in catalytic and informational RNAs as well (7, 8). This RNAs containing each of these tetraloops have been solved

Biochemistry

Functional

interchangeability

of the

structurally similar

tetranucleotide

loops

GAAA and

UUCG

in

fission

yeast

signal

recognition

particle RNA

(RNAstructure/RNA-proteininteractions)

DAVID SELINGER, XIUBEI

LIAOt,

AND Jo ANN WISEt

DepartmentofBiochemistry,UniversityofIllinoisatUrbana-Champaign, Urbana,IL61801

Communicatedby JoanA. Steitz,February10, 1993 (receivedforreviewJanuary 5, 1993)

ABSTRACT Signal recognition particle (SRP) RNA exhib-itssignificant primary sequence conservation only in domain IV, abulgedhairpin capped by a GNRA (N, anynucleotide; R,

purine)tetranucleotide loop except inplanthomologs. Tetra-loopsconforming tothis sequence or to the consensus UNCG

enhance the stability of synthetic RNA hairpins and have

strikinglysimilar three-dimensionalstructures. To determine the biological relevance of thissimilarity,aswell as to assess the

relativecontributions of sequence and structure to thefunction ofthedomain IV tetraloop,wereplacedtheGAAA sequence in

fissionyeastSRPRNA withUUCG.Haploid strains harboring this substitution are viable, providing experimental evidence for thefunctional equivalenceof thetwotetraloops. Wenext

tested the two sequences found in plant SRP RNAs at this locationforfunction in thecontext of theSchizosaccharomyces pombe RNA. While substitution of CUUC does not allow growth, a viable strain results from replacing GAAA with

UUUC. Although the viable tetraloop substitution mutants

exhibit wild-type growth under normal conditions, all three express conditionaldefects. To determinewhether thismight

be a consequence ofstructural perturbations, we performed enzymaticprobing.The resultsindicate that RNAscontaining

tetraloop substitutions exhibit subtledifferencesfrom the wild

typenotonlyinthetetraloopitself,but also in the3-basepair adjoiningstem.Todirectlyassesstheimportanceof the latter structure, wedisrupted itpartially or completely and made the compensatory mutations to restore the helix. Surprisingly,

mutantRNAs withaslittleas oneWatson-Crick basepair can supportgrowth.

Signal recognition particle (SRP)isanRNA-protein complex that targets ribosomes translating presecretory proteins to

the endoplasmic reticulum membrane (reviewed in ref. 1). The extensively studied canine SRP is composed of six

polypeptides andone300-nucleotide RNA (2, 3). SRPRNA

(alsoreferredto as7SL) hasbeen identified in avariety of organisms (reviewed in ref. 4) and can be folded into a phylogeneticallyconservedsecondarystructureconsistingof four domains:ashortbase-pairedregionatthe5'end(domain I); along central helix that includes the 3' end (domain II); andtwointernal stem-loopstructures,oneextensivelybase

paired (domain III)andonecontainingseveralinternalloops

(domain IV) (nomenclature according to ref. 5). The se-quence, aswellasthesecondary structure,ofdomainIVis conservedin SRP RNA homologsfrom bacteriatohumans (4). This helix terminates in a tetranucleotide loop that conforms to the consensus GNRA (N, any nucleotide; R,

purine)exceptinplant SRP RNAs, which have four pyrim-idines at this location. GNRA and UNCG tetraloops are highlyoverrepresentedinRNAs(6) andarefoundfrequently

Thepublicationcostsof this articleweredefrayed in part by page charge payment. This articlemust thereforebehereby marked "advertisement" in accordance with 18 U.S.C. §1734solelytoindicate this fact.

in catalytic and informational RNAs as well (7, 8). This prevalence was previously proposed to arise from their ability toincreasehairpin stability (9) but may instead bea

consequenceoftheir well-defined three-dimensional confor-mations (10). Recently, solution structures of smallsynthetic

RNAscontaining each of these tetraloopshave been solved by two-dimensional NMR spectroscopy (11, 12). Despite theirdifferent sequences, they adopt quite similarstructures

inwhich the first and fourthbases are hydrogenbonded, the second base has little interaction with the remainder of the loop, and the phosphate backbone between the second and third nucleotides is extended as a consequence of S-type sugarpuckering.

