mass spectrometry

ANALYTICALRIOCHEMISTRY

176, 269-277 (1989)

Identification of Phenylthiocarbamyl Amino Acids for Compositional Analysis by Thermospray Liquid Chromatography/Mass Spectrometry Bikash C. Pramanik,* Carolyn and Clive A. Slaughter*,’

R. Moomaw,”

Claudia

T. Evans,?

Steven A. Cohen&

*Howard Hughes Medical Institute, University of Texas Southwestern Medical Center at Dallas, Texas 75235, t V. A. Medical Center and Universitv of Texas Southwestern Medical Center at Dallas, Texas 75235, and $ Waters Chromatography Division Millipore. Milford, Makachusetts 01757.

The chromatographic separation of amino acids for compositional analysis of peptides and proteins is commonly performed by reverse-phase high-performance liquid chromatography of amino acid residues that have been derivatized with phenylisothiocyanate. The present report describes an extension of this method, which employs thermospray liquid chromatography/ mass spectrometry to confirm the identification of the resulting phenylthiocarbamyl (PTC) amino acids. A standard HPLC separation method has been adapted for use with the thermospray technique, and on-column mass spectra of standard synthetic PTC-amino acids have been acquired. These spectra show characteristic fragmentation patterns not seen in the corresponding cyclic phenylthiohydantoin amino acid derivatives. The LC/MS method has been tested on hydrolysates of bovine serum albumin, porcine insulin, and human placental collagen. In each case, the mass spectra of components eluting with the same retention times as the standard PTC-amino acids are similar to those observed in the standard amino acid mixture. Other components display mass spectra that can be interpreted in terms of known in vivo or in vitro modifications to amino acid side chains in these proteins. The LC/MS method has assisted in the identification of by-products of the derivatization reaction. It has also been applied to a study in which an enzyme, citrate synthase, isolated from porcine heart, was compared to the protein expressed by a recombinant porcine citrate synthase gene in Escherichia coli. The data showed that the recombinant protein lacks a modified residue, trimethyllysine, which is present in the enzyme expressed in mammalian c 1989 Academic Press, Inc. tissues.

’ To whom

correspondence

should

000%2697/89 $3.00 Copyright IZI 1989 by Academic Press, All rights of reproduction in any form

be addressed.

of

Amino acid composition analysis of peptides and proteins has traditionally been performed by the methods pioneered by Moore et al. (l), in which the free amino acids liberated by peptide hydrolysis are separated on a sulfonated cation-exchange resin and detected by colorimetry after postcolumn reaction with ninhydrin. Recently, the quest for enhanced sensitivity and convenience has resulted in the widespread adoption of an alternative methodology that is based upon precolumn derivatization of the amino acid residues with phenylisothiocyanate (PITC)” to yield phenylthiocarbamyl (PTC) derivatives (2-4). These compounds are separated by reverse-phase high-performance liquid chromatography and detected by ultraviolet absorption. The identification of amino acid residues in compositional studies is typically based upon comparisons of retention times with suitably derivatized standard amino acids. For the purpose of confirming identifications made in this way, mass spectrometric data are of unique value. Electron impact and chemical ionization mass spectrometry have previously been used for the analysis of amino acid mixtures (5), often with on-line separation of the residues by gas-liquid chromatography. Usually the amino acids are derivatized to permit passage through the gas chromatograph and to promote ionization in the mass spectrometer. HPLC, however, is advantageous for resolving involatile, thermally labile compounds because it avoids prolonged exposure of the analytes to elevated temperatures and permits the exploitation of interactions among analyte, mobile phase, and stationary phase to optimize separations. The thermospray technique (6,7) provides a simple and conve’ Abbreviations used: PITC, phenylisothiocyanate; thiocarbamyl; PTH, phenylthiohydantoin; TIC, PTU, phenylthiourea; EPTU, ethylphenylthiourea.

total

PTC, ion

phenylcurrent;

