Nanoviscosity Measurements Revealing Domain Formation in

Jan 11, 2017 - A common notion is to ascribe domain separation to thermodynamic phase equilibria. However, characterizing thermodynamic phases of sing...
1 downloads 6 Views 2MB Size
Article pubs.acs.org/ac

Nanoviscosity Measurements Revealing Domain Formation in Biomimetic Membranes Imad Younus Hasan and Adam Mechler* La Trobe Institute for Molecular Science, La Trobe University, Melbourne, Victoria 3086, Australia S Supporting Information *

ABSTRACT: Partitioning of lipid molecules in biomimetic membranes is a model system for the study of naturally occurring domains, such as rafts, in biological membranes. The existence of nanometer scale membrane domains in binary lipid mixtures has been shown with microscopy methods; however, the nature of these domains has not been established unequivocally. A common notion is to ascribe domain separation to thermodynamic phase equilibria. However, characterizing thermodynamic phases of single bilayer membranes has not been possible due to their extreme dimensions: the size of the domains falls to the order of tens to hundreds of nanometers whereas the membrane thickness is only a few nanometers. Here, we present direct measurements of phase transitions in single bilayers of 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC)/1,2-dipalmitoyl-sn-glycero-3phosphocholine (DPPC) phospholipid mixtures using quartz crystal microbalance-based nanoviscosity measurements. Coexisting thermodynamic phases have been successfully identified, and a phase diagram was constructed for the single bilayer binary lipid system. It was demonstrated that domain separation only takes place in planar membranes, and thus, it is absent in liposomes and not detectable in calorimetric measurements on liposome suspensions. On the basis of energetic analysis, the main transition was identified as the breaking of van der Waals interactions between the acyl chains.

C

known about the physical and thermodynamic factors that drive the formation of such domains. Domain separation in bilayer membranes is frequently correlated to bulk (calorimetric) phase diagrams of the same lipid mixture. The phase behavior of hydrated phosphatidylcholine (PC) membranes is the basis of studies of mixed lipid systems as PC lipids are ubiquitous in eukaryotic membranes.5 Differential scanning calorimetry (DSC) of hydrated neat saturated PC lipids of 15−22 carbon atom long acyl chains identified three phase transitions in the temperature range of 10−80 °C; the lipid phases are frequently compared to liquid crystals. At very low temperatures, all lipid molecules are in crystalline dehydrated state (Lc) in which the hydrocarbon chains pack very tightly and chain motion is extremely restricted. At “sub transition temperature”, the polar headgroup becomes hydrated leading to a large increase in the interfacial area that exerts a rotational excitation on the hydrocarbon tails, leading to the less tightly packed gel phase Lβ. With further increase of the temperature, the degree of hydration does not change; however, the long axis rotational rates of both the hydrocarbon chains and the polar head groups increase. This at the “pre-transition temperature” yields the ripple phase (Pβ) in

ell membranes are complex mixtures of lipids, proteins, and carbohydrates, of which lipids provide the structural platform.1 The “fluid-mosaic membrane model” in use since 1972 describes the cell membrane as a two-dimensional solution,2 originally assuming that the protein solutes float in a homogeneous lipid “solvent”. Evidence of barriers of diffusion lead to a quick reinterpretation of the model, emphasizing that it is a mosaic of domains of different molecular mobility.3 Understanding the nature of these domains has turned out to be a major challenge, attracting unwavering attention over the past four decades. The best known example of domain formation in cellular membranes is the “raft” structure. Rafts are microscopic domains rich in sphingolipids, glycerophospholipids, and cholesterol, leading to enhanced affinities toward integral and peripheral membrane proteins.4 Thus, rafts play a fundamental role in membrane trafficking and signaling.5 This biological function has fuelled interest in the partitioning of mixed lipid membranes. Consecutively, membrane domains other than rafts have been also observed in biological6 as well as biomimetic membranes. 7 These domains have distinct morphological, mechanical, and optical properties; hence, they have been identified with a range of analytical methods including atomic force microscopy,8 near-field scanning optical microscopy,9 high-resolution secondary ion mass spectrometry,10 ellipsometry,11 and fluorescence fluctuation spectroscopy.12 However, in spite of the ample empirical data, little is © XXXX American Chemical Society

Received: October 31, 2016 Accepted: January 11, 2017 Published: January 11, 2017 A

