Electrospray Mass Spectrometry of Recombinant Growth Hormones Lubomir Baczynskyj* and George E. Bronson Physical and Analytical Chemistry Research, The Upjohn Company, Kalamazoo, MI 49001, USA SPONSOR REFEREE: Dr D . H. Williams, Department of Chemistry, Cambridge University, Cambridge, UK
The electrospray mass spectra of several standard proteins were recorded and their molecular weights determined. These were compared to Literature values obtained by other laboratories. The agreement was quite good. Seven recombinant growth hormones were then investigated by this technique. The determined molecular weights were in agreement with the theoretical values. Differentiation between six of the seven analogs could be made on the basis of the molecular weight determinations. Two of these analogs differed in molecular weight by only 1 Da and the accuracy of the mass measurements was not sufficient for distinguishingthese two homologous proteins. The influence of solvent on the mass accuracy determinationswas studied. It appears that solvents form clusters with the proteins in the electrospray ionization method. These clusters broaden the peaks corresponding to the multiply charged ions of the proteins and make the detemination of centroids more difficult.
The recent application of electrospray (ES)' or ionspray' ionization techniques to large molecules such as proteins and fragments of DNA has generated great interest amongst mass spectrometrists. Many papers have appeared in a very short time and review booklets and articles have been published by mass spectrometry manufacturers. Recently we have acquired an electrospray ion source from the Vestec Corp. (Houston, TX, USA) for our model 201A combined liquid chromatography/mass spectrometry (LC/MS) instrument. The range of this instrument has been increased to mlz 2000 and the sensitivity of the instrument has also been increased by Vestec. We have applied this new instrumentation to the analysis of some proteins of interest. First we checked the system on some proteins reported in the Literature. The ES mass spectrum of horse-heart myoglobin is shown in Fig. 1. As can be seen, this spectrum shows multiply charged ions from 9' charges to 21+ charges. The center-of-mass of the envelope is around 16+ charges. This pattern is slightly different from the ones reported in the literture and from the ES mass spectrum of myoglobin published by the Vestec
Corp. However the determination of the average molecular weight was quite good (16 949.8 4 1.1 found; 16951.5 theory). The ES mass spectra and average molecular weights of other proteins were recorded: insulin (bovine); lysozyme (hen egg); cytochrome C (horse heart); globins (bovine), (3-lactoglobulin A (bovine); ribonuclease A; bovine serum albumin. All gave reasonable ES mass spectra and good average molecular weight values (all better than 8.5 Da, with the exception of bovine serum albumin which yielded the same average molecular weight as that found in Ref. 3, i.e., 66 509.5 or 242.5 Da too heavy. This could be due to a modification of the protein and will be further investigated in the near future). Next we turned our attention to some of the proteins of interest to our research community. In particular we recorded the ES mass spectra and determined the molecular weights of some growth hormones prepared by recombinant techniques: recombinant bovine somatotropin (rbST), recombinant porcine somatotropin (rpST) and five analogs of rbST. In the process we have studied some of the factors that affected the accuracy of the mass measurements on our system. In particular we have studied the effects of solvents on the clusters of rbST. EXPERIMENTAL The samples of proteins were dissolved in a mixture of methanol water (1 : 1) containing 5 % acetic acid by volume. This solvent system was used in all experiments except where otherwise indicated. The solutions were introduced into the ion source by slow infusion using a syringe pump. The concentrations of the proteins were ca 100 ng/pL and the flow rate was ca 5 pL/ min. All ES mass spectra were recorded on the Vestec 201A LC/MS instrument using the Teknivent Vector 1 data system. In a typical run, several scans of 5 s each were recorded from mlz700 to 2000 and averaged to minimize the noise. The centroids of the various clusters were determined manually and the molecular weight calculations were performed as reported in Ref. 2.
+
0 . 700
..,,...,,. 800
900
. , , . . , . , . . , , , , . , , ,
1000
1100
1200
13W
.,., 1400
,
,
,
1500
., ..., ..,, 1600
1700
1800
,
,.,.. 1900
2000
m/2
Figure 1. Electrospray mass spectrum of horse-heart myoglobin (concentration 100 ng/pL, flow rate 5 pL/min).
* Author to whom correspondence should be addressed.
