Cavalier-Smith (2009) Predación y origen de las células eucariontes. Una perspectiva coevolutiva

The International Journal of Biochemistry & Cell Biology 41 (2009) 307322

Contents lists available at ScienceDirect

The International Journal of Biochemistry& Cell Biology

journa l homepage: www.e lsev ier .com/ locate /b ioce l

Review

Predati

T. CavalieDepartment of

a r t i c l

Article history:Available onlin

Keywords:RetrotranslocaEndomembranCytoskeletonNucleusQuantum evolutionPhagocytosis unavoidably disruptingbacterial division; recovery entailed the evolutionof thenucleus andmitotic cycle;

(3) the symbiogenetic origin of mitochondria immediately followed the perfection of phagotrophy andintracellular digestion, contributing greater energy efciency and group II introns as precursors of spliceo-somal introns. Eukaryotes plus their archaebacterial sisters form the clade neomura, which evolved froma radically modied derivative of an actinobacterial posibacterium that had replaced the ancestral eubac-

Contents

1. Dating2. Eukar3. Neom4. The n5. Transi6. The n7. Prereq8. Intern9. Pseud10. DNA i11. Peroxi12. Phago13. Cilia, a14. The bo15. Concl

Refere

Abbreviatioproteins; SR, s

E-mail add

1357-2725/$ doi:10.1016/j.bterialmurein peptidoglycan byN-linked glycoproteins, radicallymodied its DNA-handling enzymes, andevolved cotranslational protein secretion, but not the isoprenoid-ether lipids of archaebacteria. I focus onthisphylogeneticbackgroundandonexplaininghowin response tonovelphagotrophic selectivepressuresand ensuing genome internalisation this prekaryote evolved efcient digestion of prey proteins by retro-translocation and 26S proteasomes, then internal digestion by phagocytosis, lysosomes, and peroxisomes,and eukaryotic vesicle trafcking and intracellular compartmentation.

2008 Elsevier Ltd. All rights reserved.

eukaryote origins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308yote phylogeny and the properties of the earliest eukaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308uran and -proteobacterial precursors of eukaryotes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311eomuran revolution and immediate archaebacteria/eukaryote divergence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311tion analysis resolves key difcult problems of bacterial phylogeny . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312eomuran ancestor of eukaryotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312uisites for evolving phagotrophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313al digestion rst by retrotranslocation and 26S proteasomes? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315opodia, prey uptake, and membrane recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315nternalisation, copII and the origin of permanent endomembranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316somes as endomembrane digestive differentiations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317trophy and novel protein targeting enabled symbiogenetic organelle additions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318novel compartment not delimited by membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319tched, piecemeal nature and rapidity of megaevolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319

usions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319nces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320

ns: LGT, lateral gene transfer; ERAD, ER-associated degradation ofignal-recognition-particle receptor; NE, nuclear envelope.ress: [email protected].

see front matter 2008 Elsevier Ltd. All rights reserved.iocel.2008.10.002on and eukaryote cell origins: A coevolutionary perspective

r-SmithZoology, University of Oxford, South Parks Road, Oxford, OX1 3PS, UK

e i n f o

e 18 October 2008

tione system

a b s t r a c t

Cells are of only two kinds: bacteria, with DNA segregated by surface membrane motors, dating back3.5Gy; and eukaryotes, which evolved from bacteria, possibly as recently as 800850My ago. The lastcommon ancestor of eukaryoteswas a sexual phagotrophic protozoanwithmitochondria, one or two cen-trioles and cilia. Conversion of bacteria ( =prokaryotes) into a eukaryote involved 60 major innovations.Numerous contradictory ideas about eukaryogenesis fail to explain fundamental features of eukaryotic cellbiology or conictwithphylogeny.Data are best explainedby the intracellular coevolutionary theory,withthree basic tenets: (1) the eukaryotic cytoskeleton and endomembrane systemoriginated through cooper-ativelyenabling theevolutionofphagotrophy; (2)phagocytosis internalisedDNA-membraneattachments,

308 T. Cavalier-Smith / The International Journal of Biochemistry & Cell Biology 41 (2009) 307322

The origin of eukaryotes (eukaryogenesis) was the largest reor-ganization of cell structure ever. To explain it we must answer sixquestions: (1) When did they evolve? (2) What was the natureof the lasteukaryote)?ical mechaninvolved? (seminal chnovel genesdramatic geet al., 2005;

Fortunatstanding euElements oin ultrastruand theoretbacteria areof eukaryotof many imics that endunderstood

Eukaryothe historybrane, skeleendoskeletoof feeding wthat eukaryexplain eacnetic sceneinterconnecand continumore detailgins associathese wereinternal digproteasome

I use thprokaryotebacteria an(Stanier, 19also archaediscussion i

1. Dating e

The oldeMelanocyrilpreserved sare almostpseudopodthey are arcamoebae (ptionable. Nosuggested tno marineMoreover, tMost likelyourished bPossibly, Crthe globe i635My agocondentlydate the mrst are red

mals date from 550My, but most phyla only appeared after530My during the Cambrian explosion (most animals, some proto-zoa, e.g. Foraminifera, Radiozoa, green algae) or substantially later

lant800of t

vativer-Sm006)markhoto

ntolovincite ofas eacterstic iotic (as o

moletimenitutionaticaicallynoveral Rrgent funh sloe mocreasthat amos

in rRg seqing rerpolardsing ning fnd H

te un

aryootes

nferthe rlt, were voa, anhaveitocher-Smst mandce p

has rwith. Allre atmitoubsecommon ancestor of all eukaryotes (the cenancestral(3) What were its ancestors? (4) What were the phys-isms of the changes? (5) What were the major steps6) What triggered such exceptionally disruptive butanges? This cannot be done in detail here, as 5000originated during the origin of eukaryotes, the most

netic explosion in history (Makarova et al., 2005; YangCavalier-Smith, 2007a), and I have only 9000 words.ely, not all genes are equally fundamental, and under-karyogenesis is not nearly as difcult as oftenmade out.f a sound explanation already exist through advancescture, molecular and cell biology, genetics, phylogeny,ical analyses over 40 years. Persisting problems are thatso small that their cell biology lags greatly behind thates and we do not know the functions or 3D structureportant structural proteins; lipid membrane dynam-ow cells with form and integrity are also insufciently.genesis poses the problem how and why, just once inof life, cells radically spatially reorganized their mem-ton, and chromosomal relationships. The origin of ann, membrane budding and fusion, and a novel modeere fundamental. Table 1 summarises 60 innovations

ogenesis theories must explain. Rather than trying toh in detail I focus on ve things: (1) setting the phyloge-; (2) showing how the different changes were logicallyted; (3) emphasizing that twin themes of disruptionity underlie a coherent explanation; (4) explaining inthan hitherto the earliest steps in endomembrane ori-ted with the evolution of phagotrophy; (5) arguing thatprobably preceded by a simpler predatory stage withestion mediated by retrotranslocation and improveds.e classical term bacteria as a simpler synonym for(Cavalier-Smith, 2007b), i.e. embracing both classicald cyanobacteria, which prior to invention of that name74) were called blue-green algae or Cyanophyta, andbacteria, renamed archae (Woese et al., 1990). See alson Cavalier-Smith, 1991a,b.

ukaryote origins

st indubitably eukaryotic fossils are vase-shaped, e.g.lium; their oldest secure date for numerous well-pecimens is 760My ago (Porter and Knoll, 2000). Theycertainly shells of testate amoebae constructed by

ial activity that never occurs in bacteria. Claims thatellinid amoebae (phylum Amoebozoa) and euglyphidhylum Cercozoa) (Porter et al., 2003) are highly ques-ne are condently morphologically euglyphids (thoseo be could be another group with agglutinated shells);arcellinids are known, yet these fossils are all marine.hey apparently became extinct before the Phanerozoic.they were an extinct group of testate amoebae thatefore Foraminifera evolved (Cavalier-Smith, in press).yogenian glaciations that largely or entirely coveredn several kilometres of ice periodically from 710 toextinguished them (snowball earth). The only fossilsassignable to a modern eukaryotic phylum all post-

elting of snowball earth (Cavalier-Smith, 2006a). Thealgae (Rhodophyta) about 600My old. Earliest ani-

(land pfrom devoidconser(Cavaliet al., 2

Ingenic ppalaeoNo conestimaas oldfrom boptimieukarybe 2.3

Nologicalof magduplicasystemepisodfast asancestIII divedistincto mucshift thfold inactinstions (AoccurtreatinAveragful intbackwprovidyet givRoger atick-ra

2. Eukeukary

To ilocatedifcuotes wprotozellatesthat m(Cavalithe mo1975)Sequen2007)gruent(Fig. 1)thatwetion inwere ss). A few poorly dated Melanocyrillium-like fossils dateMy, but relatively numerous deposits dated 850My arehem or anything denitely eukaryotic. Thus the moste estimate of the age of eukaryotes is 850800My agoith, 2002a,c). That they are as old as bacteria (Kurlandis disproved by the fossil evidence.ed contrast there is unequivocal evidence for oxy-synthetic prokaryotes as early as 2.45Gy ago; mostgists think cyanobacteria arose earlier, 2.92.7Gy ago.ng evidence shows life before 3.5Gy, currently the bestwhen life began. Thus bacteria are probably four timesukaryotes, making it certain that eukaryotes evolvedia, not the reverse (Cavalier-Smith, 2006a). Even weredentications of a few meagre fossils 1.5Gy ago asJavauxet al., 2001) justied (I thinknot), bacteriawouldld as eukaryotes.cular biological clock ticks constantly throughout geo-. Proteins evolve at rates differing over many orders

de. As new proteins all evolve from old ones by gene, ratesmust change dramatically over time. They changelly among different branches of the tree and also. When new paralogues arise, evolution is initially veryl functions are acquired, e.g. during eukaryogenesis theNA polymerase evolved into RNA polymerases I, II andtly adapted for transcribing rRNA, mRNA and tRNA; asctions became perfected initial fast evolution gave waywer more trivial divergence. The bigger the functionalre dramatic the transient initial acceleration, a >10,000e in rate being likely for molecules like tubulins androse from bacterial FtsZ and MreB by multiple duplica-et al., 2004; Erickson, 2007). Similar transient increasesNA. Many evolutionary misinterpretations stem fromuence divergence as clock-like (Cavalier-Smith, 2002c).ates of change in local parts of the tree allows use-ation between known fossil dates, but extrapolatingbeyond fossil calibration points is extremely unreliable,o useful information beyond what fossils directly say,alse condence in inferences (Graur and Martin, 2004;ug, 2006). It is scientically unsound to use a clock ofknown by a factor of 10,000.

te phylogeny and the properties of the earliest

the nature of the rst eukaryote rigorously we mustoot of the eukaryotic tree condently. This has beenith many false trails. For a century, the rst eukary-ariously postulated to be algae, fungi or protozoa; ifaerobic or aerobic amoebae, agellates or amoeboag-each been considered primitive. It is well establishedondria evolved by symbiogenetic cell enslavementith, 2006b, 2007a) from -proteobacteria, which have

itochondrion-like respiratory chain (John and Whatley,include purple non-sulphur photosynthetic bacteria.hylogeny (Keeling et al., 2005; Rodrguez-Ezpeleta et al.,evealed major clades of the eukaryote tree that are con-much ultrastructural data and helps position its root

known groups of anaerobic eukaryotes had ancestorsleast facultatively aerobicwith oxidative phosphoryla-chondria; in various protozoa and fungi mitochondriaquently polyphyletically modied as hydrogenosomes

T. Cavalier-Smith / The International Journal of Biochemistry & Cell Biology 41 (2009) 307322 309

Table 1Sixty groups of key innovations in the origin of the eukaryotic cell and cell cycle.

