Originally published In Press as doi:10.1074/jbc.M703896200 on August 3, 2007
J. Biol. Chem., Vol. 282, Issue 40, 29323-29335, October 5, 2007
Analysis of Carboxysomes from Synechococcus PCC7942 Reveals Multiple Rubisco Complexes with Carboxysomal Proteins CcmM and CcaA*
Benedict M. Long,
Murray R. Badger,
Spencer M. Whitney, and
G. Dean Price1
From the
Molecular Plant Physiology Group, Research School of Biological Sciences, Australian National University, Canberra, Australian Capital Territory 0200, Australia
Received for publication, May 11, 2007
, and in revised form, July 2, 2007.
 |
ABSTRACT
|
|---|
In cyanobacteria, the key enzyme for photosynthetic CO2 fixation, ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), is bound within proteinaceous polyhedral microcompartments called carboxysomes. Cyanobacteria with Form IB Rubisco produce 
-carboxysomes whose putative shell proteins are encoded by the ccm-type genes. To date, very little is known of the protein-protein interactions that form the basis of -carboxysome structure. In an effort to identify such interactions within the carboxysomes of the -cyanobacterium Synechococcus sp. PCC7942, we have used polyhistidine-tagging approaches to identify at least three carboxysomal subcomplexes that contain active Rubisco. In addition to the expected L8S8 Rubisco, which is the major component of carboxysomes, we have identified two Rubisco complexes containing the putative shell protein CcmM, one of which also contains the carboxysomal carbonic anhydrase, CcaA. The complex containing CcaA consists of Rubisco and the full-length 58-kDa form of CcmM (M58), whereas the other is made up of Rubisco and a short 35-kDa form of CcmM (M35), which is probably translated independently of M58 via an internal ribosomal entry site within the ccmM gene. We also show that the high CO2-requiring ccmM deletion mutant (
ccmM) can achieve nearly normal growth rates at ambient CO2 after complementation with both wild type and chimeric (His6-tagged) forms of CcmM. Although a significant amount of independent L8S8 Rubisco is confined to the center of the carboxysome, we speculate that the CcmM-CcaA-Rubisco complex forms an important assembly coordination within the carboxysome shell. A speculative carboxysome structural model is presented.
|
INTRODUCTION
|
|---|
Prokaryotic organisms are distinct from their eukaryotic counterparts by the absence of specialized membrane-bound compartments. Although it is commonly considered that prokaryotic cells do not compartmentalize cellular processes, there is growing evidence that cyanobacteria and some -proteobacteria have the means to do so when certain enzymes need to be sequestered or protected for maximal catalytic activity (1-3). Carboxysomes, for example, are polyhedral protein microcompartments found in cyanobacteria and chemoautotrophic bacteria, which contain the majority of a cell's ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) (4). These microcompartments are essential to the operation of the CO2-concentrating mechanism in cyanobacteria. The CO2-concentrating mechanism utilizes a number of inorganic carbon (Ci, the sum of CO2 and 
) transport systems to elevate cellular concentrations prior to CO2 fixation (for reviews, see Refs. 5 and 6). Rubisco is a hexadecameric enzyme with a core of eight large subunits (L8) made up of four L2 dimers, each containing two catalytic sites. Attached to each end of the L8 core are four small subunits (S8) that are necessary for catalytic competence (7). Rubisco carries out the CO2 fixation process in the initial step of the Calvin cycle, yet it is characterized by poor carboxylation kinetics, which is compounded by wasteful competing oxygenase activity. Carboxysomes overcome these shortcomings by elevating the CO2 concentration around Rubisco, thus maintaining a relatively high carboxylase activity while minimizing oxygenase activity. Cyanobacteria, as a class, are highly productive organisms on a global scale, and much of this productivity can be attributed to the efficiency of the CO2-concentrating mechanism.
Cyanobacterial cells with Form IB Rubisco contain this enzyme within -carboxysomes (8). The putative shell proteins of these protein microcompartments are coded for primarily by an operon of ccm-type genes (i.e. ccmKLMNO), and this operon is often closely associated with the genes for the large (rbcL) and small (rbcS) subunits of Rubisco (4). In addition, in most -cyanobacteria (9), a carboxysomal carbonic anhydrase (CA)2 is coded for elsewhere in the genome by the gene ccaA (10) (also known as icfA (11)). The required structural organization of the proteins encoded by these genes, in order to form mature carboxysomes, is a matter that has not yet been resolved. Nonetheless, progress is being made into understanding the protein complexes that make up -carboxysomes. From Synechocystis PCC6803, for example, Kerfeld et al. (12) have provided crystal structures of CcmK homologues and suggested that CcmK hexamers could form sheets that form the outer shell of the carboxysome. Likewise, So et al. (13) have provided evidence that the CcaA (carboxysomal CA) forms dimers via a particular portion of the C-terminal region. In 
-carboxysomes (species of cyanobacteria and proteobacteria possessing Form 1A Rubisco (8)), structural information is now available for the shell-bound CsoSCA (formerly known as CsoS3) protein (14), and various protein interactions within -carboxysomes have been identified by bacterial two-hybrid approaches (15). Despite these recent advances, much remains to be described in relation to the protein-protein interactions that operate to allow the formation of carboxysomes.
In this study, we have focused on CcmM as an important -carboxysomal shell protein, consistent with the observation that deletion of ccmM results in an absence of carboxysomes and a high CO2-requiring phenotype (16-18). In previous studies, we have consistently found CcmM to exist as 58- and 35-kDa protein products (referred to hereafter as M58 and M35, respectively) (19, 20), whose relative composition is not affected by protease inhibitors (20). It is therefore probable that both forms are translated from the same gene transcript and that each form has specific functions within the carboxysome shell. Here we present a novel gene complementation approach that utilizes the carboxysomeless Synechococcus PCC7942 mutant ccmM (16, 17) and wild type (WT) cells to investigate the binding partners of CcmM within carboxysomes. Plasmid DNA constructs containing chimeric forms of ccmM, coding for His6 tags at both the N and C termini of CcmM, were introduced into both WT and ccmM cell types. This allowed for unique tagging of both M58 and M35 within intact carboxysomes and the novel identification of Rubisco and CcaA as binding partners.
|
EXPERIMENTAL PROCEDURES
|
|---|
Expression of N- and C-terminally Tagged His6 CcmM Mutants—Tagged or untagged versions of CcmM where expressed from various shuttle vectors (detailed below) and expressed in either WT Synechococcus PCC7942 cells or in the ccmM markerless deletion mutant (17).
