Discrete mobility of single-stranded DNA in non-denaturing gel electrophoresis

Abstract

Gel electrophoresis is the standard method to separate, identify and purify nucleic acids. SSCP detects single base changes by altered mobility of single-stranded segments electrophoresed through non-denaturing polyacrylamide gels. Herein, changes in electrophoretic mobilities due to single base substitutions were measured for single-stranded segments of lengths ranging from 333 to 547 nt. A 484 nt segment in exon H of the human factor IX gene was studied most intensively. After SSCP, mobilities were determined by scanning autoradiograms at very high resolution (1200 d.p.i.), which allowed precise measurement of mobilities. When the mobilities of 46 single base substitutions were characterized, the distribution of mutant segments relative to a wild-type control was found to be discrete, i.e. the observed mobility values occurred in distinct ranges. Discrete mobility distributions were seen at different electrophoretic temperatures, buffer concentrations, segment lengths and segment sequences. In addition: (i) single base substitutions caused discontinuous distributions between highly dispersed and sharp bands; (ii) at least one single-stranded segment produced two sharp bands of similar intensity. These observations suggest that: (i) the single base changes in DNA segments in the size range 333–547 nt result in discrete conformational changes; (ii) individual DNA molecules of the same DNA segment can occasionally adopt two or more discrete conformations.

INTRODUCTION

Single-stranded conformation polymorphism (SSCP) is the most widely used method for mutation scanning (1). Typically, a 200 bp genomic region is amplified by PCR and the single-stranded product is electrophoresed through a non-denaturing polyacrylamide gel. Single base sequence changes can be detected by altered electrophoretic mobility of one or both of the single DNA strands. The shifted mobility is attributed to a conformational change in the single-stranded DNA. Previously, the mobility distribution of many different single base changes was hypothesized to be a continuous function, such as a Gaussian distribution (2). Herein, we have used high resolution scanning to analyze the effects of single base substitutions in single-stranded DNA segments on band mobility, band width and band number.

MATERIALS AND METHODS

PCR

PCR (3) was performed with primers F9(1)16D and F9(547)18U to amplify a 547 bp region of exon H of the human factor IX gene, with primers F9(1)16D and F9(484)20U to amplify a 484 bp region, with primers F9(1)16D and F9(333)17U to amplify a 333 bp region [for F9(1)16D the 5′-end of the primer begins at nt 1, corresponding to nt 30883 of Yoshitake et al. (4); primer length, 16 nt; D, downstream, i.e. in the direction of transcription]. In order to analyze each strand, one of the two primers was fully phosphorylated at the 5′-end by T4 kinase. The mixture contained: 70 mM Tris–HCl (pH 7.6 at 25°C), 10 mM MgCl₂, 5 mM DTT, 100 pmol primer, 200 µCi of [γ-³²P]ATP (3000 Ci/mmol, 10 mCi/ml; Amersham), 80 pmol ATP (the molar ratio of primer to total ATP was 1:1.4), 25 U of T4 polynucleotide kinase (New England Biolabs). The 20 µl reaction was incubated at 37°C for 1 h and inactivated by heating at 65°C for 5 min (5). The PCR mixture contained, in a volume of 15 µl: 50 mM KCl, 10 mM Tris–HCl pH 8.3, 1.5 mM MgCl₂, 200 µM each dNTP, 0.1 µM primer, 0.6 U of AmpliTaq (Perkin-Elmer) and 0.5 ng of a 1 kb purified DNA template which was pre-amplified from human genomic DNA with primers of F9(–237)34D and F9(763)31U. The cycling conditions were: denaturation at 94°C for 1 min, annealing at 55°C for 2 min and elongation at 72°C for 3 min for a total of 15 cycles, followed by a final step at 72°C for 10 min. The loading buffer (2) was added at a ratio of 16:1 (v/v).

