Knockdown of Asparagine Synthetase A Renders Trypanosoma brucei Auxotrophic to Asparagine

Asparagine synthetase (AS) catalyzes the ATP-dependent conversion of aspartate into asparagine using ammonia or glutamine as nitrogen source. There are two distinct types of AS, asparagine synthetase A (AS-A), known as strictly ammonia-dependent, and asparagine synthetase B (AS-B), which can use either ammonia or glutamine. The absence of AS-A in humans, and its presence in trypanosomes, suggested AS-A as a potential drug target that deserved further investigation. We report the presence of functional AS-A in Trypanosoma cruzi (TcAS-A) and Trypanosoma brucei (TbAS-A): the purified enzymes convert L-aspartate into L-asparagine in the presence of ATP, ammonia and Mg2+. TcAS-A and TbAS-A use preferentially ammonia as a nitrogen donor, but surprisingly, can also use glutamine, a characteristic so far never described for any AS-A. TbAS-A knockdown by RNAi didn't affect in vitro growth of bloodstream forms of the parasite. However, growth was significantly impaired when TbAS-A knockdown parasites were cultured in medium with reduced levels of asparagine. As expected, mice infections with induced and non-induced T. brucei RNAi clones were similar to those from wild-type parasites. However, when induced T. brucei RNAi clones were injected in mice undergoing asparaginase treatment, which depletes blood asparagine, the mice exhibited lower parasitemia and a prolonged survival in comparison to similarly-treated mice infected with control parasites. Our results show that TbAS-A can be important under in vivo conditions when asparagine is limiting, but is unlikely to be suitable as a drug target.


Introduction
Asparagine is a naturally occurring non-essential amino acid found in many proteins. Due to its high nitrogen/carbon ratio, asparagine is likely to be linked to nitrogen homeostasis and protein biosynthesis [1]. AS is the protein involved in asparagine biosynthesis. There are two distinct types of AS, AS-A and AS-B, encoded by asnA and asnB genes, respectively. AS-A encoding genes have been reported in archaea [2,3], prokaryotes [4][5][6][7], and in the protozoan parasite Leishmania [8]. The AS-B encoding gene is present in prokaryotes [5,9] and also in eukaryotes, including mammalian cells [10,11], yeasts [12], algae [13], and higher plants [14]. Both types catalyze the ATP-dependent conversion of aspartate into asparagine. While AS-B can use both ammonia and glutamine (reaction B) as amide nitrogen donors [5,[15][16][17][18][19][20], Escherichia coli (E. coli) AS-A was reported to be dependent strictly on ammonia (reaction A) [21,22].
A) ATP + L-aspartate + NH4 + = . AMP + diphosphate + Lasparagine B) ATP + L-aspartate + L-glutamine [or NH4 + ] = . AMP + diphosphate + L-asparagine + L-glutamate AS-A and AS-B share no sequence or structural similarities. Their three-dimensional structures provided important information concerning their distinct catalytic mechanisms [2,[23][24][25]. AS-A exists as a dimer where each monomer has a core of eight bstrands flanked by a-helices, resembling the catalytic domain of class II aminoacyl-tRNA synthetases such as aspartyl-tRNA synthetase [24]. AS-A synthesizes asparagine in two steps: the bcarboxylate group of aspartate is first activated by ATP to form an aminoacyl-AMP, followed by amidation by a nucleophilic attack with an ammonium ion [2]. The AS-B enzyme also forms a dimer, but each monomer contains two distinct domains, each of which contains a catalytic site. The N-terminal site catalyzes the conversion of glutamine into glutamic acid and ammonia, while aspartate reacts with ATP in the C-terminal site, generating the intermediate b-aspartyl-AMP [26,27]. Similarly to other glutamine dependent amidotransferases, ammonia released in the Nterminal domain of the enzyme travels through an intramolecular tunnel connecting the active sites, and reacts with the reactive acyladenylate intermediate to produce asparagine [28].
An open reading frame encoding a putative AS-A is present in the genome of the protozoan parasites, Trypanosoma cruzi (T. cruzi) and Trypanosoma brucei (T. brucei) [29][30][31]. T. cruzi and T. brucei are transmitted to a mammalian host through an invertebrate vector, and are responsible for Chagas disease and African sleeping sickness, respectively. Disease control is dependent on drug therapy, but treatment options are limited, both by high toxicity and recent emergence of drug resistance [32][33][34]. Vaccines for T. brucei infections are unlikely to be developed not only because of extensive antigenic variation [35], but also because infections compromise host humoral immune competence [36].
Trypanosome AS-A might be a drug target due to the absence of a homologue in humans [8]. AS-A is important in other microorganisms. For example, asnA is an essential gene in Haemophilus influenzae (DEG10050178) [37], and is strongly upregulated in Pasteurella multocida during host infection [38], and when Klebsiella aerogenes is grown in an amino acid-limited but ammonia rich environment [5]. We therefore undertook biochemical and genetic studies of AS-A in trypanosomes to ascertain its biological role and evaluate its potentiality as drug target.

