Biochemical Journal

Research article

Identification of two independent nucleosome-binding domains in the transcriptional co-activator SPBP

Sagar Darvekar , Sylvia Sagen Johnsen , Agnete Bratsberg Eriksen , Terje Johansen , Eva Sjøttem


Transcriptional regulation requires co-ordinated action of transcription factors, co-activator complexes and general transcription factors to access specific loci in the dense chromatin structure. In the present study we demonstrate that the transcriptional co-regulator SPBP [stromelysin-1 PDGF (platelet-derived growth factor)-responsive element binding protein] contains two independent chromatin-binding domains, the SPBP-(1551–1666) region and the C-terminal extended PHD [ePHD/ADD (extended plant homeodomain/ATRX-DNMT3-DNMT3L)] domain. The region 1551–1666 is a novel core nucleosome-interaction domain located adjacent to the AT-hook motif in the DNA-binding domain. This novel nucleosome-binding region is critically important for proper localization of SPBP in the cell nucleus. The ePHD/ADD domain associates with nucleosomes in a histone tail-dependent manner, and has significant impact on the dynamic interaction between SPBP and chromatin. Furthermore, SPBP and its homologue RAI1 (retinoic-acid-inducible protein 1), are strongly enriched on chromatin in interphase HeLa cells, and both proteins display low nuclear mobility. RAI1 contains a region with homology to the novel nucleosome-binding region SPBP-(1551–1666) and an ePHD/ADD domain with ability to bind nucleosomes. These results indicate that the transcriptional co-regulator SPBP and its homologue RAI1 implicated in Smith–Magenis syndrome and Potocki–Lupski syndrome both belong to the expanding family of chromatin-binding proteins containing several domains involved in specific chromatin interactions.

  • chromatin binding
  • extended plant homeodomain/ATRX-DNMT3-DNMT3L (ePHD/ADD)
  • retinoic acid inducible protein 1 (RAI1)
  • stromelysin-1 PDGF (platelet-derived growth factor)-responsive element binding protein (SPBP)
  • transcriptional co-activation


Chromatin structure imposes profound and ubiquitous effects on almost all DNA-related metabolic processes, including transcription. The state of chromatin is influenced by histone marks which are read individually or in patterns. The histone marks are recognized by conserved protein fold motifs (reviewed in [1]) which often are found in multiple copies or in tandem with other chromatin-associated domains in a single protein or protein complex. Transcriptional co-regulators include multisubunit complexes that can read specific histone marks and are recruited to promoters by gene-specific activators to facilitate initiation of transcription either by direct interaction with general transcription factors or RNA polymerase II, or indirectly through modification of chromatin structure [2]. SPBP [also named TCF20; stromelysin-1 PDGF (platelet-derived growth factor)-responsive element binding protein] is a 220 kDa multidomain protein containing features of a transcriptional regulator. It constitutes an N-terminal transactivation domain, a novel DNA-binding domain with an AT-hook motif, three NLSs (nuclear localization signals) and a C-terminal cysteine-rich region with similarity to the ePHD (extended plant homeodomain)/ADD (ATRX-DNMT3-DNMT3L) domain (Figure 1A). SPBP was originally found to be involved in transcriptional activation of the MMP3 (matrix metalloprotease 3) promoter via a specific DNA sequence [3]. Later it was shown to act as a transcriptional co-regulator that enhances [46] or represses [7] the transcriptional activity of certain transcription factors or cofactors such as Sp1 (specificity protein 1), Ets1 (E twenty-six 1), Pax6 (paired box protein 6), SNURF (small nuclear RING-finger)/RNF4, c-Jun, AR (androgen receptor) and ERα (oestrogen receptor α). The ePHD/ADD domain is shown to interact with the BRCT6 domain of the checkpoint mediator TopBP1 (DNA topoisomerase II binding protein 1) [8]. TopBP1 and SPBP co-operated to enhance Ets1-mediated transcription and both were found to be associated with the Ets1-regulated promoters c-myc P1P2 and MMP3.

Figure 1 SPBP is associated with chromatin

(A) Schematic representation of the domain structure of human SPBP. TAD, transactivation domain; DBD, DNA-binding domain; Q1/Q2, glutamine-rich stretches. Boxes designated A–G of SPBP represent evolutionarily conserved regions with more than 50% sequence identity with human RAI1. Numbers below indicate amino acid positions. (B) SPBP is mainly associated with euchromatin. Equivalent nuclear aliquots from the samples indicated in Supplementary Figure S2 (at were subjected to SDS/PAGE and Western blotting using specific anti-SPBP (top panel), anti-(CTD phosphorylated RNA polymerase II) (middle panel) or anti-(histone H3) (bottom panel) antibodies. The molecular mass in kDa is indicated on the right-hand side. (C) SPBP and RAI1 co-localize with histones and are enriched in chromatin-rich regions of the nucleus. HeLa cells were transfected with plasmids expressing EGFP–SPBP, mCherry–H2A, EGFP–RAI1 and/or mCherry–SPBP as indicated, fixed and stained with DraQ. For visualizing the endogenous expression pattern, HeLa cells were fixed and immunostained with anti-SPBP antibody and DraQ.

SPBP has a similar gene and domain structure to, and shares seven regions of strong homology with, the dosage-sensitive gene RAI1 (retinoic-acid-inducible protein 1) (Figure 1A) [5]. Deletion of, or mutations in, RAI1 is clinically associated with Smith–Magenis Syndrome [9] and Potocki–Lupski syndrome [10]. RAI1 is also reported to be associated with non-syndromic autism [11], schizophrenia [12] and spinocerebellar ataxia type 2 [13]. RAI1 is localized to the nucleus and contains a modest transcriptional activation activity localized to the N-terminal region [14].

