Insulators Targets Pcg Proteins to a Downstream Promoter

Developmental Cell 11, 117–124, July, 2006 ª2006 Elsevier Inc. DOI 10.1016/j.devcel.2006.05.009 PRE-Mediated Bypass of Two Short Article Su(Hw) Insulators Targets PcG Proteins to a Downstream Promoter

Itys Comet,1 Ekaterina Savitskaya,2,3 dependent on the chromatin surrounding the transgene Bernd Schuettengruber,1 Nicolas Ne`gre,1,4 (barrier activity). Second, when an insulator is placed Sergey Lavrov,1,5 Aleksander Parshikov,2 between an enhancer and a promoter, it can prevent François Juge,1,6 Elena Gracheva,2,7 the enhancer from activating the promoter (enhancer Pavel Georgiev,2,* and Giacomo Cavalli1,* blocking activity). 1 Institute of Human Genetics The Drosophila Su(Hw) insulator, from the gypsy retro- Centre National de la Recherche Scientifique transposon, contains 12 binding sites for the Su(Hw) 141 rue de la Cardonille protein and can block enhancer-promoter interactions F-34396 Montpellier Cedex 5 in a Su(Hw)-dependent manner when inserted between France them (Cai and Levine, 1995; Geyer and Corces, 1992; 2 Department of the Control of Genetic Processes Scott et al., 1999). On the other hand, two intervening Institute of Gene Biology Su(Hw) insulators restore enhancer access to down- Russian Academy of Sciences stream promoters (Cai and Shen, 2001; Muravyova Moscow 119334 et al., 2001). This ability, which is shared by other insula- Russia tor elements (Gruzdeva et al., 2005; Melnikova et al., 2004), was suggested to depend on pairing of the two insulators that might bring the upstream enhancer in the Summary vicinity of the downstream promoter. In addition to blocking enhancers, one copy of the Drosophila Polycomb group response elements (PRE) Su(Hw) insulator can also counteract the repressive ef- silence neighboring genes, but silencing can be fect of a Polycomb response element (PRE) (Mallin blocked by one copy of the Su(Hw) insulator element. et al., 1998; Sigrist and Pirrotta, 1997). PREs are bound We show here that Polycomb group (PcG) proteins can by proteins of the Polycomb group (PcG), inducing si- spread from a PRE in the flanking chromatin region lencing of both endogenous target genes and reporter and that PRE blocking depends on a physical barrier genes (Chan et al., 1994; Dejardin et al., 2005; Sigrist established by the insulator to PcG protein spreading. and Pirrotta, 1997). Here, we investigated molecular On the other hand, PRE-mediated silencing can by- mechanisms of insulator-dependent PRE blocking, pass two Su(Hw) insulators to repress a downstream and we analyzed the effect of the presence of two insu- reporter gene. Strikingly, insulator bypass involves lators on PRE-mediated silencing. Our data show that targeting of PcG proteins to the downstream pro- the spreading of PRE bound proteins into neighboring moter, while they are completely excluded from the in- chromatin domains is blocked by the Su(Hw) insulator. tervening insulated domain. This shows that PRE-de- In contrast, two copies of the insulator are bypassed pendent silencing is compatible with looping of the by the PRE, and this bypass involves the establishment PRE in order to bring PcG proteins in contact with of PcG binding at the repressed promoter. These data the promoter and does not require the coating of the strongly suggest that insulator pairing induces a contact whole chromatin domain between PRE and promoter. between PRE bound proteins and the downstream promoter, resulting in its silencing.

