Langmuir 2004, 20, 6407-6413
6407
Influence of DNA Adsorption and DNA/Cationic Surfactant Coadsorption on the Interaction Forces between Hydrophobic Surfaces Alan D. Braem, Jose´ Campos-Tera´n,* and Bjo¨rn Lindman Department of Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund University, Box 124, SE-22100, Lund, Sweden Received January 13, 2004. In Final Form: May 6, 2004 The forces between hydrophobic surfaces with physisorbed DNA are markedly and irreversibly altered by exposure to DNA/cetyltrimethylammonium bromide (CTAB) mixtures. In this colloidal probe atomic force microscopy study of the interactions between a hydrophobic polystyrene particle and an octadecyltrimethylethoxysilane-modified mica surface in sodium bromide solutions, we measure distinct changes in colloidal forces depending on the existence and state of an adsorbed layer of DNA or CTAB-DNA complexes. For bare hydrophobic surfaces, a monotonically attractive approach curve and very large adhesion are observed. When DNA is adsorbed at low bulk concentrations, a long-range repulsive force dominates the approach, but on retraction some adhesion persists and DNA bridging is clearly observed. When the DNA solution is replaced with a CTAB-DNA mixture at relative low CTAB concentration, the length scale of the repulsive force decreases, the adhesion due to hydrophobic interactions greatly decreases, and bridging events disappear. Finally, when the surface is rinsed with NaBr solution, the length scale of the repulsive interaction increases modestly, and only a very tiny adhesion remains. These pronounced changes in the force behavior are consistent with CTAB-induced DNA compaction accompanied by increased DNA adsorption, both of which are partially irreversible.
Introduction Complexes between DNA and cationic lipids have been receiving much study recently due to their possible application as gene delivery vehicles.2 In these complexes, called lipoplexes, compacted DNA molecules are protected from environmental factors and are thus more efficiently delivered to target cells.3 Lipoplexes do not have as high of a transfection efficiency as viral vectors, but they are less prone to provoke immunological response and they do not represent any infection threat to patients.4 The factors that control DNA compaction upon interaction with lipids, polymers, and multivalent ions have been extensively studied recently.5 For DNA-cationic surfactant systems, the nature of the DNA (length, conformation)6 and the surfactant (headgroup and tail composition)7-9 was shown to influence the structure and phase-separation limits of the DNA-cationic surfactant complexes. In vivo, the lipoplexes will encounter different types of interfaces in the blood stream (walls of blood vessels, blood cells, proteins, and other “colloidal” particles) before being delivered. Interactions with such interfaces are likely to occur and therefore influence the uptake behavior of the * Corresponding author. E-mail:
[email protected]. (1) Ca´rdenas, M.; Braem, A. D.; Nylander, T.; Lindman, B. Langmuir 2003, 19, 7712. (2) Pedroso de Lima, M. C.; Simo˜es, S.; Pires, P.; Faneca, H.; Du¨zgu¨nes, N. Adv. Drug Delivery Rev. 2001, 47, 277. (3) Somia, N.; Verma, I. M. Nature 1997, 389, 239. (4) Lasic, D. D.; Templeton, N. S. Mol. Biotechnol. 1999, 11, 175. (5) Lindman, B.; Mel’nikov, S.; Mel’nikova, Y.; Nylander, T.; Eskilsson, K.; Miguel, M.; Dias, R.; Leal, C. Prog. Colloid Polym. Sci. 2002, 120, 52. (6) Shirahama, K.; Takashima, K.; Takisawa, N. Bull. Chem. Soc. Jpn. 1987, 60, 43. (7) Koltover, I.; Dalditt, T.; Ra¨dler, J. O.; Safinya, C. R. Science 1998, 281, 78. (8) Gorelov, A. B.; McLoughlin, D. M.; Jacquier, J. C.; Dawson, K. A. Nuovo Cimento 1998, 20, 2553. (9) Dias, R.; Mel’nikov, S.; Lindman, B.; Miguel, M. Langmuir 2000, 16, 9577.
