- Email: [email protected]
Birth of the snoRNPs: the evolution of RNase MRP and the eukaryotic pre-rRNA-processing system
TiBS 20 - FEBRUARY1995 EUKARYOTIC RIBOSOME biogenesis takes place In the nucleolus, a specialized compartment of the nucleus. In both higher and lower eukaryotes, a large number of small nucleolar ribonucleoprotein particles (snoRNPs) have been localized to the nucleolus~,2.Some of these snoRNPs, vertebrate U3, U8 and U22 and yeast U3, U14 and snR30, are involved in pre-ribosomal RNA (prerRNA) processing. Although these results firmly establish the requirement for snoRNPs in pre-rRNA processing, the precise functions of the particles are still not known. Recently, it has emerged that an endoribonuclease, ribonuclease MRP (RNase MRP), is also required for nucleolar pre-rRNA processing (reviewed in Ref. 3). RNase MRP is a ribonucleoprotein (RNP) particle that is structurally related to the ubiquitous RNP RNase P, but is distinct from other snoRNPs. Here we review the role of RNase MRP in pre-rRNAprocessing and discuss evidence that indicates that RNase MRP is evolutionarily derived from RNase P. A model is also presented that suggests how other aspects of eukaryotic pre-rRNA processing may have evolved from a slmpler, ancestral system present in bacteria and archaea. RNm MRP RNase MRP was purified, initially from mouse cells4 and later from yeast5, through Its ability to cleave RNA primers used in the replication of mltochondrlal DNA in vitro, It subsequently became clear, however, that most cellular RNase MRP is in the nucleolus6, and genetic evidence obtained from 5accharomyces cerevisiae has shown unequivocally that RNase MRP plays an important role in pre-rRNAprocessing. Like most eukaryotes, S. cerevisiae synthesizes Its 18S, 5.8S and 25S rRNAs as a single precursor, the 35 S pre-rRNA, whlch is processed through a number of intermediates to yield the mature rRNAs (see Fig, 1 for a description of the yeast pre-rRNA processing pathway). There are two alternative pathways for processing In internal transcribed spacer I 0TSI), which give rise to two forms of the 5.8 S rRNA, designated 5.8S(L) and 5.8S(S)T. Formation of 5.8 S(S) but not 5.8 S(L) requires cleavage in ITS1 at site A3.
{. P. L ~ s e y and D. Toileweyare at the EuropeanMolecularBiologyLaboratory (EMBL),Postfach1.02209,D-6g012 Heidel~rg, Germany.
Birth of the snoRNPs:the evolution of RNaseMRP the eukaryotic pre-r A-Foc ing system John P. Morrissey and David Tollervey The ribonucleoprotein particle RNase MRP is required for the processing of yeast pre-ribosomal RNA (pre-rRNA). A structurally related particle, RNase P, is univers~;tliyrequired for processing of pre-tRNA, but in bacteria and archaea also cleaves a site in the pre-rRNA. This suggests that RNase MRP may have arisen in eukaryotes as a form of RNase P specialized for pre-rRNA processing. Other eukaryotic small nucleolar RNAs may have arisen as trans-acting factors that functionally replace cis-acting prerRNA interactions in bacteria and archaea. The RNA component of RNase MRP (MRP RNA or 7-2 RNA) is encoded by the NMEI gene in yeast, which is a single-copy gene essential for cell viabilitys. It transpires that NMEI is allellc with RRP2 (Ref. 9), temperaturesensitive (is) alleles of which had previously been isolated independently by two groupsl°.~L Strains carrying the ts rrp2-1 allele or strains genetically depleted of the RNA component of RNase MRP~ are Impaired in the synthesis o{ 5.8 S(S) rRNA but can synthesize 5.8S(L) rRNA normally. The same phenotype is seen in a mutant carrying a deletion of the A3 cleavage site7. A protein from S. cerevisiae, POP1, which is a common component of RNase MRP and RNase P, has been characterized and its gene has been cloned. Cleavage at the A3 site is lost in strains carrying either a ts popl-I allele or the rrp2.1 allele~. Taken together, these data indicate that RNase MRP cleaves the pre-rRNAin ITSI at site A3.
protein is required for fullactivityin vitro, and is essential in vivo (for a review of the role of the bacterialprotein, see Ref. 17").The only nuclear RNase P that has been purified to homogeneity is that of the fission yeast, Schizosaccharomyces pombe, and this particle also contains one protein, although it has a considerably larger mass than Its bacterial counterpart is. Although there is very little sequence conservation between the RNA components of RNase P from different organisms, the tertiary structure, particularly the so-called 'cage domain' formed by long-range base-pairing interactions, is conserved ~9. The conservation of an antigenic epitope between bacterial and human RNase P proteins has also been reported 2°, emphasizing the ancient nature of RNase P. The activity of RNase P in pre-transfer RNA (pre-tRNA) processing appears to be universally conserved, but in bacteria RNase P also cleaves the precursors of 4.5S RNA2~ and of 10S RNA22. Biochemical studies on the RNA Mm P and the holoenzyme (reviewed in Ref. RNase P is present in all three phylo- 17) have led to the proposal that ancesgenetic domains (i.e. in bacteria, tral RNase P consisted entirely of RNA archaea and eukaryotes), as well as and cleaved tRNA-like molecules, and in mitochondria, demonstrating its that later in evolution the enzyme ancient evolutionary origin (for reviews acquired a protein component, and see Refs 14-16). Bacterial RNase P has with it the ability to recognize other been best studied, and consists of a substrates, such as 4.5 S RNA. A recent short RNA molecule and one protein. study by Liu and Altmanxl supports this The RNA molecule alone has some model. These workers randomized enzymatic activity in vitro, but the nucleotides of an RNase P substrate, © 1995,ElsevierScienceLtd 0968-0004/95/$09.50
REVIEWS
TgBS 20 = FEBRUARY 1995
and then selected substrates that were cleaved efficiently, either by the RNA
A0 A1
A2
35S
moiety alone, or by the holoenzy~e
consisting of the RNA and the protein. It was found that the RNA enzyme selected substrates that structurally resembled tP~NA, whereas the holoenzyme selected substrates resembling both tRNA and 4.5 S rRNA. Bacteria, like eukaryotes, general|y transcribe their rRNA as a single precursor, which is subsequently processed to yield the mature rRNA molecules. There are, however, significant differences between bacterial and eukaryotic pre-rRNA transcripts (Fig. 2). One difference is that in most eukaryotes the large-subunit rRNA is split into two molecules, 5.8 S rI~NAand 25 S rRNA, by the insertion of a second internal transcribed spacer (ITS2), whereas bacteria have a single 23S rRNA molecule. Despite this, the rRNAs are homologous and there is a high degree of conservation between 5.8 S rRNA and the 5'-end of 23 S rRNA. A further difference in bacterial pre-rRNA is the presence of a tRNA molecule in the spacer between the small (16 S) and large (23 S) subunit rRNAs. The prerRNA-processing pathway in bacteria is well characterized 23,24. The spacers immediately flanking the 16 S and 23 S rRNAs are capable of forming long helices, causing the rRNA molecules to be looped out, and bringing the ends of the mature rRNA regions into close proximity. An endoribonuclease, RNase 111, recognizes an undefined structural feature of these helices and endonucleolytically cleaves both strands, generating 17 S and 23 S pre-rRNA molecules that have spacer-sequence extensions at the 5'- and 3'-ends. These extensions are subsequently removed by exo- and endonucleolytic enzymes. As with all tRNAs, the tRNA in the spacer between the 16 S rRNA and the 23 S rRNA is processed at the 5'-end by RNase P. Mutants impaired in RNase Ill function are viable because RNase P cleavage of the pre-rRNA can still occur; the 5'-product of this cleavage is an efficient substrate for the 16 S rRNA maturation enzymes, and the 23S rRNA, although extended at the 5'- and 3'ends, can still be assembled into functional ribosomes 23. Since the majority of the tRNA genes are not located in rDNA operons, the conservation of a tRNA molecule in the rRNA spacer of most bacteria and archaea suggests a functional significance. As there is no conservation in the species of tRNA that is
°3 1
U14 snR30 20S
18S
CIeavage A0, A1 A2 A3
27SA
/
q
A3 81(S)
27SA
[32
B1(L)
27SA
B2
/
RNase MRP
C~eavage A3 27SA'
.
