Temperature Gradient Focusing with Field-Amplified Continuous

Matthew S. Munson,* Gre´ goire Danger, Jonathan G. Shackman, and David Ross. Biochemical Sciences Division, National Institute of Standards and ...
0 downloads 0 Views 264KB Size
Anal. Chem. 2007, 79, 6201-6207

Temperature Gradient Focusing with Field-Amplified Continuous Sample Injection for Dual-Stage Analyte Enrichment and Separation Matthew S. Munson,* Gre´goire Danger, Jonathan G. Shackman, and David Ross

Biochemical Sciences Division, National Institute of Standards and Technology, 100 Bureau Drive, MS 8313, Gaithersburg, Maryland 20899

We describe the serial combination of temperature gradient focusing (TGF) and field-amplified continuous sample injection (FACSI) for improved analyte enrichment and electrophoretic separation. TGF is a counterflow equilibrium gradient method for the simultaneous concentration and separation of analytes. When TGF is implemented with a low conductivity sample buffer and a (relatively) high conductivity separation buffer, a form of sample enrichment similar to field-amplified sample stacking (FASS) or field-amplified sample injection (FASI) is achieved in addition to the normal TGF sample enrichment. FACSI-TGF differs from FASI in two important respects: continuous sample injection, versus a discrete injection, is utilized; because of the counterflow employed for TGF, the stacking interface exists in a pseudo-stationary region outside of the separation column. Notably, analyte concentration enrichment factors greater than the ratio of separation and sample conductivities (γ) were achieved in this method. For γ ) 6.1, the concentration factor for one model analyte (Oregon Green 488) was found to be 36-fold higher with FACSI-TGF as compared to TGF without FACSI. A separation of five fluorescently labeled amino acids is also demonstrated with the technique, yielding an average enrichment of greater than 1000-fold. The recent expansion of interest in the field of micro total analysis systems (µTAS) is a result of the benefits promised by miniaturization. It is often claimed that miniaturization will lead to the reduction in sample volumes, acceleration of diffusionlimited processes, more efficient heat transfer, and ease of integration and automation. With respect to analytical techniques, many of these advantages have been realized.1-5 However, because of the length scales employed in microfluidic systems, * Author to whom correspondence should be addressed. Phone: 301-9754125. Fax: 307-975-8246. E-mail: [email protected]. (1) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (2) Dittrich, P. S.; Tachikawa, K.; Manz, A. Anal. Chem. 2006, 78, 3887-3907. (3) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623-2636. (4) Weibel, D. B.; Whitesides, G. M. Curr. Opin. Chem. Biol. 2006, 10, 584591. (5) Yi, C. Q.; Li, C. W.; Ji, S. L.; Yang, M. S. Anal. Chim. Acta 2006, 560, 1-23. 10.1021/ac070689r CCC: $37.00 Published on Web 07/07/2007