complexes. Hence, the interfacial behavior of these complexes may partly determine how efficiently DNA is delivered to the target cells. In a previous paper from our group,1 the interfacial behavior of DNA and DNA/cetyltrimethylammonium bromide (CTAB) mixtures at a hydrophobized silica surface was studied by ellipsometry. That paper contained a detailed examination of the effects of DNA length, DNA conformation (double- or single-stranded), CTAB concentration, order of addition, and irreversibility effects. The current paper narrows the focus to a single length and conformation of DNA (2000 base pair, double-stranded), surfactant concentration, and order of addition. Given an understanding of the adsorbed amount and thickness of adsorbed layers under these conditions, we now turn our attention to the effect the adsorbed layers have on colloidal forces. We consider first bare hydrophobic surfaces in the presence of 10 mM sodium bromide (NaBr) background electrolyte, followed by 0.02 mg/mL DNA in 10 mM NaBr. The key observations from Ca´rdenas et al.1 under these conditions are that a sparse adsorbed layer (0.2 mg/m2) is formed with an ellipsometric thickness of approximately 31 nm. The adsorbed layer is sufficiently negatively charged that when challenged with sodium dodecyl sulfate (SDS) solution, none of the negatively charged surfactant adsorbs. Next, we expose the interface to a mixture with the same DNA and salt concentration as above, with 10-6 M CTAB added. Under these conditions, the ellipsometry revealed a much larger adsorbed amount (2.0 mg/m2) and a much thinner adsorbed layer (3-5 nm). SDS could adsorb to this surface, indicating the layer is positively charged, neutral, or only slightly negatively charged. Finally, we consider surfaces that have been processed as above, followed by rinsing with 10 mM NaBr. The ellipsometry measurements indicated that a relatively large adsorbed amount (1.0 mg/m2) persists, while the thickness relaxes to an intermediate value (9 nm). The layer is again
10.1021/la049882w CCC: $27.50 © 2004 American Chemical Society Published on Web 06/24/2004
6408
Langmuir, Vol. 20, No. 15, 2004
Braem et al.
negative (no SDS adsorbs), indicating that most or all of the CTAB has been removed. The fact that the adsorbed amount and thickness do not revert to either zero or their original values in the presence of DNA alone indicates that the CTAB-processed layer remains kinetically trapped in a nonequilibrium configuration. Our aim in this paper is to determine the effect of the different interfacial configurations described above on colloidal interactions. Where possible, the force curves are used to extract further information on the layer structure. The results we obtain correlate well with the observations from ellipsometry. Experimental Details Materials. Salmon sperm DNA was purchased from Gibco, BRL, and was used as received. As stated by the manufacturer, this DNA is double-stranded and 2000 ( 500 base pairs (bp) in length, as determined by 1% TAE Agarose gel analysis, and free from Dnase and Rnase. Cetyltrimethylammonium bromide (Merck, pa quality) and sodium bromide (Aldrich, extra pure quality) were used as received. Water purified by a Milli-Q system (Millipore Corp., Bedford, MA) was used in all measurements. All DNA and CTAB solutions were prepared in solutions containing 10 mM NaBr. Green muscovite mica was obtained from S & J Trading Inc. (Glen Oaks, NY) and prepared as described below. Hydrophobic polystyrene (PS) particles (uncharged, surfactant stabilized) with a nominal diameter of 6.17 µm were obtained from Bangs Laboratories, Inc., and were attached to AFM cantilevers as described below. Results from a recent paper10 confirm that DNA/CTAB adsorption behavior on PS particles and that on flat hydrophobized silica are in agreement. Surface Preparation. Mica surfaces were hydrophobically modified by Langmuir-Blodgett deposition of octadecyltrimethylethoxysilane (OTE) as described by Wood and Sharma.11 This procedure was selected over self-assembled hydrophobic layers of other silanes because it produces a stable, neutral, and flat monolayer. Briefly, the mica surfaces were first cleaved and then plasma cleaned in a Harrick Scientific Corp. model PDC3XG plasma cleaner. They were then immersed into a pH 2 HNO3 solution in a clean Langmuir trough. OTE dissolved in chloroform was spread on the water surface and allowed to react for 15 min. The barriers of the trough were then closed until a surface pressure of 12 mN/m was reached. The mica surfaces were then withdrawn from the trough at constant surface pressure, and finally baked in a vacuum oven at 100 °C for 2 h. The resulting surface had roughly circular condensed domains approximately 5-10 µm in diameter that were randomly