Developmental Regulation of Heat Shock Protein 83 inLeishmania

Developmental gene regulation in trypanosomatids proceeds exclusively by post-transcriptional mechanisms. Stability and abundance of heat shock protein (HSP)70 and HSP83 transcripts in Leishmania increase at mammalian-like temperatures, and their translation is enhanced. Here we report that the 3′-untranslated region (UTR) of HSP83 (886 nucleotides) confers the temperature-dependent pattern of regulation on a chloramphenicol acetyltransferase (CAT) reporter transcript. We also show that the majority of the 3′-UTR sequences are required for increasing mRNA stability during heat shock. Processing of the HSP70 and HSP83 primary transcripts to poly(A)+ mRNA was more efficient during heat shock; therefore, even when stability at 33 °C was reduced by deletions in the 3′-UTR, transcripts still accumulated to comparable and even higher levels. Translation of heat shock transcripts in Leishmania increases dramatically upon temperature elevation. Unlike in other eukaryotes in which the 5′-UTR confers preferential translation on heat shock transcripts, we show that translational control of HSP83 in Leishmaniaoriginates from its 3′-UTR. The 5′-UTR alone cannot induce translation during heat shock, but it has a minor contribution when combined with the HSP83 3′-UTR. We identified an element located between positions 201 and 472 of the 3′-UTR which is essential for increasing translation of the CAT-HSP83 reporter RNA at 33–37 °C. This region confers preferential translation during heat shock even in transcripts that were less stable. Thus, investigating the traditionally conserved heat shock response reveals that Leishmania parasites use unique pathways for translational control.

Unique and unusual features characterize the genomes of trypanosomatids that comprise an ancient group of eukaryotes. Genes transcribed by RNA polymerase II are found in polycistronic transcription units, and gene clusters repeated in tandem frequently encode abundant proteins. There is no evidence for transcriptional activation of developmentally regulated genes, and mRNA abundance is determined exclusively by post-transcriptional mechanisms. These include differential RNA processing (1) and control of mRNA decay (2)(3)(4). Polycis-tronic transcripts mature by trans-splicing that adds a short capped leader (39-mer) to the 5Ј-end of mRNAs and by 3Ј processing that includes cleavage and polyadenylation. Transsplicing and polyadenylation are coupled mechanistically in trypanosomatids and share regulatory signals that consist of polypyrimidine tracks and potential AG splice sites (5,6). The 3Ј-untranslated sequences lack the consensus eukaryote AAUAAA signal for cleavage and polyadenylation, and no other consensus element of that nature has been identified.
Leishmania parasites exist in the alimentary canal of female sand flies as flagellated promastigotes. They develop into virulent metacyclics that are uniquely adapted for transmission by the fly into a mammalian host where they differentiate into amastigotes, an obligate intracellular life form that resides within macrophages. Developmental gene expression is induced by environmental changes that are inflicted by the switch of hosts, including alterations in temperature (26 to 33-37°C) and extracellular pH (7 to 5.5) (4, 7). HSP70 1 and HSP83 are expressed constitutively throughout the life cycle of Leishmania. However, despite the absence of transcriptional activation (2), the abundance of these transcripts increased 3-4-fold in mammalian-like temperatures because of an increase in their stability. Moreover, translation of heat shock transcripts increases by at least 10-fold upon temperature elevation (8). Above a certain threshold of temperature which varies among different Leishmania species, synthesis of cellular proteins ceases (7).
Expression of heat shock genes in eukaryotes is regulated at a variety of levels that include transcription (9), RNA processing (10) followed by export to the cytoplasm, RNA stability, and translation. In view of the conserved and universal nature of the stress response it serves as an ideal system for studying the unusual features of molecular mechanisms used by trypanosomatids and for comparing them with other eukaryotes. HSP70 transcripts in Drosophila and in yeast are degraded rapidly during recovery from heat shock at 26°C (11,12), with deadenylation initiating this process (13). Translation of the human and Drosophila HSP70 increases dramatically during heat shock, and it is the 5Ј-UTR that confers this pattern of regulation in Drosophila (14,15). Little is known about the molecular components that control mRNA stability and its accumulation in trypanosomatids, although recent reports indicate the involvement of the 3Ј-UTR in the regulation of HSP70 in Leishmania (16) and in Trypanosoma brucei (17). Here we show for HSP83 in Leishmania that regulated RNA processing combined with differential decay plays a key role in determining mRNA abundance at elevated temperatures. Furthermore, the downstream intergenic region (IR) of HSP83 confers this pattern of regulation, and the 3Ј-UTR of HSP83 contains an element that induces preferential translation during heat shock.