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

these domains have not been identified thus far due to the challenge of performing calorimetry on unilamellar membranes. Recently, measurements with Quartz Crystal Microbalance “with Dissipation” (QCM) were used to identify phase transition temperatures of supported lipid vesicles42−45 and single bilayer membranes14,19,20 with high accuracy from the viscoelastic changes upon chain melting. Using this method, cholesterol rich domains have been identified in nontensioned “flat” unilamellar supported membranes; these domains were absent in sonicated liposomes of the same composition, suggesting that domain separation is affected by membrane curvature tension.46 Hence, in this work, we compare the phase transition temperatures of neat and mixed lipids in three different model systems: suspended MLVs by differential scanning calorimetry (DSC), supported unilamellar liposomes with QCM, and supported single lipid bilayer also with QCM to highlight the relation between membrane curvature tension and domain separation.

which the polar head groups have almost free rotation and the hydrocarbon chains pack into a hexagonal lattice. Finally, the ripple phase converts to liquid-crystalline phase (Lα) as a result of “chain melting” at the so-called “main transition temperature” (Tm).6 PC lipids of linear saturated acyl chains with 14 carbon atoms exhibit only the pretransition and main transition,7 whereas PC lipids with 10 to 13 carbon atoms in the acyl chains have only a single, direct transition from crystalline dehydrated phase to liquid-crystalline phase.8 PC lipids with less than 9 carbon atoms do not show any phase transitions at all.9 There is some ambiguity in the literature about the actual values of these phase transition temperatures, likely due to the lack of a consensual reference state. In a hydrated sample, depending on the method of preparation and the amount of excess water, phospholipids form bulk lamellar phases as well as single- and multilamellar vesicles (liposomes) of various sizes. In calorimetry measurements, multilamellar lipid vesicles (MLVs) usually exhibit a narrow main transition peak whereas unilamellar vesicles exhibit a broad main transition peak at slightly lower temperatures than MLVs.13,14 Furthermore, MLVs have a clear, albeit broad, pretransition that is absent in unilamellar vesicles of the same composition. For example, multilamellar 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC) MLVs have two prominent phase transition peaks at 34.6 and 41.2 °C while DPPC small unilamellar liposomes exhibit a single transition at 37 °C.15,16 The size of unilamellar vesicles is another factor that affects the phase transitions; e.g., DPPC vesicles of diameters less than 35 nm show a chain melting transition at ∼37 °C that increases with increasing vesicle size and is largely identical to the “bulk” phase transition temperature in large unilamellar and multilamellar vesicles.17,18 This shift is likely related to membrane curvature tension as the molecular self-assembly model as well as molecular dynamics simulation results predict that increasing the membrane tension by 1 mN/m will decrease the temperature of the gel−liquid crystalline transition for DPPC by ∼1.0 K.19,20 Hence, calorimetric measurements depend on the structure and geometry of the lipid assembly and that accounts for the broad variation of phase transition data reported in the literature. In mixed lipids of differing acyl chains, equilibria persist between different phases over a broad temperature range as it is described in phase diagrams21−35 However, many uncertainties and variances accompany the data.36,37 In unilamellar systems, mixed membranes can laterally separate into different domains that have distinct composition; these domains may exhibit longrange order, such as circular, stripe, or even ring geometries.38 For example, saturated DPPC mixtures with unsaturated 1,2dioleoyl-sn-glycero-3-phosphatidyl serine (DOPS) or -choline (DOPC) mix homogeneously in buffer; however, DPPC rich domains form once divalent ions are introduced.39 There is correlation between domain composition and bulk lipid phases. However, additional factors also have to be considered. Most importantly, domain composition depends on the local surface curvature.38 In two component giant lipid vesicles of DOPC and DPPC, formation of membranes domains was reported in response to changing membrane tension.40 Contemporarily, Parikh and co-workers imaged oscillatory domains in semipermeable giant lipid vesicles where domain separation correlated to membrane tension modulations through swell− burst lytic cycles.41 However, the thermodynamic phases of



