Gene expression signatures and biomarkers of noninvasive and invasive breast cancer cells: comprehensive profiles by representational difference analysis, microarrays and proteomics

We have characterized comprehensive transcript and proteomic profiles of cell lines corresponding to normal breast (MCF10A), noninvasive breast cancer (MCF7) and invasive breast cancer (MDA-MB-231). The transcript profiles were first analysed by a modified protocol for representational difference analysis (RDA) of cDNAs between MCF7 and MDA-MB-231 cells. The majority of genes identified by RDA showed nearly complete concordance with microarray results, and also led to the identification of some differentially expressed genes such as lysyl oxidase, copper transporter ATP7A, EphB6, RUNX2 and a variant of RUNX2. The altered transcripts identified by microarray analysis were involved in cell-cell or cell-matrix interaction, Rho signaling, calcium homeostasis and copper-binding/sensitive activities. A set of nine genes that included GPCR11, cadherin 11, annexin A1, vimentin, lactate dehydrogenase B (upregulated in MDA-MB-231) and GREB1, S100A8, amyloid beta precursor protein, claudin 3 and cadherin 1 (downregulated in MDA-MB-231) were sufficient to distinguish MDA-MB-231 from MCF7 cells. The downregulation of a set of transcripts for proteins involved in cell-cell interaction indicated these transcripts as potential markers for invasiveness that can be detected by methylation-specific PCR. The proteomic profiles indicated altered abundance of fewer proteins as compared to transcript profiles. Antisense knockdown of selected transcripts led to inhibition of cell proliferation that was accompanied by altered proteomic profiles. The proteomic profiles of antisense transfectants suggest the involvement of peptidyl-prolyl isomerase, Raf kinase inhibitor and 80 kDa protein kinase C substrate in mediating the inhibition of cell proliferation.

Racial differences in prostate cancer related to loss of heterozygosity on chromosome 8p12-23

PURPOSE

To determine if there is a racial difference in prostate cancer related to loss of heterozygosity (LOH) on chromosome 8p12-23, the region most frequently altered in prostate cancer.

METHODS AND MATERIALS

A total of 51 prostate cancer patients, consisting of 23 African Americans and 28 Caucasians, were included in this study. All patients underwent radical prostatectomy, and patients in the two racial subgroups were matched for median serum PSA, Gleason score, and pathological stage of cancer. Paired normal prostate and cancer tissue DNA was isolated and amplified with 13 polymorphic markers mapped to 8p12-23 by radiolabeled polymerase chain reaction. The amplified products were resolved by polyacrylamide gel electrophoresis, autoradiographed, and analyzed for allelic losses.

RESULTS

The overall incidence of LOH at 8p12-23 was 53%, and 16% showed homozygous deletions. The incidence of LOH in Caucasians was 68% compared to 35% in African Americans. On univariate (p = 0.02) and multivariate logistic regression analysis (p = 0.02), only Caucasian race was a significant predictor for LOH. The other clinicopathologic parameters did not have any significant effect on incidence of LOH.

CONCLUSION

These results highlight the independent influence of Caucasian race on incidence of LOH at 8p12-23, and suggest that genetic differences at specific tumor suppressor loci may be a factor responsible for racial variations observed in prostate cancer.

A receptor tyrosine kinase, UFO/Axl, and other genes isolated by a modified differential display PCR are overexpressed in metastatic prostatic carcinoma cell line DU145

We have used a modified differential display PCR protocol for isolating 3' restriction fragments of cDNAs specifically expressed or overexpressed in metastatic prostate carcinoma cell line DU145. Several cDNA fragments were identified that matched to milk fat globule protein, UFO/Axl, a receptor tyrosine kinase, human homologue of a Xenopus maternal transcript, laminin and laminin receptor, human carcinoma-associated antigen, and some expressed sequence tags. The transcript for milk fat globule protein, a marker protein shown to be overexpressed in breast tumors, was elevated in DU145 cells. The expression of UFO/Axl, a receptor tyrosine kinase, was considerably higher in DU145 cells as compared to normal prostate cells and prostatic carcinoma cell line PC-3. The overexpression of UFO oncogene in DU145 cells is discussed in the context of prostate cancer metastasis.

