1396
Anal. Chem. 1992, 64, 1396-1399
Sub-Femtomole Determination of Phenylthiohydantoin-Amino Acids: Capillary Electrophoresis and Thermooptical Detection Karen C. Waldron and Norman J. Dovichi* Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Twenty phenylthlohydantoln-amlno aclds (PTH-AA) are separated by mlceilar capillary electrophoresis in 13.5 mln. A low-energy krypton fluoride laser Is used In a thermooptlcai absorbance detector to produce typical detectlon limits (30) of 9 X lo-' M PTHglyclne injected In a 0.6-nL volume, correspondingto 0.5 fmol (100 fg) of PTHglyclne InJectedonto the capillary, Detectionilmltsare roughly 3 orders of magnitude superior to conventionalliquid chromatographkdetermlnatlons. Linear dynamlc range for thls ultraviolet absorbance detector Isgreater than 3000 In peak height and at least 20 000 In peak area.
The determination of the amino acid sequence of minute amounts of proteins remains fundamental in biochemistry. Current technology requires roughly 1 pmol of protein to obtain sequencinginformation. As stated by Smith in a review article, "If it were possible to obtain sequence information on (femtomole quantities of) proteins, the impact on biology and medicine would be tremendous and we would see the establishmen of new fields of research."' While mass spectrometry offers hope for sub-picomole protein sequencing,2 it is profitable to investigate improvements in current technology. Currently, protein sequence is determined by repetitive application of the Edman degradation reaction. In this reaction, an isothiocyanate is coupledto the N-terminal amino group of the protein under basic conditions to form the thiocarbamoyl derivative. The thiocarbamoyl is treated with acid to produce the cyclic thiazolinone amino acid derivative; the protein is truncated by one amino acid residue. Last, the thiazolinone is transferred to a separate conversion flask, where exposure to aqueous acid produces the stable thiohydantoin. The isothiocyanate originally used by Edman? phenylisothiocyanate (PITC), is still universally Phenylthiohydantoin- (PTH-) amino acids must be analyzed as the last step in protein sequencing by the Edman degradation reaction. Currently, the cleaved terminal PTH-amino acid is identified with liquid chromatographic separation and ultraviolet absorbance detection. It is difficult to analyze less than 0.2 pmol of PTH-amino acid with this te~hnology;~ absorbance sensitivity limits the detection limit. Alternate isothiocyanates have been proposed for highsensitivity sequencing,including (dimethy1amino)azobenzene isothiocyanate and fluorescein i s o t h i o ~ y a n a t eWhile . ~ ~ ~ these (1)Smith, L. M. Anal. Chem. 1988,60, 381A-390A. (2) Loo, J. A.; Edwards, C. G.; Smith, R. D. Science 1990,248,201-204. ( 3 ) Edman, P. Acta Chem. Scand. 1950,4, 277-283. (4) Tempst, P.; Riviere, L. Anal. Biochem. 1989, 183, 290-300.
(5) Kent, S.; Hood, L.; Aebersold, R.; Teplow, D.; Smith, L.; Farnsworth, V.; Cartier, P.; Hines, W.; Hughes, P.; Dodd, C. BioTechniques 1987,5, 314-321. (6) Jeno, P.; Chang, J.-Y. In Laboratory Methodology in Biochemistry: Amino Acid Analysis and Protein Sequencing; Fini, C., Floridi, A., Finelli, V. N., Wittman-Liebold, B., Eds.; CRC Press Inc: Boca Raton, 1990; p 63. (7) Muramoto, K.; Kamiya, H.; Kawauchi, H. Anal. Biochem. 1984, 141, 446-450.
