Domain swapping and SMYD1 interactions with the PWWP domain of human hepatoma-derived growth factor

The human hepatoma-derived growth factor (HDGF), containing the chromatin-associated N-terminal PWWP domain capable of binding the SMYD1 promoter, participates in various cellular processes and is involved in human cancers. We report the first crystal structures of the human HDGF PWWP domain (residues 1–100) in a complex with SMYD1 of 10 bp at 2.84 Å resolution and its apo form at 3.3 Å, respectively. The structure of the apo PWWP domain comprises mainly four β-strands and two α-helices. The PWWP domain undergoes domain swapping to dramatically transform its secondary structures, altering the overall conformation from monomeric globular folding into an extended dimeric structure upon DNA binding. The flexible loop2, as a hinge loop with the partially built structure in the apo PWWP domain, notably refolds into a visible and stable α-helix in the DNA complex. The swapped PWWP domain interacts with the minor grooves of the DNA through residues Lys19, Gly22, Arg79 and Lys80 in varied ways on loops 1 and 4 of the two chains, and the structure becomes more rigid than the apo form. These novel structural findings, together with physiological and activity assays of HDGF and the PWWP domain, provide new insights into the DNA-binding mechanism of HDGF during nucleosomal functions.


Colony formation assay
In the assay of the colony formation, SK-Hep1 cells were seeded with 1 x 10 3 cells per well in the 6-well plate and fed with DMEM medium containing calf serum (10%).
Post cell adhesion, recombinant HDGF (10 ng/mL) and the PWWP domain (100 ng/mL) were added in DMEM medium containing calf serum (1%). Cells were cultured at 37 °C in humidified atmosphere with CO 2 (5%), and the fresh medium and recombinant proteins were replaced every three days. At the tenth day, culture plates were fixed with paraformaldehyde (4%), and then stained with crystal violet. Images of the colonies were captured, and colony numbers were counted with the statistical analysis.

Secondary structure analyses of the apo PWWP domain and HDGF
Circular-dichroism (SRCD) experiments were performed using a nitrogen-flushed instrument with X-rays from a synchrotron at TLS beamlines BL04B and BL04C at NSRRC. Samples were prepared in Tris buffer (20 mM, pH 7.5) before experiments. SRCD spectra were recorded near 23 °C for HDGF and the PWWP domain with concentrations 6 and 3 mg/mL, respectively, and the sample volume 20 µL. The measurements were repeated with three scans. The spectrum of background buffer was subtracted from the experimental data. Dominant monomeric PWWP domain was obtained at a low concentration (< 1.5 mg/mL). Also, the dimeric PWWP domain would change to monomers in buffer containing ionic-strength salts, such as NaCl (150 mM) within three days (D) The dimeric PWWP domain was acquired at a high protein concentration (> 1.5 mg/mL). (E) The symmetry-related apo PWWP domains in the crystals of hexagonal space group P6 4 22. The monomeric PWWP domain is shown in blue, whereas other symmetrically generated PWWP domain molecules related by crystallographic twofold axes (arrows) are shown as green, red and orange, respectively. The detail contacts are presented in enlarged boxes. Figure S2. Distances between loop1 and loop4 in the PWWP-SMYD1 complex. The distances (12.3 and 12.8 Å, both are from the N atom at Gly22 to the N atom at Lys80) between loop1 and loop4 from the C/N (green and marine) and the N/C termini (limon and blue; generated with the crystallographic symmetry of another complex) are near the spatial dimension (~ 12 Å) of the minor groove. Green and marine represent chains A and B, respectively. DNA molecules are shown in orange for the complex structure and gray for symmetry.  Treatments with HDGF (100 ng/mL) significantly enhanced the colony number of SK-Hep1 cells as compared with the control. In the treatment group with the apo PWWP domain, the colony numbers of cells were not notably changed. Values that differ significantly from controls are indicated as *P < 0.05 by one-way ANOVA. ns: non-significant.   The conserved PWWP motif is included in the black box. The DNA-binding residues in the HDGF PWWP-SMYD1 complex are indicated with *, whereas the binding residues from the symmetry-related HDGF PWWP-SMYD1 complex are marked with #. The conserved residues are highlighted with red (identical) and blue (low similarity) within the DNA-binding regions, loop1 ( 19 KMKG 22 ) and loop4 ( 77 NKRK 80 ). All those PWWP domains contain a highly conserved DNA-binding region, loop1 (especially residues Lys19 and Gly22). The other DNA-binding region, loop4 shares a low similarity among those PWWP domains except the isoforms from the HDGF family (HDGF, HDGF2 and HDGF3). The involved DNA-binding residues detected by NMR chemical-shift perturbation are marked with yellow and cyan (mouse and human HDGF PWWP domain share the same sequence). The alignment was generated with Clustal W with manual adjustment 47 . (B) The superimposition of the DNA-binding regions among the crystal structure of the PWWP-SMYD1 complex (gray), the NMR solution structure of the mouse HDGF PWWP domain (PDB entry: 2B8A, green), the NMR solution structure of the MSH6 PWWP domain (PDB entry: 2GFU, salmon) and the NMR solution structure of the LEDGF/p75 PWWP domain (PDB entry: 2M16, cyan). The DNA-binding regions in the PWWP-SMYD1 complex are the loops (black) that are similar to other PWWP domains (palegreen, red and blue).  4673-4680 (1994).

No. pair interactions
Chain A Chain B  Table S1. The numbers of pair interactions correspond to those in Fig. 2F.

Hydrogen-bonding residue pairs (The PWWP-SMYD1 complex) Distance (Å) Hydrogen-bonding residues pairs (The apo PWWP domain) Distance (Å)
Chain A Chain B  Table S4. The numbers of pair interactions correspond to those in Fig. S5B.