J. Atherton1, J. Ju2, Y. Wang2, B. Carpenter1,
R.A. Mathies2, and G.F. Sensabaugh1
1 Forensic Science Group, School of Public Health, University of
California, Berkeley, CA
2 Chemistry Department, University of California, Berkeley, CA 94720
DNA profiling has been established as a valuable tool in the investigation and prosecution of violent crimes, particularly sex offenses. It has also proven of great use in the identification of skeletal and other body remains in criminal, civil, and human rights cases. The next great challenge is the development and implementation of DNA profile identification databanks that can be used in the same way fingerprint identification databanks are currently used. Many states have legislatively mandated the establishment of databanks containing DNA profiles of certain categories of offenders, notably sex offenders, and the framework of a national system for the U.S. has been established under CODIS (1).
For the full potential of DNA databanks to be realized, two conditions must be met. First, high throughput DNA profiling technology will be required to process the large number of samples submitted for databank entry. Some states have inputs exceeding 1000 databank samples per month to process. Second, casework laboratories will have to begin doing DNA profiling on every case in which there is viable biological evidence. Needless to say, DNA profile databanks are anticipated to be of great value in helping solve suspectless cases, cases which at present are rarely worked up in crime laboratories due to limited personnel and technology.
These two conditions impose constraints on both the genetic markers and analytical technologies to be used in profiling. The markers used must be capable of being typed on multiple platforms-high throughput platforms for databank process laboratories and low cost platforms for routine casework laboratories. The current consensus favors the use of polymorphic short tandem repeat (STR) loci for profiling, in great because this method meets both conditions. Typing throughput-a primary consideration for databank labs-is also of concern for casework labs; the greater the information that can be generated per unit analysis, the greater the efficiency. Typing throughput, however, means little without better approaches to sample preparation. Sample preparation remains a rate limiting step; no matter how rapid the DNA typing technology, the generation of profiles can proceed no faster than the rate of DNA isolation. Advances in sample preparation will benefit throughput in both databank and casework laboratories, particularly if sample preparation can be integrated with typing technologies. Finally, it is likely that today's high throughput analytical systems will be superseded by more advanced technologies in the future. Accordingly, the analytical typing technologies developed to meet present needs should be capable of evolving. Advanced throughput technologies will bring future benefit to casework throughput as well, much as advances in computer chip technology have brought computers into the home.
To meet high-end throughput needs, we are evaluating capillary array electrophoresis (CAE) with laser excited confocal fluorescence detection (2-4). We have previously demonstrated the feasibility of using CAE for the rapid analysis of STR polymorphisms (5,6); this provides a foundation for high throughput DNA profiling based on these markers. We have also investigated the use of an intermediate technology-detection of fluorescently labeled STRs on a fluorescence imager (FlourImager, Molecular Dynamics) (7). In both these studies, we have exploited fluorescence energy transfer (ET) to optimize multi-color detection. The concept of using PCR primers labeled with fluorescent ET dyes has been recently described by Ju et al. (8,9) who demonstrated their superior properties for automated DNA sequence analysis. In this paper, we summarize our studies on STR analysis using ET primers.
CAPILLARY ARRAY ELECTROPHORESIS (CAE)
Capillary electrophoresis (CE) is capable of providing rapid, high resolution separations of DNA fragments; it has been used for the analysis of DNA restriction fragments and polymerase chain reaction (PCR) products (10-18) and for DNA sequencing (2,3,19-25). Typical CE runs are 15-120 minutes in duration, depending on the size range of the DNA fragments to be separated and resolution desired. CE requires very small sample loads and has a loading format that is easily automated (8-10). The presentation by McCord at this symposium describes CE in more detail. Capillary array electrophoresis (CAE) is CE with many capillaries run simultaneously in parallel array (2-4). The design of CAE, coupled with the high speed of CE separations, allows very high throughput. The CAE unit used in this study, a prototype model built in the Mathies laboratory (4), has an array of 10 capillaries (75 um i.d.) with an effective run length of 25 cm. Fluorescence excitation is provided by an argon ion laser at 488 nm with confocal detection at 510-530 nm (green channel) and >590 nm (red channel). Output is stored in digitized form and processed using the programs IPLab, Kaleidagraph, and Canvas. The confocal detection system is quite sensitive, with a routine sensitivity limit of about 10 attomoles DNA per band. A schematic of the laser excited confocal fluorescence CAE system set up for two-color detection is shown in Figure 1.
With CE, it is necessary to co-electrophoresis the test sample with a typing standard. The two-color detection system allows a test sample and typing standard, each labeled with a different fluorescent dye, to be analyzed in the same capillary. The CAE unit illustrated in Figure 1 has also been constructed in a four-color detection mode; the four-color unit can be used both for DNA sequence analysis and STR typing analysis.