XRNI RAT1
1
E×onuclease
A3 --, B1 (S)
27SB(S)
J 5.8S(8)
25S
27SB(L)
ITS2 processing 5.88(L)
25S
Figure 1 Pre-rRNA processing in S. cerevisiae. For clarity the pre-rRNA processing pathway has been summarized; a more detailed figure can be found in Ref. 13. The first steps in pre-rRNA processing involve cleavage of 35 S pre.rRNAat the AO and A1 sites in the 5'-external transcribed spacer (5'-ETS) and at the A2 site in the first internal transcribed spacer (ffS1), generating the 20 S and 27 SA pre-rRNAs. Tile small nucleolar RNAs (snoRNAs} U3, U14 and snR30, and the associated small nucleolar ribonucleoprotein particle (snoRNP) proteins NOP1, SOFI and GAR1 are required for these cleavage reactionst,2'2°. The 20 S pre-rRNA is transported to the cytoplasm where it is matured to 18 S rRNA, whereas the 27 SA pre-rRNA remains in the nucleolus where it is further processed by one of two alternative pathways. Approximately 90% of the 27 SA pre-rRNA molecules are cleaved at the A3 site, 77 nucleotides upstream of the BI(S) site. Following this cleavage, the XRN1 and RAT1 exonucleases rapidly digest to the BI(S) site 7. The remaining 10% of 27 SA pre.rRNA molecules are processed at the BI(L) site by an undetermined mechanism. In S. cerevisiae there are two predominant forms of 5.8 S rRNA, differing by seven nucleotides at the 5"-end; the BI(S) and BI(L) sites are the 5'-ends of the short, 5.8 S(S) rRNAand the long, 5.8 S(L) rRNA, respectively. Contemporaneously with the processing reactions forming the BI(S) and BI(L) ends, the B2 site in the 3'-external transcribed spacer (3;ETS) is cleaved. The mechanism and precise relative timing of this cleavage is not known. The 27 SB(S) and 27 SB(L) pre.rRNAs are processed to mature 5.8 S(S) and 5.8 S(L) rRNA, respectively, and to 25 S rRNA, by removal of the second internal transcribed spacer (ITS2) in a set of reactions not known to require snoRNPs.
present in the pre-rRNAs, it is likely that the presence of a tRNA per se, rather than any particular tRNA, is of functional importance. We propose that a tRNA has been maintained in this location so as to preserve the RNase P cleavage site in the rRNA spacer. RNase P cleavage of the pre-rRNA could then serve as an alternative processing pathway for the synthesis of 16 S rRNA. It is not unusual to find redundant mechanisms
for performing a reaction crucial to a cell's survival; having alternative mechanisms for carrying out a reaction may be the most effective way of ensuring that the reaction proceeds with high efficiency.
Model for the evolution of RNase MRP RHase MRP, which has been found only in eukaryotes, is structurally related to RHase P. Phylogenetic comparison
19
TIBS 2 0 -
RNase P
1 E. colt
~
i
16S
't I lt
I
S. cersv/s/se 5'-ETS ~ 18S
-
:,
'i tRNA '
I !
i I
I
I
23S
.
", 5S
t l
ITS1 ITS2 ~ 5,8S
%
*
3-E'rS_ 25S
RNase MRP Rgure 2
Comparison of E. colt and S. cerevisiae precRNAs. Although the basic arrangement of the rRNA is generally conserved between bacteria and eukaryotes, there are three major differences. Rrst, in most eukaryotes the large-subunit rRNA is split by the insertion of a spacer, ITS2. Second, the 5 S rRNA is part of the bacterial pre-rRNAbut is synthesized independently in eukaryotes. Third, bacteria have a tRNA molecule in the spacer between the 16 S and 23 S rRNAs. The species of tRNA varies, but its presence is conserved. Some bacterial pre-rRNAs have further tRNAs downstream of the 5 S rRNA, but the presence of these tRNAs is not conserved. The positions of the RNase P and RNase MRP cleavage sites are indicated.
