Proc. Natl. Acad. Sci. USAVol. 79, pp. 6636-6640, November 1982Genetics

A phylogeny for the principal alleles of the humanphosphoglucomutase-1 locus

(evolution/gene frequencies/intraallelic recombination/mutation)

NORIO TAKAHASHI*, JAMES V. NEELt, CHIYOKO SATOH*, JUNKO NISHIZAKI*, AND NAOMI MASUNARI**Radiation Effects Research Foundation, 5-2 Hijiyama Park, Hiroshima 730, and 164 Sakurababa-machi, Nagasaki 850, Japan; and tDepartment of Human Genetics,University of Michigan Medical School, Ann Arbor, Michigan 48109

Contributed by James V. Neel, July 28, 1982

ABSTRACT The results of phosphoglucomutase-1 (PGM1)typings by starch gel electrophoresis and subtypings by isoelectricfocusing are presented for a sample of Japanese. A distinctionmade on the basis of isoelectric focusing (termed "+" and "-") isnonrandomly associated with each ofthe products of the four mostcommon electrophoretic alleles (PGM11, PGM12, PGM13, andPGM17). The isoelectric trait cosegregates with the allele; the de-gree of nonrandomness of the association varies from allele to al-lele. Thus, the four alleles become eight. On the basis ofthese factsplus the additive nature of the pI differences between allele prod-ucts and the geographical distribution of the alleles, an allele phy-logeny can be constructed. This postulates that the eight allelesmay be explained by three nucleotide substitutions involving thestem allele plus four intragenic recombinations between these sub-stitutions. The potential of intragenic recombination as a cause ofmutation has been insufficiently appreciated.

Some 6 years ago, Bark et al. (1) and Kuhnl et aL (2) reportedthat in Caucasoid populations the phenotypes associated withthe two common electrophoretic alleles of phosphoglucomu-tase-1 (PGM1), PGM/ and PGMI2 , could be subdivided by iso-electric focusing with thin-layer polyacrylamide gel as the sup-porting medium into two phenotypes, one migrating morecathodically than the other. These phenotypes were termed"+" and "-". From the manner in which this property coseg-regated with the two original phenotypes, it was intrinsic tothe PGMI locus. Thus, they postulated four alleles, subsequentlytermed PGMI/+, PGMI/-, PGM12+, and PGM12-. Their dataindicated that these two properties ofthe PGM1 alleles [namely,I vs. 2 and (+) vs. (-)] were not associated at random.

In the present communication, we will confirm this findingfor a Japanese population but then extend this isoelectric sub-division to the products of two rarer alleles encountered in theAsia-Pacific Ocean area, PGM13 and PGM17. Each of these lat-ter alleles can also be subclassified as (+) or (-), and again,these properties, although cosegregating, are not associated atrandom with reference to the electromorphs under study. Thisnonrandom association, taken in conjunction with the isoelec-tric focusing data, will be shown to permit the construction ofa gene phylogeny for the eight alleles that must be postulated.In this phylogeny, presumed intragenic recombination eventsplay a role equal to that of presumed nucleotide substitutions.

SAMPLES AND METHODSSamples. The blood samples on which the PGM1 typings

were performed were obtained from individuals included in astudy of the potential genetic effects of the atomic bombs beingconducted in Hiroshima and Nagasaki (3, 4) and from individ-uals in a preceding pilot study (5). Because no putative muta-tions involving the PGM1 locus have as yet been encounteredin this study, the samples have been analyzed without referenceto the radiation histories of the parents of the individuals in

question. Samples examined by isoelectric focusing had beenkept in liquid nitrogen since aliquots of them had been typedby starch gel electrophoresis. The 184 samples whose pheno-type was PGM1 1-7 were drawn from both of the previouslyreferenced studies. Because of ambiguities in subtyping thePGM17 electromorph when it occurs in combination with thePGM12 electromorph, we present only the results ofsubtypingindividuals with genotypes PGMI1/PGMI7. The 28 sampleswith phenotype either PGM1 1-3 or PGM1 2-3 were drawnonly from the study of the genetic effects of the atomic bombs(3, 4). In the pilot study (5), an allozyme whose mobility wassimilar to that of the PGM13 reported by Hopkinson and Harris(6) was named as PGM1 3NGS1 because, given the existence ofthe heterogeneity in PGM13 as detected by starch gel electro-phoresis in the different populations in the world (7), its identitywith the original PGM13 could not be assured (8). Therefore,the phenotypes of the PGM13 samples subtyped by isoelectricfocusing in this report would have been referred to either asPGM1 1-3NGS1 or PGM1 2-3NGS1 in our first report. For thesake of simplicity, however, in this paper we will use the termPGM13 as synonymous with PGM1 3NGS1-

