13102 DOI: 10.1021/la101403j Langmuir 2010, 26(16), 13102–13109Published on Web 07/14/2010

pubs.acs.org/Langmuir

© 2010 American Chemical Society

Interactions between DNA and Nonionic Ethylene Oxide Surfactants

are Predominantly Repulsive

Alexandra H. E. Machado,*,†,‡ Dan Lundberg,*,§,‡ Ant�onio J. Ribeiro,† Francisco J. Veiga,†

Maria G. Miguel,§ Bj€orn Lindman,§,‡ and Ulf Olsson‡

†Laboratory of Pharmaceutical Technology, Faculty of Pharmacy, University of Coimbra, 3000-548 Coimbra,Portugal, ‡Division of Physical Chemistry, Center for Chemistry and Chemical Engineering, Lund University,

P.O. Box 124, SE-221 00 Lund, Sweden, and §Department of Chemistry, University of Coimbra,3004-535 Coimbra, Portugal

Received April 8, 2010. Revised Manuscript Received June 24, 2010

In the present work, the interactions between double-stranded (ds) or single-stranded (ss) DNA and nonionicethylene oxide (EO) surfactants, with special attention to the possible contributions from hydrophobic interactions,have been investigated using amultitechnique approach. It was found that the presence of ss as well as dsDNA induces aslight decrease of the cloud point of pentaethylene glycol monododecyl ether (C12E5). Assessment of the partitioning ofDNA between the surfactant-rich and surfactant-poor phases formed above the cloud point showed that the polymerwas preferably located in the surfactant-poor phase. Surface tensiometry experiments revealed that neither of the DNAforms induced surfactant micellization. Finally, it was shown by DNA melting measurements that another EO surfactant(C12E8) did not affect the relative stabilities of ss and dsDNA. To summarize, all experiments suggest that the netinteraction betweenDNAand nonionic surfactants of the EO type is weakly repulsive, which can be attributedmainly tosteric effects. In general, the results were practically identical for the ds and ss forms ofDNA, except those from the cloudpoint experiments, where the decrease of the cloud point was less pronounced with ssDNA. This finding indicates thepresence of an attractive component in the interaction, which can reasonably be ascribed to hydrophobic effects.

Introduction

The fine details in the highly ordered structure of the familiarDNA double helix are directed and stabilized by hydrogen bondsand stacking of the flat bases. On a somewhat coarser level, how-ever, the formation of double-stranded DNA (dsDNA) from thecomplementary single strands (ssDNA) is largely controlled by thebalance of hydrophobic interactions between the bases, which pro-mote assembly, and electrostatic repulsions between the phosphategroups, which counteract it.1 The sometimes neglected amphiphilicnature of DNA is manifested in a number of parallels to the beha-vior of more conventional amphipihiles, such as different aspects ofthe assembly of water-soluble surfactants into micelles. For in-stance, the stability of dsDNA in aqueous solution increaseswith anincreasing ionic strength; in fact, if DNA is present in very low con-centrations in pure water, the double helix spontaneously disas-sembles into ssDNA.2 Other similarities include a pronounced co-operativity in the double helix formation, which results in a narrowtemperature range of coexistence of the ss and ds forms in denatu-ration experiments,1 and the disruption of dsDNAwhen it is trans-ferred from an aqueous solution to a nonaqueous polar solvent.3

The phosphate groups present on each of the nucleotidesrender DNA a high negative charge. This can, in turn, induce astrong attractive interaction with oppositely charged molecules,which often leads to associative phase separation.4 Complexes

formed by DNA and cationic amphiphiles have been extensivelystudied. One of the main incentives for understanding suchsystems is the widespread use of cationic amphiphiles in genedelivery formulations,5,6 but theyhave alsobeenused in a range ofother applications, for instance, in extraction and purification ofDNA,7,8 renaturation and ligation of complementary strands,9

and separation, by precipitation fractionation, of native anddenatured DNA.2

Most work onDNA-cationic amphiphile coassembly has beenperformed with dsDNA. However, there are a number of investi-gations that allow comparisonbetween the behaviors of the ss andds forms of DNA, and clear differences can indeed be observed.Complexes formed by dsDNA and cetyltrimethylammoniumbromide (CTAB) reveal a 2D structure, with elongated micellesand practically straight DNA chains arranged in a hexagonalorder,10,11 whereas the corresponding complexes with ssDNAadopt a micellar cubic structure with the surfactant aggregatesdecorated by the polymer.11 The difference in structure of therespective complexes can largely be attributed to a higher con-formational flexibility of ssDNA, which has a persistence lengthin the range 1-6 nm,12-14 as compared to the relatively stiff

*To whom correspondence should be addressed. (A.H.E.M.) E-mail:alexandra.machado@fkem1.lu.se. (D.L.) E-mail: dlundberg@qui.uc.pt.(1) Evans, D. F.; Wennerstr€om, H. The Colloidal Domain Where Physics,

Chemistry, Biology, and Technology Meet, 2nd ed.; Wiley-VCH: New York, 1999.(2) Rosa, M.; Dias, R.; Miguel, M. G.; Lindman, B. Biomacromolecules 2005, 6,

2164–2171.(3) Cui, S.; Yu, J.; K€uhner, F.; Schulten, K.; Gaub, H. E. J. Am. Chem. Soc.

2007, 129, 14710–14716.(4) Dias, R.; Mel’nikov, S.; Lindman, B.; Miguel, M. G. Langmuir 2000, 16,

9577–9583.

(5) Lasic, D. D.; Templeton, N. S. Adv. Drug Delivery Rev. 1996, 20, 221–266.(6) Wasungu, L.; Hoekstra, D. J. Controlled Release 2006, 116, 255–264.(7) Trewavas, A. Anal. Biochem. 1967, 21, 324–329.(8) McLoughlin, D. M.; O’Brien, J.; McManus, J. J.; Gorelov, A. V.; Dawson,

K. A. Bioseparation 2000, 9, 307–313.(9) Pontius, B. W.; Berg, P. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 8237–8241.(10) Leal, C.; Wads€o, L.; Olofsson, G.; Miguel, M.; Wennerstr€om, H. J. Phys.

