Proc. Natl. Acad. Sci. USAVol. 89, pp. 2703-2707, April 1992Biochemistry

Gene expression in Leishmania: Analysis of essential 5'DNA sequences

(protozoa/Kinetoplastida/trans-splicing/transcription)

MARIA A. CUROTTO DE LAFAILLE, AVRAHAM LABAN, AND DYANN F. WIRTH*Department of Tropical Public Health, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115

Communicated by Phillips W. Robbins, December 18, 1991

ABSTRACT A major unanswered question in Kinetoplas-tida parasites is the mechanism of regulating gene expression.Using a transfection system, we have previously shown that theintergenic region of the a-tubulin gene of Leishmania enrieticontained sequences required for gene expression. The goal ofthe work reported here was to determine whether the Leish-mania-derived sequences were providing transcriptional con-trol signals or functioning at a post-transcriptional level, mostlikely in trans-splicing. The chloramphenicol acetyltransferase(cat) gene was used as the reporter gene and was stablyintroduced into L. enrketti as part of an extrachromosomalelement by transfection. We show here that the production ofcat mRNA was dramatically dependent on the presence of theintergenic region 5' to the cat gene. The intergenic region couldbe substituted by a smaller fragment (222 base pairs) thatcontained the trans-splice acceptor site and an adjacent poly-pyrimidine tract. This native fragment could be replaced by asynthetic polypyrimidine tract containing an AG site. Thenative and the synthetic fragments had unidirectional activity.No effect on transcription of the cat gene by the wild-typefragment or the synthetic polypyrimidine was detected. Theresults indicate that both regions contain signals that affectRNA stability, probably sequences involved in trans-splicing.

Leishmania are flagellated protozoan parasites of the orderKinetoplastida. Many of the aspects of gene expression andgene regulation in these organisms are still poorly under-stood. For example, little is known about the relative con-tributions of transcriptional and post-transcriptional pro-cesses to the control of gene expression. A general feature ofgene expression in Kinetoplastida organisms is the process-ing ofmRNAs by trans-splicing (refs. 1 and 2; reviewed in ref.3). Trans-splicing is similar to cis-splicing in its requirementfor U2, U4, and U6 (4) and is related to self-splicing of groupII introns by its independence of U1 (5). A possible functionof trans-splicing in Kinetoplastida organisms is the process-ing ofpolycistronic transcripts to create a capped 5' end in themRNAs (6, 7).Many highly expressed genes in Leishmania and other

Kinetoplastida organisms are present in multiple copies or-ganized in tandem repeats-for example, the tubulin genes(8-10). The intergenic regions are short and contain the sitesfor polyadenylylation ofthe upstream gene and trans-splicingofthe downstream gene (11, 12). Promoters have not yet beenidentified in Leishmania but because of the hypothesizedpolycistronic transcription it is likely that the promoters willbe located upstream of the tandem arrays (13-16). Thedevelopment of transfection vectors for Leishmania in whichbacterial genes were expressed under the control of inter-genic regions from the parasite raised the possibility thatthese regions could contain signals for transcription (17, 18).

In this work we have studied the role of the intergenicregion ofLeishmania enriettii a-tubulin in the control ofgeneexpression. Our main purpose was to determine if the activityof the intergenic region was at the transcriptional or post-transcriptional level and to identify sequences essential forgene expression. The chloramphenicol acetyltransferase(cat) gene (19) was used as the reporter gene and was stablyintroduced into L. enriettii in an extrachromosomal elementby transfection (18). Expression of the cat gene was depen-dent on the presence of either the native intergenic region ora synthetic polypyrimidine tract and splice acceptor site. Thenative and synthetic fragments had unidirectional activity,and neither had transcriptional activity. The results indicatethat both regions contain signals that affect RNA stability,probably sequences involved in trans-splicing.