© 2007 American Chemical Society

there is still much room for improvement with respect to limits of detection (LOD) when optically based detection schemes are utilized.2,6 The approaches to solving this problem can be classified as either improvements to the detection method (i.e., improvement of instrumentation, design of novel flow cells, and the use of detectors that are not optically based)7-9 or through coupling of analyte enrichment techniques to the analysis (e.g., sample preconcentration,10-12 gradient focusing methods,13-18 etc.). Field-amplified sample stacking (FASS) and field-amplified sample injection (FASI) are two of the most commonly used methods for the preconcentration of analytes prior to separation by capillary electrophoresis (CE).19-27 FASS and FASI are typically implemented by preparing the sample in a lower conductivity buffer than the run buffer. A portion (ranging from 1% to 90% of the total length, depending on the method) of the separation channel is filled with the low conductivity sample solution, and the remainder of the capillary is filled with the high conductivity run buffer forming an interface between the high and low conductivity solutions inside the capillary. Because of the con(6) Effenhauser, C. S.; Bruin, G. J. M.; Paulus, A.; Ehrat, M. Anal. Chem. 1997, 69, 3451-3457. (7) Chervet, J. P.; Vansoest, R. E. J.; Ursem, M. J. Chromatogr. 1991, 543, 439-449. (8) Mogensen, K. B.; Klank, H.; Kutter, J. P. Electrophoresis 2004, 25, 34983512. (9) Zarrin, F.; Dovichi, N. J. Anal. Chem. 1985, 57, 2690-2692. (10) Lin, C. H.; Kaneta, T. Electrophoresis 2004, 25, 4058-4073. (11) Sentellas, S.; Puignou, L.; Galceran, M. T. J. Sep. Sci. 2002, 25, 975-987. (12) Song, S.; Singh, A. K. Anal. Bioanal. Chem. 2006, 384, 41-43. (13) Kelly, R. T.; Li, Y.; Woolley, A. T. Anal. Chem. 2006, 78, 2565-2570. (14) Liu, J. K.; Sun, X. F.; Farnsworth, P. B.; Lee, M. L. Anal. Chem. 2006, 78, 4654-4662. (15) Petsev, D. N.; Lopez, G. P.; Ivory, C. F.; Sibbett, S. S. Lab Chip 2005, 5, 587-597. (16) Shackman, J. G.; Munson, M. S.; Ross, D. Anal. Bioanal. Chem. 2007, 387, 155-158. (17) Shackman, J. G.; Ross, D. Electrophoresis 2007, 28, 556-571. (18) Huang, Z.; Ivory, C. F. Anal. Chem. 1999, 71, 1628-1632. (19) Mala, Z.; Krivankova, L.; Gebauer, P.; Bocek, P. Electrophoresis 2007, 28, 243-253. (20) Gong, M. J.; Wehmeyer, K. R.; Limbach, P. A.; Arias, F.; Heineman, W. R. Anal. Chem. 2006, 78, 3730-3737. (21) Yang, H.; Chien, R. L. J. Chromatogr. A 2001, 924, 155-163. (22) Lichtenberg, J.; Verpoorte, E.; de Rooij, N. F. Electrophoresis 2001, 22, 258271. (23) Chien, R. L.; Burgi, D. S. Anal. Chem. 1992, 64, A489-A496. (24) Chien, R. L.; Burgi, D. S. J. Chromatogr. 1991, 559, 141-152. (25) Burgi, D. S.; Chien, R. L. Anal. Chem. 1991, 63, 2042-2047. (26) Breadmore, M. C. Electrophoresis 2007, 28, 254-281. (27) Chien, R. L. Electrophoresis 2003, 24, 486-497.

Analytical Chemistry, Vol. 79, No. 16, August 15, 2007 6201

ductivity differences, the electric field strength, and therefore electrophoretic velocities of the analytes, are higher in the low conductivity zone than in the high conductivity zone. As the analyte ions migrate from the low conductivity zone to the high conductivity zone, their velocity is reduced, causing them to be “stacked up” into a narrower, more concentrated band. Neglecting the effects of diffusion and differences in electroosmotic flow in the two zones, the concentration enhancement from stacking is given by:

γ≡

σ2 C2 ) σ1 C1

(1)

where C1 and C2 are the analyte concentrations before and after the stacking, respectively, σ1 and σ2 are the conductivities of the low and high conductivity buffers, respectively, and γ is the conductivity ratio.25 In addition to the stacking methods such as FASS and FASI, there are a variety of methods for analyte focusing that have been demonstrated for the concentration of analytes. Examples include isotachophoresis (ITP), isoelectric focusing (IEF), and electric field gradient focusing (EFGF).10,11,19 Stacking and focusing techniques both lead to an increase in analyte concentration based on an axial variation in analyte velocity; however, it is important to differentiate the two techniques. The primary difference between the two lies in how analytes are transported to/through the region where concentration occurs. With stacking techniques, the concentration occurs at a boundary where there is a change in the magnitude of the analyte velocity but no change in direction. As such, the concentration enhancement is limited to the velocity ratio as described in eq 1. However, with focusing methods, there is a change in direction of the analyte velocity so that analytes move toward the accumulation point from both directions. Consequently, the analyte will continue to accumulate at that point as long as the focusing field is maintained or until the supply of analyte is exhausted. The concentration enhancement in focusing techniques is limited only by the rate at which analyte is transported to the focusing zone and the duration of the application of the focusing field. Temperature gradient focusing (TGF) is a recently developed equilibrium gradient focusing technique that has been demonstrated for the simultaneous separation and concentration of chemical analytes.16,28-35 In TGF, electrophoresis is conducted in the opposite direction of a bulk flow (i.e., under counterflow) generated by a combination of electroosmotic flow and pressure driven flow. Because of continuity in the fluid phase, the bulk (28) Kamande, M. W.; Ross, D.; Locascio, L. E.; Lowry, M.; Warner, I. M. Anal. Chem. 2007, 79, 1791-1796. (29) Shackman, J. G.; Munson, M. S.; Kan, C. W.; Ross, D. Electrophoresis 2006, 27, 3420-3427. (30) Kim, S. M.; Sommer, G. J.; Burns, M. A.; Hasselbrink, E. F. Anal. Chem. 2006, 78, 8028-8035. (31) Hoebel, S. J.; Balss, K. M.; Jones, B. J.; Malliaris, C. D.; Munson, M. S.; Vreeland, W. N.; Ross, D. Anal. Chem. 2006, 78, 7186-7190. (32) Balss, K. M.; Vreeland, W. N.; Phinney, K. W.; Ross, D. Anal. Chem. 2004, 76, 7243-7249. (33) Balss, K. M.; Vreeland, W. N.; Howell, P. B.; Henry, A. C.; Ross, D. J. Am. Chem. Soc. 2004, 126, 1936-1937. (34) Balss, K. M.; Ross, D.; Begley, H. C.; Olsen, K. G.; Tarlov, M. J. J. Am. Chem. Soc. 2004, 126, 13474-13479. (35) Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 2556-2564.