Introduction Results

Chromatin insulators are thought to regulate gene ex- Enhancer-Mediated Insulator Bypass pression by defining independent chromatin domains in the Presence of a PRE (Brasset and Vaury, 2005; Capelson and Corces, 2004; We first asked whether an enhancer can activate its pro- West et al., 2002). These elements are characterized moter across a chromosomal domain maintained in a si- mainly by two activities that have been identified in lenced state by a PRE and bordered by two Su(Hw) insu- transgenic assays. First, when an enhancer and its cog- lators. In the (E)S(P)YSW transposon (Figure 1A), the nate promoter are flanked by two insulators, they are white minigene and the yellow gene served as reporters isolated from activating or repressing position effects responsible, respectively, for red eye pigmentation and for dark pigmentation of the larval and adult cuticle and its derivatives. Silencing was dependent on a PRE *Correspondence: [email protected] (G.C.); georgiev_p@ from the bxd region of the homeotic gene Ultrabithorax mail.ru (P.G.) 3 Present address: Centre of Bioengineering, Russian Academy of (Ubx)(Sigrist and Pirrotta, 1997) (hereafter, called PRE). Sciences, Prosp. 60-let Oktiabria, bld.7-1, Moscow 117312, Russia. The upstream region of the transposon carried the eye- 4 Present address: Department of Genetics, Yale University School specific enhancer, separated from the white reporter by of Medicine, New Haven, Connecticut 06520. an intervening domain containing the PRE and the yel- 5 Present address: DMGC, Institute of Molecular Genetics, RAS, 2, low gene bordered by two Su(Hw) insulators. The eye- Kurchatov sq., Moscow, 123182, Russia. specific enhancer was flanked by loxP sites (lox), and 6 Present address: Institute of Molecular Genetics, CNRS-UMR 5535. 1919, route de Mende, F-34293 Montpellier Cedex 5, France. the PRE was flanked by frt sites, allowing excision by 7 Present address: Department of Biology, Washington University, crossing transgenic flies with flies expressing Cre- or St. Louis, Missouri 63130. Flp-recombinase (Golic et al., 1997; Siegal and Hartl, Developmental Cell 118

that could be excised are indicated between brackets in all constructs throughout the manuscript). In five lines with single copy insertions at independent genomic sites, the PRE silenced yellow as excision of the PRE induced yellow derepression (Figure 1A). In contrast, white was not repressed in four of these lines since the eye color ranged from brown to red and was generally not influenced by excision of the PRE, although eye pigmentation was reduced after excision of the eye enhancer. These data show that the eye-specific enhancer can activate white, bypassing a PRE-containing domain bordered by two Su(Hw) insulators. We note here a slight increase in eye pigmentation upon PRE excision in the I-3 and the III-1 lines. This might suggest a partial ability of the PRE to overcome the Su(Hw) insulator in these lines, consistent with previous reports of partial enhancer blocking by chromatin insulators (Hag- strom et al., 1996; Savitskaya et al., 2006). In the four lines showing insulator bypass, strong reduction in eye pigment levels was observed and eye color variegation appeared in a su(Hw) mutant background (data not shown). These characteristic features of PRE-mediated silencing indicated that the PRE gained the ability to repress white when the insulator was neutralized. Taken together, these results suggest that an insulated PRE is unable to affect insulator bypass by capturing an enhancer on its way to the downstream promoter.

PRE-Mediated Insulator Bypass We next tested whether PRE-mediated silencing can bypass a domain bordered by two Su(Hw) insulators. The transposons PSYSW, (P)(S)YSW, and (P)SY(S)W (Fig- ures 1B–1D) carry in their upstream regions the yellow body- and wing-specific enhancers and the PRE. These elements are followed by a domain in which two Su(Hw) insulators border the bristle-specific enhancer and the yellow coding sequence and, further downstream, by the white reporter. The phenotypes of PSYSW, (P)(S)YSW, and (P)SY(S)W were similar, suggesting that the frt or loxP sequences (absent in the PSYSW transposon) did not affect any functional element. In all transgenic lines, the yellow gene was expressed at Figure 1. Transposons Used in This Study and Phenotypic Analysis high levels by the bristle enhancer (Figures 1B–1D) of Transgenic Lines and, as expected, was only weakly expressed in the (A–D) (A) (E)S(P)YSW, (B) PSYSW, (C) (P)(S)YSW, (D) (P)SY(S)W. body and the wing, as the proximal insulator separates Brackets indicate elements that were excised by using Flp- or Cre- the yellow coding sequence from the body and wing en- recombinase. The transgene maps, not drawn to scale, indicate the yellow promoter and coding region (y), the yellow body (Eb), hancers. In contrast, the eye pigmentation was strongly wing (Ew) and bristle enhancer (Ebr), the white minigene (w), the reduced, ranging from complete white to pale yellow in eye-specific enhancer (Ee), the PRE, and the Su(Hw) insulators (S). most lines (Figures 1B–1D), suggesting that the PRE The P element ends, the frt, and lox sites are indicated by empty tri- strongly repressed white across the two Su(Hw) insula- angles, gray triangles, and gray lozenges, respectively. The trans- tors. Some of these lines showed variegation and in- genic line name and the derivative lines obtained after excision of creased silencing in the homozygous state, typical fea- PRE, enhancer, or insulator (DPRE, DEe, DS, respectively) are indicated in (A), (C), and (D). Phenotypic data are schematically repre- tures of PRE-mediated silencing (Kassis, 2002). Other sented above each construct. The number of lines showing the lines escaped silencing, but the frequency of strongly re- same phenotype is indicated in (B). For each line, pigmentation pressed lines was over 50%, similar to previous reports levels reflecting expression of the yellow (in the bristles) or the white of PRE-mediated silencing of white-containing trans- (eye color) reporters are indicated by boxes with a five level scale. genes (Chan et al., 1994; Sigrist and Pirrotta, 1997). White indicates no expression, and black indicates maximal expres- The excision of the PRE strongly derepressed white sion. ‘‘L’’ indicates that the corresponding lines are lethal. in numerous lines but did not affect yellow expression (Figures 1B–1D). In most of these lines, excision of one 1996). This allowed us to compare the effects of the ab- Su(Hw) insulator copy derepressed white to a similar sence or presence of either the eye-specific enhancer or extent as the PRE excision. In two lines, (P)(S)YSW II-1 the PRE at a given transgene insertion site (elements and II-3, the derepression upon insulator excision was PcG Protein-Mediated Insulator Bypass 119

