Accession codes

The accession codes of HBS1L proteins for sequence alignments, obtained from UniProt and/or NCBI, were as follows: Mus musculus (Q69ZS7); Homo sapiens (Q9Y450); Gallus gallus (E1BW59); Chrysemys picta bellii (A0A8C3HXA3); Xenopus laevis (Q6GNS6); Danio rerio (F1Q5Z6); Drosophila melanogaster (Q9W074); Gordionus sp. (XP_065321722.1); Schizosaccharomyces pombe (O74774); Mycena sanguinolenta (KAF7355041.1); S. cerevisiae (P32769); C. elegans (P90922/NP_001021556.2).

Sample preparation

DNA encoding residues 51–120 of mouse HBS1L (M. musculus; Q69ZS7) was subcloned from a full-length mouse cDNA clone by polymerase chain reaction. The DNA fragment was cloned into the expression vector pCR2.1 (Invitrogen) to produce a fusion protein containing an N-terminal His affinity tag, a tobacco etch virus (TEV) protease cleavage site, and a GSSGSSG linker sequence preceding the HBS1L sequence. For ubiquitin, the encoded DNA was cloned into the pCR2.1 vector and expressed as a fusion protein with the same N-terminal sequence. Consequently, the expressed protein containing UBAh used in this study included the artificial tag-derived sequences GSSGSSG at the N-terminus and SGPSSG at the C-terminus, whereas the ubiquitin protein contained only the GSSGSSG sequence at its N-terminus.

¹³C/¹⁵N-labeled fusion proteins were synthesized using a cell-free protein expression system [35,36]. The lysate was clarified by centrifugation at 16,000 × g for 20 min and filtered through a 0.45-µm membrane (Merck Millipore). The clarified lysate was applied to a His-Trap column (Cytiva) and eluted using a 12–500 mM imidazole gradient. The affinity tag was then removed by incubation with TEV protease for 1 h at 30°C. The tag-free proteins were further purified using Superdex 75 gel filtration chromatography (Cytiva). The purified HBS1L and ubiquitin samples were each concentrated to approximately 1.0 mM in 20 mM sodium phosphate buffer (pH 6.0), containing 100 mM NaCl, 1 mM 1,4-dl-dithiothreitol-d₁₀, and 0.02% NaN₃ (in 90% H₂O/10% D₂O) using an Amicon Ultra-15 filter (3,000 MWCO; Merck Millipore). For NMR measurements of the complex structure, HBS1L and ubiquitin solutions were mixed at a molar ratio of 1:1 and concentrated to approximately 1.0 mM both for each protein.

NMR spectroscopy and structure calculations

All NMR data were acquired at 25°C on Bruker 600 MHz and Bruker 800 MHz spectrometers and processed using NMRPipe [37]. Processed data were analyzed using NMRView [38] and KUJIRA [39]. Two-dimensional (2D) ¹H–¹³C and ¹H–¹⁵N HSQC spectra, as well as three-dimensional (3D) HNCO, HN(CA)CO, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, HBHA(CO)NH, H(CCCO)NH, (H)CC(CO)NH, HCCH-TOCSY, HCCH-COSY, and CCH-TOCSY spectra acquired on a Bruker 600 MHz spectrometer were used to assign all carbon, nitrogen, and hydrogen atoms of the ¹³C/¹⁵N-labeled samples [40,41].

Furthermore, 2D ¹H–¹³C and ¹H–¹⁵N HSQC spectra, and 3D ¹⁵N- and ¹³C-edited NOESY spectra acquired on a Bruker 800 MHz spectrometer were used for structural calculations. The proton shifts were indirectly referenced using the ¹H resonance frequency of the methyl group of DSS (the chemical shift reference of ¹H was based on the proton of water (4.821 ppm at 298 K)), and the ¹³C and ¹⁵N resonances were indirectly calibrated based on their gyromagnetic ratios, according to IUPAC recommendations [42]. The nuclear Overhauser effect (NOE) peaks from the ¹⁵N- and ¹³C-edited NOESY spectra with a mixing time of 80 ms were converted into distance constraints for structure calculations. The 3D structure of the protein complex was determined by combining automated NOESY cross-peak assignment and structural calculations with torsion angle dynamics [43] implemented in CYANA 2.1 [44]. The dihedral angle restraints for ϕ and ψ were obtained from the main-chain and ¹³C^β chemical shift values using the TALOS program [45] and by analyzing the NOESY spectra. Stereospecific assignments for the isopropyl methyl and methylene groups were determined based on the patterns of the inter- and intra-residual NOE intensities [46]. Hydrogen-bond restraints were not used in this study. The structure calculation started with 400 randomized conformers using a standard CYANA-simulated annealing schedule with 40,000 dynamic torsion angle steps per conformer [47]. Further refinements by restrained molecular dynamics simulations, followed by restrained energy minimization, were performed for the 80 conformers with the lowest final CYANA target function values using the Amber24 program with an ff19SB force field and generalized Born model [48], as previously described [49]. Finally, 20 conformers with the lowest Amber energy values were selected as the final structures for deposition in the Worldwide Protein Data Bank. PROCHECK-NMR [50] and MOLMOL [51] were used to validate and visualize the deposited structures.