shows that the MRP RNA can be folded into a secondary structure that is similar to the structure of the RNase P RNA~. This predicted structure includes the highly conserved cage domain, which has been proposed to be Involved In catalysis~9. Human autoImmune antibodies, the Th/To sera, coprecipitate both MRP RNA and RNase P RNA, suggesting that the two particles contain a common proteln epltope. This was confirmed by the identlflcatlon of the POPI protein in S. cerevistae, This protein Is a component of both RNase P and RNase MRP: Immunopreclpltatlon of epltope-tagged POP1 co-preclpltates RNase P RNA and MRP RNA, and a strain carrying a mutant popl.1 allele Is Impaired In the function of both RNase P and RNase MRPm. From the structural similarities it is clear that RNase P and RNase MRP are evolutlonarlly related. Given that RNase P is ubiquitous and of ancient origin, whereas RNase MRP has been found only In eukaryotes, it seems likely that RNase MRP was deriv~ from RNase P. We envisage that, shortly after the divergence of the eukaryotes and the archaea, the RNA component of RNase MRP arose by duplication of an ancestral R N a s e P RNAgene. Alignment of the bacterial and e~aryotlc pre-rRNAs shows that the bacterial RNase P cleavage site and the eukaryotlc RNase MRP cleavage site are in similar positions (Fig. 2). Since a tRNA molecule is found in the pre-rRNA spacer of both bacteria and archaea, it is likely that this represents the ancestral state. Duplication of the RNase P RNA gene in an early eukaryote would
have allowed co-evolution of the enzyme and pre-rRNA cleavage site. We propose that this gave rise to RNase MRP and site A3 in modern eukaryotes. In both S. cerevisiae and E. colt, the RNase MRP and the RNase P cleavages represent alternative processing pathways for the pre-rRNA. Conservation of alternative processing pathways throughout evolution supports the idea that this is a means of ensuring high processing efficiency. This model explains the origin of RNase MRP and tile reason for its cleavage of the pre~rRNA, but does not exclude the possibility that RNase MRP has acquired other functions In the cell, including a role In mitochonddal DNA replication. OdgJns of eukaryotic pre.~NA procesdng
In S. cerevisiae, the snoRNPs U3, U14 and snR30 are required both for process!n~ at site A1, the 3'-end of the 5'external transcribed spacer (5'-ETS), and for processing at site A2 in ITS1 (Fig. 1)i,2,29. Genetic depletion of the RNA or protein components of any of these particles abolishes cleavage at both A1 and A2, preventing synthesis of the 18 S rRNA but allowing synthesis of both iarge-subunit rRNAs, 5.8S rRNA and 25 S rigA. At least one of the snoRNPs, U3, binds to a sequence in the 5'-ETS27, and it is postulated that the snoRNPs also interact with sequences in ITS1, forming a complex in which the 5'-ETS and ITS1 are in close proxJ~.ity with the looped out 18S rRNA. Consistent with this, deletions in the 5'-ETS also inhibit cleavage at both AI and A2
FEBRUARY
1995
(Refs 27, 28). Similarly, depletion of the U22 snoRNA in Xenopus oocytes inhibits processing in both the 5'-ETS and ITS1 (Re[. 29). In S. cerevisiae, coupling has also been proposed between processing at site BI, the 3'-end of ITS1, and at site B2, the 5%rid of the 3'-ETS (Fig. 1)3°. There are also data from other eukaryotes that suggest a connection between processing in the 3'-region of ITS1 and in the 3'-ETS. Depletion of the U8 snoRNA in Xenopus oocytes impairs processing both at the 3'-end of ITS1 and at site T1 in the 3'-ETSm. The consequence of this inhibition is that the 5.8S and 28S rRNAs are not synthesized, whereas synthesis of the 18 S rRNA continues. In oocytes to which partial processing activity has been restored by injection of U8, pre-rRNAs cleaved at only one of these sites are not detected, leading to the suggestion that processing in ITS1 and in the 3'-ETS is coupled and that this coupling requires the U8 snoRNP 2. In S. pombe, mutations in the 3"ETS also impair processing in ITS1 and abolish synthesis of the large-subunit rRNAs33, supporting the idea that processing in ITS1 and the 3'-ETS are interdependent events. A general model for eukaryotic prerRNA processing can be postulated, in which snoRNP complexes bring the ends of the rRNA molecules into close proximity, and in this way substitute for the terminal base-pairing interactions found in bacteria and archaea (Fig. 3). This would enable the eukaryotlc cell to co-ordinate processing at the 5" and 3" ends of an rRNA molecule. Given that the archaeal and bacterial systems are so similar (for a review of archaeal rRNA operons see Ref. 34), it is likely that the ancestral pre-rRNA relied on spacer-sequence base-pairing for its processing system. In yeast, there are clear structural and functional differences between MRP RNA and the 14 other snoRNAs characterized to date. With the exception of MRP RNA, all of the snoRNAs are precipitable by antibodies against the NOP1 protein (the yeast homologue of fibrillarin; Ref. 1; D. Tollervey, unpublished). Functionally, RNase MRP is an endoribonuclease in vitro, whereas no other snoRNP has been shown to have nuclease activity, although it is possible that multiple snoRNPs assemble together with the pre-rRNA to form an RNAbased cleavage complex. It seems likely that these differences are related to a fundamental difference between the
REVIEWS
TIBS 20 = FEBRUARY 1 9 9 5 evolutionary origin of RNase MRP and the other snoP~Ps. There is a precedent for the function of cis-acting RNA
sequences being usurped by trans-acting small nuclear ribonucleoprotein particle (snRNP) factors. Group |l introns, which are found in mitochondria and chloroplasts, are autocatal~ic and are responsible for their own removal from the pre-mPdqA and the coupled ligation of the resulting exons. Nuclear pre~messenger Pd~IA ~remRNA) splicing proceeds by a similar pathway but requires the UI, U2, U4/U6 and U5 snRNPs. It is believed that the
Euka~otes
SHORN
Bacteria
oRNPs
RNas
ITS1
ase roll tRNA
Figure 3 Model comparing eukaryotic and bacterial pre-rRNA processing. The spacers flanking the bacterial rRNA molecules form extensive helices, which contain the recognition sites for RNase ill. We propose that in eukaryotes small nucieolar ribonucleopretein particle (snoRNP) complexes have replaced intramolecular base-pairing in bringing the 5" and 3;flanking sequences of the mature rRNAstogether. As discussed in the text, there is evidence from S. cerevisiaethat the U3, Ut4 and snR3O small nucleolar ribonucleoproteins (snoRNPs) may be part of the complex looping out the small subunit rRNA and that, i~ Xenopus,U8 may be part of the complex looping out the large subunit rRNAs.
group I1 self-~plicing introns represent the ancestral state and that the transacting snP,NPs evolved from, and took over the functions of, cis-acting intron sequences (for a recent review of splicing see Ref. 35). An intermediary state may be found in trypanosomatids, where U1 function is provided in cis, but some of the other snRNP functions are provided in trans36. The model that the eukaryotic ITSI is related to the bacterial internal spacer, and that the 5'- and 3'-ETS are related to the bacterial external spacers, implies that these spacers are more ancient than ITS2. Consistent with this is the observation that a deeply branched eukaryote, the microsporidium Vairimorpha necatrix, does not have an ITS2 and has a single 23S-like large-subunlt rRNA36. The suggestion that the ITS2 spacer became inserted into the iarge-subunit rRNA after the divergence of eukaryotes is supported by the case of trypanosomatids, where multiple spacers appear to have been inserted during evolution; for example, the large-subunit rRNA of Crithidia fasiculata is interrupted by six spacers 38. if ITS2 is unrelated to the other eukaryotic spacers it might be expected that it would be processed by an independent mechanism. This would provide an explanation for why so many yeast mutants, including the snoRNP mutants, affect processing in the 5'-ETS and ITS1, but do not affect ITS2 processing.
Concludingremarks We have presented two models that offer explanations for a number of recent observations on the function of eukaryotic snoRNP particles. In both,
we envisage that, following divergence of the eukaryotes from the archaea and bacteria, elaboration of existing, simpler processing systems gave rise to the pre-rRNA processing components seen in modern eukaryotes. These models postulate quite different evolutionary origins for different eukaryotic snoPd~Ps. Pd~lase MP& is envisaged as being derived from the ancient ribonuclease Pd~lase P. By contrast, we propose that the other snoRNPs involved in processing arose as transacting factors that functionally replace the interactions between the spacer sequences in bacterial and archaeal pre-rRNAs. Whether the snoRNAs are directly derived from the spacers or represent an independent solution to the problem of bringing the ends of the mature rILNAs together, is unclear at present. It is also still unclear whether the snoPd~IPs function in processing by presenting the pre-rPd~lA in the correct conformation for ~roteinaceous) processing nucleases, or whether multiple snoP.d~Ps assemble with the pre-rPd~lA to form an PallIAbased processing complex. Finally, it is notable that a further group of snoRNAs, which are encoded in the introns o[ other genes and are characterized by high sequence complementarity to mature rPd~lAsequences, has been described39; it is possible that these snoPd~IAs have a different evolutionary origin from the snoPd~lAsshown to be involved in pre-rP,NA processing. More detailed functional characterization may shed further light on the origins of the various groups of snoRNPs.
