• Open Access

Structural and kinetic analysis of an MsrA–MsrB fusion protein from Streptococcus pneumoniae


  • Re-use of this article is permitted in accordance with the Creative Commons Deed, Attribution 2.5, which does not permit commercial exploitation.

*E-mail hykim@ynu.ac.kr; Tel. (+82) 53 620 4347; Fax: (+82) 53 654 6651;

**E-mail chahong@korea.ac.kr; Tel. (+82) 2 3290 3009; Fax: (+82) 2 923 3229.


Methionine sulphoxide reductases (Msr) catalyse the reduction of oxidized methionine to methionine. These enzymes are divided into two classes, MsrA and MsrB, according to substrate specificity. Although most MsrA and MsrB exist as separate enzymes, in some bacteria these two enzymes are fused to form a single polypeptide (MsrAB). Here, we report the first crystal structure of MsrAB from Streptococcus pneumoniae (SpMsrAB) at 2.4 Å resolution. SpMsrAB consists of an N-terminal MsrA domain, a C-terminal MsrB domain and a linker. The linker is composed of 13 residues and contains one 310-helix and several hydrogen bonds interacting with both MsrA and MsrB domains. Interestingly, our structure includes the MsrB domain complexed with an SHMAEI hexa-peptide that is the N-terminal region of neighbouring MsrA domain. A kinetic analysis showed that the apparent Km of SpMsrAB for the R-form-substrate was 20-fold lower than that for the S-form substrate, indicating that the MsrB domain had a much higher affinity for the substrate than the MsrA domain. Our study reveals the first structure of the MsrAB by providing insights into the formation of a disulphide bridge in the MsrB, the structure of the linker region, and the distinct structural nature of active site of each MsrA and MsrB domain.


Methionine (Met) is one of the most oxidation-sensitive amino acids. Oxidation of Met to methionine sulphoxide (Met-SO) damages the proteins and can alter protein function (Hoshi and Heinemann, 2001; Weissbach et al., 2002; Moskovitz, 2005; Kim and Gladyshev, 2007). However, these damaged proteins can be reversed by the repair enzymes, methionine sulphoxide reductases (Msr). Msr enzymes are essential to protect cells, such as bacteria, mammals and plants, against oxidative stress (Skaar et al., 2002; Alamuri and Maier, 2004; Bechtold et al., 2004; Kantorow et al., 2004), and are also implicated in ageing and neurodegenerative diseases (Gabbita et al., 1999; Moskovitz et al., 2001; Ruan et al., 2002; Friguet, 2006; Wassef et al., 2007). Two distinct Msr enzymes have been classified for the reduction of Met-SO: MsrA reduces the S-form of Met-SO and MsrB reduces the R-form (Brot et al., 1981; Sharov and Schoneich, 2000; Grimaud et al., 2001). MsrA and MsrB have their active sites that display an essentially mirror-image-like relationship, reflecting the observed stereo-specificity of these enzymes (Lowther et al., 2002). The MsrA and MsrB sequences show no homology to each other. However, enzymatic studies demonstrate that both the enzymes share a common catalytic mechanism based on sulphenic acid chemistry involving two (or three) cysteine residues (Boschi-Muller et al., 2000; Kumar et al., 2002; Antoine et al., 2003; Olry et al., 2004; Kauffmann et al., 2005).

Most organisms contain MsrA and MsrB typically as separate enzymes. However, MsrA and MsrB exist as domains in a single fused protein (MsrAB) in some bacteria such as Streptococcus pneumoniae, Neisseria gonorrhoeae and Haemophilus influenza (Kryukov et al., 2002; Delaye et al., 2007). Why some bacteria possess the fusion protein instead of separate MsrA and MsrB forms is not clear. Analysis of the sequence alignment of MsrA, MsrB and MsrAB reveals that the active sites are conserved in each protein. To date, individual MsrA and MsrB structures from eight species have been reported (Lowther et al., 2000; 2002; Tête-Favier et al., 2000; Taylor et al., 2003; Rouhier et al., 2007; Ranaivoson et al., 2008) or deposited (1XM0, 3CXK and 3CEZ). However, there is no structural information on MsrAB. Compared with either the MsrA or MsrB structure, MsrAB structure may be somewhat different since a linker region exists between the MsrA and MsrB domains.

