AUG-proximal nucleotides regulate protein synthesis in Leishmania tropica


*E-mail; Tel. (+1) 706 542 3355; Fax (+1) 706 542 4271.


Gene expression in the Leishmania is controlled post-transcriptionally, and is likely to be impacted by both 5′ and 3′ untranslated regions (UTRs). We have investigated the effects of trinucleotides in the AUG-proximal region (APR) (i.e. positions −3 to −1 upstream of an AUG) on two reporter genes in the context of an endogenous intergenic region of Leishmania tropica. The effects of APRs on protein expression were determined in stable transfectants in vivo. Three APRs, namely, C−3C−2C−1, ACC and GCC, yielded robust translation, whereas GTA produced low amounts of proteins. A purine at −3 of an APR was not crucial for efficient translation. Steady-state level of reporter mRNA did not correlate directly with the amount of protein detected. Polysome analysis revealed that APRs modulate translation, at least in part, by influencing mRNA association with ribosomes. An analysis of genomic UTRs in L. major showed that (i) the consensus APR is N−3N−2C−1 (where N = any nucleotide), and (ii) the most frequently used APRs include ACA, ACC, ATC, GCC, GCG, GTC and CAC, some of which were translation enhancers in our experimental studies.


Gene expression in trypanosomatids is generally controlled at the post-transcriptional level because most promoters are active by default (reviewed in Clayton, 2002; Beverley, 2003). Genes are transcribed into polycistronic premRNAs which are processed into monocistronic mRNAs (reviewed in Vanhamme and Pays, 1995; 1998; McDonagh et al., 2000; Clayton, 2002; Martinez-Calvillo et al., 2003; Worthey et al., 2003). Protein abundance frequently does not parallel mRNA levels (Gale et al., 1994; Tomas and Kelly, 1996; Saas et al., 2000). Consequently, premRNA processing, mRNA stability, and translatability contribute immensely to regulation of gene expression in trypanosomatids (e.g. Leishmania species). Post-transcriptional control of gene expression may be influenced by 5′ or 3′ untranslated regions (UTRs) (reviewed in Mignone et al., 2002).

The life cycle of Leishmania species involves differentiation of an extracellular promastigote stage in sand flies to an intracellular amastigote form within a vertebrate cell. In trypanosomatids, 3′ UTRs are important for controlling gene expression during differentiation (reviewed in Clayton, 2002). However, little is known about sequences that regulate translatability of mature mRNAs in the absence of differentiation, although this aspect of gene control is likely to contribute significantly to Leishmania responses to (and communication with) its human and fly hosts. 5′ UTRs are a logical control point for post-transcriptional regulation of gene expression (reviewed in de Moor and Richter, 2001; Keene and Tenenbaum, 2002), because ribosomes start synthesis of protein from initiation codons (AUGs) located at the 5′ end of mRNA-coding sequences.

AUG-proximal regions (APRs) (i.e. sequences in the −3 to −1 positions upstream of initiation codons) influence translation to variable degrees in vertebrates and in the yeast Saccharomyces cerevisiae (Kozak, 1984; Yun et al., 1996). However, in Leishmania no studies have been reported on this region of 5′ UTRs. To test the effects of 5′ UTRs on translation in Leishmania, seven 12-nucleotide sequences (12-mers) (Teilhet et al., 1998) were positioned just upstream of the initiation codon (AUG) of two reporter-coding sequences; namely, luciferase (LUC) (de Wet et al., 1987) and glycosylphosphatidylinositol-specific phospholipase C (GPI-PLC) (Hereld et al., 1988; Carrington et al., 1989; Mensa-Wilmot et al., 1995). The 12-mers differed only in the −3 to −1 positions, and were studied in the context of a 350-nucleotide dihydrofolate reductase/thymidylate synthase (DHFR/TS) intergenic region of Leishmania (Kapler and Beverley, 1989; LeBowitz et al., 1993).

APRs produced large differences in translation of reporter mRNAs, and probably modulated mRNA–ribosome interactions. These observations are the first report that APRs from a trypanosomatid 5′ UTR regulate translation in vivo in the absence of differentiation cues. Relevance of these data for post-transcriptional control of gene expression in Leishmania is discussed.


Effects of APRs on reporter protein synthesis

To examine the effect of AUG-proximal sequences on protein production in Leishmania, seven 12-mers (Fig. 1B) were tested in the context of a 350-nucleotide UTR from a DHFR/TS intergenic region (LeBowitz et al., 1993). Only the −3 to −1 sequences, defined as the APR (Fig. 1A), differed in the constructs, which contained either a GPI-PLC- or LUC-coding region as the reporter gene (Fig. 1C and D). In preliminary studies, we observed that −3A−2C−1C (i.e. ACC) yielded high amounts of protein, whereas GTA resulted in low amounts of protein. To determine whether single nucleotide changes in APRs could convert ‘effective’ APRs into ‘ineffective’ ones and vice versa, 14 APR constructs were studied. Seven APRs are presented in this paper because (i) they spanned the range of protein expression that we observed, and (ii) the data set is representative of general trends discerned.

Figure 1.

AUG-proximal region and expression plasmids for Leishmania.
A. The 5′ UTR is defined as the region between the splice leader acceptor site and the translation initiation codon (AUG). The APR is the sequence at −3, −2 and −1. SL, spliced leader.
B. Sequences near the translation initiation codon in the 5′ UTR. The −12 to −1 regions of the 5′ UTR contain three sequences: (i) a polypurine hexamer (bold), (ii) an Escherichia coli lz spacer (underlined), and (iii) nucleotides of the APR at −3, −2 and −1 (bold italics). 5′ UTRs are named according to the spacer and APR. For example, Tb-GTA refers to a Trypanosoma brucei GPI-PLC 5′ UTR with GTA at −3
to −1.
C. The −12 to −1 sequences described above were placed upstream of a GPI-PLC-coding region, and cloned into a BamHI site of pX63NEO.
D. Plasmid pX63NEO-LUC is identical to pX63NEO-GPI-PLC except that the GPI-PLC-coding region has been replaced with LUC. DHFR/TS intergenic regions [upstream and downstream sequences (US and DS)] from Leishmania are required for processing of NEO, LUC and GPI-PLC premRNAs.

