Human cell lines engineered for tetracycline-regulated expression of tumor suppressor candidate genes from a frequently affected chromosomal region, 3p21

BACKGROUND

We modified a tetracycline-regulated system that can control the activity of individual genes quantitatively and reversibly in transgenic mammals. Despite these advances, there remained one problem in the intensive use of the tet-system: the limited range of acceptor cell lines, expressing a tetracycline-controlled transcriptional activator (tTA). This study describes in detail new vectors and a unifying strategy to generate tTA-expressing cell lines.

METHOD

Two retroviral vectors pLNCtTA-hCMV and pLNCtTA-EF1alpha coding for the tTA were used to engineer cell lines to constitutively express tTA. New expression vectors pETE-Hyg and pETE-Bsd were also created that replicate in episomal form in human cells and facilitate tetracycline-regulated expression of targeted genes.

RESULTS

The primate-tropic retroviruses efficiently delivered the regulatory tTA gene into 12 selected human cancer cell lines. Two candidate tumor suppressor genes from the human 3p21-p22 region MAPKAPK3 (3pK) and MLH1 were cloned into the episomal vector and transfected into engineered A9 and KRC/Y cells. The transfectants were subcutaneously grown in SCID mice, and the expression of the transgene was successfully controlled in vivo by tetracycline administered ad libitum in drinking water. The experiments demonstrated that both transgenes did not antagonize the tumorous growth of these cells.

CONCLUSIONS

New retroviral and episomal vectors appear particularly suited for tight regulation of genes that cause suppression of cell growth. The generated cell lines can be used in various applications to study the effect of an inducible transgene in human cancer cells.

hUNC93B1: a novel human gene representing a new gene family and encoding an unc-93-like protein

We have identified a novel human gene UNC93B1 encoding a protein related to unc-93 of Caenorhabditis elegans. The combined sequence derived from several cDNA clones is 2282 bp and comparison with genomic sequence shows that the gene contains 11 exons. The longest open reading frame encodes a deduced sequence of 597 amino acids. Homology analysis shows that the hUNC93B1 gene is highly conserved and related to sequences in Arabidopsis thaliana, C. elegans, Drosophila melanogaster, chicken and mouse. Structural analysis of the deduced amino acid sequence of hUNC93B1 points to possible existence of multiple membrane-spanning domains. hUNC93B1 protein also displays some similarities to the bacterial ABC-2 type transporter signature and to ion transporters of Deinococcus radiodurans and Helicobacter pylori. As revealed by Northern analysis, the level of expression varies significantly between tissues, with the highest level detected in the heart. The gene was mapped to chromosomal band 11q13 by fluorescence in situ hybridization. We suggest that this gene is a member of a novel hUNC93B-related gene family.

A new approach to genome mapping and sequencing: slalom libraries

We describe here an efficient strategy for simultaneous genome mapping and sequencing. The approach is based on physically oriented, overlapping restriction fragment libraries called slalom libraries. Slalom libraries combine features of general genomic, jumping and linking libraries. Slalom libraries can be adapted to different applications and two main types of slalom libraries are described in detail. This approach was used to map and sequence (with approximately 46% coverage) two human P1-derived artificial chromosome (PAC) clones, each of approximately 100 kb. This model experiment demonstrates the feasibility of the approach and shows that the efficiency (cost-effectiveness and speed) of existing mapping/sequencing methods could be improved at least 5-10-fold. Furthermore, since the efficiency of contig assembly in the slalom approach is virtually independent of length of sequence reads, even short sequences produced by rapid, high throughput sequencing techniques would suffice to complete a physical map and a sequence scan of a small genome.

