Michelle GallowaySteven A. Soper

Department of Chemistry,Choppin Laboratories,Louisiana State University,Baton Rouge, LA, USA

Contact conductivity detection of polymerase chainreaction products analyzed by reverse-phase ionpair microcapillary electrochromatography

We describe the development of an integrated microelectrophoretic system consistingof a contact conductivity detector mounted on-chip for monitoring the separation ofdouble-stranded (ds) DNA fragments produced via the polymerase chain reaction(PCR) using microcapillary electrochromatography as the separation mode. Theseparation was carried out in a polymer-based microfluidic device, hot-embossedinto poly(methylmethacrylate) (PMMA), whose walls were functionalized to produce aC18-terminated surface to act as the stationary phase (open channel format). Thecarrier electrolyte contained the ion-pairing agent, triethylammonium acetate (TEAA)to allow the separation to be carried out using reverse-phase ion-pair capillary electro-chromatography (RP-IPCEC). The microelectrophoretic separations were investigatedutilizing various solvent strengths (acetonitrile/water) with 25 mM TEAA to observe theeffects on the separation efficiency as well as the chromatographic development timeand detector performance. The field strength significantly affected the quality of theseparation, with no separation observed at 333 V/cm for a low mass dsDNA sizingladder, but baseline separation achieved using a field strength of 67 V/cm. It wasobserved that the solvent strength affected the retention behavior of the polyanionicmolecules as well as the electroosmotic mobility. Higher acetonitrile compositions inthe run buffer resulted in reduced plate numbers, which produced lower chromato-graphic resolution. The use of conductivity detection allowed mass detection sensitiv-ities in the range of 10�21 mol with a separation efficiency of 104 plates and the per-formance of the detector independent of the acetonitrile content used in the carrierelectrolyte.

Keywords: Contact conductivity detector / Polymerase chain reaction / Poly(methylmethacrylate) /Reverse-phase ion-pair microcapillary electrochromatography EL 5216

1 Introduction

The PCR is a valuable tool used frequently for eitherDNA sequencing or molecular diagnostic applications.The PCR procedure is typically utilized to amplify targetregions of DNA enzymatically and can produce �105-fold increases in the amount of starting target sequences[1, 2]. Following PCR amplification of target DNAs, theresults of the amplification must be validated, which isusually carried out via an electrophoretic analysis, suchas capillary gel electrophoresis (CGE) with either UV ab-

sorbance or laser-induced fluorescence detection [3–10].This separation step not only serves to validate the cor-rect production of the amplicon, but also can be used topurify the PCR product prior to subsequent analysis, suchas sequencing of the PCR product or determining itsallelic composition for mutational analysis. Purification ofthe PCR product is required to remove unwanted salts,primers, and enzymes and is important if down streamprocessing of the amplicon is required, since the reagentspresent in the PCR cocktail can potentially interfere withsubsequent steps.

Numerous strategies have been used to perform purifica-tion of PCR-generated DNA. These include gel filtration[11], gel electrophoresis [7], ultrafiltration [12] or ion-ex-change [13, 14]. Unfortunately, these techniques requirelarge amounts of solvents, can be time-consuming, areplagued by low sample recoveries and often cannot beeasily automated. In addition, most of the gel-basedmethods require the amplicon to be excised from the gelmatrix following separation.

Correspondence: Professor Steven A. Soper, Department ofChemistry, Choppin Laboratories, Louisiana State University,Baton Rouge, LA 70803-1804, USAE-mail: chsope@lsu.eduFax: +225-578-3458

Abbreviations: PMMA, poly(methylmethacrylate); RP-IPCEC,reverse-phase ion-pair capillary electrochromatography; TEAA,triethylammonium acetate

3760 Electrophoresis 2002, 23, 3760–3768

2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 0173-0835/02/2111–3760 $17.50�.50/0

Electrophoresis 2002, 23, 3760–3768 Contact conductivity detection of PCR products 3761

An attractive alternative to the aforementioned purificationtechniques for PCR products is a reverse-phase separa-tion, which can readily purify PCR products with highrecoveries, typically exceeding 97% [15–18]. In addition,the use of reverse-phase chromatography is conduciveto automation, making it amenable to implementation inhigh-throughput DNA sequencing applications. For mostreverse-phase separations of DNA amplified via PCR, anion-pairing agent is used in conjunction with a packed-col-umn containing for example, a C18 stationary phase. Thecommon ion pairing agents that have been used are qua-ternary ammonium salts, such as tetrabutylammoniumphosphate or triethylammonium acetate (TEAA) [19, 20].