Ethics statement
All experiments involving animals were carried out in accordance with the IBMC.INEB Animal Ethics Committees and the Portuguese National Authorities for Animal Health guidelines, according to the statements on the directive 2010/63/EU of the European Parliament and of the Council. IL, JT and ACS have an accreditation for animal research given by the Portuguese Veterinary Direction (Ministerial Directive 1005/92).

Parasite culture
Procyclic and bloodstream forms of Trypanosoma brucei brucei Lister 427 were used. Procyclic forms were grown in MEM-Pros medium supplemented with 7.5 mg/ml hemin, 10% fetal calf serum (FCS) and 100 IU/mL of penicillin/streptomycin at 27uC, with cell densities between 5610 5 cells/ml to 1-2610 7 cells/ml. Bloodstream forms were grown in complete HMI-9 medium (supplemented with 10% FCS and 100 IU/mL of penicillin/ streptomycin) [39] in vented tissue culture flasks; these cultures were diluted when cultures reached the cell density of 2610 6 /ml and incubated in a humidified atmosphere of 5% CO2, at 37uC. Bloodstream RNAi cell cultures were supplemented with 7.5 mg/ ml hygromycin and 0.2 mg/ml phleomycin.

Expression and purification of poly-His-tagged recombinant TbAS-A and TcAS-A
The TbASA and TcASA genes were subcloned into pET28a(+) expression vector (Novagen). The recombinant 6-His-tagged proteins were expressed in E. coli BL21DE3 by induction of logphase cultures with 0.5 mM IPTG (NZYTech) for 3 h at 37uC (TcAS-A) and at 18uC, overnight (O/N) (TbAS-A). Bacteria were harvested and resuspended in buffer A [0.5 M NaCl (Sigma), 20 mM Tris.HCl (Sigma), pH 7.6]. The sample was sonicated and centrifuged to obtain the bacterial crude extract. The recombinant proteins were purified using Ni 2+ resin (ProBond) and washing and elution with increasing levels (25  To generate rat and rabbit polyclonal antibodies against TbAS-A, each animal was first immunized with 150 mg of recombinant TbAS-A protein. After 2 weeks, 4 boosts with 100 mg of recombinant TbAS-A were given weekly. The collected blood samples were centrifuged to obtain the serum.
For Western blotting, 2 mg of recombinant TbAS-A and TcAS-A proteins, 10 mg of total soluble cell extract, or 1610 7 parasites, were resolved in SDS/PAGE and transferred on to a nitrocellulose

Author Summary
The amino acid asparagine is important not only for protein biosynthesis, but also for nitrogen homeostasis. Asparagine synthetase catalyzes the synthesis of this amino acid. There are two forms of asparagine synthetase, A and B. The presence of type A in trypanosomes, and its absence in humans, makes this protein a potential drug target. Trypanosomes are responsible for serious parasitic diseases that rely on limited drug therapeutic options for control. In our study we present a functional characterization of trypanosomes asparagine synthetase A. We describe that Trypanosoma brucei and Trypanosoma cruzi type A enzymes are able to use either ammonia or glutamine as a nitrogen donor, within the conversion of aspartate into asparagine. Furthermore, we show that asparagine synthetase A knockdown renders Trypanosoma brucei auxotrophic to asparagine. Overall, this study demonstrates that interfering with asparagine metabolism represents a way to control parasite growth and infectivity.

