|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: XC SC KZ. Performed the experiments: SC XC KC CCJ. Analyzed the data: SC XC. Contributed reagents/materials/analysis tools: TR. Wrote the paper: XC SC KZ. Edited the manuscript: MTF.
Polycomb group (PcG) proteins are key chromatin regulators implicated in multiple processes including embryonic development, tissue homeostasis, genomic imprinting, X-chromosome inactivation, and germ cell differentiation. The PcG proteins recognize target genomic loci through cis DNA sequences known as Polycomb Response Elements (PREs), which are well characterized in Drosophila. However, mammalian PREs have been elusive until two groups reported putative mammalian PREs recently. Consistent with the existence of mammalian PREs, here we report the identification and characterization of a potential PRE from human T cells. The putative human PRE has enriched binding of PcG proteins, and such binding is dependent on a key PcG component SUZ12. We demonstrate that the putative human PRE carries both genetic and molecular features of Drosophila PRE in transgenic flies, implying that not only the trans PcG proteins but also certain features of the cis PREs are conserved between mammals and Drosophila.
Polycomb group (PcG) proteins, together with the functionally antagonizing trithorax group proteins (TrxG), maintain a pre-determined state of transcription, which constitutes a cellular memory stable over many cell divisions , , , , , , , . The PcG proteins act in at least two distinct but interacting protein complexes in mammals, Polycomb Repressive Complex 1 (PRC1, containing BMI1, RING1A, RING1B, CBX, and PHC) and PRC2 (EZH2, SUZ12, and EED) , . The core complex of PRC1 in Drosophila consists of Polycomb (Pc), Polyhomeotic (Ph), Posterior sex combs (Psc), and Drosophila Ring (dRing) ; while PRC2 contains Enhancer of Zeste [E(z)], Extra sex combs, Suppressor of Zeste 12 [Su(z)12], Nurf55, and several other components . In Drosophila, two additional PcG complexes were identified as Pho repressive complex (PhoRC, containing DNA binding proteins Pho/Phol and dSfmbt) and Polycomb repressive deubiquitinase (PR-DUB) , . But the mammalian functional counterparts for PhoRC and PR-DUB remain unclear , . The PcG complexes may use multiple mechanisms to silence transcription, for example, by making the chromatin more compact , or by interfering with transcription initiation ,  and/or elongation . It is generally agreed that PcG complexes employ epigenetic mechanisms that alter chromatin state to repress gene expression. A widely accepted model of PcG action is initiated by the PRC2 complex, which contains an enzymatic component EZH2 to trimethylate histone H3 at Lysine 27 (H3K27me3) , , , . The methylated histone recruits PRC1, which binds to H3K27me3 through the chromodomain of the PC (Polycomb) protein , , leading to nucleation of the entire PcG complex. Although PRC2 and PRC1 should have overlapping binding sites according to this model, other studies revealed some exceptions, suggesting that PRC1 and PRC2 may have independent functions , . In addition to the histone methyl-transferase activity of EZH2 in the PRC2 complex, the RING1B protein in the PRC1 complex acts as an E3 ubiquitin ligase, which ubiquitinates histone H2A at Lysine119 (H2AK119ub) . The H2AK119ub may affect transcription by blocking efficient elongation . In contrast to the transcriptional repressive activity of the PcG complex, the active H3K4me3 mark is generated by the TrxG complex , ,  whose function opposes the PcG function.
Since PcG complex acts through regulation of chromatin structure, it is important to understand how they are recruited to chromatin, in order to characterize the molecular mechanism of PcG-mediated gene silencing. In Drosophila, the PcG proteins are recruited by sequence-specific DNA binding factors, such as PHO (homolog of YY1 ), GAF, PSQ, Zeste, and DSP1 , , , , , , to their target sites known as Polycomb Response Elements (PRE) to silence transcription of target genes , . Although several target genes of the PcG proteins have been identified in mammals , , , mammalian PREs have remained elusive until recently , , partially due to the fact that recruiters such as GAF, PSQ, and Zeste are not conserved in vertebrates. Although two recent studies that identify mammalian PREs ,  show the function of YY1 (homolog of Drosophila PHO) to be important, previous studies have not revealed many regions where YY1 and PcG proteins colocalize . This suggests that the recruitment of mammalian PcG to their target sites may use mechanisms other than YY1. Therefore it is important to identify more functional mammalian PRE to study PcG recruitment in vivo. The lack of knowledge of the cis regulatory elements, the PREs, has hindered our understanding of the critical PcG regulation during mammalian development.
