Gradient Elution Isotachophoresis for Enrichment and Separation of

Jonathan G. Shackman and David Ross*. Biochemical Sciences Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899...
1 downloads 0 Views 504KB Size
Anal. Chem. 2007, 79, 6641-6649

Gradient Elution Isotachophoresis for Enrichment and Separation of Biomolecules Jonathan G. Shackman and David Ross*

Biochemical Sciences Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899

A novel format for performing capillary isotachophoresis (ITP) is describedsgradient elution ITP (GEITP). GEITP merges the recently described electrophoretic separation technique of gradient elution moving boundary electrophoresis (GEMBE) with an ITP enrichment step. GEMBE utilizes a combination of continuous sample injection with a pressure-controlled counterflow; as the counterflow is reduced, analytes are sequentially eluted onto the separation column and detected as boundary interfaces. By incorporating leading electrolytes into the counterflow and terminating electrolytes into the sample matrix, an ionic interface can be formed near the capillary inlet. The discontinuous buffer system forms highly enriched analyte zones outside of the capillary, which are then eluted onto the separation capillary as the counterflow is reduced. Separation of fluorescent analytes was achieved either through discrete electrolyte spacers added to the sample or by using ampholyte mixtures to form a continuum of spacers. As the ITP process occurs off-column, extremely short length separations can be achieved, as demonstrated by a separation in 30 µm. The effects of various parameters on the GEITP enrichment process are investigated, including initial counterflow rates, electric field, leading electrolyte concentration, and counterflow acceleration, which is an adjustable parameter allowing for highly flexible separations. Typical enhancements in limits of detection and sensitivity were greater than 10 000-fold and were achieved in less than 2 min, yielding lowpicomolar detection limits using arc lamp illumination and low-cost CCD detection. An optimized system afforded greater than 100 000-fold improvement in detection of carboxyfluorescein in 8 min. Specific examples of enrichment and separation demonstrated include the following: small dye molecules, DNA, amino acid mixtures, and protein mixtures. Capillary electrophoresis (CE) is a powerful analytical tool that is currently maturing from mainly a research technique to a routine method, perhaps propelled by the explosion of interest in microfluidic devices.1-3 While affording rapid, high-resolution separations with small sample requirements, CE suffers from high * To whom correspondence should be addressed. Phone: 301-975-2525. Fax 301-975-4845. E-mail: [email protected]. (1) Lee, S. J.; Lee, S. Y. Appl. Microbiol. Biotechnol. 2004, 64, 289-99. (2) Dittrich, P. S.; Tachikawa, K.; Manz, A. Anal. Chem. 2006, 78, 3887-907. (3) Dittrich, P. S.; Manz, A. Nat. Rev. Drug Discovery 2006, 5, 210-8. 10.1021/ac070857f Not subject to U.S. Copyright. Publ. 2007 Am. Chem. Soc.

Published on Web 08/04/2007

concentration limits of detection (LOD) when using the most common optical detection techniques (i.e., absorbance and fluorescence).4,5 The detection problem has been exacerbated when translated to microfluidic devices, which frequently employ thick borosilicate glass or polymeric materials, as well as extremely small channel dimensions that further degrade concentration LOD when compared to results obtained with fused-silica capillaries. While alternative detection strategies and microcolumn geometries have been employed to improve LOD,6-8 sample enrichment methods, both preconcentration and in-line concentration, have also been very successfully utilized to bring analytes up to the detection limits of the common detectors using standard capillary instruments.9-12 Electrophoretically based sample enrichment (as opposed to extraction, filtration, or chromatographically based enrichment) exploits differences in the electrophoretic velocities of the analytes during the separation. These methods can loosely be grouped into two categories: equilibrium focusing and dynamic enrichment. Focusing relies upon creating a point within the separation region where the analyte experiences zero net velocity; this can be achieved through pH gradients, such as in isoelectric focusing (IEF), or electric field-related gradients, as found in electric field gradient focusing (EFGF), temperature gradient focusing (TGF), and electrocapture techniques.11,12 While IEF is typically limited to proteins and peptides, which exhibit accessible isoelectric points, the other gradient methods have proven to be more universal with concentration improvements on the order of 10010 000-fold; however, they necessitate specialized equipment or complicated chip fabrication strategies in order to generate the gradients employed, as well as requiring tens of minutes to hours to achieve these enrichments. Of the dynamic enrichment methods, field-amplified sample stacking and field-amplified sample injection (FASI) are the most commonplace.9,10,13,14 The former two methods typically afford 10(4) Weinberger, R. Practical Capillary Electrophoresis; Academic Press Inc.: San Diego, CA, 1993. (5) Gotz, S.; Karst, U. Anal. Bioanal. Chem. 2007, 387, 183-92. (6) Hempel, G. Electrophoresis 2000, 21, 691-8. (7) Viskari, P. J.; Landers, J. P. Electrophoresis 2006, 27, 1797-810. (8) Mogensen, K. B.; Klank, H.; Kutter, J. P. Electrophoresis 2004, 25, 3498512. (9) Lin, C. H.; Kaneta, T. Electrophoresis 2004, 25, 4058-73. (10) Breadmore, M. C. Electrophoresis 2007, 28, 254-81. (11) Wang, Q. G.; Tolley, H. D.; LeFebre, D. A.; Lee, M. L. Anal. Bioanal. Chem. 2002, 373, 125-35. (12) Shackman, J. G.; Ross, D. Electrophoresis 2007, 28, 556-71. (13) Osbourn, D. M.; Weiss, D. J.; Lunte, C. E. Electrophoresis 2000, 21, 276879.