Mutations

The mutations are indicated on the sense strand: 1, A18G; 2, C46A; 3, A47T; 4, T48C; 5, C51T; 6, T54G; 7, T63C; 8, C91A; 9, T103G; 10, G119T; 11, C126T; 12, T157A; 13, T159C; 14, G165A; 15, G169C; 16, G170A; 17, G171A; 18, C198A; 19, C198T; 20, C209T; 21, C214A; 22, C214G; 23, C236G; 24, C236T; 25, G237T; 26, G237A; 27, T245A; 28, C248T; 29, T249C; 30, C251T; 31, C258G; 32, T270G; 33, A279T; 34, G283A; 35, T284A; 36, G293A; 37, G294A; 38, G305T; 39, G321T; 40, G329C; 41, G329T; 42, T334A; 43, G336A; 44, G338A; 45, G338C; 46, A345G; 47, T392A; 48, G394A; 49, G396A; 50, A399G. The anti-parallel nature of DNA duplexes implies that sense strand sequence changes near the 5′-end are associated with complementary strand sequence changes near the 3′-end of the antisense strand.

Gel electrophoresis and autoradiograph analysis

Gel electrophoresis (45 cm × 37.5 cm × 0.35 mm) was performed at 12 W constant power at 20 or 8°C. To eliminate band smearing that often occurs with polyacrylamide gels, the gel matrix was 7.5% GeneAmp™ (Perkin-Elmer). A total of 4000 ml of 90 mM TBE buffer (90 mM Tris, 90 mM borate, 1 mM EDTA, pH 8.3 at 25°C) or 50 mM TBE buffer (50 mM Tris, 50 mM borate, 1 mM EDTA, pH 8.3 at 25°C) was used. After pre-running for 30 min, 1.5 µl of sample was loaded and electrophoresed for 6–20 h, during which the buffer was remixed between the upper and lower chambers every 4 h. The gel was dried and exposed to Kodak X-OMAT AR film. The autoradiogram was scanned with a transmittance light source to an image by a Bio-Rad Model GS-700 Imaging Densitometer at 600 or 1200 d.p.i. resolution. The mobility was assigned as the distance from the origin to the position of the peak. The peak height was assigned as the peak OD after subtracting the background OD. Analysis was performed using Molecular Analyst™ v.1.4 software (Bio-Rad).

RESULTS

Discrete mobility distribution

A total of six single-stranded segments of 333, 484 and 547 nt in exon H of the human factor IX gene were examined. For each segment, 38–46 samples, each containing a different single base substitution, were tested under various electrophoretic conditions. The mutations were randomly selected in position and type, so the observed distribution was not attributed to sampling biases. After scanning the autoradiograms at a resolution of 1200 d.p.i., the mobility was determined much more accurately than by visual inspection (Fig. 1A and B); the mobility difference was <0.1 mm between the two wild-type lanes separated by a distance of 102 mm. For gels electrophoresed at 8°C, the mobilities were sorted in 0.4 mm bins, since mobility differences were larger and the band widths were greater. At 20°C, the bins were set at 0.1 or 0.2 mm, since with bins of 0.4 mm the discrete peaks of mobility frequencies tended to fuse.

Figure 1.

Discrete mobility distribution. (A) Gel autoradiogram. The 484 nt antisense segment was ³²P-labeled and electrophoresed through a 7.5% GeneAmp gel with 50 mM TBE at 8°C. Lane C is the wild-type; lanes 1–46 each contain a single base substitution. Δ mobility = sample mobility – wild-type mobility (mm). The wild-type mobility was 243.5 mm with a variance of <0.1 mm between the two C lanes, which were separated by 102 mm. (B) Scanning of single lanes. Lanes 18, C and 23 were shown to be symmetrical bands from left to right. The x-axis is the difference in mobility from the wild-type mobility (0) in mm. The y-axis is the band intensity in OD units. The peak height and band width are indicated. (C) Mobility distribution. The mobilities of 46 single base substitutions were categorized into 0.4 mm bins. The x-axis is the difference in mobility (mm) from the wild-type mobility (assigned as zero). The y-axis is the frequency of mutations at each mobility.