In an attempt to shed more light on the mechanism of SPBP-mediated transcriptional activation, we have studied its subnuclear localization, mobility and interaction with chromatin. In the present study we show that SPBP localized to chromatin-rich regions in the nucleus of HeLa cells, mainly associated with euchromatin. Two regions of SPBP, amino acids 1551–1666 and the ePHD/ADD domain, were identified to contain nucleosome-binding activity. The binding mediated by the 1551–1666 region was independent of the histone tails. This region, which shows no homology with other characterized nucleosome-binding domains, is localized next to the DNA-binding domain of SPBP and is critically important for proper localization of SPBP in the nucleus of HeLa cells. It encompasses a bipartite NLS, is highly conserved and has homology with the corresponding region in RAI1. The SPBP ePHD/ADD domain bound nucleosomes in a histone tail-dependent manner, and was important for the dynamic interaction between SPBP and chromatin, and also for complete translocation of SPBP from the cytoplasm to the nucleus. RAI1 was found to be enriched in the same chromatin speckles as SPBP, and had low nuclear mobility. Also, the RAI1 ePHD/ADD domain bound nucleosomes, but with significantly lower affinity than the SPBP ePHD/ADD domain. Hence, both SPBP and its homologue RAI1 are chromatin-binding proteins implicated in regulation of gene expression.



cDNA constructs were subcloned into Gateway entry vectors and expression clones made according to the manufacturer's instructions (Invitrogen). All constructs were verified by DNA sequencing. The following constructs were made in the present study: pDONR221-H2A, pDONR221-SUMO3, pDONR221SPBP-(Δ1344–1601), pDONR221-SPBP-(Δ1551–1666), pDONR-SPBP-(1333–1666, ΔP2N2), pDONR221-SPBP-(1333–1960), pDONR221-SPBP-(1333–1551), pDONR221SPBP-(1551–1666), pDONR221-SPBP-(1601–1666), pDONR221-SPBP(NLS2), pDONR221-SPBP(PEST2). Detailed information on the constructs are available upon request. All other SPBP constructs, RAI1 constructs and pENTR1A-H2B have been described previously [5,8,15].


The following primary antibodies were used: rabbit anti-SPBP [8] (1:500 dilution), rabbit anti-[CTD (C-terminal domain) phosphorylated RNA polymerase II] (sc-13583, Santa Cruz Biotechnology) (1:500 dilution), anti(histone H3) (ab1791, Abcam) (1:1000 dilution), anti-(histone H3K36me3) (ab9050, Abcam) (1:500 dilution), anti(histone H3K4me3) (ab8580, Abcam) (1:2000 dilution), anti-(acetylated histone H3) (Upstate Biotechnology) (1:1000 dilution), anti-GST (glutathione transferase) (Santa Cruz Biotechnology) (1:1000 dilution), mouse anti-PML (promyelocytic leukaemia) (Santa Cruz Biotechnology) (1:500 dilution) and mouse anti-FLAG (Stratagene) (1:2000 dilution). Secondary antibodies used were: HRP (horseradish peroxidase)-conjugated goat anti-rabbit IgG and anti-mouse IgG antibodies (BD Biosciences) (1:2000 dilution) and Alexa Fluor® 568-conjugated goat anti-mouse IgG (Molecular Probes) (1:500 dilution).

Cell culture

HeLa cells (A.T.C.C. CCL2) were grown in Eagle's MEM (minimum essential medium) supplemented with 10% FBS (fetal bovine serum), 100 units/ml penicillin and 100 μg/ml streptomycin.

Cell imaging

Cells were seeded at a density of 104 cells/well in eight-well coverglass slides (Nunc) and transfected with 50 ng (H2A) or 200 ng (SPBP, RAI1) of plasmid expression vectors using Lipofectamine™ PLUS (Invitrogen). At 1 day after transfection, cells were analysed by live-cell confocal microscopy, or fixed for 10 min in PBS containing 4% paraformaldehyde. The cells were permeabilized using 0.1% Triton X-100 for 5 min at room temperature (23°C) and blocked in PBS with 3% goat serum. Subsequently, the cells were incubated at room temperature with primary and secondary antibodies for 60 and 30 min respectively. DNA was stained using Draq5™ (Biostatus) (1:1000) for 5 min at room temperature. Images were collected using a Zeiss Axiovert 200 microscope with a 40×, 1.2 W C-Apochroma objective, equipped with an LSM510-META confocal module using LSM 5 software version 3.2 (Carl Zeiss), or a Leica TCS SP5 confocal microscope, 63×, 1.2 W objective, equipped with an incubation chamber with CO2 and temperature control. Images were processed using Canvas version 9 (ACD Systems).

FRAP (fluorescence recovery after photobleaching)

Cells were seeded at a density of 104 cells/well in eight-well coverglass slides and transfected with 50–200 ng of plasmid expression vectors 24 h later using Lipofectamine™ PLUS (Invitrogen). The FRAP experiments were carried out essentially as described previously [15].

Chromatin association assay

Approximately 1.5×107 HeLa cells were harvested, washed in PBS and lysed in 300 μl of buffer N [15 mM Tris/HCl (pH 7.5), 60 mM KCl, 15 mM NaCl, 15 mM MgCl2, 1 mM CaCl2, 1 mM DTT (dithiothreitol) and protease inhibitor cocktail (Roche)] with 0.3% Nonidet P40 for 5 min on ice. A fraction of the lysed cells were removed as WCE (whole cell extract), before centrifugation at 400 g for 10 min at 4°C. The supernatant was saved as the cytoplasmic extract, and the pelleted nuclei were resuspended in 300 μl of buffer N with 0.6% Nonidet P40 and incubated 10 min on ice. The resuspended nuclei were centrifugated at 400 g for 10 min at 4°C, and the supernatant was saved as nuclear extract wash 1. The pelleted nuclei were resuspended in 300 μl of buffer N with 0.6% Nonidet P40 and prewarmed for 10 min at 37°C. The fraction was divided into two and digested with 2 or 5 units of micrococcus nuclease respectively, for 10 min at 37°C. The digested nuclei were cooled on ice for 10 min before centrifugation at 16000 g for 10 min at 4°C. The supernatants were saved as nuclear extract wash 2 and the amount of micrococcus degradation was investigated by agarose gel electrophoresis. The nuclear pellets were resuspended in ice-cold 2 mM EDTA (pH 8.0) and incubated on ice for 10 min, followed by centrifugation at 16000 g for 10 min at 4°C. The supernatants were saved as chromatin extracts and the pellets were resuspended in 150 μl 2× SDS loading buffer. Equal amounts (in terms of cell number) of WCE, the supernatants and the pellets were subjected to Western blotting as described previously [8].

