TABLE 7.1 Sequencing of cosmid subclones using octamers.
CHAPTER 8: GENERAL DISCUSSION
The main aim of this thesis was to research further the use of short oligonucleotides for the detection of coding regions from amongst bulk DNA. Preliminary studies using hybridization o f short oligos have been carried out (Estivill and Williamson, 1987; Melmer and Buchwald, 1990; Melmer et al, 1990) which have explored the possibilities of using oligos based on rare- cutter restriction sites for selecting clones for long range restriction mapping projects. Studies have also been performed in which short degenerate oligos based on splice site consensus sequences were hybridized to cloned DNA, specifically to sequences bearing splice sites (Melmer and Buchwald, 1992). Possibilities exist for the extension of these methods for the detection o f CpG islands and coding regions, either in bulk genomic DNA or in uncharacterized DNA clones covering regions of the genome that are of interest with regards to genetically linked hereditary disorders. Whilst there are already methods for detection of CpG islands and coding regions, as discussed in Chapter 1, these methods can often be slow
and labour intensive. The development of additional technologies would seem appropriate.
8.1 DETECTION OF CPG-RICH SEQUENCES USING SHORT OLIGOS
In Chapter 4 conditions were established for the amplification by PCR, with reasonable specificity, between two octadeoxyribonucleotides based on rare cutter sequences. PCR using such short primers may have a number of potential applications. For instance this method was used on human total genomic DNA under relaxed stringency conditions to generate a CpG island "mini-library" by cloning the resulting PCR product. Enrichment of over 60-fold was observed. This could possibly be increased if longer primers were used, with perhaps some
G / c degeneracies towards the 5’ end of the primers. Such a library could be used to establish many Sequence Tagged Sites (STS’s), with a high probability of being adjacent to or within coding regions. For a sequence to be useful as a STS, it should be short (200-500bp) with enough sequence data available from the flanks to design unique primers for single locus PCR amplification (Olsen et al, 1989).
Most mapping projects using STS’s have so far involved the use of STS’s derived from well characterized clones or sequences. In order to map the less well characterized regions of the genome it will be necessary to develop methods of generating large numbers of novel STS’s from undefined DNA fragments specific to selected chromosomal regions. The PCR based method used here to develop CpG rich mini-libraries from genomic DNA could easily be applied to this task, perhaps using microdissected chromosomal fragments as template. Several other methods have been developed for generating STS’s using Alu-Alu PCR on rodent/human radiation hybrids (Cole et al, 1991) and Notl/Alu PCR (Patel et al, 1991). Because the Alu primers are specific for the human repeat, rodent sequences are selected out. Notl/Alu-PCR which uses ligation of Notl linker/primer to Notl digested DNA, followed by Notl-Alu PCR has the added advantage that the sequences can then be used to identify CpG islands within the region. However only CpG islands that have a conveniently close Alu repeat nearby and in the correct orientation can be identified this way. Alu repeats are not distributed randomly across the genome and are found more frequently in light staining G- bands, whereas LINE repeats occur more in dark G-bands (Korenberg and Rykowski, 1988; Moyzis et al, 1989; Chen and Manuelidis, 1989). LINE PCR primers could be used for isolating sequences from dark staining G-bands and other rare-cutter linker/primers could be used to improve the efficiency of these methods. The PCR method developed in Chapter 4 could be used to detect CpG islands in any region of interest, regardless of the presence of repeat elements and has the advantage over the method sescribed by Patel et al (1991) by
virtue of being simpler and requiring fewer steps.
