Progressive maturation of chromatin structure regulates HSP70.1 gene expression in the preimplantation mouse embryo

The packaging of DNA into chromatin has important consequences for the regulation of gene activity in eukaryotic organisms. The fundamental unit of chromatin organization is the histone octamer, in which 146 base pairs of DNA are wound around a central (H3/H4)₂ tetramer, flanked by two (H2A/H2B) dimers. Positioning of nucleosomes with respect to the DNA helix can have repressive effects on gene expression by blocking transcription factor access (Bogen-hagen et al., 1982; Knezetic and Luse, 1986; Workman et al., 1991), or stimulatory effects through juxtaposition of activating regulatory elements (Thomas and Elgin, 1988; McPherson et al., 1993; Schild et al., 1993). Incorporation of DNA into higher order chromatin structures is mediated by binding of histone H1 to the linker DNA between nucleosomes. Histone H1 has been considered as a repressor of transcription (Shimamura et al., 1989; Laybourn and Kadonga, 1991) but recent studies suggest caution in the generalisation of such a role (Sandaltzopolous et al., 1994; Bouvet et al., 1994).

In early embryonic development, switching of H1 subtypes appears to be a common theme, while alteration of core histone subtypes, as observed in the sea urchin, is much less wide-spread (Poccia, 1986). In Drosophila, H1 is not detected until the 9th or 10th cleavage cycle, coincident with the major activation of zygotic transcription (ZGA). Up to this point, H1 appears to be replaced by the high mobility group protein HMG-D, a protein which may generate a less compact chromatin structure, and facilitate rapid condensation and decondensation in the short early cleavage cycles (Ner and Travers, 1994). During Xenopus embryogenesis, H1 is replaced by the maternal variant, histone B4, until the mid-blastula transition (MBT), again coincident with the ZGA. In the latter organism, H1 has been shown to be involved in the switch from oocyte to somatic 5S rRNA gene expression, through specific repression of the oocyte 5S rRNA gene even in the presence of an abundance of the transcriptional activator TFIIIA (Bouvet et al., 1994).

Levels of histone H1 may play a role in regulating gene expression in early mouse embryos, as immunofluorescence studies have shown that H1 is not present until S-phase of the 4-cell stage (Clarke et al., 1992). An essential difference in early mammalian development, compared to that of Drosophila or Xenopus, is the absence of an extended series of nuclear divisions prior to activation of the zygotic genome. This is particularly true of the mouse, where activity of a cAMP-dependent protein kinase (Latham et al., 1992) has been implicated in establishing a transcriptionally permissive state as early as the 1-cell stage. During this minor activation of the genome (Schultz, 1993), transcriptional activity has been detected from the male pronucleus (Ram and Schultz, 1993; Christians et al., 1995). This then leads to the major ZGA at the 2-cell stage (Flach et al., 1982). Enhancers do not seem to have an important role in the regulation of episomal gene expression during the 1-cell stage, in S-phase arrested, aphidicolin-treated embryos, but become necessary to prevent repression of episomal expression from weak promoters at the 2-cell stage (Majumder et al., 1993). These results have led to the proposition that chromatin structure is important in the selective regulation of early zygotic gene expression in the mouse, perhaps via modulation of acetylation levels of histones in the nucleosomal core (Wiekowski et al., 1993). Further analysis of these hypotheses is limited by the fact that gene regulation data have been generated by microinjection of various DNA constructs into early embryos. The consequences of this approach are that it is impossible to control variability in the number of DNA copies introduced, large numbers of copies are injected, subsequent embryonic development is compromised by the injection process, and it is not trivial to relate the chromatin structure of supercoiled episomal plasmid templates to that of native embryonic chromatin.

A further indication that chromatin structure and nuclear architecture are involved in regulating the program of gene expression during ontogeny is the observation that a number of developmentally regulated genes are bordered by scaffold attachment regions (SARs). These include Adh, ftz, and Sgs-4 in Drosophila (Gasser and Laemmli, 1986), and the μ-globin gene locus in humans (Jarman and Higgs, 1988). Originally identified as AT-rich DNA sequences which bind to preparations of the nuclear matrix and frequently map to or near domain boundaries of gene loci (Gasser et al., 1989), SARs do not appear to be as consistent in insulating transgenes from position effects as Drosophila specialized chromatin structures (scs) (Kellum and Schedl, 1992) nor to act as dominant activators of tissue-specific gene expression as do locus control regions (LCRs) (Dillon and Grosveld, 1993). More recently it has been proposed that they may be involved in mediating chromatin accessibility, through synergistic action with enhancers (Forrester et al., 1994), or via their interaction with histone H1 (closed) or proteins such as the high mobility group protein, HMG-I/Y, capable of displacing H1 from the AT-tracts (open) (Zhao et al., 1993). We have investigated the effect of SAR sequences on gene expression in the early mouse embryo, a period when significant changes in chromatin structure, and particularly histone H1 concentration occur.

