Conserved terminal hairpin sequences of histone mRNA precursors ...

Proc. Nati. Acad. Sci. USA Vol. 86, pp. 4345-4349, June 1989 Biochemistry

Conserved terminal hairpin sequences of histone mRNA precursors are not involved in duplex formation with the U7 RNA but act as a target site for a distinct processing factor (histone gene expression/RNA processing/U7 small nuclear ribonucleoprotein particle/anti-Sm antibodies/sea urchin embryos)

ALAIN P. VASSEROT*, FREDERICK J. SCHAUFELEt*, AND MAX L. BIRNSTIEL* *Research Institute of Molecular Pathology, Dr. Bohr-Gasse 7, A-1030 Vienna, Austria; and tInstitut fur Molekularbiologie II der Universitat Zurich,

Hcnggerberg HPM-2, CH-8093 Zurich, Switzerland

Contributed by Max L. Birnstiel, February 16, 1989

ABSTRACT The hairpin loop structure and the downstream spacer element of histone mRNA precursors are both needed for efficient 3' end formation in vivo and in vitro. Though generally considered as a single processing signal, these two motifs are involved in different types of interaction with the processing machinery. Whereas RNA duplex formation between the downstream spacer element and the U7 small nuclear RNA is essential for processing, we show here that base pairing between the histone stem-oop structure and the U7 RNA is not relevant. Our experiments demonstrate that a processing factor other than the U7 RNA makes contact with the highly conserved hairpin structure of the histone precursor. The recognition of the target site by the processing factor is structure and sequence specific. Prevention of this interaction results in an 80% decrease of 3' cleavage efficiency in vitro. The hairpin binding factor is Sm-precipitable and can be partially separated from the U7 small nuclear ribonucleoprotein particle on a Mono Q column.

There are many cases where RNA hairpins mediate control of gene expression. In prokaryotes, stem-loop structures are involved in attenuation (1), transcription termination (1), rRNA processing (2), mRNA turnover (3), retroregulation (3), and translational regulation (4). Their role in ribosome assembly has also been demonstrated (5). In eukaryotes, hairpin structures have been shown to mediate 3' end formation (6, 7), the regulation (8) or stability (9) of histone mRNAs, the iron response of ferritin (10) and transferrin receptor (11) biosynthesis, as well as modulation of translation (12). A possible role of a folded structure at the trans-esterification active site in self-splicing (13), in recognition events (14), or in determining the amount ofalternative splicing (15) has been suggested. Finally, reports have emphasized the role of stem-loop structures in small nuclear ribonucleoprotein particle (snRNP) assembly (16, 17). In some cases, the hairpin structure, rather than the primary sequence perse, seems to be of major importance for function. In bacterial attenuation, for instance, synthetic terminator hairpins are able to promote transcription termination irrespective of their primary sequence (18). In general increasing stability of the stem-loop structure correlates positively with the efficiency of transcription termination (1) although some mutations do not follow this rule. The primary sequence of the sea urchin U7 small nuclear RNA (snRNA) hairpin, but not its structure, can be extensively modified without affecting the assembly into an Sm-precipitable particle or the processing function of the reconstituted U7 snRNP (17). The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

4345

A second class of stem-loop structures comprises those hairpins that have in common a requirement for specific sequences. In this case, the loop sequences and, sometimes, protruding unpaired nucleotides in the stem are a feature essential for function. Here, the primary sequence ofthe stem does not seem to be required for recognition, as exemplified by the interaction between the R17 coat protein and its palindromic binding site (19). Although the protein-RNA interaction has not been directly investigated, a similar situation probably holds for the binding site of the Li ribosomal protein (20), for the iron-responsive element of ferritin and transferrin receptor mRNAs (11), and for the sequence within the human immunodeficiency virus tar site (21). We have undertaken a detailed analysis of the highly conserved terminal stem-loop structure of histone mRNA precursors (pre-mRNA). We show that in contrast to the hairpins above, the histone stem-loop must retain not only its structure but also its overall primary sequence for efficient 3' processing. We further suggest that the hairpin structure is not involved in Watson-Crick type base pairing with the U7 snRNA, as considered possible (22), but makes contact to another processing factor. This factor is Sm-precipitable and, upon fractionation of nuclear extract, is present in a fraction lacking the U7 snRNP and the heat-labile factor previously shown to promote 3' processing of histone pre-mRNAs (for review, see refs. 6 and 7).

