Proc. Natl. Acad. Sci. USA Vol. 90, pp. 5409-5413, June 1993 Biochemistry Functional interchangeability of the structurally similar tetranucleotide loops GAAA and UUCG in fission yeast signal recognition particle RNA (RNA structure/RNA- interactions) DAVID SELINGER, XIUBEI LIAOt, AND Jo ANN WISEt Department of Biochemistry, University of Illinois at Urbana-Champaign, Urbana, IL 61801 Communicated by Joan A. Steitz, February 10, 1993 (receivedfor review January 5, 1993)

ABSTRACT Signal recognition particle (SRP) RNA exhib- in catalytic and informational as well (7, 8). This its significant primary sequence conservation only in domain prevalence was previously proposed to arise from their IV, a bulged hairpin capped by a GNRA (N, any nucleotide; R, ability to increase hairpin stability (9) but may instead be a purine) tetranucleotide loop except in plant homologs. Tetra- consequence of their well-defined three-dimensional confor- loops conforming to this sequence or to the consensus UNCG mations (10). Recently, solution structures of small synthetic enhance the stability of synthetic RNA hairpins and have RNAs containing each of these tetraloops have been solved strikingly similar three-dimensional structures. To determine by two-dimensional NMR spectroscopy (11, 12). Despite the biological relevance ofthis similarity, as well as to assess the their different sequences, they adopt quite similar structures relative contributions of sequence and structure to the function in which the first and fourth bases are hydrogen bonded, the of the domain IV tetraloop, we replaced the GAAA sequence in second base has little interaction with the remainder of the fission yeast SRP RNA with UUCG. Haploid strains harboring loop, and the phosphate backbone between the second and this substitution are viable, providing experimental evidence third nucleotides is extended as a consequence of S-type for the functional equivalence of the two tetraloops. We next sugar puckering. tested the two sequences found in plant SRP RNAs at this We previously reported the in vivo effects of point muta- location for function in the context of the Schizosaccharomyces tions in the domain IV tetraloop of fission yeast SRP RNA pombe RNA. While substitution of CUUC does not allow (nucleotides 160-163; wild-type sequence GAAA) (13). Both growth, a viable strain results from replacing GAAA with lethal alleles identified were transversions at G-160, which UUUC. Although the viable tetraloop substitution mutants had been implicated in SRP19 protein binding by RNase exhibit wild-type growth under normal conditions, all three protection studies (14, 15). However, a G at this position is express conditional defects. To determine whether this might also critical to the integrity of the tetraloop, and there is a be a consequence of structural perturbations, we performed strong correlation between the phenotypes of the remaining enzymatic probing. The results indicate that RNAs containing point mutants we examined and predicted perturbations of tetraloop substitutions exhibit subtle differences from the wild the structure. To gain further insight into the role of domain type not only in the tetraloop itself, but also in the 3-base pair IV, as well as to assess the relevance of recent in vitro adjoining stem. To directly assess the importance of the latter structural data to the situation in vivo, we analyzed both the structure, we disrupted it partially or completely and made the effects of en bloc tetraloop substitutions and the conse- compensatory mutations to restore the helix. Surprisingly, quences ofdisrupting and restoring the adjoining stem. Taken mutant RNAs with as little as one Watson-Crick base pair can together, the results of our studies imply that the in vivo support growth. function of this region is determined by its structure. Signal recognition particle (SRP) is an RNA-protein complex EXPERIMENTAL PROCEDURES that targets ribosomes translating presecretory to Materials. Enzymes were purchased from BRL and New the endoplasmic reticulum membrane (reviewed in ref. 1). England Biolabs; mutagenesis reagents were from Amer- The extensively studied canine SRP is composed of six sham; DNA sequencing reagents were from United States polypeptides