6202

Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

counterflow is constant along the length of the separation channel. A background electrolyte which exhibits a temperature-dependent ionic strength is used. A temperature gradient along the length of the separation channel is applied to induce a gradient in the ionic strength of the separation buffer,29,35 leading to a gradient in the electrophoretic velocity of the analytes along the length of the separation channel. Separations are conducted so that at the inlet end of the separation channel, electromigration of the analyte into the channel is faster than convection out of the channel (i.e., constant sample injection with net analyte flux into the channel), while at the outlet end of the separation channel, convection is greater than electromigration. The result is that the analyte accumulates at some point in the separation channel determined by its electrophoretic mobility. TGF has been used for the concentration and separation of DNA, peptides, small molecules, proteins, beads, and cells.35 In this paper we present results for the combination of fieldamplified continuous sample injection (FACSI) and TGF. As with conventional FASI, this new method is implemented simply by preparing the sample in a lower conductivity buffer than the run buffer. However, because of the bulk counterflow required for TGF, the interface between the low and high conductivity buffers is located outside the separation channel. Our results suggest that because of the reduction in the magnitude of the convective velocities outside the capillary, the conductivity interface is stabilized by diffusion and the field-amplified injection is changed from a stacking phenomenon to a focusing phenomenon. As a result, concentration enhancements greater than the conductivity ratio, γ, are obtained. Additionally, the continuous nature of the sample introduction allows for further sample enrichment in the TGF stage of the analysis. Key operational parameters and advantages of the FACSI-TGF technique are explored, and it is demonstrated for the concentration and separation of a mixture of amino acids. EXPERIMENTAL SECTION36 Chemicals and Reagents. A fused silica capillary column (30 µm i.d.; 360 µm o.d.) was obtained from Polymicro Technologies, LLC (Phoenix, AZ). Polycarbonate (PC) sheets were from McMaster-Carr (Atlanta, GA). Oregon Green 488 carboxylic acid (OG488) and 5-carboxyfluorescein, succinimidyl ester (5-FAM, SE) were purchased from Invitrogen (Carlsbad, CA). All solutions were made from Milli-Q (Millipore, Bedford, ME) g 18 MΩ cm deionized water. Tris(hydroxymethyl)aminomethane (Tris) and boric acid were purchased from Amresco (Solon, OH). All other chemicals were obtained from Sigma-Aldrich (St. Louis, MO) and were of the highest purity available. All TGF buffers were made from dilution of 2 mol L-1 Tris, 2 mol L-1 boric acid (TB) stock solution. OG488 samples were prepared from a 6 mmol L-1 stock solution of dye in 20 mmol L-1 TB buffer. Stock solutions of amino acids (aspartic acid, glutamic acid, serine, valine, and glycine) were made at 1 mmol L-1 in 15 mmol L-1 sodium borate buffer (pH 9.2). A stock solution of 0.1 mol L-1 5-FAM, SE in dimethyl sulfoxide (DMSO) was made. For (36) Certain commercial equipment, instruments, or materials are identified in this report to adequately specify the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.

