Application of continuous-flow fast atom bombardment mass spectrometry to cyclic nucleotide biochemistry

177

Analytrca Chlmrca Acta, 247 (1991) 177-185 Elsev1er Science Publishers B.V , Amsterdam

Application of continuous-flow fast atom bombardment mass spectrometry to cyclic nucleotide biochemistry James I. Langridge *, Andrea M. Evans, Dipankar Ghosh, Terence J. Walton, A. Gareth Brenton, Frank M. Harris and Russell P. Newton Blochemlstry

Research Group, School of Bzologlcal Sciences, and Mass Spectrometry Swansea SA2 8PP (UK.) (Received

20th August

Research Umi, Unrversrty College Swansea,

1990)

Abstract The modification and optmuzat1on of a continuous-flow fast atom bombardment system for the mass spectrometnc analysis of cychc nucleot1des 1s described.. The most effective solvent matnx was water-methanol-glycerol (70 20 10, v/v/v) and the optrmum probe t1p operating temperature was 40 o C. Modification of the probe t1p was necessary to facilitate solvent evaporation and a solvent degassing and filter system were found to be essential Comparison of the theoretical sensltlvles of the continuous and static systems with cyt1d1ne 3’,5’-cychc monophosphate as analyte indicated a increase of between one and two orders of magnitude, wtth a s1gmf1cant increase 1n signal-to-noise ratio also observed. Analysis of sample tmxtures simulating enzyme incubates has proved successful and the potentml apphcation of dynamtc fast atom bombardment systems 1n quantitative analysts of enzymes 1s further dtscussed here Keywords

Mass spectrometry;

Cyclic nucleotldes,

Fast atom bombardment

While the more established mass spectrometric techniques have been successful in yielding mass spectra of both cyclic and non-cyclic nucleotides, for example electron ionization MS (EI-MS) of the trimethylsilyl derivatives of adenine nucleotides [4], the spectra obtained had a molecular ion of relatively low abundance and it had been necessary to convert the parent nucleotide to a volatile derivative prior to generation of the spectra. The advent of softer means of ionization has however greatly increased the applicability of MS to analysis of biological extracts. Fast atom bombardment (FAB) [5] has arguably proved the most sigmficant in tlus context, enabling MS analysis of a wide range of compounds of biochemical importance including penicillins [6], oligopeptides [7] and a diversity of low M, compounds endogenous to living tissue such as phosphatidyl choline [8] and nucleotides [9,10]. With cyclic nucleotide analysis, a form of tandem mass spectrometry

Cyclic nucleotides are compounds which exert a profound influence upon metabolic regulation [l], and as such the enzymes which catalyze cyclic nucleotide biosynthesis, nucleotide cyclases, and cyclic nucleotide degradation, cyclic nucleotide phosphodiesterases, are significant pharmacological targets (for a review, see Ref. 2). Current assays of these enzymes have the disadvantages that they require radioisotopically labelled substrates and a chromatographic separation of substrate and product; as a consequence they are discontinuous assays and only one datum is obtained from each reaction incubation. In an attempt to obtain a continuous assay, 31P NMR has been utilized to determine adenylate cyclase activity, but this approach has not been widely applied because of its relatively low sensitivity [3]. Recent developments in mass spectrometry (MS) methodolgy suggest that MS may provide a more effective system. 0003-2670/91/$03

