Introduction

Bacteria use many strategies to optimize metabolic processes. Polyhedral organelles are perhaps the most sophisticated, and they appear to be emerging as a general mechanism of metabolic optimization. This review first describes the various types of polyhedral organelles that have been identified then summarizes current knowledge about their structure, mechanism, assembly, turnover, evolution, and unique biochemistry.

Carboxysomes

Carboxysomes were the first bacterial polyhedral organelle identified. They were discovered by electron microscopy as cytoplasmic bodies resembling bacteriophage (Fig. 1a) (Gantt and Conti 1969; Shively et al. 1970). They have been observed in cyanobacteria and many chemoautotrophic bacteria, but they have not been found in eukaryotes (Price et al. 1998; Shively et al. 1998b; Kaplan and Reinhold 1999; Cannon et al. 2001). Carboxysomes are polyhedral in shape and are usually about 100–150 nm in cross section, which gives them a volume 1,000× greater than a ribosome. They have a 3- to 4-nm protein shell which is composed of six to ten different polypeptides. The lumen of the carboxysome is filled with ribulose bisphosphate carboxylase/oxygenase (RuBisCO) (Shively et al. 1973), which catalyzes the CO2 fixation step of the Calvin cycle, the carboxylation of ribulose bisphosphate to form two molecules of 3-phosphoglycerate. Also associated with the carboxysome is carbonic anhydrase (CA), which converts \(HCO_{3} ^{ - }\) to CO2, the substrate for RuBisCO (\(HCO_{3} ^{ - }\) is not used by RuBisCO) (Price et al. 1992; Yu et al. 1992; So et al. 2004). As yet, there are no confirmed reports that lipids, DNA, or RNA are associated with carboxysomes. Available information indicates carboxysomes are composed completely of protein and glycoprotein subunits. The proposed function of carboxysomes is to improve carbon fixation via the Calvin cycle by enhancing the activity of RuBisCO at low CO2 concentrations.

Fig. 1
figure 1

Electron micrograph of polyhedral organelles: a the carboxysomes of H. neapolitanus; b the organelles formed during growth of Salmonella enterica on 1,2-propanediol. Triangles point to the organelles

The structure of carboxysomes varies somewhat among the different bacterial species that produce these structures. Based on their constituent proteins, carboxysomes can be divided into two subtypes (α and β) (Badger et al. 2002; Cannon et al. 2002). The best-studied α-type carboxysome is that of the facultative chemoautotroph, Halothiobacillus neapolitanus. This organelle was purified, and a number of its protein components were identified (Cannon and Shively 1983; Holthuijzen et al. 1986; English et al. 1994). It is composed of at least eight polypeptides, including the large and small subunits of RuBisCO (CbbL and CbbS), and six shell proteins (CsoS1A, CsoS1B, CsoS1C, CsoS2A, CsoS2B, and CsoS3). The shell proteins CsoS2A and CsoS2B are encoded by the same gene but differ in their degree of glycosylation (Baker et al. 1999). CsoS1A–C are very closely related in amino acid sequence and are encoded by three paralogs. Currently, CsoS1- and Csos2-type shell proteins are only known to have structural roles. However, the CsoS3 protein was recently shown to be a novel class of carbonic anhydrase (ɛ-class) that is an integral shell component (So et al. 2004).

The genes encoding the known carboxysome polypeptides of H. neapolitanus are found at a single genetic locus having the following structure: cbbL, cbbS, csoS2, csoS3, orfA, orfB, csoS1C, csoS1A, and csoS1B (Cannon et al. 2002). The orfA and orfB sequences are required for carboxysome function, but their role is unknown. Their encoded proteins were not identified as components of purified organelles, so they are either absent from the assembled structures or present in low amounts.

The second known type of carboxysome (type β) is found in the β subdivision of the cyanobacteria. In cyanobacteria, carboxysome proteins are designated CCM for their role in a carbon dioxide concentrating mechanism. Studies of partially purified β-type carboxysomes as well as genetic and bioinformatic analyses indicate that their protein constituents commonly include RbcL and RbcS, CcmK CcmO, CcmL, CcmM, and CcmN and CcaA (IcfA) (Kaplan and Reinhold 1999; Badger et al. 2002; Cannon et al. 2002; Badger and Price 2003). RbcL and RbcS are the large and small subunits of RuBisCO. CcaA (IcfA) is a putative carboxysome-associated carbonic anhydrase (Fukuzawa et al. 1992; Yu et al. 1992). CcmK and CcmO have homology to CsoS1, which is a known carboxysome shell protein (English et al. 1994). CcmL is related in sequence to the OrfA and OrfB proteins, which are of unknown function. The N-terminal portions of CcmM and CcmN have homology to carbonic anhydrases, and their C-terminal regions have similarity to RuBisCO or to several enzymes and cell surface proteins, respectively. The functions of CcmM and CcmN are unknown. They may be catalytically active. Alternatively, the observed sequence similarities may provide domains that allow specific protein–protein interactions.