Isolation of soluble mono- and di-nucleosomes

Soluble nucleosomes were prepared mainly as described for the chromatin association assay except that the micrococcus nuclease digestion was increased to 5 units per 107 cells with an incubation time of 30 min at 37°C. The chromatin extract contained soluble mono- and di-nucleosomes. To obtain tail-less nucleosomes, the nucleosomes were trypsinized in trypsin digestion buffer [25 mM NaCl, 10 mM Tris/HCl (pH 8.0) and 1 mM EDTA] with trypsin (7 ng/μl) for 30 min at room temperature. The trypsinization reaction was stopped by adding a 20-fold excess (140 ng/μl) of soya bean trypsin inhibitor.

GST–nucleosome pull-down assay

The GST-fusion proteins were expressed and purified from Escherichia coli as described previously [8]. The nucleosomes were precleared by GST immobilized on gluthatione–Sepharose beads (1 μg) by incubating with rotation for 30 min at 4°C in 400 μl of GST–nucleosome-binding buffer [20 mM Tris/HCl (pH 7.8), 1 mM EDTA, 150 mM NaCl, 20% glycerol, 1 mM DTT, 0.1% Nonidet P40 and protease inhibitor cocktail (Roche)]. The beads were precipitated by centrifugation at 300 g for 5 min at 4°C. The supernatant was transferred to a new tube and GST-fusion protein (1–5 μg) and gluthatione–Sepharose beads to a total bead volume of 20 μl was added. The tube was incubated on a rotating wheel at 4°C for at least 1 h to overnight. The beads were washed three times in GST–nucleosome-binding buffer. Bound proteins were separated by SDS/PAGE and subjected to Western blotting as described previously [8].

Nucleosome gel-shift assay

Gel-shift asssays were carried out in 20 μl reaction volumes containing 10 mM Hepes (pH 8.0), 100 mM NaCl, 5% glycerol, 0.03% Nonidet P40, 1 mM DTT and 150 ng/μl BSA. Approximately 2 μM GST–SPBP(ePHD/ADD) protein was incubated with 60 nM 32P-labelled mononucleosomes at room temperature for 20 min. Samples were then subjected to a non-denaturating PAGE in 0.4×TBE [Tris/borate/EDTA (1×TBE=45 mM Tris/borate and 1 mM EDTA)], 5% acrylamide (40:1), 5% glycerol and 0.03% Nonidet P40 overnight at 100V and 4°C. The gels were dried and visualized by phosphorimaging.

GST pull-down assays

Expression of GST-fusion proteins and GST pull-down assays were carried out as described previously [8]. For expression of GST-fusion proteins containing ePHD domains, 0.1 mM ZnCl2 was added to the growth medium.

Partial trypsin digestion

The His–SPBP-(1551–1666) protein was expressed and purified from E. coli BL21 (DE3) cells mainly as described previously [8], except that nickel beads were used as the solid phase. The His-fusion protein was eluted from the beads by adding 50 mM Tris/HCl, 200 mM NaCl, 2 mM 2-mercaptoethanol and 500 mM imidazole. Equal amounts of HeLa nucleosomes and His-tagged protein were mixed together and partial trypsin digestion was carried out at an enzyme/substrate ratio of 1:50. His–SPBP-(1551–1666) with or without nucleosomes were incubated with trypsin at different time intervals ranging from 0 to 120 min. Digested proteins were separated by SDS/PAGE (15% gel) and visualized by Imperial Blue protein staining.

SUMOylation assay

HeLa cells (105 per well) were seeded in six-well dishes 20 h before transfection. The cells were transiently co-transfected with expression vectors for 3× FLAG-tagged SPBP deletion constructs (500 ng) and SUMO3 (500 ng) using Metafectene® Pro (Biontex Laboratories). The cells were harvested in 60 μl of 2× SDS loading buffer 1 day post-transfection. Proteins were separated by SDS/PAGE and FLAG-tagged proteins were detected by Western blotting.

Reporter gene assays

HeLa cells were seeded at a density of 105 cells/well in six-well plates, transfected using the calcium phosphate co-precipitation method and assayed further as described previously [8].


SPBP is localized to euchromatin regions in HeLa cells

We have shown previously that SPBP is a nuclear protein [5]. To evaluate its subnuclear localization, chromatin association assays were performed. Cytoplasmic, nucleoplasmic, euchromatin and chromatin fractions were extracted from HeLa cells as illustrated in Supplementary Figure S1 (at Western blot analysis of the various extracts shows that a fraction of SPBP is localized in the nucleoplasm, whereas most of the protein is released upon partial micrococcus nuclease digestion (Figure 1B). Partial micrococcus nuclease digestion releases proteins that are associated with more loosely organized chromatin, mainly the euchromatin. CTD-phosphorylated RNA polymerase II was found in the same fractions as SPBP (Figure 1B), but also in the insoluble pellet, indicating binding to highly packed chromatin or other insoluble structures. Histone H3 on the other hand, was mainly localized to the chromatin fraction (Figure 1B). Furthermore, confocal microscopy analysis showed that SPBP is localized to the nucleus and excluded from the nucleoli with enrichment in punctated structures close to the nuclear membrane and around the nucleoli (Figure 1C). Moreover, it is enriched in several speckles throughout the nucleoplasm. The SPBP distribution resembles the DNA staining pattern and the distribution of histone H2A (Figure 1C), which further indicates that SPBP is associated with chromatin. Also, analysis of its nuclear mobility by FRAP suggests that SPBP is associated with chromatin and/or other complexes in the nucleus. SPBP displayed a relatively low mobility and was not completely recovered within 5 min, whereas the transcription factors Sp1, AR and GTFIIH (general transcription factor IIH) were completely recovered at this time [15] (Supplementary Figure S2 at Fluorescence recovery rate depends on the size of the diffusing protein and the retardation of the mobility of the diffusion protein due to association with immobile nuclear structures [16]. Thus association with chromatin will have a profound impact on the diffusion rate of a nuclear protein complex. The punctated distribution pattern of SPBP was completely recovered following photobleaching (Supplementary Figure S2) indicating that SPBP is specifically recruited to these subnuclear structures. Furthermore, we found that the SPBP homologue RAI1 displayed a very similar nuclear distribution pattern as SPBP and H2A (Figure 1C). The RAI1 association with nuclear structures seemed to be rather strong since FRAP analysis revealed that RAI1 has very low nuclear mobility, with only approximately 50% recovery after 5 min (Supplementary Figure S2). Taken together, these results show that SPBP and its homologue RAI1 are recruited to chromatin structures in the nucleus of living cells.