Figure 2. Chromosomal Landscape at the Site of Insertion in the (P)(S)YSW-22E and the (P)SY(S)W-93D Lines (A) The distribution of PH (top) and PC (bot- tom) along the chromosomal region at the (P)(S)YSW-22E insertion site is shown. The y axis shows the fold change expressed in

alog2 scale, for each oligonucleotide present along the chromosomal walk synthesized on the chip. The genes annotated along the walk (Flybase release 4.3) are indicated be- low the graphs. The 50 and 30 sequence coor- dinates (Flybase release 4) are indicated. (B) Distribution of PH and PC along the chromosomal region at the (P)SY(S)W-93D insertion site. (C–D) Distribution of PH and PC along the chromosomal regions containing the en and the ph loci. Known PREs are indicated by as- terisks (Ne` gre et al., 2006).

slightly lower than upon PRE excision. This might de- (Figure 1D), while proximal insulator excision in pend on a weak ability of the PRE to bypass the remain- (P)(S)YSW lines triggered yellow repression (Figure 1C). ing insulator. In two further lines, (P)SY(S)W 36C8 and All lines showing insulator bypass were tested in III-1, white expression was almost absent, and even a su(Hw) mutant background. A reduction in yellow ex- upon excision of the PRE, the derepression was barely pression similar to levels obtained upon excision of the detectable, preventing thorough assessment of insula- proximal insulator was observed (data not shown), indi- tor bypass. This indicates that the construct landed in cating that although two insulators were bypassed, they a strongly repressive chromatin environment in these were still functional in isolating the intervening domain lines. PRE-containing constructs have been previously from PRE-mediated silencing. Therefore, these data shown to have a tendency to insert into other PRE-con- show that an active chromatin domain bordered by taining loci (Kassis, 2002). Therefore, we mapped the two Su(Hw) insulators can be bypassed by a PRE to si- sites of transgene insertion in these two lines by re- lence a distal gene. verse-PCR and then tested whether PcG proteins bind in the vicinity of these regions. This was done by com- PcG-Mediated Insulator Bypass Involves PcG bining chromatin immunoprecipitation (ChIP) with cus- Binding to the Downstream Promoter tom-made long oligonucleotide DNA microarrays (ChIP Two representative lines, named (P)(S)YSW-22E and on chip). The data show no signiﬁcant binding of Poly- (P)SY(S)W-93D, were analyzed in more detail. We map- comb (PC) and Polyhomeotic (PH) in a region of 75 kb ped the site of transgene insertion in these two lines surrounding the transgene insertion site of the two lines by reverse-PCR. ChIP on chip analysis showed that analyzed (Figure S1). This is consistent with the absence the PcG proteins PC and PH are absent from these re- of PcG binding sites in polytene chromosomes (Zink and gions in the absence of the transgenes (Figures 2A and Paro, 1989) and suggests that the repressive chromatin 2B), while their distribution in two control regions con- environment at the site of insertion in these two lines de- taining known PcG targets, the en and the ph loci, pends on other regulatory factors. strongly resembles a previously published proﬁle (Ne` gre We then analyzed yellow, the second reporter gene et al., 2006). Moreover, PC and PH show an association present in the construct. No effect was observed after to the cytological sites of transgene insertion in polytene excision of the distal Su(Hw) insulator in (P)SY(S)W lines chromosome stainings, which is lost upon excision of Developmental Cell 120

