Nonspecific Adsorption of Charged Quantum Dots on Supported

Feb 4, 2011 - Nonspecific Adsorption of Charged Quantum Dots on Supported Zwitterionic Lipid Bilayers: Real-Time Monitoring by Quartz Crystal Microbal...
5 downloads 10 Views 4MB Size
Download PDF
ARTICLE pubs.acs.org/Langmuir

Nonspecific Adsorption of Charged Quantum Dots on Supported Zwitterionic Lipid Bilayers: Real-Time Monitoring by Quartz Crystal Microbalance with Dissipation Xinfeng Zhang and Shihe Yang* Department of Chemistry, William Mong Institute of Nano Science and Technology, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China

bS Supporting Information ABSTRACT: Understanding how the composition and environmental conditions of membranes influence their interactions with guest species is central to cell biology and biomedicine. We herein study the nonspecific adsorption of charged quantum dots (QDs) onto a supported zwitterionic lipid bilayer by using quartz crystal microbalance with dissipation (QCM-D). It is demonstrated that (1) the adsorption of charged QDs is chargedependent in a way similar to but much stronger than that of the capping molecules by reason of size effect; (2) the adsorption behavior of charged QDs is dominated by electrostatic interaction, which can be well described by an “adsorption window”; (3) the “adsorption window” can be broadened by exploiting the bridge role of Ca2þ ions; and (4) by introducing a cationic lipid into the zwitterionic lipid bilayer, one can achieve preferential adsorption of anionic QDs but suppression of the cationic QD adsorption. Our QCM-D data also indicates that these different adsorption traits effect different changes in the dissipation of supported lipid bilayers (SLBs) after adsorption of the charged QDs. The different adsorption propensities of cationic and anionic QDs on SLBs have reinforced the picture of electrostatic interactions. We believe that these findings provide important information on QD-lipid membrane interactions, which will help to develop new drug molecules and efficient drug delivery systems, and to predict and unravel their potential toxicities if any.

’ INTRODUCTION The interaction of nanomaterials with cell membranes is critical to their applications in phototherapy, imaging, and drug/gene delivery.1-5 These applications require a deep understanding of and firm control over nanoparticle-cell membrane interactions, which are mainly dictated by sizes, shapes, and surface properties of the nanomaterials.1,2 Better understanding of the nanoparticle-cell membrane interactions is also essential to assessing the potential cytotoxicity of nanomaterials, which has become one of the greatest concerns in the development of nanomaterials and nanotechnology.6,7 Despite the promise of nanoparticles in biomedical applications, our understanding on the interactions between nanoparticles and cell membranes is still inadequate at the present.1,2 Experimental and computational works have implicated the importance of the size, shape, and surface properties of nanomaterials in determining their interactions with cell membranes.8-11 Particularly important are surface charges.1 For example, the uptake of charged nanoparticles by cells is more facile than that of neutral nanoparticles, r 2011 American Chemical Society

and cationic nanoparticles have the greatest efficiency in cellmembrane penetration and cellular internalization.12-14 Cell culture assays indicated that cationic particles are more toxic than anionic particles.15,16 This was explained by the fact that the negative membrane potential of most cells favors the adsorption of cationic nanoparticles. However, there have also been reports that anionic nanoparticles could get into cells more efficiently than cationic nanoparticles.17,18 Experimental and theoretical researchers working on biomimetic membranes maintained that anionic nanoparticles interact more strongly than cationic nanoparticles with model zwitterionic lipids.19-21 These controversies most likely originate from the complexity of the cell constituents and the cell environments,18,22 which make it rather difficult to dissect the nanoparticle-cell membrane interactions and thus demand Received: November 8, 2010 Revised: January 5, 2011 Published: February 04, 2011 2528

dx.doi.org/10.1021/la104449y | Langmuir 2011, 27, 2528–2535

Langmuir

ARTICLE

systematic studies on the nature and mechanistic details of these interactions. While tremendous efforts have been dedicated to biomedical study with various nanoparticles, little in situ monitoring work has focused on how the composition and environmental conditions of lipid membranes affect their cell internalizations. In this article, we report the first in situ study on the adsorption behavior of charged quantum dots (QDs) onto supported lipid bilayers (SLBs) by focusing on the influence of charge states and environmental conditions (e.g., pH, Ca2þ ions) of the bilayers (e.g., zwitterionic and charged lipids). SLBs is a model system for fundamental studies of biological membranes,23,24 whereby lipid composition and environment can be set and controlled to imitate biological conditions. Here the adsorption of both anionic and cationic QDs on SLBs under different conditions were monitored as a function of time by quartz crystal microbalance with dissipation monitoring (QCM-D), which is a very sensitive method for tracking the binding processes of nanoparticles at surfaces and providing time-resolved information about not only the amounts adsorbed but also the morphologies of surface-confined films.25,26 Different adsorption behaviors for cationic and anionic QDs have been elucidated, which are useful for developing drug delivery systems and for understanding cytotoxicities of the nanomaterials.

