conformational changes associated with splicing

Biochemistry

Ribosomal RNA introns in archaea and evidence for RNA

conformational

changes associated with splicing

(extremethermophile/23S rRNA/RNA rearrangement)

J0RGEN

KJEMSt*

AND ROGER A. GARRETT§

tBiostructuralChemistry, KemiskInstitut,AarhusUniversity,8000AarhusC,Denmark;andtlnstitutforBiologiskKemi B,CopenhagenUniversity,

S0lvgade 83, 1307 Copenhagen K,Denmark

CommunicatedbyCarl R. Woese, August29, 1990

ABSTRACT The single 23S rRNA gene of the archaeon

Staphylothermus marinus exhibits two introns which, at the RNAlevel, are located in highly conservedregions ofdomains IVand V. The RNA introns, which are 56 and 54 nucleotides long,respectively, canform singlehairpinstructures. Invivo,

RNA splicing occurs efficiently, whereas in vitro pre-rRNA transcripts containing each intron were cleaved efficiently

when incubated with archaeal cell extracts but were poorly ligated. The introns arecleaved byamechanism which differs fromthemechanisms ofeukaryoticrRNA introns but resem-bles thoseofthe rRNA intronofDesulfurococcusmobilis and thearchaeal tRNA introns. The cleavage enzymerecognizes

and cuts aputativebulge-helix-bulgestructurethat canform at the archaeal exon-intron junctions. Using a phylogenetic se-quence comparison approach, we defime the parts of this structural feature that are essential for cleavage. We also provide evidencefor conformational changesoccurringin the S. maninus23SRNA,after cleavage,atboth exon-exon junc-tions,whichmay account for the lowyieldsofligationobserved in vitro.

Intronsoccurwidelyamongeukaryoticnuclei andorganelles ingenesforproteins, ribosomal RNAs (rRNAs) and transfer RNAs (tRNAs) (1, 2). Among archaea they have been detected inonerRNAgene(3) and insometRNAgenes(4-9). So far, they have notbeen detected in bacterial genomes,

although theyoccurinproteingenesofsomebacteriophages (10, 11).

Theonly archaeal rRNA intron detectedtodatelies within the 23S RNA gene of Desulfurococcus mobilis (3). On excision, it yields acircular RNA which is large [622 base pairs (bp)] and stable andprobablyencodesasingle protein (12).Although itssplicingmechanism is special and differs in

many respectsfrom the mechanisms ofeukaryotic group I

introns, it shows similarities tothe splicing of the intron in pre-tRNATrpofextreme halophiles (13) and other archaeal tRNA introns (4, 6-9); thus all can generate a bulge-helix-bulge structure attheexon-intronjunction thatmay

consti-tute asubstrate for the cleavageenzyme (8, 9, 13, 14). Apossiblebasisfor this similar mechanism ofrRNAand tRNAcleavageis that theD.mobilis intron lies ina regionof 23SRNAthat appearsisostructuralwith the D and anticodon armsoftRNA(12, 15).Therefore, to establish whether this

intron and its location were an exception among archaeal

pre-rRNAs, we searched for new rRNA introns among extreme thermophiles. The 23S rRNA genes from four

re-cently discovered organismswereexaminedandone ofthem, fromStaphylothermus marinus (16), exhibited two

introns.¶

Their mechanisms ofcleavage are shown to be similar to

thoseof theD.mobilisintron,and both require a

rearrange-mentof basepairing, aftercleavage, togeneratethemature

23SRNA structure.

MATERIALS AND METHODS

Preparation ofCells,CellularDNA, andRNA;Cloning and DNA Sequencing. Cells from S. marinus, Pyrobaculum is-landicum, Pyrococcusfuriosus, and Pyrodictium occultum

were kindly provided by Karl Stetter, and cells from D. mobilis, Thermoproteus tenax, Thermoflum pendens, and Sulfolobus B12 were gifts from Wolfram Zillig. Isolation proceduresforgenomic DNA and total RNAweredescribed earlier(17).Cloningwasperformed byreplacingthe internal BamHIfragment of A-EMBL3 (18) withgenomic DNAcut

with BamHI. The rRNA genes were selected by using a

mixtureof randomly labeled RNA fragments from the 3' half of 23S RNA from otherextremethermophiles(17,19,20).An

-8-kbp BamHIfragmentwasisolated from S.marinusanda

2.2-kbpBamHI-Pvu IIfragment, containing 1.8 kbp of the 23S RNA gene, was subcloned in M13mpl9. Exonuclease Bal-31 (Amersham) was used togenerate clonesexhibiting increasingly large deletions from either end (21), such that continuousreading could beobtainedon both DNAstrands by usingthedideoxynucleotide sequencing procedure (22). Southernblottingwasperformed underhighstringency

con-ditions (23).

