journal menu
research papers
accessNew structural forms of a mycobacterial adenylyl cyclase Rv1625c
aMolecular Biophysics Unit, Indian Institute of Science, C. V. Raman Avenue, Bangalore, Karnataka 560 012, India, and bMolecular Reproduction, Development and Genetics, Indian Institute of Science, C. V. Raman Avenue, Bangalore, Karnataka 560 012, India
*Correspondence e-mail: suguna@mbu.iisc.ernet.in
Rv1625c is one of 16 adenylyl cyclases encoded in the genome of Mycobacterium tuberculosis. In solution Rv1625c exists predominantly as a monomer, with a small amount of dimer. It has been shown previously that the monomer is active and the dimeric fraction is inactive. Both fractions of wild-type Rv1625c crystallized as head-to-head inactive domain-swapped dimers as opposed to the head-to-tail dimer seen in other functional adenylyl cyclases. About half of the molecule is involved in extensive domain swapping. The strain created by a serine residue located on a hinge loop and the crystallization condition might have led to this unusual domain swapping. The inactivity of the dimeric form of Rv1625c could be explained by the absence of the required catalytic site in the swapped dimer. A single mutant of the enzyme was also generated by changing a phenylalanine predicted to occur at the functional dimer interface to an arginine. This single mutant exists as a dimer in solution but crystallized as a monomer. Analysis of the structure showed that a salt bridge formed between a glutamate residue in the N-terminal segment and the mutated arginine residue hinders dimer formation by pulling the N-terminal region towards the dimer interface. Both structures reported here show a change in the dimerization-arm region which is involved in formation of the functional dimer. It is concluded that the dimerization arm along with other structural elements such as the N-terminal region and certain loops are vital for determining the oligomeric nature of the enzyme, which in turn dictates its activity.
Keywords: Rv1625c; adenylyl cyclases; Mycobacterium tuberculosis.
PDB references: Rv1625c, wild type, 4p2m; 4p2x; F363R mutant, 4p2f
1. Introduction
In cell signalling, the importance of second messengers such as cyclic AMP and cyclic GMP in regulating cellular function and homeostasis is well recognized. Adenylyl cyclases convert ATP to cyclic AMP, and these enzymes are classified into six types based on their sequence similarity. Among these six classes of adenylyl cyclases (ACs), class III is the universal class and is present in eukaryotes, bacteria and archaebacteria (Shenoy et al., 2004![[Shenoy, A. R., Sivakumar, K., Krupa, A., Srinivasan, N. & Visweswariah, S. S. (2004). Comp. Funct. Genomics, 5, 17-38.]](../../../../../../logos/arrows/iucrj_arr.gif "Shenoy, A. R., Sivakumar, K., Krupa, A., Srinivasan, N. & Visweswariah, S. S. (2004). Comp. Funct. Genomics, 5, 17-38.")
). There are about ten isoforms of mammalian adenylyl cyclases, of which nine are membrane-bound (mAC; Sunahara et al., 1996) and one is cytoplasmic and is therefore referred to as soluble adenylyl cyclase (sAC; Buck et al., 1999).
Mammalian membrane-bound adenylyl cyclases have a complex structure with 12 transmembrane helices and two cytoplasmic domains. These proteins are arranged in two tandem repeats, with each repeat having six transmembrane helices followed by a cytoplasmic domain. The first cytoplasmic domain is called C1 and the second one is called C2 (Taussig & Gilman, 1995). The structure of a heterodimer comprised of the C1 and C2 domains crystallized in the presence of the GS α subunit and a forskolin molecule was the first structure of the active cyclase enzyme to be reported (Tesmer et al., 1997). The C1 and C2 domains in this heterodimer are arranged in a head-to-tail fashion, like a wreath. The C1 domain contributes the metal-binding aspartates and the C2 domain contributes the substrate-specifying and transition-state-stabilizing residues to the active site formed at the dimer interface. Thus, the C1–C2 heterodimer has only one potential active site, with the other site being a pseudo-active site which is occupied by a forskolin molecule. In an earlier report, an inactive homodimer of C2 domains which lacks the residues required for metal binding was shown to form the same type of wreath-like dimeric structure with two bound forskolin molecules (Zhang et al., 1997). The recent structures of human sAC in complex with a substrate analogue, products and a regulatory bicarbonate molecule revealed the mechanism of catalysis and the activation of sACs by bicarbonate (Kleinboelting et al., 2014).
Bacteria harbour a number of class III nucleotide cyclases. In contrast to the mACs, the bacterial enzymes have a single that contains all of the residues necessary for substrate binding and catalysis. Thus, these proteins form homodimers with two binding sites at the interface as seen in the structures of Spirulina platensis CyaC (sAC; Steegborn et al., 2005), in two mycobacterial ACs, Rv1264 (Tews et al., 2005) and Rv1900c (Sinha et al., 2005), and in the Pseudomonas aeruginosa enzyme (Topal et al., 2012). Crystal structures of the catalytic domains of three soluble guanylyl cyclases (GCs), CYG12 from Chlamydomonas reinhardtii (Winger et al., 2008), Cya2 from Synechocystis PCC6803 (Rauch et al., 2008) and from human (Allerston et al., 2013), have been reported. One exception to the dimeric form of the cyclase enzyme is the trypanosomal enzyme, which crystallized as a monomer (Bieger & Essen, 2001) despite being a catalytically active dimer in solution.