Conventional reverse-phase separations use liquid chro-matographic equipment, which consists of dual recipro-cating pumps to allow gradient elution, a liquid chromato-graphic column (4.1 mm ID�10 cm length) and a UVdetector to allow direct detection of the DNA. There arethree potential shortcomings associated with using LCequipment for PCR product separations or purifications;(i) sample dilution due to the use of macrobore columns;(ii) large consumption of solvents, and (iii) difficult tomultiplex the assay (i.e. run many columns in parallel). Inorder to circumvent the limitations associated with (i) and(ii) listed above, we have developed a �-reverse-phasehigh-pressure liquid chromatography separation of PCRproducts with detection accomplished using a miniatur-ized conductivity detector [21]. The packed column was15 cm in length and possessed an ID of 300 �m. The de-tector used two Pt wires and a bipolar waveform appliedto the electrode pair from which the conductivity of thesolution could be measured. Efficient separation of thePCR amplicon from the other reagents present in the PCRcocktail was achieved in less than 4 min with a capacityfactor of 2.5 and separation efficiency of 9.1�103 plates.Unfortunately, this chromatographic format is not feasiblefor high-throughput assays, which would require manyseparation columns running in parallel.

Recently, we reported the use of a microfluidic devicefabricated in poly(methylmethacrylate) (PMMA) to per-form reverse-phase ion-pair separations of polyanionicmaterials using an electrochromatographic format [22].The PMMA chip was hot-embossed from a Ni master toallow rapid production of microfluidic chips. In order tocreate a hydrophobic stationary phase, surface modifica-tion chemistry was used to produce a monolayer cover-age of C18 over the PMMA surface. The native PMMAsurface was unable to separate DNA fragments from amass sizing ladder, but addition of the C18 phase resultedin the separation of the components comprising this massladder. In addition, this C18 phase protected the under-lying polymer substrate from attack by the organic modi-fier included in the reverse-phase separation.

The microfluidic chip was comprised of an integratedconductivity detector for transduction of the componentseluting from the chromatographic channel. Some attrac-tive features of conductivity detection compared to UVabsorbance or fluorescence are: conductivity detectionis universal since no labeling of the target analyte isrequired, which can significantly simplify upstream anddownstream processing of the sample. In addition, con-ductivity is amenable to small volume detection con-sidering the sensitivity can be optimized by the distancebetween the electrode pair. A prominent disadvantage ofUV absorbance is the detection limit depends on thepathlength, which can translate into poor detection limitsfor microcolumn-based separations. Indeed, the con-centration detection limit for DNA by UV absorbance(260 nm) has been reported to be on the order of 10�4–10�5 M [23].

In this paper, we wish to report on the isolation/separationof PCR amplified double-stranded (ds) DNA fragmentsusing reverse-phase, ion-pair capillary electrochromatog-raphy (RP-IPCEC) with contact conductivity detection.The electrochromatography was carried out in a PMMA-based microfluidic device, whose surfaces were modifiedwith a C18 phase to permit effective partitioning of theion-pair complexes into the stationary phase. As part ofthe method development, the mobile phase compositionwas altered to optimize and examine the effects on thereverse-phase separation in terms of chromatographicdevelopment time, separation efficiency and separationresolution. In addition, the effects of the mobile phasecomposition on the performance of the contact conduc-tivity detector was also investigated.

2 Materials and methods

2.1 Microfluidic device fabricated in PMMA forintegrated conductivity detection

The microfluidic devices were fabricated using tech-niques developed in our laboratories, which consisted offabricating a nickel mold master using LIGA and hot-embossing polymer parts from that master [22, 24]. Theembossing system consisted of a PHI Precision Pressmodel number TS-21-H-C(4A)-5 (City of Industry, CA,USA). Prior to hot-embossing, all residual water presentin the polymer must be removed. At room temperature,water adsorption to PMMA is �0.4% and this amountmust be reduced to 0.1% for proper embossing. ThePMMA wafers (Goodfellow, Berwyn, PA, USA) were133 mm in diameter and were baked in an oven at 80�Cfor 8 h. The molding master was coated with a releaseagent, MoldWiz, to improve demolding. The master was

