Genes are made up of DNA and contain the instructions to make molecules called proteins. This information is stored as a genetic code consisting of four bases: adenine (A), cytosine (C), guanine (G) and thymine (T). The order of these bases and their different combinations serves as a blueprint for making thousands of different proteins and to assemble living cells.

Converting the information in the genes into proteins requires several steps. First, the code from the DNA needs to be transcribed into RNA and then processed to make messenger RNA, or mRNA for short, which in turn is translated into proteins. Cells decode mRNAs by reading the bases as groups of three, also called codons. Most codons specify an amino acid – the building blocks of proteins – but certain codons also mark the start and end point of a protein. This ensures that the mRNA is read in the correct ‘frame’ and the desired proteins are made.

Any cell contains thousands of different proteins and each protein has its own unique level. The mechanisms used to set the number of each different type of protein can operate at every point in the process. In many organisms, the number of times a gene is transcribed to make an mRNA, underpins differences in protein levels. Trypanosomes, for example, are parasites that cause a range of devastating diseases in humans and livestock. They lack the ability to set individual mRNA levels by regulating how often the gene is transcribed. This suggests that the expression of thousands of mRNAs is regulated by a common control mechanism later in the process ending in protein synthesis. However, until now, it was unclear what these mechanisms are.

Most amino acids are encoded by more than one codon. The different codons for one amino acid are not equivalent and using a different codon can lead the mRNA to yield more or less protein. Evolution acts on these differences between codons, and the ‘codon choice’ in any one mRNA represents the outcome of natural selection. Now, Nascimento, Kelly et al. found that codon choice directs both the levels of mRNAs and the level of translation.

For the experiments, a new metric that enables a prediction of the level of expression for each mRNA was created. This metric (known as the ‘gene expression codon adaptation index’ or geCAI for short) could relate the codon choice to mRNA levels. For example, mRNAs with a low index score had shorter half-lives, i.e., how long that mRNA remained in the cell before being broken down.

Nascimento, Kelly et al. confirmed this by measuring mRNA levels in specific genes tagged with distinguishable markers and revealed that the codon choice indeed dictated the rate at which an mRNA would be broken down. A separate study by Jeacock, Faria and Horn looked more closely at how codon choice contributes to the control of the copy number of proteins. However, genes and mRNAs involved in development could deviate from the levels predicted by the geCAI metric, which suggests that other mechanisms may be in place to control the stability these mRNAs.

The importance of codon choice in setting mRNA levels has now been demonstrated in several organisms, including yeast and trypanosomes, which suggests that this process is more widespread than previously realised.

In trypanosomatids, RNA polymerase II dependent transcription of protein coding genes is polycistronic (Van der Ploeg, 1986; Worthey et al., 2003). Individual monocistronic mRNAs are produced after co-transcriptional processing by trans-splicing a 39 nucleotide, capped, exon to the 5’ end and linked endonucleolytic cleavage and polyadenylation of the upstream mRNA (Sutton and Boothroyd, 1986; Murphy et al., 1986; Ullu et al., 1993; Matthews et al., 1994). This mechanism of mRNA transcription has co-evolved with the structure of the genome; protein coding genes occur in tandemly arrayed clusters containing up to hundreds of genes in one transcription unit (Ivens et al., 2005; Berriman et al., 2005). The transcription start sites for these gene clusters have been identified (Kolev et al., 2010) and are marked by histone variants (Siegel et al., 2009) and although the nature of promoters remains incompletely defined, a GT-rich promoter element has been identified that can both initiate transcription and cause histone variant deposition (Wedel et al., 2017). There is remarkable synteny between the genomes of trypanosomatid species that diverged hundreds of millions of years ago (El-Sayed et al., 2005) suggesting a functional organisation of genes within polycistronic transcription units. In most cases, this is not due to the presence of an equivalent of prokaryotic operons but appears to be linked to the distance from the transcription start site. For example, genes up-regulated during heat shock tend to be distal to the transcription start site so that they continue to be transcribed by elongating RNA polymerase II for >60 min after transcription initiation has stopped as a consequence of the stress (Kelly et al., 2012). As yet, there is no convincing evidence for selective regulation of RNA polymerase II transcription of individual genes or transcription units.

In the absence of regulation based on selective transcription of individual genes, the disparate levels of individual mRNAs within a trypanosome cell is dependent solely on post-transcriptional processes. In trypanosomatids, some very abundant RNA polymerase II transcribed mRNAs are encoded by tandemly arrayed multigene families so one pass of the RNA polymerase transcribes multiple copies of the genes in series producing increased numbers of pre-mRNAs. For example, α- and β-tubulins are encoded by a tandem array of 19 copies of each gene (Ersfeld et al., 1998). However, the majority of genes are single copy and are expressed in the cell at different levels (Kolev et al., 2010) so mechanisms for regulation of mRNA abundance must exist and these must act post-transcriptionally.

It is probable that the abundance of each mRNA is regulated at several levels during its production/degradation cycle: the rate of maturation against the decay rate of the pre-mRNA in the nucleus; the rate of export to the cytoplasm against the half-life of the mature mRNA in the nucleus; the half-life of the mature mRNA in the cytoplasm (Haanstra et al., 2008). Most work aimed at understanding determinants of mRNA levels has investigated cis- and trans-acting factors that modulate the cytoplasmic half-life of the minority of mRNAs that alter in response to a developmental or an environmental trigger. In Trypanosoma brucei there are a series of developmental forms that are adapted to different host niches and the variation in mRNA expression levels between the mammalian bloodstream form (BSF) and the insect midgut procyclic form (PCF) has been used to assay for factors necessary for a decreased half-life in one developmental form. Cis-elements have been identified in the 3’UTRs that differentially affect stability and/or rate of translation of mRNAs in specific developmental stages; the best characterised being those in procyclin mRNAs (Furger et al., 1997; Hehl et al., 1994), cytochrome oxidase subunits (Mayho et al., 2006), and a short stem loop, necessary and sufficient for response to external purine concentration (Fernández-Moya et al., 2014). Manipulation of these elements rarely quantitatively recapitulates the observed variations in mRNA levels in vivo whereas transfer of an entire 3’UTR can (Webb et al., 2005a), this may reflect a combinatorial mechanism necessary for complete developmental regulation.

Trans-acting factors that affect mRNA levels have been identified. The most spectacular examples of these include the RNA recognition motif-containing proteins RPB6 and RPB10. Over expression of RBP6 is sufficient to drive PCFs through several successive developmental transitions (Kolev et al., 2012) and over expression of RBP10 causes a PCF transcriptome to switch to BSF (Mugo and Clayton, 2017). Other RNA binding proteins have been shown to have specific effects on sets of mRNAs, for example RBP42 binds within the open reading frame (ORF) of many mRNAs that encode proteins involved in energy metabolism (Das et al., 2012). Similarly, ZPF3 binds the LII cis-element in the EP1 procyclin mRNA 3’ UTR and displaces a negative regulator of translation leading to increased EP1 protein abundance (Walrad et al., 2009; Walrad et al., 2012).

These observations provide strong evidence for co-ordinated regulation of mRNA cohorts and have led to a model in which the fate of cytoplasmic mRNAs is regulated by a combinatorial mechanism dependent on a set of interactions between RNA-binding proteins and their cognate sites in individual mRNAs. In this model, the information for the range of possible fates for any mRNA is contained in discrete cis-elements. However, the model is derived from a consideration of a small subset of mRNAs prone to dramatic regulation and further, it fails to consider work that demonstrated a codon bias in highly expressed genes in T. brucei and related trypanosomatid species (Horn, 2008; Seward and Kelly, 2016). The use of favoured codons in highly expressed genes correlated with cognate tRNA gene numbers and led to the suggestion that translational selection was likely to act on highly abundant genes. Moreover, selection was also acting in proportion to mRNA abundance to reduce the biosynthesis requirements of those transcripts (Seward and Kelly, 2016). More recently, it has been shown that codon choice is a major determinant of mRNA levels in yeast (Presnyak et al., 2015) operating through a system that monitors the rate of translation elongation involving displacement of the RNA helicase DHH1 which in turn affects the activity of the NOT1 deadenylase complex (Radhakrishnan et al., 2016). Further, codon choice is also an important determinant of maternal mRNA turnover in zebra fish (Mishima et al., 2016) and Drosophila (Bazzini et al., 2016).

