Study on the quality of recombinant proteins using matrix-assisted laser desorption ionization time of flight mass spectrometry

Abstract

AIM: To study the possibility of matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) for controlling the quality of recombinant proteins.

METHODS: By using MALDI-TOF MS, the molecular weights and purity of recombinant bioactive proteins were analyzed.

RESULTS: The molecular weights and purity were obtained in nine recombinant bioactive proteins, including interleukin 2, tumor necrosis factor α, granulocyte-macrophage colony stimulating factor, interferon α2b, interferon α1, erythropoietin, calmodulin and its fragment, and neuronal nitric oxide synthase were obtained. MALDI-TOF MS was also used to assay specific proteins in the mixtures and to characterize the erythropoietin tryptic digests.

CONLUSION: The results showed that MALDI-TOF MS can be employed for the effective quality control of recombinant proteins.

Key Words: recombinant proteins; molecular weight; flight mass spectrometry; erythropoietin tryptic digests

INTRODUCTION

Since Hillenkamp et al[1] first introduced matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) to analyze proteins with molecular masses greater than Mr 10000, MALDI-TOF MS has been widely used to study different classes of biomolecules such as proteins, oligonucleotides, polysaccharides and polymers[2-5]. Compared to the traditional techniques, such as sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), MALDI-TOF MS has several advantages in the determination of protein molecular weight (Mr), peptide mapping, and purity.

In the present report, MALDI-TOF MS was used to accurately determine the Mr and purity of nine biological samples including recombinant bioactive proteins, such as: interleukin-2 (IL-2), interferon-alpha 2b (IFNα2b), interferon-alpha 1 (IFNα1), erythropoietin (EPO), granulocyte-macrophage-colony stimulating factor (GM-CSF), tumor necrosis factor alpha (TNFα), calmodulin (CaM) and its fragment, and neuronal nitric oxide synthase (nNOS). In addition, the protein mixtures and EPO peptide mapping were characterized by the technique.

MATERIALS AND METHODS

Materials

Chemicals 3,5-dimethoxy-4-hydroxycinnamic acid (Sinapinic acid), 2,5-dihydroxybenzoic acid (DHB) and acetonitrile were produced by Sigma (St, Louis, USA). Trifluoroacetic acid (TFA) was from Merk-Schuchardt. Ammonium hydrogencarbonate, acetic acid and ethanol were produced by Nanjing Chemical Reagent Plant (Nanjing, China). All water used in the experiment was ultra high quality produced by a Milli-Q Plus Ultra Pure Water System.

Proteins Horse heart myoglobin, carbonic anhydrase B, β-lactoglobulin, bovine serum albumin and trypsin were obtained from Sigma (St, Louis, USA). IL-2 (E.coli), IFNα1 (Silkworm cell), GM-CSF (E.coli ), IFN α2b ( E.coli), EPO (CHO), nNOS, and CaM and its fragments were supplied by the East China Institute for Medicine and Biotechnology (Nanjing, China).

Methods

Tryptic digestion of EPO Approximately 200 µg EPO protein was lyophilized and dissolved in 200 µL 1% ammonium bicarbonate (pH 8.5). Trypsin protease was added to the solution which was incubated at 37 °C for 16h in a substrate: enzyme ratio of 50:1 (w:w). The digestion was quenched by storing the sample at -70 °C. The sample was lyophilized and reconstituted in 50 µL 10% acetic acid before use.

MALDI-TOF MS Analyses were performed on a Finnigan Laser MAT 2000 time-of-flight mass spectrometer (Finnigan MAT, Hemel Hempstead, Herts, UK). The system used a nitrogen laser (337 nm, 2 ns pulse) to des orb ions from the sample specimen . The desorbed ions were accelerated to 20kV into a free long tube. The time recorded for a molecule to travel the length of the tube to a detector was proportional to the mass of the ion, which was its molecular weight. All spectra were obtained using the positive-ion mode. Standard stainless-steel targets (with a sample application area of about 3.14 mm²) obtained from the manufacturer were employed for all analyses. The lasermat software allowed the user to irradiate one of the four possible target regions or quadrants of about 0.02 mm². The spectra in this study were calibrated using instrumental calibration, based on the parameters determined from analysis of a number of standard proteins and peptides.

