796

NUCLEIC ACID CONSTITUENTS

[44]

line is opened slowly during evaporation of reagents and solvent to prevent "bumping" the sample from the crucible. Operation in the chemical ionization (CI) mode is performed by adjusting the reagent gas pressure until the following ion ratios are obtained: (a) m/z 17/15 approximately 3 : 1 for methane, (b) m/z 57/43 approximately 30 : 1 for isobutane, and (c) m/z 18/35 approximately 25 : 1 for ammonia. 27 Other chromatographic and mass spectrometer operating conditions are as described above. Analysis of nucleoside and nucleotide TMS derivatives in the FAB mode is performed as followsg'~°: 1 /zl of the derivatization mixture is added to 1/xl of an appropriate matrix, e.g., diglyme (tetraethyleneglycol dimethyl ether), previously applied to the tip of the FAB probe. The probe is introduced into the mass spectrometer and the mass range of interest scanned at 11 sec/decade. Samples are ionized using an Ion Tech FAB 11N saddle field atom gun (Ion Tech Ltd., Teddington, England) operating at 8 kV and 1 mA with argon as the bombarding gas. The major advantages9'~° of analysis of the TMS-derivatized nucleosides in the FAB mode include (1) the presence of an intense MH ÷ ion which allows easy assignment of molecular weight, (2) the presence of a large number of structurally relevant ions related to the intact molecule, the sugar moiety, and the aglycon with portions of the sugar residue attached, (3) considerably better sensitivity relative to the FAB analysis of the free sample, and (4) the acquisition of the mass spectrum at ambient temperature, which decreases the possibility of thermal degradation. 27 M. S. Wilson and J. A. McCloskey, J. Am. Chem. Soc. 97, 3436 (1975).

[44] Analysis of R N A H y d r o l y z a t e s b y Liquid Chromatography-Mass Spectrometry

By STEVEN C. POMERANTZand JAMES A. McCLOSKEV Reversed-phase high-performance liquid chromatography (HPLC) is an experimentally effective technique for the separation of nucleosides, ~ particularly for the analysis of enzymatic digests of RNA 2,3 and DNA. 4,5 I C. W. Gehrke and K. C. Kuo, J. Chromatogr. 188, 129 (1980). 2 C. W. Gehrke and K. C. Kuo, J. Chromatogr. 471, 3 (1989). 3 M. Buck, M. Connick, and B. N. Ames, Anal. Biochem. 129, 1 (1983). 4 K. C. Kuo, R. A. McCune, and C. W. Gehrke, Nucleic Acids Res. 8, 4763 (1980). 5 C. W. Gehrke, R. A. McCune, M. A. Gama-Sosa, M. Ehrlich, and K. C. Kuo, J. Chromatogr. 301, 199 (1984).

METHODSIN ENZYMOLOGY,VOL. 193

Copyright© 1990by AcademicPress,Inc. All rightsof reproductionin any formreserved.

[44]

ANALYSISOF RNA HYDROLYZATESBY LC/MS

797

Depending on the problem at hand, identification of nucleoside constituents can often be made on the basis of retention times and UV absorbance characteristics. As is common, in general, when chromatographic methods are used for purposes of identification, as opposed to simply for separation, the reliability of the method suffers as the complexity of the mixture increases, or when components of unknown or unexpected identity are encountered. The development of directly combined HPLC-mass spectrometry [liquid chromatography (LC)/MS] based on the thermospray interface 6,7 provides a method which can be effectively applied to the analysis of nucleosides in nucleic acid digests, s and is a powerful extension of the capabilities of either technique alone. Identification of structurally known nucleosides present in mixtures using LC/MS is based on: (a) relative HPLC retention times, which are generally reproducible to -0.15 min from run-to-run and -+0.50 min from day-today, or better if precautions such as rigorous temperature control are taken I; (b) characteristic UV absorbance, 2,9,1°for example, recorded using a dual wavelength detector or photodiode array detector; (c) complete mass spectra, recorded every 2-3 sec over the duration of a chromatographic run, or in the form of selected-ion recordings, 11 made several times per second. In the case of structurally unknown components, the thermospray mass spectrum will establish whether a given constituent is a nucleoside or nucleotide, as opposed to a UV-absorbing impurity. For unknown nucleosides, the mass spectrum will, in general, establish the masses of the molecule and base moiety, and whether ribose is methylated or otherwise substituted. In some cases, LC can be carded out using deuterated mobile phase, thus exchanging all heteroatom-bound hydrogen atoms by deuterium.12 The resulting mass spectrum can be used to differentiate isomers having different numbers of active hydrogens without isolation of the component of interest, and provides constraints in the characterization of structural unknowns. In the case of the unknown nucleosides, LC/MS data may, in favorable cases, provide sufficient leads to permit structural candidates to be proposed, and then tested by synthesis or other means. 6 C. R. Blakley, J. J. Carmody, and M. L. Vestal, Anal. Chem. 52, 1636 (1980). 7 M. L. Vestal, this volume [5]. 8 C. G. Edmonds, M. L. Vestal, and J. A. McCloskey, Nucleic Acids Res. 13, 8197 (1985). R. H. Hall, "The Modified Nucleosides in Nucleic Acids," Chap. 2. Columbia Univ. Press, New York, 1971. 10 H. Ishikura, K. Watanabe, and T. Ohishima, in "Handbook of Biochemistry" (Japanese Biochemical Society, ed.), Vol. I, p. 1032. Tokyo Kagaku Dozin, Tokyo, 1979. It j. T. Watson, this volume [4]. 12 C. G. Edmonds, S. C. Pomerantz, F. F. Hsu, and J. A. McCloskey, Anal. Chem. 60, 2314 (1988).