Here, we provide evidence that codon use is also central to the regulation of mRNA levels in trypanosomes. A novel codon usage statistic, the gene expression codon adaptation index (geCAI), was developed that maximises the relationship between codon usage and the measured transcript abundance for the full range of mRNAs present in a cell. Then geCAI predictions were tested using differently coded GFPs and mRNA levels were predicted with >90% accuracy over a ~ 25 fold dynamic range. An investigation of the mechanism of how geCAI determined mRNA levels showed that mRNAs with low geCAI scores had shorter half-lives and that turnover was prevented if translation was blocked. These observations and measurements show that for most mRNAs the majority of the information determining mRNA levels resides in the ORF and is accessed by the translating ribosome.

The experiments described below originated from an observation made during attempts to reduce expression of a transgene encoding a MS2 binding protein-green fluorescent protein-nuclear localisation signal (MS2bp-GFP-nls) fusion protein. Of the approaches used, adjusting the open reading frame (ORF) to codons that were under-represented in highly expressed mRNAs was more effective than other modifications including mutating the AG acceptor dinucleotide to AA in the trans-splicing site (Siegel et al., 2005), using a synthetic 5’UTR (Siegel et al., 2005) and using a 3’UTR from a low abundance mRNA (Supplementary file 1). The relative levels of the MS2bp-GFP-nls mRNA were then estimated by northern blotting (Figure 1), this showed a correlation between levels of MS2bp-GFP-nls mRNA and protein (Figure 1—figure supplement 1) levels, including the codon altered transgene mRNA. This observation implied that codon choice could determine mRNA levels and the relationship between the two was investigated further.

Figure 1 with 1 supplement see all

Download asset Open asset

Altering codon use results in a range of GFP protein expression levels

All subsequent experiments used procyclic form trypanosomes and GFP as a reporter. A range of GFP transgenes were integrated into the tubulin locus and were expressed after endogenous transcription by RNA polymerase II (Figure 2). The transgene-derived GFP mRNA contained an α-tubulin 5’UTR, the GFP ORF and the actin 3’UTR (Supplementary file 2A). Different GFP ORFs were selected from a pre-existing library (Kudla et al., 2009) and others were newly synthesised (Supplementary file 3). On analysing transgenic cell lines, it was immediately apparent that different GFP ORFs produced a range of fluorescence levels (Figure 2—figure supplement 1).

Figure 2 with 2 supplements see all

Download asset Open asset

Prior to further experiments, the use of fluorescence as a proxy for GFP protein was investigated by comparing GFP expression levels measured using flow cytometry with estimates from western blotting. Three independent clones of cell lines expressing GFPs with different codon use (eGFP, GFP 065, GFP 226 and GFP 102; Supplementary file 3) were used. GFP levels were measured by flow cytometry of live cells and lysates from the same cultures were used to estimate GFP expression using densitometric scanning of a western blot and comparison with a standard curve of recombinant GFP (Figure 2—figure supplement 2). The Pearson’s correlation coefficient between the measurement of GFP by flow cytometry and estimates of GFP expression from western blotting was R² = 0.965; validating the use of GFP fluorescence as a measure of total GFP protein.

Gene expression codon adaptation index, geCAI, a codon usage statistic to predict mRNA levels

In initial experiments using transgenes with different ORFs, the expression of GFP measured by flow cytometry did not correlate as well as might be expected with CAI (Carbone et al., 2005) or tAI (Tuller et al., 2010) scores (described in more detail below). We decided to re-visit the calculation of the association between codon scores and expression levels. Building on the observation that altering codon use affected both protein and mRNA levels for the MS2bp-GFP-nls transgene described above, the approach taken was to make an assumption that codon use directly affected mRNA levels for most constitutively expressed genes and to derive a codon statistic for genes that best correlated with their measured mRNA abundance levels.

To estimate the abundance of individual mRNAs, an RNA-Seq analysis of poly(A) enriched RNA was performed with three biological replicates of mid-log phase PCF T. brucei cells (Kelly et al., 2017) (EBI Array Express E-MTAB-3335). The level of each mRNA was then determined and expressed as the mean of the transcripts per million transcripts (TPM) across all three replicates (Supplementary file 4).

To develop a set of numerical codon values that could explain mRNA levels, an approach was used in which the codon usage statistic learnt a codon value by maximising the Spearman’s rank correlation coefficient between the measured mRNA level of a set of transcripts and the codon usage statistics for the cognate ORFs. Spearman’s rank correlation coefficient was used to avoid distributional assumptions about the hidden distribution of per-gene translational efficiencies and the observed distribution of mRNA abundance estimates. To calculate the codon values, the measured expression levels of mRNAs from 5136 single copy genes were used. This set of genes represented the genome with the following exclusions: (i) mRNAs with >2 fold differential expression when BSF and PCF trypanosomes were compared (Kelly et al., 2017) as the mRNA abundance are likely to be modulated by cis-acting RNA binding proteins and thus obscure the codon score learning process. (ii) mRNAs encoded by single copy genes without a homologue in at least one other kinetoplastid species, this was a mechanism to exclude any potentially spurious ORFs that may not encode proteins. (iii) mRNAs encoded by multicopy genes families as the accuracy of TPM calculations is compromised by uncertainty of copy number and associated errors in allocation of many sequence reads to individual genes.

The calculation of codon weights Ci proceeded by randomly generating a set of codon scores S were Ci [0,1]. Here, a score of 1 means that a codon has the maximum positive effect on mRNA levels and a score of 0 the minimum effect. The geCAI score of an ORF was evaluated as the geometric mean of the scores for all codons in that ORF. The Spearman’s rank correlation coefficient between geCAI scores for the 5136 ORFs and mRNA abundances was then computed. A Markov chain Monte Carlo algorithm was then employed to introduce stochastic changes into the codon score matrix and the entire set of genes was rescored and the Spearman’s rank correlation coefficient recalculated. If in a given generation of the Markov chain, a matrix was found that produced a higher correlation coefficient than the current best matrix, then it replaced the current best matrix and formed the basis for subsequent generations of stochastic modification. One thousand chains each starting from a different random starting matrix were initiated and allowed to run for 5000 generations. In all cases, stationary phase was reached between 1500 and 2000 generations (Figure 3A). The gene expression codon adaptation index (geCAI) values for each codon were calculated as the median of the values obtained from the 1000 chains (Figure 3B and Table 1).

Figure 3 with 2 supplements see all

Download asset Open asset

Table 1

Amino acid	Codon	geCAI codon weight	Amino acid	Codon	geCAI codon weight
A	GCA	0.55	N	AAC	1.00
A	GCG	0.33	N	AAT	0.55
A	GCC	0.32	P	CCC	0.82
A	GCT	0.21	P	CCA	0.61
C	TGC	0.20	P	CCG	0.34
C	TGT	0.02	P	CCT	0.26
D	GAC	0.78	Q	CAA	0.93
D	GAT	0.45	Q	CAG	0.51
E	GAG	0.81	R	CGC	0.63
E	GAA	0.32	R	AGA	0.50
F	TTC	0.58	R	CGT	0.22
F	TTT	0.15	R	CGA	0.17
G	GGA	1.00	R	CGG	0.08
G	GGG	0.58	R	AGG	0.06
G	GGC	0.52	S	AGC	0.19
G	GGT	0.18	S	TCA	0.18
H	CAC	0.68	S	TCC	0.16
H	CAT	0.17	S	TCG	0.15
I	ATT	1.00	S	TCT	0.13
I	ATC	0.82	S	AGT	0.04
I	ATA	0.52	T	ACG	1.00
K	AAG	1.00	T	ACC	0.93
K	AAA	0.72	T	ACA	0.78
L	CTC	0.55	T	ACT	0.66
L	TTG	0.19	V	GTG	0.25
L	CTG	0.15	V	GTT	0.15
L	CTT	0.14	V	GTC	0.13
L	CTA	0.12	V	GTA	0.12
L	TTA	0.09	W	TGG	0.22
M	ATG	1.00	Y	TAC	1.00
			Y	TAT	0.24

When mRNA abundance expressed as TPM was plotted against the ORF geCAI scores (Supplementary file 4), the Spearman’s rank correlation coefficient for the 5136 mRNAs was ρ = 0.55 (Figure 3C). A mid-log phase PCF cell contains approximately 50000 mRNA molecules (Dhalia et al., 2006) and so 20 TPM is equivalent to one mRNA molecule/cell, so the range of transcript abundance for the vast majority is between ~1 and 40 mRNAs/diploid gene pair/cell. The use of geCAI allows a prediction of a range of expression level for an mRNA based on the sequence of the ORF alone, for example an ORF with a geCAI score of 0.35 will be expressed at 2.5 mRNAs/cell on average with the vast majority of mRNAs in a range between 1 and 5.

