Anal. Chem. 1995, 67,2974-2980
Design of a High=PrecisionFraction Collector for Capillary Electrophoresis Odilo Muller, Frantlsek Foret,t and Barry L. Karger*
Bamett Institute, Nottheastem University, Boston, Massachusetts 02 7 15
A high-precision fraction collector for capillary electrophoresis has been developed. The device utilizes detection close to the end of the capillary and a sheath liquid at the exit of the capillary, allowing continuous collection (Le., unintenupted applied electrical field) of multiple species. The role of the sheath liquid flow rate and position of detection in the column on the collection precision was assessed. Fiber-optic detection at -1 cm before the exit end of the capillary was found effective for precise timing of the collection. Up to 60 fractions of microliter or smaller volumes could be automatically collected into capillaries used as collection vials. The collection capillaries were placed on a cylinder, and a computer-controlledstepping motor w e d the appropriate capillary with the column exit. The eSeCtiveness of the fraction collector was demonstrated in the collection of all 11 fragments of the HueIII restriction digest of QX-174 plasmid DNA Polymerase chain reaction amplification of the 271 and 281 bp fragments revealed an inversion of the size-dependent migration order. The ability to identify and characterize separated bands in capillary electrophoresis (CE) is important for the full development of this rapidly growing method. On-line coupling to structure elucidating detectors such as mass spectrometry (CE/MS) is one obvious approach. A second method is collection of separated fractions for purification of complex mixtures and further characterization (e.g., sequencing, enzyme digestion, blotting, biological function, etc.). At first glance, fraction collection in CE may appear impractical, given the small quantities of material separated. Indeed, since nanogram or lower quantities may be handled, the term "nanopreparative"may be appropriate to classify this operation. However, in the case where DNA fragments of short to moderate length (Le., less than a few kilobase pairs) are considered, small collected amounts do not represent a problem, since polymerase chain reaction (PCR) amplification is readily available when necessary. Moreover, even in the case of small molecules, proteins, and carbohydrates, where such amplification procedures do not exist, the collected amounts are still sufficient for protein microsequencing, immunoassays, enzyme digestion, blotting, etc. Several designs of CE collection devices have already been described. In an early approach, a sweep liquid through a standard liquid chromatographic detector was employed.' Since the HPLC connections and the detection cell were of relatively large dead volume, the performance of this device was not suitable for high-resolution CE separations. In later sample collection ' On leave from the Institute of Analytical Chemistry. Bmo. Czech Republic. (1)Hjerten. S.;Zhu, M.-D. J. Chromatogr. 1985,327,157-164. 2974 Analytical Chemistry, Vol. 67, No. 17, September 1, 7995
systems, a dedicated instrument with a set of collection vials was used.2 Whenever a separated zone was calculated to exit the capillary, the separation current was turned off, and the capillary was moved into a collection vial containing a small volume of buffer and an electrode. By turning the separation current on, the ions of interest were collected.2 This procedure, described as electroelution,is presently the most often used procedure, since it is readily adaptable to commercial instrumentationor laboratorybuilt systems.3-'3 Although quite simple, electroelution has, however, some inherent disadvantages. First, the need to interrupt the field each time the fraction is collected makes the procedure potentially imprecise, especially when many fractions are to be collected. Second, a relatively large volume of the collection buffer (typically 10 pL or more) is used, resulting in significant sample dilution. Another approach employs pressure to exit the zone from the column, once the zone is close to the end of the ~apillaty.'