Search tips
Search criteria 


Logo of actafjournal home pagethis articleInternational Union of Crystallographysearchsubscribearticle submission
Acta Crystallogr Sect F Struct Biol Cryst Commun. 2010 October 1; 66(Pt 10): 1354–1364.
Published online 2010 July 27. doi:  10.1107/S1744309110021214
PMCID: PMC2954226

Structure of the γ-d-glutamyl-l-diamino acid endopeptidase YkfC from Bacillus cereus in complex with l-Ala-γ-d-Glu: insights into substrate recognition by NlpC/P60 cysteine peptidases


Dipeptidyl-peptidase VI from Bacillus sphaericus and YkfC from Bacillus subtilis have both previously been characterized as highly specific γ-d-glutamyl-l-­diamino acid endopeptidases. The crystal structure of a YkfC ortholog from Bacillus cereus (BcYkfC) at 1.8 Å resolution revealed that it contains two N-terminal bacterial SH3 (SH3b) domains in addition to the C-terminal catalytic NlpC/P60 domain that is ubiquitous in the very large family of cell-wall-related cysteine peptidases. A bound reaction product (l-Ala-γ-d-Glu) enabled the identification of conserved sequence and structural signatures for recognition of l-Ala and γ-d-Glu and, therefore, provides a clear framework for understanding the substrate specificity observed in dipeptidyl-peptidase VI, YkfC and other NlpC/P60 domains in general. The first SH3b domain plays an important role in defining substrate specificity by contributing to the formation of the active site, such that only murein peptides with a free N-terminal alanine are allowed. A conserved tyrosine in the SH3b domain of the YkfC subfamily is correlated with the presence of a conserved acidic residue in the NlpC/P60 domain and both residues interact with the free amine group of the alanine. This structural feature allows the definition of a subfamily of NlpC/P60 enzymes with the same N-terminal substrate requirements, including a previously characterized cyanobacterial l-­alanine-γ-d-glutamate endopeptidase that contains the two key components (an NlpC/P60 domain attached to an SH3b domain) for assembly of a YkfC-like active site.

Keywords: γ-d-glutamyl-l-diamino acid endopeptidase, cell-wall recycling, NlpC/P60, SH3b, cysteine peptidases, enzyme specificity

1. Introduction

Cell-wall turnover, an enzymatic process that results in the loss of peptidoglycan (PG) components, has been reported in many bacteria, including Escherichia coli and Bacillus subtilis (Doyle et al., 1988 [triangle]). The products of the turnover are generally re-utilized through a process known as PG recycling (Park & Uehara, 2008 [triangle]). The molecular processes involved in cell-wall turnover and recycling are not currently well understood in comparison to cell-wall synthesis, particularly in bacteria other than E. coli (Park & Uehara, 2008 [triangle]; Uehara & Park, 2003 [triangle], 2004 [triangle], 2007 [triangle]; Uehara et al., 2005 [triangle]). In E. coli, the cell wall is degraded by lytic transglycosylases that release anhydromuropeptides (GlcNAc-anhMurNAc-l-Ala-γ-d-Glu-DAP-d-Ala, where DAP is meso-diaminopimelic acid), which are imported into the cytoplasm, primarily by AmpG permease, and subsequently processed by N-acetyl-anhydromuramyl-l-alanine amidase (which cleaves between GlcNAc-anhMurNAc and l-Ala) and ld-carboxy­peptidase LdcA (which cleaves between DAP and d-Ala). Two fates are possible for the generated murein tripeptide l-Ala-γ-d-Glu-DAP. Under normal growth conditions, these tripeptides are recycled by Mpl ligase and returned to the peptidoglycan-biosynthetic pathway. During nutrient-limiting conditions, an additional pathway (Fig. 1 [triangle]) is likely to be involved in the murein tripeptide metabolism, as proposed in E. coli (Uehara & Park, 2003 [triangle]). MpaA endopeptidase, a metallocarboxypeptidase, specifically cleaves l-Ala-γ-d-Glu-DAP to produce l-Ala-γ-d-Glu and DAP. l-Ala-γ-d-Glu is then converted to l-Ala-l-Glu and subsequently to l-Ala and l-Glu by YcjG epimerase and PepD peptidase, respectively.

Figure 1
Proposed metabolic pathways for murein peptides in E. coli and B. subtilis.

Some of the enzymes in the cell-wall recycling of E. coli, such as AmpG, AmpD, Mpl and MpaA, have no orthologs in B. subtilis (Park & Uehara, 2008 [triangle]), suggesting that the mechanism of cell-wall recycling may differ between the two bacteria. In B. subtilis, PG was proposed to be cleaved by a muramidase and an amidase to produce GlcNAc-MurNAc and stem peptides (Park & Uehara, 2008 [triangle]). The free peptides are imported into the cytoplasm by an unidentified permease and subsequently processed by YkfABC enzymes. YkfA, an ld-carboxy­peptidase, removes the terminal d-Ala. The generated tripeptide is further metabolized by a pathway that is functionally equivalent to that of E. coli, with YkfC as the γ-d-Glu-DAP endopeptidase and YkfB as the l-Ala-d-Glu epimerase (Fig. 1 [triangle]) (Schmidt et al., 2001 [triangle]). Interestingly, while YkfB is homologous to YcjG of E. coli, YkfC is unrelated to MpaA in sequence and structure despite having an equivalent function.

YkfC contains a C-terminal NlpC/P60 cysteine peptidase domain. NlpC/P60 is a large family of cell-wall related cysteine peptidases that are broadly distributed in bacteria, viruses, archaea and eukaryotes (Anantharaman & Aravind, 2003 [triangle]; Bateman & Rawlings, 2003 [triangle]; Rigden et al., 2003 [triangle]). Characterized NlpC/P60 enzymes are almost all γ-d-Glu-DAP (or γ-d-Glu-Lys) endopeptidases. While their bio­chemical function seems to be conserved, the physiological roles of NlpC/P60 proteins are diverse, including involvement in cell separation, expansion, differentiation, cell-wall turnover, cell lysis, protein secretion and virus infection (Smith et al., 2000 [triangle]). Secreted NlpC/P60 proteins also have other roles in pathogenesis. The autolysin P60 of Listeria monocytogenes is involved in host-cell invasion (Kuhn & Goebel, 1989 [triangle]), enterotoxin FM of B. cereus in food poisoning (Asano et al., 1997 [triangle]), and SagA of Enterococcus faecium is a secreted antigen that binds to extracellular matrix proteins (Teng et al., 2003 [triangle]).

