Cell Cycle-Dependent Recruitment of Polycomb Proteins to the ASNS Promoter Counteracts C/ebp-Mediated Transcriptional Activation in Bombyx mori

Epigenetic modifiers and transcription factors contribute to developmentally programmed gene expression. Here, we establish a functional link between epigenetic regulation by Polycomb group (PcG) proteins and transcriptional regulation by C/ebp that orchestrates the correct expression of Bombyx mori asparagine synthetase (BmASNS), a gene involved in the biosynthesis of asparagine. We show that the cis-regulatory elements of YY1-binding motifs and the CpG island present on the BmASNS promoter are required for the recruitment of PcG proteins and the subsequent deposition of the epigenetic repression mark H3K27me3. RNAi-mediated knockdown of PcG genes leads to derepression of the BmASNS gene via the recruitment of activators, including BmC/ebp, to the promoter. Intriguingly, we find that PcG proteins and BmC/ebp can dynamically modulate the transcriptional output of the BmASNS target in a cell cycle-dependent manner. It will be essential to suppress BmASNS expression by PcG proteins at the G2/M phase of the cell cycle in the presence of BmC/ebp activator. Thus, our results provide a novel insight into the molecular mechanism underlying the recruitment and regulation of the PcG system at a discrete gene locus in Bombyx mori.


Introduction
Polycomb group (PcG) proteins are conserved transcriptional repressors involved in regulating body formation during embryonic development [1], and have also been characterized as chromatin modifiers required for the epigenetic regulation of numerous cellular processes, including cell cycle control, tumorigenesis, X-inactivation, cell fate decisions and differentiation [2,3,4,5,6].
PcG proteins function via three key multiprotein complexes: Polycomb repressive complex 1 (PRC1), Polycomb repressive complex 2 (PRC2), and Pleiohomeotic repressive complex (PhoRC) [7,8,9]. The PRC1 complex, which includes Polycomb (Pc), Polyhomeotic (Ph), Sex combs extra (Sce), and Posterior sex combs (Psc) in Drosophila or their counterparts in mammals, is involved in recognizing the chromatin marked with tri-methylated histone H3 on lysine 27 (H3K27me3) [8]. The PRC2 core subunits, which consist of Enhancer of zeste (E(z)), Extra sex combs (Esc), and Suppressor of zeste 12 (Su(z)12), are responsible for catalyzing the tri-methylation of H3K27 to produce H3K27me3 [7]. Importantly, PRC2 complex-mediated establishment of H3K27me3 is orchestrated by the recognition of Pleiohomeotic (Pho, a DNA-binding protein in the PhoRC complex) on specific DNA sequences called Polycomb responsive elements (PREs) in target genes [9], and by subsequent recruitment of other PcG components to the PREs region. A genomewide search in Drosophila has revealed a conserved target site for Pho with a 17 bp binding sequence containing a core ''CCATTTT'' motif [9], which is the same as the binding sequence for Yin and Yang 1 (YY1), the mammalian ortholog of Pho [10]. More recently, a potential PRE region containing YY1binding sites between the HOXD11 and HOXD12 locus has been identified in human embryonic stem cells, and the YY1-binding sites have been reported to contribute to the repression of this locus, but are not required for it [11]. On the other hand, several studies have revealed that long non-coding (lnc) RNAs, short RNAs, or even CpG islands in target genes specifically regulate the recruitment of PcG complexes in mammals [12,13,14,15]. Therefore, there is insufficient evidence that an YY1-binding sequence is required for the recruitment of PcG complexes in mammals.
We previously identified the conserved 13 PcG genes in the silkworm, Bombyx mori [16]. Interestingly, our microarray analysis revealed that the asparagine synthetase (ASNS) gene is similarly upregulated after RNA interference (RNAi)-meditated knockdown of some PcG genes, such as BmSCE, BmESC, BmPHO, and BmSCM (encoding a protein that can interact with BmPho and contribute to transcriptional repression), in silkworm cells [17]. The ASNS gene encodes an enzyme product that catalyzes the biosynthesis of asparagine using glutamine and aspartate as substrates, and is extensively expressed in mammalian cells [18]. It has been reported that under the condition of amino acid or glucose deprivation, human ASNS gene transcription is activated by the amino acid response (AAR) or endoplasmic reticulum stress response (ERSR) pathway, respectively, through two response elements, nutrient-sensing response elements (NSRE)-1 and 22 within the ASNS promoter region [19,20]. In mouse hematopoietic stem cells, ASNS expression can be activated by a basic leucine zipper transcription factor, CCAAT/enhancer-binding protein (C/ebp) [21]. One isoform of C/ebp, C/ebpb, has also been shown to bind to the NSRE sequence within the human ASNS promoter and activate ASNS expression in response to nutrient stress [19]. Much is known about the activation of the ASNS gene, but relatively little is known about how this gene is negatively regulated in normal cells in which the basal transcription is maintained. Recently, however, a microarray screening confirmed that ASNS transcription was induced in mouse cells by a deletion of either BMI1 or SU(Z)12, which are members of the PcG protein family in mammals [21]. This finding, together with our microarray data, implies that ASNS gene may be a conserved target gene for PcG-mediated epigenetic repression. However, the molecular mechanism underlying this potential regulation remains largely unknown.
In addition, the emerging evidence has also shown that most cancer cells display high levels of ASNS expression, suggesting that the ASNS protein has an important function in cancer progression [22]. Although the relationship between cancer progression and ASNS expression is not yet well defined, it would be reasonable to suppose that ASNS expression is precisely regulated so as to prevent cancer development in normal mammalian cells, especially by a potential suppression of PcG regulation.
To define the role of PcG proteins in the regulation of ASNS gene expression, we here analyzed the structure and function of the ASNS promoter in the context of the ASNS regulation in Bombyx. Our results demonstrated that the Bombyx ASNS (BmASNS) gene is regulated by PcG-mediated repression and BmC/ebpmediated activation in a cell cycle-dependent manner. BmPHO and BmSCM RNA interference (RNAi) had little effect on BmASNS expression at the late G1 phase, suggesting that the dissociation of PcG proteins from chromatin would occur before DNA replication. A CpG island on the BmASNS promoter, and, to a lesser degree, several YY1-binding motifs are considered to play a crucial role in the transcriptional repression by recruiting PcG complexes. These findings, for the first time, reveal an epigenetic and cell cycle-specific regulation of ASNS gene expression by a PcG system.

