Transcription factor EKLF (KLF1) recruitment of the histone chaperone HIRA is essential for β-globin gene expression

Significance

The β-like globin locus has provided a long-standing model for the study of cell-specific and developmental control of transcription and chromatin structure. We have previously shown that the replication-independent histone H3.3 variant is enriched at the active adult β-globin promoter. Although the erythroid Krüppel-like factor (EKLF), which is also known as Krüppel-like factor 1 (KLF1), transcription factor interacts with histone H3, it does not distinguish between the H3.1 or H3.3 variants. We now show that EKLF interacts with the H3.3 chaperone named histone cell cycle regulation defective homolog A (HIRA) and that this enables its selective recruitment of HIRA to the promoter. To our knowledge, our studies implicate HIRA for the first time in establishment of erythropoiesis and explain how critical protein interactions can lead to directed changes in histone variants at restricted sites.

Abstract

The binding of chromatin-associated proteins and incorporation of histone variants correlates with alterations in gene expression. These changes have been particularly well analyzed at the mammalian β-globin locus, where transcription factors such as erythroid Krüppel-like factor (EKLF), which is also known as Krüppel-like factor 1 (KLF1), play a coordinating role in establishing the proper chromatin structure and inducing high-level expression of adult β-globin. We had previously shown that EKLF preferentially interacts with histone H3 and that the H3.3 variant is differentially recruited to the β-globin promoter. We now find that a novel interaction between EKLF and the histone cell cycle regulation defective homolog A (HIRA) histone chaperone accounts for these effects. HIRA is not only critical for β-globin expression but is also required for activation of the erythropoietic regulators EKLF and GATA binding protein 1 (GATA1). Our results provide a mechanism by which transcription factor-directed recruitment of a generally expressed histone chaperone can lead to tissue-restricted changes in chromatin components, structure, and transcription at specific genomic sites during differentiation.

Efficient packaging of DNA in a highly organized chromatin structure inside the cell is a remarkable characteristic of all eukaryotic organisms. Chromatin assembly is a stepwise process that requires histone chaperones to deposit histones in forming nucleosomes. These chaperones are essential to facilitate ordered assembly of nucleosomes for both replication-dependent and -independent events (1).

The incorporation of histone variants also modulates chromatin dynamics. Histone variant H3.3 differs from its canonical counterpart H3.1 in only five amino acid residues but has a very distinct function. H3.3 is incorporated into nucleosomes independent of DNA replication and also serves as an epigenetic mark of active chromatin (2, 3). H3.3 incorporation into nucleosomes contributes to their destabilization, thus facilitating transcription (4). Hence, H3.3 is highly enriched at active promoters and gene bodies of actively transcribed genes and also at regulatory sites of both active and inactive genes (5–7). The importance of H3.3 is further highlighted by the recent observation that H3FA3, one of the two genes encoding H3.3, is mutated in pediatric malignant brain tumors and that these mutations are proposed to drive tumor formation (8, 9).

H3.3 deposition is mediated by distinct factors at specific genomic regions, with histone cell cycle regulation defective homolog A (HIRA) being involved in deposition of H3.3 at promoters and in the body of active genes (10–12). Further supporting these observations, the biochemical isolation and characterization of protein complexes containing preassembled histone H3.1 and H3.3 from human cells revealed that H3.1 associates with the chaperone chromatin assembly factor-1, whereas H3.3 is incorporated into chromatin by the HIRA chaperone complex consisting of HIRA, CABIN1, and UBN1 together with the histone-binding protein ASF1 (10, 13–17).