Methods. Hemolysates were prepared from frozen packederythrocytes that had been stored in liquid nitrogen as de-scribed by Ueda et aL (9). Starch gel electrophoresis was con-ducted as described by Satoh et aL (5). For the isoelectric fo-cusing, an Ampholine-containing polyacrylamide gel plate[acrylamide in gel = 5% (wt/vol); N,N'-methylenebisacrylam-ide in total acrylamide = 3% (wt/vol)] with a dimension of 12x 11 x 0.1 cm was prepared by the method in LKB applicationnote 250 with slight modifications. Riboflavin and N,N,N',N'-tetramethylethylenediamine were used instead of ammoniumpersulfate, and polymerization was carried out under an ultra-violet light. Final concentrations of the components in this gelwere as follows: riboflavin, 0.0005%; glycerol, 10% (vol/vol);Ampholine (pH 5-7), 2% (wt/vol); N,N,N',N'-tetramethyleth-ylenediamine, 0.056%. Isoelectric focusing was performed onthe LKB 2117 Multiphor apparatus with the LKB 2103 powersupply at 4°C. The electrode solutions were 0.01 M NaOH(cathode) and 1% aqueous acetic acid (anode). Prefocusing wasof 1-hr duration at the maximum limits of 1,200 V, 10 W, and10 mA. Hemolysates were applied 3 cm from the anodal elec-trode strip by using a 5 x 5 mm piece of Whatman 3MM filterpaper; focusing was carried out for 2 hr at the maximum limitsof 10 mA and 2,000 V. The application strips offilter paper werethen removed, and the electrofocusing was terminated afteranother3hratthemaximum limits.

For the population survey, we used the methods mentionedabove with slight modification. Gel plates with a dimension of

Abbreviation: PGM1, phosphoglucomutase-1.

6636

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

021

Proc. NatC Acad. Sci. USA 79 (1982) 6637

24 x 11 x 0.1 cm were used. Prefocusing was done at themaximum limits of 1,200 V, 20 W, and 20 mA, and focusing wasdone with 2,000 V and 20 mA.

Staining solution was prepared as described by Spencer etaL (10) and applied to the gel surface with a brush; the gel wasincubated for 1 hr at 37°C in the dark. A combination micro-electrode, type DE 103 (Toko Chemical Laboratories, Tokyo),connected to a Hitachi-Horiba digital pH meter, F-7 AD (Hor-iba, Kyoto, Japan), was used for the pH measurements. Beforemeasurement ofeach gradient, the pH electrode was calibratedwith standard buffers (pH 4 and pH 7) at 4°C. For the pH mea-

surements, the, polyacrylamide gels were cut into segments (15x 5 x 1 mm), and the segments were eluted in 0.5 ml ofdeion-ized water overnight at 4°C.

RESULTSThe Standard- PGM1 Alleles in Japanese. Table 1 presents

our experience thus far with this population in typing PGM,with starch gel electrophoresis. We. refer to the phenotypesrecognized by this approach as "standard." The appellation"other" embraces at least 10 different phenotypes in which avery rare allele occurs in combination with one ofthe four morecommon alleles, as well as two examples of homozygosity forthe PGMI4 allele. Table 1 also presents the allele frequenciesthat were calculated not only on the basis of the eight pheno-types shown but also from the number of the common and rarealleles observed in other rare phenotypes. These frequenciesare similar to those previously reported from Japan by others(reviewed in ref. 7) and ourselves (5).

