86

Reviews TIBS 1 3 - March 1988

[i !

Heterogeneous nuclear ribonucleoprotein particles and the

pathway of mRPIA formation Gideon Dreyfuss, Maurice S. Swanson and Serafm PifioI-Roma

Heterogeneous nuclear ribonucleoprotein (hnRNP) particles, the amaures that package hnRNA, are one of the major constintents o f the nucleus. Recent work has led to the immunopurification o f hnRNP particles and the identification o f their proteins, and demonstrated a role for hnRNP proteins in mRNA splicing. The molecular doning and sequencing of cDNAs for RNP proteins made possible the discovery of a conserved RNA-

binding domain and a RNP consensus sequence.

In eukaryotic cells, messenger RNAs (mRNAs), the functional translatable intermediates of gene expression, are generally formed by extensive process- ing of primary gene transcripts. Histori- ca,y, the primary gene transcripts were discovered in higher eukaryotes and termed heterogeneous nuclear RNAs (hnRNAs). hnRNAs can be distin- guished from other RNAs on the basis of their size, their subcellular compart- mentation, and the characteristics (e.g. antibiotic sensitivity) of the RNA polymerase that transcribes them. The terms hnRNA and pre-mRNA are often used interchangeably, although it is possible that only a subpopulation of hnRNAs are actually mRNA precur- sors. Typically, pre-mRNAs contain 5'- cap structures (m7Gpp), polyadenylated tails, and the majority of them contain intervening sequences that are later spliced out. Little is known about the ensuing events except that the RNA is translocated through nuclear pores and that spliced mRNAs accumulate in the cytoplasm. One of the ultimate goals of molecular and cell bi~logy is to under- stand all of these processes in terms of both molecular detail and cellular topolo~.

Work from numerous laboratories over the past 20 years has revealed that hnRNAs in cells do not normally occur as naked polynucleotides but are found in complexes with specific proteins (re- viewed in Refs 1 and 2), which are

G. Dreyfuss, M. S. Swanson and S. Pihol-Roma are ta the Department of Biochemistry, Molecular Biology and Cell Biology, Northwestern University, Evan- ston, IL 60208, USA.

1988, Elsevier Publications Cambridge 0376- 5067188/$02,00

termed hnRNP complexes or hnRNP particles. Interest in hnRNP complexes stems from the fact that the processing of pre-mRNA probably takes, place on these ribonucleoprotein complexes. Therefore, to understand how the post- transcriptional portion of the pathway of expression of genetic information oper- ates in the cell, we need to learn more about the ribonucleoprotein complexes.

The proteins of hnRNP particles hnRNP particles are one of the major

components of the nucleus. The proteins of the hnRNP particles are as abundant as histones in the nucleus of growing cells and comprise --80% of the mass of hnRNP particles. The association of hnRNA with hnRNP proteins begins as the hnRNA is still a nascent transcript and persists throughout its nuclear resi- dency. Biochemical analyses of nuclear fractions containing hnRNA or hnRNA fragments obtained by velocity sedimen- tation revealed a group of proteins (of about 30-43 kDa) as consistent compo- nents of hnRNP complexes3-7. These proteins are detected by SDS-poly- acrylamide gel electrophoresis (SDS- PAGE) as six bands and classified into three groups of proteins: A, B and C (Ref. 3). Thus it was thought that hnRNPs were composed predominantly of these six proteins. Although numer- ous additional proteins of higher molecu- lar mass have been frequently observed in hnRNP fractions 6-7 there has been no consensus about their authenticity. This is due to the intrinsic limitations of the velocity sedimentation method; it does not afford complete separation of hnRNP particles from other nuclear

components, and non-specific associa- tion of proteins with the hnRNA after cell fractionation can not be ruled out.

