484
M . J . Kochele,iu
CI
ul
Marie J. Rocheleau Ronda J . Grey Da Yong Chen Heather R. Harke Norman J. Dovichi Department of Chemistry, University of Alberta, Edmonton, Alberta
Formamide modified polyacrylamide gels for DNA sequencing by capillary gel electrophoresis Compressions are occasionally found during the separation of DNA sequencing fragments, particularly in G/C-rich regions and in gels operated at room temperature. Addition of at least 10% formamide to urea/polyacrylamide sequencing gels improves the denaturing capacity of the gel, minimizing compressions. Addition of 20% or more formamide decreases the separation rate, theoretical plate count, and resolution for normally migrating fragments. An optimum concentration of 10% formamide improves resolution of compressed regions without degrading the other characteristics of the gel. Operation of gels at room temperature simplifies the engineering associated with automated sequencers based on capillary gel electrophoresis.
1 Introduction DNA sequencing requires separation of labeled DNA fragments by gel electrophoresis. The separation rate of these fragments is proportional to the electric field strength; high electric fields lead to high separation rates. However, Joule heating generates a temperature rise that limits the maximum electric field strength. The temperature rise creates thermal gradients across the gel, band broadening, and degradation of the separation 111. Band broadening is produced by the strong temperature dependence of viscosity, 2% per degree at room temperature, of aqueous solutions. The velocity of fragments in the gel mirrors the viscosity profile; fragments at the center of the gel migrate faster than fragments near the cooler walls. The high surface-tovolume ratio of thin gels produces efficient heat loss and reduces both thermal gradients and band broadening. Conventional DNA sequencing slab gels are -0.5 m m thick.Thinner gels, -0.1 mm, generate fast and efficient seperations 121. Difficulties in automation, in maintaining a uniform gel thickness across the slab, and in detection have retarded widespread applications of this technology. On the other hand, capillary gel electrophoresis offers highly uniform chambers and high sensitivity detection technology. Typical capillaries of 50 pm inner diameter offer excellent thermal properties. Finally, the highly flexible nature of the capillaries simplifies automation. Several groups have developed DNA sequencers based on capillary gel electrophoresis and laser-induced fluorescence detection 13-10]. In these systems, the capillaries are operated at room temperature. At room temperature, G/C-rich samples are sometimes difficult to sequence and fragments tend to coe1ute.These regions of anomalously poor resolution, or compressions, will produce more typical separations if conditions are used that improve the denaturing capacity of the gel. Conventionally, gels are operated at elevated temperatures to minimize formation of compressions. Unfortunately, our capillary system is not thermostated. Following suggestions in the literature [ll], formamide was added to the gels to improve their denaturing capacity [lo]. These gels produce improved resolution at the expense of decreased sequencing rate. In this paper, we report the effect Correspondence: Dr. Norman J . Dovichi, Department of Chemistry, University of Alberta. I?lnionton, Alberta, Canada T6G 2G2
of formamide on the sequencing rate, resolution, and efficiency in capillary gel electrophoresis at room temperature.
