NMR reveals structural rearrangements associated to substrate insertion in nucleotide‐adding enzymes

Abstract

The protein NP_344798.1 from Streptococcus pneumoniae TIGR4 exhibits a head and base‐interacting neck domain architecture, as observed in class II nucleotide‐adding enzymes. Although it has less than 20% overall sequence identity with any member of this enzyme family, the residues involved in substrate‐recognition and catalysis are highly conserved in NP_344798.1. NMR studies showed binding affinity of NP_344798.1 for nucleotides and revealed μs to ms time scale rate processes involving residues constituting the active site. The results thus obtained indicate that large‐amplitude rearrangements of regular secondary structures facilitate the penetration of the substrate into the occluded nucleotide‐binding site of NP_344798.1 and, by inference based on sequence and structural homology, probably a wide range of other nucleotide‐adding enzymes.

Abbreviations

AMP

Adenosine monophosphate

APSY

Automated projection spectroscopy

ATP

Adenosine triphosphate

CSP

Chemical shift perturbations

CTP

Cytidine triphosphate

DUF

Domain of unknown function

H/D

Hydrogen/Deuterium

HSQC

Heteronuclear single quantum coherence

K_D

Dissociation constant

NMR

Nuclear magnetic resonance

NOESY

Nuclear Overhauser effect spectroscopy

NTP

Nucleoside triphosphate

P

Protection factor

RMSD

Root mean square deviation

UTP

Uridine triphosphate.

Introduction

The addition of the CCA tri‐nucleoide to the 3′ terminus is fundamental for maturation of tRNAs and protein biosynthesis.1, 2 This reaction is performed by the CCA‐adding family of nucleotidyl‐transferases, which on the one hand catalyze the high‐fidelity incorporation of the three nucleotides in the absence of a nucleic acid template,3 and on the other hand act by proofreading unstable tRNAs.4 Archeal genomes encode class I CCA‐adding enzymes,5 while class II is found in eubacteria and eukaryotes. The typical architecture of the CCA‐adding enzymes consists of four domains denoted, from the N‐ to the C‐terminus, as head, neck, body and tail domains.6, 7, 8, 9, 10 Although individual domains are very similar in the two protein families, class I CCA‐adding enzymes adopt a U‐shaped architecture,11 while class II prefers a seahourse‐like structure.6, 7 In both families the head domain contains the catalytic site, while the neck domain presents residues that interact with nucleotides7, 10 and, jointly with the body and tail domains, is involved in tRNA recognition.

The catalytic mechanism of the addition of the CCA tri‐nucleotide in the absence of a RNA template has been the object of intense structural and biochemical studies, which have yielded more than 45 crystal structures, representing different complexes and catalytic conformations. Crystal structures of a class I CCA‐adding enzyme in ternary complexes with tRNA and nucleoside triphosphates have shown that selective addition of CTP or ATP in this family requires a large conformational rearrangement, which modifies the size and shape of the nucleotide binding site once two cytidines have been added to the tRNA.12, 13 This conformation favors adenosine insertion by preventing proper geometry for a complex that would contain a third cytosine in the catalytic site.11 In class II, comparative studies of CCA‐adding enzymes from T. maritima and an A‐adding enzyme from A. aeolicus suggest that changes in the orientation of the side chain of a highly conserved arginine located in the neck domain are responsible for the fine control of the number and types of nucleotides added.8, 10, 14 Despite these important advances in dissecting different steps in the mechanisms of CCA addition to tRNA, it is still not well established how nucleotides enter the occluded binding site prior to the reaction.

The structure determination of the first representative of the protein family PF06042, NP_344798.1, has enabled the use of NMR spectroscopy to investigate this important step. NP_344798.1 belongs to a new family of class II CCA‐adding‐like proteins that is present in more than 700 bacterial and fungal species.15 Members of PF06042 exhibit the catalytic and NTP recognition elements of class II CCA‐adding enzymes, including the head and the nucleotide‐interacting component of the neck domains, but lack the body and tail domains. Here, we show that NP_344798.1 interacts with nucleotides in a similar fashion as CCA‐adding enzymes, which indicates that these proteins exhibit not only structural but also functional similarities. These similarities provide a platform for the use of the free NP_344798.1 as a model for characterizing the behavior of CCA‐adding enzymes in solution prior to nucleotide insertion in the catalytic site. We focus on the dynamics of structural elements involved in nucleotide recognition, and provide details on those parts of the structure which experience local rearrangements that facilitate the penetration of nucleotides into the binding site.