Shuttle Plasmid for N-terminal Hexahistidine Tagging of CcmM—The ccmM gene from Synechococcus PCC7942 was PCR-adapted with primers: forward, 5'-ttgggaggtggatccatggcgagcccaac (adding BamHI and NcoI sites at the underlined start codon); reverse, 5'-ggattcccggaattcttacggcttttg (adding an EcoRI site downstream of the underlined stop codon). The resulting 1.6-kb fragment was cloned into the NcoI and EcoRI sites of the Escherichia coli expression plasmid pTrcHisA (Invitrogen) and verified by DNA sequencing. This trc promoter (hybrid lac and trp promoter) in this plasmid is inducible with isopropyl 1-thio--D-galactopyranoside and features an expressed 4-kDa, N-terminal hexahistidine (His6) fusion protein region upstream of the NcoI insertion site. This expression plasmid was converted to a shuttle vector capable of replication in E. coli and Synechococcus PCC7942 via an addition of the 5.1-kb BamHI fragment (blunted) from pSK6B (21) into the ScaI site of the ampicillin resistance marker. The shuttle vector harbors a chloramphenicol resistance marker and was designated ptrc-H6ccmM (Fig. 1B).
Shuttle Plasmid for C-terminal Hexahistidine Tagging of CcmM—The ccmM gene from Synechococcus PCC7942 was PCR-adapted with primers: forward, 5'-ttagcatatgccgagcccaacaac (adding an NdeI site at the underlined start codon); reverse, 5'-atctagattactcgagcggcttttgaatcaacagttc (adding an in-frame XhoI site upstream of the underlined stop codon and XbaI downstream). The resulting 1.6-kb fragment was cloned into the NdeI and XbaI sites of the shuttle plasmid pSE4 (22, 23) and verified by DNA sequencing; this shuttle plasmid, known as pSE4-ccmM (Fig. 1B), possesses a nirA promoter from Synechococcus PCC7942 that is repressed in the presence of ammonium and derepressed in the presence of nitrate. A tag designated CH6 was added at the XhoI site to produce a C-terminal 27-kDa fusion tag comprising the structural gene for chloramphenicol acetyltransferase (CAT) and a C-terminal His6 affinity tag; the resulting shuttle plasmid was known as pSE4-ccmM-CH6 (Fig. 1B). Briefly, this involved the PCR adaptation of the CAT gene from pSK6B (21) with a forward primer 5'-aactcgagagatctatggagaaaaaaatcactggatatac (adding XhoI and BglII sites in frame with ATG) and a reverse primer 5'-tctcgagagatctgtgatgatgatgatgatgcgccccgccctgccactc (adding a His6 tag upstream of the stop codon and XhoI and BglII sites downstream) (Fig. 1B).
Co-expression of Rubisco and M35 in E. coli—Synechococcus PCC7942 Rubisco was expressed in E. coli XL1Blue cells transformed with ptrc7942LS as previously described (24). These cells were co-transformed with pACYCUb-M35, which directs expression of an N-terminally tagged His6-ubiquitin-M35 fusion (H6-Ub-M35). This was made by amplifying the M35 coding region with primers 5'-SacIIccmM1 (5'-ctccgcggtggaatgagcgcttataacggccaaggccgactc) and 3'-HindIIIccmM (5'-aagcttacggcttttgaatcaacagttc) and cloning the SacII-HindIII fragment into plasmid pHue (25) and then ligating the 1263-bp NcoI-HindIII H6-Ub-M35 coding sequence downstream of the trc promoter in pACYCtrc (24).
Culture Conditions—WT and mutant strains of Synechococcus PCC7942 were grown in modified BG11 medium (26) containing 20 mM HEPES-KOH, pH 8.0, at 30 °C and 
80 µmol of photons m-2 s-1. For carboxysome isolations, cultures were grown in 1-liter flattened Roux bottles and sparged with 2% (v/v) CO2 in air. For mutants harboring spectinomycin or chloramphenicol resistance markers, antibiotics were added to liquid cultures at 7 and 5 µg ml-1, respectively.
Carboxysome Isolations—Carboxysomes were enriched to the Triton-Percoll pellet stage (TP pellets) using the Percoll/Mg2+ method as described previously (20, 27) except that lysozyme treatment was carried out using the recombinant rLysozyme (Novagen, Madison, WI) at a concentration of 8 µg ml-1. TP pellets resulting from 1.0-liter cultures were resuspended with 0.5 ml of buffer (30 mM EPPS-NaOH, pH 8.0, 20 mM MgSO4, 20% (v/v) glycerol), separated into 50-µl aliquots, and snap-frozen in liquid N2 and then stored at -80 °C. Protein in TP pellets was estimated using the Bradford method (28).
Growth Rates—Cells were pregrown at 2% (v/v) CO2 and then diluted in fresh modified BG11 medium to a final A730 nm of 0.1, and triplicate 50-ml cultures were then grown at 30 °C, 80 µmol of photons m-2 s-1 illumination, and bubbled either with air or air enriched with 2% (v/v) CO2. Growth was monitored by hourly A730 nm measurements. Maximum growth rates were determined from the slope of logarithmic regressions of the data and transformed into doubling times.
Rubisco Assays—Rubisco activity in soluble protein extracts and purified complexes was determined using the 14CO2 fixation assay method of Badger and Price (29). The protein extracts were prepared from overnight cultures (80 ml) grown at 2% CO2. Cells were treated with lysozyme and lysed using a cold French pressure cell according to the method described by Satoh et al. (30). Protein and chlorophyll content of extracts was determined by the methods of Bradford (28) and Porra et al. (31), respectively. Synechococcus PCC7942 Rubisco small subunits (RbcS) used in complementation studies were purified using the procedure described previously (25).
Mass Spectrometric Measurements—Cells were prepared and analyzed in the mass spectrometer as previously described at 700 µmol of photons m-2 s-1 illumination (32, 33) with the modifications of Woodger et al. (17).
Immobilized Metal Affinity Chromatography (IMAC) Purification of Tagged Complexes—Enriched carboxysomes (TP pellets) containing 0.5 mg of protein were collected by centrifugation (16,100 x g, 1 min) and dissolved in 300 µl of IMAC binding buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 20 mM imidazole, 1 M urea). Purifications were also carried out under native conditions without urea. Insoluble material was removed by centrifugation (16,100 x g, 1 min), and the partially dissociated carboxysomes were transferred to 0.2 ml (packed bed volume) of prewashed nickel-charged Profinity IMAC resin (Bio-Rad) and incubated for 30 min at 4 °C with gentle inversion. The unbound fraction was retained for analysis, and the resin was transferred to a 14-cm column and washed with 20 ml of IMAC binding buffer by gravity flow. Bound proteins were eluted with 1 ml of IMAC elution buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 250 mM imidazole, 1 M urea). The eluate was maintained on ice and analyzed for Rubisco activity, protein content (see above), and SDS-PAGE (see below). Prior to SDS-PAGE analysis, the purified complexes were concentrated 20-fold using Biomax Ultrafree 30-kDa molecular mass cut-off membrane centrifuge filters (Millipore, Billerica, MA) and stored at -20 °C in 1% (w/v) SDS until required. IMAC purification, using the same procedure as above under native conditions, was also carried out on whole cell lysates from isopropyl 1-thio--D-galactopyranoside-induced E. coli cells expressing either Rubisco alone or Rubisco and H6-Ub-M35.