Isoelectric Eocusing of a Random Subsample of the Popu-lation. Table 2 presents phenotype and allele frequencies on

the basis of subtyping by isoelectric focusing, a random sub-sample of the individuals contributing to Table 1. All of thesesamples had been typed previously by starch gel electropho-resis. Note that the number ofpersons with the standard PGMI7allele who were subtyped was quite small and that no persons

having the standard PGM13 allele were included in this partic-ular subsample. Considering only the results of subtyping ofthe standard PGMI/ and PGM12 alleles, we note that the non-random association of the (+) and (-) attributes with the elec-trophoretic classification is quite similar to that already reportedfor Caucasoid populations (11). Extensive pedigree data (notpresented) demonstrated that the isoelectric trait cosegregateswith the PGM1 type classified by starch gel electrophoresis,thus indicating an intrinsic property of the gene. In conse-

quence, 85.5% of the PGMI/ alleles would be classified asPGM11+, and 70.4% of the PGMI2 as PGM12 . The frequencyof the observed isoelectric phenotypes is in good agreementwith the expectation predicted from the allele frequencies on

the assumption of Hardy-Weinberg equilibrium.

Table 1. The PGM1 phenotypes and allele frequencies in a

sample of 14,575 Japanese from Hiroshima and NagasakiPhenotype No. Percent Allele Frequencies

1 8,477 58.161-2 4,919 33;752 693 4.75

1-7 316 2.172-7 76 0.52 PGM,'- 0.76337 8 0.05 PGM12 0.2196

1-3 34 0.23 PGM7 0.01402-3 9 0.06 PGM13 0.0015

Other rare phenotypes 43 0.30 Other 0.0015

14,575 99.99 0.9999

Table 2. The PGM1 phenotypes and allele frequencies in asubsample of 788 unrelated Japanese from Hiroshima andNagasaki, as revealed by isoelectric focusing

Number NumberPhenotype observed expected Gene frequencies*1+ 359 359.881+, 1- 120 121.63 PGM11+, 0.67581- 13 10.28 PGM.1-, 0.11421+, 2+ 151 152.73 PGM12+,0.14341+, 2- 70 64.22 PGM12-, 0.06031-, 2+ 25 25.81 pGM/7+,0.00571-, 2- 6 10.85 PGM/7-,0.00062+ 19 16.202+,2- 12 13.632- 3 2.871+, 7+ 6 6.071-,7+ 3 1.032-,7- 1 0.06

Others 0 2.76Total 788 788.02

Phenotypes with n(expected) below 3 were combined for x2 calculation.* 2= 4.1683; 0.50 < P < 0.75; degrees of freedom = 5.

Isoelectric Focusing of a Selected Subsample of Products ofthe Standard PGM13 and PGM17 Alleles. The results of sub-typing by isoelectric focusing the allelic products of the PGMI3and PGM17 alleles thus far encountered- in our laboratory areseen in Table 3. Of the 28 PGMI3 gene products subjected toisoelectric focusing, 23 (82.1%) were "+" in type, whereas ofthe 184 PGM17 gene products typed by such focusing, 176(95.7%) were "+." This difference is significant (X2 = 7.71;degrees offreedom = 1; 0.01 > P > 0.005); intragenic disequi-librium is more marked for the (+) and (-) forms of PGMI7.The pI from Isoelectric Focusing of These Eight Pheno-

types. Table 4 presents the pI values in our system of the eightprimary allozymes produced by the alleles under considerationand the average pI differences between them. These values arebased on 16 determinations for the 2+, 7+ phenotype (3 per-sons), 17 determinations for the 2+, 2- phenotype (2 persons),11 determinations on a 1-, 7- individual, 11 determinationson a 1-, 3- individual, 11 determinations on a 1+, 3+ indi-vidual,. 6 determinations on a 1+, 7+ individual, 6 determi-nations on a 1-, 7+ individual, and 2 determinations on a 3+,7+ individual, for a total of 80 determinations. Fig. 1 shows thevarious PGM1 phenotypes ofhuman hemolysates examined byisoelectric focusing. Fig. 2 is a schematic of the data of Table4 for an idealized focusing gel. With the PGM11 pattern as a

Table 3. Results of subtyping by isoelectric focusing a selectedsample of persons having the PGM13 or PGMI7 alleles

PGM13 PGM17Phenotype No. Phenotype No.

