Thermococcus species produce MVs (membrane vesicles) into their culture medium. These MVs are formed by a budding process from the cell envelope, similar to ectosome formation in eukaryotic cells. The major protein present in MVs of Thermococci is a peptide-binding receptor of the OppA (oligopeptide-binding protein A) family. In addition, some of them contain a homologue of stomatin, a universal membrane protein involved in vesiculation. MVs produced by Thermococcus species can recruit endogenous or exogenous plasmids and plasmid transfer through MVs has been demonstrated in Thermococcus kodakaraensis. MVs are frequently secreted in clusters surrounded by S-layer, producing either big protuberances (nanosphere) or tubular structures (nanotubes). Thermococcus gammatolerans and T. kodakaraensis produce nanotubes containing strings of MVs, resembling the recently described nanopods in bacteria, whereas Thermococcus sp. 5-4 produces filaments whose internal membrane is continuous. These nanotubes can bridge neighbouring cells, forming cellular networks somehow resembling nanotubes recently observed in Firmicutes. As suggested for bacteria, archaeal nanopods and/or nanotubes could be used to expand the metabolic sphere around cells and/or to promote intercellular communication.
The production of extracellular MVs (membrane vesicles) (50–200 nm in size) is a universal, but poorly understood, mechanism for cell–cell communication that has been only appreciated recently [1–5]. OMVs (outer membrane vesicles) in Bacteria, as well as exosomes/ectosomes in Eukarya can transfer toxins, quorum-sensing agents, pathogenicity factors, RNA and possibly DNA. The production of MVs also seems to be widespread in the archaeal domain. MVs accumulate in the inter-membrane compartment of Ignicoccus species, and could be involved in the transfer of metabolites and macromolecules between the nanoarchaeon Nanoarchaeum equitans and its host, Ignicoccus hospitalis . In Sulfolobus species, MVs are associated with proteinic toxins (Sulfolobicins) that have been identified and characterized at the genetic level [7–9]. Notably, Sulfolobus MVs contain homologues of eukaryotic ESCRTIII (endosomal sorting complex required for transport III) proteins involved in vesicle traffic and cell division . The universality of MVs suggests that MV production is a very ancient process that was possibly already operating in LUCA (the last universal common ancestor) and possibly in even more ancient cells, such as RNA-based cells .
For a long time, interest in MVs was limited to a few laboratories. Recently, this situation has changed, especially after MVs were found to be critical for many important physiological functions in Eukarya, such as microRNA transport or in pathological processes such as metastasis [11,12]. The recent creation (in 2011) of a scientific society (the International Society for Extracellular Vesicles) entirely devoted to this topic testifies to this new interest.
Thermococcales produce MVs resembling eukaryotic ectosomes
Hyperthermophilic archaea of the order Thermococcales produce high amounts of MVs . Electron microscopic analysis of thin cell sections revealed that these MVs were produced by budding of the cell envelope, a process that is reminiscent of the production of ectosomes (microparticles) by eukaryotic cells . This has been confirmed by biochemical analyses showing that MVs and cell membranes from the same species have quite similar protein and lipid compositions . The most abundant proteins present in MVs produced by Thermococcales are peptide receptors of ABC (ATP-binding cassette) transporter systems, OppA (oligopeptide-binding protein A) . These proteins are also present and abundant in MVs from Sulfolobus species . Notably, OppA is also abundant in Thermococcus membranes and MVs when cells are cultivated in a medium without peptide, suggesting that this protein not only is used to scavenge peptides but also could be involved in peptide-based signalling pathways and/or in binding toxic peptides. Interestingly, a protein homologous with eukaryotic stomatin was also identified in MVs of Thermococcus kodakaraensis (TK0348)  and a related protein, flotilin, is present in MVs from Sulfolobales . These membrane proteins, which are widely distributed in the three domains of life, are members of the SPFH (stomatin, prohibitin, flotillin and HflK/C) family . In Eukarya, they are found in microdomains called lipid rafts and could be involved in membrane bending and vesiculation .