Sample preparation for MALDI-TOF MS determination The samples of each protein were prepared to 0.8 g/L-1.0 g/L in dilute TFA (0.05%. 0.09 mol/L sinapinic acid in acetonitrileethanol-water(60:4:36, v/v/v) was used as protein matrix and 0.1 mol/L 2,5-dihydroxybenzoic (DHB) in formic acid-water (9:1, v/v) was used as peptide matrix. The matrix solution was kept in the dark and prepared fresh every few days.

The samples for mass spectrometric analysis were mixed with the matrix in an Eppendorf tube. Typically, 5 µL of sample solution was added to a tube containing 10 µL of matrix solution. The solution was stirred in a vortex mixer, and then approximately 0. 5 µL of sample solution was applied to the target and followed by drying at room temperature and atmospheric pressure. For some protein samples in which the buffer contained salts at high concentrations, the dried sample/matrix preparation on the target was washed by depositing a droplet of cold distilled water for a few seconds on the sample spot surface to dissolve excess salt crystals, and followed by removing the droplet.

RESULTS AND DISCUSSION

Mr measurement for purified proteins

Nonglycosylated proteins The Mrs of recombinant proteins, IL-2, TNFα, GM-CSF, TNFα-2b, CaM fragments and nNOS expressed in E.coli were determined using MALDI-TOF MS. Table 1 shows the number of amino a cids, theoretical Mr, measured Mr and relative determination error for these nonglycoproteins.

Table 1 Mrs of non-glycoproteins measured by MALDI-TOF MS.

Protein	Number of amino acids	Theoretical Mr	Measured Mr	Relative error (%)
IL-2	134a	15478	15610	2.8
GM-CSF	127	14477	15451	5.56
TNF-α	158a	17484	17517	0.18
nNOS	199a	22248	22306	0.26
IFN-α2b	166a	19378	19533	0.79
CaM fragment	95	10846	10851	0.05

^aAn additional methionine at the NH-2-terminal.

As Table 1 shows, the M_r relative error of the protein GM-CSF was 5.56%, which was much higher than the systematic error of the instrument. In addition, the peak in the GM-CSF mass spectrum was very sharp, indicating that the product was very pure. However, the results of Edman sequencing showed that there were 8 additional amino acids, MMKSDNSH, at the NH-2-terminus. The relative error was 0.12% when the 8 amino acids were added to the regular GM-CSF amino acid sequence. The relative errors of interleukin-2 and interferon α2b were approximately 0.8%. The increased mass was approximately equal to the M_r of an amino acid, which may have been added during the modifications, such as phosphorylation and cystinylation. The details were not achieved in the present report. An asymmetric single charged peak was observed in the interferon α2b mass spectrum (Figure 1), with an unresolved impurity peak on the left side that could be seen when the mass spectrum was amplified partially. This may be caused by partially removing N terminal methionine residue in the interferon α2b.

Figure 1 MALDI-TOF mass spectrum of interferon α2b using sinapinic acid as matrix.

Glycosylated proteins Proteins expressed in mammal cells are generally glycosylated. Because of the microheterogeneity of the carbohydrates, the final product is a complicated heterogeneous mixture. In some cases, the functions of the biomolecules depend on the carbohydrate components[11,12]. For example, recombinant human erythropoietin will lose its in vivo bioactivity wi thout sialic acid residues[13,14]. Thus, it is very important to characterize the carbohydrate parts in glycoproteins[15,16]. In this article, MALDI-TOF MS was used to determine M_r and carbohydrate percentages of erythropoietin (EPO) and interferon α1 (Table 2).

Table 2 Mrs of glycoproteins measured by MALDI-TOF MS.