A more stringent definition of genuine hnRNP proteins has been provided by photochemical (UV light-induced) co- valent cross-linking of proteins to RNA in vivo, which allows identification of proteins that are bound to the RNA in the cell 8-1o. This method relies on the fact that UV light photo-activates RNA and converts it to an extremely reactive, short-lived molecule which reacts vir- tually indiscriminately with other mol- ecules (including proteins) that are in direct contact with it. In effect this is photo-affinity labeling of the RNA-bind- ing proteins in vivo. The cross-linked hnRNA-protein and mRNA-protein complexes can then be isolated from the nuclear and cytoplasmic fractions, respectively, after boiling in SDS and mercaptoethanol, by affinity chromato- graphy on the oligo(dT)-cellulose, which selects them through binding to the poly(A) tail. These protein-denaturing

(1)

UV (254nm)

sos [ ~ME ~ 90*

(2) oligo(dT)-cellulose

(el polylAI mRNA-protelnlmRNPI (N) poly (A) hnRNA-protein (hnRNP)

131

~ - myelomo (SP2/O)

fusion I

hybridomo

MONOCLONAL ANTIBOOIE$ Fig. 1. The experimental strategy used for the iden- tification and production of monoclonal antibodies to RNA-binding proteins in vivo t0. Step (1) protein- RNA cross-linking; step (2) RNA selection; and step (3) immunization. C, cytoplasmic fraction; N, nuclear fraction.

TIBS 13-March 1988 87

cond;tions ensure that only proteins which are covalently linked to the RNA are purified with it, thus eliminating non- specific associations of proteins with the RNA. The cross-linked proteins can be released from the complexes by diges- tion with RNases and analysed by SDS- PAGE. These studies showed numerous proteins, including the proteins that cor- respond in SDS--PAGE mobility to A, B and C proteins 8-10. This provided further evidence for the authenticity of the A, B and C proteins and also indicated that the hnRNP protein profile is much more complex and includes proteins of higher molecular mass. The limitations of the photo-chemical cross-linking are that the cross-linked proteins are no longer use- ful for biochemical studies and that pro- teins which bind RNA, but do not cross- link to it efficiently, could be lost. Nevertheless, because of the specificity of this method, we have used UV-cross- linked hnRNPs as a source of genuine hnRNP proteins for immunization of mice to produce monoclonal antibodies to these proteins 1°,11 in order to obtain highly specific probes for the study of individual hnRNP proteins and of intact hnRNP particles.

This strategy (Fig. 1) has now been used successfully for several different cell types including human 1°,11 and yeast 12 and for adenovirus-infected cells 13. It is a general approach for the identification and production of monoclonal anti- bodies to nucleic acid binding proteins/n vivo. One of the most important applica- tions of the monoclonal antibodies to hnRNP proteins was the immunopurifi- cation of hnRNP particles 14,~s. The method involves a brief incubation of antibodies bound to agarose beads with nuclear lysate (nudeoplasm) and rapid washing of the beads. The procedure is efficient, specific, rapid and mild. It yields pure hnRNP particles and does not require either nuclease degradation or labeling of the hnRNA.

Detailed analysis of the protein com- position of immunopurified hnRNP particles from human HeLa cells by two- dimensional gel electrophoresis reveals an assortment of at least 24 polypeptides in the molecular mass range of 34-120 kDa (Fig. 2) is. Some of the complexity seen in this two-dimensional gel may be due to post-translational modification of a smaller number of proteins 6, but most of the proteins which have an assigned letter are distinct (by immunological and other criteria). The previously described, abundant 30-40 kDa pro- teins, A, B and C, are a subset of these polypeptides.

H + OH-

U /

tt6K ' ~ T - /

95K-- R . . \

L M - ~ Q / P 68K-- K~e ~ \ - "

-} J J" ,,H . e " I N q ) e l

4 K- -:o2 - ' I " °' J'J' Ct'O :.I-E .

I A2

3OK-- At Fig. 2. Two-dimensional gel electrophoresis of hnRNP particles immunopurified with a monoclonal anti- body, 4F4, to the C proteins. The hnRNP particles were immunopurified from the nucleoplasm of [3SS]methionine-labeUed HeLa cells u.15 . The proteins were separated by non-equilibrium pH gel electrophoresis in the first dimension and by SDS-PAGE in the second dimension and visualized by fluorography.

