Solution NMR structure of the 30S ribosomal protein S28E from Pyrococcus horikoshii

Abstract

We report NMR assignments and solution structure of the 71-residue 30S ribosomal protein S28E from the archaean Pyrococcus horikoshii, target JR19 of the Northeast Structural Genomics Consortium. The structure, determined rapidly with the aid of automated backbone resonance assignment (AutoAssign) and automated structure determination (AutoStructure) software, is characterized by a four-stranded β-sheet with a classic Greek-key topology and an oligonucleotide/oligosaccharide β-barrel (OB) fold. The electrostatic surface of S28E exhibits positive and negative patches on opposite sides, the former constituting a putative binding site for RNA. The 13 C-terminal residues of the protein contain a consensus sequence motif constituting the signature of the S28E protein family. Surprisingly, this C-terminal segment is unstructured in solution.

The field of structural genomics aims at elucidating the structures of representative proteins from sequence families to provide more complete coverage of protein fold space and at spurring advances in bioinformatic, biotechnological, and biophysical techniques that will make high-throughput protein production and structure determination feasible (Montelione et al. 2000; Burley and Bonanno 2002). The Northeast Structural Genomics Consortium (NESG; www.nesg.org), a pilot project funded by the National Institutes of Health Protein Structure Initiative, is focused on sequence-based clusters of eukaryotic protein domain families. Our efforts target one or more members of protein domain families from eukaryotic proteomes (Liu et al. 2003) and their homologs in bacteria and archaea, particularly proteins with low sequence similarity to proteins of known structure.

The 71-residue S28E protein from the 30S ribosomal unit of Pyrococcus horikoshii (SWISS-PROT ID: RS28_PYRHO, Boeckmann et al. 2003; NESG target i.d. JR19; Wunderlich et al. 2003) is a member of a conserved protein domain family ubiquitous in archaea and eukaryotes. In spite of an abundance of structural information for ribosomal proteins from recent X-ray crystal structures of a bacterial 30S subunit (Wimberley et al. 2000), an archaeal 50S subunit (Ban et al. 2000), and an intact bacterial ribosome (Yusupov et al. 2001), little is known about the role of the S28E family in archaeal and eukaryotic ribosomes. Moreover, no three-dimensional structure of any member of the S28E domain family has yet been described in the literature. The multiple sequence alignment of S28E with homologs from other model organisms that constitute the NESG cluster of targets for this protein family (http://cubic.bioc.columbia.edu/cgi-bin/var/database/pep/getPepEntry.cgi?clusterId=18021 [Liu et al. 2003]), including that from Methanobacterium thermoautotrophicum (NESG target TT744; 64% sequence identity to S28E), is shown in Figure 1 ▶. In this paper, we describe the NMR solution structure of S28E, discuss the insights gained from the structure into the function of the S28E family of ribosomal proteins, and compare it to the solution NMR structure of M. thermoautotrophicum S28E described in the accompanying paper (Wu et al. 2003).

Figure 1.

Multiple sequence alignment of the S28E protein from Pyrococcus horikoshii (NESG target JR19) with proteins (SWISS-PROT ID shown) from NESG cluster 18021 (Liu et al. 2003; http://cubic.bioc.columbia.edu/PEP), aligned using Clustal X (Jeanmougin et al. 1998). Identical and similar residues are box shaded in black and gray, respectively. Regions of secondary structure observed in the solution NMR structure of S28E are indicated above the sequence.