The distribution and composition of carboxysomes has been investigated by genomic and phylogenetic analyses (Badger et al. 2002; Cannon et al. 2002; Badger and Price 2003). Type α carboxysomes are present in proteobacteria such as Halothiobacillus and in α-cyanobacteria (cyanobacteria the express Form 1A RuBisCO), while type β carboxysomes are found in β-cyanobacteria (cyanobacteria that produce Form 1B RuBisCO) (Badger et al. 2002; Cannon et al. 2002). Studies conducted thus far indicate that the csoS2 and csoS3 genes are specific to type α, while ccmM and ccmN are characteristic of type β (Badger et al. 2002; Cannon et al. 2002). Comparisons of the molecular phylogeny of carboxysomes and RuBisCO suggests that α and β carboxysomes may have coevolved with Form 1A and Form 1B RuBisCO, respectively (Badger et al. 2002; Badger and Price 2003).

The function of carboxysomes is to enhance autotrophic CO2 fixation at low CO2 levels. This role is supported by the findings that carboxysomes are induced by CO2 limitation (Beudeker et al. 1979; McKay et al. 1993), and that mutant strains of cyanobacteria and chemoautotrophs which are unable to form carboxysomes properly require high CO2 for autotrophic growth (the HCR phenotype) (Marcus et al. 1986; Friedberg et al. 1989; Price and Badger 1989b; Kaplan 1990; English et al. 1995). The rationale for the HCR phenotype is that the carboxysome is an essential part of a CCM (Berry et al. 1978; Badger et al. 1980; Price et al. 1998; Kaplan and Reinhold 1999; Badger and Price 2003). The initial steps of the CCM are transport processes that concentrate \(HCO_{3} ^{ - }\) within the cellular cytoplasm where equilibrium with CO2 is not reached due to a lack of CA. Subsequently, carboxysomal CA converts \(HCO_{3} ^{ - }\) to CO2 and releases it within the organelle which acts as a diffusion barrier. These processes elevate CO2 levels in the immediate vicinity of RuBisCO, increasing the rate of CO2 fixation and suppressing photorespiration (a nonproductive process in which O2 replaces CO2 as a substrate RuBisCO competitively inhibiting carboxylation). The association of CA with carboxysomes was predicted by quantitative models for CCM (Price et al. 1998; Kaplan and Reinhold 1999; Kaplan et al. 2001) and is supported by three lines of experimental evidence: (1) CA is associated with carboxysomes and, in some cases, is an integral component of the protein shell (Fukuzawa et al. 1992; So et al. 2004); (2) mutants defective in carboxysomal CA have an HCR phenotype (Yu et al. 1992; So et al. 2002); and (3) ectopic expression of CA in the cytoplasm Synechococcus leads to massive loss of CO2 and HCR phenotype (Price and Badger 1989a). Thus, the association of CA with the carboxysome and the importance of this association to the CCM seems well established. However, thus far, the means by which carboxysomes act as a barrier to CO2 diffusion is undetermined and is a very interesting question.

1,2-propanediol utilization (pdu) organelles

Carboxysomes were isolated in the early 1970s, and for many years, they were the only known polyhedral organelles. Recently, however, Salmonella enterica was found to form a polyhedral organelle during growth on 1,2-propanediol (1,2-PD) as a sole carbon and energy source (Shively et al. 1998a,b; Bobik et al. 1999). These organelles had been overlooked because they are specifically formed during growth on 1,2-PD but not during growth on standard laboratory media. Their discovery resulted from genetic studies which showed genes specifically involved in 1,2-PD utilization (pdu) encode homologs of carboxysome shell proteins (Chen et al. 1994). It is now known that the pdu organelles have components that are homologous and distantly related to carboxysome proteins but that their enzyme complement and physiological functions are distinctly different (Bobik et al. 1999; Havemann et al. 2002; Havemann and Bobik 2003; Leal et al. 2003; Johnson et al. 2004). Carboxysomes enhance autotrophic CO2 fixation, and the pdu organelles support the coenzyme B12-dependent degradation of 1,2-PD as a carbon and energy source.

