A proteomic analysis of arsenical drug resistance in Trypanosoma brucei

RESEARCH ARTICLE

A proteomic analysis of arsenical drug resistance

in Trypanosoma brucei

Aude L. Foucher1, Anne McIntosh1, Gill Douce1, Jonathan Wastling1,Andy Tait2 and C. Michael R. Turner1, 2

1 Division of Infection and Immunity, Institute of Biomedical and Life Sciences, University of Glasgow,Glasgow, UK

2 Wellcome Centre for Molecular Parasitology, Anderson College, University of Glasgow, Glasgow, UK

We have undertaken 2-DE and MS to identify proteins associated with arsenical drug resistancein Trypanosoma brucei. This parasite causes sleeping sickness in humans, and arsenical drugresistance is a significant potential problem. Comparative analysis of approximately 2000 spotsresolved by 2-DE in the soluble proteomes of drug-sensitive and drug-resistant isogenic lines ofT. brucei identified a protein spot whose absence associated with resistance to the arsenical drug,Cymelarsan. MS matched this protein to an identical pair of tandem genes Tb09.211.0120 and0130 that encode a putative nascent polypeptide associated complex subunit. This protein alsooccurs as an isoform located in both resistant and sensitive lines at a similar molecular weight,but different pI. The difference between isogenic lines was confirmed by Western blot using anantibody against recombinant protein. Both genes were identical in sequence between drug-sensitive and drug-resistant lines and both were transcribed as determined by RT-PCR. Wepostulate that the missing protein isoform arose due to the lack of a PTM.

Received: June 6, 2005Revised: November 28, 2005

Accepted: November 29, 2005

Keywords:

Cymelarsan / Drug resistance / Parasite / Proteome / Trypanosoma brucei

2726 Proteomics 2006, 6, 2726–2732

1 Introduction

Two subspecies of the protozoan parasite Trypanosoma brucei,T. b. rhodesiense and T. b. gambiense, are responsible for thedeath of 150 000 people every year in Africa from sleepingsickness [1]. This disease is fatal if left untreated and only ahandful of drugs are available for treatment. The most effec-tive drug, a melaminophenyl arsenical drug, melarsoprol, isactive against both subspecies of the parasite and was intro-duced in 1949. However, treatment failure with this drug is

increasing with some 30% of melarsoprol-treated sleepingsickness cases in Uganda [2] and 25% in Angola [3] currentlyshowing treatment failure which is thought to arise fromdrug resistance [4]. It is becoming increasingly important todefine the genes/proteins associated with arsenical resis-tance in order to understand the mechanisms of resistanceand develop simple tests for resistant parasites.

The best studied potential mechanism of arsenical resis-tance involves mutations in the P2 adenosine transporter [5,6], which is responsible for the transport of the melamino-phenyl arsenical drugs into the trypanosome cell [7]. How-ever, deletion of the TbAT1 gene encoding the P2 transporterconfers only a threefold resistance to melarsoprol [8] andthere is incomplete association of P2 mutations with drugresistance in field studies [4, 9]. Overexpression of a thiolconjugate transporter, TbMRPA, has also been associatedwith arsenical drug resistance and conferred a ten-foldincrease in resistance [10]. Whether TbAT1 and TbMRPA arethe only mechanisms of generating resistance available tothe parasite remains an open question.

Correspondence: Professor C. Michael R. Turner, Division ofInfection and Immunity, Institute of Biomedical and LifeSciences, Joseph Black Building, University of Glasgow, Glas-gow G12 8QQ, UKE-mail: [email protected]: 144-141-330-2041

