Multigenic Response to Ethanol in Drosophila melanogaster

Author(s): Douglas R. Cavener and Michael T. Clegg

Source: Evolution, Vol. 35, No. 1 (Jan., 1981), pp. 1-10
Published by: Society for the Study of Evolution
Stable URL: http://www.jstor.org/stable/2407937
Accessed: 24-01-2017 15:27 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to
Evolution

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
 EVOLUTION
INTERNATIONAL JOURNAL OF ORGANIC EVOLUTION

PUBLISHED BY

THE SOCIETY FOR THE STUDY OF EVOLUTION

Vol. 35 January, 1981 No. 1

Evolution, 35(1), 1981, pp. 1-10

MULTIGENIC RESPONSE TO ETHANOL IN

DROSOPHILA MELANOGASTER

DOUGLAS R. CAVENER AND MICHAEL T. CLEGG

Department of Botany and Department of Molecular and Population Genetics,
University of Georgia, Athens, Georgia 30602

Received November 15, 1979. Revised June 20, 1980

The use of enzyme polymorphisms for
evolutionary and population genetic in-
vestigations has greatly facilitated the
study of genetic structure both within and
between species. Despite their utility as
genetic markers, fundamental problems
concerning the adaptive nature of enzyme
variation have remained refractory to in-
vestigation. Numerous indirect statistical
studies give support to the notion that en-
zyme polymorphisms are subject to selec-
tion, but direct evidence for selection is
rare. Notable exceptions are the alcohol
dehydrogenase (ADH) and the amylase
polymorphisms in Drosophila melanogas-
ter which have been intensively investi-
gated under controlled environments and
with randomized genetic backgrounds (for
ADH see Gibson, 1970; Van Delden et al.,
1975; Oakeshott, 1976; Cavener and
Clegg, 1978; and Cavener, 1979; for am-
ylase see Dejong et al., 1972; Hickey,
1977). In the case of both ADH and am-
ylase the specific substrates of the enzymes
were used in high concentrations in ex-
perimental populations to induce selec-
tion. Complimentary to these population
experiments, in vitro kinetic differences
have been demonstrated between the al-
ternative allozymes and these differences
appear to be consistent with the direction
of gene frequency change observed in cage
populations.
It can be argued that the ADH and am-
ylase polymorphisms are exceptional be-
cause they are involved in the processing
of external substrates. In an effort to de-
termine the validity of this view and to
investigate adaptation at the biochemical
level, we have chosen a somewhat differ-
ent approach based upon the elementary
proposition that metabolism is a closely
integrated process. Because of the coor-
dinated nature of biochemical pathways,
a perturbation introduced at one point is
likely to have manifold effects. Thus we
have focused upon the manipulation of
simple environmental factors (e.g., alco-
hol) and we have attempted to trace the
genetic consequences arising from pro-
cessing high levels of the stress factor.
We have chosen to investigate adapta-
tion to high concentrations of ethanol in
Drosophila melanogaster. This is not an
artificial choice because the major food
resource of D. melanogaster is fermenting
fruit which contains moderate to high con-
centrations of ethanol. Indeed, D. mela-
nogaster has actually been found to
undergo a part of its life cycle on floating