Results and Discussion

Resonance assignments

Using the automated assignment program AutoAssign (Zimmerman et al. 1997; Moseley et al. 2001) with a three-rung (C^α, C^β, H^α) matching strategy for establishing sequential spin system relationships, together with spin-system-type assignment constraints (STACs; Zheng et al. 2003), we obtained ~98% complete assignment of backbone nuclei of S28E, excluding the C-terminal tag (H^N-N: 67/68; C′: 69/71; C^α: 70/71; H^α: 78/79). The resulting AutoAssign connectivity map for S28E is shown in Figure 2 ▶. The assignments from AutoAssign were manually extended into the side chains, yielding ~92% complete side chain assignment, excluding the C-terminal tag: (C^β: 62/63; C^γ: 68/70; C^δ: 34/35; C^ɛ: 5/6; H^β: 100/102; H^γ: 104/106; H^δ: 54/54; H^ɛ: 10/24; N^ɛ: 1/14). In addition, stereospecific isopropyl methyl assignments were obtained for all valines and leucines in the protein (12 in total).

Figure 2.

AutoAssign connectivity map for S28E. Intraresidue (i) and sequential (s) connectivities are shown as horizontal black and gray lines, respectively. Resonances of residues A73 and A74, as well as H78 through H82 in the C-terminal tag, could not be unambiguously assigned due to degeneracy. Sequential spin-system type assignment constraints (STACs) used in the AutoAssign calculations are represented by filled triangles. ³J(H^N-H^α) values are depicted as open circles (<6 Hz), half-filled circles (6 ≤ J ≤ 8 Hz), and filled circles (>8 Hz). Residues with slowly exchanging backbone amide protons are represented by filled diamonds. Interresidue NOE connectivities identified by AutoStructure calculations are shown as thin, medium, and thick black lines, corresponding to weak, medium, and strong NOE interactions. Bar graphs of the consensus CSI (Wishart and Sykes 1994) and ¹H-¹⁵N heteronuclear NOE data are shown. Secondary structural elements in the final S28E structure and predicted by PHD (Rost 1996) are also shown.

Description of the three-dimensional structure

The three-dimensional solution structure of S28E was determined by automated analysis of NMR data using the programs AutoStructure (Huang 2001; Huang et al. 2003; Zheng et al. 2003) and DYANA (Güntert et al. 1997). The solution structure is represented by the 10 conformers that best fit the resulting NMR constraint data. The backbone atoms of the superimposed ensemble, together with ribbon and electrostatic surface potential diagrams of a representative structure, are shown in Figure 3A–C ▶, and structural statistics are listed in Table 1. The structure of S28E is comprised of a Greek-key motif with four β-strands, encompassing residues Pro 8–Arg 17 (which includes β-bulges at Ile 12 and Ile 15), Val 24–Ile 31, Val 40–Arg 46, and Ile 54–Ile 56. These strands interact to form two antiparallel β-sheets, one four-stranded (3–2–1–4) and one three-stranded (3–2–1) β-sheet, packing into an overall β-barrel fold (Fig. 3A,B ▶). This topology is strongly corroborated by several lines of spectral evidence, including the pattern of sequential and medium range NOEs, the slow amide exchange pattern, and local backbone dynamical information indicated by ¹H-¹⁵N heteronuclear NOE data (Fig. 2 ▶). Both β-bulges in β-strand 1 are characterized by strong sequential H^N-H^N NOEs, and a small ³J(H^N-H^α) scalar coupling constant was observed for Val 11. The electrostatic surface potential (Fig. 3C ▶) reveals that the molecule has two distinct charged surfaces: a positively charged surface formed by basic residues from β-strands 2 and 3, and a negatively charged surface due to a cluster of acidic residues from β-strands 1 and 4 and the loops between β-strands 2 and 3, and between β-strands 3 and 4. Several of these charged residues, including Lys 28, Arg 30, Arg 39, Arg 43, Glu 10, and Asp 53, are highly conserved across the S28E family (Fig. 1 ▶). Interestingly, the C-terminal ~13 residues of S28E (plus the C-tag) are essentially disordered in the structure, and there is little spectroscopic evidence (i.e., NOE patterns, CSI, etc.) for the α-helix that is predicted (Rost 1996) in this region of the molecule (Fig. 2 ▶). Conformational flexibility in this C-terminal region of S28E is also indicated by ¹H-¹⁵N heteronuclear NOE data (Fig. 2 ▶). Small ³J(H^N-H^α) scalar coupling constants were measured throughout the C-terminal polypeptide segment, suggesting the possible formation of a transient α-helix. Interestingly, this unstructured C-terminal region features a PROSITE (Falquet et al. 2002) sequence, E-[S/T]-E-R-E-A-R-x-[L/I], that is a signature of the S28E family of proteins.

