Aminoacylation of Plasmodium falciparum tRNAAsn and Insights in the Synthesis of Asparagine Repeats*

Background: Proteins from the malaria parasite Plasmodium contain extensive asparagine repeats. Results: Plasmodium tRNAAsn concentration and in vitro asparaginylation catalytic parameters were analyzed to explain the frequency of asparagine repeats. Conclusion: The requirement for high levels of asparaginylated tRNAs hinders efficient protein translation. Significance: Limiting the availability of asparaginylated tRNAs represents a plausible regulatory mechanism for the production of properly folded proteins. Genome sequencing revealed an extreme AT-rich genome and a profusion of asparagine repeats associated with low complexity regions (LCRs) in proteins of the malarial parasite Plasmodium falciparum. Despite their abundance, the function of these LCRs remains unclear. Because they occur in almost all families of plasmodial proteins, the occurrence of LCRs cannot be associated with any specific metabolic pathway; yet their accumulation must have given selective advantages to the parasite. Translation of these asparagine-rich LCRs demands extraordinarily high amounts of asparaginylated tRNAAsn. However, unlike other organisms, Plasmodium codon bias is not correlated to tRNA gene copy number. Here, we studied tRNAAsn accumulation as well as the catalytic capacities of the asparaginyl-tRNA synthetase of the parasite in vitro. We observed that asparaginylation in this parasite can be considered standard, which is expected to limit the availability of asparaginylated tRNAAsn in the cell and, in turn, slow down the ribosomal translation rate when decoding asparagine repeats. This observation strengthens our earlier hypothesis considering that asparagine rich sequences act as “tRNA sponges” and help cotranslational folding of parasite proteins. However, it also raises many questions about the mechanistic aspects of the synthesis of asparagine repeats and about their implications in the global control of protein expression throughout Plasmodium life cycle.

they may be excised by some unknown mechanism (6). Yet, experimental evidence suggesting that LCRs play a functional role is rare. In addition to these hypotheses, we proposed that LCRs in general and asparagine-rich LCRs in particular modulate the rate of translating ribosomes and thus participate in cotranslational folding of plasmodial proteins (14). This proposed process would be driven by limiting access to asparaginylated transfer RNA Asn (tRNA) in the parasite during translation of LCRs.
It is generally accepted that variations in ribosome decoding rates are a consequence of the abundance of tRNAs in the cell; translation is optimal at codons decoded by highly abundant tRNAs and limited at codons read by rare tRNAs, as the ribosome pauses until the cognate tRNA becomes available. Still, it has been shown that tRNA abundance is not the only determinant of elongation rates. Interactions between the mRNA and the 16S ribosomal RNA as well as the presence of positively charged amino acids in the nascent peptide decelerate the ribosomes along transcripts in bacteria and yeast, respectively (15,16). Among prokaryotic and eukaryotic genomes, genes encoding a given tRNA are usually present in multiple copies. Plasmodia are exceptions to this rule: the plasmodial nuclear genomes display 45 tRNA genes, with only one gene copy per tRNA isoacceptor (1). Thus, this minimal set of tRNA genes leads to the appealing assessment that the product of the unique tRNA Asn gene would decode Ͼ10% of the Plasmodium proteome. Translation specificity and efficiency depend on the accuracy of tRNA aminoacylation catalyzed by their cognate aminoacyl-tRNA synthetases (17) and on the availability of charged tRNAs, respectively. In the present study, P. falciparum cytosolic asparaginyl-tRNA synthetase (AsnRS) and tRNA Asn were cloned and their kinetic parameters were determined in vitro to estimate asparaginylation efficiency in this parasite. Moreover, the relative concentrations of each molecular partners involved in P. falciparum asparaginylation were compared with other plasmodial aminoacylation systems to deduce the relative accumulation of asparaginyl-tRNA Asn in Plasmodium and eventually its impact on the rate of synthesis of asparagine-rich LCRs.
