It has been suggested that ethanol metabolism in the strict anaerobe Clostridium kluyveri occurs within a metabolosome, a subcellular proteinaceous bacterial microcompartment. Two bacterial microcompartment shell proteins [EtuA (ethanol utilization shell protein A) and EtuB] are found encoded on the genome clustered with the genes for ethanol utilization. The function of the bacterial microcompartment is to facilitate fermentation by sequestering the enzymes, substrates and intermediates. Recent structural studies of bacterial microcompartment proteins have revealed both hexamers and pentamers that assemble to generate the pseudo-icosahedral bacterial microcompartment shell. Some of these shell proteins have pores on their symmetry axes. Here we report the structure of the trimeric bacterial microcompartment protein EtuB, which has a tandem structural repeat within the subunit and pseudo-hexagonal symmetry. The pores in the EtuB trimer are within the subunits rather than between symmetry related subunits. We suggest that the evolutionary advantage of this is that it releases the pore from the rotational symmetry constraint allowing more precise control of the fluxes of asymmetric molecules, such as ethanol, across the pore. We also model EtuA and demonstrate that the two proteins have the potential to interact to generate the casing for a metabolosome.
- bacterial microcompartment
- Clostridium kluyveri
- ethanol utilization shell protein B (EtuB)
- protein sheet
- shell protein
Bacterial microcompartments are prokaryotic organelles made up of a protein shell that encapsulates a metabolic process [1–5]. These polyhedral structures are typically 100–200 nm in diameter and represent novel bioreactors that are dedicated to a specific cellular pathway. The best characterized of the bacterial microcompartments is the carboxysome, which is a structure associated with carbon fixation that is found in cyanobacteria and many chemotrophic bacteria [6–9]. Here, polyhedral shell proteins encase the enzymes carbonic anhydrase and RuBisCo (ribulose-1,5-bisphosphate carboxylase/oxygenase). The role of the microcompartment is thought to help concentrate carbon dioxide to overcome the inefficiency of RuBisCo, which is the rate-limiting step in the Calvin cycle [10–14].
Carboxysomes are icosahedral structures with an outer shell that contains somewhere around 10000 protein subunits [3,5,9]. In the last few years, progress has been made on the structural biology of these interlocking shell proteins [4,7,9,15]. In the carboxysome, there appear to be approximately five proteins that make up the outer shell. The proteins have a recognizable sequence that is referred to as the BMC (bacterial microcompartment) motif. Most of the individual shell proteins form cyclical hexamers, which further assemble in a side-by-side fashion to generate large protein sheets. This probably forms the facet of the structure [7,9,15]. Some of the shell proteins appear to form a pentameric structure, and these have been proposed to act as the vertices of the icosahedral shell [5,9]. Within some of the hexameric and pentameric protein assemblies, a central pore is observed and this is thought to serve as the major route for the diffusion of substrates and products [4,7,9,15]. The proteins have a very distinctive surface charge distribution, with many positively charged amino acid residues surrounding the central pore, and these again are thought to assist in the diffusion of substrates and products.
Sequencing studies revealed the presence of genes encoding shell proteins in operons associated with metabolic processes other than carbon fixation. Proteins with BMC motifs were found encoded in the operons associated with propane-1,2-diol metabolism and ethanolamine utilization in Salmonella enterica [16–18]. Indeed, growth of S. enterica on media containing either propane-1,2-diol or ethanolamine as an energy source resulted in the appearance of microcompartments when the bacteria were analysed by electron microscopy after thin sectioning . Both these metabolic processes require adenosylcobalamin (coenzyme B12)-dependent enzymes and both proceed via aldehyde intermediates. It has been suggested that the main purpose of the microcompartment here is to reduce the toxicity of the aldehyde intermediate by sequestering the compound within the confines of the organelle , although it has also been proposed that the organelle could conserve the volatility of the intermediate . The aldehyde is then disproportionated by alcohol and aldehyde dehydrogenases within the structure [1,3,21].