Abbreviation: NAC, nascent polypeptide associated complexsubunit, putative

DOI 10.1002/pmic.200500419

2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com

Proteomics 2006, 6, 2726–2732 Microbiology 2727

The use of proteomics approaches to investigate drugresistance in T. brucei has not been reported previously but ithas provided interesting insights into methotrexate resistancein the related parasite, Leishmania major [11]. Such approachesare timely with the recent completion of the genome sequenceof T. brucei from the reference line TREU 927 [12]. Proteins ofinterest can thus be readily identified by MS [13] and wholeproteome comparisons made to analyse proteins associatedwith the acquisition of drug resistance. In this study we haveundertaken a proteomic analysis using comparative 2-DE oftwo isogenic lines of T. b. gambiense, Type 2: one sensitive tokilling by arsenical drugs and the other resistant, being derivedfrom the sensitive line by in vivo selection in increasing con-centration of cymelarsan [14]. Cymelarsan is a drug that isclosely related to melarsoprol, both structurally and function-ally [15]. We have previously demonstrated that the cyme-larsan-resistant line is crossresistant to melarsoprol and thatthe resistance phenotype is stable in the absence of drug pres-sure and after cyclical transmission through tsetse flies [14].The drug sensitivity/resistance phenotype is expressed both inbloodstream-form parasites that infect mammals and in pro-cyclic forms; this life cycle stage proliferates in the tsetse mid-gut and can be readily grown in vitro.

In this study a comparative analysis using 2-DE of solu-ble protein extracts from sensitive (386) and resis-tant (386Mr) isogenic lines of T. brucei was used to addressthe following questions. Are there any consistent alterationsin protein expression associated with arsenical resistance?Can differentially expressed protein spots and the corre-sponding genes be identified? Are alterations in arsenicalresistance associated with changes in the sequence andexpression of identified genes?

2 Materials and methods

2.1 Sample preparation

Two isogenic lines of the cloned stock STIB 386 that differ insensitivity to killing by arsenical drugs [14] were grown asprocyclic-form trypanosomes in SDM 79 (GibcoBRL) sup-plemented with 10% heat-inactivated foetal calf serum (Gib-coBRL) and 0.2 mg/mL Gentamycin (Sigma) at 257C.

Protein extracts were made from procyclic-form trypa-nosomes essentially as described previously [16]. Cultureswere grown to late log phase, and approximately 56108 try-panosomes were harvested by centrifugation at 12 0006g for8 min. Cells were washed twice with PBS with 1% glucose(PSG) pH 7.4, lysed in 470 mL of lysis/rehydration buffer(9 M urea, 2 M thiourea, 2% CHAPS, 65 mM DTT,0.5% IPG buffer, trace of bromophenol blue) with the addi-tion of 10 mL of nucleases (2000 U/mL DNAase 1750 U/mLRNAase A, 50 mM MgCl2) and 35 mL of proteinase inhibitors(complete mini protease inhibitor cocktail, Roche). Lysateswere then incubated for 1 h at room temperature with fre-quent vortexing and stored at 2207C.

2.2 2-DE and image analysis

Samples were thawed and centrifuged at 30006g for3 min to remove insoluble materials, and the super-natant was used for analysis. In the first dimension,samples were run on 24 cm IPG dry strips (pH 4.0–7.0,6.0–11.0, 4.0–5.0, 4.5–5.5 and 5.5–6.7; Amersham Bio-sciences) on an IPGphore (Amersham Biosciences) at207C and using 50 mA/strip. Rehydratation was per-formed at 30 V for 15 h, followed by isoelectric focus-sing at 500 V for 500 V/h, 1000 V for 1000 V/h, 8000 Vfor 3975 V/h by gradient and a final stage at 8000 V/hfor 6 h. Strips were equilibrated for 15 min in equilibra-tion buffer (50 mM Tris-HCl pH 8.8, 6 M urea, 30% gly-cerol, 2% SDS) containing 1% DTT and then for 15 minin equilibration buffer containing 2.5% a-iodoacetamide(Calbiochem) and sealed onto 12% acrylamide gels using0.5% agarose in electrophoresis buffer. Second-dimen-sion SDS-PAGE gels were run at 120 V (157C) in aDALT VI tank (Amersham Biosciences) using the elec-trophoresis buffer kit (Amersham Biosciences) until thetracking dye ran off the gel.