In Drosophila, PcG proteins have enriched binding at PRE sites , , , . Therefore, in this study we sought to identify potential PREs using a candidate approach based on the hypothesis that a functional PRE is associated with H3K27me3 and PcG proteins, and is localized near silenced genes. The first candidate we chose was an H3K27me3-enriched region (SLC) downstream of the SLC17A7 gene , which encodes a sodium-dependent inorganic phosphate co-transporter  and is silent in T cells . Since Hox genes are potentially regulated by PcG proteins, we also selected two regions (A3 and A13) from the HoxA gene locus (Table 1). Using ChIP-PCR (chromatin immunoprecipitation followed by PCR using isotope labeled primers) assays, we found that SLC, A3 and A13 displayed differential levels of H3K27me3 binding (Fig. 1A). To test for enrichment of PRC1 and PRC2 components at these loci, we used human resting CD4+ T cells to perform ChIP experiments using antibodies against PRC1 components BMI1 and RING1B, as well as PRC2 component SUZ12. Our data indicated that the three DNA elements SLC, A3 and A13 had differential binding of PcG proteins. The SLC element was highly enriched with all three PcG proteins: SUZ12, BMI1 and RING1B. The A3 element was associated with intermediate levels of all three PcG proteins, whereas relatively low PcG protein binding was detected at the A13 region (Fig. 1B). Similar results were obtained from HeLa cells (Fig. 1C) and SW-13 cells (Fig. 1D).
To examine whether the enrichment of H3K27me3 and PcG proteins at the putative PREs requires normal function of PcG proteins, we knocked down SUZ12, an essential component of the PRC2 complex , in cell culture system. As shown in Figure S1, the siRNA construct targeting SUZ12 sequences decreased the protein level by over 80%. Global H3K27me3 level was also significantly reduced, probably due to the important function of SUZ12 in PRC2 complex to generate the H3K27me3 modification . We then analyzed the level of PcG proteins at the endogenous SLC and A3 regions using ChIP assays. Consistent with the reduction in its global expression level, SUZ12 binding at both SLC and A3 regions decreased significantly in the SUZ12 knockdown cells (Fig. 2A and 2B). The binding of PRC1 proteins BMI1 and RING1B were also significantly reduced, consistent with the idea that the H3K27me3 mark generated by the PRC2 complex is required for recruitment of the PRC1 complex to the potential PRE sites . In summary, in SUZ12 knockdown cells, the chromatin state at both the SLC and A3 putative PREs changed from high PcG binding and activity to low PcG binding and activity.
To investigate whether these changes at the chromatin level affect expression of endogenous genes near the SLC and A3 loci, we examined the mRNA levels of the genes at the vicinity (Table 1). Interestingly, in HeLa cells, knockdown of SUZ12 resulted in an increased expression level of SLC17A7 (near SLC element) and HoxA3 (near A3 element). In contrast, no obvious increase of HoxA13 (near A13 element) expression was observed (Fig. 2). Together, these results suggested that normal function of PcG proteins is required for the repressive activities of the putative human PREs.
To investigate the functional conservation of the putative human PREs, we tested their activities in Drosophila by assaying the PRE-mediated silencing effect on the miniwhite reporter gene expression in adult fly eyes. In this experiment, a 3-kb genomic DNA fragment surrounding the SLC, A3 or A13 region was placed next to the miniwhite gene (within 100-bp) individually in the pCasper3 expression vector (Table 1, Table 2, and Figure S2). Each of these reporter constructs was incorporated as a single transgene into the w67c23 fly genome, which has a deletion of the promoter region of the endogenous white gene and is a transcript-null allele (, Flybase and data not shown). To better control the eye color difference in males vs. females (males usually have darker eye color than females even if the transgene is on autosomes), we used males for all experiments in Figures 3 and and4.4. The silencing effect was then evaluated by examining the eye color of male flies and by measuring the white mRNA levels with quantitative RT-PCR (qRT-PCR). The qRT-PCR is a more direct and quantitative method to monitor the transcriptional levels of the white gene. In this experiment, newly enclosed heterozygous males (0–1 day old in adulthood) were used for RNA extraction and quantification using an internal control rp32L, a constitutively expressed gene. To rule out the positional effect in gene expression, 5–6 independent transgenic lines were generated from each construct and the results obtained from all lines were analyzed and shown in Figure 3. The repressive effects of miniwhite reporter gene expression by the putative human PRE elements in Drosophila eyes correlated well with their binding affinity with PcG proteins and repressive activities in human cells (Fig. 1 and and2).2). The SLC element, which was a strong PRE candidate in human cells, had the strongest silencing effect in the Drosophila reporter assay (Fig. 3). The A3 element showed modest repressive activity in the Drosophila assay (Fig. 3), consistent with its moderate PcG protein binding in human cells (Fig. 1B and 1C). In contrast, the A13 region did not show any obvious repression activity in Drosophila (Fig. 3), consistent with its lack of PcG binding in human cells (Fig. 1B and 1C).