The microcompartment associated with propane-1,2-diol metabolism has been characterized more than its ethanolamine orthologue. The actual physical structure is not so well defined as the clear icosahedral shape observed with the carboxysome. However, the operon encodes 21 gene products, and nine of these are thought to constitute shell proteins [1,3,16,22]. This larger number of shell proteins may give rise to a more complex rounded structure. In contrast with the carboxysome, only one (non-essential) shell protein has had its structure determined, PduU, which reveals that it adopts a similar fold to the other BMC-containing proteins and forms a hexamer . The microcompartment also contains more metabolic enzymes than are found in the carboxysome as well as a number of re-activation factors for the adenosylcobalamin coenzyme [3,22,24–26]. For this reason, the structure has been referred to as a metabolosome [27,28]. Significantly, it has been shown that the propanediol utilization operon can be genetically transferred between bacterial species to allow production of a functional structure in the new host . This has significant potential for the design of bespoke bioreactors and metabolic engineering.
More recently, genome sequencing projects have revealed that bacterial microcompartments are more widespread than originally thought [3,4,29]. Indeed, it has been estimated that approx. 25% of prokaryotes have the genetic capacity to form such structures. Significantly, though, only the carboxysome and the ethanolamine and propanediol metabolosomes have been physically characterized and the remaining systems remain hypothetical. In the present paper we report on a potentially novel bacterial microcompartment that is found in the anaerobic organism Clostridium kluyveri. This bacterium is able to grow on ethanol and acetate as sole carbon sources, and it has been proposed that the organism encloses ethanol and acetaldehyde dehydrogenases within a microcompartment . Indeed, a serendipitous electron micrograph of a thin section of C. kluyveri grown on ethanol and acetate reveals the presence of such a polyhedral structure . The genes for ethanol metabolism are encoded within a potential operon, which includes genes for two ethanol dehydrogenases, three aldehyde dehydrogenases and two putative shell proteins (Figure 1A). The enzymes are known to associate to allow the oxidation of ethanol to acetyl-CoA (Figure 1B) . The two shell protein genes have been termed etuA (ethanol utilization shell protein A) and etuB. EtuA contains 92 residues and shares 60% sequence identity with the carboxysome protein CsoS1A from Halothiobacillus neapolitanus (2ewh) and PduA from the propanediol metabolosome. These proteins have a single BMC motif and, in the case of CsoS1A, form a hexamer . The second C. kluyveri protein, EtuB, is unusual in that it is much larger (304 residues), has a tandem sequence repeat of the BMC motif and shares approx. 60% similarity with PduB from the propanediol metabolosome. In the present paper we report on the cloning of etuA and etuB, the effect of their recombinant production in Escherichia coli and the structure determination of the more unusual shell protein, EtuB.
Cloning and protein production
The etuA gene was amplified by PCR from C. kluyveri genomic DNA using primers CK1 (cttcatatgggacaggaagcattagg) and CK2 (cttggatccgctagcttagccaatgttaggc) and cloned into the NdeI and BamHI site of pET3a and pET14b (the NdeT and BamHİ restriction sites within the primers are underlined). The etuB gene was amplified by PCR from C. kluyveri genomic DNA using primers CK3 (cttcatatgaataatgagttaatcg) and CK4 (cttactagtttaaaaagtaggtttagac). The PCR product was cloned into pET14b into the NdeI and SpeI site. This cloning strategy allowed the encoded proteins to be overproduced with N-terminal hexahistidine tags.
