Proc. Natl. Acad. Sci. USAVol. 90, pp. 5409-5413, June 1993Biochemistry

Functional interchangeability of the structurally similartetranucleotide loops GAAA and UUCG in fission yeastsignal recognition particle RNA

(RNA structure/RNA-protein interactions)

DAVID SELINGER, XIUBEI LIAOt, AND Jo ANN WISEtDepartment of Biochemistry, University of Illinois at Urbana-Champaign, Urbana, IL 61801

Communicated by Joan A. Steitz, February 10, 1993 (received for review January 5, 1993)

ABSTRACT Signal recognition particle (SRP) RNA exhib-its significant primary sequence conservation only in domainIV, a bulged hairpin capped by a GNRA (N, any nucleotide; R,purine) tetranucleotide loop except in plant homologs. Tetra-loops conforming to this sequence or to the consensus UNCGenhance the stability of synthetic RNA hairpins and havestrikingly similar three-dimensional structures. To determinethe biological relevance of this similarity, as well as to assess therelative contributions of sequence and structure to the functionof the domain IV tetraloop, we replaced the GAAA sequence infission yeast SRP RNA with UUCG. Haploid strains harboringthis substitution are viable, providing experimental evidencefor the functional equivalence of the two tetraloops. We nexttested the two sequences found in plant SRP RNAs at thislocation for function in the context of the Schizosaccharomycespombe RNA. While substitution of CUUC does not allowgrowth, a viable strain results from replacing GAAA withUUUC. Although the viable tetraloop substitution mutantsexhibit wild-type growth under normal conditions, all threeexpress conditional defects. To determine whether this mightbe a consequence of structural perturbations, we performedenzymatic probing. The results indicate that RNAs containingtetraloop substitutions exhibit subtle differences from the wildtype not only in the tetraloop itself, but also in the 3-base pairadjoining stem. To directly assess the importance of the latterstructure, we disrupted it partially or completely and made thecompensatory mutations to restore the helix. Surprisingly,mutant RNAs with as little as one Watson-Crick base pair cansupport growth.

Signal recognition particle (SRP) is an RNA-protein complexthat targets ribosomes translating presecretory proteins tothe endoplasmic reticulum membrane (reviewed in ref. 1).The extensively studied canine SRP is composed of sixpolypeptides and one 300-nucleotide RNA (2, 3). SRP RNA(also referred to as 7SL) has been identified in a variety oforganisms (reviewed in ref. 4) and can be folded into aphylogenetically conserved secondary structure consisting offour domains: a short base-paired region at the 5' end (domainI); a long central helix that includes the 3' end (domain II);and two internal stem-loop structures, one extensively basepaired (domain III) and one containing several internal loops(domain IV) (nomenclature according to ref. 5). The se-quence, as well as the secondary structure, of domain IV isconserved in SRP RNA homologs from bacteria to humans(4). This helix terminates in a tetranucleotide loop thatconforms to the consensus GNRA (N, any nucleotide; R,purine) except in plant SRP RNAs, which have four pyrim-idines at this location. GNRA and UNCG tetraloops arehighly overrepresented in RNAs (6) and are found frequently

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.

in catalytic and informational RNAs as well (7, 8). Thisprevalence was previously proposed to arise from theirability to increase hairpin stability (9) but may instead be aconsequence of their well-defined three-dimensional confor-mations (10). Recently, solution structures of small syntheticRNAs containing each of these tetraloops have been solvedby two-dimensional NMR spectroscopy (11, 12). Despitetheir different sequences, they adopt quite similar structuresin which the first and fourth bases are hydrogen bonded, thesecond base has little interaction with the remainder of theloop, and the phosphate backbone between the second andthird nucleotides is extended as a consequence of S-typesugar puckering.We previously reported the in vivo effects of point muta-

tions in the domain IV tetraloop of fission yeast SRP RNA(nucleotides 160-163; wild-type sequence GAAA) (13). Bothlethal alleles identified were transversions at G-160, whichhad been implicated in SRP19 protein binding by RNaseprotection studies (14, 15). However, a G at this position isalso critical to the integrity of the tetraloop, and there is astrong correlation between the phenotypes of the remainingpoint mutants we examined and predicted perturbations ofthe structure. To gain further insight into the role of domainIV, as well as to assess the relevance of recent in vitrostructural data to the situation in vivo, we analyzed both theeffects of en bloc tetraloop substitutions and the conse-quences ofdisrupting and restoring the adjoining stem. Takentogether, the results of our studies imply that the in vivofunction of this region is determined by its structure.