Images of ribosome-containing structures were generated using PyMOL (PyMOL Molecular Graphics System, version 2.0, Schrödinger, LLC). Other structures were generated using MOLMOL, unless otherwise stated.

The atomic coordinates for the ensemble of 20 energy-refined NMR conformers representing the solution structures of free UBAh and the UBAh–ubiquitin complex were deposited in the Worldwide Protein Data Bank under the accession codes 9WPR and 9WPS, respectively. Chemical shift assignments for free UBAh and the UBAh–ubiquitin complex were deposited in the BMRB database under accession numbers 36786 and 36787, respectively.

Chemical shift perturbation experiments

¹⁵N-labeled UBAh (0.1 mM) was dissolved in NMR buffer as described above, and unlabeled ubiquitin was added stepwise to a final molar ratio of 1:3. At each step, a 2D ¹H-¹⁵N HSQC spectrum was acquired. The weighted chemical shift change in the amide ¹H and ¹⁵N for each residue, Δδ(¹⁵N + ¹H_N), was calculated as [(Δδ_15N/6.5)² + Δδ_1H²] ^1/2, where Δδ_15N and Δδ_1H represent the observed chemical shift changes of the amide nitrogen and proton, respectively. The dissociation constant (K_d) for UBAh–ubiquitin binding was determined by fitting the chemical shift perturbation data to the following equation [52]:

where Δδ_obs is the weighted chemical shift change from the free state; x is the molar ratio of UBAh to ubiquitin; P₀ is the initial concentration of the labeled protein (UBAh); and Δδ_max is the maximum chemical shift change. Global nonlinear regression analysis, assuming a single binding-site model, was performed using GraphPad Prism 6 based on titration data from three residues.

Determination of the solution structure of the UBAh–ubiquitin complex

The structure of UBAh was determined using residues 51–120 of mouse HBS1L, whereas that of ubiquitin was determined using its full-length form (79 residues) with an N-terminal seven-residue tag. Mouse UBAh and ubiquitin share 84% and 100% sequence identity with their human counterparts, respectively. Each protein was overexpressed in Escherichia coli and labeled with ¹⁵N and/or ¹³C or left unlabeled, depending on the type of experiment.

The solution structure of the UBAh–ubiquitin complex was determined at a 1:1 molar ratio of the two proteins using standard multidimensional heteronuclear NMR spectroscopy. The completeness of the backbone and side-chain resonance assignments is summarized in S1 Table. The assigned 2D ¹H–¹⁵N HSQC spectrum of the tagged UBAh protein in the presence of ubiquitin is shown in S2 Fig. Tertiary structures were calculated using the CYANA software package [44] based on 3,558 ¹H–¹H distance restraints derived from NOESY spectra, 184 backbone torsion angle restraints, and nine χ-dihedral angle restraints. Representative intermolecular NOE cross-peaks are shown as strip plots in S3 Fig, and the intermolecular NOEs used for the structural calculations are summarized in S4A Fig. Of the 100 independently calculated structures obtained in the final cycle, 20 conformers with the lowest CYANA target function values were further refined by restrained energy minimization and subsequently used for all downstream analyses. The statistics of the structures and distance and torsion angle restraints used in the CYANA program are summarized in Table 1. The root-mean-square deviation (RMSD) from the mean structure was 0.65 ± 0.20 Å for the backbone atoms (N, C^α, C′) and 1.08 ± 0.17 Å for all heavy (non-proton) atoms in the well-ordered region, comprising residues 70–118 of UBAh and residues 1–71 of ubiquitin. For the well-ordered regions of the individual domains, the RMSD values were 0.39 ± 0.12 Å for the backbone and 1.32 ± 0.18 Å for all heavy atoms in UBAh, and 0.20 ± 0.05 Å and 0.77 ± 0.08 Å, respectively, in ubiquitin. Highly disordered regions were observed in the following regions: the N-terminal tag (GSSGSSG), the adjacent segment (Glu51–His67), the C-terminal segment (Asp119–Gly120), and the C-terminal tag (SGPSSG).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Summary of conformational constraints and structural statistics for 20 energy-refined conformers of the complex between UBAh of HBS1L and ubiquitin.

https://doi.org/10.1371/journal.pone.0348877.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Structure of the UBAh–ubiquitin complex.

