key: cord-0834030-3a7yrpkt
authors: Staup, Andrew J.; De Silva, Ivon U.; Catt, Justin T.; Tan, Xuan; Hammond, Robert G.; Johnson, Margaret A.
title: Structure of the SARS-Unique Domain C From the Bat Coronavirus HKU4
date: 2019-05-27
journal: Nat Prod Commun
DOI: 10.1177/1934578x19849202
sha: 2bfcdf5569f14eacd54b0ea1593a8158bc1ee5d1
doc_id: 834030
cord_uid: 3a7yrpkt

Coronaviruses (CoVs) that cause infections such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome phylogenetically originate from bat CoVs. The coronaviral nonstructural protein 3 (nsp3) has been implicated in viral replication, polyprotein cleavage, and host immune interference. We report the structure of the C domain from the SARS-Unique Domain of bat CoV HKU4. The protein has a frataxin fold, consisting of 5 antiparallel β strands packed against 2 α helices. Bioinformatics analyses and nuclear magnetic resonance experiments were conducted to investigate the function of HKU4 C. The results showed that HKU4 C engages in protein-protein interactions with the nearby M domain of nsp3. The HKU4 C residues involved in protein-protein interactions are conserved in group 2c CoVs, indicating a conserved function.

Coronaviruses (CoVs) are known as single-stranded enveloped RNA viruses which possess positive-sense RNA genomes. 1 They are responsible for potentially lethal infections related to the human respiratory system, such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS). Additionally, CoVs cause other types of infections, including acute respiratory distress and acute lung injury syndrome; upper and lower respiratory disease ranging from mild to severe; and gastrointestinal disease. 2 While SARS and MERS CoVs are similar to other CoVs in genomic composition, they belong to two different phylogenetic lineages. Betacoronaviruses can be divided into 4 lineages known as A, B, C, and D. The SARS CoV belongs to the lineage B, 3 while the MERS CoV belongs to the C lineage. These two types of betacoronaviruses are predicted to originate zoonotically from bats, which are known to be a reservoir host for CoVs. 3 Additionally, some bat species such as the lesser bamboo bat (Tylonycteris pachypus) can act as hosts for the bat CoV HKU4, which is a betacoronavirus in the C lineage. 4 Since this virus exhibits a high similarity to MERS with respect to the spike protein, ribonucleic acid (RNA) polymerase, and nucleocapsid protein, HKU4 is a relative of the MERS CoV. 5 CoVs contain one of the largest RNA genomes, which encodes two important polyproteins known as the replicase polyproteins 1a and 1ab. 6 Viral proteases catalyze the cleavage of these polyproteins into several nonstructural proteins (nsp) which are involved in the formation of the replicase-transcriptase complex (RTC). This complex carries out RNA synthesis, RNA processing, and interference with the host cell innate immune system. 7, 8 The largest nsp present in this complex is nsp3 which is the most variable region in the CoV genome. 9 nsp3 is a complex protein which may contain one or several macrodomains, several conserved uncharacterized domains, a papain-like cysteine protease, and transmembrane regions. Therefore, it can be identified as a multifunctional protein. 10 The macrodomain is known to be a conserved region of nsp3 present in different viral species including CoVs. 8 This macrodomain plays a major role in the virulence of CoVs since they can infect through protein-protein interactions related to host immune system. 11 There is an additional macrodomain called the SARS-unique domain (SUD) which contains two divergent domains and a conserved domain. This three-domain structure is capable of binding G-quadruplex nucleic acids 10, 12 and of interacting with p53 to target it for degradation by stabilizing the RCHY E3 ligase. 13 We report here the first structural study of the HKU4 SARS-unique region.

We show that HKU4 C adopts a frataxin-like fold characteristic of the SUD-C domain. There is minimal conservation relative to group 2b viruses. Therefore, the function of this protein is likely to be specific to group 2c viruses including the Tylonycteris CoVs and MERS. We also characterized the folding of the neighboring SUD-M macrodomain and show that the M domain is a 150-residue independently folded, monomeric domain for which the fold does not change in the presence of the C domain.