Min

iatu

riza

tio

n

3762 M. Galloway and S. A. Soper Electrophoresis 2002, 23, 3760–3768

heated to 150�C and pressed into the PMMA wafer witha force of 1000 lbs for 4 min. During this process, themaster was heated to 160�C. After 4 min, the press wasopened and the embossed microdevice removed andcooled. The PMMA wafer was maintained at 85�Cthroughout the demolding process. The embossed micro-fluidic device consisted of a 15 �m wide separation chan-nel of 4 cm total length (effective length = 3 cm). For theintegrated conductivity detection, guide channels weredesigned with a width of 130 �m for the Pt wires (127 �mdiameter) used for conductivity detection. The Pt wireswere end-polished using a 600 �m grit sandpaper fol-lowed by 25, 15, 3, and 1 �m diamond pastes (Buehler,Lake Bluff, IL, USA). After each polishing, the Pt wireswere rinsed with deionized water (18 M��cm) and soni-cated for 2 min, then inspected under the 20� stereo-microscope to ensure proper end polishing and insertedinto the guide channels embossed in the microfluidicPMMA substrate to accommodate the electrode pair.Once in place (0.5 cm before the anode on the micro-separation channel), the Pt wires were situated via manualmanipulation with viewing under a microscope into theguide channels and terminated with an end-to-endspacing of �15 �m. The Pt wires and the entire fluidicchannels were then covered with a PMMA coverplate bythermally annealing at 107�C for 12 min in a GC oven withthe PMMA pieces clamped between two glass plates.The bipolar pulse waveform for the conductivity detectorwas generated by an in-house fabricated circuit, whichwas controlled by a National Instruments board (PCI-1200; National Instruments, Austin, TX, USA) [21, 22].The pulse frequency was typically 5.0 kHz, and the pulsewidth 100 �s. The pulse width was selected to be shortcompared to the cell time constant defined by the dou-ble-layer capacitance and the solution resistance to mini-mize Faradaic currents that could be produced followingdouble-layer formation. The potential of one electrodewas maintained at virtual ground while the potential atthe other electrode was controlled by the generated bi-polar waveform. The current between the electrodepair was measured 5 �s prior to the rising edge of everybipolar pulse (� 0.5 V) and averaged over the electro-phoresis sampling time (1 s) to improve S/N. The dataacquisition and controlling software was written in Lab-view (National Instruments).

2.2 PMMA surface modification to createC18-phase

The C18-modified PMMA (5) was developed in our labora-tories using the chemistry shown in (Fig. 1) [25]. Prior tochip assembly, the chip surfaces were modified by flood-ing the channels with the reagents required for the mod-

Figure 1. Synthetic scheme for the preparation of C18-modified PMMA used to create a hydrophobic stationaryphase for reverse-phase separations.

ifications. The surface modification procedure consistedof converting the methyl ester groups of the polymerbackbone into an amine-terminated surface (3) by addingN-lithioethylenediamine (1) (Sigma-Aldrich Chemical, St.Louis, MO, USA) to the PMMA (2) microfluidic channelsfor approximately 5 min followed by subsequent washeswith 2-propanol (Sigma-Aldrich Chemical) and deionizedwater (18 M��cm). The amine-terminated microdevicewas placed in an airtight vessel which was purged withnitrogen for 20 min and then exposed to neat N-octa-decane 1-isocyanate (4) (Sigma-Aldrich Chemical) for10 min followed by subsequent washes with copiousamounts of hexanes, toluene, and acetone (Sigma-Aldrich Chemical) and finally dried with nitrogen priorto use.

2.3 Current monitoring procedures formeasuring EOFs

EOF measurements were conducted using the proceduredescribed by Zare and co-workers [26] in which thechange in current flow produced by a discontinuouselectrolyte system served as an indicator of the linearvelocity under the application of an applied field. Briefly,the procedure involved filling the microdevice with a solu-tion of 30 mM TEAA and then replacing one reservoir with28 mM TEAA solution. The electric field was applied and

Electrophoresis 2002, 23, 3760–3768 Contact conductivity detection of PCR products 3763

the current measured. After there was no observablechange in current, the linear velocity could be calculatedfrom the time in which this occurred. Dividing the linearvelocity by the electric field strength produced theelectroosmotic mobility (cm2/Vs). While the electroosmo-tic mobility can be measured using the migration time of aneutral marker moving through the PMMA channel (nativeor modified), it was not adopted in the present applicationbecause it may experience partitioning into the stationaryphase producing errors in the electroosmotic mobilityvalue. While there may be errors produced in the electro-osmotic mobility deduced from using a discontinuousbuffer system due to ionic strength effects on the doublelayer thickness, these effects were minimized by using asmall change in the ionic strength of the buffer.