Plasmid Construction-The IRs within the genomic HSP83 cluster extend from the termination codon of one gene to the first translated ATG of the following gene. Plasmid pHC included the CAT coding gene in pBluescript fused with a complete upstream HSP83 IR that provided the signals for trans-splicing of the CAT primary transcript. An HSP83 IR was then ligated downstream to the CAT gene providing the signals for 3Ј cleavage and polyadenylation, resulting in plasmid pHCH (formerly denoted pICI (2)). This gave a CAT coding gene that was flanked by HSP83 IRs at both ends. A series of constructs with deletions spanning the 3Ј-UTR was created by digestion of the 2-kb BglII-SalI genomic fragment (pKS83BS) that encodes the HSP83 intergenic region with different restriction enzymes. This fragment contained the complete IR (1,471 bp) flanked by short coding sequences that extended from position 1692 of one coding gene (SalI) to position 80 of the following coding gene in the cluster ((BglII; see Ref. 18). Because the downstream IR extended beyond the first translated ATG codon of the subsequent gene it included the complete HSP83 endogenous signals for 3Ј processing. The following sites were used to introduce the deletions described: SmaI (201), NruI (472), SphI (59, 655) and NarI (873). Partial digests were performed for SphI. The modified IRs were cloned downstream of the CAT gene in plasmid pHC, and the complete fused gene (HCH⌬ "deletion") was cloned as a BclI-BamHI insert into the BamHI site of pX (5), at an orientation opposite that of the Neo r gene. This ensured that only the HSP83 signals would direct processing of the CAT primary transcript. A parallel set of deletions was constructed by a similar cloning approach except that the 5Ј-UTR was derived from the ␣-tubulin gene. The modified HSP83 IR was cloned downstream of the CAT gene in pALT1 (19), resulting in the pTCH-based deletion plasmids.
Transfections-Plasmid DNA was electroporated into L. amazonensis parasites as described (19) except that a double electrical pulse of 5.5 kV/cm at 25 microfarads in a Bio-Rad Gene Pulser apparatus was applied. Stably transfected lines were selected in the presence of 60 g/ml Geneticin (G418, Sigma). Neomycin-resistant parasites appeared 10 -14 days after transfection and were grown in the presence of 200 g/ml Geneticin.
Isolation of RNA and Northern Analysis-Total RNA was extracted from cells incubated at 26°C and at different time points after their transfer to 33°C with the TRI reagent (Scientific Research Laboratories). To measure the T1/2 of specific mRNAs 10 g/ml actinomycin D was added to stop transcription, cells were distributed to 26 and 33°C, and RNA levels were sampled at four different time points. RNA was fractionated over 1% denaturing agarose gels, blotted, and hybridized with specific probes. The intensity of hybridization was determined by analysis in a Molecular Dynamics PhosphorImager. The probes used for hybridization were a 1.2-kb PstI fragment from the HSP83 coding gene of L. amazonensis (20), the CAT coding gene, and a rDNA probe of Leishmania major, 2 which was used to control RNA loads. The hybridization intensities of HSP83 and CAT were normalized according to the rRNA values. The T1/2 of CAT and HSP83 transcripts at 26 and 33°C was calculated from the corresponding decay curves that were obtained by nonlinear regression, and the 33/26 ratio between the T1/2 at each temperature was determined. A 33/26 ratio of T1/2 Ͼ 1 indicated that the transcript was more stable at 33°C.
Mapping of RNA Ends by 3Ј-RACE-To determine the site of poly(A) ϩ addition in the HSP83 transcript, cDNA was generated by reverse transcription of total RNA using a (dT) 18 primer attached to an adaptor sequence (5Ј-GACTCGAGTCGACATCGATT-3Ј). The cDNA was amplified by a set of primers derived from the adaptor sequence (reverse primer) and a gene-specific sequence (forward primer) that corresponded to positions 478 -507 (5Ј-GTCGATGCTGCCCGAGCT-CACCCCCACC-3Ј) in the HSP83 IR. The 3Ј-boundaries of the CAT transcripts were determined using the adaptor (reverse) primer and a forward primer corresponding to positions 595-615 of the CAT coding region (5Ј-CAGGTTCATCATGCCGTCTGT-3Ј). Amplification products were cloned into the TA cloning vector (Promega) and sequenced to determine the site of polyadenylation.
