Structure of human DPPA3 bound to the UHRF1 PHD finger reveals its functional and structural differences from mouse DPPA3

DNA methylation maintenance is essential for cell fate inheritance. In differentiated cells, this involves orchestrated actions of DNMT1 and UHRF1. In mice, the high-affinity binding of DPPA3 to the UHRF1 PHD finger regulates UHRF1 chromatin dissociation and cytosolic localization, which is required for oocyte maturation and early embryo development. However, the human DPPA3 ortholog functions during these stages remain unclear. Here, we report the structural basis for human DPPA3 binding to the UHRF1 PHD finger. The conserved human DPPA3 85VRT87 motif binds to the acidic surface of UHRF1 PHD finger, whereas mouse DPPA3 binding additionally utilizes two unique α-helices. The binding affinity of human DPPA3 for the UHRF1 PHD finger was weaker than that of mouse DPPA3. Consequently, human DPPA3, unlike mouse DPPA3, failed to inhibit UHRF1 chromatin binding and DNA remethylation in Xenopus egg extracts effectively. Our data provide novel insights into the distinct function and structure of human DPPA3.

In addition to its well-established role in DNA methylation maintenance, UHRF1 has emerged as a factor in oocyte and preimplantation embryo development [21][22][23] .A maternal factor, developmental pluripotencyassociated 3 (DPPA3), also known as Stella/PGC7, has been identified in mice as a strict inhibitor of chromatin binding of UHRF1 and regulation of its cytosolic localization, in cooperation with exportin-1 [24][25][26] .Expression of mouse DPPA3 (mDPPA3), an intrinsically disordered protein, is restricted to primordial germ cells, oocytes, and preimplantation embryos 24,27,28 .mDPPA3 plays an important role in the formation of oocyte-specific DNA methylation patterns by preventing excessive de novo DNA methylation mediated by UHRF1 24 .Using nuclear magnetic resonance (NMR) solution structural analysis of mouse the UHRF1 PHD finger (mPHD) bound to mDPPA3, we recently revealed that the C-terminal region of mDPPA3 binds to mPHD utilizing a VRT motif at residues 88-90 ( 88 VRT 90 ), which is conserved in the motifs of other binding partners, histone H31 ART 3 and PAF15 1 VRT 3 with two subsequent α-helices unique to mDPPA3 29 .Owing to this multifaceted interaction, the binding affinity of mDPPA3 to mPHD (K D of 0.0277 μM) is significant stronger than those of histone H3 and PAF15 (K D of 1.59 μM and 3.52 μM, respectively), indicating that the mechanism by which mDPPA3 inhibits chromatin-binding of UHRF1 involves the competitive binding of between mDPPA3 and histone H3/ PAF15 to UHRF1 29 .The biological functions of mDPPA3 as a demethylation factor and UHRF1-inihibitor in oocyte and preimplantation embryos have been extensively studied in mouse models.A recent report has shown that UHRF1 is enriched in the cytoplasmic lattices of human oocytes 30 .However, it is unclear if the biological function of mDPPA3 is conserved in human DPPA3 (hDPPA3), and its role in human oocytes and preimplantation embryos is unknown.Two α-helices in mDPPA3 which are induced upon binding to mPHD has been shown to be required for the interaction with mUHRF1 29 .However, the amino acid sequences corresponding to these helices are poorly conserved between human and mouse DPPA3 (Fig. 1a), which raises a question of whether hDPPA3 also binds to the hUHRF1 PHD finger in a manner similar to their mouse counterparts, and whether hDPPA3 can inhibit chromatin binding of UHRF1.
In this study, we determined the crystal structure of the human UHRF1 PHD finger complexed with the C-terminal hDPPA3 fragment.The structure clearly showed that the binding mode of hDPPA3 to the human UHRF1 PHD finger differs markedly from that of the mouse proteins and explains why hDPPA3 binds to the human UHRF1 PHD finger with low binding affinity, comparable to the binding of histone H3 and PAF15.Biochemical assays using Xenopus egg extracts demonstrated that the inhibitory effect of hDPPA3 on chromatin-binding of UHRF1 is relatively modest compared to the strong inhibition by mouse DPPA3.Our findings shed light on the unexpected role of hDPPA3 in epigenetic regulation during early embryonic development, which differs from the evidence in mice.