Identification of two independent nucleosome-binding domains in SPBP

On the basis of the above data, we investigated whether SPBP has the ability to interact directly with chromatin. GST pull-down assays using nucleosomes isolated from HeLa cells and various regions of SPBP fused to GST, showed that the region from amino acid 1333 to 1666 of SPBP, and both the short and the long splice variant [5] of the SPBP ePHD/ADD domain associated with nucleosomes (Figure 2A). A protocol based on cell fractionation and micrococcus nuclease digestion was used to generate mono- and di-nucleosomes from HeLa cells (Figure 2B, right-hand panel). GST pull-down assays using the ePHD/ADD domain, the 1333–1666 region or both [SPBP-(1333–1960)] fused to GST confirmed nucleosome binding activity within all three constructs (Figure 2B). Next, the impact of the chromatin-binding domains 1333–1666 and ePHD/ADD on the nuclear mobility of SPBP was analysed by FRAP. Surprisingly, removing the ePHD/ADD domain from SPBP resulted in a significant decrease of the recovery rate compared with full-length SPBP. In contrast, removing most of the novel nucleosome-binding region [SPBP-(Δ1344–1601)] had no significant effect on the SPBP recovery curve (Figure 2C). Moreover, the ePHD/ADD domain alone displayed a significantly faster recovery curve than the 1333–1666 region (Figure 2D). These data suggest that the on/off rate of the association between the ePHD/ADD domain and chromatin is much higher compared with the association between the 1333–1666 nucleosome-binding region and chromatin. Hence the ePHD/ADD domain may be important for the dynamic interaction between SPBP and chromatin, and contribute to a continuous sampling mechanism which is reported for chromatin-associated proteins [17].

Figure 2 SPBP contains two independent nucleosome-binding domains

(A) The ePHD/ADD domain and the region from amino acid 1333 to 1666 of SPBP bind to nucleosomes in vitro. Nucleosomes isolated from HeLa cells were incubated with GST or GST-fusion proteins covering different regions of SPBP as indicated. -s and -l indicate short and long splice variants of the SPBP ePHD/ADD domain respectively. Bound proteins were detected by Western blotting using an anti-histone H3 antibody (top panel). The GST proteins were visualized by Coomassie Brilliant Blue staining (bottom panel). (B) Chromatin extracts containing mono- and di-nucleosomes isolated from HeLa cells (right-hand panel) were incubated with GST, GST–SPBP-(1333–1960), GST–SPBP-(1333–1666) or GST–SPBP(ePHD/ADD) immobilized on gluthatione–Sepharose beads. Bound proteins were visualised by Imperial Blue protein stain (Pierce). The histone bands were analysed by MS. The molecular mass in kDa is indicated on the left-hand side. MW, molecular mass; WB, Western blot. (C) The SPBP(ePHD/ADD) domain has an effect on the dynamic association between SPBP and chromatin. Recovery curves of HeLa cells transfected with plasmids expressing EGFP–SPBP and the deletion mutants EGFP–SPBP-(Δ1344–1601) and EGFP–SPBP(ΔePHD/ADD). Cells were imaged and the fluorescent signal was calculated before and after photobleaching a portion of the nucleus. Graphs show the recovery kinetics after double normalization of the fluorescence, correcting for background and loss of fluorescence caused by photobleaching during imaging. The graph shows the average intensity of the bleach spot for the indicated time point. Each experiment was performed three independent times bleaching eight to ten cells per EGFP-fusion protein each time. (D) The SPBP(ePHD/ADD) domain displays a faster recovery curve than SPBP-(1333–1666). Recovery curves of HeLa cells transfected with plasmids EGFP–SPBP(ePHD/ADD) and EGFP–SPBP-(1333–1666) were obtained as described in (C).

A novel core nucleosome-binding domain is located adjacent to the AT-hook motif

To further characterize the association of SPBP-(1333–1666) with nucleosomes, we mapped the minimal nucleosome-binding region using deletion constructs of the 1333–1666 region in nucleosome pull-down assays (Figure 3A). The region spanning amino acids 1551–1666 binds nucleosomes equally well as the complete region (Figure 3B, top panel). Amino acids 1551–1666 span a predicted PEST sequence (PEST2) (amino acids 1551–1576) and a bipartite NLS, NLS2 (amino acids 1576–1599) [5]. To determine whether the nucleosome binding was dependent on the histone tails, the HeLa nucleosomes were exposed to partial trypsin digestion (Supplementary Figure S3A, right-hand panel, at and used in nucleosome pull-down assays. Removing the histone tails had no effect on the nucleosome-binding affinity of the 1551–1666 region (Figure 3B, bottom panel). These in vitro data show that SPBP exhibits a core nucleosome-binding activity in the region spanning amino acids 1551–1666. To explore whether a stable domain structure may be located in this region, recombinant His-tagged SPBP-(1551–1666) was exposed to partial trypsin digestion for various time intervals. Within 1 min His–SPBP-(1551–1666) was degraded to an approximate 10 kDa product which was completely degraded after 20 min of trypsin digestion (Figure 3C, bottom panel). This indicated the absence of any stable domain structure in the SPBP-(1551–1666) region. However, when nucleosomes were added before SPBP-(1551–1666) was exposed to trypsin, the 10 kDa fragment was stabilized and completely resistant to further degradation for at least 120 min (Figure 3C, top panel). MS analysis of the 10 kDa fragment identified the region SPBP-(1553–1623) (indicated by arrows in Figure 3D). A BLAST search ( did not reveal any other proteins with similarity to the SPBP-(1553–1623) fragment. This suggests that SPBP-(1551–1666) encodes a novel nucleosome-binding domain, including a bipartite NLS. Sequence analysis of the SPBP homologue RAI1 identified nearly 60% similarity in this region (Figure 3D), suggesting that RAI1 may also exhibit nucleosome-binding activity.

