Ann. Hum. Genet.

(1981), 45, 323-330 323

Printed i n Great Britain

Genetics of paraoxonase

BY H. EIBERG AND J. MOHR

University Institute of Medical Genetics, Tagensvej 14, DK-2200 Copenhagen, Denmark

SUMMARY

Danish family material comprising 1664 unrelated individuals (parents) and 3169 children,
as well as 699 grandparents of the same families, were examined for paraoxonase activity. A
micro-autoanalyser method, comprising a primary testing in tris buffer at p H 7.5 and, in the
case of primarily intermediate individuals, a secondary testing a t pH 10,was applied. This gave
a better discrimination than testing only at pH 7.5, because individuals around the low mode
of the primary activity distribution had their p H optimum at pH 10, while the optimum of
individuals around the high was a t p H 8-5. By this combined testing all individuals could be
unequivocally classified as ‘low’ or ‘high’, and the family material was compatible with the
low phenotype representing homozygosity for an autosomal recessive gene with a frequency
plow= 0.726. Out of 5532 individuals, 5 showed an almost complete lack of paraoxonase
activity.

INTRODUCTION

Paraoxonase (EC 3 . 1 . l .2) is an arylesterase in serum which hydrolyses paraoxone into

p-nitrophenol and diethylphosphate (Aldridge, 1953). I n Caucasians the enzyme activity of
paraoxonase in human serum has been found to be polymorphic (Geldmacher-v. Mallinckrodt
et al. 1972) and to have either a bimodal (Playfer et al. 1976; Zech & Zurcher, 1974) or a trimodal
distribution (Geldmacher-v. Mallinckrodt et al. 1973, 1978, 1979) ; accordingly, there may be
either an autosomal dominant (Playfer et al. 1976) or intermediate (Geldmacher-v.Mallinckrodt,
1978) mode of inheritance. I n the latter study the familial distribution showed some deviation
from the intermediate model; this was attributed to a broad overlapping of the presumed three
distribu ti ons.
This article reports our results from a study of the dependency of the paraoxonase activity
upon p H in tris buffer, for the purpose of deciding which p H gives the most pronounced
bimodality (and therefore the best prospect of genetic analysis); it was found, that a modified
procedure, with a pH 7.5 tris buffer instead of the pH 9.5 tris/glycine buffer used by Playfer
et al. (1976) gives an improved resolution.
We also report the results of testing a family material for paraoxonase by our modified
procedure, t o provide information concerning the genetics of the paraoxonase polymorphism.
For comparison, the material was tested both with the modified procedure and with that
applied by Playfer et al. (1976).
I n particular, the following questions were considered: ( 1 ) if modification of the test
procedure, especially concerning pH, gives a better recognition of genotypes than earlier
procedures; (2) if the familial distribution of paraoxonase activities is compatible with a two
324 AND J. MOHR
H. EIBERG
allele model, or if some more complicated model is required; (3) how Danish gene frequencies
compare with those of other populations; and finally, (4) if individuals with aberrant
paraoxonase activity show any peculiarity that could give a clue to the physiological significance
of paraoxonase.

MATERIALS AND METHODS

Blood serum from 832 two- and three-generation families (nos. 604-1505), stored for 5 years
in liquid nitrogen, were tested for paraoxonase activity (and for about 50 other genetic marker
systems).
For the purpose of discrimination between genotypes the paraoxonase was measured by two
different autoanalysis methods: the one described by Playfer et al. (1976) and our own modified
micromethod version of this method. The modified points were as follows. We diluted 50 p1 of
serum with 150 pl water (instead of not diluting) and used narrower autoanalyser tubes (half
volume) ;instead of buffered substrate we used water, which stabilized paraoxone and gave lower
background; and instead of glyeine buffer a t pH 10.5 we used 0 2 M tris/HCl buffer a t pH 7.5.
For assaying the dependence of paraoxonase activity upon pH, the activity was measured
for eight values of p H from 6 5 to 10 in 0.2 M tris buffer and from p H 9 to 11 in 0 2 M glycine
buffer. p H above 10 in tris buffer was not measured because of the weak buffer capacity in this
area. p H above 11 in glycine buffer gave too high a background in our hands.

