Characterization of epidermal growth factor-like domain 7 (EGFL7) expression in normal endometrium and in the endometrium of women with poor reproductive outcomes

STUDY QUESTION

Could epidermal growth factor-like domain 7 (EGFL7) be a factor involved in the preparation of the endometrium for implantation and could its dysregulation be implicated in poor reproductive outcomes?

SUMMARY ANSWER

EGFL7 is highly expressed in the endothelium and glandular epithelium throughout the menstrual cycle; it is upregulated by stromal cells in secretory phase and appears strongly reduced in endometrial biopsies and isolated stromal cells of women with unexplained recurrent pregnancy loss (uRPL) and recurrent implantation failure (RIF).

WHAT IS KNOWN ALREADY

The secreted factor EGFL7, originally identified as a gene primarily expressed in endothelial cells, is also expressed by the mouse blastocyst and by mouse and human trophoblast cells. It regulates trophoblast migration and invasion by activating NOTCH1 signaling. NOTCH1 has been demonstrated to play a fundamental role in endometrial receptivity and its dysregulation may be involved in selected pregnancy complications characterized by altered endometrial receptivity, such as uRPL.

STUDY DESIGN, SIZE, DURATION

This is an exploratory study for which 84 endometrial biopsies were collected from normally fertile women, as well as from women with uRPL and RIF.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Samples were collected from women in both the proliferative and secretory phases of the menstrual cycle and stratified into three sub-groups according to the patient clinical history: 20 fertile women (8 in proliferative and 12 in secretory phase), 41 women with uRPL (6 in proliferative and 35 in secretory phase), and 27 women with RIF (8 in proliferative and 19 in secretory phase). Immunohistochemistry, real-time PCR, and western blot analyses were performed to study the expression of EGFL7 and NOTCH1, as well as the NOTCH target genes.

MAIN RESULTS AND THE ROLE OF CHANCE

Analysis of spatial and temporal distribution of EGFL7 in endometrial biopsies from fertile women revealed higher levels of EGFL7 in samples from the secretory phase compared to proliferative phase. The expected expression of EGFL7 in endothelial cells was shown as well as the novel, not previously reported, expression in endometrial glands and stromal cells. EGFL7 was significantly reduced in the endometrium of women with uRPL and RIF in the secretory phases and this was associated with a downregulation of the NOTCH1 signaling pathway. Human recombinant EGFL7 was able to activate the NOTCH1 signaling pathway in endometrial stromal cells (EndSCs) obtained from fertile women but not in cells from uRPL or RIF patients. EndSCs from fertile women and decidualized in vitro for three days showed an upregulation of EGFL7 expression, whereas cells obtained from women with uRPL and RIF and decidualized in vitro did not.

LIMITATIONS, REASONS FOR CAUTION

This study was conducted with a relatively small number of patient samples. Although results are highly reproducible and consistent, additional observations from multicentric cohorts would strengthen the relevance of the data. Moreover, this is an in vitro study, which might only partially represent the in vivo conditions.

WIDER IMPLICATIONS OF THE FINDINGS

Our results demonstrate for the first time that EGFL7 is new player involved in decidualization and provide new insights into the pathophysiology of selected implantation defects and early pregnancy complications. Our studies have revealed that alterations in EGFL7 expression and the consequent dysregulation of NOTCH signaling are potential underlying causes of RIF and uRPL. Our results might have therapeutic relevance, as the EGFL7/NOTCH pathway may represent a potential target for medical intervention.

STUDY FUNDING/COMPETING INTEREST(S)

This study has been supported by the Grant for Fertility Innovation 2017 (Merck KGaA). There are no competing interests to disclose.

TRIAL REGISTRATION NUMBER

Not applicable.

Epidermal growth factor-like domain 7 promotes migration and invasion of human trophoblast cells through activation of MAPK, PI3K and NOTCH signaling pathways