Uniformly 15 N, 13 C-labeled HKU4 C domain was expressed and purified from Escherichia coli. Domain boundaries were predicted employing secondary structure prediction with Jpred 14 and multiple sequence alignment with Fold and Function Assignment System, 15 resulting in a construct comprising the residues 1445 to 1522 of nsp3. A 6xHis fusion tag was employed to assist purification, leaving the non-native residues Ser-His-Met at the N-terminus after cleavage.

Assignment of 92% of the observable resonances of the backbone and side chains was achieved using triple resonance experiments. 16 All backbone 15 N, 1 H, and C′ atoms excluding the Ser-His-Met tag remaining from cloning were assigned. The assigned chemical shift list was deposited in the BioMagResBank 17 under the identifier 30531. Structure determination was accomplished utilizing 3D 15 N-and 13 C-resolved nuclear Overhauser effect spectroscopy (NOESY) experiments.

HKU4 C adopts a frataxin-like fold, 18 consisting of 5 antiparallel β strands packed against 2 α helices ( Figure 1a ). The helices are located at the N-and C-termini of the protein and are oriented parallel to each other. The hydrophobic core of the protein is comprised of residues from β strands 2 to 4 as well as α helices 1 and 2. The hydrophobic core residues of the ensemble (Figure 1b) are well defined across all conformers. The fold is classified in Structural Classification of Proteins 19 as similar to the N-terminal domain of CyaY. 20 The fold has also been identified in the CyaY family of regulatory proteins; frataxins, which are involved in the formation of iron-sulfur complexes in mitochondria; and C domains of the coronaviral SARS-unique region.

The structure consists of a β sheet composed of 5 β strands flanked by an N-terminal and a C-terminal α helix. The helix α1 consists of the residues 1448 to 1456. This is followed by several loop residues leading into the first β strand β1 (1471-1473) and is immediately followed by a β turn (1474-1476). A curved sheet formed by the remaining 4 β strands connected to one another by β turns is observed in the following residues: β2 1476 to 1483, β3 1487 to 1491, β4 1496 to 1499, β5 1504 to 1506. A final C-terminal helix α2 consists of residues 1510 to 1519. Table 1 summarizes the input and statistics of the solution nuclear magnetic resonance (NMR) structure determination. The target function of the ensemble of 20 conformers of Figure 1b is 2.20 ± 0.82 Å 2 . The backbone root-mean-square deviation (RMSD) is 0.37 ± 0.08 Å 2 measured from residues 1449 to 1519. Validation of the NMR structure was carried out on the basis of calculation input requirements, agreement with NMR observables, local residue geometry, and the fold of the protein. 21 The ensemble of 20 conformers representing the solution structure (RMSD 0.37 Å 2 ) is well defined by the experimental NOE and dihedral angle constraints, including 9 long-range NOEs per residue. Taken together, the results indicate a high-quality structure determination.

Alignment of the sequence and structure of HKU4 C with protein databases identified similarities to coronavirus C domains of SARS, mouse hepatitis virus (MHV), and HKU9 ( Table 2 ). The highest similarity was obtained to the SARS C domain, with a distance matrix alignment (DALI) score of 9.6 being obtained.

DALI and TM-align were used to identify other proteins which are structurally similar to HKU4 C. Both DALI and TM-align predict the SARS C domain, the HKU9 C domain from the Rousettus BtCoV, and the MHV nsp3 C domain as the top three homologous proteins for HKU4 C. Both DALI and TM-align predict almost a similar sequence identity (25%) for HKU9 C, while the lowest sequence identity can be observed for SARS C according to both servers ( Table 2) .

