J. Embryol. exp. Morph. 78, 127-140 (1983)
Printed in Great Britain © The Company of Biologists Limited
The transition from oocyte-coded to embryo-coded
glucose phosphate isomerase in the early mouse
embryo
By JOHN D. WEST 1 ' 2 AND J. F. GREEN 1
From the Sir William Dunn School of Pathology, Oxford
SUMMARY
The proportions of glucose phosphate isomerase (GPI-1) allozymes produced by early
(Gpi-lsa/Gpi-lsb 9 x Gpi-lsc/Gpi-lsc 6)F\ mouse embryos were analysed by quantitative
cellulose acetate electrophoresis. Technical controls showed that this system is extremely
sensitive, quantitatively reproducible and quite accurate. Genetic controls established that the
Gpi-lsa/Gpi-lsb mothers were homozygous for the Gpi-ltb temporal allele, that produces
relatively high GPI-1 activity in the oocyte. The oocyte-coded enzyme lasted until about 5£
days post coitum ip.c.) or shortly thereafter. The maternally derived, embryonic Gpi-ls allele
was expressed no earlier than the paternally derived allele. This was first expressed between
2i and 3£ days p.c. In this cross, most of the transition from oocyte-coded to embryo-coded
GPI-1 occurred between 2\ and 5£ days p.c.
INTRODUCTION
Chapman, Whitten & Ruddle (1971) showed that the paternally derived allele
of Gpi-ls, which codes for the dimeric enzyme glucose phosphate isomerase,
GPI-1 (E.C. 5.3.1.9), was expressed in mouse embryos by the late blastocyst
stage. By using groups of 500 embryos, Brinster (1973) was able to show that the
paternally derived allele was expressed as early as the 8-cell stage at 2\ d&y&post
coitum (p.c). These important observations provided some of the first evidence
for expression of the embryonic genome in preimplantation mouse embryos.
In both experiments embryos were produced by crossing homozygous
Gpi-ls*/Gpi-lsa females with homozygous Gpi-lsb/Gpi-lsb males. The onset of
paternal gene expression was demonstrated by the observation of both the
GPI-IB homodimer and the GPI-1AB heterodimer. This indicates synthesis of
both paternally coded GPI-IB and maternally coded GPI-IA monomers at this
time (assuming that GPI-IA monomers are not recycled from GPI-IA
homodimers). Prior to this, the embryo produced only the GPI-IA allozyme. It
1
Authors' address: Sir William Dunn School of Pathology, South Parks Road, Oxford
0X1 3RE, U.K.
2
Author's present address: M.R.C. Radiobiology Unit, Harwell, Didcot, Oxon 0X11ORD,
U.K.
128
J. D. WEST AND J. F. GREEN
is likely that most, if not all, of this is synthesized in the diploid oocyte but some
may be produced in the early embryo, either from oocyte messenger RNA or by
the early transcription of the maternally derived, embryonic Gpi-lsa allele.
The discovery of a third allele of Gpi-ls by Padua, Bulfield & Peters (1978)
provides an opportunity to examine the contributions of the oocyte-coded
enzyme and both the maternally and paternally inherited forms of the embryocoded enzyme. In this paper we report the results of an electrophoretic analysis
of GPI-1 in embryos produced by crossing heterozygous Gpi-lsa/Gpi-lsb
females to homozygous Gpi-lsc/Gpi-lsc males. Qualitative observations on the
time of disappearance of the GPI-1 AB allozyme and the time of appearance of
the GPI-1C allozyme respectively allow us to monitor the disappearance of the
oocyte-coded enzyme and the onset of embryonic, paternally derived gene expression. In addition, the quantitative increase in the ratio of GPI-1 A: GPI-1AB
allozymes indicates the onset of embryonic, maternally derived gene expression
in Gpi-ls&/Gpi-lsc embryos and allows us to evaluate whether the maternal
allele is expressed before the paternal allele.
MATERIALS AND METHODS
(i) Mice
Two new partially congenic strains were produced by backcrossing mice,
carrying different Gpi-ls alleles to the C57BL/Ola inbred strain for eight
generations. (C57BL/Ola mice are homozygous for Gpi-lsb.) Mice carrying the
Gpi-lsc allele were kindly supplied by Dr G. Bulfield and used to produce the
C57BL/Ola-Gp*-is c /Ws stock (abbreviated to B-Gpi-lf). The C57BL/
Ola.l29-Gp/-is a /Ws strain (abbreviated to B-Gpi-lsa) was derived from a 129
stock (129/Sv-CP/Pas-FukiVe) kindly provided by Dr V. E. Papaioannou.
