The function of pseudouridylic acid in transfer ribonucleic acid. IV. Cyanoethylation of fragments of Escherichia coli formylmethionine transfer ribonucleic acid and reconstitution of acceptor activity.

Abstract The function of pseudouridylic acid and of 4-thiouridylic acid in formylmethionine transfer RNA of Escherichia coli has been explored by cyanoethylation with acrylonitrile. By splitting the molecule into two complementary parts before chemical reaction, it was possible to modify specifically one of these nucleotides without reaction of the other. Recombination of the two fragments reconstituted acceptor activity only when the pseudouridylic acid-containing fragment was unmodified, and the rate of inactivation of this fragment could only be accounted for by reaction of its pseudouridylate residue. On the other hand, cyanoethylation of the 4-thiouridylate-containing fragment did not appreciably affect the methionine-accepting capacity of the recombined fragments. The pseudouridylate residue appears to be important in some way for proper transfer RNA-enzyme recognition but the 4-thiouridylate still has no known function.


SUMMARY
The ability of acrylonitrile to cyanoethylate 13 nucleosides was examined.
The variation of rate constant with pH showed that the rapidly reacting nucleosides must be in the anionic form for reaction to occur. The spectral, chromatographic, and electrophoretic properties of the derivatives were determined and used to assign the structures as given.
The stability of the cyanoethyl derivatives to alkali and acid was examined and found to be sufficient to allow the application of standard procedures for the detailed nucleotide analysis of cyanoethylated transfer RNA.
In the biosynthesis of protein, amino acyl transfer RNA faces three critical recognition problems.
They are recognition of the correct amino acyl RNA synthetase, recognition of ribosomal binding sites, and recognition of the correct codon in messenger RNA.
Presumably, specific conformations of tRNAl determine the selectivity of these interactions, and it seems reasonable to assume that the high content of unusual nucleotides found uniquely in tRNA may play an important role in maintaining this conformation.
Most of the unusual nucleotides occur infrequently however, less than once per chain on the average, with the exception of pseudouridylic acid and ribothymidylate which are found two or more times per chain on the average and * This work was supported in part by Grant GM-11506 from the National InsLitutes of Health. 1 The abbreviations used are: tRNA, transfer ribonucleic acid; CE-, cyanoethyl; CEZ-, dicyanoethyl; $-, pseudouridine.
once per chain, respectively (l-5). Dihydrouridylate is also present two or more times per chain in each of the five tRNAs for which the sequence is known.
While there is little doubt that even the nucleotides which occur in only a limited number of chains have an important function, those which are to be found in every chain would appear to be of more general significance.
Of these, pseudouridylic acid appears to be of greatest interest because of its location in tRNA and its chemical structure.
It is the only unusual nucleotide found both in a sequence common to all chains, GprTp#-pCpGp (6) and also elsewhere in the molecule.
In addition, its synthesis must represent a relatively complex task for the cell since in pseudouridylate the N-C glycosidic bond has been replaced by a C-C bond (7). Nevertheless, it is synthesized in all species and is found in large amounts only in tRNA (8,9) (with minor exceptions, l&12).
Reasoning that this unique structure and cellular localization of pseudouridylic acid must indicate an important role for this nucleotide, we have initiated studies intended to clarify its function in tRNA.
The approach chosen was to search for a method of chemically modifying pseudouridylate residues in intact tRNA in a specific and stable manner which would then allow subsequent assay for the retention of biological activity.
Cyanoethylation with acrylonitrile appears to be a suitable reaction for this purpose since (a) reaction occurs with pseudouridylate at the N1 atom blocking the very group which distinguishes pseudouridylate from uridylic acid, (b) it can be carried out under conditions which preclude phosphodiester bond hydrolysis, and (c) the reagent has minimal reactivity with most other nucleotides.
In this paper, we describe the cyanoethylation of nucleosides as a model for the cyanoethylation of intact tRNA.
Here we consider the kinetics and specificity of the cyanoethylation reaction for various nucleosides, the chemical characterization of the cyanoethyl nucleoside adducts, and their chemical stability. In the next paper of this series, we will describe the effect of cyanoethylation on the structure and function of tRNA.2 Preliminary reports of part of this work have already ap-peared (13)(14)(15). In addition, Chambers (16), Yoshida and Ukita (17), and Rake and Tener (18) have reported observations similar to some of the findings described here and in general agreement with them.