50

0 1991 - Elsemer

Science Publishers

mass spectrometry

B.V

178

(MS-MS), comprising FAB-MS together with collisionally induced dissociation followed by massanalysed ion kinetic energy spectrum scanning (CID-MIKES), has provided a system which has enabled the differentiation of 3’,5’- and 2’,3’-cyclic nucleotides [11,12]. This approach has now been applied to identify endogenous cyclic nucleotides m tissue extracts [13-151 and to elucidate the structure of both naturally occurring and synthetic cyclic nucleotide derivatives [16,17], while quantitative application of the FAB-MIKES technique to cyclic nucleotide samples has included kinetic analysis of cyclic nucleotide phosphodiesterases [ 181 and cyclic nucleotide-responsive protein kinases [2]. A modification of the static FAB system was the development of dynamic or continuous-flow FAB (CF-FAB) [19], which was first employed to identify bile acids chromatographically separated using a packed fused-silica column, with the eluent introduced directly to and analysed by a FAB source [19]. A modified version of this dynamic system was later used to analyse a number of peptides [20] and then applied to a quantitative study of peptidase activity [21]. In view of this successful determination of kinetic parameters of peptidase by CF-FAB, and in view of the success we had obtained in the use of static FAB-MIKES in both quantitative and quantitative application to cyclic nucleotide studies, it was decided to develop a continuous assay for phosphodiesterase and nucleotide cyclase activity utilizing a dynamic FAB system in our laboratories. In devising and developing this system it has to be taken into consideration that in comparison with the peptidase, these two classes of enzyme are much less stable, as are their substrates and products, with an additional problem being the much closer structural similarity and 44, of substrate and product. Here we describe our efforts to apply CF-FAB analysis to cyclic nucleotides as a stage in the development of a reaction cell from which samples can be continuously removed and processed into a FAB source, with the reaction profile being determined by multiple ion monitoring. Comparative studies between static and dynamic FAB-MS analysis were concentrated initially upon cytidine 3’,5’-cyclic monophosphate

J I LANGRIDGE

ET AL

(cyclic CMP), in view of its greater solubility in water in the free acid form relative to adenosine and guanosine cyclic monophosphates, and also upon its product of hydrolysis, cytidine S’-monophosphate (CMP). These two compounds comprise substrate and product respectively of the enzyme cyclic CMP phosphodiesterase [22] and thus provide a basis for determining the potential of applying a continuous FAB quantitation method for estimating phosphodiesterase activity in an analogous manner to that used with static FAB [18]. NH, <; 0 0

0 \ HO-P w

6’0

3

N

HtO 2

;

Phosphodmterase

oLI

0 no-a-o 6

0

H -u

w cyclic CMP

EXPERIMENTAL

R

CMP

AND RESULTS

Static positive ion FAB spectra were obtained on a VG ZAB-2F mass spectrometer fitted with a VG FAB source and ion gun, using Xenon as the bombarding gas under conditions previously described [ll-181. CID-MIKE spectra were generated by using nitrogen as collision gas in the second field-free region gas cell at a pressure which gave a reading of 6 x 10e6 Torr on the nearby ion gauge. MIKE spectra were obtained by selecting the ion to be collisionally dissociated with the magnetic sector and then scanning the electric sector voltage under data system control as previously described [ 11-l 81. Dynamic FAB mass spectra and MIKE spectra were obtained on the same instrument, but with the VG standard FAB source replaced by the VG Universal FAB source which includes a heater and thermocouple, thereby enabling the probe tip temperature regulation necessary to prevent solvent freezing, a dynamic FAB probe and a reference probe tip. The flow of sample to the dynamic FAB probe tip was achieved via an uncoated fused-silica capillary along the centre of the probe shaft which emerged flush at a fine capillary hole in the centre of the stainless-steel tip. The capillary was sealed

FAB-MS

OF CYCLIC

179

NUCLEOTIDES

with a ferrule in the coupling at the base of the handle of the probe, leaving a free length of capillary protruding. It was necessary to optimize a series of experimental parameters prior to analytical use of the dynamic FAB system. Initially it was attempted to run the system without the use of a matrix pump. Flow-rates through a capillary tubing of 80 cm X 75 pm i.d. were determined in preliminary experiments for a series of mixtures of methanol, glycerol and water at a pressure of 14.7 p.s.i. The most effective mixture was found to be water-methanol-glycerol (70 : 20 : 10, v/v/v) which provided a flow-rate of 6.7 ~1 mm-‘, but when this matrix was supplied to the probe tip at this rate the source diffusion pump was unable to accommodate it and frequently cut out. A further problem with the initial set up was the occurrence of sudden surges of the source pressure while the solvent was flowing into the source housing of the mass spectrometer. This appeared to be a result of either bubbles which form in the solvent by hydraulic action during passage through the fine capillary causing a rise in source pressure on arrival at the probe tip, or the solvent not evaporating rapidly enough from the probe tip but periodically dripping into the source housing. In order to overcome these problems it was necessary to make several modifications to the initial system. The rate of evaporation of the solvent from the probe tip was increased by packing the inside of the copper section of the probe with solid copper and passing the silica capillary through a tight fitting stainless-steel tube (Fig. 1); these modifica-

