Nuclear magnetic resonance studies on transfer ribonucleic acid: assignment of AU tertiary resonances.

The hydrogen-bonded ring NH nuclear magnetic resonance (NMR) spectra of several transfer ribonucleic acid (RNA) species have been examined with particular emphasis on the extreme low-field portion. Betwen --13.8 and --15 ppm there are two extra resonances which are not derived from cloverleaf base pairs. A combined approach involving undermodified tRNAs, chemical modification, and hairpin fragment studies has assigned the T54--A58 resonance at --14.3 ppm in yeast tRNAPhe and Escherichia coli tRNA1 Val., the U8--A14 resonance has been assigned at --14.3 ppm, and the s4U8--A14 resonance in bacterial tRNAs has been assigned at --14.9 ppm. The T54--A58 resonance shifts between --14.3. and --13.8 ppm depending on the surrounding nucleotide sequence in the ribothymidine loop.     

High-resolution nuclear magnetic resonance determination of transfer RNA tertiary base pairs in solution. 1. Species containing a small variable loop.

Eight class I tRNA species have been purified to homogeneity and their proton nuclear magnetic resonance (NMR) spectra in the low-field region (-11 to -15 ppm) have been studied at 360 MHz. The low-field spectra contain only one low-field resonance from each base pair (the ring NH hydrogen bond) and hence directly monitor the number of long-lived secondary and tertiary base pairs in solution. The tRNA species were chosen on the basis of their sequence homology with yeast phenylalanine tRNA in the regions which form tertiary base pairs in the crystal structure of this tRNA. All of the spectra show 26 or 27 low-field resonances approximately 7 of which are derived from tertiary base pairs. These results are contrary to previous claims that the NMR spectra indicate the presence of resonances from secondary base pairs only, as well as more recent claims of only 1-3 tertiary resonances, but are in good agreement with the number of tertiary base pairs expected in solution based on the crystal structure. The tertiary base pair resonances are stable up to at least 46 degrees C. Removal of magnesium ions causes structural changes in the tRNA but does not result in the loss of any secondary or tertiary base pairs.     

High-resolution nuclear magnetic resonance determination of transfer RNA tertiary base pairs in solution. 2. Species containing a large variable loop.

The number of base pairs in the solution structure of several class III D3VN tRNA species from E. coli has been determined by analyzing the number of low-field (-15 to -11 ppm) proton resonances in their nuclear magnetic resonance spectra at 360 MHz. Contrary to previous reports indicating the absence of tertiary resonances, all the spectra exhibit the expected number of secondary base pair resonances plus approximately ten extra resonances derived from tertiary base pairs in the three-dimensional folding of these molecules. The possible origins of some of these tertiary resonances are discussed; none of the spectra exhibits the characteristic resonance of the 8-14 tertiary base pair seen in class I D4V5 tRNA spectra.     

Proton nuclear magnetic resonance study of the effect of pH on tRNA structure.

The low-field 220-MHz proton nuclear magnetic resonance (NMR) spectra of four tRNA molecules, Escherichia coli tRNAPhe, tRNA1Val, and tRNAfMet1, and yeast tRNAPhe, at neutral and mildly acidic pH are compared. We find a net increase in the number of resonances contributing to the -9.9-ppm peak (downfield from sodium 4,4-dimethyl-4-silapentanesulfonate) in three of these tRNAs at pH 6, while tRNAfMet1 does not clearly exhibit this behavior. The increase in intensity at this resonance position is half-completed at pH 6.2 in the case of yeast tRNAPhe. An alteration at the 5'-phosphate terminus is not involved, since removal of the terminal phosphate does not affect the gain in intensity at -9.9 ppm. Based on a survey of the tertiary interactions in the four molecules, assuming that they possess tertiary structures like that of yeast tRNAPhe at neutral pH, we tentatively attribute this altered resonance in E. coli and yeast tRNAPhe to the protonation of the N3 of the adenine residue at position 9 which results in the stabilization of the tertiary triple A23-U12-A9. This intepretation is supported by model studies on the lowfield proton NMR spectrum of AN oligomers at acid pH, which reveal an exchanging proton resonance at -9.4 ppm if the chain length N greater than or equal to 6.     

