Enlaces NCL

7.1 Introduction

This thesis consists o f the study o f two regions showing loss o f heterozygosity (LOH) o f genom ic D N A in hum an breast cancer on chrom osom e I p S l.l and lp l3 .1 . The positional analysis o f genom ic DNA from the long range genom ic m ap to the genes and their possible role as tum our suppressors is described. A t the beginning, this thesis involved the construction o f long range physical m aps spanning approxim ately 1 Oh4b and 1.3Mb at the lp31.1 and the lp l3 .1 loci, respectively. O verlapping clone maps w ere first built up in Y AC vectors for rapid coverage o f the regions and th en converted into cosm id pocket m aps derived from a flow sorted chrom osom e 1 library. T he second step w as the generation o f a contig o f overlapping PAC, BAC and cosm id clones at chrom osom al region lp l3 .1 , w hich then served as the fram ew ork for gene hunting strategies (exon trapping and cDNA selection). As a result, tw o genes w ere localised to this region and a prelim inary study o f their potential involvem ent w ith breast cancer w as investigated by sequencing o f tum our and normal tissue DNA material from patients show ing LO H at this locus.

7.2 Evaluation of the chromosome 1 specific cosmid pocket maps as a

molecular resource

A t the tim e this w ork began, the chrom osom e 1 sequencing project at the S ang er Centre, H inxton, U K , relied on the construction o f physical m aps in Y AC vectors, th at w ould then be used as a fram ew ork for assem bling contigs o f bacterial clones, w hich w ould constitute the sequencing tem plate. Due to the high rate o f chimaerism an d instability o f Y AC inserts and the difficulty o f preparing large quantities o f pure D N A from Y AC clones, it was useful to obtain further coverage in bacterial based cloning system s such as cosmids, PI and bacterial artificial chromosomes.

The construction o f a cosm id-pocket map spanning the two regions in chro m oso m e Ip show ing loss o f heterozygosity (LOH) in hum an breast cancer is described in C hapter 3. A total o f 1589 cosm ids w ere assigned into 26 different pockets in ch rom o som e lp 3 1 .1, w hile 278 clones in chrom osom e lp l3 .1 defined three pockets. Taking th e entire length

o f chrom osom e 1 into account (263M b) we have produced cosmid pockets defining up to 5% o f the total genom ic size o f chrom osom e 1. The entire set o f cosm id pockets was subm itted to the Sanger Centre, H inxton, Cam bridge, U K to be integrated into the C hrom osom e 1 database, producing a prelim inary representation o f these regions already covered in YACs. The com plete set o f data can be visualised at http://w ebace.sanger.ac.uk/cgibin/w ebace?db=acedbl& class=Laboratorvcfeobiect=U niv ersitv% 20of% 2Q London.

This cosm id pocket m ap constitutes not only a significant contribution for the H um an G enom e Project effort but also a valid resource to overcom e future problem s in obtaining the full sequence o f hum an chrom osom e 1. The usefulness o f this resource has already been dem onstrated:

• in confirm ing the Y AC m ap integrity, by defining the overlapping segm ents and the degree o f overlap betw een adjacent YAC clones

• in overcom ing the problem o f chim aerism observed in some o f the Y A C s used, e.g. Y A C 904 _c_ l and YAC 883_e_l 1

• in extending bacterial contigs and closing gaps, as described in C hapter 4

The chrom osom e specific cosm id pocket map should also be a valuable resource to study regions that are duplicated or shared betw een chrom osom e 1 and other chrom osom es. For exam ple, areas o f hom ology betw een clirom osom e 1 and other chrom osom es have been described - chrom osom es 6 and 9 have sho w n to possess regions w ith D N A sequences hom ologous to the pericentrom eric region o n chrom osom e 1 (H ughes et al, 1998). These ordered sets o f cosm ids could turn out to be o f great assistance in distinguishing betw een the sequences originating fro m a specific chrom osom e w hen the tim e to assem ble and confirm the genom ic sequences arrives.

7.3 Construction of a contig of overlapping bacterial clones

It is necessary to build overlapping clone maps in order to facilitate an in-depth study o f the genom ic regions o f interest using readily accessible cloned DNA. The resulting high resolution physical maps may then be integrated with genetic and cytogenetic m aps to create a m olecular framework for the identification expressed sequences. T h e next part o f this thesis, which is described in Chapter 4, has involved the construction o f an

overlapping set o f bacterial clones spanning 1.1Mb o f hum an chrom osom e lp l3 .1 , and encom passing the sm allest region o f overlap show ing loss o f heterozygosity in hum an breast cancer. This m inim al region has been defined betw een the loci D 1S440 and D 1S453. These LOH studies in the region are thought to indicate the presence o f a putative tum our suppressor gene associated w ith breast cancer (H oggard et al, 1995a; Brintnell et al, 1997). Furthermore, this region has also been strongly associated with other types o f cancer, e.g. head and neck carcinomas (Jin Y et al, 1998), suggesting a more general role for a possible tum our suppressor gene m apping to the area.