798

NUCLEIC ACID CONSTITUENTS

[44]

In any event, LC/MS provides a rapid means of screening RNA or DNA digests so that specific components of interest can be rigorously identified, 13-~7 or structural unknowns can be targeted for isolation and further characterization.18,19 In favorable situations in which modifications are relatively simple, preliminary structure assignments to unknown nucleosides can be made, without isolation of components, directly from digests of unfractionated tRNA. 2°,2~ The thermospray LC/MS technique can also be utilized in conjunction with RNA sequencing studies by analysis of oligonucleotides produced by selective cleavage by ribonuclease T~,17 and is effective for rapid screening of synthetic nucleoside reaction mixtures,22 and hydrolyzates of DNA in vitro reaction products. 23 The use of a mass-specific HPLC detector (the mass spectrometer) introduces a high degree of structural selectivity, such that minor components of interest can generally be detected without chromatographic resolution from other constituents. On the other hand, thermospray mass spectra are less reproducible in terms of ion yields and relative abundances than ions produced by other means (in particular, electron ionization) and so are less well suited for quantitative analysis. In common with other ionization methods which produce parent molecular species by protonation, z4 and in distinct contrast to electron ionization (EI), 25 thermospray generates relatively few ions and thus less structural detail, although the 13 T. G. Hagervall, C. G. Edmonds, J. A. McCloskey, and G. R. Bjork, J. Biol. Chem. 262, 8488 (1987). 14 G. M. Kirtland, T. D. Morris, P. H. Moore, J. J. O'Brian, C. G. Edmonds, J. A. McCloskey, and J. R. Katze, J. Bacteriol. 170, 5633 (1988). 15T. G. Hagervall, Y. H. Jonsson, C. G. Edmonds, J. A. McCIoskey, and G. R. Bjork, J. Bacteriol. 172, 252 (1990). t6 R. P. Martin, A.-P. Sibler, C. W. Gehrke, K. Kuo, C. G. Edmonds, J. A. McCIoskey, and G. Dirheimer, Biochemistry 29, 956 (1990). 17 p. F. Crain, T. Hashizume, C. C. Nelson, S. C. Pomerantz, and J. A. McCIoskey, in "'Biological Mass Spectrometry" (A. L. Burlingame and J. A. McCIoskey, eds.), p. 509. Elsevier, Amsterdam, 1990. t8 D. W. Phillipson, C. G. Edmonds, P. F. Crain, D. L. Smith, D. R. Davis, and J. A. McCloskey, J. Biol. Chem. 262, 3462 (1987). 19j. A. McCloskey, P. F. Crain, C. G. Edmonds, R. Gupta, T. Hashizume, D. W. Phillipson, and K. O. Stetter, Nucleic Acids Res. 15, 683 (1987). a~ C. G. Edmonds, P. F. Crain, T. Hashizume, R. Gupta, K. O. Stetter, and J. A. McCIoskey. J. Chem. Soc. Chem. Commun. p. 909 (1987). 21 j. A. McCloskey, C. G. Edmonds, R. Gupta, T. Hashizume, C. H. Hocart, K. O. Stetter, Nucleic Acids Res. Symp. Ser. 20, 45 (1988). -'-' T. Hashizume, C. C. Nelson, S. C. Pomerantz, and J. A. McCIoskey, Nucleosides Nucleotides 9, 355 (1990). .,3 S. M. Musser, S.-S. Pan, and P. S. Callery, J. Chromatogr. 474, 197 (1989). 24 A. G. Harrison and R. J. Cotter, this volume [1]. 25 j. A. McCloskey, this volume [45].

[44]

ANALYSISOF RNA HYDROLYZATESBY LC/MS

799

principal side chains common to posttranscriptionally modified RNA can usually be recognized26 (see below). Materials Reagents HPLC-grade water Deuterium oxide, 98 atom % D or higher Ammonium acetate Acetonitrile Acetic acid Trifluoroacetic anhydride Reagents should be the highest purity obtainable (HPLC-grade). Water. Water quality is of paramount importance for both chromatography and mass spectrometry. Optimum thermospray performance for molecular ion production requires low concentrations of alkali cations in the mobile phase. Water that is free of organic impurities and has a resistivity of at least 16 MI~-cm -~ is required. In this laboratory, water is obtained from the house-distilled water feed stock, which is subsequently treated by passage through an ion-exchange bed, an activated-carbon filter, a Nanopure II ion-exchange system (Barnstead, Boston, MA), a second activated-carbon filter, and a final particulate filter (0.2/zm). Deuterium Oxide. Previous experience with commercially available D20 indicated a purity problem, causing thermospray vaporizers to become occluded by deposits, probably siliceous in nature, within 30 to 60 min of operation. This problem can be avoided if D20 is purchased that has been stored and shipped in non-glass containers. Low conductivity, 99.8% minimum isotopic purity D20 from Merck Isotopes (St. Louis, MO) is satisfactory. Ammonium Acetate Buffer. The buffer (A) is prepared by adding 77.08 g of deliquescent ammonium acetate crystals (HPLC-grade, J. T. Baker, Phillipsburg, N J) to 4 liters of water. While stirring vigorously, the pH of the solution is adjusted to 6.0 with glacial acetic acid. The buffer is vacuum degassed and filtered through a 0.2 tzm Nylon-66 filter in a single step with a solvent filter/degasser (EM Science, Cherry Hill, N J). Aqueous buffer should be stored in polyethylene containers to prevent leaching of cations from glass vessels, and should be used within 2 weeks of preparation. 26 C. G. Edmonds, T. C. McKee, and J. A. McCloskey, Proc. 33rd ASMS Conf. Mass Spectrom. Allied Topics, San Diego, CA p. 514 (1985).

800

NUCLEIC ACID CONSTITUENTS

inject J

HPLC

~

UV detector

chromatogram (UV)

[44]

thermosprey vaporizer

mass spectrometer

--J

mass spectrum and chromatogram (UV or mass)

FIG. l. Configuration of instrument components for thermospray LC/MS; see text for discussion.

Organic Modifier. Organic modifier (B) is prepared by adding 2.4 liters of water to 1.6 liters of acetonitrile (HPLC-grade, American Burdick and Jackson, Muskegon, MI), and degassed by stirring vigorously under vacuum for 20 min. Alternatively, a filter/degasser, ultrasonication, or helium sparging may be employed to degas the organic modifier. Deuterated Buffer. Deuterated ammonium acetate (ND4OAc) is prepared by three successive evaporations from D20 solution. The deuterated ammonium acetate crystals are stored under dry nitrogen to prevent backexchange with atmospheric water. The buffer solution is made immediately prior to use, with pD adjustment to 7.25 by deuterated trifluoroacetic acid (prepared by deuterolysis of trifluoroacetic anhydride). The pD of the buffer is measured with a standard pH electrode, and corrected by the relation pD = pH + 0.4 units. 27 Neat acetonitrile is used for organic modifier in both isocratic and gradient elutions to maximize deuterium exchange levels.

