The cultures reminded him of Swiss cheese, says Egbert Hoiczyk. It was 2009 and the microbiologist and his colleagues at Johns Hopkins University were growing the motile bacterium Myxococcus xanthus on agar plates. “What we observed was something quite bizarre,” recalls Hoiczyk, now at the University of Sheffield in the UK.

Bacteria had grown out in a yellowish smear coating the plates, but then holes appeared in the microbial film. The empty spaces looked like plaques created by bacteria-killing viruses, so Hoiczyk’s team concluded there must be a viral invader in their cultures. When the researchers scraped the cleared areas and examined the material under an electron microscope, they observed 32-nanometer-wide structures resembling 12-sided dice—the right size and shape for a bacteriophage.

It’s becoming accepted, a lot, in the last 10 or 20 years that the prokaryotic cytoplasm is highly organized.

—Cheryl Kerfeld, Michigan...

Hoiczyk, undergraduate Alan Lam, and graduate student Colleen McHugh set out to isolate the virus. But something was off. When the researchers attempted to purify the phages by centrifuging them in standard cesium chloride gradients, where viruses usually show up in the middle, the particles always dropped to the bottom of the columns. The investigators eventually realized they weren’t purifying viruses, but tiny, iron-laden nanocompartments native to the bacteria. (They never did figure out why the holes appeared in the cultures.)

The structures were relatively new to science, having been described for the first time just a year earlier in the bacterium Thermotoga maritima.1 At 25–66 nm across, they are typically too small to notice unless you’re looking for them, Hoiczyk explains. But in recent years, thanks to modern genomics and bioinformatics, scientists have found nanocompartments across a range of bacterial phyla.

SWISS CHEESE: In 2009, microbiologist Egbert Hoiczyk, then at Johns Hopkins University, observed strange holes in cultures of Myxococcus xanthus. He and his colleagues thought the spaces were caused by viruses killing the bacteria, but closer investigation revealed something quite different: 32-nanometer-wide structures known as nanocompartments.

It’s not the first time researchers have identified what is essentially an organelle in Bacteria, a kingdom traditionally thought to lack subcellular compartments. Decades earlier, scientists had first described larger structures called microcompartments—100–600 nm in diameter—that appeared in micrographs of cyanobacteria, though researchers are only building a detailed understanding of those compartments today. “It’s becoming accepted, a lot, in the last 10 or 20 years that the prokaryotic cytoplasm is highly organized,” says Cheryl Kerfeld, a structural biologist who splits her time between one lab at Michigan State University and another at the Lawrence Berkeley National Laboratory.

Preliminary studies suggest that nanocompartments assist with stress responses, while microcompartments often have roles in metabolism. And although the details remain murky, it’s becoming clear that these structures create a specialized microenvironment for a specific purpose, says Kerfeld: “I would define them as organelles.” A growing number of researchers are now working to understand what these organelles do. Kerfeld predicts that eventually, the population of people studying bacterial compartments “is going to be as big as any eukaryotic organelle community.”

The oddity of bacterial microcompartments

It was the geometry of microcompartments that first caught Kerfeld’s eye. Attending cyanobacteria meetings in the early 2000s, she noticed that researchers would exhibit micrographs of bacteria containing “crazy-looking bodies.” The three-dimensional symmetry of the so-called “polyhedral bodies” was irresistible to the structural biologist, who set out to crystallize the proteins that made up their exterior.

Kerfeld, then a research scientist at the University of California, Los Angeles (UCLA), focused on carboxysomes, microcompartments first observed in the 1950s.2 They were known to perform carbon fixation, and under the microscope, they looked like viral shells, or capsids, protein-based structures that encapsulate RNA or DNA. But it wasn’t clear, at that time, whether carboxysomes were true shells with an inner space for reaction components, or just geometrical clumps of relevant enzymes and reactants, says UCLA structural biologist Todd Yeates, Kerfeld’s former PhD advisor.

Members of the Yeates lab debated whether the carboxysome protein structure, given its similarity to that of viruses, would resemble the folds in viral capsids. While Kerfeld did find that the carboxysome proteins of the cyanobacteria Synechocystis formed shell-like proteins, those who had argued against any similarity to viruses were right: the amino acid–scale architecture of the carboxysomes differed from that of viral capsids. Individual proteins formed hexagonal tiles that composed the 12 faces of the bacterial structures. Many of these tiles contained pores, presumably to let reactants in and products out.3 “It was clear they were going to be really sophisticated machines,” says Yeates.