Materials
Nucleosides-Adenosine, guanosine, and cytidine were obtained from Sigma Chemical Company as Sigma grade material.
Inosine, xanthosine, thymidine, and orotidine were purchased from Calbiochem as A grade compounds.
On chromatographic analysis in Solvent B, the uridine was shown to contain less than 0.5 % of ribothymidine, but was contaminated with about 1% of pseudouridine.
Pseudouridine C and B (the p and (Y anomers, respectively) were purified by the method of Ofengand and Schaefer (19) from the commercial mixture of isomers also obtained from Calbiochem.
One sample of ribothymidine, prepared by the method of Fox et al. (20), was purchased from Cycle Chemical Corporation, and another sample was isolated from E. coli transfer RNA (21) by standard methods.
The RNA was digested in 0.3 N NaOH at 37" for 15 hours and the uridine-like nucleotides were isolated by passage through Dowex 50-H+ in 0.01 N HCl.
Salt was removed by the method of Uziel and Cohn (22). After enzymatic removal of the phosphate group, ribothymidine was isolated by preparative thin layer chromatography in Solvent B which readily separates ribothymidine, uridine, and pseudouridine from each other (Table I). Both the isolated and synthesized samples of ribothymidine appeared identical with each other in their ultraviolet absorption spectra at pH 2 and 12 and showed a single spot when chromatographed three times in Solvents B or C. Since both yielded N-cyanoethyl derivatives when treated with acrylonitrile (see below), they have not been further distinguished in the text. 4-Thiouridine disulfide, dihydrouridine, and Nl,a-dimethyl-C thiouracil were obtained from Cycle. 4-Thiouridine was prepared from the disulfide by reduction with mercaptoethanol, followed by chromatographic purification in butanol-Hz0 (86: 14) (23). The purified product possessed the spectral properties described by Lipsett (23) (X,,,, (pH 12), 317 mp; X,,, (pH 2), 331 w; &:,/A~,, = 0.93) and was chromatographically homogenous in Solvents A, B, and C of Table I.
N1,3-dimethyl-4-thiouracil, prepared by the method of Elion and Hitchings (25), was chromat.ographically homogenous in solvent systems A, B, and C (Table I). In addition, the ultraviolet spectrum was invariant over the pH range 2 to 12, showing the absence of any ionizable protons, and was very similar to a slightly displaced acid spectrum of 4-thiouridine (A,,,, 328 mp).

2-Thiouridine
was the gift of Dr. John Carbon. It was chromatographically homogenous in Solvents B and C. Cyanoethyl Nucleosides-In general, cyanoethyl derivatives were synthesized by the procedure previously described (13).
A reaction mixture was prepared containing nucleoside at a concentration of 0.04 M or less, 1 M acrylonitrile, and 0.05 M sodium carbonate buffer, pH 9.6 to 10. After incubation for an appropriate time at 30" in well sealed tubes, the reaction was terminated by the adjustment of the pH to 6 to 7 with HCl.
In some cases, carbonate buffer was omitted, and the pH held constant by periodic additions of acid or alkali.
Purification of the desired compound from the reaction mixture was achieved by thin layer chromatography in the solvent which gave optimal separation of the unreacted nucleoside from its cyanoethyl adduct (see Table I). In some cases, multiple development with the same or different solvents was used. CE-nucleosides were recovered by elution with water.
When subsequent spectral analysis showed the presence of a contaminant ionizable at alkaline pH, purification by chromatography in Solvents A or E was used, since in this alkaline medium, the unionizable CE-nucleosides separated well from ionizing contaminants.
In addition, the purity of all CE-nucleosides was assessed by chromatography in Solvents B, C, and E and was greater than 95%.
Other Chemicals-Acrylonitrile was obtained from Matheson, Coleman, and Bell. After redistillation it was stored in the cold. Except as otherwise indicated, all other chemicals used were of reagent quality.

Methods
Chromatography-Ascending thin layer chromatography on layers of cellulose (0.5 to 1 mm) was performed as described previously (19). In most cases, volatile solvents were used to permit multiple development with the same or different solvents in order to obtain improved resolution of compounds with similar RF values. The RF values in several solvents for all of the CEnucleosides and related compounds described in this paper are summarized in Table I.