Analytical Chemistry, Vol. 79, No. 17, September 1, 2007 6641

Table 1. Selected Focusing and ITP-Based Enrichment Methods technique

analyte

analysis timea

improvementb

TGF37 EFGF38 tITP volume coupling with static counterflow25 static counterflow tITP23 tITP17

small dye protein small organic

90 min 40 min 2.5 h

10 000-fold in signal 10 000-fold in signal 200-fold in signal

small organic small dye

10 min 3 min

GEITP (this work)

small dye, amino acid, DNA, protein

2-10 min

200-fold in signal 60 000-200 000-fold in concentration 300-100 000-fold in LOD

a Approximate analysis times are either those as reported by the authors or estimated from the time for the analyte of interest to pass the final detector (i.e., the CE detector for tITP). b Improvements are either those as reported in the original article or estimated from nonenrichment versus enrichment conditions.

1000-fold enhancements in detection, although there are instances of much higher enrichment such as using FASI with matrix removal by an electroosmotic flow (EOF) pump yielding 100010 000-fold enrichments.15 While isotachophoresis (ITP) itself is a form of focusing, it is frequently employed as a transient step (tITP) prior to a zonal CE separation, making tITP closer to a dynamic enrichment method. tITP has been demonstrated both in single- and coupled-column formats yielding 100-10 000-fold concentration enhancement.4,13,16 Recently, Santiago’s group has published work demonstrating 60 000-200 000-fold improvements in concentration of small dye molecules in less than 3 min using a dual-channel microfluidic device and tITP.17,18 Selected methods and the technique discussed herein are compared in Table 1. The reader is also directed to current reviews (e.g., refs 10, 12, and 19), which survey the figures of merit of the various methods for analyte enrichment. It is important to note that improvement values are frequently reported only as signal enhancement (e.g., peak height) as opposed to improvement in signal-to-noise or LOD. Counterflow has been used previously in ITP applications, although it has been mainly employed either to lengthen the residence time of analytes on-column to produce higher enrichment and resolution in shorter separation lengths or to affect transfer of the separation from an ITP mode to a CE mode. Vacik and Zuska demonstrated an ITP instrument employing constant electrolyte counterflow aimed at improving the separation of small anions.20 Everaerts et al. detailed a means of producing counterflow in narrow bore (0.5-mm i.d.) polymer tubes to improve the separation of high-concentration, small-molecule samples under ITP conditions. The applied counterflow was performed at a static flow rate (determined by maintaining constant current) during the enrichment, followed by removal of the counterflow for detection. While the effect of counterflow on enrichment was not studied, they did note improved resolution of analytes.21 Reinhoud et al. developed an automated back-pressure system for coupling (14) Breadmore, M. C.; Haddad, P. R. Electrophoresis 2001, 22, 2464-89. (15) Liu, S. H.; Li, Q. F.; Chen, X. G.; Hu, Z. D. Electrophoresis 2002, 23, 33927. (16) Gebauer, P.; Mala, Z.; Bocek, P. Electrophoresis 2007, 28, 26-32. (17) Jung, B.; Bharadwaj, R.; Santiago, J. G. Anal. Chem. 2006, 78, 2319-27. (18) Jung, B. G.; Zhu, Y. G.; Santiago, J. G. Anal. Chem. 2007, 79, 345-9. (19) Osbourn, D. M.; Weiss, D. J.; Lunte, C. E. Electrophoresis 2000, 21, 276879. (20) Vacik, J.; Zuska, J. J. Chromatogr. 1974, 91, 795-808. (21) Everaerts, F. M.; Verheggen, T. P. E. M.; Vandevenne, J. L. M. J. Chromatogr. 1976, 123, 139-48.