Chem. B 2004, 108, 3044–3050.(11) Zhou, S.; Liang, D.; Burger, C.; Yeh, F.; Chu, B. Biomacromolecules 2004,

5, 1256–1261.(12) Smith, S. B.; Cui, Y.; Bustamante, C. Science 1996, 271, 795–799.(13) Tinland, B.; Pluen, A.; Sturm, J.; Weill, G.Macromolecules 1997, 30, 5763–

5765.(14) Desruisseaux, C.; Long, D.; Drouin, G.; Slater, G. W. Macromolecules

2000, 34, 44–52.

DOI: 10.1021/la101403j 13103Langmuir 2010, 26(16), 13102–13109

Machado et al. Article

dsDNA, with a persistence length of about 50 nm at low ionicstrengths.15 There are also a number of findings that indicate astronger attractive interaction between cationic surfactants andssDNA than with dsDNA. For instance, denatured DNA hasbeen shown to precipitate with lower concentrations of dodecyl-trimethylammonium bromide (DTAB) than the correspondingnative DNA.2 Furthermore, the addition of cationic surfactantsto cross-linked DNA gels reveals a more pronounced deswellingand collapse in the case of the denatured form of DNA.16 Inrelated studies on DNA gel particles prepared with CTAB,assessment of DNA release kinetics shows faster release fromparticles containing dsDNA as compared to the case with gelsprepared from ssDNA.17,18 Single-stranded DNA has lowerlinear charge density than dsDNA. The linear charge density ofdsDNA is ∼0.59 negative charges/A,19 whereas it is roughly halffor ssDNA, presuming that the length remains approximatelyunchanged, which means that the electrostatic attraction tocationic surfactant aggregates is expected to be lower in theformer case than in the latter. This suggests that a strongerattractive interaction in the case of ssDNA can be attributed toa combination of the higher flexibility20 and a larger contributionfrom hydrophobic interactions.

Very few studies have been dedicated to the study of interac-tions between the double- or single-stranded forms of DNA andnonionic surfactants, where strong electrostatic interactions areabsent. In general, the interactions between polyelectrolytes andnonionic surfactants are weak. However, if the polyelectrolytecarries hydrophobic domains or groups, rather strong interac-tions between the components can be induced, which can, in turn,have a dramatic influence on the behavior of the system.21,22 Forinstance, a water-soluble polymer grafted with a small fraction ofhydrophobic side chains, often referred to as a hydrophobicallymodified polymer (HMP), can induce a strong increase in theviscosity of a polymer-surfactant solution, as compared to thecase with the unmodified polymer. This is due to a transientphysical cross-linking of the surfactant micelles by the amphiphi-lic polymer.23 The strong adsorption of these polymers to non-ionic surfactants is also reflected in its incorporation in lamellar24

ormicroemulsionphases25 formed in systems containingnonionicsurfactant and by an increase in the cloud point.24 Thus, aninvestigation of the effects of ss and dsDNA on the behavior ofnonionic surfactants can provide information on the importanceof hydrophobic contributions to the interactions of DNA withamphiphiles.

An assessment of the extent of hydrophobic interactionsbetween DNA and amphiphiles is of interest from both technicaland fundamental points of view. It has already been mentionedabove that formulations of DNA containing amphiphiles are ofgreat importance for pharmaceutical applications. Furthermore,

the cell nucleus contains a significant fraction of polar lipids.26 Thepossible functions, in addition to their role as structural compo-nents, of the largely net-uncharged nuclear lipids are largelyunknown. It is found, however, that a fraction of the nuclearlipids are colocalizedwith the chromatin and potential indicationson strong interactions between DNA and nonionic amphiphilescan be taken to suggest a significance of direct interplay betweenDNA and lipids. In a larger perspective, findings on systemscontaining amphiphiles can, to a certain extent, be extrapolated tothe importance of hydrophobic interactions between DNA andother entities carrying hydrophobic domains.

In this study, the interactions between double-stranded orsingle-stranded DNA and nonionic surfactants with ethyleneoxide (EO) headgroups were investigated by utilizing establishedmethods for identifying polymer-surfactant interactions.19,23

The effects of the presence of ss or dsDNA on the behavior ofthe surfactant in aqueous solutions were assessed by determiningchanges in the cloud point of the surfactant, the partitioning ofDNA between the surfactant-rich and surfactant-poor phasesabove the cloud point in samples of different compositions, andthe dependence of the surface tension on surfactant concentra-tion in the presence of DNA. Furthermore, the influence of thepresence of surfactant on the melting temperature of DNA wasinvestigated.

Experimental Section

Materials. Deoxyribonucleic acid (DNA) sodium salt fromsalmon testes and Trizma (Tris) base (reagent-grade) were pur-chased from Sigma (USA). Pentaethylene glycol monododecylether (C12E5) and octaethylene glycol monododecyl ether (C12E8)were obtained fromNikko Chemicals (Japan); all surfactants wereused as received. Seakem LE agarose and Gelstar nucleic acid gelstain were supplied from Cambrex (USA) and Tris acetate-EDTA(TAE) buffer 10� concentrate from Sigma (USA). 6� orangeloading dye solution and ZipRuler express DNA ladder were kindgifts by Fermentas (Sweden). All solutions were prepared usingwater purified with Millipore Milli-Q equipment.