2 Materials and methods The capillary electrophoresis system has been described before [8,10]. The polyimide-coated, fused silica capillary is 50 pm inner diameter, 190 pm outer diameter. The gel formulae are described below.The injection end of the capillary is held in a Plexiglas box equipped with a safety interlock.The other end of the capillary is inserted into the flow chamber of a sheath flow cuvette. Fluorescence is excited with a low power helium-neon laser ( " G r e - N E ) operating in the green at 543.5 n m ; fluorescence is collected at right angles with a microscope objective, imaged onto a pinhole, passed through a spectral filter, and detected with a photomultiJlier tube that is cooled to -2O"C.The current from the pho.omultiplier tube is dropped across a resistor, digitized with d Rainin analog-to-digital converter, and recorded with a Llacintosh Plus computer. The samples are prepared based on the peak-height-encoded sequencing strategy develo7ed by Tabor and Richardson [12].The samples are synthesized with 16 pmol ofeither ROX-orTAMRA-labeled primer, 4 pmole of M13mp18 single-stranded template, and a nucleotide mixture in the following ratio: 8d/ddA: 4d/ddC: Zd/ddG: Id/ddTwith deoxynucleotide to dideoxynucleotides in a 300:l ratio. The samples are resuspended in a 49: 1 mixture of formamide-EDTA. Three pL aliquots are taken from the resuspended samples, heated to 95°C for 2 min, and injected onto the capillary by applying a 200 V/cm electric field for 20 s. Prior to preparation ofthe gels, about 1 cm of the detection end of the capillary is dipped into y-methacryloxypropyltrimethoxysilaneand allowed to air-dry. Gels are prepared in 5 m L aliquots from carefully degassed acrylamide/N,N'-methylene bisacrylamide (6% T, 5 % C), X Tris-borate-EDTA (TBE) buffer, 7~ urea, and varying amounts of formamide (0-30% by volume). Polymerization is initiated b y addition of 2 pL of N,N,N',N'-tetramethylethylenediamine (TEMED) and 20 pL of 10% ammonium persulfate. The solution is injected into the capillary by use of a syringe. Although polymerization appears complete in 30 min, the capillaries are typically stored overnight before use. With the use ofhigh purityreagents and careful degassing, more than 95 O/o of the gel-filled capillaries are bubble-free and generate useful data. 01 73-0835/92/0808~0484 $3.50+.25/0
485
Forniuniide modified polyacryiamide gels for DNA 5equenciiig
3 Results and discussion A total of 26 sequencing gels (all 6 % T, 5 O/o C, 7~ urea, and O-30°/o formamide) were poured and evaluated with an M13mp18 sequencing sample. Either five or six capillaries were evaluated for each formamide concentration; the relative standard deviation in sequencing rate ranged from 1 to 13O/o and presumably was due to minor fluctuations in acrylamide concentration. During the degassing step, the total volume of the solution would decrease; reconstitution of the solution to the original volume introduced a variation in acrylamide concentration. Figure 1 presents the sequencing rate as a function of formamide concentration at the 95% confidence interval. The sequencing rate was constant for gels containing less than 20% formamide; a 30 cm long gel, operated at 200 V/cm, produced a sequencing rate of 275 bases/h. However, addition of 30% formamide decreased the sequencing rate to 175 baseslh.
A sequencing reaction was performed with only a C-terminator. This sample was separated on a 0% formamide gel (Fig. 3a). Fragments 56-59 are baseline resolved, whereas fragments 65-66 are compressed. Addition of'30°/o formamide degraded slightly the separation of the 56-59 fragments, but improved the resolution of fragments 65-66 (Fig. 3b). That is, fragments that are free of compression are better resolved in 0 O/o formamide gels whereas compressions are better resolved in 30% formamide gels. Resolution is definegas the ratio of peak spacing, A T to average peak width, W
The theoretical plate count was measured for fragments 47, 84 and 154 nucleotides in length. Theoretical plate count was estimated as N = 5.54
($)
where T, is the retention time and W,,, is the full-width at half-height of the peak. Figure 2 presents the theoretical plate count for the fragments 47 nucleotides in length as a function of formamide concentration; again, the data are presented at the 95 Vo confidence interval. Theoretical plate count is constant for 0.5, and 10% formamide gels. There is a significant loss in plate count for gels containing 30% formamide. Longer fragments show similar trends. The degradation in plate count for gels with higher formamide concentration may reflect the poor thermal properties of the additive. Formamide has poorer thermal conductivity and heat capacity than water, and formamide-modified gels will operate at higher temperature and thermal gradient than aqueous gels.
0.0
10.0
20.0
30.0
% Formamide Figurr2. Theoretical plate count for47-merfragment,95% confidence interval. The same gels as in Fig. I were used in this study.
325 300 27 5 0.351
;
:
I
33.0
34.0 35.0 Time (Min.)
36.0
33.0
34.0 35.0 Time (Min.)
36.0
250 225 200 175 150
0.0
10.0
20.0
2
% Formamide Figure 1. Sequencing rate as a function offormamide concentration in 7M urea, 6%T, 5%C polyacrylamide gels. One cm of the gel was covalently bound to the capillary wall. The fused silica capillary was 30 crn long, SO vm inner diameter,and operated at an electric field strength of 200 V/cm. The data are shown at the 95% confidence interval, with either 5 or 6 data averaged for each point.
