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.
Human hepatoma-derived growth factor (HDGF) is highly expressed in the developing heart, tumour cell lines and normal tissue ubiquitously with mitogenic and angiogenic activities1,2,3. Previous studies have demonstrated that HDGF participates in various cellular processes, such as the formation of vascular lesion4, astrocyte proliferation5 and cardiovascular differentiation1. Over-expressed HDGF can be found in several type of human cancers, including hepatocellular carcinoma6, pancreatic cancer7, oral cancer8, breast cancer9 and non-small cell lung cancer9. Human HDGF has also been shown to be involved in the phosphatidylinositol 3-kinase (PI3K)/Akt pathway, which is related to breast cancer and hepatocellular carcinoma. This pathway regulates post-translational histone modification6,10,11. Moreover, HDGF is involved in cardiovascular development and transcriptional regulation, resulting in its function as a transcriptional repressor3,12.
HDGF generally comprises two portions, an N-terminal domain with a highly conserved PWWP motif (residues 1–100 for human HDGF) and a variable C-terminal domain (residues 101–240)13. These two domains play distinct roles in physiology. The dominant nuclear location signal 2 (NLS2) located in the C-terminus in HDGF might assist the translocation of HDGF into the nucleus and subsequently regulates its mitogenic activity14,15. The PWWP motif of the N-terminal domain (PWWP domain) is named after the Pro-Trp-Trp-Pro motif, but only the third (Trp) and fourth (Pro) residues are highly conserved among various PWWP domain-containing proteins, whereas the first and second residues are variable16. The sequence of the PWWP motif of human HDGF is 24Pro-His-Trp-Pro27, of which the second residue is variable. The highly conserved PWWP domains are discovered in numerous eukaryotes, from unicellular organisms to humans. For instance, bromo and plant homeodomain finger–containing protein 1 (brpf1) that promotes histone acetylation utilizes its C-terminal PWWP domain to recognize histone modifications17,18. The mammalian DNA methyltransferase, DNMT3B, which contains the PWWP domain, is not only linked to human immunodeficiency, centromere instability and facial anomalies syndrome but also to distinct cancers19,20,21. Another PWWP-containing protein, Lens epithelium-derived growth factor (LEDGF/p75), plays an essential role in the integration of HIV-1 cDNA into human chromosomes22,23,24,25. To date, more than twenty PWWP domain-containing proteins from human genomes have been discovered, most of which are chromatin-associated26.
The PWWP domain can interact with heparin, a linear polymer consisting of repeating units of uronic acid and glucosamine residues on the cell surface16,27. The PWWP domain of HDGF may first bind to heparin through its positively charged residues and then be internalized into a cell27,28. In addition, nucleolin (NCL) located on the plasma membrane functions as a receptor for HDGF through interactions with of the HDGF PWWP domain6. Because of its chromatin-associated properties, the PWWP domain exhibits a methylated histone and DNA-binding capability16,29,30. The PWWP domain of DNA methyltransferase DNMT3B has been demonstrated to show no preference for double-stranded DNA (dsDNA), based on DNA binding assays29,31. Previous work has shown that the PWWP domain has greater binding affinities towards various types of dsDNA, instead of single-stranded DNA, indicating that the PWWP domain can interact with dsDNA nonspecifically32. In addition to the nonspecific DNA binding, the HDGF PWWP domain is capable of recognizing the SMYD1 promoter (5′-CAGGCTGGTCTTGAACTCCTGACCTCAGATGATCCATG TGCCTCGGCCTCCCAAGGTGGGGATTACAGGCGTGAGCCACC-3′)12. HDGF has been demonstrated to bind to the SMYD1 promoter containing a HDGF core-binding sequence with a shortest length of 37 bp from human chromatin and functions as a transcriptional repressor3,12. HDGF interacts with the SMYD1 promoter through its N-terminal PWWP domain; however, the precise interaction of the PWWP domain with SMYD1 remains unclear12.
Several structures of the PWWP domains of various proteins from different species have been reported using NMR and X-ray approaches, but they represent only either the apo form or complexes with histone-related peptides16,28,29,30. The NMR structures of a monomeric form of the apo soluble HDGF PWWP domain and one domain-swapped dimeric form of the apo refolded PWWP domain have been determined, respectively (PDB entries: 1RI0, 2NLU)27,28. Despite the preceding work, the mechanisms of DNA binding and interactions with the PWWP domain remain puzzling because of a lack of essential knowledge regarding the intact structure in complex with DNA molecules. Here, we report the first crystal structures of the human HDGF PWWP domain in complex with a 10-bp SMYD1 and its unbound apo form. Our structural studies, with physiological assays, provide new insights into the PWWP-DNA interaction pattern, which might facilitate research into the role of the PWWP domain in nucleosomal functions.
Overall structure of the HDGF apo PWWP domain
We obtained crystals of the apo PWWP domain after several crystallization attempts, and determined the crystal structure at 3.3 Å resolution (Table 1). Crystals belong to the hexagonal space group P6422 with one PWWP domain in an asymmetric unit. The overall structure of the apo PWWP domain comprises mainly four β-strands and two α-helices. The truncated protein sequence consists of the first 100 amino acids of HDGF, but the first N-terminal eight residues are disordered without an interpretable electron density. The structure thus begins from a short N-terminal turn followed by two β-strands (β1 and β2) (Fig. 1A,B). The 24PWWP27 motif is located on loop1 between the β1 and β2 strands. The structure of loop2, after the strand β2, is only partially built because of its diminished electron density from Ala37 to Lys39, indicating that this loop is flexible in the apo PWWP domain (Fig. 1C). The other two β-strands (β3 and β4) follow loop2, with one connecting loop linked to the C-terminal α-helices (αA and αΒ) and another connecting loop (Fig. 1A,B). The X-ray structure of the apo HDGF PWWP domain exhibits several notable structural features that differ from the NMR solution structure (PDB entry: 1RI0): the length and amounts of β-strands as well as the locations of the α-helices. One additional short β-strand (Leu63 – Phe64) exists in the solution structure. The positions of the defined first α-helix (αA) in the two structures are also distinct (Fig. 1D).