RESULTS

(A) p H dependency
Using 0.2 M Tris we obtained a maximum activity a t about pH 10 for samples with low
activity and a t pH 8.5-9 for persons with high activity (Fig. 1). I n 0 2 M glycine we got a pH
optimum at p H 11 for samples with low activity and a t pH 10 for samples with high activity.
The best separation between low and high activity was attained in tris in the pH range pH
7.5-8.5. The shape of the curves for dependence of activity upon pH was different, for highs and
lows, the former being ‘cupola-shaped’ and the latter nearly linear (Fig. 1 ) . For samples with
low paraoxonase activity it was found that the activity doubled by a change of p H from 7-5
to 10 ; while for persons with high activity the activity remained roughly unchanged throughout
the interval pH 8-9. This could be used for a sharper categorization of phenotypes.

(B) Paraoxonase activity

The distribution of paraoxonase activity by determination a t pH 7.5, as compared with the
distribution obtained a t p H 9.5.
Paraoxonase activity measured by the technique of Playfer et al. (1976) at pH 9.5 for 400
of the 832 pairs of parents in our family material is shown in Fig. 2. A bimodal distribution
is recognizable, although less pronounced than that observed by Playfer et ab. (1976), in a
material of 190 unrelated British blood donors.
The distribution of paraoxonase activity as determined by the procedure presented here, i.e.
in tris buffer a t pH 7.5 for 1664 individuals (832 pairs of parents) is shown in Fig. 3. A sharp
 Genetics of paraoxonase 325

Activity
yM P N P / L

1500.

1200

900

600

300

I I I
7 8 9 10-
Fig. 1 . The paraoxonase activity of six individuals, three of the ‘low’ category ( 0 )and three of the
‘high ’ category (A)as a function of pH (curve no. 7 represents a 1 : 1 dilution of the sample represented
by curve no. 6). It appears that curves 1, 2 and 3 show the same kind of pH dependence, different
from that shown by curves 4-7.

Nu m be r of p e r s o n s

Fig. 2. Paraoxonase activity of 800 independent individuals (400 pairs of parents) assayed a t pH
9 5 in tris/glycine buffer.
326 AND J. MOHR
H. EIBERG
Number of p e r s o n s

100- I .

50-

a
0 600 1200 1800 Activity
pM P N P / I
Fig. 3. Paraoxonase activity of 1664 independent individuals assayed at pH 7.5 in 0 2 M tris
buffer; 878 persons with low activity and 786 with high activity.

Table 1. Distribution of ' low ' and 'high' paraoxonase activity among males and females and among
children, parents and grandparents in our family material (nos. 604-1505) :and gene frequencies
in the same categories on the assumption of a two allele model
Low High Plnw

Male Female Total Male Female Total Male Female Total

Children 850 844 I694 759 716 I475 0'727 0.736 0'730
Parents 422 456 878 410 376 786 0.710 0.740 0.726
Grandparents 136 224 360 123 216 339 0.725 0.714 0720
Total (5532) 1408 1524 2932 1292 1308 2600 0.725 0.733 0728

bimodality is immediately apparent, with not more than some 20 individuals so positioned in
the intermediate range, as to make their classification doubtful.
The sharper bimodality is due to persons with low paraoxonase activity having a lower
activity a t pH 7-5 than at p H 95, and persons with high activity having a higher activity a t
pH 7.5 than a t pH 95, so that the whole distribution is expanded.

( C ) Qene frequencies, age and sex

Among the 1664 unrelated individuals, 878 showed a 'low' paraoxonase activity, which on
the assumption of low activity representing homozygosity for a 'low activity allele ', low, would
give a gene frequency plow= 0726, and qhigh = 0.274. The limit between highs and lows was
placed a t a point slightly below the antimode (in accordance with the range of the high part
of the distribution being wider than that of the low part, i.e. so as t o minimize the number of
erroneous classifications). The point was, by a somewhat arbitrary judgment made by inspection
of the distribution (Fig. 3), put at 499/500 ,LAMPNP/1.
Consideration of a possible dependency upon age by comparison of the distributions of
grandparents, parents and children (Table 1), did not show any significant heterogeneity
 Genetics of paraoxonase 327

Table 2. Families of mating type high x low paraoxonase activity

r = no. of
children c = no. of children in family tested for paraoxonase
of low
activity I 2 3 4 5 6 7 8 I2 Total
0 0 9 I8 53 9 3 0 0 0 92
I 0 I2 32 43 4 2 0 0 0 93
-
2 8 33 84 8 2 0 0 0 ‘35
3 - - I0 52 9 3 I 0 0 75
4 - - - I 0 4 I 2 0 0 77
5 - - - - 0 I I 0 I 3
6 - - __ - - - 0 I 0 I

Total 29 93 242 34 I2 4 I I 416

(&, = 0.93, P > 0.05). Nor was there any sign of a sex influence, a comparison between the
distribution among males and females within the parental group showing no significant
difference (x$,
= 2.79, P > 0*05),and the total material of 2700 males and 2832 females showed
an even smaller difference (xt,
= 1.54, P > 0.05).