Instrumentation The configuration of instrument components is shown schematically in Fig. 1. Liquid Chromatograph. The chromatographic system employed for analyses described here is a Beckman Instruments (Fullerton, CA) Model 342. This system consists of a Model 114M sin#e-piston pump for delivery of aqueous buffer, and a Model 100A dual-piston pump for delivery of 27 p. K. Glasoe and F. A. Long, J. Phys. Chem. 64, 188 (1960).

[44]

ANALYSIS OF R N A HYDROLYZATES BY L C / M S

801

organic modifier. These pumps are connected to a Model 400 dual-chamber high-pressure mixer with a 2.8 ml mixing volume. Because high-pressure mixing and gradient formation is generally more reproducible than lowpressure mixing, two pumps and a high-pressure mixer are used. In general the pumps should be the most pulse-free available, and capable of operation at 5000 psi back-pressure. The injector is a Beckman Model 210 fitted with a 100 /~1 loop. Microparticulates are prevented from entering the injector by the insertion of a 0.5 txm frit assembly (Scientific Systems, State College, PA) between the mixer and injector. Chromatographic separations are performed on a 4.6 × 250 mm LC-18S analytical column (Supelco, Bellefonte, PA) preceded by a 4.6 x 30 mm RP-18 guard column (Applied Biosystems, Foster City, CA). Reproducibility of retention times is increased by maintaining the analytical and guard columns at 30° in a thermostatically controlled column heater (Model TCM, Waters Associates, Milford, MA). UV absorbance detection at 254 and 280 nm is performed using a Model 440 detector (Waters Associates) serially connected between the column and the mass spectrometer, as shown in Fig. 1. In principle, any UV detector can be used, but the detector flow cell must be capable of withstanding a pressure differential of approximately 1500 psi. If possible, the absorbance detector output should be digitized by the mass spectrometry data system to facilitate time alignment of the UV chromatogram and mass spectra, as discussed in a later section. All interconnecting tubing is 0.01-inch id, and lengths are minimized to the greatest extent possible. Mass Spectrometer. All spectra presented here were obtained on a noncommercial quadrupole mass spectrometer and data system previously described. 8 The quadrupoles are maintained at 150° to prevent vapor condensation in the analyzer section. Mass spectra are scanned every 1.7 sec from m/z 100 to m/z 360. The UV absorbance signal is digitized and recorded by the mass spectrometer data system each time a mass spectral scan is made. A UV chromatogram is, therefore, produced by the data system for accurate time alignment with mass channels. In the resulting chromatograms, the mass channels time scale lags the UV time scale by approximately 6 sec, due to the transit time between UV and mass detectors. HPLC Gradient. The gradient profile used for nucleoside separations is a slight modification of the gradient developed by Buck et al. 3 and is listed in Table I. If only a narrow range of elution times is of interest, or for simple mixtures, isocratic conditions or a modified version of the gradient shown can be used to reduce the analysis time. Thermospray Vaporizer Controller and Operating Conditions. The vaporizer controller used is from Vestec Corp. (Houston, TX). Most mass

802

NUCLEIC ACID CONSTITUENTS

[44]

TABLE I GRADIENT PROFILE FOR HPLC OF NUCLEOSIDES Interval time (min)

Composition (%B)

Start

End

Start

End

0.0 3.0 4.4 5.8 7.2 8.6 10.0 25.0 30.0 34.0 37.0 45.0 48.0

3.0 4.4 5.8 7.2 8.6 10.0 25.0 30.0 34.0 37.0 45.0 48.0 53.0

0.0 0.0 0.2 0.8 1.8 3.2 5.0 25.0 50.0 75.0 75.0 100.0 100.0

0.0 0.2 0.8 1.8 3.2 5.0 25.0 50.0 75.0 75.0 100.0 100.0 0.0

spectrometer manufacturers offer thermospray interface options, either as part of a new instrument or for retrofitting to an existing instrument. The controller should provide complete temperature control of the vaporizer probe, and should ideally include provision for automatic temperature compensation for the variable heat capacity of the gradient eluant. The operational parameters of the vaporizer for nucleoside analysis are in general accord with most other classes of compounds. Sensitivity is maximized when the tip temperature (i/'2) is maintained within 2° to 5° of the "burnout" temperature, i.e., the temperature at which 100% of the effluent is vaporized within the vaporizer. The determination of this operating point from the T1/T2 curve has been previously described. 7 Operation at this point for vaporizers of average orifice size (75/.~m diameter) yields tip temperatures of approximately 240°-270 °, and control temperatures (T1) in the range of 125°-140 °. The source block should be maintained at approximately 350 ° to provide sufficient heat to complete the vaporization process. The resultant vapor temperature is approximately 270o-300 °. In thermospray ion sources equipped with vaporizer tip heaters, the tip should be maintained at approximately 300 °. Optimization of the operating conditions can be accomplished by pumping 2 ml/min of a solution of ammonium acetate buffer containing 1 ppm adenosine and 5 ppm guanosine directly into the ion source, bypassing the column. A broadly useful operating temperature is obtained by maximizing the response of the guanosine molecular ion (MH ÷, m/z 284), without significant sacrifice of

[44]

ANALYSISOF RNA HYDROLYZATESBY LC/MS

803

the adenosine response (MH +, m/z 268). The intensity ratio is a moderately strong function of temperature, and is highly vaporizer-dependent. Generally, the best achievable response ratio is approximately 10 : 1 (268 : 284), although in many cases, maximum performance will only yield a response ratio of 20 : 1 or less. Enzymatic Hydrolysis o f R N A . Procedures for nuclease digestion of RNA to ribonucleotides, followed by dephosphorylation using alkaline phosphatase, are followed which, in general, permit direct injection of the crude digest directly into the HPLC. Detailed protocols and discussion are given elsewhere in this volume. 28 I n t e r p r e t a t i o n of R e s u l t s

H P L C Retention Times and UV Absorbance

The elution positions in the reversed-phase HPLC system of Buck et al. 3 of ribonucleosides from RNA are listed in Table II; some common contaminants which may occur in enzymatic digests of RNA are given in Table III. These retention times data were compiled from this laboratory and the earlier listing of Buck et al. 3 Because individual retention time values were, in general, determined from multiple experiments and not necessarily from the same analysis, their values in relation to each other may exhibit minor variance. In some cases, elution order has been estimated from the standard reversed-phase retention data published by Gehrke, z based on a phosphate buffer system. In general, retention times can be referenced to the elution times of known nucleosides within -+0.50 min on a day-to-day basis. The major ribonucleosides (cytidine, uridine, guanosine, adenosine) can be used as internal reference points, although peak centers are often difficult to accurately define if large quantities have been injected. Depending on the source of RNA, common modified nucleosides present in lower amounts, such as pseudouridine, N2-methyl guanosine, or N~,Nr-dimethyladenosine, can also be used as convenient reference points. Common contaminants such as the four major 2'-deoxynucleosides from traces of DNA can usually be recognized from their retention times (Table III), in particular, for deoxycytidine and deoxyadenosine, which are generally well resolved in the presence of ribonucleosides. Mononucleotides resulting from incomplete dephosphorylation (see also Ref. 28) constitute a more difficult problem because the elution times for most of the monophosphates of the nucleosides listed in Table II are not known. 28p. F. Crain, this volume [42].

