss of heterozygosity need to be mapped in detail.

Proximal 6q, a region showing allele loss in primary breast cancer.

To define regions of deletion of chromosome 6q in breast cancer, we scored 18 (CA)n microsatellites for allelic imbalance (AI) in 42 paired blood/tumour samples. Heterozygosity frequencies of the markers in the sample population ranged from 31% to 92% (mean 68%). Two regions of the chromosome arm showed AI values greater than the background range of 10-22% (mean 17%) of informative cases that was observed with five markers spanning 6q21-q25.2. Firstly, seven markers gave AI values that averaged 35% in a region flanked by D6S313 (AI = 10%) at 6q13 and D6S283 (AI = 17%) at 6q16.3-21. The second region showed marginally increased AI at 6q25.2-q27 and included D6S193, previously shown to be close to a tumour-suppressor gene involved in ovarian carcinoma. Since AI of 6q in breast cancer was shown previously to be due predominantly to loss of heterozygosity, our results suggest the presence of at least two tumour-suppressor genes on 6q that are involved in breast cancer. The proximal region has not been recognised in breast cancer before and is involved in a higher frequency of tumours than the distal region.

Nucleotide ectoenzyme activities of human and Chinese hamster fibroblasts in tissue culture.

We have previously assigned human ecto-5'-nucleotidase (NT) to chromosome 6 on the basis of conversion of exogenously supplied [14C]AMP to adenosine by whole cells of human and Chinese hamster hybrids carrying chromosome 6. In this paper we demonstrate that the activity on human MRC-5 fibroblasts is typical of previously described and purified ecto-5'-nucleotidases. In contrast to MRC-5 cells, Chinese hamster V79A2 cells weakly express an AMPase activity that is not NT. The cytosolic form of NT in human and hybrid fibroblasts is similar to the ectoenzyme in substrate specificity. Hybrids that lack chromosome 6 express neither the ecto- nor the cytosolic enzyme, suggesting that both forms may be coded by the same gene on chromosome 6. Ecto-ATPase, ecto-ADPase, and ecto-ADP kinase activities are each expressed at similar levels in MRC-5 and V79A2. The ATPase, ADPase and NT activities of MRC-5 cells act sequentially to generate adenosine. A similar cascade acts on V79A2 cells but the lack of NT causes the accumulation of AMP.

Regional localization of human ecto-5' nucleotidase to chromosome 6q14-q21.

Human and mouse hybrids that contain fragments of human chromosome 6 as translocations were analysed for expression of ecto-5' nucleotidase enzymic activity measured by the conversion of AMP to adenosine and for antigenicity recognized by a monoclonal antibody specific for the human isozyme. Both methods allow a regional assignment of ecto-5'nucleotidase to 6q14-q21.

Tandem dye-ligand chromatography and biospecific elution applied to the purification of glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides.

1. A total of 65 immobilized triazine dyes were screened for their ability to purify the dual-nucleotide-specific glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides. From this screen a 'negative' (Matrex Gel Purple A) and a 'positive' (Matrex Gel Orange B) adsorbent were found to be the best in terms of overall purification and yield and were therefore combined to give the best purification. 2. Glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides was purified approx. 56-fold in a two-step tandem chromatographic system using Matrex Gel Purple A followed by Matrex Gel Orange B chromatography to a specific activity of 228 units/mg of protein in a final yield of 73%. 3. A study of the elution characteristics of glucose-6-phosphate dehydrogenase bound to Matrex Gel Orange B by KCl (pulse and gradient) and biospecific eluents (pulse) was carried out. NADP+, NADPH and adenosine 2',5'-bisphosphate were found to be the only effective biospecific eluents. A pulse of 50 microM-NADP+ (1/2 column vol.) was found to give a better purification than a 0-1 M-KCl gradient and therefore was the preferred method of elution. 4. Presaturation of the enzyme with various nucleotides was carried out to determine the effect on the subsequent binding of glucose-6-phosphate dehydrogenase to Matrex Gel Orange B. The results of these and biospecific-elution studies lead us to propose two possible schemes to explain the mechanism of the dye-protein interaction. 5. Reusability, capacity of the adsorbent and effect of varying the ligand concentration were also studied in the purification of glucose-6-phosphate dehydrogenase on Matrex Gel Orange B.