Protein	Number of amino acids	Measured Mr	Carbohydrate content (%)
EPO	166	28707	35.6
IFNα1	166	20465	5.2

As can be seen in Figure 2, the half-intensity-width of EPO was much larger, indicating that EPO was a heterogeneous protein. The carbohydrate percentage was so high that electrospray mass spectrometry could not be employed to determine the heterogeneity. The Mr obtained using MALDI-TOF MS was 28707, while the Mr obtained using the traditional SDS-PAGE technique was about 34000. The mass difference of 5293 suggested that SDS-PAGE could measure the apparent Mr of the glycoprotein with high carbohydrate content. Since the results of SDS-PAGE for the molecular weight were related to the gel concentration, the determination procedure was difficult to control. The degree of glycosylation for biomolecules expressed in silkworm cells was generally lower, with only 5.2% carbohydrates found in recombinant protein interferon α1 by MALDI-TOF MS, which was consistent with theoretical calculations.The TOF mass spectra for EPO and IFNα1 showed that the half-intensity-widths were 8000 and 3000, respectively, much wider than that of nonglycoproteins in Table 1, Figure 3 and Figure 4. Ordinarily the half-intensity-widths are produced by isotopes, and are no more than 500[17]. Therefore, half-intensity-widths of 500 in singly charged MS peaks may be employed as a simple criterion to evaluate the heterogeneity of glycoproteins. The MS peak width will increase as glycoprotein becomes more heterogeneous.

Figure 2 MALDI-TOF mass spectrum of recombinant human erythropoietin using sinapinic acid as matrix.

Figure 3 MALDI-TOF mass spectrum of mixtures containing calmodulin using sinapinic acid as matrix.

Figure 4 MALDI-T OF mass spectrum of calmodulin fragment using sinapinic acid as matrix.

Determination of protein mixture

Mixture composed of known standard proteins Methods used to analyze each component in multi-protein mixtures with out tedious separation are of great value. The advantage of MALDI-TOF MS in multicomponent characterization is the ability to simultaneously determine the M_r of each ingredient during one scan. However, singly charged oligomers with multiply charged monomers and multiply charged oligomers would complicate the mass spectra. Peaks could be specified only according to their M_r and the number of charges.

Figure 5 shows the MALDI-TOF mass spectrum for mixed proteins composed of myogl obin from horse heart, carbonic anhydrase B, β-lactoglobin and bovine serum albumin. Both molecular ion and the multiply charged monomers or oligomers were observed. The M_r measurement accuracy for the four proteins was 0.0%, 0.04%, 0.12%, and 0.18%, respectively.

Figure 5 MALDI-TOF mass spectrum of standard protein mixtures using sinapinic acid as matrix. M: Myoglobin; C: Carbonic anhydrase B ; L: β-lactoglobulin; B: Bovine serum albumin.

Determination of a specific protein in mixtures The mass spectrum of E.coli expressed products containing calmodulin in Figure 3, showed a protein mixture containing a protein with an observed M_r of 16934, together with a doubly charged monomer and a singly charged dimer, indicating that the mixture contained calmodulin with a theoretical M_r of approximately 17000. In addition, the spectrum showed that there was very little singly charged dimer, because calmodulin seldom formed dimers.

Measurement of the purity of expressed products and the relative content of impurities Since MALDI-TOF MS can give the M_r of all components in a mixture, the relative purity of the expressed product, along with the M_r of the impurity and its abundance, can be obtained from the mass spectrum. Figure 4 shows the mass spectrum of calmodulin fragment expressed in E.coli had small amounts of impurities with M_r of 9643.8 and 9074.0. In addition, few impurities were observed in the recombinant proteins, IL-2, TNFα, GM-CSF, IFNα2b, nNOS, IFNα1 and EPO.