In this study, we report the first crystal structure of MsrAB from S. pneumoniae (SpMsrAB). The SpMsrAB consists of an N-terminal MsrA domain (SpMsrA), a C-terminal MsrB domain (SpMsrB) and a linker. The linker, iloop, contains conserved residues that participate in significant hydrogen bonds perhaps to maintain structural stability. Interestingly, our structure includes an SpMsrB complex with a hexa-peptide (SHMAEI) that is the N-terminal region of neighbouring SpMsrA through crystal packing. This is the first report of a complex form with a peptide including the product Met in the active site of MsrB structure. Kinetic analysis revealed that the apparent Km value of SpMsrAB for Met-R-SO is 20-fold lower than that for Met-S-SO and the catalytic efficiency (kcat/Km) for the R-form is sevenfold higher than that for the S-form.

Results and discussion

Overall structure and comparison of SpMsrAB

The orthorhombic structure of SpMsrAB was determined using the method of single-wavelength anomalous dispersion (SAD) to a resolution of 2.4 Å with seleno-Met. Phasing and refinement statistics are shown in Table 1. The asymmetric unit contains four molecules that show two conformations (Fig. 1A and Fig. S1). Molecule A (Mol A) and molecule C (Mol C) are the same conformation (conformation 1) while molecule B (Mol B) and molecule D (Mol D) are the other conformation (conformation 2). Interestingly, Mol A contains an SHMAEI hexa-peptide and Mol B and Mol C have an SHMA peptide in their MsrB domains, but there appears no peptide bound in Mol D. This structure results from the crystal packing. Thus far, no MsrB structures in complex with a peptide substrate (or product) have been reported. The SHMAEI hexa-peptide in Mol A was donated from the symmetric Mol D molecule in the neighbouring unit cell (Figs S1 and S2). The Ser and His are originated from the His-tag following thrombin treatment cleavage in the protein purification. The Met, Ala, Glu and Ile of the hexa-peptide are identical to the N-terminal residues of the native SpMsrAB sequence. The molecules donating SHMA peptides bound in Mol B and Mol C are not clear.

Table 1.  Data collection and refinement statistics.
  • a. 

    Rmerge = ΣhΣj|〈Ih − Ihj|/ΣhΣj|Ihj, where 〈Ih is the mean intensity of symmetry-equivalent reflections.

  • b. 

    Rwork = Σh|Fo − Fc|/Σh|Fo|, where Fo and Fc are the observed and calculated structure factor amplitudes of reflection h.

  • c. 

    Rfree is the same as Rwork, but calculated on the reflections set aside from refinement.

Data collection  
 Space groupP21212P21212
 Cell dimension, a, b, c (Å)158.8 165.0 77.9158.5 165.5 77.3
 Molecules per AU44
 Wavelength (Å)0.979501.00000
 Resolution range (Å)20.0–2.70 (2.80–2.70)20.0–2.4 (2.48–2.40)
 No. of measured reflections1 533 2001 036 517
 No. of unique reflections55 59990 336
 Completeness (%)99.6 (98.8)91.9 (77.5)
 Average I/σ(I)26.9 (3.5)9.3 (1.3)
 Rmergea (%)11.1 (36.8)10.5 (46.5)
 Resolution range (Å)20.0–2.720.0–2.4
 No. of reflections (work/test)56 761 (51 027/5734)77 162 (70 101/7061)
 B-factors (Å2) (protein/solvent)44.0/38.047.3/50.4
 No. of atoms (protein/ligand/water)10 088/92/41310 104/107/305
 Root mean square deviations  
  Bond lengths (Å)0.0080.007
  Bond angles (degree)1.31.4
Ramachandran plot  
 Most favoured (%)82.987.3
 Additionally allowed (%)16.612.6
 Generously allowed (%)0.50.1
 Disallowed (%)0.10.0
Figure 1.

Overall structures of SpMsrAB.
A. In the crystallographic orthorhombic form, four molecules of SpMsrAB show two conformations in an asymmetric unit. Mol A and Mol C are the same conformation while Mol B and Mol D are the other conformation. SHMAEI and SHMA peptides are shown by stick models. The SHMAEI hexa-peptide in Mol A is originated from the N-terminal region of Mol D in the neighbouring unit cell. The molecules that donate the SHMA in Mol B and Mol C are not clear.
B. Ribbon diagram of the structure of the SpMsrAB–SHMAEI hexa-peptide complex, with individual elements of secondary structure labelled. All figures were created using the pymol (http://pymol.sourceforge.net/) (left). Topology diagram of SpMsrAB was drawn with tops (Westhead et al., 1999). α-Helices (circles) are labelled α1–α15 and β-strand (triangles) are labelled β1–β16 (right).