Figure 3.

NMR solution structure of S28E determined using AutoStructure and DYANA. For the purposes of display, the disordered C-terminal segment and His-tag have been omitted, and only residues 1 through 60 are shown. (A) Stereoview showing the backbone atom superposition of 10 conformers representing the solution structure; β-strand elements are shown in green. (B) Stereoview of a ribbon representation of a representative conformer (lowest DYANA target function) from the ensemble. The four β-strands comprising the Greek-key motif are labeled. (C) Electrostatic potential surfaces (Nicholls et al. 1991) showing the “positive” (blue) and “negative” (red) faces of S28E. (D) Ribbon diagrams of representative models from the solution structures of P. horikoshii S28E (1NY4; residues 1 to 60), E. coli CspA (3MEF), and M. thermoautotrophicum S28E (1NE3; residues 1 to 58). The superpositions (A) were made using the program MOLMOL (Koradi et al. 1996) and ribbon diagrams (B and D), were generated by the program Ribbons (Carson 1991).

Table 1.

Summary of NMR data and structural statistics for S28E

Completeness of resonance assignmentsa
Assignable backbone (%)	98
Assignable side chain (%)	92
	3D 15N	3D 13C	4D 13C/13C
NOESY spectral datab
Total number of peaks	747	1462	920
Number of “assignable” peaks	744	1352	910
Number of peak assignments	618	1198	771
Percent assignment (%)	83	89	85
Conformationally restricting constraints
Distance constraints
Total		828
Intraresidue (i = j)		121
Sequential (|i − j| = 1)		268
Medium range (1 < |i − j| ≤ 5)		97
Long range (|i − j| > 5)		342
Distance constraints per residue		11.7
Dihedral angle constraints		69
Hydrogen bond constraints
Total		30
Long range (|i − j| > 5)		28
Number of constraints per residue		13.1
Number of long range constraints per residue		5.2
Residual constraint violationsc
DYANA target function (Å2)		0.48 ± 0.067
Average number of distance violations per structure
0.1–0.2 Å		6.5
0.2–0.5 Å		2.4
>0.5 Å		0
Average r.m.s. distance violation per constraint (Å)		0.02
Maximum distance violation (Å)		0.28
Average number of dihedral angle violations per structure
0–10°		0.7
>10°		0
Average r.m.s. dihedral angle violation per constraint (°)		0.13
Maximum dihedral angle violation (°)		1.30
van der Waals violations per structure
Average van der Waals violation (Å)		0.10
Maximum van der Waals violation (Å)		0.12
r.m.s.d. from average coordinates; ordered residuesd
Backbone atoms		0.3
Heavy atoms		1.1
Ramachandran plot statistics; ordered residuesd
Most favored regions (%)		82.0
Additional allowed regions (%)		18.0
Generously allowed (%)		0.0
Disallowed regions (%)		0.0
r.m.s. deviations from ideal geometry
Bond length (Å)	0.011
Bond angle (°)	1.7

Structural statistics were compared for the 10 structures with lowest DYANA target function out of 56 calculated in the final cycle of AutoStructure analysis.

^a Computed using AVS software (H.N.B. Moseley, G. Sahota, and G.T. Montelione, in prep.) from the expected number of peaks, excluding highly exchangeable protons (N-terminal, Lys, and Arg amino groups, hydroxyls of Ser, Thr, Tyr), carboxyls of Asp and Glu, and nonprotonated aromatic carbons.