These two stocks and the inbred strains C57BL/Ola, 129/Sv-SiJ-CP
(abbreviated to B and 129 respectively) and DB A/2Ola were maintained at the
Sir William Dunn School of Pathology. (The C57BL/Ola stock is also known as
C57BL/6Ola. We use the former name because it now seems likely that the
stock is derived from the British C57BL line rather than the American C57BL/6
line.)
(ii) Sample preparation
Unfertilized eggs and some preimplantation embryos were produced by induced ovulation following injections of 5i.u. pregnant mares' serum gonadotrophin, PMS (Folligon, Intervet), followed 48 h later by 5 i.u. human chorionic
gonadotrophin, HCG (Chorulon, Intervet). These injections were given at
approximately 11a.m. The time of mating is taken to be 11 p. m., when ovulation
is expected to begin. Most embryos were produced by natural matings that were
timed from 1a.m., which is the middle of the dark period (7 p.m. to 7 a.m.)
preceding the detection of the vaginal plug.
GPI-1 expression in mouse embryos
129
Unfertilized eggs and embryos were collected at approximately 10 a,m.
Cumulus cells were removed from the eggs in a solution containing 100 units of
hyaluronidase (Sigma) per ml of phosphate-buffered saline, (PBS). Preimplantation embryos were flushed from the reproductive tract and postimplantation embryos (5£ days p.c.) were removed from decidual swellings
under a dissecting microscope. Eggs and embryos were washed twice in PBS or
0-9 % saline and stored at - 2 0 °C in less than 1 /A of PBS or saline, held between
two small volumes of paraffin oil, in finely drawn-out Pasteur pipettes as recommended by Dr M. Buehr (personal communication). All samples were frozen
and thawed three times in liquid nitrogen vapour before analysis.
Adult livers were homogenized in distilled water (3 mis per gram of tissue)
using a hand-held glass homogenizer. After low-speed centrifugation the supernatant was stored at - 2 0 °C and used undiluted for starch gel electrophoresis or
diluted 1/80 with distilled water for cellulose acetate electrophoresis. Supernatants from mice homozygous for different Gpi-ls alleles were mixed in various
proportions and the mixtures coded and stored at — 20 °C prior to
electrophoresis. These mixtures were used for various technical controls.
(iii) Electrophoresis and densitometry
Starch gels were prepared in rectangular (100x80x2 mm) moulds from 12 %
electrostarch and run with a pH6-4 Tris-citrate buffer, essentially as described
by Chapman, Whitten & Ruddle (1971). Bands of GPI were detected by staining
an overlying nitrocellulose membrane, applied to the blotted surface of the
inverted gel, as described by Peterson, Friar & Wong (1978). A 90 x 30 x 5 mm
staining well was formed by sealing a rectangular perspex frame to the
100 x 40 mm overlay (cut from a 200 x 160 mm sheet of Sartorius SM11306) with
petroleum jelly (Vaseline). Approximately 6ml of aqueous stain, prepared as
described by Peterson et al. 1978 but with a different grade of glucose-6phosphate dehydrogenase (Sigma Type XI) were used per well. Normally,
twelve samples were stained per well.
Cellulose acetate electrophoresis was done on 76 x 60 mm Helena, Titan III
electrophoresis plates, using the 'pH 8-5' Tris-glycine buffer (3-0 g Tris base plus
14-4 g glycine per litre) described by Eicher & Washburn (1978). (In our hands,
this recipe produced a pH 8-1 buffer which we used without any pH adjustment.)
After electrophoresis (1 h at 200 V) the plates were stained for 3 to 30 min in
62 x 78 mm staining dishes, using the same aqueous staining mixture that was
used for starch gels. To avoid problems of overstaining, samples with widely
differing GPI-1 activities were not run on the same plate. Both the cellulose
acetate plates and nitrocellulose membranes were washed, fixed and dried after
staining.
The proportions of the different GPI-1 allozymes were quantified by scanning
densitometry using an absorbance maximum of 550 nm. A Vitatron MPS, fitted
with a densitometer unit, analogue integrator and chart recorder was used for the
130
J. D. WEST AND J. F. GREEN
earlier experiments (control mixtures of liver homogenates run on starch gels)
and a Gelman DCD-16 computing densitometer with automatic integration was
used for the later experiments (all of the cellulose acetate plates). For densitometry with the Vitatron MPS system the nitrocellulose overlays were soaked
in glacial acetic acid, methanol and water (1:1:3 by volume) and sandwiched
between two glass plates. For densitometry with the Gelman DCD-16 both the
nitrocellulose membranes and cellulose acetate plates were scanned dry and
uncleared.