As the White protein is required for eye pigmentation, the eye color of flies reflects the White protein level and activity. Consistent with the order of the white transcript levels in SLC, A3 and A13 driving transgenes (Fig. 3), the eye color of the corresponding transgenic heterozygous flies ranged from light color for the SLC and A3 transgenes to a much darker color for the A13 transgene (Fig. 4A, upper panels).
A characteristic phenomenon for PRE-mediated gene silencing in flies is the pairing-sensitive silencing (PSS), in which the PRE silencing effect is enhanced in the homozygotes for the transgene. The PSS effect has been suggested as a result of dimerization of PcG proteins at the paired PRE loci . Because of the extremely light eye color of the SLC flies, we used 3-day old males for experiments in Figure 4A–B. As shown by eye color of transgenic flies with an autosomal transgene (Fig. 4A) and the quantification of the white gene mRNA levels (Fig. 4B), the SLC-driving transgene showed the strongest PSS, in which the homozygous eye is lighter than the heterozygous one (Fig. 4A) and by quantification, the white transcript reduced ~40% in homozygotes compared to heterozygotes (Fig. 4B). The A3 human PRE elements showed weaker PSS effect, in which the homozygous eye color is about the same as the heterozygous one (Fig. 4A). In contrast, the A13 element exhibited an approximately 2-fold increase of miniwhite expression in homozygous flies, typical of non-repressive cis-acting sequences. In addition, eye color variegation, often associated with PRE-mediated repression, was also apparent for SLC (Fig. 4A).
To confirm that the silencing of the miniwhite reporter gene by the human PREs in Drosophila is mediated by PcG proteins, we crossed the same human PRE-conjugated miniwhite transgenes (on the second chromosome) to a ph (polyhomeotic) mutant background. Ph is one of the four core components of PRC1 complex in flies , . We then analyzed the eye color and white transcript level using newly enclosed males (0–1 day old). Consistent with the involvement of PcG in silencing, the ph401 mutation (on the X chromosome) relieved SLC-mediated repression 9.6-fold (Fig. 4C). A derepression of 1.5-fold was observed for the moderate repressor A3 (Fig. 4C). In contrast, there was no significant derepression for the non-repressor A13 (Fig. 4C).
Transcriptional repression by the PcG silencing machinery is counteracted by the action of the TrxG protein complex . To test whether mutations in TrxG proteins have an opposite effect compared to the ph mutation, we crossed the same PRE-conjugated miniwhite transgene (on the second chromosome) to a trx temperature sensitive allele trx1 (on the third chromosome) . By shifting the larvae to restrictive temperature at 29°C, we then analyzed the eye color and white transcript level in newly enclosed males (0–1 day old). We observed a 3.1-fold repression of the miniwhite expression in the SLC line in the trx1 background at restrictive temperature compared to the same transgene at permissive temperature at 25°C (Fig. 4D). However, such repression is not significant for either the moderate repressor A3 or the non-repressor A13 (Fig. 4D).
We next examined the relative enrichment of the H3K27me3 modification and PcG proteins at the putative human PREs in transgenic flies. To have a better internal control for ChIP experiments, we generated fly strains that either has P[SLC-w+]; P[A13-w+] or P[A3-w+]; P[A13-w+] dual transgenes. As shown in Figure 5, ChIP with the anti-H3K27me3 antibody using the P[SLC-w+]; P[A13-w+] strain enriched the SLC region 4.3-fold compared to the A13 region. Using the P[A3-w+]; P[A13-w+] strain to ChIP with anti-H3K27me3, the A3 region was also enriched 2.7-fold relative to the A13 region. Likewise, ChIP with antibodies against the PRC1 component Pc or the PRC2 component E(z) enriched SLC region sequence about 2-fold compared to the A13 region. However, this enrichment of Pc and E(z) was not significant at the A3 region, consistent with its less effective PRE activity. Therefore, we concluded that the stronger PRE activities of the SLC element were associated with higher levels of the H3K27me3 modification and the binding of both PRC1 and PRC2 proteins.
