Biosynthesis of Isoprenoids via Mevalonate in Archaea: The Lost Pathway

18 Pages·2000·9.21 MB·English

Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Article
Biosynthesis of Isoprenoids via Mevalonate
in Archaea: The Lost Pathway
Arian Smit1 and Arcady Mushegian2
1InstituteforSystemsBiology,Seattle,Washington98195,USA;2AkkadixCorporation,LaJolla,California92037
Isoprenoid compounds are ubiquitous in living species and diverse in biological function. Isoprenoid side chains
of the membrane lipids are biochemical markers distinguishing archaea from the rest of living forms. The
mevalonate pathway of isoprenoid biosynthesis has been defined completely in yeast, while the alternative,
deoxy-D-xylulose phosphate synthase pathway is found in many bacteria. In archaea, some enzymes of the
mevalonate pathway are found, but the orthologs of three yeast proteins, accounting for the route from
phosphomevalonate to geranyl pyrophosphate, are missing, as are the enzymes from the alternative pathway.
To understand the evolution of isoprenoid biosynthesis, as well as the mechanism of lipid biosynthesis in
archaea, sequence motifs in the known enzymes of the two pathways of isoprenoid biosynthesis were analyzed.
New sequence relationships were detected, including similarities between diphosphomevalonate decarboxylase
and kinases of the galactokinase superfamily, between the metazoan phosphomevalonate kinase and the
nucleoside monophosphate kinase superfamily, and between isopentenyl pyrophosphate isomerases and MutT
pyrophosphohydrolases. Based on these findings, orphan members of the galactokinase, nucleoside monophos-
phate kinase, and pyrophosphohydrolase families in archaeal genomes were evaluated as candidate enzymes for
the three missing steps. Alternative methods of finding these missing links were explored, including physical
linkage of open reading frames and patterns of ortholog distribution in different species. Combining these
approaches resulted in the generation of a short list of 13 candidate genes for the three missing functions in
archaea,whoseparticipationinisoprenoidbiosynthesisisamenabletobiochemicalandgeneticinvestigation.
A challenge for computational biology in the “post- ofsetsofgenesthataremissinginothers(Tatusovetal.
genomic”eraistoreconstructcellularmetabolicpath- 1997; Pellegrini et al. 1999). We are interested in ap-
ways,inasmuchdetailaspossibleandappropriate,by plying all available methods to reconstruct pathways
analysisofgenomesequences.Ifgenesinagivenpath- in poorly characterized species, in order to develop a
way have been characterized in one species, and a highlyautomatedstrategyfacilitatingthisprocessand
newly sequenced species has the complement of the to get an insight in general and specific trends in the
orthologous genes, characterized by high sequence evolutionofbiochemicalpathways.
similarityandconsistentpositioninthephylogenetic More than 25,000 naturally occurring isoprenoid
tree(Tatusovetal.1996;Eisen1998;Yuanetal.1998), derivatives are known, including such important
thenmodelingoftheunknownmetabolismisstraight- classes of bioactive compounds as vitamin A and re-
forward. However, many instances of gaps in meta- lated light-capturing and ultraviolet-protecting carot-
bolic pathways have been reported, especially when enoids,steroidsthatmodifylipidmembranesandpar-
arbitrary similarity cutoffs were imposed in the data- ticipate in signal transduction in eukaryotes, phytols
basesearches.Comparativeproteinsequenceanalysis, that form the membrane-anchoring side chains of
withtheattentiontobothcloseandmoredistantsimi- chlorophyll, and plant protective isoprenoids (Eisen-
larities,remainstheprincipalmethodoffunctionpre- reichetal.1998).Pathwaysofisoprenoidbiosynthesis
diction and pathway reconstruction (Tatusov et al. are perturbed in such human diseases as mevalonic
1996; Koonin et al. 1997; Bork et al. 1998). In the at- aciduria (OMIM Entry 251170) and hyperimmuno-
tempt to break through the “similarity barrier”, globulinaemiaDwithperiodicfeversyndrome(OMIM
complementary approaches have been proposed, in- Entry260920;Houtenetal.1999a,b),butcanbemodi-
cluding identification of genes that are physically fiedtotheadvantageofhumanhealth,asinthecaseof
close,andpresumablycoordinatelyregulated,inmul- inhibition of HMG-CoA reductase by cholesterol-
tiple genomes (Overbeek et al. 1999), and the use of reducing drugs (Moghadasian 1999). From an evolu-
phyleticprofiles,i.e.,theoccurrenceinsomegenomes tionary perspective, isoprenoids are notable as a bio-
chemical marker that distinguishes archaeal lipids
2Correspondingauthor. from bacterial and eukaryotic lipids, which have fatty
[email protected];FAX[phone].
acidsasthesidechains(Kates1993).