Characterization of tryptic peptide mapping for erythropoietin

The MALDI-TOF peptide mapping strategy has been used successfully to analyze proteins in enzymatic digests and has thus been applied to verify the primary structure of some proteins[18]. As a technique, it offers a number of advantages over conventional peptide mapping with HPLC since it reduces the actual analysis time and also provides a means of assigning peptide fragments from their respective M_r. Peptide mapping is the “fingerprint” of the protein, so it can be employed to rapidly identify the correctness of the primary structure of protein during expression or purification[19]. This report introduces the peptide map of recombinant human erythropoietin by MALDI-TOF MS. Protease trypsin can theoretically digest the erythropoietin into 21 peptide fragments, including three glycopeptide fragments and two disulfide bonds (one between fragments T₁ and T₂₀, and another inside T₅). The proteolytic digests were subjected directly to MALDI-TOF MS analysis without any prior purification. Figure 6 shows the mass spectrum of the tryptic digest with 13 peaks corresponding to peptide fragments. The observed mass values of the digests were consistent with the theoretical mass values as calculated from the amino acid sequence of rhEPO (Table 3). The mass value of 1612.7 indicated the formation of a disulfide bond between T₁ nor T₂₀, which was verified by the fact that neither T₁ nor T₂₀ was observed. Peptide mixtures generated by enzymatic digestion do not always allow for complete evaluation of peptide fragments by MALDI-TOF MS. As can be seen in Table 3, about 76.5% of the entire amino acid sequence was confirmed. In the lower mass range (below m/z 400), analyte molecule s could not be identified since the matrix peaks overlap with the signals from the analytes. In addition, the three glycosylated peptides were also not observed, because EPO was heterogeneous and the MS sensitivity to carbohydrates was much lower than that of peptides. Several matrices, including 2,5-dihydroxy benzoic acid (DHB), 2-aminobenzoic acid and α-cyano-4-hydroxycinnamic acid, were tested to obtain the best resolution and sensitivity in the mass spectrum measurement. The results in Figures 6, 7 and 8, show that the best matrix was DHB, which produced large analytical signals with good resolution. The 2-aminobenzoic acid matrix gave signals with lower sensitivity and poorer resolution, but it was better than the α-cyano-4-hydroxycinnamic acid matrix that is more suitable for “big” proteins rather than “small” peptides. The matrix concentration also affects the mass resolution and analyte signals. For example, with saturated DHB solution employed as the matrix, only two main peaks with mass values of about 480 and 2550 were obtained, possibly due to poor crystallization over the entire sample area in the target.

Table 3 MALDI-TOF-MS analysis of the tryptic digests of EPO.

rhEPO peptide	Position	Calculated mass value	Observed mass value	Relative error (%)
T1	1-4	439.5	440.5	0.23
T3	11-14	515.6	517.8	0.39
T4	15-20	735.9	735.5	-0.05
T5	21-45	glycopeptide
T6	46-52	927.1	928.4	0.14
T7	53	174.2
T8	54-76	2525.3	2525.5	0.01
T9	77-97	glycopeptide
T10	98-103	601.7	600.8	-0.15
T11	104-110	802.5	803.4	0.11
T12	111-116	586.3	585.5	-0.14
T13	117-131	glycopeptide
T14	132-139	923.0	923.5	0.05
T15	140	146.2
T16	141-143	434.5	*
T17	144-150	898.0	899.1	0.12
T18	151-152	203.2
T19	153-154	259.3
T21	163-166	447.4	*
T2-S-S-T20	1615.8	1612.7		-0.19

*A wide unresolved peak caused by T₁, T₁₆ and T₂₁.

Figure 6 MALDI-TOF mass spectrum of tryptic digests of recombinant human erythropoietin using 100 mM DHB as matrix.

Figure 7 MALDI-TOF mass spectrum of tryptic digests of recombinant human erythropoietin using 2-Aminobenzoic acid as matrix.

Figure 8 MALDI-TOF mass spectrum of tryptic digests of recombinant human erythropoietin using α-cyano-4-hydroxycinnamic acid as matrix.

In general, the choice of MALDI-TOF matrix is very critical to the homogeneous crystallization and the homogeneous embedding of the analyte molecules in the matrix. In addition, the matrix is responsible for the high mass resolution and high analyte signals.

CONCLUSION

The present investigations clearly show that MALDI-TOF MS is a powerful tool for measuring protein molecular weight, identifying unknown components in mixtures, determining purity and the relative content of impurities, and characterizing peptide mapping. The MALDI-TOF analysis validated that SDS-PAGE is unreliable in determining the Mr of glycoproteins. Since the broad MS peaks reflect the carbohydrate heterogeneity, half-intensity widths, MALDI-TOF MS peaks are recommended as a criterion to evaluate the heterogeneity of glycoproteins. The advantages of MALDI-TOF MS were described in this paper, such as its accuracy, small voluem sample consumption and rapid response. It is suggested that this technique should be the first choice when studying the quality of scarce or precious samplesduring the early stages in the development of engineered proteins.