The gels were stained with either CBB G-250 (Sigma) orSYPRO Ruby stain (BioRad). Images from Coomassie-stained gels were acquired using Labscan sv3.0 software onan Umax flatbed scanner with integrated transparencyadapter. Images from SYPRO Ruby-stained gels wereacquired using Amersham Biosciences Typhon scan-ner 9400 (variable-mode imager) under the following set-tings: emission filter set at 560 LP green/purple, PMT = 550,laser = green 532, sensitivity normal, pixel size = 10 mm.Images were analysed using ImageMaster 2DElite software.Our method of sample preparation selectively extracts solu-ble proteins and our approach to protein separation and im-aging is likely to select for medium- and high-abundanceproteins (13). These factors need to be borne in mind in theinterpretation of the results.

2.3 MS

A protein spot that differed in abundance between drug-sen-sitive and drug-resistant lines, as well as a sample of spots inthe same region of the gel, were excised from the gel usingthe Amersham Ettan robotic station. In-gel trypsin digestionwas performed as previously described [17]. The resultingpeptides were desalted and concentrated using ZipTip pi-pette tips containing C18 RP media as described by the man-ufacturer (Millipore). One microlitre of the concentratedsample was loaded into a capillary needle (Protana) forinjection by ESI in the Tandem TOF-MS (Applied Biosys-tems). The resulting peptide MS/MS spectra were inter-preted using the MASCOT search engine (http://www.matrixscience.com/search_form_select.html) to search theNCBInr and a local T. brucei database downloaded fromGeneDB release 4 (http://www.genedb.org/).


2728 A. L. Foucher et al. Proteomics 2006, 6, 2726–2732

2.4 Molecular cloning

The ORFs of genes Tb09.211.0120 and 0130 were amplifiedfrom genomic DNA of both parasite lines by PCR usingORF-specific primers 274F (5’-CAGCGAATTTATCTA-CATGC) with 274R (5’-TGGCCACGGAAACACTCCTA) forORF 0120 and 277F (5’-GTGCTACTACATTCGCCTCA) with277F (5’-CATCACAATGGATGGAGGAT) for ORF 0130.PCRs were performed in 40 mL reaction volumes containing5 mM of each primer, 200 ng DNA template, 4 U Taq poly-merase and previously described buffer [18]. PCR conditionswere as follows: 30 cycles of 957C for 50 s, 637C for 50 s and687C for 2 min, with a final 15 min at 687C. Amplicons werecloned into the TOPO vector (Invitrogen) following themanufacturer instructions. Clone inserts were sequenced(MBSU, University of Glasgow) and the sequences reportedhere were defined by multiple analyses of cloned ampliconsfrom independent PCRs.

2.5 RT-PCR

Trypanosomes (approximately 26108) were centrifuged, thepelleted cells were washed once in PSG and then resus-pended in 1 mL of Trizol (Gibco BRL). Following 5 minincubation at room temperature, 0.2 mL of chloroform wasadded, and the tubes were shaken vigorously and then incu-bated for further 2 min at room temperature. The phaseswere separated by centrifuging for 15 min at 12 0006g (47C)and the aqueous phase was transferred into a clean Eppen-dorf to which 0.5 mL of isopropanol was added, mixed welland incubated at room temperature for 10 min. The RNAwas pelleted by centrifuging the tubes for 10 min at12 0006g (47C), and the pellet washed in 75% ethanol pre-pared in diethyl pyrocarbonate-treated water. The pellet wasair-dried, resuspended in 100 mL of treated water and treatedwith DNA-free kit (Ambion) according to the manufacturerinstructions.

Complementary DNA (cDNA) was prepared from RNAusing a kit (Promega) according to the manufacturerinstructions. cDNA was then used as template in PCR reac-tion using primers 274F/274R and 277F/277R. As controls,primers for the triose phosphate isomerase (TIM) gene wereused as previously described [19].

2.6 Antiserum production

An antiserum against ORF 0120 was generated in miceusing a DNA prime/protein-boost immunisation strategy.This strategy was adopted to maximise the specificity of theantibody generated.