MATERIALS AND METHODS Buffer Preparation. Twenty mM phosphate buffer solution containing 100 mM sodium chloride (Merck) at pH 6.6 was used in all experiments. Potassium dihydrogen phosphate (KH2PO4) and dipotassium hydrogen phosphate (K2HPO4) were purchased from Fluka (Switzerland) at ACS Reagent grade. Ultrapure water purification system (Sartorius AG, Germany) was used to provide 18.2 MΩcm deionized water for all solutions. Preparation of Liposomes. 1,2-Dimyristoyl-sn-glycero-3phosphocholine (DMPC) and 1,2-dipalmitoyl-sn-glycero-3phosphocholine (DPPC) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Chloroform (spectrophotometric grade >99.8%, Sigma−Aldrich, Castle Hill, NSW, Australia) was used to create individual stock solutions by dissolving lyophilized DMPC and DPPC in chloroform. Dry films were formed in round-bottom test tubes by evaporating the solvent upon continuous vortexing under a stream of dry nitrogen. The lipid layers were hydrated in the assay buffer (20 mM phosphate buffer containing 100 mM NaCl at pH 6.6). All the liposome suspensions used in differential scanning calorimetry measurements were prepared by 30 min incubation at 37 °C, followed by brief vortexing and 30 s sonication to yield a mixed population of unilamellar and multilamellar vesicles (ULVs and MLVs, respectively) in an equilibrium size distribution.47,48 The liposomes used in the QCM experiments to form intact liposome layers as well as flat lamellar lipid bilayers on gold surface were hydrated at 60 °C in a water bath for 30 min and then vortexed and sonicated for 30 min to yield mostly ULVs. For fluorescence microscopy imaging, liposomes were colabeled by mixing lipophilic 0.1% DMPE-atto-594 (λfl = 627 nm, ATTO-TEC, Germany) with the lipids in an organic solvent; liposomes formed from this mixture were loaded with 10 mM 5(6)-carboxyfluorescein (CF) (λfl = 517 nm, SigmaAldrich) solution by rehydrating the dry lipids in the aqueous phosphate buffered saline (PBS) solution of the fluorophore. The excess 5(6)-carboxyfluorescein was removed via dialysis. For tonicity measurements, DMPC/DPPC (70:30) liposomes were loaded with 200 mM glucose (>99.5%, Sigma-Aldrich, Castle Hill, NSW, Australia) solution; the excess glucose was removed via dialysis. QCM Sensor Surface Modification. To oxidize the gold surfaces, the chips were placed into 1:1:3 mixture of ammonium hydroxide solution (28%), hydrogen peroxide B

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry solution (30%), and water, at 70 °C for 15−20 min. The chemicals were purchased from Sigma-Aldrich (Castle Hill, NSW, Australia). After the oxidative treatment, the chips were thoroughly rinsed with ultrapure water. For preparing vesicular supported layers, the chips were dried under a gentle stream of nitrogen gas and immediately assembled into the QCM chamber. For preparing partially supported bilayers, the chips were placed in 3-mercaptopropionic acid (MPA) (HPLC grade >99%, Fluka, Switzerland) solution overnight to give sufficient time for the formation of a self-assembled monolayer on the gold surface. After functionalization, the chips were rinsed with isopropanol and dried under a gentle stream of nitrogen gas and assembled into the QCM chamber. Quartz Crystal Microbalance Experiments. All experiments were performed in a Q-SENSE E4 system (Q-SENSE, Sweden) using AT-cut gold coated quartz sensors (chips) with a fundamental resonance frequency of 5 MHz. The frequency shift (Δf) and energy dissipation (ΔD) were recorded simultaneously at five eigenmodes of the sensor (3, 5, 7, 9, and 11th) corresponding to 15, 25, 35, 45, and 55 MHz, respectively. However, all values reported in this work for Δf and ΔD are that of the fifth overtone (25 MHz) for liposome deposition and ninth overtone (45 MHz) for temperature sweeping experiments. The raw data was analyzed using QTools (Q-SENSE) and OriginPro 9 (Origin Lab, USA) software. All deposition experiments were performed by pumping 1 mL of the liposome suspension (containing 10 μmole lipid) through the QCM cell at 50 μL/min flow rate followed by rinsing with buffer solution until a stable (constant) signal was reached in both frequency and dissipation. All depositions on MPA modified gold have been performed at temperatures higher than the main phase transition temperatures of the respective lipid mixtures in order to promote liposome rupture and reduce the residual amount of intact liposomes. The quality of the deposit was confirmed from the signals as described previously.47,49 In some cases, high dissipation signal was observed after deposition, that is the sign of the presence of residual liposomes;47 these can be removed by applying osmotic stress,49−58 and hence, 500 μL of deionized water was flushed in the chamber that forms a largely defect-free lamellar membrane as established in a previous work.59 For more technical details, see Figure S1. For all temperature sweeping experiments, temperature was initially maintained constant at 15 °C for 30 min; then, it was increased at a rate of 0.2 °C/min to 45 °C. Temperature was maintained constant at 45 °C for 30 min after which it was decreased at the same rate as before to 15 °C. This cycle was repeated up to two times. Generally, the second cycle gave a clearer trend for the dissipation signal, and therefore, the second and third cycles were used for analysis. Differential Scanning Calorimetry. Experiments have been performed in a SETARAM μDSC Evo3 instrument, using 700 μL hastelloy pressure cells. In each experiment, three zones were recorded with the same parameters. 50 μmol of lipid suspended in PBS was loaded into the batch cell against a reference of the same volume of PBS. The temperature profile of each zone was as follows: the system was equilibrated at 15 °C for 30 min; then, the temperature was increased at a rate of 0.33 °C min−1 to 35 °C; next, the system was equilibrated for 30 min at 35 °C, and then, the temperature was returned to 15 °C at a rate of 0.33 °C min−1 followed by 30 min equilibration. Fluorescence Microscopy Imaging. Experiments were performed with a Nikon Eclipse TM100 inverted epifluor-