6642 Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

ITP to CE and utilized the counterflow to remove the terminating electrolyte (TE) prior to the CE dimension. The constant back pressure maintained the enriched analyte zones (fluorescently labeled amino acids) on column during the ITP process; the counterflow was stopped and the buffer solution containing TE was switched to CE buffer. Once the TE was removed from the capillary, the field polarity was inverted to perform CE separation. A 100-fold LOD improvement was determined.22 A similar method was utilized by Enlund et al. yielding signal improvements of ∼200fold due to the ITP preconcentration step.23 We have recently described a new separation technique termed gradient elution moving boundary electrophoresis (GEMBE), which combines continuous sample injection with a variable bulk counterflow during electrophoresis within a microcolumn.24 In GEMBE, the counterflow, which has EOF as the dominant component, is manipulated by controlling the pressure applied to the ends of the separation channel or capillary. The counterflow is initially great enough to exclude analytes from entering the separation column; as the bulk fluid flow is decreased from high to low rates, analytes are sequentially eluted onto the column and detected as boundary interfaces. GEMBE exhibits the advantages of short separation lengths (typically 1 cm or less) and simplicity of design because there is no need to form an injection plug, making it highly amenable to high-throughput and high-density microdevices. Here we describe the combination of GEMBE with ITP. By introducing leading electrolytes (LE) into a GEMBE counterflow buffer and TE into the sample matrix, a novel form of ITP results, which we term gradient elution ITP (GEITP). As opposed to conventional capillary ITP methods, GEITP enrichment occurs outside of the capillary at a counterflow rate great enough to push leading ions into the sample reservoir, forming an ionic interface near the capillary inlet. Similar to GEMBE, as the counterflow rate is reduced, the enriched plug of analyte is introduced onto the column. The improvement in sensitivity over conventional ITP or tITP is analogous to performing volume-coupled ITP.25-27 (22) Reinhoud, N. J.; Tjaden, U. R.; Vandergreef, J. J. Chromatogr. 1993, 641, 155-62. (23) Enlund, A. M.; Schmidt, S.; Westerlund, D. Electrophoresis 1998, 19, 70711. (24) Shackman, J. G.; Munson, M. S.; Ross, D. Anal. Chem. 2007, 79, 565-71. (25) Mazereeuw, M.; Tjaden, U. R.; Vandergreef, J. J. Chromatogr., A 1994, 677, 151-7. (26) Verheggen, T. P. E. M.; Everaerts, F. M. J. Chromatogr. 1982, 249, 22130. (27) Chen, S. J.; Lee, M. L. Anal. Chem. 1998, 70, 3777-80.

Previous work has demonstrated that combining a small-bore detection capillary with a larger bore focusing capillary allows for improved LOD by allowing for much larger sample injection volumes. Mazereeuw et al. combined a 50-cm-long, 300-µm-i.d. focusing capillary with a 60-cm, 100-µm-i.d. detection capillary; injections on the order of 20 µL were made, and ∼200-fold improvement in signal was achieved after 2.5 h of focusing.25 GEITP is similar to extreme volume coupled ITP separation with counterflow, with the “focusing column” being the sample reservoir itself, giving i.d. ratios between the focusing and separation columns on the order of 200 (as opposed to 3 used by Mazereeuw et al.). Additionally, the sample reservoir has essentially zero pressure generated or EOF-generated flows associated with it, which can hamper the focusing and separation steps. Due to the gradient in counterflow, a wide range of analyte electrophoretic mobilities can be enriched and separated in a single analysis on a short capillary; while rapidly migrating analytes are being introduced on-column for detection, slower migrating analytes continue to focus outside of the capillary inlet. GEITP allows for high-resolution separations in short lengths, involves only a single microcolumn, and does not require a discrete injection method, which will aid in transfer to microdevices. Additionally, as compared to other ITP methods, GEITP does not necessitate any buffer or polarity switching, yielding a more reliable and automatable system. The work described herein investigates the operational parameters of GEITP, typically obtaining greater than 10 000-fold improvements in LOD and sensitivity within 2 min, with an optimized system demonstrating enhancement of over 130 000-fold; also, examples of enrichment and separation of DNA, amino acids, and proteins are presented. EXPERIMENTAL SECTION28 Chemicals and Reagents. Fluorescein, 6-carboxyfluorescein (FAM), and 5-carboxyfluorescein, succinimidyl ester (5-FAM, SE) were purchased from Invitrogen (Carlsbad, CA). Fluoresceinlabeled ssDNA of 30 cytosine bases was from Sigma-Genosys (St. Louis, MO). Green fluorescent protein (GFP) and DsRed were obtained from BD Biosciences (San Jose, CA). All solutions were made from Milli-Q (Millipore, Bedford, ME) g18 MΩ cm deionized water. All other chemicals were obtained from SigmaAldrich (St. Louis, MO), including high-resolution ampholyte mixture (pH 3.0-10.0), and were of the highest purity available. Stock solutions of amino acids (aspartic acid, glycine, serine, valine) were 1 mmol/L in 15 mmol/L sodium borate buffer (pH 9.2). 5-FAM, SE stock was 0.1 mol/L in dimethyl sulfoxide. For 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 were carried out on the apparatus previously described for GEMBE24 and shown in Figure (28) Certain commercial equipment, instruments, or materials are identified in this report to specify adequately 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.