packed together, with areas of lower silane density existing between the condensed domains.12 An AFM height image of a typical surface is shown in Figure 1. Atomic Force Microscopy. All measurements were performed on a Digital Instruments NanoScope III MultiMode AFM that was equipped with a fluid cell. V-shaped Microlevers from ThermoMicroscopes, with a nominal spring constant of 0.06 N/m, were used in all of the experiments. The 6.17 µm hydrophobic PS particle was attached to the cantilever by conventional means.13 All of the force data presented here were acquired sequentially with a single cantilever and surface. Reproducibility was confirmed by subsequent experiments. After each injection of a new solution into the fluid cell, an equilibration time of 1 h elapsed prior to acquisition of force curves. For the data presented here, one full force curve (approaching and retracting data) was obtained every 5 s (the scan rate was 0.2 Hz). The amplitude of the piezo travel (ramp size) varied from 3 µm to 300 nm, depending on the length scale of the interactions upon retraction. Thus, velocities ranged from 1200 to 120 nm/s. Scan rates of 0.1 and (10) Ca´rdenas, M.; Schille´n, K.; Nylander, T.; Jansson, J.; Lindman, B. Phys. Chem. Chem. Phys. 2004, 6, 1603. (11) Wood, J.; Sharma, R. Langmuir 1994, 10, 2307. (12) Wood, J.; Sharma, R. Langmuir 1995, 11, 4797. (13) Ducker, W. A.; Senden, T. J.; Pashley, R. M. Langmuir 1992, 8, 1831.
Figure 1. AFM contact mode image of a typical OTE-modified mica surface. The roughness (rms) of the monolayer is 0.37 nm. 0.5 Hz were also used to check if the results were sensitive to velocity. No significant differences were present. Force curves were taken repeatedly on a single spot as well as on several different locations with continuous data acquisition to ensure the data were representative of the overall interactions between the surfaces. All data presented here are representative of the entire set. Variations with time on a single location were not observed. Occasionally, qualitatively different force curves were observed when moving to a different location. These outliers are attributed to the existence of regions of lower silane density between the well-compressed OTE islands on the mica surface. Only the predominant behavior, which can reasonably be interpreted as the behavior of the OTE condensed domains, is reported here. The cantilever deflection versus piezo travel data are converted to force versus separation data by the standard method.13 The separation distances reported here are apparent separations, where the zero point is defined as the point at which the force gradient exceeds the spring constant of the cantilever. This means that at an apparent separation of zero, the surfaces may not have achieved contact due to the presence of stiff material adsorbed in the gap between the surfaces.
Results and Discussion Bare Hydrophobic Surfaces. Figure 2 shows the forces between the 6.17 µm PS sphere and a OTE-modified mica surface across 10 mM NaBr solution. On approach of the two surfaces, a long-range, monotonically attractive force is observed below 150 nm of apparent separation until the surfaces reach apparent contact. On retraction, there is a very large hysteresis loop, which corresponds to a large force of adhesion. Considering the retracting curve first: after the surfaces are brought into contact, they stick until the piezo travels 1200 nm beyond the point of first contact. At this point, the restoring force due to flexion of the cantilever exceeds the adhesive force, the colloidal particle pulls off from the surface, and the cantilever snaps back to its baseline position. The data between 100 and 1200 nm appear to be a flat line because the deflection went off the instrument’s scale. This problem occurs due to the weak spring used here, which we chose because it is most appropriate for the other force curves presented here (allows for greatest sensitivity to weak interactions).
Forces between Hydrophobic Surfaces: DNA/CTAB
Figure 2. Forces between a bare OTE-modified mica surface and a bare PS particle. Solid circles are the force measurements on approach of the surfaces, and hollow triangles are the forces upon retraction of the surfaces. The main figure is the data at low apparent separation, and the inset is the entire data set. The arrow in the inset marks a jump from strong attractive interactions (large cantilever deflections) to complete separation of the surfaces. There is an apparent inward change in separation distance that appeared in all of the acquired force curves. This is most likely due to mechanical or optical effects after contact of the cantilever with the surface that creates an imperfect compliance region (e.g., slipping or twisting of the cantilever).