The organbation of proteins in hnRNP particles

Little is known about the structure of hnRNP particles other than that the bulk of the proteins are organized into monoparticles - complexes that after limited cleavage of the ImRNA with nu- clease sediment at 30S-40S and appear by election microscopy as globular particles of about 200A, 1,2. The composition of all monoparticles is presumed to be the same but more direct evidence for this is needed. Most of the proteins shown in Fig. 2 remain together as a complex after digestion with nuclease that generates monoparticles (S. Pifiol-Roma and G. Dreyfuss, unpublished). The size of RNA recovered with monoparticles is in the range of 125 to 800 nucleotides s,4,14. The integrity of the monoparticles is dependent on RNA in that, when the RNA is further degraded, monoparticles dissociate. The prevailing model of the overall structure of hnRNP particles is that of a beads-on-a-string motif with monoparticles forming the beads and the hnRNA itself the string. The simplest interpretation of the pattern of nuclease lability of the hnRNA in hnRNP parti- cles is that the inter-monoparticle (linker) hnRNA is hypersensitive because it is not covered with much pro- tein. Since the monoparticle-associated hnRNA is also RNase sensitive (al- though less than the linker), it is likely to occupy an exposed position. The 5' cap

and the 3' poly(A) tails, when present, are associated with other (non-monopar- ticle) specific: p~'oteins.

Two major questions about the struc- ture of hnRNP particles arise: (1) what is the arrangement of the proteins in monoparticles? and (2) where are monoparticles positioned on any particu- lar pre-mRNA? Related to these ques- tions are the issues of how hnRNP par- ticles assemble and how the packaging of the pre-mRNA into a specific hnRNP structure is related to the processing of pre-mRNA into mRNA. The clarifica- tion of the picture of the protein compo- sition of hnRNP monoparticles and the availability of antibody probes to many of these proteins should facilitate the investigation of the arrangement of pro- teins in monoparticles.

Arrangement of monoparticles? It is believed (but cannot be stated

with certainty) that the positioning of monoparticles on the pre-mRNA is spe- cific and unique for each pre-mRNA. This view is derived primarily from elec- tron micrographs which show protein particles (of similar dimensions to those of monoparticles isolated on sucrose gradients) positioned uniquely on specific transcripts 16. The inherent difficulty in interpreting these studies is that it cannot be ascertained that the observed parti- cles are indeed hnRNP monoparticles. In fact, monoparticles would not nor-

88 TIBS 13-March 1988

mall), be expected to survive the treat- ment involved in the preparation of the samples for such electron microscopy.

Nevertheless, if a specific arrange- ment of monoparticles exists, drastically different mechanisms could be en- visioned which would give rise to it. The hnRNP proteins could have RNA sequence-dependent binding speci- ficities, although it is widely held that they do not. On the other hand, specific monoparticle positioning could also be achieved either by a processive assembly of monoparticles or by the direction of

other factors which do recognize specific RNA sequences, such as small nuclear (sn)RNPs. Knowledge of the RNA- binding properties of the hnRNP pro- teins and of the assembly process of hnRNP particles is crucial for under- standing the structure of hnRNP particles and its relationship to mRNA formation.

The function of hnRNP proteins in mRNA biogenesis

An obvious function for hnRNP pro- teins is in the packaging of hnRNA. The need for proteins to compact hnRNAs in

30s monoportlcle (ribonucleosome) chromoUn F (5OO±lOOnt)

hnRNP proteins

RNA.__.. pol II 5' hnRNP

3 ' - POLY(~A) addition > snRNPs

1

spliceosorne 3 ,.~

IVS * snRNPs

1

~ ~ cop

I mRNA t ronsporL~ [ _ _ , ~ . p r o t e i n

PABP mRNP proteins

exchange NUCLEUS

CYTOPLASM

cap

Fig. 3. Schematic presentation of a generalized model of hnRNP particle structure and its involvement h~ pre-mRNA splicing. For simplicity the spliceosome is illustrated in this scheme as a combh~ation of the snRNP and hnRNP particles. Not all of the hnRNP proteins, which comprise the bulk of hnRNP particles as seen bz Fig. 2, are necessarily bl the spliceosonze. The double set of arrows for snRNP addition simply illustrate that snRNPs may be associated before, and/or after, addition of hnRNP proteins. E, exon; IVS, intervening sequence (huron); IVS* bttron in lariat form; snRNPs, small nuclear ribonucleoprotein parti- cles; PA BP, poly(A) binding protein.

the nucleus can be understood in terms similar to those for the need to package DNA in chromatin. The packaging must be done, however, in such a way that the pre-mRNA is folded into a structure that can be spliced. This includes the require- ment that pre-mRNA is also accessible for interaction with snRNPs. With the development of/n vitro cell-free systems which faithfully splice mRNA precursors and with the availability of specific anti- bodies to proteins of the hnRNP com- plexes, it has become possible to address the long-standing issue of the involve- ment of the hnRNP proteins and of the hnRNP complex in pre-mRNA splic- ingl7,18.