EXPERIMENTAL PROCEDURESMaterials. Enzymes were purchased from BRL and New

England Biolabs; mutagenesis reagents were from Amer-sham; DNA sequencing reagents were from United StatesBiochemical; RNases were from Pharmacia; and calf intes-tinal alkaline phosphatase was from Boehringer Mannheim.Sequencing primers and mutagenic oligonucleotides weresynthesized at the Biotechnology Center at the University ofIllinois. Radiolabeled [y_32P]ATP was from ICN.

Site-Directed Mutagenesis. Targeted mutations were intro-duced into the cloned SRP7 gene carried on the phagemidpWEC4.2 (16) by standard methods (17, 18) with the follow-ing oligonucleotides: STL1 (5'-ATGTGCATTSC-GAASAACCTCCATC-3'), replaces nucleotides 159-164with SUUCGS sequences where S is G or C; STL2 (5'-ATTCCGAAGAACCTCCATC-3'), generates a variant notobtained with STL1; PTL1 (5'-TGTGCATTGGAAR-CAACCTCCA-3'), replaces nucleotides 160-163 with thecorresponding segment of plant SRP RNA; PTL2 (5'-

Abbreviation: SRP, signal recognition particle.tPresent address: Department of Molecular Biology, Research In-stitute of Scripps Clinic, La Jolla, CA 92037.1To whom reprint requests should be addressed.

5409

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

0

5410 Biochemistry: Selinger et al.

TGTGCATTTGAARCAACCTCCA-3'), places a G-A pairadjacent to the plant tetraloops; M3a (5'-ATGTGCAT-TG*TT*TCC*AACCTCCAT-3'), creates mutations at posi-tions 159, 162, and 164 (45% degeneracy was allowed at thepositions marked with an asterisk); BP2 and -3 (5'-GATGTGCAT1T2GTTTCCA3A4CCTCCATCG-3'), createsmismatches in the second and third base pairs flanking thetetraloop and changes the A-U pairs to C-G pairs (T1 = 50%T/50% G, T2 = 50% T/50% C, A3 = 17% A/83% G, and A4= 17% A/83% C); BP-C (5'-TGATGTGCAGCGTTTC-CGCCC-3'), replaces both A-U pairs with G-C pairs; D4-E(5'-GATGTGCATTGCGTTTC*CGCAACCTCCATC-3'),inserts two G-C base pairs into the stem adjacent to thetetraloop, in the context of either the wild-type tetraloop ora lethal point mutant (G16OC) (50% C/50% G at the positionmarked with an asterisk).

Yeast Methods. Mutant alleles were confirmed by DNAsequencing and introduced into Schizosaccharomycespombe strain RM2a, heterozygous for disruption of the SRP7gene (19). Transformation, random spore analysis, and test-ing for sensitivity to high or low temperature and/or in-creased osmotic strength were performed as described (13).To determine generation times, cells were grown in richmedium at 30°C and their density was monitored by countingin a hemacytometer.

Construction of the p77 Plasmid Series. Domain IV wasamplified by 25 cycles ofPCR under standard conditions withthe two primers D4-PCRI (5'-CTGGCAGTTAGGCCTTG-TAGTACCGA-3'; identical to nucleotides 125-150, exceptfor the underlined changes to create a Stu I site) and D4-PCRII (5'-GCACTGCCCAGGATC-CT*GTAGTGATG-3';complementary to positions 172-196 except at the underlinednucleotides, which create a BamHI site, and at the positionmarked with an asterisk, which restores pairing to the Stu Isite) on pWEC4.2 DNA containing wild-type or mutant 7SLsequences. The products were phenol extracted, digestedwith Stu I and BamHI, and ligated into the same sites inpDW19 (a gift of Norman Pace, Indiana University), whichhas a T7 promoter positioned such that transcription starts atthe first G of the Stu I site.In Vitro Transcription. BamHI-digested p77 DNA was

transcribed with T7 RNA polymerase according to publishedprocedures (20), and a 10-pmol sample was dephosphorylatedand 5'-end-labeled with polynucleotide kinase in the presenceof [y-32P]ATP.Enzymatic Structure Probing. Labeled transcripts

(.200,000 cpm) were subjected to partial digestion withRNases Ti and Vi, nuclease S1, and alkali by publishedprocedures (21) with minor modifications.

