Knowledge is accumulating rapidly about the structural details of protein synthesis in contemporary systems. This remarkable body of knowledge may eventually shed light on how this process evolved in prebiotic times.
Exobiological research is oriented toward finding those conditions which led to the establishment of life on this planet. In this work one starts with an environment comparable to that believed to have existed in prebiotic times, and using natural processes we attempt to create those chemicals which have given rise to living systems. Research in this field may be used to interpret the evolutionary state of planets to be explored in the Solar System, especially in assessing the molecular complexity found on planets and its relevance to a particular stage of biological evolution.
A number of workers in exobiological research have made significant advances. Using a reducing atmosphere, Miller, Urey, and many other investigators have established conditions that give rise to the production of amino acids, nucleotides, sugars, and other biological molecules. We now have a number of plausible prebiological chemical models that not only aid our understanding of the development of these monomeric constituents of living systems, but even lend some insight into the production of polynucleotides, polysaccharides, and polypeptides. Work on polymerization has proceeded in several directions. Most interesting in the polynucleotide field is the development of complementary polynucleotides that use the hydrogen bonding found in the double-helical nucleic acids of contemporary biological systems (Lohrmann et al., 1980). This system contains its own molecular constraints, which allow a single polynucleotide strand to aid in selecting individual  monomeric nucleotides to build up a double helix. Strand separation and repetition of this process lead to a plausible prebiological model for nucleic acid replication, and this reproduces an important stage in the early evolution of life on this planet.
There are many ways to polymerize amino acids, including the use of heat alone. Although such prebiotic polymers may have had some role in the early establishment of living systems, they are not biological products because they do not yield multiple copies of molecules with the same amino acid sequence. The central event of contemporary biological systems is the polynucleotide-directed synthesis of polypeptides. In this system, information found in nucleotide sequences directs the sequential assembly of amino acids. The next step in the development of a prebiological origin of life is the elucidation of this system. There is today a great chasm in our understanding. Although we have a general comprehension of the role of messenger RNA codons in bringing about the sequential assembly of amino acids in ribosomes, we do not know the detailed nature of the interactions that made this possible. The contemporary system outlined in figure 1. In attempting to create a model for protein synthesis is of amino acid polymerization in protein synthesis, it is not at all clear in which direction efforts should be expanded. One of the central research goals is the definition of the physical nature of the interactions which occur between messenger RNA (mRNA) and transfer RNA (tRNA). We need to understand the important constraints found in contemporary biological systems to gain insight into the nature of the models needed to develop prebiological research. We need to know how information coded in the randomly polymerized nucleic acids can be translated into polypeptides in the absence of ribosomes, or at least their protein constituents. If we can create a prebiotic, protein-free system of polynucleotide-directed polypeptide synthesis, we will in principle have reached a stage sensitive to the selection pressures that are the basis of biological evolution.
Transfer RNA is the central actor in protein synthesis today. It is also an ancient component of biological systems. It is likely that it arose some 4 billion years ago in the early prebiotic period, at which time the increasing complexity of organic molecules led to the creation of polynucleotide chains that had the capacity to carry out molecular self-replication. Although the nucleic acids are effectively selected for carrying genetic information and carrying out self-replication, they cannot express genetic information. The major mode for expressing genetic information is the synthesis of proteins, which, through their varied amino acid side chains, are able to provide the large variety of complex chemical environments needed to create the catalytic activities and structural assemblies that are the molecular basis of living systems. The tRNA molecule acts at the crossroads between the information-containing polynucleotide chains and the proteins that express genetic information. At one end of the tRNA molecule, the three anticodon bases....
 ....interact with mRNA; at the other end, the molecule is attached to the growing polypeptide chain during protein synthesis (fig. 1). This molecule is probably as ancient a component of biological systems as the system of expressing genetic information , through the polymerization of amino acids. Here we consider the two types of tRNA molecules found in protein-synthetic systems and speculate on how the system may have started.
