3148
Biomacromolecules 2009, 10, 3148–3150
Notes Molecular Weight and Polydispersity of Calf-Thymus DNA: Static Light-Scattering and Size-Exclusion Chromatography with Dual Detection Bedrˇich Porsch,* Richard Laga, Jirˇ´ı Horsky´, ˇ estmı´r Konˇa´k, and Karel Ulbrich C Institute of Macromolecular Chemistry, Academy of Sciences of the Czech Republic, v.v.i., 162 06 Prague, Czech Republic Received July 8, 2009 Revised Manuscript Received September 9, 2009
Plausible calf-thymus DNA molecular weight distribution can be obtained by size-exclusion chromatography with dual lowangle light scattering/refractometric detection at sufficiently low flow rate. The distribution extends over three decades of molecular weight and is characterized by weight average molecular weight Mw ) 8418000 and polydispersity index Mw/ Mn ) 5.2. After strongly scattering impurities had been removed from the sample using adsorption properties of the 3 µm mixedcellulose-ester filter membranes, static light-scattering measurement in flow injection mode was feasible and gave Mw ) 8580000, corroborating the veracity of SEC results.
Introduction Calf-thymus (CT) DNA, which has been commercially available for a long time from various vendors at a reasonable price, is widely used in diverse biophysical and biochemical studies (Web of Science has provided 1140 references to CTDNA during the latest five years). The commercial samples are referred to as “highly polymerized”, “polydisperse”, “fibrous preparation”, “containing low amount of RNA and proteins”, and characterized by water content and UV absorbance per mass. The information about their molecular weight and polydispersity is not available from the suppliers and cannot be obtained from electrophoresis, routinely used for DNA characterization. In the literature, Mw of CT-DNA samples appears scarcely (80000001,2 and 6000000,3 that is, the values are in the ultrahigh molecular weight range); quantitative information on CT-DNA polydispersity is, to the best of our knowledge, missing completely. This poses a major impediment to physicochemical studies utilizing CT-DNA as experimentally accessible quantities (obtained by methods such as static and dynamic light scattering (LS), viscometry, etc.) are related to the molecular weight in different ways and thus difficult to compare because they are affected by the sample polydispersity differently. Recently, we studied4 formation and transformation of DNA-poly(lysine) complexes as models for gene delivery vectors, using, among other methods, static and dynamic LS, and found the lack of the information on CT-DNA polydispersity rather frustrating. Polydispersity of macromolecular samples is routinely assessed by the size-exclusion chromatography (SEC); however, * To whom correspondence should be addressed. E-mail: porsch@ imc.cas.cz.
the molecular weight of CT-DNA is considered too high for successful SEC analysis.1 In our recent SEC studies of ultrahigh molecular weight poly(ethylene oxide)5 and ultrahigh molecular weight sodium hyaluronate,6 we found that their SEC separation is substantially disturbed by diverse flow-retardation effects, including slalom chromatography but that the nonbiased molecular weight distribution might be obtained using sufficiently low mobile phase flow rate. To verify that this approach is valid also for CT-DNA, we performed and optimized SEC of CTDNA. We found that CT-DNA can indeed be analyzed by SEC if low enough flow rate was used. The procedure described in this note can be used also for other polydisperse DNA samples. To verify SEC results and especially to check that no sample is “lost” within the chromatography column, we also determined Mw of CT-DNA by static LS experiments in flow-injection mode. The light-scattering signal from solutions of water-soluble polymers prepared from biological sources is generally degraded by the presence of strongly scattering compact impurities. We found that hydrophobic adsorption on a suitable filtration membrane effectively removes these strongly scattering particles from CT-DNA and makes static LS possible.