As lz-GTA resulted in the lowest expression level, this was considered a ‘base line’ to which the performance of other APRs was compared. Further, all but one APR used in this study contained the lz sequence at the −4 to −12 (Fig. 1B) their effects on reporter protein synthesis could be compared with lz-GTA. The expression plasmid pX63NEO is maintained extrachromosomally in Leishmania (LeBowitz et al., 1993). Stable cell transfectants were used in enzyme assays for GPI-PLC polypeptide (GPI-PLCp) or LUC polypeptide (LUCp).

Amounts of GPI-PLCp enzyme produced were influenced by the APR (Fig. 2A). For both Tb-GTA and lz-GTA, which contained an identical APR but differed in the −4 to −12 nucleotides (Fig. 1B), the level of reporter protein was equivalent, suggesting that the −3 to −1 nucleotides (i.e. the APR) exerted a dominant effect on 5′ UTR regulation of protein synthesis in Leishmania.

Figure 2.

Effect of AUG-proximal sequences on GPI-PLC protein expression.
A. GPI-PLC enzyme activity assay. Stably transfected Leishmania (108 cells) were lysed hypotonically, centrifuged, and the pellet extracted with detergent. The detergent extract was centrifuged, and the supernatant was used in a GPI-PLC assay with [3H]myristate-labelled membrane form variant surface glycoprotein ([3H]mfVSG) as substrate. Total protein in the lysate was determined with a bicinchoninic acid assay (Pierce). Enzyme assays were performed in triplicate. Data from a representative experiment are shown.
B. APRs affect translation of GPI-PLC. Leishmania stably transfected with GPI-PLC expression plasmids were labelled with [35S]Met/Cys. Cells were hypotonically lysed and centrifuged, and GPI-PLC protein was immunoabsorbed using a monoclonal antibody and protein-A sepharose beads. Immunoabsorbed proteins were eluted and separated by SDS-PAGE (14%). The relative amount of [35S]Met/Cys-labelled GPI-PLC was quantified by phosphorimaging. Lanes: (1) pX63NEO; (2) lz-ATC; (3) lz-CCA; (4) lz-CCC; (5) lz-GTA.
C. Single cell assays of APR-driven GPI-PLC protein expression in Leishmania. Stably transfected Leishmania expressing GPI-PLC with different APRs were fixed and permeabilized. Anti-GPI-PLC antibody was added, followed by Alexa Fluor 488-conjugated anti-rabbit IgG for detection of GPI-PLC (green). DAPI was used to stain nuclei and kinetoplast DNA (blue). A representative field of cells is shown for the APRs tested.
The colour version of this figure is available as supplementary material Fig. S1.

Five APRs, namely, CCC, GCC, CTA, ACC and ATC, produced at least fourfold more protein than GTA. For example, CCC gave rise to 10-fold more protein than GTA. Similarly, CTA generated sixfold more GPI-PLCp than GTA (Fig. 2A).

To verify that the level of enzyme activity reflected the amount of GPI-PLCp produced, Leishmania were metabolically labelled with [35S]Met/Cys, and [35S]Met/Cys-labelled GPI-PLCp was immunoprecipitated from cell lysates and quantified (Fig. 2B). The amount of GPI-PLCp produced with different APRs correlated with the observed level of enzyme activity (Fig. 2B). For example, CCC resulted in comparatively high amounts of [35S]Met/Cys-labelled GPI-PLCp (Fig. 2B) in agreement with the high enzyme activity noted previously (Fig. 2A). In contrast, GTA produced relatively low quantities of [35S]Met/Cys-labelled GPI-PLCp (Fig. 2B), consistent with the low enzyme activity observed when APR was placed upstream of the GPI-PLCp-coding sequence (Fig. 2A).

It was important to determine whether expression of the reporter enzyme was relatively uniform within the cell population. Single-cell assays were performed to monitor expression of GPI-PLCp by immunofluorescence. Three representative APRs were studied for this objective. The amount of GPI-PLCp detected with an anti-GPI-PLC antibody was generally uniform within each population of cells, and the relative intensity of fluorescence was consistent with the results of the enzyme assays (Fig. 2A and C). For example, CCC and CTA produced high levels of GPI-PLC protein while GTA led to a low level of GPI-PLC protein, as detected by the anti-GPI-PLC antibody. From these data we conclude that differences in protein expression observed in enzyme assays (Fig. 2A) are a general property of the cell population (Fig. 2C) rather than a minor group of overexpressors.

It was important to determine whether the effect of APRs on GPI-PLC translation could be generalized to other protein-coding sequences. To address this issue, a second reporter gene, firefly luciferase (LUC) (de Wet et al., 1987), was studied in stable transfectants of L. tropica. Enzyme activity of LUCp was affected greatly by different APRs (Fig. 3). GTA and ATC resulted in comparatively low LUCp translation, while CTA produced at least sixfold more LUCp than GTA. Astonishingly, three APRs (GCC, CCC, ACC) produced at least 2700-fold more protein than GTA. These results confirm the earlier conclusion (Fig. 2A) that APRs can influence protein expression in Leishmania. The dynamic range of protein expression observed with the two reporter genes was not identical. For GPI-PLCp, the difference between the most active (CCC) and base line APR (GTA) was 10-fold, whereas in the case of LUCp the highest performing APR (GCC) was 3700 times higher than the base line (GTA). The precise reason for the 100-fold difference in the response of the two reporters has not been studied. It is possible that the assay method for LUCp is more sensitive than the detection system for GPI-PLCp. Nevertheless, numeric differences in the effects of APRs for each reporter must be considered in terms of the observed range of protein expression for that reporter. That is, a fourfold difference in translation of LUCp is less significant than a fourfold difference in translation of GPI-PLCp.