Pulse Field Gel Electrophoresis and Southern Analysis-Chromosomal DNA embedded in agarose blocks (10 7 parasites) was digested overnight with NcoI and separated by pulse field gel electrophoresis in a CHEF II DR system (Bio-Rad) on 1% agarose gels in 0.5 ϫ TBE, along with blocks of nondigested chromosomes. Running conditions included a gradient of pulses ranging from 1 to 6 min for 11 h at 6 V/cm. The gels were soaked in 0.25 M HCl, blotted, and hybridized with the 3.5-kb HSP83 SalI genomic fragment.
Cloning of the Last Gene Copy in the HSP83 Cluster-A cosmid library of L. amazonensis was the generous gift of D. Smith from Imperial College, London. The library was subjected to a double screen with probes derived from positions Ϫ3/ϩ80 in the HSP83 coding gene (5Ј-probe) and positions 1691 in the coding gene to 200 in the subsequent IR (3Ј-probe). A genomic clone that contained part of the last copy of the HSP83 cluster was expected to hybridize only with the 3Ј-probe and not with the corresponding 5Ј-probe. Clones that hybridized differentially with a 3Ј-probe and not with a 5Ј-probe were isolated. The first gene copy in the HSP83 genomic cluster was cloned previously using a similar approach (21).
Metabolic Labeling-Parasites were grown in RPMI with supplements to a cell density of 3 ϫ 10 7 /ml. Cells (1 ml) were preincubated at 26, 33, and 37°C for 1 h and then labeled with 20 Ci of [ 35 S]methionine/cysteine protein labeling mix (1,175 Ci/mmol) for 30 min at the corresponding temperatures. After labeling, the cells were harvested at 4°C, washed twice with cold phosphate-buffered saline, and lysed in SDS-polyacrylamide gel sample buffer. Incorporation of the 35 S-labeled amino acids was measured by precipitation with trichloroacetic acid. Samples of proteins containing the same amount of incorporated radiolabel corresponded to a similar number of cells (22) and were separated over 15% SDS-polyacrylamide gels. The gels were dried and processed further for fluorography.

RESULTS
The HSP83 Gene Cluster Contains 18 Gene Copies-To evaluate the size of the HSP83 genomic cluster, L. amazonensis chromosome blocks were digested with an enzyme that does not cut within the genomic cluster but cleaves at sites located in the immediate flanking sequences. NcoI sites are found 1.2 and 0.8 kb upstream and downstream from the first and last gene copies, respectively. Southern analysis of L. amazonensis pulse field gel electrophoresis blots revealed a single 67-kb fragment that hybridized with the HSP83 probe ( Fig. 1), confirming the presence of a single HSP83 genome cluster, mapped to chromosome 28 of L. major (23). Based on the size of the repeat unit (3,580 bp (20)), the 67-kb fragment contains 18 tandem repeats of the HSP83 gene.
The HSP83 3Ј-UTR Confers RNA Stability at Elevated Temperatures-The temperature-dependent abundance of the 2 C. L. Jaffe, unpublished data.
FIG. 1. Size estimation of the HSP83 genomic cluster by pulse field gel electrophoresis. L. amazonensis cells were embedded in agarose blocks, digested overnight with NcoI that cuts only in the flanks of the cluster, and separated by pulse field gel electrophoresis on 1% agarose gels in 0.5 ϫ TBE. Nondigested (N.D.) blocks were separated in parallel. Running conditions included a gradient of pulses ranging from 1 to 6 min for 11 h at 6 V/cm. Gel blots were hybridized with the HSP83 probe. A mixed digest of DNA was used as size markers.
HSP83 transcript is controlled post-transcriptionally, with RNA stability playing an important regulatory role. mRNA Abundance at 26°C and after incubation at 33°C for 4 h is depicted as histograms. Decay curves were obtained from RNA samples of cells incubated with actinomycin D at ambient and elevated temperatures, and the corresponding T1/2 at 33 and 26°C and the ratio between them were calculated. A 33/26 ratio Ͼ 1 indicated that the mRNA was more stable at 33°C. The endogenous HSP83 mRNA level increased 3-fold at 33°C, and the ratio between the T1/2 measured at both temperatures indicated that it was 3.4-fold more stable at elevated temperatures ( Fig. 2

and Ref. 2).