Interest in the use of CpG islands as gene markers is growing. A recent study (Larsen et al, 1992) showed that 46% of all first exons and 14% of all exons are CpG rich, suggesting that CpG island sequences could be used to identify transcripts since they often extend into exons. A CpG island library could provide an unbiased collection of DNA segments corresponding to the promoters of approximately 60% of human genes. CpG islands have previously been made by cloning the small fragments from the digestion of genomic DNA with Hpall. However this method fragments the CpG islands into small segments which are of little use. A recent article discusses an attempt to generate a CpG island library by removing methylated CpG regions by passing DNA through an affinity matrix that contains the methyl-CpG binding domain from the rat chromosomal protein MeCP2, leaving intact CpG islands (Cross et al, 1994). The method for generating CpG islands discussed in this thesis (4.6.3) would give larger segments than the Hpall method, but would not give whole intact CpG islands as with the affinity column method. In addition, the assessment of the CpG islands in Chapter 4 by sequencing part of the clones was inconclusive, since most of the sequences were under 200 bp- the minimum size for a genuine CpG island (Larsen et al, 1992). If only the sequences over 200 bp are taken into account, then the enrichment for CpG content is only 36-fold (rather than 6 6-fold, as mentioned in Chapter 4). This compares poorly to the
enrichment of 80-fold claimed for the affinity column method (Cross et al, 1994). Further analysis of some of the TA clones obtained would have been useful. It might also have been appropriate to screen a cosmid library using several clones and to analyse the selected cosmids for conserved regions, in order to provide evidence of the effectiveness of this method. Also further attempts at using the TA clones for hybridzation against northern blots or zoo blots would have been useful.
Amplification between GC oligos was unable to identify CpG islands within cosmid clones, being incapable of discriminating between genuine CpG islands and the strongly CpG rich regions of the vector DNA. Amplification was not occurring between rare cutter sites within the vector since none are present, but was probably taking place at areas of high, but less than 100% homology, with the primers. Annealing stringency during the PCR was relaxed with the intention of allowing such low specificity primer extensions to occur within the CpG islands of the cloned genes because these did not have many rare cutter sites close enough for specific amplification. The cosmid cosGSTrp? had single Notl sites in the CpG islands of both the glutathione-s-transferase and NADH-ubiquinone oxidoreductase genes. Cosmid cosCTl had no Notl sites. In addition low stringency was necessary because the purpose of the experiment was to establish conditions which could be applied to any cosmid clone prior to any knowledge of rare cutter restriction sites and therefore required a degree of flexibility in annealing stringency. The method may be workable if the vector DNA could be excised prior to amplification. For the cosmids used in these experiments (cos202- Kioussis et al, 1987) this was not practicable, but may be possible for cosmid clones in other vectors such as pWE15 (Wahl et al, 1987) and SuperCosl (Stratagene) and lambda clones. The method may be applied effectively to YAC’s, whose vector DNA is a relatively minor component.
8.2 SEQUENCING DIRECTLY INTO CPG ISLANDS IN CLONED DNA
In Chapter 7 correct sequence from the CpG islands o f the model cosmid clones was achieved by using direct approaches, either by:-
i. PCR-sequencing from rare cutter sites with octamer primers, having first cut the clone close by to allow primer extension in only one direction (subclones were used, but the method could be used on whole cosmid clones).
ii. Ligase mediated PCR-sequencing, by attaching a linker/primer to rare-cutter sites in the cosmid.
The first method has the drawback of requiring extensive restriction mapping before it can be used on a new clone, whereas the second only requires knowledge of the presence of a rare-cutter site within the clone and this could be achieved by selecting clones by screening cosmid libraries with octanucleotides based on rare-cutter sites, as has been done previously (Estivill and Williamson, 1987; Melmer and Buchwald, 1990; Melmer et al, 1990). This second method gave one correct sequence from cosGSTrp? and another sequence assumed to be the result of two sequences from two Notl restriction fragments of the same size. The method needed refinement and could have benefited from the use of biotin labelled primer linkers. Streptavidin coated magnetizable particles could then be used to separate, clean and obtain single-stranded primer-linked DNA fragments prior to sequencing.
These methods cut out the labour intensive procedures involved in cloning the PCR products and would not produce a high background of sequences from the vector genome (see Chapter 4). Clone-specific STS’s at CpG islands could be generated this way, either for cosmids or YACs. The STS would then be mapped back onto the clone and the position of the clone within the genome could be mapped using the STS. Sequence from the STS could then be used to design primers for sequencing the flanking regions which may contain coding regions.
8.3 IDENTIFICATION OF CODING REGIONS USING SHORT CONSENSUS