Cleavage of Peptides and Proteins Using Light-Generated Radicals

Barbara J. Jones,* Matthew J. Vergne, David M. Bunk, Laurie E. Locascio, and Mark A. Hayes. Arizona State University, Tempe, Arizona 85257, and Nation...
5 downloads 0 Views 243KB Size
Anal. Chem. 2007, 79, 1327-1332

Cleavage of Peptides and Proteins Using Light-Generated Radicals from Titanium Dioxide Barbara J. Jones,* Matthew J. Vergne, David M. Bunk, Laurie E. Locascio, and Mark A. Hayes

Arizona State University, Tempe, Arizona 85257, and National Institute of Standards and Technology, Gaithersburg, Maryland 20899

Protein identification and characterization often requires cleavage into distinct fragments. Current methods require proteolytic enzymes or chemical agents and typically a second reagent to discontinue cleavage. We have developed a selective cleavage process for peptides and proteins using light-generated radicals from titanium dioxide. The hydroxyl radicals, produced at the TiO2 surface using UV light, are present for only hundreds of microseconds and are confined to a defined reagent zone. Peptides and proteins can be moved past the “reagent zone”, and cleavage is tunable through residence time, illumination time, and intensity. Using this method, products are observed consistent with cleavage at proline residues. These initial experiments indicate the method is rapid, specific, and reproducible. In certain configurations, cleavage products are produced in less than 10 s. Reproducible product patterns consistent with cleavage of the peptide bond at proline for angiotensin I, Lysbradykinin, and myoglobin are demonstrated using capillary electrophoresis. Mass characterization of fragments produced in the cleavage of angiotensin I was obtained using liquid chromatography-mass spectrometry. In addition to the evidence supporting cleavage at proline, enkephalin and peptide A-779, two peptides that do not contain proline, showed no evidence of cleavage under the same conditions. Many techniques currently used in proteomics research rely on protein cleavage for identification of proteins; in fact, there are few alternatives to the use of protease digestion of separated proteins to accomplish this task. Protease cleavage, while specific and reliable, requires much effort and time either in preparation or in interfacing the digestion media with analytical devices. Tryptic digest is most commonly used as it provides highly specific fragmentation at arginine and lysine residues and its use has been refined over a period of decades.1 The rate of digestion is improved with greater protease-to-protein contact and limited by difficulties with self-digestion. Strategies such as surface attachment1-6 and * To whom correspondence should be addressed. E-mail: [email protected]. (1) Greiner, D. P.; Hughes, K. A.; Gunasekera, A. H.; Meares, C. F. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 71-75. (2) Malmsten, M.; Xing, K. Z.; Ljunglof, A. J. Colloid Interface Sci. 1999, 220, 436-42. (3) Ge, S. J.; Bai, H.; Zhang, L. X. Biotechnol. Appl. Biochem. 1996, 24, 1-5. (4) Kulik, E. A.; Kato, K.; Ivanchenko, M. I.; Ikada, Y. Biomaterials 1993, 14, 763-69. 10.1021/ac0613737 CCC: $37.00 Published on Web 01/20/2007