~-~~ Again, the separation current must be turned off and the end of the capillary moved into a collection vial. As in the previous approach, potential loss in resolution and difficulty in precise collection of multiple bands within a run may be limiting factors. Other collection procedures include the use of moving mem~ * ~ glass connections,or an on-column brane~'~*'~ or d r ~ m s , 'porous frit s t r u ~ t u r e . ~These ~ - ~ ~approaches have the advantage that all zones can be continuously collected without the need to interrupt the electric current. Potential disadvantages, however, stem from (2)Rose, D.J.; lorgenson, J. W. J. Chromatogr. 1988,438,23-24. (3)Albin, M.; Chen, S.-M.; Louie, A; Pairaud. C.; Colbum, J.; Wiktorowicz,Anal. Biochem. 1992,206,382-388. (4)Atria, K D.; Dave Y. K. J. Chromatogr. 1993,633,221-225. (5) Banke, N.;Hansen, IC; Diers, I. J. Chromatogr. 1991,559,325-335. (6) Bergman. T.; Agerberth, B.: Jomvall, H. FEES Lett. 1991,283,100-103. (7)Camillen, P.; Okafo, G. N.; Southan, C.; Brown, R Anal. Biochem. 1991, 198,36-42. (8)Guzman, N. A;Trebilcock, M. A; Advis, J. P. Anal. Chim. Acta 1991,249, 247-255. (9)Lecoq, A F.;Di Biase, S.; Montanarella, L.1. Chromatogr. 1993,638,363373. (10)Nashabeh, W.; Smith, J. T.: El Rassi, 2. Electrophoresis 1993,14,407-416. (11)Takigiku, R;Keough, T.; Lacey, M. P.; Schneider, R E. Rapid Commun. Mass Spectrom. 1990,4, 24-29. (12)Schwer, C.: Lottspeich, F.J. Chromatogr. 1992,623,345-355. (13)Guttman, A; Cohen, k S.; Heiger. D. N.; Karger, B. L. Anal. Chem. 1990, 2,137-141. (14)Camillen, P.;Okafo, G. N.; Southan, C. Anal. Biochem. 1991, 196,178182. (15)Herold, M.; Wu, S. L. LC/CC 1994,12 (7), 531-533. (16)Grimm, R;Ross, G.; Herold, M.; Bek, F. Presented at HPCE '95,Wurzburg, Germany, Jan 29-Feb 2,1995 Poster P-633. (17)Cheng, Y.-F.; Fuchs, M.; Andrews, D.; Carson, W. J. Chromatogr. 1992, 608,109-116. (18)Warren, W.J.; Cheng, Y. F.;Fuchs. M. LC/GC 1994, 12 (l), 22. (19)Huang, X.;Zare, R N. J. Chromatogr. 1990,516,185-189. (20)&linger, L. Electrophoresis. A Survey of Techniques and Applications;Joumal of Chromatography Library 18 Elsevier: Amsterdam, 1979. 0003-2700/95/0367-2974$9.00/0 0 1995 American Chemical Society
sheath liquid
1
A
septum
I
h I
7
to collection vials
CE capillary
light out Flgure I. Exploded view of the sheath flow connection. For further details, see Experimental Section.
the deposition of the zones on the solid surface (membrane) and the need for electroosmotic bulk flow (e.g., for porous glass, frit structure), Recently, the use of a coaxial sheath liquid interface, similar to that developed for coupling of CE to electrospray mass spectrometry, has been suggested for fraction c~llection.*~**~ In this case, similar to the first CE collection technique,' multiple zones can be collected without interruption of the electtic current, since the high-voltage electrode is in contact with the sheath buffer. As the separated bands exit the CE column, the sheath buffer transports the species into appropriate collection vials. An advantage of this approach is continuous operation and an effective isolation of the electrode from the collection end of the capillary. The goal of this work was to design a precise automated fraction collector suitable for any mode of CE, that could be used for continuous collection of a large number of zones, with less dilution than that for collection in a vial. A sheath liquid arrangement was selected to allow for continuous collection. Glass capillaries were used for collection to simplify handling of small liquid volumes. Precise collection was achieved by detecting the analyte zones close to the end of the CE capillary, employing a fiber-optic-based UV detection cell. The detection signal was used as input for computer control of the collection procedure. The performance of the device has been evaluated in the isolation of the 11fragments of the Hue111 digest of OX-174 plasmid DNA, with subsequent identification of several collected fragments by PCR using specific primers. (21) Fujimoto, C.; Fujikawa, T.; Jinno, K. J High Resolut. Chmmatogr. 1992. 15, 201-203. (22) Fujimoto, C.; Muramatsu, Y.; Suzuki, M.; Jinno, K. J. High Resolut. Chromatogr. 1991,14, 178-180. (23) Huang, X; Zare, R N. Anal. Chem. 1990,62,443-446. (24) Schwer, C.; Lottspeich, F. Presented at HPCE '95, Wiirzburg, Germany, Jan 29-Feb 2, 1995: Poster P-140. (25) Weinmann, W.; Parker, C. E.; Deterding, L. J.; Papac, D. I.; Hoyes, J.; Przybylski, M.; Tomer, K. B. J. Chromatogr. A 1994,680,353-361.