The effect on pre-mRNA splicing of several monoclonal antibodies to hnRNP proteins was investigated 17. It was found that a monoclonal antibody to the C proteins, two of the major proteins of hnRNP particles (see Fig. 2), inhibits the splicing/n v/tro of a mRNA precursor (pre-mRNA). It was also found that the splicing complex (spliceosome) contains C proteins and that the splicing complex can be immunoprecipitated with this antibody. Furthermore, immunodeple- tion of C proteins from the nuclear extract abolishes its capacity to splice RNA and to form spliceosomes. Thus, these experiments suggest that at least two hnRNP proteins, in addition to snRNPs (see Refs 19 and 20 for reviews), are important for mRNA splicing. A model depicting the overall structure of hnRNP particles and their involvement in mRNA splicing is presented in Fig. 3. Although the specific function of these proteins in the splicing reaction is not yet known, it appears that the antibodies to hnRNP proteins will be valuable tools for studying the splicing complex and the function of hnRNP proteins in mRNA biogenesis.

The transition from hnRNPs to mRNPs and nucleo-cytoplasmic transport

Cytoplasmic mRNAs are also associ- ated with proteins (mRNP proteins) although with much less protein than are h n ~ A s . The proteins are likely to be involved in modulating the translation and stability of the mRNA and in its cellular localization. In spite of much effort, the picture of the mRNP protein constituents and their arrangement on the mRNA is less well developed than that for the nuclear hnRNP counter- parts, but it is clear that the mRNP pro- teins are different from hnRNP proteins. All of the hnRNP proteins for which antibodies are available (so far) are con- fined to the nucleus as determined by

TIBS 13-March 1988 89

immunofluorescence microscopy. These include the A and B proteins 2~, the C proteins, the 120 kDa U protein~= and the L and M proteins (S. Pifiol-Roma and G. Dreyfuss, unpublished). With regard to the mRNP proteins, at least the mRNA 72 kDa poly(A) binding protein (PABP) is localized to the cytoplasm and is not cross-linked to nuclear RNAt2,2L No proteins have yet been found that are bound to both hnRNA in the nucleus and mRNA in the cytoplasm, but some hnRNP and mRNP proteins may be related as it has been suggested that nudear and cytoplasmic PABPs in yeast are derived from the same gene 23. It therefore appears that concomitant with the transport of mRNA from the nucleus to the cytoplasm there is a process of exchange of RNP proteins. To under- stand the exchange process fully, it is necessary to know what proteins are bound to the spliced pre-transport

nuclear transcript. Actual transport and RNP protein exchange are probably somehow coupled.

Structure of RNP proteins and the RNP consensus sequence

Once the sequence of more than one RNP protein became available it was possible to examine whether there is any common sequence motif to RNP pro- teins. The first two RNP proteins for which amino add sequence was obtained were the .yeast mRNA PABp~2,:3 and the mamraalian A1 hnRNP protein25.26. The amino-terminal portion of the PABP contains four similar domains of about 100 amino acids each. The A1 pro- tein contains two similar domains of about 100 amino acids each. Approxi- mately in the middle of these repeating domains is a segment of eight amino adds which is the most strikingly conserved sequence between these

domains, and it was termed the RNP consensus sequence~2, ~4 (RNP1 in Fig. 4). Examination of the sequences for 18 such domains in eight different RNP pro- teins sequenced up to September 1987 (including hnRNA, mRNA, snRNA and pre-rRNA binding proteins) is shown in Fig. 4 and further highlights the conservation of this sequenceE4. RNP1 is found also in three additional snRNP proteins recently sequenced (van Ven- rooij, pers. commun.). Figure 4 also depicts several additional characteristic features that emerge from examination of the 100 amino acid RNA-binding domains, in particular a second con- served element which is located about 30 amino acids amino terminal to RNP1 and designated RNP2. It is mostly aliphatic and aromatic in the character of amino adds but it is less strictly con- served than RNP1.