VESPEL

INSULATOR \

I

STAIWSS-STEEL COPPER

SOLID

PROBE

PROBE TIP

/

COPPER

TIP

\

.

STAINLESS-STEEL

NEEDLE

Fig 1 CF-FAB probe tip modlfled to Improve thermal conductwty by the mcluslon of a stamless-steel needle surrounded by sohd copper

tions improved the heat conductivity into the solvent, thereby increasing the rate of evaporation significantly. To obviate problems related to bubble formation in the solvents, the solvents were degassed in an ultrasonic bath for 15 min after mixing and immediately before use; in order to obtain a slower and more controlled flow-rate into the FAB source a silica capillary column of narrower internal diameter (50 pm) and greater length (1.5 m) was used and was connected to a Jasco Familic-100N syringe pump. This pump was capable of providing a flow-rate range of 1~1 mm’ to 90 ~1 min-‘; the optimum flow-rate was found to be 2-3 ~1 min-’ as judged from obtaining good quality spectra while reducing to a minimum the frequency of instrument shut down due to vacuum pump cut out. As a means of preventing possible blockage of the capillary column by dust particles and other impurities each solvent was filtered through a Acrodisc filter (Gelman), and as a secondary measure a Valco lo-pm stainless-steel filter of zero dead volume (Vici) was incorporated m line at the Junction of the capillary to the 50-~1 syringe in the syringe pump. Under these modified conditions the optimum temperature was determmed by applying cytidine 3’,5’-cyclic monophosphate at a concentration of 3 pg ~1~’ and attempting to obtain FAB mass spectra at a range of indicated temperatures between 30 and 60°C. Above 50°C the mass spectrometer consistently cut out, at temperatures over 45 o C and below 35 o C poor quality spectra were obtained, thus 40 o C was selected as the probe tip operating temperature. To compare FAB mass spectra and CID-MIKE spectra generated by static and dynamic FAB, two series of sample solutions containing 10 ng-10 pg /.L-’ of cyclic CMP and CMP were separately made up in glycerol-water (1: l), and 3 ~1 of these solutions applied to the static FAB system and the positive ion FAB mass spectra, and the CID-MIKE spectra of [MH]+ from the mass spectrum of each compound (m/z 306 for cyclic CMP and m/z 324 for CMP) determined. For generation of the analogous FAB-MS spectra by the dynamic FAB system, 50-/.~1aliquots of the samples at the same concentration range in water-methanol-glycerol (70 : 20 : 10) were intro-

180

J I LANGRIDGE

duced via the Jasco syringe pump and spectra obtained at a flow-rate of 3 ~1 mm-‘; CID-MIKE spectra were then generated from selected parent ions in identical manner to that described above for the static FAB system. On qualitative examination, the FAB mass spectra obtained for cyclic CMP and CMP showed clear similarities, reflecting the fact that the only structural difference between them is the presence of a phosphodiester bond in the former and a monoester in the latter. The spectra obtained from cyclic CMP (Fig. 2) contain an intense ion at W-4 +r m/z 306, with major peaks corresponding to [M + Na]+, [MH + Gro]+ and [MNa + Gro]+ present at m/z 328, 398 and 420 respectively (Gro = glycerol). Matrix-derived peaks are evident at m/z 321, 369, 461 and 483. In the FAB mass

spectra obtained from CMP (Fig. 3), the same pattern of peaks at [MH]+, [M + Na]+, [M + Gro] + and [MNa + Gro] + is apparent at m/z 324, 346, 416 and 438 respectively; matrix-derived peaks at m/z 321,369,391,461 and 483 are again present. This similarity of spectra extends to the CID-MIKE spectra (Fig. 4); fragmentation of + in each case provides a MIKE spectrum W-U with the major peak being the protonated base at m/z 112, with other common ions being [BH, 17]+ at m/z 95 corresponding to loss of NH, from the protonated base, at [MH - 17]+ corresponding to the same loss from the parent ion, and at m/z 140 and 154 corresponding to the diagnostic S, and S, fragments [9-121, with the peaks at m/z 214 and 232 representing loss of glycerol from the matrix-derived species isobaric