Most of the class III ACs have been found to be fused with various other domains which are involved in signal perception and regulation (Shenoy & Visweswariah, 2006; Linder & Schultz, 2003). An analysis of the mycobacterial genome by Shenoy et al. (2004) showed the presence of at least 16 class III adenylyl cylases. Crystal structures of the functional dimers of two of these cyclases, Rv1264 and Rv1900c, are available. The structure of Rv1264 revealed that the N-terminal domain regulates the activity of the enzyme in different pH conditions (Tews et al., 2005) by causing drastic structural changes. In Rv1900c, a loop at the interface with two different conformations in the two protomers of the dimer acts as a regulator by allowing the substrate to enter only one of the two active sites (Sinha et al., 2005).
The of a third AC from Mycobacterium tuberculosis, Rv1625c, has been cloned and demonstrated to be biochemically active (Guo et al., 2001; Shenoy et al., 2003). Since it has a slightly higher sequence identity (37%) to mammalian guanylyl cyclases (UniProt ID P51841) than to mammalian adenylyl cyclases (UniProt ID P30803; 35.5%), attempts were made to switch the substrate specificity of this enzyme from ATP to GTP (Shenoy et al., 2003). In this process several mutants were generated and biochemically characterized, and the of one such mutant, the triple mutant KFD→ERC, has been reported (Ketkar et al., 2006). This mutant crystallized as a monomer and showed significant movements in the loops harbouring the mutated residues, preventing dimer formation and thus making the enzyme catalytically inactive. A homology model of the dimeric structure of Rv1625c based on mACs (PDB entries 1ab8 and 1azs ; Zhang et al., 1997; Tesmer et al., 1997) revealed the presence of a phenylalanine residue at the subunit interface (Shenoy et al., 2003). This phenylalanine residue was predicted to be involved in dimer stabilization by forming a stacking interaction (Shenoy et al., 2003). To further elucidate the role of this residue in dimer stabilization, a single mutant F363R of Rv1625c (Rv1625c-F363R) was generated. This mutant was shown to be dimeric in solution, with a 3000-fold reduced activity (Ketkar et al., 2006). This may be caused by structural changes at the dimeric interface and/or at the active site. We attempted to crystallize Rv1625c and its mutant Rv1625c-F363R, expecting to obtain crystals of the active dimeric forms of the enzyme. However, Rv1625c-F363R crystallized as a monomer, while the wild-type Rv1625c enzyme crystallized as a dimer with extensive domain swapping. Here, we report these two structures of Rv1625c, which appear to be the forms that the enzyme adopts when it exists in the inactive state. This study also reveals the regulatory roles played by the N-terminal segment and other loops of the structure in modulating the oligomeric state and thus the activity of the enzyme.
2. Materials and methods
2.1. Protein preparation
The of wild-type Rv1625c (Rv1625c-Wt) encompassing residues Met212–Val443 was cloned into pPRO-EX-HT vector to generate a hexahistidine at the N-terminus. The F363R mutant of Rv1625c (Rv1625c-F363R) was generated by performing site-directed mutagenesis as described previously (Shenoy et al., 2003). Expression and purification of the protein was carried out as described previously (Ketkar et al., 2004). The protein obtained after Ni–NTA purification was further purified and its oligomeric state was assessed using gel filtration. Gel filtration was performed using a Superose 12 10/300 GL column (Amersham Pharmacia Biotech) equilibrated with buffer consisting of 20 mM Tris pH 7.4, 10%(v/v) glycerol, 5 mM β-mercaptoethanol (β-ME), 10 mM NaCl on an ÄKTA fast protein (FPLC) system at a flow rate of 0.25 ml min−1. 10% glycerol was required to keep the protein in a stable form. The column was calibrated with standard protein molecular-weight markers from Bio-Rad: bovine γ-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa) and vitamin B12 (1.3 kDa).
2.2. Rv1625c-F363R
2.2.1. Crystallization
The major peak (dimer) of the single mutant eluted from the gel-filtration column (Supplementary Fig. S1a) was concentrated to 3.3 mg ml−1 and used for crystallization. Initial screening for crystals was carried out with commercially available crystal screens from Hampton Research by mixing 2 µl protein solution with 2 µl crystallization condition using the microbatch method. Plate-like crystals were obtained in some of the conditions. These conditions were chosen for further optimization. The ammonium acetate that was present in one of the conditions (condition No. 65) in which crystals appeared was replaced by ammonium sulfate and potassium acetate in further setups. After one week, crystals started growing in a condition consisting of 0.15 M ammonium sulfate, 0.1 M bis-tris pH 5.5, 30% PEG 10K and were allowed to grow further over a period of three weeks before harvesting them.