A more detailed analysis of conservation patterns was carried out using the program ConSurf. 25, 26 The residues %ID is the percentage of identities among amino acid sequences of different proteins. DALI scores which are higher than 2.0 indicate high similarities to the query sequence. 23 TM-align score is a measurement of the similarity in 3D fold among proteins. 24 F1451, G1459, W1464, K1480, S1485, K1500, L1505, F1507, R1514, L1517, S1519, and R1520 were most conserved with respect to orthologues. In addition, the following regions of the protein were most conserved overall: 1459 to 1467, 1480 to 1491, and 1513 to 1520 ( Figure 2) . A comparison with secondary structure and surface accessibility properties showed that the highly conserved regions of the protein correspond with the β strands 2 and 3 (residues 1480-1491) and α helix 2 (residues 1513-1520). Some of these residues, such as F1451, W1464, and F1507, are buried in the hydrophobic core of the protein, indicating that their interactions are likely to be necessary to maintain the fold of the protein.

However, the conserved region 1459 to 1467, located between α helix 1 and β strand 2, contains several residues on the surface of the protein.

To facilitate further discussion, we define the surface regions of the protein as follows: the α face, which is composed of solvent-exposed residues from both α helices; the β face, which is located on the opposite side of HKU4 C relative to the α face and is composed of β strand and β turn residues; the conserved face, which contains several conserved residues in the C lineage; and the variable face, which contains several non-conserved residues and is located on the opposite side of HKU4 C relative to the conserved face.

HKU4 C was subjected to functional annotation using the prediction servers COACH, COFACTOR, 27 TMSite, 28 and RaptorX. 18 The results indicated that HKU4 C may bind ligands such as peptides and ions (Table 3) . Based on the confidence scores (C-scores) predicted by both TMSite and COACH, binding of acetate ion to HKU4 C shows the highest probability. Both TM-Site and COACH predict the same binding site for acetate ion on HKU4 C, which is located on the variable face, while Raptor X predicts a binding site on the β face.

A second functional possibility is that of a peptide-binding site. Peptides are also possible ligands according to TM-Site and COFACTOR. The binding site for peptides predicted by TM-Site would be situated on the α face, near the C terminus (Table 3) , while multiple binding sites are predicted by COFACTOR.

In addition to these ligands, calcium ions are also capable of binding with HKU4 C according to the predictions generated by TM-Site and Raptor X ( Table 3 ). The probability of binding calcium ions to HKU4 C has been indicated by TM-Site with a C-score value of 0.21. Additionally, Raptor X is also able to predict that calcium ions can bind to the amino acid residues such as K1480 and G1486 which are located on the β face.

We next investigated how the HKU4 C protein might interact with other domains within the nsp3 protein. A prime candidate for such interactions is the neighboring domain of the SARS-unique region, the M domain. The C domain occurs together with the M domain in several CoVs 9,10,27 and has been found to modulate the binding specificity of the M domain. 10 Figure 2 . (a) Molecular surface of HKU4 C colored according to residue conservation among orthologues. Conservation levels are as follows: 9 represents the most highly conserved residues, while 1 represents the most variable residues. The structure is shown with the conserved face closest to the viewer, with the α face and β face indicated on the structure. The variable face of HKU4 C is on the opposite side of the protein relative to the viewer, and therefore cannot be seen from this perspective. (b) Multiple sequence alignment. Secondary structures of HKU4 C residues are indicated above them by rectangles and arrows, which represent α helices and β strands, respectively. Solvent-exposed residues of HKU4 C are represented by blue Xs, while residues buried in the core are represented by red Bs. Finally, HKU4 C residues marked by a "δ" experienced a chemical shift above the threshold in the HKU4 MC construct. (c) A graphical key of conservation levels determined by ConSurf is shown. 25 Residues in the sequence alignment that are yellow lack sufficient data to determine their conservation reliably.