Identification of tertiary base pair resonances in the nuclear magnetic resonance spectra of transfer ribonucleic acid.

The low-field hydrogen-bond ring NH proton nuclear magnetic resonance (NMR) spectra of several transfer ribonucleic acids (tRNAs) related to yeast tRNAPhe have been examined in detail. Several resonances are sensitive to magnesium ion and temperature, suggesting that they are derived from tertiary base pairs. These same resonances cannot be attributed to cloverleaf base pairs as shown by experimental assignment and ring current shift calculation of the secondary base pair resonances. The crystal structure of yeast tRNAPhe reveals at least six tertiary base pairs involving ring NH hydrogen bonds, which we conclude are responsible for the extra resonances observed in the low-field NMR spectrum. In several tRNAs with the same tertiary folding potential and dihydrouridine helix sequence as yeast tRNAPhe, the extra resonances from tertiary base pairs are observed at the same position in the spectrum.     

Direct assignment of the dihydrouridine-helix imino proton resonances in transfer ribonucleic acid nuclear magnetic resonance spectra by means of the nuclear Overhauser effect

The NMR resonances from the hydrogen-bonded ring NH protons in the dihydrouridine stem of Escherichia colt tRNA1Val have been assigned by experiments involving the nuclear Overhauser effect (NOE) between adjacent base pairs. Irradiation of the 8-14 tertiary resonance produced a NOE to base pair 13. Irradiation of the CG13 ring NH produced NOEs to base pairs 12 and 14. Similarly, base pair 12 was shown to be dipolar coupled to 11 and 13, and base pair 11 was found to be coupled to 10 and 12. These sequential connectivities led to the assignment of CG13 at -13.05 ppm, UA12 at -13.84 ppm, CG11 at -12.23 ppm, and GC10 at -12.60 ppm. The results are compared with previous, less direct assignments for these four base pairs and with the expected proton positions from the crystal structure coordinates for this helix.     

Assignment of the magnetic resonances of the imino protons and methyl protons of Bombyx mori tRNA(GlyGCC) and the effect of ion binding on its structure.

The magnetic resonances in the low-field H-NMR spectra of Bombyx mori tRNA(GlyGCC), corresponding to the hydrogen-bonded imino protons of the helical stems and tertiary base pairs, could be tentatively assigned by means of the sequential nuclear Overhauser effects. While B. mori tRNA(GlyGCC) does not contain the G19C56 tertiary base pair, the D20G57 base pair exists between the D and T loops, which was not found in the X-ray crystal structure of yeast tRNA(Phe). The effects of Mg2+, spermine and temperature on the conformation of this tRNA have also been examined based on the behavior of the assigned resonance signals. Mg2+ stabilize the D and T stems and the tertiary structure between the D and T loops. Spermine affects the resonances of the D and anticodon stems, and A23G9, but does not stabilize them. While the acceptor stem melts sequentially from both ends (G7C66 and G1C72) with increasing temperature, the anticodon stem melts from only one end (G39C31) and the G26C44 base pair is the most stable. In the tertiary structure between the variable loop and D stem, G10G45 melts first and G22G46 last. Yeast tRNA(Phe) has also been examined, and the results were compared with those for B. mori tRNA(Gly).     