269 Inc. reserved.

270

PRAMANIK

nient approach to interfacing a liquid chromatograph with a mass spectrometer and has been used as a “soft” ionization technique to analyze a variety of mixtures containing involatile, thermally labile components (8). Its suitability for the analysis of amino acids and their derivatives was suggested by recent studies of phenylthiohydantoin (PTH) amino acids (10). These derivatives, which are cyclic analogs of the open-chain PTC amino acids, are produced during the Edman degradation, which is used for amino acid sequence analysis (9). The present report describes methods by which thermospray LC/MS can be used to provide information for identifying PTC amino acids during compositional analysis of peptides and proteins. Conditions for on-line reverse phase HPLC of PTC amino acids are presented, and tests of the method by application to hydrolysates of bovine serum albumin, porcine insulin, and human placental collagen are reported. The utility of the method is exemplified in a study of the amino acid composition of a mammalian enzyme, citrate synthase, expressed by a recombinant gene in Escherichia coli. These results have been presented in preliminary form.3 Similar studies have also been initiated by others to analyze hydrolysates of covalently modified synthetic peptides. MATERIALS

AND

METHODS

HPLC was performed with a system from the Waters Chromatography Division of Millipore configured as described previously (10). Separations were performed on a Waters/Millipore 3.9 mm X 15 cm Nova Pak Cl8 steel column. Mass spectrometry was performed with a VG Masslab Model 30-250 quadrupole mass spectrometer equipped with a standard VG Masslab thermospray interface. Standard amino acids and quantitative amino acid mixtures (standard H), PITC, and constant-boiling hydrochloric acid were purchased from Pierce Chemical Co. N’-Trimethyl-L-lysine was from Behring Diagnostics (Calbiochem). Phenol was from Mallinkrodt, triethylamine (Gold Label) from Aldrich Chemical Co., ethanol from Aaper Alcohol and Chemical Co. (Shelbyville, KY), and PTC-amino acid diluent from Waters/Millipore. Bovine insulin, bovine serum albumin, human placental collagen (type IV), and porcine heart citrate synthase were from Sigma Chemical Co. HPLC-grade water was produced by a Millipore Mini-Q water system and HPLC-grade acetonitrile was purchased from Burdick and Jackson. HPLC-grade ammonium acetate and acetic acid were from J. T. Baker Chemicals. Protein samples were dried; hydrolyzed at 120°C for 18 h with vapor-phase, constant-boiling hydrochloric 3 Pramanik, B. C., Moomaw, C. R., Slaughter, C. A., Evans, C. T., and Cohen, S. A. 36th ASMS Conference on Mass Spectrometry and Allied Topics, June 5-10, 1988, San Francisco, CA. 4 Chen, T.-M., and Coutant, J. E. 38th ASMS Conference on Mass Spectrometry and Allied Topics, June 5-10, 1988, San Francisco, CA.

ET

AL.

acid containing 0.5% (v/v) phenol in a sealed, evacuated vessel; redried; and neutralized as described in (3). PTC derivatives were formed from the resulting free amino acids as described previously (3). Briefly, 20 ~1 of a freshly made reagent consisting of 7:l:l:l (v/v) methanol:triethylamine:water:PITC was added to the dried amino acid mixture, and the derivatization reaction was allowed to proceed for 20 min at room temperature. The reaction mixtures were dried thoroughly on a lyophilizer and finally redissolved in PTC amino acid diluent for chromatography. HPLC separation of the PTC amino acids was achieved by a modification of the method described in (3). Gradients were produced by mixing solvent A, which contained 94 parts by volume 0.145 M ammonium acetate, pH 6.4, plus 6 parts acetonitrile, with solvent B, which contained 60 parts by volume acetonitrile plus 40 parts water. Column temperature was maintained at 43°C and chromatographic solvent flow at 0.7 ml/min. Initial conditions for each separation were 100% A/O% B and the composition was changed to 40% A/60% B over 15 min using a convex gradient (curve 5). Each separation was followed with a 2-min column cleaning at 0% A/100% B before returning to initial conditions. Postcolumn “make-up” solvent containing 0.2 M ammonium acetate was introduced upstream of the mass spectrometer at a rate of 0.3 ml/min to promote ionization. Thus a combined flow of 1 ml/min was introduced to the LC/MS interface. The mass spectrometer was tuned to record positive ion mass spectra at unit mass resolution by conventional adjustments to the ion filter and detector. Thermospray nozzle and chamber temperatures were optimized at 320 and 34O”C, respectively, and the mass analyzer was scanned from either m/z 110 to 430 or 120 to 450 with a scan time of 1 s and an interscan delay of 0.1 s. RESULTS

