Report

Chromatin Loosening by Poly(ADP)-Ribose Polymerase (PARP) at Drosophila Puff Loci

See allHide authors and affiliations

Science  24 Jan 2003:
Vol. 299, Issue 5606, pp. 560-562
DOI: 10.1126/science.1078764

Abstract

Steroid response and stress-activated genes such ashsp70 undergo puffing in Drosophila larval salivary glands, a local loosening of polytene chromatin structure associated with gene induction. We find that puffs acquire elevated levels of adenosine diphosphate (ADP)-ribose modified proteins and that poly(ADP)-ribose polymerase (PARP) is required to produce normal-sized puffs and normal amounts of Hsp70 after heat exposure. We propose that chromosomal PARP molecules become activated by developmental or environmental cues and strip nearby chromatin proteins off DNA to generate a puff. Such local loosening may facilitate transcription and may transiently make protein complexes more accessible to modification, promoting chromatin remodeling during development.

Cells within developing multicellular eukaryotes build complex tissue-specific chromatin architectures to express certain genes and silence others (1). The enzyme PARP is thought to play a critically important role in preserving differentiated chromatin during DNA repair (2). The protein's zinc fingers specifically recognize DNA damage, and PARP activity increases strongly upon binding to such sites. The activated enzyme modifies nearby chromatin proteins with ADP-ribose moieties, disrupting their macromolecular complexes, and causing the affected chromatin to decondense. The newly repaired region returns to a normal state after PARP down-regulates its own activity via automodification, and the chromatin proteins, freed of ADP-ribose groups by a specific glycohydrolase, reassemble. Recent genetic studies with Drosophila melanogaster show that PARP is an essential gene required to organize chromatin throughout the life cycle (3). However, the relation between PARP action during repair and during development remains unclear.

Drosophila chromatin normally undergoes many programmed changes that could be mediated by PARP (1). In particular, chromatin alterations manifest as salivary gland polytene chromosome puffs occur at the sites of ecdysone response genes before molting (4). To look for specific loci where PARP may act, we surveyed the levels of PARP protein, ADP-ribosylation activity, and ADP-ribose polymers during embryonic and larval development. Epitope-tagged PARP protein is widely distributed throughout the euchromatin of both diploid and polyteneDrosophila nuclei (Fig. 1, A and B). However, the activity state of these PARP molecules varies widely because injected biotinylated nicotinamide adenine dinucleotide (NAD), the source of ADP-ribose adducts, differentially labels only a limited number of polytene chromosome regions, and this labeling is blocked by the PARP inhibitor 3-aminobenzidine (3-AB) (Fig. 1C). ADP-ribose polymers detected with a specific antibody are also found only within particular chromosome bands, including early ecdysone puff loci (Fig. 1D). For example, in the region surrounding the 75A-75B puffs, PARP protein is widespread (Fig. 1E), but high levels of poly(ADP-ribose) only appear at the time of puffing, (Fig. 1F). These observations suggest that PARP becomes active during puff induction and modifies local proteins, leading to an accumulation of ADP-ribose moieties. PARP activation in ecdysone puffs may be functionally important because Parp mutant larvae frequently arrest at ecdysis (3).

Figure 1

Differential accumulation of poly(ADP-ribose) in polytene chromosome puffs. DNA shows as green in (A) through (D), red in (E) and (F). PARP-DsRed (red) (3) is widespread in the euchromatin (arrowheads) of (A) larval diploid brain cells (B) and polytene salivary gland cells as well as in nucleoli (arrows). (C) PARP activity, indicated by incorporation of biotinylated-NAD (red) 1 hour after larval injection, is high at certain sites (arrowhead) and in the nucleolus (arrow). Incorporation is due to PARP because it is abolished by injection of 3-AB (lower panel) 30 min before assay. (D) Poly(ADP-ribose) polymer (red) is enriched at specific polytene chromosomes sites, including puffs. Note early ecdysone-induced puffs at 2B (arrow) and 75A and 75B (arrowheads). (E) PARP-GFP (3) is present throughout this chromosome region (green). (F) However, poly(ADP-ribose) (green) is not found at 75A or 75B before puffing (left) but is found after puffing (right) (fig. S1).