MATERIALS AND METHODS Multistep Mutagenesis, Microi'Jection, and Mutant Analysis. H3 histone mutants were generated by BAL-31 deletion from a restriction site on the 3' side of the hairpin structure to the first dinucleotide of the stem (GG) resulting in the formation of a "left arm" of the H3 gene. A similar procedure resulted in the formation of a "right arm" starting with the last stem dinucleotide (CC). Ligating oligonucleotides in between the left and right arms of the H3 gene resulted in the mutants depicted in Fig. 2A. Sea urchin microinjections (23) and RNA analysis were carried out essentially as described

(24).

In Vitro Processing, Processing Substrates, and Competitor RNAs. Preparation of nuclear extracts (25), SP6 transcription (26), in vitro pre-mRNA processing (27), and depletion of Sm-type snRNPs in nuclear extract (28) were performed as described (27, 29). T7 transcription was performed under SP6 buffer conditions. To test for complementation, a 10-fold molar excess of a construct with a wild-type palindromic Abbreviations: snRNA, small nuclear RNA; snRNP, small nuclear ribonucleoprotein particle; pre-mRNA, mRNA precursor; HBF, hairpin binding factor. tPresent address: Metabolic Research Unit, 1141-HSW, University of California, San Francisco, CA 94143.

4346

Biochemistry: Vasserot et al.

Proc. Natl. Acad. Sci. USA 96

sequence (wtpal compRNA; see text) was added to the reaction. Wild-type and mutant pre-mRNA hairpin sequences are shown in Fig. 3A. Oligonucleotides were ligated into a HindIII/Pst I-digested pSP64 (processing substrates) or pSPT19 vector (competitor RNAs). The processing substrates have the following overall sequence: 5'-m7GpppGA-

A

\

H3Ba1

AUACAAGCUUUCCCUAAC-(hairpin)-AACCACAGUCUCUUCAGGAGAGCUGACACUGAC-3'. The deltapal and

Cpal constructs (see Fig. 3A) have been described elsewhere (30). The competitor RNAs have the following overall sequence: 5'-GGGAGACCCAAGCUUCCCUAAC-(hairpin)AACCACAGUCUCUUC-3'. The falsepal compRNA (see text) whose hairpin sequence is 5'-CUGAGAGAAAUCUCAG-3' is not shown in Fig. 3A.

RESULTS 3' Processing in Sea Urchin Embryos Injected with Cloned Histone Genes. In early experiments, it was shown that deletion of the terminal hairpin of sea urchin H2A precursors results in a processing deficiency in the frog oocyte (31). A possible interpretation was that hairpin deletion prevents essential RNARNA contacts in the hypothetical processing complex between the U7 snRNA and the histone pre-mRNA (ref. 22 and Fig. 1). We investigated this hypothesis by introducing base changes into the stem-loop structure of the sea urchin Psammechinus miliaris H3 early histone gene in such a way that the capacity of the mutant transcripts to base pair with the U7 snRNA was severely reduced (Fig. 2A, potential base pairings are marked with an open circle). The mutants were tested in vivo by microinjection into sea urchin eggs of a closely related species, Paracentrotus lividus (23). At early blastula, total RNA was extracted, the 3' end of the different mutants was analyzed by SP6 mapping, and the injection efficiency was standardized by comparison to the HISTONE PRE -mRNA

HISTONE PRE - mRNA

U

GA(U)A

5'

U7 RNA

CAAU

3N N

U7RNA

CC C A U G A

u UU

CU

G

CA

UC

U

I G *C /A k

U -A - GUGCCCMC(U)-3' C *G G-C

CGC

,C *G, GU AGU 'U A' C*G C*G G*C AUb 'A-G' SEA URCHIN

A A ,,/

G A A

A

AC&

,3 U C

MOUSE

FIG. 1. Base complementarity between the 3' region of the histone pre-mRNA and the U7 snRNA. The hypothetical RNA duplex between the histone hairpin structure and the U7 snRNA has an estimated free energy (32) of -11 kcal in sea urchin and -7 kcal in mouse (1 cal = 4.184 J). The two histone conserved elements are boxed and the hairpin loop is marked by inverted arrows. The mature 3' end of the histone mRNA is designated by an arrowhead.