Isolation and characterization of two 5-fluorouracil-substituted Escherichia coli initiator methionine transfer ribonucleic acids

Escherichia coli initiator methionine tRNA labeled in vivo with 5-fluorouracil (FUra) has been isolated and characterized. The tRNA, with essentially all its uracil and uracil-derived minor bases replaced by FUra, was purified by sequential chromatography, first on diethylaminoethylcellulose (DEAE-cellulose), at pH 8.9, followed by chromatography on Sepharose 4B, using a reverse salt gradient, then on DEAE-Sephadex A-50, and finally on benzoylated DEAE-cellulose. The last step resolved two FUra-substituted tRNAfMet-iso-accepting species, each with a specific activity over 1500 pmol/A260. Kinetic analysis shows both are aminoacylated at the same rate; apparent KmS for the two are 0.92 and 0.94 microM, compared with 1.7 microM for normal tRNAfMet. Chromatographic differences between the two forms of fluorinated tRNAfMet persist after aminoacylation, and the two tRNAs are not interconverted by denaturation and renaturation. The isoacceptors have nearly identical nucleoside composition, and both contain 7-methylguanosine and 2'-O-methylcytidine as the only modified nucleosides. Analysis of complete RNase T1 digests of the two methionine tRNAs shows that they differ in only one oligonucleotide. The sequence 20FpApGp, derived from the dihydrouridine loop and stem region, which is found in one of the isoaccepting forms of the tRNA, is replaced by an oligonucleotide containing adenine and guanine, but no FUra in the other. A modified FUra, with the properties of a 5-fluoro-5,6-dihydrouracil derivative, is detected in this tRNA. 19F NMR spectra of the two species of FUra-substituted initiator tRNA show 9-10 resolved resonances for the 12 FUra residues incorporated. The spectra differ primarily in the shift of one peak in the form lacking the sequence 20FpApGp, from 4.8 ppm downfield from free FUra (= 0 ppm) to 14.9 ppm upfield from the standard.     

19F NMR of 5-fluorouracil-substituted transfer RNA transcribed in vitro: resonance assignment of fluorouracil-guanine base pairs.

5-Fluorouracil is readily incorporated into active tRNA(Val) transcribed in vitro from a recombinant phagemid containing a synthetic E. coli tRNA(Val) gene. This tRNA has the expected sequence and a secondary and tertiary structure resembling that of native 5-fluorouracil-substituted tRNA(Val), as judged by 19F NMR spectroscopy. To assign resonances in the 19F spectrum, mutant phagemids were constructed having base changes in the tRNA gene. Replacement of fluorouracil in the T-stem with cytosine, converting a FU-G to a C-G base pair, results in the loss of one downfield peak in the 19F NMR spectrum of the mutant tRNA(Val). The spectra of other mutant tRNAs having guanine for adenine substitutions that convert FU-A to FU-G base pairs all have one resonance shifted 4.5 to 5 ppm downfield. These results allow assignment of several 19F resonances and demonstrate that the chemical shift of the 19F signal from base-paired 5-fluorouracil differs considerably between Watson-Crick and wobble geometry.     

Nuclear magnetic resonance study of exchangeable protons in ferrocytochrome c

Five well-resolved exchangeable proton resonances have been observed in horse ferrocytochrome c in the low-field region between ?10.0 and ?12.0 ppm. A resonance at ?11.6 ppm is assigned to an amino proton of His 26 on the basis of its magnetic field position, pH dependence and its correlation with histidine at this position in eight species. A resonance at ?10.9 ppm observed in horse and donkey ferrocytochrome c is assigned to a hydrogen-bonded ?-amino proton of Lys 60. This resonance shifts upfield with increasing salt and decreasing pH, with the shifts increasing as one goes from chloride to bromide to iodide. With less assurance, a resonance at ?10.6 ppm is assigned to the indole amino of Trp 59 and a resonance at ?10.3 ppm to the amino proton of the His 18. These resonances have been used to study the binding of small anions to ferrocytochrome c, the results of which show that inorganic phosphate, ADP and ATP all bind in the immediate vicinity of His 26, and that the inorganic phosphate has the greatest effect upon the imidazole amino exchange time.     