T7 RNA Polymerase Transcription in Vitro.Templates for transcriptionwere prepared by cloning, first,a 136-bpHae IIIfragment containing intron 1 and, second, a 106-bp Alu I-HaeIIIfragment containing intron2in theunique Stu I site ofan M13mpl9construct behind the T7 RNA polymerase

promoter. Therecombinantphagewaslinearized withPstI and blunt ends were then produced by using the Klenow fragment of DNA polymerase (Amersham). Transcription using T7 RNA polymerase (24) produced RNA with the cloned sequenceplus an extraGG atthe 5' end and CCN, CCNN,orCCNNN atthe 3' end.

Conditions for inVitro Splicing. Crude cellextractsfrom all organismswerepreparedasdescribed earlier (12).Atypical

in vitrosplicing mixture contained1-5

jig

of RNAwhichwas

randomly labeled with [a-32P]GTP in 20

Al

of standard splicing buffer (50mM Tris HCI, pH 7.5/5 mM EDTA/100

mMNaCl/2mM spermidine)and 2

Al

of crude cell extract

from S.marinuscells containing about 25 ,ug of total protein. Themixture wasincubatedat37°C for30min. For splicing oflarge amounts of nonradioactive RNA, the reaction

vol-ume was scaled up and the RNA concentration was main-tained.

5'- and 3'-End-Labeling of RNAFragments and RNA

Se-quencing. 5'-Phosphates were labeled byfirst removing

5'-*Present

address: Center for Cancer Research, Massachusetts In-stitute ofTechnology, Cambridge, MA 02139.

fThe sequence reported in this paper has been deposited in the GenBank data base (accession no. M38363).

Thepublication costs ofthisarticle weredefrayed in part by page charge payment.This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

440 Biochemistry: Kjems andGarrett

phosphates with alkaline phosphatase (12) and then labeling with 50 ,Ci of

[y-32P]ATP

(6000Ci/mmol, NEN; 1 Ci = 37 GBq) and polynucleotide kinase (Amersham) (23). The 3' ends were labeled with [32P]pCp by using T4 RNA ligase (Amersham) (25). End-labeled RNA was sequenced enzy-matically (26). Primer-directed sequencing of RNA using reversetranscriptase and dideoxynucleotides was performed asdescribed earlier (12).

RESULTS

Splicing in Vivo. An 8.0-kbp BamHI fragment from S. marinus chromosomal DNA which hybridized to a 23S RNA-specific probe was cloned in A-EMBL3. A 2.1-kbp BamHI-Pvu II subclone wasisolated which contained 1.8 kbp coding for domains II to VI of the 23S RNA and 0.3 kbp ofthedownstreamregion. It was sequenced on both strands and acomparison of the sequence of the noncoding strand with thesequences of other 23S RNAs revealed the presence oftwointronlikeinsertsof56bpand54bp within domains IV and V, respectively (Fig. 1). The remainder of the gene, coding for domain I and part of domain II, was not se-quenced.CorrespondingBamHIfragments were cloned from thechromosomal DNAsof Pyrobaculum islandicum, Pyro-coccusfuriosus, and Pyrodictium occultum and subclones

Proc. Natl. Acad. Sci. USA 88(1991)

starting within domain II and extending beyond domain VI were sequenced. No intronlike inserts were detected.

Thenumber of 23S RNA genes and intronlike inserts in the genome of S. marinus were investigated by the Southern blotting procedure. Total cellular DNA was digested with

BamHI,

EcoRI, HindIII, or Pst I and hybridized to a ran-domly labeled restrictionfragment from the 23S RNA gene. Total DNAdigested with eachrestrictionenzymeyielded a single band (data not shown), which implies that only one copy of the gene, containing two inserts, occurs in the genome.

Toestablish whether the inserts were introns, the 23S RNA gene wassequencedacross theputativeexon-exonjunctions. Two primers,

P1

(5'-GGGGACCTCGTTGATCC-3') and P2

(5'-CCCGTTCCTCTCGTACT-3'),

which are complemen-tary tosequencesdownstream from putative introns 1 and 2, respectively, were annealed with total cellular RNA and upstream sequences were determined by the dideoxynucle-otide procedure (22). Only sequences corresponding to li-gated exons weredetected, and notermination occurred at theligation site(Fig. 2). We conclude that the inserts were, indeed, introns.

