Addressing microbial organelles: a short peptide directs enzymes to the interior.

M any bacterial species contain protein shell-bound polyhedral inclusions referred to as bacterial microcompartments (BMCs). The earliest of these to be described and analyzed in detail was the carboxysome (1), an organelle that is essential for efficient CO2 fixation in cyanobacteria and many chemoautotrophs. By cosequestering ribulose 1,5-bisphosphate carboxylase/oxygenase (RubisCO) and a carbonic anhydrase, the carboxysome provides a compartment with elevated CO2 levels that overcome some of the kinetic deficiencies of RubisCO (2). Reverse genetic studies of isolated carboxysome proteins revealed that homologues of the major shell protein genes occur in the genomes of heterotrophic bacteria within operons for metabolic pathways that are completely unrelated to autotrophic CO2 fixation (3). Recent comparative genomics analysis suggested that these bacterial organelles play a much broader role than originally surmised, because the genetic potential for BMC formation is widespread among the bacterial lineages (4). Salmonella enterica uses two different BMCs (Pdu and Eut) for the degradation of propanediol and ethanolamine, respectively. Sequestering as many as four sequential enzymes from the respective pathways in polyhedral protein shells is thought to prevent diffusion and consequent loss of volatile aldehyde intermediates that might damage cellular structures. In a recent issue of PNAS, a report by Fan et al. (5) identifies a short peptide signal, present on the N terminus of some enzymes contained within the Pdu BMC of S. enterica, as directing the placement of these enzymes inside the organelle. Their study constitutes a major step toward elucidating the selfassembly processes that bring about BMC formation within the bacterial cytoplasm. Despite their function as organelles and the apparent permselectivity of their shells, BMCs are not surrounded by a lipid bilayer-based membrane like their eukaryotic counterparts but are instead bordered by a thin protein shell. The shells of BMCs that have been purified and studied directly were shown to be composed of several thousand copies of orthologous polypeptides with conserved domains (Pfam 00936); recent crystallographic studies revealed that individual monomers of the predominant BMC shell proteins are arranged into thin hexamers that strongly interact in an “edgeon” fashion to form the flat facets of the polyhedral compartments (6–8). Another, albeit minor, shell constituent (Pfam 03319) that appears to occur in all BMCs forms the pentamers thought to occupy the vertices of the polyhedral structures (9). One of the great mysteries surrounding BMCs is the mechanism by which their shells, all of which apparently are built from proteins that belong to the same two structural families, function in widely diverse metabolic pathways and package a large variety of different and seemingly structurally unrelated enzymes. Because BMCs lack a bounding lipid bilayer-based membrane, the mechanisms that target polypeptides to their interior are likely different from those that guide newly synthesized proteins to the plasma membrane. Related to the targeting question is that of the mechanism by which the proper stoichiometric ratio of the encapsulated enzymes, which is likely crucial to BMC function, is ensured during assembly of the organelle. Two pathways were previously put forth for the biogenesis of BMCs that are based on electron micrographs of apparent carboxysome assembly intermediates. One of these proposes that a preassembled RubisCO core guides assembly of the polyhedral shell (10). The other model posits the existence of a preassembled (partial) shell that directs the packaging of RubisCO into the BMC interior (11). Clearly, carboxysome shell formation is not dependent on the presence of RubisCO, as was shown for a Halothiobacillus neapolitanus RubisCO deletion mutant that forms empty BMC shells of otherwise normal shape and composition (12). However, that same study provided strong evidence for specific interactions between payload and shell proteins by identifying the large subunit of the enzyme as the factor that determines whether a particular RubisCO species can be packaged into the carboxysome. Contrary to the N-terminal peptide that is identified by Fan et al. (5) on some enzymes of the Pdu BMC, sequence comparisons between carboxysomal and noncarboxysomal large subunits of RubisCO from a variety of autotrophic bacteria did not reveal an obvious packaging signal (12). As Fan et al. (5) point out, a variety of short peptide sequences have been reported to mediate assembly of larger protein complexes. Perhaps most relevant to BMC packaging is the landmark study by Sutter et al. (13) that Fig. 1. Targeting of sequestered proteins to the interior of BMCs. As described in PNAS (5), a short extension of ≈20 amino acids found on the N terminus of PduP directs the enzyme to the lumen of the Pdu BMC. (A and B) Fan et al. (5) propose that the N-terminal segment binds specific sites on the interior surface of shell protein assemblies. (C) Additional protein interactions are likely to occur between shell-anchored proteins and those found deeper in the BMC interior and would account for the solid appearance of BMC interiors observed with various imaging methods.