In mammals, HIRA was originally identified as a gene that is deleted in DiGeorge syndrome (18). HIRA plays an important role at gastrulation, and HIRA-null mice are grossly abnormal and die early in embryogenesis, suggesting that it is essential for proper development and survival (19, 20). The precise nature of HIRA’s absolute requirement for vertebrate development, however, remains to be elucidated. HIRA family member proteins are characterized by seven tryptophan-aspartic acid (WD) repeats conserved at the amino terminus, predicted to form a β-propeller structure, and the presence of nuclear localization signals. The carboxyl-terminal region of HIRA is responsible for its interaction with proteins Pax-3 (21), HIRIP, and core histones (22). Although mutational studies have identified key residues that play an important role in specifying H3.3 deposition (3, 5), and more recently two key H3.3 residues were shown to be important for its recognition by DAXX protein (23, 24), it is still unclear how H3.3 is targeted to transcriptionally active regions.

Erythroid Krüppel-like factor (EKLF), which is also known as Krüppel-like factor 1 (KLF1), an erythroid cell-specific zinc finger protein, is a key player in activating mammalian β-globin gene transcription by virtue of its binding ability to its cognate CACCC sequence element at the β-globin promoter (25). Genetic ablation of EKLF leads to a loss of specific DNaseI hypersensitive site in the proximal β-globin promoter and a lack of DNase hypersensitivity at hypersensitive site 3 at the distal locus control region (26), indicating that EKLF is required for the chromatin reorganization at the β-globin promoter. EKLF-null embryos die of anemia at embryonic day (E)14.5, because definitive erythroid cells fail to produce β-globin transcripts in vivo, leading to a profound β-thalassemia (27, 28). It is now recognized that EKLF is a global regulator of erythroid gene expression (29–31) because its activation target repertoire includes protein-stabilizing, heme biosynthetic pathway, red cell membrane protein, cell cycle, and transcription factor genes in both primitive and definitive cells.

EKLF can undergo multiple modifications, including phosphorylation (32), sumoylation (33), ubiquitination (34), and acetylation (35, 36), and these in turn alter its ability to interact with modifiers (e.g., CBP/p300, Sin3A) and chromatin remodelers (e.g., SWI/SNF). Lys-288 acetylation is critical for recruitment of CBP to the β-globin locus, modification of histone H3, occupancy by EKLF, opening of chromatin structure, and transcription of adult β-globin (37). EKLF helps to coordinate this process by the specific association of its zinc finger domain with the histone H3 amino terminus. These interactions likely play a crucial role in establishing the correct 3D structure at the β-like globin locus (38) and transcription factories in vivo that enable efficient coordinate expression of select EKLF target genes (39).

Previous work from our laboratory demonstrated that the replication-independent H3.3, but not the replication-dependent H3.1, is enriched on the β-globin promoter after the induction of differentiation of erythroid MEL cells (37). Because only one of the five amino acid differences reside in this region, affinity differences with EKLF might not account for the differential H3.3 recruitment to the actively transcribing region of the globin gene.

Although broad distribution and binding correlations have been established for HIRA and H3.3 genome-wide (5, 12), a major unresolved question in the field is the mechanism by which HIRA and H3.3 are enriched at specific, developmentally critical sites, not just in the erythroid program but for any transcriptional output (5, 40). Studies showing selective H3.3 enrichment at the β-globin locus, the critical importance of EKLF for its optimal chromatin and transcriptional configuration, and its direct interaction with histone H3, all converge on the likelihood that these observations are operationally linked. With this in mind, we have identified a novel interaction between HIRA and erythroid-specific transcription factor EKLF by in vitro and in vivo approaches. Importantly, we also find that depletion of HIRA impairs hematopoietic development in mouse ES cells. Our data show that HIRA is not only required for transcriptional activation of globin genes but also for activation of erythropoietic regulators, such as EKLF and GATA-1, during erythroid differentiation.

Results

EKLF and HIRA Interact in Vivo.

Because only one of the five amino acid differences between the H3.1 and H3.3 variants reside in its region of interaction with EKLF (37), we tested whether a modified histone H3 might alter the interaction, particularly given that most of its modifications are localized to the amino tail (41) that overlaps the EKLF interaction region (37). However, using an in vitro array containing all known modifications of H3 and H4 (Active Motif), we find no discrimination by EKLF under conditions whereby the CBX7 chromodomain (42) discriminates its modified H3 targets (Fig. S1). As a result, we investigated whether EKLF might recruit histone H3.3 to the β-globin promoter via its chaperone, HIRA.