Results
Bombyx Asns is a functionally conserved ortholog of human ASNS In order to understand the potential function and regulation of the BmASNS gene in Bombyx, we first cloned the full-length cDNA of the BmASNS gene according to the reported nucleotide sequence (NP_001037414.1). Amino acid sequence-based multiple alignment of Asns proteins from Bombyx, Drosophila, human, and mouse revealed high identities between insects and mammals ( Figure 1A), raising the possibility that these Asns proteins share some similar functions.
In human, ASNS exhibits a functional response to nutrient stress [19]. Accordingly, we next examined whether BmASNS expression could also be induced under a deprivation of amino acid. To this end, we cultured silkworm cells using modified IPL-41 medium lacking all amino acid components and harvested at the indicated time points after the deprivation of amino acids for semi-quantitative PCR analysis. As shown in Figure 1B, the expression of BmASNS began to increase at 6 h after incubation with modified IPL-41 medium and peaked at 24 h. This result reveals that BmAsns is able to respond to amino acid limitation and may possess an enzymatic activity for amino acid biosynthesis. We also noted that the further loss of expression of the BmASNS gene mediated by RNAi under the condition of amino acid starvation significantly decreased the cell growth (data not shown), indicating an important role of BmAsns in cell proliferation. Taken together, these results led us to identify Asns as an evolutionarily conserved protein in Bombyx that can contribute to amino acid metabolism.