Figure 3. PRE-Mediated Insulator Bypass in the (P)(S)YSW-22E and the (P)SY(S)W-93D Lines at the Pupal Stage (A) Eye pictures of 7 day old homozygous males from the (P)(S)YSW-22E line and its derivatives obtained after excision of the PRE or of the proximal Su(Hw) insulator. (B) PC binding on the constructs from the homozygous (P)(S)YSW-22E line and its respective derivatives analyzed by ChIP, followed by real-time PCR quantification. Each ChIP was performed on 0–1 day old pupae in at least two independent experiments. Enrichment factors are shown in the y axis. Red squares indicate enrichment for the intact line. Green triangles and blue circles indicate enrichments for derivatives obtained, respectively, after excision of the proximal Su(Hw) insulator or of the PRE. Triangles and circles are drawn under transgene maps to designate the excised elements. Error bars represent 95% confidence intervals corresponding to standard deviations calculated on enrichment factor values from at least three independent PCR quantifications. The black arrow on the right of each chart indicates the enrichment level obtained for a positive control region from the Fab-7 PRE, an endogenous PcG protein target located in the Bithorax complex. The empty arrow indicates enrichment of the negative control rp49. (C) PH binding on the constructs from the homozygous (P)(S)YSW-22E line and its derivatives, analyzed as in (B). (D) Eye pictures of 7 day old homozygous males from the (P)SY(S)W-93D line and its derivatives. (E–F) PC and PH binding on the constructs from the homozygous (P)SY(S)W-93D line and its derivatives, as in (B). the PRE (data not shown). These data strongly suggest strong white derepression (Figures 3A and 3D). Support- that the distribution of PcG proteins in these lines de- ing this conclusion, both lines showed white derepres- pends strictly on the insertion of the PRE. sion in a mutant background for the PcG gene polyho- Eye phenotype analysis showed that the eye color of meotic (data not shown). In both lines, excision of one both lines is white variegated, with a few yellow to or- insulator derepressed white to levels similar to PRE ange patches (Figures 3A and 3D). Silencing of the white excision, indicating that one intervening insulator copy reporter depends on the PRE, as its excision induced can block the PRE (Figures 3A and 3D). But how is PcG Protein-Mediated Insulator Bypass 121