Nuclear magnetic resonance study of hydrogen-bonded ring protons in oligonucleotide helices involving classical and nonclassical base pairs.

A study of the exchangeable ring nitrogen protons in aqueous solutions of oligonucleotide complexes involving Watson-Crick base pairs as well as Hoogsteen pairs and other nonclassical hydrogen bonding schemes shows that resolvable resonances in the low-field (-10 to -16 ppm from sodium 4,4-dimethyl-4-silapentanesulfonate) region can be detected in a variety of structures other than double stranded helices. Ring nitrogen proton resonances arising from the following hydrogen-bonding situations are reported: (1) AT and GC Watson-Crick base pairs in a self-complementary octanucleotide, dApApApGpCpTpTpT; (2) U-A-U base triples in complexes between oligo-U15 and AMP; (3) C-G-C+ base triples in complexes between oligo-C17 and GMP at acid pH; (4) s4U-A-s4U base triples in complexes between oligo-s4U15 and AMP, all of which involve both Watson-Crick and Hoogsteen base pairing to form triplexes; (5) C-C+ base pairing between protonated and unprotonated C residues in oligo-C17 at acid pH; and (6) I4 base quadruples in the four strand association among oligo-I at high salt. The behavior of the dA3G-CT3 helix is consistent with both fraying of the terminal base pairs and presence of intermediate states as the helix opens. In the monomer-oligomer complexes, under the conditions used here, the exchange appears to be governed by the dissociation rate of monomer from the complex. These findings suggest that those tertiary structure hydrogen bonds in tRNA involving ring nitrogen protons should have representative resonances in the low-field (11-16 ppm) proton NMR region in H2O.     

Changes in tertiary structure accompanying a single base change in transfer RNA. Proton magnetic resonance and aminoacylation studies of Escherichia coli tRNAMet f1 and tRNAMet f3 and their spin-labeled (s4U8) derivatives.

The properties of Escherichia coli tRNAMet f1 and tRNAMet f3 that differ by only one base change, m7G to A at position 47, have been compared structurally by proton magnetic resonance and functionally by the aminoacylation reaction. The NMR spectra of the two tRNA species in the region between 0 and 4 ppm below 4,4-dimethyl-4-silapentane-1-sulfonic acid (DSS) (methyl and methylene region) were the same except for the absence of the lowest field peak at 3.8 ppm in tRNAMet f3, thus unequivocally identifying this resonance at the methyl group of m7G47 of tRNAMet f1. The same resonance disappears in tRNAMet f1 spin-labeled at s4U8 and reappears in the diamagnetic reduced spin-labeled tRNAMet f1 from which the average distance between the spin-label and the methyl protons of m7G is estimated to be less than 15 A. The proximity of m7G47 but not T55 to s4U8 in the structure of E. coli tRNAMet f1 in solution is consistant with the crystallographic model for yeast tRNAPhe. A spectral comparison of the hydrogen-bond regions (11-14 ppm below DSS) of tRNAMet f1 and tRNAMet f3 reveals major shifts of four resonances previously assigned to tertiary hydrogen bonds. Of the four, the one at lowest field (14.8 ppm) had been assigned by chemical modification to the tertiary (s4U8-A14) hydrogen bond and the one at 13.3 ppm had been tentatively assigned to the tertiary hydrogen bond G23-m7G47 of the 13-23-47 triple. A more positive assignment of the G23-m7G47 at 13.3 ppm could be made from the additional evidence that this resonance, which was first observed in the difference spectrum between spin-labeled tRNAMet f1 and its reduced form, is the only one missing in the analogous difference spectrum of tRNAMet f3. At low ionic strength and in the absence of magnesium ions, the differences in the hydrogen-bonded region of the NMR spectra of tRNAMet f1 and tRNAMet f3 are much greater than in the presence of magnesium ions. The optimal magnesium concentration required for maximal initial velocities is also higher for tRNAMet f3 than for tRNAMet f1. The perturbation caused by the spin-label in destabilizing hydrogen bonds in the region between 13 and 14 ppm is greater for tRNAMet f3 than tRNAMet f1 but the distance relations for the hydrogen bonds in the region between 12 and 13 ppm (the major paramagnetic perturbations) are conserved in the two species. The disruption of one hydrogen bond relative to native tRNAMet f1 either by spin-labeling (s4U8-A14) or by substitution of m7G by A in tRNAMet f3 has little effect on the aminoacyl acceptor activity or the velocity of the aminoacylation reaction at optimal magnesium concentration, but the absence of both tertiary hydrogen bonds in the augmented D-helix region in the spin-labeled tRNAMet f3 results in approximately 60% reduction both in acceptance activity and in initial velocity of the aminoacylation reaction.     