The tRNA molecules are active in protein synthesis where they participate in two major activities: aminoacylation, which takes place on the surface of the aminoacyl synthetases, and protein synthesis, which takes place inside ribosomes. In a sense, there is a paradox in trying to understand this central biochemical function of tRNA. On the one hand, the tRNA molecules must be sufficiently unique to be discriminated by different ammoacyl synthetases. On the other hand, they must be sufficiently identical that they can all go through the same ribosomal apparatus during protein synthesis.
We would like to understand the manner in which this twofold chore is carried out in an effective and relatively error-free mode. Further, a distinction has to be made between the initiator tRNA, which lays down the first amino acid, and chain-elongation tRNAs, which add subsequent residues. Initiator tRNAs go to the ribosomal P site (fig. 1), while chain-elongation tRNAs go to the ribosomal A site.
Holley et al. (1965) reported the first nucleotide sequence of a tRNA molecule. They noted that there were segments of the polynucleotide chain that appeared complementary, which suggested that the chain folded back on itself. One of these foldings has given rise to the cloverleaf diagram in which sequences are represented as a series of stems and loops, the stems largely composed of complementary bases with Watson-Crick hydrogen bonding much as in the DNA double helix. As additional sequences were reported, it became apparent that this cloverleaf folding was expressing something of a fundamental nature concerning the molecule. At present, over 100 different tRNA molecules have been sequenced; figure 2 shows the cloverleaf sequence for yeast phenylalanine tRNA (yeast tRNAPhe) (Sprinzl et al., 1980). There are a number of invariant and semi-invariant nucleotides in all tRNA sequences. The reasons behind this large number of conserved residues remained unknown until the elucidation of the three-dimensional structure of yeast tRNAPhe (Rich and Rajbhandary, 1976).
The cloverleaf arrangement plus the large number of invariant positions imply considerable constraints on the three-dimensional form of the molecule. Prior to the elucidation of the three-dimensional structure of yeast tRNAPhe, a number of attempts were made to anticipate the conformation by building models of tRNA folding. These models usually maintained the stem regions as double helices, while additional tertiary base-base interactions were used to stabilize the three-dimensional conformation. None of the....
....proposed models was correct. The principal reason was that they all emphasized Watson-Crick hydrogen bonding in the tertiary interactions between residues that were not in stems. There are 9 base-base tertiary hydrogen-bonding interactions in the three-dimensional structure of yeast tRNAPhe, as indicated by the solid lines in figure 2. Only one of these is a Watson-Crick interaction; the other 8 involve alternative types of hydrogen bonding. Watson-crick hydrogen bonds are of great utility in building a regular helical structured but the hydrogen-bonding potential of bases is much more varied  and it is utilized in many different ways in building the highly irregular and nonrepeating interactions found in the globular coiling of the tRNA polynucleotide chain.
Transfer RNA molecules were first crystallized in 1968 independently in several different laboratories. A large number of species were found to form crystals, and this gave rise to optimism that the three-dimensional structure would be known in a relatively short time. However, it was not generally appreciated that all the crystal forms had one defect in common: they were all disordered. In protein crystallography, it is recognized that one must have an x-ray diffraction pattern with a resolution of between 2 and 3 Å to trace the polypeptide chain. To fix the position of the bases, sugars, and phosphate groups, a comparable resolution is required.
The origin of the disorder in tRNA crystals is not readily apparent. A large part of it must result from the polyelectrolytic nature of the molecule. These molecules have from 75 to 90 negative charges, and the exact ordering of the tRNA molecules in the crystal lattice is very sensitive to the nature and number of cations found in the crystallizing mixture. It is likely that the polyelectrolytic nature of the molecule gives rise to frequent mistakes in the building of the lattice. After much effort, we reported in 1971 that if one added the oligo cation spermine to yeast tRNAPhe, an orthorhombic crystal could be formed which yielded an x-ray diffraction pattern with a resolution of nearly 2 Å (Kim et al., 1971).