To overcome the limitations of transient expression assays, we generated transgenic mice in which the promoter of the heat shock gene HSP70.1 (Hunt and Calderwood, 1990) directed expression of a luciferase reporter cDNA (de Wet et al., 1987). HSP70.1 is part of the multigenic, inducible, hsp70 family known to be expressed constitutively in the 2-cell mouse embryo and to be heat inducible at the blastocyst stage (Bensaude et al., 1983). A second series of transgenic lines was then produced with the above construct flanked by 5′ and 3′ SARs from the human μ-interferon locus (Bode and Maass, 1988). Significant stage-specific differences were observed in the preimplantation expression profiles of the two different transgenic constructs. Modification of expression profiles in the presence of trichostatin A (TSA), which specifically inhibits histone deacetylases (Yoshida et al., 1990) and leads to hyperacetylation of core histones, was then examined and compared with the nuclear immunofluorescence staining pattern of acetylated histone H4 during early development. At embryonic stages where SAR+ lines showed increased activity, studies carried out with the topoisomerase II inhibitor, VM-26, suggested that recruitment of topoisomerase II to SAR sequences was not the mechanism responsible for the observed effect. Finally, preimplantation HMG-I/Y mRNA levels were studied by RT-PCR and its cellular localisation was analysed by immunofluorescence, to determine whether there was any correlation between the presence of HMG-I/Y and the transgene expression profiles. Taken together, the data characterize a progressive maturation of chromatin structure in the zygotic nucleus which plays an important role in regulating gene expression during the early cleavage stages of mouse development.

The transgenic mouse lines used in this study have been previously described (Thompson et al., 1994). Four lines, F2, F27, F29 and F31, with the murine HSP70.1 promoter directing firefly luciferase reporter expression, were compared with 5 lines, SF2, SF3, SF4, SF6 and SF9, harbouring the same promoter-reporter construct flanked by 5′ and 3′ SAR elements obtained from the human μ-interferon locus. The ‘F’ lines contained 2, 3, 1 and 1 transgene copies and the ‘SF’ lines, 4, 3, 6, 4 and 2 transgene copies, respectively.

For transient expression experiments, plasmid DNA was prepared by cesium chloride density gradient ultracentrifugation. Supercoiled plasmid DNA was microinjected as previously described (Hogan et al., 1986) at a concentration of 50 ng/μl using Narishige micromanipulators (Nikon), and an Eppendorf microinjector.

Mating of transgenic male mice to superovulated F₁ hybrid C57BL6×CBA females (Iffa Credo), recovery of 1-cell embryos, in vitro culture conditions, and luciferase assays were as described by Thompson et al. (1994). Constitutive expression of the transgene was determined at the 2-cell, 4-cell, 8-cell and blastocyst stages. Dark current photometer background was 150±20 relative light units (RLU) and therefore, 170 RLU were subtracted from all measured values. Timing of the anlaysis was with respect to hours post human chorionic gonadotropin (HCG) injection: 2-cell, 41-42 hours; 4-cell 63-64 hours; 8-cell 73-74 hours; and blastocyst 111-112 hours post HCG. For induced expression at the blastocyst stage, embryos at 111-112 hours post HCG were heat-shocked at 43°C for 30 minutes and allowed to recover at 37°C for 5-6 hours before lysis.

Treatment of transgenic embryos with drugs and inhibitors was carried out as shown in Fig. 1. Embryos were cultured from the 1-cell stage in drops of M16 medium (Hogan et al., 1986), and at the points indicated T1-T4 in the figure, were split into two groups. The test group was transferred to an equilibrated drop of M16 medium, containing a specified concentration of a given compound. The control group was transferred into a fresh equilibrated drop of M16 medium. Transfers took place towards the end of a given cleavage stage, embryos divided through to the next cell cycle, and were then harvested for analysis at the points indicated A1-A4 in Fig. 1. Stock solutions of 600 μM trichostatin A (TSA) (a gift from M. Yoshida, University of Tokyo) in 100% dimethylsulfoxide (DMSO), and 1 mM teniposide VM-26 (a gift from Sandoz laboratories) in 100% DMSO were prepared and diluted to working concentrations in M16 medium just prior to use.

An exception to the above protocol was treatment of embryos with VM-26, which was only studied in SAR+ lines, at the 2-cell stage and in heat-shocked blastocysts. When this inhibitor was introduced before cell cleavage, embryos were unable to pass through to the next cell cycle. Therefore, 2-cell embryos were picked off following cleavage from the 1-cell stage and 10 μM VM-26 was introduced. Assay of luciferase activity was performed at 41-42 hours post HCG. At the blastocyst stage, embryos at 111 hours post HCG were incubated for 1 hour in 10 μM VM-26 prior to heat shock, recovery, and assay as described above. VM-26 was maintained in the culture medium at 10 μM during heat shock and subsequent recovery.

Statistical analyses of expression data were performed with Statview II software (Abacus Concepts, Inc.)

Pools of 50 embryos of a given cleavage stage, were resuspended in 7.5 μl lysis buffer containing 1% NP40, 1.3× MMLV-RT buffer (GibcoBRL) and then heated at 100°C for 1 minute. Synthesis of cDNA was initiated by addition of 2.5 μl of a mixture containing 100 ng oligo-dT, 200 mM of each dNTP, 10 units of RNAse inhibitor (Boehringer), and 200 units of MMLV reverse transcriptase (GibcoBRL). Following a 1 hour incubation at 37°C, mixtures were heated for 5 minutes at 95°C. PCR amplifications of the cDNA for heat shock factor 2 (HSF2) mRNA (equivalent to 25 embryos), and HMG-I/Y mRNA (equivalent to 30 embryos), were carried out in 50 μl of 10 mM Tris-HCl, 1.5 mM MgCl₂, 50 mM KCl, 0.1% Triton X-100, 200 mM of each dNTP, 1 unit of TAQ DNA polymerase (Bioprobe), and 25 pmoles of each primer. After an initial denaturation step at 94°C for 10 minutes, samples were subjected to 35 cycles of amplification (94°C 30 seconds, 58°C 30 seconds, 72°C 30 seconds). This was in the linear range of amplification product versus cycle number. Aliquots of 20% of each sample were run on 1.8% agarose gels and transferred to nylon membranes (hybond-N+, Amersham) for probe hybridization. Autoradiographic exposure times at −80°C were 3 hours for HSF2 and overnight for HMG-I/Y.