0.40
Figure 3 Separation of a C-terminated sequencing sample in a 7 M urea, b%T, 50ioC polyacrylamide gel operating at 200 V/cm. ( b ) Separation o f a C-terminated sequencing sample in a 30% formamide, 7~ urea, 6%T, SO/aC polyacrylamide gel operating at 200 V/cm.
486
M J Kuclicleau
PI
iil.
Figure 4 presents resolution as a function of formamide concentration. Gels containing 10Yo and less formamide generate excellent resolution ( R = 1.9) of fragments 56-59 that are free of compression (solid circles, 95 Yo confidence limits pointing to the right). Up to 10% formamide can be added before a decrease in baseline is observed.The 20 and 30% gels produce poorer resolution, about 1.3. Decreased resolution for the 20 and 30% formamide gels is related to the decrease in plate count noted for the higher formamide concentration gels. Resolution is proportional to the square root of plate count. The factor-of-two decrease in plate count noted produces the factor-of-0.7 decrease in resolution. Figure 4 also presents resolution of fragments 65-66 that show compression (solid squares and 95Yo confidence limits pointing to the left). Gels containing 10-30°/o formamide generate a factor-of-two improvement in resolution compared with the 0% gel. The denaturing properties of formamide, which acts to disrupt secondary structure of the fragments, leads to improved resolution of the compressions. The compressions are not completely relaxed under these conditions and the resolution is less than that produced by normally migrating fragments.
4 Concluding remarks Addition offormamide to sequencing gels improves the resolution of DNA compressions. However, gels containing higher concentrations of formamide have lower sequencing rate, theoretical plate count, and resolution for normal fragments. A 10[l/oformamide concentration appears optimum. Sequencing rate, theoretical plate count, and resolution of normally migrating fragments are similar to that ob-
r 2.0
1.5
1.0
0.5
- 0.0.
.
This w o r k was supported i n p a r t by the Department qf'Ener
M . J . Kochele,iu
CI
ul
Marie J. Rocheleau Ronda J . Grey Da Yong Chen Heather R. Harke Norman J. Dovichi Department of Chemistry, University of Alberta, Edmonton, Alberta
Formamide modified polyacrylamide gels for DNA sequencing by capillary gel electrophoresis Compressions are occasionally found during the separation of DNA sequencing fragments, particularly in G/C-rich regions and in gels operated at room temperature. Addition of at least 10% formamide to urea/polyacrylamide sequencing gels improves the denaturing capacity of the gel, minimizing compressions. Addition of 20% or more formamide decreases the separation rate, theoretical plate count, and resolution for normally migrating fragments. An optimum concentration of 10% formamide improves resolution of compressed regions without degrading the other characteristics of the gel. Operation of gels at room temperature simplifies the engineering associated with automated sequencers based on capillary gel electrophoresis.
1 Introduction DNA sequencing requires separation of labeled DNA fragments by gel electrophoresis. The separation rate of these fragments is proportional to the electric field strength; high electric fields lead to high separation rates. However, Joule heating generates a temperature rise that limits the maximum electric field strength. The temperature rise creates thermal gradients across the gel, band broadening, and degradation of the separation 111. Band broadening is produced by the strong temperature dependence of viscosity, 2% per degree at room temperature, of aqueous solutions. The velocity of fragments in the gel mirrors the viscosity profile; fragments at the center of the gel migrate faster than fragments near the cooler walls. The high surface-tovolume ratio of thin gels produces efficient heat loss and reduces both thermal gradients and band broadening. Conventional DNA sequencing slab gels are -0.5 m m thick.Thinner gels, -0.1 mm, generate fast and efficient seperations 121. Difficulties in automation, in maintaining a uniform gel thickness across the slab, and in detection have retarded widespread applications of this technology. On the other hand, capillary gel electrophoresis offers highly uniform chambers and high sensitivity detection technology. Typical capillaries of 50 pm inner diameter offer excellent thermal properties. Finally, the highly flexible nature of the capillaries simplifies automation. Several groups have developed DNA sequencers based on capillary gel electrophoresis and laser-induced fluorescence detection 13-10]. In these systems, the capillaries are operated at room temperature. At room temperature, G/C-rich samples are sometimes difficult to sequence and fragments tend to coe1ute.These regions of anomalously poor resolution, or compressions, will produce more typical separations if conditions are used that improve the denaturing capacity of the gel. Conventionally, gels are operated at elevated temperatures to minimize formation of compressions. Unfortunately, our capillary system is not thermostated. Following suggestions in the literature [ll], formamide was added to the gels to improve their denaturing capacity [lo]. These gels produce improved resolution at the expense of decreased sequencing rate. In this paper, we report the effect Correspondence: Dr. Norman J . Dovichi, Department of Chemistry, University of Alberta. I?lnionton, Alberta, Canada T6G 2G2
of formamide on the sequencing rate, resolution, and efficiency in capillary gel electrophoresis at room temperature.
