With Kathi J Ulfelder


The analysis of double-stranded DNA (dsDNA) fragments, such as those produced by the polymerase chain reaction* (PCR) and enzymatic digestion, has led to considerable advances in molecular biology. Since its advent in 1985 (Saiki et al., 1985; Mullis and Faloona, 1987), PCR technology has been used to directly detect and quantify viruses, to track inheritance patterns in a family, to diagnose numerous genetic diseases, to identify individuals in forensic applications, and to aid in mapping the human genome.

The classic technique for analyzing the DNA produced by PCR (i.e., slab gel electrophoretic separation, followed by stain or probe detection) has its share of drawbacks. Slab gel electrophoresis is often time-consuming, labor intensive, and difficult to quantitate much less automate. It is, however, accepted as standard methodology for the molecular biology field. Recently, capillary electrophoresis (CE) has been used successfully for the separation of PCR products and DNA restriction fragments with high reproducibility and efficiency (Ulfelder et al., 1992; Landers et al., 1993). The introduction of laser-induced fluorescence detection (LIF) increased the sensitivity of CE (Schwartz and Ulfelder, 1992; Schwartz et al., 1994), a necessary improvement in any competition with autoradiography for low level detection. Given these advances in methodology, CE has become an attractive alternative to the standard method for separation and quantitation of PCR products. Additional information on HPCE is given in Chapter 3.


Since its introduction, the PCR technique (Saiki et al., 1985; Mullis and Faloona, 1987) has been used extensively in molecular biology to amplify specific

* PCR is covered by U.S. patents owned by Hoffmann-LaRoche, Inc.

DNA sequences that are present in trace quantities. While the design seems simple—repeated cycles of denaturation, primer binding, and DNA synthesis—this powerful technique can result in 106-fold amplification from a single copy of target DNA.

Essentially, all forms of DNA (and RNA) can be amplified by PCR. The amount of DNA template required for amplification depends on the complexity of the genome. Typically, 0.1 to 1 mg of mammalian genomic DNA is required for PCR, while only picogram to nanogram quantities are utilized for bacteria (Sambrook et a!., 1989).

A standard PCR reaction mix consists of many components, (Sambrook et al., 1989; Cha and Thilly, 1993). Besides the template, two single-stranded (ss) oligonucleotide "primers," 15 to 30 bases long and having sequences that are complementary to regions on the target, are required to initiate DNA synthesis. The primers are present in excess in the reaction mix—typically between 0.3 and 3 mM of each primer is used. Usually, the ratio of primer to target is kept at least to 108: 1. Too high a ratio creates nonspecific product and primer-dimer; too low a ratio produces very little product. Deoxynucleo-side triphosphates (dNTPs) are necessary as the components for primer extension, and are also present in the reaction in excess (37-1.5 mM of each dNTP). The reaction mix is buffered using 50 mM KC1, 10 mM Tris (pH 8.3), and 1.5 mM MgCl2. This buffer is normally optimized for target, template, and DNA polymerase variations.

DNA polymerase is required for synthesis. There are several different polymerases available, each with its 6Wn characteristics that will affect its efficacy in the PCR. Since many labs have studied the properties of the different polymerase in current use, much is known about their efficiencies in PCR and rate of base misincorporation (Table 7.1). The thermostable polymerases (Taq, Vent) are preferred in PCR over the heat-labile enzymes

TABLE 7.1 Fidelity and Efficiency of DNA Polymerases Used in PCR


Error rate


Efficiency per

Number of



fraction" (%)

cycle (%)

cycles requiredb


2 x 10"4





7.2 x 10"5



45 optimized



1.3 X 10"4





3.4 X 10"5





3 X 10"6





4.5 x 10"5




" Fraction of PCR-induced noise following 106-fold amplification of a 200-bp target sequence given the error rate.

b Number of cycles required to obtain 106-fold amplification given the efficiency per cycle. Adapted with permission from Cha and Thilly, (1993) PCR Methods Applic., 3:S18-S29.

" Fraction of PCR-induced noise following 106-fold amplification of a 200-bp target sequence given the error rate.

b Number of cycles required to obtain 106-fold amplification given the efficiency per cycle. Adapted with permission from Cha and Thilly, (1993) PCR Methods Applic., 3:S18-S29.