ORF 0120 was amplified from genomic DNA by PCRand inserted into pcDNA 3.11 (Invitrogen) in frame withthe promoter, according to the manufacturer instructionsand Top 10 Escherichia coli competent cells (Invitrogen)were transformed with the plasmid. The recombinant

plasmid was purified from 100 mL of cell culture usingthe EndoFree plasmid Maxi kit (Invitrogen), according tothe manufacturer instructions and then coated onto 1 mmgold particles, which were used to prepare firing car-tridges for the Gene Gun (BioRad). Conditions for load-ing were those described in the manufacturer instructionguidelines (BioRad) and random testing of inoculationcartridges indicated a loading of approximately 1 mg ofDNA per shot.

A recombinant ORF 0120 protein was produced usingthe Gateway technology (Gibco BRL) to boost the productionof antibody in mice. A gel-purified PCR amplicon was in-serted into the entry clone, pDONR211, according to themanufacturer instructions. The insert was then transferredinto the pDEST 17 vector containing a T7 promoter and sixhistidines at the N-terminus. BL21AI-competent cells weretransformed with the recombinant pDEST17 vector andrecombinant protein production induced by addition ofL-arabinose. The recombinant protein was purified frombacterial cell lysate using a Ni-NTA column.

Female C57/b6 mice, aged 6–8 wk were obtained fromHarlan Olac (Bicester, UK). These animals were immunisedon day 1 and day 14 of the experiment and primed usingeither the control DNA alone (pcDNA3.11) or the DNAencoding the 0120 reading frame pcDNA3.1/0120. Approxi-mately 1 mg of DNA was delivered per animal by Gene Gunfrom a distance of 2.5 cm (400 psi). On day 28 and day 36,half of the animals immunised with pcDNA3.1/0120 wereboosted s.c. with 10 mg of Ni-purified protein. Serumresponses were followed using test bleeds taken on day 0, 21,36 and terminally on day 42.

2.7 Western blotting

Protein extracts (prepared as previously described) wereseparated on 7 cm, pH 4.0–7.0 2-D gels and transferred toHybond-C NC membrane by electroblotting in transferbuffer (20 mM Tris, 225 mM glycine) at 110 V for 1 h30 min at 47C using a mini transblot cell (BioRad). The NCmembrane was then placed in blocking solution consistingof 5% skimmed milk in TBS-Tween (0.02 M Tris, 0.137 MNaCl pH 7.6, 0.05% Tween 20) and incubated overnight at47C. The NC membrane was washed for 10 min in TBS-Tween at room temperature three times and then incu-bated for 1 h 30 min in antiserum diluted 1/10 000 in TBS-Tween. The membrane was washed three times for 10 minin TBS-Tween to remove any unbound antibody; horse-radish peroxidase-conjugated anti-mouse IgG (DiagnosticsScotland, 1/2000 dilution) in TBS-Tween was added andincubated for 1 h 30 min at room temperature. The blotwas washed as above, incubated with ECL Plus Chemilu-minescence substrate (Amersham Biosciences), andexposed to X-ray film. Images from the Western blots werescanned and matched to gel images using ImageMaster2Delite software.



3 Results

3.1 Comparative 2-DE

Protein extracts from drug-sensitive and drug-resistant lines,386 and 386Mr respectively, were separated by 2-DE andcompared. In the first dimension, various pH range gels(pH 4–7, 6–11, 4–5, 4.5–5.5 and 5.5–6.7) were used in orderto increase protein resolution. An illustration of the separa-tion of trypanosome soluble protein extracts is shown inFig. 1, in which approximately 750 spots are resolved.Approximate values for apparent molecular weight and pIwere determined by comparison of these figures with ourprevious analyses that identified most spots on gels ([13],http://pantar.dcs.gla.ac.uk/RAPAD/StudyAnnotator/). Theanalysis of all the various pH range gels indicated that ap-proximately 2000 protein spots could be resolved but thecomparative analysis identified only two potential differ-ences between resistant and sensitive lines, both detectableon pH 4–7 gels. To confirm the consistency of this result afurther nine replicates of these gels were run using materialindependently prepared from separate experiments. Analy-sis of these replicate experiments showed that only one pro-tein spot was reproducibly differentially expressed. This pro-tein spot was visible in the drug-sensitive line with a pI ofapproximately 4.5, but was consistently absent from prep-arations of the drug-resistant parasites (Fig. 1, arrow). Thedifferentially expressed spot was located adjacent to another

spot with the same molecular weight but with a more acidicpH that was present in both resistant and sensitive lines; thisprotein was subsequently shown to be an isoform of the dif-ferentially expressed protein. Protein differences could beclearly visualised with both SYPRO Ruby and CBB staining,indicating that the differentially expressed protein was rea-sonably abundant in drug-sensitive procyclic trypanosomes.

