Anal. Chem. 2001, 73, 1610-1613
Technical Notes
Attomole-Level Protein Fingerprinting Based on Intrinsic Peptide Fluorescence Eric Okerberg and Jason B. Shear*
Department of Chemistry & Biochemistry, and the Institute for Cellular & Molecular Biology, University of Texas, Austin, Texas 78712
Protein identification has relied heavily on proteolytic analysis, but current techniques are often slow and generally consume large quantities of valuable protein sample. We report the development of a rapid, ultralow volume protein analysis strategy based on tryptic digestion within the tip of a 1.5-µm capillary channel followed by separation of the proteolytic fragments using capillary electrophoresis (CE). Two-photon excitation is used to probe the intrinsic fluorescence of peptide fragments through “deep-UV” excitation of aromatic amino acid residues at the outlet of the CE channel. Detection limits using this technique are 0.7, 2.4, and 23 amol for the aromatic amino acids tryptophan, tyrosine, and phenylalanine, respectively. In these studies, we demonstrate the capacity to differentiate bovine and yeast cytochrome c variants using less than 15 amol of protein through tryptic fingerprinting. Moreover, the detection of a single amino acid substitution between bovine and canine cytochrome c illustrates the sensitivity of this approach to minor differences in protein sequence. The 2-pL sample volume required for this on-column tryptic digestion is, to our knowledge, the smallest yet reported for a proteolytic assay. Current optical techniques for protein identification typically rely on time-consuming and/or costly derivatization procedures to generate a fluorescent product that is excitable using visible wavelengths. These assays generally require some fractionation procedure, such as capillary electrophoresis (CE), to correlate signal with protein identity. Proteolytic digests can be analyzed in such a manner to “fingerprint” the parent protein, providing a means to identify proteins that may be difficult to differentiate based solely on the migration velocity of the parent compound. However, reagent blanks and multiple peak formation (caused by incomplete fluorescence labeling of multiple reactive sights) often complicate protein identification using this strategy.1-3 * Corresponding author. E-mail:
[email protected]. (1) Krull, I.; Strong, R.; Cho, B.; Beale, S.; Wang, C.; Cohen, S. J. Chromatogr. B 1997, 699, 173-208. (2) Xhao, J.; Waldron, K.; Miller, J.; Zhang, J.; Harke, H.; Dovichi, N. J. Chromatogr. 1992, 608, 239-242. (3) Craig, D.; Dovichi, N. Anal. Chem. 1998, 70, 2493-2494.
1610 Analytical Chemistry, Vol. 73, No. 7, April 1, 2001
Conventional one-photon excitation (1PE) of intrinsic protein fluorescence has been demonstrated with ultraviolet (UV) lasers, but relatively large samples are required to generate signal that is detectable over the significant background typically associated with UV excitation.4-7 To date, femtomole amounts of proteins have been required for tryptic analyses using CE with 1PE.4,7 The lowest detection limits achieved in such studies have relied on detection of tryptophan (Trp)-containing peptides, but the relative infrequence of this residue limits the utility of this approach for protein identification assays. Although fluorescence from the more common amino acids tyrosine (Tyr) and phenylalanine (Phe) has been used for peptide measurements, detection limits suffer dramatically because of their relatively weak fluorescence.5 We report the application of CE with two-photon excited fluorescence for ultralow-volume proteolytic analyses sa strategy that provides large improvements in mass detection limits for Tyrand Phe-containing fragments. In these experiments, picoliter protein samples are subjected to tryptic digestion in the entrance tip of a 1.5-µm electrophoresis capillary. After proteolysis, peptide fragments are separated using a large electric field and are detected at the outlet using two-photon excited intrinsic deep-UV fluorescence. The nonlinear nature of two-photon excitation (2PE) restricts fluorescence generation to three-dimensionally defined, femtoliter focal volumes when using a high numerical aperture (NA) objective to tightly focus the laser light. As a consequence, essentially no autofluorescence is produced within optical components, and extremely narrow separation channels (capable of accommodating ultralow-volume samples) can be used without exciting significant emission from the fused silica capillary. Importantly, because elastically scattered laser light in these studies has a substantially longer wavelength (445 nm) than peptide fluorescence (∼300-350 nm), it can be readily removed from the detection path by colored-glass filters that efficiently transmit UV fluorescence to the detector. We demonstrate the feasibility for using this strategy to differentiate between closely related cytochrome c variants using only attomole quantities of sample. (4) Chang, H.; Yeung, E. Anal. Chem. 1992, 65, 2947-2951. (5) Timperman, A.; Oldenburg, K.; Sweedler, J. Anal. Chem. 1995, 67, 34213426. (6) Lee, T.; Yeung, E. Anal. Chem. 1992, 64, 3045-3051. (7) Park, Y.; Zhang, X.; Rubakhin, S.; Sweedler, J. Anal. Chem. 1999, 71, 49975002. 10.1021/ac0012703 CCC: $20.00
© 2001 American Chemical Society Published on Web 02/27/2001
Figure 1. Fluorescence measurements are made at the outlet of a separation capillary using a high numerical aperture objective (obj) in an epi-collection geometry; the imaging beam splitter (BS) was used to ensure alignment of the 2PE probe volume with the capillary outlet. SHG, second-harmonic generator; DM, dichroic mirror; PMT, photomultiplier tube; F, filters; CCD, video camera; cap, reaction/ separation capillary.