In this study, we demonstrated that the PcG-enriched DNA cis-elements in human primary CD4+ T cells have roles in repressing transcription of neighboring genes, and such repressive activities depend on normal function of the trans-acting PcG proteins. Despite high homology of PcG proteins among different organisms , it is unclear to which extent the cis-acting elements, namely the PREs, are also conserved during evolution. Interestingly, we showed that the PcG-enriched sequences not only repress transcription in human cells, but also carry this ability to repress reporter gene expression in Drosophila. Recent research has suggested other molecular mechanisms that target PcG proteins to specific genomic loci in mammalian cells, such as the non-coding RNAs  and the pRB family proteins . Identification of more functional PREs in mammals will allow for sequence comparison and functional analysis to explore whether transcription factor-mediated recruitment ensures PcG-PRE interaction in mammals, which is still under a lot of debates .
Recently, two other groups also reported identification of human PREs in mouse  and human , which revolutionarily changed the view of the molecular mechanisms underlying PcG function in mammals. Interestingly, in one of these reports , it was also shown that the mouse PRE can repress reporter gene expression in Drosophila, and such repressive function can be further modified by mutations in Drosophila PcG genes, which is probably due to the fact that mouse PRE can recruit fly PcG proteins. These data are highly consistent with what we report here. PcG proteins have been found to play multiple roles in stem cell maintenance and tumorigenesis in mammals. Several key developmental regulators are associated with PRC complexes as well as H3K27me3 modification in human and mouse embryonic stem cells , ,  and in human embryonic fibroblasts . Failures in PcG function have profound effect on diseases, such as cancers and tissue dystrophy . Therefore, understanding the mode of action of PcG proteins is essential for understanding mammalian development and PcG dys-regulations during pathological processes. Our successful identification of genomic regions that mediate PcG-dependent transcriptional repression demonstrates evidence for the existence of human counterparts of Drosophila PREs and provides an opportunity for further characterization of the PcG targeting mechanisms in mammalian cells. Studying potential defects by deleting these elements in mouse will be the next step to definitely establish their functional roles as mammalian PREs during development.
Human resting T cells were purified from the whole blood using the lymphocyte separation medium (Mediatech) and Pan T-cell isolation kit II (Miltenyi Biotech) as described previously . The T cells were from healthy donors through the blood bank of National Institutes of Health and do not require any IRB and consent.
Total RNAs from human cells were isolated as described previously . Total RNAs from fly heads were extracted using TRIzol reagent (Cat#15596) according to manufacturer's suggestion (Life Technologies Inc.). The equivalent of 0.5-head was used per PCR reaction. cDNA was synthesized using SuperScript III RNase H reverse transcriptase (Invitrogen). To quantify gene expression levels, real-time PCR was carried out with primers and TaqMan probes from Applied Biosystems Inc. using the Universal RT-PCR Master Mix (Cat# 4309169, Applied Biosystems Inc.). Each PCR reaction was performed in duplicates or triplicates and the Ct numbers for each reaction were collected. Quantification was carried out by the absolute quantification method using standard curves.
ChIP assays using human cells (SW-13 cells are obtained according to , Hela cells were obtained from ATCC, Inc.) were performed as described previously . The antibodies used were histone H3K27me3 (Upstate, 07-449), SUZ12 (Abcam, ab12201), BMI1 (Upstate, 05-637), and RING1B (Abcam, ab3832). Quantification of the ChIP samples from human cells was carried out by the comparative Ct method . Briefly, the target sequences in the ChIP and the input DNA samples were amplified with primers specific for the potential PRE regions or the control regions and the fold difference between the ChIP and the input DNA were calculated. As a control for the ChIP experiment a locus upstream of the human BRG1 gene which showed very little enrichment of H3K27me3 was used.