MVs are associated with DNA and could be confused with virions in environmental studies
MVs from Thermococcales are associated with cellular DNA [10,13]. As a consequence, they are fluorescent when labelled with dyes such as SYBR Green or Hoechst, producing fluorescent dots in EFM (epifluorescent microscopy) . This suggests that MVs could be confused with virions in environmental studies, when EFM is used to count virions . The association of MVs and DNA has been indeed previously reported in Bacteria  and MVs have been observed in natural environments as well as under laboratory conditions . MVs could be also confused with virions in virome analyses when EFM is used to verify that virome samples are not contaminated with cellular organisms . This could partly explain the presence of a relatively high amount of bacterial reads in viromes . Using electron microscopy to evaluate the purity of viromes can also be misleading since MVs cannot be distinguished from virions from their morphology, and many pictures of environmental ‘virions’ published in microbial ecology papers could easily be MVs instead of bona fide virions.
MVs from Thermococcales can transfer DNA in controlled experiments
MVs from Thermococcus nautilus harbour the endogenous cryptic plasmid pTN1 . We have shown recently that MVs from the genetically tractable strain T. kodakaraensis transformed with a pTN1 derivative, the shuttle vector pCL70 , produce MVs containing this plasmid . MVs can then be used to transfer pLC70 to plasmid-free T. kodakaarensis cells, indicating that MVs can fuse with cells and deliver foreign DNA . Since MVs also protect DNA against thermodegradation at high temperature (90°C) , they could facilitate HGT (horizontal gene transfer) between hyperthermophiles both by stabilizing DNA and by serving as vehicle for the transfer itself. To investigate this possibility, it is important to determine whether MVs from a given Thermococcus species can also deliver DNA to other species of Thermococcus or even to cells from different archaeal genera or even to thermophilic bacteria, such as Thermotogales. Indeed, comparative genomic analyses have shown that HGT has indeed occurred between Thermococcus and Thermotoga , whereas a recent analysis of several new plasmids from Thermococcales (Thermococcus and Pyrococcus) identified cryptic plasmids that have been likely transferred by HGT between Thermococcales and Methanococcales .
Thermococcus gammatolerans and T. kodakaraensis produce tubular structures with a row of internal vesicles, resembling bacterial nanopods
In addition to free MVs, supernatants from cultures of Thermococcales often contain flexible filamentous or tubular structures . Some of them look like strings of MVs encapsulated in a continuous membrane derived from the cell envelope . These unusual structures are especially abundant in cultures of Thermococcus sp. 5-4  (Figure 1). SEM (scanning electron microscopy) of whole Thermococcus cells and TEM (transmission electron microscopy) of thin cell cross-sections (50 nm) have shown that these tubular structures are not formed by secondary association of free MVs present in culture supernatant, but originate directly by budding from growing cells (Figures 2b, 2e, 2g and 2h). Whereas these tubular structures appear rather flexible when observed by TEM in culture supernatant (Figure 1), they appear as rigid stick-like structures pointing perpendicular to the cell envelope when observed by SEM. They have rather constant diameters of approximately 60–80 nm and their length is variable, which suggests a growing process, with some filaments (hereinafter called nanopods or nanotubes) being more than 1 μm long. Thin cross-sections reveal that tubular structures produced by T. gammatolerans and T. kodakaraensis are formed by strings of MVs surrounded by a hairy layer probably corresponding to the S-layer covered by lipopolysaccharides (Figures 2e and 2g). They strikingly resemble the so-called nanopods that were recently observed in the soil bacterium Delftia (a deltaprotebacterium) (Figure 2B in ). These nanopods also contain chains of MVs (in that case, OMVs) within a tubular element containing S-layer material. However, these bacterial nanopods are thinner (20–30 nm in diameter from Figure 2B in ) than those produced by T. gammatolerans and T. kodakaraensis. In addition to nanopods, SEM analyses showed the presence of large protuberances (200–300 nm) on the surface of T. gammatolerans and T. kodakaraensis cells (Figures 2a and 2b). The analysis of thin cross-sections has revealed that, similar to nanopods, these nanospheres (200–300 nm in diameter) correspond to clusters of MVs surrounded by S-layer (Figures 2c and 2d). Figure 2(f) shows a nanosphere that seems to be escaping from the surface of a Thermococcus cell.