804

NUCLEIC ACID CONSTITUENTS

[44]

TABLE 11 THERMOSPRAY LC/MS DATA FOR RNA NUCLEOSIDES Mass spectrum, m / z h Nucleoside 5,6-Dihydrouridine Pseudouridine 5-Carboxymethylaminomethyluridine Cytidine 5-Hydroxyuridine 5-Carboxymethoxyuridine 3-{3-Amino-3-carboxypropyl)uridine Uridine I-Methylpseudouridine 2-Thiocytidine 2'-O-Methylpseudouridine 3-Methylcytidine N4-Methylcytidine 5-Methylcytidine 5-Carboxymethylaminomethyl2-Ihiouridine I-Methyladenosine 5-Methylaminomethyl-2-thiouridine 2'-O-Methylcytidine Inosine 5-Methoxyuridine 5-Methyluridine (ribosylthymine) 2-Thiouridine Guanosine 7-Methylguanosine 2'-O-Methyluridine 4-Thiouridine 3-Methyluridine 4.2'-O-Dimethylcytidine 5,2'-O-Dimethylcytidine Epoxyqueuosine I-Methylinosine Queuosine I-Methylguanosine 2'-O-Methylguanosine Lysidine NLAcetylcytidine N-'-Methylguanosine Adenosine 5-Methyl-2-thiouridine 5-Met hoxycarbonylmethoxyuridine 2'-O-Ribosyladenosine 1,2'-O-Dimethylinosine

Symbol

Retention time (min)"

MH ÷

BH2 +

S - H.NH4 +

D • cmnmSU C hoSU cmoSU acp3U U m~ s2C ~m m3C m4C mSC cmnm~s-'U

3.2 3.5 4.7 5.2 5.2 5.6 6.5' 7.4 8.6 8.7 11.9 12.0 12.0 12.4 12.7

247 245 332 244 261 319 346 245 259 260 259 258 258 258 349

115 -202 i 12 129 187 214 113 -128 -126 126 126 216

150 -150 150 150 150 150 150 -150 -150 150 150 150

mtA mnmSs-'U Cm I moSU mSU (T) s2U G mTG Um s4U m3U m4Cm mSCm oQ mq Q m~G Gm acp-'C ac4C m-'G A mSs2U (s-'T) mcmoSU Ar mqm

13.9 14. I 14.3 15.2 15.2 15.4 15.5 15.7 16.0 17.0 17.0 17. I 17.8 18.0 18. I 18.2 18.3 18.7 18.7 19.0 19.4 19.5 19.9 20.2 2{I.2 21.6 21.9

282 304 258 269 275 259 261 284 298 259 261 259 272 272 426 283 410 298 298 372 286 298 268 275 333 400 297

150 172 112 137 143 127 129 152 166 113 129 127 126 126 294 151 278 166 152 240 154 166 136 143 201 136 151

150 150 164 150 150 150 150 150 150 164 150 150 164 164 150 150 150 150 164 150 150 150 150 150 150 282 164

[44]

ANALYSIS OF R N A HYDROLYZATES BY L C / M S

805

TABLE 11 (continued) Mass spectrum, m/z h Nucleoside

Symbol

Retention time (min)"

MH +

BH, +

S - H • NH4 +

N-',N 2-Dimethylguanosine 2'-O-Methyladenosine N6-Threonylcarbamoyladenosine 5-Met hoxycarbonylmethyl-2thiouridine

m_~G Am t6A mcmSs-'U

22.0 22.8 23.0 23.2'

312 282 413 333

180 136 281 201

150 164 150 150

N4-Acetyl-2'-O-methylcytidine

ac4Cm m2A m6A s2Um mt6A

23.9 24.4 25.0 25.3 27.0'

300 282 282 275 427

154 150 150 129 295

164 150 150 164 150

ms2t6A

27.8

459

327

150

ms2A m6Am m~Gm m6A mimG

28.2 28.5' 28.7 30.6 31.5

314 296 326 296 350

182 150 180 164 218

150 164 164 150 150

io6A

32.5"

352

220

150

io6A

32.5'

352

220

150

yW i6A ms2io6A

34.5' 36.0 36.2'

509 336 398

377 204 266

150 150 150

ms2i6A

45.0'

382

250

150

2-Methyladenosine N~-Methyladenosine 2-Thio-2'-O-methyluridine N~-MethyI-N6-threonylcarbamoyladenosine N6-Threonylcarbamoyl-2-methyl thioadenosine 2-Met hylthioadenosine

N~,2'-O-Dimethyladenosine N'-,N'-,2'-O-Trimethylguanosine N".N~-Dimethyladenosine 3-{fl-D-Ribofuranosyl)-4,9-dihydro4,6,7-t rimet hyl-9-oximidazo[ I, 2-alpurine

N~-ltrans-4-Hydroxy-3-methyl -2-butenyl)adenosine (zeatin riboside) N6-lcis-4-Hydroxy-3-methyl 2-butenyl)adenosine (zeatin riboside) Wyobutosine N~-13-Methyl-2-butenyl)adenosine N~-lcis-4-H ydroxy-3-methyl 2-butenyl)-2-methylthioadenosine N~-{3-Methyl-2-butenyl)2-methylthioadenosine

" Nucleosides listed with the same retention times are given in the order of elution. h Values given are calculated, and do not imply that the ion will be observed under typical operating conditions. ' Time taken from M. Buck, M. Connick, and B. N. Ames, Anal. Biochem. 129, I (1983).