2.4 PCR amplification protocols

The 500-bp (PCR-1) and 1000-bp (PCR-2) products of�-DNA (10 ng/mL) (Applied Biosystems, Branchburg, NJ,USA) were amplified by PCR in a Perkin-Elmer GeneAmpPCR System 2400 thermocycler (Foster City, CA, USA).For DNA amplification, the temperature program wasinitiated at 94�C for 1 min to completely denature thetarget DNA and the cycling temperature steps were 94�Cfor 15 s, 55�C for 15 s, and 72�C for 1 min. These stepswere repeated for the appropriate number of thermalcycles and then, the temperature was held at 72�C for7 min to complete chain extension. Samples were cooledto 4�C to conclude the program. The PCR reaction cock-tail contained 10 mMTris-HCl (pH 8.3), 50 mM KCl, 1.5 mM

MgCl2, 0.001% gelatin, 200 �M each of deoxynucleoside

triphosphates, 25 units/mL AmpliTaq DNA polymerase,1.0 �M each primer, and 10 ng/mL �-DNA. The primerset used to amplify PCR-1 was PCR01 forward primer(5’-GATGAGTTCGTGTCCGTACAACTGG-3’) and a PCR02reverse primer (5’-GGTTATCGAAATCAGCCACAGCGCC-3’) (Perkin-Elmer). The primer set used to amplify PCR-2was PCR03 forward primer (5’-GGATATGGGCCGCAGTGAGGAGAA-3’) (Life Technologies, Gaithersburg, MD,USA) and a PCR02 reverse primer (5’-GGTTATCGAAATCAGCCACAGCGCC-3’).

2.5 DNA analysis via RP-IPCEC

Instrumentation used for RP-IPCEC is seen in Fig. 2.The PCR products were loaded directly from the thermalcycler into one reservoir of the microfluidic device, whilethe other three reservoirs were filled with carrier electro-lyte solution. The samples were electrokinetically loadedinto a cross offset “T” injector by applying a 200 V/cmfield between the sample and waste reservoirs with thesample at ground for 3 s. During electrophoresis, a fieldof 67 or 333 V/cm was applied between the cathode andanode (injection end was cathodic), while using 10% pullback voltages for the sample and waste reservoirs toprevent leakage of sample or waste into the separationchannel during the separation. The DNA moleculesmigrated down the separation channel (3 cm effectivelength, 15 �m wide, � 85 �m height) and were subse-quently detected within the conductivity cell operated at5.0 kHz using a bipolar pulse amplitude of � 0.5 V. AllRP-IPCEC separations were conducted using various

Figure 2. Schematic diagram ofthe microelectrophoresis appa-ratus with a C18-modified PMMAmicrodevice and an integratedconductivity detector for theanalysis of polyanionic mole-cules.

3764 M. Galloway and S. A. Soper Electrophoresis 2002, 23, 3760–3768

solvent strengths of acetonitrile/water with 25 mM TEAA(pH 7.4) acting as an ion-pairing agent (Fluka, Milwaukee,WI, USA).

3 Results and discussion

3.1 EOF in modified PMMA microchips

In conventional CEC using a C18 or other stationary phasechemically tethered to a fused-silica support via siloxanechemistry, the EOF serves as the primary driving forcefor the mobile phase and analytes moving through thecolumn. Typically, the EOF is significantly greater thanthe electrophoretic mobility of most ionic species and assuch, all material moves from anode to cathode, evenmost anionic species. The linear velocity (veo, cm/s) ofthe EOF can be calculated from the Smoluchowski equa-tion given by:

veo � ��

�0�rRT

2F2

1I

� �1�2

E (1)

where �0 is the permittivity of vacuum, �r is the dielectricconstant of the mobile phase, � is the solution viscosity,� is the charge density of excess ions in the Gouy-Chap-man layer, R is the gas constant, F is Faraday’s constant,T is the temperature and E is the field strength (V/cm). Ascan be seen from the above equation, the excess chargefound on the wall of the closed channel plays a significantrole in determining the magnitude of the EOF as well asthe directional flow of the EOF.

In the present case we are using a PMMA surface, whichis known to have a substantially smaller electroosmoticmobility compared to fused-silica (�eof = 2.0�10�4 cm2/Vs at pH � 7.0 for PMMA) [27]. In addition, we are addingan amine scaffold to the PMMA surface and then reactingthat with a neutral moiety (C18) producing an electro-osmotic mobility different from that of native PMMA. Themeasured electroosmotic mobility for this type of surfacewas found to be �1.75�10�4 cm2/Vs, where the negativesign indicates an EOF that moves from cathode to anode(pure aqueous phosphate buffer, pH � 7.0) [25]. In Fig. 3is shown the EOF profile for the C18-modified PMMAsurface used in these reverse-phase ion-paired electro-chromatographic separations as a function of the per-centage of organic modifier (acetonitrile) found in thecarrier electrolyte. At 0% acetonitrile (pure aqueous solu-tion with TEAA), our electroosmotic mobility was found tobe higher than that reported previously by our group [25],most likely due to the lower ionic strength of the carrierelectrolyte used in the present case. As can be seenfrom Fig. 3, increases in the organic acetonitrile compo-nent to the carrier electrolyte resulted in increases in theelectroosmotic mobility for this C18-modified PMMA sur-