Figure 3 The region 1551–1666 interacts with the nucleosome core and is critically important for proper localization of SPBP in the nucleus of HeLa cells

(A) Schematic representation of the 1333–1666 region of SPBP. Deletion constructs are indicated by arrows. DBD, DNA-binding domain. (B) The interaction between SPBP-(1333–1666) and HeLa nucleosomes maps to amino acids 1551–1666, and is independent of the histone tails. Intact nucleosomes (top panel) or tail-less nucleosomes (bottom panel) were incubated with GST, or various GST–SPBP deletion constructs as indicated. Bound proteins were subjected to Western blotting using an anti-(histone H3) antibody and an anti-GST antibody (Supplementary Figure S3A at The molecular mass (MW) in kDa is indicated on the left-hand side. WB, Western blot. (C) The region of SPBP from amino acid 1553 to 1623 is stabilized upon nucleosome binding. His–SPBP-(1551–1666) with (top panel) or without (bottom panel) HeLa nucleosomes were incubated with trypsin for various time points as indicated. Digested proteins were resolved by SDS/PAGE and stained with Imperial Blue protein stain (Pierce). The approximate 10 kDa band was cut out of the gel and analysed by MS. L, molecular mass ladder (in kDa); N, nucleosomes. (D) Alignment of the novel nucleosome-binding region of SPBP with a similar region in the homologue RAI1. Arrows indicate the protected region identified by trypsin digestion. Dots indicate SUMOylation sites. (E) The core nuclosome-binding region 1551–1666 interacts directly with histones. Labelled SPBP-(1551–1666) (top panel), SPBP-(1601–1666) (middle panel) and SPBP-(1333–1551) (bottom panel) were incubated with GST, GST–H2A or GST–H2B as indicated. Bound proteins were separated by SDS/PAGE and visualized using a PhosphorImager. (F) SPBP is SUMOylated in the nucleosome-binding region 1333–1666. HeLa cells were transiently transfected with plasmids expressing 3× FLAG-tagged deletion constructs of SPBP and SUMO3 as indicated. SUMOylated proteins were visualized by Western blotting using an anti-FLAG antibody. The arrow indicates SUMOylated SPBP-(1333–1666). The molecular mass in kDa is indicated on the right-hand side. (G) Deletion of the nucleosome-binding region 1551–1666 redistributes SPBP. HeLa cells were transfected with plasmids encoding EGFP–SPBP, EGFP–SPBP-(Δ1344–1601) and EGFP–SPBP-(Δ1551–1666) as indicated.

In Figure 3(E) we show that SPBP-(1551–1666) has the ability to bind directly to the histone proteins H2A and H2B. This binding is specific since the flanking regions SPBP-(1601–1666) and SPBP-(1333–1551) do not interact (Figure 3E). Similar results were obtained for histone H3.1 and H3.3 (results not shown). Hence, SPBP-(1551–1666) may bind to nucleosomes via direct interactions with the core histone proteins. Interestingly, a SUMOylation assay performed in HeLa cells indicated that the SPBP-(1333–1666) region is a target for SUMOylation (Figure 3F). SUMOylation of nuclear proteins often promotes regulation of their interaction with nuclear stuctures, and is also shown to affect nuclear mobility. Bioinformatic analysis has predicted several putative SUMOylation sites in the 1333–1666 region. Mapping of these sites by site-directed mutagenesis has identified several target sites located N-terminal and C-terminal of the stable (1553–1623) nucleosome-binding region (results not shown). Two of the C-terminal sites are indicated by round dots in Figure 3(D). This indicates that the SPBP interaction with the core nucleosome may be a target for regulation.

Finally, we performed confocal microscopy analysis to investigate the impact of the novel nucleosome-binding region on subnuclear localization of SPBP. Expression of the SPBP-(Δ1344–1601) and SPBP-(Δ1551–1666) constructs in HeLa cells showed that deletion of the nucleosome-binding region resulted in complete redistribution of SPBP. The deletion mutants accumulated in round dots, and cells containing large dots seemed to undergo apoptotic cell death (Figure 3G, middle panel). This was most prominent when the 1344–1601 region, containing the DNA-binding domain and most of the novel nucleosome-binding domain, was deleted, but similar dot formation was observed when only the novel nucleosome-binding domain was deleted (Figure 3G, right-hand panel). Furthermore, these dots were completely recovered in FRAP experiments (results not shown), indicating that they are highly dynamic structures and not insoluble aggregates of misfolded proteins. Co-staining with an anti-PML antibody showed that the round dots were not due to recruitment of the SPBP deletion mutants to PML bodies (Supplementary Figure S4 at Together these images suggest that the novel nucleosome-binding region SPBP-(1551–1666) is critically important for proper localization of SPBP in the nucleus of HeLa cells.