A number bacterial genera degrade 1,2-PD as a sole carbon and energy source in a coenzyme B12-dependent manner. This processes appears to be particularly important to the enterica group since genes for 1,2-PD degradation are present in the genomes of Salmonella, Klebsiella, Shigella, Yersinia, and at least one species of Escherichia coli (E24377A). 1,2-PD may be especially important in anaerobic or microaerobic environments since it is a major product of the fermentation rhamnose and fucose, common sugars in plant cell walls, bacterial capsules, and glycoconjugates found on the surfaces of eukaryotic cells. In all genomes we have examined, the genes for 1,2-PD degradation are proximal to homologs of carboxysome shell genes.

Coenzyme B12-dependent diol dehydratase (DDH) catalyzes the first step of 1,2-PD, the conversion of 1,2-PD to propionaldehyde (Toraya et al. 1979; Obradors et al. 1988; Bobik et al. 1997). The propionaldehyde is then converted to 1-propanol and propionate, by alcohol dehydrogenase, coenzyme A (CoA)-dependent propionaldehyde dehydrogenase (PduP), phosphotransacylase, and propionate kinase (Toraya et al. 1979; Obradors et al. 1988). This pathway generates one adenosine triphosphate (ATP), an electron sink, and a 3-carbon intermediate (propionyl-CoA), which can feed into central metabolism via the methyl-citrate pathway (Horswill and Escalante-Semerena 1997).

Coenzyme B12 (adenosylcobalamin, AdoCbl) is required for DDH activity and, thus, essential for 1,2-PD degradation (Abeles and Lee 1961). The AdoCbl needed for 1,2-propanediol degradation can be obtained either by de novo synthesis or by the assimilation of exogenous cobalamin (Cbl). In S. enterica, de novo synthesis occurs only under strictly anaerobic conditions (Jeter et al. 1984). However, cyanocobalamin (vitamin B12, CN-Cbl) can be assimilated both aerobically and anaerobically (Jeter 1990). The conversion of CN-Cbl to AdoCbl requires several enzymatic steps: CN-Cbl is decyanated to OH-Cbl, reduced to cob(II)alamin, further reduced to cob(I)alamin, and lastly, adenosylated to AdoCbl (Brady et al. 1962; Vitols et al. 1965; Friedmann 1975; Huennekens et al. 1982).

Genetic analyses of 1,2-PD degradation by S. enterica identified a single contiguous cluster of genes specifically involved in 1,2-PD utilization (pdu) (Jeter 1990). DNA sequence analysis of this locus identified 23 putative genes: pocR, pduF, pduA, pduB, pduC, pduD, pduE, pduF, pduG, pduH, pduJ, pduK, pduL, pduM, pduN, pduO, pduP, pduQ, pduS, pduT, pduU, pduV, pduW, and pduX (Chen et al. 1994; Bobik et al. 19971999). Currently, the known pdu genes are assigned to eight categories (Bobik et al. 1999); the pduF gene encodes a 1,2-PD diffusion facilitator (Chen et al. 1994). The pocR gene encodes a positive transcriptional regulator of the araC family that coinduces both pdu genes and genes for the de novo synthesis of coenzyme B12 (Bobik et al. 1992; Rondon and Escalante-Semerena 1992). Four pdu genes have unknown functions (pduLMVX), and six encode the enzymes needed for the pathway of 1,2-PD degradation (pduCDEPQW). The pduCDE genes encode the large, medium, and small subunits of coenzyme B12-dependent DDH (Bobik et al. 1997). The pduP gene encodes PduP (Leal et al. 2003), and the pduW gene encodes propionate kinase (Palacios et al. 2003). The pdu locus also encodes a protein related to the DDH reactivation factor from Klebsiella (PduGH) (Bobik et al. 1999) as well as two genes that are used for the biosynthesis of coenzyme B12 from complex precursors (pduO and pduS) (Johnson et al. 20012004; Sampson et al. 2005). The most surprising finding to come out of sequence analyses of the pdu locus is that it encodes seven proteins related to those involved in the formation of carboxysomes (pduABJKNTU) (Bobik et al. 1999). The PduA and PduJ proteins are closely related to carboxysome shell proteins Csos1, CcmK, and CcmO (50–60% identity). The PduK and PduT proteins are 25–30% identical to CsoS1. The PduN protein is 46% identical to carboxysome protein CcmL from Synechococcus elongatus PCC 6301 and distantly related to the putative carboxysome proteins encoded by orfA and orfB of H. neapolitanus. In addition, position-specific iterated–basic local alignment search tool (PSI-BLAST) analyses show that the PduB and PduU proteins are distantly related to the carboxysome shell proteins CcmK and CsoS1.