1+, 3+ 19 1+, 7+ 1431-, 3+ 1 1-, 7+ 331+, 3- 3 1+, 7- 6

1-,3- 1 1-,7- 22+,3+ 2 1842-,3+ 1

2+,3- 128

Because of difficulties in subtyping PGM17 electromorphs in thepresence of the PGM12' electromorph, only individuals whose geno-types are PGM,'/PGM,7 have been subtyped.

Genetics: Takahashi et aL

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

021

Proc. NatL Acad. Sci. USA 79 (1982)

Table 4. The pI values of the products of the eight alleles underconsideration in the study, with an analysis of thecomplementarity of the observed differencesIsoelectric Isoelectric

type pI value type pI value1+ 6.3 1- 6.42+ .6.1 2- 6.27+ 6.0 7- 6.13+ 5.8 3- 5.9

Differences

~~~0.5 +-= .1+ - 7+= 0.5 1-- 7-- 0.31+ -3+ =0.5 1- -3- =0.5

reference, for either the (+) or the (-) subtype, the pI differ-ence between PGM11 and PGM13 is close to the sum of thedifferences between PGMll and PGM17 and between PGM11and PGM12. Conversely, of course, the pI difference betweenPGM11 and PGM17 is obtained by subtracting the differencebetween PGM11 and PGM12 from the difference betweenPGM11 and PGM13. As will shortly become apparent, we donot believe this finding to be a coincidence.

DISCUSSIONShortly after the subdivision of the PGM11 and PGM12 phe-notypes by isoelectric focusing (1, 2), Carter et aL (11) recog-

+

41

......'l"".'

0* 0 0 0o

7w

0

.....o'EP"o

",..

4 5} 6

FIG. 1. Isoelectric patterns of six PGM1 isoenzymes from hemo-lysates. Phenotypes: 1+, 1- (lane 1); 1+, 7+ (lane 2); 1+,7- (lane3); 1+, 3+ (lane 4), 1+, 3- (lane 5); 2+, 2- (lane 6). The main PGM1bands are marked with closed squares; and the PGM21 bands aremarked with open circles.

FIG. 2. Position after isoelectric focusing of the two PGM1 bandsin the eight phenotypes described in this paper. For references, theposition of the two PGM21 bands (which also appear in these prepa-rations) is shown.

nized that the intragenic nonrandom association of the (+) and(-) attributes with the previously recognized alleles permittedthe construction of a probable allele phylogeny-i. e., a recon-struction of the sequence in which the four alleles (PGMI/+,PGMI1-, PGM12+, and PGM12-) could have arisen by a com-bination of two mutations and one subsequent intragenic re-combination. Their observations included those made in a seriesof 14 higher primates [Hominoidea: Pan (7), Gorilla (4), andPongo (3)]; in typings by starch gel electrophoresis and isoelec-tric focusing, all samples appeared identical and similar to thePGM1L+ phenotypic pattern. This observation suggested thattheir phylogeny be rooted in the PGM11+ gene, with a timedepth of no more than 6-10 million years, as estimated fromthe antiquity ofthe evolutionary divergence ofthe lines leadingto Homo and to the other Hominoidea. Given this root and theobserved allele frequencies, they postulated that two indepen-dent mutations resulted in the PGM/'- and PGM12+ alleles,with the PGM12- allele the result of an intragenic crossover.

The present data, in conjunction with the known facts con-cerning the geographic distributions of these alleles, suggest aconsiderable extension of that four-allele phylogeny. The exist-ing data on the distribution of the PGM1 and PGMI7 alleles inthe Asia-Pacific area, based on the summary of Blake and Om-oto (7), are schematically presented in Fig. 3. Although scat-tered instances of these two electromorphs have been encoun-tered all over the world, both these electromorphs (which oncloser analysis are everywhere proving to be heterogeneous),on the basis of current information, show local (but different)maxima in the Asia-Pacific area, with overlapping distributions.-The PGM13 localization principally involves Melanesians,whereas the (partially overlapping) PGMI7 localization extends

3+Second band _of PGM21i = = = = = ==

3-

7+Main band . = = = = = X---of PGM21 - __

2+ 7-

2-

1+1- e

6638 Genetics: TakAashi et aL

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

021

Proc. NatL Acad. Sci. USA 79 (1982) 6639