Database Screening for AsnRS Sequences-The P. falciparum genome encodes two copies of AsnRS, a cytoplasmic version (PlasmoDB ID PF3D7_0211800) and an apicoplastic one (Plas-moDB ID PF3D7_0509600) (14). The cytoplasmic P. falciparum AsnRS gene was identified by blasting the sequence of yeast cytoplasmic AsnRS (NP_011883.1) against the PlasmoDB database. Sequence alignments were computed with Tcoffee software.
RNA Seq Data of aaRS mRNA Expression in Sexual and Asexual Stages-The study was performed in the laboratory of Xinzhuan Su (18), and data were retrieved from PlasmoDB for each P. falciparum cytosolic aminoacyl-tRNA synthetase gene.
Proteomic Data of P. falciparum Life Cycle Stages-Data were retrieved from Florens et al. (19). We considered proteins (i) expressed specifically in a single developmental stage and (ii) with sequences covered by at least three peptides . Thus, we  used 27, 18, 31, and 66 protein sequences, corresponding to  50,027, 9629, 24,152, and 55,394 amino acids, identified in the  sporozoite, merozoite, trophozoite, and gametocyte stages, respectively.
Cloning and Expression of Recombinant P. falciparum and H. sapiens AsnRSs-Genes coding for both the P. falciparum and human AsnRSs were amplifed by PCR from either P. falciparum (provided by H. Vial, Montpellier, France) or human liver cDNA libraries, respectively. Primers for PCR were designed according to the sequences of the AsnRS genes. In the case of the parasite AsnRS, the PCR reaction was performed with a customized dNTP mixture adapted to the P. falciparum genome composition (80% AT-rich): 240 M dATP, 240 M dTTP, 80 M dCTP, 80 M dGTP, and elongation was performed at 65°C. Amplified DNA fragments were entirely sequenced and cloned into pQE30 (Qiagen, Courtaboeuf, France) to yield N-terminal His 6 -tagged recombinant proteins. Escherichia coli Top10 strains, separately expressing each construct, were grown at 37°C in Luria broth with 2% glucose and 50 g/ml ampicillin until they reached an absorbance of 2.0 (1-cm path length). The medium was then substituted with the same medium lacking glucose and AsnRS expression was induced for 3 h at 37°C with 1 mM isopropyl-␤-D-thiogalactopyranoside. Recombinant proteins were purified on nickelnitrilotriacetic acid resin according to the manufacturer's instructions (Qiagen). Fractions containing AsnRS were pooled and dialyzed overnight at 4°C against 50 mM KH 2 PO 4 / K 2 HPO 4 , pH 7.4, 10 mM ␤-mercaptoethanol, 150 mM KCl, and 50% glycerol. AsnRS proteins were quantified by Bradford assay (Bio-Rad) and stored at Ϫ20°C. Preparations yielded one band on SDS-polyacrylamide gel and were estimated to be at least 95% pure.
Cloning and Expression of Recombinant P. falciparum AspRS and TyrRS-Purification of recombinant P. falciparum AspRS was performed as described in Ref. 20. The gene coding for P. falciparum TyrRS was amplified by PCR from P. falciparum cDNA, cloned into pQE30, and purified as described for P. falciparum AsnRS.
Dynamic Light Scattering-P. falciparum and H. sapiens AsnRS samples (20 M) were incubated in HEPES-KOH, pH 7.0, 50 mM NaCl, 10% glycerol and were analyzed with a Zetasizer Nano S (Malvern, Orsay, France). Five intensity measurements were recorded at 20°C, and the data were processed with the manufacturer's software, assuming that the enzyme particles were spherical. This method measures distributions of particle diffusion coefficients (D) that were transformed via the Einstein equation (D ϭ kT/6p 0 R h ) into hydrodynamic radii (R h ) and thus apparent particle sizes. Corrections for solvent refractive index and viscosity were applied, and the contribution of the solvent components was subtracted.
