a functionally divergent hydrogenosomal peptidase with protomitochondrial ancestry

A functionally divergent hydrogenosomal peptidasewith protomitochondrial ancestry

Mark T. Brown,1 Heather M. H. Goldstone,2

Felix Bastida-Corcuera,1 Maria G. Delgadillo-Correa,1

Andrew G. McArthur2 and Patricia J. Johnson1*1Department of Microbiology, Immunology and MolecularGenetics, University of California, Los Angeles, 609Charles E. Young Drive East, Los Angeles, CA90095-1489, USA.2Josephine Bay Paul Center for Comparative MolecularBiology and Evolution, Marine Biological Laboratory, 7MBL Street, Woods Hole, MA 02543-1015, USA.

Summary

Matrix proteins of mitochondria, hydrogenosomesand mitosomes are typically targeted and translocatedinto their respective organelles using N-terminalpresequences that are subsequently cleaved by apeptidase. Here we characterize a ~47 kDa metallo-peptidase, from the hydrogenosome-bearing, unicel-lular eukaryote Trichomonas vaginalis, that containsthe active site motif (HXXEHX76E) characteristic of theb subunit of the mitochondrial processing peptidase(MPP) and localizes to hydrogenosomes. The purifiedrecombinant protein, named hydrogenosomal pro-cessing peptidase (HPP), is capable of cleaving ahydrogenosomal presequence in vitro, in contrast toMPP which requires both an a and b subunit for activ-ity. T. vaginalis HPP forms an ~100 kDa homodimer invitro and also exists in an ~100 kDa complex in vivo.Our phylogenetic analyses support a common originfor HPP and bMPP and demonstrate that geneduplication gave rise to aMPP and bMPP before thedivergence of T. vaginalis and mitochondria-bearinglineages. These data, together with published analy-ses of MPPs and putative mitosomal processingpeptidases, lead us to propose that the length of tar-geting presequences and the subunit composition oforganellar processing peptidases evolved in concert.Specifically, longer mitochondrial presequences mayhave evolved to require an a/b heterodimer for accu-rate cleavage, while shorter hydrogenosomal andmitosomal presequences did not.

Introduction

A variety of microaerophilic, unicellular eukaryotes lackmitochondria and instead contain energy producingorganelles called hydrogenosomes (Dyall et al., 2004a;Martin, 2005; Dolezal et al., 2006; Embley, 2006; Hack-stein et al., 2006). These double membrane-boundedorganelles in Trichomonas vaginalis and a relatedorganelle called the mitosome in Giardia lamblia (Tovaret al., 2003; Regoes et al., 2005), are the sites of FeScentre biosynthesis (Sutak et al., 2004) like their mito-chondrial counterparts. Hydrogenosomes also generateATP through the metabolism of pyruvate, albeit using ananaerobic pathway absent in mitochondria and mito-somes, which results in molecular hydrogen production.Determining whether hydrogenosomes and mitosomesarose from the same a-proteobacterial endosymbiont thatgave rise to mitochondria has stimulated much debate(Dyall et al., 2004b; Hrdy et al., 2004; Martin and Embley,2004; Gray, 2005; Embley and Martin, 2006), and hasbeen hampered by the absence of organellar genomes(Gray et al., 1999). Thus, assessing the phylogeny ofthese organelles must rely on analyses of nuclear-encoded, organellar proteins.

The ability to translocate proteins into an organellewould be a prerequisite for conversion of an endosym-biont to an organelle. Thus, organellar biogenesis pro-vides insight into the origin of organelles, as those with acommon origin will share certain biogenetic properties.Nuclear-encoded hydrogenosomal, mitosomal and mito-chondrial proteins are synthesized in the cytosol and sub-sequently directed to the organelle. All three organellestypically use an N-terminal presequence to target andtranslocate proteins into the organelles (Bradley et al.,1997; Williams et al., 2002; Pfanner et al., 2004; Wiede-mann et al., 2004; Dolezal et al., 2006). These N-terminalpresequences are cleaved by a processing peptidase togive rise to the mature protein. Hydrogensomal, mito-chondrial and mitosomal presequences have a similarbiochemical composition, notably the conservation of Argat the -2 or -3 position relative to the cleavage site andthe presence of Leu and Ser residues (Bradley et al.,1997; Gakh et al., 2002). In mitochondria, the processingpeptidase has been well characterized (Wiedemannet al., 2004), but little is known about this activity in hydro-genosomes or mitosomes. T. vaginalis hydrogenosomal

Accepted 24 March, 2007. *For correspondence. E-mail [email protected]; Tel. (+1) 310 825 4870; Fax (+1) 310 206 5231.

Molecular Microbiology (2007) 64(5), 1154–1163 doi:10.1111/j.1365-2958.2007.05719.x

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd

lysates have been shown to cleave mitosomal precursorproteins, suggesting the presence of a conserved pepti-dase in all mitochondria-related organelles (Dolezal et al.,2005).

The mitochondrial processing peptidase (MPP) thatcleaves the N-terminal presequence is a heterodimercomposed of two structurally related proteins (Tayloret al., 2001), bMPP and aMPP. The bMPP subunit con-tains the zinc-binding active site motif HXXEHX76E foundin metallopeptidases of the pitrilysin family (Luciano et al.,1998) and directly mediates catalysis. The secondsubunit, aMPP, plays a role in substrate recognition(Kitada et al., 2003) and contains a conserved glycine-rich loop region that has been associated with release ofthe cleaved substrate (Nagao et al., 2000). Biochemicalstudies have shown that a and bMPP together form thefunctional peptidase and that neither subunit is activealone (Luciano et al., 1997).

Mitochondrial processing peptidases have been highlyconserved between yeast and mammals. The a and bsubunits exhibit ~36% and ~45% amino acid identity,respectively, over this distance (Gakh et al., 2002). More-over, there is a 20–30% identity between the a and bMPPsubunits, and aMPP has retained a degenerate active sitemotif. Their similarity has led to the hypothesis that bothsubunits originated from a single gene that underwentduplication and divergence (Pollock et al., 1988).