Enzyme assays
AS activity was assessed by quantification of asparagine formation [41]. The reactions were performed in a total of 150 ml of enzyme assay mixture in 85 mM Tris-HCl (Sigma) containing aspartate (Sigma), ammonia (Sigma), ATP (Sigma) and 8.4 mM Mg 2+ (Sigma). Following incubation for set times at 37uC, enzymatic reactions were terminated by boiling 4 min, and then centrifuged at maximum speed for 30 s. 100 ml of the reaction mixture supernatant was added to 900 ml of ninhydrin 0.05% in ethanol. The resulting mixtures were boiled at 100uC for 5 min, then centrifuged for 30 s and maintained on ice. 300 ml of clear supernatant fluids were transferred to 96-well plates, and the absorbance at 340 nm determined [41]. Based on reaction linearity studies, 7.5 mg of enzyme and 15 min incubation at 37uC were selected as final conditions. To determine K m s, the concentrations of substrates were varied in the following ranges: .50 mM). K m for glutamine was determined using a concentration range of 1.5625-25 mM, while ATP and aspartate were maintained in excess. Measurements were performed in triplicate, and the initial rate was analyzed to obtain values of V max and K m by curve fitting using GraphPad Prism (5.0 version).
Using a query based on L-cysteine-S-sulfinic acid inhibitor [42], the ZINC database was screened using the program ROCS (version 2.3.1) to find compounds that have good shape similarity (measured by 3D Tanimoto) and similar functional group overlap to the query molecule. L-cysteine-S-sulfate (Sigma; PubChem Substance ID 24892471) was used under the following conditions: 2.5 mM aspartate, 1.25 mM ATP, 12.5 mM ammonia, and 8.4 mM Mg 2+ . The characterization of the mechanism of inhibition consisted in the determination of K m and V max for each substrate, in the presence of four inhibitor concentrations (0.025, 0.050, 0.1 and 0.2 mM). The following substrate concentration ranges 1.25-10 and 1.25-50 mM were used for aspartate and ammonia, respectively, while to determine K m for ATP, a range from 0.625 to 10 mM (TbAS-A) or from 0.3125 to 5 mM (TcAS-A) was assayed. K i was determined by ''K m app Method'' [43].

Generation of TbAS-A RNAi cell lines
The ''stuffer strategy'' was used to generate RNAi-mediated AS-A depletion. First, the TbASA fragment (amplified with a sense oligo with a BglII -SphI linker 59 -GAGAAGATCTGCA-TGCGCGACGACGGTTATTCGTCATAC -39, and an antisense oligo with a EcoRI -SalI 59 -CGGAATTCGTCGA-CACTCCGTTTTTCGGATTGCGGC -39) was cloned twice in opposite direction on either sides of a 'stuffer' fragment of the pHD1144 vector (also digested with SphI and SalI) (Fig. S1A). The The residues are colored according to ALSCRIPT Calcons (Aline version 011208) using a predefined colour scheme (red: identical residues; orange to blue: scale of conservation of amino acid properties in each alignment column; white: dissimilar residues). Secondary structure elements of EcAS-A crystal structure (grey) and of TcAS-A (purple) and of TbAS-A (blue) homology models are depicted above the alignment. In all protein sequences, asparagine (squares) and AMP binding residues (circles) were identified. The dashed box indicates a structurally divergent region. doi:10.1371/journal.pntd.0002578.g001 resulting [(target)-stuffer-(reverse-complement target)] construct obtained through HindIII and BglII digestion, which generates a stem-loop RNA, was cloned into pHD1145 (also digested with HindIII and BglII) (Fig. S1B). The final construct was linearized with NotI and 10 mg of DNA was transfected into 2610 7 /ml bloodstream form cell line carrying pHD1313 plasmid (contains two copies of the tet repressor and a phleomycin resistance cassette) by electroporation using Amaxa Basic Parasite Nucleofector Kit (Lonza). Transcription occurs on induction with tetracycline (100 ng/ml), hence producing mRNA homologs to the target the gene. Stable individual clones were selected 5 to 7 days after transfection with 7.5 mg/ml of hygromycin.

In vitro analysis of TbAS-A RNAi
To analyse growth, T. brucei RNAi cell line and cells expressing the tet repressor only (wt), were seeded at 2610 5 cells/ml of complete HMI9 medium, in the presence and absence of tetracycline. Cell growth was monitored microscopically on a haemocytometer (Marienfeld) and the culture diluted back to 2610 5 cells/ml daily. The same protocol was repeated in complete HMI9 medium with basal IMDM without asparagine, complete HMI9 medium with basal IMDM without asparagine supplemented with 6.7610 4 nM of asparagine (levels found in human plasma [49]), and complete HMI9 medium with basal IMDM without asparagine supplemented with 1.67610 5 nM of asparagine (levels found in normal medium).