PcG-mediated silencing blocked, and how can a PRE again similar to the (P)(S)YSW-22E line. In contrast to bypass a domain bordered by two insulators? PcG pro- (P)(S)YSW-22E, however, PC and PH were still blocked teins have been suggested to spread from a PRE into by the proximal insulator upon removal of the distal insu- flanking domains (Strutt et al., 1997), and they can asso- lator, instead of invading the yellow domain. These data ciate with promoters where they may interfere with the show that the proximal Su(Hw) insulator can block the transcription machinery (Dellino et al., 2004). Therefore, spread of the PcG proteins. we postulated that the Su(Hw) insulator might act as a ‘‘roadblock’’ to prevent the spread of PcG proteins PcG-Mediated Insulator Bypass Is Established from the PRE into regions downstream of the insulator, during Embryonic Development and that adding a second insulator copy might restore Since the PRE used in this study first functions during PcG-dependent silencing via insulator pairing, bringing midembryogenesis (Orlando et al., 1998; Poux et al., the PRE into physical contact with the downstream 1996), we also performed ChIP on 0–12 hr old embryos promoter. from both the (P)(S)YSW-22E and the (P)SY(S)W-93D This hypothesis was tested by ChIP experiments with lines and their respective derivatives (Figures 4A–4D). antibodies against the PC and PH proteins on the The patterns of PC distribution on transgenes at embry- (P)(S)YSW-22E line and two derivatives obtained by ex- onic stage reproduced those observed at pupal stage cising the PRE or the proximal Su(Hw) insulator (Figures (compare Figures 4A and 4C with Figures 3B and 3E). 3B and 3C). Both PC and PH are required for PRE-medi- As for PC, no change in PH binding is observed at em- ated silencing and are core components of the PcG mul- bryonic stage except on the PRE sequence, where PH tiprotein complex PRC1 (Shao et al., 1999). Furthermore, binding was clearly stronger than at pupal stage for PC has previously been shown to be recruited in vitro both lines (compare Figures 4B and 4D with Figures and in vivo by the PRE used here (Horard et al., 2000; 3C and 3F). This indicates that PC and PH may be pres- Strutt et al., 1997). ChIP experiments were performed ent at different ratios on their target regions at different on 0–1 day old pupae because adult eye pigmentation developmental stages, hinting to a dynamic develop- requires white expression during the first two days of mental function for these two members of the PRC1 puparation (Steller and Pirrotta, 1985). The PRE and its complex in PRE-mediated silencing. In summary, these flanking regions were strongly bound by PC and PH in data show that PRE-mediated insulator bypass as well both the (P)(S)YSW-22E line and its derivative lacking as insulator-dependent blocking of PcG protein spread- the proximal insulator (Figures 3B and 3C), suggesting ing are established early in development. that insulator-mediated blocking of PRE-mediated silencing does not involve direct interference with binding Discussion of PcG proteins to the PRE. In contrast, both PC and PH were completely removed from the transgene upon PRE The present work suggests two complementary mecha- excision, showing that PcG protein association with the nisms for promoter silencing by PcG proteins. First, our transgene strictly depends on the PRE. Remarkably, the data show directly that PcG proteins recruited at a PRE white promoter region was efficiently bound by PC and can spread over several kilobases along the flanking PH in the (P)(S)YSW-22E line, whereas no binding was chromatin. Therefore, promoters located within short detected within the intervening domain bordered by distances from PREs (Americo et al., 2002; Bloyer the two Su(Hw) insulators. Thus, in the presence of et al., 2003) might be silenced by PcG spreading and in- two insulators, PcG proteins bound at the PRE by- terference with the transcription machinery (Dellino passed the Su(Hw) insulator-bordered domain, which et al., 2004). However, PcG spreading induced by the was shielded from PcG association, and reached the Ubx PRE did not extend beyond few kilobases in our ex- downstream white promoter for silencing. periments, and ChIP on chip also showed limited exten- Excision of the proximal Su(Hw) insulator allowed PC sion of PcG protein binding from known PREs. This lim- and PH to invade the yellow-containing domain, show- ited spreading might depend on genomic sequences or ing that PcG proteins spread from the PRE in neighbor- proteins bound to them that might attenuate chromatin ing chromatin regions in the absence of an insulator that association of PcG proteins. Thus, spreading alone blocks PcG spreading (Figures 3B and 3C and Table S1). might not be sufficient for silencing promoters located These data also show that the yellow promoter did not several tens of kilobases away, as in the case of the block PcG protein spreading in the absence of the prox- Ubx gene, suggesting that additional mechanisms imal insulator, although it might be able to attenuate allow PcG proteins to gain access at distant promoters. spreading to some extent. PcG binding ended at the dis- We found that pairing of two Su(Hw) insulators can in- tal insulator and no PC and PH proteins were detected at duce promoter association of PcG complexes without the white promoter (Figures 3B and 3C). In order to con- PcG-mediated coating of the insulated domain. This firm that the Su(Hw) insulator can prevent PcG proteins suggests an additional mechanism of PRE-dependent from invading a neighboring chromatin domain, we per- promoter silencing, whereby PREs located at large dis- formed a similar ChIP analysis on the (P)SY(S)W-93D tances from their promoters might contact them via line, in which the distal—rather than the proximal— looping of intervening domains. This looping might be Su(Hw) insulator was excisable (Figures 3E and 3F). favored by natural regulatory elements present at these The distribution of PC and PH on the transgene in this loci (Parnell et al., 2003; Zhou and Levine, 1999), which line reproduced the pattern of the (P)(S)YSW-22E line might play a role similar to the pair of Su(Hw) insulators (compare Figures 3E and 3F with Figures 3B and 3C). used in this study. Moreover, excision of the distal Su(Hw) insulator re- The endogenous distribution of PcG proteins might moved the PC and PH proteins from the white promoter, reflect spreading from a PRE into the flanking genomic Developmental Cell 122

Figure 4. PRE-Mediated Insulator Bypass in the (P)(S)YSW-22E and the (P)SY(S)W-93D Lines at the Embryonic Stage (A–D) Binding of PC and PH in the homozygous lines and their derivatives analyzed by ChIP, followed by real-time PCR quantiﬁcation, as in Figure 3. (E) A model depicting a possible mechanism for PcG-mediated bypass of two insulator elements with exclusion of the intervening chromatin domain from PcG protein spreading.