1. Phagocytosis (Cavalier-Smith, 1987c, 2002e)2. Actin and ac3. Pinocytosis4. SNAREs (Ko5. 3+ myosins6. Tubulin evo rotub7. 13 kinesins (8. Dynein for s rt (Cav9. Exocytosis a10. Arf1 and Sa11. Rab GTPase12. Rho GTPas13. Ras GTPase14. Centrioles15. Cilia (nine16. Clathrin co17. COPI vesicl18. COPII coats19. Peroxisom20. Endosome21. Lysosomes22. Golgi comp23. Sphingolip24. Trans-Golg25. Centrin (Ca26. Phosphatid27. Calmodulin28. Delrin prot29. Ubiquitin a30. 26S protea31. SR of sign32. Plasma me33. Cell cycle r34. Massive ex35. Formins fo36. AAA lysine37. Mcm replic38. Internalisa39. Cell divisio40. 4-histone n41. Separate RN42. Four-modu43. Chromatin44. Centromer45. Meiosis an46. Telomerase47. Post-transc48. Proteinace49. Nuclear lam50. Nuclear po51. Nucleolus a52. Ran GTP/G53. Karyopher54. Ribosome55. mRNA cap56. polyA tran57. Nuclear en58. Mitochond59. Spliceosom60. Nonsense-

The referenceslike chloroplasmarine (Cavalithe exclusionissues. These apopulation gen

or mitosomand ribosomery that ev(Embley, 20

The ideadonedwhentin-related proteins (Arps 1,2, 3) (Cavalier-Smith, 2006a)(Field et al., 2007)umandou et al., 2007; Yoshizawa et al., 2006)(Odronitz and Kollmar, 2007; Richards and Cavalier-Smith, 2005)lution from FtsZ and subsequent triplication: for centrosome and and for micCavalier-Smith, 2006a; Wickstead and Gull, 2006))liding surface-attached astral microtubules and related midasin for ribosome expo

nd exocysts (Koumandou et al., 2007)r1 GTPases (Dacks and Field, 2004)s (Jkely, 2004)es (Jkely, 2004)s (Jkely, 2004)and , , and tubulins (Beisson and Wright, 2003; Feldman et al., 2007; Keller et al., 200doublets, dynein arms and centre pair spokes, ciliary transport) (Jkely and Arendt, 2006ats and adaptinse coats (Stagg et al., 2008)(Grkan et al., 2006; Stagg et al., 2008)

es (Tabak et al., 2006)s (early, late and multivesicular bodies) (Leung et al., 2008)

lexid synthesisi network++ contractility) (Salisbury, 2007)ylinositol/kinase signaling, Ca++ and inositol triphosphate second messenger systems.ein extrusion channel for ER-associated degradation (ERAD)nd polyubiquitin labelling systemsomes with 19S regulatory subunit (Cavalier-Smith, 2006c)al-recognition-particle receptor (Schwartz and Blobel, 2003; Schwartz et al., 2006)mbrane phosphatidylinositol anchor proteins (Oriol et al., 2002)esetting by anaphase proteolysis (Cavalier-Smith, 2005, 2006a,c; de Lichtenberg et al., 20pansion of serine/threonine kinase controls (Shiu and Bleecker, 2001)r positioning actomyosin (Grunt et al., 2008; Higgs and Peterson, 2005)biosynthetic pathway (Cavalier-Smith, 1987b; Sumathi et al., 2006)ation licensing system controlled by cyclins (de Lichtenberg et al., 2007; Krylov et al., 20tion of DNA attachment sites as protoNE/roughER (Cavalier-Smith, 1975, 1987b, 1991a, 20n by actomyosin not FtsZ (Cavalier-Smith, 1975, 1981, 1987b, 1992, 2002e, 2006a)ucleosomesA polymerases I, II and IIIle 30-subunit mediator complex regulating polII transcription (Bourbon, 2008)condensation cycle: histone phosphorylation, methylation, acetylation; heterochromatines/kinetochores (CenpA from core histone) for attaching DNA to microtubules (Cavalier-Sd synaptonemal complex (Cavalier-Smith, 1981, 1987a, 1995a, 2002b,d,e)s and telomeres (Cavalier-Smith, 1981, 1987b, 1988, 2002e)riptional gene silencing, dicer and argonaut nucleases (Molnar et al., 2007)ous interphase nuclear matrix with bound DNA-topoisomerase II and its ability to reorgaina (Cavalier-Smith, 1982a, 2005)

re complexes (NPCs) (Bapteste et al., 2005; Cavalier-Smith, 2004b, submitted for publicand more complex rRNA processing (e.g. 5.8S rRNA) (Cavalier-Smith, 2002e)DP cycle for directionality of NE export/import (Bapteste et al., 2005; Jkely, 2004; Mansins (Bapteste et al., 2005; Mans et al., 2004)subunit nuclear export machinery (Bapteste et al., 2005; Mans et al., 2004)ping and export machinery (Bapteste et al., 2005; Cavalier-Smith, 1981; Mans et al., 2004scription termination system (Cavalier-Smith, 1981)velope fusion and syngamy (Cavalier-Smith, 1995a, submitted for publication)ria (Cavalier-Smith, 2006b, 2007a)es and spliceosomal introns (Cavalier-Smith, 1981, 1985, 1987b, 1991c, 1993)mediated mRNA decay (Maquat, 2004; Cavalier-Smith, submitted for publication)

discuss their evolution or molecular basis in more detail. This list includes only charactts or intermediate laments. It is probably not exhaustive. Another possible one is contraer-Smith, in press), eukaryotes originated in soil or freshwater. Note that only the last throf the 57 primary, purely autogenous, and often far more complex, innovations, too manre explaining coordinated radical structural change in almost every non-metabolic aspeetics or sequence phylogeny, is required to understand the transition.

es, double membrane organelles typically lacking DNAes but retaining the Tom/Tim protein-import machin-

olved in the last common ancestor of all eukaryotes06; Gill et al., 2007; Stechmann et al., 2006).of aphotosynthetic ancestorof all eukaryoteswasaban-the symbiogeneticoriginof chloroplastsbypermanent

intracellula1905) was pof the glaucpeptidoglycbiciliate hosalgae aroseules xing it to cell surface (Cavalier-Smith, 1987b, 2002e, 2006a)

alier-Smith, 2006a)5; Marshall, 2007; Zamora and Marshall, 2005); Mitchell, 2007)

07) and mitosis

03; Nasmyth, 1995)02e)

(Cavalier-Smith, 2005)mith, 1987b, 2002e, 2006a)

nize as mitotic chromosome cores (Cavalier-Smith, 1982a)

tion; Devos et al., 2004, 2006; Mans et al., 2004)

et al., 2004)

)

ers that evolved before the cenancestor and excludes later additionsctile vacuoles, if, contrary to the deduction that they were originallyee innovations depended on symbiogenesis; by focusing on them toy speculators have put the cart before the horse and evaded the keyct of the cell. Molecular cell biology, not metabolic biochemistry or

r enslavement of a cyanobacterium (Mereschkovsky,roved in the late 1970s by discovering that chloroplastsophyte alga Cyanophora retain a typically eubacterialanwall. Oneprimary cyanobacterial symbiogenesis in at created the plant kingdom, whereas other eukaryoticby secondary symbiogenesis: enslavement of a pre-


Fig. 1. The treeof life, emphasizing the fundamental differencebetween theancestral negibacteria,with twoboundingmembranes, and theadvancedunibacteria (Posibacteriaplus Archaebacteria), which evolved from them by losing the outer membrane before giving rise to eukaryotes. Marked changes in membrane and wall chemistry were ofkey importance in bacterial evolution and in preparation for eukaryogenesis. The neomuran revolution, when core histones evolved and N-linked glycoproteins that areglycosylated during translation replaced the peptidoglycan murein, created a fundamental division between the ancestral eubacteria (Negibacteria plus Posibacteria), whichare a grade of organization not a taxon, and the derived clade neomura (Archaebacteria plus Eukaryota), which is also not a taxon. Symbiogenetic enslavements of twonegibacteria (-proteobacteria, cyanobacteria) to form mitochondria and chloroplasts, are shown by thin grey arrows; their organellar envelope outer membranes evolvedfrom thenegibacterial outermembranes, not from the food vacuolesmembranewhichwas lost, liberating them into the cytosol (Cavalier-Smith, 1982c, 1983, 2000, 2006a,b,c).A secondary symbiogenesis generated all chromophyte chloroplasts from an enslaved red alga (thick grey arrow), adding two extra membranes across which proteins haveto be targeted; the food vacuole membrane was retained and signal peptides added upstream of transit peptides for crossing it; additionally the ERAD extrusion channelduplicated and was recruited for protein import across the extra periplastid membrane (the former red algal plasma membrane) (Maier, in press). Two secondary symbiosesnot shown by arrows implanted green algal chloroplasts into chlorarachnean Cercozoa and euglenoid excavate agellates (asterisks). For more on the bacterial tree and itsrooting see Cavalier-Smith (2006a, 2006c); for more on the eukaryote tree see Cavalier-Smith (2003b, submitted for publication, in press).