AND

Chromatographic

DISCUSSION

Separation

for On-Line

LC/MS

A reverse-phase chromatographic method to resolve PTC amino acids for on-line mass-spectral analysis was developed by modifying a previously described method (3). The prototype method, which has been described commercially by Waters/Millipore as the Pica-Tag procedure, employed a sodium acetate buffer, pH 6.4, containing triethylamine as an organic modifier. Because thermospray requires a volatile mobile phase, sodium acetate was replaced by ammonium acetate. Triethylamine was omitted because early attempts to include it resulted in rapid mass spectrometer source contamination and loss of signal. Elution was affected by an acetonitrile gradient as before. No change of buffer salt concentration or pH was required, but the column temperature was increased from 38 to 46°C to resolve PTC-Arg from PTC-Thr. A representative uv absorption chromatogram is shown in Fig. 1. The results with the new

MASS

SPECTROMETRIC

200 4 40

PHENYLTHIOCARBAMYL

400 6 20

AMINO

600 1200

ACID

271

IDENTIFICATION

800 1539

1000 1919

SCAN TIME

1000 1919

SCAN TIME

PHE

200 4 40

FIG. 1. component, By-product

400 8 20

Ultraviolet Absorption (254 nm) and TIC (m/z 110-430) normalized to the highest peak in each chromatopram. due to ammonia; UNK. unknown by-product.

600 1200

600 1539

chromatograms of’ a standard PTC amino acid mixture containing 1 nmol/ Numerical peak annotations represent scan numbers at peak maxima. NH,],

method differed from the old in only one respect. A uvabsorbing derivative of ammonia, which eluted after PTC-Pro in the sodium acetate system, emerged before PTC-Pro in the ammonium acetate system. In some experiments, in which the chromatographic column had been used for many separations, a second difference was observed. A uv-absorbing by-product of the PITC coupling reaction, which eluted before PTC(PTC)-Lys in the sodium acetate system, emerged after PTC(PTC)Lys in the ammonium acetate system. All the PTC amino acids eluted in less than 17 min at a flow rate of 0.7 ml/min. For on-line thermospray mass spectrometry, 0.3 ml/min of 0.2 M ammonium acetate was introduced into the solvent flow upstream of the mass spectrometer to promote analyte ionization. The chromatographic method was then used to perform LC/MS experiments with a mixture of PTC amino acids produced by derivatization of a standard quantitative amino acid mixture (Pierce standard H). Figure 1 also shows the total ion current (TIC) chromatogram obtained with 1 nmol/component by scanning from m/z 110 to 430. Each peak on the uv absorption chromatogram was accompanied by a peak in the TIC chromatogram, signalling the successful ionization of all the PTC amino acids in the mixture. The comparative uniformity of the peak areas in the TIC chromatogram indicated that PTC amino acids with different side chains ionized with relatively uniform efficiency. Reconstructed single-ion mass chromatograms corresponding to the protonated molecular ion for each PTC amino acid in the mixture, wit.h a single excep-

tion, displayed a peak coinciding with the elution of the corresponding PTC amino acid. The exception was PTC-cystine, for which the largest ion observed within the scanned range was at m/z 255. This ion represents the monomer product formed from the PTCderivitized cysteine dimer by fragmentation. Aside from PTC-Cys, the signal-to-noise ratio for the protonated molecular ion signals was 1OO:l or more in all cases except that of PTC-Asp (m/z 269). In this case, an except,ionally high noise level attributable to a high background at that m/z value was observed and resulted in a signal-to-noise ratio occasionally as low as 4:l. Most of the reconstructed single mass chromatograms displayed more than one peak. The additional peaks coincided with the elution of various other components of the PTC amino acid mixture, coincidentally producing signals at the same mass as the protonated molecular ion of the PTC amino acid of interest. PTC-Leu and -1le appear on the same chromatogram because they are isobaric. The results underscored the necessity of performing chromatography to distinguish unambiguously the components of the PTC amino acid mixture. On-Column