The conclusion that the 100 amino

PROTEIN DOMAIN PABP YEAST I ESQSVENSSASLWGDLEPSVSEAHLYD, FSP, GSVSS' | | ' l l l l l l f |TKTSLGYAYVNFINDHEAGRKA' EQLI~fl'TP I KGRLCR' I~S~OPSLR PABP HUMAN I NPSAPSYPI~SLYV~DLHPDVTEAMLYEKFSPAGPI LSI RVCRDM TRRS~.GYAYVNF~PADAERALDTIVlNFDVI

PABP HUMAN II DPSLRKSGVGNI KNLDKSl DNKALYNTFSAFGNI LSCKWCDENG..S~GYGFVHF]L='rQGAAERAI EKMNGMLLNDRKVFVGRFKSRKERE

PABP YEAST III QLEETKAHYTNLYVKNI NSETTDEQFQELFAKFGPI VSASLEKDADGK.LRGFGFVNYIEKHEDAVKAVEALNDSELNGEKLYVGRAQKKNERM

PABP HUMAN IV

A1 A1 DROSOPHILA I R A T I DIIT I [ IH ~ I [ ; ( I I [DY ~ I ; I N [ K A H I I i ~ IV l~/MK;~ GFGF! TYISHSSMI I~QKSI;HI, I ; I ~ I ; I ~ V P I Q I ID

A1 DROSOPHILA II DSPNAGMVKKLFVC, ALKDDHDEQSi RDYFQHFGNI VD VI bKET(~K~DYDPVbKWLQKQHQ.LNGKMVDVKKALPKQNDQ

Cl/C2 H U M A N NKTDPRSMNSRV,I GNLNTLWKKSDVEA, ,SKYGKI VG" . . . . . . CSVPIKGFAFVQ~VNERNARAAVAGEDGRM, AGQVLDI NLAAEPKVNR

E (UP2) HUMAN KAI~KEPVKK, FVGGLSPDTPEEKI REYFGGFGEVES, ELPMI)NI(rNKI~RGFCFi TI~IQEEP

U1 70K H U M A N DPNAQGDAFKTLFVARVNYDTTESKLRREFEVYG,I KRI HMVYSKRSGKRRGYAR EYIEHERDMHSAYKHADGKKI DGRRVLVDVERGRTVKG

NUCLEOLIN HAMSTER I VEGSESTTPFNLFI GNLNPNKSVAELKVAI SEPFAKNDLAW°DVRTGTHRKFGYVDFIESAEDLEKALELTGLK,VFGNEI KLEKPKGRDSI<K NUCLEOMN HAMSTER :I KNST=GESKTLVLSNLSYSATEETLQEVF:KATFI KVPQNQQGKS . . . . ~(GYAFI EF~ASFEDAAKEAIN $CNKIVEI EGRTI RLELQGPRG~P

NUCLEOLIN HAMSTER III SPNARSCPSKTLFVKGLSEDTTEETLKESFF;;GSVRAR. • ° I VTDRE'rGSSKGFGFVD'F1NSEEDAKAAKEAMEDGEI DGNK~ fLDWa.KPKGEGG /

rLEU PHE VAL X X LEU I F LYS ~_, v PHE GLY PHE ua~ ~ PHE 1 CONSENSUS ILE TYR ILE ILE ARG v~" TYR ALA TYR "~"- ~ 1YR

PHE VAL LEU VAL VAL THR ASN

RNP 2 RNP 1 Fig. 4. The 93 amino acid RNA binding domains in hnRNP, mRNP, snRNP and pre-ribosomal RNA-binding protehzs. The cytoplasmic mRNP poly(A ) binding protein (PABP) from human z7 and yeasd 2,z¢ and the nuclear hnRNP AI proteb~s from rat and Drosophila -'s are illustrated in pair=vise comparisons with the vertical lines between amino acids (single letter code) indicating matching pairs. Four of these domab~s exist in the case of the PABP and two for AI. Also shown are tize hnRNP proteins C!1C224, the 70 kDa protein from rite UI snRNP, and the three domahts of the nucleolar pre-ribosomal RNA-bhzding protein, nucleolin a°. Two regions of extensive amino acid similarity are highlighted by shading. The boxed region, RNPI, corresponds to dze RNP consensus sequence, a stretch of eight amino acids described previouslfl 2,24 which is the most highly conserved region b~ the 93 ambm acid domab~s. The tinted strip b~dicates RNP2 which is the next most conserved region. These regiods were assigned by the criterion of havb~g four or fewer different amh~o acids at that position - other positions meeting this restriction are indicated by an asterisk. Sequences were aligned first by pair-wise comparisons and then visually by including gaps bzdicated by dots within a sequence. Only a partial sequence is available for the E(UP2) proteb~s. The citation for the original references for the sequences of the hnRNP AI, the snRNP UI 70 kDA, and the UP2 proteins, is found in Ref. 24.