Article and publication are at www.genome.org/cgi/doi/10.1101/
gr.145600. Themevalonatepathwayofisoprenoidbiosynthe-
1468 GenomeResearch 10:1468–1484©2000byColdSpringHarborLaboratoryPressISSN1088-9051/00$5.00;www.genome.org
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Isoprenoid Biosynthesis in Archaea and Elsewhere
sis provides isopentenyl pyrophosphate (IPP) and di- of all relevant yeast genes. Genes coding for the en-
methylallyl pyrophosphate (DMAPP), the essential zymes comprising the alternative DXP pathway also
five-carbonbuildingblocksfromwhichthemorecom- are present in higher plants, in agreement with the
plexisoprenoidsareformed.Geneticdeterminantsfor biochemicalevidenceofbothpathwaysingreenalgae
the complete mevalonate pathway for IPP synthesis and in dicots (Disch et al. 1998; Lange et al. 1998;
have been defined in yeast (Saccharomyces cerevisiae), LangeandCroteau1999).
followedbythebiochemicalanalysisoftheindividual Inmostcompletelysequencedbacterialgenomes,
gene products. An overview of the mevalonate path- the mevalonate pathway is missing and is substituted
way and its genetic control in yeast is shown in Fig- bytheDXPpathway.However,insomebacteria,such
ure1. as the completely sequenced spirochete Borrelia burg-
The central role of isoprenoids in living cells has dorferi and the almost completely covered gram-
promptedthesearchfortheenzymesofthemevalon- positivecocci,mostofthemevalonatepathwaygenes
atepathwayindifferentspecies.Biochemicalevidence are found, with the exception of isopentenyl diphos-
has suggested the occurrence of this pathway in vari- phateisomerase.Anothercaseofa“one-bit”difference
ouseukaryotesandinarchaea,anditsapparentsubsti- with the yeast pathway is presented by Metazoa, in-
tutioninbacteriabyanalternativerouteinvolvingbio- cluding the completely sequenced Caenorhabditis el-
synthesis and reductive isomerization of deoxy-D- egansandDrosophilamelanogaster,bothofwhichhave
xylulose phosphate (DXP) (Eisenreich et al. 1998; a single phosphomevalonate kinase unrelated to the
Fig.1). yeastenzyme.Inarchaea,theorthologsforonlythree
We used comparative sequence analysis to show of six committed enzymes could be found, leaving
that most of the species synthesizing isoprenoids lack phosphomevalonatekinase(PMK),diphosphomevalo-
one or more of the orthologs of the yeast mevalonate nate decarboxylase, and isopentenyl diphosphate
pathway genes. In particular, the route from phos- isomerase(IPPI)unaccountedfor.Giventhebiochemi-
phomevalonate to geranyl pyrophosphate, requiring calevidencethatbiosynthesisofthesidechainsofar-
threeseparateenzymesinyeast,hasnotbeenfoundin chaeal lipids proceeds via the mevalonate route, at
archaea. In an attempt to find the missing links in leastinthecaseofhalophiles(Kates1993;Tachibanaet
archaeaandtoreconstructtheevolutionofisoprenoid al. 1996), this gap in the downstream portion of the
biosynthesis,weanalyzedthesequencesoftheknown trunkpathwayneedstobeexplained.
