sign in sign up
<<
Home , Microbody, Protein biosynthesis

12

The of mRNA: synthesis

12.1 AN OVERVIEW OF PROTEIN involves the successive reading of the codons BIOSYNTHESIS of the mRNA by the aminoacyl-tRNAs in an ordered manner, and the linking of the In the preceding chapters the reader has amino acids to form a polypeptide chain. already encountered the concept that the This is a complex process and takes place on mRNA (messenger RNA) is an intermedi• an elaborate , the . The ary in the expression of that portion of direction of growth of the polypeptide chain the genetic information in the DNA that is from the N-terminus to the C-terminus encodes . The present chapter pre• [2], and the direction of reading of the sents a detailed consideration of the process mRNA is 5' ~ 3' [3, 4]. Figure 12.1 provides of translation of the mRNA. Although this a schematic summary of the interactions of will be prefaced with a summary of the main mRNA, tRNA and ribosome. It represents a features of translation, the reader is directed stage in protein biosynthesis just before to suitable textbooks of (e.g. the aminoacyl ester bond of the peptidyl• [1]) for a more elementary account of this tRNA is broken and the polypeptide chain topic. transferred to the a-amino group of the In essence, protein biosynthesis involves aminoacyl-tRNA. Also represented in Fig. translating the information of the sequence 12.1 are two sites on the ribosome to which of the four different of DNA or tRNA can bind, the A-site and the P-site, RNA into a protein sequence with twenty and the integral ribosomal enzymic activity different possible units. The that catalyses the formation of provides the conceptual basis of bonds, peptidyltransferase. this in the relationship of single amino acids The overall length of the mRNA and the to groups of three nucleotides in the mRNA rate of initiation are usually such that a (triplet codons); whereas tRNA (transfer second ribosome can attach to the mRNA RNA) provides its physical basis through before the first one has completed its poly• possessing an anticodon complementary to peptide chain. In fact, several a mRNA codon, and a specific covalently are normally found on a given molecule attached amino acid. Protein biosynthesis of mRNA, translating different parts of it

R. L. P. Adams et al., The Biochemistry of the Nucleic Acids © Roger L. P. Adams, John T. Knowler and David P. Leader 1992 516 The translation of mRNA

N amino acids s-1 for globin chains in rabbit 50S reticulocytes [8] have been reported. These are, however, much lower than the rates estimated for DNA or RNA synthesis (800 and 50 nucleotides s-1, respectively, in pro karyotes). It should be mentioned that the synthesis of certain small bacterial occurs in a manner not dependent on mRNA and ribosomes. The reader interested in this tRNA subject is directed elsewhere [9].

mRNA s' 12.2 THE GENETIC CODE

305 12.2.1 The standard genetic code

Fig. 12.1 Diagrammatic representation of a The elucidation of the genetic code repre• prokaryotic ribosome. Two tRNA molecules are bound to the ribosome in response to the mRNA sented one of the major breakthroughs codons designated n and n + 1. The tRNA in modern biology. Here we shall merely bearing the growing polypeptide chain is describe the features of the code, as accounts occupying the peptidyl site (rectangular area, of the history of the code are to be found in marked P), and the tRNA bearing an amino acid a number of reviews (e.g. [10-12]). the aminoacyl site (rectangular area, is occupying The genetic code is a triplet code with marked A). The peptidyltransferase centre, where the formation is catalysed, is individual amino acids represented in the represented by the semicircular area, marked X. mRNA by code words (codons) of three Note the exaggeration of the amino acid (1 A - nucleotides. Although certain codons also 0.1 nm) relative to the tRNA (75 A- 7.5nm), specify initiation and termination signals, the shape of the and the misrepresentation of the code is uninterrupted, with no 'commas' ribosomal subunits (cf. Fig. 12.21). between codons, and these follow one another in succession and do not overlap. The code was at one time thought to be simultaneously, and such groups of ribo• universal, i.e. each triplet codon had the somes are termed polyribosomes or poly• same meaning, regardless of the species. somes (Fig. 12.2). The size of the polysomes This assumption derived from comparison increases with the size of the mRNA: poly• between E. coli and higher mammals, where somes synthesizing haemoglobin P-chains initially it was shown that the tRNAs re• (Mr = c. 16000) contain four to five ribo• cognized the same codon triplets in vitro somes [5], whereas those synthesizing myosin [13], a result subsequently corroborated by heavy chains (Mr =c. 200000) contain about comparison of protein and 50-60 ribosomes [6]. sequences. It is now known that in certain The rates of protein synthesis in pro• organisms and in the mitochondria of karyotes and appear to be quite eukaryotes the genetic code differs from similar: values of 15 amino acids s-1 for that first established in E. coli (section P-galactosidase in E. coli [7], and seven 12.2.3). However, the latter code is so wide- The genetic code 517

Fig. 12.2 Electron micrograph showing the translation of silk fibroin mRNA on polysomes. The extended fibrous fibroin molecules can be seen emerging from the ribosomes (dark irregular particles). The length of the fibroin molecules increases from the top right to the bottom left of the frame, indicating that this is the 5'- 3' direction along the mRNA (courtesy of Dr Steven L. McKnight and Dr Oscar L. Miller, Jr).

spread (it is found in the vast majority of It can be seen from Fig. 12.3 that 20 dif• and eukaryotes, animals and ferent amino acids are specified by the plants) that we shall refer to it as the standard genetic code. Other amino acids are found genetic code. It is presented in Fig. 12.3. in proteins, but almost all of these are 518 The translation of mRNA

Second letter u c A G uuu } UAU } UGU u uuc Phe ucc UAC Tyr UGC l Cys c u UUA } ""]UCA Ser UAA } UGA StoP A UUG Leu UCG UAG Stop UGG Trp G CAU u ..... CAC l His ::r wulcue ccuCCC CGUCGC c -, c CUA Leu CCA lPro CGA lArg c. I... CAA } Gln A .....Q) ..... CUG CCG CAG CGG G ,...CD Q) ,... u CD ..... AUU AAU } AGU } -, en AUC l Ile AWlACC AAC Asn AGC Ser c I... A ACA Thr -u.. AUA AAA ) LYS AGA } Arg A AUG Met ACG AAG AGG G GAU u GAC l Asp G wuGUC l Val GCC~u l Ala GGUGGC l Gly c GUA GCA GAA J Glu GGA A GUG GCG GAG GGG G

Fig. 12.3 The standard genetic code. Termination codons are indicated as 'Stop'. generated by post-translational enzymic ation codons. These tRNAs occur in the modification of these 20 genetically-defined 'suppressor' strains of E. coli, so called for amino acids. The one known exception is the their ability to suppress particular classes rare amino acid, selanocysteine, which is of 'nonsense' mutants. sequence encoded by UGA, otherwise a termination analysis of such a mutant of a tRNA Tyr from codon (see below). As this is not a complete an amber suppressor strain showed that its reassignment of the meaning of the codon, anticodon is changed from 3'-AUG-5' to 3'• but an alternative translational possibility, AUC-5'. Thus, it is able to insert thought to depend on the broader context of into a polypeptide chain in response to the the mRNA, we shall consider it in section termination codon UAG, rather than to the 12.9.6 with other examples of 'suppression' tyrosine codon UAC [14]. The reason that of termination codons. such suppressor mutants are viable, and do It is also apparent from Fig. 12.3 that not exhibit premature termination of the three ofthe codons (UAA, UAG and UGA, bulk of normal proteins, is because the designated 'Stop') do not normally specify occur in the minor, otherwise an amino acid, but are all signals for the redundant, representatives of certain pairs termination of the polypeptide chain. For of isoaccepting tRNAs. historical reasons relating to the type of Although inspection of Fig. 12.3 does not suppressor mutations (see below) that reveal a codon, the sole role of which is to characterized them, UAG and UAA are specify the start of a polypeptide chain, the also referred to as 'amber' and 'ochre', codon AUG fulfills this role as well as that of respectively; and UGA is sometimes called residues in the body 'opal' or 'umber'. Although the termination of the polypeptide chain. In E. coli the process does not involve tRNA (section initiation of protein synthesis involves the 12.4.3), certain mutant tRNAs have been AUG codon being decoded by a unique found that can recognize individual termin- species, N-formylmethionyl-tRNA (fMet- The genetic code 519

tRNA) [15]. The tRNA that inserts the Triplets coding for the same amino acid are initiating fMet into polypeptide chains, not distributed at random, but are grouped tRNAret, has a different nucleotide together so that they generally share the sequence from the tRNA that inserts Met same 5' and middle base (although there are internally: tRNA~et [16, 17]. Both species two separate groups of codons in the cases accept methionine, but only the Met-tRNAr of Ser and Leu). This has the consequence can then be formylated by a transformylase that mutations producing a change in the that has N10-formyltetrahydrofolic base at the 3'-position of the codon often acid as a cofactor [18]. Although have no effect on the amino acid specified. of Met-tRNAret is normally an absolute Furthermore, the different amino acids are requirement for initiation in E. coli and segregated to a considerable extent on the many other bacteria, the specific recognition basis of chemical similarity (hydrophobicity, of the initiator tRNA in the initiation process hydrophilicity, acidity and basicity). Thus (sections 12.4.1 and 12.5.1) is most certainly a in the 5'-base of any of the influenced by the structure of the tRNA six codons would give a codon itself. Thus, in eukaryotes the methionyl re• specifying another hydrophobic amino acid. sidue of the initiator tRNA is not formylated, Such a change might not impair the function and the initiator tRNA of at least one of a particular if the altered bacterium can compensate for lack of amino acid merely performed a structural formylation by undergoing a base substi• role in the hydrophobic core of this protein. tution [19]. It has therefore been argued that the specific In bacteria AUG, although the main arrangement of codons in the genetic code initiation codon, is not the sole one. The serves to reduce the potentially harmful usage of GUG and certain other much rarer effect of mutations. codons for initiation is discussed in section It might have been expected that each de• 12.4.1; but at this point it is important to generate codon would require its own tRNA stress that these minor initiation codons, like with a corresponding anticodon. Although AUG, are recognized by fMet-tRNA. there are some such discrete isoaccepting Although fMet or Met initiate polypeptide tRNAs, recognizing the same amino acid, chains in prokaryotes and eukaryotes, their number is less than the 61 required if respectively, these amino acids are not all codon-anticodon interactions involved necessarily found at the N-terminus of the three standard Watson-Crick base pairs. mature protein. In bacteria the formyl group This situation is viable because there is a is removed by a deformylase [20] and the degree of latitude or wobble in the com• methionine group is frequently removed by plementary base-pairing between the base in an aminopeptidase [21], the action of which the 3' -position of the codon and that in the is governed by the nature of the penultimate 5'-position of the anticodon. Rules govern• residue in the polypeptide chain [22]. ing the extent of this latitude were proposed by Crick in his wobble hypothesis [23]. He considered the stereochemistry of possible 12.2.2 The degeneracy of the genetic non-standard base-pairing that might allow code the two pairs of 3'-codon bases that are grouped together in degenerate codons, U Since many of the 20 amino acids are and C, and A and G, to be recognized by encoded by more than one triplet (Fig. a common 5'-anticodon base. He found 12.3), the code is said to be degenerate. that relatively small movements from the H I G G o---H-:h (~-r--H-Nh Q) }-N u /N f N-H·--·Nf ' /N f N-H--- 0 \ N=< r-N c N=< N-H ---0 \ N-H I HI (a) H (b)

(c) (d)

H H \ I N-H (N);-1(0 ---H-Nh L rl H /N--1( ;N-H---·( ) ~=(--- H-J ""/A Nd }-N C 0 \ I ~N-H---K"'\ '-=N (e) (f)

(g) (h) Fig. 12.4 Some proposed patterns of hydrogen bonding involving nucleosides in the 'wobble' position of the anticodon of tRNA. In each case the 5' -nucleoside of the tRNA anticodon is shown on the left, and that of the 3' -nucleoside of the mRNA codon on the right, with the ribose represented as •. (a) and (b) 'wobble' for guanosine; (c), (d) and (e) 'wobble' for inosine; (f) proposed base-pairing for lysidine, R = -CH2CH 2CH 2CH 2CH(NHt)COo-; (g) and (h) possible interactions of uridine with pyrimidines in mitochondrial tRNAs. The genetic code 521

Fig. 12.5 Stereoscopic view of the anticodon bases 34-36 of yeast tRNAPhe and the 31-hypermodified base yW37 as viewed from the interior of the molecule. It can be seen that the stacking between resi~ues yW37, A36 and A35 is greater than that between A35 and the 'wobble' base G34. (From (49], with permission.)

1 standard positions might allow the 5 - seems to lie in the fact that the base imme• anticodon base, G, to ·pair with both diately 31 to the anticodon, which is always pyrimidines, U and C, in the 31 -position of a purine (Fig. 12.7) and frequently heavily the codon; and allow the 51 -anticodon base, modified, has strong base-stacking inter• U, to pair with both purines, A or G, in the actions with the two adjacent bases of the 3'-position of the codon. Moreover it had anticodon (Fig. 12.5). This serves to anchor already been found that several tRNA anti• these bases in a helical conformation that codons contained the nucleoside inosine only allows standard Watson-Crick base (which pairs only with C in a double helix), pairs between the anticodon and the two first and he suggested that this might be able to codon positions. pair with A, U or C in the 31 -position of Although the basic concept of the wobble the mRNA codon. The hydrogen bonding hypothesis has been validated, it is import• patterns of these wobble interactions are ant to realize that not all its detailed pre• shown in Fig. 12.4. The elucidation of the dictions have turned out to be correct. three-dimensional structure of tRNA Furthermore, one must consider the rules (section 12.3) helps explain how this wobble for the base-pairing of important modified occurs. It is possible to accommodate a small 51 -nucleosides of tRNA which were unknown conformational change into the tertiary at the time of the formulation of the wobble structure of the anticodon of yeast tRN A Phe, hypothesis. Table 12.1 presents a compari• such that the 51 -anticodon wobble base Gm son of Crick's proposals with the observed (the hydrogen bonding pattern of which is pattern of codon-anticodon interaction in identical to G) could move into an alter• prokaryotes and the of eukaryotes, native position that would allow pairing with on the one hand, and mitochondria (which U rather than C. The reason that such a show a unique pattern of interaction), on change is confined to the 51 -anticodon base the other hand. The wobble predictions that 522 The translation of mRNA

Table 12.1 Predicted and observed 'wobble' base-pairing

3' -Codon nucleoside

Prokaryotes, 5' -Anticodon Crick's and eukaryotic Non-plant nucleoside prediction cytoplasm mitochondria

Unmodified c G G G* G C,U C,U C,U A u n.f. t n.f. u A,G n.U A,U,G,C

Modified (C) m5C, Cm, ac4C G n.f. L A n.f. (G) Gm,Q c c,u§ (A) I U,C,A U,C,A n.f. (U) mcm5U, mcm5s2U, mnm5s2U A n.f. mo5U, cmo5U, mcmo5U A,G,U n.f. cmnm5U n.f. A,U

-:No prediction was made for the modified 5' -nucleoside (unknown at that time). n.f.: The 5' -nucleoside is not found in tRNAs of that category. • Except for that occurring in tRN A~et, which decodes both A and G. t Except for isolated examples in which the prediction in probably incorrect (see text). *Except for isolated examples in which all four 3'-nucleosides are decoded (see text). §With 0 in the 5'-anticodon position oft RNA.

have been substantiated are the ability of the the wobble position are in the same family as 5'-anticodon nucleoside G to recognize 3'• a codon with C in that position. Likewise codon nucleosides U and C (but not A or in three- or four-codon families, an A in G), and the 5' -anticodon nucleoside I to the initial transcript is replaced by an I recognize 3'-codon nucleosides U, C and in the mature tRNA, which can decode three A (but not G). Furthermore, with one codons. In the rare cases in which A does exception (considered below), 5' -anticodon occur in the wobble position of the anticodon nucleoside C only recognizes 3'-codon (a tRNAThr of mycoplasma and a tRNA Arg nucleoside G. The problems arise with the of yeast mitochondria are examples) the wobble predictions for the recognition pro• pattern of base-pairing is unclear, however. perties of 5'-anticodon nucleosides A and For example, the tRNAArg of yeast mito• U, neither of which occur in an unmodified chondria with anticodon 3'-GCA-5' would form in the vast majority of prokaryotic or seem to have to decode the Arg codons eukaryotic cytoplasmic tRNAs. in the 5'-CGN-3' family, all of which are The almost complete absence of A from represented in the yeast mitochondrial the 5'-position of anticodons has been genome. Thus, a suspicion existed that there rationalized on the basis that it is more was some more fundamental reason for the economical to use G, as all codons with U in rarity of an A in the wobble position of The genetic code 523

tRNA anticodons. This has been confirmed whereas mo5U and cmo5U decode all three by experiments in which a 3'-CCA-5' of A, G and U (these abbreviations are anticodon was engineered into an E. coli defined at the front of the book). Structural tRNA Gly, which did not then show the explanations for the extension or restriction predicted discrimination against of wobble recognition with some of these codons other than 5'-GGU-3' [24]. It there• modified U residues have been proposed in fore appears likely that the reason that A terms of the adoption of the flexible C3'• is avoided in the wobble position of anti• en do or the rigid C2' -en do conformation codons is that, contrary to prediction, it (Fig. 2.16) of the ribose [27, 28]. It is in• would wobble. teresting, however, that mitochondria have The situation with U and the 5 '-position evolved tRNAs with a derivative of U, of anticodons is somewhat similar. In cmnm5U, that allow them to decode codon prokaryotes and the cytoplasm of eukaryotes pairs, NMA and NMG, according to the U in the anticodon position of tRNAs is wobble rules, hence minimizing their com• almost always modified. The reason for this plement of tRNAs [29]. Because this would seem to be that the wobble that could modified base has not been found in the occur with unmodified U would exceed the wobble position of non-mitochondrial tRNAs limitation predicted by the hypothesis. In one suspects that there are other features in two-codon amino acid families this would the structure of the mitochondrial tRN As cause violation of the genetic code; in four• that contribute to the pattern of decoding. codon families the extended wobble is Similar extra-anticodon structural features generally not efficient enough to be utilized. must be invoked to explain the ability of The basis for this viewpoint is as follows. the C in the 3'-UAC-5' anticodon of yeast Unmodified U does occur in mitochondria mitochondrial tRNA~et to recognize both (and also in mycoplasma [25]), where it can AUA and AUG [30]. One possible structure decode all four 3'-codon bases. In mito• for the C:A base-pair that is presumed to chondria this has the important consequence form is that found in crystals of an artificial that fewer tRNAs are needed to decode deoxyoligonucleotide [31]. four-codon families (section 12.7 .2). When The essence of the wobble hypothesis is the 3' -CCC-5' anticodon of the afore• the possibility of alternative base-pairing mentioned E. coli tRNA Gly was mutated to interactions involving the 5'-base of the 3'-CCU-5', the unmodified wobble U was anticodon. It is evident from the foregoing able to decode Gly codon GGA, but was discussion that a sound structural basis exists much less efficient than the mycoplasma for the standard wobble base-pairings of tRNA Gly with the same anticodon in trans• unmodified and modified anticodon bases. lating the codons GGG, GGU or GGC [26]. What, however, of the decoding of four• Thus, other structural features of the tRNA codon families described above for the besides the anticodon are required to allow unmodified U found in some mitochondrial unmodified U to decode effectively all four and mycoplasma tRNAs? One possibility is bases in the third position of the codon. that the unique structure of these tRNAs Even the modified derivatives of U em• allow a relaxed pattern of base-pairing in ployed by the majority of prokaryotes and which, by rotation of the C4' -C5' and P-0 the eukaryotic cytoplasm do not decode bonds, U:U and U:C become close enough according to the wobble rules: mcm5s2U, together to form satisfactory hydrogen mcm5U and mnm5s2U decode only A; bonds (Fig. 12.4) [32]. However, an alter- 524 The translation of mRNA

native possibility is that there is no inter• dria the most widespread of these is the action in the third position in this situation, reassignment of one or two of the three and that decoding is by the 'two-out-of-three' termination codons to an amino acid. Thus, mechanism proposed by Lagerkvist [33]. in some ciliated protozoa and algae UAA This 'two-out-of-three' hypothesis was and VAG code for rather than actually developed before the situation in acting as chain-termination codons [35], mitochondrial tRNAs became apparent, and whereas in several species of the , proposes that in certain circumstances a Mycoplasma, these codons are chain ter• single tRNA that can recognize the first two minators but UGA codes for codons of a four-codon family has the [36]. In Euplotes octocarinatus, another potential to decode the whole of such a ciliate, UGA codes for , with UAA family. The four-codon families for which functioning as a termination codon and such 'two-out-of-three' recognition was pro• VAG apparently not being employed in any posed are UCN, CUN, CCN, CGN, ACN, capacity [310]. Species of tRNA with appro• GUN, GCN and GGN (Fig. 12.3). All of priate anticodons exist in these species to these involve at least one strong G:C base decode these erstwhile termination codons. pair in the first two positions. Some experi• How can these changes be accounted for mental evidence to support this hypothesis in evolutionary terms? The first stage is was obtained from the successful translation thought to have been the loss of the par• of phage MS2 RNA in a -free system ticular termination codon or codons from from E. coli in which the individual tRNAs the repertoire in use. This could have been that can normally translate particular codons facilitated by a small coding capacity in were omitted. These could be replaced by the case of mitochondria, and by pressure tRNAs that would be predicted to translate against usage of G in the extremely AT-rich them according to the 'two-out-of-three' genome of Mycoplasma. Next, the for• hypothesis, but not according to the 'wobble' mation of a new tRNA or a mutation in the hypothesis [34]. The efficiency of this pro• anticodon of a tRNA would have been cess was such that it would be unable to required to allow recognition of the codon compete with the normal tRNA that can as an amino acid when it re-emerged in the make a 'three-out-of-three' interaction, and mRNA. For example, the mutation of the would, therefore, be most unlikely to par• tryptophan anticodon from 3'-ACC-5' to ticipate in contemporary codon-anticodon 3' -ACU-5' (with appropriate post-trans• interactions in prokaryotes (or eukaryotic criptional base modification) would have cytoplasm). Nevertheless it cannot be ex• allowed it to recognize the codon UGA as cluded that in the peculiar environment of well as UGG (Table 12.1). Such a change the similar interactions might would have had to be accompanied by a loss be involved in the decoding of four-codon of, or alteration in specificity of, the termin• families by anticodons containing unmodified ation factor that had previously recognized U in the wobble position. the codon. In Mycoplasma the loss of RF-2 (section 12.4.3) could have accomplished this. 12.2.3 Alternative genetic codes The second type of divergence from the standard genetic code involves changes in There are two types of divergence from the the assignment of codons specifying amino standard genetic code. Outside mitochon- acids. Most of the examples are encountered The genetic code 525

UGU (i) UGU }Cys UAU }Tyr (iv) UAU} UGC}Cys ___.. UGC UAC ___.. UAC Tyr UGA Stop UGA}Trp UAA} UAA UGG Trp UGG UAG Stop UAG Stop

AGU AGC }Ser (ii) AGU} cuu} cuu} (v) ___.. AGC Ser cue Leu ___.. cue Thr AGA AGA CUA CUA AGG}Arg AGG CUG CUG

AUU} AAU} AUU} (iii) AUC lie AAC Asn (vi) AAU} AUC lie ___.. ___.. ~ Asn AUA AUA AAA AUG Met AUG }Met AAG }Lys AAG Lys

(a) Simplifying changes (b) Other changes

Fig. 12.6 Some deviations from the standard genetic code found in non-plant mitochondria. The figure shows the changes in relation to the other members of the four codon family involved. The changes indicated are found in: (i) all non-plant mitochondria examined; (ii) echinoderms, molluscs, nematodes and platyhelminths; (iii) vertebrates, arthropods, molluscs, nematodes and yeasts; (iv) platyhelminths; (v) yeasts; (vi) echinoderms and platyhelminths. Details of other changes can be found in [37].

in non-plant mitochondria, which will be also be envisaged in terms of temporary the focus of consideration, but a non• abandonment of use (all the examples are mitochondrial example is known. In the degenerate, i.e. other codons for the same fungus, Candida cylindracea, CVG encodes amino acid remain) followed by reassign• rather than leucine [35]. Some of the ment. In the case of a gross change, such changes found in non-plant mitochondrial as that in which all the codons in yeast mito• genetic codes [37] are illustrated in Fig. chondria of the type CUN changed from Leu 12.6. It should be stressed that there is no to Thr, all that would be required would be single mitochondrial genetic code, implying for a mutation (outwith the anticodon) in that different mitochondrial genomes have the single tRNA that decodes this family to undergone separate evolution. Indeed, plant cause a change in aminoacylation specificity mitochondria (and ) employ the (section 12.3.2). In the case of the con• standard genetic code. Examination of the version of the AAA codon from Lys to Asn nature of the changes suggest that although in echinoderm mitochondria, mutation of some appear quite gratuitous (e.g. Leu to the Asn anticodon from 3'-UUG-5' (which Thr for CUN in S. cerevisiae), others can can only decode AAC or AAU) to 3'• be regarded as simplifying the code into UUI-5' could have accomplished this if four-member or two-member families (e.g. accompanied by the loss of the tRNA Lys change from Arg for AGR). This facilitates uniquely decoding AAA. In other cases the decoding of the mitochondrial genome there has been no change in the anticodon, by a smaller number of tRNAs according to but a change in the decoding capacity of the the special rules already described in section tRNA, presumably as a result of mutation 12.2.2. elsewhere in the molecule. Thus, the Met• Such changes in the value of individual tRNA anticodon 3'-UAC-5' normally codons from one amino acid to another can decodes only AUG, but in yeast mitochon- 526 The translation of mRNA

dria it has acquired the capacity to decode There are two main ideas as to how the AUA as well (section 12.2.2). pattern of codon usage might regulate trans• lation. The first suggestion, for which the indirect evidence is strongest, is that the 12.2.4 Differential codon usage speed of translation of synonymous codons varies with the abundance of the correspond• Because the genetic code is degenerate, ing isoaccepting tRNAs (section 12.2.2). A there exist, in principle, alternative choices study of the iso-accepting tRNAs in E. coli between 'synonymous' codons for most of showed a strong correlation between tRN A the amino acids, for the termination signal, abundance and codon choice in the and even for the initiation signal. The usage of highly expressed proteins [40]. A similar of alternative initiation (section 12.4.1) and correlation between codon usage in highly termination codons (section 12.4.3) is dis• expressed genes and the relative abundance cussed later: the present section considers of iso-accepting tRNAs was found in yeast the pattern of choice of amino acid codons [41] and for the fibroin mRNA of the silk• and its implications. worm, B. mori [42]. Nevertheless, in E. coli The usage of such synonymous amino there is also a difference in usage of codons acid codons is definitely non-random, and of the type NMU and NMC, which cannot differs between prokaryotes and eukaryotes, be explained in terms of abundance of iso• and their hosts. In some cases, e.g. accepting tRNAs as they are both decoded certain eukaryotic viruses and mycoplasma, by a tRNA with G or I in the 'wobble' codon choice seems to be determined by the position (Table 12.1). It was observed in pressure to a particular extreme base com• such pairs that the usage correlated well position. However, in bacteria (and to a with the predicted codon-anticodon inter• certain extent in yeast) there is evidence action energy, those with either maximum that the reason for non-random codon usage (e.g. CGC) or minimum (e.g. AUU) energies is that it can influence the rate of translation. being less frequent in the genes for highly In E. coli it was observed that certain codons expressed proteins than those (e.g. CGU that were rarely used in the genes for highly or AUC) with intermediate energies [38]. expressed proteins (e.g. ribosomal proteins) It was argued that intermediate codon• were found at a significantly greater fre• anticodon interaction energies would pro• quency in the genes for weakly expressed duce more efficient decoding by allowing proteins (e.g. the lac and other repressors) the optimal balance between binding and [38]. Circumstances where such translational, release of tRNA. Although this latter idea rather than transcriptional, regulation is has not found universal acceptance, it is important are nicely illustrated by the case striking that a similar pattern of preference of DNA primase. The for this protein in NMY codons has been reported in yeast is in the same unit as those for [41 ]. S21 and the a factor of Such correlations between codon usage and RNA polymerase, proteins that are expressed , although highly suggestive, under similar circumstances but approxi• cannot be taken as proof of a causal effect of mately 800 and 50 times more strongly, codon usage on translation. However, strong respectively. The difference in codon usage experimental support has been obtained in between this gene and its cotranscribed both bacteria and yeast. One elegant experi• neighbours is most striking [39]. ment involved site-directed mutagenesis of The structure and aminoacylation of tRNA 527