ChIP assays using the transgenic flies were performed as described previously  except the following changes. About 20 fly heads were isolated from a strain with double transgenes P[SLC- w+];P[A13- w+] or P[A3-w+]; P[A13-w+]. The equivalent of 1.5-head was used per PCR reaction per antibody. The following amounts of antibodies were used: 5 µl anti-H3K27me3 (Upstate, 07-449), 2.5 µl anti-E(z), and 2.5 µl anti-Pc (from R. Jones and R. Kingston, respectively). For quantification of ChIP DNA samples, input DNA, mock precipitated DNA (no antibody) and ChIP DNA with specific antibodies were all analyzed by real-time PCR using the primers obtained from Integrated DNA Technologies. The ChIP and mock DNA were normalized with the input DNA amount. The values from ChIP DNA were further corrected by subtraction of the non-specific signal derived from the mock precipitate (ChIP DNA- mock DNA)/Input, and compared with each other.
For silencing human SUZ12, the target sequence from the SUZ12 cDNA (GGACCTACGTTGCAGTTCACT; position 1053–1073) was inserted into pBS-U6 vector. An unrelated sequence was used as control. The cloned SUZ12 sequence and the control sequence along with the U6 promoter were then subcloned into pREP4-puro as described previously . For RNA interference analysis, HeLa cells (ATCC, Inc.) were transfected with the siRNAs or control and selected with 2 µg of puromycin/ml for 72 hours.
Flies were raised on standard cornmeal molasses agar medium at 25°C unless stated otherwise. The w, ph401 and trx1 strains were obtained from the Bloomington Stock Center (stock numbers are BL-5392 and BL-2114, respectively).
Potential human PREs elements were subcloned into the multi-cloning sites within 100-bp from the white gene in a pCasper3 expression vector. Each plasmid bearing either a potential human PRE or a control element was introduced into the w67c23 fly genome via standard P-element-mediated transformation. The w67c23 (w) mutation deletes the promoter and the first exon region including the start codon of white gene, therefore represents a null allele of white ( and Flybase). On average, 5–6 independent transgenic lines were generated and analyzed for each construct. To test for the pairing-sensitive silencing effect, a double-balanced stock was generated for each transgenic strain with a 2nd or 3rd- chromosomal insertion (Cyo used as the 2nd chromosomal balancer and TM6B used as the 3rd chromosomal balancer). The stock was self-crossed and crossed with the parental w strain simultaneously. The resulting homozygotes and heterozygotes (with no balancer) were compared with each other. To test the derepression of miniwhite by ph401, male flies with the autosomal transgenes were crossed to virgin females either from a w, ph401 strain, or a w strain. Newly enclosed (0–1 day old) male progenies from each cross were compared with each other. To test the repression of miniwhite by the temperature-sensitive trx1 allele, double-balanced males with the second-chromosomal transgenes were crossed to virgin females from a w, trx1 strain. The resulting w; P[w+]/+; trx1/TM6B males were backcrossed to the w, trx1 strain to obtain w; P[w+]/+; trx1 males. The larvae were shifted to restrictive temperature at 29°C and newly enclosed (0–1 day old) males were obtained and compared with newly enclosed males at permissive temperature at 25°C.
SLC-TaqMan Probe ACCCCTGGCTCCTGCCCCATT
A3-TaqMan Probe ATTGAGAGGCAAAGTGCAGGATGG
A13- TaqMan Probe CTGGGCTTGGGCTTTTATCTG
BRG1-TaqMan Probe TGCAGTGAGCCAAGATCTCGACA
SUZ12 small interference RNA inhibits SUZ12 expression and the H3K27me3 signals in HeLa cells. HeLa cells were transfected with pREP4-Puro-siSUZ12 or a control vector with an unrelated sequence and analyzed using Western blotting with antibodies against SUZ12, H3K27me3. Histone H3 was used as loading control.
Index of eye color averaged from 3–6 independent transgenic lines for each of the 12 human PRE tested for repressing white gene expression in Drosophila. −4 is for the lightest eye color and +4 is for the darkest eye color, all data are from Table 2.
We thank Dr. Rick Jones for anti-E(z) antibodies, Dr. Robert Kingston for anti-Pc antibody, Bloomington center for fly strains, and Matt Fish for generating transgenic flies.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was supported by Leukemia and Lymphoma Society Special Fellowship (grant LLS-3598-06), National Institutes of Health (NIH) grants K99/R00 HD055052 Pathway to Independence Award and 1R01HD065816 to XC, NIH grant 1R01GM61986 to MTF, and National Heart, Lung and Blood Institute (NHLBI), NIH intramural grants to KZ. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.