MICROCOMPARTMENT STRUCTURE: In 2017, Cheryl Kerfeld and colleagues published the crystal structure of an assembled 40-nm microcompartment from a myxobacterium. Kerfeld estimates the microcompartment could hold approximately 300 interior proteins of 30 kiloDaltons each, though no one knows for sure how such contents are organized.

At the time, carboxysomes and another type of microcompartment called metabolosomes, discovered in the ’90s, were thought to be rare oddities, says Kerfeld. To see if they could find more, she and Yeates each went hunting through microbial genomes for sequences encoding the proteins that make up microcompartment shells.

Conveniently, researchers already knew that the genes for the shell proteins tend to colocalize with the DNA that codes for the enzymes the compartments will house.4 Yeates’s group took advantage of this fact to identify putative microcompartment groupings of shell and inner enzyme genes and extrapolate their potential functions. In 2013, the team delineated seven categories of microcompartments, including known carboxysomes and metabolosomes as well as novel types, such as one apparently involved in the metabolism of amino alcohols.5 Separately, Kerfeld, by now at Michigan State and Berkeley Lab, and her team used a similar approach to identify 23 different types of microcompartments spread across 23 bacterial phyla, as they reported the following year.6

Now, Kerfeld and Markus Sutter in her Berkeley lab are repeating the bioinformatic analysis and incorporating more genomes, including those from uncultivated species. They’ve already found more microcompartments, Kerfeld says. “The proportion of bacteria that seem to make these is rising.” A couple of species possess genes for six different kinds of microcompartments, potentially giving them access to a complex metabolism.

Why house certain reactions in tiny containers? Computer modeling indicates that microcompartments should maximize the turnover of metabolites by keeping reaction intermediates close and interfering chemicals at a distance.7 For instance, concentrating the carbon-fixing enzymes RuBisCO and carbonic anhydrase in a carboxysome makes the processing of carbon dioxide more efficient.8 And cordoning off toxic reactions, such as those that produce aldehyde intermediates—a hypothesized, though unproven, job of metabolosomes—would protect the rest of the cell’s interior.

Not all bacteria can make nano- and microcompartments. In fact, most denizens of mammalian intestines seem to lack microcompartments, although many pathogens possess them. For example, Salmonella’s microcompartments metabolize the organic compounds propanediol and ethanolamine, which are found in processed foods and the human gut. Compartmentalizing the reactions is thought to allow the bacterium to digest nutrients that members of the human intestinal microbiome cannot, allowing the pathogen to outcompete them. Other pathogens, such as Listeria and Clostridium, also contain metabolosomes.

Microcompartment biology

Bioengineers are eager to use bacterial compartments too (see “Practical Applications” below), and to do that, they’ll need a better understanding of the organelles’ biology. Thanks to genetics and cell biology experiments, “very basic design rules are starting to fall into place,” says Danielle Tullman-Ercek, a chemical engineer at Northwestern University in Illinois.

In 2010, for example, Yeates, Thomas Bobik of Iowa State University, and colleagues reported how the inner enzymes and shell proteins find each other. Examining the genomes of several bacterial species, they identified sequences for short peptides on the enzymes inside propanediol metabolosomes. These peptides allow the enzymes to find and be packaged into the microcompartment shell.9

Types of Bacterial Compartments

Compartment typeFunctionSize range
CarboxysomeThese bacterial microcompartments contain RuBisCO and carbonic anhydrase, allowing them to fix carbon efficiently by concentrating the substrate, CO2, and excluding O2, which can interfere by competing with CO2 for RuBisCO. There are two types: alpha-carboxysomes and beta-carboxysomes.100–600 nm
MetabolosomeThis type of microcompartment contains enzymes that break down certain metabolites, allowing the structures to sequester toxic aldehyde intermediates from the rest of the cell. There are several known types of metabolosomes.40–150 nm
Encapsulin nanocompartmentThese structures are smaller than microcompartments and are made of a different type of shell protein. They appear to sequester toxins or store enzymes that help bacteria deal with stressors. Scientists have identified at least three classes so far, based on their associated enzymes.25–66 nm

In a separate study, Kerfeld and colleagues engineered cyanobacteria to express fluorescently tagged RuBisCO so they could observe the carboxysome assembly process. The researchers watched the enzymes coalesce into aggregates in the cytoplasm under the microscope. A few hours later, they saw smaller clusters of RuBisCO proteins begin to detach from the clumps. These smaller enzyme particles were within polyhedral carboxysomes. Kerfeld’s team concluded that the inner enzymes first aggregate into a “procarboxysome,” and then individual microcompartment shells assemble around some of those enzymes and bud off as mature carboxysomes.10 (See infographic below.)