(T4, T7, Klenow) because of their ease of use, especially when automating PCR.

Finally, the reaction mix may contain glycerol for formamide to enhance specificity (Cha et al., 1992; Sarkar et al., 1990), as well as gelatin, Triton X-100, or bovine serum albumin to stabilize the polymerase, and mineral oil to prevent water evaporation during the reaction.

The typical PCR run (Fig. 7.1) consists of three cycles: denaturation (12 min at 3 94°C), primer annealing (1-2 min at 50-55°C), and extension (1-2 min at 72°C). This design also requires optimization for each particular PCR. The desired blunt-ended duplex product does not appear until after the third cycle, whereupon it accumulates exponentially in subsequent cycles. The number of cycles required will depend on the efficiency of the reaction per cycle. Once the desired product has reached about 1012 copies, PCR efficiency drops significantly, and product stops amassing exponentially. This is the plateau phase; continuing PCR beyond this point often results in contaminating by-products rather than more product (Cha and Thilly, 1993).


The separation of nucleic acids by CE has become a steadily growing area of interest, especially since the inception of the Human Genome Initiative. This interest originally stemmed from the use of polyacrylamide (PA) or agarose slab gel electrophoresis as the accepted standard for nucleic acid separation (Stellwagen, 1987).

7.2.1 Separation Mechanism

In free solution, the constant linear charge density of DNA molecules affords them a mobility that is independent of molecular weight (Olivera et al., 1964). However, in a support medium such as a slab gel electrophoretic separation by molecular weight occurs. Although the exact mechanism of this molecular sieving effect on nucleic acids is not clear, two theories have been proposed. The first was based on the Ogston model (Ogston, 1958), namely, in a random network of enmeshed fibers, the mobility of a macromolecule will be directly proportional to the volume fraction of pores of a gel it can enter. With increasing gel concentration, the average gel pore size decreases; thus, larger molecules will have difficulty entering the gel pores and therefore show retarded mobility. The second theory is based on the "snakelike" migration of DNA through the pores of gel, known as (biased) reptation (Lumpkin and Zimm, 1982). Depending on the size of the DNA fragment and/or the magnitude of the electric field, nucleic acid species can become aligned parallel to the field and may comigrate in a manner independent of size. Thus, if a DNA fragment is too large, or the field strength too high, no resolution between small or large DNA pieces may be seen.













N0 copies of template DNA

Target sequence Primers

After n-th cycle









Figure 7.1 Schematic representation of PCR: N0 copies of duplex template DNA are subjected to n cycles of PCR. During each cycle, duplex DNA is denatured by heating, allowing primers (arrows) to anneal to the targeted sequence (hatched square). In the presence of DNA polymerase and dNTPs, the primers are extended. The desired blunt-ended duplex product (thick bars with arrows) does not appear until after the third cycle, and accumulates exponentially during subsequent cycles. After n cycles of PCR, N0 (1 + Y)""1 copies of duplex product are present. [Reprinted with permission from Cha and Thilly, PCR Methods Appl 3:S18 (1993).]

7.2.2 Classical Methods of DNA Analysis

In classical electrophoresis, cross-linked PA or agarose slab gels are typically used in a flat bed or cylindrical tube format, where the Joule heating generated in the electrophoresis process can be more readily dissipated. This arrangement allows the use of higher field strengths to improve analyte resolution. Linear PA at low concentrations also provides a molecular sieving mechanism for nucleic acids (Bode, 1977; Tietz et al., 1986).

The many new nucleic acid amplification techniques and methods of determining DNA composition and structure require a final detection step involving PA or agarose electrophoresis. Since slab gel electrophoresis is an established technology, many "cookbook" procedures are available to aid in the techniques of nucleic acid separation (Sambrook) et al,, 1989). However, these procedures can be quite time-consuming (> 2 h), technically demanding, labor intensive, and not amenable to automation. Although several samples at a time may be run on a single gel, the gel can be used only once. Reproducibility may be poor, and separated analytes difficult to quantify. Usually, microliters of precious sample are required for analysis; and there is also the problem of waste disposal when toxic substances such as acrylamide and radioactive probes must be used for autoradiography. Because of these problems, CE becomes an attractive alternative to agarose or PA slab gel electrophoresis.