Acknowl~gemenL~ We would like to thank E. [rare and many colleagues at EMBL for their helpful discussions and comments on
the manuscript.
Re,fences 1 Rlipowicz, W. and Kiss, T. (1993) Mol. Biol. Rep. 18, 149-156 2 Fournier, M. J. and Maxwell, E. S. (1993) Trends Biochem. Sci. 18, 131-135 3 Clayton, D. A. (1994) Proc. Nati Acad. Sci. USA 91, 4615-4617 4 Chang, D. D. and Clayton, D. A. (1987) EMBOJ. 6, 409-417 5 Stohl, L. L. and Clayton, D. A. (1992) Mol. Cell. Biol. 12, 2561-2569 6 Kiss, T., Marshallsay, C. and Rlipowicz, W. (1992) EMBOJ. 11, 3737-3746 7 Henry, Y. et al. (1994) EMBOJ. 13, 2452-2463 8 Schmitt, IVl. E. and Clayton, D. A. (1992) Genes Dev. 6, 1975-1985 9 Chu, S., Archer, R. H., Zengel, J. M. and Lindaill, L. (1994) Proc. Natl Acad. ScL USA 91, 659-663 10 Shuai, K. and Warner, J. W. (1991) Nucleic Acids Res. 19, 5059-5064 11 Lindahl, L., Archer, R. H. and Zengal, J. M. (1992) Nucleic Acids Res. 20, 295-301 12 Schmitt, M. E. and Clayton, D. A. (1993) Mol. Cell. BioL 13, 7935-794.1 13 Lygerou, Z. et aL (1994) Genes Dev. 8, 1423-1433 14 Airman, S. (1989) Adv. EnzymoL62, 1-36 15 Darr, S. C., Brown, J. W. and Pace, N. R. (1992) Trends Biochem. ScL 17, 178-182 16 Altman, S., Kirsebom, L. and Talbot, S. (1993) FASEBJ. 7, 7-14 17 Gopolan, V., ralbot, ~. J. and Altman, S. in RNA--Protein Interactions (Nagai, K. and Mattaj, I. W., eds), Oxford University Press (in press) 18 Zimmerly, S., Drainas, D., Sylvers, L. A. and SBII, D. (1993) Eur. J. Biochem. 217,501-507 19 Forster, A. C. and Altman, S. (1990) Cell 62, 407-409 20 Mamula, M. J., Baer, M., Craft, J. and Altman, S. (1989) Prec. Natl Acad. Sci. USA 86, 8717-8721 21 Liu, F. and Altman, S. (1994) Cell 77, 1093-1100
81
TiBS 2 0 - FEBRUARY1995 22 Komine, Y, et at, Proc. Natl Acad. ScL USA(in
press) 23 King, 1', C,, Sirdeskmukh, R. and Schlessinger, D. (1986) MicrobioL Ray. 50, 428451 24 Apidon, D, and Miczak, A, (1993) Bioessays15, 113-13) 25 Schmitt, M. E., Bennett, J. L, Dairaghi, D. J. and Clayton,D. A. (1993) FASEB J. 7, 208-213 26 Mordssey, J. P. and Tollervey, D. (1993) MoL Cell. Biol. 13, 2469-2477 27 Beltreme, M., Henry, Y. and ToUervey,D. (1994) Nucleic tCids Res. 22, 5139-5147
The three.dimensional structural information available for macromolecules (proteins, nucleic acids and sugars) is expanding rapidly, and is becoming increasingly important to molecular biologists, cell biologists and biochemists as they develop molecular hypotheses for biological processes, it is not easy for the majority of scientists to access these structures because of the large quantity of information and a general lack of familiarity with on-llne computer database systems. Therefore, bench scientists and students, most of whom do not have easy access to the sophisticated software required to visuallze molecular structures, frequently remain unaware that such data even
28 Musters, W. et al. (1990) EMBOJ. 9, 3989-3996 29 Tycowski, K. T., Shu, M.D. and Steitz, J. A. Science (in press) 30 Veldman, G. M., Klootwijk, J., van Heerihuizen, H. and Planta, R. J. (1981) Nucleic Acids Res. 19, 4847-4862 31 Peculis, B. A. and Steitz, J. A. (1993) Cell 73, 1233-1245 32 Peculis, B. A. and Steitz, J. A. (1994) Genes Dev. 8, 2241-2255 33 Melekhovets, t'. F., Good., L., Elala, S. A. and Nazar, R. N. 0994) J. MoL BioL 239, !70-180
34 Garrett, R. G. J., Larsen, N., Kjems, J. and Mankin, A. S. (1991) Trends Biochem. Sd. 116, 22-26 35 Newman, A. (1994) Curt. Opin. Cell Biol. 6, 360-367 36 Agabian, N. (1990) Cell 61, 1157-1160 37 Vossbrinck, C. R. and Woese, C. R. (1986) Nature 320, 287-288 38 Spencer, D. F., Coliings, J. C., Schnare, M. N. and Gray, M. W. (1987) EMBOJ. 6, 1063--1071 39 Soliner-Webb, B. (1993) Cell 75, 403-405
The Swiss-3DImagecollection and PDB-Browseron the World-WideWeb
under TCP/IP), more recently through Gopher, and now via the World-Wide Web (WWW)~. These tools provided access to the individual entries of the PDB and allowed limited searches of the data bank. exist. Two new features have been added For a number of years, the Protein to the PDB and ExPASy WWW servers, Data Bank (PDB) archive of the which should enhance their accessiBrookhaven National Laboratory (BNI) bility and usability: first, a PD~Browser has been available on the Internet for has been introduced, which allows the access via FTP (file transfer protocol submission of complex search queries
and the downloading of selected entries via the WWW; and second, the Swiss-3Dimage collection, consisting of a large number of annotated images describing the PDB entries, has been added.
