Anal. Chem. 1998, 70, 2704-2709
Monitoring Protein Expression in Whole Bacterial Cells with MALDI Time-of-Flight Mass Spectrometry Michael L. Easterling, Christopher M. Colangelo, Robert A. Scott, and I. Jonathan Amster*
Department of Chemistry, University of Georgia, Athens, Georgia 30602-2556
We report the application of matrix-assisted laser desorption ionization (MALDI) to monitor recombinant protein expression in whole bacteria. This technique is characterized by rapid sample preparation that provides analysis of samples extracted directly from the growth media in less than 10 min. The mass spectrometric method holds several advantages over gel electrophoresis, the conventional method for examining the protein content of cells. Comparisons between the two methods of analysis are presented in terms of increased speed, efficiency, resolution, and mass accuracy. Delayed extraction time-of-flight mass spectrometry identifies posttranslational modifications and other changes in the expected structure which are not recognized by gel electrophoresis. The utility of this method is demonstrated for proteins with molecular masses ranging from 5 to 50 kDa. Low molecular mass proteins ( 10 000). A 10 µL aliquot of the broth solution was mixed with an equal volume of a saturated solution of sinapinic acid matrix in a 50:50 (v:v) water/acetonitrile mixture. The samples were prepared by the method of Beavis in which 1-2 µL of the saturated matrix solution was applied to the target and allowed to dry.33 The dried crystals were smeared with a lintfree cloth. A 1-1.5 µL amount of the broth/matrix mixture was placed on the smeared crystals. After drying, the sample spot 2706 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
was rinsed repeatedly with pure water (>18 MΩ) to remove excess salts. This method appeared to be more efficient at producing ions from sonicated cells than the more widely used dried drop method,34 and was used exclusively for the sonicated samples. Recombinant proteins with molecular weights less than 10 000 were subjected to a second method of sample preparation, in which1 mL of broth was centrifuged and decanted to remove the excess liquid. A 10-15 µL amount of pure water or a 50:50 (v:v) water/acetonitrile mixture was added to the bacterial “pellet”. A 1-2 µL volume was pipetted into a saturated matrix solution. Sample spots were prepared as for the above procedure, but rinsing was not required. RESULTS AND DISCUSSION MALDI analysis of unpurified bacterial cultures provides strong signals for recombinantly expressed proteins. Figure 2 shows the MALDI-TOF mass spectrum of E. coli which have been transformed with the gene for the N-terminal domain of human transcription factor IIB (hTFIIB-NTD). Two hours after induction of the protein expression, 1 mL of the growth medium was extracted and centrifuged in an identical fashion to the method used for the collection of samples for SDS-PAGE analysis. The most intense peak in the linear mode mass spectrum at m/z 6493 corresponds to the desired protein, which is present as the most abundant species in the sample. In comparison, SDS-PAGE of the same sample produced a band that appears at an anomalously higher molecular weight, as can be seen in lane 7 of Figure 1a. A peak appearing 609 mass units lower than the base peak most likely indicates a product of posttranslational degradation rather than a metastable fragment, as the latter do not appear in the linear mode of analysis when using continuous extraction of ions from the source. It is interesting to note that the proteins of similar molecular weight native to E. coli do not appear in the mass spectrum. This is probably due to the low abundance of the naturally occurring proteins compared to the overexpressed (33) Xiang, F.; Beavis, R. C. Rapid Commun. Mass Spectrom. 1994, 8, 199-204. (34) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301.
Figure 3. MALDI-TOF mass spectrum of bacterial cells used for overexpression of PfTFB-NTD+Met (average mol wt ) 5635), obtained 4 h after induction. The resolution improvement provided by delayed extraction provides distinction of the expressed protein and the N terminal methionine cleavage product that occurred during expression.