Bombyx PcG proteins negatively regulate BmASNS expression
Due to the potential role of BmAsns in regulating cell development and the abnormal expression of human ASNS in cancer cells, we sought to investigate the regulatory mechanism for this gene expression using the silkworm as a model. Our previous microarray analysis revealed that in silkworm cells, RNAimediated knockdown of PcG genes, such as BmSCE, BmESC, BmPHO, or BmSCM, all increased BmASNS expression [17]. Here, we also observed this up-regulation of BmASNS expression in PcG gene-silenced BmN4-SID1 cells by using a semi-quantitative PCR assay ( Figure 2A).
To survey the effect of transcriptional regulation of PcG proteins on BmASNS expression, we isolated and cloned a 4,974 bp region upstream of the translational initiation codon of the BmASNS gene, as the promoter of BmASNS gene. Several prominent binding sites for YY1 protein were found in this region by the TFSEARCH program (http://www.cbrc.jp/research/db/ TFSEARCH.html) ( Figure 2B), and four conserved YY1-binding motifs in the region were shown to be identical to the Pho-binding profile in Drosophila ( Figure 2C) [9]. Generally, YY1-binding sites mediate the regulation of YY1 protein as transcription factor on the expression of various genes including several PcG targets [23], thus we speculated that these YY1-binding sites may be similarly involved in the regulation of the BmASNS gene via the binding through BmPho, a Bombyx ortholog of the Drosophila Pho protein [16].
We also constructed a luciferase reporter system under the control of the BmASNS promoter and then introduced it into BmN4 cells to confirm this transcriptional regulation by measuring luciferase activity at 72 h after transfection. This promoter exhibited a stronger promoter activity compared with that of the promoter-less luciferase reporter plasmid ( Figure S1). In agreement with the endogenous BmASNS expression in BmN4-SID1 cells (Figure 2A), knockdowns of each of four PcG genes, BmSCE, BmESC, BmPHO, or BmSCM, all increased the BmASNS promoter activity on the reporter plasmid ( Figure 2D). Together, these results showed a negative regulation of PcG proteins on BmASNS expression and gave us a clue to elucidate the role of the PcG system in regulating the expression of a specific target gene in Bombyx.
Transcription factor BmC/ebp activates BmASNS expression putative C/ebp-binding sites on the BmASNS promoter ( Figure 2B). Importantly, transcription of BmASNS was greatly decreased after the RNAi of the BmC/EBP gene in BmN4-SID1 cells ( Figure 3A), which showed an important activation activity of BmC/ebp on BmASNS expression in the presence of the PcG proteins. In agreement with this endogenous promoter activity, a luciferase reporter assay revealed that the BmASNS promoter activity was decreased significantly after the silencing of BmC/EBP expression ( Figure 3B) and increased after the BmC/ebp overexpression ( Figure 3C). These findings thus indicated that transcription factor BmC/ebp is involved in activating BmASNS expression in Bombyx, as shown in mammals.

YY1-binding sites on the BmASNS promoter are involved in the regulation of PcG proteins
To understand how PcG complexes repress BmASNS expression and how the different YY1 recognition sites on the BmASNS promoter are involved in this regulation, we performed a serial deletion analysis for the YY1-binding elements in the promoter ( Figure 4A). Luciferase detection for each deletion construct of YY1-binding elements showed that the upstream region designated as p2285 with deletions of both YY1-3 and YY1-4 displayed the highest activity, suggesting that the activators are bound primarily between YY1-1 and YY1-2 ( Figure 4B), the density binding for BmC/ebp activator in this region ( Figure 2B), and the putative repressors will bind to the distal promoter region. In contrast, the p645 construct in which all four YY1-binding motifs were deleted showed very weak activity, indicating that the activator binding to the upstream sequence is necessary for basal transcription of the BmASNS gene. Together with the aforementioned findings, this result led us to speculate that PcG proteins would mainly bind to the YY1-3 and/or YY1-4 regions and repress BmASNS expression.
To examine this possibility, we performed RNAi experiments for four PcG genes, i.e., BmPHO, BmSCM, BmSCE, and BmESC, as well as the activator gene BmC/EBP. The RNAi of the PcG genes increased the activities of truncated constructs in p4974, p3614, p2285, and p1056, but not p645, whereas depletion of BmC/EBP decreased their activities ( Figure 4C and Figure S2). Furthermore, we co-transfected these deletion constructs with overexpression plasmids for each of the PcG genes or BmC/ebp to test the effects of PcG genes or BmC/ebp overexpression on the activities of the BmASNS promoter. As shown in Figure 4D, the overexpression of BmC/ebp clearly up-regulated the promoter activities of all the constructs. This is consistent with the results obtained from the knockdown experiment, and may also suggest that its putative binding sites of the C/ebp protein are not occupied completely by the endogenous BmC/ebp and the ectopic expression of the BmC/ebp product will facilitate its own local recruitment to the unoccupied binding sites. On the other hand, the overexpression of BmPho or BmScm further repressed the promoter activity in p2285 and slightly repressed the promoter activity in p4974 and p3614. Combined with the RNAi experiments, these results suggested that the principal binding site present on p4974 and p3614 critical for the repression is pre-occupied by the PcG proteins.
Bombyx PcG proteins catalyze a tri-methylation of H3K27 at the BmASNS promoter locus PcG complexes are responsible for the tri-methylation of H3K27 into H3K27me3 on its targets and this H3K27me3 mark is known to be required for PcG-mediated gene silencing [24]. To understand the mechanism by which PcG complexes suppress the BmASNS promoter activity, we analyzed the H3K27me3 patterns within the BmASNS promoter by ChIP assay with a specific antibody for H3K27me3. Pairs of primers for qPCR of the BmASNS promoter were designed as shown in Figure 5A. The results revealed that, although there was a broad domain of H3K27me3 enrichment across the BmASNS promoter locus, it was particularly enriched in the P2 region, namely between YY1-2 and YY1-3 ( Figure 5B). This observation addressed why deletions of YY1-3 greatly increase the promoter activity ( Figure 4B).
To determine whether the tri-methylation of H3K27 at the P2 region was mediated by the PcG system, a ChIP-qPCR assay in PcG gene-silenced BmN4-SID1 cells was performed. We found that the depletion of PRC2 components, such as BmE(Z), BmESC, or BmSU(Z)12, greatly reduced the H3K27me3 marks ( Figure 5C), which revealed that the PRC2 complex is required for the trimethylation of H3K27 on the BmASNS promoter. In contrast, down-regulation of other genes, i.e., BmPHO, BmSCM, and BmSCE (one gene belongs to the PRC1 complex), did not induce a drastic reduction of H3K27me3 levels in comparison with the RNAi for PRC2 components. Therefore, it was shown that PcG complexes bind predominantly to the high-affinity binding site at the nearby P2 region of the chromosome, probably the YY1-3 site.