Figure 3. EOF profiles of modified PMMA (monolayercoverage of C18-phase) taken at different percentagesof the organic modifier, acetonitrile. The electroosmoticmobility was measured by monitoring a change in currentversus time with an applied electric field and the appropri-ate carrier electrolyte solution. The carrier electrolyte con-sisted of 30 mM TEAA (higher ionic strength) or 28 mM

TEAA (lower ionic strength), which possessed a pH of7.4. Error bars shown represent the standard deviation inthe measurements.

face. The increases in the electroosmotic mobility withorganic composition in the carrier electrolyte are consis-tent with previous data for fused-silica surfaces with a C18

phase [28–30]. In any case, the fact that the EOF movesfrom cathode to anode for this modified surface is veryconvenient, since the analytes for our separations areall polyanionic (dsDNAs) and thus, possess an electro-phoretic mobility that directs the movement from cathodeto anode, in the same direction as the EOF.

3.2 RP-IPCEC of dsDNAs

In Fig. 4A is shown an RP-IPCEC analysis of the dsDNAlow mass-sizing ladder using a C18-modified PMMAmicrodevice with a field strength of 333 V/cm and 15%acetonitrile in the running buffer. As can be seen, onlyone peak was evident for this ladder with no apparentseparation. The lack of separation may have resultedfrom minimal interactions between the ion-paired DNAcomplexes and the stationary phase during the courseof the elution process since lateral diffusion must carrythe analytes to the wall of the device for effective parti-tioning into the stationary phase to affect the separation.The short elution time and small diffusion coefficientsassociated with these DNA molecules minimizes theinteraction time with the stationary phase. A lack ofseparation for these DNAs also indicates that the free

Electrophoresis 2002, 23, 3760–3768 Contact conductivity detection of PCR products 3765

Figure 4. (A) RP-IPCEC separation of a DNA low masssizing ladder (100 ng/mL and (B) PCR product profiles at0 PCR cycles and 10 PCR cycles using a C18-modifiedPMMA microdevice. The sizing ladder consisted of 100,200, 400, 800, 1200 and 2000 bp fragments. The primer-dimer peak along with the 500-bp (PCR-1) and 1000-bp(PCR-2) product peaks are shown as the dark thick line.The separation was accomplished using 25 mM TFAA(ion-paring agent, pH 7.4) with a mobile phase composi-tion comprised of acetonitrile/water (15/85 v/v) at a fieldstrength of (A) 333 V/cm or (B) 67 V/cm. Detection wascarried out using contact conductivity detection. Theconductivity cell was operated at 5.0 kHz and a pulseamplitude of �0.5 V.

draining behavior (molecular conformational state of theDNA molecule which results in the frictional coefficientdependence of the electrophoretic mobility to scaledirectly with the number of bases comprising the DNAmolecule; coupled to the fact that the charge of the DNAscales directly with the number of bases, the mobility infree solution is independent of base number) of thedsDNA remains intact even for these ion-paired com-plexes and, therefore, they have the same effectiveelectrophoretic mobility in free solution. From the appar-ent mobility in the absence of chromatographic retention,we can estimate a linear velocity of 0.25 cm/s at this field

strength, giving an apparent mobility of �7.5�10�4 cm2/Vs. We can calculate the free solution electrophoreticmobility of the ion-paired complexes using the electro-osmotic mobility obtained from Fig. 3 at 15% acetonitrileand the apparent mobility of these DNA complexes calcu-lated above, which produced a value of �1.9�10�4 cm2/Vs. The free solution mobility of dsDNAs with base pairnumbers � 400 have been measured to be �4.35�10�4 cm2/Vs [31]. Our smaller value for the DNA electro-phoretic mobility is most likely due to the presence ofthe associated cationic species, triethylammonium,which neutralizes some of the negative charge found onthe DNA fragment as well as adding mass to the DNAfragment increasing its frictional coefficient (i.e., lowerelectrophoretic mobility).

We next attempted to do this separation in the openchannel format using a lower applied field. The resultingRP-IPCEC separation is shown in Fig. 4B, which used amobile phase consisting of 15% acetonitrile and 85%water with 25 mM TEAA (pH 7.4) at a field strength of67 V/cm. As can be seen in this electrochromatogram,the sizing ladder chromatogram was effectively devel-oped within 80 s with near baseline resolution for allfragments comprising this ladder. The ability to do thisseparation on a 3 cm length channel even at this lowfield strength is rather surprising given the fact that in thisopen channel format, the analytes must laterally diffuseto the wall in order to interact with the stationary phase.Comparison of Figs. 4A and B indicate that the migrationtime changed from 10 s to 80 s for the 2000 bp DNA frag-ment. Using these values and a diffusion coefficient of2.2�10�8 cm2/s, gives an average diffusional distanceof 7.4 and 18.7 �m for 10 and 80 s, respectively. Clearly,the differences observed in the chromatographic resultsdepicted in Figs. 4A and B cannot be due solely to differ-ences in diffusional distances as calculated with a con-stant diffusion coefficient, which determine the potentialnumber of interactions of the solute between the station-ary and mobile phases. At the present time, we are un-certain about what other effects may be contributing tothe observed results, but one possible cause may be thealtered diffusion coefficient observed by these DNA com-plexes at higher applied fields and the defined orientationthey adopt when placed in a high field. Large DNA mole-cules at high fields were found to have enhanced longitu-dinal diffusion coefficients due to orientational effects inthis field, which may result in reduced lateral diffusion(rectified diffusion) [33]. At lower fields, the DNAs aremore randomly oriented.

