D irect selection technique

D irect selection, is a fairly new technique, developed in 1991 by two independent groups for rapid identification of cDNAs (Lovett et aL,

and which map to a specific chromosome(s) or smaller regions of chrom osom es.

In brief, cDNAs are prepared from either a single tissue/cell type or a m ixture of different tissues. Prior to selection, cDNAs are either cloned, or attached by ligation to PCR primers. cDNAs can then be am plified en m asse by PCR to provide relatively large quantities as starting m aterial for the direct selection procedure. If cDNAs come from more than one tissues they may be attached to different linkers and separated at a later stage.

Highly repetitive elements present in the cDNAs are suppressed by hybridisation with either Cotl DNA or total genomic DNA, to a low C o t,/2

value. However, low copy repeats like Y specific DYZ5 (section 1.2.2) may not be efficiently blocked and can be present in the resulting cDNA

selection library. The levels of contam ination from low copy repeat units can vary according to the chrom osom al region and the tissue of interest, but in general have been found to represent around 10% of the selection library (Lovett, 1994; Tassone er <3/., 1995).

cDNAs are selected by hybridisation to cloned genom ic DNA from a specific chrom osom al region. W hen the technique was first developed, the genom ic target cDNA clones were immobilised onto filters. However, experience has shown that the hybridisation steps are more efficient when the reaction is carried out in solution (Lovett, 1994; Del M astro et aL, 1995; Del M astro and Lovett, 1997). The target genomic DNA is tagged with biotin 16-UTP by nick translation, to allow capture of genom ic D N A and associated cDNAs by streptavidin-coated magnetic beads, after liquid hybridisation. Flow-sorted chromosomes (Rouquier et al., 1995),

chrom osom e or chrom osom e region specific YACs or cosmids (G uim era et aL, 1997; Lahn and Page, 1997; Simmons et aL, 1997) are the m ost

com m only used genomic target.

Tissue derived mRNA made into double stranded cDNAs

i

PCR amplified pool of cDNAs

Biotinylated, chromosome specific genomic DNAs

A

block repeats with Cot-I

Cot-I

Cot-I

Mix and hybridise

A:

■ Ill

cDNA/genomic target products captured by using streptavidin coated paramagnetic beads

Streptavidin) paramagnetic bead >

•Beads are washed and captured primary cDNAs are eluted •Re-PCR amplified for second round of selection

•Secondary selected cDNAs cloned and further analysed

H ybridisation is carried out at high stringency at 65° C, to an interm ediate C ot]/2 of between 100 and 200 (mole x sec/1). After

hybridisation, streptavidin-coated param agnetic beads are used to isolate the biotinylated genomic DNA and their associated cDNAs (Tagle et aL, 1993). The m agnetic beads are washed to rem ove any unbound single stranded sequences, and then the cDNAs are eluted from the beads. These cDNAs represent the primary selected cDNAs and are PCR am plified for a second time and passed through a further round of selection, exactly as above (Fig

1.7). Experim ents have shown that two rounds of direct selection are usually sufficient to yield an efficient enrichm ent of up to 100,000-fold; this

enrichm ent is not increased substantially by additional rounds of selection. The secondary selected cDNAs are then cloned into a plasm id vector and can be further analysed (Lovett, 1994; Del M astro and Lovett, 1997).

The direct selection method appears to have many advantages over the techniques described in the early part of this section. This method rapidly identifies coding sequences, which map to large genomic regions in a way, which does not depend on num ber and size of introns or cryptic splice sites. The method does not depend upon the presence o f CpG islands or transcription in hybrid cell lines (Parimoo et aL, 1991). Since this

m ethod allows great flexibility regarding both tissue and genom ic region, it can be used to identify genes that are active at different stages of

developm ent and body locations (M atsubara and Okubo, 1993). The cDNA fragm ents are collected as ESTs and can be used for the isolation of whole cDNAs and genes (Parimoo et aL, 1991).

cDNA direct selection was used by Osbourne-Law rence et al (1995), for isolation of expressed sequences within a 1Mb region of hum an

chrom osom e 17q21 that flanks B R C A l gene. In this case, a direct

com parison of efficiency of direct selection with exon trapping was m ade (Brody et aL, 1995). Results indicated sim ilar efficiency in identification of

expressed sequences between direct selection and exon trapping (>90% of sequences were covered by both techniques). In addition, direct selection enabled the identification of two extra low copy transcripts. Since the direct selection method relies on PCR amplification, only small amounts of RNA or cDNA as a starting m aterial are required. However, the selected PCR am plified products consist only of short fragments. As a result, full-length cD N A sequence needs to be obtained by further screening o f full-length cD N A libraries or by 5 ’ RACE.