Preparation of L-[ 14 C]Asparagine-This procedure was adapted from Roy and collaborators (21). The reaction mixture (200 l), containing 50 mM HEPES-NaOH, pH 7.2, 10 mM MgCl 2 ,2 mM NH 4 Cl, 2 mM KF, 240 M L-[ 14 C]Asp (300 cpm/ pmol), and 10 mM ATP, was incubated at 70°C for 2 h in the presence of 2 M Pyrococcus abyssi asparagine synthetase A. Under these conditions, 95% of the initial aspartic acid was amidated into asparagine (as analyzed by thin layer chromatography). The reaction was stopped by phenol extraction, the L-[ 14 C]Asn was further purified on an Illustra NAP-5 column (GE Healthcare Life Sciences) and concentrated by cryo-desiccation. The pellet was recovered in H 2 O to reach a final concentration of 200 M.
Preparation of tRNA Samples-Transfer RNA transcripts were produced in vitro using either the classical method for P. falciparum tRNA Asn (22) or the "transzyme" system for H. sapiens tRNA Asn (23). Briefly, synthetic genes made of overlapping oligonucleotides reconstructed the T7 RNA polymerase promoter, the optional hammerhead ribozyme, the tRNA, and a BstNI site matching the CCA 3Ј-end of the tRNA sequence. When transcribed, the hammerhead ribozyme selfcleaved and released the mature tRNA that was further purified on a denaturing polyacrylamide gel. Before any assay, tRNA transcripts were heated at 65°C for 2 min in water and refolded at 37°C in the presence of MgCl 2 (10 mM).
Total RNA from P. falciparum 3D7 strain parasites were cultured at 37°C in 1% hematocrit as described previously (24), and cultures were harvested with 12-15% parasitemia. Infected erythrocytes were washed with incomplete growth medium and lysed with 0.15% saponin, and the parasites were used immediately for total RNA isolation by the Tri®-reagent BD (Sigma) extraction method according to the manufacturer's instructions.
ATP/PP i Exchange and tRNA Aminoacylation Assays-ATP/ PP i exchange assays were performed at 37°C in 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1.5 mM [ 32 P]PP i (3 cpm/pmol), 2 mM KF, and 250 nM AsnRS. The concentrations of either L-Asn or ATP varied from 0.05 to 1.5 mM. The synthesized [ 32 P]ATP was quantified as described by Campanacci and collaborators (25).
Aminoacylation assays were conducted in 50 mM HEPES-NaOH, pH 7.2, 10 mM MgCl 2 , 2 mM ATP, 30 mM KCl, and appropriate amounts of tRNA Asn and AsnRSs (diluted in 100 mM HEPES-NaOH, pH 7.4, 1 mM DTT, and 10% glycerol for the plasmodial enzyme and the same buffer containing 5 mg/ml bovine serum albumin for the human enzyme). The mixture was incubated for 15 min at 37°C and the reaction was initiated by adding 20 versus 10 M L-[ 14 C]asparagine for the plasmodial and the human enzymes, respectively. Apparent kinetic parameters were determined from Lineweaver-Burk plots. Data were expressed as averages of at least three independent experiments.
Aminoacylation plateaus of Plasmodium total RNA were performed in the presence of 0.5 M of the relevant aminoacyl-tRNA synthetases (aaRS; P. falciparum recombinant AspRS, TyrRS, and AsnRS). Values were measured after 60-min incubations at 37°C.

