Bruce R. McCord1, Alice R. Isenberg2, John M. Butler3,
and Ralph O. Allen2
1 FBI Laboratory, FSRU, FBI Academy, Quantico, VA 22135
2 Department of Chemistry, University of Virginia, Charlottesville, VA 22903
3 National Institute of Standards and Technology, Gaithersburg, MD 20899
With the increasing application of polymerase chain reaction (PCR)-based methods for human identification, there is a concomitant desire for improved automation and sample handling techniques. Techniques have been developed to automate the detection and storage of genetic data. However, such systems do not address the entire process of analysis from sample loading to detection to data processing (1). The recent development of capillary electrophoresis (CE) for DNA typing provides an avenue for this process (2-7). In capillary electrophoresis, DNA analysis takes place in a thin capillary, 50-100mm wide. The capillary permits effective heat dissipation allowing high voltages to be used, greatly reducing analysis time. The capillary also permits efficient sample loading since it can be easily manipulated by a robotic sampling device. Detection usually uses fluorescence analysis and the sample can be analyzed with no further preparation following the PCR.
The capillary electrophoresis system in our laboratory utilizes a separation process that eliminates the necessity for crosslinked or "chemical gels" (3,5,18). Instead, a buffer solution containing a water soluble polymer or "physical gel" is pumped into the capillary prior to each analysis and pumped out at its conclusion. Separation occurs due to the retardation of the DNA molecule as it migrates through transient pores in the entangled polymer matrix (8). The ability to renew the separation media after each run permits a single capillary to be used hundreds of times before replacement is needed. Typical polymers used in DNA analysis include hydroxyethyl cellulose, linear polyacrylamide, and polyethylene oxide (9). Applicable concentrations for hydroxyethyl cellulose are typically 0.5% to 1.0%. Molecular weights of this polymers can range from 50,000 to 200,000 amu.
To detect the DNA, a fluorescent intercalating dye is used. The dye, oxazole yellow (YO-PRO-1), has a very low background fluorescence permitting it to be added directly to the buffer system (16). The fluorescence of the molecule is activated through the intercalation process. In addition, intercalation enhances the resolution of the DNA fragments and limits conformational effects, improving the accuracy of the sizing of DNA fragments.
While the potential for the application of CE to PCR analysis has long been recognized, only recently has the technology evolved to the point where precise analysis of the reaction products is possible. The development of laser fluorescence and the availability of purified internal standards has allowed for the advancements in CE. These two developments have improved the precision and sensitivity of CE to the point that routine use of this technique for genetic analysis is now possible (5,18).
In this laboratory, we are developing the analysis of PCR products by CE for application to human identification. With the increasing use of genetic typing in the forensic arena, rapid and more automated techniques are required. This paper will discuss the advantages of CE in this context and give examples of its efficacy by comparing results obtained on population samples typed for the D1S80 and HUMTHO1 loci using both CE and slab gels.
Trizma base and boric acid were obtained from the Sigma Chemical Co. (St. Louis, MO), EDTA was from Fisher Scientific (Fair Lawn,), and the hydroxyethyl cellulose was from the Aldrich Chemical Co. (Catalog No. 30,863-3, Milwaukee, WI) and Polysciences Inc. (140,000-160,000amu, Warrenton, PA). Primers were obtained from Operon Technologies (Alameda, CA) or Promega Corporation (Madison, WI). AmpliTaq DNA polymerase was obtained from the Perkin Elmer Corp. (Norwalk, CT). Population samples were extracted for genomic DNA (10), and the extracted DNA was quantitated by slot-blot hybridization (11).
Samples were prepared by amplifying approximately 5 ng of DNA template via the PCR using previously described protocols (14,15). Standard allelic ladders were prepared by mixing alleles from individual samples with appropriate genotypes. The 150, 300 and 1000 bp DNA standards were obtained separately from BioVentures (Murfreesboro, TN) or from Gensura (Del Mar, CA) and diluted to a concentration of 1-3 ng/mL. For CE analysis, 1 µL of each internal standard was added to 1 mL of the PCR product and diluted to 50 mL with deionized water.