Several previously developed codon usage statistics were tested to determine if these could explain the measured mRNA levels. Of the tested methods, CAI (Carbone et al., 2005) and tAI (Tuller et al., 2010) provided the best correlation but the predictive capacity was limited with Spearman’s rank correlation coefficients of ρ = 0.16 and ρ = 0.13 respectively (Figure 3—figure supplement 1). These measures may be unsuitable for this type of analysis as each assumes the effectiveness of the 'best' codon for each amino acid is equivalent, for example that the most efficient codon for glycine will have the same effect as the most efficient codon for alanine. There is evidence that this is not the case (Gardin et al., 2014).

The geCAI values for each codon in Table 1 are derived from the mRNA levels measured in the transcriptome analysis used in this study. There is variation in the measurements of mRNA levels in transcriptome analyses from different studies that probably has a range of origins: identity of the cell line, growth conditions, methodology of RNA preparation, RNA-Seq and data analysis and similar variation has been found in studies in yeast (Harigaya and Parker, 2016). The consequence is that any table of geCAI codon values contains an element reflecting how mRNA levels were measured, but the underlying principle that codon choice is a major determinant of mRNA levels is unaffected. To illustrate this point, transcriptome data from three other studies were analysed to determine geCAI values as above. The Pairwise Pearson correlation (r) between log₂(mRNA abundance) measures from this study and three others ranged from 0.68 to 0.86 (Supplementary file 5A). The calculation of geCAI values for each codon showed variation derived from the differences in transcriptome measurements (Supplementary file 5B and C). Each set of geCAI codon values was then used to derive geCAI scores for each of the 5136 mRNAs and these were plotted against the mRNA expression level determined in the cognate study (Figure 3—figure supplement 2). In each case, there was still a relationship between geCAI scores for ORFs and expression levels with a range of Spearman’s rank correlation coefficients from ρ = 0.35 to ρ = 0.50.

geCAI score predicts GFP protein and mRNA levels over a range similar to endogenous transcripts

Twenty-two different GFP ORFs (Supplementary file 3) were used to systematically test the predictions of expression levels made by geCAI scores. For GFP protein, expression level was measured by flow cytometry of three independent clones for each GFP transgene, in every case there was less than 5% variation in GFP protein abundance between the three clones (Supplementary file 6A and B). The GFP protein abundance was expressed relative to eGFP and plotted against geCAI score (Figure 4A); the geCAI score was able to predict GFP protein abundance with 84% accuracy (Pearson’s correlation coefficient, r² = 0.84). This compared favourably with the correlation coefficients between GFP protein abundance and either CAI (r² = 0.25) or tAI (r² = 0.61) (Supplementary file 6B).

Figure 4

Download asset Open asset

Next, the mRNA abundance for four GFP transgenes with different coding sequences with a range of geCAI scores was determined by RNA-Seq analysis. Equal numbers of cells of the four cell lines were mixed, RNA extracted and expression levels for each GFP mRNA measured in transcripts per million transcripts (TPM). Three separate analyses were performed using three independent clones for each GFP transgene. In analysing the results, the measured expression level of each GFP mRNA was adjusted to compensate for the effect of gene copy number and the dilution effect from the mixing of four different cell lines. Trypanosomes are diploid and any endogenous single copy gene will be present in two copies per cell, each of the four cell lines contained a haploid copy of a different GFP transgene so in the mixed population used to prepare RNA there are eight copies of an endogenous gene for one copy of each of the individual GFP transgenes, effectively an 8-fold lower copy number. After adjusting for the effective difference in copy number within the sequenced pool, GFP mRNA abundance was plotted against geCAI values (Figure 4B). The Pearson’s correlation coefficient between GFP mRNA levels and geCAI was r² = 0.92 (Figure 4B and Supplementary file 7). Therefore, as the genomic integration site, UTRs and amino acid sequence were all identical, codon use was the major determinant of steady state levels for the transgene derived GFP mRNA and protein.

The ~25 fold range of GFP mRNA levels, 2.5 to 60 mRNAs/diploid gene pair equivalent/cell (Figure 4B) is similar to the range present in mRNAs from single copy genes in the transcriptome. The measurements for the GFP mRNAs are in the higher end of the endogenous transcriptome, probably because the range of GFP geCAI scores included values higher than those present in most endogenous single copy genes. Thus, altering codon use can account for the range of steady state level of most mRNAs and the geCAI score is a good predictor of mRNA level for single copy genes not subject to developmental regulation. Furthermore, discrepancy between geCAI score for a gene and the observed mRNA or protein abundance for that gene is an indicator that other regulatory mechanisms, such as cis-acting RNA binding proteins may be acting on that gene to modulate mRNA abundance. mRNAs for cytosolic ribosomal proteins have significantly higher geCAI scores than those encoding mitochondrial ribosomal proteins

The mRNA abundance measurements used to obtain geCAI codon weights did not include mRNAs for 67 of the 75 cytosolic ribosomal proteins as these are encoded by gene families with two or three members. Comparison of the geCAI scores for mRNAs encoding cytoplasmic and mitochondrial ribosomal proteins provided a test of geCAI scores for endogenous genes with orthologous function but different expression levels, there are many more ribosomes in the cytoplasm than in the mitochondrion. Lists of ribosomal protein genes were derived from the structures of both types of ribosome (Hashem et al., 2013; Zíková et al., 2008) and geCAI scores calculated for each mRNA encoding a ribosomal protein. The more abundant cytosolic ribosomal protein mRNAs had higher geCAI scores than the less abundant mitochondrial ribosome protein mRNAs (Figure 5 and Supplementary file 8). The mean geCAI scores (±SD) were 0.439 (±0.034) for cytoplasmic and 0.378 (±0.033) for mitochondrial ribosomal protein mRNAs respectively. The probability of this difference arising by chance is <0.00001 (unpaired two-sample t test with equal variance). These measurements are consistent with more abundant proteins are encoded by mRNAs with higher geCAI scores.

Figure 5

Download asset Open asset

geCAI is a predictor of mRNA turnover rates

The origin of the differences in GFP mRNA levels from transgenes with differently coded ORFs was investigated by measuring turnover of five different GFP mRNAs by quantitative northern blotting after the inhibition of trans-splicing, and thus mRNA maturation, with sinefungin (Pugh et al., 1978) (Figure 6A). The two GFPs with the highest geCAI scores (eGFP: 0.547 and GFP P3: 0.520) had no detectable decay over the time course of 60 min, these both have geCAI scores greater than nearly all endogenous mRNAs. The three other GFPs with lower geCAI scores (GFP P2: 0.493, GFP 226: 0.432 and GFP 102: 0.375) had a rate of decay inversely proportional to the geCAI value (Figure 6B and Supplementary file 9). For these three GFP mRNAs decay to 50% occurred between 10 and ~80 min, similar to the range of half-lives reported for endogenous mRNAs (Fadda et al., 2014).

Figure 6 with 1 supplement see all

Download asset Open asset

One study has estimated the half-lives of most mRNAs in procyclic form trypanosomes (Fadda et al., 2014), and these measurements were plotted against geCAI scores determined from the cognate transcriptome and this study (Figure 6—figure supplement 1). The Spearman’s rank correlation coefficient was ρ = 0.22 for the cognate geCAI values and ρ = 0.33 for the geCAI values used in this study. Thus, geCAI scores calculated using either study can explain a significant proportion of variance in mRNA turnover rates at a transcriptome wide level.

Preventing translation through a hairpin in the 5’UTR blocks geCAI score-mediated instability

The observations above provide evidence that codon use is a major determinant of mRNA half-life, which in turn implies translation is involved. This was tested directly by blocking or reducing translation through the inclusion of secondary structures in the 5’UTR. Five different GFP transgene constructs with a range of geCAI scores were modified by insertion of a 48 base hairpin (24 base pairs) in the 5’UTR located 59 nucleotides from the cap and 88 nucleotides from the initiation codon (Supplementary file 2B). The same approach has been used previously in trypanosomes to block translation of a specific mRNA (Webb et al., 2005b). Measurements of GFP protein and mRNA expression were made in three independent clones for each transgene. As expected, there was no detectable expression of GFP by flow cytometry in any of the cell lines containing a transgene with a hairpin (Supplementary file 6C and 6D), therefore the hairpin-containing mRNAs were not translated. Measurements of the mRNA levels by quantitative northern blotting revealed that the effect of inclusion of the hairpin in the 5’UTR was to stabilise the GFP mRNAs (Figure 7A and Supplementary file 6E) and the lower the geCAI score the greater the relative increase in abundance of the hairpin to control mRNAs from 1.4-fold for eGFP (geCAI = 0.547) to 14-fold for GFP 102 (geCAI = 0.375) (Figure 7B). Could this observation be explained by the persistence of decapped mRNA stabilised by the hairpin blocking 5’ to 3’ decay? This is very unlikely as an mRNA with a similar size and GC-content hairpin in the 5’UTR was readily degraded in procyclic trypanosomes (Webb et al., 2005b).