Toward expression mapping of albinism-deafness syndrome (ADFN) locus on chromosome Xq26

We have employed a direct cDNA selection methodology to isolate transcribed sequences encoded in the human chromosomal interval Xq26 that contains the gene for X-chromosome linked albinism deafness syndrome (ADFN). ADFN had been previously mapped to an 8 centi Morgan region on chromosome Xq26. We have constructed six cDNA libraries specific to six YACs mapping to a 1.5 mb span at the distal boundary of the ADFN locus. The YAC specific libraries were characterized for the presence of unique cDNAs. We have identified 15 transcribed sequences from the selected cDNA libraries. These cDNAs matched to three well characterized sequences corresponding to steroid 5-alpha reductase, ribosomal protein L28, and a short transcript that has been shown to be expressed in human brain cortex. Seven of the cDNAs matched to expressed sequence tags or other sequences of unknown function, and five cDNAs shared no homology with sequences in the public data bases. Each one of these sequences was represented as 3-10 clones in the set that was subjected to sequencing. Further characterization of these transcribed sequences may indicate potential candidates responsible for ADFN. We have discussed the utility of cDNA selection methodology in assembling transcript maps and identifying potential candidates for genetic deafness.

Molecular cloning of a zinc finger gene eZNF from a human inner ear cDNA library, and in situ expression pattern of its mouse homologue in mouse inner ear

We have isolated and characterized the cDNA for eZNF, a zinc finger gene expressed in human inner ear, from a kinetically enriched human inner ear cDNA library. The sequence of full length cDNA was determined and its expression pattern characterized. A high degree of homology is shared between eZNF and rat transcription factor Kid-1. It belongs to the C2H2 class of zinc finger genes, contains a Kruppel-associated box (KRAB) domain near the N-terminus, and has consensus sites for phosphorylation. The gene is expressed in kidney and inner ear structures of mouse and human as determined by Northern blot analysis. In situ hybridization was used to demonstrate specific expression of the mouse eZNF homologue in epithelial layers of the saccule, semicircular canals, and the cochlea of newborn mice. The genomic clone corresponding to the cDNA was isolated and used for fluorescence in situ hybridization to localize it to human chromosome 5qter. The identification of genes expressed in human inner ear by representational difference analysis, their chromosomal location, and expression pattern of their homologues in developing mouse inner ear comprise a strategy that can potentially identify genes important in hearing and deafness.

Expression of protein kinase regulator genes in human ear and cloning of a gamma subtype of the 14-3-3 family of proteins

We have used oligonucleotides corresponding to conserved regions of protein kinase regulators of 14-3-3 gene family as primers to amplify these genes from cDNAs constructed from the human fetal inner ear. Sequence characterization of clones revealed that the 14-3-3 cDNA library from the fetal inner ear has high abundance of clones encoding a protein kinase regulator (theta subtype), a member of 14-3-3 gene family, and relatively lower abundance of clones for two other members of the gene family. One of these genes is identical to the eta subtype of human 14-3-3 genes; there is no cloned gene for the other subtype in the human 14-3-3 gene family in the nucleic acid data bases. A sequence homology search revealed that the latter shared significant homology with the gamma subtype of the rat 14-3-3 family. On the basis of the sequence data, it appears that this clone represents a human homolog of the rat gamma subtype. The results demonstrate the expression of 14-3-3 genes in the inner ear, characterize a human homolog of the rat gamma subtype of 14-3-3, implicate these proteins in ear development, and indicate the utility of gene family polymerase chain reaction (PCR) for investigating gene expression in specific tissues.

Isolation of human ear specific cDNAs and construction of cDNA libraries from surgically removed small amounts of inner ear tissues

We have used representational difference analysis (RDA) for subtractive hybridization of oligo dT primed directionally cloned cDNA libraries from human inner ear tissue and a B-lymphoblast cell line. Two rounds of subtraction-amplification, followed by differential hybridization of selected clones led to the isolation of genes which were specific to the ear. Sequence analysis of randomly chosen clones revealed the presence of a histidine rich Ca2+ binding protein, human dynamin, collagen type 1A1, collagen type 2A1, SPARC, human growth hormone, and several specific genes which had no sequence homology in the data base. Furthermore, to apply these techniques for isolating genes specific to distinct inner ear structures and/or cell types of inner ear for which the starting tissue material is limiting, we have used a modified PCR based protocol to construct representative cDNA libraries. We have characterized a cDNA library constructed from small amounts of inner ear tissues recovered by ablative surgical procedure involving labyrinthectomy. The potential application of these protocols for isolating genes involved in hearing and deafness is discussed.