RESULTSPhenotypes of Additional Point Mutants Are Consistent with

the NMR Structure. In our earlier point mutagenesis of thedomain IV tetraloop, we did not obtain substitutions at thethird nucleotide. Using a more focused strategy, we isolatedA162C, which is viable but has a mild conditional growthdefect (Table 1, line 2). This phenotype presumably resultsfrom disrupting an interaction with the ribose at position 160;a purine is required at the third position of the tetraloopbecause its N-7 serves as a hydrogen bond acceptor (12). Thethree double mutants involving A-162 have more severephenotypes than any of the component single mutants (Table1, lines 3-5; ref. 13), suggesting that these residues functioncooperatively.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 frompoint mutations in the domain IV GAAA tetranucleotide loopparallel predicted perturbations of the structure, these datado not definitively rule out a sequence-specific role for the

Table 1. Phenotypes conferred by mutations targeted to thedomain IV terminal regionLine Allele Sequence Growth OTS1 Wild type GGAAAC2 A162C GGACAC Viable +3 G159U/A162C UGACAC Dead4 G159U/A162U UGAUAC Dead5 A162C/C164G GGACAG Viable +++6 STL1-3 GUUCGC Wild type +++7 STL2-1 CUUCGG Viable ++8 STL1-1 GUUCGG Dead9 STL1-2 CUUCGC Dead10 PTL1-1 GUUUCC Viable ++11 PTL1-2 GCUUCC Dead12 PTL 2C GCUUCA Dead13 PTL 2U GUUUCA Dead14 D4E-1 UUGCGG ... CGCAA Dead15 D4E-2 UUGCCC ... CGCAA Dead16 D4E-3 UUGG... CGCAA Viable17 G159C/C164G UUC ... GAA Cold

sensitive + + +18 G159C/C164A UUC ... AAA Dead19 G159A UUA... CAA Viable +++20 G159U UUU ... CAA Viable +++21 C164A UUG... AAA Viable ++22 G159A/C164U UUA ... UAA Viable23 G159U/C164A UUU ... AAA Viable24 U157G GUG... CAA Viable25 U158C UCG ... CAA Viable26 A165G UUG... CGA Viable27 A166C UUG ... CAC Viable ++28 U157G/U158C GCG. .. CAA Viable29 U157G/A165G GUG ... CGA Viable +30 U158C/A166C UCG ... CAC Dead31 BP-Comp GCG . CGC Viable

critical residues. The striking similarities between the re-cently determined solution structures of UUCG (11) andGAAA (12) tetranucleotide loops prompted us to ask whetherthey are interchangeable in vivo. Both UUCG substitutionmutants in which the flanking residues do not form Watson-Crick base pairs are inviable (Table 1, lines 8 and 9). Thelethality of unpaired G residues at positions 159 and 164 inconjunction with the UUCG tetranucleotide loop is interest-ing, since C164G produced only a mild conditional growthdefect in an otherwise wild-type RNA; the C-C appositionwas lethal in combination with the GAAA tetraloop (13), aswith UUCG. Most importantly, our data demonstrate thatreplacing the wild-type GAAA loop at positions 160-163 withUUCG produces an RNA that supports growth as the onlyform of 7SL in the cell (Table 1, lines 6 and 7). Thisobservation has two important implications. First, the twotetraloops that adopt similar structures in vitro are function-ally interchangeable in vivo, at least in this context. Second,the role of the domain IV tetraloop in SRP RNA is not basespecific, since these sequences have no nucleotides in com-mon. Although the UUCG tetraloop mutant has a generationtime indistinguishable from that of a wild-type strain undernormal conditions (data not shown), its growth is impairedunder restrictive conditions (Table 1 legend). Interestingly,the substitution allele carrying the wild-type G-C base pairhas a more severe conditional growth defect than the mutantwith the reversed C-G base pair.Only One of the Two Plant Tetranucleotide Loop Sequences

Can Support Growth in S. pombe. In light of the conservationof the domain IV GAAA tetraloop from Escherichia coli tohumans, it was surprising that the corresponding sequencesin several recently characterized plant SRP RNAs are ex-clusively pyrimidine (reviewed in ref. 4). We therefore

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 5411

wanted to determine whether these tetraloops could, likeUUCG, 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). Incontrast, the CUUC tetraloop substitution confers a reces-sive lethal phenotype (Table 1, line 11). Since plant SRPRNAtetraloops are flanked by a noncanonical GA base pair, weattempted to rescue this mutant by replacing the closing G-Cpair with this combination. This substitution not only failedto support viability with the CUUC tetraloop but was alsolethal in combination with the otherwise viable UUUC mu-tant (Table 1, lines 12 and 13).