Two additional constructs were cloned, purified, and subjected to preliminary NMR analysis: first, a construct spanning the residues 1319 to 1445 (M domain); and second, a didomain construct spanning the nsp3 residues 1319 to 1522, HKU4 MC. Both proteins were purified using similar protocols to HKU4 C and were monomeric in solution as shown by gel filtration chromatography (Supplemental Figure 1) . Figure 3 shows analyses of the heteronuclear single quantum coherence (HSQC) spectra of these constructs. The assigned HSQC spectrum of HKU4 C is shown in Figure 3a , with 95 crosspeaks. The HSQC profile is shown in Figure 3b . The profile indicates a pure, well-folded protein sample that is suitable for NMR characterization due to the homogeneity of peak distributions and intensities. 29 A peak count and profile of the HKU4 M domain revealed a globular protein with high chemical shift dispersion and even peak intensities, indicating that the domain is independently folded in the absence of the C domain (Figure 3b,c) . A total of 141 peaks were observed, which is within 10% of the expected 145.

A detailed comparison of the spectra showed that the C domain overlays closely in the absence and in the presence of the covalently attached M domain (Figure 3e ). This indicates that the fold of the protein is maintained in the didomain construct and there are no major conformational changes. However, significant chemical shift perturbations to the C spectrum were observed, indicating either an interaction between the two domains or minor conformational changes that occur in the didomain construct. These perturbations were mapped on the HKU4 C structure in Figure 4a . The perturbations are most intense on the conserved face of the protein. A similar analysis was carried out for the M domain, for which resonance assignments have not been obtained. Again, the HSQC spectra of this domain overlay closely in the absence and in the presence of the covalently attached C domain (Figure 3f ). There are 31 residues in HKU4 C that experienced a chemical shift perturbation greater than or equal to 0.022 ppm (Figure 4c ). There are 50 residues in HKU4 M that experience a chemical shift perturbation greater than or equal to 0.008 ppm. These observations are consistent with an interaction between the two domains in the didomain protein.

The flexibility of the linker region between the domains of HKU4 MC was further investigated using NMR relaxation experiments. 15 N{ 1 H}-NOE plots of HKU4 C and the individual domains of HKU4 MC are shown in Figure 5 . In the HKU4 C residues shown in Figure 5a , residues with NOE values less than 0.6 are the terminal residues Q1445 and A1521, indicating no flexible residues within HKU4 C. Although HKU4 C maintains its overall stability in the didomain protein, Q1445 loses flexibility as it is no longer a terminal residue, while R1520 becomes more flexible, as shown in Figure 5b . These experiments probed fast dynamics on the ps-ns timescale and do not preclude slower timescale conformational exchange.

The HSQC spectrum of HKU4 MC also shows a protein that is primarily globular (Figure 3d) ; 229 peaks were observed for this protein, which is close to the sum of the 95 peaks observed for HKU4 C and 141 observed for HKU4 M, minus the 3 peaks of the shared Q1445 residue.

There are 4 residues in the HKU4 M domain of the didomain protein with I rel values within the 0.2 to 0.4 range, designating N-terminal or C-terminal residues that are flexible in the HKU4 M domain, 10 as shown in Figure 5c . There are 7 more residues within HKU4 M with values below 0.6, bringing the total number of flexible residues to 11. These data suggest a short linker between HKU4 M and HKU4 C domains no longer than 4 residues in length.

Additionally, there are 16 residues in the HKU4 MC didomain NOE plot that are flexible, as shown in Supplemental Figure 1 . The additional 2 residues not accounted for in the HKU4 MC spectrum could be perturbed peaks that could not be assigned to either of their individual domain. Further characterization of the didomain would need to be performed for elucidation. The solution structure of the HKU4 C domain revealed a frataxin-like fold. 30 The fold family has similarity to other proteins present in CoV SUDs, implying the potential for a conserved function. This observation adds to other lines of evidence suggesting the SUD provides essential functions for CoV virulence 31 and replication. 13, 32 Despite this, sequence similarity to other viral proteins is low, with the maximum similarity being 25% to the HKU9 C protein (Table 2) . Thus, while the C domain clearly belongs to a conserved fold as indicated by DALI scores of 9.6, 7.0, and 6.5 and TM-Align scores of 0.70, 0.63, 0.64 predicted for HKU9 C, SARS C, and MHV C respectively, it may not necessarily retain conserved functions among CoV phylogenetic groups.