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
2 D. R. CAVENER AND M. T. CLEGG
mats on top of open wine vats in Spain
(Briscoe et al., 1975). In addition D. mel-
anogaster larvae are attracted to high con-
centrations of ethanol (Parsons and King,
1977), unlike its sibling species, D. sim-
ulans.
The experiments described in this re-
port involved monitoring ten enzyme
polymorphisms in population cages con-
taining media augmented with ethanol
over a period of 56 generations. Five of
these enzymes (ADH, aGPDH, MDH,
GOT, and 6PGD) may be influenced by
high dietary ethanol (Scholtz et al., 1974;
Cederbaum et al., 1977). The results of
the population cage experiments indicate
that 6PGD, MDH, aGPDH, and ADH
responded to ethanol selection. (The initial
response of the latter two polymorphisms
was reported earlier [Cavener and Clegg,
1978].)
Joint viability estimates for Adh and
aGpdh are also presented together with
expected gene frequency trajectories com-
puted using these estimates. The viability
estimates and computer runs are consis-
tent with the hypothesis that the dynamic
of aGpdh is influenced by an epistatic in-
teraction between aGpdh and Adh.
Data are also presented on a series of
relaxed selection experiments established
from the ethanol selected populations at
generation 29. The relaxed selection pop-
ulations have been maintained an addi-
tional 30 generations on control medium.
Periodic sampling pf the relaxed selection
populations indicates that the loci exhib-
iting gene frequency change under ethanol
stress display a quite different dynamic
when they are taken off ethanol supple-
mented media.
MATERIALS AND METHODS
The two ethanol (denoted El and E2)
and two control populations (denoted Cl
and C2) monitored in this investigation are
those reported previously (Cavener and
Clegg, 1978). These were derived from a
sample of a natural population collected
in Providence, Rhode Island, in 1974.
Equal numbers of full sibs from 11 single
pair matings were used to initiate the four
This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
replicate populations. Thus the initial
gene frequencies between the four popu-
lations were similar. The control repli-
cates have been maintained on a standard
cornmeal-molasses-agar diet and the eth-
anol replicates maintained on the same
diet but augmented with 10% ethanol for
57 generations. All populations have been
maintained on an 18 day discrete genera-
tion schedule.
Immediately after a five day period of
egg production at generation 29, the
adults from El and E2 (the ethanol repli-
cates) were divided in half and trans-
ferred to new population cages with ten
scintillation vial food cups containing con-
trol medium. These populations, denoted
relaxed ethanol selection replicates (RElA
and RElB from El and RE2A and RE2B
from E2), were maintained for 30 gener-
ations under the same 18 day discrete gen-
eration schedule as their "parental" etha-
nol selection populations.
The ethanol selection and control pop-
ulations were periodically monitored over
a 57 generation period for the following
enzymes: glutamate oxaloacetate trans-
aminase, GOT (2-4.8); alpha-glycerophos-
phate dehydrogenase, a-GPDH (2-20.5);
malate dehydrogenase, MDH (2-41.5); al-
cohol dehydrogenase, ADH (2-50.1); es-
terase-6, E6 (3-36.8); esterase-C, E, (3-
49.0); octanol dehydrogenase, ODH (3-
49.2) and superoxide dismutase (formerly
tetrazolium oxidase TO), SOD (3-32.5). In
addition male specific glucose oxidase, GO
(3-48.0), and 6-phosphogluconate dehy-
drogenase, 6PGD (x-0.64), were electro-
phoretically monitored at generations 50
and 57, respectively.
- The relaxed selection replicates were
electrophoretically typed for ADH and
aGPDH for generations 1 through 12, and
generations 15, 22, 27, and 30. In addition
allozyme frequencies at MDH and 6PGD
were determined during generation 57.
The electrophoretic assays on the adult
stage were conducted as previously re-
ported for GOT, MDH, aGPDH, ADH,
SOD, E6, E,, and ODH (Cavener and
Clegg, 1978). GO was assayed on the Pou-
lik buffer system (Poulik, 1957) utilizing a
 ETHANOL RESPONSE IN DROSOPHILA 3

10

9 *-_C1
u-- C2
0-0 E21
.8 - -'

7 -

:6 x K

0
E 15;

5 1 0 1 5 20 25 3 30 35 40 45 50 55 6 0 5 10 15 20 25 30 35 40 45 50 55 60
GENERATION GENERATION

FIG. 1. Adhs-frequencies for two control repli- FIG. 2. cxGpdhs frequencies for two control rep-
cate populations (Cl and C2) and two ethanol selected licate populations (Cl and C2) and two ethanol se-
replicate populations (El and E2). lected replicate populations (El and E2).

