8950
Langmuir 2008, 24, 8950-8953
Visualization of Hydrophobic Polyelectrolytes Using Atomic Force Microscopy in Solution Axel Gromer, Michel Rawiso, and Mounir Maaloum* Institut Charles Sadron CNRS - UPR 22 23, rue du Loess BP 84047, 67034 Strasbourg Cedex 2, France ReceiVed March 26, 2008. ReVised Manuscript ReceiVed May 12, 2008 We have used atomic force microscopy to study the morphology of hydrophobic polyelectrolytes adsorbed on surfaces. The polyelectrolytes consisted of polystyrene sulfonate (PSS) chains made with three charge densities: 32%, 67%, and 92%. They were adsorbed on two types of surfaces: mica, and phospholipid bilayers made of mixed neutral and cationic lipids. We show that the chains with a low charge density (32%) are collapsed in spherical globules while highly charged chains (67% and 92%) are fully extended. End-to-end distances and contour lengths of the extended chains were measured. Statistical analysis shows that the persistence length of these chains depends on the surface where they adsorb. On lipid bilayers, highly ordered monolayers are formed upon increase of the proportion of cationic phospholipids. These results show that highly charged PSS chains behave in a similar manner than the stiffer, hydrophilic DNA when adsorbed on surfaces. It could lead to the design of new types of nanostructured surfaces using polyelectrolyte molecules synthesized with specific properties.
Introduction Polyelectrolytes are polymers that carry groups which can dissociate in a polar solvent, leaving the chain charged. When the dissociation does not depend on the pH, the polyelectrolyte is called “strong”; otherwise it is a “weak” polyelectrolyte. These molecules can adsorb on oppositely charged surfaces, which has applications such as colloidal stabilization1 or multilayer film making.2 Interestingly, polyelectrolytes are also found in nature. The most famous example is DNA, which is a strong negatively charged polyelectrolyte. In our cells’ nucleus, it wraps around the positively charged histones to form the compact structure of chromosomes. Atomic force microscopy (AFM) has been successfully used for studying the morphologies of biopolymers on surfaces.3 It allows working directly in solution, and it is even possible to study dynamic properties such as conformational changes. Another characteristic of this method is the two-dimensional representation of the polymer molecules. In an ideal 2D-solution, the conformation of a polymer chain can be described by the worm-like chain model:
(
〈R2〉2D ) 4lpL
L 2lp 11 - e- 2lp L
(
))
fonate (PSS). Although PSS is negatively charged like DNA, there are important differences between these two polyelectrolytes. One of them is the chain’s flexibility. While DNA is a stiff molecule with a persistence length of ∼50 nm (in 0.1 M Na Cl), neutral polystyrene is a flexible molecule with a persistence length of only ∼1 nm. Also, because of polystyrene’s hydrophobicity, PSS chains in water might be collapsed in compact globules if the charge density is low. When the charge density is high enough, the polymer behaves as a hydrophilic polyelectrolyte. In this work, we compared the structure of PSS chains which had three charge densities: 32, 67, and 92%. Adsorption and conformation of the chains were studied directly in water so that their true behavior in solution could be seen. Moreover, adsorption was studied on two types of surfaces: mica and cationic phospholipid bilayers, to assess the influence of the substrate on chain conformation. The chains’ persistence length was evaluated, when possible, using statistical analysis and interpolation by the worm-like chain model. Finally, the proportion of cationic phospholipids in the bilayer was tuned, which led to dramatic changes in the organization of the polymer layer.