Primers labeled with different fluorescent dyes are routinely used in automated sequencing and in multiplex STR typing on the Applied Biosystems automated sequencer (26-32). A limitation of these dyes is that each fluorophore has its own distinctive fluorescence excitation and emission properties; detection systems require compromise excitation wavelengths and software processing to unscramble emission signals. ET labeling represents a significant advance on this situation (6,8,9). Primers are labeled with two dye molecules, one of which functions as a energy donor and the other as an acceptor; the design of a typical ET primer is shown below.
F6R
FAM-5'-GTGGGCTGAAAAGCTCCCGATTAT-3'
|
(CH)2(CO)-NH-(CH2)6-NH-ROX
The designation F6R indicates that 6 nucleotides intervene between the donor dye FAM and the modified T to which the acceptor dye ROX is attached. Incident radiation at a single wavelength (in our case, 488 nm) is absorbed by the donor which transfers energy to the acceptor; the acceptor then emits at its characteristic fluorescence emission wavelength. Use of different acceptor dyes allows generation of distinctive emission spectra to be obtained from excitation at a single wavelength (Fig. 2). For example, FNF labels have their emission maximum in the green at 525 nm; FNR labels emit maximally in the red at 605 nm. Due to the high efficiency of energy transfer, there is no need to make primer concentration adjustments to compensate for fluorescence intensity differences among the primers as must be done for primers labeled with single fluors; ET primers with different acceptors have roughly comparable fluorescence emission properties (8,9).
STR TYPING USING CAE AND ET PRIMERS
Our initial efforts focused on the STR marker THO1, a tetranucleotide STR; this is one of the best studied STRs and is a consensus marker in many forensic STR typing systems (32,33). Samples of previously determined THO1 type were amplified by PCR using ET primers; PCR products were desalted by flotation dialysis on VCWP membranes and were loaded on the capillary columns by electrokinetic injection.
Initial experiments demonstrated that THO1 alleles could be separated with near baseline resolution on capillary columns containing the non-denaturing separation matrix 0.8% hydroxyethylcellulose (HEC) and the non-fluorescent intercalator 9-aminoacridine; under non-denaturing conditions, the presence of an intercalator is needed to achieve near single base resolution. ET primers differing in the number of nucleotides between donor and acceptor dyes were tested to determine effects on mobility. A pair of primers, F14F and F6R, were found to yield PCR products with coincident mobilities and these primers were used in subsequent experiments involving non-denaturing separations. Two samples amplified using one ET primer, F6R, were combined to form an allelic ladder containing the 6, 7, 8 and 9 alleles. Test samples were amplified using the other ET primer, F14F. Test samples were co-electrophoresed with the allelic ladder. Typing the test samples was straightforward; the allelic ladder peaks that coincide with the test sample peaks indicate the type. Electrophoresis runs were completed in under 20 min. Statistical analysis of replicate data shows mean sample sizing on each allele to be accurate within 0.4 bp with standard deviations less than 0.7 bp. This accuracy and precision makes possible straightforward type calling by data processing software. This work has been published (6).
Electrophoresis under non-denaturing conditions detects heteroduplex DNA as well as homoduplex DNA (34). The heteroduplex bands migrate outside the region of the homoduplex allelic bands and hence do not interfere with typing. They do, however, potentially intrude on migration zones that might be occupied by other markers in a multiplex set. To avoid heteroduplex formation, we have recently evaluated capillary electrophoresis under denaturing conditions employing a separation matrix containing 2% HEC and 7M urea. Under these conditions, baseline resolution is achieved for the tetranucleotide repeat alleles and alleles differing by a single nucleotide (such as the 9.3 and 10 alleles in THO1) can be distinguished. Mobility shifts due to the primer labeling configuration are insubstantial and typing against an allelic ladder is straightforward.
We are currently working toward multiplex typing by CAE using previously characterized markers, e.g., the markers developed by Promega and the Home Office Forensic Science Service (27-32,33,35-38). In some cases, primers have been modified to improve amplification properties. CAE typing exploiting ET primers has been developed for the loci shown in table 1. Multiplex sets have been constructed using non-overlapping markers; an example using the markers vWFA, TPOX, THO1, and D18S51 is illustrated in Figure 3.
Overall, the results of these studies establish the feasibility of using CAE for STR typing. With electrophoresis cycles of 30 minutes, and projection to a 48 capillary unit, it should be possible to process more than 500 samples per day; if samples are typed with a 4 locus multiplex, this translates to more than 2000 typings per day. These figures are well within the throughput specifications required for databank development. If desired, throughput can be increased using 4-color detection. Two 4-color configurations are possible. First, two multiplexes can be typed in concert against the corresponding allelic ladders, each test sample multiplex and each allelic ladder being detected in a different color. Alternatively, three multiplexes can be typed in concert against a standard sizing ladder; this approach requires allele-calling based on size measurement rather than comparison to the allelic ladder.
The potential for even higher throughput is on the horizon. Preliminary experiments show that STR typing can be done using the CAE on a chip developed by Woolley and Mathies (39); a result is shown in Figure 4. With a run time of under 5 minutes, the CAE on a chip is at least 4 times faster than conventional CAE. This result demonstrates the feasibility of STR typing on a next generation technology.