We previously reported the in vivo effects of point

muta-tions in the domain IVtetraloop of fission yeast SRP RNA (nucleotides 160-163; wild-type sequenceGAAA) (13). Both lethal allelesidentified were transversions at G-160, which had been implicated in SRP19 protein binding by RNase

protection studies (14, 15). However, a G at thispositionis also critical to the integrity of the tetraloop, and there is a

strong correlationbetween the phenotypes of the remaining point mutants we examined and predictedperturbations of thestructure.Togainfurtherinsight into the role of domain IV, as well as to assess the relevance of recent in vitro structural datato thesituation in vivo, we analyzed both the effects of en bloc tetraloop substitutions and the conse-quences ofdisrupting and restoring the adjoining stem. Taken together, the results of our studies imply that the in vivo function of this regionisdeterminedby its structure.

EXPERIMENTALPROCEDURES

Materials.Enzymes werepurchased from BRL andNew

England Biolabs; mutagenesis reagents were from Amer-sham; DNA sequencing reagents were from United States Biochemical; RNases were from Pharmacia; and calf

intes-tinalalkalinephosphatase was fromBoehringer Mannheim. Sequencing primers and mutagenic oligonucleotides were

synthesizedattheBiotechnology Center at the University of Illinois. Radiolabeled

[y_32P]ATP

wasfrom ICN.

Site-DirectedMutagenesis.Targetedmutations were intro-duced into the cloned SRP7gene carried on the phagemid pWEC4.2 (16) by standard methods (17, 18) with the follow-ing oligonucleotides: STL1 (5'-ATGTGCATTSC-GAASAACCTCCATC-3'), replaces nucleotides 159-164 with SUUCGS sequences where S is G or C; STL2

(5'-ATTCCGAAGAACCTCCATC-3'), generates a variant not obtained with STL1; PTL1

(5'-TGTGCATTGGAAR-CAACCTCCA-3'), replaces nucleotides 160-163 with the

corresponding segment of plant SRP RNA; PTL2

(5'-Abbreviation: SRP,signalrecognitionparticle.

tPresentaddress: Departmentof MolecularBiology, Research In-stituteofScripps Clinic,LaJolla, CA 92037.

TGTGCATTTGAARCAACCTCCA-3'), places a G-A pair adjacent to the plant tetraloops; M3a

(5'-ATGTGCAT-TG*TT*TCC*AACCTCCAT-3'), createsmutationsat posi-tions 159, 162, and 164(45%degeneracywas allowed atthe

positions marked with an asterisk); BP2 and -3

(5'-GATGTGCAT1T2GTTTCCA3A4CCTCCATCG-3'), creates mismatches in the second and third basepairsflankingthe

tetraloop and changesthe A-UpairstoC-Gpairs (T1 = 50%

T/50%G,T2 = 50%T/50% C,A3 = 17%A/83% G,andA4

= 17% A/83% C); BP-C

(5'-TGATGTGCAGCGTTTC-CGCCC-3'), replaces both A-U pairswith G-C pairs; D4-E

(5'-GATGTGCATTGCGTTTC*CGCAACCTCCATC-3'),

inserts two G-C base pairs into the stem adjacent to the

tetraloop, in the context of either thewild-type tetraloopor

a lethalpointmutant(G16OC) (50%C/50%G at theposition marked withanasterisk).

Yeast Methods. Mutant alleles were confirmed by DNA

sequencing and introduced into Schizosaccharomyces

pombestrain RM2a,heterozygous fordisruptionof the SRP7 gene(19). Transformation, random spore analysis,and

test-ing for sensitivity to high or low temperature and/or in-creasedosmoticstrengthwereperformedasdescribed(13). To determine generation times, cells were grown in rich mediumat30°Cand theirdensitywasmonitoredby counting

in ahemacytometer.

Construction ofthe p77 Plasmid Series. Domain IV was

amplified by25cycles ofPCR under standardconditionswith the two primers D4-PCRI (5'-CTGGCAGTTAGGCCTTG-TAGTACCGA-3'; identical tonucleotides 125-150, except for the underlined changes to create a Stu I site) and

D4-PCRII

(5'-GCACTGCCCAGGATC-CT*GTAGTGATG-3';

complementarytopositions172-196 exceptatthe underlined

nucleotides, which createaBamHIsite,andattheposition

markedwith anasterisk, which restores pairing to the Stu I

site)onpWEC4.2DNAcontaining wild-typeor mutant7SL sequences. The products were phenol extracted, digested

with Stu I and BamHI, and ligated into the same sites in

pDW19(agift of Norman Pace, IndianaUniversity),which has a T7 promoterpositioned such thattranscription starts at the first Gofthe Stu I site.