The splicing sites lie within regions of highly conserved sequence and secondary structure in

domains

IVand V, close tothe sites of other known rRNA introns (Fig. 1). They occur after nucleotides A-2098 and C-2769 and both precede the

A DOMAIN IV AA U A A A ACA U AUU AA G A C- G C- G C- G G U A G A U C-G G- C C -- G C-G C- G C- G U ° G G-C A G ° U -GU CUG\ 6 \\CU A 3' A A C U A G A G-C G-C G- C C- 6 AC-G_C A G A A U GA G-C CC - G U UG-C A G-C C-G G-C G °UG U-AA AC- G 6 CC-G A G-C 2080 AUC C UGC G G GUA C II oII U A C Au Pc.3 U C G U GA T p. Uelntron 1 C G CG -C U GU 2130-G-C C GC C-G G C G6 PP-2 A U U G A G 'KD.m. B DOMAIN V A6G A A 5'C C CC C CGC GGGCGAU III 1 3'GG G G G GCGCCC CU AU C G Intron 2 G AkA G U U CCG\\ C U '2760 UU C~r.chip .375n. G-C C- G C-G C- G A C -G Au Pp. 1 fAA S.c.mit C2620 G C AG U G U CG.-II . 25n UCG UGGC..' A UU C A U _C G G a UGU G-C U o G C-G G-C A-U C A- U , G o U G-C G-C G-C 49n* INTRON 1 5G6GGUA AC U C CCA AAJU6 CCUU|3' G-G- C G- C A U G C C- G C- G C- G-G- C U-A U-A AG- C AC -G -C- G C-G Uc A G-uAUAu INTRON 2 S. Az cu 3' G-CA A-U~ G-C G- C G-C C C -U-A G- C-G-C G-C G-C U-A C-G G-C G-C G-C -G- C A A-A A A A uU AA

FIG. 1. Secondary structures withindomain IV (A)anddomainV (B)of the mature 23SRNAfrom S. marinus, showing the intron loca-tions. The structures were derived fromcomparative sequence analyses and the nucleotide numbers of D. mobilis23S RNA areused(19). Only basepairswhicharephylogenetically

supported in the mature rRNAs are

indicated by lines or dots. Putative secondary structures are presented

foreach intron andexon-intron

junc-tionbesidetherespective domain;the cleavage sitesare indicatedbythick arrows.Thoseregionswhichweinfer

areinvolved inabase-pair

rearrange-mentafter cleavageareboxed and the sequences whichare base paired in the mature 23S RNA arejoinedby arrows. A-denotesevery 10th

nucle-otide in the vicinities of thesplicing

sites and in the introns. The locations of other known rRNA intronsarealso indicated with thefollowing abbrevi-ations:C.r.chlp,Chlamydomonas

re-inhardii chloroplast (27); D.m., D. mobilis; P.c.3, Physarum carolina

(28); P.p.1 and -2, Physarum polycephalumintrons 1 and 2(29, 30); S.c.mit, Saccharomyces cerevisiae mitochondria (31); T.p., Tetrahy-menapigmentosa(32).

Biochemistry: Kjems and Garrett

A

A G C U db'4 .. MM c _ u m c

qju

_u

41w

A

_W

G

B

A G C U G A G A G G __ t ~U r ~~* C

r

_

G* 4w RNA 1 A 12 34 A.-W 155---Bt- _- 129 -B 1 23 4 56 C- 82-_ E1- GM 73 s- -l C2- 79-_ B,2 _ 86

FIG. 2. Sequences complementary to the exon-exon boundary regionsafter excision of intron 1 (A) and intron 2 (B) and ligationin

vivo. Sequences were obtained by annealing 0.5 pmol of 5'-end-labeledprimerP1orP2,respectively, to 20,ugof total cellular RNA. Theprimers were extended by using reverse transcriptase. Ligation sites are marked witharrowsand the RNA sequencesacrossthese sitesareshown.In Baputativeguanosinemodification,denotedby

anasterisk, caused strong termination in all four lanes.

sequenceAGGU.Theintrons differ insequencebut bothcan

form17- to18-bp helices exhibiting small internalloopsand A+ U-rich terminalloops (Fig. 1).

