Centrifugal ultrafiltration is a well-established method for concentrating and purifying DNA. Here, we describe the use of centrifugal ultrafiltration for the separation of plasmid DNA isoforms based on differences in elongational flexibility of the supercoiled, open-circular, and linear plasmids. Transmission of each isoform is minimal below a critical value of the filtration velocity, which is directly related to the magnitude of the centrifugal speed and the system geometry. A discontinuous diafiltration process was used to enrich the desired isoform, as determined by agarose gel electrophoresis. The simplicity and efficacy of this membrane-based separation are attractive for multiple applications requiring the use of separated DNA isoforms.
Centrifugal ultrafiltration (UF) is used extensively for laboratory-scale puri f ication, concentration, and isolation of a variety of biomolecules. This includes the use of centrifugal UF for rapid concentration of Legionella antigen in urine (1), the isolation of free benzalkonium chloride (BAC) molecules from micelle solutions (2), the analysis of low-molecular weight proteins in the human plasma proteome (3) and saliva (4), and the analysis of plant extracts (5).
Centrifugal UF is also a well-established method for concentrating and purifying DNA. Krowczynska and Henderson (6) described the use of centrifugal UF for the removal of unincorporated nucleotides and PCR primers after DNA amplification by the polymerase chain reaction. Schratter et al. (7) summarized the use of centrifugal UF in the construction of cosmid libraries (hybrid plasmids containing cos sequences), removal of restriction enzymes, and concentration of RNA. Centrifugal UF is also widely used in forensics laboratories for isolation and concentration of genomic DNA (8).
All of these applications of centrifugal UF involve recovery of large DNA in the retentate, with smaller molecules, impurities, and buffer components passed through the membrane and into the filtrate. However, recent work (9-11) has demonstrated that large DNA molecules with lengths of 3–17 kb (kilobase pairs) can actually be transmitted through the small pores in traditional ultrafiltration membranes under certain conditions because of flow-induced elongation of the DNA. This opens up exciting opportunities for DNA purification, e.g., for removal of large cell debris, viruses, or bacteriophage, with the DNA recovered in the filtrate. In addition, Latulippe and Zydney (12) demonstrated that pressure-driven ultrafiltration could also be used to separate the linear, supercoiled, and open circular isoforms of a given plasmid based on differences in elongational flexibility of these topological isoforms.
The objective of this work was to demonstrate the feasibility of using centrifugal UF for DNA separation, including removal of unwanted plasmid isoforms. Initial experiments were performed using the individual isoforms to identify the critical centrifugal conditions for retention and transmission of the plasmid. DNA separations were then performed using commercially available centrifugal UF devices, with the composition of the DNA in the filtrate and retentate analyzed by agarose gel electrophoresis (AGE).
Materials and methods
Experiments were performed using linear, open circular, and supercoiled isoforms of a 9.8 kb plasmid in which a 6.84 kb fragment was inserted into the SalI site of the pBluescript II KS+ plasmid vector. Stock solutions of the supercoiled plasmid were provided by Dr. Jeff Chamberlain at the University of Washington. The linear and open circular isoforms were obtained by enzymatic digestion using 2 U/µg of respectively KpnI and Nt.AlwI enzymes (New England Biolabs, Ipswich, MA, USA). BSA and the recommended buffer were added to 27.5 µg of supercoiled plasmid DNA with the final volume of the digestion solution adjusted to 275 µL by addition of deionized water obtained from a Barnstead Nanopure water purification system (Thermo Scientific, Dubuque, IA, USA) with a resistivity greater than 18 MΩ. The digestion mixtures were incubated at 37°C for 3 h and then purified using a QIAquick PCR purification kit (QIAGEN, Valencia, CA, USA) which uses a silicagel membrane to bind the DNA in high-salt buffer while the impurities are washed away. The DNA was then eluted with low-salt Tris-EDTA (TE) buffer made by diluting a TE concentrate (Sigma Aldrich, St. Louis, MO, USA) with deionized water. The final concentrations were measured by a NanoDrop 2000c UV spectrophotometer (Thermo Scientific) using the absorbance at 260 nm. The purified plasmid solutions were stored at -20°C and warmed to room temperature immediately prior to use.