existing eukaryotic plant cell by a phylogenetically distinct host,which adds extra membranes (requiring additional translocationmachinery) and sometimes even a foreign nucleus (Cavalier-Smith,1995b, 1999protein-targtae (Glaucogreen plantcyanobactesecondary sbiciliate hoall chloropous heterotphotosynthpseudofungtargeting mplus compeduplicate hosecondaryplasts. Thregreen algalone small dacquired ha

The rooteukaryote cterium to mchromalveoas unique pso can a unanaerobic dnot lie betwtrees subjecroot is thoualternativeAmoebozoaless likely b

Whichevcenancestowas unquessarily followtree belowthe presencdeepest bifuand is not mally possibitmight havand syntropseen in planorigin of plaoteswere inthe rst pla

3. Neomureukaryotes

Eukaryophagotrophton (Cavalieconverted iArchaebactforming a cArchaebactare their siearlier arguplaces arch

not ancestral to them (Yutin et al., 2008). Thus >20 features sharedby both groups but absent from eubacteria (e.g. N-linked glyco-proteins, more complex RNA polymerases, core histones) are not

callyommionly froc). Caal (Sncritut euous aia anandrly aleriumll chauniqtersag

008)ncesagearyocamolecuste etzymly lo0002002

neobact

econcest

eobat inerialclade

andd thibact200ike nbyubac

raneavaliologiandthatactemicualiere evacti

rols.dedra, aical pith s

myce, 2000, 2002a, 2003a). Demonstration of shared Toc/Ticeting machinery in plastids throughout kingdom Plan-phyta, Rhodophyta (red algae) and Viridiplantae, i.e.s) proved a common origin for Plantae by a uniquerial enslavement. The most important of four knownymbiogeneses was the enslavement of a red alga by ast to form chromalveolates, a major clade comprisinghyll-c containing algae (chromophytes) and numer-rophic descendants whose ancestors secondarily lostesis, e.g. ciliate protozoa, malaria parasites, oomycetei, (Cavalier-Smith, 1999, 2004a). Common protein-echanisms in chromalveolates (Cavalier-Smith, 2003a)lling evidence of sharedplastid protein replacements byst genes (Keeling, in press) rmly establishes the single

symbiogenetic acquisition of chromalveolate chloro-e independent secondary symbiogeneses implantedplastids into chlorarachnean Cercozoa, euglenoids, andinoagellate clade. Another small dinoagellate cladeptophyte plastids by tertiary symbiogenesis.of the eukaryote tree cannot be within Plantae becauseells must have existed before one enslaved a cyanobac-ake the rst plant. Likewise, the root cannot be withinlates, as the enslaved red alga evolved after Plantae. Justlastid enslavement excludes the root from these clades,ique lateral gene transfer (LGT); several shared by theiplomonads and parabasalids show that the root can-een them (Andersson et al., 2005), where early rRNAt to gross long-branch artefacts falsely put them. Theght to be between unikonts and bikonts (Fig. 1); twopositions, within unikonts (between opisthokonts and) or within bikonts (between excavates and others), areut not completely unreasonable.er is correct, we can conclude that the eukaryote

r had one or two cilia and aerobic mitochondria, andtionably a phagotroph that ate other cells. This neces-s from the absence of any branch on the eukaryoticthe last common ancestor of animals and plants ande of these homologous characters on both sides of thercation. As secondary loss of photosynthesis does occurrmly excluded even for the ancestral unikont, it is for-le that the eukaryotic cenancestorwasphotosynthetic;e already enslaved aplastid, inwhich case coadaptationhy amongplastids,mitochondria andperoxisomesnowts dates from it (Cavalier-Smith, 1987c). However, laterstids in an early bikont is more likely; arguably eukary-itially benthic heterotrophs, plastids only evolvingwithnkton (Cavalier-Smith, in press).

an and -proteobacterial precursors of

te cells are evolutionary chimaeras of an ancestrallyic cell with nucleus, endomembranes, and endoskele-r-Smith, 2002e) and an enslaved -proteobacteriumnto a mitochondrion (Fig. 1) (Cavalier-Smith, 2006b).eria are closely related to the eukaryote host (togetherlade called neomura: Cavalier-Smith, 1987b, 2002c).

eria are not direct ancestors of eukaryotes. Instead theysters (Cavalier-Smith, 2002c, 2006a,c). Besides thosements, comprehensive 136-gene analysis convincinglyaebacteria as a holophyletic clade, sister to eukaryotes

specitheir crevolutarguab2006a,bacterisuch uing abonumerchondrMartinfor neaeubactwith abut nocharacuniqueet al., 2in the aote lineby eukthat beface m(Bapteand enprobablost 1Smith,

4. Thearchae

To rtical an-protpreseneubactare asurfacededuceria. Pos1987b,brane lmerelyOther emembcult (Cmorphsporesnositolother bria (Firsee Cavnot hav(unlikeand ste

Forneomucylindrteria, wActinoarchaebacterial, but neomuran characters evolved byon ancestor, which itself arose during the neomurancaused by loss of the ancestral eubacterial murein wall,m an actinobacterium (Cavalier-Smith, 1987b, 2002c,lling the host for mitochondrial enslavement archae-earcy, 1992; Yutin et al., 2008) is therefore wrong;ical terminology has extremely distorted much think-karyogenesis. Archaebacterial holophyly disproves therchaebacteria-as-host hypotheses of the origin ofmito-d eukaryotes, e.g. Lpez-Garca and Moreira (2006);Mller (1998), which are also all explanatorily emptyl eukaryotic characters (Table 1). The ancestorwas not aeither, but an extinctmissing link anearly neomuran

racters shared by both archaebacteria and eukaryotes,uely archaebacterial properties. Purely archaebacterial(isoprenoid-ether lipids, archaeosine modied rRNAs,ella, duplicate versions of DNA polymerase B (Rogozin) aremostparsimoniously interpretedashavingevolvedtral archaebacterium after it diverged from the prekary-(Cavalier-Smith, 2002c, 2006c). Moreover, genes sharedtes and eubacteria, but not archaebacteria (e.g. MreBe actin (Cavalier-Smith, 2002e, 2006a), eubacterial sur-les that became nuclear envelope lamin B receptorsal., 2005), cytochromeP450ancestors of ER respiration,es making acyl ester phospholipids and sterols), werest by the ancestral archaebacterium, which apparentlygenes when adapting to hyperthermophily (Cavalier-c, 2006c, 2007b).

muran revolution and immediateeria/eukaryote divergence

struct the nature of the transient early neomuran ver-or of eukaryotes (i.e. excluding what came from thecterium), we must consider not only what genes werethe ancestral archaebacterium, but also those of theancestors of neomura. When arguing that neomuraderived from eubacteria by radical changes in cellmode of DNA supercoiling, Cavalier-Smith (1987b)

at their closest eubacterial relatives were posibacte-eria (Actinobacteria plus Endobacteria: Cavalier-Smith,2c) are the only eubacteria with one bounding mem-eomura and thus the potential to evolve into neomurareplacing the murein wall by N-linked glycoproteins.teria (collectively Negibacteria) all have an extra outerthat would have to be lost, which is extremely dif-er-Smith, 1987b). I proposed that actinobacteria, oftencally complex aerobes, with cell differentiation, aerialvery diverse lipids, including invariably phosphatidyli-was crucial for eukaryote origins but absent from all

ria, were most likely ancestral to neomura. Endobacte-tes in its confused modern sense minus Eurybacteria:-Smith, 2006c) have specialised endospores that couldolved into protozoan cysts and fungal and plant sporesnobacterial exospores) and lack phosphatidylinositol

ucing that actinobacteria are the closest relatives ofkey consideration was the evolution of proteasomes,roteolytic chambers found in neomura and actinobac-impler precursors in other bacteria (Gille et al., 2003).te proteasomes were considered ancestral to those of


neomura, and derived from HslV of other eubacteria (Cavalier-Smith, 2006c). Two recent ndings complicate that inference.First, a second eubacterial proteasome precursor was discovered:Anbu (Valamore similearlier cons20S proteaprobably antype proteametaproteo2007); unleteasome geactinobactenetic positia separate ptain) of no snot invalidaprobably annger actinparate char

Among tthe evolutioabsent fromteria). Thesole) reasonof eukaryotactinobactesimply thement of arc-proteoba(Martin andand unlikelplace. Far ma common

Shared ncharactersmurein waglycoproteiby core histFtsZ:Gardineubacterialarguably loones in theall novel feare explicabthermophilcompared wno sense (C

5. Transitibacterial p

Criticalacters canstates. Sucing majorgenerally caSmith, 199prokaryoticparalogue rin eubacterand neomu2006c). Itbut are an

DNA/ribosome-related paralogue trees (Gogarten et al., 1989;Iwabe et al., 1989) misrooted the tree because of long-branch arte-facts. It supports actinobacteria as the closest relatives of neomura,

onglynegit al.,el daons.oban Chum Cbactece allderukar2006nec

er wal trra, b. Noreliatrueoutardethe

reco008)d fotheres anetabasseslutiomislougacte

tes (eriaCavainserbranet alin Ru

neo

s thbacte, butter coN-linatidNA-hhet

s (aces. Aly peg intnt ean

rd mer line tos and Bourne, 2008), three-dimensionally somewhatar to both 20S proteasomal subunits than is HslV,idered the ancestor of actinobacterial/archaebacterialsomes (Cavalier-Smith, 2006c). Anbu, not HslV, wascestral to 20S proteasomes. Second, actinobacterial-some genes occur in environmental metagenomes andmes dominated by Leptospirillum negibacteria (De Mot,ss they come from contaminating actinobacteria, pro-nes probably underwent lateral gene transfer fromria to Leptospirillum, a lineage of uncertain phyloge-on, likely a very deep-branching proteobacterium (nothylum, as many assume because its relatives are uncer-ignicance for eukaryote origins. Possibility of LGT doeste the conclusion that actinobacterial proteasomes arecestral to those of neomura, as many other charactersobacteria as ancestral to neomura; concordance of dis-acters is convincing.hese are phosphatidylinositol, with pervasive roles inn of phagocytosis and endomembrane trafcking butall other bacteria, and sterols (never in archaebac-

uniqueness of archaebacterial lipids was a major (notfor rejecting the idea of an archaebacterial ancestry

es (Van Valen and Maiorana, 1980) in favour of therial/neomuran interpretation where archaebacteria aresisters of eukaryotes (Cavalier-Smith, 1987b). Replace-haebacterial lipids by acyl ester lipids from the enslavedcterial ancestor of mitochondria is a formal possibilityMller, 1998), but evolutionarily extremely onerous

y; phylogeny gives no reason to assume it in the rstore likely, archaebacteria and eukaryotes evolved fromprokaryotic ancestor with acyl ester lipids.eomuran characters are best interpreted as derived

stemming from two key changes to eubacteria afters lost: origin of cotranslational synthesis of N-linkedns and evolution of passive negative DNA supercoilingones (thermal ratchets possibly derived from MreB orer et al., 2008) insteadof activenegative supercoilingbyDNAgyrase (Cavalier-Smith, 2002c). Both changesweregical novelties adapting cells to hotter environments,reverse direction being incomprehensible. Likewise

atures of archaebacteria, especially lipids and agella,le as adaptations to even higher temperatures (hyper-y). Archaebacterial extremophily is a derived characterith eubacterial mesophily; a reverse transition makes

avalier-Smith, 2002c).