D

B

fL =T H3Baf2 H3Ba10

(1989)

0

H3wt

H3 3

To~~ 0 +0-*read through

Duw~e pow

.4- H3 3'

14H2B 3'

FIG. 2. In vivo processing of Psammechinus miliaris H3 wild-

type and mutant pre-mRNAs. (A) Hairpin loop sequences. Mutations are shown white on black and potential base complementarities with the U7 snRNA are marked with an open circle. (B) SP6 mapping. H2B 3', 3' end mapping of H2B transcripts as a positive injection control; H3 3', properly processed H3 transcripts; H3 3' readthrough, unprocessed H3 transcripts; P, undigested H3 probe; M, Hpa II-digested pBR322 marker.

transcripts from a coinjected H2B gene. Fig. 2B shows that processed RNA was obtained from both wild-type and all mutant H3 genes, although the ratios between the mature H3 and H2B mRNA pools vary. Interestingly, the H3Ba2 mutant, in which 11 out of 13 possible complementary bases were exchanged (while maintaining a hairpin structure) still yielded processed RNA in appreciable amounts.

It can be concluded from these experiments that production of mature histone H3 mRNA is not critically dependent on the base-pairing capacity between the histone hairpin and the U7 snRNA, although it has been shown (24) to be dependent on duplex formation between the downstream histone spacer element and the 5' terminal sequences of the U7 snRNA. It should be noted that for all of the mutants, but not the wild-type gene, unprocessed RNA molecules appeared. This, combined with a somewhat reduced level of mature H3 mRNA by comparison to the H2B standard, suggests that processing is affected for the mutants but to an extent that is difficult to quantify precisely because the incubation time required to obtain mappable transcripts is relatively long and the injected developing embryos are a dynamic system exhibiting important RNA turnover (ref. 33 and references therein). We, therefore, decided to extend the above observations by studying 3' processing of mutant transcripts in a homologous in vitro processing system where RNA turnover and RNA degradation are not occurring (27), so that the amounts of processed RNA should be a bona fide measure of the actual processing rate, allowing us to comprehend the finer details of the reaction. 3' Processing in Vitro. SP6 polymerase-generated transcripts from H4 mouse histone constructs with various mutations or a deletion of the 3' region (Fig. 3A) were processed in vitro in a nuclear extract from mouse hybridoma cells as described (27). The reaction products were analyzed on polyacrylamide gels and the processing efficiency was quantified. Depending on the experiment, 50-60% ofthe wild-type input RNA was processed (Fig. 3B, lane 2). This amount was then taken as 100% for comparison with other processing reactions.

Biochemistry: Vasserot et al.

Proc. Natl. Acad. Sci. USA 86 (1989)

A 0

IC

..

G1

(Cc1011

0

'lGc*?JG.0I~

nc.oGC

~ ~~~a

.0 Ce

H4wt fstemlwtloop false pal open pal delta pal C pal

B

a

3

_so m

* INPUT

[

INPUT

] PROC.

-_

PROCI[

1 2 3 4

5

6 7 89

FIG. 3. In vitro processing of mouse H4 wild-type and mutant pre-mRNAs. (A) Hairpin loop sequences. Mutations are shown white on black, potential base complementarities with the U7 snRNA are marked with an open circle. (B) In vitro processing. Lanes: 1, wild-type pre-mRNA input; 2, 4-7, and 9, processing efficiency of the various mutants and their controls; 3 and 8, Hpa II-digested pBR322. Proc., processed RNA.