2.2.2. Data collection
The crystals (Supplementary Fig. S1b) were soaked in a cryoprotectant solution (20% ethylene glycol in the mother liquor) prior to mounting. One of the crystals diffracted to a resolution of 2.05 Å at the home-source X-ray facility with Cu Kα radiation generated using a rotating-anode X-ray generator (Bruker Microstar) and focused with Osmic mirrors. 180 frames were collected with a crystal-to-detector distance of 175 mm at an oscillation angle of 1° per image using a MAR345 detector. The data were processed and scaled using MOSFLM and SCALA from the CCP4 suite, respectively (Winn et al., 2011). The final statistics of data collection and processing are given in Table 1.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2.2.3. Structure solution and refinement
(MR) was attempted by searching for a single molecule in the using Phaser (McCoy et al., 2007). Monomers of mAC (PDB entry 1cjk ; Tesmer et al., 1999) and the triple mutant KFD→ERC of Rv1625c (PDB entry 1yk9 ; Ketkar et al., 2006) that crystallized as a monomer were used as search models, but no solution was obtained. During this time, the of a GC from C. reinhardtii (PDB entry 3et6 ; Winger et al., 2008) became available. The sequence of GC is 45% identical to the sequence of Rv1625c-F363R, whereas that of mAC is only 35% identical. Hence, MR trials were also carried out with the monomer of GC from C. reinhardtii. A unique solution was obtained (Z-score = 8.0 and LLG = 66) with this model.
The MR solution was subjected to 20 cycles of rigid-body followed by ten cycles of using REFMAC5 from the CCP4 suite (Murshudov et al., 2011). Manual model building was carried out using Coot (Emsley et al., 2010) followed by using REFMAC5, and the model was subsequently subjected to automated model building using ARP/wARP (Perrakis et al., 1999) followed by many cycles of manual building and Ethylene were added manually based on a careful examination of the electron-density map. Addition of water molecules and further yielded an R factor and Rfree of 17.9 and 23.3%, respectively. The final are given in Table 1.
2.3. Rv1625c-Wt
2.3.1. Crystallization
Rv1625c-Wt eluted predominantly as a monomeric species on gel filtration (Fig. 1a) and this monomeric fraction was initially used for crystallization. Crystallization trials were performed with the His-tagged protein. No crystals were obtained, probably owing to the flexible nature of the long N-terminal region. Hence, 23 residues at the N-terminus, including the His tag, were removed from the protein using TEV protease. Rv1625c-Wt devoid of His tag was used in further crystallization trials. To stabilize the dimeric form of Rv1625c-Wt, the metal ion Mg2+ (10 mM MgCl2) and the substrate analogue cordycepin (3′-deoxyadenosine) were added. The metal ion was added during purification, while the inhibitor was added in a ten-molar excess to the protein prior to crystallization. Small crystals were obtained after mixing 2 µl protein solution with 2 µl crystallization condition consisting of 0.2 M ammonium sulfate, 0.1 M Tris pH 8.5, 25% PEG 3350 using the microbatch method. This condition was optimized in order to obtain diffraction-quality crystals (Fig. 1b). The lower amount of dimer obtained from gel filtration (Fig. 1a) was subsequently concentrated and used for crystallization without TEV protease treatment.
![[Figure 1]](jt5004fig1thm.gif) | Figure 1 (a) Gel-filtration analysis of Rv1625c-Wt shows that it exists predominantly as a monomer (∼30 kDa) with a small fraction of dimer (∼60 kDa). (b, c) Crystals of Rv1625c-Wt obtained during initial screening of (b) the monomeric fraction and (c) the dimeric fraction. |
2.3.2. Data collection
One of the crystals of the monomeric fraction obtained in the optimized condition diffracted to a resolution of 3 Å at the home source. 120 frames were collected with a crystal-to-detector distance of 275 mm using a MAR345 detector. The same crystal was also subjected to X-ray diffraction at a synchrotron source on ESRF beamline 14, where it diffracted to 2.7 Å resolution. 360 frames with an oscillation angle of 1° per image were collected with a crystal-to-detector distance of 270 mm using a MAR Mosaic 225 CCD detector. Data were processed and scaled using MOSFLM and SCALA from the CCP4 suite, respectively (Winn et al., 2011). The data-collection and processing statistics are given in Table 2.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2.3.3. Structure solution and refinement
Molecular replacement was carried out using the structure of Rv1625c-F363R as a search model. A clear solution with two molecules in the was obtained using Phaser (Z = 27.4 and LLG = 639). During the initial stages of model building, a continuous density between the two subunits was observed. Even after several rounds of model building and the continuous density persisted. It was realised that this density appeared because of extensive domain swapping between the subunits and that the two chains were connected across the subunits. Since the data showed the presence of (twin operator k, h, −l and twin fraction = 0.42), twin correction, as available in REFMAC5.5, was applied in all cycles. The final R factor and Rfree were 20.0 and 25.4%, respectively (Table 2).
Crystals of the dimeric fraction were obtained using the initial screening condition consisting of 0.2 M lithium sulfate, 0.1 M Tris pH 8.5, 25% PEG 4000 (Fig. 1c). One of the crystals diffracted to 2.4 Å resolution at the synchrotron source. The data-collection and processing statistics are given in Table 2. The structure was solved following the procedure that was used to obtain the structure of the monomeric fraction. The final R factor and Rfree are 19.6 and 25.9%, respectively (Table 2). was not detected in this data set.