The monomeric structure of SpMsrAB has the overall dimensions of approximately 74.5 × 37.5 × 39.7 Å3 and consists of two major domains, MsrA (residues 1–158) and MsrB (residues 172–312), and a linker domain (residues 159–171) (Fig. 1B). The N-terminal SpMsrA and C-terminal SpMsrB domains are structurally similar to the previously reported individual MsrA and MsrB proteins respectively. The N-terminal MsrA domain of SpMsrAB is composed of eight rolled mixed β-strands (β1–β8) and seven α-helices (α1–α7), similar to Bos taurus MsrA (BtMsrA) (Lowther et al., 2000), Escherichia coli MsrA (EcMsrA) (Tête-Favier et al., 2000), Mycobacterium tuberculosis MsrA (MtMsrA) (Taylor et al., 2003), Neisseria meningitides MsrA (NmMsrA) (Ranaivoson et al., 2008) and Populus trichocarpa MsrA (PtMsrA) (Rouhier et al., 2007) (Fig. 2A). The MsrB domain of SpMsrAB is composed of seven antiparallel β-strands (β9–β16) and seven α-helices (α9–α15), similar to the structures of MsrB from N. gonorrhoeae (NgMsrB) (Lowther et al., 2002), Burkholderia pseudomallei (BpMsrB; 3CEZ, 3CXK) and Bacillus subtilis (BsMsrB; 1XM0). We also determined the crystal structure of BsMsrB (3E0O) for comparison of SpMsrB (Fig. S3 and Table S1; Park et al., 2008). It should be noted that the crystal structures of BpMsrB (3CEZ, 3CXK) and the NMR structure of BsMsrB (1XM0) had not been reported when we initiated this study. Our crystal structure of BsMsrB was found to be identical to the NMR BsMsrB structure and was further used for structure comparison. Comparison of SpMsrB with BpMsrB, BsMsrB and NgMsrB revealed that SpMsrB is more highly conserved than SpMsrA (Fig. 2B). A sequence alignment analysis indicated that SpMsrAB shows a high sequence identity to other bacterial MsrABs: 52% (to N. meningitides), 53% (to N. gonorrhoeae), 60% (to Helicobacter pylori) and 59% (to H. influenza) (Fig. 3). These bacterial MsrABs are linked by a hinge region; NgMsrAB and NmMsrAB contain additional 12 residues in their linker regions compared with those of other species. However, the role of the linker in MsrABs is not known. The linker, iloop, of SpMsrAB consists of 13 residues (residues 159–171) and, interestingly, contains one 310-helix and several hydrogen bonds to both the MsrA and MsrB domains that could restrict the positions of both domains.

Figure 2.

Structural comparison of MsrAs and MsrBs.
A. Comparison of MsrAs. The backbone models for E. coli, Bos taurus, Mycobacterium tuberculosis, Neisseria meningitides (3BQH), Popular trichocarpa and Streptococcus pneumoniae are shown in green, red, yellow, orange, cyan and blue respectively.
B. Comparison of MsrBs. The backbone models for Bacillus subtilis, Neisseria gonorrhoaea, Burkholderia pseudomallei (3CXK) and S. pneumoniae are shown in orange, green, yellow and blue respectively.

Figure 3.

Multiple sequence alignment of MsrABs. Multiple sequence alignment of HpMsrAB, HiMsrAB, NmMsrAB, NgMsrAB and SpMsrAB is shown (Hi, Haemophilus influenza; Hp, Helicobacter pylori; Nm, Neisseria meningitides; Ng, Neisseria gonorrhoae). Sequence alignment was accomplished with BioEdit (http://www.mbio.ncsu.edu/BioEdit/bioedit.html). In the sequence alignment, red and blue box indicate the catalytic and recycling cysteine residues respectively; yellow and cyan box indicate the conserved residues participating in the catalytic mechanisms of MsrA and MsrB respectively. The linker region is shown in green box and Asp-156 and Lys-159 are shown as an inverted red triangle.