Figure 1. Instrumentation and theory of FACSI-TGF. (a) Photograph of experimental apparatus. The capillary/polycarbonate composite chip is clamped by two copper blocks to impose the temperature gradient. The hot block is heated using a PID-controlled thermoelectric heater while the cold block is chilled using a circulating water bath. The Delrin sample reservoir is open to ambient pressure. The polysulfone run buffer reservoir is sealed and connected to a pressure controller for precise control of the counterflow. High voltage is applied to the reservoirs using platinum electrodes. Detection is via fluorescence microscopy. (b) Schematic view of the sample reservoir. The end of the capillary protrudes into the sample reservoir. (c) The conductivity interface generated by the counterflow is visualized by adding rhodamine B to the run buffer and imaging sample cup (voltage: 3 kV, pressure 0 Pa). The interface is observed to be the located outside the capillary and is pseudo-stationary. (d, e) False color plots of focusing on the conductivity interface at two different applied pressures (d: 5000 Pa, e: 1000 Pa). Images are the pixel-by-pixel difference of the intensity with the voltage on minus the intensity with the voltage off.

labeling, 10 µL of dye stock was added to 990 µL of amino acid stock and incubated at room temperature for 12 h in the dark. Following incubation, labeled amino acid solutions were stored at 5 °C until used. These solutions were used without removal of free dye as stock for separation experiments. Each amino acid was labeled separately to allow for peak identification. The degree of labeling was not characterized. Instrumentation. Experiments utilized an apparatus previously described for use in scanning TGF31 (Figure 1a). Briefly, a ca. 7 mm wide optical window was burned into a 3 cm long capillary prior to enclosure between two PC sheets; the sandwich assembly was then placed in a hydraulic press at 500 kg, heated to 200 °C, and then cooled to 120 °C prior to releasing pressure. This process ensured good mechanical stability and thermal conductivity of the exposed silica region. The device was then placed in a holding apparatus consisting of two copper anchor blocks, maintained at disparate temperatures. The temperature of the hot anchor was maintained through proportional-integralderivative (PID) controlled thermoelectric heaters (TE-Tech, Traverse City, MI) while the temperature of the cold anchor was maintained using a circulating water bath. One capillary end was inserted into a 360 µm diameter hole drilled into a Delrin sample reservoir containing 100 µL while the other end was inserted through a Teflon backed silicone septum into a polysulfone run buffer reservoir containing 1400 µL of buffer connected to a precision pressure controller (Series 600, Mensor, San Marcos, TX) with a range of ( 13.8 kPa. The capillary was conditioned with 1 mol L-1 sodium hydroxide for 15 min followed by deionized water prior to the first use and thoroughly flushed with running buffer between analyses. Capillary experiments were performed on a Leica DMLB fluorescent microscope (Leica Microsystems, Bannockburn, IL) equipped with a long-working distance 10× objective, 100 W Hg arc lamp, 12-bit color CCD camera (CFW-1312C, Scion Corporation, Fredrick, MD), and appropriate fluorescence filter sets. All

instrument control and data acquisition was performed using Java 5.0 software (Sun Microsystems, Santa Clara, CA) written inhouse. The detection spot (25 pixels long × 10 pixels wide) was located 500 pixels from the hot edge of the temperature gradient. The camera exposure time was 0.75 s at a gain of 5 dB. Buffer conductivity as a function of buffer concentration was measured using a circulating water bath and Traceable meter (Control Company, Friendswood, TX) with the temperature coefficient set to zero. All conductivity measurements were made at a temperature of 20.8 °C. Separation Conditions. Characterization of FACSI-TGF was performed by constructing calibration curves of peak height as a function of analyte (OG488) concentration using the scanning TGF procedure described previously.31 The separation buffer used for all experiments was 500 mmol L-1 TB, while sample buffers varied from 31 mmol L-1 to 1 mol L-1 TB. OG488 concentrations varied from 31 pmol L-1 to 64 nmol L-1. A temperature gradient of 30 °C mm-1 (80 °C to 20 °C) was applied across a 2 mm segment of a 30 mm long capillary. A voltage of -3 kV was applied at the sample reservoir (run buffer reservoir grounded), giving an approximate field strength of -1 kV cm-1. During each experiment the pressure was held for 30 s with the voltage applied at a pressure of +250 Pa. The pressure was then reduced to -375 Pa in -5 Pa increments and held for 2 s at each pressure. Amino acid separations were performed as follows. Each amino acid was diluted to 1 nmol L-1 in either 500 mmol L-1 TB (TGF only) or 31 mmol L-1 TB (FACSI-TGF). A temperature gradient of 2.7 °C mm-1 (30 °C to 3 °C) was applied across a 10 mm segment of a 40 mm long capillary. A voltage of -4 kV was applied at the sample reservoir (run buffer reservoir grounded), giving an approximate field strength of -1 kV cm-1. During each experiment the pressure was held for 30 s with the voltage on at a pressure of +1200 Pa. The pressure was then reduced to -100 Pa in -1 Pa increments and held for 2 s at each pressure. Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