derivatives may be detected with high sensitivity,8,9neither has found acceptance in protein sequencing because of inefficient reaction chemistry.435 To sequence smaller amounts of proteins, improvements are necessary in the determination of PTH-amino acids. Capillary electrophoresis is a useful technique for the rapid and efficient separation of small amounts of analyte.10 Typical capillary dimensions are 50-pm-i.d. and 50-cm-length. Zone electrophoresis has been used to determine phenylthiocarbamoyl- (PTC-)amino acids; ultraviolet absorbance produces detection limits of -0.2 pmol of PTC-amino acids.11 However, zone electrophoresis does not separate the neutral PTHamino acids. Otsuka et al. reported the use of micellar capillary electrophoresis for the separation of PTH-amino acids;'* distribution of the PTH-amino acids between sodium dodecyl sulfate micelles and aqueous separation buffer leads to separation. No data was provided on PTH-amino acid detection limit, but it is expected to be similar to that for the PTC-amino acids. Because of the short optical path length across the capillary, conventional absorbance detection leads to limited sensitivity.13 Thermooptical absorbance techniques are well-established methods for the determination of small absorbance.14-16 In these methods, nonradiative relaxation following absorbance of a laser beam produces a temperature rise within a sample. This temperature rise is proportional to both laser power and absorbance; a high-power laser may be used to study very weakly absorbing samples. In thermooptical methods, the temperature rise is detected as a change in the refractive index of the heated sample;the refractive index of most liquids changes by a few ppt for a '1 temperature rise. This group has developed a thermooptical absorbance detector for micrometer capillaries." A modulated pump laser beam periodically illuminates the sample at a point near the exit of the capillary. Complicated deflection and diffraction effects occur at the capillary-solution interface. Perturbation of the refractive index at the interface changes the intensity of the probe beam, measured after the capillary with a small photodiode. Phase-sensitive detection is used to demodulate the intensity change. There are two examples of ultraviolet laser pumped thermooptical detection for capillary separations. A high-energy excimer laser-pumped dye laser (A = 400 nm) was used to determine nitropyrene by capillary liquid chromatography; (8) Waldron, K. C.; Wu, S.; Earle, C. W.; Harke, H. R.; Dovichi, N. J. Electrophoresis 1990, 11, 777-780. (9) Wu, S.; Dovichi, N. J. Talanta 1992, 39, 173-178. (10) Jorgenson, J. W.; Lukacs, K. D. Science 1983,222, 266-272. (11)Rohlicek, V.; Deyl, Z. J. Chromatogr. 1989, 494, 87-89. (12) Otsuka, K.; Terabe, S.; Ando, T. J . Chromatogr. 1985,332,219226. (13) Green, J. S.; Jorgenson, J. W. J. Liq. Chromatogr. 1989,12,25272561. (14) Leite, R. C. C.; Moore, R. S.; Whinnery, J. R. Appl. Phys. Lett. 1964, 5, 141-143. (15) Long, M . E.; Swoford, R. L.; Albreicht, A. C. Science 1976,191, 183-185. (16)Dovichi, N. J. CRC Crit. Reu. Anal. Chem. 1987, 17, 357-423. (17) Bornhop, D. J.;Dovichi, N. J. Anal. Chem. 1987,59,1632-1636.
0003-2700/92/0364-1396$03.00/0 0 1992 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992 6
1397
1
Helium-Neon Laser h = 632.8 nm
Capillary
2
I I I I I I
4
6
8
10
12
14
TIME (MIN)
I I I I
I I
Reference
Lock-in Amplifier
Photodiode
Flgure 1. Thermooptical absorbance detector for capillary electro-
phoresis.
detection limits were in the mid 10-6M range.l8 Bornhop and co-workers reported the first application of ultraviolet laserbased thermooptical absorbance detection in capillary electrophoresis.lg They used a frequency-doubled argon ion laser (A = 257 nm) to determine three Dansyl-amino acids by highspeed capillary electrophoresis; detection limits were in the low lo-* M range. Both the excimer laser-pumped dye laser and the frequency-doubled argon ion laser are quite expensive and rather temperamental to operate. Neither is likely to see routine use in the analytical laboratory. -.
EXPERIMENTAL SECTION A block diagram of the thermooptical absorbance detector is shown in Figure 1. The capillary was illuminated with a 5-mW average power, 10-pJ pulse energy KrF excimer laser (Potomac Photonics Model GX-500) operating at X = 248 nm, 610-Hz pulse repetition rate, and 50-ns pulse width. The excimer laser beam was focused with a 15-mm focal length quartz biconvex lens at right angles to a 50-pm-i.d., 190-pm-0.d. fused silica capillary; the polyimide coating of the capillary was burnt from the detection region with a gentle flame. A 3-mW helium neon laser beam (Melles Griot Model 05-LHP-151) was focused at right angles to both the capillary and the excimer laser beam with a 7 X microscope objective. The probe beam intensity change was detected 30 cm after the capillary with a 1-mm2silicon photodiode. The photodiode output was conditioned with a currentto-voltage converter (1MQ feedback resistor in parallel with a 47 pF capacitor) and sent to a two-phase lock-in amplifier (Ithaco Model 3961), phase referenced to the excimer laser pulse repetition rate. An 80386 PC collected data from the lock-in amplifier over the IEEE-488bus. Sneakernet was used to transfer the data to a Macintosh IIsi computer. Matlab was used to convolute the data with a Gaussian-shaped filter. A 39-cm-longpolyimide coated fused silica capillary was used for this separation;the distance from the injector to the detector was 34 cm. The separationproceeded in a 12.5mM pH 7.0 borate/ phosphate buffer that contained 35 mM sodium dodecyl sulfate. Electrokinetic injection of 5 s at 500 V was used. Separation proceeded at 8 kV. (18)Kettler, C.N.;Sepaniak, M. J. Anal. Chem. 1987,59,1733-1736. (19)Bruno, A. E.;Paulus, A.; Bornhop, D. J. Appl. Spectrosc. 1991, 45,462-467.