An inspection of the distribution of surface charge of the apo PWWP domain reveals that the positively charged region and the slightly negatively charged and neutral region are separated on two opposite sides of the molecular surface (Fig. 1E). This charge distribution might result in the apo PWWP domain possessing a high value of pI, 9.19. The positively charged surface is generally thought to be the primary area interacting with the negatively charged DNA molecule, but our structure of the DNA complex shows an unexpected result and constitutes a detailed DNA-binding region, which is described in the following sections.
Multiple forms of the HDGF apo PWWP domain
In general large-scale culture procedures, the apo PWWP domain could be collected from two fractions with distinct elution times in the size-exclusion chromatography, as confirmed by SDS-PAGE (Fig. S1A,S1B). These two fractions indicated that the apo PWWP domain could exhibit both monomeric and dimeric forms. The ratio of these two forms depended on the initial protein concentration during the size-exclusion chromatography. The apo PWWP domain is present as the monomer with an initial concentration of less than ~1.5 mg/mL, whereas the proportion of dimer increased as the concentration increases (Fig. S1C,S1D). Moreover, based on our observations, the dimeric PWWP domain in buffer containing the ionic-strength salt, such as NaCl (150 mM), could be transformed into the monomers within three days. In crystals of the apo PWWP domain, monomers related to crystallographic two-fold symmetries exhibit few interactions with residues on β4 and αB regions, respectively, with the average hydrogen-bond distance larger than 3.1 Å (Fig. S1E). Therefore, the weak interactions between monomers may occur when the concentration of the PWWP domain is increased and these contact monomers may be dissociated when the concentration decreases.
Overall structure of the domain-swapped HDGF PWWP-SMYD1 complex
Examination of the crystal structure of the apo PWWP domain reveals that the positively charged area on the surface of the apo PWWP domain covers ~45 Å in length that might accommodate and interact with DNA of a size up to 13 bp (Fig. 1E). We have therefore designed SMYD1 of various lengths accordingly for complex formation. To facilitate crystallization of the HDGF PWWP-SMYD1 complex, we have tested three SMYD1 with sizes of 5, 10 and 15 bp. The complex formation is confirmed by size-exclusion chromatography; the elution profiles of these complexes with SMYD1 of various lengths are all shifted compared to that of the apo PWWP domain (Fig. 2A). The respective fractions of various complexes are subsequently collected for crystallization. Among the three complexes, crystals of the PWWP complex with 10-bp SMYD1 have been obtained. The crystal structure of the HDGF PWWP-SMYD1 complex is determined at 2.84 Å resolution (Table 1). The asymmetric unit contains one swapped PWWP dimer and one 10-bp SMYD1, indicating that the PWWP domain dimerises with domain swapping to interact with one DNA molecule in the PWWP-SMYD1 complex.
Based on the unambiguous electron density, the conformation of the PWWP domain in the complex has altered unexpectedly from a globular shape in the apo form structure to an extended shape (Figs 1A and 2B,C,D). The composition of the secondary structure contents in the PWWP-SMYD1 complex is distinct from that in the apo PWWP domain (Figs 1B and 2E). In particular, the flexible loop2 with the partially built structure in the apo PWWP domain notably refolds into a visible α-helix (αC) based on the omit map (Figs 1C and 2C,D). This intrinsic flexible loop is apparently capable of exhibiting distinct conformations in both the monomer of the apo PWWP domain as well as the complex dimer, where it functions as a hinge loop 33. The C-terminus of one protein molecule (chain A) is elongated further after this newly formed αC helix and interacts tightly with the N-terminal region of the other molecule (chain B) to form a dimer, which undergoes 3D domain swapping 33 (Fig. 2F). To explain the swapped domains in the following sections, we define one swapped domain consisting of the C-terminus from chain A and the N-terminus from chain B as the “swapped C/N terminus” and the other swapped domain as the “swapped N/C terminus” (Fig. 2B).
Several residues are involved in the dimer formation of the PWWP domain in the SMYD1 complex. Thr41 located on each hinge loop from the A and B chains can interact with each other. Tyr10, Leu15, Phe17, Lys19, Gly22, Trp26, Pro27, Arg29 and Asp31 in chain A interact with Tyr45, Gln46, Phe48, Phe50, Asp62, Phe64, Tyr66, Arg79 and Pro93 in chain B to maintain the dimeric structure of the complex (Fig. 2G). Among these interacting residues, Arg29 presumably plays an important role in protein dimerization because it is involved in four hydrogen bonds through interactions with two residues, Phe48 and Pro93, whereas the other residues form only one or two hydrogen bonds. All residues involved in protein dimerization of the PWWP-SMYD1 complex with the corresponding interaction distances are shown in Table S2.
Interactions between the HDGF PWWP domain and 10-bp SMYD1
Based on the structure of the HDGF PWWP-SMYD1 complex, two major DNA-binding regions are identified: loop1 (19KMKG22) from chain B and loop4 (77NKRK80) from chain A in the swapped C/N terminus (Fig. 2B). Accordingly, chain A utilizes Arg79 and Lys80 on loop4 to interact with DNA; chain B uses Lys19 and Gly22 on loop1 to bind DNA at the C/N terminus (Fig. 3A).
Because of the pseudo two-fold symmetry within the structure of the HDGF PWWP-SMYD1 complex, another potential DNA-binding region may exist on loop1 and loop4 of the swapped N/C terminus (Fig. 3A). Therefore, we have generated the symmetry-related PWWP-SMYD1 complex molecules to illuminate the DNA-binding regions on swapped C/N and N/C termini in more detail. At the swapped N/C terminus, chain A interacts with DNA through Lys19 and Gly22, and chain B interacts with DNA via Asn77 and Lys80 (Fig. 3A). All residues involved in DNA interactions and their corresponding interaction distances are depicted in Table 2.
Considering both the swapped C/N and N/C termini, a total of eight residues participate in DNA interactions, including Lys19, Gly22, Arg79, Lys80 in chain A and Lys19, Gly22, Asn77, Lys80 in chain B. Among them, Gly22 and Lys80 in both chains contact DNA through the main-chain N atoms rather than their side chains. The PWWP domain thus may interact with DNA through residues with varied interactions in addition to the charge-charge interactions (Fig. 3A and Table 2). The distance (12.3 or 12.8 Å) tip-tip between loop1 and loop4 at the swapped C/N or N/C terminus is similar to the spatial dimension (~12 Å) of a minor groove of DNA (Fig. S2).