and one 300-nucleotide RNA (2, 3). SRP RNA Biochemical; RNases were from Pharmacia; and calf intes- (also referred to as 7SL) has been identified in a variety of tinal alkaline phosphatase was from Boehringer Mannheim. organisms (reviewed in ref. 4) and can be folded into a Sequencing primers and mutagenic oligonucleotides were phylogenetically conserved secondary structure consisting of synthesized at the Biotechnology Center at the University of four domains: a short base-paired region at the 5' end (domain Illinois. Radiolabeled [y_32P]ATP was from ICN. I); a long central helix that includes the 3' end (domain II); Site-Directed Mutagenesis. Targeted mutations were intro- and two internal stem-loop structures, one extensively base duced into the cloned SRP7 gene carried on the phagemid paired (domain III) and one containing several internal loops pWEC4.2 (16) by standard methods (17, 18) with the follow- (domain IV) (nomenclature according to ref. 5). The se- ing oligonucleotides: STL1 (5'-ATGTGCATTSC- quence, as well as the secondary structure, of domain IV is GAASAACCTCCATC-3'), replaces nucleotides 159-164 conserved in SRP RNA homologs from bacteria to humans with SUUCGS sequences where S is G or C; STL2 (5'- (4). This helix terminates in a tetranucleotide loop that ATTCCGAAGAACCTCCATC-3'), generates a variant not conforms to the consensus GNRA (N, any nucleotide; R, obtained with STL1; PTL1 (5'-TGTGCATTGGAAR- purine) except in plant SRP RNAs, which have four pyrim- CAACCTCCA-3'), replaces nucleotides 160-163 with the idines at this location. GNRA and UNCG tetraloops are corresponding segment of plant SRP RNA; PTL2 (5'- highly overrepresented in RNAs (6) and are found frequently Abbreviation: SRP, signal recognition particle. The publication costs of this article were defrayed in part by page charge tPresent address: Department of , Research In- payment. This article must therefore be hereby marked "advertisement" stitute of Scripps Clinic, La Jolla, CA 92037. in accordance with 18 U.S.C. §1734 solely to indicate this fact. 1To whom reprint requests should be addressed. 5409 Downloaded by guest on September 26, 2021 5410 Biochemistry: Selinger et al. Proc. Natl. Acad. Sci. USA 90 (1993) TGTGCATTTGAARCAACCTCCA-3'), places a G-A pair Table 1. Phenotypes conferred by mutations targeted to the adjacent to the plant tetraloops; M3a (5'-ATGTGCAT- domain IV terminal region TG*TT*TCC*AACCTCCAT-3'), creates mutations at posi- Line Allele Sequence Growth OTS tions 159, 162, and 164 (45% degeneracy was allowed at the positions marked with an asterisk); BP2 and -3 (5'- 1 Wild type GGAAAC GATGTGCAT1T2GTTTCCA3A4CCTCCATCG-3'), creates 2 A162C GGACAC Viable + mismatches in the second and third base pairs flanking the 3 G159U/A162C UGACAC Dead tetraloop and changes the A-U pairs to C-G pairs (T1 = 50% 4 G159U/A162U UGAUAC Dead T/50% G, T2 = 50% T/50% C, A3 = 17% A/83% G, and A4 5 A162C/C164G GGACAG Viable +++ = 17% A/83% C); BP-C (5'-TGATGTGCAGCGTTTC- 6 STL1-3 GUUCGC Wild type +++ CGCCC-3'), replaces both A-U pairs with G-C pairs; D4-E 7 STL2-1 CUUCGG Viable ++ (5'-GATGTGCATTGCGTTTC*CGCAACCTCCATC-3'), 8 STL1-1 GUUCGG Dead inserts two G-C base pairs into the stem adjacent to the 9 STL1-2 CUUCGC Dead tetraloop, in the context of either the wild-type tetraloop or 10 PTL1-1 GUUUCC Viable ++ a lethal point mutant (G16OC) (50% C/50% G at the position 11 PTL1-2 GCUUCC Dead marked with an asterisk). 12 PTL 2C GCUUCA Dead Yeast Methods. Mutant alleles were confirmed by DNA 13 PTL 2U GUUUCA Dead sequencing and introduced into Schizosaccharomyces 14 D4E-1 UUGCGG ... CGCAA Dead pombe strain RM2a, heterozygous for disruption ofthe SRP7 15 D4E-2 UUGCCC ... CGCAA Dead gene (19). Transformation, random spore analysis, and test- 16 D4E-3 UUGG ... CGCAA Viable ing for sensitivity to high or low temperature and/or in- 17 G159C/C164G UUC ... GAA Cold creased osmotic strength were performed as described (13). sensitive + + + To determine generation times, cells were grown in rich 18 G159C/C164A UUC ... AAA Dead medium at 30°C and their density was monitored by counting 19 G159A UUA... CAA Viable +++ in a hemacytometer. 