To investigate whether exposure to sublethal drug con-centration could cause changes in protein expression, wegrew 386Mr parasites in the presence of 1 mM cymelarsanand then analysed the proteome profile on pH 4–7 gels. Allspots were the same as for 386Mr grown in the absence ofdrug (data not shown).

3.2 Identification of the differentially expressed

protein

The differentially expressed protein spot and several adjacentspots were analysed (as shown in Fig. 2) by MS/MS and theresults are shown in Table 1. The results for spots 1, 3, 4 and5 in gels of both drug-sensitive and drug-resistant parasitesconfirm that the same regions of the gels were compared,based on spot identification as well as the pattern of spots.The identification of three of the spots as b-tubulin is con-sistent with previous data on the locations of isoforms of thisprotein in this region of the gel under the preparation andrunning conditions we use [13]. Spot 2 from the drug-sensi-tive line was matched to two putative, identical ORFs;

Figure 1. 2-DE of proteins extracted from isogenic cymelarsan-sensitive (A) and cymelarsan-resistant (B) lines separated on pH 4–7 rangegels and SYPRO Ruby stained. The differentially expressed protein spot is indicated by an arrow. No other differences were consistentlyobserved in all ten replicate experiments.



Table 1. MS/MS analysis of polypeptide spots shown in Fig. 2. Results in bold show spots associated witharsenical drug resistance. A MOWSE score .53 indicates identification with 95% confidence. Identifica-tions and gi numbers are those given in GeneDb; NAC,nascent polypeptide associated complex subunit,putative

Sample No. of peptidessequenced

%coverage

MOWSEscore

ID gi number

386/1 3 22 123 NAC Tb09.211.0120

386/2 4 37 178 NAC Tb09.211.0120

386/3 6 31 74 b tubulin Tb927.1.2330386/4 5 16 204 b tubulin Tb927.1.2330386/5 11 38 469 b tubulin Tb927.1.2330386Mr/1 5 40 217 NAC Tb09.211.0120

386Mr/2 Absent – – – –386Mr/3 5 25 143 b tubulin Tb927.1.2330386Mr/4 5 16 142 b tubulin Tb927.1.2330386Mr/5 9 18 437 b tubulin Tb927.1.2330

Figure 2. Analysis of the differentially expressed protein andnearby spots in drug-sensitive (A) and drug-resistant (B) lines oftrypanosomes. Spots indicated with arrows were identified byMS with the results shown in Table 1.

Tb09.211.0120 and Tb09.211.0130 in the T. brucei database inGeneDb, hereafter referred to as ORFs 0120 and 0130respectively. One or both of these genes also encodes thespot 1 polypeptide in both drug-sensitive and drug-resistanttrypanosome lines. In the genome reference line, 927, thesetwo ORFs have identical sequences and are identified as‘nascent polypeptide associated complex subunit, putative’(NAC) on the basis of the presence of a NAC domain asdetected using Pfam and Interpro algoritms. Another algo-rithm, PredictProtein (http://cubic.bioc.columbia.edu/pre-dictprotein/), also predicts a ubiquitin-associated (UBA) do-main near the C-terminus of this protein. Thus, the resistantline lacks one of the NAC polypeptide isoforms.