Materials and Methods (1)
where R is the end-to-end distance, L is the contour length, and lp is the persistence length. In AFM experiments, however, molecules are often trapped onto the surface during their deposition. It may result in their end-to-end distance being significantly smaller than that at equilibrium. Therefore, comparison between measured R and theoretically calculated R has been used, in the case of DNA adsorbed on mica,4 to distinguish between the trapped and equilibrium states. In-depth studies of other types of polyelectrolytes are scarcer though. Here, we present the first AFM measurements of a synthetic strong hydrophobic polyelectrolyte: polystyrene sul-
Polymer Preparation. Polystyrene chains with a molecular weight of mw ) 81036 were synthesized in our laboratories. The polydispersity was determined by size exclusion chromatography (polydispersity ) 1.05). Then, chains were sulfonated according to a method based on a patent from Exxon.5,6 PSS with three fractions of sulfonation, or charge densities, were made: 32%, 67%, and 92% (determined from elemental analysis). For AFM experiments, the lyophilized PSS samples were diluted in ultrapure water to a concentration of 2.5 µg/L to 2.5 µg/mL. Adsorption on Mica (in Water). In these experiments, the PSS was diluted in 2 mM MgCl2 before the solution was injected directly into the AFM fluid cell (sealed on freshly cleaved mica). After waiting for a few minutes, the sample was imaged in Tapping in solution. Adsorption on Lipid Bilayers. The lipids, neutral DPPC and cationic DPTAP, were purchased from Avanti Polar Lipids
* Corresponding author. E-mail:
[email protected]. (1) Bonekamp, B. C.; Alvarez, R. H.; Nieves, F. J. D.; Bijsterbosch, B. H. J. Colloid Interface Sci. 1987, 118, 366–371. (2) Decher, G. Science 1997, 277, 1232–1237. (3) Czajkowsky, D. M.; Shao, Z. FEBS Lett. 1998, 430, 51–54. (4) Rivetti, C.; Guthold, M.; Bustamante, C. J. Mol. Biol. 1996, 264, 919–932.
(5) Essafi, W.; Lafuma, F.; Williams, C. E. Macroion characterization from dilute solutions to complex fluids; ACS Symposium Series 548; Schmitz, K. S. , Ed.; American Chemical Society: Washington, DC, 1993; Chapter 21. (6) Makowski, H. S.; Lundberg, R. D.; Singhal, G. S. U.S. Patent 3870841 to Exxon Research and Engineering Company, 1975.
10.1021/la8009139 CCC: $40.75 2008 American Chemical Society Published on Web 06/27/2008
Visualization of Hydrophobic Polyelectrolytes
Langmuir, Vol. 24, No. 16, 2008 8951
Figure 1. AFM images showing the adsorption of PSS chains in water on mica and cationic bilayers. Chain structure depends on PSS charge density: chains are collapsed in globules at low charge density (a and x) while they are extended for higher charge densities (b, c, y, and z). The chains adsorb on mica in the presence of 2 mM MgCl2. The proportion of cationic lipids (DPTAP)/neutral lipids (DPPC) in the bilayers are: 1:1 (x), 3:7 (y), and 1:9 (z). These ratios were chosen because they resulted in comparable polymer surface densities.