© 2007 American Chemical Society

increased surface-adsorbed protease7 are successful at the cost of increased experimental complexity. Other cleavage agents such as cyanogen bromide,8 photochemical cleavage,9 and cleavage using radical species,1,10-16 among others, have all been used successfully to provide identification and structural information, though these techniques vary in their usefulness and facility. We have developed an approach to site-specific cleavage of the peptide backbone using locally generated hydroxyl radicals (OH•) with products consistent with cleavage at proline residues. Rather than using proteases or chemical cleavage agents, the cleavage method takes advantage of the photocatalytic property of the photoactive semiconductor, titanium dioxide (TiO2), for the production of OH•.17 The hydroxyl radicals are produced in a “cleavage zone”, which is confined by the area of illumination and the short lifetimesand hence limited diffusional distancesof the radicals, and exists only during illumination of the TiO2 (Figure 1). This is an initial presentation of this approach and several aspects will need to be investigated further, but many advantages can be envisioned as it is more fully developed. No additional soluble reagents are required, except native levels of dissolved oxygen, thus avoiding potential contamination and experimental complexity. The stringent requirements and high cost for use of enzymes are avoided, eliminating the need for a temperaturecontrolled environment for movement, storage, and application of enzymes. This enables applications that are permanently integral to instrumentation or microfabricated devices that can (5) Bahadur, A.; Bahadur, P. Indian J. Biochem. Biophys. 1985, 22, 107-10. (6) Yanyshpolsky, V. V.; Tertykh, V. A.; Lyubynsky, G. V. Ukr. Biokhim. Zh. 1979, 51, 324-29. (7) Gao, J.; Xu, J. D.; Locascio, L. E.; Lee, C. S. Anal. Chem. 2001, 73, 264855. (8) Kuhn, K.; Thompson, A.; Prinz, T.; Muller, J.; Baumann, C.; Schmidt, G.; Neumann, T.; Hamon, C. J. Proteome Res. 2003, 2, 598-609. (9) Buranaprapuk, A.; Kumar, C. V.; Jockusch, S.; Turro, N. J. Tetrahedron 2000, 56, 7019-25. (10) Platis, I. E.; Ermacora, M. R.; Fox, R. O. Biochemistry 1993, 32, 12761-67. (11) Hoyer, D.; Cho, H.; Schultz, P. G. J. Am. Chem. Soc. 1990, 112, 3249-50. (12) Heyduk, E.; Heyduk, T. Biochemistry 1995, 34, 15388. (13) Heyduk, T.; Baichoo, N.; Heyduk, E. In Probing of Proteins by Metal Ions and Their Low-Molecular-Weight Complexes; Sigel, A., Sigel, H., Eds.; Marcel Dekker: New York, 2001. (14) Heyduk, T.; Heyduk, E.; Severinov, K.; Tang, H.; Ebright, R. H. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 10162-66. (15) Ermacora, M. R.; Ledman, D. W.; Hellinga, H. W.; Hsu, G. W.; Fox, R. O. Biochemistry 1994, 33, 13625-41. (16) Rana, T. M.; Meares, C. F. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 1057882. (17) Fujishima, A.; Kohayakawa, K.; Honda, K. J. Electrochem. Soc. 1975, 122, 1487-89.

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007 1327

Figure 1. Illustration of cleavage of angiotensin I at the surface of TiO2 via hydroxyl radical attack of proline.

provide specific cleavage products by simply controlling the illumination of the reaction zone. We anticipate that full development of this tool will provide researchers with a facile, inexpensive, and rapid protein cleavage process. The phenomenon of hydroxyl radical production at the surface of TiO2 has been extensively studied17-24 and is employed in a variety of processes including wastewater treatment and air purification, as well as an array of products such as self-cleaning glass and self-sanitizing surgical room tiles. Hydroxyl radical production by TiO2 results from an energy difference of 3.2 eV between the valance and conducting bands of TiO2 which, with the exposure of UV light energy to the semiconductor, will result in the excitation and migration of electrons. The electrons leave behind electron holes, which can then migrate to the surface of the material. The highly oxidizing holes will strip electrons from adsorbed or bound hydroxyl groups, which results in the formation of hydroxyl radicals.17 Though specific peptide bond cleavage by radical species had been noted by researchers characterizing radical damage of proteins,25-35 previous attempts at harnessing radical oxygen species for analytical cleavage of peptide bonds have been limited (18) Gu, Z. Z.; Hayami, S.; Kubo, S.; Meng, Q. B.; Einaga, Y.; Tryk, D. A.; Fujishima, A.; Sato, O. J. Am. Chem. Soc. 2001, 123, 175-76. (19) Ishibashi, K.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J. Photochem. Photobiol. A 2000, 134, 139-42. (20) Sun, R. D.; Nakajima, A.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J. Phys. Chem. B 2001, 105, 1984-90. (21) Tatsuma, T.; Tachibana, S.; Fujishima, A. J. Phys. Chem. B 2001, 105, 698792. (22) Tryk, D. A.; Fujishima, A.; Honda, K. Electrochim. Acta 2000, 45, 236376. (23) Wang, R.; Hashimoto, K.; Fujishima, A.; Chikuni, M.; Kojima, E.; Kitamura, A.; Shimohigoshi, M.; Watanabe, T. Adv. Mater. 1998, 10, 135-+. (24) Al Bastaki, N. M. Chem. Eng. Process. 2004, 43, 935-40. (25) Davies, K. J. Biol. Chem. 1987, 262, 9895-901. (26) Davies, M. J. Arch. Biochem. Biophys. 1996, 336, 163-72.