Figure 2. High-speed micellar capillary electrophoresis separation of PTHaminoacids. A 50-pm-i.d., 39-cm-long capillary was used for the separation; the distance from the injector to the detector was 34 cm. The separation proceeded at 8 kV. Injection was for 5 s at 500 V. A 12.5 mM, pH 7.0 borate/phospohatebuffer, containing35 mM sodium dodecyl sulfate, was used for the separation. Threonine produced a small peak that overlapped asparagine.
PTH-aminoacids were purchased from Sigma. Stock solutions of lod2M concentration were prepared by dissolving each amino acid in a 50% acetonitrile and 50% pH 7.5 mM phosphate/borate buffer. Mixtures of amino acids were prepared by pipetting 10-pLaliquots into 1.5 mL of 12.5 mM pH 7.0 borate/phosphate buffer that contained 35 mM sodium dodecyl sulfate.
RESULTS Figure 2 presents the separation of 20 PTH-amino acids by micellar capillary electrophoresis and thermooptical absorbance detection. A 1.1-mV base-line signal is due to absorbance of the excimer laser beam by trace impurities in the separation buffer. The disturbance at 4.5 min is due to the elution of trace amounts of acetonitrile, added to the analyte to effect dissolution. Acetonitrile produces a refractive index change that perturbs the optical alignment, producing the base-line disturbance. The PTH-amino acids are nearly base line resolved, with slight overlap of alanine and glutamic acid, and histidine and tryptophan. The separation takes 13.5 min at 8 kV. Separation efficiencyin capillary electrophoresis improves with higher potentials. Unfortunately, higher electric field produces significant Joule heating, which generates bubbles in this buffer and destroys the separation. A longer capillary, operated at higher potential (15 kV) but the same electric field (215 V/cm) as the previous figure, produces base-line separation of the PTH-amino acids in 40 min. Alternatively, urea or another additive could be used to improve the separation without a penalty in speed. As noted above, the waveguide excimer laser has nearly ideal wavelength, pulse repetition rate, beam spatial coherence, and pulse energy for thermooptical detection of PTHamino acids in capillary electrophoresis. At the laser wavelength, 248 nm, the molar absorptivity of PTH-glycine is 6700 L mol-1 cm-1,60% of the maximum absorptivity a t 268 nm. The pulse repetition rate, 610 Hz, is well matched to the thermal relaxation time for aqueous solvent in a 50-pm capillary. The time-resolved thermooptical signal, Figure 3, is nearly sawtooth in shape; most of the signal is generated at the fundamental modulation frequency. As a result, lockin amplification is quite useful in signal demodulation. Because of the waveguide design of the laser, the beam has unusually high spatial coherence for an excimer laser and may be focused easily to a spot size less than the capillary
1398
ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992
Table I.
Detection Limits for PTH-Amino Acidss
amino acid
Time (mS) Figure 3. Time resolved thermooptical signal generated by a pulsed excimer laser. Two cycles of the waveform are shown.
