Proc. Nati. Acad. Sci. USA
`Vol. 91, pp. 11348-11352, November 1994
`Biophysics
`
`Ultra-high-speed DNA fragment separations using microfabricated
`capillary array electrophoresis chips
`(microchenical analysis/microfabrication/confocal fluorescence detection/alelic fragment sizing/DNA sequencing)
`
`ADAM T. WOOLLEY AND RICHARD A. MATHIES*
`Department of Chemistry, University of California, Berkeley, CA 94720
`
`Communicated by Sung-Hou Kim, August 12, 1994 (received for review June 10, 1994)
`
`Capillary electrophoresis arrays have been
`ABSTRACT
`fabricated on planar glass substrates by photolithographic
`masking and chemical etching techniques. The photolitho-
`graphically defined channel patterns were etched in a gass
`substrate, and then capillaries were formed by thermally
`bonding the etched substrate to a second glass slide. High-
`resolution electrophoretic separations of OX174 Hae m DNA
`restriction fragments have been performed with these chips
`using a hydroxyethyl cellulose sieving matrix in the channels.
`DNA fragments were fluorescently labeled with dye in the
`running buffer and detected with a laser-excited, confocal
`fluorescence system. The effects of variations in the electric
`field, procedures for injection, and sizes of separation and
`injection channels (ranging from 30 to 120 #m) have been
`explored. By use ofchannels with an effective length of only 3.5
`cm, separations of #X174 Hae III DNA fragments from =70 to
`1000 bp are complete in only 120 sec. We have also demon-
`strated high-speed sizing of PCR-amplified HLA-DQa alleles.
`This work establishes methods for high-speed, high-
`throughput DNA separations on capillary array electrophore-
`sis chips.
`
`Capillary electrophoresis (CE) is a powerful method for DNA
`sequencing, forensic analysis, PCR product analysis, and
`restriction fragment sizing (1, 2). CE provides faster and
`higher-resolution separations than slab gel electrophoresis
`because higher electric fields can be applied. However, unlike
`slab gel electrophoresis, conventional CE allows analysis of
`only one sample at a time. Mathies and Huang (3) have
`introduced capillary array electrophoresis, in which separa-
`tions are performed on an array of parallel silica capillaries,
`and demonstrated that it can be used to perform high-speed,
`high-throughput DNA sequencing (4, 5) and DNA fragment
`sizing (6). This method combines the fast electrophoresis times
`of CE with the ability to analyze multiple samples in parallel.
`The underlying concept behind the approach was to increase
`the information density in electrophoresis by miniaturizing the
`"lane" dimension to =-100 ,um. The further miniaturization of
`electrophoretic separations to increase the number of lanes,
`the speed, and the throughput would be valuable in helping to
`meet the needs of the Human Genome Project (7, 8).
`The electronics industry routinely uses microfabrication to
`make circuits with features < 1 Aum in size. Microfabrication
`would allow the production of higher density capillary arrays,
`whose current density is limited by the capillary outside
`diameter (4-6). In addition, microfabrication of capillaries on
`a chip should make it feasible to produce physical assemblies
`not possible with glass fibers and to link capillaries directly to
`other devices on the chip. However, few devices for chemical
`separations have been made by microfabrication technology.
`A gas chromatograph (9) and a liquid chromatograph (10) have
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement"
`in accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`been fabricated on silicon chips, but these devices have not
`been widely used. Recently, several groups have fabricated
`individual CE devices on chips and performed capillary zone
`electrophoresis separations of fluorescent dyes (11, 12) and
`fluorescently labeled amino acids (13-15). However, it is not
`known whether high-resolution separations of DNA can be
`performed with these devices or whether multiple separation
`channels can be fabricated in a single chip.
`We were therefore interested in microfabricating CE chan-
`nels on planar glass substrates and exploring their use for
`DNA separations. We show here that photolithography and
`chemical etching can be used to make large numbers of CE
`separation channels on glass substrates. Procedures have
`been developed to fill these channels with hydroxyethyl
`cellulose (HEC) separation matrices, and we have been able
`to separate DNA restriction fragment digests on these chips
`in <2 min with excellent resolution. We have also charac-
`terized the injection techniques, the dependence of the sep-
`aration on channel geometry, and the reproducibility of
`separations. The demonstration that high-speed DNA sepa-
`rations can be performed on microfabricated CE channel
`arrays establishes the feasibility of integrated devices for
`electrophoretic DNA analysis.
