CcaA is a β-carbonic anhydrase (CA) that is a component of the carboxysomes of a subset of β-cyanobacteria. This protein, which has a characteristic C-terminal extension of unknown function, is recruited to the carboxysome via interactions with CcmM, which is itself a γ-CA homolog with enzymatic activity in many, but not all cyanobacteria. We have determined the structure of CcaA from Synechocystis sp. PCC 6803 at 1.45 Å. In contrast with the dimer-of-dimers organization of most bacterial β-CAs, or the loose dimer-of-dimers-of-dimers organization found in the plant enzymes, CcaA shows a well-packed trimer-of-dimers organization. The proximal part of the characteristic C-terminal extension is ordered by binding at a site that passes through the two-fold symmetry axis shared with an adjacent dimer; as a result, only one of a pair of converging termini can be ordered at any given time. Docking in Rosetta failed to find well-packed solutions, indicating that formation of the CcaA/CcmM complex probably requires significant backbone movements in at least one of the binding partners. Surface plasmon resonance experiments showed that CcaA forms a complex with CcmM with sub-picomolar affinity, with contributions from residues in CcmM's αA helix and CcaA's C-terminal tail. Catalytic characterization showed CcaA to be among the least active β-CAs characterized to date, with activity comparable with the γ-CA, CcmM, it either complements or replaces. Intriguingly, the C-terminal tail appears to partly inhibit activity, possibly indicating a role in minimizing the activity of unencapsulated enzyme.

Introduction

Cyanobacteria actively concentrate inorganic carbon by use of energy requiring CO2 or pumps, thereby increasing the efficiency of carboxylation by ribulose-1,5-bisphophate carboxylase/oxygenase (RubisCO) [13]. The seeming straightforwardness of this mechanism is complicated by the simultaneous constraints that only can be effectively concentrated in the cytosol (due to the high lipid permeability of CO2), but only CO2 is a substrate for RubisCO. Cyanobacteria resolve this dilemma by encapsulating RubisCO within a CO2 impermeable protein shell, forming a carboxysome. Inorganic carbon is accumulated in the cytosol as and diffuses through the carboxysome shell through small pores. The need for the entering the carboxysome to be efficiently converted into CO2 requires that carbonic anhydrase (CA; E.C. 4.2.1.1.) be co-encapsulated within the carboxysome. On the other hand, the presence of CA activity in the cytosol is highly deleterious as it converts the accumulated into its membrane-permeable CO2 counterpart outside of the protective barrier afforded by the shell [4]. Intracellular CAs in cyanobacteria are, therefore, required to possess interaction determinants that target them to the carboxysome, as well as mechanisms that minimize CA activity prior to the completion of the shell.

Carboxysomes come in two deeply divergent variants termed α- (characterized by encapsulating type Ia RubisCO) and β-carboxysomes (characterized by encapsulating type Ib RubisCO). In α-carboxysomes, the role of CA is fulfilled by CsoSCA [5], a highly divergent β-CA [6] which is tightly associated with the α-carboxysome shell [7]. In β-carboxysomes, the situation is more complicated in that there are two proteins, CcmM and CcaA (sometimes present as two copies), which can individually or together fulfill the role of the CA. CcaA was discovered early in the history of carboxysome research as a protein encoded in a DNA fragment that rescues high CO2 requiring mutants [8,9]; however, the corresponding gene is only found in about two-thirds of available β-cyanobacterial genomes. This protein has historically been considered a fairly close homolog of plant β-CAs, though the recent explosion in sequenced bacterial genomes has revealed the closest homologs outside of cyanobacteria being from bacteria such as Methylocystis rosea (56% identity) or Pseudomonas aeruginosa (where PA0102 is 48% identical). The most closely related protein with a known structure remains, however, the pea chloroplast β-CA, which has only 32% sequence identity with Synechocystis sp. PCC 6803 CcaA [10] (Figure 1). Sequence comparisons between CcaA and these close homologs reveal that CcaA possesses a characteristic C-terminal extension of ∼80 amino acids where only the first and last 15–20 amino acids show significant conservation (Figure 1). The role of this extension is unknown.

Multiple sequence alignment of CcaA homologs.

Figure 1.
Multiple sequence alignment of CcaA homologs.

CcaA sequences are SYNY3 Synechocystis sp. PCC 6803 (gi|16330758), SYNE7 Synechococcus elongatus PCC 7942 (gi|46129892), NOSP7 Nostoc punctiforme PCC 73102 (gi|23128982), MICAE Microcystis aeruginosa PCC 9443 (_1 is gi|389730592, _2 is gi|488860576) and LEPBY Leptolyngbya boryana (_1 is gi|515855206, _2 is gi|515859155), whereas PEA denotes the chloroplastic β-CA from P. sativum. Cyan circles mark zinc-binding residues and blue circles mark known catalytic residues. Orange circles denote residues that mediate interactions between dimers, yellow circles denote residues that map to the potential interface with CcmM (see Figure 3). A green box marks the C-terminal conserved sequence, whereas the magenta box marks the residues that, in Synechocystis CcaA, are ordered at the two-fold symmetry axis. Note that in strains with two CcaA sequences one is typically ∼270 amino acids, whereas the other is ∼210 amino acids and is missing the C-terminal extension normally characteristic of CcaA.

Figure 1.
Multiple sequence alignment of CcaA homologs.

CcaA sequences are SYNY3 Synechocystis sp. PCC 6803 (gi|16330758), SYNE7 Synechococcus elongatus PCC 7942 (gi|46129892), NOSP7 Nostoc punctiforme PCC 73102 (gi|23128982), MICAE Microcystis aeruginosa PCC 9443 (_1 is gi|389730592, _2 is gi|488860576) and LEPBY Leptolyngbya boryana (_1 is gi|515855206, _2 is gi|515859155), whereas PEA denotes the chloroplastic β-CA from P. sativum. Cyan circles mark zinc-binding residues and blue circles mark known catalytic residues. Orange circles denote residues that mediate interactions between dimers, yellow circles denote residues that map to the potential interface with CcmM (see Figure 3). A green box marks the C-terminal conserved sequence, whereas the magenta box marks the residues that, in Synechocystis CcaA, are ordered at the two-fold symmetry axis. Note that in strains with two CcaA sequences one is typically ∼270 amino acids, whereas the other is ∼210 amino acids and is missing the C-terminal extension normally characteristic of CcaA.

CcmM is the second potential β-carboxysomal CA and is universally present in β-carboxysomes as it functions as a central nexus for organizing the carboxysome's interior. CcmM is built as two distinct regions; the C-terminal region of CcmM consists of three to five repeats of a RubisCO small subunit-like domain, separated by flexible linker regions; this region (which can be translated independently from an internal ribosome-binding site) binds RubisCO [1114]. The N-terminal domain is clearly homologous to γ-CAs, but in at least some strains, including Synechocystis sp. PCC 6803 and Synechococcus sp. PCC 7942, this domain lacks measurable CA activity [14,15]. CA activity is, however, present in CcmM from other strains, such as Thermosynechococcus elongatus BP-1 and Nostoc sp. PCC 7120 [16,17]. The critical determinant for activity is a motif associated with the N- and C-termini of the domain that acts as a redox sensitive switch. In many (but not all) strains that include the β-CA, CcaA, this motif has devolved, resulting in a CcmM protein that cannot be activated [11,14,16]. CcmM's N-terminal domain also mediates interactions, with both CcmN [14] (whose role appears to be to link this complex with the carboxysome shell [18]) and, where present, CcaA, recruiting them to the carboxysome [11,14]. Given that an active CcmM was probably present in the last common ancestor of β-cyanobacteria, it is unclear what competitive advantage CcaA offers to those cyanobacterial lineages that have subsequently adopted it. One explanation lies in the order of magnitude advantage in catalytic rates β-CAs typically show relative to γ-CAs; however, no kinetic data for CcaA itself have been reported to support or refute this idea.

