In some lines of transgenic mice, the methylation of Mspl sites within or adjacent to the transgene locus is affected by the sex of the parent from which the transgene is inherited. These differences are consistent with a role for DNA methylation in genome imprinting. In a previous report, we noted that in one such line, all offspring of females exhibited hypermethylation of the transgene while only some offspring of males carried a hypometh-ylated transgene. In this report, we provide evidence that this phenomenon is controlled by at least two factors, one of which acts in cis and is dependent on the transgene locus, and one of which acts in trans and is supplied by the maternal genome. We also provide evidence that there are genetic differences between inbred mouse strains in the trans-acting factor.

The methylation state of some transgene loci changes in a predictable manner depending upon their maternal or paternal inheritance (Reik et al. 1987; Sapienza et al. 1987; Swain et al. 1987; Hadchouel et al. 1987). A direct relationship between this type of ‘methylation imprinting’ and imprinting as defined by pronuclear transplantation (McGrath and Solter, 1984; Surani et al. 1984; Solter, 1988) or genetic experiments (Cattanach and Kirk, 1985; Searle and Beechey, 1985; Solter, 1988) has not been established. However, these observations are consistent with a role for differential DNA methylation in the process of genome imprinting.

We have previously demonstrated (Sapienza et al. 1987) that particular Mspl sites within a quail troponin I transgene always became methylated in the somatic tissues of offspring that inherited the locus from a female. However, when the same locus was inherited from a male, some, but not all, offspring exhibited transgene hypomethylation in their somatic tissues (Sapienza et al. 1987).

In this report, we demonstrate that the transgene methylation phenotype observed in the somatic tissues of offspring of transgenic males is affected by both the transgene methylation phenotype exhibited by the male, and the strain background of the non-transgenic female. These data imply that both cis-acting and transacting elements are involved in the establishment of the methylation imprint.

All tissue DNA preparations, restriction endonuclease digestions, gel eletrophoresis, in vitro32P-labelling of hybridization probes, hybridization and autoradiography were performed as described in Sapienza et al. 1987. The derivation of the 379 troponin 1 transgenic line is described in Hallauer et al. 1988. Embryos used for injection of DNA were the result of matings between B6D2F1 mice. C57BL/6J, DBA/2J and B6D2F! mice were obtained from The Jackson Laboratory, Bar Harbor, ME.

A transgenic mouse line, designated line 379, was established by microinjection of a 7-1 kb DNA fragment which contained the entire coding sequence of the quail troponin I gene, as well as 375 bp of pBR322 at the 5’ end and 275 bp of pBR322 at the 3’ end (Fig. 1 and Sapienza et al. 1987). This fragment contains sequences sufficient to direct the expression of the quail gene in fast twitch muscle fibers of the mouse (Hallauer et al. 1988). As previously reported (Sapienza et al. 1987) this transgene locus, consisting of 15–20 copies of the injected sequence, displayed gamete-of-origin dependent methylation changes at some Mspl sites within the transgene array. While the hypermethylated phenotype appeared consistent when the transgene was inherited from a female, inheritance of the transgene from males did not always result in hypomethylation of the transgene (Sapienza et al. 1987).

Fig. 1.

Partial restriction endonuclease cleavage map of the micro-injected DNA fragment containing the quail troponin I gene. The open box shows the position of the BamHI–KpnI fragment used as a hybridization probe in Figs 2–4. E: EcoRI, B: BamHI, M: MspI, S: Sall.

Fig. 1.

Partial restriction endonuclease cleavage map of the micro-injected DNA fragment containing the quail troponin I gene. The open box shows the position of the BamHI–KpnI fragment used as a hybridization probe in Figs 2–4. E: EcoRI, B: BamHI, M: MspI, S: Sall.

Blot hybridization analysis of more than 200 transgenic offspring from crosses between non-transgenic females and males hemizygous for the transgene locus has revealed three apparently distinct transgene methylation phenotypes. Representatives of these pheno-types are shown in Fig. 2. Individuals 64, 427, 429, and 431 represent the ‘low’ transgene methylation pattern, characterized by prominent 0·87 kb and 1-3 kb bands, the absence of a 1·35 kb band, and less prominent and approximately equally intense 2·9 kb and 3·2 kb bands. The ‘intermediate’ phenotype is represented by individuals 433, 434, 435, 437 and 298 and is characterized by a prominent 3·2 kb band, a faint 1·35 kb band and the continued presence of the 0·87 kb band. The ‘high’ phenotype is shown by individuals 438, 439, 442, and 56 and is characterized by a virtual absence of the 0·87 kb band, a prominent 3·2 kb band, and 1·3 kb and 1·35 kb bands which are of equal intensity. Hybridization of the same probe used in Fig. 2 to Aispl-cleaved DNAs results in the appearance of only the 0·87 kb fragment (data not shown). Additional matings and hybridization analyses revealed no evidence of either somatic or germline loss of transgene sequences (data not shown).

