Cryo-EM structure of the nucleosome containing the ALB1 enhancer DNA sequence

Pioneer transcription factors specifically target their recognition DNA sequences within nucleosomes. FoxA is the pioneer transcription factor that binds to the ALB1 gene enhancer in liver precursor cells, and is required for liver differentiation in embryos. The ALB1 enhancer DNA sequence is reportedly incorporated into nucleosomes in cells, although the nucleosome structure containing the targeting sites for FoxA has not been clarified yet. In this study, we determined the nucleosome structure containing the ALB1 enhancer (N1) sequence, by cryogenic electron microscopy at 4.0 Å resolution. The nucleosome structure with the ALB1 enhancer DNA is not significantly different from the previously reported nucleosome structure with the Widom 601 DNA. Interestingly, in the nucleosomes, the ALB1 enhancer DNA contains local flexible regions, as compared to the Widom 601 DNA. Consistently, DNaseI treatments revealed that, in the nucleosome, the ALB1 enhancer (N1) DNA is more accessible than the Widom 601 sequence. The histones also associated less strongly with the ALB1 enhancer (N1) DNA than the Widom 601 DNA in the nucleosome. Therefore, the local histone–DNA contacts may be responsible for the enhanced DNA accessibility in the nucleosome with the ALB1 enhancer DNA.


Introduction
In eukaryotes, genomic DNA is organized into chromatin, in which the nucleosome is the fundamental unit [1]. The protein components of the nucleosome are two histone H2A -H2B dimers and two histone H3 -H4 dimers, which associate as a histone octamer [2]. A DNA segment of about 145 base pairs is then wrapped around the histone octamer in the nucleosome [2]. In chromatin, nucleosomes are connected by linker DNAs, and linker histones bind to the DNA region lying on the dyad axis of the nucleosome together with the linker DNAs [1,3,4].
Nucleosome formation generally restricts the binding of transcription factors (TFs) to the DNA wrapped around the histone octamer. Therefore, in active promoters, TFs predominantly bind nucleosome-free DNA regions [5]. By contrast, pioneer transcription factors bind their target sequences within nucleosomes, and create open chromatin conformations around their target sequences [6][7][8] initial step in the gene activation cascades required for tissue development [9][10][11].
In order to understand how FoxA functions as a pioneer TF, it is essential to determine the three-dimensional structure and physical characteristics of its target nucleosome. However, the nucleosome structure containing a native DNA sequence with pioneer TF-binding sites has not been reported yet. In the present study, we reconstituted the nucleosome containing the specific ALB1 enhancer (N1) sequence, and reconstructed the ALB1 nucleosome structure by cryogenic electron microscopy to near-atomic resolution.

Nucleosome positioning on the ALB1 enhancer (N1) DNA
To map the nucleosome positioning, the ALB1 nucleosome reconstituted with the 180 base-pair ALB1 enhancer (N1) DNA sequence was fixed with 0.5% formaldehyde, and was then treated with MNase, which preferentially digests linker DNA regions. As shown in figure 2a, about 145 basepair DNA fragments of the ALB1 enhancer (N1) DNA were protected from the MNase attack, suggesting that the protected DNA region is tightly wrapped within the nucleosome. We then sequenced 20 MNase-resistant DNA fragments, and found that 65% of the DNA fragments were mapped to 2 base pairs away from one end of the 180 basepair DNA (figure 2b, right position), although 25% of the MNase-resistant DNA fragments were mapped to the opposite end of the 180 base-pair DNA (figure 2b, left position). Therefore, the right position may be the predominant translational nucleosome position in the 180 base-pair ALB1 enhancer (N1) DNA.
To confirm these nucleosome positions, we reconstituted the nucleosome with the 180 base-pair ALB1 enhancer (N1) DNA sequence containing three additional base pairs at both ends (the 186 base-pair ALB1 DNA). We performed the MNase treatment assay without cross-linking, and the sequences of the resulting DNA fragments were analysed by massive parallel sequencing. The centres of the MNaseresistant DNA fragments, which may correspond to the nucleosomal dyad, were plotted (figure 2c the MNase mapping data with cross-linking (figure 2b), two nucleosome positions, corresponding to the right and left positions, were found on the ALB1 enhancer (N1) sequence in the non-cross-linked nucleosomes (figure 2c).