FIG. 3. The known distributions of the PGMI3 and PGM17 allelesin the Asia-Pacific area. E, PGMI3; A, PGM17; E,PGMI3 andPGM17.

to Malaysians, Thais, Chinese, and Japanese. Large areas re-

main to be explored in this respect, as does the question of thesubtypes of PGM13 in the areas already studied. Thus, on thebasis of a single sample from New Guinea, the PGM13 phe-notype encountered there is reported to differ slightly from thePGM13 of Okinawans (7) and from Japanese on Kyushu andHonshu (unpublished data); clearly a much more extensivecomparison is called for.

In an extension of the phylogeny, the chief constraint is theneed to observe the additive (or subtractive) pI relationshipsbetween alleles as described in Table 4. There are a numberof different solutions to this "phylogeny problem. " We will, tobegin with, accept the phylogenetic relationships of PGMI"+,PGM11-, PGMI2+, andPGMI - as proposed by Carter et aL (11),although, as these authors point out, other interpretations are

possible. The other four alleles (PGM13+, PGMI3-, PGMI7+,and PGMI7-) can then be explained by one additional nucleo-tide substitution and three intragenic recombinational events.These latter can in each instance result from at least two typesof intragenic recombination. The resulting phylogeny is shownin Fig. 4. A very approximate site for the postulated intragenicrecombination is indicated for each ofthe recombination events.

In theory, additional evidence concerning the most probablesequence of events is provided by allele frequencies and theextent ofthe "linkage disequilibrium. " It is appropriate that themost common ofthe four Asia-Pacific alleles. (PGM1 7+) is shownas the oldest, although, because ofthe possible roles ofselectionand drift in establishing allele frequency, we do not regard thefrequency criterion as particularly compelling. Another crite-rion in setting the sequence of events could be the amount ofintragenic linkage disequilibrium. Other factors equal, themore recent recombinational possibilities should present thegreatest disequilibrium. However, the precise time curve for

- 2 pL.-6.2 + 7 1

AGOF2.- PGt7M

-OVER (ALSOWITH PGMh)WER 7 1

(ALSO FROm PGM-) PG17 2 GN

_ 7 2pI 5.9

PGMf-

FIG. 4. A proposed phylogeny relating eight alleles in the PGM1system. The symbol 1u indicates origin through mutation. The positionof the postulated crossover is in each instance indicated by an x.

the decay of the disequilibrium depends on the relative nu-cleotide distances between the three postulated nucleotide sub-stitutions; because these are unknown, no firm use can be madeofthe decay function, the more so because genetic selection anddrift intervene between recombination frequency and the ob-served phenotypes.

This phylogeny must be recognized as only a beginning forthe alleles of this locus. Thus, there are at least eight other al-leles (ref. 5;. unpublished data) not yet studied with respect totheir subdivision by isoelectric focusing, and, in addition, Scoz-zari et aL (12) have reported a common polymorphism affectingthe thermostability of the products of the PGMI/, PGM/,PGM12+, and PGM12- alleles, the site for which, from its segre-gation pattern, must be intragenic to the PGMI locus. In prin-ciple, most of the alleles waiting to be defined with respect toisoelectric focusing can be fitted into the present phylogeny bythe type of logic we have used.An attraction of this phylogeny is that it is ultimately subject

to verification or disproof by DNA sequence determination.When a probe becomes available for the PGMI locus, not onlywill the postulates ofthe phylogeny be subject to empirical test-ing, but the prevalence of DNA restriction enzyme-site poly-morphisms should provide additional markers within and nearboth ends of the gene. Indeed, the opportunities for sophisti-cated evolutionary biology should provide a stimulus to developa probe for this locus. In this connection it should be borne inmind that, in the present context, the important issue is not thecorrectness in detail ofthe phylogeny of Fig. 4 but the principlethat intragenic recombination has been as important as nucleo-tide substitution in the genesis ofthe alleles encountered at thislocus. It is very likely that DNA sequence determination of thePGMI locus will result in some revisions in the phylogeny.