A 484 nt antisense strand segment was ³²P-labeled and electrophoresed through a 7.5% GeneAmp hydro-linker polyacrylamide gel with 50 mM TBE buffer at 8°C. The autoradiogram was scanned at 1200 d.p.i. to determine the mobilities with a resolution of 0.02 mm. The mobilities of 46 single base substitutions were observed to have a discrete mobility distribution, even when they were grouped into bins of 0.4 mm, 20-fold larger than the scan resolution (Fig. 1). (The meaning of discrete used herein is that the observed mobility values occurred in distinct ranges; Fig. 1C.) A bin of 0.4 mm corresponds to only 0.16% of the wild-type mobility. Similar results were obtained when the 484 nt sense strand or the 547 nt antisense strand were analyzed under the same electrophoretic conditions. In order to normalize mobilities from different gels, relative mobility (R_m = mutant mobility/wild-type mobility) was also used.

The mobility distribution was examined as a function of buffer, size and sequence. The gels in Figure 2 analyze: (i) TBE at 50 and 90 mM at 20°C; (ii) segment lengths of 484 and 333 nt; (iii) sense and antisense strands. The mobility distributions were also observed to be discrete with 0.1 or 0.2 mm bins. A 0.1 mm bin corresponds to 3.34 × 10^–4 to 5.30 × 10^–4 of the wild-type mobility, depending on experimental conditions. The discrete distribution was more pronounced with 50 than with 90 mM TBE (Fig. 2A and B) and with the longer segment of 484 nt than with a shorter segment of 333 nt (Fig. 2A and D). Both sense and antisense strands had discrete distributions (Fig. 2A and C).

Figure 2.

Mobility distributions as a function of buffer, size and sequence. (A) The 484 nt sense segment was electrophoresed with 50 mM TBE. (B) The 484 nt sense segment was analyzed with 90 mM TBE. (C) The 484 nt antisense segment was analyzed with 50 mM TBE. (D) The 333 nt sense segment was analyzed with 50 mM TBE. (A–D) 7.5% GeneAmp gel at 20°C. The band mobilities of 46 single base substitutions (A–C) or of 38 single base substitutions (D) were categorized into 0.1 mm bins. The wild-type mobilities were 227.7, 188.7, 299.3 and 243.0 mm, respectively; and the variance between the two control lanes was <0.1 mm.

The mobility of a segment was not affected by the amount of loaded sample or the presence of the complementary strand, i.e. when a 16-fold excess of the non-radioactively labeled PCR product was added or when the sample was diluted, no change in the mobility occurred. A darker exposure of the X-ray film did not alter mobility but did alter peak height and band width of a segment (data not shown).

The observed data are not consistent with a Gaussian distribution. The Shapiro–Wilkes test of normality reveals significant deviation (P < 0.01) from a normal distribution or a uniform distribution in the range –6.2 to +6.2 and –1.5 to +1.5 (mm) for Figures 1 and 2, respectively.

Band number and band width

In most cases, each single-stranded segment produced one band. However, one segment may produce two distinct bands (see the 73 nt segment in Fig. 3, lane 3). When the electrophoresis temperature was increased from 8 to 20°C, this segment in lane 3 generated only one band (data not shown). Two bands also occurred in a 219 nt segment at 8°C (data not shown).

Figure 3.

Two bands generated by one segment. The Sanger termination reactions were performed with ³²P-labeled primer F9(1)16D and ddCTP terminator, denatured and then electrophoresed through a 7.5% GeneAmp gel at 8°C with 50 mM TBE (2). Lane C is the wild-type and lanes 1–4 are mutants. Lane 1, C46A; lane 2, A47T; lane 3, T48C; lane 4, C51T. The segment in lane 3 with two bands of ~50% intensity is shown by arrows.

Band width is also an important parameter of electrophoresis performance. Dispersed bands of single-stranded and double-stranded DNA are observed on non-denaturing polyacrylamide gels at cold temperatures (6,7). Herein we report that single base substitutions can condense dispersed bands.