2.3.
Cryo-EM structure of the nucleosome containing the ALB1 enhancer (N1) DNA sequence The reconstituted nucleosome with the 186 base-pair ALB1 enhancer (N1) sequence was fixed with paraformaldehyde by the GraFix method [17]. We then collected images of the ALB1 nucleosome by the cryo-electron microscopy (cryo-EM) method (figure 3a,b). The cryo-EM structures of the ALB1 nucleosome were reconstructed, and the best three-dimensional class was selected and refined to 4.0 Å resolution (figure 3c-e and table 1). In this analysis, we prepared the sample in the presence of FoxA1(170 -472), which includes the DNA-binding and histone-binding domains [15]. However, we could not detect the extra volume corresponding to FoxA1(170 -472) around the predicted FoxA-binding sites, suggesting that it may have dissociated during the sample preparation process. Although the threedimensional classification seeks to separate unique threedimensional structures, the orientation of the DNA sequence cannot be clearly distinguished in the original nucleosome images. We then symmetrized the images. In addition, our cryo-EM nucleosome structure may represent a mixture of two differently positioned nucleosomes ( figure 2). Therefore, the DNA may appear as an average structure. In the structure, the nucleosomal DNA corresponding to 146 base pairs was clearly visible, but the 40 base-pair linker DNA segments disappeared, probably due to the symmetry imposition (figure 3c).

Cryo-EM structure of the ALB1 nucleosome with a linker DNA
To visualize the linker DNA region, we next reconstructed the cryo-EM structure of the ALB1 nucleosome without the symmetrizing process. The ALB1 nucleosome was then successfully reconstructed at 4.5 Å resolution (figure 4a,b). This time, the linker DNA region was clearly visualized, extending the reconstructed electron potential of the DNA ends beyond the limits of the symmetrized reconstruction (figure 4b,c). As shown in figure 2, the right position is the major position of the ALB1 nucleosome. Therefore, we superimposed the ALB1 enhancer DNA sequence on the structure by orienting the linker sequence towards the protruding DNA end, and thus mapped the possible locations of the two FoxA-binding sites on the nucleosome structure (figure 4b). We used the ALB1 nucleosome structure for further comparisons with the known nucleosome structure.
2.5. The ALB1 enhancer (N1) DNA structure in the nucleosome We then compared the ALB1 enhancer (N1) DNA structure with the Widom 601 DNA structure in the nucleosomes. The Widom 601 sequence is a well-known nucleosome positioning sequence that is tightly wrapped around the histone octamer [18,19]. A previous cryo-EM analysis revealed that, up to a resolution of 3.9 Å , the Widom 601 DNA structure in the nucleosome is identical to that in the crystal structure [20]. Therefore, we superimposed the crystal structure of the nucleosomal Widom 601 DNA onto the ALB1 enhancer (N1) DNA structure. As shown in figure 5a, the DNA path in the ALB1 nucleosome is quite similar to that in the Widom 601 nucleosome, and no obvious difference is apparent. The DNA path of the ALB1 nucleosome is also similar to that in the nucleosome containing the 3 0 -LTR of the mouse mammary tumour virus sequence (figure 5b) [21]. The local resolution map revealed that the ALB1 enhancer (N1) DNA contained DNA regions with low resolution, suggesting that these DNA regions are flexible in the nucleosome (figure 5c (right panel), coloured red). By contrast, these flexible regions are not obvious in the local resolution map of the cryo-EM Widom 601 nucleosome structure [20]. Interestingly, the flexible DNA regions of the ALB1 nucleosome are located near the direct binding sites for histones (figure 5c (left panel)). This fact suggested that when compared with the Widom 601 DNA, the ALB1 enhancer (N1) DNA may be loosely bound to the histones in the nucleosome.
2.6. The nucleosome containing the ALB1 enhancer (N1) DNA is more accessible to DNaseI than that containing the Widom 601 DNA To test whether the ALB1enhancer (N1) DNA is actually loosened, we performed a DNaseI treatment assay. In this assay, the endonuclease DNaseI attacks more efficiently if the nucleosomal DNA is loosened. We found that the nucleosomal ALB1 enhancer (N1) DNA (186 base pairs) was more susceptible to the DNaseI than the nucleosomal Widom 601 DNA (figure 6a). We then identified the DNaseI-sensitive sites of the ALB1 enhancer (N1) DNA in the nucleosome, by denaturing polyacrylamide gel electrophoresis. In the ALB1 nucleosome with the right position, the DNA regions around 20 bases and 40 bases from the labelled end were DNaseI-sensitive sites (figure 6b). These regions coincide with the low-resolution DNA regions of the ALB1 nucleosome structure (figure 6c). Therefore, we conclude that the ALB1 enhancer (N1) DNA is locally loosened in the nucleosome.
We then compared the thermal stabilities of the ALB1 nucleosome and the Widom 601 nucleosome. To eliminate the effect of the nucleosome positions, we reconstituted the ALB1 (right position) and Widom 601 nucleosomes with 147 base-pair DNAs. In the thermal stability assay, the dissociations of the H2A -H2B and H3 -H4 dimers from the nucleosome are independently monitored, as the first and second peaks, respectively. In the Widom 601 nucleosome, the H2A -H2B dimers dissociated at 70 -758C, while the second peak for the H3 -H4 dimer dissociation was observed rsob.royalsocietypublishing.org Open Biol. 8: 170255 at 86-908C (figure 6d). Interestingly, in the ALB1 nucleosome, the first and second peaks were both shifted towards lower temperatures (figure 6d). These data indicate that the histones of the ALB1 nucleosome associate more weakly with the DNA than those of the Widom 601 nucleosome, consistent with our structural analysis. Therefore, the ALB1 enhancer (N1) DNA sequence may be more accessible to DNA-binding proteins, because of the loosened association of the DNA with the histones. This characteristic of the nucleosomal ALB1 enhancer DNA may play a role in accommodating the pioneer TFs within the nucleosome.