Cloning of deleted sequences (CODE): A genomic subtraction method for enriching and cloning deleted sequences

The deletion of specific genomic sequences is believed to influence the pathogenesis of certain diseases such as cancer. Identification of these sequences could provide novel therapeutic avenues for the treatment of disease. Here, we describe a simple and robust method called cloning of deleted sequences (CODE), which allows the selective cloning of deleted sequences from complex human genomes. Briefly, genomic DNA from two sources (human normal and tumor samples) was digested with restriction enzymes (e.g., BamHI, BglII, and BclI), then ligated to special linkers, and amplified by PCR. Tester (normal) DNA was amplified using a biotinylated primer and dNTPs. Driver (tumor) DNA was amplified using a non-biotinylated primer, but with dUTP instead of d7TP After denaturation and hybridization, all the driver DNA was destroyed with uracil-DNA glycosylase (UDG), and all imperfect hybrids were digested with mung bean nuclease. Sequences deleted from the driver DNA but present in the tester DNA were purified with streptavidin magnetic beads, and the cycle was repeated three more times. This procedure resulted in the rapid isolation and efficient cloning of genomic sequences homozygously deleted from the driver DNA sample, but present in the tester DNA fraction.

Initial isolation and analysis of the human Kv1.7 (KCNA7) gene, a member of the voltage-gated potassium channel gene family

A novel human potassium channel gene was identified and isolated. The maximal open reading frame encodes a protein of 456 amino acids. The predicted product exhibits 91% amino acid identity to the murine voltage-gated potassium channel protein Kv1.7 (Kcna7), which plays an important role in the repolarization of cell membranes. Based on the high similarity, the human gene has been classified as the ortholog of the mouse Kcna7 and given the name Kv1.7 (KCNA7). A structural prediction identified a pore region characteristic of potassium channels and six membrane-spanning domains. Northern expression analysis revealed the gene is expressed preferentially in skeletal muscle, heart and kidney. However, it is expressed at lower level in other tissues, including liver. A single mRNA isoform was observed, with a size of approximately 4.5 kb. Using fluorescence in situ hybridization, the gene was mapped to chromosomal band 19q13.4 (269.13 cR(3000)). A genomic sequence was identified in the database from this region, and the KCNA7 gene structure determined. Computational analysis of the genomic sequence reveals the location of a putative promoter and a likely muscle-specific regulatory region. Initial comparison to the published murine Kcna7 cDNA suggested a different N-terminal sequence for the human protein, however, further analysis suggests that the original mouse sequence contained an error or an unusual polymorphism.

The candidate tumor suppressor gene, RASSF1A, from human chromosome 3p21.3 is involved in kidney tumorigenesis

Clear cell-type renal cell carcinomas (clear RCC) are characterized almost universally by loss of heterozygosity on chromosome 3p, which usually involves any combination of three regions: 3p25-p26 (harboring the VHL gene), 3p12-p14.2 (containing the FHIT gene), and 3p21-p22, implying inactivation of the resident tumor-suppressor genes (TSGs). For the 3p21-p22 region, the affected TSGs remain, at present, unknown. Recently, the RAS association family 1 gene (isoform RASSF1A), located at 3p21.3, has been identified as a candidate lung and breast TSG. In this report, we demonstrate aberrant silencing by hypermethylation of RASSF1A in both VHL-caused clear RCC tumors and clear RCC without VHL inactivation. We found hypermethylation of RASSF1A's GC-rich putative promoter region in most of analyzed samples, including 39 of 43 primary tumors (91%). The promoter was methylated partially or completely in all 18 RCC cell lines analyzed. Methylation of the GC-rich putative RASSF1A promoter region and loss of transcription of the corresponding mRNA were related causally. RASSF1A expression was reactivated after treatment with 5-aza-2'-deoxycytidine. Forced expression of RASSF1A transcripts in KRC/Y, a renal carcinoma cell line containing a normal and expressed VHL gene, suppressed growth on plastic dishes and anchorage-independent colony formation in soft agar. Mutant RASSF1A had reduced growth suppression activity significantly. These data suggest that RASSF1A is the candidate renal TSG gene for the 3p21.3 region.