While the fact that CcmM and CcaA interact is experimentally well established, the organization of this complex remains unclear. In general, hetero-oligomeric complexes rarely form between oligomeric proteins with incompatible symmetries [19]. Symmetry mismatches preclude taking advantage of avidity effects (where binding energy across symmetrically arrayed subsites is additive), require a given surface to mediate multiple independent binding modes and can lead to runaway aggregation. CcmM's N-terminal domain is trimeric [16], while structurally characterized homologs of CcaA are all either dimeric [20], tetrameric [21] or octameric [10], leading to an obvious symmetry mismatch. We report here that the structure of CcaA from Synechocystis sp. PCC 6803 has a novel hexameric organization; this structure suggests that CcaA's recruitment to the carboxysome relies upon interacting with CcmM by aligning their mutual three-fold symmetry axes.

Experimental procedures

Molecular biology and protein expression, purification and quantification

CcaA220 was amplified using the primers 5′-GAGCTACATATGCAAAGACTCATCGAGGGAC-3′ and 5′-GATGCTGGATCCTATTAATGGAGAGCGTACTCATCCTCC-3′, purchased from Sigma Genosys, using Synechocystis sp. PCC 6803 genomic DNA as a template. CcmM206 was amplified using the primers 5′-GAACTATCATATGGGATCTCGCACCGCCTTGGC-3′ and 5′-GAAGCTTCTAGGGGAGGGACTCCGGCAACAC-3′. After digestion, amplified DNA was ligated into the BamHI and either NdeI (CcaA) or HindIII (CcmM206) sites of pET-28a, which was then transformed into DH5α competent cells. The nucleotide sequences encoding CcaA274 (822 bp) and CcmM206-4A (621 bp) from Synechocystis sp. PCC 6803 were synthesized by GenScript (Piscataway, NJ) and subsequently cloned into the NdeI or BamHI–HindIII restriction site of the pET-28b expression vector. Plasmids from selected colonies were purified using the QIAprep Spin Miniprep kit (QIAGEN) and sequenced. Authenticated plasmids were transformed into BL21(DE3) cells (EMD Biosciences) and grown in LB or 2× yeast tryptone (2YT) media with 34–50 µg ml−1 of kanamycin at 30 or 37°C. Expression was induced by adding 100 µg/l isopropyl-1-thio-β-d-galactopyranoside once A600 reached 0.8, and the cells were then grown overnight at 15 or 23°C. Cells were harvested by centrifugation (4000 g, 4°C, 20 min), resuspended in loading buffer [20 mM Tris (pH 7.9), 10% glycerol and 0.5 M NaCl] and were lysed by French Press. After centrifugation (15 000 g, 4°C, 40 min), the 0.45 μm filtered supernatant was loaded onto a His-Trap Nickel Nickel-Agarose column (Sigma) pre-equilibrated with loading buffer. After washing with loading buffer supplemented with 10 mM imidazole, the protein was eluted using loading buffer with 0.5 M imidazole added. Protein was concentrated using spin columns and the purity of the final prep was confirmed by visual inspection of an overloaded SDS–PAGE gel. Purified protein was stored in elution buffer at −80°C until needed.

Crystallization and structure determination

Crystals of CcaA220 were grown by sitting drop vapor diffusion with 1 µl of well solution and 1 µl of protein. CcaA crystals grew from 7.1 mg/ml protein, 2.0 M Na formate and 0.1 M Na acetate (pH 4.6) well solution. Surface liquid was removed by immersion in paratone N oil, and crystals were then frozen in liquid nitrogen. Data were collected at the Canadian Light Source (beam line 08ID) at 100K and processed in XDS [22]. Crystals proved to be of the tetragonal space group I4, with cell dimensions a = b = 202.95 Å, c = 70.51 Å and diffracted to 1.45 Å. The structure was determined by molecular replacement using Phaser [23] in Phenix, with a catalytic dimer (residues 121–318) from the Pisum sativum β-CA structure (1ekj) as a search model [10]. The structure was rebuilt in Coot [24] and refined in Phenix [25], using TLS parameters. Data collection and final refinement statistics are reported in Table 1. Individual chains have two to nine additional residues derived from the histidine tag ordered at the N-terminus, where they form a short extension to αA. At the C-terminus, chains A, B and E are ordered to residue 220, whereas, for chains C, D and F, all residues C-terminal to 204 or 205 (F) are disordered. Structured figures were prepared using PyMol.

Table 1
Data collection, model refinement and final structure statistics
Crystallographic data collection statistics 
Pdb ID 5SWC 
Space group I
Cell dimensions 
  a 203.720 
  c 70.880 
Wavelength (Å) 0.97888 
Resolution (Å) 1.45 
Redundancy 4.7 
Completeness (last shell)1 0.999 (1.00) 
(last shell)1 17.5 (2.55) 
 Rsym (last shell)1 0.054 (0.613) 
X-ray structure refinement statistics 
 Rcryst 0.1308 
 Rfree 0.1465 
Asymmetric unit contents 
Protein chains 
Water molecules 1038 
Other molecules 6 Zn2+ 
5 Cl 
22 formate 
Average ADPs (Å2
Protein 26.8 
Water 35.1 
r.m.s.d. bond lengths (Å) 0.013 
r.m.s.d. bond angles (°) 1.418 
Ramachandran favored (%) 99.4 
Ramachandran outliers (%) 
Crystallographic data collection statistics 
Pdb ID 5SWC 
Space group I
Cell dimensions 
  a 203.720 
  c 70.880 
Wavelength (Å) 0.97888 
Resolution (Å) 1.45 
Redundancy 4.7 
Completeness (last shell)1 0.999 (1.00) 
(last shell)1 17.5 (2.55) 
 Rsym (last shell)1 0.054 (0.613) 
X-ray structure refinement statistics 
 Rcryst 0.1308 
 Rfree 0.1465 
Asymmetric unit contents 
Protein chains 
Water molecules 1038 
Other molecules 6 Zn2+ 
5 Cl 
22 formate 
Average ADPs (Å2
Protein 26.8 
Water 35.1 
r.m.s.d. bond lengths (Å) 0.013 
r.m.s.d. bond angles (°) 1.418 
Ramachandran favored (%) 99.4 
Ramachandran outliers (%) 
1

The last shell includes all reflections between 1.50 and 1.45 Å.

Docking

All water and His-tag residues were removed from the CcaA structure. A model of the Synechocystis PCC 6803 CcmM N-terminal domain was built using SwissModel, with 3KWD (a T. elongatus CcmM construct lacking the C-terminal disulfide bond-forming residues) serving as a template. After both structures were oriented with their three-fold symmetry axes aligned along the Z axis, and subjected to predock relaxation, candidate poses were generated by rotating CcmM around (1° increments) and translating along (0.3 Å increments) the Z-axis. Poses were then, minimized, repacked and evaluated in RosettaDock 3.5 [26] using high-resolution docking. A second search using the opposite end of CcmM was also attempted, and a similar procedure was followed to dock T. elongatus CcmM (3KWD) to the same Synechocystis CcaA model.