’ EXPERIMENTS Materials. 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC, purity 99%) and dioleoyltrimethylammonium propane (DOTAP, purity 99%) were purchased from Avanti Polar Lipids (Alabaster, AL). 3-Mercaptopropionic acid (MPA, 99%) was purchased from Aldrich. 2-Aminoethanethiol hydrochloride (AET, 98%) was purchased from Acros. Selenium powder (99%) was provided by Aldrich. Cadium choloride (99%) was provided by Aldrich. Other reagents used including chloroform (99.8%, Laboratory-Scan Ltd.), tris(hydroxymethyl)aminomethane (Tris, 99.9þ%, Aldrich), sodium chloride (99.8%, Riedel-deHaen), cacium chloride-2-hydrate (99%, Riedel-deHaen), and sodium borohydride (Merck, 96%) were of the highest purity available. All the chemicals were used without further purification. The buffer solutions used throughout the experiment were Tris buffer containing 10 mM Tris and 100 mM NaCl. The pH was adjusted to 8.0 with a 2.0 M HCl solution. Milli-Q water (Barnstead, compact ultrapure water system) with a resistivity of 18.3 MΩ 3 cm was used. The percentages mentioned throughout the paper are in molar fraction unless otherwise stated. Preparation of Vesicles. Vesicles were prepared by extrusion. Briefly, phospholipid mixtures were obtained by mixing the appropriate volumes of phospholipid solutions in chloroform and allowed to dry in a stream of nitrogen, followed by desiccation under vacuum for 8 h to remove the residual organic solvent. The resulting phospholipid film was resuspended in Tris buffer overnight. The solution was then extruded 15 times through polycarbonate membranes (100 nm pore size) to produce uniform vesicles. The final concentration of the vesicles was 1 mg/ mL. The vesicles were stored under 4 °C, and used within two weeks. Preparation of QDs. CdSe QDs were prepared by aqueous synthesis method reported by Rogach et al.27,28 Briefly, for preparation of MPA or AET-capped QDs, 2 mmol CdCl2 was dissolved in 100 mL of water with 1 mL of MPA or AET; the pH of the solution was adjusted to 10.0 or 6.0 with 1 M NaOH.

Figure 1. (a,b) Chemical structures of DOPC and DOTAP, respectively. (c) Typical QCM-D curve monitoring the formation process of an SLB through the vesicle fusion. (d,e) Typical AFM images of a DOPC bilayer. (e) Zoom-in image taken from the center of image d. Inset of e: Cross-sectional profile taken from the red line in image e.

Then the fresh 1 mmol NaHSe solution was quickly injected into the N2-saturated Cd2þ solution under fast stirring. The solution was refluxed at 100 °C for 2 h. The CdSe nanocrystals were precipitated out by ethanol, and collected by centrifugation. It was stored under 4 °C, and diluted with tris buffer before use. Dissipative QCM. Dissipative QCM measurements were performed with a QCM-Z500 system (KSV Instruments, Finland) equipped with a temperature control unit QCM-501. The technique is based on the resonant oscillation of a piezoelectric quartz crystal disk at a frequency (f) and energy dissipation (D), which, respectively, characterize the mass and the viscoelastic property of the molecules adsorbed on the crystal surface. In vacuum or air, if the layer is rigid, evenly distributed, and much thinner than the crystal, Δf is related to Δm by the Sauerbrey equation, Δm = CfΔf/n, where n = 1, 3, 5, ..., is the overtone number, and Cf = -17.7 ng 3 cm-2 3 Hz-1 at f = 5 MHz is the mass-sensitivity constant. The dissipation factor is defined by ΔD = Ed/2πEs, where Ed and Es are, respectively, the energies dissipated and stored during one cycle of oscillation. If not stated otherwise, dissipations and changes in normalized frequency of the third overtone (n = 3, i.e., 15 MHz) will be presented. For a 2529

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir given experiment, the silica crystal was initially exposed to Tris buffer and rinsed several times until the baseline was stable. Then, the buffer was replaced by a vesicle solution, and the subsequent deposition of vesicles was monitored in time. After the formation of a bilayer, different amounts of QD solution were introduced into the QCM chamber. All of the dissipative QCM measurements were carried out at 25 ((0.1 °C). Sample Characterization. TEM images were recorded by a JEOL-2010 electron microscope operated at 200 kV. Ultraviolet-visible (UV-vis) spectra were measured on a MILTON ROY Spectronic 3000 Array instrument. ζ-Potential was measured by a Zeta Potential Analyzer (Brookhaven Instruments Corporation). Atomic force microscopy (AFM) images were obtained at room temperature using a Innova atomic force microscope (Veeco) in tapping-mode. Silicon tips (SNL-10 type, Veecoprobe) with spring constants of ∼0.32 N/m and resonance frequencies between 10 and 25 kHz in aqueous solution were used.