3. Results and discussion
3.1. Rv1625c-F363R
The single mutant eluted predominantly as a dimer in gel filtration (Supplementary Fig. S1a) which showed much lower activity (63 ± 13 pmol cAMP min−1 mg−1) compared with the wild-type enzyme (199 ± 30 nmol cAMP min−1 mg−1) (Ketkar et al., 2006; Shenoy et al., 2003). The structure solved by MR has one monomer in the It cannot form crystallographic dimers as the protein crystallized in P212121, thus indicating that the protein crystallized as a monomer in spite of being a dimer in solution. The only other AC that crystallized as a monomer is the trypanosomal enzyme (Bieger & Essen, 2001). It was observed that there was a difference in the pH of the buffer used for gel filtration (pH 7.4), at which the enzyme is active, and the pH at which it crystallized (pH 5.5). Thus, it appears that Rv1625c-F363R exists as a monomer at a pH lower than the active pH. However, our attempts to confirm the presence of monomers in solution by gel filtration were not successful since the protein was found to precipitate at the acidic pH of 5.5, close to its pI of 5.7.
3.1.1. of Rv1625c-F363R
Clear electron density was observed for the region Glu237–Lys425, with 42 and 18 residues missing at the N- and C-termini, respectively. The overall fold of the monomer is similar to other reported class III ACs with the conserved β1–α1–α2–β2–β3–α3–β4 core. The N-terminal residues form an extended β-strand by hydrogen bonding to the β-strands of the dimerization arm to give a three-stranded β-sheet (β1′–β4b–β5; Fig. 2a). The β7 and β8 strands along with the interconnecting loop have moved dramatically away from the core of the molecule. This region is also referred to as an `arm' subdomain (Kamenetsky et al., 2006). The α1 helix was found to be shorter than that in mACs (Fig. 2b).
![[Figure 2]](jt5004fig2thm.gif) | Figure 2 Comparison of the Rv1625c-F363R structure with that of mAC (PDB entry 1cjk ). (a) Cartoon representation of the Rv1625c-F363R structure (blue) with the secondary-structural elements labelled. (b) The Rv1625c-F363R structure (blue) superposed on that of mAC (light orange). The regions of Rv1625c-F363R present in different conformations compared with those of mAC are labelled. Images of the structures shown in Figs. 2 to 7 and Supplementary Figs. 2 and 3 were prepared with PyMOL (DeLano, 2002 ). |
The density for the mutated residue Arg363 could be seen clearly (Fig. 3a). Arg363 forms salt bridges with the Glu239 and the substrate-specifying residue Asp365 (Fig. 3b). Because of this salt-bridge interaction between the N-terminal Glu239 and the mutated Arg363 residue, the N-terminal segment is pulled towards the dimer-interface region. When the monomers of Rv1625c-F363R are individually superposed on different subunits of dimeric ACs or GCs, the extended N-terminal segment of one monomer was found to make steric clashes with the residues of helix α2 and strand β2 of the other subunit. Thus, the extended region of the N-terminus of this structure, stabilized by the mutated residue, hinders dimer formation. Interestingly, these interactions appear to resemble those occurring at a spatially equivalent location in the inactive form of Rv1264, another AC present in M. tuberculosis. As shown in Fig. 4(a), a histidine residue in the N-terminal α-helix, called the αN10-switch, of the regulatory domain of Rv1264 is found in a position similar to Glu239 of the N-terminal segment of Rv1625c (Tews et al., 2005). This histidine residue interacts with the catalytic aspartate. Owing to a conformational change in this αN10-switch region, it occupies a different position from that in the active state of the enzyme, preventing the formation of the head-to-tail functional dimer in the inactive form (Fig. 4b).
![[Figure 3]](jt5004fig3thm.gif) | Figure 3 The mutated residue Phe363Arg in Rv1625c-F363R. (a) Arg363 is shown in stick representation with the 2Fo − Fc electron-density map contoured at the 1.5σ level. (b) Residues involved in salt-bridge and hydrogen-bond interactions with Arg363 are shown in stick representation. |
![[Figure 4]](jt5004fig4thm.gif) | Figure 4 (a) Hydrogen-bond interaction of the N-terminal histidine residue of Rv1264 with the substrate-specifying residues. (b) Active (blue) and inactive (yellow–orange) forms of Rv1264. The subunit which moves from the inactive state to the active state is shown with 60% transparency. The arrows show the switch elements α1-switch and αN10-switch of the active and inactive forms, which are highlighted in darker shades. |
The conformation of the β2–β3 loop of Rv1625c-F363R was found to be different when compared with that in mAC (Fig. 2b). The β2–β3 loops meet at the centre of the dimeric interface region and contribute to dimer stabilization (Sinha & Sprang, 2006). Because of this altered conformation, the side chain of Ser298 present in this region makes steric clashes with the β2–β3 loop of the other protomer in the modelled dimer of Rv1625c-F363R.
The β7–β8 loop was found to interact with a symmetry-related molecule. The β3′–α3′ loop (where ′ indicates residues of a symmetry-related molecule) was found to be placed between the β7–β8 loop and the α4 helix (Supplementary Fig. S2a) and thus prevents the arm subdomain from adopting the conformation which is seen in other ACs. This β7–β8 loop conformation is stabilized by numerous van der Waals interactions with the symmetry-related molecule and is further stabilized by the hydrogen bond between His407 and Asp399′ of the β3′–α3′ loop (Supplementary Fig. S2b). The α1 helix was found to interact with the α2′ helix by forming (i) salt bridges from Arg263 to Asp288′ and Glu285′ and (ii) a hydrogen bond between Thr261 and Glu285′. These interactions (Supplementary Fig. S2b) make the α1-helix shift from its position compared with mAC (Fig. 2b).