Surface plasmon resonance

Protein/protein-binding analysis was performed using an OpenSPR localized surface plasmon resonance (LSPR) biosensor (Nicoya Life Science, Inc., Kitchener, Canada). CcaA274 or CcaA220 served as the ligand and was immobilized on gold nanoparticle, COOH sensor chips using standard 1-ethyl-3-(3-dimethylpropyl)-carbodiimide plus N-hydroxysuccinimide coupling chemistry. Typically, 100 µl of CcaA (15–50 µg ml−1) in activation buffer was introduced into the sensor chip at a flow rate of 20 µl min−1 with 20 mM Tris–HCl (pH 7.4) as the running buffer. Any remaining activated carboxyl groups were deactivated by treatment with OpenSPR deactivation buffer. The analyte running buffer consisted of 20 mM Tris–HCl (pH 7.4), 210 mM NaCl, 0.05% w/v Tween 20 and 0.1% w/v BSA. During optimization experiments, analyte (100 µl, CcmM206 or CcmM206-4A; 0.12–29.9 µg ml−1) was introduced into a ligand-bound sensor chip at flow rates between 20 and 150 µl min−1. Sensorgram traces of CcmM/CcaA interaction were recorded and analyzed using the built-in TraceDrawer software package (Ridgeview Instruments, Uppsala, Sweden). Under optimal conditions, the association phase was 4.2 min, whereas the disassociation phase was up to 15 min. Ligand/sensor chips were regenerated between each injection of analyte using 10 mM HCl at a flow rate of 100 µl min−1.

Enzyme kinetics

Initial rates of CO2 hydration were determined at 25°C with an Applied Photophysics SX20 stopped-flow spectrometer using the changing pH-indicator method [2628]. Saturated CO2 solutions were prepared by bubbling CO2 gas (100% v/v) into distilled water in a temperature-controlled (25°C) vessel. To achieve lower CO2 concentrations, the saturated stock was diluted with degassed water by coupling two gas-tight syringes, containing the appropriate volumes, via a luer-lock connector. The CO2 concentrations were calculated based on a 33.8 mM saturated solution at 25°C. CcaA was assayed at a final concentration of 0.25–1 μM. The following buffer/indicator combinations and absorbance wavelengths were used: at pH 6.0–6.5, 2-(N-morpholino)ethanesulfonic acid (MES; pKa = 6.1)/chlorophenol red (574 nm); pH 7.0–8.1, 3-morpholinopropane-1-sulfonic acid (MOPS, pKa = 7.2)/p-nitrophenol (400 nm); pH 8.0–9.5 Bicine (pKa = 8.3)/m-creosol purple (578 nm), and pH 9.0–9.5 3-[(1,1-dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulfonic acid (AMPSO; pKa = 9.0)/thymol blue (596 nm). All measurements were made at a final buffer concentration of 40 mM and 10 μM EDTA. The steady-state kinetic constants kcat and kcat/Km were determined by fitting the initial rates to the Michaelis–Menten equation using the built-in algorithm in SigmaPlot 12.5 (Systat Software).

Results

CcaA structure

The crystal structure of Synechocystis sp. PCC 6803 CcaA220 (i.e. a construct with the C-terminal residues 221–274 deleted) was determined at 1.45 Å (Figure 2A and Table 1). Individual chains within the structure are largely identical, with pairwise r.m.s.d. values of 0.2 Å or lower. Searching the protein databank with pdbEfold shows that the P. sativum chloroplast β-CA 1ekj is most closely related with a Z-score of 13.8 and an r.m.s.d. of 1.24 Å [10], followed by the fungal enzymes from Saccharomyces cerevisiae (3eyx; Z = 10.4, r.m.s.d = 2.15 Å) [27] and Cryptococcus neoformans (2w3n; Z = 11.9, r.m.s.d. = 1.68 Å) [20]; bacterial β-CA structures are more distantly related, the most closely superimposing of which is that of Haemophilus influenzae (2a8d; Z = 10.3, r.m.s.d. = 1.8 Å) (Figure 2B; the corresponding structure-based sequence alignment is shown in Supplementary Figure S1) [28]. While CcaA has an overall 32% sequence identity with plant chloroplast β-CA, differences include the presence of an additional short 310 helix in CcaA between αE and αG (Figure 2A; the equivalent residues in other β-CA structures form a loop) and an additional β-strand, β6 antiparallel to β5 (the equivalent residues in P. sativum β-CA form a long extension to β5 that stabilizes the octamer by hydrogen bonding with equivalent residues from an adjacent protomer [10]). In CcaA, three of the six chains also have residues 205–220 ordered as an extended, irregular loop along the protein's surface (see below); this motif is also unique to CcaA.

The structure of CcaA.

Figure 2.
The structure of CcaA.

(A) Organization of the protomer, colored blue to red in a gradient from N- to C-terminus; the zinc ion is shown as a gray sphere. Secondary structure elements are labeled. Residues C-terminal to β6 are ordered in only three of the six protomers. (B) Superposition of CcaA protomer on representative β-CA structures. Shown are CcaA (5swc, red), P. sativum (1ekj, yellow), S. cerevisiae (3eyx, green), C. neoformans β-CA (2w3n; cyan) and H. influenzae β-CA (2a8d; blue); the corresponding sequence alignment is shown as Supplementary Figure S1. Note that the structures generally superimpose closely, with the main variations being in the organization of the C-terminus, which in hexameric and octameric variants plays an important role in organizing the oligomer. (C) The organization of the CcaA hexamer. One protomer is shown as in A, with its dimeric partner in beige. Note the trimer-of-dimers organization, where each protomer only mediates dimeric interactions on one face, and trimeric interactions on the other. The lower subpanel is the orthogonal view, with a dimeric pair shown as a cartoon and the other four protomers shown as a white surface. (D) Organization of the CcaA catalytic site. One protomer is shown in white and the second protomer of the catalytic dimer is in beige. The formate ion is shown in green. The catalytic water molecule is bound to the zinc ion and makes hydrogen bonds with Asp41, Gln30 and Gly102. The fourth hydrogen-bonding group points into the catalytic pocket, where it forms an additional hydrogen bond with a well-ordered formate molecule bound in the catalytic site. (E) Details on the proximal C-terminal extension at the oligomeric two-fold symmetry axis. Note that this sequence passes through the molecular two-fold symmetry axis, allowing only one chain to be ordered at a time.

Figure 2.
The structure of CcaA.

(A) Organization of the protomer, colored blue to red in a gradient from N- to C-terminus; the zinc ion is shown as a gray sphere. Secondary structure elements are labeled. Residues C-terminal to β6 are ordered in only three of the six protomers. (B) Superposition of CcaA protomer on representative β-CA structures. Shown are CcaA (5swc, red), P. sativum (1ekj, yellow), S. cerevisiae (3eyx, green), C. neoformans β-CA (2w3n; cyan) and H. influenzae β-CA (2a8d; blue); the corresponding sequence alignment is shown as Supplementary Figure S1. Note that the structures generally superimpose closely, with the main variations being in the organization of the C-terminus, which in hexameric and octameric variants plays an important role in organizing the oligomer. (C) The organization of the CcaA hexamer. One protomer is shown as in A, with its dimeric partner in beige. Note the trimer-of-dimers organization, where each protomer only mediates dimeric interactions on one face, and trimeric interactions on the other. The lower subpanel is the orthogonal view, with a dimeric pair shown as a cartoon and the other four protomers shown as a white surface. (D) Organization of the CcaA catalytic site. One protomer is shown in white and the second protomer of the catalytic dimer is in beige. The formate ion is shown in green. The catalytic water molecule is bound to the zinc ion and makes hydrogen bonds with Asp41, Gln30 and Gly102. The fourth hydrogen-bonding group points into the catalytic pocket, where it forms an additional hydrogen bond with a well-ordered formate molecule bound in the catalytic site. (E) Details on the proximal C-terminal extension at the oligomeric two-fold symmetry axis. Note that this sequence passes through the molecular two-fold symmetry axis, allowing only one chain to be ordered at a time.