To confirm this identification we made an antiserumagainst the recombinant protein using a prime-boost strate-gy. The antiserum detected a single 34 kDa band on Westernblots of 1-D gel separations of protein extracts from bothisogenic lines; serum from vector only-immunised mice wasused a negative control and detected no bands (data notshown). Western blots of mini 2-D gel separations detectedonly a single spot in the drug-resistant 386Mr line but threeadjacent spots in the drug-sensitive 386 line, all in theexpected region of the blot (Fig. 3). The detection of three

spots in 386 rather than two was an unexpected finding. Itmay indicate perhaps that the antiserum has greater sensi-tivity compared with Coomassie Blue or SYPRO Ruby stain-ing and so detects a third isoform not visible by direct proteinstaining. This finding raises the possibility that drug resis-tance is associated with loss of either one isoform, detectedby direct staining of protein, or two isoforms, detected byWestern blotting. The two visualisation methods agree,however, that isoform(s) have been lost and one isoformremains.

3.3 Molecular analysis of NAC expression

Identification of two paralogues in the genome referencestrain and two protein products in 386 naturally leads to thehypothesis that each of the two paralogous genes in

Figure 3. Western blot of NAC protein in isogenic cymelarsan-sensitive (A) and cymelarsan-resistant (B) lines separated by 2DEwith a pH 4–7 range in the first dimension. Three isoforms appearto be detected in panel A but only one in panel B.



386 codes for a polypeptide that differs slightly in pI due to aslight sequence difference. To test this possibility, we firstinvestigated whether there were two copies of nac in 386 asfound in the genome reference strain. To do this, we madeuse of the observation that the ORFs 0120 and 0130 in927 differ in their flanking sequences suggesting that itshould be possible to amplify each separately. Primers basedon the 927 sequence successfully amplified DNA fragmentsof the expected size in both 386 and 386Mr and RFLP analy-sis indicated that each of these fragments contained a nacgene (data not shown). Therefore, there were two copies ofthe gene in 386Mr. Sequence analysis of these ampliconsindicated that the ORFs of both paralogues of nac in 386 wereidentical to each other and to the genome reference sequence(data not shown). These data show that the two spots detect-ed on gels were not the products of separate genes with aslight difference in sequence resulting in a difference in pIand also indicate that the loss of the spot in 386Mr was notcaused by the deletion of one of the genes.

To determine whether both gene copies were transcribedin 386Mr as well as 386, RT-PCR analysis was undertaken,using the pairs of primers that differentiate the two genes.Products of the expected sizes were identified, indicating thatboth genes were transcribed in both trypanosome lines(Fig. 4). This result demonstrates that the absence of spot 2in 386Mr could not be ascribed to absence of transcripts ofone of the gene paralogues. Although the formal possibilityof a change in quantitative level of transcription leading toabsence of a protein isoform cannot be excluded, it seemsunlikely.

4 Discussion

We have undertaken a proteomic analysis to compare twoisogenic lines of the human infective pathogen T. b. gam-biense that differ in sensitivity to cymelarsan. This drug is a

Figure 4. RT-PCR analysis of both copies of the nac gene,Tb09.211.0120 and 0130, showing that each gene copy is ex-pressed in both drug-sensitive and drug-resistant lines of theparasite. mRNA was extracted from the drug-sensitive (tracks 1,3, 5 and 7) or drug-resistant lines (tracks 2, 4, 6 and 8). RT-PCRreactions were conducted using primers specific for ORF 0120(tracks 1 and 2) or ORF 0130 (tracks 3 and 4). RT-PCR of the TIMgene before (tracks 7 and 8) and after (tracks 5 and 6) reversetranscription acted as negative and positive controls.

melaminophenyl arsenical compound and is used clinicallyfor the treatment of veterinary trypanosomiasis in camels. Itis closely related to melarsoprol, the most widely used drugfor treatment of sleeping sickness. Crossresistance betweencymelarsan and melarsoprol has been demonstrated in theparasite lines used in these current [14, 20] and other [8, 21]studies. Different mechanisms of resistance have beenreported [5, 6, 10] suggesting that alternative routes to thegeneration of resistance are possible but none of these stud-ies have been undertaken with a human infective isolate. Thedrug target for arsenical drugs is as yet unknown [22] butresistance can be caused by modification of a drug target, andso the study of the mechanisms of resistance could poten-tially lead to the identification of the mechanism of drugaction [23].