(iv) Calculations
Statistical tests were done on a Hewlett Packard 97 programmable calculator,
programmed, by Mr D. G. Pap worth of the M.R.C. Radiobiology Unit, for
Student's t-test and Smith & Satterthwaite's variation of Welch & Aspin's t-test
for samples with different variances (Satterthwaite, 1946).
RESULTS AND DISCUSSION
(i) Technical controls
The sensitivity of the two electrophoresis systems was tested by comparing the
time taken before GPI-1 staining was visible for samples of eggs. In all such tests
the enzyme staining developed much more quickly and fewer samples failed to
stain on the cellulose acetate plates than on the starch gels with nitrocellulose
overlays.
Both electrophoresis systems produced good estimates of the proportions of
GPI-1 allozymes in coded mixtures of liver homogenates (Fig. 1) although
there is a tendency for the minor component to be overestimated particularly
on the cellulose acetate plates. Little variation occurred when the same samples
were run repeatedly on the same or different gels or plates and even less
variation occurred when the same staining pattern was repeatedly scanned.
There was also excellent agreement between results obtained using the two
densitometers.
For both electrophoresis systems, however, deliberate overstaining produced
an overestimation of the minor component in the mixtures where the two allozymes were present in unequal proportions. This so-called stain saturation effect
occurs in other systems at high enzyme activities or at lower activities when
staining is prolonged (Markert & Masui, 1969) and probably contributes to the
sigmoidal shape of the curves shown in Fig. 1.
These technical controls indicate that the starch-gel method offers slightly
more accurate quantitation, at least for liver homogenates. This advantage is
outweighed, however, by the greater sensitivity of the cellulose acetate plates.
For this reason the cellulose acetate plates were used for the analysis of GPI-1
allozymes in early embryos.
GPI-1 expression in mouse embryos
131
Cellulose acetate electrophoresis
GPI-1A + GPMB mixture
Starch gel electrophoresis
GPI-1A + GPMB mixture
100
80
60
40
20
40
60
80
% Gpi-lsa/Gpi-lsa
100
0
20
40
60
80
100
liver in mixture (by weight)
Fig. 1. Comparison of two electrophoresis systems for the relationship between the
expected percentage of GPI-1A in a mixture of liver homogenates and the percentage estimated by quantitative electrophoresis. Each point represents the mean of
three separate, coded mixtures. The line shows the expected relationship if GPI-1 A
and GPI-1B have equal specific activities.
(ii) Genetic controls
The activity of GPI-1 in the oocyte is regulated by a cis-acting temporal gene,
Gpi-lt which is closely linked to the structural gene (Peterson & Wong, 1978;
McLaren & Buehr, 1981). This gene will, therefore, affect the relative
contributions of oocyte- and embryo-coded enzyme. Oocytes from females that
are heterozygous for both the structural Gpi-ls and temporal Gpi-1 tloci produce
a skewed distribution of GPI-1 allozymes. The proportions of the allozymes
produced by unfertilized eggs from females of various genotypes were compared
with those from eggs from (DBA x B)Fi females, which are known to be
heterozygous, Gpi-lsa Gpi-lf /Gpi-lsb Gpi-lP, in order to determine the Gpi-lt
genotype of the stocks of mice used in our experiments.
The results, shown in groups 1 and 2 in Table 1, indicate that eggs from B-Qpilsa/Gpi-lsb heterozygotes produce allozymes in an approximately 1:2:1 ratio as
expected if the B-Gpi-ls* strain carries the Gpi-lft allele which is characteristic
of the B strain. Certainly B-Gpi-lsa/Gpi-lsb eggs do not show the dramatic
skewing of the (DBA x B)Fi eggs that is produced by unequal proportions of
GPI-1 A and GPI-IB monomers (Peterson & Wong, 1978; group 2 Table 1).
The Gpi-lf allele also affects GPI-1 activity because the GPI-1C allozyme is
thermolabile. Padua et al. (1978) showed that the specific activity of GPI was
lower in a variety of tissues from Gpi-1 sc/Gpi-1 sc mice than in those from other
132
J. D . WEST AND J. F . GREEN
Table 1. Quantitative analysis of GPI-1 allozymes in eggs from heterozygous
females
Strain"
No. of eggs
per samples
No. of
samples
Mean proportion of
allozymes
1
3
10
9
12
7
13
0-23A:0-47AB:0-30B
0-06A:0-32AB:0-62B
0-28A:0-68AC:0-05C
0-25A:0-70AC:0-05C
0-32B:0-68BC:0-00C
B-Gpi-lsa/Gpi-lsh
(DBAxB)Fi
(129 x B-Gpi-l^Fi
(B-Gpi-ls* x B-Gpi-lsc)¥i
B-Gpi-lsb/Gpi-lsc
1.
