Developmentally regulated instability of the GPI-PLC mRNA is dependent on a short-lived protein factor.

The expression of the vast majority of protein coding genes in trypanosomes is regulated exclusively at the post-transcriptional level. Developmentally regulated mRNAs that vary in levels of expression have provided an insight into one mechanism of regulation; a decrease in abundance is due to a shortened mRNA half-life. The decrease in half-life involves cis-acting elements in the 3' untranslated region of the mRNA. The trans-acting factors necessary for the increased rate of degradation remain uncharacterized. The GPI-PLC gene in Trypanosoma brucei encodes a phospholipase C expressed in mammalian bloodstream form, but not in the insect procyclic form. Here, it is reported that the differential expression of the GPI-PLC mRNA also results from a 10-fold difference in half-life. Second, the instability of the GPI-PLC mRNA in procyclic forms can be reversed by the inhibition of protein synthesis. Third, specifically blocking the translation of the GPI-PLC mRNA in procyclic forms by the inclusion of a hairpin in the 5' untranslated region does not result in stabilization of the mRNA. Thus, the effect of protein synthesis inhibitors in stabilizing the GPI-PLC mRNA operates in trans through a short-lived factor dependent on protein synthesis.

INTRODUCTION

Protein coding genes in Trypanosoma brucei and other kinetoplastid protozoa are arranged in arrays (1,2) and tens of genes are co-transcribed from distant promoters (3–7). Individual mRNAs are monocistronic and are co-transcriptionally processed from a pre-mRNA by two steps, a trans-splicing reaction that adds a capped 39 base mini-exon (8) to the 5′ end of the mRNA and a cleavage and polyadenylation reaction at the 3′ end. The trans-splicing is functionally linked to the polyadenylation of the 5′ flanking mRNA (9–11). The trans-splicing reaction is mechanistically similar to cis-splicing (12) and orthologues of many of the factors required for polyadenylation in crown group eukaryotes are present in the genomes of kinetoplastids (13), but the canonical AAUAAA motif for the site of polyadenylation is not present.

With some exceptions, there does not appear to be extensive clustering of genes into operon-like arrays of obviously functionally linked genes. In contrast, mRNAs derived from adjacent genes are often present at different levels and can show different developmental expression patterns [see (14) for an example]. In trypanosomes, there is regulation of transcription of the two dominant developmentally regulated cell surface proteins, the variant surface glycoprotein and procyclin (15–18), but these are transcribed by RNA polymerase I and represent a special case (3,19–21). The vast majority of the protein coding genes are transcribed by RNA polymerase II and there is no definitive evidence for transcriptional control of these genes (22). If adjacent genes are transcribed at the same rate, then the regulation of individual mRNA levels must be post-transcriptional. The mechanism of post-transcriptional regulation that has attracted most investigation is mRNA half-life and especially the involvement of regulation of half-life in differentially expressed mRNAs that are more abundant in one life cycle stage than in another (22). In trypanosomes, there are two life cycle stages amenable to experimental manipulation: the bloodstream form trypanosome, normally resident in a mammal, and the procyclic form present in the insect midgut. The bloodstream form differentiates into the procyclic form in the insect and the process can be reproduced in culture (23). Most effort has concentrated on mRNAs that are more abundant in procyclic forms and the identification of the cis-acting sequences necessary for instability in bloodstream forms (24–26). This work has identified U-rich elements (UREs) within the 3′UTR of three mRNAs that are required for instability of these mRNAs in bloodstream forms (27,28). The pathways by which unstable mRNAs are degraded have received less attention. There is evidence for involvement of the exosome (29), a complex of 3′ to 5′ exonucleases (30), and other nucleases as both 5′ to 3′ and 3′ to 5′ exonuclease intermediates have been observed (31). The most striking inhibition of degradation of unstable mRNAs occurred after inhibition of protein synthesis; procyclic form-specific EP procyclin mRNA in bloodstream form trypanosomes (32–34) and logarithmic phase-specific GP63 mRNA in stationary phase Leishmania chagasi promastigotes (35).