(Alabaster, AL). Solutions of small unilamellar vesicles (containing a mixture of both lipids) were prepared according to the following procedure. The lipid powders were first dissolved in chloroform before being mixed in desired amounts. The fraction of cationic DPTAP was varied in the range: 0-65%. The solution was then dried under a steam of nitrogen and left under vacuum overnight to remove the residual solvent. Dried lipids were dispersed in buffer solution (10 mM Tris in ultrapure water; pH ) 7.4) to a concentration of 0.5 mg/mL and sonicated for 12 min in continuous mode using a tip sonicator (Branson Sonifier B15 cell disrupter, Danbury, CT). This resulted in the formation of small unilamellar vesicles. Finally, the vesicle solution was rediluted to a final concentration of 0. 2 mg/mL. Supported lipid bilayers were formed on mica using the vesicle fusion method.7 The vesicle solution was heated up to 59 °C, i.e., over the melting temperature of both lipids, which are in the gel phase at room temperature. This hot solution was then rapidly injected into the AFM fluid cell (sealed on mica). As the cell was at room temperature, the temperature of the lipids decreased rapidly. The supported bilayer nevertheless had enough time to form by vesicle fusion before the solution equilibrates to room temperature. Therefore, the supported bilayer was afterward in the gel phase. The sample was allowed to incubate for 15 min before rinsing with 1 mL of buffer solution (i.e., about 10 times the volume of the cell) to remove unfused vesicles. The sample was then imaged in Tapping in solution to check the quality of the bilayer. Finally, 100 µL of PSS solution was injected into the cell and incubated for 10 min before imaging. Atomic Force Microscopy. AFM measurements were performed using a Nanoscope III (Veeco) fitted with a 10 µm scanner. All experiments were made in Tapping mode and at ambient temperature (25 °C). For imaging in solution, a Tapping fluid cell (Veeco) was sealed on freshly cleaved mica using a silicon ring. The cantilever (Budget Sensor) had a resonance frequency of 75 kHz and a nominal spring constant of 3 N/m. The scanning rate was 1.5 Hz, and the setpoint was set to reduce as much as possible the force acting on the samples. All recorded images were flattened using the Nanoscope software. Statistical Analysis. In order to measure automatically the lengths and end-to-end distances of a large number of chains, some images were skeletonized using the software Visilog. (7) Richter, R.; Mukhopadhyaya, A.; Brisson, A. Biophys. J. 2003, 85, 3035– 3047.
Results and Discussion In water, the surface of mica is negatively charged. Therefore, in order for the negatively charged PSS chains to adsorb, 2 mM MgCl2 was added to the polymer solution. Indeed, it has been shown that the divalent cations make electrostatic bridges that bind negatively charged DNA to mica.4 Images showing individual PSS molecules adsorbed on the surface were recorded a few minutes after injection of the PSS solution inside the AFM cell. For low charge density (32%), only globules with a regular size (height ∼ 3.5 nm) were seen all over the surface (Figure 1a). For higher charge densities however (67%, 92%), we observed chains lying flat (Figures 1b, 1c). The two charge densities, 67% and 92%, have similar features. The height of the molecules is constant (∼0.6 nm) and in good agreement with the radius of PSS chain (∼0.7 nm). On the phospholipid bilayer, PSS interact electrostatically with the cationic and zwiterrionic lipid head groups. Chain adsorption is driven by the gain of entropy resulting from the release of counterions. Here again, images showing individual molecules were recorded soon after injection of the PSS solution inside the cell. It should be noted that, in a few experiments, the surface of the lipid bilayer appeared to have two slightly different height levels (as seen on the top and bottom left-hand corner of Figure 1x). This feature might result from a phase separation between the two different types of lipids. Such samples were not considered in the analysis presented later in this article. We also occasionally observed holes and cracks in the bilayers (as seen on Figures 1x, 1y, and 1z). These defects probably result from an incomplete fusion of the lipid vesicles on the mica surface. They do not, however, represent a problem for imaging the polymer chains. Thus, the morphology of the PSS chains at different charge densities is similar to that on mica: low charge molecules always appear as globules (Figure 1x), but high charge molecules are lying flat on the surface (Figures 1y and 1z). The height of these chains is constant (∼0.6 nm). Interestingly, the chains are more extended than on mica. Some chains appear almost rod-like (near the middle of Figure 1z). These observations of the structure of PSS in water suggest that the chains with a low charge density (32%) are collapsed by the hydrophobic interaction, in agreement with X-ray scattering
8952 Langmuir, Vol. 24, No. 16, 2008
Gromer et al.
Figure 3. Mean-square end-to-end distance, 〈R2〉, as a function of the contour length for high charge densities (67% is represented in black squares, 92% in circles). The data are fitted with the two-dimensional worm-like chain model (curves), which gives the persistence lengths. On mica (a), lp ∼ 12 nm. On the cationic bilayer (b), the chains are more extended: lp ∼ 34 nm (92%) and lp ∼ 19 nm (67%).