In Silico Identification of Plasmodial AsnRSs
Six plasmodial genomes were screened for cytosolic AsnRS genes. Complete AsnRS genes were found in primate-infecting (P. falciparum, P. vivax, and P. knowlesi) and rodent-infecting (P. yoelii, P. berghei, and P. chabaudi) plasmodia ( Fig.  1). The gene coding for the cytosolic P. falciparum AsnRS (PF3D7_0211800) is 1833 base pairs, is located on chromosome 2, and codes for a 610-amino acid protein. AsnRS is classified with both the aspartyl-and lysyl-tRNA synthetases as a class IIb aaRSs (26). Plasmodia AsnRSs were compared with homologues from H. sapiens, the vertebrate host of P. falciparum, and from Brugia malayi, for comparison with the crystal structure determined by Crepin and collaborators (27). Structurebased sequence alignments show characteristic eukaryotic AsnRS features, in particular in the C-terminal catalytic domain that contains the three class II aaRS motifs. All AsnRSs exhibit N-terminal appendices, which are predicted to fold as helical modules. The plasmodial modules are shorter (75 to 82 amino acids), and their sequences diverge from the N-terminal appendices of B. malayi AsnRS (110 amino acids) (28) and H. sapiens AsnRS (103 amino acids) (29). Yet, they all contain a helical lysine-rich motif (KMKKLK) that is similar to the RNAbinding motif identified in eukaryotic AspRSs and LysRSs (30 -32). Lastly, the N-terminal module of plasmodia AsnRSs does not contain the amino acid sequence involved in the cytokine activity of B. malayi AsnRS (28).
All plasmodial AsnRSs are characterized by the presence of an insertion in the first half of the catalytic domain, between motifs 1 and 2 ( Fig. 1), which likely does not disturb the active site architecture. This insertion corresponds to an LCR and has not been previously identified in AsnRSs of other phylogenetic origins. The insertion site of this LCR is strictly conserved among plasmodial AsnRSs, but the size and sequence vary from one species to another. In P. falciparum, P. berghei, P. chabaudi, and P. yoelii, this insertion contains several asparagine stretches.
To test the contribution of both the N-terminal extension and the LCR, P. falciparum AsnRS variants missing these Decoding Asparagine in Plasmodium DECEMBER 20, 2013 • VOLUME 288 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 36363 domains were designed and constructed. However, they were not tested because recombinant mutant proteins were not soluble (data not shown).

Physicochemical Analysis of P. falciparum and H. sapiens AsnRSs by Dynamic Light Scattering
Dynamic light scattering spectroscopy is a convenient and non-invasive method for protein sizing (33) and has been applied to measure the aggregation states of aaRSs (e.g. Ref. 20). Measured diffusion coefficients of P. falciparum and H. sapiens AsnRS samples showed that both recombinant enzymes are dispersed at the concentrations used (20 M monomeric protein). Only a small portion of the human AsnRS (13%) was aggregated (particle diameter, 110 nm; data not shown). Analysis of dynamic light scattering data allowed the determination of the size of AsnRS particles. Distributions of scattered intensities indicated that the average diameter of the P. falciparum and the H. sapiens AsnRSs were 13.8 and 12.5 nm, respectively, which correspond to spherical particles of 295 and 210 kDa (versus 142 and 127 kDa, theoretically). Because the crystallographic structure of the B. malayi AsnRS is more elongated than spherical (27), these data indicate that, as expected, both AsnRSs organize as dimers.

Catalytic Constants for the ATP/PP i Exchange Activity
The apparent K m and k cat values for asparagine activation were determined under identical conditions for the P. falciparum and H. sapiens AsnRSs ( Table 1). The kinetic parameters for both enzyme orthologues are similar to the values determined for the AsnRSs from E. coli and CHO cells (34,35). However, the K m for asparagine is 11-fold higher for the human AsnRS (460 M) than for the other enzymes (40 M for P. falciparum and E. coli AsnRS and 60 M for CHO AsnRS).

Kinetic Analyses of tRNA Aminoacylation by P. falciparum and H. sapiens AsnRSs
Both recombinant P. falciparum and H. sapiens AsnRSs were tested for tRNA charging under the same conditions (Fig.  2). They aminoacylated their cognate tRNA Asn transcripts at plateau levels of 65 and 35%, respectively (Fig. 2C). The apparent K m for their cognate tRNA transcripts were the same, but the velocity (apparent k cat ) was 13-fold lower for P. falciparum AsnRS than for H. sapiens AsnRS, significantly decreasing (13fold) the catalytic efficiency of the parasite AsnRS (Table 2).