Once the microcompartments have assembled, they somehow selectively obtain reactants and release products, so the shell must function in a manner akin to the semipermeable membranes of eukaryotic organelles. In a 2015 study, Bobik and colleagues fiddled with the propanediol metabolosome pore to show that the wild-type version selectively permits propanediol into the microcompartment, while blocking the exit of the toxic reaction intermediate propionaldehyde.11

MICROCOMPARTMENTS UP CLOSE: This micrograph shows empty shells of Halothece carboxysomes, which self-assemble in the presence of shell proteins but not the enzymatic cargo and have potential for various biotechnological applications. These structures tend to be smaller than natural, fully packed bacterial microcompartments. (Scale bar = 50 nm)
Nature Commun, 9:2881, 2018

Taking a closer look at the pores from Salmonella ethanolamine and propanediol microcompartments, Tullman-Ercek’s group found functional differences between two very similar shell proteins, one from each metabolosome, called EutM and PduA, respectively. These hexagonal proteins, which tile the sides of the microcompartments, contain a six-angstrom hole in the middle. (See infographic below.) When the researchers put EutM in place of the PduA gene in the propanediol-processing operon in Salmonella enterica, the bacteria didn’t grow normally. When cultured with propanediol, the transgenic strain grew slowly at first, then caught up and eventually outpaced unmodified S. enterica.

The researchers suspect that the growth was different because of the EutM pore’s slightly more negative charge. Sure enough, introducing a point mutation in the gene—one that made the charge of the pore more similar to that of EutM—was sufficient to alter Salmonella’s growth, indicating that the pore’s structure somehow affects the bacteria’s metabolism.12

“We actually think we were changing how much of the toxic intermediate is getting out,” Tullman-Ercek says. “I think [the microcompartment is] holding in intermediates, and letting out products.”

Even smaller compartments

Back in Baltimore in 2009, McHugh and Lam were puzzling over their virus-like structures that weren’t viruses at all. (And at 32 nanometers across, they were too small to be microcompartments.) When the researchers analyzed the structures’ four component proteins using mass spectrometry, they noticed that one had similarities to the shell protein of the first nanocompartment to have been described, just the previous year, by Kerfeld’s Berkeley Lab research associate Sutter.

As a PhD student at ETH Zurich, Sutter had characterized nanocompartments. The lab, run by Nenad Ban, had been trying to crystallize ribosomes, and wound up with mysterious particles twice the size of the protein-building machines. Sutter discovered that they were in fact protein shells, built out of a single protein, which he called encapsulin A. As with microcompartments, Sutter found, the enzymes within the structures attach to the shell via short peptide tags. In T. maritima, the thermophilic bacteria Sutter was studying, the encapusulin nanocompartments (sometimes referred to simply as encapsulins) housed an iron-storage protein that potentially protected the cells from oxidative stress.1

Microcompartment Form and Function

Many different bacterial species contain small, protein-based compartments that expand cells’ metabolic repertoires by sequestering chemical reactions. Depending on their enzyme contents, the compartments can fix carbon, break down molecules for energy, or protect cells from stressful conditions.


Bacterial microcompartment shells are built out of thousands of protein subunits that fall into three basic structural motifs:

© thom graves

BMC: The principal shell protein comes in two main forms, BMC-H, which forms a hexamer, and BMC-T, which forms a trimer. These hexagon-shape components tile together to form the 20 sides of the icosahedron and have a central pore to allow substrates in and products out. The pores in the BMC-H tiles are small, allowingmolecules of just one or a few carbon atoms in and out of the microcompartment. BMC-T pores, on the other hand, are larger, presumably for movement of bigger molecules, and these can be opened or closed. Some of the tiles are thought to bend to form the edges of the icosahedron’s faces, though it’s still unclear which types of BMC protein take on this conformation.

BMV: These pentameric proteins form the vertices of the icosahedron.


Some microcompartments form when the inner components aggregate (1) the shell proteins assemble around enzymes and other molecules (2) to generate the completed compartment (3). Other types of microcomparments form when the core proteins and shell come together at the same time (not shown).