Transcribed sequences encoded in the region involved in contiguous deletion syndrome that comprises X-linked stapes fixation and deafness

We have used a direct cDNA selection protocol to isolate expressed sequences from yeast artificial chromosome clones that contain approximately 900 Kb of genomic DNA from Xq21 band that is deleted in contiguous gene syndromes comprising of mixed deafness associated with stapes fixation (DFN3). In addition to identifying Brn4 (POU3f4), a POU domain containing transcription factor that is involved in DFN3 phenotype, we have isolated seven short fragment cDNAs mapping to the deleted region. Some of the selected fragments showed X-chromosome specificity and hybridized to autosomal DNA fragments, indicating the presence of a low abundance interspersed repeat in the cDNAs or their homology to some uncharacterized family of genes. In conformity with the inertness of Xq21 band our results demonstrate that the region encodes far less than the average density of genes in other parts of the genome.

Isolation of expressed sequences that include a gene for familial breast cancer (BRCA2) and other novel transcripts from a five megabase region on chromosome 13q12

A proportion of familial breast cancer has recently been shown by genetic linkage analysis to map to chromosome l3q12 (Wooster et al, 1994). This locus contains a tumor suppressor gene BRCA2, mutations in which lead to tumorigenesis. Genetic alterations at this locus have also been shown in pancreatic adenocarcinoma and in hepatocellular carcinoma. In an effort to isolate the BRCA2 gene, we have cloned 73 non overlapping cDNAs from a set of nine YACs spanning 6 cM interval on chromosome 13q12 by using a direct cDNA selection method. One of the selected cDNAs corresponds to a region of the 3' portion of BRCA2 mRNA, the sequence of which was published recently (Wooster et al, 1995). Northern analysis of BRCA2 transcripts from a variety of cell lines showed altered sizes of the mRNA in a breast cancer cell line (MCF7) and a prostate carcinoma cell line (DU145). Furthermore, BRCA2 transcript was present in cDNA libraries from total fetus as well as adult human tissues. Fifteen unique cDNA fragments encode genes/ESTs that are already known, of which only two have been mapped to this region. The other 12 cDNAs include genes for RPL6/mRNA for TAX REB 107, elongation factor-1 delta, 26S protease S4 regulatory subunit, small cytoplasmic 7SL RNA, a full length open reading frame (ORFU), brain thiol specific antioxidant protein, ribosomal protein, L35, and lipoxygenase activating protein. Six cDNAs represent human homologs of genes known in other species, namely, mouse HSPE71, Rat RhoGAP protein, S cerevisiae leucyl tRNA synthetase and S cerevisiae chromosome II ORF YBLO44W. The remaining 52 cDNAs showed either weak similarity or no similarity to sequences in the nucleotide data base and hence would represent novel genes. The plausible functions of some of these genes based on their sequence similarity to other known genes is discussed.

Uniform amplification of a mixture of deoxyribonucleic acids with varying GC content

A PCR method for uniform amplification of a mixture of DNA templates differing in GC content is described using the two enzyme approach (Klentaq1 and Pfu DNA polymerase) and a combination of DMSO and betaine. This method was applied to amplify the CGG repeat region from the fragile X region.

A cyclophilin gene-like sequence maps to human X-chromosome

We isolated cDNA fragments encoded in an X-chromosome specific YAC from the locus DXS995 by using direct cDNA selection method. Several of the selected cDNA fragments were identical to various exons of cyclophilin A gene. Hybridization of the selected cyclophilin cDNA fragments to the target YAC, presence of these sequences in an X-chromosome specific phage library and absence of a cross hybridizing fragment in a cell line (XL-45) that contains deletions of the interval Xq21, demonstrates that a cyclophilin like sequence is present in the human X-chromosome.