1328 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

primarily by experimental complexity. Peptide cleavage with reactive oxygen species produced by metal complexes covalently tethered to the protein showed some promise as a site-specific chemical technique; however, the logistics of synthesis, purification, attachment, and finally the cleavage reaction make this method largely impractical.10,11,15 In addition to these techniques that use metal complexes, conjugation of a free radical initiator at the N-terminus of a peptide was used for cleavage when combined with collisionally activated dissociation within an ion trap mass spectrometer.36 Though both of these techniques are innovative, they require complex sample preparation. In this work, consistent and predictable reaction products from peptides and the protein myoglobin were accomplished using locally generated hydroxyl radicals produced by illumination of TiO2. In these experiments, only the dissolution of lyophilized peptides in aqueous solutions was required, where the reactions were performed without the addition of other reagents. Reactions were carried out in a variety of conditions and formats; five specific device configurations are presented: (1) a well device with TiO2coated glass forming the bottom of the well; (2) a well device containing TiO2 particles suspended in solution; (3) a microfluidic channel device with a square channel profile with one TiO2-coated (27) Dean, R. T.; Fu, S. L.; Stocker, R.; Davies, M. J. Biochem. J. 1997, 324, 1-18. (28) Hawkins, C. L.; Davies, M. J. Biochim. Biophys. Acta 1997, 1360, 84-96. (29) Hawkins, C. L.; Davies, M. J. J. Chem. Soc., Perkin Trans. 2 1998, 261722. (30) Hawkins, C. L.; Davies, M. J. Biochim. Biophys. Acta 2001, 1504, 196219. (31) Headlam, H. A.; Davies, M. J. Free Radical Biol. Med. 2002, 32, 1171-84. (32) Puchala, M.; Schuessler, H. Int. J. Radiat. Biol. 1993, 64, 149-56. (33) Puchala, M.; Schuessler, H. Int. J. Pept. Protein Res. 1995, 46, 326-32. (34) Schuessler, H.; Herget, A. Int. J. Radiat. Biol. 1980, 37, 71-80. (35) Schuessler, H.; Schilling, K. Int. J. Radiat. Biol. 1984, 45, 267-81. (36) Hodyss, R.; Cox, H. A.; Beauchamp, J. L. J. Am. Chem. Soc. 2005, 127, 12436-37.