Mass Spectra

On-column mass spectra were acquired for each PTC amino acid. Examples representing amino acids with different chemical classes of side-chain are shown in Fig. 2. The spectra were strikingly different from those for the corresponding PTH amino acids acquired under similar conditions ( 10). For the PTH derivatives, the spec-

272

PRAMANIK

ET

AL.

CH3 Cl+-CHz-S-W3

S s II Ph-NH-C-NH-CH-COOH

PTC - VALINE (m.w. = 252.3)

II

PTC - METHIONINE (m.w. = 284.4)

CltCH3 I

I

Ph-NH-C-NH-CH-COOH I

100

1

[Mnl+ = 285 -3-4 60 *

[M” _ tq,+

= 2(n

-%1 ,MH - H*Slf = 251 -28 1 ,MH - H*S co]+ = 223 - 421 [MH - H*S - co. CH3SH]f = 175

so . 60. 60 . 285 I

0

lk., 100

220

260

300

Ph I CH2

220

300

340 MASS

S

[MH]+ = 301 -18

G-b

II

Ph-NH-C-NH-CH-COOH 2

100 -

JMHJ+ = 2% -)B

,MH - H*ol+ = 283

341 ,MH

60 -

-344 ,MH H*S]* = 257 - *ai [MH - Ii*0 . co]+ = 239

SO-

. 260

PTC - HISTIDINE (m.w. = 290.3)

120

loo-

180

340 MASS

S II Ph-NH-C-NH-CH-COOH

PTC - PHENYLALANINE (m.w. = 300.4)

140

60 -

W-

li*si+

,MH - li*q+

= 257

1w

““explained

350

400

-34 ,MH

= 273

H*O - H*Sl+ = 239

411 ,MH - H*O

H*S - CHNHCH]+ = 198

do-

20.

O+. 100

1 150

200

250

305

PTC - SERINE (m.w. = 240.3)

350

400

150

MASS

s

CHzOH

II

I

Ph-NH-C-NH-CH-COOH

341 (MH - tl*sj+ = 207 -18 161

ax

173

“nexplslned

,MH H*O H*o]+ = 205 - IS ,MH t H*O]+ = 223 - 34 1 ,MH Ii*0 - tt*q+ = 189 - 2s 1 ,MH H*O. H*S -cot+

250

‘91 >. 3w

= 16,

MASS

S II

PTC - GLUTAMICACID (m.w. = 282.3)

205

[MH]+ = 241 -

200

PTC-SW L

PlTC + se,

[WI+= 283-18 -4 IMH- li*q+ = *a

60 .

60.

- 321 1.97 ,“nexPl~md,

0 140

160

220

260

300

340 MASS

FIG. 2.

100

130 142 I 140

187

IS0

On-column positive ion thermospray mass spectra of PTC-Val, -Phe, -Ser, -Met, at least six scans acquired during analysis of a mixture containing 1 nmol/component.

tra for all but two (PTH-Thr and PTH(PTC)-Lys) showed the protonated molecular ion as the most abundant peak. For the PTC amino acids, however, fragmentation was much more extensive. Only in one, PTC-Ala,

PTC-01” d

219 20 -

100

,MH- H*ol+= 286 1 [MH- “20 - n,o1+= 247 -4 IMH- H*O- H*O- co]+= 219

40.

+I+ I 136

0

- CH2 - COOH

165

IW-

241 223

7”’

Ph-NH-C-NH-CH-COOH

220

-His,

Pm + 0,”

+ti 247

. 2c$0

I 14s-- 18 ,142 - n*oj+ = ra

283 I

and -Glu.