The fate of the excised intronsin vivo wasinvestigated by annealing primers P3(5'-CCCAGCATATATATGGG-3')and

P4 (5'-GGCAGGGGTGGGGTCGGGG-3'), which are

com-plementary to sequences within RNA introns 1 and 2,

re-spectively,tototal cellular RNA.Extension of these primers

at the 5' end by reverse transcriptase revealed no longer products, which suggests that the introns are rapidly de-graded after excisionin vivo (datanotshown).

Splicing in Vitro. Yields of the intron-containing pre-23S RNA isolated from cells were too low forin vitro splicing studiesand, therefore, RNAtranscripts 1and2, containing their respective introns andflankingexonsequences(Fig.3),

weresynthesized byusing theT7RNApolymerase. RNA 1 was transcribed from aHae III fragment and RNA 2 was

derived fromanAlu I-Hae IIIfragment. Neithertranscript showed self-splicing activity when incubated over a wide

rangeof solution andtemperatureconditions, including those tested for the rRNA intron of D. mobilis (12). However, incubation of bothtranscripts witha crudeextractfrom S.

marinuscells generated various cleavageproducts

compat-ible withintron excision(Fig. 4,gels A). Cleavage occurred between30°C and 80°C andwasefficientat75°C and37°Cfor both RNAs, with a slightly higher rate ofcleavage at the highertemperature. Cleavage wasefficientbetween 50 and 150mMNaCl,was morespecificintheabsence ofMg2+,and

was enhancedbyspermidine (2 mM); NTPs hadnoeffect. The various cleavage products were extracted from the denaturing polyacrylamide gels (Fig. 4, gels A), labeled at

their5' or3'termini,andsequenced byusingribonucleases (26).Owingtothehighly stablestructures, we wereunableto getcompleteterminalsequencesfor all thefragments.

Nev-23SDNA B P 126 56 73 ppp Intron 1 RNA1 \25 54 m-OH pp OH Intron 2 RNA2

FIG. 3. Construction of the transcripts RNA 1 and RNA 2 in vitro.The upperline shows theorganizationof the 2.2-kbp

Bam-HI(B)-Pvu 11(P)DNAfragmentwhichwassequenced. Below,the

compositionsof the derivedtranscripts,RNA and2,areillustrated.

Thick linesdenote sequences corresponding tomature 23S RNA. The numberof nucleotides in eachtranscriptisindicated.

D,_- 56--_

D2- _ 54 -E2 - 32

[ALg

F.- 26 F2- _,_ 25 -i -.

FIG.4. Autoradiogramsshowingthe in vitrocleavage productsof RNA 1andRNA 2fractionatedondenaturing lo polyacrylamide

gels. Reactionswere performed bymixingrandomly labeled

tran-scriptswith crudecellextractsofS.marinus andincubatingat37°C

instandardcleavagebuffer(50mMTris'HCI, pH 7.5/5mMEDTA/

100 mMNaCI/2mMspermidine)unless otherwise stated. Gels A: Lanes1,control RNA without cellextract.Lanes 2-4 containedcell extract. Lanes2,10 mMMg2'added and EDTAomitted;lanes3,

standardbuffer;lanes4,5mMspermidine.Gels B: Purifiedcleavage

intermediates B1 and B2 (intron-3'-exon) or C1 and C2

(5'-exon-intron), from RNAs 1 and 2, were incubated at37°C in standard

cleavagebuffer.Lanes1,cleaved RNAmarker;lanes2,intermediate B without cellextract.Lanes 3 and4contained cellextract. Lanes

3, intermediate B;lanes4, intermediate B plus 5 ,ugofunlabeled RNA.Lanes5, control, intermediateC withnocellextract;and lanes

6, intermediate C with cell extract. The nucleotide lengthsof the

cleavage products are given and their compositions are indicated

schematically;thick andthin lines correspondtointrons andexons,

respectively.

ertheless, by sequencing both ends of each fragment the cleavage sites could be assigned unambiguously (data not shown). Most of the possible cleavage products were

de-tected, including the free intron but not the ligated exons;

their identities are illustrated in Fig. 4. For RNA 1, the electrophoretic mobilities correlate with size, whereas for RNA 2, intron-containing bands migrated anomalously, probably due to incomplete denaturation of the putative 9 adjacent G*C base pairs of the intron (Fig. 1B).The3'exon

ofRNA 2 (E2) yielded multiple bands, which may reflect residualconformational heterogeneity.