AoH 3' AoH 3' c c c c A 5' pG • C 5, P • • • Acceptor C • G • . • Stem G • C ••• 70 G-U ••• A • U U • A ••• U • A C U ••• 1 •. • ILJY e A D G A U G A C A C m A * u •••• c R A 10 • .. • • A 5 • • • • • G I • Dl R D C U C.JG m C U G U G T C *G I ••Y• ylt••GT,,C G G A G C 'G C Um 7G 'VI R e, T G A m.C·GAG G A • • •. • m'· .. ·····• ...) 20 .. . C • G ••• A • U ••• G • m5C 30• • e•o A • l" • • • Cm A y II e U YW U R* Gm A A e anticodon Cal (b)

Fig. 12.7 Secondary structure of tRNA. (a) Yeast tRNAPhe; (b) generalized structure in which only the invariant or semi-invariant bases are named, the others being represented by filled circles. An asterisk indicates that a base may be modified. Hydrogen bonds in the helical stems are indicated by dots, except for the G:U pair in (a) which is indicated by a line for emphasis. The dotted lines indicate a variable number of nucleotides. RandY stand for purine and pyrimidine, respectively, other symbols being defined at the beginning of the book. An explanation for the numbering of the variable loops can be found in [45]. rare Leu codons to more highly used codons in section 12.2.3, this is one of the prere· in the attenuating leader of the leucine quisites for the evolution of variant genetic of Salmonella typhimurium (cf. codes. section 10.2.1), and this was found to have a profound effect on the attenuation [43). In yeast the highly expressed phosphoglycerate 12.3 THE STRUCTURE AND gene was mutated so that up to 39% AMINOACYLATION OF tRNA of major codons were replaced by minor ones, and this was found to cause a dramatic 12.3.1 The structure of tRNA [27, 45] decline in the extent of expression [44). In concluding this discussion of the usage The key role of transfer RNA in decoding of amino acid codons it should be mentioned the genetic information has already been that in the small genomes of mitochondria, mentioned. To reiterate, each tRNA can some codons are completely unrepresented covalently attach one specific amino acid to in the repertoire of mRNAs. As mentioned its 3' -end, and each possesses a sequence of 528 The translation of mRNA

T '1/tC stem

Dloop I I

Variable loop

'--~ Fig. 12.8 Tertiary structure of yeast tRNAPhc. The sugar- backbone is shown as a coiled tube, the numbers referring to the nucleotide residues starting from the 5'-end. Hydrogen bonding interactions between bases are shown as cross-rungs, tertiary interactions being shaded solid black. Bases not involved in hydrogen bonding are shown as shortened rods attached to the backbone (from [49] , with permission). three bases, the anticodon, complementary 1. A stem, the acceptor or aminoacyl stem, to and able to interact by hydrogen-bonded containing the 5' and 3 '-extremities of the base pairing with a mRNA codon for this, molecule. It consists of a helix of seven so-called, cognate amino acid. pairs of bases generally making Watson• The tRNAs range in length from 73 to Crick base pairs (i.e. A: U or G:C) but 94 nucleotides and are characterized by a occasionally (e.g. Fig. 12.7(a)) with a relatively large proportion of modified or G:U (cf. Fig. 12.4), together non-standard nucleosides (sections 2.1.2 and with an unpaired sequence of four bases 11.7.2). Almost all these primary structures at the end of the 3 '-strand of the stem. (mitochondrial tRNAs are the exception - The last three bases are always CCA and section 12.7.2) allow themselves to be it is to the 3' -terminal adenosine residue arranged into a common secondary struc• that the amino acid is attached. ture, specific and generalized examples of 2. An arm, the dihydrouridine (D) arm, which are shown in Fig. 12.7(a) and (b), comprising a stem of three or four base respectively. The common features of these pairs together with the D loop (loop I) structures are as follows . of eight to eleven nucleotides, some of The structure and aminoacylation of tRNA 529

which are invariant. Although loop I junction of the two domains comprises the generally contains one or more dihy• D-loop (I) and the T\{IC-loop (IV). The drouridine residues (hence one of its dimensions of the domains are similar: common names), several examples are approximately 60 A (6nm) from the T\{IC• known in which this base is absent. loop at the corner to either anticodon or 3. An arm, the anticodon arm, comprising a 3' -acceptor end; whereas the distance be• helical stem of five base pairs together tween these extremities is approximately 75 with the anticodon loop (loop II) of seven A (7.5nm). nucleotides. It is worth pointing out that The inner surface of the two domains the base 5' to the centrally placed anti• (broken line in Fig. 12.8) rather belies the codon is always U, and that two other description, L-shaped, that applies to the bases in the loop are semi-invariant (i.e. outer surface. It comprises a more or less restricted to being purines in one case and planar region that in two dimensions can be pyrimidines in the other). The 5' -base of the anticodon, the base in the 'wobble'

position, is frequently a modified or non• I AOH 3 standard base. c 4. An extra arm (III) of extreme variability, c ranging from four to 21 nucleotides. This may be either a helical stem together with s'p • a loop of three or four nucleotides ( 13-21 ·-· nucleotides in all), or merely a loop of ·-· three to five nucleotides. ·-· 5. An arm, the TqJC arm, comprising a ·-· helical stem of five base pairs and a loop ·-· (loop IV) of seven nucleotides. ·-· c •

.cJA G To date, X-ray crystal structures have I been determined for four tRNAs uncom• o plexed to any protein [27, 45-48]. The GTr structure of yeast tRNAPhc, for which greatest resolution has been obtained, is shown in Fig. 12.8. One way of regarding it is as having an L-shape, with two angularly ·-· oriented domains: one comprising the TqJC ·-· stem and acceptor stem with the 3' -terminal (~--··-· CCA at its furthest extremity; and the other U yW comprising the D-stem, the variable loop 0 0 0 (III) and the anticodon arm, with the anti• Fig. 12.9 Relationship of tertiary hydrogen codon loop (II) at its furthest extremity. The bonding interactions to the cloverleaf secondary manner in which the individual helical stems structure of yeast tRNA Phe. The generalized augment one another to form these two convention of Fig. 12. 7(b) has been applied to the structure in Fig. 12.7(a) to highlight the domains of extensive base stacking is perhaps involvement of invariant or semi-invariant bases the most fundamentally important feature in the tertiary interactions (extended lines). of the structure. The corner made by the (After [49], with permission.) 530 The translation of mRNA

CH3 I FN ' 0 N~ A9 ~N)' N~ ~ -N G22 H-~ + "J- 'e H-N N) N:.iN---· ~ )=N I I H : H II __ H-N m 7G46 I H 'N~N o-· \ ,' I I I H A23 (NHN-H---Oh H, H: H , : I N-1( ~---H-N~ ) U12 OV~N~ N,H .... Nd 'q-N C13 I 0 .. r"N .0

(0) (b) Fig. 12.10 Examples of hydrogen-bonding in yeast tRNAPhe that involve three bases. The numbering is as in Fig. 12.7(b). regarded as the top of a T, the base of which phosphate backbone. Although these (at about nucleotide 56) is equivalent to the hydrogen bonds may at first sight seem angle of the L. This inner surface contains esoteric, their occurrence in tRNAs merely many of the variable bases of the D-stem and reflects the fact that the bases involved variable loop (III), which, furthermore, are are not confined to the relative spatial not involved in any of the tertiary inter• orientations which they are forced to adopt actions discussed below. They are thus avail• in a standard RNA A-helix. The potential able for interaction with other molecules, for the 21 -OH of the ribose to participate in particularly the aminoacyl-tRNA synthetase hydrogen bonding may be the basis of the (section 12.3.2). ability of RNA molecules to form such non• The tertiary structure of yeast tRNAPhe is helical structures, not possible in the case of maintained by a large number of hydrophobic DNA. stacking interactions in the augmented The general base-stacked two-domain helices, together with additional specific structure stabilized by tertiary interactions hydrogen-bonding interactions between described above for yeast tRNAPhe also nucleotide residues that are often widely obtains in yeast tRNA Asp [ 48], and in yeast separated in the secondary structure clover• and E. coli tRNAret (section 12.2.1), leaf (Fig. 12.9). Many of these interactions suggesting that it is a general feature of are between invariant or semi-variant bases, tRNAs. (There are certain differences in the suggesting a rationale for their restricted conformation of the anticodon loop in the variance and a generality for the yeast initiator tRNAs which are discussed in tRNAPhe tertiary structure. These hydrogen• section 12.4.1.) There is also evidence in• bonding interactions are in no way confined dicating that the crystal structure corresponds to Watson-Crick base pairs, but include a to the structure adopted in solution [25]. variety of non-standard interactions, some However, as described in the next section, involving three bases (Fig. 12.10). Indeed, changes in this structure may occur on inter• some of the interactions involve the sugar- action with proteins. The structure and aminoacylation of tRNA 531

R R I N H -CH-I CO"" ®P - . Adenosme pp i N~-CH-COOH + ATP 2 +

(a)

OH OH R I ® . AMP + NH2-CH-CO- P -Adenosme

A CH2 0 a 0

(tRNA) (

(b)

Fig. 12.11 The activation of amino acids and their attachment to tRNA. The reaction occurs in two stages, (a) and (b), both of which are catalysed by the same enzyme, and to which the intermediate is bound (see text). The symbol - represents a bond with a relatively high standard free energy of .

12.3.2 The aminoacylation of tRNA the appropriate tRNA to form an aminoacyl• tRNA: Before a tRN A molecule can act as an adaptor by interacting with its corresponding (amino acid -AMP)E + tRNA ~ anticodon in the decoding process, it must 1 1 1 first be 'charged' with its cognate amino acid. tRNAramino acid 1 +AMP+ E1 The responsible for this process are called aminoacyl-tRNA synthetases (amino Although the amino acid is located at the acid-tRNA ligases EC 6.1.1.) and catalyse 3'-OH of the ribose of the terminal adenosine the reaction illustrated in Fig. 12.11. moiety of tRNA during peptide bond for• The reaction occurs in two stages [50, mation, the initial point of attachment can 51], in the first of which the amino acid be the 2'-0H, the 3'-0H, or either, depend• is activated by ATP to form an aminoacyl ing on the amino acid adenylate: [52]. After attach• ment, rapid migration between the two positions is possible [53]. The aminoacyl ATP + amino acid1 + E1 ~ ester linkage has a relatively high standard free energy of hydrolysis, derived from the (amino acid 1-AMP)E 1 + PPi ATP hydrolysed during its activation. This is important as it provides the necessary The aminoacyl-adenylate complex then energy for the subsequent peptide-bond reacts with the terminal adenosine moiety of formation to occur [54]. 532 The translation of mRNA

On the basis of primary structure it is possible to divide the aminoacyl-tRNA syn• thetases into two distinct classes, which have no discernible relationship between one another. Furthermore, it is observed that the class I synthetases generally catalyse attach• ment of the amino acid to the 2'-position of the ribose, whereas the class II synthetases generally catalyse attachment to the 3'• position. The evolutionary implication of this - that there were seperate origins for the two classes of synthetase - although startling, has been supported by the finding that the three-dimensional structures of representatives of the two classes are quite distinct [321]. Despite the fact that multiple species of tRNA (isoaccepting tRNAs) exist for a single amino acid (section 12.2.2), there appears to be only one aminoacyl-tRNA Fig. 12.12 Comparison of the conformation of synthetase for each amino acid. Even the free yeast tRNAPhc with that of E. coli Gln• different initiating and elongating methionyl• tRNA Gin in a complex with glutaminyl-tRNA tRNAs are recognized by the same enzyme. synthetase. The s olid lines and open lines represent the path of the backbones of tRNA G in A fundamental question that arises is what and tRNAPhc, respectively (After [57], with features of these isoaccepting tRNAs are permission.) recognized by individual aminoacyl-tRNA synthetases to ensure that no misacylation occurs. Chemical and genetic manipulation illustrate this point. A G: U base-pair unique of tRNAs has now led to a clearer general to position 3:70 of the anticodon loop of understanding of this, and the determination tRNA Ala determines recognition by the of the X-ray crystal structure of tRNA• alanyl-tRNA synthetase. Mutation of this synthetase complexes has provided detailed causes loss of Ala-acceptor activity, and information. mutation of the c orresponding C:G base• In general, the studies with artificial tRNA pair of tRNA Cys to G: U conveys Ala• constructs [55] have identified two main acceptor activity to this latter tRNA. In the regions on tRNAs that act as determinants case of tRNA Me t, conversion of the for recognition by synthetases: the anticodon of the 3'-UAC-5' anticodon to its modified loop and the acceptor stem. Which of these derivative, lysidine (found in tRNA 11e), regions provides the major determinant conveys lie-acceptor activity on the tRNA; varies from tRNA to tRNA, and it has been whereas conversion of the 3' -CAU-5' anti• suggested that this may reflect the structural codon of the tRNA Val to 3'-UAC-5' conveys classes of aminoacyl-tRNA synthetase, men• upon it Met-acceptor activity. tioned above [56]. Although not all tRNAs The first tRNA-synthetase complex to show such clear-cut determinants, the cases have its structure determined was that for of tRNA Ala and tRNA Met will be used to E. coli tRNA G in , a class I enzyme [57]. Both The structure and aminoacylation of tRNA 533

Fig. 12.13 Stereoscopic view of the complex between E. coli glutaminyl-tRNA synthetase and Gln-tRNA. (From [57], with permission.) the central U of the anticodon of this tRNA tRNA does not lie in direct interaction with and its acceptor stem are determinants of the synthetase, the most striking feature of synthetase recognition. The striking feature which in this region are amino acid residues of the X-ray crystal structure is the distortion that disrupt the helix. However, it should be of both these features of the tRNA com• stressed that there are specific interactions pared with tRNAPhe (unfortunately the between the enzyme and the tRNA in these structure of tRNA Gin alone has not yet been regions. For example, the carboxyl group of determined) and the interaction of these Asp235 of the synthetase makes hydrogen regions of the tRNA with the synthetase bonding interactions with the the 2-amino (Figs 12.12 and 12.13). In the complex the group of the of the G3:C70 base• base pairing of the acceptor stem is actually pair, and the importance of this is indicated broken, explaining why only a weak G:U by the fact that mutations that convert this pair (or a non-pairing C:A) can be tolerated Asp to Asn or Gly cause misacylation [58]. in the 1:72 position. In addition, the obser• Furthermore, the anticodon loop, which is vation that this distorted structure is stabilized also involved in specifying the identity of the by a to the phosphate back• tRNA, makes multiple specific interactions bone made by the 2-amino group of the with amino acid residues in the protein unpaired residue G73 explains the require• [322). ment for a guanine residue in this position The three-dimensional structure of a com• of the acceptor stem. It is perhaps a little plex of a class II synthetase with its tRNA surprising that the major role of the deter• has also been determined, that of tRNAAsp mining residues of the acceptor stem in this [323). This complex also shows interaction 534 The translation of mRNA

between the synthetase and the acceptor 12.4 THE EVENTS ON THE stem and anticodon loop of the tRNA. BACTERIAL RIBOSOME [61-63] Although in this case there is no separation of the acceptor stem of tRNAAsp (this is not Bacterial protein synthesis will be described surprising as most of the identity deter• using the traditional two-site model of the minants are in the anticodon loop) the ribosome (cf. Fig. 12.1) as a convenient distortion of the anticodon loop found in visual aid. There are, however, several tRNA 01" is also observed, and in this case different proposals for additional sites there is a structure for tRNA Asp alone on · the ribosome which will be discussed (rather than that of tRNAPhc) as a basis for subsequently. direct comparison. The recognition of the correct amino acid by the aminoacyl-tRNA synthetase is 12.4.1 Chain initiation [64-66] equally as important for the fidelity of trans• lation as the recognition of the correct In polypeptide chain initiation fMet-tRNA is tRNA, just described. This is because if an bound to the initiation codon of the mRNA incorrect amino acid is enzymically attached on the 30S ribosomal subunit and the result• to the tRNA it will be misincorporated into ing 30S initiation complex then reacts with protein on the basis of the codon-anticodon the 50S ribosomal subunit to give a 70S interaction. (This was shown in the classic initiation complex. This process requires experiment in which Cys-tRNACys was GTP and the initiation factors, IF-1, IF-2 reduced with Raney nickel to give Ala• and IF-3, and in polycistronic mRNAs can tRNA cys, which then incorporated alanine occur independently at several different in response to codons for cysteine in a cell• initiation sites. free system [59]). There are several pairs of The exact sequence of events in initiation amino acids differing in structure by no more and the precise roles of all the factors is not than a single , and this poses a entirely certain, but that thought most likely real problem in discrimination for the syn• is presented in Fig. 12.14. The initiating 30S thetases. For example, it was observed that subunit most probably has bound to it IF-3 valine bound appreciably to Ile-tRNA syn• , and IF-1 (Fig. 12.14(a)) which are involved thetase. There is good evidence to support in generating free ribosomal subunits after the existence of a 'proofreading' or 'editing' polypeptide chain termination (section mechanism in those synthetases for which 12.4.3). Their role in initiation is distinct there are inappropriate isosteric or smaller from this latter as they are required for the amino acids with which the tRNA may be formation of a 30S initiation complex (Fig. mischarged. This involves a hydrolytic site 12.14(b)) even when 50S subunits are not on the enzyme, close to, but distinct from, present. IF-3 (Mr = 21 000) is primarily the acylation site, to discharge such inappro• involved in binding mRNA to the ribosome. priate aminoacyl-tRNAs [60]. The accuracy Although it is needed for translation of of recognition at the hydrolytic site can, of natural mRNAs, there is no absolute re• course, be no greater than that at the initial quirement for IF-3 in either the AUG• acylation site; but by requiring the amino dependent ribosome binding of fMet-tRNA acid to be recognized twice, an error fre• or the translation of artificial quency of (e.g.) 1 in 102 would be reduced such as AUGAn. This suggests that, either to 1 in 104• directly or indirectly, IF-3 facilitates the The events on the bacterial ribosome 535

fMe~-~RNA GTP X X tMet IF-2 I Met fMet (GTP (GTP ( IF-I ) IF-2 IF-1 IF-2 IF-1 3' AUG AUG AUG ( IF-3 IF-3 p A p 7s: A mRNA IF-3 IF -1 IF-2 (a) (b) (c) (d) Fig. 12.14 A schematic diagram of prokaryotic polypeptide chain initiation. The ribosome and tRNA are represented as in Fig. 12.1.

16S rR A 3' HoAUUCCUCCACUAG ...... f--7±2--1 Phage cpX 174 A 5' .... AAUCUUGGAGGCUUUUUUAOG .. Phage MS2 coat protein 5'. ... UCAACCGGAGUUUGAAGCAOG .. Phage~ cro 5' . .. . AUGUACUAAGGAGUUGUAOG . . . galE 5' . .. . CCUAAUGGAGCGAAUUAUG ... . ~lacmmase 5' .. .. AUUGAAAAAGGAAGAGUAUG .. . Ribosomal protein Sl2 5' .... AAACCAGGAGCUAUUUAAOG .. . RNA pol ymerase~ 5' ... . GCGAGCUGAGGAACCCUAOG .. . lad 5' .... CUUCAGGGUGGUGAAUGUG .. . .

Fig. 12.15 Some ribosome binding sites of E. coli and bacteriophage mRNAs. The key polypyrimidine region in the l6S rRNA and the bases complementary to this are shown with background shading, and the initiation codons are indicated in bold. recognition of the untranslated 'leader' nicely prove the rule, as these occur in sequences that precede the initiating AUGs mRNAs coding for extremely weakly ex• of natural mRNAs. Such bacterial 'leader' pressed proteins: the cl repressor protein of sequences contain short polypurine regions phage lambda, and the DNA primase (the complementary to a polypyrimidine sequence particular codon usage pattern of which was at the 3'-end of the 16S rRNA (Fig. 12.15); mentioned in section 12.2.4) and trp re• and it was suggested by Shine and Dalgarno pressor of E. coli. The polypurine region is [67] that base-pairing between these regions generally found at quite a specific distance is the means by which bacterial ribosomes 5' to the initiation codon (Fig. 12.5), as select the correct AUG codon for initiation. would be expected if its function is to bring There is now overwhelming evidence (re• the initiation codon into the P-site of the viewed in [68]) that this hypothesis is correct, ribosome. although other factors, such as secondary The primary role of IF-2 (Mr = 97 000) is structure, may prevent every potential to bind fMet-tRNA to the ribosome in a 'Shine and Dalgarno' sequence being used reaction that requires GTP and is stimulated for initiation. Three known exceptions in by the other initiation factors, especially which initiation codons lack such a poly• IF-1. The fMet-tRNA most probably binds purine region in their 'leader' sequence to the ribosome after the mRNA has done 536 The translation of mRNA

so, in contrast to the situation in eukaryotes It will be evident from the foregoing that (section 12.5.1), and most likely in the form although IF-1 (Mr = 8000) is absolutely of a ternary complex between IF-2, fMet• required for initiation, its role is not clearly tRNA and GTP, although such a complex is defined. The fact that it cycles on and off the experimentally much less stable than those ribosome during initiation established that it formed by evolutionally related tRNA• is indeed an , rather than binding factors. IF-2 must recognize some a loosely bound ribosomal protein such as specific structural feature of the initiator S1 (section 12.6). Although IF-1 seems tRNA as it will not interact with aminoacyl• especially to facilitate the action of IF-2, tRNAs, not even Met-tRNA. One such it is clear that it does not have a function feature may be the unique unpaired end of analogous to that of EF-Ts in elongation the acceptor stem, C1:A72, as mutation of (section 12.4.2). this to a C:G base-pair allowed the tRNA Lower-Mr subspecies of IF-2 and IF-3 to function in elongation [69]. Another, exist, lacking portions of the N-termini of possibly relevant, structural difference of the larger species. The smaller form of IF-2 the initiator tRNA from elongator tRNAs is (IF-2P), rather than being a proteolytic seen in the X-ray crystal structure, where fragment of the larger form (IF-2a), is the the anticodon loop has an external-facing product of a second initiation on the mRNA rather than internal-facing disposition [47], a [71]. These subspecies appear to be func• feature also conserved in eukaryotic initiator tionally equivalent to their parents. tRNA [46]. The crucial role of the Shine and Dalgarno Once a 30S initiation complex, containing sequence in the selection of the initiation mRNA and fMet-tRNA, has been formed codon helps explain the fact, mentioned in (Fig. 12.14(b)) the 50S ribosomal subunit section 12.2.1, that this latter is not always can associate with it (Fig. 12.14(c)) causing AUG. The most common variant, GUG, is the release of IF-3. The non-hydrolysable used at about 3-4% the frequency of AUG. analogue of GTP, 5' -guanylylmethylene There are, in addition, some examples of diphosphonate, has been used to show that UUG [68] and one of AUU [72] functioning hydrolysis of GTP is not required for this as initiation codons in natural bacterial step, but GTP hydrolysis is required for mRNAs. The other initiation codons are the fMet-tRNA to become available for translationally less efficient than AUG: reaction with puromycin. This reactivity relative activities of AUG > GUG > UUG with puromycin is used to define occupancy have been demonstrated [73]. However, the of the P-site, as puromycin is an analogue occurrence of GUG and UUG codons in of aminoacyl-tRNA, and hence binds to the highly expressed mRNAs confirms that this A site. After hydrolysis of GTP, IF-2 and effect is not large and may be compensated (probably) IF-1 are released. As GTP for by a strong Shine and Dalgarno sequence. hydrolysis does not cause relative movement In the case of the AUA initiation codon, of mRNA and the ribosome, fMet-tRNA which is only known to occur in the mRNA must be bound directly at the P-site [70]. for IF-3, the low intrinsic activity of this The role of the GTP hydrolysis, which is appears to be important for autoregulation discussed in more detail below (section by the cellular concentration of IF-3 itself. It 12.4.2), cannot, therefore, be the provision has been suggested that the translation of the of energy for movement of the fMet-tRNA mRNA for IF-3 is independent of IF-3, and from A-site to P-site. that a low intrinsic initiation activity because The events on the bacterial ribosome 537

X (a) fMe~

Ala-~RNA ( s' AUGGCCUCG-3' p EF-Tu~EF-Tu• • A GTP GTP • Ala-~RNA i X (b) fMe~ Ala EF-Ts () GTP 5' AUGGCCUCG- 3' p A EF-Tu • EF-Tu • EF-Ts GOP c

t GOP +pi (c)

t EF-G + G D P ~------.1.