The encoded proteins were overproduced in E. coli BL21(DE3)plysS cells with an N-terminal hexahistidine tag. Cells were grown in LB (Luria–Bertani) broth containing 100 mg/l ampicillin and 34 mg/l chloramphenicol with aeration at 37 °C. Upon reaching a D600 of 0.6, the protein overproduction was induced with IPTG (isopropyl β-D-thiogalactoside; 0.4 mM), and growth was continued overnight at 16 °C. The cells were harvested by centrifugation (15 min, 4000 g) and resuspended in 15 ml of binding buffer (0.5 M NaCl, 5 mM imidazole and 20 mM Tris/HCl, pH 8.0). Cell lysis was achieved by sonication using a Sonics Vibracell Ultrasonic processor with an output of 30 W for four 30 s bursts interspersed with equal periods of cooling. The recombinant protein was purified using immobilized metal affinity chromatography. The supernatant was applied to a nickel-charged Sepharose column. Unbound protein was washed off with binding buffer (0.5 M NaCl, 5 mM imidazole and Tris/HCl, pH 8.0), wash buffer I (0.5 M NaCl, 50 mM imidazole and Tris/HCl, pH 8.0) and wash buffer II (0.5 M NaCl, 100 mM imidazole and Tris/HCl, pH 8.0). Proteins were eluted with buffer containing 0.5 M NaCl, 400 mM imidazole and Tris/HCl, pH 8.0 and finally buffer exchanged into buffer containing 50 mM Tris/HCl, pH 8.0, and 100 mM NaCl.
Prior to crystallization, EtuB was further purified on a size-exclusion column (Superdex 200 Global 10/30) which was connected to an Akta® FPLC chromatography system. Protein was eluted with a flow rate of 0.5 ml/min in 50 mM Tris/HCl buffer, pH 8.0, containing 100 mM NaCl.
TEM (transmission electron microscopy): preparation of samples for thin sectioning
Dehydration and embedding
Samples of E. coli strains overproducing EtuA, EtuB or PduB were prepared for thin sectioning to visualize the interior of the cells. The E. coli strains harbouring the appropriate plasmids were grown in 20 ml cell cultures at 37 °C to a D600 of 1.5. The protein production was induced with 400 mM IPTG at 18 °C overnight. The cells were harvested by centrifugation for 10 min at 2700 g and fixed for 2 h in 5 ml of 2.5% glutaraldehyde in PBS. Subsequently, the cells were pelleted and washed twice with PBS to remove traces of the fixing solution. The cells were stained for 2 h in 1% osmium tetroxide and then washed twice with PBS before dehydration. This was accomplished by subjecting the samples to a solvent gradient of: 60% IMS (industrial methylated spirit) overnight, 90% IMS for 15 min and 100% IMS for 15 min. The cells were then washed in 100% ethanol for 2 h, a procedure that was repeated a further two times. The cells were embedded by incubation in 30% LV resin (low viscosity resin) in ethanol overnight followed by three LV resin changes with 100% resin for 2 h. This produced blocks with medium hardness. The samples were placed in 0.5 ml BEEM® capsules, centrifuged for 5 min at 4000 g to concentrate the cells to the tip and incubated at 60 °C overnight to polymerize.
Sectioning and visualization of samples
Sections 80 nm thick were cut with an Ultra 450 MX 3768 diamond knife and placed on 300-mesh copper grids. The samples were stained with 4.5% uranyl acetate in 1% acetic acid for 30 min and with Reynolds lead citrate for 10 min at room temperature (21 °C). Sections were observed and photographed with a JEOL-1230 transmission electron microscope.
Crystallization, preparation of heavy atom derivatives and data collection
EtuB crystals were grown using the hanging drop method using 4 M sodium formate reservoirs and a protein concentration of 8 mg/ml. The 10 μl drops comprised equal volumes of protein solution and reservoir. Single crystals were harvested in litholoops, transferred through reservoir supplemented with 10% PEG200 [poly(ethylene glycol) 200] as cryoprotectant, and stored in liquid nitrogen prior to data collection. Five heavy atom reagents were screened at two concentrations, the most successful of which was a 5 mM methylmercury chloride soak for 3 h. The heavy atom derivative was prepared because the diffraction from crystals of Se-Met (selenium-methionine) labelled EtuB was too poor to enable protein phases to be calculated from anomalous differences. Data were collected at 100 K using station ID29 at the European Synchrotron Radiation Facility and IO4 at Diamond Light Source.