The behavior of aGpdh is similar to that

histochemical stain after Cavener (1980). of Adh for the first 18 generations (Fig. 2).
6PGD was assayed on the JRP buffer sys- However, after generation 18 a reversal
tem (Ayala et al., 1974). in the direction of gene frequency change
All samples sizes were relatively large, occurs in the ethanol replicates followed
ranging from a low of 90 genomes to a by a rapid increase in the frequency of the
high of 120 genomes with a mean of 112.4 aGpdhS allele. The aGpdhS dynamic is
genomes per sample. remarkably different in the control repli-
cates, where the aGpdhS allele displays a
mild decline from generation 18 to 50.
RE,SULTS
Mdh gene frequencies also appear to
Ethanol Selection Experiments have diverged between control and etha-
Four of the ten enzyme polymorphisms nol replicates. No differentiation was ob-
monitored exhibit a gene frequency re- served during the first 12 generations
sponse to ethanol selection. These four (Cavener and Clegg, 1978) and MDH was
polymorphisms, ADH, axGPDH, MDH, not assayed again until generation 57, at
and 6PGD, respond quite differently in which point both ethanol replicates were
both rate of change and in complexity of fixed for the MdhS allele while both con-
the gene frequency trajectory. trol replicates have remained polymorphic
Consider first the Adh locus which dis- (Table 1).
plays a simple and replicable change over Allele frequencies for 6Pgd were not
time. The Adh locus exhibits a series of determined until generation 57, conse-
rapid increases in the fast allele followed quently the pattern of gene frequency be-
by periods of temporary stability in the havior for this polymorphism is complete-
ethanol environment (Fig. 1). One repli- ly unknown. Nevertheless, data obtained
cate has become fixed for the Adkh' allele from generation 57 show considerable dif-
while the other replicate has remained ferentiation between the ethanol and con-
polymorphic but close to fixation. In con- trol replicates (Table 1). The ethanol rep-
trast, the Adhs allele has remained in high licates are both close to fixation for 6PgdF,
frequency throughout the 50 generations while the control replicates are character-
spanned by the data from the control pop- ized by intermediate frequencies. Consid-
ulations. The gene frequency response at ering both 6Pgd and Mdh, the ethanol
the Adh locus is very different in control replicates do not differ from each other
and ethanol environments (Fig. 1). but are significantly different from the

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
4 D. R. CAVENER AND M. T. CLEGG

TABLE 1. 6Pgds and MdhS gene frequencies for the

control and ethanol selection populations (generation
8 - RE1A
57) and for the relaxed selection populations (gen- A--A RElB
7 *-_ RE2A
eration 30). 1 --RE2B
6 E0 E
6 Pgds SE Mdhs SE
wZ
Control 1 .263 ? .057 .767 ? .039
w 4-
2 .662 ? .059 .933 ? .023

Ethanol 1 .022 ? .015 1.00

2 .067 ? .024 1.00 32

Relaxed 1 A .322 ? .049 1.00

1 B .033 ? .019 1.00
2 A .033 ? .019 1.00 5 1 0 15 2 0 25 3 0 35 40 4 5 50 5 5 60
(1) (6) (11) (16) (21) (26) (31)
2 B .289 ? .048 1.00 GENERATION

FIG. 3. Adhs frequencies for two ethanol selected

controls. However, the control replicates
are also significantly different from each
other.
The gene frequencies for Got, Sod, E6
Ec) Odh, and Go determined at genera-
tion 50 are given in Table 2. Considerable
between replicate heterogeneity exists
among these six loci, but no replicable dif-
ferences occur between the ethanol and
control populations. In general the non-
selected loci in the ethanol populations
and all the loci in the control populations
show more between replicate heterogene-
ity than the four selected loci in the eth-
anol populations.
Relaxed Selection Experiments
The relaxed selection lines show a gene
frequency behavior different from the
contemporaneous ethanol selected lines.
Figure 3 displays the relaxed selection
data for the Adh locus superimposed on
the data from the ethanol selected lines.
The gene frequency behavior in the four
relaxed selection populations is erratic and
poorly replicated, but despite this varia-
replicate populations (El and E2) and four relaxed
ethanol selected populations (RE1A, RE1B, RE2A,
and RE2B). The RElA and RElB replicates were
derived from the E, replicate. The RE2A and RE2B
replicates were derived from the E2 replicate.
tion, the frequency of the AdhS allele is on
average higher than in the ethanol envi-
ronment.
The aGpdh gene frequency behavior,
shown in Figure 4, is different. The pair
of replicates derived from the E1 (ethanol)
population show some between replicate
variation, but there is still substantial
agreement among replicates in gene fre-
quency behavior. The among replicate
agreement is especially strong for the re-
laxed selection replicates derived from the
E2 population. What is particularly strik-
ing is the very different gene frequency
response of the ethanol selected popula-
tions in comparison with their relaxed se-
lection derivatives. Clearly the ethanol en-
vironment is causing a response at the
aGpdh locus and equally clearly the gene
frequency response stops when ethanol is
removed from the environment.

TABLE 2. Gots, Sods, E6S, Ecs, Gos, and Odhsgene frequencies for the control and ethanol replicate
populations, generation 50.