20 G159U UUU ... CAA Viable +++ Construction of the p77 Plasmid Series. Domain IV was 21 C164A UUG... AAA Viable ++ amplified by 25 cycles ofPCR under standard conditions with 22 G159A/C164U UUA ... UAA Viable the two primers D4-PCRI (5'-CTGGCAGTTAGGCCTTG- 23 G159U/C164A UUU ... AAA Viable TAGTACCGA-3'; identical to nucleotides 125-150, except 24 U157G GUG... CAA Viable for the underlined changes to create a Stu I site) and D4- 25 U158C UCG ... CAA Viable PCRII (5'-GCACTGCCCAGGATC-CT*GTAGTGATG-3'; 26 A165G UUG... CGA Viable complementary to positions 172-196 except at the underlined 27 A166C UUG ... CAC Viable ++ nucleotides, which create a BamHI site, and at the position 28 U157G/U158C GCG. .. CAA Viable marked with an asterisk, which restores pairing to the Stu I 29 U157G/A165G GUG ... CGA Viable + site) on pWEC4.2 DNA containing wild-type or mutant 7SL 30 U158C/A166C UCG ... CAC Dead sequences. The products were phenol extracted, digested 31 BP-Comp GCG . CGC Viable with Stu I and BamHI, and ligated into the same sites in pDW19 (a gift of Norman Pace, Indiana University), which critical residues. The striking similarities between the re- has a T7 promoter positioned such that transcription starts at cently determined solution structures of UUCG (11) and the first G of the Stu I site. GAAA (12) tetranucleotide loops prompted us to ask whether In Vitro Transcription. BamHI-digested p77 DNA was they are interchangeable in vivo. Both UUCG substitution transcribed with T7 RNA polymerase according to published mutants in which the flanking residues do not form Watson- procedures (20), and a 10-pmol sample was dephosphorylated Crick base pairs are inviable (Table 1, lines 8 and 9). The and 5'-end-labeled with polynucleotide kinase in the presence lethality of unpaired G residues at positions 159 and 164 in of [y-32P]ATP. conjunction with the UUCG tetranucleotide loop is interest- Enzymatic Structure Probing. Labeled transcripts ing, since C164G produced only a mild conditional growth (.200,000 cpm) were subjected to partial digestion with defect in an otherwise wild-type RNA; the C-C apposition RNases Ti and Vi, nuclease S1, and alkali by published was lethal in combination with the GAAA tetraloop (13), as procedures (21) with minor modifications. with UUCG. Most importantly, our data demonstrate that replacing the wild-type GAAA loop at positions 160-163 with RESULTS UUCG produces an RNA that supports growth as the only Phenotypes of Additional Point Mutants Are Consistent with form of 7SL in the cell (Table 1, lines 6 and 7). This the NMR Structure. In our earlier point mutagenesis of the observation has two important implications. First, the two domain IV tetraloop, we did not obtain substitutions at the tetraloops that adopt similar structures in vitro are function- third nucleotide. Using a more focused strategy, we isolated ally interchangeable in vivo, at least in this context. Second, A162C, which is viable but has a mild conditional growth the role of the domain IV tetraloop in SRP RNA is not base defect (Table 1, line 2). This phenotype presumably results specific, since these sequences have no nucleotides in com- from disrupting an interaction with the ribose at position 160; mon. Although the UUCG tetraloop mutant has a generation a purine is required at the third position of the tetraloop time indistinguishable from that of a wild-type strain under because its N-7 serves as a acceptor (12). The normal conditions (data not shown), its growth is impaired three double mutants involving A-162 have more severe under restrictive conditions (Table 1 legend). Interestingly, phenotypes than any ofthe component single mutants (Table the substitution allele carrying the wild-type G-C base pair 1, lines 3-5; ref. 13), suggesting that these residues function has a more severe conditional growth defect than the mutant cooperatively. with the reversed C-G base pair. A Different Stabilizing Tetraloop Can Functionally Substi- Only One of the Two Plant Tetranucleotide Loop Sequences tute for the GAAA Sequence in Domain IV of SRP RNA. Can Support Growth in S. pombe. In light of the conservation Although the severity of the growth defects resulting from of the domain IV GAAA tetraloop from Escherichia coli to point mutations in the domain IV GAAA tetranucleotide loop humans, it was surprising that the corresponding sequences