Coimmunoprecipitation assays were performed after cotransfection of Flag-tagged EKLF and HA-tagged HIRA (or their empty vector controls) into 293T cells. EKLF but not HIRA alone can be precipitated efficiently by the anti-Flag antibody (Fig. S2, Left). However, only in the presence of EKLF protein can the anti-Flag antibody also coprecipitate HIRA. To confirm this interaction, the immunoprecipitation was also performed with anti-HIRA. HA-tagged HIRA protein is precipitated efficiently by anti-HIRA antibody (Fig. S2, Right). However, EKLF can be precipitated by anti-HIRA antibody only in the presence of ectopic HIRA protein. These results indicate that EKLF readily interacts with HIRA after exogenous cotransfection in vivo.

The interaction between endogenous EKLF and HIRA was examined in two ways. First, we used murine erythroleukemia (MEL) cells that express EKLF and HIRA (25) (Fig. S3). MEL whole-cell extracts were immunoprecipitated with an anti-EKLF antibody (or the IgG antibody) and then blotted and probed with anti-EKLF and anti-HIRA antibodies. A human nonerythroid cell line that does not express EKLF but does express HIRA (293T) was used as a negative control. We find that endogenous EKLF and HIRA interact (Fig. 1A, Left). Second, cell extracts prepared from embryonic mouse fetal liver cells isolated at a stage when they are primarily erythroid (E13.5) were immunoprecipitated using anti-EKLF antibody or control IgG antibody. Our results show that EKLF interacts with HIRA in these primary erythroid cells (Fig. 1A, Right). Together these data support the cotransfection results and further demonstrate that endogenous EKLF and HIRA proteins interact in vivo.

Fig. 1.

Endogenous and deletion analysis of EKLF and HIRA interaction. (A) Whole-cell extracts from erythroid MEL and (untransfected) 293T cells (Left) or E13.5 murine fetal livers (Right) were prepared and subjected to immunoprecipitation with an anti-EKLF antibody or an IgG antibody, and immunoblots were probed with an anti-HIRA antibody. Ten percent of total protein extract was loaded to serve as “input.” For cotransfection studies in 293T cells, schematic diagrams of full-length EKLF or HIRA and the deletion constructs used are shown at the top of B, C, and D. Cells were transfected with HA-tagged HIRA, Flag-tagged full-length EKLF, its deleted derivatives, or vector control, as indicated. Samples were then immunoprecipitated with anti-Flag (B) or anti-HA (C and D) antibodies and probed with anti-EKLF, anti-HA, or anti-Flag (indicated on right). Expression of EKLFΔpro was separately analyzed on a higher percentage acrylamide gel (Fig. S4A). Ten percent of total protein extract was loaded to serve as “input.” ns, a nonspecific band observed after probing with anti-HA (D).

The EKLF Zinc Finger Domain Interacts with HIRA.

To map the subdomain of EKLF that is responsible for interacting with HIRA, two EKLF deletion constructs were used. The first construct, EKLFΔPro (Fig. 1B), contains the zinc finger domain of EKLF (amino acids 287–376) along with a Flag tag at the amino terminus. The second construct, EKLFΔZnF (Fig. 1C), contains the proline-rich domain of EKLF (amino acids 19–305; no tag). We performed two immunoprecipitations, one using the Flag antibody coupled to M2 agarose beads to pull down EKLFΔPro and another using a HA-specific antibody to pull down HIRA. The results show that HIRA can be immunoprecipitated with EKLF proteins that contain the zinc finger domain (full-length and EKLFΔPro) (Fig. 1B and Fig. S4A), but not with EKLF containing only its proline-rich domain protein (EKLFΔZnF) (Fig. 1C). These data suggest that the zinc finger domain of EKLF is primarily responsible for its interaction with HIRA.