806

NUCLEIC ACID CONSTITUENTS

[44]

TABLE III THERMOSPRAYLC/MS DATA

FOR COMMON CONTAMINANTS IN

RNA

HYDROLYZATES

Mass spectrum, m/z" Compound

Symbol

Retention time (min)

Cytidine 5'-monophosphate Uridine 5'-monophosphate Guanosine 5'-monophosphate Adenosine 5'-monophosphate 2'-Deoxycytidine 5-Methyl-2'-deoxycytidine 2'-Deoxyguanosine 2'-Deoxyinosine Thymidine 2'-Deoxyadenosine Nr-Methyl-2'-deoxyadenosine

pC pU pG pA dC mSdC dG dI dT dA m6dA

2.0 2.3 3.4 7.4 8.3 16.2 16.4 16.8 17.8 20.8 26.2

MH +

BH2 ÷

S - H" NH4 ÷

324 325 364 348 228 242 268 253 243 252 266

112 113 152 136 1t2 126 152 137 127 136 150

150b 150b

150b 150b

134 134 134 134 134 134 134

`"Values given are calculated, and do not imply that the ion will be observed under typical operating conditions. b In the mass spectra of mononucleotides, the value of S is defined as that of the corresponding nucleoside. See text for discussion. In the case of v e r y c o m p l e x nucleoside mixtures, such as digests of unfractionated t R N A , identification of minor c o m p o n e n t s based solely on retention time is unreliable (see discussion elsewhere in this volume'-9), due not only to the large n u m b e r o f k n o w n natural nucleosides, but also to the possible p r e s e n c e of small oligonucleotides or to UV-absorbing impurities. U V a b s o r b a n c e characteristics, 2'9'~° for example, in the form of 254/ 280 nm ratios, can be useful in limiting cases for confirmation of structure or as an aid in recognizing mononucleotides through the characteristic a b s o r b a n c e of the corresponding nucleoside. H o w e v e r , chromatographic overlap f r o m minor c o m p o n e n t s or impurities, which m a y be unrecognized, seriously limits the reliability of U V absorbance, particularly in the m o s t c r o w d e d region o f the c h r o m a t o g r a m b e t w e e n about 17 and 23 min. With the exception o f 5,6-dihydrouridine, all natural nucleosides f r o m R N A and D N A absorb strongly in the UV, and are readily detected by U V a b s o r b a n c e at levels d o w n to about 1 ng. Dihydrouridine, the earliest eluting nucleoside (Table II), is easily detected by mass s p e c t r o m e t r y , as discussed below. In principle, a U V detector is not required for detection, and mass s p e c t r o m e t e r total ion current can be used to monitor elution of nucleosides. In practice the ion current signal f r o m t h e r m o s p r a y is sufficiently 29j. A. McCloskey, this volume [41].

[44] O

ANALYSISOF RNA IaYI~ROLYZATESBY LC/MS B 1'

~

HO

R M

thermospray ionization > NH4OAc

B.H +

HO~i~

+ HO

R

MH+

BH2+ +

807

HO+IH

+NH4

HO R

R ra/z

OH 150 OCH3 164 H 134

SCHEME1. Principal products from thermospray ionization of nucleosides.

noisy that fine detail of chromatographic resolution is lost and occasional interferences are encountered from nonnucleoside (non-UV-absorbing) impurities which contribute to the total ion current signal. The use of a UV detector placed in series between the chromatograph and mass spectrometer as shown in Fig. 1 is, therefore, strongly recommended.

Thermospray Mass Spectra of Nucleosides The principal products of thermospray ionization of nucleosides in the presence of NH4 + , are schematically shown in Scheme 1,8 and are similar to those resulting from N H 3 chemical ionization. 3° These are the protonated molecule (MH+), the protonated free base (BH2+), and a sugar ion [(S - H)" NH4 +] corresponding to loss of one hydroxyl hydrogen from the sugar fragment and formation of an adduct with NH4 + from the mobilephase buffer. Thermospray mass spectra of the four major ribonucleosides from RNA are shown in Fig. 2. Molecular ion (MH ÷) abundances vary widely, and are influenced by vapor temperature and other experimental and physical conditions which affect droplet size and rate of evaporation of the spray, such as flow rate and wear at the tip of the vaporizer nozzle. Among the common nucleosides, guanosine and 7-methylguanosine produce extensive dissociation of MH ÷ to form BH2 +, a tendency exhibited by some other derivatives of guanine. MH ÷ abundance is not strictly related to sample size, and for large quantities (> 1/zg), MH ÷ peaks may be observed at higher relative intensity values than is the case in the range below - 5 0 ng. The principal sugar ions of all nucleosides are usually of low abundance and assume one of the three values shown in Scheme I, characteristic of the two principal sugars known in natural RNA (ribose or 2'-O-methylribose), or DNA (2'-deoxyribose). 3oM. S. Wilson and J. A. McCloskey, J. Am. Chem. Soc. 97, 3436 (1975).

NUCLEIC ACID CONSTITUENTS

808 a Zi,i z

[44]

NH2

100.

I-'"

HOH. H ~ 50

Mr 243

(S-H20+NH3)+

(BH)2H+

J(S-H20)+ 132

223 MH+

/ 'r

,,,

i,i

i/ll I ' '1 1 O0

(S-H)'NH4+ I'

'""1'''

' I'''"l''"l

150

b

250

245

f--

~

~

113

n,"

i'

i "' I' 1 't.., ' '

I00

,,,

300

o

100t

Z LLI I'-Z

j,i,1 I

I ' '''

200

I ''""

''''

150

''

''1

'1 '

'''

200

C

100 I

JiM+" ' I

300

0

BH2+ 152

N~I

>

H= HOH _ H ~

50

Mr 283

150

MH+ 284 /

W

I'

100

['1

'l,'

' '1'')"1''''

150

d

50

I'

' ' '1'

'''1

200

....

I'"

250

' I

300

NH2

~ I00" I--

I ~

Z D-l

Z ,,, ~_~

' r [

250

HO._I~O,~[

MH+ 268 Mr 267

BH2+ I Ht I 100

150

200

250

I 300

m/z FIG. 2. Thermospray mass spectra of(a) cytidine, (b) uridine, (el guanosine, (dl adenosine. Some minor sugar ions are denoted in panel (a). See text for experimental details.

Minor ions, which vary in abundance and depend in part on operating conditions, and are not always observed, include the sugar moiety (S) ions 31 as marked in Fig. 2a, and adduct ions of the molecule and base 31 F. F. H s u , Ph.D. Dissertation, University of Utah, Salt Lake City, U t a h , 1986.