Epidermal growth factor-like domain 7 (Egfl7) is a gene that encodes a partially secreted protein and whose expression is largely restricted to the endothelia. We recently reported that EGFL7 is also expressed by trophoblast cells in mouse and human placentas. Here, we investigated the molecular pathways that are regulated by EGFL7 in trophoblast cells. Stable EGFL7 overexpression in a Jeg3 human choriocarcinoma cell line resulted in significantly increased cell migration and invasiveness, while cell proliferation was unaffected. Analysis of mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K) pathways showed that EGFL7 promotes Jeg3 cell motility by activating both pathways. We show that EGFL7 activates the epidermal growth factor receptor (EGFR) in Jeg3 cells, resulting in downstream activation of extracellular regulated kinases (ERKs). In addition, we provide evidence that EGFL7-triggered migration of Jeg3 cells involves activation of NOTCH signaling. EGFL7 and NOTCH1 are co-expressed in Jeg3 cells, and blocking of NOTCH activation abrogates enhanced migration of Jeg3 cells overexpressing EGFL7. We also demonstrate that signaling through EGFR and NOTCH converged to mediate EGFL7 effects. Reduction of endogenous EGFL7 expression in Jeg3 cells significantly decreased cell migration. We further confirmed that EGFL7 stimulates cell migration by using primary human first trimester trophoblast (PTB) cells overexpressing EGFL7. In conclusion, our data suggest that in trophoblast cells, EGFL7 regulates cell migration and invasion by activating multiple signaling pathways. Our results provide a possible explanation for the correlation between reduced expression of EGFL7 and inadequate trophoblast invasion observed in placentopathies.

Heterogeneous expression of Gap junction channels in the heart leads to conduction defects and ventricular dysfunction

BACKGROUND

- Heterogeneous remodeling of gap junctions is observed in many forms of heart disease. The consequent loss of synchronous ventricular activation has been hypothesized to result in diminished cardiac performance. To directly test this hypothesis, we designed a murine model of heterogeneous gap junction channel expression. Methods and Results-- We generated chimeric mice formed from connexin43 (Cx43)-deficient embryonic stem cells and wild-type or genetically marked ROSA26 recipient blastocysts. Chimeric mice developed normally, without histological evidence of myocardial fibrosis or hypertrophy. Heterogeneous Cx43 expression resulted in conduction defects, however, as well as markedly depressed contractile function. Optical mapping of chimeric hearts by use of voltage-sensitive dyes revealed highly irregular epicardial conduction patterns, quantified as significantly greater negative curvature of the activation wave front (-1.86+/-0.40 mm in chimeric mice versus -0.86+/-0.098 mm in controls; P<0.01; n=6 for each group). Echocardiographic studies demonstrated significantly reduced fractional shortening in chimeric mice (26.6+/-2.3% versus 36.5+/-1.6% in age-matched 129/SvxC57BL/6F1 wild-type controls; P<0.05).

CONCLUSIONS

- These data suggest that heterogeneous Cx43 expression, by perturbing the normal pattern of coordinated myocardial excitation, may directly depress cardiac performance.

Conduction slowing and sudden arrhythmic death in mice with cardiac-restricted inactivation of connexin43

Cardiac arrhythmia is a common and often lethal manifestation of many forms of heart disease. Gap junction remodeling has been postulated to contribute to the increased propensity for arrhythmogenesis in diseased myocardium, although a causative role in vivo remains speculative. By generating mice with cardiac-restricted knockout of connexin43 (Cx43), we have circumvented the perinatal lethal developmental defect associated with germline inactivation of this gap junction channel gene and uncovered an essential role for Cx43 in the maintenance of electrical stability. Mice with cardiac-specific loss of Cx43 have normal heart structure and contractile function, and yet they uniformly (28 of 28 conditional Cx43 knockout mice observed) develop sudden cardiac death from spontaneous ventricular arrhythmias by 2 months of age. Optical mapping of the epicardial electrical activation pattern in Cx43 conditional knockout mice revealed that ventricular conduction velocity was significantly slowed by up to 55% in the transverse direction and 42% in the longitudinal direction, resulting in an increase in anisotropic ratio compared with control littermates (2.1+/-0.13 versus 1.66+/-0.06; P:<0.01). This novel genetic murine model of primary sudden cardiac death defines gap junctional abnormalities as a key molecular feature of the arrhythmogenic substrate.

Retroviral promoter-trap insertion into a novel mammalian septin gene expressed during mouse neuronal development