`
`MATERIALS AND METHODS
`Electrophoresis Chip Fabrication. Electrophoresis chips
`were made by bonding a chemically etched glass bottom
`substrate to a drilled glass top substrate to form capillaries.
`The etched pieces were produced by coating a glass substrate
`with a photoresist film and then transferring the channel
`pattern to the film by exposure to UV radiation through a
`patterning mask. The exposed portions of the film were
`dissolved, and the remaining film was hardened by heating.
`The exposed glass was chemically etched, and then the
`etched substrate was thermally bonded to the top glass plate,
`which had access holes drilled in it.
`Fig. 1A shows the dimensions and layout of the separation
`chips. Fifteen CE devices were fabricated on each chip with
`all possible combinations of 30-, 50-, and 70-jum-wide sepa-
`ration channels and 30-, 70-, and 120-tkm cross channels. The
`separation channels connect reservoirs 2 and 4, while the
`cross channels connect reservoirs 1 and 3. Precleaned mi-
`croscope slides (75 x 50 x 1 mm, catalogue no. 12-550C;
`Fisher Scientific) made of soda lime glass were used for the
`top and bottom pieces. Four rows of 15 access holes 0.8 mm
`in diameter were drilled in the top pieces with a diamond-core
`drill. The glass pieces were first cleaned by spraying with
`H20, submerging in a bath ofhot H2SO4/H202 for 10 min, and
`then thoroughly rinsing with H20. The bottom pieces were
`dried in a furnace at 1500C for 10 min, exposed to hexameth-
`
`Abbreviations: CE, capillary electrophoresis; HEC, hydroxyethyl
`cellulose; TO, thiazole orange; T06, (NN'-tetramethylpropanedi-
`amino)propylthiazole orange.
`*To whom reprint requests should be addressed.
`
`11348
`
`Agilent Exhibit 1271
`Page 1 of 5
`
`

`

`Biophysics: Woolley and Mathies
`
`Proc. Nati. Acad. Sci. USA 91 (1994)
`
`11349
`
`A
`
`photomultiplier tube-
`
`A_
`
`i
`
`/
`
`/
`
`50
`
`(A) Schematic of the CE chip and the laser-excited,
`FIG. 1.
`confocal fluorescence detection system. The size of the features in
`the channel intersection area is exaggerated, and only every third
`channel on the chip is shown. (B) Low-magnification (x25) electron
`micrograph of a 70-pm separation channel intersected by a 120-pam
`cross channel and a buffer reservoir (type 3). (C) High-magnification
`(x500) electron micrograph of the intersection of a 50-pm separation
`channel with a 30-pm cross channel.
`
`yldisilazane vapor for 5 min, coated with a layer of Micro-
`posit S1400-31 positive photoresist (Shipley, Newton, MA)
`on a Headway photoresist spinner (6000 rpm), and then soft
`baked at 900C for 25 min.
`The photomask was designed with the computer-assisted
`design system Kic on a Sun SPARC 1 workstation and
`fabricated by Berkeley Microfabrication Laboratory staff
`with a GCA 3600F pattern generator and an APT chrome
`mask developer. The mask pattern was transferred to the
`substrate by exposing the photoresist to UV radiation
`through the mask in a Kasper contact mask aligner. The
`photoresist was developed in Microposit developer concen-
`trate (Shipley)/H20, 1:1. The substrate was hard baked at
`1500C for 60 min and then etched for 15 min in a 1:1 mixture
`of two aqueous NH4F/HF etchants (BOE 5:1 and BOE 10:1,
`J. T. Baker, Phillipsburg, NJ). Etch depth was profiled with
`an Alphastep profilometer (Tencor, Mountain View, CA) and
`was controlled by monitoring the etch time.