Langmuir, Vol. 20, No. 15, 2004 6409
Figure 4. Forces between an OTE-modified mica surface and PS particle after DNA adsorption. Solid circles are the force measurements on approach of the surfaces, and hollow triangles are the forces upon retraction of the surfaces.
longer-ranged than van der Waals forces are expected to be in such systems.15 There is even spacing between the points on the curve in Figure 3, until a separation of 50 nm is reached. The increasing spacing of the points below 50 nm is indicative of a jump to contact due to mechanical instability. Such jumps are frequently found in the approaching force measured between hydrophobic surfaces,15 due to the force gradient exceeding the spring constant of the cantilever. The black curve in Figure 3 is a fit of the data to the equation:
F/R ) C exp(-D/λ)
Even if the data had not been off-scale, it is difficult to relate adhesion values in an AFM colloidal probe experiment to theoretical calculations due to uncertainty in the true contact region. In general, if the surfaces exhibit roughness, the contact region may be determined by high curvature asperities rather than by the particle radius.14 Thus, the contact forces will be weaker than expected. For the purposes of this study, it suffices to note that the adhesive force in the absence of adsorbed layers is large, as expected for hydrophobic surfaces.15 Figure 3 is a plot of the approaching force curve presented in Figure 2, with the scale reduced so that the features of the attractive force may be seen more clearly. The onset of the attractive force occurs somewhere between 100 and 150 nm. This is at least an order of magnitude
where F/R is the force divided by radius, D is the apparent separation, and C and λ are fitting parameters. C and λ are -1.11 ( 0.01 and 49.5 ( 0.7 nm, respectively. In a recent review of experimental studies of the forces between hydrophobic surfaces, Christenson and Claesson15 note that in many cases the force curve fits a single or double exponential decay. In cases where the force is well represented by a single exponential, the decay length was found to vary between 5 and 50 nm for different systems. Thus, with a decay length of 49.5 nm, we can see that the force between PS and OTE mica is relatively long-ranged in comparison with other systems. Additionally, we note that Wood and Sharma,12 using a surface force apparatus (SFA), found no evidence of similarly long-ranged attraction on the symmetric LB-deposited OTE layer system, only a jump to contact at 17 nm. It is beyond the scope of the current work to explain why the attractive force has this form and magnitude. We do, however, note that the most plausible explanation suggested by the current literature for the long range of the interaction appears to be the formation of one or more nanobubble bridges between the surfaces.15-17 These nanobubbles will quite probably be present in our experiments considering the method of injection of the electrolyte solution in the fluid cell of the AFM. Surfaces with Adsorbed DNA. After the bare surface measurements were acquired, the fluid cell was injected with 0.02 mg/mL DNA in 10 mM NaBr solution. A typical force curve under such conditions, collected after 1 h of equilibration, is shown in Figure 4. Three major changes from the bare surface interactions are immediately
(14) Claesson, P. M.; Dedinaite, A.; Rojas, O. J. Adv. Colloid Interface Sci. 2003, 104, 53. (15) Christenson, H. K.; Claesson, P. M. Adv. Colloid Interface Sci. 2001, 91, 391.
(16) Ishida, N.; Sakarnoto, M.; Miyahara, M.; Higashitani, K. J. Colloid Interface Sci. 2002, 253, 112. (17) Carambassis, A.; Jonker, L. C.; Attard, P.; Rutland, M. W. Phys. Rev. Lett. 1998, 80, 5357.
Figure 3. Forces between a bare OTE-modified mica surface and a bare PS particle, plotted to highlight the attractive interaction upon approach. The line is a curve fit to eq 1. Filled circles are the points used in the curve fitting, while the hollow circles are the remainder of the data set.