The HMG-I/Y primers were derived from the published cDNA sequence (Johnson et al., 1988): sense primer, nucleotides 154 to 172; anti-sense primer, nucleotides 546-564. HSF2 primers were from the published cDNA sequence (Sarge et al., 1991): sense primer, nucleotides 1353-1373; anti-sense primer, nucleotides 1874-1894. The HSF2 probe was the mouse cDNA (gift from R. Morimoto, Northwestern University). The HMG-I/Y probe was cloned by RT-PCR from mouse fibroblast total RNA using the primers described above.

Foetuses obtained by mating transgenic male mice to non-superovulated F₁ hybrid C57BL6×CBA females were recovered 12-13 days post coitum and immediately killed by decapitation. Livers were removed and bodies were chopped into fine pieces with a scalpel. Subsequent preparation of fibroblasts was as described for primary culture of ear fibroblasts (Thompson et al., 1994), except that the final concentration of trypsin (Gibco) was 0.25% and collagenase was not used. Isolated fibroblasts were used in experiments after one or two passages except when aged preparations were desired. In the latter case, cells were passed four times, and then plated at low density and cultured for several days, with medium changes, until cell division had ceased. When foetal fibroblasts were treated with TSA, the incubation was for a period of 11-12 hours.

Prior to fixation, the zona pellucida was removed by treatment with 5 μg/ml pronase (Boehringer) in M2 medium (Hogan et al., 1986). Subsequent treatment of embryos and foetal fibroblasts was then identical, except that fibroblasts were transferred between baths attached to a coverslip, whereas embyros were unattached and transferred by mouth pipette. Two fixation protocols were used. In the first, cells were fixed in 2.5% paraformaldehyde (Sigma)/0.02% Triton X-100/phosphate-buffered saline (PBS) at 37°C for 15 minutes. In the second, cells were fixed in ethanol/H₂O/acetic acid, 95:4:1, for 1 hour at 4°C and then rehydrated in a series of 10-minute baths: 85% ethanol, 70% ethanol, 50% ethanol, and PBS/1% bovine serum albumin (BSA)/0.02% sodium azide. Further preparation was the same for both fixation protocols. Fixed cells were blocked in PBS/10% fetal calf serum or 10% sheep serum/0.2% Triton X-100. First antibodies and preimmune sera were diluted in PBS/2% fetal calf or sheep serum/0.2% Triton X-100 and incubations were performed overnight at 4°C. Cells were rinsed at 37°C in PBS/2% fetal calf or sheep serum/0.2% Triton X-100 for 30 seconds, 10 minutes, and 30 seconds and then incubated with fluorescein isothiocyanate-(FITC) conjugated sheep anti-rabbit IgG (1:400) (Sigma Immunochemicals) for 1 hour at 37°C. Cells were rinsed as above, counterstained in 10 μg/ml propidium iodide (Sigma) for 15 minutes at 37°C and mounted in Moviol 4-88 (Hoechst). Observations were made using a Zeiss confocal laser scanning microscope LSM-310 with a Zeiss plan neofluor 100× (NA 1.3) oil immersion objective. Images were analysed using NIH Image 1.54 software (National Institute of Health, USA). A minimum of 50 embryos were examined in 2 to 3 replicates for each cleavage stage and for each experimental treatment.

Two antibodies against acetylated forms of histone H4 were used. The first recognized all acetylated histone H4 isoforms (Lin et al., 1989) (a gift from D. Allis, Syracuse University), and was used at a 1:1000 dilution. The second antibody, specific for histone H4 acetlyated at lysine 5 (Turner and Fellows, 1989) (a gift from B. Turner, University of Birmingham), detects only the most highly acetylated isoforms in mammalian somatic cells and was used at a 1:700 dilution. The anti-HMG-I/Y antibody (a gift from R. Reeves, Washington State University) was used at a 1:100 dilution. Preimmune sera were provided with the first anti-acetylated histone H4 antibody and the anti-HMG-I/Y antibody. For the second anti-lysine 5 acetylated histone H4 antibody, the control consisted of overnight incubation in the absence of first antibody, followed by incubation with the FITC-conjugated second antibody.

The HSP70.1 gene is one of the few genes for which a preimplantation profile of expression has been characterized in early mouse development. RT-PCR analysis revealed that HSP70.1 is constitutively transcribed in the 2-cell embryo and is induced by heat shock at the blastocyst stage (Christians et al., 1995). Zygotic transcription of HSP70.1 begins during G₂ of the 1-cell stage and increases during G₁ of the 2-cell stage. The arrival of S-phase during the second cell cycle signals a reduction of HSP70.1 transcription to basal levels. Constitutive basal levels of endogene expression were also observed in 4- and 8-cell embryos and in non-heat shocked blastocysts as well as in a variety of differentiated tissues and cells in culture (Thompson et al., 1994).

The firefly luciferase transgene provides a sensitive single embryo assay with which to monitor HSP70.1 expression. Using purified firefly luciferase (Sigma) for calibration, 1 relative light unit (RLU) was produced by approximately 1.5 fg of luciferase. The lability of the luciferase is useful in following the kinetics of preimplantation expression. When purified luciferase was microinjected into early zygotes, to attain levels of expression similar or superior to those generated by transgenic embryos, the half-life of the luciferase was less than 2 hours (Christians et al., 1995).