EXPERIMENTAL PROTOCOL Sepaniak and Yeung first introduced 2PE fluorescence as a detection mode for analytical separations,8 and recent developments in laser technology have prompted renewed interest in this field. The CE-2PE instrumentation used in the present studies is depicted in Figure 1 and is similar to that described previously.9 A Coherent Mira 900-F femtosecond mode-locked titanium: sapphire (Ti:S) oscillator pumped by a 10-W Coherent Verdi solidstate frequency-doubled Nd:Vanadate (Nd:YVO4) laser was used for all studies. The Mira was operated at 890 nm, producing a pulse width of ∼100-150 fs at a repetition rate of ∼76 MHz. To achieve desired powers, the Mira output was attenuated using a half-wave plate/polarizer pair. Extracavity second-harmonic generation was accomplished using a Super Doubler (Super Optronics) equipped with an LBO doubling crystal. Approximately 120140 mW of 445-nm light was directed through a 1.3 NA 100× UltraFluar microscope objective (Zeiss) for two-photon excitation. Fluorescence signal was collected by the same objective and was reflected from the beam path using a long-pass dichroic mirror (Chroma Technologies, 375DCLP). Residual laser scatter and other long-wavelength background were filtered using three 3-mmthick UG11 filters. Signal was measured using a UV-sensitive bialkali photomultiplier tube connected to a photon counter, and data were transferred from the counter (0.2-s bins) to a Macintosh running LabView-based software. A manually operated highvoltage power supply (Spellman, CZE1000R) was used for injections and separations. Fused-silica capillary columns (1.5-µm i.d., 360-µm o.d.) were obtained from Polymicro Technologies. Amino acid samples were prepared fresh daily in the electrophoresis running buffer (10 mM Hepes, pH 7.6) and were used as received from Sigma. Data for amino acid standards were obtained using 2-kV, 3-s electrokinetic sample injections followed by 20-kV separations in a Polymicro capillary (29.8-cm length). Cytochrome c variants from bovine heart, canine heart, and Saccharomyces cerevisiae were obtained from Sigma. Trypsin and protein samples were prepared in the electrophoresis buffer (10 mM borate, pH 9.2), and on-column tryptic digests were performed/ analyzed in a 38.6-cm-long capillary. On-column proteolysis was achieved by introducing a plug of 0.1 mg/mL trypsin with 10 mM CaCl2 (3-kV, 5-s electrokinetic injection), followed by the cyto(8) Sepaniak, M. J.; Yeung, E. S. Anal. Chem. 1977, 49, 1554-1556. (9) Gostkowski, M.; Shear, J. J. Am. Chem. Soc. 1998, 120, 12966-12967.
chrome c sample plug (3-kV, 3-s injection), and then a second trypsin/CaCl2 plug (3-kV, 5-s injection). Reactions typically were allowed to proceed for 10 min at 50 °C before fragments were separated using 20 kV. Identification of EDLIAYLK and EDLIAY fragments was accomplished using synthetic peptides as standards and was verified by LC-MS. Detection limits for amino acids and peptide fragments were calculated as the mean peak height from at least three separations, and represent the amount of analyte needed to produce a signalto-rms noise value of 3. For calculations, tryptic digest data were subjected to a three-point running smooth. Typical relative standard deviations (RSDs) for amino acid analyses were