The GPI-PLC gene encodes a non-essential glycosylphosphatidylinositol-specific phospholipase C expressed in bloodstream form, but not procyclic form trypanosomes (36–38). The mRNA is moderately abundant in bloodstream forms, but almost undetectable in procyclic forms (37,39), and there is no difference in the transcription rate in the two life cycle stages (39). In this paper, the differential expression of GPI-PLC mRNA in the two life cycle stages is investigated and it is shown that the developmental regulation of steady-state mRNA levels is underpinned by a 10-fold reduced half-life in procyclic forms. Inhibition of protein synthesis causes a rapid stabilization of GPI-PLC mRNA in procyclic forms resulting from an increase in mRNA half-life. However, specifically blocking the translation of the GPI-PLC mRNA in procyclic forms does not lead to stabilization of the mRNA. Thus, it is likely that the stabilization of the GPI-PLC mRNA caused by inhibition of protein synthesis results from the dissipation of a polypeptide with a very short half-life that is necessary for the degradation of the unstable GPI-PLC mRNA in procyclic forms.

MATERIALS AND METHODS

Trypanosomes

The stock Lister 427 was used throughout (40). Bloodstream forms expressing either VSG MITat 1.5 (118) or MITat1.2 (221) were grown in culture in HMI-9 medium (41) containing 10% foetal bovine serum. In vitro differentiation of bloodstream forms to procyclic forms was performed using DTM with 15% foetal bovine serum containing 3 mM citrate and 3 mM cis-aconitate at 27°C (23). Procyclic forms derived in vitro from bloodstream forms were grown in DTM containing 15% foetal bovine serum. Established procyclic form cell lines (Lister 427) were grown in SDM-79 containing 10% foetal bovine serum (42). For the tetracycline-inducible expression of the GPI-PLC gene, Lister 427 procyclic forms with the plasmids pLEW29 and pLEW13 integrated into the tubulin locus (427 29-13) were used (43) (a gift from Professor George Cross, Rockerfeller University). Sinefungin was used at a final concentration of 2 μg/ml and cycloheximide at 50 μg/ml, this latter concentration is sufficient for maximal inhibition of [35S]-methionine incorporation into trichloroacetic acid precipitable material (44).

Transgenic trypanosomes

The over-expression of GPI-PLC mRNA in procyclics was achieved using a plasmid derived from pLEW100 (43). First, the EP procyclin 5′UTR and luciferase gene were removed using a complete BamHI and partial Xma1 digest and replaced with a polylinker containing BstBI and HindIII sites to produce p1865. Second, a ClaI HindIII fragment containing all of the GPI-PLC gene (37) plus processing signals for the flanking genes (Figure 4a) was cloned into the BstBI and HindIII sites of p1865 to produce p1885. The plasmid p1885 was linearized with NotI and electroporated into Lister 427 29-13 procyclic forms and integrants selected using zeocin.

Figure 4

(a) Diagram showing the GPI-PLC genomic locus which is flanked by the HSP100 and β′-COP genes (49) and the GPI-PLC gene replacement strategy. The two major mRNAs are shown above the GPI-PLC gene with the mini-exon addition sites indicated by filled circles and the approximate location of the polyadenylation site by a vertical line. The ClaI HindIII restriction enzyme fragment cloned into the expression vector p1865 is indicated. The starting cells are GPI-PLC −/− and targeted insertion of a neo gene results in the return of a GPI-PLC gene to the endogenous locus. (b) Sequence detail around the initiation codon of the GPI-PLC showing the location and sequence of the hairpin introduced into the 5′UTR and the location of the premature stop codon.

GPI-PLC −/− trypanosomes were generated from bloodstream form trypanosomes expressing VSG MITat 1.5 using constructs described previously (38) except that a bleomycin-resistance gene was used in place of a neomycin-resistance gene. The GPI-PLC gene was returned to the endogenous locus using the construct shown in Figure 4b. The neomycin-resistance cassette contained the EP1 procyclin mini-exon addition signal and 5′UTR, and the β-tubulin 3′UTR and polyadenylation site. It was located 400 bp to the 3′ side of the putative GPI-PLC polyadenylation site.

Cell harvesting

For most experiments, cells—typically 10 ml of procyclic form culture at 5 × 106 cells/ml—were harvested by centrifugation at 1200 g for 7 min, the pellet was resuspended in 1 ml phosphate-buffered saline, transferred to an Eppendorf tube and cells recovered by centrifugation at 8000 g for 30 s, the supernatant was removed and the cell pellet lysed in the appropriate buffer from the RNA preparation kit (see below). This procedure took between 9 and 10 min from culture to lysis. For the determination of GPI-PLC mRNA half-life in procyclic cells (Figure 1c), cells were harvested by centrifugation at 6000 g for 1 min, the supernatant was aspirated away and the cell pellet lysed in the appropriate buffer from the kit. This procedure took ∼3 min from culture to lysis.