mevalonate pathway proteins, positional linkage of
relevantgenes,andpatternsoforthologdistributionin Detailing the Motifs: New Sequence Relationships
completely sequenced genomes. Frequent functional for Three Enzymes in the MevalonatePathway
takeovers, i.e., gene replacements by a gene with the It has been shown that mevalonate kinases and the
same function but distantly related or unrelated se- yeast PMK belong to a large superfamily, which also
quence, appear to be a major problem in reconstruct- includesgalactokinasesandhomoserinekinases(Bork
ing metabolism of a poorly studied species. In such et al. 1993). Recently, a role in the DXP pathway has
cases, combination of computational approaches en- been proposed for another member of this superfam-
ablessignificantreductionofthespaceofgenecandi- ily, called YchB in E. coli (indicated by 12 in Fig. 1),
dates, providing short lists of targets for biochemical whichiscapableofformingIPPbyphosphorylationof
characterization. an isopentenyl monophosphate (IMP) precursor, and
also phosphorylates the essential intermediate phos-
RESULTS AND DISCUSSION phocytidyl-2-C-methylerythritol (Lange and Croteau
1999; Luttgen et al. 2000). We found that related se-
Defining the Orthologs: Only Higher Plants Have
quencemotifsarepresentindiphosphomevalonatede-
the Full Complement of Yeast Mevalonate carboxylases. A PSI-BLAST search initiated by a puta-
PathwayEnzymes tive mevalonate kinase sequence from Enterococcus
WesearchedDNAandproteinsequencedatabasesfor faecalis, retrieved, at the second iteration, a diphos-
candidate orthologs of yeast mevalonate pathway en- phomevalonate decarboxylase sequence from Arabi-
zymes and Escherichia coli DXP pathway enzymes in dopsis with a probability of matching by chance of
completely sequenced or extensively covered ge- 10–4. Other members of the galactokinase family also
nomes. The results of the ortholog definition are wereobservedinthesesearches.Allkinases(EC2.7)in
showninFigure1. the superfamily transfer a phosphate to a hydroxyl
Inspectionofthecommittedportionofthemeva- group on a substituted tetra-, penta-, or hexacarbon
lonate pathway, leading from acetoacetyl-CoA to the linear scaffold. Diphosphomevalonate decarboxylases
two main isoprenoid precursors IPP and DMAPP (col- work on a similar substrate, but belong to a different
umns 2–7), revealed substantial differences between enzymeclass(EC4.1.1.33).Anexplanationforacom-
evolutionarylineages.Higherplantscontainorthologs monoriginofthesedecarboxylasesandkinasesispro-
Genome Research 1469
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Smit and Mushegian
d.)
n
e
g
e
l
or
f
e
g
a
p
g
n
ci
a
f
e
e
S
(
1
e
r
u
g
Fi
1470 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Isoprenoid Biosynthesis in Archaea and Elsewhere
vided by the understanding of the decarboxylation monophosphate kinases (Teplyakov et al. 1996). The
mechanism,inwhichthe3(cid:1)-positionofdiphosphom- regionofhighestsequencesimilarityinnucleosideki-
evalonateundergoesATP-dependentphosphorylation, nases (motif I in Fig. 2B) also is found in many other
followed by elimination of phosphate and CO (Dhe- ATP-binding proteins where it corresponds to a
2
Paganon et al. 1994). Thus, affinities for ATP and a Walker-typeP-loop(Koonin1993).Itismodifiedinthe
substituted aliphatic chain are the common denomi- bacteriophagekinaseandinmetazoanphosphomeva-
natorsforthewholesuperfamily. lonate kinases, but the essential lysine residue is con-
Multiple sequence alignment (Fig. 2A) revealed served (Fig. 2B). The C-terminal motif II in T4 deoxy-
four conserved sequence motifs in the galactokinase nucleosidemonophosphatekinaseconsistsofastrand,
superfamily. A role in binding of ATP gamma- aturn,andahelix,whichcomprisethegraspholding
phosphatehasbeensuggestedforthelysine-13residue the adenine moiety of the bound nucleoside phos-
of rat mevalonate kinase, based on affinity mapping phate(Teplyakovetal.1996).Onlymarginalsequence
andsite-directedmutagenesis(Potteretal.1997a).This similaritybetweennucleosidemonophosphatekinases
residue,foundwithinmotifI(markedbyanasteriskin andPMKwasdetectedinthemiddle,substrate-binding
Fig.2A),isconservedinmanykinasesbutreplacedin regions of these proteins, reflecting the difference in
diphosphomevalonate decarboxylases and in a subset thephosphorylationtargets.
ofuncharacterizedarchaealproteins(Fig.2A).Another Another new sequence relationship was observed
residue in rat mevalonate kinase, aspartate-204, is in the case of IPPI. At the second PSI-BLAST iteration,
thoughttoserveasthebasethatfacilitatesprotonex- statisticallysignificantmatches(p<10–6)weredetected
traction from the substrate (Potter et al. 1997b). This between the C-terminal halves of IPPI and the MutT
amino acid (asterisk in motif 3, Fig. 2A) is substituted familyofnucleosidepyrophosphatases(Fig.2C).Mul-
by a similarly nucleophilic serine in decarboxylases. tiple sequence alignment of MutT-like proteins and
Glycine-rich loops commonly form the phosphate- IPPI, and analysis of the known solution structure of
binding sites in ATP-dependent enzymes (e.g., Bork the E. coli MutT protein indicate that the conserved
andKoonin1994),andmotifs2and4arenotablecan- motifs correspond to the alpha helix I and connected
didatesforsuchafunction.Verificationofthespecific loops(Mildvanetal.1999).IntheMutTprotein,these
roles of the four conserved motifs awaits the site- structural elements are involved in interactions with
directed mutagenesis of additional representatives of theboundnucleotideandwithdivalentmetalcations
thisfamilyandcrystallographicstudies. essential for the nucleophilic attack on the pyrophos-
The known PMK enzymes encoded by metazoan phate bond. Although a pyrophosphate bond is pres-
genomesareunrelatedtothegalactokinasefamily,and ent in isopentenyl pyrophosphate, its hydrolysis
their evolutionary origin was unclear. We found that would be unexpected of and not sufficient for the
metazoan PMK are distantly related to nucleoside isomerase reaction. Possibly, MutT-like motifs in the
monophosphatekinasesfoundinallsuperkingdomsof IPPI enzymes are catalytically inefficient and are used
lifeandinviruses.Thedatabasesearchesrevealedmod- to bind the pyrophosphate moiety of the substrate,
erate similarity (p = 10–3 upon first-time passing the whiletheisomerizationrequirestheaidoftheunique
cutoffinPSI-BLASTanalysis)betweenanimalPMK,un- N-terminaldomain.