Kinetic Measurements-The rate of reaction of nucleoside with acrylonitrile was determined by one of two methods. 1. Spectral technique.
This method was used for rapidly reacting compounds.
Buffer (20 ml), usually 0.05 M carbonate or pyrophosphate plus NaCl or MgOAc as indicated, was prepared at the desired pH and the tare weight was determined.
Acrylonitrile was then added with stirring, and after solution was achieved the flask was reweighed to determine the true amount of acrylonitrile added. In all cases, kinetic measurements were made with solutions approximately 1 M in acrylonitrile. (There is less than 0.5% volume contraction on mixing these quantities of acrylonitrile and water.) After readjustment of the pH at 30", the solution was placed in both reference and sample cells and equilibrated.
The reaction was started by the addition of nucleoside in 1% or less of the total volume. pH measurements were taken before and during the course of the kinetic run and did not vary by more than 0.04 unit.
The average value was used. Recordings were made either as complete spectra on a Cary model 14 spectrophotometer as a function of time (see Figs. 1 to 4) or automatically at a single wavelength in a Gilford model 2000/Beckman DU monochromator system. Both methods gave equivalent results (see Figs. 5 to 8). A pseudo first order plot of the absorbance change, Dt -Dm was used according to Equation 15 or 16 (see the "Appendix") to determine k,,,. Alternatively, the method of Guggenheim (26) was used when the reaction rate was too slow to allow complet,ion of the reaction to be conveniently reached, or when a second slower reaction followed the   Table I In the one case where the two kinetic methods were compared 290 300 310 320 (see Table III spectra of pseudouridine C and 13, inosine, and 4-thiouridine as a function of time. There are several features common to all four of these reactions. First, each set of spectral curves shows clear isosbestic points, a good indication of a single reactant-product system without spectral complications. The shift in the pseudouridine spectral curves at later times is due to the secondary slow conversion of N&E-pseudouridine to N1, Na-CE2pseudouridine (13) (note especially the dashed curves). Second, the rate of reaction with time is accurately pseudo first order in each case. Third, in the case of inosine at least, the final spectrum obtained is the same as that of the chromatographically isolated reaction product. The reaction does not come to a stop because an equilibrium is reached (except in the case of 4-thiouridine at high pH) since all four of the adducts can be isolated and show no signs of decomposition when reincubated under the same conditions in the absence of acrylonitrile. (CEl-thiouridine shows some decomposition, see below.) Also, the reaction of both pseudouridine B and C is not due to isomeriaation of one to the other (7), since incubation of each isomer separately at pH 10 for 24 hours in the absence of acrylonitrile resulted in no evidence for a rate of isomerization of either isomer which was more than 1% of the cyanoethylation rate seen here. In this experiment, conversion of isomers was measured spectrally at pH 12 by taking advantage of the distinct alkaline spectra of the two forms (7,19).
In order to examine more carefully the hypothesis proposed previously (13) that ionization must precede reaction of a nucleoside with acrylonitrile, the rate of reaction was measured as a function of pH. The predicted rate expression for the general case is derived in the "Appendix" and is given in Equation 12, and in Equation 15 in a more useful form. From this derivation it is clear that the apparent form of the reaction will be pseudo first order (as is observed) and k,, will vary with pH in a way which depends on the values chosen for kl, kz, kB, and kd. kS was known for 4-thiouridine and was negligible for the other nucleosides, while kk was negligible for all of the nucleosides (see below). However, no a priori relative estimate of kl and lcz was available.
In order to evaluate the experimental data provided in Fig. 5 to 8, two cases were considered. In the first case, kl = 0. That is, reaction only occurs via the ionized nucleoside. In the other case, 20 to 30% of the reaction was assumed to proceed via the unionized nucleoside, i.e. kl = 0.2 to 0.3 kz. The theoretical curves for each case were then obtained in the following way and compared with the experimental values. For each k,,,, a lcz could be calculated by means of Equations 13 and 14, since pK and kS were known, and kl was defined in terms of i%:. The average value of kz was computed in this way, using all of the data, and then used to construct the theoretical curves shown in Figs. 5 to 8 by going back through Equations 13 and 14 again to obtain kapp (calculated).