Fig. 2.

Blot hybridization analysis of transgene methylation phenotype of individuals resulting from crosses between non-transgenic females and transgenic males. Open symbols are non-transgenic individuals, half-filled symbols are individuals that are hemizygous for the transgene. DNA prepared from tail biopsies was cleaved with Hpall and BiwnHI. All restriction endonuclease digestions were checked for completion of digestion by incubating 0·75 μg of a test plasmid with an aliquot of the sample digest. Digestions were judged to be complete when the cleavage pattern of the test plasmid incubated with the sample was identical to the cleavage pattern of the test plasmid alone.

Fig. 2.

Blot hybridization analysis of transgene methylation phenotype of individuals resulting from crosses between non-transgenic females and transgenic males. Open symbols are non-transgenic individuals, half-filled symbols are individuals that are hemizygous for the transgene. DNA prepared from tail biopsies was cleaved with Hpall and BiwnHI. All restriction endonuclease digestions were checked for completion of digestion by incubating 0·75 μg of a test plasmid with an aliquot of the sample digest. Digestions were judged to be complete when the cleavage pattern of the test plasmid incubated with the sample was identical to the cleavage pattern of the test plasmid alone.

Which of the three phenotypes appears in the off spring of transgenic males mated to non-transgenic females depends on both the methylation phenotype of the transgenic male, and the strain background of the non-transgenic female. This is illustrated by the crosses in Fig. 2.

When mated to DBA/2J females, male 64 (low phenotype) gave rise to offspring that displayed the low somatic transgene methylation pattern, identical to his own. However, when mated to C57BL/6J females, his offspring displayed the intermediate pattern. Because all of these offspring were sired by the same male, these differences most likely arise as a consequence of differences in the genetic background of the non-transgenic female. Similarly, male 56 (high phenotype) gave rise to offspring that displayed one of two transgene methylation phenotypes. When mated to C57BL/6J females, his offspring displayed a high methylation phenotype, identical to his own. However, when mated to DBA2/J females, his offspring displayed the intermediate phenotype (Fig. 2 and data not shown). The difference between these two classes of offspring again appears to reflect differences in the genotype of the non-transgenic female.

To determine whether such differences were genetic, and based upon single or multiple loci, a series of crosses was analyzed. Transgenic males with the high methylation phenotype were mated to B6D2F1 females, and the methylation phenotype of the offspring assayed. Fig. 3 shows some of the results obtained. The apparent strain-specific differences observed in Fig. 2 segregate in the ova of F1 females, giving rise to offspring with either the high phenotype or the intermediate phenotype within the same litter. Of 36 offspring sired by high methylation phenotype males mated to B6D2F1 females, 17 exhibited the intermediate phenotype and 19 showed the high phenotype (Fig. 3 and data not shown). This ratio does not differ significantly from 1:1 and is consistent with an allelic difference at a single genetic locus.

Fig. 3.

Blot hybridization analysis of transgene methylation phenotype of individuals resulting from crosses between B6D2F] females and transgenic males. Lanes under halffilled triangles contain DNA samples extracted from the skin of day 14 embryos. DNAs were cleaved with HpaII and BamHI as in Fig. 2.

Fig. 3.

Blot hybridization analysis of transgene methylation phenotype of individuals resulting from crosses between B6D2F] females and transgenic males. Lanes under halffilled triangles contain DNA samples extracted from the skin of day 14 embryos. DNAs were cleaved with HpaII and BamHI as in Fig. 2.

To determine whether the differences observed between the offspring of high and low methylation phenotype males (male 98 and male 64, respectively) reflected differences in the methylation state of the transgene in the gametes of each male, we analyzed the methylation phenotype of tissues derived from each primary germ layer (Hogan et al. 1986) of each individual, including testes. Fig. 4 shows that within each individual, all somatic tissues analyzed exhibit the same transgene methylation phenotype. Thus the difference in the somatic methylation phenotype is consistent between all somatic tissues of the two males. However, no difference was observed in the methylation phenotype of their testes. Therefore, the transgene methylation phenotype of their gametes does not correlate with the differences observed between their offspring.