Discussion
The nucleosomal DNA binding of the proteins that regulate genomic DNA functions, such as replication, recombination, repair and transcription, depends on the sequence-dependent conformations and physical properties of the DNA wrapped in nucleosomes, as well as their histone compositions. As a consequence, structural studies of nucleosomal DNA by X-ray crystallography have been severely limited, because the crystallization of the nucleosome is highly contingent on the DNA sequence. In fact, many crystal structures of nucleosomes have been deposited in the Protein Data Bank, but most of them contained the palindromic a-satellite or Widom 601 DNA sequence with 145-147 base pairs [2,19,22 -24]. These DNA sequences form relatively stable nucleosomes, which are properly packed in the crystal. However, to understand the mechanism by which the genomic DNA is regulated in chromatin, the structures and physical properties of nucleosomes containing native genome sequences must be studied. In this context, a crystal structure of the nucleosome containing the 3 0 -LTR of the mouse mammary tumour virus sequence has been reported [21]. This is the only nucleosome structure containing a native regulatory DNA sequence published so far.
Currently, the cryo-EM method is becoming increasingly used for structural studies of nucleosomes. A high-resolution cryo-EM nucleosome structure has been reported at 3.9 Å resolution [20]. The cryo-EM structure of the nucleosome complexed with a retroviral integrase has been determined at 7.8 Å resolution [25]. The nucleosome with H2A lysine 15 (H2AK15) monoubiquitination and H4 lysine 20 (H4K20) methylation was reconstructed in a complex form with 53BP1, which functions in the double-strand break repair process, by cryo-EM at 4.5 Å resolution [26]. The nucleosome structure complexed with the domains of a nucleosome remodeller, yeast Chd1, has also been reconstructed at 4.8 Å resolution [27]. In addition to these mononucleosome structures, a cryo-EM structure of the polynucleosome containing a linker histone H1 has been determined at 11 Å resolution [28]. These outstanding studies have greatly advanced the structural biology of chromatin. However, these cryo-EM structures of nucleosomes were reconstituted with the Widom 601 DNA sequence, except for the nucleosome complexed with a retroviral integrase [25].
In this study, we successfully reconstructed the nucleosome structure with 186 base pairs of DNA containing the ALB1 enhancer (N1) sequence by the cryo-EM method at near-atomic resolution. This method allows us to analyse the nucleosome structure with a native DNA sequence. In addition, the cryo-EM method can avoid the potential restriction of the DNA structure by the effects of crystal packing.
We found that the ALB1 enhancer (N1) DNA exhibited higher accessibility to DNaseI than the Widom 601 sequence in the nucleosome (figure 6), although the DNA-binding paths of both nucleosomes are not significantly different (figures 3 -5). This higher accessibility suggests that the ALB1 enhancer (N1) DNA in the nucleosome may allow the efficient binding of pioneer TFs, such as FoxA, to the nucleosomal DNA. Interestingly, in the nucleosome, the ALB1 enhancer (N1) sequence contains flexible regions, and the DNA regions are located near a putative FoxA-binding region (figures 4 and 6). We also found that the histones are more weakly associated in the ALB1 nucleosome than in the Widom 601 nucleosome (figure 6). Therefore, the enhanced DNA accessibility and the weaker histone association found in the ALB1 nucleosome may be induced by the reduced local histone -DNA contacts. Unfortunately, the resolution of our cryo-EM structures (4.0-4.5 Å ) is not high enough to clarify the detailed histone -DNA interactions in the ALB1 nucleosome. Further structural studies will be required to reveal the mechanism by which the association of the histones is weakened and how the ALB1 enhancer (N1) DNA sequence becomes accessible to pioneer TFs in the nucleosome.