To determine whether puff formation requires PARP, we studied stress-induced puffs. A heat shock strongly induces loci containing stress response genes to puff (5), including the clustered genes encoding Hsp70 chaperone at 87A and 87C (Fig. 2A). Before heat shock, PARP is widespread along polytene chromosomes (Fig. 2B), but little poly(ADP-ribose) is present at the 87A, 87C, and surrounding loci (Fig. 2C). In contrast, after a brief heat shock, ADP-ribose polymers accumulate differentially throughout the large, newly formed puffs. After 25 min of recovery, the amount of poly(ADP)-ribose begins to fall, and the puff itself begins to regress. These observations reveal a strong correlation between ADP-ribose accumulation and puffing.

Figure 2

Parp is required for heat shock–induced puffing and hsp70 expression. (A) The 87A and 87C polytene chromosome region before (left) and after (right) a 30-min heat shock (37°C); large, induced puffs contain Hsp70 chaperone genes. (B) PARP-GFP is widespread in the 87A and 87C puffs (arrows) and in the surrounding chromosome region. (C) However, poly(ADP-ribose) polymers are sparse before heat shock (0'), but are abundant in the induced puffs after a 20-min heat shock followed by 10 min at 25°C (30'). Fifteen minutes later (45)' the amount of poly(ADP-ribose) has decreased. (D) Puffs at 87A and 87C are absent in larvae fed the PARP inhibitor 3-AB for 1 hour before heat shock. (E) Heat shock puff size in second instar larval salivary glands measured as the area (in arbitrary units) labeled by hybridization in situ with an hsp70 probe in excess of controls that were not heat shocked. (F) A histogram shows that 87A and 87C puff sizes average three times as large as in wild-type (red) compared withParpCH1 mutant (black) cells. (G). A Western blot shows that 5 to 10 times more Hsp70 protein is produced in wild-type (wt) compared with ParpCH1 larvae when normalized to Actin 5C levels (fig. S2).

To determine if the Parp gene encodes the enzyme responsible for elevating ADP-ribose levels during puffing and if this increase is functionally important, we asked whether heat shock–induced puffs arise normally in Parp mutant larvae.Parp-defective animals die during the second instar, when salivary gland polytene chromosomes are too small to analyze cytologically (3). However, we found that puffs in these animals can be quantitated by in situ hybridization with anhsp70 probe (Fig. 2, E and F). The average size of the 87A and 87C puffs is reduced threefold in Parp mutant larvae. Complete loss of PARP likely has an even bigger effect, as the mutant animals continue to express low levels of PARP (6). Consistent with this expectation, puff formation at 87A and 87C was completely blocked in wild-type third instar larvae that were fed 3-AB for 1 hour (Fig. 2D), and these larvae failed to recover from heat shock. Western blots showed that 5 to 10 times less Hsp70 protein was induced by heat shock in mutant compared with wild-type larvae (Fig. 2G). Thus, PARP is needed to form normal heat shock puffs, to express normal levels of puff-encoded proteins, and to normally resist the deleterious effects of heat stress.

The association of PARP activity with developmental puffs and its requirement for heat shock–induced puffs suggests that some genes undergo chromatin loosening during induction. Immune response genes may belong to this group because Parp1knockout mice display dramatic immune defects and cannot induce genes controlled by nuclear factor kappa B (NF-κB) transcription factors (7). We found that Parp mutantDrosophila larvae frequently develop intracellular bacterial infections (Fig. 3A), a condition rarely observed in wild type. Spontaneous infection correlates with low PARP levels, because in larval tissues retaining variable amounts of PARP only cells with little or no PARP function become infected (Fig. 3B). Furthermore, Parp mutant larvae are sensitive to experimental infection. Ninety-five percent died after a dose of injected Escherichia coli bacteria that killed less than 7% of size-matched wild-type controls. Thus, like Parp1-deficient mice,Drosophila lacking normal PARP levels display immune defects.