Fig. 4.

Blot hybridization analysis of transgene methylation phenotype of tissues of a high phenotype male (male 98, see also Fig. 3) and low phenotype male (male 64, see also Fig. 2). DNAs were cleaved with HpaII and BamHI as in Fig. 2.

Fig. 4.

Blot hybridization analysis of transgene methylation phenotype of tissues of a high phenotype male (male 98, see also Fig. 3) and low phenotype male (male 64, see also Fig. 2). DNAs were cleaved with HpaII and BamHI as in Fig. 2.

Analysis of the transgene methylation phenotype of more than 200 offspring sired by transgenic males has revealed that the methylation phenotype of offspring is controlled by both an epigenetic factor and a genetic factor. The epigenetic factor reflects the methylation phenotype of the sire; those sires with the low somatic phenotype giving rise to offspring with either the low or intermediate phenotype; and those sires with the high somatic phenotype giving rise to offspring with either the high or the intermediate phenotype. Which of the phenotypes appears among their progeny is dependent on the inbred strain background of the non-transgenic female to which the male was mated.

These data imply that the process of transgene methylation imprinting involves both cw-acting elements, which reflect the methylation state of the transgene in the previous generation, and trans-acting factors, which are supplied by the maternal genome after fertilization. Our data also demonstrate that there are genetic differences among inbred strains in the trans-acting factor. The trans-acting factor that gives rise to these phenotypes is apparently controlled by a single genetic locus because the segregation ratio of phenotypes produced by F1 females is 1:1.

In order to account for the segregation of these phenotypes from diploid F1, ova, the expression of this locus must either be subject to allelic exclusion (if expression begins prior to meiosis in the oocyte) or expression must not begin until the completion of meiosis, i.e. after formation of the second polar body, at fertilization. The existence of a trans-acting factor that does not operate until after fertilization is consistent with data obtained on the somatic methylation pattern of endogenous loci, which indicate that postfertilization changes take place (Sanford et al. 1987).

The cis-acting element that operates within this system must logically be due to either the transgene itself or its integration site. One must presume that integration site has some effect because different lines that carry the same transgene may behave differently, with respect to whether a gamete of origin methylation effect is observed (Reik et al. 1987; Sapienza et al. 1987), but these data do not eliminate the possibility that sequences endogenous to the transgene also have an effect. In this regard, it is interesting to note that four of seven lines transgenic for the same quail troponin I sequences displayed gamete-of-origin dependent methylation changes (Sapienza et al. 1987; McGowan et al. 1989 and unpublished) while only one of eight lines transgenic for an immunoglobulinchloramphenicol acetyltransferase construct showed such characteristics (Reik et al. 1987).

Because we are unable to demonstrate transgene methylation differences in the gametes of males that give rise to offspring with different transgene methylation phenotypes, it seems unlikely that the cw-acting signal that gives rise to these differences is methylation of the transgene perse, unless the sites that dictate these differences are not revealed by our assay (Bird, 1986). While this is possible, it does not explain how such differences are established at an allele that is identical by descent within all individuals.

In some respects, these differences are reminiscent of the differences observed in the inheritance of the fragile X-linked mental retardation syndrome in the human, where only some individuals who inherit the fragile X chromosome express the mental retardation phenotype (Camerino et al. 1983; Sherman et al. 1985). In a recent model (Laird, 1987), this variability has been proposed to arise from a cis-acting mutation, which leads to a local block in the reactivation of a previously inactivated fragile X. A similar mechanism may explain the variability in transgene methylation phenotype that we observe.

One conclusion that may be drawn from this study is that comparisons between different methylation imprinting studies are difficult, if not impossible, in the absence of genetic analyses. All published reports on gamete-of-origin dependent changes in transgene methylation have been carried out on non-inbred mouse populations (Reik et al. 1987; Sapienza et al. 1987; Swain et al. 1987 ; Hadchouel et al. 1987). While some of these transgene loci seem relatively immune to strain background effects (Swain et al. 1987), others (Hadchouel et al. 1987) apparently show the same strain dependence. Because such differences sometimes affect the expression of transgenes (Swain et al. 1987), this caveat is not restricted to studies on genome imprinting, but may be relevant to any investigation involving the expression of exogenous DNA sequences in transgenic mice.

We are grateful to Catherine Italiano, Mira Puri, Irene Tretjakoff and Susan Gauthier for technical assistance, Susan Caluori and Monique Roger for typing the manuscript and Linda Sapienza and Robert Derval for artwork.