In early 1973, the folding of the polynucleotide chain was traced from an electron density map at 4-Å resolution (Kim et al., 1973). The tracing was possible because the electron-dense phosphate groups could be seen even at this low resolution. The folding of the chain (fig. 3) was highly unusual and had not been anticipated by any of the model builders. The stem regions, which had been assumed to be in the form of RNA double helices from the sequence data, were indeed in that form; but they had an unusual organization. The molecule was L-shaped, with the acceptor stem and T stem forming one limb of the L while the D stem and anticodon stem formed the other limb. The 3' terminal adenosine to which the amino acid was attached was at one end of the L while the anticodon was at the other end, some 76 Å away. The molecule is fairly flat, about 20 to 25 Å thick. Although the coiling of the chain could be seen at 4-Å resolution, the details of the tertiary interactions awaited the results of a higher resolution analysis.
In 1974 the analysis was extended to 3 Å and revealed a number of tertiary interactions (figs. 2 and 3). In particular, it showed the detailed....
 .....manner in which the T and D loops interact to stabilize the corner of the molecule (Kim et al., 1974a). Robertus et al. (1974) also presented 3-Å results for the same spermine-stabilized yeast tRNAPhe in the monoclinic crystal lattice. This showed a virtually identical folding of the molecule. It can be seen (fig. 3) that many of the base-base tertiary hydrogen-bonding interactions involve the invariant or semi-invariant nucleotides. This suggests that the structure of yeast tRNAPhe may be a generalized model for understanding the structure of all tRNAs (Kim et al.,1974b).
The analysis has since been extended to 2.5-Å resolution, and crystallographic refinement calculations have been carried to the point where many details concerning the tertiary interactions in the molecule can be observed (Quigley and Rich, 1976).
The major structural unit in yeast tRNAPhe is the RNA double helix that forms the cloverleaf stems. These are very close to an RNA helix with approximately 11 residues per turn. This conformation is maintained due to the presence of the ribose 2' OH group in the backbone. DNA lacks this group, and its double helix is substantially different.
The presence of a 2' hydroxyl group in the ribose phosphate backbone contributes to the stabilization of the standard ribose 3' endo conformation generally found in RNA molecules. Its absence in the DNA molecule leads to a deoxyribose 2' endo conformation. This is responsible for the difference in the overall shape and form of the RNA double helix compared to the DNA double helix. The RNA double helix differs from the familiar DNA double helix in that the base pairs in the double helix are not perpendicular to the helix axis, but rather are tilted about 15°. Furthermore, they are not found on the axis of the helix, but rather are displaced away from the center. A hole approximately 6 Å in diameter extends through the center of the RNA double helix, so that the molecule looks more like a band wrapped around an imaginary cylinder than a double-helical twisted molecule that fills its axis. Consequently, the deep groove of the RNA double helix is extremely deep, whereas the narrow groove is extremely shallow.
The helix joining the acceptor stem and T stem seems almost uninterrupted. The sequence of yeast tRNAPhe shows a G-U pair in the acceptor stem. This introduces only a slight perturbation in the helix. Examination of the electron density map at 2.5 Å shows that these bases are held together by two hydrogen bonds in a typical "wobble" pairing (Crick,1966).
One of the outstanding features of the yeast tRNAPhe molecule is the fact that most of the bases are involved in base-stacking as well as hydrogen-bonding interactions. That this involves the bases in the double-helical stem  regions was not a great surprise, but the extent to which these same interactions also involve the bases in the nonhelical loop regions was of considerable interest. Virtually all the bases in the molecule are organized into two large base-stacking domains along each of the two limbs of the L-shaped molecule. The horizontal stacking domain in figure 3 includes bases of the acceptor and T loop and some from the D loop. Most of the remaining bases are involved in the vertical stacking domain, which extends down to and includes bases in the anticodon loop. Only 5 of the 76 bases are not involved in stacking interactions. The two dihydrouracil residues, D16 and 17 and G20, are not stacked, nor is U47 from the variable loop. These all come from regions with variable numbers of nucleotides, as can be seen in a survey of different tRNA sequences (Sprinzl et al., 1980). In addition, the 3'terminal adenosine A76 is unstacked. It is interesting that most of the unstacked bases are found in the segments of the polynucleotide chain that do not have constant numbers of nucleotides in all tRNA molecules. This suggests a structural explanation for this variability. These bases are found in regions in which there is a bulging or arching of the polynucleotide chain backbone. This arching makes it possible to accommodate variable numbers of nucleotides in different tRNA molecules; a larger arch is likely to be found in molecules containing larger numbers of nucleotides. The structure thus provides information relative to the question of whether yeast tRNAPhe can serve as a model for all tRNAs. In this case, something special in the structure is found in regions of nucleotide variability.