Since 1991, a num ber of laboratories have used cDNA direct

selection with various modifications of the original m ethod to successfully identify and characterise novel genes. A modification of the technique was used to generate selected sublibraries, which contained expressed sequences conserved between hum an and mouse or pig (Sedlacek et al., 1993). In this experim ent the genomic tem plate com prised cosmids covering -8 0 K b from hum an chrom osom e Xq28. M ouse and pig cDNAs were used as

hybridisation probe and allowed the identification of novel cDNA sequences that share hom ology in hum ans and mouse and/or pig. This procedure has the potential to identify interesting sequences conserved betw een species, but is restricted to the isolation of strongly conserved sequences only, with num erous other less well conserved sequences going unidentified.

Tassone et al (1995) used 16 non-overlapping YACs, covering a total o f an -1 0 M b of hum an chrom osom e 21q to select 16 pools o f cDNA (one from each Y AC) from hum an foetal brain, whole hum an foetus, adult testis, thymus, liver and spleen. Analysis of 60 clones indicated that 19 sequenced cDNAs, matched to sequences in databases with a probability of p<0.01 and most of these detected mRNA transcripts by Northern blotting. The num ber of matched clones would be much higher if the experim ent were carried out today. It is of interest to note that when the corresponding

prediction, only 3 of the 19 sequences were predicted to contain easily recognised exons.

Del M astro et al (1995) produced five cDNA selection libraries using as genomic target a human chrom osom e 5 cosmid library covering around

174Mb of DNA. As cDNA template five different sources cDNA pools were used; hum an placenta, foetal brain, thymus, activated T-cells and HeLa cells. 261 clones from the HeLa selected cDNAs, were analysed and more than 50% mapped back to chrom osom e 5 and represented ESTs, rRNAs and novel cDNAs.

Simmons et al (1995) used a cDNA selection library to identify candidate genes involved in cri-du-chat syndrome. In this case the genomic target DNA was composed of 30 cosmids corresponding to hum an

chrom osom e 5 p l5 .2 These cosmids were used to select cDNAs from a m ixture of placenta, activated T cells and hum an cerebellum. Out of the 121 independent cDNAs selected, 21 were found to represent 5 separate novel cDNA sequences, emphasising the need to exam ine a large num ber of cDNA clones. Two more cDNA selected libraries were prepared from the same region of chrom osom e 5 by using two YACs spanning -2 M b to select cDNAs from a foetal brain cDNA selection library (Del M astro et al., 1995), and a norm alised infant brain library (Simmons et al., 1997). In this case though, the selected cDNA libraries were generated by hybridising the purified YACs onto filters containing high-density arrays of the cDNA clones isolated from the previous cDNA selection experim ent (Del M astro et al., 1995). Sixteen cDNAs were identified from the libraries and arrays of these four clones m apped back to the cri-du-chat region and becam e candidate genes for cri-du-chat.

The project, which is the topic of this thesis, began in 1997 and during the course of that year two further reports appeared, indicating that the direct selection approach was a pow erful tool in gene identification.

G uim era et al (1997) chose YACs from three regions of chrom osom e 21, the D ow n syndrom e critical region, and used them to construct a cosmid library. The pooled cosmids were used to select 576 cDNAs from a human, total foetus, cDNA library. 45 cDNAs were sequenced, some of which showed hom ology with genes, ESTs and STSs, while others were characterised as novel.

Finally and relevant to the present project, Lahn and Page (1997) constructed a cDNA selection library using as genomic target 3600 cosmids containing flow -sorted Y chromosomes to select adult testis cDNA. 3600 clones were sequenced. O f these 308 were putative novel, Y-specific sequences. O f those, Lahn and Page (1997) more fully characterised 12 clones and obtained full-length cDNA for 10 of them. The expression pattern of these 12 clones was investigated by Northern analysis and all w ere expressed in testis. These data are described in more detail in chapter 3.