NlpC/P60 proteins can be lethal to bacteria owing to their ability to compromise cell-wall integrity or cell-wall biosynthesis. Therefore, their activities are tightly controlled through multiple mechanisms (Smith et al., 2000 [triangle]); their expression is regulated at the transcription level and their cellular localization is dependent on their physio­logical roles. Furthermore, their atomic structures are highly optimized to precisely define their substrate specificity (Xu et al., 2009 [triangle]). NlpC/P60 proteins are often fused to auxiliary domains, many of which are known cell-wall binding modules (e.g. LysM and the choline-binding domain). Thus, it is generally assumed that these auxiliary domains function as targeting domains which localize their proteins to the cell wall. The functional synergy between the NlpC/P60 domains and their auxiliary domains is currently not fully understood. We have previously determined the crystal structure of a γ-d-Glu-DAP endopeptidase from cyanobacteria (AvPCP/NpPCP; Anabaena variabilis/Nostoc punctiforme PG cysteine peptidase; Xu et al., 2009 [triangle]) and showed that it contained an N-terminal bacterial SH3 (SH3b) domain and a C-terminal NlpC/P60 domain. We proposed that the SH3b domain of this enzyme is important in defining the substrate specificity of the peptidase domain. However, the mechanism of substrate recognition by NlpC/P60 and SH3b was not firmly established. Here, we report the crystal structure of YkfC from B. cereus (BcYkfC) in complex with l-Ala-γ-d-Glu. BcYkfC shares 40% sequence identity with YkfC from B. subtilis, which has previously been biochemically characterized (Schmidt et al., 2001 [triangle]). Thus, we now have the first detailed view of substrate recognition by an NlpC/P60 protein.

2. Material and methods

2.1. Sequence analysis

Homologs of BcYkfC were identified using PSI-BLAST (Altschul et al., 1997 [triangle]; three iterations) against the nonredundant (nr) protein sequence database at the National Center for Biotechnology Information (NCBI) using the sequence of the catalytic domain of BcYkfC as the probe (residues 210–333). An alignment length of ≥70 and an E value of ≤0.02 were used to extract a subset of hits. These proteins (2599 sequences) were aligned using HMMALIGN (Eddy, 1998 [triangle]) against the default global alignment profile of NlpC/P60 domains (PF00877) from the PFAM database v.23 (Bateman et al., 2004 [triangle]). YkfC-subfamily candidates were extracted from the above aligned subset based on the presence of an aspartate corresponding to position 256 of BcYkfC. The full-length sequences were then aligned and clustered using PIPEALIGN (Plewniak et al., 2003 [triangle]). Only sequences that also contained a conserved tyrosine corresponding to position 118 of BcYkfC were classified into the YkfC subfamily. Plots of sequence conservation in the active site were prepared using WEBLOGO (Crooks et al., 2004 [triangle]).

2.2. Protein expression and purification

Clones were generated using the Polymerase Incomplete Primer Extension (PIPE) cloning method (Klock et al., 2008 [triangle]). The gene encoding BcYkfC (GenBank NP_979181; Swiss-Prot Q736M3) was amplified by polymerase chain reaction (PCR) from B. cereus NRS248 ATCC 10987 genomic DNA using PfuTurbo DNA polymerase (Stratagene) and I-PIPE (Insert) primers (forward primer, 5′-ctgtacttccagggcGAAGAGAAGAAAGATAGTAAGGCGT-3′; reverse primer, 5′-aattaagtcgcgttaAGGTAAGTAACGACGCGCAC­CAGCG-3′; target sequence in upper case) that included sequences for the predicted 5′ and 3′ ends. The expression vector pSpeedET, which encodes an amino-terminal tobacco etch virus (TEV) protease-cleavable expression and purification tag (MGSDKIHHHHHHENLYFQ/G), was PCR-amplified with V-PIPE (Vector) primers (forward primer, 5′-taacgcgacttaattaactcgtttaaacggtctccagc-3′; reverse primer, 5′-gccctggaagtacaggttttcgtgatgatgatgatgatg-3′). V-PIPE and I-PIPE PCR products were mixed to anneal the amplified DNA fragments together. E. coli GeneHogs (Invitrogen) competent cells were transformed with the I-PIPE/V-PIPE mixture and dispensed onto selective LB–agar plates. The cloning junctions were confirmed by DNA sequencing. Using the PIPE method, the gene segment encoding residues Met1–Ala23 was omitted as these residues were predicted to form a signal peptide. Expression was performed in a selenomethionine-containing medium with suppression of normal methionine synthesis. At the end of fermentation, lysozyme was added to the culture to a final concentration of 250 µg ml−1 and the cells were harvested and frozen. After one freeze–thaw cycle, the cells were sonicated in lysis buffer [50 mM HEPES pH 8.0, 50 mM NaCl, 10 mM imidazole, 1 mM tris(2-carboxyethyl)phosphine–HCl (TCEP)] and the lysate was clarified by centrifugation at 32 500g for 30 min. The soluble fraction was passed over nickel-chelating resin (GE Healthcare) pre-equilibrated with lysis buffer, the resin was washed with wash buffer [50 mM HEPES pH 8.0, 300 mM NaCl, 40 mM imidazole, 10%(v/v) glycerol, 1 mM TCEP] and the protein was eluted with elution buffer [20 mM HEPES pH 8.0, 300 mM imidazole, 10%(v/v) glycerol, 1 mM TCEP]. The eluate was buffer-exchanged with TEV buffer (20 mM HEPES pH 8.0, 200 mM NaCl, 40 mM imidazole, 1 mM TCEP) using a PD-10 column (GE Healthcare) and incubated with 1 mg TEV protease per 15 mg of eluted protein. The protease-treated eluate was run over nickel-chelating resin (GE Healthcare) pre-equilibrated with HEPES crystallization buffer (20 mM HEPES pH 8.0, 200 mM NaCl, 40 mM imidazole, 1 mM TCEP) and the resin was washed with the same buffer. The flowthrough and wash fractions were combined and concentrated to 18.8 mg ml−1 as determined using the Coomassie Plus Protein Assay Reagent (Pierce) by centrifugal ultrafiltration (Millipore) for crystallization trials. The oligomeric state of BcYkfC was determined using a 0.8 × 30 cm Shodex Protein KW-803 column (Thomson Instruments) pre-calibrated with gel-filtration standards (Bio-Rad).

