|Home | About | Journals | Submit | Contact Us | Français|
Like most other DNA sequencing core facilities, one of our continuing goals is to improve our sequence output without substantially adding to cost. To minimize sample-to-sample variability in template DNA concentration, we implemented the rolling circle amplification (RCA) procedure for preparing our DNA templates. In addition to saving time and reducing the number of steps in template DNA preparation, the RCA method has the potential to normalize the DNA concentration in samples that can be sequenced directly without additional purification. In the present study, we used RCA-generated templates to test a recently reported procedure that increased sequence quality by resuspending the sequenced products in low concentrations of agarose before capillary electrophoresis (CE) on a MegaBACE 1000 platform. Although we did not obtain the expected result using the specified procedure, a modification resulted in up to 60% increase in total sequence yield per sample plate. A combination of agarose and formamide-EDTA in the resuspension solution enabled us to generate long-read and high-quality sequences for more than 38,000 templates with minimal additional cost.
Capillary DNA sequencing revolutionized the sequencing process. It played a major role in the early completion of the draft sequence of the human genome. Increased throughput through automation of gel preparation and sample loading have made capillary electrophoresis (CE)-based machines a mainstay not only in large genomic research laboratories but also in small core service facilities. Although CE sequencing offers advantages such as ease of operation and reduced user interaction, some factors like DNA quality and concentration are still a major issue in CE sequencing. The small volume of the separation matrix (typically 2–3 μL), the nature of electrokinetic injection, and the presence of residual high molecular weight DNA template and protein contaminants in the sample even after the postreaction cleanup are several factors that may cause ion overloading of the capillaries and subsequent degradation in sequence quality.1–3 Deciding which electrokinetic injection parameters to apply to any given sample plate is not always easy, especially in a core service facility where users’ sample DNA concentration and quality vary widely among wells, plates, projects, and laboratories. To get the maximum sequence output from any given run, one must use optimum injection conditions to minimize or at least balance the number of wells giving low sequence signals (i.e., underloading) and those showing low capillary current (i.e., overloading). Maintaining a balance between overloaded and underloaded samples has been a problem and, oftentimes, leads to several reruns of the same sample plate. This obstacle is probably more accentuated in the MegaBACE 1000 DNA sequencing platform, which is known to be sensitive to the amount and purity of the template.4,5
We approached the DNA concentration problem initially by including plasmid purification using the Qiagen 9600 robotic system as part of our sequencing service in order to control DNA quality of the templates fed through our sequencing pipeline. After plasmid purification, we normalized template DNA concentration among wells to a narrow range.4 Our semiautomated protocol involved (a) taking OD280/260 readings of the robotically prepared plasmids, (b) converting these readings to DNA concentrations, and (c) using the diluting capability of the Qiagen robot to normalize the concentrations within a 2-fold range before setting up the sequencing reactions. Although implementation of this procedure resulted in improved sequence yields, this multistep normalization protocol is time-consuming, rendering its utility in a high-throughput production-sequencing environment impractical.
A recently introduced DNA amplification kit based on the replication of circular bacteriophage DNA molecules seems to be the answer to the problem of sample-to-sample variability in DNA concentration. The rolling circle amplification (RCA) method,6,7 which utilizes the TempliPhi reagent kit (Amersham Biosciences, Piscataway, NJ),8 is an isothermal amplification method enabling production of microgram quantities of DNA (from picograms of starting material) that can be used directly for sequencing without additional postreaction purification. The RCA procedure effectively reduces the time and number of steps in template DNA preparation. In addition, if the amplification process proceeds for 16–18 h, the RCA method will, in theory, normalize the DNA concentration in all samples, which makes it an ideal step before capillary sequencing reaction.9 Although the RCA method generates high yields of amplified DNA in almost every sample, we still notice some variation in DNA concentrations (Fig. 11).). Increasing injection times while keeping the voltage the same or lower helps improve the signals from templates with lower DNA concentration3,10 but can also cause poor resolution of sequence signals due to ion overloading of templates with higher initial DNA concentration.3 Elkin et al.11 solved this template-overloading problem by binding template to magnetic beads used for postreaction cleanup. A simpler solution using agarose, however, was reported in a recent study. Researchers found improvement in sequence quality and read length when sequencing reaction products were resuspended in low concentrations of agarose prior to electrokinetic injection.12 They attributed this effect to the ability of agarose to “gel-purify” the sequencing products by preventing large template DNA and protein contaminants from entering the capillary. We tested this method on production DNA templates prepared by the RCA method in conjunction with our MegaBACE 1000 sequencing system. Contrary to previous reports, the sequence data generated from samples resuspended in the specified agarose solution were generally inferior compared with those in the standard MegaBACE loading solution, which is basically a 70% formamide/1 mM EDTA mixture (FE). Surprisingly, when the agarose solution was added to samples already resuspended in FE, we noticed a significant increase in sequence yield.