EXPERIMENTAL SECTION Instrumentation. A laboratory-built instrument was employed in this work, using a high-voltage power supply (CZE 1OOOR; Spellman, Plainview, NY) and a m o ~ e liquid d chromate graphic UV detector (LC 90; Perkin-Elmer, Cupertino, CA) equipped with optical fibers. (A safety lock system can be used with this instrument as well as the collection device to be described below.) Siice the intensity of the detector UV source (deuterium lamp) was not sufficient to obtain good signal to noise ratio, an external mercury UV lamp (Model SC-1; UVP, San Gabriel, CA) served as a light source at a wavelength of 254 nm. This lamp provided -100 times higher intensity compared to the original source. The performance of the detector equipped with the optical fibers was practically identical to the original specification of the instrument, with the noise level of 5 x AU. No focusing optics were necessary for coupling the lamp and optical fibers. The 254 nm line was isolated by a narrow band pass filter (10 nm; Barr Associates,Westford, MA) and guided by two optical fibers cFvp300 330 360; Polymicro Technologies, Phoenix, AZ), one each for the sample and reference beams, respectively. The length of the fibers, shielded from the ambient light with black shrinkable tubing, was -1.5 m. For proper alignment, stainless steel tubes (15 mm pieces of Hypo S/S 316 21G4 Small Parts, Miami Lakes, FL) were used as guides for both the fibers and the CE capillary. Once aligned, the steel tubes were permanently glued by a quick-set epoxy to the face plate of the collection interface (see Figure 1). The original flow cell of the UV detector was replaced by a holder positioning the sample and reference optical fibers in front of the photodiodes inside the detector. The detection signal was amplified 10-fold to match the input range ( f 5 V) of the A/D board (AT-MIO-16L9National Instruments, Austin, TX) and analyzed by data acquisition software based on Labview (National Instruments). The software included on-line data processing and control of a stepper motor (SAS; Hurst, Princeton, IN), which was connected to the computer by means Analytical Chemistty, Vol. 67, No. 17, September 1, 1995
2975
waste (to vacuum)
Figure 2.
of a modified controller (EPC-015; Hurst). This device will be described in detail in the Results and Discussion section. A second CE system was used to analyze the collected DNA fragments and was equipped with a laser-induced fluorescence detector.26 For fluorescence detection, ethidium bromide (Sigma, St. Louis, MO) was added to both the separation matrix and the anodic buffer reservoir at a concentration of 1pg/mL. A He-Ne laser (PMS Electrooptics, Boulder, CO) was used for excitation at 543 nm, and fluorescence was detected by a photomultiplier after passing 543 nm blocking and 610 nm band pass filters (Oriel, Stratford, CT). In the laser-based system, data were processed using Turbochrom software (Perkin-Elmer Nelson, Cupertino, CA) . Capillary Electrophoresis. Capillary electrophoresis was performed using either 100pm i.d. fused silica capillaries supplied by Polymicro Technologies and coated with linear polyacrylamideZ7 or 50 pm i.d. DE1 capillary columns & W Scienac, Folsom, CA). The Hue111 digest of QX-174 plasmid DNA (New England Biolabs, Beverly, MA) was separated using 1%(w/v) methylcellulose (Sigma) dissolved in 40 mM Tris/TAPS buffer 0 & W capillaries) or in 5%(w/v) linear polyacrylamide prepared by diluting a 10% (w/v) solution (Mw 700 000-1 000 OOO; Polysciences, Warrington, PA) in 2x Tridborate buffer. Electrokinetic sample injection was used throughout the study. DB-1 capillaries, 50 ,um i.d., filled with a 1%(w/v) MC solution in l x TBE, were used for the analyses on the laser-based system. Collection Device. The collection device consisted of two major parts: (a) the detection interface, comprising a fiber-optic detection cell and the sheath flow unit, and (b) the fraction collector, consisting of a holder of collection capillaries operated by a computer-controlled stepper motor. Figure 1 shows an exploded view of the detection interface. A CE separation capillary and the optical fibers were attached to