In summary, the HDGF PWWP domain utilizes loop1, which occurs before the PWWP motif, and loop4, before the αB helix, on both swapped C/N and N/C termini to interact with DNA. The molecular dimension of the swapped dimer with the positively charged surface of the PWWP-SMYD1 complex is ~ 60 Å, but only the partial region is involved in DNA interactions (Fig. 3B). Together, the charge-charge interactions may thus play an essential role in facilitating the PWWP domain to approach DNA, and subsequently the residues with varied interactions of the swapped C/N and N/C termini interact and bind to DNA to form the stable PWWP-SMYD1 complex (Fig. 3).
Structural and B-factor comparisons of the PWWP-SMYD1 complex and apo form
A structural comparison between the apo PWWP domain and the PWWP-SMYD1 complex reveals some major conformational differences, including alterations of the hinge loop region, changes in αA, the appearance of strand β5 and the distinct folding of monomers and dimers. The overall B-factor of the PWWP-SMYD1 complex is much smaller than that of the apo form (Fig. 4A). We further compared the average B-factors of the regions with notable differences between two structures: the hinge loop; the DNA-binding region; αA and β5; the hydrogen-bonding residue pairs for dimerization of the complex and the corresponding residues of the monomeric apo form. To properly compare the B-factors between two structures, we normalised B-factor at the above-mentioned regions. The normalised B-factor for each region is a ratio calculated by dividing the average value for the region by the overall B-factor for the whole molecule. All normalised B-factor ratios of the regions involved in structural differences between the complex and the apo form are shown in Table S3.
For the largest structural differences at the hinge loop (Asp31 – Lys44) (Fig. 2F), the normalised B-factor of the newly formed helix αC is 1.0 at each chain in the PWWP-SMYD1 complex. In contrast, the normalised B-factor of this same region is 1.51 in the apo PWWP domain (Fig. 4B), suggesting that the hinge loop region is relatively rigid in the PWWP-SMYD1 complex but much flexible in the apo PWWP domain. This similar structural flexibility could be observed from the electron density map at this region of two structures (Fig. S3). The density map is broken from Asp31 to Lys44 in the apo PWWP domain (Fig. S3A), whereas the electron density is continuous in this region in the complex (Figs 2D and S3B). Next, the region from αA to β5 (Pro60 – Pro65) and the DNA-binding region, loop1 (Lys19 – Gly22) and loop4 (Asn77 – Lys80), are also relatively flexible in the apo form (Fig. 4C,D). Finally, the normalised B-factor (0.8) of the hydrogen-bonding residues for swapped dimerization of the PWWP-SMYD1 complex is lower than that (0.94) of the corresponding residues of the monomeric apo PWWP domain (Fig. 4E). Taken together, the structure of the HDGF PWWP-SMYD1 complex is more rigid and stable than that of the apo PWWP domain.
SMYD1 binding attenuates the interaction of HDGF with NCL
HDGF is known to be related to various cancers6,7,8,10. The colony formation assay, a well-used indication for cancer research34, was performed for confirming whether the recombinant HDGF and PWWP domain retains biological functions. The results showed that HDGF exhibited functional ability to promote colony formation better than the PWWP domain because of the dominant NLS2 in the C-terminus of HDGF (Fig. S4).
Moreover, it has recently been demonstrated that HDGF interacts with membrane nucleolin (NCL) via the PWWP domain6. In addition to the plasma membrane, such an interaction mainly occurs in the nucleus, where HDGF and NCL are abundant. Because HDGF also binds to SMYD1 12, we would like to delineate whether SMYD1 binding affects the interaction between HDGF and NCL. It has been found that HDGF and the PWWP domain selectively bind to the NCL-coated matrix by the solid phase binding assay (Fig. 5A). Interestingly, the HDGF-SMYD1 and PWWP-SMYD1 complexes exhibited a significantly reduced binding capacity to NCL compared with unbound apo HDGF and the apo PWWP domain, respectively. Subsequently, we have employed the competition assay and have revealed that the excessive PWWP domain potently disrupts the interaction between HDGF and NCL, whereas the PWWP-SMYD1 complex is less effective (Fig. 5B). Together, these results indicate that SMYD1 binding interferes with the interaction of HDGF with NCL in the nucleus.
The multiforms of the PWWP domain for DNA binding
According to our results, the protein concentration can affect the formation of multiple forms of the apo PWWP domain, and the dimeric apo PWWP domain can be readily obtained by increasing the protein concentration during purification (Fig. S1C, S1D). The apo PWWP domain has also been demonstrated to potentially form a dimer to bind to heparin on the cell surface with a binding affinity greater than that of the monomer; its transformation between dimers and monomers depends on the freshness of the sample preparation27. The dimeric and monomeric forms of the apo PWWP domain can hence co-exist, depending on the environment with variable conditions.
The PWWP domain is well known to be able to interact with the promoter and nonspecific DNA12,31. Our results show that the apo PWWP domain can form complexes with SMYD1 of various lengths (Fig. 2A). Size calibration chromatography indicates that the apo PWWP domain and the PWWP-5 bp SMYD1 complex exhibit as monomers, whereas the PWWP-10 bp and PWWP-15 bp SMYD1 complexes exist as dimers. These observations might explain how the monomeric PWWP domain can still possess functions similar to dimers, such as heparin28 and DNA-binding abilities31 (Fig. 2A). It is thus reasonable to assume that the monomeric apo PWWP domain is capable of binding to the 5-bp SMYD1 through its two intrinsic loops— loop1 (Lys19 – Gly22) and loop4 (Asn77 – Lys80) (Figs 2A and S5A). In contrast, the PWWP domain may utilize a dimeric structure to interact with DNA larger than 10 bp during its contact with DNA in the nucleosome under physiological conditions (Fig. 2A,B).
Correspondingly, structural comparison of the HDGF PWWP-SMYD1 complex and the apo PWWP domain reveals several significant conformational distinctions (Fig. 2F). The complex stretches its overall structure into an extended shape as a domain-swapped dimer from the globular shape of the apo form in the monomer. The distance between two DNA fibres in a nucleosome is ~5.5 nm, which is about the dimension (~6 nm) of the extended main-chain structure of the PWWP-SMYD1 complex, suggesting that the PWWP domain may explore the methylated histone and the promoter with both swapped C/N and N/C termini35.