A comparison of the curves with the data indicates that, in every case, the best fit is obtained for kl = 0, although there is sufficient experimental scatter to prevent the detection of about 10% reaction via the unionized nucleoside. It should also be noted that different buffers, changes in ionic strength, or the presence of 1Mg ions have no effect on the kinetics observed here. This is true for all of the nucleosides, although it has been most lknction. of Pseudouridylic Acid in Transfer RNA. extensively studied with inosine ( Fig. 8 and Table II).
In addition, Table II shows that inosinic 5'-phosphate with or without Mg reacts at the same rate as inosine.
Other Nucleosides-An examination of a number of other minor nucleosides known or likely to be found in tRNA was carried out to see if still other highly reactive nucleosides existed. For these experiments, the thin layer chromatography method was used, as exemplified in Fig. 9, which shows the rate of reaction of ribothymidine at a single pH. In this case the react,ion was very slow with a rate constant about like that for uridine (Table  III).
In similar experiments, the reaction of acrylonitrile with xanthosine, dihydrouridine, and 2-thiouridine was studied by the thin layer chromatography method, and reaction with orotidine by the spectral method. The results of these experiments plus data on the reaction of adenosine, guanosine, and cytidine are summarized in Table III. Characterization of Cyanoethyl Nucleosides Spectral Properties-In view of the mechanism of the cyanoethylation reaction presented above, it was expected that a stable cyanoethyl group would become substituted for the ionizable proton of the heterocyclic ring, resulting in a product the spectrum of which would no longer be affected by pH changes. This prediction was borne out with every nucleoside studied except adenosine, guanosine, and cytidine. Chromatographically purified CE-inosine has a spectrum very similar to that of the acid form of inosine (see Fig. 4, D, curve) and is unaffected by pH changes from 2 to Il.
The same thing is true of CEz-inosine formed slowly as a secondary product in the reaction.
(Evidence presented below indicates that the second cyanoethyl group adds bo the 2'(3')-hydroxyl of ribose.) Similarly, t,he spectra of CE-uridine (13) and CE-ribothymidine (Table IV) resemble that of the unionized nucleosides and do not vary over the pH range 2 to 12.
The spectrum of CE-4-thiouridine (Fig. '3, D, curve) is also invariant over the pH range 2 to 10.6, indicating that here also replacement of the ionizable proton has occurred.
(The pK for 4-thiouridine is 8.0.) It was not possible to make spectral measurements at higher pH because of the rapid rate of alkaline hydrolysis.
In all other cases st,udied, the spectrum at pH 12 and 30" was stable for at least 20 min. Unlike the previous examples, however, the spectrum of CE-4-thiouridine does not resemble that of the unionized form of 4-thiouridine.
This is because cyanoethylation occurs not on Na but on the sulfur atom instead (see below).
The spectral variation with pH of the pseudouridine adducts was also as expected, bearing in mind that cyanoethylation at the N1 atom would produce a CE-nucleoside with spectral properties like that of uridine which is also an Ni-substituted uracil. That is, there should be a decrease in extinction coefficient with pH due to ionization at NO, but virtually no shift in the wavelength of maximum absorption, in marked contrast to the 24 mp bat'hochromic shift seen at pH 12 with pseudouridine itself (Fig. 10). Moreover, the pK calculated for the remaining ioniaation, 9.2 to 9.3, was very close to that for uridine, 9.25.
The data from this figure can also be used to calculate the extinction coefficient (E) for both CE-pseudouridine C and CEpseudouridine B as follows.  The control lacked acrylonitrile.
Analysis was performed by the thin layer chromatographic method in Solvent C. k,,, = 0.28 X lo+ min+. kz was calculated from Equat.ion 17 of the "Appendix" to be 0.47 X IO-3 min-I, assuming /cl = 0. uridine can be obtained by combining the spectral data of Ofengand and Schaefer (19) with the E values given by Shapiro and Chambers (31) for pseudouridine C at pH 7 and by Chambers (32) for pseudouridine C and pseudouridine B at pH 12. 2. From the reaction with acrylonitrile shown in Figs. 1 and 2, an isosbestic point can be located at the given pH of the reaction where eCE$ is equal to efi which can be determined as in (1).