`After the bottom piece was etched, the film of photoresist
`was removed by immersing the slide in a mixture of hot
`H2SO4/H202 for 10 min. Prior to thermal bonding, the drilled
`top slide and etched bottom slide were again submerged in
`hot H2SO4/H202 for 10 min, rinsed thoroughly with H20,
`
`dried with N2 gas, and then aligned. The slides were thermally
`bonded in a model 6-525 programmable furnace (J. M. Ney
`Co., Yucaipa, CA) using the following temperature program:
`ramp 50C/min to 5000C and hold for 30 min, ramp 50C/min to
`5500C and hold for 30 min, ramp 50C/min to 6000C and hold
`for 2 hr, ramp -50C/min to 5500C and hold for 1 hr, ramp
`-50C/min to 5000C and hold for 30 min, and finally, cool to
`room temperature.
`Electrophoresis Procedures. Channel surfaces were coated
`by a modified version of the Hjerten coating protocol (16).
`Surfaces were derivatized by pumping a 0.4% (vol/vol)
`-methacryloxypropyltrimethoxysiane in H20
`solution of
`(pH adjusted to 3.5 with acetic acid) through the channels for
`1 hr, rinsing with H20, allowing an aqueous solution of 4%
`(wt/vol) acrylamide to polymerize for 5 min to coat the
`channel surfaces, and then rinsing with H20. Following refs.
`6 and 17, the separation matrix consisted of TAE buffer (40
`mM Tris/40 mM acetate/1 mM EDTA, pH 8.2), 0.75%
`(wt/vol) HEC (Mn, 438,000; Aquilon, Hopewell, VA), and
`either 1 pM thiazole orange (TO) or 0.1 ;&M (NN'-
`tetramethylpropanediamino)propylthiazole orange (TO6,
`ref. 18). HEC was added to TAE buffer and stirred overnight
`at room temperature. The dye was added to the HEC buffer,
`which was degassed under vacuum for 20 min, centrifuged in
`a tabletop centrifuge for 30 min, aliquoted, and then centri-
`fuged for 5 min (12,000 rpm) in a microcentrifuge. Buffer
`reservoirs were formed by inserting micropipette tips into the
`drilled holes; electrical contact was made by inserting small
`Pt wires into the buffer reservoirs.
`DNA restriction digest samples (+X174 Hae III fragments;
`New England Biolabs) were diluted in 1 mM Tris/0.1 mM
`EDTA, pH 8.2. PCR amplification was done on a hypervari-
`able region in the second exon of the HLA-DQa locus (19)
`that can be encompassed by a single 242-bp PCR amplifica-
`tion fiagment (20). A DNA sample (HLA-DQa genotype
`1.2/3) was PCR-amplified with an AmpliType HLA-DQa
`forensic DNA amplification and typing kit (Perkin-Elmer)
`and provided by George Sensabaugh of the School of Public
`Health, University of California, Berkeley. The sample was
`precipitated with ethanol and then resuspended in 1 mM
`Tris/0.1 mM EDTA, pH 8.2, prior to injection.
`The sieving matrix was vacuumed into the separation
`channel via reservoir 4. The cross channel, and the separa-
`tion channel between reservoir 2 and the cross channel, was
`filled with TAE buffer lacking HEC. The channels were
`preelectrophoresed for 10 min at 180 V/cm. Samples were
`introduced into the cross channel by rinsing and filling
`reservoir 3 first with 1 mM Tris/0.1 mM EDTA, pH 8.2;
`applying vacuum to reservoir 1; rinsing and filling reservoir
`3 with sample; and then applying vacuum to reservoir 1.
`Samples were injected by either a "stack" (21) or a "plug"'
`(Fig. 2) injection method. The stack injection involved ap-
`plying a field of 180 V/cm between reservoirs 3 and 4, with
`reservoir 3 at ground and reservoirs 1 and 2 floating. For the
`plug injections, a field of 170 V/cm was applied between
`reservoirs 1 and 3, with reservoir 3 at ground and reservoirs
`2 and 4 floating. Electrophoresis was at 180 V/cm, except
`where otherwise noted.