STR TYPING ON SLAB GELS USING ET PRIMERS
ET labeling offers many advantages for STR typing on conventional slab gels using fluorescence detection (7). This is demonstrated by comparison of the several detection formats that can used with conventional slab gel electrophoresis: silver staining, post-staining with the DNA binding fluorescent dye SYBR Green, single fluorescent dye labeling, and ET labeling. For fluorescence detection, we have employed a fluorescence imager (Molecular Dynamics FlourImager 575) with a two-color detection potential similar to the CAE unit, i.e., excitation at 488 nm with detection using 530 nm bandpass and 610 nm longpass filters. SYBR Green labeled products and FAM labeled products were detected with the 530 nm bandpass filter; FNR labeled products were detected with the 610 nm longpass filter.
Slab gel electrophoresis was performed on 6% polyacrylamide gels (19:1 acrylamide/bis), 32 cm in length, for 2.25 hours at 40 watts. Near single base resolution was achieved for THO1; the 9.3 and 10 alleles could be resolved when run in adjacent lanes or when in different colors in the same lane. Initial studies with THO1 PCR products showed that the primer labeling configuration affected mobility. For example, the F14F labeled primer exhibited an approximate +1 base shift relative to the F6F and F6R labeled primers; the two F6 primers were nearly identical in mobility. In contrast, use of the THO1 primer singly labeled with fluorescein (Promega) gave an approximate -1 base shift. This consideration is important if test samples and allelic ladder standards are to be run in the same lane.
Electrophoresis under denaturing conditions separates the DNA strands of PCR products. As a consequence, both silver stained and SYBR Green stained alleles are detected as band doublets. In contrast, use of fluorescently labeled primers, whether in the single fluor or ET mode, presents a single band for each allele. This simplifies pattern interpretation, particularly in situations involving PCR artifacts such as non-templated base addition and stutter banding. With regard to detection sensitivity, fluorescent primers and SYBR Green staining are comparable; all are about 3 times more sensitive than silver staining. Use of the FlourImager for detection allows relative band densities to be quantitated, a benefit when assessing mixtures.
A singular advantage of ET labeling coupled with fluorescence imager detection is the throughput gain offered by two-color detection: it is possible to co-electrophorese two multiplexes, each labeled with a different color, in a single lane. Because there is negligible cross talk between the red and green detection channels, the red and green labeled samples are completely distinguished. FNF and FNR labeled products are detected with comparable sensitivity. The simultaneous typing of seven loci in two multiplexes is shown in Figure 5.
ET labeling represents a significant advance on fluorescent labeling technology. We have demonstrated here the advantages of using ET labeling in high throughput STR typing by CAE and in conventional slab gel electrophoresis coupled with fluorescence imager detection.
Locus |
Repeat |
Het. |
Alleles |
Reference |
| THO1 | AATG |
0.77 |
8 (179-203) |
(33, 34) |
| CSF1PO | AGAT |
0.74 |
9 (299-323) |
(35) |
| TPOX | AATG |
0.66 |
8 (232-248) |
(36, 37) |
| vWFTA31 | AGAT |
0.74 |
11 (139-167) |
(32) |
| FIBRA | AAAG |
0.82 |
9 (256-284) |
(32) |
| FES/FPS | AAAT |
0.70 |
8 (222-250) |
(38) |
| D6S502 | TCTA |
0.78 |
8 (152-210) |
(32) |
| D18S51 | AAAG |
0.89 |
14 (267-319) |
(32) |
| D20S85 | AAAG |
0.70 |
6 (123-147) |
(32) |
| D21S11 | TCTA/G |
0.82 |
12 (172-264) |
(32) |
Figure 1. Capillary array electrophoresis unit with two-color confocal fluorescence detection.
Figure 2. Fluorescence energy transfer.
The donor molecule is excited at the wavelength indicated by the arrow (top panel). The excited donor molecule transfers its excitation energy to an acceptor molecule which then emits fluorescence at the acceptors characteristic emission wavelength (bottom panel). Pairing a single donor with different acceptors permits efficient fluorescence emission at different wavelengths from excitation at a single wavelength.
Figure 3. Multiplex typing of four loci on a single capillary run.
Allelic ladders for the vWFA, THO1, TPO, and D18S51 loci are shown.
Figure 4. THO1 typing with CAE on a chip.
Two samples, a 6,9.3 and a 7,10, were mixed to show 4 bp and 1 bp resolution.
Figure 5. Two-color detection of two multiplexes on a slab gel.
The triplex CSF1PO- FES/FPS-D20S85 was labeled using FNF primers and detected in the green channel (530 nm bandpass filter). The quadruplex D18S51-TPO-THO1-vWFA was labeled with FNR primers and detected in the red channel (610 nm longpass filter). Each multiplex was run separately (the G lanes for the triplex and the R lanes for the quadruplex) and combined (the B lanes). The test sample was typed by comparison to the allelic ladder lane. Note the absence of fluorescence crosstalk between the green and red detection channels.
Go to proceedings home page