Based on domain swapping to exchange the similar domains between two protein molecules (Fig. S5A), the pattern of hydrogen-bonding residue pairs of one swapped terminus in the PWWP-SMYD1 complex is similar to that in internal interactions of the monomeric apo PWWP domain (Fig. S5B). The amount and average bond length of these hydrogen-bond pairs are, however, different in the two forms. This condition thus implies that fewer hydrogen bonds (9 vs. 13) with a larger bond distance in the apo PWWP domain may be just sufficient to maintain the globular structure of a monomer with larger temperature factors, and provide the possibility for domain swapping to form a much more stable dimer of the complex (Fig. 2F and Table S1 and S4).
Dimerization by domain swapping of the HDGF PWWP domain for DNA binding
The intrinsic flexibility of the hinge loop with partial density of the apo PWWP domain is consistent with multiple conformations of loop2 in the NMR solution structure (PDB entries: 2B8A, 1RI0, 2NLU) (Fig. 6A,B). The refolded dimeric PWWP domain without DNA in the NMR structure (PDB entry: 2NLU) exhibits domain swapping with the flexible loop2 (Glu32 – Lys44) under an ultrahigh concentration (100 mg/mL). In our PWWP-DNA complex structure, however, the flexible loop alters its conformation into an α-helix. Moreover, the structural orientations of the apo dimeric PWWP domain obtained with NMR and the dimeric PWWP-SMYD1 complex with X-ray are variable, beginning with its flexible loop2 upon superimposing these two structures on one swapped terminus (Fig. 6C). The flexible loop2 in the apo form is transformed to a rigid helix αC in the PWWP-SMYD1 complex, indicating that loop2 may be involved in the domain swapping. In some domain-swapped proteins, the transformation of a hinge loop to an α-helix or β-strand indicates that the dimer or oligomer form is favoured over the monomer33,36. The energy barrier for domain swapping of oligomers could be reduced under several conditions, such as temperature changes, protein mutations, increased protein concentrations and the binding of a ligand33. The PWWP domain should hence be stable with a swapped dimeric form to interact with DNA more than with a monomeric form. The domain-swapped PWWP domain thus increases larger DNA-binding surfaces, which may benefit the DNA interactions.
The structure of the apo PWWP domain shows an interpretable electron density but with an overall B-factor that is larger than for the complex (Fig. 4A). It is thus reasonable to assume that the activation energy of the flexible monomeric PWWP domain may be decreased upon altering its conformation to form a more rigid dimer during attachment to DNA. Additionally, the B-factor values of the relevant residues that participate in protein dimerization and DNA interactions in the PWWP-SMYD1 complex are much smaller than those in the apo PWWP domain. These observations indicate that DNA might help not only to alter the conformation of the hinge loop in the PWWP domain but also to stabilize the complex conformation (Fig. 4A and Table S3).
The overall secondary-structure content of HDGF contains a large fraction of random coils as shown by SRCD, whereas the structure of the apo PWWP domain comprises mostly β-strands, indicating that the C-terminus of HDGF may be flexible with a large portion of the coil structures (Figs 1A and S6). The previous work has shown that a protein containing a large portion of random coils can undergo domain swapping without interference37. The coiled C-terminus of HDGF hence may not affect domain swapping of the N-terminal PWWP domain as observed in the structure of the PWWP-SMYD1 complex.
Structural comparison with other PWWP domains and implications
A structural comparison of the apo HDGF PWWP domain with other PWWP domains from mouse and human HDGF, HDGF2, mismatch repair protein MSH6, LEDGF/p75, DNMT3B (PDB entries: 2B8A, 2NLU, 3EAE, 2GFU, 2M16 and 5CIU) shows that these PWWP domains comprise four or five stranded β-barrel and α-helices in their C-terminus, indicating that the main structural features of those PWWP domains are conserved, especially for the β-barrel region (Fig. 6D).
Before our work, no structure of the PWWP-SMYD1 complex was available, but several research groups had attempted to map the DNA-binding region with NMR chemical-shift perturbation experiments31,32,38. Previous studies of the DNA-binding model of the mouse HDGF PWWP domain (PDB entry: 2B8A) indicated that the N-terminus (Lys8 – Tyr10), the loop region (Lys19 – Tyr23) and two α-helix regions (Gly59 – Leu63; Tyr66 – Phe73), which contribute to the positively charged surface, were involved in DNA binding, especially with Glu9, Tyr10, Gly22, Lys61 and Arg7931. Based on our crystal structure of the PWWP-SMYD1 complex, the DNA-binding regions are loop1 (Lys19 – Gly22), which is before the highly conserved PWWP motif, and a partial loop4 (Asn77 – Lys80), differing from the prediction obtained with NMR chemical-shift perturbation. This result potentially suggests that the PWWP domain interacts with a longer region of DNA through larger regions (Fig. 3 and S7A).
The identified interactions between two DNA-binding loops and the minor groove of DNA, which comprises ~ four base pairs with an axial dimension ~12 Å in the B-form DNA, may explain the obvious chemical shift of interactions between the PWWP domain and the 15-bp DNA detected only in the middle 5 bp of DNA31 (Fig. S2). Both NMR chemical-shift perturbation analyses of the PWWP domain of Human Mismatch Repair Protein -MSH6 (PDB entry: 2GFU) with 35-bp DNA and the PWWP domain on the Lens epithelium-derived growth factor (LEDGF) (PDB entry: 2M16) with DNA suggested that some residues involved in DNA binding occur in the loop1 region and the residues of the corresponding loop4 region32,40 (Fig. S7A). Our human HDGF PWWP-SMYD1 complex further provides direct evidence and interaction details of loop1 (Lys19 – Gly22) and loop4 (Asn77 – Lys80) with DNA in the PWWP domain (Fig. S7A). In addition, the PWWP domain can interact with a longer region of DNA through its basic patch32,40.
In our HDGF PWWP-SMYD1 complex, the PWWP domain contacts DNA with residues of varied types that are surrounded by a positively charged area, especially loop1 and loop4. The sequence alignment of various PWWP domains shows that the DNA-binding loop1 contains highly conserved residues, whereas loop4 comprises only partially conserved residues. The sequences 77NKRK80 of loop4 in the isoforms of HDGF, HDGF2 and HDGF3 are, however, identical, indicating that all isoforms of HDGFs may also interact with DNA via this loop region (Figs 3A and S7A).