’ RESULTS AND DISCUSSION The carboxyl group-capped (COOH-QDs) and amine groupcapped (NH2-QDs) CdSe QDs were prepared by using MPA and AET as stabilizers, respectively, in aqueous solution synthesis.27,28 The sizes of these QDs determined from transmission electron microscopy (TEM) and UV-vis adsorption were of about 2-3 nm, as shown in Figure S1, Supporting Information, which is very consistent with reports of other researchers.27,28 The ζ-potentials of COOH-QDs and NH2-QDs at pH 8 were measured to be -22.86 and þ15.99 mV, respectively, confirming the expected charge states of the nanoparticles. To study the interactions of nanoparticles with lipid membranes, a zwitterionic lipid DOPC and a cationic lipid DOTAP were chosen to construct SLBs as a model system. The relevant molecular structures are shown in Figure 1a,b. The formation of the SLBs on silica substrates through vesicle fusion, a well established process, was monitored by QCM-D and AFM. The distinctive U-shaped feature in the QCM-D Δf response could be used to indicate the formation and judge the quality of the SLBs, as shown in Figure 1c. As found previously, the equilibrium Δf is typically around 25 Hz for a good SLB on a QCM crystal.29 The kinetics of SLB formation was only slightly changed at different ionic strengths or by mixing in a charged lipid such as DOTAP. The SLBs were further imaged by AFM. Figure 1d,e shows typical AFM images of a DOPC SLB deposited on a silica substrate. The silica surface was found to be fully covered by the SLB formed from vesicle fusion (Figure 1d). Sometimes, cracks and holes could be found in the bilayer as shown in the zoom-in AFM image in Figure 1e. The thickness of the DOPC bilayer is estimated to be around 5 nm from the cross-sectional profile, in accord with the literature reports.30 DOPC SLBs were first targeted to study the adsorption of COOH-QDs and NH2-QDs (0.1 mg/mL, pH 8) by QCM-D. Control experiments were also performed with MPA and AET molecules. Generally, the changes in frequency (Δf) and energy dissipation (ΔD) of the QCM-D characterize the mass and the viscoelastic properties of the molecules adsorbed on the crystal surface, respectively. Under ideal conditions, i.e., when the adsorbed layer on the crystal surface is rigid, there is a linear relationship between the change in frequency and the adsorbed mass according to the Sauerbrey equation. However, hydrodynamically entrained water also contributes to the Δf and could

ARTICLE

Figure 2. QCM-D responses of (a) MPA, (b) COOH-QDs, (c) AET, and (d) NH2-QDs exposed to DOPC SBLs at pH 8.0. The blue arrows indicate the injection of pH 8.0 tris buffer.

lead to significant deviations from the linear Sauerbrey relation. In the case studied here, we found that the binding of QDs mostly resulted in large dissipation changes of the lipid bilayer, namely, the ΔD of the viscoelastic layer was found to be larger than 10-6 over a 5 Hz range of the Δf, weakening the applicability of the Sauerbrey equation.26,27 For this reason, we did not transform the Δf values into mass variations. Nevertheless, the Δf values provide a semiquantitatively analysis of the adsorption processes of QDs onto SLBs. Figure 2a,b shows the time profiles of Δf and ΔD during the exposure of MPA and COOH-QDs to the QCM-D sensor crystal. There are hardly any changes in Δf and ΔD for COOH-QDs and MPA, indicating unfavorable adsorption of COOH-QDs and MPA to the DOPC bilayer at pH 8. By contrast, exposure of AET and NH2-QDs to the DOPC bilayer resulted in obvious decreases in Δf and increase in ΔD, as shown in Figure 2c,d, respectively, signifying the attractive interaction between these species and the bilayer. In the case of NH2-QDs, the Δf reached an equilibrium value of -32 Hz in about 15 min after the first exposure, and incurred a further decrease of about 40 Hz after the second exposure. The subsequent rinsing with Tris buffer did not discernibly change the Δf, indicating firm binding of NH2-QDs to the bilayer. However, the AET molecules adsorbed on the DOPC bilayer were easily removed by simply rinsing with buffer. The much stronger binding of NH2QDs to SLBs than that of their surface capping ligand AET can be understood simply by their much larger size, which spawns a cooperative effect of binding and trims down unbinding-prone Brownian motion. A similar size-dependent adsorption behavior has been noticed for polycationic dendrimers.31,32 Taken together; these results show that the adsorption of QDs is charge-dependent in a way somewhat similar to that of their capping agents. Two groups of different adsorption behaviors can be discerned from Figure 2: (1) rapid decrease/increase of Δf, associated with a fast reversible adsorption/desorption process (see Figure 2a-c); (2) slow decrease of Δf reaching an equilibrium in tens of minutes, corresponding to an activated irreversible binding process (see Figure 2d). It appears that the first group is associated with weak adsorption and the second is associated with strong adsorption because the cationic QDs bind to SLBs much more strongly than the anionic QDs. 2530

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir

ARTICLE

Figure 3. The “adsorption window” for estimation of the adsorption behavior of COOH-QDs (a) and NH2-QDs (b) on a DOPC bilayer. (c,d) QCM-D responses of the COOH-QDs and NH2-QDs exposed to the DOPC bilayer at different pH values.