We first investigated the importance of the highly conserved hairpin nucleotides (34). Providing a template with a false palindrome (referred to as the falsepal mutant, see Fig. 3A) that maintains only two potential contact sites with the U7 snRNA allows processing of the precursor at 17-22% of the wild-type activity (Fig. 3B, compare lanes 2 and 5). Thus, as in the sea urchin injection experiments described above, extensive base complementarity between the U7 snRNA and the histone hairpin cannot be said to be an absolutely essential element for 3' processing. This is supported by the finding that some processing (5-10%6) is possible in vitro even when the histone hairpin structure has been deleted (compare lane 7 to the control lane 9; ref. 30). In the openpal construct (Fig. 3A), 10 out of 12 nucleotides capable of base pairing with the U7 snRNA have been

preserved, but two base alterations in the stem result in a sequence that cannot be folded into a stem-loop structure. Despite the presentation of most of the potential contact sites in a readily accessible linear form, processing of this precursor RNA is reduced by a factor of 5 (Fig. 3B, lane 6). Hence it appears that both in vivo (35) and in vitro (this paper) presentation of the wild-type sequences in a stem-loop configuration is required for maximal efficiency of 3' processing, whereas deletion, mutation, or linearization of the hairpin reduces processing to 5-20% of the control value. We designed the fstemlwtloop mutant (Fig. 3A) to investigate the relative contribution of the stem sequences. While maintaining the wild-type loop nucleotides, this construct contains a false stem sequence with 12 nucleotides altered. This mutant supports 3' processing at a level of about 40%o (Fig. 3B, lane 4) of the control value suggesting that the wild-type primary sequence of the stem is important for maximal rate of 3' processing (see below). However, it should be noted that the requirement for wild-type stem

4347

sequences is not absolute because hairpin revertants (see figure 1 of ref. 35) with four stem nucleotides altered yield mature RNA in the frog oocyte system (35). All the processed mutant RNAs migrate differently relative to the processed wild-type RNA. However, it was shown by the pCp-tagging approach (27) that all mutants produce mature RNA species with genuine 3' ends (A.P.V. and M.L.B., unpublished results). Similar aberrant migration of RNAs has also been reported and results probably from differential destabilization of RNA secondary structure during electrophoresis (36). Competition Experiments. Since all alterations of the histone hairpin lead to some reduction of processing, it is not possible to rule out from the above experiments that base complementarity may play a minor role in 3' processing. However, the hairpin nucleotides could interact with a processing factor by mechanisms that do not involve RNA-RNA duplex formation. The strict conservation of the stem-loop nucleotides (34) suggests that such an interaction would be extremely specific. Thus, it should be possible to prevent this interaction by adding a suitable unlabeled competitor RNA and, hence, to inhibit processing of the labeled reporter substrate. To trap a putative hairpin binding factor(s) [HBF(s)], competitor RNAs with various hairpin sequences were used that are themselves not capable of being processed since they lack the downstream spacer element (30). When the competition is carried out with an RNA containing the wild-type palindrome sequence (referred to as wtpal compRNA), processing of the labeled wild-type substrate is reduced, with competition leveling off at about 20% of the value obtained in the absence of competitor (Fig. 4 B and C). Some processing component accounting for 80%o of the processing rate is apparently being preempted by the addition of the wtpal compRNA. That this competition is sequence specific and pertains to the hairpin structure can be concluded from the inability of a competitor RNA with a different palindrome sequence (i.e., falsepal compRNA; see Fig. 4C) to reduce processing. The downstream spacer element alone is a competitor at high input (Fig. 4C) whereas the complete wild-type RNA (having both the conserved stem-oop and the downstream spacer motif) reduces processing to levels close to zero, even at relatively low molar excess (Fig. 4 A and C). This more effective competition of the complete wild-type competitor RNA as compared to that of either conserved motif alone is consistent with the concept that both elements cooperate in directing histone 3' end formation (6, 7). As seen in Fig. 3B, the mutant falsepal pre-mRNA is a poor processing substrate (17-22% of wild type). From the above results it appears possible that the processing rate for this mutant is reduced to -20%6 because this pre-mRNA is not capable of binding the HBF due to the mutation of its target sequence. If this is the case, processing ofthe labeled falsepal pre-mRNA substrate should be insensitive to competition with a vast excess of wtpal compRNA that is presumed to bind the HBF. As shown in Fig. 4C, this is indeed what is observed. Thus these experiments suggest that the residual 20% of the processing activity can be attained either by preempting the HBF with excess of wtpal compRNA or by alterations in the stem-loop sequence that prevent binding of the HBF to its target site. Interestingly, when the labeled wild-type processing substrate is challenged by an excess of fstem/wtloop compRNA, no inhibition can be observed (Fig. 4C). Apparently, more than just the loop sequence is required for efficient recognition by the HBF; rather the stem nucleotides of the wild-type hairpin appear to play a role during 3' processing, perhaps by providing necessary contact points for the presumptive HBF. Rescue Experiments. If the competition experiments reflect the sequestration or preemption of a factor that can then no