Experimental Section Materials. CT-DNA sodium salt, type 1, “highly polymerized” (Lot No. 091K7030) having 16.9 A260 units/mg solid was from Sigma. Analytical reagent grade NaCl was obtained from Merck (Darmstadt, Germany) and used without further purification. Water was from a Millipore Milli-QPLUSUF ultrapure water purification unit (Millipore Corp., Bedford, MA). Methods. Modular SEC chromatograph consisted of a Shimadzu LC-10ADVp pump (Shimadzu Corp., Kyoto, Japan), a vacuum degassing unit DEGASYS (Sanwa Tsusho, Ltd., Tokyo, Japan), a Pharmacia injection valve V-7 with 500 µL loop (Pharmacia and Upjohn, Uppsala, Sweden), a Chromatix KMX-6 low-angle light-scattering detector (LDC/Milton Roy, Sunnyvale, CA) and a Waters 2410 differential refractometer (Waters Assoc., Milford, MA) connected through a Black Star (Huntingdon, UK) 2308 A/D converter to an IBM-compatible computer. The separation column was a TSKgel GMPW linear (7.5 × 600 mm) column, particle size 17 µm, (Watrex, Prague, CR). Aqueous sodium chloride (0.1 M) was used as a mobile phase in all experiments. This SEC setup was transformed to a flow-injection static LS system using a Teflon capillary (length 60 cm, inner diameter 0.5 mm) instead of the SEC column and a 10 mL Superloop (Pharmacia & Upjohn, Uppsala, Sweden), which acts as a mobile-phase-driven syringe, was mounted instead of a capillary loop. The injected volume in flowinjection experiments was usually 4 mL. A detailed description of both techniques is given elsewhere.5,6 0.1% CT-DNA in mobile phase was prepared by gentle shaking for 3 days. This stock solution was diluted to a nominal working concentration of 25 µg/mL for SEC and static LS experiments (3 h gentle mixing). Sample Filtrations. Filtrations prior to injections were performed using a “Genie” programmable syringe pump (Kent Scientific Corporation, Torrington, CT) at a flow rate of 0.25 mL/min. Hydrophilic PVDF and hydrophobic PTFE membrane filters having porosity 1 µm (Puradisc 13 mm, Whatman, Maidstone, U.K.) and MCE (mixed cellulose esters) membranes with a porosity of 1.2 and 3 µm (Millipore) in a 13 mm Teflon holder (Whatman) were used because coiled macromolecules of CT-DNA are too large for commonly used filters
10.1021/bm900768j CCC: $40.75 2009 American Chemical Society Published on Web 10/09/2009
Notes
Biomacromolecules, Vol. 10, No. 11, 2009
3149
Figure 1. Light-scattering response of CT-DNA solutions after filtration with adsorbing MCE and nonadsorbing PVDF filter.
having 0.2 µm porosity or less. Sample concentrations before and after filtration were verified from the UV absorbance at 260 nm using a Helios R spectrophotometer (Thermo Scientific, MA).
Results and Discussion Strongly scattering impurities were found detrimental to static LS experiments when CT-DNA solutions were either unfiltered or filtered by 1 µm PVDF filter. Such impurities were successfully removed from solutions of hyaluronic acid by hydrophobic adsorption on a PTFE filter in our previous study.6 Not surprisingly, this filter was completely blocked by DNA, known to exhibit significant hydrophobic interactions (e. g., solubilizes carbon nanotubes in aqueous environment7). MCE filters of lower hydrophobicity were found adequate. The efficiency of particle removal gradually decreased after passage of approximately 2 mL of sample solution through a single MCE filter membrane. By stacking four MCE filter membranes in the holder, more than 15 mL of solution could be purified. Hydrophobicity of DNA implies its possible coadsorption with impurities. We checked the loss of mass after filtration by UV absorption. MCE filters trapped 17-19% of sample mass; on the other hand, the UV absorption was almost unchanged by filtration through PVDF filters, as expected. Both DNA and impurities being hydrophobic, their mutual hydrophobic interaction and aggregation may be expected in solution. To verify that mainly DNA is adsorbed on the MCE filter, we determined the DNA loss after filtration using DNA specific fluorescence of its ethidinium bromide complex and found values between 16 and 18%. Hence, the amount of impurities could be fairly low. By a Bradford method using Coomassie protein assay kit (Pierce, IL), we determined the content of protein in the sample to be about 1% in unfiltered solution and zero in the filtered one. Figure 1 illustrates the effect of strongly scattering impurities on static LS experiment when CT-DNA solution was filtered with hydrophilic (nonadsorbing) 1 µm PVDF filter and compares this behavior with static LS response when these impurities are properly removed by adsorption on 3 µm MCE membranes. A pronounced spiking in the case of PVDF filter indicates nonergodic behavior of impurities present at rather low concentration and makes the calculation of Mw impossible. Despite of the low amount of impurities, a dramatic decrease of scattering intensity after MCE filtration is observed (Figure 1). The content of water in the sample must be known because the calculation of Mw requires the true sample concentration. We used Karl Fischer titration and found the value 16.8%. Pure double stranded DNA should have the unit absorbance (260 nm) at 50 µg/mL.8 If we assume that all nonabsorbing part of the sample is water and use the supplier’s absorbance value, we obtain 15.5% in a good agreement with the Karl Fischer value. The literature values of refractive index increment dn/dc
Figure 2. SEC distribution analysis of CT-DNA: Corresponding LS and RI signals, calculated composite log M ) f(Ve) calibration with polynomial fit (a) and resulting molecular weight distribution (b).