(A) Left: C^α traces of the backbone atoms for the 20 superimposed lowest-energy conformers of the complex. Residues 68–119 of UBAh (green) and residues 1–72 of ubiquitin (gray) are shown. Right: ribbon representation of UBAh in complex with ubiquitin in the same orientation as in the left panel. (B) Characteristic structures of the α1/α2 (left) and α2/α3 (right) loops. In this sequence, the types of β-turns and hydrogen bond connections (blue lines above the sequences) are shown. Side chains of the loop residues are displayed in blue in the ribbon representations. In the stick representations, the red dotted lines indicate hydrogen bonds within the β-turns.

https://doi.org/10.1371/journal.pone.0348877.g003

For the free form of UBAh, the corresponding data together with the structure are presented in S2 Table and S4C Fig. A comparison of the free and ubiquitin-bound forms of UBAh showed that significant structural changes upon ubiquitin binding occur in the α1/α2 loop and at the C-terminal ends of α1 and α3, as described below (S4D Fig). Hereafter, only the UBAh structure of this complex is shown.

Finally, with regard to the ubiquitin structure bearing the N-terminal seven-residue tag, the NMR results indicate that it adopts the canonical ubiquitin fold, also known as the β-grasp fold, consisting of one α-helix and five β-strands (αA: Ile23–Glu34; β1: Met1–Lys6; β2: Thr12–Val17; β3: Gln41–Phe45; β4: Lys48–Gln49; β5: Thr66–Leu71) (S4E Fig). This structure was virtually identical to that of untagged free ubiquitin. N-terminally tagged ubiquitin has been reported to function properly in structure–function studies [53] and in the determination of complex structures [54]. However, as reported by Nelson et al. [53], slight chemical shift differences between tagged and untagged ubiquitin were confined to a limited region near the N-terminus (Met1), specifically within part of the β-sheet and on one side of αA, suggesting minor structural changes in this region. Because this region lies on the face opposite the Ile44 hydrophobic patch, the seven-residue tag is unlikely to substantially affect the UBAh–ubiquitin interaction mode.

Structural features of the UBAh domain: conservation across eukaryotes

The NMR results show that the UBAh structure consists of three tightly packed α-helices (α1: 71–88; α2: 94–103; and α3: 108–118), forming a three-helix bundle, with the connecting α1/α2 and α2/α3 loops being well ordered (Fig 3A). Within the UBAh structure, an extensive hydrophobic interaction network forms a large hydrophobic core comprising one residue upstream of α1, six in α1, two in the α1/α2 loop, six in α2, one in the α2/α3 loop, and six in α3, collectively accounting for 44% of the 50 residues in the ordered region (S5A Fig). A portion of this core exposed on the surface contributes substantially to ubiquitin binding, as described below.

The two well-defined loops connecting the α-helices are each characterized by distinct β-turn types. The α1/α2 loop, together with Leu88 in α1, forms a six-residue segment (⁸⁸LGDAVP⁹³) that adopts an overlapping β-turn conformation (referred to as a double β-turn). Within the segment, residues ⁸⁸LGDA⁹¹ and ⁹⁰DAVP⁹³ correspond to Type II′ and Type VIII β-turns, respectively, according to the β-turn classification [55] (Fig 3B, left; S6A Fig). The type II′ β-turn is canonical, with glycine at the i + 1 position, and the main-chain amide of Ala91 (i + 3) forms a hydrogen bond with the main-chain carbonyl oxygen of Leu88 (i). The type VIII β-turn represents one of the most common nonclassical β-turn categories. The four-residue segment satisfies the geometric criterion for a β-turn (Cα(i)–Cα(i + 3) < 7 Å) and is classified as a type VIII β-turn based on the backbone dihedral angles of the i + 1 and i + 2 residues. Sequence alignment suggests that residues forming the α1/α2 loop are highly conserved across diverse eukaryotic lineages, including vertebrates, insects, gordiids, and the fission yeast S. pombe, with the exceptions of M. sanguinolenta and S. cerevisiae (see Discussion) (Fig 2B). However, at position 93 (position i + 3), the residue can vary (Pro in most species, Met in X. laevis, and Ser in D. melanogaster and S. pombe). Proline is indeed frequently found at the i + 3 position in type VIII β-turns, but no strict residue preference is observed at this position [55]. In addition, because the Pro side chain is oriented outward, it does not contribute to the hydrophobic interactions that stabilize the structure. Taken together, these observations suggest that the type VIII β-turn, namely, the double β-turn forming the α1/α2 loop, is largely conserved across eukaryotes, with only a few exceptions, such as these two species. Furthermore, DALI analysis [56], based on the α1–α1/α2 loop–α2 region, indicated that the α1/α2 loop is uniquely characteristic of UBAh domains in both length and architecture among loops connecting two α-helices, including those of the UBA and CUE domains. Accordingly, this loop serves as a hallmark feature distinguishing the UBAh domain from the UBA and CUE domains.