In Vitro Enzymatic Probing Data Indicate That TetraloopSubstitution Mutants with Conditional Growth Defects ExhibitSubtle Structural Differences from the Wild-Type RNA. Al-though a UUCG tetraloop can functionally replace the wild-type GAAA sequence under normal conditions, the growthdefect of the mutant under extreme conditions suggests thatthe substitution may subtly perturb the structure. To confirmthis hypothesis, we performed in vitro enzymatic probing onT7 transcripts corresponding to domain IV with either aGAAA or a UUCG tetraloop. In the substitution mutant, thetetraloop was flanked by a C-G base pair, which has a lesssevere conditional growth defect in vivo. Each transcript wasdigested with RNase Vi, which cleaves double-strandedRNA and nucleotides involved in tertiary structure (see ref.22 for a discussion of Vi specificity), and with nuclease Si,which is specific for single-stranded nucleic acids; somerepresentative results are shown in Fig. 1 and data fromseveral experiments are summarized in Fig. 3. The digestionpatterns for the wild-type transcript are generally consistentwith the structure deduced from phylogenetic analysis of theintact RNA, except for a few anomalous nuclease Si cutsproduced only at the higher enzyme concentration (Fig. 1A).Interestingly, not only does the structure of the tetraloopitself change, but the surrounding region is also altered. Inparticular, in the C(UUCG)G transcript, RNase Vi cuts atpositions 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 El0*_* 153 * ,0152 01G52 *

iA _ *

G149 G14 9

FIG. 1. Products of enzymatic structure probing for wild-typeand C(UUCG)G substitution mutant RNAs. 5'-End-labeled domainIV transcripts were digested with the enzyme indicated and resolvedon a 10% polyacrylamide/8 M urea sequencing gel. Alk, partialalkaline hydrolysis products. In the nuclease Si and RNase Vi lanes,L and H are low and high concentrations ofenzyme. NE, no enzymecontrol. Products of partial digestion with RNase Ti were also runas a control to allow location of cleavage products in the RNAsequence; G residues are numbered according to ref. 19. (A) Resultsfor wild-type RNA. (B) Products from mutant transcript.

pair, only nucleotides 157 and 158 show significant cleavage.The simplest interpretation of these observations is that theC(UUCG)G sequence promotes stronger base pairing in theshort stem. Within the tetraloops themselves, nuclease Sicleaves all nucleotides in GAAA but not in UUCG; the lattersequence is, in contrast, cleaved by RNase Vi at the secondand third positions. Thus, it appears that the UUCG tetraloopadopts a more helix-like conformation than the wild-typetetraloop.We carried out similar nuclease probing experiments with

domain IV transcripts carrying plant tetraloop substitutionsto determine whether structural differences might account forthe ability of UUUC, but not CUUC, to functionally replaceGAAA in fission yeast SRP RNA; some representativeresults are shown in Fig. 2 and data from several experimentsare summarized in Fig. 3. As expected, the upper part of thedomain IV structure showed only minor variations in the sitesof cleavage between the CUUC substitution mutant and thewild-type RNA, while differences in the tetraloop regionwere more dramatic. In both RNAs carrying plant tetraloops,RNase Vi cleaves residues within the loop, in common withthe C(UUCG)G RNA and in contrast to the pattern observedwith the wild-type transcript. The terminal domain IV stemis cleaved by both nuclease Si and RNase Vi at the U-157 A-166 base pair in the CUUC RNA, while only Vi cleaves atthese positions in the UUUC mutant. Notably, the latterpattern is the same as in the GAAA and UUCG transcripts,consistent with this mutant's ability to support growth. Inaddition, the CUUC substitution increases the susceptibilityofpositions 154-156 to cleavage by nuclease Si relative to theviable mutants. Although these bases, which are critical forSRP54 protein binding (23, 24), are predicted to be singlestranded, A-154 and G-155 are not accessible to nuclease Siin the other three transcripts. Thus, the inability ofthe CUUCtetraloop to function in the context offission yeast SRP RNAmay be due either to destabilization of the adjoining helix orto a conformational change in the 5' internal loop, since thetetraloop structure itself is similar to that of the viablesubstitution mutants.The Tetraloop Does Not Function Solely to Stabilize the