300

340 MASS

The spectra

represent

averages

from

was the protonated molecular ion the most abundant peak in the spectrum. The mass spectra could be interpreted in terms of fragmentation and pyrolytic cleavages (Fig. 2). The

MASS

SPECTROMETRIC

PHENYLTHIOCARBAMYL

principal features showed a high degree of commonality among the different PTC amino acids. All but PTC(PTC)-Lys showed a prominent (MH-18)’ peak, which represented the loss of a molecule of water. This loss may represent either a fragmentation or a thermally induced intramolecular rearrangement to yield the PTH derivative. However, the absence of ions in the PTH amino acid spectra that are prominent in the PTC amino acid spectra (see below) suggests that a fragmentation mechanism predominates. All PTC amino acids except PTC-Pro and PTC(PTC)-Lys showed additionally an abundant (MH-34)+ ion, representing the loss of a molecule of hydrogen sulfide. The hydrophobic and aromatic amino acids, as well as PTC-Thr and -Met showed a further loss of 28 Da from (MH-34)‘, representing the loss of a molecule of carbon monoxide. Such common losses indicate processes that affect the PTC skeleton, but to which the PTH ring is not susceptible. In addition, evidence for a pyrolytic cleavage process that yielded PITC and the free amino acid was obtained in all cases except PTC-Cys. The free amino acid was observed as the protonated molecular ion and PITC as a species at m/z 136. However, the intensity of the signals for these moieties varied from residue to residue. All the PTC amino acids displayed a signal for the free amino acid moiety. The corresponding signals for Ala and Gly were below the mass range scanned routinely, but were detected in separate experiments. This signal was generally prominent, and in only one case, PTC-Ser, was it less than 5% of the base peak. The signal for PITC, however, was generally less prominent and exceeded 10% in only five cases, PTC-Ala, -Gly, -Ser, -Tyr, and -His. Fragmentation involving the amino acid side chain was observed in the acidic, basic, hydroxy and sulfurcontaining amino acids (Fig. 2). These fragmentations were of a kind that may prove useful for identifying nonstandard amino acids in future studies. The on-column mass spectra were then compared to those acquired for standard amino acid mixtures that had been subjected to the conditions used to hydrolyze peptides and proteins. The spectra were indistinguishable (data not shown). However, the hydrolysis conditions did produce certain chemical modifications of amino acid side chains. Asparagine and glutamine were converted to the corresponding free acids, and tryptophan was destroyed (data not shown). A new component was observed eluting immediately before PTC-Cys. This component showed a mass spectrum identical to that of PTC-Cys and probably represented the meso isomer of cystine. A minor component was also observed eluting between PTC-Pro and -Tyr (Fig. 1). This component showed a mass spectrum similar to that of PTC-Glu in all respects except that it lacked the peak at m/z 283 for the protonated molecular ion. It co-chromatographed with PTH-Glu, which, at equivalent concentrations, produced a similar mass spectrum. These observations

AMINO

ACID

S II

PHENYLTHIOUREA (m.w. = 152.2)

Ph-NH-C-NH2

s

ETHYLPHENYLTHIOUREA Im.w. = 180.3) 100

80

1

273

IDENTIFICATION

II Ph-NH-C-NH-CH2-CH,

161 [MH]+ = 161 -34+ [MH - H$]+

60 -

= 147

DMPTU 1

PITC t Ethylamine -H

J 136

40 -

20. 147 -

136

0,.

. 120

FIG. PITC

160

200

240

260

3. On-column mass spectra coupling reaction.

320

MASS

of uv-absorbing

by-products

of the

indicated that a small amount of PTC-Glu had been converted to the cyclic thiohydantoin derivative prior to chromatography. Sensitivity

of the LCIMS

Method

In preliminary investigations of the sensitivity of the method in the full scanning mode, different quantities of the standard PTC amino acid mixture were subjected to LC/MS. These experiments revealed that, whereas signals for the protonated molecular ions of all the components except PTC-Cys could be distinguished reliably at the 1-nmol level, observation of characteristic fragment ions permitted identification of all the amino acids at the 250-pmol level with a signal-to-noise ratio of 8O:l or better. The levels of detection were less sensitive than

274

PRAMANIK

ET

AL.