TheWWWPD~Browser The PDB-Browser ~. R. Stamp|, C. E. Felder and J. L. Sussman, unpublished) accessed by the WWW agows the user to search over various PDB record
TaMe !, How to download Swiss~3DIm~es end GiF viewers Connection
ExPASy
Brookhaven Data Bank
WWWa Uniform resource Iocator (URL)
tlttp:/ / expasy,hcuge.ch/pub/Graphics/
ftp://pdb.pdb.bnLgov/images/GIF ftp://pdb.lxIb.bnLgov/images/RGB
Connect using FrP (TCP!IP)b Login as: Password to use: Set ftle type to binary: Change directory for GIFformat (Mac and P C ) : Stlicon Graphics RGBf o r m a t : List the directory content: Download the selected files: Disconnect:
FTPexpasy.hcuge.ch or connect to expasy.hcuge.chb anonymous Youremall address binary
ftp pdb.pdb.bnl.gov or connect to pdb.pdb.bnl.govb anonymous Your email address binary
cd/pub/Graphics/IMAGES/GIF cd/pub/Graphics/IMAGES/SGI Is get file_name quit or disconnectb
cd/images/GIF cd/images/RGB Is get file_name
l)~mbsdl~ GIF dewi~ Get the prg~ram: Macintosh° Silicon Graphics PC-Windowsb.c
Connection, Iogin and set binary file type as described above get/pub/Graphics/mac/quick~[if.hqx get/pub/Graphics/SGI/xv3 get/pub/Graphics/WIndows/Iview31.zip get/pub/Graphics/Windows/pkunzip.exe
aThe Swise-3Olmagesare also referenced on the stop WWW server (httP:l/www.bio.cam.ac.uk/scop). bTheMacintosh program Fetch and the PC-Windowsprogram WinFtp offer user-friendly interfaces for file transfer under TCP/IP. Fetch will automatically generate an executable from the quick-I[if.hqxfile. qvlew31.zip must be uncompmssed under DOS using 'pkunzip'.
82
© 1995,Elsevier Science Ltd 0968-0004/95/$09.50
{. P. L ~ s e y and D. Toileweyare at the EuropeanMolecularBiologyLaboratory (EMBL),Postfach1.02209,D-6g012 Heidel~rg, Germany.
Birth of the snoRNPs:the evolution of RNaseMRP the eukaryotic pre-r A-Foc ing system John P. Morrissey and David Tollervey The ribonucleoprotein particle RNase MRP is required for the processing of yeast pre-ribosomal RNA (pre-rRNA). A structurally related particle, RNase P, is univers~;tliyrequired for processing of pre-tRNA, but in bacteria and archaea also cleaves a site in the pre-rRNA. This suggests that RNase MRP may have arisen in eukaryotes as a form of RNase P specialized for pre-rRNA processing. Other eukaryotic small nucleolar RNAs may have arisen as trans-acting factors that functionally replace cis-acting prerRNA interactions in bacteria and archaea. The RNA component of RNase MRP (MRP RNA or 7-2 RNA) is encoded by the NMEI gene in yeast, which is a single-copy gene essential for cell viabilitys. It transpires that NMEI is allellc with RRP2 (Ref. 9), temperaturesensitive (is) alleles of which had previously been isolated independently by two groupsl°.~L Strains carrying the ts rrp2-1 allele or strains genetically depleted of the RNA component of RNase MRP~ are Impaired in the synthesis o{ 5.8 S(S) rRNA but can synthesize 5.8S(L) rRNA normally. The same phenotype is seen in a mutant carrying a deletion of the A3 cleavage site7. A protein from S. cerevisiae, POP1, which is a common component of RNase MRP and RNase P, has been characterized and its gene has been cloned. Cleavage at the A3 site is lost in strains carrying either a ts popl-I allele or the rrp2.1 allele~. Taken together, these data indicate that RNase MRP cleaves the pre-rRNAin ITSI at site A3.
protein is required for fullactivityin vitro, and is essential in vivo (for a review of the role of the bacterialprotein, see Ref. 17").The only nuclear RNase P that has been purified to homogeneity is that of the fission yeast, Schizosaccharomyces pombe, and this particle also contains one protein, although it has a considerably larger mass than Its bacterial counterpart is. Although there is very little sequence conservation between the RNA components of RNase P from different organisms, the tertiary structure, particularly the so-called 'cage domain' formed by long-range base-pairing interactions, is conserved ~9. The conservation of an antigenic epitope between bacterial and human RNase P proteins has also been reported 2°, emphasizing the ancient nature of RNase P. The activity of RNase P in pre-transfer RNA (pre-tRNA) processing appears to be universally conserved, but in bacteria RNase P also cleaves the precursors of 4.5S RNA2~ and of 10S RNA22. Biochemical studies on the RNA Mm P and the holoenzyme (reviewed in Ref. RNase P is present in all three phylo- 17) have led to the proposal that ancesgenetic domains (i.e. in bacteria, tral RNase P consisted entirely of RNA archaea and eukaryotes), as well as and cleaved tRNA-like molecules, and in mitochondria, demonstrating its that later in evolution the enzyme ancient evolutionary origin (for reviews acquired a protein component, and see Refs 14-16). Bacterial RNase P has with it the ability to recognize other been best studied, and consists of a substrates, such as 4.5 S RNA. A recent short RNA molecule and one protein. study by Liu and Altmanxl supports this The RNA molecule alone has some model. These workers randomized enzymatic activity in vitro, but the nucleotides of an RNase P substrate, © 1995,ElsevierScienceLtd 0968-0004/95/$09.50
REVIEWS
TgBS 20 = FEBRUARY 1995
and then selected substrates that were cleaved efficiently, either by the RNA
A0 A1
A2
35S
moiety alone, or by the holoenzy~e
consisting of the RNA and the protein. It was found that the RNA enzyme selected substrates that structurally resembled tP~NA, whereas the holoenzyme selected substrates resembling both tRNA and 4.5 S rRNA. Bacteria, like eukaryotes, general|y transcribe their rRNA as a single precursor, which is subsequently processed to yield the mature rRNA molecules. There are, however, significant differences between bacterial and eukaryotic pre-rRNA transcripts (Fig. 2). One difference is that in most eukaryotes the large-subunit rRNA is split into two molecules, 5.8 S rI~NAand 25 S rRNA, by the insertion of a second internal transcribed spacer (ITS2), whereas bacteria have a single 23S rRNA molecule. Despite this, the rRNAs are homologous and there is a high degree of conservation between 5.8 S rRNA and the 5'-end of 23 S rRNA. A further difference in bacterial pre-rRNA is the presence of a tRNA molecule in the spacer between the small (16 S) and large (23 S) subunit rRNAs. The prerRNA-processing pathway in bacteria is well characterized 23,24. The spacers immediately flanking the 16 S and 23 S rRNAs are capable of forming long helices, causing the rRNA molecules to be looped out, and bringing the ends of the mature rRNA regions into close proximity. An endoribonuclease, RNase 111, recognizes an undefined structural feature of these helices and endonucleolytically cleaves both strands, generating 17 S and 23 S pre-rRNA molecules that have spacer-sequence extensions at the 5'- and 3'-ends. These extensions are subsequently removed by exo- and endonucleolytic enzymes. As with all tRNAs, the tRNA in the spacer between the 16 S rRNA and the 23 S rRNA is processed at the 5'-end by RNase P. Mutants impaired in RNase Ill function are viable because RNase P cleavage of the pre-rRNA can still occur; the 5'-product of this cleavage is an efficient substrate for the 16 S rRNA maturation enzymes, and the 23S rRNA, although extended at the 5'- and 3'ends, can still be assembled into functional ribosomes 23. Since the majority of the tRNA genes are not located in rDNA operons, the conservation of a tRNA molecule in the rRNA spacer of most bacteria and archaea suggests a functional significance. As there is no conservation in the species of tRNA that is
°3 1
U14 snR30 20S
18S
CIeavage A0, A1 A2 A3
27SA
/
q
A3 81(S)
27SA
[32
B1(L)
27SA
B2
/
RNase MRP
C~eavage A3 27SA'
.