Adjoining Helix. Since our in vivo phenotypic analysis and invitro structure probing data both suggest that a primary role

A I___Al -L

0168

-C41m.t_

G15-_it U. 00_

G156 - l - :G155 _ _l

G153 .* 'mG152 _* *El

G149 _ El

B L-L L i

Gt68 *=3

o C = _ .

G156_ 1G155 A

G153 _ w ow *G152 * op,*t

G149 44j_ b*,

FIG. 2. Products of enzymatic structure probing for CUUC andUUUC substitution mutant RNAs. Lanes are labeled as described inFig. 1. (A) Products of partially digesting CUUC mutant RNA. (B)UUUC transcript.

Biochemistry: Selinger et al.

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

0

5412 Biochemistry: Selinger et al.

3' 3'c 5 c 5'

u g u g

a g a g

G-C-A *G-c1810G -C *G-c

*A-U A-uU- u

C-G*141 C -g

A-U A -u

U-Am U-aC-G-144 C-g

U- u

A- aA U A

Co cC-G-149 C-GU -A U -AA-UU A-U

*C-G-152 OC-GoA GA153 A GA

C A AC A168 G G 155 *G G

UIG' 156 UIG.*OA- U * A - U-AA-U- A-U-C - G * 159 G-CO

AA GA16o G UAA C U

A A *

Wild-Type CUUCGG

3'c 5'u g

AA g

AG-CA* G-CA

-C

A-U

* C-G-& A-U -U -Av

* C - G m

U-n

A ACC 0

C-C-G-

AU -AA

A-UA

OC-GEA 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-cAG - cHG-c

A - uu

C - gA - u

. C-g

u

A A-AAC

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. Summary of enzymatic structure probing data. Nucleo-tides cleaved by nuclease Si are denoted by triangles and sites ofRNase Vl cuts are denoted by circles, with the extent of cleavageindicated by the size of the symbol. Positions that were cleaved byboth enzymes are denoted by squares. Data for the 5' end of thewild-type and CUUC mutant RNAs were obtained by resolving thecleavage products on 15% sequencing gels (data not shown). Posi-tions that could not be read from our gels are indicated by lowercaseletters. Numbering of G residues is as described in Fig. 1.

of the tetraloop is to stabilize the short adjoining helix, wedecided to test whether stabilizing the structure by anothermeans would serve the same purpose. To this end, weinserted two additional G-C base pairs into the helix, either incombination with the wild-type tetraloop or with a lethalpoint mutation, G160C (13). If destabilization of the helix isthe sole cause of the functional defect in this mutant, exten-sion of the helix is predicted to rescue it. This prediction isnot borne out; moreover, extending the helix results in alethal phenotype even in combination with the wild-typetetraloop (Table 1, lines 14 and 15). During the mutagenesisprocedure, an unexpected mutant was isolated, D4E-3, inwhich two bases were inserted 3' to the tetraloop. Surpris-ingly, this mutant exhibits no growth defect under anycondition tested (Table 1, line 16). Taken together, these dataindicate that the length of the helix adjoining the tetraloop isimportant, although some modifications ofthe terminal struc-ture can apparently be accommodated.Mutants with Only a Single Watson-Crick Base Pair in the

Helix Adjoining the Tetraloop Are Viable. To complete our

analysis of the relative importance of sequence vs. structurein this region of fission yeast SRP RNA, we tested the effectsof disrupting the terminal domain IV helix and restoring itwith a different sequence. First, since the identity of the basepair adjacent to the tetraloop had a marked effect on thephenotype of the UUCG substitution mutants, we examinedthe consequences ofjuxtaposing various nucleotides flankingthe wild-type tetraloop. Our results demonstrate that basepairing at these positions is critical for SRP RNA function,since the lethal phenotype of G159C (13) can be rescued by