Templates used in this study were either (a) plasmids generated through the RCA process using the TempliPhi kit or (b) purified pBluescript plasmid vector as standards. We purchased the 10,000-reaction TempliPhi kit. Each of the five bottles containing 20 mL TempliPhi solution was thawed on ice, divided into 1-mL aliquots, flash-frozen in dry ice, and stored in −80°C. Each aliquot was thawed on ice just before use. Any leftover TempliPhi in the tube was kept at 4°C and used within 12 h. Our scaled-down (1/4 the original) version of the RCA protocol has a final volume of 5 μL, which included 0.5 μL of diluted (1:10) glycerol stock or bacterial culture containing the plasmid DNA. Any problem associated with reproducibility and stability of the RCA kit encountered initially and during the development of the scaled-down procedure was kept to a minimum by (a) strictly adhering to the aforementioned storage guidelines, and (b) avoiding repeated freezing and thawing of the TempliPhi aliquots. All samples used were in 96-well format. Ten microliters of the diluted (1:5) amplified product was sequenced directly (without purification) in a 20-μL ET terminator cycle sequencing reaction (Amersham Biosciences). The 20-μL reaction mixture was divided into two 10-μL aliquots prior to ethanol precipitation postreaction purification. After ethanol precipitation, the air-dried extension products in one plate were resuspended in 10 μL of 0.06% aqueous solution of Seakem Gold agarose (Cambrex, Rockland, ME) at 65°C. The products in the other replicate plate were resuspended in 5-μL FE. Both plates were vortex-mixed, spun, and run under the same injection and running conditions. The quality of the sequences were determined using the Phred Q >20 (Phred 20) criteria (http://www.phrap.org). Passing rate was defined as the percentage of samples that give 50 Phred 20 bases or higher.12
The result of our initial experiment using purified pBluescript plasmid as our template is shown in Figure 22.. Contrary to our expectation, the samples resuspended in agarose alone gave shorter Phred 20 read lengths compared with those in FE (average success read length of 291 ± 80 vs. 525 ± 107, respectively). In fact, the total number of Phred 20 bases per plate was only half as much in the plate with agarose as in FE (23,241 vs. 46,754). Detailed analysis of the electropherograms of the raw (Fig. 33)) and analyzed (Fig. 44)) sequence data revealed several things. First and most obvious is the poor resolution of the peaks in samples resuspended in plain agarose as evidenced by the excessive tailing of the peaks (Fig. 3A3A)) compared with those in FE (Fig. 3B3B).). Second, these poorly resolved peaks have greater signal intensities (2–3 times) than those in FE. Surprisingly, poor resolution occurred only in the first 250 bases, as shown in the analyzed sequence (Fig. 4A4A),), beyond which sequence quality improved. On the other hand, high-quality sequences were present even in the first 50 bases of samples resuspended in FE alone (Fig. 4B4B).
The loss of high-quality sequence in the agarose-suspended samples was unexpected. Before abandoning agarose resuspension, we considered adding agarose to samples already suspended in FE solution and running them under similar conditions. Contrasting with the previous result, we observed an unanticipated 21% increase in total Phred 20 bases per plate (56,678 vs. 46,674). This result suggested that, under our sequencing conditions, FE must be present in the resuspension solution in order for agarose to work. We, therefore, predicted that adding FE to samples previously resuspended in agarose should also increase their sequence yield. Indeed, the total Phred 20 bases per plate increased by 127% (52,762 vs. 23,241) simply by adding FE to agarose-suspended samples.
In a core service facility like ours, variation in DNA sample quality among samples or projects is common. The problem we most regularly encounter is ion overloading. Sample plates showing more than 10 overloaded wells are aborted and rerun with less injection time. Foreseeing the potential benefits agarose addition might bring to problematic plates, we tested archived sequencing plates that were less than one month old to see if we could improve the total Phred 20 bases output per plate. Sequencing plates that have been stored in FE resuspension medium at −20°C for 3–6 months do not show any signs of degradation of sequence quality (data not shown). Ten microliters of 0.06% agarose (0.04% final concentration) was added to FE (5 μL)-suspended samples and run under the same electrokinetic injection and running parameters (voltage, time) used previously for each plate. Table 11 shows the data obtained before and after addition of agarose to 10 previous production plates and a standard pBluescript plate. A highly significant percent increase (37.0 ± 12.6%) in total Phred 20 bases per plate was obtained when agarose was added to samples previously run with FE solution alone. Additionally, sample plates with lower initial total sequence yield (e.g., samples A and B) gave higher percentage increases, and vice versa. This inverse relationship is evident when percent increase was plotted against the initial total Phred 20 bases per plate (Fig. 55).