2.
3.
4.
5.
3
3
1
* Females in groups 1 and 5 were heterozygous backcross progeny produced while making
the partially congenic strains B-Gpi-lsa and B-Gpi-lsc respectively.
GPI-IA + GPI-IC mixture
GPI-IB + GPI-IC mixture
100
100 r
80
80
60
60
40
40
20
20
o
0
20
40
60
80
100
% Gpi-lsc/Gpi-hc liver in mixture (by weight)
Fig. 2. Relationship between the expected percentage of GPI-IC in mixtures of liver
homogenates and the percentage estimated by quantitative cellulose acetate
electrophoresis. Each point represents the mean of three separate coded mixtures.
The straight line shows the expected relationship if the specific activity of GPI-IC is
equal to the other allozyme in the mixture. .The curved line shows the expected
relationship if the specific activity of GPI-IC is only 50 % of the other allozyme in
the mixture.
genotypes. This is illustrated in Fig. 2 which shows the results of quantitative
electrophoresis of mixtures of GPI-IA or GPI-IB and GPI-IC in crude liver
homogenates. The results indicate that under our experimental conditions the
specific activity of GPI-1 in Gpi-lsc/Gpi-lsc is about 50 % of that in Gpi-lsa/Gpilsa livers and less than 50 % of that in Gpi-lsb/Gpi-lsb livers.
GPI-1 expression in mouse embryos
133
Table 2. Quantitative analysis of GPI-1 allozymes in brains and livers from
heterozygous male mice
Genotype
Organ
• 7 b/r • 7 c
upi-is /upi-is
liver
6
brain
6
r
r, • 7 a / ^ • 7 b
Gpi-ls»/Gpi-lsb
liver
brain
No. of samples
6
6
Mean proportion of
allozymes
0-29B:0-60BC:011C
0-30B:0-62BC:0-08C
0-27A:0-47AB:0-26B
0-31A:0-46AB:0-23B
Results of quantitative electrophoresis of brain and liver homogenates from
Gpi-lsb/Gpi-lsc males are shown in Table 2. The GPI-1C allozyme is significantly reduced, as expected, but the other two allozymes survive in a 2:1 ratio which
suggests that they are equally active and stable. Essentially similar results were
obtained for unfertilized eggs (Table 1, groups 3 to 5). In this case the GPI-1 C
allozyme was more completely reduced and the heteropolymer was present in
slight excess of the expected 2:1 ratio in Gpi-ls^/Gpi-ls0 eggs. (It is unclear
whether this small excess is trivial or, for example, represents a novel Gpi-lt
allele associated with Gpi-lsc.)
These controls are relevant, in two respects, to the analysis of (Gpi-ls^/Opib
ls 2 X Gpi-lsc/Gpi-lsc Cf )Fi embryos, discussed in the next section. First, they
Gpi-lh
show that the Gpi-lsa/Gpi-lsb mothers are Gpi-lsa, Gpi-lP/Gpi-lsb,
and so will produce oocytes with relatively high GPI-1 activities. Second, the
controls indicate that the GPI-1 AC and GPI-1BC allozymes are stable and so can
be used in the quantitative analysis. The GPI-1C allozyme, however, is unstable
and cannot easily be used in the quantitative analysis because the amount of
activity lost in embryos cannot be predicted from experiments on other tissues.
(iii) GPI-1 allozymes in (Gpi-ls7Gpi-ls fc $ x Gpi-ls c /Gpi-ls c ($)Fi embryos
a) Qualitative analysis
The results from the qualitative analysis of GPI-1 allozymes in (B-Gpi-ls*/
Gpi-lsb $ x B-Gpi-lsc/Gpi-lsc cf)Fi embryos is shown in Table 3. Both Opilsa/Gpi-lsc and Gpi-lsb/Gpi-lsc embryos are produced so samples containing
embryos of both genotypes may produce GPI-1AC, GPI-1BC and GPI-1C allozymes as well as the three oocyte allozymes (GPI-1 A, GPI-1 AB and GPI-1B).
In practice the GPI-1AC allozyme co-migrates with GPI-1B so only five bands
are distinguishable.