combining it with C164G but not with C164A (Table 1, lines17 and 18). However, although the G159C/C164G doublemutant is viable, it exhibits a severe growth defect at elevatedtemperature and osmotic strength and is also unable to growat low temperature. Mutating G-159 to A and U or C-164 toA produced conditional growth defects, which were amelio-rated by restoration of base pairing (Table 1, lines 19-23). TheA-U combination exhibited wild-type growth, while U'Aretained a mild conditional phenotype. Taken together withour earlier data (13), it appears that the identity of the residueat position 164 is less critical than that at position 159 and, inparticular, growth defects arising from a pyrimidine at posi-tion 159 partially persist even upon restoration of basepairing.Because the length of the helix adjoining the tetraloop is

phylogenetically conserved (except in plants), we also testedthe effects of mutating the other two base pairs (positions 157,158, 165, and 166). Of the four single mutants examined, onlyA166C exhibits a conditional phenotype; the others growunder all conditions tested despite the loss of a Watson-Crickbase pair (Table 1, lines 24-27). Even more remarkable, oneof the three double mutants shows no detectable growthdefect 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 thethird, U158C/A166C, is lethal (Table 1, lines 29 and 30). Theinviability of the latter mutant can be rescued by changingboth U-157 and A-165 to (G Table 1, line 31). This mutant, inwhich both original A*U pairs are replaced with G-C pairs,exhibits fully wild-type growth. Thus, the identity of thebases in this helix is unimportant for SRP RNA function.

DISCUSSIONOur finding that a UUCG tetraloop can functionally replaceGAAA in SRP RNA is consistent with extensive analysis ofphylogenetically diverse 16S rRNA sequences, which re-vealed that GNRA and UNCG tetraloops are sometimesfound substituted en bloc even between closely relatedorganisms (6). However, the viability of our UUCG tetraloopmutants, particularly the allele with a C-G flanking base pair,conflicts with our earlier conclusion that the 5' nucleotide ofthe closing pair and the G residue of the tetraloop were likelyto be sequence-specific components of the SRP19p bindingsite (13). This inference was based on the inviability of S.pombe mutants harboring transversions at positions 159 and160, together with the results of in vitro RNase protectionexperiments on mammalian components (14, 15). The closecorrelation between our earlier and present phenotypic datafor point mutations in the SRP RNA tetraloop and therecently determined NMR structure (12) suggest instead thatthe tetraloop is a structural entity and that the lethality of thepoint mutants G159C, G160C, and G16OU arises from con-formational alterations. This conclusion is reinforced by ourfinding that the UUCG substitution mutants, in which thetetraloop is predicted to have a structure similar to that ofthewild-type GAAA despite its completely different sequence,support growth. The ability of a UUCG tetraloop to func-tionally replace the wild-type sequence indicates that, if afission yeast homolog of the SRP19 protein does in factcontact this region, it must recognize the ribose-phosphatebackbone and not the bases. Consistent with our observa-tions, Zwieb (25) has recently shown by an in vitro assay thatreplacement of the domain IV GAAA tetraloop in humanSRP RNA with UUCG is compatible with SRP19p binding.In contrast to the situation in SRP RNA, not a single residuewithin the GNRA tetranucleotide in the large rRNA thatserves as a critical recognition element for the cytotoxin ricincan be altered without loss of recognition by the protein (26).The data presented here are incompatible with an earlier

report that positions 157-160 are part of a tertiary interaction

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 5413

with nucleotides 63-66 in the S. pombe RNA (27). First, inthe viable C(UUCG)G substitution mutant, two of the threeWatson-Crick pairs in the proposed pseudoknot cannotform. In addition, in the U157G/U158C mutant, which ex-hibits no growth defect under any condition tested, the othertwo proposed pairs are disrupted. Although the potential forbase pairing between these regions appears to be phyloge-netically conserved (27), we note that, in the fission yeastRNA, the two sequences involved are constrained by otherinteractions: nucleotides 63-66 are part of the B box requiredfor RNA polymerase III transcription (15), while nucleotides157-160 are critical to the structure whose importance wehave demonstrated here.The viability of the UUUC substitution mutant was initially