90 TIBS 13- March 1988

YFAST PABP YEAST PABP

i II III IV I II III

_ , 1 o 32 27 33 _ '1 -32 28 32 = , 34 [ ] z4 37 32 - 36 3 8 n n

Z III 29 50 [ ] 46 ~ III 28 36 - 42 .¢¢ =E "¢[ "1 UJ "1- IVI 28 41 41 [ ] >- IV I 32 38 42 -

RAT A1 RAT A!

~, I II -~ I II

a. I ~ 29 ~ I - 33 f~ , 3o [ ] :33 -

PABP A1

RNP2 RNP; NHzr-F I,] ~ ~ ~ ~ _ ~ C O O H

I II III IV

RNP2 RNP1 NH2-1 H 1 H Iq 1 I - C , ~ C O O H

I II

Rg. 5. The percentage similarity of amino acids within RNA.binding domains of RNP proteins. The individual RNA-binding domains of the human and yeast PABP and the rat and Drosophila hnRNP A2 proteins are contrasted with self comparisons (yeast to yeast PABP ar, d rat to rat AI ). The percentages indicate amino acid identities between individual domains using pairwise comparisons as shown in Fig. 4. The four RNA binding domains of the PABP and the proline-rich carboxyl terminus (hexagon), and the two domains of the A1 protein and its glycine-rich carboxyl terminus (oval) are also illustrated.

acid segments are RNA-binding domains is based on the following con- siderations: (1) this feature is common to all of the RNA-binding proteins se- quenced; (2) UP1, a 24 kDa amino-termi- nal portion of A1, which is comprised only of the two 100 amino acid repeats, binds RNA25, 26. Similar findings have been made for the C proteins and PABP (E. Mortenson and G. Dreyfuss, unpub- lished). Since the 100 amino acid repeats are the RNA-binding domains of the proteins, and since the RNP consensus peptides (RNP1 and RNP2) are the most highly conserved section of these larger domains, it seems likely that they rep- resent important common structures necessary for the RNA-binding domain. It is not yet clear, based on recent experi- ments for the yeast PABP, whether RNP1 is essential for specific high-affin- ity RNA binding 30. Furthermore, the high degree of conservation within the domain suggests that all the proteins that contain it evolved from a common ancestral gene 12.

In a pairwise comparison of the PABP between yeast and human and those of A1 between rat and Drosophila we have noticed that each domain (indicated as I, 1I, etc., from the amino terminus of the protein) is significantly more similar to the corresponding domain in a divergent species than it is to any other domain in the same protein (within the same organism). This indicates that each domain has been conserved indepen- dently of the others and therefore suggests that the domains in the same protein are not simply redundant repeats but rather serve a different function,

.J

albeit probably only subtly so, from the others.

Conclusions and perspectives Several important advances made

over the past several years have led to considerable progress in the knowledge and understanding of hnRNP particles and hnRNP proteins. Thus, whereas sev- eral years ago the unambiguous identifi- cation of RNP proteins posed consider- able difficulty, it may now be possible to classify new proteins as RNP proteins on the basis of the RNP identifier sequence. However, much remains to be investi- gated about the structure, assembly and disassembly of hnRNP particles, about the detailed mechanism of the participa- tion of hnRNP proteins in mRNA pro- cessing and transport and about the structure of the hnRNP proteins and how they interact with hnRNA. The availability of specific antibody and DNA probes as well as the/n v~o sys- tems for transcription, RNA splicing and polyadenylation should make additional progress possible.

Acknowledgements This work was supported by grants

from the National Institutes of Health (GM31888 and GM37125). G. Dreyfuss is an Established Investigator of the American Heart Association.