on analysis resolves key difcult problems ofhylogeny

analysis of major transitions involving complex char-polarise them and distinguish ancestral from derivedh transition analysis is important for understand-evolutionary transitions, because sequence analysisnnot distinguish ancestral and derived states (Cavalier-1a, 2006c). Applying transition analysis to majortransitions roots the tree of life more reliably thanooting, which gave two contradictory results; a rootia (most metabolic enzymes) or between eubacteriara (DNA/ribosome-related enzymes) (Cavalier-Smith,strongly indicates that eubacteria are not a cladecestors of archaebacteria, conrming suspicions that

andstrtral to(Lake ethe indinsertiterial EbetweebacteriChloroevidentimes othan eSmith,

It istogethbacterineomu2006a)tion forThis isonly abeven hago. Of136aet al., 2ber useare anoenzymwithmular clthe evoducing

Althactinobproteanobactlikely (gyrasedo notServinabsent

6. The

Thuarchaegyrasecharacmadephosphtype Daerobicproteinenzymprobabsplittindivergeran hadstandadaughttive zoindicates thatposibacteria arederived from,notances-bacteria (Cavalier-Smith, 2006c); contrary arguments2007; Skophammer et al., 2007) are extremely weak,tabeingexplicablebyconvergentdeletionsorunrelatedI concluded that the root of all life is within negibac-cteria (Hadobacteria plus Chlorobacteria), specicallylorobacteria (e.g. the lamentous non-sulphur greenhloroexus) and all other organisms or possibly withinria (Cavalier-Smith, 2006c). Critical evaluation of fossil

so ts this conclusion, showing negibacteria as over fourthan eukaryotes; archaebacteria are probably no olderyotes, thus the youngest bacterial phylum (Cavalier-a), not the oldest as often assumed.essary to use transition analysis to root the tree and,ith discrete character cladistic analysis, to deduce theee topology, including which eubacteria are closest toecause sequence trees are indecisive (Cavalier-Smith,single gene retains enough deep phylogenetic informa-bly reconstructing relationships amongbacterial phyla.even for eukaryotes where deepest divergences were

600500My ago. Unsurprisingly for bacteria the task isr, as their major divergences were probably 2.53.5Gyfew thousand genes in most free-living bacteria, onlynservedenough formultiphylummultigene trees (Yutin; even these have less resolution than the greater num-r deep eukaryotic phylogeny. Biased evolutionary ratesproblem. Because of marked changes to DNA-handlingd ribosome-related enzymes of neomura compared

olic enzymes (Cavalier-Smith, 2002c), these twomolec-give discordant results, e.g. rRNA greatly exaggeratesnary distance between neomura and eubacteria, intro-eading long-branches into trees.h posibacteriawere probably ancestors to neomura, andria plus neomura together are probably a clade: theCavalier-Smith, 2006c), it is less clear whether acti-are direct ancestors of neomura as I argued is mostlier-Smith, 1987b, 2002c,e, 2006c) or just their sisters. Ation shows that Actinomycetes are a clade, so neomurach within them, but could be their sisters; contrary to. (2008) that insertion is not in all Actinobacteria, beingbrobacter.

muran ancestor of eukaryotes

e immediate ancestor of eukaryotes was neither anrium, nor a eubacterium with murein wall and DNAa transitional prokaryotic intermediate with a uniquembination: a surface coat or wall of cotranslationally-ked glycoproteins, acyl ester phospholipids, includingylinositol, sterols, H3/H4 core histones and neomuran-andling enzymes. It was probably a large facultativelyerotrophic cell with huge secretome of hundreds oftinobacteria have 800) including numerous digestives no bacteria have this combination of characters, itrsisted but briey historically, surviving only througho two further-altered daughter lineages: the radicallyukaryotes and archaebacteria. Thus the rst neomu-unstable phenotype unable to compete effectively withesophilic bacteria. The special strength of its survivingneages was arguably that each entered a novel adap-tally free of eubacterial competitors: phagotrophy for


eukaryotes; hyperthermophily for archaebacteria. This ancestralneomuran itself probably arose in a thermophilic (not hyper-thermophilic) environment for which histones and cotranslationalrather than(Cavalier-Smhabitat in wteins is usemarine is uote tree ofmore parsima planktonitat for the osubmarine hredox potenwas small. Tthe rst (hyphagotroph

I suggesshallows ofand proteointense witmarine matwould havemuran to sconsequentpenicillin. Mavailable alcient eukarchase or shfreshwaterfor losing aor acquirinosmoticallydrying.

Prekaryoface skeleta(Arps) ablestabilise theby pseudopby actomyobecome tubWright, 200mitosis andtated eukardivision Ftsplasmids, w2008); thisporarily whBy contrasttheir surfacteins and evof more rigArguably, stthe transitiester phospbacteria anhandful ofall acquiredhave a relatfor the actiet al., 2007mophilic enprobably cabacteria andon by Mar

are therefore irrelevant to eukaryogenesis. Ancestral hyperther-mophilic Archaebacteria also lost the related Hsp70; secondarilymesophilic lineages reacquired it from eubacteria by LGT.

requ

gotroroteicatioion,c enzanysion,exosand

eletoe wiled, be canermecy thheng aneria.coms,whitan

ry tout ocet al.dantTPasus wey cin n

-Garcs unpt euket alarke

iablend Rathe cce coevol

ocho.re isgwothe vproted wrane,roseer-Smranesmawhoevo

ptedxternacterareter mpost-translational secretion were specic adaptationsith, 2002c). Many actinomycetes are thermophilic, ahich their proteasomal destruction of denatured pro-

ful. Whether the habitat was terrestrial, freshwater ornclear, but from the distribution across the eukary-habitat preferences, a marine ancestry is somewhatonious; a benthic ancestry is much more likely than

c one (Cavalier-Smith, in press). The most likely habi-rigin of neomura was a benthic thermal gradient near aydrothermal vent,where extremes of temperature andtial closely juxtaposed and the transitional populationhere, one daughter lineage could have rapidly becomeperthermophilic) archaebacterium, the other the rst.t that this occurred within a dense microbial mat inthe photic zone, where cyanobacteria, actinomycetes,

bacteria abounded, making competition for resourcesh marked chemical warfare by antibiotics. In a densehigh salt and organic solutes exuded by phototrophsbeen osmoprotectants, enabling the ancestral neo-

urvive loss of the murein wall, possibly favoured byresistance to peptidoglycan-targeted antibiotics likeoreover, in dense mats, prey would be immediately

l around, so phagocytosis could evolve without ef-yotic motility; they just had to bathe in their food notit, as necessary for plankton. At the opposite extreme,

plankton would be the most difcult environmentcell wall, with higher likelihood of osmotic rupture,

g prey or evolving hyperthermophily. Soil would bedangerous, requiring surviving rain and/or extreme

tes happened to undergo gene duplications in the sur-l protein Mreb, making actin and actin-related proteinsto cross-link actin in a 3D meshwork and osmoticallycell. This preadapted them for amoeboid locomotion

odia, ingestion of prey by phagocytosis and cytokinesissin, freeing bacterial FtsZ protein (Osawa et al., 2008) toulin and then duplicate to six paralogues (Beisson and3; McKean et al., 2001; Tuszynski et al., 2006) enablingcilia to evolve. An intriguing idea that possibly facili-

yogenesis is that tubulins evolvednot frombacterial cellZ, but from related proteins that segregate posibacterialith properties more like tubulin (Chen and Erickson,could have enabled cell division by FtsZ to persist tem-ilst microtubules evolved for chromosome segregation., prearchaebacteria became miniaturized and increasede stability by making a rigid wall from their glycopro-olving a membrane with a covalently bonded unilayerid C40 isoprenoid-ether lipids (Cavalier-Smith, 2002c).erols in their membranes helped rigidify them duringonal stage when ancestral neomuran sterols and acylholipids were replaced by isoprenyl ethers. Archae-cestrally retained FtsZ for division, but lost MreB. Aarchaebacteria have MreB-related proteins, probablyby LGT from eubacteria, possibly via plasmids, which

ed protein ParM for their DNA segregation, as proposedn-like protein of Thermoplasma and Ferroplasma (Hara). These protein sequences are closer to those of ther-dobacteria than to actin or actinobacterial MreBs, some by LGT from an endobacterium long after archae-eukaryotes evolved. Contrary to Searcy (1987), seized

gulis et al. (2005), Thermoplasma actin-like proteins

7. Pre

Phaglycopmodicle fusby lyticling mand fuface byon thoucytoskthat wit evolvbest wkey intefcientially, wbuddinin bactand beGTPaseconcomContraotes, b(DongdescenRarD Gand ththat thwidely(Lpez1998) ifact tha(Dongtheir mnot relRarD aor forevidenbranesto mithistory

Thebuddintargetaptedadoptemembbrane a(Cavalimembthe pla

Thestartedpreadaprey emyxobbut allond ouisites for evolving phagotrophy

phy requires binding prey to the predators surfacens, pseudopodial engulfment by actomyosin and lipidn, successiveprocessingofphagosomecontentsbyvesi-e.g. of acidosomes, fusion with lysosomes, digestionymes, active import of products to the cytosol, recy-membranes to other compartments by vesicle buddingand of the residual digestive vacuoles to the cell sur-cytosis. Evolution of efcient phagocytosis dependeds of newgenesmaking the endomembrane systemand

n, many of unknown functions. The complexity is suchl never reconstruct in detail every pathway by whichecause so many things had to happen in parallel. Thedo is to distil its central logic and suggest plausible

diates (Fig. 2). Much of its present complexity concernsat would not have been important or even possible ini-core functions began. Central is well-controlled vesicled fusion and actomyosin motility; neither ever occursBoth so intertwine that each must have started simplye more complex together. Both are controlled by smallosemultiplication into numerous paralogueswas a keyt and partial cause of endomembrane differentiation.past ideas, small GTPases are not restricted to eukary-cur throughout bacteria as four different paralogues, 2007b). As eukaryotes arose from an early neomuranof actinobacteria, all most likely evolved from smalles (present only in actinobacteria and archaebacteria)ere vertically inherited by the eukaryote host. The ideaame from the less similar BglA GTPase family foundegibacteria via cell fusion with a -proteobacteriuma and Moreira, 1999, 2006; Moreira and Lpez-Garca,arsimonious, superuous and entirely implausible. Thearyotic paralogues form two branches on a GTPase tree

