Solution structure of ribosomal protein L40E, a unique C4 zinc finger protein encoded by archaeon Sulfolobus solfataricus

Abstract

The ribosomal protein L40E from archaeon Sulfolobus solfataricus is a component of the 50S ribosomal subunit. L40E is a 56-residue, highly basic protein that contains a C4 zinc finger motif, CRKC_X₁₀_CRRC. Homologs are found in both archaea and eukaryotes but are not present in bacteria. Eukaryotic genomes encode L40E as a ubiquitin-fusion protein. L40E was absent from the crystal structure of euryarchaeota 50S ribosomal subunit. Here we report the three-dimensional solution structure of L40E by NMR spectroscopy. The structure of L40E is a three-stranded β-sheet with a simple β2β1β3 topology. There are two unique characteristics revealed by the structure. First, a large and ordered β2–β3 loop twists to pack across the one side of the protein. L40E contains a buried polar cluster comprising Lys19, Lys20, Cys22, Asn29, and Cys36. Second, the surface of L40E is almost entirely positively charged. Ten conserved basic residues are positioned on the two sides of the surface. It is likely that binding of zinc is essential in stabilizing the tertiary structure of L40E to act as a scaffold to create a broad positively charged surface for RNA and/or protein recognition.

Sulfolobus solfataricus is a thermoacidophilic archaeon that grows optimally at 80°C and over a pH range of 2 to 4 (Zillig et al. 1980). As a model organism for the domain of crenarchaeotes, its information-processing pathways—including DNA replication, transcription, and translation—have been studied extensively (Bell and Jackson 1998, 2001). The ribosome as a complex and dynamic cellular machine responsible for protein synthesis has attracted growing attention (Ramakrishnam 2002). The genome sequencing of the S. solfataricus P2 revealed 68 ribosomal proteins encoded within the genome (28 and 37 proteins in the 30S and 50S subunits, respectively). As with all archaea, these proteins are more similar to their eukaryotic homologs than to their bacterial counterparts (She et al. 2001). The 50S ribosomal protein, L40E, was selected by the Northeast Structural Genomics Consortium (http://www.nesg.org) as a target for NMR study (NESG ID: SST91). The 56-residue L40E protein from S. solfataricus is conserved in all sequenced archaeal genomes but is not present in bacteria. Homologs of L40E are found in yeast, plants, parasites, and mammals. L40E was identified in eukaryotic genomes as a hybrid protein in which ubiquitin is fused at its C-terminus to L40E (Ozkaynak et al. 1987; Chen et al. 1995; Redman and Burris 1996; Kato et al. 2001). Rat L40E was reported to be methylated at Lys22 (Williamson et al. 1997). S. solfataricus L40E and its homologs are highly basic and characterized by the presence of a conserved C4 zinc finger motif, CX₂C_X₁₀_CX_2–4C, for which no structural models currently exist. Surprisingly, there is little information on L40E and its role in the ribosome. L40E was not modeled in the crystal structure of the 50S ribosomal subunit from euryarchaeota Haloarcula marismortui (Ban et al. 2000) which shares 43% sequence identity to the S. solfataricus L40E.

Here we report the solution structure of the ribosomal protein L40E from S. solfataricus. The structural characteristics of L40E, particularly its unusual fold and buried polar residue packing feature, are investigated, and its putative functional roles in the 50S ribosomal subunit are discussed.

Results and Discussion

S. solfataricus L40E is a zinc-binding protein

The S. solfataricus L40E possesses a C4-type zinc-binding sequence of CRKC_X₁₀_CRRC. The four putative metal-binding Cys residues are completely conserved among all members of the L40E family (Fig. 1A). More interestingly, eukaryotic L40E has two basic residue insertions between Cys36 and Cys39.

Figure 1.