CIS--cloning of identical sequences between two complex genomes

Development of the methods permitting cloning of identical sequences between two sources of DNA can be very useful for many purposes, including isolation of disease genes. Here we describe a new method called CIS (cloning of identical sequences). A combination of digestion with MvnI, treatment with mung bean nuclease, UDG (uracil-DNA glycosylase) and PCR with 5'-methyl-dCTP and dUTP was used to isolate identical sequences between two micro-cell hybrid lines (MCH). In a control experiment, mouse MCH903.1 and MCH939.2 containing human chromosome 3 from different individuals, were compared using the CIS procedure. Only background fluorescence in-situ hybridization (FISH) was achieved. In another experiment, mouse MCH903.1, containing complete human chromosome 3, and rat MCH429.11, containing a part of human 3q from the same chromosome were compared. The experiment showed that the original MCH429.11 and the DNA purified using the CIS procedure had identical FISH patterns to human metaphase chromosomes, thus demonstrating the efficiency of CIS.

NotI clones in the analysis of the human genome

Not I linking clones contain sequences flanking Not I recognition sites and were previously shown to be tightly associated with CpG islands and genes. To directly assess the value of Not I clones in genome research, high density grids with 50 000 Not I linking clones originating from six representative Not I linking libraries were constructed. Altogether, these libraries contained nearly 100 times the total number of Not I sites in the human genome. A total of 3437 sequences flanking Not I sites were generated. Analysis of 3265 unique sequences demonstrated that 51% of the clones displayed significant protein similarity to SWISSPROT and TREMBL database proteins based on MSPcrunch filtering with stringent parameters. Of the 3265 sequences, 1868 (57.2%) were new sequences, not present in the EMBL and EST databases (similarity < or =90%). Among these new sequences, 795 (24.3%) showed similarity to known proteins and 712 (21.8%) displayed an identity of >75% at the nucleotide level to sequences from EMBL or EST databases. The remaining 361 (11.1%) sequences were completely new, i.e. <75% identical. The work also showed tight, specific association of Not I sites with the first exon and suggest that the so-called 3' ESTs can actually be generated from 5'-ends of genes that contain Not I sites in their first exon.

Combined LOH/CGH analysis proves the existence of interstitial 3p deletions in renal cell carcinoma

We have recently developed an allele titration assay (ATA) to assess the sensitivity and influence of normal cell admixture in loss of heterozygosity (LOH) studies based on CA-repeat. The assay showed that these studies are biased by the size-dependent differential sensitivity of allele detection. Based on these data, we have set up new criteria for evaluation of LOH. By combining these new rules with comparative genome hybridization (CGH) we have shown the presence of interstitial deletions in renal cell carcinoma (RCC) biopsies and cell lines. At least three out of 11 analysed RCC cell lines and three out of 37 biopsies contain interstitial deletions on chromosome 3. Our study suggests the presence of several regions on human chromosome 3 that might contribute to tumor development by their loss: (i) 3p25-p26, around the VHL gene (D3S1317); (ii) 3p21. 3-p22 (between D3S1260 and D3S1611); (iii) 3p21.2 (around D3S1235 and D3S1289); (iv) 3p13-p14 (around D3S1312 and D3S1285). For the first time, AP20 region (3p21.3-p22) was carefully tested for LOH in RCC. It was found that the AP20 region is the most frequently affected area. Our data also suggest that another tumor suppressor gene is located near the VHL gene in 3p25-p26.

Cloning of polymorphisms (COP): enrichment of polymorphic sequences from complex genomes

Here we describe a new procedure (cloning of polymorphisms, COP) for enrichment of single nucleotide polymorphisms (SNPs) that represent restriction fragment length polymorphisms (RFLPs). COP would be applicable to the isolation of SNPs from particular regions of the genome, e.g. CpG islands, chromosomal bands, YACs or PAC contigs. A combination of digestion with restriction enzymes, treatment with uracil-DNA glycosylase and mung bean nuclease, PCR amplification and purification with streptavidin magnetic beads was used to isolate polymorphic sequences from the genomes of two human samples. After only two cycles of enrichment, 80% of the isolated clones were found to contain RFLPs. A simple method for the PCR detection of these polymorphisms was also developed.