At 2-4 and 2-5 days p. c. (8-cell stage) only three allozyme bands were detected
(Groups 1-4 in Table 3). Assuming that the bands of GPI-1B activity produced
by groups 1-4 do not contain any GPI-1AC allozyme this implies that, at this
stage, all of the detectable GPI-1 activity is oocyte coded.
134
J. D. WEST AND J. F. GREEN
Table 3. Qualitative analysis of GPI-1 allozymes produced by (7?-Gpi-lsa/
sft 9 xfl-Gpi-ls c /Gpi-ls c J)Fi embryos
Age
Presence of GPI-1 allozyme
(days
p.c.)*
No. of
embryos in
sample
No. of
samples
1
2
3
4
5
6
7
8
9
2-4
2-4
2-5
2-5
3-4
4-4
4-4
5-4
5-4
5
9
5
10
6
3
3
1
1
2
1
4
1
3
1
2
5
1
10
5-4
1
1
11
5-4
1
3
Group
AB
B and/
or AC
BC
* The age is the expected interval between mating and embryo collection as explained in the
Materials and Methods sections. The time of mating was not closely controlled so the age given
is imprecise. Although 2-4- and 2-5-day embryos do not differ significantly in age, the distinction is made because 2-5-day embryos were produced by induced ovulation and all of the other
embryos resulted from spontaneous ovulations.
Some samples of older embryos produced five bands of GPI-1 as shown in Fig.
3 and groups 5, 7 and 9 in Table 3. This indicates that both oocyte-coded and
embryo-coded enzymes are present in these samples. Other samples, containing
small numbers of embryos, produced no GPI-IBC heteropolymer presumably
because all of the embryos in the sample were Gpi-ls*/Gpi-lsc (groups 6, 8 and
10 in Table 3). The GPI-1 AB allozyme is produced entirely by the diploid oocyte
genome and was absent from four of the ten 5-4-day embryos. It, therefore,
seems likely that no oocyte-coded enzyme remained in these samples.
The qualitative results indicate that the oocyte-coded enzyme is exhausted by
about 5 | days p.c. and the embryonic genome is expressed by 3-4 days p.c. The
presence of the GPI-IC allozyme in 3-4-day embryos indicates expression of the
paternally derived allele. The formation of the GPI-IBC allozyme suggests that
the maternally derived, embryo-coded Gpi-lsb allele is also expressed in 3-4-day
Gpi-ls°I Gpi-lsc embryos although it is possible that GPI-1B monomer is
produced from oocyte-coded mRNA.
The observation that one group of three 4-4-day embryos has no GPI-IBC
allozyme suggests that no significant oocyte GPI-1 mRNA remained after the
paternally derived Gpi-lsc allele was expressed.
b) Quantitative analysis
The quantitative analysis of the allozyme patterns is shown in Table 4 and a
Age
*&»*
(daysp.c.)
unfertilized eggs
2-4
2-5**
3-4
4-4
4-4
5-4
5-4
5-4
5-4
a/c
a/c
b/c
b/c
b/c or mixed
a/c
N.A.t
N.A.
N.A.
b/c or mixed
Embryo
genotype*
6
3
3
1
1
1
1
5-10
1
5-9
10
3
5
3
1
2
5
1
1
3
* Embryo genotype is deduced from distribution of GPI-1 allozymes.
** Produced by induced ovulation.
fN.A. = not applicable.
1
2
3
4
5
6
7
8
9
10
Group
No. of
embryos No. of
in sample samples
47 ± 0-4
48 ±1-3
49 ±0-7
39 ±0-3
14
13 ±2-0
4 ±0-3
0
6
0
23 ±0-5
22 ±0-8
22 ±1-4
20 ± 0-3
14
1
0
12 ±2-0
18 ±0-3
17
AB
A
30 ±0-6
30 ±1-7
29 ±0-9
31 ±0-7
53
44 ±9-7
59 ±0-6
79
17
21 ±2-2
B and/or AC:
A
0
0
0
8± 0-6
0
21 ±12-1
0
0
61
67 ± 4-3
BC
0
0
0
2 ±0-2
14
10 ±1-6
19 ±0-3
8
13
12 ±2-2
Percentage of GPI-1 allozymes (Mean ± standard error)
-
Table 4. Quantitative analysis of GPI-1 allozymes produced by f5-Gpi-lsfl/Gpi-ls/) 9 x 5-Gpi-lsc/Gpi-lsc d)Fi embryos
TO
TO
o2s:
6'
Co
^J
136
J. D. WEST AND J. F. GREEN
scan from a group of three 4-4-day embryos is illustrated in Fig. 3. The quantitative results provide estimates of the proportion of embryo-coded enzyme at
different stages and gives us some indication of when the maternally derived,
embryonic Gpi-ls allele is first expressed.