4348

Biochemistry: Vasserot et al.

Proc. Natl. Acad Sci. USA 86 (1989)

B

A 2

0)~~ oo 000 00 000

0XCry

')s

T

Co)

_EoQz+1Q.. _ _

A _

oXo

5o o o o D3 X cX CO) - XI CL Z a .. .. .... C + Ir _ _ _ _

cam)

....

_

] INPUT[

B

mm

a) e

C CI ..

_UEE_!UE~iIIIP ]INPUT [ EII

Inu _ _inbw O

]PROC.E ] PROC.E

_

qW

_NW

_

-p

wtpal comp RNA (-o-)

complete wt comp RNA (---) C processing

activity R

A

100%

1 80% 60% 40%

:X X~~~~~~~~

20% 0% 1:1

1:3

1:10

4b 1 2 3 4 5 6 7 8 91011

1

2 3 4 5 6

FIG. 5. Rescue experiment. (A) HBF is Sm-precipitable. Lanes: 1, pre-mRNA input; 2, processing in 7.5 ,4l of nuclear extract; 3, same as lane 2 but in presence of 10-fold molar excess of wtpal compRNA (referred to as processing in preempted extract); 4, 7.5 JAI of untreated extract restores processing of 7.5 t.l of preempted extract; 5, Hpa II-digested pBR322; 6-8, each fraction tested for complementation is by itself inactive in processing; 9-11, ability of these fractions to restore processing of preempted extracts. (B) HBF can be separated from U7 snRNA. Lanes: 1, input RNA; 2, processing in nuclear extract; 3, processing in preempted extract; 4, Hpa IT-digested pBR322; 5, restoration of processing activity by Mono Q fraction 15 (compare with lane 3); 6, complementation between an Sm-depleted

nuclear extract and Mono Q fraction 15.

U

1:0

a

1:30

1:100 1:300 1:1000 1:3000

FIG. 4. Effect of competitor RNAs on processing activity. The competitor ratios are given in molar excess over the labeled substrate RNA. NX, nuclear extract; Hpa marker, Hpa II-digested pBR322; Proc., processed RNA. (A) Autoradiogram of the labeled wild-type substrate in self-competition. (B) Autoradiogram of the labeled wild-type substrate in presence of wtpal compRNA. (C) m, Labeled wild-type substrate in self-competition; o, in the presence of wtpal compRNA; *, in the presence of downstream spacer element compRNA; ci, in the presence of falsepal compRNA; A, in the presence of fstem/wtloop compRNA; A, labeled falsepal mutant substrate in presence of wtpal compRNA.

longer bind to the reporter substrate, one should be able to restore processing by adding back HBF-containing fractions to the preempted processing extract and use this rescue experiment to characterize the relevant component(s) in a preliminary way. When additional untreated nuclear extract is added to the preempted extracts, processing is restored to near normal levels, as expected (Fig. 5A, lane 4). Addition of the cytoplasmic supernatant (S100 fraction) has no effect (lane 11), suggesting that the HBF is a nuclear factor. Since the histone hairpin is known to mediate regulation of the replication variant mRNA pools (37, 38) and since the heatlabile factor (29) may play a role in this regulatory circuit (39), we wanted to know whether the HBF and the heat-labile factor were one and the same. This appears not to be the case because nuclear extracts preheated to 50'C for 15 min [a procedure that completely inactivates the heat-labile factor but leaves the snRNPs apparently unaffected (29)] are still capable of enhancing processing of a preempted extract (lane 10). By contrast, nuclear extracts depleted with anti-Sm antibodies do not rescue processing activity (lane 9). This leads us to the conclusion that the HBF removed during Sm depletion carries an Sm determinant or is associated with a particle (perhaps U7 snRNP) precipitable with this antiserum (see Discussion). U7 snRNA Does Not Participate in the Restoration of Processing Activity in HBF-Depleted Extracts. To definitely rule out the participation of the U7 snRNA in the observed phenomenon, we attempted to separate it from the HBF