A uv 254nm

LEU

200 4:40

loo-

400 6:20

600 12:oo

TIC 120-150Da.

600 1539

SCAN TIME

800 15:39

SCAN TIME

752

80.

200 4:40

400 a:20

600 12.00

B uv 254nm loo-

GLX 85

c’y’s

60. 60. 40. 20. 0* 200 4:40

TIC 420-450

400 8:20

600 15’39

600 12:oo

SCAN TIME

Da 751 660

SCAN TIME

200 4:40

400 e:20

600 12-00

600 1539

SCAN TIME

FIG. 4. LC/MS analysis of hydrolysates of (1 nmol) (A) bovine serum albumin; (B) bovine insulin; and (C) human placental collagen, type IV. Ultraviolet absorption (254 nm) and TIC (m/z 120-450) are shown for all three proteins. Also shown are the mass chromatograms for the protonated molecular ions of PTC-pyridylethylcysteine (m/z 362) from insulin, and hydroxyproline (m/z 267) and hydroxylysine (m/z 433) from collagen. Each chromatogram is normalized to the highest peak, but the TIC chromatograms are represented with auto-ranging.

those for PTH amino acids, which underwent less extensive fragmentation (10). By-products of the Coupling Reaction The LC/MS method afforded an opportunity to investigate two commonly observed uv-absorbing by-products of the PITC coupling reaction (3). These by-products appear at uv scans No. 431 and 837, respectively, in Fig. 1. One, labeled NH3 (uv scan No. 431) in Fig. 1,

eluted between PTC-Ala and PTC-Pro in the present chromatographic system and results presumably from the presence of free ammonia prior to derivatization (3). The other, labeled UNK (uv scan No. 837), eluted between PTC-Phe and PTC(PTC)-Lys. The mass spectra of these two by-products are shown in Fig. 3. The spectrum of the first is consistent with the structure of phenylthiourea (PTU), which represents an adduct of PITC with ammonia. The spectrum of the second is consistent with the structure of ethylphenylthiourea (EPTU), con-

MASS

SPECTROMETRIC

PHENYLTHIOCARBAMYL

200 4 41

TIC 120-450 100,

Da.

AMINO

ACID

215

IDENTIFICATION

400 8 20

600 1200

800 1540

1000 IS20

SCAN TIME

400 6 20

600 12 00

800 1540

1000 1920

SCAN TIME

148 I

80 60 40 20 0

200 4 41

267

Da.

,1_

IOO80 Go-

I

1 134

40

729

20.0

II

433

,,i

200 4 41

400 8 20

600 ,200

800 ,540

200 4 41

400 6 20

600 1200

800 1540

SCAN TIME

4--Continued

firming a previous tentative identification (13). This compound represents an adduct of PITC with ethylamine, which may be formed by decomposition of triethylamine in the derivatization buffer. Redistillation of the reagent prior to use eliminated this by-product (data not shown). to Hydrolysates

1000 1920

I \

FIG.

Application

SCAN TIME

Da.

IOO80 60 40 20 0'

1000 1920

of Proteins

The LC/MS method was tested on hydrolysates of 1 nmol each of three proteins, bovine serum albumin, bovine pancreatic insulin, and human placental collagen, type IV. The quantity of protein was chosen to be s&icient for recognition of unusual amino acids at the level of one residue per protein or peptide molecule. The results are illustrated in Fig. 4, which shows the chromatogram for uv absorption at 254 mm and the TIC chromatogram, together with reconstructed mass chromatograms for the protonated molecular ions of some components of special interest. The uv absorption due to the phenylthiocarbamyl moiety of each component permitted quantitation of the various amino acids relative to external standard mixtures in the conventional way (3). The data were consistent with the known compositions of these polypeptides except as noted below.