Results and Discussion

NP_344798.1 and CCA‐adding enzymes exhibit similar structures and nucleotide binding modes

Structural comparisons with NP_344798.1 using the DALI16 server revealed 18 similar structures in the Protein Data Bank, all of which belong to Class II nucleotide‐adding enzymes (Fig. 1). NP_344798.1 superimposes well with the head and the N‐terminal region of the neck domain of these structures [Fig. 1(A)], with the exception of the loop between the strand β3 and the helix α5, which is the most variable structural element in this class of proteins. Despite exhibiting less than 20% sequence identity with other Class II nucleotide‐adding enzymes, NP_344798.1 contains structurally conserved key catalysis residues (D62, D64 and E94) and substrate recognition amino acids (G41, V43 and R44) [Fig. 1(B)] in the same positions and orientations as observed for other family members, such as in the Thermatoga maritima protein [Fig. 1(C)]. These observations prompted us to evaluate the ability of NP_344798.1 to bind nucleotides.

Figure 1.

Structure comparison of NP_344798.1 with nucleotide‐adding enzymes. (A) Ribbon diagram of the protein NP_344798.1 (pdb code: 2LA3)¹⁵ and several nucleotide adding‐enzymes identified by DALI¹⁶ with a Z‐score > 6.5. The structure coordinates of these five proteins belong to apo forms of the catalytic and nucleobase‐interacting neck domains of class II polyA polymerase (PDB_Id 3AQL),25 CCA‐adding enzymes (PDB_Ids 1MIV,7 1OU5,6 3H38⁸ and and 3WFP.26 The base‐interacting elements of the neck domain are shown in color yellow. Polypeptide segments between red dots were not observed in the crystals structures. (B) Structure‐based sequence alignments of the protein domains in (A) using DALI. Structurally equivalent and non‐equivalent residue positions relative to the protein NP_344798.1 are shown in uppercase and inserted dashed line, respectively. Framed residues are conserved in all structures. Asterisks indicate catalytic carboxylate residues. Vertical bars indicate the starting position of the neck domain for 3AQL, 1MIV, 3WFP, 3H38 and 1O5U structures. Residue “X” indicates Selenomethionine. (C) Superposition of the catalytic and nucleobase‐interacting neck domains of the class II CCA‐adding enzyme from Thermotoga maritima (pdb code: 3H38‐A) and NP_344798.1.

NMR chemical shift perturbation studies with increasing concentrations of ATP revealed four polypeptide segments in NP_344798.1 which are affected by the presence of the nucleotide [Fig. 2(A)]; these are clustered around the putative binding site [Fig. 2(B)] and yielded a K_D of 1.88 ± 0.46 mM [Fig. 2(C)]. Three of these binding segments (residues G41−R44, T61−D64 and Q92−V128) are located in the head domain, and the fourth segment, with residues Y170−E181, belongs to the neck domain. Similar chemical shift perturbations patterns were obtained with monophosphate nucleotides such as AMP [Fig. 2(A) and Supporting Information Fig. S2(A‐B)], albeit with a ∼5‐fold reduction in binding affinity when compared to ATP. This suggests that the triphosphate group is involved in additional interactions with the protein. Chemical shift differences induced in NP_344798.1 by sodium phosphate, when compared with NP_344798.1 dissolved in MES buffer, revealed residues in the neck domain which are affected by the phosphate groups of bound nucleotides, for example, K176 and W179 [Fig. 2(A) and Supporting Information Fig. S1]. The fact that the phosphate groups are located near to the end of helix α8 in the binding site is further supported by the behavior of the indole ¹⁵N–¹H group of residue W179 in the presence of different nucleotides, for which chemical shift changes in the [¹⁵N,¹H]‐HSQC spectra of NP_344798.1 in the presence of ATP, CTP and UTP have the same sign (upfield or downfield), but differ in the magnitudes (Supporting Information Figs. S1, S3). While the magnitude depends on the binding affinity, which is primarily determined by interactions with nucleobases far from W179, the uniform sign of the chemical shift changes is a consequence of the similar chemical environments generated by the triphosphate groups of the different nucleotides. In order to further evaluate the role of the nucleobase on the affinity, we performed a titriation with UTP, which resulted in a ∼3‐fold reduction in affinity when compared with ATP [Supporting Information Fig. S2(C‐D)]. The reduced affinity can be rationalized based on crystal structures of CCA‐adding enzymes, which revealed that two residues in the neck domain are responsible for discriminating between different nucleotides. In NP_344798.1 this role was attributed to R173, which is conserved in CCA‐adding enzymes, and to [170, which occupies a position which is typically occupied by aspartic acid [Fig. 1(B)]. The presence of tyrosine instead of aspartic acid could be responsible for the limited selectivity between ATP and UTP, when compared with other nucleotide‐adding enzymes. Overall, these results indicate that the NP_344798.1 molecular architecture is closely related to that of previously characterized nucleotide‐adding enzymes,8, 10, 14 and that NP_344798.1 also exhibits the same nucleotide binding mode [Fig. 3(A,B)]. Chemical shift perturbation data of NP_344798.1 further suggest that the AMP and UTP binding modes are similar to that for ATP (Supporting Information Fig. S3). This is in agreement with the charge distribution observed in the binding site [Fig. 3(C)], which shows that there is a positively charged patch in the neck domain that interacts with the negative phosphate group. These similarities indicate that observations on the nucleotide‐binding site of NP_344798.1 can be extrapolated to other members of the class II nucleotide‐adding enzyme family.