In unfertilized eggs (Group 1, Table 4) approximately 50 % of the GPI-activity
is in the GPI-1AB allozyme, so in embryos the proportion of oocyte-coded
enzyme is approximately twice the proportion of enzyme represented by the
GPI-1AB allozyme. This is likely to be overestimated, however, because the
GPI-IC allozyme is thermolabile. The genetic controls discussed in the previous
section do not allow us to predict the extent of GPI-IC loss so in Table 5 two
estimates of the proportion of embryo-coded enzyme are given. One assumes
C
BC
B
AB
A
AC
Fig. 3. The five bands of GPI-1 activity produced by three (Gpi-lsa/Gpi-lsb 9 x
Gpi-lsc/Gpi-lsc S)F\ 4-4-day mouse embryos (below) and the densitometer tracing
(above). Migration is towards the cathode (left) and in this sample the proportions
(from right to left) were estimated as 9-9 % GPI-1A, 11-2 % GPMAB, 34-7 % GPI1B plus GPMAC, 32-7 % GPI-1BC, 11-5 % GPI-IC.
GPI-1 expression in mouse embryos
137
Table 5. Estimated proportion of embryo-coded GPI-1 produced by (5-Gpi-ls a /
Gpi-ls b x5-Gpi-ls c /Gpi-ls c )F 7 embryos
Estimated proportion of embryo-coded GPl-1
Age (daysp.c.)
2-4
2-5
3-4
4-4
4-4
5-4
5-4
Embryo genoty
N.A.f
N.A.
b/c or mixed
a/c
b/c or mixed
a/c
b/c
Qualitative*
Quantitative"
0
0
N.A.
N.A.
N.A.
5=1-00
sSl-00
N.A.
N.A.
0-22-0-25
0-72-0-78
0-74-0-78
0-92-0-94
0-88-0-90
* From qualitative allozyme patterns shown in Table 1.
** Quantitative estimates assume proportion of embryo-coded GPI-1 is ( 1 — T
, J where
AB is the proportion of GPI-lAB allozyme. The first estimate assumes no disproportionate
loss of GPI-1C homodimer and the second estimate assumes that only 33 % GPI-1C activity
remains. The quantitative estimates for 5-4-day embryos excludes those that qualitatively have
no GPI-lAB allozyme.
tN.A. = not applicable.
that no GPI-1 C is preferentially lost and the other assumes that 67 % of this
allozyme is lost.
These calculations indicate that at 3-4 days p. c. 22-25 % of the GPI-1 activity
is embryo coded. Presumably the paternally derived Gpi-lsc allele is active
earlier than this time and may be expressed as early as 2\ daysp.c. This is when
Brinster (1973) detected paternally derived GPI-IA allozyme in pooled samples
of 500 Gpi-lsa/Gpi-lsb embryos. The absence of GPI-1C in our 2i-day embryo
samples may reflect the small sample size even with our more sensitive
technique.
The results shown in Table 5 also indicate that 88-94 % of the GPI-1 is embryo
coded in the six 5-4-day embryos that showed some residual oocyte-coded GPIlAB heteropolymer (groups 8 and 9 in Table 3 and groups 7 and 9 in Table 4).
Since the other four 5-4-day embryos had no oocyte GPI-lAB heteropolymer,
oocyte GPI-1 is normally exhausted at or shortly after 5 | daysp.c. in the embryos
studied. This conclusion is not necessarily valid for all embryos since the oocyte
genome will have a profound effect on both the initial activity and stability of the
oocyte-coded enzyme. For example, Gpi-lf/Gpi-lf
females will produce
oocytes with much less GPI-1 enzyme which is likely to be exhausted before 5\
days p.c. unless oocyte mRNA is translated in early embryos.
The expression of the embryonic, maternally derived Gpi-ls* allele will result in
an elevated ratio of GPI-IA: GPI-lAB allozymes in those samples containing
Gpi-lsaI Gpi-lsc embryos. The results, shown in Table 6, indicate thatfor4-4'day
138
J. D. WEST AND J. F . GREEN
Table 6. Expression of maternally and paternally derived Gpi-ls alleles in
(£-Gpi-lsfl/Gpi-ls* 9 x £-Gpi-lsc/Gpi-lsc 6)Fi embryos
Age
(days p. c.)
