A PHYLOGENOMIC FALSIFICATIONOF THE CHROMALVEOLATE HYPOTHESIS

Denis BAURAIN†, Henner BRINKMANN, Jörn PETERSEN, Naiara RODRÍGUEZ-EZPELETA, Alexandra STECHMANN,Vincent DEMOULIN, Andrew J. ROGER, Gertraud BURGER, B. Franz LANG, and Hervé PHILIPPE

Département de Biochimie, Centre Robert-Cedergren, Université de Montréal, Montréal, Québec, CanadaDépartement des Sciences de la Vie, Université de Liège, Liège, Belgium

†Present address: Unit of Animal Genomics, GIGA-R and Faculty of Veterinary Medicine, Université de Liège, Liège, Belgium

What are ’Chromalveolates’?

ANRV329-GE41-08 ARI 27 September 2007 14:38

Protist: microbialeukaryote notincluding plants andfungi

Algae:photosyntheticeukaryotes (protists)not including plants

‘Chromalveolata’:putativemonophyletic groupdescended from aprotist commonancestor thatcaptured a red algaand maintained it asa secondaryendosymbiont

INTRODUCTION

The Eukaryotic Tree of Life asBackdrop for Plastid Origin

Multigene phylogenetics and genome datafrom microbial eukaryote (protist) lineageshave provided a renewed impetus to resolv-ing the eukaryotic tree of life (e.g., 11, 71,90), culminating recently in a formal classi-fication of eukaryotes into 6 “supergroups”(3, 44). These supergroups (see Figure 1)contain the protistan roots of all multi-cellular eukaryotes and are currently de-fined as ‘Opisthokonta’ (e.g., animals, fungi,choanoflagellates), ‘Amoebozoa’ (e.g., loboseamoebae, slime molds), ‘Archaeplastida’ or‘Plantae’ [red, green (including land plants),and glaucophyte algae], ‘Chromalveolata’(e.g., diatoms, ciliates, giant kelps), ‘Rhizaria’

Excavata

Rhizaria

ChromalveolataPlantae

Amoebozoa

EuglenidsParabasalidsDiplomonads

Jakobids

RadiolariaCercozoa

AlveolatesStramenopilesHaptophytesCryptophytes

Red, GlaucophyteGreen algae

EntamoebaeAmoebae

Slime molds

Opisthokonta

AnimalsChoanozoa

FungiMicrosporidia

Figure 1Schematic view ofthe eukaryotic treeof life showing theputative sixsupergroups. Thebroken linesdenote uncertaintyof branch positionsin the tree. Forexample, the‘Rhizaria’ are likelymonophyletic butmay branch withinchromalveolatesand the ‘Excavata’may comprise atleast two distinctlineages. Thepresence ofplastid-containingtaxa in thesupergroups isshown with thecartoon of an alga.

(e.g., cercomonads, foraminifera), and ‘Ex-cavata’ (e.g., diplomonads, parabasalids). Al-though the supergroups broadly capture thediversity of eukaryotes, there are in factonly two that currently have robust sup-port from molecular phylogenetic analyses,the ‘Opisthokonta’ and the ‘Amoebozoa’ (71).Therefore in this review all supergroups aremarked with ‘ ’ to denote their provisional na-ture. Of the remaining lineages, the ‘Plantae’is gaining the most support from multigenetrees (83) and features associated with thephotosynthetic organelle (plastid) in thesetaxa (e.g., 63, 78, 99). This group is verylikely to be monophyletic, a key feature thatplays an important role in understanding plas-tid evolution. The ‘Rhizaria’ includes pho-tosynthetic amoebae (chlorarachniophytesand Paulinella chromatophora) and receives

148 Reyes-Prieto ·Weber · Bhattacharya

Annu. R

ev. G

enet

. 2007.4

1:1

47-1

68. D

ow

nlo

aded

fro

m a

rjourn

als.

annual

revie

ws.

org

by U

NIV

ER

SIT

E D

E L

IEG

E o

n 0

4/1

0/1

0. F

or

per

sonal

use

only

.

CHROMALVEOLATES are a large and diverse putative super-group of eukaryotes that include a mix of photosynthetic

and heterotrophic lineages. According to the Chromalveolatehypothesis (Cavalier-Smith 1999), which regards plastid lossas much more common than plastid gain, these organismsvertically descent from a single chl. c-containing ancestor thatacquired its plastid early via a secondary endosymbiosis witha red alga (Reyes-Prieto et al. 2007).