Figure 3

PARP is required to express innate immunity genes. (A) A single ParpCH1 larval cell is shown with clusters of intracellular rod-shaped bacteria (Oligreen staining of DNA). (B) ParpCH1 brain cells that retain some PARP activity (enclosed by dashed region) contain nucleoli that stain with antibodies to Fibrillarin (red) and are free of intracellular parasites (arrows). (C) Western blot using epitope-specific antibodies quantitating NF-κB–dependent innate immunity gene expression (Diptericine-LacZ and Drosomycin-GFP) 40 hours after injection with a sublethal dose of E. colibacteria (+); control indicated by (–). Both immunity genes are strongly induced by infection (relative to Actin 5C) in wild-type (wt) but not in Parp mutant larvae (12).

Inducible innate immunity genes constitute a major resistance mechanism to microbial infection in insects such asDrosophila (8). The rapid production of antibacterial proteins by these genes is controlled by NF-κB–related transcription factors. We compared the ability of wild-type and Parp mutant animals to induce two innate immunity genes, Diptericine and Drosomycin, using Western analysis of Diptericine-lacZ and Drosomycin-GFP (where GFP is green fluorescent protein) reporter genes (Fig. 3C). Both genes were strongly induced in wild-type, 20 to 40 hours after bacterial challenge, but protein levels increased little if at all in Parp mutant animals under the same conditions. These experiments argue that PARP's role in NF-κB–dependent immune response gene expression has been conserved during evolution. Our observations indicate that PARP may act on the chromatin organization of NF-κB target loci.

PARP's proposed action during DNA repair suggests a model for its role during puffing (Fig. 4). Signals other than DNA lesions, including steroid hormones, stress, and infection, may activate PARP molecules at specific chromosome sites. After modification with poly(ADP-ribose), local chromatin proteins, transcription factors, and PARP molecules themselves are proposed to dissociate from the DNA and/or from pre-existing protein complexes, and to complex with nearby branched ADP-ribose polymers. While in this expanded state, the altered chromosome domain may be transcribed and its chromatin constituents modified. Subsequently, the ADP-ribose would be cleaved, allowing chromatin to reform in an unchanged or altered configuration as dictated by the modified constituents.

Figure 4

Model of PARP-mediated puffing and chromatin remodeling. (Top) Chromatin domain contains nucleosomes (blue), PARP molecules (closed green rings), scaffold-attachment complexes (yellow), and other chromatin proteins (see box). (Middle) After an inducing signal, PARP becomes activated (open green rings) and modifies itself and other proteins with poly(ADP-ribose) chains (red), causing chromatin proteins to dissociate and the DNA backbone to extend. Released proteins remain nearby, bound to poly(ADP-ribose) forming a loose structure corresponding to the puff. After possible chromatin modification, autoinactivation of PARP activity, and removal of the inducing signal, a constitutive glycohydrolase removes poly(ADP-ribose) adducts and the chromatin reassembles in an unchanged or reprogrammed state (bottom).

These results help clarify the biological importance of puffs, which has remained unclear because under certain conditions puffing and gene expression are separable (9). Formation of an expanded chromatin state, the puff, would require PARP. However, only the configuration of factors pre-existing at the target site would determine whether chromatin loosening leads to gene activation. Our experiments also suggest a specific molecular mechanism for puffing. Nucleosomes in puffs lose their regular arrangement (10), and hsp70 region genomic DNA becomes greatly extended (11). Nucleosomal and linker histones may be targets of ADP ribosylation by activated PARP molecules, causing chromatin in puffed regions to expand.

Our experiments also suggest that genes located within repressed chromatin could be exposed and made more susceptible to reprogramming by intentionally activating nearby chromosomal PARP molecules. Understanding how PARP is activated within normal, undamaged chromatin will advance our knowledge of developmental gene regulation and facilitate the development of methods to experimentally reprogram genes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/299/5606/560/DC1

Materials and Methods

SOM Text

Fig. S1 and S2

References

REFERENCES AND NOTES

View Abstract

Navigate This Article