The CpG island cooperates with YY1-binding sites and contributes to the repressive role of PcG proteins
In mammals, compared with the YY1-binding element, the CpG islands appear to play a more important role in recruiting of the PRC2 complex; and also, the CpG islands participate in the regulation of PcG complexes [14,15]. However, there has been no report about this role of the CpG islands in insect species, maybe because insect genomes are poorly methylated [25]. Interestingly, we found a putative CpG island between YY1-2 and YY1-3 on the BmASNS promoter ( Figure 6A). In order to check whether or how this CpG island is involved in the regulation of PcG complexes, we constructed the luciferase reporters in which YY1-2, YY1-3, or the CpG island was deleted from the BmASNS promoter construct, respectively ( Figure 6A).
The luciferase promoter assay showed that removal of the CpG island significantly increased the luciferase activity compared to that of the YY1-3 deletion construct, and that the levels of both activities were higher than that by the deletion of YY1-2, which findings together suggested a critical repressive effect of the CpG island and YY1-3 on the BmASNS promoter activity ( Figure 6B). BmC/EBP RNAi greatly decreased the promoter activity in each of these deletion constructs. This might be because there are only two C/ebp-binding sites between the region of YY1-3 and YY1-2 ( Figure 2B) and the remaining C/ebp-binding sites could provide enough binding opportunity to BmC/ebp protein. Importantly, deletion of the YY1-2 site increased the promoter activity, but this construct still responded to RNAi for BmPHO or BmSCM ( Figure 6B and 6C). In contrast, the activity in constructs resulting from deletion of the CpG island or YY1-3 exhibited a partial loss of the responses to BmPHO or BmSCM silencing. These findings indicated that the CpG island within the BmASNS promoter is involved in the PcG-mediated repression of BmASNS expression and may facilitate the binding of BmPho protein to the YY1-3 site near the H3K27me3-enriched P2 region, and the YY1-2 site may play only an auxiliary role in the repression of BmASNS expression. Bombyx PcG proteins counteract BmC/ebp activity to repress BmASNS expression PcG proteins and BmC/ebp were demonstrated to act as repressors and to act as an activator, respectively, of BmASNS expression. We further examined whether these two regulators could interact with each other to act on the BmASNS promoter. First, we performed a co-immunoprecipitation experiment that confirmed an interaction between BmPho and BmScm ( Figure 7A), as reported previously [17]. However, neither BmPho nor BmScm interacted with BmC/ebp ( Figure 7A), suggesting that the repression and activation of the BmASNS promoter are accomplished independently by two complexes.
We then analyzed the effects of double knockdowns of these repressors and activator on BmASNS promoter activity. As expected, the depletions of both BmPHO and BmSCM increased the promoter activity ( Figure 7B, Lane 2). Interestingly, the depletions of both BmPHO and BmC/EBP resulted in a higher BmASNS expression than the BmPHO and BmSCM depletions (Lane 3), while knockdowns of BmSCM and BmC/EBP (Lane 4) resulted in a slight reduction compared with BmPHO and BmC/ EBP knockdowns, but still higher than the control treatment, and together these findings indicated the significant repressive effects of the PcG proteins on BmASNS promoter activity. It should also be noted that there was a reduced expression of BmC/EBP in response to the BmPHO and BmSCM depletions (Lane 2), although the mechanism of these effects was unknown. However, this weak decrease of BmC/EBP did not affect the up-regulation of BmASNS expression (Lane 2) because the higher expression of BmASNS was observed in the other two lanes (Lanes 3 and 4) when suffered from a significant loss of BmC/EBP expression. In contrast to the transcription from the chromosomal BmASNS locus, the effect of BmPHO knockdown on the deletion constructs was different from that of BmSCM RNAi in the absence of BmC/ebp ( Figure 7C). These facts implied that the DNA-binding protein of BmPho has a dominant effect on the negative regulation of BmASNS expression; namely, BmPho alone is enough to exclude the putative transactivators other than BmC/ebp. Moreover, the formation of PcG complexes at the BmASNS locus via the BmPho recognition can antagonize the BmC/ebp-mediated activation, since BmPHO and BmSCM RNAi in the presence of BmC/ebp caused similar upregulation of the promoter activity (Figure 2 and Figure 4).
To further estimate the relative power balance between these two transcriptional regulators, we performed RNAi assays for PcG genes under the enforced expression of the BmC/EBP gene. Overexpression of BmC/EBP up-regulated the BmASNS promoter activity on all the deletion constructs (see Figure 7D and 7C). Interestingly, the knockdown of BmPHO or BmSCM still significantly increased transcription of the p4974 and p3614 constructs under the condition of BmC/EBP overexpression, and slightly increased transcription of the p2285, but not of the p1056 construct. These results showed that the BmC/ebp transactivator can basically compete with PcG proteins in the presence of the CpG island region.
Bombyx PcG proteins regulate BmASNS expression at the appropriate time during the cell cycle The above results have demonstrated that the PcG system is involved in the repression of BmASNS expression, but there remains a question as to why PcG complexes are able to regulate a ''housekeeping'' gene that differs from other well-documented PcG targets such as the Hox genes in Drosophila as well as in mammals. The repression might not proceed in the canonical fashion of PcG regulation on the BmASNS gene.
Following the clue that human cells can be induced to express the ASNS gene at the G1 phase or before DNA replication [26], we examined the expression profile of the BmASNS gene during the cell cycle of silkworm cells. Because there was no effective method to synchronize the cell progression in silkworm cells, we developed a novel strategy that took advantage of cell cycle-regulated factors with the ability to induce cell arrest at different phases upon RNAi treatment [27]. After checking the knockdown efficiency of each gene in the BmN4-SID1 cells by semi-quantitative PCR experiments ( Figure S3), we further confirmed, using flow cytometry analysis, that the cell cycle distribution in control treatment was consistent with the previous pattern [27,28], and that cells lacking the BmMYC gene primarily accumulated at the G1 phase of the cell cycle, and BmCDT1 knockdown resulted in an arrest at the late-G1 phase, whereas BmRNRS RNAi could affect entry of cells into the G2 phase. We also observed that the loss of BmCDK1 would arrest cells at the G2/M phase and the loss of BmCDC27 led to an effective arrest at the metaphase ( Figure 8A) [28]. Based on the arrest of these cells, a semi-quantitative PCR estimation revealed that, as shown in Figure 8B, the expression levels of BmASNS were very low at the M phase (BmCDC27 knockdown), and after cell division, the expression was up-regulated, particularly at the late-G1 phase (BmCDT1 knockdown), consistent with the profile of the ASNS gene in human cells [26]. These observations confirmed that the BmASNS expression level differs during different phases of the cell cycle, and indicated that the PcG system allows timely control of BmASNS expression to accommodate the needs of cells.  To test this possibility, we further performed separate knockdowns of BmPHO, BmSCM, or BmC/EBP in the knockdown cells from the foregoing experiment, respectively. As shown in Figure 8C, when cells were arrested at the G1 or S phase (BmMYC, BmCDT1, or BmRNRS RNAi), the concomitant silencing of BmPHO or BmSCM did not greatly change the expression of the BmASNS gene. By contrast, a significant increase of BmASNS levels were shown in the cells arrested at the G2/M phase (BmCDK1 or BmCDC27 RNAi) when there was an additional depletion of BmPHO or BmSCM. On the other hand, the loss of BmC/EBP expression only caused significant decrease of BmASNS transcription in the cells that underwent BmMYC, BmCDT1, or BmRNRS RNAi treatment, but not in those that underwent BmCDK1 or BmCDC27 RNAi treatment. Collectively, these results revealed that PcG proteins cooperate with BmC/ebp to regulate the BmASNS promoter activity in a cell cycle-dependent manner.