According to Table 2 , HKU4 C is similar in tertiary structure to other betacoronaviruses such as HKU9 C, SARS C, and MHV C. Even though SARS C is the lowest in sequence similarity toward HKU4 C, it has the highest structural similarity with HKU4 C. This is indicated by the low RMSD values according to both servers ( Table 2 ). SARS and MHV CoVs belong to lineage B and A, respectively, while HKU4 and HKU9 CoVs belong to lineage C and D, respectively. 3, 5, 33 According to the structural data generated during our bioinformatics analysis (Table 2 ), HKU4 C has a similar fold as SARS C, MHV C, and HKU9 C. Therefore, it can be hypothesized that HKU4 C should have the same origin as the C domain present in other betacoronaviruses of the C lineage.

The results of functional annotation using prediction servers indicate that HKU4 C may have the ability to bind peptides and/or small ions such as acetate and calcium. Additionally, it can be hypothesized that the β face and the region opposite to the M domain, the variable face, have possible binding sites for small ligands such as acetate and calcium ions. Based on the C-score value predicted (0.7) for the binding reaction between acetate and HKU4 C by COACH, acetate should have the highest probability to bind HKU4 C. COACH is a meta server which is specific in predicting possible ligands and their corresponding binding sites for the given amino acid sequence of a particular protein. In order to generate these predictions it uses other predictions made by other online severs like TM-Site, S Site, COFACTOR, and FINDSITE among others. 26 TM-Site predicts the same C-score value for both acetate and calcium (Table 3) . Therefore, both acetate and calcium have the same tendency for binding HKU4 C according to the results generated by TMSite.

Another possible function of HKU4 C is binding proteins or peptides. According to the results (Table 3) , possible binding sites involved in protein-protein interactions can be found all over the protein, including the β face and ɑ helices situated close to the C-terminus. TM-Site predicts binding of an amino acid sequence (Tyr-Ser-Arg-Ser-Pro-Thr-Ala) in peptides to the α face of HKU4 C and that prediction has a C-score value of 0.22 which is the highest C-score predicted for binding reactions between HKU4 C and peptides. Therefore, it can be considered as the most reliable prediction for protein-protein binding.

COFACTOR is an online server which can predict biological functions, homologous proteins, and amino acid residues of the possible binding site for a given protein structure. COFACTOR makes all these predictions based on other online servers and databases including TM-Align, Bio Lip Protein Database, Protein Data Bank (PDB), and UniProt. 25, 34 Additionally, COFACTOR predicts binding of two other I rel lower than 0.6 indicates flexibility in the residue. Q1445 and A1521 in the HKU4 C single domain fall below this threshold, as they are residues found in the N-and C-terminal respectively. In the didomain, R1520 gains greater flexibility, while Q1445 loses flexibility due to no longer being a terminal residue. There are 11 residues in the HKU4-M domain that are flexible. NOE, nuclear Overhauser effect.

oligopeptides and both binding reactions have a C-score value of 0.1. Therefore, those two protein-protein binding reactions should occur with a less probability than the protein-protein binding reaction predicted by TM-Site. One of the oligopeptides predicted by COFACTOR contains only 5 alanine residues which may bind with the β face of HKU4 C. The other oligopeptide which contains an amino acid sequence of Gln-Arg-Lys-Try-Pro-Leu-Arg-Pro may have the ability to bind with amino acid residues which are situated all over the protein. For this specific peptide, it is hard to predict a specific binding site within HKU4 C.