The band width was examined as a function of buffer, electrophoretic temperature, segment size and sequence. The 547 nt sense segment was electrophoresed at 8°C with 90, 50 or 30 mM TBE. With 90 mM TBE, the bands of four mutants, 6, 7, 48 and 49, were condensed, while all the other 46 mutant and the wild-type bands were dispersed (Fig. 4A). The ratio of band width to peak height (mm/OD) was a discontinuous distribution, varying from 423.3 in lane 44 for a dispersed band to 8.9 for a condensed band in lane 48. There were no segments with intermediate ratios ranging from 9 to 63. In addition, within dispersed bands there were sometimes several small peaks and troughs. In order to explore the interaction of the dispersed bands along the same strand or between the two complementary strands, a 16-fold excess of the non-radioactively labeled PCR product was added. The band width of a segment was not more dispersed.

Figure 4.

Discontinuous distribution of band width. The 547 nt sense segment was electrophoresed through a 7.5% GeneAmp gel at 8°C. (A) 90 mM TBE; (B) 50 mM; (C) 30 mM. Lane C is the wild-type and lanes 39–49 are mutants. The autoradiogram was scanned at 600 d.p.i. resolution × 8 pixel depth. (A–C) The wild-type mobilities were 132.7, 206.0 and 195.3 mm, respectively. The ratio of band width to peak height and the relative mobility (R_m) are indicated at the top of the gel.

Relative to 90 mM TBE, the R_m with 30 mM TBE changed greatly for many mutations, and each band was condensed (band width to peak height ratios varying from 11.3 in lane 44 to 10.4 in lane 48) (Fig. 4C). With 50 mM TBE, an intermediate state was observed; all the bands were condensed except for the bands of mutants 43–45. Similar results were observed when the experiment was repeated with the 5′-co-terminal 484 nt sense strand. The one difference is condensation of the bands of mutants 43–45 with 90 mM TBE. When the 5′-co-terminal 333 nt sense strand was analyzed, all the bands were condensed with 30, 50 and 90 mM TBE. For all the segments tested, all the bands were condensed at 20°C.

DISCUSSION

Scanning of autoradiograms at very high resolutions revealed a discrete distribution of mobilities, a discontinuous distribution of band width and the occasional presence of two distinct, condensed bands for a single-stranded segment. The underlying cause of these observations remains to be defined. An attractive hypothesis is that single-stranded DNA, especially when its size is >400 nt, adopts one of a discrete number of conformational states. The limited number of conformational states would produce a limited number of mobility changes associated with single base substitutions. A sharp or condensed band is consistent with either a unique conformation or a group of rapid interconverting conformations, while slower interconversion may be the explanation for the dispersed bands. Two bands from one segment implies two unique conformations or two groups of conformations which do not interconvert.

If the conformation(s) of each single base variant could be measured and identified with very high structural resolution, a correspondence between the mobility and the conformation(s) may be established. However, DNA conformation in solution may not be the same as that in a polyacrylamide gel in the process of electrophoresis. It is intriguing that discrete segregation of large T4 and λ double-stranded DNA within agarose gels can occur under a high electric field (8). Buffer concentration, buffer pH, gel matrix and temperature can dramatically affect the mobility of a single-stranded segment, suggesting that sugar–base and sugar–sugar interactions may play a more important role than secondary structure in determining the conformation associated with mobility (2,9,10). End-to-end distance of single-stranded segments is unlikely to be a major factor, because a single-stranded segment >20 nt can be considered as a polymer behaving as a random flight (free-jointed) chain with a statistical length of 4 nm and a persistence length of 2 nm (11). Since the phosphate-to-phosphate distance of a single nucleotide is ~6.5 Å, a single nucleotide substitution of the polymer should not result in a significant change in the statistical segment length. In summary, analysis of the SSCP phenomenon with high resolution scanning demonstrates an unanticipated distribution of single-stranded DNA segment mobilities.