Structure determination and analysis
EtuB crystallizes in space group I213 with a=174.3 Å (1 Å=0.1 nm) with a single molecule in the asymmetric unit and a solvent content of 80%. Native data were collected to 3.0 Å and protein phases calculated to 3.3 Å using the monomethyl mercury derivative. The CCP4 program suite was used for structure solution . MOSFLM  was used for data processing and SCALA  for scaling. Two mercury sites were found using SHELX , refined using MLPHARE , and the phases improved and extended using DM . ARP/wARP  was used for automated model building with some rebuilding completed using COOT  and REFMAC . The final model comprises residues 75–304 and one water molecule. Data collection statistics and the quality of the final model are presented in Table 1. Comparative modelling used MODELLER [41,42] and structural comparisons made use of DALI .
Cloning of etuA and etuB and shell protein production
The genes corresponding to the putative C. kluyveri ethanol metabolosome shell proteins (etuA, etuB; database entry numbers CKL_1072 and CKL_1073 respectively) were amplified from C. kluyveri genomic DNA by PCR using primers that allowed the amplified fragments to be cloned into pET14b and pET3a vectors to permit the production of the shell proteins with and without an N-terminal His-tag. When transformed into E. coli BL21(DE3)plysS cells, the resulting strains harbouring etuA were found to overproduce the protein, but the protein was found largely in the insoluble fraction upon cell lysis. EtuA could be solubilized with chaotrophic agents such as urea, but the protein precipitated as the salt was removed. In contrast, cells harbouring EtuB produced large quantities of soluble protein, which was easily purified by immobilized metal affinity chromatography using an Ni2+-column.
Formation of long axial filaments in E. coli by EtuA production
Analysis of thin sections of the E. coli strain containing etuA in pET3a by TEM revealed some interesting features. Cells grown to stationary phase revealed the presence of ordered long axial filaments (Figure 2A). These filaments appear to interfere with cell division, apparently preventing septation, and also take up a significant amount of the cell cytoplasm. Similar, though smaller, structures have previously been reported with the overproduction of PduA from the propanediol metabolosome [28,44]. It would appear that EtuA has the ability to self assemble into large macromolecular structures. Overproduction of EtuB did not produce any internal structures when cells were analysed by TEM (Figure 2B), although overproduction of the orthologous protein from the pdu operon (PduB) did reveal the presence of protein filaments wound around the inside of the cell membrane (Figure 2C)
Purification and crystallisation of the Etu shell proteins
Attempts were made made to purify EtuA, but the protein was found to be sparingly soluble. However, as EtuA has 60% sequence identity with the carboxysome protein CsoS1A from H. neapolitanus (2ewh), whose structure has previously been determined , a model of EtuA (Figure 3A) can be readily built by comparative modelling. This suggests that EtuA is likely to form a hexamer. In contrast with EtuA, EtuB was much more soluble and could be purified by metal affinity and size-exclusion column chromatography. The protein was then used in crystallization trials using Molecular Dimensions crystal screens I and II. Several conditions were found that produced small crystals. One of these was optimized to give large crystals that were used for structural studies.
Structure of EtuB
The crystal structure of EtuB was determined using single isomorphous replacement with anomalous scattering and refined to a resolution of 3.0 Å, with final R and R-free values of 0.18 and 0.20 (see the Experimental section). The quality of the final model is good as judged by the stereochemistry and the validation of results. The EtuB molecular trimer sits on the crystallographic three-fold axis of the I213 cell so that there is a single molecule in the asymmetric unit (Figure 3B). The close and extensive association of three subunits around the crystallographic three-fold axis reveals that this is the biologically authentic oligomer (Figure 4A).