Our finding above seems to contradict some recent reports proposing that negatively charged nanoparticles bind more strongly than positively charged nanoparticles to a zwitterionic lipid bilayer.19-21 It was suggested from a simple geometrical relation between the P--Nþ headgroup dipole of the zwitterionic lipid and the charged nanoparticle that anionic nanoparticles would adsorb more strongly than cationic ones.19-21 However, such a crude assumption overlooks more complicated factors of the membrane environment such as pH value. Actually, the zwitterionic DOPC was found to have an isoelectric point of ∼4, caused by the unsymmetrical adsorption of hydroxide and hydronium ions on the outer leaflet of the membrane.33 This means that the DOPC bilayer should be negatively charged at pH 8, the working condition of our experiment. Thus a simple electrostatic line of reasoning can nicely explain our experimental result: the negatively charged DOPC SLBs at pH 8 favor the adsorption of cationic NH2-QDs but disfavor the adsorption of anionic COOH-QDs. In analysis of the different adsorption behaviors of the two types of QDs on the DOPC bilayer, we notice that the surface carboxyl groups of COOH-QDs have a pKa value of about 4-6, while for the surface amino groups of NH2-QDs, the pKa is approximately 9-10. Bearing in mind the isoelectric point of DOPC (∼4),33 we can expect an “adsorption window” for the charged nanoparticles on the zwitterionic DOPC bilayer. This is illustrated in Figure 3a,b: the adsorption of COOH-QDs is favored at about pH 4-6 (depending on the exact pKa of carboxyl group under certain conditions), while the adsorption of NH2-QDs is favored at pH 4-9. This compellingly explains why, at pH 8.0, the NH2-QDs could adsorb strongly on the DOPC bilayer, whereas COOH-QDs could not. To further verify the electrostatic picture of the nanoparticle-lipid interaction in terms of the “adsorption window”, we have investigated the effect of pH in the buffer solution. In this effort, both COOH-QDs and NH2-QDs were exposed to the bilayer at different pH values, as shown in Figure 3c and 3d,

Figure 4. QCM responses showing the effect of Ca2þ ions on the adsorption behavior of COOH-QDs (a) and NH2-QDs (b) on the DOPC SLBs at pH 8.

respectively. One can clearly see that exposure of the COOHQDs to the SLBs at pH 5.5 resulted in a Δf decrease by about 15 Hz, due to the strengthened adsorption of COOH-QDs at pH 5.5. Rinsing with Tris buffer did not cause noticeable changes in Δf, indicating an irreversible binding of COOH-QDs at this pH value just like the binding of NH2-QDs at pH 8. However, rinsing with a pH 3.3 buffer induced an increase in Δf and decrease in ΔD, suggesting that the binding of COOH-QDs is weaker at pH 3.3. On the other hand, the adsorption of NH2-QDs on the DOPC bilayer became unfavorable at pH 3.5, whereas their adsorption started at pH 4.5 and picked up at pH 5.5 (see Figure 3d). These pH-controlled experimental results clearly substantiate the “adsorption window” put forth above. Our ζpotential measurements of the DOPC vesicles provide another testimony of the “adsorption window”, which were conducted under the same working conditions with varying pH values (see Figure S2 in the Supporting Information). The results clearly indicate that the DOPC vesicles are positively charged at pH 3.5 and negatively charged at pH > 5.5. Similar results were also reported previously on the pH-dependent charging states of DOPC bilayers.33 2531

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir In complex media such as in biological systems, many environmental factors could also change the “adsorption window” described above one way or another. To address this issue, we opted to study the influence of Ca2þ ions on the adsorption behavior of QDs, since Ca2þ ions are widely extant in living cells and can drastically alter the physical organization of phospholipid membranes. This experiment was performed with a buffer solution containing 2 mM CaCl2 at pH 8.0. Figure 4 shows the adsorption profiles of QDs on a DOPC bilayer in the presence of Ca2þ ions detected by QCM-D. Whereas no frequency change was observed for the COOH-QDs without Ca2þ, there was a large frequency decrease (∼18 Hz) in the presence of Ca2þ (Figure 4a), clearly pointing to the enhancement effect of Ca2þ on the QD adsorption on the lipid bilayer. On the contrary, the adsorption of NH2-QDs on the bilayer was greatly suppressed by the Ca2þ ions: the Ca2þ-induced frequency decrease was only 3 Hz, which is much smaller than that of 34 Hz in the absence of Ca2þ (Figure 4b). Notice that when a Tris buffer solution containing Ca2þ ions was exposed to the SLBs without the addition of QDs solutions, no detectable frequency change was found, confirming that the frequency changes in these cases were mainly due to binding of the QDs (see Supporting Information Figure S3). It is remarkable that the addition of Ca2þ ions has flipped over the adsorption aptitude of both COOH-QDs and NH2-QDs on the lipid bilayers. Most likely, the presence of Ca2þ ions has changed the surface charge state of the zwitterionic DOPC bilayer. Indeed, McManus et al. reported that binding of Ca2þ ions can change the polarity of a zwitterionic DOPC bilayer from negative to positive,34 by bridging two neighboring lipid molecules through the phosphate groups and thus neutralizing their

Figure 5. QCM responses showing the effect of introducing cationic DOTAP to the DOPC bilayers on the adsorption behavior of COOHQDs (a) and NH2-QDs (b) at pH 8.