(D) Segregation
The material comprised 231 low x low matings with a total of 891 children, of whom 886 wcre
low, while 5 were primarily classified as high, or about 0.5 yo of the children, i.e. not more than
is to be expected from the character of the antimodal region (a later special testing of the pH
dependency of these five individuals, showed them to behave as lows, the activity rising
markedly by a change of pH from 7.5 t o 100).
The distribution of children in mating types high x low and high x high is shown in Tables
2 and 3. An analysis of segregation according to the method of Smith (1956) is represented in
Tables 4-8.
In Table 4 a, represents the expected number of lows per family of c children (with a t least
one low), assuming that ’low’ is in fact recessive, and b, is the expected variance per family.
The values of a, and b, are taken from tables in Smith (1956).This method of tabulation allows
a comparison, by eye or otherwise, of observed and expected numbers of lows for each family
size. The totals are compared by calculating
[total observed -total expected ( h , a , ) ] 2
xs1, = Total variance (Xm,b,)
A similar explanation applies to Table 5 as regards high x high matings with a t least one low
child.
In Table 6 we investigate whether the number of families with at least one low is a reasonable
fraction of all families with high x low parents. Here the values of d, are tabulated by Smith
(1956) while u is obtained from the estimated allele frequencies.
Table 7 does the same for high x high matings.
There was no significant deviation from the distribution expected if ‘low’ is due to an
autosomal recessive gene with frequency 0.726 (combined xf4)= 2.887; 0.50 < P < 0.70) (Table
8).
328 H. EIBERG
AND J. MOHR

Table 3. Families of mating type high x high paraoxonase activity

r = no. of
children c = no. of children tested for paraoxonase
of low -
activity Total
0 I01

I 46
2 33
3 3
4 I
5 0
6 I

Total 16 185

Table 4. Families of mating type high x low paraoxonase with at least one low child; no. of low
children
No. in No. of Expected
family, families, 0bserved lows, Variance
C mc lows mc a, mc bc
2 20 28 26.66 444
3 75 I 28 128.55 36.75
4 189 407 403.14 147.80
5 25 63 64'53 27.05
6 9 24 27'43 12.41
7 4 r6 I 4 1I 6.67
8 I 6 402 1.95
I2 I 5 6.00 2'99
Total 324 677 67444 24006
XJ~, = 0.027

Table 5. Families of mating type high x high paraoxonase phenotype with at least one low child :
number of low children
No. in No. of Expected
family, families, 0bserved lows, Variance,
c MC lows Mc A c Mc Bc
I I I 1'000 0'000
2 2 2 2.286 0.244
3 19 28 24643 4997
4 51 75 74'613 2 I .420
5 9 18 14'751 5.328
6 I I 1.825 0.776
8 I 6 2.223 1.172
Total 84 131 121.341 33-937
X t l ) = 2'749

(E) Lack of paraoxonase activity

Two out of the 3153 children in our material showed an almost complete lack of paraoxonase
activity, namely about 1/50 of the average 'low' activity. These two were sisters, 7 and 9 years
old, when the original samples were taken, and phenotypically normal a t 10 and 12 years of
 Genetics of paraoxonase 329

Table 6. Families of mating type high x low paraoxonase phenotype; division of matings between
those having low children and those without
No. of families with at least one low
No. of Total no. -
children, of families, Observed Expected Variance
C n
C mC nc udc n,ud,(I -udcf
2 29 20 I 8.30 6.75
3 93 75 68.46 18.06
4 242 189 190.97 40.27
5 34 25 27'72 51'22
6 I2 9 9'93 1.71
7 4 4 3'34 055
8 I I 0.84 0.14
I2 I I 084 013
Total 416 324 32040 I 18.83
xG, = 0.109
Table 7. Families of mating type high x high paraoxonase phenotype; division of matings
between those having low children and those without
No. of families with at least one low
No. of Total no. -
children, of families, Observed Expected Variance
C NC MC Ncu2Dc N,u2DC(I -u'D,)
I 2 I 0354 0291
2 I1 2 3.410 2'353
3 54 I9 22-092 13.054
4 98 5' 47'445 24475
5 16 9 8.641 3'974
6 3 I 1.745 0730
8 1 I 0637 0.23 I
Total 185 84 84324 45.108
x& = 0002

Table 8. Combined x2for the comparisons in Tables 4-7

Test Mating XP D.F.
No. of low children, given mc High x low 0027 I
No. of low children, given M, High x high 2'749 I
Em,,given n, High x low 0109 I
CM,, given N , High x high 0'002 I

Total 2.887 4
xX, = 2.887;0.50 < P < 0.70.
age, with no apparent sign of any disease. New blood samples, taken two years after the original
ones, showed the same extremely low paraoxonase activities.
Neither their parents, nor any other parents in our material, showed such low values; but
among 699 grandparents three had similarly low values (two men and one woman). These three
did not show any clinical deviation either that could be related t o the low paraoxonase activity.