., 2007b) is probably a phylogenetic artefact caused bydly contrasting divergence from a common ancestor,evidence that only Sar/Arf came from actinobacterials/Rab/Rho from-proteobacterial BglA (less like Rabs),ytologically untenable fusion theory or even LGT. Nontradicts the idea that phagocytosis and endomem-ved purely autogenously in an early neomuran priorndria, without input from LGT or any more complex

amarkedchicken-and-eggproblem: specicmembraneuld lack anobvious advantagewithout specic fusion toesicle, and vice versa. Since each involves several coad-ins, how did either start? The explanation previouslyas that actomyosin accidentally internalised surfaceand crude membrane re-fusion with the surface mem-prior to the origin of endomembrane vesicle buddingith, 1987b, 2002e). Because phagocytosis internalises

it cannot have evolved without means for recycling it tomembrane, i.e. exocytosis.le process was so complex that the neomuran thatlving phagotrophy for predation must have beenas a non-phagotrophic predator that digested its

ally. Several such bacterial predators exist today, e.g.ia, Vampirococcus, Daptobacter (Guerrerro et al., 1987),negibacteria, bounded by a murein wall and a sec-embrane, whose rigidity and complexity prevented


Fig. 2. Logic awhich is divideoccurs, and smlysosomes (L)endosome-desPeroxisomes (hypothetical senzymes; preyovals) and assphagosome (Pphase of endopre-existing SNand some Sec6of phagosomaendomembranevolved, qualitCOPII and adahomotypic COCOPIa,b differeenvelope and(2007).

any of themever a neoand digestthe difculsurface coadigestive enthe prey anucts. At thacotranslatiotionally glythe ribosomcarriers to Nhaving evolnd phagotrophic origin of the eukaryotic endomembrane system. (a) Modern cells. Md into rough regions (RER) bearing signal-recognition-particle receptors and therefore ribooth regions whence Golgi-destined COPII vesicles bud. From the trans-Golgi network (Tand endosomes (E); phagosomes (P) enclosing prey phagocytosed by actomyosin pseudtined clathrin-coated vesicles bud fromplasmamembrane coated pits; TGNnon-clathrinPer) divide and/or form from vesicles budded from ER; COPI vesicles budding from Golgtages in evolution of prekaryote endomembranes. (b) Predation probably began by preylysis liberated proteins (Pr). The rst simplest form of intracellular digestion perhaps in

ociated 26S proteasome digestion chambers (top right). Soon thereafter, actin-driven pse). After digestion, the phagosomal membrane could be re-fused with the plasma membranmembrane stabilisation involved evolution of the rst undifferentiated coated vesicle (cvAREs (not shown). By chance some protoendomembranes carried DNA, some TAT protei1 channels, SR and Derlin (protoER/Golgi: pER/Golgi). As coated vesicle driven growth ofl membrane, the plasma membrane became depleted of ribosome receptors (SR) and Des, peroxisomes and DNA were permanently internal and absent from the surface membry and cell cycle control superseding their original digestive function. (d) In the third phaptin/clathrin (and of cognate SNAREs) made separate ER and protoGolgi. Evolution of rPII fusion (H) generated Golgi cisternae (specialising in complex saccharide and lipid synntiated proximal and distal cisternae (Donohoe et al., 2007), and new structural protein

pore complexes (Fig. 3), after which many proteasomes functioned within the nucleus als

evolving phagocytosis and internal digestion. If how-muran prekaryote evolved a similar capacity to bindbacteria externally, it would have been preadapted fort transition to phagocytosis. Initially its glycoproteint could be modied to bind prey, enabling secretedzymes and membrane destabilisers to kill and digestd surface membrane transporters to import the prod-t stage digestive enzymes would have been secretednally across its single surfacemembrane, andcotransla-cosylated by oligosaccharyl transferase associated withe, transferring oligosaccharides from their isoprenoid-asparagines on the nascent protein, this cooperation

ved in the ancestral neomuran before prekaryotes and

prearchaebto the cell sdiffusing awdigestion isand competis smaller tsoup of digdoes Daptobanswer to textra large penabled efbefore phagincrease latembrane lipids and proteins are made in the nuclear envelope/ER,osomes, where alone cotranslational protein synthesis/glycosylationGN) bud clathrin-coated vesicles for supplying digestive enzymes foropodia (A) are acidied by acidosomes (a) and fuse with lysosomes;vesicles bud for exocytosis (secretion and plasmamembrane growth).i cisternae mediate retrograde transport towards the ER. (bc) Threeadhesion to the surface and secretion of externally bound digestivevolved just surface membrane protein-import channels (Derlin; greyudopodia (A) sometimes accidentally fused, enclosing prey within ae by ancestral V-SNAREs (black) and t-SNAREs (grey). (c) The second) budding by a novel protocoat (ancestral to COPI and adaptins) and

n translocation machinery (black rectangle; protoperoxisomes: pPer)the plasma membrane increasingly predominated over direct returnNA attachment proteins; when totally devoid of them peroxisomes,ane. Proteasomes associated with the ER and ERAD/ubiquitin controlsse of endomembrane differentiation, divergence of vesicle coats intoetrograde transport by COPI sharpened that differentiation. Finallythesis) distinct from the TGN; COPI duplications and divergences tos stacked them. Separate homotypic copII fusion made the nuclearo. For additional discussion see Dacks and Field (2007), Grkan et al.

acteria diverged. Digestive enzymes probably attachedurface by membrane anchors to prevent waste throughay uselessly. A central inefciency of such externalthe loss of much digested material to the environmentitors before absorption. This is reduced if the predatorhan its host, bores into its cytoplasm, and bathes in aestion products still enclosed by prey membranes, asacter (Guerrerro et al., 1987). Though not a generalisedhe problem as it limits predator size, constraining it torey,whenmost potential prey is too small, it couldhavecient evolution of prey binding and external digestionocytosis arose, facilitating internal digestion and sizeer.


One can envisage parallel ways for predation to improve grad-ually by evolving several characteristic eukaryotic properties. Oneis to control cotranslational secretion, secreting enzymes only inresponse tomaking thegralmembrthis by dire(SR) on theeukaryotes,which can ssecretory por alternatibecause theN-terminalcould haveinitiate coat2006; Pasqulater internboth modiing to memN-terminalpositively cterminal amal., 2008). Iotic noveltifrom surfacbrane tubuand eukaryotion via maand phagoc

8. Internalproteasom

The ERAunfolded prpolyubiquitfor degrada2008; Raasition evolvedmembrane,the cell andmore efcieThis idea pforce for evmuch strontallymisfoldthey now mfoodproteinbe no needuse of this mtheir originsomal degrcomplexitymore basic

ERAD de1, possiblythe 19S regcytosolic fastrates (Wabe importesomes. Mudigest extradifferent m

uitinated before entry into the proteasome. But such complexpolyubiquitin selection is necessary only because ERAD now justexerts quality control over the cells own newly made proteins

ust nve thestedbyomemesprotsibiliealbiqu

y moe presubu, 200se suotesve caytosouldl comly reunk

udop

evolvpseut aredayr exho cs inwhi

rane-attacendsphaptaticteri) as datiddosoans-osis,ulti

raneembrs andInvay maacidimpe ef, 200t seeternae rolon, we prdid tng thprey adhesion to the membrane carrying ribosomesm, not constitutively. Conformational changes in inte-ane glycoproteins bindingprey externally could achievectly affecting the signal-recognition-particle receptorcytosolic face of the membrane. Possibly this was whybut not archaebacteria, evolved SR, a small GTPasewitch between binding to SR, allowing it to transfer arotein into the trimeric Sec61 ER membrane channel,vely forming homodimers that prevent translocationy cannot bind SR (Schwartz et al., 2006). SR has anhydrophobic helix attaching it to the membrane. SRbeen ancestral to related GTPases, Sar1 and ARF, whichassembly on membrane transport vesicles (Kahn et al.,alato et al., 2002), but evolved only after phagocytosis

alised parts of the cell surface as protoendomembranes;ed the membrane-embedded helix, Sar1 only attach-branes in its GTP, not GDP state, and Arfs adding anmyristoyl fatty acid also as membrane anchor. Arfsurve membranes by GTP-driven insertion of their N-phipathic helix into the lipid bilayer (Lundmark etsuggest that prior to phagocytosis four other eukary-

es evolved to improve digestion and protein absorptione-bound prey: actomyosin-based pseudopodia, mem-lation, ER-associated degradation of proteins (ERAD),tic 26S proteasomes. Any of thesewould benet preda-ny fewer innovations than the endomembrane systemytosis, and be easier to get started.

digestion rst by retrotranslocation and 26Ses?