6203

Excel 2003 (Microsoft, Redmond, WA) was used to determine peak heights by subtracting the average baseline intensity from the peak maximum. The ‘LINEST’ function was used to determine the slope and standard error of the peak height vs analyte concentration data. Nonlinear regression of the slope vs conductivity ratio data was performed using Origin 7.5 (OriginLab Corp., Northampton, MA). RESULTS AND DISCUSSION In scanning TGF,31 the counterflow of separation buffer from the run buffer reservoir toward the sample is controlled by the applied pressure at the run buffer reservoir. The applied pressure, and therefore the magnitude of the counterflow, is slowly reduced until the analytes of interest are able to elute onto the capillary and focus along the temperature gradient. As the pressure is further reduced, the focused peaks move from the inlet end of the gradient toward the outlet end. The fluorescence intensity is then measured at a fixed detection spot along the gradient and plotted as a function of the applied pressure to give a result similar to a conventional electropherogram. The implementation of FACSI in combination with TGF was used as a method for the simultaneous separation and dual-stage enrichment of charged analytes. The magnitude of the concentration factor achieved was greater in FACSI-TGF than in an otherwise equivalent TGF experiment and was achieved simply by constituting the sample in a lower conductivity buffer than the separation buffer. Under the counterflow conditions, with net fluid flux moving from the run buffer to sample reservoirs, the high conductivity buffer exits the capillary and diffuses into the low conductivity sample reservoir, creating a conductivity interface outside of the capillary (Figure 1C). Run buffer containing rhodamine B as a fluorescent tracer was observed to form a stable zone of high conductivity fluid where the separation channel protrudes into the sample reservoir. The lower conductivity sample (visualized with OG488) was excluded from this region. The conductivity interface was observed to remain outside the capillary over a relevant range of applied pressures. We term the interface pseudo-stationary, as while it remains outside of the capillary, some slight movement occurs as the counterflow was varied. The interface only entered the capillary under flow reversal conditions at extremely low pressures, which were not utilized. Upon application of voltage, at a relatively high applied pressure, the magnitude of the bulk flow velocity was greater than the magnitude of the electrophoretic velocity inside the capillary and near the capillary entrance. Consequently, the analyte did not enter the capillary. Nevertheless, the analyte moved toward the conductivity interface and capillary entrance where it was observed to form a high concentration band on the hemispherical surface of the conductivity gradient outside the capillary as shown in Figure 1D. At constant applied pressure, the high concentration band was observed to remain stationary and stable for at least several minutes, suggesting that the analytes were being focused on the conductivity gradient rather than stacked. As the applied pressure was reduced, the focused band became more intense and moved toward the capillary entrance (Figure 1E). Upon further reduction of the applied pressure, the focused analyte band passes into the capillary where it is subsequently focused on the temperature gradient and detected. 6204 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

Figure 2. Conductivity ratio (γ) as a function of TB buffer concentration. The conductivity is normalized to the separation buffer (0.5 mol L-1 TB, 1699 µS cm-1). All measurements were made at a probe temperature of 20.8 °C. Error bars are (1 standard deviation (n ) 3).