`Fluorescence Det i Apparatus. The detection apparatus
`was similar to that described earlier (6, 17). An excitation
`beam (1 mW, 488 nm) from an air-cooled Ar ion laser was
`passed into a confocal microscope (Axioplan, Zeiss) and
`reflected with a dichroic beam splitter (FT 510, Carl Zeiss) to
`a 40 x 0.60 n.a. objective (LD Epiplan, Carl Zeiss), which
`focused the beam to an --10-jum spot within the channel, -3.5
`cm from the intersection of the separation channel with the
`injection channel. Fluorescence was collected by the objec-
`tive, passed through the dichroic beam splitter, filtered by a
`bandpass filter (530DF30, Omega Optical, Brattleboro, VT),
`and focused on a 400-pm confocal pinhole followed by
`
`Agilent Exhibit 1271
`Page 2 of 5
`
`

`

`_,.-ZZws~~~~~~~~~~;-X
`r .10- i'
`
`>
`
`fX1Lt
`
`il]jet. 11011]r -..
`
`~~~~~~~~~~~~~
`: ~
`21)-I, \
`
`1)
`
`f-1)
`
`70
`
`As
`
`,A)(
`sec lld"l
`
`80
`lilli
`
`Vw 'We
`
`00
`
`A
`
`.i h '!i,
`
`.
`
`'u
`
`0I-- --k
`
`.1
`
`I111Wju
`
`3
`
`m~~~~~~~~~~~~1X
`
`02
`
`) (2 --9
`3
`
`a.l
`
`t,
`
`C.
`
`(Top) Electropherograms comparing the stack and plug
`FIG. 2.
`injection methods. A sample containing 4bX174 Hae III fragments at
`10 ng/,ul was injected for 1 sec in each experiment. The buffer
`consisted of the standard TAE/HEC sieving medium with 1 ,uM TO.
`These experiments were performed with a 50-Eum separation channel
`and a 30-pm cross channel. (Middle) Schematic diagram of stack
`injection method. (Bottom) Schematic diagram of plug injection
`method. The diagonal lines indicate the HEC in the separation
`channel. A signal of 8000 photons per second over background
`corresponds to 100 pg of DNA per ILI in the separation channel. [Our
`on-column detection limit (1000 photons per second over back-
`ground) is -2 fg of DNA for a typical band (1-sec width, 100-sec
`migration time), corresponding to a sample limit of detection of 100
`pg/Al for plug injections and 50 pg/ILI for stack injections.]
`
`11350
`
`Biophysics: Woolley and Mathies
`
`photomultiplier detection. Photoelectron pulses were ampli-
`fied and discriminated by a photon counter (model 1106,
`Princeton Applied Research) and counted with a PCA II data
`acquisition card (The Nucleus, Oak Ridge, TN) in a 486
`personal computer.
`
`RESULTS
`To characterize the capillary electrophoresis chips, electron
`micrographs of the channel features were obtained prior to
`bonding. Fig. 1B shows a low-magnification electron micro-
`graph of the intersection of a 70-gm separation channel with
`a 120-lim cross channel, as well as an injection reservoir. Fig.
`1C shows a higher-magnification electron micrograph of the
`intersection of a 50-tm separation channel with a 30-1m
`cross channel. The sloping sidewalls and flat bottoms of the
`channels, as well as the quality of the etch, can be seen
`clearly. Channel depth was 8 Am for a 15-min etch time;
`channel top widths were measured as 27, 47, 66, and 118 um
`for mask line widths of 10, 30, 50, and 100 lzm, respectively.
`With etch times of 15 min, features of this size were made
`reliably and uniformly. Deeper, 16-pum etches were obtained
`by increasing the etch time to 30 min, but with nonuniform
`undercutting of the photoresist.
`The development of a protocol for reliable injection of
`samples is critical. The electropherograms in Fig. 2 compare
`stack and plug injections using a 4X174 Hae III DNA sizing
`standard. In the stack injection, DNA is continuously stacked
`
`StackIn'ectIMI
`
`~
`
`~
`
`II~
`
`Proc. Natl. Acad Sci. USA 91 (1994)
`
`into the separation channel during the injection time. Sepa-
`ration of the 4X174 Hae III bands occurs in <120 sec at 180
`V/cm. The fluorescent signal is strong, but the resolution is
`not as good as in typical CE separations. For example, the
`271- and 281-bp bands, as well as the three largest bands, are
`not resolved with stack injection. In the plug injection
`method, the size of the injection zone is determined by the
`geometry of the channel intersection. With a 1-sec plug
`injection, the fluorescent signal is lower than for the stack
`injection, but the resolution is superior; the 271- and 281-bp
`fragments, as well as the three largest fragments, are re-
`solved. With plug injections, separations as good as those
`obtained with conventional CE can be completed in <2 min,
`using an effective separation distance of only 3.5 cm!