CcaA oligomeric organization

β-CA structures are organized around a tight, catalytic dimer (or, in several examples, a pseudo-dimer); CcaA is no exception with a dimeric interface that is predicted to be very stable, burying 3643 Å2 per protomer. CcaA presents a novel oligomeric variation for the β-CA quaternary structure, where a trimer of dimers forms a hexamer (Figure 2C). This hexamer can be thought of as a more tightly packed variant on the rather open P. sativum octamer (Supplementary Figure S2D), with dimers twisted ∼20° to achieve optimal packing. In this oligomeric state, the dimers are arranged with their longest axis orthogonal to the molecular three-fold axis, so interactions between adjacent dimers utilize contributions only from a single protomer on either side. Much of the interaction surface between dimers is afforded by helices αH, αI and αJ, with a total of 1490 Å2 buried per protomer. Interestingly, this general region of the protein surface also mediates interactions in the tetrameric and octameric variants of β-CAs.

The oligomerization interface in CcaA is supplemented by additional interactions provided by the proximal portion of the C-terminal extension (Figure 2E). Residues 205–220 are ordered in only three of the chains (A, B and E), adopting an extended conformation that passes through the two-fold molecular symmetry axis where two dimers meet. Since there is only a single binding site formed by the dimer interface, with Asp215 sitting at the two-fold symmetry axis, residues from only one of the two adjacent protomers can bind, leaving the equivalent residues in the other protomer disordered. Despite the intrinsic symmetry of this site, the loop does not take on an especially symmetric conformation, and hence even where equivalent residues are contacted in the two half sites, the interactions are quite different. For example, Leu211 interacts with Leu159 on the same chain, whereas Tyr217 interacts with the same residue on the second protomer, but interacts in a nonequivalent manner. Residues from the C-terminal extension make a series of hydrogen bonding and van der Waals interactions, and contribute almost 400 Å2 of the 1490 Å2 interface between adjacent dimers; using PISA [29], this region is estimated to contribute 5 kcals/mol to the stability of the interface.

Catalytic site

The catalytic site of CcaA is located at the interface between two protomers, with the zinc ion bound to Cys39, His98 and Cys101 (Figure 2D). A water molecule (at pH 4.5, the water should retain its proton) forms the fourth zinc ligand, with a 2.05 ± 0.02 Å bond length and tetrahedral geometry. The water donates a hydrogen bond to Asp41 (2.63 ± 0.01 Å; positioned by Arg43) and accepts one from the amide nitrogen of Gly102 (2.89 ± 0.02 Å). A fourth hydrogen bond (2.54 ± 0.07 Å) is made to a well-ordered formate molecule which sits at the back of the catalytic pocket, making van der Waals contacts with Phe58′ and Tyr83′ from the second protomer, as well as hydrogen bonds to Ala103 and Gln30′. Kinetic analysis of mutated variants of the plant enzymes confirm that these residues together constitute the critical catalytic machinery [3032]. The formate molecule binds near identically with CO2 binding in the PsCA3 structure, with the oxygen atoms superimposing to within 0.1 Å [33]. The structure of CcaA in complex with formate is, therefore, an excellent mimic of the β-CA–OH–CO2 catalytic complex. Tyr83′ is known from work in the plant enzymes to play the role of a proton shuttle [31]; the histidine residue that assists in this role is missing in CcaA, but possibly His100, which is also immediately adjacent to the catalytic site, may substitute.

In most bacterial β-CA structures determined to date, the catalytic Asp (here Asp41) is found to bind the zinc ion, blocking substrate binding and hence catalysis; these are known as ‘type II’ enzymes [34]. This type II structure is stabilized by low pH and bicarbonate binding in a regulatory site. CcaA was crystallized at a pH of 4.5 but remains in the active, type I conformation, which is typified by a salt bridge between Asp41 and Arg43 that allows the zinc ion to remain free for substrate binding. In addition, CcaA does not show the co-operative inactivation at low pH characteristic of the type II bacterial enzymes (see below), and the residues that form the non-catalytic (regulatory) bicarbonate-binding site (Arg–Trp–Tyr) are absent. The main difference here may lie in the substitution of Phe36 for the key tryptophan residue that mediates a hydrogen bond in the H. influenzae structure. CcaA is, therefore, one of only two structurally characterized bacterial β-CAs that display type I active sites (the other being Rv3588) [35].

Docking the CcaA and CcmM structures

CcaA is recruited to the β-carboxysome by binding the N-terminal domain of CcmM, as previously demonstrated by yeast two-hybrid studies, and IMAC based pull-down studies using both recombinant proteins and tagged complexes isolated from carboxysomes [11,14]. The details of how exactly these two proteins interact are currently unknown; however, the symmetry match strongly suggests that this interaction occurs along their mutual three-fold axes. Inspection of the CcmM structure shows that the C-terminal end of the β-helix is dominated by the protruding αA helix and the αA–αB loop (Figure 3A). The corresponding surface on CcaA has a prominent depression around the three-fold axis, with a pocket of conserved residues that seems well positioned to bind the exposed helix (Figure 3B). To investigate the interaction geometry, we attempted to generate a complex by docking the two proteins. We generated a homology model of Synechocystis CcmM, and then systematically generated poses that conform to the overall three-fold symmetry of the complex, sampling at 1° rotations and 0.3 Å translations along the Z-axis. These 2500 candidate poses were then energy-minimized (including shifts of the molecules and side chain repacking) and evaluated using RosettaDock [26]. While the highest observed RosettaDock integrated score (I_sc) achieved was −7.33, closer inspection showed that the generated solutions were unlikely to reflect the biological complex: several quite different poses lead to similar scores, and the surfaces do not interact strongly, with only a relatively small area buried at the interface [729 Å2 per CcaA(dimer)–CcmM pair]. It is unlikely that we failed to find an accessible solution due to undersampling, as we finely sampled all possible starting poses that conform to the symmetry constraints. Docking attempts where CcmM interacts with the N-terminal end of the β-helix, or with TeCcmM (which will bind CcaA at least in vitro, [16]), also produced no convincing solutions.

The interaction determinants of the CcaA–CcmM complex.

Figure 3.
The interaction determinants of the CcaA–CcmM complex.

(A) Surface representation of Synechocystis PCC 6803 CcmM (a homology model based on 3KWD), looking at the C-terminal end of the β-helix. αA and the αA–αB loop, shown in yellow, form a prominent ridge of residues that protrudes from the surface. The four residues mutagenized to test the role of this surface in CcaA binding are indicated in brighter yellow and labeled. (B) Surface of CcaA looking down the three-fold interface. Coloring is by sequence conservation as calculated in Consurf, with magenta being absolutely conserved and cyan being unusually variable. The scale is identical with A. (C) CcaA colored by ADPs, with blue denoting the most ordered parts of the structure and orange the least ordered. Note that both the αG–αH hairpin and 310F motifs show elevated ADPs.

Figure 3.
The interaction determinants of the CcaA–CcmM complex.

