Epigenetic Inheritance of Active Chromatin After Removal of the Main Transactivator

See allHide authors and affiliations

Science  29 Oct 1999:
Vol. 286, Issue 5441, pp. 955-958
DOI: 10.1126/science.286.5441.955


The Drosophila Polycomb and trithorax group proteins act through chromosomal elements such as Fab-7 to maintain repressed or active gene expression, respectively. AFab-7 element is switched from a silenced to a mitotically heritable active state by an embryonic pulse of transcription. Here, histone H4 hyperacetylation was found to be associated withFab-7 after activation, suggesting that H4 hyperacetylation may be a heritable epigenetic tag of the activated element. ActivatedFab-7 enables transcription of a gene even after withdrawal of the primary transcription factor. This feature may allow epigenetic maintenance of active states of developmental genes after decay of their early embryonic regulators.

During early embryogenesis, spatially restricted expression patterns of Drosophilahomeotic genes are established by a regulatory cascade involving the products of segmentation genes (1, 2). Subsequently, Polycomb (PcG) and trithorax (trxG) group proteins maintain these patterns (3, 4) at the level of chromatin structure (5, 6) by acting through overlapping cis regulatory elements (7–13). We previously reported that one of these elements, the Fab-7 DNA from the bithorax complex (14–16), could be specifically switched into a repressing or an activating mode during embryogenesis (17). Once established, both regulatory modes are mitotically inherited throughout development and at a certain frequency through meiosis. For this reason, Fab-7 was termed a cellular memory module (CMM). The Fab-7–dependent transmission of a chromatin state permissive for gene expression through multiple cell divisions suggested that an active process is involved in the maintenance of open chromatin. Local hyperacetylation of histones H3 and H4 in fission yeast centromeres can be epigenetically inherited through mitosis and meiosis (18). Here, we tested the roles of PcG and trxG proteins in maintaining theFab-7 state and whether epigenetic control involves similar histone modifications in Drosophila.

Fab-7–dependent chromosomal memory of silent or open chromatin states was previously described in transgenicDrosophila lines such as FLW-1 and FLFW-1 (19). These lines carry a heat shock–inducible GAL4 driver (hsp70-GAL4) regulating a GAL4-dependent lacZ reporter (UAS-lacZ) flanked by Fab-7 and the mini-white gene. Silencing imposed by Fab-7 on the flanking reporter genes was dependent on the components of the PcG, as heterozygous mutant PcG genes showed a relief of white gene repression (Fig. 1A). Conversely, white gene activity required the trxG, as heterozygous mutations in the different members tested (Fig. 1A) resulted in a down-regulation of expression. A GAL4 pulse during embryogenesis can impose a mitotically stable reprogramming of the Fab-7 CMM from a silenced to an open chromatin state (17). The maintenance of the activated Fab-7 state was dependent on trithorax(trx) but not on Polycomb (Pc) (Fig. 1, B to D). In a heterozygous Pcbackground, Fab-7 can be switched by a GAL4 pulse and be stably maintained, resulting in strong white expression (Fig. 1C; HSE). In contrast, a trxmutation completely abolished the mitotic transmission (Fig. 1D).

Figure 1

Role of PcG and trxG genes inFab-7–mediated epigenetic inheritance. (A) The eye color of flies of the 5F24 25,2 line (19) was compared with that of flies carrying the same Fab-7 transgene insertion and heterozygous mutations in genes of the PcG or the trxG, as indicated. Darker eye color with respect to wild-type 5F24 25,2 was interpreted as derepression in the mutant background, whereas lighter eyes indicated repression. The relative strength of the effect is indicated on the right. (B toD) 5F24 25,2 embryos carrying one copy of hsGAL47-1 in the second chromosome and the heterozygous mutations PcXT109 or trxE2 were either grown at 18°C or activated by an embryonic pulse of GAL4 (HSE). The extent of transmission of the activated state was measured by comparing the eye color of the HSE with that of the 18°C F0 progeny 1 day after hatching. The trxE2 mutation prevented transmission of the Fab-7–activated state. At 18°C, all eyes show a similar color, because of background pigmentation caused by the mini-white gene present in the hsGAL47-1transgene. In its absence, PcXT109 flies show darker eyes, and trxE2 show lighter eyes than their wild-type (wt) siblings.