Figure 7

Download asset Open asset

These observations provide evidence that: (i) the translation of an ORF is necessary for the variation in mRNA half-life conferred by the geCAI, (ii) the different GFP mRNA levels result from active destabilisation, and (iii) none of the mRNAs with differently coded GFPs are inherently unstable due to sequence alone.

Insertion of a shorter secondary structure in the 5’UTR reduces GFP ORF translation but has little effect on state levels of GFP mRNA

The experiments above indicated that translation is necessary for geCAI score-mediated mRNA turnover. The 5’UTR hairpin approach above was modified by decreasing the length of double stranded RNA inserted into the 5’UTR until eGFP protein was expressed. Reduction of the length of the hairpin from 24 bp to 18 bp still blocked translation and no GFP fluorescence was observed. In contrast, a 12 bp stem loop resulted in eGFP protein expression that was 11% of that from the parental transgene without a hairpin in the 5’UTR (Supplementary file 6F). The mRNA levels from three independent clones expressing GFP transgenes with the wild type 5’UTR, 18 bp hairpin containing 5’UTR and 12 bp stem loop containing 5’UTR were measured (Figure 8 and Supplementary file 6G), the presence of the 12 bp stem loop resulted in a modest reduction to 76% in the steady state GFP mRNA levels (p=0.13 for a difference between the mRNAs with and without a 12 bp stem loop; unpaired two-sample t test). This observation that a nine-fold reduction in the frequency of translation has only a small effect on the GFP mRNA levels favours a model in which geCAI score is interpreted by the translating ribosome by a mechanism mostly independent of ribosome density/frequency of translation, possibly the rate of ribosome progression.

Figure 8

Download asset Open asset

Two sets of ribosomal profiling data are available that contain values for ribosome density for >5000 of the 5136 non-developmentally regulated mRNAs expressed from single copy genes that were used to derive geCAI values (Vasquez et al., 2014; Jensen et al., 2014). There is a correlation between ribosome density and mRNA levels but there is also great variation in the ribosome density on different mRNAs with similar expression levels (Jensen et al., 2014) indicating frequency of translation/ribosome density alone is not sufficient to determine translationally regulated mRNA turnover. The geCAI scores of mRNAs were plotted against ribosomes density for the two ribosome profiling datasets (Supplementary file 10). The Pearson’s correlation coefficient between ribosome footprint levels and geCAI was R² = 0.141 and 0.244 for the two sets of ribosome density values. Thus occupancy/frequency of translation can explain a significant component of the geCAI measure, but suggests a model in which geCAI mediated mRNA turnover is also dependent on other factors.

Many developmentally regulated mRNAs do not conform to geCAI predictions

The geCAI values for each codon were calculated using a set of values for mRNA expression levels that excluded developmentally regulated mRNAs, defined in the case as having a twofold or greater difference in abundance when PCF cells were compared with BSF cells. Analysing only single copy genes, the geCAI scores for 372 mRNAs upregulated in PCF cells, and 176 mRNAs upregulated in BSF cells were calculated and plotted against expression levels in PCF cells (Figure 9). mRNAs upregulated in PCFs are largely expressed at higher levels than would be predicted by geCAI alone. In contrast, mRNAs downregulated in PCFs (upregulated in BSFs) are expressed at levels lower than predicted by geCAI. Taken together, these measurements provide evidence that developmental regulation results from both stabilisation and destabilisation pathways that act in concert to modulate the geCAI determined mRNA levels.

Figure 9

Download asset Open asset

Trypanosomatids have evolved separately from other eukaryotic lineages for ~500 million years (Lukeš et al., 2014). Over this time the core gene expression machinery present in the last eukaryotic common ancestor has been largely conserved but selective transcription of individual genes by RNA polymerase II has been lost and individual mRNAs arise by co-transcriptional processing of polycistronically transcribed precursors. Tens or hundreds of genes, mostly with no obvious functional link to each other, are transcribed from a single start site and monocistronic mRNAs arise by co-transcriptional trans-splicing and polyadenylation. The downstream consequence is that individual mRNA levels must be set post-transcriptionally.

An observation that altering the codon use of a transgene to codons under-represented in highly expressed mRNAs not only reduced protein expression but also the cognate mRNA to a similar extent led us to investigate the effect of codon use on mRNA levels. The major finding presented here is that codon use is central to the regulation of mRNA levels in trypanosomes. A novel codon usage metric, the gene expression codon adaptation index (geCAI), was developed that maximises the relationship between codon usage and the measured mRNA abundance. The geCAI score for an ORF allows a prediction of mRNA levels for constitutively expressed genes. Differently coded GFPs were used as a test and mRNA expression levels were successfully predicted over a ~ 25 fold range, similar to the range of endogenous mRNAs. The differences in GFP mRNA levels resulted from differences in the half-lives of the mRNAs, the lower the geCAI score, the shorter the half-life. An investigation of the mechanism of geCAI mediated mRNA half-life showed that translation was necessary as mRNA turnover was reduced/prevented if translation was blocked using a hairpin in the 5’UTR. Further, insertion into the 5’UTR of secondary structure that greatly reduced but did not block translation had only a small effect on steady state mRNA levels, providing evidence that the sensor for geCAI mediated turnover was detecting ribosome progression rather than frequency of translation initiation.

1. 1. Kelly S
   2. Ivens A
   3. Mott GA
   4. O'Neill E
   5. Emms D
   6. Macleod O
   7. Voorheis P
   8. Tyler K
   9. Clark M
   10. Matthews J
   11. Matthews K
   12. Carrington M

   (2017) Continuous culture of bloodstream form and procyclic form of Trypanosoma brucei

   Publicly available at ArrayExpress (accession no. E-MTAB-3335).

   https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-3335/

1. 1. Bazzini AA
   2. Del Viso F
   3. Moreno-Mateos MA
   4. Johnstone TG
   5. Vejnar CE
   6. Qin Y
   7. Yao J
   8. Khokha MK
   9. Giraldez AJ

   (2016) Codon identity regulates mRNA stability and translation efficiency during the maternal-to-zygotic transition

   The EMBO Journal 35:2087–2103.

   https://doi.org/10.15252/embj.201694699

   • PubMed
   • Google Scholar
2. 1. Berriman M
   2. Ghedin E
   3. Hertz-Fowler C
   4. Blandin G
   5. Renauld H
   6. Bartholomeu DC
   7. Lennard NJ
   8. Caler E
   9. Hamlin NE
   10. Haas B
   11. Böhme U
   12. Hannick L
   13. Aslett MA
   14. Shallom J
   15. Marcello L
   16. Hou L
   17. Wickstead B
   18. Alsmark UC
   19. Arrowsmith C
   20. Atkin RJ
   21. Barron AJ
   22. Bringaud F
   23. Brooks K
   24. Carrington M
   25. Cherevach I
   26. Chillingworth TJ
   27. Churcher C
   28. Clark LN
   29. Corton CH
   30. Cronin A
   31. Davies RM
   32. Doggett J
   33. Djikeng A
   34. Feldblyum T
   35. Field MC
   36. Fraser A
   37. Goodhead I
   38. Hance Z
   39. Harper D
   40. Harris BR
   41. Hauser H
   42. Hostetler J
   43. Ivens A
   44. Jagels K
   45. Johnson D
   46. Johnson J
   47. Jones K
   48. Kerhornou AX
   49. Koo H
   50. Larke N
   51. Landfear S
   52. Larkin C
   53. Leech V
   54. Line A
   55. Lord A
   56. Macleod A
   57. Mooney PJ
   58. Moule S
   59. Martin DM
   60. Morgan GW
   61. Mungall K
   62. Norbertczak H
   63. Ormond D
   64. Pai G
   65. Peacock CS
   66. Peterson J
   67. Quail MA
   68. Rabbinowitsch E
   69. Rajandream MA
   70. Reitter C
   71. Salzberg SL
   72. Sanders M
   73. Schobel S
   74. Sharp S
   75. Simmonds M
   76. Simpson AJ
   77. Tallon L
   78. Turner CM
   79. Tait A
   80. Tivey AR
   81. Van Aken S
   82. Walker D
   83. Wanless D
   84. Wang S
   85. White B
   86. White O
   87. Whitehead S
   88. Woodward J
   89. Wortman J
   90. Adams MD
   91. Embley TM
   92. Gull K
   93. Ullu E
   94. Barry JD
   95. Fairlamb AH
   96. Opperdoes F
   97. Barrell BG
   98. Donelson JE
   99. Hall N
   100. Fraser CM
   101. Melville SE
   102. El-Sayed NM