Preparation of DNA Solutions and Characterization of

the DNA. DNA stock solutions were extensively dialyzed(typically for 2 days) against water using dialysis tubing with amolecular weight cutoff of 6000-8000 (Spectrum, USA). The pHof the DNA solutions was adjusted to 7-8, by the addition ofNaOH,2 before dilution with water and/or mixing with a Trisbuffer concentrate. The pH of the buffer was adjusted, by theaddition of HCl, to be 8 in the final solution in 2 mMTris buffer.As will be further discussed in the Results andDiscussion section,the sizeof theDNAwasdeterminedbyagarose gel electrophoresisto be approximately 10000 base pairs (bp). DNA concentrationswere determined by measuring the absorbance at 260 nm, con-sidering that the molar extinction coefficients are 6600M-1.cm-1

for double-stranded DNA and 8700 M-1.cm-1 for single-stranded DNA.27 The average molecular weight of a nucleotidephosphate residuewas considered to be 330 g/mol. The ratioA260/A280 was determined to be higher than 1.8, which suggests that theDNA is essentially free from protein contamination.27

The solutions of ssDNA were prepared using a previouslydescribed procedure, where solutions of dsDNAwere first heatedat 90-95 �C for 15 min and then rapidly cooled on ice for30 min.28 The fast cooling limits renaturation of the DNA.Whenpreparing the ssDNA, the solutions of dsDNAweredilutedto the desired final concentrations prior to denaturation. Anefficient conversion to ssDNA was confirmed by an increase of

(15) Lu, Y.; Weers, B.; Stellwagen, N. C. Biopolymers 2002, 61, 261–275.(16) Costa, D.; Miguel, M. G.; Lindman, B. J. Phys. Chem. B 2007, 111, 10886–

10896.(17) Mor�an, M. C.; Miguel, M. G.; Lindman, B. Biomacromolecules 2007, 8,

3886–3892.(18) Mor�an, M. C.; Miguel, M. G.; Lindman, B. Langmuir 2007, 23, 6478–6481.(19) Dias, R.; Lindman, B. DNA Interactions with Polymers and Surfactants;

John Wiley & Sons: Hoboken, 2008.(20) Wallin, T.; Linse, P. Langmuir 1996, 12, 305–314.(21) Brackman, J. C.; Engberts, J. B. F. N. Chem. Soc. Rev. 1993, 22, 85–92.(22) Saito, S.; Anghel, D. F. In Polymer-surfactant systems; Kwak, J. C. T., Ed.;

Marcel Dekker: New York, 1998; pp 357-408.(23) Holmberg, K.; J€onsson, B.; Kronberg, B.; Lindman, B. Surfactants and

Polymers in Aqueous Solution, 2nd ed.; John Wiley & Sons: Chichester, 2003.(24) Iliopoulos, I.; Olsson, U. J. Phys. Chem. 1994, 98, 1500–1505.(25) Kabalnov, A.; Olsson, U.; Thuresson, K.; Wennerstr€om, H. Langmuir

1994, 10, 4509–4513.(26) Ledeen, R. W.; Wu, G. J. Lipid Res. 2004, 45, 1–8.

(27) Sambrook, J.; Russell, D. W. Molecular cloning: a laboratory manual, 3rded.; Cold Spring Harbor Laboratory Press: New York, 2001; Vol. 1.

(28) Yang, A. Y.; Rawle, R. J.; Selassie, C. R. D.; Johal, M. S. Biomacromo-lecules 2008, 9, 3416–3421.

13104 DOI: 10.1021/la101403j Langmuir 2010, 26(16), 13102–13109

Article Machado et al.

approximately 30% in the absorbance at 260 nm, as compared tothe value for the original dsDNA solution.27 Since thermallydenatured DNA will gradually renature (although slowly), allssDNA solutions were used shortly after preparation.

Gel Electrophoresis. Agarose gels (1%) were prepared in1�TAE buffer and stained with 2.5 μL of the dye Gelstar, whichcan detect both conformational forms of DNA with high sensi-tivity. This dye confers green and orange fluorescence to ds andssDNA, respectively. DNA-containing samples were diluted with2 mM Tris buffer to a concentration in the range 20-25 μg/mLand mixed with 2.5 μL of 6� loading dye. Fifteen microlitersof sample were then added to each well. A DNA ladder (100 to20000 bp) was used for size reference. Electrophoresis wasperformed at 90 V, using 1�TAE as running buffer. Whencompleted, the gel was placed in a detector (Dark Reader, ClareChemicals, USA) that allows visualization of the DNA bands byirradiation with ultraviolet light.

Cloud Point Determination.The cloud points ofC12E5 in theabsence or presence of ds or ssDNAwere determined by measur-ing the variation in turbidity at 530 nm, as the temperature wasincreased from 20 to 60 �C at a rate of 0.5 �C/min. The experi-ments were performed on a Cary 300 Bio UV-visible spectro-photometer (Varian, USA) equipped with a Peltier device fortemperature control, and the sample was placed in a 1 cm cuvette.

In a typical result from a cloud point experiment, one cangenerally observe an initial gradual increase in turbidity, whichcan be attributed to critical fluctuations when approaching phaseseparation, followed by a very steep increase in the slope as phaseseparationoccurs. The cloud point is identified as the temperatureat the point of intersection of the two linear portions of the plot(exemplified in Figure S1, Supporting Information).

Partitioning of DNA between the Surfactant-Poor and

Surfactant-Rich Phases above the Cloud Point. Samplescontaining ds or ssDNA and 2.5 or 10 wt % C12E5 were placedin water baths at 34.5 and 38.3 �C for the low and high surfactantconcentrations, respectively. Separation of the two phases in therespective samples was slow, which can be explained by smalldensity differences and quite low surface tensions between thephases. Thus, the samples were left to equilibrate in the waterbaths for up to one month to obtain visually clear phases. Afterseparation, the upper and lower phases of each sample werecollected and diluted with 2 mM Tris buffer whereafter theamount of phosphorus in the respective samples was determinedby inductively coupled plasma mass spectrometry (ICP-MS)(performed in the Ecology Department at Lund University,Sweden). The amount of DNA was calculated from the amountof phosphorus using eq 1

DNA mass ¼ av mol wt nucleotide

mol wt P�mass of P ð1Þ

The calculations were performed using an average molecularweight per nucleotide of 330.29

Surface Tension Measurements. Surfactant solutions withor without ds or ssDNA were vigorously stirred to ensurecomplete mixing and then transferred to Pyrex cells for equilibra-tion. Surface tension measurements were performed using a duNo€uy ring tensiometer (Kr€uss,Germany).Allmeasurementswereperformed at 25 �C and in triplicate.