We have identified an association between acquisition ofarsenical resistance and loss of one or two isoform(s) of aputative NAC protein. Interestingly, the identification of thisresistance-associated gene product would have been difficultto detect using more traditional biochemical approaches orLC-MS approaches such as MudPIT that do not readily dis-tinguish protein isoforms.

Comparative proteome scanning approaches to investi-gate specific alterations associated with changes in pheno-type in trypanosomes have not previously been undertaken,despite the demonstrated utility of this strategy in investi-gation of methotrexate resistance in the related pathogen,L. major [11] and to compare expression in different lifecycle stages in L. mexicana [24], L. major [25] and L. donovani[26]. We have been able to compare approximately2000 polypeptide spots on 2-D gels between the drug-sensi-tive and drug-resistant trypanosome lines. This represents asubstantial portion of the expressed proteins of T. brucei,which contains approximately 9000 predicted genes [12]. Amore accurate estimate of the proportion of the genomecovered in our study is difficult to discern for a number ofreasons, not least because some 75% of gene products haveco translational or post-translational modifications leadingto detectable differences in pI and/or molecular weight andmost spots (at least on pH 4–7 range gels) comprised two ormore products of separate genes [13]. Also, it is unknownwhat proportion of the protein complement has its expres-sion stage-regulated in the trypanosome life cycle. Ourmethod of sample preparation is likely to have led to anunderrepresentation of very low abundance and membrane-bound proteins, although we have not formally analysedthis issue. We could find no evidence of induction of anychanges in proteome profile by incubation of parasites insublethal concentrations of cymelarsan, although inductionhas been observed in other drug-resistant pathogens (see[27] for example).

The potential function of the NAC protein in drugresistance can, at this stage, be inferred only from com-parative bioinformatic analyses that indicated the presenceof two domains (NAC and UBA). Both domains are pres-ent in NAC proteins: abundant cytosolic protein com-



plexes, heterodimeric in eukaryotes and homodimeric inarchaea [28]. There is agreement in the literature as to thelocation of NAC, associated with ribosomes and in prox-imity to new polypeptides emerging from the ribosome,but the biological function(s) of NAC remain controversial.NAC may act to shield nascent polypeptides from othercytosolic elements and/or to regulate translocation into theendoplasmic reticulum and/or mitochondria. Individualsubunits of NAC have also been implicated in transcrip-tional regulation [29]. NAC proteins have been reported tobe downregulated in response to salt stress in rice [30].Thus, they may have a role in translational and/or tran-scriptional regulation and/or protein trafficking, but it isuncertain how changes in these processes might generatearsenical drug resistance.

In our view, it is unlikely that loss of one isoform of NACin T. brucei is directly responsible for arsenical drug resis-tance/sensitivity as there are no changes in the ORFsequences or their transcription between the resistant andsensitive lines. To test directly for a causal link, however, isextremely difficult. In a reverse genetic knockdown/knock-out experiment for example, three outcomes can be envi-saged: lethal, not lethal and no change in resistance pheno-type or not lethal and change in resistance. All three out-comes are uninformative, the first two for reasons that areself-evident. The third outcome would indicate a directinvolvement of the NAC gene in drug resistance. But, be-cause there are multiple isoforms of the protein and one ofthese is not changed, it neither refutes nor supports theconjecture that a change in only one or two isoforms deter-mines resistance as suggested by the proteomics evidence.The evidence we present suggests that it is more likely thatNAC is a marker for resistance. Given that PTMs of proteinsare apparently widespread in T. brucei [13] we would suggestthat a more likely explanation for the absence of NAC iso-form(s) in drug-resistant trypanosomes is the loss of a PTM.There are several candidate sites for PTMs in the sequence,including one for N-glycosylation, one for SUMO-based pro-tein binding and seven for phosphorylation. Our data leadsto the hypothesis that the causal event determining arsenicalresistance is an alteration in the activity of the enzyme thatgenerates the PTM.

We thank our colleagues in the Sir Henry Wellcome Func-tional Genomics Facility for their help with MS. We are gratefulto The Wellcome Trust for financial support.