glass wall and channel depths of 180 µm; (4) a microfluidic channel device with a square channel profile with two TiO2-coated glass walls and channel depth of 100 µm; and (5) a microflow reactor filled with TiO2-coated silica beads. EXPERIMENTAL SECTION Materials. Titanium tetraisopropanate (TPT) was provided by DuPont Chemicals (Wilmington, DE). Horse skeletal myoglobin and N-(tris(hydroxymethyl)methyl)-3-aminopropanesulfonic acid (TAPS) were obtained from Sigma-Aldrich (St. Louis, MO). Angiotensin I, Lys-bradykinin, dynorphin , peptide A-779, and [D-Ala2, D-Leu5] enkephalin were obtained from American Peptide Co., Inc. (Sunnyvale, CA). Additional angiotensin I samples were obtained from Penninsula Labs and from Beckman Labs, no longer supplying peptides. Poly-N-hydroxyethyacrylamide (PHEA) was provided in a 1% (w/v) experimental solution from Cambrex Bioscience (Walkersville, MD) and then diluted with deionized, ultrafiltered (DIUF) water to 0.1% (w/v). Sylgard-184 poly(dimethylsiloxane) (PDMS) was obtained from Dow Corning (Midland, MI). Acetate buffers were prepared from acetic acid, sodium acetate, and sodium chloride, all obtained from Aldrich Chemicals (St. Louis, MO). Titanium dioxide-coated glass was provided by Pilkington Glass (Activ Self-Cleaning Glass, Perrysburg, OH). EXPERIMENTAL PROCEDURE Titanium Dioxide Coating. A clean 3 cm × 5 cm microscope slide was washed for 1 min with ethanol and then rinsed with copious amounts of water. The slide was blast air-dried until all visible water was absent and then further dried in a 120 °C oven for 30 min. TiO2 was deposited on the slide by the spin coat method. Briefly, 1.0 mL of TPT was dropped onto the slide while spinning at 100 rpm. The slide was allowed to spin for 10 s and then was removed from the spin coater and allowed to react with atmospheric water until coated with white powder. The slide was annealed for 2 h at 400 °C. After cooling, the slide was rinsed with water and sonicated for 2 min. Device Fabrication. Four devices were used over the course of the experiments detailed here. Experiments were carried out in a well device, two fluidic channel devices fabricated using TiO2coated glass with channel depths of 180 and 100 µm, respectively, and finally a TiO2-coated silica bead microflow reactor. The well device was fabricated by adhering a 3-mm well, cut from a 3-mm-thick sheet of PDMS, to a TiO2-coated glass slide. In this device, the sample was pipetted directly into the well and illuminated from the top. The fluidic channel devices were fabricated by bonding two glass cover slides, either 180 or 100 µm in thickness, to a TiO2coated section of glass 3 mm apart to form the walls and bottom of the channel. The device was rinsed with acetone to remove any residual bonding agent. A top piece of TiO2-coated glass was then bonded to the cover slides to form the top of the channel. The glass was bonded using a UV-curable methylacrylic acid that was exposed to UV light for 10-25 s. The device was again rinsed with acetone and copious amounts of water and dried with a nitrogen stream for 2 min; 10 cm of 75-µm-i.d. capillary was then epoxied into each end of the device. The device was rinsed with 10× the total volume of the device with DIUF water using a syringe pump.

The microflow reactor was fabricated by first coating 114-µmdiameter silica beads (Bangs Laboratories, Inc., Fisher, IN) with TiO2. The beads were rinsed with DIUF water, filtered, and then dried for 1 h at 120 °C. TPT was pipetted onto the beads until they were submerged. The beads were then placed in a 100 °C oven along with a beaker of water (to provide a humid environment) for 1 h and shaken every 10 min to allow air movement around each bead. The beads were then annealed in a 400 °C oven for 2 h. The TiO2-coated beads were then rinsed with DIUF water, filtered, and dried for 1 h in a 150 °C oven. A 10-cm section of capillary (75-µm i.d.) was then sealed with epoxy into one end of a glass disposable pipet, and, after completely curing, the beads were placed into the pipet to a height of 1 cm. The pipet was cut just above the height of the beads and another 10-cm section of capillary (75-µm i.d.) was then sealed with epoxy into the open end of the reactor. Finally, the reactor was rinsed with 10 times the volume with DIUF water. Protein Cleavage. All well experiments were carried out by placing 20 µL of the 2 mg/mL aqueous peptide solutions in a PDMS well and illuminating by placing under a microscope with a 10× objective, total illumination area of 3 mm in diameter. The light source was a mercury arc lamp with a 360-nm bandpass filter set. The sample was place at a distance of 2 mm from the objective lens and illuminated for the specified time. The same illumination setup was used for all experiments. After illumination, the samples were extracted from the well using a pipet and diluted in a 1:9 ratio (v/v) with 3 mM pH 5.5 TAPS buffer for the capillary electrophoresis (CE) separation and detection. All samples were diluted in this manner for CE. The microfluidic channel experiments were carried out by flowing 2 mg/mL aqueous peptide solutions through the channel using a syringe pump. The flow rates of 5 and 10 µL/min corresponded to a residence time in the illuminated area of 60 and 30 s, respectively. The device was placed at a distance of 2 mm from the microscope objective as previously described. The samples were collected in a vial at the outlet of the device and diluted for CE. The microflow reactor experiments were carried out by flowing 2 mg/mL aqueous peptide solutions through the reactor using a syringe pump. The flow rates of 5 and 10 µL/min corresponded to a residence time in the illuminated area of 60 and 30 s, respectively. The device was placed at a distance of 2 mm from the microscope objective as previously described. The samples were collected in a vial at the outlet of the device and diluted for CE. Cleavage Using TiO2 Anatase Particles. Peptide samples were also fragmented in solution in a test tube using TiO2 particles. For these experiments, 20 mg of TiO2 particles was placed in a microcetrifuge filter tip. A 20-µL aliquot of 2 mg/mL angiotensin was pipetted into the TiO2 particles. The tip was placed in a vial and the TiO2 was illuminated using a 200-W light source with a 365-nm bandpass filter (Exfo Omnicure Series 2000, Frederick, MD) for the prescribed times. The light source was positioned at 1 cm above the sample, and samples were illuminated for 30 s, 2 min, 15 min, and 30 min. After illumination, the sample was centrifuged for 5 min at 13 200 rpm (90000g). Finally, 20 µL of DIUF water was added to the filter tip and the sample was again centrifuged for 8 min at 13 200 rpm. The filtered solution was Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1329