Influence of Surfactants on DNA Melting Temperature.DNAmelting transitions in the absence or presence ofC12E8wereinvestigated by monitoring the changes in the UV absorptionat 260 nm as the temperature was gradually increased from 25 to90 �C at a constant rate of 0.5 �C/min. The experiments wereperformed on a Cary 300 Bio UV-visible spectrophotometer(Varian, USA) equipped with a Peltier device for temperaturecontrol. The melting point was identified as the temperature

corresponding to the midpoint of the slope in the melting curve(Supporting Information, Figure S2).

Results and Discussion

Choice of Materials. In this study, the experiments wereperformed with nonionic surfactants carrying ethylene oxide(EO) polar headgroups. These surfactants, usually designatedas CmEn, with m indicating the number of carbon atoms in thealkyl chain and n the number of EO units in the headgroup, showa characteristic temperature dependence of their physicochemicalproperties. With increasing temperature, the EO groups are lesshydrated and the surfactant becomes more hydrophobic.23 Thisleads to lowered aqueous solubility at higher temperature. At theso-called cloud point temperature, there is a transition from ahomogeneous micellar solution into two isotropic phases, onesurfactant-rich and one surfactant-poor, which is manifested byan increased turbidity of the solution.23

The cloud point is often rather sensitive to the presence ofcosolutes in the system, which can, by different mechanisms,cause increases as well as decreases in its value. For instance, theaforementioned HMPs, which associate to the surfactant aggre-gates, and thus stabilize the homogeneous micellar solution, caninduce a considerable increase in the cloud point.24A solute that isdepleted from the micellar pseudophase, on the other hand, willpromote phase separation and thus cause a decrease of the cloudpoint. Thus, an investigation of the effects of a certain cosolute onthe cloud point of an EO surfactant provides a sensitive toolfor identifying attractive or repulsive interactions between thecomponents.

C12E5 was chosen as the main surfactant in the investigations,since its behavior in aqueous systems is very well characterized andit shows a cloud point in a suitable temperature range, e.g., 31.9 �Cat a concentration of 1 wt % in pure water.30 Literature values ofthe critical micelle concentration (CMC) of this surfactant are64-65 μM, at 25 �C.23,31 In most experiments, the surfactantconcentration was significantly higher than the CMC, implyingthat a major fraction of the surfactant resides in micelles.

Since the cloud point of C12E5 is below the expected range ofDNA denaturation temperatures and therefore can overlap withthe melting curve of the DNA, another surfactant of the sameclass, C12E8, was used in the DNA melting experiments. TheCMC of C12E8 is 71 μM, at 25 �C.23

Ideally, to simplify interpretation of the results, one would likethe DNA used to be as well-defined and close to monodisperse aspossible. However, considering the amount of DNA required inthe herein discussed experiments this option was not reasonable.Therefore, a quality of DNA from salmon sperm, which has beenpreviously used in a number of related studies, was chosen for thiswork. In order to remove salts, small fragments and low mole-cularweight impurities thatmaybepresent, theDNAwas purifiedby dialysis.

As was mentioned in the introduction, dsDNA is inherentlyunstable at very low concentrations. The limiting concentrationfor dsDNA stability without added salt has been reported to be0.56 mM for DNA from herring sperm.2 Furthermore, dissolu-tion ofCO2 from the air might decrease the pHand thus affect theDNA protonation state.32 Thus, to ensure integrity of the DNAdouble helix and stability of the pH, most experiments were

(29) M€ulhardt, C. Molecular Biology and Genomics; Elsevier Academic Press:Burlington, 2007.

(30) Strey, R.; Schom€acker, R.; Roux, D.; Nallet, F.; Olsson, U. J. Chem. Soc.Faraday Trans. 1990, 86, 2253–2261.

(31) Rosen, M. J.; Cohen, A. W.; Dahanayake, M.; Hua, X. Y. J. Phys. Chem.1982, 86, 541–545.

(32) Korolev, N. I.; Vlasov, A. P.; Kuznetsov, I. A. Biopolymers 1994, 34, 1275–1290.

DOI: 10.1021/la101403j 13105Langmuir 2010, 26(16), 13102–13109

Machado et al. Article

performed in the presence of a Tris buffer of pH 8. On the otherhand, the presence of salts in the system can affect the behavior,notably the cloud point, and it can reduce or eliminate effectsrelating to the osmotic pressure. Thus, to minimize its effects onthe physicochemical behavior, the buffer concentration was keptas low as possible, specifically at 2 mM. In addition, certainexperiments were performed both in buffer and in pure water inorder to assess possible differences.Characterization of DNA andVerification of DNAStorage

Stability. Before the interactions between DNA and nonionicsurfactants were investigated, the DNA used was characterizedwith respect to its size, and it was verified that its conformationaland/or storage stability was not affected by the presence of sur-factant. These parameters were assessed by means of gel electro-phoresis, which allows separation of molecules according to theirsize and assessment of changes in the DNA conformation.

As is shown in Figure 1, the native DNA, which, as expected,causes the dye to fluoresce green, is large and shows a notablepolydispersity, with a mean size of approximately 10 000 bp (lane2). The denatured DNA, however, which gives orange fluores-cence, shows a higher mobility and appears to be more poly-disperse (lane 3). The higher mobility of ssDNA as compared tothat of dsDNA can be ascribed to its higher flexibility. The extentof DNA degradation in the used temperature and time ranges isexpected to be of minor influence.33 The apparent higher poly-dispersity can likely be attributed to the formation of a wide rangeof secondary structures within and between the ssDNA strands,by fractional intra- or intermolecular base pairing.34 This effectcan reasonably be explained by similar arguments as the smearingassociated with the transitory trapping of circular DNA, ascompared to the band arising from linearDNAof the same size.35

As is also shown in Figure 1, the results on solutions ofpure ds and ssDNA were compared with those obtained with

corresponding solutions containing 1 wt % of C12E5. It can beseen that there are no significant differences in the bands obtainedin the absence or presence of surfactant for dsDNA(lanes 2 and 4)and for ssDNA (lanes 3 and 5) and that the characteristicfluorescence of the bands is conserved in the presence of surfac-tant. These observations suggest that C12E5 does not notablyaffect the DNA conformation or stability at the conditions used.