The first bacterial MPP-like protein was recently char-acterized from Rickettesiae prowazekii (Kitada et al.,2007), an a-proteobacterium that is evolutionarily relatedto the endosymbiont that gave rise to mitochondria(Andersson et al., 1998; Gray, 1998). This protein, namedrickettsial putative peptidase (RPP), is encoded by asingle gene and is similar to both a and bMPP (Kitadaet al., 2007), consistent with the hypothesis that the twomitochondrial MPP subunits arose from a single gene.

Here we identify and describe the T. vaginalis hydro-genosomal processing peptidase (HPP) and show thatrecombinant HPP is an active metallopeptidase com-posed of two identical bHPP subunits. Phylogeneticanalyses indicate that HPP and MPP have a commonancestry and reveal that a gene duplication gave rise toaMPP and bMPP before the divergence of mitochondriaand hydrogenosomes.

Results

Search for T. vaginalis bMPP and aMPP homologues

Nuclear-encoded hydrogenosomal and mitochondrialmatrix proteins are typically targeted to these organellesvia N-terminal extensions that are proteolytically cleavedupon translocation into the organelle (Bradley et al., 1997;Williams et al., 2002; Pfanner et al., 2004; Wiedemann

et al., 2004; Dolezal et al., 2006). To determine whetherT. vaginalis uses a homologue of MPP to process hydro-genosomal proteins, we searched the 7.2 ¥ coverage ofthe T. vaginalis genome database (http://tigrblast.tigr.org/er-blast/index.cgi?project=tvg) for sequences similar to aand bMPP using BLAST (Carlton et al., 2007). We foundone protein containing a putative N-terminal targeting pre-sequence (MSIISRY) and the metallopeptidase active sitemotif (HXXEHX76E) which shares 24% identity and 53%similarity to Saccharomyces cerevisiae bMPP (Fig. S1).We amplified and cloned the complete predicted openreading frame (ORF) encoding a ~47 kDa protein. Subse-quent Northern blot analysis showed that the gene,named bHPP, is expressed (data not shown).

The best hit obtained by BLAST analyses of theT. vaginalis genome database using yeast aMPP wasalso the bHPP protein (E-value = 7.7e-17). An additionalprotein that contains a glycine-rich region similar to astrictly conserved motif found in all aMPPs was also found(Fig. S2). However, none of the other conserved M16peptidase domains (Luciano et al., 1998) found in a andbMPPs are present in this protein and it was not possibleto make a robust full-length alignment to any eukaryoticMPP or related bacterial M16 family peptidases. Analysisof the N-terminus of the protein did not identify a putativehydrogenosomal presequence. The ORF encodes a~46 kDa protein and Northern blot analysis confirmed thatthe gene is expressed (data not shown). Given the pres-ence of the glycine-rich loop motif, the protein was namedthe glycine rich loop protein (GRLP).

We also attempted to identify an aMPP homologue inT. vaginalis by a hidden Markov model (HMM) methodusing the Pfam database (Finn et al., 2006). This analysisdid not identify GRLP as an aMPP homologue nor did itidentify any other candidate proteins. In contrast, repeat-ing the same analyses for bMPP clearly identifiedT. vaginalis bHPP (data not shown). Although GRLP is notpredicted to be an aMPP homologue, we could notexclude the possibility that it is a highly divergent homo-logue of aMPP. As all examined MPPs require both sub-units for activity, we included GRLP in our subsequentanalyses.

Subcellular targeting of bHPP and GRLP in T. vaginalis

To assess the cellular localization of bHPP and GRLP inT. vaginalis, C-terminally HA-tagged bHPP (bHPP-HA)and GRLP (GRLP-HA) were expressed in T. vaginalis.Immunostaining of the transformants showed thatbHPP-HA colocalized with the hydrogenosomal markerHsp70, demonstrating its presence in hydrogenosomes(Fig. 1). Despite the absence of a typical N-terminalhydrogenosomal presequence, immunostaining of theGRLP expressing transformants showed that this protein

Trichomonas hydrogenosomal processing peptidase 1155

© 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 64, 1154–1163

http://tigrblast.tigr.org/er-blast/index.cgi?project=tvg

http://tigrblast.tigr.org/er-blast/index.cgi?project=tvg

is also found in the hydrogenosome (Fig. 1). The lack ofperfect colocalization of these hydrogenosomal pro-teins is likely the result of variability in membranepermeabilization. This might also be explained by someorganelles lacking Hsp70 and others lacking HPP and/orGRLP, however, we think this less likely.

bHPP and GRLP are not in the same hydrogenosomalcomplex

To determine whether endogenous GRLP and bHPPassociate in the same complex, as predicted if they arethe functional equivalents of a and bMPP, wild-type (wt)hydrogenosomal extract was fractionated on a 2–20%sucrose gradient and the resulting fractions subjected toimmunoblot analysis using antibodies raised againstrecombinant bHPP-His and recombinant GRLP-His.bHPP sedimented below the 140 kDa marker (Fig. 2A),indicating that it exists in a higher molecular weightcomplex, possibly a dimer. The majority of endogenousGRLP did not cofractionate with bHPP, but instead sedi-mented with a predicted mass of ~67 kDa. The minimaloverlap in the sedimentation of the two proteins demon-strates that they do not form a stable complex, unlike a

and bMPP (Ou et al., 1989; Luciano et al., 1998). On theother hand, the sedimentation profile of both proteinsindicates neither exist solely as a free monomer in thehydrogenosome, as both migrate in SDS-PAGE as47 kDa proteins (Fig. 3A; data not shown).

To further characterize endogenous bHPP and GRLPusing an independent method, we subjected wt hydro-genosomal lysate to size exclusion chromatography(SEC). Immunoblot analyses of the eluted fractions usinganti-bHPP and anti-GRLP antibodies revealed the elutionof the proteins in different peaks, with calculated molecu-lar masses of 100 kDa and 67 kDa for bHPP and GRLP,respectively (Fig. 2B). These data are consistent withthose obtained using sucrose gradient fractionation andconfirm that the two proteins exist in independentcomplexes.

bHPP forms a homodimeric complex

Consistent with the observation that bHPP exists in a100 kDa molecular weight complex devoid of GRLP, weobserved that purified recombinant bHPP (rbHPP-His)also forms a complex of ~100 kDa (Fig. 3A). rbHPP-Hiswas purified from bacterial lysate, dialysed, and analysedby blue native PAGE under native and denaturingconditions. Immunoblot analysis showed that denaturedrbHPP-His ran as expected for a 47 kDa protein.