Structural positions of critical cysteines in catalysis in SpMsrAB

Cys-10 and Cys-284 are charged for catalytic attack of Met-S-SO and Met-R-SO respectively. Cys-150 and Cys-229 (called recycling cysteines) interact with the oxidized catalytic cysteines (sulphenic acid intermediates), respectively, to form disulphide bonds. Cys-10 and Cys-150 of SpMsrA were compared with those of EcMsrA, BtMsrA, MtMsrA, PtMsrA and NmMsrA (Fig. 4A). The structural positions of Cys-10 and Cys-150 were similar to those of other MsrAs. As expected from the structures of EcMsrA, BtMsrA, MtMsrA and PtMsrA, the catalytic Cys-10 of SpMsrA is remote from the recycling Cys-150 which is located within the flexible loop (residues 145–153). The distances between the two cysteine residues are 8.0 Å (SpMsrA), 8.6 Å (BtMsrA), 12.8 Å (EcMsrA), 6.8 Å (MtMsrA) and 7.1 Å (PtMsrA). To form a disulphide bridge, the two catalytic and recycling cysteine residues in such a long distance may undergo conformational changes that bring the residues within approximately 3.0 Å of each other, as shown in the structure of NmMsrA (Ranaivoson et al., 2008). As shown in Fig. 4B, unlike SpMsrA, the catalytic Cys-284 and recycling Cys-229 of SpMsrB are structurally identical to those of the NgMsrB, BpMsrB and BsMsrB. The distance (4.1 Å) between Cys-284 and Cys-229 is sufficient to enable formation of a disulphide bridge. To form a disulphide bridge, the β-carbon (Cβ) of Cys-284 must rotate towards the substrate Met-R-SO and the angle of the Cα of Cys-229 must change. These observations suggest that MsrB can easily form a disulphide bridge once the protein binds a substrate.

Figure 4.

Structural positions of active-site cysteine residues.
A. The positions of cysteines in SpMsrA. The catalytic Cys-10 and recycling Cys-150 are labelled. The stick models for the active-site amino acids for EcMsrA (green), BtMsrA (red), MtMsrA (yellow), NmMsrA (orange), PtMsrA (cyan) and SpMsrA (blue) are superimposed. For NmMsrA (3BQH), the approximate position of unresolved sequence was added and connected with the dotted line.
B. The positions of cysteines in SpMsrB. The catalytic Cys-284 and recycling Cys-229 are labelled. The stick models for the active-site amino acids for BsMsrB (red), NgMsrB (cyan), BpMsrB (green; 3CXK) and SpMsrB (yellow) are superimposed.

Complex formation with the SHMAEI hexa-peptide of SpMsrAB

Most MsrA structures contain a complex with either a small molecule such as dithiothreitol, mercaptoethanol and arsenate group (Lowther et al., 2000; Tête-Favier et al., 2000; Rouhier et al., 2007), or a Met residue through a crystal packing (Taylor et al., 2003), or even a Met-SO substrate (3BQF) (Ranaivoson et al., 2008) in their active pocket. NgMsrB structure also includes a cacodylate molecule in its active site (Lowther et al., 2002). Interestingly, as mentioned, our SpMsrAB structure reveals a complex between the active pocket of SpMsrB and the SHMAEI hexa-peptide, and is the first report of the structure of an MsrB complex with the product Met. The hexa-peptide is surrounded by four loops, A, B, C and D (Fig. 5A). The structure of the SpMsrB complex indicates that the catalytic Cys-284 faces the sulphur atom of the Met residue. His-269 and one water molecule also interact with the sulphur atom of the Met residue. Thr-192, His-266 and Asn-286 are stabilized by their interaction with water (Fig. 5B). The Thr-192, His-266, His-269 and Asn-286 have been proposed to be important for MsrB catalysis and these residues are well conserved in all MsrBs. The structurally ordered water molecule, which interacts with Thr-192, Asn-286 and Met-SO during the enzymatic reaction, is also suggested to play a key role in the enzymatic reaction of MsrB as shown in the structure of NgMsrB (Lowther et al., 2002). Our complex structure with the SHMAEI peptide supports the catalytic mechanism model of MsrB proposed earlier (Lowther et al., 2002).

Figure 5.