parallel predicted perturbations of the structure, these data in several recently characterized plant SRP RNAs are ex- do not definitively rule out a sequence-specific role for the clusively pyrimidine (reviewed in ref. 4). We therefore Downloaded by guest on September 26, 2021 Biochemistry: Selinger et al. Proc. Natl. Acad. Sci. USA 90 (1993) 5411 wanted to determine whether these tetraloops could, like pair, only nucleotides 157 and 158 show significant cleavage. UUCG, function in the context of S. pombe SRP RNA. The simplest interpretation of these observations is that the Remarkably, the UUUC replacement mutant is viable, al- C(UUCG)G sequence promotes stronger base pairing in the though conditionally growth defective (Table 1, line 10). In short stem. Within the tetraloops themselves, nuclease Si contrast, the CUUC tetraloop substitution confers a reces- cleaves all nucleotides in GAAA but not in UUCG; the latter sive lethal phenotype (Table 1, line 11). Since plant SRP RNA sequence is, in contrast, cleaved by RNase Vi at the second tetraloops are flanked by a noncanonical GA base pair, we and third positions. Thus, it appears that the UUCG tetraloop attempted to rescue this mutant by replacing the closing G-C adopts a more helix-like conformation than the wild-type pair with this combination. This substitution not only failed tetraloop. to support viability with the CUUC tetraloop but was also We carried out similar nuclease probing experiments with lethal in combination with the otherwise viable UUUC mu- domain IV transcripts carrying plant tetraloop substitutions tant (Table 1, lines 12 and 13). to determine whether structural differences might account for In Vitro Enzymatic Probing Data Indicate That Tetraloop the ability of UUUC, but not CUUC, to functionally replace Substitution Mutants with Conditional Growth Defects Exhibit GAAA in fission yeast SRP RNA; some representative Subtle Structural Differences from the Wild-Type RNA. Al- results are shown in Fig. 2 and data from several experiments though a UUCG tetraloop can functionally replace the wild- are summarized in Fig. 3. As expected, the upper part of the type GAAA sequence under normal conditions, the growth domain IV structure showed only minor variations in the sites defect of the mutant under extreme conditions suggests that of cleavage between the CUUC substitution mutant and the the substitution may subtly perturb the structure. To confirm wild-type RNA, while differences in the tetraloop region this hypothesis, we performed in vitro enzymatic probing on were more dramatic. In both RNAs carrying plant tetraloops, T7 transcripts corresponding to domain IV with either a RNase Vi cleaves residues within the loop, in common with GAAA or a UUCG tetraloop. In the substitution mutant, the the C(UUCG)G RNA and in contrast to the pattern observed tetraloop was flanked by a C-G base pair, which has a less with the wild-type transcript. The terminal domain IV stem severe conditional growth defect in vivo. Each transcript was is cleaved by both nuclease Si and RNase Vi at the U-157 A- digested with RNase Vi, which cleaves double-stranded 166 base pair in the CUUC RNA, while only Vi cleaves at RNA and nucleotides involved in tertiary structure (see ref. these positions in the UUUC mutant. Notably, the latter 22 for a discussion of Vi specificity), and with nuclease Si, pattern is the same as in the GAAA and UUCG transcripts, which is specific for single-stranded nucleic acids; some consistent with this mutant's ability to support growth. In representative results are shown in Fig. 1 and data from addition, the CUUC substitution increases the susceptibility several experiments are summarized in Fig. 3. The digestion ofpositions 154-156 to cleavage by nuclease Si relative to the patterns for the wild-type transcript are generally consistent viable mutants. Although these bases, which are critical for with the structure deduced from phylogenetic analysis of the SRP54 protein binding (23, 24), are predicted to be single intact RNA, except for a few anomalous nuclease Si cuts stranded, A-154 and G-155 are not accessible to nuclease Si produced only at the higher enzyme concentration (Fig. 1A). in the other three transcripts. Thus, the inability