Anal. Chem. 1998, 70, 2493-2494
Multiple Labeling of Proteins Douglas B. Craig† and Norman J. Dovichi*
Department of Chemistry, University of Alberta, Edmonton, Alberta T6G 2G2, Canada
Fluorescent dyes are often used to label proteins before analysis by capillary electrophoresis. Fluorescent labeling produces spectacular improvements in sensitivity compared with UV absorbance detection of the native protein. However, labeling of the protein can lead to significant band broadening. This band broadening is interpreted as a result of multiple labeling of the protein, wherein one or more fluorescent molecules are bound to the protein. The heterogeneous reaction products, which are presumed to have different mobilities, generate a broad peak during electrophoresis. There has been little direct evidence for multiple labeling as the cause of band broadening of proteins. In this paper, we perform electrophoresis on native green fluorescence protein, along with the reaction products produced by fluorescence labeling. For short incubations, a series of regularly spaced components are resolved by free-zone electrophoresis; upon longer incubation, the product peaks merge together, forming a broad envelope. Fluorescence labeling is often used to improve the sensitivity of protein analysis by capillary electrophoresis. However, precolumn labeling often results in poor electrophoretic resolution compared with absorbance detection of the native protein. It is generally assumed that this peak broadening arises from multiple labeling of the protein. Most labeling reactions rely on reagents that attack primary amines. These reagents do not distinguish efficiently between the N-terminal amine and the -amine group associated with lysine residues. Labeling usually does not go to completion, and as a result, the labeling reaction produces a complex mixture of products, corresponding to attachment of different numbers of labels at different sites.1 We have pointed out that there are 2n - 1 possible fluorescent reaction products where n is the number of primary amines on the protein.2 It has been demonstrated that multiple labeling of peptides leads to a set of distinct electrophoresis peaks, one for each product molecule.2 However, multiple labeling of proteins has not been demonstrated to yield distinct products. While it is assumed that multiple labeling of proteins leads to peak broadening, evidence is rather circumstantial. For example, it could be that protein conformation changes during labeling and differences in migration time are related to these conformation † Present address: Department of Chemistry, University of Winnipeg, Winnipeg, Manitoba, Canada. (1) Liu, J.; Hsieh, Y. Z.; Wiesler, D.; Novotny, M. Anal. Chem. 1991, 63, 408412. (2) Zhao, J. Y.; Waldron, K. C.; Miller, J.; Zhang, J. Z.; Harke, H. R.; Dovichi, N. J. J. Chromatogr. 1992, 608, 239-242.
S0003-2700(97)00856-1 CCC: $15.00 Published on Web 05/20/1998
© 1998 American Chemical Society
changes. Analyses of proteins have been hampered by the complex electropherograms generated by most proteins. For example, a series of peaks is generated by the electrophoretic analysis of native ovalbumin; these peaks are due to the different glycoforms of the molecule. These peaks merge into a broad envelope after fluorescent labeling.3 In this paper, we demonstrate multiple labeling of green fluorescent protein (GFP), a high-purity 27 kDa recombinant protein. The molecule generates a single, narrow peak in zone electrophoresis, with a theoretical plate count of greater than 100 000.4 We use zone electrophoresis to follow the reaction of the protein with 3-(2-furoyl)quinoline-2-carboxaldehyde (FQ), the fluorogenic labeling reagent of choice.5 The excitation and emission wavelengths of GFP are blue-shifted with respect to the wavelengths that characterize the fluorescence of FQ-labeled proteins. By monitoring the fluorescence from GFP, rather than the labeled product, we can monitor both the native protein and the products. EXPERIMENTAL SECTION 3-(2-Furoyl)quinoline-2-carboxaldehyde (FQ) was supplied by Molecular Probes (Eugene, OR) and Aequorea victoria green fluorescent protein was from Clonetech (Palo Alto, CA). A 10 µL aliquot of a 3.7 × 10-5 M solution of GFP in 5 mM borate and 1.5 mM NaCN was added to 100 nmol of dry FQ and incubated at room temperature for 1, 5, and 10 min. The sample was immediately diluted by 4 orders of magnitude to quench the reaction. Solutions were analyzed by capillary zone electrophoresis on an uncoated