Gots Sods E,s Ecs Gos Odhs

Control 1 .650 + .043 .225 + .038 .617 + .044 .833 ? .034 1.000 .217 + .038
2 .958 + .018 .083 + .025 .350 ? .043 .867 ? .031 .957 ? .024 .000

Ethanol 1 .672 ? .041 .200 ? .035 .612 ? .043 .821 ? .037 .821 ? .043 .133 ? 0.030
2 .625 + .043 .000 .484 + .044 .789 ? .036 .988 ? .011 .430 ? 0.044

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
 ETHANOL RESPONSE IN DROSOPHILA 5

Gene frequencies for 6Pgd and Mdh I 0

were not determined until generation 30. 9 - RElA

h-ARElB
However, MdhS is fixed in all four relaxed U- RE2A
1 RE2B A
selection replicates (Table 1) as well as in 5 10 10El
both ethanol replicates, hence it is prob-
able that Mdhs became fixed before the
relaxed selection lines were split off from W 5
a:

their "parental" ethanol selection popula- 4

tions. The pattern of response to relaxed

selection at 6Pgd is ambiguous. Replicates
2
RElB and RE2A are not significantly dif-
ferent from their parental ethanol selec- 5 10 15 2-0 25 30 35 40 45 50 55 60
(1) (6) (11) (16) (21) (26) (31)
tion replicates, whereas RElA and RE2B GENERATION

are significantly different. FIG. 4. aGpdhs frequencies for two ethanol se-
lected replicate populations (El and E2) and four re-
Adh-aGpdh Joint Viability Estimates laxed ethanol selected populations (RE1A, RE1B,
During the first 12 generations of the RE2A, and RE2B). The RElA and RElB replicates
were derived from the El replicate. The RE2A and
control and ethanol populations gametic
RE2B replicates were derived from the E2 replicate.
frequencies were estimated by progeny
testing. Approximately 60 males were
sampled from the ethanol and control
populations in each generation. Because puted for every sample to test the hy-
pothesis that the adult numbers corre-
of the labor and time involved in progeny
sponded to the numbers expected under
testing individual males, the El and Cl
populations were sampled on odd num- random mating and no selection (i.e., Ho:
bered generations and the E2 and C2 pop- nu' = NP'). Considering the 12 gener-
ulations were sampled on even numbered ations of data for the control experiment,
generations. Individual males were then only one generation (generation 4, C2
test mated to virgin females homozygous population) was significant (P < .005).
for Adh and aGpdh. The rate of recom- The total fit to all 12 generations of data
bination in male D. melanogaster is give a test statistic falling at the 10%
known to be very small; consequently probability level. The null hypothesis can
both gametic types received by the tested be rejected, at or below the 5% level,
male, from his parents, can be determined in three out of 12 generations for the
by the assay of the male and a single test ethanol populations (generations 2, 6 and
cross progeny. 11 for E2, E2 and El, respectively). The
Denote the relative gametic frequency total fit to all 12 generations of data also
estimates for the FF, FS, SF and SS types leads to rejection of the null hypothesis
by g, g12, g21 and g22, respectively. As- (.01 > P > .005). Thus the ethanol ex-
suming random mating the expected two- periments have adult genotypic frequency
locus genotypic frequencies at the point of distributions which frequently depart
union of gametes are, from random mating. This departure is
most likely a result of viability selection,
P j7 = gij i, j = 1, 2 because the random union of gametes
2 Pfj = 2AiAk1 i, j, k, 1 = 1, 2 assumption is usually consistent with
genotypic frequency data obtained closely
and i # k or j # 1.following the mating cycle in D. melano-
Observed genotypic numbers in the adult gaster. In this regard, it is noteworthy
stage are denoted 2 nkjfor heterozygotes that the evidence for a perturbation in
and nU for homozygotes, where N= the adult two-locus genotypic frequency
distribution is much stronger in the etha-
X nj. A likelihood ratio test was com-
i,j,k,1 nol populations.

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
6 D. R. CAVENER AND M. T. CLEGG

The indication of a viability difference

among genotypes prompted us to calculate
approximte two-locus viability selection
estimates. Viability selection estimates can
+1+1 +1
be calculated as,

-kjl = i, j, k, I-1, 2
kl NPiki
00C*

To standardize the estimates to a particu-

0-1-
lar genotype, we have pooled the two
rt~~~~o c t tl t double heterozygous types to obtain rela-
tive estimates given by,
C -o +1 +1+

0~~~~) ~ ~ S0 - rk - nkl(P22 + P2')