ofthe CUUC Interestingly, not only does the structure of the tetraloop tetraloop to function in the context offission yeast SRP RNA itself change, but the surrounding region is also altered. In may be due either to destabilization of the adjoining helix or particular, in the C(UUCG)G transcript, RNase Vi cuts at to a conformational change in the 5' internal loop, since the positions corresponding to 156-159 in the full-length RNA, tetraloop structure itself is similar to that of the viable while for the wild-type G(GAAA)C tetraloop and closing base substitution mutants. The Tetraloop Does Not Function Solely to Stabilize the Adjoining Helix. Since our in vivo phenotypic analysis and in A B_ vitro structure probing data both suggest that a primary role

A AlI-L___ B L-L L i

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

G156 G168 _ do Gt68 *=3 0168 -C41m.t_ o C = _ . 0153 0*_*El 153 * , 0152 01G52 * G15-_it U. 00_ iA _ * G156_ 1 G156 - l - : G149 G14 9 G155 A G155 _ _l G153 _ w ow * G153 .* 'm FIG. 1. Products of enzymatic structure probing for wild-type G152 * op,*t and C(UUCG)G substitution mutant RNAs. 5'-End-labeled domain G152 _* *El IV transcripts were digested with the enzyme indicated and resolved on a 10% polyacrylamide/8 M urea sequencing gel. Alk, partial alkaline hydrolysis products. In the nuclease Si and RNase Vi lanes, G149 _ El G149 44j_ b*, L and H are low and high concentrations ofenzyme. NE, no enzyme control. Products of partial digestion with RNase Ti were also run FIG. 2. Products of enzymatic structure probing for CUUC and as a control to allow location of cleavage products in the RNA UUUC substitution mutant RNAs. Lanes are labeled as described in sequence; G residues are numbered according to ref. 19. (A) Results Fig. 1. (A) Products of partially digesting CUUC mutant RNA. (B) for wild-type RNA. (B) Products from mutant transcript. UUUC transcript. Downloaded by guest on September 26, 2021 5412 Biochemistry: Selinger et al. Proc. Natl. Acad. Sci. USA 90 (1993)

3' 3' 3' 3' combining it with C164G but not with C164A (Table 1, lines c 5 c 5' c 5' C 5' 17 and 18). However, although the G159C/C164G double u g u g u g u g mutant is viable, it exhibits a severe growth defect at elevated g a g AA g a a g temperature and osmotic strength and is also unable to grow G-C-A *G-c AG-CA AG-c AG - c at low temperature. Mutating G-159 to A and U or C-164 to 1810G - C *G-c * G-CC HG-c - A produced conditional growth defects, which were amelio- *A-U A-u AA-U A - u rated by restoration ofbase pairing (Table 1, lines 19-23). The U- u u A-U combination exhibited wild-type growth, while U'A C-G*141 C -g * C-G- C - g retained a mild conditional phenotype. Taken together with A-U A - u & A-U - A - u our earlier data (13), it appears that the identity of the residue U-Am U-a U - A v m at position 164 is less critical than that at position 159 and, in C-G-144 C-g * C - G . C-g particular, growth defects arising from a pyrimidine at posi- U- u U-n u A- a tion 159 partially persist even upon restoration of base A A A AC A U 0 A-A C AC pairing. Because the length of the helix adjoining the is Co c C- C tetraloop phylogenetically conserved in we also tested C-G-149 C-G C-G- * C -G (except plants), - two U - A U - A AU -AA U -AA the effects ofmutating the other base pairs (positions 157, * C - G v A-UU A-U A-UA 158, 165, and 166). Of the four single mutants examined, only *C-G-152 OC-Go OC-GE A166C exhibits a conditional phenotype; the others grow A GA153 A GA A GA A GA under all conditions tested despite the loss of a Watson-Crick

C A AC A AC AA AC A base pair (Table 1, lines 24-27). Even more remarkable, one 168 G G 155 *G G AG G - AG G of the three double mutants shows no detectable growth UIG' 156 UIG. AUIGA AUIG defect even though both base pairs are disrupted (Table 1, *OA- U * A - U- *A-U - line 28). A second double mutant in this series, U157G/ AA-U- A-U- *A-UO* A165G, exhibits a mild conditional phenotype, while the C - C - G * 159 G-CO * C-GEGU- *C-Gm third, U158C/A166C, is lethal (Table 1, lines 29 and 30). The AA GA16o G U *C Cm *C UA inviability of the latter mutant can be rescued by changing AA C U UU. U U both U-157 and A-165 to (G Table 1, line 31). This mutant, in A A * CUU 0 * which