COMMENTARY

Bacteria with a eukaryotic touch: A glimpse of ancient evolution?

Patrick Forterrea,b,1 and Simonetta Gribaldoa aInstitut Pasteur, 25 rue du Docteur Roux, 75015 Paris, France; and bUniversité Paris Sud, Institut de Génétique et Microbiologie (IGM), Centre National de la Recherche Scientifique (CNRS), 91405, Orsay Cedex, France

he discovery in 1991 by Fuerst B A E includes all known with cell plans and Webb that the chromosome involving compartmentalization of the cy- T of the bacterium Gemmata ob- PVC toplasm by an intracytoplasmic membrane scuriglobus (a member of Planc- ICM ER (ICM). The double membrane of the tomycetales) is surrounded by a double G. obscuriglobus nucleusisformedbyin- membrane, mimicking a eukaryotic nu- NM vagination of this ICM, much like the double cleus, challenged the traditional pro- Gemmata membrane of the eukaryotic nucleus is karyote/eukaryote classification of living NM formed by invagination of the endoplasmic organisms based on cell structure (1). fi __ reticulum (ER) (3). Signi cantly, Devos and Since then, such an unexpected observa- colleagues detected MC in com- tion has puzzled evolutionists, leading to partmentalized PVC bacteria only. In par- different reactions. Most considered ticular, they did not find any MC proteins G. obscuriglobus a curiosity of the bacterial in , which in fact do not have an world. Until recently, this view has been ICM. Using antibodies prepared by Devos supported by the absence of obvious eu- and colleagues against one of the eight karyotic signatures in the MC proteins of G. obscuriglobus, gp4978, of Planctomycetales and related bacteria LUCA Fuerst and colleagues found that this pro- (2). For a few others, the “nucleus” of tein colocalizes with vesicles containing G. obscuriglobus should, one way or an- Fig. 1. A scenario for the origin of modern com- partmentalized cells (Eukarya and PVC bacteria) proteins imported by endocytosis (4). They other, testify as some sort of evolutionary from a compartmentalized LUCA. A, Archaea; B, also obtained images showing that gp4978 link between and modern Bacteria; E, Eukarya. The cytoplasm and chromo- associates with the cytoplasmic mem- eukaryotes (3). Two landmark papers somes of Archaea and Eukarya are shown with the brane in the first stage of membrane in- now seem to strengthen this opinion. same color in reference to the high similarity of vagination. These data strongly suggest that In PNAS, Fuerst and colleagues show these two domains for informational proteins and gp4978 is involved in the membrane- that G. obscuriglobus can perform a func- several membrane-bound processes (e.g., ATP syn- remodeling process that triggers endocyto- thesis). For the sake of simplicity, eukaryotic or- tion previously thought to be restricted to sis. It is tempting to think that, more gen- eukaryotes: endocytosis (4). Whereas in ganelles have not been indicated. Also for simplicity, the ICM of PVC bacteria and the ER of Eukarya have erally, the multiple MC proteins present in traditional bacteria, such as Escherichia been schematically drawn as intracellular circles. several PVC bacteria are required to ma- coli, the capture of proteins involves their The horizontal black bars indicate loss of MC pro- nipulate their complex endomembrane sys- transport through the cell membrane via teins. NM, nuclear membrane. tems. In G. obscuriglobus, the nuclear double pore-like structures, the data obtained membrane segregates the nucleoid from by Fuerst and colleagues strongly suggest a large fraction of ribosomes (3). This sup- that, as in eukaryotes, proteins can be in- Of course, endocytosis by G. obscuriglobus poses the existence of nuclear pores allowing ternalized by G. obscuriglobus via the could be considered another curiosity the transport of messenger RNA from the “ ” formation of endosome-like structures. of this bacterium with no link to true nucleus to these ribosomes. By analogy with They show that imported proteins (GFP, endocytosis, a hallmark of eukaryotes. nucleoporin (one of the eukaryotic MC Ig, or streptavidin) accumulate and are However, a recent work, by Devos and col- proteins), it is possible that some of the later degraded in the ribosome-free leagues from the European Molecular Bi- MC proteins encoded by G. obscuriglobus region of the G. obscuriglobus cytoplasm ology Laboratory (Heidelberg), nicely could be used to build these pores. (the so-called paryphoplasm). Ultracen- complements the results of Fuerst and col- The analogies between the membrane- trifugation of a cell extract shows that in- leagues (5). Using sensitive structure- trafficking systems of PVC bacteria and ternalized proteins copurify with a cell based in silico search approaches, these au- Eukarya, both at the cytological and pro- fraction containing both membrane debris thors screened all available archaeal and tein structure levels, are thus strikingly and vesicles. These vesicles are visible bacterial proteomes for the presence of evident. It now seems impossible to ignore within the paryphoplasm by electron mi- structural analogs of eukaryotic membrane these data when discussing the origin of croscopy, and some of them appear to coat (MC) proteins. In eukaryotes, MC the eukaryotic nucleus. A major objective be formed by invagination of the cyto- proteins are involved in both vesicle- fi of future research should now be to de- plasmic membrane. Immunogold labeling traf cking systems and in the formation of termine whether bacterial MC proteins revealed the presence of internalized the nuclear pore. Strikingly, Devos and col- are only structural analogs of eukaryotic GFP lining these vesicles and membrane leagues only detected eukaryotic MC-like ones (a case of convergent evolution) or invagination, suggesting that vesicle for- proteins (from 5 to 16 copies) in the pro- whether instead they are homologous. mation is indeed linked to protein uptake. teomes of G. obscuriglobus and several Although genetic experiments lack de- related bacteria of the Planctomycetes- finitive proof (there are presently no -Chlamydiae (PVC) su- Author contributions: P.F. and S.G. wrote the paper. available genetic tools for Planctomyce- perphylum. This superphylum groups The authors declare no conflict of interest. tales), these data strongly suggest that this Planctomycetes, Verrucomicrobia, Chlamy- See companion paper 10.1073/pnas.1001085107. is indeed a demonstration of endocytosis diae, Lentispherae, and Poribacteria based 1To whom correspondence should be addressed. E-mail: in a bacterium. on rRNA phylogeny (2). Importantly, it [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1007720107 PNAS Early Edition | 1of2 Downloaded by guest on October 1, 2021 This cannot be tested through sequence Thaumarchaeota, because these Archaea based on the slowest evolving positions similarity (even between eukaryotic are mostly mesophiles and exhibit (8). If this is so, only two losses of MC pro- MC proteins), because these proteins eukaryotic features that are missing in teins (one in Bacteria after the divergence evolve too rapidly at the sequence level. other archaeal phyla (6). However, sym- of the PVC lineage and another in the However, MC proteins have retained their biotic hypotheses for the origin of Eukarya branch leading to Archaea) would be core architecture during evolution, which remain difficult to understand in terms necessary to explain the present distribu- is formed by a unique combination of of known biological mechanisms. For ex- tion of MC proteins in the three domains two protein domains composed of long ample, they imply a specific association of life (Fig. 1). stretches of β-strands in the N terminus between a bacterium and an archaeon for If the LUCA already harbored MC pro- and α-helices in the C terminus. The re- which there are no examples in nature, teins, it was probably compartmentalized. spective lengths of the two domains, as and assume a very unlikely process where This idea can appear odd to many biolo- well as the respective numbers of β-strands all of the of the bacterial host coding gists who use to think of the LUCA and and α-helices in each domain, are strik- for informational proteins would have all its contemporaries as very primitive en- ingly similar in PVC bacteria and Eukarya been replaced by those of the archaeal tities. However, the formation of vesicles (5). In their paper, Devos and colleagues symbiont. Alternatively, under the evolu- and membrane manipulation may be very thus argue in favor of homology instead tionary scenarios where modern Eukarya ancient features of life. For instance, eu- of convergence. The final proof will originated from an ancestral proto- karyotic RNA viruses encode proteins that probably require solving the structures of eukaryotic lineage [urkaryote, sensu can manipulate the host ER to produce several MCs from Bacteria and Eukarya viral factories surrounded by membranes and reconstructing (if possible) the struc- (9), suggesting, by analogy, that even an- ture of the ancestral MC protein in Endocytosis of proteins cient cells with RNA genomes could have each domain. Preliminary results have might well be an had such capacity and therefore already be nevertheless already provided important compartmentalized. If MC proteins were information, suggesting in fact an ancient ancient trait. already around at the time of the LUCA, origin of these proteins in both PVC the ancient biosphere might have been bacteria and in Eukarya, because several more diversified than usually suspected, copies of MC proteins were probably Woese and Fox (7)], two hypotheses can with various lineages of compartmentalized already present in their respective last be proposed. In the first one, MC proteins cells, some of them with nuclei (which could common ancestors (5). appeared in the lineage leading to PVC be named synkaryotes) and others with- If we assume that bacterial and eukary- bacteria and were subsequently trans- out (akaryotes), thriving in various envi- otic MC proteins have a common origin, ferred to the lineage leading to Eukarya, ronments. Endocytosis of proteins might how can this information be fitted with or the reverse. In the second one, MC well be an ancient trait that was lost in current theories on the origin of eukar- proteins would have already been present bacteria with rigid cell walls. Although PVC yotes? Three scenarios can be imagined. in the last universal common ancestor bacteria are bona fide members of the First, in models suggesting that eukaryotes (LUCA) and were inherited in Eukarya bacterial domain, they might therefore have originated from the association of a bacte- and PVC bacteria, whereas they were conserved some ancestral features in terms rium and an archaeon (symbiotic hy- lost in all other and in Ar- of cellular structure and function that potheses), it appears reasonable now, in chaea. The second hypothesis (favored by open up new avenues of thinking about the light of the data by Fuerst and Devos and Fuerst and Webb) is supported by the fact nature of our cellular ancestors. Further colleagues, to suggest a compartmental- that the loss of MC proteins appears to exploration of microbial diversity will most ized PVC bacterium as the host. The ar- have occurred recurrently in the PVC su- likely bring surprises. Other compartmen- chaeal symbiont would have lost its cell perphylum, whereas lateral transfer talized cells could in fact exist among the membrane, and its chromosome would of MC proteins to other bacteria or Eu- vast numbers of still uncultivated archaeal have been surrounded by a nuclear mem- karya (and vice versa) has not been ob- and bacterial lineages. In any case, the re- brane coming from the ICM of the bac- served. Moreover, Brochier and Philippe sults of Fuerst and Devos and colleagues terial host. In this scenario, it is tempting reported several years ago that Plancto- remind us that we should definitely stop to suggest that the archaeal symbiont mycetales are the earliest diverging bac- thinking of bacteria in terms of simple belonged to the newly recognized phylum terial lineage in a 16S rRNA phylogeny “lower” organisms.