The EtuB trimer has a concave and a convex face (Figures 4A and 4C respectively); both faces are characterized by significant negative potential. The convex face has positive potential close to the three-fold axis and on the periphery. The concave face is punctuated by a number of small positive patches. Both the pseudo-hexameric structure and the three-fold symmetry of the EtuB trimer manifest in the electrostatic potential are clear in Figure 3(B). In contrast with the acidic faces, the edges of the EtuB trimer, which are anticipated to be involved in interactions forming the shell of the microcompartment (Figure 4B), are more neutral in nature.
The total solvent-accessible surface area which is buried on forming the trimer is 7850 Å2. Surprisingly, the atomic structure of EtuB lacks the first 74 residues, but is otherwise complete. To test the possibility that the N-terminal region of the EtuB structure might not comply with the 3-fold symmetry of the C-terminal domain, the diffraction data were processed in the lower symmetry sub-group I212121 with a trimer, rather than a single subunit in the asymmetric unit. In the resulting electron density map, there was no evidence for the N-terminal 74 residues, even in electron density maps calculated at low resolution. It is likely that the first 74 residues are cleaved during crystallization. SDS/PAGE, Western and MALDI (matrix-assisted laser-desorption ionization) MS analysis of solubilized protein crystals all support this conclusion (results not shown). The EtuB sequence is closely similar, over 50% sequence identity, to that of the Citrobacter freundii PduB (Figure 5A). The major area of homology aligns between positions 75 and 304 of EtuB i.e. from where EtuB is truncated during the crystallization process. The N-terminal regions of EtuB (1–74) and PduB (1–43) share a much lower level of similarity. In both S. enterica and C. freundii, PduB is produced in two forms, termed PbuB and PduB′, which result from the presence of two translation start sites on the mRNA, giving rise to two protein species that differ in the nature of the first 37 amino acid residues [22,28]. It is not known if EtuB is also produced in more than one form. Nonetheless, the high degree of sequence similarity between EtuB and PduB suggest that they have a very similar structure, a conservation that includes many of the residues that are found around the three pores in the complex. Conversely, it is not known if the first 70 residues of EtuB have any functional significance. In EtuB, the presence of the first 70 amino acids appears to make the protein more soluble in comparison with PduB. Indeed, PduB appears to form stable filaments within E. coli that form an inner insulating layer (Figure 2C). The filaments are possibly formed through a helical arrangement of trimers within the fibre.
A tandem repeat of the BMC fold
The EtuB subunit is wedge shaped and comprises a four-layer βαβα sandwich (Figure 5B). There is a tandem repeat of the BMC motif within the single polypeptide chain. The domain architecture is most closely similar to that of PduU  (Figure 5C) with an rmsd (root mean square deviation) of 2.2 Å over 89 equivalenced Cα atoms with sequence identity of 19%; DALI server. PduU is described as a circularly permuted BMC domain with the N-terminal region contributing the secondary structure elements β2, and the short first α-helix typically contributed by the C-terminal region. The principal elements of the BMC fold are a four-stranded antiparallel β-sheet with two α-helices on one side. PduU has an additional short β-strand, β6. Compared with PduU, the N-terminal polypeptide chain of the N-terminal BMC repeat of EtuB replaces the C-terminal polypeptide of PduU running antiparallel to β2 and β1 is missing (the strands are labelled according to the PduU assignment). The C-terminal polypeptide of the N-terminal BMC repeat of EtuB then extends as a bent α-helix and continues to form a short first β-strand (β6′) of the second BMC fold (Figure 5B). The topology of this second fold is the same as the first, the C-terminal polypeptide folds down and across the surface of the second BMC repeat (Figure 5B).