2.3. Crystallization

BcYkfC was crystallized by mixing 200 nl protein solution with 200 nl crystallization solution and equilibrating against a 50 µl reservoir solution using the nanodroplet vapor-diffusion method (Santarsiero et al., 2002 [triangle]) with standard Joint Center for Structural Genomics (JCSG; crystallization protocols (Lesley et al., 2002 [triangle]). The crystallization solution was composed of 0.2 M sodium chloride, 50%(v/v) PEG 200 and 0.1 M phosphate–citrate pH 4.2. A needle-shaped crystal of approximate dimensions 100 × 15 × 15 µm was harvested after 29 d at 277 K. No additional cryoprotectant was added to the crystal. Initial screening for diffraction was carried out using the Stanford Automated Mounting system (SAM; Cohen et al., 2002 [triangle]) at the Stanford Synchrotron Radiation Lightsource (SSRL, Menlo Park, California, USA).

2.4. Data collection, structure solution and refinement

Multi-wavelength anomalous diffraction (MAD) data were collected at wavelengths corresponding to the peak, high-energy remote and inflection wavelengths of a selenium MAD experiment at 100 K using a MAR CCD 325 detector (Rayonix) on SSRL beamline 11-1. Processing of the diffraction data and initial structure solution were carried out using the automatic structure-solution script autoXDSp developed at the JCSG (unpublished work). This script shepherds the structure-determination process, as summarized below, using preset rules through a decision-tree that mimics that used by an experienced crystallographer. The calculations are parallelized on a computer cluster such that initial maps and models can usually be obtained within 1 h of the completion of data collection. In summary, the MAD data were integrated and reduced using XDS and then scaled with the program XSCALE (Kabsch, 1993 [triangle], 2010 [triangle]). Selenium sites were located with SHELXD (Sheldrick, 2008 [triangle]). Phase refinement and automatic model building were performed using autoSHARP (Bricogne et al., 2003 [triangle]) and ARP/wARP (Cohen et al., 2004 [triangle]). This automated process produced an initial model that was 92% complete. Further model completion and refinement were performed manually with Coot (Emsley & Cowtan, 2004 [triangle]) and REFMAC (Murshudov et al., 1997 [triangle]) from the CCP4 suite (Collaborative Computational Project, Number 4, 1994 [triangle]). Data and refinement statistics are summarized in Table 1 [triangle]. Analysis of the stereochemical quality of the model was accomplished using MolProbity (Chen et al., 2010 [triangle]). All molecular graphics were prepared with PyMOL (DeLano Scientific) unless specifically stated otherwise. Atomic coordinates and experimental structure factors for BcYkfC at 1.8 Å resolution have been deposited in the PDB with accession code 3h41.

Table 1
Data-collection, phasing and refinement statistics (PDB code 3h41)

2.5. Molecular modeling

Molecular docking was performed using the same protocol as described previously (Xu et al., 2009 [triangle]) using Glide v.5.0 (Schrödinger LLC). The positions of the bound ligand were used as restraints such that the l-Ala-γ-d-Glu portion of the docked substrate adopted a similar conformation as seen in the crystal structure. In order to perform the docking studies, the tri-oxidized cysteine (OCS238) in the crystal structure was replaced with the reduced form, which is needed for reaction in the papain family of cysteine peptidases. Furthermore, only the side-chain conformer with higher occupancy in the crystal structure was considered in the docking experiments when multiple conformations were observed for an active-site residue.

3. Results and discussion

3.1. Genomic context

Full-length BcYkfC (strain B. cereus ATCC 10987) contains 333 residues (molecular weight 37.3 kDa), the first 23 of which are pre­dicted to be a signal peptide by the Phobius web server (Kall et al., 2007 [triangle]). Despite being homologous to BcYkfC, B. subtilis YkfC (296 residues) does not contain a signal peptide, suggesting that the two proteins function at different cellular locations.

As in B. subtilis, the ykfC gene (BCE_2878) is adjacent to the ykfB l-Ala-d-Glu epimerase gene (BCE_2879) in the B. cereus genome. The YkfB epimerases (sequence identity 50%) from both bacteria do not contain predicted signal peptides. The genome association of ykfB and ykfC is also observed in other bacteria (e.g. Listeria innocua, Acidobacterium sp., Solibacter usitatus, Bacteroides thetaiota­omicron and Gramella forsetii). However, the genome context of ykfBC is different in B. subtilis compared with B. cereus (Fig. 2 [triangle]). The YkfA–D genes of B. subtilis are located next to the dppB–E dipeptide ABC transporter operon, whereas the YkfBC genes of B. cereus are located downstream of divIC, which encodes a putative cell-division protein. Downstream of ykfC is an oppA gene that is homologous to dppE (34% sequence identity). Both genes are predicted to encode periplasmic dipeptide-binding proteins.

Figure 2
Genomic context of the ykfB and ykfC genes in B. subtilis and B. cereus.

Based on the presence/absence of the signal peptide on YkfC and YkfB, as well as their genomic contexts, we suggest that the strategy for metabolizing murein peptides is likely to differ between B. subtilis and B. cereus. In B. cereus, the murein peptides are likely to be broken down outside the cell, with the resulting dipeptide l-Ala-γ-d-Glu being imported into cytoplasm for further processing by YkfB, while the reactions catalyzed by both YkfC and YkfB are likely to occur in the cytoplasm in B. subtilis.