Figure 3.

Effect of AUG-proximal sequences on luciferase enzyme activity. Stably transfected Leishmania (107 cells) were lysed with 0.2% Triton X-100, and the lysate was used in a luciferase assay with luciferin as substrate. Relative light units were measured and normalized for total protein. Enzyme assays were performed in triplicate. Data from a representative experiment are shown.

In most cases the APRs had similar effects on translation of GPI-PLC and LUC; ACC, CCC and GCC resulted in prolific translation of both GPI-PLCp and LUCp, whereas GTA yielded low amounts of both proteins (Figs 2A and 3). However, in one instance reporter dependence of APR effects was noted: ATC produced high GPI-PLCp but little LUCp. Such reporter dependence of 5′ UTR effects on translation have been noted in S. cerevisiae (Gallie et al., 1991).

Abundance of protein does not correlate with steady-state mRNA level

Differences in the amount of reporter protein synthesized (Figs 2 and 3) could, in theory, arise from changes in the amounts of mRNA. To test this hypothesis, steady-state levels of reporter transcripts were measured by Northern blot analysis (Fig. 4). In control experiments, no significant difference in ribosomal RNA (rRNA) amount was detected by ethidium bromide staining of gels in which total RNA from the cells had been separated, implying that similar amounts of RNA were used in the Northern analysis (Fig. 4).

Figure 4.

Effect of AUG-proximal sequences on steady-state levels of GPI-PLC and LUC mRNA.
A. Northern analysis of GPI-PLC transcripts. Total RNA was isolated from 108Leishmania with TriReagent (Sigma). Twenty micrograms of RNA was separated on a 1% formaldehyde agarose gel, and electroblotted to a BrightStar-Plus membrane (Ambion). The agarose gel was stained with ethidium bromide to detect rRNA as a loading control. [In Leishmania the large subunit (LSU) rRNA is fragmented into LSUα (1900 nucleotides), LSUβ (1600 nucleotides) and four smaller fragments. The small subunit rRNA is 2500 nucleotides long.] After UV cross-linking, the membrane was probed with a psoralen-labelled probe for GPI-PLC, and then probed for ribosomal protein S8 (rpS8), as an additional loading control. Probes were detected using BioDetect (Ambion), and then the membrane was exposed to BioMax MR film (Kodak). GPI-PLC and rpS8 mRNA signals were quantified directly from the membrane with a Genegnome Chemiluminescence Imager and Genetools software (Syngene). GPI-PLC mRNA signals were normalized by dividing by the corresponding mRNA signal for the loading control rpS8.
B. Northern analysis of luciferase transcripts. Northern analysis was performed as described above except that a probe for luciferase (LUC) was used. Detection of rRNA and rpS8 was performed as described earlier.

Quantities of reporter mRNA did not parallel protein levels. For example, with GPI-PLC as the reporter-coding sequence, GTA and CCA have similar level of mRNA (Fig. 4A), but expression of GPI-PLCp is fourfold higher with CCA as compared with GTA (Fig. 2A). Similarly, there was more steady-state mRNA for ATC than for CCC (Fig. 4A), but the amount of GPI-PLCp produced by CCC was significantly higher than that for ATC (Fig. 2A). Overall, mRNA varied less than twofold for the GPI-PLC transcripts (Fig. 4A), while the difference in GPI-PLCp was 10-fold (Fig. 2A). The discrepancy between the relative amount of protein synthesized and steady-state mRNA suggests that APRs affect gene expression post-transcriptionally.

Results similar to that obtained with GPI-PLC transcripts (Fig. 4A) were produced when LUC mRNAs were analysed (Fig. 4B). For example, mRNA for GTA and GCC was found at similar levels, but the LUCp produced with GCC was 3700-fold higher than that obtained with GTA. In the same vein, the amount of mRNA for CCC exceeded that for GCC (Fig. 4B), but more protein was obtained with GCC than CCC (Fig. 3). Overall the relative level of LUC mRNA varied twofold (Fig. 4B). Yet, the range of protein (LUCp) produced by different APRs differed 3700-fold (Fig. 3). These data support our earlier proposal that APRs influence the utilization of mRNA for protein synthesis.

Distribution of reporter mRNA between cytosol and ribosomes

During translation initiation in eukaryotes, a 43S ribosomal initiation complex binds to mRNA to form a 48S complex. A 60S large ribosome subunit associates with a 48S complex at the initiator AUG, to form the 80S ribosome (or monosome). Polysomes are produced when multiple ribosomes bind (concurrently) to one mRNA.

As neither GPI-PLC nor LUC mRNA levels correlated directly with the amount of protein produced (Figs 2–4), we hypothesized that APRs that led to synthesis of large amounts protein (e.g. CCC) caused (or stabilized) mRNA association with ribosomes. To test this concept, the distribution of reporter mRNA between cytosol and ribosomes in vivo was investigated. Three APRs that result in varying levels of GPI-PLCp expression were studied: CCC (very high), CTA (high) and GTA (low) (Fig. 2). Ribosomes from cells stably expressing the GPI-PLC constructs were separated on a 10–40% sucrose gradient (Brecht and Parsons, 1998), and RNA was extracted in order to determine whether APRs altered association of GPI-PLC mRNAs with ribosomes (Fig. 5C and D).