Superimposing the two BMC folds of EtuB reveals they have 21% sequence identity over 94 aligned residues, which have an rmsd of Cα atoms of 2.7 Å (Figure 5D). The conserved β-core with strand-order 2,3,5,4 can be clearly seen. The greatest structural difference between the two BMC domains occurs in the loops between β4 and β5. There are good structural and functional reasons for the differences between these loops. In the N-terminal BMC domain, the β4/β5 loop forms one side and contributes His156, Thr150 and main chain amides and carbonyls to the pore. In the C-terminal BMC domain, the β4/β5 loop packs with the equivalent loops from the two other subunits close to the molecular 3-fold axis. An obvious residue from this loop involved in packing close to the rotation axis is Tyr259. The duplication of the BMC motif facilitates the specialized roles of these two loops.
Assembly into an icosahedral shell
The EtuB trimer has the appearance of a hexameric disc of size 42 Å with the subunits packing tightly around the 3-fold axis (Figures 3B and 4). The N- and C-termini are on the convex side of the pseudo-hexameric disc (Figures 4B and 4C) and the other side is concave (Figure 4A). Close to the 3-fold axis, the disc is approx. 25 Å thick, increasing to 35 Å towards the edge of the disc. The packing in this crystal form gives no indication of how the hexamers may pack to form a flat sheet of molecules required to form an icosahedral face. There is also no indication as to which face of the oligomer is inside, which out, or indeed if there is a preferred inside or outside face. The sloped packing surfaces of the oligomer (Figure 4B) suggests that to form a flat strip these molecules would have to pack alternatively face in followed by face out. These strips could then be staggered to form a sheet. The missing N-terminal 74 residues might also play a part in the assembly of the pseudo-hexamers, either by binding them together or by filling the wedge between oligomers when packing into a sheet in which the oligomers all have the same orientation. The N-terminus is located close to the edge of the trimer, such that it could have a role in an embrace between adjacent oligomers. It is correctly positioned such that it could fill the wedge left when the oligomers are packed in a consistent orientation. If this were the case, then perhaps it is more likely that the convex side of the oligomer is in. It appears that the EtuB trimer can be accommodated within a sheet of EtuA hexamers and this may be a plausible model for a facet (see the Discussion).
As mentioned above, a structure of EtuA can be readily built by comparative modelling (Figure 3A). On the basis of this model, EtuA is expected to form a hexamer with the packing of subunits as seen in CsoS1A. The β4/β5 loop of EtuA occurs close to the symmetry axis and, although the model suggests this loop may pack close to the symmetry axis, it is equally probable that this flexible four-residue β-hairpin (Ile-Gly-Ser-Gly) packs alternatively, creating a pore close to the symmetry axis. Below the β4/β5 loop there is a ring of lysines (Lys37), resulting from the 6-fold symmetry axis, that might contribute to the selectivity of the pore. It is notable that in the C. kluyveri ethanol utilization gene cluster there is no gene encoding a product that could form a pentamer. Pentamers would occupy each of the vertices of a regular icosahedral shell. There is, however, a candidate for the pentamer elsewhere in the C. kluyveri genome associated with the apparently defunct glycerol utilization microcompartment (database entry number CKL_0849) . This protein has 52% sequence identity with the carboxysome pentamer CcmL from Synechocystis 6893 (2qw7) . More work on isolated microcompartments from C. kluyveri is needed to determine their precise composition so that models of the pseudo-icosahedral shell can be constructed.