306

(a)

461

328 321 I

100

ET AL

483

-

z H

306

(b)

s E 2. n z w 2 ;I 1

369 328

398

461

H

483 o-

r~

I

550

300 m/z Fig 2 Positwe ion mass spectra of cyclic CMP obtaned concentration m each case was 1 5 gg ml-’

followng

lonlzatlon

by (a) static FAB and (b) continuous

FAB. Cychc CMP

FAB-MS

OF CYCLIC

181

NUCLEOTIDES

100 2 ‘+I

324 369

346

: ;:

416 391

438

321

P

461 463

x

324

346

369 416

321

391

1 300

Fig. 3 Posltlve eon mass spectra of CMP obtamed concentration m each case was 2.85 pg pl-’

439

461

7 5io

m/z followmg

wtth [MH]+ for cyclic CMP and CMP respectively. In addition to the facility for continuous monitoring of selected ions, the potential advantages of the use of dynamic FAB are an increase in sensitivity and an increase in signal-to-noise ratio [1921]. A method of comparing the sensitivity of static and dynamic FAB is to compare the ion current obtained from a sample of known concentration applied to a static FAB probe, with the sensitivity for the contmuous FAB system calculated as recommended in [23] by: I x 60 -1 sensitivity = C x F C pg where Z = ion current (A); C = concentration of sample (pg PI-‘); and F= flow-rate (~1 mm-‘). Over the range of concentrations examined the sensitivity of the continuous FAB system with cyclic CMP was 50 to 80 fold greater than with

lonlzatlon

by (a) static

FAB

and (b) continuous

FAB.

CMP

static FAB; with CMP the increase in sensitivity was less pronounced, with the increase being 30 to 50 fold. With the CID-MIKE spectra the effect was similar, with the cyclic CMP spectra showing an increase in sensitivity of 60 to 70 fold for cyclic CMP and 35 to 55 fold for CMP. Comparison of signal-to-noise ratio (S/N) in the mass spectra showed an improvement of 1.5 to 3 fold over the series of concentrations for both cyclic CMP (Fig. 2) and CMP (Fig. 3) when determined by comparing [MH]+ to the ten most adjacent peaks [21]. When the relative abundance of [MH]+ is compared to that of a strong matrix derived peak at m/z 369 ([3Gro + HI+), a similar signal-to-noise improvement is observed, 1.3 to 2.8 fold, in both cases. The potential advantages of the continuous system are more apparent when the CID-MIKE spectra are compared with those produced from the static FAB generated [MH]+

182

J I LANGRIDGE

ET AL

11,

(a)

214

1.0

154

289

r

I

4000.00

8000.00

111

(b)

4000.00

Fig 4. CID-MIKE

spectra obtamed from [MH]+ of (a) 3.2 pg gl-’

cychc CMP (m/z 306) and (b) 5 25 pg ~1~’ CMP (m/r 324)

FAB-MS

OF CYCLIC

NUCLEOTIDES

183

(a)

214

(b)

140 + 0.00

289

154 I ~001.01

(

_‘I

214 I

,

1

.I, 8000.00

VOLTS

Fig. 5 CID-MIKE

spectra obtained from [MHI+ of 2.85 gg p1-’ cyclic CMP after (a) static FAB and (b) contmuous FAB.

184

J I LANGRIDGE

ples, a series of solutions containing mixtures of O-3 pg pl-’ of cyclic CMP and CMP were subjected to the CF-FAB process and the relative proportions of the two components calculated from the resultant CID-MIKE spectra as previously described for the static FAB quantitation of phosphodiesterase activity [18]. Good correlation between actual and calculated [cyclic CMP] to [CMP] ratios were obtained with a S.E.M. of f 8.9%.