(d)

Fig. 12.16 A schematic diagram of prokaryotic polypeptide chain elongation. For convenience the 70S initiation complex of Fig. 12.14( d) has been taken as the starting point, (a), although the scheme applies equally for 70S ribosomes bearing peptidyl-tRNA in the P-site (e.g. (d)). Likewise, the designation of the mRNA triplet in the A-site as coding for Ala, and the third triplet for Ser are purely arbitrary.

of the AU A codon will mean that the that the recogmtwn of the alternative in• mRNA cannot compete for ribosomal sub• itiation codons, GUG and UUG, involves units unless a low concentration of IF-3 'wobble' at the first, rather than the third, prevents other mRNAs from doing so [74]. position of the codon. It is noteworthy that, Consistent with this idea, mutation of the in contrast to elongator tRNAs (section AUA codon to AUG causes a 40-fold 12.2.2), the base 3' to the anticodon in increase in translation and abolishes the tRNAret is unmodified. Hence the base• autoregulation [75]. stacking that facilitates fidelity in the first Some comment is necessary on the fact two codon positions of elongator tRN As is 538 The translation of mRNA

decreased. It is interesting in this regard that an EF-Tu.EF-Ts complex, from which the in eukaryotic cytoplasmic tRNAret, which EF-Tu.GTP complex can be regenerated does not normally show such first position (Fig. 12.16). 'wobble' (section 12.5.1), this base is Analysis of its primary structure shows modified. that EF-Tu, like IF-2 and EF-G (below), belongs to the family of GTP-binding proteins that include the signal-transducing 12.4.2 Chain elongation [76-79] G-proteins [81]. The X-ray crystal structure of EF-Tu has been determined [82, 83], but, In polypeptide chain elongation an as yet the structure of the complex with aminoacyl-tRNA binds to the A-site of the aminoacyl-tRNA has not been solved, so ribosome and reacts with the peptidyl-tRNA that at present only the position at which (Fig. 12.16) or fMet-tRNA (Fig. 12.14) in GTP binds is known. There are two, ap• the P-site, accepting the growing poly• parently functionally equivalent forms of peptide chain. The tRNA is then moved EF-Tu in E. coli, the products of separate across to the P-site (translocated), with con• genes (tufA and tufB), differing only in their comitant movement of the mRNA and C-terminal amino acids. EF-Tu is extremely expulsion of the deacylated tRNA, in order abundant, constituting some 5% of total to make the A-site available for the next bacterial cell protein, and occurring in ap• aminoacyl-tRNA. Elongation requires three proximately sixfold excess over ribosomes soluble factors, EF-Tu, EF-Ts, EF-G, and and other elongation factors. The significance the hydrolysis of two molecules of GTP. of this is not known. EF-Tu (Mr = 43000 The aminoacyl-tRNA bound in the A-site [126]) is responsible for the ribosomal bind• (Fig. 12.16(b)) can now be linked to the ing of the aminoacyl-tRNA corresponding to carboxyl group of the fMet or nascent the mRNA codon in the A-site (arbitrarily peptide, through the catalytic activity of the designated as Ala in Fig. 12.16), before which intrinsic peptidyltransferase centre of the 50S it forms a soluble ternary complex with the ribosomal subunit. As already mentioned, tRNA and GTP. All elongator aminoacyl• the thermodynamic free energy for peptide tRNAs will form this complex, but fMet• bond formation comes from the hydrolysis tRNA will not [80]. The non-hydrolysable of the 'energy-rich' acyl-ester bond of the analogue of GTP, 5'-guanylylmethylene aminoacyl-tRNA, which in its turn is derived diphosphonate, will allow the aminoacyl• from the A TP hydrolysed during amino• tRNA to bind to the 70S ribosome, but the acylation. The fact that GTP and supernatant GTP must be hydrolysed before peptide• factors are not required for the trans• bond formation can occur. The GTP peptidation was perhaps most convincingly hydrolysis is not required for the peptidyl• confirmed when it was discovered that, in transferase reaction itself and its possible the presence of ethanol (about 50%), a role is discussed below. The EF-Tu and 3'-hexanucleotide fragment of fMet-tRNA GOP are released from the ribosome as a could react with puromycin on the isolated complex. In this form the EF-Tu cannot 50S ribosomal subunit [84]. Extension of this react with GTP or aminoacyl-tRNA, and it 'fragment reaction' to even smaller oligo• is the function of EF-Ts (Mr = 30000) nucleotide fragments has shown that CCA• to displace GOP from the EF-Tu.GOP fMet is the smallest species that can occupy complex. This results in the formation of the P-site of the peptidyltransferase, and The events on the bacterial ribosome 539

that puromycin can be replaced by CA-Gly generally to terminate the action of the at the A-site. protein by causing its dissociation from its The translocation of the peptidyl-tRNA target [81]. Against this context the idea has from the A-site to the P-site requires the emerged of a unitary role for GTP hydrolysis elongation factor EF-G (Mr = 77000) and in protein biosynthesis to expel the factors GTP. This reaction has been shown to (including IF-2) from the ribosome after allow movement of the peptidyl end of the they have fulfilled their functions. The pre• tRNA so that it becomes reactive towards sence of the factor on the ribosome can be puromycin, movement of the mRNA rela• regarded as preventing the next reaction tive to the ribosome, and ejection of the from occurring, and the role of EF-G deacylated-tRNA from the P-site. The reac• envisaged in this model is to prime the ribo• tion requires hydrolysis of the GTP, the some for subsequent translocation which non-hydrolysable analogue being inactive does not, in itself, require GTP. In the case although it will allow EF-G to bind to ribo• of EF-Tu, the time taken for the GTP to be somes. The molecular mechanism underlying hydrolysed after the ternary complex has this process is perhaps the most intriguing bound to the ribosome is important in the and the least understood aspect of protein kinetic 'proofreading' that ensures accuracy biosynthesis. It is clearly possible that a of the codon-anticodon interaction (section large part of the structural complexity of the 12.6.3). However, assertions that the func• ribosome, including even the division of tion of GTP hydrolysis is to provide the the ribosome into subunits, may be a con• energy for this proofreading are, in our view, sequence of the need for this specific and misleading. concerted movement of . As already stated, there are models of the After translocation (Fig. 12.16( d)), one ribosome containing additional sites, and cycle of elongation has been completed these have implications for models of the (cf. Fig. 12.16(a)). The vacant A-site now elongation cycle. Lake [85] proposed that 'contains a new mRNA codon, to which a before occupying the A-site proper, the corresponding aminoacyl-tRNA can bind, aminoacyl-tRNA entered through a recog• starting another round of elongation. nition or R-site, which would allow the anti• The function of GTP hydrolysis in the codon to be proofread for correspondence elongation reactions was unclear for many to the codon. Nierhaus (reviewed in [86]), years. The earliest ideas were influenced on the other hand has proposed a model of by the biochemical precedents of A TP the ribosome in which the deacylated tRNA hydrolysis and the GTP-utilizing phos• is first transferred to an exit or E-site (i.e. pheno/pyruvate carboxykinase, and focused after stage (c) in Fig. 12.16), before being on the provision of energy for movement expelled from the ribosome. More recently of molecules: the translocation of peptidyl• Moazed and Noller [87] obtained evidence tRNA to the P-site in the case of EF-G, and for the employment of an additional site the 'accommodation' of the aminoacyl• during translocation; although one rather tRNA into the A-site in the case of EF-Tu. different from that proposed by Nierhaus. However, as mentioned above, it is now Chemical footprinting indicated that the clear that the GTP-utilizing protein-synthesis translocation occurred in two discrete steps factors are members of a larger family of involving first the acceptor end and then the which includes the G-proteins, in anticodon end of the tRNA. To explain their which the function of GTP hydrolysis is results they proposed an extra E-site, solely 540 The translation of mRNA

E P A

a b

aa-tRNA• EF-Tu· 50S GTP .. 30S tEF-Tu·G state: P/P P/P A/T

c ~ d e ( a a peptidyl EF-G transfer .. GTP ..

P/P A/A P/E A/P E P/P

Fig. 12.17 Model for the movement of tRNA during translocation involving hybrid occupancy of A• or P-sites on the 30S subunit, and A-, P- or E-sites on the 50S subunit of the ribosome. In state A/T, no site on the 50S subunit is occupied. (From [87], with permission.)

on the 50S ribosomal subunit, with hybrid vented from doing so by IF-3 and IF-1 occupancy by the tRNA of various 50S and (section 12.4.1). 30S sites occurring during the elongation In contrast to the other 61 codons, the cycle (Fig. 12.17). It is interesting that there three specific terminators are not read by is structural similarity between the RNA in tRNAs. This was shown using RNA from the proposed E-site and that of part of the an amber mutant of bacteriophage R17 in RNA component of E. coli ribonuclease P which the first six codons of the coat protein (section 11.3.4(a)) involved in interaction are followed by a . Only the six with the 3r-end of precursor tRNAs [319]. appropriate aminoacyl-tRNAs and super• natant proteins were required for release of the hexapeptide. This same system was 12.4.3 Chain termination [88-90] subsequently used with purified elongation factors and release factors to show that In polypeptide chain termination (Fig. release of the peptide required the peptidyl• 12.18), the ester linkage of the peptidyl• tRNA to be at the P-site of the ribosome tRNA is hydrolysed in response to one of [91]. To study the factor requirements for three termination codons (section 12.2.1) in termination at all three codons, an assay was a reaction involving two of the three re• developed in which the termination codons lease factors, RF-1, RF-2 and RF-3. The could direct the release of fMet-tRNA, deacylated tRNA and the mRNA are ex• previously bound to ribosomes in the pre• pelled from the ribosome in the presence sence of the triplet AUG. This led to the of , RRF, and EF-G, liberating resolution of two release factors, RF-1 free ribosomal subunits. The subunits will (Mr = 36000) and RF-2 (Mr = 38000), of associate to form 70S ribosomes unless pre- different codon specificities: The events on the bacterial ribosome 541

705 (a) IF/ ( (d)----

GTP s' ~ ( lOS ) 50S (b) mRNA (c) + IF~ t RNA Arg ("30S"l ~ (e)

Fig. 12.18 A schematic diagram of prokaryotic polypeptide chain termination. The amino acid designation in the P-site is purely arbitrary. Other possible termination codons in the A-site are UAG and UGA (see text).

those of the known GTPases discussed in RF-1 for UAA or UAG section 12.4.2. There are quite strong grounds for think• RF-2 for UAA or UGA ing that the actual hydrolysis of the peptidyl ester linkage is catalysed by the peptidyl• The third release factor, RF-3 (Mr = transferase centre of the ribosome, the 46000), is not codon-specific and has no reaction specificity of which has been release activity in the absence of the other modified by the binding of the release factors. It enhances the release of poly• factors. This was suggested by the finding peptide promoted by the other factors and that the peptidyltransferase would catalyse seems to stimulate both binding and release the formation of an ester link to fMet-tRNA of these latter from the ribosome. Its activity or its hexanucleotide fragment (section is increased by GTP, but a requirement 12.4.2) if certain alcohols were presented to for GTP hydrolysis during termination in the ribosome instead of aminoacyl-tRNA. prokaryotes is not firmly established as it If the hydroxyl groups of an alcohol could has been reported that GOP can replace replace the a-amino group of an aminoacyl• GTP in the reaction in vitro. In this respect tRNA as a reactive nucleophile, it seemed it would be useful to know the primary struc• possible that the hydroxyl group of water ture of RF-3 to see whether it is related to might do likewise. This suggestion was sup- 542 The translation of mRNA

ported by the fact that a number of antibiotics these species, without altering the position (e.g. and ) of its equilibrium [94]. and ionic conditions known to inhibit the Although all three termination codons are peptidyltransferase reaction were also found found to occur in mRNAs, the distribution to inhibit the termination reaction in vitro. of these is not random. The codon UAG is More recently a new perspective has been highly disfavoured, and in a sample of c. 800 generated on the termination process in the E. coli termination codons the relative fre• form of a possible role for ribosomal RNA. quencies of occurrence, UAA:UAG:UGA, A ribosomal mutation that specifically sup• were approximately 10:4:1 [95, 96]. Further• pressed UGA termination codons was found more, the context of termination codons is to be the deletion of a cytosine (C1054) from a also not random, with a tendency for a highly conserved region of the 16S rRNA, following U being the most obvious feature and it was suggested that this mutation might of this. These features appear to provide prevent codon-specific base-pairing to a protection against readthrough by natural proximal 3'-ACU-5' sequence [92]. There suppressor tRNAs (section 12.9.6), rather is, as yet, no direct evidence for such an than contributing to the recognition of the interaction, nor is there any indirect evi• codons by termination factors [97, 98]. dence for analogous interactions involving the other two termination codons. After the release of the peptide, the 12.5 THE EVENTS ON THE mRNA and deacylated tRNA are still EUKARYOTIC RIBOSOME [99, 100] attached to the ribosome (Fig. 12.18(b)) and must be removed before subunits can be Eukaryotic ribosomes catalyse essentially regenerated for another round of protein the same process as prokaryotic ribosomes. synthesis. This requires GTP, EF-G and Although the details of eukaryotic protein ribosome release factor, RRF (Mr = 18000). synthesis are less well understood, it is clear Although it might be expected that the that the differences from prokaryotic protein primary role of EF-G in this process would synthesis are relatively minor for elongation be the expulsion of deacylated tRNA, and termination, but much greater for in• there appear to be no data bearing on this itiation. The following discussion relates to question. the nucleo-cytoplasmic protein-synthesizing The released ribosomes can be in the form system, the protein-synthesizing systems of of 70S particles or 30S and 50S subunits. The mitochondria and chloroplasts being dealt supply of isolated subunits for reinitiation with in section 12.7. is controlled by IF-3, acting as an anti• association factor, preventing the 50S subunit from associating with the 30S.IF-3 complex 12.5.1 Chain initiation [101, 102] [93]. Inactive 70S ribosomes do accumulate in cells, especially when inhibition of in• One way in which eukaryotic initiation itiation results in a relative excess of 30S differs from that in prokaryotes is in the subunits over IF-3. To regenerate subunits initiating amino acid: instead of fMet, the when conditions improve there must be an initiating amino acid is Met. Nevertheless, equilibrium between 70S ribosomes and there is a specific species of methionine ribosomal subunits. Initiation factor IF-1 is tRNA for initiation, distinct from that used thought to accelerate the interconversion of in elongation. The eukaryotic initiator The events on the eukaryotic ribosome 543

tRNA will be referred to here as tRNArct three subunits (a, p and y). The ternary (and the elongator as tRNA~e 1 ), although complex then binds to a 40S ribosomal sub• it is sometimes designated tRN Af"1ct. This unit bearing eiF-3 and eiF-4C, giving a 43S nomenclature emphasizes the unity in struc• preinitiation complex. One of the roles ture between the eukaryotic and prokaryotic of eiF-3 is as an anti-association factor, initiator tRNAs: despite the fact that there generating free 40S subunits; but it should is no transformylase in the eukaryotic cyto• be stressed that eiF-3 is an extremely com• plasm, the eukaryotic Met-tRNAr can be plicated species, consisting of at least seven formylated in vitro by E. coli transformylase antigenically discrete polypeptide chains [103]. with an aggregate Mr of approximately The more fundamental difference in 700000 [109]. eukaryotic initiation is the mode of selection The next step is a most distinctive feature of the initiating AUG, and it is clear that of eukaryotic initiation, the binding of the it is this aspect of initiation, although still 43S preinitiation complex to the 5' -end of incompletely understood, that accounts for the mRNA, involving melting of mRNA a plethora of eukaryotic initiation factors, secondary structure and recognition of the far exceeding the number in prokaryotes. 5'-cap structure. Four factors are involved at Instead of the 'Shine and Dalgarno' inter• this stage: the three factors eiF-4A, eiF-4E, action (eukaryotic 18S rRNA lacks the key and p220 (Mr = 220000), which together CCUCC sequence involved in this), allowing make up a complex known as eiF-4F, and a independent internal initiations on a poly• fourth factor, eiF-4B. The factor responsible cistronic mRNA, in eukaryotic initiation for promoting unwinding of the secondary there is a different mechanism: the Kozak structure of mRNA is eiF-4A, which is 'scanning mechanism' [68, 104, 105]. The an A TP-dependent . (At least in 40S ribosomal subunit binds to the 5' -end mouse, there are two distinct functional of a monocistronic mRNA and 'scans' genes for eiF-4A, but the significance of this along this until it encounters an appropriate is unclear [110].) The recognition of the (usually the first) AUG initiation codon, 5'-cap structure is the property eiF-4F. when attachment of the 60S ribosomal sub• Although all three components of the factor unit can occur. In contrast to the situation in are required for this, purification on 'cap' prokaryotes, the initiator tRNA binds to the affinity columns has shown that it is eiF-4E small ribosomal subunit before this attaches (Mr = 24000) that is the actual cap-binding to the mRNA [106, 107], and experiments protein. The factor eiF-4E must play a in which the 3'-UAC-5' anticodon of the pivotal role in the regulation of eukaryotic tRNAret was mutated to 3'-UCC-5' have protein biosynthesis, for transfected fibro• demonstrated that the anticodon is involved blasts over-expressing this factor exhibit in the scanning for the appropriate AUG malignant transformation [111]. The precise codon [108]. role of eiF-4B at this step of initiation has A model for the mechanism of eukaryotic yet to be defined. initiation is presented in Fig. 12.19. The first The 40S initiation complex, having bound stage is the formation of a (stable) ternary to the mRNA, scans for the correct AUG complex between Met-tRNAret, eiF-2 and codon, when the 60S subunit is incor• GTP. Although eiF-2 appears functionally porated, forming the 80S complex. Factor analogous to prokaryotic IF-2, the eukaryotic eiF-5 appears to be necessary to promote factor differs from the latter in comprising GTP hydrolysis and the release of the other 544 The translation of mRNA

Me~-~RNAt'eiF-2·GTP~---- Met"-~RNAt+ eiF-2 + GTP (a) r40S•e!F-3·e!F -4C

Mer

AUG (b) ~+()rum\ 7 NpppGJ eiF-4FeiF·4B { eJF·4EeiF·4A p220 nATP Mer nADP + nPi e1F-4F, -4B

5' m7 GpppNa AUG-- (c )

M" j 7 5' m GpppN c!:J AUG (d)

1 Mer 5' m7 GpppN ~ (e) {wseiF-5 eiF·2•GDP eiF-3,-4C

5' m7GpppN ( f ) ~p A Fig. 12.19 A schematic diagram of eukaryotic polypeptide chain initiation. factors before this can occur. The eiF-2 role has not been clearly defined and it is released at this stage is complexed to GOP not shown in Fig. 12.19. This protein is which must be displaced by a factor, eiF-2B interesting, however, as it contains a unique (also called eRF and GEF), before it can modified derivative of , , continue to function (Fig. 12.35). This re• which is essential for its biological activity action, which is analogous to that between [112]. EF-Tu and EF-Ts (Fig. 12.16), is subject to The key tenets of the original 'scanning regulation and is discussed in more detail model' of Kozak [104], presented above, are in section 12.9.5. Another factor, termed attachment of the ribosome to the 5' -end of eiF-40, has been implicated in the for• the mRNA and initiation at the first AUG mation of the first peptide bond, but its codon; and the corollary of these is that The events on the eukaryotic ribosome 545

NSP4 t------lt-----1 lntermediCte f } ond moture ~ ------H------t~--.....:.....-----i polypeptides } Reodthrough polyprotem AUG UGA UAG AUG t UGA '} 495 RNA s' 7mGppp lll•••••••••••:::::==::tl... ~----• IIII;Jpoly A 3 ( 11 703 nt) AUG UGA } 265 RNA III••••••IIIIII::JPolyA (4106 nt) ~ } Polyprote~ns

NSPl NSP2 NSP3 - t-----'"~----:----1t lInt ermediate t---~ H t----; ond moture t 6K El polypeptides H 1------1 E3 E2

Fig. 12.20 Translational strategy of Sinbis [113]. Translation of the 49S genomic RNA occurs only from the first AUG, a second internal initiation site being recognized only on the subgenomic 26S RNA, where it has now become the first AUG. The initial transcripts are polyproteins which are proteolytically processed to give the mature non-structural and structural proteins indicated. Suppression (section 12.9.6) of the first termination codon produces a larger polyprotein from the 49S RNA, the processing of which generates NSP4 (and perhaps more of the other non-structural proteins: indicated by broken lines).

there should be no polycistronic mRNAs realized that the initiation codons actually with internal initiation, in contrast to the used have a similar 'context', the consensus situation in prokaryotes. Certainly, in the being: vast majority of cases this holds true, and it is striking how many RNA viruses (e.g. GCCGCCRCCAUGG [ 113]) resort to strategies such as the pro• duction of polyproteins or the generation of nested mRNA subspecies that allow then to and it was therefore proposed that this be functionally monocistronic despite having context somehow influenced the selection of structurally polycistronic genomes (Fig. the initiation codon, with AUG codons in 12.20). It is true that there are some minor weak contexts being bypassed. Not all the deviations from the original predictions, consensus nucleotides exert equal influence, but these can be accommodated by slight but the presence of either a purine at refinement of the scanning model [105). position '-3' or a Gat position '+4' to the However, there are also some rare major start of the AUG seem to be particularly violations of the predictions that can only be· important. The majority of those 5-10% explained by the existence of an alternative of mRNAs that did not initiate at the first mechanism of eukaryotic initiation [114). AUG were found to initiate at the first AUG We shall consider each of these situations occurring in a strong context. Moreover, in turn. some rare cases where two consecutive In approximately 700 vertebrate mRNAs AUG codons are used as alternative in• analysed, the most 5'-terminal AUG is the itiation codons are explicable in terms of initiation codon in 90-95% of cases . It was this context effect if it is assumed that a 546 The translation of mRNA

proportion of 43S initiation complexes ignore example of how deviations from the optimal the first AUG because of a relatively poor situation for operation of the scanning context which, nevertheless, does not deviate mechanism are used to regulate translation. totally from the consensus. The mode of In the instances cited above there is no action of the consensus sequence is unknown, question that the basic mechanism of 'scan• although the occurrence of purines with a ning' from the 5'-end of the mRNA still periodicity of three suggests that some kind holds. However, there is one case, that of 'phasing' of the reading-frame may be of the picornaviruses, where an entirely occurring. different mechanism has to be invoked. There are two other refinements that have Picornaviruses, such as polio, are peculiar to be made to the 'scanning' model in order in having uncapped mRNAs which do not to account for cases where initiation does not require eiF-4F for their translation; and occur at the first AUG codon in a strong indeed this is fundamental to the manner context. One is when the first AUG occurs in which they subvert the protein synthetic within 10 nucleotides of the 'cap', in which machinery of the cell (section 12.9.5). case it is assumed to be inaccessible to Although this does not itself preclude scan• the 43S initiation complex. The second is ning from the 5'-end of the mRNA, this where the AUG codon from which the appeared unlikely because of the long 5'• protein initiates is preceded by a short open leader with multiple AUG codons, some in reading-frame. It appears that the first open strong contexts, preceding the actual in• reading-frame is translated, and that, after itiation codon of the polyprotein that polio termination, continued scanning and re• encodes. There now seems no doubt that initiation can occur [115]. The explanation there is internal initiation in this case, as of this phenomenon, which is restricted to it has been shown that a 450-nucleotide short open reading-frames culminating in a segment of this leader, not including the termination codon, is unclear. Such preced• extreme 5' -end, can allow translation of ing short open reading-frames have the the erstwhile silent second member of effect of decreasing translation of the laboratory-generated bicistronic mRNAs authentic initiation reading-frame, and when inserted in front of the latter [114]. Nature appears to have used this as a device The secondary structure of the internal entry to repress the translation of certain proteins. site provided by this 450 nucleotide segment The most studied example of this is the seems to be important, and there is some mRNA for the GCN4 regulatory protein of evidence that this may be recognized by a yeast, where four such short open reading• cellular protein [120], but otherwise the frames are found preceding the initiation details of the mechanism of such internal codon [116]. initiation are obscure. It is clear, however, There is ample experimental evidence that a completely distinct mechanism exists that mutation of AUG initiation codons for the complex and fundamental process of markedly decreases translation (reviewed in initiation of protein synthesis, raising the [105]), so that the first reports of initiation question whether this could have evolved at non-AUG codons generated considerable in picornaviruses. It is perhaps easier to surprise. These, the best cellular examples of envisage the virus acquiring a mechanism which involve CUG codons, appear to be for of initiation evolved by a few host mRNAs very weakly expressed proteins [117-119], to enable them to operate the type of and therefore seem to represent a further described in section The events on the eukaryotic ribosome 547

12.9.5, and the existence of a functional One surpnsmg feature of eukaryotic internal initiation sequence has recently elongation is a species difference: yeast and been demonstrated for one host mRNA some fungi have a third elongation factor, [121]. EF-3, not required by a range of other eukaryotes [122]. Although yeast EF-3 has been purified and its gene cloned, its precise 12.5.2 Chain elongation [122] function is as yet unclear.

The eukaryotic factors required for binding aminoacyl-tRNA to the ribosome during the 12.5.3 Chain termination elongation cycle are analogous to EF-Tu and EF-Ts of prokaryotes. EF-1a, which is In eukaryotic termination there is a single structurally related to EF-Tu [81], forms a factor, RF, to recognize all three termin• ternary complex with aminoacyl-tRNA and ation codons, VAA, VAG and VGA, and GTP; and a second activity, EF-1py (or hence this is functionally equivalent to both perhaps EF-1Pyb), functionally analogous RF-1 and RF-2 of prokaryotes [129]. A to the monomeric EF-Ts, promotes release second eukaryotic release factor, equivalent from EF-1a of GOP, and exchange for GTP. to bacterial RF-3, has also been reported Like prokaryotic EF-Tu, eukaryotic EF-1a and warrants further study in view of the fact is an extremely abundant cellular protein that the eukaryotic termination reaction [123], and is encoded by two separate genes shows a clear requirement for GTP hydrolysis [124]. in vitro. The amino acid sequence of The elongation factor involved in the eukaryotic RF shows no similarity to those translocation reaction in eukaryotes, EF-2, of prokaryotic RF-1 and RF-2 (which are is structurally and functionally related to quite similar to one another [130]), al• EF-G in prokaryotes. One point of interest though, intriguingly, it contains a region is the specific inactivation of EF-2 (but not homologous to one in the tryptophanyl• EF-G) by bacterial [125], tRNA synthetases [131]. which transfers ADP-ribose from NAD+ to Although one imagines that there is a an unusual modified residue [126] eukaryotic factor analogous to RRF, none in EF-2. Although there has been a report has so far been described. Eukaryotic ribo• of an enzyme with an analogous activity to somes are liberated from polysomes as sub• that of diphtheria toxin in eukaryotic cells units, and a pool of inactive 80S monomers [127], it is still uncertain whether ADP• is present in eukaryotic cells [132]. In ribosylation of EF-2 represents a normal contrast to prokaryotes there is both an anti• mechanism of cellular regulation. association activity, eiF-3 (section 12.5.1), Eukaryotic, like prokaryotic, ribosomes which binds to 40S subunits, and an 80S possess an intrinsic peptidyltransferase subunit dissociation factor, eiF-6 [101]. activity, which has also been studied using The usage of the three termination codons the 'fragment' reaction (section 12.4.2) in eukaryotes is much more random than in [128]. Although the eukaryotic peptidyl• prokaryotes (section 12.4.3), although VAG transferase is inhibited by certain antibiotics is still the codon least frequently used. (e.g. sparsomycin) that inhibit prokaryotic In vertebrates the relative occurrence of peptidyltransferase, it is resistant to the VGA: VAA: VGA is approximately2.5:1.7: 1 action of others (e.g. chloramphenicol). [133]. 548 The translation of mRNA

12.6 THE RIBOSOME

The focus of this section will be E. coli, the ribosomes of which have been the subject of extensive study [134-138). Further information about eukaryotic ribosomes can be found elsewhere [139, 140) .