Moreover, DNA solutions with or without surfactant indifferent concentrations were shown to be stable for at least2 months. The bands in the agarose gel remain practicallyunchanged, showing that no considerable degradation of theDNA occurs with time.Effect of Double-Stranded and Single-Stranded DNA on

the Cloud Point of C12E5.Asmentioned above, an investigationof possible changes in the cloud point of an oxyethylene surfactanton additionof a cosolute is a useful approach for identifying, and tosome extent classifying, interactions between the components.

The cloud point is usually detected by visual inspection of thesurfactant solution as the temperature is increased, and identifiedas the temperature where the sample turns turbid. In our study,this parameter was determined by measuring turbidity in aUV-visible spectrophotometer, which allows a better controlof the temperature gradient and is expected to give a moreaccurate value of the cloud point.

Figure 2 shows the dependence of the cloud point of a solutionof 1 wt % C12E5 (24.6 mM) in the presence of differentconcentrations of ds or ssDNA in either 2 mM Tris buffer orpure water. The values obtained for the cloud point of 1 wt %C12E5 in the absence of DNA are in accordance with literaturevalues.30,36,37 The value obtained in buffer is slightly lower thanthe one in water (32.8 and 32.9 �C in 2 mMTris buffer and water,respectively). It can be seen that both forms of DNA have only asmall effect on the surfactant cloud point under the conditionsused. When compared to what is generally found for highlycharged polyelectrolytes, these findings are not unexpected, atleast for dsDNA.23 However, for ssDNA, with its more dis-tinct hydrophobic domains, the very limited effect on the cloudpoint is in significant contrast to what is found in corresponding

Figure 1. Electrophoresis gel showing bands for dsDNA in theabsence (lane 2) or presence (lane 4) of 1wt%C12E5, ssDNA in theabsence (lane 3) or presence (lane 5) of the same surfactant, and areference ladder (lane 1).

Figure 2. Variation in the cloud point of 1 wt % C12E5 in 2 mMTris buffer (diamonds) or inwater (squares)with the concentrationof double-stranded (solid symbols) and single-stranded DNA(open symbols). The DNA concentration is expressed in terms ofphophate groups. The experimental error was estimated to bewithin(0.2 �C.

(33) Freifelder, D.; Dewitt, R. Gene 1977, 1, 385–387.(34) Dimitrov, R. A.; Zuker, M. Biophys. J. 2004, 87, 215–226.(35) Akerman, B.; Cole, K. D. Electrophoresis 2002, 23, 2549–2561.

(36) Feitosa, E.; Brown, W.; Hansson, P.Macromolecules 1996, 29, 2169–2178.(37) Fonseca, S. M.; Eus�ebio, M. E.; Castro, R.; Burrows, H. D.; Tapia, M. J.;

Olsson, U. J. Colloid Interface Sci. 2007, 315, 805–809.

13106 DOI: 10.1021/la101403j Langmuir 2010, 26(16), 13102–13109

Article Machado et al.

mixtures with HMPs. For instance, hydrophobically modifiedpoly(sodium acrylate) (HMPA) associates strongly with C12E5

aggregates and induces a considerable increase of the cloud point;at a concentration of 0.1% of a HMPA with 3 mol % of C18chains randomly grafted to its backbone, an increase of the cloudpoint of approximately 5 �Cwas observed in a 2 wt% solution ofC12E5 in D2O.24

The small effect of ssDNA on the cloud point of C12E5 incomparison to that ofHMPs inevitably raises questions about theeffective hydrophobicity of ssDNA. To explain the difference oneneeds to take a closer look at the structures of the respectivepolymers. In the case of a typical HMP, hydrophobic chains aregrafted to the hydrophilic polymer backbone in a way that leavesthem highly exposed to the surrounding and available to interactwith the surfactant aggregates. Their hydrophobic moieties can,therefore, be efficiently anchored in the micelles. The single-stranded DNA, on the other hand, has a very different structure.Each nucleotide has a polar end at the phosphate group, whereasthe base constitutes a more hydrophobic domain. When thenucleotides are connected into a polymer, however, this cannotnecessarily be considered as a ribbon with one hydrophobic andone hydrophilic side. In fact, ssDNA is known to retainmuch of ahelical structure due to base stacking interactions,38-40 whichlimits the effective exposure of its hydrophobic parts. Thisconformation could potentially be disrupted if the interactionwith the surfactant was strong enough, but this does not seem tobe the case in the herein studied system. In addition to the hiddenhydrophobic groups, there is a steric effect associated with thebulky and highly hydrated headgroups of the surfactant, whichcan likely prevent the hydrophobic parts of the DNA fromreaching the hydrophobic micellar core. It is probable that acombination of the factors discussed above impair the hydro-phobic interactions between the ssDNA and the surfactant.

Although the effect of ssDNAon the cloud point of C12E5 is, incomparison to the effects observed with typical HMP, practicallynegligible, some small but reproducible changes in the cloud pointare observed and these differ somewhat between the differentconditions studied. For all conditions used, there was first a slightdecrease of the cloud point followed by an increase in its valuewith increasingDNA concentration. The weak minimum is likelya consequence of a subtle balance between several factors, and itsorigin was not further investigated. More importantly, there is ingeneral a slight decrease of the cloud point in the presence ofDNA, which can be attributed to a repulsive interaction betweenthe surfactant and the polymer. This repulsion can reasonably beascribed to steric interactions between the micelles and the DNA;the repulsion isweak, but strong enough to overshadow the loss inentropy expected on confinement of theDNAand its counterionsin a smaller volume.

Interestingly, one can note that, although the general trend forboth ss and dsDNA is similar, the denatured DNA consistentlygives slightly higher cloud point values than the native one in thewhole investigated range of DNA concentration and in bothmedia used. This observation can be taken to suggest that theremight actually be a small contribution from hydrophobic attrac-tion between ssDNA and the surfactant that partially counter-balances the steric repulsion. However, in the direct quantitativecomparison between ss and dsDNA one should also recall that as

the DNA concentration is expressed in terms of bases, theconcentration in terms of polymer chains is twice as high inssDNA compared to dsDNA, which may affect, for example, thesteric excluded volume interactions.