Molecular cloning and expression pattern of genes from a 470 Kb region near BRCA1 locus on chromosome 17q21

We have used direct cDNA selection to isolate expressed sequences from a set of four overlapping YACs, spanning approximately 470 kb of the chromosomal interval 17q21 centromeric to BRCA1 gene. Thirty-eight nonoverlapping unique cDNA fragments were identified in this region. Twenty-two of the selected cDNAs showed complete identity with known genes and expressed sequence tags. Two of these cDNAs shared sequence homology with genes known to encode potential DNA binding motifs and hence could function in transcriptional regulation. The remaining 16 unique cDNA fragments showed no significant similarity to sequences in the databases and could potentially encode novel genes. Northern analysis of the novel cDNAs showed differential expression in various tissues, supporting the transcribed nature of these sequences. Our results place the gene for a G-protein coupled receptor (GPR2) previously mapped to 17q within a 400 kb region on 17q21.

Construction of libraries enriched for sequence repeats and jumping clones, and hybridization selection for region-specific markers

We describe a simple and rapid method for constructing small-insert genomic libraries highly enriched for dimeric, trimeric, and tetrameric nucleotide repeat motifs. The approach involves use of DNA inserts recovered by PCR amplification of a small-insert sonicated genomic phage library or by a single-primer PCR amplification of Mbo I-digested and adaptor-ligated genomic DNA. The genomic DNA inserts are heat denatured and hybridized to a biotinylated oligonucleotide. The biotinylated hybrids are retained on a Vectrex-avidin matrix and eluted specifically. The eluate is PCR amplified and cloned. More than 90% of the clones in a library enriched for (CA)n microsatellites with this approach contained clones with inserts containing CA repeats. We have also used this protocol for enrichment of (CAG)n and (AGAT)n sequence repeats and for Not I jumping clones. We have used the enriched libraries with an adaptation of the cDNA selection method to enrich for repeat motifs encoded in yeast artificial chromosomes.

Construction and characterization of a NotI-BsuE linking library from the human X chromosome

We describe the construction and characterization of methylation-resistant sequence-tagged NotI linking clones specific for the X chromosome, referred to as NotI-BsuE linking clones. The approach consists of methylating the X-chromosome-specific cloned DNA with BsuE methylase (M. BsuE), an enzyme that methylates the first C residue in the CGCG sequence, followed by selection of the methylation-resistant NotI sites by insertion of a kanamycin-resistance gene in the clones cleavable by NotI. The frequent occurrence of NotI sites in CpG islands is expected to cause methylation of a large number of NotI sites with BsuE methylase, thereby rendering them resistant to NotI cleavage. Thus, the combination of M. BsuE and NotI yields less frequent cutting than the NotI alone. We have isolated, partially sequenced, and characterized 113 NotI-BsuE linking clones, and mapped 50 clones to various regions along the chromosome.

A polymerase chain reaction approach for constructing jumping and linking libraries

Images

Selective enrichment of a large size genomic DNA fragment by affinity capture: an approach for genome mapping

A method to enrich large size DNA fragments obtained by digestion with rare cutting restriction endonucleases was developed and applied for the isolation of a 150 kb SfiI fragment containing the beta-globin gene cluster. The digested DNA is rendered single stranded at the ends by diffusing a strand specific exonuclease into an agarose plug containing DNA. The plug is melted and solution hybridization is then performed with a bridge RNA containing specific sequences from the end of a desired fragment linked to a common probe sequence. The common probe sequence is annealed to a biotinylated RNA and the resulting tripartite hybrid is retained onto a solid matrix containing avidin and specifically released by ribonuclease action. Enrichments of greater than 350 fold have been achieved consistently. Such directed purification of large DNA fragments without cloning can considerably expedite mapping and gene localization in a complex genome and facilitate the construction of sublibraries from defined regions of the genome.