Initial analysis of transgenic embryos at the 2-cell and heat-shocked blastocyst stages showed a statistically significant, copy number-dependent stimulation of transgene expression in SAR+ lines relative to SAR− lines (Thompson et al., 1994). Further analysis of constitutive expression at the 4-cell, 8-cell and blastocyst stages is presented in Fig. 2. At the 4-cell stage there was a drop in transgene expression as observed for the HSP70.1 endogene. In SAR− lines, the basal constitutive level was attained with the exception of the line F29, which continued to express at a slightly elevated level. All SAR+ lines exhibited absolute expression above the basal level, this being particularly true of the line SF4 in which expression was about half of that observed during the early 2-cell stage. However, at the 4-cell stage, correlation of expression with copy number was lost in SAR+ lines and on a per copy expression basis, only the line SF4 showed a clear difference when compared to SAR− lines. At the 8-cell stage, all lines had attained the basal transcriptional level. The SAR+ line, SF4, retained a slightly elevated absolute expression level but showed no difference on a per copy basis. At the blastocyst stage, low constitutive levels similar to those at the 8-cell stage were observed, though if the substantial increase in cell number, and therefore transgene copy number, is taken into account, basal levels per transgene copy were much lower. When blastocysts were heat-shocked the stimulatory effect of SAR sequences was again observed as was a positive correlation with copy number in SAR+ lines.

When present in culture medium, TSA, a specific inhibitor of histone deacetylases, causes hyperacetylation of core histones, a condition usually associated with transcriptionally active chromatin (Hebbes et al., 1988; Jeppesen and Turner, 1993). The strategy used to examine the stage-specific effect of histone hyperacetylation is outlined in Fig. 1. Embryos were exposed to TSA towards the end of a given cell cycle and expression was analysed during the following cell cycle. The results are summarized in Fig. 3 and the statistical significance is assessed in Table 1. First, we compared the effects of TSA at 0 nM, 30 nM, 75 nM, and 150 nM concentrations. Variability in response was observed among different lines, but for most a near maximal expression signal was obtained at 75 nM.

In some lines a non-linear increase in signal was observed at 150 nM relative to 75 nM but in others the response remained flat or actually decreased. Therefore, 75 nM was selected as the working concentration for further experiments.

A clear stimulatory effect of TSA was observed on constitutive expression at the 4-cell, 8-cell and blastocyst stages. Absolute levels of expression were much higher in SAR+ lines with mean values of 4,694, 947, and 447 RLU at the 4-cell, 8-cell, and blastocyst stages, respectively, compared to mean values of 293, 183, and 75 RLU for the SAR− lines. It was also informative to compare the expression observed in the presence of TSA with that obtained in its absence at each cleavage stage and for each transgenic line (Fig. 3). At the 2-cell stage there was no clear effect of TSA on transgene expression, and no obvious differences in the response of SAR+ and SAR− lines, with mean relative stimulation values of 1.4 and 0.82 respectively. At the 4-cell stage, hyperacetylation of histones led to significant stimulation of transgene expression in all lines except F2. The mean stimulation relative to untreated embryos in SAR− lines was 24-fold, similar to the mean 29-fold increase in SAR+ lines. At the 8-cell stage there was a significant increase in gene expression in all lines when embryos were incubated in TSA. Here however, there was a divergence in the response of SAR+ lines, with a mean 160-fold increase in expression, compared to the mean 10-fold stimulation observed in SAR− lines. At the blastocyst stage there was again a uniformly significant stimulation of transgene expression; SAR+ lines dropping back to the mean 29-fold stimulation observed at the 4-cell stage, while in SAR− lines the mean increase was reduced to only 3.4-fold.

Thus, at the 2- and 4-cell stages both SAR+ and SAR− lines showed a similar response to incubation in the presence of TSA. SAR− lines attained a maximal relative response at the 4-cell stage whereas in SAR+ lines, the maximal relative response occurred one cleavage cycle later, at the 8-cell stage. A further distinction between SAR+ and SAR− lines is illustrated in Fig. 4, where the luciferase synthesis by 4-cell and 8-cell embryos in the presence of TSA is compared to the synthesis by control 2-cell embryos, on a per transgene copy basis. All SAR+ lines showed an increased synthetic capacity in the TSA-treated 4-cell embryo relative to the 2-cell control embryo, the difference being significant in all but the line SF3. In contrast, among SAR− lines, only F29 showed increased synthesis, while F31 and F2 exhibited significantly reduced levels of synthesis. At the 8-cell stage, TSA was unable to stimulate synthesis to levels observed in the 2-cell embryo in any transgenic line.

The pattern of histone H4 acetylation was then investigated during the preimplantation period and in transgenic fibroblasts isolated from foetuses 12-13 days post coitum. Two different antibodies were used, one which recognized all acetylated forms of H4 and one which was specific for H4 acetylated at lysine-5. Both control and TSA-treated transgenic embryos were examined, using incubation protocols identical to those used in the analysis of luciferase expression. Results using ethanol-acetic acid fixation are shown for the lysine-5-specific antibody, in Fig. 5. At all stages, treatment with TSA increased the intensity of the immunofluorescent signal, though the increase was less significant in blastocysts and foetal fibroblasts. It is important to note that the paired control and treated images shown in Fig. 5 are not taken at the same brightness and contrast settings on the confocal microscope. This is in order to avoid barely visible control images, or alternatively, oversaturated TSA images, in the earlier embryonic stages. Digital analyses of signal intensity on images acquired at identical contrast and brightness settings are presented in Fig. 6.