^b Peak assignment statistics for the 3D ¹⁵N, combined aliphatic and aromatic 3D ¹³C, and 4D ¹³C/¹³C NOESY spectra from the final cycle of AutoStructure. The total number of peaks refer to the raw input to AutoStructure; the number of “assignable” peaks refer to peaks that can be matched to the chemical shift list within the tolerances used; the number of peak assignments represents the total (unambiguous + ambiguous) number of peaks assigned by AutoStructure.

^c Average distance violations were calculated using the sum over r⁻⁶.

^d Ordered residue ranges: 7–17, 24–33, 36–47, 50–56.

Comparison to related structures

A Dali (Holm and Sander 1993) search of the Protein Data Bank (PDB) revealed that the structure of S28E is distantly related to the classic OB-fold (βββαββ) found in many oligosaccharide and nucleic acid binding proteins (Murzin 1993; Arcus 2002), although it lacks two structural elements commonly seen in OB-folds, the α-helix between β-strands 3 and 4 and the fifth β-strand. The closest structural relative in the PDB is the 70-residue CspA cold shock protein from Escherichia coli, whose structure has been solved by both X-ray crystallography (Schindelin et al. 1994; 1MJC; Dali Z score = 4.4) and NMR (Newkirk et al. 1994; Feng et al. 1998; 3MEF). In spite of the very low sequence identity between S28E and CspA (11%), the four β-strands of S28E superimpose well with the first four strands of the OB-fold of CspA, including the β-bulge at Ile 12 in β-strand 1. However, instead of a fifth β-strand folding back to close the β-barrel, as in the CspA structure, the C terminus of S28E juts away from the core of the molecule (Fig. 3D ▶). Like CspA, S28E lacks an α-helical segment between β-strands 3 and 4, typical of other OB-fold proteins. The positively charged surface of S28E is similar to the positively charged nucleic-acid-binding epitope of CspA, formed by conserved basic and surface aromatic residues predominantly from β-strands 1, 2, and 3. However, in contrast to CspA, S28E, and indeed the entire S28E family, is largely devoid of aromatic residues, which play a key role in nucleic acid binding by CspA (Schröder et al. 1995). Other proteins containing OB-fold motifs structurally similar to S28E include the laminin-binding domain of agrin (Stetefeld et al. 2001; 1J7C), the maltose transport protein MalK (Diederichs et al. 2000; 1G29), and the molybdate-binding protein ModG (Delarbre et al. 2001; 1H9J).

The OB-fold is employed for RNA binding by several proteins, including ribosomal proteins S1, S12, and S17 (Draper and Reynoldo 1999; Brodersen et al. 2002). Interestingly, in the crystal structure of the 30S subunit from Thermus thermophilus, ribosomal protein S17 (104 amino acids) features a long C-terminal α-helix following the five-stranded β-barrel, with residues in loops as well as in this α-helix making contacts with specific RNA elements at the interface between the 5′ and central domains of the 30S subunit (Wimberly et al. 2000; Brodersen et al. 2002). The unstructured C-terminal segments of S28E and TT744, which are predicted to form α-helices (Rost 1996), may play a similar role in RNA binding, becoming ordered only upon making specific interactions with rRNA substrate and/or other proteins.

The solution NMR structure of S28E is also very similar to that of the related S28E protein from Methanobacterium thermoautotrophicum (NESG TT744) described in the accompanying paper (Wu et al. 2003); the proteins share 64% sequence identity. The ordered regions of these homologs are highly superimposable, with a pairwise backbone (N, C′, C^α) root-mean-square deviation of 0.72 ± 0.14 Å over the ensembles (Fig. 3D ▶). It is interesting to note that these two solution NMR structures were derived using different automated approaches for determining NOESY cross-peak assignments. In both proteins the highly conserved C-terminal region predicted (Rost 1996) to be α-helical is disordered, and it is reasonable to predict that it may become structured upon binding to other proteins and/or rRNA in the ribosome.