3.1.2. The active site
The side-chain conformations of some of the substrate-binding residues in this monomeric structure appear to be different from those observed in the active forms of the other reported structures. These side chains are likely to acquire conformations that are optimal for substrate binding in the active form of the enzyme, as observed in the case of human sAC (Kleinboelting et al., 2014) and mAC (Tesmer et al., 1999).
The side chain of the substrate-specifying residue Lys296 is in a different orientation (Fig. 5a) when compared with its mammalian counterpart that was bound to the substrate analogue (active form). The interaction of the carbonyl group of Lys296 with the side chain of the mutated residue Arg363 might be causing this altered orientation (Fig. 3b). The other substrate-binding residue Asp365 is in a similar conformation to its counterpart in active mammalian ACs.
![[Figure 5]](jt5004fig5thm.gif) | Figure 5 (a) Comparison of the active-site residues of Rv1625c-F363R (blue) with active (PDB entry 1cjk ; yellow) and inactive (PDB entry 1ab8 ; pink) forms of mAC. The active-site residues of Rv1625c-F363R are shown in grey and labelled. The substrate analogue ATPαS-Rp and metal ions bound in the active mAC structure are shown in stick and sphere representations, respectively. (b) Salt bridges formed by the catalytic aspartate Asp296 with Arg344 are shown. |
Electron density for the side chain of one of the metal-binding residues, Asp300, was not observed. The second metal-binding residue, Asp256, has a minor conformational change compared with that of the active mammalian enzyme (Figs. 5a and 5b). This conformation of the aspartate is stabilized by a salt bridge to the side chain of Arg344.
The residue Arg376, the predicted function of which is to bind the negatively charged phosphate group of ATP, has its side chain in a different orientation (Fig. 5a) that is not optimal for binding to the phosphate group of the substrate. A similar orientation of this transition-state residue arginine is seen in the case of the inactive mAC C2–C2 homodimer. The ribose-orienting transition-state-stabilizing residue Asn372 is in a similar orientation (Fig. 5a) to Asn1029 of the active mACs (PDB entry 1cjk ).
3.2. Rv1625c-Wt: the domain-swapped dimer
3.2.1. Form 1
Rv1625c-Wt eluted predominantly as a monomer in gel filtration (Fig. 1a) and this monomeric fraction was initially used for crystallization. Two subunits (A and B) were obtained in the with extensive domain swapping. Clear electron density is observed for residues 240–426 of chain A (except for the β1–α2, α3–β4 and β7–β8 loops) and for residues 240–428 of chain B (except for the β1–α2 and β7–β8 loops). All of the residues for which electron density was not observed belong to loop regions of the molecule.
Rv1625c-Wt crystallized as a domain-swapped dimer. In general small regions are involved in domain swapping, but in this case about half of the residues (β5–α4–β6–α5–β7–β8) are swapped between the subunits, leading to a large number of interactions between the subunits (Fig. 6). The dimer has a head-to-head arrangement instead of the head-to-tail arrangement which is seen in the case of dimeric active ACs or GCs. The intact substrate-binding pockets at the active dimer interface are not formed in the swapped dimer. Metal ions and the substrate analogue cordycepin added during crystallization were not seen in the structure because of the altered domain arrangement that is not conducive for binding ligands. Thus, this form of Rv1625c-Wt is an inactive form in spite of being a dimer. Density for the α1 helix was also not seen, indicating that it was disordered in the structure.
![[Figure 6]](jt5004fig6thm.gif) | Figure 6 Domain-swapped dimer of Rv1625c-Wt. The two subunits are coloured differently. Secondary-structural elements are labelled for one of the subunits of Rv1625c-Wt. An OMIT map contoured at the 2.5σ level is shown where the domain swapping occurred. (This view, which is different from that shown in the box, was chosen for clarity.) |
Domain swapping is a frequent phenomenon in oligomeric proteins where one or more structural elements are exchanged between two identical subunits, keeping the overall globular form of the domains unaltered. Several examples of domain swapping can be found in protein crystal structures (Carey et al., 2007; Gronenborn, 2009; Shingate & Sowdhamini, 2012). In a number of cases proline residues were found to occur in hinge loops where swapping often takes place. A few residues at such hinge regions appear to have considerable strain which is released upon domain swapping, rendering more favourable conformations of these residues. In the case of Rv1625c-Wt the dimerization-arm loop (β4–β5 loop) opened up and exchanges the part of the molecule consisting of the β5–α4–β6–α5–β7–β8 structural elements with the other subunit (Fig. 6). Consequently, a new open interface is formed between the α2 helices of each subunit.
A sulfate ion found in the new open interface neutralizes the positive charge present at the interface, interacting with the arginine side chains from both subunits (Supplementary Fig. S3). The open interface is also stabilized by a hydrogen bond formed between Arg274 of chain A and Ser281 of chain B.
There is no proline residue in the hinge-loop region (β4–β5 loop) where the domain swapping takes place. The residue Ser359 present in the hinge loop of Rv1625c-F363R (Fig. 7a), with φ = −123.4°, ψ = −93.0°, falls at the edge of the allowed region of the Ramachandran plot (Fig. 7b). In the Rv1625c-Wt structure, Ser359 in chain A has φ = −76.9°, ψ = 175.2° and that in chain B has φ = −144.6°, ψ = 168.9°, which fall in the favourable regions of the Ramachandran plot (Fig. 7b). Thus, this serine residue with a slightly unfavourable conformation possibly creates a strain in the hinge region which is released upon formation of the swapped dimer.