Figure 2.

Nucleotide binding to NP_344798.1. (A) Lower panel: Chemical shift perturbations indyced by the presence of 20 mM phosphate using MES buffer as reference. Upper panel: Chemical shift perturbations with a phosphate‐free solution of the protein NP_344798.1 in the presence of 10 mM ATP (NP_344798.1: ATP = 1: 10), 30 mM UTP (NP_344798.1: UTP = 1: 30) and 100 mM AMP (NP_344798.1: AMP = 1: 100). (B) Neon diagram of the 20 NMR conformers representing the structure of NP_344798.1. Residues experiencing chemical shift perturbations larger than 0.1 ppm in the presence of ATP are shown in red. (C) Titration curves showing the effect of ATP addition to different NP_344798.1 residues. K_D value was computed using a one‐site binding hyperbola nonlinear regression analysis as shown in Eq. 2.

Figure 3.

NP_344798.1 and CCA‐adding enzymes exhibit a similar binding site. (A) Ribbon diagram of protein NP_344798.1 (pdb code: 2LA3;¹⁵ green) superimposed with two structural homologs (pdb code: 3H3A‐Chain B in pink and 3H39‐Chain A in white).8 Structural alignments were done using DALI. The base and ribose moieties of CTP and ATP are shown in blue and cyan sticks, respectively. The phosphate chain is shown in red and W179 side‐chain of NP_344798.1 is shown as a green stick. The figures were prepared using the program PyMol. (B) Glide docking of NP_344798.1 with ATP. 7 ATP molecules with lowest Glide score are shown in cyan sticks. (C) Electrostatic surface potential of the conformer closest to the mean coordinates of NP_344798.1, identifying a charged pocket analogous to the catalytic pocket of class II CCA‐adding enzymes. Positive and negative electrostatic potentials are represented in blue and red, respectively. A green ribbon presentation of the protein is superimposed to guide the eye. (D) Ribbon representation of the NP_344798.1 structure shown in C. Residues experiencing the largest chemical shifts in the presence of ATP are colored dark green. Some amino acid positions are indicated to guide the eye.

Admixture of lowly‐populated conformational variants in free NP_344798.1 facilitate nucleotide insertion in the biding site

Analysis of the 2D [¹⁵N,¹H]‐HSQC spectra of NP_344798.1 in the absence of nucleotides revealed rate processes involving residues located in the loop region between the strand β3 and the helix α5 (residues Q98−Y111), and in the helix α8 (residues R173−Q180). Line shape analysis of backbone amide signals in this loop and the helix α8 regions, revealed line‐broadening for residues Y100, M101, Q103, H104, T172, S175 and Q180, while the signals of residues H102 and R173 were very weak or broadened beyond detection [Fig. 4(A)]. The severe broadening of signals indicates local conformational exchange on the μs to ms time scale, which has been observed in other enzyme active sites.17, 18 The evaluation of the ¹⁵N{¹H}‐NOE experiment revealed motions on the ns timescale only at the N‐terminus, which are not related to the slower rate processes displayed by helix α8 and the loop region located between the strand β3 and the helix α5. These data show that elements of the nucleotide‐binding site do not manifest sub‐nanosecond dynamics [Fig. 4(B)].