The ePHD/ADD domain interaction with nucleosomes is dependent on the histone tails

As described previously, the cysteine-rich C-terminal end of SPBP can be viewed as a PHD-like region (zinc ligands 5–12) with an additionally N-terminal zinc-finger (zinc ligands 1–4) [8]. Such extended PHD finger-like structures are also present in the C-terminus of RAI1, and in eight other groups of proteins: (i) the MLL (mixed lineage leukaemia) proteins MLL, MLL2 and MLL 4; (ii) the trithorax homologues AF10, AF17 and MLL6; (iii) enhancer of polycomb family members BRD1/BRL and BR140; (iv) the JADE family members JADE 1, JADE 3 and JADE 4; (v) Jumonji family transcription factors and demethylases including GASC-1 (JMJD2C); (vi) the chromatin-associated protein ATRX; (vii) the demethyl DNA transferases DNMT3L, DNMT3A and DNMT3B; and (viii) the SET domain protein histone methyltransferase NSD1 (Figure 4A). These structures have been named ePHD [8] or ADD domains [18]. Alignment of the various ePHD/ADD domains identified so far has shown that they can be divided into three subgroups (Figure 4A). ATRX and the DNMT3A family constitute one subgroup representing C4 and C8 fingers with the invariant tryptophan/aromatic residue two positions N-terminal to Cys7 in the PHD finger (indicated by an arrow in Figure 4A). This aromatic residue is characteristic of PHD fingers and believed to be structurally important [19]. The other two subgroups represent C2HC plus C4HC2H fingers lacking the aromatic residue characteristic for PHD fingers. Instead, they contain an invariant aromatic residue two positions N-terminal to the third zinc ligand in the extended (GATA-1 like) zinc finger. SPBP and RAI1 contain a very long loop region between zinc ligands 2 and 3 in the GATA-1-like finger, with insertions of 118 amino acids and 110 amino acids respectively, as indicated in Figure 4(A). Hence, they represent one subgroup. There are no obvious sequence similarities between the loop regions of SPBP and RAI1. However, the SPBP loop region is highly conserved and contains a functional NLS (NLS3) [5]. Moreover, four serine/threonine residues clustered N-terminal to NLS3 is reported to be phosphorylated (, suggesting that the loop region is dynamic and may have a regulatory role.

Figure 4 SPBP and RAI1 ePHD/ADD domains contain long loop regions between zinc ligands 2 and 3

(A) Alignment of ePHD/ADD domains identified in SPBP, RAI1, MLL2 (GenBank® accession number NP003473) , BRD1/BRL (GenBank® accession number AAH47508), ATRX [18], DNMT3A [21], JADE1 (GenBank® accession number CAE30500), AF10 (GenBank® accession number AAT47519), GASC1 (GenBank® accession number BAB16102) and NSD1 (GenBank® accession number AAK92049). The loop region between zinc ligands 2 and 3 in SPBP and RAI1 is excluded from the alignment, but indicated above. Cysteine and histidine residues serving as zinc ligands are indicated by numbers. Dots indicate conserved amino acids not involved in zinc co-ordination. The arrow indicates the position of the conserved tryptophan residue normally present in PHD fingers. (B) The RAI1 ePHD/ADD domain binds weakly to nucleosomes. HeLa nucleosomes were incubated with GST, GST–SPBP(ePHD/ADD) or GST–RAI1(ePHD/ADD) as indicated. The GST proteins were immobilized on gluthatione–Sepharose beads. The bound proteins were subjected to SDS/PAGE, and detected by Western blotting using an anti-(histone H3) antibody. The GST proteins are visualized using an anti-GST antibody (Supplementary Figure S3B at WB, Western blot. (C) SPBP(ePHD/ADD) is more mobile than RAI1(ePHD/ADD). Recovery curves of HeLa cells transfected with plasmids GFP–SPBP(ePHD/ADD) and GFP–RAI1(ePHD/ADD). Cells were imaged and fluorescent signals were calculated as described in Figure 2(C).

The ePHD/ADD domain of ATRX binds specifically to histone H3 tails containing trimethylation in position Lys9 and the absence of H3K4me3 and H3K4me2 marks [20]. DNMT3A and DNMT3L ePHD/ADD domains bind to histone H3 tails that are unmethylated in position Lys4 [21,22], and thereby link DNA methylation to H3K4me0. The RAI1 ePHD/ADD domain is predicted to recognize histone H3 tails unmodified at Lys4 [23]. We performed nucleosome pull-down experiments using nucleosomes isolated from HeLa cells, and in-vitro-constituted nucleosomes containing histone H2A or H2AZ to compare the nucleosome-binding activity of SPBP ePHD/ADD and RAI1 ePHD/ADD. Interestingly, SPBP ePHD/ADD bound nucleosomes with a much higher affinity than RAI1 ePHD/ADD in vitro (Figure 4B), but displayed significantly higher mobility than the RAI1 ePHD/ADD in HeLa cells (Figure 4C). This suggests that even if the SPBP and RAI1 ePHD/ADD domains show approximately 50% sequence identity (except the loop region), they represent nucleosome-binding regions with distinct properties.

To investigate the SPBP ePHD/ADD interaction with nucleosomes further, pull-down experiments using trypsin-treated HeLa cell nucleosomes were performed. Removal of the histone tails abolished the interaction between the SPBP ePHD/ADD domain and the HeLa nucleosomes (Figure 5A, bottom panel). Similar results were obtained when the ePHD domains of Hrx2/MLL4 and Alr/MLL2 were exposed to nucleosome pull-down assays (see Supplementary Figure S5 at The PHD3 domain of MLL, which is shown to bind H3K4me3 [24], was included as a control (Supplementary Figure S5). Surprisingly, mapping of the nucleosome-binding activity within the ePHD/ADD domain revealed that the conserved region F (amino acids 1678–1741) alone had the ability to bind nucleosomes (Figure 5A, top panel). The PHD-like region (encompassing zinc ligands 5–12) showed some affinity for nucleosomes, but bound more weakly than the complete ePHD/ADD domain. These data suggest that the nucleosome-binding activity of the ePHD/ADD domain of SPBP mainly is due to the N-terminal-extended region, and that the histone tails are important for high-affinity binding.