) 0.166 mL/g and second virial coefficient A2 ) 0.00037 mol · mL/g2 were used in all calculations.9 We note that the effect of A2 was quite low at injection concentration used, comparable with static LS experimental error and completely negligible in SEC due to additional on-column dilution. Hence, a weak molecular weight dependence of A2 may be safely neglected. The resulting Mw from static LS experiments with MCE-filtered solutions was 8580000. Having previous experience with ultrahigh molecular weight polymers,5,6 we anticipated a pronounced effect of flow rate on the SEC experiment also for CT-DNA. The lowest flow rate used in our studies of poly(ethylene oxide) and hyaluronic acid (0.09 mL/min) did not facilitate reliable separation. Further reduction of flow rate to 0.068 mL/min was necessary to obtain coincidence of calibration curves log M ) f(Ve) calculated from the dual detection data obtained for original and sonicated CTDNA samples to one “master” curve as seen previously when the hydrodynamic retardation phenomena were absent.6 The absence of a sufficient light-scattering signal at higher elution volumes (Ve) is clearly visible in Figure 2a. This phenomenon is typical for broad polymer samples due to different sensitivities of differential refractometer (responding to concentration) and low-angle light-scattering (responding to product c · M) detectors. To obtain experimental points for the missing part of the calibration log M ) f(Ve), we added data obtained for a sonicated sample having suitably reduced Mw. Figure 2a shows that calibrations for the original and degraded samples fall on the common “master” curve as expected in the case of correct SEC behavior. The nonlinearity of the common calibration curve seen in Figure 2a reflects gradual changes of shape of the DNA molecule from a rod-like structure to a worm-like coil when its M increases. The detailed discussion of these phenomena is beyond the scope of this note. Briefly, rigidity of DNA is described in terms of persistence length of 50 nm.10 With the base pair having the length of 0.34 nm and a molecular weight of 663, we get M corresponding to a persistence length around 100000. At least several persistence lengths are needed for the onset of worm-like coil behavior; hence, the curvature interval seen in Figure 2a appears reasonable. Otherwise, the part of the calibration in Figure 2a in the vicinity of the exclusion limit of the column packing is similar to that observed with hyaluronic
3150
Biomacromolecules, Vol. 10, No. 11, 2009
acid, indicating mixed SEC, hydrodynamic, and slalom chromatography modes.6 Thus, a third order polynomial fit of the calibration in Figure 2a was used to calculate molecular weight distribution (Figure 2b). This broad distribution extends over three decades in M and its descriptive values are Mw ) 8418000 and polydispersity index ) 5.2. We believe that this high polydispersity is mainly a result of mechanical and shear degradation of CT-DNA during its isolation. The good agreement of Mw from static LS and SEC further supports reliability of both techniques. When MCE filtered samples were injected, the calculated sample recovery values from SEC were around 65% of the nominal injected concentration, in good agreement with the sum of the water content and DNA adsorption on the filter. To check possible effects of unfiltered impurities seen in static LS (Figure 1), we also injected PVDF-filtered solutions and obtained Mw and molecular weight distributions identical within the experimental error with those for MCE-filtered samples. The only difference was an increase in sample recovery to the values between 80-85%, roughly corresponding to the sample water content. Hence, the column used (known to exhibit weak hydrophobic interaction) retains only minor amounts of impurities. This explains why impurities not removed by filtration from the sample do not appear on the SEC lightscattering peak. Nevertheless, the MCE filtration should be preferred because after a few injections using PVDF filtrate, worsening of the LS baseline was visible due to a gradual elution of impurities from previous injections.
Conclusion We have shown that SEC with dual detection can be used to estimate the molecular weight distribution of CT-DNA if the
Notes
flow rate used is low enough. The proposed use of MCE membrane filters as adsorption media, suitable for removal of strongly scattering impurities contained in CT-DNA, allowed reliable measurements of static light scattering in flow injection mode and provided the Mw of CT-DNA in agreement with SEC. Acknowledgment. This work was supported by the Grant Agency of the Academy of Sciences of the Czech Republic (Grants A100500501, A4050403, and KAN 200200651).
References and Notes (1) Tanigawa, M.; Suzuzuto, M.; Fukudome, K.; Yamaoka, K. Macromolecules 1996, 29, 7418–7425. (2) Mason, T. G.; Dhople, A.; Wirtz, D. Macromolecules 1998, 31, 3600–3603. (3) Sundaresan, N.; Suresh, C. H.; Thomas, T.; Thomas, T. J.; Pillai, C. K. S. Biomacromolecules 2008, 9, 1860–1869. (4) Sˇubr, V.; Konˇa´k, Cˇ.; Laga, R.; Ulbrich, K. Biomacromolecules 2006, 7, 122–130. (5) Porsch, B.; Wellinder, A.; Ko¨rner, A.; Wittgren, B. J. Chromatogr. A 2005, 1068, 249–260. (6) Porsch, B.; Laga, R.; Konˇa´k, Cˇ. J. Liq. Chromatogr. Relat. Technol. 2008, 31, 3077–3093. (7) Chun, J.; Fagan, J. A.; Hobbie, E. K.; Bauer, B. J. Anal. Chem. 2008, 80, 2514–2523. (8) Mora´n, M. C.; Miguel, M. G.; Lindman, B. Biomacromolecules 2007, 8, 3886–3892. (9) Krasna, A. I.; Dawson, J. R.; Harpst, J. A. Biopolymers 1970, 9, 1017– 1028. (10) Laib, S.; Robertson, R. M.; Smith, D. E. Macromolecules 2006, 39, 4115–4119.
BM900768J