Do ’Chromalveolates’ really exist?

would have to have been acquired before the split betweenthe haptophyte + cryptophyte clade from alveolates + -stramenopiles + Rhizaria (Figure 3b). Plastids would thenpresumably have been lost, independently, in Rhizaria,some stramenopiles, ciliates, early diverging dinoflagel-lates (e.g. Oxyrrhis) and many or most apicomplexans[39]. Finally, the subsequent uptake of a green algal endo-symbiont in the ancestor of chlorarachniophytes wouldproduce the distribution of plastids observed today(Figure 3b). Like the original chromalveolate hypothesis(Box 2), this scenario would require that plastid loss be farmore common than gain. Although the prevalence of plas-tid loss (as opposed to loss of photosynthesis) amongeukaryotes is unknown, the nuclear genomes of two Phy-tophthora species [29] (stramenopiles) and the apicom-plexan Cryptosporidium [41] encode plastid-derivedgenes, despite these organisms lacking plastids, an indica-tion of at least two instances of plastid loss in the ancestorsof these different organisms. Additionally, the recentlydiscovered photosynthetic eukaryote Chomera velia [42],which is closely related to apicomplexans, stronglyindicates a shared photosynthetic ancestor of both Apicom-plexa and dinoflagellates and subsequent loss in the plas-tid-lacking members of these groups.

If the new position of Rhizaria as a part of Chromalveo-lata reflects the true evolutionary history of this lineage,one would predict that genes of red algal ancestry mightpersist in the nuclear genomes of this group as remnants ofthe red algal genomes that were present in the rhizarian

common ancestor. Interestingly, red algal-derived plastidgenes were discovered in the nuclear genome of the greenalgal plastid-containing rhizarian Bigelowiella natans[43], and were interpreted as having been acquired bylateral gene transfer rather than vertically inherited froma red algal plastid-containing ancestor. A complete genomesequence for B. natans will soon be available (http://www.jgi.doe.gov) and will make it possible to test whetheror not this red algal ‘footprint’ is (at least in part) the resultof ancient endosymbiotic gene transfer. However, mostRhizaria are recalcitrant to laboratory experimentation,and significant amounts of sequence data from diversemembers of this lineage will be slow in coming. At anyrate, if analyses eventually show that two (ormore) distinctplastids were harbored by the ancestors of extant organ-isms, as has been previously shown in some dinoflagellates(see Ref. [37]), then determining the organismal history ofsuch eukaryotes might be even more difficult than cur-rently appreciated.

Phylogenetic hope in light of EGT?Although we have focused on chromalveolates and ignoredthe potentially significant role of lateral gene transfer ineukaryotic evolution (e.g. Ref. [44]), the reality of EGT andits phylogenetic implications can be extended to many ofthe eukaryotic supergroups. The relationships within andbetween chromalveolate and rhizarian taxa are not onlyimportant for understanding amajor component of the treeof life but also for understanding organelle evolution and

Figure 3. Two hypotheses to explain the distribution of secondary plastids, based on competing scenarios of eukaryotic evolution. A green algal-derived secondary plastidhas been acquired by two separate lineages, in independent endosymbiotic events (thin dashed lines). (a) A single red algal endosymbiosis occurred in the commonancestor of Chromalveolata, necessitating multiple plastid losses at the base of the various nonphotosynthetic lineages. (b) If Rhizaria evolved fromwithin chromalveolates,it is most parsimonious to assume that the red algal secondary plastid was lost before the diversification of this lineage. A green algal secondary plastid has been acquiredby chlorarachniophytes more recently.

Opinion Trends in Ecology and Evolution Vol.23 No.5

273

Amoebozoa

Opisthokonta

Plantae

!Chromalveolata"(Hacrobia)

!Chromalveolata"(Stramenopiles+Alveolates)

Rhizaria

Excavata

SAR

RECENT phylogenomic studies have led to dramatically ex-panded ‘Chromalveolates’. To account for the growing

collection of heterotrophic lineages apparently related to chl.c-containing algae, the Chromalveolate hypothesis has to pos-tulate ever more plastid losses. Further, it suffers from a re-current lack of support in most molecular phylogenies, eventhose based on large data sets. Therefore, alternative evolu-tionary scenarios have been proposed to explain the origin of’Chromalveolate’ organisms (Lane and Archibald 2008).

HERE, we present a falsifying experiment that comparesthe Chromalveolate hypothesis to serial models invoking

higher-order eukaryote-eukaryote endosymbioses (EEEs).

How old are chl. c-containing plastids?