The SpMsrB–SHMAEI complex.
A. A complex structure of the SpMsrB–SHMAEI hexa-peptide. Four loops that surround the SHMAEI hexa-peptide are shown in surface model and the SHMAEI hexa-peptide is shown by 2Fo − Fc electron density map at 1.0 σ.
B. Interaction in the SpMsrB–SHMAEI hexa-peptide complex. The detail interaction between the Met of SHMAEI hexa-peptide and the active-site residues of SpMsrB is displayed. The water molecule is shown as a red sphere. The residues involving the interactions are shown in stick models. The Met residue is shown as a green stick model.

Similar to the general structural features of the MsrBs, the structure of SpMsrB demonstrates that the active site is exposed to solvent and that its substrate, Met-R-SO, can readily form a complex with the enzyme. The active-site pocket of SpMsrA is negatively charged while that of SpMsrB is positively charged, similar to the previously known MsrA and MsrB structures. These opposite charge distributions in the substrate binding pockets of SpMsrA and SpMsrB may contribute to a factor for discriminating their stereo-specific substrates, Met-S-SO and Met-R-SO, in which the positions of oxygen of the sulphoxide moiety are different from each other. Unlike SpMsrB, the residues forming the active-site pocket in SpMsrA are so bulky that the steric hindrance may be produced from the aromatic ring structures of Phe-11, Trp-12 and Tyr-91 when a substrate attempts to access the active pocket (Fig. 6A). Despite no significant conformational changes between the reduced NmMsrA (3BQE) and the complexed form with a Met-SO substrate (3BQF) (Ranaivoson et al., 2008), the flanked loop, including β4 and β5, of the active pocket appears to be somewhat different between these two forms (Fig. S4). It is more ‘open’ state in the reduced form than in the complexed form. This flanked loop of the SpMsrA was more ‘closed’ state than those of the complexed NmMsrA and MtMsrA forms (Fig. S4). Taken together, the results suggest the ‘closed’ nature of the MsrA active site relative to the ‘open’ MsrB active site in SpMsrAB. However, there is a possibility that the crystal contacts and packing of the four molecules within the crystal lattice may have preferentially stabilized the ‘closed’ state of active site of SpMsrA domain.

Figure 6.

Surface models of active pockets of MsrA and MsrB domains and MsrB catalytic mechanism model.
A. A surface model of SpMsrA active pocket. The stick models of residues (except Cys-10 and Cys-150) show the steric hindrance produced when a substrate attempts to access the pocket of SpMsrA. The active pocket is negatively charged.
B. A surface model showing the SpMsrB complex. The green stick model is the SHMAEI hexa-peptide and the blue stick models are the residues that interact with the hexa-peptide. The active pocket is positively charged.
C. A model for MsrB catalytic mechanism. When MsrB accesses a substrate, catalytic Cys-284 attacks the sulphur atom of its substrate (i.e. Met-R-SO) (a). Recycling Cys-229 and the oxidized Cys-284 (sulphenic acid form) change their positions (b) to establish a disulphide bridge (c). A complete disulphide bridge is formed in SpMsrB (d). The model shows SpMsrB and the SHMAEI hexa-peptide. Cys-229 and Cys-284 are shown in stick models (blue). The red stick model represents the shift of Cys-284 after MsrB attacks a substrate and the green stick model is the SHMAEI hexa-peptide.

As described above, the formation of the product Met–SpMsrB complex is correlated to the kinetics of SpMsrAB. The surface model of SpMsrB demonstrates that the SHMAEI hexa-peptide fits within the active-site pocket (Fig. 6B). Based on these results, we propose a structural model for the disulphide bridge formation in MsrB (Fig. 6C). When SpMsrB accesses Met-R-SO, (i) the catalytic Cys-284 attacks the sulphur atom of Met-R-SO and then a sulphenic acid intermediate is formed, (ii) once the product Met is released, the Cβ angle of Cys-284 rotates towards the recycling Cys-229, (iii) the Cys-229 also moves towards the Cys-284 and (iv) a disulphide bridge forms between these residues. Finally, the oxidized MsrB is associated with thioredoxin to reduce the disulphide bond.