the face plate of the interface, as described above. A protection plate was slid over the CE capillary, covering the detection point. The sheath flow tee connection was machined from a 5 mm thick Plexiglas plate. The exit of the separation capillary ended flush with a collection tip made of a short piece of Teflon tubing, 5 mm x 0.5 mm i.d. x 1.6 mm 0.d. (Upchurch Scientific, Oak Harbor, WA) . The hydrophobic nature of the collection tip led to formation of small, stable droplets, without a danger of spreading the liquid around the tip. The buffer flowed into the tee connection by means of a syringe pump (Model 341B; Sage Instruments,Boston, MA) and then was forced to flow around the end of the CE capillary by a seal at the opposite exit by means of a silicone septum. To eliminate migration of electrolysis products into the liquid sheath, the ground electrode was inserted into a buffer reservoir, separated from the sheath liquid by a semipermeable membrane.28 The detection unit was assembled in a modular fashion and fixed with two screws. The collection unit consisted of an aluminum cylinder with precisely machined grooves to accommodate 60 collection capillaries. This holder was attached to a stepper motor, enabling precise positioning of the collection capillaries in front of the collection tip. The first position on the cylinder accommodated a fused silica capillary connected to a diaphragm pump (Thomas Scientiiic,Swedesboro, NJ) for the waste removal of sheath liquid prior to collection. When a collection capillary was aligned with the collection tip (-300 pm spacing), the small droplets containing the eluted sample fractions were taken up by capillary action. The motion of the stepper motor was directed by a controller, connected through an optical coupler to the analog output of the data acquisition board. The analog output signal was created by a laboratory-written program in the Labview environment Figure 2 shows the complete design of the collection device.
(26) Chen, D. Y.; Swerdlow, H. P.; Harke, H. R.; Zhang, J. Z.; Dovichi N. J. J. Chromatogr. 1991,559, 237-246. (27) Hjerten, S. J Chromatogr. 1985,347, 191-198.
(28) Everaets, F. M.; Becker, J. L: Verheggen. T.P. E. M. Isotuchophoresk-nteoy, Instrumentation and Applications; Journal of Chromatography Library 6 Elsevier: Amsterdam, 1976.
-
2976 Analytical Chemistry, Vol. 67, No. 17, September 1, 1995
The s o h a r e controlling the stepper motor consisted of two parallel working loops. The b t loop read the voltage signal from the detector at a data acquisition frequency of 2 points/s. Incoming voltage data were compared to a threshold value which could be set before or during the run. If the detector output exceeded the threshold value, the elapsed time since the start of the run was read and transferred to the second loop. The second loop calculated the exit time of the detected species, according to the migration time to the detection window and the capillary dimensions. After the calculated time had elapsed, an analog signal (square wave) was sent to the controller of the stepper motor to move a collection capillary to the desired position. Furthermore, the collection could be executed in 2-10 s steps, in order to split one zone into several aliquots or heartcut unresolved bands. The software also allowed correction for systematic and random errors by widening and shifting the collection window (see Results and Discussion). PCR For identification of collected DNA fragments (271,281, and 310 bp), PCR with fragment-specificprimers was carried out on a Perkin-Elmer Cetus thermal cycler (Perkin-Elmer, Norwalk, CT). The PCR mixture had a final volume of 100p L and contained sterile PCR buffer, Pfu polymerase (both from Stratagene Cloning Systems, La Jolla, CA), nucleotides (Perkin-Elmer), and customsynthesized primers (Ransom Hill Biosciences, Ramona, CA). Primers were designed using Oligo software (National Biosciences Inc., Plymouth, MN). The DNA fragments were amplified for 25 cycles. The melting temperature was set to 95 "C, annealing temperature 48 "C, and extension temperature 72 "C. Each temperature was maintained for 45 s. The product lengths were 115 bp for the 271 bp fragment, 96 bp for the 281 bp fragment, and 147 bp for the 310 bp fragment. RESULTS AND DISCUSSION The goal of this work was to design a precise and broadly applicable collection device that could maintain the resolving power of all modes of CE. To preserve resolution and purity of the collected fractions, the interface had to minimize cross contamination during collection and also collect fractions in a reproducible manner. A key factor in reaching these goals was that the exit times of the separated zones be precisely known so that any desired zone or fraction of a zone could be isolated. Collection Precision. In HPCE, high-speed separations are frequently performed, and narrow peak widths in the order of 2-5 s are commonly achieved. Therefore, the precise determination of the exit time of the sample is an impojant issue in order to allow accurate collection during high-speed separations. If zones are detected on-column before leaving the capillary, their exit