Figure 1

Estimation of GPI-PLC mRNA half-life following inhibition of mRNA maturation with sinefungin. Northern blots showing decay of GPI-PLC mRNA after sinefungin addition in (a) bloodstream forms and (b) procyclic forms and (c) transgenic procyclic forms expressing a GPI-PLC gene from a strong promoter. In each case, the same blot was probed with α- and β-tubulin to show the sinefungin-dependent appearance of dicistronic tubulin mRNAs and then for an rRNA to estimate comparative loading. The probe is indicated under each blot. The inset in panel (b) shows a five times longer exposure of region of the blots containing the GPI-PLC mRNA. (d) The GPI-PLC mRNA and rRNA were quantitated in (a) and (c) using a phosphorimager and the GPI-PLC mRNA determined as a concentration relative to the amount of rRNA. RNA size standards are shown in panels (a) and (b).

For western blot analysis, cells were washed twice with 0.5 culture volumes of phosphate-buffered saline prior to lysis in SDS–PAGE sample buffer (45).

RNA and protein analysis

RNA was prepared using the RNAeasy kit from Qiagen and analysed using vertical agarose gels after denaturing the RNA with glyoxal (46). Northern blotting was as described in (47). After hybridization, all blots were washed in 15 mM sodium chloride 1.5 mM trisodium citrate (0.1× SSC), 0.1% sodium dodecyl sulphate at 60°C. Routinely, blots were probed sequentially for GPI-PLC, then α- and β-tubulin sometimes including EP procyclin at the same time, then rRNA. All time courses were performed twice and gave the same results within the accuracy of the phosphorimager measurements (see below).

SDS–PAGE (45) and western blots (48) were performed using standard procedures.

Phosphorimager analysis

Data were collected using a Molecular Dynamics Storm 840 and analysed using the accompanying ImageQuant software. Individual bands were measured and background subtracted. Loading differences were measured using the signal from an rRNA overprobe of the same blot and the mRNA signal adjusted for loading. The error in the measurement procedure was estimated by comparing values for the same sample loaded two or three times on the same gel. The range of values fell within ±15% of the mean value.

RESULTS

A difference in mRNA half-life underlies the developmentally regulated expression of the GPI-PLC gene

Expression of the GPI-PLC gene is developmentally regulated, the mRNA is readily detected in bloodstream form trypanosomes, but is barely detectable in procyclic forms (37). As the GPI-PLC gene is located in a presumed polycistronic transcription unit on chromosome 2 (2) and is flanked by two constitutively expressed genes HSP100 and β'-COP (49,50), it appeared likely that the developmental regulation of the GPI-PLC mRNA is post-transcriptional. To test this, the half-life of the GPI-PLC mRNA in the two developmental stages was estimated by determining the rate of decay after blocking mRNA maturation using sinefungin, which blocks methylation of the 5′ cap on the mini-exon donor RNA, thus inhibiting trans-splicing which in turn blocks polyadenylation (9,11,51). Over the course of the experiments, the sinefungin had no effect on cell motility and the cells retained normal morphology as judged by phase contrast microscopy. One effect of sinefungin was to cause an immediate accumulation of incompletely processed tubulin mRNAs as previously described (11), most visible as dicistronic RNAs on the northern blot (Figure 1a). The use of sinefungin to follow mRNA decay was used in preference to an inhibitor of transcription, such as actinomycin D, as the point of action in blocking mRNA maturation removes any need to consider the fate of partially processed transcripts. The effect of sinefungin was very rapid and allowed the measurement of mRNA half-lives of as little as 3 min (see below).

The major form of the GPI-PLC mRNA detected on northern blots was 3.6 kb (Figure 1a). The half-life of the 3.6 kb GPI-PLC mRNA was 30 ± 5 min in bloodstream forms (Figure 1d). In procyclic forms, the GPI-PLC mRNA was detected only after a long exposure of the northern blot and was absent at the 10 min sinefungin time point suggesting a very short half-life (Figure 1b). The low steady-state levels in procyclic forms made an estimate of the half-life technically difficult, so a procyclic cell line with a GPI-PLC transgene expressed from the strong EP procyclin promoter (43) was produced with the aim of increasing the steady-state level of the GPI-PLC mRNA at the beginning of the time course. In addition, the cell harvesting procedure was altered so that cells were lysed in 3 min from culture instead of the 10 min taken in the previous experiments. The over-expression strategy was successful and permitted the estimation of half-life after addition of sinefungin (Figure 1c). The half-life of the GPI-PLC mRNA was 3 ± 1 min (Figure 1d). Thus, the half-life of GPI-PLC mRNA is approximately 10-fold greater in bloodstream forms than in procyclics.