The α2/α3 loop (¹⁰⁴HKFD¹⁰⁷) adopts a type I′ β-turn conformation (Fig 3B, right; S6B Fig). Although type I′ is among the most common β-turns and has no strict residue preference, Phe106 and Asp107 are highly conserved. This is likely because they contribute to the stabilization of the helical bundle. The aromatic ring of Phe106 interacts with the aliphatic portion of Arg76 in α1 and with Leu69 in the upstream region of α1, forming hydrophobic contacts that likely stabilize the interface between the α2/α3 loop and α1 (S5B Fig). Phe/Tyr residues are observed at position 106; Leu/Ile/Thr (methyl group) or Lys (aliphatic side chain) at position 69; and Arg/Lys/Phe/Leu at position 76 (with exceptions such as Cys at position 106 in D. rerio and Cys at position 76 in Gordionus sp.) (Fig 2B). Thus, similar hydrophobic interactions may be conserved in most eukaryotes. However, because the residues at positions i and i + 1 are not conserved, it remains unclear whether the type I′ β-turn conformation is widely conserved across eukaryotes. Nevertheless, given that the putative turn length was consistently four residues, the overall turn conformation could be maintained. Moreover, the side chain of Asp107 forms a hydrogen bond with the backbone amide and/or the side chain amide of residue 109, thereby functioning as an N-terminal helix cap of α3 (S5B Fig). As Asp at position 107 is invariant and Gln/Glu/Ala/His are observed at position 109 (Fig 2B), the hydrogen bond with the backbone amide is likely to be more consistently preserved and may play a more central role in helix capping.

Structural basis of the UBAh–ubiquitin interaction: comparison with UBA and CUE domains

A schematic representation of the main residue–residue interactions is provided in S7 Fig. The complex reveals that the hydrophobic patch centered on Ile44 in β3 of ubiquitin constitutes the primary binding interface with UBAh (Fig 4A and 4B). The hydrophobic patch in ubiquitin is formed by Leu8 (β1/β2 loop), Ile44 (β3), His68 (β5), and Val70 (β5), whereas the corresponding hydrophobic surface in UBAh is composed of Val87 and Leu88 (α1), as well as Leu112 and Leu116 (α3). A similar interaction is observed in UBA and CUE domains [57,58]. The key hydrophobic residues involved in ubiquitin recognition are Val87, Leu112, and Leu116 in UBAh; Met19, Ile43, and Leu47 in CUE; and Met342, Leu365, and Leu369 in UBA (Fig 5A, 5B, and 5C). Although the relative positions of these residues are largely conserved, their overall spatial arrangement is not identical, as the three helices are subtly shifted relative to one another (Fig 5D and 5E). Notably, because the side chain of Val87 in UBAh, the penultimate residue of α1, directly contacts that of Ile44 in β3 of ubiquitin, Val87 appears to be a critical determinant of this hydrophobic interaction. This position is invariably occupied by hydrophobic residues such as Val, Met, or Leu across all members of the THB–Ub group (S8 Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Interaction of UBAh with ubiquitin.