0951-4198/90/0533-0535$05.00
0
John Wiley & Sons Limited, 1990
RAPID COMMUNICATIONS IN MASS SPECTROMETRY. VOL. 4, NO. 12, 1990 533
ES MASS SPECTROMETRY OF RECOMBINANT GROWTH HORMONES
1200
7700
1300
1600
1500
1400
1700
1800
1900
2000
m/z
Figure2. Electrospray mass spectrum of recombinant bovine somatotropin (rbST) (concentration 100ng/pL in water with 5% acetic acid).
RESULTS AND DISCUSSION The electrospray mass spectrum of recombinant bovine somatotropin (rbST) recorded in methanol water (1: 1) with 5% acetic acid, is shown in Fig. 2. This spectrum is the result of averaging 10 scans and moving the baseline from 0 to 250 counts. As is apparent, clusters of 11+through 16' charges are observed. The cluster having 17' charges is barely visible. The measured average molecular weight of this protein was 21 816.1 k 1.6 and the theoretical average molecuar weight is 21 816.1. However during the measurements of the centroids of the various clusters it was clear that the peaks were not symmetrical and a large tail was present at the high mass side of the cluster, as seen in Fig. 3. This makes the determination of the centroids of such clusters difficult and introduces errors. It was interesting to investigate the origin of these tails and to see if corrections are necessary to the centroid determinations. This will be discussed below. Next the ES mass spectrum of the recombinant porcine somatotropin (rpST) was recorded and the average molecular weight determined. The observed average molecular weight was 21 801.3 k 1.6 and the theoretical average molecular weight is 21 801.9. Another series of clusters was present at the highermass side of the main clusters. The centroids of these
+
"i 6000
",'
. ..
were measured and yielded an average molecular weight of 21 897.1 f 1.4. The presence of this apparent impurity will be discussed below. Five homologs of rbST at position 99 were also investigated by ES mass spectrometry. In normal rbST, the amino acid Asn is at position 99. The five homologs had Gly, Pro, Ser, Glu and Asp at this position. The ES mass spectra for all five protein analogs looked very similar to that of rbST. The average molecular weights recorded for these proteins were: rbST-99-Gly, 21 759.0); rbST-99-Pro, 21 755.5 f 1.6 (theory 21 794.7 f 1.9 (theory 21 799.0); rbST-99-Ser, 21 782.4 +_ 1.9 (theory 21 789.0); rbST-99-Asp, 21 809.7 f2.9 (theory 21 817.0); and rbST-99-Glu, 21 826.6+ 1.7 (theory 21 831.0). From these results it is clear that we could not differentiate between the rbST and its homolog r b s T - 9 9 - A ~since ~ ~ they differ only by 1mass unit. But we could differentiate rbST from all its other analogs solely by this molecular weight determination. The largest error between the found and theory values (7.3 mass units) was for rbST-99-Glu. Note that the replacement of one amino acid by another requires greater mass measurement accuracy than the accuracy necessary to differentiate two proteins which differ by the addition or deletion of an amino acid residue, since the weight of the smallest amino acid residue (Gly, 57.0) is greater than the errors obtained on the above measurements. The values reported above were obtained by determining the centroids of the clusters manually and are therefore from a single measurement. The peaks were generally skewed and the determination of the centroids of each cluster was greatly influenced by the tuning of the instrument, the calibration and the judgement of the operator. The skewing of the peaks towards the high-mass end of the mass spectrum was of great concern and we suspected the solvent to be the main contributor. For this reason the ES mass spectrum of rbST was recorded in several solvents: water, methanol, ethanol, propano1 + water (1 :1). In each case 5% of acetic acid by volume was also present in the solvent. The concentration was also maintained at ca 100ng/yL in all cases. The high-mass end of the ES mass spectrum of rbST in ethanol + 5% acetic acid is shown in Fig. 4. Here we see that each cluster is a composite of four or five peaks. The three most abundant of them (Fig. 5) were mass
...
;.~,~, ._,,-, , , , , ,
1818
,
,
1815
m/z
Figure 3. Profile of 12+ cluster of recombinant bovine somatotropin (rbST) in water with 5% acetic acid.
Figure4. High-mass end of the electrospray mass spectrum of recombinant bovine somatotropin (rbST) run in ethanol with 5% acetic acid.