target protein, as the expressed protein makes up 10-50% of the total protein produced by the bacteria.1 Although the simple linear mass spectrum adequately indicates the presence and relative abundance of the target protein, the higher mass resolution provided by delayed extraction of the ion source coupled with an ion reflector, can provide more detailed information.12,35,36 Figure 3 shows a MALDI mass spectrum of an E. coli sample containing a plasmid for overexpression of the N-terminal domain of transcription factor B (PfTFB-NTD),37,38 a protein responsible for transcription start-site selection and recruitment of RNA polymerase in Pyrococcus furiosus. The mass spectrum was acquired in the reflector mode with a delayed extraction pulse from the TOF source and exhibits much greater mass resolution than that of Figure 2, allowing the identification of two peaks closely spaced in mass. The target protein appears with significant intensity at m/z 5510, indicating that the overexpression was successful. The peak occurring 131 m/z units higher indicates a large relative concentration of methionine-attached protein (Met-Pf TFB-NTD), produced during the expression as a result of the transcription of the initiating AUG codon. The resolution for this mass range (m/∆m ≈ 550fwhh) provides unambiguous mass assignment exclusive of closely spaced adducts or losses, greatly enhancing the analytical capabilities of this method over the electrophoretic method. Figure 4 shows the MALDI-TOF mass spectrum of E. coli transformed with a plasmid containing the gene for P. furiosus rubredoxin (Pf Rd), acquired 3 h after induction in which three forms of the protein are observed in significant abundance. In addition to the target protein and its methionine variant, a formylated methionine (Metf-Pf Rd) product, common to prokaryotic expression, is observed in equal abundance to the product (35) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67, 1998-2003. (36) Vestal, M. L.; Juhasz, P.; Martin, S. A. Rapid Commun. Mass Spectrom. 1995, 9, 1044-1050. (37) Zeng, Q. D.; Lewis, L. M.; Colangelo, C. M.; Dong, J.; Scott, R. A. J. Biol. Inorg. Chem. 1996, 1, 162-168. (38) Zhu, W. L.; Zeng, Q. D.; Colangelo, C. M.; Lewis, L. M.; Summers, M. F.; Scott, R. A. Nat. Struct. Biol. 1996, 3, 122-124.
Figure 4. Mass spectrum showing three forms of the protein present in bacterial cells 3 h after induction of a Pf Rd (average mol wt ) 5895) overexpression. The pure protein occurs at the lowest abundance, while the methionine and formylated methionine attached species are observed to be more abundant.
with the N-terminal methionine. Although these proteins differ by only 28 Da, they are easily resolved by delayed extraction TOF mass spectrometry. In contrast to the mass spectrometry results, SDS-PAGE analysis of this sample, shown in lane 3 of Figure 1a, produces a single band, thus concealing the presence of the extra components. Although protein expression monitoring by MALDI-TOF is an offline method, the time-dependence of protein expression may be examined by sampling the fermentation at regular intervals after induction. This is illustrated in Figure 5a-c for the overexpression of Pf Rd, examined before induction, and at 1 h intervals afterward for 4 h. Figure 5a shows a mass spectrum of a sample taken before induction, in which the Met form of the target protein is observed to have a very low abundance. The appearance of a background peak, occurring near the molecular weight of the native E. coli ribosomal protein, rpL29, with 30% abundance relative to the Metf form suggests that the recombinant protein concentration is on the order of native bacterial proteins before induction. The sample obtained 1 h after induction shown in Figure 5b indicates Metf-Pf Rd as the most abundant form of the protein, while the target protein is present at less than 10% relative abundance. The abundance of the target protein relative to that of the native E. coli protein collected at four times during the induction is shown in Table 2. The ratios indicate a linear increase of expressed protein over time, consistent with the expected growth rate for this type of system. While the ratio of abundances of expressed Pf Rd to that of the background protein shows a linear increase with time, the absolute value of the signal from the target protein shows fluctuations in abundance. This highlights the importance of using an internal standard for quantitative analysis by using MALDI.39,40 Figure 6a and b shows an example of this type of analysis applied to a higher molecular weight protein, a fusion of a maltose binding protein affinity purification tag and a periplasmic mercury binding protein (MalE/MerP), Mr ) 50 500, kDa, expressed without a signal peptide. Although the low resolution of MALDI(39) Pitt, J. J.; Gorman, J. J. Anal. Biochem. 1997, 248, 63-75. (40) Hensel, R. R.; King, R. C.; Owens, K. G. Rapid Commun. Mass Spectrom. 1997, 11, 1785-1793.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2707
(a)
Table 2. Absolute Intensities for the Target Protein and Background Protein at Various Times during Overexpression time after induction
Pf Rd+Metf peak intensity
background protein peak intensity
ratio
0h 1h 2h 3h
190 540 912 700
69 5 5 3
2.8 108 182 233
a The ratio of the intensities increases linearly with time, with a slope of 1.27 min-1 and a correlation constant, R ) 0.99. Peak intensities were obtained after smoothing the mass spectra with a seven-point weighted average.