3. The data of the above figure (Fig. 10) can then be used to convert this ehrH into a more useful value, eaax. When this was done, a value of 9800 was obtained for pseudouridine C, with a corresponding e7 max of 8900 for pseudouridine B. For comparison, the value for uridine is 9900 (28). These values are in good agreement with that reported for the Ni-cyanoethyl mixed isomers of pseudouridine (9100) by Yoshida and Ukita (17).
As indicated above and previously (13), monocyanoethyl pseudouridine can react further to give a presumed dicyanoethyl product which is chromatographically and electrophoretically distinct (see below).
Presumably, reaction occurs at the availa- ble Na atom. It would be expected, therefore, that the spectrum of this compound, like CE-uridine, would be invariant with pH from 2 to 12. This result was obtained previously for the mixed isomers of pseudouridine and has been confirmed for the separated C and B isomers. Thus, the X,,, occurs at 268 and 269 rnp, respectively, for the C and B isomers, with 280:260 ratios of 1.20 and 1.10, and the spectra do not change with pH.
The findings obtained above can be contrasted with the following examples which indicate that cyanoethylation can occur without modifying spectral properties.
This was the case for the reaction with adenosine, cytidine, and guanosine. affected by cyanoethylation and suggests that addition had occurred elsewhere in the molecule.
Chromatographic Behavior-The chromatographic properties summarized in Table I also provide some insight into the structure of the derivatives synthesized.
First, it is clear that Solvent C and to a lesser extent D is able to distinguish mono-from dicyanoethyl derivatives by virtue of the increased affinity of the nucleoside for the mobile organic phase due to the addition of cyanoethyl groups. Thus, both CEz-pseudouridine and CEe inosine move further than their corresponding monocyanoethyl derivatives even though pseudouridine has both groups on the ring and inosine has one on the ribose moiety.
Second, the lack of ionizability predicted from the failure to show spectral shifts with pH can be seen directly in every applicable instance by comparison of the RF values for the CE-nucleoside in Solvent C (isobutanol-water) versus Solvent A (butanol-NHZ-water) with the same Rp values for the parent nucleoside.
While the Rr goes down from 0.32 in Solvent C to 0.14 in the NHB-containing Solvent A owing to ionization, the Rp of the CE-nucleoside actually increases in going from Solvent C (0.41) to Solvent A (0.57).
Electrophoretic Mobility-Paper electrophoresis in the presence and absence of borate was performed in order to detect cyanoethylation of the 2'(3')-hydroxyl group since only nucleosides possessing free vicinal hydroxyl groups can form a complex with borate and so acquire increased negative charge (Table VI, Experiment 1). Clearly this has occurred with cytidine, adenosine, and guanosine (Experiment 2) and with CEz-inosine (Experiment 3, line 3). Equally clearly, no 2'(3')-O-cyanoethyl derivatives were produced from ribothymidine, 4-thiouridine, or uri-dine, nor in the formation of CE-inosine. Although the increased mobility due to borate is less marked in the pseudouridine series because of the greater extent of ionization at this pH, Experiments 5 and 6 show clearly that in neither the mono-nor dicyano- ethyl derivatives has there been any cyanoethylation on the vicinal hydroxyls of ribose. Electrophoresis in borate has also been used by Yoshida and Ukita to detect cyanoethylation on the ribose moiety of nucleosides in their study, although no details were presented (17).
Additional evidence for the lack of ionizability was obtained by comparing the mobilities of the CE-nucleosides with their parent compounds in the absence of borate since these runs were performed at a pH at which the parent nucleosides were partially ionized.
The results were particularly clear with CE-inosine, CE-4-thiouridine, CE-uridine, and CEs-pseudouridine and show that cyanoethylation markedly reduced the electrophoretic mobility.
Since ribothymidine itself is only slightly ionized at this pH, the effect is less striking.
In a converse way, since CEpseudouridine itself is approximately half-ionized at this pH, only a small effect is to be expected from the slightly lower pK of pseudouridine compared to CE-pseudouridine.