times can be calculated from the migration velocity and known distance between the detection point and the end of the capillary. Since there is always a variation in migration velocity in CE, the calculated exit time will be imprecise. The extent of this imprecision will be iduenced by factors such as the quality of the capillary, electroosmoticflow, temperature,pH changes during the run, etc., as well as the migration time and width of the peak. The absolute value of the variation of the exit time will clearly be related to the actual distance between the detection point and the exit of the capillary. For a detection point placed at a distance X from the exit of the capillary,the time te to travel the distance X can be expressed as
where I is the distance from injection to detection point and is the migration time to the detector. Since the variation in the migration velocity in CE is typically on the order of f l - a , the uncertainty of the exit time, Ate, will be equal to ~(0.01-O.02)te. Clearly, the shorter X , the lower will be Ate. As an example, consider a 30 cm long capillary and a zone with a migration time of 10 min. If the detection point is 10 cm from the end of the capillary, the collection uncertainty will be -6-12 s. While this value might be acceptable for low-resolution separations with relatively broad peaks (>20 s wide), the collection of closely spaced or overlapping zones of high efficiency (such as often occurs in CE analysis, peaks -5 s wide) would be difficult. The uncertainty in exit times can be decreased by placing the detection point close to the exit end of the capillary. If, for example, the distance Xis decreased to 1 cm, the uncertainty would decrease proportionately to 0.6-1.2 s. A convenient means to detect close to the end of the capillary is to use an optical fiber to direct the source light on the column. In this work, optical fibers were placed 1 cm from the capillary exit, resulting in a collection precision of -2 s. When sieving matrices were employed, a systematic decrease of the migration times in the order of 2-4 s could be observed. It was found that this shift would increase in cases where the polymer solution was not replaced after each run (-9-13 s shift). This result led to the assumption that the sheath liquid diluted the separation matrix at the end of the capillary, thereby accelerating the migration velocity of the molecules near the exit of the column. This assumption is supported by the fact that the surface tension of the liquid sheath droplet creates a pressure drop from the exit end of the capillary to the injection side which is in the buffer reservoir. This pressure drop is resisted by the sieving matrix in the column; however, the pressure difference could cause the above dilution of the matrix at the end of the column. This systematic error in exit time was minimized by calibration and by replacement of the sieving matrix after each run. Of course, for open-tube CE, the effect of the pressure drop will be greater. It would thus be important to counterbalance this effect in opentube operation by raising the height of the injection side of the capillary. With electroosmotic flow, this effect will be less. One possible means of minimizing the collection time uncertainty would be to place the detector after the CE column and to detect the zones in the stream of the collection buffer. Indeed, as noted, this approach has been used for collection of peptides separated by free solution CE.24 In our study, we have found that for collection of DNA fragments separated with a sieving matrix and no electroosmotic flow, oncolumn detection provided both a higher detection signal and less band broadening. It is important to note that in order to avoid formation of moving ionic boundaries,30the sheath liquid was typically the same as the running buffer of the CE unit. In those CE procedures where the migration velocity is not constant prior to detection (e.g., ITP transient focusing, voltage programming), two optical fibers could be placed near the exit end of the capillary to allow precise prediction of the exit time. In addition, in the case of capillary IEF with pressure m0bilization,2~where all bands are (29) Chen, S. M.; Wiktorowicz, J. E. Anal. Biochem. 1992, 206 (I), 84-90, (30) Foret, F.; Thomson, T.; Vorous, P.; Karger, B. L.; Gebauer, P.; Boceck, P. Anal. Chem. 1994, 66, 4450-4458. Analvtical Chemisttv Val. 67. No. 17. Seotember 1. 1995
2977
Sheath tube
I
/
sheah flow V
CEespillsry /
I di
d, [di
%B,
ECB
I
Figure 3. Model for estimation of the minimum sheath flow necessary to avoid migration of separaied zones into the sheath tube. For details, see text.