Despite the similarity between the P. falciparum and H. sapiens tRNA Asn sequences (Fig. 2, A and B), cross-species aminoacylation was not efficient (Fig. 2C). The human enzyme did not recognize the plasmodial tRNA Asn transcript, and the extent of asparaginylation of the human tRNA Asn transcript by the P. falciparum AsnRS was reduced 8-fold compared with its cognate tRNA Asn . The only remarkable feature distinguishing these two tRNAs is the presence of a UU mismatch at base pairs 4 -69 in the acceptor arm of the P. falciparum tRNA Asn ; it is conserved in all plasmodial tRNA Asn sequences (Fig. 2B), which is an indication that the acceptor stem might play a role in this speciesspecific barrier. However, a human tRNA Asn mutant, where the UG base pair was replaced by UU did not show any improvement in its aminoacylation capacity with P. falciparum AsnRS (data not shown).

Expression of Asparaginylation Molecular Partners and Comparison with Other Aminoacylation Systems in P. falciparum
tRNA Concentrations-We investigated the expression levels of tRNA Asn , tRNA Tyr , tRNA Asp , and tRNA Lys in non-synchronized in vitro cultures of P. falciparum. Transfer RNA Tyr and tRNA Asp are unique isoacceptors, which are each encoded by a single tRNA gene similar to tRNA Asn , but they decode fewer codons (4.4 and 6.4%, respectively) than tRNA Asn (11.2%) (Codon Usage Database). On the contrary, frequencies of asparagine and lysine codons are comparable (11.2 and 10.4%, respectively), but tRNA Lys is encoded by two tRNA genes (tRNA Lys AAA and tRNA Lys AAG ). Northern blots and aminoacylation detection showed that two of the four selected tRNAs, tRNA Asn and tRNA Tyr , are globally expressed at the same level, independent of their use in translation (Fig. 3). Logically, the two tRNA Lys genes result in twice as much tRNA Lys compared with either tRNA Asn or tRNA Tyr . Surprisingly, tRNA Asp appeared ϳ3-fold more concentrated than the others, despite only one copy of the tRNA Asp gene.
aaRS Expression-RNA-Seq coverages from different P. falciparum development stages (18) were used to estimate the  relative expression levels of AsnRS compared with other aaRSs in P. falciparum (Fig. 4A). Concentrations of mRNAs varied from one aaRS to another. For example, trophozoites had 300fold less ValRS mRNA than TyrRS mRNA. They also differed from one developmental stage to another: the mRNA encoding AsnRS was 4-fold higher in trophozoites than it was in gametocytes. This mRNA belongs to a group of three aaRS mRNAs (AsnRS, SerRS, and TyrRS) that are, on average, more concen-trated than the remaining 18 aaRS mRNAs (Fig. 4A). Thus, the AsnRS mRNA has a propensity to accumulate noticeably; however, it is not the most abundant mRNA at any stage growth. Amino Acid Availability-Based on genome analysis, Plasmodium possesses biosynthetic pathways for Asn, Glu, Asp, Gly, Pro, and Gln (36). However, its growth depends on a supply of a complete set of exogenous amino acids. Thus, during the blood stage, Plasmodium development relies mostly on hemo- A, cloverleaf structure of tRNA Asn transcripts: the motifs involved in tRNA folding are present in both the plasmodial (green) and the human (blue) tRNA Asn sequences. The few differences are spread across their sequences and do not involve residues strategic to the structure and are presumed to be harmless for folding and aminoacylation. B, sequence alignments of human and plasmodial tRNA Asn sequences: the human genome codes for 35 tRNA Asn genes partitioned into two isoaccepting families either with an AUU anticodon (two genes coding for a unique isodecoder tRNA sequence) or a GUU anticodon (GtRNAdb, Genomic tRNA Database). The GUU isoacceptor tRNA family contains 14 different isodecoders encoded by the 33 other genes. Among them, the sequence of the major tRNA Asn isodecoder (asterisk) is represented 11 times. This sequence