Discussion
In this study, we report the first evidence that the PcG system directs gene regulation in a given locus in Bombyx and elucidate a new mechanism under the regulation of the PcG complexes on its target gene BmASNS, namely, PcG complexes can counteract with the C/ebp transactivator in the specific cell phase of the cell cycle through the cis-regulatory elements of YY1-binding motifs and the CpG island present on the BmASNS promoter.
Drosophila PREs containing YY1-binding motifs for Pho binding serve as a platform for the binding of the Pho protein and other PcG proteins that participate in the modification of histone and compaction of chromatin in the local region [29]. Moreover, the YY1-binding sites in the silkworm BmASNS promoter have also been shown to be involved in recruiting the PcG complexes. However, what is different from the model in Drosophila is that the CpG island in the BmASNS promoter could facilitate this repressive effect from PcG proteins through a potential mechanism of increasing the quantity of PcG proteins recruited or elevating the repressive H3K27me3 marks on the neighboring histone. This is in agreement with the well-known observation on the involvement of CpG islands in the regulation of PcG complexes in mammals [14,15]. It remains to be determined whether such CpG islandmediated regulations are common in insects.
In Drosophila and mammals, the PcG system generally controls the expression of target genes by influencing the status of H3K27 methylation deposited by the PRC2 complex in their promoter regions. Our ChIP assays revealed a high level of H3K27me3 on the BmASNS promoter, especially near the CpG island. Moreover, the H3K27me3 was maximally reduced after down-regulating the expression of PRC2 components, revealing that it is the PRC2 complex rather than other complexes that contributes to the trimethylation of H3K27 in Bombyx. Thus, it is reasonable to consider that the formation of a condensed chromatin structure through H3K27me3 mediated by the PRC2 complex will result in a high repressive effect of the PcG system on the prompter activity, just as in our previous observation of the depletion of BmESC [17], one component of the PRC2 complex. In addition, it is noted that H3K27me3 can be inherited as epigenetic memory at the chromosomal locus [30], and may provide a repressive signal for the subsequent restructuring of the PcG system mediated by the PRC1 complex [31]. However, because there is no epigenetic memory in the episomal construct newly introduced into the cell, our results demonstrated that BmPho or BmScm should play a more critical role in the creation of new epigenetic memory at the episomal construct, compared to the chromosomal locus. We know that this hypothesis is not sufficient to explain why the knockdown of each PcG gene induces similar effects on the chromosomal locus. In other words, this observation raises the possibility that the repression of BmASNS may be ascribed to the binding of PcG proteins per se or the established H3K27me3. We have previously suggested the presence of at least two types of regulatory complexes, namely, the whole PcG system and Pho/ Scm complex [17]. Taken together with the current results, this finding makes it apparent that the Bombyx PcG system is not enough to suppress the BmASNS promoter activity without the continuous expression of the BmPho or BmScm protein, suggesting that the complex of BmPho/BmScm will be required to facilitate this repression from the intact PcG complexes. Indeed, we observed a similarly repressive effect of BmPho or BmScm overexpression on BmASNS promoter activity using the episomal constructs. Therefore, all these findings suggest that the removal of PcG proteins from this region is more crucial than the loss of histone modification to achieve the activation of BmASNS expression.
In addition to H3K27 methylation on the target promoter mediated by PcG complexes, some studies have also shown that several mammalian PcG proteins, such as EZH2 as a mammalian PRC2 component and CBX7 as one component of the PRC1 complex, could recruit DNA methyltransferases (DNMTs) on the promoter of target gene to repress its expression [32,33]. Therefore, it would be of interest to examine whether the repression of BmASNS expression from PcG complexes could also be regulated by DNMTs, though it should be noted that the silkworm genome was shown to be poorly methylated in a previous study [34]. Our present data excluded this possibility because the knockdowns of Bombyx DNMT1 and DNMT2 (only two DNMT genes are found in the silkworm genome) could not induce the expression of BmASNS ( Figure S4), indicating that the repression of BmASNS should be attributed only to H3K27 methylation from the PcG system rather than to DNA methylation from the DNMTs.