moving with the same speed, two optical fibers would be a good approach for collection accuracy. Here, the sheath liquid would of necessity not fully match the gradient buffer system leaving the column. However, the pressurized flow should sufficiently minimize the formation of moving ionic boundaries." Sheath Flow Rate. In this work we have used sheath flow of the collection buffer for hsport of the separated zones into appropriate vessels. With no bulk electroosmotic flow in the capillary, the sheath flow rate will determine the volume and thus the dilution of the collected species. Since the CE separation current is passed through the sheath buffer, a minimum flow rate is necessary to prevent the sample ions from moving into the space between the walls of the CE capillary and the sheath tube. It is clear that the velocity of the sheath collection buffer, UCB, must be higher than the electrophoretic velocity of the sample ions, 'CB
'hECB
(2)
where p is the electrophoretic mobility of the species to be collected and ECSis the electric field strength in the sheath liquid (see F i r e 3). The flow rate, V,of the liquid sheath is given by
v= vc,s
(3)
where S is the cross section area between the liquid sheath and the separation capillary. Considering Ohm's law,
E =~/SK
(4)
(where I is the electric current and K is the conductivity) and the migration velocity, UCB, of the fastest sample ion inside the sheath tube, VCB
= @CB = N
S K C B
(5)
(where PCCB is the mobility of fastest sample molecule, ECBis the electric field strength, and KCB is the conductivity in the sheath liquid), the minimum flow rate, Vmin,necessary to avoid migration of the fastest species can be expressed as
v .= mm
p1lKCB
(6)
By substituting for I, we obtain 'mi"
2978 Analytical
= npEBGEKBGEd2/4KCB
(7)
Chemistfy, Voi. 67, No. 17, September 1, 1995
where d3 is the intemal diameter of the CE capillary, KBGE and KCB are the conductivities of the background electrolyte and collection buffer, respectively, and EBCEis the elechic field strength for the CE separation. Equation 7 predicts that the minimum flow rate of the sheath buffer depends on the inner diameter of the separation capillary, the elechic field strength across the capillary, the electrophoretic mobilities of the collected ions, and the ratio of conductivitiesof the separation and collection buffers. mote that the minimum flow rate is independent of the outer diameter of the separation capillary (62) and the inner diameter of the Teflon tubing (dJ.) To estimate the m i n i u m flow rate, consider a 100 pm i.d. separation capillary operated at 300 V/cm, collection of highly mobile ions with a mobility of 50 x 10.' cmz/Vs, and the same buffer for the collection as for separation, i.e., KBGE = KCB. For such a case, the minimum flow rate can be calculated to be 10 pL Collection into Capillaries. Small volumes of the collected fractions pose bigb demand on proper handling and storage. After several arrangements were tested with glass or plastic microvials, glass capillaries (20 pL),as supplied for bot air FCR thennocyclers (Idaho Technology, Idaho Falls, ID), were selected for DNA fraction collection. The use of capillaries instead of collection vials bas several advantages.10 First, the size of the collection device can be maintained small and yet capable of storing a number of fractions. Second, once the liquid is inside the capillary, solvent evaporation is negligible. Third, glass capillaries can be coated to minimize adsorption of collected material; indeed, coated capillaries are commercially available. Since both ends of the capillary can be sealed, the samples can easily be stored. Finally, capillary action can be used for active transport of the liquid fractions into the collection capillaries,without the need for any pumping device. Capillary action also makes the collection system more rugged, as it allows a misalignment of the collection capillaries on the order of several hundred micrometers without any loss of sample. Collection Procedure. Two methods have been applied to the sample of interest. Whenever baseline resolved peaks were to be isolated, a collection window was calculated according to the actual peak width (Figure 4.4). To compensate for random and systematic error, the collection window could be increased and even shifted. 'Typically, the collection window was increased by f2 s on Figure 44,x = 2 s). The offset, which compensated for systematic errors, was determined from appropriate calibration runs (see Collection Precision section). The stepper motor brought a collection capillary into position at tl - x and removed it at h + x. In the second method, when peaks overlapped, the rotor holding the collection capillaries could be moved in predefined steps (typically 5 s duration: see Figure 4B). After a peak was detected and the elution time calculated, as described above, the offset value was subtracted from the calculated elution time. After this time, n fractions of equal time widths were collected. In this manner, unresolved bands could be cut into several portions, and the pure fractions could be used for further analysis. Of course, more sophisticated programs could be used to h e a r t a t peaks.
4
t
A
11- x
y tl
................