Characteristics of the formation of enzyme-bound ATP from medium inorganic phosphate by mitochondrial F1 adenosinetriphosphatase in the presence of dimethyl sulfoxide

Addition of dimethyl sulfoxide promotes the formation of enzyme-bound ATP from medium Pi by mitochondrial F1 adenosinetriphosphatase that has tightly bound ADP present. Measurements are reported of medium Pi in equilibrium H18OH exchange and of the dependence of formation of enzyme-bound ATP on Pi concentration. Attainment of an apparent equilibrium between medium Pi and bound ATP requires longer than 30 min, even though the rates of Pi binding and release after apparent equilibrium is reached would suffice for a faster approach to equilibrium. Slow protein conformational changes or other unknown modulating factors may be responsible for the slow rate of bound ATP formation. After apparent equilibrium is reached, each Pi that binds to the enzyme reversibly forms ATP about 50 times before being released to the medium. The rate of interconversion of bound ATP to bound ADP and Pi is much slower than that in the absence of dimethyl sulfoxide as measured with sufficiently low ATP concentrations so that single-site catalysis is favored. Although the interconversion rate is slowed, the equilibrium constant for bound ATP formation from bound ADP and Pi is not far from unity. Dimethyl sulfoxide favors the formation of enzyme-bound ATP by promoting the competent binding of Pi to enzyme with ADP bound at a catalytic site rather than by promoting formation of bound ATP from bound ADP and Pi.

Escherichia coli F1 ATPase is reversibly inhibited by intra- and intersubunit crosslinking: an approach to assess rotational catalysis

Reaction of the multisubunit F1 ATPase from Escherichia coli (EF1) with a bifunctional cleavable crosslinker, 3,3'-dithiobis(succinimidylpropionate) (DSP), has been used to explore the possibility that during catalysis a rotational movement of catalytic subunits relative to noncatalytic subunits occurs. The premise is that such rotational catalysis is tenable if intersubunit crosslinking of a major subunit with one of the minor subunits inhibits the enzyme activity and if upon cleavage of the crosslinks, the enzyme regains activity. The results presented in this paper show that crosslinking of about 5-6 reactive groups on EF1 with DSP is accompanied by a loss of 2/3 of the enzyme activity. Both intra- and intersubunit crosslinks are formed. The most prominent intersubunit crosslinks are those of gamma and delta subunits with the alpha subunit. Nearly complete recovery of activity can be attained by cleaving the disulfide bond in the crosslinker with dithiothreitol. Because the chemical modification of enzyme groups remains after the crosslinker is cleaved, the loss in activity before cleavage can be ascribed to conformational restraints. The results show that catalysis by the EF1 ATPase is highly sensitive to the restrictions of crosslinking, and are consistent with the view that catalysis is accompanied by appreciable movements of the major subunits with respect to the minor subunits, as suggested for rotational catalysis.

Mitochondrial F1-ATPase will bind and cleave ATP but only slowly release ADP after N,N'-dicyclohexylcarbodiimide or 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole derivatization

The ATPase from the inner mitochondrial membrane is known to be inhibited by modification of one of the three catalytic subunits with N,N'-dicyclohexylcarbodiimide (DCCD) or 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole. An experimental approach described in this paper shows that most of the residual ATPase activity observed after the usual DCCD or 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole modification is due to the presence of unmodified enzyme, although the large fraction of modified enzyme retains a weak catalytic activity. This weak catalytic activity can be stimulated by methanol or dimethyl sulfoxide. When the modified enzymes are exposed to Mg2+ and [3H]ATP, about equal amounts of [3H]ATP and [3H]ADP appear at catalytic sites. The turnover rate for these enzymes is less than 1/1000 that of the native enzyme when it is calculated from the rate at which the enzyme becomes labeled at the catalytic sites with [3H]ATP and [3H]ADP during steady state hydrolysis. In addition, a higher ATP concentration is required for steady state turnover and, after ATP binding, the principal rate-limiting step is the capacity of the derivatized enzyme to undergo the binding changes necessary for the release of ADP and Pi. When the modified enzymes are not hydrolyzing ATP, they convert to form(s) that show a distinct lag in the replacement of bound nucleotides at catalytic sites. The replacement of bound nucleotides is still promoted by MgATP, even though the enzymes have been converted to sluggish forms. Contrary to a recent suggestion based on the study of the DCCD-modified enzyme (Soong, K.S., and Wang, J.H. (1984) Biochemistry 23, 136-141), our data provide evidence for the existence of catalytic cooperatively between at least two alternating sites in the modified enzyme and are consistent with continued sequential participation of all three sites.