Automated High Resolution MassSpectrometry for the Synthetic Chemist

George Perkins, Frank Pullen, and Christine ThompsonPhysical Sciences, Pfizer Central Research, Sandwich, Kent, UK

Exact mass measurement at high resolution is an important tool alongside other spectroscopicmethods to help confirm the structure of a novel compound prepared by the synthetic chemist.Exact mass measurement is used in the pharmaceutical industry to confirm the expectedempirical formula of a product when problems have been experienced using elementalanalysis. Because of the amount of manual intervention necessary when acquiring exact massmeasurements, especially when using probe ionization techniques such as fast atom bombard-ment ionization or electron ionization, this method has been seen to be time consuming andlabor intensive for the mass spectrometrist. An automated high resolution mass spectrometricmethod has been developed at Pfizer Central Research which has streamlined exact massmeasurement. The method, which uses electrospray ionization on a double focusing massspectrometer, is described. The samples are analyzed using a flow injection technique, withsodiated polyethylene glycol present in the mobile phase to provide mass reference peaks. Thedata are acquired and processed using a macro developed “in house.” This automatedtechnique can process 15–20 samples an hour including data processing and report generation,using very small amounts of compound (;25 mg), but more importantly it can be left to rununattended overnight. This allows the instrument to be used for more complex experimentsduring the day when it is important to have a mass spectrometrist present. The resultspresented here demonstrate that this method gives exact mass measurements within anacceptable limit of 5 ppm, and the variation on one sample, injected 10 times, is not excessivelyhigh (21.8 to 11.6 mDa). (J Am Soc Mass Spectrom 1999, 10, 546–551) © 1999 AmericanSociety for Mass Spectrometry

Mass spectrometry and nuclear magnetic reso-nance (NMR) are the major spectroscopictechniques used by synthetic chemists in the

pharmaceutical industry to monitor key stages in asynthetic pathway and to characterize the final product.The use of open access NMR and mass spectrometryinstruments is now widespread and this approach tospectroscopic analysis has greatly increased the amountof structural information readily available to the syn-thetic chemist, while significantly reducing the timetaken to obtain these data [1, 2].

Exact mass measurement [3] at high resolution is animportant tool alongside these other spectroscopicmethods to help confirm the structure of novel com-pounds prepared by the synthetic chemist [4]. It hasbeen extensively used since the advent of high resolu-tion mass spectrometers to determine empirical formu-las of ions from a number of ionization techniques suchas electron ionization, chemical ionization, thermosprayionization, fast atom bombardment (FAB) ionization,and more recently atmospheric pressure chemical ion-ization and electrospray [5–7].

Elemental analysis has historically been used by thesynthetic chemist to confirm that the product has theexpected empirical formula. Elemental analysis cangive results that are difficult to match to the expectedempirical formula. This can be due to impurities beingpresent in the sample such as solvents, salts, or com-pound-related impurities. Another possible source oferror is the inaccuracy of weighing samples, especiallyif the sample is a gum or a hygroscopic compound. Afurther disadvantage of elemental analysis is the rela-tively large amount of sample needed (;2 mg).

The use of high resolution mass spectrometry hastraditionally overcome these problems, but the method,especially using FAB, has often been labor intensive,needing an expert mass spectrometry instrument oper-ator, and time consuming, as all the data acquisition,processing, and interpretation had to be carried out bythe operator. There was a clear need to automate exactmass measurement which would provide a signifi-cantly higher throughput by minimizing the manualinput for this important technique [8, 9].

An automated high resolution mass spectrometricmethod has been developed at Pfizer Central Researchwhich has streamlined exact mass measurement. Themethod, which uses electrospray ionization [10, 11] on adouble focusing mass spectrometers [12, 13] will be

Address reprint requests to George Perkins, Physical Sciences, PfizerCentral Research, Ramsgate Road, Sandwich, Kent CT13 9NJ, UK. E-mail:George_Perkins@sandwich.pfizer.com

© 1999 American Society for Mass Spectrometry. Published by Elsevier Science Inc. Received June 22, 19981044-0305/99/$20.00 Revised January 22, 1999PII S1044-0305(99)00014-8 Accepted February 8, 1999

described. The samples are analyzed in a batch modeusing a flow injection technique, and the data areprocessed using macros developed “in house.” Thisautomated technique can process 15–20 samples anhour, using very small amounts of compound, but moreimportantly, it can be left to run unattended overnight.This allows the instrument to be used for more complexexperiments such as exact mass electron ionization orhigh resolution collision-induced dissociation MS-MSduring the day when it is important to have a massspectrometrist present.