Two regions in the yeast tRNAPhe molecule are marked by moderate structural complexity. One is the corner of the molecule where the T and D loops interact to stabilize the L-shaped conformation. The other is in the core of the molecule near the D stem, where there is considerable additional hydrogen bonding.
The T loop is organized in such a manner as to stabilize its interaction with the D loop and thereby maintain the two limbs of the molecule at approximately right angles to each other. This is accomplished using both detailed hydrogen-bonding and base-stacking interactions. All bases in the T loop are stacked parallel to the bases in the T stem, except U59 and C60, which are excluded from the stacking interactions . Bases U59 and C60 are oriented at right angles to the rest of the bases in the loop; they serve to nucleate the stacking interactions on which the vertical stacking domain in figure 3 is built.
In the corner of the molecule where the T and D loops come together, we see many features that have the overall effect of bringing about a rather  intricate or detailed fitting together of different components to stabilize the unusual conformation. Broadly, the tRNA molecule has the appearance of a molecule in which the double helix has been engineered to make a right angle turn. The corner of the molecule has several features that appear to be designed to stabilize this conformation.
The core of the molecule immediately beneath the T stem includes some complex hydrogen-bonding interactions. Four segments of polynucleotide chains come into this region; two of the chains are part of the D stem, and the others are from the variable loop and the section of chain joining the acceptor stem to the D stem (residues 8 and 9). These chains are hydrogen-bonded in several ways, including groups of three bases hydrogen-bonded in triplets.
The anticodon stem is an RNA double helix with about 11 residues per turn. The conformation of the bases in the anticodon loop is somewhat similar to that suggested by Fuller and Hodyson (1967) in that the three anticodon bases are at the end of a stacked series of bases at the 3' end of the loop. The constant residue U33 plays an interesting role in maintaining the conformation of the loop since it is hydrogen-bonded through N3 to the phosphate group of P36. Indeed, there is great similarity between the conformation of the polynucleotide chain in the region of the T loop and the conformation in the anticodon loop (Quigley and Rich, 1976). In both places the polynucleotide chain makes a sharp bend and a uridine residue (U33 or U55) plays a key role in stabilizing this sharp turn through the formation of a hydrogen bond to a phosphate residue on the other side of the loop.
The configuration of the anticodon bases is shown in figure 4 as viewed from the bottom of the molecule. The three anticodon bases have the form of a right-handed helix with approximately 8 residues per turn. They are in a conformation such that they can form hydrogen-bonding interactions with the codon. At present it is not clear whether the detailed anticodon conformation seen in the crystal is maintained when it interacts with mRNA.
The orthorhombic crystal of yeast tRNAPhe contains approximately 75% water. This suggests that the transformation from the crystalline state to a completely aqueous solution is not likely to have enormous structural consequences. The available evidence strongly supports this interpretation. A variety of experimental techniques have been used to correlate the structure observed in the crystal with that found in solution. All of these studies reach the same general conclusion: the structure appears to be the same in solution as it is in the crystalline state (Rich and Rajbhandary, 1976).
There is good reason to believe that the molecular structure of yeast tRNAPhe can be used as a model for understanding the molecular structure of all tRNAs (Kim et al.,1974b). Differences in the number of nucleotides in different tRNA sequences can be accommodated by variably sized arches looping out of the molecule. However, there is uncertainty about the conformation of those tRNAs that have very large variable loops containing 13 to 21 nucleotides These undoubtedly form a stem and loop structure that is likely to project away from the molecule, but further work will be necessary before the details of their conformation are known.