In Vitro Transcription. BamHI-digested p77 DNA was transcribed with T7 RNA polymeraseaccordingtopublished procedures (20), and a 10-pmol sample was dephosphorylated and5'-end-labeled withpolynucleotidekinase in the presence of[y-32P]ATP.

Enzymatic Structure Probing. Labeled transcripts

(.200,000

cpm) were subjected to partial digestion with

RNases Ti and

Vi,

nuclease S1, and alkali by published procedures (21) with minor modifications.

RESULTS

PhenotypesofAdditionalPoint MutantsAre Consistentwith

theNMRStructure. In ourearlier point mutagenesis of the domain IVtetraloop, wedidnotobtain substitutions at the thirdnucleotide. Usingamorefocusedstrategy, we isolated

A162C, which is viable but has a mild conditional growth defect(Table 1, line 2). This phenotype presumably results

fromdisruptinganinteractionwith the ribose at position 160;

apurine is required at the third position of the tetraloop because itsN-7servesas ahydrogen bond acceptor (12). The three double mutants involving A-162 have more severe

phenotypesthanany of thecomponent single mutants (Table 1, lines3-5; ref. 13), suggesting that these residues function

cooperatively.

A Different

Stabilizing

Tetraloop Can Functionally

Substi-tute for the GAAA Sequence in Domain IV of SRP RNA. Although the severity of the growth defects resulting from

pointmutations in the domainIVGAAAtetranucleotide loop parallel predicted perturbations of the structure, these data do notdefinitively rule outasequence-specific role for the

Table 1.Phenotypes conferredby mutationstargetedtothe domain IVterminal region

Line Allele Sequence Growth OTS

1 Wild type GGAAAC

2 A162C GGACAC Viable +

3 G159U/A162C UGACAC Dead

4 G159U/A162U UGAUAC Dead

5 A162C/C164G GGACAG Viable +++

6 STL1-3 GUUCGC Wildtype +++

7 STL2-1 CUUCGG Viable ++ 8 STL1-1 GUUCGG Dead 9 STL1-2 CUUCGC Dead 10 PTL1-1 GUUUCC Viable ++ 11 PTL1-2 GCUUCC Dead 12 PTL2C GCUUCA Dead 13 PTL 2U GUUUCA Dead

14 D4E-1 UUGCGG...CGCAA Dead

15 D4E-2 UUGCCC... CGCAA Dead

16 D4E-3 UUGG...CGCAA Viable

17 G159C/C164G UUC ...GAA Cold

sensitive + + +

18 G159C/C164A UUC ...AAA Dead

19 G159A UUA... CAA Viable +++

20 G159U UUU...CAA Viable +++

21 C164A UUG...AAA Viable ++

22 G159A/C164U UUA...UAA Viable

23 G159U/C164A UUU...AAA Viable

24 U157G GUG... CAA Viable

25 U158C UCG...CAA Viable

26 A165G UUG...CGA Viable

27 A166C UUG...CAC Viable ++

28 U157G/U158C GCG...CAA Viable

29 U157G/A165G GUG...CGA Viable +

30 U158C/A166C UCG...CAC Dead

31 BP-Comp GCG . CGC Viable

critical residues. The striking similarities between the re-cently determined solution structures of UUCG (11) and GAAA (12)tetranucleotide loops prompted us to ask whether they are interchangeable in vivo. Both UUCG substitution

mutantsinwhich theflanking residues do not form Watson-Crick base pairs are inviable (Table 1, lines 8 and 9). The lethality ofunpaired G residues at positions 159 and 164 in conjunctionwiththeUUCG tetranucleotideloop is

interest-ing, since C164G produced only a mild conditional growth defect in an otherwise wild-typeRNA; the C-C apposition was lethal in combination with the GAAA tetraloop (13), as with UUCG. Most importantly, our data demonstrate that

replacingthewild-typeGAAA loop at positions 160-163 with UUCGproduces an RNA that supports growth as the only

form of 7SL in the cell (Table 1, lines 6 and 7). This observation has two important implications. First, the two

tetraloopsthat adopt similar structures in vitro are function-allyinterchangeable in vivo, at least in this context. Second,

theroleof the domain IV tetraloopin SRP RNA is not base

specific, sincethese sequences have no nucleotides in

com-mon.AlthoughtheUUCG tetraloop mutant has a generation timeindistinguishablefrom that of a wild-type strain under normal conditions (data not shown), its growth is impaired underrestrictive conditions (Table 1 legend). Interestingly, the substitution allele carrying the wild-type G-C base pair hasa moresevereconditional growth defect than the mutant with thereversed C-G base pair.