7.3.1 CE Principles Related to Nucleic Acids

In a capillary, conditions can be produced that are very similar to a slab gel environment. Using high viscosity buffers or polymerizing a gel inside the capillary creates a matrix not unlike the pores of a slab gel. DNA molecules, negatively charged, migrate toward the anode but must also navigate through the polymer matrix. Smaller molecules will be able to travel with greater ease through this environment and will reach the detector first; the larger the molecule, the more hindered its passage through the matrix. A separation based on molecular mass is thus achieved.

7.3.2 Buffer Systems

The choice of a buffer system becomes the most significant variable in developing a separation of PCR products by CE. Smaller DNA components (such as dNTPs and primers) have been separated by means of a partitioning approach based on affinity for a micelle in solution (Cohen et al., 1987). In separations of larger dsDNA components by CE, a sieving mechanism is required in the capillary to separate species with the same linear charge density. In optimizing a sieving matrix, two approaches have been used: chemical gels and physical gels.

A chemical gel consists of a capillary filled with a cross-linked sieving matrix, such as polyacrylamide. With their well-defined pore structure and size, these gels produce high resolution separations comparable to those obtainable from sequencing gels. A chemical gel, however, cannot withstand high temperatures or high field strengths. Because of this, separations of double-stranded PCR products usually take more than 45 minutes. In addition, electrokinetic sample introduction is the only injection mode feasible, since the high viscosity gel prevents aqueous sample plugs from entering the capillary. As a result, some sample preparation, (e.g., desalting) may be necessary to be sure that enough sample is loaded onto the capillary.

Although chemical gels have been applied to the separation and sequencing of nucleic acids, common molecular biology buffers (Tris-borate-EDTA, Tris-acetate-EDTA) containing organic additives such as low- or zero-cross-linked polyacrylamide, polyethylene glycol, or cellulose, derivatives (e.g., hy-droxypropylmethylcellulose, hydroxyethylcellulose) have become increasingly popular for most PCR product separations (Zhu et al., 1989; Heiger et al., 1990; Schwartz et al., 1991; Nathakarnkitkool et al., 1992). These viscous buffers, known as "physical gels," simulate the pore structure of a gel and function as a sieving matrix similar to a chemical gel. The degree of sieving can be controlled by changing the type (e.g., chain length or derivative) and/ or concentration of linear polymer additive used, thereby optimizing resolution for a specific DNA size range. These gels are easily replaceable with a simple rinse through the capillary by pressure or vacuum, yielding identical separation conditions from run to run. Moreover, extremely high temperatures (up to 70°C) and high field strengths (1000 V/cm) can be applied to these matrices without damage to the sieving polymer. This ruggedness is beneficial when one is optimizing separation for a specific DNA size range, since manipulation of temperature and/or field strength can improve resolution (Guttman and Cooke, 1991b; Guttman et al., 1992). Most often, a polyacrylamide or polysiloxane-coated capillary is used in conjunction with these buffers to control electroosmotic flow and enhance peak efficiency.

Agarose solutions have also been applied to the separation of DNA fragments (Bocek and Chrambach, 1991a, 1991b, 1992). To take advantage of the light-scattering and -absorbing properties of agarose gels, clear agarose solutions have been used to separate DNA fragments up to 12 kb in length; for this application, molten low-melting agarose is heated and maintained in a coated capillary at 40°C. Alternatively, urea may be added to agarose solutions to prevent gel setup by disrupting hydrogen bond formation.

7.3.3 Intercalators

Intercalators (DNA-binding dyes) have been used for many fluorometric DNA assays, and as a DNA stain in slab gel electrophoresis. Typical dyes include ethidium bromide, a bisintercalator that binds one dye molecule for every 5 base pairs (bp) of DNA; and thiazole orange, a monointercalator having one dye for every 2 bp. A wide variety of dyes has been recently developed (Glazer and Rye, 1992; Benson et al., 1993). Interestingly, addition of these dyes to a CE gel buffer can actually improve dsDNA resolution (Schwartz et al., 1991; Guttman and Cooke, 1991a). These molecules insert themselves between the base pairs of DNA, changing the molecular persistence length, conformation, and charge of the DNA molecules. These modifications result in a change in electrophoretic behavior: Larger DNA molecules move relatively more slowly, allowing the separation time window to widen, thus increasing peak capacity. In some cases, separation is achieved for dsDNA fragments of the same size but different base composition, (Fig. 7.2).