(A) Surface representation of Synechocystis PCC 6803 CcmM (a homology model based on 3KWD), looking at the C-terminal end of the β-helix. αA and the αA–αB loop, shown in yellow, form a prominent ridge of residues that protrudes from the surface. The four residues mutagenized to test the role of this surface in CcaA binding are indicated in brighter yellow and labeled. (B) Surface of CcaA looking down the three-fold interface. Coloring is by sequence conservation as calculated in Consurf, with magenta being absolutely conserved and cyan being unusually variable. The scale is identical with A. (C) CcaA colored by ADPs, with blue denoting the most ordered parts of the structure and orange the least ordered. Note that both the αG–αH hairpin and 310F motifs show elevated ADPs.

Inspecting candidate solutions shows that the topographies of the surfaces do not match ideally. Part of the challenge is that in order to be computationally tractable, docking typically assumes that the backbone is relatively rigid. However, inspection of atomic displacement parameters (ADPs) in the two structures shows that several key motifs in this region are quite mobile — in particular, CcaA's 310F and the αG–αH loop have backbone ADPs in the 30–40 range (Figure 3C). The interaction region, therefore, has considerable capacity to reorganize co-operatively during binding. The αG–αH helix pair in particular is surface-exposed and seems poised to shift so as to maximize packing interactions against the outside edge of bound CcmM.

CcaA–CcmM interactions by LSPR

To quantitatively analyze the CcmM–CcaA interaction, we investigated binding using LSPR. To highlight contributions from the C-terminal tail of CcaA, we determined the binding kinetics for both CcaA274 and CcaA220 binding to CcmM206. We also investigated the binding of these CcaA constructs to CcmM206-4A — a variant that has four amino acid substitutions: Q159A/H161A/D168A/Q170A. These residues are located in exposed positions in αA and the αA–αB loop at the C-terminal end of the CcmM β-helix; this construct, therefore, is designed to test the hypothesis that this is the interaction face of CcmM. In the LSPR sensorgram traces, the wavelength shift upon the introduction of CcmM206 into the biosensor containing tethered CcaA confirms binding (Figure 4A,B). No significant wavelength shift occurred if commercial bovine CA was used as the analyte (negative control) or the CcaA ligand was absent from the biosensor, demonstrating minimal nonspecific binding. Global analysis of the sensorgram traces using a 1:1 binding model yielded an association rate constant, kon, of 1.54 × 105 M−1 s−1 ± 23 for the interaction of CcmM206 with the tethered CcaA274 ligand (Table 2). Once formed, the Cca274–CcmM206 complex was quite stable with a dissociation rate constant, koff, of 5.81 × 10−5 s−1 ± 3.25 × 10−5 and KD of 4.82 × 10−10 M ± 2.7 × 10−10 (0.48 pM). Estimates of koff were prone to more error than kon due to the slow dissociation and the extended time period required to complete the measurements. CcaA274 binds the CcmM206-4A construct (Figure 4A and Table 2) with a nine-fold reduction in affinity (KD of 4.3 pM). This suggests that the side chains of these residues in and around CcmM αA play a role in the mechanism that mediates binding of CcaA by CcmM, although these individual side chains are not indispensable. Characterization of the interaction between CcaA220 and CcmM206 (Figure 4B and Table 2) yielded a KD of 4.65 pM, indicating that the C-terminal tail of CcaA also contributes to the complex formation. The interaction of CcaA220 was again weaker with the CcmM206-4A construct (KD of 33.3 pM), indicating that these two interactions are probably independent.

Representative LSPR sensorgram traces illustrating CcaA–CcmM association/dissociation kinetics.

Figure 4.
Representative LSPR sensorgram traces illustrating CcaA–CcmM association/dissociation kinetics.

Traces have been overlaid for comparison. CcaA274 (A) or CcaA220 (B) served as the ligand with either CcmM206 (black lines) or CcmM206-4A (red lines) as the analyte. The nominal ligand concentration was 15 µg ml−1. The analyte (1–30 µg ml−1 as indicated) was introduced into the biosensor at a flow rate of 20 µl min−1 for 252 s followed by buffer alone. The green vertical line (252 s) delineates the boundary between the association and the dissociation phases. Commercial bovine CA (30 µg ml−1, blue trace) served as a negative control.

Figure 4.
Representative LSPR sensorgram traces illustrating CcaA–CcmM association/dissociation kinetics.

Traces have been overlaid for comparison. CcaA274 (A) or CcaA220 (B) served as the ligand with either CcmM206 (black lines) or CcmM206-4A (red lines) as the analyte. The nominal ligand concentration was 15 µg ml−1. The analyte (1–30 µg ml−1 as indicated) was introduced into the biosensor at a flow rate of 20 µl min−1 for 252 s followed by buffer alone. The green vertical line (252 s) delineates the boundary between the association and the dissociation phases. Commercial bovine CA (30 µg ml−1, blue trace) served as a negative control.

Table 2
Kinetic constants for CcaA–CcmM interactions derived from the global analysis of LSPR sensorgram traces
Ligand Analyte Kon, M−1 s−1 (±SD)1 Koff, s−1 (±SD) KD, M (±SD) 
CcaA274 CcmM206 1.54 × 105 (±23) 5.81 × 10−5 (±3.25 × 10−54.82 × 10−10 (±2.7 × 10−10
CcmM206-4A 6.87 × 104 (±2.7) 2.97 10–4(±1.07 × 10−74.33 × 10–9 (±1.73 × 10−12
CcaA220 CcmM206 2.97 × 104 (±8.6) 1.38 × 10–4 (±3.18 × 10−74.65 × 10–9 (±1.2 × 10−11
CcmM206-4A 7.40 × 104 (±65.0) 2.46 × 10−3 (±3.98 × 10−53.33 × 10−8 (±8.3 × 10−10
Ligand Analyte Kon, M−1 s−1 (±SD)1 Koff, s−1 (±SD) KD, M (±SD) 
CcaA274 CcmM206 1.54 × 105 (±23) 5.81 × 10−5 (±3.25 × 10−54.82 × 10−10 (±2.7 × 10−10
CcmM206-4A 6.87 × 104 (±2.7) 2.97 10–4(±1.07 × 10−74.33 × 10–9 (±1.73 × 10−12
CcaA220 CcmM206 2.97 × 104 (±8.6) 1.38 × 10–4 (±3.18 × 10−74.65 × 10–9 (±1.2 × 10−11
CcmM206-4A 7.40 × 104 (±65.0) 2.46 × 10−3 (±3.98 × 10−53.33 × 10−8 (±8.3 × 10−10
1

n = 3–5 samples.

Enzyme kinetics

We characterized the CO2 hydration reaction of Synechocystis CcaA constructs CcaA220 and CcaA274 over a pH range of 6.0–9.5 via stopped-flow spectrophotometry using standard buffer–indicator pairs (Figure 5). Catalytic characterization of the full-length CcaA274 enzyme in a low ionic strength buffer gave maximum turnover, with a kcat of 1.37 × 104 s−1 and a kcat/Km of 1.73 × 107 M−1 s−1, at the pH of 9.5, the highest pH measured. At a pH of 7.5, close to the cytosolic pH of a photosynthetically active cyanobacterial cell, kcat is an even more modest 3.34 × 103 s−1. Interestingly, the CcaA220 construct is an order of magnitude more active at this pH, with a kcat of 3.1 × 104 s−1. However, in all stopped-flow reaction buffers, CcaA220 catalytic activity was observed to decrease to near zero over a time span of 20 min. This phenomenon was not observed for CcaA274, suggesting that the C-terminal tail plays a role in the long-term maintenance and modulation of catalytic activity. The Km for both constructs is in the low millimolar range, consistent with what has been observed for other β-CAs.

Enzyme kinetics analysis of CcaA.