Only Oneofthe TwoPlantTetranucleotide Loop Sequences

CanSupport GrowthinS. pombe. Inlight of the conservation of the domain IVGAAAtetraloopfrom Escherichia coli to

humans,it wassurprising that thecorresponding sequences in several recently characterized plant SRP RNAs are

wanted to determine whether these tetraloops could, like UUCG, function in the context of S. pombe SRP RNA.

Remarkably, the UUUC replacement mutant is viable, al-though conditionally growth defective (Table 1, line 10). In contrast, the CUUC tetraloop substitution confers a

reces-sive lethal phenotype(Table 1, line 11). Since plant SRP RNA tetraloops are flanked by anoncanonical GA basepair, we attempted to rescue this mutant byreplacing the closing G-C pair with this combination. This substitution notonly failed to supportviability with the CUUC tetraloop but was also lethal incombination with the otherwise viable UUUC mu-tant(Table 1, lines 12 and 13).

In VitroEnzymatic Probing Data Indicate That Tetraloop

SubstitutionMutantswithConditionalGrowthDefectsExhibit Subtle Structural Differences from the Wild-TypeRNA. Al-thoughaUUCGtetraloopcanfunctionally replacethe wild-type GAAA sequenceunder normalconditions, the growth defect of the mutant under extreme conditions suggests that thesubstitutionmaysubtlyperturb the structure. To confirm this hypothesis, we performed in vitro enzymaticprobing on T7 transcripts corresponding to domain IV with either a

GAAAor aUUCGtetraloop.In thesubstitution mutant, the tetraloop was flanked byaC-G base pair,which has a less severeconditionalgrowthdefect in vivo. Eachtranscript was digested with RNase Vi, which cleaves double-stranded RNAandnucleotides involved intertiarystructure(seeref. 22foradiscussion of Vi specificity), and with nucleaseSi, which is specific for single-stranded nucleic acids; some representative results are shown in Fig. 1 and data from severalexperiments are summarized in Fig. 3. The digestion

patternsfor thewild-typetranscript are generally consistent with thestructurededuced from phylogenetic analysis of the

intact RNA, except for a few anomalous nuclease Si cuts produced only at the higher enzyme concentration (Fig. 1A).

Interestingly, not only does the structure of the tetraloop itself change, but the surroundingregion is also altered. In particular, in the C(UUCG)G transcript, RNase

Vi

cuts at

positions corresponding to 156-159 in the full-length RNA, while for the wild-type G(GAAA)C tetraloop and closing base

A~~~~~~~~~~~A A

B_

G156 G168 _ do 0153 0*_*El 153 * , 0152 01G52 * iA _ * G149 G149

FIG. 1. Products of enzymatic structure probing for wild-type

and C(UUCG)G substitutionmutant RNAs. 5'-End-labeled domain

IV transcripts were digested with the enzyme indicated and resolved on a 10% polyacrylamide/8 M urea sequencing gel. Alk, partial alkalinehydrolysis products. In the nucleaseSiand RNaseVilanes,

L andH arelow andhigh concentrations of enzyme. NE, no enzyme control. Productsofpartialdigestion with RNase

Ti

werealso run as a control to allow location of cleavage products in the RNA sequence;G residuesarenumbered according toref. 19. (A)Results forwild-typeRNA. (B) Products from mutant transcript.

pair, onlynucleotides157 and 158showsignificantcleavage. Thesimplest interpretationof theseobservations is that the

C(UUCG)Gsequence promotes strongerbasepairingin the short stem. Within the tetraloops themselves, nuclease Si

cleaves all nucleotides inGAAA butnotinUUCG;the latter sequenceis, in contrast, cleaved by RNase Vi at the second and thirdpositions. Thus,it appears thatthe UUCGtetraloop

adopts a more helix-like conformation than the wild-type

tetraloop.