characterized proteins encoded by the catfish herpes
virus, and the deoxynucleoside monophosphate Reconstructing the Pathway in Archaea: Paralogous
(NMP) kinase of bacteriophage T4. The latter protein Candidates for MissingFunctions
shares significant sequence similarity and the same Archaeal genomes encode members of each of the
three-dimensional fold with a number of nucleoside threenewlydelineatedsuperfamilies.Forexample,the
Figure1 Themevalonateanddeoxy-D-xylulose(DXP)pathwaysofisoprenoidbiosynthesis.GenenamesandGenBankaccessionnos.
for the prototype yeast proteins of the mevalonate pathway are shown. Green shading indicates genes orthologous to the yeast
prototypes.Genedisplacementsareshowninyellowor,whenthereplacingenzymeshavenotbeencharacterized,inred.Blueshading
indicatestheenzymesoftheDXPpathway.Noshadingindicatesthatthesefunctionsaremorelikelytobeabsentinagivengenus.
CompoundsareindicatedbyRomannumerals:I,acetyl-CoA;II,acetoacetyl-CoA;III,hydroxy-3-methylglutaryl-CoA;IV,mevalonate;V,
phosphomevalonate;VI,diphosphomevalonate;VII,isopentenylpyrophosphate;VIII,dimethylallylpyrophosphate;IX,geranylpyrophos-
phate;X,pyruvate;XI,glyceraldehyde3-phosphate;XII,2-deoxy-D-xylulose5-phosphate;XIII,2C-methyl-D-erythritol4-phosphate;XIV,
4-diphosphocytidyl-2C-methyl-D-erythritol; XV, 4-diphosphocytidyl-2C-methyl-D-erythritol 2-phosphate; XVI, 2C-methyl-D-erythritol
2,4-cyclodiphosphate; XVII, isopentenyl monophosphate. Arabic numerals indicate enzymes. Mevalonate pathway: 1, acetoacyl-CoA
synthetase;2,hydroxy-3-methylglutaryl-CoAsynthase;3,hydroxy-3-methylglutaryl-CoAreductase;4,mevalonatekinase;5,phosphom-
evalonate kinase; 6, diphosphomevalonate decarboxylase; 7, isopentenyl pyrophosphate delta-isomerase; 8, geranyl pyrophosphate
synthasefamily(the†signindicatesthatorthologsandparalogsarenotwelldistinguishedinthisfamily,whichiscompatiblewiththe
observation that substrate specificity of these enzymes is modulated easily by small number of point mutations). DXP pathway: 9,
deoxy-D-xylulosephosphatesynthase;10,deoxy-D-xylulosephosphatereductoisomerase;11,2C-methyl-D-erythritol4-phosphatecy-
tidylyltransferase(YgbP);12,isopentenylmonophosphatekinase;13,2C-methyl-D-erythritol2,4-cyclodiphosphatesynthase(YgbB).