A Beckman P/ACETM 2050 capillary electrophoresis instrument with a Laser Module 488 argon ion laser (Beckman Instruments, Fullerton, CA) was used with a 520 nm bandpass filter. The capillary was a 50 mm i.d. DB-17 coated (0.1 mm phase thickness) capillary (J&W Scientific, Folsom, CA) 27 cm in length (20 cm to detector). The run buffer consisted of 100 mM tris(hydroxmethyl)-aminomethane, 100 mM boric acid and 2 mM EDTA. HUMTHO1 PCR products were analyzed using the run buffer at a pH of 8.2 with additives of 1% hydroxyethyl cellulose (HEC) and 500 ng/ml of the fluorescent intercalating dye YO-PRO-1 (Molecular Probes, Eugene, OR). D1S80 samples were run at a pH of 8.6 (adjusted with CsOH) with 0.3% of the Aldrich HEC and 0.28% of the Polysciences HEC. The Aldrich HEC was dissolved in the Tris-Borate-EDTA buffer at a concentration of 1% by shaking vigorously overnight. The solution was then filtered through a 0.45 mm cellulose acetate disposable filter (Corning Inc., Corning, NY). The larger HEC from Polysciences was not filtered.
Fresh buffer solutions containing the YO-PRO-1 dye were prepared each day. Prior to each individual run, the column was rinsed for 1 minute with Methanol to clean the capillary wall and 2 minutes with buffer in order to replenish the separation media. Following the methanol and buffer washes, the samples were electrokinetically injected at 1-2 kV. Separations took place at 4.5-5kV (167-185 V/cm).
Data points were recorded at a rate of 10 Hz on Waters Millennium 2010 software version 2.0 (Waters Chromatography, Bedford, MA). For the HUMTHO1 samples, the base pair (bp) size of each allele was estimated by fitting its migration time to a regression line fit between 150 bp and 300 bp internal standards (5). D1S80 samples were sized using a cubic spline fit of the data obtained from the allelic ladder separation. The calibration obtained with this procedure was applied to each sample run using a 1000 bp internal standard as a reference point (18). Calculations were performed using an optional GPC sizing package available with the Millennium software.
The PCR-amplified samples were also analyzed using polyacrylamide slab gels as described previously (13,17). Run time was approximately 3 hours at a constant 1000 volts. Silver staining was used to visualize the DNA bands (12,13). Typing of the gel run samples was performed visually by comparing the migration of sample alleles with that of the allelic ladder.
The goal of this project was to determine whether or not CE is a valid and robust technique for use in genetic typing of PCR-amplified products. In this study, we examined the sensitivity, precision, and reproducibility of the CE technique and performed optimization studies of each of these performance factors. The object of these studies was to produce a method which could be applied for long periods of uninterrupted operation (3,5,18).
The experimental data we used were 2 sets of 100 individuals which also had been typed by slab gel electrophoresis. To test the efficacy of the CE system, we typed one of the two populations using the STR HUMTHO1 and the second using the VNTR D1S80.
In previous studies performed with CE analysis of DNA, the following items were deemed important for reproducible detection: fluorescent detection, intercalating dye concentration, proper sample preparation, polymer size and concentration, capillary length and diameter, and separation voltage (2,3). In examining CE separation of DNA, we have obtained optimum results using internally coated, short, narrow diameter (50mm) capillaries with polymer concentrations of 0.5%-1.0% HEC. The voltage used was selected through a process of optimizing two competing factors-speed and resolution. Figure 1 which shows the effect on resolution and run time of D1S80 and amelogenin as voltage is increased. In general we have found the best combination of resolution and efficiency to occur at 160-190 V/cm.
A substantial difference between CE and slab gel electrophoresis is the fact that samples must be analyzed sequentially in CE. Thus, a direct comparison of sample migration time to that of an allelic ladder, cannot be performed by the CE system unless sample and allelic ladder are commingled. This type of analysis is only possible using DNA labeled with different fluorescent dyes and instruments equipped with multichannel detection. At the time of these studies, only a single channel detector was at our disposal. Thus for effective genetic typing, run to run variations in DNA migration time had to be minimized or compensated using internal standards whose migration time did not conflict with sample DNA. Sizing was then performed by estimating the size of each allele with reference to the internal sizing standards and comparing these results with those obtained from a standard allelic ladder.