Figure 5.
Enzyme kinetics analysis of CcaA.

pH dependence of the steady-state kinetic parameters kcat (A), kcat/Km (B) and Km (C) for the CO2 hydration reaction at 25°C catalyzed by CcaA274 (full length), black circles, CcaA220, red squares. Note that CcaA220 shows an order of magnitude bump in kcat and kcat/Km near neutral pH.

Figure 5.
Enzyme kinetics analysis of CcaA.

pH dependence of the steady-state kinetic parameters kcat (A), kcat/Km (B) and Km (C) for the CO2 hydration reaction at 25°C catalyzed by CcaA274 (full length), black circles, CcaA220, red squares. Note that CcaA220 shows an order of magnitude bump in kcat and kcat/Km near neutral pH.

Discussion

Kinetic characterization of Synechocystis sp. PCC 6803 CcaA reveals that this enzyme has surprisingly modest catalytic activity. Even at the optimal pH of 9.5 in a minimally inhibitory low ionic strength buffer, CcaA274 has a kcat of 1.37 × 104 s−1 and a kcat/Km = 1.73 × 107 M−1 s−1. Plant CAs are typically an order of magnitude more active. More physiologically relevant is the activity at pH 7.5, where kcat is an even more modest, 3.34 × 103 s−1, and kcat/Km is 1.19 × 106 M−1 s−1. However, it is possible that, in the assembled carboxysome, the inhibitory effect of the C-terminal tail is somehow relieved, allowing an order of magnitude improvement in the pH 7.5 catalytic constants. It is interesting to compare these results with CcmM from Nostoc PCC 7120, which has a kcat of 1.9 × 104 s−1 and a kcat/Km of 1.9 × 106 M−1 s−1 at physiological pH [17]. Surprisingly, given the importance of CA activity to models of carboxysome functioning, this is the first time that catalytic characterizations have been available for representative examples of both enzymes. On the surface, these data would indicate that CcmM is somewhat more active than CcaA. It is unclear whether this difference in activity has physiological consequences for carboxysomes that use CcmM, CcaA or both as their CAs; possibly, the available CA activity is in such large excess relative to RuBisCO's activity that there is little selective pressure to maximize CA activity. Furthermore, it may be advantageous to express the minimal required levels of CA activity, particularly in the cytosol during carboxysome biogenesis, as this would diminish the deleterious effects that CA has on the concentrated pool.

While the fundamental catalytic unit of β-CAs is a dimer, considerable variation is observed in their large-scale oligomeric organization (Supplementary Figure S2). The fungal and the ‘cab’-type β-CAs [20,27,36] are dimeric, most bacterial β-CAs are dimer-of-dimer tetramers [21,28], the plant chloroplast β-CA forms an irregular dimer-of-dimers-of-dimers octamer [10] and the closely related enzyme carbon disulfide hydrolase interlocks two such rings to form a hexadecameric catenane [37]. There are extensive data — from yeast two-hybrid experiments, IMAC-based interactions of recombinant proteins [14] and tagged pull-downs from carboxysomal preparations [11] — showing that CcaA interacts with the N-terminal domain of CcmM. The difficulty in rationalizing this interaction is that CcmM is trimeric, while no known β-CA structure has a matching three-fold symmetry axis. Understanding that CcaA forms a novel trimer-of-dimers hexamer, therefore, provides an immediate resolution to the CcaA–CcmM symmetry mismatch dilemma and strongly suggests that CcmM and CcaA interact along their respective three-fold symmetry axis. Each CcaA hexamer would then have two CcmM-binding sites located on opposite faces; by default, one would anticipate a symmetric CcmM3CcaA6CcmM3 complex, though symmetry breaking by the C-terminal tail could potentially stabilize an asymmetric CcmM3CcaA6 complex.

LSPR indicates that this interaction is very strong, with KD of 0.48 pM. The CcmM-4A construct has a significant (if somewhat modest) impact on the binding affinity, confirming that the C-terminal end of CcmM's β-helix is probably the main interaction determinant. This orientation is perhaps to be expected, as the N-terminal end of CcmM's β-helix is required to undergo a major reorganization upon encapsulation and oxidation [16], a reorganization that might be arrested if CcaA were to be tightly bound to the inactive conformation. Despite the strong constraints afforded by the geometry of the proteins, RosettaDock fails to find a convincing CcaA–CcmM docking solution. Instead, interactions occur only between side chains along exposed ridges of residues. Inspection of the relevant regions of the structure strongly suggests that at minimum, the exposed ends of αG–αH in CcaA need to shift outward slightly to allow CcmM to approach CcaA more closely. Elevated backbone ADPs for this motif (which are triple those of residues in the well-packed hydrophobic core) indicate that this region has elevated flexibility that should readily allow repacking. Our failure to find rigid-backbone docking solutions would, therefore, seem to indicate that CcaA binding to CcmM probably involves local reorganization of the backbone at the interface. However, the fast binding we observe in the LSPR experiments probably precludes large-scale domain reorganization, which typically occurs on a millisecond time scale.

CcaA differs from other β-CAs by the presence of a long C-terminal extension (Figure 1). The first 20 amino acids of this motif (195–214 in Synechocystis sp. PCC 6803) show some conservation, and deleting these residues compromises the protein's activity and ability to form oligomers [38]. Our structure shows that the proximal part of this extension (residues 205–220) binds a site along the two-fold axis, with two copies of this motif in competition for a single binding site. The reason why these residues are essential for stabilizing the active conformation of CcaA is not immediately obvious, though interactions mediated by this region may stabilize the β4–β5 hairpin. This essential region is followed by up to 80 residues that are poorly conserved, hydrophilic and predicted to be unstructured; the variable length of this linker, as well as the fact that it is wholly absent from some strains (e.g. Microcoleus sp. PCC 7113), suggests that its role is nonessential. This linker region is followed by a second conserved motif of ∼15 amino acids that has been suggested to possibly mediate interactions with one of the shell proteins [18], though direct evidence for this idea is presently lacking. Deleting this region results in a construct that cannot rescue a ccaA deletion strain [39]. This region is, therefore, functionally indispensable, a finding that is still difficult to fully rationalize from our biochemical data. Our LSPR results indicate that this region contributes to CcmM binding, increasing the association rate, but is not essential. More intriguing is the suggestion from kinetics that this region is auto-inhibitory at physiological pH. CA activity in the cytosol of cyanobacteria causes leakage of concentrated by allowing it to convert back into membrane-permeable CO2 (as demonstrated by the high CO2 dependence of Synechococcus sp. PCC 7942 expressing human α-CAII [4]). As full-length CcaA274 constructs are an order of magnitude less active than CcaA220 constructs lacking the C-terminal tail, the tail may act to attenuate activity prior to encapsulation. In preliminary experiments, adding CcmM to CcaA does not relieve this inhibition, but interactions with some other protein in the mature carboxysome may possibly be required to fully activate the enzyme. Other mechanisms seem to also contribute to CcaA's regulation: in vivo CcaA levels seem to closely track CcmM levels as they vary in recombinant Synechococcus sp. PCC 7942 cells [13]. This implies that CcaA is not stable in vivo unless it forms a complex with CcmM; CcmM may therefore protect CcaA from rapid turnover, but also commits it to encapsulation.

It is interesting to note that in those cyanobacterial strains that have two CcaA paralogs, one CcaA copy possesses a C-terminal extension, whereas the second CcaA homolog consistently lacks both the linker and conserved C-terminal region (e.g. Leptolyngbya and Microcystis in Figure 1). Of note is the observation that the shorter CcaA variant never occurs alone, even in strains that contain a catalytically competent CcmM variant. This implies that only a subset of CcaA copies require the C-terminal tail (and there may be an advantage to only a subset having it). One possibility is that the two CcaA proteins form an obligatory tight heterodimer, reinforcing the inherent asymmetry we observe even in the single protein hexamer.