Fig. 1. Subcellular localization of bHPP and GRLP transformants.Transformed T. vaginalis cells expressing bHPP-HA or GRLP-��were probed with mouse anti-HA and goat anti-mouse antibodies(green). The hydrogenosomal marker protein Hsp70 was localizedusing a rabbit anti-T. vaginalis Hsp70 and goat anti-rabbitantibodies (red). Merged images show colocalization of bHPP-HAand GRLP-HA with Hsp70. Nuclei were stained with DAPI.

Fig. 2. Fractionation of bHPP and GRLP.A. Solubilized hydrogenosomal lysate (HS) fractionated on a2–20% sucrose gradient, blotted and probed with anti-GRLP(top panel) and anti-bHPP antibodies (bottom panel).B. Elution profile of bHPP and GRLP from a Superdex 200 columnloaded with 4 mg of wt hydrogenosomal lysate. Proteins weredetected using anti-GRLP (�) and anti-bHPP antibodies (�), andthe resulting immunoblots were scanned by ImageJ software todetermine intensity values. Arrows indicate the migration ofmolecular weight markers.

1156 M. T. Brown et al.


However, rbHPP-His migrated as an ~100 kDa bandunder native conditions. The molecular mass of the nativerbHPP-His corresponds to that observed for the endog-enous bHPP complex (Fig. 2) and reveals that the purifiedprotein forms a homodimer in vitro.

Given that the observation of a b subunit homodimer isnovel among MPPs, we further addressed this issue usingaffinity chromatography. Hydrogenosomal lysate fromT. vaginalis expressing bHPP-HA was applied to anrbHPP-His affinity column and collected fractions weresubjected to immunoblot analysis using an anti-HA anti-body (Fig. 3B). These data show that bHPP-HA binds tothe rbHPP-His affinity column (left panel). In contrast, nobHPP-HA was found to bind to a negative control columncontaining the nuclear protein IBP39 (Schumacher et al.,2003) (right panel). Thus, the binding of bHPP-HA to therbHPP-His affinity column is specific.

We next tested whether endogenous T. vaginalis bHPPbinds to affinity columns of either rbHPP-His or rGRLP-His. A hydrogenosomal lysate derived from non-transfected (wt) cells was applied to both columns andcollected fractions were analysed by immunoblotting.bHPP was found only in the flow-through fraction col-lected from the rGRLP-His column, indicating that it didnot bind (Fig. 3C, lower left panel). In contrast, analysesof fractions collected from the rbHPP-His column showedbHPP was present in the elution fraction, confirming theability of the endogenous protein to form homodimers(Fig. 3C, upper left panel). To ensure that the protein inthe elution fraction was not rbHPP-His leaching fromthe column, the samples were rerun and the blot probedwith an anti-His antibody; no signal was detected in thisanalysis (Fig. 3C, upper right panel). Together, thesedata provide strong evidence for bHPP existing as a

homodimer, in contrast to the a-b heterodimer formed byMPPs (Taylor et al., 2001).

Recombinant bHPP-His forms an active peptidase

We have tested whether the homodimeric bHPP is anactive enzyme using a fluorometric assay designed tomeasure peptidase activity. The fluorogenic substrate,Abz-pAK, contains the cleavable N-terminal presequencefrom precursor to adenylate kinase (pAK) (Lange et al.,1994; Bradley et al., 1997) and fluoresces when cleaved.To establish the ability of the assay to detect presequencecleavage, Abz-pAK was incubated with wt hydrogenoso-mal lysate supplemented with 2 mM MnCl2 to stimulatemetallopeptidase activity, and the increase in fluores-cence intensity was measured over time. To ensure thatthe activity was due to a metallopeptidase, EDTA wasadded to the reaction and the activity was found to beeffectively inhibited (Fig. 4A).

rbHPP-His was then incubated with Abz-pAK in thepresence or absence of MnCl2, to definitively test whetherthe purified homodimer is an active metallopeptidase. Weobserved that rbHPP-His was capable of cleaving thesubstrate in a MnCl2-dependent reaction (Fig. 4B) andwas inhibited by EDTA. To eliminate the unlikely possibilitythat activity required a contaminating complementarysubunit or cofactor from E. coli, not detectable by Coo-massie staining of the purified protein (Fig. 3A), rbHPP-His was purified under denaturing conditions andrenatured during dialysis. The renatured peptidase wasalso found to cleave Abz-pAK in the presence of MnCl2(Fig. 4B).

Because a functional b subunit homodimer is unprec-edented, we next tested whether addition of GRLP

Fig. 3. bHPP forms a homodimer.A. Coomassie stained gel of purified rbHPP-His from transformed bacteria (Bac). rbHPP-HIS was analysed under native and denaturingconditions by BN-PAGE (right panel). Arrows indicate molecular weight markers.B. Hydrogenosomal lysate (HS) from T. vaginalis expressing bHPP-HA was applied to affigel columns containing rbHPP-His or thenon-hydrogenosomal protein rIBP39-HIS, flow-through (FT), wash (W) and elution (E) fractions were taken, and the resulting immunoblotswere probed with anti-HA antibody.C. Wild-type hydrogenosomal lysate was applied to affigel columns containing either rbHPP-His (top) or rGRLP-HIS (bottom) and immunoblotsof HS, flow-through (FT), wash (W) and elution (E) fractions were probed with anti-bHPP antibody or anti-His, as indicated. rbHPP-His (b)provided a control for the reactivity of the anti-His antibody.



enhanced bHPP activity in vitro. We first established thatsubstrate hydrolysis by rbHPP-His is concentration-dependent (Fig. 4C), then monitored presequence cleav-age at a non-saturating concentration of bHPP andvarying amounts of added GRLP. The addition of rGRLP-His had no significant effect on the reaction (Fig. 4C).These data show that rbHPP-His alone forms an activepeptidase, and that addition of rGRLP-His does not stimu-late its activity. The ability of purified bHPP alone to cleavea hydrogenosomal presequence is consistent with ourcompositional analyses (Figs 2 and 3) that show bHPP isnot found in the same complex as GRLP but insteadexists as a homodimer.