(Fig. 5). The absolute intensity of the major peak at m/z 112 ([BH,]+) was almost doubled in the continuous FAB generated spectrum, and the intensity of the characteristic peaks at m/z 95, 140 and 154 have increased to a similar extent. The peak at m/z 214, corresponding to [MH - 92]+, which provides an indication of the glycerol-matrix contribution to the parent ion m/z 306, has on the other hand decreased by a factor of 6.9 fold in the dynamic spectrum, thus in the CID-MIKE spectra shown the S/N ratio, calculated on the basis of major sample derived ion relative to major matrix derived ion, has improved by 12.9 fold in the dynamic spectrum. This improvement was observed over the range of cyclic CMP concentrations, with an increase in S/N of between 9.5 and 13-fold, and similar effects were observed with the CMP series. An important observation in the context of the development of the continuous enzyme assay was that there was no significant difference in the ratio of relative intensity of m/z 306 to m/z 324 in the two forms of mass spectra of cyclic CMP, indicatmg that no additional hydrolysis of the cyclic nucleotide had taken place during the passage of sample from the syringe pump to the continuous FAB probe tip. A minor difference, consistently observed, was that the formation of sodium adducts of both sample and matrix peaks was reduced in the dynamic FAB system. Both these features are of significance to the potential application of CF-FAB to cyclic nucleotide-related enzyme studies. To simulate enzyme incubation sam-

DISCUSSION

While the nominal increase in sensitivity m continuous FAB relative to static FAB seems very impressive, it is derived from the concentration of sample required to produce a particular level of ion current. In practice the total amount of sample required is approximately the same for both systems since a larger volume of sample solution is required for the dynamic system, and thus the limits of detection for static FAB-MS-MS of cyclic CMP of 180 ng for MIKE spectra and 23 ng for mass spectra, previously obtained under optimal conditions and at an S/N ratio of 3 : 1 [12], will not be appreciably decreased by the implementation of a continuous system. Nevertheless the improved S/N ratios obtained and the absence of any increased hydrolysis of cyclic CMP during sample transit from pump to probe tip confirm the viability of the system, and the effective sensitivity of the system is fully compatible with mea-

lhermocouple

CONTINUOUS

quark

FAB

column

shield

thermocouple

REACTION Fig 6. Reactlon

malrlx

CELL

cell and dynarmc

FAB system

deslgned

ET AL

for continuous

lnpul

momtonng

of enzyme

reactlon

rate (not to scale)

FAB-MS

OF CYCLIC

185

NUCLEOTIDES

surement of the activity of the cyclic nucleotide-re lated enzyme preparations which typically have activities in the nmol to pmol min-’ range [22]. Our current efforts centre on a reaction cell of the type represented in fig. 6, with a thermostatically controlled, stirred cell having inlet and outlet systems both regulated by pumps and the outlet from the cell being mixed with water-methanol-glycerol prior to passage to the probe. In addition to problems relating to the need for adequate mixing, stirring, balanced flow-rates etc., a range of difficulties arising from the enzyme properties must also be addressed in the development of this system. For example as enzymes are proteins they may create problems due to denaturation during the process and, at least in the case of particulate enzymes, it may be necessary to selectively remove the protein from the sample solution after removal from the reaction cell but pnor to pumping through the capillary to the probe tip. Buffers will have to be selected which are effective in the enzyme incubation medium at low concentration and that will not hamper spectrometric analysis by yielding fragments which reduce the usefulness of the spectra. A further major problem is the fact that a large number of enzymes have an absolute requirement for cations as cofactors, cations which can produce adducts with sample or matrix derived peaks and thereby impair spectral interpretation. Unless cations can be selected which do not form such adducts but which still activate the enzyme under examination it will be necessary to devise a means of selective cation removal. Nevertheless the potential advantages of a continuous assay system fully justify the efforts necessary to overcome these obstacles, since a continuous assay based upon MS analysis will enable the monitoring of several enzyme incubation components simultaneously thereby providing valuable evidence relating to enzyme mechanism of action. This is not currently possible with conventional assays of cyclic nucleotide-related enzymes, nor indeed with many other enzymes, thus any such system successfully developed will have a very wide applicability. The authors gratefully acknowledge financial support from SERC and the Wellcome Trust.