( (n2 + n 12)PA

These estimates are approximate because

C~~~~~~C
the gametic frequency estimates come
e ~ Q
from the adult population following via-
C~~~~~~~/
bility selection. Viability selection alters
the joint genotypic frequency distribution
and, if the system is not at an equilibrium,
gametic frequencies will also change.
Thus the use of gametic frequency esti-
mates obtained from the adult sample un-
C..) C ) derstates the strength of selection to the
extent that gametic frequencies have also
been altered by the selective process.
O~ ~~~C 00 Vso Table 3 gives the viability selection es-
timates averaged over generations within
each experiment (i.e., within control and
within ethanol experiments). The data
suggest a different pattern of viabilities
00 IC
between control and ethanol environ-
ments. For example, in the ethanol envi-
*0

ronment the Adhss-aGpdhss genotype has

00 VO a low viability in contrast to the control
14~~~~~~~- -114 data. The fate of the aGpdhs allele may
be different depending on whether Adh is
fixed at SS or FF in ethanol containing
0~~~~~0 environments.
g~+1+1 +1 A striking feature of the aGpdh gene
0% co%
frequency behavior is the nonrnonotone
dynamic observed in the alcohol environ-
Of) 0000
ment. This contrasts sharply with the con-
0')~~~~~~C/ f
trol populations, which exhibit little
change in the aGpdhs frequency. To see
whether there is any qualitative similarity
between the dynamic observed and the
one predicted by the viability estimates,
the general two-locus recursion system
was iterated upon using the control and