Genome Research 1471
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Smit and Mushegian
d.)
n
e
g
e
l
or
f
e
g
a
p
g
n
ci
a
f
e
e
S
(
2
e
r
u
g
Fi
1472 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Isoprenoid Biosynthesis in Archaea and Elsewhere
Figure2 Conservedsequencemotifsinmevalonatepathwayenzymes.Blocksofhigh-sequencesimilarityareshown.Uniqueidentifiers
inSWISSPROTorGenBankaregivenforeachsequence.Yellowshadingindicatesconservedbulkyhydrophobicresidues(I,L,F,M,V,Y,
andW),redtypeindicatesconservedsmallsidechainresidues(A,G,andS),andbluetypeindicatesotherconservedresidues.Secondary
structures predicted with reliability of eight or higher (PHD program) are shown; h indicates a helix, and s indicates a strand. (A)
Diphosphomevalonatekinasebelongstothegalactokinasesuperfamily.Secondarystructuresforyeastphosphomevalonatekinase(ERG8)
anddiphosphomevalonatedecarboxylase(ERG19)predictedwithreliabilityofeightorhigherareshown.(B)ConservedATP-binding
motifsofnucleotidemonophosphatekinasetypeinmetazoanphosphomevalonatekinases.Secondarystructureelementsobservedinthe
T4 bacteriophage deoxynucleoside monophosphate kinase (pdb code 1DEL) and predicted for human phosphomevalonate kinase
(PMKA_HUMAN)areshown.Greenshadingindicatesresidueslocatedwithin3ådistancefromtheboundADP.(C)MutT-likepyrophos-
phate-binding motifs in isopentenyl pyrophosphate delta-isomerases. Secondary structure elements observed in Escherichia coli MutT
protein(pdbcode1TUM)andpredictedforyeastIPPI(IDI1_YEAST)areshown.
genome of Methanococcus jannaschii codes for at least domain rearrangements between proteins. Therefore,
five members of the galactokinase superfamily, six blocksofhigh-sequencesimilaritywerecollectedfrom
members of the nucleoside monophosphate kinase thealignmentsofeachofthethreesuperfamilies,and
family (as well as a large number of remotely related phylogenetictreeswereconstructedfromfourconcat-
kinases and ATPases with Walker-type P-loops), and enated blocks, in order to trace the evolutionary rela-
one member of the MutT/IPPI superfamily. The bio- tionships of some of the “orphan” members of the
chemical functions of these proteins cannot be estab- familiesofinterest.
lished by database searches alone; nevertheless, for a Using this approach, we were able to reliably re-
subset of archaeal proteins, orthologous relationships solve most branches in the galactokinase superfamily
with biochemically characterized proteins from other tree(bootstrapsupport75%orhigher).Theknownga-
species could be reliably inferred by analysis of best lactokinases,homoserinekinases,mevalonatekinases,
matches.Inmanycases,however,orthologdefinition phosphomevalonate kinases, isopentenyl monophos-
wascomplicatedbynontransitiverelationshipsand/or phate kinases, and diphosphomevalonate decarboxyl-
Genome Research 1473
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Figure 3 (A) Phylogenetic tree of the galactokinase superfamily. The results of neighbor joining analysis are shown, which fully
correspondtothemaximumlikelihooddata.Thetreewasbuiltwith116galactokinasesuperfamilymembers,all<90%identicaltoone
another. Forty-seven less-informative proteins have been weeded out afterwards. Thick lines indicate a bootstrap value >75% for the
corresponding node. Eukaryotic branches are in blue, eubacterial in green, and archaeal in red lines. Three orphan archaeal-specific
familieswithinthesuperfamilystandout,andmayincludephosphomevalonatekinases,isopentenylmonophosphatekinases,and,less
likely, diphosphomevalonate decarboxylases (see text).(B) Blowup of the central region of the tree in Fig. 3A. Numbers indicate the
bootstrapsupport,outof1000replicates,forthedeepbranchesinthetree.
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Isoprenoid Biosynthesis in Archaea and Elsewhere
Table1. ShortListofCandidatesthatMayReplacethe“Missing”EnzymesofMevalonatePathwayinArchaea,Borrelia,
andGram-PositiveCocci
Missingfunction Suggestedcandidates Criteriaforselection Comments
Phosphomevalonate Familyofproteinsrelatedto Linkedtomevalonatekinasein Amorelikelycandidate.
kinase(archaea) uridylateandacetylglutamate allarchaea,similarlytothe
kinases linkageobservedbetween
3258052_PyrhoY044_METJA mevalonatekinaseand
2621082_Metth2648231_Arcfu phosphomevalonatekinasein
5105458_Aerpe Borelliaandcocci;isolated
phylogeneticposition.
Familyoforthologswithin Sequencesimilarityandisolated
galactokinasesuperfamily phylogeneticpositionwithin
3258263_Pyrho3024949_Metja thesuperfamily.
2621655_Metth2648904_Arcfu
5105835_Aerpe
Familyoforthologswithin Sequencesimilarityandisolated Noorthologsintwoofthree
galactokinasesuperfamily phylogeneticpositionwithin Pyrococcusspecies.
2128859_Metja2621895_Metth thesuperfamily.