Because the zinc finger domain is also the DNA binding domain of EKLF, we determined whether the ability of EKLF to bind to DNA and interact with HIRA could be uncoupled. For this we used a mutant EKLF that is compromised in its ability to bind to DNA (43). Wild-type EKLF or EKLF ZnF(m123) (a full-length variant containing mutations that alter critical histidines known to be important for coordinating zinc within each finger) were cotransfected with HA-tagged HIRA, immunoprecipitated using anti-EKLF, and analyzed using anti-HIRA. We find that the wild type and the zinc finger mutant of EKLF attain a slightly reduced level of interaction with HIRA after normalization with the input (Fig. S4B). We conclude that the DNA-binding and HIRA-binding activities of EKLF are separable.

EKLF Interacts with the HIRA Carboxyl Terminus.

Complementation analysis of chicken HIRA-null cells revealed that the amino- and carboxyl-terminal halves of HIRA have distinct roles resulting from region-specific interactions with chromatin and transcriptional modulators (44). To locate the binding region(s) between HIRA and EKLF, we constructed two truncation variants of HA-tagged HIRA containing either the amino-terminal half (amino acids 1–443, which overlaps the seven HIRA WD40 repeats) or the carboxyl-terminal half (444–1017). These constructs were transfected into 293T cells either alone or together with wild-type full-length EKLF. Although HIRA/1–443 exhibited no interaction, HIRA/444–1017 exhibited similar binding activity to that of the wild-type full-length HIRA protein (Fig. 1D). These data show that EKLF interacts with the region of HIRA that is also known to interact with CABIN1 and the bulk of the ASF1a binding B domain, but not with the WD40 region known to interact with UBN1.

HIRA Alters EKLF Transcriptional Activity at Erythroid Promoters.

To elucidate the functional consequence of the interaction between EKLF and HIRA, we analyzed its effect on EKLF transcriptional activity by cotransfection with a human β-globin gene promoter/luciferase reporter into K562 erythroleukemic cells (45). As expected, EKLF activates the reporter; although transfection of HIRA alone had no effect, it enhanced EKLF activity (Fig. 2) in a manner similar to that seen with CBP/P300 (36).

Fig. 2.

HIRA alters EKLF-mediated transcriptional activity of its target promoters. K562 cells were transfected with luciferase reporters containing the adult β-globin promoter together with expression plasmids for EKLF and HIRA. Luciferase activity was normalized against Renilla activity from a cotransfected control vector. The relative luciferase activity reflects the values obtained from triplicate experiments.

Recent studies of EKLF transactivation mechanisms have demonstrated that not all targets are equivalently affected (29). We find this to also be true for HIRA’s effect on EKLF. For example, EKLF activation of the BKLF promoter is enhanced in the presence of HIRA but is not at the AHSP or p21 promoters (Fig. S5). The β-globin and BKLF promoters share some similarities in their promoter architecture (e.g., INI and DPE elements), unlike the AHSP promoter (46). In addition, EKLF/TAF9 interactions are critical for β-globin but not AHSP promoter activity. We conclude that HIRA–EKLF interactions play an important role at selected erythroid promoters by modulating EKLF activity.

HIRA Is Required for Erythropoiesis in ES Cells.