We have characterized a retroviral promoter-trap insertion into a novel mammalian septin gene, Sep3. Its predicted amino acid sequence shares significant homology to that of Saccharomyces cerevisiae CDC3, CDC10, CDC11, CDC12, the Drosophila genes Pnut, Sep1, Sep2, and the mammalian genes BH5, CDC10, Nedd5, Diff6, and Sep2, which are implicated in cytokinesis and cell polarity. Sep3 encodes a protein of 465 amino acids, and contains an evolutionary conserved ATP/GTP-binding motif, two coiled-coil domains, and a highly hydrophobic domain at the C terminus. Alkaline phosphatase reporter gene expression in transgenic embryos was first detected at E8.5 in the neural fold, and high levels of expression continued throughout embryogenesis in the neural tube and brain. In addition, a low level of transient expression was detected in the somites, gut, and branchial arches of mouse embryos. Overall, reporter gene expression recapitulated Sep3 mRNA expression during mouse embryogenesis. In adults, Sep3 transcripts were only detected in the brain and testis. Zoo blot analysis revealed that Sep3-related sequences exist in several vertebrate species including zebrafish, frog, chicken, mouse and human. Consistent with the retroviral insertion into the 3' UTR of the Sep3 gene, no obvious phenotypes associated with the promoter trap were detected in transgenic embryos or adult mice. In summary, we report the first isolation of a novel full-length Sep3 cDNA and extensive characterization of its expression during mouse embryogenesis and in adult tissues.

Use of developmental marker genes to define temporal and spatial patterns of differentiation during embryoid body formation

Mouse embryonic stem cells are pluripotent cells that are derived from the inner cell mass of blastocysts. When induced to synchronously enter a program of differentiation in vitro, they form embryoid bodies that contain cells of the mesodermal, hematopoietic, endothelial, muscle, and neuronal lineages. Here, we used a panel of marker genes with early expression within the germ layers (oct-3, Brachyury T, Fgf-5, nodal, and GATA-4) or a variety of lineages (flk-1, Nkx-2.5, EKLF, and Msx3) to determine how progressive differentiation of embryoid bodies in culture correlated with early postimplantation development of mouse embryos. Using RNA in situ hybridization, we found that the temporal and spatial relationships existing between these marker genes in vivo were maintained also in vitro. Studying the onset of marker gene expression allowed us also to determine the time course of differentiation during the formation of embryoid bodies. Thus, stages equivalent to embryogenesis between implantation and the beginning of gastrulation (4.5-6.5 d.p.c.) occur within the first two days of embryoid body differentiation. Between days 3 and 5, embryoid bodies contain cell lineages found in embryos during gastrulation at 6.5 to 7.0 d.p.c., and after day 6 in culture, embryoid bodies are equivalent to early organogenesis-stage embryos (7.5 d.p.c.). In addition, we demonstrate that the panel of developmental markers can be applied in a screen for stage- or lineage-specific genes. Reporter gene expression from entrapment vector insertions can be co-localized with expression of specific markers within the same cell during embryoid body formation as well as during embryogenesis. Our results thus demonstrate the power of embryoid body formation as an in vitro model system to study early lineage determination and organogenesis in mammals, and indicate that they will prove to be useful tools for identifying developmental genes whose expression is restricted to particular lineages.

Vezf1: A Zn finger transcription factor restricted to endothelial cells and their precursors

Using retroviral entrapment vectors, we identified a novel mouse gene whose expression is restricted to vascular endothelial cells and their precursors in the yolk sac blood islands. A 3.68-kb cDNA corresponding to the endogenous transcript was isolated using genomic DNA flanking the entrapment vector insertion as a probe. We have named this gene Vezf1 for vascular endothelial zinc finger 1. Vezf1 encodes a protein with a predicted molecular mass of 56 kDa and that contains six putative zinc finger domains and shows high homology to a previously identified human gene, DB1, that is believed to be involved in regulating expression of cytokine genes such as interleukin-3. In situ hybridization analysis revealed the onset of expression in advanced primitive streak-stage embryos being located in the extraembryonic mesodermal component of the visceral yolk sac and in the anteriormost mesoderm of the embryo proper. During head-fold and somite stages, expression was restricted to vascular endothelial cells that arise during both vasculogenesis and angiogenesis. Vezf1-related sequences were found to be highly conserved among higher vertebrate species that have acquired extraembryonic yolk sac membranes during evolution. The Vezf1 locus mapped to the proximal part of mouse chromosome 2, a region which has homology to human chromosome 9q. Vezf1 expression correlates temporally and spatially with the early differentiation of angioblasts into the endothelial cell lineage and the proliferation of endothelial cells of the embryonic vascular system. Thus, Vezf1 may play an important role in the endothelial lineage determination and may have an additional role during later stages of embryonic vasculogenesis and angiogenesis.