12.6.1 The structure of the ribosome

(a) Overall features

The overall size and shape of ribosomes have been analysed by several different tech• 10nm niques. At the most basic level w:;~s the physical separation of the ribosomes into 2 Fig. 12.21 Model of the 70S ribosome of E. coli subunits (by removal of the Mg + that holds based on electron-microscopic studies by Lake them together) and the determination of [141]. The 30S subunit is light, and the 50S their sedimentation coefficients in the ultra• subunit is dark. The cleft (C) and platform (P) of centrifuge: 30S and 50S in the case of the 70S the 30S subunit, and the stalk (S) and central bacterial ribosome, and 40S and 60S in protuberance of the 50S subunit, referred to in the case of the 80S eukaryotic ribosome the text, are indicated. of mammals. (The sizes of the ribosomes of other eukaryotes may differ from this.) diffraction methods. Three-dimensional Initial small-angle X-ray scattering studies crystals of ribosomes which diffract X-rays provided values for the overall dimensions of have been obtained, but these have not yet the ribosomes and their subunits, but their yielded sufficient resolution to be useful. gross features have been deduced from elec• However, significant progress has been tron microscopy of individual negatively made with three-dimensional image recon• stained particles. Although inherent prob• struction of the electron diffraction patterns lems in this latter method limit the detail obtained with two-dimensional crystalline that can be obtained, useful models have sheets of ribosomes [143). One striking been derived from such work, and the one feature revealed by this technique is a presented in Fig. 12.21 is generally used as a 'tunnel' through the centre of the 50S ribo• reference for studies on the organization of somal subunit (Fig. 12.22), and a similar the ribosomal components [141]. Its most feature has been observed in independent obvious features are a 'cleft' and 'platform' studies of eukaryotic 60S ribosomal subunits on the 30S subunit, and an elongated 'stalk' [144). This tunnel is large enough to accom• protruding to one side of the 'central pro• modate a nascent peptide of the length tuberance' in the 50S subunit. (25-40 amino acid residues) that the ribo• Refinements have been made in the some is known to protect against proteolytic methodology for interpreting such electron digestion [145], and it is most likely that this micrographs [142], but the way to more sub• feature represents an exit channel for the stantial advances would seem to lie through nascent peptide. The ribosome 549

Fig. 12.22 Computer graphic representation of a reconstructed model of the 50S ribosomal subunit of Bacillus stearothermophilus obtained from two-dimensional sheets, negatively stained with gold thioglucose (courtesy of Dr A. Yonath).

What of the physical location of other could represent 'bends' in the mRNA, and species interacting with the ribosome? The have been interpreted in terms of the mRNA 'length' ofthe tRNA (75 A -7.5nm) is such 'looping' around the ribosome to re-emerge that it could be accommodated in the space near its point of entry [148]. between the two ribosomal subunits; but this is purely speculation. Immune electron (b) Components microscopy (see below) suggests that puromycin, and hence the acceptor stem of The composition of the ribosome is approxi• aminoacyl-tRNA, may interact with the 50S mately 60% RNA and 40% protein. The subunit near the base of the 'central pro• smaller ribosomal subunit contains a single tuberance' [146). Related experiments species of RNA (16S or 18S RNA for 30S suggest that the anticodon of the tRNA and 40S subunits, respectively) together with interacts with the mRNA at the 'platform' ribosomal proteins; the larger ribosomal of the 30S subunit [147). It has long been subunit contains a major species of RNA known from nuclease protection experi• (23S or 28S RNA for 50S and 60S subunits, ments that 35-50 nucleotides of mRNA (i.e. respectively), an additional small 5S RNA, 12-17 codons) are in contact with the ribo• together with a number of proteins exceed• some. A more recent sophisticated analysis ing that in the small subunit (Fig. 12.23). revealed two internal cleavage sites that In eukaryotes the portion of the primary 550 The translation of mRNA

235 RNA (2904 nt) + 34 proteins 55 RNA ( 120 nt) (a)

165 RNA 6 {2·75•10 Do) G { (1542 nt) + 21 proteins (0. 95 x 106 Do)

!Ribosome I Isubunits I IProtein I

285 + 5.85 RNA (4718 + 158 nt) + c. 45 proteins 55 RNA (b) B (121 nt)

185 RNA {4.5•106 Do) G { (1874 nt) + c. 30 proteins ( 1.5 x 106 Do)

Fig. 12.23 A schematic diagram of the components of the ribosomes of (a) E. coli (b) rat. The molecular masses of the particles are the mean of physical determinations. The chemical values for the M. of 70S, 50S and 30S ribosomal particles from E. coli, based solely on RNA and protein content, are 2.3 x 106, 1.45 x 106 and 0.85 x 106• The discrepancy between the two sets of values can be accounted for by the presence of metal ions and spermidine.

transcript that gives rise to the major species (149, 150). In the case of E. coli, 10 base of rRNA in the 60S subunit undergoes have been identified in both further nucleolytic cleavage after it has 16S and 23S rRNA; and the 23S rRNA also started adopting its secondary structure contains three pseudouridine residues and (section 8.4.3). This results in a 5.8S RNA a few ribothymidine residues and ribose species hydrogen-bonded to the 28S RNA methylations. In mammals there are 46 and in the case of mammals. Similar 5' -post• 71 methylated groups on the 18S and 28S transcriptional processing gives rise to a 2S RNA, respectively, most of which involve RNA species in Drosophila and a 4.5S RNA the 2'-0 of the ribose moiety. They are species in plant chloroplasts (sections 9.4.2, predominantly clustered in the 5'-half of 9.4.3 and 9.7.2). As discussed further below, 18S rRNA and the 3'-half of 28S rRNA from a functional point of view these (151). Eukaryotic rRNAs also contain are better considered to be integral parts of pseudouridine residues: approximately 37 the larger rRNA species. and 60 for 18S and 28S rRNA, respectively. The ribosomal RNAs contain a small Neither eukaryotic nor prokaryotic 5S number of specific modified (largely rRNA contains modified nucleotides. methylated) nucleotides (section 11.6.3) The ribosomes of eukaryotes, prokaryotes, A

3' (a) 125 RNA (Human mitochondrion)

3''

(b) 165 RNA (E.coli)

B

(c) 18 5 RNA ( X.laevis)

Fig. 12.24 Comparison of the secondary structures of the smaller ribosomal RNA from (a) a human mitochondrion, (b) a prokaryote and (c) a . Each structure has been separated into two parts for clarity of presentation. These early proposals have undergone subsequent detailed refinement, but without in any way invalidating the comparative approach. (After [152], with permission.) 552 The translation of mRNA

mitochondria and chloroplasts catalyse which can be done using certain nucleases, similar reactions. It seems reasonable, there• chemical reagents and oligonucleotide fore, to assume that their RNAs serve probes. A particularly extensive study used similar functions, whatever these might be. chemical probes for all four bases, and In comparing rRNAs that diverge markedly identified the resulting modified exposed in size (from c. 600 to 2000 nucleotides for bases by their property of terminating small subunit rRNAs, and from c. 1200 to reverse transcription, primed by a battery 4700 nucleotides for large 'subunit rRNAs of oligonucleotides complementary to rRNA [149]) attention has therefore centred on [154]. The other, more laborious approach, common structural features that may be was identification of double-stranded regions, crucial to such functions. The most evident which involved RNA-RNA cross-linking similarity lies in the secondary structure, followed by isolation and identification of in which a common core is present which base-paired fragments. Such cross-links have becomes progressively more elaborated as been important in building models for the the rRNA increases in size (Fig. 12.24). It is tertiary structure of the rRNAs [155, 156], such comparison of secondary structures that one of which is shown in Fig. 12.25. Models makes it apparent that the 5.8S rRNA of for the secondary and tertiary structure of 5S mammalian ribosomes, when hydrogen• rRNA have also been proposed [157]. bonded to the 28S rRNA, is equivalent to a One feature of the hairpin loops in the structural feature present in E. coli 23S rRNA secondary structure (section 2.6) is rRNA [153]. At corresponding regions in the worth reiterating. This is the frequency with secondary structures of rRNAs from differ• which the loops are found to consist of ent species (predominantly in single-stranded the tetranucleotides, GNRA or UNCG. regions) .it is possible to identify a small Analysis of these has revealed particular number of relatively short, highly conserved, structures in which there is a base-pair regions of primary structure. The significance between the first and fourth nucleotide of these will be discussed in section 12.6.2. (G:A and U:G, respectively) [317]. The The initial models of a common secondary resulting stability imparted to the loops by structure for the rRNAs were primarily these sequences explains their frequent based on comparisons of the sequences of occurrence. rRNAs of different phylogenetic origins. Originally the ribosomal proteins of the Evidence was obtained from situations ribosomes of E. coli were enumerated where there is poor conservation of primary S1-S21 and Ll-L34 for the small and large structure in a feature of proposed secondary subunits, respectively. It is still thought that structure, which is maintained because of the 30S ribosomal subunit has these 21 compensatory base changes (e.g. an A~ G distinct proteins, one copy per 30S subunit. change in one strand of the RNA is com• However, there has been subsequent revision pensated for by a U ~ C change in the of the status of the proteins of the 50S sub• complementary base of the other strand). unit. It emerged that the species designated Experimental confirmation and refinement L8 was, in fact, a complex of L7/L12 and of the models was then necessary, and was of LlO, and that L26 is identical to S20, there especial importance where there was either being, on average, 0.2 copies of L26 and extreme conservation or divergence of 0.8 copies of S20 per 70S ribosome [158]. primary structure. One approach was to Furthermore, two additional proteins, A and identify regions of single-stranded RNA, B, were detected on the ribosome, and as The ribosome 553

plete amino acid sequences of all the 3' Domain ribosomal proteins of E. coli have been determined. The structures of a couple of ribosomal proteins have been determined by X-ray crystallography and, on the basis of these, possible regions for interaction with rRNA have been proposed (160, 161]. Amino acid sequences are at present avail• able for approximately half the eukaryotic 51 Domain ribosomal proteins, but so far only eight of these have been shown to be statistically significantly related to the sequences of proteins of E. coli (to S3, S5, SlO, S14, S16, S17, S19 and L6 of the latter). In addition, there are grounds for thinking that certain ribosomal proteins of eukaryotes are func• Central tionally homologous to L7/L12 of E. coli. Domain These include the size of these proteins, their acidic nature and their occurrence in multiple copies. There is also an indirect primary structure link through mutual re• latedness to a corresponding archaebacterial (section 12.7.1) ribosomal protein [162]. Two forms of this acidic ribosomal protein, with similar but distinct primary structures, Fig. 12.25 Model proposed by Noller and co• occur in eukaryotes. A third related, but workers for the folding of E. coli 16S rRNA. The 3' -domain is at the top of the figure and the larger, protein may be functionally equi• 5'-domain at the right. (From [155), with valent to LlO (section 12.6.2) [163]. All permission.) these eukaryotic acidic proteins are phos• phorylated to various extents; a modification these are chemically basic and have genes found in another eukaryotic ribosomal that map in clusters of other ribosomal protein (section 12.9.4) but not m pro• proteins (section 12.9.2) they have now been karyotic ribosomal proteins (164]. redesignated as ribosomal proteins L35 and L36 (159] . Thus it is best regarded that there (c) Organization are 34 distinct proteins on the 50S subunit. All but two of these proteins are completely We now turn to the question of the organ• dissimilar and are present as single copies. ization of RNA and ribosomal proteins in The exceptions are L7 and L12, L7 being the the ribosome. The relative positions of ribo• N-acetylated form of Ll2. These two somal proteins to one another have been proteins together are present in a total of determined by cross-linking with cleavable four copies per 50S subunit. Apart from bifunctional reagents [165] and by neutron proteins S1, S6, L7 and L12, the ribosomal scattering [166]. In the latter, extremely proteins are all chemically basic, with values powerful, technique, ribosomes are recon• of Mr in the range 9000-35 000. The com- stituted from their components, but with 554 The translation of mRNA

between the results obtained by these dif• ferent techniques. To integrate models such as that in Fig. 12.26 with the three-dimensional models for the structure of rRNA both direct and indirect approaches have been used. Studies of the direct interactions between proteins and RNA have shown that a subset of proteins is involved in the primary inter• action with RNA during assembly in vitro, and probably also in vivo. For 16S RNA and 5S RNA these proteins are fairly well defined as S4, S7, S8, S15, S17 and S20; and L5, L18 and L25 , respectively. In the case of 23S RNA an original list of 10 proteins (Ll, L2, L3, L4, L6, L13, L16, L20, L23 and L24) was subsequently extended to about half the proteins of the 50S ribosomal subunit. This may reflect a greater complexity of structure, but probably also illustrates the fact that most proteins in the assembled ribosome are in contact with RNA. Thus, using chemical cross-linking reagents [156], it was possible to demonstrate interaction Fig. 12.26 The relative positions of the proteins with RNA for proteins other than those of the 30S ribosomal subunit of E. coli derived from the neutron scattering studies of Moore and involved in the assembly of ribosomes co-workers. (From [166], with permission.) [134, 166]. This approach has provided more precise information than obtained previously by nuclease-protection experi• two of their proteins replaced by ones of a ments. The indirect approach to determining neutron density that contrasts with that of the relative positions of RNA and protein the rest of the particle. This is achieved by has been to compare their electron micro• growing E. coli in media with appropriately scopic locations on the ribosome. Initially different proportions of 0 20. The results electron-microscopic visualization was only of such measurements for the proteins of possible for regions of rRNA against which the 30S ribosomal subunit are shown in antisera could be raised, such as the 5'- and Fig. 12.26. A more indirect method is 3' -ends and certain methylated bases. 'immune electron-microscopy', although this However, after the secondary structure of has the advantage that it relates the positions rRNA had been established it became of the proteins to the gross features of the possible to use biotinylated oligonucleotides ribosomes. This method involves electron complementary to single-stranded regions, microscopic examination of ribosomal and to visualize these after cross-linking particles cross-linked by specific antibodies with avidin [168]. to individual proteins [141, 167]. It is re• On the basis of all these different types of assuring that there is broad agreement information, models of the relative positions The ribosome 555

proteins deleted have been found to be viable [171). Although it is not yet possible to describe the operation of the ribosome in terms of its constituent parts, it has been possible to identify components of import• ance to different aspects of ribosomal func• tion. These are present in different domains on the ribosome, and we shall concentrate our discussion on three of these, with emphasis on RNA, rather than on proteins. This is because the experimental evidence is increasingly in line with the evolutionary speculation [172, 173) that RNA has the primary functional role in ribosomes.

(a) The elongation factor domains

There is a domain on the 50S subunit in• volved in the action of EF-G and its intrinsic GTPase activity. It is associated with the 'stalk' of the subunit (Fig. 12.21), which is made up of the elongated L7/L12 tetramer, at the base of which is ribosomal protein LlO, and in the vicinity of which is ribosomal Fig. 12.27 Computer-generated model of the protein Lll. Proteins L7/Ll2 are essential for 30S ribosomal subunit of E. coli proposed by the binding and GTPase activity of EF-G, Brimacombe and co-workers. Helical regions of the rRNA are represented as cylinders and the and Ll1, although not functionally indis• proteins as spheres, as in Fig. 12.26. (From [156], pensable, is labelled by affinity analogues with permission.) of GOP, and has an rRNA binding site ( nucleotides 1052-1112) that overlaps the of RNA and protein in the 30S subunit have region (1055-1081) to which EF-G can be been proposed [155, 156), one of which is cross-linked (Fig. 12.28( a)). Strong evidence shown in Fig. 12.27. that this region of 23S RNA is functionally involved in the action of EF-G comes from studies involving the antibiotic thiostrepton, 12.6.2 Ribosomal structure-function a specific inhibitor of the action of this relationships [169, 170] factor. The natural producer of thiostrepton, Streptomyces azureus, unlike E. coli, is What is the relationship between the struc• methylated at nucleotide A 1067 • Mutation of ture of the ribosome already described this same nucleotide from A to a pyrimidine, and its function in catalysing protein bio• either under antibiotic selection or by site• synthesis? It is evident that some of the directed mutagenesis, conveys resistance to components - proteins and portions of the thiostrepton. Furthermore, this region of RNA - play only structural roles; moreover rRNA (in Domain II of Noller [155]) con• bacterial ribosomes with certain individual tains highly conserved portions of primary 556 The translation of mRNA

!------~~~~-~ : r------l~ ~~ ~ ~ 1 110~0G lOBO IU A I * : : A c A G c c Au c l uLu 1100'/!~ : ,-cx-sarcin 1 1 A I I I I I I I I .--, \ A ~ I I I G r:--;\ U C G G U U G U A G i CG : CG % A ' G I I £0 I r .!3r c A c u I I I e 1060 I A-U I U G I L_____ ------1 C- G : Jf.YG A .. A C I 1050 C-G 1 Uc-GC L ______------';(;1• U I C-G"'-2680 I A C I C 26SQ-U-A G : G-1110 : C-G A I I G·U C C C.!._G A I U·G EF-Gcross-link L----- __ J fragment G • U C-G G-C 1040-A G A G G-C '----~

(a) (b)

Fig. 12.28 The proposed elongation factor domains of the 50S ribosomal subunit of E. coli (after [169] and (155], with permission). (a) Part of Domain II of Noller: e nucleotide which is methylated in Streptomyces azureus and confers resistance to thiostrepton; (b) part of Domain II of Noller: *, base protected from chemical modification by EF-Tu; A, base protected from chemical modification by EF-G. The boxed areas are highly conserved in eubacterial and organelle ribosomes. structure, consistent with a functional role. that these nucleotides are important for There is a body of evidence suggesting triggering the GTPase activities of the two that EF-G and EF-Tu interact with the ri• elongation factors to similar effect. In the bosome in the same region. However, it case of EF-G this would also need to involve appears that EF-Tu does not bind to the the other domain, even though the two region just described, consistent with the regions of 23S rRNA are far apart in both fact that its action is not inhibited by primary and secondary structure. thiostrepton. It has been shown by RNA In focusing on the domains on the ribo• 'footprinting' with chemical probes that some associated with the action of EF-G, there is a distinct region of the ribosome one should not neglect the dynamic aspects where there is overlap between the binding of the translocation of peptidyl-tRNA from of EF-G and EF-Tu, and this involves the the A- to the P-site. This may involve conserved loop in the region of nucleotide changes in the conformation of the ribosome 2600 (in Domain VI of Noller (155]). More• as a result of events on the ribosome. It is over this loop (Fig. 12.28(b)) is known to possible that those changes of conformation be functionally important because it is the that have been detected (by distal changes target of the cytotoxins a-sarcin (which in protected nucleotides) may involve cleaves between 0 2661 and A2662) and ricin 'switches' between alternative rRNA (which removes A2660). One possibility is secondary structures. The ribosome 557

mA- U A 2010--A- U C-2440 lmG- c UG -C C-G C -G G

2050 A G A C \A C : A • • • C C C G C G GC A G A G G A* I I I • I I I I I I I I I •e • G G G U GC C G U C U C C*~ I AU fu)-2610 2620 ~ c u \ 2500 U \ G A 2600 A G \ A U p-Azldopuromyc!n "A c, 'c \ (A-s! tel , G \ G 2590 A \ U G C BP-Phe-tRNA U G (A and P-sltesl G- C-2510 A-U G C G- C GAU-A G • U /· 2570 •

Fig. 12.29 The proposed peptidyltransferase domain of the 50S ribosomal subunit of E. coli (after [174], with permission).*· Site of base change conferring resistance to erythromycin; A, site of base change conferring resistance to chloramphenicol; +. site of base change conferring resistance to anisomycin; e nucleotide which is methylated in Streptomyces erythraeus and confers resistance to erythromycin; BP-Phe-tRNA, 3-(4'-benzoylphenyl}propionyi-Phe-tRNA. The boxed nucleotides are highly conserved between different organisms.

(b) The peptidyltransferase domain in the peptidyltransferase centre by affinity labelling with antibiotic derivatives, none The peptidyltransferase catalytic activity is of them satisfied the criteria for catalytic the most fundamental functional feature of activity. Although, likewise, it has not been the ribosome, and, before the discovery of possible to demonstrate catalytic activity ribozymes (section 11.3), it was assumed associated solely with the rRNA, attention that it must reside in an individual ribosomal became focused on the latter because it protein. The major tools in probing this was discovered to be the site of point activity have been the drugs that specifically mutations that confer certain mitochondrial inhibit it; but, although several proteins ribosomes with resistance to drugs such as (especially L2, L16 and L27) were implicated chloramphenicol or erythromycin. These 558 The translation of mRNA mutations fall in a non-hydrogen-bonded 12.30). This region contains a nucleotide loop (in Domain V of Noller [155]) that (C1400) that can be cross-linked to the modi• forms a junction from which several hairpin fied base, cmo5U, at the wobble position of stems emanate (Fig. 12.29), and many tRNA Val when this is bound to the P-site nucleotides in this junction are conserved. of E. coli or yeast ribosomes [177]. Further• Furthermore, affinity-labelling and 'foot• more, resistance to certain antibiotics that printing' experiments with aminoacyl- or cause misreading was associated with peptidyl-tRNAs or their derivatives allow or mutation at positions 1405, assignment of parts of the A- and P-sites to 1408 and 1409 [178, 179]. Site-directed conserved nucleotides in this central loop mutagenesis suggests that C1400 is important [155, 174, 175]. The key role of the 3'• for ribosome assembly, but that the adjacent terminal CCA of tRNAs in the peptidyl• base (G 1401 ) is necessary for function [180]. transferase 'fragment' reaction (section Also in this general area is the highly 12.4.2) suggests that this might base-pair conserved sequence m~Am~A (nucleotides to complementary RNA sequences at the 1518-1519). The functional importance of A-and P-sites of the peptidyltransferase this was suggested by the characteristics of centre. There are, however, no totally con• an E. coli mutant resistant to the antibiotic served 3'-UGG-5' sequences in the 'central kasugamycin, which inhibits initiation and loop', and the results of footprinting experi• causes misreading. The ribosomal RNA ments with tRNAs lacking increasing from this mutant was unmethylated at these portions of the CCA sequence suggest that positions [181]. any nucleotides that base-pair with this It should be emphasized, however, that region would be physically separated from not all antibiotics that cause misreading one another in the primary structure and interact with this domain in the 30S subunit. require to be brought together in the tertiary The most thoroughly studied of these is structure [176]. Models for the tertiary struc• streptomycin, mutations conferring resist• ture of 23S rRNA are starting to emerge ance to which occur in both the 900 region (e.g. [318]) and should help clarify this (C912) and the 530 region (A523 and C525) of question. 16S RNA, the latter involving a pseudoknot [320]. The role of these latter regions of 16S (c) The decoding domain rRNA may relate to the participation of other species in the control of ribosomal This is the domain on the 30S subunit where accuracy, as is described in the next section. the codon-anticodon interaction occurs, and this is close to the region of the 'Shine 12.6.3 Ribosomal optimization of and Dalgarno' interaction between the translational accuracy [182] mRNA and the 3'-end of the 16S RNA (section 12.4.1), which may be modulated by We have already seen how a mechanism ribosomal protein S21. Although the 'Shine exists to minimize misreading of the genetic and Dalgarno' interaction is specific to pro• message as a result of mischarging of tRNA karyotes (section 12.4 .1), the decoding (section 12.3.2). Misreading can also occur domain has features that are conserved in as the result of incorrect codon-anticodon prokaryotes and eukaryotes, especially the recognition, and reference has been made in non-base-paired region from nucleotides section 12.6.2(c), above, to the effects of cer• 1392-1407 of the 16S RNA of E. coli (Fig. tain antibiotics on this. Although some mu- The ribosome 559

G u A c Peptl dy I tRNA A anticodon c aff!nl ty label

r --- 1 c 1-1400 " -- j G mC c c e G u c 14,0 1420 1430 A'fl I A/ PoJypyrlmldlne e CACCAUGGGAGUGGGUUGCAAAG··· 1111 region ••I •11••1••11111 GUGG 6 GUCA 6 UACUUAGUGUUUC··· binding mRNA A •I I I A 1490 1480 1470