1H NMR of valine tRNA modified bases. Evidence for multiple conformations.

Methyl and methylene protons of dihydrouridine 17 (hU), 6-methyladenosine 37 (M6A), 7-methylguanosine 46 (m7G), and ribothymidine 54 (rT) give clearly resolved peaks (220 MHz) for tRNA1val (coli solutions in D2O, 0.25 m NaCl, at 27 degrees C. Chemical shifts are generally consistent with a solution structure of tRNA1val similar to the crystal structure of tRNAphe (yeast). At least 3 separate transitions are observed as the temperature is raised. The earliest involves disruption of native tertiary structure and formation of intermediate structures in the m7G and rT regions. A second transition results in a change in structure of the anticodon loop, containing m6A. The final step involves unfolding of the m7G and rT intermediates and melting of the TpsiC helix. Low salt concentrations produce multiple, partially denatured conformations, rather than a unique form, for tRNA1val. Native structure is almost completely reformed by addition of Na+ but Mg2+ is required for correct conformation in the vicinity of m7G.     

Nuclear magnetic resonance studies on yeast tRNAPhe I. Assignment of the iminoproton resonances of the acceptor and D stem by means of Nuclear Overhauser Effect experiments at 500 MHz

Resonances of the water exchangeable iminoprotons of the acceptor and D stem of yeast tRNAPhe have been assigned by means of Nuclear Overhauser Effects (NOE's). Assignments were made for spectra recorded from tRNA dialysed against a buffer with 110 mM sodium and 5 mM magnesium ions and against a buffer with 430 mM sodium and no magnesium ions. Remarkable is the assignment of a resonance at 13.6 - 13.7 ppm to the iminoproton of C11G24. This assignment as well as those of G1C72, G3C70, U7A66, U12A23 and C13G22 are different from those made previously on the basis of less direct evidence. NOE experiments performed at 45 degrees C support the view that the D stem together with the tertiary interaction U8A14 is one of the most stable parts of the molecule in the presence of magnesium ions. A comparison of the spectra recorded under the two different buffer conditions shows that an excess of 320 mM sodium ions is not capable to force the tRNA in the same conformation as 5 mM magnesium ions can do.     

Structure of transfer RNA by carbon NMR: resolution of single carbon resonances from 13C-enriched, purified species

Methyl carbon-13 NMR spectra of purified tRNA species are presented for the first time. In addition, these spectra of tRNA species specific for phenylalanine, tyrosine, and cysteine exhibited the first resolution of single methyl carbon resonances. Carbon-13 enriched methyl groups of ribothymidine (T) and 7-methylguanosine (m7G) and the methylthio group of 2-methylthio-N6-(delta2-isopentenyl) adenosine (ms2i6A) were resolved. The T methyl signal of tRNAPhe shifted from 12.3 ppm at 45 degrees in the absence of added Mg2+ to 11.1 ppm at 30 degrees in the presence of 10mM MgCl2. The same change in conditions led to a 0.4 ppm shift of the m7G methyl signal in the opposite direction. The relative ease in obtainment of single carbon resonances of purified tRNA species, and display of the sensitivity of their chemical shifts to changes in local structure, are requisite criteria for 13C-NMR to be a useful technique in probing tRNA conformation and its changes during interaction with proteins and other nucleic acids.     

