Solution structure of a BolA-like protein from Mus musculus

Abstract

The BolA-like proteins are widely conserved from prokaryotes to eukaryotes. The BolA-like proteins seem to be involved in cell proliferation or cell-cycle regulation, but the molecular function is still unknown. Here we determined the structure of a mouse BolA-like protein. The overall topology is αββααβα, in which β₁ and β₂ are antiparallel, and β₃ is parallel to β₂. This fold is similar to the class II KH fold, except for the absence of the GXXG loop, which is well conserved in the KH fold. The conserved residues in the BolA-like proteins are assembled on the one side of the protein.

Escherichia coli BolA and its homologs constitute a widely conserved protein family, called the BolA-like protein family. BolA-like proteins are found in most bacteria and eukaryotes, and in an archaeon. The E. coli bolA gene was isolated as the morphogene induced in stationary phase (Aldea et al. 1988, 1989, 1990; Lange and Hengge-Aronis 1991), and was also found to be induced by various stresses (Santos et al. 1999). The mutation or the overexpression of the bolA gene affects the transcription level of the penicillin binding protein (PBP) genes, which are involved in cell wall biosynthesis (Aldea et al. 1989; Santos et al. 2002). In Schizosaccharomyces pombe, a bolA homolog, uvi31⁺, which was isolated as a UV-inducible gene, shows cell-cycle and growth phase-dependent expression (Lee et al. 1994; Kim et al. 1997). In addition, the deletion mutant of uvi31 proliferates faster than the wild-type cell, and shows abnormal septation after UV-induced cell cycle arrest (Kim et al. 2002). Although these results imply that BolA-like proteins play some roles in cell proliferation by controlling the transcription of other genes, their molecular functions remain unknown.

Most of the BolA-like proteins consist of about 100 amino acid residues, as do the three BolA-like proteins found within the mouse full-length cDNA libraries, FANTOM and FANTOM2 (Kawai et al. 2001; Okazaki et al. 2002). We designated the mouse BolA-like proteins as BolA1 (gi|12841442), BolA2 (gi|26325949), and BolA3 (gi|26389531). Among the three proteins, BolA1 shows the highest homology to E. coli BolA. Close homologs of the mouse BolA2 were found in most eukaryotes, including Homo sapiens, Drosophila melanogaster, Caenorhabditis elegans, Arabidopsis thaliana, and Saccharomyces cerevisiae. Therefore, the BolA2 proteins constitute a eukaryotic subfamily of the BolA-like proteins (Fig. 1A ▶). Mouse BolA3 and its homologs compose another eukaryotic subfamily, with lower similarity to E. coli BolA than the BolA2 subfamily.

Figure 1.

(A) Multiple sequence alignment of the BolA2 subfamily. Highlighted and shaded residues indicate identical and similar residues, respectively. Mm: BolA2 from Mus musculus. Hs: My016 from Homo sapiens. Dm: CG16804-PB from Drosophila melanogaster. Ag: agCP4532 from Anopheles gambiae. Sp: SPAC8C9.11 from Schizosaccharomyces pombe. The multiple alignment was achieved with CLUSTAL_X (Thompson et al. 1997). The secondary structures of the mouse BolA2 are indicated at the top of the alignment. (B) Ribbon diagram of BolA2. (C) Class II KH domains of ribosomal protein S3 (top, PDB ID: 1J5E) and Era GTPase (bottom, PDB ID: 1EGA). Each class II KH domain is colored in the rainbow order, from red (N terminus) to blue (C terminus).

In the present study, we determined the solution structure of the mouse BolA2, which is the first tertiary structure of a BolA-like protein family member. The BolA2 protein has the αββααβα topology, which is the same as that of the class II KH fold. However, the BolA-like proteins lack the GXXG sequence, which is involved in binding to nucleic acids in the authentic KH-fold proteins. Instead of the GXXG loop, BolA2 has a turn, which results in a helix-turn-helix (HTH) motif. The surface residues around the HTH motif are well conserved in the BolA2 subfamily.

Results and Discussion

The overall topology of BolA2 is αββααβα, in which β₁ and β₂ are antiparallel, and β₃ is parallel to β₂ (Fig. 1B ▶). A nine-residue loop, inserted between β₁ and β₂, has several amide protons with very low intensities in the ¹H/¹⁵N-HSQC spectrum. The α₄ helix is anchored on one side of the β-sheet, while the other helices are on the other side. A one-turn 3₁₀-like helix exists between α₃ and β₃. The α₂ and α₃ helices form an HTH motif, and are arranged at an angle of about 60° to each other. In E. coli BolA, another HTH motif had been predicted by the Chou-Fasman algorithm (Aldea et al. 1989). However, the corresponding region of BolA2 comprises the β₁/β₂ loop and the β₂ strand.

According to the DALI (Holm and Sander 1993) search, BolA2 shows significant structural similarity to the ribosomal protein S3 N-terminal domain (Wimberly et al. 2000) and the Era GTPase C-terminal domain (Chen et al. 1999), with Z-values of 5.4 and 4.7, respectively, which belong to class II KH fold (Fig. 1C ▶). The KH fold is commonly found among nucleic acid binding proteins, and consists of two classes with distinct topologies, βααββα for class I and αββααβ for class II (Grishin 2001). Most of the KH-fold proteins have a well-conserved GXXG sequence on the loop between two adjacent helices. Some of the class I KH-fold proteins bind to nucleic acids using their GXXG loop (Lewis et al. 2000; Liu et al. 2001; Braddock et al. 2002a,b). The corresponding region of BolA2 forms an HTH motif, and the BolA-like proteins lack the conserved GXXG sequence in this region.