Since DNA intercalators fluoresce strongly when excited by an appropriate light source, use of these dyes presents the opportunity of detecting low levels


ra 0 015


0 000

ra 0 015



18 20

Time (min)

Figure 7.2 Effect of ethidium bromide on the separation of Msp I-digested pBR322 DNA. (j4) No ethidium bromide; 100 V/cm. (S) 1 /ug/mL ethidium bromide in the gel-buffer system; 200 V/cm. Peaks (in bp): 1, 26; 2, 34; 3, 67; 4, 76; 5, 90; 6, 110; 7, 123; 8, 147; 9, 147; 10, 160; 11, 160; 12, 180; 13, 190; 14, 201; 15, 217; 16, 238; 17, 242. [Reprinted with permission from Guttman and Cooke, Anal. Chem. 63:2038 (1991). Copyright: American Chemical Society.]

18 20

Time (min)

Figure 7.2 Effect of ethidium bromide on the separation of Msp I-digested pBR322 DNA. (j4) No ethidium bromide; 100 V/cm. (S) 1 /ug/mL ethidium bromide in the gel-buffer system; 200 V/cm. Peaks (in bp): 1, 26; 2, 34; 3, 67; 4, 76; 5, 90; 6, 110; 7, 123; 8, 147; 9, 147; 10, 160; 11, 160; 12, 180; 13, 190; 14, 201; 15, 217; 16, 238; 17, 242. [Reprinted with permission from Guttman and Cooke, Anal. Chem. 63:2038 (1991). Copyright: American Chemical Society.]

of DNA using LIF, without the need for precolumn derivatization of the sample (Schwartz and Ulfelder, 1992).

7.3.4 Typical Instrument Parameters

Capillary electrophoresis of PCR-amplified products is usually performed in the reverse polarity mode (negative potential at the injection end of the capillary). A coated capillary (100 mm i.d., 37-57 cm total length) is filled with a gel buffer system. PCR samples are introduced hydrodynamically or, after desalting, electrokinetically. The PCR sample and a DNA marker of known size may be injected sequentially and allowed to comigrate in the capillary. With a capillary temperature set at 20 to 30°C, separation of PCR products is accomplished at field strengths of 200 to 500 V/cm. Detection is on-line, measuring either UV absorbance at 260 nm, or LIF.

7.3.5 Detection

Although it is clear that CE of nucleic acids with UV detection has many advantages over slab gel techniques—high efficiency peaks (107 plates/m) have been demonstrated (Guttman and Cooke, 1991a—a more sensitive approach is again needed to compete with autoradiography for low level detection. The detection limit on the average for double-stranded DNA by UV detection is about 8 ng/mL sampling concentration, (roughly 58 fg on-column). The use of on-column sample stacking (Chien and Burgi, 1992) together with intercalating agents in the buffer system (Ulfelder et al., 1992) can improve sensitivity fivefold. The addition of fluorescent intercalators to the buffer mixture, in conjunction with LIF, has also demonstrated enhanced sensitivity (Schwartz and Ulfelder, 1992).

Under conditions of LIF detection, a fluorogenic intercalator is added to the buffer to identify and quantify the PCR products by means of a laser that excites the resulting DNA-dye complex; the fluorescence emission wavelength of the complex will depend on the fluor chosen. For example, with thiazole orange intercalation, excitation of the complex is accomplished using the 488 nm line of an argon ion laser, with subsequent emission at 530 nm (Haugland, 1992). The DNA-dye complex fluoresces when excited by the appropriate wavelength of a laser, whereas the intercalator alone (as well as non-DNA sample components) will not. Primer and primer-dimer peaks are observed, since they too complex with the dye. However, dNTPs and other PCR reaction species such as Taq polymerase are not intercalated and therefore not detected. In all, this mode of LIF detection results in improved resolution of base pairs between the closely migrating fragments, as well as a sensitivity enhanced up to three orders of magnitude compared to UV detection (Schwartz and Ulfelder, 1992; Ulfelder and Shieh, 1993; Zhu et al., 1994).

For example, Figure 7.3 compares spectra from LIF detection of a fluorescent monointercalator and from UV detection. Note that by LIF, the pattern

Sabin 3

Was this article helpful?

0 0

Post a comment