Figure 5 The SPBP ePHD/ADD interacts with nucleosomes in a histone tail-dependent manner, and is important for the transcriptional activation potential of SPBP

(A) The interaction between SPBP(ePHD/ADD) and nucleosomes is dependent on the histone tails. Intact nucleosomes (top panel) or tail-less nucleosomes (bottom panel) were incubated with GST–SPBP(ePHD/ADD) or two different deletion constructs of GST–SPBP(ePHD/ADD) as indicated. Bound proteins were visualized by Western blotting using an anti-(histone H3) antibody (left-hand panels) and an anti-GST antibody (Supplementary Figure S3C at The molecular mass (MW) in kDa is indicated on the left-hand side. WB, Western blot. (B) Gel-shift assay showing SPBP(ePHD/ADD) binding to various types of nucleosomes. Increasing amounts of GST–SPBP(ePHD) proteins were incubated with reconstituted mononucleosomes containing histone H2A, histone H2AZ or histone H3 with modifications on the histone tail (H3K4me3 or H3K9me3) as indicated. The mononucleosomes contained DNA with incorporated 32P-labelled nucleotides. GST–SPBP-(4–493) is a negative control. The protein–nucleosome complexes were resolved on a non-denaturating polyacrylamide gel, and visualized by a PhosphorImager. (C) SPBP(ΔePHD/ADD) co-localises with full-length SPBP in the nucleus of HeLa cells. HeLa cells co-transfected with GFP–SPBP(ΔePHD/ADD) and mCherry–SPBP expression plasmids were analysed by confocal laser scanning fluorescence microscopy. (D) The SPBP ePHD/ADD domain contributes to the co-activation potential of SPBP. Transient transfection assays were carried out in HeLa cells using the reporter MMP3–luciferase, together with expression vectors for Ets68 (0.1 μg), HA (haemagglutinin)–SPBP (2 μg), HA–SPBP(ΔePHD) (2 μg) and/or HA–SPBP-(Δ1344–1601) (2 μg) as indicated. The data represent the mean of three independent experiments performed in triplicate.

Next, gel mobility-shift assays were performed to determine whether the SPBP ePHD/ADD domain interacts with nucleosomes with specific post-translational modifications on the histone H3 tail, or specific histone H2 variants. As displayed in Figure 5(B), SPBP(ePHD/ADD) bound all types of reconstituted nucleosomes tested. To further explore the histone tail recognition mediated by SPBP(ePHD/ADD), peptide array experiments using the MODified Histone Peptide Array and Analysis Software from Active Motif ( were performed. Using this assay we found that the SPBP(ePHD/ADD) domain actually recognizes the histone H4-(11–30) tail, especially histone H4 tails with methylations in the Arg19 and Lys20 positions (results not shown). Thus SPBP(ePHD/ADD) binding to nucleosomes seems to be dependent on the histone H4 tail.

To examine the importance of the ePHD/ADD domain for the nuclear localization of SPBP, the fusion protein GFP (green fluorescent protein)–SPBP(ΔePHD/ADD) and wild-type mCherry–SPBP were expressed in HeLa cells. GFP–SPBP(ΔePHD/ADD) and mCherry–SPBP show a completely overlapping distribution pattern in the cell nucleus (Figure 5C), suggesting that proper nuclear localization of SPBP is not dependent on the ePHD/ADD domain. Finally, we identified the importance of the two nucleosome-binding domains 1551–1666 and ePHD/ADD for the transcriptional co-activator function of SPBP. We used the MMP3 promoter and co-activation of Ets1 binding to this promoter as a model system [8]. Deletion of the SPBP(ePHD/ADD) domain resulted in a significant decrease of the co-activation potential of SPBP mediated by Ets1, whereas deletion of the region encompassing the novel nucleosome-binding region resulted in a modest decrease (Figure 5D).


The nucleosomal DNA presents a barrier for proteins that need to contact DNA. The main regulatory mechanisms described to increase the accessibility of DNA in chromatin are chromatin-modifying activities and chromatin-remodelling activities (reviewed in [25]). These activities often reside in large multi-protein complexes classified as transcriptional co-regulators (reviewed in [2,26,27]). In the present study, we show that the transcriptional co-regulator SPBP contains two independent nucleosome binding regions, SPBP-(1551–1666) and SPBP(ePHD/ADD). The interaction between SPBP-(1551–1666) and nucleosomes is independent of the histone tails, and this region of SPBP is very important for proper localization of SPBP in the cell nucleus. The region SPBP-(1551–1666) is adjacent to the DNA-binding domain of SPBP containing an AT-hook motif [5]. The AT-hook is a small positively charged DNAbinding motif centred on the invariant tripeptide GRP. AT-hooks bind the minor groove of DNA with a preference for AT-rich sequences [28]. AT-hooks have been described to constitute an important part of the chromatin-binding domain of chromatin-associated proteins, such as BRAHMA [29], LEDGF/p75 [30] and BRM1 [31] suggesting that the AT-hook motif contributes to the interaction between these proteins and nucleosomes, either by co-operating with other nucleosome/DNA-binding activities or facilitating changes in the DNA structure. Twenty-three amino acids spanning the AT-hook motif within the SPBP DNA-binding domain are 100% conserved between zebrafish, mouse and human, indicating that an important and conserved function is retained within this region. The invariant GRP tripeptide is also found in RAI1 (Figure 3D). However, the amino acids located especially C-terminal, but also N-terminal, to the tripeptide are diverged from the consensus [28] and any DNA-binding activity of RAI1 has so far not been reported. Interestingly, we have found that the region of SPBP spanning the AT-hook motif is highly SUMOylated, suggesting that the interaction between SPBP and DNA is regulated. Post-translational modifications of AT-hook-containing regions have been reported by others and are suggested to have a regulatory role [32]. The novel nucleosome-binding domain 1551–1666 of SPBP is completely conserved in mouse and human. Partial trypsin digestion and MS analysis revealed a protected region spanning amino acids 1553–1623 when this region was incubated with nucleosomes. This suggests that a stable structural fold is formed upon nucleosome binding, and structural predictions using the GlobPlot module of the ELM database ( indicates a globular domain located between amino acids 1578 and 1664. An interesting subject for future studies would be to determine the structure of this specific part of SPBP.