Figure 5.

Distribution of gp63 mRNA and large subunit rRNA (controls) and GPI-PLC on ribosomes. Leishmania stably expressing GPI-PLC were lysed, and a supernatant from a 14 000 g centrifugation was separated on a 10–40% sucrose gradient. Twelve 1 ml fractions were collected. RNA was extracted from the fractions, and 1/5 of the RNA was spotted onto a Bright-Star Plus membrane. After UV cross-linking, the membrane was probed with a psoralen-labelled probe as described in the legend to Fig. 4.
A. DNA probes were as follows: gp63 or large subunit rRNA (γ subunit). Representative blots are presented.
B. Quantification of gp63 mRNA and large subunit rRNA. Signals were quantified directly from the membrane with a Genegnome chemiluminescence imager and Genetools software (Syngene). The proportion of the RNA associated with cytosol, 40S ribosomes, 60S/80S ribosomes, and polysomes is presented. Representative graphs are shown. Distribution of GPI-PLC mRNA on ribosomes.
C. Separation of ribosomes and detection GPI-PLC mRNA with a GPI-PLC DNA probe were done as described above.
D. Quantification of GPI-PLC mRNA signal was performed as described above. The proportion of the GPI-PLC mRNA signal associated with cytosol, 40S ribosomes, 60S/80S ribosomes, and polysomes is shown. Representative graphs are presented.

Polysome analysis has not been used frequently to study Leishmania ribosomes. Therefore, we included experiments to: (i) determine the distribution of an endogenous mRNA, and (ii) to verify the location of ribosome subunits in the sucrose gradients. To simplify data analysis, mRNAs found in locations indicative of association with two or more ribosomes were grouped as the ‘polysome fraction’. Further, the peaks corresponding to the 80S monosome and the 60S large subunit were not easily resolved on the sucrose gradient. Consequently, mRNAs associated with these two forms of ribosomes were grouped together as the 60/80S fraction, for purposes of quantification.

Control RNAs were detected in expected positions of the sucrose gradient. The transcript for gp63, the most abundant protein in Leishmania promastigotes, was found mostly (72%) associated with polysomes (Fig. 5A and B). Large subunit rRNA was detected in both 60S/80S (32%) and polysome fractions (50%) (Fig. 5A and B).

For APRs that led to efficient translation of GPI-PLCp, i.e. CCC and CTA (Fig. 2A), most of the mRNA (54–61%) associated with polysomes (Fig. 5C and D). Conversely, for GTA, which produced low amounts of GPI-PLCp (Fig. 2A), only 29% of the mRNA was found on polysomes (Fig. 5C and D). The polysome profiles for the GPI-PLC mRNAs with CTA and GTA are strikingly different although the APRs differ only in the −3 nucleotide (Fig. 5C and D); mRNA for the translational-enhancing CTA is shifted to polysomes in comparison with GTA. Thus, the efficiency of protein synthesis directed by CCC, CTA and GTA correlates with the relative ability of the corresponding mRNA to associate stably with polysomes in vivo.

Genomic analysis of APRs in L. major 5′ UTRs

We were interested in finding out whether the APRs that we studied experimentally were present upstream of Leishmania protein-coding sequences. To achieve this objective, 5′ UTRs of 200 protein-coding sequences of L. major were obtained from the genome sequencing project ( and analysed. In this study, we only used coding sequences with homologues in other organisms, so that the correct initiator methionine could be identified by alignment of the protein-coding sequences. Such selective use of the genome sequence data is preferred because over 70% of the protein-coding sequences in Leishmania are ‘hypothetical’ (Myler and Stuart, 2000; Myler et al., 2000), and have never been studied experimentally. For such genes it would be difficult to correctly assign the initiator AUGs that in turn define APRs. (A compilation of the genes whose UTRs were analysed is presented in Table S1).

We attempted to derive a consensus sequence for the L. major APRs by aligning the 200 5′ UTRs and applying the ‘50/75 Consensus Rule’ (Cavener, 1987; Cavener and Ray, 1991). The frequency of a sole consensus nucleotide is: (i) greater than 50%, and (ii) greater than twice the frequency of the next most frequent nucleotide (Cavener, 1987). However, two nucleotides may share a consensus if their combined frequency is greater than 75% but neither meets the criterion of a sole consensus. Using these criteria, a degenerate consensus APR of protein-coding genes L. major is N−3N−2C−1 (where N = any nucleotide) (Fig. 6A).

Figure 6.

A. Consensus APR from Leishmania 5′ UTRs. The frequency of each nucleotide (A, C, G or T) at positions −3, −2 and −1
was analysed for the 200 5′ UTRs.
B. Frequency of trinucleotides in the APR of Leishmania 5′ UTRs. Sequences at −3 to −1 were obtained from the Sanger Centre Leishmania major genome project ( The 200 genes are found on chromosomes 1, 3, 4, 5, 19 and 23. 5′ UTRs were selected only if the initiator AUG could be predicted by protein sequence similarity to a polypeptide from another organism, to ensure that the translation initiation codon was unambiguous.

We investigated whether the APRs that we studied experimentally were present in the genome of L. major. Further, we wanted to determine which APRs were used most frequently by L. major and also to assess whether abundant proteins in the parasite employed those APRs. Towards this goal, we analysed trinucleotides in the −3 to −1 positions of 200 L. major protein-coding sequences (Table S1). The maximum number of times a trinucleotide occurred was 17 (Fig. 6B). Based on this observation, we defined ‘frequently used trinucleotides’ as those whose APRs whole frequencies were at least half of the most popular trinucleotide (GCC) (i.e. eight times or more). Using this approach, the frequently used trinucleotides in the L. major−3 to −1 regions are: ACA, ACC, ATC, CAC, GCC, GCG and GTC (Fig. 6B), which includes some APRs that we studied experimentally (i.e. ACC, ATC, GCC). Also of interest, some abundant proteins in L. major possess APRs that we studied experimentally. Examples include GCC for α-tubulin, ribosomal protein P1, and gp63, and ACC for ribosomal protein S7 and translation initiation factor 3 (Table S1).