In vivo analysis of TbAS-A RNAi
Wild-type and transgenic bloodstream T. brucei parasites were cultured in the absence of selecting drugs (hygromycin and phleomycin) for 24 h, then tetracycline was added. After a further 48 h, parasites were inoculated in mice. For each experiment, 4 groups of BALB/c mice (6-8 weeks old, n = 4) (Harlan Laboratories, United Kingdom) were infected by intraperitoneal injection of 10 4 T. brucei bloodstream forms. 2 groups were injected with wt strain (with or without tetracycline) and the other 2 groups were injected with RNAi cell line (with or without tetracycline). 48 h prior infection the 2 RNAi induced groups were given doxycycline (treated with 1 mg/ml doxycycline hyclate and 5% sucrose containing water). The 2 non-induced groups were given standard water. To evaluate the virulence of RNAi induced parasites in mice with reduced plasmatic levels of asparagine, animals were treated with 50 IU of E. coli L-asparaginase (ProSpec-Tany TechnoGene) 48 h before injection and every 48 h. According to the literature, L-asparagine could not be detected in the blood 48 h following an intravenous injection of E. coli L-asparaginase, at a dose of 50 IU/mouse [50]. Mice were monitored every day for general appearance and behaviour. Parasitemia was monitored daily from the fifth day post-infection, using tail-vein blood, in a haematocytometer under a microscope. Animals with a parasitemia greater than 10 8 parasites/ml were euthanized, as previous studies had established that these levels were consistently lethal within the next 24 h.

Immunofluorescence
T. brucei bloodstream forms from log-phase cultures, with or without RNAi, were fixed in m-Chamber 12 well (Ibidi) for 15 min, at room temperature, in PBS containing 3% p-formaldehyde, washed twice with PBS, and then permeabilized in PBS containing 0.1% of Triton X-100. Fixed cells were incubated for 60 min in PBS containing 10% FCS at room temperature (RT), in a humidified atmosphere, then washed twice with PBS/2% FCS. Cells were then incubated with primary rat or rabbit polyclonal antibody against TbAS-A (1:100 and 1:5000 respectively, both  Fig. 1. (B) Superposition of EcAS-A structure (grey) (PDB accession code 12AS), with TbAS-A (blue) and TcAS-A (purple) homology models (obtained from the SWISS-MODEL server, using PDB 12AS as a template). A small structurally divergent region is marked by a dashed rectangle. Ligand color schemes: asparagine is shown in yellow (oxygen, red; nitrogen blue) and AMP is shown in brown (oxygen, red; nitrogen blue; phosphorous orange). doi:10.1371/journal.pntd.0002578.g002 diluted in blocking solution) overnight at 4uC, followed by two washes with PBS/2% FCS. Subsequently, cells were incubated with Alexa Fluor 647 conjugated goat anti-rat or Alexa Fluor 488 conjugated goat anti-rabbit secondary antibodies (Molecular probes from Life technologies) (1:500 diluted in blocking solution) for 1 h at RT in an humidified atmosphere, then washed twice with PBS. Next, the slides were stained and mounted with Vectashield-DAPI (Vector Laboratories, Inc.). Images were captured using fluorescence microscope AxioImager Z1 and software Axiovision 4.7 (Carl Zeiss, Germany). Pseudo-coloring of images was carried out using ImageJ software (version 1.43u).
In case of TbAS-A immunolocalization, T. brucei wt bloodstream forms cells were co-stained using rat anti-TbAS-A antibody (1:100 diluted in blocking solution), rabbit anti-aldolase antibody (1:5000 diluted in blocking solution), anti-BiP antibody (kindly provided by Dr. Jay Bangs, 1:500 diluted in blocking solution), anti-enolase antibody (kindly provided by Dr. Paul Michels, 1:5000 diluted in blocking solution) or anti-GRASP antibody (kindly provided by Dr. Graham Warren, 1:200 diluted in blocking solution). Alexa Fluor 647 conjugated goat anti-rat (1:500) and Alexa Fluor 488 conjugated goat anti-rabbit (1:500) were used as secondary antibodies. Staining with MitoTracker Orange (Invitrogen) followed by Alexa Fluor 488 conjugated goat anti-rabbit (1:500),  as a secondary antibody. The labelling of parasites with MitoTracker was done by adding 250 nM to the cell culture medium (without FCS) for 30 minutes, prior to washing, fixing and staining using the protocol described above. Images were captured using the confocal microscope Leica TCS SP5II and LAS 2.6 software (Leica Microsystems, Germany). Again, image analysis was done using ImageJ version 1.43U software.