XRNI RAT1
1
E×onuclease
A3 --, B1 (S)
27SB(S)
J 5.8S(8)
25S
27SB(L)
ITS2 processing 5.88(L)
25S
Figure 1 Pre-rRNA processing in S. cerevisiae. For clarity the pre-rRNA processing pathway has been summarized; a more detailed figure can be found in Ref. 13. The first steps in pre-rRNA processing involve cleavage of 35 S pre.rRNAat the AO and A1 sites in the 5'-external transcribed spacer (5'-ETS) and at the A2 site in the first internal transcribed spacer (ffS1), generating the 20 S and 27 SA pre-rRNAs. Tile small nucleolar RNAs (snoRNAs} U3, U14 and snR30, and the associated small nucleolar ribonucleoprotein particle (snoRNP) proteins NOP1, SOFI and GAR1 are required for these cleavage reactionst,2'2°. The 20 S pre-rRNA is transported to the cytoplasm where it is matured to 18 S rRNA, whereas the 27 SA pre-rRNA remains in the nucleolus where it is further processed by one of two alternative pathways. Approximately 90% of the 27 SA pre-rRNA molecules are cleaved at the A3 site, 77 nucleotides upstream of the BI(S) site. Following this cleavage, the XRN1 and RAT1 exonucleases rapidly digest to the BI(S) site 7. The remaining 10% of 27 SA pre.rRNA molecules are processed at the BI(L) site by an undetermined mechanism. In S. cerevisiae there are two predominant forms of 5.8 S rRNA, differing by seven nucleotides at the 5"-end; the BI(S) and BI(L) sites are the 5'-ends of the short, 5.8 S(S) rRNAand the long, 5.8 S(L) rRNA, respectively. Contemporaneously with the processing reactions forming the BI(S) and BI(L) ends, the B2 site in the 3'-external transcribed spacer (3;ETS) is cleaved. The mechanism and precise relative timing of this cleavage is not known. The 27 SB(S) and 27 SB(L) pre.rRNAs are processed to mature 5.8 S(S) and 5.8 S(L) rRNA, respectively, and to 25 S rRNA, by removal of the second internal transcribed spacer (ITS2) in a set of reactions not known to require snoRNPs.
present in the pre-rRNAs, it is likely that the presence of a tRNA per se, rather than any particular tRNA, is of functional importance. We propose that a tRNA has been maintained in this location so as to preserve the RNase P cleavage site in the rRNA spacer. RNase P cleavage of the pre-rRNA could then serve as an alternative processing pathway for the synthesis of 16 S rRNA. It is not unusual to find redundant mechanisms
for performing a reaction crucial to a cell's survival; having alternative mechanisms for carrying out a reaction may be the most effective way of ensuring that the reaction proceeds with high efficiency.
Model for the evolution of RNase MRP RHase MRP, which has been found only in eukaryotes, is structurally related to RHase P. Phylogenetic comparison
19
TIBS 2 0 -
RNase P
1 E. colt
~
i
16S
't I lt
I
S. cersv/s/se 5'-ETS ~ 18S
-
:,
'i tRNA '
I !
i I
I
I
23S
.
", 5S
t l
ITS1 ITS2 ~ 5,8S
%
*
3-E'rS_ 25S
RNase MRP Rgure 2
Comparison of E. colt and S. cerevisiae precRNAs. Although the basic arrangement of the rRNA is generally conserved between bacteria and eukaryotes, there are three major differences. Rrst, in most eukaryotes the large-subunit rRNA is split by the insertion of a spacer, ITS2. Second, the 5 S rRNA is part of the bacterial pre-rRNAbut is synthesized independently in eukaryotes. Third, bacteria have a tRNA molecule in the spacer between the 16 S and 23 S rRNAs. The species of tRNA varies, but its presence is conserved. Some bacterial pre-rRNAs have further tRNAs downstream of the 5 S rRNA, but the presence of these tRNAs is not conserved. The positions of the RNase P and RNase MRP cleavage sites are indicated.
shows that the MRP RNA can be folded into a secondary structure that is similar to the structure of the RNase P RNA~. This predicted structure includes the highly conserved cage domain, which has been proposed to be Involved In catalysis~9. Human autoImmune antibodies, the Th/To sera, coprecipitate both MRP RNA and RNase P RNA, suggesting that the two particles contain a common proteln epltope. This was confirmed by the identlflcatlon of the POPI protein in S. cerevistae, This protein Is a component of both RNase P and RNase MRP: Immunopreclpltatlon of epltope-tagged POP1 co-preclpltates RNase P RNA and MRP RNA, and a strain carrying a mutant popl.1 allele Is Impaired In the function of both RNase P and RNase MRPm. From the structural similarities it is clear that RNase P and RNase MRP are evolutlonarlly related. Given that RNase P is ubiquitous and of ancient origin, whereas RNase MRP has been found only In eukaryotes, it seems likely that RNase MRP was deriv~ from RNase P. We envisage that, shortly after the divergence of the eukaryotes and the archaea, the RNA component of RNase MRP arose by duplication of an ancestral R N a s e P RNAgene. Alignment of the bacterial and e~aryotlc pre-rRNAs shows that the bacterial RNase P cleavage site and the eukaryotlc RNase MRP cleavage site are in similar positions (Fig. 2). Since a tRNA molecule is found in the pre-rRNA spacer of both bacteria and archaea, it is likely that this represents the ancestral state. Duplication of the RNase P RNA gene in an early eukaryote would
have allowed co-evolution of the enzyme and pre-rRNA cleavage site. We propose that this gave rise to RNase MRP and site A3 in modern eukaryotes. In both S. cerevisiae and E. colt, the RNase MRP and the RNase P cleavages represent alternative processing pathways for the pre-rRNA. Conservation of alternative processing pathways throughout evolution supports the idea that this is a means of ensuring high processing efficiency. This model explains the origin of RNase MRP and tile reason for its cleavage of the pre~rRNA, but does not exclude the possibility that RNase MRP has acquired other functions In the cell, including a role In mitochonddal DNA replication. OdgJns of eukaryotic pre.~NA procesdng
In S. cerevisiae, the snoRNPs U3, U14 and snR30 are required both for process!n~ at site A1, the 3'-end of the 5'external transcribed spacer (5'-ETS), and for processing at site A2 in ITS1 (Fig. 1)i,2,29. Genetic depletion of the RNA or protein components of any of these particles abolishes cleavage at both A1 and A2, preventing synthesis of the 18 S rRNA but allowing synthesis of both iarge-subunit rRNAs, 5.8S rRNA and 25 S rigA. At least one of the snoRNPs, U3, binds to a sequence in the 5'-ETS27, and it is postulated that the snoRNPs also interact with sequences in ITS1, forming a complex in which the 5'-ETS and ITS1 are in close proxJ~.ity with the looped out 18S rRNA. Consistent with this, deletions in the 5'-ETS also inhibit cleavage at both AI and A2
FEBRUARY
1995