Structure and protein composition of nanotubes present in culture supernatant from Thermococcus sp. 5-4
Nanopods and nanospheres produced by Thermococcus species observed by SEM (a, b, h), or by TEM of thin cell cross-sections (c, d, e, f, g)
Thermococcus sp. 5-4 produces nanotubes lacking visible internal vesicles but with typical vesicle protein composition
Supernatants from cultures of Thermococcus sp. 5-4 are especially rich in flexible filaments, which, unlike the one produced by T. gammatolerans and T. kodakaraensis do not seem to contain MVs (Figure 1). Observation of Thermococcus sp. 5-4 thin cell cross-sections confirms that these nanotubes do not contain visible MVs, but are formed by a continuous layer of cell envelopes that protrude from the cell with an internal cytoplasmic component and an external S-layer (Figures 3a–3c). Although Thermococcus sp. 5-4 nanopods do not contain visible MVs, their protein content (Figure 1) is quite similar to those of MVs from other Thermococcales. MV preparations enriched in nanotubes mainly contain membrane proteins (S-layer protein, oligopeptide-binding protein, ABC transporter components, amylopullulanases and V-type ATPase subunit) (M. Gaudin, unpublished work). Notably, the major protein present in these peparations is a closely related homologue of the OppA proteins detected in T. gammatolerans and T. kodakaraensis MVs  (Figure 1). Stomatin, previously detected in MVs from T. kodakaraensis , is also present in these preparations (Figure 1), confirming that this universal membrane protein might be an important component of archaeal MVs, as it is the case in eukaryotic MVs.
Nanotubes from Thermococcus sp. 5-4, fine structure and network formation
MVs are frequently observed by TEM at the tips of Thermococcus sp. 5-4 nanotubes (Figures 1 and 3c). This suggests that nanotubes could be used for MV production. Figure 3(c) shows a thin section of an MV located at the tip of a nanotube produced by Thermococcus sp. 5-4. This image shows a direct connection between the cytoplasmic membrane of the nanotube and those of the vesicle, suggesting that nanotubes could be involved in MV production.
Archaeal nanopods can form networks bridging cells of Thermococcus sp. 5-4
In concentrated cultures, most cells of Thermococcus sp. 5-4 are connected by rigid nanotubes visible by SEM (E. Marguet, unpublished work; an example can be seen in Figure 3d). These nanotubes seem to connect the cells via their S-layers. Such connections are never observed in thin sections, probably because they require cutting precisely all along the tubule connecting two cells. Two neighbouring cells can be linked by more than one nanotube (up to three) of various lengths (from 100 nm up to 1.5 μm). The same cell can produce both nanotubes that connect to other cells and nanotubes with a free end. Figure 3(d) shows the example of one cell producing nine nanotubes (large white arrow), with only three of them being connected to the neighbouring cell. In many cases, several free-ended nanotubes seem to deliver grapes of MVs that spread on to the coverslip surface (small white arrows).
Dubey and Ben-Yehuda  reported recently the existence of ‘nanotubes’ bridging neighbouring Bacillus subtilis cells together as well as B. subtilis and Staphylococcus aureus. They visualized a transfer of cytoplasmic fluorescent molecules between adjacent cells and reported that plasmids can be transferred from cell to cell via these nanotubes. These bacterial nanotubes resemble those of Thermococcus sp. 5-4, although they are more variable in diameter (from 30 to 130 nm) compared with those of Thermococcus sp. 5-4 (from 60 to 80 nm). We suggest that archaeal nanotubes are also used for cell–cell communication, in connection with the production of MVs
The formation of giant networks in which archaeal cells are connected by tubular structures has been observed previously with hyperthermophilic Crenarchaea of the genus Pyrodictium [26,27]. These Pyrodictium nanotubes (25 nm in diameter) are much thinner than those produced by Thermococcus sp. 5-4, and it is not known whether or not they are linked to MV production. Among Thermococcales, Näther and Rachel  reported that Pyrococcus furiosus cells are connected in artificial biofilm by cable-like structures formed by the aggregation of flagella. These bunches of flagella are wide structures with a diameter that varies around 200 nm. We have indeed observed this type of structure with T. gammatolerans (Figure 4). Flagella (~20 nm in diameter) are visible at the tip or within large tubular structures that can also sometimes connect different cells (Figure 4). This suggests that Thermococcales can produce different types of tubular structures to connect cells in natural biofilms, some more specialized in the transport of material between cells (nanotubes), and others more specialized in sticking to surfaces (flagellar aggregates).