4
offset
t
t+x
collection window '
,
'
14
I
'
16
,
'
I
'
,
20
18
'
I
'
24
22
migration time (min)
Figure 5. Electropherogram for the collection of 11 fragments. Conditions: 100 ,um i.d. (J & W Scientific), L = 25 cm, I = 24.1 cm, 1% (w/v) methylcellulose, 40 mM T r i f l A P S buffer, E = 167 V/cm, I = 10 PA, sheath flow collection (sheath flow = 40 mM TrislTAPS buffer), UV detection at 254 nm.
.............
I
9 ,
I
1
otfset
I
'
I
'
I
'
I
'
I
I
'
I
'
I
'
I
'
I
I
.
'
I
'
I
'
I
8
Figure 4. Collection modes for CE. (A) The peak elution times t1 and b can be calculated according to the voltage, the preset signal threshold value, and the capillary dimensions. For compensation of random and systematic error, the collection window can be widened ( h - x; f2 x) and shifted (offset). (B) Unresolved band can be cut into several portions by collecting ih predefined steps. An offset shift can compensate for systematic errors.
+
The ability to achieve this cutting will depend on the time resolution available. After the collection step, the fragments were either stored for later use inside the collection capillaries or eluted into an Eppendorf vial for reinjection. In some cases, the collected fractions were desalted with Centri-Sep columns (Princeton Separations, Adelphia, NJ) prior to the reinjection. Collection of RestrictionFragments of CPX-174 Plasmid. To illustrate the performance of the fraction collector, the DNA restriction fragment standard, (PX-l74/HueIII, was separated under two different conditions, and fractions of the sample were isolated using the peakactivated approach. The Hue111 digest of (PX-174plasmid DNA contains 11 fragments in the size range from 72 to 1356 bp. Figure 5 shows the electrophoretic separation of this sample in a 1%(w/v) methylcellulose matrix. Since the peaks were well separated, all 11 fragments could be obtained in separate capillaries. In order to determine the purity of the collected fractions, a more sensitive detection technique, i.e., laser-induced fluorescence (LIF),was used. Figure 6 shows the CELIF analysis of each of the 11 collected fragments, in which added ethidium bromide (EtBr), an intercalating dye, was excited with a He-Ne laser at 543 nm. All fragments yielded a single peak, demonstrating the usefulness of the approach for fraction collection. Only the 271 bp fragment showed a slight impurity (fifth run). Each of the sample components was now available for further analysis. Identification of Collected Fractions. It is well known that double-stranded DNA fragments may exhibit anomalous migration behavior in which larger fragments migrate faster than smaller ones in cross-linked polyacrylamide slab gels.31 This behavior,
1
I 0
2
1
6 8 migration time (min) 4
I
10
Flgure 6. Reinjection of 11 fragments collected during the run in Figure 4. Conditions: 50 pm i.d. (J & W Scientific), L = 20 cm, I = 12 cm, 1% (w/v) methylcellulose, 1x TBE buffer, E = 250 V/cm, I = 7 PA, 1 ,ug/mL EtBr, LIF detection at (5431610) nm.
related to conformational or structural effects, can also be observed in CE, where the anomaly may even be amplified by the high electric field strength used for separation. The extent of a sequenceinduced migration shift depends on experimental condition^?^-^^ In this study, we investigated inversions occurring in a linear polyacrylamide matrix in order to link to previous experiments performed in this laboratory. However, it is to be understood that inversions are a ubiquitous phenomenon which can be observed in many different matrices. Figure 7 shows the separation of (PX-l74/HueIII digest, this time performed in a 5%(w/v) linear polyacrylamide matrix. Since (31) Stellwagen, N. C. Biopolymen 1990,30, 309-324. (32) Wenz, H. M. Nucleic Acids Res. 1994, 22, 4002-4008. (33) k r h , J.; Pariat, Y.; Muller, 0.; Heknbrock, K; Heiger, D.; Foret, F.; Karger, B. L. Electrophoresis 1995, 16, 377-388.
Analytical Chemistry, Vol. 67, No. 17, September 1, 1995
2979
1
14
'
1
'
16
I
*
I
20
18
'
I
22
-
1
'
24
1
.
1
26
28
'
1
'
30
migration lime (min)
Figure 7. Determination of the peak migration order of the 271, 281, and 310 bp fragments of the @X-l74/Haelll digest. Conditions: 100 pm i.d. (Polymicro Technologies), L = 25 cm, I = 24.1 cm, 5% (w/v) LPA, 40 mM 1 x TB buffer, E = 160 V/cm, I = 11 pA, sheath flow collection (sheath liquid = 1 x TB buffer), UV detection at 254 nm. The inset shows an expanded view of the collection window.