In order to assess the contribution from electrostatics to theresults obtained, the cloud point of C12E5 in the absence orpresence of DNA was also determined in 0.5 M NaCl, where theelectrostatic interactions between the DNA strands will beeffectively screened. As expected, the cloud point of the surfactantdecreased, to a value of 27.6 �C, in the salt solution, due to thesalting-out effect of Cl-.41 In the presence of dsDNA (1.5 mM),the cloud point decreased even further to 22.6 �C. These findingscan be explained by the entropic penalty fromconfiningDNA inaseparate phase being substantially reduced in the presence ofadded salt. Thus, the increase in ionic strength promotes phaseseparation. The slight decrease in the cloud point values in Trisbuffer as compared to the values in water can likely be ascribed tothe same mechanism.DNA Partitioning between Surfactant-Rich and Surfac-

tant-Poor Phases Formed above the Cloud Point. As dis-cussed in the previous section, the general decrease in the cloudpoint of the surfactant on the addition of ds or ssDNA suggests aweak net repulsion between the components, although there areindications of an attractive component in the case of ssDNA. Togain further understanding of the nature and relative importanceof these components to the interaction between DNA and C12E5,we chose to assess the partitioning of DNA between the twophases formed above the cloud point.

The partition coefficient is a sensitive parameter that is usefulfor investigating the balance between repulsive and attractiveinteractions between a polymer and a surfactant. Substances thatassociate with the surfactant aggregates, such as the previouslymentioned HMP,42 are expected to show a preference for thesurfactant-rich phase, whereas large molecules with no attractiveinteraction with the micelles or substances that are depleted fromthe micellar surface generally remain in the less crowded surfac-tant-poor phase. Partitioning between the two phases above thecloudpoint has been successfully utilized to separate biomoleculesaccording to hydrophobicity and size.43,44

In order to obtain similar volumes of the surfactant-rich and thesurfactant-poor phases, the samples were prepared with highersurfactant concentrations than those used in the cloud pointexperiments, specifically 2.5 and 10 wt % (which correspond to61 and 246 mM). Samples with these surfactant concentrationswere left to phase-separate at 34.5 and 38.3 �C, respectively.30 Theuse of two different C12E5 concentrations aimed to assess thepossible dependence of partitioning on surfactant concentration.

The presence of the surfactant was found to interfere with the260 nmUV absorbance from DNA, likely due to scattering fromthe surfactant aggregates. To obtain accurate values of the DNAconcentration, this was thus instead determined by elementalanalysis of phosphorus, using ICP-MS.

The partition coefficient of the DNA between the two phases(K) was calculated according to eq 2

K ¼ ½DNA�t½DNA�b

ð2Þ

(38) Arnott, S.; Chandrasekaran, R.; Leslie, A. G. W. J. Mol. Biol. 1976, 106,735–748.(39) Camerman, N.; Fawcett, J. K.; Camerman, A. J.Mol. Biol. 1976, 107, 601–

621.(40) Vesnaver, G.; Breslauer, K. J. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 3569–

3573.

(41) Kabalnov, A.; Olsson, U.;Wennerstr€om, H. J. Phys. Chem. 1995, 99, 6220–6230.

(42) Zhao, G.; Chen, S. B. Langmuir 2006, 22, 9129–9134.(43) Hinze, W. L.; Pramauro, E. Crit. Rev. Anal. Chem. 1993, 24, 133–177.(44) Liu, C. L.; Nikas, Y. J.; Blankschtein, D. Biotechnol. Bioeng. 1996, 52, 185–

192.

DOI: 10.1021/la101403j 13107Langmuir 2010, 26(16), 13102–13109

Machado et al. Article

where [DNA]t and [DNA]b denote the concentrations of DNA inthe top, surfactant-rich, and in the bottom, surfactant-poorphases, respectively. The coefficients were determined in 2 mMTris buffer as well as in pure water, and the results are shown inTable 1.

For all conditions tested, the partition coefficients obtainedwere clearly below 1, indicating that DNA is preferably located inthe bottom, surfactant-poor phase. The coefficient was lowerwhen the surfactant concentration was higher. These resultssuggest that the repulsive component of the interaction betweenDNA and C12E5 micelles is dominant in all cases. The absence ofsignificant differences between ds and ssDNA show that, eventhough there might be a small hydrophobic interaction in the caseof ssDNA, this is overshadowed by the repulsive interactions. Itcan be noted that therewere no significant differences between theresults obtained in water or in Tris buffer.

The net repulsive character of the interaction may be ascribedto excluded volume in the top phase by the large volume fractionof surfactant aggregates and to the steric effect from the largesurfactant headgroups. These findings are consistent with thosepreviously obtained for samples containing DNA and 7.8 wt %Triton X-114 (which carries 9-10 EO units as its headgroup) inphosphate-buffered saline (PBS),45 where the clear preference ofDNA for the surfactant-poor phase was explained by repulsive,steric, excluded volume interactions.

In Table 1, it can also be noted that the presence ofDNA in thesystem induced a shift in the relative volumes of the phases, withthe volume of the bottom, surfactant-poor phase, increasing atthe expense of the top, surfactant-rich phase, for all conditionstested. The deviation is roughly of the same order for bothinvestigated surfactant concentrations. Consequently, the finalsurfactant concentrations in the top phases are higher in thepresence of DNA than in the binary samples. Presuming that thebottomphase is free from surfactant after separation, a surfactantconcentration of approximately 9 wt % in the top phase wasobtained for the samples containing in total 2.5wt%of surfactant(at 34.5 �C), whereas the corresponding value was 27wt% for thesamples with a total surfactant concentration of 10 wt % (at38.3 �C). These values should be compared to those reported forthe binary system of C12E5-H2O by Strey et al.,30 where finalsurfactant concentrations of 7 and 16 wt % in the top phase arefound for the corresponding samples kept at 34.5 and 38.3 �C,respectively. The differences in relative volumes in sampleswith orwithout DNA can be explained by the increase of the osmoticpressure exerted by the DNA in the surfactant-poor phase. Theincrease in the volume of the bottom phase is more pronounced

when the partitioning coefficient is low, asmoreDNA is confinedin the bottom phase.