The ePHD/ADD domains of SPBP and RAI1 contain a cysteine-rich C-terminal related to the PHD finger, a common structural motif predominantly found in proteins that function at the chromatin level [33] and reside in large multiprotein complexes [34]. PHD fingers have emerged as a histone mark ‘reader’ module, differentially recognizing either methylated or unmodified lysine residues on histone tails (reviewed in [35]). Many histone mark ‘reader’ complexes also exhibit histone mark ‘writer’ or ‘eraser’ activities, which may be involved in nucleation, spreading, maintenance or erasure of epigenetic marks [36]. Furthermore, PHD fingers are also found to contain enzymatic activity such as histone acetyl transferase activity [37,38], ubiquitin E3 ligase activity [39,40] and SUMO E3 ligase activity [4143]. In the present study we showed that the ePHD/ADD domains of SPBP and RAI1 had the ability to bind to nucleosomes. The SPBP ePHD/ADD binding was dependent on the histone tails, and peptide array analyses indicated that SPBP(ePHD/ADD) preferably binds to the histone H4 tail methylated in the Arg19 position. Interestingly both the C-terminal PHD finger-like part and the N-terminal part of ePHD/ADD exhibited nucleosome-binding activity independent of each other. Structure determinations of the ePHD/ADD domains of ATRX [18], DNMT3L [21] and DNMT3A [44] identifies three modules forming a single globular domain; an N-terminal zinc-finger structure related to the erythroid transcription factor GATA-1, followed by a PHD-like structure and a C-terminal α-helix. The ePHD/ADD domain is reported to be involved in protein–protein interactions [8,45,46] and chromatin binding [2022]. Mutations within these structural motifs are associated with diseases [18,20,46,47]. Mutational analysis of the SPBP ePHD/ADD domain indicates that the N-terminal part forms a zinc-finger-like structure similar to the ATRX and DNMT3A/3L proteins (C. Rekdal, personal communication). However, an especially long loop region between zinc ligands 2 and 3 is present in SPBP and RAI1. This loop may, at least for SPBP, have a regulatory role, since a cluster of phosphorylation sites is identified in more than five published high-throughput mass spectrometric identifications of phosphorylation sites in cellular proteins (

The ePHD/ADD domains of DNMT3A/L and ATRX are reported to recognize histone modification marks associated with transcriptional repression [20,21,44]. However, some ePHD/ADD-containing proteins are associated with transcriptional activation, such as the MLL family and trithorax homologues which are associated with establishment of H3K4 methylation (reviewed in [48]). We have found SPBP to act as a transcriptional co-activator of several sequence-specific transcription factors [5,6,8], whereas RAI1 acted more like a transcriptional repressor [8]. This may indicate that these homologous proteins have distinct roles and are associated with distinct protein complexes. However, overexpression of GFP- and mCherry-tagged proteins demonstrate that they are enriched in the same subnuclear structures.

Establishment and maintenance of correct chromatin structure seem to be critically important in brain development and function, as well as in cell proliferation and differentiation. Interestingly, all ePHD/ADD-containing proteins described so far are important regulators of chromatin and chromosome architecture (reviewed in [49,50]). They are associated with cancer or diseases that include deficits in brain development and function. Hence, mutations in these ePHD/ADD-containing proteins most probably lead to misinterpretation, misreading or miswriting of the epigenetic code, as described for PHD fingers (reviewed in [48,50]). An exception is SPBP, which so far has not been reported to be connected to any specific disease or cancer type. However, SPBP is expressed in certain cell types in most tissues with strong expression in the CNS (central nervous system) ( Several reports show that SPBP acts as a transcriptional co-regulator [36,8], and in the present stud we show that SPBP is associated with chromatin, mainly euchromatin, and that a novel chromatin-binding domain located adjacent to the DNA-binding domain is critically important for its nucleosome-binding activity. In addition, SPBP contains an ePHD/ADD domain important for the dynamic interaction between SPBP and chromatin. A significant question to address in future studies is how SPBP enhances transcriptional activity. Is it associated with specific protein complexes and contributes in guiding these complexes to chromatin, or does it sit on chromatin and recruit various protein complexes depending on specific epigenetic marks?


Sagar Darvekar, Sylvia Johnsen, Agnete Eriksen and Eva Sjøttem carried out the experiments; Eva Sjøttem and Terje Johansen analysed the data. Sagar Darvekar, Terje Johansen and Eva Sjøttem participated in the writing of the paper. Eva Sjøttem co-ordinated the study, and conceived and supervised most of the work.


This work was supported by the Norwegian Research Council, Familien Blix Fond and the Norwegian Cancer Society (to T.J.).


We thank the group of Dr R. Kingston (Massachusetts General Hospital, Boston, MA, U.S.A.) for H2A and H2AZ expression plasmids and reconstituted nucleosomes. We thank Jack A. Bruun for MS/MS (tandem mass spectrometry) analysis and Tromsø BioImaging Platform for assistance with the confocal microscopes.

Abbreviations: ADD, ATRX-DNMT3-DNMT3L; AR, androgen receptor; CTD, C-terminal domain; DTT, dithiothreitol; ePHD, extended plant homeodomain; Ets1, E twenty-six 1; FRAP, fluorescence recovery after photobleaching; GFP, green fluorescent protein; EGFP, enhanced GFP; GST, glutathione transferase; MLL, mixed lineage leukaemia; MMP3, matrix metalloprotease 3; NLS, nuclear localization signal; PML, promyelocytic leukaemia; RAI1, retinoic-acid-inducible protein 1; Sp1, specificity protein 1; SPBP, stromelysin-1 PDGF (platelet-derived growth factor)-responsive element binding protein; TopBP1, DNA topoisomerase II binding protein 1; WCE, whole cell extract


View Abstract