`Fig. 3 presents electropherograms obtained with four dif-
`ferent channel geometries to explore their effects on the
`injection and separation. In the electropherogram obtained
`with a 30-.um separation and 30-Ium cross channel, the 271- and
`281-bp fragments are not resolved, nor are the 1078- and
`1353-bp fragments. Generally, it was difficult to fill 30-pum
`separation channels with the HEC solution by vacuum and to
`obtain reproducible separations with any 30-ium separation
`channels. Thus, to see any signal at all, it was necessary to
`perform 5-sec plug injections of DNA at 100 ng/pl. With a
`50-pam separation and 30-um cross channel, all fragments were
`observed and resolved. The better performance allowed us to
`achieve satisfactory signal strength with 10 times less DNA
`and only a 1-sec injection. In the separation with a 50-pm
`separation and 120-pm cross channel, all peaks are resolved
`except for the 271- and 281-bp fragments. Separations per-
`formed with a cross channel more than twice the width of the
`separation channel (such as this one) did not give reproducible
`
`A. 30x30
`
`60
`
`70
`
`80
`
`90
`
`100
`
`.V-
`
`16-
`
`12
`
`Q
`
`3
`
`0.
`
`21
`
`,0-
`
`80
`70
`C. 50 x120 gm
`
`90
`
`100
`
`110
`
`1
`
`0 x u 00
`
`I1
`
`11
`
`I
`
`LIiTt
`
`6-i A l
`-,JL.4vrI-UL-
`2
`L
`K180
`
`L
`
`2
`300-
`
`200-
`10o
`
`100
`80
`D. 70x 120 gm
`
`120
`
`140
`
`160
`
`180
`
`100
`
`120
`160
`140
`Time (seconds)
`
`200
`
`Electropherograms obtained with various separation and
`FIG. 3.
`cross channel sizes. Samples were Hae III digests of 40X174 phage
`DNA. (A) Separation channel, 30 /Lm; cross channel, 30 pm; sample
`(plug injected for 5 sec), 100 ng/pul; separation in the presence of 1
`,uM TO. (B) Separation channel, 50 pm; cross channel, 30 ,.m;
`sample (plug injected for 1 sec), 10 ng/fkl; separation in the presence
`of 1 )zM TO. (C) Separation channel, 50 jum; cross channel, 120 um;
`sample (stack injected for 1 sec), 10 ng/gl; separation in the presence
`of 0.1 AM T06. (D) Separation channel, 70 pum; cross channel, 120
`,um; sample (plug injected for 1 sec), 10 ng/Ad; separation in the
`presence of0.1 IuM TO6. Sensitivities of DNA detection with TO and
`T06 are comparable at the concentrations used.
`
`Agilent Exhibit 1271
`Page 3 of 5
`
`

`

`Biophysics: Woofley and Mathies
`migration times, and the channel current decreased with each
`successive run. We attribute this to the dilution of the ionic
`strength and the HEC in the separation channel by the
`lower-ionic-strength solution in the cross channel, which
`produced longer separation times and lowered the efficiency of
`the electrokinetic injection. The bottom electropherogram,
`obtained with a 70-Am separation and 120-pm cross channel,
`has all the bands well-resolved. The high signal strength in this
`electropherogram can be attributed to the fact that this chan-
`nel's surface was coatedjust prior to use, which minimized the
`loss of dye and DNA by adsorption to the surfaces of the
`channels (17).
`Fig. 4 illustrates the effect of the electric field on CE chip
`separations. The electropherograms obtained at 100 and 140
`V/cm exhibit baseline resolution of the 1078- and 1353-bp
`fragments, while the 180-V/cm separation exhibits nearly
`baseline resolution of those fragments. The resolution of the
`271- and 281-bp fragments is little affected by the electric
`field. In summary, the resolution of the 4X174 Hae III
`fragments is not affected significantly by the electric field for
`these field strengths, but the separation is much faster at 180
`V/cm.
`To examine the reproducibility and stability of the chan-
`nels, multiple runs were performed on the same channel. Fig.