For this analysis scheme to be effective, the instrumental parameters must be kept consistent. Periodic replenishment of buffer and system calibration is necessary. Otherwise migration shifts due to changes in pH and dye concentration will result in a loss of precision. Specific factors in the separation mechanism also affect calibration. In the range from 100-400 bp, the relationship between migration and fragment size is linear with CE (19). Thus, for the analysis of STRs within this range, allele size may be estimated by interpolation between two standards. This is illustrated in Figure 2, which shows two separate injections of amplified HUMTHO1 products. In this figure, an allelic ladder is overlaid on a sample containing alleles 6 and 9, illustrating the stability of the system (5).
For the larger VNTR D1S80, which has a 16 bp repeat and fragment sizes ranging from 360-900 bp, more complex analysis is required due to a non-linear relationship between size and migration (2,18). In this situation, a cubic spline fit is applied to the data obtained from the D1S80 allelic ladder. Unknown samples are then analyzed by utilizing a 1000 bp standard as an internal control to adjust the calibration data (18). Figure 3 illustrates the analysis of an amplified sample of D1S80 and ameloginin overlayed on an allelic ladder. In this situation, the 150 and 300 bp standards are used to size the amelogenin products.
Upon developing a system for sizing the two systems, tests were undertaken to evaluate the reproducibility of the experimental protocols. System stability was measured by repetitive analysis of the allelic ladders. For the HUMTHO1 system the standard deviation for fragment size measured for 50 repetitive analyses was 0.1% (5). A similar experiment run using 30 D1S80 analyses resulted in a standard deviation for fragment sizes of 0.3% (18). The increased standard deviation for the D1S80 ladder was the result of the nonlinearity of the system and the larger size of the fragments.
For testing the HUMTHO1 system, samples of liquid blood from nearly 100 African-Americans was extracted and analyzed using both slab gel electrophoresis and CE (5). In a blind study, 94 of 97 types measured by CE agreed completely with the gel analysis. Three samples showed extra peaks which interfered with sample alleles. Upon reanalysis of freshly prepared samples, correct results were obtained. The tests of D1S80 were carried out on liquid blood samples obtained from a Vietnamese population. All Vietnamese samples typed were in agreement with slab gel results. The pooled standard deviations from the HUMTHO1 and D1S80 studies were 0.3 bp and 1.2 bp respectively (5,18).
The above results demonstrate that CE is an effective and efficient technique for use in genetic typing of PCR-amplified DNA. Through the development of precise and robust procedures, nearly 200 short tandem repeats and VNTR samples have been analyzed. The data which were processed using calculations based on in-lane standards were in agreement with those obtained from visually from slab gels. The results demonstrate the capability of CE to rapidly and precisely type DNA.
This research was supported in part by NIJ Grant 93-IJ-CX-0030 from the National Institute of Justice. The authors would like to thank Bruce Budowle, Kathy Keys, Jill Smerick, Barbara Koons, and Janet Jung for their valuable assistance in support of this research.
This is a publication number 96-02 of the Laboratory Division of the Federal Bureau of Investigation. Names of commercial manufacturers are provided for identification only, and inclusion does not imply endorsement by the Federal Bureau of Investigation.
Figure 1. Effect of applied voltage on the migration time and resolution of D1S80 and amelogenin.
Migration time is in minutes. Resolution for D1S80 is determined for alleles 36 and 37. Conditions: 27cm, 50mm ID DB-17 capillary; 100mM TBE buffer with 0.58% of a mixture of hydroxyethyl cellulose polymers and 500ng/ml YO-PRO-1; 25ºC; 5 sec 2 kV injection; detection, LIF @ 520nm.
Figure 2. Separation of HUMTH01 alleles overlayed with an allelic ladder.
The calculated sizes of the alleles were determined from interpolating between the 150 bp and 300 bp peak migration times. Conditions: 27cm 50mm ID DB-17 capillary; 100mM TBE buffer with 1% HEC and 500 ng/mL YO-PRO-1; 25ºC; 5s @ 1 kV injection; 5 kV separation; detection, LIF @ 520 nm.
Figure 3. Separation of D1S80 and amelogenin alleles overlayed with allelic ladders.
150, 300, and 1000 bp internal lane standards are used for sizing the alleles and for calibration purposes. Conditions as in figure 1 with 4.5 kV separation voltage.
Go to proceedings home page