5 References

[1] Barrett, M. P., The Lancet 1999, 353, 1113–1114.

[2] Legros, D., Fournier, C., Gastellu, E. M., Maiso, F., Szumilin, E.,Bull. Soc. Pathol. Exot. 1999, 92, 171–172.

[3] Stanghellini, A., Josenando, T., Trop. Med. Int. Health 2001, 6,330–334.

[4] Matovu, E., Geiser, F., Schneider, V., Maser, P. et al., Mol.Biochem. Parasitol. 2001, 117, 73–81.

[5] Carter, N. S., Fairlamb, A. H., Nature 1993, 361, 173–176.

[6] Maser, P., Sutterlin, C., Kralli, A., Kaminsky, R., Science 1999,285, 242–244.

[7] De Koning, H. P., Jarvis, S. M., Mol. Pharmacol. 1999, 56,1162–1170.

[8] Matovu, E., Stewart, M. L., Geiser, F., Brun, R. et al., Eukar-yotic Cell 2003, 2, 1003–1008.

[9] Matovu, E., Enyaru, J. C., Legros, D., Schmid, C. et al., Trop.Med. Int. Health 2001, 6, 407–411.

[10] Shahi, S. K., Krauth-Siegel, R. L., Clayton, C. E., Mol. Micro-biol. 2002, 43, 1129–1138.

[11] Drummelsmith, J., Girard, I., Trudel, N., Ouellette, M., J.Biol. Chem. 2004, 279, 33273–33280.

[12] Berriman, M., Ghedin, E., Hertz-Fowler, C., Blandin, G. et al.,Science 2005, 309, 416–422.

[13] Jones, A., Faldas, A., Foucher, A., Hunt, E. et al., Proteomics2006, 6, 259–267.

[14] Scott, A. G., Tait, A., Turner, C. M., Acta Trop. 1996, 60, 251–262.

[15] Berger, B. J., Fairlamb, A. H., Antimicrob. Agents Chemo-ther. 1994, 38, 1298–1302.

[16] Van Deursen, F. J., Thornton, D. J., Matthews, K. R., Mol.Biochem. Parasitol. 2003, 128, 107–110.

[17] Hellman, U., Wernstedt, C., Gonez, J., Heldin, C. H., Anal.Biochem. 1995, 224, 451–455.

[18] MacLeod, A., Turner, C. M. R., Tait, A., Mol. Biochem. Para-sitol. 1999, 102, 237–248.

[19] MacLeod, A., Turner, C. M. R., Tait, A., Mol. Biochem. Para-sitol. 1997, 84, 267–70.

[20] Scott, A. G., Tait, A., Turner, C. M. R., Exp. Parasitol. 1997, 86,181–190.

[21] Popischal, H., Brun, R., Kaminsky, R., Jenni, L., Acta Tropica1994, 58, 187–197.

[22] Denise, H., Barrett, M. P., Biochem. Pharmacol. 2001, 61, 1–5.

[23] Barrett, M. P., Fairlamb, A. H., Parasitol. Today 1999, 15, 136–140.

[24] Nugent, P. G., Karsani, S. A., Wait, R., Tempero, J., Smith, D.F., Mol. Biochem. Parasitol. 2004, 136, 51–62.

[25] El Fakhry, Y., Ouellette, M., Papadopoulo, B., Proteomics2002, 2, 1007–1017.

[26] Bente, M., Harder, S., Wiesgigl, M., Heukeshoven, J. et al.,Proteomics 2003, 3, 1811–1829.

[27] McAtee, C. P., Hoffman, P. S., Berg, D. E., Proteomics 2001, 1,516–521.

[28] Spreter, T., Pech, M., Beatrix, B., J. Biol. Chem. 2005, 280,15849–15854.

[29] Rospert, S., Dubaquie, Y., Gautschi, M., CMLS, Cell. Mol. LifeSci. 2002, 59, 1632–1639.

[30] Yan, S., Tang, Z., Su, W., Sun, W., Protoemics 2005, 5, 235–244.


Documents

A proteomic analysis of arsenical drug resistance in Trypanosoma brucei