ARTICLE

negative charges. Ainalem et al.35 also found that Ca2þ associates with a surface-deposited lipid bilayer, and the observed PC:Ca2þ ratio of 2:1 is in perfect agreement with the model proposed by McManus et al. that one Ca2þ cation associates with two lipid HGs.34 Next, we investigate the effects of membrane constituents on the adsorption of QDs by incorporating a cationic lipid DOTAP to form a mixed DOTAP:DOPC bilayer. Figure 5 shows the QCM-D responses after exposing the COOH-QDs and NH2QDs to the DOTAP:DOPC mixed bilayers with two different molar compositions (10:90 and 50:50). A decrease of Δf was detected at about 3 and 53 Hz, respectively, when the COOHQDs were exposed to the 10:90 and 50:50 DOTAP:DOPC bilayers. A reverse trend was found for the NH2-QDs: Δf was decreased by 50 Hz on the 10:90 DOTAP:DOPC bilayers, and by 3 Hz on the 50:50 DOTAP:DOPC bilayers. In short, the presence of the cationic DOTAP in the bilayer enhances the adsorption of the COOH-QDs but suppresses the adsorption of the NH2-QDs. This result is again explicable in terms of the simple electrostatic picture, and is actually in line with the practice of using cationic lipids to bolster the uptake of anionic nanoparticles.36,37 A closer look at Figure 5a is necessary. With 10% DOTAP, equilibrium has almost been reached at only -3 Hz even at a high concentration of COOH-QDs (0.6 mg/mL). When DOTAP was increased to 50%, however, even a lower concentration of COOH-QDs (0.3 mg/mL) could cause a much larger change in frequency by about 16-fold. This means that the amount of the adsorbed COOH-QDs is proportional to the percentage of DOTAP not in an order much higher than linearity, suggestive of an amplification effect of the DOTAP on the QD adsorption. Here the added DOTAP may not simply play the role of binding site for the COOH-QDs, but it could alter the surface potential of DOPC bilayer by, for example, changing the headgroup structure of DOPC.38 Previous experiments and molecular dynamics simulations have shown that the P-N dipoles of phosphatidylcholine can change their orientation from 10° (relative to the bilayer membrane surface) to around 60° in the presence of DMTAP, due to the electrostatic repulsion between cationic TAP groups and the Nþ moiety of the choline groups.38,39 Such a DOTAP-mediated reorientation of the DOPC headgroup should result in a more positive surface potential of the SLB, thus greatly augmenting adsorption of the COOH-QDs. This is also consistent with the concomitant observation that adsorption of the NH2-QDs was largely suppressed.

Figure 6. ΔD vs Δf plots for the adsorption processes of COOH-QDs and NH2-QDs onto DOPC SLBs under different conditions: (a) at pH 5.5; (b) with Ca2þ for COOH-QDs and without Ca2þ for NH2-QDs; (c) with 10% DOTAP. 2532

