Generation of large deletion mutants from plasmid DNA

In the course of our work on the vitamin D receptor (VDR) protein, we became interested in the generation of several large deletion mutant versions of this protein for structure-function relationship studies. One goal was the deletion of a 50 amino acid sequence from the 427 amino acids of the native protein in addition to the introduction of several point mutations throughout the remaining part of the cDNA. At the DNA level, this deletion of 50 amino acids is equivalent to deleting 150 nucleotides from a total of about 1300 bp. These 150 nucleotides had to be removed and the resulting gap in the sequence closed without altering the overall reading frame of the gene. The starting material available to us was a 6200-bp plasmid, designated pcVDR, containing the entire wild-type VDR cDNA. Several methods for the introduc-tion of such relatively large deletions have been described. Clearly, if prop-erly located unique restriction enzyme recognition sites are present in the starting gene, the desired fragment can be excised and the gap closed by the use of a suitable adapter molecule that would also restore the correct reading frame. Such conveniently located unique restriction sites can be introduced by design if the gene of interest is generated by chemical synthesis. Unfortunately, such a situa-tion occurs rarely in the case of native sequences, including our starting VDR cDNA. Probably the most widely used method for introducing large deletions is inverse PCR (1), where a circular DNA molecule is used as the template, the primers are designed to hybridize in a tail-to-tail fashion at locations immediately adjacent to the sequence to be deleted, and the remaining larger fragment of the molecule is amplified. The PCR product can then be recircu-larized by enzymatic ligation, result-ing in a product that has the desired sequence deleted. The main chal-lenges in this scheme are the relatively large size of the PCR product and, more importantly, the low efficiency of its intramolecular self-ligation. The efficiency of this self-ligation is espe-cially low in the case of a blunt-ended PCR product. To increase the efficien-cy of this step, unique restriction en-zyme cleavage sites can be engineered into the PCR primers. Following the amplification reaction, treatment of the PCR product with the restriction endonuclease(s) generates comple-mentary single-stranded overhangs that can facilitate the ligation reaction. One must take care to ensure that the restriction sites thus introduced are not present elsewhere within the PCR product. Additional difficulties arise when, as in our case, the reading frame of the final recircularized product is to be preserved, and no additional amino acids are to be introduced. Our first attempts to generate the desired deletion mutant by blunt-ended recircularization of the 6050-bp PCR product obtained by inverse PCR were unsuccessful. Various ligation conditions were tested at different PCR product concentrations and in the presence of compounds reported to enhance the efficiency of blunt-end ligations, such as polyethylene glycol (PEG) (2). No recombinant colonies could be obtained after the transfor-mation of competent bacterial cells with the ligation products. Next, we attempted to generate complementary sticky ends at the PCR termini by in-troducing linker molecules having a BamHI recognition site. These linkers were blunt-end ligated onto the PCR product and digested with BamHI. Since our PCR product also contains a single BamHI site, it was first treated with BamHI methyltransferase to make this internal site resistant to enzymatic hydrolysis. This approach also proved unsuccessful, and no transformants were obtained. We then turned our attention to the use of partially phosphorothioate-modified PCR primers and T7 gene 6 exonuclease (Amersham Biosciences, Piscataway, NJ, USA) digestion as an alternative method for introducing complementary sticky ends at the ends of the PCR product. Phosphorothio-ate residues are known to be resistant to various endonucleases and exo-nucleases (3). This property has been exploited, for example, in a method where one of the PCR primers contains several phosphorothioate residues at its 5′ end that become incorporated into the PCR product. The treatment of this PCR product with T7 gene 6 exonucle-ase, a 5′ to 3′ double-stranded specific exonuclease, results in the complete hydrolysis of the unprotected DNA strand, leaving the phosphorothioate-protected strand intact (4). For our needs, we modified this method as illustrated (Figure 1). In this figure, the amino acid residues shown are those of the VDR protein, flanking the two ends of the 50 amino acid fragment that we wanted deleted (Δ165–215). Both PCR primers were designed to incorporate four consecutive phosphorothioate residues, located 12 nucleotides from the 5′ ends of the prim-ers. This allows the generation of single-stranded 3′ overhangs of the same length at both ends of the PCR product, follow-ing the digestion of the initial blunt-end-ed product by the 5′ to 3′ exonuclease activity of the T7 gene 6 exonuclease. Importantly, the 6 outermost nucleotides of each PCR primer are complemen-tary to the 6 nucleotides of the opposite strand primer located immediately to the 5′ end of its phosphorothioate residues. Thus, these 6 outermost nucleotides at the 5′ ends of the PCR primers do not initially hybridize to the DNA template. Their presence, however, assures that the 12-nucleotide long single-stranded 3′ tails obtained after the T7 gene 6 exo-nuclease treatment of the PCR product