AUG-proximal regions of 5′ UTRs affect protein synthesis in Leishmania

Gene expression in Leishmania spp. is unusual, as compared with other eukaryotes, for several reasons. First, the process is regulated largely at post-transcriptional stages, because most promoters are constitutively active (reviewed in Shapira et al., 2001; Clayton, 2002; Beverley, 2003). Second, transcription units are polycistronic (Lee et al., 1988; Button et al., 1989; Worthey et al., 2003). Third, transcripts of both sense and antisense strands of chromosomal DNAs are detectable (Liniger et al., 2001; Martinez-Calvillo et al., 2003). Fourth, genes encoded on the same polycistronic premRNA can produce variable amounts of protein (reviewed in Vanhamme and Pays, 1995; 1998; McDonagh et al., 2000; Clayton, 2002; Martinez-Calvillo et al., 2003; Worthey et al., 2003). Finally, protein abundance frequently does not track steady-state mRNA levels (Gale et al., 1994; Tomas and Kelly, 1996; Saas et al., 2000). Given this background, it is important to examine the factors that contribute to post-transcriptional utilization of mRNAs in trypanosomatids. As protein synthesis initiates from the 5′ regions of mRNAs, it is logical to find out whether sequences upstream of translation initiation codons (AUGs) contribute to protein production in Leishmania species.

The first three nucleotides upstream of AUGs can have significant effects on translation in vertebrates (reviewed in Kozak, 1999). Less dramatic effects of these APRs occur in the yeast S. cerevisiae (Yun et al., 1996) (reviewed in McCarthy, 1998). In trypanosomatids, where post-transcriptional processes dominate gene expression, 5′ UTRs are highly likely to be a major regulatory point, in contrast to other eukaryotes where isolated, but highly significant, instances of 5′ UTR regulation have been described (reviewed in de Moor and Richter, 2001; Keene and Tenenbaum, 2002).

AUG-proximal sequences influenced Leishmania protein synthesis significantly, in the context of a DHFR/TS 5′ UTR (Kapler and Beverley, 1989). APRs tested with a trypanosomal reporter (GPI-PLC) produced a 10-fold difference in translation (Fig. 2A), and with LUC, even larger differences were obtained (Fig. 3). This striking effect of APRs on protein synthesis in the absence of corresponding changes in steady-state mRNA (Fig. 4) emphasizes the importance of post-transcriptional mechanisms for gene regulation in Leishmania (Aly et al., 1994; ter Kuile, 1999; Shapira et al., 2001; Clayton, 2002).

Response of the Leishmania translation apparatus to APRs was generally independent of the reporter protein-coding sequences (compare Figs 2A and 3). Three APRs (i.e. CCC, GCC and ACC) that stimulated translation of both reporters (Figs 2 and 3) may be regarded as ‘translation enhancers’ that can be used for efficient synthesis of heterologous and native proteins in Leishmania. Consistent with this suggestion, the APRs were found upstream of abundant endogenous proteins. For example, GCC is the APR for α-tubulin, ribosomal protein P1 and also gp63 (Table 1). Interestingly, the effect of one APR (i.e. ATC) on translation was modulated by the coding sequence (Fig. 2). Such effects of UTRs on reporter gene expression, although not widely known, have been documented in Chinese hamster ovary cells (Gallie et al., 1991).

Purine at −3 of AUG is not critical for efficient translation in Leishmania

In vertebrates cells, the consensus sequence for the APR is A/GNC (N = any nucleotide) as determined by application of the application of the ‘50/75 Consensus Rule’ (Kozak, 1987; Cavener and Ray, 1991; Peri and Pandey, 2001). [The Kozak consensus A/GCC is found upstream of 30% of vertebrate initiation codons (Cavener and Ray, 1991; Schneider et al., 2001)]. Detailed analysis of variants of the sequence revealed that the purines A or G at the −3 position of an AUG fostered optimal translation of mRNAs (Kozak, 1997). This fundamental concept about vertebrate translation may not be applicable to the Leishmania, which are ancient unicellular eukaryotes. In L. tropica, some APRs with A or G at −3 (e.g. ACC and GCC) enhanced protein synthesis (Figs 2 and 3). However, in contrast to vertebrate systems, C at −3 does not inhibit translation, as illustrated in the following examples. First, CCC and CTA, which do not contain a purine at the −3 position, yielded amounts of LUCp that were 2900- and sixfold respectively, higher than GTA, which contains G at −3 (Fig. 3). Second, conversion of an A or G at −3 of ACC (or GCC) to C (forming CCC) does not inhibit translation of either LUCp or GPI-PLCp (Figs 2 and 3). Therefore, some of features vertebrate translation may not be applicable to Leishmania protein synthesis. These observations are consistent with the frequency with which nucleotides occupy the −3 position in vertebrates and L. major. In the former, a purine is present at the −3 position with a frequency of 90% (Cavener and Ray, 1991; Schneider et al., 2001) whereas in Leishmania a purine occupies −3 less than 70% of the time (Fig. 6A).