somewhat surprising, since this sequence, unlike UUCG andGAAA, is not overrepresented in rRNA. However, interac-tion of the U and C at positions 1 and 4, as has been observedin intermolecular duplexes (28), could result in a conforma-tion similar to that of the two tetraloops whose structureshave been solved. Consistent with the viability of this mutant,our structure probing data show that the adjoining helixexhibits a pattern of cleavage similar to that of the wild-typeand the UUCG substituted RNA and distinct from that of theinviable CUUC mutant. In the CUUC transcript, the adjoin-ing stem appears to be destabilized relative to the wild-typeand viable mutant RNAs. While the sites of enzymaticcleavage in both plant tetraloop transcripts overlap more withthose in the UUCG substitution mutant than in the wild-typeRNA, the resemblance is greater between the two viablemutants. Although replacement of the S. pombe SRP RNAdomain IV GAAA with a CUUC tetraloop is lethal, thissequence is presumably functional in the plant RNAs, sinceall 11 of the corn (Zea mays) cDNAs sequenced have aCUUC tetraloop, as do two of three wheat (Triticum aesti-vum) cDNAs and the tomato (Lycopersicon esculentum)RNA; only the SRP RNA from cineraria hybrids (Seneciocruentus) has exclusively UUUC at this location (4). Theability of a CUUC tetraloop to function in the context ofplant, but not fission yeast, SRP RNA may be related to theextra noncanonical GA base pair in the plant domain IVterminal helix.The effects of mutations in the terminal domain IV helix

depend on their proximity to the tetraloop and ability to forma noncanonical base pair. Disrupting the pair immediatelyadjacent to the tetraloop has the most severe effects, whileeliminating either or both of the other 2 base pairs is toleratedunder normal growth conditions with one exception, whichproduces a noncanonical AC pair and a C U juxtaposition.Although disrupting the closing pair is always deleterious,only those mutants with a C in place of the wild-type G atposition 159 are inviable. Even when the G159C mutation iscompensated by C164G, the cells display a severe growthdefect under restrictive conditions. The significant deleteri-ous effect of a C-G pair flanking the GAAA tetraloop con-trasts with the less severe phenotype of a C-G relative to aG-C base pair in combination with the UUCG tetraloop. Thepreference in the latter case may be related to the fact thatRNA hairpins capped by 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 rRNAtaken as a whole shows little selectivity regarding the closingbase pair (6), we note that at any given location, there isgenerally a strong preference for a particular sequence. Therelatively severe phenotype of the CG closing pair mutant incombination with the wild-type tetraloop in SRP RNA pre-sumably reflects a structural or sequence requirement that wedo not yet fully understand, perhaps related to recognition bySRP19p. Binding of the SRP54 protein, which also interactswith domain IV, is unaffected by reversal of the closing base

pair or by point mutations within the tetraloop (23, 24) but isreduced by mutations that disrupt the stem (23). The stabilityof this helix may be critical for maintaining the 5' internalloop, which is recognized in a sequence-specific manner bySRP54p (23, 24), in a productive conformation.

In summary, the data presented here, together with ourearlier mutagenesis results (13), indicate that the function ofthe domain IV GAAA tetranucleotide loop and adjoiningstem in SRP RNA is to promote formation of a particularstructure, which can be adopted by several dramaticallydifferent primary sequences. In addition to defining thestructural features required for a functional RNA, these dataimpose constraints on the properties of factors that interactwith this region. The imperfect correlation between stabilityof the region, which appears to be the major determinant ofSRP54p binding (23), and the phenotypes of mutants, sug-gests that the tetraloop region does indeed interact with othercellular components, among which may be the SRP19 pro-tein, in a functionally important manner.

We are grateful to Olke Uhlenbeck and Art Pardi (University ofColorado) for helpful discussions and for reading an earlier versionof this manuscript. We thank Claudia Reich and Steve Althoff forcritical reading of both the original and revised versions. We ac-knowledge the efforts ofMin Ma in creating the UUCG mutant seriesand Anne Chiang in assaying the plant tetraloop mutants. Thisresearch was supported by National Science Foundation Grant DCB88-16325 awarded to 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. Cell Biol. 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) Cell 50, 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.USA 85, 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.USA 86, 4137-4141.Siegel, V. & Walter, P. (1988) Proc. Natl. Acad. Sci. USA 85,1801-1805.Poritz, M. A., Siegel, V., Hansen, W. B. & Walter, P. (1988) Proc.Natl. Acad. Sci. USA 85, 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. USA 82, 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.

Biochemistry: Selinger et al.

Dow

nloa

ded

by g

uest

on

Apr

il 8,

202

0