`5 shows the first three and the last three in a series of eight
`consecutive separations of 4PX174 Hae III fragments. All runs
`were performed in a 50-pkm separation channel with a 30-pim
`cross channel. The same HEC in the separation channel
`continued to separate the DNA fragments reproducibly, even
`after eight runs. The 271- and 281-bp fragments were partially
`resolved in all experiments, with the resolution improving
`when the sampling rate was increased from 5 Hz (runs 1-3)
`to 10 Hz (runs 6-8). Although the migration times and the
`fluorescent signal of the fragments varied slightly from run to
`run, the reproducibility of the separations was excellent. The
`small variations in mobilities (2-3% relative standard devia-
`tion) are typical for multiple capillary runs (6) and are
`
`180 V/cm
`
`140 V/cmn
`
`200-
`
`100-
`
`0-
`
`80
`
`0
`
`00
`
`0-
`
`40-
`
`50
`
`100
`
`150
`
`200
`300
`250
`Time (seconds)
`
`350
`
`400
`
`450
`
`Effect of electric field on CE chip separations. Samples
`FIG. 4.
`contained 4X174 Hae III fragments at 10 ng/gl; the running buffer
`consisted of the standard TAE/HEC with 0.1 AM T06. Experiments
`were performed with a 70-pum separation channel and a 120-Aum cross
`channel.
`
`Proc. Natl. Acad. Sci. USA 91 (1994)
`
`11351
`
`6020 1
`
`1
`
`0~~~~~~~~~~
`
`Time (seconds)
`
`FIG. 5. The first three and the last three in a series of eight
`consecutive separations obtained with the same channel. The sep-
`arations were performed with a 50-pgm separation channel and a
`30-ptm cross channel, with 1.0 puM TO in the TAE/HEC running
`buffer; the sample was a 4X174 Hae III digest at 10 ng/pl. Data
`points were collected at a sampling rate of 5 Hz for runs 1-3 and at
`10 Hz for runs 6-8.
`
`attributed to the ionic strength and HEC dilution effects
`mentioned above. We have performed up to 75 separations in
`a channel with a single HEC filling.
`
`DISCUSSION
`We have demonstrated that high-speed DNA fragment sep-
`arations can be performed with capillary arrays microfabri-
`cated on glass chips. Electrophoresis ofa restriction firgment
`digest on a 3.5-cm microfabricated channel exhibits resolu-
`tion as good as that obtainable with fiber capillaries that are
`10 times longer. Electrophoretic separations from 72 to 1353
`bp are complete in only 120 sec, -10 times faster than with
`typical CE. We have also characterized two different injec-
`tion methods, the effects of channel geometry and electric
`field, and selected conditions and parameters that lead to
`reliable devices. Sizing with CE chips is as fast as fluores-
`cence burst sizing by flow cytometry (22) but is also appli-
`cable to DNA fragments much smaller than the current lower
`limit of -1000 bp with the fluorescence burst methodology.
`In our analysis of the effects of channel geometry, we
`found it easier to fill wide (>50 pum) separation channels with
`the TAE/HEC sieving buffer. When the cross channel was
`more than about twice as wide as the separation channel, the
`devices generally had short usable lifetimes, irreproducible
`mobilities, and lower signal strengths. These effects are most
`likely caused by dilution of the buffer in the separation
`channel by the lower-ionic-strength solution in the cross
`channel. Wide (50 and 70 pgm) separation channels combined
`with narrow (30 pum) cross channels gave the most reproduc-
`ible separations over the longest periods of time.
`
`Agilent Exhibit 1271
`Page 4 of 5
`
`

`

`11352
`
`Biophysics: Woolley and Mathies
`
`Proc. Natl. Acad Sci. USA 91 (1994)
`
`tion of integrated devices that incorporate DNA preparation,
`amplification, and analysis on a single chip. Coupling of our
`technology with recent developments in photolithographic
`DNA synthesis (28) and microfabricated cell analysis devices
`(29) should lead to powerful microchemical DNA analysis
`systems.
`
`We thank George Sensabaugh for providing the PCR-amplified
`HLA-DQa DNA sample and the members of the Berkeley High-
`Sensitivity DNA Analysis Project for many valuable interactions.