Conductivity as a Function of Concentration. The variation of conductivity with buffer strength for TB buffers is shown in Figure 2 with data shown normalized to the conductivity of the separation buffer (0.5 mol L-1 TB), which had a conductivity of (1699 ( 1) µS cm-1 (standard deviation, n ) 3). The conductivity ratio (γ) was shown to vary from a value of 6.1 for 31 mmol L-1 TB to 0.78 for 1 mol L-1 TB. The nonlinearity of the conductivity/ buffer concentration relationship is likely due to many factors, such as variations in viscosity, self-association of borate ions, and the relatively weak ionization of the buffer species. Investigation of these phenomena is outside the scope of this report. Impact of Starting Pressure. As previously described for scanning TGF,31 to achieve the maximum degree of analyte concentration before a focused peak passes the detection spot, it must be swept across the entire range of the focusing gradient. Consequently, the initial applied pressure (Pstart) for a scanning TGF analysis must be high enough to prevent the fastest analyte of interest from entering the capillary or microchannel. With FACSI-TGF, analytes are focused in two stages, first on the conductivity gradient outside the capillary and then on the temperature gradient inside the capillary. To determine the starting pressure necessary for maximum focusing, a series of scanning TGF measurements with different starting pressures were conducted. Figure 3 shows the measured peak height for OG488 as a function of the difference between Pstart and the applied pressure at the peak maximum (Pmax) for both normal TGF and FACSI-TGF. For relatively low start pressures, the peak was only focused for a short time before it reached the detector, and the peak heights were relatively small. As the start pressure was increased, the peak height increased until it reached a plateau value (peak height no longer increased with increasing Pstart). This indicates that the focused peak is swept across the entire focusing gradient, giving the maximal amount of focusing for a given set of experimental conditions. For normal TGF (γ ) 1), it was observed that the peak height reached the plateau value when the start pressure was approximately 300 Pa greater than the focusing pressure. With FACSI-TGF (γ ) 6.1), the plateau was not reached until 600 Pa. The additional concen-

Figure 3. Effect of starting pressure on the peak heights for TGF and FACSI-TGF. As the pressure difference between the starting pressure (Pstart) and the focusing pressure (Pmax) was increased, the peak height eventually reaches a plateau for both methods. Sample was 1 nmol L-1 OG488 in either 0.5 mol L-1 (TGF; γ ) 1; O) or 31 mmol L-1 (FACSI-TGF; γ ) 6.1; 0) TB; separation buffer was 0.5 mol L-1 TB; temperature gradient was 30 °C mm-1; field was -1 kV cm-1; applied pressure was reduced in 5 Pa increments and held 2 s at each pressure.

tration factor due to FACSI is equal to the ratio of the peak height for FACSI-TGF to that for normal TGF. For the data of Figure 3, that ratio varied from approximately 12 for low start pressure to approximately 30 for Pstart - Pmax g 600 Pa. In all cases the concentration factor due to FACSI was greater than the conductivity ratio, γ. The two curves shown in Figure 3 were measured with the same capillary for direct comparison. However, the start pressure required to reach the plateau value varied from capillary to capillary. For FACSI-TGF with γ ) 6.1, the value of Pstart - Pmax at the onset of the plateau ranged from approximately 400 to 1000 Pa. The plateau value of the peak height, on the other hand, was more repeatable. For FACSI-TGF with γ ) 6.1, the plateau value of the peak height was typically between 30-fold and 35-fold greater than the case for normal TGF. Note however, that for a given capillary, and a given set of experimental parameters, the reproducibility of both FACSI-TGF and normal TGF were found to be quite good, with a relative standard deviation (RSD) of between 1% and 10%. Concentration Enhancement in FACSI-TGF. To characterize the concentration enhancement due to FACSI, the peak height was measured for a range of OG488 concentrations in sample buffers of varying conductivity to give a range of γ from 0.78 to 6.1 (see Figure 2). All samples were analyzed in triplicate, using a fresh aliquot of sample for each analysis. Blank scans were run between each change in sample buffer to ensure that there was no cross-contamination. All camera and microscope parameters were kept constant for all experiments. The limited dynamic range of the 12-bit camera forced the use of a different concentration range of analyte for each sample buffer, although one analyte