MATERIALS AND METHODSPlasmid Construction. Standard cloning techniques were

used (20). The structure of the plasmids was determined byrestriction analysis and DNA sequence analysis. All plasmidswere derivatives of p50. p50 was constructed using plasmidsp50-neo and pLCLC. p50-neo contained a 909-base-pair (bp)BstBI fragment from pALT-neo (18) end filled and clonedinto the Sma I site of pBluescript KS' (Jim Tobin, personalcommunication). The insert in pSO-neo contained 50 bp fromthe 3' end of the a-tubulin intergenic region ofL. enriettii (12)and the neomycin-resistance (neor) gene (21). pLCLC con-tained two copies of the intergenic region of a-tubulin (LT1)and two copies of the cat gene in tandem (5'-LT1-cat-LT1-cat-3'). A 1.6-kilobase (kb) BstBI fragment from pLCLC wascloned into the Cla I site of pSO-neo. The resultant plasmid,p5O, contained, in 5' -- 3' order, a 50-bp fragment from theintergenic region of a-tubulin, the cat gene, the intergenicregion of a-tubulin, the neor gene, and pBluescript KS'vector. p2-3 was derived by the deletion of the Sma I/HincIIfragment from p50. pLCLN was obtained by replacing theSal I/Sma I fragment from p50 by the intergenic region ofa-tubulin (Xho I/Sma I fragment from pALT1.1, ref. 17).pR-S and pS-R were obtained by cloning the Rsa I/Sma Ifragment from pALT1.1 into HincII/Sma I-digested p50.The plasmids containing the synthetic polypyrimidine

tracts were obtained by insertion of the double-strandedoligonucleotide OPy43 (5 '-CCTCGAGCTCTCTCTCT-CTTCTTCCCCTCTCCTCTCTCTCTCTCTTCTCCA-GACGCGTT-3') upstream from the cat gene. The double-stranded OPy43 fragment was digested with Xho I and clonedinto the Sal I/Sma I sites of p50. Alternatively, blunt double-stranded OPy43 was cloned into the HincII/Sma I sites ofp5O.

Cell Cultures and Transfection. Promastigotes of L. enri-ettii were grown and transfected as described (18). Trans-

Abbreviations: neor, neomycin-resistance; CAT, chloramphenicolacetyltransferase; SL, spliced leader.*To whom reprint requests should be addressed.

2703

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

2704 Biochemistry: Curotto de Lafaille et al.

fected cells were grown in medium containing 150 ug ofG418per ml. The presence of the plasmid DNA as a circularextrachromosomal element (episome) in the transfected celllines was determined by Southern analysis of chromosomegels and standard Southern analysis of restricted DNA (datanot shown). No evidence of chromosomal integration wasobserved. The copy number of plasmid DNA in the trans-fected cell lines was calculated by comparison of the hybrid-ization signals in Southern blots of the Pst I fragment of theintergenic region of a-tubulin derived from the plasmid withthose derived from the chromosomal tandem repeat (p2-3, 93copies; pLCLN, 62 copies; pR-S, 50 copies; pS-R, 62 copies;p114, 39 copies; pl2n, 56 copies; pl-5, 56 copies; pl-2, 86copies). The copy number of plasmid DNA has been stablein these transfected cells.CAT Activity. CAT activity was assayed in supernatants of

cell lysates (17) using the two-phase fluor-diffusion method(22). The results presented correspond to the CAT activity inlysates containing 3 pg of soluble protein.

Analysis of Steady-State RNA. Total RNA and poly(A)+RNA were obtained as described (12). Northern analysis (20)was performed using either total RNA (8 ,g in Fig. 2 or 1 ,ugin Fig. 3) or poly(A)+ RNA (0.1 ,ug in Fig. 3). cat RNA wasdetected by hybridization with a 32P-labeled antisense RNA.To detect neor RNA a pool of the following 32P-labeledantisense oligonucleotides was used: NEO 1 (5'-GGGCAC-CGGACAGGTCGGTCTTGAC-3'), NEO 3 (5'-ACCATGA-TATTCGGCAAGCAGGCA-3'), NEO 4 (5'-CCGGAGAA-CCTGCGTGCAATC-3'). A 3-tubulin probe was obtained byrandom primer labeling of the 800-bp Sal I fragment frompLEB3 (12). Hybridization intensities were quantified in abio-imaging analyzer, Fujix BAS 2000.