Results
Interaction between hDPPA3 and hUHRF1 PHD finger Our previous NMR structural analysis of mDPPA3 complexed with mUHRF1 PHD (mPHD) revealed that residues 85-118 of mDPPA3 are essential for its interaction with mPHD (Figs.1a and 2b) 29 .Thus, we identified the corresponding region of hDPPA3 by sequence alignment (residues 81-118: hDPPA3 81-118 ) (Fig. 1a, b), and evaluated whether this region binds to the human UHRF1 PHD finger, residues 299-366 (hPHD).Isothermal titration calorimetry (ITC) demonstrated that hDPPA3 81-118 could bind to hPHD with a K d of 0.868 μM (Fig. 1c and Supplementary Data 1), which is approximately 30-fold weaker than the binding affinity between mDPPA3 and mPHD (K d = 0.0277 μM) 29 .The binding affinity of hDPPA3 81-118 to hPHD is comparable with the previously reported binding affinity between hPHD and the histone H3 N-terminal tail (residues 1-15; K D = 1.7 μM) or PAF15 (residues 1-10; K D = 2.2 μM) 17,31 .To further investigate the interactions at an atomic resolution, we performed NMR titration experiments.We successfully assigned 1 H- 15 N heteronuclear single quantum coherence (HSQC) spectra for [ 15 N]-hPHD in the free and complex states with non-labeled hDPPA3 81-118 (Supplementary Fig. 1a).The 1 H-15 N HSQC spectra of [ 15 N]-hPHD titrated with non-labeled hDPPA3 81-118 showed that the HSQC signals shifted in the intermediate exchange regime on a chemical shift timescale, supporting the modestly weak interaction between hDPPA3 81-118 and hPHD (Fig. 1d).These data indicate that the binding of hDPPA3 to hPHD was not significantly stronger than that of the other binding partners, histone H3 and PAF15.Chemical shift differences (CSD) between the free and complexed states showed relatively large values for Asp330, Met332, Asp337, Glu355, and Asp356, suggesting the contribution of the main chain of these amino acid residues to the hPHD-hDPPA3 interaction (Supplementary Fig. 1b).
In contrast to the hPHD moiety, the binding mode of hDPPA3 shares both similarities and dissimilarities with that of mDPPA3 (Fig. 2a, b).The conserved VRT motif at residues 85-87 of hDPPA3 81-118 is accommodated on the acidic surface of hPHD, the binding site for 1 ART 3 of histone H3, and 1 VRT 3 of PAF15, in a manner concordant with the motif in mDPPA3 ( 88 VRT 90 ) (Fig. 2b).The side chain of Val85 in hDPPA3 81-118 forms a hydrophobic interaction with Leu331, Val352, Pro353, and Trp358 in hPHD (Fig. 2a).The positively charged guanidino group at Arg86 of hDPPA3 81-118 forms hydrogen bonds with the side chains of Asp334 and Asp337 of hPHD (Fig. 2a, c).The side chain methyl and hydroxyl groups of Thr87 in hDPPA3 81-118 forms hydrophobic interactions with Leu331 and Val352 of hPHD and hydrogen bonds with the main chain amide of Ser90 of hDPPA3 (Fig. 2a).The latter potentially functions as a helical cap for the N-terminus of the following α-helix (Fig. 2a).Leu88 of hDPPA3 81-118 is surrounded by the side chains of Ala317, Gln330, Met332, and Ala339 in hPHD, in which the side chain of Met332 functions as a separation between the side chains of Arg86 and Leu88 of hDPPA3 81-118 (Fig. 2a).
When mPHD binds to mDPPA3, the two α-helices following the VRT motif of mDPPA3 form an L-like shape, in which the long α-helix binds to the shallow groove between the pre-and core-PHD fingers (Fig. 2b).However, the C-terminus of the 85 VRT 87 motif of hDPPA3 81-118 forms a unique conformation that differed from that of mDPPA3.Residues 88-101 of hDPPA3 forms a four-turn single α-helix, which is not kinked and markedly differs from mDPPA3 complexed with mPHD (Fig. 2a, b).The contact area between the hPHD and hDPPA3 (ca.449 Å 2 ) was smaller than that of the mouse protein (ca.1360 Å 2 ) 32 , which is concordant with the weaker dissociation constant of the human proteins.
Structural feature of hPHD:hDPPA3 in solution Intriguingly, the α-helix of hDPPA3 81-118 has no contact with the hPHD moiety in the crystal (Fig. 2a, c).Instead, the α-helix interacts with the corresponding part of a symmetry molecule related to a crystallographic two-fold axis (Supplementary Fig. 2c).This interaction in the crystal gives rise to two possibilities: the helical structure formation of hDPPA3 is an artifact of crystal packing, or the hPHD:hDPPA3 complex forms a dimer structure via the interaction mediating the α-helix of hDPPA3.
Next, we examined the structure of hDPPA3 81-118 in solution using circular dichroism (CD) and size-exclusion chromatography in line with small angle X-ray scattering (SEC-SAXS) which can analyze the solution structure, oligomeric state, conformational changes and flexibility of biomacromolecules at a scale ranging from a few Å to hundreds of nm (Supplementary Fig. 3a-c and Supplementary Table 1) 33 .The CD spectrum exhibited that hDPPA3 81-118 alone showed a typical random-coil spectrum (Fig. 3a and Supplementary Data 2).The CD spectrum of hDPPA3 81-118 mixed with hPHD showed a negative peak at 222 nm, which was lower than the sum of the spectra of hPHD and hDPPA3 81-118 alone (Fig. 3a and Supplementary Data 2), indicating that the binding of hDPPA3 to hPHD involved a coupled folding and binding mechanism.The SEC-SAXS data also supported the coupled folding and binding mode of hDPPA3.The dimensionless Kratky plot showed the unfolding state of sole hDPPA3 81-118 , whereas the hPHD:hDPPA3 81-118 complex was in a globular state (Fig. 3b and Supplementary Data 3).
SEC-SAXS experiments also revealed that the molecular mass of the measured proteins was estimated by the empirical volume of correlation Vc 34 , resulting in a 13.0 kDa hPHD:hDDPA3 81-118 complex, which was highly similar to the molecular weight calculated from the amino acid sequence of the hPHD:hDPPA3 81-118 complex with 1:1 stoichiometry (12.2 kDa) (Supplementary Table 1).The ab initio model of the measured proteins showed clear results.The overall shape of the bead model was well superimposed on the crystal structure of the hPHD:hDPPA3 81-118 complex in the asymmetric unit (Fig. 3c, d and Supplementary Data 3).These data indicated that hDPPA3 binds to hPHD at 1:1 stoichiometry with the induction of a four-turn single α-helix.