concentration (1 nmol L-1) was present for all concentration ranges. Sample buffers more dilute than 31 mmol L-1 TB were anticipated to give values for γ that were greater than 6, leading to greater concentration factors. However, as the buffering capacity of the sample was reduced, confounding experimental effects began to appear, leading to an upper limit in γ. Because FACSI-TGF is a counterflow technique, there is a flux of buffer ions from the run buffer into the sample reservoir. As the sample buffer concentration is decreased, this flux becomes increasingly significant, making the value of γ variable over the course of the experiment. In addition, as the buffer strength was reduced, an increased tendency for bubbles to nucleate inside or on the end of the capillary was observed. For these reasons, the experiments conducted here did not pursue greater values of γ. Representative intensity versus applied pressure plots for the analysis of 1 nmol L-1 OG488 are shown in Figure 4. The peak position was observed to vary by less than one peak width for all scans. For a given channel geometry and separation conditions, the peak position is a reflection of the magnitude of the electroosmotic flow. The electroosmotic mobility has been shown to vary significantly with buffer concentration.37 The stability of the peak position with changes in the sample buffer implies that the capillary remained filled with the 500 mmol/L run buffer and that the interface between the run buffer and sample was outside the capillary entrance. In addition, significant variation in the peak width was not observed. The peak height ( the standard deviation (n ) 3) increased from (79 ( 0.97) counts at γ ) 1.0 to (2760 ( 88) counts for γ ) 6.1, an enrichment of 35-fold above TGF alone and 1140-fold over the unfocused sample input. Additionally, under destacking conditions (γ ) 0.8) the peak height was reduced to (25 ( 2.2) counts. These measurements alone could have been used to determine the impact of the conductivity ratio on focusing. However, measurement uncertainty was reduced by comparing slopes of calibration curves, as they contain more data points and the influence of analyte concentration on focusing efficiency can be ruled out. Figure 5 shows calibration curves constructed for each sample buffer concentration. The peak height was observed to be linear with respect to input concentration, as previously reported for scanning TGF,31 indicating that the analyte concentration does not impact the focusing efficiency. The maximum RSD observed was 8.8%; the average RSD across all data points was 3.6%. All correlation coefficients were above a value of 0.99. The slope of the regression lines ( the standard error increased from (75 ( 0.71) counts (nmol L-1)-1 at γ ) 1.0 to (2700 ( 55) counts (nmol L-1)-1 for γ ) 6.1, yielding a maximum enrichment of 36fold above TGF alone and 1170-fold relative to the unfocused input sample. Destacking conditions (γ ) 0.8) reduced the slope to (24 ( 0.25) counts (nmol L-1)-1. Figure 6 is a plot of the regression line slope normalized to the slope for γ ) 1.0 to show the impact of conductivity ratio on focusing. For FASI in CE, at perfect efficiency, it is anticipated that the concentration enhancement would be equivalent to the conductivity ratio (eq 1).19 However, our results indicate concentration enhancements significantly greater than γ and, for the range of γ examined here, data seem to be best fit by a power(37) Kirby, B. J.; Hasselbrink, E. F. Electrophoresis 2004, 25, 187-202.

Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

6205

Figure 5. FACSI-TGF calibration curves. The peak height as a function analyte concentration was determined using identical separation conditions except varying the sample buffer conductivity (0: γ ) 0.78, O: γ ) 1.0, 4: γ ) 1.5, 3: γ ) 2.4, ): γ ) 3.7, ×: γ ) 6.1). Relative slopes of the calibration curves give the concentration enhancement due to FACSI. Inset: close-up of data set. Error bars are (1 standard deviation (n ) 3). Sample was OG488 in TB buffer varied from 31 mmol L-1 (γ ) 6.1) to 1 mol L-1 (γ ) 0.78); separation buffer was 0.5 mol L-1 TB; temperature gradient was 30 °C mm-1; field was -1 kV cm-1; applied pressure was reduced from 250 Pa in 5 Pa increments and held 2 s at each pressure.

Figure 4. TGF and FACSI-TGF results for various values of γ. Sample was 1 nmol L-1 OG488 in TB buffer varied from 31 mmol L-1 (γ ) 6.1) to 1 mol L-1 (γ ) 0.78); separation buffer was 0.5 mol L-1 TB; temperature gradient was 30 °C mm-1; field was -1 kV cm-1; applied pressure (relative to atmospheric pressure) was reduced from 250 Pa in 5 Pa increments and held 2 s at each pressure.

law relationship, slope ratio ) γb, with b ) 1.98 ( 0.01, R2 ) 0.995. 1000-fold Concentration and Separation of Amino Acids. We applied the use of FACSI-TGF to the enrichment and separation of fluorescently labeled amino acids. The separation of five amino acids and the free dye leftover from the labeling reaction under TGF and FACSI-TGF conditions (γ ) 6.1) are shown in Figure 7. Each peak was identified by spiking in the individual labeled components. On average, the FACSI-TGF peak heights were 8.3-fold greater than the standard TGF peaks. While the stacking concentration enhancement is not as great as was seen in the characterization experiments performed using OG488, it still is greater than the conductivity ratio. In addition, the concentration factor from TGF alone is estimated (ratio of peak height to baseline shift) to be between 80-fold and 200-fold, giving a total 6206 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