Primer extension reactions were performed as described(23). The primers used were CAT 20 (5'-CAACGGTGG-TATATCCAGTG-3') and NEO 4 (sequence as above). Forthe polymerase chain reaction (PCR) analysis of the 5' end ofthe cat mRNAs 1 ,ug of total RNA was annealed to oli-go(dT)12.18 and incubated with avian myeloblastosis virusreverse transcriptase under standard conditions. One-hundredth of the reaction was then used for amplificationthrough the PCR (24) using a primer homologous to thespliced leader (SL) sequence (ME, 5'-AAGCTTAGTAT-CAGTTTCTGTACT-3'; sequence from ref. 23) and the CAT20 primer (sequence above). Southern blots of the PCRsamples were hybridized to a 32P-labeled internal oligonucle-otide (SCAT 61, 5'-ATTTTAGCTTCCTTAGCTCCTG-3').Direct sequencing of PCR products was performed as de-scribed (25).Run-On Transcription Analysis. Nuclei were obtained by

lysis of promastigotes in hypotonic buffer (26) with 0.8%Nonidet P-40. Nascent RNA was elongated in vitro asdescribed (26, 27). Incubations of nuclei for RNA elongationwere done at 27°C for 5 min. Hybridizations were performedat 42°C as described (26). Radioactivity in the slots wasquantified in a Fujix BAS 2000 analyzer.DNA fragments homologous to the coding regions of

plasmid or chromosomal genes were obtained by PCR and gelpurified. Equimolar amounts ofDNA for each fragment weredenatured and applied to a nylon filter using a slot blotapparatus (cat and neor, 0.35-kb fragment, 200 ng per slot;a-tubulin, 1-kb fragment, 570 ng per slot; Pro-gl, ref. 28,0.5-kb fragment, 290 ng per slot; Plasmodium falciparummdr, ref. 19, 1-kb fragment, 570 ng per slot). Single-strandedsense DNA and antisense DNA were obtained by PCRamplification of specific DNA fragments, followed by Aexonuclease treatment (25) and gel purification. The follow-ing oligonucleotides were used to generate DNA fragmentshomologous to the coding regions of cat (19), neor (21), L.enriettii a-tubulin (unpublished sequence, D.F.W.), Pro-gl(28), and mdr from P. falciparum (P.f.) (29): CAT 2, sense

(5'-ACCGTTCAGCTGGATATTACGGCCTTT-3'); CAT 3,antisense (5'-CACCCAGGGATTGGCTGAGACGAA-AAAC-3'); NEO 2, sense (5'-GCTGTGCTCGACGTTGT-CACTGAAG-3'); NEO 3, antisense (as above); ALPHA 3.1,sense (5'-ACAAGTGCATCGGTGTCGAGGATGAC-3');ALPHA 3.3, antisense (5'-TGTGGTGTTGCTCAGCATA-CACAC-3');PRO61, sense(5'-GACGGAGACCAGAAGGT-GATGG-3'); PRO 71, antisense (5'-ACAGCAGAATGCCG-GTGATGG-3'); P.f. mdr 2484, sense (5'-AAAAT-TAATAATGAGGGT-3'); P.f. mdr 3038, antisense (5'-AGTGTTTCTGAAATGAACATAGCAATAGCA-3'); P.f.mdr 510, sense (5'-AGAGAAAAAAGATGGTAACCTCAG-3'); P.f. mdr 1487R, antisense (5'-TGGTAAGATAATTGT-TAACAT-3').

RESULTSTo analyze 5' DNA sequences required for the expression ofa foreign gene in Leishmania, we constructed the vectorpLCLN. pLCLN contained the neor gene as a selectablemarker and the cat gene as the reporter gene. Both geneswere flanked at their 5' ends by the intergenic region ofa-tubulin (Fig. 1A). pLCLN and the derivative plasmids (Fig.1) were stably transfected into L. enriettii using conditionsthat resulted in the extrachromosomal amplification of mul-timers ofthe transfected plasmid (18). The number ofplasmidcopies was determined for each cell line, and expression ofneor and cat mRNA in all subsequent experiments wasnormalized to plasmid copy number in each cell line.We have previously shown that cells transfected with a

plasmid that carried the neor gene flanked by a-tubulinintergenic regions produced neor mRNA trans-spliced at thea-tubulin AG acceptor site and polyadenylylated within thea-tubulin intergenic region 3' to the neor gene (18). PlasmidpLCLN does not contain Leishmania sequences immediately3' to the neor gene. In cells transfected with pLCLN,polyadenylylation of the neor mRNA occurs in a cryptic sitewithin pBluescript, at -600 bp from the polylinker, gener-ating a 1.6-kb mRNA (data not shown). Expression of theneor gene was analyzed for each cell line and compared to theexpression of f8-tubulin, a chromosomal gene. (The ratioswere as follows: p2-3, 0.82; pLCLN, 1.03; pR-S, 0.84; pS-R,0.75; p114, 1.39; pl2n, 0.89; pl-5, 1.24; pl-2, 1.00.) Theexpression of neor mRNA was similar in all cell lines exam-ined.