Proteins S1 ,S21 cross-link

Fig. 12.30 The proposed decoding domain of the 30S ribosomal subunit of E. coli. The RNA secondary structure is from [156] and diverges from that of [155) with respect to the hydrogen bonding of the polypyrimidine region at the 3' -end. *, Nucleotides m~A 1518 and m~A 1519 , the demethylation of which confers resistance to kasugamycin; A, site of base change conferring resistance to paromomycin; ., site of base change conferring resistance to ; e nucleotides which are methylated (as m7G1405 and m1A 1408) in organisms producing certain aminoglycoside antibiotics, and confer them with resistance to these. The boxed nucleotides are those not involved in secondary structure that are highly conserved between different organisms. tations conveying resistance to streptomycin ribosomes with different error frequencies have now been shown to involve RNA, the are possible. Thus, many streptomycin• mutants most pertinent to the question of resistant mutants have alterations in ri• the accuracy of decoding involve ribosomal bosomal protein S12 that confer greater proteins. These led to the realization that accuracy to the ribosomes; whereas revert- 560 The translation of mRNA

ants from streptomycin-dependence (also The general view of the nature of ribo• involving S12) have mutations in ribosomal somal accuracy described above made it protein S4 that result in increased mis• possible to resolve an apparent paradox. reading [183]. Indeed, a role for ribosomal This was that, if it is possible to select for protein S12 in modulating translational ribosomes that have increased accuracy, why accuracy is indicated by the fact that ribo• has Nature not already done this? The most somes lacking S12 are more accurate than reasonable answer (and the one that is those of the wild-type (184]. thought to be correct) is that perfect accuracy The accuracy of decoding found in vivo is only possible if translation is infinitely is difficult to reconcile with the energy dif• slow, and hence the actual error frequency is ference between cognate and non-cognate the best compromise between accuracy and codon-anticodon interactions. Although speed. However, studies in vitro showed recognition may involve more than the anti• that ribosomes of different accuracy ex• codon of the tRNA (there is a suppressor hibited no differences in their maximum tRNA with an unaltered anticodon, but a speed of translation. This paradox was re• base change in loop I (185]) it seems that solved when it was realized that what was the accuracy of the ribosome is achieved relevant to kinetic proofreading was not the by the multiplicative effect of several steps of maximum rate that ribosomes could achieve 'kinetic' proofreading (cf. section 12.3.2), as with saturating ternary complex, but the rate proposed by Hopfield (186] and Ninio (187]. at which the ternary complex could interact There is evidence that the first step of proof• with the ribosome. When the values of Km reading (some workers prefer to call this were measured for the interaction of the stage 'recognition', and reserve 'proofread• ternary complex with ribosomes it was found ing' for the subsequent checking step) occurs that the Km increased (and hence the rate of in the binding of the ternary complex, cellular protein synthesis would be expected aminoacyl-tRNA.EF-Tu.GTP, when the to decrease) as the accuracy of the ribosome complex either binds to the ribosome, with increased (182, 188]. hydrolysis of GTP, or dissociates; whereas the second proofreading is of the bound aminoacyl-tRNA.EF-Tu.GDP, which can 12.7 OTHER PROTEIN-SYNTHESIZING either form a peptide bond, with liberation SYSTEMS of EF-Tu.GDP, or dissociate from the ribo• some [188]. In effect, the recognition of the cognate aminoacyl-tRNA by the ribosome is 12.7.1 Archaebacteria [189, 190] a question of whether the aminoacyl-tRNA dissociates from the ribosome before the In the discussion of protein-synthesizing GTP is hydrolysed or the EF-Tu.GDP is systems so far, a distinction has been made released: the time taken for GTP hydrolysis between the nucleo-cytoplasmic system of provides a 'kinetic yardstick' for proofread• eukaryotes and the system found in those ing. It has been suggested that the hydrolysis prokaryotes represented by the commonly of GTP during the action of EF-G is also studied bacteria (e.g. E. coli) and their involved in proofreading, but, although viruses. There is good reason, however, to there are some data that are consistent with think that the prokaryotes comprise not one, this view, it cannot be regarded as proven but two, kingdoms. These are the eubacteria, [182]. into which kingdom the majority of bacteria Other protein-synthesizing systems 561

fall, and the archaebacteria, a kingdom con• certain archaebacterial ribosomes are insen• taining methanogens, extreme halophiles and sitive to some antibiotics (chloramphenicol certain thermoacidophiles. It is appropriate and kanamycin) that had previously been to discuss briefly the protein-synthesizing regarded as specific for prokaryotic ribo• system of archaebacteria; not only because somes, and sensitive to at least one inhibitor this shows certain differences from eubacteria (anisomycin) that had been regarded as and eukaryotes, but because analysis of the specific for eukaryotic ribosomes. The archaebacterial translation apparatus has position of archaebacteria in relation to provided the strongest evidence that the eukaryotes and eubacteria is also apparent archaebacteria do in fact constitute a separate from the sequences of their ribosomal pro• kingdom. teins [192, 193]. Although the genes for Archaebacteria have certain structural these are organized in clusters similar to characteristics that distinguish them from those in E. coli, only some of the genes eubacteria. Their cell walls (where these are in these clusters encode proteins that are found) lack peptidoglycan, and their related to eubacterial ribosomal proteins. are unique in consisting of branch-chain fatty Some are related to eukaryotic ribosomal acids, and being linked to glycerol by an proteins, and others appear to be unique ether, rather than an ester, link. This, in (although a relationship to eukaryotic pro• itself, would hardly be sufficient to establish teins not yet characterized cannot be ex• them as a separate kingdom. The evidence cluded at present). Two proteins of particular that does so comes from phylogenetic studies interest are the equivalents of E. coli LlO on ribosomal RNA. Woese and Fox [191] and L7/L12, the components of the 'stalk' on showed that the sequence divergence be• the large ribosomal subunit important for tween the smaller (16S or 18S) rRNAs of translocase function (section 12.6.2(a)). eukaryotes and eubacteria clearly dis• Although it is difficult to see a relationship tinguished these two kingdoms. (This work between the eukaryotic and eubacterial was actually done by analysis of oligo• proteins here, the similarity of the archae• nucleotides before complete sequences were bacterial proteins to both eubacterial and available.) When the rRNAs of the different eukaryotic proteins establishes this and archaebacteria were examined it was clear argues for a functionally conserved role. that they were related to one another, but The sequences of the elongation factors were no more closely related to eubacteria from archaebacteria bear much greater than they were to eukaryotes. similarity to those of eukaryotes than they The protein-synthesizing systems of dif• do to those of E. coli. This is reflected in ferent archaebacteria are not all identical, the fact that, as in eukaryotes, the trans• but, as a generalization, they can be said locase from archaebacteria has the modified to show some similarities to those of histidine residue that can be ADP-ribosylated eubacteria, some to those of eukaryotes, by diphtheria toxin, and an EF-Tu/EF-1a and to have some unique features. Although equivalent that is insensitive to the anti• archaebacterial ribosomes have a sediment biotics, pulvomycin and kinomycin. Another coefficient of 70S, rather than 80S, they have apparent difference from eubacteria is the an additional morphological feature (a 'bill') lack of formylation of the methionyl residue also present in eukaryotic ribosomes. Most on the initiator tRNA, although further striking, however, is the sensitivity of archae• characterization of the initiation process is bacterial ribosomes to antibiotics. Thus, lacking at present. 562 The translation of mRNA

12.7.2 Mitochondria [194-196] of mitochondrial ribosomes to antibiotics shows an extensive similarity to that of The majority of mitochondrial proteins are eubacteria. The mitochondrial ribosomal encoded by mRNAs transcribed from nuclear proteins (encoded in the nucleus) are not genes, and these mRNAs are translated on well characterized, but appear to exceed 80S ribosomes in the , the resulting those of bacteria in number. The mito• proteins being transported into the mito• chondrial elongation factor responsible for chondrion (section 12.8). The mitochondrion binding aminoacyl-tRNA is functionally has its own genome (section 3.4.1) and interchangeable with E. coli EF-Tu, but its own machinery for transcription (section the translocase is not interchangeable with 8.5) and translation. Although the size and EF-G. Mitochondria, like eubacteria, use coding capacity of mitochondria vary quite fMet-tRNA to initiate protein synthesis widely between species (from c. 17 kb in [199), and there is some evidence that the man, to c. 2500kb in some plants) most mitochondrial initiation factors resemble mitochondria encode their own rRNAs and those of prokaryotes [198]. However, the tRNAs, together with a variable subset of smaller rRNAs from mitochondria lack the the proteins of the electron-transfer and bacterial polypyrimidine 3' -sequence; and oxidative systems. The in any case there is no opportunity for amount of coding DNA in the larger mito• 'Shine and Dalgarno'-type base-pairing to chondrial genomes is little greater than that mammalian mitochondrial mRNAs that start in the smaller ones, as much of the extra directly, or almost directly, with AUG. The DNA in the former is simple-sequence DNA unique post-transcriptional editing of protist (chapter 3). mitochondrial mRNAs has been described in It is generally thought that the precursors section 11.8. of mitochondria were aerobic bacteria that Perhaps the most striking features of the were engulfed by, and evolved in a symbiotic mitochondrial translational system (again, relationship with, an anaerobic host cell. with the exception of higher plants) are the Comparisons of rRNA sequences indicate deviations from the standard genetic code that mitochondria are most closely related to (section 12.2.3) and the unique manner in purple photosynthetic bacteria [197). This which the codons are read by their tRNAs. relationship between eubacteria and mito• This seems to be a consequence of the re• chondria is seen in certain functional aspects stricted set of mitochondrial tRNAs. The of the translational machinery, although actual number varies between species there are unique features to the mitochon• (mammals have 22, yeast 25), but the drial translational apparatus [198]. Mito• number is less than the 32 required to read chondrial ribosomes, other than those of the standard code (Fig. 12.3) according to plants, differ from the ribosomes of pro• the wobble hypothesis (Table 12.1). In the karyotes and eukaryotes in lacking the case of mammalian mitochondria (although otherwise essential SS rRNA, vestiges of not in those with larger genomes) there are which can be seen in the rRNA of the large apparently no separate genes for tRNArct subunit, which may subserve its function. and tRNA~ct, even though both Met-tRNA Mitochondrial rRNAs show fewer secondary and fMet-tRNA are found. It is assumed modifications ( methylations, pseudouridine that the single gene gives rise to both substitutions) than other rRNAs. As already tRNA species through differential post• mentioned (cf. Fig. 12.29), the sensitivity transcriptional modification. As already The control of the cellular location 563

described, these restricted sets of tRNAs are mation than those of mitochondria, and the able to decode the mitochondrial genome apparent lack of a corresponding intense by a different set of wobble rules (Table pressure for space has left them with a full 12.1). However, there are some protists complement of normal-sized rRNAs (in• (Tetrahymena pyriformis, Trypanosoma cluding 5S rRNA). Although the standard brucei, Chlamydomonas reinhardtii) in genetic code is employed in chloroplasts, which the number of tRNAs encoded by the there are not quite enough tRNAs (31 in mitochondrial genome is too small even for tobacco) to decode the mRNAs according this. In these cases it must be assumed that to the standard pattern of observed wobble the tRNAs are encoded in the nucleus and shown in Table 12.1. Unmodified Us are imported from the cytoplasm. employed in certain anticodons to allow The mitochondrial tRNAs do not conform reading of all members of certain four-codon to the general structural pattern described in families [37, 202]. section 12.3.1 and summarized in Fig. 12.7, The genome encodes other and it is possible that these deviations are components of its protein-synthetic ap• important for their unique decoding pro• paratus, including an initiation factor (IF-1 ), perties. In particular the mitochondrial an elongation factor (EF-Tu, although this is tRNAs violate some or all of the following absent in some cases) and somewhat less general features previously enumerated: the than half the ribosomal proteins. These conservation of the GT'PCRA sequence in latter show an average of 44% identity loop IV, the constant seven-nucleotide with clearly analogous proteins of E. coli, length of loop IV, the pattern of conserved although there is one example of a chloro• bases in loop I. In individual cases even plast ribosomal protein with no eubacterial more striking violations are observed: the homologue [203]. The polypyrimidine mammalian mitochondrial tRNA Ser com• sequence capable of 'Shine and Dalgarno' pletely lacks loop I and its stem, and in base-pairing is present at the 3' -end of several tRNAs from Caenorhabditis elegans chloroplast 16S RNA, but some chloroplast loop IV and its stem have been replaced by a mRNAs seem to lack canonical Shine and loop of variable size. Nevertheless, these Dalgarno sequences. If initiation codon tRNAs may adopt a conformation similar to selection involves a mechanism similar to that of yeast tRNAPhe (Fig. 12.8) [200]. that in eubacteria it may perhaps operate in a modified form.

12.7.3 Chloroplasts [201-203] 12.8 THE CONTROL OF THE The protein-synthetic apparatus of chloro• CELLULAR LOCATION OF THE plasts resembles that of bacteria even more PRODUCTS OF TRANSLATION [204] clearly than does that of mitochondria. Indeed, rRNA sequence comparison (cf. Proteins synthesized on the cytoplasmic section 12.7 .1) shows that chloroplasts are ribosomes of eukaryotic cells can be either specifically related to cyanobacteria, with retained in the cytoplasm, transferred to which they also share similarities in the struc• subcellular , inserted into the ture of their chlorophylls and carotinoids plasma membrane, or secreted. Although, [197]. The genomes of chloroplasts (120- of course, prokaryotes do not have sub• 200 kb) contain much more genetic infor- cellular organelles, the other possibilities 564 The translation of mRNA

mentioned are also available to them, N-terminus of the nascent protein [210]. In together with sequestration in the peri• most cases the signal peptide is an extension plasmic space. The molecular mechanisms of the N-terminus of the mature protein, governing these possible fates of proteins approximately 15-30 amino acids in length, involve recognition of features in their and is later removed by proteolytic cleavage. amino acid sequences - targetting signals. (In some cases the signal peptide is internal, There are two main classes of such signals: and in some cases it is not cleaved.) The signals that direct the translocation of pro• core of this signal peptide is predominantly teins across a membrane separating two hydrophobic, but there are usually basic or compartments, and signals that cause the neutral residues at the extreme N-terminus. retention of a protein in a particular cellular It seems reasonable to suppose that the compartment. Some of the signals consist of hydrophobic nature of the signal peptide is a conserved amino acid sequence, but more necessary for it to cross the membrane. frequently the targetting sequence can only However, in order for this hydrophobic be described in terms of its general chemical region to interact with the membrane it must nature and position within the protein. remain exposed, and this means that the nascent protein must not be allowed to adopt its native conformation before the inter• 12.8.1 Translocation across the action can occur. It is now known that there or the bacterial are two mechanisms to achieve this in pro• inner membrane [204-209] karyotes and lower eukaryotes (such as yeast), one of which is similar to the single Much work has centred on the execution of well-studied mechanism that operates in the signal for the translocation of proteins higher eukaryotes. In higher eukaryotes the across the membrane of the endoplasmic signal peptide interacts with an 11S signal reticulum of eukaryotes, or across the inner recognition particle (SRP), which contains membrane of bacteria. This pathway is six polypeptide chains and a 7S RNA species, taken by secreted proteins, which have been 260 nucleotides in length, to which the the main focus of study, but it is also the Alu family of repeated DNA sequences is common first step in the processes leading related (Fig. 7.36, section 7.7.4(c)). This to compartmentation of other proteins (see interaction, which involves the 54 kDa sub• below). Ribosomes synthesizing proteins unit of SRP (SRP54), may temporarily arrest destined for (unlike the majority protein synthesis, but, in any case, maintains of ribosomes synthesizing proteins to be the exposure of the signal peptide until the retained intracellularly) are located on the SRP interacts with a specific receptor ('SRP membranes of the rough endoplasmic re• receptor' or 'docking protein') on the inner ticulum of eukaryotes. The proteins are face of the membrane. This allows the ribo• extruded through the membrane as they are some to attach to the membrane, after which synthesized and pass from the cisternae of the SRP and its receptor are released, trans• the endoplasmic reticulum, via the Golgi lation is resumed, and the nascent peptide is apparatus, to secretory . The translocated through a specific channel [311] chemical basis for the initial membrane into the intracisternal space. The signal translocation, and that occurring in pro• peptide is, in most cases, cleaved by a karyotes, lies in a largely hydrophobic peptidase as it emerges on the luminal side peptide, the 'signal peptide', at or near the of the endoplasmic reticulum (Fig. 12.31). The control of the cellular location 565

Fig. 12.31 Model for the segregation of secretory proteins into the lumen of the endoplasmic reticulum. SRP, signal recognition particle; DP, docking protein (signal recognition particle receptor); P signal peptidase. For further details see text.

There is evidence that GTP (and presumably many years), and a structural counterpart its hydrolysis) is involved in the release of to SRP54 in the product of the ffh gene. the SRP and its receptor [211], and SRP54 Furthermore, a structural counterpart to and the SRP receptor contain conserved the eukaryotic SRP receptor exists in the motifs found in other GTPases [208]. It product of the fts Y gene [209]. One of the would be consistent with the precedents set reasons that the SRP escaped detection for by many other GTPases (including the so long in bacteria was that the genetic elongation factors of protein synthesis) if the approach that was applied to the problem role of the GTP hydrolysis were to disengage only revealed components of the second these proteins after they had fulfilled their mechanism for preventing function. By analogy with EF-Tu, there have (see below). However, it did lead to the also been speculations that the time taken identification of integral membrane proteins for GTP hydrolysis to occur may act as a necessary for the process of translocation, 'kinetic standard' to allow dissociation of the and thus common to both mechanisms [212]. N-terminal peptides of erroneously bound There is evidence that two of these, the proteins, not destined for translocation products of genes secE (priG) and sec Y across the membrane. (prlA), constitute the actual translocase, but The SRP of prokaryotes escaped detection the functions of the other two, the products for a considerable time, even though it was of seeD and secF, remain to be elucidated. known that bacterial signal peptides were Much less is currently known about the mem• functionally interchangeable with those of brane translocase proteins in eukaryotes, eukaryotes. The overall structure of bacterial and there may be some differences expected SRP is still unclear, but it contains a func• here, because in prokaryotes, but not in tional counterpart of7S RNA in the essential eukaryotes, the interaction of the negatively and abundant bacterial 4.5S RNA (the charged inner surface of the bacterial cell function of which had remained elusive for membrane with a positively charged residue 566 The translation of mRNA

near the N-terminus of the signal peptide and it has been suggested that an alternative is absolutely required for attachment to mechanism must operate in these cases occur [213]. It also appears that the signal [216]. peptidase in eukaryotes is more complex than the general signal peptidase in pro• karyotes, although their specificity is similar 12.8.2 Fate of non-secretory proteins [214]. This cleavage specificity cannot be after translocation across the membrane characterized simply, but involves recog• of the endoplasmic reticulum nition of an amino acid with a small side• chain in a relatively hydrophilic environment [210]. Proteins destined for secretion are thought to pass by default from the cisternae of The second bacterial mechanism to pre• the endoplasmic reticulum to the secretory vent protein folding before translocation vacuoles. However, proteins destined for (also found in yeast) involves proteins termed other locations have additional signals to 'molecular chaperones' or 'chaperonins'. direct them to these or retain them in them. Some of these (the GroE protein product of the groEL, groES genes, and the product of dnaK) are heat-shock proteins, which have (a) Retention as membrane proteins [217] a more general role in binding partially denatured proteins. However, one, the Certain plasma membrane proteins do not product of the secB gene, is specific for the complete passage through the membrane secretory pathway, and transfers the nascent of the endoplasmic reticulum, but remain peptide to SecA, a peripheral membrane embedded in it. The eventual fusion of protein with ATPase activity [212]. The membrane derived from the endoplasmic proteins protected from folding by this reticulum with the plasma membrane results mechanism, unlike those protected by the in the internally facing N-termini of such mechanism involving SRP, pass through the proteins becoming exposed to the external membrane post -translationally, rather than surface of the plasma membrane. What stops co-translationally. The relative roles of the such membrane proteins being secreted is two pathways to the membrane still remains a hydrophobic sequence at or near the C• to be resolved. It seems likely that the one terminus, the 'stop transfer' sequence, that involving SRP is the more ancient, and it anchors the protein into the membrane. This has been suggested that this pathway is still is especially well illustrated in the case of absolutely required for certain proteins (yet the membrane-bound and secreted forms of to be identified), although other proteins immunoglobulin heavy chain (Fig. 11.38), have evolved the ability to use the second which differ only in approximately 20 amino pathway [215]. If this is so, it should be acid residues at their C-termini [218]; and possible to identify features of the signal this C-terminal segment has been shown to peptide that confer the ability to use the convey membrane location when genetically second pathway. 'grafted' onto a protein that is normally Finally it should be mentioned that a small secreted [219]. minority of secretory proteins do not appear It should be stressed that there are other to possess a signal peptide of the type orientations of proteins in the plasma mem• required by the mechanism described above, brane: external orientation of the C-terminus The control of the cellular location 567

or multiple loops back and forth through 12.8.3 Translocation into the the membrane. However, proteins with such mitochondrion and direction to orientations are inserted into the membrane submitochondrial locations [223-225] in specialized manners involving internal signal sequences, for a description of which Although mitochondria encode some of the reader is directed elsewhere [217, 220]. their own proteins (section 12.7.2) the vast majority are encoded in the nucleus and synthesized on free cytoplasmic ribosomes. (b) Retention in the lumen of the Irrespective of their ultimate submitochon• endoplasmic reticulum [221 1 drial location, most of these bear a targetting signal that directs them post-translationally The lumen of the endoplasmic reticulum to a translocation system which leads through contains a number of resident soluble pro• the outer and inner mitochondrial mem• teins (e.g. protein disulphide isomerase, branes (at a point of contact between the prolyl-4-hydroxylase and others that interact two) into the mitochondrial matrix. Proteins with newly synthesized proteins). These are of the outer and inner mitochondrial mem• prevented from being secreted by a specific brane have additional 'stop transfer' signals C-terminal sequence, which in animal cells that arrest their transfer before they reach is KDEL, and in yeast is generally HDEL. the matrix. Proteins of the intermembrane It is thought that this sequence is bound space are re-exported to that compartment by a membrane receptor in a salvage com• on reaching the matrix. partment, on route to the Golgi, and this The mitochondrial targetting sequence receptor recycles the proteins to the endo• takes the form of an N-terminal extension plasmic reticulum. of the mature protein, but differs from the signal peptide described in 12.8.1 in being (c) Transfer to the [222 1 more variable in length (c. 12 to > 70 residues) and being much more polar in Proteins such as hydrolases that are resident nature. An important property of the ta:rget• in the lysosomes of mammalian cells follow ting sequence is a potential for forming the secretory pathway as far as the Golgi. a positively charged amphiphilic a-helix. In the cis Golgi an unknown feature of Soluble cytoplasmic chaperonins (including these proteins is recognized by N-acetyl• a family of 70-kDa heat-shock proteins, glucosaminyl-1-phosphotransferase, which cytoplasmic hsp70) prevent the mitochon• modifies N-linked oligosaccharides on the drial protein from folding completely after proteins, and further modification by other translation. The protein is then transferred enzymes generates manose 6-phosphate from the cytoplasmic hsp70 to a receptor groups. These latter constitute the signal on the outer mitochondrial membrane, that directs the proteins, bound by specific with the concomitant hydrolysis of ATP. receptors, to the lysosomes. Receptor components have been identified A different mechanism to the above [224, 225), but as yeast bearing mutants in operates in yeast, and, although the details these are viable it is assumed that there are have not yet been completely elucidated, alternative receptor systems. The receptor is this involves the action of a protein kinase thought to transfer the protein to the actual [312). transfer pore, one component of which 568 The translation of mRNA would appear to be an essential integral is cytochrome c, which does not have an outer- (ISP42 in yeast, N-terminal extension and is inserted directly MOM38 in Neurospora). An electric poten• into the intermembrane space, rather than tial across the inner membrane is required first passing through the matrix. for transfer of the protein to the mitochon• drial matrix, although this is of opposite polarity to that required for periplasmic 12.8.4 Translocation across the sequestration in bacteria. Within the matrix chloroplast membrane [226, 227] the protein first interacts with one of the essential components of the translocation The problem of translocation to different machinery, mitochondrial hsp70. It is then compartments of the chloroplast has not transferred to a second essential matrix received the same intense study accorded to chaperonin, mitochondrial hsp60 (a protein the mitochondrion, but the picture emerging structurally related to GroEL of bacteria), is not dissimilar. Initial targetting to the with hydrolysis of A TP. There is evidence chloroplast is conveyed by an N-terminal that hsp60 is required for correct folding of extension somewhat resembling the mito• matrix proteins. The final step for matrix chondrial one, although with the difference proteins is the cleavage of the signal peptide that it is not predicted to form an amphiphilic by a (MAS protease), which has a-helix. This signal delivers the protein two non-identical subunits. through the double membrane of the chloro• It appears that the mechanism for the plast envelope into the stroma. There is re-export of proteins to the intermembrane evidence for a requirement for A TP for space is basically similar to that employed import into the stroma, but not a membrane in bacterial periplasmic sequestration: the potential, and a stromal signal peptidase relationship of the mitochondrial matrix to has been described. From the stroma certain the intermembrane space is similar to that of proteins need to be transported further into the bacterial cytoplasm to the periplasmic the lumen of the thylakoids, and this process space, and the directionality of the membrane formally resembles that of mitochondrial polarity is, of course, the same. In fact, the transport from the matrix to the inter• presumed evolutionary pre-existence of this membrane space. It, too, requires a second, translocation system allows one to rationalize hydrophobic, signal sequence, and, indeed, the initial transfer of proteins to the matrix certain proteins synthesized in the stroma on before transfer to the intermembrane space. chloroplast ribosomes have just this single A second, hydrophobic, portion of the signal targetting signal. sequence is required for this latter process, and this is finally removed by a protease, perhaps inner-membrane protease I, which 12.8.5 Translocation to microbodies is structurally related to bacterial signal [228] peptidase. Before concluding this section it should be Microbodies, which include , stressed that the general mechanism de• glyoxisomes and , are structurally scribed above does not apply to all mito• related organelles which share the possess• chondrial proteins, some of which have ion of a P-oxidation system. Microbody specialized individual means of trans• proteins are synthesized on cytoplasmic location. One example of such an exception ribosomes and their transport is directed by The regulation of translation 569

targetting signals that do not undergo sub• the sites of synthesis and usage of mRNA sequent cleavage. In the case of some of are physically separated, and this, and the these proteins this targetting signal is a relative stability of most eukaryotic mRNAs, C-terminal tripeptide sequence, SKL. provide scope for the translational control However, other proteins appear to carry a of protein synthesis. Moreover, even in different, but as yet undefined, internal prokaryotes, examples of translational re• targetting signal. gulation are to be found. The following account of translational control classifies these examples primarily in terms of the me• 12.8.6 Translocation across the nuclear chanisms involved. The post-transcriptional membrane [229, 230, 313] editing of mRNAs (section 11.8), and regu• lation of translation by antisense RNAs At present two methods of post-translational (section 10.7) and by codon usage (section translocation of proteins across the nuclear 12.2.4) have already been described. membrane have been identified. Small proteins pass through the nuclear pores by simple diffusion, and are then concentrated 12.9.1 mRNA secondary structure in the nucleus by specific binding to other nuclear proteins. Large proteins, however, (a) RNA bacteriophages [231, 232] require a targetting signal. The targetting signal, which may be as short as a penta• The influence of mRNA secondary structure peptide, does not consist of a precise on translation differs in prokaryotes and sequence, but contains three or more basic eukaryotes because of their different modes amino acids (lysine being more frequent of initiation. In the small bacteriophages, than ). An example of such a such as MS2 and Qp, there is a striking sequence is PKKKRKVE of SV 40 large T example of the role of the secondary structure antigen, which can cause nuclear trans• of their mRNA in the differential expression location when artificially introduced into of different cistrons of a polycistronic proteins normally resident in the cytoplasm. mRNA. These single-stranded RNA phages The nuclear targetting sequence does not direct the translation of four proteins from appear to have to be located at any particular their structural RNA, and these include the position in the protein, provided that it is coat protein and the viral subunit of the exposed. The translocation appears to be a replicase. During infection, or translation in two-step process: initial rapid binding to vitro, there is much greater synthesis of coat nuclear pores, probably involving an acidic protein than of replicase; and it was shown receptor region, followed by slower ATP• that destruction of the secondary structure of dependent transfer through the pores. RNA by mild formaldehyde treatment or by heating redressed the rate of synthesis of replicase to that of coat protein. The pro• 12.9 THE REGULATION OF posed secondary structure for MS2 RNA TRANSLATION [233] shows the initiation site of the replicase subunit hydrogen-bonded to part of the coat The regulation of protein synthesis by alter• protein cistron, the initiation site of the ing the rate of transcription of mRNA has latter being well exposed (Fig. 12.32). Thus, been described in chapter 10. In eukaryotes it appears that initiation of replicase trans- 570 The translation of mRNA