(A) Ribbon representation of the UBAh–ubiquitin complex. The center panel is shown in the same orientation as Fig 3A. In the left and right panels, UBAh and ubiquitin are shown separately, each in the rotated orientation indicated in the figure. Side chains of residues involved in complex formation are shown, as described in the text. Lys63 in ubiquitin, shown in black, is the linkage residue in ubiquitin chains on the small subunit of stalled ribosomes. (B) Surface representations of the UBAh–ubiquitin complex in the same orientations as in Fig 3A. Hydrophobic residues (Ala, Ile, Leu, Met, Phe, Pro, Trp, and Val) are shown in green, while Gly47 and His68 of ubiquitin are shown in magenta and khaki, respectively. The cavity and protrusion formed mainly by the α1/α2 loop in UBAh and by Gly47 in the ubiquitin β-turn, respectively, are indicated by red dotted circles. (C) Close-up view of the putative interaction network formed by salt bridges and hydrogen bonds between α3 of UBAh and β3 of ubiquitin, with interactions shown as red dotted lines. (D) Close-up view of the putative interaction between the main-chain carbonyl oxygen (C = O) of Val87 in UBAh and the main-chain amide (N–H) of Gly47 in ubiquitin, mediated by a water molecule (blue sphere, position estimated from the UBA crystal structure [PDB ID: 2QHO]). The side chain of Val87 makes direct contact with that of Ile44 in ubiquitin; this hydrophobic contact represents one of the most crucial interactions.

https://doi.org/10.1371/journal.pone.0348877.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Structural comparisons among the three-helix bundles in complex with ubiquitin.

(A–C) Ribbon models of ubiquitin (white) complexes with UBAh (pale green), CUE (yellow) [PDB ID: 1OTR], and UBA (pink) [2QHO], shown in the same view as (A). Ile44 in the hydrophobic ubiquitin patch is highlighted in magenta in this panel and in all other related panels. Other key residues involved in ubiquitin binding are shown in various colors. (D) Superimposition of ribbon models of ubiquitin complexes with UBAh, CUE, and UBA aligned on ubiquitin. (E) Superposition of ribbon models of the three complexes, aligned on α2. (F) Close-up view of the α1/α2 loops in the three complexes. The superposition was rotated 50° forward relative to that shown in (E).

https://doi.org/10.1371/journal.pone.0348877.g005

Additionally, a salt bridge and a hydrogen bond may form between α3 of UBAh and β3 of ubiquitin, positioned over the hydrophobic patch–patch interface (Fig 4C). The guanidinium group of Arg42 (β3) in ubiquitin may form a hydrogen bond with the hydroxyl group of Ser113 (α3) and a salt bridge with the carboxylate group of Glu117 (α3), both in UBAh; moreover, it may form an intramolecular hydrogen bond with the amide carbonyl group of Gln49 (β2). Thus, these interactions appear to establish a network centered on Arg42 in ubiquitin. This network likely contributes to the positional shift of α3 relative to α1 and α2 upon ubiquitin binding (S4C and S4D Fig). Notably, even in vertebrates, position 113 is occupied by Ser or Asp, whereas position 117 is occupied by Glu, Ser, Ala, or Lys (Fig 2B). At least one electrostatic interaction is likely to be conserved. Similar electrostatic interactions have been reported in UBA and CUE domains despite differences in position and residue identity [59,60].

Furthermore, two characteristic interactions were observed between the C-terminal region of α1 (Leu88) and the α1/α2 loop of UBAh (⁸⁹GDAVP⁹³), and the β-turn between β3 and β4 of ubiquitin (⁴⁵PAGK⁴⁸): (i) Based on the angle and distance between the C = O and N–H groups, the C = O group of Val87 in UBAh appears to form a water-mediated hydrogen bond with the N–H group of Gly47 in ubiquitin (Fig 4D). This interaction has also been observed in the X-ray ubiquitin complex structures of the UBA domain of the ubiquitin ligase EDD [2QHO] [59] and the CUE-like domain of N4BP1 [8T48] [61] (S9A Fig). Conversely, in the NMR and X-ray ubiquitin complex structures of Dsk2 UBA [1WR1; 4UN2] [62,63], the C = O of the corresponding residue in UBA forms a direct hydrogen bond with the N–H of Gly47 in ubiquitin. Thus, whether a water molecule participates in this interaction appears to vary among the members of the THB–Ub group. This hydrogen bond may be regarded as functioning as an N-terminal capping of α1. (ii) Residues in the α1/α2 loop are involved not only in the interaction but also in maintaining the UBAh structure. Leu88 and the first three residues (⁸⁹GDA⁹¹) of the α1/α2 loop form a Type II′ β-turn, whose conformation orients the side chain of Ala91 to establish a van der Waals contact with the other H^α proton of Gly47 in ubiquitin (S9B Fig). In contrast, the conformation of the other β-turn (Type VIII; ⁹⁰DAVP⁹³) orients the side chain of the following residue, Val92, inward, enabling hydrophobic interactions with Ile96 in α2 and Val115 in α3 (S9B Fig). Furthermore, the Type II′ β-turn contributes to the formation of a hydrophobic, dish-like cavity (Fig 4B). This cavity accommodates the slight protrusion of the ubiquitin β-turn between β3 and β4; in other words, the Type II′ β-turn is positioned to avoid steric clashes with this ubiquitin β-turn.