Structure of Cyanoethyl
Q-Thiouridine-Although the spectrum of this derivative was stable to pH, the marked difference between it and the unionized spectrum of 4-thiouridine suggested that in this case the cyanoethyl group had been added to an atom other than the Ns originally carrying the ionizable proton. All of the possible structures are shown in Fig. 11, bearing in mind that the ionizable proton must be replaced by a stable cyanoethyl group. Originally, Structure I was expected by analogy with the other nucleoside derivatives, but the observed spectrum did not support this view. As a control the spectrum of N1,3-dimethyl-4-thiouracil was examined and proved to be very similar to the acid form of 4-thiouridine with a X,,, at 328 rnp instead of 335 rnp but stable to pH changes from 2 to 12.
Since the derivative spectrum was similar to that of 4-thiouridine disulfide, it was important to rule out Structure IV occurring by some unusual oxidative reaction catalyzed by acrylonitrile. This was done in two ways. First, the kinetics of formation observed in Fig. 3 were first order in 4-thiouridine while an oxidative reaction should be second order. Second, exposure of the isolated derivative to 0.1 M 2-mercaptoethanol for 45 min at pH 7.7 had no effect on the spectrum.
if-Thiouridine disulfide is reduced almost immediately by this procedure (23).
It was possible to rule out both remaining structures, I and III, on the basis of the iodine-azide reaction of Feigl, since this reaction is catalyzed both by thiols and thioketones, but not by 5042 Function of Pseudouridylic Acid in Transfer RNA. I Vol. 242,No. 21 thioethers. The experiment is given in Fig. 12. 4-Thiouridine was used as an -SH control since it can enolize to this form, and the fully blocked dimethyl-4-thiouracil was used as a thioketone control to show that thioketones incapable of enolization can also react. /3-(2-Thienyl)-alanine was included as an additional control. It is quite clear from an examination of the left of Fig. 12 (Before Alkali) that CE-4-thiouridine contained neither an available thiol nor thioketone functional group. However, in order to be sure that it contained sulfur at all, the experiment shown on the right of Fig. 12 (Ajter Alkali) was performed. It was known that the 4-thiouridine derivative was unusually alkali-labile (see below) and on hydrolysis regenerated a spectrum identical with that of 4-thiouridine itself. While this alone is good evidence that 4-thiouridine had not been destroyed by cyanoethylation, the experiment shown in Fig. 12 offers confirmatory evidence for the: regeneration by alkali of t)he sulfur-containing reactive group. It is concluded from these experiments that Structure II is the correct one for GE-4-thiouridine.

Chemical Stability
ClL$-thiouridine-The alkaline lability of CE-4-thiouridine can be readily shown spectrophotometrically since it results in the generation of exactly the same set of curves shown in Fig. 3, except in the reverse time sequence. Using the spectral change at 325 ml* as a measure of the loss of CE-4-thiouridinc, the rate of hydrolysis was determined at 30" and several hydroxide concentrations. A general hydrolysis equation for CE-4-thiouridine can be derived from the sum of Equations 2 and 3 of the "Appen- Thus k4 = 0. From the slope of the line, a value of 1.81 f 0.05 min-1 was obtained for Jc3. This corresponds to a half-life of 38 min at pH 12. Although not extensively studied, CE-4-thiouridine appears relatively acidstable. There was no spectral change after 30 min at 23" in 0.01 N HCl, and only about 10 to 20% breakdown to another compound detected chromatographically in Solvent B after exposure to 0.01 N HCl for 2 hours at 23".
C&pseudouridine-The Ni-cyanoethyl derivatives of both pseudouridine isomers are quite stable to both alkali and acid. Treatment with 0.3 N NaOH at 37" for 16 hours or 0.05 N HCl at 23" for 30 mm had no effect on the pH 12 spectrum which is a sensitive measure of hydrolysis to free pseudouridine, nor did it alter the Rp of the treated sample as compared to a control in Solvent,s B or D (alkali-treated sample), or B, C, or isopropanol-NHI-H20 (7 : 1: 2) (acid-treated sample).
The N, , N3-dicyanoethyl derivatives are more alkali-labile. CEn-pscudouridine C decomposed in 0.1 N NaOH at 30" in a pseudo first order reaction showing an alkaline spectral change from 268 1nl.c to 286 mp. This shift was used to follow the reaction and, in addition, strongly suggest that the product had lost the cyanocthyl group from position N1. Since the maximum absorption increased 20% after 95% hydrolysis had occurred, it is likely that the Ns-cyanoethyl group was still intact by analogy with the spectral properties described for the methylated uracils (28). However, no further identification of the product of hydrolysis has been carried out. The rate constant for this hydrolysis at 30" assuming proportionality with hydroxide concentration was 0.036 min-i.