This antagonistic role of the PcG and trxG suggests a competitive interaction of the protein components regulating either the repressed or the active state of Fab-7. Indeed, counteracting PcG-mediated silencing by a transient GAL4 pulse during the third instar larval stage induced displacement of Polycomb (PC) protein from a transgene carrying Fab-7 in polytene chromosomes (Fig. 2, G to I) (9). Displacement of PC is accompanied by strong UAS-lacZ expression in salivary glands. However, after the disappearance of GAL4, PC protein was rebound toFab-7 and UAS-lacZ became repressed again (Fig. 2I). To assess whether the epigenetically activated Fab-7 state correlates with a permanent loss of PcG proteins from the chromatin template, we administered a strong GAL4 induction pulse at embryogenesis in the FLFW-1 line. Polytene chromosomes of third instar larvae were immunostained with antibodies directed against PcG proteins. Surprisingly, all of the PcG proteins tested, PC andPosterior Sex Combs (PSC) (Fig. 2, A to F),Polyhomeotic (PH), and Polycomb-like (PCL) (20), were still strongly bound to the Fab-7transgene irrespective of the epigenetic state. Thus, an epigenetically activated state can be stably propagated in the presence of the protein components of the PcG. These data support previous observations that demonstrated binding of PC at cytological sites containing potentially active genes in polytene chromosomes (21) and binding of PH and PSC proteins at an actively transcribed gene inDrosophila Schneider cells (22). It has been reported that certain PcG genes may function as activators in specific tissues and at specific developmental times by genetic analyses (23, 24). Although we did not observe a role for PC protein in the maintenance of the activated state ofFab-7 (see Fig. 1C), it may be possible that other PcG proteins are involved in this process.

Figure 2

Maintenance of the active state in the presence of PcG proteins. (A to F) Binding of the PcG proteins PC and PSC at Fab-7 at 61C9 in different functional states was analyzed. The position of integration of the Fab-7–containing transgene is indicated by a tick in the absence of PcG protein binding or by an arrow in the case of PcG binding. PC and PSC proteins colocalize in two endogenous bands at cytological position 61C1,2 and 61F2 (A and D). In the FLFW-1 line grown at 18°C, both proteins bind at Fab-7 (61C9) without GAL4 induction (B and E), as well as upon a GAL4 pulse at embryogenesis (HSE) (C and F). (G to I) A time course of PC displacement upon a GAL4 pulse in third instar larvae is shown. PC is bound at the transgene at 18°C (G), it is displaced upon a 45-min heat shock followed by 80 min of recovery at 18°C (H)(HS+80), and it reassociates to the transgene upon a heat shock pulse at mid third instar larvae followed by an overnight recovery at 18°C (I) (HS+ON).

If it is not the removal of PcG repressors on the template, what is the epigenetic tag that marks the activated Fab-7 state? A single embryonic GAL4 pulse was administered to FLFW-1 embryos, and histone acetylation of the Fab-7 transgene as a possible mark was analyzed by immunostaining polytene chromosomes (25) of third instar larvae with specific antibodies against the tetra-acetylated form of H4 and H3 histones (Fig. 3, A to H). Hyperacetylation of histone H4 was detected at the Fab-7 transgene location in larvae derived from activated embryos, but not from control embryos raised at 18°C (compare Fig. 3D with Fig. 3C). Similarly, no H4 hyperacetylation at Fab-7 was detected in control larvae that either did not carry the transgene (Fig. 3A, wt) or carried the target transgene in the absence of the hsp70-GAL4 driver (GCD-6 line;Fig. 3B). Furthermore, a line carrying the hsGAL4 driver and a control transgene with UAS-lacZ but no Fab-7 (19) did not show increased acetylation upon an embryonic pulse of GAL4 (26). In contrast to H4, no hyperacetylation of histone H3 could be detected in activated FLFW-1 individuals (Fig. 3, E to H).

Figure 3

Hyperacetylation of H4 marks epigenetically maintained open states of Fab-7. (A to H), Tetra-acetylated histone H3 and H4 were visualized on polytene chromosomes by immunostaining. Strains (19) and growth conditions were as indicated. The position of the Fab-7–containing transgene is indicated by a tick (when no histone hyperacetylation was detected) or by an arrow (in the presence of histone hyperacetylation). Hyperacetylation of histone H4 is observed at the transgene in larvae derived from embryos submitted to a GAL4 activation pulse (D) but not in the absence of GAL4 activation pulses (A to C). H3 hyperacetylation is not observed in any condition (E to H). (I to P) Time course of histone H4 hyperacetylation during GAL4-mediated activation by a heat shock pulse to third instar larvae. Immunostainings (I to L) and phase contrast (M to P) are shown. HS is a 45-min heat shock at 37°C; HS+40 and HS+80 are a 45-min heat shock followed by 40 or 80 min of recovery at 18°C, respectively.

To determine whether H4 histone acetylation also marks a difference between embryonically activated heritable Fab-7 states and larval transiently activated states, we analyzed H4 acetylation in a time course of GAL4-mediated transient transcriptional activation of the Fab-7 transgene in third instar larvae (Fig. 3, I to P). In contrast to the hyperacetylation observed after GAL4 activation at embryogenesis, H4 hyperacetylation of the transgene was observed only weakly and transiently (Fig. 3J) at the end of the heat shock induction of GAL4 in third instar larvae. Forty minutes after the end of the heat shock pulse, H4 hyperacetylation was no longer detected (Fig. 3, K and L), even though the presence of a small chromosomal puff (Fig. 3, O and P) and the recruitment of large amounts of RNA polymerase II indicate continuing transcription (27). Therefore, the absence of maintenance of the hyperacetylated states correlates with the failure to reprogram the chromatin of the transgene. This suggests that postembryonic tissues are more refractory to histone acetylation of silent Fab-7 templates.