3.2. Structure determination and quality of the model

The crystal structure of BcYkfC was determined using the high-throughput structural genomics pipeline implemented at the JCSG (Lesley et al., 2002 [triangle]). The selenomethionine derivative of BcYkfC was expressed in E. coli with an N-terminal TEV-cleavable His tag and purified by metal-affinity chromatography. In order to improve the chance of obtaining crystals, the predicted N-terminal signal peptide (residues 1–23) was not included in the cloned construct. The data were indexed in space group C2 and the structure was determined at 1.79 Å resolution with one molecule per asymmetric unit using the MAD method (R cryst = 16.3%, R free = 19.7%). The electron density for the main chain was well defined throughout the entire molecule. The mean residual error of the coordinates was estimated to be 0.11 Å by the diffraction-component precision index (DPI) method (Cruickshank, 1999 [triangle]). The model of BcYkfC displays good geometry, with an all-atom clash score of 5.07, and the Ramachandran plot produced by MolProbity (Chen et al., 2010 [triangle]) shows that all residues, but one, are in allowed regions, with 97.7% in favored regions. Only one residue is flagged as a rotamer outlier. The Ramachandran (Pro305) and rotamer (His303) outliers are supported by well defined electron density. Since these residues are either close to or part of the active site, these structural deviations from ideality are likely to be of functional relevance. Additionally, two cis-peptides (Asn57–Pro58 and Asn154–Pro155) are also supported by clear electron density. The final model of BcYkfC contains residues 29–333, one dipeptide l-Ala-γ-d-Glu, one phosphate, six polyethylene glycol (PEG) fragments from the crystallization solution and 265 waters. The residual residue (Gly0) from the cleaved N-terminal purification tag, residues 24–28 and the side chains of Glu201 and Arg270 were disordered and were not included in the final model. Data-collection, refinement and model statistics are summarized in Table 1 [triangle].

3.3. Overall structure

BcYkfC is likely to be a monomer in solution, as supported by crystal-packing analysis and analytical size-exclusion chromatography. The structure of BcYkfC consists of three domains: two SH3b domains (SH3b1, residues 29–129; SH3b2, residues 130–207) and a C-terminal NlpC/P60 cysteine peptidase domain (residues 208–333) (Figs. 3 [triangle] a and 3 [triangle] b). The two SH3b domains are similar to each other, with an r.m.s.d. of 2.2 Å for 57 aligned Cα atoms (sequence identity of 13%). A structural similarity search using DALI (Holm & Sander, 1995 [triangle]) did not find any other structures with the same three-domain architecture. However, a number of significantly similar substructures were identified and are summarized in Table 2 [triangle]. The SH3b domains of BcYkfC are similar to other SH3b domains (Holm & Sander, 1995 [triangle]), including the N-­terminal domain of cyanobacterial γ-d-Glu-DAP endopeptidases (Xu et al., 2009 [triangle]), the GW domains of internalin B (Marino et al., 2002 [triangle]), the cell-wall-targeting domain of glycylglycine endopeptidase ALE-1 (Lu et al., 2006 [triangle]), PhnA-like protein (Srisailam et al., 2006 [triangle]) and endolysin PlyPSA (Korndorfer et al., 2006 [triangle]), as well as many eukaryotic SH3 domains. A β-hairpin within the so-called ‘RT loop’ (i.e. the loop between βA and βB) region appears to be a common and unique structural feature of SH3b domains compared with their eukaryotic counterparts (Xu et al., 2009 [triangle]). The BcYkfC SH3b domains both contain this conserved structural motif (βA1–βA2). However, SH3b1 contains a novel helical insertion (α1–α3) that is not seen in previous SH3b structures or in SH3b2 (Fig. 3 [triangle] b). Prokaryotic SH3-like domains have also been implicated in polypeptide binding (Wylie et al., 2005 [triangle]) and metal binding (Pohl et al., 1999 [triangle]). Although these SH3-like domains contain a similar five-stranded (βA–βE) core, they display much larger structural differences (and are not among significant DALI hits with Z > 2.0) and lack the β-­hairpin in the RT-loop region when compared with SH3b domains. The C-terminal NlpC/P60 catalytic domain of BcYkfC is remotely related to the papain family of cysteine peptidases (Anantharaman & Aravind, 2003 [triangle]), with highest similarity to cyanobacterial γ-d-Glu-DAP endopeptidases (Xu et al., 2009 [triangle]), E. coli lipoprotein Spr (Aramini et al., 2008 [triangle]) and two uncharacterized proteins (PDB codes 2p1g and 2im9; NYSGXRC, unpublished work) (Table 2 [triangle]).

Figure 3
Crystal structure of YkfC from B. cereus in complex with l-Ala-γ-d-Glu. (a) Ribbon representation of BcYkfC, highlighting its domain organization. SH3b1 is depicted in blue, SH3b2 in green and NlpC/P60 in red. The bound l-Ala-γ-d-Glu is ...
Table 2
Structural comparisons of BcYkfC and other bacterial proteins that share at least one common domain

The three domains of BcYkfC are arranged in a triangle such that each domain interacts with the two other domains. The interface (~570 Å2 per domain) between the two SH3b domains is mostly hydrophobic and is centered on interactions between the βA–βA1 and βA2–βB loops of SH3b1 and the βA2 and βB strands and βD–βE loop of SH3b2. The SH3b1–NlpC/P60 domain interface (903 Å2 buried surface per domain) is mediated through α2–α3–βA1 and the βC–βD loop of SH3b1, and α1–α2–α3 and the β6–β7 loop of NlpC/P60. The active site is located at this SH3b1–NlpC/P60 interface, whereas the SH3b2 domain is distal to the active site. A multiple sequence alignment of full-length homologs of YkfC (37 sequences; average sequence identity of 54%) indicates that most of the highly conserved residues are either buried inside the protein or clustered around the active site (Fig. 3 [triangle] c).

3.4. Active site

The catalytic triad of the peptidase domain consists of Cys238, His291 and His303. The catalytic Cys238 is oxidized (OCS) in the crystal based on the electron density (Fig. 4 [triangle] a). Since an oxidized cysteine can no longer function as a nucleophile in the reaction (Storer & Ménard, 1994 [triangle]), the enzyme in the crystal is inactive. The con­formation of the cysteine side chain is not significantly affected by the oxidation as its side chain is in a similar location and conformation as in other NlpC/P60 structures (Xu et al., 2009 [triangle]).

Figure 4
Active site and recognition of l-Ala-γ-d-Glu by BcYkfC. (a) Stereoview of a 2F oF c OMIT map, where l-Ala-γ-d-Glu and OCS238 were omitted from phasing/refinement, contoured at 1.5σ. ...

The dipeptide l-Ala-γ-d-Glu was identified from well defined electron density in the active site of BcYkfC (Fig. 4 [triangle] a). As the dipeptide was not added during protein purification or crystallization, it was most likely to have been obtained during protein expression in E. coli. Since this dipeptide is one of the reaction products of BcYkfC (Fig. 2 [triangle]), it unequivocally identifies the S2–S1 binding-site cavity, which is formed by residues from both the SH3b1 (Glu83, Thr84 and Tyr118) and NlpC/P60 domains (Tyr226, Trp228, Ala229, Asp237, Arg255, Asp256, Ser257, His290 and His291). Two charged residues, Asp237 and Arg255, which are highly conserved in NlpC/P60 domains, are involved in a hydrogen-bond network that connects many residues in the active site (Fig. 4 [triangle] b).