To show that the HSP83 IRs could confer a temperature-dependent pattern of regulation, the CAT gene was cloned between two HSP83 IRs, resulting in pHCH. To examine the individual role of the 5Ј-and 3Ј-UTRs in control of mRNA stability we placed the CAT reporter gene between IRs derived from either the HSP83 or ␣-tubulin gene clusters (24). This gave constructs in which the CAT gene was flanked by an upstream HSP83 IR and a downstream tubulin IR (pHCT) or by the upstream tubulin UTR and the downstream HSP83 UTR (pTCH). The IRs extended from the translational termination codon of one gene to the translational initiation of the downstream gene in the cluster, to maintain the 3Ј processing signals. Each of the fused genes was stably introduced into L. amazonensis cells via the pX transfection vector (25), and regulation of the CAT mRNA was investigated. When both the upstream and downstream IRs were derived from HSP83 (HCH), the pattern of regulation was similar to that of the endogenous gene. When the 3Ј-UTR was derived from HSP83 and the 5Ј-UTR was obtained from ␣-tubulin (TCH), the T1/2 increased 2.4-fold at 33°C, and the steady-state level at 4 h was 1.8-fold higher. When the 3Ј-UTR was derived from ␣-tubulin and the 5Ј-UTR was obtained from HSP83 (HCT), the T1/2 of the resulting transcript and its steady-state level decreased by 2-fold at 33°C (33/26 ϭ 0.5; Fig. 2). The presence of the HSP83 3Ј-UTR was therefore required for increasing the T1/2 of HSP83 during heat shock.
Most Deletions in the 3Ј-UTR Destabilize RNA, but Abundance of These Unstable Transcripts Does Not Decrease-To map regions that may control the differential decay of HSP83 mRNA at different temperatures, we constructed a series of IRs with deletions that spanned the 3Ј-UTR and cloned them downstream of the CAT reporter gene (Fig. 3). The deletions did not interfere with sequences downstream to the 3Ј-UTR which contained the signals for 3Ј processing.
Initially we created small deletions that removed sequences 1-201, 201-473, 473-655, and 655-873 of the 3Ј-UTR. To examine whether the 5Ј-UTR was involved in some way in the regulation processes during heat shock, two parallel sets of the same deletions were generated using different 5Ј-IRs. In the TCH deletion plasmids the 5Ј-IR was obtained from ␣-tubulin, and in the HCH deletion plasmids the 5Ј-IR was obtained from HSP83.
3Ј-RACE was used to map the polyadenylation site of the wild type HSP83 and the fused CAT-HSP83 genes. This ensured that the 3Ј-UTRs retained their original site of polyadenylation and that the processing of the major CAT transcript occurred as expected 250 nucleotides upstream of the AG splice site of the successive gene. The poly(A) ϩ addition site of the endogenous HSP83 transcript was mapped 886 nucleotides downstream from the translation termination codon and 250 nucleotides upstream from the AG splice site of the subsequent gene copy in the cluster. The poly(A) ϩ sites of the CAT-HSP83 deletion transcripts also occurred as expected 250 nucleotides upstream from the successive site for trans-splicing. To examine the effect of the deletions in the 3Ј-UTR on mRNA stability at ambient and elevated temperatures, decay curves were obtained as described above, and steady-state levels were monitored at 4 h (Fig. 3, A-C). Except for ⌬1-201, each of the deletions in the HSP83 3Ј-UTR abolished differential stability of the CAT transcript at elevated temperatures, when the FIG. 2. The 3-IR of HSP83 confers temperature-dependent regulation on the CAT reporter mRNA. A scheme of the HSP83 genomic cluster is shown at the top, including the last gene copy and its flanking sequences (vertical bars). Cells were transfected with plasmids that contained the CAT coding gene cloned between the HSP83 5Ј-and 3Ј-IRs (white rectangles, pHCH), an ␣-tubulin 5Ј-IR (gray rectangles) combined with an HSP83 3Ј-IR (pTCH) and an HSP83 5Ј-IR and an ␣-tubulin 3Ј-IR (pHCT). RNA was extracted from cell lines grown at 26°C and after incubation (4 h) at 33°C. The black arrows under the DNA constructs show the RNA generated and its boundaries. Parallel cultures were incubated with actinomycin D at 26 or at 33°C, and aliquots were removed after 1, 2, 4, and 6 h for RNA extraction. Northern blots of 10-g RNA samples were hybridized with the CAT coding gene. The same blots were stripped and rehybridized with probes that corresponded to the HSP83 coding region and rDNA. The values of the hybridization intensity were measured in a PhosphorImager and normalized according to the rRNA. Accumulation of the mRNA at 33°C relative to 26°C is shown as histograms. The mRNA decay curves at 26°C (black lines, E) and 33°C (gray lines, ϫ) were obtained by nonlinear regression. The T1/2 values at 33 and 26°C were calculated, and the ratio between them is shown (right of the decay curves).