Electron microscopy showed that S. enterica forms structures similar in size and shape to carboxysomes during growth on 1,2-PD but not during growth on other carbon sources (Fig. 1b) (Bobik et al. 1999; Havemann et al. 2002). Immunoelectron microscopy indicated that coenzyme B12-dependent DDH is a major component of the pdu organelles, and that the PduA protein is a shell component (Bobik et al. 1999; Havemann et al. 2002). Recently, the pdu organelles were purified and found to be composed of 15 different polypeptides. Fourteen of these are encoded in the pdu operon, but the gene for the 15th was not identified (Havemann and Bobik 2003). The components of the pdu polyhedra include PduP, the three subunits of DDH (PduCDE), a putative DDH reactivation factor (PduGH), a cob(I)alamin adenosyltransferase that catalyzes the last step of vitamin B12 assimilation (PduO), and seven polypeptides related to carboxysome proteins (PduABB′JKNTU). PduB′ is a truncated version of PduB lacking 37 N-terminal amino acids. The PduN protein (which is related to CcmL, OrfA, and OrfB) was not identified as a component of the purified pdu organelles.

Fig. 2
figure 2

Organelle-minus mutants show growth arrest on 1,2-PD minimal medium. ○, wild type; ▴, organelle-minus mutant. Growth was measured by determining the optical density of cultures at 600 nm

Genetic studies indicate that the pdu organelles function to minimize the harmful effects of a toxic intermediate of 1,2-PD degradation (propionaldehyde) (Havemann et al. 2002; Havemann and Bobik 2003; Leal et al. 2003). Mutants unable to form pdu organelles undergo a 20-h period of growth arrest during degradation of 1,2-PD, whereas wild-type S. enterica grows normally under similar conditions (Havemann et al. 2002) (Fig. 2). The length and severity of growth arrest increases at higher 1,2-PD concentrations, suggesting it results from the accumulation of a toxic metabolite derived from 1,2-PD (Havemann et al. 2002). The major metabolites of 1,2-PD degradation by S. enterica are 1-propanol, propionaldehyde, and propionate (unpublished results). Of these, propionaldehyde is known to be cytotoxic. Thus, it was proposed that the function of the pdu organelles is to minimize propionaldehyde toxicity during growth of S. enterica on 1,2-PD (Havemann et al. 2002).

Fig. 3
figure 3

Model for the pdu organelles. The dashed line represents the protein shell of the organelle. PduCDE, diol dehydratase; PduP, propionaldehyde dehydrogenase; PduO, adenosyltransferase; PduGH, diol dehydratase reactivating factor; PduQ, 1-propanol dehydrogenase; PduW, propionate kinase; Cbl, cobalamin; AdoCbl, coenzyme B12

Based on the findings outlined above, a model for the pdu organelles is proposed (Fig. 3). The dashed line indicates the shell of the organelle. It is composed of the PduABB′JKTU proteins, each of which has sequence similarity to carboxysome shell proteins (Bobik et al. 1999). 1,2-PD traverses the shell by an unknown mechanism and enters the lumen of the organelle, where it is converted to propionaldehyde and then to propionyl-CoA by the combined actions of DDH and PduP, both of which are major components of the pdu organelles (Havemann and Bobik 2003). Propionyl-CoA diffuses into the cytoplasm of the cell, where it is converted to propionate by phosphotransacylase (not yet identified) and propionate kinase (PduW). The proposed function of these organelles is to protect cytoplasmic components from propionaldehyde. To do so, they sequester propionaldehyde and regulate the activity of DDH (propionaldehyde producing). Also associated with the pdu organelles are the PduGH and PduO proteins (Havemann and Bobik 2003). These enzymes are needed to maintain the activity of DDH. Based on sequence similarity, the PduGH protein is proposed to be a DDH reactivation factor (Bobik et al. 1999). The PduO adenosyltransferase catalyzes the last step of the conversion of vitamin B12 to coenzyme B12, the required cofactor of DDH.