This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
 ETHANOL RESPONSE IN DROSOPHILA 7
ethanol viability estimates, and estimates
of the initial gametic frequency distribu-
tion. Figure 5 displays the gene frequency
trajectory, for Adh and aGpdh, which
would obtain if the estimated viabilities
completely accounted for the selective
process. There is a general qualitative
similarity between the observed and pre-
dicted behaviors for both control and eth-
anol environments. aGpdh does exhibit a
nonmonotone dynamic, although the re-
versal in the trajectory occurs earlier and
is less dramatic than in the data from the
ethanol experiments. The Adhs frequency
declines in the iterations using the viabil-
ities determined in the ethanol environ-
ments, but appears to stabilize at a fre-
quency of 0.3; whereas, it may be going
to fixation in the experimental data. A
point to emphasize is that the viability es-
timates were obtained from the first 12
generations, yet they predict the increase
in frequency of aGpdhs which does not
begin until after generation 18 in the ex-
perimental data.
The viability estimates must be inter-
preted with caution because it remains a
theoretical possibility that the actual tar-
gets of selection are loci other than the
ones under observation. Moreover, other
components of selection, such as fertility
and mating interactions, are not taken
into account. Still the viability estimates
do suggest an interaction between Adh
and aGpdh, or between the blocks of
genes adjacent to these loci, despite the
fact that linkage disequilibrium among
this pair of loci is small and inconsistent
among generations (Cavener and Clegg,
1978).
DISCUSSION
The rapid increase in the AdhF allele is
consistent with the fact that the AdhFF
genotypes in these populations have twice
the in vitro ADH activity as the Adhss
genotypes (Cavener, 1979). Thus high eth-
anol stress selects for high ADH activity.
This interpretation is consistent with the
results of several other investigations of
the ADH polymorphism under ethanol
stress (Gibson, 1970; Morgan, 1974; Van
This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
o0 a Gpdh'-E
9
8 < == = = = Adhs-c
7-
i,j6 w _ - - ~ ~ -- _apd5-
Z 6
w5
4-
Adhs-E
3-
5 10 15 20 25 30 35 40 45 50
GENERATION
FIG. 5. Simulations of Adhs and aGpdhS fre-
quencies based upon the control and ethanol viability
estimates of the joint Adh-aGpdh genotypes.
Delden et al., 1975; Oakeshott, 1976;
Cavener, 1979).
Inasmuch as the relationship between
aGPDH and ethanol metabolism is less
apparent, the possible explanations for the
gene frequency behavior at the aGpdh lo-
cus in the ethanol treated populations
must be evaluated in more detail.
In considering the possible explanations
for the gene frequency behavior observed
at aGpdh we must take the following facts
into account: 1) population sizes were
moderately large (400 or more adults per
generation in each cage); 2) the experi-
ments were initiated from a large sample
and initial linkage disequilibria were small
for second and third chromosome marker
loci regardless of the degree of linkage; 3)
estimates of linkage disequilibria re-
mained small for the first 12 generations
of the experiment; and 4) the response ob-
served at oaGpdh is unlikely to be a result
of hitchhiking with Adh because there is
very little disequilibrium between these
loosely linked loci and because the early
(generations 1-5) changes in Adh frequen-
cy do not produce contemporaneous
changes at aGpdh (Cavener and Clegg,
1978). There remains, of course, the pos-
sibility that hitchhiking between aGpdh
and other closely linked, but unobserved
loci, could account for the aGpdh dynam-
ic. While it is impossible to exclude hitch-
8 D. R. CAVENER AND M. T. CLEGG
hiking we can examine the requirements
of this hypothesis.
Let us suppose that the different aGpdh
genotypes are selectively neutral. Then we
know from two-locus theory for infinite
populations that linkage disequilibrium
(D) between a selected locus and a neutral
locus, linked to the extent r($A0), will go
to zero. Suppose in the construction of the
experimental populations that random
linkage disequilibrium was generated be-
tween aGpdh and a linked selected locus
(locus B). The dynamic of gene frequency
change at aGpdh is then given by,
P~=P+D(W1 w2), (1)
w
where p is the gene frequency at aGpdh,
W= PBWII + (1 - PB)W12, W2 = PBW12 +
(1 PB)W22, W = PBW1 + (1 - PB)W2 and we can easily calculate that for r = 0.02
w1l, w12, w22 are the fitness values of
genotypes BB, Bb, bb, respectively. The
rate of gene frequency change at aGpdh
can never exceed the rate at the selected
locus.
For ease of discussion let us first confine
our attention to the portion of the dynamic
where aGpdhs increases in frequency
(generations 28 through 50) and assume
genic selection so that w1 - = s. We
can roughly estimate the apparent value
of s assuming continuous generations
from the generation 28 and generation 50
gene frequencies in the ethanol replicates.
The mean value of s over both replicates
is 0.11. Consequently, the value of s at