Cleavage Mechanism. Fractionation of stable 5'-exon-intron and intron-3'-exon intermediates (Fig. 4, gels A) enabled us to test for formation of the bulge-helix-bulge featureatthe exon-intronjunction, since the5'-exon-intron product can generate a stable 5'-cleavage bulge while the intron-3'-exon cannot form the 3'-cleavage bulge. These intermediateswereisolated from both RNA 1 and 2, purified,

and incubated withcrude cellextract.The resultsingelsB of Fig. 4 demonstrate that the 5'-exon-introns (intermediatesC1 and C2)are, indeed, cleaved, while the intron-3'-exons (in-termediates B1and B2)arenot,evenafteraddingexcessRNA transcript. This supportstheidea that thebulge-helix-bulge structureisasubstrate forthecleavage enzyme.

Finally, weinvestigated whethereach pair ofexons was

ligated and whether the intronswerecircularized.Purified 5' and3'exonsfrom RNA 1 andRNA2,and their introns(Fig.

4, gels A), were extracted and incubated with crude cell

extract undervarious conditions (+ Mg2, + spermidine,

Proc. Natl. Acad. Sci. USA 88 (1991) 441

RNA2 A 1234 A2- Who 111_ B 123456 -a -ama 4

442 Biochemistry: Kjems and Garrett

NTP) and examined electrophoretically. Neither exon liga-tion nor introncircularizationwasdetected (data not shown). Previously, we demonstrated thatcleavage ofthe D. mo-bilis pre-rRNA yields products with 5'-OH and 3'-P04 groups. The following experiments on the S. marinus pre-rRNAindicate thatthe samegroups areformed: (i) electro-phoretic mobility ofthe 3' exon which exhibits a 3'-OH4-(Fig. 3) was unchanged byalkaline phosphatase treatment; (ii) yields of the 5'-end-labeled intron and 3'-exon did not increaseafter phosphatase treatment; and(iii) the 5' exon and intron werelabeledefficiently attheir 3'ends by [32P]pCp and RNA ligaseonly if theRNA waspretreated with phosphatase (data not shown).

Other Archaea Contain the Cleavage Enzyme. Cleavage efficiencies of cell extracts from the extreme thermophiles D. mobilis, Thermoproteus tenax, Thermofilum pendens, Pyro-baculum islandicum, Pyrococcus furiosus, and Sulfolobus B12, from the methanogen Methanobacterium thermoau-totrophicum Marburgstrain,and from theextreme halophile Halococcus morrhuae were tested, using 25 ug of total protein in each assay. The extreme thermophile extracts were tested in standard splicing buffer containing 10 mM MgCl2 instead of EDTA for 20minat56°C, while the methanogen extract was tested in the same buffer at37°C

4nd

theextreme halophile was tested with and without 5 M NaCl at 37°C. All except the H. morrhuae extract produced cleavage (data not shown). None of the digests was complete and, in general, the yield of the5'-exon-intron was lower than that of the intron-3'-exon, which is compatible with the former intermediate of S. marinus being more susceptible to cleavage (see above).

DISCUSSION

The discovery of twoadditional small rRNAintrons among the archaea shows that the D. mobilis rRNAintron(3) was not an exception. Moreover, the exon junctions of the S. marinus rRNA do not apparently generate tRNA-like struc-tures and, therefore, the presence of such a structure at the splicing site of the D. mobilis 23S RNA (12) is neither a prerequisite for archaeal rRNA introns nor anexclusive basis for any mechanistic similarities betweenarchaeal rRNA- and tRNA-intron splicing.

Defining the Cleavage Substrate. Extracts from extreme thermophile cells that lack rRNAintrons can inducesplicing of D. mobilis rRNAtranscripts and we inferred, therefore, that the cleavage and ligation enzymes have other cellular functions (12). Thisinference isreinforcedby thefindinghere that cell extracts from several extreme thermophiles and a methanogen have similar activities. Daniels and coworkers (13) noted that a bulge-helix-bulge structure that had been proposed earlier as a processing site on the long stems of archaeal 16S and 23S RNAs (33) could also form at the exon-intronjunctionsof all archaealpre-tRNAs (4-8, 14)and the pre-23S RNAof D.mobilis.Theexon-intronjunctions of the S. marinus introns can generate the same feature, and experimental support isprovidedfor their formation. Totest

the generality ofthis concept, we undertook aphylogenetic analysis of the 16 available archaealpre-tRNA and pre-23S RNA sequences that contain introns.