The pores in the shell proteins solved to date are located around the 5-fold or 6-fold rotation axis that relates subunits. EtuB has no such pore, the 12-residue β4/β5 loop and β4′/β5′ loops are relatively complex in structure, certainly compared with the β-hairpin in EtuA, and pack tightly around the three-fold axis. Arg148 and Tyr259 are two residues that pack around the symmetry axis of the EtuB trimer. EtuB does, however, have a pore within the subunit, facilitated by the gene duplication that has given rise to the tandem repeat of EtuB. No subunit pore has previously been observed in any BMC shell protein characterized structurally to date. Three histidine residues (at positions 156, 195 and 224) and two glutamate residues (at positions 197 and 262) line the pore (Figure 6A) and four out of these five residues are conserved in the PduB sequence, highlighting their significance (Figure 5A). Additionally some hydrophobic residues line the pore, including Phe212, Val256 and Pro124. The ethanol utilization operon of C. kluyveri contains two aldehyde dehydrogenases and three alcohol dehydrogenase genes (Figure 1A). The size and characteristics of the subunit pore are consistent with ethanol diffusion into the BMC. The pore in the crystal structure has a single water molecule trapped within it (Figure 6B). Clearly, the tandem duplication affords the possibility of evolving an asymmetric pore, potentially one with greater selectivity than one that must conform to strict rotational symmetry.
Shell proteins for a potential metabolosome within C. kluyveri
It has been proposed that the ethanol utilization enzymes in C. kluyveri are contained within a BMC. Indeed, the genes for the ethanol and acetaldehyde dehydrogenases are clustered on the genome together with two genes for potential shell proteins of such a proteinaceous organelle. However, although the formation of such a metabolosome is attractive, hard experimental evidence for its presence has not yet been produced and its existence cannot therefore be assumed. Nonetheless, in this paper we report the structure of one of these two shell proteins, EtuB, and model the other, EtuA.
Bacterial shell proteins characterized to date form hexamers and pentamers, with the single pore per oligomer centred on the symmetry axis [4,5]. The EtuB subunit has a domain duplication and the EtuB trimer has pseudohexameric appearance due to the structural similarity of the two BMC domains present in the subunit. The core structure of the domain is restrained to the ancestral hexameric template essential for forming a facet of the microcompartment. In fact, the conservation of the pseudohexameric structure and of residues that may be key in forming contacts with neighbouring molecules provides support for the occurrence of EtuB in the C. kluyveri microcompartment. Both pseudohexamer edges have the strongly conserved lysine residue at position 23, characteristic of the canonical BMC domain (see Pfam family PF00936), which corresponds to Lys135 and Lys240 in EtuB. This lysine is found at the 2-fold axis between adjacent hexamers in the CcmK1, CcmK2 and CsoS1A layers . The lysine makes a hydrogen bond to the main chain carbonyl of the lysine in the opposite hexamer and to the side chain of aspartate within the same subunit (position 19 in the canonical BMC domain). This aspartate is also conserved in each BMC domain of EtuB (Asp131 and Asp237). The conservation of these surface residues further supports the view that EtuB forms a part of the microcompartment facet (Figure 7).
Consequences of the novel features of the EtuB pore
More peripheral structural elements are less restrained by the symmetry of the ancestral protein, particularly those structural elements that do not affect the surfaces involved in packing to form the microcompartment facet. In EtuB, the internal β4/β5 loop adopts different conformations in each of the two copies of the BMC domain. These two loops contribute residues that pack around the symmetry axis as well as residues that contribute to the subunit pore. The subunit pore is a novel feature of EtuB, which may have two advantages. The first is that it is present at three times the concentration of pores that are coincident with the 6-fold symmetry axis; this increase in concentration (or density of pores) may be important if the rate of exchange of substrate or product molecules across the microcompartment shell is rate-limiting, but only if EtuB is abundantly present in the shell. The second advantage is that the architecture of the pore is released from the symmetry constraint which may allow it greater potential to become more selective. The pore is not large but may allow the passage of ethanol.