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir The COOH-capped and NH2-capped nanoparticles are the two representative prototypes of charged particles widely investigated for targeting cells,14,18,40,41 since they are readily amenable to further conjugation with biomolecules. It is widely believed that surface charges have great effect on the internalization process of QDs, but its direct proof and systematic studies have been difficult. Our result clearly shows that the different adsorption behaviors of COOH-QDs and NH2-QDs onto SLBs are driven by the electrostatic force and can be easily modulated by cationic lipid or Ca2þ ions and by changing the pH. Additional information on how the different adsorption behaviors possibly induce morphological changes of the SLBs can be inferred from the dissipation data. Figure 6 presents the QCM-D results by plotting ΔD versus Δf under different conditions. Note that the ΔD-Δf plots can provide information on the energy dissipation per unit mass added to the quartz crystal. A large dissipation change is commonly associated with coupled water and flexible conformation of the attached objects; whereas a small dissipation reflects dehydrated and compact layers.42 At pH 5.5 (Figure 6a), the different adsorption patterns COOH-QDs and NH2-QDs onto DOPC SLBs are striking: the adsorption of COOH-QDs resulted in large changes in ΔD, whereas only slight increases in ΔD were detected upon the adsorption of NH2-QDs. This means that adsorption of NH2QDs at pH 5.5 formed a rigid layer on SLBs; whereas adsorption of COOH-QDs at pH 5.5 softened the bilayer. A likely explanation is that the charge-density of NH2-QDs at pH 5.5 is much higher than that of COOH-QDs, in keeping with the more firm binding of NH2-QDs with SLBs than that of COOH-QDs as well as the resulting higher dissipation of the latter than the former. This is further supported by the fact that the dissipation change per mass of NH2-QDs at pH 5.5 (red line in Figure 6a) is even lower than that at pH 8 (red line in Figure 6b), arising from the progressively increasing protonation of the NH2- group on the surface with decreasing pH.25 Our result is not unlike what has been observed previously: weakly bound protein resulted in large energy dissipation onto a liquid crystal phthalocyanine surface.42 A similar pattern was observed for the adsorption of COOHQDs in the presence of Ca2þ and for the adsorption of NH2-QDs in the absence of Ca2þ (Figure 6b), namely, the adsorption of COOH-QDs softened the SLBs to a much greater extent than the adsorption of NH2-QDs. This is reasonable since the adsorption of negative COOH-QDs caused larger changes in the P-N dipoles of the phosphatidylcholine in the presence of Ca2þ, resulting in local structural reorganization of the bilayer.43 Also the redistribution of Ca2þ at the bilayer surface upon the binding of QDs would also lead to structural variations of the bilayer. Both of these factors are likely to increase the SLBs dissipation. It has been reported that Ca2þ ions can regulate the binding of DNA to zwitterionic lipid bilalyers.34,35 Considering the similarly negative charge states of the COOH-QDs to those of DNA, one would expect a similar effect of Ca2þ on the adsorption of the COOH-QDs onto the zwitterionic SLBs. Finally, with 10% DOTAP in the SLBs, the adsorption of COOH-QDs, albeit small, brought about a greater change in the SLBs morphologies than the adsorption of NH2 -QDs (Figure 6c). Similar to the effect of Ca2þ, the presence of DOTAP could also greatly change the headgroup configurations of DOPC, as discussed above. Conceivably, before the adsorption of QDs, the DOTAP should be more uniformly distributed

ARTICLE

in the mixed bilayer simply due to the electrostatic repulsion between the cationic lipids. However, the binding of QDs onto the DOTAP-DOPC mixed bilayer could alter the configurations of the bilayer and render it more dissipative. In this connection, we notice a recent report that t-RNA could induce segregation in mixed cationic-zwitterionic lipid bilayers.44 To get one step further into the detail, the COOH-QDs, because of the negative charges they carry, are likely to cause more extensive structural rearrangements of the positively changed SLBs than the NH2-QDs, leading to the larger dissipation changes. Nevertheless, in the presence of 10% DOTAP, the more profuse binding of NH2-QDs ultimately induced a much larger dissipation. Collectively, our results clearly illustrate that environmental conditions such as pH value, Ca2þ ions, and cationic lipid can not only change the adsorption propensities of both COOH-QDs and NH2-QDs but also greatly mold the after effects of their binding to the SLBs. By using the SLBs as model lipid membranes and in situ monitoring, we have demonstrated that electrostatic force dominates the adsorption of cationic and anionic QDs onto the lipid bilayers, and that various parameters, e.g., pH, Ca2þ ions, and/or positively charged lipids, can influence the adsorption processes. The ability to modulate the QD-lipid interactions will open up new approaches to the development of targeted drug delivery systems. Take as an example the report on pH-controlled stimuliresponsive liposome fusion mediated by adsorption/desorption of COOH-capped gold nanoparticles.3 Here only at acidic environments (e.g., pH < 5) the liposome could be used for controlled release of drugs from liposomes. In light of our present work, the working window might be appropriately changed by employing Ca2þ ions and charged lipids. It is striking that the QDs-SLBs interactions manifested in our in situ monitoring experiments can be understood in terms of a simple electrostatic force framework. Recently, Laurencin et al. argued that a simple electrostatic picture falls short of capturing giant unilamellar vesicle (GUV)-nanoparticle interactions drawing on their finding that cationic nanoparticles tended to bind to positively charged bilayers, whereas anionic nanoparticles remained inert.45 In point of fact, this experimental result falls well within the electrostatic framework we have proposed above. The GUVs used by the authors consisted of a 9/1 (w/w) ratio of DOPC/DOTAP. This lipid composition is the same as we used in the DOPC SLBs with 10% DOTAP, for which there was still a large amount of cationic nanoparticles adsorbed but the adsorption amount of anionic nanoparticles was very small. In this sense, the results of Laurencin et al. and ours are congruent considering the better mass resolution of QCM-D than confocal microscopy. However, with a higher percentage of DOTAP in SLBs up to 50%, a scenario which was not researched by Laurencin et al., both the suppression of cationic QD binding and the enhancement of anionic QD binding became prominent. Therefore, we have confirmed through our present work in tandem with that by Laurencin et al. the dominant role of electrostatic forces in the nonspecific interaction between charged nanoparticles and lipid bilayers.