region as well as their ability to bypass insulators. At the site downstream to the PRE, in addition to the previously two endogenous target loci en and ph, where PREs are described peaks corresponding to the PRE and the Ubx located in the promoter region, the distribution of PC promoter (Orlando et al., 1998; Ringrose et al., 2004). and PH suggests spreading from the PREs. We then Furthermore, binding of PH and PC drops between the characterized the distribution of PC and PH at Ubx, a lo- bxd peak and the Ubx promoter and rises again at the cus where the PRE is over 20 kb upstream from the Ubx promoter (Figure S2). This distribution is consistent promoter (Figure S2). In addition to Ubx, this region con- with spreading from the PRE for short-distance chroma- tains the bxd locus, driving the production of noncoding tin silencing, and direct targeting of PRE bound PcG transcripts (Lipshitz et al., 1987). We found that PC and complexes to the downstream promoter to drive silenc- PH binding shows a peak at the bxd transcription start ing over larger distances. PcG Protein-Mediated Insulator Bypass 123

The sharp transitions in PcG protein binding detected was supported by the Ministe` re de la Recherche and the Ligue at insulators are surprising, especially considering that Nationale Contre le Cancer. E.S. was supported by a stipend from the PRC1 complex is larger than 1 MDa (Shao et al., the Center for Medical Studies, University of Oslo. B.S. was supported by the Austrian FWF (Fonds zur Fo¨ rderung der wissenschaflichen For- 1999), a size equivalent to several nucleosomes. The schung). N.N. was supported by the Ministe` re de la Recherche, block in PcG spreading might depend on a physical bar- the European Molecular Biology Organization (EMBO), and the As- rier imposed by protein complexes tightly bound to the sociation pour la Recherche sur le Cancer. P.G. was supported by insulator. The bypass of the insulated domain might be a Molecular and Cellular Biology Program of the Russian Academy explained by topological features imposed by insulators of Sciences grant and by an International Research Scholar award on three-dimensional chromatin folding (Figure 4E). The from the Howard Hughes Medical Institute. G.C. was supported by grants from the Centre National de la Recherche Scientifique, the Su(Hw) and Mod(mdg4) proteins that regulate the Human Frontier Science Program Organization, the European Union Su(Hw) insulator are organized into discrete ‘‘insulator FP 6 (Network of Excellence ‘‘The Epigenome’’ and STREP ‘‘3D Ge- bodies’’ in the cell nucleus (Gerasimova et al., 2000; Xu nome’’), the Ministe` re de la Recherche (ACI BCMS 2004), and by the et al., 2004). PcG proteins are also organized into Association pour la Recherche sur le Cancer. ‘‘PcG bodies’’ that might be the sites of PRE-mediated silencing (Grimaud et al., 2006). A single Su(Hw) insula- Received: August 6, 2004 tor located near a PRE might thus exclude the down- Revised: April 21, 2006 stream domain from the PcG body associated to the Accepted: May 30, 2006 Published: July 10, 2006 PRE. A second insulator paired with the first one in the insulator body (West et al., 2002) might bring the down- References stream promoter at the PRE-associated PcG body, while excluding from it the intervening chromatin do- Americo, J., Whiteley, M., Brown, J.L., Fujioka, M., Jaynes, J.B., and main (Figure 4E). This type of regulation of three-dimen- Kassis, J.A. (2002). A complex array of DNA-binding proteins re- sional chromatin folding by insulator elements might quired for pairing-sensitive silencing by a polycomb group response modulate gene expression at a number of loci in Dro- element from the Drosophila engrailed gene. Genetics 160, 1561– sophila and other species. 1571. Bloyer, S., Cavalli, G., Brock, H.W., and Dura, J.M. (2003). Identifica- tion and characterization of polyhomeotic PREs and TREs. Dev. Experimental Procedures Biol. 261, 426–442.