1a. Briefly, 3 cm of fused-silica capillary (30-µm i.d., 360-µm o.d.; Polymicro Technologies, LLC, Phoenix, AZ) with a 5-mm-wide optical window burned through the polyimide coating was enclosed between two polycarbonate sheets (McMaster-Carr, Atlanta, GA). The sandwich assembly was placed in a hydraulic press at 500 kg, heated to 200 °C, and then cooled to 120 °C prior to releasing pressure, ensuring good mechanical stability of the exposed silica region. One end of the device was inserted into a 360-µm-diameter hole drilled into a 110-µL polypropylene sample reservoir maintained at high voltage while the other end was inserted through a Teflon-backed silicon septum into a grounded polypropylene buffer reservoir containing 1.4 mL of leading electrolyte buffer. The buffer reservoir was connected to a (14 kPa (2 psi) precision pressure controller (Series 600, Mensor, San Marcos, TX) for regulation of the counterflow velocity. Fluorescence detection was performed on an upright Axioskop 2 Plus microscope (Carl Zeiss, Thornwood, NY) equipped with a 5× objective (0.13 numerical aperture), Hg arc lamp, and 8-bit color CCD camera (DXC-390, Sony, New York, NY). The fluorescence intensity was acquired at 60 Hz within a region of interest detection window with a width of 6 µm (in the axial direction) and height of 24 µm (to approximately match the inner diameter of the capillary) located 1 cm from the capillary inlet (except when otherwise noted). A 470 ( 20 nm excitation and 515-nm-long pass emission filter set was used for all experiments except those involving both GFP and DsRed, which utilized 485 ( 20/546 ( 12 dual band-pass excitation and 522 ( 7 and 605 ( 25 nm bandpass emission filters. All instrument control and data acquisition was performed using Java 5.0 software (Sun Microsystems, Santa Clara, CA) written in-house. Data analysis was performed using Cutter software.29 Peak identification of mixtures was accomplished through individual runs and spiking of analytes. RESULTS AND DISCUSSION GEITP relies upon forming an ionic interface near the capillary inlet through the use of counterflow. The mechanism is similar to that recently described for field-amplified continuous sample injection coupled to TGF whereby a high-conductivity buffer is pushed by bulk convective flow into a sample reservoir containing the same buffer at reduced concentration; the off-capillary conductivity interface allowed for introduction of higher analyte concentration into the temperature gradient region on-capillary, where further enrichment occurred.30 To create an ITP interface for anionic analyses, a high concentration of a leading anion is incorporated into the separation buffer located in the pressurecontrolled buffer reservoir (see Figure 1a); a terminating anion with an electrophoretic mobility less than the analyte is included in the sample matrix. When negative high voltage is applied at the sample reservoir, electrophoresis is counteracted by EOF and the hydrodynamic flow, pushing LE into the sample reservoir and forming a discontinuous buffer boundary. Analytes and TE are driven by electrophoresis into the ionic interface, forming enriched ITP layers (Figure 1b). As the counterflow is lowered through reduction in applied pressure, the interface and accumulated analytes are sequentially introduced onto the capillary, where they can then be detected. To achieve discrimination of fluorescent (29) Shackman, J. G.; Watson, C. J.; Kennedy, R. T. J. Chromatogr., A 2004, 1040, 273-82. (30) Munson, M. S.; Danger, G.; Shackman, J. G.; Ross, D. Anal. Chem. In press.

Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

6643

Figure 1. Instrumentation for and concept of GEITP. (a) A 3-cm section of capillary connects a pressure-controlled and grounded reservoir containing 1.4 mL of leading electrolyte (LE) buffer solution to a 110-µL sample reservoir maintained at high voltage (HV). Detection was performed by fluorescence microscopy 1 cm from the sample inlet. (b) It is hypothesized that the LE creates an ionic interface at the sample inlet upon which analytes and trailing electrolyte (TE) are enriched; the interface and analyte zones are subsequently pulled into the capillary as the hydrodynamic flow is reduced. (c-f) False color images were taken of the capillary sample inlet with a sample solution containing 100 nmol/L each of carboxyfluorescein (FAM) and fluorescein (F) and 1 mol/L glycine spacer in 0.5 mol/L tris-borate buffer at (c) 1800, (d) 850, (e) 730, and (f) 590 Pa of applied pressure. LE solution contained 100 mmol/L citrate balanced to pH 8.3 with tris; field strength was -600 V/cm. (g) Electropherogram showing separation of FAM and F detected 30 µm from the capillary inlet. -10 Pa/s from starting pressure of 850 Pa; other conditions as above.