(A) Sequence alignment of L40E from Crenarchaeota (top), Euryarchaeota (middle), and eukaryota (bottom) genomes. Crenarchaeota: Sulfolobus solfataricus (gi:48428673), Sulfolobus acidocaldarius (gi:68567210), Sulfolobus tokodaii (gi:15920384), Aeropyrum pernix (gi:6685882), Pyrobaculum aerophilum (gi:18313942). Euryarchaeota: Methanosarcina acetivorans (gi:20092765), Pyrococcus horikoshii (gi:3257728), Pyrococcus furiosus (gi:18977847), Methanothermobacter thermautotrophicus (gi:3122754). Eukaryota: Saccharomyces cerevisiae (gi:133107), Leishmania tarentolae (gi:312488), Tetrahymena pyriformis (gi:417679), Entamoeba histolytica (gi:67471049), Homo sapiens (gi:51702821). Identical and similar residues are highlighted in black and gray, respectively. The NMR-derived secondary structural elements of L40E are illustrated above the alignment. (B) Ribbon representation of the lowest-energy structure of L40E. (C) Ensemble of 20 refined structures represented in an orientation similar to (B). (D) Zinc-binding site in L40E. A magenta sphere indicates zinc ion. The four Cys residues whose side chains coordinate the zinc ion are shown by a stick. (E) Stereoview of an ensemble expansion showing conserved structure core in the protein (residues 15–48). Hydrophobic residues Val21, Ala27, Pro30, Ile31, Ala33, and Leu44 are colored blue. Polar residues Cys22, Asn29, and Cys36 are shown in red. Charged positive residues Lys19 and Lys20 are in green. Zinc ion is shown in magenta.

S. solfataricus L40E was expressed in Escherichia coli using M9 minimal medium supplemented with ZnSO₄ and other trace metals. Likewise, all the buffers used in the preparation of the NMR sample also included zinc. Metal analysis was performed using inductively coupled plasma optical emissions spectrometry (ICP-OES). The result showed that the protein fraction contained 39 times more zinc ion concentration than the buffer fraction, suggesting that the L40E sequestered the zinc from the buffer. All other metals tested showed no detectable difference between the protein and the buffer fraction. This result confirmed that L40E preferentially bound zinc.

The presence of a bound zinc ion was further supported by Cys chemical shift and structure calculations. The ¹³Cα and ¹³Cβ chemical shifts of the four Cys residues are typical for those zinc-coordinated cysteines with probabilities 0.98, 0.95, 0.95, and 0.96 for Cys22, Cys25, Cys36, and Cys39, respectively (Kornhaber et al. 2006). Structure calculations without zinc distance restraints revealed that the overall protein fold was identical to those calculated with zinc restraints. The side chains of the four Cys residues thought to be involved in zinc binding were in the same orientation, regardless of whether zinc restraints were present. Taking these results together, we conclude that L40E is stabilized by the coordination of the four conserved Cys residues to zinc ion.

Overall fold

The solution structure of L40E is represented in Figure 1, B and C, and the structure parameters are summarized in Table 1. The ensemble of NMR-derived structures shows a well-ordered core consisting of a three-stranded antiparallel β-sheet with a loop packed against the β-sheet side. The strands (residues 20–22, 27–29, and 44–46) are arranged with a β2β1β3 topology. The nine N-terminal residues are less well-ordered, and only a few sequential NOEs were detected. Residues 13–15 were not detected in the ¹⁵N-HSQC spectrum, and assignments of Gln13–Lys15 were missing, likely due to line broadening arising from conformational change on an intermediate timescale. Thus, residues 12–16 are poorly defined. The seven residues at the C terminus (residues 49–56) are completed disordered, which showed no NOEs with the rest of the protein or whose assignments for residues 54–56 were not available.

Table 1.

Structural statistics for the ensemble calculated for L40E^a

The overall fold of L40E is largely defined by the zinc ion and β-sheet. Four conserved Cys residues tetrahedrally coordinate to a zinc ion, thereby stabilizing the tertiary structure of the protein. Residue Cys22 lies at the end of strand β1 and Cys25 is located in a rubredoxin-like turn connecting strands β1 and β2. The other two zinc-coordinating Cys residues, Cys36 and Cys39, reside within the β2-β3 loop. Additional stabilization is achieved by hydrophobic interactions and hydrogen bonds within the core of the protein. Residues with a low accessible surface area (<15%) include Gln10, Ile11, Lys19, Lys20, Val21, Cys22, Ala27, Asn29, Pro30, Ile31, Ala33, Cys36, and Leu44 (Fig. 1E). All these residues except Gln10 and Ile11 are well conserved throughout the entire L40E family.