Protein Electrophoresis and Western Blotting—For SDS-PAGE, TP pellets or purified complexes were separated on NuPAGE 4-12% Bis-Tris gels (Invitrogen) according to the manufacturer's instructions using the MES-based electrophoresis buffer system. For native PAGE, 4-12% Bis-Tricine NuPAGE gels (Invitrogen) were run overnight at 4 °C using Tris-glycine native running buffer (25 mM Tris, 192 mM glycine, pH 8.3) at 80 V. Proteins were visualized using the GelCode Blue (Pierce) Coomassie Blue stain reagent or the PlusOne Silver Staining kit (GE Healthcare) according to the manufacturer's instructions. Western blots of SDS-PAGE-separated proteins were probed with Rubisco (34), CcmM, and CcaA antibodies and detected as described previously (19, 35).
N-terminal Sequencing of CcmM Short Forms from Synechococcus PCC7942 and Anabaena PCC7120—Carboxysomes enriched from both Synechococcus PCC7942 and the filamentous -cyanobacterium Anabaena PCC7120, using the Percoll/Mg2+ technique, were separated in duplicate by SDS-PAGE and electroblotted onto polyvinylidene difluoride membrane as described above. The short forms of CcmM from each species were first identified on one replicate blot by Western analysis using polyclonal rabbit serum raised against M35 from Synechococcus PCC7942. The remaining polyvinylidene difluoride blots were stained briefly with Coomassie Blue, and the bands corresponding to the short form of CcmM from each species were excised with a clean scalpel blade. The excised bands were destained with methanol and then air-dried before N-terminal sequencing analysis by seven cycles of Edman degradation chemistry (Australian Proteome Analysis Facility, Sydney, Australia).
Transmission Electron Microscopy—Cells were prepared for electron microscopy essentially as described by Price and Badger (32). Stained sections were viewed using a Hitachi H-7100FA transmission electron microscope at 75 kV. The maximum cross-sectional width of WT carboxysomes (n = 42) was determined from digital images of micrographs from sections of fixed cells, and the data were used to determine carboxysome dimensions for modeling purposes.
|
RESULTS
|
|---|
A series of gene complementation strategies in both Synechococcus PCC7942 WT and ccmM mutant cells were undertaken to elucidate the binding partner proteins that associate with the putative carboxysome shell protein CcmM. The carboxysomeless, high CO2-requiring ccmM mutant has a markerless deletion of the ccmM gene that does not interfere with the functional expression of the remaining ccmKLNO operon, thereby enabling carboxysome production to be restored by complementing the ccmM cells with ccmM genes expressed on a shuttle vector (16, 17) (Fig. 1A). We have previously described that the CcmM protein within enriched carboxysome fractions from WT cells is present as a full-length form (M58) and a short form (M35), the latter comprising the C-terminal region that contains three RbcS-like domains (19, 20, 36). Our novel complementation strategy therefore utilized both N-terminal and C-terminal His6-tagged variants of both the M35 and M58 peptides to identify their binding partner proteins in intact carboxysomes (summarized in Fig. 1B). The 27-kDa CH6 fusion was used as a large spacer intended to extend the His6 tag beyond the surface of CcmM.
Cell Growth and Physiology—Growth rate analysis of each cell type showed that doubling times at ambient and elevated CO2 concentrations are similar in all CcmM complementation mutants compared with WT cells (Table 1). Although there was some variation in the doubling times determined for each mutant, it is important to note that the ability for growth in air was recovered in all of the CcmM (both His6-tagged and untagged) complementation mutants of ccmM. Measurements of inorganic carbon (Ci)-dependent O2 evolution also showed that although ccmM cells have an increased requirement for Ci (12, 13), complementation with both WT and chimeric forms of ccmM reduced the Ci requirement to WT levels, although maximal O2 evolution rates did require higher Ci concentrations than WT, suggesting slightly impaired carboxysome formation (Fig. 2). Typically, rates at 1 mM Ci were around 20% lower than for WT. Rubisco activity, relative to both chlorophyll and protein concentrations, was similar in ccmM and its complemented mutant cell types compared with the WT and complemented WT controls, indicating that Rubisco content was not adversely affected by the variant CcmM complementations (Table 1).
View this table:
[in this window]
[in a new window]
|
TABLE 1
Doubling times and Rubisco content of WT, ccmM, and CcmM complementation mutant cell types Doubling times were determined from triplicate 50-ml cultures grown at 30 °C and 80 µmol of photons m-2 s-1. Data shown are mean values ± S.D. Rubisco activity in cell extracts from the various WT and mutant cell types was determined as described under "Experimental Procedures." Data shown are mean values ± S.E. of triplicate measurements.
|
|
View larger version (30K):
[in this window]
[in a new window]
|
FIGURE 1.
Synechococcus PCC7942 and ccmM operons and complementation gene constructs. A, a graphical depiction of the ccm operons of Synechococcus PCC7942 (WT) and the deletion mutant ccmM (17) showing the organization of the genes for the putative carboxysomal shell proteins CcmK1, -L, -M, -N, and -O. A putative internal ribosomal entry site (IRES) within ccmM (GTG) and the resulting restriction site (BamHI) in the ccmM ccm operon are indicated. B, ccmM gene complementation constructs were introduced using the vectors ptrcHisA (Invitrogen) and pSE4 (22, 51) with the recombinant genes under the control of trp-lac and nirA promoters, respectively. The IRES within the ccmM gene gives rise to a short, 35-kDa form (M35) of CcmM (see "Results"). The ptrc-H6ccmM construct codes for an N-terminal His6-tagged full-length form of CcmM, whereas pSE4-ccmMCH6 introduces a C-terminal chimera with a His6-tagged CAT protein, providing a 27-kDa spacer to allow projection of the tag away from the protein core. The protein products from each construct and their predicted molecular masses are indicated. An unintended IRES in CAT-6 x His (ATG) and the resulting protein product are also shown.
|
|
Carboxysome Formation—Although ccmM deletion mutants are unable to produce carboxysomes (16, 18, 36, 37), transmission electron micrographs of sectioned cells revealed that carboxysomes were produced in all of the CcmM complementation mutants of both ccmM and WT genetic backgrounds (Fig. 3) (data not shown for the ccmM mutant expressing untagged CcmM). All carboxysomes appeared to have a normal shape, although some larger sized carboxysomes were evident in the His6-tagged CcmM mutants compared with the more uniformly sized carboxysomes in WT (and ccmM expressing untagged CcmM; data not shown) cells that had an average maximum cross-sectional width of 172 ± 26 nm (n = 42).