The approach used for construction o f physical clone contigs relied on screening PAC/BAC and cosmid libraries with the YACs spanning the region. The positive FACs constituted the lp l3 .1 enriched sub-library. Genetic markers, available at the tim e that this work begun, w ere also used to enable the PGR verification o f small num bers o f clones previously positive by the YAC probe hybridisation. Further PACs from the lp l3 .1 sub­ library w ere identified and the small contigs thus generated w ere subsequently linked using PAC insert probes. However, one gap o f a m axim um o f 50 -100Kb still remained. Screening o f the flow sorted chromosome 1 cosmid library identified several cosmids, w hich had been previously positioned in the cosmid pocket map. These cosm ids extended further into the gap thus reducing its size. The resulting contig o f overlapping clones consists o f approxim ately 1.1Mb o f genom ic D NA , w ith one gap o f approxim ately 50Kb.

The difficulty in bridging both sides o f the gap m ay be due to extensive num ber o f repetitive sequences in the area, w hich m ake the m ap hard to resolve. D irect sequencing, vectorette and inverse PC R techniques w ere also used to generate end probes that could then be used for identifying further clones. H ow ever, even though these techniques w ere successful at generating clone-end fragm ents, the p ro b es derived from them failed to detect any new specific clones in the libraries screened. In order to close this gap, it may be therefore necessary to use other resources, su ch as hum an genom ic libraries w ith a m uch larger total coverage - e.g. RPCI-11 (a library w ith a coverage o f 25.3 genom e equivalents - SC and W U G SC , 1998).

7.4 Identification of expressed DNA sequences in lp l3 .1

A s the com plete sequence o f the human genome is m ade publicly available, it becom es the logical substrate for targeted in silico identification o f novel sequences. Som e large- scale sequencing centres release the data already fully annotated by gene prediction program s used on the genom ic sequence for the identification o f novel genes. Ideally, this strategy w ould ghastly and com pletely identify all hum an genes. H ow ever, the softw are used by this strategy is derived from self-know ledge acquiring “neuronal" netw ork algorithm s, and the current degree o f know ledge about all gene and gene- specific elem ents is far from 100% accurate. Therefore, in order to confirm the predictions obtained in silico it is necessary to re-exam ine these data in vitro (for m ore details see section 1.2.7). D espite the huge effort by the Sanger Centre to fully sequence hum an chrom osom e 1, data for the Ip 13.1 region are still scarcely available at the tim e o f w riting this thesis. This w ay, the search for transcribed sequences in the region could only be perform ed by m eans o f traditional gene hunting strategies - cD N A selection and exon trapping. Gene detection strategies can be quite efficient w hen traditional techniques such as cDNA selection and exon trapping are used in com bination. In the case o f identification o f any cDNAs and exons, w hen large scale D N A sequence analysis data becom e available, it may allow to further elucidation on gene and protein structure.

In this thesis, nine bacterial clones from the m inim al tilling path o f the bacterial contig described in C hapter 4 w ere employed as the framework for expressed sequence identification w ithin the smallest region o f overlap on chromosome Ip 13.1 show ing LOH in hum an breast cancer. Exon trapping was carried out in conjunction w ith cDNA selection. W hile exon trapping is not expression dependent, requiring only th e presence o f a functional splice acceptor and donor site, cDNA selection is an expression dependent technique and only the breast and bone marrow derived cDNA libraries w ere available. It was hoped that these two methods w ould complement each other and thus enable good coverage o f the region. Also, a safe theoretical assumption would be that a putative breast cancer tum our suppressor is expressed in normal breast tissue.

The exon trapping experim ents described in Chapter 5 identified two ex o n s from one gene: a l,2 -m a n n o sid ase IB, localised w ithin the sm allest region sh o w in g loss o f heterozygosity in hum an breast cancer on chrom osom e lp l3 .1 , b etw een the two

cytogenetic m arkers that delim it the SRO (D1S440 and D 1S453). H owever, the overall efficiency o f this m ethod to identify the m axim um num ber o f exons w ithin the region studied w as ham pered by incom plete B s tX l digestion o f the am plified products derived from transfected C 0 S 7 cells. At the beginning o f this study, no genes w ere know n to m ap to the clones used in this experim ent, thus not providing any direct positive control o f the gene-identification efficiency. W ith effective optim isation o f the exon trapping strategy, it is predicted that an even higher num ber o f exons can be identified w ithin this region.

Even though carefully validated, the cD N A selection protocol w as not sufficiently optim ised to detect rare transcripts, and contam inants such as ribosom al D N A w ere co­ selected w ith the cDNAs, thereby m asking true results. N evertheless, in this w ork (described in Chapter 5), cD N A selection led to the isolation o f cloned cD N A s derived from breast tissue m apping to the region. BLA ST searches perform ed through the N C B I web site o f available databases o f nucleotide and protein sequences that w ere not associated w ith repetitive elem ents w ere perform ed on these clones. H om ology to a probable novel gene coding for a prostaglandin p2a receptor regulatory protein w as this w ay identified.

These prelim inary screens led to the identification o f two genes, representing partial success at detection o f expressed sequences from the region by m eans o f exon trapping and cD N A selection techniques. Searches for cD N A selected clones w ith in the norm al

breast cD N A library corresponding to the a 1,2-m annosidase IB gene w ere

unsuccessful, probably due to the low expression o f this gene in breast tissue. Future efforts w ill therefore concentrate on optim ising both protocols, enabling the identification and isolation o f any new putative exons/cD N A s from th e constructed libraries. I f any m ore transcripts are found m apping to the area, their p ossible roles as candidate tum our suppressor genes should be assessed.

7.5 Could al,2-m annosidase IB be a possible tumour suppressor gene