A recent study reported that the mouse ASNS promoter contained the di-methylated histone H3 lysine 4 (H3K4me2) and tri-methylated histone H3 lysine 9 (H3K9me3) [35]. H3K9me3 is associated with the formation of constitutive heterochromatin, which contributes to the repression of gene expression, whereas H3K4me2 at the promoter is related to the activation of gene transcription. Although there has been no report concerning an increase of H3K4me2 in response to amino acid or protein deprivation, contrarily, increased H3K9me3 level and elevated acetylation of histone H4 are observed [35], suggesting that the multiple histone modifications on the mouse ASNS promoter may be involved in the regulation of its own expression. To determine whether the H3K9me3 also contributes to the regulation of BmASNS expression, we performed RNAi against Bombyx H3K9me3-related genes, BmHP1a, BmHP1b, and BmSU(-VAR)3-9 [36]. The semi-quantitative PCR examination of BmASNS expression revealed no significant changes compared with the control RNAi treatment ( Figure S5). Although this observation cannot directly rule out the possibility that the H3K9me3 modification also appeared on the BmASNS promoter, it points to a role of PcG complexes-mediated H3K27me3 as a key repressive modification for tightly modulating the transcription of the Bombyx BmASNS gene.
Genome-wide expression analysis in mouse hematopoietic stem cells has shown that PcG proteins and C/ebp can regulate a large set of genes in a positive or negative manner [21]. In this study, we identified that Bombyx PcG proteins negatively regulate the BmASNS expression by counteracting the transactivator BmC/ ebp. This case of the BmASNS gene in Bombyx is in agreement with the negative correlation of PcG proteins and C/ebp on target gene expression in mouse [21], and therefore suggests a common model in which most target genes orchestrated by PcG complexes tend to be cooperatively regulated by the C/ebp protein. It would be worth further investigation to determine the target genes of BmC/ ebp in a genome-wide scale and to analyze their correlations in conjunction with our microarray data from the knockdowns of PcG genes [17].
Importantly, we have demonstrated that PcG proteins play roles in the regulation of BmASNS promoter activity at the specific phase of the cell cycle and provide a model for the dynamic regulation of BmASNS expression involved by PcG repressors and BmC/ebp activator as shown in Figure 8D. According to this model, before cells enter into the S phase, uptake of various nutrient elements, including amino acids, is required. To increase the BmASNS transcription at the late G1 phase, the cells have to remodel the chromatins around the BmASNS promoter followed by the release of PcG complexes and the increase of accessibility of the promoter to transcriptional activators, including BmC/ebp. We speculate that a small number of H3K27me3 marks will be maintained in the promoter at this cell phase, since the previous studies have revealed that tri-methylation of H3K27 can be maintained at sites of DNA replication and even during the cell division [37], and moreover, the remaining H3K27me3 can help the PRC1 complex recognize the locus on the BmASNS promoter in the next cell phase. After completion of the S phase, the expression of the BmASNS gene will return to a low level mediated by the rebinding of the PcG system and/or the Pho/Scm complex at the G2 and M phases. The presence of PcG complexes may in turn increase the levels of H3K27me3 and promote the compaction of local chromatins, whereas the Pho/Scm complex assists in the repression by the whole PcG complexes. The comprehensive impact of this procedure will lead, to a certain extent, to the loss of BmC/ebp binding and then attenuate the promoter activity. Notably, it will require a balance between the repressor and activator to maintain this expression state. This appears to be a reasonable model for the regulation of BmASNS by these two distinct complexes, although it remains to be determined what signal triggers the release of PcG complexes from this locus.
In conclusion, the present results have clarified a novel epigenetic regulation in which PcG complexes regulate BmASNS expression involving H3K27me3. Our data confirmed that PcG proteins suppress the transcription of the BmASNS gene by recruiting themselves to the putative YY1-binding motifs and the CpG island within the BmASNS promoter. It is therefore tempting to speculate that both the YY1-binding element and the CpG island are sufficient to recruit PcG complexes and subsequently deposit H3K27me3 to repress the target gene expression in Bombyx. In particular, this study provides important new insights into the mechanism underlying the dynamic regulation of PcG target gene by the PcG system during the cell cycle. Our data will also shed fresh light on the regulation of the ASNS gene in human, which could explain a potential mechanism for the tumorigenesis mediated by the PcG system via the ASNS functions.
For the amino acid deprivation assay, the modified IPL-41 medium was prepared according to the manufacturer's protocol (Sigma) but without addition of all amino acid components.