One of the most unusual features of L40E is the large fairly well-defined β2-β3 loop (residues 30–43) which twists to pack across one side of the protein and includes residues with dihedral angles φ and ψ in both α and β space. The loop interacts with the β-sheet through extensive hydrophobic contacts. The side chains of residues Lys20 and Cys22 in strand β1, Ala27 and Asn29 in strand β2, and Leu44 in strand β3 cluster together with side chains of Ile31, Ala33, Cys36, Arg37, and Arg38 in the β2-β3 loop. Most prominent are long-range NOEs involving conserved polar residues Lys19, Lys20, Cys22, Asn29, Cys36, Arg37, and Arg38. The aliphatic moieties of Lys19 and Lys20 participate in a hydrophobic core that includes Leu28, Ile31, Ala33, Leu44, and Leu46. The buried Asn29 located in strand β2 is a key structural residue in linking the β-sheet to the loop. The side chain of Asn29 is well packed against Lys20 in strand β1, those of Ala33, Cys36, Arg37, and Arg 38 in the β2-β3 loop, and that of Leu44 in strand β3. Overall, these interactions are critical to ensure the unique orientation of the β2-β3 loop with respect to the β-sheet and to complete the optimal folding of the protein around the zinc ion.

L40E structurally resembles Rsgi Ruh-035

Among known zinc finger structures, the topology of L40E differs significantly from those of classical zinc finger proteins which adopt a ββα fold (Wolfe et al. 2001) and also from the three-stranded zinc ribbon module of TFIIS (Qian et al. 1993; Olmsted et al. 1998), TFIIB (Zhu et al. 1996; Chen et al. 2000), and rubredoxin (Schweimer et al. 2000).

The Protein Data Bank was searched for structures similar to L40E via the DALI server (Holm and Sander 1993) using either the entire L40E or just the well-structured core (residues 15–48). One significant and two weaker structural homologs were found with Z scores of 3.9, 2.1, and 2.1, respectively. The best match was obtained for Rsgi Ruh-035 (PDB ID: 2CON, Z score 3.9) (Y. Doi-Katayama, Y. Hirota, T. Tomizawa, S. Koshiba, T. Kigawa, and S. Yokoyama, in prep.). Rsgi Ruh-035 from mouse is a newly predicted RNA binding protein which contains a characteristic CXXC-X₁₁-CXXC motif. The tertiary topology of L40E (residues 15–48) and Rsgi Ruh-035 (residues 12–46) is strikingly similar with an RMSD of 1.6 Å over 33 aligned residues (Fig. 2A), though sequence identity among the structurally equivalent residues is only 15%. As with L40E, Rsgi Ruh-035 has a three-stranded antiparallel β-sheet with a long loop which packs onto the opposite face of the β-sheet. In addition, the positions of the four Cys residues which coordinate the zinc ion in Rsgi Ruh-035 appear to be equivalent to those of Cys residues in L40E. Though Rsgi Ruh-035 is less charged than L40E whose basic resides spread over all the surface, the pattern of the basic residues displayed on the surface is quite similar (Fig. 2B). These data suggest that L40E and Rsgi Ruh-035 may adopt the same nucleic acid binding mode and have a possible evolutionary relationship.

Figure 2.

(A) Ribbon diagram depicting L40E (residues 15–50), the Rsgi Ruh-035 (residues 12–48, PDB ID: 2CON), and human MAT1 ring finger domain (PDB ID: 1G25). The β-sheet and helices are shown in blue and red, respectively. Zinc ions are shown as magenta spheres. (B) Surface electrostatic potential of L40E and the Rsgi Ruh-035 as calculated by MOLMOL. Blue and red colors represent positive and negative electrostatic potential, respectively. The left view is the same orientation as in A. The right view is rotated by +180° along the Y-axis.

The server found no significant similarity with known zinc ribbon folds (Krishna et al. 2003). However, an unexpected weaker related structure was identified as the human MAT1 RING finger domain (PDB ID: 1G25; Z score 2.1) (Gervais et al. 2001). The MAT1 Ring finger domain adopts a classical ββαβ topology with a cross-brace arrangement of the eight zinc-binding ligands. The similarity occurs in the region of the three-stranded antiparallel β-sheet which is involved in the second zinc binding (Fig. 2A). Although L40E and the MAT1 RING finger domain share the same topology (an RMSD for 44 Cα atoms is 3.8 Å), the β2-β3 loop varies in length. In addition, the two structures differ in the loop orientation relative to the β-sheet. A large insertion which forms an α-helix and comprises the two zinc ligands, Cys31 and Cys34, of the first zinc-binding site is present in the ring finger domain loop.