Phylogenetic analysis of bHPP

We conducted phylogenetic analyses to determine therelationship of T. vaginalis bHPP to eukaryotic MPPs,related cytochrome bc1 core proteins (Glaser and Dessi,1999) and bacterial M16 metallopeptidases (see Fig. S3for alignments) using Bayesian analyses. Our data indi-cate that bHPP shares a common ancestry with mitochon-drial peptidases, specifically the b subunit (Fig. 5). Threeindependent analyses strongly supported reciprocalmonophyly of eukaryotic and bacterial peptidases (poste-rior probability scores � 0.99). As expected, within theeukaryotic peptidases, mitochondrial a and b subunitsformed reciprocally monophyletic clades with absoluteposterior probability support. These data support thehypothesis (Pollock et al., 1988; Alper et al., 2006) thatthe a and bMPP genes arose from a gene duplication thatoccurred early in eukaryotic history.

We found that the T. vaginalis bHPP gene is positionedwithin the mitochondrial b subunit clade with strong pos-terior probability support (� 0.95). Furthermore, these

data soundly reject the hypothesis that the bHPP genediverged prior to the gene duplication that produced themitochondrial a and b subunits (Bayes factor = 34.07). Totest the possibility that long branch attraction (LBA) influ-ences the phylogenetic position of bHPP, bHPP andneighbouring taxa were removed individually and the treewas reconstructed. An identical tree topology wasobserved regardless of these omissions, demonstratingthat the positioning of bHPP was not an LBA artefact (datanot shown). Attempts to include GRLP in the data set usedto generate the tree shown in Fig. 5 were unsuccessfuldue to its lack of sequence similarity. Thus, while theevolution of bHPP has paralleled that of mitochondrial bsubunit genes, there is no evidence of a T. vaginalis genecorresponding to aMPP.

Discussion

We have characterized a metallopeptidase that is capableof cleaving the targeting presequence of a T. vaginalishydrogenosomal matrix protein. This enzyme, namedbHPP, is a novel protein related to the b subunit of theMPPs. bHPP was found to contain the pitrilysin active sitemotif HXXEHX76E and to localize to the hydrogenosome.In contrast to MPP, which is a heterodimeric proteinrequiring both an a and b subunit for activity (Lucianoet al., 1997), bHPP forms an active homodimer in vitro.Consistent with these observations, genome databasesearches for a T. vaginalis homologue with significantsimilarity to an aMPP subunit were unproductive. Similar-ity between aMPP and the only candidate gene identified,GRLP, is limited to a short, glycine-rich loop (GRL)domain. To eliminate the possibility that GRLP is a highlydivergent homologue of aMPP, we tested whether GRLPphysically interacts with bHPP in vivo and in vitro or is

Fig. 4. bHPP forms an active peptidase (A) Wild-type hydrogenosomal lysate was incubated with the presequence-containing fluorogenicsubstrate Abz-pAK in the presence of MnCl2 or MnCl2 + EDTA. Mean � SD (n = 3).B. Processing of Abz-pAK (expressed in pmole min-1) by rbHPP-His (50 nM) in the presence or absence of MnCl2, � EDTA. rbHPP-Hispurified under denaturing conditions, renatured and assayed in presence of MnCl2 is also shown.C. Abz-pAK incubated in the presence of MnCl2 and rbHPP-His � increasing concentrations of GRLP-His. Mean � SD (n = 3).



capable of stimulating bHPP peptidase activity in vitro.GRLP does not appear to associate with bHPP, nor did itstimulate bHPP activity. Together, the data presented hereprovide strong support that the T. vaginalis HPP is com-posed of a homodimeric bHPP that has common ancestrywith mitochondrial bMPP.

While this work was in progress, the first bacterial MPP-like protein was characterized from the a-proteobacteriumR. prowazekii (Kitada et al., 2007). Interestingly thisprotein, RPP, is composed of a single subunit containingthe active site motif HXXEHX76E found in bMPP and iscapable of cleaving short mitochondrial presequencesand basic peptides (Kitada et al., 2007), similar to thatreported here for HPP.

Phylogenetic analyses show that bHPP clusters witheukaryotic bMPPs, consistent with mitochondria andhydrogenosomes originating from a common endosym-biont (Martin and Embley, 2004; Embley and Martin,2006), contrary to that indicated by our previous analyses

of hydrogenosomal oxidoreductase proteins (Dyall et al.,2004b). Additionally, the phylogeny of HPP and MPPsshows that a and bMPPs form reciprocally monophyleticclades demonstrating that they arose from a gene dupli-cation event that occurred before the divergence ofT. vaginalis and mitochondria-bearing lineages (Pollocket al., 1988; Alper et al., 2006). Analyses conducted onRPP (Kitada et al., 2007) and those reported here alsoindicate that processing peptidases in mitochondrial-likeorganelles, including hydrogenosomes, originate from theancestral a-proteobacterial endosymbiont.

Examination of available bacterial genomes for a andbMPP homologues reveals genes containing the activesite motif HXXEHX76E found in bMPP as discussed abovefor Rickettsiae RPP (Kitada et al., 2007), and no geneswith the hallmark GRL of aMPP. Nevertheless, RPP wasshown to be capable of replacing yeast aMPP to form anactive peptidase together with yeast bMPP. The ability ofRPP to act as an aMPP, although it is more similar to

Fig. 5. Phylogenetic analyses of a and bMPPsubunits reveal that the hydrogenosomal HPPis a b subunit and that the duplication givingrise to a and bMPP subunits occurred beforethe divergence of mitochondrial andhydrogenosomal lineages. Consensus treetopology and posterior probabilities recoveredfrom three independent MC3 analyses areshown. Posterior probabilities for individualnodes varied by � 0.02 between analyses;the most conservative posterior probabilityestimate for each node is presented. Branchlengths are mean maximum likelihoodestimates calculated from 11241 trees withthe consensus topology.



bMPP, is consistent with ancestral organellar processingpeptidases functioning as homodimers, as described herefor T. vaginalis HPP. Should this be the case, the evolutionof an exclusive a-b heterodimer would be an adaptationspecific to later (mitochondrial) lineages, whereas ahomodimeric peptidase (bHPP) continued to do the job intrichomonad hydrogenosomes. Alternatively, the retentionof more ancestral qualities by the b subunit may haveallowed T. vaginalis to revert to the homodimeric state.