Fly Methods Brasset, E., and Vaury, C. (2005). Insulators are fundamental compo- Transgenes were obtained with standard cloning techniques (see nents of the eukaryotic genomes. Heredity 94, 571–576. Supplemental Data). Flies were grown on standard rich media. For Cai, H., and Levine, M. (1995). Modulation of enhancer-promoter in- germline transformation, the y1,w1118 strain was used. For establish- teractions by insulators in the Drosophila embryo. Nature 376, 533– ment of transgenic lines and of derivatives with excision of func- 536. tional elements and for phenotype analysis, see Supplemental Data. Cai, H.N., and Shen, P. (2001). Effects of cis arrangement of chromatin insulators on enhancer-blocking activity. Science 291, 493–495. ChIP Experiments Capelson, M., and Corces, V.G. (2004). Boundary elements and nu- Chromatin immunoprecipitation from staged 0–12 hr old embryos clear organization. Biol. Cell. 96, 617–629. and from 0–1 day old pupae were performed as described previously (Ne` gre et al., 2006). For quantitative ChIP, after immunoprecipitation Chan, C.S., Rastelli, L., and Pirrotta, V. (1994). A Polycomb response and DNA purification, enrichment of specific DNA fragments was an- element in the Ubx gene that determines an epigenetically inherited alyzed by real-time PCR, with Roche Light Cycler equipment and ac- state of repression. EMBO J. 13, 2553–2564. cessories and a Light Cycler FastStart DNA Master SYBR green I kit. Dejardin, J., Rappailles, A., Cuvier, O., Grimaud, C., Decoville, M., The detailed method for real-time PCR analysis of relative PcG pro- Locker, D., and Cavalli, G. (2005). Recruitment of Drosophila Poly- tein enrichments is described in the Supplemental Data. The oligos comb group proteins to chromatin by DSP1. Nature 434, 533–538. used for PCR reactions and the sequences of each amplicon are de- Dellino, G.I., Schwartz, Y.B., Farkas, G., McCabe, D., Elgin, S.C., and scribed in Tables S2–S4. For microarray analysis, DNA samples out Pirrotta, V. (2004). Polycomb silencing blocks transcription initiation. of specific ChIP (PC and PH antibodies) or control ChIP (mock), done Mol. Cell 13, 887–893. in a w1118 line, were amplified by adaptor-mediated PCR (Ne` gre Gerasimova, T.I., Byrd, K., and Corces, V.G. (2000). A chromatin in- et al., 2006). Amplified DNA was labeled by random priming with sulator determines the nuclear localization of DNA. Mol. Cell 6, 1025– Cy3 or Cy5 and hybridized to custom oligonucleotide tiling arrays 1035. (NimbleGen Systems of Iceland, LLC). Tiling arrays containing iso- thermal probes (Tm: 76ºC) were designed to provide 100 bp resolu- Geyer, P.K., and Corces, V.G. (1992). DNA position-specific repres- tion with an average 40 bp spacing throughout the regions of inter- sion of transcription by a Drosophila zinc finger protein. Genes Dev. est. The signal ratio between specific ChIP and control ChIP (fold 6, 1865–1873. changes) were log2-transformed and schematically presented with Golic, M.M., Rong, Y.S., Petersen, R.B., Lindquist, S.L., and Golic, the SignalMap program (NimbleGen Systems). K.G. (1997). FLP-mediated DNA mobilization to specific target sites in Drosophila chromosomes. Nucleic Acids Res. 25, 3665–3671. Supplemental Data Grimaud, C., Bantignies, F., Pal-Bhadra, M., Ghana, P., Bhadra, U., Supplemental Data include four tables and two figures and are avail- and Cavalli, G. (2006). RNAi components are required for nuclear able at http://www.developmentalcell.com/cgi/content/full/11/1/ clustering of Polycomb group response elements. Cell 124, 957–971. 117/DC1/. Gruzdeva, N., Kyrchanova, O., Parshikov, A., Kullyev, A., and Geor- giev, P. (2005). The Mcp element from the bithorax complex contains Acknowledgments an insulator that is capable of pairwise interactions and can facilitate enhancer-promoter communication. Mol. Cell. Biol. 25, 3682–3689. We thank R. Paro for the generous gift of PC and PH antibodies and P. Geyer and V. Pirrotta for kindly providing DNAs. We thank S. He- Hagstrom, K., Mu¨ ller, M., and Schedl, P. (1996). Fab-7 functions as nikoff for the design of oligonucleotides used in ChIP on chip exper- a chromatin domain boundary to ensure proper segment specifica- iments. We thank the reviewers for criticism that helped improve the tion by the Drosophila bithorax complex. Genes Dev. 10, 3202–3215. analysis of real-time ChIP experiments. We also thank G. Uze´ and Horard, B., Tatout, C., Poux, S., and Pirrotta, V. (2000). Structure of G. Lutfalla for their help in establishing real-time PCR analysis. I.C. a polycomb response element and in vitro binding of polycomb Developmental Cell 124