In 2- and 4-cell control embryos a more or less even granular staining was observed throughout the nucleus with a varying degree of increased staining at the nuclear periphery. In one quarter of these embryos no increase in fluorescence was observed at the nuclear periphery. The tendency to increased peripheral staining was much reduced in control 8-cell embryos and blastocysts and was rarely observed in foetal fibroblasts. When incubated in the presence of TSA, all 2- and 4-cell embryos showed a marked increase in the immunofluorescent signal at the nuclear periphery (compare Fig. 6A,F and B,G) with no obvious increase in staining elsewhere in the nucleus. In 8-cell embryos, the change in immunofluorescence pattern induced by TSA showed less contrast between the periphery and more central portions of the nucleus, because of increased staining in the nuclear interior (Fig. 6C,H). This difference, relative to 2- and 4-cell embryos, was confirmed by optical sectioning throughout nuclei from these developmental stages (data not shown). The peripheral pattern of staining in 2- and 4-cell embryos was not due to insufficient TSA concentrations. Increasing TSA to 300 nM did not alter the staining pattern, though it did damage nuclear morphology in 2-cell embryos, resulting in numerous invaginations and distortion of the normal rounded shape. Contrast between staining of the nuclear periphery and interior was further reduced in blastocysts and foetal fibroblasts (compare Fig. 6D,I and E,J). Staining patterns very similar to those described above were obtained at each stage, with the antibody capable of recognizing all forms of acetylated H4, the only essential difference being that at identical confocal brightness and contrast settings, the staining was more intense with this reagent for both control and TSA treated samples (data not shown). Similar immunofluorescence patterns were also obtained using a paraformaldehyde fixation protocol, although in control 2-, 4-, and 8-cell embryos, a uniform nuclear staining was always observed with no increased intensity at the nuclear periphery. When embryos were incubated in the presence of TSA, and then fixed with paraformaldehyde, the evolution in the pattern of peripheral staining relative to the nuclear interior was the same as that observed with alcohol fixation. However, overall nuclear morphology was not as well preserved in paraformaldehyde-fixed embryos and the immunofluorescence pattern was less finely detailed than that observed after alcohol fixation. Preimmune serum gave a low nonspecific background with a negligible signal at the contrast and brightness settings used for acquisition of the control and TSA images (data not shown).

A possible explanation for reduced expression of the HSP70.1 endogene, and the SAR− and SAR+ transgenes, at the 4-cell stage, could be that an open chromatin structure is maintained at these loci, but that there is a reduction in the nuclear concentration of specific transcriptional activators present at the 2-cell stage. The stimulation of transgene expression in the presence of TSA might then be attributed to a stimulation of the synthesis of these activators. To test this hypothesis, microinjection experiments were performed as shown in Fig. 7. A supercoiled plasmid containing the HSP70.1 promoter linked to the luciferase cDNA was injected at 50 ng/μl. When one pronucleus of 1-cell embryos was microinjected, an increase in luciferase activity was observed through to the 2-cell stage with a subsequent drop at the 4-cell stage. However, when one nucleus of 2-cell embryos or two nuclei of 4-cell embryos were micoinjected with the plasmid construct, luciferase activity was easily detected at the 4-cell stage. Therefore, transcription factors, capable of activating gene expression from the HSP70.1 promoter were not limiting in the 4-cell mouse embryo.

SAR sequences contain multiple consensus sequences for topoisomerase II cleavage and the 5′ μ-interferon SAR contains 31 such repeats (Bode and Maass, 1988). It has been proposed that SARs might stimulate gene expression through topoisomerase II-mediated reduction of the torsional strain generated in the DNA template by an actively transcribing RNA polymerase (Cockerill and Garrard, 1986). To study whether the SAR effect observed in 2-cell embryos and heat-shocked blastocysts might be explained by such a mechanism, the topoisomerase II inhibitor teniposide VM-26 was used. At 10 μM VM-26, embryo cleavage was arrested, in conformity with the requirement for topisomerase II activity in chromatid disjunction and chromosome segregation at anaphase (Uemara et al., 1987; Holm et al., 1989). Therefore, the protocol used for TSA could not be used with this inhibitor during the early cleavage stages, and embryos that had recently cleaved from the 1-cell to 2-cell stage were picked off and transferred to M16 medium containing 10 μM VM-26. No significant reduction in SAR-flanked transgene expression was observed (Table 2). In blastocysts, heat shocked in the presence of VM-26, the line SF3 showed a slight increase in transgene expression, with no effect in the other 4 SAR+ lines. These results suggest that recruitment of topoisomerase II is unlikely to explain the increased expression observed in SAR+ lines at the 2-cell stage and in heat-shocked blastocysts.

The high mobility group HMG-I/Y proteins are expressed at elevated levels in profilerating cells and in those that are relatively undifferentiated (Vartiainen et al., 1988; Johnson et al., 1990). They are found in AT-rich regions of human and mouse metaphase chromosomes (Disney et al., 1989; Saitoh and Laemmli, 1994) and stimulate in vitro transcription from SAR-associated templates in the presence of histone H1 (Zhao et al., 1993). We examined the preimplantation expression profile of HMG-I/Y by RT-PCR and its cellular localisation by immuno-fluorescence.

The temporal expression pattern of HMG-I/Y during preim-plantation mouse embryogenesis is shown in Fig. 8. High mRNA levels were present in the oocyte, followed by a decrease up to the 8-cell stage, where HMG-I/Y mRNA was never detected. The mRNA was then again observed in blastocysts. The high levels in oocytes suggest that the mRNA detected up to the 4-cell stage was of maternal origin while that present in blastocysts was transcribed from the zygotic genome.

The pattern of immunofluorescent staining for the HMG-I/Y protein in preimplantation embryos and foetal fibroblasts is shown in Fig. 9. In the 1-cell embryo, HMG-I/Y was found in both male and female pronuclei, with more predominant staining in the nucleolar region. In a majority of 1-cell embryos, there was also a localised area of more intense staining along part of the cell periphery. The polar body was brightly immunofluorescent in all embryos. Staining was essentially nuclear in 2-cell embryos. In 4-cell embryos, nuclear staining was somewhat reduced and there was a variable intensity of immunofluorescent signal in the cellular cortex. At the 8-cell stage, nuclear, and overall staining were greatly reduced. In blastocysts, large numbers of HMG-I/Y foci were seen in the majority of nuclei in all embryos.