From a structural perspective, the DNA-binding residues in the HDGF PWWP-SMYD1 complex are located on loop1 and loop4, which are within the proposed DNA-binding area in those structures monitored by NMR chemical shift perturbation (PDB entries: 2B8A, 2M16, 2GFU), implying that these PWWP domains may specifically contact DNA through the identified residues at the two loop regions (Fig. S7B). The orientation of loop4 in the MSH6 PWWP domain is different from other PWWP domains. Although the flexibility of this loop from all NMR models of the MSH6 PWWP domain is relatively rigid, it may also alter its orientation when interacting with DNA.
The importance of the positively charged distribution in the complex
Previous work using the electrophoresis mobility shift assay (EMSA) to examine truncated HDGF with 80-bp SMYD1 showed that the truncated proteins HDGF1-35, HDGF1-70, HDGF70-237 cannot interact with 80-bp SMYD1, whereas HDGF35-237 retains a partial binding ability with 80-bp SMYD1 12. Assuming no severe influence on protein folding by truncation, all of those truncated regions lack one of two DNA-binding loops, either loop119-22 or loop477-80. As mentioned, the PWWP domain is equipped with a positively charged surface on only one side of the molecular surface (Fig. 1E). We found that the region from Lys39 to Lys80 contributes to a positively charged surface, which is included in the 35HDGF237 (Figs 1E and 3B). In addition, Lys8, Lys19 – His25, Lys39 – Lys44, Leu58 – Phe64 and Lys72 – Lys80 compose the basic patch in the PWWP domain, which may be involved in interactions with the longer DNA. We therefore reasonably assume that DNA should be first recruited by the positively charged force on one side of the exposed surface and subsequently interacts with the DNA-binding regions, which is identified in this work, in the PWWP domain.
Hypothesis of the mechanism of DNA binding and the functional process
Taken together, we propose a hypothesis for the mechanism of DNA binding and the functional process of HDGF in gene regulation (Fig. 7). The growth factor HDGF is expressed in the highly developed foetal tissue, and over-expressed HDGF causes a greater risk of various cancers4,39,40. Hence, HDGF might first interact with heparin on the cell surface with its dimeric and monomeric forms with distinct binding affinities27, or bind to NCL, a plasma membrane receptor of HDGF6. After internalizing into the cell, the HDGF recognizes the receptor on the nucleus pole by its specific sequence of NLS2 in the variable C-terminus to translocate itself into the nucleus14. The interactions of monomeric or dimeric HDGF with the receptor remain unclear. Subsequently, HDGF of dimers or monomers may gather together in the nucleus under physiological conditions. Next, HDGF may undergo domain swapping with the PWWP domain, and the swapped dimeric HDGF in the nucleus transforms its flexible hinge loop further into a rigid hinge α-helix when the protein approaches DNA fibres with loop1 and loop4. Once the PWWP domain is activated by methyl peptides of histone and recognizes the promoter, the synergistic binding mechanism may enhance the binding affinity with the corresponding histone and DNA and participate further in the regulation of gene expression through the C-terminus of HDGF12,15,30.
In summary, we have determined the first crystal structures of the HDGF PWWP domain in complex with SMYD1 of 10 bp and its apo form, respectively. The PWWP domain undergoes domain swapping to markedly transform its secondary structures and to alter the overall conformation from monomeric globular folding into an extended dimeric structure upon binding to DNA. The PWWP-DNA complex becomes more rigid and stable than the apo form. The PWWP domain utilizes residues with variable characters at loop1 and loop4 from two chains of the swapped dimer to interact with the minor grooves of DNA. Together with physiological assays, these novel structural findings provide new insights into the mechanism of DNA binding and the functional process of HDGF.
Construction of the HDGF PWWP1-100 domain
The construct of human HDGF was amplified from the library of human fetal brain cDNA (Stratagene, La Jolla, CA) utilizing the polymerase chain reaction (PCR) as described previously41. After the DNA sequencing analysis, the PCR-amplified PWWP domain was subcloned into NdeI and EcoRI sites of the pET28a vector and transformed into Escherichia coli (E. coli) BL21-Codon Plus®-RIL for expression and purification of the recombinant PWWP protein. All constructed plasmids were confirmed by DNA sequencing.
Expression and purification of the HDGF PWWP1-100 domain from E. coli
A single colony of the bacteria was selected and placed into Luria-Bertani (LB) broth (20 mL) with kanamycin (50 μg/mL) at 37 °C overnight. The culture was then transferred into fresh LB medium (2 L) with kanamycin (50 μg/mL). After the OD600 reached 0.4–0.6, isopropyl β-D-thiogalactopyranoside (IPTG, final concentration 1 mM) was added to the culture for induction with rotary shaking at 37 °C for another 20 h. The cells were harvested by centrifugation (6,000 × g) at 4 °C for 10 min.
The cells were resuspended in lysis buffer (20 mM Tris-HCl, pH 7.5) and then sonicated by ultrasonication (SONICS VCX 750) using a pulsation cycle (2 s on, 4 s off) for a total duration of 12 min of sonication at 40% energy on ice. The soluble PWWP protein extract was collected by centrifugation (12,000 × g) at 4 °C for 30 min; the supernatant was loaded onto a Ni2+-NTA agarose column (GE healthcare) that had been pre-equilibrated with binding buffer (20 mM Tris-HCl, pH 7.5). The unwanted proteins were washed out with binding buffer containing imidazole (20–50 mM); the desired target protein of the PWWP domain was eluted with binding buffer containing imidazole (300 mM). The protein was further purified by anion-exchange chromatography (Hitrap Q) with buffer A (20 mM Tris-HCl, 5 mM EDTA, 5 mM β-ME pH 7.5) and buffer B (20 mM Tris-HCl, 1 M NaCl, 5 mM EDTA, 5 mM β-ME pH 7.5) while monitoring the ratio of A260/280 based on the markedly distinct pI values of the apo PWWP domain and DNA during purification. This purification step removed endogenous DNA from E. coli and contaminated proteins to ensure that there was no interference and contamination of endogenous DNA for further structural studies and physiological assays28,32. The homogeneity of the PWWP domain protein was confirmed by size-exclusion chromatography (Superdex-200) with buffer (20 mM Tris-HCl, 5 mM EDTA, 5 mM β-ME, pH 7.5). The ratio of A260/280 was examined, and showed that the purified apo PWWP domain was not contaminated with endogenous DNA from E. coli. The PWWP protein was concentrated (on Centricon, MWCO 10,000; Satorius Vivapin 20) before crystallization. The purity of the final PWWP domain was greater than 95% based on SDS-PAGE (15%) and staining (Coomassie Brilliant Blue R-250).