Implication for L40E function

L40E is an essential protein required in the large ribosomal subunit. Its role in ribosome function remains unknown, but its conservation across the archaea and eukaryotes suggest that it is likely important. The NMR structure shows that L40E has a simple β2β1β3 fold and uses a zinc ion to maintain its structural integrity. L40E represents a novel member of the growing and diverse family of zinc finger proteins that have a β-strand fold as an element of their structure (Laity et al. 2001; Krishna et al. 2003). An antiparallel β-sheet is frequently found in RNA-binding proteins (Brown 2005; Chen and Varani 2005). Furthermore, the distribution of the electrostatic surface potential of L40E reveals the presence of clustered basic residues typical for RNA recognition. L40E is a highly basic protein with 30% of the total residues being Lys and Arg. The surface of the molecule core (residues 15–50) is almost entirely positively charged except Glu50 at the C termini. A total of 15 out of 17 basic residues cluster on the two sides of the surface (Fig. 2B). Seven of these basic residues, including Lys24, Arg32, Arg37, Arg38, Arg45, Arg47, and Lys49, are strictly conserved in L40E proteins. Residues Lys23 and Lys48 are completely conserved in crenarchaeota and eukaryota, whereas in euryarchaeota, Lys23 is replaced by Met or Leu and Lys49 by Ala or Ser. Residues Arg15 and Lys35 are found only in archaea. Therefore, the positively charged surface shows a remarkably high conservation, indicating its importance in the biological function of L40E. The role of zinc ion is likely to be structural, bridging three strands of the β-sheet with the β2-β3 loop to create a broad positively charged surface for RNA binding.

A number of ribosomal proteins contain a Cys-rich sequence motif and function as a zinc finger protein (Hoof et al. 1992; Chen et al. 1993, 1995; Kondoh et al. 1996; Tsiboli et al. 1998; Boysen and Hearn 2001; Makarova et al. 2001; Nanamiya et al. 2004). The H. marismortui 50S zinc finger ribosomal proteins L24E, L37E, L37Ae, and L44E were clearly visualized in its crystal structure (Ban et al. 2000; Klein et al. 2004). L37E, L37Ae, and L44E along with bacterial L36 (Hard et al. 2000; Harms et al. 2001) all fold into a central β-sheet similar to the zinc ribbon fold, even though they share no primary sequence homolog, and differ in overall structure. Intriguingly, though these proteins possess a CX₂C-X_11–56-CX₂C (CX₂C-X₁₂-CX₄H in L36) sequence motif resembling that of L40E and are structurally related to the β-sheet, their roles in 50S ribosomal subunit are quite different. L37E and L44E have the most broad contacts with RNA in the H. marismortui 50S subunit. The zinc ribbon fold of L37Ae, on the other hand, acts as one of the largest 50S ribosomal protein recognition motifs. L37Ae, which makes minor protein–RNA contacts, packs extensively against L2 (Klein et al. 2004). Bacterial L36 is essential for specific RNA recognition. It has been reported that L36 disrupts the network of tertiary interactions that are responsible for the proper folding of 23S RNA by the removal of L36 from the ribosome (Maeder and Draper 2005). The large subunit proteins function primarily to organize and stabilize the RNA tertiary structure. The fusion of eukaryotic L40E with ubiquitin and the methylation of rat L40E imply that L40E may have an important regulatory role in ribosome assembly (Finley et al. 1989; Williamson et al. 1997). As the structure of L40E is distinct significantly from any of these characterized zinc finger ribosomal proteins, it will be of great interest to elucidate the nature of L40E implicated in RNA and/or protein binding.

In conclusion, we have used NMR spectroscopy to determine the first three-dimensional structure of 50S ribosomal protein L40E among L40E family. L40E is characterized not only by a C4 zinc finger motif but also by an unusual protein fold consisting of an antiparallel β-sheet with a β2β1β3 topology and a long loop containing two of the four Cys residues coordinated to the zinc ion. The structure presented here will provide the basis for the design of the further studies to clarify the function of the L40E family of proteins in ribosome.