6410
Langmuir, Vol. 20, No. 15, 2004
apparent. First, the approaching interaction has changed from a long-range attractive force to a shorter-ranged repulsive force. The shape of this approaching curve will be considered in more detail below. Second, the retracting curve still exhibits a significant hysteresis shortly after zero separation, although the magnitude of this force is greatly decreased from the bare surface case. This adhesion event occurs in all of the recorded force curves. Finally, a second, smaller hysteresis event is evident at long separations. Such events occur in approximately 50% of the force curves. In some cases, two or three similar events occur in the same force curve, although the data presented in Figure 4 are more representative of the typical behavior. The minimum of the initial hysteresis event does not occur at zero separation. The surfaces begin to pull apart while the force is increasing, and the minimum in the force curve occurs at 31 nm. This may be significant because it demonstrates the stretching force of some bond between the surfaces, rather than a general (nonspecific) adhesive interaction, which would show increasing force while still in contact. At separations larger than the 31 nm, the force begins to decrease in a stepwise fashion. This may indicate that the breaking of several bonds or bridges is occurring. Hugel et al.18 observed similar behavior in the forces between polyvinylamines and charged surfaces and attributed this behavior to stretching and desorption of many polymer strands. In this case, another possibility exists: given that the surface coverage is quite low under these conditions (0.2 mg/m2),1 there may yet be free hydrophobic interface where nanobubbles may exist on the free surface. Thus, the bridges may be nanobubble bridges rather than polymer bridges. The stepwise behavior in that case would be due to multiple bubbles bridging the gap, or stick-slip type action of the air-water-mica contact line during the shrinking of the bubble’s contact area, due to surface heterogeneity. We favor the polymer-bridging hypothesis for two reasons: the length scale of the hysteresis loop corresponds remarkably well with the average contour length of the DNA chains (680 nm) and is longer than typically observed for hydrophobic surfaces.15 Second, to our knowledge, there is no experimental evidence of bubble bridging on surfaces with adsorbed polymer layers (even sparsely covered surfaces). The second, longer-ranged hysteresis event is more clearly interpretable as a polymer-bridging event. The length scale is much longer than one typically observes for a hydrophobic interaction.15 Although the average contour length of the DNA chains in this sample is only 680 nm, longer chains may exist in this polydisperse DNA sample. Furthermore, the shape is typical for single-chain stretching followed by desorption.18 Whether polymer bridges or air bubbles are the cause of the hysteresis events, it is clear that the measured force curve is indicative of a sparsely adsorbed DNA layer, where free interface is available for DNA segments or bubbles to create a bridge. Figure 5 is a plot of the approaching data from Figure 4, with the x-axis rescaled to focus on the repulsive interaction and with the y-axis having a natural log scale. Several features become evident. First of all, the outer region of the repulsive-force data is exponential in nature, with a decay length of 20.7 nm. This is much larger than the Debye length in the system (3.0 nm), indicating that the interaction is a steric or electrosteric repulsion. One (18) Hugel, T.; Grosholz, M.; Clausen-Schaumann, H.; Pfau, A.; Gaub, H.; Seitz, M. Macromolecules 2001, 34, 1039.
Braem et al.
Figure 5. Approaching forces between an OTE-modified mica surface and PS particle after DNA adsorption plotted on a semilog scale. Solid lines are linear regression fits to the solid circles. Hollow circles are the nonfitted points.
may interpret the decay length of the outer region of force profile as being related to the polymer layer thickness.19,20 Thus, we can use this value as a means of comparing layer thickness in the presence or absence of surfactant. Also, the apparent onset of repulsion in this case is between 60 and 70 nm (corresponding to two 30-35 nm thick layers). Between this outer region and a second exponential region, there is a flat region in the data. This region appears in all of the recorded force curves and occurs repeatably over a distance of 10-12 nm and at a force of approximately 0.25 mN/m. It is possible that this is an inward jump, indicative of an attractive interaction with a force gradient greater than the spring constant, and not a constant-force approach. There are several obvious possibilities for an attractive interaction in the system, notably the aforementioned bridging and hydrophobic interactions. The outer edge of the plateau region occurs approximately 30 nm from the onset of repulsive interaction, a distance that corresponds to the ellipsometric thickness of 31 nm found for DNA layers under these conditions on hydrophobized silica.1 This may be viewed as support for the hypothesis that a bridging attraction occurs at this distance; that is, DNA segments from one surface become close enough to the other surface to adsorb. If the bridging mechanism is responsible for the plateau, it suggests that the adsorbed layer is sparse or patchy, with free interfacial area available to chains adsorbed to the opposing surface. This would be consistent with the low surface coverage observed by Cardenas et al.1 We can see, however, that the spacing of the points in this region is roughly equal to the spacing of the points in the rest of the force curve. This is evidence that the flat region is indeed a plateau (however, it is insufficient to entirely rule out the possibility of an attractive jump). The energy of interaction can be constant over some distance if the osmotic pressure of polymers and counterions ceases to increase with decreasing separation. This could happen, for example, if polymer chains were free to escape the gap between the particle and the surface. Simulations21,22 and experiment23 have demonstrated that (19) Braem, A. D.; Biggs, S.; Prieve, D. C.; Tilton, R. D. Langmuir 2003, 19, 2736. (20) Luckham, P. F. Adv. Colloid Interface Sci. 1991, 34, 191. (21) Murat, M.; Grest, G. S. Macromolecules 1996, 29, 8282. (22) Steels, B. M.; Koska, J.; Haynes, C. A. J. Chromatogr., B 2000, 743, 41.