2850101_Arcfu5105835_Aerpe
Familyoforthologswithin Sequencesimilarityandisolated P.horikoshiiandAeropyrumhave
galactokinasesuperfamily phylogeneticpositionwithin asetofparalogs,generally
YE27_METJA3256616_Pyrho thesuperfamily. notseenformevalonate
3257645_PyrhoY830_METTH pathwayenzymes.
YK89_ARCFU5106179_Aerpe
5105197_Aerpe
Familyoforthologswithin Sequencesimilarityandisolated ClosertoCMP/UMPkinasesand
nucleosidemonophosphate phylogeneticpositionwithin mayberequiredforthe
kinasesuperfamily thesuperfamily. nucleotidekinasefunction
3257422_Pyrho2128835_Metja (judgedfromthenumbersof
2621499_Metth2649179_Arcfu paralogswithsuchspecificity
inbacterialgenomes).
Diphosphomevalonate Homologs(probablyparalogs)of Eliminationofother Relatedgenesinsomebacteria,
decarboxylase SAMdecarboxylasefromE.coli decarboxylasesinarchaea, butaninternaldeletionin
(archaea) 5103467_Aerpe2648951_Arcfu eitherasmisannotations,or archaealgenessuggeststhat
2495908_Metja3258436_Pyrho becausetheyarepredictedto thesubstrateinarchaeamay
haveaclearlyunrelated bedifferent.Nohomologin
biologicalfunction. M.thermoautotrophicum.
Isopentenyl FMN-dependentdehydrogenases, InBorrelia,foundwithinan Oxidoreductaseactivity
pyrophosphate includingglycolateoxidase “operon”withallother predictedbysequence
delta-isomerase (NCBICOG1304) enzymesofmevalonate similarity.Inplants,some
(archaea,Borrelia, 2688617_Borbu5105455_Aerpe pathway;neighborswith oxidoreductasesareknownto
cocci) 2648236_Arcfu1591547_Metja relevantfunctionsinsome haveadesaturaseactivity,
2621084_Metth3257619_Pyrho archaea.Consistentphyletic which,inthiscase,couldbe
pattern,althoughparalogsare exploitedtoisomerizethe
widespread. doublebondsinIPP.
Uncharacterizedfamily(NCBI Exceptionallyconsistentphyletic InEnterococcus,plasmid-borne
COG1916) patterninmicroorganisms copyisinvolvedinresponse
2688316_Borbu2650316_Arcfu (Borrelia,Enterococcus,only topeptidepheromone,which
2129184_Metja2622291_Metth oneadditionalbacterium, controlsconjugation.
3257569_Pyrho Treponema,andarchaea Orthologsinplants,C.elegans
exceptforAeropyrum). andhumans.
Uncharacterizedfamily(NCBI Phyleticpattern(archaea+ Presenceofseveralinvariant
COG0327) Borrelia),althoughorthologs histidinessuggeststhatthis
2688374_Borbu2648773_Arcfu arefoundinmanybacteria. maybeametalloenzyme.
Y927_METJA726074_Metth
3257033_Pyrho
Uncharacterizedfamily(NCBI Phyleticpattern(archaea+ Glycine-rich(phosphate-
COG0061) Borrelia),althoughorthologs binding?)motifsharedwith
2688218_Borbu5104774_Aerpe arefoundinmanybacteria. diacylglycerolkinasesand6-
2650718_Arcfu2621965_Metth phosphofructokinases.
2826350_Metja3257490_Pyrho
Undecaprenyldiphosphate Phyleticpattern(archaea+ Multiplefunctionsinthesame
synthase(NCBICOG0020) Bacteriaexceptfor pathwaywouldhavetobe
UPPS_BORBU5105068_Aerpe Mycoplasmae),notethatthe assumed;nobiochemical
UPPS_ARCFUUPPS_METJA DXPS-pathwaybacteriamay evidence.
UPPS_PYRHO needsuchactivity,too(see
text).
(Continuesonfollowingpage)
Genome Research 1475
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Smit and Mushegian
Table1. (Continued)
Missingfunction Suggestedcandidates Criteriaforselection Comments
Geranylgeranylpyrophosphate Phyleticpattern(archaea+ Multiplefunctionsinthesame
synthasebeta-subunit Bacteriaexceptfor pathwaywouldhavetobe
2688215_Borbu5105454_Aerpe Mycoplasmae),notethatthe assumed;nobiochemical
4633648_ArcfuIDSA_METJA DXPS-pathwaybacteriamay evidence.
IDSA_METTH3257488_Pyrho needsuchactivity,too(see
text).