Two active-site residues, Ser257 and His291, display two discrete side-chain conformations (Fig. 4 [triangle] b). The first conformer of His291 (occupancy modeled as 0.7), which facilitates hydrogen bonding to His303, is identical to the corresponding histidine of the catalytic dyad in papain and other NlpC/P60 enzymes. The second rotamer points His291 towards the solvent. The side-chain isomerism in the active site is likely to be a consequence of the observed oxidized state of the catalytic Cys238, since the oxidized Cys238 makes multiple hydrogen-bond interactions with nearby side chains, including Tyr226, Ser257 and His291 (Fig. 4 [triangle] b).

3.5. Recognition of l-Ala-γ-d-Glu

The active-site residues form a pocket that is highly complementary in shape and chemical properties to the ligand (Figs. 4 [triangle] c and 4 [triangle] d). The average B value for the bound ligand is 26 Å2 (the overall B value of the protein is 18 Å2), indicating that it is well ordered with almost full occupancy in the active site. The interface between the dipeptide and the protein buries a total surface area of 490 Å2. The dipeptide is stabilized by multiple hydrogen bonds. The free amine of l-Ala in the S2 pocket makes hydrogen bonds to Glu83 O and the side chains of Tyr118 (OH) and Asp256 (Oδ1), which are highly conserved in the YkfC subfamily of NlpC/P60 enzymes (see below). The α-NH group of γ-d-Glu in the S1 site forms a weak hydrogen bond to Asp237 Oδ2. The α-carboxyl of d-Glu is stabilized by hydrogen-bonding interactions with Ser239 and Ser257 (Fig. 4 [triangle] d). On the solvent-exposed side of the substrate, waters are also involved in the hydrogen-bond network with the substrate and the enzyme (not shown). The l-Ala methyl side chain points towards a hydrophobic pocket defined by the side chains of Trp228 and Ala229. Additionally, the aliphatic C atoms of d-Glu are involved in hydrophobic contacts with Trp228. As a result, Trp228 contributes to both the S2 and S1 binding sites.

3.6. YkfC is specific for murein peptides with free N-terminal l-Ala

The active site of B. subtilis YkfC is highly conserved compared with that of BcYkfC (Fig. 5 [triangle]). Dipeptidyl peptidase VI (DPP VI) from B. sphaericus is a γ-d-Glu-DAP(Lys) dipeptidase that is found in the cytoplasm during sporulation (Vacheron et al., 1979 [triangle]). DPP VI has strict specificity for murein peptides with an N-terminal l-Ala. The crystal structure of BcYkfC provides a structural basis for this specificity. DPP VI and BcYkfC have only 21% sequence identity, but the same sets of key residues are conserved, as expected given their similar folds and function (Fig. 5 [triangle]). Thus, DPP VI is clearly a homolog of BcYkfC, except that its SH3b1 domain has no large helical insertion (α1–α3) between βA1 and βA2 and it has a shorter βC–βD loop (Fig. 5 [triangle]). Furthermore, the S2 and S1 binding sites for l-Ala-γ-d-Glu are highly conserved between DPP VI and BcYkfC, including the conserved tyrosine (Tyr118 of YkfC), which is the only residue from SH3b1 whose side chain interacts with l-Ala. The residues from the catalytic domain that contribute to the S2 and S1 sites are identical in BcYkfC and DPP VI (except for an Ala/Ser mutation at position 257 of BcYkfC). Therefore, we conclude that both the B. subtilis and the B. cereus YkfCs are likely to have very similar substrate specificity for murein peptides that contain an N-terminal l-Ala. Enzymes with this specificity cannot cleave PG-biosynthesis precursors such as UDP-MurNAc-pentapeptide and thus are not likely to interfere with the PG-synthesis pathway.

Figure 5
Sequence alignment of YkfC from B. subtilis and B. cereus, dipeptidyl-peptidase VI (DPP VI) from B. sphaericus and a γ-d-glutamyl-l-diamino acid endopeptidase from A. variabilis (AvPCP). The sequence numbering and secondary-structure elements ...

3.7. S′ sites of YkfC and docking studies

B. subtilis YkfC has previously been shown to process the tri­peptide l-Ala-γ-d-Glu-l-Lys, the tetrapeptide l-Ala-γ-d-Glu-l-Lys-d-­Ala and the pentapeptide l-Ala-γ-d-Glu-l-Lys-d-Ala-d-Ala (Schmidt et al., 2001 [triangle]). DPP VI also displayed no particular specificity towards the additional residues attached after DAP (or Lys; Vacheron et al., 1979 [triangle]). Residues in BcYkfC that potentially form the S′ sites include His290, His291, Arg306, Glu308, Arg309, Tyr321 and Glu324. These residues are generally less conserved among close BcYkfC homologs (Fig. 5 [triangle]), suggesting that the specificity of YkfC is determined primarily at the S2 and S1 sites as described above. The catalytic mechanism of NlpC/P60 is currently unknown, but it is likely to be similar to that of papain based on the similar arrangement of their catalytic residues (Anantharaman & Aravind, 2003 [triangle]; Aramini et al., 2008 [triangle]; Xu et al., 2009 [triangle]), except for two interesting variations of important catalytic residues. The third polar residue of the catalytic triad in NlpC/P60 is more often a histidine rather than the asparagine in papain (Bateman et al., 2004 [triangle]). A tyrosine (Tyr226 of BcYkfC) is located at the position equivalent to Gln119 in papain that is thought to be important for catalysis by stabilizing the transition state (Xu et al., 2009 [triangle]; Aramini et al., 2008 [triangle]). Therefore, we docked the tripeptide l-­Ala-γ-d-Glu-l-DAP into the active site of BcYkfC in order to deduce a likely mode of interaction and to assess the effects of the differences (Fig. 6 [triangle] a). The Cδ carboxyl group of γ-d-Glu is displaced owing to the oxidation of Cys238 in the crystal structure. In the modeled structure, Tyr226 OH interacts with the carboxyl group of γ-­d-Glu (Fig. 6 [triangle] b). This docked structure places the Cδ atom close to Cys238 SH (distance ~3.6 Å) and could represent the chemically productive conformation. The electrostatic surface of the binding site is highly complementary to that of the substrate, indicating that the polar and charged interactions are likely to be important for substrate recognition. The basic His290 and Arg306 residues could interact with carboxyl groups. It has previously been reported that the catalytic efficiency of B. subtilis YkfC for l-Ala-γ-d-Glu-Lys is significantly less than that for tetrapeptides or pentapeptides (Schmidt et al., 2001 [triangle]). This observation could be explained by the unfavorable basic environment in the S′ site. Two residues (Arg306/Glu308) at the BcYkfC S1′ site are substituted by Lys/Gly in B. subtilis YkfC (Fig. 5 [triangle]), resulting in a subsite with a positive charge but no negative charge that would be less likely to accommodate the positively charged Lys of l-­Ala-γ-d-Glu-Lys (Fig. 6 [triangle] a).

