Environmental and Molecular Mutagenesis 22:152-156 (1 993)

Classification of Mutations at the HLA-A Locus by Use of the Polymerase Chain Reaction

Grant Joseph, Scott Grist, Frank Firgaira, David Turner, and Alec Morley Department o f Haematology, Flinders University and Medical Centre,

Bedford Park, South Australia, Australia

We investigated whether the polymerase chain reaction (PCR) could be used to determine the mechanism of mutation in lymphocyte clones mutated at the HLA-A locus. Three polymor- phisms, at Factor XIIIA, D6S109, and intron 3 of the HLA-A gene, were used to study a series of clones previously characterised by Southern blotting (SB) at multiple loci on chromosome 6. For detection of loss of heterozygosity, the re- sults of PCR and SB were concordant in 140 of 141 clones when polymorphism in the Factor XI- HA region was studied and in 144 of 145 clones when polymorphism in the HLA-A gene was studied. For classification of the mechanism of mutation, PCR and SB gave the same result in 88

of 92 clones (96%) when a combination of the HLA-A and Factor XlllA polymorphisms was used and in 46 of 47 clones (98%) when a com- bination of the HLA-A and D6S109 polymor- phisms was used. The results indicate that PCR provides a simple and reliable method for cate- gorising mutations at the HLA-A locus as arising from mitotic recombination, deletion, or from presumptive minor changes within the gene. Rare events such as gene conversion, nondis- junction, or large deletions extending to the te- lomere will be misclassified. However, such events are rare for mutations at this locus. 0 1993 Wiley-Liss, Inc.

Key words: mutations, lymphocytes, PCR

INTRODUCTION

Of the various methods developed for measuring in vivo mutations in humans, the HLA-A system has been particu- larly valuable in providing information on mutations at an autosomal gene locus and indicating the importance of seg- regation following mitotic recombination as a mechanism for gene loss. Use of Southern blotting (SB) to study restric- tion fragment length polymorphisms and to estimate gene dosage has permitted the classification of mutations at the HLA-A locus into those that show no change (presumptive point mutations, microdeletions, and microconversions) and into those that are due to gene deletion. mitotic recombina- tion, gene conversion, and nondisjunction [Turner et a]., 1988; Morley et al., 19901. Classified in this way 60-65% of mutants fall into the no change category. 30-35% result from mitotic recombination, and 2-5% result from gene deletion. Rare cases of gene conversion but no cases of nondisjunction have been detected. The absolute numbers of mutants in the no change and recombination categories have been found to increase with age [Grist et al.. 19921.

The use of SB for classification of mutations has several disadvantages. It is time consuming and technically de- manding, requires a relatively large amount of DNA, and cannot precisely measure gene dosage. For these reasons we have investigated whether mutations could be satisfactorily classified using length polymorphisms detectable by the

0 1993 Wiley-Liss, Inc.

polymerase chain reaction (PCR). We chose a number of PCR polymorphisms localized to chromosome 6 and used them to analyse mutant clones which had previously been classified by SB.

MATERIALS AND METHODS

Clones

A total of 177 mutant and 12 wild type clones isolated from 17 individuals were studied. Mutant clones were iso- lated by immunoselection using a monoclonal antibody di- rected against either HLA-A2 or HLA-A3 and complement, followed by cloning at limiting dilution [Janatipour et al.. 1988; McCarron et al., 19891. Mutant clones included some in which no change was observable on analysis by SB and some in which the mechanism was mitotic recombination, gene conversion, or gene deletion; clones resulting from nondisjunction have not yet been observed. The criteria for classification of mutant clones by SB have been previously described [Morley et al.. 19901 and are shown in Table I .

Received November 30, 1992; revised and accepted June 2. 1993.

Address reprint requests to Grant Joseph, Department of Haematology. Flinders University and Medical Centre, Bedford Park. South Australia, 5042 Australia.

HLA Mutations and PCR 153

TABLE 1. Expected Findings From SB or PCR in Mutants of Various TvDes

Gene dosage' LOH" of nonselected LOH at

at HLA-A allele telomeric loci

No change 1 - SB

PCR Deletion

SB + 1 PCR + SB + 2 PCR + SB + 2 PCR +

-

Recombination

Gene conversion

-(small) + (large) -(small) + (large)

+ +

d l o ~ s of heterozygosity: " I = hemizygosity , 2 = homozygosity; '- = LOH absent, + = LOH present.

Polymorphisms

The polymorphisms and PCR systems used in this study were as follows:

(a) a tetranucleotide microsatellite polymorphism at the Factor XIIIA (FXIIIA) locus [Polymeropoulos et al., 19911, which is situated at 6p 25-p24 near the telomere of the short arm of chromosome 6.

