Structure of the PII signal transduction protein of Neisseria meningitidis at 1.85 Å resolution

The structure of the PII signal transduction protein of N. meningitidis at 1.85 Å resolution is described.


Introduction
The signal transduction protein P II (GlnB) is best known for its role in the regulation of nitrogen assimilation in enteric bacteria. However, closely related proteins are found in a wide variety of organisms including bacteria, archaea and plants (Arcondeguy et al., 2001;Ninfa & Jiang, 2005;Ninfa & Atkinson, 2000). In Escherichia coli, P II and its paralogue GlnK, which is only induced under nitrogen limitation, are involved in two signalling pathways that regulate the activity of glutamine synthetase. Firstly, the bifunctional uridylyltransferase/uridylyl-removing enzyme (UTase/UR, the product of glnD) uridylates a conserved tyrosine (Tyr51) of P II under nitrogenstarvation conditions, whereas under circumstances of nitrogen excess the enzyme removes UMP from P II (Jaggi et al., 1996;Jiang et al., 1998a). Native P II activates the adenylation activity of ATase, which in turn inhibits glutamine synthetase, whilst UMP-P II reverses this effect (Jaggi et al., 1997;Mangum et al., 1973). Secondly, P II affects glutamine synthetase transcription by modulating the activity of the kinase/phosphatase NRII (the ntrB gene product), which controls the phosphorylation status of the transcription factor NRI (the ntrC gene product). In its phosphorylated form, NRI stimulates transcription of the glutamine synthetase gene, whilst the unphosphorylated NRI acts as a repressor. Native P II binds to NRII, preventing phosphorylation of NRI and hence the activation of GS transcription (Liu & Magasanik, 1995). UMP-P II , which is generated under nitrogen-limiting conditions through uridylation of the T-loop, which extends beyond an otherwise compact trimeric structure, does not bind NRII and hence activation of GS via phosphorylated NRI can proceed (Jiang et al., 1998b). In addition, E. coli P II binds cooperatively to two effectors, ATP and -ketogluturate (Jiang et al., 1998a). The binding of -ketogluturate regulates the interaction between P II and ATase, thus integrating signals from carbon and nitrogen metabolism.
Although in E. coli the P II paralogue GlnK (van Heeswijk et al., 1995Heeswijk et al., , 1996 appears to have similar properties to P II and can form heterotrimers with P II (van Heeswijk et al., 2000), GlnK also has distinct roles. In many bacteria, including E. coli, GlnK is associated with regulation of the amtB gene, which encodes the ammonium transporter (Coutts et al., 2002). GlnK is also involved in the NifL-NifA regulatory system of nitrogen-fixing bacteria (Little et al., 2000).
The X-ray crystal structures of both P II (Cheah et al., 1994;Xu et al., 2001) and GlnK (Xu et al., 1998) from E. coli, as well as a number of other P II proteins [those from Herbaspirillum seropedicae (Machado Benelli et al., 2002), Synechoccus sp. PCC7942 and Synchocystis sp. PCC6803 (Xu et al., 2003), Thermus thermophilus (Sakai et al., 2005) and Thermotoga maritima (Schwarzenbacher et al., 2004)], have been solved. All share a highly conserved monomer structure arranged into a tightly associated trimer. A key feature of the structures is the so-called T-loop which contains the regulatory uridylylation site (Tyr51) or, in the case of cyanobacteria, a phosphorylation site (Ser49; Forchhammer & Tandeau de Marsac, 1995). The T-loop has been implicated in the protein-protein interactions of P II in E. coli Martinez-Argudo & Contreras, 2002) and shows differences in conformation in the crystal structures.
Neisseria spp. are Gram-negative -protobacteria which include many species found only in humans, including successful pathogens. In recent years, the genomes of N. meningitidis serotypes A (strain Z2491; Parkhill et al., 2000) and B (strain MC58; Tettelin et al., 2000) and N. gonorrhoeae (strain FA1090; currently unpublished work) have been sequenced and annotated. Each contain a single P II protein encoded by a monocistronic operon (NMB1955 in N. meningitidis strain MC58). The three neisserial P II proteins have 98% identical sequences and share 73% identity to the P II from E. coli. As part of a structural proteomics approach to the study of Neisseria, we have solved the structure of P II from N. meninigitidis (gene locus NMB1995) to 1.85 Å resolution using the semi-automated pipeline of the Oxford Protein Production Facility (OPPF).