Sequence alignments across the exon-intron boundaries are presented in Fig. 5A. There are no strictly conserved nucleotides, although there is a low level of conservation (=70%)for themotif 5'-R*A... intron. . . GAR*A-3' with cleavagesites (*) located in thebulges (Fig. 5B).Examination of eachputative base pair revealed that thetwopairsflanking each bulge, eight in total, are well supported by multiple compensating base changes with very few mismatches (Fig. SB). Incontrast, basepairs 1 and 2 and 11 and 12inhelices Iand III,respectively, weresubjecttomultiplemismatches, which implies either that they do not form or, iftheydoform

A

D.m. 23S RNA

5'-Exon Intron 3'-Exon

UUGCCGGGUATAGGGC 607n GCCCCCCGAG AGUUC *

0** v0

S.m. 235 RNA 1 GACUCUCUUA ACGGG 40n CCCGGGAGAAAGGUA *

s D D T s s s s s 7 7 7 H H ;.m. 23SRNA 2 ).m. tRNACYs ).m. tRNAMet r.p. tRNAGlY ;.. tRNASer ;.s. tRNALeu i.s. tRNAGl y ;.. tRNAMet ;.s. tRNAPhe r.t. tRNAMet r.t. tRNA~la r.t. tRNALeu i.v. tRNATrp H.v. tRNAMet 70% Consensus

GUCGUGAGAC AGCAG 37n CUGCCUCAAA AGGUC * CGGGCUGCAG GGCAC 15n GUUCACGGAG ACCCG (*)

CGGACUCAUA CGGGG 24n CCCAGGGGAG AUCCG (*)

CAAGCCGCAG AGCCU 4n AGGCGCGGAA AACCC (*)

UAGGCUCGAG ACCAU iOn AAGGCGGGCG ACCUA

CGGGCUCAGG ACCAU UGGCCUCCCA UUCUU

UAGGUGAGAC CCCCG UAGAGUGGAA AGCCA CCGGCUCAUA AACUA 4n AUGUGGAGAG ACCGG CCGGCUGAAG UUUUU in AUACGCAGUG ACCGG GGGCUCAUAA AGGGC 3n AGGCCCGAGG AGCCC (*)

AAGGACGCCC AGGCU ln UGCCGCGUAG GCCCU (*)

GGCAGGGUUC ACAGC 3n GCUGGCCCAG AGGUC * CUGACUCCAG AGGCU 89n CACCGGAGAU AUCAG *

CGCACUCAUA GGGUU 60n CACCUGGGAC AUGCG R A B 5' 3, 1 N- N Helix I 2 N - N 3 N- N 4 N- NNA N 5 N- N N Helix II 6 N - N 7 N- N 8 N N- N N 9 ifN N -N Helix III 10 H - N 11 N- N 12 N- N I Introns GAR A CBC 6 5 4 6 4 7 7 5 Mismatch 2 4 1 1 0 0 2 2 Consensus G-C C-G 7 1 8 2 7 6 6 7

FIG. 5. (A)Alignmentof sequences at theexon-intronjunctions of the archaeal rRNA and tRNA introns. D.m.,D. mobilis; S.m.,S. marinus;T.p., Thermofilumpendens;S.s., Sulfolobussolfataricus; T.t., Thermoproteustenax;H.v.,Halobacterium volcanii;n, nucle-otides. Two arrowheads on the upperlinedenote positionsofthe cleavage sites andthe nucleotides with dots underneatharebulgedin thesecondarystructureshown below. Fororganismsmarked withan * thecleavage sites havebeen determined accurately, while those

markedwith(*)have been determined within±1nucleotide.The lack of an asterisk indicates that the sites are derived from sequence alignments.Nucleotidescommon tomorethan70%ofthe sequences are presented below as a consensus sequence. (B) Phylogenetic

sequencecomparison evidence forthe formation of theconsensus

structure attheexon-intronboundary.Nsdenoteanyof the sequences givenin A and the arrows indicate thecleavagesites. Basepairsare

numbered 1-12 and the numbers ofcompensatingbasechanges(CBC)

and mismatchesareindicated for each basepair.Theconservedbase pairregion which receives strong phylogenetic support is boxed. Literaturereferences forthe sequenceswhichare notgiveninthetext areD.m. tRNAs,ref. 8 andunpublished results; T.p. tRNA, ref. 9;

S.s. tRNAs,ref.4; T.t. tRNAs,ref.6;andH.v. tRNAs, refs. 5 and 34.

in some pre-RNAs, that their presence is not critical for substrate recognitionandcleavage.