534 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4. NO. 12. 1990
@ John Wiley & Sons Limited, 1990
ES MASS SPECTROMETRY OF RECOMBINANT GROWrH HORMONES
should improve the determination of centroids and thus the accuracy of mass determinations.
m/z
Figure 5. Profile of 12' cluster of recombinant bovine somatotropin (rbST) in ethanol with 5% acetic acid.
measured and gave the values 21 814.3, 21 870.6 and 21 928.6. The differences between these masses (56.3 and 58.0) suggest that each hump is due to the addition of one, two and probably more molecules of ethanol. This of course holds for other solvents as well. In particular for the case of water + methanol (1:1) the high-mass end of the ES mass spectrum exhibits additions of both water and methanol for some clusters. This is less visible at the low-mass end of the spectrum but certainly has an appreciable influence on the accuracy of the centroid determinations and therefore the mass measurements. Certain instruments have already incorporated devices for declustering the compound peaks from the solvent clusters. This should make the mass measurements of the average molecular weight more accurate and make the appearance to the penks in the ES mass spectra narrower. In addition, better diitil processing software, such as peak deconvolution.
@ John Wiley & Sons Limited, 1990
CONCLUSIONS In this study we have shown that it is possible to determine the molecular weights of certain proteins on an instrument of modest resolution and price. The mass accuracy can be improved by better data acquisition and data processing of the clusters. Averaging and peak deconvolution should,improve the accuracy of the mass measurements. The choice of solvent can influence the shape of the clusters and the accuracy of the measurements. Better pumping of the regions where the clusters are formed should also improve the peak shapes and the accuracy of mass determination.
Acknowledgements We would like to thank Drs R. L. Garlick, J . G. Hoogerheide and S. B. Lyle of the Bioprocess Research unit for providing the samples used in this study.
REFERENCES I . J. B. Fenn, M. Mann, C . K. Meng, S . F. Wong and C. M. Whitehouse, Science 246, 64 (1989). 2. T. R. Covey, R. F. Bonner, B. I. Shushan and J . Henion, Rapid Commun. Mass Spectrom. 2, 249 (1988). 3 . S . K. Chowdhury, V. Katta and B. T. Chait, Rapid Commun. Mass Spectrom. 4, 81 (1990). Received 5 October 1990; accepted 6 October 1990.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4.NO. 12. 1990 535
The electrospray mass spectra of several standard proteins were recorded and their molecular weights determined. These were compared to Literature values obtained by other laboratories. The agreement was quite good. Seven recombinant growth hormones were then investigated by this technique. The determined molecular weights were in agreement with the theoretical values. Differentiation between six of the seven analogs could be made on the basis of the molecular weight determinations. Two of these analogs differed in molecular weight by only 1 Da and the accuracy of the mass measurements was not sufficient for distinguishingthese two homologous proteins. The influence of solvent on the mass accuracy determinationswas studied. It appears that solvents form clusters with the proteins in the electrospray ionization method. These clusters broaden the peaks corresponding to the multiply charged ions of the proteins and make the detemination of centroids more difficult.