REFERENCES

1 R.H. Herman and 0 D. Taunton, m R.H Herman, R.M. Cohn and P.D McNamara (Eds ), Prmciples of Metabohc Control m Mammahan Systems, Plenum Press, New York, 1980, p. 424. 2 R.P. Newton, A.G. Brenton, D. Ghosh, T.J Walton, J I. Langndge, F M Hams and A.M. Evans, Anal. Chtm. Acta, in press 3 J Shirhattt, E. Sokoloskt, S Eng, S. Heuch, E. Rrccardi and G. Krtshna, J Cychc Nucleotrde Protem Phosphorylatron Res., 11 (1906) 463. 4 A.M. Lawson, RN Stillwell, M.M. Tacker, K Tsuboyama and J.A. McCloskey, J. Am. Chem. Sot , 93 (1971) 1014. 5 M. Barber, R.S. Bordoh, R.D. Sedgewtck and A.N Tyler, J. Chem. Sot. Chem. Commun , (1981) 325 6 M. Barber, R.S. Bordoh, R.D Sedgewtck, A.N. Tyler, U.N. Green, V.C Parr and J.L Gower, Boomed. Mass Spectrom., 9 (1982) 11. 7 M. Barber, R.S Bordoli, R.D. Sedgewtck and E.T Whalley, Boomed. Mass. Spectrom., 8 (1981) 337. 8 W Aberth, K M Straub and A.L. Burhngame, Anal. Chem., 54 (1982) 2029. 9 F.W Crow, K.B. Tomer, M L. Gross, J.A McCloskey and D.E. Bergstrom, Anal. Bmchem., 139 (1984) 243. 10 D.L. Slowikowskr and KY. Schram, Nucleostdes Nucleotides, 4 (1985) 309 11 E.E. Kmgston, J.H. Beynon and R.P. Newton, Boomed Mass Spectrom., 11 (1984) 367. 12 E.E. Kmgston, J.H. Beynon, R.P. Newton and J.G. Ltehr, Boomed Mass Spectrom., 12 (1985) 525 13 R.P. Newton, S.G Sahh, B.J Salvage and E.E. Kmgston, Bmchem. J., 221 (1984) 665. 14 R.P. Newton, E.E. Kmgston, N.A. Hakeem, S.G. Sahh, J.H. Beynon and C.D. Moyse, Btochem J., 236 (1986) 431. 15 R.P Newton, D. Chiatante, D Ghosh, A.G. Brenton, T.J. Walton, F M. Hams and E.D Brown, Phytochemtstry, 28 (1989) 2243. 16 R.P. Newton, N.A. Hakeem, B.J. Salvage, G Wassenaar and E E. Kmgston, Rapid Commun Mass Spectrom, 2 (1988) 118. 17 R.P. Newton, T.J. Walton, S.A. Basarf, A.M. Jenkms, A G. Brenton, D. Ghosh and F.M. Hams, Org. Mass Spectrom , 24 (1989) 679 18 R.P Newton, T.J Walton, A.G. Brenton, E.E. Kmgston and F.M. Harm, Rapid Commun. Mass Spectrom., 3 (1989) 178. 19 Y. Ito, T. Takeuchr, D. Ishr and M. Goto, J Chromatogr., 346 (1985) 161 20 R.M. Capnoh and T. Fan, Bmchem. Btophys. Res. Commun., 141 (1986) 1058. 21 R.M Capnoh, Mass Spectrom. Rev, 6 (1987) 237 22 R.P. Newton and S.G. Sahh, Int. J Bmchem., 18 (1986) 743 23 Dynamrc FAB Accessory Operation, V G. Analytical, Wythershaw, 1988, Sect. 6.