Figure 6
Models of substrate recognition by BcYkfC. (a) l-Ala-γ-d-Glu-DAP was docked into the active site of BcYkfC. The protein surface is colored according to a gradient in electrostatic potential from negative (red) to positive (blue) (MOE 2008.10; ...

3.8. Recognition of γ-d-Glu by NlpC/P60 cysteine peptidases

NlpC/P60 cysteine peptidases are ubiquitous in bacteria, with ~5000 NlpC/P60 domains in the current Pfam database when metagenomics data are included (Bateman et al., 2004 [triangle]). However, only a few of these domains have been biochemically characterized. In the light of the much clearer picture that we now have of substrate binding in BcYkfC, we examined the sequence conservation of NlpC/P60-family proteins around the active site in order to obtain further insights into substrate specificity across the entire family.

The nonredundant (nr) database of protein sequences was searched using the PSI-BLAST program (Altschul et al., 1997 [triangle]) with the NlpC/P60 domain of BcYkfC as a search probe. We extracted a total of 2599 protein sequences using the criteria of an E value of >0.02 and an alignment length of >70 residues from 3985 total hits, representing a significant subset of NlpC/P60 proteins. Among these proteins, only ~1% do not possess the conserved Cys/His dyad and an additional 9% (with catalytic dyad intact) do not contain the conserved tyrosine that is believed to be essential for catalysis. These proteins are likely to have the NlpC/P60 fold, but either have lost cysteine peptidase activity or are remote homologs that hydrolyze different substrates (e.g. the CHAP family; Anantharaman & Aravind, 2003 [triangle]; Bateman & Rawlings, 2003 [triangle]; Rigden et al., 2003 [triangle]). The sequence-conservation pattern of active-site residues of the remaining 2277 proteins (average sequence identity of 19%) is shown in Fig. 6 [triangle](c). The S sites (S2 and S1) are significantly more conserved than the S′ sites in these proteins. The most conserved site is S1, which is tailored to recognize γ-d-Glu (Fig. 6 [triangle] b). The three most highly conserved residues (Asp237, Ser239 and Tyr226), aside from the catalytic dyad, are involved in hydrogen bonding with γ-d-Glu (Fig. 6 [triangle] b). At position 257, residues with small side chains (Ala/Ser/Thr) are observed, creating a cavity for the carboxyl group of γ-d-Glu. Trp228 contributes to both the S1 and S2 sites and a conserved hydrophobic residue is usually found at this position, which may facilitate interaction with the substrate Ala Cβ. Therefore, the conservation of active-site residues indicates that a significant percentage of the NlpC/P60 family is likely to be specific for γ-d-Glu. More than 1300 proteins which contain the strictly conserved Trp228, Asp237, Ser239 and Tyr226 residues (and catalytic triads) are most likely to bind X-l-Ala-γ-d-Glu moieties (where X is H or other moieties).

3.9. Structural comparison to the cyanobacterial NlpC/P60 l-Ala-γ-d-Glu endopeptidases

We have previously determined the crystal structures of two closely related NlpC/P60 l-Ala-γ-d-Glu endopeptidases from the cyanobacteria A. variabilis (AvPCP) and N. punctiforme (NpPCP). These enzymes contain an N-terminal SH3b domain and a C-terminal NlpC/P60 catalytic domain (Xu et al., 2009 [triangle]). Surprisingly, the structures of the cyanobacterial enzymes are essentially a substructure of YkfC: the two domains of the cyanobacterial enzymes are structurally equivalent to the first and third domains of BcYkfC (Figs. 5 [triangle], 7 [triangle] a and 7 [triangle] b). The full-length AvPCP can be superimposed onto BcYkfC with an r.m.s.d. of 2.3 Å and 29% sequence identity for 193 aligned Cα atoms (Table 2 [triangle]). Thus, the individual SH3b and NlpC/P60 domains, as well as their relative arrangements, are highly similarly in AvPCP and BcYkfC. Furthermore, the residues in the S2 and S1 pockets are nearly identical (except for S257A; Fig. 7 [triangle] c). The striking similarity between the cyanobacterial enzymes and YkfC was not previously detected by sequence analysis owing to the presence of two insertions: the entire SH3b2 domain and the α1–α3 region of SH3b1.

Figure 7
Structural comparisons between BcYkfC and the cyanobacterial NlpC/P60 endopeptidase AvPCP. (a) BcYkfC and AvPCP (PDB code 2hbw; Xu et al., 2009 [triangle]) are shown with the same orientation of their common SH3b1 and NlpC/P60 domains. The structurally ...

The βC–βD loop and the βA1–βA2 loop of SH3b1 are longer compared with SH3b of AvPCP. The βC–βD loop of YkfC is located upstream of the S2 site. The βA1–βA2 loop of YkfC contains three helices (α1–α3) and is located on the S′ side of the binding groove in the catalytic domain. This insertion packs against the surface of the NlpC/P60 domain and contributes to stabilizing the interface between the SH3b1 and NlpC/P60 domains. Furthermore, it is also defines the shape of the S′ binding site, which appears to be more restricted for BcYkfC than for AvPCP (Fig. 7 [triangle] d). Without these insertions to stabilize this domain interface, AvPCP uses a different strategy to achieve a similar purpose: a longer C-terminal loop in NlpC/P60 extends to interact with the SH3b domain.