Experimental

Sample Preparation

All chemicals, except test compounds, were purchasedfrom Aldrich Sigma (Gillingham, Dorset, UK). Thechemist’s sample is dissolved in a solution of methanolcontaining 1% v/v acetic acid and 0.005% w/w poly-ethylene glycols 200 and 600. The solution concentra-tion should ideally be between 0.5 and 1 mg/mL. Afinal liquid sample of 50 mL volume is the minimumvolume, which means the minimum sample consump-tion per analysis is 25 mg.

Mobile Phase Conditions

A Gilson 305 HPLC pump and a Gilson 306 HPLCpump fitted with 5 mL pump heads (Anachem, Luton,Beds, UK) are connected to a Gilson 231 autosamplerwhich has an integral Rheodyne injector valve 7125with a 20 mL loop. The Gilson 305 pump provides themobile phase [methanol/water/acetic acid 80/20/1v/v containing 0.005% w/w polethylene glycol (PEG)200, 0.005% w/w PEG 600 and 0.005% sodium acetate]at 30 mL/min flow rate to the electrospray source of themass spectrometer. The Gilson 306 pump providesmethanol at 100 mL/min as a wash solution at the endof the batch. Control of the HPLC pumps and a startsignal for the mass spectrometer is provided by theGilson autosampler.

Mass Spectrometry

Analyses are performed on a Micromass Autospec-Qdouble focusing magnetic sector mass spectrometerequipped with an electrospray interface (Micromass,Manchester, UK), controlled by a Digital AlphaStation255/233 2D CAD (CSF Systems, London, UK) usingOPUS software and the OPAL macro language. Allsamples are analyzed in the positive ion mode usingvoltage scans and the data collected in continuummode. The mass range scanned is determined by themacro and is related to the mass of the compound to beanalyzed, the selected range is scanned in 5 s with aninterscan delay of 1 s. The mass spectrometer is run atan accelerating voltage of 4 kV and the source heatercurrent is set to 1.2 A. The resolution is set to approxi-

mately 10,000 at 5% valley at a mass of 525. The bath gasand nebulizer gas is dry nitrogen at flow rates of 400and 12 L/h, respectively.

Sample Acquisition

The operator is asked for the number of samples in thebatch (batch size 1 to 80) and then for each sample itsnominal mass, empirical formula, and sample identifi-cation information. The macro uses this information toconstruct a voltage scan for each sample which willencompass the MH1 and MNa1 for the sample and therelevant lower and upper mass reference peaks. Onceall the information for a batch of samples has beenentered, the macro instructs the mass spectrometer tostart acquiring data on receipt of a signal from theautosampler.

The sample is introduced by the autosampler intothe mobile phase at the same time a start signal is sentby the autosampler to the mass spectrometer and ac-quisition commences using the predetermined voltagescan range automatically set during the log-in phase ofthe process. At the end of the first sample analysis thespectrometer waits for the next signal from the au-tosampler to begin collecting data for the next sample.At the end of the batch the macro checks that theacquisition data for all the samples have been obtainedand moves into the processing section of the macro.

Sample Processing

After acquisition the macro examines each sample’sdataset in turn. First, a total ion current chromatogramis created (see Figure 1). The macro determines the peakapex and a spectrum is created by averaging the fourscans on either side of the apex. As the analysis iscarried out by loop injection then the peak apex occursat approximately the same time point for each injection,and so if a sample does not produce a peak the macrotakes a preset time point for the peak apex and averagesthe four scans on either side of that time point to createthe spectrum. The macro converts the continuum dataspectrum to a centroid data spectrum, using a saved setof parameters for smoothing, peak detection, etc.; itthen mass assigns the spectrum, before printing (seeFigure 2).

The m/z values for the MH1 and MNa1 ions aretransferred individually by the macro to the elementalfit program. The elemental fit program is setup by themacro to allow a maximum error of 5 ppm and to havea predetermined maximum and minimum number ofatoms from the formula supplied. The elemental fitprogram calculates the possible matches, and the errorsassociated with these for the observed mass, and printsthis report. Both the data for the MH1 and MNa1 ionsare sent to the synthetic chemist.