Treatment of embryos with TSA increased nuclear HMG-I/Y staining in 2-cell, 4-cell, and 8-cell embryos. The change in signal relative to that seen in control embryos was most strongly evident at the 8-cell stage. At the blastocyst stage, there was no obvious increase in the nuclear HMG-I/Y signal in treated versus control embryos though it did appear that the average size of the nuclear foci was reduced in embryos incubated in the presence of TSA. A similar effect was observed in foetal fibroblasts; TSA did not increase overall staining intensity but did appear to reduce the average size of the nuclear foci (Fig. 9J,K). In aged foetal fibro-blasts, where mitotic figures were extremely rare in the population, the great majority of cells exhibited large nuclear foci of HMG-I/Y. The preimmune serum gave a low nonspecific background with a negligible signal at the contrast and brightness settings used for acquisition of the images in Fig. 9 (data not shown).

Formation of the mammalian zygote involves the union of two specialized cells, the oocyte and the sperm. Prior to breakdown of the germinal vesicle, the oocyte nucleus is the site of active transcription of maternally stocked mRNAs. The sperm nucleus, on the other hand, is designed for compact delivery of the male’s genetic contribution. Thus, the early stages of embryogenesis are a unique period of extensive chromatin remodelling of the zygotic nucleus, prior to the complex series of cellular and nuclear differentiations that occur during subsequent development. In early Xenopus development, the importance of chromatin structure in transcriptional silencing prior to the MBT has been demonstrated (Prioleau et al., 1994; Almouzni and Wolffe, 1995), though the latter study also implicates deficiency in the activity of transcriptional activators.

To investigate chromatin structure, molecular ap-proaches such as DNA foot-printing, analysis of nucleosome positioning, and more recently, chromatin immuno-precipitation (Gould et al., 1990; Orlando and Paro, 1993) have been developed. Although the prerequisite isolation of nuclei in most of these protocols may lead to some artefactual repositioning of nuclear proteins (Käs et al., 1993), these strategies provide important information about the organisation of protein complexes on DNA regulatory regions which control gene expression. The abundant embryonic material available at the beginning of development in model organisms such as Drosophila, and Xenopus, make them amenable to the use of such techniques for probing early chromatin organisation.

The mammalian embryo differs from the above model organisms in that the early cleavage phase is already characterized by regulated cell cycles with specific zygotic gene expression. Mammalian embryologists, however, are confronted with the problem of a considerably more limited experimental resource. Nonetheless, using alternative approaches, advances have been made in beginning to define aspects of chromatin structure involved in regulating gene expression in early mammalian embryos. A qualitative study showed that very early mouse embryos are permissive for gene expression from a variety of microinjected promoter-reporter constructs and suggest that there are few restrictions on the use of promoter sequences in 2-cell embryos (Bonnerot et al., 1991). DePamphilis and colleagues have specifically focussed on the interplay between promoter and enhancer elements in the regulation of gene expression from episomal templates in 1-, and 2-cell mouse embryos. They have shown that enhancers are dispensable in S-phase arrested, 1-cell embryos, but are required to prevent repression of weak promoters in 2-cell embryos (Majumder et al., 1993). Butyrate stimulates expression from plasmids injected into maternal pronuclei but not from those injected into paternal pronuclei. After passage to the 2-cell stage, butyrate increases plasmid expression in the zygotic nucleus (Wiekowski et al., 1993). They suggest that chromatin of 1-cell embryos is deficient in H1, H2A and H2B, and enriched in hyperacetylated core histones, while in 2-cell embryos a mature chromatin structure is attained with a reduction in histone hyperacetylation (Majumder and DePam-philis, 1994).

To create an integrated model that would allow analysis of the effects of modifications of chromatin structure on early gene expression, we established transgenic mouse lines in which the luciferase reporter was coupled to the murine HSP70.1 promoter. The 800 bp HSP70.1 promoter contains a TATA box, an inverted CCAAT box, two binding sites for the transcription factor Sp1, one binding site for the transcription factor AP2, and four tandemly arranged heat shock elements for the binding of specific heat shock transcription factors (HSFs) (Hunt and Calderwood, 1990). Both the SAR− and SAR+ HSP70.1 transgenes were constitutively expressed in the 2-cell embryo. The HSE elements do not appear to play a role in 2-cell constitutive hsp70 expression, as gel retardation assays show no evidence of bound complexes (Mezger et al., 1994). However, it has recently been shown that the concentration of Sp1 decreases during oocyte growth and then increases in nuclear concentration upon fertilization and continues to increase through to the 2-cell stage (Worrad et al., 1994). Therefore, it is likely that Sp1 binding sites play a more important role than HSE elements in constitutive 2-cell hsp70 expression. When core histones were hyperacetylated by incubating embryos in the presence of the histone deacetylase inhibitor, TSA, there was no effect on constitutive transgene expression at the 2-cell stage in either SAR+ or SAR− lines.

Thus, nucleosomes are either positioned such that they do not interfere with access to promoter binding sites, or they are already sufficiently acetylated in the absence of TSA. In our experimental protocol, the transgenes were contributed via the sperm, and this result is consistent with the idea that chromatin remodelling of the sperm nucleus may involve hyperacetylated histones (Wiekowski et al., 1993). We were unable to compare the effect of TSA on the expression of transgenes contibuted by the female, as very high transgene expression during oogenesis masked early zygotic expression.