The asymmetry of the EtuB oligomer, compared with the carboxysome hexamer, might also account for the more pleiomorphic shape of the C. kluyveri shell compared with that of the carboxysome. In reality, little is known about the proposed ethanol utilization bacterial microcompartment. Electron-microscopic evidence is suggestive of such bodies , and biochemical evidence points to a close association between the alcohol and aldehyde deydrogenases, since the various enzymes and isoenzymes are isolated in a large macromolecular complex [31,45] as would be expected for a metabolosome. This, coupled with the organization of the respective genes into a cluster with the genes for the shell proteins , provides compelling evidence that this ethanol utilization system is incorporated into a bacterial microcompartment. With this in mind, the structure of the microcompartment must also reflect the biochemical role that it plays within the metabolism of the cell. The structure must be able to allow ethanol into the macromolecular assembly together with coenzyme A (CoASH) and NAD+. The products of the reaction have to be able to leave the complex, so acetyl-CoA and NADH have to be able to exit the structure. Access to and entry from the complex is likely to be via the pores in the shell proteins. The pores in EtuB are comparatively small and polar, and thus should be able to allow the passage of ethanol. The pores are too small to allow larger more complex molecules through, such as the coenzymes or their derivatives. The structure of EtuB is likely to be very similar to PduB, which is involved in propanediol metabolism. Here the pores in PduB could also allow the substrate propane-1,2-diol into the bacterial microcompartment. In both cases, however, the other substrates/coenzymes/cofactors must access by another shell protein, or via specific channels formed between different shell unit components. In the propanediol metabolosome, there are seven different shell protein subunits, providing a high level of complexity. In the ethanol utilization system, there appear to be only two components. The EtuA shell has been modelled and does not appear to have a central pore (Figure 3A). A conformational change has to be invoked to permit the passage of larger molecules. In the large macromolecular assembly formed between EtuA and EtuB (Figure 7), a certain amount of breathing could be envisaged that could also permit the transfer of metabolites from outside to inside and vice versa. Much more detail on the composition of the ethanol utilization bacterial microcompartment is required as well as detail on the interaction between EtuA and EtuB in order to advance our current understanding of these structures.
Recent trimeric shell protein structures
Remarkably, a paper appeared at the time of writing the present paper describing the structure of a trimeric carboxysome shell protein CsoS1D . Both the BMC architecture and the tandem fusion of BMC motifs appear to be closely similar to EtuB (CsoS1D co-ordinates are on hold at PDB). The major difference is that CsoS1D has a large pore, apparently gated, on the three-fold symmetry axis, as opposed to the small subunit pore of EtuB. Duplication of the BMC motif has therefore allowed the evolution of at least two types of novel pore. The crystallization of another shell protein, EutL, which also contains a duplicated BMC domain has also been reported recently, and this protein displays many structural similarities to EutB, including the presence of three pores . Together, all of this recent information will greatly assist in providing further molecular detail on the structure and function of these remarkable proteinaceous bacterial organelles and may allow, in due course, the construction of bespoke bioreactors.
Dana Heldt cloned etuA and etuB, purified EtuB and performed the initial crystallization. Stefanie Frank analysed the cells producing recombinant EtuA and EtuB after thin sectioning and electron microscopy. Arefeh Seyedarabi helped with the X-ray diffraction data acquisition and subsequent data analysis. Dimitrios Ladikis optimized the crystallization and helped with data collection. Joshua Parsons helped with the thin sectioning and electron microscopy. Martin Warren devised the project and helped write the paper. Richard Pickersgill solved the structure of EtuB and helped write the paper.
This work was supported by the Biotechnology and Biological Sciences Research Council [grant number BB/E01563] and the Higher Education Funding Council for England.
Synchrotron data were collected at ESRF (Grenoble) and Diamond (Oxford).
Co-ordinates and structure factor amplitudes for EtuB have been deposited in the Protein Data Bank with the accession code 3IO0.
Abbreviations: BMC, bacterial microcompartment; Etu, ethanol utilization shell protein; IMS, industrial methylated spirit; IPTG, isopropyl β-D-thiogalactoside; LV resin, low viscosity resin; rmsd, root mean square deviation; RuBisCo, ribulose-1,5-bisphosphate carboxylase/oxygenase; TEM, transmission electron microscopy
- © The Authors Journal compilation © 2009 Biochemical Society