(Refs 27, 28). Similarly, depletion of the U22 snoRNA in Xenopus oocytes inhibits processing in both the 5'-ETS and ITS1 (Re[. 29). In S. cerevisiae, coupling has also been proposed between processing at site BI, the 3'-end of ITS1, and at site B2, the 5%rid of the 3'-ETS (Fig. 1)3°. There are also data from other eukaryotes that suggest a connection between processing in the 3'-region of ITS1 and in the 3'-ETS. Depletion of the U8 snoRNA in Xenopus oocytes impairs processing both at the 3'-end of ITS1 and at site T1 in the 3'-ETSm. The consequence of this inhibition is that the 5.8S and 28S rRNAs are not synthesized, whereas synthesis of the 18 S rRNA continues. In oocytes to which partial processing activity has been restored by injection of U8, pre-rRNAs cleaved at only one of these sites are not detected, leading to the suggestion that processing in ITS1 and in the 3'-ETS is coupled and that this coupling requires the U8 snoRNP 2. In S. pombe, mutations in the 3"ETS also impair processing in ITS1 and abolish synthesis of the large-subunit rRNAs33, supporting the idea that processing in ITS1 and the 3'-ETS are interdependent events. A general model for eukaryotic prerRNA processing can be postulated, in which snoRNP complexes bring the ends of the rRNA molecules into close proximity, and in this way substitute for the terminal base-pairing interactions found in bacteria and archaea (Fig. 3). This would enable the eukaryotlc cell to co-ordinate processing at the 5" and 3" ends of an rRNA molecule. Given that the archaeal and bacterial systems are so similar (for a review of archaeal rRNA operons see Ref. 34), it is likely that the ancestral pre-rRNA relied on spacer-sequence base-pairing for its processing system. In yeast, there are clear structural and functional differences between MRP RNA and the 14 other snoRNAs characterized to date. With the exception of MRP RNA, all of the snoRNAs are precipitable by antibodies against the NOP1 protein (the yeast homologue of fibrillarin; Ref. 1; D. Tollervey, unpublished). Functionally, RNase MRP is an endoribonuclease in vitro, whereas no other snoRNP has been shown to have nuclease activity, although it is possible that multiple snoRNPs assemble together with the pre-rRNA to form an RNAbased cleavage complex. It seems likely that these differences are related to a fundamental difference between the
REVIEWS
TIBS 20 = FEBRUARY 1 9 9 5 evolutionary origin of RNase MRP and the other snoP~Ps. There is a precedent for the function of cis-acting RNA
sequences being usurped by trans-acting small nuclear ribonucleoprotein particle (snRNP) factors. Group |l introns, which are found in mitochondria and chloroplasts, are autocatal~ic and are responsible for their own removal from the pre-mPdqA and the coupled ligation of the resulting exons. Nuclear pre~messenger Pd~IA ~remRNA) splicing proceeds by a similar pathway but requires the UI, U2, U4/U6 and U5 snRNPs. It is believed that the
Euka~otes
SHORN
Bacteria
oRNPs
RNas
ITS1
ase roll tRNA
Figure 3 Model comparing eukaryotic and bacterial pre-rRNA processing. The spacers flanking the bacterial rRNA molecules form extensive helices, which contain the recognition sites for RNase ill. We propose that in eukaryotes small nucieolar ribonucleopretein particle (snoRNP) complexes have replaced intramolecular base-pairing in bringing the 5" and 3;flanking sequences of the mature rRNAstogether. As discussed in the text, there is evidence from S. cerevisiaethat the U3, Ut4 and snR3O small nucleolar ribonucleoproteins (snoRNPs) may be part of the complex looping out the small subunit rRNA and that, i~ Xenopus,U8 may be part of the complex looping out the large subunit rRNAs.
group I1 self-~plicing introns represent the ancestral state and that the transacting snP,NPs evolved from, and took over the functions of, cis-acting intron sequences (for a recent review of splicing see Ref. 35). An intermediary state may be found in trypanosomatids, where U1 function is provided in cis, but some of the other snRNP functions are provided in trans36. The model that the eukaryotic ITSI is related to the bacterial internal spacer, and that the 5'- and 3'-ETS are related to the bacterial external spacers, implies that these spacers are more ancient than ITS2. Consistent with this is the observation that a deeply branched eukaryote, the microsporidium Vairimorpha necatrix, does not have an ITS2 and has a single 23S-like large-subunlt rRNA36. The suggestion that the ITS2 spacer became inserted into the iarge-subunit rRNA after the divergence of eukaryotes is supported by the case of trypanosomatids, where multiple spacers appear to have been inserted during evolution; for example, the large-subunit rRNA of Crithidia fasiculata is interrupted by six spacers 38. if ITS2 is unrelated to the other eukaryotic spacers it might be expected that it would be processed by an independent mechanism. This would provide an explanation for why so many yeast mutants, including the snoRNP mutants, affect processing in the 5'-ETS and ITS1, but do not affect ITS2 processing.
Concludingremarks We have presented two models that offer explanations for a number of recent observations on the function of eukaryotic snoRNP particles. In both,
we envisage that, following divergence of the eukaryotes from the archaea and bacteria, elaboration of existing, simpler processing systems gave rise to the pre-rRNA processing components seen in modern eukaryotes. These models postulate quite different evolutionary origins for different eukaryotic snoPd~Ps. Pd~lase MP& is envisaged as being derived from the ancient ribonuclease Pd~lase P. By contrast, we propose that the other snoRNPs involved in processing arose as transacting factors that functionally replace the interactions between the spacer sequences in bacterial and archaeal pre-rRNAs. Whether the snoRNAs are directly derived from the spacers or represent an independent solution to the problem of bringing the ends of the mature rILNAs together, is unclear at present. It is also still unclear whether the snoPd~IPs function in processing by presenting the pre-rPd~lA in the correct conformation for ~roteinaceous) processing nucleases, or whether multiple snoP.d~Ps assemble with the pre-rPd~lA to form an PallIAbased processing complex. Finally, it is notable that a further group of snoRNAs, which are encoded in the introns o[ other genes and are characterized by high sequence complementarity to mature rPd~lAsequences, has been described39; it is possible that these snoPd~IAs have a different evolutionary origin from the snoPd~lAsshown to be involved in pre-rP,NA processing. More detailed functional characterization may shed further light on the origins of the various groups of snoRNPs.
Acknowl~gemenL~ We would like to thank E. [rare and many colleagues at EMBL for their helpful discussions and comments on
the manuscript.