The possibility that intragenic recombination could be acause of "point" mutation in higher eukaryotes has been ap-parent ever since the nature of the genetic code was established;the nature of the products of the human haptoglobin allelesprovided an example of the consequences of an error in suchrecombination some 20 years ago (13). The recognition of in-tervening sequences (introns) in genes increases the potentialrole for this type of event. In this context it is worth noting thatintragenic recombination imparts a self-generating aspect tomutation: if three or four nucleotide substitutions (n) resultingfrom mutation become established in any substantial frequency

Genetics: Takahashi et al

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

021

Proc. NatL Acad. Sci. USA 79 (1982)

in a population, intragenic recombinational events create theopportunity for a much larger number of alleles, the total num-ber ofpossible alleles (nucleotide substitutions plus subsequentrecombinational events) increasing as 2". Thus, it is not sur-prising that our phylogeny is based on three nucleotide sub-stitutions (which have polymorphic frequencies) but four cross-over events. We have suggested on the basis of the number ofdifferent alleles that the PGM1 locus appeared more mutablein a Japanese than in an English population (14); the presentanalysis provides a mechanism for the phenomenon. Other hu-man loci with complex allelic structures which with further datashould lend themselves to a treatment similar to that presentedin this paper are those responsible for the MNSs and Rh bloodgroups and the HLA-A and HLA-B histocompatibility factors.The present findings suggest that the appearance of a new prop-erty at these loci (i.e., mutation) is as likely to result from in-tragenic recombination as nucleotide substitution.Some might prefer to refer to the consequences of the pos-

tulated crossover events as the evolution of haplotypes ratherthan as mutations. We use the term 'point" mutation in theclassical sense-as the appearance in an offspring ofan inheritedphenotype not present in either parent, the genetic basis forwhich can be localized to a specific point on a chromosome, aswould be the case ifphenotype PGM13+ appeared for the firsttime in a child of a PGM12+/PGM17+ heterozygote.

Each of the alleles postulated to arise through intragenic re-combination has a substantially higher probability of multipleorigins than an allele resulting from a specific nucleotide sub-stitution. From its molecular weight, PGM1 should consist of=510 amino acids, thus requiring for its coding 1,530 nucleo-tides in exons, these probably. interrupted by a considerablenumber of nucleotides in intervening sequences. In the so-called complex loci of Drosophila, the frequency of "within-lo-cus" recombination varies from about 10'- to 105 per locus pergeneration, depending on which characteristics of the locus areunder consideration (reviewed in ref. 16). Let us assume thatintragenic recombination at the PGMI locus occurs with a fre-quency similar to that observed at these complex loci (whosemolecular basis is still poorly understood); the resulting muta-tions would have relatively high frequencies. On the otherhand, the probability ofmutation per nucleotide per generationin humans is of the order of 10-8 (17). Although all mutation isofcourse repetitive, the probability ofrandom loss ofa mutationin the small, highly subdivided populations in which our ances-tors lived was so high (cf. ref. 18) that many ofthe establishedpolymorphisms due to nucleotide substitutions may be unifocalin origin. Thus, whereas the nucleotide substitution postulatedto give rise to the "7" attribute of the PGMI locus may haveoccurred and persisted only once in the Asia-Pacific area, thealleles arising from the postulated intragenic recombinationevents should be shown by the appropriate analysis ofDNA tohave multiple origins.