We carriedoutsimilar nucleaseprobing experimentswith domain IVtranscripts carryingplanttetraloop substitutions todetermine whether structural differences might account for theability of UUUC, butnotCUUC,tofunctionallyreplace

GAAA in fission yeast SRP RNA; some representative

resultsareshown in Fig. 2 and data from severalexperiments

are summarized inFig. 3. Asexpected, the upper part of the domain IV structure showed only minor variations in the sites ofcleavage between the CUUC substitution mutant and the wild-type RNA, while differences in the tetraloop region weremoredramatic. In both RNAscarrying planttetraloops,

RNase Vi cleaves residues within theloop, in common with theC(UUCG)G RNA and incontrast tothe pattern observed with the wild-type transcript. The terminal domain IV stem

is cleavedby both nuclease Si and RNase ViattheU-157 A-166basepair in the CUUC RNA,whileonly Vi cleavesat

these positions in the UUUC mutant. Notably, the latter pattern is the same as in the GAAA and UUCG transcripts, consistent with this mutant's ability to support growth. In addition, the CUUC substitution increases the susceptibility ofpositions 154-156 tocleavage by nuclease Si relative to the viable mutants. Although these bases, which are critical for SRP54 protein binding (23, 24), are predicted to be single stranded, A-154 and G-155 are not accessible to nucleaseSi in the other three transcripts. Thus, the inability of the CUUC tetraloop to function in the context offission yeast SRP RNA maybe due either todestabilizationof the adjoining helix or toaconformational change inthe 5' internal loop, since the tetraloop structure itself is similar to that of the viable substitution mutants.

The Tetraloop Does Not Function Solely to Stabilize the

Adjoining Helix. Since our in vivo phenotypic analysis and in vitro structureprobing data both suggest that a primary role

A Al

I ___

-L 0168 -C41m.t_ G15-_it U. 00_ G156 - l - : G155 _ _l G 153 .* 'm G152 _* *El G149 _ El B L-L L

i

Gt68 *=3 o C = _ . G156_1 G155 A G153 _ w ow * G152 * op,*t G149 44j_ b*,

FIG. 2. Products ofenzymatic structure probing for CUUC and UUUCsubstitutionmutantRNAs. Lanes arelabeled as described in

Fig. 1.(A) Products of partially digesting CUUC mutant RNA. (B) UUUCtranscript.

3' 3' c 5 c 5' u g u g a g a g G-C-A *G-c 1810G-C *G-c *A-U A-u U- u C-G*141 C-g A-U A-u U-Am U-a C-G-144 C-g U- u A- a A U A Co c C-G-149 C-G U-A U-A A-UU A-U *C-G-152 OC-Go A GA153 A GA C A AC A 168G G155 *G G UIG' 156 UIG. *OA- U* A- U-A A-U- A-U-C-G* 159 G-CO AA GA16o G U AA C U A A * Wild-Type CUUCGG 3' c 5' u g AA g AG-CA *G-C A -C A-U * C-G-&A-U -U-Av * C-Gm U-n A AC C 0 C- C-G-AU -AA A-UA OC-GE A GA AC AA AG G -AUIGA *A-UO * C-

GU-*C-GE *C Cm UU. CUU CUUJC 3' C 5' u g a g

AG-c

AG-c HG-c A-u u C-g A-u .C-g u A

A-A

AC C * C -G -U -AA * C-G v A GA AC A AG G AUIG *A-U -*C-Gm *C UA U U * 0 UUUC

FIG.3. Summaryofenzymaticstructure probing data.

Nucleo-tides cleavedbynuclease Si are denoted by triangles and sites of

RNase Vlcutsaredenotedby circles, withtheextentofcleavage

indicatedbythesizeofthesymbol.Positionsthatwerecleavedby

both enzymesare denotedby squares. Datafor the 5' endof the wild-typeandCUUC mutantRNAswereobtainedby resolving the

cleavageproductson15%sequencing gels (datanotshown).

Posi-tionsthatcouldnotberead fromourgelsareindicatedby lowercase

letters. Numberingof GresiduesisasdescribedinFig. 1. of the tetraloop is to stabilize the short adjoining helix, we

decided to test whetherstabilizingthe structureby another

means would serve the same purpose. To this end, we

insertedtwoadditional G-Cbasepairsinto thehelix, either in

combination with the wild-type tetraloop or with a lethal

point mutation, G160C (13).Ifdestabilizationof the helix is

thesolecauseofthefunctionaldefect inthismutant, exten-sionof thehelixis predictedtorescueit. Thisprediction is

not borne out; moreover, extending the helix results in a

lethal phenotype even in combination with the wild-type tetraloop (Table 1,lines 14and 15). Duringthemutagenesis procedure, an unexpected mutant was isolated, D4E-3, in

whichtwo bases were inserted 3' tothetetraloop. Surpris-ingly, this mutant exhibits no growth defect under any

conditiontested(Table 1,line16).Takentogether, these data

indicatethatthelengthofthehelixadjoining thetetraloop is important, althoughsomemodificationsoftheterminal struc-turecanapparently beaccommodated.