Inhibition of protein synthesis in procyclic forms leads to a rapid accumulation of GPI-PLC mRNA due to an increase in half-life of the mRNA

Inhibition of protein synthesis in bloodstream form trypanosomes leads to a rapid accumulation of the normally unstable EP procyclin mRNA (32–34). A similar effect was observed when procyclic forms were treated with cycloheximide and the steady-state levels of GPI-PLC mRNA determined (Figure 2). There was readily detectable accumulation 30 min after cycloheximide addition, and the mRNA reached a steady-state level after 2 h. A similar stabilization of the GPI-PLC mRNA was also observed with anisomycin, another protein synthesis inhibitor (data not shown). The effect of protein synthesis inhibitors on the GPI-PLC mRNA was not a general phenomenon; over the time course of cycloheximide treatment, the levels of α- and β-tubulin mRNAs increased but less than 2-fold whereas the levels of EP procyclin mRNA decreased (Figure 2).

Figure 2

Inhibition of protein synthesis results in the appearance of GPI-PLC mRNA. Northern blot showing a time course after the addition of cycloheximide. The same blot was probed for EP procyclin and α- and β-tubulin to demonstrate that mRNA stabilization is not a general consequence of blocking translation and for an rRNA to estimate comparative loading. The probe is indicated under each blot. The GPI-PLC mRNA and rRNA were quantitated using a phosphorimager, and the GPI-PLC mRNA determined as a concentration relative to the amount of rRNA and is expressed in arbitrary units.

The data from this experiment were also used to estimate the ratio between the bloodstream and procyclic form steady-state levels of GPI-PLC mRNA. The result, corrected for rRNA as a measure of loading, was that the GPI-PLC mRNA is at 50-fold higher levels in bloodstream than in procyclic forms. After cycloheximide treatment, the difference was reduced to just over 2-fold.

The half-life of GPI-PLC mRNA after inhibition of protein synthesis treatment by incubation with cycloheximide for 2 h was determined by addition of sinefungin and preparation of RNA over an 80 min time course (Figure 3). At the end of the time course, the cells had retained motility and appeared morphologically normal by phase contrast microscopy. The sinefungin resulted in an accumulation of incompletely processed tubulin mRNAs, again most visible as the dicistron on a northern blot (Figure 3). The rate of accumulation of the tubulin dicistron was similar to procyclic cells treated with sinefungin without prior cycloheximide treatment (Figure 1b) suggesting that the 2 h treatment with cycloheximide had little effect on transcription. After addition of sinefungin, the half-life of the GPI-PLC mRNA was 30 ± 5 min. Thus, the inhibition of protein synthesis had resulted in a 10-fold increase in the half-life of the GPI-PLC mRNA in procyclic forms.

Figure 3

Estimation of the half-life of cycloheximide-induced GPI-PLC mRNA in procyclic forms. Cells were incubated with cycloheximide for 2 h, and then sinefungin added and samples were taken over a time course and analysed by northern blotting. After probing for GPI-PLC, the same blot was probed with α- and β-tubulin to show the sinefungin-dependent appearance of dicistronic tubulin mRNAs and then for an rRNA to estimate comparative loading. The probe is indicated under each blot. The predominant GPI-PLC mRNA indicated with an arrow and rRNA were quantitated using a phosphorimager and the GPI-PLC mRNA determined as a concentration relative to the amount of rRNA. RNA size standards are shown.

Blocking the translation of the GPI-PLC mRNA in procyclic forms does not result in stabilization

The GPI-PLC gene is non-essential (38) and bloodstream trypanosomes derived from the Lister 427 cell line with both copies of the GPI-PLC gene deleted proliferate at the same rate in culture as wild-type cells (52) (H. Webb and M. Carrington, unpublished). The strategy used to specifically block translation of the GPI-PLC mRNA was to reintroduce versions of the GPI-PLC gene that could not be translated into the endogenous locus of GPI-PLC null bloodstream trypanosomes (Figure 4). The bloodstream trypanosomes were then differentiated to procyclic forms in vitro and the steady-state level of the GPI-PLC mRNA was compared in the two life cycle stages.