activity using a column chromatography fractionation procedure developed by Kramer et al. (40). When Mono Q fractions are assayed for their ability to restore processing in preempted extracts, the HBF activity shows a broad distribution noticeably different from, but overlapping with, the other two known processing components, the heat-labile factor and the U7 snRNP. A particular fraction can be obtained that, in a complementation test, shows HBF activity (Fig. 5B, compare lanes 3 and 5) but lacks U7 snRNP (lane 6). Furthermore, SP6 mapping failed to reveal the presence of any U7 snRNA in this Mono Q fraction, demonstrating that the HBF activity does not require the presence of the U7 snRNA to restore processing. Hence, the HBF appears to be separable from the U7 snRNP. It could represent some still unidentified snRNP or could be, at its simplest, a processing protein (see Discussion).

DISCUSSION The role of the two conserved RNA elements in producing the mature 3' end of histone mRNAs has been extensively investigated and their paramount importance unanimously demonstrated (for review, see refs. 6 and 7). The understanding of their respective contribution to the processing reaction began with the demonstration that the downstream spacer motif is a target for RNA-RNA duplex formation with the 5' terminal sequence of the U7 snRNA (24), as proposed (22). The present report focuses on the molecular mechanisms mediated by the evolutionary highly conserved hairpin structure (34) that is included in both histone pre-mRNA and the mature mRNA. Our competition and rescue experiments suggest that the hairpin loop provides another anchor point for the processing machinery. Indeed, a HBF appears to specifically recognize the histone stem-loop structure. The interaction between the inferred HBF and its RNA target is of major importance since it contributes to as much as 80% of the processing efficiency under the in vitro experimental conditions used. Apparently, the binding of the HBF to the histone pre-mRNA can be interfered with in one of several ways: addition of a wild-type hairpin competitor RNA, changes in the primary sequence, or linearization or deletion of the HBF target site. We hypothesize that in all these cases the processing apparatus