(b) a dinucleotide microsatellite at the D6S109 locus, which is situated at 6p 21.3-p24 [Ranum et al., 19911.

(c) A polymorphism within intron 3 of the HLA-A gene on 6 ~ 2 1 . 3 . which was detected using the following primers:

5' ACC AGA GAG TGA CTC TGA GGT T 3'

5' GCA CAG AAC CCA GAC ACC AGC 3'

In informative individuals, these primers produced 2 ampli- fied fragments differing in length by 21 bp [Firgaira et al., 19921. Informative individuals comprise approximately 49% of the population.

Primers were synthesized on a DNA synthesizer (Applied Biosystems Model 38 I A) according to the manufacturer's instructions. The polymerase chain reaction was performed as described by Saiki et al. [1988]. PCR products for the HLA-A and FXIIIA polymorphism were detected by elec- trophoresis through 6% polyacrylamide gels and staining with ethidium bromide. The D6S 109 dinucleotide polymor- phism was detected by incorporating "S adATP into the PCR reaction. electrophoresis of the amplified products on a 6% polyacrylarnide urea gel, and detection of the amplified material by autoradiography .

Classification by PCR

Mutant clones were classified by PCR by regarding as a mitotic recombinant any clone in which heterozygosity

present in wild type clones from the lymphocyte donor had been lost at either of the telomeric FXIIIA or D6S109 loci. Where informative, the FXIIIA polymorphism was pre- ferred for scoring heterozygosity, as detection of the ampli- fied products could be simply performed by ethidium bro- mide staining. Clones which retained heterozygosity at one of these telomeric loci were regarded as deletion mutants if heterozygosity was lost at HLA-A and as "no change" mu- tants if heterozygosity was retained at HLA-A. It was recog- nised that, since it is difficult to determine gene dosage by PCR, cases of gene conversion would be misclassified as deletions, and large deletions extending from HLA to the telomere would be misclassified as recombinations (Table I). However, our previous experience has shown that these types of mutations were rare [Morley et al., 19901, and it was felt that this error was acceptable.

RESULTS

Examples of the results observed for FXIIIA in wild type, no change, and recombinant clones are shown in Figure 1. In wild type and no change mutants heterozygosity in infor- mative individuals was demonstrated by the presence of two (occasionally one, if resolution was inadequate) bands, which represented homodimers for the two alleles and also, in nearly all cases analysed by polyacrylamide gels. by the presence of two slowly migrating additional bands, presum- ably heterodimers, each containing one DNA strand from each of the two alleles. In contrast, mitotic recombinant clones showed a single band, derived from the remaining single FXIIIA allele syntenic with the unselected HLA al- lele. In these cases no heterodimer bands were observed. Examples of the results observed by autoradiography for the D6S 109 polymorphism are shown in Figure 2.Wild type and no change mutants showed 2 main bands, representing the 2 alleles, together with several shadow bands, whereas re- combinant clones showed only one major band of the allele syntenic with the unselected HLA-A allele and several shadow bands.

Comparison Between Southern Blotting and PCR for Detection of Loss of Heterozygosity

At FXIIIA and HLA-A it was possible to examine corre- lations between SB and PCR for detection of loss of het- erozygosity, as for each locus there was a close spatial relationship between the DNA sequences responsible for polymorphism detected either by SB or PCR. The results for Factor XIIIA are shown in Table I1 and those for HLA-A are shown in Table 111. There was agreement between the re- sults of SB and PCR in all but one of the 14 I clones studied at FXIIIA (99.3%) and in all but one of the 145 clones studied at HLA-A (99.3%). The two clones for which the results were discordant were (MMM4) for the FXIIIA re- gion and (OMBM2.7) for the HLA region, and are dis- cussed below.

154 Joseph et al.

Fig. 1. Results of PCR using the Factor XllIA primers. Lanes 1-6, 7-12, and 13-17 represent clones from 3 individuals. Lanes 6 , 12, and 16 are wild-type clones; lanes 3-5, 9, I I, and I S are “no change” mutants. and lanes I . 2. 7 , 8, 10, 13, 14, and 17 are mutants due to recombination, FXIIIA alleles are clearly resolved in the second and third individuals but not in the first, in whom loss of heterozygosity is indicated by loss of heteroduplexes

TABLE 11. Comparison of Heterozygosity in the Region of Factor XIIIA as Assessed bv Southern Blotting or PCR