Validation of the structural data by mutational analysis
To validate our structural data and confirm the contribution of individual residues to complex formation, ITC experiments were conducted using hPHD and hDPPA3 harboring mutations in the VRT sequence.Mutations with deleterious effects on the interaction were R86A and T87A of hDPPA3, which reduced the dissociation constant to 20.0 and 16.5 µM, respectively, and a double mutation (R86A/T87A), which resulted in a more severe reduction in the interaction, with a K D exceeding 85.0 µM (Fig. 1c and Supplementary Data 1).In contrast, alanine mutations at Val85 and Leu88 in hDPPA3 had less marked effects on the hPHD:hDPPA3 interaction (Fig. 1c and Supplementary Data 1).
We further investigated mutants of DPPA3 that affect dimer formation as observed in the crystal structure (Supplementary Fig. 4a and Supplementary Data 1).R98A/M102A, located at C-terminal region in the α-helix of hDPPA3 and potentially interacting with hPHD of the symmetrical molecule in the crystal, did not reduced the binding affinity.Similarly, M96A/L99A, which contribute to the formation of the helix bundle of hDPPA3 in the crystal, also had no effect on the interaction with hPHD, validating the 1:1 stoichiometry of the hPHD:hDPPA3 complex in solution.Interestingly, the introduction of proline residue, known as a helix breaker, at both Arg93 and Ala97 in hDPPA3 (R93P/A97P) significantly reduced the binding affinity to hPHD, with K D of 9.39 µM (Supplementary Fig. 4a and Supplementary Data 1), indicating that helical structural formation following the VRT motif in hDPPA3 is crucial for its interaction with hPHD.
Next, mutations were introduced into hPHD.Concordant with the hDPPA3 mutants, the D334A/D337A mutations in hPHD, which form an ionic-pair with Arg86 of hDPPA3, had a severe effect, reducing the binding affinity to a K D exceeding 115 µM.The M332A hPHD mutation showed a decreased binding affinity, with a K D of 8.07 µM (Supplementary Fig. 4a and Supplementary Data 1).ITC data based on mutant proteins indicate that the VRT motif of hDPPA3 is important for its interaction with the UHRF1 hPHD finger.