![[Figure 7]](jt5004fig7thm.gif) | Figure 7 (a) The Rv1625c-F363R structure (yellow) superposed on the domain-swapped dimer of Rv1625c-Wt (green). Ser359 is shown in stick representation in both of the structures. (b) Ser359 in the structures of Rv1625c-F363R (red dot) and Rv1625c-Wt (blue dots) is depicted in the Ramachandran plot. |
Domain swapping can be triggered by changes in external conditions such as solubility and pH. When the solubility of the protein decreases, the protein might open up to release the strain present in the hinge region, allowing the polypeptide chain beyond the hinge to detach from the subunit and interact with another subunit, forming new interactions. This is a slow process since an energy barrier has to be overcome by breaking interactions within the same subunit (Nagradova, 2002). This could be the reason for the long time (two months) taken for the formation of crystals of domain-swapped Rv1625c-Wt. It is likely that this strain, along with the crystallization conditions, which cause a decrease in the solubility of the protein, would have caused the domain swapping in the case of Rv1625c-Wt.
Most of the active residues of the Rv1625c-Wt are in a similar conformation to the Rv1625c-F363R active-site residues except for Arg376, which has a conformation close to its counterpart in active AC structures.
Structure-based sequence alignment of Rv1625c-Wt with other nucleotide cyclases shows the conservation of dimer-interface residues, with the exception of Ser298 (Supplementary Fig. S4). This residue is present in the β2–β3 loop that forms the inner edge of the dimer interface. The structurally equivalent residue to Ser298 of Rv1625c in other class III AC is a hydrophobic residue that forms a hydrophobic interaction with the corresponding residue in the β2–β3 loop of the dimeric partner. This residue thus strengthens the dimer interaction. The residue in this position also contributes to the of the enzyme with the nitrogen base of the substrate. Ser298 of Rv1625c cannot make such interactions and is possibly involved in a different kind of interaction.
3.2.2. Form 2
Crystallization of the dimeric fraction of Rv1625c-Wt obtained during gel filtration was carried out to check whether it forms a domain-swapped dimer like that described above or a different type of dimer. Surprisingly, this dimeric fraction also crystallized as a domain-swapped dimer that was similar to the swapped dimer formed by the monomeric fraction. Clear electron density is observed for residues 240–428 of chain A (except for the β1–α2, β2–β3, α3–β4 and β7–β8 loops) and for residues 240–425 of chain B (except for the β1–α2, α3–β4 and α4–β6 loops). Comparison of forms 1 and 2 of Rv1625c-Wt shows that they form similar structures with small differences. When the protomers of each form are superimposed, it was observed that the other protomers are related by a rotation of 19°. Because of this rotation, the α2 helices of the domains come closer in form 2, precluding the space to accommodate a sulfate ion present in form 1. The possible explanation for the observed inactivity of the dimeric fraction is that it exists as a domain-swapped dimer in solution and cannot convert to the active dimeric state owing to strong interactions in the swapped form. The monomeric form, on the other hand, has no such constraints and forms the active dimer upon substrate binding.
In an earlier study, we reported that the dimeric fraction of Rv1625c-Wt did not show any AC activity (Ketkar et al., 2006). In the absence of the availability of structural information, an attempt was made to explain this lack of activity by speculating that the protomers of the dimer might not be in the correct juxtaposition required for catalysis as in the case of the inhibited form of Rv1264 (Tews et al., 2005). The current study showing the presence of the swapped dimer with an entirely new mode of dimerization which completely breaks the active-site architecture not only provides an explanation for our earlier results but is also consistent with our hypothesis about the lack of activity of the dimer.
To check for monomer–dimer transitions and their stabilities, we collected the monomeric and dimeric fractions separately and performed gel-filtration experiments on each of them. As expected, the monomeric state showed a transition to the dimeric state (Supplementary Fig. S5), whereas the dimeric form was found to be stable over time. The appearance of both fractions initially indicates that dimers are also produced during expression of the recombinant protein. In the presence of substrates and the required metals, the monomers convert to active dimers transiently. The dimers, whether coexisting with the monomers or formed from the monomers, have the same structure as the swapped dimers with no activity.
The two structures reported here along with the triple-mutant structure KFD→ERC reported previously (Ketkar et al., 2006) have different conformations of the dimerization-arm region that are also different from the conformation of this region in active mAC (Fig. 8). In the case of the mutant proteins, the mutated residues present in the dimerization-arm region stabilize the altered conformations of this loop and the β2–β3 loop (Fig. 2b), both of which are involved in dimer stabilization (Sinha & Sprang, 2006). Similar to the triple mutant, the residues present in the β2–β3 loop of Rv1625c-F363R also make steric clashes with the subunit when superposed on the dimeric AC structure in addition to the severe steric clashes made by the extended N-terminal loop. The hinge loop of Rv1625c-Wt involved in domain swapping is also present in the dimerization-arm region. These observations imply that the N-terminal segment, the dimerization arm and the β2–β3 loop play crucial roles in the of the enzyme, thereby affecting its activity.