Digitonin permeabilization
For each sample condition, 1.0610 7 bloodstream cells were washed once with cold trypanosome homogenization buffer (THB), containing 25 mM Tris (Sigma), 1 mM EDTA (Sigma) and 10% sucrose (Sigma), pH = 7.8. Just before cell lysis, leupeptin (Sigma) (final concentration of 2 mg/ml) and different digitonin (Calbiochem) quantities (final concentrations of 5, 12.5, 25, 50, 100, 150 and 200 ug/ml) were added to 500 ml of cold THB, for cell pellet resuspension. Untreated cells (0 mg/ml of digitonin) and those completely permeabilized (total release, the result of incubation in 0.5% Triton X-100) were used as controls. Each sample was incubated 60 min on ice, and then centrifuged at 2000 rpm, 4uC, for 10 min. Supernatants were transferred to new chilled tubes and 500 ml of cold THB was added to each pellet and then mixed. All fractions were analysed through Western blot as described above.
Cell cycle analysis T. brucei bloodstream forms were analyzed by flow cytometry for DNA content following RNAi induction. Cells were collected by centrifugation and washed twice with PBS containing 2% FCS. Each 2610 6 cells were resuspended in 1 mL of PBS/2% FCS and 3 mL of cold absolute ethanol was added while vortexing. Cells were fixed for 1 hour at 4uC and then washed twice in PBS. 1 mL of staining solution [3.8 mM sodium citrate dehydrate (Sigma), 50 mg/mL propidium iodide (Sigma), 0.5 mg/mL RNAse A (Sigma) in PBS] was added to the cell pellets and vortex. Samples were analysed by FACS (Becton Dickinson) after a incubation at 4uC for 30 min. Data was analyzed by FlowJo software (Ashland, OR).

Statistical analysis
One-way ANOVA and two-tailed Student's test were used for statistical analysis. Statistical analysis was performed using GraphPad Prism Software (version 5.0), and p values#0.01 were considered to be statistically significant. Asterisks indicate statistically significant differences (*** p#0.001, ** p#0.01).
Like EcAS-A, TbAS-A and TcAS are predicted to be dimeric, as seen from superimposed homology models with the EcAS-A crystal structure [24] (Fig. 2A). The only structurally divergent region (area marked by dashed rectangle) (Fig. 1, 2B), is present in both TbAS-A (from residues Q232 to S250) and TcAS-A (from residues D232 to S247), but absent in EcAS-A. This region is distant from the enzyme active site and the dimer interface and its functional and structural significance are unknown. The amino acids involved in asparagine binding are all strictly conserved, while in the AMP binding pocket, the majority of the residues are conserved, except for three residues (Fig. 1). In EcAS-A, E103 (D106 and E106 in TbAS-A and TcAS-A, respectively) and L109 (I112 and T112 in TbAS-A and TcAS-A, respectively) ( Fig. 1) are not directly involved in polar interactions with the nucleotide base, but form part of the outer wall of the binding site [24]. The main chain of L249 in EcAS-A (V271 and L268 in TbAS-A and TcAS-A, respectively) is directly involved in hydrogen bonds with ribose from AMP, however the different side chains of leucine and valine do not affect the shape of AMP binding site.