In summary, CcaA introduces several unique twists to the well-studied β-CA family. It is weakly active, hexameric, lacks the typical bacterial pH/bicarbonate regulation motifs and, instead, has a unique C-terminal tail that breaks local symmetry, and appears to play a role in both auto-regulation and protein–protein interactions. What remains unclear, especially given its modest CA activity, is what selective advantage CcaA offers cyanobacteria over the ancestral CcmM as a carboxysomal CA. These findings will help inform efforts to engineer cyanobacterial carbon-concentrating mechanisms into higher plants to increase crop photosynthetic efficiency.

Abbreviations

     
  • ADPs

    atomic displacement parameters

  •  
  • AMPSO

    3-[(1,1-dimethyl-2-hydroxyethyl)amino]-2-hydroxypropanesulfonic acid

  •  
  • CA

    carbonic anhydrase

  •  
  • RubisCO

    ribulose-1,5-bisphophate carboxylase/oxygenase

  •  
  • LSPR

    localized surface plasmon resonance

  •  
  • MES

    2-(N-morpholino)ethanesulfonic acid

  •  
  • MOPS

    3-morpholinopropane-1-sulfonic acid.

Author Contribution

G.S.E. and M.S.K. designed the study; M.M.-G. and M.S.K. designed and coordinated experiments; L.D.M., M.M.-G., S.A.W., M.S.K., T.S., B.B. and J.Q.T. performed experiments; L.M.D., M.M.-G., G.S.E. and M.S.K. analyzed data and wrote the manuscript.

Funding

This work was funded by Discovery Grants from the National Science and Engineering Research Council of Canada to G.S.E. and to M.S.K. [#04045-2015].

Acknowledgments

The authors thank Scott Mazurkewich and Jason Carere for synchrotron data collection, and Charlotte de Araujo and Alex Paquette for completing preliminary studies. The atomic co-ordinates and structure factors (code 5SWC) have been deposited in the Protein Data Bank (http://www.rcsb.org).