The fact that DNA is excluded from the phase concentrated insurfactant is consistent with results from previous studies performedwith EO-based surfactants and polymers. Using the method ofsingle-molecule observation with fluorescence microscopy, it wasobserved that athigh concentrationsofTritonX-100 (50-90%)46orpoly(ethylene glycol) (PEG, 230 mg/mL)47 compaction of DNAmacromolecules through a discrete coil-globule transition could beinduced. This finding was explained by the increase in osmoticpressure in the surfactant solution and depletion interactions.

Since the DNA used in these experiments is polydisperse, it isconceivable that there is an uneven distribution of DNA sizesbetween the two phases. Particularly, one might imagine thatsmaller fragments could be accumulated in the top phase. Thispossible effect was assessed for samples prepared with 2 mMTrisbuffer by performing agarose gel electrophoresis.

The DNA size distributions in the separated surfactant-richand surfactant-poor phases from the samples prepared in 2 mMTris buffer are shown in Figure 3. It can be seen that there are nonotable differences in the sizes of the DNA between the top andbottom phases of each sample, with neither ds nor ssDNA. Thus,the distribution of DNA between the two phases is not signifi-cantly influenced by a molecular size selectivity. The slightbroadening of the bands of the separated samples relative to thestarting DNA solutions might be ascribed to a minor extent ofdegradation due to storage at elevated temperatures.Surface Tension Measurements in the Presence of ds or

ssDNA. The presence of polymers in aqueous surfactant solu-tions might influence the aggregation into micelles. If the inter-action between the surfactant and the polymer is attractive, thepolymer can induce surfactant micellization at concentrationssignificantly lower than the critical micelle concentration (CMC)in a binary solution.23,48,49 In addition, a noncooperative bindingstarting at very low surfactant concentrations can also occur, as,for instance, in the presence of HMP.50-52

A convenient way to assess the effect of a polymer on the CMCof a surfactant is to determine the dependence of the surfacetension on surfactant concentration, as the formation of aggre-gates in the bulk is reflected by changes in the surface tension.

Table 1. Partitioning of DNA in the Surfactant-Rich and Surfactant-Poor Phases above the Cloud Pointa

sample medium temperature (�C) total conc. C12E5 (wt %) partition coefficientb,c,d volume ratioe,f

dsDNA Tris buffer 34.5 2.5 0.62( 0.32 0.4ssDNA Tris buffer 34.5 2.5 0.54( 0.22 0.4dsDNA Tris buffer 38.3 10 0.09( 0.07 0.6ssDNA Tris buffer 38.3 10 0.23( 0.09 0.6dsDNA water 34.5 2.5 0.46( 0.18 0.4ssDNA water 34.5 2.5 0.64( 0.00 0.4dsDNA water 38.3 10 0.20( 0.11 0.6ssDNA water 38.3 10 0.12( 0.12 0.6

aThe totalDNAconcentrationwas 2.8mM (in terms of phosphate groups) in all samples. bThe partition coefficient is defined as the ratio of theDNAconcentrations in the top, surfactant-rich phase, and in the bottom, surfactant-poor phase, respectively. cThe results are represented as themean value(standard deviation (n=2). dThe experimental error associatedwith the determination of phosphorus by ICP-MS is within 5%. eThe ratio of the volumeof the top phase to the volume of the bottomphase. fThe experimental error associatedwith the determination of the volume bymeasuring the heightwasestimated to be within 10%.

(45) Mashayekhi, F.; Meyer, A. S.; Shiigi, S. A.; Nguyen, V.; Kamei, D. T.Biotechnol. Bioeng. 2009, 102, 1613–1623.

(46) Mel’nikov, S. M.; Yoshikawa, K. Biochem. Biophys. Res. Commun. 1997,230, 514–517.

(47) Kojima, M.; Kubo, K.; Yoshikawa, K. J. Chem. Phys. 2006, 124, 024902.(48) Goddard, E. D. J. Colloid Interface Sci. 2002, 256, 228–235.(49) Taylor, D. J. F.; Thomas, R. K.; Penfold, J.Adv. Colloid Interface Sci. 2007,

132, 69–110.(50) Chang, Y.; Lochhead, R. Y.; McCormick, C. L. Macromolecules 1994, 27,

2145–2150.(51) Johnson, K.M.; Fevola,M. J.; Lochhead, R. Y.;McCormick, C. L. J. Appl.

Polym. Sci. 2004, 92, 658–671.(52) Deo, P.; Somasundaran, P. Langmuir 2005, 21, 3950–3956.

13108 DOI: 10.1021/la101403j Langmuir 2010, 26(16), 13102–13109

Article Machado et al.

Figure 4 shows the surface tension as a function of surfac-tant concentration in the presence of 1 or 5 mM of ds or ssDNA.TheCMCofC12E5 in 2mMTris buffer (pH8) (i.e., in the absenceofDNA) was determined to be 76 μM, which is close to theaforementioned literature values for theCMCofC12E5 inwater.

23,31

The surface tension of the pure DNA solutions in 2 mM Trisbuffer was also assessed. In comparison to the surface tension ofwater (72.2 mN/m) and the Tris buffer (71.2 mN/m), there was a

small decrease of the surface tension values to 69.8 and 69.0mN/mfor 1 and 5 mM dsDNA and 68.0 and 66.8 mN/m for 1 and5 mM ssDNA, respectively. The decrease in surface tension withdsDNA is somehow unexpected, as it has been previously shownthat DNA does not exhibit surface activity and thus does notaffect the surface tension of water.53 The denatured DNA isexpected to be more surface active, due to its more amphiphiliccharacter. Our results indicate that ssDNA does indeed induce adecrease of the surface tension that is slightly larger than thatobserved with dsDNA. However, the observed decrease is verysmall in comparison to that promoted by, e.g., the anionichydrophobically modified poly(maleic acid/octyl vinyl ether)(PMAOVE), which has been shown to decrease the surfacetension values by ∼25%.52 The very small decrease in surfacetension induced by ssDNA can be explained by the same argu-ments discussed in connection to the cloud point experiments, bythe ssDNA retaining a helical structure where the hydrophobicdomains are largely hidden.