[44]

ANALYSIS OF R N A HYDROLYZATES BY L C / M S

-14-6

100-

BH2+

_

809

MH÷ 258

50iI

I--

0

tl H O ~ HO

i,i rF

I

Mr257

OCH3

164 l II '1' ' I"

I00

''

' I ''"

150

I I ' ' ''

I '//'

200

' I

250

' ' 1 ' ' ' ' 1 ' ' ' ' 1

300

~/~ FIG. 3. Thermospray mass spectrum of 2'-O-methylcytidine, showing 146 u difference characteristic of methylribose.

with NH4 +, Na, + or protonated acetonitrile. In some cases, particularly derivatives of cytidine, proton-bound base-base dimers may be formed (see Fig. 2a, m/z 223) and recognized by their exact chromatographic correspondence with the corresponding BH2 ÷ or MH ÷ ions. Simple Modification. Methylation and other simple forms of modification are most easily recognized by mass shifts: in the base by differences between the observed value for BH2 + and the corresponding values of the four major nucleosides, and in the sugar by differences between MH ÷ and BH2 ÷ in the same spectrum (e.g., 146 mass units for 2'-O-methylation) as seen in Fig. 3. Note that hydroxylation and thiation (replacement of O by S) in the base each result in a 16 mass unit shift, so isomers such as 5hydroxyuridine and 2- or 4-thiouridine must be differentiated by other means such as relative retention time (Table II). Pseudouridine and its derivatives are distinguished by the absence of BH2 + ions(Fig. 4), a characteristic of C-nucleosides resulting from increased glycosidic bond strength) 2 Fragmentation energy is, therefore, channeled into other dissociation routes which follow analogous pathways reported in the NH3-chemical ionization spectrum 3° involving expulsion of H20 (m/z 227, 209) and cleavage through the ribose ring (m/z 155, 185) with corresponding ammonia complexes (m/z 172, 202). Complex Modification of the Base. Posttranscriptional processing of RNA produces more than 50 forms of base modification, mostly in tRNA and many with side chains that produce characteristic fragment ions. Thermospray mass spectra of some of the major classes of modified nucleosides from tRNA are shown in Fig. 5, and serve as examples of the influence of side chains on the dissociation of MH ÷ .26 In this respect, two processes are represented: one is protonation of basic sites in the side chain which strongly directs side chain fragmentation; the other involves 32 j. M. Rice and G. O. Dudek, Biochem. Biophys. Res. Commun. 35, 383 (1969).

810

NUCLEIC ACID CONSTITUENTS

[44]

o

245

LO P-

~

Z I,I

o

Ho 50

r 244

I~

HO

OH

+NH3 155~ +NH3 172 1 8 5 ~

LLI ~-

I

,

I~'''l 100

''''

. 150

.

.

202

.

.

. . 200

209 227

''

'''l''''l 250

500

m/~ FIG. 4. Thermospray mass spectrum of pseudouridine, showing the absence of BH, + ions characteristic of C - - C glycosidic bonds.

thermal effects, which contributes to the absence of MH ÷ ions in the spectra of some compounds having complex side chains (Fig. 5c,d). N-Acetylation of cytidine, a modification found in many tRNAs, leads to elimination of ketene (42 u) from both the molecular and base ions (Fig. 5a), a reaction which is broadly characteristic of O- and N-acetyl derivatives in the case of other ionization modes. In 5-methylaminomethyluridine (Fig. 5b), loss of CH 3 from the base produces a stable protonated 5-methylamino ion, m/z 158. This ion is characteristic of several related 5-substituted uridines and, although of relatively low abundance, can be effectively monitored for detection in hydrolyzates of unfractionated tRNA by LC/MS.~3 Very polar nucleosides dissociate completely in the heated spray and [in contrast to fast atom bombardment (FAB)] molecular species are of low abundance or are absent, although this disadvantage is, in part, compensated for by the occurrence of characteristic and structurally informative fragment ions. Two leading examples are represented by the highly modified nucleosides N-[(9-fl-n-ribofuranosyl-9H-purine-6-yl)carbamoyl]threonine (t6A) (Fig. 5c) and queuosine (Q) (Fig. 5d), common constituents of tRNA. In both cases, side chain ions constitute the most abundant fragments formed as shown in Scheme 2: m/z 120 and 116, respectively. Formation of protonated adenosine (m/z 268) and adenine (m/z 136) from t6A (Fig. 5c) clearly marks the spectrum as that of an adenosine derivative. With the exception of the minor ion m/z 294, structural detail of the threonyl side chain is absent, in contrast to the El mass spectrum of the volatile trimethylsilyl derivative. 25,33

33 H. Kasai, K. Murao, S. Nishimura, J. G. Liehr, and P. F. Crain, Eur. J. Biochem. 69, 444 ll976). (Microfilm Suppl. AO-553.)

[44]

ANALYSIS OF RNA HYDROLYZATESBY LC/MS a

811

HNCOCH3

100Z Ld FZ

~

HO

r

so BH2+ 154

~.

k~

I ....

I"'''ll['""l

''

150

100

],

I

I'

L [

200

b BH2 +

LdZ

ZN-

_

'1''

250

I

HO

I, J

i 'i'i'' . . . . . . 150

I ; 200

M r 303

OH

',

MH 3o,+

''"1 250

tr'"'l 500

'

'1 350

c Threonine.H + 120

5

Mr 412

Adenosine'H +

268

50 Adenine .H+

136 150

IO0

--

150

200

250

300

116

100

t d ,,,

294

425

Mr 4,09

~L,,I ' '

50 163

\

hi fY

150 \ iiii

1 O0

i

i'1

I

; ,2'

150

180

295

/

/

I 't', , I I 4

200

.

.

~/=

.' I ' ' . ' 250

~ I .' L ' I'l. ' ' ' "//'

I

300

425

FIG. 5. Thermospray mass spectra of side chain containing nucleosides from RNA. (a) N4-Acetylcytidine; (b) 5-methylaminomethyl-2-thiouridine; (c) t6A: (d) queuosine. See also Scheme 2.