![[Figure 8]](jt5004fig8thm.gif) | Figure 8 Comparison of the structures of Rv1625c-Wt and its mutants Rv1625c-F363R and KFD→ERC with mAC. The dimerization-arm region is shown in green in each of the structures except for Rv1625c-Wt, in which it is highlighted with a circle. This figure was prepared with UCSF Chimera (Petterson et al., 2004 ). |
Some of the reported crystal structures of cyclases suggest that the head-to-tail dimers of the enzymes adopt different strategies to shift from the inactive to the active form. These strategies include side-chain conformational changes in the binding-site residues (Kleinboelting et al., 2014; Tesmer et al., 1997, 1999), shifts in α-helices and loops (Steegborn et al., 2005; Winger et al., 2008) and rotation of the entire subunit (Sinha et al., 2005; Tesmer et al., 1997; Allerston et al., 2013). The two crystal structures reported here indicate that the mechanism of conversion between the active and inactive forms is different in the case of Rv1625c.
4. Conclusions
Crystal structures of Rv1625c and its single mutant Rv1625c-F363R have been determined. Both of them crystallized in inactive forms. The mutant crystallized as a monomer even though it exists as a dimer in solution. The extended N-terminus of the mutant, stabilized by the mutated residue Phe363Arg, hinders dimer formation. The wild-type enzyme exists predominantly as a monomer in solution but crystallized as a domain-swapped dimer. The geometry of the active site is lost in the head-to-head arrangement of the domain-swapped dimer, rendering it inactive. The minor inactive dimeric fraction in solution also has a similar swapped arrangement in the crystalline state. These observations lead to the conclusion that the transition of the monomeric enzyme to the active dimeric form takes place upon activation, whereas the preformed dimers in solution cannot be triggered to become functional dimers by the mere addition of the substrate. At this stage the biological significance of the swapped dimers is unknown. It is clear that the swapped dimers cannot perform the adenylyl cyclase activity but it can be speculated that they might play a different role such as interaction with other molecules. The crystal structures presented here also provide new insights into the inherent flexibility of the molecule, which is likely to be related to its function. The function of the molecule appears to be regulated by the N-terminal region, the dimerization arm and the β2–β3 loop, which affect the oligomeric nature of the enzyme.
Guo et al. (2001) observed that at higher concentrations of the of Rv1625c, the presence of dimers and other higher oligomeric states increases, causing reduced activity. It is not possible to guess whether these dimers were formed by the swapped catalytic domains, but one cannot rule out the possibility that these could be swapped dimers in the light of the present study. In the full-length protein with an additional six transmembrane helices, the linker region between the last helix and the is about 40 residues in length and can easily facilitate formation of either the head-to-tail domain or the swapped domain. However, determination of the nature of the dimers in vivo, whether swapped or in the active head-to-tail form, may not be trivial but can be attempted with techniques such as single-molecule fluorescence resonance (smFRET). This method has successfully been used to distinguish between two conformational states of a recombinant SNARE protein (Sakon & Weninger, 2010). As it has been shown that full-length Rv1625c could be targeted to the cell membrane of HEK293 cells (Guo et al., 2001), it is possible to specifically label the protein by cysteine mutations at appropriate positions with acceptor and donor molecules using the available information from the crystal structures of both active and swapped forms of the dimers. These labelled molecules can then be microinjected into the cell and the resulting amount of FRET signal could be examined to distinguish between the type of oligomers formed in the cell.
The coordinates and structure factors have been deposited in the Protein Data Bank with accession codes 4p2m and 4p2x for Rv1625c-Wt forms 1 and 2, respectively, and 4p2f for Rv1625c-F363R.
Supporting information
PDB references: Rv1625c, wild type, 4p2m; 4p2x; F363R mutant, 4p2f
Supplementary files. DOI: 10.1107/S2052252514016741/jt5004sup2.pdf
Acknowledgements
Preliminary data were collected at the National X-ray Data Collection Facility at the Molecular Biophysics Unit, Indian Institute of Science and the high-resolution data were collected at beamline station BM14 of the European Synchrotron Radiation Facility (ESRF) funded by the Department of Biotechnology (DBT). Financial support from the Council of Scientific and Industrial Research, Government of India is acknowledged.