Materials and methods
Cloning, expression and protein purification followed standard OPPF pipeline protocols, as described previously (Ren et al., 2005). Briefly, the P II gene (NMB 1955) was amplified from genomic DNA by PCR with the forward primer ggggacaagtttgtacaaaaaagcaggcttcctggaag-ttctgttccagggcccgATGAAAAAAATCGAGGCGATTGTC and the reverse primer ggggaccactttgtacaagaaagctgggtctcaTCAGACTGC-CGCGTCCGAAC incorporating an N-terminal His tag followed by a 3C protease-cleavage site and inserted into the expression vector pDEST17 using Gateway recombinatorial cloning (Invitrogen). Expression was induced by the addition of 0.5 mM IPTG and the protein was purified by a combination of Ni-NTA affinity chromatography and gel filtration. The N-terminal His tag was removed by cleavage with 3C protease prior to gel filtration. The protein was crystallized using the nanodrop crystallization procedure with standard OPPF protocols (Walter et al., 2003). Hits from this initial screening exercise were then tested in-house using a MAR 345 structural genomics communications Acta Cryst.  (a) ClustalW alignment of P II paralogue protein sequences, numbered relative to the N. meningitidis P II sequence, with the T-loop (amino acids 37-55), B-loop (amino acids 81-90) and C-loop (amino acids 96-112) marked. The alignment view was generated with JALVIEW, with colouring according to the Zappo colour scheme, but applied only to residues showing a population identity of !80%. (b) 'New-cartoon' format C -trace overlay of N. meningitidis P II and five other selected P II paralogue structures, illustrating tight conservation of the core region and the location of the B-, C-and T-loops relative to the ATP-binding cleft. N. meningitidis P II , PDB code 2gw8, green; E. coli P II /GlnB, PDB code 2pii, red; E. coli GlnK, PDB code 2gnk, orange; H. seropedicae GlnK, PDB 1hwu, yellow; Synechococcus sp. GlnB, PDB code 1qy7, purple; T. thermophilus TT021, PDB code 1v9o, dark blue. (c) 'New-cartoon' format C trace of N. meningitidis P II coloured by chain showing a close-up view of the ATP-binding site, with conserved residues labelled according to the standard single-letter sequence code and displayed in 'liquorice' format coloured purple. ATP is also shown in 'liquorice' format coloured by name; the coordinates were derived from a C -trace overlay of E. coli ATP-bound structure PDB code 2gnk and N. meningitidis P II . (d) Cutaway view 'new-cartoon' format C -trace of N. meningitidis P II coloured by chain, with selected residues labelled according to the standard single-letter sequence code and displayed in 'liquorice' format and coloured by name, illustrating the central Lys60/Glu62 oligomerization contact and the conserved hydrophobic pocket. ATP is also shown in 'liquorice' format coloured orange, with coordinates derived from a C -trace overlay of E. coli ATP-bound structure PDB code 2gnk and N. meningitidis P II . image-plate system on a Rigaku generator equipped with a Cu anode and Osmic multilayer optics, giving Cu K radiation with = 1.5418 Å . The best diffraction observed (d min < 2.0 Å ) was from a 75 Â 25 Â 25 mm rod-shaped crystal grown in Hampton Cryo Screen I condition No. 31 [0.17 M ammonium sulfate, 25.5% PEG 4000, 15%(v/v) glycerol]. This crystal was therefore frozen and dry-shipped to Daresbury for data collection. Indexing, integration and merging of data images were carried out with DENZO and SCALEPACK (Otwinowski & Minor, 1997). Rotation-function searches, translation searches and initial rigid-body Patterson correlation refinement were carried out using CNS (Brü nger et al., 1998) and molecularreplacement solutions were checked by displaying the transformed coordinates in O, as described in Jones et al. (1991). Rigid-body, positional and B-factor refinement, simulated annealing and initial water picking were carried out in CNS. Manual rebuilding, including insertion of ions, ligands and extra water molecules, was carried out using the program O. Homologous sequences to N. meningitidis P II were selected from the Integrated Microbial Genomes (IMG) database (http://img.jgi.doe.gov/cgi-bin/pub/main.cgi), aligned with ClustalW (Thompson et al., 1994;Chenna et al., 2003) and visualized using JALVIEW (Clamp et al., 2004). Model quality was assessed using PROCHECK (Laskowski et al., 1993). The final N. meningitidis P II model was overlaid with the previously released GlnB and GlnK structures using TOPP (Collaborative Computational Project, Number 4, 1994) and the results were compared visually in O and VMD (Humphrey et al., 1996). Final figures were prepared from VMD screenshots using Corel11.