To make the GPI-PLC null cell line, the entire GPI-PLC mRNA coding sequence, including 5′ and 3′UTR, was replaced with an antibiotic-resistance cassette. The two alleles were replaced sequentially, first using a hygromycin and then a bleomycin-resistance gene (Figure 4). Once the GPI-PLC deletants had been verified, a range of wild-type and modified GPI-PLC genes were returned to the locus (Figure 4b). The re-introduction construct returned the entire GPI-PLC mRNA coding sequence, restored the intergenic sequence to the 5′ side of the GPI-PLC gene and introduced a neomycin-resistance cassette in the intergenic sequence between the GPI-PLC gene and the downstream gene.

Three GPI-PLC genes were returned to the endogenous locus: (i) a wild-type GPI-PLC gene as a control for expression; (ii) a GPI-PLC gene with a hairpin in the 5′UTR; (iii) a GPI-PLC gene containing a stop codon instead of codon 15 to assay for nonsense-mediated decay of the GPI-PLC mRNA. The hairpin added to the 5′UTR contained 23 bp (Figure 4), and was based on constructs used in other systems that have been shown to be sufficient to block translation (53).

The level of expression of GPI-PLC protein was determined by western blotting bloodstream form clones containing a transgene (Figure 5). A titration of wild-type cell lysate was used to calibrate the sensitivity of the assay and on a long exposure of the western blot, it was possible to detect GPI-PLC when 1/80 of the cell equivalents of wild-type cells (+/+) were loaded indicating that the assay was sensitive down to ∼1% of wild-type levels. Expression from the wild-type GPI-PLC transgene (−/−::GPI-PLC) was readily detected, but no expression was detected from either the GPI-PLC gene containing a hairpin in the 5′UTR (−/−::GPI-PLC hairpin) or, as expected, from the GPI-PLC gene containing a premature stop codon (−/−::GPI-PLC early stop) (Figure 5). These results demonstrate that the hairpin in the 5′UTR of the GPI-PLC mRNA blocked translation.

Figure 5

Expression of GPI-PLC protein from GPI-PLC transgenes returned to the endogenous locus. The expression of GPI-PLC was tested in the parental GPI-PLC −/− cell line and two independent clones of cell lines containing transgenes encoding wild-type GPI-PLC or GPI-PLC with a hairpin in the 5′UTR or GPI-PLC with a stop codon replacing codon 15. A titration of 2 × 106 (1 cell equivalent) to 2.5 × 104 bloodstream form cells (1/80 cell equivalent) is shown as a measure of sensitivity. The antiserum used is indicated under each blot, anti-ISG65 was used as a loading control.

Each of the bloodstream trypanosome clones was differentiated to procyclic forms in vitro and RNA prepared from both life cycle stages for each clone. Two distinct clones containing the GPI-PLC transgene with a 5′UTR hairpin and two containing a GPI-PLC transgene with a premature stop codon were analysed. The level of the GPI-PLC mRNA was determined by northern blotting (Figure 6).

Figure 6

Comparison of GPI-PLC expression from transgenes. RNA from both bloodstream (B) and procyclic (P) forms of the same cell lines was analysed by northern blotting. The probe is indicated under each blot. One clone containing a wild-type GPI-PLC transgene was compared with two independent clones of each of two cell lines containing either GPI-PLC with a hairpin in the 5′UTR or GPI-PLC with a stop codon replacing codon 15. GPI-PLC +/+ and +/− bloodstream form RNA is shown for comparison, rRNA was used as a loading control and RNA size standards are shown.

GPI-PLC mRNA was detected in all bloodstream form clones containing a transgene. In all cases, the mRNA was a similar size to that from GPI-PLC +/+ cells and had the characteristic pattern of two distinct mRNAs indicating that the transcript was correctly processed. The steady-state level of mRNA in the clone containing the wild-type GPI-PLC transgene (−/−::GPI-PLC) was less than in a heterozygote (+/−). The only difference between the transgene and the endogenous locus is the antibiotic-resistance cassette to the 3′ side of GPI-PLC gene, which may affect the efficiency of polyadenylation of the GPI-PLC mRNA. The steady-state level of GPI-PLC mRNA in bloodstream forms with the transgene containing a premature stop codon (−/−::GPI-PLC early stop) was similar to that from the wild-type transgene indicating that there is little or no nonsense-mediated decay of the GPI-PLC mRNA. There were increased levels of the GPI-PLC mRNA in bloodstream trypanosomes containing a transgene with a hairpin in the 5′UTR (−/−::GPI-PLC hairpin) presumably due to interference with the normal mRNA turnover pathway.

