(Received for publication, September 27, 1996, and in revised form, January 13, 1997)
From the Department of Bioscience, National
Cardiovascular Center Research Institute, 5-7-1, Fujishiro-dai, Suita,
Osaka 565, Japan, the ¶ Institute of Molecular Embryology and
Genetics, Kumamoto University Medical School, 4-24-1 Kuhonji, Kumamoto,
Kumamoto 862, Japan, the
Department of Pathology, Saga Medical
School, 5-1-1 Nabeshima, Saga 549, Japan, and the ** Research Institute
for Microbial Disease, Osaka University, 3-1 Yamada-oka, Suita,
Osaka 565, Japan
Genomic imprinting refers to the parental allele-specific expression of genes. The precise mechanism underlying this phenomenon, which may involve DNA methylation, is not yet known. U2af1-rs1(SP2) is an imprinted gene expressed from the paternal allele and is methylated on the maternal allele. Here we report an artificial system in which expression and methylation of the endogenous imprinted gene U2af1-rs1 can be affected by interaction with its own transgene in the testis. We suggest that there is a mechanism in male gametogenesis by which the U2af1-rs1 gene is kept unmethylated to be expressed in the offspring in addition to a mechanism in female gametogenesis by which the U2af1-rs1 gene is methylated and is not expressed in the offspring.
The epigenetic phenomenon of genomic imprinting determines an inequality of the maternal and paternal genomes. In mammals, genomic imprinting ensures functional inequality of paternal and maternal genomes in the fertilized egg and causes developmental failure of embryos produced by parthenogenesis or by gynogenesis or androgenesis (1-6). The precise mechanism underlying this phenomenon, which may involve DNA methylation, is not yet known. Previously, we cloned an imprinted gene U2af1-rs1(SP2) in a search for parental origin-specific CpG methylation using restriction landmark genomic scanning (7-9). U2af1-rs1 encodes a putative 51-kDa protein with significant similarity to U2 small nuclear ribonucleoprotein auxiliary factor, an essential mammalian splicing factor. This gene is methylated on the maternal allele and expressed from the paternal allele. The maternal allele is nearly completely methylated at CpG sites within and around the gene1 (9). The gametic imprint of this gene has not yet been found. In this study, we made transgenic mice of U2af1-rs1 to study the mechanism of genomic imprinting.
The DNA construct Xb5-sfi was prepared for
the pronuclear microinjection of DNA into fertilized eggs as
illustrated in Fig. 1a. The DNA was derived from the mouse
U2af1-rs1 gene, Xb5-1S (9). To distinguish the transgene
from the endogenous mouse gene, an EcoRV linker (Takara) was
inserted into the SfiI site which is located 688 bp2 upstream from the ATG initiation
codon.
Transgenic Mice
The insert of Xb5-sfi was purified from agarose gel and microinjected into fertilized eggs of C57BL/6 by a previously described method (10).
Southern and Northern AnalysisSouthern and Northern hybridization analysis was performed for 20 h at 65 °C in a solution of 6 × SSC (0.9 M NaCl and 0.09 M sodium citrate), 0.5% (w/v) of SDS, 5 × Denhardt's solution (1% each of bovine serum albumin, Ficoll, and polyvinylpyrrolidone), 100 µg/ml heat-denatured herring-sperm DNA, and heat-denatured probes (1 × 106 cpm/ml). Membranes were washed for 20 min at 65 °C in 0.1 × SSC, 0.1% SDS. The intensity of the band was measured by a BAS2000 image analyzer (Fuji).
RT-PCRTotal RNA, prepared by the acid-phenol method (11), was applied to RT-PCR as described previously (12).
To study the trans-acting mechanism of genomic imprinting, we made
transgenic mice of an imprinted gene, U2af1-rs1 using the XbaI fragment (Fig. 1a) (9). We
expected that the multicopy transgene would titrate the trans-acting
factor which is involved in genomic imprinting, and as a result the
machinery would be disturbed. The XbaI fragment cloned from
C57BL/6 was the 8.3-kb genomic fragment that contained the whole coding
of U2af1-rs1 plus 2.9 kb upstream from the transcription
initiation site and 2.0 kb downstream from the poly(A) signal. This
fragment was injected into fertilized eggs of C57BL/6. We obtained five
transgenic F1 animals but only two transgenic
F1 gave offspring (Tg8 and Tg28). Both of these founder
animals had 20-30 transgene copies, and the expression level was twice
to three times that in the wild-type mouse (data not shown). We could
distinguish endogenous and transgene U2af1-rs1 by digestion
with EcoRV because the transgene was constructed to have an
additional EcoRV site in the 5-noncoding region (Fig. 1a). There is a SmaI site at the 199-bp upstream
position from the initiation site of transcription. This site on the
maternal allele is methylated and then not digested by SmaI
in the wild-type mouse. On the contrary, the site on the paternal
allele is not methylated and is then completely digested by
SmaI (Fig. 2).