1. Fuerst JA, Webb RI (1991) Membrane-bounded nucleoid bus. Proc Natl Acad Sci USA, 10.1073/pnas.100108 7. Woese CR, Fox GE (1977) Phylogenetic structure of the in the eubacterium Gemmata obscuriglobus. Proc Natl 5107. prokaryotic domain: The primary kingdoms. Proc Natl – Acad Sci USA 88:8184 8188. 5. Santarella-Mellwig R, et al. (2010) The compartmentalized Acad Sci USA 74:5088–5090. 2. Wagner M, Horn M (2006) The Planctomycetes, Verru- bacteria of the Planctomycetes-Verrucomicrobia-Chlamydiae 8. Brochier C, Philippe H (2002) Phylogeny: A non- comicrobia, Chlamydiae and sister phyla comprise a superphylum have membrane coat-like proteins. PLoS Biol 8: hyperthermophilic ancestor for bacteria. Nature 417: superphylum with biotechnological and medical rele- e10000281. 244. vance. Curr Opin Biotechnol 17:241–249. 6. Brochier-Armanet C, Gribaldo S, Forterre P (2008) A 9. Netherton C, Moffat K, Brooks E, Wileman T (2007) A 3. Fuerst JA (2005) Intracellular compartmentation in Planc- DNA topoisomerase IB in Thaumarchaeota testifies for guide to viral inclusions, membrane rearrangements, fac- tomycetes. Annu Rev Microbiol 59:299–328. 4. Lonhienne TGA, et al. (2010) Endocytosis-like pro- the presence of this enzyme in the last common ances- tories, and viroplasm produced during virus replication. tein uptake in the bacterium Gemmata obscuriglo- tor of Archaea and Eucarya. Biol Direct 3:54. Adv Virus Res 70:101–182.

2of2 | www.pnas.org/cgi/doi/10.1073/pnas.1007720107 Forterre and Gribaldo Downloaded by guest on October 1, 2021