Regulation of HOXA2 gene expression by the ATP-dependent chromatin remodeling enzyme CHD8

2.1. Cell culture and reagents 

Dulbecco’s Modified Eagle Medium (DMEM) (Invitrogen) with 10% fetal bovine serum (Hyclone) and 1X penicillin–streptomycin–glutamine (Invitrogen) was used to culture NTERA2cl. D1 (NT2/D1) cells at 37°C in 10% CO2. For ATRA induction, cells were treated with 10μM ATRA while control cells were left untreated. Cells were grown an additional 24h before harvesting.

CHD8 antibodies were previously described [10]. The anti-Flag M2 (F3165) antibody and rabbit normal IgG immunoglobulin (I8140) were purchased from Sigma. The anti-RbBP5 (A300-109A) and anti-Ash2L (A300-489A) antibodies were purchased from Bethyl. The anti-WDR5 (22512-100) antibody was purchased from Abcam. The anti-histone H3 (39163) and the anti-trimethyl histone H3 lysine 4 (39159) antibodies were purchased from Active Motif. All oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA).

2.2. Chromatin immunoprecipitation (ChIP) assay 

ChIP experiments were performed essentially as described in the ChIP Assay Kit (Millipore) as described previously [10].

2.3. RT-PCR and quantitative PCR (qPCR) 

Total RNA was isolated using the RNeasy Kit (Qiagen). Reverse transcriptase reactions employed random decamers (Ambion) and Superscript II (Invitrogen) following manufacturers’ protocols. qPCR analysis employed the indicated primers (Table S1), the iQ SYBR Green Supermix or iQ Supermix (Bio-Rad), and a MyiQ Real-Time PCR Detection System (Bio-Rad). All PCR reactions were performed in triplicate. For RNA analysis, Ct values were normalized to the levels of the Pol III transcribed H1 RNA while ChIP experiments were expressed relative to input.

2.4. RNAi knockdown experiments 

RNAi experiments utilized the UI2-SIBR short hairpin RNA (shRNA) vectors [11] modified to express the alpha subunit of the interleukin 2 receptor as a selection marker (Table S1). CHD8 and control shRNA cassettes were described [10]. Ten micrograms were transfected into cells using lipofectamine 2000 (Invitrogen) as described by the manufacturer. Twenty four hours post-transfection, cells were either treated with ATRA or not and grown an additional 24h before harvesting. Cells were trypsinized and resuspended in DMEM. Twenty microliters of Dynabeads-CD25 slurry (Invitrogen) were added and rotated at 4°C for 30min. Following two PBS washes, the cells were lysed on the beads for RNA extraction or Western blotting, or, for ChIP experiments, crosslinked on the beads in 1% formaldehyde for 10min.

3.1. CHD8 forms a complex with the core WDR5/Ash2L/RbBP5 complex 

Previous work from our lab has identified several CHD8 associated polypeptides [10]. We reported that CHD8 can interact with WDR5 both in vivo and in vitro. Initially we focused in on WDR5 as this polypeptide has been reported to be part of a MLL histone methyltransferase complex [4], [5], [6], [7], [8], [9], possibly linking chromatin remodeling and chromatin modification in a single complex. Further evidence for this link was provided by the discovery of CHD8 in a WDR5-containing complex [9]. As WDR5, Ash2L, and RbBP5 (WAR) form a subcomplex in the absence of MLL [12], [13], we asked whether CHD8 could also associate with WDR5 in the context of the subcomplex. This association was tested through the reconstitution of the complex using a baculovirus expression system. Co-infection of SF9 cells with WDR5, Ash2L, RbBP5, and CHD8 followed by affinity purification demonstrated that indeed CHD8 can associate with WDR5 in the context of the WAR subcomplex (Fig. 1A).

View Large Image Download to PowerPoint

Fig. 1. CHD8 directly interacts with the core WDR5/Ash2L/RbBP5 complex. (A) Cellular extracts were prepared from SF9 cells following co-infection with the indicated viruses. Immunoprecipitations were performed with anti-Flag-M2 or control beads. After extensive washing, purified samples were subjected to SDS–PAGE followed by Western blotting analysis using the indicated antibodies. (B) Pairwise coinfections were performed as in (A) using the indicated baculoviruses.