Expression and purification of HDGF from E. coli
The expression and purification procedures for full-length HDGF were similar to that for the apo PWWP domain described in the above section. A pre-equilibrated binding buffer (20 mM Tris-HCl, 150 mM NaCl, pH 7.5) was prepared for a Ni2+-NTA agarose column. The unwanted proteins were washed out with binding buffer containing imidazole (20–50 mM), and the desired target protein of HDGF was eluted with a binding buffer containing imidazole (150 mM). The subsequent step with anion-exchange chromatography (Hitrap Q) was similar to that for the apo PWWP domain. The homogeneity of the full-length HDGF protein was confirmed by size-exclusion chromatography (Superdex-200) with buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 5 mM β-ME). HDGF was concentrated (on Centricon, MWCO 10,000; Satorius Vivapin 20) before colony-formation and Western-blot assays and structural studies. The purity of HDGF was greater than 95% based on SDS-PAGE (12%) and staining (Coomassie Brilliant Blue R-250).
Purification of the HDGF PWWP-SMYD1 complex
After purification by the ion-exchange chromatography to exclude endogenous DNA from E. coli, if any, we incubated DNA of various lengths with the sequence of the SMYD1 (15, 10 and 5 bp) and the PWWP domain at 4 °C overnight. The SMYD1 sequences were 5′-CAGGCTGGTCTTGAA-3′, 5′-CAGGCTGGTC-3′ and 5′-CAGGC-3, respectively. Complementary single-stranded DNA molecules ordered from the commercial company (Mission Biotech) were mixed together and then heated at 95 °C for 10 min. After heating, the DNA mixtures were left near 23 °C overnight to form the double-stranded DNA. The monomeric PWWP domain and SMYD1 were incubated at a molar ratio of 1:1. The complex of the PWWP domain and SMYD1 was purified again by size-exclusion chromatography (Superdex-200) in a buffer (20 mM Tris-HCl, 5 mM EDTA, 5 mM β-ME, pH 7.5) to ensure that the PWWP-SMYD1 complex of high homogeneity was properly collected and that unbound DNA was removed to avoid a perturbing effect on the subsequent crystallization.
Solid-phase binding assay
The solid-phase binding assay was performed to detect the binding ability of HDGF, HDGF-15 bp SMYD1 complex, the PWWP domain and the PWWP-15 bp SMYD1 complex to the nucleolin (NCL) protein. Human NCL cDNA was constructed using the pET15b vector and transformed into E. coli BL21 (DE3) cells. After IPTG (1 mM) induction, the protein was purified using the Ni2+-NTA agarose column. For the solid-phase binding assay, NCL coating was applied at 10 µg/mL at 4 °C overnight. Each well was blocked with BSA (3%) in PBS buffer containing Tween-20 (0.05%) at 25 °C for 1.5 h. Antigen proteins (HDGF and PWWP) were reacted at 25 °C for 1.5 h. The PWWP-specific antibody and the HRP-conjugated antibody were used to detect HDGF and PWWP domain proteins. Each well was reacted in TMB substrate at 25 °C, and H2SO4 (1 N) was applied as a stop solution. Finally, the optical density of each well was measured at 450 nm using an ELISA reader.
The competitive assay was performed to validate whether the PWWP domain was competitive for the HDGF-NCL interaction. For this assay, the human HDGF cDNA was constructed in the pcDNA3.1 expression vector using a fusion of the human antibody-Fc fragment. For protein production, the plasmid was transfected into human embryonic kidney 293 T cells using Lipofectamine® 2000 reagent. The conditional medium was collected, and the HDGF-Fc protein was purified using a protein G Sepharose® column. In the competitive assay plate, NCL was coated at a concentration 10 µg/mL at 4 °C overnight. After BSA (3%) blocking, the HDGF-Fc fusion protein (0.01 μM) was reacted at ten times the molar concentration of the PWWP domain (0.1 μM) and PWWP-SMYD1 (0.1 μM) for competition. The goat anti-human-HRP antibody was used for detection of the HDGF-Fc fusion protein. Finally, the TMB substrate and H2SO4 (1 N) as a stop solution were used for the colour reaction.
Crystallization of the apo PWWP domain and HDGF PWWP-SMYD1 complex
The purified PWWP domain and HDGF PWWP-SMYD1 complex proteins were concentrated to 10–20 mg/mL in a storage buffer (20 mM Tris-HCl, 5 mM EDTA, 5 mM β-ME, pH 7.5) containing protease inhibitor cocktail (1%, Roche). Crystallization trials were performed in 96-well plates (JET Biofil) using several crystal-screening kits based on the hanging-drop vapour-diffusion method utilizing hanging drops (0.4 μL) containing equal volumes of protein solution (0.2 μL) and a reservoir solution (0.2 μL) against the bottom reservoir solution (100 μL) at 8 and 18 °C, respectively. The apo PWWP in storage buffer with spermine tetrahydrochloride (10 mM), which functioned as a kind of precipitant for DNA to remove soluble DNA, was crystallized under conditions containing sodium chloride (200 mM), Tris (100 mM, pH 8.5), and polyethylene glycol (PEG) 4000 (25%, w/v). The PWWP-SMYD1 complex was crystallized under conditions containing sodium nitrate (90 mM), sodium phosphate dibasic (90 mM), Tris; Bicine (100 mM, pH 8.5), PEG 1000 (12.5%) and PEG 4000 (12.5%, w/v).