To directly ascertain the role of HIRA in erythroid genetic regulation, we first compared normal ES cells and genetically modified ES cells lacking HIRA (47). We conducted a 6-d time course for the differentiation of ES cells into embryoid bodies (EBs) as a cell culture model of hematopoietic development (48) and evaluated the expression of β-globin and the transcriptional regulators GATA-2, GATA-1, and EKLF. GATA-2 plays a critical role early in hematopoietic development (49), whereas EKLF and GATA-1 are downstream targets that are essential for establishment of erythroid differentiation. EBs from HIRA-null cells are smaller in comparison with WT (Fig. S6A), but most importantly we find that the onset of β-globin mRNA is greatly reduced in differentiating HIRA^−/− ES cells (Fig. 3A). However, there is also a dramatic reduction in the levels of the erythroid regulators EKLF and GATA-1 (Fig. 3B) and in expression of the Ter119 terminal differentiation marker (Fig. S6B). Because there is no change in levels of the hematopoietic regulator GATA-2 in the HIRA^−/− ES cells compared with WT (Fig. 3B), we conclude that depletion of HIRA exerts an erythroid, but not hematopoietic, cell-specific effect.

Fig. 3.

Depletion of HIRA effects hematopoietic differentiation during EB differentiation. mRNA expression levels of adult β-globin (A), EKLF, GATA-1, and GATA-2 (B) at indicated days (d) after EB differentiation of wild-type and HIRA^−/− ES cells were monitored by quantitative RT-PCR (normalized to GAPDH levels). Values are presented relative to expression level at day 0, which was set to 1. Results are the average of triplicates from a single experiment that is representative of two experiments. The error bars reflect SD.

HIRA Is Necessary for Activation of β-Globin During Cell Differentiation.

Because the absence of HIRA led to a down-regulation of not only β-globin expression but also of the upstream transcription factors EKLF and GATA-1, we determined whether MEL cells could be alternatively used to address whether HIRA is required for β-globin expression. Treatment of MEL cells with hexamethylene bis-acetamide (HMBA) or DMSO results in high-level induction of erythroid genes, including β-globin, and eventual terminal differentiation (50, 51). The expression of EKLF, GATA-1, and importantly, HIRA proteins, are stable during the differentiation of MEL cells (Fig. S3). MEL cells were individually or coinfected with short hairpin (sh)RNA-expressing lentiviruses that target different regions of HIRA (52) (scheme outlined in Fig. S7A). The expression of HIRA was reduced in the shHIRA #2 and in the shHIRA #1+2 MEL cells at both the protein and mRNA levels but not by shHIRA #1 alone (Fig. S7B). Importantly, depletion of HIRA had no effect on the expression of EKLF and GATA-1 (Fig. S7B).

We used these and two additional stably infected cell lines to study the effect of HIRA depletion on the expression of β-globin after their treatment with DMSO to induce differentiation and expression of adult β-globin. As monitored after 4 d of differentiation, the shHIRA#2-, #3-, and #4-expressing MEL cells exhibit a dramatic drop in β-globin activation in proportion to the level of RNA and protein knockdown (Fig. 4). No effect on β-globin induction was observed in cells infected with the empty vector, a scrambled shRNA, or shHIRA #1 alone (Fig. 4). These results are also supported by a detailed time course of differentiation and monitoring RNA isolated every 24 h after differentiation induction (Fig. S8). We conclude that HIRA reduction leads to a significant drop in the activation of β-globin, independent of any effect on EKLF or GATA-1 expression. Coupled with the data in Fig. 3, we can also conclude that HIRA is thus not only required for erythroid onset but also continuously required for maintenance of proper expression even after erythropoiesis has been underway.

Fig. 4.

HIRA is required for β-globin expression. MEL cells were infected individually with lentiviral shRNAs directed against different regions of HIRA (1, 2, 3, 4) empty vector (V), or a scrambled control (Scr). Results from two separate experiments are shown (Left and Right). Protein and mRNA levels of HIRA (Insets; β-actin is protein loading control) were monitored in the infected cell lines for this experiment. Stably infected shHIRA MEL cells were subjected to treatment with DMSO for 96 h to induce differentiation, and β-globin mRNA levels were monitored by quantitative RT-PCR in comparison with the uninduced cells. Results are the average of triplicates from a single experiment that is representative of up to three experiments. The error bars reflect SD.

HIRA Recruitment to the β-Globin Promoter Requires EKLF.