Also shown in this figure is the electrochromatogram forthe two PCR products, which was overlaid on the sizingladder chromatogram. The PCR products appeared at45 and 54 s, for PCR-1 and PCR-2, respectively, and

3766 M. Galloway and S. A. Soper Electrophoresis 2002, 23, 3760–3768

possessed migration times that confirmed their size withrespect to the migration times for the components of thesizing ladder. As can be seen from this chromatographictrace, we could readily resolve the two PCR productsusing these chromatographic conditions and also, sepa-rate them from other components found within the PCRcocktail, such as dNTPs, salts, primers and polymerase.

The PCR products in this case were produced from onlyten thermal cycles and clearly discernible signals weregenerated from the contact conductivity detector. Wecan estimate the mass of DNA analyzed from knowledgeof the fixed volume injector dimensions (425 pL) and theconcentration of the PCR products. The DNA concentra-tion can be determined from PCR amplification using:

CDNA � N En� �2n

NAV(2)

where N is the starting copy number of DNA molecules, nis the number of PCR thermal cycles, NA is Avogadro’snumber, En is the efficiency of a particular amplificationstep (depends on supply of dNTPs and integrity of poly-merase enzyme) and V is the volume of the PCR reaction.Since we are using only ten cycles for this calculation,En was assumed to be 1 for each cycle. Using a startingcopy number of 1.9�106 (10 ng/mL of �-DNA, 1.0 �Lused in PCR reaction), the DNA concentration analyzedin the peaks pictured in Fig. 4B was 15.9 pM. For the425 pL injection volume, this represents 6.8�10�21 mol.The signal-to-noise ratio (S/N) from the data presented inFig. 4B was estimated to be 17, giving a mass detectionlimit equal to 1.2�10�21 mol at an S/N = 3. To the best ofour knowledge, this is the lowest detection limit reportedto date using conductivity detection.

3.3 Mobile phase composition effects onchromatographic behavior

Figure 5 illustrates the effects of mobile phase composi-tion, which was varied from 5 to 30% acetonitrile, on theretention factor (k�) for several DNA fragments of the siz-ing ladder and the two PCR products. The separation pro-cess in CEC is determined by the partitioning betweentwo immiscible phases as well as differences in solutes’electrophoretic mobilities. The migration velocity of anycharged solute (v) in CEC can be calculated from [28]:

v � 11 � k� veo � vep

� �(3)

where the sum veo � vep is equal to the apparent velocityvapp and k� is the retention factor. In the present case, allof the ion-paired DNA fragments possess similar electro-phoretic mobilities (see Fig. 4A) and as such, the separa-

Figure 5. The influence of the mobile phase compositionon the retention factor in RP-IPCEC of a DNA low masssizing standard ladder and the PCR-1 and PCR-2 prod-ucts. The mobile phase consisted of acetonitrile/waterwith 25 mM TEAA (ion-pairing agent, pH 7.4). The electro-phoretic conditions were 3 s electrokinetic injection timeand a field strength for the separation of 67 V/cm. Detec-tion was accomplished using contact conductivity detec-tion. The conductivity cell was operated at 5.0 kHz and apulse amplitude of � 0.5 V.

tion is determined primarily by selective partitioning be-tween the C18 and mobile phases. The retention factorswere calculated by rearrangement of Eq. (3):

k� � vep � veo

v� 1 (4)

The electrophoretic mobility of all ion-paired DNA com-plexes was calculated from the results depicted inFig. 4A, where no stationary phase interactions wereassumed and the velocity of the EOF was taken from theresults depicted in Fig. 3. As can be seen from Fig. 5,k� decreased as the amount of acetonitrile was increasedin the mobile phase, similar to what has been observedin other studies using CEC with C18-based stationaryphases and fused-silica capillaries [28–30, 34, 35]. Inaddition, while increasing the acetonitrile concentrationdoes reduce the analysis time by lowering the partitioningof the ion-paired complexes to the stationary phaseand increasing the electroosmotic mobility, the ability toresolve DNAs becomes more problematic (i.e. smallerdifferences in k� values for the solutes). In fact, at 30%acetonitrile, the k� value for the 400 bp fragment wasfound to be approximately 0, indicating no partitioning tothe stationary phase. Therefore, for short DNAs (� 400 bp)using these chromatographic conditions, acetonitrilecompositions exceeding 30% would not be effective inperforming the desired separation.