Although the thermospray mass spectrum of queuosine was shown earlier to exhibit an MH ÷ peak, 18the abundance of the MH ÷ ion is dependent on vaporization conditions and the spectrum in Fig. 5d is more representative of that usually obtained. The most important ions, shown in Scheme 2, result from initial protonation of the side chain nitrogen, in

812

NUCLEIC ACID CONSTITUENTS 294 --I

1

[44]

~H3 CHOH

b

NH--CO-4-,NH--CH

~H3

C02H

CHOH +

I

H3N--CH I C02H

HO

rrr/z 120

OH

t6A OH

+

m/z 163 ct~/z 295

R R = H R = ribosyl

NH

+NH3

CH2

rr~/z 11 6

I

I~.~ I C H 2 H2N

OH

Q H2N . ~ HO

0 H ~~N I

HO

OH Q

H2N

CH2

H

~'rr/z 180

SCHEME2. Ions from thermospray ionization of hypermodified nucleosides t6A and Q.

accord with the solution-phase properties of queuosine. 34 The abundant side-chain ion m/z 116 played a major role in the discovery of the epoxide derivative (oQ), in which the analogous ion occurs at m/z 132. ~8

Thermospray Mass Spectra Acquired in D20 If LC/MS is carded out in D20 or similarly labeled mobile phases, active hydrogen will be rapidly exchanged for deuterium, leading to very high exchange levels of approximately 97%. Details of column equilibration and other experimental procedures are given by Edmonds et al. ~2The exchange procedure is useful under two circumstances: (1) to distinguish 34H. Kasai, Z. Ohashi, F. Harada, S. Nishimura, N. J. Oppenheimer, P. F. Crain, J. G. Liehr, D. L. von Minden, and J. A. McCloskey, Biochemistry 14, 4198 (1975).

[44]

ANALYSIS OF R N A HYDROLYZATES BY L C / M S

813

isomers which differ in the number of exchangeable hydrogen atoms, a circumstance most likely to occur when one or both reference isomers are unavailable to establish HPLC elution times, or when chromatographic separation is not possible; and (2) for characterization of structural unknowns, as an aid in developing plausible structures which can then be tested by other means. In either case, caution should be exercised that some exchange of "nonactive" carbon-bound protium is not effected. Such exchange occurs in the heated thermosprayjet ~2'35and is characteristically quantitative at C-8 of guanosine derivatives, and to the extent of 45 to 59% at C-8 in adenosine derivatives) 6 The D20-LC/MS method is particularly useful because exchange data can readily be obtained, on nanogram-level quantities of material in complex mixtures, without necessity of isolation of components of interest. An example of the method is given by the differentiation of the isomers N4,2'-O-dimethylcytidine and 5,2'-O-dimethylcytidine in digests of unfractionated Pyrodictium occultum tRNA (Scheme 3). The isomers, which could not be reliably chromatographically distinguished or resolved from another nucleoside, both would exhibit the same MH ÷ value. Upon LC/MS analysis in D20, the molecular species shifted very clearly from m/z 272 to 277, as shown in Scheme 3, indicating four exchangeable hydrogens in the neutral molecule, in favor of the 5,2'-O-dimethyl isomer. The same approach applies to other nucleoside or base isomers which differ by N or O vs. C substitution.

Analysis of RNA Hydrolyzates Because of the added dimension of mass as a HPLC detection parameter, both reliability of identification and increased dynamic range of analysis are markedly enhanced using LC/MS (as configured in Fig. 1) compared with UV detection alone. If the analysis is oriented toward detection of structurally known or expected nucleosides, a direct combination of retention time and mass (usually MH + and BH2 ÷ ions) for direct comparison with reference data as in Table II, will usually suffice. The HPLC chromatogram shown in Fig. 6 is typical of a complex RNA hydrolyzate, with the exception that enzymatic digestion was carded out so as to produce several dinucleotides (for discussion in a subsequent section). The assignments marked were made by comparison of mass and retention time data with those in Table II, and following the guidelines for 3~ M. M. Siegel, Anal. Chem. 60, 2090 (1988). C. G. Edmonds, S. C. Pomerantz, F. F. Hsu, and J. A. McCIoskey. Proc. 36th A S M S Conf. Mass Spectrom. Allied Topics, San Francisco, CA p. 1256 (1988).

814

NUCLEIC ACID CONSTITUENTS

HO

oo_

~>

HO MH +

OCH 3 =

MH +

DO

272

=

CH 3

276

ND2

OCH 3 =

OCH3

MD +

.~]2

HO

[44]

DO

272

OCH 3

MD +

=

277

SCHEME 3. Mass shifts from D20-LC/MS used to distinguish N- and C-methyl nucleoside isomers.

1oo

3

b 2 19

50

22

7 8

©

13

25

16

28

2

21

23

29

J

0

11

.£3

18

24



Ld rY

100- C

ac4C~

rn,/z

154

Z Ld Z

7 A

50

Ld

>

5 Ld

L~A

¢Y

100-

d

......

.... t . , . . r . . u . . . . r . . u . . . . r . . u . . . . i . . , u . . , .

.... i . . . . r . . . ~ . . . . r . . u . . . T . . U . . . T . . . ~ . .

ac4C

rr~/z

I

286

Z LtJ Z

50

UJ

Ld ¢Y

.... l,, ~,iI,,

15

III,I,II,I1,]I,,,l,

17

, JIIIIIIII'''I

19

'

~ . . h . t . T . . . ~ . . . . r . . . , . . . . 1 . . . . . ¢..'..i..,.~....1

21

23

25

Time, minutes

FIG. 10. Effect of minor ions associated with major nucleoside components. (a) UV detection; (b) m/z 268 channel, marking the elution position of adenosine; (c) m/z 154, characteristic of NLacetylcytidine (ac4C), showing "false positive" at 20.4 min; (d) m/z 286 channel, characteristic of ac4C.

bly 5-methyl-2'-deoxycytidine, Table III) under the present circumstance they will usually be at undetectable levels, and the only problem from DNA contamination will be the four major deoxynucleosides, at predictable elution times (Table III). As shown by the spectrum of thymidine in Fig. 12, deoxynucleosides are readily recognized by the 116 mass unit

[44]

ANALYSIS OF RNA HYDROLYZATESBY LC/MS 7.7 Z Ld I,-Z

.3.9

>

Ld

254 nm

__LL_J L__J

Ld ¢Y

J

i'`.'~.'.'~''.u'.''~.'u.'.'~'.u''''~'''u..,.~'.'.~'''~'''u'~'~.'.u.'.'~.'~u'~.'~.''u.''.~

100Z LQ I-Z

821

b

IU

C

rr~/z

247

,o

Ld

ILl or"

~""'"'i ..... ~'i""'" 2 4

~, ' ,;,,, I ;,,, i ,g,, I / . . L. .i . .^. ,I ;,. ,. . . . . . . .I 6 8 Time, minutes

1000

2,$7

_z

','