Parasites transfected with pLCLN expressed high levels ofCAT activity and 1.5-kb cat mRNA (Fig. 2 A and B). Thepresence of the a-tubulin intergenic region 5' to the cat genewas essential for expression, since only minimal CAT activitywas detected in cells transfected with a plasmid lacking theintergenic region 5' to the cat gene (p2-3, Fig. 2A). CATactivity in cells transfected with pLCLN was -1000-foldhigher than in cells transfected with p2-3. Cells transfectedwith plasmid p2-3 did not express detectable levels of the1.5-kb cat mRNA (Fig. 2B), whereas the levels of neormRNA were similar to those observed in cells transfectedwith pLCLN.

Polypyrimidine Fragments Can Substitute for the IntergenicRegion 5' to the cat Gene. To determine more precisely whatsequences were required for cat expression, plasmids con-taining insertions of either native or synthetic sequences 5' tothe cat gene were transfected into L. enriettii and catexpression was analyzed in the transfected cell lines. Fig. 2shows that the 222-bp Rsa I-Sma I fragment from theintergenic region (pR-S, Fig. 1A) could replace the completeintergenic region 5' to cat. CAT activity and cat mRNA inparasites transfected with pR-S were expressed at similarlevels as those in cells transfected with pLCLN. The effectof the Rsa I-Sma I fragment was unidirectional, since noCAT activity or 1.5-kb cat mRNA was detected in cells

Proc. Natl. Acad. Sci. USA 89 (1992)

Proc. Natl. Acad. Sci. USA 89 (1992) 2705

BACat p1-2A

.I.I i

p2-3

PLCLN- G AG cat_p_.

p 2 n

t p1-5IN

1.--

pR-S

pS-P

C vpLCLN / . . /cgggccccccctcgaggTCGAGGGAGG/ . .. /TCCCTCTCCCCTCTCCCCGCACACACACGCAGT

CCTTCGCTTTCACTCTTCGAACAAACACCTCAAACCATCTTATCACCTTTCTCTCTTCCTTGTCggggGAmTCCCCCGGGAGCTTGGCGAGATTTTCAGGAGCTAAGGAAGCCTATG/ ... /

p2-3 /... /cgggccccccctcgaggtcGGGAGCTTGGCC-AGATTTTCAGGAGCTAAGGAAGCTAAAATG/. . . /

p114 /... /cgggqcccCcctcgaggTCGAGCTCTCTCTCTCTTCTTCCCCTCTCCTCTCTCTCTCTCTTCTCCAGAACGCGTTCCTCTCCTCTCTCTCTCTCTTCTCCAGACGCGTTGGGAGCTTGGCGAGATTTTCAGGAGCTAAGGAAGCTA AATG/... /

pl2n /... /cgggccccccctcgaggtcAACGCGTCTGC-AGAAGAGAGAGAGAGAGGAGAGGGAAGAAGAGAGAGAGAGCTCGAAACGCGCTCGGAGAAGAGAGAGAGAGAGGAGAGGGGAAGAAGAGAGAGAGAGCTCGAGGGGAGCTTGGCGAGATTTTCAGG-AGC-iAGGAAGCTA*GATG/ ... /

FIG. 1. (A) Structure of the plasmids used to transfect L. enriettii. pBluescript sequences are represented by a line; inserts are boxes. Arrowsbelow the DNA fragments indicate the orientation of the cloning. T3 and T7 promoters are shown. The AG splice acceptor site of a-tubulin isshown in plasmid pR-S. S1, Sal I; X, Xho I; R, Rsa I; S, Sma I; E, EcoRI; Sc, Sac I; M, Mlu I; P, Pst I. Restriction sites that were destroyedduring the cloning are shown in parentheses. (B) Plasmids containing the synthetic polypyrimidine tract OPy43 (striped box). Restriction sitesas in A. The AG site at the 3' end of OPy43 is shown. (C) Sequences upstream of the cat gene in pLCLN and derivative plasmids. The junctionsof the 5' end of the cat gene and immediate upstream sequences are shown. The sequences are shown from left to right in 5' -. 3' direction.