Carboxysome Enrichment—Carboxysome-enriched preparations (TP pellet fractions) from all cell types excluding ccmM were isolated using the Percoll/Mg2+ technique developed for Synechococcus PCC7942 cells (20, 27). There was no appreciable variation in yield or quality of carboxysomes from any of these cell types. Analysis of TP pellet fractions by SDS-PAGE revealed a consistent co-purification of several carboxysomal proteins from each mutant that were identified as the Rubisco large subunit (RbcL) and RbcS, both forms of CcmM, and, identified for the first time in TP pellets, CcmK1 and CcaA (Fig. 4A). CcmK1 was unambiguously identified by proteomic techniques from carboxysome preparations (Fig. 4 and Table 2) after identification of a clear protein band at the estimated CcmK1 mass (8 kDa). Due to its size, this protein is often diffuse on protein gels, and resolving it by Coomassie staining required sufficient fixation and minimal washing steps. As a result, its relative composition in carboxysome-enriched preparations is often variable (Table 3 and Fig. 4). The identification of this protein in TP pellets, where it has not been identified in the past, was made possible by refinements of the enrichment technique, such as the use of recombinant lysozyme and care in avoiding excessive contamination from cell debris after the cell lysis step (20). Likewise, CcaA is visible only as a faint band on Coomassie-stained gels of WT carboxysome preparations and usually requires Western blotting analysis for identification (Figs. 4, A and B, and 5, B and C) (19). In both ccmM and WT cells expressing H6ccmM (N-terminally tagged M58), however, CcaA is visibly more abundant (Fig. 4A), enabling its identity to be confirmed by proteomic analysis of the protein in the TP pellets from these cell types (Table 2).
View this table:
[in this window]
[in a new window]
|
TABLE 2
Proteomic identification of CcmK1 and CcaA protein bands from SDS-polyacrylamide gels of enriched carboxysomes Protein bands from Coomassie Blue-stained gels were excised, destained, reduced with iodoacetamide, and digested with trypsin using standard protocols prior to separation using reverse phase chromatography and analysis on a Thermo-Finnigan LCQ DecaXP Plus ion trap mass spectrometer. Peptides were identified using the MASCOT MS/MS ion search algorithm and the annotated Synechococcus PCC7942 protein sequence data base dated 10 November 2005 (available on the World Wide Web). The data base was searched with variable methionine oxidation (+16) and cysteine carbamidomethylation (+57) with up to two missed trypsin cleavages per peptide.
|
|
View this table:
[in this window]
[in a new window]
|
TABLE 3
Protein stoichiometries in enriched carboxysomes and IMAC-purified Rubisco-CcmM complexes Relative quantities of carboxysomal proteins in enriched carboxysomes (TP pellets) and from IMAC-purified CcmM-protein complexes estimated by densitometric analysis following separation by SDS-PAGE and staining with Coomassie Blue. The stoichiometries are normalized relative to RbcL levels that are assumed to be correctly assembled into structurally stable octameric (L8) cores. Values are presented as means ± S.D. with the number of samples analyzed by SDS-gels in parentheses.
|
|
CcmM Fusion Proteins—As has been reported previously for Synechococcus PCC7942, CcmM from WT carboxysomes is consistently observed as two polypeptide forms (19, 20) with apparent masses of 38 and 58 kDa on SDS-polyacrylamide gels (Fig. 4A, WT carboxysomes). The smaller of the two, M35, corresponding to a predicted sequence mass of 35 kDa based on N-terminal sequence analysis, represents the C-terminal region of the polypeptide, containing three RbcS-like repeats (36, 38). This form is consistently observed as the most abundant form in enriched carboxysome fractions from Synechococcus PCC7942 (Fig. 4A, WT carboxysomes) and is possibly translated independently from the same transcript (20). The larger of the two polypeptides (M58; Fig. 4A, WT carboxysomes) is of lower abundance and includes both the RbcS-like repeats domain and the N-terminal 
-CA-like domain (36, 39). M35 and M58 are now commonly observed at a ratio of between 4 and 5:1 in high quality WT TP pellets in our hands (Table 3).
In the absence of carboxysomes, no carboxysomal proteins could be enriched from the ccmM mutant using the Percoll/Mg2+ technique (data not shown). When complemented with untagged and His6-tagged CcmM variants, carboxysome proteins were recovered in the TP pellets of ccmM (Fig. 4A). Both M35 and M58 were recovered from ccmM cells expressing untagged CcmM, and the 62-kDa H6-M58 and M35 were recovered from ccmM cells expressing N-terminally tagged CcmM. From WT expressing N-terminally tagged CcmM, both H6-M58 and M58 were isolated in the enriched TP pellets as well as M35 (Fig. 4A). Expression of the CcmM-CH6 fusion in ccmM and WT cells led to enrichment of both an 85-kDa M58-CH6 and 63-kDa M35-CH6 peptide in the TP pellets, the latter being produced at levels >5-fold that of any of the other His6-tagged CcmM variants (Fig. 4A). In WT cells expressing the C-terminal tag, both M35 and M58 where also isolated, the production of the former confirmed using Western blot analyses with an antibody to CcmM to distinguish between the similarly sized M58 and M35-CH6 (58 kDa on SDS-polyacrylamide gels) peptides in diluted TP pellet protein preparations (data not shown).
Rubisco-CcmM Subcomplexes—The most interesting outcome of this study is that RbcL and RbcS were co-purified with both N- and C-terminally tagged forms of CcmM (from both WT and ccmM genetic backgrounds), even in the presence of up to 4.0 M urea (Figs. 4B and 6), indicating that CcmM and Rubisco strongly interact in whole carboxysomes. Control applications of WT carboxysome preparations to IMAC yielded neither CcmM nor Rubisco proteins in the eluate (Fig. 4B); the same observation was made using ccmM cells expressing untagged CcmM (data not shown), indicating that His6 tags are required for purification. A small amount of free CH6 protein was also enriched in IMAC eluates from C-terminally tagged mutant carboxysome preparations due to an unintentional start codon and ribosomal binding site (Figs. 1, 4B, and 5A). The co-purification of only very small amounts of untagged M35 with N-terminally tagged complexes (from both ccmM and WT cells) and the C-terminally tagged complex (from WT cells) (Fig. 5A) indicates only a weak or indirect interaction between the two forms of CcmM. The only short form of CcmM found in complexes purified from ccmM expressing the C-terminal tag was M35-CH6 (Fig. 5A).
Rubisco-CcmM-CcaA Subcomplex—In both WT- and ccmM-expressing H6-M58, significant amounts of CcaA were co-purified with Rubisco and CcmM peptides by IMAC and to a lesser extent from carboxysome preparations of C-terminally tagged CcmM expressed in WT cells (Fig. 5A). To examine the binding partner of CcaA, its stoichiometry relative to RbcL and M58 was examined on immunoblots. CcaA was purified in relatively large quantities by IMAC from carboxysomes (both from WT and ccmM lines) complemented with H6-M58 (Fig. 5). In the TP pellets from these cells, the quantities of M58, relative to Rubisco, were 7-8-fold greater compared with their C-terminally tagged counterparts (Table 3 and Fig. 4A). By comparing their relative abundance, a reproducible H6-M58/CcaA stoichiometry of 2:1 in the TP pellets was measured (Table 3). In the IMAC-purified protein preparations, the amount of CcaA relative to RbcL was significantly greater in ccmM carboxysomes containing H6-M58 than in those containing M58-CH6 (Table 3 and Fig. 5B). In contrast, the level of CcaA recovered by IMAC was proportional to the amount of His6-tagged M58 peptides (Fig. 5C), suggesting that CcaA interacts directly with M58, not Rubisco, in carboxysomes. In addition, CcaA was often observed as a dimer on immunoblots if not fully reduced prior to separation by SDS-PAGE, consistent with suggestions that it forms dimers in vivo (13, 19).