analytes with similar emission spectra, either discrete nonfluorescent components can be added to the sample matrix31 or, in the case of multianalyte separations, a continuum of spacers can be achieved by introducing ampholyte mixtures.32,33 The GEITP process can be visualized in Figure 1c-f. Images were taken through the sample reservoir at the capillary inlet, where the polyimide coating was removed. The sample buffer, 0.5 mol/L tris-borate (pH 8.3), contained 100 nmol/L each of fluorescein and FAM and 1 mol/L glycine (acting as a spacer). The large analyte concentrations were used to better visualize the capillary inlet and to assist detection through the several millimeters of fluorescent solution. The pressure-controlled waste reservoir contained 100 mmol/L citrate balanced to pH 8.3 with tris. In this instance, citrate acted as the LE and borate as the TE, although other systems utilizing chloride or acetate as LE and HEPES as a TE were also successfully used (data not shown). (Note that borate at the concentration and pH used here forms a variety of different ions in solution, including tetrahydroxyborate, tetraborate, triborate, and pentaborate, all of which exist in a complicated equilibrium.34 For simplicity, in this work we will refer to the collection of borate ions and other borate compounds as “borate”.) At high counterflow rates (1800 Pa applied pressure; Figure 1c) no enrichment was detectable, likely due to the high (31) Marak, J.; Nagyova, I.; Kaniansky, D. J. Chromatogr., A 2003, 1018, 23349. (32) Inano, K.; Tezuka, S.; Miida, T.; Okada, M. Ann. Clin. Biochem. 2000, 37, 708-16. (33) Chartogne, A.; Reeuwijk, B.; Hofte, B.; van der Heijden, R.; Tjaden, U. R.; van der Greef, J. J. Chromatogr., A 2002, 959, 289-98. (34) Becker, P. Z. Kristallogr. 2001, 216, 523-33.

6644

Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

rate of LE efflux and rapid dispersion. As the counterflow rate was reduced (850 Pa applied pressure; Figure 1d), a single concentrated fluorescent spot outside of the capillary was observed. When the flow rate was further reduced (730 Pa applied pressure; Figure 1e), FAM was introduced onto the capillary. Note that the use of the spacer excluded fluorescein from entering the capillary until further reduction in counterflow (590 Pa applied pressure; Figure 1f). Although not optimal in the capillary format, with detection through the sample reservoir, Figure 1g shows that extremely short separations can be realized using GEITP, with separation of FAM and fluorescein within 30 µm. Enrichment was not calculated for this experiment owing to the difficulties of calibration through the sample solution, and all other experiments were performed at the detection window outside of the reservoir, 1 cm from the inlet. A systematic evaluation of various parameters on detection enhancement in GEITP were investigated (Figure 2) including: initial pressure, electric field strength, LE concentration, and change in counterflow rate (denoted as acceleration in units of Pa/s). As suggested by Figure 1c-f, the counterflow rate at the onset of the analysis plays a significant role in the enrichment rate of the analyte; too low a flow would result in limited enrichment, as the ionic interface cannot exit the capillary, while too high an initial rate would simply extend the analysis time. This can be seen in Figure 2a, which is a plot of the peak height (at fixed FAM concentration) versus ∆P, which we define as the difference in initial pressure and the pressure at which the peak passes the detector. All points were less than 6% relative standard deviation (RSD; n ) 3). There is a steep rise in the measured

Figure 2. Effects of various parameters upon GEITP enhancement. Detection enhancement was determined by calculating the ratio of the GEITP calibration curve slope to the non-GEITP calibration curve slope. (a) Peak height versus initial pressure (∆P). ∆P is defined as the difference between starting and peak migration pressures. LE buffer: 50 mmol/L citrate balanced to pH 8.3 with tris. Sample: 500 pmol/L FAM in 0.5 mol/L tris-borate buffer, pH 8.3. Other conditions: -1000 V/cm; -10 Pa/s acceleration. (b) Example electropherogram at ∆P ) 942 Pa. (c) Calibration curves as the field strength was varied from -200 to -1000 V/cm. LE buffer: 50 mmol/L citrate balanced to pH 8.3 with tris. Sample buffer: 0.5 mol/L tris-borate, pH 8.3. Acceleration: -10 Pa/s. (d) Effect of field strength on analyte detection enhancement based on LOD determined from (c). A linear trend was fit excluding the point at -200 V/cm. (e) Calibration curves as the LE concentration was varied from 25 to 200 mmol/L citrate (balanced to pH 8.3 by tris). Sample buffer: 0.5 mol/L tris-borate, pH 8.3. Other conditions: -500 V/cm; -10 Pa/s acceleration. (f) Effect of citrate LE concentration on analyte detection enhancement based on LOD determined from (e). A logarithmic trend was fit. (g) Calibration curves as the acceleration was varied from -1 to -100 Pa/s. LE buffer: 200 mmol/L citrate balanced to pH 8.3 with tris. Sample buffer: 0.5 mol/L tris-borate. Field strength: -600 V/cm. (h) Effect of acceleration on analyte detection enhancement based on LOD determined from (g). A power law trend was fit. All error bars are (1 standard deviation (n ) 3).

Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

6645

Figure 4. GEITP enrichment of fluorescein-labeled DNA. (a) Example electropherogram of 150 pmol/L of a 30-mer of polycytosine (polyC). LE buffer: 100 mmol/L citrate balanced to pH 8.3 with tris. Sample buffer: 0.5 mol/L tris-borate, pH 8.3. Other conditions: -600 V/cm; -10 Pa/s acceleration. (b) Calibration curve for polyC. Separation conditions the same as in (a). Error bars are (1 standard deviation (n ) 3).

Figure 3. Enrichment and separation of FAM and F using a discrete spacer. (a) Sample: 1 nmol/L each of FAM and F and 0.5 mol/L glycine in 0.5 tris-borate, pH 8.3. LE buffer: 100 mmol/L citrate balanced to pH 8.3 with tris. Other conditions: -600 V/cm; -2 Pa/s acceleration. (b) Calibration of F while holding FAM constant at 500 pmol/L. -10 Pa/s acceleration; other conditions the same as in (a). (c) Effect of acceleration on resolution of 1 nmol/L each of FAM and F in 0.5 mol/L tris-borate buffer with 0.5 mol/L glycine. Other separation conditions the same as in (a). A power law trend was fit. All error bars are (1 standard deviation (n ) 3).

peak height as ∆P is increased from 0 to ∼750 Pa. For ∆P greater than ∼750 Pa, the rise in peak height is more gradual. For improved reproducibility, all subsequent experiments were run with a ∆P g 950 Pa. An example electropherogram with ∆P ) 942 Pa is shown in Figure 2b. Panels c and d in Figure 2 show the effect of electric field strength on detection enhancement, all other parameters being held constant. At the LE concentration chosen (50 mmol/L citrate balanced to pH 8.3 with tris), an upper limit of -1000 V/cm was found due to excessive Joule heating and bubble formation above this value. Calibration curves all had R2 values of at least 0.99, and each calibration point had less than 6% RSD for peak heights (n ) 3). To determine enhancement over non-GEITP conditions (i.e., GEMBE), standard solutions were flushed through the capillary and the detector response was measured. The detector response was linear across the concentration range used (2-20 µmol/L) with a nominal LOD (at a signal of three times the standard deviation of the noise) of 300 nmol/L for FAM. It was noted that the enrichment process had no effect on the noise level 6646 Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

of the signal; the ratio of the GEITP LOD to the non-GEITP LOD and the ratio of the slopes of the GEITP and non-GEITP calibration curves yielded equivalent values so that both sensitivity and detection limit were enhanced equally. Enhancement values in detection (Figure 2d, f, and h) were reported based on improvements in sensitivity (the ratio of the slopes of the calibration curves) between the GEITP and non-GEITP conditions. The maximum value observed for the electric field series of calibrations was 29 000 ( 2000-fold improvement (LOD ) 10 pmol/L), with each analysis completed within 2 min. With the exclusion of the calibration point at a field strength of -200 V/cm, it appeared that enhancement was linear with respect to electric field, up to the value limited by excessive current and heating. A similar process was followed to determine the effect of LE concentration (citrate) on detection enhancement. The electric field strength was held constant at -500 V/cm, which limited the current passed at the highest concentration (200 mmol/L) to a reasonable level. Calibration curves are shown in Figure 2e, with all R2 values greater than 0.99 and less than 6% RSD at each point (n ) 3). No effort was made to model the trend in enhancement as LE concentration increased (Figure 2f), although it did appear to follow a logarithmic trend, with a maximum detection improvement of 40 000 ( 1900-fold (LOD ) 4 pmol/L). Finally, the acceleration of flow rate effect on detection enhancement was investigated using 200 mmol/L citrate and -600 V/cm. As slower acceleration leads to longer enrichment times prior to detection, this parameter was predicted to have a significant effect on LOD enhancement. Similar figures of merit were found in the calibration curves (Figure 2g; R2 > 0.99; RSD < 6%; n ) 3), and a maximal enhancement of 130 000 ( 8600 (LOD ) 2 pmol/L) was found. Again, no effort was made to model