The destiny in T. vaginalis of the duplicated gene thatbecame aMPP in other eukaryotes is unclear. It remainspossible that the GRLP identified in this study is a relic ofthe duplication event. Alternatively, the duplicated genemay exist in an unsequenced portion of the genome, ormay have been completely lost in the trichomonadlineage.

The observation that recombinant bHPP forms anactive homodimer, whereas all characterized MPPsrequire both an a and b subunit for activity also providesclues into mechanistic variation for processing peptidasesbased on the length of their presequences substrates.The presequences that target proteins to the matrix ofhydrogenosomes, mitosomes and mitochondria sharesimilarities, however, hydrogenosomal and mitosomalpresequences are typically shorter (< 15 residues)than their mitochondrial counterparts (20–60 residues)(Bradley et al., 1997; Gakh et al., 2002). The b subunit ofMPP has been shown to interact directly with short mito-chondrial presequences, while the a subunit providesdistal binding sites only necessary for longer presequencesubstrates (> 20 residues) (Taylor et al., 2001; Kitadaet al., 2003). The glycine-rich loop (GRL) of aMPPappears to function in retaining longer mitochondrial pre-sequences until cleavage occurs (Nagao et al., 2000).Deletion of the GRL of yeast aMPP was recently shown toabolish the cleavage of a long mitochondrial presequence(58 residues) whereas cleavage of a 16 residue prese-quence was reduced by only 3-fold (Kitada et al., 2007).This is consistent with the data presented here indicatingthat short (< 15 residues) hydrogenosomal presequencesdo not require an aHPP subunit for cleavage.

Comparative analyses of protein translocation intohydrogenosomes, mitosomes and mitochondria highlightboth conserved and divergent traits (Dyall et al., 2004a;Wiedemann et al., 2004; Dolezal et al., 2006; Burri andKeeling, 2007). With regard to presequence processingpeptidases, a varied picture is emerging. Examination ofthe complete genome of the diplomonad Giardia, a modelorganism for studying mitosomes (Tovar et al., 1999;Tovar et al., 2003) reveals only one MPP-like gene con-taining the active site conserved in bHPP and bMPPs(Dolezal et al., 2005; Regoes et al., 2005) mirroring thatfound in the Trichomonas genome and supporting ourhypothesis that organelles that use short targeting prese-

quences do not require a heterodimeric peptidase. More-over, no MPP-like gene was found in the completegenome of the microspordian Encephalitozoon cuniculi,described to contain a mitochondrial remnant (Williamset al., 2002; Burri et al., 2006), leading Burri and Keeling(2007) to propose that E. cuniculi mitosomal proteins thatcontain short presequences may not need processing tofunction in the organelle.

Processing peptidases and protein translocases likelyplayed a vital role in the conversion of an ancestrala-proteobacterial endosymbiont into an intracellularorganelle. Hydrogenosomes are present in phylogeneti-cally diverse lineages and appear to have evolved inde-pendently, multiple times (Muller, 1993; Boxma et al.,2005; Hackstein et al., 2006). As the functional differencebetween HPP and MPP described here illustrates, tri-chomonad hydrogenosomes lack many propertiesconserved between ciliate hydrogenosomes and mito-chondria (Akhmanova et al., 1998; Tielens et al., 2002;Boxma et al., 2005). Whereas hydrogenosomes in theciliate Nyctotherus ovalis are converted mitochondriaadapted to function in anaerobic conditions (Henze andMartin, 2003), mitosomes and Trichomonas hydrogenos-omes (Tovar et al., 1999; Tovar et al., 2003; Dolezal et al.,2005), are less similar to mitochondria, and having under-gone reductive evolution, are less complex. Proteomicanalyses of different hydrogenosomes and mitosomes willlikely reveal additional components with common ances-try with mitochondria, as well as ones with independentorigins (Andersson, 2005; Likic et al., 2005; Mukherjeeet al., 2006).

Experimental procedures

Cloning T. vaginalis bHPP and the GRLP genes

Preliminary sequence data for the T. vaginalis strain G3genome database (Carlton et al., 2007) was obtained fromThe Institute for Genomic Research (http://www.tigr.org). The1260 bp ORF encoding bHPP, corresponding to the predictedgene product 88141.m00154 (TVAG_233350), and the1257 bp GRLP ORF, corresponding to the predicted geneproduct 84270.m00683 (TVAG_119710), were amplified byPCR from G3 T. vaginalis DNA and inserted into the pET29bvector (Novagen) for E. coli expression with a C-terminal Histag. These ORFs were also used to replace the hmp31cassette of the pHmp31-(HA)2 plasmid (Dyall et al., 2000) togenerate T. vaginalis expression constructs with twoC-terminal haemagglutinin (HA) tags.

Expression and purification of rbHPP-His andrGRLP-His in E. coli

BL21(DE3) cells containing the pET29b-bHPP-His plasmidor pET29b-GRLP-His plasmid grown in the presence of50 mg ml-1 kanamycin at 37°C were induced to express the



http://www.tigr.org

protein using 1 mM IPTG. Cells lysed in 20 mM Tris-HCl(pH 8.0), 200 mM NaCl, 0.2% Triton X-100, 200 mg lysozyme,and 1¥ protease inhibitor cocktail EDTA-free (Roche) werecentrifuged at 16 000g for 15 min at 4°C and the supernatantwas added to a Ni-NTA column (Qiagen). RecombinantbHPP-His (rbHPP-His) was eluted under native conditionsusing standard buffers containing 10 mg ml-1 leupeptin,50 mg ml-1 TLCK and dialysed in 20 mM Tris-HCl (pH 8.0),200 mM NaCl, 1 mM octylglucoside, 10 mg ml-1 leupeptin and50 mg ml-1 TLCK. Rabbit polyclonal antibodies were raisedagainst the purified protein (Animal Pharm).