The recent application of electrospray (ES)' or ionspray' ionization techniques to large molecules such as proteins and fragments of DNA has generated great interest amongst mass spectrometrists. Many papers have appeared in a very short time and review booklets and articles have been published by mass spectrometry manufacturers. Recently we have acquired an electrospray ion source from the Vestec Corp. (Houston, TX, USA) for our model 201A combined liquid chromatography/mass spectrometry (LC/MS) instrument. The range of this instrument has been increased to mlz 2000 and the sensitivity of the instrument has also been increased by Vestec. We have applied this new instrumentation to the analysis of some proteins of interest. First we checked the system on some proteins reported in the Literature. The ES mass spectrum of horse-heart myoglobin is shown in Fig. 1. As can be seen, this spectrum shows multiply charged ions from 9' charges to 21+ charges. The center-of-mass of the envelope is around 16+ charges. This pattern is slightly different from the ones reported in the literture and from the ES mass spectrum of myoglobin published by the Vestec
Corp. However the determination of the average molecular weight was quite good (16 949.8 4 1.1 found; 16951.5 theory). The ES mass spectra and average molecular weights of other proteins were recorded: insulin (bovine); lysozyme (hen egg); cytochrome C (horse heart); globins (bovine), (3-lactoglobulin A (bovine); ribonuclease A; bovine serum albumin. All gave reasonable ES mass spectra and good average molecular weight values (all better than 8.5 Da, with the exception of bovine serum albumin which yielded the same average molecular weight as that found in Ref. 3, i.e., 66 509.5 or 242.5 Da too heavy. This could be due to a modification of the protein and will be further investigated in the near future). Next we turned our attention to some of the proteins of interest to our research community. In particular we recorded the ES mass spectra and determined the molecular weights of some growth hormones prepared by recombinant techniques: recombinant bovine somatotropin (rbST), recombinant porcine somatotropin (rpST) and five analogs of rbST. In the process we have studied some of the factors that affected the accuracy of the mass measurements on our system. In particular we have studied the effects of solvents on the clusters of rbST. EXPERIMENTAL The samples of proteins were dissolved in a mixture of methanol water (1 : 1) containing 5 % acetic acid by volume. This solvent system was used in all experiments except where otherwise indicated. The solutions were introduced into the ion source by slow infusion using a syringe pump. The concentrations of the proteins were ca 100 ng/pL and the flow rate was ca 5 pL/ min. All ES mass spectra were recorded on the Vestec 201A LC/MS instrument using the Teknivent Vector 1 data system. In a typical run, several scans of 5 s each were recorded from mlz700 to 2000 and averaged to minimize the noise. The centroids of the various clusters were determined manually and the molecular weight calculations were performed as reported in Ref. 2.
+
0 . 700
..,,...,,. 800
900
. , , . . , . , . . , , , , . , , ,
1000
1100
1200
13W
.,., 1400
,
,
,
1500
., ..., ..,, 1600
1700
1800
,
,.,.. 1900
2000
m/2
Figure 1. Electrospray mass spectrum of horse-heart myoglobin (concentration 100 ng/pL, flow rate 5 pL/min).
* Author to whom correspondence should be addressed.
0951-4198/90/0533-0535$05.00
0
John Wiley & Sons Limited, 1990
RAPID COMMUNICATIONS IN MASS SPECTROMETRY. VOL. 4, NO. 12, 1990 533
ES MASS SPECTROMETRY OF RECOMBINANT GROWTH HORMONES
1200
7700
1300
1600
1500
1400
1700
1800
1900
2000
m/z
Figure2. Electrospray mass spectrum of recombinant bovine somatotropin (rbST) (concentration 100ng/pL in water with 5% acetic acid).
RESULTS AND DISCUSSION The electrospray mass spectrum of recombinant bovine somatotropin (rbST) recorded in methanol water (1: 1) with 5% acetic acid, is shown in Fig. 2. This spectrum is the result of averaging 10 scans and moving the baseline from 0 to 250 counts. As is apparent, clusters of 11+through 16' charges are observed. The cluster having 17' charges is barely visible. The measured average molecular weight of this protein was 21 816.1 k 1.6 and the theoretical average molecuar weight is 21 816.1. However during the measurements of the centroids of the various clusters it was clear that the peaks were not symmetrical and a large tail was present at the high mass side of the cluster, as seen in Fig. 3. This makes the determination of the centroids of such clusters difficult and introduces errors. It was interesting to investigate the origin of these tails and to see if corrections are necessary to the centroid determinations. This will be discussed below. Next the ES mass spectrum of the recombinant porcine somatotropin (rpST) was recorded and the average molecular weight determined. The observed average molecular weight was 21 801.3 k 1.6 and the theoretical average molecular weight is 21 801.9. Another series of clusters was present at the highermass side of the main clusters. The centroids of these
+
"i 6000
",'
. ..