the hypothetically selected locus must be
greater than 0.11 and probably much
greater.
Now the dynamic at aGpdh also de-
pends on D. The recursion for D under
the model of associative genic selection is
approximately
D' = D(1 - r)[1 + s(2 - PB)]- (2)
D may initially increase (Thomson, 1977),
however, for PB > ?2, D will decay to
zero at a rate greater than the neutral
rate. Asmussen and Clegg (unpubl.) pro-
vide a complete analysis of the dynamic
of D under a wide variety of selection
models. They show that if selection is
This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
recessive, acceleration of decay will occur
for PB >1 - (2 ) < . Experi-
mental investigations show that decay
rates in D. melanogaster may be as much
as 1.8 times the neutral rate (Clegg et al.,
1980). Judging from the range of gene
frequencies at aGpdh, the hypothetically
selected locus must increase in gene fre-
quency beyond PB = 12. Thus a rapid de-
cay of D should occur.
Using these facts we can put some
rough bounds on the value of r between
aGpdh and the hypothetical locus under
selection. The fact that the experiments
span a period of 50 generations is very
useful in this context. Assuming a decay
rate equal to the neutral rate and t = 50,
more than 60% of the initial disequilibri-
um will have decayed. If we assume an
acceleration of 1.5 times the neutral rate
then nearly 80% of the initial disequilib-
rium will have been lost in 50 generations.
Thus we must assume the existence of a
polymorphic locus under strong selection
in ethanol environments within 3 or 4 map
units of aGpdh.
Even this assumption is not sufficient
to account for the observed dynamic at
aGpdh. Selection at a single locus will not
generate a nonmonotone trajectory like
that observed at aGpdh. Thus we must
assume the existence of yet a second linked
locus in initial disequilibrium with aGpdh
and under strong selection in ethanol con-
taining environments. It may very well be
that there are two polymorphic loci tightly
linked to aGpdh which are under strong
selection in ethanol environments, how-
ever, it seems equally parsimonious to
consider the possibility that aGpdh may
itself be influenced by ethanol.
Although little is known about the de-
tails of ethanol metabolism in Drosophila,
the relationship between aGPDH and eth-
anol metabolism in mammals is well es-
tablished. Ethanol oxidation in mamma-
lian species causes a sharp increase in the
NADH/NAD ratio as a result of the oxi-
dation of ethanol via ADH (Smith and
 ETHANOL RESPONSE IN DROSOPHILA 9
Newman, 1959; Cherrick and Leevy,
1965; McCaffrey et al., 1974; Williamson
et al., 1974). Further the rate limiting step
in ethanol oxidation is the reoxidation of
NADH via the a-glycerophosphate and
the malate-aspartate shuttle systems and
the mitochondrial respiratory chain (Vi-
dela and Israel, 1970; Lindros et al., 1972;
Lundquist et al., 1974; Williamson et al.,
1974; Meijer et al., 1975; Cederbaum et
al., 1977). The enzymatic components of
these shuttle systems are aGPDH, aGPO
(a-glycerophosphate oxidase), MDH, and
GOT.
It is known that the a-glycerophosphate
shuttle system is operative in the flight
muscles of adult D. melanogaster (Obrien
and MacIntyre, 1972). However, the
NADH reoxidation shuttle systems func-
tioning in Drosophila larvae are not
known. Consequently, the hypothesis that
aGpdh is responding to ethanol selection
as a result of its role in ethanol metabolism
must await further biochemical investi-
gations of aGPDH function at both the
larval and adult stages under ethanol
stress. The possible interaction in viability
between the Adh and aGpdh loci certainly
provide motivation for testing this bio-
chemical hypothesis.
Because of the role of MDH in the ma-
late-aspartate shuttle system and the
repression of 6PGD function under etha-
nol stress (Hoek and Ernster, 1974; Scholtz
et al., 1974) similar biochemical hypoth-
eses can be constructed for the apparent
ethanol selection on the Mdh and 6Pgd
loci.
SUMMARY
The results of a series of long term ex-
periments on the genetic response of Dro-
sophila melanogaster to ethanol environ-
ments are presented. The experiments
include pairs of replicate cage populations
raised for 57 generations on medium sup-
plemented with 10% ethanol and pairs of
control populations raised on a normal
medium. Relaxed selection experiments
are also presented, where populations
raised for 29 generations on ethanol me-
dium were reared an additional 30 gen-
This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
erations on control medium. Periodically,
during the course of these experiments,
samples of flies from each population were
assayed for ten enzyme polymorphisms
(GOT, aGPDH, MDH, ADH, E6, Ec,
ODH, SOD, GO and 6PGD).
Four out of the ten enzyme polymor-
phisms responded to ethanol selection.
The response is particularly strong for
ADH and aGPDH, and clearly differs be-
tween control populations and relaxed se-
lection populations. Two-locus joint via-
bilities estimated from the Adh-aGpdh
data differ markedly between control and
ethanol environments and indicate a fit-