22 H Q E 45
330 AND J. MOHR
H. EIBERG

DISCUSSION

( 1 ) Concerning the first question under consideration, namely whether the present technique
gives a better recognition of genotypes than previous methods (Playfer el al. 1976) (Fig. 2), it
is, firstly, apparent that a sharper bimodality is attained, with a very sparse antimodal region;
secondly, the 0-5% of children of low x low parents who had a high phenotype by the primary
classification ( 5 children out of 891) belonged to the ‘low’ category by a secondary criterion,
namely the test of whether the activity shows a pronounced increase where the pH is raised
(they respond).
Regarding the practical procedure in testing for the purpose of classifying individuals into
the two major categories, high and low, the material should thus first be tested by measuring
the activity a t only one pH, in the region pH 7.5-8.5 in tris buffer; it is then sufficient to test
further only individuals with an intermediate activity, as to an effect of a pH rise (preferably
from pH 7.5 to p H 10).
This first question may then be answered affirmatively, since our family material shows no
real exception from the low phenotype representing homozygosity for a low allele.
(2) Concerning the second question, of whether a two allele hypothesis is sufficient, the
distribution of unrelated individuals could well be compatible with this possibility, since there
is no sign of any bimodality among the highs (Fig. 3).
(3) Concerning how the Danish ‘low ’ gene frequency relates to that of other populations, a
close correspondence was found to the results obtained by others for Caucasian populations,
namely 0726 in our study, 0.7595 for Germany (Geldmacher-v. Mallinckrodt, 1978) and 07034
for England (Playfer et al. 1976). These three European estimates, taken together, deviate
somewhat from each other (x&, = 7.74, 002 < P < 0.05).
(4) The five individuals who were found t o have almost complete absence of paraoxonase
activity in serum did not show any conspicuous disorder, which again underlines our still
complete ignorance concerning the biochemical function of paraoxonase.

This work was supported by the Danish Medical Research Council.

REFERENCES
ALDRIDOE,W. N. (1953). Serum esterases. 1. Two types ofesterase (A and B) hydrolysingp-nitrophenol acetate,
propionate and butyrate, and a method for their determination. Biochem. J. 53, 110-117.
GELDMACHER-V. MALLINCKRODT, M., BAUMGARTNER, W., PETENYI, M., BUROIS,H., LINDORF, H. H. & METZER,
H. (1972).Korrelation zwischen der unterschiedlichen Vergiftbarkeit der Serum - Cholinesterase durch E 600
und der Aktivitiit des E600-spaltenden Enzym-Systems in mensohlichen Seren. 2.Physiol. CJhem.353,217-220.
GELDMACHER-V. MALLINCKRODT,M., LINDORF,H. H., PETENYI, M., FLUGEL, M., FISCHER,T. & HILLER,T.
(1973). Genetisch determinierter Polymorphismus der menschlichen Serum-Paraoxonase (EC 3 . 1 . 1 . 2 ) .
Humangenetik 17, 331-335.
GELDMACHER-V. MALLINCKRODT, K. (1978).Polymorphism of human paraoxonase. Hum. Genet. Suppl. 1,65-68.
GELDMACHER-V. MALLINCKRODT,M., HOMMEL, G. &, DUMBACH, J. (1979). On the genetics of the human
paraoxonase (EC 3 . 1 . 1 .2). Hum. Genet. 50, 313-326.
PLAYFER, J. R., EZE,L. C., BULLEN,M. F. & EVANS, D. A. P. (1976). Genetic polymorphism and interethnic
variability of plasma paraoxonase activity. J. Med. Genet. 13, 337-342.
SMITH,C. A. B. (1956). A test for segregation ratios in family data. Ann. Hum. Genet. 20, 257-265.
ZECH,R. &, ZURCHER,K . (1974). Organophosphate splitting serum enzymes in different mammals. Comp.
Biochem. Physiol. 48 B, 427-433.