2 Materials and methods The capillary electrophoresis system has been described before [8,10]. The polyimide-coated, fused silica capillary is 50 pm inner diameter, 190 pm outer diameter. The gel formulae are described below.The injection end of the capillary is held in a Plexiglas box equipped with a safety interlock.The other end of the capillary is inserted into the flow chamber of a sheath flow cuvette. Fluorescence is excited with a low power helium-neon laser ( " G r e - N E ) operating in the green at 543.5 n m ; fluorescence is collected at right angles with a microscope objective, imaged onto a pinhole, passed through a spectral filter, and detected with a photomultiJlier tube that is cooled to -2O"C.The current from the pho.omultiplier tube is dropped across a resistor, digitized with d Rainin analog-to-digital converter, and recorded with a Llacintosh Plus computer. The samples are prepared based on the peak-height-encoded sequencing strategy develo7ed by Tabor and Richardson [12].The samples are synthesized with 16 pmol ofeither ROX-orTAMRA-labeled primer, 4 pmole of M13mp18 single-stranded template, and a nucleotide mixture in the following ratio: 8d/ddA: 4d/ddC: Zd/ddG: Id/ddTwith deoxynucleotide to dideoxynucleotides in a 300:l ratio. The samples are resuspended in a 49: 1 mixture of formamide-EDTA. Three pL aliquots are taken from the resuspended samples, heated to 95°C for 2 min, and injected onto the capillary by applying a 200 V/cm electric field for 20 s. Prior to preparation ofthe gels, about 1 cm of the detection end of the capillary is dipped into y-methacryloxypropyltrimethoxysilaneand allowed to air-dry. Gels are prepared in 5 m L aliquots from carefully degassed acrylamide/N,N'-methylene bisacrylamide (6% T, 5 % C), X Tris-borate-EDTA (TBE) buffer, 7~ urea, and varying amounts of formamide (0-30% by volume). Polymerization is initiated b y addition of 2 pL of N,N,N',N'-tetramethylethylenediamine (TEMED) and 20 pL of 10% ammonium persulfate. The solution is injected into the capillary by use of a syringe. Although polymerization appears complete in 30 min, the capillaries are typically stored overnight before use. With the use ofhigh purityreagents and careful degassing, more than 95 O/o of the gel-filled capillaries are bubble-free and generate useful data. 01 73-0835/92/0808~0484 $3.50+.25/0
485
Forniuniide modified polyacryiamide gels for DNA 5equenciiig
3 Results and discussion A total of 26 sequencing gels (all 6 % T, 5 O/o C, 7~ urea, and O-30°/o formamide) were poured and evaluated with an M13mp18 sequencing sample. Either five or six capillaries were evaluated for each formamide concentration; the relative standard deviation in sequencing rate ranged from 1 to 13O/o and presumably was due to minor fluctuations in acrylamide concentration. During the degassing step, the total volume of the solution would decrease; reconstitution of the solution to the original volume introduced a variation in acrylamide concentration. Figure 1 presents the sequencing rate as a function of formamide concentration at the 95% confidence interval. The sequencing rate was constant for gels containing less than 20% formamide; a 30 cm long gel, operated at 200 V/cm, produced a sequencing rate of 275 bases/h. However, addition of 30% formamide decreased the sequencing rate to 175 baseslh.