Heterologous expression and functional characterization of a mouse renal organic anion transporter in mammalian cells

Organic anion transporters play an essential role in eliminating a wide range of organic anions including endogenous compounds, xenobiotics, and their metabolites from kidney, thereby preventing their potentially toxic effects within the body. The goal of this study was to extend our previous study on the functional characterization and post-translational modification of a mouse kidney organic anion transporter (mOAT), in a mammalian cell system, COS-7 cells. The transporter-mediated p-aminohippurate (PAH) uptake was saturable, probenecid-sensitive, and inhibited by a wide range of organic anions including vitamins, anti-hypertensive drugs, anti-tumor drugs, and anti-inflammatory drugs. Tunicamycin, an inhibitor of asparagine-linked glycosylation, significantly inhibited the transport activity. Immunofluorescence provided evidence that most of the protein remained in the intracellular compartment in tunicamycin-treated cells. Diethyl pyrocarbonate (DEPC), a histidine residue-specific reagent, completely blocked PAH transport. The inhibitory effect by DEPC was significantly protected (90%) by pretreating the cells with excess unlabeled PAH, suggesting that the histidine residues may be close to the PAH binding sites. Finally, in situ mRNA localization was studied in postnatal mouse kidney. The expression was observed in proximal tubules throughout development. We conclude that COS-7 cells may be useful in pharmacological and molecular biological studies of this carrier. The carbohydrate moieties are necessary for the proper trafficking of mOAT to the plasma membrane, and histidine residues appear to be important for the transport function.

Large-scale screening for developmental genes in embryonic stem cells and embryoid bodies using retroviral entrapment vectors

Mammalian development is orchestrated by a variety of cellular proteins with expression that is regulated precisely. Although some of the genes encoding these factors have been identified, largely by homology to those of simpler organisms, the majority of them presumably remain unknown. We report here on the results of a large-scale genetic screen that can potentially lead to the identification of many of these unidentified genes in mice. The method we developed takes advantage of the fact that many of the factors that regulate early development are expressed at highly specific stages of early embryogenesis. We therefore established a method for tagging candidate developmental genes by virtue of their expression in a stage-specific manner during formation of embryoid bodies without a bias for their expression in undifferentiated embryonic stem (ES) cells. Of 2,400 ES cell clones with random insertions of retroviral vectors carrying a human placental alkaline phosphatase reporter gene (AP), 41 clones exhibited stage-specific reporter gene expression during embryoid body formation. The majority of these insertions were in genes that are not expressed in undifferentiated ES cells. Eleven ES cell clones with characteristic patterns of AP reporter gene expression in vitro were chosen for further examination in vivo for AP expression in developing embryos. Ten ES cell clones exhibited AP expression between day 7.5 and day 10.5 of development. Clones that showed restricted reporter gene expression in vitro also exhibited similar temporally and spatially restricted AP expression in vivo. Sequence analysis of genomic DNA flanking several vector insertions and corresponding cDNAs suggested that several of the insertions identified a previously unidentified gene. Thus, screening for reporter gene expression during embryoid body formation provides an efficient means of enriching clones that contain vector insertions into potentially novel genes that are important for regulating different stages of early postimplantation development.

Retrovirus-mediated gene transfer of 6-pyruvoyl-tetrahydropterin synthase corrects tetrahydrobiopterin deficiency in fibroblasts from hyperphenylalaninemic patients

Tetrahydrobiopterin (BH4) deficiency, a variant form of hyperphenylalaninemia with progressive neurological dysfunction, is primarily caused by autosomal recessive mutations in the gene encoding the 6-pyruvoyl-tetrahydropterin synthase (PTPS). PTPS is a biosynthetic enzyme for the BH4 co-factor, and its deficiency is associated with a malfunction of the phenylalanine catabolism in the liver and a lack of biogenic amine neurotransmitters dopamine and serotonin in the brain. We have previously isolated the wild-type PTPS cDNA and identified several mutations responsible for a decreased enzyme in patients. This study reports the in vitro correction of BH4 deficiency by using retrovirus mediated transfer of the PTPS cDNA into primary fibroblast cultures established from different patients. The Bing packaging cell line was used for amphotropic virus production. Following PTPS gene transfer, stimulation with cytokines restored biosynthesis of BH4 in originally defective cells to values comparable to those of heterozygous fibroblasts from clinically healthy subjects. These results not only provide a direct proof that the mutations in PTPS were causative for the mutant phenotype, but they are also the first step toward gene therapy as a potential alternative approach to treat BH4 deficiency.

Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection

Rare individuals have been multiply exposed to HIV-1 but remain uninfected. The CD4+ T-cells of two of these individuals, designated EU2 and EU3, are highly resistant in vitro to the entry of primary macrophagetropic virus but are readily infectable with transformed T-cell line adapted viruses. We report here on the genetic basis of this resistance. We found that EU2 and EU3 have a homozygous defect in CKR-5, the gene encoding the recently described coreceptor for primary HIV-1 isolates. These individuals appear to have inherited a defective CKR-5 allele that contains an internal 32 base pair deletion. The encoded protein is severely truncated and cannot be detected at the cell surface. Surprisingly, this defect has no obvious phenotype in the affected individuals. Thus, a CKR-5 allele present in the human population appears to protect homozygous individuals from sexual transmission of HIV-1. Heterozygous individuals are quite common (approximately 20%) in some populations. These findings indicate the importance of CKR-5 in HIV-1 transmission and suggest that targeting the HIV-1-CKR-5 interaction may provide a means of preventing or slowing disease progression.

Homologous recombination of copackaged retrovirus RNAs during reverse transcription

According to prevailing models, the high frequency of recombination in retroviruses occurs during reverse transcription of two genetically different genomes copackaged into virion particles. This view has been tested in our studies of the mechanism of recombination within homologous sequences of two retroviral genomes during a single round of virus replication and in the absence of helper virus. The recombination substrates were Moloney murine leukemia virus-based vectors, each of which contains an altered defective neomycin gene (neo) under the transcriptional control of the 5' long terminal repeat; the 3' sequences of each construct contain either the Moloney murine leukemia virus or simian virus 40 large-T polyadenylation sequence. One neo gene contained a linker insertion mutation at the 5' end (neo minus), and the other contained a deletion and linker insertion at the 3' end (neo delta 3). Each of the mutant neo constructs was introduced into the packaging helper cell line psi 2 by sequential cotransfection, and individual psi 2 double transformants were selected. Supernatant fluids from the cloned psi 2 double transformants were used to infect NIH 3T3 cells, and recombinant neo+ proviruses were detected by their ability to confer G418 resistance during infection of NIH 3T3 cells. Our results show that (i) recombination between a homologous sequence of about 560 bp occurred with a frequency of about 10(-4) per virus replication cycle; (ii) recombination occurred only after the viral RNAs had been packaged into particles, i.e., recombination between the two vector DNAs or between viral RNAs prior to packaging was not detected; and (iii) copackaging of two different genomic RNAs as a heterodimer is a prerequisite for recombination. Furthermore, our results indicate that recombination can occur during the DNA negative-strand synthesis of reverse transcription.

Transduction of cellular neo mRNA by retrovirus-mediated recombination

Transduction of cellular oncogenes by retroviruses is thought to be a multistep process, involving transcriptional activation of a cellular gene by upstream proviral integration and joining of cellular DNA to retroviral transcriptional signals, followed by copackaging and recombination with a helper virus genome during reverse transcription. To examine the molecular mechanism of the reverse transcriptase-mediated recombination, we introduced into mouse fibroblast cells a variety of constructs in which the neo selectable marker was joined to flanking retroviruslike or cell-like sequences. After superinfection and copackaging with a replication-competent Mo-MuLVsupF virus, the formation of recombinant neo transducing viruses was assessed in a second round of virus infection by the ability to confer G418 resistance to infected cells. Our results showed that recombinant neo proviruses were generated from neo RNA containing either a 5' or 3' retroviral end, implying that one recombination event with helper virus RNA was sufficient to incorporate the neo gene into proviral DNA. Recombination occurred with an apparent frequency of 10(-4) to 10(-5) per replication cycle in the absence of homology between the two recombining partners. This frequency, however, increased at least 100-fold if homology was provided at the site of recombination. Our results support the hypothesis that neo-transducing viruses arise via reverse transcriptase-mediated recombination of RNA rather than by recombination proceeding through DNA intermediates. Unexpectedly, removal of the retroviral packaging site psi reduced the number of neo recombinants only slightly. Our data indicated that although RNAs lacking the psi site are poorly packaged into virions, those RNAs that are included in the virions undergo frequent recombination, even if there is no selection for recombination. Many of the neo recombinants formed with the psi- constructs had undergone additional recombinations and often incorporated the psi site from the helper RNA.