`Microfabrication was performed at the University of California,
`Berkeley, Microfabrication Laboratory. This research was sup-
`ported by the Director, Office of Energy Research, Office of Health
`and Environmental Research of the U.S. Department of Energy
`under contract DE-FG-91ER61125. A.T.W. was supported in part by
`a fellowship from the Fannie and John Hertz Foundation and in part
`by a National Science Foundation predoctoral fellowship.
`
`1.
`
`2.
`
`3.
`
`4.
`
`5.
`
`6.
`
`7.
`
`8.
`9.
`
`10.
`
`11.
`
`12.
`
`13.
`
`14.
`
`15.
`16.
`17.
`
`18.
`
`19.
`
`20.
`
`21.
`22.
`
`23.
`
`24.
`
`25.
`
`26.
`
`27.
`
`28.
`
`29.
`
`Landers, J. P., Oda, R. P., Spelsberg, T. C., Nolan, J. A. &
`Ulfelder, K. J. (1993) BioTechniques 14, 98-111.
`Dubrow, R. S. (1992) in Capillary Electrophoresis: Theory and
`Practice, eds. Grossman, P. D. & Colburn, J. C. (Harcourt
`Brace Jovanovich, San Diego), pp. 133-157.
`Mathies, R. A. & Huang, X. C. (1992) Nature (London) 359,
`167-169.
`Huang, X. C., Quesada, M. A. & Mathies, R. A. (1992) Anal.
`Chem. 64, 967-972.
`Huang, X. C., Quesada, M. A. & Mathies, R. A. (1992) Anal.
`Chem. 64, 2149-2154.
`Clark, S. M. & Mathies, R. A. (1993) Anal. Biochem. 215,
`163-170.
`Hunkapiller, T., Kaiser, R. J., Koop, B. F. & Hood, L. (1991)
`Science 254, 59-67.
`Smith, L. M. (1993) Science 262, 530-532.
`Terry, S. C., Jerman, J. H. & Angell, J. B. (1979) IEEE Trans.
`Electron Devices ED-26, 1880-1886.
`Manz, A., Miyahara, Y., Miura, J., Watanabe, Y., Miyagi, H.
`& Sato, K. (1990) Sens. Actuators Bi, 249-255.
`Manz, A., Harrison, D. J., Verpoorte, E. M. J., Fettinger,
`J. C., Paulus, A., Ludi, H. & Widmer, H. M. (1992) J. Chro-
`matogr. 593, 253-258.
`Jacobson, S. C., Hergenroeder, R., Koutny, L. B. & Ramsey,
`J. M. (1994) Anal. Chem. 66, 1114-1118.
`Harrison, D. J., Fluri, K., Seiler, K., Fan, Z., Effenhauser,
`C. S. & Manz, A. (1993) Science 261, 895-897.
`Effenhauser, C. S., Manz, A. & Widmer, H. M. (1993) Anal.
`Chem. 65, 2637-2642.
`Fan, Z. H. & Harrison, D. J. (1994) Anal. Chem. 66, 177-184.
`Hjerten, S. (1985) J. Chromatogr. 347, 191-198.
`Zhu, H., Clark, S. M., Benson, S. C., Rye, H. S., Glazer,
`A. N. & Mathies, R. A. (1994) Anal. Chem. 66, 1941-1948.
`Benson, S. C., Mathies, R. A. & Glazer, A. N. (1993) Nucleic
`Acids Res. 21, 5720-5726.
`Horn, G. T., Bugawan, T. L., Long, C. M. & Erlich, H. A.
`(1988) Proc. Natl. Acad. Sci. USA 85, 6012-6016.
`Saiki, R. K., Walsh, P. S., Levenson, C. H. & Erlich, H. A.
`(1989) Proc. NatI. Acad. Sci. USA 86, 6230-6234.
`Burgi, D. S. & Chien, R.-L. (1991) Anal. Chem. 63, 2042-2047.
`Goodwin, P. M., Johnson, M. E., Martin, J. C., Ambrose,
`W. P., Marrone, B. L., Jett, J. H. & Keller, R. A. (1993)
`Nucleic Acids Res. 21, 803-806.
`Luckey, J. A., Norris, T. B. & Smith, L. M. (1993) J. Phys.
`Chem. 97, 3067-3075.