Biochemistry: Vasserot et aL is stalled by blockage of the ability of HBF to make contacts with the histone hairpin and this results in a loss of 80-95% of processing rate, the residual processing activity being presumably directed, in the main, by the downstream spacer element. The question remains unanswered whether the HBF is a free entity or whether under physiological conditions it is part of the U7 snRNP. The slightly different chromatographic behavior of the HBF and U7 snRNP on a Mono Q column leads to a partial separation of these two activities (although this could be a consequence of dissociation due to salt concentration); thus HBF as a distinct entity, free from U7 snRNP, can be recovered. Since an Sm-depleted nuclear extract cannot complement preempted extracts, we suggest that the HBF carries an Sm determinant. Mowry and Steitz (41) have described a non-snRNP factor binding upstream of the mouse H3 precursor processing site. This factor, present in a fraction that contains proteins between 100 and 500 kDa but does not contain (most of the) snRNPs and is anti-Sm reactive, and its binding is insensitive to prior micrococcal nuclease treatment (41). We do not know at present whether the HBF described here and their non-snRNP factor are identical. The HBF could also be a U7 snRNP protein or a factor associated with this particle. In this context, it is noteworthy that in sea urchin and mouse, both histone and U7 snRNA hairpins are brought in exactjuxtaposition when RNA duplex formation between the downstream spacer motif and the U7 snRNA has occurred, even though the length of the small RNAs are quite divergent (22, 30, 42). This strict topological relationship also brings the Sm-binding site of the U7 snRNP in close proximity to the hairpin of the histone pre-mRNA (for illustration, see refs. 7 and 42). There is no evidence at the moment that one of the Sm proteins of U7 snRNP is the same as the HBF we have detected. However, there are precedents for snRNP proteins mediating substrate binding. For instance, a micrococcal nuclease-treated (inactive) RNase P retains its capacity to bind tRNA precursors, suggesting an important role of a protein component (43). Moreover, the 5' splice site selection seems to also require U1 snRNP proteins (44, 45), the intron binding protein appears to be U5-associated (46, 47), and the 64-kDa protein that recognizes the AAUAAA sequence (48) could be snRNP-associated as well (49). Clearly, further characterization of the HBF is needed and will provide insights into the exact mechanism of the processing reaction. We are very grateful to Gotthold Schaffner and Klaus Kalusa for oligonucleotide syntheses, to Eddy De Robertis for his generous gift of anti-Sm sera, to Octavian Gick-Schatz for providing the deltapal and Cpal constructs, to Martin Nicklin for the gift of a powerful T7 polymerase, and to the members of the Service Department of the Research Institute of Molecular Pathology for sequencing. We also thank Matt Cotten for his valuable help and advice and Margaret Chipchase and Meinrad Busslinger for critical reading of the manuscript. The EB1 mouse cells were used with the kind permission of Peter Swetly (Bender, Vienna). We thank Ingeburg Hausmann and Fritz Ochsenbein for the preparation of the figures, and Marianne Vertes for secretarial help. This work was supported by the Research Institute of Molecular Pathology and in part by grants of the Swiss National Science Foundation and the Kanton of Zurich. This work was partial fulfillment of the Ph.D. requirements for A.P.V. and for F.J.S. who was supported by a postgraduate scholarship from the Natural Science and Engineering Research Council of Canada. 1. Platt, T. (1986) Annu. Rev. Biochem. 55, 339-372. 2. Bram, R. J., Young, R. A. & Steitz, J. A. (1980) Cell 19, 393-401. 3. Brawerman, G. (1987) Cell 48, 5-6.

Proc. Natl. Acad. Sci. USA 86 (1989)

4349

4. Nomura, M., Gourse, R. & Baughman, G. (1984) Annu. Rev. Biochem. 53, 75-117. 5. Noller, H. F. (1984) Annu. Rev. Biochem. 53, 119-162. 6. Birnstiel, M. L., Busslinger, M. & Strub, K. (1985) Cell 41, 349-359. 7. Bimstiel, M. L. & Schaufele, F. J. (1988) in Structure and Function ofMajor and Minor Small Nuclear Ribonucleoprotein Particles, ed. Birnstiel, M. L. (Springer, Heidelberg), pp. 155-182. 8. Schumperli, D. (1986) Cell 45, 471-472. 9. Marzluff, W. F. & Pandey, N. B. (1988) Trends Biochem. 13, 4952. 10. Aziz, N. & Munro, H. N. (1987) Proc. Natl. Acad. Sci. USA 84, 8478-8482. 11. Casey, J. L., Hentze, M. W., Koeller, D. M., Caughman, S. W., Rouault, T. A., Klausner, R. D., & Harford, J. B. (1988) Science 240, 924-928. 12. Kozak, M. (1986) Proc. Natl. Acad. Sci. USA 83, 2850-2854. 13. Cech, T. R. & Bass, B. R. (1986) Annu. Rev. Biochem. 55, 599-629. 14. Hall, K. B., Green, M. R. & Redfield, A. G. (1988) Proc. Natl. Acad. Sci. USA 85, 704-708. 15. Solnick, D. & Lee, S. I. (1987) Mol. Cell. Biol. 7, 3194-3198. 16. Mattaj, I. W. (1988) in Structure and Function of Major and Minor Small Nuclear Ribonucleoprotein Particles, ed. Birnstiel, M. L. (Springer, Heidelberg), pp. 100-114. 17. Gilmartin, G. M., Schaufele, F. J., Schaffner, G. & Birnstiel, M. L. (1988) Mol. Cell. Biol. 8, 1076-1084. 18. Christie, G. E., Farnham, P. J. & Platt, T. (1981) Proc. Natl. Acad. Sci. USA 78, 4180-4184. 19. Romaniuk, P. J., Lowary, P., Wu, H. N., Stormo, G. & Uhlenbeck, 0. C. (1987) Biochemistry 26, 1563-1568. 20. Thomas, M. S. & Nomura, M. (1987) Nucleic Acids Res. 15, 3085-