Figure 2A ▶ shows the conserved surface residues of BolA2. Most of the conserved residues are assembled on one side of the protein, especially around the HTH motif. The “conserved” side consists of an electrically neutral region surrounded by several basic residues (Fig. 2B ▶). There are significantly conserved solvent-exposed hydrophobic residues (L50, L55) near the HTH motif. On the other hand, the other side of BolA2 is highly acidic. Some residues are still conserved on the “variable” side of BolA2.

Figure 2.

(A) The conserved residues on the surface of the mouse BolA2. The identical and the similar residues defined in Figure 1A ▶ are colored dark blue and light blue, respectively. The right panel is viewed from the opposite side of the left panel. (B) The surface electrostatic potential of BolA2. (C) Ribbon diagram in the same orientation as A and B. The HTH motif is indicated by the blue circle.

The nucleic-acid binding property of the BolA-like proteins is implicated by the structural similarity to the KH fold and the assembly of conserved surface residues around the HTH motif. However, the existence of some conserved surface residues on the “variable” side and the absence of the GXXG sequence suggest the possibility that the BolA-like proteins have other functions than those of the KH-fold proteins.

Materials and methods

Sample preparation and NMR spectroscopy

The mouse BolA2 protein was produced as the 113-amino-acid recombinant protein with a 27-amino-acid purification tag sequence (MKGSSHHHHH HSSGASLVPR GSEGAAT) at the N terminus of the 86-amino-acid coding sequence. The uniformly ¹³C- and ¹⁵N-labeled protein was produced by the E. coli cell-free protein synthesis system (Kigawa et al. 1999), and then 5.0 mg of the purified protein were obtained by polyhistidine affinity chromatography followed by anion exchange chromatography. All NMR spectra were acquired on a Bruker AVANCE600 spectrometer at 298K, with a 1.2 mM ¹³C/¹⁵N BolA2 protein solution in 50 mM Na-phosphate buffer (pH 6.0) containing 100 mM NaCl, 0.02% NaN₃, 1 mM DTT-d₁₀, and 10% ²H₂O. Spectra were processed and analyzed with NMRPipe (Delaglio et al. 1995) and NMRview (Johnson and Blevins 1994).

Structure determinaiton

The chemical shifts were assigned using ¹H/¹⁵N-HSQC, ¹H/¹³C-HSQC, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, HNCO, HN(CA)CO, HBHA(CO)NH, CC(CO)NH, H(CCCO)NH, HCCH-COSY, HCCH-TOCSY, NOESY-¹H/¹⁵N-HSQC (τ_m = 100 msec), and NOESY-¹H/¹³C-HSQC (τ_m = 100 msec) spectra. In the BolA2 coding region, 97% of the observable proton signals were assigned. The NOESY cross-peaks were assigned with DYANA/CANDID (Herrmann et al. 2002), following manual correction. The total number of assigned NOEs was 1546, but 931 meaningful distance constraints were used for the final structure calculation. The dihedral-angle constraints of the main chain were used in the secondary structure regions predicted by TALOS (Cornilescu et al. 1999; φ = −60° ± 15°, ψ = −45° ± 15° for the helical regions, φ = −120° ± 30°, ψ = 150° ± 30° for the β-strand regions). The side-chain χ angle constraints (180°, 60°, or −60° with a tolerance of ±30°) were used for the residues for which the χ angles could be obviously determined from the NOESY spectra. To obtain hydrogen bond information, ¹H/¹⁵N-HSQC spectra were measured in ²H₂O. Twenty-nine peaks in the α-helix or the β-strand regions were used for the structure calculation, while a total of 37 peaks were observed. Hydrogen bond constraints were used as two distance constraints (3.5 Å for N and O, 2.5 Å for HN and O) for each hydrogen bond. The final structure was calculated with CNS (Brünger et al. 1998). The calculation statistics are summarized in Table 1. The coordinates of the best 20 structures have been deposited in the Protein Data Bank, under the accession ID 1IW5. The ribbon diagrams were drawn with MOLMOL (Koradi et al. 1996). The molecular surface presentations were generated by GRASP (Nicholls et al. 1991).

Table 1.

Calculation statistics for the ensemble of 20 structures

NOE distance constraints
All	931
Intraresidual	467
Sequential	214
Medium range (2 ≤ |i − j| ≤ 4)	97
Long range (|i − j| > 4)	157
Hydrogen bond constraints	29
Dihedral angle constraints
φ	54
ψ	54
χ	25
Number of violations
Distance violations (>0.2 Å)	0
Dihedral angle violations (>2°)	0
Coordinate RMSD (Å)a
Backbone atoms	0.38 ± 0.09
All nonhydrogen atoms	0.87 ± 0.08
Ramachandran plotb
Most favored regions	92.4%
Additional allowed regions	7.5%
Generously allowed regions	0.1%
Disallowed regions	0.0%
CNS energy term (kcal • mole−1)
Eoverall	46.9 ± 0.6
Ebond	0.735 ± 0.03
Eangle	39.1 ± 0.2
Eimprop	1.05 ± 0.03
EvdW	4.53 ± 0.29
ENOE	1.39 ± 0.28
Edihed	0.0260 ± 0.0079

^a For residues 5–14, 19–26, 35–41, 51–62, 69–75. RMSD: root mean square deviation.

^b From PROCHECK-NMR (Laskowski et al. 1996). For residues 4–25, 35–82.