X-ray data collection, structure determination and refinement
The crystals were mounted on a synthetic nylon loop (0.05–0.1 mm, Hampton Research Co.) and then flash-cooled in liquid nitrogen. The protein crystals were first tested with X-rays from synchrotrons at TLS beamlines BL13B1, BL13C1 and BL15A1 at the National Synchrotron Radiation Research Center (NSRRC) in Taiwan and BL12B2 at SPring-8 in Japan. For complete data collection of the apo PWWP domain at the best resolution, a total rotation of 120° with oscillation of 1.0° were measured at BL44XU of SPring-8 using an X-ray wavelength of 0.9 Å with an exposure duration of 1 s and distance of 500 mm from the crystal to the detector at 110 K under a nitrogen stream provided by a cryo-system (X-Stream, Rigaku/MSC, Inc.). For complete data collection of the PWWP-SMYD1 complex, a total rotation of 200° with oscillation of 0.5° were measured at the Taiwan Photon Source (TPS) beamline 05A equipped with a CCD detector (MX300-HS, Rayonix) using an X-ray wavelength of 1.0 Å, exposure duration of 0.1 s and distance of 350 mm from the crystal to the detector at 110 K under a nitrogen stream provided by a cryo-system. All data were indexed, integrated and scaled using the programme suite HKL-2000 42.
The crystal of the apo PWWP domain belonged to the space group P6422, with unit cell dimensions a = b = 79.49 Å and c = 105.10 Å, and diffracted to 3.3 Å resolution. The crystal of the PWWP-SYMD1 complex belonged to the space group P31, with unit-cell dimensions a = b = 32.40 Å and c = 205.82 Å, and diffracted to 2.84 Å resolution. Structures of both the apo PWWP domain and the PWWP-SMYD1 complex were determined using the molecular replacement method with the Molrep programme43. The crystal structure of the apo PWWP domain was first determined with the X-ray structure of the PWWP domain of human HDGF2 (PDB entry: 3QBY) as a search model. The HDGF PWWP-SMYD1 complex was later determined with the solved structure of the HDGF apo PWWP domain as the protein model template and the structure of a 10-bp DNA as the DNA search model, respectively. All structures were built with COOT 44. Refinement was performed with REFMAC5 and CCP4 45. The correctness of the stereochemistry of each structure was validated with MolProbity 46. The calculations of root-mean-square deviation from angles, bonds and dihedral and improper angles showed the satisfactory stereochemistry. All details of the crystallographic data and refinement statistics are shown in Table 1. The figures were prepared in PyMOL (http://www.pymol.org/).
Coordinates and structure factors of the HDGF PWWP-SYMD1 complex and the apo PWWP domain have been deposited in PDB under the accession codes 5XSK and 5XSL, respectively.
Everett, A. D. Identification, cloning, and developmental expression of hepatoma-derived growth factor in the developing rat heart. Dev. Dyn. 222, 450–458 (2001).
Wanschura, S. et al. Mapping of the Gene Encoding the Human Hepatoma-Derived Growth Factor (HDGF) with Homology to the High-Mobility Group (HMG)-1 Protein to Xq25. Genomics 32, 298–300 (1996).
Yang, J. & Everett, A. D. Hepatoma-derived growth factor represses SET and MYND domain containing 1 gene expression through interaction with C-terminal binding protein. J. Mol. Biol. 386, 938–950 (2009).
Everett, A. D., Lobe, D. R., Matsumura, M. E., Nakamura, H. & McNamara, C. A. Hepatoma-derived growth factor stimulates smooth muscle cell growth and is expressed in vascular development. J. Clin. Invest. 105, 567–575 (2000).
Crossin, K. L., Tai, M. H., Krushel, L. A., Mauro, V. P. & Edelman, G. M. Glucocorticoid receptor pathways are involved in the inhibition of astrocyte proliferation. Proc. Natl. Acad. Sci. USA 94, 2687–2692 (1997).
Chen, S. C. et al. Hepatoma-derived growth factor/nucleolin axis as a novel oncogenic pathway in liver carcinogenesis. Oncotarget 6, 16253–16270 (2015).
Uyama, H. et al. Hepatoma-derived growth factor is a novel prognostic factor for patients with pancreatic cancer. Clin. Cancer Res. 12, 6043–6048 (2006).
Lin, Y. W. et al. The expression and prognostic significance of hepatoma-derived growth factor in oral cancer. Oral Oncol. 48, 629–635 (2012).
Iwasaki, T. et al. Hepatoma-derived growth factor as a prognostic marker in completely resected non-small-cell lung cancer. Oncol. Rep. 13, 1075–1080 (2005).
Kung, M.-L. et al. Hepatoma-derived growth factor stimulates podosome rosettes formation in NIH/3T3 cells through the activation of phosphatidylinositol 3-kinase/Akt pathway. Biochem. Biophys. Res. Commun. 425, 169–176 (2012).
Liu, X. et al. AURKA induces EMT by regulating histone modification through Wnt/beta-catenin and PI3K/Akt signaling pathway in gastric cancer. Oncotarget 7, 33152–33164 (2016).
Yang, J. & Everett, A. D. Hepatoma derived growth factor binds DNA through the N-terminal PWWP domain. BMC Mol. Biol. 8, 101–101 (2007).
Dietz, F. et al. The family of hepatoma-derived growth factor proteins: characterization of a new member HRP-4 and classification of its subfamilies. Biochem. J. 366, 491–500 (2002).
Kishima, Y. et al. Hepatoma-derived growth factor stimulates cell growth after translocation to the nucleus by nuclear localization signals. J. Biol. Chem. 277, 10315–10322 (2002).
Wang, C. H. et al. Cell surface heparan sulfates mediate internalization of the PWWP/HATH domain of HDGF via macropinocytosis to fine-tune cell signalling processes involved in fibroblast cell migration. Biochem. J. 433, 127–138 (2011).
Wu, H. et al. Structural and histone binding ability characterizations of human PWWP domains. PloS One 6, e18919 (2011).
Stec, I., Nagl, S. B., van Ommen, G.-J. B. & den Dunnen, J. T. The PWWP domain: a potential protein–protein interaction domain in nuclear proteins influencing differentiation? FEBS Lett. 473, 1–5 (2000).
Laue, K. et al. The multidomain protein Brpf1 binds histones and is required for Hox gene expression and segmental identity. Development 135, 1935 (2008).
Robertson, K. D. et al. The human DNA methyltransferases (DNMTs) 1, 3a and 3b: coordinate mRNA expression in normal tissues and overexpression in tumors. Nucleic Acids Res. 27, 2291–2298 (1999).