D system and proteasomes are coadapted for extrudingoteins from ER lumen into the cytosol, concomitantlyinating them, then passing them to 26S proteasomestion within (Brodsky, 2007; Li et al., 2008; Lipson et al.,and Wolf, 2007; Tamura et al., 2008). If this coadapta-whilst ribosomeswere still attached to the cell surface

itwould takepartly digestedpreyproteins fromoutsidedigest them inside with 100% retention, necessarily farnt than external digestion with some inevitable loss.rovides the rst obviously extremely strong selectiveolving ERAD, 26S proteasomes, and polyubiquitination,ger and more basic than quality control over acciden-edproteins or temporal controls over the cell cycle thatediate. Moreover, as the function was just convertinginto aminoacids, digestion alone sufced. Therewouldalso to evolve all or any other processes necessary forachinery in cellular controls, previously postulated as

al function (Cavalier-Smith, 2002e). The basic protea-adative machinery probably had to evolve its presentbefore its co-option for multifarious controls. What isthan simple digestion of prey proteins?pends on a small integral membrane protein Derlin-constituting the ER membrane channel for extrusion;ulatory subunit of the 26S proteasome is the only

ctor essential for translocating non-ubiquitinated sub-hlman et al., 2007). Any soluble external protein couldd and digested just by Derlin and cytosolic protea-ch simpler than phagocytosis, this would efcientlycellular proteins internally. Modern cells use numerousotifs to select different substrates, mostly polyubiq-

and mto evolbe dighelpedthe endendosocation/responpiecemgin of uinitiallbecamlatoryMorrisproteaeukarydigestiphagocunits ccontrodistantmostly

9. Pse

Byextendcontacucts. Toboth fotion. RGTPase2004),membbranediverseby phopreadaonly banositolPhosphandenlates trexocytsome/mmembthe mtubule2004).usefullamino

It isincreas(Jkelycient. Ithat increativsecretiwas th(2004)implyiot digest the wrong thing. It would have been easiere basic machinery when any external protein should. Adding polyubiquitin tags by polyubiquitin kinasesATPase Cdc48 was a renement only necessary aftermbrane system formed, when ER became distinct fromand lysosomes and phagocytosis replaced retrotranslo-easomes for internal prey digestion. Relieved of thatty, ubiquitin-controlled proteasomes were recruitedfor disparate cellular controls by proteolysis. An ori-itin-controlled degradation simply to make predationre efcient, explains why archaebacteria, which neverdators, did not evolve it, why the 19S proteasomal regu-nit became complex (Chen et al., 2008; da Fonseca and8; Lipson et al., 2008; Nickell et al., 2007), and whybunits underwent gene duplication and divergence inalone: increased protease diversity quickly increasedpabilities for utilising more prey proteins. Later, afteris and lysosomes tookoverpreydigestion, different sub-digest different cellular proteins, enabling full modernplexity to evolve easily. Derlin probably evolved fromlated bacterial membrane proteins (e.g. ABJ85348) ofnown function.

odia, prey uptake, and membrane recycling

ing actin and Arps to branch it, the prekaryote coulddopods partially around the prey cell, much increasinga and thus absorption efciency of digestion prod-Rho small GTPases (e.g. Rac) control actin remodelling,tending phagocytic pseudopods and prey internalisa-an weakly group on sequence trees with related Rabvolved in transport vesicle target specicity (Jkely,ch may also have evolved from SR by losing itsattaching peptide, getting lipid tails instead for mem-hment. Rabs all are prenylated and associate withomembranes; Rac is attached to the plasma membranetidic acid. Outer leaet phosphatidylinositol was a keyon of the prekaryote, derived from actinobacteria (theapreadapted for phagocytosis byhavingphosphatidyli-iverse phosphoinositides are vital formembrane trafc.ylinositol 3-phosphate is essential for phagocytosismal trafcking, phosphatidylinositol 4-phosphate regu-Golgi secretion, phosphatidylinositol 4,5-bisphosphateand phosphatidylinositol 3,5-bisphosphate late endo-vesicular body trafcking (Deli et al., 2008). SurfaceARF and actomyosin could probably together tubulateane even before microtubules evolved (ARF, micro-

kinesin alone are sufcient for tubulation: Jkely,ginating surface membrane as narrow tubules wouldgnify surface area for protein uptake by Derlin andactive import.lausible that surface invaginations evolved instead tociency of secreting digestion enzymes to the prey4); direct secretion at the surface would be more ef-ms odd to accept the prekaryote as predatory, but denyl digestion (entirely novel for eukaryotes) had the keye in eukaryogenesis, and to argue instead that enzymehich prokaryotes have done perfectly well for 3.5Gy,imary force in eukaryogenesis (Jkely, 2004). Jkelyhat because he interpreted his GTPase sequence tree asat secretion evolved before phagocytosis; that conclu-


sion was unjustied (see Cavalier-Smith, submitted for publicationfor more details).

Actin evolved from MreB, present in the neomuran ancestor,but not inas phosphaHsc70 is essing endomefor eukaryoits distantfusion betwet al., 2002ing the tailuniversallyeukaryoticdisjunction(Cortes-LedHaering, 20before theRichards antosis in uniphagocytic

Coiled-ceukaryotictarget memfusion (Ant2007; Fasshformins forthat SNAREre-fuse mordental fusioprey. Afterto its boundentally reinefcientinitial innovprimitive exface probabeasier for Swith only oRabs themspresent inthe last steevolve, wasreemphasizeukaryogen

As soonby SNAREsproteins libribosomes s(Fig. 2c). ThThe rst entor of COPI(Donohoe e(now used fding); COPI(Stagg et alby these prfusion of eand plasmabenecial, tbrane. For esimple uncoefciency isduplicatedthan surfac

vesicles. If different membranes to be targeted increased one ata time and selectivity is advantageous, as on my scenario, its evo-lution is comprehensible. This stage would have worked well, with

rary(whind faids anlgi (

.l con, alrecteri2007ruitemitoy wasaro eenntosisrom

A inemb

yclinent.embally;agmen cohrommakecellse ofuplicd thmbraprodmpoes wAREembnceserie

iffereranound tachmall cicubulonly

r porcom-soloundes aade

sicleed b198

raneclingof rend comost archaebacteria, whose ancestor lost it as welltidylinositol, sterols, and Hsp70 chaperones (relatedential for clathrinuncoating), all prerequisites for evolv-mbranes,making archaebacteria implausible ancestorstes. Phagocytosis also needs myosin. Together withrelative kinesin, myosin probably arose by domaineen ATPases derived from a bacterial GTPase (Leipe), forming the head, and a coiled-coil protein form-. Plausible coiled-coil tail ancestors are Smc proteinsimportant for chromosome segregation, from which

cohesins (holding sister chromatids together for proper) and condensins (compacting chromatin) evolvedesma et al., 2007; Hirano, 2005, 2006; Nasmyth and05). Myosin tails diversied into at least three kindseukaryotic cenancestor (Odronitz and Kollmar, 2007;d Cavalier-Smith, 2005); myosin II mediates phagocy-konts, but myosin I does in bikonts; probably the rstmyosin was less differentiated.oil motifs are rare in bacteria but important in manystructural proteins, e.g. tethers anchoring vesicles tobranes before fusion, SNAREs that mediate speciconin et al., 2002; Bassham and Blatt, 2008; Cai et al.,auer et al., 1998; Liu et al., 2008; Mima et al., 2008),positioning actin, and intermediate laments. I suggests rst evolved in the prekaryote to help phagosomese efciently with the plasma membrane (Fig. 2b). Acci-n of pseudopodia would inevitably often internalisedigestion by enzymes made by ribosomes attachedding membranes, the spent phagosome could acci--fuse with the plasma membrane, but this would bewithout fusogenic proteins (SNAREs), probably a keyation to make internal digestion more efcient. Thus,ocytosis to recycle phagosome membrane to cell sur-ly evolved before internal membrane budding. It wasNAREs and Rab-based GTPase controls to evolve thenne possible target membrane, the plasma membrane.elves are controlled by activators, with three classesthe cenancestor; the activator of Rab4 that controlsp in exocytosis (Novick et al., 2006), possibly rst toa catalytic coiled-coil, Sec2p (Dong et al., 2007a),

ing the importance of novel coiled-coil proteins inesis.as phagosome formation by actomyosin and recyclingevolved, coated vesicles could evolve for pinocytosis oferated by external lysis of prey by enzymes made bytill present on the inner face of the surface membraneus, early predators had three ways of internalising food.docytic vesicle coats were probably the common ances-coats (now used for retrograde transport from Golgit al., 2007), which did not then exist), and clathrin coatsor plasma membrane, endosome and trans-Golgi bud-coats are homologous to adaptins that bind clathrin., 2007). The target membrane for vesicles generatedotocoats was the primitive phagosome. Even randomndocytosed vesicles promiscuously with phagosomesmembrane, despitewaste if the latter, would have beenhe more so the higher the proportion of internal mem-xplaining evolutionary transitions it is important thatntrolled versions of processes have strong advantages;initially irrelevant, but could immediately improve if

SNAREs that preferentially targeted phagosome rathere membrane associated specically with the coated

temposomesacids apholipand Gomade)

FinaGTPaseposibaet al.,ally recnally

Pre(Danov1993; Hendocygenes f

10. DNendom

Recinefciendomgeneticthey frregatiobear cwouldnomicabsencgene dsupplieingmewouldtein coenzymand SNendompreferemachinSuchdmembthem aity. Attthe smby spemicrot

Notnucleahave aily of never fmachinWhat mtory veexcludSmith,membof recynationsites aphagosomes ancestral ultimately to endosomes, lyso-ch now digest proteins), peroxisomes (where D-aminotty acids aredigested) andER (whereproteins andphos-d sterols are now made, and to which DNA is attached)where complex carbohydrates and sphingolipids are

striction of clathrin-coated vesicles requires dynaminadypresent inbacteria (LowandLwe,2006), includinga. Spiral arrays of dynamins divide membranes (Mears); duplications generated many paralogues, eventu-d for dividing endosomes, vacuoles, peroxisomes, andchondria.probably not only bacteria; viruses are more abundantt al., 2008) and protozoa eat them (Gonzlez and Suttle,emuth et al., 2008); probably some manipulated earlyfor entering the host, which perhaps even got useful

them.

ternalisation, copII and the origin of permanentranes

g phagosomes to the surface was originally inevitablySome membranes failed to re-fuse, becoming potentialranes, but without division would not be perpetuateddynamin provided the division mechanism the morented into numerous small pieces themore randomseg-uld maintain them. Some of these membranes wouldosomes; until mitosis evolved, random segregationsomeDNA-less cells thatwould die andmanymultige-(ineffective for making numerous paralogues in the

efcient multichromosomal segregation, so individualation, not whole genome duplication, probably mainlye burst of new paralogues). Primitive dynamins divid-nes and coats budding them, neither as specic as now,uce a plethora of internal membranes of disparate pro-sition (and lipid composition unless lipid synthesisingere distributed uniformly). Poorly specic coat proteinss would equilibrate lipids and proteins among theseranes. As specicity developed through initial randomamong paralogues of the dividing, budding, and fusions, different endomembrane types would be stabilised.ntiationalso requiredphysical linksamong theevolvings organelles, because such attachments, and betweenhe cytoskeleton, are central to eukaryotic cell hered-ents linking all organelles are striking exemplied inagellate Sainouron (Cavalier-Smith et al., 2008a), andbinding of centromeres by nuclear envelope protein toes in yeast (King et al., 2008).