Increase in total Phred 20 bases per 96-well plate may be due to either an increase in Phred 20 read length or passing rate, or both. Plots of Phred 20 read length versus passing rate for sample plate A and the standard pBluescript template are shown in Figures 6A and BB.. Detailed analysis of sequence data revealed that percent increases in passing rate and Phred 20 read length for plate A and the pBluescript standard are 18 and 55, and 7 and 81, respectively. The results indicate that a higher percentage of samples in low-sequence yielding plates (e.g., plate A) showed greater passing rate after agarose addition compared to high-sequence yielding plates, such as the pBluescript standard plate. This was expected because of the already high initial passing rate (93%) of samples in the standard plate even before agarose addition. Consequently, a greater portion of the increase in total Phred 20 bases in the standard plate was attributed to increase in read length.
How does agarose addition augment total sequence output? A further experiment helped explain the positive effects of agarose addition. It is widely known that ion overloading has a detrimental effect on the resolution of the labeled fragments. Reduction in capillary current is almost always associated with ion overloading. We determined the number of samples with reduced capillary current in plates injected in the presence or absence of agarose (Table 22).). Reduced currents seen in FE-suspended samples were almost entirely eliminated after agarose addition (19 ± 13 vs. 1 ± 1).
Believing our core service facility would benefit tremendously from the improved passing rate and longer read length attainable with the agarose addition, we immediately incorporated this method into our routine sequencing protocols. Utilizing this procedure, we have processed more than 38,000 samples and produced high-quality and long-read sequences from RCA-generated DNA templates. The 37% average boost in our total sequence output tremendously outweighs the added cost of agarose (less than $0.01 per 96-well plate) and the 5–10 additional minutes invested in plate preparation.
Increased automation and higher throughput gives CE-based DNA sequencing instruments more advantages than slab-gel systems. However, the exceptionally small volume of separation matrix and the nature of electrokinetic injection are a few reasons for the high sensitivity to ion overloading in CE sequencing. In addition, the large variation in the quality of the DNA library and, consequently, concentration of DNA among plates and projects faced by service core facilities help aggravate this overloading problem in CE sequencing, especially in MegaBACE 1000 systems.4,5 To minimize this problem, we currently have in place two protocols, namely the RCA method for template amplification8 and the addition of low concentrations of agarose12 to the FE-suspended samples prior to electrokinetic injection. The ease of DNA preparation and the ability to directly sequence DNA templates without purification are some advantages of the RCA method, in addition to normalization of DNA concentration when the amplification reaction is allowed to proceed to completion.10 Although the original agarose procedure suggested by Vatcher and co-workers12 did not work successfully under our running conditions (Figs. 11 and 22),), a modification allowed us to generate longer sequence read lengths and higher passing rates, which resulted in higher total sequence output per plate (Table 11).). The positive agarose effect is attributed to its ability to “gel-purify” the samples during the electrokinetic injection process. This particular type (Seakem Gold) of agarose at low concentrations (preferably between 0.02 and 0.2%)12 produces pores that are large enough to retard the passage of large molecules, such as protein and template DNA, which lead to poor-quality sequence when present in high concentrations.3,11,12,14 In our study, agarose effectively reduced and almost eliminated ion overloading as evidenced by the disappearance of more than 90% of overloaded samples after agarose addition (Table 22).
Resuspending samples in aqueous agarose solution alone and running them under our conditions did not improve sequence quality and output. In fact, our results showed poor resolution of peaks in the first 300 bases (Fig. 4A4A),), which is contrary to results previously reported.12 It is possible that (a) our ethanol precipitation procedure to purify samples prior to sample injection reduced the salt concentration below 50 μM, and (b) agarose did not completely remove the template DNA but only reduced the amount to a level that did not cause any ion overloading of the capillaries. Salas-Solano and co-workers showed that when template DNA is present in the resuspension solution, salt has to be present in at least 50 μM concentration in order to achieve good resolution of the sequence signals, especially the first 300 bases.3 In our case, enhanced peak resolution observed beyond 300 bases in samples resuspended in plain agarose probably indicates that a certain amount of time is needed for salts in the running buffer to equilibrate and subsequently raise the salt concentration created by the passage of the desalted sample through the capillary. On the other hand, the optimum range of salt concentration might have been achieved in samples resuspended in FE even before electrokinetic injection, resulting in longer Phred 20 quality read lengths. The same reasoning could explain the long-read sequences obtained in the previous study using only aqueous solution of agarose for resuspension.12 Since salt concentration in ethanol-precipitated samples can vary from 20 to 230 μM,13 it is possible that the concentration of salt remaining in their ethanol-precipitated samples is within the optimum range for good peak resolution.
It is our continuing goal to achieve higher quality and longer read sequences while keeping the cost to a minimum. In the present study, we obtained sequences from RCA-generated templates with increased passing rates and longer Phred 20 read lengths by including a combination of FE and low concentrations of agarose in the resuspension medium prior to electrokinetic injection. This cost-effective method improved our total sequence output by up to 60%. Since implementation of this protocol, we have processed and generated high-quality sequences for more than 38,000 DNA templates.
We thank Regina Shaw for her expert technical assistance.
There are no known conflicts of interest on the part of any of the authors.