Loss of heterozygosity in tumor cells requires re-evaluation: the data are biased by the size-dependent differential sensitivity of allele detection

Normal tissue contamination of tumors may eclipse the detection of loss of heterozygosity (LOH) by microsatellite analysis and may also hamper isolation of tumor suppressor genes. To test the potential impact of this problem, we prepared artificial mixtures of mouse-human microcell hybrid lines that carried different alleles of the same chromosome 3 marker. After performing an allele titration assay, we found a consistent difference between the LOH of a high molecular weight (H) allele and the LOH of a low molecular weight (L) allele of the same CA repeat marker. It follows that normal tissue admixtures will be less of a problem when LOH affects a H allele than with a L allele. Random screening of 100 papers published between 1994 and 1999 revealed that the loss of a L allele was recorded at about half the frequency (52%) of loss of a H allele. To avoid this bias, we have developed rules for the evaluation of LOH data. We suggest that the loss of a L allele should be given more weight than the loss of a H allele in LOH studies using microsatellite markers.

Analysis of NotI linking clones isolated from human chromosome 3 specific libraries

We have partially sequenced more than 1000 NotI linking clones isolated from human chromosome 3-specific libraries. Of these clones, 152 were unique chromosome 3-specific clones. The clones were precisely mapped using a combination of fluorescence in situ hybridization (FISH) and hybridization to somatic cell or radiation hybrids. Two- and three-color FISH was used to order the clones that mapped to the same chromosomal region, and in some cases, chromosome jumping was used to resolve ambiguous mapping. When this NotI restriction map was compared with the yeast artificial chromosome (YAC) based chromosome 3 map, significant differences in several chromosome 3 regions were observed. A search of the EMBL nucleotide database with these sequences revealed homologies (90-100%) to more than 100 different genes or expressed sequence tags (ESTs). Many of these homologies were used to map new genes to chromosome 3. These results suggest that sequencing NotI linking clones, and sequencing CpG islands in general, may complement the EST project and aid in the discovery of all human genes by sequencing random cDNAs. This method may also yield information that cannot be obtained by the EST project alone; namely, the identification of the 5' ends of genes, including potential promoter/enhancer regions and other regulatory sequences

Identification of new tumor suppressor genes based on in vivo functional inactivation of a candidate gene

As a step towards developing a new functional test for the identification of tumor suppressor genes, human wild type and mutant RB genes were expressed in the mouse A9 fibrosarcoma cell line under the transcriptional regulation of the tetracycline repressor using two new vectors: pLNCtTA and pETI. Following passage of the transfectants in immunodeficient SCID mice, the wild type RB gene was deleted or functionally inactivated already after the first passage in all 20 tumors tested. In contrast, a non-functional mutant RB gene was maintained in all 10 tumors studied. These results suggest that tests for the identification of tumor suppressor genes may be based on their functional inactivation in vivo, rather than on growth suppression.

A cosmid and cDNA fine physical map of a human chromosome 13q14 region frequently lost in B-cell chronic lymphocytic leukemia and identification of a new putative tumor suppressor gene, Leu5

B-cell chronic lymphocytic leukemia (B-CLL) is a human hematological neoplastic disease often associated with the loss of a chromosome 13 region between RB1 gene and locus D13S25. A new tumor suppressor gene (TSG) may be located in the region. A cosmid contig has been constructed between the loci D13S1168 (WI9598) and D13S25 (H2-42), which corresponds to the minimal region shared by B-CLL associated deletions. The contig includes more than 200 LANL and ICRF cosmid clones covering 620 kb. Three cDNAs likely corresponding to three different genes have been found in the minimally deleted region, sequenced and mapped against the contigged cosmids. cDNA clone 10k4 as well as a chimeric clone 13g3, codes for a zinc-finger domain of the RING type and shares homology to some known genes involved in tumorigenesis (RET finger protein, BRCA1) and embryogenesis (MID1). We have termed the gene corresponding to 10k4/13g3 clones LEU5. This is the first gene with homology to known TSGs which has been found in the region of B-CLL rearrangements.