Forces between Hydrophobic Surfaces: DNA/CTAB
Langmuir, Vol. 20, No. 15, 2004 6411
Figure 6. Forces between an OTE-modified mica surface and PS particle after DNA and CTAB adsorption. Solid circles are the force measurements on approach of the surfaces, and hollow triangles are the forces upon retraction of the surfaces.
Figure 7. Approaching forces between an OTE-modified mica surface and PS particle after DNA and CTAB adsorption plotted on a semilog scale. Solid lines are linear regression fits to the solid circles. Hollow circles are the nonfitted points.
this is possible in an AFM experiment. We view this possibility as unlikely, however, due to the relatively large radius of curvature of our colloidal probe (as compared to a standard tip). Another possibility is that some other configuration change is occurring at the interface at constant free energy. We note that the process responsible for this apparent plateau must be reversible, because the same plateau occurs in subsequent approaches to the same spot on the surface. After the apparent plateau regime, there is an inner regime that also appears to be approximately exponential in nature, with a decay length of 8.6 nm. This is also significantly greater than the Debye length, indicating that it is most probably produced by a steric interaction as well. The existence of two distinct force regimes may indicate the presence of two distinct structural regimes in the DNA adsorbed layer. Such an explanation has been invoked to interpret similar force data between protein layers24 as well as polyvinylamine layers.25 The proposed structure is an outer, “hairy” layer with a lower density, and an inner, denser layer. In this case, most of the DNA segments would be adsorbed as trains or in loops near the surface, with a few larger loops or tails extending further away from the surface in the outer layer. We note that the onset of the outer regime occurs at approximately 60 nm and the onset of the inner regime is approximately 20 nm. These correspond to an outer layer thickness of 30 nm (which agrees with the ellipsometric thickness,1 as noted above) and an inner regime of 10 nm. Surfaces with a DNA-CTAB Coadsorbed Layer. When the surface in the above section is exposed to a solution containing 1 × 10-6 M CTAB, 0.02 mg/mL DNA, and 10 mM NaBr and allowed to equilibrate for 1 h, the interaction forces change significantly. Figure 6 shows a typical approaching and retracting force measurement. Because there are no very long-range forces present, the scale of the apparent separation axis is much smaller than in the previous figures. Two obvious differences with DNAcoated surfaces present themselves: (1) the repulsive force upon approach is significantly shorter-ranged (recalling that the apparent onset of repulsion occurred at approximately 60 nm in the DNA-only case), and (2) a much
smaller hysteresis loop is observed. Again, considering the second difference first: approximately 50% of the observed force curves displayed hysteresis loops, with some variation existing in the magnitude of the hysteresis. The curve presented here shows the average behavior; the largest hysteresis loop observed had a maximum difference between the approaching and retracting curve of 1.5 mN/ m. In addition to the more noticeable hysteresis between 0 and 10 nm, there is also a very slight hysteresis between 10 and 60 nm. The latter is so small that it may be due to baseline noise. None of the observed force curves had clearly distinguishable hysteresis events beyond 20 nm. The comparatively small magnitude of the hysteresis and the absence of any clear hysteresis at separations beyond 20 nm suggest that a fundamentally different mechanism of interaction is at work under these conditions. Possibilities include entanglements between polymer chains on each surface, hydrophobic attraction between any remaining bare patches of surface, and molecular associations (tail-tail interactions) between surfactants present in each layer. The present data are insufficient to distinguish between these possibilities. The disappearance of bridging and long-ranged hydrophobic interactions is indicative of an increase in the adsorbed amount (less bare interfacial area is available for bridges to form, and fewer hydrophobic patches exist). Figure 7 shows the approaching data on a natural log scale and with a smaller scale on the x-axis. Again, we clearly see a roughly exponential outer region of the force curve (although we note that below 0.6 mN/m the curve deviates from exponential behavior), an apparent plateau, and