Polyhedral organelles involved in coenzyme B12-dependent ethanolamine degradation

Recent studies have shown that a polyhedral organelle is involved in ethanolamine utilization (eut) by S. enterica (Kofoid et al. 1999). The initial step of this process is catalyzed by coenzyme B12-dependent ethanolamine ammonia lyase, which converts ethanolamine to acetaldehyde (Chang and Chang 1975). Subsequently, acetaldehyde is converted to ethanol and acetate by a series of reactions analogous to those used for 1,2-PD degradation (Roof and Roth 1988). This suggests that the eut and pdu organelles might have the common function of protecting cells against aldehyde toxicity, which was first proposed when homologs of carboxysome shell proteins were found in the eut operon (Stojiljkovic et al. 1995). In addition, recent studies (which refer the eut organelles as metabolosomes) suggest that these structures function to concentrate both metabolic enzymes and their substrates to allow more efficient growth while minimizing acetaldehyde toxicity and maintaining coenzyme A balance (Brinsmade et al. 2005). Two key findings that support this proposal are that (1) the elimination of the eut organelles by deletion of multiple shell gene homologs prevents growth of S. enterica on ethanolamine and (2) elevated ectopic expression of two ethanolamine catabolic enzymes (ethanolamine ammonia lyase and acetaldehyde dehydrogenase) in the absence of the organelle is sufficient for growth on ethanolamine (Brinsmade et al. 2005). Alternatively, the primary function of the eut organelles might be to prevent carbon loss: acetaldehyde is volatile and passes relatively freely through lipid membranes (J.R. Roth and J.T. Penrod, personal communication). Overall, it is clear that the eut organelles (and the pdu organelles) are complex structures; hence, it would be reasonable that they enhance/optimize ethanolamine and 1,2-PD degradation by multiple mechanisms.

Distribution and functional diversity bacterial polyhedral organelles

Studies that demonstrated polyhedral organelles are involved in coenzyme B12-dependent ethanolamine, and 1,2-PD degradation raised questions about the distribution and functional diversity of related structures. Sequence analyses show that the genomes of over 40 different genera of bacteria contain homologs of carboxysome shell genes (unpublished results). Such homologs are not detected in the genomes of archaea or eukarya. Among the bacteria, shell gene homologs are often found proximal to associated enzymes. Hence, multiple shell gene homologs interspersed with (or adjacent to) putative enzymes serve as an organelle signature. Examination of these “genomic signatures” tentatively identifies seven functionally distinct organelles (unpublished results). These include carboxysomes, the pdu organelles and eut organelles (which are well established), and four putative organelles involved in unknown metabolic process. Each of the presumptive organelles contains shell gene homologs interspersed with a distinct enzyme complement. Among these, one is conserved in three genera of bacteria (Desulfitobacterium, Desulfovibrio, and Shewanella), but currently, the remaining three are only found in a single genome: Solibacter, Carboxydothermus, or Rhodopirellula. In each case, carboxysome shell protein homologs are found proximal to a different set of putative enzymes, tentatively suggesting distinct physiological functions. However, to our knowledge, no experimental evidence supports the production of polyhedral organelles in these latter four cases.

X-ray crystallography studies

Recently, the crystal structure of two carboxysome shell proteins (CcmK2 and CcmK4) was reported (Kerfeld et al. 2005). The CcmK2 and CcmK4 monomers are wedge-shaped and fit tightly together, forming a solid hexamer having a central pore. The pores of the CcmK2 and CcmK4 hexamers are 7 and 4 Å, respectively, and have a large net positive electrostatic potential. The sides of the hexamers are nearly flat and pack tightly into sheets that could serve as facets of the polyhedral shell. There are 4- to 6-Å-wide gaps between the hexamers that also feature conserved charged amino acids. Both the hexamer pores and the gaps between hexamers could potentially serve as conduits for metabolites, and it was proposed that these pores might selectively allow passage of negatively charged molecules such as the substrates and products of RuBisCO while restricting uncharged molecules such as CO2 and O2 (Kerfeld et al. 2005).