References
Allerston, C. K., von Delft, F. & Gileadi, O. (2013). PLoS One, 8, e57644. CrossRef PubMed Google Scholar
Bieger, B. & Essen, L.-O. (2001). EMBO J. 20, 433–445. Web of Science CrossRef PubMed CAS Google Scholar
Buck, J., Sinclair, M. L., Schapal, L., Cann, M. J. & Levin, L. R. (1999). Proc. Natl Acad. Sci. USA, 96, 79–84. CrossRef CAS PubMed Google Scholar
Carey, J., Lindman, S., Bauer, M. & Linse, S. (2007). Protein Sci. 16, 2317–2333. Web of Science CrossRef PubMed CAS Google Scholar
DeLano, W. L. (2002). PyMOL. https://www.pymol.org . Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501. Web of Science CrossRef CAS IUCr Journals Google Scholar
Gronenborn, A. M. (2009). Curr. Opin. Struct. Biol. 19, 39–49. Web of Science CrossRef PubMed CAS Google Scholar
Guo, Y. L., Seebacher, T., Kurz, U., Linder, J. U. & Schultz, J. E. (2001). EMBO J. 20, 3667–3675. Web of Science CrossRef PubMed CAS Google Scholar
Kamenetsky, M., Middelhaufe, S., Bank, E. M., Levin, L. R., Buck, J. & Steegborn, C. (2006). J. Mol. Biol. 362, 623–639. Web of Science CrossRef PubMed CAS Google Scholar
Ketkar, A. D., Shenoy, A. R., Kesavulu, M. M., Visweswariah, S. S. & Suguna, K. (2004). Acta Cryst. D60, 371–373. Web of Science CrossRef IUCr Journals Google Scholar
Ketkar, A. D., Shenoy, A. R., Ramagopal, U. A., Visweswariah, S. S. & Suguna, K. (2006). J. Mol. Biol. 356, 904–916. Web of Science CrossRef PubMed CAS Google Scholar
Kleinboelting, S., Diaz, A., Moniot, S., van den Heuvel, J., Weyand, M., Levin, L. R., Buck, J. & Steegborn, C. (2014). Proc. Natl Acad. Sci. USA, 111, 3727–3732. CrossRef CAS PubMed Google Scholar
Linder, J. U. & Schultz, J. E. (2003). Cell. Signal. 15, 1081–1089. Web of Science CrossRef PubMed CAS Google Scholar
McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674. Web of Science CrossRef CAS IUCr Journals Google Scholar
Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355–367. Web of Science CrossRef CAS IUCr Journals Google Scholar
Nagradova, N. K. (2002). Biochemistry, 67, 839–849. Web of Science PubMed CAS Google Scholar
Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Nature Struct. Biol. 6, 458–463. Web of Science CrossRef PubMed CAS Google Scholar
Pettersen, E. F., Goddard, T. D., Huang C. C., Couch. G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E. (2004). J. Comput. Chem. 25, 1605–1612. Google Scholar
Rauch, A., Leipelt, M., Russwurm, M. & Steegborn, C. (2008). Proc. Natl Acad. Sci. USA, 105, 15720–15725. Web of Science CrossRef PubMed CAS Google Scholar
Sakon, J. J. & Weninger, K. R. (2010). Nature Methods, 7, 203–205. CrossRef CAS PubMed Google Scholar
Shenoy, A. R., Sivakumar, K., Krupa, A., Srinivasan, N. & Visweswariah, S. S. (2004). Comp. Funct. Genomics, 5, 17–38. Web of Science CrossRef PubMed CAS Google Scholar
Shenoy, A. R., Srinivasan, N., Subramaniam, M. & Visweswariah, S. S. (2003). FEBS Lett. 545, 253–259. Web of Science CrossRef PubMed CAS Google Scholar
Shenoy, A. R. & Visweswariah, S. S. (2006). FEBS Lett. 580, 3344–3352. Web of Science CrossRef PubMed CAS Google Scholar
Shingate, P. & Sowdhamini, R. (2012). PLoS One, 7, e39305. CrossRef PubMed Google Scholar
Sinha, S. C. & Sprang, S. R. (2006). Rev. Physiol. Biochem. Pharmacol. 157, 105–140. Web of Science CrossRef PubMed CAS Google Scholar
Sinha, S. C., Wetterer, M., Sprang, S. R., Schultz, J. E. & Linder, J. U. (2005). EMBO J. 24, 663–673. Web of Science CrossRef PubMed CAS Google Scholar
Steegborn, C., Litvin, T. N., Levin, L. R., Buck, J. & Wu, H. (2005). Nature Struct. Mol. Biol. 12, 32–37. Web of Science CrossRef CAS Google Scholar
Sunahara, R. K., Dessauer, C. W. & Gilman, A. G. (1996). Annu. Rev. Pharmacol. Toxicol. 36, 461–480. CrossRef CAS PubMed Google Scholar
Taussig, R. & Gilman, A. G. (1995). J. Biol. Chem. 270, 1–4. CrossRef CAS PubMed Google Scholar
Tesmer, J. J. G., Sunahara, R. K., Gilman, A. G. & Sprang, S. R. (1997). Science, 278, 1907–1916. CrossRef CAS PubMed Web of Science Google Scholar
Tesmer, J. J. G., Sunahara, R. K., Johnson, R. A., Gosselin, G., Gilman, A. G. & Sprang, S. R. (1999). Science, 285, 756–760. Web of Science CrossRef PubMed CAS Google Scholar
Tews, I., Findeisen, F., Sinning, I., Schultz, A., Schultz, J. E. & Linder, J. U. (2005). Science, 308, 1020–1023. Web of Science CrossRef PubMed CAS Google Scholar
Topal, H., Fulcher, N. B., Bitterman, J., Salazar, E., Buck, J., Levin, L. R., Cann, M. J., Wolfgang, M. C. & Steegborn, C. (2012). J. Mol. Biol. 416, 271–286. Web of Science CrossRef CAS PubMed Google Scholar
Winger, J. A., Derbyshire, E. R., Lamers, M. H., Marletta, M. A. & Kuriyan, J. (2008). BMC Struct. Biol. 8, 42. Google Scholar
Winn, M. D. et al. (2011). Acta Cryst. D67, 235–242. Web of Science CrossRef CAS IUCr Journals Google Scholar
Zhang, G., Liu, Y., Ruoho, A. E. & Hurley, J. H. (1997). Nature (London), 386, 247–253. CrossRef CAS PubMed Web of Science Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.
access