547J Am Soc Mass Spectrom 1999, 10, 546–551 AUTOMATED HIGH RESOLUTION MASS SPECTROMETRY

Results and Discussion

As an example, five different compounds, which aretypical of the structures being synthesized within thepharmaceutical industry, were analyzed by thismethod. These compounds produced both protonatedand sodium cationized molecules which were massmeasured (see Table 1). For the compoundC20H25ClN2O5, 10 replicate injections were carried outin 45 min (see Table 2). The mean result for the MH1 ionis 409.1528 Da and for the MNa1 ion is 431.1345 Da,giving average errors of 20.2 and 20.5 mDa, respec-tively, from the calculated mass. The range of error

observed between the 10 injections was 21.8 to 11.6mDa for the MH1 and 20.6 to 10.9 mDa for the MNa1.This range of error compared favorably to the errorobserved when carrying out these replicate injectionsmanually.

Electrospray ionization was chosen as it is the mostuniversal ionization technique for the vast majority ofcompounds synthesized at Pfizer Central ResearchSandwich. Sodium acetate is included in the mobilephase to promote the formation of the sodiated ion PEGreference peaks. We have found that when the mobilephase is prepared without sodium acetate, a mixture ofthe protonated, ammoniated, and sodiated PEG refer-

Figure 1. Total ion chromatogram of a sample showing peak apex.

548 PERKINS ET AL. J Am Soc Mass Spectrom 1999, 10, 546–551

ence peaks are present in the spectrum. This makesmass calibration more difficult as the intensity of thesepeaks varies and sometimes one of the three expectedions for the reference compound is not present.

One potential problem that can occur with the sys-tem is when a sample compound yields an ion which isisobaric with one of the PEG reference peaks. This hasproved not to be as much of a problem as initiallythought. The fact that sodium acetate is present in themobile phase but not in the sample, usually assures thatboth protonated and sodiated ions of the sample arepresent and a match can usually be found for thenonisobaric ion. The error on this match is usuallyhigher than expected as one of the three reference peakmass is incorrectly assigned because of the presence ofthe isobaric peak. Another possible source of errorderives from an inappropriate ratio of the reference andsample ion intensities. For the best elemental fit theseions should be of similar abundance, but this is difficultto obtain, as samples have a large variation in theirrelative proton or sodium affinities. This does make adifference to the errors recorded for a sample, butresults to date suggest the errors are still within the 5ppm limit. Another perceived problem was the lifetimeof the source and therefore the amount of down time.During the development and testing of this system themobile phase was running for 8 h a day and thesensitivity of the source was not substantially impairedover a two week period. This is helped by the secondGilson HPLC pump which is used to pump methanol at

a flow rate of 100 mL/min for 30 min at the end of eachbatch/day. This is triggered by the autosampler so nomanual interaction is necessary.

The results shown in Table 1 all meet the 5 ppm errorlimit (observed range for the data presented is 0.2–4.8ppm) and this has been observed across the mass range.Compounds with molecular weights as low as 250 Daand as high as 800 Da have been analyzed on thissystem and the results show that this method is appli-cable across this mass range. This is the mass range thathas been shown to be optimal for therapeutic com-pounds which have good absorption or permeationwhen administered orally and fit the “rule of 5” criteriawhich is a method in common use within the pharma-ceutical industry [14]. The reproducibility of the injec-tions is within expected limits. This gives us confidencein this technique’s ability to cope with “real” sampleswhich may contain impurities.

Conclusions

Exact mass measurement is an important technique inidentifying the elemental formula for ions of interest ina mass spectrum. Soft ionization promotes the forma-tion of protonated and sodium catonized molecules. Byusing exact mass measurement, the elemental formulaof the compound(s) can be determined with a highdegree of confidence. Historically this method has beentime consuming as each sample had to be introducedmanually into the mass spectrometer. Using electros-

Figure 2. Typical spectrum showing MH1 and MNa1 ions for the sample and MNa1 ions for the tworeference peaks.

549J Am Soc Mass Spectrom 1999, 10, 546–551 AUTOMATED HIGH RESOLUTION MASS SPECTROMETRY

pray ionization at high resolution with the referencecompound being infused constantly into the source ofthe mass spectrometer, means that samples can beanalyzed faster. Macros are in place to automaticallyacquire and process the data. We have demonstratedthat there is no compromise in the data quality recordedby this method in comparison to the old manual sampleintroduction technique. It can be seen from the results

that the technique is reproducible with a standarddeviation of 0.94 mDa for the MH1 and 0.86 mDa forthe MNa1 for 10 replicate injections. This approach toexact mass measurement provides a significantly higherthroughput than the conventional approach and isroutinely operated in an unattended mode overnight. Ithas proved to be a robust and routine system and hashad a major impact on the sample throughput in a busypharmaceutical chemistry environment.