MutantswithOnlyaSingleWatson-Crick Base Pair inthe

Helix Adjoining theTetraloop AreViable. To completeour

analysisoftherelativeimportanceofsequence vs. structure inthisregionoffissionyeast SRPRNA,wetested the effects

ofdisruptingthe terminal domainIV helix and restoring it

withadifferentsequence.First,sincetheidentityofthe base

pair adjacent to the tetraloop had a marked effecton the

phenotypeoftheUUCGsubstitutionmutants,weexamined

theconsequencesofjuxtaposing various nucleotides flanking

the wild-type tetraloop. Our results demonstrate that base

pairing atthesepositions iscriticalfor SRPRNAfunction,

sincethelethal phenotypeofG159C(13)canbe rescuedby

combiningit withC164G butnot withC164A

(Table 1,

lines

17 and 18). However, although the

G159C/C164G

double

mutantisviable,itexhibitsa severe

growth

defectatelevated

temperatureandosmoticstrength andisalsounabletogrow

atlow temperature.

Mutating

G-159 toA and UorC-164to

Aproducedconditional growthdefects, whichwere amelio-ratedbyrestoration of basepairing (Table 1,lines

19-23).

The A-U combination exhibited wild-type growth, while

U'A

retainedamild conditional phenotype. Takentogether with

ourearlier data(13), it appears that theidentity oftheresidue

atposition 164 is less critical than thatatposition 159and, in particular, growth defects arising fromapyrimidine at

posi-tion 159 partially persist even upon restoration of base pairing.

Because the length ofthe helix adjoining thetetraloop is phylogenetically conserved (exceptinplants),wealso tested theeffects ofmutatingtheothertwobase

pairs

(positions

157,

158, 165,and 166). Of thefour singlemutantsexamined,only

A166C exhibits a conditional phenotype; the others grow underallconditions testeddespite the lossofaWatson-Crick

basepair(Table 1, lines24-27). Evenmoreremarkable, one

of the three double mutants shows no detectable growth

defect even though both base pairs are disrupted (Table 1,

line 28). A second double mutant in this series,

U157G/

A165G, exhibits a mild conditional phenotype, while the third, U158C/A166C, is lethal(Table 1,lines 29 and 30). The

inviability of the latter mutant can be rescued by changing

both U-157 and A-165 to (G Table 1,line 31). This mutant,in

which both original A*U pairs are replaced with G-C pairs,

exhibits fully wild-type growth. Thus, the identity of the bases in this helix is unimportant for SRP RNAfunction.

DISCUSSION

Ourfinding thataUUCG tetraloop canfunctionally replace

GAAA in SRP RNA isconsistentwithextensive analysis of

phylogenetically diverse 16S rRNA sequences, which

re-vealed that GNRA and UNCG tetraloops are sometimes found substituted en bloc even between closely related organisms (6). However, the viabilityof our UUCGtetraloop

mutants, particularly the allele with aC-Gflanking basepair, conflicts with our earlier conclusion that the 5' nucleotideof

the closing pair and the G residue of the tetraloop were likely tobe sequence-specific components of the SRP19pbinding site (13). This inference was based on the inviability of S. pombe mutants harboring transversions at positions 159 and 160, together with the results of in vitro RNase protection experiments on mammalian components (14, 15). The close correlation between our earlier and present phenotypic data for point mutations in the SRP RNA tetraloop and the recently determined NMR structure (12) suggest instead that the tetraloop is a structural entity and that the lethality of the point mutants G159C, G160C, andG16OU arises from con-formational alterations. This conclusion is reinforced by our finding that the UUCG substitution mutants, in which the tetraloop is predicted to have a structure similar to that of the wild-type GAAA despite its completely different sequence, support growth. The ability ofa UUCG tetraloop to func-tionally replace the wild-type sequence indicates that, if a fission yeast homolog of the SRP19 protein does in fact contact thisregion, it must recognize the ribose-phosphate backbone and not the bases. Consistent with our observa-tions, Zwieb (25) has recently shown by an in vitro assay that replacement of the domain IV GAAA tetraloop in human SRP RNA with UUCG is compatible with SRP19p binding. In contrast to the situation in SRP RNA, not a single residue within the GNRA tetranucleotide in the large rRNA that serves asacritical recognitionelement for the cytotoxin ricin canbe altered without loss of recognition by the protein (26). The datapresented here are incompatible with an earlier report thatpositions 157-160 are part of a tertiary interaction

with nucleotides 63-66 in the S. pombe RNA (27). First, in theviable C(UUCG)G substitution mutant, two of the three