Effect of hDPPA3 on UHRF1 function
Next, to analyze whether hDPPA3 affects the biological functions of UHRF1, ubiquitination of histone H3, and chromatin binding, we performed NMR titration assays and in vitro biochemical experiments using recombinant proteins and Xenopus egg extracts.First, we examined the competitive binding of hDPPA3 and histone H3 to hPHD because both hDPPA3 and histone H3 mainly bind to the acidic surface of hPHD via the 85 VRT 87 and 1 ART 3 motifs, respectively.We conducted NMR titration experiments using 1 H- 15 N labeled hPHD and non-labeled hDPPA3 81-118 and/or histone H3 peptides (residues 1-37W; the H3 1-37W peptide).The HSQC spectrum of hPHD mixed with hDPPA3 81-118 and the H3 1-37W peptide (hPHD:hDPPA3:H3 = 1:2:2) showed most of the signals, with weakened or no intensity by the broadening due to chemical exchange, suggesting that, as expected, hDPPA3 81-118 and the H3 1-37W peptide competitively bound to the acidic surface of hPHD as the shared binding site (Fig. 4a, upper).In the presence of excess H3 1-37W peptide (hPHD:hDPPA3 81-118 :H3 = 1:2:8), hDPPA3 81-118 could not bind to hPHD (Fig. 4a, lower).This differed from the situation with mDPPA3, which bound to mPHD even in the presence of excess H3 1-37W peptide (mPHD:mDPPA3:H3 = 1:2:8) 29 .An in vitro ubiquitination assay of C-terminal FLAG-tagged H3 1-37W with full-length human UHRF1 also supported the weak inhibitory effect of hDPPA3.hDPPA3 did not effectively inhibit ubiquitination of the histone H3 tail, whereas mDPPA3 showed a markedly negative effect on ubiquitination (Supplementary Fig. 4b and Supplementary Data 4).The addition of hDPPA3 to a 1-2 equimolar excess of histone H3 only slightly inhibited histone H3 ubiquitination (Fig. 4b and Supplementary Data 4).Mutant forms of hDPPA3, which exhibited decreased binding to hPHD, failed to inhibit ubiquitination of histone H3 (Fig. 4b and Supplementary Data 4).These findings indicate that the binding of hDPPA3 81-118 to UHRF1 inhibits the ubiquitination activity of UHRF1 on histone H3; however, the inhibitory effect was moderately weak due to the low binding affinity between hDPPA3 81-118 and UHRF1.
Finally, we tested the ability of hDPPA3 to inhibit UHRF1 chromatin binding in Xenopus egg extracts (Fig. 5a).As previously reported, the addition of 0.5 µM recombinant mDPPA3 to interphase extracts was sufficient to block UHRF1 chromatin loading, UHRF1-dependent PAF15 ubiquitylation, and DNMT1 recruitment (Fig. 5b and Supplementary Data 4).In contrast, hDPPA3 did not inhibit the chromatin binding of UHRF1 and DNMT1 recruitment, even at 1.0 μM (Fig. 5b and Supplementary Data 4).Consistently, hDPPA3 did not show significant inhibitory activity on DNA methylation in Xenopus egg extracts compared to mDPPA3 (Fig. 5c).
Taken together, the binding of hDPPA3 to hUHRF1 PHD competes with that of histone H3.However, it is noteworthy that the inhibitory effect exerted by hDPPA3 was relatively modest, implying that hDPPA3 does not appear to function as a strong inhibitor of UHRF1 chromatin binding, unlike mouse DPPA3.