CEs-pseudouridine B decomposed in alkali differently.
Instead of losing the Nr-cyanoethyl group as evidenced by a spectral shift, the ultraviolet spectrum as a whole disappeared with time with a less than 1 rnp shift of the absorption peak. The nature of the product formed has not been investigated. The rate constant for this reaction was faster than for CE2pscudouridine C, being 0.20 mini at 30", again assuming proportionality with hydroxide ion concentration.
The acid stability of the CEz-pseudouridine compounds has not been carefully studied.
They were spectrally stable at pH 2 and 23" for 10 to 20 min but it is not known for how much longer this treatment could be safely extended.
C&uridine-The alkaline stability was studied spectrally by looking for the expected decrease in absorption at pH 12 on hydrolysis to uridine.
This reaction did not occur, but instead the compound underwent a slight rise in absorption with time on exposure to 0.01 N NaOH at 30" which then markedly decreased on reneutralization of the sample. The decrease was about 6 times the initial increase. Although the nature of these reactions has not been studied, they were very slow and it was calculated that less than 5% loss of CE-uridine occurs after lf hours in 0.01 N NaOH at 30". CE-uridine was also stable to 0.1 N HCl at 23" for 2 hours, as shown by chromatography in Solvent B. CIC-ribothymidine-No spectral changes were observed with this derivative either on exposure to alkali or after reneutraliza-tion, and it was calculated that less than 5% hydrolysis occurs after 18 hours in 0.03 N NaOH at 23". The acid stability of this derivative was not studied.
CZGnosine-The alkaline stability of the cyanoethyl derivatives of inosine were studied making use of the expected spectral changes shown in Fig. 4, but in reverse. By this method it was found that Nr-CE-inosine showed less than 5% hydrolysis after 18 hours at pH 11 and 30", while less than 5% hydrolysis was observed for CEz-inosine after 4 hours at 30" in 0.001 M NaOH. The Nr-cyanoethyl compound was also stable to 0.1 N HCl at 23' for 2 hours as shown by chromatography in Solvent B. The acid stability of CEa-inosine was not examined. DISCUSSION The general nature of the reaction of nucleosides and nucleotides with acrylonitrile appears to be that of a typical anionic Michael addition reaction across a polarized double bond. Thus the data presented in this paper show that only ionized nucleosides react with acrylonitrile.
This fact can explain in part the variation in apparent reactivity of nucleosides which have different pK values like ribothymidine and orotidine; and the unreactivity of dihydrouridine, which has a pK greater than 11.0, (33) might also be explained in this way. However, it clearly cannot account for the marked differences in reactivity between uridine and pseudouridine 13, which have almost identical pK values. The reasons for the great.ly increased reactivity of pseudouridine, inosine, and 4-thiouridine are not known. Empirically, Spector and Keller (34) observed that acetylation of uracil gave exclusively l-acetyl uracil, and Shvachkin et al. (35) found that cyanoethylation of 2-thiouracil gave exclusively the Ni-cyanoethyl compound, suggesting an inherently greater reactivity of the Nr atom. However, others have found an equal reactivity for both nitrogen atoms (cf. 16) so that it seems more likely that some other explanation need be sought for the high reactivity of pseudouridine.
In any case, the reactivity of inosine and 4-thiouridine require independent explanations. The reaction of inosine and pseudouridine with acrylonitrile has also been reported by Yoshida and Ukita (17) in a brief communication.
They also observed that kapp increased with pH in an approximately sigmoid manner, but they did not attempt to compare their results with a theoretical prediction. The approximately three-fold higher karp values reported by them are probably due to the use of a higher temperature and a higher but unspecified acrylonitrile concentration.
The kinetic specificity shown by acrylonitrile for pseudouridine, inosine, and 4-thiouridine has proven useful in studies on the function of these minor nucleotides in tRNA.% In addition, the fact that ksgp depends on both solution pH and nucleotide pK has been of value in studies using cyanoethylation as a chemical probe of the involvement of these minor nucleotides in secondary structure.