Figure 5. Enrichment and separation of a mixture of carboxyfluorescein-labeled amino acids using an ampholyte mixture as continuous spacers. (a) Low-resolution separation of aspartic acid (Asp), glycine (Gly), serine (Ser), valine (Val), and free dye using a constant, rapid counterflow acceleration. LE buffer: 50 mmol/L citrate balanced to pH 8.3 with tris. Sample: 10 nmol/L of each amino acid and 5% (v/v) ampholyte mixture in 0.5 mol/L tris-borate buffer. Other conditions: -900 V/cm; -10 Pa/s acceleration. (b) Moderate resolution separation using a lower counterflow acceleration (-5 Pa/s). Other separation conditions the same as in (a). (c) High-resolution separation using a two-stage counterflow gradient (-2.5 to 400 Pa/s followed by -5 Pa/s). Other separation conditions the same as in (a).

the trend in enhancement. A power law fit to the data gave an exponent of -0.6 ( 0.1, although the fit was not particularly good (R2 ) 0.94). The trend nevertheless suggested a tradeoff between analysis time and concentration enhancement, as an acceleration of -5 Pa/s yielded 95 000 ( 4900-fold improvement in ∼3 min while -2 Pa/s yielded 130 000 ( 8600-fold improvement in ∼8 min. To summarize, the following generalities can be made from the above parametric study results. An optimal starting pressure is required to initiate enrichment of the highest mobility analyte. The counterflow generated must be high enough to extend the LE interface into the sample well, although excessively high initial counterflows will only extend the analysis time. Higher electric fields allow for more rapid enrichment due to the higher velocity of the analytes into the LE interface, with an upper boundary being where heating and bubble formation occurs. Increasing the LE concentration leads to a larger difference in electric field between the LE and the sample matrix, with more of the field being dropped across the sample. This also allows for higher analyte flux to the sample-LE interface. Generally, it is recommended to use the highest electric field possible with a moderate LE concentration (hundreds of millimolar). However, the short separation lengths can lead to electrical arcing at very high fields, in which case further increasing LE concentration is an alternative.

Slower acceleration of the counterflow increases the time allowed for enrichment before the analyte enters the capillary and leads to increased concentration at the detector. Similar to other analytical methods, the optimal balance between detection limits and total analysis times will depend on the specific application. To demonstrate simultaneous enrichment and separation, mixtures of fluorescein and FAM with glycine utilized as a spacer were analyzed (Figure 3a). A calibration of fluorescein was obtained while holding FAM at a constant concentration of 500 pmol/L (Figure 3b). Similar figures of merit were obtained for fluorescein as compared to FAM, with R2 ) 0.99 and less than 6% RSD in peak heights (n ) 3), and enhancement of 25 000 ( 1500-fold (LOD ) 10 pmol/L) in 2 min. Additionally, the effect of acceleration on resolution was measured (Figure 3c) and appeared to follow a power law function with an exponent of -0.66 ( 0.05, similar to what was found for GEMBE.24 GEITP was next applied to the enrichment of fluoresceinlabeled single-stranded DNA composed of 30 cytosine bases. While exhibiting high electrophoretic mobility due to charge density, the DNA was roughly 20 times the molecular weight of the small dye molecules. An example electropherogram for the rapid enrichment of 150 pmol/L DNA is shown in Figure 4a, with a calibration shown in Figure 4b. The tailing observed may have been due to impurities (shorter DNA lengths) left from the Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