   (2005) The genome of the african trypanosome trypanosoma brucei

   Science 309:416–422.

   https://doi.org/10.1126/science.1112642

   • PubMed
   • Google Scholar
3. 1. Bloom-Ackermann Z
   2. Navon S
   3. Gingold H
   4. Towers R
   5. Pilpel Y
   6. Dahan O

   (2014) A comprehensive tRNA deletion library unravels the genetic architecture of the tRNA pool

   PLoS Genetics 10:e1004084.

   https://doi.org/10.1371/journal.pgen.1004084

   • PubMed
   • Google Scholar
4. 1. Bolger AM
   2. Lohse M
   3. Usadel B

   (2014) Trimmomatic: a flexible trimmer for Illumina sequence data

   Bioinformatics 30:2114–2120.

   https://doi.org/10.1093/bioinformatics/btu170

   • PubMed
   • Google Scholar
5. 1. Brun R
   2. Jenni L

   (1977)

   A new semi-defined medium for Trypanosoma brucei sspp

   Acta Tropica 34:21–33.

   • PubMed
   • Google Scholar
6. 1. Bruske EI
   2. Sendfeld F
   3. Schneider A

   (2009) Thiolated tRNAs of Trypanosoma brucei are imported into mitochondria and dethiolated after import

   Journal of Biological Chemistry 284:36491–36499.

   https://doi.org/10.1074/jbc.M109.064527

   • PubMed
   • Google Scholar
7. 1. Carbone A
   2. Képès F
   3. Zinovyev A

   (2005) Codon bias signatures, organization of microorganisms in codon space, and lifestyle

   Molecular Biology and Evolution 22:547–561.

   https://doi.org/10.1093/molbev/msi040

   • PubMed
   • Google Scholar
8. 1. Carrington M
   2. Roditi I
   3. Williams RO

   (1987) The structure and transcription of an element interspersed between tandem arrays of mini-exon donor RNA genes in Trypanosoma brucei

   Nucleic Acids Research 15:10179–10198.

   https://doi.org/10.1093/nar/15.24.10179

   • PubMed
   • Google Scholar
9. 1. Christiano R
   2. Kolev NG
   3. Shi H
   4. Ullu E
   5. Walther TC
   6. Tschudi C

   (2017) The proteome and transcriptome of the infectious metacyclic form of Trypanosoma brucei define quiescent cells primed for mammalian invasion

   Molecular Microbiology 106:74–92.

   https://doi.org/10.1111/mmi.13754

   • PubMed
   • Google Scholar
10. 1. Chu D
    2. Barnes DJ
    3. von der Haar T

    (2011) The role of tRNA and ribosome competition in coupling the expression of different mRNAs in Saccharomyces cerevisiae

    Nucleic Acids Research 39:6705–6714.

    https://doi.org/10.1093/nar/gkr300

    • PubMed
    • Google Scholar
11. 1. Chu D
    2. Kazana E
    3. Bellanger N
    4. Singh T
    5. Tuite MF
    6. von der Haar T

    (2014) Translation elongation can control translation initiation on eukaryotic mRNAs

    The EMBO Journal 33:21–34.

    https://doi.org/10.1002/embj.201385651

    • PubMed
    • Google Scholar
12. 1. Clayton C

    (2013) The regulation of trypanosome gene expression by RNA-binding proteins

    PLoS Pathogens 9:e1003680.

    https://doi.org/10.1371/journal.ppat.1003680

    • PubMed
    • Google Scholar
13. 1. Coller JM
    2. Tucker M
    3. Sheth U
    4. Valencia-Sanchez MA
    5. Parker R

    (2001) The DEAD box helicase, Dhh1p, functions in mRNA decapping and interacts with both the decapping and deadenylase complexes

    RNA 7:1717–1727.

    https://doi.org/10.1017/S135583820101994X

    • PubMed
    • Google Scholar
14. 1. Darnell JC
    2. Van Driesche SJ
    3. Zhang C
    4. Hung KY
    5. Mele A
    6. Fraser CE
    7. Stone EF
    8. Chen C
    9. Fak JJ
    10. Chi SW
    11. Licatalosi DD
    12. Richter JD
    13. Darnell RB

    (2011) FMRP stalls ribosomal translocation on mRNAs linked to synaptic function and autism

    Cell 146:247–261.

    https://doi.org/10.1016/j.cell.2011.06.013

    • PubMed
    • Google Scholar
15. 1. Das A
    2. Morales R
    3. Banday M
    4. Garcia S
    5. Hao L
    6. Cross GA
    7. Estevez AM
    8. Bellofatto V

    (2012) The essential polysome-associated RNA-binding protein RBP42 targets mRNAs involved in Trypanosoma brucei energy metabolism

    RNA 18:1968–1983.

    https://doi.org/10.1261/rna.033829.112

    • PubMed
    • Google Scholar
16. 1. Dhalia R
    2. Marinsek N
    3. Reis CR
    4. Katz R
    5. Muniz JR
    6. Standart N
    7. Carrington M
    8. de Melo Neto OP

    (2006) The two eIF4A helicases in Trypanosoma brucei are functionally distinct

    Nucleic Acids Research 34:2495–2507.

    https://doi.org/10.1093/nar/gkl290

    • PubMed
    • Google Scholar
17. 1. dos Reis M
    2. Savva R
    3. Wernisch L

    (2004) Solving the riddle of codon usage preferences: a test for translational selection

    Nucleic Acids Research 32:5036–5044.

    https://doi.org/10.1093/nar/gkh834

    • PubMed
    • Google Scholar
18. 1. El-Sayed NM
    2. Myler PJ
    3. Blandin G
    4. Berriman M
    5. Crabtree J
    6. Aggarwal G
    7. Caler E
    8. Renauld H
    9. Worthey EA
    10. Hertz-Fowler C
    11. Ghedin E
    12. Peacock C
    13. Bartholomeu DC
    14. Haas BJ
    15. Tran AN
    16. Wortman JR
    17. Alsmark UC
    18. Angiuoli S
    19. Anupama A
    20. Badger J
    21. Bringaud F
    22. Cadag E
    23. Carlton JM
    24. Cerqueira GC
    25. Creasy T
    26. Delcher AL
    27. Djikeng A
    28. Embley TM
    29. Hauser C
    30. Ivens AC
    31. Kummerfeld SK
    32. Pereira-Leal JB
    33. Nilsson D
    34. Peterson J
    35. Salzberg SL
    36. Shallom J
    37. Silva JC
    38. Sundaram J
    39. Westenberger S
    40. White O
    41. Melville SE
    42. Donelson JE
    43. Andersson B
    44. Stuart KD
    45. Hall N

    (2005) Comparative genomics of trypanosomatid parasitic protozoa

    Science 309:404–409.

    https://doi.org/10.1126/science.1112181

    • PubMed
    • Google Scholar
19. 1. Ersfeld K
    2. Asbeck K
    3. Gull K

    (1998) Direct visualisation of individual gene organisation in Trypanosoma brucei by high-resolution in situ hybridisation

    Chromosoma 107:237–240.

    https://doi.org/10.1007/s004120050302

    • PubMed
    • Google Scholar
20. 1. Fadda A
    2. Ryten M
    3. Droll D
    4. Rojas F
    5. Färber V
    6. Haanstra JR
    7. Merce C
    8. Bakker BM
    9. Matthews K
    10. Clayton C

    (2014) Transcriptome-wide analysis of trypanosome mRNA decay reveals complex degradation kinetics and suggests a role for co-transcriptional degradation in determining mRNA levels

    Molecular Microbiology 94:307–326.

    https://doi.org/10.1111/mmi.12764

    • PubMed
    • Google Scholar
21. 1. Färber V
    2. Erben E
    3. Sharma S
    4. Stoecklin G
    5. Clayton C

    (2013) Trypanosome CNOT10 is essential for the integrity of the NOT deadenylase complex and for degradation of many mRNAs

    Nucleic Acids Research 41:1211–1222.