Figure 2. The PSS chains are polydisperse, as shown by AFM measurements of the chain lengths (a and b) and flux-force-field chromatography curves (c). The histograms show that the majority of chains consist of small fragments (∼10 nm). The shapes of chromatography curves confirm the existence of small molecular weights (plateau) next to the main ones (peaks).
results.8 On the contrary, for high charge densities (67% and 92%), the chains behave like hydrophilic polyelectrolytes which are in an extended-coil conformation. On a side note, the chains that are lying flat appear to be polydisperse. We measured chain lengths for statistical ensembles using image treatment (i.e., skeletonization of the chains backbones). This analysis was made in the case where the chains are adsorbed on the bilayer. The reason was that image treatment gave more accurate results when the chains were much extended. Histograms for the charge densities 67% and 92% are shown on Figures 2a and 2b. They show that the majority of molecules consist of small chain fragments which average length, about 10 nm, is clearly inferior to the length expected for the PS precursors (185 nm, assuming a monomer length: 0.25 nm). Furthermore, flux-force-field chromatography (FFF) confirmed the existence of these small chains in solution (Figure 2c). The results suggest that the sulfonation of PS damaged the chains. Therefore, we believe that, in research using PSS, particular care should be taken when interpreting data based on molecular weight. Ideally, characterization by size exclusion chromatography should always be undertaken after sulfonation. We benefitted from chains polydispersity when calculating the mean-square end-to-end distance: 〈R2〉, as a function of chain length. In this way, the persistence length of the chains can be (8) Spiteri, M.-N. The`se de doctorat, Université de Paris XI, 1997.
determined, assuming they are at thermodynamic equilibrium.4 On mica, end-to-end distance and chain length were measured manually using the Nanoscope software (400 chains for each charge density). Averages (〈R2〉) were obtained by considering chain lengths within 10 nm intervals. The results are shown in Figure 3a. The data can be interpolated by the two-dimensional wormlike chain model (continuous curve), giving a persistence length lp ∼ 12 nm, for 67% as for 92% charge (curves are superposed). The value is in perfect agreement with the OSF model of persistence length9,10 applied to the PSS chains in the bulk with 2 mM MgCl2 (also lp ∼ 12 nm). It suggests that PSS chains adsorbed on mica with a conformation similar to thermodynamic equilibrium. Actually, a similar conclusion was found for DNA adsorbed on mica in presence of MgCl2,4 suggesting that the behavior might be universal for polyelectrolytes. However, we note that the PSS chains are not really at thermodynamic equilibrium in our experiments on mica. In fact, their conformation appeared frozen on successive scans, and no desorption could be seen (i.e., the adsorption was irreversible). Thus, the chains must have relaxed to a conformation near equilibrium before being immobilized on the surface. We then analyzed the chains adsorbed on a bilayer containing 10% of cationic lipids. This time, the end-to-end distances and chain lengths were automatically determined using skeletonization (Figure 3, bottom). Interpolation by the worm-like chain model gave the persistence lengths lp ∼ 19 nm for 67% (1500 chains), and lp ∼ 34 nm for 92% (300 chains). The use of the worm-like chain model for chains at equilibrium is here justified by the fact that chains are mobile on the bilayer. Indeed, while no desorption was observed on successive scans, small conformation changes did occur. The increase of persistence length observed might (9) Odijk, T. J. Polym. Sci. Polym. Phys. 1977, 15, 477–483. (10) Skolnick, J.; Fixman, M. Macromolecules 1977, 10, 944–948.