E
0
2
4
6
0
10
12
migration time (min)
Figure 8. Reinjection of the fragments collected during the run in Figure 7. Conditions: 50 pm i.d. (J & W Scientific), L = 28 cm, I = 20 cm, 1% (w/v) methylcellulose, 1 x TBE buffer, E = 250 Vkm, I = 6 pA, 1 pg/mL EtBr, LIF detection at (543/610) nm.
these fragments all arise from a common plasmid, the peak areas should increase with number of base pairs. However, the area of peak 5 is greater than that of peak 6 (supposedly 271 and 281 bp fragments, respectively), suggesting an inversion in migration order. For identification purposes, peaks 5, 6, and 7 (fragment lengths 271, 281, and 310 bp, respectively) were collected and reinjected into an EtBrcontainimg matrix, shown in Figure 8. I t is useful to note that the collection procedure for the three fragments was repeated 20 times, each yielding the same results as in Figure 8. Thus, each of the three fragments could be reproducibly collected. Under the conditions in Figure 8, peak 6 migrated faster than peak 5, further suggesting an inversion of these two species, since EtBr is known to reduce significantly the effect of sequence influenced migration anomalies.33 To prove the anomalous behavior, identification of the separated fragments was necessary. Since the sequences of the fragments are d%erent, specific primers for exclusive amplification of each of the three fragments (34) Muller, 0.; Foret, F.; Karger. B. L.J. Chromatogr., in press.
2980 Analytical Chemistry, Vol. 67, No. 17,September 1, 7995
were synthesized. The following primers were selected: GAT TAG AGG CGT TIT ATG (upper) and TAG CAG TCG GCG TGT GAA (lower) for the specific amplification of the 271 bp fragment, resulting in a 115 bp fragment; AAT GTG CTC CCC CAA CTT (upper) and CTG CGT AAC CGT ClT CTC (lower) for the specific amplification of the 281 bp fragment, resulting in a 96 bp fragment; and AAG AAA ACG TGC GTC AAA (upper) and CCA CCT ACA TAC CAA AGA (lower) for the specific amplification of the 310 bp fragment, resulting in a 147 bp fragment. Each of the three collections was amplified, using all three pairs of primers. The resulting PCR products were analyzed by CE with LIF detection,after desalting and diluting the amplification products 1Wfold. The first collected peak beak 5) could only be ampiified with primers suitable for the amplification of the 281 bp fragment, whereas the second collected peak beak 6) could solely be amplified with the primer pair specific for the ampucation of the 271 bp fragment. The inversion of the lengthdependent migration order was therefore demonstrated, in agreement with recent literature rep0rts.3~~~~ These results illustrate the value of fraction collection in CE, where subsequent PCR analysis for identification can be achieved. One could also amplify and subsequently sequence the DNA fragments for further identification purposes. CONCLUSlONS The sheath flow collection device presented in this paper has the capability of reproducibly collecting up to 60 fractions during a single CE run without interruption of the electric field. High collection precision is achieved by detecting the migrating zones close to the end of the column using an optical fiber-based W detector. The collection into capillaries facilitated the handling of submicrolitervolumes. All 11fragments of (PX-174IHueIIIwere isolated into separate vials to demonstrate the feasibility of multiple peak collection. The inversion of the peak migration order of fragments 271 and 281 bp of this sample, using specific PCR amplifkation primers, illustrated the importance of peak collection for identification purposes. The collection device is applicable for different modes of CE without changing the hardware configuration. Recently, for example, we collected protein fractions separated by capillary isoelectric focusing with subsequent analysis by MALDI-TOF.34 In addition, in cases where collection into capillaries is not required, the capillary holder can easily accommodate either a membrane for blotting analysis or a target for direct IvlALDI-TOF analysis. ACKNOWLEDGMENT The authors acknowledge NIH GM15847 for support of this work. They also acknowledge the work of Jan Berka and Yvan Pariat in the field of sequence induced migration behavior, which initiated this project. This work is contribution 636 from the Barnett Institute. Received for review March 28, 1995. Accepted June 15, 1995.@ AC950309E Abstract published in Advance ACS Abstracts, August 1, 1995.