Patterning transcription factors, like the products of many segmentation genes, act only shortly on their downstream genes during early Drosophila development, whereas the PcG/trxG memory system subsequently maintains the embryonically programmed patterns (3). For this reason, we tested whether embryonically activated Fab-7 can maintain expression of the reporter genelacZ in the absence of the primary transcription factor GAL4 (28). The leakiness of the hsp70 promoter prevented the complete disappearance of GAL4 protein during single fly development. To overcome this problem, we made use of the fact that activatedFab-7 can be efficiently propagated through meiosis in the line FLW-1 (17). This allowed us to cross out the GAL4 driver to test lacZ expression in the complete absence of GAL4 in subsequent generations (28). Upon crossing out GAL4 in activated flies, 20 to 25% of the GAL4-less embryos showed substantial levels of homogeneous β-galactosidase (β-Gal) expression in all embryonic cells in two consecutive generations (Fig. 4, D and E). This percentage correlates well with the fraction of adults showing meiotically stablewhite derepression (17). Unfortunately, it was not possible to also test for the meiotic inheritance of H4 hyperacetylated states because of a staining pattern with endogenous bands at the insertion site of the transgene in the FLW-1 line. However, the functional analysis demonstrates that epigenetic inheritance of an active Fab-7 chromatin state results in transcriptional activity of the UAS-lacZ reporter even in the absence of the GAL4 transactivator.

Figure 4

Inheritance of the active state of Fab-7 maintains expression of lacZ in the absence of the specific GAL4 transactivator. β-Gal staining of embryos is shown. (A) 5F24 25,2 line, carrying no GAL4 transgene. (B) Strong homogeneous expression is observed in the FLW-1 line upon a 45-min heat shock followed by 2 hours of recovery. (D and E) GAL4 cross-out F2 and F3 generations, derived from the FLW-1 line activated by a GAL4 pulse followed by crossing out of the GAL4 transgene (GAL4 cross-out F1) and recrossing either once (D) or twice (E). A weak but homogeneous β-Gal expression is observed in all tissues. Below each panel, the mean percentage of embryos showing the depicted pattern ± standard deviation of two independent experiments is indicated. (C and F) Control lines U/l5 1,1 (C) and U/l5 2,1 (F), which carry the p/Ul5 construct, containing neither Fab-7 nor GAL4. In these lines, most of the embryos are either not stained or show a very weak, variegated staining pattern. Only about 2% of the embryos showed a more homogeneous UAS-lacZ expression resembling the one in (D) and (E).

A weak expression of lacZ in the absence of GAL4 may arise from a heritable loss of PcG-mediated repression, thereby neutralizing the silencing ability of Fab-7 and consequently reflecting the ground state of a nonrepressed chromatin template. If this were the case, it may be expected that in flies carrying an UAS-lacZ construct without Fab-7 (pU/l5) (9,17), a similar weak homogeneous β-Gal staining pattern would be observed in all embryos in the absence of GAL4. To test this point, we analyzed β-Gal staining in two independent lines carrying the pU/l5 construct but no GAL4 driver: the U/l5 1,1 and U/l5 2,1 transgenic lines (9). In both cases, most of the embryos were not stained or were stained in a weakly variegated fashion in random cells (Fig. 4, C and F). This strongly suggests that meiotic inheritance of Fab-7 CMM-activated states does not simply reflect lifting of PcG-mediated silencing but rather the inheritance of an active chromatin state, which is competent for transcriptional activation.

In a recent report, Cosma et al. (29) measured the time and interdependence of transcription and chromatin remodeling factor recruitment to the yeast HO promoter. They found that the transcription factor Swi5p only fleetingly binds to the HO promoter before the recruitment of remodeling factors such as Swi/Snf (related to the Drosophila trxG) and SAGA take over. This suggests that Swi5p might only be transiently necessary for activation. Our results support and extend these findings by showing in a functional manner that trxG protein complexes recruited at a CMM relieve the requirement for the activating factor for transcriptional maintenance. We identify hyperacetylation of histone H4 as an epigenetic mark for the activated Fab-7 state. Unlike the short-lived H4 hyperacetylation induced by transient gene activation at late developmental stages, the mark set at embryonic stages is mitotically stable and inheritable. An important maintenance function of the PcG and trxG protein complexes at CMMs might be to protect epigenetic marks from erasure.

  • * To whom correspondence should be addressed. E-mail: paro{at}


View Abstract

Stay Connected to Science

Navigate This Article