    https://doi.org/10.1093/nar/gks1133

    • PubMed
    • Google Scholar
22. 1. Fernández-Moya SM
    2. Carrington M
    3. Estévez AM

    (2014) A short RNA stem-loop is necessary and sufficient for repression of gene expression during early logarithmic phase in trypanosomes

    Nucleic Acids Research 42:7201–7209.

    https://doi.org/10.1093/nar/gku358

    • PubMed
    • Google Scholar
23. 1. Fiebig M
    2. Kelly S
    3. Gluenz E

    (2015) Comparative life cycle transcriptomics revises leishmania mexicana genome annotation and links a chromosome duplication with parasitism of vertebrates

    PLOS Pathogens 11:e1005186.

    https://doi.org/10.1371/journal.ppat.1005186

    • Google Scholar
24. 1. Fluitt A
    2. Pienaar E
    3. Viljoen H

    (2007) Ribosome kinetics and aa-tRNA competition determine rate and fidelity of peptide synthesis

    Computational Biology and Chemistry 31:335–346.

    https://doi.org/10.1016/j.compbiolchem.2007.07.003

    • PubMed
    • Google Scholar
25. 1. Furger A
    2. Schürch N
    3. Kurath U
    4. Roditi I

    (1997) Elements in the 3' untranslated region of procyclin mRNA regulate expression in insect forms of Trypanosoma brucei by modulating RNA stability and translation

    Molecular and Cellular Biology 17:4372–4380.

    https://doi.org/10.1128/MCB.17.8.4372

    • PubMed
    • Google Scholar
26. 1. Gardin J
    2. Yeasmin R
    3. Yurovsky A
    4. Cai Y
    5. Skiena S
    6. Futcher B

    (2014) Measurement of average decoding rates of the 61 sense codons in vivo

    eLife 3:e03735.

    https://doi.org/10.7554/eLife.03735

    • Google Scholar
27. 1. Gaston KW
    2. Rubio MA
    3. Spears JL
    4. Pastar I
    5. Papavasiliou FN
    6. Alfonzo JD

    (2007) C to U editing at position 32 of the anticodon loop precedes tRNA 5' leader removal in trypanosomatids

    Nucleic Acids Research 35:6740–6749.

    https://doi.org/10.1093/nar/gkm745

    • PubMed
    • Google Scholar
28. 1. Haanstra JR
    2. Stewart M
    3. Luu VD
    4. van Tuijl A
    5. Westerhoff HV
    6. Clayton C
    7. Bakker BM

    (2008) Control and regulation of gene expression: quantitative analysis of the expression of phosphoglycerate kinase in bloodstream form Trypanosoma brucei

    The Journal of biological chemistry 283:2495–2507.

    https://doi.org/10.1074/jbc.M705782200

    • PubMed
    • Google Scholar
29. 1. Harigaya Y
    2. Parker R

    (2016) Analysis of the association between codon optimality and mRNA stability in Schizosaccharomyces pombe

    BMC Genomics 17:895.

    https://doi.org/10.1186/s12864-016-3237-6

    • PubMed
    • Google Scholar
30. 1. Hashem Y
    2. des Georges A
    3. Fu J
    4. Buss SN
    5. Jossinet F
    6. Jobe A
    7. Zhang Q
    8. Liao HY
    9. Grassucci RA
    10. Bajaj C
    11. Westhof E
    12. Madison-Antenucci S
    13. Frank J

    (2013) High-resolution cryo-electron microscopy structure of the Trypanosoma brucei ribosome

    Nature 494:385–389.

    https://doi.org/10.1038/nature11872

    • PubMed
    • Google Scholar
31. 1. Heck AM
    2. Wilusz J

    (2018) The Interplay between the RNA Decay and translation machinery in eukaryotes

    Cold Spring Harbor Perspectives in Biology a032839.

    https://doi.org/10.1101/cshperspect.a032839

    • PubMed
    • Google Scholar
32. 1. Hehl A
    2. Vassella E
    3. Braun R
    4. Roditi I

    (1994) A conserved stem-loop structure in the 3' untranslated region of procyclin mRNAs regulates expression in Trypanosoma brucei

    PNAS 91:370–374.

    https://doi.org/10.1073/pnas.91.1.370

    • PubMed
    • Google Scholar
33. 1. Horn D

    (2008) Codon usage suggests that translational selection has a major impact on protein expression in trypanosomatids

    BMC Genomics 9:2.

    https://doi.org/10.1186/1471-2164-9-2

    • PubMed
    • Google Scholar
34. 1. Hutchinson S
    2. Glover L
    3. Horn D

    (2016) High-resolution analysis of multi-copy variant surface glycoprotein gene expression sites in African trypanosomes

    BMC Genomics 17:806.

    https://doi.org/10.1186/s12864-016-3154-8

    • PubMed
    • Google Scholar
35. 1. Ivens AC
    2. Peacock CS
    3. Worthey EA
    4. Murphy L
    5. Aggarwal G
    6. Berriman M
    7. Sisk E
    8. Rajandream MA
    9. Adlem E
    10. Aert R
    11. Anupama A
    12. Apostolou Z
    13. Attipoe P
    14. Bason N
    15. Bauser C
    16. Beck A
    17. Beverley SM
    18. Bianchettin G
    19. Borzym K
    20. Bothe G
    21. Bruschi CV
    22. Collins M
    23. Cadag E
    24. Ciarloni L
    25. Clayton C
    26. Coulson RM
    27. Cronin A
    28. Cruz AK
    29. Davies RM
    30. De Gaudenzi J
    31. Dobson DE
    32. Duesterhoeft A
    33. Fazelina G
    34. Fosker N
    35. Frasch AC
    36. Fraser A
    37. Fuchs M
    38. Gabel C
    39. Goble A
    40. Goffeau A
    41. Harris D
    42. Hertz-Fowler C
    43. Hilbert H
    44. Horn D
    45. Huang Y
    46. Klages S
    47. Knights A
    48. Kube M
    49. Larke N
    50. Litvin L
    51. Lord A
    52. Louie T
    53. Marra M
    54. Masuy D
    55. Matthews K
    56. Michaeli S
    57. Mottram JC
    58. Müller-Auer S
    59. Munden H
    60. Nelson S
    61. Norbertczak H
    62. Oliver K
    63. O'neil S
    64. Pentony M
    65. Pohl TM
    66. Price C
    67. Purnelle B
    68. Quail MA
    69. Rabbinowitsch E
    70. Reinhardt R
    71. Rieger M
    72. Rinta J
    73. Robben J
    74. Robertson L
    75. Ruiz JC
    76. Rutter S
    77. Saunders D
    78. Schäfer M
    79. Schein J
    80. Schwartz DC
    81. Seeger K
    82. Seyler A
    83. Sharp S
    84. Shin H
    85. Sivam D
    86. Squares R
    87. Squares S
    88. Tosato V
    89. Vogt C
    90. Volckaert G
    91. Wambutt R
    92. Warren T
    93. Wedler H
    94. Woodward J
    95. Zhou S
    96. Zimmermann W
    97. Smith DF
    98. Blackwell JM
    99. Stuart KD
    100. Barrell B
    101. Myler PJ

    (2005) The genome of the kinetoplastid parasite, Leishmania major

    Science 309:436–442.

    https://doi.org/10.1126/science.1112680

    • PubMed
    • Google Scholar
36. 1. Jensen BC
    2. Ramasamy G
    3. Vasconcelos EJ
    4. Ingolia NT
    5. Myler PJ
    6. Parsons M

    (2014) Extensive stage-regulation of translation revealed by ribosome profiling of Trypanosoma brucei

    BMC Genomics 15:911.

    https://doi.org/10.1186/1471-2164-15-911

    • PubMed
    • Google Scholar
37. Website

    1. Kelly S
    2. geCAI

    (2017) geCAI

    GitHub.

    https://github.com/SteveKellyLab/geCAI
38. 1. Kelly S
    2. Ivens A
    3. Mott GA
    4. O'Neill E
    5. Emms D
    6. Macleod O
    7. Voorheis P
    8. Tyler K
    9. Clark M
    10. Matthews J
    11. Matthews K
    12. Carrington M

    (2017) An alternative strategy for trypanosome survival in the mammalian bloodstream revealed through genome and transcriptome analysis of the ubiquitous bovine parasite trypanosoma (megatrypanum) theileri

    Genome Biology and Evolution 9:2093–2109.

    https://doi.org/10.1093/gbe/evx152

    • PubMed
    • Google Scholar
39. 1. Kelly S
    2. Kramer S
    3. Schwede A
    4. Maini PK
    5. Gull K
    6. Carrington M