both original A*U pairs are replaced with G-C pairs, Wild-Type CUUCGG CUUJC UUUC exhibits fully wild-type growth. Thus, the identity of the in FIG. 3. Summary of enzymatic structure probing data. Nucleo- bases this helix is unimportant for SRP RNA function. tides cleaved by nuclease Si are denoted by triangles and sites of DISCUSSION RNase Vl cuts are denoted by circles, with the extent of cleavage indicated by the size of the symbol. Positions that were cleaved by Our finding that a UUCG tetraloop can functionally replace both enzymes are denoted by squares. Data for the 5' end of the GAAA in SRP RNA is consistent with extensive analysis of wild-type and CUUC mutant RNAs were obtained by resolving the phylogenetically diverse 16S rRNA sequences, which re- cleavage products on 15% sequencing gels (data not shown). Posi- vealed that GNRA and UNCG tetraloops are sometimes tions that could not be read from our gels are indicated by lowercase found substituted en bloc even between closely related letters. Numbering of G residues is as described in Fig. 1. organisms (6). However, the viability of our UUCG tetraloop mutants, particularly the allele with a C-G flanking base pair, of the tetraloop is to stabilize the short adjoining helix, we conflicts with our earlier conclusion that the 5' nucleotide of decided to test whether stabilizing the structure by another the closing pair and the G residue of the tetraloop were likely means would serve the same purpose. To this end, we to be sequence-specific components of the SRP19p binding inserted two additional G-C base pairs into the helix, either in site (13). This inference was based on the inviability of S. combination with the wild-type tetraloop or with a lethal pombe mutants harboring transversions at positions 159 and point mutation, G160C (13). If destabilization of the helix is 160, together with the results of in vitro RNase protection the sole cause of the functional defect in this mutant, exten- experiments on mammalian components (14, 15). The close sion of the helix is predicted to rescue it. This prediction is correlation between our earlier and present phenotypic data not borne out; moreover, extending the helix results in a for point mutations in the SRP RNA tetraloop and the lethal phenotype even in combination with the wild-type recently determined NMR structure (12) suggest instead that tetraloop (Table 1, lines 14 and 15). During the mutagenesis the tetraloop is a structural entity and that the lethality ofthe procedure, an unexpected mutant was isolated, D4E-3, in point mutants G159C, G160C, and G16OU arises from con- which two bases were inserted 3' to the tetraloop. Surpris- formational alterations. This conclusion is reinforced by our ingly, this mutant exhibits no growth defect under any finding that the UUCG substitution mutants, in which the condition tested (Table 1, line 16). Taken together, these data tetraloop is predicted to have a structure similar to that ofthe indicate that the length of the helix adjoining the tetraloop is wild-type GAAA despite its completely different sequence, important, although some modifications ofthe terminal struc- support growth. The ability of a UUCG tetraloop to func- ture can apparently be accommodated. tionally replace the wild-type sequence indicates that, if a Mutants with Only a Single Watson-Crick Base Pair in the fission yeast homolog of the SRP19 protein does in fact Helix Adjoining the Tetraloop Are Viable. To complete our contact this region, it must recognize the ribose-phosphate analysis of the relative importance of sequence vs. structure backbone and not the bases. Consistent with our observa- in this region of fission yeast SRP RNA, we tested the effects tions, Zwieb (25) has recently shown by an in vitro assay that of disrupting the terminal domain IV helix and restoring it replacement of the domain IV GAAA tetraloop in human with a different sequence. First, since the identity ofthe base SRP RNA with UUCG is compatible with SRP19p binding. pair adjacent to the tetraloop had a marked effect on the In contrast to the situation in SRP RNA, not a single residue phenotype of the UUCG substitution mutants, we examined within the GNRA tetranucleotide in the large rRNA that the consequences