Figure 4.

Line‐shape and ¹⁵N{¹H}‐NOE data analysis identifies local structural dynamics on NP_344798.1. (A‐B) Evidence of slow conformational exchange in the NMR structure of NP_344798.1. ω₂‐dimension cross sections of selected NH signals from the 2D [¹⁵N,¹H]‐HSQC spectrum of NP_344798.1. Intensities of residues M101, Q103, H104, S175 and Q180 are shown relative to the reference signals of F72 and Q90. H102 and R173 backbone amide signals broaden beyond detection. (B) ¹⁵N{¹H}‐NOE values versus amino acid sequence of NP_344798.1. Low values indicating higher amplitude motion on the ns‐ps time scale are indicated in red. (C) Neon diagram of the bundle of 20 conformers representing the NMR structure of NP_344798.1 with indication of the chain ends. Residues with global backbone displacement values greater than 0.8 Å (1–4, 100–113, 173–177) are shown in blue. (D) Same bundle as in (C) with indication of residues experiencing slow conformational exchange (magenta) as determined from (A) and those experiencing fast motions (red) as identified from (B).

The analysis of the backbone amide proton protection factors also revealed a distinct behavior of helix α8, which, along with α9, belongs to the neck domain and exchanges with D₂O at a much faster rate than the rest of the regular secondary structures in NP_344798.1 [Fig. 5(A)]. This indicates that residues located in the helices α8 and α9 are in a dynamic equilibrium with locally different conformers that do not include the typical i+4⃯i hydrogen bond connectivities of regular α‐helices. In contrast, the ¹³C_α + ¹³C_β chemical shifts for the helices α8 and α9 resemble the profiles observed for the rest of the regular secondary elements in the protein [Fig. 5(B)], suggesting that the helices α8 and α9 are nonetheless highly populated. It thus appears that the solution structure of NP_344798.1 represents a highly populated state that corresponds to those observed in crystal structures of similar proteins [Fig. 1(A)], but the NMR data also revealed the presence of lowly populated conformations manifesting increased plasticity in the helices α8 and α9 and in the loop region Y100−S113. These “open” conformations are likely to enable the incorporation of nucleotides by facilitating the access to the binding site, which is confined between the head and neck domains [Fig. 3(B)]. The plasticity of helix α8, which contains residues responsible for nucleobase recognition, and the ensuing reorganization of the binding site, may also play a role in nucleotide discrimination. This hypothesis will be further assessed in follow‐up NMR studies involving NP_344798.1:nucleotide complexes. The data presented here reveal low‐frequency conformational motions in the active site of NP_344798.1 and, by inference based on sequence and structural homology, in other nucleotide‐adding enzymes, suggesting that the observed structural plasticity has an important role in the mechanism by which nucleotides are selected to be added to tRNA sequences.

Figure 5.

Analyses of the ¹³C^αβ secondary shifts and hydrogen/deuterium exchange protection factors in NP_344798.1. (A) Plot of the ¹³C^αβ secondary shifts against the amino acid sequence of NP_344798.1. Blue horizontal lines indicate cutoffs to identify population of helical and β‐strand structures, which exhibits positive and negative values, respectively. Regular secondary structures present in the NMR bundle of NP_344798.1 are shown at the bottom. (B) Plot of hydrogen/deuterium exchange protection factors (LogP) amino acid sequence of NP_344798.1. Vertical broken lines indicate the boundaries of helices α8 and α9 to facilitate comparison with panel. (C) Ribbon diagram of the conformer closest to the mean coordinates of the 20 NMR conformers of NP_344798.1. Regular secondary structures are labeled as well as the polypeptide chain ends. Color code: Residues with Log P ≥ 5, green; 3 ≥ Log P < 5, blue; 2 ≥ P < 3, red; and P < 2, grey.