    (2012) Genome organization is a major component of gene expression control in response to stress and during the cell division cycle in trypanosomes

    Open Biology 2:120033.

    https://doi.org/10.1098/rsob.120033

    • PubMed
    • Google Scholar
40. 1. Kolev NG
    2. Franklin JB
    3. Carmi S
    4. Shi H
    5. Michaeli S
    6. Tschudi C

    (2010) The transcriptome of the human pathogen Trypanosoma brucei at single-nucleotide resolution

    PLoS Pathogens 6:e1001090.

    https://doi.org/10.1371/journal.ppat.1001090

    • PubMed
    • Google Scholar
41. 1. Kolev NG
    2. Ramey-Butler K
    3. Cross GA
    4. Ullu E
    5. Tschudi C

    (2012) Developmental progression to infectivity in Trypanosoma brucei triggered by an RNA-binding protein

    Science 338:1352–1353.

    https://doi.org/10.1126/science.1229641

    • PubMed
    • Google Scholar
42. 1. Kramer S
    2. Queiroz R
    3. Ellis L
    4. Hoheisel JD
    5. Clayton C
    6. Carrington M

    (2010) The RNA helicase DHH1 is central to the correct expression of many developmentally regulated mRNAs in trypanosomes

    Journal of Cell Science 123:699–711.

    https://doi.org/10.1242/jcs.058511

    • PubMed
    • Google Scholar
43. 1. Krog JS
    2. Español Y
    3. Giessing AM
    4. Dziergowska A
    5. Malkiewicz A
    6. Ribas de Pouplana L
    7. Kirpekar F

    (2011) 3-(3-amino-3-carboxypropyl)-5,6-dihydrouridine is one of two novel post-transcriptional modifications in tRNALys(UUU) from Trypanosoma brucei

    FEBS Journal 278:4782–4796.

    https://doi.org/10.1111/j.1742-4658.2011.08379.x

    • PubMed
    • Google Scholar
44. 1. Kudla G
    2. Murray AW
    3. Tollervey D
    4. Plotkin JB

    (2009) Coding-sequence determinants of gene expression in Escherichia coli

    Science 324:255–258.

    https://doi.org/10.1126/science.1170160

    • PubMed
    • Google Scholar
45. 1. Langmead B
    2. Salzberg SL

    (2012) Fast gapped-read alignment with Bowtie 2

    Nature Methods 9:357–359.

    https://doi.org/10.1038/nmeth.1923

    • PubMed
    • Google Scholar
46. 1. Li B
    2. Dewey CN

    (2011) RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome

    BMC Bioinformatics 12:323.

    https://doi.org/10.1186/1471-2105-12-323

    • PubMed
    • Google Scholar
47. 1. Lukeš J
    2. Skalický T
    3. Tyc J
    4. Votýpka J
    5. Yurchenko V

    (2014) Evolution of parasitism in kinetoplastid flagellates

    Molecular and Biochemical Parasitology 195:115–122.

    https://doi.org/10.1016/j.molbiopara.2014.05.007

    • PubMed
    • Google Scholar
48. 1. Matthews KR
    2. Tschudi C
    3. Ullu E

    (1994) A common pyrimidine-rich motif governs trans-splicing and polyadenylation of tubulin polycistronic pre-mRNA in trypanosomes

    Genes & Development 8:491–501.

    https://doi.org/10.1101/gad.8.4.491

    • PubMed
    • Google Scholar
49. 1. Mayho M
    2. Fenn K
    3. Craddy P
    4. Crosthwaite S
    5. Matthews K

    (2006) Post-transcriptional control of nuclear-encoded cytochrome oxidase subunits in Trypanosoma brucei: evidence for genome-wide conservation of life-cycle stage-specific regulatory elements

    Nucleic Acids Research 34:5312–5324.

    https://doi.org/10.1093/nar/gkl598

    • PubMed
    • Google Scholar
50. 1. McCulloch R
    2. Vassella E
    3. Burton P
    4. Boshart M
    5. Barry JD

    (2004) Transformation of monomorphic and pleomorphic Trypanosoma brucei

    Methods in Molecular Biology 262:53–86.

    https://doi.org/10.1385/1-59259-761-0:053

    • PubMed
    • Google Scholar
51. 1. Mishima Y
    2. Tomari Y
    3. Usage C

    (2016) Codon Usage and 3' UTR Length Determine Maternal mRNA Stability in Zebrafish

    Molecular Cell 61:874–885.

    https://doi.org/10.1016/j.molcel.2016.02.027

    • PubMed
    • Google Scholar
52. 1. Mugo E
    2. Clayton C

    (2017) Expression of the RNA-binding protein RBP10 promotes the bloodstream-form differentiation state in Trypanosoma brucei

    PLOS Pathogens 13:e1006560.

    https://doi.org/10.1371/journal.ppat.1006560

    • PubMed
    • Google Scholar
53. 1. Murphy WJ
    2. Watkins KP
    3. Agabian N

    (1986) Identification of a novel Y branch structure as an intermediate in trypanosome mRNA processing: evidence for trans splicing

    Cell 47:517–525.

    https://doi.org/10.1016/0092-8674(86)90616-1

    • PubMed
    • Google Scholar
54. 1. Novoa EM
    2. Pavon-Eternod M
    3. Pan T
    4. Ribas de Pouplana L

    (2012) A role for tRNA modifications in genome structure and codon usage

    Cell 149:202–213.

    https://doi.org/10.1016/j.cell.2012.01.050

    • PubMed
    • Google Scholar
55. 1. Novoa EM
    2. Ribas de Pouplana L

    (2012) Speeding with control: codon usage, tRNAs, and ribosomes

    Trends in Genetics 28:574–581.

    https://doi.org/10.1016/j.tig.2012.07.006

    • PubMed
    • Google Scholar
56. 1. Padilla-Mejía NE
    2. Florencio-Martínez LE
    3. Figueroa-Angulo EE
    4. Manning-Cela RG
    5. Hernández-Rivas R
    6. Myler PJ
    7. Martínez-Calvillo S

    (2009) Gene organization and sequence analyses of transfer RNA genes in Trypanosomatid parasites

    BMC Genomics 10:232.

    https://doi.org/10.1186/1471-2164-10-232

    • PubMed
    • Google Scholar
57. 1. Presnyak V
    2. Alhusaini N
    3. Chen YH
    4. Martin S
    5. Morris N
    6. Kline N
    7. Olson S
    8. Weinberg D
    9. Baker KE
    10. Graveley BR
    11. Coller J

    (2015) Codon optimality is a major determinant of mRNA stability

    Cell 160:1111–1124.

    https://doi.org/10.1016/j.cell.2015.02.029

    • PubMed
    • Google Scholar
58. 1. Pugh CS
    2. Borchardt RT
    3. Stone HO

    (1978)

    Sinefungin, a potent inhibitor of virion mRNA(guanine-7-)-methyltransferase, mRNA(nucleoside-2'-)-methyltransferase, and viral multiplication

    The Journal of Biological Chemistry 253:4075–4077.

    • PubMed
    • Google Scholar
59. 1. Radhakrishnan A
    2. Chen YH
    3. Martin S
    4. Alhusaini N
    5. Green R
    6. Coller J

    (2016) The DEAD-Box protein Dhh1p Couples mRNA decay and translation by monitoring codon optimality

    Cell 167:122–132.

    https://doi.org/10.1016/j.cell.2016.08.053

    • PubMed
    • Google Scholar
60. 1. Ragone FL
    2. Spears JL
    3. Wohlgamuth-Benedum JM
    4. Kreel N
    5. Papavasiliou FN
    6. Alfonzo JD

    (2011) The C-terminal end of the Trypanosoma brucei editing deaminase plays a critical role in tRNA binding

    RNA 17:1296–1306.

    https://doi.org/10.1261/rna.2748211

    • PubMed
    • Google Scholar
61. 1. Rubio MA
    2. Pastar I
    3. Gaston KW
    4. Ragone FL
    5. Janzen CJ
    6. Cross GA
    7. Papavasiliou FN
    8. Alfonzo JD

    (2007) An adenosine-to-inosine tRNA-editing enzyme that can perform C-to-U deamination of DNA

    PNAS 104:7821–7826.

    https://doi.org/10.1073/pnas.0702394104

    • PubMed
    • Google Scholar
62. 1. Rubio MA
    2. Ragone FL
    3. Gaston KW
    4. Ibba M
    5. Alfonzo JD