was first chosen to produce the human tRNA Asn transcript in vitro but was a poor aminoacylation substrate. Another sequence was thus chosen, tRNA Asn -89, whose transcript is a better substrate for the human AsnRS. The sequences of plasmodial tRNA Asn are highly conserved, except for a U-G base pair that becomes the C-G in the T-stem of P. chabaudi and P. knowlesi tRNA Asn . The U 4 ⅐U 69 mismatch conserved in the acceptor stem is highlighted in gray. C, aminoacylation plateaus of tRNA Asn transcripts. Reactions were performed in the presence of 80 pmol of transcript and 25 pmol pf of P. falciparum AsnRS or 3.5 pmol of H. sapiens AsnRS. The color code is the same as described in A. globin catabolism and circulating amino acids (37). Based on the amino acid concentrations measured in human blood, the amount of asparagine available for translation in Plasmodium corresponds to 3.5% of all amino acids (Fig. 4B).

Asparagine Content of P. falciparum Stage-specifically Expressed Proteins
Data from stage-specific proteomes were retrieved from Ref. 19. They relate to four distinct stages of the parasite: sporozoites (the infectious form injected by the mosquito), merozoites (the invasive stage of the erythrocytes), trophozoites (the form multiplying in erythrocytes), and gametocytes (the sexual stage). For each stage, amino acid contents and amino acid repeats were determined. Global proteome asparagine content is not strictly conserved among the different stages; for example, there is 1.8-fold more asparagine in sporozoite proteins than in trophozoites proteins (Fig. 5A). This observation is strengthened when looking at the frequency of asparagine repeats (Fig. 5B): asparagine repeats are more abundant in sporozoites (2.7-and 4.2-fold) and in gametocytes (2-and 3-fold) than in merozoites and trophozoites, respectively, and longer. Glutamate and valine contents are also stage-specific, but unlike asparagine, these amino acids do not cluster in long repeats (data not shown).

AsnRS Activity
aaRSs are diverse in sequence, size, and structure but perform the same function: they catalyze specific tRNA aminoacylation (17). In turn, the efficiency and accuracy of these reactions govern the efficiency and accuracy of protein synthesis. Crystal structures of Thermus thermophilus and B. malayi AsnRSs have been reported (27,38). Yet, biochemical and enzymatic studies are rare. The only exception relates to the T. thermophilus asparaginylation system (39).
In the present work, we studied P. falciparum AsnRS and its host H. sapiens AsnRSs to directly compare both enzymes. Neither the plasmodial nor the human enzymes are optimal in charging their cognate tRNA Asn transcripts in vitro, as compared with the asparaginylation kinetics measured for the T. thermophilus system with native tRNA Asn . The kinetic parameters determined for the prokaryotic system are 2000and 150-fold more efficient than the parasite and the human systems, respectively. Similarly, the K m of the CHO AsnRS for its cognate tRNA Asn (35) is ϳ30-fold lower (0.06 M) than the apparent K m values determined in the present study (2 M). These differences could be explained by the use of in vitro transcripts in this study, which are deprived of post-transcriptional modifications. Occasionally, these modifications have been shown to be involved in tRNA aminoacylation (i.e. Ref. 40). In particular, the absence of a queuosine at position 34 in both the Plasmodium and the human tRNA Asn could explain the reduced K m and k cat values. The presence of two genes coding for putative queuosine tRNA ribosyltransferases in the P. falciparum genome (PF3D7_1242200, probably targeted to the apicoplast, and PF3D7_0717400, the putative cytosolic enzyme) support the existence of this particular post-transcriptional modification in P. falciparum tRNA Asn , as it is the case in all organisms except yeast (41).