Trypanosome AS-A catalyze asparagine synthesis using either ammonia or glutamine as nitrogen donors
TbAS-A and TcAS-A coding sequences were cloned into the bacterial expression vector pET28a. Histidine-tagged fusion proteins were purified under non-denaturing conditions (Fig. 3A,  B). As expected, the recombinant proteins were recognized by anti-His Tag monoclonal antibody (Fig. 3A, B). Rabbit polyclonal antibodies produced against recombinant TbAS-A recognized the protein in total extracts from two different parasite developmental stages, bloodstream forms (mammalian host parasite stage) and procyclic forms (insect vector parasite stage) (Fig. 3C).
The capacities of TbAS-A and TcAS-A to produce asparagine from aspartate in the presence of ATP, ammonia and Mg 2+ were determined using a specific quantitative colorimetric assay for Lasparagine [41]. The pH optimum was 7.6, with detectable activity from 6.0 to 9.0 (data not shown). Mg 2+ was an essential cofactor for TbAS-A and TcAS-A (data not shown), as previously described for EcAS-A [22]. We included 8.4 mM Mg 2+ in the final reaction mixture. Lower concentrations (2, 4 and 6 mM) gave lower activity while increased concentrations (up to 16 mM) resulted in no substantial activity improvement (data not shown). TbAS-A and TcAS-A showed similar K m s for aspartate and ammonia (p.0.01), while TcAS-A showed higher K m for ATP than TbAS-A (p = 0.0042) ( Table 1). ATP is the substrate required for the generation of the b-aspartyl adenylate intermediate, which reacts with ammonia, releasing asparagine. In its absence, the reaction did not occur (data not shown). To our surprise, both TbAS-A and TcAS-A could also use glutamine as a nitrogen donor ( Table 2). TbAS-A showed higher K m for this nitrogen donor than TcAS-A, however not statistically significant (p.0.01). Both enzymes present higher K m values for ammonia than for glutamine, but these differences were not statistically significant (p.0.01) ( Table 1 and 2). TbAS-A had a higher V max than TcAS-A, for both ammonia (p,0.0001) and glutamine (p = 0.0043) dependent-activities (Table 1 and 2). TbAS-A had similar catalytic rates for both glutamine and ammonia-dependent activities (p.0.01), whereas TcAS-A presented a slightly higher, but not statistically significant, rate for glutamine-dependent activity (p.0.01) ( Table 1 and 2). The high conservation of the active sites and the small amino acid differences identified in the homology models do not allow an accurate structural interpretation of the small differences observed. Indeed, these might have been due to slight differences in the proportion of protein that was correctly folded. L-cysteine-S-sulfate, considered a putative AS-A inhibitor from a virtual screening, inhibited both enzymes, with IC 50 s of 126 and 100 mM for TbAS-A and TcAS-A, respectively (Fig. 4A). For both enzymes, the kinetic characteristics suggested competition with ATP binding (Fig. 4C, D). No changes in the K m s and V max s for aspartate and ammonia were observed (p.0.01) (Fig. 4E, F, G, H), suggesting the inhibition is exclusively due to ATP binding interference (p#0.01) . K i values of 137 and 128.9 mM were determined for TbAS-A and TcAS-A, respectively (Fig. 4B).

AS-A localizes in the cytosol of T. brucei bloodstream forms
The subcellular localization of TbAS-A was determined by immunofluorescence and digitonin fractionation in bloodstream forms. As expected, induction of RNAi resulted in a decrease in the fluorescence intensity (Fig. S2A, B, C). TbAS-A is in the cytosol, as revealed by colocalization with the cytosolic enzyme enolase [51] (Fig. 5A) and no colocalization with aldolase, BiP, GRASP or mitotracker (Fig. S3), markers for glycosomes [52],  Circles and squares represent wild-type growth in the absence or presence of tetracycline, respectively, while up triangles and down triangles represent clones growth in the absence or presence of tetracycline, respectively. Cumulative cell numbers are plotted as the product endoplasmic reticulum [53], Golgi and mitochondria compartments [54], respectively. Controls performed with rat or rabbit pre-immune sera and secondary antibody alone, showed no detectable signal (data not shown). Digitonin fractionation also resulted in similar profiles for AS-A and enolase (cytosolic marker) and no similarity to aldolase (glycosomes marker) (Fig. 5B).

