Human Molecular Genetics

RESULTS

The overall strategy is described here briefly and in more detail below. It begins with the cloning of a PCR product corresponding to p53 mRNA into a DNA binding domain vector (Fig. 1A). The two-hybrid interaction of wild-type p53 with the p53-interacting protein 53BP1 (6) results in transcription of the HIS3 reporter gene, whereas a mutant p53 that fails to bind 53BP1 does not activate this reporter gene (Fig. 1B). Colonies are patched onto plates that select only for the presence of the two-hybrid plasmids (Fig. 1C, left) and replica-plated on to plates that indicate whether the two-hybrid interaction occurs (Fig. 1C, right). A representative example of the results is shown in Figure 1D.

In order to enable one-step cloning of p53 alleles we generated a Gal4 DNA binding plasmid, pGBT53GAP, which contains two 120 bp segments of DNA encoding p53 flanking a unique restriction site; these encode residues 73-113 and 353-393. Upon linearization by digestion at this site the plasmid was unable to transform yeast efficiently. Co-transformation of this linearized vector with a PCR product corresponding to the entire p53 open reading frame results in circularization by recombination (Fig. 1A). Transformations carried out in the presence of the PCR product resulted in at least a 20-fold and often a 50- to 100-fold increase in the number of transformants (data not shown). The p53-interacting protein 53BP1 was encoded as an activation domain hybrid by plasmid pSEBP1. We transformed the yeast reporter strain Y153 (8), which contains a HIS3 gene under regulation of Gal4 binding sites. This strain produces a low level of His3 protein in the absence of any plasmids. However, His3 activity can be inhibited using 3-amino-1,2,4-triazole (3AT). Thus at 25 mM 3AT it was possible to achieve a level of stringency such that only a wild-type, or nearly wild-type, level of interaction between p53 and 53BP1 was sufficient to yield growth of transformants in the absence of histidine (Fig. 1B, left).

Seven clinical samples were independently evaluated by the transcriptional activation assay (9) and by the two-hybrid interaction assay. The transcriptional activation assay is a functional assay that relies on the ability of the p53 protein to behave as a site-specific transcription factor in yeast. p53 cDNA is cloned by gap repair into a yeast expression vector by transformation of a yeast strain carrying an integrated ADE2 reporter gene under control of a p53-responsive promoter. The utility of the yeast ADE2 gene as reporter is that Ade⁺ colonies that result from wild-type p53 expression are white, whereas Ade^- colonies from mutant p53 expression are red, because of accumulation of an intermediate in adenine metabolism in ade2 yeast. RNA was isolated from cell lines derived from a malignant melanoma and other tumors and from frozen tissue of a head and neck tumor and reverse transcription-PCR (RT-PCR) with p53 primers was performed as described (9). Yeast yIG397, containing the ADE2 reporter gene, was co-transformed with a linearized p53 expression vector that requires gap repair to reconstitute a functional p53 coding sequence and the ability of the p53 proteins derived from the seven samples to activate ADE2 transcription was determined (9; Table 1). This assay indicated that one sample is derived from an individual who is homozygous wild-type, four from individuals who are homozygous mutant and two from individuals who are heterozygous.

The seven RT-PCR products, blindly coded, were used to co-transform yeast Y153 with linearized pGBT53GAP (Fig. 1A). The yeast strain contained the plasmid pSEBP1 encoding the Gal4 activation domain fused to 53BP1. Transformants were selected on plates lacking leucine and tryptophan, selecting for the LEU2 and TRP1 genes present in pSEBP1 and pGBT53GAP respectively. Control transformations included the linearized vector with no insert with a PCR product derived from wild-type p53 or with PCR products derived from p53 mutants R175H or R273H, neither of which can bind 53BP1. Colonies from each transformation were patched onto -Leu -Trp plates and replica-plated onto medium lacking histidine and containing 3AT to identify transformants in which the two-hybrid interaction occurred (Fig. 1C). Samples were scored to determine the ratio of transformants able to grow on the two-hybrid selective medium (i.e. His⁺) to the total number of transformants (Fig. 1D). Of the seven samples, six were in agreement with the transcriptional assay, while the seventh, FA10, was interpreted to be homozygous mutant by transactivation and homozygous wild-type by two-hybrid assay (Table 1). FA10 was sequenced and found to be homozygous for the mutation R267W, found in other human tumors. This mutation abolishes the ability of p53 to activate transcription in yeast but not to bind 53BP1, a property observed previously with other p53 mutants (10); other p53 mutations eliminate protein binding without affecting transactivation activity (10). For the six samples that yielded consistent results in the two assays confirmation of the mutations inferred from the transactivation and two-hybrid assays could also be obtained by DNA sequence analysis.

Table 1. Comparison of transactivation and two-hybrid assays

Patient	% Wild-type by transactivationa	Genotype by transactivation	Ratio wild-type by two-hybrid	% Wild-type by two-hybrid	Genotype by two-hybrid
FA1	49.6	+/-	89/224	39.7	+/-
FA5	5.8	-/-	0/104	0	-/-
FA6	5.2	-/-	0/104	0	-/-
FA7	100	+/+	133/143	93.0	+/+
FA8	5.4	-/-	0/104	0	-/-
FA9	42	+/-	72/184	39.1	+/-
FA10	10	-/-	99/104	95.2	+/+

^aBetween 200 and 500 colonies were scored for transactivation.

DISCUSSION

We present in this report a simple yeast-based assay to determine whether a mutation that affects protein function is present in a gene. This two-hybrid assay should prove to be a useful additional means for screening of human genes implicated in disease. It has some disadvantages, including the inability to detect all point mutations of interest or alleles present in a heterozygous state with the wild-type that result in loss of expression of the protein. The assay cannot be used if an interacting protein is not available or if the interaction is not sensitive to the mutant state of the protein that is to be tested. However, this method is inexpensive and rapid to perform and requires no DNA purification, ligation or bacterial transformation steps or any equipment beyond that necessary for simple yeast procedures. The screen detects heterozygosity at the RNA level, avoids identification of irrelevant mutations and should be able to identify mutant proteins even if the clinical sample is derived from a mixture of wild-type and mutant tissues.

For large proteins implicated in disease, small domains may exist that are responsible for binding to different proteins, such that a panel of two-hybrid interactions could be used in the assay. For example, domains of the large Brca1 (1863 residues) and Brca2 (3418 residues) proteins implicated in human breast cancer susceptibility interact by two-hybrid analysis with a RING domain-containing protein (11) and Rad51 protein (12) respectively. However, as functional assays in yeast, such as the one described here, do not provide the site of mutation, these simple screens may prove most useful as initial assays, with results that indicate the presence of a mutation being followed up by additional methods to yield more precise mutational identification.

MATERIALS AND METHODS

ACKNOWLEDGEMENTS

This work was partially supported by a grant from Amgen Inc. S.F. is an investigator of the Howard Hughes Medical Institute.