Results and discussion
The structure of P II from N. meningitidis was determined by molecular replacement using the PDB model 1pil (E. coli P II ) to a resolution of 1.85 Å ( Table 1). As with the E. coli crystal structure, P II from N. meningitidis has a single molecule in the asymmetric unit. Each molecule of the trimer (the normal biological oligomeric state) is therefore in a crystallographically equivalent environment, indicating that all three chains have the same conformational state. The flexible T-loop (residues 37-55) is semi-disordered from residues 38 to 52 in the N. meningitidis P II crystal structure, although sufficient residual density was still visible in low-contoured difference maps (2.2 F o À F c density) to allow an approximate fit based on a rigidbody overlay of the T-loop from E. coli P II . This section is thus included in Figs. 1(b), 1(c) and 1(d) for comparative purposes, but is omitted from the final deposited coordinates (PDB code 2gw8).
Comparing C -trace overlays of N. meningitidis P II with five additional P II structures obtained from the PDB, the mean C r.m.s.d. for the core section, excluding the flexible T-loop, is just 0.9 Å , with a range of 0.7-1.0 Å (Fig. 1b; N. meningitidis P II is in green, E. coli P II /GlnB is in red, E. coli GlnK is in orange, H. seropedicae GlnK is in yellow, Synechococcus sp. GlnB is in purple and T. thermophilus TT021 is in dark blue). The overall fold is thus very tightly conserved across the whole population of P II proteins. The N. meningitidis P II T-loop tracing also indicates a similar conformation to that observed for E. coli P II and the T-loop uridylation site (Tyr51) is also conserved.
As discussed by Xu et al. (1998), analysis of the total population of available P II sequences shows that the most highly conserved residues map to the P II ATP-binding site, which is formed by the B-loop of one subunit together with the C-loop and sequences at either end of the T-loop from the adjacent subunit (Xu et al., 1998;Schwarzenbacher et al., 2004; the T-loop, B-loop and C-loop clusters are shown in Fig. 1a). As expected, the key contact residues identified by this analysis (Gly27, Thr29, Gly35, Lys58, Gly87, Gly89, Lys90, Arg101 and Arg103) are conserved in the N. meningitidis P II sequence and the crystal structure shows they form a pocket equivalent to that seen in E. coli (Fig. 1c), indicating that the Neisseria P II protein is likely to bind to ATP. Analysis of mutants (Jiang, Zucker, Atkinson et al., 1997) and the mode of binding of the inhibitor 2-oxo-3-pentynoate to 4-oxalocrotonate tautomerase (Taylor et al., 1998) have also previously been combined to suggest that Gly37, Arg38, Gln39, Lys40, Thr83, Gly84, Gly89, Lys90 and Arg101 form the 2-oxoglutarate binding site (Machado Benelli et al., 2002). As can be seen in Fig. 1(a), all of these residues are conserved in the N. meningitidis P II sequence apart from Thr83, which has undergone a neutral mutation to Ser83, thus indicating that the Neisseria P II protein is also likely to bind 2-oxoglutarate.