Materials and Methods

Protein production for NMR studies

The protein NP_344798.1 was expressed using a pSpeedET‐NP_344798.1 plasmid produced by the JCSG's crystallomics core and was purified following our standard protocol.19 Details of NMR sample preparation for protein structure determination have been reported elsewhere.15

NMR data acquisition and analysis

All NMR binding studies were performed at 298 K on a Bruker DRX‐700 spectrometer equipped with a 1 mm TXI microprobe and processed using Topspin 2.1. Chemical shift perturbations (CSP) were computed using the Eq. (1):20

$$\text{CSP} \left(\Delta \delta \right) = {\left({\left({\Delta \delta }_{\mathrm{H}}\right)}^2+ {\left(0{.2\Delta \delta }_{\mathrm{N}}\right)}^2\right)}^{1/2}$$ (1)

where ${\Delta \delta }_{HN}$ and ${\Delta \delta }_N$ are the difference of amide proton and amide nitrogen chemical shifts between the apo and holo protein forms. The equilibrium dissociation constant was computed using one‐site binding hyperbola nonlinear regression analysis as shown in Eq. (2):20, 21

$$\text{CSP}\left(\Delta \delta \right)\hspace{0.17em}=\hspace{0.17em}\Delta {\delta }_{\max }/2[\left(\text{Prot}\hspace{0.17em}+\hspace{0.17em}\text{Lig}\hspace{0.17em}+\hspace{0.17em}{\mathrm{K}}_{\mathrm{D}}\right)-\sqrt{\left(\text{Prot}\hspace{0.17em}+\hspace{0.17em}\text{Lig}\hspace{0.17em}+\hspace{0.17em}{\mathrm{K}}_{\mathrm{D}}^2-4(\text{Prot}\ast \text{Lig}\right)}]$$ (2)

where Lig and Prot are the total concentrations of ligand and protein, and K _D is the equilibrium dissociation constant. The data were analyzed using the program GraphPad Prism.

Line shape analysis was performed with 2D [¹⁵N,¹H]‐HSQC experiments acquired at 298 K on a Bruker Avance 600 MHz spectrometer equipped with a CPTCI HCN z‐gradient cryoprobe with ¹H^direct (2048) × ¹⁵N(256) complex points covering a ¹⁵N spectral width of 30 ppm centered at 118 ppm. Prior to Fourier transformation, the data were zero‐filled to ¹H^direct (2048) × ¹⁵N(512) and multiplied with a cosine window in both dimensions, and processed using Topspin 2.1. The line shape in ¹H direct dimension was extracted for the non‐overlap backbone amide cross peaks of F72, Q90, Y100, M101, Q103, H104, T172, S175, Q180 and analyzed by Topspin 2.1. ¹⁵N{¹H}‐NOE experiments were acquired on a Bruker Avance 600 MHz spectrometer at 298 K and processed using Topspin 2.1. The heteronuclear NOE values were computed from the intensity ratio of each assigned cross peak in two [¹⁵N,¹H]‐HSQC spectra, collected with and without 3 seconds proton presaturation. The recycle delay was set to 2 seconds.

Hydrogen/deuterium (H/D) exchange measurements were performed with a protein sample previously lyophilized and reconstituted in D₂O. 2D [¹⁵N,¹H]‐HSQC spectra at different times (T) were collected up to 7.5 days. The 2D [¹⁵N,¹H]‐HSQC experiments were acquired with two transients in the first hour and during the the following 24h with four. Subsequent measurements were recorded with 32 transients. The intensity (Signal‐to‐Noise) of each cross peak was normalized with respect to the number of transients and a single exponential model was used to analyze the time dependence of the intensity decay. All the spectra were acquired at 298 K on a Bruker Avance 600 MHz spectrometer. The data were analyzed in the limit of EX2 exchange regime and the protection factor was computed using the Eq. (3):22

$${\mathrm{P} = \mathrm{K}}_{\text{in}}{/\mathrm{K}}_{\text{ex}}$$ (3)

where k _ex is the observed H/D exchange rate constant, and k _in is the intrinsic H/D exchange rate constant when the amide proton is fully exposed to the solvent. k _in of each residue was calculated by taking into account the neighboring side‐chains effect and corrected for pH and temperature.22

Docking Protocol

Glide23 “ligand docking” calculation was performed with default settings available within Maestro (Maestro, v9.7; Schrödinger: New York, NY, 2014). NP_344798.1 (conformer closest to the mean coordinates of 20 NMR conformers) and ATP structures were prepared for Glide using a previously described protocol.24 Based on chemical shift perturbation data [Fig. 2(A)], four segments with residues G41−R44, T61−D64, Q92−V128 and Y170−E181 were selected to define the receptor grid box of length X=Y=Z=10 Å. Top seven conformers were selected based on the Glide score and analyzed with PyMOL.

Supporting information

Supporting Information