Specific activity and kinetic analysis

The specific activities and kinetic parameters of SpMsrAB and BsMsrB are summarized in Table 2. HPLC for kinetic analysis and the preparation of dabsyl-Met-R-SO and dabsyl-Met-S-SO were conducted according to established procedures (Minetti et al., 1994; Kumar et al., 2002; Etienne et al., 2003). The MsrB activity of SpMsrAB was slightly higher than its MsrA activity at 200 μM substrate. However, the apparent kcat value for dabsyl-Met-R-SO was threefold lower than that for dabsyl-Met-S-SO. The total Msr activity was consistent with the sum of the MsrA and MsrB activities. The specific MsrA and MsrB activities of SpMsrAB were similar to those of mammalian MsrA and MsrB (Kim and Gladyshev, 2004; Kim and Gladyshev, 2005). The specific activity of BsMsrB was slightly lower than the MsrB activity of SpMsrAB and of mammalian MsrBs. Km values of SpMsrAB were 0.86 mM for dabsyl-Met-S-SO and 0.038 mM for dabsyl-Met-R-SO. These data suggest that SpMsrAB can efficiently catalyse the Met-R-SO to Met reaction at low substrate concentrations. BsMsrB also has a low Km value, similar to that of SpMsrAB for dabsyl-Met-R-SO. The catalytic efficiency (kcat/Km) of SpMsrAB for dabsyl-Met-R-SO is sevenfold higher than that for dabsyl-Met-S-SO. These results suggest that the MsrB domain contains a more efficient catalytic activity than the MsrA domain in the bacterial MsrAB. Moreover, the higher affinity for the R-form of the substrate may have been correlated with the SpMsrB complex formation with the SHMAEI hexa-peptide in the crystal structure.

Table 2.  Specific activities and kinetic parameters of SpMsrAB.
ProteinSubstrateSpecific activity [nmol min−1 (mg protein)−1]Km (mM)kcat (s−1)kcat/Km (M−1 s−1)Reference
  1. The specific activities and kinetic parameters were determined using dabsylated substrates as described in Experimental procedures.

  2. NA, not assayed.

S. pneumoniae MsrABMet-S-SO250 ± 70.86 ± 0.050.80 ± 0.04930 ± 10This study
Met-R-SO327 ± 30.038 ± 0.0020.24 ± 0.016320 ± 150 
Met-R,S-SO580 ± 7NANANA 
B. subtilis MsrBMet-R-SO264 ± 140.041 ± 0.0020.10 ± 0.0052440 ± 70This study
Mouse MsrAMet-S-SO238 ± 260.34 ± 0.040.28 ± 0.02820 ± 50Kim and Gladyshev (2005)
Mouse MsrB2Met-R-SO3530.170.231350Kim and Gladyshev (2004)
Human MsrB3Met-R-SO4232.92.29790Kim and Gladyshev (2004)

The linker region of SpMsrAB

SpMsrA and SpMsrB form a single protein joined by a linker. Why some bacteria contain an MsrAB rather than separate MsrA and MsrB is not clear, and raises a question as to whether these two domains behave differently in the fused protein. In this work, we first demonstrated the structure of MsrA–MsrB fusion protein including the linker region (see also the electron density map of the linker region in Fig. S5). Unexpectedly, two structural conformations of SpMsrAB were revealed in an asymmetric unit. We compared these two conformations and found that SpMsrA occurs in two dramatically distinct conformations relative to the centre of the iloop, while SpMsrB forms a complex with Met (Fig. 7A). SpMsrB is complexed with the SHMAEI hexa-peptide, and the movement of SpMsrB is therefore restricted. In contrast, SpMsrA may move over approximately 10 Å. As described above, the hinge loop iloop, which connects MsrA and MsrB, is composed of residues 159–171. The movement of Lys-159 is remarkable in the two conformations of the structure (Fig. 7B).

Figure 7.

The movement of MsrA and the hinge loop, iloop, motion of SpMsrAB.
A. Two conformations appeared in the crystallographic asymmetric unit. The position of SpMsrA moves approximately 10 Å. Con 1 and Con 2 show the distinct positions of the SpMsrA domain in an asymmetric unit when superposed by optimizing the linker regions of two conformations. The blue is shown as a conformation 1 (Con 1) and the white as a conformation 2 (Con 2). The N- and C-terminal of Con 2 is presented with N′ and C′ respectively. The red box is a linker region. The hexa-peptide SHMAEI from Con 1 is shown in green stick model.
B. The interaction change of Lys-159 between two conformations (Con 1 and Con 2). Asp-156 and Lys-159 of Con 2 is presented with Asp-156′ and Lys-159′ respectively. The magenta is the linker region of Con 1 while the white is the linker of Con 2. The amino acids of linker region are shown as the stick models.