It can be seen in Figure 4 that, despite a slight decrease of thesurface tension values for lower C12E5 concentrations in thepresence of DNA, which can be attributed to the lower surfacetension values of the pure DNA solutions, the addition of 1 or5 mMof either ss or dsDNAdid not significantly affect the CMCof the surfactant. Thus, neither of the polymers induced surfac-tant micellization. This behavior differs from, e.g., that reportedby Deo et al. with PMAOVE.52 In mixtures of this polymer withC12E5, clear changes were observed in the surface tension plot dueto the incorporation of C12E5 into the hydrophobic domains ofthe polymer.52

The results from the surface tension experiments are in accor-dancewith the findings discussed in previous sections and theweakinteractionbetween ss or dsDNAwith the surfactantmonomers, asthe one with the surfactant aggregates, can be ascribed to the influ-ence of the large headgroup that prevents the interaction betweenthe surfactant tail and the hydrophobic domains of DNA.Effect of Surfactant on theMeltingTemperature ofDNA.

All the previous experiments were performed to identify theinteractions between DNA and surfactant by assessing the effectsof the presence of ds or ssDNAon the surfactant behavior. In thislast section, the possible influence of the surfactant on DNAstability will be addressed.

The determination of the melting temperature is a straightfor-ward way to assess the relative stability of the different DNAconformations in the presence of cosolutes. It is known thatcationic surfactants can have either a stabilizing or destabilizingeffect on DNA, depending on several factors, for instance, thesurfactant alkyl chain length.2 It is also known that phospholipidscan stabilize or destabilize the double helix under differentconditions.54,55 As a general rule, compounds that are able tostabilize dsDNA are expected to induce an increase of the meltingpoint, whereas the opposite is expected for substances thatstabilize ssDNA.

The influence of surfactant on the DNA melting temperaturewas, as mentioned above, investigated using C12E8, instead ofC12E5. Due to the higher hydrophilicity associated with theadditional ethylene oxide groups, C12E8 has a higher cloud point(∼80 �C), which is above the range of expected DNA meltingtemperatures.

Figure 3. Electrophoresis gel showing the DNA size distributionsin the two phases of separated samples prepared in 2 mM Trisbuffer, for the two surfactant concentrations used, i.e., 2.5 wt %(lanes 4-7) and 10 wt % (lanes 8-11). The gel also shows thedifferences between the two forms of DNA, dsDNA top (lanes4 and 8) and bottom (lanes 5 and 9) phases, and ssDNA top (lanes6 and 10) and bottom (lanes 7 and 11) phases. For reference, resultsfrom a DNA ladder (lane 1) as well as from solution of puredsDNA (lane 2) or pure ssDNA (lane 3) are also shown.

Figure 4. Surface tensionas a functionof surfactant concentrationfor two concentrations, 1 mM (squares) and 5 mM (diamonds) ofnative (closed symbols) and denatured (open symbols) DNA. Thecurve for the pure C12E5 in 2 mM Tris buffer (crosses) is alsoshown. DNA concentration is expressed in terms of phophategroups. The standard deviation is below 2% (n= 3).

(53) McLoughlin, D.; Langevin, D. Colloids Surf., A: Physicochem. Eng.Aspects 2004, 250, 79–87.

(54) Manzoli, F. A.; Muchmore, J. H.; Bonora, B.; Sabioni, A.; Stefoni, S.Biochim. Biophys. Acta 1972, 277, 251–255.

(55) Manzoli, F. A.; Muchmore, J. H.; Bonora, B.; Capitani, S.; Bartoli, S.Biochim. Biophys. Acta 1974, 340, 1–15.

DOI: 10.1021/la101403j 13109Langmuir 2010, 26(16), 13102–13109

Machado et al. Article

Figure 5 shows the melting curve for DNA in the presence of1 or 2 wt % (18.6 and 37.1 mM, respectively) C12E8, as well asreference data for solutions of DNA or C12E8. The results forDNA in the presence of 1 wt % C12E8 are practically identical tothe curve obtained by adding the melting curve for DNA in theabsence of surfactant and the surfactant cloud point curve.Experiments performed with 2 wt % of the same surfactant alsoled to similar results. Additionally, it could be observed that the

presence of DNA did not affect the surfactant cloud point, whichis consistent with the previously discussed experiments. Theseresults give further support to the notion that DNA does notadsorb to nonionic EO surfactant micelles.

Conclusions

All of the results discussed herein show that the net interactionsbetween DNA and nonionic surfactants of the EO type areweak, but effectively repulsive. The weak repulsion can probablybe attributed mainly to a combination of steric effects associatedwith the bulky surfactant headgroup, which can hinder effectivecontact between the DNA and the surfactant tails, and excludedvolume interactions.

The results were practically identical for the ds and ss forms ofDNA, except in the cloud point experiments, where there wereindications on an attractive component to the ss DNA-surfactantinteraction. This can reasonably be ascribed to hydrophobiceffects.

Acknowledgment. The authors thank the Swedish ResearchCouncil (VR) and the Fundac-~ao para a Ciencia e a Tecnologia,Portugal, for their financial support (SFRH/BD/41424/2007,A.H.E.M., and SFRH/BPD/48522/2008, D.L.). Tommy Olssonis acknowledged for performing the ICP-MS experiments. BrunoMedronho and Wei Wang are acknowledged for valuable inputon this work.

Supporting Information Available: Representative plots ofcloud point and DNA melting temperature determinationare included in this section. This material is available free ofcharge via the Internet at http://pubs.acs.org.

Figure 5. Effect of 1 wt%C12E8 (red) and 2 wt%C12E8 (blue) onthe melting temperature of dsDNA (solid line). For reference, theDNA melting curve obtained in the absence of surfactant (black,solid line), the surfactant cloud point curve (dotted line), and thecurve resulting from the addition of the two (dash dotted line) arealso presented.