6647

synthesis of the 30-mer. The calibration figures of merit were similar to the small dye molecules (R2 ) 0.99; < 6% RSD; n ) 3) and 11 000 ( 700-fold improvement in LOD was observed within 2 min. To demonstrate multianalyte separations across a wide range of electrophoretic mobilities, a series of FAM-labeled amino acids (aspartic acid, glycine, serine, valine) were analyzed by GEITP using a 5% (v/v) ampholyte mixture added to the sample matrix to form a continuous band of ITP spacers. As was the case with GEMBE,24 the acceleration of the counterflow rate becomes an adjustable separation parameter in GEITP, analogous to gradient composition in liquid chromatography; additionally, nonlinear or multistage gradients can be employed to optimize resolution and total analysis time. Under modest acceleration conditions similar to those used above for the small dye molecules (-10 Pa/s), glycine, serine, and valine are nearly baseline resolved while aspartic acid and the free dye comigrate (Figure 5a). Reducing the acceleration by half yielded baseline resolution of the slow analytes and partial resolution of aspartic acid from the free dye (Figure 5b). Utilizing a two-stage gradient (Figure 5c), baseline resolution of all components could be achieved in ∼10 min. Further optimization of the spacers utilized could also minimize the analysis time through the reduction of wasted separation space between the high- and low-mobility analytes. The degree of enhancement was not evaluated, as the efficiency of the labeling reaction was not measured and the free dye not removed. However, the electrophoretic mobility of the FAM-labeled aspartic acid is similar to that of FAM, as can be inferred by the migration times of the free dye peaks, so that similar enrichment (10 000100 000-fold) should be possible at least for that analyte. Although optimized for relatively high mobility analytes using the above conditions, it was still possible to achieve separation and enrichment for very slow migrating species (e.g., proteins) with GEITP. As an example, GFP was analyzed using citrate as LE and borate as TE (Figure 6a). The extreme tailing of this protein (MW ≈ 27 kDa) suggests that, under these conditions, the TE may actually have been of higher mobility than the analyte, reducing the enhancement as measured by peak height. The calibration obtained showed only a modest 300 ( 20-fold improvement (LOD ) 4 nmol/L) over non-GEITP conditions (LOD ) 1200 nmol/L), although this enhancement occurred in 2 min. Additionally, these results are on par with those obtained using conventional modes of capillary ITP, albeit in a simpler format.35,36 To demonstrate separation of similar MW proteins, 100 nmol/L each of GFP and DsRed were analyzed; although a discrete spacer would be optimal, 5% (v/v) ampholyte solution was used instead to simplify the separation. Figure 6c shows an electropherogram that is the result of averaging the red and green channel outputs of the CCD detector. The sharper tail of the GFP peak may be indicative that a component within the ampholyte mixture was acting as a more effective TE, as opposed to borate (compare Figure 6a and c). (35) Olvecka, E.; Kaniansky, D.; Pollak, B.; Stanislawski, B. Electrophoresis 2004, 25, 3865-74. (36) Busnel, J. M.; Descroix, S.; Godfrin, D.; Hennion, M. C.; Kasicka, V.; Peltre, G. Electrophoresis 2006, 27, 3591-8. (37) Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 2556-64. (38) Humble, P. H.; Kelly, R. T.; Woolley, A. T.; Tolley, H. D.; Lee, M. L. Anal. Chem. 2004, 76, 5641-8.

6648 Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

Figure 6. Enrichment and separation of natively fluorescent proteins. (a) Example electropherogram of 10 nmol/L green fluorescent protein (GFP) in 0.5 mol/L tris-borate buffer, pH 8.3. LE buffer: 100 mmol/L citrate balanced to pH 8.3 with tris. Other conditions: -600 V/cm; -10 Pa/s acceleration. (b) Calibration curve for GFP. Separation conditions the same as in (a). Error bars are (1 standard deviation (n ) 3). (c) Separation of 100 nmol/L each of DsRed and GFP using 5% ampholyte mixture in the sample buffer. Trace is the averaged signal of the red and green CCD detector channels. LE buffer: 50 mmol/L citrate balanced to pH 8.3 with tris. Sample buffer: 0.5 mol/L tris-borate, pH 8.3. Other conditions: -1000 V/cm; -5 Pa/s acceleration.

CONCLUSIONS GEITP is a novel format for performing simultaneous capillary ITP enrichments and separations. Requiring only a short, singleseparation column, combined with the continuous injection mechanism, GEITP is a method that can easily be implemented and automated in microcolumn designs. While the nominal separation length used was 1 cm, microfluidics should allow even shorter separation distances, as evidenced by obtaining resolution of two analytes within 30 µm in the nonoptimal capillary format. The improvements in LOD obtained can also ease detection hardware requirements, resulting in lower cost of complete instrument systems. Indeed, low-picomolar detection limits were routinely obtained in 2 min using only arc lamp illumination and a low-cost imaging CCD detector. In addition to providing rapid enrichment of a variety of analytes, spanning the range of mobilities from rapid DNA to slow proteins, the adjustable parameter of counterflow acceleration affords a new dimension

of flexibility to electrophoretic separations, which can be utilized to optimize separation resolution and total analysis time. As an alternative to utilizing spacing compounds, GEITP could be coupled to a CE separation step to provide even higher resolution separations in addition to rapid concentration enhancement. While this report has made some initial general statements regarding fundamental GEITP parameters effects upon enrichment, further theoretical modeling and experimental validation would provide a deeper understanding of the process and improved optimization of the technique.

ACKNOWLEDGMENT The authors acknowledge the financial support of the NIST/ NRC Postdoctoral Research Program and Dr. Gre´goire Danger for assistance in the amino acid labeling. This research was supported in part by the NASA Astrobiology Science and TechnologyInstrumentDevelopment(ASTID)programgrantNNH06AE121.

Received for review April 26, 2007. Accepted June 22, 2007. AC070857F

Analytical Chemistry, Vol. 79, No. 17, September 1, 2007

6649