At the 2-cell stage, expression from SAR+ lines was significantly elevated relative to that observed in SAR− lines. The topoisomerase II inhibitor, VM-26, did not reduce SAR+ transgene expression suggesting that the increased luciferase activity observed in these lines was not due to recruitment of topoisomerase II to SAR sequences in order to relieve torsional strain associated with transcription. However, consistent with the in vitro observation that HMG-I/Y stimulates transcription from templates containing SAR sequences (Zhao et al., 1993), HMG-I/Y was present in the nucleus of 2-cell embryos. HMG-I/Y has been shown to bind to preferred regions on the front face of core nucleosomes (Reeves and Nissen, 1993) and stimulates transcription from the human μ-interferon promoter through DNA bending and synergistic interactions with the transcription factors NF-kB, and activating transcription factor 2 (ATF-2) (Thanos and Maniatis, 1992; Du et al., 1993). Immunolocalisation studies show that HMG-I/Y is associated with AT-rich regions of metaphase chromosomes (Disney et al., 1989; Saitoh and Laemmli, 1994) and our results show concentrations of HMG-I/Y in discrete foci in interphase nuclei of mouse embryos and foetal fibroblasts. Should the AT-rich sequences of SARs serve to increase the local concentration of HMG-I/Y, this would favour the type of interactions described above, and could account for the increased expression observed in the SAR+ lines at the 2-cell stage. While the affinity of HMG-I/Y for the minor groove of AT-tracts is well established (Solomon et al., 1986; Fashena et al., 1992), we are at present, technically unable to provide direct molecular evidence linking this protein specifically to the SAR+ transgene loci in the early mouse embryo.

A major event in the 4-cell mouse embryo is the arrival of somatic histone H1 (Clarke et al., 1992). At the 4-cell stage, SAR− transgene expression was reduced to basal levels. SAR+ expression was also considerably reduced but remained above basal levels, particularly in the line SF4. HMG-I/Y transcripts were less abundant at the 4-cell stage relative to the 2-cell stage suggesting a depletion of maternal stocks. The HMG-I/Y protein was still present in the nucleus, but at lower density, and also showed a more or less intense staining in the cellular cortex of 4-cell embryos. Thus, the equilibrium between histone H1 and HMG-I/Y changes in the nucleus of the 4-cell embryo in favour of H1. In response, SAR+ expression was reduced but did not attain the basal levels observed in SAR− lines.

In the 4-cell embryo, core histone hyperacetylation had a strong positive influence on transgene expression. Reduced constitutive expression of the transgene at the 4-cell stage might be interpreted as being caused by a decrease in the concentration of HSP70.1-specific transcription factors. The effect of TSA could then be explained as a general stimulation of the synthesis of these factors. However, microinjection of HSP70.1-luciferase plasmid templates into 2-cell and 4-cell nuclei demonstrated that transcription directed by the HSP70.1 promoter was not limited by transcription factor concentration. Taken together, these results indicate that nucleosome positioning had an important repressive effect on transgene expression at the 4-cell stage, which could be relieved by hyperacetylation of core histones. It is important to note however, that global levels of acetylation did not appear to change significantly between the 2-cell and the 4-cell stages in control embryos (Figs 5 and 6). This suggests that finer regional controls are already operational.

By the end of the 8-cell stage, histone H1 has been present for one full cell cycle, and all transgenic lines were reduced to a basal level of expression. HMG-I/Y was present in very low amounts in the nuclei of 8-cell embryos and transcripts could no longer be detected by RT-PCR. At this cleavage stage, the response of the two types of transgenic construct to core histone hyperacetylation was quite different, with a mean relative stimulation in SAR+ lines of 160-fold compared to a mean response of only 10-fold in SAR− lines. Incubation in the presence of TSA also resulted in a considerable increase in the nuclear concentration of HMG-I/Y. However, TSA was now unable to stimulate per copy transgene expression in 8-cell embryos to levels observed in 2-cell embryos in any line. Thus, chromatin structure continued to mature, with the presence of SAR sequences appearing to delay final repression of the transgene loci to basal expression levels by one cleavage cycle.

In blastocysts, basal expression was observed in all lines with per copy expression much lower than at the 8-cell stage. The effect of TSA was diminished compared to that observed in 8-cell embryos, though it continued to be more important in SAR+ lines. HMG-I/Y transcripts were again detected and the protein was present in large quantities in blastocyst nuclei. Upon heat shock of blastocysts, the stimulatory effect of SAR sequences was recovered. At the blastocyst stage, heat shock causes the formation of a protein complex on the HSE elements suggesting this as a major regulatory event in the induced response (Mezger et al., 1994). It is probable that heat shock results in the phosphorylation of an HSF complex (Morimoto, 1993), which then binds to HSE elements in the promoter. This could be considered as an enhancer factor binding to the promoter which then in cooperation with SARs, in the presence of HMG-I/Y, extended an open chromatin structure and resulted in increased levels of expression. Such a mechanism of SAR action has been proposed in other experimental systems (Forrester et al., 1994).

Thus, we propose the following series of events in the chromatin regulation of HSP70.1-luciferase transgene expression. In the 2-cell embryo, nucleosomes are positioned or acetylated such that key promoter elements, likely the Sp1 sites, are available to transcription factors. Upon passage to the 4-cell stage, local changes in nucleosome positioning or acetylation status block transcription factor access to the promoter. Hyperacetylation of core histones at the 4-cell stage permits per copy expression levels that, in the case of the SAR+ lines, exceed that observed in the 2-cell embryo. The arrival of histone H1 during the 4-cell stage then completes the organisation of a chromatin structure that reduces all transgenic lines to a basal level of expression at the 8-cell stage. At this point, hyperacetylation of core histones alone is no longer sufficient to increase per copy expression levels to those attained at the 2-cell stage in any transgenic line. Therefore, these results indicate that a mature chromatin structure is not attained until the 8-cell stage rather than at the 2-cell stage as previously proposed (Majumder and DePamphilis, 1994).