Immunofluorescent microscopy

Trichomonas vaginalis T1 transfectants were methanol-fixed,preincubated in blocking buffer (PBS/0.25% BSA) priorto incubation with mouse anti-HA monoclonal antibody(Covance) and/or rabbit anti-Hsp70 polyclonal antibodydiluted 1:5000 (v/v), in blocking buffer. Secondary AlexaFluor-488 (green) anti-mouse and Alexa Fluor-594 (red) anti-rabbit antibodies (Molecular Probes), diluted 1:5000 (v/v)were added for 1 h. Cells were mounted in Prolong anti-fademounting medium (Molecular Probes). Images were taken ona Zeiss Axioscope 2 microscope and analysed using Axiovi-sion LE software.

Isolation and subfractionation of hydrogenosomes

Trichomonas vaginalis strain T1 hydrogenosomes were iso-lated as described (Bradley et al., 1997). Hydrogenosomeswere lysed in 20 mM Tris-HCl (pH 8.0), 200 mM NaCl and0.1% Triton X-100 by sonication (lysis buffer). Hydrogenoso-mal lysate was centrifuged at 16 000g and the soluble lysate(HS) was collected, followed by resuspension of the insolublepellet (HP) in lysis buffer.

Fractionation of hydrogenosomal soluble fraction bysucrose gradient

Hydrogenosomes, equivalent to 2 mg of protein, were lysedin 20 mM Tris-HCl, 200 mM NaCl, 0.1% Triton X-100, 2%sucrose, 10 mg ml-1 leupeptin and 50 mg ml-1 TLCK. HS wasoverlaid on a step gradient of 4–20% sucrose, 20 mM Tris-HCl (pH 8.0), 200 mM NaCl, 0.1% Triton X-100, 10 mg ml-1

leupeptin and 50 mg ml-1 TLCK. Centrifugation was per-formed in an SW41 rotor at 38 000 r.p.m. for 16 h at 4°C.HMW Native Marker Kit (GE Healthcare) were dissolved inhydrogenosomal lysis buffer and fractionated in a parallelgradient. Blots were generated and probed with either rabbitanti-bHPP antibody diluted 1:5000 or mouse anti-GRLPserum diluted 1:10 followed by anti-rabbit HRP diluted1:50 000 (Jackson Laboratories) or anti-mouse HRP diluted1:25 000 respectively. Bands were visualized using ECL.

Fractionation of hydrogenosomal lysate by SEC

Hydrogenosomes were solubilized as described above in20 mM Tris-HCl, 200 mM NaCl, 0.1% Triton X-100,10 mg ml-1 leupeptin and 50 mg ml-1 TLCK. HS was loaded

onto the SEC column at 0.1 ml min-1, the column was washedfor 16 h with solubilization buffer at 0.1 ml min-1 and 2 mlfractions were collected. A 250 ml aliquot from each fractionwas TCA precipitated and the samples were run on an 8–16%SDS-PAGE gel. Blots were generated and probed with eitherrabbit anti-bHPP antibody diluted 1:5000 or mouse anti-GRLP serum diluted 1:10 followed by anti-rabbit HRP diluted1:50 000 or anti-mouse HRP diluted 1:25 000 respectively.Bands were visualized using ECL. The immunoblot wasanalysed using ImageJ software (National Institute of Health,USA).

Affinity matrix pull-down of bHPP-HA

rbHPP-His, rGRLP-His or the non-hydrogenosomal proteinrIBP39-His were conjugated to 1 ml of Affigel-10 beads(Bio-Rad) as per instructions, washed to remove unconju-gated proteins and equilibrated in lysis buffer. HS fromT. vaginalis hydrogenosomes was loaded on the beads,washed with 10 vols lysis buffer, and bound proteins wereeluted with 20 mM Tris-HCl, 200 mM NaCl, 8 M urea,10 mg ml-1 leupeptin and 50 mg ml-1 TLCK and separated ona 12% SDS-PAGE gel. Blots were generated and analysedusing the primary anti-bHPP antibody diluted 1:5000 andsecondary anti-rabbit HRP or anti-His HRP (Novagen) anti-bodies diluted 1:25 000 and visualized using ECL.

Blue native gel electrophoresis of the rbHPP-His

rbHPP-His was solubilized at a protein concentration of200 mg ml-1 in 20 mM Tris-HCl (pH 8.0), 200 mM NaCl,10 mg ml-1 leupeptin, 50 mg ml-1 TLCK, 1% Triton X-100 and10% glycerol. The solubilized proteins were analysed by bluenative electrophoresis on an 8–16% linear polyacrylamidegradient. Control and denatured samples were generated byheating at 95°C for 5 min in the same buffer with 0.5% SDS.

Peptidase activity assays

Fluorescent cleavage assays were conducted usingsubstrate 2-aminobenzoyl-MLSTLAKRFAY(3-NO2)GKKDRM(Abz-pAk) (Synpep, CA) solubilized in dimethyl formamideand assayed in 10 mM Tris-HCl (pH 8.0), 150 mM NaCl,2 mM MnCl2 at 37°C with either hydrogenosomal lysate (HS),solubilized in 20 mM Tris-HCl (pH 8.0), 200 mM NaCl and 1%Triton X-100, or purified rbHPP-His. Fluorescence was mea-sured in a PTI fluorometer (emission wavelength = 320 nmand excitation wavelength = 410 nm).

Phylogenetic analyses

Homologues of T. vaginalis bHPP were identified by BLAST

searching NCBI’s non-redundant protein database and othergenome databases. An alignment was produced usingMUSCLE (Edgar, 2004), with manual curation (Fig. S2). Phy-logenetic analysis of the final alignment of 323 amino acidpositions after masking was performed using Mr Bayesversion 3.1.1 (Ronquist and Huelsenbeck, 2003), the WAGamino acid substitution matrix, allowance for invariant sites,



and a gamma distribution model of among-site rate variation.In three independent analyses, four differentially heatedMetropolis-coupled Markov chains were run to 10 milliongenerations with sampling every 100 generations. Afterremoval of burn-in, Markov chain convergence was con-firmed using AWTY (http://king2.scs.fsu.edu/CEBProjects/awty/awty_start.php). Statistical support for alternative treetopologies was assessed using the Bayes factor test (Kassand Raftery, 1995).