,

,,,,, . . . . . . . . . . . . . . . I i I i

HO

IL!d~. l,I

i,, 150

,o.,,,! t d j t , 200

,

.... '

I

i

'

'

M~ 248

"

''

i

250

12

HLN~

50

100

I

10

J

Z Ld F--

L~

i

"

' I

300

i

350

Fro. I I . Detection of the non-UV-absorbing nucleoside dihydrouridine (D). (a) UV detection; (b) m/z 247 channel (MH ÷ for D); (c) mass spectrum recorded at 3,1 min.

difference between MH ÷ and BH2 + ions, compared with 132 or 146 units for ribonucleosides (Scheme 1). Mono- and dinucleotides may result from incomplete enzymatic digestion (cf. elsewhere in this volume28), and may be difficult to recognize under some circumstances due to the absence or very low abundance of molecular species. The monophosphates of the major nucleosides--those most prevalent as a result of incomplete hydrolysis--will elute earlier than the corresponding nucleosides (compare data in Tables II and III), but will exhibit essentially the mass spectrum of the nucleoside, which is undoubtedly formed in the heated spray. 31 A more difficult problem arises in the (less likely) event that monophosphates of modified nucleosides are present at detectable levels, because very few reference elution times are

822

NUCLEIC ACID CONSTITUENTS

15..3- a ~--~

'~ ~ ~-~

Z Ld I-Z

--

lad

[44]

13

254 nm

16

7.7

i,i . . . . •••••••••••••••r•••••'•••••••l••••••••••••••••••••••••r'••P•••••'••••••'r•••••••••••'••••••

5

17

]

19

21

Time,

23

25

minutes

100Z hi I'-Z Ld

HO

50-

M r 242

BH2 + 127

Ld r'r"

J.J~ I

100

'

I ,i . '

. '

I

150

. '

. '

'

. '

. I"'

200

. '

r

~

I

250

'

'

'

'

I

300

. . . .

I

350

~/z FIG. 12. Thermospray mass spectrum of thymidine resulting from contamination of the S. solfataricus tRNA isolate by DNA. (a) UV detection; (b) mass spectrum recorded at 18.4 min, showing 116 u difference characteristic of deo×ynucleosides.

known. Here also, tentative assignment as a nucleotide can be made by finding ions corresponding to the nucleoside, but at the wrong elution time. Dinucleotides may be formed principally due to marked resistance of some modified residues to nuclease digestion.28 Among these modifications is ribose methylation, which renders the 3'-phosphate group to slowed digestion by nuclease P1 so as to form nucleotides of the type NmpN. In general, dinucleotides can be characterized by the observation of ions corresponding to both constituent nucleosides and their bases, having identical elution time profiles, but at a retention time corresponding to neither nucleoside. A typical result (from intentional underdigestion) is shown in Fig. 13a, in which is plotted the mass spectrum of the 30.7-min HPLC peak from Fig. 6, from which ions were chosen for profiling. The matched elution profiles for cytosine (m/z 112), the MH + ion of a trimethylated guanosine (m/z 326), and the base dimethylguanine (m/z 180) are experimentally indistinguishable, leading to the probable partial structure m2GmpC. In some instances such as this, one of the four diagnostic ions is absent (MH ÷ of cytidine in the present case), an effect probably associated with the chemistry of thermally assisted dissociation in the spray, for which there is presently insufficient information to permit pre-

[44]

ANALYSIS OF R N A HYDROLYZATES BY LC/MS

823

100 Z Z LO

uJ

50:

1;o

IM

i

i

i

00

I

i

,

, i

326

i

,

,

,

,

200

150

I

,

,

250

,

i

,

,I,

,

,300

i

,

,350

~/~ 100-

m/z

112

Z t.d Z LtJ

50

W

.... i . , , i , . , u , , , , r , , , i , , , , r , . u , , , T , , U , , , , i , , ,

r,,u.,,T,,U.,,.r,,u.,,T,,U.,,.i

u.

100"

rn,/z

180

Z Ld Z hi

50

Ld .... I.,.,I.,,u,,,.r,.,~.,.,r,.u,,,,I,,,u,.,W,,

~2 100"

u,

r,,u,,,,r,,,i,,.,r,,u-,,r..u,,,,i

d "rn,/z

326

Z bJ

z h.I

5O

~

I

25

I

i

A I

27

i

AJ

I

I

e

I

29 Time,

i

I

A

i

I

31

i

I

i

I

33

i

A^

I

A

A i

I

35

minutes

FIG. 13. Recognition of dinucleotides as impurities in RNA digests. Ca) Mass spectrum recorded at 30.7 min (Fig. 6, peak 28). Mass chromatogram from ions shown in panel Ca) (data from channels not exhibiting chromatographic correspondence not shown); (b) mlz 112; (c) m/z 180; (d) m/z 326.

824

NUCLEIC ACID CONSTITUENTS

[44]

diction of occurrence. From the evidence in hand, there is no direct means to place the two methyl groups in guanine, although a parallel approach in such cases is to collect the HPLC peak, and digest it completely to nucleosides for subsequent LC/MS analysis. 3a In the present example, the guanosine derivative is probably Na,N2,2'-O-trimethylguanosine (m2Gm), based on the occurrence of this unusual nucleoside in S. solfataricus tRNA 2° and limited sequence data. 3a Other responses in the m/z 180 and 326 channels around 31.7 min suggest the presence of related nucleotides, but at levels too low for further characterization. If dinucleotides are formed which do not contain ribose methylation, the possibility of two sequence isomers results, which cannot be distinguished on the basis of the data in hand. For example, the HPLC peak at 31.2 min in Fig. 6 exhibits ion current profiles (not shown) clearly demonstrating presence of the elements of adenosine and the tricyclic guanine derivative mimG, 19 but the dinucleotide isomers mimGpA and ApmimG cannot be distinguished. Acknowledgments The authors acknowledge numerous contributions of C. G. Edmonds, F. F. Hsu, and M. L. Vestal in early stages of development of the methods described, under support by the National Instituteof General Medical Sciences ( G M 21584). Hydrolyzates for the presently described experiments were prepared by P. F. Cmin. Samples of lysidinc and 2'-O-ribosyladenosine were generously provided by S. Yokoyama and G. Keith, respectively.

N. Takeda, unpublished experiments, 1989. 39 y. Kuchino, M. Ihara, Y. Yabusaki, and S. Nishimura, Nature (London) 298, 684 (1982).

3s