group complexes containing GAGA factor. Mol. Cell. Biol. 20, 3187– Zink, B., and Paro, R. (1989). In vivo binding pattern of a trans-regu- 3197. lator of homeotic genes in Drosophila melanogaster. Nature 337, Kassis, J.A. (2002). Pairing-sensitive silencing, polycomb group re- 468–471. sponse elements, and transposon homing in Drosophila. Adv. Genet. 46, 421–438. Lipshitz, H.D., Peattie, D.A., and Hogness, D.S. (1987). Novel transcripts from the Ultrabithorax domain of the bithorax complex. Genes Dev. 1, 307–322. Mallin, D.R., Myung, J.S., Patton, J.S., and Geyer, P.K. (1998). Poly- comb group repression is blocked by the Drosophila suppressor of Hairy-wing [su(Hw)] insulator. Genetics 148, 331–339. Melnikova, L., Juge, F., Gruzdeva, N., Mazur, A., Cavalli, G., and Georgiev, P. (2004). Interaction between the GAGA factor and Mod(mdg4) proteins promotes insulator bypass in Drosophila. Proc. Natl. Acad. Sci. USA 101, 14806–14811. Muravyova, E., Golovnin, A., Gracheva, E., Parshikov, A., Belenkaya, T., Pirrotta, V., and Georgiev, P. (2001). Loss of insulator activity by paired Su(Hw) chromatin insulators. Science 291, 495–498. Ne` gre, N., Hennetin, J., Sun, L.V., Lavrov, S., Bellis, M., White, K.P., and Cavalli, G. (2006). Chromosomal distribution of PcG proteins during Drosophila development. PLoS Biol. 4, e170. Orlando, V., Jane, E.P., Chinwalla, V., Harte, P.J., and Paro, R. (1998). Binding of trithorax and Polycomb proteins to the bithorax complex: dynamic changes during early Drosophila embryogenesis. EMBO J. 17, 5141–5150. Parnell, T.J., Viering, M.M., Skjesol, A., Helou, C., Kuhn, E.J., and Geyer, P.K. (2003). An endogenous suppressor of hairy-wing insulator separates regulatory domains in Drosophila. Proc. Natl. Acad. Sci. USA 100, 13436–13441. Poux, S., Kostic, C., and Pirrotta, V. (1996). Hunchback-independent silencing of late Ubx enhancers by a Polycomb Group Response Element. EMBO J. 15, 4713–4722. Ringrose, L., Ehret, H., and Paro, R. (2004). Distinct contributions of histone H3 lysine 9 and 27 methylation to locus-specific stability of polycomb complexes. Mol. Cell 16, 641–653. Savitskaya, E., Melnikova, L., Kostuchenko, M., Kravchenko, E., Pomerantseva, E., Boikova, T., Chetverina, D., Parshikov, A., Zoba- cheva, P., Gracheva, E., et al. (2006). Study of long-distance functional interactions between Su(Hw) insulators that can regulate enhancer-promoter communication in Drosophila melanogaster. Mol. Cell. Biol. 26, 754–761. Scott, K.C., Taubman, A.D., and Geyer, P.K. (1999). Enhancer blocking by the Drosophila gypsy insulator depends upon insulator anat- omy and enhancer strength. Genetics 153, 787–798. Shao, Z., Raible, F., Mollaaghababa, R., Guyon, J.R., Wu, C.T., Bender, W., and Kingston, R.E. (1999). Stabilization of chromatin structure by PRC1, a Polycomb complex. Cell 98, 37–46. Siegal, M.L., and Hartl, D.L. (1996). Transgene Coplacement and high efficiency site-specific recombination with the Cre/loxP system in Drosophila. Genetics 144, 715–726. Sigrist, C.J., and Pirrotta, V. (1997). Chromatin insulator elements block the silencing of a target gene by the Drosophila polycomb response element (PRE) but allow trans interactions between PREs on different chromosomes. Genetics 147, 209–221. Steller, H., and Pirrotta, V. (1985). Expression of the Drosophila white gene under the control of the hsp70 heat shock promoter. EMBO J. 4, 3765–3772. Strutt, H., Cavalli, G., and Paro, R. (1997). Co-localization of Poly- comb protein and GAGA factor on regulatory elements responsible for the maintenance of homeotic gene expression. EMBO J. 16, 3621–3632. West, A.G., Gaszner, M., and Felsenfeld, G. (2002). Insulators: many functions, many mechanisms. Genes Dev. 16, 271–288. Xu, Q., Li, M., Adams, J., and Cai, H.N. (2004). Nuclear location of a chromatin insulator in Drosophila melanogaster. J. Cell Sci. 117, 1025–1032. Zhou, J., and Levine, M. (1999). A novel cis-regulatory element, the PTS, mediates an anti-insulator activity in the Drosophila embryo. Cell 99, 567–575.