Bird
,
A. P.
(
1986
).
CpG-rich islands and the function of DNA methylation
.
Nature, Lond
.
321
,
209
213
.
Camerino
,
G.
,
M. G.
Mattei
,
J. F.
Mattei
,
M.
Jaye
&
J. L.
Mandel
. (
1983
).
Close linkage of fragile X-mental retardation syndrome to haemophilia B and transmission through a normal male
.
Nature, Lond
.
306
,
701
704
.
Cattanach
,
B. M.
&
M.
Kirk
. (
1985
).
Differential activity of maternally and paternally derived chromosome regions in mice
.
Nature, Lond
.
315
,
496
498
.
Hadchouel
,
M.
,
H.
Farza
,
D.
Simon
,
P.
Tiollias
&
C.
Pourcel
. (
1987
).
Maternal inhibition of hepatitis B surface antigen gene expression in transgenic mice correlates with de novo methylation
.
Nature, Lond
.
329
,
454
456
.
Hallauer
,
P. L.
,
K. E. M.
Hastings
&
A. C.
Peterson
. (
1988
).
Fast skeletal muscle-specific expression of a quail troponin I gene in transgene mice
.
Molec. cell. Biol
.
8
,
5072
5079
.
Hogan
,
B.
,
F.
Costantini
&
E.
Lacy
. (
1986
).
Manipulating the mouse embryo
.
Cold Spring Harbor Laboratory
.
Laird
,
C. D.
(
1987
).
Proposed mechanism of inheritance and expression of the human fragile-X syndrome of mental retardation
.
Genetics
117
,
587
599
.
McGowan
,
R.
,
T. H.
Tran
,
J.
Paquette
&
C.
Sapienza
. (
1989
).
Transgene methylation imprints are established post-fertilization
.
In Nucleic Acid Methylation, UCLA Symposium on Molecular and Cellular Biology (G. Clawson, D. Willis, A. Weissbach & P. Jones eds
.),
Alan R. Liss, Inc
.,
New York
,
in press
.
McGrath
,
J.
&
D.
Solter
. (
1984
).
Completion of mouse embryogenesis requires both the maternal and paternal genomes
.
Cell
37
,
179
183
.
Reik
,
W.
,
A.
Collick
,
M. L.
Norris
,
S. C.
Barton
&
M. A. H.
Surani
. (
1987
).
Genomic imprinting determines methylation of parental alleles in transgenic mice
.
Nature, Lond
.
328
,
251
254
.
Sanford
,
J. P.
,
H. J.
Clark
,
V. M.
Chapman
&
J.
Rossant
. (
1987
).
Differences in DNA methylation during oogenesis and spermatogenesis and their persistence during early embryogenesis in the mouse
.
Genes Dev
.
1
,
1039
1046
.
Sapienza
,
C.
,
A. C.
Peterson
,
J.
Rossant
&
R.
Balling
. (
1987
).
Degree of methylation of transgenes is dependent on gamete of origin
.
Nature, Lond
.
328
,
251
254
.
Searle
,
A. G.
&
C. V.
Beechey
. (
1985
).
Noncomplementation phenomena and their bearing on nondisjunctional effects
.
In Aneuploidy
. (
V. L.
Dellarco
,
P. E.
Voytek
&
A.
Hollaender
. Eds.) pp.
363
376
.
Plenum Press
.
New York/London
.
Sherman
,
S. P.
,
Jacobs
,
N.
Morton
,
U.
Froster-Iskenius
,
P.
,
Howard-Peebles
,
K.
Nielsen
.
M.
Partington
,
G.
Sutherland
,
G. Turner
&
M.
Watson
. (
1985
).
Further segregation analysis of the fragile X syndrome with special reference to transmitting males
.
Hum. Genet
.
69
,
289
299
.
Solter
,
D.
(
1988
).
Differential imprinting and expression of maternal and paternal genomes
.
A. Rev. Genet.
22
,
127
146
.
Surani
,
M. A. H.
,
S. C.
Barton
&
M. L.
Norris
. (
1984
).
Development of reconstituted mouse eggs suggests imprinting of the genome during gametogenesis
.
Nature, Lond
.
308
,
548
550
.
Swain
,
J. L.
,
T. A.
Stewart
&
P.
Leder
. (
1987
).
Parental legacy determines methylation and expression of an autosomal transgene: a molecular mechanism for parental imprinting
.
Cell
50
,
719
727
.