To systematically test the assembly of the WAR/CHD8 complex, further baculovirus co-infection experiments were performed, each time omitting one component. As shown in Fig. 1A, the removal of any one of the subunits does not preclude a stable association of the remaining subunits, although omission of Ash2L results in a diminished quantity of associated RbBP5. To further explore these interactions, pairwise baculovirus co-infection experiments were performed. As shown in Fig. 1B, CHD8 is capable of interacting directly with each component of the WAR complex. These results demonstrate that CHD8 has extensive contacts with the WAR subcomplex, unlike the interactions with MLL that are dependent on WDR5 [12], [13], [14], [15]. The precise means by which these interactions occur, including interacting domains between each protein, remains to be elucidated.

3.2. CHD8 is bound to the promoters of the HoxA genes 

Previous studies have demonstrated that WAR is recruited to the HOX locus and is an important regulator of HOX gene expression [8], [9], [12], [13]. As CHD8 has been shown to interact with WAR [9], [10], we investigated the possibility that CHD8 is also recruited to the HOX locus where it could function in conjunction with WAR in the regulation of transcription. The NT2/D1 embryonal carcinoma cell line was chosen as treatment of these cells with ATRA induces the expression of many genes, including the HOX genes [16], and ultimately induces cellular differentiation into neuronal-like lineages [17]. Treatment of NT2/D1 cells with ATRA results in over 500 fold activation of HOXA1, and ∼75-, ∼10-, and ∼4-fold activations of HOXA2-A4, respectively (Fig. 2A). This system therefore presents an opportunity to study the role of CHD8 in Hox gene activation.

View Large Image Download to PowerPoint

Fig. 2. CHD8 is bound to the HoxA gene cluster. (A) NT2/D1 cells were treated with ATRA for 36h. Cells were harvested and RNA was isolated. Expression of the indicated genes was analyzed by RT-PCR. (B) Chromatin from NT2/D1 cells was crosslinked with formaldehyde. Cells were lysed, and ChIPs were performed with anti-CHD8 antibodies. Bound DNA was detected by quantitative PCR with primer pairs to the promoter region of the indicated gene. (C) ChIP experiments were performed using NT2/D1 cells that had been treated with ATRA or not. Results are expressed relative to untreated cells. Control IgG precipitated samples in all experiments were less than 0.001% of input and not shown.

To test whether CHD8 is bound to these promoters, we first examined CHD8 occupancy in uninduced cells using ChIP assays. We found CHD8 bound to the HOXA1 and HOXA4 promoters, and to a lesser extent, the HOXA2 and HOXA3 promoters (Fig. 2B). Due to the upregulation of these HoxA genes upon treatment with ATRA, it is reasonable to suspect that proteins involved in the regulation of these genes would associate with these promoters in a dynamic manner. Therefore, we compared the occupancy of CHD8 at the HOXA1 through HOXA4 promoters before and after ATRA treatment. As shown in Fig. 2C, 24h treatment with ATRA causes a striking relocalization of CHD8 at the HOXA1 through HOXA4 promoters, with the greatest change being the ∼20-fold increase of CHD8 at the HOXA2 promoter. This result suggests that CHD8 is indeed functioning in gene regulation at the HoxA locus.

3.3. CHD8 regulates HoxA gene expression 

To directly test the effect of CHD8 upon the regulation of the HoxA locus, an shRNA strategy was employed to deplete endogenous CHD8. NT2/D1 cells were transfected with control shRNA constructs or shRNA constructs targeting CHD8. Following selection of transfected cells, cDNA was isolated for analysis by qPCR. Using this selection protocol, we were unable to detect measureable levels of HoxA gene expression in the absence of ATRA; therefore, all experiments were performed with ATRA treatment.

Depletion of endogenous CHD8 results in an increased expression of HOXA2, while the expression of HOXA1, 3, and 4 does not change in a statistically significant manner (Fig. 3A). The efficacy of CHD8 knockdown is demonstrated via Western blot in Fig. 3B. It is important to compare the increase of HOXA2 transcription in Fig. 3A to 2C which illustrates that, upon ATRA treatment, the HOXA2 promoter showed the greatest increase in CHD8 occupancy. Taken together, these data demonstrate that CHD8 indeed functions in the regulation of HOXA2 gene expression. Furthermore, under activating conditions at the HOXA2 promoter, the normal function of CHD8 appears to be the negative regulation of HOXA2 gene transcription. This negative effect of CHD8 on HOXA2 expression is consistent with our prior studies of CHD8 and β-catenin mediated transcription where depletion of CHD8 results in an increase in target gene expression under activating conditions [10].