Experiments were performed using NanoSep centrifugal UF tubes obtained from Pall Corporation (Ann Arbor, MI, USA). Each tube contains a feed reservoir with an encapsulated Omega polyethersulfone ultrafiltration membrane with an effective filtration area of 0.28 cm2 and a nominal molecular weight cutoff of 100 kDa. Membranes were initially flushed with 450 µL of 0.05 N NaOH followed by 1 mL of deionized water. A 5424 microcentrifuge (Eppendorf, USA) with a fixed-angle rotor (FA 45–24–11) of 45° was used for centrifugation. The centrifugal speed could be varied up to 15,000 rpm, which corresponds to approximately 16,520 × g (for a centrifuge tube filled with 450 µL of liquid) based on the rotor arm length of 0.066 m.
The NanoSep centrifuge tube was initially filled with 450 µL of the feed, the rotation speed was set, and the tube was spun for a given period of time. The tube was then removed from the microcentrifuge, with the filtrate collected and weighed using a MS104S analytical balance (Mettler Toledo, Columbus, OH, USA) with an accuracy of 10−4 g. A small sample of the filtrate and retentate were collected for analysis of the plasmid concentration as described subsequently.
Multistep discontinuous diafiltration experiments were performed as follows. The NanoSep centrifuge tube was prepared as above and a centrifugal UF was performed to reduce the volume from 450 µL to approximately 50 µL. The tube was then removed from the centrifuge with the filtrate collected and weighed. The retentate remaining within the tube was then diluted with additional TE buffer back to the initial volume. This entire procedure was repeated multiple times using the same centrifugal conditions, with the filtrate removed at the end of each UF step. A final sample was obtained from the retentate sample at the end of the experiment.
The total DNA concentration was evaluated using Quant-iT PicoGreen dsDNA assay kit (Life technologies). 75 µL samples of the DNA solution were loaded into the wells of a 96-well GENiosFL microplate reader (TECAN, Durham, NC, USA) along with an equal volume of the PicoGreen reagent that was first diluted 200:1 with analytical TE buffer (Life Technologies). Samples were mixed for 180 s at 36°C using an orbital shaker, with the fluorescence intensity measured at 535 nm. Actual concentrations were determined by comparison with a calibration curve constructed using known standards.
The composition of solutions containing mixtures of the different isoforms was determined by AGE. A 0.7% (w/w) agarose solution was prepared by dissolving 0.98 g of agarose powder (EMD Chemicals Inc., Gibbstown, NJ, USA) in 140 mL of diluted 10X Tris-Acetate-EDTA (TAE) buffer (Mediatech Inc., Manassas, VA, USA). An agarose gel was cast in an Owl B2 EasyCast mini gel electrophoresis system (Thermo Scientific) using a 12-tooth comb (B2–12-EA) with tooth thickness of 1.5 mm. Each well was loaded with 18 µL of the DNA mixture with 6× loading dye. A 1-kb DNA ladder (New England BioLabs) was loaded into at least one well as a reference. An electric field of 70–75 V was applied with a PowerPac Basic power supply (Bio-Rad, Hercules, CA, USA) for 2-2.5 h. The voltage was removed after sufficient time, and the gel was stained in diluted 10,000× SYBR Gold nucleic acid gel stain solution (Life Technologies) for 60–70 min. Gels were imaged using a Fluorchem FC2 image system (ProteinSimple, Santa Clara, CA, USA).
Results and discussion
Figure 1 shows typical experimental data obtained using NanoSep centrifugal UF tubes fitted with an Omega polyethersulfone ultrafiltration membrane with nominal molecular weight cut-off of 100 kDa. The tubes were initially filled with 450 µL of the supercoiled, linear, or open circular isoform of a 9.8 kb plasmid suspended in 10 mM Na2EDTA Tris buffer containing 10 mM NaCl at a concentration of 1 µg/mL. Each data point represents results from a separate centrifugation. In each case, the centrifugation time was adjusted to collect 160 ± 10 µL of filtrate based on the model presented in the Supplementary materials. The y-axis shows the concentration of the plasmid isoform collected in the filtrate solution for each centrifugation, determined using the measured fluorescence of the PicoGreen stain. The x-axis is the set value of the angular velocity used for the centrifugation. At any given angular velocity, the filtrate concentration for the linear isoform was significantly greater than that of the supercoiled isoform, with relatively little of the open circular isoform transmitted through the membrane out to even 5500 rpm.