Fig. 12.32 Model for the secondary structure of part of the coat protein cistron of bacteriophage MS2. The positions of the coat protein initiation and termination codons are indicated, as well as the initiation codon for the replicase. The numbers 6, 50 and 70 indicate codons which have been mutated in the related bacteriophage f2 to produce premature termination (see text). (After [233), with permission.) lation is restricted because it is only accessible 50th or subsequent mutant termination when the coat protein cistron is being codons would. It now appears that the ex• translated (which would temporally disrupt pression of a fourth, minor, protein of MS2 this hydrogen bonding). Experimental (the lysis protein) is also constrained by support for this model came from a study the secondary structure (and not by a re• of chain-termination mutations in the coat quirement for frameshifting, as originally protein gene. An amber mutation at the proposed). codon specifying amino acid position 6 of the coat protein severely repressed express• (b) Eukaryotic mRNAs ion of the cistron for the replicase subunit, but this effect was abolished if the phage Although initiation of eukaryotic protein RNA was treated with formaldehyde. In synthesis involves the RNA helicase action contrast, amber mutations at the codons of eiF-4A (section 12.5.1), there is ex• specifying the 50th, 54th and 70th amino perimental evidence that the translational acids did not have this effect. It can be seen efficiency of mRNAs decreases if artificial from inspection of Fig. 12.32 that translation stem -loop structures are inserted in the of only the first six codons of the coat pro• 5'-untranslated region [234). It is possible tein cistron would not break the putative that a difference in secondary structure be• hydrogen bonds to the replicase subunit tween different mRNAs could affect their initiation site, but that translation to the translation if eiF-4A were limiting. Examples The regulation of translation 571

might be the a- and p-globin mRNAs, where than phage mRNAs, is the autogenous a difference in translation was eliminated control of the synthesis of bacterial ribo• by addition of eiF-4A and eiF-4B (235], somal proteins. The syntheses of the ri• and reovirus mRNAs, which are translated bosomal proteins of E. coli are closely to different extents during infection (236]. co-ordinated with one another and with that It has recently been suggested that such of rRNA. There are approximately 25 secondary structure may be a reason for the transcriptional units (also referred to as translational repression of maternal mRNAs ) for the 54 ribosomal protein genes, of Xenopus oocytes; and evidence has been some containing several ribosomal protein obtained that at fertilization, when trans• genes, whereas others contain only one. lation of maternal mRNAs occurs, oocytes Genes for other proteins involved in macro• suddenly become capable of translating molecular synthesis are also found in these artificial mRNAs with extensive secondary units, some of which are shown in Fig. 12.33. structure (237]. In this respect it is interest• When transducing phages or plasmids were ing that an inhibitor of eiF-4F has been used to increase the copy number of the reported in unfertilized sea urchin eggs, ribosomal protein genes, without a con• which also contain translationally repressed comitant increase in the number of rRNA maternal mRNA (238]. genes, it was found that there was no in• crease in synthesis of the corresponding ribosomal proteins. This indicated that some 12.9.2 RNA-protein interaction form of feedback repression was occurring, and it transpired that a single protein in each Another aspect of the role of mRNA transcription unit was responsible for re• secondary structure in translation control pressing the synthesis of the other members is that it can provide specific sites for the of the unit in vivo, as indicated in Fig. 12.33. binding of regulatory proteins, and there are Experiments in vitro (which show consistent, several well studied examples of this in both if not identical, results to those in vivo) prokaryotes and eukaryotes. demonstrated that the target of the regulation was the translation of the ribosomal protein (a) Bacteriophage coat protein mRNA, and that the repressor proteins also inhibited their own translation. In the case of the translation of the cistrons A possible mechanism for the regulation of bacteriophage RNAs, discussed above, it was suggested by the fact that many of the is thought that the decline in synthesis of proteins responsible for the feedback in• the replicase at later times of infection is hibition (e.g. S4, S7 and S8) had been shown caused by the binding of the coat protein to bind to rRNA early in the assembly of (as the concentration of this increases) to a ribosomes (section 12.6.1). Thus, it was region of secondary structure containing the proposed that these proteins were also able ribosome-binding site [239). to recognize a site on the polycistronic mRNA that resembled their rRNA-binding (b) Autogenous regulation of bacterial site, and the binding to which would block ribosomal protein synthesis [240, 241] translation. Binding to this site would be of lower affinity than that to rRNA so as not to A more intensively studied prokaryotic interfere with assembly of ribosomes, but example, and one involving cellular rather would occur when the ribosomal protein was 572 The translation of mRNA

L11 operon p L11 lb1l In vitro + + In vivo + (+)

Rif operon p luol L7/12 (3 (3' In vitro + + In vivo (+) +

str operon p 512 ~ EF-G EF-Tu In vitro + + In vivo + (+) + S10 operon p 510 L3 ~ L23 L2 519 L22 53 L16 L29 517 In vitro + + + + ± In vivo + + (+) + + + + + + + +

Spc operon p L14 L24 LS 514 ~ L6 L18 55 L30 LIS secY L36 In vitro + + NO NO NO NO In vivo + + + + (+) + + + + + NO NO oc operon p 513 511 ~ 0( L17 In vitro + + + In vivo + + (+) + + 520 operon p ~ In vitro + In vivo (+)

Fig. 12.33 Autogenous regulation of ribosomal protein synthesis in E. coli. Each operon is indicated by an arrow, the direction of which is that of its transcription. The promoters are indicated by P and the individual genes by the names of their protein products. Regulatory ribosomal proteins are boxed, and the effects of these on the synthesis of proteins in the same operon in vivo and in vitro are indicated ( +, inhibition; -, no effect; ( + ), presumed to inhibit in vivo; ND, not determined). The autogenous regulation of the synthesis of proteins L14, L24 and S12 operates by retroregulation, involving nucleolytic cleavage of the mRNA [324]. The non-ribosomal proteins in the operons are a, p, and P', subunits of RNA polymerase, and secY, a component of the protein export machinery. (After [240], with permission.)

produced in excess of the available rRNA. mutagenic studies have established that As the cotranscribed 16S and 23S rRNAs the target of the translational feedback is would regulate the uptake of all rRNA• generally a single site, located near the binding proteins in the same way, the feed• translational start of the first gene of the back of the excess of these would explain transcriptional unit (the strand spc operons how the translation of the polycistronic are exceptional: Fig. 12.33). The binding mRNAs from different transcription units sites on rRNA of certain ribosomal proteins was also co-ordinately regulated. Extensive have been determined (section 12.6.1) and, The regulation of translation 573

in the case of S8, L1 and LlO, similar sec• the translation of the mRNA for which in• ondary structures can be predicted for their creases when the concentration of iron regulatory binding sites on the mRNA [241 ]. increases. A secondary structure element in A more complex structure, a pseudoknot the 5' -untranslated region of the mRNA has (Fig. 2.27), has been demonstrated as the been demonstrated by mutagenic experi• recognition site for ribosomal protein S4 ments to be sufficient and necessary for the in the region of the S13 mRNA initiation regulation; and this element has been codon, and it is assumed, although not yet termed the iron-responsive element (IRE). established, that the S4 binding site on A 90-kDa protein that specifically binds this rRNA has a similar tertiary structure. element in vitro has been identified, and this Indeed, in this case, the deduction of the represses translation of the mRNA, which structure of the mRNA has provided an may also lead to its degradation. It is as• impetus to studies on the more complex sumed that in vivo the binding of this protein tertiary structure of rRNA. to the IRE is regulated by the concentration This proposed mechanism of translational of available iron. There is suggestive evi• regulation (analogues of which appear to dence that this regulation may be mediated operate on the mRNAs for a range of other by the iron-, haemin, which may RNA-binding proteins) raises the question oxidize free sulphydryl groups on the protein of how the binding of a ribosomal protein [243, 244]. This may be a general mechanism to a single site on a polycistronic mRNA for iron-binding proteins, as an IRE has also can prevent translation starting at separate been identified in the 5'-untranslated region initiation sites on the same molecule, and of the mRNA for the transferrin receptor why certain members of the transcription [245]. units (e.g. EF-Tu and the f3 and /3' subunits of RNA polymerase) are not subject to this feedback regulation (Fig. 12.33). The 12.9.3 Degradation of mRNA and most likely explanation is that the initiation modulation of its half- [246, 247] sites of all but the first mRNA are masked by secondary structure and only become The destruction of a mRNA will obviously accessible when the first mRNA is translated prevent its translation. Although the induc• ( cf. the coat protein and replicase genes tion of ribonuclease is used as a regulatory of the RNA phages- section 12.9.1). This mechanism in certain circumstances, the would explain those proteins that escape structural features of individual mRNAs can feedback regulation (EF-Tu etc.) in terms of influence their half-, and it is this aspect exposed translational initiation regions, and that will be considered first. explain the lack of repression in vitro for certain distal mRNA cistrons in terms of (a) The role of the poly(A) tail of the mRNA exposure through nuclease attack in vitro. [248, 249]

(c) Ferritin mRNA [242] There is a large body of evidence indicating that the 3'-poly(A) segment of mRNAs The best characterized example of the influences their degradation, with the loss of regulation of a eukaryotic mRNA by inter• this segment or its reduction to a certain action with a protein is that of ferritin minimum size leaving the mRNA vulnerable mRNA. Ferritin is an iron-storage protein, to exonuclease action. Thus, the poly(A) 574 The translation of mRNA sequences of cytoplasmic mRNAs become must be some structural feature of different progressively shorter with time; mRNAs to explain the differences in their mRNAs, which lack poly(A), have shorter half-lives. In the case of some very short• half-lives than most eukaryotic mRNAs; lived mRNAs, e.g. that for the granulocyte• artificially deadenylated mRNAs are trans• monocyte colony stimulating factor, this has lated less efficiently than normal mRNAs been shown to be an AU-rich component of in vivo and in vitro; and histone mRNAs the 3' -untranslated region of the mRNA. have been stabilized by artificially poly• Several other short-lived mRNAs (including adenylating them. Although initial poly• the cellular oncogenes, fos and myc) have a adenylation occurs in the nucleus, it is similar sequence, and transfer of this to a possible for the length of the poly(A) seg• stable mRNA, such as globin, dramatically ment of mRNAs to be extended in the decreases its half-life. cytoplasm [250]. This occurs during oogenesis At present nothing is known of the way and embryogenesis in certain maternal in which this AU-rich sequence conveys mRNAs that contain U-rich regions in their instability on the mRNA. One suggestion 3' -untranslated regions. This is thought to is that it competes with the 3' -poly{ A) seg• selectively increase the translation of these ment for the poly(A)-binding protein [251], mRNAs. although a different protein that specifically Poly(A)-containing mRNA decays expo• binds AU-rich sequences has been described nentially, and this implies that the suscepti• [253]. A further intriguing aspect of the bility of such mRNA to degradation does instability of these mRNAs is that their not increase with decreasing size of poly( A). half-lives can be dramatically increased by Rather, there is evidence that when the size various mitogenic stimuli [252], and in view of the poly( A.) segment falls below a certain of the potential oncogenic nature of the threshold value the mRNA suddenly be• products of some of these mRNAs there is comes extremely susceptible to degradation. considerable interest in elucidating the An approach to understanding the mechan• mechanism of this regulation. ism of this effect has come through the characterization of a specific poly{ A)-binding (c) Interferon and the degradation of viral protein containing a four-times tandemly RNA {254-257] repeated RNA- [251]. It is the association of this protein with the 3'• Animal cells infected with viruses produce poly(A) segment that is thought to protect interferon, a family of glycoproteins which it from nuclease attack, and the critical act on adjacent uninfected cells, enabling poly(A) size below which rapid degradation them to resist infection by inhibiting the of mRNA occurs is thought to be the replicative cycle of the virus. Interferons are minimum size for binding of the protein. species-specific but are effective against a wide spectrum of animal viruses. (They also (b) The A U-rich instability sequence {252] have inhibitory effects on which may be of importance in uninfected cells.) Apart from the cases in which the extent Although in this section we are only con• of is increased in the cyto• cerned with the ability of interferon to inhibit plasm, the 3' -poly{ A) segment would appear translation of viral mRNAs, it appears prob• to be a common determinant of the half-life able that the transcription is also inhibited. of eukaryotic mRNAs. Nevertheless, there The common signal for the induction of The regulation of translation 575

Virus ..... B ~ Interferon ~ Cell 2

[2'- 5' A Synthetase]. . [Protein kinase] inactive mact1ve Virus _ ...... ,. ..~ ds RNA----+- t ~ ~dsRNA..,..4:~...... ;-..virus 2'- 5' A Synthetase Protein kinase ATP t lO 2'-5' A elF-2 A ..- eiF-2 ATP ADP ® )1 ~ [RNase L). t' ---'---~ RNase L mac 1ve +

RNA -----'-~Degraded RNA

Fig. 12.34 Pathways by which double-stranded RNA and interferon, resulting from infection of cells by viruses, may cause the inhibition of viral protein synthesis. 2' -5' A is an abbreviation for ppp(A2'p5' A2'p5' A)n· For other details, see text.

interferon by different viruses is thought to specific in its mode of action (section 12.9.4). be double-stranded RNA, which is also the Nevertheless, there is a considerable body signal that activates the gene-products of of evidence that both these mechanisms can interferon that counter the translation of be important for the translational inhibition viral mRNA in the 'protected' cell. One in vivo, their relative importance depending of these gene-products is a protein kinase, on the virus. In the case of the action of and this is discussed in section 12.9.4. The 2-5A synthetase, the importance of which other is an enzyme called 2-5A synthetase, appears restricted to certain picornaviruses, which catalyses the synthesis of the tri• it has been suggested that the 2' -5' -oligo• nucleotide, pppA2'p5' A2'p5' A (Fig. 12.34). adenylate required for the activation of the This nucleotide, in turn, activates the re• ribonuclease is rapidly degraded as it dif• latively non-specific ribonuclease L (section fuses away from its site of synthesis, the 4.3.1(e)). double-stranded RNA [254]. It is argued Although it can be seen how this nuclease that its action would be largely restricted to could prevent the translation of viral mRNA, this viral double-stranded RNA, thought it is not immediately obvious how host to consist of replicative intermediates in mRNA could escape degradation. Nor is the the case of these RNA viruses. (In the interferon-induced protein kinase any more case of DNA viruses, overlapping mRNA 576 The translation of mRNA transcripts are thought to be the source of was inhibited when there was a build up of double-stranded RNA.) However, it is more unpolymerized tubulin monomers, and that difficult to extend this idea to the action of this involved degradation of their mRNAs. the protein kinase. An alternative way of In the case of the mRNA for fJ-tubulin, it resolving the problem is possible if one was found that this regulation was conveyed accepts that the inhibition need not, in fact, by the section of the mRNA encoding the be specific. The cell, unlike the virus, may four N-terminal amino acids, MREI. be able to compensate for the effect of the Mutations in this region that did not alter nuclease by increasing its synthesis of RNA, the coding potential did not destroy the and by resuming protein synthesis on de• regulation, whereas the opposite result was phosphorylation of the cellular target protein obtained with those that did. It appears after the virus has been destroyed. that, in some way yet to be determined, the interaction of unpolymerized tubulin mono• (d) Herpes simplex virus and the degradation mers with this peptide activates nucleolytic of host mRNA [258, 259] degradation of the mRNA. Although it might be expected that other proteins with A situation that appears to be the converse the same N-terminal sequence might bring of that described above is found in the case into question the specificity of this mechan• of the inhibition of host protein synthesis by ism, no other example of an N-terminal the large DNA virus, herpes simplex type-1. MREI sequence has yet been found. The N• It has been demonstrated genetically that terminal sequence of a-tubulin, MREC, the product of viral gene UL41 is responsible is similar enough to expect that it is also for the early phase of the shut-off of host recognized by unpolymerized tubulin, al• protein synthesis. However, if this non• though this has not yet been demonstrated. essential gene is deleted there is an increase in the half-life of virus mRNA as well as host mRNA. Thus, it would appear that the ribo• 12.9.4 nuclease responsible for the degradation (there are some indications that this is a (a) Phosphorylation of e/F-2 [261-263] cellular enzyme, rather than the product of gene UL41 itself) is not intrinsically specific Initial studies of the phosphorylation of for host mRNA. In this case it may be the eiF-2, performed in reticulocyte lysates, concomitant selective shut-off of host trans• will be summarized before considering the cription that produces an apparent specific more general but similar phenomenon in effect on translation. cells responding to interferon. Reticulocyte lysates provide a very active protein• (e) Autoregulation of the stability of tubulin synthesizing system, but if haem is not added mRNAs [260] there is a rapid cessation of protein syn• thesis. (Haem is rapidly converted in aqueous The co-ordinated regulation of the a- and solution into an oxidized form, haemin, fJ-tubulins (the principal subunits of the which is what is actually used in such studies.) microtubules) involves the regulation of the Although one can rationalize this effect in degradation of their mRNAs by a mech• terms of the co-ordination of the synthesis anism that is, to date, unique to them. It was of the predominant reticulocyte protein, demonstrated that the synthesis of tubulin globin, with the availability of its prosthetic The regulation of translation 577

Met-tRNAf

--f----...,-----;.,... eiF-2•GTP \._ ., MeHRNAf•eiF-2•GTP

GTP ~405

eiF-2•eiF-2B 40S•Met-tRNA1• elF-2• GTP mRNA 605 GOP Pi --~---eiF-2•GDP.,._ ___.-

eiF-28-....---< ATP=i Pro~ein kinase ADP --~--- eiF-2 .. GDP ® 80S•Met-tRNA1•mRNA I I f [eiF-2•GDP•eiF -2 B] ® Inactive complex

Fig. 12.35 A scheme for the role of factors elF-2 and elF-2B in the initiation of eukaryotic protein synthesis, and the effect of phosphorylation of the a-subunit of eiF2. The point at which the phosphorylation of eiF-2 has been indicated is purely arbitrary.

group, haem, it must be emphasized that, in phorylated eiF-2 has a much greater affinity fact, the synthesis of all reticulocyte proteins for eiF-2B than the unphosphorylated factor; is similarly affected. It was demonstrated in effect sequestering the exchange protein. that the step of protein synthesis affected Phosphorylation of only approximately 20% in reticulocytes deprived of haem was the of eiF-2 can cause complete inhibition of binding of Met-tRNAr to the 40S ribosomal initiation because there are only 20-25% as subunit, the reaction that requires eiF-2 many molecules of eiF-2B as eiF-2. (Fig. 12.19). An inhibitor (known as HCR, It was also found in reticulocytes that haem-controlled repressor; or HRI, haem• low concentrations of double-stranded RNA regulated inhibitor) was purified from the could also provoke the phosphorylation of lysates and shown to be a cytoplasmic pro• eiF-2, but that a different protein kinase tein, pre-existing in an inactive form in (although of the same substrate-specificity) normal cells. It was eventually found to be a was activated. This double-stranded RNA• specific protein kinase that phosphorylates activated protein kinase is also found in the a-subunit of eiF-2. This has the effect of nucleated cells, although, depending on the preventing the liberation of GDP from the cell type, its endogenous concentration may eiF-2 released from the ribosome, although be quite low. However, when cells are treated eiF-2B still forms a complex with the released with interferon there is an induction of species (Fig. 12.35). Although the formation the synthesis of this protein kinase [256]. of the complex is readily reversible, phos- Attempts have been made to explain why 578 The translation of mRNA low concentrations, but not high concentra• (b) The phosphorylation of eukaryotic EF-2 tions, of double-stranded RNA are able to activate this enzyme. It is known that Although the dogma of translational regu• activation of the protein kinase by double• lation insists that control can only be exerted stranded RNA is accompanied by autophos• at initiation, there is now clear evidence that phorylation of the protein kinase, and it has the eukaryotic translational translocase, been suggested that this is an intermolecular EF-2, is the target of regulatory phos• reaction that is essential for the activation. If phorylation. Thus, it has been shown that this is so, then at high concentrations the the major (perhaps sole) substrate for Ca2+ I ratio of double-stranded RNA to protein calmodulin-dependent protein kinase III kinase would be such that any single mol• among cellular proteins in vitro is EF-2 ecule of RNA would be unlikely to bind the [267], that this phosphorylation inactivates two molecules of protein kinase required for the factor [268], and that there are changes intermolecular autophosphorylation [264]. in the extent of phosphorylation of EF-2 in The ability of the double-stranded RNA• certain situations in vivo. It is rather curious, dependent protein kinase to inhibit viral however, that several of these situations protein synthesis in vivo has been clearly are typified by an increase, rather than a shown for adenovirus [265]. Adenovirus syn• decrease, in protein biosynthesis (e.g. stimu• thesizes a small RNA, VAl RNA, which is lation of quiescent cells by [269]); and required to allow translation of late mRNAs, it must be remarked that there is no obvious and this acts by binding to the protein kinase rationale for the control of protein synthesis and inhibiting its activity. Other viruses also by changes in the cellular concentration of have mechanisms to inactivate this kinase calcium ions. [265]. In HIV-1 (human immunodeficiency virus type 1), it is possible that the stem• (c) The phosphorylation of other loop structure of the Tat-responsive region components of the protein synthetic RNA acts in a similar way to adenovirus apparatus VAl RNA [314]. In contrast, vaccinia virus encodes a truncated homologue of eiF-2a Other eukaryotic initiation factors besides lacking the target serine residue phos• eiF-2 have been shown to be phosphorylated phorylated by the double-stranded RNA• in vivo [266]. Most strongly implicated in dependent protein kinase. This has been regulation is the phosphorylation of eiF-4E, shown to abrogate the effect of interferon, which in certain circumstances shows a cor• presumably by binding to the protein kinase relation with protein synthetic activity. as a pseudo-substrate inhibitor [315]. In the However, as yet, there has been no demon• case of influenza virus the precise details of stration of a direct effect of phosphorylation the anti-interferon mechanism are not yet on the activity of the factor in vitro. known, but it is interesting that this involves Certain eukaryotic ribosomal proteins are a cellular, rather than viral, protein, which also phosphorylated in vivo [270, 271]. On would suggest a role in normal the 60S subunit, the acidic proteins related [316]. Certainly, there is a substantial body to E. coli L7/L12 (section 12.6.1) are phos• of evidence that uninfected cells can regulate phorylated, and there is evidence that this is their protein synthesis through the double• a prerequisite for their incorporation into stranded RNA-dependent eiF-2 protein the subunit, an event that appears to occur in kinase [266]. the cytoplasm for these proteins [272, 273]. The regulation of translation 579

On the 40S subunit, the basic ribosomal responsible for the degradation of the factor. protein, designated (eukaryotic) S6, can It appears that an, as yet uncharacterized, accept up to five phosphoryl residues per cellular protease is the enzyme responsible, molecule in vivo. The phosphorylation of and this raises the possibility that there exist this protein is greatest in rapidly growing circumstances where this is used cells and there is much interest in the possible as a cellular mechanism for the regulation of role of this phosphorylation in the trans• protein synthesis. duction of extracellular signals for growth inside the cell. Two have been characterized [274, 275], 12.9.6 Suppression and frameshifting one with two different protein kinase [278,279] domains, but the function of the phos• phorylation of ribosomal protein S6 is still The methods of regulation of protein syn• uncertain. As it is clear that the phosphoryl• thesis discussed so far operate within the ation is not an obligatory requirement for framework of rigid adherence to the genetic ribosome function, some more subtle regu• code. However, in both eukaryotes and latory role is indicated. One intriguing prokaryotes there are cases, predominantly possibility is of an interaction with the although not exclusively involving RNA system controlling the half-life of those viruses, where regulation involves the dis• unstable mRNAs that are induced following regard of a termination codon (suppression) a mitotic stimulus (section 12.9.3). or its avoidance by change of Finally it should be mentioned that certain ( frameshifting). eukaryotic aminoacyl-tRNA synthetases are also phosphorylated, although the regu• (a) Suppression [278-281] latory significance of this is as yet unclear [276]. Suppression of termination codons is used by certain RNA viruses to effect the syn• thesis of an alternative extended version of 12.9.5 Initiation factor proteolysis a particular protein, containing additional [264,277] functional potential. An example of this has already been encountered in the case of the The unorthodox mode of initiation adopted alphavirus, Sinbis virus (Fig. 12.20); and by poliovirus mRNA, which lacks the 5'• bacteriophage op, mentioned in section 'cap' structure, has been described in section 12.9.1, synthesizes minor amounts of a 12.5.1. Its lack of requirement for a func• fourth protein, thought to be involved in tional eiF-4F complex is the basis of its release of progeny phage particles from the inhibition of the translation of capped host bacterium, as a result of readthrough of mRNAs, which involves proteolytic cleavage the coat protein termination codon. Other of p220, the 220-kDa component of eiF-4F. examples include several plant viruses, such The poliovirus mRNA, like that of Sinbis as tobacco mosaic virus, and certain retro• virus (Fig. 12.20), is translated into a poly• viruses, such as Moloney murine leukaemia protein which is cleaved by two virally virus, where a gag-pol 'fusion protein' is coded . One of these, protease produced by the relatively efficient sup• 2A, has been shown to be necessary for the pression of the gag termination codon. In the proteolysis of p220, but is not itself directly case of the alphaviruses and retroviruses, 580 The translation of mRNA

there is proteolytic processing of the fusion a G, and in tobacco mosaic virus a UAG protein to generate one or more separate followed by a C. Furthermore, in the case of functional polypeptides. Moloney murine leukaemia virus, mutation There are two requirements for such of the UAG termination codon to UAA did specific suppression of termination codons: not abolish its suppression. Thus, it seems there must be some distinguishing feature in likely that a wider context is important, as the 'context' of the codon, and there must be has been found for frameshifting (below), a tRNA (a suppressor tRNA) to insert an although there are few other data available amino acid in response to this. As regards a at present that bear on this question. Instead, context effect, it has long been known to the focus of much work in this area has been bacterial geneticists that amber mutations identifying the tRNAs responsible for the (i.e. artificially induced UAG stop codons) suppression in vivo. are particularly inclined to be 'leaky' (i.e. to A normal mouse tRNA Gin has been im• be subject to readthrough suppression) but plicated as the suppressor of the UAG that the degree of 'leakiness' varied with the termination codon of Moloney murine position of the mutation. It has been shown leukaemia virus, and this has an anticodon in experiments in bacteria that UAG codons 3'-GUC-5', implying a G:U interaction at followed by a purine, especially A, are well the first position of the codon [283]. In suppressed, whereas those followed by a the case of tobacco mosaic virus, the UAG pyrimidine are weakly suppressed; and it is suppressor in tobacco leaves has been shown precisely a UAG codon followed by an A to be a normal tRNATyr with an anticodon that is suppressed in the readthrough of 3'-A'I'G-5', implying abnormal G:G hydro• the Q{J coat protein termination codon. gen bonding in the 'wobble' position. It is This, incidentally, allows one to rationalize interesting that another natural tRNATyr in terms of efficient termination the high does not act as a suppressor. This has the 5 '• frequency with which U follows natural anticodon G replaced by queuosine (Q), a bacterial stop codons, and the low frequency modified form of G, which shows the same with which UAG occurs (section 12.4.3). pattern of base-pairing with C in the wobble The tRNA responsible for the suppression position. One imagines that the conformation in Q{J has been shown to be a normal E. coli adopted by the termination codon must be tRNATrp with a 3'-ACC-5' anticodon, which different from that usually adopted in order would predict a C:A mismatch in the middle to allow such abnormal decoding to occur. position of the codon. Although an attempt Other potential natural suppressors ident• was made to explain the context effect of the ified include two calf tRNALcu species A, 3' to the suppressed termination codon, with anticodons 3' -GAC-5' and 3' -AAC-5', in terms of base-pairing with the (universal) both of which are able to suppress UAG in U, 5' to the anticodon, this has now been tobacco mosaic virus RNA in vitro. excluded, and at present the molecular basis It is convenient in this section to describe of the abnormal codon-anticodon inter• one case in which there is suppression of action is unknown [282]. certain cellular UGA codons, even though it In eukaryotic viruses there is no consistent does not have any regulatory significance. pattern to the suppressed termination codon Rather, it is the means by which the genetic and the following base: in Sinbis virus it is a code is extended to a twenty first amino acid, UGA codon followed by a C, in Moloney selanocysteine, an analogue of cysteine in murine leukaemia virus a UAG followed by which selenium replaces the sulphur atom The regulation of translation 581