Watson-Crick pairs in the proposed pseudoknot cannot

form. Inaddition, in the U157G/U158C mutant, which

ex-hibits no growth defect under anycondition tested, the other twoproposed pairs are disrupted. Although the potentialfor base pairing between these regions appears to be phyloge-netically conserved (27), we note that, in the fission yeast RNA, the two sequences involved are constrained by other interactions: nucleotides 63-66 are part of the B box required

forRNApolymerase III transcription (15),while nucleotides 157-160 are critical to the structure whose importance we have demonstrated here.

Theviability oftheUUUCsubstitution mutant was initially somewhatsurprising,since this sequence, unlike UUCG and GAAA,isnotoverrepresented in rRNA. However, interac-tion of the U and C at posiinterac-tions 1 and 4, as has been observed inintermolecular duplexes (28), could result in a conforma-tion similar to that of the two tetraloops whose structures have been solved.Consistent with the viability of this mutant, our structure probing data show that the adjoining helix

exhibitsapattern of cleavagesimilar to that of thewild-type and the UUCGsubstitutedRNAanddistinct from that of the inviableCUUC mutant. In the CUUC transcript, the

adjoin-ing stemappears tobe destabilized relative to the wild-type and viable mutant RNAs. While the sites of enzymatic cleavageinbothplant tetraloop transcriptsoverlap morewith those in theUUCGsubstitutionmutant thanin the wild-type RNA, the resemblance is greater between the two viable mutants. Although replacement of the S. pombe SRP RNA

domain IV GAAA with a CUUC tetraloop is lethal, this sequence is presumablyfunctionalin the plant RNAs, since all 11 of the corn (Zea mays) cDNAs sequenced have a

CUUC tetraloop, as do two ofthreewheat (Triticum aesti-vum) cDNAs and the tomato (Lycopersicon esculentum) RNA; only the SRP RNA from cineraria hybrids (Senecio cruentus) has exclusively UUUC at this location (4). The

ability of a CUUC tetraloop to function in the context of plant, but notfissionyeast,SRP RNA may be related to the extra noncanonical GA base pair in the plant domain IV

terminal helix.

The effects ofmutations in the terminal domain IV helix

dependontheir proximityto the tetraloop andability to form

a noncanonical base pair. Disrupting the pair immediately

adjacent to the tetraloop has the most severe effects, while

eliminating eitherorboth ofthe other 2 base pairs is tolerated under normalgrowth conditionswith one exception, which

produces anoncanonical AC pair and a C U juxtaposition.

Although disrupting the closing pair is always deleterious,

onlythose mutants with aC in place of the wild-type G at

position159 areinviable. Even when the G159C mutation is

compensated by C164G, the cells display a severe growth

defectunderrestrictive conditions. Thesignificant

deleteri-ous effect ofa C-G pair flanking theGAAA tetraloop

con-trasts with the less severe phenotype of a C-G relative to a G-C basepair in combinationwith the UUCG tetraloop. The

preferencein the latter case may be related to the fact that RNAhairpins cappedby C(UUCG)G are both more common (6) and more stable (29) than those containing G(UUCG)C. Although the population of GAAA tetraloops in 16S rRNA takenas awhole shows little selectivity regarding the closing base pair (6), we note that at any given location, there is generallyastrongpreference for a particular sequence. The

relativelysevere phenotype of the CGclosing pair mutant in combination with the wild-type tetraloop in SRP RNA

pre-sumablyreflects a structural or sequence requirement that we

donotyet

fully understand,

perhapsrelatedtorecognition by

SRP19p. Binding

of the SRP54

protein,

which also interacts

withdomainIV,is

unaffected

byreversal of theclosingbase

pair orby point mutationswithin thetetraloop (23, 24)but is reduced by mutations that disrupt the stem(23).Thestability

of this helix may be critical formaintaining the 5' internal

loop, which is recognized in a sequence-specific mannerby

SRP54p (23, 24), in a productiveconformation.