To address the functional consequence of the interaction of HIRA with EKLF, first we studied HIRA occupancy at the β-globin promoter by chromatin immunoprecipitation (ChIP) of MEL cells by a benzonase-based, non–cross-linked protocol, using primers that overlap known EKLF binding sites at the β-promoter and the locus control region (LCR) (53). As shown in Fig. 5A, HIRA is present, and its occupancy increases after induction of differentiation, solely at the adult β-globin promoter but not at the upstream hypersensitive sites 2 and 3 (HS2 and HS3; critical components of the LCR) or the necdin promoter (used as a negative control). This demonstrates that HIRA interacts precisely within the same region as EKLF at the promoter but not at the EKLF-bound upstream enhancer elements.

Fig. 5.

EKLF mediates the binding of HIRA to the β-globin promoter. We used a benzonase-based ChIP protocol in the absence of crosslinker and sonication with a mixture of four anti-HIRA antibodies that led to a highly enriched and specific signal. (A) Quantitative ChIP analysis of HIRA occupancy at the β-globin promoter, globin LCR hypersensitive sites HS2 and HS3, or necdin was performed with anti-HIRA antibodies or control IgG in MEL cells without differentiation (day 0) or after DMSO-induced differentiation (72 h), as indicated. (Inset) Agarose gel of the product after end-point PCR. (B) Protein levels of EKLF after Dox-inducible knockdown (2 d) of stable MEL cells that express inducible shEKLFs (46) (Inset; β-actin is the protein loading control). HIRA occupancy at the β-major promoter was measured by quantitative ChIP using HIRA or IgG antibodies on untreated (−) or Dox-treated (+) control (parental) or shEKLF MEL cell lines 1 or 2. Results are the average of biological triplicates from a single experiment that is representative of two experiments.

Next we addressed whether the presence of HIRA on the β-globin promoter is EKLF dependent by monitoring its chromatin occupancy after using RNA interference to knock down EKLF levels. We used two MEL stable cells lines expressing two different doxycycline (Dox)-inducible shEKLF RNAs that have been previously shown to decrease EKLF expression and thus its occupancy at and activation of the β-globin promoter (54). We find that HIRA occupancy at the β-globin promoter is adversely affected in proportion to the extent of EKLF knockdown in the two MEL lines, even without induction of differentiation (Fig. 5B). These results, together with our interaction assays, support the idea that EKLF is required for recruitment of HIRA to the β-globin promoter.

Discussion

We have shown that ablation of HIRA leads to a deficiency in erythropoiesis due to lower levels of the critical EKLF and GATA-1 transcription factors. In addition, depletion of HIRA disrupts expression of an important erythroid target, β-globin, even in the presence of normal levels of EKLF and GATA-1. As a result, HIRA plays critical roles in both the onset and maintenance of erythropoiesis. Its control of specific targets during terminal erythroid differentiation is brought about, at least in part, by its interaction with and recruitment by EKLF to these restricted sites. Knowledge of the regulation and mode of action of these molecules raises a number of interesting biochemical and developmental issues.

Developmental Convergence of EKLF and HIRA Function.

EKLF is highly restricted in its developmental expression pattern, because it is first expressed in the mesodermal blood islands of the yolk sac at E7.5, then solely in the fetal liver as the site of erythropoiesis is changed during early development (55). Although ablation of EKLF yields visibly “normal” primitive yolk sac cells with slight effects on embryonic β-like globin expression, morphologically aberrant definitive erythroid cells that are profoundly deficient in adult β-globin expression appear after the switch to fetal liver, leading to embryonic lethality by E14.5 (27, 28). HIRA also plays a critical role in early developmental processes: null embryos exhibit lethality by E10–11 due to defects that begin during gastrulation at E6 and alter subsequent formation of the mesoderm and primitive streak (19). This can alter expression of important tissue-restricted regulators in development, such as MyoD during myogenic differentiation (56) and VEGFR1 in vascular development and angiogenesis (52). Analogous to our observations showing that erythroid lineage differentiation (EKLF, GATA-1, and β-globin) but not early hematopoiesis (GATA-2) is affected, HIRA-null ES cells are able to generate neuronal precursor cells but not mature neurons (47). Although we have focused on specific targets, absence of HIRA likely has additional effects deleterious to erythroid maturation.