Thus,

theboxed areain Fig. SB represents the essential structural core of the

sub-strate.

A more extensive

analysis

of the structures ofthe exon-intron regions revealed twomain types of helices I and III. Mostof thepre-tRNAsexhibitastablehelixI

and,

generally,

ashorthelixIII,whilethe

pre-rRNAs

fromD. mobilisand S.

marinus, and the

pre-tRNAGIY

from

Thermofilum

pendens

exhibitashortorirregularhelix I anda

long

stable helixIII. This suggests that eitherastable helix I or astable helix III is required, together with helix II, to generate afunctional

substrate. Thesetwosubstrateclasses also correlate withthe

type of cleavage intermediate that accumulates in vitro. Pre-RNAs exhibitingastronghelixI/weak helix III

accu-mulate the 5' exon-intron and are exemplified by pre-tRNATrP from Halobacterium volcanii (13). When the 3' bulge is cutfirst, helices IIand IIIwill dissociate, thereby disrupting the 5' bulge and preventing its cleavage. In con-trast, prior cutting of the 5' bulge will not appreciably destabilize helicesIand II orthe3'bulge. The second class of pre-RNAs withaweakhelixI/strong helixIIIaccumulates the intron-3'-exon, and is typified by rRNA intron 2 of S. marinus. Here, prior cutting of the 5' bulge will lead to

dissociation of helices I and II, loss of the 5' exon, and disruption of the 3' bulge, while prior cutting of the 3' bulge willnot appreciably destabilize helix II, helix III, or the 5' bulge.

Insummary, (i) the cleavagesubstrate constitutes a con-served core structure consisting of eight base pairs and two bulgesattheexon-intronjunction (boxedin Fig. 5B); (ii)a

long stable helix isrequired on one side of thesubstrate;and (iii) the bulges arenot cutin agiven orderin vitro.

23S RNA Rearranges After Intron Excision. Initially, we weresceptical aboutformation of the bulge-helix-bulge struc-ture inpre-23S RNA of D. mobilis because it would require asubstantial RNA rearrangement, aftercleavage,togenerate the mature 23S RNA (8, 19). However, subsequently, we foundanintron-containing pre-tRNA (9), inadditionto two

described earlierforThermoproteus tenax(6),and nowthe

pre-rRNAfrom S. marinus, all of which form a bulge-helix-bulge featureatthe expenseof thematureRNA structure (35, 36). Wealsoprovidedexperimentalsupportfor formationof thebulge-helix-bulgestructure in thepre-rRNA ofD. mobilis (8).Thus,for six of theknownarchaealexon-intronjunctions both the bulge-helix-bulge feature (Fig. 5B) and thelocally rearranged structure of the mature RNA, aftersplicing, are strongly supported phylogenetically (19, 37). Formation of thesealternative conformers is alsosupportedbyfree energy calculations (ref. 13; J.K., unpublished results).

Theputative conformational changemay accountforthe low yields ofligated exons produced in vitro where, under suboptimal conditions, there may be time for the RNA to rearrange prior to ligation, which could render the ligation reaction moredifficult.

We are grateful to Prof. Karl Stetter for providing extreme thermophile cells. We thank Tina Olesenfor excellent technical assistance and JacobDalgaard,ArneLindahl,and VickaNissen for helpwith thefiguresandmanuscript.Grants were received from the Danish Natural Science Research Council and the Novo and Carls-bergFoundations.

1. Cech, T. R. & Bass, B. L. (1986) Annu. Rev. Biochem. 55, 599-629.

2. Reed, R. &Maniatis, T. (1986) Cell 46, 681-690.

3. Kjems,J. &Garrett, R. A. (1985) Nature (London) 318, 675-677.

4. Kaine, B. P., Gupta, R. & Woese, C. R. (1983) Proc. Nat!.

Acad. Sci. USA 80, 3309-3312.