Acknowledgements

This work was supported by a National Institutes of Health(NIH) grant and a Burroughs-Wellcome Molecular Parasitol-ogy Award to P.J.J., and an NIH Microbial PathogenesisTraining Grant (2-T32-AI-007323) to M.B.. A.M. and H.G.were supported by the Marine Biological Laboratory’sProgram in Global Infectious Disease, funded by the EllisonMedical Foundation. Computational resources were providedby the Josephine Bay Paul Center for Comparative MolecularBiology and Evolution (Marine Biological Laboratory) throughfunds provided by the W.M. Keck Foundation and the G.Unger Vetlesen Foundation. Sequencing of the T. vaginalisgenome was accomplished with support from the NationalInstitute of Allergy and Infectious Diseases (NIAID).

References

Akhmanova, A., Voncken, F., van Alen, T., van Hoek, A.,Boxma, B., Vogels, G., et al. (1998) A hydrogenosome witha genome. Nature 396: 527–528.

Alper, B.J., Nienow, T.E., and Schmidt, W.K. (2006) Acommon genetic system for functional studies of pitrilysinand related M16A proteases. Biochem J 398: 145–152.

Andersson, J.O. (2005) Lateral gene transfer in eukaryotes.Cell Mol Life Sci 62: 1182–1197.

Andersson, S.G., Zomorodipour, A., Andersson, J.O.,Sicheritz-Ponten, T., Alsmark, U.C., Podowski, R.M., et al.(1998) The genome sequence of Rickettsia prowazekii andthe origin of mitochondria. Nature 396: 133–140.

Boxma, B., de Graaf, R.M., van der Staay, G.W., van Alen,T.A., Ricard, G., Gabaldon, T., et al. (2005) An anaerobicmitochondrion that produces hydrogen. Nature 434:74–79.

Bradley, P.J., Lahti, C.J., Plumper, E., and Johnson, P.J.(1997) Targeting and translocation of proteins into thehydrogenosome of the protist Trichomonas: similaritieswith mitochondrial protein import. EMBO J 16: 3484–3493.

Burri, L., and Keeling, P.J. (2007) Protein targeting in para-sites with cryptic mitochondria. International J Parasitol 37:265–272. doi:10.1016/j.ijpara.2006.1012.1004.

Burri, L., Williams, B.A., Bursac, D., Lithgow, T., and Keeling,P.J. (2006) Microsporidian mitosomes retain elements ofthe general mitochondrial targeting system. Proc Natl AcadSci USA 103: 15916–15920.

Carlton, J.M., Hirt, R.P., Silva, J.C., Delcher, A.L., Schatz, M.,Zhao, Q., et al. (2007) Draft genome sequence of the sexu-ally transmitted pathogen Trichomonas vaginalis. Science315: 207–212.

Dolezal, P., Smid, O., Rada, P., Zubacova, Z., Bursac, D.,Sutak, R., et al. (2005) Giardia mitosomes and tri-chomonad hydrogenosomes share a common mode ofprotein targeting. Proc Natl Acad Sci USA 102: 10924–10929.

Dolezal, P., Likic, V., Tachezy, J., and Lithgow, T. (2006)Evolution of the molecular machines for protein import intomitochondria. Science 313: 314–318.

Dyall, S.D., Koehler, C.M., Delgadillo-Correa, M.G., Bradley,P.J., Plumper, E., Leuenberger, D., et al. (2000) Presenceof a member of the mitochondrial carrier family in hydro-genosomes: conservation of membrane-targeting path-ways between hydrogenosomes and mitochondria. MolCell Biol 20: 2488–2497.

Dyall, S.D., Brown, M.T., and Johnson, P.J. (2004a) Ancientinvasions: from endosymbionts to organelles. Science 304:253–257.

Dyall, S.D., Yan, W., Delgadillo-Correa, M.G., Lunceford, A.,Loo, J.A., Clarke, C.F., and Johnson, P.J. (2004b) Non-mitochondrial complex I proteins in a hydrogenosomal oxi-doreductase complex. Nature 431: 1103–1107.

Edgar, R.C. (2004) MUSCLE: a multiple sequence alignmentmethod with reduced time and space complexity. BMCBioinformatics 5: 113.

Embley, T.M. (2006) Multiple secondary origins of theanaerobic lifestyle in eukaryotes. Philos Trans R Soc LondB Biol Sci 361: 1055–1067.

Embley, T.M., and Martin, W. (2006) Eukaryotic evolution,changes and challenges. Nature 440: 623–630.

Finn, R.D., Mistry, J., Schuster-Bockler, B., Griffiths-Jones,S., Hollich, V., Lassmann, T., et al. (2006) Pfam: clans,web tools and services. Nucleic Acids Res 34: D247–D251.

Gakh, O., Cavadini, P., and Isaya, G. (2002) Mitochondrialprocessing peptidases. Biochim Biophys Acta 1592:63–77.

Glaser, E., and Dessi, P. (1999) Integration of themitochondrial-processing peptidase into the cytochromebc1 complex in plants. J Bioenerg Biomembr 31: 259–274.

Gray, M.W. (1998) Rickettsia, typhus and the mitochondrialconnection. Nature 396: 109–110.

Gray, M.W. (2005) Evolutionary biology: the hydrogenos-ome’s murky past. Nature 434: 29–31.

Gray, M.W., Burger, G., and Lang, B.F. (1999) Mitochondrialevolution. Science 283: 1476–1481.

Hackstein, J.H., Tjaden, J., and Huynen, M. (2006) Mitochon-dria, hydrogenosomes and mitosomes: products of evolu-tionary tinkering!. Curr Genet 50: 225–245.

Henze, K., and Martin, W. (2003) Evolutionary biology:essence of mitochondria. Nature 426: 127–128.

Hrdy, I., Hirt, R.P., Dolezal, P., Bardonova, L., Foster, P.G.,Tachezy, J., and Embley, T.M. (2004) Trichomonas hydro-genosomes contain the NADH dehydrogenase module ofmitochondrial complex I. Nature 432: 618–622.

Kass, R.E., and Raftery, A.E. (1995) Bayes Factors. J AmStatist Assoc 90: 733–795.