Transfer of a mutant dihydrofolate reductase gene into pre- and postimplantation mouse embryos by a replication-competent retrovirus vector

In order to explore the potential of retrovirus vectors for efficiently transferring foreign genes into mouse embryos, a replication-competent recombinant Moloney murine leukemia virus (Mo-MLV) vector carrying a mutant dihydrofolate reductase (DHFR) cDNA insert in the U3 region of the viral long terminal repeat was used to infect pre- and postimplantation embryos. When preimplantation mouse embryos were infected with the vector, as expected, the provirus integrated into the embryos and the germ line with the same efficiency as that observed with wild-type Mo-MLV, leading to inactivation of the recombinant virus. In contrast, when postimplantation mouse embryos were microinjected with virus-producing cells, between 90 to 100% of the surviving animals proved to be infected with the virus. The recombinant virus spread as efficiently as wild-type Mo-MLV in the infected embryos, resulting in up to three to five proviral copies per genome in heart, thymus, and brain tissues. Substantial expression of mutant DHFR*-coding viral message was found in all somatic tissues analyzed, the amounts correlating with the proviral copy number in the respective organ. These results suggest that replication-competent vectors are useful for efficient transfer and expression of foreign genes into tissues or whole animals when virus spread is needed.

Construction and properties of replication-competent murine retroviral vectors encoding methotrexate resistance

A series of replication-competent Moloney murine leukemia virus vectors was constructed in which each vector contained a mutant dihydrofolate reductase (DHFR) cDNA insert in the U3 region of the viral long terminal repeat. Two of the resulting viruses, MLV (murine leukemia virus) DHFR*-5 and MLV DHFR*-7, were able to stably transfer methotrexate resistance to infected fibroblast cells upon multiple rounds of virus replication and in the absence of drug selection. Cell lines producing recombinant virus with high titers were established, which indicated that the insert did not grossly interfere with viral replication functions. These vectors should be useful for introducing and expressing foreign genes in vivo in tissues and whole animals in which virus spread is needed for efficient infection.

Introduction of a selectable gene into different animal tissue by a retrovirus recombinant vector

The potential use of retrovirus vectors to transduce foreign genetic information into cells of different tissues of an animal was explored by introducing a recombinant genome carrying the Eco gpt gene into postimplantation mouse embryos. To obviate the need for preparing concentrated virus stocks, psi 2-2-5 cells producing the replication-defective murine sarcoma virus (MSV)-gpt virus were microinjected directly into embryos. The psi 2-2-5 cells were mixed with cells producing replication-competent Moloney murine leukemia virus (Mo-MuLV) to facilitate spread of the vector. A high percentage of the manipulated embryos continued to develop without disturbance and were analyzed either prior to birth or as adults for expression of both helper and Eco gpt virus. Microinjection of as few as 10 Mo-MuLV-producing cells resulted in viremia of greater than 50% of the embryos or adults, 25%-30% of which produced MSV-gpt recombinant virus in a variety of organs including thymus, spleen, lung, kidney, and brain. The fraction of vector-producing cells, however, was 3 to 5 orders of magnitude lower than that of helper-virus-producing cells. Our results demonstrate that a selectable gene can be introduced by retroviral vectors into animals and can be expressed in a wide variety of different somatic tissues.

Retrovirus genomes methylated by mammalian but not bacterial methylase are non-infectious

The biological importance of DNA methylation for gene expression in eukaryotes is becoming increasingly evident, and a direct role of methylation in gene expression has been suggested by an analysis of the infectivity of integrated retroviral genomes in a transfection assay. These studies, however, did not address whether specific methylatable residues are involved in gene regulation. Methylation by sequence-specific bacterial DNA methylases has been shown to suppress the expression of some genes, but not others. To investigate the effect of methylation on gene expression without having to rely on sequence-specific methylases, a rat liver enzyme was used to methylate in vitro all C-G dinucleotides of a proviral genomic clone. This treatment reduced the biological activity of Moloney murine leukaemia virus (M-MuLV) proviral DNA by more than three orders of magnitude, whereas complete methylation of 35 HpaII sites in the same DNA had only a marginal effect. The rat methylase-induced inactivation was reversible, as treatment of recipient cells with 5-azacytidine rendered the non-infectious viral genomes biologically active. This suggests that methylation in other C-G dinucleotides than those detectable with restriction enzymes can be crucial for gene expression.

Endogenous Moloney leukemia virus in nonviremic Mov substrains of mice carries defects in the proviral genome

Substrains of mice carrying Moloney murine leukemia virus as a Mendelian gene (Mov locus) have been derived previously. Some of these strains, i.e., Mov-3 and Mov-9, develop viremia, whereas others, i.e., Mov-2, Mov-7, and Mov-10, do not regularly activate virus. We previously have molecularly cloned the respective Mov loci and shown that proviral clones derived from the different viral loci were either infectious (Mov-3, Mov-9) or failed to induce infectious virus (Mov-2, Mov-7, Mov-10) in a transfection assay. To analyze the sites affecting infectivity of the latter clones, complementation assays, in vitro recombinations, and marker rescue experiments were performed. Our results show that the three endogenous Moloney murine leukemia virus clones derived from Mov-2, Mov-7, and Mov-10 carry different mutations in the gag-pol region of the proviral genome. No inhibitory effect of flanking mouse sequences on provirus infectivity was observed.