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Espie
,
G.S.
and
Kimber
,
M.S.
(
2011
)
Carboxysomes: cyanobacterial RubisCO comes in small packages
.
Photosyn. Res.
109
,
7
20
doi:
2
Rae
,
B.D.
,
Long
,
B.M.
,
Whitehead
,
L.F.
,
Förster
,
B.
,
Badger
,
M.R.
and
Price
,
G.D.
(
2013
)
Cyanobacterial carboxysomes: microcompartments that facilitate CO2 fixation
.
J. Mol. Microbiol. Biotechnol.
23
,
300
307
doi:
3
Kupriyanova
,
E.V.
,
Sinetova
,
M.A.
,
Cho
,
S.M.
,
Park
,
Y.-I.
,
Los
,
D.A.
and
Pronina
,
N.A.
(
2013
)
CO2-concentrating mechanism in cyanobacterial photosynthesis: organization, physiological role, and evolutionary origin
.
Photosyn. Res.
117
,
133
146
doi:
4
Price
,
G.D.
and
Badger
,
M.R.
(
1989
) Expression of human carbonic anhydrase in the cyanobacterium Synechococcus PCC7942 creates a high CO2-requiring phenotype: evidence for a central role for carboxysomes in the CO2 concentrating mechanism.
Plant Physiol.
91
,
505
513
doi:
5
So
,
A.K.-C.
,
Espie
,
G.S.
,
Williams
,
E.B.
,
Shively
,
J.M.
,
Heinhorst
,
S.
and
Cannon
,
G.C.
(
2004
)
A novel evolutionary lineage of carbonic anhydrase (epsilon class) is a component of the carboxysome shell
.
J. Bacteriol.
186
,
623
630
doi:
6
Sawaya
,
M.R.
,
Cannon
,
G.C.
,
Heinhorst
,
S.
,
Tanaka
,
S.
,
Williams
,
E.B.
,
Yeates
,
T.O.
et al. 
(
2006
)
The structure of β-carbonic anhydrase from the carboxysomal shell reveals a distinct subclass with one active site for the price of two
.
J. Biol. Chem.
281
,
7546
7555
doi:
7
Heinhorst
,
S.
,
Williams
,
E.B.
,
Cai
,
F.
,
Murin
,
C.D.
,
Shively
,
J.M.
and
Cannon
,
G.C.
(
2006
)
Characterization of the carboxysomal carbonic anhydrase CsoSCA from Halothiobacillus neapolitanus
.
J. Bacteriol.
188
,
8087
8094
doi:
8
Fukuzawa
,
H.
,
Suzuki
,
E.
,
Komukai
,
Y.
and
Miyachi
,
S
. (
1992
)
A gene homologous to chloroplast carbonic anhydrase (icfA) is essential to photosynthetic carbon dioxide fixation by Synechococcus PCC7942
.
Proc. Natl. Acad. Sci U.S.A.
89
,
4437
4441
doi:
9
Yu
,
J.-W.
,
Price
,
G.D.
,
Song
,
L.
and
Badger
,
M.R.
(
1992
)
Isolation of a putative carboxysomal carbonic anhydrase gene from the cyanobacterium Synechococcus PCC7942
.
Plant Physiol.
100
,
794
800
doi:
10
Kimber
,
M.S.
and
Pai
,
E.F.
(
2000
)
The active site architecture of Pisum sativum β-carbonic anhydrase is a mirror image of that of α-carbonic anhydrases
.
EMBO J.
19
,
1407
1418
doi:
11
Long
,
B.M.
,
Badger
,
M.R.
,
Whitney
,
S.M.
and
Price
,
G.D.
(
2007
)
Analysis of carboxysomes from Synechococcus PCC7942 reveals multiple Rubisco complexes with carboxysomal proteins CcmM and CcaA
.
J. Biol. Chem.
282
,
29323
29335
doi:
12
Long
,
B.M.
,
Tucker
,
L.
,
Badger
,
M.R.
and
Price
,
G.D.
(
2010
)
Functional cyanobacterial beta-carboxysomes have an absolute requirement for both long and short forms of the CcmM protein
.
Plant Physiol.
153
,
285
293
doi:
13
Long
,
B.M.
,
Rae
,
B.D.
,
Badger
,
M.R.
and
Price
,
G.D.
(
2011
)
Over-expression of the β-carboxysomal CcmM protein in Synechococcus PCC7942 reveals a tight co-regulation of carboxysomal carbonic anhydrase (CcaA) and M58 content
.
Photosyn. Res.
109
,
33
45
doi:
14
Cot
,
S.S.-W.
,
So
,
A.K.-C.
and
Espie
,
G.S.
(
2008
)
A multiprotein bicarbonate dehydration complex essential to carboxysome function in cyanobacteria
.
J. Bacteriol.
190
,
936
945
doi:
15
Espie
,
G.S.
and
So
,
A.K.-C.
(
2005
) Cyanobacterial carbonic anhydrases.
Can. J. Bot.
83
,
721
734
doi:
16
Peña
,
K.L.
,
Castel
,
S.E.
,
de Araujo
,
C.
,
Espie
,
G.S.
and
Kimber
,
M.S.
(
2010
)
Structural basis of the oxidative activation of the carboxysomal γ-carbonic anhydrase, CcmM
.
Proc. Natl Acad. Sci. U.S.A.
107
,
2455
2460
doi:
17
de Araujo
,
C.
,
Arefeen
,
D.
,
Tadesse
,
Y.
,
Long
,
B.M.
,
Price
,
G.D.
,
Rowlett
,
R.S.
et al. 
(
2014
)
Identification and characterization of a carboxysomal γ-carbonic anhydrase from the cyanobacterium Nostoc sp. PCC 7120
.
Photosyn. Res.
121
,
135
150
doi:
18
Kinney
,
J.N.
,
Salmeen
,
A.
,
Cai
,
F.
and
Kerfeld
,
C.A.
(
2012
)
Elucidating essential role of conserved carboxysomal protein CcmN reveals common feature of bacterial microcompartment assembly
.
J. Biol. Chem.
287
,
17729
17736
doi:
19
Ahnert
,
S.E.
,
Marsh
,
J.A.
,
Hernández
,
H.
,
Robinson
,
C.V.
and
Teichmann
,
S.A.
(
2015
)
Principles of assembly reveal a periodic table of protein complexes
.
Science
350
,
aaa2245
doi:
20
Schlicker
,
C.
,
Hall
,
R.A.
,
Vullo
,
D.
,
Middelhaufe
,
S.
,
Gertz
,
M.
,
Supuran
,
C.T.
et al. 
(
2009
)
Structure and inhibition of the CO2-sensing carbonic anhydrase Can2 from the pathogenic fungus Cryptococcus neoformans
.
Structure
385
,
1207
1220
doi:
21
Cronk
,
J.D.
,
Endrizzi
,
J.A.
,
Cronk
,
M.R.
,
O'neill
,
J.W.
and
Zhang
,
K.Y.J.
(
2001
)
Crystal structure of E. coli β-carbonic anhydrase, an enzyme with an unusual pH-dependent activity
.
Protein Sci.
10
,
911
922
doi:
22
Kabsch
,
W
. (
2010
)
XDS
.
Acta Crystallogr. Sect. D Biol. Crystallogr.
66
,
125
132
doi:
23
McCoy
,
A.J.
,
Grosse-Kunstleve
,
R.W.
,
Adams
,
P.D.
,
Winn
,
M.D.
,
Storoni
,
L.C.
and
Read
,
R.J.
(
2007
)
Phaser crystallographic software
.
J. Appl. Crystallogr.
40
,
658
674
doi:
24
Emsley
,
P.
and
Cowtan
,
K
. (
2004
)
Coot: model-building tools for molecular graphics
.
Acta Crystallogr. Sect. D Biol. Crystallogr.
60
,
2126
2132
doi:
25
Adams
,
P.D.
,
Grosse-Kunstleve
,
R.W.
,
Hung
,
L.-W.
,
Ioerger
,
T.R.
,
McCoy
,
A.J.
,
Moriarty
,
N.W.
et al. 
(
2002
)
PHENIX: building new software for automated crystallographic structure determination
.
Acta Crystallogr. Sect. D Biol. Crystallogr.
58
,
1948
1954
PMID:
[PubMed]
26
Gray
,
J.J.
,
Moughon
,
S.
,
Wang
,
C.
,
Schueler-Furman
,
O.
,
Kuhlman
,
B.
,
Rohl
,
C.A
et al.  et al.  (
2003
)
Protein–protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations
.
J. Mol. Biol.
331
,
281
299
doi:
27
Teng
,
Y.-B.
,
Jiang
,
Y.-L.
,
He
,
Y.-X.
,
He
,
W.-W.
,
Lian
,
F.-M.
,
Chen
,
Y
et al.  et al.  (
2009
)
Structural insights into the substrate tunnel of Saccharomyces cerevisiae carbonic anhydrase Nce103
.
BMC Struct. Biol.
9
,
67
doi:
28
Cronk
,
J.D.
,
Rowlett
,
R.S.
,
Zhang
,
K.Y.J.
,
Tu
,
C.
,
Endrizzi
,
J.A.
,
Lee
,
J.
et al. 
(
2006
)
Identification of a novel noncatalytic bicarbonate binding site in eubacterial β-carbonic anhydrase
.
Biochemistry
45
,
4351
4361
doi:
29
Krissinel
,
E.
and
Henrick
,
K
. (
2007
)
Inference of macromolecular assemblies from crystalline state
.
J. Mol. Biol.
372
,
774
797
doi:
30
Smith
,
K.S.
,
Ingram-Smith
,
C.
and
Ferry
,
J.G.
(
2002
)
Roles of the conserved aspartate and arginine in the catalytic mechanism of an archaeal β-class carbonic anhydrase
.
J. Bacteriol.
184
,
4240
4245
doi:
31
Rowlett
,
R.S.
,
Tu
,
C.
,
McKay
,
M.M.
,
Preiss
,
J.R.
,
Loomis
,
R.J.
,
Hicks
,
K.A.
et al. 
(
2002
)
Kinetic characterization of wild-type and proton transfer-impaired variants of β-carbonic anhydrase from Arabidopsis thaliana
.
Arch Biochem. Biophys.
404
,
197
209
doi:
32
Rowlett
,
R.S.
,
Tu
,
C.
,
Murray
,
P.S.
and
Chamberlin
,
J.E.
(
2004
)
Examination of the role of Gln-158 in the mechanism of CO2 hydration catalyzed by β-carbonic anhydrase from Arabidopsis thaliana
.
Arch Biochem. Biophys.
425
,
25
32
doi:
33
Aggarwal
,
M.
,
Chua
,
T.K.
,
Pinard
,
M.A.
,
Szebenyi
,
D.M.
and
McKenna
,
R
. (
2015
)
Carbon dioxide ‘Trapped’ in a β-carbonic anhydrase
.
Biochemistry
54
,
6631
6638
doi:
34
Rowlett
,
R.S.
(
2010
)
Structure and catalytic mechanism of the beta-carbonic anhydrases
.
Biochim. Biophys. Acta
1804
,
362
373
doi:
35
Covarrubias
,
A.S.
,
Bergfors
,
T.
,
Jones
,
T.A.
and
Högbom
,
M
. (
2006
)
Structural mechanics of the pH-dependent activity of β-carbonic anhydrase from Mycobacterium tuberculosis
.
J. Biol. Chem.
281
,
4993
4999
doi:
36
Strop
,
P.
,
Smith
,
K.S.
,
Iverson
,
T.M.
,
Ferry
,
J.G.
and
Rees
,
D.C.
(
2001
)
Crystal structure of the ‘cab-’ type β class carbonic anhydrase from the archaeon Methanobacterium thermoautotrophicum
.
J. Biol. Chem.
276
,
10299
10305
doi:
37
Smeulders
,
M.J.
,
Barensds
,
T.R.M.
,
Pol
,
A.
,
Scherer
,
A.
,
Zandvoort
,
M.H.
,
Udvarhelyi
,
A.
et al. 
(
2011
)
Evolution of a new enzyme for carbon disulphide conversion by an acidothermophilic archaeon
.
Nature
478
,
412
416
doi:
38
So
,
A.K.-C.
,
Cot
,
S.S.-W.
and
Espie
,
G.S.
(
2002
)
Characterization of the C-terminal extension of carboxysomal carbonic anhydrase from Synechocystis sp. PCC6803
.
Funct. Plant Biol.
29
,
183
194
doi:
39
So
,
A.K.-C.
(
2004
)
The role of carbonic anhydrase in the CO2-concentrating mechanism of cyanobacteria. Ph.D. Thesis, University of Toronto

Author notes

*

These authors contributed equally to the manuscript.

Supplementary data