’ CONCLUSIONS To conclude, our work has established a QCM-D platform for in situ tracking of nanoparticle-membrane interactions. We have demonstrated how the composition and environmental conditions of membranes can affect the adsorption behaviors of 2533

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir cationic and anionic QDs on lipid bilayers by the in situ monitoring technique. First, we have shown that the adsorption of charged QDs is via a process similar to that of their capping ligands, but binding is much stronger than that of the capping molecules, seemingly as a result of the size-effect. Second, we have established that the adsorption behavior of charged QDs is dominated by electrostatic interaction characterized by an “adsorption window”. Third, we have broadened the “adsorption window” by exploiting the bridging role of the Ca2þ ions. Finally, by introducing a cationic lipid into the zwitterionic lipid bilayer, we have achieved a preferential adsorption of anionic nanoparticles but a suppression of the cationic nanoparticle adsorption. It is of the utmost importance to reveal that the electrostatic interactions between the QDs and SLBs dominated in all of the nonspecific adsorption pathways. Furthermore, we have shown that electrostatic interactions not only determined the adsorption pathways of QDs but also the outcome of the QDsSLBs bindings. These results provide valuable information for understanding, evaluating, and predicting the efficiencies of nanoparticles internalization and their cytotoxicity. Significantly, all of the results have shown a good correlation with many in vitro and in vivo cellular uptake studies, suggesting the critical role of lipid bilayer-nanoparticle interaction in the cellular uptake processes. The molecular mechanisms of such interactions could be effectively used for optimizing the characteristics of nanomaterials for biological applications, such as in the development of drugs and new schemes for controlled drug deliveries and therapies.

’ ASSOCIATED CONTENT

bS

Supporting Information. Details of the QDs characterization, QCM data of monitoring Ca2þ effect on SLBs, and the ζpotential of DOPC vesicles. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION *E-mail: [email protected].

’ ACKNOWLEDGMENT Support from HK-RGC GRF (604107 and 604608) is acknowledged. The authors also thank the Consulate General of France in Hong Kong, the France-Hong Kong (MAE) PROCORE program, and the Hong Kong University of Science and Technology (CGF07/08.SC01 and CGF07/08.SC01-M1). ’ REFERENCES (1) Verma, A.; Stellacci, F. Small 2010, 6, 12–21. (2) Nel, A. E.; Madler, L.; Velegol, D.; Xia, T.; Hoek, E. M.; Somasundaran, P.; Klaessig, F.; Castranova, V.; Thompson, M. Nat. Mater. 2009, 8, 543–557. (3) Pornpattananangkul, D.; Olson, S.; Aryal, S.; Sartor, M.; Huang, C. M.; Vecchio, K.; Zhang, L. ACS Nano 2010, 4, 1935–1942. (4) Xia, T.; Kovochich, M.; Liong, M.; Meng, H.; Kabehie, S.; George, S.; Zink, J. I.; Nel, A. E. ACS Nano 2009, 3, 3273–3286. (5) Peer, D.; Karp, J. M.; Hong, S.; Farokhzad, O. C.; Margalit, R.; Langer, R. Nat. Nanotech. 2007, 2, 751–760. (6) Lewinski, N.; Colvin, V.; Drezek, R. Small 2008, 4, 26–49. (7) Tarantola, M.; Schneider, D.; Sunnick, E.; Adam, H.; Pierrat, S.; Rosman, C.; Breus, V.; Sonnichsen, C.; Basche, T.; Wegener, J.; Janshoff, A. ACS Nano 2009, 3, 213–222. (8) Cho, E. C.; Au, L.; Zhang, Q.; Xia, Y. N. Small 2010, 6, 517–522.