Kitada, S., Yamasaki, E., Kojima, K., and Ito, A. (2003) Deter-mination of the cleavage site of the presequence by mito-chondrial processing peptidase on the substrate bindingscaffold and the multiple subsites inside a molecular cavity.J Biol Chem 278: 1879–1885.



http://king2.scs.fsu.edu/CEBProjects/awty/awty_start.php

http://king2.scs.fsu.edu/CEBProjects/awty/awty_start.php

Kitada, S., Uchiyama, T., Funatsu, T., Kitada, Y., Ogishima,T., and Ito, A. (2007) A protein from a parasitic microor-ganism, rickettsia prowazekii, can cleave the signalsequences of proteins targeting mitochondria. J Bacteriol189: 844–850.

Lange, S., Rozario, C., and Muller, M. (1994) Primary struc-ture of the hydrogenosomal adenylate kinase of Trichomo-nas vaginalis and its phylogenetic relationships. MolBiochem Parasitol 66: 297–308.

Likic, V.A., Perry, A., Hulett, J., Derby, M., Traven, A., Waller,R.F., et al. (2005) Patterns that define the four domainsconserved in known and novel isoforms of the proteinimport receptor Tom20. J Mol Biol 347: 81–93.

Luciano, P., Geoffroy, S., Brandt, A., Hernandez, J.F., andGeli, V. (1997) Functional cooperation of the mitochondrialprocessing peptidase subunits. J Mol Biol 272: 213–225.

Luciano, P., Tokatlidis, K., Chambre, I., Germanique, J.C.,and Geli, V. (1998) The mitochondrial processing pepti-dase behaves as a zinc-metallopeptidase. J Mol Biol 280:193–199.

Martin, W. (2005) The missing link between hydrogenosomesand mitochondria. Trends Microbiol 13: 457–459.

Martin, W., and Embley, T.M. (2004) Evolutionary biology:early evolution comes full circle. Nature 431: 134–137.

Mukherjee, M., Brown, M.T., McArthur, A.G., and Johnson,P.J. (2006) Proteins of the glycine decarboxylase complexin the hydrogenosome of Trichomonas vaginalis. EukaryotCell 5: 2062–2071.

Muller, M. (1993) The hydrogenosome. J Gen Microbiol 139:2879–2889.

Nagao, Y., Kitada, S., Kojima, K., Toh, H., Kuhara, S.,Ogishima, T., and Ito, A. (2000) Glycine-rich region ofmitochondrial processing peptidase alpha-subunit isessential for binding and cleavage of the precursorproteins. J Biol Chem 275: 34552–34556.

Ou, W.J., Ito, A., Okazaki, H., and Omura, T. (1989) Purifi-cation and characterization of a processing protease fromrat liver mitochondria. EMBO J 8: 2605–2612.

Pfanner, N., Wiedemann, N., Meisinger, C., and Lithgow, T.(2004) Assembling the mitochondrial outer membrane.Nat Struct Mol Biol 11: 1044–1048.

Pollock, R.A., Hartl, F.U., Cheng, M.Y., Ostermann, J.,Horwich, A., and Neupert, W. (1988) The processing pep-tidase of yeast mitochondria: the two co-operating compo-nents MPP and PEP are structurally related. EMBO J 7:3493–3500.

Regoes, A., Zourmpanou, D., Leon-Avila, G., van der Giezen,M., Tovar, J., and Hehl, A.B. (2005) Protein import, repli-cation, and inheritance of a vestigial mitochondrion. J BiolChem 280: 30557–30563.

Ronquist, F., and Huelsenbeck, J.P. (2003) MrBayes 3:Bayesian phylogenetic inference under mixed models.Bioinformatics 19: 1572–1574.

Schumacher, M.A., Lau, A.O., and Johnson, P.J. (2003)Structural basis of core promoter recognition in a primitiveeukaryote. Cell 115: 413–424.

Sutak, R., Dolezal, P., Fiumera, H.L., Hrdy, I., Dancis, A.,Delgadillo-Correa, M., et al. (2004) Mitochondrial-typeassembly of FeS centers in the hydrogenosomes of theamitochondriate eukaryote Trichomonas vaginalis. ProcNatl Acad Sci USA 101: 10368–10373.

Taylor, A.B., Smith, B.S., Kitada, S., Kojima, K., Miyaura, H.,Otwinowski, Z., et al. (2001) Crystal structures of mito-chondrial processing peptidase reveal the mode for spe-cific cleavage of import signal sequences. Structure(Camb) 9: 615–625.

Tielens, A.G., Rotte, C., van Hellemond, J.J., and Martin, W.(2002) Mitochondria as we don’t know them. TrendsBiochem Sci 27: 564–572.

Tovar, J., Fischer, A., and Clark, C.G. (1999) The mitosome,a novel organelle related to mitochondria in the amitochon-drial parasite Entamoeba histolytica. Mol Microbiol 32:1013–1021.

Tovar, J., Leon-Avila, G., Sanchez, L.B., Sutak, R., Tachezy,J., van der Giezen, M., et al. (2003) Mitochondrial remnantorganelles of Giardia function in iron-sulphur proteinmaturation. Nature 426: 172–176.

Wiedemann, N., Frazier, A.E., and Pfanner, N. (2004) Theprotein import machinery of mitochondria. J Biol Chem279: 14473–14476.

Williams, B.A., Hirt, R.P., Lucocq, J.M., and Embley, T.M.(2002) A mitochondrial remnant in the microsporidian Tra-chipleistophora hominis. Nature 418: 865–869.

Supplementary material

The following supplementary material is available for thisarticle:Figure S1. Alignment of bHPP and bMPPs from S. cer-eviseae (P10507), C. elegans (NP_501576), M. musculus(Q9CXT8).Figure S2. Alignment of GRLP with putative aMPPs from C.hominus (EAL38279), P. chabaudi (CAH88238) and T. gondii(AAF00541).Figure S3. Final alignment used in Bayesian analysis.

This material is available as part of the online article from:http://www.blackwell-synergy.com/doi/abs/10.1111/j.1365-2958.2007.05719.x(This link will take you to the article abstract).

Please note: Blackwell Publishing is not responsible for thecontent or functionality of any supplementary materials sup-plied by the authors. Any queries (other than missing mate-rial) should be directed to the corresponding author for thearticle.



http://www.blackwell-synergy.com/doi/abs/10.1111/j.1365-2958.2007.05719.x

http://www.blackwell-synergy.com/doi/abs/10.1111/j.1365-2958.2007.05719.x

a functionally divergent hydrogenosomal peptidase with protomitochondrial ancestry

Documents