It is likely that some altered forms of hydrogen bonding will be found in other sequences. Basically, those hydrogen bonds in yeast tRNAPhe which do not involve invariant or semi-invariant positions are probably altered in  other tRNA molecules. In several cases it is possible to guess the detailed nature of these modifications, but many will need to be determined by further experimentation.
As mentioned above, the initiator tRNA is special in that it goes to the P site of the ribosome, whereas all other chain-elongation tRNAs go to the ribosomal A site. The initiator tRNA is specifically designed to recognize the initiator codon AUG and to insert a methionine in the N terminus of the polypeptide chain of all proteins. Initiator tRNAs have several features in common. Although their nucleotide sequences differ from species to species, they can nonetheless be substituted for each other in in vitro proteinsynthetic systems. Furthermore, when cleaved by S1 nuclease, which cuts single strands and loops of nucleic acids, a common cleavage pattern is found for all initiator tRNAs (Wrede et al., 1979). This pattern differs from the cleavage pattern that is common for all chain-elongation tRNAs.
The structural basis for this modification in cleavage pattern has recently been revealed in the three-dimensional structure of the E. coli , which has been solved to a resolution of 3.5 Å (Woo et al., 1980). A comparison of the folding patterns of the chain-elongation yeast tRNAPhe and the initiator E. coliis shown in figure 5. In the side views, it can be seen that the general folding patterns are very similar, except a difference in the conformation at the 3' acceptor end of E. coli . The 3' OH end is folded over, and this may be associated with a modification of the hydrogen-bonding pattern in the acceptor stem of that molecule. An interesting and apparently functional change is found in the folding of the anticodon loop. The three anticodon bases have a similar stacking, as seen in the upper part of figure 5; but there are differences, which can be seen in the lower part of figure 5. The tracing of the anticodon loop in yeast tRNAPhe is rather rounded. It can be seen that residue 33, which is emphasized in the figure, has a conformation such that the uracil residue is hydrogen-bonded to phosphate 36. In contrast, the nucleotide U33 has a different orientation in E. coli . The nucleotide is rotated relative to its position in yeast tRNAPhe. Instead of having U33 hydrogen bonding to P36, the altered conformation has the 2'-hydroxyl group of the ribose in a position where it may form this hydrogen bond. This change in the conformation of uridine is associated with a radically different folding of the anticodon loop. This in turn is no doubt responsible for the differences observed in the S1 nuclease cleavage pattern (Wrede et al., 1979).
 Initiator tRNAs go into the ribosomal P site, while chain-elongation tRNAs go into the ribosomal A site. Thus it is reasonable to suggest that these two conformations may be similar to the conformations seen in these ribosomal sites. If this is true, then the tRNA may undergo a conformational change in going from the A site to the P site in the ribosome. This process may be facilitated by the formation of a ternary complex involving the two tRNAs and the mRNA.
Knowledge of the three-dimensional structure of the tRNA-mRNA complex is essential to understand the manner in which these molecules may come together in a prebiotic system. Protein synthesis had to be initiated in the absence of a ribosome, although not necessarily in the absence of ribosomal RNA. Observing how this is done in contemporary biochemical systems may provide clues that allow us to carry out this reaction in a manner that seems plausible for a prebiotic environment.
Now we return to the paradox described at the beginning of this paper. What is the mechanism whereby nature differentiates between different tRNA species during aminoacylation, and at the same time allows all these molecules to pass through the same ribosomal apparatus? In short, where are the components of uniqueness and the components of commonality?
There has been considerable effort toward understanding the mechanism of aminoacylation and finding the regions of the tRNA molecule that may be recognized by synthetase enzymes. Different workers have suggested the acceptor stem, the D stem, or the anticodon (Kisselev and Favorova, 1974). Much experimental work on synthetase-tRNA interactions is in accord with the suggestion that synthetases recognize varying aspects of the tRNA structure along the diagonal side of the molecule (Rich and Schimmel, 1977). It is clear that the synthetases are probably all different, and there is not likely to be a common recognition system even if they all approach the same side of the tRNA molecule. It is likely that tRNA-synthetase recognition takes two parts. One is a recognition of the ribose-phosphate chain, which would be sensitive to the folding of the tRNA molecules. Second, there must be a recognition by the protein of specific bases in the double-helical stems or among the unpaired segments of the molecules. There are several ways in which proteins can recognize nucleic acid sequences (Seeman et al., 1976). It is likely that the basis of specificity resides in this detailed sequence recognition rather than in any conformational differences between tRNA molecules. Furthermore, there is an adequate basis for specificity in  these sequences. The number of nucleotides or base pairs recognized by the enzyme need not be very large to obtain the requisite specificity.