Localization of the structural change induced in tRNA fMET (Escherichia coli) by acidic pH.

We have compared the molecular mechanism of thermal unfolding for native tRNA fMet (Escherichia coli) and the denatured species produced by annealing at pH 4.3. Relaxation kinetic measurements reveal that the transitions assigned to melting of TphiC, anticodon, and acceptor stem helices at neutral pH remain essentially unaltered at pH 4.3, but the transition corresponding to coupled melting of tertiary structure and dihydrouridine helix is greatly affected. The Tm of this region is more than 20 degrees higher at pH 4.3 and it has a larger enthalpy formation than in the native state. The transition dynamics are also considerably changed. In contrast to the native structure, tRNA fMet1 and tRNA fMet3 have similar tertiary structure stabilities at pH 4.3. We conclude that the structural difference between native and acid-denatured forms is localized in the tertiary structure-dihydrouridine helix cooperative interaction region of the molecule.     

Correlations between fluorine-19 nuclear magnetic resonance chemical shift and the secondary and tertiary structure of 5-fluorouracil-substituted tRNA.

To complete assignment of the 19F nuclear magnetic resonance (NMR) spectrum of 5-fluorouracil-substituted Escherichia coli tRNA(Val), resonances from 5-fluorouracil residues involved in tertiary interactions have been identified. Because these assignments could not be made directly by the base-replacement method used to assign 5-fluorouracil residues in loop and stem regions of the tRNA, alternative assignment strategies were employed. FU54 and FU55 were identified by 19F homonuclear Overhauser experiments and were then assigned by comparison of their 19F NMR spectra with those of 5-fluorouracil-labeled yeast tRNA(Phe) mutants having FU54 replaced by adenine and FU55 replaced by cytosine. FU8 and FU12, were assigned from the 19F NMR spectrum of the tRNA(Val) mutant in which the base triple G9-C23-G12 substituted for the wild-type A9-A23-FU12. Although replacement of the conserved U8 (FU8) with A or C disrupts the tertiary structure of tRNA(Val), it has only a small effect on the catalytic turnover number of valyl-tRNA synthetase, while reducing the affinity of the tRNA for enzyme. Analysis of the 19F chemical shift assignments of all 14 resonances in the spectrum of 5-fluorouracil-substituted tRNAVal indicated a strong correlation to tRNA secondary and tertiary structure. 5-Fluorouracil residues in loop regions gave rise to peaks in the central region of the spectrum, 4.4 to 4.9 parts per million (p.p.m.) downfield from free 5-fluorouracil. However, the signal from FU59, in the T-loop of tRNA(Val), was shifted more than 1 p.p.m. downfield, to 5.9 p.p.m., presumably because of the involvement of this fluorouracil in the tertiary interactions between the T and D-loops. The 19F chemical shift moved upfield, to the 2.0 to 2.8 p.p.m. range, when fluorouracil was base-paired with adenine in helical stems. This upfield shift was less pronounced for the fluorine of the FU7.A66 base-pair, located at the base of the acceptor stem, an indication that FU7 is only partially stacked on the adjacent G49 in the continuous acceptor stem/T-stem helix. An unanticipated finding was that the 19F resonances of 5-fluorouracil residues wobble base-paired with guanine were shifted 4 to 5 p.p.m. downfield of those from fluorouracil residues paired with A. In the 19F NMR spectra of all fluorinated tRNAs studied, the farthest downfield peak corresponded to FU55, which replaced the conserved pseudouridine normally found at this position.     

Chemical and physical properties of mammalian mitochondrial aminoacyl-transfer RNAs. II. Analysis of 7-methylguanosine in mitochondrial and cytosolic aminoacyl-transfer RNAs.