Rubisco Activity in Purified Complexes—Importantly, the IMAC-purified complexes exhibited Rubisco activity, and this activity could be enhanced to varying degrees by the addition of purified RbcS (Table 4). This strongly suggests the RbcL associated with both M35 and M58-CcaA complexes is likely to be assembled in stable L8 cores. Recovery to near control levels of Rubisco activity was achieved in the C-terminally tagged complex (predominantly M35-Rubisco) after supplementing with additional RbcS (Table 4). In contrast, the addition of saturating RbcS to IMAC-purified N-terminally tagged complexes (predominantly M58-CcaA-Rubisco) did improve Rubisco activity almost 2-fold but was not enough to recover full activity (Table 4), suggesting that diminished activity is due to the presence of CcaA and M58. Attempts to disrupt CcmM-Rubisco complex binding by the addition of an excess of purified RbcS during the binding phase of IMAC did not affect either the complement of the eluted protein or its Rubisco activity (data not shown).
View this table:
[in this window]
[in a new window]
|
TABLE 4
Rubisco activities of purified carboxysome complexes Relative activities of L8S8 Rubisco (control) and several purified carboxysomal CcmM-Rubisco complexes from Synechococcus PCC7942. Activities were determined as described previously (29). Specific activity of the L8S8 control was 0.59 ± 0.02 µmol/min/mg protein. The cell type from which each complex has been purified using IMAC is indicated. The purified complexes have been analyzed either with (+RbcS) or without RbcS added to the reaction mixture.
|
|
View larger version (126K):
[in this window]
[in a new window]
|
FIGURE 3.
Cell ultrastructure. Transmission electron micrographs of sections of Synechococcus PCC7942 cells (WT) and ccmM complementation mutants in both WT and ccmM (carboxysomeless) genetic backgrounds. Carboxysomes are indicated by arrowheads. Note the tendency for larger carboxysomes in tagged CcmM mutants, although small carboxysomes are also evident. Scale bar, 500 nm.
|
|
View larger version (67K):
[in this window]
[in a new window]
|
FIGURE 4.
Carboxysome enrichment and CcmM complex purification. A, Coomassie Blue-stained SDS-polyacrylamide gel showing typical protein banding patterns from enriched carboxysome preparations (TP pellets) from WT and tagged CcmM fusion mutants in both WT and ccmM genetic backgrounds. The peptide terminal location (N or C) of the tags is indicated. WT carboxysomes and the WT form of ccmM containing no tag expressed in ccmM serve as controls. Note that N-terminal tagging gives rise to the production of both H6-M58 and untagged M35; C-terminal tagging results in both M58-CH6 and M35-CH6; WT genetic backgrounds also produce untagged M58 and M35 (see Fig. 1). The positions of tagged and WT forms of CcmM and several identified carboxysomal proteins are indicated. The identities of M58, RbcL, M35, and RbcS were confirmed previously by proteomic analysis (20), as were CcaA and CcmK1 in this study. Specific antibodies to Rubisco, CcmM, and CcaA were also employed. His6-tagged forms of CcmM and CH6 were identified using both CcmM-specific and His6-specific antibodies and by proteomic analysis (data not shown). Approximately equal amounts of RbcL have been applied to the gel in each case. B, Coomassie Blue-stained SDS-polyacrylamide gel showing the IMAC purification of His6-tagged CcmM and associated RbcL and RbcS enriched from carboxysome fractions (TPp) from N- and C-terminally tagged forms of CcmM in both WT and ccmM genetic backgrounds. Purification was carried out in the presence of 1 M urea. The unbound IMAC fractions (Ub) and IMAC eluate fractions (E) are indicated, as are the identities of purified proteins. Note the absence of protein in the eluate fraction from WT carboxysomes. Approximately equal amounts of Rubisco from TP pellets of each cell type were applied to IMAC, and equivalent volumes of the enriched carboxysomes and unbound fractions were loaded onto the gel. One-fifth of each eluate fraction was loaded onto the gel. Note that considerably less protein is purified from the N-terminally tagged CcmM constructs.
|
|
Protein Stoichiometries in the Subcomplexes—The similar mobility of Rubisco in the IMAC-purified proteins compared with purified Synechococcus PCC7942 Rubisco through nondenaturing PAGE (Fig. 7B) suggested that the isolated RbcLs were assembled into comparably sized oligomeric enzyme complexes (presumably in the form of stable octameric (L8) cores). Notably, the L8 cores were not fully complemented with eight RbcS, as evidenced from densitometry estimates of relative subunit levels (Table 3); the addition of RbcS stimulated Rubisco activity in the IMAC-purified Rubisco-CcmM complexes (Table 4) and their slightly faster mobility through nondenaturing PAGE (Fig. 7B). Notably, the yield of Rubisco that co-purified with the His6-tagged CcmM forms by IMAC decreased with increasing urea concentrations (Fig. 6). A linear correlation was found between the amount of RbcL (assumed to be assembled into octamers up to 4 M urea) that co-purified with M35-CH6 by IMAC and the urea concentration used during purification. Estimates of Rubisco/CcmM stoichiometries from M35-CH6 complexes purified in the absence of urea gave an L8/M35-CH6 ratio of 1:4. However, due to extremely low protein yields when urea was not used (Fig. 7A), these data were both difficult to obtain and confirm. Nonetheless, purification of the same complex over a range of urea concentrations and extrapolation to 0 M urea confirmed this stoichiometry (Fig. 6). Notably, this ratio does not account for an association between RbcL and the >20-fold less abundant M58-CH6 peptide or the possibility a proportion of the Rubisco-M35-CH6 subcomplexes have dissociated during IMAC purification. Thus, we postulate from these preliminary analyses a binding ratio of between one and four CcmM subunits per L8 Rubisco.
Optimum Release of Carboxysome Complexes—It is important to note that protein complexes could be purified intact in the absence of urea but that yields were either unacceptably low or highly variable under these conditions (Fig. 7A). Inclusion of urea during IMAC purification was required to consistently liberate substantial amounts of protein subcomplexes from whole carboxysomes. Although repeated freeze-thawing of carboxysomes can be used to enrich shell proteins (e.g. from Synechocystis PCC68033 and Halothiobacillus neapolitanus (40)), it failed to enhance the recovery of shell proteins from TP pellets of either His6-tagged CcmM complementation mutant or WT Synechococcus PCC7942 cells (data not shown). Only partial denaturation of carboxysomes using 1 M urea provided a consistent and significant increase in the yield of purified tagged protein without substantially disrupting their association with other proteins or the oligomeric Rubisco complex (Fig. 7). The requirement for mild denaturation to consistently retrieve the His6-tagged CcmM variants suggests the possibility that both the N terminus of M58 and the C terminus of both M35 and M58 are not accessible to the carboxysome surface. Alternatively, IMAC binding of the His6 tags on intact carboxysomes is too weak, but urea breaks up the carboxysome sufficiently to allow binding. Consistent with this idea is the finding that the same protein complement is purified in the absence of urea (Fig. 7A), suggesting that the material purified under native conditions represents complexes that have been mechanically disrupted from semi-intact carboxysomes. Importantly, in 1 M urea, effectively all of the His6-tagged CcmM peptides isolated in the TP pellets were recovered by IMAC along with reproducible relative quantities of Rubisco, despite most of the Rubisco in the TP pellets (60-95%) not being retained by IMAC (Figs. 4B and 5A). This is consistent with the idea that most Rubisco is internalized within carboxysomes and is not bound to shell proteins.