Molecular mechanisms

The molecular mechanism of the carboxysome is not fully understood. A simple mechanism would be a shell that enhances CO2 fixation by altering the activity or efficiency of RuBisCO. However, studies conducted thus far indicate that free RuBisCO and carboxysome-associated enzyme have similar kinetic properties with respect to both catalysis and activation (Cannon and Shively 1983; Holthuijzen et al. 1987). It has also been suggested that carboxysomes enhance CO2 fixation by minimizing inhibition of RuBisCO by molecular oxygen (photorespiration). This idea is supported by in vitro studies of free and carboxysome-associated RuBisCO (Marcus et al. 1992). However, it raises the question of how carboxysomes could be freely permeable to ribulose bisphosphate and 3-phosphoglycerate (the substrates and products of RuBisCO) (Satoh et al. 1997) and yet restrict the movement of O2 and CO2. Interestingly, recent structural studies (described above) suggest that needed selectivity could be mediated by a pore within the protein shell that has a positive electrostatic potential (Kerfeld et al. 2005). Furthermore, it was recently shown that CA is an integral component of some carboxysome shells, raising the possibility that it might act as a primitive transport system that translocates CO2 into the lumen of carboxysomes (So et al. 2004). A charged pore combined with a system for CO2 translocation provides a plausible mechanism for the selective concentration of CO2 within the carboxysome via the CCM.

The pdu organelles have also been proposed to protect lumen enzymes from inhibition by molecular oxygen (coenzyme B12-dependent DDH is inhibited by oxygen) (Chen et al. 1994). However, in the studies reported to date, DDH activity was not significantly impaired in organelle-minus mutants under aerobic growth conditions (Havemann et al. 2002). It has also been proposed that the pdu organelles function to sequester/concentrate propionaldehyde to reduce cytotoxicity (Havemann et al. 2002). The use of charged pores in the shell to obtain selectivity is a more complex problem in the case of the pdu organelles compared to carboxysomes. The substrate for DDH (1,2-PD) is uncharged, and the lumen enzymes of the pdu organelles require NAD, coenzyme B12, and coenzyme A as cofactors, which are large molecules in comparison to propionaldehyde. The use of charged pores in conjunction with added systems that move enzyme cofactors across the shell is a possibility. However, there are several reasonable alternatives. I will suggest one. The pdu organelles and the cell cytoplasm may act as a biphasic system, with aldehydes being more soluble inside the organelles than in the cell cytoplasm. This would effectively concentrate propionaldehyde within the organelles. In turn, this might enhance the PduP activity while inhibiting DDH via product inhibition. Overall, propionaldehyde production and consumption would be tightly coordinated.

Assembly

Currently, little is known about assembly of polyhedral organelles. Analogy with other multiprotein complexes tentatively suggests self-assembly guided by chaperones and assembly factors. The gene cluster required for formation of the eut organelles (the eut operon) encodes two proteins distantly related to the DnaK family of heat shock chaperonins (EutA and EutJ). The EutA protein is involved in ethanolamine ammonia lyase reactivation (Mori et al. 2004; Sheppard et al. 2004), but the role of the EutJ protein is unknown, making it a candidate organelle-assembly chaperone. Assembly factors for polyhedral organelles have not been identified. Some proteins required for proper carboxysome function OrfA and OrfB are either minor components of the purified particles or absent and, thus, may fit the definition of an assembly factor (a protein needed for assembly that is not part of the finished structure). With respect to organelle assembly, one might also imagine that the shell forms around the lumen enzymes which act as a scaffold. This may be the case for carboxysomes since studies indicate that the small subunit of RuBisCO is required for their assembly (Orus et al. 1995), and that replacement of the form I RuBisCO with the L2-type prevents carboxysome formation (Pierce et al. 1989). However, in the case of the pdu organelles, genetic deletion of each lumen enzyme individually does not preclude the formation of the shell (Havemann and Bobik 2003; Leal et al. 2003; Johnson et al. 2004). Hence, no single lumen enzyme of the pdu organelles is required as a scaffold for shell formation.

A particularly interesting aspect of organelle assembly which is currently poorly understood is the manner in which proteins are targeted to the lumen. Several proteins that reside in the lumen of polyhedral organelles have N- or C-terminal extensions by comparison to homologous proteins found in the cell cytoplasm. For example, two of the three subunits of B12-dependent DDH which are found in the lumen of the pdu organelles have short N-terminal extensions compared to their cytoplasmic homologs. Removal of these extensions dramatically increases the solubility of DDH but does not affect its catalytic activity appreciably (Tobimatsu et al. 2005). This raises the possibility that these N-terminal extensions could affect localization as was tentatively suggested (Tobimatsu et al. 2005). Presumably, this would occur via protein–protein interactions, but a targeting apparatus that directs proteins to the lumen of polyhedral organelles is not excluded.