The mass spectral data were used to confirm identifications. In all casesthe mass spectra of components in the hydrolysates were similar to the spectra of the standard components at the corresponding retention times. The absence of extra mass spectral peaks in these components indicated that there were no novel amino acids cochromatographing with the conventional ones. In collagen (Fig. 4C), however, two abundant chromatographic components not present in the standard PTC amino acid mixtures were observed. One eluted at TIC scan No. 825, between PTC-Leu and -Phe. This component gave a signal at m/z 433 (Fig. 4C) which was consistent with the protonated molecular ion of PTChydroxylysine. The various fragment ions (not shown) were also similar to those given by synthetic PTC-hydroxylysine. The other component eluted at TIC scan No. 134, between PTC-Glu and PTC-Ser. The mass spectrum (not shown) was consistent with PTC-hydroxyproline and the signal at highest m/z, 267 (Fig. 4C), was identified as the protonated molecular ion. The insulin sample (Fig. 4B) had been reduced and alkylated with vinyl pyridine by the method of Andrews and Dixon (12) prior to hydrolysis. PTC-Cys was accordingly absent, and a new component (TIC scan No. 727) eluting prior to PTC-Ile was observed. Its mass

276

PRAMANIK

1 324

Da

II

50 i

04, 200

I ‘.\

300

400

500

600

700

SCAN

300

400

500

600

700

SCAN

a

200 FIG. 5.

LC/MS analysis of hydrolysates of (1 nmol) (A) porcine heart citrate synthase and (B) the product of a recombinant porcine gene encoding citrate synthase expressed in E. co&. Ultraviolet absorption (254 nm) and TIC chromatograms (m/z 110-430) are shown for both proteins, plus the reconstructed mass chromatograms for the protonated molecular ion of PTC-trimethyllysine, TMK, (m/z 324), as mentioned in the text.

spectrum was consistent with PTC-pyridylethylcysteine, and a peak at m/z 362, representing the protonated molecular ion for this compound, was observed (Fig. 4B). Both bovine serum albumin and collagen showed components representing PTC-Cys and PTC-meso-Cys as expected for nonalkylated protein hydrolysates. All three proteins showed a low abundance component eluting between PTC-Pro and PTC-Tyr (TIC scan No. 561 in Fig. 4A, scan No. 551 in Fig. 4B, and scan No. 565 in Fig. 4C). This component represented PTH-Glu (see above). Also present in the derivatized hydrolysate of bovine serum albumin was an extra compound eluting slightly later than PTH-Glu (TIC scan No. 601 in Fig. 4A). This extra compound showed a mass spectrum similar to that of PTC-Glu, including the presence of a peak at m/z 283. The identity of this speciesis presently unknown. These LC/MS experiments illustrated the use of the LC/MS method for detecting nonstandard amino acids in protein and peptide hydrolysates. Application Synthase

to Study

of Trimethyllysine

in Citrate

Thermospray LC/MS was applied in a study in which the enzyme citrate synthase isolated from pig heart was

ET

AL.

compared to the protein product of a recombinant porcine gene for citrate synthase expressed in E. coli. The mammalian enzyme contains a single trimethylated lysine residue (13), which is associated with the catalytic site (14), suggesting the possibility that it may take part in the catalytic process. The recombinant protein expressed a catalytic activity that was quantitatively and qualitatively similar to the mammalian enzyme.5 Yet, covalent processing events that characteristically affect proteins in mammalian cells often do not occur in bacterial cells. It was, therefore, of interest to determine whether the recombinant citrate synthase underwent lysine trimethylation in a process similar to its mammalian counterpart. Preliminary experiments indicated that synthetic PTC-trimethyllysine co-chromatographed with PTC-Thr under the conditions of separation described above. Various modifications to these conditions were tested. Resolution of PTC-trimethyllysine from all the other components was achieved by reducing the temperature of the chromatography column from 43°C to room temperature. Hydrolysates of porcine citrate synthase were tested under the modified LC conditions, and a minor component (TIC scan No. 543) cochromatographing with synthetic PTC-trimethyllysine was identified by uv absorption. LC/MS experiments were then performed on a hydrolysate of pig citrate synthase to verify the identity of this component. A mass peak at m/z 324 was observed, consistent wit,h the protonated molecular ion for PTC-trimethyllysine (Fig. 5). Fragment ions were also seen at m/z 130,144,175,245, 247, 276, and 292. These fragments were indistinguishable from masses of the synthetic standard (data not shown). Hydrolysates of recombinant citrate synthase were then tested and found to be indistinguishable from those of the porcine enzyme, except that they lacked the trimethyllysine component (Fig. 5). Dimethyl- and monomethyllysine were also absent (data not shown). These results showed that the recombinant protein does not undergo lysine methylation and indicated that this modification is not required for the expression of catalytic activity by citrate synthase. CONCLUSIONS