Unfertilized eggs
2-4
2-5
3-4
4-4
Presence
No. of
of C
Ratio A/AB allozymes
samples allozyme (Mean ± standard error)
10
3
5
3
3
—
+
+
0-48 ±0-01
0-46 ± 0-02
0-45 ±0-03
0-51 ±0-01
1-01 ±0-11
Significance*
-
N.A
N.A
tn == 1-51;N.S.
ti-07 = 4-68; P<0-05
* Statistical significance of A/AB ratios that are larger than the ratio for unfertilized eggs
are evaluated using a one-tailed Student's t-test or Welch & Aspin's t-test (t' value).
N.A. = not applicable. N.S. = not statistically significant.
embryos the ratio is significantly higher than for unfertilized eggs from Gpi-ls*/
Gpi-lsb females so the embryonic Gpi-ls* allele is expressed. At 3-4 days, when
the expression of the paternally derived Gpi-lsc allele is first detected, the ratio
is higher than for eggs but not significantly so. At 2-4 and 2-5 days the ratio is no
higher than for eggs.
Although earlier expression of the embryonic Gpi-ls* allele may go undetected if the samples contained mainly Gpi-lsb/Gpi-lsc embryos, these results suggest that the maternally derived, embryonic Gpi-ls* allele is expressed no earlier
than the paternal Gpi-lsc allele.
One unexplained peculiarity of the results shown in Table 4 is that the proportion of GPI-1 AC or GPI-IBC heteropolymer is more than twice the proportion of
GPI-1 A or GPI-1B in 5-4-day Gpi-ls*/Gpi-lsc and Gpi-lsb/Gpi-lsc embryos respectively (groups 7-10 in Table 4). This is clearest in groups 8 and 10 which have no
residual oocyte-coded GPI-1 activity. This effect was not seen in adult brain and
liver from Gpi-lsb/Gpi-lsc mice (see Table 2) and is more extreme than the slightly
elevated heteropolymer levels seen in oocytes (groups 3-5 in Table 1). It, therefore, seems unlikely to be caused by a difference in allozyme stability but could
reflect a greater production of GPI-1 C monomers at this stage of development.
The expression of a number of paternally derived genes, in addition to Gpi-ls,
has been demonstrated in the mouse and this subject has been extensively
reviewed (Epstein, 1975; McLaren, 1976; Chapman, West & Adler, 1977; Sherman, 1979; Johnson, 1981; Kidder, 1981; Magnuson & Epstein, 1981). As far as
we know the X-chromosome-linked Pgk-1 gene is the only other system in the
mouse where consideration has been given to the relative timing of the onset of
expression of the maternally and paternally derived genes. Krietsch et al. (1982)
interpret their results to suggest that the maternally derived allele is expressed
before the paternal allele and this issue has also been raised by Papaioannou,
West, Bucher & Linke (1981).
GPI-1 expression in mouse embryos
139
Considerable attention has been paid to the transitions from oocyte-coded to
embryo-coded mRNA and protein (see reviews by Epstein, 1975; Sherman,
1979; Johnson, 1981; Magnuson & Epstein, 1981). Although most oocyte
mRNA is degraded by the 4-cell stage (see for example Johnson, 1981) some
mRNA persists until the early blastocyst stage (Bachvorova & De Leon, 1980).
As yet little is known about specific mRNA or protein gene products. Harper
& Monk (1983) have suggested that oocyte mRNA for hypoxanthine phosphoribosyl transferase (HPRT) is present until the late 2-cell and possibly until the 4to 8-cell stage. Although our experiments with GPI-1 allozymes indicate that
oocyte-coded GPI-1 mRNA is absent by the time the paternally derived Gpi+lsc
allele is expressed they shed no light on whether oocyte GPI-1 mRNA is present
in younger embryos.
The recent experiments of Harper & Monk (1983) are also relevant to the
duration of oocyte-coded HPRT enzyme which, they suggest, is depleted by the
early morula stage (2?-3 days p.c). This is based on extrapolation of a biphasic
curve, for increasing HPRT activity per embryo with time, to derive hypothetical
curves for oocyte-coded and embryo-coded contributions. Our more direct
approach conclusively shows that oocyte-coded GPI-1 enzyme persists
somewhat longer.
CONCLUSIONS
Oocyte-coded and embryo-coded enzyme both contribute to the glucose
phosphate isomerase activity in preimplantation embryos. In our experiments
oocyte-coded enzyme was present throughout preimplantation development but
had disappeared by 5 | daysp.c. in some embryos and was almost exhausted in
other embryos at this stage. We confirmed that embryo-coded GPI-1 is produced
before 3h days p.c. and found no evidence that the maternally derived Gpi-ls
allele is expressed before the paternally derived Gpi-ls allele in our crosses.