tRNA Asn Availability
The genomes of most organisms display multiple copies of tRNA genes; the number of copies for each tRNA isoacceptor diverges from species to species (42). It is accepted that (i) tRNA gene content determines the relative amount of each tRNA in the cell, (ii) the most abundant codons are decoded by the most abundant tRNAs, and (iii) translation efficiency is governed by the relative abundance of each tRNA (reviewed in Ref. 43). In this context, the fact that P. falciparum but also all other plasmodia genomes encodes only one gene copy per tRNA isoacceptor is confusing. Assuming that each gene is transcribed to similar extents, one can calculate the theoretical availability of Plasmodium tRNA species compared with their codon usage in P. falciparum protein synthesis (Fig. 4C). Strikingly, unlike other organisms, there is no correlation between the number of tRNA genes and its codon usage in this parasite (or the related amino acid frequency). For example, the ratios of tRNA gene copy number to amino acid frequency varies at most by 4-fold among all H. sapiens tRNA/amino acid pairs. In contrast, in P. falciparum, these ratios vary over a 20-fold range, with the lowest value assigned to tRNA Asn and asparagine (Fig. 4C). Thus, tRNA Asn is the most limiting tRNA in the translation of P. falciparum. This hypothesis was confirmed by Northern blots and aminoacylation tests on total RNA isolated from in vitro cultures of different P. falciparum blood stages (Fig. 3). In these experiments, tRNA Asn did not accumulate more than any of the three other tRNAs tested (tRNA Asp , tRNA Lys , and tRNA Tyr ). Because the codons that are decoded by tRNA Asn are highly represented in the Plasmodium genome, translation efficiency from that particular tRNA is expected to be compromised. This scenario might be partially compensated for by an increased local concentration of this specific tRNA near the translation site of asparagine stretches, as has been suggested for other organisms (44).

Asn-tRNA Asn Availability
The regulating action of tRNA Asn depends above all on the proportion of asparagine-charged tRNA Asn available for translation. Based on the RNA seq data, the mRNA encoding AsnRS is highly expressed in the cell, suggesting that, in the absence of specific post-transcriptional inhibitory mechanism, AsnRS would substantially accumulate in the parasite. Moreover, P. falciparum AsnRS shows an optimal affinity for asparagine (Table 1), indicating that even if the aspara- gine concentration is getting low, it should not affect significantly the rate of the first step of the asparaginylation reaction. These observations favor efficient asparaginylation of the tRNA Asn pool in Plasmodium.

Open Questions
Slippery Sequences?-The relative scarcity of tRNA Asn in plasmodia could sequentially decrease the rate of Asn-tRNA Asn selection at the ribosome A-site, trigger ribosomal frameshifting, and finally lead to a reading frameshift when translating repeated asparagine codons. Moreover, one can wonder whether the post-transcriptionally added queuosine modification in the tRNA Asn anticodon would also influence the fidelity of asparagine insertion in asparagine repeats. It was previously shown that queuosine does not influence frameshifting in bacteria (45) or when decoding a specific sequence of the infectious bronchitis virus (46). However, this hypothesis has not been tested on the long AU-rich sequences that characterize Plasmodium mRNAs or in the presence of limiting concentrations of Asn-tRNA Asn .
Asparagine Repeats and Protein Aggregation?-In P. falciparum, the presence of asparagine repeats could be deleterious because it leads to the expression of a large number of prionlike domains (ϳ25% of the proteome) (47), which in turn make the corresponding proteins suitable for aggregation. Recently, it was shown that P. falciparum Hsp110c, with other chaperones, plays an essential role in overcoming the aggregation tendencies of asparagine-rich proteins (48). Furthermore, we propose that accumulation of such prion-like sequences could be permitted through the presence of charged amino acids, lysine, glutamate, and aspartate, which also cover Ͼ25% of the proteome. The presence of so many charged amino acids could, in theory, improve the production of more soluble proteins (49,50).