Electrophoresis 2002, 23, 3760–3768 Contact conductivity detection of PCR products 3767

Figure 6. The effects of organic modifier on (A) the selec-tivity (– – –), plate numbers (–––) and (B) resolution forthe 400/800 bp (�), 800/1200 bp (�) and PCR-1/PCR-2(�) peak pairs. The CEC conditions are similar to thoseused in Fig. 5.

We also looked at the effects of the organic compositionin the run buffer on other chromatographic parameterssuch as selectivity, plate numbers and resolution (Fig. 6).The selectivity factors were calculated for the peak pairsof 400/800 bp and 800/1200 bp fragments from the sizingladder and for PCR-1/PCR-2 as well. The selectivity wassimply calculated from the ratio of the two retention timesof the peaks within a pair with the later eluting peak’sretention time placed in the numerator of this ratio. Theplate numbers were calculated from the retention timeand width of the peak at the base using the assumptionof purely Gaussian peak shapes. Resolution values (R)were calculated from the plate numbers, retention andselectivity factors using the equation:

R � N1�2

2

� ��� 1�� 1

� �k�

ave

k�ave � 1

� �(5)

where k’ave is the average retention factor for the particu-lar peak pair, N is the plate number and � is the selectiv-ity. As can be seen from the data in Fig. 6A, the selectivityfor all three peak pairs show a maximum at �20% aceto-

nitrile, while the plate numbers decreased as the acetoni-trile concentration increased, with the maximum platenumbers occurring at 5% acetonitrile. In all cases, thelonger DNA fragments (800/1200 bp) produced largerplate numbers due to their higher affinity for the stationaryphase (see Fig. 5). The fact that the plate numbers werehigher for these larger DNA fragments also suggeststhat mass transfer artifacts into and out of the stationaryphase contributed minimally to the dispersion of thebands, most likely a direct consequence of the fact thatthe stationary phase is a monolayer. Inspection of Fig. 6Balso shows that R decreased continuously as the amountof acetonitrile was increased due mostly to the loss ofplate numbers at these higher acetonitrile compositionsas well as reductions in the retention factor.

We also monitored the S/N ratio for several peaks in thecapillary electrochromatogram as a function of the com-position of the running buffer. We found no substantialchanges in the S/N ratio as the amount of acetonitrilewas increased in the running buffer (data not shown)when keeping the ionic strength constant. However, wenoticed that the S/N did degrade slightly when the ion-pairing agent concentration was increased in the runningbuffer, but this increase did improve the dynamic rangeof the response. Therefore, it seems that the compositionof the run buffer in terms of the percentage of organicmodifier can be altered to optimize the chromatographicperformance without significantly affecting the responseelicited by the conductivity detector.

4 Concluding remarks

RP-IPCEC can be used as an effective tool for the isola-tion and/or purification of PCR products. Due to the freedraining behavior of the dsDNA complexes, the separa-tion is affected predominately by partitioning into the sta-tionary phase, which in this case was a C18 wall-coatedsurface. Chemically modifying the surface of a PMMAmicrochip with a C18 stationary phase provided a protec-tive layer for PMMA against the organic solvent (acetoni-trile) and could effectively separate DNAs in the sizerange of 400–2000 bp in an open channel when thefield strength was low. By varying the solvent strength(increasing the organic modifier concentration), the chro-matographic performance could be altered, with maxi-mum performance, as indicated by the plate numbersand chromatographic resolution, afforded by low amountsof organic modifier. However, chromatographic develop-ment times did decrease with increasing solvent strengthdue not only to the higher elution powers of the mobilephase but also due to increases in the electroosmoticmobility.

3768 M. Galloway and S. A. Soper Electrophoresis 2002, 23, 3760–3768

We thank the National Institutes of Health (R01-HG01499)for financial support of this research. We also would liketo thank CAMD for technical assistance in performing theX-ray lithography, Dr. Wieslaw Stryjewski for helpful dis-cussions on Labview program writing, Don Patterson forthe construction of the conductance detector electronicsand James Barrow for hot-embossing the PMMA.

Received July 17, 2002

5 References

[1] Saiki, R. K., Scharf, S., Faloona, F., Mullis, K. B., Horn, G. T.,Erlich, H. A., Arnheim, N., Science 1985, 230, 1350–1854.