were measured and yielded an average molecular weight of 21 897.1 f 1.4. The presence of this apparent impurity will be discussed below. Five homologs of rbST at position 99 were also investigated by ES mass spectrometry. In normal rbST, the amino acid Asn is at position 99. The five homologs had Gly, Pro, Ser, Glu and Asp at this position. The ES mass spectra for all five protein analogs looked very similar to that of rbST. The average molecular weights recorded for these proteins were: rbST-99-Gly, 21 759.0); rbST-99-Pro, 21 755.5 f 1.6 (theory 21 794.7 f 1.9 (theory 21 799.0); rbST-99-Ser, 21 782.4 +_ 1.9 (theory 21 789.0); rbST-99-Asp, 21 809.7 f2.9 (theory 21 817.0); and rbST-99-Glu, 21 826.6+ 1.7 (theory 21 831.0). From these results it is clear that we could not differentiate between the rbST and its homolog r b s T - 9 9 - A ~since ~ ~ they differ only by 1mass unit. But we could differentiate rbST from all its other analogs solely by this molecular weight determination. The largest error between the found and theory values (7.3 mass units) was for rbST-99-Glu. Note that the replacement of one amino acid by another requires greater mass measurement accuracy than the accuracy necessary to differentiate two proteins which differ by the addition or deletion of an amino acid residue, since the weight of the smallest amino acid residue (Gly, 57.0) is greater than the errors obtained on the above measurements. The values reported above were obtained by determining the centroids of the clusters manually and are therefore from a single measurement. The peaks were generally skewed and the determination of the centroids of each cluster was greatly influenced by the tuning of the instrument, the calibration and the judgement of the operator. The skewing of the peaks towards the high-mass end of the mass spectrum was of great concern and we suspected the solvent to be the main contributor. For this reason the ES mass spectrum of rbST was recorded in several solvents: water, methanol, ethanol, propano1 + water (1 :1). In each case 5% of acetic acid by volume was also present in the solvent. The concentration was also maintained at ca 100ng/yL in all cases. The high-mass end of the ES mass spectrum of rbST in ethanol + 5% acetic acid is shown in Fig. 4. Here we see that each cluster is a composite of four or five peaks. The three most abundant of them (Fig. 5) were mass
...
;.~,~, ._,,-, , , , , ,
1818
,
,
1815
m/z
Figure 3. Profile of 12+ cluster of recombinant bovine somatotropin (rbST) in water with 5% acetic acid.
Figure4. High-mass end of the electrospray mass spectrum of recombinant bovine somatotropin (rbST) run in ethanol with 5% acetic acid.
534 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4. NO. 12. 1990
@ John Wiley & Sons Limited, 1990
ES MASS SPECTROMETRY OF RECOMBINANT GROWrH HORMONES
should improve the determination of centroids and thus the accuracy of mass determinations.
m/z
Figure 5. Profile of 12' cluster of recombinant bovine somatotropin (rbST) in ethanol with 5% acetic acid.
measured and gave the values 21 814.3, 21 870.6 and 21 928.6. The differences between these masses (56.3 and 58.0) suggest that each hump is due to the addition of one, two and probably more molecules of ethanol. This of course holds for other solvents as well. In particular for the case of water + methanol (1:1) the high-mass end of the ES mass spectrum exhibits additions of both water and methanol for some clusters. This is less visible at the low-mass end of the spectrum but certainly has an appreciable influence on the accuracy of the centroid determinations and therefore the mass measurements. Certain instruments have already incorporated devices for declustering the compound peaks from the solvent clusters. This should make the mass measurements of the average molecular weight more accurate and make the appearance to the penks in the ES mass spectra narrower. In addition, better diitil processing software, such as peak deconvolution.
@ John Wiley & Sons Limited, 1990
CONCLUSIONS In this study we have shown that it is possible to determine the molecular weights of certain proteins on an instrument of modest resolution and price. The mass accuracy can be improved by better data acquisition and data processing of the clusters. Averaging and peak deconvolution should,improve the accuracy of the mass measurements. The choice of solvent can influence the shape of the clusters and the accuracy of the measurements. Better pumping of the regions where the clusters are formed should also improve the peak shapes and the accuracy of mass determination.
Acknowledgements We would like to thank Drs R. L. Garlick, J . G. Hoogerheide and S. B. Lyle of the Bioprocess Research unit for providing the samples used in this study.
REFERENCES I . J. B. Fenn, M. Mann, C . K. Meng, S . F. Wong and C. M. Whitehouse, Science 246, 64 (1989). 2. T. R. Covey, R. F. Bonner, B. I. Shushan and J . Henion, Rapid Commun. Mass Spectrom. 2, 249 (1988). 3 . S . K. Chowdhury, V. Katta and B. T. Chait, Rapid Commun. Mass Spectrom. 4, 81 (1990). Received 5 October 1990; accepted 6 October 1990.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4.NO. 12. 1990 535