Fig. 2. Autoradiograph of polyacrylamide gel electrophoresis of poly- morphisms analysed by PCR at the D6S109 locus. The 10 tracks show results obtained from control (unselected) clones, mutant clones displaying no change of heterozygosity at the HLA-A (selected) locus and mutant clones exhibiting loss of heterozygosity at HLA-A locus extending to the DS6109 locus. The results are from clones derived from S different individ- uals. The individual’s initials are underlined in the clone designation. In each clone either 2 major bands or I major hand plus several shadow bands are seen: track 1, nonselected control clone (MMCI) heterozygous; track 2, recombinant clone ( M M M J ) homozygous; track 3, no change mutant (MMcM6) heterozygous; track 4, recombinant clone ( M M c M S ) homozy- gous; track 5, no change mutant (OTBM2) heterozygous; track 6, recom- binant clone (OTBMS) homozygous: track 7, no change mutant (OWM3) heterozygous; track 8, recombinant clone (OVWM2) homozygous; track 9, nonselected control clone (OGTC) heterozygous; track 10, recombinant clone (OGTM13) homozygous.

Comparison Between Southern Blotting and PCR for Classification of Mutant Clones

Of all the mutant clones studied, 92 mutants previously classified by SB were also able to be classified by PCR using the FXIIIA and HLA-A polyrnorphisrns and the criteria stated in the Methods. The results are shown in Table IV and indicate 96% concordance between the 2 classifications, with discrepant results being obtained in only 4 clones (MMM4, OMRM2.2. OMRM4, and OMBM2.7). Forty- seven mutant clones were similarly classified using the

~

Southern blotting PCR heterozygosity

heterozygosit y Retained Lost

Retained 19 0 Lost 1 61

TABLE Ill. Comparison of Heterozygosity in the Region of HLA-A as Assessed by Southern Blotting or PCR

PCR heterozygosity Southern blotting heterozy gosity Lost Retained

Ret ined 91 I Lost 0 53

TABLE IV. Comparison of the Classification of Mutations by Southern Blotting or PCR Using Polymorphisms at Factor XIIIA and HLA-A

Southern PCR PCR blotting concordant discordant

Recombination 43 42 1 (MMM4) Deletion 6 5 1 (OMRM2.2) Gene conversion 1 0 1 (OMRM4) No change 42 41 1 (OMBM2.7)

Clones for which the PCR classification did not agree with SB results are shown in parentheses.

D6S109 and HLA-A polyrnorphisms. The results are shown in Table V , and they indicate 98% concordance. with only 1 clone (OMBM2.7) being discordant.

The results and interpretations in the four clones showing discordant results were as follows.

MMM4

This clone had been scored as a recombinant by SB on the basis of loss of heterozygosity at HLA-A and FXIIIA, and

HLA Mutations and PCR 155

TABLE V. Comparison of the Classification of Mutations by Southern Blotting or PCR Using Polymorphisms at D6S109 and HLA-A -

Southern PCR PCR blotting concordant discordant

Recoinbi nation 26 26 0 Deletion 4 4 0 Gene conversion -

No change 17 16 1 (OMBM2.7)

The clone for which the PCR classification did not agree with the SB classification is shown in parenthesis.

- -

gene dosage showing homozygosity at HLA-A. The PCR results showed loss of heterozygosity at HLA-A and D6S109, but not at FXIIIA, and taken by themselves would have suggested a large deletion. The explanation for these discordant results is uncertain, particularly as the discor- dance persisted after repetition of both SB and PCR. If the FXIIIA PCR locus is telomeric to the FXIIIA SB locus, then it is possible that assessment of gene dosage by SB was incorrect, so that MMM4 actually represented a deletion, or that strand migration during recombination moved a sub- stantial distance into the FXIIIA region of chromosome 6. The results could also be explained by a double mitotic recombinational event. Whether mitotic recombinational frequencies are elevated with respect to physical distance near the telomere, as is seen in meiotic recombination, is not known.

OMRM2.2

Both SB and PCR agreed ir showing loss of heterozygos- ity at both HLA-A and FXIIIA. This clone was scored as a recombinant by PCR but as a large deletion by SB as the result of the measurement of gene dosage. The individual from whom this clone originated was not informative at D6S 109.

OMRM4

Both SB and PCR agreed in showing loss of heterozygos- ity at HLA-A and retention of heterozygosity at FXIIIA. This clone war; scored as a deletion by PCR but as a case of gene conversion by SB as the result of measurement of gene dosage. The individual from whom this clone originated was not informative at D6S 109.

OM BM2.7

PCR showed loss of heterozygosity at HLA-A and reten- tion at D6S 109 and FXIIIA. However, SB showed retention of heterozygosity for HLA-A and FXIIIA so that the clone was scored as a “no change” mutant. We believe the results of both SB and PCR are correct, and this clone probably represents a micro deletion, or conceivably a point mutation leading to a mismatch with the 3’ end of one of the HLA-A PCR primers.