[284, 309]. This amino acid is rare, but is termination that generates the 47-kDa y found in E. -coli and subunit of DNA polymerase in addition to mammalian glutathione peroxidase. In both the 71-kDa r subunit [290]; and frame• eukaryotes [285] and prokaryotes [286] a shifting is necessary in the normal translation specific tRNA, tRNAISer]Sec, has been of the E. coli mRNA for RF-2 [291]. Al• identified as responsible. In each case the though mutant tRNAs have been described has an anticodon of that can cause abnormal frameshifting, sequence 3' -ACU-5', although the U in the the natural frameshifting involves normal 'wobble' position undergoes a variety of tRNAs, and the focus of research has been modifications in different organisms. The on defining the mRNA context responsible. tRNAs have other unusual structural features The frameshifting in all the above cases is that presumably ensure that they only to the '-1' reading frame, with the exception recognize UGA codons in the specific con• of Ty and RF-2, where it is to the '+ 1' text of the mRNAs for proteins containing reading frame. In the case of shifting to the selanocysteine. In the case of formate '-1' frame, a common feature of the mRNA dehydrogenase this context includes 39 in the region immediately preceding the nucleotides 3', and 9 nucleotides 5', to the position of shift is a sequence containing codon [287], and a common feature of the tandemly repeated nucleotides. This has been context in different cases is a stem-loop shown to be necessary for the frameshifting, structure immediately 3' to the UGA and is thought to facilitate slippage of the codon. It has been shown in E. coli that the (normal) tRNA. The most striking examples tRNAISer]Sec is first charged with serine by of such 'slippery sequences' are A AAAAAC the normal seryl-tRNA synthetase, and the and U UUU UUA in certain of the retro• serine then undergoes enzymic modification viruses (the codon recognized by the tRNA to selanocysteine [288]. A special elongation involved in the frameshift is underlined), factor, structurally related to EF-Tu (section although G GGA AAC is also encountered. 12.4.2), is required to bind this tRNA to the It is not merely the tandem repetition that is ribosome [289]. important, as individual nucleotides are also essential in particular cases. However, such (b) Frameshifting [278, 279] slippery sequences, though necessary, are not sufficient for frameshifting. A second The synthesis of extended fusion-proteins by feature is required that is thought to cause frameshifting is encountered in retroviruses, the ribosome to pause for a sufficient length such as Rous sarcoma virus, mouse mammary of time to allow the frameshifting to occur. tumour virus and human immunodeficiency In some cases this takes the form of 3'• virus (HIV); in coronaviruses, such as the secondary structure: in many retroviruses avian infectious bronchitis virus (IBV); in (although not in HIV) a stem-loop, and in hepadnaviruses, such as mouse hepatitis IBV a pseudoknot (Fig. 2.27). In Ty (where virus; and in certain transposons such as Ty the slippery sequence, CUU AGG C, does (section 7.7.3). However, there are also two not conform to the pattern of tandem examples known in which frameshifting is repeats) the pausing is thought to be caused necessary for the synthesis of normal cellular by a preceding AGG codon, which is proteins. Frameshifting in the translation of decoded by a rare tRNAArg (cf. section the mRNA encoded by the E. coli dnaX 12.2.4). In the case of RF-2 two features are gene is responsible for the out-of-frame involved: a 5' -Shine and Dalgarno sequence 582 The translation of mRNA

ATP AMP ~

gpp pppG pppGpp ----::p-:-:rod....,.u-ct:---'1~ ppGpp I I I? spoT I . product p A I I + Stringent factor ppGp• ppG Fig. 12.36 Pathways for the synthesis and degradation of guanine nucleotides involved in the 'stringent response'.

and a UGA termination codon immediately 12.10 OTHER FUNCTIONS OF THE 3' to the CUU at which the shift occurs. The PROTEIN SYNTHETIC APPARATUS mechanism by which the Shine and Dalgarno sequence can facilitate frameshifting in this case is obscure, although pausing at the 12.10.1 Ribosomes and the stringent termination codon is easier to envisage. reaction Indeed, the rationale for the frameshifting here is thought to be to regulate the syn• In addition to their role in protein synthesis, thesis of RF-2 through control of the ratio of the ribosomes of bacteria such as E. coli frameshifting to termination. Termination are able to synthesize guanine nucleotides would be expected to vary inversely with the which act as 'alarmones' to integrate the cellular concentration of RF-2, which is regulation of RNA and protein synthesis. required to read the UGA codon. It is This phenomenon (which is not found in curious, however, that there is no compar• eukaryotes) occurs when amino acids are able regulation in the case of RF-1. limiting, and is known as the 'stringent Finally, we should mention some cases response' [293, 294). It results in the selective where the frameshifting does not involve inhibition of the synthesis of rRNA and slippage of the tRNA and ribosome, but tRNA, but not of the bulk of mRNA; hence 'hopping' over considerable distances. An avoiding wasteful synthesis of more ribo• example of this is gene 60 of bacteriophage somes and tRNA, and conserving the amino T4, where 50 nucleotides are bypassed with acids provided by protein turnover for the almost 100% efficiency. No rationalization synthesis of essential proteins. of this apparent translational perversity has Certain mutations preventing the stringent yet been proposed, but some requirements response map to the relA locus. This encodes for its occurrence have been defined, and the 'stringent factor', which is required for these include matching sequences at either the ribosome to respond to the presence of a side of the region bypassed, and a particular deacylated tRNA cognate to the codon in region of nascent peptide [292). the A-site with the synthesis of guanosine 5'- Other functions of the protein synthetic apparatus 583

triphosphate 3 '-diphosphate (pppGpp) in an coat protein, and it has been suggested that 'idling' reaction (Fig. 12.36). There is further the presence of this protein in the replicase non-ribosomal conversion of pppGpp to might either be connected with the temporal yield what is probably the more physiologi• lagging of coat protein synthesis behind cally important 'alarmone', guanosine 5'• replicase synthesis, or to play a role in diphosphate 3'-diphosphate (ppGpp); and preventing collision between replicase and other guanine nucleotides are generated ribosomes on the RNA [231, 232]. during the degradation of these so-called An additional function is suggested in 'magic spots'. normal cells for the particular eukaryotic The concentration of ppGpp generally ribosomal protein that is termed S27a in correlates well with the inhibition of stable mammals and S37 in yeast. In all eukaryotes RNA synthesis, but there has been a long examined this protein is generated by post• controversy regarding the effects of this translational processing of a larger species nucleotide in vitro [294, 295]. It is a striking which has N-terminal sequences specifying reflection of the shift of scientific interest the protein [296]. Ubiquitin is (and funding) from prokaryotic to eukaryotic perhaps best known for its role in targetting systems that, over 20 years after its dis• proteins for degradation [297], but in certain covery, the mechanism of action of ppGpp is circumstances can apparently have the still unknown. opposite effect. There is evidence to suggest that the ubiquitin portion remains with the ribosomal protein until it is incorporated 12.10.2 Other functions of individual into the ribosome [298], and it has been ribosomal proteins and elongation found that a yeast mutant with the ubiquitin factors tail removed from the ribosomal protein is defective in assembling the 40S ribosomal In prokaryotes, there are situations in which subunit [299]. It is therefore possible that the certain ribosomal proteins and elongation ubiquitin sequence acts as a 'chaperonin' for factors are recruited for other roles. It was this protein or, a more provoking thought, mentioned in section 12.9.1 that the RNA plays a role in the assembly of the 40S sub• phages encode a replicase enzyme; however unit as a whole. this also requires three , host subunits in order to replicate the RNA 'minus' strand. These are EF-Tu, EF-Ts, and ribosomal 12.10.3 Other functions of protein Sl. This may merely be a case of aminoacyl-tRNA synthetases the bacteriophage adapting for its own advantage the RNA-binding properties of The ribosome is not the only component EF-Tu and ribosomal protein Sl. In the case of the protein synthetic machinery with the of EF-Tu, it has been suggested that this may capacity to synthesize a nucleotide in an alter• involve specific binding to putative tRNA• native to the reaction normally catalysed. Iike features at the 3' -end, although the Under abnormal conditions such as heat RNA phages do not show the strong re• shock or oxidative stress, the aminoacyl semblance to tRNA seen in certain plant adenylate produced in the first step of the viruses (section 12.10.3). For ribosomal aminoacyl-tRNA synthetase reaction can protein S1 there is the fact that it is able to react with A TP to give diadenosine 5' ,5"'-P1, inhibit the synthesis of the bacteriophage p4-tetraphosphate, AppppA [300]: 584 The translation of mRNA

(amino acid-pA)E + pppA cell walls [303]. A more recently described J, related function [304] appears to be in the post-translational tagging of the N-terminus E amino acid AppppA + + of proteins with chemically basic amino Other related nucleotides have also been acids to prepare them for proteolysis by the reported to be synthesized by this reaction, ubiquitin system [297]. which can occur in both prokaryotes and Somewhat more esoteric is the involve• eukaryotes. The function of the nucleotide ment of tRNA Glu in the synthesis of £5- is still in dispute, however. It has been aminolevulinate from glutamate in plants, suggested that AppppA may act as an where it is an important precursor of chloro• 'alarmone' to induce the heat shock response phyll [305]. For its NADPH-linked reduction [300], but this idea is not supported by ex• to glutamate 1-semialdehyde, glutamate periments with E. coli mutants in the apaH must be in the form of Glu-tRNA. As there gene, which specifies the hydrolase that seems no obvious chemical advantage in this degrades the nucleotide [301]. it has been suggested that the use of the More recently it has been found that tRNA may be a mechanism for co-ordinating specific aminoacyl-tRNA synthetases are chloroplast protein synthesis and chlorophyll components of the -splicing machinery synthesis. of fungal mitochondria [302]. It is possible The above examples all involve the amino to produce a mutant Tyr-tRNA synthetase acid attached to the tRNA. However, there that has lost its ability to charge tRNA are two virus-related examples involving with Tyr, but still functions in the splicing tRNA structures which, although apparently reaction. However, this does not exclude a quite different, may be 'molecular fossils' of role in RNA recognition of those parts of a role of this structure that preceded protein the enzyme already evolved to recognize biosynthesis. The first example is the use features of tRNA Tyr, especially as we have of particular cellular tRNAs by different already seen that the acceptor stem, with its retroviruses to prime reverse transcription, 5'- and 3' -termini, is one of the features that where a specific sequence on the tRNA are recognized by certain tRNA-synthetases hybridizes to the viral 5'-LTR [306], as has (section 12.3.2). The possibility that such already been mentioned in section 6.14. At recognition is a very ancient feature of life is first sight the second example is quite dis• discussed in the final section of this chapter. similar from the first. It is the role of the tRNA-like structures that are found at the 3'-ends of the genomic RNAs of certain 12.10.4 Other functions of tRNAs and plant viruses, such as tobacco mosaic virus tRNA-Iike structures [307]. These can be aminoacylated by specific amino acids (His in the virus mentioned), In addition to their role on the ribosome, but for many years no plausible function specific tRNAs are important for several could be suggested for them. An attractive other cellular processes. Some of these hypothesis has now been proposed by appear quite natural extensions of the major Weiner and Maizels [308], in which it is contemporary role of tRNAs. Thus, certain suggested that the tRNA structures might aminoacyl-tRNAs act as donors of amino serve a telomeric function (section 6.16.4). acids in the synthesis of the peptide com• This would be through their recognition by ponents of the peptidylglycans of bacterial the tRNA nucleotidyl transferase, which References 585 adds CCA to the 3' -end of tRNA transcripts 9 Kleinkauf, H. and von Dohren, H. (1983) (and the virus RNAs) - a process formally Trends Biochem. Sci., 8, 281. similar to the addition to telomeres of the 10 Crick, F. H. C. (1967) Proc. R. Soc. London, Ser. B, 167, 331. polymeric CmAn sequences by telomerases. 11 Woese, C. R. (1967) The Genetic Code. The point of convergence between these The Molecular Basis for Genetic Expression, two examples is Weiner and Maizels' further Harper and Row, New York. speculation that tRNAs may have evolved 12 Jukes, T. H. (1977) in Comprehensive Bio• from the 3' -terminal structures of primitive chemistry (eds M. Florkin and E. H. Stotz), Elsevier, Amsterdam, vol. 24, p. 235. RNA genomes in the 'RNA world' believed 13 Marshall, R. E., Caskey, C. T. and to have existed before proteins (and protein Nirenberg, M. (1967) Science, 155, 820. synthesis) emerged; and which the proteins 14 Goodman, H. M., Abelson, J., Landy, A., that subsequently took over the major part Brenner, S. and Smith, J.D. (1968) Nature of the replicative function would have had to (London), 217, 1019. be able to recognize. In this case the use of 15 Marcker, K. A. and Sanger, F. (1964) J. Mol. Bioi., 8, 835. extra-genomic tRNA by retroviruses might 16 Dube, S. K., Marcker, K. A., Clark, be more than just fortuitous use of a cellular B. F. C. and Cory, S. (1968) Nature RNA for its base-complementarity to their (London), 218, 232. genome, but reflect an ancient ability of the 17 Cory, S., Marcker, K. A., Dube, S. K. and replicative enzymes of the virus to recognize Clark, B. F. C. (1968) Nature (London), 220, 1039. the tertiary structure of the tRNA molecule. 18 Marcker, K. A., Clark, B. F. C. and Indeed, as the original replicative enzymes Anderson, J. S. (1966) Cold Spring Harbor in an RNA world would have been RNA, a Symp. Quant. Bioi., 31, 279. specific base-pairing interaction would also 19 Delk, A. S. and Rabinowitz, J. C. (1974) have been a feature of their interaction with Nature (London), 252, 106. 20 Adams, J. M. (1968) J. Mol. Bioi., 34, 571. the 3' -end of the genome. 21 Miller, C. M., Strauch, K. L., Kukral, A. M., Miller, J. L., Wingfield, P. T. et at. (1987) Proc. Natl. Acad. Sci. USA, 84, REFERENCES 2718. 22 Sherman, F., Stewart, J. W. and Tsunasawa, 1 Stryer, L. A. (1989) Biochemistry (3rd edn), S. (1985) BioEssays, 3, 27. Freeman, New York. 23 Crick, F. H. C. (1966) J. Mol. Bioi., 19, 2 Dintzis, H. M. (1961) Proc. Nat/. Acad. Sci. 548. USA, 47, 247. 24 Boren, T., Elias, P., Samuelsson, T., 3 Thach, R. E., Cecere, M.A., Sundrararajan, Claesson, C., Barciszewska, M. et al. (1992) T. A. and Doty, P. (1965) Proc. Nat[. Acad. J. Bioi. Chern. Sci. USA, 54, 1167. 25 Offengand, J. (1982) in Protein Biosynthesis 4 Terzaghi, E., Okada, Y., Streisinger, G., in Eukaryotes (ed. R. Perez-Bercoff), Emrich, J., Inouye, M. and Tsugita, A. Plenum, New York, p. 1. (1966) Proc. Natl. Acad. Sci. USA, 56, 500. 26 Lustig, F., Boren, T., Guindy, Y., Elias, P., 5 Lodish, H. F. and Jacobsen, M. (1972) J. Samuelsson, T. et al. (1989) Proc. Nat[. Bioi. Chern., 247, 3622. Acad. Sci. USA, 86, 6873. 6 Heywood, S. M., Dowben, R. M. and Rich, 27 Hillen, W., Egert, E., Lindner, H. J. and A. (1967) Proc. Nat/. Acad. Sci. USA, 57, Gassen, H. G. (1978) FEBS Lett., 94, 361. 1002. 28 Yokoyama, S., Watanabe, T., Marao, K., 7 Lacroute, F. and Stent, G. (1968) J. Mol. Ishikura, H., Yamaizumi, Z. et al. (1985) Bioi., 35, 165. Proc. Natl. Acad. Sci. USA, 82, 4905. 8 Knopf, P. M. and Lamfrom, H. (1965) 29 Martin, R. P., Sibler, A.-P., Gehrke, Biochim. Biophys. Acta, 95, 398. C. W., Kuo, K., Edmonds, C. G. et at. 586 The translation of mRNA

(1990) Biochemistry, 29, 956. 53 Griffin, B. E., Jarman, M., Reese, C. B., 30 Sibler, A. P., Dirheimer, G. and Martin, Sulston, J. E. and Trentham, D. R. (1966) R. P. (1985) Nucleic Acids Res., 13, 1341. Biochemistry, 5, 3638. 31 Hunter, W. N., Brown, T. and Kennard, 0. 54 Zachau, H. G. and Feldman, H. (1965) (1987) Nucleic Acids Res., 15, 6589. Prog. Nucleic Acid Res. Mol. Bioi., 4, 217. 32 Grosjean H. J., de Henau S. and Crothers, 55 Yarus, M. (1988) Cell, 55, 739. D. M. (1978) Proc. Natl. Acad. Sci. USA, 56 Schimmel, P. (1991) Trends Biochem. Sci., 75, 610. 16, 1. 33 Lagerkvist, U. (1978) Proc. Natl. Acad. Sci. 57 Rould, M. A., Perona, J. J., Soli, D. and USA, 75, 1759. Steitz, T. A. (1989) Science, 246, 1135. 34 Samuelsson, T., Axberg, T., Boren, T. and 58 Perona, J. J., Swanson, R. N., Rould, Lagerkvist, U. (1983) J. Bioi. Chern., 258, M. A., Steitz, T. A. and Soli, D. (1989) 13178. Science, 246, 1152. 35 Caron, F. (1990) Experientia, 46, 1106. 59 Chapeville, F., Lipmann, F., von 36 Osawa, S., Muto, A., Ohama, T., Andachi, Ehrenstein, G., Weisblum, B., Ray, W. J. Y., Tanaka, R. et al. (1990) Experientia, 46, et al. (1962) Proc. Nat[. Acad. Sci. USA, 48, 1097. 1086. 37 Jukes, T. H. and Osawa, S. (1990) 60 Fersht, A. R. (1979) in Transfer RNA: Experientia, 46, 1117. Structure, Properties and Recognition (eds 38 Grosjean, H. and Fiers, W. (1982) Gene, P.R. Schimmel, D. Soli and J. N. Abelson), 18, 199. Cold Spring Harbor Laboratory Monograph 39 Konigsberg, W. and Godson, G. N. (1983) Series, p. 247. Proc. Natl. Acad. Sci. USA, 80, 687. 61 Lengyel, P. (1974) in Ribosomes (eds M. 40 lkemura, T. (1981) J. Mol. Bioi., 146, 1. Nomura, A. Tissieres and P. Lengyel), Cold 41 Bennetzen, J. L. and Hall, B. D. (1982) Spring Harbor Laboratory Monograph J. Bioi. Chern., 257, 3026. Series, p. 13. 42 Garel, J.-P. (1976) Nature (London), 260, 62 Weissbach, H. (1979) in Ribosomes: 805. Structure, Function and (eds 43 Carter, P. ·w., Bartkus, J. M. and Calvo, G. Chambliss, G. R. Craven, J. Davies, J. M. (1986) Proc. Natl. Acad. Sci. USA, K. Davis, L. Kahan and M. Nomura), 83, 8127. University Park Press, Baltimore, p. 377. 44 Hoekma, A., Kastelein, R. A., Vasser, M. 63 Clark, B. F. C. and Petersen, H. U. (1984) and de Boer, H. A. (1987) Mol. Cell. Bioi., Gene Expression. The Translational Step 7, 2914. and its Control, Munskgaard, Copenhagen. 45 Goddard, J. P. (1977) Prog. Biophys. Mol. 64 Grunberg-Manago, M. (1980) in Ribosomes: Bioi., 32, 233. Structure, Function and Genetics (eds 46 Schevitz, R. W., Podjarny, A. D., G. Chambliss, G. R. Craven, J. Davies, Krishnamachari, N., Hughes, J. J. and K. Davis, L. Kahan and M. Nomura), Sigler, P. B. (1979) Nature (London), 278, University Park Press, Baltimore, p. 445. 188. 65 Maitra, U., Stringer, E. A. and Chaudhuri, 47 Woo, N. H., Roe, B. A. and Rich, A. A. (1982) Annu. Rev. Biochem., 51, 869. (1980) Nature (London), 286, 346. 66 Gualerzi, C. 0. and Pon, C. L. (1990) Bio• 48 Westhof, E., Dumas, P. and Moras, D. chemistry, 29, 5881. (1985) J. Mol. Bioi., 184, 119. 67 Shine, J. and Dalgarno, L. (1974) Proc. 49 Quigley, G. J. and Rich, A. (1976) Science, Natl. Acad. Sci. USA, 71, 1342. 194, 796. 68 Kozak, M. (1983) Microbiol. Rev., 47, 1. 50 Hoagland, M. B., Stephenson, M. L., Scott, 69 Seong, B. L. and RajBhandary, U. L. J. F., Hecht, L. I. and Zamecnik, P. C. (1987) Proc. Natl. Acad. Sci. USA, 84, (1958) 1. Bioi. Chern., 231, 241. 8859. 51 Lagerkvist, U., Rymo, L. and Waldenstrom, 70 Kuechler, E. (1971) Nature (London) New J. (1966) J. Bioi. Chern., 241, 5391. Bioi., 234, 216. 52 Julius, D. J., Fraser, T. H. and Rich, A. 71 Morel-Deville, F., Vachon, G., Sacerdot, (1979) Biochemistry, 18, 604. C., Cozzone, A., Grunberg-Manago, M. References 587

et al. (1990) Eur. J. Biochem., 188, 605. and Evolution (eds W. E. Hill, A. Dahlberg, 72 Sacerdot, C., Fayat, G., Dessen, P., R. A. Garrett, P. B. Moore, D. Schlessinger Springer, M., Plumbridge, J. A. et al. and J. R. Warner), American Society for (1982) EMBO J., 1, 311. Microbiology, Washington, p. 393. 73 Reddy, P., Peterkofsky, A. and McKenny, 91 Capecchi, M. R. and Klein, H. A. (1969) K. (1985) Proc. Nat/. Acad. Sci. USA, 82, Cold Spring Harbor Symp. Quant. Bioi., 5656. 34, 469. 74 Gold, L., Stormo, G. and Saunders, R. 92 Mugola, E. J., Huazi, K. A., Garinger, A. (1984) Proc. Nat/. Acad. Sci. USA, 81, E. and Dahlberg, A. E. (1988) Proc. Nat/. 7061. Acad. Sci. USA, 85, 4162. 75 Butler, J. S., Springer, M. and Grunberg• 93 Kaempfer, R. (1974) in Ribosomes (eds Manago, M. (1987) Proc. Nat/. Acad. Sci. M. Nomura, A. Tissieres and P. Lengyel), USA, 84, 4022. Cold Spring Harbor Monograph Series, 76 Harris, R. J. and Pestka, S. (1977) in Mol• p. 679. ecular Mechanisms of Protein Biosynthesis 94 Naaktgeboren, N., Roobol, K. and Voorma, (eds H. Weissbach and S. Pestka), Academic H. 0. (1977) Eur. J. Biochem., 72, 49. Press, New York, p. 413. 95 Kohli, J. and Grosjean, H. (1981) Mol. 77 Miller, D. L. and Weissbach, H. (1977) in Gen. Genet., 182, 430. Molecular Mechanisms for Protein Biosyn• 96 Brown, C. M., Stockwell, P. A., Trotman, thesis (eds H. Weissbach and S. Pestka), C. N. A. and Tate W. P. (1990) Nucleic Academic Press, New York, p. 324. Acids Res., 18, 2079. 78 Bermek, E. (1978) Prog. Nucleic Acid Res. 97 Miller, J. H. and Albertini, A. M. (1983) J. Mol. Bioi., 21, 63. Mol. Bioi., 164, 59. 79 Clark, B. F. C. (1983) in DNA Makes RNA 98 Bossi, L. (1983) J. Mol. Bioi., 164, 73. Makes Protein (eds T. Hunt, S. Prentis 99 Perez-Bercoff, R. (1982) Protein Biosyn• and J. Tooze), Elsevier Biomedical, thesis in Eukaryotes, Plenum, New York. Amsterdam, p. 213. 100 Moldave, K. (1985) Annu. Rev. Biochem., 80 Ono, Y., Skoultchi, A., Klein, A. and 54, 1109. Lengyel, P. (1968) Nature (London), 220, 101 Pain, V. M. (1986) Biochem. J., 235, 625. 1304. 102 Rhoads, R. E. (1988) Trends Biochem. Sci., 81 Bourne, H. R., Sanders, D. A. and 13, 52. McCormick, F. (1991) Nature (London), 103 Caskey, C. T., Redfield, B. and Weissbach, 349, 117. H. (1967) Arch. Biochem. Biophys., 120, 82 La Cour, T. F. M., Nyborg, J., Thirup, S. 119. and Clark, B. F. C. (1985) EMBO J., 4, 104 Kozak, M. (1978) Cell, 15, 1109. 2385. 105 Kozak, M. (1989) J. Cell Bioi., 108, 229. 83 Jurnak, F. (1985) Science, 230, 32. 106 Darnborough, C., Legon, S., Hunt, T. and 84 Monro, R. E. (1967) J. Mol. Bioi., 26, 147. Jackson, R. (1973) J. Mol. Bioi., 76, 379. 85 Lake, J. A. (1977) Proc. Nat/. Acad. Sci. 107 Schreier, M. H. and Staehelin, T. (1973) USA, 74, 1903. Nature (London) New Bioi., 242, 35. 86 Nierhaus, K. H. (1990) Biochemistry, 29, 108 Cigan, A.M., Feng, L. and Donahue, T. F. 4997. (1988) Science, 242, 93. 87 Moazed, D. and Noller, H. F. (1989) Nature 109 Milburn, S. C., Duncan, R. F. and (London), 342, 142. Hershey, J. W. B. (1990) Arch. Biochem. 88 Caskey, C. T. (1977) in Molecular Mech• Biophys., 276, 6. anisms of Protein Biosynthesis ( eds H. 110 Nielsen, P. J. and Trachsel, H. (1988) Weissbach and S. Pestka), Academic Press, EMBOJ., 1, 2097. New York, p. 443. 111 Lazaris-Karatzas, A., Montine, K. S. and 89 Tate, W. P. (1984) in Peptide and Protein Sonenberg, N. (1990) Nature (London), Reviews (ed. M. T. W. Hearn), Marcel 345, 544. Dekker, New York, vol. 2, p. 173. 112 Smit-McBride, Z., Schnier, J., Kaufman, 90 Tate, W. P., Brown, C. P. and Kastner, B. R. J. and Hershey, J. W. B. (1989) J. Bioi. (1990) in The Ribosome. Structure, Fuction Chern., 264, 18527. 588 The translation of mRNA

