Carboxysome proteins CcmK2 and CcmK4 crystallized in tightly packed hexameric units that assembled, also tightly, into sheets. “The six subunits leave very little space down the middle,” says Yeates. “And the space left is highly [positively] charged.” The proximity of so many charges is more than coincidence. For instance, the pore might regulate ion passage, allowing through small negatively charged molecules such as bicarbonate (which is converted to CO2) but blocking oxygen (which inhibits RuBisCO). Mutagenesis of the pore-lining residues should allow the group to test this hypothesis.
Shells formed by CcmK homologues do more than fix carbon. The breakdown of propanediol in Salmonella, for instance, occurs inside polyhedral bodies, which might confine a toxic pathway intermediate. As they are not membrane bound, the compartments might not be considered true organelles. But they are, Yeates says, “if organelles are defined by purpose—spatial separation and control of what comes in and out. The bacteria are trying to achieve the same level of sophistication [found in eukaryotic organelles].”
Structurally, carboxysomes and other polyhedral bodies resemble viral particles, suggesting that perhaps bacteria co-opted viruses in the same way that led eukaryotic cells to contain bacteria-like organelles such as mitochondria. But small proteins with CcmK-like architectures seem to have evolved independently multiple times. There is, therefore, no sure way to know whether the primitive bacterial organelles were originally viruses or arose on their own.