dimensions. Although the thermooptical signal increases with pump laser power, a modest laser power is appropriate for this experiment; higher energy beams decompose the sample, boil the solution, or destroy the capillary, leading to highly nonlinear response.ls The peak power for this laser, 200 W, is quite low for an excimer laser because of the modest pulse energy, 10 J , and long pulse width, 50 ns. The time-resolved signal of Figure 3 presents a measure of the thermal relaxation rate of solutions in a capillary. The shape of the waveform is dominated by heat flow and not electronic response time. Heat flow from the solution to the capillary is fast and must be accounted for in any model of the temperature rise induced in zone electrophoresis. The origin of the 200-ps rise time is speculative, but may reflect heat flow from the solution to the capillary wall. As we have shown elsewhere, a parabolic temperature rise is created along the pump beam axis by heat flow to cuvette windows.20 This temperature rise, and its associated refractive index gradient, acts to defocus the probe beam in the plane containing the pump laser beam. Detection limit is the amount of analyte that generates a signal 3 times larger than the uncertainty of the base-line signal. Data in Figure 2 were recorded at 0.d-s intervals from a lock-in amplifier that operated with 0.3-s time constant. Peaks are Gaussian in shape with a typical full-width a t halfheight of 2.5 s. Optimal processing of the data was achieved by convoluting the data with a Gaussian filter function with 1-s half-width; noise is reduced and separation efficiency is increased by less than 10%. It is not appropriate to use the standard deviation in the background signal as an estimate of noise; the filter function ensures that successive data are correlated. Instead, detection limits are estimated by the maximum deviation from the mean base-line signal, measured over a time period given by 50 times the peak width.*I Detection limits for the 20 PTH-amino acids, Table I, range from 0.2 fmol for tryptophan and lysine to 5 fmol for threonine, injected onto the capillary. Poor sensitivity for threonine is due to decomposition of the PTH derivative; improved chemistry will bring the detection limit of all the PTH-amino acids to the sub-femtomole range. These detection limits are 2-3 orders of magnitude superior to those produced by liquid chromatography and UV absorbance detectiom2 Concentration detection limits, less than 10-6 M, are more than 2 orders of magnitude superior to the other UV thermooptical absorbance detector in capillary electroph~resis.~gThe use of a high-repetition rate excimer laser produces excellent performance. One millivolt of thermooptical signal equals 10-4 absorbance units. The base-line signal corresponds to an absorbance of 10-4 measured across the 50-pm-diameter capillary. (20)Wu, S.; Dovichi, N. J. J. Appl. Phys. 1990, 67, 1170-1182. (21) Knoll, J. E. J . Chromatogr. Sei. 1986, 23, 422-425.
alanine (A) arginine (R) asparagine (N) aspartic acid (D) cysteic acid iC) serine (S) glutamic acid (E) glutamine (Q) glycine (G) histidine (H) isoleucine (I) leucine (L) PTH 6-PTC lysine (K) methionine (M) phenylalanine (F) proline (P) threonine (T) tyrosine (Y) tryptophan (W) valine (V)
mol injected, fmol
vol injected, nL
detection limit, fmol
detection limit, WM
66 22 68 58 51 73 65 65 68 30 27 43 17
0.55 0.26 0.68 0.53 0.52 0.66 0.54 0.59 0.62 0.33 0.34 0.33 0.28
0.5 0.6 0.7 08 2 2 0.7 0.7 0.5 0.3 0.7 0.3 0.2
1.0 0.9 1.0 1.4 3 3 1.2 1.1 0.9 0.8 2 0.8 0.6
34 29 41 48 42 24 41
0.40 0.31 0.41 0.44 0.46 0.33 0.41
0.4 0.3 0.4 5 0.3 0.2 0.4
1.0 1.0 1.0 12 0.7 0.5 0.8
a Detection limit is defined as that amount of analyte that produces a peak 3 times larger than the noise in the background signal.
The absorbance per unit length of the electrophoresis buffer cm-', a factor of 4 greater than the (decadic) abis 2 X sorbance per unit length of (6 f 2) X cm-' for water a t 254 nm quoted by Roivin et a1.22 Absorbance due to buffer components, and impurities in those components, leads to the observed background signal. The relative noise on the base line, - O B % , presumably is proportional to fluctuations in the pump laser pulse energy. The absorbance detection limit (3a) for the thermooptical detector is 2.5 X lo4. This detection limit is 1-2 orders of magnitude superior to that produced by commercial transmission detectors in capillary electrophoresis with 50-pm-diameter capillaries. Improved detection limits may be produced for the thermooptical detector by use of higher purity reagents, more stable pump laser, and solvent with a larger change in refractive index with temperature. While heavy water (DzO)has greater thermooptical sensitivity than normal water, it has a factor of 2 greater background absorbance and is not useful as a solvent for high-sensitivity ultraviolet absorbance detectors. Mixed acetonitrile-water solvents have proven valuable in capillary zone electrophoresis separationand thermooptical absorbance detection of (dimethy1amino)azobenzenesulfonyl chlorideamino acids.23 The dynamic range of the system was evaluated by injecting a series of PTH-Gly samples. Improved optical alignment gave better sensitivity than in Figure 2; detection limits were 0.2 fmol injected. The peak height increased linearly ( r > 0.997, n = 11)from the detection limit to 800 fmol injected; linear dynamic range of peak height extends a factor of 3500 in analyte concentration. At higher concentrations,the peak showed pronounced tailing, presumably due to distortions in the local electric field due to the ionic strength of the analyte.Z4 However, peak area is conserved and is linearly related to analyte concentration from the detection limit to a t least 5 pmol, a 20 000 range in analyte amount. Precision was investigated by repeatedly injecting the same sample; the relative standard deviation in peak height was (22) Boivin, L. P.; Davidson, W. F.; Storey, R. S.; Sinclair, D.; Earle, E. D. A p p l . Opt. 1986, 25, 877-882. (23) Yu, M.; Dovichi, N. J. Anal. Chem. 1989, 61, 37-40. (24) Mikkers, F. E. P.; Everaertz, F. M.; Verheggen, T. P. E. M. J . Chromatogr. 1979, 169, 1--10.