It is important to note that the Nsl complex was also identified as similar protein for HKU9 C, suggesting a possible surface and functional similarity. It may also be relevant that protein-protein binding has also been identified as a biochemical function of the divergent SUD-M domain of the SARS-CoV. This protein binds to host cell p53 and targets this protein for ubiquitination and degradation. 13 Hence, protein or peptide binding by the neighboring C domain of bat CoVs could have potential biological significance.

A predicted functional site previously identified in the HKU9 C domain 33 was found to not be conserved in HKU4 C. The predicted functional site of HKU9 C is comprised of residues 1360 to 1362 (RDW) and 1381 to 1383 (KRG), with all of the residues except W1362 being solvent exposed. The sequence alignment of Figure 2b shows that the analogous residues of HKU4 C are 1462 to 1464 (HSW) and 1483 to 1485 (HWS), where the conserved W1464 is inaccessible to solvent in both proteins. The aligned residues occupy similar regions of their respective proteins, though the lack of conservation of residues from the HKU9 C-predicted functional site suggests that its function is not retained in HKU4 C. When compared to group 2c CoVs such as the MERS CoV, residues 1464 (W) and 1483 to 1485 (HWS) of HKU4 C are highly conserved (Figure 2b) .

The HKU4 C residues that are aligned with the HKU9 C predicted functional site residues experienced significant chemical shift perturbations in the HKU4 MC didomain, indicating that they form protein-protein interactions with HKU4 M (Figures 2 and 4) . These residues are well conserved in group 2c CoVs including MERS, implicating protein-protein interactions with the M domain as a conserved function. Additionally, the residues of the HKU9 C-predicted functional site are not well conserved in HKU4 C. The different composition of the HKU4 C conserved face (HSW, WHS in HKU4 as opposed to RDW, KRG in HKU9) and its involvement in protein-protein interactions with HKU4 M both allude to a divergent function of the HKU4 C domain.

Based on the results of chemical shift perturbation of HKU4 C residues in the HKU4 MC didomain, we propose that HKU4 C engages in protein-protein interactions with HKU4 M on its conserved face. The conservation of residues in HKU4 C's conserved face in group 2c CoVs suggests that their purpose is the maintenance of an interface with their respective M domains. The establishment of this interaction surface as well as the lack of conservation of HKU9 C's predicted functional site indicates the possibility of a divergent function for group 2c C domains.

The residues 1445 to 1522 of nsp3 were amplified from a codon-optimized synthetic gene (Genscript, Piscataway, NJ, USA) 35 and cloned into the vector pET-15b-TE (Northeast Structural Genomics Consortium and DNASu) encoding a N-terminal 6xHis tag. The construct was confirmed by DNA sequencing. The protein was expressed in E. coli strain BL21 (DE3) pLysS employing 1 g/L NH 4 Cl and 4 g/L 13 C-glucose for isotope labeling. Following purification by Ni 2+ affinity chromatography, fusion tag cleavage with 4.0 mg tobacco etch virus protease was performed overnight at room temperature; in a second Ni 2+ affinity step and gel filtration, the protein was concentrated into 20 mM sodium phosphate, 150 mM NaCl, and 5 mM d6-DTT using Sartorius Vivaspin 20 ultrafiltration concentrators (Littleton, MA, USA). The protein was monomeric as assessed by size-exclusion chromatography on a HiLoad 26/600 Sephadex 75 pg column (Buckinghamshire, United Kingdom).

The HKU4 M domain and HKU4 MC didomain were expressed using the same vector and E. coli strain. These domains spanned residues 1319 to 1445 and 1319 to 1522 of nsp3, respectively. Each protein was purified using analogous steps, with the exception that HKU4 MC used 7.5 mg tobacco etch virus protease for cleavage, and was then concentrated into 20 mM sodium phosphate, 300 mM NaCl, and 3 mM d 6 -DTT buffer.