ness interaction between these two loci or
between blocks of genes marked by these
loci.
ACKNOWLEDGMENT
This research has been supported by
National Science Foundation grants DEB-
76-82479 and DEB-79-00822.
LITERATURE CITED
AYALA, F. J., J. R. POWELL, AND M. L. TRACEY.
1972. Enzyme variability in the Drosophilia wil-
listoni group. IV. Genic variation in natural pop-
ulations of Drosophila willistoni. Genetics 70:
113-139.
BRISCOE, D. A., A. J. ROBERTSON, AND J. M. MAL-
PICA. 1975. Dominance at the ADH locus in re-
sponse of adult Drosophila melanogaster to en-
vironmental alcohol. Nature 225:148.
CAVENER, D. R. 1979. Behavioral preference for
ethanol in Drosophila melanogaster associated
with the alcohol dehydrogenase polymorphism.
Behav. Genet. 9:359-365.
. 1980. The genetics of male specific glucose
oxidase and the identification of other unusual
hexose enzymes in Drosophila melanogaster. Bio-
chem. Genet. In press.
CAVENER, D. R., AND M. T. CLEGG. 1978. Dy-
namics of correlated genetic systems. IV. Multi-
locus effects of ethanol stress environments. Ge-
netics 90:629-644.
CEDERBAUM, A. E., E. DICKER, AND E. RUBIN.
1977. Transfer and reoxidation of reducing
equivalents as the rate-limiting steps in the oxi-
dation of ethanol by liver cells isolated from fed
and fasted rats. Arch. Biochem. Biophys.
183:638-646.
CHERRICK, G. R., AND C. M. LEEVY. 1965. The
effect of ethanol metabolism on levels of oxidized
and reduced nicotinamide-adenine dinucleotide
in liver, kidney, and heart. Biochem. Biophys.
Acta 107:29.
CLEGG, M. T., J. F. KIDWELL, AND C. R. HORCH.
10 D. R. CAVENER AND M. T. CLEGG
1980. Dynamics of correlated genetic systems. V.
Rates of decay of linkage disequilibria in exper-
imental populations of Drosophila melanogaster.
Genetics 94:217-234.
DEJONG, G., A. J. W. HOORN, E. W. THORIG, AND
W. SHARLOO. 1972. Frequencies of amylase
variants in Drosophila melanogaster. Nature
238:453-454.
GIBSON, J. 1970. Enzyme flexibility in Drosophila
melanogaster. Nature 227:959-960.
HICKEY, D. A. 1977. Selection for amylase allo-
zymes in Drosophila melanogaster. Evolution
31:800-804.
HOEK, J. B., AND L. ERNSTER. 1974. Mitochon-
drial transhydrogenase and the regulation of cy-
tosolic reducing power, p. 351-364. In R. G.
Thurman, T. Yonetani, J. R. Williamson, and
B. Chance (eds.), Alcohol and Aldehyde Metab-
olizing Systems. Academic Press, N.Y.
LINDROS, K. O., R. VIHMA, AND 0. A. FORSANDER.
1972. Utilization and metabolic effects of acet-
aldehyde and ethanol in perfused rat liver. Bio-
chem. J. 126:945-952.
LUNDQUIST, F., S. E. DAMGAARD, AND L. SESTOFT.
1974. Interrelation between the pathways of
fructose and ethanol carbon metabolism in per-
fused pig liver, p. 405-416. In R. G. Thurman,
T. Yonetani, J. R. Williamson, and B. Chance
(eds.), Alcohol and Aldehyde Metabolizing Sys-
tems. Academic Press, N.Y.
MCCAFFREY, T. B., AND R. G. THURMAN. 1974.
Mechanism of the adaptive increase in ethanol
utilization due to chronic prior pretreatment with
alcohol, p. 483-492. In R. G. Thurman, T.
Yonetani, J. R. Williamson, and B. Chance
(eds.), Alcohol and Aldehyde Metabolizing Sys-
tems. Academic Press, N.Y.
MEIJER, A. J., G. M. VAN WOERKOM, J. R. WIL-
LIAMSON, AND J. M. TABER. 1975. Rate-limit-
ing factors in the oxidation of ethanol by isolated
rat liver cells. Biochem. J. 150:205-209.
This content downloaded from 130.91.128.93 on Tue, 24 Jan 2017 15:27:23 UTC
All use subject to http://about.jstor.org/terms
OAKESHOTT, J. G. 1976. Selection at the alcohol
dehydrogenase locus in Drosophila melanogaster
imposed by environmental ethanol. Genet. Res.
26:265-174.
OBRIEN, S. J., AND R. J. MACINTYRE. 1972. The
a glycerophosphate cycle in Drosophila melano-
gaster. II. Genetic aspects. Genetics 71:127.
PARSONS, P. A., AND S. B. KING. 1977. Ethanol:
larval discrimination between two Drosophila
sibling species. Experientia 33:898-899.
POULIK, M. D. 1957. Starch gel electrophoresis in
a discontinuous system of buffers. Nature
180:1477.
SCHOLTZ, R., A. KALSTEIN, U. SCHWABE, AND R.
G. THURMAN. 1974. Inhibition of fatty acid
synthesis from carbohydrates in perfused rat liver
by ethanol, p. 315-318. In R. G. Thurman, T.
Yonetani, J. R. Williamson, and B. Chance
(eds.), Alcohol and Aldehyde Metabolizing Sys-
tems. Academic Press, N.Y.
SMITH, M. E., AND H. W. NEWMAN. 1959. The
rate of ethanol metabolism in fed and fasted an-
imals. J. Biol. Chem. 234:1544.
THOMSON, G. 1977. The effect of a selected locus
on linked neutral loci. Genetics 85:753-788.
VAN DELDEN, W., A. KAMPING, AND M. VAN DIJK.
1975. Selection at the alcohol dehydrogenase lo-
cus in Drosophila melanogaster. Experientia
31:418-419.
VIDELA, L., AND Y. ISRAEL. 1970. Factors that
modify the metabolism of ethanol in rat liver and
adaptive changes produced by its chronic admin-
istration. Biochem. J. 118:275-281.
WILLIAMSON, J. R., K. OHKAWA, AND A. J. MEI-
JER. 1974. Regulation of ethanol oxidation in
isolated rat liver cells, p. 365-382. In R. G.
Thurman, T. Yonetani, J. R. Williamson, and
B. Chance (eds.), Alcohol and Aldehyde Metab-
olizing Systems. Academic Press, N.Y.
Corresponding Editor: G. B. Johnson