In addition to these features, our analysis of multiple ClustalW alignments using sets of sequences from diverse species such as those illustrated in Fig. 1(a) has revealed a number of other patterns. Firstly, apart from residues such as Thr29 or Gly35, which are involved in both ATP binding and oligomerization, most residues forming oligomerization contacts are not strictly conserved across the whole population of P II proteins; in part, this reflects the fact that many such contacts are formed between backbone atoms such that the nature of the side chain does not substantially affect the oligomerization. However, for the remaining contacts that do involve side-chain interactions then, as seen by Machado Benelli and coworkers in their comparison of the H. seropedicae and E. coli GlnK structures, each pair of structures compared does have some contacts in common (Machado Benelli et al., 2002). P II oligomerization-interface contacts therefore tend to show clustering of residue type, with significantly retarded genetic drift relative to the mean difference, e.g. the central oligomerization contact between Lys/Arg60 and Asp/Glu62 (Fig. 1d). This pattern suggests a strong selective pressure for maintaining the viability of normal trimer formation, which is in line with expectation as such an assembly is believed to be important for the function of P II proteins (Zhang et al., 2004). Secondly, when the residues not involved in ATP or 2-oxyglutarate binding but showing >80% identity are mapped onto our crystal structure, they outline a hydrophobic pocket on the other side of the -sheet to the ATP-binding site, together with Ile7 on the same face as the ATP site (Fig. 1d). This pattern is strongly conserved, implying that there is a functional significance to the arrangement. It is possible that the conserved residues define a binding site for an unknown ligand. Since the potential pocket is on the opposite side of the -sheet to the ATP site, such a ligand might function as an allosteric effector regulating P II function by modulating access to the ATP-binding site. An alternative and more speculative interpretation is that the conserved residues are important in stabilizing a conformational change in the molecule. Such a change might occur during interaction with partner proteins. In this case, the hydrophobic side chains currently separated from one another and forming an open pocket might be brought into contact, forming an interdentate cross-link and making the P II protein significantly more rigid.
The interacting partners of the P II in Neisseria are currently unknown. However, given the structural conservation of the protein, some candidates can be proposed based on E. coli GlnB/GlnK. The genomes of Neisseriae encode putative UTase (e.g. NMB1203, 31% sequence identity to E. coli) and GS-ATase enzymes (e.g. NMB0224, 35% sequence identity to E. coli) and we suggest that these interact with N. meningitidis P II . The nature of the downstream effectors that functionally correspond to E. coli NRI/NRII in Neisseria spp. is less clear. The likely candidates are the co-transcribed genes that encode NtrX and NtrY (NMB0114/NMB0115), although NtrY only shows 24% sequence identity to E. coli NRII. Interestingly, the Neisseria NtrY protein is very similar (45% sequence identity) to another 2-component regulator, the atoS gene in the related species Chromobacterium violaceum. In E. coli, AtoS regulates the expression of AtoC involved in short-chain fatty-acid metabolism. In C. violaceum, which has both PII and GlnK homologues, atoS is located adjacent to ntrX in the genome, reminiscent of NtrB/NtrC in E. coli and NtrX/ NtrY in Neisseria species. Since Neisseria spp. do not appear to have separate NtrA, NtrB, atoS and atoC genes, we speculate that the single N. meningitidis P II may be involved in the regulation of more than one sensor system via NtrY.
The model for the regulatory role of P II is an example of direct sensing and action in which the mechanism of activation and action is dependent upon the direct modification of the sensing protein rather than upon the transcriptional control of the protein itself. Consistent with this, transcriptional profiling normally does not detect the transcript from this gene unless very high data depths (>80%) are obtained, suggesting that it is a relatively low-abundance transcript. Furthermore, it is seldom shown to alter its expression, with the only observed changes to date being between early-log and late-log and between mid-log and late-log phase cultures, but these changes are between 1.9-fold and 1.4-fold induction and are not highly significant (p < 0.05) (unpublished observations). This suggests that this gene is induced in conditions in which protein/amino-acid supply is restricted. A low constitutive expression that does not often change under differing conditions is consistent with its primary regulatory role being controlled by changes in a pre-formed protein.