DISCUSSION

The present results show that use of PCR at several loci can satisfactorily determine the molecular nature of the great majority of spontaneous in vivo mutations detected at the HLA-A locus. Hitherto, SB has been the only technique available for classification of HLA-A mutations. Theoreti- cally it should be able to correctly classify all mutations, as it provides information on both loss of heterozygosity and gene dosage, but in practice it falls short of this ideal due to its imprecision in measurement of gene dosage. PCR will misclassify cases of gene conversion and large deletions, but both these types of mutation are rare and account for only 15 of 434 in vivo HLA mutations ( 1.1 %) which we have stud- ied and classified by SB. Since recombination is responsible for approximately 30% of HLA mutations, it is even possi- ble that misclassification by SB o f a small proportion of such mutations may be a greater source of error than misclassifi- cation by PCR.

In 4 clones the interpretations of SB and PCR were discor- dant. In 2 of these there was no discordance between the results of SB and PCR insofar as loss of heterozygosity was concerned, and the different interpretations of the 2 methods resulted from the measurement of gene dosage by SB. This led to one clone being classified by SB as being due to deletion and one to gene conversion, whereas by PCR the former was classified as being due to a recombination and the latter to a deletion. The measurement of gene dosage by SB has been discussed [Turner et al., 19881, and it is possi- ble that one or both clones had been misclassified by SB based on the statistical criterion used for homozygosity. In the other 2 clones the results for heterozygosity were dis- crepant between SB and PCR. The findings in one clone could be explained by an unusual recombination or deletion, and in the other by the occurrence of a small deletion. Even if these 4 clones are regarded as having been incorrectly classified, PCR using FXIIIA and the HLA-A polymor- phism was correctly able to classify 96% of mutations, and PCR using D6S109 and the HLA-A polymorphisms was correctly able to classify 98% of mutations. However, since the classification based on SB may have been incorrect and that based on PCR may have been correct in one or more of these clones for the reasons previously discussed, PCR on the whole may result in correct classification in more than 9698% of mutations.

The chief limitation of PCR for classification of mutation as described here is the availability of informative polymor- phisms. Approximately 78% of individuals are heterozy- gous at Factor XIIIA, 92% at D6S109, and 49% at Intron 3 of HLA-A. However, the difficulty in having informative polymorphisms for an individual will decrease as new poly- morphisms are progressively discovered.

Compared to SB, PCR is simple, quick, and cheap. It has the additional advantage that much less preceding tissue culture is required. as only a small amount of DNA is needed. For these reasons our current approach is to use

156 Joseph et al.

PCR to study in vivo mutations. There does remain the possibility that certain mutagenizing agents cause a high frequency of mutations due to gene conversion or large gene deletion and. in this circumstance, that misclassification by PCR might be numerically important. Accordingly, it may be necessary when using new mutagenizing agents that clones displaying loss of heterozygosity should initially be analysed for gene dosage before the approach we present here for classification is routinely used.

ACKNOWLEDGMENTS

This study was supported by the National Health and Medical Research Council. We thank Mr. D. Bourel for supplying the antibodies.

REFERENCES

Firgaira FA, McEvoy C, Dempsey J, Turner DR, Morley AA (1992): HLA-A allele specific length polymorphism detectable by PCR. Submitted.

Grist SA. McCarron M, Kutlaca A,Turner DR, Morley AA (1992): In vivo human somatic mutation: Frequency and spectrum with age. Mutat Res 266:189-196.

Janatipour M, Trainor KJ, Kutlaca R. Bennett G , Hay J , Turner DR. Morley AA (1988): Mutations in human lymphocytes studied by an HLA selection system. Mutat Res 198:221-226.

McCarron MA, Kutlaca A, Morley AA (1989): The HLA-mutation assay: Improved technique and normal results. Mutat Res 225: 189-193.

Morley AA, Grist SJ, Turner DR. Kutlaca A. Bennett G (1990): The molecular nature of in vivo mutations in human cells at the autoso- ma1 HLA-A locus. Cancer Res 50:45844587.

Polymeropoulos MH, Rath DS, Xiao H. Merril CR (199 I ): Tetranucleotide repeat polyniorphistn at the human coagulation factor XIIIA subunit gene (F13A1). Nucleic Acids Res 19:4306.

Ranum LP, Chung MY, Duvick LA. Zoghbi HY, Orr HT (1991): Dinucle- otide repeat polymorphism at the D6S109 locus. Nucleic Acids Res 19: 1171.

Saiki R K , Gelfand DH. Stoffel S , Scharf SJ, Higuchi R. Horn GT, Mullis KB. Erlich HA ( 1988): Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239:487491.

Turner DR, Grist SA, Janatipour M. Morley AA (1988): Mutations in human lymphocytes commonly involve gene duplication and re- semble those seen in cancer cells. Proc Natl Acad Sci USA 85:3 189.

Accepted by- J. Cole