a higher slope inner region. The decay length of the outer region is 5.1 nm, and the apparent onset of repulsion is between 15 and 20 nm (corresponding to two 7.5-10 nm thick layers). Once again, the decay length is larger than the Debye length (3.0 nm) and thus can be correlated to the polymer layer thickness. Both the decay length and the apparent onset of attraction indicate a 3-4-fold decrease in the adsorbed layer thickness. This observed decrease in layer thickness is corroborated by the ellipsometry data of Cardenas et al.1 The introduction of CTAB to the system clearly causes a compaction of the adsorbed DNA layers. The plateau region in Figure 7 occurs repeatably, over a distance of 2-3 nm. The same arguments for what may cause the plateau region given in the above section may be applied here. Attraction due to bridging appears to be a less plausible explanation in this case, in light of the
(23) Overney, R. M.; Leta, D. P.; Pictroski, C. F.; Rafailovich, M. H.; Liu, Y.; Quinn, J.; Sokolov, J.; Eisenberg, A.; Overney, G. Phys. Rev. Lett. 1996, 76, 1272. (24) Nylander, T.; Wahlgren, N. M. Langmuir 1997, 13, 6219. (25) Poptoshev, E.; Rutland, M. W.; Claesson, P. M. Langmuir 2000, 16, 1987.
6412
Langmuir, Vol. 20, No. 15, 2004
Braem et al.
Figure 8. Forces between an OTE-modified mica surface and PS particle after DNA/CTAB coadsorption followed by rinsing. Solid circles are the force measurements on approach of the surfaces, and hollow triangles are the forces upon retraction of the surfaces.
Figure 9. Approaching forces between an OTE-modified mica surface and PS particle after DNA/CTAB coadsorption followed by rinsing, plotted on a semilog scale. Solid lines are linear regression fits to the solid circles. Hollow circles are the nonfitted points.
absence of the large hysteresis expected when a bridge is stretched and forced to desorb. If the plateau is assumed to represent a balancing of repulsive and attractive forces, the observation of the plateau having a shorter length scale in this case as compared to the DNA-only case agrees well with the shorter-ranged hysteresis that is observed. There is a region of higher slope to the left of the flat region of the force curve, with an exponential decay length of 1.3 nm. It is again reasonable to postulate that this portion of the curve corresponds to a denser core region of the adsorbed layer, while the outer, lower slope region of the force curve corresponds to a less dense outer region of the adsorbed layer. Surfaces after Coadsorption and Rinsing. Figure 8 is a plot of the approaching and retracting data after the surface has been rinsed with 10 mM NaBr solution and equilibrated for 1 h. Once again, the interaction between the surfaces is predominately repulsive, a clear indication that rinsing with 10 mM NaBr does not result in complete desorption of the DNA or DNA-surfactant complex (a return to purely attractive forces would be expected in that case if we accept that nanobbubles return after DNA desorption, or almost no forces if we assume that DNA and the DNA-surfactant complexes removed these bubbles). The decay length of the repulsive force confirms the presence of a DNA layer and is discussed below. The curves in Figure 8 show a complete absence of significant hysteresis, which is what was observed in 100% of the recorded force curves. Another clear difference between this and the preceding cases is the lower magnitude of the repulsive interaction. This may reflect the presence of less material in the outer, “hairy” region of the interface and more material at the inner region. The lower force magnitude does not imply that a lower total adsorbed amount is present in this experiment as compared to the DNA-only layers probed in Figures 4 and 5. This is because the measurable forces are those in the regime where the force gradient is less than the spring constant of the cantilever. If the inner region of the layer is very compact and dense, it may be stiff enough to cause the force gradient to exceed the spring constant of the cantilever, and higher magnitude forces are not observed. The approaching data are replotted on a natural log scale in Figure 9. A structure similar to that in the previous cases is evident here, although it is difficult to determine if there is or is not a plateau region (the data are noisier at the transition between the inner and outer regimes