5Ј-UTR was derived from HSP83 (Fig. 3A, HCH constructs). It is possible that all of the tested modifications could alter the structure of the 3Ј-UTR and therefore change the accessibility of regulatory determinants. If the 5Ј-UTR was derived from ␣-tubulin all of the deletions eliminated differential stability, including ⌬1-201 (Fig. 3B, TCH constructs). Furthermore, the destabilizing effect of deletions in the 3Ј-UTR as elucidated by the 33/26 ratio of T1/2 was more evident in combination with the ␣-tubulin 5Ј-UTR. This could indicate possible interactions between the 5Ј-and 3Ј-UTRs of HSP83. In both sets of constructs, deletion 201-472 in the 3Ј-UTR had the most profound effect on eliminating differential stability. The black arrows under the DNA constructs show the RNA generated. The accumulation of CAT mRNA, the decay curves, T1/2 at both temperatures, and the ratio between them were obtained as described in the legend of Fig. 2. Data for the small deletions were obtained with pHCH-based (panel A) and pTCH-based (panel B) constructs. Data for the larger deletions and the LC IR were obtained with pHCHbased (panel C) constructs. The IR of the last gene copy in the HSP83 cluster (white rectangle) and its flanking region (vertical lines) were cloned from a cosmid library following a differential screen, using 3Јand 5Ј-probes (black bars above the last copy of the genomic cluster shown in Fig.  2). The IR of the LC was fused downstream to the CAT coding gene, resulting in pHCH/LC (panel C). In the absence of the endogenous signals for 3Ј processing the resulting CAT transcript generated a short 3Ј-UTR with 73 nucleotides. A similar transcript was also obtained in cells transfected with pHCH⌬59 -873.
We noted that although the tested deletions (200 -250 nucleotides) eliminated differential stability of the CAT RNA at elevated temperatures, the resulting transcripts accumulated to comparable or slightly higher levels at both temperatures (Fig. 3, A and B). This suggested that temperature-dependent degradation of mRNAs was not the only mechanism acting post-transcriptionally to control mRNA abundance in Leishmania.
To increase the instability of the 3Ј-UTR at 33°C and to examine the effect on accumulation of the CAT transcript, we introduced larger deletions into the 3Ј-UTR, spanning sequences ⌬59 -472, ⌬472-873, ⌬59 -655, and ⌬59 -873. These deletions led to a marked destabilization of the CAT transcripts at elevated temperatures, decreasing the 33/26 T1/2 ratio to 0.3-0.5. However, despite this dramatic reduction in mRNA stability, the CAT transcripts accumulated to comparable (0.97, pHCH⌬59 -873) or mildly reduced (0.67, pHCH⌬59 -472; 0.79, pHCH⌬472-873; 0.74, pHCH⌬59 -655) steady-state levels at 33°C (Fig. 3C). This could suggest that the abundance of the CAT transcripts was determined not only by differential mRNA stability. If 3Ј cleavage and polyadenylation were more efficient at 33°C, their influence could overcome the effect of reduced mRNA stability at elevated temperatures because of the deletions in the 3Ј-UTR. Thus, transcript abundance most probably resulted from the combination between the different regulatory processes.
The Last HSP83 Gene Copy Contains Sequences Homologous to the 3Ј-UTR but Lacks the Endogenous Signals for 3Ј Processing-To investigate further the effect of specific 3Ј processing signals on transcript abundance during heat shock we used the IR of the terminal HSP83 gene in the genomic cluster. This last gene copy and 3 kb of its nonhomologous flanking sequences were cloned from a cosmid library of L. amazonensis. This was done by a differential screen using probes derived from the 5Ј-and 3Ј-ends of the coding region ( Fig. 2A, scheme at the top describing the HSP83 cluster), as described previously for cloning of the first HSP83 repeat unit (21). The terminal IR sequence is almost identical (99%) to that of an internal repeat unit (1,471 bp) until position 840, from which point the two sequences are totally divergent.