Materials and Methods

Protein purification

The gene sso5336 encoding the ribosome protein L40E (56 amino acids) from S. solfataricus was subcloned into the pET-11 expression vector with an N-terminal His tag. It was expressed in E. coli strain BL21-gold (DE3) growing in 2× M9-minimal medium supplemented with zinc sulfate, biotin, ¹⁵N- ammonium chloride, and ¹³C-glucose. The protein was purified to homogeneity using metal affinity chromatography as described previously (Yee et al. 2002). The purified protein contained the complete sequence of L40E plus His₆ affinity tag (MGTSHHHHHHSSGRENLYFQGH) at the N terminus of the protein. The concentration of protein samples ranged from 1.0 to 1.5 mM in an aqueous solution containing 10 mM MOPS (pH 6.5), 450 mM NaCl, ∼10 μM ZnSO₄, 10 mM DTT, 1 mM benzamidine, 0.01% NaN₃, 1× inhibitor cocktail, and 95% H₂O/5% D₂O.

Metal analysis using inductively coupled plasma optical emissions spectrometer (ICP-OES)

The L40E sample was analyzed by ICP-OES (PerkinElmer 3000 DV) for Mg, Al, Ca, Ti, Cr, Mn, Fe, Co, Ni, Cu, Zn, and Pb; 200 μL of the sample was centrifuged using a Vivaspin 5000-MWCO PES concentrator (Vivascience) to separate out the protein and any protein-associated metals from the buffer. An aliquot of the buffer flowthrough was used as a control. Both protein and control samples were prepared for ICP-OES by digestion in 2% nitric acid.

NMR spectroscopy

All NMR spectra were collected at 25°C on Bruker Avance 600-MHz instrument equipped with a z-shielded gradient triple-resonance cryoprobe. Chemical shifts were referenced to external DSS. Spectra were processed using the program NMRPipe (Delaglio et al. 1995) and analyzed with the program SPARKY (http://cgl.ucsf.edu/home/sparky). The backbone assignments were obtained using HNCO, HN(CA)CO, CBCA(CO)NH, HNCB, HNCA, HN(CO)CA, HNHA, and ¹⁵N-edited NOESY-HSQC spectra. Assignments were made initially with the automated program MONTE (Hitchens et al. 2003) and followed by manual analysis. Aliphatic side-chain assignments relied on HBHA(CO)NH, (H)CC(CO)NH-TOCSY, H(CC)(CO)NH-TOCSY, and H(C)CH-COSY spectra (Bax et al. 1994; Kay 1997).

Structure calculation

Distance restraints for structure calculations were derived from cross-peaks in ¹⁵N-edited NOESY-HSQC (τ_m = 150 ms), ¹³C-edited aliphatic NOESY-HSQC in H₂O (τ_m = 150 ms) and D₂O (τ_m = 100 ms), respectively. NOE peaks were picked with intensities using the program SPARKY. Automated NOE assignment and structure calculations were performed using the NOEASSIGN module implemented in the program CYANA, version 2.1 (Güntert 2004). The NOEASSIGN/CYANA input included the chemical shift list, peak lists from ¹⁵N-edited NOESY, ¹³C-edited NOESY in H₂O, and ¹³C-edited NOESY in D₂O, and dihedral angle restraints derived from the program TALOS. In addition, the zinc ion was included in the CYANA library. Distance restraints from zinc ion to the putative zinc ligands were introduced as pseudo-NOEs with a Cys Sγ–Zn distance of 2.25–2.35 Å and a Cys Cβ-Zn distance of 3.28–3.38 Å. Upper and lower distance limits were also set to 3.71–3.93 Å for Cys Sγ–Cys Sγ (Hammarstrom et al. 1996). Seven iterative cycles of automated CYANA structure calculation were applied. A total of 93% of the NOE cross peaks from ¹⁵N-edited NOESY and ¹³C-edited NOESY were assigned in cycle 7. In the final NOEASSIGN cycle, CYANA automatically analyzed methylene protons and (isopropyl)methyl groups to identify 40 stereospecific assignments which were used in the final round of structure calculation. The quality of NOEASSIGN/CYANA calculation was assessed by NMR structure quality assessment scores (NMR PRF scores) (Huang et al. 2005). The best 20 of 100 CYANA structures from the final cycle were selected and subjected to molecular dynamics simulation in explicit water by the program CNS (Brunger et al. 1998; Linge et al. 2003). The structures were inspected by PROCHECK (Laskowski et al. 1996) and MolProbity (Lovell et al. 2003) using the NESG validation software package PSVS (Bhattacharya et al. 2007). Structures were visualized using the program MOLMOL (Koradi et al. 1996).

Accession numbers

The chemical shifts have been submitted to the BioMagResDB (BMRB accession number 6747), and the structure ensemble and restraint files have been deposited to the Protein Data Bank (PDB ID: 2AYJ).