It is intriguing that BMP4 expression is significantly down-regulated in HIRA-null embryos (19), because this signaling pathway plays a critical role in directing the onset of EKLF and GATA1 expression (57, 58). As a result, HIRA is appropriately placed to exert a significant biological effect in tissues within which EKLF expression and erythropoietic differentiation begins.

Constraints on Protein Interactions/Structural Considerations at the β-Globin Promoter.

Our studies demonstrate that directed recruitment of the HIRA histone H3.3 chaperone to the developmentally controlled adult β-globin gene is accomplished by its interaction with the EKLF transcription factor. However, EKLF recruitment is discriminatory: HIRA was selectively recruited to the β-promoter and not to the upstream HS2 and HS3 sites within the LCR. These data also provide an explanation for the selective enrichment of histone H3.3 at the promoter during erythroid differentiation (37), particularly because a large majority of HIRA binding sites are enriched for H3.3 (12). Depletion of HIRA decreases β-globin expression, and reduction of EKLF decreases HIRA recruitment. Although a number of transcription factors have been shown to interact with HIRA (12, 25, 59), their ability to recruit it to a specific site had not been addressed. In this context, it is of interest that the EKLF zinc fingers are primarily responsible for its interaction with HIRA, because they are also critical for EKLF contact with histone H3 (37). Protein binding irrespective of an intact finger structure raises the speculation that EKLF/HIRA interactions may remain relevant at sites that do not contain a cognate EKLF DNA binding element.

The EKLF zinc finger region is also its interaction site for protein Brg1 (60, 61). EKLF acetylated at K288 shows a higher affinity for the SWI/SNF chromatin remodeling complex (that minimally includes Brg1 and Baf155) and is a more potent activator of the β-globin promoter in in vitro reconstituted chromatin (36, 62). These observations tie together well with HIRA, which also forms a protein complex with Brg1, Baf170, and Baf155, as shown by endogenous coimmunoprecipitation in HeLa cells and by their preferential coenrichment at active promoters and enhancers (12).

HIRA is part of stable multiprotein complex that consists of UBN1, CABIN1, and ASF1a proteins (16). UBN1 interacts with the HIRA WD40 region (63) and ASF1a with a central “B” region adjacent to this module (13). On the other hand, CABIN1 interacts with the non-WD40, C-terminal region of HIRA (16), which we have shown also binds EKLF. As a result, it is possible that these interactions are antagonistic, as has been shown for Mef2 in myogenesis (64). The ASF1a component of the complex directly interacts with histone H3.3, as does EKLF. In this case these interactions are nonoverlapping, because the EKLF zinc fingers interact with the amino terminal tail of H3 (37), but ASF1a interacts with the globular region of H3 (65, 66). HIRA interacts with yet a different surface on ASF1a (13).

These latter structural considerations enable a relatively straightforward visualization of a HIRA/ASF1a/H3.3/EKLF interaction. However, placing this within the larger context of chromatin remodelers, and given that the multitasking EKLF zinc fingers also confer DNA recognition specificity at the β-globin promoter, is clearly more complex. The end result, however, of such recruitment by EKLF of HIRA is incorporation of H3.3 and establishment of an open chromatin domain at the β-locus (37, 38, 59). Because nucleosomal core particles containing H3.3 are relatively less stable than those with H3.1 (4), transcriptional activation of the β-globin gene is more readily accomplished (37).

Materials and Methods

A non–cross-linked/benzonase protocol was used for ChIP. Lentiviral shRNA-expressing particles were used for RNA interference. Details of these and other experimental approaches are provided in SI Materials and Methods.