TPR-repeatproteinsincluding Phyleticpattern(archaea+ Multiplefunctionsinthesame
geranylgeranylpyrophosphate Bacteriaexceptfor pathwaywouldhavetobe
synthasealpha-subunit(large Mycoplasmae),notethatthe assumed;nobiochemical
NCBICOG0457) DXPS-pathwaybacteriamay evidence.
needsuchactivity,too(see
text).
aseseachformadistinctclusteronthetree.Inthecase seems a less likely candidate for phosphomevalonate
of the five M. jannaschii proteins in this superfamily, kinase function in archaea. Representative proteins,
theinitialassignmentsofgi2497515tothehomoser- however,areprovisionallyincludedintheshortlistof
ine kinase family and gi 2497517 to the mevalonate alternativesinTable1.
kinase family1 were confirmed. The three remaining Suggesting a candidate for the diphosphomevalo-
paralogs(gi2128859,gi3024949,andgi3183371)be- nate decarboxylase function in archaea based on se-
long to ancient, archaea-specific clusters. Distribution quencesimilaritytotheyeastprototypeisnolesschal-
of orphan paralogs in other archaea is very similar to lenging. Orphan paralogs remain in the galactokinase
whatisobservedinM.jannaschii(Fig.3). superfamily even after reserving one homolog per ar-
Theexactbiochemicalfunctionofanymemberof chaealgenomeforthephosphomevalonatekinaseac-
theseclustersisunknown,anditispossiblethatoneof tivity. Arguing against these candidates, however, is
themisthemissingPMK(or,inaconvergencewiththe the fact that the C-terminal halves of eukaryotic di-
DXP pathway, the IMP kinase). None of the paralogs phosphomevalonate decarboxylase sequences are
groups together with either the PMK or IMP kinase unique for this family of enzymes, and they do not
(IMPK)families;infact,twooftheparalogsareslightly retrieve any other classes of sequences in database
closer related to homoserine kinases. Thus, if one of searches. Thus, it is not clear whether any orphan
these proteins indeed performs either PMK or IPMK memberofthegalactokinasesuperfamilyinarchaeais
function,thiswouldmeanthatmembersoftwodiffer- endowed with all sequence elements required for the
ent families within the same superfamily have been diphosphomevalonatedecarboxylaseactivity.
recruitedtoperformthesamebiochemicalreactionon Yetanothermissingstepofthemevalonatepath-
at least two independent occasions. However, this as- way in archaea is isopentenyl delta-pyrophosphate
sumption is plausible, as similar scenarios have been isomerase. Although MutT-like domains in IPPI en-
documented for other enzymes, including sugar ki- zymes were detected in this study, and MutT-domain
nasesfromtheunrelatedfamilies(Borketal.1993)and proteins are found in archaea (Fig. 2C), all of the ar-
aminoacyl-tRNA synthetases (Koonin and Aravind chaeal proteins are shorter than the known IPPI pro-
1998).Theidentifiersofthesearchaealcandidatesfora teins,whichcontainapparentlyuniquedomainsinad-
missingkinaseactivityarelistedinTable1. dition to the MutT-like pyrophosphate-binding sites.
Could an alternative PMK in archaea have been Thus, the IPPI function is still unaccounted for in ar-
recruitedfromtheNMPkinasesuperfamily?Members chaea. IPPI orthologs also are missing from the com-
of this superfamily are present in each of the com- pletely sequenced genome of B. burgdorferi and from
pletelysequencedarchaealgenomes,buttheytypically the extensive sequence databases of gram-positive
aremoresimilartoNMPkinasesthantothetaxonomi- cocci.
cally heterogeneous group of PMK-like proteins (data
not shown). No conservation between the substrate-
Alternative Approaches Towards Detection
binding domains of the latter group of proteins and
of the “Missing Link”—Combining theEvidence
the middle regions of the archaeal paralogs could be
Detectionofhomologsandpartitioningthemintoor-
detected. Thus, an orphan NMP kinase-like protein
thologs and paralogs are the first steps in the recon-
1Theseandsomeotherginumbersrelatetotheresubmissionofthepro- structionofmetabolicpathwaysinabiochemicallyun-
teins to the NR database as the cured SWISS-PROT entries; sequential characterized species. Whenever clear candidates can-
numberingoftheentriesdoesnotimplytheirlinkageinthegenomeunless
specificallyindicated. not be identified by such approach, alternative
1476 Genome Research
www.genome.org
Downloaded from genome.cshlp.org on January 1, 2023 - Published by Cold Spring Harbor Laboratory Press
Isoprenoid Biosynthesis in Archaea and Elsewhere
strategies are needed. It has
beensuggestedthat,assecond-
ary and tertiary structures of
proteins may be better con-
served than their sequences,
these higher-order structures,
known or predicted, should be
compared directly (Aurora and
Rose 1998; Pennec and Ayach
1998; Lehtonen et al. 1999).