During the three-day period between 2\ and 5 | days p.c, there is a transition
from almost entirely oocyte-coded enzyme to 90-100 % embryo-coded enzyme.
It seems likely that there is no oocyte-coded GPI-1 mRNA present during this
transition period.
We thank Mrs L. Ofer and Mrs M. Carey for performing many of the starch gel technical
controls, Drs M. Buehr, R. Holmes, A. C. Peterson and P. Thorogood for advice and technical demonstrations and Drs G. Bulfield and V. E. Papaioannou for supplying mice. We are
also grateful to Professor R. L. Gardner, Dr M. F. Lyon and Mr G. Fisher for reading the
manuscript. This work was supported by an M.R.C. programme grant awarded to Professor
R. L. Gardner.
REFERENCES
R. & DE LEON, V. (1980). Polyadenylated RNA of mouse ova and lo$s of
maternal RNA in early development. Devi Biol. 74, 1-8.
BACHVAROVA,
140
J. D. WEST AND J. F . GREEN
R. L. (1973). Parental glucose phosphate isomerase activity in three-day mouse
embryos. Biochem. Genet. 9, 187-191.
CHAPMAN, V. M., WEST, J. D. & ADLER, D. A. (1977). Genetics of early mammalian embryogenesis. In: Concepts in Mammalian Embryogenesis (ed. M. I. Sherman), pp. 95-135.
Cambridge Mass.: MIT Press.
CHAPMAN, V. M., WHITTEN, W. K. & RUDDLE, F. H. (1971). Expression of paternal glucose
phosphate isomerase-1 (Gpi-1) in preimplantation stages of mouse embryos. Devi Biol. 26,
153-158.
EICHER, E. M. & WASHBURN, L. L. (1978). Assignment of genes to regions of mouse
chromosomes. Proc. natn. Acad. Sci., U.S.A. 75, 946-950.
EPSTEIN, C. J. (1975). Gene expression and macromolecular synthesis during preimplantation
embryonic development. Biol. Reprod. 12, 82-105.
HARPER, M. I. & MONK, M. (1983). Evidence for translation of HPRT enzyme on maternal
mRNA in early mouse embryos. /. Embryol. exp. Morph. 74, 15-28.
JOHNSON, M. H. (1981). The molecular and cellular basis of preimplantation mouse development. Biol. Rev. 56, 463-498.
KIDDER, G. M. (1981). The analysis of mammalian development using allelic isozyme
variants. Devi Genet. 2, 385-405.
BRINSTER,
KRIETSCH, W. K. G., FUNDELE, R., KUNTZ, G. W. K., PEHLAU, M., BURKI, K. & ILLMENSEE,
K. (1982). The expression of X-linked phosphoglycerate kinase in the early mouse embryo.
Differentiation 23, 141-144.
MAGNUSON, T. & EPSTEIN, C. J. (1981). Genetic control of very early mammalian development. Biol. Rev. 56, 369-408.
MARKERT, C. L. & MASUI, Y. (1969). Lactate dehydrogenase isozymes of the Penguin
Pygoscelis adeliae. J. exp. Zool. 172, 121-146.
MCLAREN, A. (1976). Genetics of the early mouse embryo. Ann. Rev. Genet. 10, 361-388.
MCLAREN, A. & BUEHR, M. (1981). GPI expression in female germ cells of the mouse. Genet.
Res. 37, 303-309.
PADUA, R. A., BULFIELD, G. & PETERS, J. (1978). Biochemical genetics of a new glucosephosphate isomerase allele (Gpi-P) from wild mice. Biochem. Genet. 16, 127-143.
PAPAIOANNOU, V. E., WEST, J. D., BUCHER, T. & LINKE, I. M. (1981). Non random X
chromosome expression early in mouse development. Devi Genet. 2, 305-315.
PETERSON, A. C , FRIAR, P. M. & WONG, G. G. (1978). A technique for detection and relative
quantitative analysis of glucosephosphate isomerase isozymes from nanogram tissue
samples. Biochem. Genet. 16, 681-690.
PETERSON, A. C. & WONG, G. G. (1978). Genetic regulation of glucose phosphate isomerase
in mouse oocytes. Nature 276, 267-269.
SATTERTHWAITE, F. E. (1946). An approximate distribution of estimates of variance components. Biometrics 2, 110-114.
SHERMAN, M. I. (1979). Developmental biochemistry of preimplantation mammalian embryos. Ann. Rev. Biochem. 48, 443-470.
(Accepted 30 June 1983)