ANALYTICAL CHEMISTRY, VOL. 64, NO. 13, JULY 1, 1992
2.2% for the electrokinetic injection. To obtain this high level of precision, care was taken to ensure that the sample and terminal buffer containers were at the same level. However,signal amplitude is highly dependent on alignment conditions. Replacement of the capillary, and subsequent realignment, inevitably produces a significant change in sensitivity. However, the capillaries were very stable in this separation; in favorable cases, a capillary would be used for several months before it was replaced. In fact, replacement of the capillary was required only due to breakage, which usually occurred when the fragile detector window was stressed. Retention time precision, which is of great importance in sequencing,was 1.25%. Both retention time and peak height precision are dominated by reproducibility in adjustment of the power supply voltage for both separation and injection.
1399
capillary electrophoresis separation will allow routine sequencing of femtomole amounts of proteins. A number of groups, including this one, are developing miniaturized sequencers.25 The development of this high-mass sensitivity detector will facilitate the development of femtomole protein sequencers. Detection is an important issue in capillary electrophoresisagExcept for fluorescence,26most detectors provide limited sensitivity and dynamic range. There is a need for sensitive, universal detection with high dynamic range in capillary electrophoresis. The thermooptical absorbance detector comes close to satisfying this need; any molecule containing an aromatic group will absorb strongly at this excimer laser wavelength. Additional applications of micellarcapillary electrophoresis with thermooptical absorbance detection include analysis of protein mixtures, oligonucleotides,or halogenated hydrocarbons of environmental intere~t.~'-2Q
CONCLUSION Micellarcapillary zone electrophoresisseparation and thermooptical absorbance detection produces outstanding mass detection limits and wide linear dynamic range for analysis of PTH-amino acids. T o obtain excellent separation efficiency, the analyte must be injected in nanoliter volumes. As a result, interface with conventionalsequencersis problematic. The sample might be concentrated before injection by use of solid-phase extraction or isotachophoresis. Instead, a highly miniaturized sequencer may be developed that is matched to the volume of the capillary electrophoresisseparation method. Of course, there are significant experimental difficulties in the design of a miniaturized sequencer. However, because the same reagent concentration is used for the conventional and miniaturized sequencers, the reaction rate and efficiency for the miniaturized sequencer should be identical to that of the conventional scale sequencer. Improvements in mass detection limit arise only because the system volume is decreased. Combination of the miniaturized sequencer with
ACKNOWLEDGMENT This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada and unrestricted grants from Pharmacia/LKB and Waters Division of Millipore, Inc. N.J.D. acknowledges a Steacie Fellowship from the Natural Sciences and Engineering Research Council. RECEIVED for review December 16, 1991. Accepted March 6,1992. (25) Liang, S. P.; Laursen, R. A. Anal. Biochem. 1990,188,366373. (26) Cheng, Y.F.; Dovichi, N. J. Science 1988,242, 562-564. (27) Cohen, A. S.; Terabe, S.; Smith, J. A.; Karger, B. L. Anal. Chem. 1987,59, 1021-1027. (28) Liu, J.; Cobb, K. A.; Novotny, M. J. Chromatogr. 1990,519,18+ 197. (29) Terabe, S.; Otauka, K.; Ichikawa, K.; Tsuchiya, A.; Ando,T. Anal. Chem. 1984,56, 111-113.