References 1 Dreyfuss, G. (1986) Anna. Rev. Cell. Biol. 2,

459--498 2 Chung, S. Y. and Wooley, J. C. (1986) Proteins

1,195-210 3 Beyer, A. L., Christensen, M. E., Walker,

B. W. and Le Stonrgeon, W. M. (1977) Cell 11,

127-138 4 Martin, T. E., Billings, P. B., Pullman, J. M.,

Stevens, B. J. and Kinniburgh, A. J. (1978) Cold Spring Harbor Syrup. Quant. Biol. 42,899- 909

5 Lothstein, L., Arevistoff, H. P., Wooley, J. C., Chung, S. Y., Walker, B. W. and LeStourgeon, W. M. (1985)/. CellBiol. 100,1570--1581

6 Wilk, H. E., Wen', H., Friedrich, D., Kiltz, H. H. and Schaefer, K. P. (1985) Fur. J. Biochem. 146, 71-81

7 Stevenin, J. H., Gallinaro-Matfinge, R., Gattoni, R. and Jacob, M. (1977) Fur. J. Biochem. 74, 589--602

8 Van Eekelen, C. A. G. and Van Venrooij, W. J. (1981) J. Cell Biol. 88,554-563

9 Economides, I. V. and Pederson, T. (1983) Proc. Natl Acad. Sci. USA 80,1599-1602

10 Dreyfuss, G., Choi, Y. D. and Adam, S. A. . (1984) MoL Cell. Biol. 4,1104-1114 I1 Choi, Y. and Dreyfuss, G. (1984) I. Cell Biol.

99,1997-2004 12 Adam, S. A., Nakagawa, T., Swanson, M. S.,

Woodruff, T. K. and Dreyfuss, G. (1986) Mol. Cell. Biol. 6, 2932-2943

13 Adam, S. A. and Dreyfuss, G. (1987) J. ViroL 61, 3276--3283

14 Choi, Y. and Dreyfuss, G. (1984) Proc. Natl Acad. Sci. USA 81, 7471-7475

15 PifioI-Roma, S., Choi, Y. D., Matunis, M. J. and Dreyfuss, G. (1988) Genes Develop. 2,215- 227

16 Osheim, Y. N., Miller, O. L. and Beyer, A. L. (1985) CeU43, 143-151

17 Choi, Y. D., Grabowski, P. J., Sharp, P. A. anti Dreyfuss, G. (1986)Science 231,1534-1539

18 Sierakowska, H., Szer, W., Furdon, P. J. and Kole, R. (1986) Nucleic Acids Res. 14, 5241- 5254

19 Guthrie, C. (1986) Trends Biochem. Sci. !1, 430-434

20 Maniatis, T. and Reed. R. (1987) Nature 325, 673-678

21 Leser, G. P., Escara-Wiike, J. and Martin, T. E. (1984) J. Biol. Chem. 259,1827-1833

22 Setyono, B. and Greenberg, J. R. (1981) Cell 24, 775--783

TIBS 13-March 1988 91

23 Sachs, A. B., Bond, W. M. and Komberg, R. D. (1986) Cell 45, 827-835

24 Swanson, M. B., Nakagawa, T. Y., LeVan, K. and Dreyfuss, G. (1987) Mol. Ceil. Biol. 7, 1731-1739

25 Kumar, A., Williams, K. R. and Szer, W. (1986) J. Biol. Chem. 261,11266-11273

26 Pandolfo, M., Valentini, O., Biomonti, G., Morandi, C. and Riva, S. (1985) Nucleic Acids Res. 13, 6577--6590

27 Grange, T., Martin, de Sa, C., Oddos, J. and Pictet, R. (1987) Nucleic Acids Res. 15, 4771- 4787

28 Haynes, S. R., Rebbert, M. L., Mozer, B. A.,

Forquignon, F. and Dawid, I. B. (1987) Proc. Natl Acad. Sci. USA 84, 1819-1823

29 Lapeyre, B., Bourbon, H. and Amalric, F. (1987) Proc. Natl Acad. Sci. USA 84, 1472- 1476

30 Sachs, A. B., Davies, R. W. and Komberg, R. D. (1987) Mol. Cell. Biol. 7,3268-3276

EFTu provides an intemal kinetic standard for translational accuracy

Robert C. Thompson

Dta'ing polypeplide chain elongation the ribosome interacts with aminoaeyl-transfer RNA (aa-tRNA) and with elongation factor (EFTu). The rate constants characteri~g the ribo- some' s interactions with EFFu are independent of whether the aa-tRNA is cognate or noncog- nate. These rate constants act as an internal kinetic standard to measure lhe kinetics of the ribosome' s bgerac~n ~ the. aa-tRNA, and thereby help deterrm'ne whether the aa-tRNA is

cognate or noncognate.