Confirmation of Complex Formation—Confirmation of M35-Rubisco binding was achieved by co-expression of soluble hexadecameric (L8S8) Rubisco in E. coli (16) with M35 N-terminally tagged with a His6-ubiquitin fusion (H6-Ub-M35). Neither RbcL nor RbcS could be isolated by IMAC from cells expressing soluble Rubisco alone (Fig. 8). However, IMAC was successful in isolating both RbcL and RbcS when H6-Ub-M35 was co-expressed with Rubisco. We intend to develop this system for further analysis of M35 and M58 interactions with Rubisco and CcaA (work in progress). However, the interaction between CcmM and CcaA from Synechocystis PCC6803 has been independently observed in yeast two-hybrid studies.3
|
DISCUSSION
|
|---|
Carboxysomal Rubisco Complexes—We have suggested previously that the RbcS-like domains in the C-terminal region of CcmM could play a role in Rubisco/CcmM interactions and carboxysome assembly (38). Here we show, using both N- and C-terminal His6-tagged forms of CcmM, the first experimental data indicating that CcmM and Rubisco (RbcL and RbcS) form at least two independent and active protein complexes within Synechococcus PCC7942 carboxysomes. One of these complexes is also associated with the carboxysomal carbonic anhydrase, CcaA, via the N-terminal region of M58 (the 58-kDa form of CcmM). Given that a substantial quantity of carboxysomal Rubisco is not purified from CcmM-tagged carboxysome preparations using IMAC (Fig. 4B), it is likely that these two Rubisco complexes augment the independent L8S8 Rubisco holoenzymes that make up the bulk of the carboxysome. Thus, we assume that Rubisco-CcmM complexes are associated with the carboxysome shell. N-terminally tagged M58 from both WT and ccmM cell types was not capable of significantly enriching M35 (the 35-kDa form of CcmM) via IMAC (Fig. 5A), suggesting that both forms of CcmM do not substantially interact with each other. Thus, there are likely to be at least two independent CcmM-Rubisco complexes (viz. M35-Rubisco and M58-CcaA-Rubisco). Using M35 as a reference, we estimated the RbcL/CcmM stoichiometry to be at most L8M4.
View larger version (16K):
[in this window]
[in a new window]
|
FIGURE 6.
Urea-dependent dissociation of CcmM-Rubisco complexes. The apparent ratio of the tagged short form of CcmM (M35-CH6) to RbcL in the complexes purified from ccmM expressing the C-terminally tagged form of CcmM, as determined by densitometric analysis of Coomassie-stained SDS-polyacrylamide gels. The complexes were purified by IMAC over a range of urea concentrations from 0 (native conditions) to 4 M. At high concentrations of urea, there is a preferential purification of M35-CH6 due to complex dissociation. Based on SDS-PAGE and immunoblot analysis (Figs. 4 and 5), complexes containing Rubisco and the full-length CcmM tag (M58-CH6) resulting from this cell type are assumed to make an insignificant contribution to the total purified protein in this case.
|
|
We currently have no evidence to suggest that RbcS has been displaced by CcmM. However, we note that the addition of RbcS to the purified complexes increases their Rubisco activity, indicating that RbcS binding sites are available in the purified, partially dissociated complexes (Table 4). It is possible that CcmM cross-links Rubisco complexes together, occupying RbcS binding sites on one face of Rubisco. Under the purification conditions used in this study, some cross-links may be lost from purified complexes, with only the strongest binding RbcS-like repeat of CcmM remaining in contact with L8. Under such circumstances, it is conceivable that the addition of RbcS allows occupation of the now available sites and a subsequent increase in activity (Table 4). Notably, the addition of purified RbcS during the IMAC binding phase did not disassociate Rubisco from CcmM (data not shown), suggesting that the binding of CcmM to Rubisco is either unrelated to the RbcS binding domain or, more likely, that it is much stronger than the interaction between RbcL and RbcS.
View larger version (67K):
[in this window]
[in a new window]
|
FIGURE 7.
Effect of urea on carboxysomal complexes. A, typical Coomassie-stained SDS-polyacrylamide gel lanes (upper panels) of CcmM-associated complexes purified by IMAC either with (+) or without (-) urea in the IMAC buffers. The lower panels show immunoblots for CcaA from the same IMAC eluates. Note that protein recovery in the presence of 1 M urea is enhanced, but the same proteins are observed when urea is omitted, confirming that the native complexes are represented in 1 M urea eluates. In this typical example, complexes have been purified from WT cells expressing the N-terminally tagged form of CcmM (H6-M58). B, a Coomassie-stained native polyacrylamide gel showing the effect of urea concentration on carboxysomal complex stability. Aliquots (50 µg) of enriched carboxysome TP pellet fraction from WT Synechococcus PCC7942 were collected by centrifugation (1 min at 16,000 x g) and resuspended in 10 µl of 0.75 x EM buffer (30 mM EPPS-NaOH, pH 8.0, 20 mM MgSO4) at one of the urea concentrations indicated for 5 min at room temperature. Insoluble material was collected by centrifugation, as above, and supernatants were analyzed by native PAGE on a 4-12% Tris-glycine gel (Invitrogen) using native PAGE running buffer (2.9 g/liter Tris, 14.4 g/liter glycine). An aliquot of purified L8S8 Synechococcus Rubisco (2.5 µg) was analyzed as a control. Note that there is some release of predominantly Rubisco L8S8 complexes at urea concentrations as low as 0.5 M, but solubilization of Rubisco L8S8 complexes from carboxysomes is maximal in the 2-4 M range. Beyond 4 M urea, Rubisco dissociates into monomeric subunits.
|
|
View larger version (32K):
[in this window]
[in a new window]
|
FIGURE 8.
Co-expression of Rubisco and M35 in E. coli. E. coli cells expressing either Rubisco only (LS) or Rubisco and M35 (LS-M35) N-terminally tagged with a His6-ubiquitin fusion (H6-Ub-M35) were induced overnight with 0.5 M isopropyl 1-thio--D-galactopyranoside, lysed, and subjected to immobilized metal affinity chromatography, without urea, as described under "Experimental Procedures." Eluted protein was applied to SDS-polyacrylamide gels and stained with Coomassie Blue (A) or subjected to Western blotting (B and C), followed by probing with a Rubisco-specific antibody that recognizes both RbcL and RbcS. The identity of H6-Ub-M35 was confirmed by immunoblots (data not shown). No protein was recovered by IMAC from cells expressing Rubisco only, but H6-Ub-M35 and its binding partner Rubisco were recovered from cells where both were expressed.
|
|
CcaA as a Shell Protein—The finding that there is indeed an interaction between M58, Rubisco, and CcaA places CA activity in close proximity to Rubisco, allowing for rapid transfer of CO2 to the site of fixation. It also suggests that CcaA is likely to be associated with the carboxysome shell. This is analogous to the findings for the -carboxysomal CA (CsoS3), which has been localized to the carboxysome shell of H. neapolitanus (41, 42), and partly supports a model put forward by Reinhold et al. (43), which places CcaA within bundles of Rubisco.