Materials and methods

Sample preparation

The full-length S28E gene (RPS28E) from P. horikoshii was cloned into a pET21d (Novagen) derivative, generating plasmid pJR19–21. The resulting S28E open reading frame contains an additional 11 nonnative residues (AAALEHHHHHH) at the C terminus of the protein. This construct sequence was verified by standard DNA sequence analysis. E. coli BL21 (DE3) pMGK cells, a rare codon-enhanced strain, were transformed with pJR19–21. A single isolate was cultured in MJ9 minimal medium (Jansson et al. 1996), containing 100% ¹⁵N-enriched ammonium sulphate and either 5% or 100% uniformly enriched ¹³C-glucose as the sole nitrogen and carbon sources, respectively. Initial growth was carried out at 37°C until the O.D.₆₀₀ of the culture reached 1.0 units. The incubation temperature was then decreased to 17°C and protein expression was induced by the addition of isopropyl-β-D-thiogalactopyranoside at a final concentration of 1 mM. Following overnight incubation at 17°C, the cells were harvested by centrifugation.

S28E protein samples were purified using standard protocols. Cell pellets were resuspended in Lysis Buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, 5 mM β-mercaptoethanol at pH 8.0) and disrupted by sonication. The resulting lysate was clarified by centrifugation at 26,000g for 45 min at 4°C. The supernatant was loaded onto a Ni-NTA column (Qiagen) and eluted in Lysis Buffer containing 250 mM imidazole. Fractions containing partially purified S28E were pooled and loaded onto a gel filtration column (Sephadex 75, Amersham Pharmacia Biotech), and eluted in 10 mM Tris, 100 mM NaCl, and 5 mM dithiothreitol (DTT) at pH 8.0. Sample purity (>97%) and molecular weight (9718 D) were verified by SDS-PAGE and MALDI-TOF mass spectrometry, respectively. The yield of purified protein was approximately 20 mg/L. Protein samples for NMR spectroscopy were concentrated by ultracentrifugation to 1.0 mM S28E in 20 mM MES, 100 mM NaCl, 5 mM CaCl₂, 10 mM DTT (pH 6.5), containing 5% (v/v) ²H₂O (unless otherwise indicated).

NMR spectroscopy

NMR spectra were acquired at 20°C on 500, 600, 750, and 800 Varian Inova spectrometers and a 600 Varian Unity spectrometer. Chemical shifts are referenced to external DSS, whereas ¹³C and ¹⁵N chemical shifts were referenced indirectly using the gyromagnetic ratios of ¹³C:¹H (0.251449530) and ¹⁵N:¹H (0.101329118), respectively. Backbone and side chain resonance assignments were made using the following series of standard triple-resonance experiments (Cavanaugh et al. 1996; Montelione et al. 1999) obtained on U-¹³C,¹⁵N S28E: 2D ¹⁵N-¹H HSQC, 2D ¹³C-¹H HSQC, 3D HNCO, HNCACB, CBCA(CO)NH, HA(CA)(CO)NH, HA(CA)NH, CBCACO(CA)HA, (H)CC(CO)NH-TOCSY, and H(CC)(CO)NH-TOCSY, along with HCCH-TOCSY and HCCH-COSY spectra recorded on a sample prepared in 100% ²H₂O. NOE distance constraints were derived from 3D ¹⁵N-edited NOESY (τ_m = 80 ms), 3D ¹³C-edited aliphatic and aromatic NOESYs (τ_m = 80 ms), and 4D ¹³C/¹³C -NOESY in 100% ²H₂O (τ_m = 100 ms). Three-bond ³J(H^N-H^α) scalar couplings were obtained from the ratio of cross-peak to diagonal intensities in a 3D HNHA (Vuister and Bax 1993). Amide proton exchange rate information was obtained by a 1:6 dilution of 50 μL U-¹³C,¹⁵N S28E with 100% ²H₂O and monitoring the decay of the ¹⁵N-¹H HSQC signal height over time; some amide resonances in the core of the protein were still present after several days. Stereospecific Val and Leu methyl assignments were obtained from a high-resolution nonconstant time ¹³C-¹H HSQC spectrum of 5%-¹³C, U-¹⁵N S28E (Neri et al. 1989). ¹H-¹⁵N heteronuclear NOEs were obtained on 5%-¹³C, U-¹⁵N S28E using a sensitivity-enhanced version of our previously described 2D heteronuclear NOE experiment (Li and Montelione 1993), with a recycle time of 14 sec to ensure complete relaxation of solvent magnetization. All NMR spectra were processed with NMRPipe 2.1 (Delaglio et al. 1995) and analyzed with Sparky 3.106 (Goddard and Kneller 2002).