Xie, S. et al. Cloning, expression and chromosome locations of the human DNMT3 gene family. Gene 236, 87–95 (1999).
Xu, G. L. et al. Chromosome instability and immunodeficiency syndrome caused by mutations in a DNA methyltransferase gene. Nature 402, 187–191 (1999).
Cherepanov, P. et al. HIV-1 integrase forms stable tetramers and associates with LEDGF/p75 protein in human cells. J. Biol. Chem. 278, 372–381 (2003).
Ciuffi, A. et al. A role for LEDGF/p75 in targeting HIV DNA integration. Nat. Med. 11, 1287–1289 (2005).
Llano, M. et al. An essential role for LEDGF/p75 in HIV integration. Science 314, 461–464 (2006).
Shun, M.-C. et al. LEDGF/p75 functions downstream from preintegration complex formation to effect gene-specific HIV-1 integration. Genes Dev. 21, 1767–1778 (2007).
Rondelet, G., Dal Maso, T., Willems, L. & Wouters, J. Structural basis for recognition of histone H3K36me3 nucleosome by human de novo DNA methyltransferases 3A and 3B. J. Struct. Biol. 194, 357–367 (2016).
Sue, S. C. et al. PWWP module of human hepatoma-derived growth factor forms a domain-swapped dimer with much higher affinity for heparin. J. Mol. Biol. 367, 456–472 (2007).
Sue, S. C., Chen, J. Y., Lee, S. C., Wu, W. G. & Huang, T. H. Solution structure and heparin interaction of human hepatoma-derived growth factor. J. Mol. Biol. 343, 1365–1377 (2004).
Qiu, C., Sawada, K., Zhang, X. & Cheng, X. The PWWP domain of mammalian DNA methyltransferase Dnmt3b defines a new family of DNA-binding folds. Nat. Struct. Biol. 9, 217–224 (2002).
Vezzoli, A. et al. Molecular basis of histone H3K36me3 recognition by the PWWP domain of Brpf1. Nat. Struct. Mol. Biol. 17, 617–619 (2010).
Lukasik, S. M. et al. High resolution structure of the HDGF PWWP domain: A potential DNA binding domain. Protein Sci. 15, 314–323 (2006).
Laguri, C. et al. Human mismatch repair protein MSH6 contains a PWWP domain that targets double stranded DNA. Biochemistry 47, 6199–6207 (2008).
Liu, Y. & Eisenberg, D. 3D domain swapping: As domains continue to swap. Protein Sci. 11, 1285–1299 (2002).
Horibata, S., Vo, T. V., Subramanian, V., Thompson, P. R. & Coonrod, S. A. Utilization of the Soft Agar Colony Formation Assay to Identify Inhibitors of Tumorigenicity in Breast Cancer Cells. J. Vis. Exp. 20, 52727 (2015).
Olins, D. E. & Olins, A. L. Chromatin history: our view from the bridge. Nat. Rev. Mol. Cell. Biol. 4, 809–814 (2003).
Liu, Y., Hart, P. J., Schlunegger, M. P. & Eisenberg, D. The crystal structure of a 3D domain-swapped dimer of RNase A at a 2.1-A resolution. Proc. Natl. Acad. Sci. USA 95, 3437–3442 (1998).
Pop, C. et al. Removal of the pro-domain does not affect the conformation of the procaspase-3 dimer. Biochemistry 40, 14224–14235 (2001).
Eidahl, J. O. et al. Structural basis for high-affinity binding of LEDGF PWWP to mononucleosomes. Nucleic Acids Res. 41, 3924–3936 (2013).
Wang, S. & Fang, W. Increased expression of hepatoma-derived growth factor correlates with poor prognosis in human nasopharyngeal carcinoma. Histopathology 58, 217–224 (2011).
Chen, X. et al. Prognostic value of nuclear hepatoma-derived growth factor (HDGF) localization in patients with breast cancer. Pathol. Res. Pract. 208, 437–443 (2012).
Hu, T.-H. et al. Expression of hepatoma-derived growth factor in hepatocellular carcinoma. Cancer 98, 1444–1456 (2003).
Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 (1997).
Vagin, A. & Teplyakov, A. MOLREP: an Automated Program for Molecular Replacement. J. Appl. Cryst. 30, 1022–1025 (1997).
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta. Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).
Vagin, A. A. et al. REFMAC5 dictionary: organization of prior chemical knowledge and guidelines for its use. Acta. Crystallogr. D Biol. Crystallogr. 60, 2184–2195 (2004).
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta. Crystallogr. D Biol. Crystallogr. 66, 12–21 (2010).
We are indebted to Yuch-Cheng Jean and the supporting staff at beamlines TLS BL13B1, BL13C1, BL15A1 and TPS 05A at the National Synchrotron Radiation Research Center (NSRRC) in Taiwan, Hirofumi Ishii at the Taiwan contracted beamline BL12B2 and Eiki Yamashita at the Beamline BL44XU at SPring-8 in Japan for the technical assistance for technical assistance under the proposal numbers 2014B6600, 2014B6965, 2015B6600, 2015B4004, 2015B4010, 2016A6600, 2016A6659, 2016A4012, 2016B6600, 2016B4000, 2017A4000, 2017A6600 and 2017B6600. We thank KC Han-Ching Wang for valuable discussion. We also thank Sunney I. Chan for the discussion and manuscript editing. Portions of this research were carried out at the NSRRC-NCKU Protein Crystallography Laboratory at University Center for Bioscience and Biotechnology of National Cheng Kung University (NCKU). C.-J. Chen thanks Department of Biochemistry and Molecular Biology, NCKU for joint appointment. This work was supported in part by National Science Council (NSC) grants 101-2628-B-213-001-MY4, 102-2627-M-213-001-MY3, Ministry of Science and Technology (MOST) 105-2311-B-213-001-MY3 and NSRRC grants to C.-J. Chen.
The authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
About this article
Cite this article
Chen, LY., Huang, YC., Huang, ST. et al. Domain swapping and SMYD1 interactions with the PWWP domain of human hepatoma-derived growth factor. Sci Rep 8, 287 (2018). https://doi.org/10.1038/s41598-017-18510-8