capillary with a borate buffer. GFP was excited with a 457.9 nm line from an Ar+ laser, and fluorescence was filtered with a 515dF20 filter from Omega Optical (Brattleboro, VT). These are not the optimal conditions for GFP detection but were chosen to avoid detection of the attached FQ label. Separation was in a 50 cm long, 50 µm i.d., 145 µm o.d. fused silica capillary at an electric field of 400 V cm-1. Running, sheath, and sample buffers were 20 mM tricine (pH 8.0). Detection was by laser-induced fluorescence. Excitation was at 457.9 nm, and emission was measured at 515 nm. RESULTS AND DISCUSSION Figure 1 shows electropherograms resulting from native and FQ-labeled GFP. Unlabeled GFP gives a sharp peak with a (3) Pinto, D. M.; Arriaga, E. A.; Craig, D.; Angelova, J.; Sharma, N.; Ahmadzadeh, H.; Dovichi, N. J.; Boulet, C. A. Anal. Chem. 1997, 69, 3015-3021. (4) Craig D. B.; Dovichi, N. J. Biomed. Chromatogr. 1997, 11, 205-206. (5) Cobb, K. A.; Dolnik, V.; Novotny, M. Anal. Chem. 1990, 62, 2478-2483.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998 2493
Figure 1. Multiple labeling with FQ. Separation was in a 50 cm long, 50 µm i.d., 145 µm o.d. fused silica capillary at an electric field of 400 V cm-1. Running, sheath, and sample buffers were 20 mM tricine (pH 8.0). Detection was by laser-induced fluorescence. Excitation was at 457.9 nm, and emission was measured at 515 nm. Injections: 3.7 × 10-9 M native GFP; 3.7 × 10-5 M GFP labeled with FQ for 1, 5, and 10 min. The solutions were diluted by 4 orders of magnitude prior to injection.
theoretical plate count of 1.2 × 105. FQ labeling for as little as 1 min caused severe band broadening, with several distinct components visible. These components provide evidence for the addition of one or more labels to the parent molecule. The addition of one neutral FQ dye to GFP neutralizes one positively charged lysine. We measured the mobility of the first five components of the 1 min reaction time data. The mobility of these components decreases linearly with the number of labels attached to the molecule, r ) -0.9994. Only the first five or so components of the 1 min incubation peak are resolved. Attachment of the same number of labels, but at different locations on the protein, may lead to slight variations in the mobility of the product. These products would combine to generate the broad envelope. The 5 min reaction retains hints of peaks for the three- and four-label products, but larger numbers of labels are lost. Extrapolating the mobility to large numbers of
2494 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
labels suggests that the 10 min reaction product has an average of eight dye labels and that there are product molecules with up to 12 dye molecules attached. It is rather amazing that GFP still retains some fluorescent character after half of its lysines have been labeled with FQ. Peak broadening of proteins, like peptides, appears to be due to attachment of multiple labels to the analyte. In the case where the label is neutral, mobility drops linearly with the number of attached labels as the -amine of lysine is neutralized and as the size of the product molecule increases. We have previously demonstrated two methods to reduce peak broadening of fluorescently labeled proteins. First, the molecule can be taken through one cycle of the Edman degradation reaction, which removes the N-terminal amino acid, forms the phenylthiocarbamyl derivative on all -amine groups on lysine, and leaves a free N-terminal amine group for labeling.2 This chemistry is cumbersome, is difficult to perform on dilute solutions, and is unlikely to go to completion for larger proteins. As a second method, we have demonstrated that the use of submicellar concentrations of SDS helps to dramatically sharpen peaks generated by fluorescently labeled proteins.3 As a third method, it would appear useful to employ a cationic label, so that there is no change in the protein’s charge upon incorporation of the label. Unfortunately, we are not aware of any positively charged fluorogenic reagents; such compounds could be very useful as a general method of labeling trace amounts of proteins while preserving high separation efficiency. ACKNOWLEDGMENT This work was supported by the Natural Sciences and Engineering Research Council and the Department of National Defence. D.B.C. acknowledges a postdoctoral fellowship from the Alberta Heritage Foundation for Medical Research.
Received for review August 8, 1997. Accepted March 17, 1998. AC970856V