None of the transgenes produced a significant steady-state level of GPI-PLC mRNA in the procyclic forms indicating that the developmental regulation of GPI-PLC mRNA was still operating for all the transgenes. Thus, specifically blocking the translation of the GPI-PLC mRNA does not lead to stabilization of the mRNA in procyclic forms.

DISCUSSION

The regulation of mRNA stability is the major level of regulation of gene expression in trypanosomes. Developmentally regulated mRNAs, such as GPI-PLC which is moderately abundant in bloodstream trypanosomes, but almost absent in procyclic forms, provide an opportunity to investigate the regulation of stability. The major findings of this paper are that (i) the half-life of the GPI-PLC mRNA is approximately 10-fold greater in bloodstream forms than in procyclic forms; (ii) the inhibition of protein synthesis causes a rapid stabilization of GPI-PLC mRNA in procyclic forms; (iii) the stabilization of the GPI-PLC mRNA in procyclic forms after inhibition of protein synthesis results from an approximately 10-fold increase in mRNA half-life; (iv) the inclusion of a hairpin in the 5′UTR of the GPI-PLC mRNA is sufficient to block translation, but does not stabilize the GPI-PLC mRNA in procyclics; (v) a GPI-PLC mRNA with a stop codon in place of codon 15 is not subject to nonsense-mediated decay but is not stabilized in procyclic forms.

The steady-state amount of any mRNA in a particular cell will be determined by the rate of synthesis and the rate of decay. Phosphorimager measurements of the amount of any mRNA are standardized against rRNA to allow for differences in the amount of RNA loaded in each gel. This step effectively converts the estimate of mRNA amount into a concentration, and thus the measurements of steady-state levels are affected by the cell volume and ribosome density. In trypanosomes, the relative abundance of the same mRNA in the two life cycle stages will be affected by: (i) relative transcription rate, which is probably higher in bloodstream forms grown at 37°C than in procyclic forms grown at 27°C; (ii) cell volume, procyclic forms are roughly twice the volume of bloodstream forms (iii) ribosome density, no comparison of ribosomal content of the two cell types has been reported and (iv) mRNA half-life. The considerations other than half-life become apparent when a constitutive expressed mRNA is analysed. The actin mRNA is equally abundant in bloodstream and procyclic forms (54), but the half-life in bloodstream forms (31) is several-fold shorter than in procyclic forms (27). Presumably, the difference in half-life is necessary to compensate for the higher transcription rate and smaller cell volume of bloodstream forms.

The steady-state level of GPI-PLC mRNA is 50-fold higher in bloodstream than in procyclic forms. It is probable that this difference in steady-state levels can be accounted for by a half-life that is 10-fold greater in bloodstream forms combined with the effect of the higher transcription rate in and smaller volume of bloodstream forms, both factors amplifying the difference in steady-state levels. There is an interesting problem to be addressed in the future: why the low level of GPI-PLC mRNA in procyclic forms produces no detectable GPI-PLC activity (36). The possible explanations are that there is translational control of the mRNA in procyclic forms and/or the GPI-PLC protein has a short half-life.

At least two GPI-PLC mRNAs were usually detected on a northern blot; the predominant band of 3.6 kb corresponds to the mature mRNA, whereas the precise origin of all of the larger GPI-PLC mRNAs is less clear. All originate from the single GPI-PLC gene present in heterozygotes (Figure 6) and so do not represent allelic variants. One of the larger GPI-PLC mRNAs results from the use of an alternative trans-splicing site in the 5′UTR and can be reproduced by transferring the sequence to a reporter gene (H. Webb and M. Carrington, submitted). It is not clear whether the different GPI-PLC mRNAs represent different mature mRNAs resulting from the use of alternative trans-splicing and/or polyadenylation sites, or whether the larger mRNAs are partially processed precursors that are yet to complete maturation to the final GPI-PLC mRNA. After sinefungin treatment, the turnover rate of the different GPI-PLC mRNAs is not the same (Figures 1a and 3) and the major, 3.6 kb, mRNA is turned over more rapidly than the larger forms, either it has a longer half-life or the larger mRNA are precursors of the 3.6 kb mRNA.