As mentioned, the iloop contains one 310-helix and several hydrogen bonds interacting with residues at both the MsrA and MsrB domains. The interacting hydrogen bonds are summarized in Table 3. We also found changes of hydrogen bonds between Msr domains and iloop in the two conformations (Table 3). An outstanding change in hydrogen bonds is a new bond formation between Asp-156 of SpMsrA and Lys-159 of iloop in the conformation 2 (Fig. 7B). This interaction between Asp-156 and Lys-159 may induce (or result from) the distinct movement of SpMsrA. These results suggest that the movement of MsrA domain in other MsrABs may also occur while MsrB domain movement may be restricted relative to the central iloop, although the Asp-156 and Lys-159 of SpMsrAB are not conserved in other MsrABs. According to the sequence alignment, other MsrABs have Lys and Glu residues in these positions (Fig. 3), but these two residues are likely to interact with each other as well. To verify the role of these residues, further studies will be needed.

Table 3.  The hydrogen bonds and their changes between Msr domains and iloop in the two conformations.
Hydrogen bondConformation 1Conformation 2
  1. –, none.

Arg-65 (NH2)–Asp-163 (Oδ1)2.9 Å3.0 Å
Arg-65 (Nε)–Asp-163 (Oδ2)2.5 Å2.8 Å
Arg-73 (NH1)–Lys-159 (O)2.8 Å3.4 Å
Arg-73 (NH2)–Pro-160 (O)3.2 Å2.9 Å
Asp-156 (Oδ1)–Lys-159 (Nζ)2.7 Å
Ser-171 (Oγ)–Val-174 (Cγ1)2.9 Å3.0 Å
Ser-171 (O)–Leu-175 (N)3.2 Å3.1 Å
Lys-169 (Nζ)–Gln-188 (Oε1)2.8 Å
Tyr-167 (O)–Ser-262 (Oγ)2.6 Å2.6 Å
Ile-162 (O)–Arg-261 (Nη2)2.9 Å2.9 Å
Tyr-167 (OH)–Tyr-305 (OH)2.9 Å2.9 Å

The linker of SpMsrAB may stabilize the positions of the MsrA and MsrB domains relative to each other, despite the fact that two distinct conformations are observed in an asymmetric unit. In regard to the occurrence of two dramatically different MsrA conformations, the SpMsrA may have another function in the cells in addition to the catalytic function.

In summary, we determined the first crystal structure of SpMsrAB at 2.4 Å resolution. First, we suggest that the iloop region of SpMsrAB may play a role in the structural stability of the protein by hydrogen bond interactions with both the MsrA and MsrB domains. Further biochemical and structural studies will be necessary for verifying the function of the iloop. Second, the apparent Km value of SpMsrB for the substrate is 20-fold lower than that of SpMsrA, suggesting that SpMsrB forms a complex with its substrate more readily than SpMsrA. Third, in agreement with this kinetic result, we demonstrate the first structure of MsrB complexed with Met residue. As a result of examining the SpMsrB–SHMAEI complex, we propose a model for the catalytic mechanism of MsrB and this model shows how catalytic and recycling cysteine residues are involved in conformational changes to form a disulphide bridge. The structural and subsequent biochemical analysis reveals why SpMsrB can readily form a complex with its substrate and provides insights into the distinct structural nature of the active site of each MsrA and MsrB domain in MsrAB family.

Experimental procedures

Cloning, expression and protein purification

The gene encoding the full-length SpMsrAB (residues 1–312) was amplified from the S. pneumoniae genome by PCR. The PCR products were then digested with BamHI and XhoI and inserted into pET-28a (+) (Novagen), containing a His-tag. The plasmid was transformed into E. coli BL21(DE3). Cells were grown in LB medium and protein expression was induced with 0.5 mM IPTG at 18°C. After induction, the cells were harvested and disrupted by sonication in buffer A [20 mM Tris-HCl (pH 8.0), 100 mM NaCl, 5 mM imidazole and 1 mM DTT]. The lysate was then clarified by centrifugation and was applied to a 5 ml HisTrap column (Amersham Pharmacia). The protein was eluted by linear gradient with buffer A and 5–500 mM imidazole. The His-tag was removed by treatment with thrombin, followed by dialysis overnight at 4°C. The protein was loaded on a 5 ml HiTrap ion exchange column (Amersham Pharmacia) using buffer of 50 mM Tris-HCl (pH 8.0) and 5 mM DTT with a gradient of 0–1.0 M NaCl followed by gel filtration on a HiLoad 26/60 Superdex-200 column (Amersham Pharmacia) using buffer of 25 mM Tris-HCl (pH 8.0), 100 mM NaCl and 5 mM DTT. The purified protein was concentrated to 35 mg ml−1. Seleno-Met-labelled SpMsrAB was expressed in E. coli B834(DE3) and purified as described above.