5. Daniels, C. J., Gupta, R. & Doolittle, W. F. (1985) J. Biol. Chem. 260, 3132-3134.

6. Wich, G., Leinfelder, W. & Bock, A. (1987) EMBO J. 6, 523-528.

7. Kaine, B. P. (1987) J. Mol. Evol. 25, 248-254.

8. Kjems,J., Jensen,J.,Olesen,T.&Garrett,R. A.(1989)Can. J. Microbiol.35, 210-214.

9. Kjems, J., Leffers, H., Olesen, T. & Garrett, R. A. (1989) J. Biol. Chem. 264, 17834-17837.

10. Chu, F. K., Maley, G. F. & Belfort, M. (1984) Proc. Natl. Acad. Sci. USA81, 3049-3053.

11. Sjoberg, B. M., Hahne, S.,Matthews, C. Z., Matthews, C. K., Rand, K. N. & Gait, M. J. (1986) EMBO J. 5, 2031-2036. 12. Kjems, J. &Garrett, R. A. (1988) Cell 54, 693-703.

13. Thompson, L. D. & Daniels, C. J. (1988) J. Biol. Chem. 263, 17951-17959.

14. Thompson, L.D., Brandon,L. D.,Nieuwlandt, D. T.& Dan-iels, C. J. (1989) Can. J. Microbiol. 35, 36-42.

15. Andersen, A., Larsen, N., Leffers, H.,Kjems, J. &Garrett, R. A. (1986) in Structure and Dynamics of RNA, eds. Van Knippenberg, P. H. & Hilbers, C. W. (Plenum, New York), pp. 221-237.

16. Fiala, G., Stetter, K. O.,Jannasch, H. W., Langworthy, T. A. & Madon,J.(1986) Syst. Appl.Microbiol. 8, 106-113. 17. Kjems,J., Leffers,H.,Garrett,R. A., Wich, G., Leinfelder,

W. & Bock, A. (1987) Nucleic Acids Res. 15, 4821-4835. 18. Frischauf,A.,Lehrach, H., Poustka, A. & Murray, N. (1983)

J.Mol.Biol. 170, 827-842.

19. Leffers,H.,Kjems,J.,0stergaard,L.,Larsen, N. & Garrett, R. A. (1987) J. Mol. Biol. 195, 43-61.

20. Kjems,J., Leffers, H.,Olesen,T., Holz,I. &Garrett,R. A. (1990) Syst. Appl. Microbiol. 13,117-127.

21. Guo, L. H.,Yang, R.C. & Wu, R. (1983) Nucleic Acids Res. 11,5521-5540.

22. Sanger, F.,Nicklen, S. &Coulson, A. R. (1977) Proc. Natl. Acad.Sci. USA 74, 5463-5467.

23. Maniatis,T.,Fritsch, E. F.&Sambrook,J.(1982) Molecular Cloning:A Laboratory Manual (Cold Spring Harbor Lab., Cold Spring Harbor, NY), pp. 122-124.

24. Studier, F. W. & Moffatt, B. A. (1986) J. Mol. Biol. 189, 113-130.

25. Bruce, A.G. & Uhlenbeck,0. C.(1978)NucleicAcids Res.5, 3665-3677.

26. Donis-Keller,H.,Maxam, A. M. &Gilbert, W. (1977)Nucleic Acids Res.8, 2527-2528.

27. Allet,B. &Rochaix,J. D.(1979)Cell 18, 55-60. 28. Muscarella,D. E.& Vogt, V. M. (1989) Cell56,443-459. 29. Nomiyama, H., Sakaki, Y. & Takaji, Y. (1981) Proc. Natl.

Acad. Sci. USA 78, 1376-1380.

30. Nomiyama,H.,Kuhara,S., Kukita, T., Otsuka, T. &Sakaki, Y.(1981) Nucleic Acids Res. 9,5507-5520.

31. Dujon,B. (1980)Cell 20, 185-187.

32. Wild, M. A. & Sommer, R. (1980) Nature (London) 283, 693-694.

33. Mankin, A. S., Teterina, N. L., Rubtsov, P. M., Baratova, L. A. & Kagramanova, V. (1984) Nucleic Acids Res. 12, 6537-6546.

34. Datta, P. K., Hawkins, L. K. & Gupta, R. (1989) Can. J. Microbiol.35, 189-194.

35. Egebjerg,J.,Larsen, N. &Garrett,R. A.(1990)in Structure, Function andEvolutionof Ribosomes,eds. Hill, W.,Garrett, R. A., Moore, P., Schlessinger, D. & Warner, J. (Am. Soc. Microbiol., Washington, DC), pp. 168-179.

36. Rich,A.&RajBhandary, U. L. (1976) Annu. Rev. Biochem. 45, 805-860.

37. Sprinzl,M., Hartmann, T., Meissner, F., Moll, J. & Vorder-wulbecke,T.(1987) NucleicAcids Res. 15,r53-r188.