because they are at a lower magnitude). The outer region of the curve has a decay length of 6.7 nm. This decay length is much lower than the 20.7 nm decay length observed for a DNA layer without exposure to CTAB and is significantly larger than the 5.1 nm decay length observed prior to rinsing. The inner decay length is 1.8 nm, also lower than the 8.6 nm length observed prior to CTAB addition and higher than the 1.3 nm observed prior to rinsing. It is therefore clear that the layer after rinsing is neither completely frozen in the CTAB-compacted configuration nor able to relax to its original thickness. Despite evidence that only DNA remains adsorbed after rinsing,1 the effect of the CTAB partially persists even after its complete removal. Several studies on polymers at interfaces have demonstrated the importance of kinetically trapped nonequilibrium states.1,26-31 In particular, strong effects of the order of addition of polymer and surfactant, as well as very slow approach to equilibrium or irreversibility on experimental time scales, have been observed. This study confirms the importance of path-dependence and nonequilibrium states in determining the behavior of physisorbed polymer layers. Furthermore, the technologically relevant phenomenon of DNA compaction is demonstrated to occur at interfaces, with a potential benefit arising from the fact that surfactant need not remain in the system to maintain some degree of compaction. One application that comes to mind is the use of a hydrophobic or hydrophobically modified colloidal particle as a vector for gene delivery. DNA could be compacted onto the particles by intermittent exposure to cationic surfactants, many of which are not suitable for use in the body. Implications for Colloidal Stability. One may speculate how the above force profiles might indicate something about the stability of a hydrophobic colloidal system in the presence of DNA and DNA-CTAB mixtures. Clearly, bare hydrophobic colloids are expected to be unstable, due to the purely attractive interaction profile. (26) Neivandt, D. J.; Gee, M. K.; Tripp, C. P.; Hair, M. L. Langmuir 1997, 13, 2519. (27) Velegol, S. B.; Tilton, R. D. Langmuir 2001, 17, 219. (28) Pagac, E. S.; Prieve, D. C.; Tilton, R. D. Langmuir 1998, 14, 4. (29) Furst, E. M.; Pagac, E. S.; Tilton, R. D. Ind. Eng. Chem. Res. 1996, 35, 1566. (30) Dedinaite, A.; Claesson, P. M.; Bergstro¨m, M. Langmuir 2000, 16, 5257. (31) Braem, A. D.; Prieve, D. C.; Tilton, R. D. Langmuir 2001, 17, 883.
Forces between Hydrophobic Surfaces: DNA/CTAB
A hydrophobic colloidal system with DNA adsorbed under the above conditions could be stable or unstable: the force curve on approach of two particles would be initially repulsive; however, if the particles are able to overcome this repulsive energy barrier, there is significant adhesive force and the possibility of bridging. In a DNA-CTABhydrophobic colloid system, one may expect stability to be achieved due to the repulsive interaction and presence of only very little adhesion and no bridging. Furthermore, even if the DNA and CTAB were subsequently removed from the aqueous phase of a colloidal mixture, a stable state would persist, because of the partially irreversible adsorption of DNA. Conclusion DNA adsorption and DNA/CTAB coadsorption have a profound impact on the colloidal forces between hydrophobic surfaces. The interaction between DNA-free surfaces is purely attractive on approach with a consistently large adhesion upon retraction. Once DNA is adsorbed, the interaction becomes predominantly repulsive on approach, and the adhesion is significantly lessened. Bridging of DNA between the two surfaces,
Langmuir, Vol. 20, No. 15, 2004 6413
however, occurs frequently and would destabilize a colloidal system that is able to overcome the repulsive energy barrier. Once CTAB is added to the mixture, the interaction remains predominantly repulsive, although shorter in length scale, with evidence of only occasional, slight adhesion. If the surfactant is removed, the interaction remains repulsive, and no bridging and very slight adhesion remain. The persistence of a robust, stabilizing DNA adsorbed layer after rinsing highlights the importance of kinetically trapped (nonequilibrium) polymer layer configurations in determining colloidal interactions. Furthermore, it may be possible to utilize such kinetic traps in developing stable DNA/colloidal particle systems where DNA compaction can be maintained in the absence of surfactant. Acknowledgment. Many thanks go to Håkan Wennerstro¨m and Tommy Nylander for illuminating discussions. The Swedish Foundation for Strategic Research Program in Colloid and Interface Technology and the Swedish Research Council financially supported this work. LA049882W