NMR samples were prepared at 2.0 mM (HKU4 C), 2.0 mM (HKU4 M), or 3.0 mM (HKU4 MC) protein concentration; 3% D 2 O (v/v) and 0.02% NaN 3 (w/v) were added. Experiments were carried out on Bruker Avance III HD spectrometers at 600 (Faellanden, Switzerland) and 850 MHz 1 H frequencies (Karlsruhe, Germany) equipped with Bruker 5 mm TCI cryoprobes or alternatively on a Bruker Avance II spectrometer at 700 MHz 1 H frequency (Faellanden, Switzerland) equipped with a CP TCI H-C/N-D cryoprobe.

Backbone assignments employed triple resonance HNCA, HNCOCA, HNCACB, CBCA(CO)NH, and HNCO experiments analyzed with the program CARA. 36 Side chain assignments employed 3D 13 C-resolved NOESY (aliphatic and aromatic) and 3D 15 N-resolved NOESY experiments computationally analyzed using the program ASCAN. 37 These assignments were subsequently interactively verified and completed employing the 3D 15 N-TOCSY, HBHA(CO) NH, HCCH-TOCSY, aromatic 13 C-resolved NOESY, aliphatic 13 C-resolved NOESY, and 15 N-resolved NOESY experiments. 1 H chemical shifts were calibrated using DSS sample and 15 N and 13 C shifts were referenced using the Ξ ratio. 38 The CANDID 35 algorithm of the J-UNIO suite 21 was used to pick NOE resonances in the NOESY spectra to calculate the structure of HKU4 C. The first cycle of the calculation resulted in a fold that was consistent throughout the remaining cycles of the calculation. The completion of more cycles of the calculation showed a decrease in target function and RMSD of the ensemble of 20 conformers. The ensemble was then energy-minimized through simulated annealing using CYANA. 39 Validation of the structure was performed using ProCheck 3.5.4, 40 Protein Data Bank validation suite, 41, 42 Protein Structure Validation Software, 43 and MolMol v1.0.7. 44 The input and output of UNIO calculations were also evaluated in addition to the agreement of the structure with NMR observables.

The structure for HKU4 C was deposited to the PDB under the accession code 6MWM. The chemical shift assignments for HKU4 C atoms were deposited to the BioMagResBank under accession code 30531.

Chemical shift perturbations were measured by comparing the 1 H and 15 N chemical shifts of the HKU4 C and HKU4 M constructs to those of the HKU 4-MC didomain. 1 H chemical shift perturbations were measured from the center of each peak with no scaling, while 15 N chemical shift perturbations were measured from the center of each peak and were later multiplied by a scaling coefficient, α, which is equal to the quotient of the range of observed 1 H chemical shifts divided by the range of observed 15 N chemical shifts. 45 A threshold for chemical shifts that were considered perturbed was established by calculating the standard deviation of all chemical shift perturbations in each protein shifts. 45 Chemical shifts were considered perturbed in the didomain construct if the magnitude of the chemical shift perturbation was greater than the threshold value.

The 15 N{ 1 H}-NOE experiments were performed using the Bruker 600 MHz, and the results were processed with the Bruker Dynamics Center.

DALI and TM-Align 33 were used to identify homologous proteins which are very close to the primary and tertiary structures of HKU4 C. First, the amino acid sequence was submitted to COFACTOR online server which uses TM-Align to identify homologous proteins for a given protein from the PDB. The structure of HKU4 C was submitted to DALI which compares the 3D structures of proteins in the PDB with the given protein to identify homologous proteins During the second part of this bioinformatics search, it was aimed to identify possible ligands which can bind to HKU4 C and their possible binding pockets. In this step, online binding site prediction tools like TM-Site, COACH, COFACTOR, and RaptorX 12 were used.