Cells of T. gammatolerans observed by SEM
The study of archaeal MVs is still in its infancy, and many questions remain to be tackled. It will be especially important to determine whether MVs can indeed play the role of vehicle for HGT in natural environments and, if it does, what its relative importance is compared with transformation, transduction or transfer via conjugative plasmids. Since MVs can carry toxins, they cannot be simply viewed as one more tool for gene exchanges, but as complex biological entities that should also play a role in inter-species competition. The evolutionary connection between MVs and virions is another line of research worth considering. It has been shown that MVs can prevent bacteriovirus infection by trapping virions , and the observation of archaeal viruses (virions) attached to MVs  suggests that archaeal MVs could also somehow reduce the load of viral infection in nature, playing the role of lures.
The mechanisms of MV formation and fusion with cells have been studied extensively in eukaryotes, but it is still not known how this operates in the case of Bacteria. As for Archaea, it has been suggested that ESCRTIII-like proteins detected in MVs from Sulfolobus are involved in MV formation , but this remains to be demonstrated genetically. Thermococcus species do not harbour ESCRTIII homologues, but contain several ATPases homologous both with Vps4 (vacuolar protein sorting 4), the ATPases that energize the ESCRTIII machinery, and with bacterial FtsH (, and P. Forterre, unpublished work). Interestingly, in Escherichia coli, both homologues of stomatin, HflK and QmcA, also called prohibitins, interact with FstH [32,33]. This suggests that one or several Vps4/FtsH-like proteins in Thermococcus species could interact with stomatin in MV formation. The structure of stomatin from Pyrococcus horikoshii has been solved  and genome context analysis has shown that the stomatin genes of Thermococcales are linked to a gene encoding a protease that should be involved in stomatin processing . Vps4/FtsH, as well as stomatin and associated protease, should be priority targets for future genetic studies aiming at deciphering the mechanism of MV formation. As a first step in that direction, we have constructed a knockout mutant of the T. kodakaraensis stomatin-encoding gene. The mutant still produces MVs, but with a slightly altered size (M. Gaudin, unpublished work). On a more general basis, the fact that all cells produce MVs suggest a common basic mechanism, with additional processes specific for different lineages. We suggest that overproduction of cell envelope material compared with cytoplasmic components could produce an excess of cell envelope material that could be the primary force for extrusion of MVs, nanotubes or nanopods. Natural selection could have built on such general phenomenon to produce more specialized structures with selectable physiological functions. Shetty et al.  suggested, for instance, that nanopods are used by soil bacteria ‘to project OMVs over long distance from the cell independently of diffusion’, allowing them ‘to expend their metabolic sphere of influence within their environment’. Similarly, archaeal nanopods could project MVs from one cell to the other, facilitating cellular communication at high temperature within the porous structure of hydrothermal vents.
Molecular Biology of Archaea 3: An Independent Meeting held at the Max Planck Institute for Terrestrial Microbiology, Marburg, Germany, 2–4 July 2012. Organized and Edited by Sonja-Verena Albers (Max Planck Institute for Terrestrial Microbiology, Germany), Bettina Siebers (University of Duisberg-Essen, Germany) and Finn Werner (University College London, U.K.).
endosomal sorting complex required for transport III
horizontal gene transfer
outer membrane vesicle
oligopeptide-binding protein A
scanning electron microscopy
transmission electron microscopy
vacuolar protein sorting 4
We thank Egle Conforto (La Rochelle) and the TEMSCAN platform (Toulouse) for help with SEM and Ludivine Houel-Renault (Orsay) for help in the preparation of thin sections and TEM.
This work was funded by the program ‘Marine Genome-Marine Biotechnologie’ of the Centre National de la Recherche Scientifique and the Japan Science and Technology Agency.