clathrin vesicle adaptins and COPI coats, but alsoe complexes (NPCs) and COPII coats that bud from ERmon origin, as they all belong to a major superfam-enoid/-propeller proteins (Devos et al., 2004, 2006),in bacteria, whose origin as novel membrane-curving

rguably began in early phagotrophs as outlined above.endomembranes permanent was the origin of secre-budding from endomembranes such that the vesiclesoth ribosome- and DNA-attachment sites (Cavalier-7b, 2002e). If vesicles carried SNAREs for plasmafusion with them, they would be an alternative meansphagosome membrane to the cell surface. The combi-cycling biased against ribosome- and DNA-attachmentntinued phagocytosis would steadily deplete the sur-


face of ribosome- and DNA-binding sites until all were lost fromit, remaining only on endomembranes. New phagosomes wouldthereafter lack such binding sites; therefore, whether their mem-brane wasbudding, thotic state beplasma memgenes, but wtor loss from(Cavalier-Smthe cell wasexist. One many modernmuch less dsome depleto evolve, atially ineffusion, withmembraneance betweinternal meensuring nesecondpossgence in biand endocyitselfmakediately follounimpededtion prevenplasma mem

The logrequires mespecic tra(SNAREs); oplest explancognate coaalready hava formofmeity (Cavaliemembrane,and Rapopomatically gstrength offor non-cogevolved begest that prmembranesproteins antially genersurface ribphagosomedigestive enmembranevesicles tartem for exomethod invtant relativ(Harsay anAvl9 acts incoat compl(Sanchatjat

This supthan once satively efccould evolv

simplest way of evolving them is for recycling phagosomal mem-branes, outlined above. This exemplies for membrane trafckingthe established principle that complex metabolic pathways evolve

asilyonarerennlyy homughithnetiubulm: mchro

ationr envatin.the nytoss andsubm. 3 sut onlgi (FOPIIn ER006)st intheias trcestrcati

easeoate

roxisntia

xisog peurso2005roxise bassionriginntiated iner-Smmemondrfcieo acure

uct,prot

er-Smgregas TAldedhe ts Sectworetairecycled to the surface by direct re-fusion or vesiclee surface membrane could never regain its prokary-aring ribosomes andDNA. This differentiation betweenbrane and endomembrane was not directly coded byas a membrane mutation caused by permanent recep-the cell surface that no DNA mutation could reverseith, 2004b). A new era of cell evolution had begun, butstill not eukaryotic, as a nuclear envelope did not yetay not be able to equate the original vesicle coat withcomponent because the system must then have been

ifferentiated. There are two obvious possibilities. Ribo-tion might have been achieved by the very rst coats suggested above for pinocytotic feeding. Even if ini-cient and more indiscriminate for both budding and

vesicles fusing with and budding from both surfaceand phagosome, this would provide a quantitative bal-en both membranes; just by mass action an excess ofmbrane caused by phagocytosis would bias transfer,t movement from endomembrane to cell surface. Theibility is thatduplicationof coats andSNAREsanddiver-nding properties separated more polarised secretorytic paths. If such duplication of vesicle targeting did nota permanent endomembrane, it probably almost imme-wed it to make digestion more efcient and ensureplasma membrane growth after ribosome internalisa-ted them inserting membrane proteins directly into thebrane.

ic of evolutionary endomembrane differentiationchanistic linkage of coat-biogenesis of compartment-fcking vesicles and of their targeting machinerytherwise their evolutionwould be impossible. The sim-ation is that SNAREs bind directly and specically tots, ensuring their automatic targeting to membranesing the same SNAREs; this perpetuated target identity mbranehereditydependingonprotein complementar-

r-Smith, 2004b). T-SNAREs would remain in the targetbut V-snares must be recycled to the donor. Heinrichrt (2005) showed that such complementarity can auto-enerate distinct membrane compartments if bindingcognate SNAREs and coats is substantially greater thannate ones. Probably >23 different coat/SNARE partnersfore the cenancestor (Kloepper et al., 2007). I sug-imary endomembrane differentiation was into internalbearing ribosomesandDNA(protoER/NE/Golgimakingd lipids) and some not (protoendosomes), both ini-ated automatically by phagocytic internalisation. Onceosome depletion made endomembranes permanent,s would have to fuse with protoendosomes bearingzymes. To make this efcient two different kinds ofbudding arose from the protoER/Golgi: clathrin-coatedgeted to endosomes and a largely uncharacterised sys-cytotic secretory vesicles. One widespread exocytosisolves Avl9, a universal eukaryotic protein with dis-es involved in wall attachment of mainly posibacteriad Schekman, 2007). Unfortunately it is unknown iftrans-Golgi network vesicle budding, as does Chs5/6

ex for budding secretory chitosomes in higher fungie and Schekman, 2006).poses a relatively later origin of COPII budding (Fig. 2d)uggested (Cavalier-Smith, 2002e, 2004b) because rel-ient SNAREs for exocytosis must exist before COPIIe (otherwise it would lack selective advantage) and the

more eevoluti

Diffnent onovo bboth dagated wcopygemicrotate thebind tosegregnucleachroming byphagocnucleuSmith,but Figcles nothe Gotypic Cbetweeet al., 2not exilack it;actingthe ansimplito incrother c

11. Pediffere

Peroexistinof precet al.,ing pealso thby diviably odifferediscuss(Cavaligeneticmitochmore eD-aminterial mbyprodsoluble(Cavaliand setosis, afully foThus, tsystemsomes,the ERbackwards (Horowitz, 1945), because this requires noy foresight.tiated endomembrane compartments become perma-if they divide, e.g. by dynamin, or can be made de

otypic membrane fusion. They must also segregate toters. Protoendosomes perhaps initially randomly segre-sufcient accuracy given high copy numbers. The lowc/biosynthetic compartment evolved specic binding toes or the -tubulin-containing centrosomes that nucle-ost important for the chromosome, but rough ER couldmatin surfaces by specic proteins for segregation. ERwas probably an early reason for thus originating aelope, which does not merely surround but binds toAnother was probably protection of DNA from shear-ew molecular motors, initially myosin that rst helpedis, then vesicle transport and cytokinesis. Origins of the

mitosis are discussed in detail elsewhere (Cavalier-itted for publication); I lack space to repeat arguments,mmarises key points. Homotypic fusion of COPII vesi-

ly generated the nuclear envelope and NPCs, but alsoig. 2da); as can still happen in modern cells. Homo-fusion can generate the compartment intermediateand Golgi in many opisthokonts and plants (Bentley

. However, this intermediate compartment possibly didthe cenancestor. Many small protozoa of diverse phylar Golgi attaches directly to a smooth region of the NE,ansitional ER (Cavalier-Smith et al., 2008a,b), possiblyal state for eukaryotes rather than multiple secondaryons. Retrograde COPI transport probably evolved laterretention efciency for ER functions, as did numerousd-vesicle budders and cognate SNAREs.

omes as endomembrane digestivetions

mes form both by dynamin-powered division of pre-roxisomes (Hoepfner et al., 2001) and by ER buddingrs containing the membrane protein PEX3 (Hoepfner), which imports other peroxisomal proteins, endow-ome proteins with their organelle identity. PEX3 isis of peroxisomal membrane heredity when they form(Cavalier-Smith, 2004b). Therefore, peroxisomes prob-ated from the protoendomembrane by budding andion as originally proposed (Cavalier-Smith, 1975) anddetail (Tabak et al., 2006). An origin by endosymbiosisith, 1990) iswrong; theyareprobablynot irreplaceablebranes (Cavalier-Smith, 2004b;DeDuve, 2007), unlike

ia, plastids and ER. Peroxisomes made prey digestionnt by segregating fatty acid -oxidation enzymes andid oxidases (for digesting hydrolytic products of eubac-in walls) plus catalase for destroying their harmfulhydrogen peroxide. The PEX system for importing sucheins probably evolved from the eubacterial TAT systemith, 2006a), accidentally removed from the cell surfaceted into ancestral peroxisomes (not ER) by phagocy-T and PEX both import proteins post-translationallyafter recognising related C-terminal targeting signals.

wo contrasting ancestral neomuran protein-targetingand TAT segregated respectively into ER and peroxi-

different autogenously evolved respiratory organelles;ned the actinobacterial surface membrane cytochrome


Fig. 3. Originpermanently imyosin motorpore complexetransport. (d)and/or-propdomain bearinwere narroweby three carriecisternae prevwhich gives th

P450 systemsoon be disevolved a nprotein to u

12. Phagotsymbiogen

Arguablyof related mmembrane-proteobaesis was the-proteobaproteins; seably had oxit could losof mitochonhost, but mthrough comof the cytoskeleton, mitosis (left) and nucleus (right). (a) By the time peroxisomes (P) anternalised (Fig. 2c), early mitosis must also have evolved to segregate the chromosoms. (b) Dynein evolution improved segregation and preadapted the cell for ciliary origins; Es (NPCs), mitochondria and cilium made the cenancestral eukaryote. (df) Two-phasePhase I: nucleoporins (Nups) forming the octagonal cylindrical scaffold evolved by dupleller domains, being attachedby integralmembraneNupsdescended fromactinobacterialg ribosome receptors (SR) and the inner membrane domain bearing integral membrane cd by FG-repeat-rich Nups, preventing passive diffusion of macromolecular complexes anr complexes (karyopherins; ribosome subunit and mRNA transporters). (f) Phase I in sented by COPII coat proteins (black blobs) remaining in place to become octagonal NPCe detailed rationale.

for oxidative sterol synthesis. A novel coat type mightcovered for budding PEX3 precursors. Peroxisomes alsoovel multispanning integral membrane ADP/exchangese fatty acid oxidation energy better.

rophy and novel protein targeting enabledetic organelle additions

the peroxisomal ADP/ATP exchanger was the ancestoritochondrial, and thereby secondarily of plastid, innercarriers, which had the primary role in later enslavingcteria and cyanobacteria. Central to mitochondriogen-origin of novel protein-import machinery, partly from

cterial outer membrane proteins and partly from hoste Cavalier-Smith (2006b, 2007a). Initially thehost prob-idative phosphorylation in its surfacemembrane,whiche after enslaving the -proteobacterium. Acquisitiondria permanently added no novel metabolism to theade its phagotrophy and respiration more efcientpartmentation (Cavalier-Smith, 2002e, 2006b, 2007a).

Incidentallythat after mrampantly iing transcribecause nu(unlike bactive skeletaenvelope orspliceosomsomesevolvlater.