Visualization of Hydrophobic Polyelectrolytes
Langmuir, Vol. 24, No. 16, 2008 8953
Figure 4. Variation of the density of chains on the bilayer as the proportion of cationic lipids (DPTAP)/neutral lipids (DPPC) increases. On these images (size: 300 nm), highly charged PSS chains (92%) are adsorbed on bilayers containing: 0% (a), 30% (b), 50% (c), and 65% (d) of cationic lipids. The interchain distance decreases (curve), leading to a highly ordered layer.
result from the release of counterions as the chains approach bilayer’s charged surface. This would explain why the chains with a higher charge fraction (92%) have a stronger persistence length, being more stretched by electrostatic repulsion. In contrast, on mica, chain charge is screened by the mechanism of counterion condensation. 11 It results in the same effective charge for 67% and 92% chains and, consequently, the same persistence length. It will be interesting in the future to check if an increase of persistence length occurs when individual DNA molecules are adsorbed on a cationic bilayer. When the concentration of PSS was high enough (2.5 µg/mL), the highly charged chains rapidly formed a flat layer on the bilayer. On bilayers that do not contain cationic lipids, the chains remain at a large distance from each other (∼10-50 nm) and the layer is much disordered (Figure 4a). This behavior appears similar to that of DNA on the same substrate,12 except that the great size of the DNA chains is likely to put more constraint on the layer organization. Indeed, while performing parallel experiments with DNA, we noticed more important conformational changes than in the case of PSS. When the proportion of cationic phospholipids increases, the distance between PSS chains decreases. Consequently, neighboring chains tend to align in the same direction and the layer becomes highly ordered (Figure 4b to 4c). When the interchain distance falls below 10 nm (at over 50% of cationic lipids), the AFM tip cannot penetrate completely in the layer. The surface, therefore, appears smoother (Figure 4d). The average interchain distance was determined using the spectral density analysis tool of the Nanoscope software (Figure 4, right). A variation inversely proportional to surface charge density is found. These results should be compared to previous work on DNA adsorbed on cationic bilayers.12–17 The organization of the (11) Manning, G. S. J. Chem. Phys. 1969, 51, 924–939. (12) Malghani, M. S.; Yang, J. J. Phys. Chem. B. 1998, 102, 8930–8933. (13) Mou, J.; Czajkowsky, D. M.; Yiyi, Z.; Zhifeng, S. FEBS Lett. 1995, 371, 279–282.
chains and the scaling law for the distance between them is identical. This demonstrates once again that highly charged PSS chains have a similar behavior to that of DNA when adsorbed on surfaces.
Conclusion We have studied the behavior of strong hydrophobic polyelectrolytes (PSS) adsorbed on surfaces by atomic force microscopy in solution. The variation of chain structure was studied for three charge densities. Low charge densities (32%) are collapsed in globules by hydrophobic interaction. On the contrary, high charge densities (67% and 92%) are extended coils. The conformation of highly charged chains depends of the substrate where they adsorb. On mica (in presence of MgCl2), statistical analysis shows that the conformation is comparable to thermodynamic equilibrium, with the persistence length lp ∼12 nm. However, on cationic lipid bilayers, chains are more extended, lp ∼34 nm (92%), and can form highly ordered monolayers. This work shows that previous results for DNA adsorbed on mica and bilayers can be extended to other highly charged polyelectrolytes, including synthetic ones. It could lead to the design of new types of nanostructured surfaces, using, for example, macromolecules having different geometries. Acknowledgment. We acknowledge Franc¸ois Isel for synthesizing the PSS samples, Christophe Contal for assistance with AFM, Pascal Marie for guidance with image treatment, and Alain Rameau for help with FFF. Finally, Albert Johner is acknowledged for helpful discussions and advice. LA8009139 (14) Ra¨dler, J. O.; Koltover, I.; Salditt, T.; Safinya, C. R. Science 1997, 275, 810–814. (15) Fang, Y.; Yang, J. J. Phys. Chem. B 1997, 101, 441–449. (16) Fang, Y.; Yang, J. J. Phys. Chem. B 1997, 101, 3453–3456. (17) Clausen-Schaumann, H.; Gaub, H. E. Langmuir 1999, 15, 8246–8251.