Far more puzzling is the question of what goes on in the ribosome. Our information in this field is scanty. As mentioned above, there are two sites within the ribosome, one occupied by the peptidyl-tRNA and the other by the aminoacyl-tRNA. It is likely that both sites are occupied at the same time, and for this reason the tRNA molecule may have been designed to have the form of a double helix that turns a corner. The L shape of the molecule may make it possible for two adjacent tRNAs to come close together at one end near their anticodons where they are interacting with adjacent codons of the mRNA. At the same time, the CCA acceptor ends may be able to come close together because of the L shape. These two CCA ends must come close to allow the ribosomal peptidyl transferase to transfer the peptide chain from one tRNA molecule to the other. It is not at all clear how this is accomplished, although it has been suggested that the two codons become unstacked during the reading process so that, in effect, the messenger "turns a corner" while it is read (Rich, 1974). This remains an exciting area for further research work.
Work by Erdmann and his colleagues (1973) has suggested that the T loop may become disengaged from the D loop inside the ribosome so that it can hydrogen-bond with the ribosomal 5S RNA. This interaction may be an important component in the translocation of tRNA from the aminoacyl site to the peptidyl site. Further, as mentioned above, the anticodon loop may differ in the A and P sites. Thus the tRNA molecule undergoes a conformational change within the ribosome. Determination of the nature of these events remains an important goal of research in this area. It is possible that this conformational change is triggered by codon-anticodon interactions, and this may be a consequence of the fact that the tRNA molecule as a whole exhibits long-range order (Rich and Rajbhandary, 1976). Altered interactions at one end of the molecule may give rise to a change in reactivity or in conformation at a more remote part of the molecule.
At present we have good models for understanding the early stages in the origin of life, including the accumulation of organic chemicals and the formation of nucleotides and polynucleotides. We do not have convincing models for the origin of polynucleotide-directed polypeptide formation; but it is reasonable to believe that the molecular mechanics of mRNA reading in the ribosome as well as peptide bond formation may provide a clue to its origin We can assume that the conformation of the anticodon does not have  a large change between the two ribosomal sites, but is similar to that shown in figure 5. We postulate a specific tRNA-tRNA interaction in the ribosome when they are in the aminoacyl and peptidyl sites, possibly using residue U33 in the ribosomal P site to bind to the tRNA in the A site. This interaction is such that it yields a spacing between the two anticodons that is directly responsible for the triplet code.
Thus, if two tRNAs and an mRNA form a transiently stabilized ternary complex, perhaps they would have the requisite geometry to bring aminoacylated ends of the tRNAs close together. An inefficient system of catalysis could then form a dipeptide, and the stage would be set for a repetition Knowledge of the structural details of protein synthesis in contemporary systems may shed light on how this process evolved in the prebiotic era.
- Crick, F. H. C.: Codon-Anticodon Pairing: the Wobble Hypothesis. J. Molecular Biol., vol. 19, Aug. 1966, pp. 548-555.
- Erdmann, V. A.; Sprinzl, M.; and Pongs, O.: The Involvement of 5S RNA in the Binding of tRNA to Ribosomes. Biochem. Biophys. Res. Commun. vol. 54, Oct. 1,1973, pp. 942-948.
- Fuller, W.; and Hodgson, A.: Conformation of the Anticodon Loop in tRNA. Nature, vol. 215, Aug. 19,1967, pp. 817-821.
- Holley, R. W.; Apgar, J.; Everett, G. A.; Madison, J. T.; Marquisse, M.; Merrill, S. H.; Penswick, J. R.; and Zamir, A.: Structure of a Ribonucleic Acid. Science, vol. 147, Mar. 1965, pp. 1462-1465.