In summary, the data presented here, together with our earliermutagenesis results(13),indicate that thefunction of the domain IV GAAA tetranucleotide loop and adjoining

stem in SRP RNA is to promote formation of a particular

structure, which can be adopted by several dramatically

different primary sequences. In addition to defining the structural features required for afunctional RNA, thesedata

impose constraints on the properties offactors that interact withthis region. The imperfect correlation betweenstability of the region, which appears to be the major determinant of SRP54p binding(23), and the phenotypes of mutants, sug-geststhat the tetraloopregion does indeed interact with other cellular components, among which may be the SRP19 pro-tein, in a functionally important manner.

We are grateful to OlkeUhlenbeck andArtPardi(Universityof Colorado) for helpfuldiscussions and forreadinganearlierversion

of this manuscript. We thankClaudiaReichand SteveAlthofffor

critical reading ofboth the original andrevised versions. We ac-knowledgetheeffortsofMinMa in creating the UUCGmutantseries and Anne Chiang in assaying the plant tetraloop mutants. This research wassupportedbyNationalScienceFoundation GrantDCB 88-16325 awardedto J.A.W. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29.

Walter, P. & Lingappa, V. (1986) Annu. Rev.CellBiol. 2, 499-516. Walter, P. & Blobel, G. (1980) Proc. Natl.Acad. Sci. USA 77, 7112-7116.

Walter, P. & Blobel, G. (1982) Nature (London) 299, 691-698. Larsen, N. & Zwieb, C. (1991) Nucleic Acids Res. 19, 209-215. Poritz, M. A., Strub, K. & Walter, P. (1988) Cell 55, 4-6. Woese, C. R., Winker, S. &Guttell,R. R.(1990) Proc.Natl. Acad. Sci. USA 87,8467-8471.

Jacquier, A. & Michel, F. (1987)Cell50,17-29.

Tuerk, C., Gauss, P., Thermes, C., Groebe, D. R., Gayle, M., Guild, N., Stormo, G., d'Aubenton-Carafa, Y., Uhlenbeck,0.C., Tinoco,I., Jr.,Brody, E. & Gold, L. (1988) Proc.Natl. Acad. Sci. USA85,1364-1368.

Uhlenbeck, 0.C. (1990) Nature (London) 346,613-614. SantaLucia,J.,Jr., Kierzek, R. &Turner,D. H. (1992)Science 256, 217-219.

Cheong,C., Varani, G. & Tinoco, I., Jr. (1990) Nature(London)

346,680-682.

Heus, H. A. & Pardi, A. (1991) Science 253, 191-194.

Liao, X.,Brennwald, P. & Wise, J. A. (1989) Proc.Natl.Acad. Sci. USA86,4137-4141.

Siegel,V. & Walter, P. (1988) Proc. Natl. Acad. Sci. USA85, 1801-1805.

Poritz, M. A.,Siegel, V., Hansen, W. B. & Walter, P. (1988) Proc. Natl. Acad. Sci. USA85,4315-4319.

Liao, X.,Selinger, D. A., Althoff, S., Chiang, A., Hamilton, D. & Wise,J. A.(1992) Nucleic Acids Res. 20, 1607-1615.

Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA82,488-492. Taylor, J. W., Ott, J. & Eckstein, F. (1985) Nucleic Acids Res. 13, 8765-8785.

Brennwald,P., Liao, X., Holm, K., Porter,G.& Wise, J. A. (1988) Mol. Cell.Biol. 8, 1580-1590.

Yisraeli, J. K. & Melton, D. A. (1989) Methods Enzymol. 180, 42-51.

Knapp, G. (1989) Methods Enzymol. 180, 192-212.

Lowman, H. B. & Draper, D. E. (1986) J. Biol. Chem. 261, 53%-5403.

Selinger,D. A., Brennwald, P. J., Liao, X. & Wise, J. A. (1993) Mol.Cell. Biol. 13, 1353-1362.

Wood, H., Lurink, J. &Tollervey, D. (1992) Nucleic Acids Res. 20, 5919-5925.

Zwieb, C. (1992) J. Biol. Chem. 267, 15650-15656.

Gluck, A., Endo, Y. & Wool, I. G.(1992)J. Mol. Biol.226,411-424. Andreazzoli,M. & Gerbi, S. (1991) EMBO J. 10,767-777. Holbrook, S. R., Cheong, C., Tinoco, I., Jr., & Kim, S.-H. (1991) Nature (London) 353,579-581.

Antao, V. P.,Lai, S. Y. & Tinoco, I., Jr. (1991) Nucleic Acids Res. 19,5901-5905.