Initially it was unclear whether the last copy (LC) was expressed because a putative polypyrimidine track and AG sites that could serve as signals for 3Ј processing of the last transcript in the cluster were observed 500 bp downstream to position 840 in the terminal IR. The IR of the last gene copy and 0.5 kb of the flanking sequences were therefore fused downstream to the CAT coding gene (pHCH/LC, Fig. 3C) and stably introduced into L. amazonensis cells. The polyadenylation site of the corresponding CAT transcript was examined by 3Ј-RACE and quite surprisingly, was mapped 73 nucleotides after the translational termination codon. This indicated that the flanking genomic sequences located downstream from the LC failed to direct 3Ј processing at the accurate site, and the resulting transcript was devoid of most of its 3Ј-UTR. Cleavage and polyadenylation that took place 73 nucleotides down-stream from the translational termination codon were most probably directed by cryptic signals. The boundaries of the CAT transcripts produced by the pHCH⌬59 -873 and pHCH/LC constructs were therefore similar, each containing a very short 3Ј-UTR of 73-84 nucleotides. However, despite the structural similarity between the two transcripts, their pattern of regulation varied (Fig. 3C). Both mRNAs were equally unstable at 33°C with a 33/26 ratio for T1/2 of 0.3, but they differed in their ability to maintain high steady-state levels of RNA at elevated temperatures. Although the HCH⌬59 -873 CAT transcript accumulated to a comparable steady-state level at both temperatures (33/26 ratio ϭ 0.97 at 4 h), the abundance of the HCH/LC mRNA decreased after incubation at elevated temperatures (33/26 ϭ 0.25). This could be explained by a reduction in the processing efficiency of 3Ј-ends in the HCH/LC transcript at elevated temperatures, which affected its RNA steady-state levels. Thus it appeared that the last gene copy in the HSP83 cluster lacks the endogenous signal for 3Ј processing and is most probably not expressed at elevated temperatures.
Processing of Heat Shock RNA Transcripts Is More Efficient at Elevated Temperatures-To examine the effect of temperature elevation on processing of heat shock transcripts, total RNA was isolated from cells grown at 26°C and after exposure to 33°C (4 h). The RNA was fractionated over a poly(dT)cellulose column, and the poly(A) ϩ and poly(A) Ϫ fractions were blotted and hybridized with HSP70, HSP83, and ␣-tubulin probes. As expected, the mature HSP83 transcripts (3.1 and 3.3 kb, respectively), although present at both temperatures were more abundant at 33°C (Fig. 4). Hybridization with the poly(A) Ϫ fractions revealed the presence of larger faint bands only in RNA obtained from 26°C. These bands were not observed in samples extracted from 33°C and could correspond to unprocessed transcripts, possibly dimers and trimers. Their abundance at the lower temperature suggested that processing of heat shock transcripts under these conditions was less efficient. Processing of the tubulin transcripts did not seem to vary with temperature alteration.
Sequences within the 201-472 of the 3Ј-UTR Are Required for Preferential Translation of Heat Shock Transcripts-Heat shock transcripts in Leishmania are translated preferentially at temperatures typical of the mammalian host (8). Previously we showed for L. amazonensis that the increase in translation of HSP70 and HSP83 was observed at 33°C, whereas translation of tubulin and other cellular non-heat shock proteins ceased at 37°C (7). We observed a similar pattern of regulation for the CAT mRNA flanked by HSP83 3Ј-and 5Ј-UTRs (Fig. 5 and Ref. 18) and therefore determined the location of the regulatory elements. This was done by incorporation of radioactive amino acids into nascently synthesized proteins for 30 min in cells grown at 26°C and after exposure to 33 and 37°C for 1 h. We show for Leishmania that the 3Ј-UTR derived from the HSP83 gene regulates the efficiency of translation at different temperatures. Translation of CAT increased at 33°C if it was fused to the HSP83 3Ј-UTR (pTCH). When the HSP83 3Ј-UTR was exchanged with the ␣-tubulin 3Ј-UTR preferential trans- FIG. 4. Heat shock transcripts are processed preferentially at elevated temperatures. Total RNA was isolated from cells grown at 26°C and after incubation at 33°C for 4 h. The RNA was fractionated over oligo(dT)-cellulose columns, and the 0.5-g poly(A) ϩ and 10-g poly(A) Ϫ fractions were separated over 1% agarose gels, blotted, and hybridized with probes corresponding to HSP83, HSP70, and ␣-tubulin. The faint bands marked with an arrow represent nonprocessed transcripts. lation was lost, despite the presence of an HSP83 5Ј-UTR (pHCT, Fig. 5A). Deletion analysis was used to map the 3Ј-UTR, indicating that preferential translation of the CAT transcripts was abolished by removal of the region 201-472, whereas deletion of other sequences, 1-201, 472-655, and 655-873, did not prevent the increase in translation at elevated temperatures (Fig. 5B). These results were obtained regardless of whether or not they were derived from HSP83 (pHCH) or ␣-tubulin (pTCH). However, although the HSP83 5Ј-UTR alone could not confer preferential translation, it had a synergistic effect of increasing translation at elevated temperature when combined with the HSP83 3Ј-UTR. The requirement for sequences within the 201-472 region was in line with the analysis of the larger deletions. Removal of the proximal half (59 -472) but not the distal half (472-873) of the 3Ј-UTR abolished preferential translation of CAT, in correlation with the absence (⌬59 -472) or presence (⌬472-873) of the 201-472 fragment (Fig. 5C). Sequences 201-472 were also absent from constructs carrying the large deletions (⌬59 -472, ⌬59 -655, ⌬59 -873), and cells transfected with these plasmids could not preferentially translate their CAT transcript. DISCUSSION We show that the 3Ј-UTR confers differential stability on the HSP83 mRNA in Leishmania because CAT transcripts fused to an HSP83 3Ј-UTR and an ␣-tubulin 5Ј-UTR followed the same pattern of temperature-dependent regulation observed for the endogenous HSP83 gene. These CAT transcripts were more stable and accumulated to a higher level at 33°C than at 26°C. Reciprocal constructs that contained an ␣-tubulin 3Ј-UTR and an HSP83 5Ј-UTR were degraded faster at 33°C and were less abundant at that temperature compared with 26°C. Decay curves at different temperatures revealed that the HSP83 5Ј-UTR had a limited synergistic effect on mRNA stability during heat shock when combined with the HSP83 3Ј-UTR. The HSP83 upstream IR may differentially accelerate trans-splicing of the 5Ј-ends, as also described for the phosphoglycerate kinase genes in T. brucei (26).