A sequencing reaction was performed with only a C-terminator. This sample was separated on a 0% formamide gel (Fig. 3a). Fragments 56-59 are baseline resolved, whereas fragments 65-66 are compressed. Addition of'30°/o formamide degraded slightly the separation of the 56-59 fragments, but improved the resolution of fragments 65-66 (Fig. 3b). That is, fragments that are free of compression are better resolved in 0 O/o formamide gels whereas compressions are better resolved in 30% formamide gels. Resolution is definegas the ratio of peak spacing, A T to average peak width, W
The theoretical plate count was measured for fragments 47, 84 and 154 nucleotides in length. Theoretical plate count was estimated as N = 5.54
($)
where T, is the retention time and W,,, is the full-width at half-height of the peak. Figure 2 presents the theoretical plate count for the fragments 47 nucleotides in length as a function of formamide concentration; again, the data are presented at the 95 Vo confidence interval. Theoretical plate count is constant for 0.5, and 10% formamide gels. There is a significant loss in plate count for gels containing 30% formamide. Longer fragments show similar trends. The degradation in plate count for gels with higher formamide concentration may reflect the poor thermal properties of the additive. Formamide has poorer thermal conductivity and heat capacity than water, and formamide-modified gels will operate at higher temperature and thermal gradient than aqueous gels.
0.0
10.0
20.0
30.0
% Formamide Figurr2. Theoretical plate count for47-merfragment,95% confidence interval. The same gels as in Fig. I were used in this study.
325 300 27 5 0.351
;
:
I
33.0
34.0 35.0 Time (Min.)
36.0
33.0
34.0 35.0 Time (Min.)
36.0
250 225 200 175 150
0.0
10.0
20.0
2
% Formamide Figure 1. Sequencing rate as a function offormamide concentration in 7M urea, 6%T, 5%C polyacrylamide gels. One cm of the gel was covalently bound to the capillary wall. The fused silica capillary was 30 crn long, SO vm inner diameter,and operated at an electric field strength of 200 V/cm. The data are shown at the 95% confidence interval, with either 5 or 6 data averaged for each point.
0.40
Figure 3 Separation of a C-terminated sequencing sample in a 7 M urea, b%T, 50ioC polyacrylamide gel operating at 200 V/cm. ( b ) Separation o f a C-terminated sequencing sample in a 30% formamide, 7~ urea, 6%T, SO/aC polyacrylamide gel operating at 200 V/cm.
486
M J Kuclicleau
PI
iil.
Figure 4 presents resolution as a function of formamide concentration. Gels containing 10Yo and less formamide generate excellent resolution ( R = 1.9) of fragments 56-59 that are free of compression (solid circles, 95 Yo confidence limits pointing to the right). Up to 10% formamide can be added before a decrease in baseline is observed.The 20 and 30% gels produce poorer resolution, about 1.3. Decreased resolution for the 20 and 30% formamide gels is related to the decrease in plate count noted for the higher formamide concentration gels. Resolution is proportional to the square root of plate count. The factor-of-two decrease in plate count noted produces the factor-of-0.7 decrease in resolution. Figure 4 also presents resolution of fragments 65-66 that show compression (solid squares and 95Yo confidence limits pointing to the left). Gels containing 10-30°/o formamide generate a factor-of-two improvement in resolution compared with the 0% gel. The denaturing properties of formamide, which acts to disrupt secondary structure of the fragments, leads to improved resolution of the compressions. The compressions are not completely relaxed under these conditions and the resolution is less than that produced by normally migrating fragments.
4 Concluding remarks Addition offormamide to sequencing gels improves the resolution of DNA compressions. However, gels containing higher concentrations of formamide have lower sequencing rate, theoretical plate count, and resolution for normal fragments. A 10[l/oformamide concentration appears optimum. Sequencing rate, theoretical plate count, and resolution of normally migrating fragments are similar to that ob-
r 2.0
1.5
1.0
0.5
- 0.0.
.
This w o r k was supported i n p a r t by the Department qf'Ener