The abundance of CcaA in N-terminally tagged M58 containing carboxysomes (Fig. 4, A and B) and its co-purification from these carboxysomes (Figs. 4B and 5) allows us to conclude that CcaA interacts primarily with the N-terminal, -CA-like region of M58. This region of M58 bears significant, and probable structural, homology (60% amino acid similarity) with Cam, the trimeric -CA of Methanosarcina thermophila (36, 39, 44); notably, M58 can be threaded onto the -CA structure, suggesting that CcmM could also form a trimer (Fig. 9). In addition, the N-terminal region of M58 contains a conserved glutamine and three histidine residues required for Zn2+ coordination and CA activity in Cam (36, 39, 44). Nonetheless, CA activity in cyanobacterial CcmM proteins has not been described to date, and we would speculate that the N-terminal region of M58 instead performs a structural role as a scaffold for CcaA and perhaps other carboxysomal proteins. Thus, it is likely that a possible trimeric form of M58 has a role in recruiting CcaA into the carboxysome. This idea is supported by the apparent high abundance of CcaA in WT and ccmM cells expressing N-terminally tagged CcmM, which contain high amounts of tagged M58 (Figs. 4, A and B, and 5A), and the consistent ratio of M58 to CcaA in purified complexes (Fig. 5, B and C). The point of interaction between CcaA and M58 is speculative, but it is important to note that CcaAs (as a group of -CAs) characteristically differ from their higher plant homologues by the presence of a 75-78-amino acid C-terminal extension (13). In Synechocystis PCC6803, only a small portion of this extension (8-18 amino acids) is absolutely required for CcaA dimerization and catalysis (13). Perhaps much of the remainder of this C-terminal extension of CcaA is required for M58 interaction. So et al. (13) also showed that CcaA does not interact directly with either RbcL or RbcS, providing additional evidence that the interaction observed in this study is likely to be directly between the N-terminal region of M58 and CcaA.
As noted above, the interaction of CcaA with the lower abundance M58 has implications for the relative amount of CcaA within a carboxysome. Very little CcaA would be needed within a carboxysome to provide the required rate of bicarbonate dehydration (43). By controlling the level of M58, perhaps through both competitive translation of the transcript into M35 and control of ccaA transcript abundance (45), there can be subsequent control of CcaA content within the carboxysome.
It is significant that isolated CcmM-Rubisco complexes are functional for Rubisco activity (Table 4); however, in regard to any potential kinetic advantage ensuing from association with CcaA, it should be noted that it is technically difficult to test for a CcaA-enhanced level of Rubisco activity in isolated complexes or isolated carboxysomes. This is because the critical cytoplasmic chemical disequilibrium of Ci species (CO2 and ) achieved in vivo by the action of the CO2-concentrating mechanism, favoring a low CO2 to ratio, cannot be achieved in vitro (21, 46). Nonetheless, studies (10, 27, 47) have shown that mutants lacking CcaA activity, while forming carboxysomes, are unable to grow at ambient CO2, thus revealing the functional importance of CcaA in the complex described in this report. Likewise, disruption of CcmM results in the inability to form carboxysomes and grow at ambient CO2 conditions (17, 36).
View larger version (33K):
[in this window]
[in a new window]
|
FIGURE 9.
Carboxysome shell model. A proposed model for the formation of carboxysome shell facets based upon the interactions between Rubisco, M35, M58, and CcaA in Synechococcus PCC 7942. A, a scale diagram of an icosahedral shaped carboxysome showing detail of proposed Rubisco and CcaA complex locations within a single facet. B, an upper surface view of the proposed arrangement of Rubisco-CcmM complexes within the carboxysome shell. The small subunit-like repeats of CcmM are indicated by yellow arrows. A trimer of M58 is depicted in the center of the diagram. The triangular shape of the N-terminal region and relative arrangement within the trimer is based upon the crystal structure of the Cam trimer from M. thermophila (44) and is drawn approximately to scale. A proposed CcaA dimer is shown bound to one M58 subunit, with other potential CcaA binding sites indicated as empty circles. The arrangement of M35 around this structure indicates the potential for cross-linking to neighboring Rubiscos within the shell layer. C, a side-on view of proposed structures of the purified complexes of M35-Rubisco and M58-CcaA-Rubisco. The N-terminal region of M58 is depicted by a pink box in this side view and is bound to a CcaA dimer. The RbcS-like repeats within M35 and M58 are shown interacting with small subunit (SSU) binding domains on L8 Rubisco. To indicate the potential for binding of either Rubisco small subunits or CcmM to Rubisco large subunits, the Rubisco small subunit binding domains are shown. The size of Rubisco and the N-terminal region of CcmM are drawn approximately to scale based on crystal structures of Synechococcus PCC6301 Rubisco (52) and the M. thermophila -CA, Cam (44). D, a cross-sectional view of the shell layer indicating possible positioning of the M58 N-terminal region and CcaA. Cross-links between Rubiscos are indicated by the arrows representing small subunit-like repeats of CcmM. The shell protein CcmK1, as its pore-forming hexamer, is shown layered on top of the interconnected Rubisco complexes. This structure is drawn approximately to scale based upon the crystal structure of CcmK2 from Synechocystis PCC6803 (12).
|
|
Model of Rubisco-CcmM Complexes—It has been shown previously that ccmM codes for a protein product with a distinct N-terminal -CA moiety (39) and a C-terminal region of three RbcS-like repeats (36, 38). Indeed, it was speculated that the RbcS-like repeats may be involved in RbcL binding (38). In addition, previous work from our group suggests that the RbcS-like region of CcmM (M35) is translated independently of the full-length form of CcmM (M58), with both forms of CcmM being consistently found in carboxysome preparations from Synechococcus PCC7942 (19, 20) and now in Anabaena PCC7120 (this study). N-terminal sequences determined from the short forms of CcmM from both species confirm that this is a likely scenario (Fig. 10). We also predictably find the respective short forms of CcmM in ccmM complementation mutants, expressing both N- and C-terminal tags, in enriched carboxysome fractions (Figs. 1B and 4A). Thus, we find carboxysomes containing two forms of CcmM, both having the potential to bind RbcL. By producing both N- and C-terminally tagged forms of CcmM, thus producing tagged forms of both M35 and M58 (Fig. 1), we have shown here that the RbcS-like region of CcmM does indeed bind to Rubisco. In addition, M58 binds to the carboxysomal carbonic anhydrase, CcaA, via the N-terminal CA-like region.