Analysis of the 3Ј-UTR was performed by introducing deletions of 200 -250 nucleotides downstream from the CAT gene in constructs that contained a 5Ј-UTR derived from ␣-tubulin. This showed that there was no specific region involved in temperature-dependent control of HSP83 mRNA stability. All deletions examined generated mRNA molecules that were no longer stable at 33°C, although removal of sequences 201-472 displayed the strongest effect. This could indicate that the 3Ј-UTR created secondary structures that were involved in regulation of mRNA stability and were disrupted by each of the deletions. Sequences 1-201, a region that is least conserved among Leishmania species (27), were dispensable only if the 5Ј-UTR was derived from HSP83. This observation could indicate possible interactions between the 3Ј-and 5Ј-UTRs, despite the inability of the 5Ј-UTR by itself to confer preferential mRNA stability at elevated temperatures. The larger deletions that eliminated half (either the proximal or the distal) or larger parts of the 3Ј-UTR had a more profound effect on differential stability of mRNA, and the T1/2 of the resulting CAT transcripts decreased by 2-3-fold upon heat shock.
Although most of the deletions that removed 200 -250 nucleotides from the 3Ј-UTR abolished the increased mRNA stability at elevated temperatures, they did not prevent the accumulation of the CAT-HSP83 transcript to higher levels. Only the more dramatic decrease in T1/2 at 33°C which was observed with the larger deletions generated CAT transcripts that no longer accumulated to higher levels at 33°C. We therefore examined whether a more efficient processing of HSP83 mRNAs at elevated temperatures could compensate for the reduction in stability. This was done by analysis of two mRNA molecules that had similar boundaries but varied in their 3Ј processing signals. One was obtained from an internal HSP83 IR and the other from the last gene copy in this cluster. We found that processing signals of the internal HSP83 IR were effective and functioned at elevated temperatures, but they were absent from the last gene copy. Although putative signals for 3Ј processing were observed within 0.5 kb of the sequences that flanked the terminal gene copy, they proved to be nonfunctional. Thus, CAT transcripts in cells transfected with either plasmid were polyadenylated very close to the translational termination site. Although both transcripts were devoid of a 3Ј-UTR and had similar stability values (33/26 ratio of 0.3 for FIG. 5. Sequences 201-472 in the HSP83 3-UTR are essential for preferential translation during heat shock. Cells were grown at 26°C, transferred to 33 or 37°C for 1 h, and metabolically labeled for 30 min at the corresponding temperatures with 35 [S]methionine/cysteine. Proteins were extracted and analyzed by separation over 15% SDS-PAGE followed by fluorography (18). The protein samples contained equal incorporated radioactivity and corresponded to a similar number of cells (22). The CAT protein is marked with an arrowhead. Panel A, effect of swapping the 5Ј-and 3Ј-HSP83 UTRs with the corresponding region from ␣-tubulin on preferential translation of the CAT reporter RNA. Panel B, effect of the smaller deletions in the HSP83 3Ј-UTR on preferential translation of the CAT reporter RNA using the pHCH and pTCH series of constructs. Panel C, effect of the large deletions in the HSP83 3Ј-UTR on preferential translation of the CAT reporter RNA, with pHCH-based constructs. T1/2), they varied in their ability to accumulate at 33°C. Although the steady-state level of the CAT mRNA in HCH⌬59 -873 cells was comparable at both temperatures, abundance of the HCH/LC transcript at 33°C decreased to 25% of its level at 26°C. This could be explained by variations in the efficiency of processing at the different temperatures. Indeed in L. amazonensis, nonprocessed heat shock RNAs detected in the poly(A) Ϫ fraction were more abundant at ambient temperatures than at elevated temperatures, suggesting that transsplicing and polyadenylation of heat shock genes were more efficient at 33-35°C. Thus, the abundance of heat shock transcripts was determined by a combination between the increased stability and a more efficient RNA processing at elevated temperatures.