The methylation status of the SmaI site of endogenous U2af1-rs1 was examined in the offspring of hemizygous transgenic females crossed with wild-type males and was found to be methylated as in the wild-type mouse (n = 138, data not shown). Next we examined the methylation status of the offspring of hemizygous transgenic males crossed with wild-type females. In most offspring, endogenous U2af1-rs1 transmitted from the transgenic fathers was not methylated like that in the wild-type mouse (110/123). However, in 11% (13/123) of the offspring, the endogenous U2af1-rs1 transmitted from the transgenic fathers was methylated unlike that in the wild-type mouse (Fig. 1b). This aberrant methylation is not restricted to the SmaI site because we can also detect it with NotI, HpaII, and HhaI (data not shown). Therefore, most of the allelic methylation sites may be related with this phenomenon. This methylation of the endogenous gene was not only observed in the transgene-positive offspring, but also in the transgene-negative offspring (Fig. 1b). Therefore, this aberrant methylation may be caused by the transgene in the father, rather than in the offspring. The aberrant methylation could not be due to any mutations affecting the methylation of the endogenous U2af1-rs1 gene. If a mutation occurred in the father's germ cell and was transmitted to the offspring to cause aberrant methylation, this mutation would be further transmitted to cause aberrant methylation in their offspring. However, no aberrant methylation was observed in the offspring of the transgene-negative males with the aberrant methylation crossed with wild-type females (data not shown). This aberrant methylation is not a naturally occurring phenomenon because we could not detect it in any wild-type mice (n = 130). This phenomenon does not seem to be a special effect caused by integration of the transgene into a special site because it was observed in the offspring of two independent founder animals Tg8 and Tg28. Southern blot analysis of Tg8 and Tg28 also supports this speculation (data not shown).
To examine whether the aberrant methylation of the endogenous gene is
related to the expression of the U2af1-rs1 gene, a
transgene-negative mouse with the aberrant methylation was analyzed by
Northern blot analysis. We could not detect any signal in this mouse in
contrast to the strong signal in the wild-type and normal offspring
(Fig. 3). Despite the absence of U2af1-rs1
expression, no gross abnormality was observed in the mouse. A related
gene, U2af1-rs2 (13), may compensate U2af1-rs1.
RT-PCR using a radioisotope was performed to detect the weak expression
of U2af1-rs1. There was no signal in the mouse in contrast
to the strong signal of paternal alleles of wild-type and normal
offspring (Fig. 4, short exposure). A long
exposure revealed that both paternal and maternal alleles were equally
expressed at a low level (Fig. 4). This level was equal to the level
expressed from the maternal alleles in wild-type and normal offspring.
Therefore, the aberrant methylation is related to the suppression of
the endogenous U2af1-rs1 gene.
The methylation status was also examined in the testis and sperm of the
transgenic father. In the testis and sperm, the endogenous gene was
mostly methylated in the transgenic father in contrast to complete
demethylation in the wild-type mouse (Fig. 5). This confirms that the aberrant methylation was caused by a transgene in the
testis of the father. However, the population of methylated endogenous
U2af1-rs1 is different between the sperm of the father and
its offspring. This may be due to the demethylation of the gene in the
remaining offspring after fertilization or the disadvantage to survival
of the sperm with methylated U2af1-rs1.
Our findings suggest the existence of a trans-acting factor(s) that interacts with the U2af1-rs1 gene and prevents it from being methylated in the testis. In the testis of the father with the transgene, this factor seems to be titrated by the multicopy transgene. It is intriguing to think that a large CpG island in the U2af1-rs1, which has a very high GC and CpG content and contains a multiple tandem repeat of short sequence (8, 9), is involved in the titration. The possibility that the overproduction of U2af1-rs1 protein by the transgene elicits some kind of response to cause aberrant methylation cannot be denied. However, this seems to be unlikely because the U2af1-rs1 expression level in transgenic mouse is only twice to three times that in the wild-type mouse (data not shown). Is the machinery involved in the aberrant methylation of the testis related to that involved in gametic imprint? The XbaI fragment seems not to contain the elements necessary for maternal gametic imprint, and it should exist outside the fragment because not all of the population of the transgene is methylated in maternal transmission (data not shown). If there is an element sufficient for maternal gametic imprint in the XbaI fragment, the whole transgene population should be methylated in maternal transmission. The methylated endogenous U2af1-rs1 in the testis of the transgenic father also should be kept methylated in all offspring. However, only 11% of the offspring had an aberrant methylation. Therefore, the machinery involving this trans-acting factor(s) should be different from that making a gametic imprint in female gametogenesis. In other words, U2af1-rs1 seems to have two mechanisms for regulating the methylation status in gametogenesis. By one mechanism in female gametogenesis, the U2af1-rs1 gene is methylated and is not expressed in the offspring. By another mechanism in male gametogenesis, the U2af1-rs1 gene is kept unmethylated to be expressed in the offspring. Here, we observed that the latter mechanism was affected by a multicopy transgene. Whether this mechanism can be similarly applied to other imprinted genes remains to be determined.
This is the first observation that the endogenous methylation of an imprinting gene was affected by its own transgene. This artificial system would be useful to study the imprinting mechanism.
We thank Dr. J. Forejt for useful advice and K. Tohyama for technical assistance.