Given that the bulk of Rubisco is not bound to CcmM (Fig. 4B) and therefore likely to represent the independent Rubisco complexes within the carboxysome, we conclude that CcmM-Rubisco complexes are associated with the carboxysomal shell. Since the carboxysomal shell is likely to form a single layer of protein rather than concentric inner layers, we have assumed that CcmM binds to one side of Rubisco such that CcmM is toward the outer surface of the carboxysome. This would provide sidedness to the complexes and thus allow arrangement of a layer of Rubisco in close association with the shell (Fig. 9). This may also provide a suitable arrangement that allows subsequent packing of L8S8 Rubisco within the carboxysome interior. The presence of three RbcS-like repeats within the C-terminal region of CcmM indicates three potential sites for binding to Rubisco (Fig. 9). This provides several possible binding arrangements between Rubisco and CcmM that allow for one, two, or three of these repeats to cross-link with neighboring Rubisco complexes. Thus, we can envisage a network arrangement across the facets of the carboxysome, predominated by M35 (Fig. 9). It is reasonable to assume that M58 interacts with Rubisco in the same manner (but at a lower frequency due to its relative abundance), so that occasional complexes containing M58 and CcaA might be interspersed with M35-Rubisco complexes within the facets of the shell (Fig. 9).
It has been proposed that the N-terminal, -CA-like, region of CcmM could form trimers (44). If so, M58-CcaA-Rubisco complexes might also form trimers via trimerization of this domain. Given the composite nature of this complex, it is also possible that this may therefore form part of a specific structure within the carboxysome shell, such as a vertex, edge, or perhaps a central node or nodes within a facet (Fig. 9). Indeed, the formation of either M35- or M58-Rubisco complexes could potentially provide a seed nucleation complex for the formation of the shell or individual facets (Fig. 9). It is also of note that the putative carboxysomal protein CcmN contains the bacterial transferase hexapeptide repeat domains that are found in the N-terminal region of CcmM and are significant to the tertiary structure of Cam from M. thermophila. Whether this indicates potential structural homology between CcmM and CcmN is yet to be described.
Schmid et al. (48) recently revealed that -carboxysomes of H. neapolitanus are icosahedral in shape. Both - and -carboxysomes appear as slightly elongated hexagons in medial sections viewed by electron microscopy, so we assume that -carboxysomes could also be icosahedral. This shape suggests a carboxysome with 20 facets, each of which approximates an equilateral triangle. Based on our own transmission electron microscopy data on WT Synechococcus PCC7942 carboxysomes, we estimate an average maximum cross-sectional diameter of 172 ± 26 nm (S.D.), which equates to triangular facets that contain 68 Rubisco molecules (Table 5). In the model presented in Fig. 9, we have envisaged three trimers of M58-CcaA-Rubisco complexes within the proposed shell facets. We propose that the center of the trimer occupies a volume similar to that of L8S8 Rubisco, therefore occupying a space within regularly packed Rubisco molecules within the shell. This model allows for the formation of facets consisting of 63 Rubisco molecules interacting with 41 M35 and nine M58 polypeptides (consistent with our observed ratio of between 4 and 5:1 M35/M58 in WT carboxysomes; Table 3) with up to nine CcaA dimers per facet. Given that carboxysomal CA activity (27) and ccaA transcript abundance (45) may vary in response to external CO2 supply, we have assumed that not all potential CcaA binding sites within the shell are necessarily filled (Fig. 9). This allows for up to 180 CcaA dimers per carboxysome, although higher order oligomerization, which may be required for CA activity (13), may limit the number of active CA complexes within the carboxysome to somewhat less than this figure. We propose that the cross-linking of CcmM and Rubisco forms the basis for other proposed shell proteins (i.e. CcmK1, -L, -N, and -O) to bind, thus linking facets and completing the carboxysome shell. For example, CcmK has been shown to form plate-like hexamers and could interlock to form large sheets (12). We therefore assume that the lower face of shell-bound Rubisco allows for regular packing of independent L8S8 holoenzymes and formation of the carboxysome as a whole.
View this table:
[in this window]
[in a new window]
|
TABLE 5
Carboxysome dimension estimates Carboxysome length, area, and volume dimensions are based on an icosahedron.
|
|
Notably, we find a shortfall in the amount of CcmK1 that would be required to completely cover the surface of the carboxysome (Tables 3 and 5). Whether this is due to loss of CcmK1 proteins during isolation, as evidenced by variability in abundance between preparations (Table 3), or truly reflects the relative coverage of carboxysomes by this protein is yet to be determined.
The model presented here (Fig. 9) is based on the presence of three RbcS-like repeats in the short form of CcmM. It is worth noting that many CcmM homologues from other -cyanobacteria contain four RbcS-like repeats in this domain (36). In carboxysomes containing these forms of CcmM, it is possible to conceive of a similar formation of complexes, perhaps with one less RbcS per Rubisco/CcmM interaction. It is interesting to note, however, that although Anabaena PCC7120 also has a CcmM with three RbcS-like repeats and a -CA-like domain, to date no CcaA homologue has yet been found in this species. As such, the homologous M58-Rubisco complex that might be found in Anabaena PCC7120 may not bear strict resemblance to that described here.
Complementation of ccmM Results in Carboxysome Formation—Previous studies from this group have characterized ccmM as a carboxysomeless mutant with a high CO2-requiring phenotype (16, 17, 45). Here we have shown for the first time that complementation of this mutant with both WT and recombinant forms of ccmM results in production of carboxysomes (Fig. 3) and substantial recovery to near normal physiology (Fig. 2 and Table 1). Interestingly, complete recovery of Ci-dependent O2 evolution to WT levels was not observed (Fig. 2), and carboxysomes appeared to be somewhat more disparate in size in complementation mutants (Fig. 3). We speculate that this may be due to different transcript control between plasmid-based ccmM genes and the chromosomal genes, resulting in a mismatch of the produced levels of CcmM and other carboxysomal proteins. If we assume that recruitment of CcaA into the carboxysome is dependent upon the availability of M58, then we might expect that M58 overexpression will result in high levels of CcaA within carboxysomes. Indeed, this is the case in the ptrc-H6ccmM complementation mutants, where both M58 and CcaA content of carboxysomes appears to be enhanced (Fig. 4A). If we assume that the interaction of M58 and CcaA can occur independently of mature carboxysome formation, an overproduction of full-length CcmM may also result in an excess of non-carboxysomal CcaA. This could potentially lead to cytosolic dehydration of and suboptimal Ci-dependent O2 evolution, as has been observed in mutants expressing human CA in the cytosol (21, 37). Also noteworthy, however, is the potential for introduction of CcmM forms with a large fusion tag to disrupt the usual packing of carboxysomal proteins, resulting in oversized carboxysomes. Schwarz et al. (49), for example, reported larger than usual, nonfunctional carboxysomes in the EK6 mutant of Synechococcus PCC7942, which has a C-terminal extension on RbcS. A further, more detailed analysis of carboxysome sizes from WT, ccmM expressing untagged CcmM, and chimeric CcmM mutants is required in order to determine first if there is a statistical variation in carboxysome size and, second, if the N- or C-terminal exte
TOP
ABSTRACT
INTRODUCTION