The GPI-PLC mRNA is unstable in procyclic forms with a half-life of 3 min. Inhibition of protein synthesis increases the half-life to 30 min. Previously, it had been shown that inhibition of protein synthesis in bloodstream forms led to accumulation of the normally unstable EP procyclin mRNA (32–34) as fully mature and polysome-associated mRNA (33). The GPI-PLC mRNA that accumulates after protein synthesis inhibitor treatment appears to be fully processed, judged by mobility after northern blotting. The stabilization of GPI-PLC mRNA after treatment with cycloheximide indicated that the normal rapid turnover of the mRNA is translation dependent. The two possibilities were that the translation of the GPI-PLC mRNA itself is necessary for the rapid turnover or that translation of another mRNA is necessary and that the polypeptide product is effective for only a very short period. To distinguish between these two possibilities the translation of the GPI-PLC mRNA was blocked in two ways: the insertion of a hairpin in the 5′UTR or the inclusion of a premature stop codon. The production of an mRNA from each of the transgenes was verified in bloodstream forms, all produced readily detectable mRNA but only the wild-type transgene produced detectable GPI-PLC protein. The cell lines were differentiated to procyclic forms and the expression from the transgenes determined. This step contained the assumption that the GPI-PLC mRNA with a hairpin in the 5′UTR was also blocked for translation in procyclic forms, but it is unlikely that this would be affected by developmental stage.

In many eukaryotes, mRNAs that are not translated are degraded by the nonsense-mediated decay pathway (55), but this pathway did not appear to act on the GPI-PLC mRNA containing a premature stop codon. Neither the inclusion of the premature stop codon, which allowed translation as far as codon 14, nor a hairpin in the 5′UTR designed to prevent ribosome assembly on the initiation codon stabilized the GPI-PLC mRNA in procyclic forms. Thus, the rapid turnover of the GPI-PLC mRNA in procyclic forms is dependent on the translation of an mRNA other than GPI-PLC mRNA itself and the cycloheximide is acting in trans.

Only one other example of dependency of the rapid turnover of an mRNA on the translation of a trans-acting factor has been unambiguously demonstrated. In mammalian cells, the transferrin receptor (TfR) mRNA is stable when cells are depleted of iron but degraded when iron is plentiful. The rapid turnover of the TfR mRNA that occurs in the presence of iron can be blocked by the addition of cycloheximide (56). The action of the cycloheximide is to block the turnover of the trans-acting factor IRP-2, which along with a second trans-acting factor, IRP-1, probably stabilizes the TfR mRNA by binding to iron responsive elements within the 3′UTR. The labile factor affected by cycloheximide could be necessary for targeting IRP-2 to the proteosome as treatment with the proteosome inhibitor MG132 also stabilized the TfR mRNA (56).

In other examples, where rapid mRNA turnover can be blocked by protein synthesis inhibitors, it is the translation of the mRNA that is necessary for turnover. In the case of β-tubulin mRNA, a polysome-associated nuclease is involved (57). The c-fos mRNA is unstable and the turnover regulated by AU-rich elements (AREs) in the 3′UTR and by protein coding region determinants. Blocking translation of the major coding region determinant by the inclusion of a stem loop stabilizes the mRNA (58,59). In yeast, the turnover pathway followed by mRNAs containing AREs involves a shortening in poly A tail length followed by removal of the 5′ cap and subsequent hydrolysis by a 5′ to 3′ exonuclease. Trans-acting factors bound to AREs are believed to modulate the rate at which deadenylation occurs though whether translation of the mRNA itself is necessary for deadenylation is not clear [reviewed in (60) and (61)]. The U-rich elements identified in the 3′UTRs of unstable mRNAs in trypanosomes are believed to function in a similar manner to AREs (22,28), but at this stage it is not known whether any cis-acting sequences in the GPI-PLC mRNA are similar to U-rich elements.

In conclusion, it has been shown that the developmentally regulated expression of the GPI-PLC gene in trypanosomes can be accounted for by the regulation of mRNA half-life and that the rapid turnover of the GPI-PLC mRNA in procyclic form trypanosomes is dependent on the continuous translation of a trans-acting factor. The dependency of the rapid turnover of the GPI-PLC mRNA in trypanosomes and TfR mRNA in mammalian cells on a labile factor suggests the presence of an evolutionarily conserved pathway and a better understanding of the pathway in the future will come from the identification of its components.