ofjuxtaposing various nucleotides flanking serves as a critical recognition element for the cytotoxin ricin the wild-type tetraloop. Our results demonstrate that base can be altered without loss ofrecognition by the protein (26). pairing at these positions is critical for SRP RNA function, The data presented here are incompatible with an earlier since the lethal phenotype of G159C (13) can be rescued by report that positions 157-160 are part ofa tertiary interaction Downloaded by guest on September 26, 2021 Biochemistry: Selinger et al. Proc. Natl. Acad. Sci. USA 90 (1993) 5413 with nucleotides 63-66 in the S. pombe RNA (27). First, in pair or by point mutations within the tetraloop (23, 24) but is the viable C(UUCG)G substitution mutant, two of the three reduced by mutations that disrupt the stem (23). The stability Watson-Crick pairs in the proposed pseudoknot cannot of this helix may be critical for maintaining the 5' internal form. In addition, in the U157G/U158C mutant, which ex- loop, which is recognized in a sequence-specific manner by hibits no growth defect under any condition tested, the other SRP54p (23, 24), in a productive conformation. two proposed pairs are disrupted. Although the potential for In summary, the data presented here, together with our base pairing between these regions appears to be phyloge- earlier mutagenesis results (13), indicate that the function of netically conserved (27), we note that, in the fission yeast the domain IV GAAA tetranucleotide loop and adjoining RNA, the two sequences involved are constrained by other stem in SRP RNA is to promote formation of a particular interactions: nucleotides 63-66 are part of the B box required structure, which can be adopted by several dramatically for RNA polymerase III transcription (15), while nucleotides different primary sequences. In addition to defining the 157-160 are critical to the structure whose importance we structural features required for a functional RNA, these data have demonstrated here. impose constraints on the properties of factors that interact The viability ofthe UUUC substitution mutant was initially with this region. The imperfect correlation between stability somewhat surprising, since this sequence, unlike UUCG and of the region, which appears to be the major determinant of GAAA, is not overrepresented in rRNA. However, interac- SRP54p binding (23), and the phenotypes of mutants, sug- tion of the U and C at positions 1 and 4, as has been observed gests that the tetraloop region does indeed interact with other in intermolecular duplexes (28), could result in a conforma- components, among which may be the SRP19 pro- tion similar to that of the two tetraloops whose structures cellular have been solved. Consistent with the viability ofthis mutant, tein, in a functionally important manner. our structure probing data show that the adjoining helix We are grateful to Olke Uhlenbeck and Art Pardi (University of exhibits a pattern of cleavage similar to that of the wild-type Colorado) for helpful discussions and for reading an earlier version and the UUCG substituted RNA and distinct from that of the of this manuscript. We thank Claudia Reich and Steve Althoff for inviable CUUC mutant. In the CUUC transcript, the adjoin- critical reading of both the original and revised versions. We ac- ing stem appears to be destabilized relative to the wild-type knowledge the efforts ofMin Ma in creating the UUCG mutant series and viable mutant RNAs. While the sites of enzymatic and Anne Chiang in assaying the plant tetraloop mutants. This cleavage in both plant tetraloop transcripts overlap more with research was supported by National Science Foundation Grant DCB those in the UUCG substitution mutant than in the wild-type 88-16325 awarded to J.A.W. RNA, the resemblance is greater between the two viable 1. Walter, P. & Lingappa, V. (1986) Annu. Rev. Cell Biol. 2, 499-516. mutants. Although replacement of the S. pombe SRP RNA 2. Walter, P. & Blobel, G. (1980) Proc. Natl. Acad. Sci. USA 77, this 7112-7116. domain IV GAAA with a CUUC tetraloop is lethal, 3. Walter, P. & Blobel, G. (1982) Nature (London) 299, 691-698. sequence is presumably functional in the plant RNAs, since 4. Larsen, N. & Zwieb, C. (1991) Nucleic Acids Res. 