`Huang, X., Coleman, W. F. & Zare, R. N. (1989) J. Chro-
`matogr. 480, 95-110.
`Delinger, S. L. & Davis, J. M. (1992) Anal. Chem. 64, 1947-
`1959.
`Jacobson, S. C., Hergenroeder, R., Koutny, L. B., Warmack,
`R. J. & Ramsey, J. M. (1994) Anal. Chem. 66, 1107-1113.
`Puers, C., Hammond, H. A., Jin, L., Caskey, C. T. &
`Schumm, J. W. (1993) Am. J. Hum. Genet. 53, 953-958.
`Pease, A. C., Solas, D., Sullivan, E. J., Cronin, M. T.,
`Holmes, C. P. & Fodor, S. P. A. (1994) Proc. Natl. Acad. Sci.
`USA 91, 5022-5026.
`Parce, J. W., Owicki, J. C., Kercso, K. M., Sigal, G. B.,
`Wada, H. G., Muir, V. C., Bousse, L. J., Ross, K. L., Sikic,
`B. I. & McConnell, H. M. (1989) Science 246, 243-247.
`
`28-
`
`x2
`
`t
`
`-t
`
`; 12
`Cs ;
`.l en
`Rs
`z
`(I
`
`m
`
`8
`
`00
`C4
`r-
`r-
`00 O
`
`V
`
`° 24-
`x
`
`o 20
`
`16-
`
`1 1
`
`40
`
`80
`
`90
`
`100
`
`110
`Time (seconds)
`
`120
`
`High-speed sizing of PCR-amplified DNA fragments on a
`FIG. 6.
`capillary array electrophoresis chip. The electropherogram shows
`the separation of a mixture of standard 4X174 Hae III fragments
`spiked with PCR-amplified HLA-DQa DNA. The standard TAE/
`HEC buffer, containing 0.1 gM T06, was used as the sieving matrix.
`A comparison of the stack and the plug injection methods
`shows the advantages and disadvantages of each method.
`The stack method gives more signal than the plug injection
`method because more sample is electrokinetically loaded into
`the separation channel. With a 1-sec stack injection, the
`number of theoretical plates obtained for the 234-bp fragment
`was 6.0 x 104, corresponding to a plate height of 0.58 pm.
`With a 1-sec plug injection, the number of theoretical plates
`for the 234-bp fragment was calculated to be 1.3 x 105,
`corresponding to a plate height of 0.27 gtm. With published
`values for diffusion coefficients of DNA in gels (23), the
`widths of the initial injection plugs were estimated (15) as 480
`,um for the 1-sec stack injection and 330 gm for the 1-sec plug
`injection. These widths are consistent with previous work
`(15, 24, 25). Under our conditions, the injection plug width is
`the limiting factor for the number of theoretical plates. The
`width of the injection plug can be decreased by controlling the
`potentials at all the channels in the junction (26). The opti-
`mum method of sample injection, accordingly, will depend on
`whether signal or resolution is more important. For example,
`to perform chip-based separations requiring extremely high
`resolution, such as DNA sequencing, minimizing the length
`of the injection plug will be critical. Alternatively, the highest
`sensitivity (50-pg/tdl sample limit of detection) is achieved
`with stack injection.
`Now that high-speed DNA separations have been demon-
`strated on capillary array electrophoresis chips, a variety of
`extensions of this technique can be envisioned. It is feasible
`to fabricate up to -80 independent separation and loading
`channels on a single chip with our current channel geometry
`and lengths. This number is primarily limited by the place-
`ment and size of the access holes. If methods can be
`developed for facile loading of multiple samples, even higher
`densities of channels should be feasible. Capillary arrays on
`chips should be useful for rapid, parallel sizing of PCR
`products for genetic analysis and forensic identification. For
`example, Fig. 6 shows a separation of a mixture of a 4X174
`Hae III standard and a solution containing the HLA-DQa
`PCR product. The PCR product (shaded) was detected at
`about 90 sec and estimated to be 256 bp by using the 4X174
`Hae III fragment mobilities. This establishes the feasibility of
`performing rapid DNA typing of, for example, the
`HUMTHO1 locus, with our capillary array electrophoresis
`chips (27). Microfabrication should also allow the construc-
`
`Agilent Exhibit 1271
`Page 5 of 5
`
`