AS-A knockdown makes T. brucei bloodstream forms dependent on extracellular asparagine
To study the biological role of AS-A in T. brucei bloodstream forms, cells were stably transfected with an RNA interference plasmid construct. RNAi against asparagine synthetase A was induced in normal medium (complete HMI-9) by adding tetracycline. No difference was observed in cell proliferation between induced and non-induced cells (Fig. 6A), although AS-A protein was reduced to <13% of the normal level within 48 hours (Fig. 6B). When, however, the AS-A-depleted cells were grown in HMI-9 medium with only the asparagine from the fetal calf serum, growth was impaired, with an increase in the proportion of cells in G0/G1 (Fig. 6C and 7B, C). Presumably the asparagine from the serum allowed this slower growth. Levels of asparagine usually found in human serum (6.7610 4 nM) [49], which are somewhat lower than in normal medium (1.67610 5 nM; IMDM -Iscove's modified Dulbecco's basal medium from Gibco Invitrogen), were sufficient to overcome this defect (Fig. 6E, G). In complete HMI-9 medium, with only asparagine from the fetal calf serum, the growth defects of induced RNAi clones are abrogated at day 5 post-induction (Fig. 6C, 7D), and the percentage of cells in GO/ G1 and S phases of the cell cycle return to the ones found in noninduced cells (Fig. 7A), suggesting the appearance of RNAi revertants, as is also visible on the Western blot (Fig. 6D). Similar reversion to evade lethal RNAi in trypanosomes has been seen many times before [55]. In the presence of asparagine, low AS-A levels were maintained (Fig. 6B, F and H).
of cell number and total dilution. Error bars indicate standard deviation. The effect of RNAi on the AS-A protein levels was analyzed by Western blots with extracts of noninduced tet (2), and RNAi-induced tet(+) cells, isolated from unmodified medium (B), modified medium (complete HMI-9 medium with basal IMDM without asparagine) (D), modified medium supplemented with 6.7610 4 nM of asparagine (F), modified medium supplemented with 1.67610 5 nM of asparagine (H). Cells were collected at day 1, 2, 6 and 10 of RNAi induction. Results are representative of three independent experiments. doi:10.1371/journal.pntd.0002578.g006 Normal T. brucei parasites also showed a statistically significant slower growth under conditions of asparagine limitation (p#0.01; data not shown) (compare Fig. 6C, with A, E, G). It is therefore possible that even when the parasite has AS, it also requires external asparagine for optimal in vitro growth.

AS-A is dispensable for T. brucei infectivity in mice
To test whether AS-A is important for parasite infection in a disease model, two groups of mice (n = 4) were inoculated with the parental cell line, and other two groups with RNAi cells. Two mice groups were fed with water containing doxycycline to induce down-regulation of TbAS-A, while the remaining mice were kept as non-induced controls. Within six days of inoculation, all mice from the different groups developed high levels of parasitemia (Fig. 8A), and all had to be euthanized after seven or eight days post-infection (Fig. 8B). The results confirm that the asparagine in mouse blood is sufficient to compensate for the <87% downregulation of AS-A (Fig. 8C, D).
To assess the contribution of blood L-asparagine in vivo, mice were treated with L-asparaginase [50]. L-asparaginase treatment did not affect growth of normal parasites in mice (Fig. 9A) and consequently did not extend animal survival (Fig. 9B). However Lasparaginase treatment in mice infected with TbAS-A RNAiinduced parasites caused a decrease in the parasitemia (Fig. 9D), thus leading to an increase of mice survival (Fig. 9E). Even so, the infection resulted in death. As happened in vitro, RNAi revertants appeared during the course of infection in asparaginase-treated, but not untreated, mice (Fig. 9F). Parasites extracts from wt infected mice were used as controls (Fig. 9C).