    (2006) C to U editing stimulates A to I editing in the anticodon loop of a cytoplasmic threonyl tRNA in Trypanosoma brucei

    Journal of Biological Chemistry 281:115–120.

    https://doi.org/10.1074/jbc.M510136200

    • PubMed
    • Google Scholar
63. 1. Schwede A
    2. Ellis L
    3. Luther J
    4. Carrington M
    5. Stoecklin G
    6. Clayton C

    (2008) A role for Caf1 in mRNA deadenylation and decay in trypanosomes and human cells

    Nucleic Acids Research 36:3374–3388.

    https://doi.org/10.1093/nar/gkn108

    • PubMed
    • Google Scholar
64. 1. Seward EA
    2. Kelly S

    (2016) Dietary nitrogen alters codon bias and genome composition in parasitic microorganisms

    Genome Biology 17:226.

    https://doi.org/10.1186/s13059-016-1087-9

    • PubMed
    • Google Scholar
65. 1. Sharp PM
    2. Li WH
    3. Wh L

    (1987) The codon Adaptation Index--a measure of directional synonymous codon usage bias, and its potential applications

    Nucleic Acids Research 15:1281–1295.

    https://doi.org/10.1093/nar/15.3.1281

    • PubMed
    • Google Scholar
66. 1. Siegel TN
    2. Hekstra DR
    3. Kemp LE
    4. Figueiredo LM
    5. Lowell JE
    6. Fenyo D
    7. Wang X
    8. Dewell S
    9. Cross GA

    (2009) Four histone variants mark the boundaries of polycistronic transcription units in Trypanosoma brucei

    Genes & Development 23:1063–1076.

    https://doi.org/10.1101/gad.1790409

    • PubMed
    • Google Scholar
67. 1. Siegel TN
    2. Tan KS
    3. Cross GA

    (2005) Systematic study of sequence motifs for RNA trans splicing in Trypanosoma brucei

    Molecular and Cellular Biology 25:9586–9594.

    https://doi.org/10.1128/MCB.25.21.9586-9594.2005

    • PubMed
    • Google Scholar
68. 1. Subramanian A
    2. Sarkar RR

    (2015) Comparison of codon usage bias across Leishmania and Trypanosomatids to understand mRNA secondary structure, relative protein abundance and pathway functions

    Genomics 106:232–241.

    https://doi.org/10.1016/j.ygeno.2015.05.009

    • PubMed
    • Google Scholar
69. 1. Sutton RE
    2. Boothroyd JC

    (1986) Evidence for trans splicing in trypanosomes

    Cell 47:527–535.

    https://doi.org/10.1016/0092-8674(86)90617-3

    • PubMed
    • Google Scholar
70. 1. Sweet T
    2. Kovalak C
    3. Coller J

    (2012) The DEAD-box protein Dhh1 promotes decapping by slowing ribosome movement

    PLoS Biology 10:e1001342.

    https://doi.org/10.1371/journal.pbio.1001342

    • PubMed
    • Google Scholar
71. 1. Tan TH
    2. Pach R
    3. Crausaz A
    4. Ivens A
    5. Schneider A

    (2002) tRNAs in Trypanosoma brucei: genomic organization, expression, and mitochondrial import

    Molecular and Cellular Biology 22:3707–3717.

    https://doi.org/10.1128/MCB.22.11.3707-3716.2002

    • PubMed
    • Google Scholar
72. 1. Tuller T
    2. Waldman YY
    3. Kupiec M
    4. Ruppin E

    (2010) Translation efficiency is determined by both codon bias and folding energy

    PNAS 107:3645–3650.

    https://doi.org/10.1073/pnas.0909910107

    • PubMed
    • Google Scholar
73. 1. Ullu E
    2. Matthews KR
    3. Tschudi C

    (1993) Temporal order of RNA-processing reactions in trypanosomes: rapid trans splicing precedes polyadenylation of newly synthesized tubulin transcripts

    Molecular and Cellular Biology 13:720–725.

    https://doi.org/10.1128/MCB.13.1.720

    • PubMed
    • Google Scholar
74. 1. Van der Ploeg LH

    (1986) Discontinuous transcription and splicing in trypanosomes

    Cell 47:479–480.

    https://doi.org/10.1016/0092-8674(86)90608-2

    • PubMed
    • Google Scholar
75. 1. Vasquez JJ
    2. Hon CC
    3. Vanselow JT
    4. Schlosser A
    5. Siegel TN

    (2014) Comparative ribosome profiling reveals extensive translational complexity in different Trypanosoma brucei life cycle stages

    Nucleic Acids Research 42:3623–3637.

    https://doi.org/10.1093/nar/gkt1386

    • PubMed
    • Google Scholar
76. 1. Walrad P
    2. Paterou A
    3. Acosta-Serrano A
    4. Matthews KR

    (2009) Differential trypanosome surface coat regulation by a CCCH protein that co-associates with procyclin mRNA cis-elements

    PLoS Pathogens 5:e1000317.

    https://doi.org/10.1371/journal.ppat.1000317

    • PubMed
    • Google Scholar
77. 1. Walrad PB
    2. Capewell P
    3. Fenn K
    4. Matthews KR

    (2012) The post-transcriptional trans-acting regulator, TbZFP3, co-ordinates transmission-stage enriched mRNAs in Trypanosoma brucei

    Nucleic Acids Research 40:2869–2883.

    https://doi.org/10.1093/nar/gkr1106

    • PubMed
    • Google Scholar
78. 1. Webb H
    2. Burns R
    3. Ellis L
    4. Kimblin N
    5. Carrington M

    (2005b) Developmentally regulated instability of the GPI-PLC mRNA is dependent on a short-lived protein factor

    Nucleic Acids Research 33:1503–1512.

    https://doi.org/10.1093/nar/gki298

    • PubMed
    • Google Scholar
79. 1. Webb H
    2. Burns R
    3. Kimblin N
    4. Ellis L
    5. Carrington M

    (2005a) A novel strategy to identify the location of necessary and sufficient cis-acting regulatory mRNA elements in trypanosomes

    RNA 11:1108–1116.

    https://doi.org/10.1261/rna.2510505

    • PubMed
    • Google Scholar
80. 1. Wedel C
    2. Förstner KU
    3. Derr R
    4. Siegel TN

    (2017) GT‐rich promoters can drive RNA pol II transcription and deposition of H2A.Z in African trypanosomes

    The EMBO Journal 36:e201695323.

    https://doi.org/10.15252/embj.201695323

    • Google Scholar
81. 1. Worthey EA
    2. Martinez-Calvillo S
    3. Schnaufer A
    4. Aggarwal G
    5. Cawthra J
    6. Fazelinia G
    7. Fong C
    8. Fu G
    9. Hassebrock M
    10. Hixson G
    11. Ivens AC
    12. Kiser P
    13. Marsolini F
    14. Rickel E
    15. Rickell E
    16. Salavati R
    17. Sisk E
    18. Sunkin SM
    19. Stuart KD
    20. Myler PJ

    (2003) Leishmania major chromosome 3 contains two long convergent polycistronic gene clusters separated by a tRNA gene

    Nucleic Acids Research 31:4201–4210.

    https://doi.org/10.1093/nar/gkg469

    • PubMed
    • Google Scholar
82. 1. Zíková A
    2. Panigrahi AK
    3. Dalley RA
    4. Acestor N
    5. Anupama A
    6. Ogata Y
    7. Myler PJ
    8. Stuart K

    (2008) Trypanosoma brucei mitochondrial ribosomes: affinity purification and component identification by mass spectrometry

    Molecular & Cellular Proteomics 7:1286–1296.

    https://doi.org/10.1074/mcp.M700490-MCP200

    • PubMed
    • Google Scholar

Author details

1. Janaina de Freitas Nascimento

   Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom

   Contribution

   Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Writing—original draft

   Contributed equally with

   Steven Kelly

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1757-3441

2. Steven Kelly

   Department of Plant Sciences, University of Oxford, Oxford, United Kingdom

   Contribution

   Conceptualization, Software, Formal analysis, Funding acquisition, Investigation, Methodology, Writing—review and editing

   Contributed equally with

   Janaina de Freitas Nascimento

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8583-5362

3. Jack Sunter

   Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom

   Present address

   Department of Biological and Medical Sciences, Oxford Brookes University, Oxford, United Kingdom

   Contribution

   Conceptualization, Investigation, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2836-9622

4. Mark Carrington

   Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom

   Contribution

   Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Project administration, Writing—review and editing

   For correspondence

   mc115@cam.ac.uk

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-6435-7266

• 2,584

  views

• 389

  downloads

• 52

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.