113 Strauss, E. G., Rice, C. M. and Strauss, 135 Chambliss, G., Craven, G. R., Davies, J., J. H. (1984) Virology, 133, 92. Davis, K., Kahan, L. and Nomura, M. 114 Jackson, R. J., Howell, M. T. and (1980) Ribosomes: Structure, Function Kaminski, A. (1990) Trends Biochem. Sci., and Genetics, University Park Press, 15, 477. Baltimore. 115 Hunt, T. (1985) Nature (London), 316, 580. 136 Hardesty, B. and Kramer, G. (1986) Struc• 116 Hinnebusch, A. G. (1988) Trends Genet., 4, ture, Fuction and Genetics of Ribosomes, 169. Springer, New York. 117 Hann, S. R., King, M. W., Bentley, D. L., 137 Hill, W. E., Dahlberg, A., Garrett, R. A., Anderson, C. W. and Eisenman, R. N. Moore, P. B., Schlessinger, D. and Warner, (1988) Cell, 52, 185. J. R. (1990) The Ribosome. Structure, 118 Florkiewicz, R. Z. and Sommer, A. (1989) Function and Evolution. American Society Proc. Nat/. A cad. Sci. USA, 86, 3978. for Microbiology, Washington. 119 Bernards, A. and de Ia Monte, S. M. (1990) 138 Moore, P. B. (1988) Nature (London), 331, EMBO 1., 9, 2279. 223. 120 Meerovitch, K., Pelletier, J. and Sonenberg, 139 Bielka, H. (1982) The Eukaryotic Ribo• N. (1989) Genes Devel., 3, 1026. some, Springer, Berlin. 121 Macejak, D. G. and Sarnow, P. (1991) 140 Wool, I. G. (1990) in The Ribosome. Nature (London), 353, 90. Structure, Function and Evolution (eds 122 Riis, B., Rattan, S. I. S., Clark, B. F. C. W. E. Hill, A. Dahlberg, R. A. Garrett, and Merrick, W. C. (1990) Trends Biochem. P. B. Moore, D. Schlessinger and J. R. Sci., 15, 420. Warner), American Society for Micro• 123 Slobin, L. I. (1980) Eur. J. Biochem., 110, biology, Washington, p. 203. 555. 141 Lake, J. (1985) Annu. Rev. Biochem., 54, 124 Nagata, S., Nagashima, K., Tsunetsugu• 507. Yokota, Y., Fujimura, K., Miyazaki, M. 142 Frank, J., Verschoor, A., Wagenknecht, eta/. (1984) EMBO J., 3, 1825. T., Radermacher, M. and Carazo, J.-M. 125 Collier, R. J. (1975) Bacterial. Rev., 39, 54. (1988) Trends Biochem. Sci., 13, 123. 126 Van Ness, B. G., Howard, J. B. and 143 Yonath, A. and Wittmann, H. G. (1989) Bodley, J. W. (1980) J. Bioi. Chern., 255, Trends Biochem. Sci., 14, 329. 10710. 144 Milligan, R. A. and Unwin, P. N. T. (1986) 127 Lee, H. and Iglewski, W. J. (1984) Proc. Nature (London), 319, 693. Nat/. Acad. Sci. USA, 81, 2703. 145 Rich, A. (1974) in Ribosomes (eds M. 128 Neth, R., Monro, R. E., Heller, G., Nomura, A. Tissieres and P. Lengyel), Battaner, E. and Vazquez, D. (1970) FEBS Cold Spring Harbor Monograph Series, Lett., 6, 198. p. 871. 129 Goldstein, J. L., Beaudet, A. L. and 146 Olsen, H. McK., Grant, P. G., Cooperman, Caskey, C. T. (1970) Proc. Nat/. Acad. Sci. B. S. and Glitz, D. G. (1982) J. Bioi. USA, 67, 99. Chern., 257, 2649. 130 Craigen, W. J., Cook, R. G., Tate, W. P. 147 Olsen, H. McK., Lasater, L. S., Cann, and Caskey, C. T. (1985) Proc. Nat/. Acad. P. A. and Glitz, D. G. (1988) J. Bioi. Sci. USA, 82, 3616. Chern., 263, 15196. 131 Lee, C. C., Craigen, W. J., Muzny, D. M., 148 Kang, C. and Cantor, C. R. (1985) J. Mol. Harlow, E. and Caskey, C. T. (1990) Proc. Bioi., 181, 241. Nat/. A cad. Sci. USA, 87, 3508. 149 Noller, H. F. (1984) Annu. Rev. Biochem., 132 Kaempfer, R. (1969) Nature (London), 222, 53, 119. 950. 150 Maden, B. E. H., Khan, M.S. N., Hughes, 133 Wada, K., Aota, S., Tsuchiya, R., Ishibashi, D. G. and Goddard, J. P. (1977) Biochem. F., Gojobori, T. eta/. (1990) Nucleic Acids Soc. Symp., 42, 165. Res., 18, 2367. 151 Maden, B. E. H. (1980) Nature (London), 134 Nomura, M., Tissieres, A. and Lengyel, P. 288, 293. (1974) Ribosomes, Cold Spring Harbor 152 Brimacombe, R. (1984) Trends Biochem. Monograph Series, New York. Sci., 9, 273. References 589

153 Walker, T. A. and Pace, N. R. (1983) Cell, and J. R. Warner), American Society for 33, 320. Microbiology, Washington, p. 123. 154 Moazed, D., Stern, S. and Noller, H. F. 168 Oates, M. 1., Scheinman, A., Atha, T., (1986) J. Mol. Bioi., 187, 399. Schankweiler, G. and Lake, J. A. (1990) 155 Noller, H. F., Moazed, D., Stern, S., in The Ribosome. Structure, Function and Powers, T., Allen, P. N. et at. (1990) in Evolution (eds W. E. Hill, A. Dahlberg, The Ribosome. Structure, Function and R. A. Garrett, P. B. Moore, D. Schlessinger Evolution (eds W. E. Hill, A. Dahlberg, R. and J. R. Warner), American Society for A. Garrett, P. B. Moore, D. Schlessinger Microbiology, Washington, p. 180. and J. R. Warner), American Society for 169 Garrett, R. (1983) Trends Biochem. Sci., 8, Microbiology, Washington, p. 72. 189. 156 Brimacombe, R., Greuer, B., Mitchell, P., 170 Dahlberg, A. E. (1989) Cell, 57, 525. Osswald, M., Rinke-Appel, J. et at. (1990) 171 Dabbs, E. R., Hasenbank, R., Kastner, B., in The Ribosome. Structure, Function and Rak, K-H., Wartusch, B. et at. (1983) Mol. Evolution (eds W. E. Hill, A. Dahlberg, R. Gen. Genet., 192, 301. A. Garrett, P. B. Moore, D. Schlessinger 172 Crick, F. H. C. (1968) J. Mol. Bioi., 38, and J. R. Warner), American Society for 367. Microbiology, Washington, p. 93. 173 Darnell, J. E. and Doolittle, W. F. (1986) 157 Westhof, E., Romby, P., Romaniuk, P. J., Proc. Natl. A cad. Sci. USA, 83, 1271. Ebel, J. P., Ehresmann, C. et at. (1989) J. 174 Steiner, G., Keuchler, E. and Barta, A. Mol. Bioi., 207,417. (1988) EMBO J., 7, 3949. 158 Wittmann-Liebold, B. (1986) in Structure, 175 Hall, C. C., Johnson, D. and Cooperman, Function and Genetics of Ribosomes (eds B. (1988) Biochemistry, 27, 3983. B. Hardesty and G. Kramer), Springer, 176 Moazed, D. and Noller, H. F. (1989) Cell, New York, p. 326. 57, 585. 159 Wada, A. and Sako, T. (1987) J. Biochem. 177 Ehresmann, C., Ehresmann, B., Millon, R., (Tokyo), 101, 817. Ebel, J.-P., Nurse, K. and Ofengand, J. 160 Wilson, K. S., Appelt, K., Badger, J., (1984) Biochemistry, 23, 429. Tanaka, I. and White, S. W. (1986) Proc. 178 Li, M., Tzagoloff, A., Underbrink-Lyon, Natl. A cad. Sci. USA, 83, 7251. K. and Martin, N.C. (1982) J. Bioi. Chern., 161 Leijonmarck, M. and Liljas, A. (1987) J. 257, 5921. Mol. Bioi., 195, 555. 179 Beauclerk, A. A. D. and Cundliffe, E. 162 Lin, A., Wittmann-Liebold, B., McNally, J. (1987) J. Mol. Bioi., 193, 661. and Wool, I. G. (1982) J. Bioi. Chern., 257, 180 Denman, R., Negre, D., Cunningham, 9189. P. R., Nurse, K. eta/. (1989) Biochemistry, 163 Rich, B. E. and Steitz, J. A. (1987) Mol. 28, 1012. Cell. Bioi. , 7, 4065. 181 Helser, T. L., Davies, J. E. and Dahlberg, 164 Leader, D. P. (1980) in Molecular Aspects J. E. (1971) Nature (London) New Bioi., of Cellular Regulation (ed. P. Cohen), 233, 12. Elsevier/North-Holland, Amsterdam, vol. 182 Kurland, C. G., Jorgensen, F., Richter, A., 1, p. 203. Ehrenberg, M., Bilgin, N. et at. (1990) in 165 Traut, R. R., Lambert, J. M., Boileau, G. The Ribosome. Structure, Function and and Kenny, J. W. (1980) in Ribosomes. Evolution (eds W. E. Hill, A. Dahlberg, Structure, Function and Genetics (eds R. A. Garrett, P. B. Moore, D. Schlessinger G. Chambliss, G. R. Craven, J. Davies, and J. R. Warner), American Society for K. Davis, L. Kahan and M. Nomura), Microbiology, Washington, p. 513. University Park Press, Baltimore, p. 89. 183 Gorini, L. (1974) in Ribosomes (eds 166 Moore, P. B. and Capel, M. (1988) Annu. M. Nomura, A. Tissieres and P. Lengyel), Rev. Biophys. Biophys. Chern., 17, 349. Cold Spring Harbor Monograph Series, 167 Stoffler-Meilicke, M. and St6ffler, G. (1990) p. 791. in The Ribosome. Structure, Function and 184 Ozaki, M., Mizushima, S. and Nomura, M. Evolution (ed. W. E. Hill, A. Dahlberg, R. (1974) Nature (London), 222, 333. A. Garrett, P. B. Moore, D. Schlessinger 185 Hirsch, D. (1971) J. Mol. Bioi., 58, 439. 590 The translation of mRNA

186 Hopfield, J. J. (1974) Proc. Nat/. Acad. Sci. 205 Pugsley, A. P. (1989) , USA, 71, 4135. Academic Press, San Diego. 187 Ninio, J. (1975) Biochimie, 51, 587. 206 Walter, P., Gilmore, R. and Blobel, G. 188 Thompson, R. C. (1988) Trends Biochem (1984) Cell, 38, 5. Sci., 13, 91. 207 Verner, K. and Schatz, G. (1988) Science, 189 Woese, C. R. and Wolfe, R. S. (1985) The 241, 1307. Bacteria, vol. 8, Archaebacteria, Academic 208 Rothman, J. E. (1989) Nature (London), Press, New York. 340, 433. 190 Kandler, 0. and Zillig, W. (1986) Archae• 209 Rapoport, T. A. (1991) Nature (London), bacteria '85, Gustav Fischer, New York. 349, 107. 191 Woese, C. R. and Fox, G. E. (1977) Proc. 210 Gierasch, L. M. (1989) Biochemistry, 28, Nat/. A cad. Sci. USA, 74, 5088. 923. 192 Wittmann-Liebold, B., Kopke, A. K. E., 211 Connolly, T. and Gilmore, R. (1989) Cell, Arndt, E., Kromer, W., Hatakeyama, T. 57, 599. et a/. (1990) in The Ribosome. Structure, 212 Bieker, K. L. and Silhavy, T. J. (1990) Function and Evolution (eds W. E. Hill, A. Trends Genet., 6, 329. Dahlberg, R. A. Garrett, P. B. Moore, D. 213 Inouye, S., Soberon, X., Franceschini, T., Schlessinger and J. R. Warner), American Nakamura, K., Itakura, K. et al. (1982) Society for Microbiology, Washington, Proc. Nat/. Acad. Sci. USA, 79, 3438. p. 598. 214 Schelness, G. S. and Blobel, G. (1990) J. 193 Mathieson, A. T., Auer, J., Ramirez, C. Bioi. Chern., 265, 9512. and Bock, A. (1990) in The Ribosome. 215 Poritz, M. A., Bernstein, H. D., Straub, Structure, Function and Evolution (eds K., Zopf, D., Wilhelm, H. and Walter, P. W. E. Hill, A. Dahlberg, R. A. Garrett, (1990) Science, 250, 1111. P. B. Moore, D. Schlessinger and J. R. 216 Muesch, A., Hartmann, E., Rohde, K., Warner), American Society for Micro• Rubartelli, A., Sitia, R. et al. (1990) Trends biology, Washington, p. 617. Biochem. Sci., 15, 86. 194 Borst, P., Grivell, L. A. and Groot, 217 Wickner, W. T. and Lodish, H. F. (1985) G. S. P. (1984) Trends Biochem. Sci., 9, 48. Science, 230, 400. 195 Attardi, G. (1985) Int. Rev. Cytol., 93, 93. 218 Rogers, J., Early, P., Carter, C., Calaine, 196 Chomyn, A. and Attardi, G. (1987) Curr. K., Bond, M., Hood, L. et al. (1980) Cell, Top. Bioeng., 15, 295. 20, 303. 197 Gray, M. W. (1989) Trends Genet., 5, 294. 219 Yost, C. S., Hedgpeth, J. and Lingappa, 198 O'Brien, T. W., Denslow, N. D., Anders, V. R. (1983) Cell, 34, 759. J. C. and Courtney, B. C. (1990) Biochim. 220 Singer, S. J., Maher, P. A. and Yaffe, M. P. Biophys. Acta, 1050, 174. (1987) Proc. Nat/. Acad. Sci. USA, 84, 199 Smith, A. E. and Marcker, K. A. (1968) J. 1960. Mol. Bioi., 38, 241. 221 Pelham, H. R. B. (1990) Trends Biochem. 200 De Bruijn, M. H. L. and Klug, A. (1983) Sci., 15, 483. EMBO J., 2, 1309. 222 Pfeffer, S. R. and Rothman, J. E. (1987) 201 Whitfeld, P. R. and Bottomley, W. (1983) Annu. Rev. Biochem., 56, 829. Annu. Rev. Plant Physiol., 34, 279. 223 Douglas, M. G., McCammon, M. T. and 202 Sugiura, M. (1989) Annu. Rev. Cell Bioi., 5, Vassarotti, A. (1986) Microbiol. Rev., 50, 51. 166. 203 Subramanian, A. P., Smooker, P. M. 224 Hartl, F.-Z. and Neupert, W. (1990) (1990) in The Ribosome. Structure, Func• Science, 247, 930. tion and Evolution (eds W. E. Hill, A. 225 Baker, K. P. and Schatz, G. (1991) Nature Dahlberg, R. A. Garrett, P. B. Moore, D. (London), 349, 205. Schlessinger and J. R. Warner), American 226 Keegstra, K. and Bauele, C. (1988) Society for Microbiology, Washington, p. BioEssays, 9, 15. 655. 227 Smeekens, S., Weisbeek, P. and Robinson, 204 Das, R. C. and Robbins, P. W. (1988) C. (1990) Trends Biochem. Sci., 15, 73. Protein Transfer and Organelle Biogenesis, 228 Borst, P. (1989) Biochim. Biophys. Acta, Academic Press, San Diego. 1008, 1. References 591

229 Dingwall, C. and Laskey, R. A. (1986) 254 Baglioni, C. (1979) Cell, 17, 255. Annu. Rev. Cell Bioi., 2, 367. 255 Lengyel, P. (1982) in Protein Biosynthesis in 230 Roberts, B. (1989) Biochim. Biophys. Acta, Eukaryotes (ed. R. Perez-Bercoff), Plenum, 1008, 263. New York, p. 459. 231 Hindley, J. (1973) Prog. Biophys. Mol. 256 Pestka, S., Langer, J. A., Zoon, K. C. and Bioi., 26, 269. Samuel, C. E. (1987) Annu. Rev. Biochern., 232 Zinder, N. D. (1975) RNA Phages, Cold 56, 727. Spring Harbor Monograph Series. 257 Samuel, C. E. (1991) Virology, 183, 1. 233 Min Jou, W., Haegeman, G., Ysebaert, M. 258 Kwong, A. D. and Frenkel, N. (1987) Proc. and Piers, W. (1972) Nature (London), 237, Nat/. Acad. Sci. USA, 84, 1926. 82. 259 Kwong, A. D. and Frenkel, N. (1989) 234 Kozak, M. (1988) Mol. Cell. Bioi., 8, 2737. J. Virol., 63, 4834. 235 Kabat, D. and Chappell, M. R. (1977) J. 260 Cleveland, D. W. (1988) Trends Biochern. Bioi. Chern., 252, 2684. Sci., 13, 339. 236 Lawson, T. G., Cladaras, M. H., Ray, 261 Farrell, P. J., Balkow, K., Hunt, T. and B. K., Lee, K. A., Abrahamson, R. D. Jackson, R. J. (1977) Cell, II, 187. et al. (1988) J. Bioi. Chern., 263, 7266. 262 Ochoa, S. (1983) Arch. Biochern. Biophys., 237 Fu, L., Ye, R., Browder, L. W. and 223, 325. Johnston, R. N. (1991) Science, 251, 807. 263 Voorma, H. 0., Goumans, H., Amesz, H. 238 Huang, W.-1., Hansen, L. J., Merrick, and Benne, R. (1983) Curr. Topics Cell W. C. and Jagus, R. (1987) Proc. Nat/. Regul., 22, 51. Acad. Sci. USA, 84, 6359. 264 Schneider, R. J. and Shenk, T. (1987) 239 Romaniuk, P. J., Lowary, P., Wu, H.-N, Annu. Rev. Biochern., 56, 317. Stormo, G. and Uhlenbeck, 0. C. (1987) 265 Sonenberg, N. (1990) New Biologist, 2, 402. Biochemistry, 26, 1563. 266 Hershey, J. W. B. (1989) J. Bioi. Chern., 240 Nomura, M., Gourse, R. and Baughman, 264, 20823. G. (1984) Annu. Rev. Biochem., 53, 75. 267 Nairn, A. C. and Palfrey, H. C. (1987) 241 Draper, D. E. (1989) Trends Biochem. Sci., J. Bioi. Chern., 262, 17299. 15, 73. 268 Ryazanov, A. G., Shestakova, E. A. and 242 Starzyk, R. M. (1988) Trends Biochern. Natapov, P. G. (1988) Nature (London), Sci., 13, 119. 334, 170. 243 Haile, D. J., Hentze, M. W., Rouault, 269 Levenson, R. M. and Blackshear, P. J. T. A., Harford, J. B. and Klausner, R. D. (1989)1. Bioi. Chern., 264,19984. (1989) Mol. Cell. Bioi., 9, 5055. 270 Leader, D. P. (1980) in Molecular Aspects 244 Lin, J.-J., Daniels-McQueen, S., Patino, of Cellular Regulation, vol. 1 (ed. P. Cohen), M. M., Gaffield, L., Walden, W. E. et al. Elsevier/North Holland, Amsterdam, (1990) Science, 247, 74. p. 203. 245 Milliner, E. W. and Kiihn, L. C. (1988) 271 Kozma, S. C., Ferrari, S. and Thomas, G. Cell, 53, 815. (1989) Cell Signal., I, 219. 246 Raghow, R. (1987) Trends Biochern. Sci., 272 MacConnell, W. P. and Kaplan, N. 0. 12, 358. (1982) J. Bioi. Chern., 257, 5359. 247 Cleveland, D. W. and Yen, T. J. (1989) 273 Vidales, F. J., Robles, M. T. S. and Balleta, New Biologist, I, 121. J. P. G. (1984) Biochemistry, 23, 390. 248 Brawerrnan, G. (1981) Crit. Rev. Biochern., 274 Jones, S. W., Erikson, E., Blenis, J., 10, 1. Maller, J. L. and Erikson, R. L. (1988) 249 Jackson, R. J. and Standart, N. (1990) Cell, Proc. Nat/. Acad. Sci. USA, 85, 3377. 62, 15. 275 Kozma, S. C., Ferrari, S., Bassand, P., 250 Wickens, M. (1990) Trends Biochem. Sci., Siegmann, M., Totty, N. and Thomas, G. 15, 277. (1990) Proc. Nat/. Acad. Sci. USA, 87, 251 Bernstein, P. and Ross, J. (1989) Trends 7365. Biochem. Sci., 14, 373. 276 Clemens, M. (1990) Trends Biochern. Sci., 252 Shaw, G. and Kamen, R. (1986) Cell, 46, 15, 172. 659. 277 Sonnenberg, N. (1990) Curr. Top. Micro• 253 Malter, J. S. (1989) Science, 246, 664. bioi. Imrnunol., 161, 23. 592 The translation of mRNA

278 Valle, R. P. C. and Morch, M.-D. (1988) 300 Varshavsky, A. (1983) Cell, 34, 711. FEBS Lett., 235, 1. 301 Farr, S. B., Arnosti, D. N., Chamberlin, 279 Atkins, J. F., Weiss, R. B. and Gestland, M. J. and Ames, B. N. (1989) Proc. Nat!. R. F. (1990) Cell, 62, 413. A cad. Sci. USA, 86, 5010. 280 Murgola, E. F. (1985) Annu. Rev. Genet., 302 Benne, R. (1988) Trends Genet., 4, 181. 19, 57. 303 Soffer, R. L. (1974) Adv. Enzymol., 40, 91. 281 Eggertson, G. and Soil, D. (1988) Micro• 304 Ferber, S. and Ciechanover, A. (1987) bioi. Rev., 52, 354. Nature (London), 326, 808. 282 Ayer, D. and Yarus, M. (1986) Science, 305 Kannangara, C. G., Gough, S. P., Bruyant, 231, 393. P., Hoober, J. K. et al. (1988) Trends 283 Kuchino, Y., Beier, H., Akita, N. and Biochem. Sci., 13, 139. Nishimura, S. (1987) Proc. Natl. Acad. Sci. 306 Litvak, S. and Arya, A. (1982) Trends USA, 84, 2668. Biochem. Sci. , 7, 361. 284 Chambers, I. and Harrison, P. R. (1987) 307 Haenni, A.-L., Joshi, S. and Chapeville, F. Trends Biochem. Sci., 12, 255. (1982) Prog. Nucleic Acid Res., 27, 85. 285 Lee, B. J., Rajagopalan, M, Kim, Y. S., 308 Weiner, A. M. and Maizels, N. (1987) Proc. You, K. H., Jacobson, B. et al. (1990) Mol. Nat!. Acad. Sci. USA, 84, 7383. Cell. Bioi., 10, 1940. 309 Burk, R. F. (1991) FASEB J., 5, 2274. 286 SchOn, A., Bock, A., Orr, G., Sprinzl, M. 310 Meyer, F., Schmidt, H. J., Plumper, E., and Soil, D. (1989) Nucleic Acids Res., 18, Haslik, A., Mersmann, G. et al. (1991) 1989. Proc. Nat!. Acad. Sci. USA, 88, 3758. 287 Zinoni, F., Heider, J. and Bock, A. (1990) 311 Lingappa, V. R. (1991) Cell, 65, 527. Proc. Nat!. Acad. Sci. USA, 87, 4660. 312 Herman, P. K., Stark, J. H., DeModena, 288 Leinfelder, W., Forchhammer, K., Veprek, J. A. and Emr, S. D. (1991) Cell, 64, 425. B., Zehelein, E. and Bock, A. (1990) Proc. 313 Silver, P. A. (1991) Cell, 64, 489. Nat!. A cad. Sci. USA, 87, 543. 314 Gunnery, S., Rice, A. P., Robertson, H. 289 Forchhammer, K., Leinfelder, W. and and Mathews, M. B. (1990) Proc. Nat!. Bock, A. (1969) Nature (London), 342,453. A cad. Sci. USA, 87, 8687. 290 Tsuchihashi, Z. and Kornberg, A. (1990) 315 Beattie, E., Tartaglia, J. and Paoletti, E. Proc. Nat!. Acad. Sci. USA, 87, 2516. (1991) Virology, 183, 419. 291 Craigen, W. J. and Caskey, T. (1987) Cell, 316 Lee, T. G., Tomita, J., Hovanessian, A. G. SO, 1. and Katze, M. G. (1990) Proc. Nat!. Acad. 292 Weiss, R. B., Huang, W. M. and Dunn, Sci. USA, 87, 6208. D. M. (1990) Cell, 62, 117. 317 Heus, H. A. and Pardi, A. (1991) Science, 293 Edlin, G. and Broda, P. (1968) Bacterial. 253, 191. Rev., 32, 206. 318 Mitchell, P., Osswald, M., Schueler, D. and 294 Gallant, J. A. (1979) Annu. Rev. Genet., Brimacombe, R. (1990) Nucleic Acids Res., 13, 393. 18, 4325. 295 Lamond, A. I. and Travers, A. A. (1985) 319 Altman, S. (1990) J. Bioi. Chern., 265, Cell, 40, 319. 20053. 296 Warner, J. R. (1989) Nature (London), 338, 320 Powers, T. and Noller, H. F. (1991) EMBO 379. J.' 10, 2203. 297 Finley, D. and Varshavsky, A. (1985) 321 Leberman, R., Hartlein, M. and Cusack, S. Trends Biochem. Sci., 10, 343. (1991) Biochim. Biophys. Acta, 1089, 287. 298 Miiller-Taubenberger, A., Graak, H.-R., 322 Rould, M. A., Perona, J. J. and Steitz, Grohmann, L., Schleicher, M. and Gerisch, T. A. (1991) Nature (London), 352, 213. G. (1989) J. Bioi. Chern., 264, 5319. 323 Ruff, M., Krishnaswamy, S., Boeglin, M., 299 Finley, D., Bartel, B. and Varshavsky, A. Poterszman, A., Mitschler, A. et al. (1991) (1990) in The Ribosome. Structure, Func• Science, 252, 1683. tion and Evolution (eds W. E. Hill, 324 Mattheakis, L., Vu, L., Sor, F. and A. Dahlberg, R. A. Garrett, P. B. Moore, Nomura, M. (1989) Proc. Nat!. Acad. Sci. D. Schlessinger and J. R. Warner), USA, 86, 448. American Society for Microbiology, Washington, p. 636.