Discussion
Our structural analysis revealed that human DPPA3 binds to hPHD solely through a conserved VRT sequence motif.This finding is consistent with biochemical data showing that the binding affinity of hDPPA3 to hPHD was in the sub-micro order range of K D , with an approximately 30-fold decrease in the binding affinity of its mouse protein counterpart.The weak binding affinity of hDPPA3 was insufficient to inhibit the chromatin binding of UHRF1 in Xenopus egg extracts.Our data suggested that the inhibitory effect of hDPPA3 differs from that of mDPPA3 under similar conditions.This raises the question of whether hDPPA3 can act as an inhibitor of UHRF1 in human oocytes and early embryogenesis.There are several possibilities to consider in this regard.Intrinsically disordered protein (region) containing low complexity sequence frequently associates with formation of liquid-liquid phase separation (LLPS) 35 .Notably, sequence analysis of human and mouse DPPA3 using FuzDrop (https://fuzdrop.bio.unipd.it)indicated that human DPPA3 exhibits a higher potential for droplet formation that mouse DPPA3 (Supplementary Fig. 5) 36 .This prediction suggests that condensed hDPPA3 within the droplet may preferentially bind to UHRF1, thereby inhibiting the chromatin binding of UHRF1.In another situation, the level of hDPPA3 protein expression in human oocytes and zygotes is key to the inhibition of UHRF1 chromatin binding.Our data indicated that the binding affinity of hDPPA3 for hPHD was approximately 1.7-fold stronger than that of histone H3, suggesting that a locally high concentration of hDPPA3 contributes to its preferential binding to hUHRF1 to inhibit chromatin binding.Another possibility involves post-translational modifications of histone H3.Given that the methylation of Arg2 and phosphorylation of Thr3 in histone H3 greatly impair its binding to the UHRF1 PHD finger 31 , hDPPA3 might bind to UHRF1 even at low protein concentrations.Recently, NLRP14 (Nucleotidebinding oligomerization domain, leucine-rich Repeat and Pyrin domain containing) has emerged as a factor related to reproduction.It interacts with UHRF1 in the zygote and two-cell stages in the cytosol 21,37 .If cytoplasmic localization of UHRF1 is not mediated by hDPPA3, it may be important for UHRF1 to interact with NLRP14 immediately after its translation into the cytoplasm.Interestingly, the cytosolic localization of the mRNA of a guanine nucleotide exchange factor, NET1, has been reported to regulate protein-protein interactions after translation, ultimately determining protein localization 38 .
The VRT motif in DPPA3, which binds to the acidic surface of the UHRF1 PHD finger, is well conserved across various species (Fig. 1a).Conversely, the amino acid sequence corresponding to the α-helix following the VRT motif showed significant diversity.Interestingly, AlphaFold2 (AF2) structural prediction indicated that mDPPA3 has both short and long α- helices following the VRT motif, forming an L-like shape, consistent with our NMR structure of the mPHD:mDPPA3 complex (Supplementary Fig. 6) 39 .In contrast, human DPPA3 exhibits a single long α-helix at the same position.AF2 predictions also suggest that Homo sapience (UniProt ID: Q6W0C5), Bos taurus (A9Q1J7), Gorilla gorilla gorilla (G3RB81), Saimiri boliviensis (A0A2K6SNG1), Puma concolor (A0A6P6HCW6), Nomascus leucogenys (A0A2I3H008), Crocuta crocuta (A0A6G1B388), Physeter macrocephalus (A0A2Y9EH83), and Acinonyx jubatus (A0A6I9ZFC3) possess a single α-helix, while Mus musculus (Q8QZY3), Rattus norvegicus (Q6IMK0), and Cricetulus griseus (A0A3L7H856) have two α-helices, consisting of both short and long α-helices, as far as we could find in the database (Supplementary Fig. 6).These observations suggest the potential limitation of the two α-helices to Rodentia and underscore the utility of AF2 structural prediction for the classification of the DPPA3 function based on the helical content.The major difference in the helical region of human and mouse DPPA3 is the substitution of a proline residue with a lysine residue at the 95th position of human DPPA3 (Fig. 1a).A similar substitution is also found in the species that predictably forms as single α-helix.However, the K95P mutation in human DPPA3 did not enhance its binding affinity for hPHD (Supplementary Fig. 7a).AF2 prediction of the K95P mutant of hDPPA3suggested that a single α-helix remains the predominant conformation (Supplementary Fig. 7b), suggesting that the differences in the helical structural regions of human and mouse DPPA3 are governed by more complicated mechanisms than a simple amino acid substitution.
The distinctive α-helical arrangement in hDPPA3 revealed in our structural analysis shed light on the function of this protein in oocytes and preimplantation embryo development distinct from the mouse DPPA3.Our results encourage further investigations into the functional implications of hDPPA3, potentially paving the way for novel discoveries in this context.
For the preparation of 15 N-labeled or 13 C, 15 N-double labeled hPHD, M9 minimal media containing 0.5 g/l 15 NH 4 Cl or 0.5 g/l 15 NH 4 Cl and 1 g/l 13 C-glucose was used instead of LB media.Site-directed mutagenesis of hPHD and hDPPA3 81-118 was performed by designing two primers containing the mutations.The mutants of hDPPA3 81-118 and the labeled hPHD were purified using the same protocol.The mutants of hDPPA3 81-118 and the labeled hPHD were purified by the same protocol.