extracted and diluted with 20 µL of DIUF before capillary HPLC/ MS analysis. Capillary Electrophoresis. Capillary electrophoresis was performed on a Beckman P/ACE 5510 system with UV detection at 214 nm. A 75-µm-i.d., 37-cm-long fused-silica capillary (Polymicro Technologies, Phoenix, AZ) was used and thermostated at 30 °C. Effective length was 30 cm. Before use, the capillary was coated with PHEA (Cambrex Bio Science, Rockland, ME). The PHEA coating procedure is detailed elsewhere.37 Briefly, the capillary was rinsed with 1.0 M HCl for 15 min. A 0.1% (w/v) solution of PHEA was then rinsed through the capillary for 15 min and allowed to stand for an additional 15 min. The coating procedure was followed by a rinse with DI water for 5 min and then with running buffer for 15 min. Separations were carried out in 25 mM pH 4.5 acetate buffer at +22 kV. The sample was prepared by a 1:9 dilution in 3 mM pH 5.5 TAPS buffer to facilitate sample stacking and then introduced into the capillary by electrokinetic injection for 5 s at +10 kV. The capillary was rinsed for 5 min with running buffer between each run. Mass Analysis. Capillary HPLC/MS was accomplished using a Waters (Milford, MA) CapLC coupled to a Waters Micromass ZMD single quadruple electrospray ionization mass spectrometer. A Supelco (Bellefonte, PA) Disovery BIO Wide Pore C18 capillary HPLC column (10 cm, 0.32 mm, 3-µm particle size) was used with a 1 µL/min flow rate. A linear gradient was used to elute the peptide mixture from mobile phase A (0.1% formic acid in water) to mobile phase B (0.1% formic acid in acetonitrile). The gradient was segmented as follows: 0-15 min, 5 to 20% B; 15-75 min, 20 to 65% B; 75-85 min, 65 to 100% B; 85-90 min, 100% B; 90-95 min, 100 to 5% B. The injector syringe was washed with mobile phase A and the injection volume was set at 1 µL. For mass spectrometry, the mass region between m/z 400 and 1400 was scanned in 1.8 s. The capillary voltage was 3.5 V, and a cone ramp gradient was used from 35 to 80 V over the scanned mass range. The chromatograms and mass spectra were analyzed using Waters MassLynx software (version 3.5). RESULTS AND DISCUSSION Cleavage of Peptides. A consistent fragment pattern with each of the peptides tested indicates specificity of the peptide bond cleavage by exposure to free radicals in the cleavage zone. Figures 2 and 3 show the results from the analysis of two different peptides, Lys-bradykinin that contains three proline residues and angiotensin I that contains one proline residue. Four peaks are present in all CE separations of Lys-bradykinin exposed to TiO2 and 365-nm light in a well device (Figure 2). Peaks 1-4 increase their intensities with increasing exposure times. The observation of four peaks, and their increasing intensity, indicates the presence of four specific peptide cleavage products, supporting the hypothesis of specific cleavage at three proline sites. For angiotensin I exposed to various times in the cleavage zone, and using several different device configurations (Figure 3), at least one additional peak, indicating cleavage, is observed following all exposures. Since angiotensin I contains one proline residue, two peptides fragments were expected. In is not clear why only one significant cleavage product was observed in the electrophoretic separation. (37) Albarghouthi, M. N.; Stein, T. M.; Barron, A. E. Electrophoresis 2003, 24, 1166-75.