Sequences derived from pBluescript are shown in lowercase letters, inserts are in uppercase letters, and cat sequences (19) are shaded.Polypyrimidine tracts are in bold type. AG sites that have been identified as trans-splice acceptor sites in this work are underlined. The AGsplice acceptor site that is normally used in the a-tubulin gene is indicated by an arrowhead. Gaps in the presented sequence are indicated by/I* *'/

transfected with a plasmid containing the inverted fragment(pS-R, Figs. 1A and 2).One feature of the 222-bp fragment is the presence of

polypyrimidine tracts adjacent to the AG splice acceptor site(ref. 12; see Fig. 1C). We decided to test directly the effectof polypyrimidines on cat expression. A synthetic oligonu-cleotide containing a run of pyrimidines (OPy43) was cloned5' to the cat gene. The oligonucleotide also contained an AGdinucleotide at its 3' end. Parasites transfected with plasmidscontaining a single copy or multiple copies of the syntheticpolypyrimidine were found to express CAT activity. Para-sites transfected with p114, which contains two polyprimi-dine tracts of 43 and 26 nucleotides (Fig. 1 A and C),expressed high levels ofCAT activity and cat mRNA (Fig. 2A and B). The effect of the polypyrimidine on cat expressionwas unidirectional, since parasites transfected with a plasmidcontaining two inverted copies of the polypyrimidine (pl2n,Fig. 1 B and C) had CAT activity at background levels andno detectable 1.5-kb cat mRNA (Fig. 2 A and B).CAT activity and cat mRNA were compared in cells

transfected with plasmids containing one (pl-2), two (p114),or three (pl-5) polypyrimidine tracts (Fig. 1B). The increasein the number of polypyrimidine tracts resulted in an increasein CAT activity and increased expression of cat mRNA in thetransfected cell lines (Fig. 3). It is interesting to note that thelevel of CAT activity was lower in cells transfected with pl-5than would be predicted based on the amount of mRNAdetected. This may indicate a down-regulation of CAT ac-

tivity when high levels of mRNA are present.The Effect of the Wild-Type 222-bp DNA Fragment and the

Polypyrimidine Region on cat Expression Is Post-Transcrip-tional. Transcriptional elements have not yet been identified

in plasmids containing the intergenic region of a-tubulin. Todetermine whether the dramatic effect of the 222-bp fragmentor the polypyrimidine tract on steady-state cat mRNA levelshad a transcriptional component, we analyzed transcriptionof plasmid DNA in run-on experiments. Labeled nascentRNA from transfected cell lines and control untransfected L.enriettii was hybridized to slot blots containing DNA frag-ments from the coding regions of plasmid genes (cat andneor), L. enriettii chromosomal genes (a-tubulin and Pro-gl),and a negative control DNA from another organism (P.falciparum mdr). These results (Table 1) demonstrate thattotal transcription across the cat and the neor genes wassimilar in transfected cell lines regardless of the presence or

the orientation of the active signals 5' to the cat gene.Northern analysis demonstrated the presence of antisense

transcripts derived from the transfected DNA (data notshown). Therefore, we decided to analyze transcription fromeach DNA strand in a subset of the transfected cells. Thecoding region of the a-tubulin gene was used as a chromo-somal control and the P. falciparum mdr gene was used as a

negative control. Blots containing single-stranded sense andantisense DNA for the coding regions of plasmid genes catand neor, L. enriettii chromosomal a-tubulin and P. falci-parum mdr, were hybridized to labeled nascent RNA. Theresults (Table 2) demonstrate that both DNA strands were

transcribed in the episomal DNA, whereas transcription wasstrand specific in the a-tubulin chromosomal gene. Theresults presented here demonstrate that neither the 222-bpnative fragment nor the synthetic polypyrimidine modulatestranscription of the cat gene in the episomal elements.Gene Expression Is Associated with the Production of Trans-

Spliced mRNA. The presence and site of addition of the SL

14

Biochemistry: Curotto de Lafaille et al.