NotI linking/jumping clones of human chromosome 3: mapping of the TFRC, RAB7 and HAUSP genes to regions rearranged in leukemia and deleted in solid tumors

By applying the 'recognition mask' strategy to 300 mammalian sequences containing NotI sites we demonstrated that 5' ends of genes are highly enriched in NotI sites. A NotI linking clone NL2-252 (D3S1678) containing transferrin receptor (TFRC) gene was used as an initial point for chromosomal jumping. One of the jumping clones, J21-045 traverses 210 kbp and links NL2-252 to NL26 (D3S1632), a NotI linking clone containing highly polymorphic sequences. The TFRC gene was mapped to 3q29, close to the telomeric marker D3S2344, by linkage analysis, a panel of hybrid cell lines, GeneBridge 4 panel and FISH. Clone NLM-007 (D3S4302) was found to contain ras-homologous gene RAB7. By FISH and a panel of hybrid cell lines this gene was mapped to 3q21. This region is of particular interest due to frequent rearrangements in different types of leukemia. Clone L2-081 (D3S4283) containing new member of ubiquitin-specific proteases (HAUSP gene) was localized in 3p21 inspiring further investigation of involvement of this gene in development of lung and renal carcinomas.

Human/mouse microcell hybrid based elimination test reduces the putative tumor suppressor region at 3p21.3 to 1.6 cM

We have previously identified an approximately 7 cM long common eliminated region (CER), involving the 3p21.3 markers AP20R, D3S966, D3S3559, D3S1029, WI-7947, D3S2354, AFMb362wb9, and D3S32, in human chromosome 3/A9 mouse fibrosarcoma microcell hybrid (MCH) derived SCID mouse tumors. We now report the results of our more detailed analysis on 24 SCID mouse tumors derived from two MCH lines that originally carried intact human chromosomes 3. They were analyzed by fluorescence in situ hybridization (FISH) painting and PCR, using 24 markers covering the region between D3S1611 and D3S13235 at 3p22-p21.2. D3S32 and D3S2354 were regularly eliminated during in vivo tumor growth, whereas the other 22 markers, D3S1611, ACAA, D3S1260, WI-692, AP20R, D3S3521, D3S966, D3S1029, D3S643, WI-2420, MSTI. GNAI2, D3S1235, D3S1298, GLBI, WI-4193, D3S3658, D3S3559, D3S3678, WI-6400, WI-7947, and WI-10865, were regularly retained. We have defined a common eliminated region of approximately 1.6 cM (designated as CER1) inside the 7 cM CER described earlier. CER1 is flanked distally by D3S1029 and proximally by D3S643.

A 3p21.3 region is preferentially eliminated from human chromosome 3/mouse microcell hybrids during tumor growth in SCID mice

We have previously shown that four markers spanning the 3p24-p21.3 region, THRB, AP20R, D3S1029, and D3S32, were regularly eliminated from three human chromosome 3 (chr3)/mouse microcell hybrids (MCHs) during tumor growth in SCID mice. In an attempt to narrow down the eliminated region, we have studied 22 new SCID mouse tumors derived from 5 MCH lines carrying human chr3. They were analyzed by fluorescence in situ hybridization (FISH), Southern blotting, and polymerase chain reaction (PCR). MCHs that carried human chr1, chr8, chr13, and chr17 were examined as controls. We could identify a common eliminated region (CER) at 3p21.3, bordered distally by D3S1260 and proximally by D3S643/D3F15S2. Eight of 53 chr3-specific PCR markers, AP20R, D3S966, D3S3559, D3S1029, WI-7947, D3S2354, AFMb362wb9, and D3S32. were localized within the CER. This finding is consistent with the notion that a tumor suppressor gene may be located in this area, as suggested by frequent loss of heterozygosity (LOH) within this region observed in several types of solid tumors.