AcknowledgmentWe gratefully acknowledge Jerry Hart of Micromass Ltd. for hisadvice and encouragement on the development of the macrodescribed in this article.

References1. Bowen, D. V.; Dalton, M.; Perkins, G. L.; Pullen, F. S.; Richards,

D. An automated, open-access service to synthetic chemists.Rapid Commun. Mass Spectrom. 1994, 8, 632–636.

2. Burton, K. I.; Kiplinger, J. P.; Perkins, G. L.; Pullen, F. S.;Teague, M. S.; Ware, R. S. Putting mass spectrometry in the

Table 1. Selected compounds showing the comparison between their calculated and experimentally observed masses

Table 2. Ten replicate injections of a compound showingexperimental variation in the observed mass

Injection numberObserved mass

for MH1

Observed massfor MNa1

1 409.1510 431.13542 409.1522 431.13393 409.1517 431.13434 409.1544 431.13415 409.1535 431.13476 409.1534 431.13547 409.1531 431.13448 409.1534 431.13419 409.1528 431.1347

10 409.1527 431.1343

550 PERKINS ET AL. J Am Soc Mass Spectrom 1999, 10, 546–551

hands of the end user. J. Am. Soc. Mass Spectrom. 1995, 6,394–399.

3. McLafferty, F. W.; Venkataraghavan, R. Mass Spectral Cor-relations; American Chemical Society: Washington, D.C.,1982.

4. Roboz, J.; Holland, J. F.; McDowall, M. A.; Hillmer, M. J.Accurate mass measurement in continuous flow fast atombombardment quadrupole mass spectrometry. Rapid Commun.Mass Spectrom. 1988, 2, 64–66.

5. Biochemical Applications of Mass Spectrometry; Waller, G. R., Ed.;Wiley–Interscience: New York, 1972.

6. D’Agostino, P. A.; Hancock, J. R.; Provost, L. R.; Semchuk,P. D.; Hodges, R. S. High resolution electrospray mass spec-trometry with a magnetic sector instrument: accurate massmeasurement and peptide sequencing. Rapid Commun. MassSpectrom. 1995, 9, 597–603.

7. Chapman, J. R. Practical Organic Mass Spectrometry; Wiley;New York, 1986.

8. Klohr, S. E.; Whitney, J.; Rourick, R. A. An Analytical Strategyfor Rapid Molecular Formula Determination Using Auto-mated High Resolution Electrospray Ionisation Mass Spec-trometry. Proceedings of the 44th Annual ASMS Conference on

Mass Spectrometry and Allied Topics; Portland OR, May 12–16,1996; p 656.

9. Perkins, G. L.; Pullen, F. S.; Thompson, C. M. Semi-AutomatedAccurate Mass. Proceedings of the 45th Annual ASMS Conferenceon Mass Spectrometry and Allied Topics; Palm Springs California,June 1–5, 1997; p 962.

10. Mann, M.; Meng, C.-K.; Fenn, J. B. Interpreting mass spectra ofmultiply charged ions. Anal. Chem. 1989, 61, 1702–1708.

11. Fenn, J. B.; Mann, M.; Meng, C.-K.; Wong, S. F.; Whitehouse,C. M. Electrospray ionization for mass spectrometry of largebiomolecules. Science 1989, 246, 64.

12. Larsen, B. S.; McEwen, C. N. An electrospray ion source formagnetic sector mass spectrometers. J. Am. Soc. Mass Spectrom.1991, 2, 205–211.

13. Moore, C. Accurate mass ESI on a magnetic sector massspectrometer. Proceedings of the 44th Annual ASMS Conferenceon Mass Spectrometry and Allied Topics; Portland, OR, May12–16, 1996; p 657.

14. Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J.Experimental and computational approaches to estimate sol-ubility and permeability in drug discovery and developmentsettings. Adv. Drug Delivery Rev. 1997, 23, 3–25.

551J Am Soc Mass Spectrom 1999, 10, 546–551 AUTOMATED HIGH RESOLUTION MASS SPECTROMETRY