The results of these studies indicate that, when used in conjunction with uv detection, thermospray LC/MS analysis of protein hydrolysates derivatized with PITC is useful for establishing the identification of amino acids during polypeptide composition analysis at the high picomole to low nanomole level. The mass spectral information acquired by this method may also facilitate the recognition of unexpected components of such hydrolysates. 5 Evans, C. T., Owens, D. D., Slaughter, C. A. and Srere, P. A. Characterization mutant TMK368K of pig citrate synthase expressed in and isolated from Escherichia coli. Manuscript submitted for publication.

MASS

SPECTROMETRIC

PHENYLTHIOCARAAMYL

AMINO

5 Johnstone,

ACKNOWLEDGMENTS

I

We thank Drs. Thomas J. Lukas and D. Martin Watterson for helpful suggestions concerning modifications to the Waters Pica-Tag chromatography for resolving PTC-trimethyllysine, Ms. Lori Baker for typing the manuscript, and Ms. Martha Burgin and Mr. Norris W. Murray for preparation of the figures.

.

6.

R. Biochemistry 524, Chapman Blakley, C. R., Vestal, M. L.,

ACID

IDENTIFICATION

277

A. W., and Rose, M. E. (1985) in Chemistry and of the Amino Acids (Barrett, G. C., Ed.), pp. 480& Hall, New York. and Vestal, M. L. (1983) Anal. Chem. 55,750-754. and Ferguson, G. J. (1985) Anal. Chem. 57, 2373-

2378.

8. 4th

Symposium on LC/MS and MS/MS, Monteux, October 22J. Chromatogr. 394(l), May 81987. P. (1960) Ann. N. Y. Acad. Sci. 88,602-610. 9. Edman, 10. Pramanik, B. C., Hinton, S. M., Millington, D. S., Dourdeville, T. A., and Slaughter, C. A. (1988) Anal. Biochem., in press. in Protein Microcharacterization 11. Tarr, G. E. (1986) in Methods (Shively, J. E., Ed,), pp. 155-194, Humana Press, Clifton, NJ. 12. Andrews, P. C., and Dixon, J. E. (1987) Anal. Biochem. 161,524-

24 (1986)

REFERENCES 1. Moore. S., Spackman, 30,1185-1190.

D. H.. and Stein,

2. Koop, D. R., Morgan, E. T., Tarr, Biol. Chem. 257,8472-8480. 3. Bidlingmeyer, Chromatogr.

B. A., Cohen, 336,93-104.

W. H. (1958)

G. E., and Coon,

S. A., and

Tarvin,

Anal.

Chem.

M. J. (1982)

J.

T. L. (1984)

J.

4. Cohen, S. A., Tarvin. T. L., and Bidlingmeyer, B. A. (1987) in Proteins: Structure and Function (LItalien, J. J., Ed.), pp. 207-213, Plenum, New York.

528. 13. Bloxham,

D., Parmelee, D. C., Kumar, S., Walsh, K. A., and Titani, K. (1982) Biochemistry 2 1,2028-2036. 14. Remington, S.. Wiegand. G., and Huber, R. (1982) J. Mol. Biol. 158,111-152.