De novo methylation and expression of retroviral genomes during mouse embryogenesis

Retrovirus genomes introduced into mouse zygotes by microinjection of cloned DNA, or into morula stage pre-implantation mouse embryos by infection with Moloney murine leukaemia virus (M-MuLV), became de novo methylated and were blocked in expression. No restriction of virus expression and no de novo methylation were observed when post-implantation mouse embryos were infected with virus. Efficient de novo methylation activity may be an important characteristic of gene regulation in early mouse embryos.

De novo methylation, expression, and infectivity of retroviral genomes introduced into embryonal carcinoma cells

We have investigated the block to expression of Moloney murine leukemia virus in murine embryonal carcinoma (EC) cells. Infected EC cells were found to contain up to 100 integrated proviral genomes. However, expression of virus as measured by XC plaque and virus-specific RNA synthesis did not occur at significant levels, in contrast to productively infected differentiated cells. Analysis of the DNA in the infected EC cells revealed that the proviral genomes were highly methylated, as shown by their resistance to cleavage by Sma I. Integrated proviral genomes in infected differentiated cells were readily cut by Sma I and thus were not methylated at these sites. Transfection of DNA from infected EC cells to cells permissive for virus expression failed to induce virus expression. The proviral genomes, however, were potentially infectious because they induced XC plaques when the recipient cells for transfection were treated with 5-azacytidine. This drug is believed to interfere with DNA methylation. We conclude that expression of proviral genomes introduced into EC cells is suppressed and that this inactivation can be correlated with the de novo methylation of the viral DNA. De novo methylation activity thus may be a characteristic of early embryonic cells.

Cloning of two genetically transmitted Moloney leukemia proviral genomes: correlation between biological activity of the cloned DNA and viral genome activation in the animal

The Mov-7 and Mov-9 substrains of mice, carrying Moloney murine leukemia virus (M-MuLV) in their germ line at the Mov-7 locus and Mov-9 locus, respectively, are different with respect to virus activation. Infectious virus appears in all mice carrying the Mov-9 locus but is not activated in animals carrying the Mov-7 locus. Consequently, only Mov-9 mice develop viremia and subsequent leukemia. The endogenous M-MuLV provirus with flanking mouse sequences corresponding to the Mov-7 and Mov-9 loci was molecularly cloned. Detailed restriction maps obtained from the cloned DNAs revealed no detectable differences in the proviral genomes. The flanking mouse sequences, however, were different, confirming that the Mov-7 and Mov-9 loci represent different integration sites of M-MuLV. Both clones induced XC plaques in a transfection assay. The specific infectivity of the clones, however, was different. A total of 10(-5) XC plaques per genome equivalent were induced by the Mov-9 clone, whereas only 10(-9) XC plaques per genome equivalent were induced by the Mov-7 clone. Moreover, NIH 3T3 cells transfected with the Mov-9 clone produced NB-tropic M-MuLV, whereas cells transfected with the Mov-7 clone did not produce infectious virus. The results suggest that M-MuLV integrated at the Mov-7 locus carries a mutation which prevents synthesis of infectious virus but permits XC plaque induction by partial genome expression or synthesis of noninfectious particles. Thus, the pattern of virus expression in Mov-7 and Mov-9 mice correlates with the biological properties of the respective clones. Genomic DNA from Mov-9 mice was not infectious in the transfection assay (specific infectivity < 10(-7) PFU per genome equivalent). As the only difference between the genomic and the cloned Mov-9 DNA appears to be the presence of 5-methylcytosine in CpG sequences, our results suggest that removal of methyl groups by molecular cloning in procaryotes permits genome expression in transfected eucaryotic cells. Our results support the hypothesis that DNA methylation is relevant not only in genome expression in the animal but also in expression of genes transfected into eucaryotic cells.