See full infographic: WEB | PDF
© thom graves

McHugh and colleagues, comparing their Myxococcus xanthus protein’s sequence to that of T. maritima’s encapsulin, found the two were homologous.13 And like the T. maritima particles, the Myxo protein shells housed proteins that store iron. Two of the proteins inside the compartments acted much like another iron storage protein, ferritin, but the size of the nanocompartments meant they could hold about 30,000 atoms of iron—10 times as much as a ferritin complex.

Why would the bacterium need such iron stores? The researchers found a clue when they took away nutrients. “When we starved the cells, there were [many] more of the particles,” recalls Hoiczyk. Hungry M. xanthus cells change their activity, coming together to form a fruiting body. The nanocompartments, Hoiczyk says, “obviously play a role under that stressful transition.” Specifically, the nanocompartments appear to protect the cells from oxidative stress, as in T. maritima, adds McHugh, who now has her own lab at the University of California, San Diego.

Most denizens of mammalian intestines lack micro­compartments, although many pathogens possess them.

It turns out that, like microcompartments, encapsulin nanocompartments may be relatively common. Last year, Tobias Giessen, a postdoc in Pamela Silver’s laboratory at Harvard Medical School and the Wyss Institute for Biologically Inspired Engineering, reported he’d identified more than 900 potential encapsulin systems in banked bacterial genomes. The encapsulin shell and interior enzyme genes colocalized, or even fused together, across 15 bacterial and 2 archaeal phyla.14 In the past year, his analysis has raised those numbers to more than 2,000 encapsulin systems in 18 bacterial and 3 archaeal phyla, he says.

Based on published information about the enzymes’ likely functions, Giessen divided the compartments into three main categories. The most common type contain peroxidases, which process hydrogen peroxide and other organic peroxides. Another large group houses iron-storage proteins, as do the M. xanthus and T. maritima nanocompartments. The third group contains hemerythrins, enzymes that can protect cells from stress caused by reactive oxygen and nitrogen species. All three types are likely to play a role in the cell’s response to stress, notes Giessen.

Practical Applications 

What researchers discover about bacterial nano- and microcompartments could be of use to bioengineers, who are eyeing the structures for their own purposes. Each compartment has its advantages. Encapsulin nanocompartments, for instance, are simple systems in which the cargo enzymes co-assemble with highly stable shells. Larger microcompartment types are more complex and less stable, but they have a size advantage, allowing them to hold more cargo. Researchers ought to be able to play with their structures, altering pore size and cargoes, to create compartments for specific purposes, says Martin Warren, a biochemist at the University of Kent in Canterbury, England.

As a proof of principle, Warren’s group altered propanediol-metabolizing microcompartments, which normally package three enzymes in that pathway, to fill them with two new enzymes, a pyruvate decarboxylase and an alcohol dehydrogenase. The resulting bioreactor efficiently transformed pyruvate, the product of glycolysis, into ethanol (ACS Synth Biol, 3:454–65, 2014). Now, Warren is collaborating with companies to produce high-value chemicals such as biofuels. Other potential applications include fixing nitrogen, to make a microcompartment-based fertilizer, or carbon, perhaps to engineer plants that are super efficient at fixing carbon. Another thought is to use microcompartments in waste cleanup.

In medicine, one might engineer desirable gut bacteria to use microcompartments so that pathogens that contain them, such as Salmonella, would lose their edge, speculates Cheryl Kerfeld of Michigan State University and the Lawrence Berkeley National Laboratory. Alternatively, researchers could design antibiotics that target pathogenic bacteria’s microcompartments, leaving commensals unharmed, posits Danielle Tullman-Ercek, a chemical engineer at Northwestern University. Others suggest that with their polyhedral, often virus-like shape, nano- and microcompartments sporting pathogen-specific antigens would make ideal vaccines. Or, offers Warren, microcompartments might be good drug-delivery vehicles.

Gil Westmeyer, a physician and bioengineer at Technical University Munich, hopes to use nanocompartments as imaging probes within mammalian cells. In a recent study, his group engineered human kidney cells to express nanocompartments similar to those found in Myxococcus xanthus, and altered the interior components to produce contrast agents for various imaging platforms (Nat Commun, 9:1990, 2018). To use these, researchers might force production of nanocompartments only in certain cell types, or engineer the compartments’ exteriors to target them to specific cellular structures, says Westmeyer, adding that “there will be applications that we cannot even imagine right now.”

So far, no synthetic nano- or microcompartment applications have been commercially realized. “We need to actually show that these things have a use,” Warren says.