3096. 21. Feng, S. & Holland, E. C. (1988) Nature (London) 334, 165-167. 22. Strub, K., Galli, G., Busslinger, M. & Birnstiel, M. L. (1984) EMBO J. 3, 2801-2807. 23. Vitelli, L., Kemler, I., Lauber, B., Birnstiel, M. L. & Busslinger, M. (1988) Dev. Biol. 127, 54-63. 24. Schaufele, F., Gilmartin, G. M., Bannwarth, W. & Birnstiel, M. L. (1986) Nature (London) 323, 777-781. 25. Dignam, J. D., Lebovitz, R. M. & Roeder, R. G. (1983) Nucleic Acids Res. 11, 1475-1489. 26. Melton, D. A., Krieg, P. A., Rebagliati, M. R., Maniatis, T., Zinn, K. & Green, M. R. (1984) Nucleic Acids Res. 12, 7035-7056. 27. Gick, O., Kramer, A., Keller, W. & Birnstiel, M. L. (1986) EMBO J. 5, 1319-1326. 28. De Robertis, E. M., Lienhard, S. & Parisot, R. E. (1982) Nature (London) 295, 572-577. 29. Gick, O., Kramer, A., Vasserot, A. & Birnstiel, M. L. (1987) Proc. Natl. Acad. Sci. USA 84, 8937-8940. 30. Cotten, M., Gick, O., Vasserot, A., Schaffner, G. & Birnstiel, M. L. (1987) EMBO J. 7, 801-808. 31. Birchmeier, C., Grosschedl, R. & Birnstiel, M. L. (1982) Cell 28, 739-745. 32. Tinoco, I., Borer, P. N., Dengler, B., Levine, M. D., Uhlenbeck, 0. C., Crothers, D. M. & Gralla, J. (1973) Nature (London) 84,498502. 33. Mohun, T., Maxson, R., Gormezano, G. & Kedes, L. (1985) Dev. Biol. 108, 491-502. 34. Busslinger, M., Portmann, R. & Birnstiel, M. L. (1979) Nucleic Acids Res. 6, 2997-3008. 35. Birchmeier, C., Folk, W. & Birnstiel, M. L. (1983) Cell 35,433-440. 36. Lynn, S. P., Kasper, L. M. & Gardner, J. F. (1988) J. Biol. Chem. 263, 472-479. 37. Luscher, B., Stauber, C., Schindler, R. & Schumperli, D. (1985) Proc. Natl. Acad. Sci. USA 82, 4389-4393. 38. Stauber, C., Luscher, B., Eckner, R., Lotscher, E. & Schumperli, D. (1986) EMBO J. 5, 3297-3303. 39. Luscher, B. & Schumperli, D. (1987) EMBO J. 6, 1721-1726. 40. Kramer, A., Frick, M. & Keller, W. (1987) J. Biol. Chem. 262, 17630-17640. 41. Mowry, K. L. & Steitz, J. A. (1987) Science 238, 1682-1687. 42. Soldati, D. & Schumperli, D. (1988) Mol. Cell. Biol. 8, 1518-1524. 43. Nichols, M., So6l, D. & Willis, I. (1988) Proc. Natl. Acad. Sci. USA 85, 1379-1383. 44. Mount, S. M., Pettersson, I., Hinterberger, M., Karmas, A. & Steitz, J. A. (1983) Cell 33, 509-518. 45. Zhuang, Y. & Weiner, A. M. (1986) Cell 46, 827-835. 46. Tazi, J., Alibert, C., Temsamani, J., Reveillaud, I., Cathala, G., Brunel, C. & Jeanteur, P. (1986) Cell 47, 755-766. 47. Gerke, V. & Steitz, J. A. (1986) Cell 47, 973-984. 48. Wilusz, J. & Shenk, T. (1988) Cell 52, 221-228. 49. Takagaki, Y., Ryner, L. C. & Manley, J. L. (1988) Cell 52, 731-742.