Infectivity and structure of molecular clones obtained from two genetically transmitted Moloney leukemia proviral genomes

The Mov-2 and Mov-10 substrains of mice, each carrying Moloney leukemia virus (= M-MuLV) in their germ line at the Mov-2 and Mov-10 locus, respectively, do occasionally at a later age (Mov-2) or not at all (Mov-10) activate infectious virus. The M-MuLV proviruses with flanking mouse sequences corresponding to the Mov-2 and Mov-10 locus, respectively, were molecularly cloned. Restriction enzyme analysis revealed no major deletions or insertions in the proviral genomes of the Mov-2 and Mov-10 locus. Both cloned DNAs induced XC plaques in a transfection assay. The specific infectivity, however, was very low and 3T3 cells transfected with the Mov-2 or Mov-10 clone did not produce infectious virus. Removing part of the 5' cellular sequences from the Mov-10 clone did not increase the infectivity. The results suggest that the M-MuLV integrated at the Mov-2 and Mov-10 locus carry a mutation which prevents synthesis of infectious virus but permits XC plaque induction by partial genome expression or synthesis of non-infectious particles.

DNA methylation, retroviruses, and embryogenesis

By exposing preimplantation embryos to Moloney leukemia virus (M-MuLV), we have previously derived substrains of mice designated as Mov-1-Mov-13 which genetically transmit the virus from one generation to the next. In some of the substrains the inserted viral genome becomes activated at specific stages of embryogenesis and the available evidence suggests that these viral genomes are developmentally regulated. To investigate the effect of cellular differentiation on virus expression, M-MuLV was introduced either into preimplantation or post-implantation mouse embryos or into embryonal carcinoma (EC) cells. Whereas preimplantation embryos or EC cells are not permissive for virus expression, efficient replication occurred in postimplantation embryos or in differentiated cell lines. The viral genomes introduced into early embryonal cells were highly methylated and noninfectious when analyzed in the adult. In contrast, viral genomes introduced into postimplantation embryos or into differentiated cells remained unmethylated and were infectious in a transfection assay. These results demonstrate an efficient de novo methylation activity which appears to be involved in repression of genes introduced into pluripotent embryonal cells and which is not observed in cells of the postimplantation embryo or in differentiated cells in tissue culture.

DNA methylation and gene expression: endogenous retroviral genome becomes infectious after molecular cloning

The Mov-3 substrain of mice carries Moloney murine leukemia virus as a Mendelian gene in its germ line. All mice segregating the Mov-3 locus activate virus and develop viremia and leukemia. The integrated provirus (i.e., Mov-3 locus) was molecularly cloned from Mov-3 liver DNA as a 16.8 kilobase long EcoRI fragment. Comparison of the cloned and genomic Mov-3 specific EcoRI fragment by restriction enzyme analysis showed no differences in the size of the fragments, indicating that no major sequence rearrangements occurred during cloning. The genomic and cloned Mov-3 DNAs were compared for methylation and infectivity. Analysis with Hha I showed that the genomic proviral and the flanking mouse sequences were methylated at cytosine residues, in contrast to the cloned Mov-3 locus. The cloned Mov-3 locus, however, was highly infectious in a transfection assay (1 x 10(-3) plaque-forming unit per viral genome) in contrast to the genomic Mov-3 DNA (less than 10(-7) per viral genome). Our results suggest that genes containing 5-methylcytosine are not expressed after transfection into susceptible cells and that removal of the methyl groups by molecular cloning in prokaryotes leads to expression generating infectious proviral DNA. If gene expression of transfected DNA is controlled by mechanisms that are relevant for gene expression in the animal, this suggests that DNA methylation may play a causative role in eukaryotic gene regulation.

Infectivity and methylation of retroviral genomes is correlated with expression in the animal

We studied mechanisms controlling gene expression during animal development using retroviruses as model genes. For this, substrains of mice have been previously derived carrying the Moloney leukemia virus (M-MuLV) in their germ line. Virus activation occurs in some of these substrains at different stages of development, resulting in two classes of viral genomes. The genetically transmitted (endogenous) copy is present in every cell, whereas somatically acquired ("exogenous") copies are carried only in cells that were superinfected. We compared these two classes of M-MuLV genomes using two parameters. DNA sequences of the endogenous M-MuLV genome in all mouse substrains were highly methylated in GCGC, the recognition sequence of the restriction enzyme Hha I, and were not infectious (specific infectivity less than 10(-7) pfu per proviral genome) in a DNA transfection assay. In contrast, the "exogenous" copies were hypomethylated and infectious. These parameters are strongly correlated to genome activity in the animal: only tissues carrying exogenous copies express virus-specific RNA. With the assumption that gene expression of transfected DNA is controlled by mechanisms that are relevant for gene expression in the animal, our results suggest that DNA methylation plays a causative role in gene regulation during development and differentiation.