Purification of recombinant histones
All human histones were prepared as recombinant proteins, and were purified by the method described previously [29,30].

Mapping of nucleosome positioning by micrococcal nuclease treatment
The nucleosome containing the 180 base-pair ALB1 enhancer DNA was dialysed against 20 mM HEPES-NaOH (pH 7.5) buffer containing 1 mM DTT, and was treated with 0.5% formaldehyde at 258C for 30 s. The reaction was stopped by adding glycine to a final concentration of 200 mM, and the

Image processing
All frames in 2312 movies were aligned using MOTIONCOR2 [37], with dose weighting. The contrast transfer function (CTF) was estimated by CTFFIND4 [38] from digital micrographs, without dose weighting. In total, 1116 micrographs were selected based on good CTF fit correlation to approximately 8 Å resolution with minimal astigmatism. RELION 2.1 [39] was used for all subsequent image processing operations. A total of 1 182 985 particles of the ALB1 nucleosome were picked semi-automatically with a box-size of 140 Â 140 pixels, followed by two rounds of two-dimensional classification to discard bad particles, resulting in the selection of 626 544 particles. The crystal structure of a canonical nucleosome (PDB: 3LZ0), low-pass filtered to 60 Å , was used as an initial alignment model. After the first round of three-dimensional classification, 288 789 particles were selected for three-dimensional refinement, and C2 symmetry was applied to the three-dimensional reconstruction. Based on CTF fit correlation to approximately 5 Å resolution, 236 386 particles were further selected before three-dimensional refinement. The final C2-symmetrized map was sharpened with an exponential B-factor (2325 Å 2 ). For the threedimensional reconstruction without symmetry, a second round of three-dimensional classification was performed using a subset of 288 789 particles, and the reconstruction with the linker DNA containing 139 343 particles was selected for three-dimensional refinement. To sharpen the final unsymmetrized map, an exponential B-factor (2331 Å 2 ) was applied. The resolution of each final three-dimensional map was estimated following the gold standard Fourier Shell Correlation (FSC) at FSC ¼ 0.143 [39]. Maps were normalized with MAPMAN [40]. The model of 3LZ0.pdb was docked into the electron potential map of the ALB1 nucleosome reconstruction with UCSF CHIMERA [41]. The local resolution map of the ALB1 nucleosome was created by RESMAP [42]. Iso-electron potential surfaces were visualized with UCSF CHIMERAX [43] using the ambient occlusion shader (figure 3c) and UCSF CHIMERA [41] (figures 4 and 5).

DNaseI treatment assay
The

Thermal stability assay of nucleosomes
The thermal stability assay was performed by the method described previously [44,45]. Purified nucleosomes containing either the ALB1 enhancer DNA fragment or the Widom 601 DNA fragment (1.1 mM) were mixed with SYPRO Orange dye (Sigma-Aldrich) in 20 mM Tris -HCl buffer ( pH 7.5), containing 100 mM NaCl and 1 mM DTT. The SYPRO Orange fluorescence was monitored with a StepOnePlus TM Real-Time PCR system (Applied Biosystems), rsob.royalsocietypublishing.org Open Biol. 8: 170255 using a temperature gradient from 258C to 958C, in steps of 18C min 21 . The DNA sequences used in the thermal stability assay are described below.