2706 Biochemistry: Curotto de Lafaille et al.

A

c

4 ;s

o0

M X

uE

1 2 3 4 5 6 7

Cell lines

B

sets

FIG. 2. (A) CAT activity in stably transfected L. enriettii. Theresults shown represent the mean and standard deviation of threesamples. Numbers 1-7 indicate the plasmids used for transfection: 1,untransfected L. enriettii (L.e.); 2, p2-3; 3, pLCLN; 4, pR-S; 5, pS-R;6, p114; 7, pl2n. CAT activity (cpm/min, primary data) determinedin supernatants from cell lysates was as follows: 0 (no. 1), 41 ± 3 (no.2), 39,500 ± 600 (no. 3), 39,600 ± 1800 (no. 4), 0 (no. 5), 25,400 ± 500(no. 6), 5 ± 1 (no. 7). (B) Northern analysis of cat RNA in transfectedcell lines. An autoradiogram of a filter hybridized with a cat-specificantisense riboprobe is shown. Size standards (RNA ladder, BethesdaResearch Laboratories) are shown on the left. The number on theright indicates the calculated size of the main cat RNA. Cell lines are

as in A.

sequence on cat mRNA were determined by a combination ofprimer extension and cDNA PCR analysis, which was fol-lowed by isolation and sequencing of the PCR products (Fig.1C and data not shown). cat mRNA was spliced at the nativeAG acceptor site (Fig. 1C) in cell lines transfected withpLCLN or pR-S or at two downstream alternative AG sites(shown in Fig. 1C). No evidence for trans-spliced cat mRNAwas detected in transfected cells that expressed backgroundlevels of CAT activity (p2-3, pS-R, and pl2n), despite thepresence ofthe alternative AG sites described above. Thus the

FIG. 3. CAT activity and cat mRNA in cells transfected withplasmids containing synthetic polypyrimidine tracts 5' to the catgene. The names of the plasmids used for transfection are indicated.L.e., untransfected L. enriettii. CAT activity shown is normalized to100 copies of plasmid per cell and represents the mean and standarddeviation of three samples. The upper box shows the corresponding1.5-kb cat RNA band from an autoradiogram of a Northern blot ofRNA extracted from the same transfected cell lines. (In a longerexposure, the mRNA band from cells transfected with plasmid pl-2is visible.) Two sets of the same samples of RNA from transfectedcell lines were gel fractionated, transferred to nitrocellulose, andhybridized with probes specific for cat or ,3-tubulin. The 1.5-kb catmRNA and the 2.2-kb 3-tubulin mRNA bands were quantified.Values for cat mRNA were normalized to 100 copies of plasmid percell. The ratio of cat mRNA to /-tubulin is shown.

presence of AG sites alone was not a sufficient signal fortrans-splicing.Two sites of SL addition were identified in cat mRNA

isolated from cells transfected with plasmid pll4, the majorsite being the AG located between the two copies of thepolypyrimidine fragment and the second one corresponding tothe first site downstream of the two polypyrimidine fragments(Fig. 1C). A third downstream splice site was also identified(Fig. 1C). The results demonstrate that trans-spliced catmRNAs are present in cells expressing high CAT activity andabsent in those cell lines with background activity.

DISCUSSIONIn this work we have studied the role of the intergenic regionof the a-tubulin gene of L. enriettii in the control of geneexpression. Our goal was to determine whether the activityof the intergenic region was transcriptional or post-transcriptional and to identify sequences essential for geneexpression. We showed here that the expression of thebacterial gene cat in stably transfected L. enriettii is depen-dent on the presence ofLeishmania DNA sequences 5' to thegene. Those DNA sequences can be provided either by thefull-length intergenic region of a-tubulin, a truncated 222-bp

Table 1. Comparison of transcription in transfected cell lines

Blotted DNA pR-S pS-R p114 pl2n p2-3 L.e.