AUG-proximal sequences affect the association of mRNAs with ribosomes

Due to the potent effects of APRs on translation (Figs 2 and 3), we investigated mechanisms by which the sequences might affect translation. Steady-state levels of reporter mRNA did not correlate directly with abundance of protein (Figs 2–4). Therefore, we checked whether APRs influenced mRNA interactions with ribosomes. Two APRs that resulted in high GPI-PLCp expression, CCC and CTA, caused the reporter transcripts to associate primarily with polysomes (Fig. 5). In contrast, with GTA, which yielded low GPI-PLCp synthesis, the mRNA was predominantly cytosolic (Fig. 5), implying that the APR was ineffective at stabilizing mRNA–ribosome interactions. We propose that stable mRNA–ribosome association contributes to translation competence of GPI-PLC reporter mRNAs in L. tropica. This observation is consistent with the report that a spliced leader that is present at the 5′-terminus of nuclear-encoded mRNAs facilitates association of RNA with ribosomes (Zeiner et al., 2003).

In summary, AUG-proximal sequences modulate translation in L. tropica, at least in part, by affecting mRNA association with ribosomes. We propose that APRs may bind trans-acting factor(s) that facilitate protein synthesis. For example, the RNA binding protein La autoantigen recognizes a GCAC motif near the start codon in a hepatitis C virus mRNA (Pudi et al., 2003). Single nucleotide changes that decreased La binding to the RNA also reduced translation of the viral mRNA (Pudi et al., 2004). Using these observations as a guide, we suggest that factors that recognize APRs in Leishmania mediate the effects of the sequences on protein synthesis.

Experimental procedures


Sources of reagents were as follows: Restriction enzymes (New England Biolabs, Beverly, MA); Amplitaq DNA polymerase (Perkin Elmer, Boston, MA); T4 DNA ligase, RNAse inhibitor, and RQ1 DNAse (Promega, Madison, WI); fetal bovine serum (Hyclone, Logan, UT); medium 199, RPMI without Met/Cys (Gibco BRL, Grand Island, NY); antibiotic/antimycotic (Mediatech, Herndon, VA); protease inhibitors and Titan One Tube RT-PCR reagents (Roche, Indianapolis, IN); bicinchoninic acid assay kit and luciferin (Pierce, Rockford, IL); cycloheximide and NP-40 (Calbiochem, San Diego, CA); [35S] Promix Met/Cys (Amersham Biosciences, Piscataway, GH); Alexa Fluor 488-conjugated anti-rabbit IgG antibody, 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) (Molecular Probes, Carlsbad, CA), BrightStar psoralen-biotin, Northern Max, BrightStar Plus membrane, UltraHyb Buffer, and RNAse-free TE buffer (Ambion, Austin, TX); Biomax MR film (Kodak, Rochester, NY); Qiagen Plasmid Midi Kit (Qiagen, Valencia, CA); G418, isopropanol, phenol, chloroform, and phenol/chloroform/isoamyl alcohol (Fisher, Pittsburgh, PA). All other reagents were obtained from Sigma (St Louis, MO).

Culture of Leishmania

Leishmania tropica strain HOM/IQ/73/LCR-L32 (Chaudhuri et al., 1989) was cultured in Medium 199 (M199) at 25°C (LeBowitz, 1994).

Plasmid construction

GPI-PLC plasmids were constructed by isolating BamHI fragments containing APRs (Fig. 1B) and GPI-PLC gene from pBluescript II constructs, and ligated into BamHI digested pX63NEO (Fig. 1C).

To construct luciferase plasmids, the LUC gene was PCR amplified (Innis et al., 1988) from plasmid p220S (a gift from Dr Mary Wilson, University of Iowa) (Ramamoorthy et al., 1996) using forward primers that introduced the different APRs, such as 5′ BamHI-lz-ACC-LUC (CGCGGATCCTTAACACAGGAGGinline imageACCatggaagacgccaaaaacata) encoding a BamHI site (bold), purine hexamer (underlined), lz spacer (outlined), APR (italicized), and luciferase amino acids 1–6 (lower case). All amplifications were performed using the reverse primer 3′ LUC-BamHI (CGCGGATCCttacaatttggactttccgcc) including a BamHI site (bold) and nucleotides complementary to luciferase sequence encoding amino acids 545–551 (lower case). PCR products were digested with BamHI and ligated into a BamHI site in pX63NEO (LeBowitz et al., 1993) (Fig. 1D).

DNA transfection

Leishmania growing at a density of 8.0 × 106−1.0 ×  107 cells ml−1 were harvested, washed in phosphate buffered saline (PBS), 136 mM NaCl, 27 mM KCl, 5.3 mM Na2HPO4, 1.7 mM KH2PO4, pH 7.4, and resuspended in electroporation buffer (21 mM Hepes, pH 7.5, 137 mM NaCl, 5 mM KCl, 0.7 mM Na2PO4, 6 mM glucose) and electroporated (475 V, 800 μF, and 13 Ω; one pulse, 2 mm gap cuvette, BTX Electro Cell Manipulator 600) using 10–50 μg of DNA purified with a QIAGEN Midi plasmid kit. Cells were incubated for 8–12 h at 25°C before addition of G418 (30 μg ml−1, final concentration) for selection of stable transfectants (LeBowitz, 1994). Cells were adapted to growth in M199 containing 100 μg ml−1 G418, and used without cloning.

GPI-PLC enzyme assay

Stably transfected Leishmania (1 × 108 cells) at a density of 8.0 × 106 cells ml−1 were washed in PBS and then lysed in 1 ml of hypotonic buffer (10 mM Tris-HCl, pH 8.0, 2 mM EDTA) supplemented with a protease inhibitor cocktail (PIC) composed of 0.4 U aprotinin, 2.1 μM leupeptin, 0.1 mM TLCK (Armah and Mensa-Wilmot, 2000).The cells were incubated on ice for 20 min and then centrifuged for 20 min (14 000 g, 4°C). The pellet was resuspended in 500 μl of GPI-PLC assay buffer (50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 1.0% NP-40) and incubated on ice for 20 min. After centrifugation for 20 min (14 000 g, 4°C), 5 μl of the supernatant was used to cleave 2 μg of [3H]myristate-labelled membrane form variant surface glycoprotein ([3H]mfVSG) for 20 min (Mensa-Wilmot et al., 1994). Protein content of the lysates was determined with a bicinchoninic acid assay.