Potential Regulatory Activities?-According to the models discussed above, the tRNA pool should remain the same not only in different cell types but also throughout their life cycle. As a matter of fact, recent measurements in yeast and human tissues show that the abundance of different tRNAs deviates from the copy number of the corresponding genes (51,52). As a consequence, variations in tRNA levels during the parasite life cycle would offer an opportunity to regulate and induce wide variations in translation efficiency and could modulate the expression of stage-specific proteins depending on their content in asparagine repeats. Our analysis of P. falciparum stagespecific proteomes (19) supports this idea by revealing an increase in asparagine repeats in proteins expressed exclusively in sporozoites and gametocytes compared with erythrocytic stages. These differences suggest that translation in sporozoites and to a lesser extent in gametocytes requires more asparaginylated-tRNA Asn . When looking at trophozoite and gametocyte stages, there is no apparent correlation between the concentra- Stage-specific proteome data were retrieved from the study by Florens and collaborators (19). They correspond to four distinct stages of the parasite: sporozoites, merozoites, trophozoites, and gametocytes. A, asparagine frequencies were compared in stage specific proteins; B, asparagine repeats were sorted according to their length: five to nine consecutive residues (light gray), 10 to 19 residues (dark gray), and Ͼ20 residues (black). tions of mRNA for AsnRS (Fig. 4A) and the presence of long single amino acid repeats (Fig. 5). However, one should be aware that mRNA concentrations do not consider post-transcriptional regulatory mechanisms, which control the final cellular concentrations of these enzymes, at least in other organisms (53,54).
tRNA Asn Limitation and Cotranslational Folding of Proteins?-Our favorite hypothesis is that LCRs and asparagine repeats are intrinsic chaperones. Indeed, the low availability of Asn-tRNA Asn should decrease the rate of translating ribosomes when decoding asparagine-rich LCRs and, in consequence, govern cotranslational folding of the proteins of the parasite: ribosome pauses would give enough time to the newly synthesized N-terminal domain to fold correctly, independently from the protein C-terminal sequence not yet translated. This scenario is consistent with our previous observation that LCRs exist in boundaries between independent folding modules in the aaRSs themselves (14). LCRs are inserted in almost all Plasmodium proteins. Their own translation spatially separates the synthesis of defined portions in these proteins (Fig. 6), allowing sequential folding of individual protein domains with translation pauses or slow downs in between. Furthermore, the presence of asparagine repeats in LCRs would increase local Asn-tRNA Asn demand and further postpone translating ribosomes.

Impact on Protein Synthesis versus a Specific Function for
Asparagine-rich LCRs-Single amino acid repeats have been found for all 20 amino acids in all three domains of life. Yet, plasmodia are distinct because of the number of amino acid repeats and because they are found in all classes of proteins. The AT content of the P. falciparum, P. yoelii, P. chabaudi, and P. berghei genomes definitely plays a part in the accumulation of asparagine repeats in their proteomes; it does not explain why these codons tend to cluster together in LCRs (6). However, the predilection for alanine repeats in P. knowlesi and P. vivax suggests that the presence of amino acids stretches is important, but the nature of the incorporated amino acid is not. This species-specific usage of amino acid repeats is a further indication that repeats (and LCRs) might not have acquired functions per se but that their presence is important for the proper function of the rest of the protein. We propose that differential accumulation of asparagine repeats could direct the correct folding and/or the stage specific expression of nascent Plasmodium proteins by controlling the rate of the translating ribosome. This could replace what exists in most organisms, where the tightness of secondary and tertiary structures in mRNAs controls ribosome binding and decoding rate (i.e. Refs. 55 and 56). One may indeed wonder whether AU-rich Plasmodium mRNAs can adopt such structures and use the same posttranscriptional regulatory strategies.