cat 740 200 520 ± 60 1230 ± 410 890 ± 410 340 ± 90 11 ± 8neor 420 20 310 ± 50 590 ± 30 430 ± 70 280 ± 30 16 ± 3a-Tubulin 130 10 85 ± 6 160 ± 20 150 ± 10 129 ± 9 170 ± 30Pro-gl 40 10 34 ± 8 62 ± 9 59 ± 2 54 ± 7 73 ± 6P.f. mdr 0 1 ± 1 7 ± 6 6 ± 5 0 0

Labeled nascent RNA was hybridized to slot blots containing double-stranded DNA fragmentshomologous to the coding regions of the genes indicated in the first column. Expression of cat and neorgenes was normalized to 100 copies of plasmid per cell. No correction was made for a-tubulin, Pro-gl,or P. falciparum (P.f.) mdr values. The numbers shown are photo luminescence units in a 34-mm2 areaper 350 bp of blotted DNA. Values are expressed as mean ± SD of three filters. L.e., untransfectedcells.

UEl

Proc. Natl. Acad. Sci. USA 89 (1992)

..mmmr7---TLLa.

Proc. Natl. Acad. Sci. USA 89 (1992) 2707

Table 2. Bidirectional transcription of plasmid DNA intransfected cellsBlottedDNA Strand pR-S pl-5 p2-3 L.e.

cat + 254 1063 156 85- 213 405 114 38

neor + 219 675 154 34- 195 494 287 27

a-Tubulin + 121 425 123 446- 2 3 0 55

P.f. mdr + 0 0 0 16- 0 0 0 13

Labeled nascent RNA was hybridized to slot blots containingsingle-stranded DNA homologous to the coding regions of the genesindicated in the first column. P.f., P. falciparum. The hybridizationsignals were quantified. Expression of cat and neor genes wasnormalized to 100 copies ofplasmid per cell. No correction was madefor a-tubulin. L.e., untransfected cells. Units as in Table 1.

fragment from the same region containing theAG trans-spliceacceptor site, or a synthetic polypyrimidine tract. The native222-bp fragment and the synthetic polypyrimidine behavedsimilarly in that they led to the expression ofhigh levels ofcatmRNA, their activity was unidirectional, and neither affectedtranscription of the cat gene. From these results we concludethat these regions contain signals that modulate RNA stabil-ity. Because of the 5' location of the signals, their proximityto the AG acceptor site, and the association of the signalswith the production of trans-spliced mRNA, it seems that themost probable function of these sequences is to participate intrans-splicing. These results also imply that trans-splicingefficiency may play a major role in controlling mRNA levels.Our results suggest that unspliced pre-mRNA is unstable, inagreement with a previous report (4) of unstable pre-mRNAin experiments in which trans-splicing was inhibited.We have shown here that a polypyrimidine tract and

adjacent AG site 5' to the cat gene were sufficient to producetrans-spliced mRNA. Our results do not preclude the partic-ipation of other signals but indicate that other sequences areprobably not essential for trans-splicing. Polypyrimidinetracts are essential elements of the splice acceptor site ofmammalian introns, in which they participate in the forma-tion of the presplicing complex and the binding of U2(reviewed in refs. 30 and 31). Our results point to a possiblesimilarity between cis- and trans-splicing in the determinationof the splice acceptor site.The demonstration of a post-transcriptional effect of the

DNA signals described here still leaves open the question ofthe existence and location of transcriptional elements in theseextrachromosomal elements. Transcription of the plasmidDNA is bidirectional, in contrast to the unidirectional tran-scription observed in the chromosomal a-tubulin gene. Theresults indicate that different elements control transcriptionin both cases. Since no difference in transcription wasobserved between plasmids containing or lacking sequencesof the intergenic region 5' to the cat gene, it seems unlikelythat the intergenic region ofa-tubulin contains transcriptionalsignals. However, the presence of a weak promoter in theintergenic region cannot be ruled out.

In this report, we have identified DNA sequences that areessential for gene expression in the protozoan parasite L.enriettii and we have determined that those sequences par-ticipate in trans-splicing. The knowledge of the factors that

control gene expression in these organisms is necessary forthe development of molecular tools to study the genetic basisof infectivity and pathogenesis.