Re,fences 1 Rlipowicz, W. and Kiss, T. (1993) Mol. Biol. Rep. 18, 149-156 2 Fournier, M. J. and Maxwell, E. S. (1993) Trends Biochem. Sci. 18, 131-135 3 Clayton, D. A. (1994) Proc. Nati Acad. Sci. USA 91, 4615-4617 4 Chang, D. D. and Clayton, D. A. (1987) EMBOJ. 6, 409-417 5 Stohl, L. L. and Clayton, D. A. (1992) Mol. Cell. Biol. 12, 2561-2569 6 Kiss, T., Marshallsay, C. and Rlipowicz, W. (1992) EMBOJ. 11, 3737-3746 7 Henry, Y. et al. (1994) EMBOJ. 13, 2452-2463 8 Schmitt, IVl. E. and Clayton, D. A. (1992) Genes Dev. 6, 1975-1985 9 Chu, S., Archer, R. H., Zengel, J. M. and Lindaill, L. (1994) Proc. Natl Acad. ScL USA 91, 659-663 10 Shuai, K. and Warner, J. W. (1991) Nucleic Acids Res. 19, 5059-5064 11 Lindahl, L., Archer, R. H. and Zengal, J. M. (1992) Nucleic Acids Res. 20, 295-301 12 Schmitt, M. E. and Clayton, D. A. (1993) Mol. Cell. BioL 13, 7935-794.1 13 Lygerou, Z. et aL (1994) Genes Dev. 8, 1423-1433 14 Airman, S. (1989) Adv. EnzymoL62, 1-36 15 Darr, S. C., Brown, J. W. and Pace, N. R. (1992) Trends Biochem. ScL 17, 178-182 16 Altman, S., Kirsebom, L. and Talbot, S. (1993) FASEBJ. 7, 7-14 17 Gopolan, V., ralbot, ~. J. and Altman, S. in RNA--Protein Interactions (Nagai, K. and Mattaj, I. W., eds), Oxford University Press (in press) 18 Zimmerly, S., Drainas, D., Sylvers, L. A. and SBII, D. (1993) Eur. J. Biochem. 217,501-507 19 Forster, A. C. and Altman, S. (1990) Cell 62, 407-409 20 Mamula, M. J., Baer, M., Craft, J. and Altman, S. (1989) Prec. Natl Acad. Sci. USA 86, 8717-8721 21 Liu, F. and Altman, S. (1994) Cell 77, 1093-1100
81
TiBS 2 0 - FEBRUARY1995 22 Komine, Y, et at, Proc. Natl Acad. ScL USA(in
press) 23 King, 1', C,, Sirdeskmukh, R. and Schlessinger, D. (1986) MicrobioL Ray. 50, 428451 24 Apidon, D, and Miczak, A, (1993) Bioessays15, 113-13) 25 Schmitt, M. E., Bennett, J. L, Dairaghi, D. J. and Clayton,D. A. (1993) FASEB J. 7, 208-213 26 Mordssey, J. P. and Tollervey, D. (1993) MoL Cell. Biol. 13, 2469-2477 27 Beltreme, M., Henry, Y. and ToUervey,D. (1994) Nucleic tCids Res. 22, 5139-5147
The three.dimensional structural information available for macromolecules (proteins, nucleic acids and sugars) is expanding rapidly, and is becoming increasingly important to molecular biologists, cell biologists and biochemists as they develop molecular hypotheses for biological processes, it is not easy for the majority of scientists to access these structures because of the large quantity of information and a general lack of familiarity with on-llne computer database systems. Therefore, bench scientists and students, most of whom do not have easy access to the sophisticated software required to visuallze molecular structures, frequently remain unaware that such data even
28 Musters, W. et al. (1990) EMBOJ. 9, 3989-3996 29 Tycowski, K. T., Shu, M.D. and Steitz, J. A. Science (in press) 30 Veldman, G. M., Klootwijk, J., van Heerihuizen, H. and Planta, R. J. (1981) Nucleic Acids Res. 19, 4847-4862 31 Peculis, B. A. and Steitz, J. A. (1993) Cell 73, 1233-1245 32 Peculis, B. A. and Steitz, J. A. (1994) Genes Dev. 8, 2241-2255 33 Melekhovets, t'. F., Good., L., Elala, S. A. and Nazar, R. N. 0994) J. MoL BioL 239, !70-180
34 Garrett, R. G. J., Larsen, N., Kjems, J. and Mankin, A. S. (1991) Trends Biochem. Sd. 116, 22-26 35 Newman, A. (1994) Curt. Opin. Cell Biol. 6, 360-367 36 Agabian, N. (1990) Cell 61, 1157-1160 37 Vossbrinck, C. R. and Woese, C. R. (1986) Nature 320, 287-288 38 Spencer, D. F., Coliings, J. C., Schnare, M. N. and Gray, M. W. (1987) EMBOJ. 6, 1063--1071 39 Soliner-Webb, B. (1993) Cell 75, 403-405
The Swiss-3DImagecollection and PDB-Browseron the World-WideWeb
under TCP/IP), more recently through Gopher, and now via the World-Wide Web (WWW)~. These tools provided access to the individual entries of the PDB and allowed limited searches of the data bank. exist. Two new features have been added For a number of years, the Protein to the PDB and ExPASy WWW servers, Data Bank (PDB) archive of the which should enhance their accessiBrookhaven National Laboratory (BNI) bility and usability: first, a PD~Browser has been available on the Internet for has been introduced, which allows the access via FTP (file transfer protocol submission of complex search queries
and the downloading of selected entries via the WWW; and second, the Swiss-3Dimage collection, consisting of a large number of annotated images describing the PDB entries, has been added.
TheWWWPD~Browser The PDB-Browser ~. R. Stamp|, C. E. Felder and J. L. Sussman, unpublished) accessed by the WWW agows the user to search over various PDB record
TaMe !, How to download Swiss~3DIm~es end GiF viewers Connection
ExPASy
Brookhaven Data Bank
WWWa Uniform resource Iocator (URL)
tlttp:/ / expasy,hcuge.ch/pub/Graphics/
ftp://pdb.pdb.bnLgov/images/GIF ftp://pdb.lxIb.bnLgov/images/RGB
Connect using FrP (TCP!IP)b Login as: Password to use: Set ftle type to binary: Change directory for GIFformat (Mac and P C ) : Stlicon Graphics RGBf o r m a t : List the directory content: Download the selected files: Disconnect:
FTPexpasy.hcuge.ch or connect to expasy.hcuge.chb anonymous Youremall address binary
ftp pdb.pdb.bnl.gov or connect to pdb.pdb.bnl.govb anonymous Your email address binary
cd/pub/Graphics/IMAGES/GIF cd/pub/Graphics/IMAGES/SGI Is get file_name quit or disconnectb
cd/images/GIF cd/images/RGB Is get file_name
l)~mbsdl~ GIF dewi~ Get the prg~ram: Macintosh° Silicon Graphics PC-Windowsb.c
Connection, Iogin and set binary file type as described above get/pub/Graphics/mac/quick~[if.hqx get/pub/Graphics/SGI/xv3 get/pub/Graphics/WIndows/Iview31.zip get/pub/Graphics/Windows/pkunzip.exe
aThe Swise-3Olmagesare also referenced on the stop WWW server (httP:l/www.bio.cam.ac.uk/scop). bTheMacintosh program Fetch and the PC-Windowsprogram WinFtp offer user-friendly interfaces for file transfer under TCP/IP. Fetch will automatically generate an executable from the quick-I[if.hqxfile. qvlew31.zip must be uncompmssed under DOS using 'pkunzip'.
82
© 1995,Elsevier Science Ltd 0968-0004/95/$09.50