View Large Image Download to PowerPoint

Fig. 3. CHD8 regulates expression of the HoxA gene cluster. (A) NT2/D1 cells were transfected with control shRNA or CHD8 shRNA and induced with ATRA. Following selection, cells were harvested and total RNA was isolated. Expression of the indicated genes was analyzed by RT-qPCR. Results are normalized first to a reference RNA and then to the control shRNA transfection. ∗P<0.03 by Student’s t-test. (B) Western blot of control transfected shRNA (GFP) or CHD8 transfected shRNA cellular lysate. Actin was used as a loading control.

3.4. CHD8 is required for recruitment of the WAR complex 

We have previously shown CHD8 interacts with several proteins including the WAR complex. Because WAR is also involved in establishing methylation patterns at target genes [8], [9], [12], [13], we next asked whether CHD8 is involved in the recruitment of WAR to the promoters of target genes. We tested this hypothesis via knockdown of CHD8 and chromatin immunoprecipitation as above. In our hands, antibodies to Ash2L were the most suitable for these experiments. Because Ash2L has been demonstrated to be vital for WAR complex function, the absence of Ash2L is indicative of a lack of functional WAR at that gene. As expected, upon treatment with control shRNA, we are able to detect Ash2L at the HOXA2 promoter (Fig. 4). However, when CHD8 is depleted via shRNA treatment, Ash2L is lost from the promoter. This result is indicative of CHD8 being required for proper WAR complex recruitment and function. As WAR is a component of multiple histone methyltransferase complexes, CHD8, though the recruitment of Ash2L, could be required for proper establishment of methylation patterns at target genes.

View Large Image Download to PowerPoint

Fig. 4. CHD8 is required for the recruitment of Ash2L to the HOXA2 promoter. NT2/D1 cells were transfected with control shRNA or CHD8 shRNA and treated with ATRA. Following selection of the transfected cells, cells were cross linked using formaldehyde. Cells were then lysed, following which ChIP experiments were performed using the indicated antibodies. Bound DNA was detected by quantitative PCR with primer pairs to the promoter region of HOXA2. Control IgG precipitated samples were less than 0.001% of input and therefore are not shown. ∗P<0.05 by Student’s t-test.

3.5. CHD8 affects histone H3 methylation patterns at the HOXA2 promoter 

As mentioned previously, the WAR complex is known to associate with all members of the SET1 family of histone H3 lysine 4 methyltransferases, playing an important regulatory role in gene expression, including transcription at the Hox loci [18]. In conjunction with the data demonstrating a requirement for CHD8 in Ash2L recruitment, a suggested role for CHD8 functioning in the regulation of histone methylation patterns at target genes emerges. We performed chromatin immunoprecipitation with H3K4me3 antibodies to look for changes in methylation at the HOXA2 promoter following knockdown of CHD8 and ATRA treatment. Consistent with the loss of Ash2L, we found a significant decrease in H3K4me3 levels at the HOXA2 upon depletion of CHD8 (Fig. 5A).

View Large Image Download to PowerPoint

Fig. 5. CHD8 affects H3K4 trimethylation at the HOXA2 promoter. (A and B) ChIP experiments were performed as in Fig. 4 using the indicated antibodies. Bound DNA was detected by quantitative PCR with primer pairs to the promoter region of HOXA2. Control IgG precipitated samples were less than 0.001% of input and therefore are not shown. ∗P<0.05 by Student’s t-test and ∗∗P<0.01 by Student’s t-test.

One possible explanation for the decrease in K4 trimethylation is that CHD8 functions in histone deposition at the HOXA2 promoter, i.e. the loss of observed methylation is due to loss of total H3 at this promoter. To address this, we again performed chromatin immunoprecipitation experiments after depletion of CHD8 by shRNA and treatment with ATRA. As shown in Fig. 5B, H3 occupancy at the HOXA2 promoter increases upon CHD8 depletion. This result suggests that the regulation of transcription by CHD8 occurs via multiple histone methylation/demethylation events at the promoters of target genes.