To speed the engineering of microcompartments, Kerfeld recently developed an artificial system to load the protein shells with cargo. Researchers in her lab used a standard protein-linking system made up of two peptides, SpyTag and SpyCatcher, that bond when they meet. The scientists hooked one half of the duo to the interior of the shell, and the other half to the desired cargo proteins, effectively controlling what ended up inside. In addition, they tagged the shells for affinity purification (Nat Commun, 9:2881, 2018).

These tools will significantly accelerate research and bioengineering, says Tullman-Ercek. “There are so many applications for these little guys.”

This year, he identified a new class with enzymes that produce aldehyde intermediates. These nanocompartments might serve to sequester the toxic compounds, as has been found of related microcompartments. But for all nanocompartments, Giessen admits, “we don’t really know too much about their real biological functions yet.”

NANOCOMPARTMENTS UP CLOSE: Micrographs of encapsulin shells
tobias giessen

As scientists continue to probe the functions of these newly discovered structures, Hoiczyk is still struck by how much they look like viruses. “The similarity is kind of mind-boggling,” he says, noting that the encapsulin shell protein contains a fold that is exceedingly similar to one found in the HK97 bacteriophage capsid. It’s likely that the resemblance is not due to chance. Hoiczyk and others suspect the nanocompartments share an evolutionary history with bacteriophages, though they don’t know which came first.

The origin of the larger microcompartments is less obvious, though their component parts share some features with cellular proteins.15 For both types of structures, there’s still plenty to learn—not only about their basic biology, but also how to engineer them to perform industrially relevant functions.

And who’s to say that nano- and microcompartments are the only types of protein-encapsulated bacterial organelles? “It’s somewhat interesting to muse about some that haven’t been discovered yet,” Yeates says. “It would be cool if there were others.”


  1. M. Sutter et al., “Structural basis of enzyme encapsulation into a bacterial nanocompartment,” Nat Struct Mol Biol, 15:939–47, 2008.
  2. G. Drews, W. Niklowitz, “Cytology of Cyanophycea. II. Centroplasm and granular inclusions of Phormidium uncinatum,” Arch Mikrobiol, 24:147–62, 1956.
  3. C.A. Kerfeld et al., “Protein structures forming the shell of primitive bacterial organelles,” Science, 309:936–38, 2005.
  4. S.H. Baker et al., “Identification and localization of the carboxysome peptide Csos3 and its corresponding gene in Thiobacillus neapolitanus,” Arch Microbiol, 173:278–83, 2000.
  5. J. Jorda et al., “Using comparative genomics to uncover new kinds of protein-based metabolic organelles in bacteria,” Protein Sci, 22:179–95, 2013.
  6. S.D. Axen et al., “A taxonomy of bacterial microcompartment loci constructed by a novel scoring method,” PLOS Comput Biol, 10:e1003898, 2014.
  7. C.M. Jakobson et al., “A systems-level model reveals that 1,2-propanediol utilization microcompartments enhance pathway flux through intermediate sequestration,” PLOS Comput Biol, 13:e1005525, 2017.
  8. N.M. Mangan, M.P. Brenner, “Systems analysis of the CO2 concentrating mechanism in cyanobacteria,” eLife, 3:e02043, 2014.
  9. C. Fan et al., “Short N-terminal sequences package proteins into bacterial microcompartments,” PNAS, 107:7509–14, 2010.
  10. J.C. Cameron et al., “Biogenesis of a bacterial organelle: The carboxysome assembly pathway,” Cell, 155:1131–140, 2013.
  11. C. Chowdhury et al., “Selective molecular transport through the protein shell of a bacterial microcompartment organelle,” PNAS, 112:2990–95, 2015.
  12. M.F. Slininger Lee et al., “Evidence for improved encapsulated pathway behavior in a bacterial microcompartment through shell protein engineering,” ACS Synth Biol, 6:1880–91, 2017.
  13. C.A. McHugh et al., “A virus capsid–like nanocompartment that stores iron and protects bacteria from oxidative stress,” EMBO J, 33:1896–911, 2014.
  14. T.W. Giessen, P.A. Silver, “Widespread distribution of encapsulin nanocompartments reveals functional diversity,” Nat Microbiol, 2:17029, 2017.
  15. M. Krupovic, E.V. Koonin, “Cellular origin of the viral capsid–like bacterial microcompartments,” Biol Direct, 12:25, 2017.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?