Cyanobacteria

Viridiplantae

Glaucophyta

Rhodophyta

Cryptophyta

Haptophyta

Stramenopiles

Euglenophyta

Chlorarachniophyta

0.0 1.00.2 0.4 0.6 0.8Time

Phaeodactylum tricornutumOdontella sinensisThalassiosira pseudonanaEctocarpus siliculosusPavlova lutheriEmiliania huxleyiGuillardia thetaGracilaria tenuistipitataPorphyra yezoensisPorphyra purpureaGaldieria sulphurariaCyanidioschyzon merolaeCyanidium caldariumCyanophora paradoxaChlorokybus atmophyticusMesostigma virideNephroselmis olivaceaOstreococcus tauriEuglena gracilisChlorella vulgarisStigeoclonium helveticumScenedesmus obliquusChlamydomonas reinhardtiiOltmannsiellopsis viridisPseudendoclonium akinetumBigelowiella natansChara vulgarisStaurastrum punctulatumZygnema circumcarinatumChaetosphaeridium globosumAnthoceros formosaeMarchantia polymorphaPhyscomitrella patensNostoc sp. pcc7120Nostoc punctiforme pcc73102Synechococcus elongatus pcc6301Trichodesmium erythraeum ims101Synechococcus sp. pcc7002Synechocystis sp. pcc6803Crocosphaera watsonii wh8501Thermosynechococcus elongatus bp1Synechococcus sp. ja33abSynechococcus sp. ja23baGloeobacter violaceus pcc7421

Supplementary Figure 2.PHYLOGENETIC inference under the CAT model on 55plastid-encoded proteins (44 OTUs x 10,805 AA) resulted

in a tree that was relatively dated using a log-normal auto-correlated model. In this analysis, chl. c-containing plastidsappear to have diversified at 38.8% of the time elapsed sincethe last common ancestor of Plantae and ’Chromalveolates’.

How to test ’Chromalveolates’?

ChromalveolateHypothesis

Alternative (Serial)Hypothesis

Time

Alveolata

Haptophyta

Stramenopiles

Cryptophyta

other Rhodophyta

Cyanidiales

Streptophyta

Glaucophyta

Opisthokonta

Cyanobacteria

Chlorophyta

0.0 0.2 1.00.4 0.6 0.8

Chlorophyta

Cyanobacteria

Opisthokonta

Glaucophyta

Streptophyta

Cyanidiales

other Rhodophyta

Cryptophyta

Stramenopiles

Haptophyta

Alveolata

plastid / mitochondrion / nucleus histories

Alveolata

Haptophyta

Stramenopiles

Cryptophyta

Streptophyta

Glaucophyta

Opisthokonta

Cyanobacteria

Chlorophyta Chlorophyta

Cyanobacteria

Opisthokonta

Glaucophyta

Streptophyta

Cryptophyta

Stramenopiles

Haptophyta

Alveolata

1

3

21

23

ChromalveolateHypothesis

AlternativeHypothesis

ChromalveolateHypothesis

AlternativeHypothesis

Streptophyta

Glaucophyta

Opisthokonta

Cyanobacteria

Chlorophyta

Haptophyta

Stramenopiles

Cryptophyta

Streptophyta

Glaucophyta

Opisthokonta

Cyanobacteria

Chlorophyta

Cryptophyta

Stramenopiles

Haptophyta1 + 2 + 3

1 + 2 + 3

BOTH Chromalveolate and serial hypotheses postulate aninitial, single secondary endosymbiosis of a red alga

within a eukaryotic host (dotted arrows), leading to the emer-gence of a chl. c–containing founder. The Chromalveolate hy-pothesis then assumes that this alga gives rise to ’Chromalve-olates’ by vertical descent (top left tree). In contrast, serial hy-potheses posit multiple subsequent EEEs (plain arrows), whichhorizontally spread plastids among otherwise unrelated eu-karyotes (top right tree). As chl. c plastids emerge from withinred algae, overall histories of plastid (orange), mitochondrial(blue), and nuclear (black) genomes cannot be superimposed.However, removing red algae ‘regularizes’ plastid history bycreating a single branch out of three smaller ones (middletrees). Now, the Chromalveolate hypothesis predicts that thesignal for the monophyly of ’Chromists’ (discarding alveolates)should be similarly strong across all genomic compartments(orange arrowhead in bottom left tree), while it should bestrong only with plastid genomes in serial hypotheses (orangearrowhead in bottom right tree). To validate our approach,green plants were used as a test case, since the signal for theirmonophyly is expected to be similarly strong, regardless of thecompartment or the hypothesis (green arrowheads).

Greenalgae

Greenalgae

Greenalgae

Greenalgae

Greenalgae

Greenalgae

6 ‘Chromists’

6 green plants

1 glaucophyte 55 plastid 13 mitochondrion encoded proteins108 nucleus

ingroup

outgroup

11 cyanobacteria 11 opisthokonts

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

Preydigestion

USING the same sampling of ingroup species, we assem-bled one concatenated protein data set per compart-

ment. The strength of the phylogenetic signal was then es-timated through a Variable Length Bootstrap (VLB) strategy.