Distinct α1/α2 loop conformations in UBAh, UBA, and CUE domains: functional implications

Although the sequence and/or length of the α1/α2 loop differ among the three domains, the loop adopts distinct conformations that avoid steric clashes with the β3/β4 turn of ubiquitin (Fig 5F; S6A, S6C, S6D and S8 Figs). This loop variation is likely to subtly alter the relative positioning of the three α-helices (Fig 5D and 5E), thereby affecting the area and shape of the hydrophobic patch and, consequently, the ubiquitin-binding affinity. These findings further support the conclusion that UBAh is a novel member of the THB–Ub group. Furthermore, sequence alignments among the three domains revealed a distinct hallmark motif in UBAh, located in and downstream of the α1/α2 loop: ⁸⁷VLGD/E⁹⁰ (Fig 2B; S8 Fig).

The corresponding region contains the ³⁴²MGF³⁴⁴ motif in UBA and the ¹⁹MFP²¹ motif, also known as Hy–F/L–P–X–Hy, in CUE (Fig 5B and 5C; S6 and S8 Figs). These two motifs are hallmarks of UBA and CUE domains, respectively [62,64]. Similar to UBAh, in each domain, these specific residues within the motif also play important roles in ubiquitin recognition and in forming the characteristic structural features of each domain.

Chemical shift perturbations and affinity determination of the UBAh–ubiquitin complex

We performed chemical shift perturbation analysis using [¹⁵N-H] heteronuclear single quantum correlation (HSQC) spectroscopy. Chemical shift changes were examined by comparing the spectra of [¹⁵N]-labeled UBAh alone and in the presence of increasing concentrations of unlabeled ubiquitin (Fig 6; S10 Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. NMR chemical shift perturbations of labeled UBAh upon binding to non-labeled ubiquitin.

(A) Chemical shift perturbations of labeled UBAh upon ubiquitin binding. Perturbation values were obtained from [¹H,¹⁵N]-HSQC spectra in the absence and presence of ubiquitin (S10 Fig). The weighted chemical shift change Δδ (¹⁵N + ¹H_N) for each residue is shown. The red line indicates the average perturbation value (0.0735) plus 0.5 times the standard deviation (0.0711), and the red dotted line indicates the average plus two times the standard deviation. Prolines and residues with resonances that were not assigned after the addition of the peptide are indicated by arrowheads. Only residues with significant chemical shift changes are shown in the graph. (B) Residues with chemical shift changes mapped on the ribbon representation of free UBAh (upper panel) and onto the surface representation (lower panel). The right view was obtained by rotating the left view by 180°. Residues are colored according to the magnitude of the chemical shift change upon ubiquitin binding (blue, above the red dotted line; cyan, between the dotted and solid lines).

https://doi.org/10.1371/journal.pone.0348877.g006

First, to assess the structural integrity of the UBAh–ubiquitin complex, we mapped the chemical shift perturbations induced by ubiquitin binding onto the solution structure of UBAh determined in this study using a 1:3 molar ratio (Fig 6). Significant chemical shift changes were observed in residues at the C-terminal ends of α1 and α3, as well as in G89 in the α1/α2 loop, but not in residues in α2. These results are consistent with the binding mode between UBAh and ubiquitin observed in the complex. Furthermore, as a similar pattern was observed in the Cα RMSD values between the free and ubiquitin-bound UBAh structures (S4C Fig), UBAh likely undergoes subtle conformational changes in α1 and α3 relative to α2 upon ubiquitin binding.

To determine the dissociation constant (K_d) for the interaction between UBAh and ubiquitin, we fitted the chemical shift perturbation data to a single-site binding model [52,65]. Global fitting using data from Met84 (α1), Glu86 (α1), and Gly89 (α1/α2 loop) yielded a K_d of 52 µM (S11 Fig). Generally, the K_d for ubiquitin-binding domain–ubiquitin interactions is relatively large, reflecting their weak affinity; UBA domains typically exhibit K_d values of 10–100 µM, whereas CUE domains display weaker affinities of 50–300 µM [58,66,67]. Thus, the measured K_d for the binding of UBAh to ubiquitin is within a reasonable range, suggesting that the complex is moderately stable.