Heat shock inhibits splicing in higher eukaryotes (10, 28 -30) causing disintegration of splicosome particles (29). HSP70 genes in Drosophila are intronless, enabling their maturation during extreme conditions. The gene encoding HSP83 in Drosophila contains an intron, and its splicing is optimal under mild heat shock conditions. However, a severe temperature stress interferes with HSP83 splicing unless the cells are pretreated with a milder stress that prevents this inhibition (10). In trypanosomes, a severe heat shock (42°C) impairs splicing of tubulin transcripts, although HSP70 and HSP83 RNAs are properly matured under these conditions (1,31,32). We now show for L. amazonensis that processing of primary transcripts of HSP70 and HSP83 is more efficient during heat shock.
Exposure of Leishmania parasites to temperatures typical of mammals causes a dramatic increase (10-fold) in translation of heat shock proteins. Although this pattern of regulation is conserved among all eukaryotes, it is not yet quite understood. In Drosophila translation of a reporter gene increases upon heat shock if it is fused to an HSP70 5Ј-UTR (33,34). However, no cis-or trans-acting regulatory elements have been identified, and no differences were found for the pattern of RNAprotein interactions at alternating temperatures (35). It was suggested that the nonstructured AT-rich 5Ј-UTR of the Drosophila HSP70 mRNA enabled translation to proceed with no interruptions during heat shock. Unlike Drosophila, preferential translation at elevated temperatures occurred in Leishmania only if the CAT reporter gene was fused with an HSP83 3Ј-UTR. Replacing it with an ␣-tubulin 3Ј-UTR abolished preferential translation even if the 5Ј-UTR was derived from HSP83.
We identified a regulatory element located between positions 201 and 472 in the 3Ј-UTR of HSP83 which enhanced translation in Leishmania during heat shock. Translation increased at 33°C even if the distal half of the 3Ј-UTR (472-873) was removed but not when the proximal half of the 3Ј-UTR (59 -472) was deleted, in correlation with the presence or absence of this element. Moreover, preferential translation occurred even in transcripts that were less stable at 33°C because of mutations in the 3Ј-UTR as long as the 201-472 region was present. This was best exemplified by the CAT mRNA in pHCH⌬472-873 cells that translated more efficiently at elevated temperatures, despite being less stable. Thus, translational regulation and control of mRNA stability of HSP83 in Leishmania are independent processes. The dissociation between stability and translation efficiency of RNA is usually not common among eukaryotes; however, it has been observed in some cases. For example, an element within the 3Ј-UTR of c-fos links the translation and intracellular localization of this transcript but has no bearing on its stability (36).
Elements that control translational regulation in eukaryotes are generally found in the 5Ј-UTR. However, there are several examples of transcript-specific factors that affect translational regulation via elements in the 3Ј-UTR. The mouse quaking proteins (QKI) involved in embryogenesis and myelination and a closely related protein in Caenorhabditis elegans (GLD-1) which is necessary for germ line development are members of the STAR family of RNA-binding proteins. Both proteins were found to serve as translational repressors that act through regulatory elements called TGEs (for tra-2 and GLI elements), present in the 3Ј-UTR of their target genes (37,38). During axis formation in Drosophila, translation of hunchback is repressed by the gene products of nanos and pumilio via binding to defined elements in its 3Ј-UTR (39,40). An element within the 3Ј-UTR of the lipoxygenase transcript is responsible for its silencing during erythroid differentiation (41). In T. brucei, a 26-mer sequence in the 3Ј-UTR of the major surface antigen of procyclic trypanosomes controls the developmentally regulated abundance of the procyclic acidic repetitive antigen (PARP) transcript and the protein it encodes (42,43). Although translational regulatory elements are occasionally found within 3Ј-UTRs, the 5Ј-UTR of heat shock genes in eukaryotes accounts for their traditionally conserved translational regulation. However, the evidence presented here suggests that a 3Ј-UTR determinant is responsible for preferential translation of HSP83 in Leishmania, illustrating another aspect of unique and unusual regulatory pathways employed by trypanosomatids.