Turnover

Little is known about turnover of polyhedral organelles. In the case of the pdu organelles, high induction of the pdu operon is necessary, but not sufficient, for organelle formation (Bobik et al. 1999). Pdu organelles are not formed unless 1,2-PD is present in growth media. This requirement could be direct or indirect. However, 1,2-PD also enhances organelle stability during purification (Havemann and Bobik 2003). Thus, the pdu organelles might dissociate into polypeptide components in the absence of 1,2-PD. Then, the individual components would be degraded by cellular protein turnover pathways.

Evolution

Changes in the Earth’s atmosphere are thought to underlie the evolution of the CCM (and therefore carboxysomes) (Badger et al. 2002; Badger and Price 2003). When cyanobacteria first appeared, the Earth’s atmosphere was anoxic and contained a high percentage of CO2. Under these conditions, the CCM and the carboxysome offer no apparent advantage. Consequently, it was proposed that the historical decline in atmospheric CO2 and increase in O2 produced a selective environment that allowed development of the carboxysome and the CCM (Badger et al. 2002; Badger and Price 2003). Furthermore, based on the known efficiencies of RuBisCOs at various CO2 concentrations, it was suggested that CCM may have evolved as recently as 400 million years ago when there was a large decrease in the CO2 concentration in the Earth’s atmosphere (Badger et al. 2002; Badger and Price 2003).

While it is reasonable that CO2 limitation drove carboxysome development, relatively few studies have addressed the molecular evolution of polyhedral organelles. The shells of carboxysomes are not icosahedra, but rather, they are D6 polyhedra, and no viral capsids have been shown to have this symmetry (Cannon et al. 2001). Furthermore, significant similarity between viral capsid proteins and the shell proteins of polyhedral organelles is not detected by BLAST, PSI-BLAST, or by structural searches (Kerfeld et al. 2005). Hence, a close relationship to viruses is not supported.

The polyhedral organelles that are known have related shell components but a different complement of enzymes. It appears that a shell evolved and was adapted to multiple physiological roles. The extent to which shell function may have diverged is unknown. The basic role of polyhedral organelles seems to be metabolite concentration. Their protein shell is a simple system in comparison to a lipid bilayer, but it may provide an advantage for concentration compounds such as CO2 and aldehydes, both of which pass through lipid-based membranes with relative ease.

The uniqueness of polyhedral organelles

Bacterial polyhedral organelles are distinguished from other multiprotein complexes by their protein shells. Recent studies (see above) tentatively suggest that their shell may be selectively permeable and may have embedded enzymes that facilitate metabolite or cofactor transport. This idea raises two important questions: (1) to what extent are protein membranes biologically active in general rather than simply serving protective roles and (2) what are the biochemical principles that govern the functioning of selectively permeable protein sheaths. In addition, the distinctive architecture and complexity of polyhedral organelles raise questions about their structural and functional principles. The manner in which proteins are targeted to the lumen and the possible regulation of pore activity by switches or protein factors are two of many unanswered questions about these unique and complex macromolecular structures.

Summary

Bacteria use many strategies to optimize metabolic processes. Polyhedral organelles are perhaps the most sophisticated, and they are emerging as a general mechanism of metabolic optimization. Until recently, the diversity of these structures had been overlooked because many are not formed on standard growth media and because electron microscopy is required for their observation. However, analyses of genome sequences tentatively indicate seven functionally distinct organelles distributed among over 40 genera of bacteria, and the searches that were performed would have missed organelles having unrelated or divergent shells. Indeed, nature might have invented many classes of polyhedral organelles with varying shell types, just as many different types of viral capsids are known.

Possible biotechnology applications

Historically, studies directed at understanding diverse bacterial processes have made important contributions to biotechnology. Studies conducted thus far tentatively suggest that polyhedral organelles act as molecular concentrators, and that this capacity allows them to enhance enzymatic reactions. Hence, in the long term, they might find an application to improving the enzymatic production of chemicals, to the clean-up of environmental pollutants, or to other processes not yet realized due to our lack of a fundamental understanding of the unusual structures.