Resonance assignments

Backbone H^N, N, C′, C^α, H^α, and side chain C^β resonance assignments were obtained using AutoAssign 1.9 (Zimmerman et al. 1997; Moseley et al. 2001), employing peak lists from the ¹⁵N-¹H HSQC, HNCO, HNCACB, CBCA(CO)NH, HA(CA)(CO)NH, and HA(CA)NH spectra. Sequential STACs information (Zheng et al. 2003), obtained from 2D (H^N-N plane) Gly-phased HA(CACO)NH (Feng et al. 1996) and (H)CC(CO)NH-TOCSY data, were included in the AutoAssign analysis. These assignments were extended into the side chains by manual analysis of (H)CC(CO)NH-TOCSY, H(CC)(CO)NH-TOCSY, HCCH-TOCSY, HCCH-COSY, and ¹³C-edited NOESY spectra. Resonance assignments were validated using the Assignment Validation Suite (AVS) software package (H.N.B. Moseley, G. Sahota, and G.T. Montelione, in prep.).

Automated structure determination

Structure calculations were performed using the program AutoStructure 1.1.2 (Huang 2001; Huang et al. 2003; Zheng et al. 2003), interfaced with DYANA 1.5 (Güntert et al. 1997). Briefly, AutoStructure performs iterative NOESY assignment and structure calculations using DYANA, XPLOR, or CNS (Brünger et al. 1998) in an automated fashion for a user-defined number of cycles. After generating a reliable initial protein fold based in the first cycle of NOESY spectral analysis, AutoStructure automatically generates distance (NOE), dihedral angle (using the program HYPER; Tejero et al. 1999), and hydrogen-bond constraints in successive cycles. In this case, the input for the AutoStructure program consisted of a resonance assignment list, manually edited peak lists with intensities for the 3D ¹⁵N-edited, 3D ¹³C-edited, and 4D ¹³C/¹³C NOESY spectra, ³J(H^N-H^α) values, φ,ξ angle constraints (±40° and ±50°, respectively) derived from chemical shift data using the program TALOS (Cornilescu et al. 1999), and slow amide hydrogen exchange data. TALOS dihedral constraints were used only for residues with confidence scores of 10. Tolerances for matching NOESY peaks with resonance assignments within the NOESY assign module of the program were set to ±0.05 ppm for ¹H (±0.04 ppm for the directly detected dimension in the 3D NOESY spectra) and ±0.5 ppm for ¹³C and ¹⁵N, and the program automatically handles aliased peaks. The reported ensemble of structures comprises the best 10 of 56 structures from the final cycle of AutoStructure, on the basis of DYANA target function. Structures were analyzed using the programs PDBStat (R. Tejero and G.T. Montelione, unpubl.) and PROCHECK-NMR (Laskowski et al. 1996). The final ensemble of structures (minus the C-terminal tag) and structural constraints, as well as the chemical shift, ³J(H^N-H^α) data, and raw fids have been deposited in the Protein Data Bank (PDB ID 1NY4) and BioMagRes Bank (accession number 5691), respectively.