The NSP3 macrodomain promotes virulence in mice with coronavirus-induced encephalitis

Molecular pathology of emerging coronavirus infections

Comparative analysis of twelve genomes of three novel group 2C and group 2D coronaviruses reveals unique group and subgroup features

Coronaviruses in bats from Mexico

Bat origin of human coronaviruses

Characterization of a critical interaction between the coronavirus nucleocapsid protein and nonstructural protein 3 of the viral replicase-transcriptase complex

Plasmodium falciparum genome-wide scans for positive selection, recombination hot spots and resistance to antimalarial drugs

The conserved coronavirus macrodomain promotes virulence and suppresses the innate immune response during severe acute respiratory syndrome coronavirus infection

Proteomics analysis unravels the functional repertoire of coronavirus nonstructural protein 3

SARS coronavirus unique domain: three-domain molecular architecture in solution and RNA binding

Murine coronavirus ubiquitin-like domain is important for papain-like protease stability and viral pathogenesis

The SARS-unique domain (SUD) of SARS coronavirus contains two macrodomains that bind G-quadruplexes

A G-quadruplex-binding macrodomain within the "SARSunique domain" is essential for the activity of the SARS-coronavirus replication-transcription complex

JPred4: a protein secondary structure prediction server

FFAS03: a server for profile--profile sequence alignments

Three-dimensional Fourier spectroscopy

A multiple-template approach to protein threading

Scop: a structural classification of proteins database for the investigation of sequences and structures

Bacterial frataxin CyaY is the gatekeeper of iron-sulfur cluster formation catalyzed by IscS

The J-UNIO protocol for automated protein structure determination by NMR in solution

MolProbity : all-atom structure validation for macromolecular crystallography

Searching protein structure databases with DaliLite v.3

Scoring function for automated assessment of protein structure template quality

ConSurf 2016: an improved methodology to estimate and visualize evolutionary conservation in macromolecules

ConSurf 2005: the projection of evolutionary conservation scores of residues on protein structures

Cofactor: improved protein function prediction by combining structure, sequence and protein-protein interaction information

Protein-ligand binding site recognition using complementary binding-specific substructure comparison and sequence profile alignment

The structure and function of frataxin

SARS-unique domain" (SUD) of SARS coronavirus is an oligo(G)-binding protein

P53 down-regulates SARS coronavirus replication and is targeted by the SARS-unique domain and PLpro via E3 ubiquitin ligase RCHY1

SARS-unique fold in the Rousettus bat coronavirus HKU9

Nidovirus papain-like proteases: multifunctional enzymes with protease, deubiquitinating and deISGylating activities

Protein NMR structure determination with automated NOE assignment using the new software CANDID and the torsion angle dynamics algorithm DYANA

Optimizing the process of nuclear magnetic resonance spectrum analysis and computer-aided resonance assignment. A dissertation submitted to the Swiss federal Institute of technology Zurich (ETH Zürich) for the degree of doctor of natural sciences

Automated amino acid side-chain NMR assignment of proteins using (13) C-and (15)N-resolved 3D [ (1)H, (1)H]-NOESY

IUPAC-IUBMB-IUPAB Inter-Union task group on the standardization of data bases of protein and nucleic acid structures determined by NMR spectroscopy

Combined automated NOE assignment and structure calculation with CYANA

PROCHECK: a program to check the stereochemical quality of protein structures

Announcing the worldwide protein data bank

The worldwide protein data Bank (wwPDB): ensuring a single, uniform archive of PDB data

Evaluating protein structures determined by structural genomics consortia. Proteins

MOLMOL: a program for display and analysis of macromolecular structures

Using chemical shift perturbation to characterise ligand binding

Template-based protein structure modeling using the RaptorX web server

We thank the members of Biomolecular NMR Class for technical assistance.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The UAB Comprehensive Cancer Center NMR facility is supported by grants 1P30 CA-13148 and 1S10 RR022994-01A1. This work was supported by the National Institutes of Health grant R35GM119456, University of Alabama at Birmingham Faculty Startup Funding, and the Department of Chemistry, University of Alabama at Birmingham.

Supplemental material for this article is available online.