19, 209-215. all 11 of the corn (Zea mays) cDNAs sequenced have a 5. Poritz, M. A., Strub, K. & Walter, P. (1988) Cell 55, 4-6. CUUC tetraloop, as do two of three wheat (Triticum aesti- 6. Woese, C. R., Winker, S. & Guttell, R. R. (1990) Proc. Natl. Acad. vum) cDNAs and the tomato (Lycopersicon esculentum) Sci. USA 87, 8467-8471. RNA; only the SRP RNA from cineraria hybrids (Senecio 7. Jacquier, A. & Michel, F. (1987) Cell 50, 17-29. cruentus) has exclusively UUUC at this location (4). The 8. Tuerk, C., Gauss, P., Thermes, C., Groebe, D. R., Gayle, M., Guild, N., Stormo, G., d'Aubenton-Carafa, Y., Uhlenbeck, 0. C., ability of a CUUC tetraloop to function in the context of Tinoco, I., Jr., Brody, E. & Gold, L. (1988) Proc. Natl. Acad. Sci. plant, but not fission yeast, SRP RNA may be related to the USA 85, 1364-1368. extra noncanonical GA base pair in the plant domain IV 9. Uhlenbeck, 0. C. (1990) Nature (London) 346, 613-614. terminal helix. 10. SantaLucia, J., Jr., Kierzek, R. & Turner, D. H. (1992) Science 256, The effects of mutations in the terminal domain IV helix 217-219. on to the tetraloop and ability to form 11. Cheong, C., Varani, G. & Tinoco, I., Jr. (1990) Nature (London) depend their proximity 346, 680-682. a noncanonical base pair. Disrupting the pair immediately 12. Heus, H. A. & Pardi, A. (1991) Science 253, 191-194. adjacent to the tetraloop has the most severe effects, while 13. Liao, X., Brennwald, P. & Wise, J. A. (1989) Proc. Natl. Acad. Sci. eliminating either or both ofthe other 2 base pairs is tolerated USA 86, 4137-4141. under normal growth conditions with one exception, which 14. Siegel, V. & Walter, P. (1988) Proc. Natl. Acad. Sci. USA 85, produces a noncanonical AC pair and a C U juxtaposition. 1801-1805. 15. Poritz, M. A., Siegel, V., Hansen, W. B. & Walter, P. (1988) Proc. Although disrupting the closing pair is always deleterious, Natl. Acad. Sci. USA 85, 4315-4319. only those mutants with a C in place of the wild-type G at 16. Liao, X., Selinger, D. A., Althoff, S., Chiang, A., Hamilton, D. & position 159 are inviable. Even when the G159C mutation is Wise, J. A. (1992) Nucleic Acids Res. 20, 1607-1615. compensated by C164G, the cells display a severe growth 17. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. defect under restrictive conditions. The significant deleteri- 18. Taylor, J. W., Ott, J. & Eckstein, F. (1985) Nucleic Acids Res. 13, ous effect of a C-G pair flanking the GAAA tetraloop con- 8765-8785. 19. Brennwald, P., Liao, X., Holm, K., Porter, G. & Wise, J. A. (1988) trasts with the less severe phenotype of a C-G relative to a Mol. Cell. Biol. 8, 1580-1590. G-C base pair in combination with the UUCG tetraloop. The 20. Yisraeli, J. K. & Melton, D. A. (1989) Methods Enzymol. 180, preference in the latter case may be related to the fact that 42-51. RNA hairpins capped by C(UUCG)G are both more common 21. Knapp, G. (1989) Methods Enzymol. 180, 192-212. (6) and more stable (29) than those containing G(UUCG)C. 22. Lowman, H. B. & Draper, D. E. (1986) J. Biol. Chem. 261, 53%- Although the population of GAAA tetraloops in 16S rRNA 5403. 23. Selinger, D. A., Brennwald, P. J., Liao, X. & Wise, J. A. (1993) taken as a whole shows little selectivity regarding the closing Mol. Cell. Biol. 13, 1353-1362. base pair (6), we note that at any given location, there is 24. Wood, H., Lurink, J. & Tollervey, D. (1992) Nucleic Acids Res. 20, generally a strong preference for a particular sequence. The 5919-5925. relatively severe phenotype ofthe CG closing pair mutant in 25. Zwieb, C. (1992) J. Biol. Chem. 267, 15650-15656. combination with the wild-type tetraloop in SRP RNA pre- 26. Gluck, A., Endo, Y. & Wool, I. G. (1992)J. Mol. Biol. 226,411-424. 27. Andreazzoli, M. & Gerbi, S. (1991) EMBO J. 10, 767-777. sumably reflects a structural or sequence requirement that we 28. Holbrook, S. R., Cheong, C., Tinoco, I., Jr., & Kim, S.-H. (1991) do not yet fully understand, perhaps related to recognition by Nature (London) 353, 579-581. SRP19p. Binding of the SRP54 protein, which also interacts 29. Antao, V. P., Lai, S. Y. & Tinoco, I., Jr. (1991) Nucleic Acids Res. with domain IV, is unaffected by reversal of the closing base 19, 5901-5905. Downloaded by guest on September 26, 2021