The stimulatory effect of SAR sequences was very well correlated with the presence of HMG-I/Y in embryonic nuclei. Stimulation of transgene expression in SAR+ lines was clearly evident at the 2-cell stage, decreased at the 4-cell stage, when the equilibrium between histone H1 and HMG-I/Y was shifting, and was no longer observed at the 8-cell stage when HMG-I/Y protein levels were very low and transcripts were undetectable. The stimulation of SAR+ expression by TSA at the 8-cell stage also corresponded to a strong increase in nuclear HMG-I/Y content. At the blastocyst stage, the presence of HMG-I/Y alone was not sufficient to increase constitutive expression in SAR+ lines, suggesting that the chromatin conformation at this stage prevented access of HMG-I/Y to the SARs. However, upon heat shock, the locus was activated through binding of HSF to HSE elements in the promoter, and cooperative binding of HMG-I/Y to AT-tracts may have led to an extended region of open chromatin, possibly accounting for the stimulatory effect of SARs.

There are also indications of a more global modification of chromatin and nuclear structure during the 2-to 4-to 8-cell transition. At the 2- and 4-cell stages, treatment with TSA resulted in a strong increase in histone H4 acetylation at the nuclear periphery but not in the nuclear interior. This might suggest that the highest nuclear histone acetyltransferase activities are confined to the nuclear periphery during these stages. At the 8-cell stage, TSA still increased acetylated H4 staining most strongly at the nuclear periphery but there was also a significant increase in staining in the nuclear interior. At the blastocyst stage and in foetal fibroblasts, TSA no longer strongly increased staining at the nuclear peripery, and contrast with staining in the nuclear interior was further reduced. Interestingly, inhibition of histone deacetylases by butyrate or TSA is unable to cause the accumulation of hyperacetylated histone H4 in early cleavage stage Xenopus embryos until after the ZGA at the MBT (Dimitrov et al., 1993; Almouzni et al., 1994). In the mouse, we observed a capability to hyperacetylate histone H4 at least as early as the 2-cell stage, again coincident with the major ZGA, and demonstrated a locus specific transcriptional effect as early as the 4-cell stage. This suggests that the ability to modulate core histone acetylation levels might be an important prerequisite to a regulated program of zygotic gene expression.

Further early chromatin modifications are suggested by the preimplantation profile of the nuclear protein HMG-I/Y. HMG-I/Y was present in 1-, 2- and 4-cell embryos, but virtually absent in 8-cell embryos. The protein was again present at high levels in blastocyst nuclei. However, two differences with respect to HMG-I/Y were observed in blastocysts compared to early cleavage stages; the protein to mRNA ratio was higher in blastocysts, suggesting modification of HMG-I/Y mRNA metabolism, and where TSA increased the amount of HMG-I/Y protein in the nuclei of early cleavage stage embryos, there was no obvious TSA-stimulated increase in blastocyst nuclei. Instead, TSA reduced the average size of HMG-I/Y foci in blastocyst nuclei. This same effect was observed in foetal fibroblasts.

The proposition of a maturing chromatin structure through to the 8-cell stage also corresponds with another important developmental event. This is a major genome-wide demethylation that occurs between the 8-cell stage and the blastocyst in the mouse (Monk et al., 1987). The DNA of blastocysts, where the first cellular differentiation occurs, is reported to have the lowest levels of 5-methylcytosine of any tissue or developmental stage, apart from primordial germ cells (Chaillet et al., 1991). This global demethylation has also been shown to occur specifically for an imprinted transgene (Chaillet et al., 1991) and for several endogenous genes (Kafri et al., 1992; Brandeis et al., 1993). As methylation is usually, though not always, associated with reduced levels of gene expression, it is possible that higher levels of methylation during very early cleavage stages would reduce inappropriate gene expression in the presence of an immature chromatin structure. Once a basic chromatin structure is established at the 8-cell stage, demethylation can proceed in preparation for a precise, developmentally regulated program of gene expression.

In mammals, there is no evidence of vegetal or animal poles as in Xenopus or bicoid-like gradients found in Drosophila which have a direct influence on early patterns of gene expression. Instead, in the mouse, our results suggest a series of early cleavage cycles, ending in the formation of a basic chromatin structure at the 8-cell stage upon which a program of cellular differentiation can then be enacted. These results also support the view of the involvement of DNA sequences such as SAR elements in mediating chromatin accessibility and provide a rationale for the observation that they are found flanking some developmentally regulated or inducible genes. Upon activation of a locus via promoter or enhancer elements, flanking SARs can, through cooperative binding of proteins with affinity for AT-tracts, such as HMG-I/Y, extend a region of chromatin opening and rapidly stimulate gene expression. When a return to basal levels of expression is signalled, the affinity of SARs for histone H1 would then assure a return to a closed chromatin conformation. The use of such mechanisms would permit finer temporal regulation of the transcription of certain developmentally expressed and inducible genes.

We thank R. Reeves for the anti HMG-I/Y antibody, D. Allis and B. Turner for anti-histone H4 antibodies, R. Morimoto for the HSF2 cDNA, M. Yoshida for a sample of trichostatin A, and Sandoz laboratories for a sample of VM-26. We are grateful to P. Adenot for an introduction to confocal microscopy and F. Fort for help with the photographs. This work was supported by a grant from the Ministère de la Recherche et de la Technologie and Rhône-Merieux (Contract Rhône-Merieux-INRA-MRT 90T0968).