Finally, we note that the distribution of the PGM13 andPGMI7 alleles in Japan may shed light on the peopling of theseislands. A common version is that the prehistoric inhabitantsof Japan were Ainu-like peoples who were gradually drivennorthwards by migratory pressures originating from the souththrough the Ryukyu Archipelago (Nansei Islands) and more cen-trally through the Korean Peninsula. The Ainu do not possessthe PGM13 or PGMJ7 alleles (19, 20). The Chinese populations

examined thus far do not possess. the PGM13 allele (summaryin ref. 7). Our studies (unpublished data) demonstrate that thePGM13 allele is roughly 3 times more frequent in the Nagasaki(Kyushu) sample than in the Hiroshima (Honshu) sample. Thissuggests that the genetic influence on present-day Japanese ofthe southern migratory stream is some 3 times greater in south-western than, in central Japan, but this observation, of course,does not permit us to estimate absolute contributions. It is inthis connection of interest that the phosphohexose isomeraseallele PHI4 HIR 1 is likewise about 3. times more common in Na-gasaki than in Hiroshima (14), but much less is known con-cerning its distribution in the Asia-Pacific area than for thePGM1 alleles in question.

This study was supported by the Radiation Effects Research Foun-dation, a private, nonprofit Japanese foundation financed equally by theGovernment of Japan through the Ministry of Health and Welfare andthe Government of the United States through the National Academyof Sciences under contract with the Department of Energy. The par-ticipation of J.V.N. was also supported by Department of Energy Con-tract ACO-2-76-EV02828.

1. Bark, J. E., Harris, M. J. & Firth, M. (1976) J. Forensic. Sci. Soc.16, 115-120.

2. Kiihnl, P., Schmidtmann, U. & Spielmann, W. (1977) Hum. Ge-net. 35, 219-223.

3. Neel, J. V., Satoh, C., Hamilton, H. B., Otake, M., Goriki, K.,Kageoka, T., Fujita, M., Neriishi, S. & Asakawa, J. (1980) Proc.Natl. Acad. Sci. USA 77, 4221-4225.

4. Satoh, C., Awa, A. A., Neel, J. V., Schull, W. J., Kato, H., Ham-ilton, H. B., Otake, M. & Goriki, K. (1982). in EnvironmentalMutagens and Carcinogens, Proceedings of the Third Interna-tional Conference on Environmental Mutagens, Tokyo, eds. Sug-imura, T., Kondo, S. & Takabe, H. (Liss, New York), pp. 687-700.

5. Satoh, C., Ferrell, R. E., Tanis, R. J., Ueda, N., Kishimoto, S.,Neel, J. V., Hamilton, H. B. & Baba, K. (1977) Ann. Hum. Genet.41, 169-183.

6. Hopkinson, D. A. & Harris, H. (1966) Ann. Hum. Genet. 30, 167-181.

7. Blake, N. M. & Omoto, K. (1975) Ann. Hum. Genet. 38, 251-273.8. Ferrell, R. E., Ueda, N., Satoh, C., Tanis, R. J., Neel, J. V.,

Hamilton, H. B., Inamizu, T. & Baba, K. (1977) Ann. Hum. Ge-net. 40, 407-418.

9. Ueda, N., Satoh, C., Tanis, R. J., Ferrell, R. E., Kishimoto, S.,Neel, J. V., Hamilton, H. B. & Baba, K. (1977) Ann. Hum. Genet.41, 43-52.

10. Spencer, N., Hopkinson, D. A. & Harris, H. (1964) Nature (Lon-don) 204, 742-745.

11. Carter, N. D., West, C. M., Emes, E., Parkin, B. & Marshall,W. H. (1979) Ann. Hum. Biol 6, 221-230.

12. Scozzari, R., Trippa, G., Santachiara-Benerecetti, A. S., Terren-ato, L., Iodice, C. & Benincasa, A. (1981) Ann. Hum. Genet. 45,313-322.

13. Smithies, O., Connell, G. E. & Dixon, G. H. (1962) Nature (Lon-don) 196, 232-236.

14. Neel, J. V., Ueda, N., Satoh, C., Ferrell, R. E., Tanis, R. J. &Hamilton, H. B. (1978) Ann. Hum. Genet. 41, 429-441.

15. Race, R. R. & Sanger, R. (1968) Blood Groups in Man (Davis,Philadelphia), 5th Ed., pp. xxviii and 599.

16. Carlson, E. A. (1959) Q. 1Rev. Biol 34, 33-67.17. Neel, J. V. (1982)J. Hered., in press.18. Thompson, E. A. & Neel, J. V. (1978) Proc. Nati Acad. Sci. USA

75, 1442-1445.19. Giblett, E. R. (1969) Genetic Markers in Human Blood (Black-

well, Oxford), pp. xxvii and 629.20. Omoto, K. & Harada, S. (1972) J. Fac. Sci. Univ. Tokyo 4, 171-

211.

W%.f-xo Genetics: 'Takahashi et aL

Dow

nloa

ded

by g

uest

on

Janu

ary

29, 2

021