(a) (b)
(b) (c)
Figure 5. (a) MALDI-TOF mass spectrum of E. coli cells used for Pf Rd expression obtained before induction showing the methionine and formylated methionine forms at low signal-to-noise. A physiological protein of E. coli is observed at 40% relative abundance, indicating a low concentration of the recombinant protein. Masses below m/z 4000 were deflected to enhance the target protein signal strength. (b) Mass spectrum acquired 1 h after induction in which the pure protein cleavage product appears at low intensity. The rpL29 protein (not shown) has decreased in relative intensity to the most intense peak by about an order of magnitude. (c) Mass spectrum obtained 2 h after induction in which the relative abundance of the Pf Rd (pure) peak continues to increase. The mass spectrum obtained 3 h after induction is presented in Figure 4.
TOF in this mass range precludes the analysis of small molecular weight modifications, rapid determination of successful protein expression is still possible. For these samples, sonication of the bacterial extracts was found to greatly enhance signal-to-noise. 2708 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
Figure 6. (a) Mass spectrum of MalE/MerP fusion protein obtained from nonsonicated bacterial cells using linear mode detection. (b) Mass spectrum of the same protein obtained from sonicated cells.
Figure 6a shows the mass spectrum obtained for a sample without sonication in which bacterial proteins and possible target protein degradation products appear with significant intensity below 20 000 Da. Figure 6b shows a mass spectrum acquired after sonication of the sample in which the overexpressed recombinant protein signal-to-noise ratio is observed to be much stronger than that of the background bacterial proteins. Since sonication did not significantly enhance the signal observed for lower molecular weight target proteins (less than 10 000), it appears that the efficient production of ions from proteins in bacterial cells is size dependent. This behavior has been observed in analysis of several strains of bacteria and was previously attributed to the hydrophobicity of high molecular weight proteins of the species.29 Since
the protein in this example is water soluble, size appears to be the determining factor. Partial chemical lysis of the cell wall provided by the low pH environment during sample preparation, as well as exposure to organic solvents may effectively release small proteins from the bacterial cells into solution while those of higher molecular weight remain trapped in the cell debris. Sonication disrupts the cell wall and is observed to free the higher mass protein contents. CONCLUSIONS MALDI-TOF mass spectrometry has been successfully applied to direct, rapid, offline analysis of protein synthesis by bacterial overexpression. This analysis furnishes simple verification of induction success or failure and has been shown to be effective for the analysis of various proteins over a wide mass range. Using this method, the induction process can be analyzed within minutes, in contrast to SDS-PAGE, which requires an order of magnitude more time. This method could thus be used to provide feedback to the induction process while it is running. The resolution enhancement provided by delayed extraction TOF provides the capability to detect modifications or errors in expression that occur during protein synthesis which are not resolved by SDSPAGE. Additionally, time-dependent molecular weight and relative abundance information can be obtained with the methods reported here. This information can be used to accurately scale up growth, determine optimal induction and harvesting times, and determine when protein altering modifications might occur. Although a randomly chosen background peak provided reasonable estimates of relative abundance, an externally introduced calibrant at known concentration would be preferable since the production rate of physiological proteins may be dynamically regulated during overexpression, leading to errant concentration estimates.
With the methods described here, the analysis of protein overexpression could be extended to sequence verification of the products. This could be accomplished with proteolytic cleavages,41 tandem mass spectrometry,42,43 or a combination of the two techniques. Similarly, MALDI-Fourier transform mass spectrometry (FTMS) could be used for exact mass analysis of proteins and modifications, and sequence verification by MSn.43,44 These approaches are the subject of ongoing investigations in this laboratory. ACKNOWLEDGMENT The authors would thank Dr. Dennis Phillips for use of the Bruker Reflex instrument, Prof. D. M. Kurtz for helpful discussion and Amy Burden, Meredith Anderson, and Hung-Ta Chen for help with cell growth. The financial support of the National Science Foundation (BIR-9413918 and CHE-9412334) and the Rohm and Haas Company is gratefully acknowledged. M.L.E. performed the mass spectrometry measurements and the initial writing. C.M.C. carried out the bacterial growth and overexpression of recombinant proteins, while R.A.S. and I.J.A. provided intellectual and financial support.
Received for review December 11, 1997. Accepted April 2, 1998. AC971344J (41) Cottrell, J. S. Peptide Res. 1994, 7, 1040-5704. (42) Biemann, K. Annu. Rev. Biochem. 1992, 61, 977-1010. (43) Loo, J. A.; Quinn, J. P.; Ryu, S. I.; Henry, K. D.; Senko, M. W.; McLafferty, F. W. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 286-289. (44) Solouki, T.; Pasatolic, L.; Jackson, G. S.; Guan, S. G.; Marshall, A. G. Anal. Chem. 1996, 68, 3718-3725.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2709