- Kim, S. H.; Suddath, F. L.; Quigley, G. J.; McPherson, A.; Sussman, J. L.; Wang, A. H. J.; Seeman, N. C.; and Rich, A.: Three-Dimensional Tertiary Structure of Yeast Phenylalanine Transfer RNA. Science, vol. 185, Aug. 2,1974a, pp. 435-439.
- Kim, S. H.; Sussman, J. L.; Suddath, F. L.; Quigley, G. J.; Wang, A. H. J.; Seeman, N. C.; McPherson, A.; and Rich, A.: The General Structure of Transfer RNA Molecules. Proc. Natl. Acad. Sci. USA, vol. 71, Dec. 1974b, pp. 4970-4974.
 - Kim, S. H.; Quigley, G. J.; Suddath, F. L.; McPherson, A.; Sneden, D.; Kim, J. J.; Seeman, N. C.; Weinzierl, J.; and Rich, A.: Three-Dimensional Structure of Yeast Phenylalanine Transfer RNA: Folding of the Polynucleotide Chain. Science, vol.179, Jan.19,1973, pp. 285-288.
- Kim, S. H.; Quigley, G.; Suddath, F. L.; and Rich, A.: High-resolution X-Ray Diffraction Patterns of Crystalline Transfer RNA that Show Helical Regions. Proc. Natl. Acad. Sci. USA, vol. 68, April 1971 pp. 841-845
- Kisselev, L. Lev; and Favorova, O. Olga: Aminoacyl-tRNA Synthetases Some Recent Results and Achievements. Adv. Enzymol., vol. 40,1974 pp.141-238.
- Lohrmann, R.; Bridson, P. K.; and Orgel, L. E.: Efficient Metal-Ion CataIyzed Template-Directed Oligonucleotide Synthesis. Science, vol. 208 June 27,1980, pp.1464-1466.
- Quigley, G. J.; and Rich, A.: Structural Domains of Transfer RNA Molecules. Science, vol.194, Nov.19,1976, pp. 796-806.
- Rich, Alexander: How Transfer RNA May Move Inside the Ribosome. In: Ribosomes, M. Nomura, A. Tissieries, and P. Lengyel, eds. Cold Spring Harbor Laboratory, New York, 1974, pp. 871-884.
- Rich, A.; and Rajbhandary, U. L.: Transfer RNA -Molecular Structure Sequence and Properties. Ann. Rev. Biochem., vol. 45,1976, pp. 805860.
- Rich, A.; and Schimmel, P. R.: Structural Organization of Complexes of Transfer RNAs With Aminoacyl Transfer RNA Synthetases. Nucleic Acids Res., vol. 4, no. 5,1977, pp.1649-1665.
- Robertus, J. D.; Ladner, J. E.; Finch, J. T.; Rhodes, D.; Brown, R. S.; Clark, B. F. C.; and Klug, A.: Structure of Yeast Phenylalanine tRNA at 3 Å Resolution. Nature, vol. 250, Aug.16, 1974, pp. 546-551.
- Seeman, N. C.; Rosenberg, J. M.; and Rich, A.: Sequence-Specific Recognition of Double Helical Nucleic Acids by Proteins. Proc. Natl. Acad. Sci USA, vol. 73, March 1976, pp. 804-808.
- Sprinzl, M.; Grueter, F.; Spelzhaus, A.; and Gauss, D. H.: Compilation of tRNA Sequences. Nucleic Acids Res., vol. 8, Jan. 11,1980, pp. R1-23
 - Wrede, P.; Woo, N. H.; and Rich, A.: Initiator tRNAs Have a Unique Anti codon Loop Conformation. Proc. Natl. Acad. Sci. USA, vol. 76, July 1979, pp. 3289-3293.
- Woo, N. H.; Roe, B.; and Rich, A.: Three-Dimensional Structure of Eschechia Coli Initiator . Nature, vol. 286, July 24, 1980, pp. 346-351.