RNA extraction and semi-quantitative PCR
Total cellular RNA was isolated and reversed according to the procedure described previously [16]. The expression profile of the BmASNS gene was evaluated by semi-quantitative polymerase chain reaction (semi-quantitative PCR) using gene-specific primers (Table S1), and the silkworm glyceraldehyde-3-phosphate dehydrogenase (BmGAPDH) gene was used as an endogenous control. The relative expression levels of the BmASNS gene were quantitated by ImageJ software and normalized to the BmGAPDH levels. Cloning of the silkworm ASNS gene Based on the reported BmASNS gene sequence (NP_001037414.1), we designed primers to amplify and obtain the cDNA clone (Table S1). The cDNA product then was digested by XhoI (underlined in Table S1), cloned into a pENTR TM 11 (Invitrogen) vector, and verified by DNA sequencing.

Isolation of the silkworm ASNS promoter region
The upstream promoter region of BmASNS was obtained by PCR screening of a genomic DNA template extracted from the silkworm strain p50T using a set of primers (Table S1) to yield a 4,974 bp sequence (AB751507). Here, we designated the translation start site of the BmASNS gene as +1, and thus this fragment was 24974/-1. The resulting product was digested with ApaI (underlined in Table S1), and inserted into the ApaI-Eco47III site of pXINSECT-DEST38 (Invitrogen). The recombinant plasmid was termed pASNS-DEST38 and its nucleotide sequences were determined by DNA sequencing.
For the reporter assay, the reporter plasmid (pASNS-Luc) was made by recombinant reaction between pASNS-DEST38 and pENTR TM 11-Luc according to the gateway reaction system. pENTR TM 11-Luc vector was generated using a pENTR TM 11 (Invitrogen) vector into which the Firefly luciferase reporter gene was cloned [39].
For the constructs with specific deletion of BmASNS promoter region, such as pASNSDYY1-3-Luc, pASNSDCpG-Luc, and pASNSDYY1-2-Luc, the direct amplification was performed based on a pASNS-Luc plasmid using the primers in Table S1. Then the PCR product was directly ligated.