For example, one would predict that the well known shift of pK to higher values for nucleotide residues involved in the secondary structure of RNA would markedly decrease the observed reaction rate from that expected on the basis of nucleosides or denatured RNA, and conversely, that residues in a denatured loop should react at expected rates. Such an effect has been observed (36) and will be the subject of a separate communication.
The structures of the derivatives produced have been shown by a combination of spectral, chromatographic, and electrophoretic measurements.
Chromatography and electrophoresis showed (a) the number of cyanoethyl groups added, (b) the physical loss of ionizability following cyanoethylation which had been inferred from the spectral measurements, and (c) those derivatives which had been cyanoethylated on ribose. These analyses also ruled out the possible hydrolysis of the nitrile group to a carboxyl function during the course of the reaction or isolation since such a conversion would be expected to add a negative charge to the molecule at these pH values and none was observed.
Analysis of the spectra was then sufficient to localize the cyanoethyl group to the appropriate atom of the heterocyclic ring. Assignment of the cyanocthyl group to the N1 or NB atom of pseudouridine, uridine, and ribothymidine rather than to the 02 or 04 was based on a comparison of the ultraviolet spectra at different pH values with those of the model compounds studied by Shugar and Fox (28), using uracil as a model for pseudouridine.
More rigorous evidence for the absence of enolic cyanoethyl derivatives should be obtainable by infrared spectral analysis of the type used so successfully by Miles (37, 38) since the ketonic structures differ markedly from the enol type, but this has not been done so far. However, this method was used to show the absence of an enolic structure in CE-inosine since analysis of the ultraviolet spectra was not adequate in this case. Measurement of the infrared spectrum in D20 showed a very strong band at 1687 cm-l, very similar in frequency, shape, and intensity to that previously found for Nr-methyl inosine, but no strong bands in the region of 1586 to 1611 cm-l corresponding to the Os-methyl compound were detected. 3 The chemical stability of the pyrimidine derivatives are also consistent with the known behavior of N-substituted uracil compounds. Thus, N1-CE-pseudouridine was quite stable to alkali, as expected by analogy with uridine and Nr-methyl uracil, and the fully substituted compounds CEt-pseudouridine C, CE2pseudouridine B, and CE-uridine were less stable to alkaline decomposition, as is also the case for N1,,-dimethyl uracil (33). Moreover, the alkaline stability of CE-ribothymidine correlates well with the stability observed for Nl,s-dimethyl thymine (33). The different nature of the alkaline decomposition products of CEz-pseudouridine C and CEz-pseudouridine B are no doubt related to the different orientation of the uracil rings with respect to the ribose in the two cases, but no attempt has been made as yet to rationalize the different hydrolysis products on this basis.
It is highly unlikely that under the conditions used here anionic polymerization of acrylonitrile on a cyanoethyl nucleoside could occur since the pK of the a-hydrogen of acrylonitrile is greater even than that of water (39) and the maximum pH used in these studies was 10. Moreover, extra aliphatic groups should be detectable by chromatography in Solvents C or D. However, even in the case of reaction with pseudouridine to which at least one cyanoethyl group must be added to each N atom in view of the spectral shifts produced, chromatography in Solvent D of a reaction mixture at different times of reaction provided no evidence' for compounds other than the expected ones. Had polymerization on the cyanoethyl group occurred, additional nucleoside products would have been expected, but none were detected. Thus it appears that the added group in all the cases described here is a monocyanoethyl group.
The number of cyanoethyl groups added in the various CEnucleosides described here is emphasized because of other studies in which the specific radioactivity associated with the cyanoethyl group was used to estimate the number of particular CEnucleotides in cyanoethylated transfer RNA.2 In conclusion, the cyanoethylation reaction described in this paper appears to be a useful one for a variety of functional and chemical studies on "minor" nucleotides in transfer RNA.
Acknowledgments-1 wish to acknowledge the several assistants who have at various times participated in and made major contributions to the progress of this work. They are R. Blumberg, G. Dunn, L. Frausto, W. Lyon, E. Medeiros, H. Schaefer, and J. Senderowitz.
They have my sincerest thanks and appreciation for the quality of their work. where NH and N-are the unionized and ionized nucleosides, respectively, A is acrylonitrile, and NAN is the cyanoethyl nucleoside.