Figure 6. Concentration enhancement due to FACSI. The calibration slopes (see Figure 5) as a function of the conductivity ratio are shown. The concentration factor appears to have a power law dependence on γ. Nonlinear least-squares fitting gives b ) 1.98 ( 0.01, R2 ) 0.995. Error bars are ( standard error. The line y ) γ, which represents the case of stacking, is shown for comparison.

concentration factor for FACSI-TGF between 664-fold and 1660fold (greater than 1000-fold on average). Furthermore, the ability to resolve unconjugated dye from the labeled amino acids demonstrates that this technique could be integrated with microfluidic

Figure 7. Comparison of TGF (blue) and FACSI-TGF (red) amino acid separations. Five fluorescently labeled amino acids and two forms of unconjugated fluorophore left over from the labeling reaction were resolved. 1: FAM-asp; 2: FAM-glu; 3, 4: unconjugated dye; 5: FAM-gly; 6: FAM-ser; 7: FAM-val. Both experiments were conducted under identical conditions excluding the buffer concentration in the sample. Conditions: Sample was 1 nmol L-1 for each amino acid; sample buffer concentration was 31 mmol L-1 TB (γ ) 6.1) for FACSITGF separations and 0.5 mol L-1 TB (γ ) 1) for TGF separation; separation buffer was 0.5 mol L-1 TB; a temperature gradient of 30 °C to 3 °C was applied across a 10 mm segment of the 4 cm channel; the applied pressure was reduced from a starting pressure of 1200 Pa in 1 Pa increments held at 2 s each; a voltage of -4 kV was applied to the sample.

on-chip labeling of samples without the additional sample cleanup steps. CONCLUSIONS In this paper we have described FACSI, a new method for sample preconcentration. In FACSI, a counterflow is used to generate a pseudo-stationary interface between a low conductivity sample buffer and a high conductivity separation buffer. Analytes migrate to and focus in the conductivity interface prior to eluting into the capillary. In contrast to conventional FASS and FASI methods, the concentration enhancements that are observed with FACSI are greater than the conductivity ratio, γ. Another consequence of the concentration factor being greater than the concentration ratio is that that dilution of a sample with water can actually result in an increase in the detector signal.

In TGF, there is a three-way tradeoff between analysis time, detection limit, and resolution. By using a shallow temperature gradient, high-resolution separations can be achieved rapidly but with little focusing occurring to aid in detection. FACSI-TGF, however, is able to achieve all three goals, as stacking is not observed to have a significant impact on resolution. Alternatively, because the concentration factor in TGF is inversely proportional to scan rate, FACSI-TGF could be used to allow for more rapid analysis without a corresponding decrease in the signal-to-noise. FACSI-TGF is a simple method for the improvement of LOD in TGF separations. Except for the conductivity difference between the sample and separation buffers, implementation of FACSI-TGF is identical to TGF. It is important to note that the concentration enhancement in FACSI-TGF is calculated at the detection point. We observed a 1000-fold concentration enhancement while conducting a highresolution separation. Since the peak concentration for CE without stacking or preconcentration is always less than the sample concentration, the detection limit gain of this method relative to CE should be greater than 1000-fold (assuming a similar detector). FACSI-TGF additionally retains all the advantages associated with counterflow separations, including matrix exclusion from the separation channel and reduction in the channel length.17 Although the implementation of FACSI-TGF is similar to FASI in CE, concentration enrichment factors greater than the conductivity ratio were achieved because of the counterflow. This result could apply to other counterflow gradient focusing techniques as well. We believe that FACSI in the context of counterflow separations represents a new focusing method. Further work is ongoing to overcome the practical limitations on γ present in this work and to demonstrate simple serialization of FACSI with other counterflow techniques. ACKNOWLEDGMENT The authors acknowledge the financial support of the NIST/ NRC Postdoctoral Research Program. This research was supported by the NASA Astrobiology Science and Technology Instrument Development (ASTID) program grant number NNH06AE121.

Received for review April 9, 2007. Accepted June 7, 2007. AC070689R

Analytical Chemistry, Vol. 79, No. 16, August 15, 2007

6207