For example, proteins with an
experimentally determined
MutT-like fold, unrecognizable
atthesequencesimilaritylevel,
would be attractive candidates
for the IPPI function. Unfortu-
nately, very few three-dimen-
sional structures of archaeal
proteins have been resolved,
whileuseofinferredsecondary
structures of proteins for the
same purpose produced am-
biguous results (e.g., Aurora
and Rose 1998; A.R. Mush-
egian,unpubl.). Figure4 Conservedstringsincludegenesofthemevalonatepathwayinarchaeaandbacteria.
Recently, additional tech- Blocksconnectedbyanarrowindicateneighboringgeneswithacommontranscriptionalori-
entation,possiblyrepresentingoperons.Blockscontainingnumbersrepresentknownmevalo-
niques of inferring functional
natepathwaygenes,asinFigure1.Blocksdesignatedas2indicatehydroxy-3-methylglutaryl-
links between proteins were CoA synthase; 3 indicates 3-hydroxy-3-methylglutaryl-CoA reductase; 4 indicates mevalonate
suggested. Enright et al. (1999) kinase;5indicatesyeast-likephosphomevalonatekinase;6indicatesdiphosphomevalonatede-
carboxylase;and8indicatesoctaprenyl-diphosphatesynthase,amemberofthegeranylpyro-
andMarcotteetal.(1999)com-
phosphate synthase family. Other designations: A indicates ancient conserved protein (COG
piled databases of orthologous
#1355),Kindicatesputativekinaserelatedtouridylate-andacetylglutamatekinases,Cindicates-
domains that exist as stand- carotenoid biosynthesis protein (flavin-dependent oxidoreductase), and H indicates putative
alone open reading frames metal-dependent hydrolase. Pyrococcus abyssi has the same structure as Pyrococcus horikoshii,
withonegeneinsertionbetweenancientconservedproteinandmevalonatekinase.InStrepto-
(ORFs) in some species, yet are
coccuspyogenes,genes2and3areflankingthemevalonatekinaseoperon,butaretranscribed
fused into multidomain pro- inoppositeorientation.ThemevalonatekinasegenehasnotbeensequencedyetinSulfolobus
teins in others. These authors solfataricus. The putative metal-dependent hydrolase has no orthologs in Aeropyrum and Ar-
reasonedthatafunctionforan chaeoglobus. GenBank identification nos. are given below the boxes, where available. Appar-
ently-missingGInumbersinstringscorrespondtooverlappinggenes,typicallyshortopenread-
uncharacterized domain may
ingframes(ORFs)ontheoppositestrand.
bepredictedbasedonitsfusion
toabetter-studieddomain(the
Rosettastoneapproach).Analysisofdomainorganiza- 1999). At the other extreme, several small sets of or-
tionwaspartofourorthologdefinitionprocedure(see thologousgenesarearrayedinthesameorderinmost
Methods). Notwithstanding the two domain fusions bacteriaandarchaea,andanalysisoftheproductsen-
noted previously, one in the mevalonate pathway, codedbytheseuniversallyconservedoperonssuggests
wheremammalianHMG-CoAreductasescontainaddi- that they code for the subunits of stoichiometric pro-
tional conserved membrane domain implicated in tein complexes (Mushegian and Koonin 1996; Dan-
cholesterolsensing(Tsengetal.1999),andanotherin dekaretal.1998).Onthemediumevolutionaryscale,
DXP pathway, where some bacteria have YgbP and relative stability of gene linkages may indicate func-
YgbBfusedintoapolyprotein(Herzetal.2000),wedid tionalcoupling(Overbeeketal.1999).
not find any “Rosetta stones” that could help to find We analyzed ORFs located close to the known
missingcandidatesinarchaeaorBorrelia. genesofmevalonatepathwayinbacterialandarchaeal
Analysis of gene colinearity in completely se- genomes. Interestingly, in Borrelia, an operon-like
quencedmicrobialgenomeshasshownthatthelong- string(gi2688613–gi2688618)containsfiveORFsbe-
range conservation of the order of orthologous genes longingtothemevalonatepathway,andalsothesixth
on a chromosome is observed only between closely product (BB0684; gi 2688617), annotated in the data-
related species (Tatusov et al. 1996; Overbeek et al. base as a carotenoid biosynthesis protein. The latter
Genome Research 1477
www.genome.org