Crystallography of hPHD in complex with hDPPA3 peptide
The hPHD:hDPPA3 81-118 complex was prepared by adding an equi-molar excess of hDPPA3 peptide to hPHD prior to crystallization.The crystal was obtained using an 8 mg/ml concentration of the complex at 4 °C and the hanging drop vapor diffusion method with a reservoir solution containing 100 mM Tris-HCl (pH 8.5) and 2 M Ammonium sulfate.The crystal was directly frozen in liquid nitrogen using a cryoprotectant containing 25% (v/ v) ethylene glycol.The X-ray diffraction data were collected at a wavelength of 0.98000 Å on a Pilatus3 6 M detector in beam line BL-17A at Photon Factory (Tsukuba, Japan) and scaled at 2.40 Å resolution using the program XDS package 40 and Aimless 41 .After molecular replacement by PHASER 42 using human PHD finger (PDB: 3ASL) as a search model and several cycles of model refinement by PHENIX 43 , the final model converged at 2.40 Å resolution with a crystallographic R-factor of 23.3% and a free R-factor of 26.6%.
The crystallographic data and refinement statistics are given in Table 1.Figures were generated using PyMol (http://www.pymol.org).

NMR
All NMR experiments were performed using a Bruker BioSpin Avance III HD spectrometers with TCI triple-resonance cryogenic probe-heads and basic 1 H resonance frequency of 600.03 and 800.23 MHz.Threedimensional (3D) spectra for backbone signal assignments, including HNCACB, CACB(CO)NH, HNCA, HN(CO)CA, HNCO, and HN(CA) CO, were acquired at 293 K for 520 µM [ 13 C, 15 N]-hPHD dissolved in PBS buffer (pH 7.0) containing 1 mM DTT and 5% D 2 O.For the complex state, 260 µM [ 13 C, 15 N]-hPHD with hDPPA3 811-118 at molar ratio of 1:2 was used in the buffer same as the free state.The spectral widths (total number of data points) of each spectrum were 18 ppm (2048) for the 1 H dimension and 24 ppm (192) for the 15 N dimension.All 3D spectra were acquired using nonuniform sampling (NUS) to randomly reduce the t 1 and t 2 time-domain data points by 25%.The uniformly sampled data were reconstructed from the raw NMR data using various techniques such as IST or SMILE 44,45 .All NMR spectra were processed using NMRPipe 46 .For NMR analysis, an integrated package of NMR tools named MagRO-NMRViewJ, version 2.01.41 47 , on NMRView was used 48 .

ITC measurements
Microcal PEAQ-ITC (Malvern) was used for ITC measurements.Wild-type and mutants of hPHD and hDPPA3 were dissolved in ITC buffer (10 mM HEPES [pH 7.5] buffer containing 150 mM NaCl and 0.25 mM TCEP).All measurements were carried out at 293 K.The data were analyzed with Microcal PEAQ-ITC analysis software using a one-site model.For each interaction, at least three independent titration experiments were performed to show the dissociation constants with mean standard deviation.

CD
Far-UV circular dichroism (CD) spectra were obtained using a JASCO J-1100 model spectrometer.All samples were prepared at a concentration of 20 µM, dissolved in 10 mM HEPES [pH7.5]buffer containing 150 mM NaCl, 0.25 mM TCEP.The measurements were performed at 293 K with a path length of 1 mm.