Translation of mRNA into protein is accurate primarily because in response to a particular codon, the ribosome binds one aa-tRNA (cognate) but rejects others (noncognate). Cognate and noncognate ribosome-mRNA.aa-tRNA complexes differ in the extent of ba.~e pairing between the codon and anti- codon. Although the structural differ- ences between cognate and noncognate codon-anticodon pairs are not yet well understood, our knowledge of the way in which these differences influence the selection of aa-tRNA by the ribosome has advanced recently. A knowledge of the mechanism of this selection process turns out to be essential for understand- ing translational accuracy because the fidelity of translation of any given codon is not a constant. It can be modulated by a wide variety of genetic and environ- mental variables that include ribosomal protein and elongation thctor mutations, codon context, ppGpp levels, tempera- ture, and polycation concentrations.

the aa-tRNA, EFTu and GTP binds to the ril~,osome (kl), and either dissociates (k_l) or undergoes GTP hydrolysis (k2). Following acceptance of the ternary complex through GTP hydrolysis, the aa-tRNA is proofread and either is incorporated into the nascent protein chain (k3), or dissociates from the ribo- some (k4). Experimental evidence for the existence of this double screen against noncognate aa-tRNAs was first obtained by studying the reaction between poly(U)-programmed ribo- somes and a ternary complex of Leu- tRNA2, which is nearly cognate and makes two out of three correct base- pairs with the mRNA. A significant frac- tion of this ternary complex passes the first discrimination test leading to the hydrolysis of GTP, but most of the aa-tRNA is then rejected (k4) rather than being incorporated into peptide (k3) (Refs 5 and 6). More recent work in a full

Translational accuracy involves two recognition steps

Figure 1 illustrates the minimal mechanism that has been establi~hed for the process by which aa-tRNAs are incorporated into nascent proteinm. As originally proposed on theoretical grounds by Hopfield 3 and by Ninio 4, ribosomes discriminate against non- cognate aa-tRNAs in two distinct steps. In the first of these, known as initial recognition, a ternary complex (TC) of

R. C. Thompson is at the Department of Molecular, Cellular, and Developmental Biology, University of Colorado, Boulder, CO 80309, USA.

protein synthesis system also indicates the presence of a proofreading process in aa-tRNA selection by ribosomes 7.

Translational accuracy requires an internal kinetic standard

Although the two-step nature of the selection process was established by relatively simple experiments, a more complete understanding of the physico- chemical basis for discriminating against noncognate aa-tRNAs requires a knowl- edge of the elementary rate constants k l, k_ l, k 2, k 3 and k4. Experiments to determine these values have shown that k_ ~ and k 4 are the rate constants that dif- fer most between cognate and near-cog- nate aa-tRNAs 2. These experiments also provided a clue to the existence of an internal kinetic standard that enables the ribosome to determine whether an aa-tRNA is cognate or noncognate and prevents the reaction with noncognate aa-tRNA proceeding to the point that an error becomes irreversible. In retrospect it is easy to see the need for some internal standard for accuracy because the ribo- some must decide whether an aa-tRNA is cognate without weighing, individu- ally, the relative merits of the 60 or so ternary complexes in, for example, the E. coli cell.

The most important property of an internal standard is that it be indepen- dent of the quantity being measured, in this case the cognate or noncognate nature of the aa-tRNA. Considering first the initial recognition of

k~ k z ) J RS + TC ~ RS-TC = RS-EFTu-GDP,aa-tRNA

k-I k 4 ~

RS • pep- tRNA 4-

EFTu "GDP

ao-tRNA ÷ RS'EFTu'GDP

RS+EFTu" GDP

INITIAL RECOGNITION PROOFREADING t . . . . . . . ..=_1 I

Fig. 1. Minimal m¢char~ism by which aa.tRNAs are incorporated into nascent protein. RS, ribosome; TC, ternary complex; Li"~,, elongation factor. See text for detaiL¢.

(~ 1988, Ei.~vier Publications Cambridge 0376- 5067/8gr~02.00