We thank Dr. Susumu Tonegawa for the use of the Fujix bio-imaging analyzer, Sarah Volkman for materials, and Ramona Gonskifor help in the preparation of this manuscript. This work has beensupported by a grant from John D. and Catherine T. MacArthurFoundation and National Institutes of Health Grant A121365-05 (toD.F.W.). D.F.W. is a Burroughs Wellcome Scholar in molecularparasitology.

1. Sutton, R. E. & Boothroyd, J. C. (1986) Cell 47, 527-535.2. Murphy, W. J., Watkins, K. P. & Agabian, N. (1986) Cell 47,

517-525.3. Agabian, N. (1990) Cell 61, 1157-1160.4. Tschudi, C. & Ullu, E. (1990) Cell 61, 459-466.5. Bruzic, J. P. & Steitz, J. (1990) Cell 62, 889-899.6. Borst, P. (1986) Annu. Rev. Biochem. 55, 701-732.7. Laird, P. W. (1989) Trends Genet. 5, 204-208.8. Landfear, S. M., McMahan-Pratt, D. & Wirth, D. F. (1983)

Mol. Cell. Biol. 3, 1070-1076.9. Seebeck, T., Whittaker, P. A., Imboden, M., Hardman, N. &

Brown, R. (1983) Proc. Natd. Acad. Sci. USA 80, 4634-4638.10. Tomashow, L. S., Milhausen, M., Rutter, W. J. & Agabian, N.

(1983) Cell 32, 35-43.11. Sather, S. & Agabian, N. (1985) Proc. Natl. Acad. Sci. USA 82,

5695-5699.12. Landfear, S. M., Miller, S. I. & Wirth, D. F. (1986) Mol.

Biochem. Parasitol. 21, 235-245.13. Gonzalez, A., Lerner, T. J., Huecas, M., Sosa-Pineda, B.,

Nogueira, N. & Lizardi, P. (1985) Nucleic Acids Res. 13,5789-5804.

14. Johnson, P. J., Kooter, J. M. & Borst, P. (1987) Cell 51,273-281.

15. Imboden, M. A., Laird, P. W., Affolter, M. & Seebeck, T.(1987) Nucleic Acids Res. 15, 7357-7368.

16. Muhich, M. L. & Boothroyd, J. C. (1988) Mol. Cell. Biol. 8,3837-3846.

17. Laban, A. & Wirth, D. F. (1989) Proc. Natl. Acad. Sci. USA86, 9119-9123.

18. Laban, A., Tobin, J. F., Curotto de Lafaille, M. A. & Wirth,D. F. (1990) Nature (London) 343, 572-574.

19. Hadfiel, C., Cashmore, A. M. & Meacock, P. A. (1987) Gene52, 59-70.

20. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning:A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY), 2nd Ed.

21. Beck, E., Ludwig, G., Auerswald, E. A., Reiss, B. & Schaller,H. (1982) Gene 19, 327-336.

22. Neuman, J. R., Morency, C. A. & Russian, K. 0. (1987)BioTechnics 5, 444-447.

23. Miller, S. I., Landfear, S. M. & Wirth, D. F. (1986) NucleicAcids Res. 14, 7341-7360.

24. Saiki, R. K., Bugawan, T. L., Horn, G. T., Mullis, K. B. &Erlich, H. A. (1986) Nature (London) 324, 163-166.

25. Higuchi, R. G. & Ochman, H. (1989) Nucleic Acids Res. 17,5865.

26. Kooter, J. M., van der Spek, H. J., Wagter, R., d'Oliveira,C. E., van der Hoeven, F., Johnson, P. J. & Borst, P. (1987)Cell 51, 261-272.

27. Grondal, E. G. M., Evers, R., Kosubek, K. & Cornelissen,A. W. C. A. (1989) EMBO J. 8, 3383-3389.

28. Cairns, B. R., Collard, M. W. & Landfear, S. M. (1989) Proc.Natl. Acad. Sci. USA 86, 7682-7686.

29. Foote, S. J., Thompson, J. K., Cowman, A. F. & Kemp, D. J.(1991) Cell 57, 921-930.

30. Padgett, R. A., Grabowski, P., Konarska, M. M., Seiler, S. &Sharp, P. (1986) Annu. Rev. Biochem. 55, 1119-1150.

31. Green, M. R. (1986) Annu. Rev. Genet. 20, 671-708.

Biochemistry: Curotto de Lafaille et A