Crystallization and data collection

SpMsrAB crystals suitable for X-ray data collection were grown by the hanging-drop vapour diffusion method at 22°C in [0.1 M MES (pH 5.6), 1% (w/v) PEG 4000 and 0.2 M MgCl2]. For cryoprotection, the crystals were soaked in reservoir solution containing 25% ethylene glycol and frozen in cold nitrogen at −173°C. X-ray diffraction was performed at beam line 4A of the Pohang Light Source (PLS), Pohang, Korea. The final X-ray diffraction data were collected with a Quantum 210 CCD detector (Area Detector Systems, Poway, CA). The native SpMsrAB crystal diffracted to 2.4 Å and belongs to the space group P21212, with unit cell dimenhyqjsions: a = 158.5 Å, b = 165.5 Å and c = 77.3 Å. The SAD data using seleno-Met-labelled SpMsrAB were collected at peak wavelength (0.9795 Å). The data were processed and scaled using HKL2000 (Otwinowski and Minor, 1997).

Structure determination and refinement

The crystal structure of SpMsrAB was determined by the SAD method. Searching of 12 selenium sites and calculation of phase were carried out with the SOLVE (Terwilliger and Berendzen, 1999; Terwilliger, 2000). The first model was built into 2.7 Å resolution electron density map by using Coot (Emsley and Cowtan, 2004). After refinement with CNS (Brünger et al., 1998), the resolution was improved to 2.4 Å. Final refinement, after including a hexa-peptide and solvents, resulted in R and Rfree values of 23.9% and 28.2% (for a 10% data sample) respectively. Data collection and refinement statistics are summarized in Table 1. The atomic co-ordinates and structure factors for the SpMsrAB have been deposited in the Protein Data Bank with the accession code 3E0M. A crystal structure of BsMsrB was also determined (Table S1; Park et al., 2008) and has been deposited (3E0O). The Protein Data Bank accession codes for other Msr proteins discussed in this article are as follows: BtMsrA (1FVA), EcMsrA (1FF3), MtMsrA (1NWA), NmMsrA (3BQE, 3BQF, 3BQH), PtMsrA (2J89), BsMsrB (1XM0), BpMsrB (3CEZ, 3CXK) and NgMsrB (1L1D).

Kinetic assays

MsrA and MsrB activities were determined in the presence of DTT using dabsylated Met-SO as substrate. Briefly, a 100 μl reaction mixture contained a buffer of 50 mM sodium phosphate (pH 7.5), 50 mM NaCl and 20 mM DTT, and either 200 μM dabsyl-Met-S-SO (for MsrA assays) or dabsyl-Met-R-SO (for MsrB assays), and 1 μg of purified protein of SpMsrAB or BsMsrB. To assay for the total Msr activity of SpMsrAB, 400 μM mixed (R, S) Met-SO was used. The reactions were carried out at 37°C for 30 min and stopped by adding 200 μl of acetonitrile. The dabsyl-Met product was analysed using an HPLC procedure. Km and kcat values were determined for DTT-dependent reactions from Lineweaver-Burk plots. For determination of Km, 0.05–0.8 mM dabsyl-Met-S-SO was used and 0.05–0.2 mM dabsyl-Met-R-SO was used.


We thank Dr E.E. Kim and K.H. Kim for help with X-ray diffraction pre-testing at KIST. We thank Dr H.S. Lee and his staff at beam line 4A, Pohang Accessory Laboratory, for assistance in data collection. This work was supported by the Functional Proteomics Center, 21C Frontier Program, of the Korea Ministry of Science and Technology and by the KOSEF (Korea Science and Engineering Foundation) Grant R13-2005-005-01004-0 (H.-Y.K.). Y.K.K. is supported by the Seoul Fellowship.