Luciferase enzyme assay

Stably transfected Leishmania (1 × 107 cells) harvested at a density of 8.0 × 106 cells ml−1 were washed with 1 ml of PBS and stored at −20°C. Frozen cells were lysed in 200 μl of lysis buffer (100 mM potassium phosphate, pH 7.8, 0.2% Triton X-100 and 2 μg ml−1 leupeptin) (Ramamoorthy et al., 1996). The lysate was centrifuged for 10 min (8000 g, 4°C), and 5 μl of the supernatant was added to 75 μl of luciferase assay buffer (25 mM glycylglycine, pH 7.8, 5 mM ATP, pH 7.5, 15 mM MgSO4) in a 96 well plate. The reaction was initiated by injection of 50 μl of 0.5 mM luciferin substrate in a luciferase assay buffer. Peak light emission was measured for 1 s after a 2 s delay and recorded by a FLUOstar OPTIMA luminometer (BMG Labtechnologies).

[35S]Cys/Met-labelling and immunoprecipitation of GPI-PLCp

Leishmania (2.5 × 108 cells) cultured to a density of 8.0 × 106 cells ml−1 were washed twice with 10 ml of RPMI without Met/Cys. Cells were incubated in 20 ml of RPMI without Met/Cys for 60 min at 25°C before addition of 250 μCi of ProMix [35S]Met/Cys for 60 min at 25°C. Cycloheximide (100 μg) was added, the cells were washed in PBS, and lysed in hypotonic buffer supplemented with PIC (see GPI-PLC enzyme assay above), and a membrane pellet obtained (as described above). The pellet was solubilized in immunoprecipitation dilution buffer (1% Triton X-100, 200 mM NaCl, 60 mM Tris-HCl, pH 7.5, 6 mM EDTA) supplemented with PIC (Armah and Mensa-Wilmot, 1999). Monoclonal antibody 2A6-6 (Mensa-Wilmot et al., 1995) (1:1000 dilution) was added to the solubilized pellet and incubated overnight at 4°C with continuous inversion. A protein A-sepharose slurry was then added and the incubation continued for 2 h. The beads were washed as follows: once with PBS plus PIC, then twice with PBS plus PIC with 0.2% NP40, and finally, once with PBS plus PIC. Protein adsorbed to sepharose beads was eluted by addition of 20 μl of SDS-PAGE sample buffer [25% (v/v) glycerol, 5% (w/v) SDS, 5% (v/v) β-mercaptoethanol, 0.05% (v/v) bromophenol blue] and heating at 95°C for 1 min. Immunoprecipitated proteins were separated by SDS-PAGE. The [35S]-labelled GPI-PLC was detected with a phosphorimager and quantified with Quantity One Software (Bio-Rad).

Immunofluorescence microscopy

To detect GPI-PLCp in Leishmania, cells (1 × 107) growing at a density of 8 × 106 ml−1 were harvested, fixed in paraformaldehyde (2%) for 8 min at 4°C, washed with PBS, and permeabilized with Triton X-100 [0.25% (w/v)] (Zheng et al., 2004). Anti-GPI-PLC rabbit antibody RC300 (1:2000) was applied for 1 h at room temperature. Cells were washed sequentially with PBS, high-salt buffer (PBS with 500 mM NaCl) and PBS again before incubation with Alexa Fluor 488-conjugated anti-rabbit IgG antibody (1:2000) for 1 h at room temperature. DNA in nuclei and kinetoplasts (mitochondrial DNA) was detected with DAPI. Cells were viewed with a fluorescence microscope (Leica DMIRBE). Images were captured with an interline chip cooled CCD camera (Orca 9545: Hamamatsu) and processed with Openlab 3.1.2 software (Improvision).

RNA isolation

Leishmania (1 × 108 cells) harvested at a density of 8.0 × 106 cells ml−1 were washed with 10 ml of PBS before lysing with 1 ml of TriReagent. Two hundred microlitres of chloroform was added, and the sample centrifuged for 15 min (12 000 g, 4°C). RNA was precipitated from the upper aqueous phase with isopropanol (0.5 ml), pelleted (14 000 g, 4°C, 15 min), washed with 75% ethanol and resuspended in 20 μl of RNase-free TE. Typically, 200–400 μg of RNA was obtained using this protocol.