ARTICLE

(9) Roiter, Y.; Ornatska, M.; Rammohan, A. R.; Balakrishnan, J.; Heine, D. R.; Minko, S. Interaction of Nanoparticles with Lipid Membrane. Nano Lett. 2008, 8, 941–944. (10) Ginzburg, V. V.; Balijepalli, S. Nano Lett. 2007, 7, 3716–3722. (11) Li, Y.; Chen, X.; Gu, N. J. Phys. Chem. B 2008, 112, 16647–16653. (12) Cho, E. C.; Xie, J. W.; Wurm, P. A.; Xia, Y. N. Nano Lett. 2009, 9, 1080–1084. (13) Tan, S. J.; Jana, N. R.; Gao, S. J.; Patra, P. K.; Ying, J. Y. Chem. Mater. 2010, 22, 2239–2247. (14) Arvizo, R. R.; Miranda, O. R.; Thompson, M. A.; Pabelick, C. M.; Bhattacharya, R.; Robertson, J. D.; Rotello, V. M.; Prakash, Y. S.; Mukherjee, P. Nano Lett. 2010, 10, 2543–2548. (15) Xia, T.; Kovochich, M.; Liong, M.; Zink, J. I.; Nel, A. E. ACS Nano 2008, 2, 85–96. (16) Goodman, C. M.; McCusker, C. D.; Yilmaz, T.; Rotello, V. M. Bioconjugate Chem. 2004, 15, 897–900. (17) Wilhelm, C.; Billotey, C.; Roger, J.; Pons, J. N.; Bacri, J. C.; Gazeau, F. Biomaterials 2003, 24, 1001–1011. (18) Ryman-Rasmussen, J. P.; Riviere, J. E.; Monteiro-Riviere, N. A. Nano Lett. 2007, 7, 1344–1348. (19) Wang, B.; Zhang, L.; Bae, S. C.; Granick, S. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18171–18175. (20) Yu, Y.; Anthony, S. M.; Zhang, L. F.; Bae, S. C.; Granick, S. J. Phys. Chem. C 2007, 111, 8233–8236. (21) Li, Y; Gu, N. J. Phys. Chem. B 2010, 114, 2749–2754. (22) Patil, S.; Sandberg, A.; Heckert, E.; Self, W.; Seal, S. Biomaterials 2007, 28, 4600–4607. (23) Tanaka, M.; Sackmann, E. Nature 2005, 437, 656–663. (24) Richter, R. P.; Remi Berat, R.; Brisson, A. R. Langmuir 2006, 22, 3497–3505. (25) Richter, R. P.; Brisson, A. R. Biophys. J. 2005, 88, 3422–3433. (26) Hishida, M.; Setoa, H.; Kaewsaiha, P.; Matsuokab, H.; Yoshikawa, K. Colloids Surf., A 2006, 284-285, 444–447. (27) Park, J. J.; Lacerda, S. H. D. P.; Stanley, S. K.; Vogel, B. M.; Kim, S.; Douglas, J. F.; Raghavan, D.; Karim, A. Langmuir 2009, 25, 443–450. (28) Hook, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73 (24), 5796–5804. (29) Gaponik, N.; Talapin, D. V.; Rogach, A. L.; Hoppe, K.; Shevchenko, E. V.; Kornowski, A.; Eychmu1ller, A.; Weller, H. J. Phys. Chem. B 2002, 106, 7177–7185. (30) Rogach, A. L.; Franz, T.; Klar, T. A.; Feldmann, J.; Gaponik, N.; Lesnyak, V.; Shave, A.; Eychmu1ller, A.; Rakovich, Y. P.; Donegan, J. F. J. Phys. Chem. C 2007, 111, 14628–14637. (31) Parimi, S.; Barnes, T. J.; Prestidge, C. A. Langmuir 2008, 24, 13532–13539. (32) Mecke, A.; Majoros, I. J.; Patri, A. K.; Baker, J. R.; Holl, M. M. B.; Orr, B. G. Langmuir 2005, 21, 10348–10354. (33) Zimmermann, R.; K€uttner, D.; Renner, L.; Kaufmann, M.; Zitzmann, J.; M€uller, M. Biointerphases 2009, 4, 1–6. (34) McManus, J. J.; R€adler, J. O.; Dawson, K. A. J. Phys. Chem. B 2003, 107, 9869–9875. (35) Ainalem, M.-L.; Kristen, N.; Edler, K. J.; Hook, F.; Sparr, E.; Nylander, T. Langmuir 2010, 26, 4965–4976. (36) Liu, J. W.; Stace-Naughtona, Z. A.; Brinker, C. J. Chem. Commun. 2009, 5100–5102. (37) Cauda, V.; Engelke, H.; Sauer, A.; Arcizet, D.; Brauchle, C.; R€adler, J.; Thomas, B. T. Nano Lett. 2010, 10, 2482–2492. (38) Jurkiewicz, P.; Olzynska, A.; Langner, M.; Hof, M. Langmuir 2006, 22, 8741–8749. (39) Gurtovenko, A. A.; Patra, M.; Karttunen, M.; Vattulainen, I. Biophys. J. 2004, 86, 3461–3472. (40) Mailander, V.; Landfester, K. Biomacromolecules 2009, 10, 2379–2400. (41) Clift, M. J. D.; Rothen-Rutishauser, B.; Brown, D. M.; Duffin, R.; Donaldson, K.; Proudfoot, L.; Guya, K.; Stone, V. Toxicol. Appl. Pharmacol. 2008, 232, 418–427. (42) Paul, S.; Paul, D.; Basova, T.; Ray, A. K. J. Phys. Chem. C 2008, 112, 11822–11830. 2534

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535

Langmuir

ARTICLE

(43) Mengistu, D. H.; Bohinc, K.; May, S. J. Phys. Chem. B 2009, 113, 12277–12282. (44) Michanek, A.; Kristen, N.; H€o€ok, F.; Nylander, T.; Sparr, E. Biochim. Biophys. Acta 2010, 1798, 829–838. (45) Laurencin, M.; Georgelin, T.; Malezieux, B.; Siaugue, J.-M.; Menager, C . Langmuir 2010, 26, 16025–16030.

2535

dx.doi.org/10.1021/la104449y |Langmuir 2011, 27, 2528–2535