SEC-SAXS
SAXS data were collected on Photon Factory BL-10C using an HPLC Nexera/Prominence-I (Shimazu) integrated SAXS set-up 49 .50 µl of 12 mg/ ml hPHD and hPHD:hDPPA3 81-118 complex and 20 mg/ml hDPPA3 81-118 were loaded onto a Superdex® 200 Increase 5/150 GL (Cytiva) preequilibrated with 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM DTT, 10 µM zinc acetate and 5% glycerol at a flow rate of 0.25 ml/min at 20 °C.The flow rate was reduced to 0.025 ml/min at an elution volume of 1.9-2.8ml.X-ray scattering data were collected every 20 s on a PILATUS3 2 M detector over an angular range of q min = 0.00690 Å −1 to q max = 0.27815 Å −1 .The UV spectra at the range of 200-450 nm were recorded every 10 s.Circular averaging and buffer subtraction were carried out using the program SAngler 50 to obtain one-dimensional scattering data I(q) as a function of q (q = 4πsinθ/λ, where 2θ is the scattering angle and λ is the X-ray wavelength 1.5 Å).The scattering intensity was normalized on an absolute scale using the scattering intensity of water 41 .The multiple concentrations of the scattering data around the peak at A280, namely the ascending and descending parts of the chromatography peak, and I(0) were extrapolated to zero-concentration using MOLASS 51 .The molecular mass of the measured proteins was estimated using the empirical volume of correlation, V c , showing no aggregation of the measured sample 34 .The radius of gyration R g and forward scattering intensity I(0) were estimated from the Guinier plot of I(q) in the smaller-angle region of qR g < 1.3.The distance distribution function, P(r), was calculated using the program GNOM 52 .The maximum particle dimension D max was estimated from the P(r) function as the distance r for which P(r) = 0.The scattering profile of the crystal structure of hPHD:hDPPA3 81-118 was computed using CRYSOL 53 software.Ab initio model of hPHD:hDPPA3 81-118 was created using GASBOR and DAMAVER 54,55 .

Statistics and reproducibility
All biochemical and biophysical experiments were repeated at least three times.

Fig. 1 |Fig. 2 |
Fig. 1 | Characterization of the interaction between hUHRF1 and hDPPA3.a Amino acid sequence alignment of C-terminal part of DPPA3.Secondary structures of mouse and human DPPA3 are indicated based on PDB:7XGA and analysis of this study, respectively.b Schematic of the domain composition of human UHRF1 and DPPA3.c Isothermal titration calorimetry measurements for hPHD and wildtype (WT)/mutants of hDPPA3 81-118 .Superimposition of enthalpy change plots with standard deviations.Data were presented as mean values for n = 3. d Overlay of

Fig. 3 |
Fig. 3 | Solution structure of hDPPA3.a CD spectra of hPHD alone (red), hDPPA3 81-118 alone (blue), and the hPHD in complex with hDPPA3 81-118 (black).The sum of CD spectra of hPHD alone and hPDDA3 81-118 alone is shown as gray.b Dimensionless Kratky plots of hPHD alone (red diamond), hDPPA3 81-118 alone (blue square), and hPHD in complex with hDPPA3 81-118 (black circle) derived from small-angle X-ray scattering (SAXS) data.c Comparison of scattering curve derived from experimental data (cyan) and theoretical curve of the crystal structure of the hPHD:hDPPA3 81-118 complex (red).d Structural comparison of solution and crystal structures of the hPHD:hDPPA3 81-118 complex.Ab initio bead model of the hPHD:hDPPA3 81-118 complex derived from the SAXS scattering data (transparent gray sphere) is superimposed on the crystal structure (cartoon).

Fig. 5 |
Fig. 5 | Functional assay of DPPA3 using Xenopus egg extracts.a Experimental design for functional analysis of DPPA3 using Xenopus egg extracts.b Sperm chromatin was incubated with interphase Xenopus egg extracts supplemented with buffer (+buffer), 3×FLAG-mDPPA3, or 3×FLAG-hDPPA3.Chromatin fractions were isolated and immunoblotted using the indicated antibodies.The gel image is representative of n = 3 independent experiments.c Sperm chromatin was added to interphase egg extracts supplemented with radiolabeled S-[methyl-3 H]-adenosyl-Lmethionine and buffer (control), 3×FLAG-mDPPA3, or 3×FLAG-hDPPA3.The efficiency of DNA methylation maintenance was assessed by the incorporation of radio-labeled methyl groups from S-[methyl-3 H]-adenosyl-L-methionine ( 3 H-SAM) into DNA purified from the egg extracts.Data were presented as mean values ± SD for n = 3.

Table 1 |
Data collection and refinement statistics a () Values in parentheses are for the highest-resolution shell.