We previously proposed that the cyanobacterial enzymes have the same substrate requirement as DPP VI based on modeling studies (Xu et al., 2009 [triangle]). The crystal structure of YkfC reported here further supports DPP VI, YkfC and the cyanobacterial enzymes belonging to a subset of NlpC/P60 γ-d-Glu-DAP endopeptidases whose sub­strates possess a free l-Ala at the N-terminus. The SH3b domain helps define the S2 binding site (Tyr118 in BcYkfC and Tyr64 in AvPCP). Additionally, it sterically hinders the docking of a large moiety beyond the S2 site and directly contributes to specificity. This role constitutes a new function for SH3b domains.

3.10. Identification and distribution of YkfC homologs

BcYkfC and cyanobacterial γ-d-Glu-DAP endopeptidases have the same specificity. Other NlpC/P60 enzymes with similar properties could potentially be detected by analyzing sequence similarities in both the SH3b and NlpC/P60 domains. However, owing to the pre­sence of long insertions/deletions and highly divergent sequences in the SH3b domains, it is often difficult to detect similar enzymes using sequence searches and the NlpC/P60 region tends to dominate the hits. We examined an alternate method by examining specific residues in the NlpC/P60 domain only. An acidic residue (Asp256) is essential for YkfC specificity by neutralizing the positive charge on the free amine of l-Ala. Among the 2599 sequences obtained above, 282 proteins were identified that contained an aspartate equivalent to Asp256 and a conserved catalytic dyad. In 239 of the 282 proteins (85%), the conserved tyrosine was maintained in their SH3b domains (Tyr118 of BcYkfC), while this tyrosine was not conserved in enzymes that lacked Asp256. Thus, the presence of an aspartate residue at position 256 and a conserved tyrosine in SH3b is highly correlated (Fig. 6 [triangle] d). Interestingly, NlpC/P60 domains with the con­served aspartate are also found in a few single-domain or multiple-domain proteins that do not contain any detectable SH3b domain. The biochemical properties of these proteins are currently unknown, but they could represent novel variants of a YkfC-type enzyme.

Nevertheless, the 239 proteins above with the conserved aspartate (Asp256) and tyrosine (Tyr118) are most likely to define a YkfC-like subfamily of NlpC/P60 proteins. These NlpC/P60 enzymes are predominately distributed in four phyla of bacteria: bacteroidetes, cyanobacteria, firmicutes and proteobacteria (α-proteobacteria). They include all currently known enzymes with the same requirement for a free N-terminal l-Ala, i.e. DPP VI, YkfC and AvPCP/NpPCP. The cyanobacterial enzymes contain one SH3b domain, while the homologs in other bacteria contain two SH3b domains. As expected, the SH3b domains are highly divergent. The βD strand, in which the conserved tyrosine is located, is more conserved (Fig. 6 [triangle] d). This strand is also responsible for the interface with the catalytic domain. BcYkfC is a member of a cluster of highly conserved homologs that are present in closely related species such as B. anthracis, B. thuringiensis and B. weihenstephanensis (sequence identity of ≥95%). All members of this group contain signal peptides. Signal peptides are also present in some members of the bacteroidetes group, but not in proteobacterial and cyanobacterial homologs.

It is thus likely that the YkfC subfamily of highly specialized enzymes has evolved from a common ancestor, which was likely to have been a P60-like general-purpose enzyme with two or more SH3b domains connected to a C-terminal NlpC/P60 domain. SH3b domains may have lost the function of targeting domains and over time the nonessential SH3b domain would have been lost, as seen in the cyanobacterial enzymes.

4. Conclusions

The structure of BcYkfC in complex with l-Ala-γ-d-Glu is the first structural representative of an NlpC/P60 enzyme with a bound ligand. This structure allowed us to identify the determinants of substrate specificity, which further led to the classification of a subfamily of highly specialized NlpC/P60 enzymes. The studies here also provide structural and functional insights into the NlpC/P60 family of enzymes. Additional information about BcYkfC is available from TOPSAN (Krishna et al., 2010 [triangle]) at

Supplementary Material

PDB reference: YkfC–l-Ala-γ-d-Glu complex, 3h41


This work was supported by the NIH, National Institute of General Medical Sciences, Protein Structure Initiative grant U54 GM074898. Portions of this research were carried out at the Stanford Synchrotron Radiation Lightsource (SSRL). The SSRL is a national user facility operated by Stanford University on behalf of the US Department of Energy, Office of Basic Energy Sciences. The SSRL Structural Molecular Biology Program is supported by the Department of Energy, Office of Biological and Environmental Research and by the National Institutes of Health (National Center for Research Resources, Biomedical Technology Program and the National Institute of General Medical Sciences). Genomic DNA from B. cereus NRS248 (ATCC No. 10987D) was obtained from the American Type Culture Collection (ATCC). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of General Medical Sciences or the National Institutes of Health.


  • Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Nucleic Acids Res.25, 3389–3402. [PMC free article] [PubMed]
  • Anantharaman, V. & Aravind, L. (2003). Genome Biol.4, R11. [PMC free article] [PubMed]
  • Aramini, J. M., Rossi, P., Huang, Y. J., Zhao, L., Jiang, M., Maglaqui, M., Xiao, R., Locke, J., Nair, R., Rost, B., Acton, T. B., Inouye, M. & Montelione, G. T. (2008). Biochemistry, 47, 9715–9717. [PubMed]
  • Asano, S. I., Nukumizu, Y., Bando, H., Iizuka, T. & Yamamoto, T. (1997). Appl. Environ. Microbiol.63, 1054–1057. [PMC free article] [PubMed]
  • Bateman, A., Coin, L., Durbin, R., Finn, R. D., Hollich, V., Griffiths-Jones, S., Khanna, A., Marshall, M., Moxon, S., Sonnhammer, E. L., Studholme, D. J., Yeats, C. & Eddy, S. R. (2004). Nucleic Acids Res.32, D138–D141. [PMC free article] [PubMed]
  • Bateman, A. & Rawlings, N. D. (2003). Trends Biochem. Sci.28, 234–237. [PubMed]
  • Bricogne, G., Vonrhein, C., Flensburg, C., Schiltz, M. & Paciorek, W. (2003). Acta Cryst. D59, 2023–2030. [PubMed]
  • Chen, V. B., Arendall, W. B., Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J., Murray, L. W., Richardson, J. S. & Richardson, D. C. (2010). Acta Cryst. D66, 12–21. [PMC free article] [PubMed]
  • Cohen, A. E., Ellis, P. J., Miller, M. D., Deacon, A. M. & Phizackerley, R. P. (2002). J. Appl. Cryst.35, 720–726. [PMC free article] [PubMed]
  • Cohen, S. X., Morris, R. J., Fernandez, F. J., Ben Jelloul, M., Kakaris, M., Parthasarathy, V., Lamzin, V. S., Kleywegt, G. J. & Perrakis, A. (2004). Acta Cryst. D60, 2222–2229. [PubMed]
  • Collaborative Computational Project, Number 4 (1994). Acta Cryst. D50, 760–763. [PubMed]
  • Crooks, G. E., Hon, G., Chandonia, J. M. & Brenner, S. E. (2004). Genome Res.14, 1188–1190. [PubMed]
  • Cruickshank, D. W. J. (1999). Acta Cryst. D55, 583–601. [PubMed]
  • Doyle, R. J., Chaloupka, J. & Vinter, V. (1988). Microbiol. Rev.52, 554–567. [PMC free article] [PubMed]
  • Eddy, S. R. (1998). Bioinformatics, 14, 755–763. [PubMed]
  • Emsley, P. & Cowtan, K. (2004). Acta Cryst. D60, 2126–2132. [PubMed]
  • Holm, L. & Sander, C. (1995). Trends Biochem. Sci.20, 478–480. [PubMed]
  • Kabsch, W. (1993). J. Appl. Cryst.26, 795–800.
  • Kabsch, W. (2010). Acta Cryst. D66, 125–132. [PMC free article] [PubMed]
  • Kall, L., Krogh, A. & Sonnhammer, E. L. (2007). Nucleic Acids Res.35, W429–W432. [PMC free article] [PubMed]
  • Klock, H. E., Koesema, E. J., Knuth, M. W. & Lesley, S. A. (2008). Proteins, 71, 982–994. [PubMed]
  • Korndorfer, I. P., Danzer, J., Schmelcher, M., Zimmer, M., Skerra, A. & Loessner, M. J. (2006). J. Mol. Biol.364, 678–689. [PubMed]
  • Krishna, S. S., Weekes, D., Bakolitsa, C., Elsliger, M.-A., Wilson, I. A., Godzik, A. & Wooley, J. (2010). Acta Cryst. F66, 1143–1147. [PMC free article] [PubMed]
  • Kuhn, M. & Goebel, W. (1989). Infect. Immun.57, 55–61. [PMC free article] [PubMed]
  • Lesley, S. A. et al. (2002). Proc. Natl Acad. Sci. USA, 99, 11664–11669. [PubMed]
  • Lu, J. Z., Fujiwara, T., Komatsuzawa, H., Sugai, M. & Sakon, J. (2006). J. Biol. Chem.281, 549–558. [PubMed]
  • Marino, M., Banerjee, M., Jonquieres, R., Cossart, P. & Ghosh, P. (2002). EMBO J.21, 5623–5634. [PubMed]
  • Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Acta Cryst. D53, 240–255. [PubMed]
  • Nasertorabi, F., Tars, K., Becherer, K., Kodandapani, R., Liljas, L., Vuori, K. & Ely, K. R. (2006). J. Mol. Recognit.19, 30–38. [PubMed]
  • Pai, C.-H., Chiang, B.-Y., Ko, T.-P., Chou, C.-C., Chong, C.-M., Yen, F.-J., Chen, S., Coward, J. K., Wang, A. H. & Lin, C.-H. (2006). EMBO J.25, 5970–5982. [PubMed]
  • Park, J. T. & Uehara, T. (2008). Microbiol. Mol. Biol. Rev.72, 211–227. [PMC free article] [PubMed]
  • Plewniak, F. et al. (2003). Nucleic Acids Res.31, 3829–3832. [PMC free article] [PubMed]
  • Pohl, E., Holmes, R. K. & Hol, W. G. (1999). J. Mol. Biol.292, 653–667. [PubMed]
  • Rigden, D. J., Jedrzejas, M. J. & Galperin, M. Y. (2003). Trends Biochem. Sci.28, 230–234. [PubMed]
  • Santarsiero, B. D., Yegian, D. T., Lee, C. C., Spraggon, G., Gu, J., Scheibe, D., Uber, D. C., Cornell, E. W., Nordmeyer, R. A., Kolbe, W. F., Jin, J., Jones, A. L., Jaklevic, J. M., Schultz, P. G. & Stevens, R. C. (2002). J. Appl. Cryst.35, 278–281.
  • Schmidt, D. M., Hubbard, B. K. & Gerlt, J. A. (2001). Biochemistry, 40, 15707–15715. [PubMed]
  • Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122. [PubMed]
  • Smith, T. J., Blackman, S. A. & Foster, S. J. (2000). Microbiology, 146, 249–262. [PubMed]
  • Srisailam, S., Lukin, J. A., Lemak, A., Yee, A. & Arrowsmith, C. H. (2006). J. Biomol. NMR, 36, Suppl. 1, 27. [PubMed]
  • Storer, A. C. & Ménard, R. (1994). Methods Enzymol.244, 486–500. [PubMed]
  • Teng, F., Kawalec, M., Weinstock, G. M., Hryniewicz, W. & Murray, B. E. (2003). Infect. Immun.71, 5033–5041. [PMC free article] [PubMed]
  • Uehara, T. & Park, J. T. (2003). J. Bacteriol.185, 679–682. [PMC free article] [PubMed]
  • Uehara, T. & Park, J. T. (2004). J. Bacteriol.186, 7273–7279. [PMC free article] [PubMed]
  • Uehara, T. & Park, J. T. (2007). J. Bacteriol.189, 5634–5641. [PMC free article] [PubMed]
  • Uehara, T., Suefuji, K., Valbuena, N., Meehan, B., Donegan, M. & Park, J. T. (2005). J. Bacteriol.187, 3643–3649. [PMC free article] [PubMed]
  • Vacheron, M. J., Guinand, M., Francon, A. & Michel, G. (1979). Eur. J. Biochem.100, 189–196. [PubMed]
  • Wylie, G. P., Rangachari, V., Bienkiewicz, E. A., Marin, V., Bhattacharya, N., Love, J. F., Murphy, J. R. & Logan, T. M. (2005). Biochemistry, 44, 40–51. [PubMed]
  • Xu, Q. et al. (2009). Structure, 17, 303–313. [PMC free article] [PubMed]

Articles from Acta Crystallographica Section F: Structural Biology and Crystallization Communications are provided here courtesy of International Union of Crystallography