Only pendomembendosymbiphagocytosterium to mfor mitochofor ever, itfor proteinToc/Tic macnd permanent endomembranes (EM) evolved and DNA was therebye/EM/P complex, generating a prekaryote with separate kinesin andM attachments around DNA made a primitive NE. (c) Adding nuclearorigin of the nuclear envelope from protoER and of trans-envelopeications of coat proteins of COPII secretory vesicles with -solenoidmembraneproteins; the nucleoporins separated the outermembranehromatin binding proteins (grey rectangles). (e) Phase II: NPC lumensd mediating active specically targeted nucleocytoplasmic exchangeurface view, showing complete Ran GTPase-mediated fusion of RERscaffolds. Modied from Cavalier-Smith (submitted for publication),

it added harmful genetic parasites: group II intronsoving to the nucleus became spliceosomal introns thatnvaded genes (Cavalier-Smith, 1991c). Though increas-ptional costs, introns did not increase replication costsclear genome sizes are not minimized by selectionterial genomes) but coadapted to cell volume by posi-l functionsofnuclearnon-codingDNAcausedbynuclearigins (Cavalier-Smith, 2005). Ubiquitin is essential foral assembly (Bellare et al., 2008), showing that spliceo-edafter endomembranes, somitochondria alsoevolved

rior evolution of phagotrophy, cytoskeleton andranes, as Stanier (1970) cogently argued, madeosis and mitochondrial enslavement possible. Lateris allowed a bikont eukaryote to enslave a cyanobac-ake chloroplasts and the plant kingdom, in which asndria the negibacterial outer membrane was retaineds envelope protein-export apparatus being modiedimport with transit sequences mediating targeting viahinery. In contrast to mitochondria, plastids did add


novel metabolism (oxygenic photosynthesis) but also increasedcompartmentation efciency, e.g. by replacing cytosolic fattyacid synthetases by plastid/cyanobacterial ones. Soon after plantsdiversied ibikont hostdiatoms, hhave the m2003a, 200but inside t(chromists)to target nusignals (Cavchromalveoargued aboapparentlythe ER lumplasma mem(Maier, in p

Thus thrmembranebranes appcomplexlyvations weperoxisomesymbiogeneplasts, and

13. Cilia, amembrane

Cilia, cenbiogenesis,thenucleusany contribsis of Saganand mitosisogy; earliertestable remrecent absu2005) is phyble. MechanunexplainetobacterialfromeukaryCavalier-Sm

Ciliary o, tubulinous other cmicrotubulest eukaryoalmost ining precededynein of p(Cavalier-Sm(Iyer et al.,cenancestoal., 2008);tion bres aensuring thcilium. Theygate the enover themmove precucomplex of

less evolutionarily derived from vesicle COP/adaptins (Jkely andArendt, 2006). Thus prior endomembrane evolution was prereq-uisite for the origin of a separate ciliary compartment, despite it

ng detic oFor t1987

e bovolu

en te beat ade wold sp

72).

winolecuogenlly. Hcatiordnearoucal acompicle-wheagoc-meg. thsmicsteader wocatiowinsy pacd eastronuantons foly aa ranvolutl mac

nclu

winsxcep. I hon, tstags froteasoas tprey. Onccomall tdert int. In ms thento three lineages, enslavement of a red alga by anothergenerated chromalveolates, whose algal members (e.g.aptophytes, brown algae) dominate the oceans andost complex cell structures known (Cavalier-Smith,4a, 2007c). Their chloroplasts are not in the cytosolhe former phagosomal membrane (alveolates) or RER. The host signal and transit machineries were co-optedclear-coded proteins to plastids by bipartite targetingalier-Smith, 1999, 2003a). Intriguingly the ancestrallate duplicated the ancestral ERAD machinery, which Ive rst made prekaryotes efcient predators, and thenco-opted it for exporting nuclear-coded proteins fromen across the periplastid membrane (former red algal

brane) into the periplastid space (its former cytosol)ress).oughout eukaryote megaevolution basic principles ofheredity and protein targeting into and across mem-ly. Innovations in them were keys for making morecompartmented cells, irrespective of whether inno-re purely autogenous (e.g. origin of endomembranes,s and nucleus: the primary eukaryotic features) ortic, as in the later renements of mitochondria, chloro-

chromalveolate periplastid membrane.

novel compartment not delimited bys

triole-nucleated organelles requiring 1000 genes foralso evolved in the ancestral eukaryote together with(Cavalier-Smith, 1987b, 2002e) autogenously withoutution from symbiogenesis, contrary to the hypothe-(1967). Her spirochaete theory of the origins of ciliais incompatible with all we know of their cell biol-predictions are disproved and nothing explanatory orains in the latest version (Margulis et al., 2005). A

rdly complicated symbiotic explanation (Li and Wu,logenetically unmerited and cell biologically implausi-istically how a spirochaete could lose its membranes isd as are any advantages of transitional changes. Planc-tubulins are irrelevant to eukaryogenesis, being LGTsotes (theirmembrane invaginations are also irrelevant:ith, submitted for publication).rigin by gene duplication of pre-existing tubulin to ,s for centrioles and of kinesins, dyneins and numer-ytoskeletal proteins and recruitment of centrosomales that evolved autogenously for mitosis in the earli-te is phylogenetically, mechanistically and selectivelynitely simpler. Very likely kinesin-based ciliary glid-d swimming and feeding by cellwards movement byrey trapped on the cilium preceded motility of the cellith, in press). Dynein evolved from a bacterial ATPase

2004); duplication made 20 dyneins in the eukaryoter largely for cilia (Wickstead and Gull, 2007; Wilkes ethigher plants lost cilia, then all dyneins. Nine transi-ttach centrioles orthogonally to the plasma membrane,at outerdoubletmicrotubule growthmakes aprojectingalso make the ciliary lumen a separate compartment,

try of ciliary proteins into it, and exert quality control(Stephan et al., 2007). Ciliary transport particles thatrsors up the cilium are kinesin-driven and consist of a-solenoid/-propeller proteins related to and doubt-

not beisymbio2005).1982b,

14. Thmegae

Whoncgretimwou(18

Darrial meukaryinternamodiawkwabraneshistorioddly cent vesbad joband phsurfaceties, e.periplaone indesignmodi

Darsteadiltion anfoundrapid qradiatimarkeding taxmegaenorma

15. Co

Darapply e2006a)digestisimpleproteinfor prosystemwholesystemgone tohavingwell unefcien1987b)tion walimited by membrane, invalidating the hypothesis thatrigin of cilia initiated eukaryogenesis (Margulis et al.,he autogenous origin of cilia see: Cavalier-Smith (1978,b, 1992), Jkely and Arendt (2006), Mitchell (2007).

tched, piecemeal nature and rapidity oftion

his adaptation [to a new and peculiar line of life] haden effected, and a few species had thus acquired avantage over other organisms, a comparatively shortuld be needed to produce many divergent forms, whichread rapidly and widely throughout the world Darwin

would have been fascinated how disparate bacte-lar machines were modied so dramatically duringesis to create the rst predators on earth to digest foode rightly believed that all evolution was by piecemeal

n of inherited structures, sometimes generating bizarress as no intelligent designer would. The four mem-nd chromalveolate chloroplasts are relics of successivecidents andmakeshift jobs, not elegant design. So is thelicated endomembrane system, with dozens of differ-targetingmachineries that evolved tomake the best of an the cell surface budded off as protoendomembranesytosis-driven ribosome depletion risked preventingmbrane growth. These and other cack-handed proper-e conserved sorting pathway in mitochondria, whereproteins cross twomembranes, then recross the secondof being imported directly across one as any rationaluld effect, evidence the botched way that descent withn generates complexity.quotation shows he did not believe evolution to be

ed, but thought itmust be especially fast during perfec-rly radiation of a new body plan. Simpson (1944, 1953)g evidence in the fossil record for just such extremelyum evolution, as he called it, and geologically suddenrvirtuallyeverynewanimalgroup.Quantumevolutionpplies to the origin of major body plans distinguish-ked as order of higher: what Simpson (1944) namedion to distinguish it from qualitatively less dramaticroevolutionary divergence.

sions

and Simpsons ideas on quantum and megaevolutiontionally forcefully to eukaryogenesis (Cavalier-Smith,

ave tried to explain here how eukaryotic intracellularhough now very complicated, could have evolved ines following a few initial key mutations that allowedm externally lysed prey to be individually internalisedmal digestion prior to the origin of the endomembranehe indirect consequence of the accidental ingestion ofcells following the origin of the actomyosin motilitye started, the postulated cascade of events could havepletion and generated a fully functional eukaryotic cellhe properties of Table 1 with remarkable speed (likelya million years: over 1010 generations), with all the lessermediates rapidly becoming extinct (Cavalier-Smith,y view, the key to understanding this dramatic innova-specic nature of the precursor cells and the initiating


mutations and the logic of cytoskeleton/membrane interactionsandof stepwise increases in their specicity, all under theoverarch-ing selective forces of improving the efciency of phagotrophy andof the consethe compleconsequencfusion (Cavauli or exterrevolutiona

References

Amos LA, van drial and eu

Andersson JOan ancest90.

Antonin W, FaendosomaNat Struct

Bapteste E, Chcomplex eGenome B

BasshamDC, BPlant Phys

Beisson J, WriBiol 2003;

BellareP, Smaluitin in the

Bentley M, Litether recr2006;281:

Bourbon HM.the large,2008;36:3

Brodsky JL. Tones durinJ 2007;404

Cai H, Reinischto mediat2007;12:6

Cavalier-Smith1975;256:

Cavalier-Smithspindles a

Cavalier-SmithMJ, Collinsevolution.

Cavalier-SmithGG, editorp. 30718.

Cavalier-SmithAmos WB,sium of thp. 46593

Cavalier-SmithCavalier-Smith

Schwemm1983. p. 26


branes. CoCavalier-Smith

Sci 1987b;Cavalier-Smith

and microCavalier-SmithCavalier-Smith

V, GrenierNational d

Cavalier-Smithlife. Tokyo

Cavalier-Smithtor. Funda1991b. p. 2


origin and79106.

Cavalier-Smitheditors. Th

Cavalier-Smith T. Cell cycles, diplokaryosis, and the archezoan origin of sex. ArchivProtistenk 1995a;145:189207.

Cavalier-Smith T. Membrane heredity, symbiogenesis, and the multiple origins ofe. In: Aonal SSmithuglenly treeSmith000;5Smiths. Cur

Documents

Cavalier-Smith (2009) Predación y origen de las células eucariontes. Una perspectiva coevolutiva