RNA interference
Knockdown of genes was performed with the gene-specific double-stranded RNAs (dsRNAs) in BmN4-SID1 cells according to our previous procedure [16], and the dsRNA for the enhanced green fluorescent protein (EGFP) gene was used as a control in this study. The knockdown efficency for each RNAi gene was quantitated by ImageJ software and normalized to the BmGAPDH levels.

Transient transfection and luciferase assay
For normal transfection, the BmN4 cells seeded in 24-well plates with a density of 0.5610 5 were transfected with the luciferase reporter vectors mentioned above or the control promoter-less vector pENTR TM 11-Luc according to our previous method [17]. Cells were harvested 72 h after transfection for luciferase activity assay.
For RNAi transfection, the BmN4-SID1 cells were pre-cultured in 24-well plates with a density of 0.2610 5 and additional dsRNAs according to the requirements of experiment. After 72 h of incubation, the cells were transfected and analyzed with the same method described above.
The transfection efficiency between dishes was measured by cotransfection with the pEXP38-bgalDIE-1 vector expressing a bgalactosidase, and the b-galactosidase activity was used to normalize luciferase activity data [17]. All experiments included at least three independent transfections and data are shown as the mean 6 standard deviation (SD).

Co-immunoprecipitation assay
Co-immunoprecipitation (Co-IP) was performed according to our previous procedure [17].

Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (ChIP) analysis was performed according to a modified protocol of NimbleGen Systems, Inc. (Roche). Briefly, BmN4 cells or BmN4-SID1 cells (treated with dsRNAs) were harvested by gentle centrifugation and resuspended in growth medium on ice for 10 min. Cells were cross-linked in a 1/10 volume of crosslinking solution and mixed gently by inversion. The reaction was stopped by the addition of a 1/20 volume of 2.5 M glycine solution. Cross-linked chromatin was sonicated for 30 min with cycles of 30 s on/30 s off at an M2 setting using a Biorupter sonicator (Diagenode) to produce chromatin fragments of 0.2,0.5 kb on average. After centrifugation and quantification, 100 mg of the supernatants was used as an input and 1 mg of the supernatants was incubated with 2 mg of H3K27me3 antibody (Upstate, Cat. #17-622). A rabbit IgG (Upstate, Cat. #PP64B) was used as the nonspecific antibody control. The antibody-bound complex was precipitated by DynabeadsH Protein G (Invitrogen, Cat. no. 100.03D). The DNA fragments in the immunoprecipitated complex were captured using an MPC-S magnet and released by reversing the crosslinking at 65uC for 12 h, and then purified using a QIAquick PCR purification kit (Qiagen).

Quantitative real-time PCR analysis
Purified DNA was amplified by quantitative real-time PCR (qPCR) using SYBR Green Master Mix (Applied Biosystems) on a Thermal Cycler Dice Real Time System (Takara). The primers used for the qPCR to amplify the BmASNS promoter sequence are shown in Table S1. All ChIP assays were performed in triplicate, and the averaged data are presented with standard errors.

Flow cytometry assay
Cell cycle distributions were analyzed by measuring the cellular DNA content using a Flow cytometer (Millipore) according to the previously described procedure [28].

Statistical analysis
The statistical significance of difference between the treated and the corresponding control was evaluated by the Student's t-test, and a P-value ,0.05 was considered statistically significant.

Author Contributions
Conceived and designed the experiments: ZL TK. Performed the experiments: ZL. Analyzed the data: ZL DC JML QX TK. Contributed reagents/materials/analysis tools: ZL HM LZ JX TT JML. Wrote the paper: ZL DC QX TK.