1330 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

Figure 2. Electrophoretic separation of the cleavage products of Lys-bradykinin cleaved in a well device with surface0bound TiO2, with each sample separately illuminated for (A) 20, (B) 30, or (C) 45 min. The control (D) was illuminated for 45 min with no TiO2 present. Numbers 1-4 indicate peaks not present in the control sample that appear in the samples illuminated with TiO2 present.

Figure 3. Electrophoretic separation of angiotensin I after illumination in three different device configurations with increasing surface area of TiO2 for decreasing residence times of (A) 4 min in a single surface coated channel, (B) 30 s in a double surface coated channel, and (C) 10 s in a microflow reactor. Elution times are reflective of different length capillaries. The fragment peak(s) discussed in the text are indicated with an asterisk (*).

However, because one expected cleavage peptide has only three amino acid residues, low limits of detection might contribute to this observation. A detectable specific cleavage pattern for angiotensin I was achieved in 4 min using a 180-µm, single-surface TiO2coated microchannel, 30 s using a 100-µm, double-surface TiO2coated microchannel, and 10 s using a TiO2-coated microbead microflow reactor (Figure 3A-C, respectively). Because of the

Figure 4. Total ion chromatogram from LC/ESI-MS analysis of angiotensin I cleaved using TiO2 particles under 15-min illumination in a well with suspended TiO2 particles. Identified peak masses are those not found in abundance in the control sample. The angiotensin I parent peak is denoted with an asterisk (*). Proposed molecular structures are shown for cleavage products generating ion signals at m/z ) 528, 403, and 409.

limited diffusion distance of hydroxyl radicals in water, the improved cleavage efficiency between the 180-µm channel depth and the 100-µm channel depth is likely due to increased OH• production from two coated walls rather than sample confinement. The efficiency improvements seen in the microflow reactor are likely due to increased surface area, increased transport caused by improved mixing efficiency, and decreased diffusion distances. It has previously been reported that hydroxyl radicals generated at the photoactive semiconductor can be found in solution at a distance of up to 2 µm from the active TiO2 surface,38 thus defining the cleavage zone. This distance is controlled by both the lifetime and diffusion of the radicals. Therefore, in these experiments complete cleavage of the peptides appeared to be limited in large part by mass transport of the peptide to the surface, which is dependent on TiO2 surface area, mixing efficiency, and diffusion distance. This is encouraging as it suggests that improvement of cleavage efficiency is indeed possible through improved surfaceto-protein contact. The time required to produce detectable cleavage of angiotensin I decreased with increased TiO2 surface area, increased mixing efficiency and decreased transport time of analyte to the TiO2 surface, and decreased diffusion distances. Though the site of peptide cleavage is not fully elucidated, there is evidence to support cleavage at the amide bond of proline, as suggested by the data presented in Figures 2 and 3. In addition to the cleavage patterns shown in these figures from the Lysbradykinin and angiotensin I, we have observed an absence of cleavage in peptides that do not contain proline. Enkephalin, a peptide without proline, was illuminated for 45 min in a well device with a TiO2-coated glass bottom, and no fragment peaks were (38) Kikuchi, Y.; Sunada, K.; Iyoda, T.; Hashimoto, K.; Fujishima, A. J. Photochem. Photobiol. A 1997, 106, 51-56.

detected by capillary electrophoresis. Even more compelling is the case of peptide-779 (P779) whose peptide sequence is identical to angiotensin I up to the proline residue where P779 terminates in arginine. P779 was illuminated in a well device for 20 min. This peptide, though nearly identical to angiotensin I, did not produce the fragment peak(s) seen previously with angiotensin I (Figure 4). A few additional low-intensity peaks were apparent after exposure; however, the peaks were not evident in the magnitude of typical fragment peaks and suggest possible minimal side chain cleavage. Mass spectrometric analysis of exposed angiotensin I using LC/ESI-MS provided further evidence of cleavage at proline. A strong signal at m/z 528 is present in all cleaved samples and not seen in controls. A protonated ion at m/z 528 is consistent with cleavage at the amide bond of proline with the addition of oxygen as proposed by Kato et al.39 This cleavage mechanism is consistent with the mechanism proposed by Kato, though Kato proposed cleavage of the amide bond involving proline’s carbonyl carbon rather than its nitrogen. These data, together with the absence of cleavage of peptides without proline residues, all indicate a model with cleavage at proline as hypothesized in the literature. Mass spectra of the separated cleavage products also show mass peaks consistently at m/z values of 403 and 409 (Figure 4). The doubly charged ion at m/z 403 corresponds to a molecular mass of 804. Cleavage at the amide bond of proline and the addition of oxygen at both cleavage sites would yield products of mass 804 and 527 (m/z 528 for the singly charged ion). The observation of an ion with m/z 409 may be explained with the addition of an oxygen bonding covalently with the carboxylate oxygens rather than remaining bound to the ring of proline. This (39) Kato, Y.; Uchida, K.; Kawakishi, S. J. Biol. Chem. 1992, 267, 23646-51.

Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

1331

Figure 5. Electrophoretic separation of myoglobin cleaved on two different well devices (sample 1) and (sample 2) vs a control sample. Illumination was for 2 h. The control sample was illuminated for 2 h without TiO2 present.

is consistent with the previously reported literature that describes oxygen’s role in the cleavage of peptide bonds.25 However, this mechanism would also yield a fragment ion at m/z 513. While an ion at m/z 513 was observed at 18.4 min, it could not be distinguished in total ion count from that observed in a dilute control sample. The m/z 513 ion was, however, observed in the control sample at 19.2 min, eluting with the angiotensin I peak and not at 18.4 min. Ions with m/z values of 337 and 627 are also evident in each sample in consistent peaks; however, these masses could not be explained by simple cleavage at proline and may be due to side chain cleavage or radical mediated polymerization. Cleavage of the peptide backbone occurs only during TiO2 illumination. To demonstrate this effect, a sample of angiotensin I was collected as the eluent from the TiO2 microflow reactor before any illumination to act as a control. A second sample of angiotensin I was collected after flowing through the TiO2 microflow reactor while it was illuminated for a residence time of 10 s. Immediately after collection of the illuminated sample, a third sample was collected after having flowed through the reactor without illumination. The control and the nonilluminated sample produced identical electropherograms, with no additional peaks (data not shown), while the illuminated sample showed an additional fragment peak (Figure 3C).

1332 Analytical Chemistry, Vol. 79, No. 4, February 15, 2007

To further investigate this cleavage technique, a more complex analyte, the protein myoglobin, was fragmented. Following exposure to illuminated TiO2, electropherograms showed the presence of at least five peaks as expected by cleavage at proline residues in the protein (Figure 5). Cleavage patterns for myoglobin were reproducible and consistent. This cleavage, as in all of the samples, was accomplished without any sample preparation, other than the dissolution of the lyophilized protein in water. Cleavage experiments were carried out in a well device, with TiO2-coated glass, with transport of protein to the surface through diffusion. The reaction was allowed to proceed for 2 h. We cannot be quantitative about cleavage efficiency in these initial experiments since we cannot accurately characterize the reaction zone geometry and kinetics due to the fleeting nature of hydroxyl radical and its poorly quantified production rates. However, it is clear that a consistent set of cleavage products is produced when proline is present in the amino acid chain. We certainly recognize that the parent peak in these initial experimental systems is large compared to the product peaks. We believe that as surface area to volume ratios are reduced and reaction conditions are optimized that the efficiency can be improved. Even without quantitative assessment, the data are compelling in that they consistently give the same products in similar ratios when proline residues are present. Further, no evidence of fragmentation occurs in the absence of proline within the amino acid sequence, without illumination or absent TiO2. Further, side chain reactions are no doubt occurring; in fact, long illumination times result in complete reaction of the peptides and evolution of gas, presumably carbon dioxide. This will be a very important issue to address in followup studies along with elucidating the core reaction mechanism of the backbone cleavage reaction. CONCLUSION The promise of this technique lies in its simplicity and photochemical generation of phenomena specific reagent. Further studies and elucidation of the fundamental mechanism of cleavage will allow optimization of the technique and permit control of cleavage through mediation of radical production, illumination intensity, and analyte residence time. We anticipate that full development of this tool will provide researchers with a facile, inexpensive, rapid, and highly tunable protein cleavage process.

Received for review July 26, 2006. Accepted December 15, 2006. AC0613737