0

20

40

60

80

100

0 2,000 4,000 6,000 8,000 10,000

Boo

tstr

ap s

uppo

rt (

%)

plastid

0 3,000

# positions (AA)

mitochondrion

0 3,000 6,000 9,000 12,000 15,000

nucleus

WHATEVER the compartment considered, the monophylyof green plants is easily recovered, whereas the mono-

phyly of ‘Chromists’ is only recovered with plastid genomes.This observation falsifies the Chromalveolate hypothesis.

How to check our assumptions?

RNA Polymerase1,911 AA

Photosynthesis5,717 AA

Ribosome2,074 AA

Others1,103 AA

Ribosome8,697 AA

Proteasome3,102 AA

Varia3,980 AA

Plastid (10,805 AA)

Nucleus (15,392 AA)

SINCE heterogeneous rates may affect inference, we decidedto subdivide each of our plastid and nuclear data sets into

smaller data sets according to functional class.

THEN, to allow easy com-parison, we defined n70

as the number of positions re-quired to reach a VLB support≥70% for the monophyly ofthe group put to test (by fittinga monomolecular model). 0

20

40

60

80

100

0 400 800 1,200 1,600 2,000

Boo

tstr

ap s

uppo

rt (

%)

# positions (AA)

n70 = 172

VLBs n 70 Values# Pos f(Sub) Trees

Greens ‘Chromists’ Gre ‘Chr’

PlastidPolymerase 1,911 4.79

BS (%

)

365 1,802

Photosynthesis 5,717 1.63

BS (%

)

101 95

Ribosome 2,074 3.56

BS (%

)

208 172

Mitochondrion 3,106 3.51

BS (%

)

176 n.c.

NucleusProteasome 3,102 2.34

BS (%

)

501 n.c.

Ribosome 8,697 2.72

BS (%

)

130 n.c.

Varia 3,980 1.99

BS (%

)

305 n.c.

ALTHOUGH plastid genomes display extreme rate varia-tions, the monophyly of ’Chromists’ is recovered with n70

values that remain small relative to the number of positionsavailable. In contrast, VLB support is so low in mitochondrialand nuclear compartments that it is not even possible to fit themodel. Thus, rate heterogeneity does not impair our test.

0.20Theileria annulataBabesia bovisPlasmodium falciparum

Toxoplasma gondiiEimeria tenella

Cryptosporidium parvumPerkinsus marinusDinophyceae

Sterkiella histriomuscorumTetrahymena thermophilaParamecium tetraurelia

Blastocystis hominisPhytophthora sojaeSaprolegnia parasitica

Nannochloropsis oculataEctocarpus siliculosus

Phaeodactylum tricornutumThalassiosira pseudonana

Pavlova lutheriPrymnesium parvumIsochrysis galbanaEmiliania huxleyi

Chondrus crispusPorphyra yezoensis

G.t. nmGaldieria sulphuraria

Cyanidioschyzon merolaeGuillardia theta

Scenedesmus obliquusChlamydomonas reinhardtiiDunaliella salina

Ostreococcus tauriMicromonas sp.

Mesostigma virideClosterium peracerosum-strigosum-littorale complex

Physcomitrella patensPinus taeda

Cyanophora paradoxaGlaucocystis nostochinearum

Puccinia graminisUstilago maydis

Schizosaccharomyces pombeRhizopus oryzae

Blastocladiella emersoniiSpizellomyces punctatus

Sphaeroforma arcticaCapsaspora owczarzaki

Monosiga brevicollisMonosiga ovataReniera sp.Hydra magnipapillata

Nematostella vectensisAcanthamoeba castellanii

Hartmannella vermiformisMastigamoeba balamuthiDictyostelium discoideum

Physarum polycephalum

71

83

98 40

84

87

99

85

67

33

79

96

82

98

88

76

Figure 2.

Amoebozoa

Opisthokonta

Glaucophyta

Viridiplantae

Cryptophyta

Rhodophyta

Haptophyta

Stramenopiles

Alveolata

FINALLY, using an extended nuclear data set (57 OTUs x15,392 AA), we tested the affinity of the fast-evolving and

compositionally biased nucleomorph of Guillardia theta. In-ference under the CAT model yielded 100% bootstrap sup-port for the grouping of the nucleomorph with red algae,while providing no support for the monophyly of ’Chroma-lveolates’. This helps us to exclude phylogenetic artifacts andto conclude that the Chromalveolate hypothesis is falsified.

Want to know more?BAURAIN D. et al. (2010) Phylogenomic evidence for sepa-rate acquisition of plastids in cryptophytes, haptophytes, andstramenopiles. Mol. Biol. Evol. 27:1698-1709.

SMBE2010, Annual Meeting of the Society for Molecular Biology and Evolution, Lyon, France, July 4–8, 2010