[2] Mullis, K. B., Faloona, F. A., Methods Enzymol. 1987, 155,335–350.

[3] Jacobsen, S. C., Hergenröder, R., Koutny, L. B., Warmack,R. J., Ramsey, J. M., Anal. Chem. 1994, 66, 1107–1113.

[4] Woolley, A. T., Mathies, R. A., Proc. Natl. Acad. Sci. USA1994, 91, 11348–11352.

[5] Woolley, A. T., Sensabaugh, G. F., Mathies, R. A., Anal.Chem. 1997, 69, 2181–2186.

[6] Effenhauser, C. S., Bruin, G. J. M., Paulus, A., Ehrat, M.,Anal. Chem. 1997, 69, 3451–3457.

[7] McMahon, G., Davis, E., Wogan, G. N., Proc. Natl. Acad.Sci. USA 1987, 84, 4974–4978.

[8] Ueno, K., Yeung, E. S., Anal. Chem. 1994, 66, 1424–1431.[9] Bruin, G. J. M., Börnsen, K. O., Hüsken, D., Gassmann, E.,

Widmer, H. M., Paulus, A., J. Chromatogr. A 1995, 709, 181–195.

[10] DeDionisio, L. A., Lloyd, D. H., J. Chromatogr. A 1996, 735,191–208.

[11] Kreoitman, M., Landweber, L. F., Genet. Anal. Techn. 1989,735, 84–88.

[12] Milovilovic, M., Lee, J. E., BioTechniques 1989, 7, 14–16.[13] Katz, E. D., Haff, L. A., Eksteen, R., J. Chromatogr. 1990,

512, 433–444.[14] Warren, W., Doniger, J., BioTechniques 1991, 10, 216–222.[15] Gold, P., BioTechniques 1992, 13, 132–134.

[16] Vordam, A. V., Kerschner, J., Anal. Biochem. 1986, 152,221–225.

[17] Robert, C. A., Yang, J. L., Wu, R., Methods Enzymol. 1979,68, 176–182.

[18] Kato, Y., Sasaki, M., Hashimoto, T., Murotsu, T., Fukushige,S., Matsubara, K., J. Biochem. 1984, 95, 83–86.

[19] Huber, C. G., Oefner, P. J., Bonn, G. K., J. Chromatogr. 1992,599, 113–118.

[20] Melander, W. R., Horváth, C., in: Horváth, C. (Ed.), High-Per-formance Liquid Chromatography Academic Press, NewYork 1980 Vol. 2, pp. 113–119.

[21] McWhorter, S., Soper, S. A., J. Chromatogr. A 2000, 883,1–9.

[22] Galloway, M., Stryjewski, W., Henry, A. C., Ford, S. M., Llo-pis, S., McCarley, R. L., Soper, S. A., Anal. Chem. 2002, 74,2407–2415.

[23] Singhal, P., Kuhr, W. G., Anal. Chem. 1997, 69, 3552–3557.[24] Soper, S. A., Ford, S. M., Qi, S., McCarley, R. L., Kelly, K.,

Murphy, M. C., Anal. Chem. 2000, 72, 642A–651A.[25] Henry, A. C., Tutt, T. J., Galloway M., Davidson, Y. Y.,

McWhorter, S., Soper, S. A., McCarley, R. L., Anal. Chem.2000, 72, 5331–5337.

[26] Xiaohua, H., Gordon, M. J., Zare, R. N., Anal. Chem. 1988,60, 1837–1838.

[27] Ford, S. M., Davies, J., Kar, B., Qi, S. D., McWhorter, S.,Soper, S. A., Malek, C. K., Biomech. Eng. 1999, 21, 13–21.

[28] Cahours, X., Morin, P., Dreux, M., J. Chromatogr. A 1999,845, 203–216.

[29] Crego, A. L., Martinez, J., Marina, M. L., J. Chromatogr. A2000, 869, 329–337.

[30] Banholczer, A., Pyell, U., J. Chromatogr. A 2000, 869,363–374.

[31] Stellwagen, N. C., Gelfi, C., Biopolymers 1997, 42, 687–703.[32] Lukacs, G. L., Haggie, P., Seksek, O., Lechardeur, D., Freed-

man, N., Verkman, A. S., J. Biol. Chem. 2000, 275, 1625–1629.

[33] Schmalzing, D., Adounan, A., Koutny, L., Ziaugra, L., Matsu-daira, P., Ehrlich, D., Anal. Chem. 1998, 70, 2303–2310.

[34] Moffatt, F., Cooper, P. A., Jessop, K. M., Anal. Chem. 1999,71, 1119–1124.

[35] Wright, P. B., Lister, A. S., Dorsey, J. G., Anal. Chem. 1997,69, 3251–3259.