The 7-methylguanosine (m7G) content of two individual mitochondrial tRNAs, labelled in the aminoacyl moiety was assayed by the specific cleavage of the tRNA at this nucleotide followed by electrophoretic analysis to identify the 3'-terminal fragment of the tRNA. Neither Syriam hamster mitochondrial tRNALeu nor tRNAMet were found to contain m7G. In contrast, cytosolic tRNAMetS were cleaved indicating the presence of m7G, apparently 27--28 and 29 nucleotides from their 3' terminus. Cystolic tRNALeu was not cleaved. These results are discussed in relationship to the reported low content of methylated nucleosides in mitochondrial 4 S RNA.     

Paramagnetic ion effects on the nuclear magnetic resonance spectrum of transfer ribonucleic acid: assignment of the 15--48 tertiary resonance.

We have studied the effects of Co2+ and Mn2+ ions on the low-field nuclear magnetic resonance (NMR) spectra of pure class 1 transfer ribonucleic acid (tRNA) species. With 1.2 mM tRNA in the presence of 15 mM MgCl2 discrete paramagnetic effects were observed for Co2+ at concentrations in the range 0.02--0.1 mM and for Mn2+ in the range 0.002--0.01 mM, indicating fast exchange of these cations with tRNA. Both of these cations paramagnetically relax the s4U8--A14 resonance as well as other resonances from proximal base pairs. The Co2+ site appears to be the same site on G15 which was observed crystallographically [Jack, A., Ladner, J. E., Rhodes, D., Brown, R. S., & Klug, A. (1977) J. Mol. Biol. 111, 315-328]; the initially occupied tight Mn2+ site is the cation site involving the phosphate of U8. There are three base pairs within 10 A of both sites, namely, G15--C48, A14--s4U8, and C13--G22; this has led to the assignment of the G15--C48 and C13--G22 resonances in the NMR spectrum [Jack, A., Ladner, J. E., Rhodes, D., Brown, R. S., & Klug, A. (1977) J. Mol. Biol. 111, 315--328; Holbrook, S. R., Sussman, J. L., Warrant, R. W., Church, G. M., & Kim, Sung-Hou (1977) Nucleic Acids Res. 4, 2811--2820; Quigley, G. J., Teeter, M. M., & Rich, A. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 64--68].     

High-resolution phosphorus nuclear magnetic resonance spectra of yeast phenylalanine transfer ribonucleic acid. Melting curves and relaxation effects.

In a continuation of our studies on structural effects on the 31P chemical shifts of nucleic acids, we present 31P NMR spectra of yeast phenylalanine tRNA in the presence and absence of Mg2+. Superconducting field (146 MHz) and 32-MHz 31P NMR spectra reveal approximately 15 nonhelical diester signals spread over approximately 7 ppm besides the downfield terminal 3'-phosphate monoester. In the presence of 10 mM Mg2+, most scattered and main cluster signals do not shift between 22--66 degrees C, thus supporting our earlier hypothesis that 31P chemical shifts are sensitive to phosphate ester torsional and bond angles. At 70 degrees C, all of the signals merge into a single random coil conformation signal. Similar effects are observed in the absence of Mg2+ except that the transition melting temperature is approximately 20 degrees C lower. Measured spin-lattice and spin-spin relaxation times reveal another lower temperature transition besides the thermal denaturation process. A number of the scattered peaks are shifted (0.2--1.7 ppm) and broadened between 22 and 66 degrees C in the presence of Mg2+ as a result of this conformational transition between two intact tertiary structures. The loss of the scattered peaks in the absence of Mg2+ occurs in the temperature range expected for melting of a tertiary structure. An attempt to simulate the 31P spectra of tRNA Phe based upon the X-ray crystallographically determined phosphate ester torsional agles supports the suggestion that the large shifts in the scattered peaks are due to bond angle distortions in the tertiary structure.