Discussion
In this study we demonstrated that trypanosomes AS-A use both ammonia or glutamine as nitrogen donors for the ATP dependent conversion of aspartate into asparagine. Such hybrid activity was only previously demonstrated for type B enzymes, which prefer glutamine to ammonia [15][16][17][18][19]. The small differences in K m of TbAS-A and TcAS-A for ammonia and glutamine (1.5 and 2 fold, respectively) are lower than the difference found in most AS-B enzymes, with the exception of the human enzyme, which has similar affinities for both [16][17][18][19][20]. Purified E. coli AS-A used only ammonia as the nitrogen source, and results from Klebsiella aerogenes also suggested that AS-A preferentially uses ammonia as substrate [5,21,22]. The conclusions for AS-A enzymes of these two Gramnegative organisms relied on both biochemical and genetic analysis, but given technical limitations at the time, and the fact that background enzyme activity was seen in the absence of both ammonia and glutamine, some re-examination in bacteria would be worthwhile. Moreover the overall K m values of trypanosomes AS-A for aspartate, are 6 up to 20 fold higher than the ones found in the literature for prokaryotic asparagine synthetase type A [5,21,22]. Trypanosoma AS-A structures were not yet been solved and our protein homology models are not completely enlightening, nevertheless we can speculate that such differences may result in the fact that parasite enzymes were expressed and purified as recombinant proteins in bacteria and not purified directly from trypanosomes extracts. As a consequence, differences in protein post-transcriptional processing and/or changes in protein conformation cannot be excluded.
Our results suggest that bloodstream-form parasites rely on two major sources of asparagine to ensure normal proliferation: uptake from the extracellular medium and biosynthesis by AS-A. Bloodstream form proliferation, either in vitro or in vivo, was only significantly affected when both asparagine sources were compromised. Also consistent with this idea, in the published RNAi screen, a very slight (possibly insignificant) growth disadvantage was seen in bloodstream forms depleted of AS-A [56]. In the same way, our in vitro results are corroborated with previous studies, as mammalian cells with low expression of AS are similarly susceptible to asparagine depletion [57][58][59], and asparaginase isolated from E. coli and Erwinia carotovora act as potent antileukemic agents [60].
In the trypanosomes genome there is a second open reading frame (Tb927.3.4060) coding for a putative AS domain. However this is apparently not a classical AS, despite the presence of a good Pfam AS domain (PF0073) at the C-terminus. A BLASTp search using the T. brucei sequence revealed a variety of proteins of unknown function that aligned not only across the AS domain, but also in the N-terminal region, which contains N-terminal aminohydrolase domains. Best matches originate from extremely diverse eukaryotes including a plant, an alga, a member of the fungi and an amoeba. BLASTp against the Saccharomyces cerevisiae predicted proteome yielded YML096W, and the reciprocal BLASTp on the trypanosome genome indeed gave Tb927.3.4060 as best match. The function of YML096W is not known, and in a trypanosome RNAi screen no growth defect was seen for Tb927.3.4060 [56].
The capacity of trypanosomes to grow using asparagine from the extracellular environment, and the lack of growth defect when the levels of AS-A are reduced, show that only a combination therapy using both a TbAS-A inhibitor and an extracellular asparagine depletor (e.g. L-asparaginase) or an asparagine transport blocker could inhibit parasite growth. This is not appropriate for African sleeping sickness treatment. A combination that absolutely required simultaneous activities of two different drugs would be wide open to resistance development, and drug combination including an intravenously-introduced enzyme is likely to be both too expensive and logistically inappropriate for treatment of African trypanosomiasis. Moreover, L-asparaginase treatment in cancer results in serious adverse events [61][62][63]. We therefore conclude that AS-A is not a good candidate as a sleeping sickness drug target. Its role in Trypanosoma cruzi, however, remains to be established. Figure S1 RNAi vectors used to generate RNAi-mediated TbAS-A downregulation. (A) pHD1144 vector for stemloop cloning (pSP72 vector with a stuffer fragment); (B) pHD1145 inducible polymerase I vector for insertion of ready-made stemloops (pHD677 vector without a T7 promoter and with an inducible EP1 promoter and hygromycin resistance cassette, insertion into ribosomal spacer). (TIF) Figure S2 Validation of antibodies against TbAS-A. Immunofluorescence analysis of T. brucei wt or a representative AS RNAi clone grown in the presence or absence of tetracycline. RNAi induced and uninduced cells were grown for 48 h, then fixed and probed with rat polyclonal anti-TbAS-A (A) or rabbit polyclonal anti-TbAS-A (B) antibody and co-stained with DAPI. Bars, 5 mm. Quantification of TbAS-A fluorescence levels in induced cells (AS RNAi tet(+), n = 30) and uninduced cells (AS RNAi tet(2), n = 30), using the rat and the rabbit polyclonal anti-TbAS-A antibodies (C). Data representative of two independent experiments using two different clones. ImageJ software (version 1.43u) was used for fluorescence quantification. p value was calculated by Student's t test (*** p#0.001 and ** p#0.01). (TIF) Figure S3 TbAS-A cellular localization in T. brucei bloodstream forms. Immunofluorescence analysis by confocal microscopy of TbAS-A (red) in bloodstream forms. Aldolase (glycosome marker), GRASP (golgi marker), BiP (endoplasmic reticulum marker) and MitoTracker (labels mitochondria) are in green. DAPI locate nuclear and kinetoplast mitochondrial DNA (blue). Bars, 5 mm. Images are maximal Z-projections of 50 contiguous stacks separated by 0.1 mm.