Probes for RNA analysis

DNA to be used as probes were obtained by PCR (Innis et al., 1988) using pairs of primers specific to different genes, namely; GPI-PLC (Hereld et al., 1988; Carrington et al., 1989), LUC (de Wet et al., 1987), gp63 (Button and McMaster, 1988), large subunit rRNA (gamma subunit) (Martinez-Calvillo et al., 2001), and small subunit 18S rRNA (Martinez-Calvillo et al., 2001). For GPI-PLC, primers KCR 21 (TTTGGTGGTGTAAAGTGGTCACCGCAGTCATGGATGAGTGACACG) and KCR 8 (TATGTGGATCCTTATTAGAAATCGAGAATGACAAATTCGTTGGC) were used to amplify the sequence encoding GPI-PLC amino acids 1–146 from the plasmid pX63NEO-GPI-PLC. The sequence encoding amino acids 313–452 of luciferase was amplified from the plasmid p220S using FLUC (GCTTCTGGGGGCGCACCTCTTTC) and RLUC (AGCGGGGGCCACCTGATATCCTTTGTA) primers. Additional probe DNA was obtained by amplification of Leishmania genomic DNA using the following primer pairs: (i) for gp63: Fgp63 (GACGCGATGCAGGCACGCGTGCGGCAGTCG) and Rgp63 (GAAGTCGGTGTTGCTGAAGCCCTCGGTGAT), (ii) for ribosomal protein S8: FLtrpS8 (CTGCATAAGCGCAAGATCACC) and RLtrpS8 (CTTGTTCACGTCGTACTTGTC), (iii) for large subunit rRNA: FgamrLtRNA (CGCAAGCGCAAATGAAATACCACCACTCG) and RgamltRNA (TGCCGCCCCAGCCAAACTCCCCATCT), (iv) for small subunit rRNA: F18SrLt-RNA (CACGCGAAAGCTTTGAGGTTACAGTCT) and R18SrLtRNA (ACCCGCCGATGAGTTTGCTATTCTATGG). PCR products were isolated from the agarose gels following standard protocols. One microgram of PCR product (in 10 μl of RNase-free TE) was labelled with 0.25 μg psoralen-biotin (BrightStar, Ambion) by UV irradiation (365 nm for 45 min). Unreacted psoralen-biotin was extracted twice with water-saturated n-butanol (200 μl). Probes were stored at −80°C. Immediately before use, probes (in 25 μl RNase-free TE) were denatured at 100°C for 10 min and quick-chilled in an ice-water bath.

Northern blotting of GPI-PLC and LUC transcripts

Twenty micrograms of total RNA were denatured for 15 min at 65°C and quick-chilled in an ice-water bath. RNA was electrophoresed on a formaldehyde-agarose (1%) gel (Northern Max Formaldehyde Gel solutions, Ambion), stained with ethidium bromide and vacuum-blotted to a BrightStar Plus membrane (Ambion) using a Posiblot Transfer Apparatus (Stratagene). After cross-linking with UV (1 min at 1.2 kJ, Stratalinker, 1800, Stratagene) the membrane was pre-hybridized in 10 ml of UltraHyb buffer (Ambion) at 42°C for 2 h. Thereafter the membrane was hybridized with approximately 250–300 ng of psoralen-labelled probe for 16–18 h at 42°C. The probe was detected using a BioDetect chemiluminescence system (Ambion), and the membrane exposed to BioMax MR film (Kodak). GPI-PLC, LUC and rpS8 mRNAs were quantified directly from the membrane using a Genegnome Chemiluminescence Imager and Genetools software (Syngene).

Polysome analysis

Leishmania (5 × 108 cells) were harvested at a density of 8.0 × 106 cells ml−1 and washed sequentially with (i) M199 (5 ml, containing 100 μg ml−1 cycloheximide) and (ii) PBS (10 ml, containing 100 μg ml−1 cycloheximide). Cells were resuspended in 750 μl of Buffer A (10 mM Tris-HCl, pH 7.4, 300 mM KCl, 10 mM MgCl2, 1 mM DTT, 100 μg ml−1 cycloheximide, 1 mM PMSF, 8.5 μg ml−1 aprotinin, 50 μg ml−1 leupeptin, 1 μM pepstatin, 50 μg ml−1 TLCK, 10 μM E-64) supplemented with 1 μl of RNAse (40 U) (Promega) (Brecht and Parsons, 1998). The cells were incubated on ice for 3 min before addition of 125 μl of cell lysis buffer (Buffer A with 1.2% Triton X-100, 0.2 M sucrose, 60 U ml−1 RNAse) (Brecht and Parsons, 1998). Parasites were further fragmented with a 7 ml Dounce homogenizer (15 strokes), transferred to a 1.5 ml tube and centrifuged for 2 min (14 000 g, 4°C). The supernatant was added to a 1.5 ml tube containing 100 μl heparin (10 μg ml−1) and 1 μl RNAse (40 U), and layered onto a 10–40% sucrose gradient (12 ml in Buffer A) with a Pasteur pipette. The gradient was centrifuged for 2 h and 15 min (36 000 rpm, 4°C) in an SW41 rotor Beckman, and harvested (12 1 ml fractions) with a gradient collector [Foxy Jr (ISCO)]. OD254 readings of samples were recorded with a UA-6 detector (ISCO).

RNA extraction from sucrose gradients and dot blot analysis

RNA was obtained from 500 μl of a sucrose fraction as described (Mangus and Jacobson, 1999), with minor modifications. RNA pellets were dissolved in 1 μl of RNase-free TE for each 100 μl of sucrose extracted. For dot blotting, RNA was denatured (65°C for 15 min), the samples were quick-chilled in an ice-water bath, and spotted onto a BrightStar Plus membrane (Ambion). Cross-linking of RNA, hybridization, and detection of probes were performed as described in the section on ‘Northern blotting’, except that 60–75 ng of probe was used.

Genomic analysis of L. major APRs

DNA sequences were obtained from the Sanger Centre Leishmania major genome project ( The 200 genes used in this study are found on chromosomes 1, 3, 4, 5, 19 and 23. To avoid uncertainty about assignment of translation initiation codons, only 5′ UTRs of proteins with homologues in other organisms were included in the data set.

To obtain an APR consensus sequence from the 200 5′ UTRs, the frequency that each nucleotide occupied positions −3, −2 and −1 was calculated. A consensus nucleotide was assigned if its frequency was at least 50% and at least twice the frequency of the next popular nucleotide (Cavener, 1987).


This study was supported by NIH Grant AI58301 to KM-W. We thank Dr Zhifeng Zheng who performed the microscopy experiments.