escence microscope, using a UV lamp excitation source. Images were taken of the surface of the QCM sensor chips after membrane deposition in the Q-SENSE E4 system. The chips were placed on top of a buffer filled thin chamber on glass microscope slides (Figure S2). Dynamic Light Scattering Experiments. Liposome size measurements were performed by the dynamic light scattering (DLS) technique using a Zetasizer Nano ZS instrument (Malvern Instruments Ltd., Malvern Worcestershire, UK). All experiments were performed at 25 °C and repeated at least five times, with different solutions. For the analysis of the data, the non-negative least squares method was used after the polydispersity index of the cumulant fit was confirmed to be less than 0.7.



RESULTS AND DISCUSSION Lipid Deposition. Liposome deposition method was used to form intact liposome layers as well as planar lipid bilayer membranes on a gold surface.46,47,49,55 When in contact with a liposome suspension, the roughness and hydrophobic nature of bare metallic gold inhibit the formation of a continuous supported lipid membrane; thus, liposomes attach to the surface intact.54,60,61 However, modifying the gold surface with 3-mercaptopropionic acid (MPA) promotes the formation of fused lamellar bilayers. QCM was used to monitor the process of lipid deposition on oxidized gold and an MPA modified gold surface. Figure 1 shows the deposition of neat DMPC (top panels) and DPPC (bottom panels) on oxidized gold (left panels) and MPA modified gold (right panels).

Figure 1. Sensograms of frequency change (blue line) and energy dissipation (red line) for the 5th eigenmode of the sensor chip resonance for liposome deposition of neat DMPC and DPPC on oxidized gold (left panels) and MPA modified gold (right panels) at 25 °C.

A few seconds after injecting the liposomes into the measurement chamber, a negative frequency change was observed and simultaneously the dissipation increased for all lipid mixtures and both surfaces. On MPA modified gold, the frequency shift (Δf n=5) stabilized at −16 and −20 Hz and dissipation change (ΔD) at 2.7 and 4.0 arb. units for DMPC and DPPC, respectively, were consistent with previous observations.47,49 As expected, on oxidized gold, Δf and ΔD curves stabilized at higher values in comparison to MPA modified gold; the deposition of DMPC and DPPC yielded Δf n=5 −23 and −56 Hz and ΔDn=5 7.7 and 9.5 arb. units, respectively. These results clearly demonstrate that the deposition of liposomes on oxidized gold lead to the formation C

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Lipid Phase Transition Measurements. Lipid phase transition temperatures for neat and mixed lipids were measured in three different model systems: in suspended multilamellar liposomes by using differential scanning calorimetry (DSC), in supported unilamellar vesicles using QCM, and in supported single lipid bilayers also with QCM. Phase Transition of Suspended Multilamellar Liposomes. In DSC measurements (Figure 3), neat DPPC produced two

of a highly dissipative layer of intact vesicles but modification of the surface with MPA promoted liposome rupture and fusion continuously to form single bilayer membranes. Nearly identical sensograms have been observed for depositing mixtures of DMPC and DPPC in different ratios. Table 1 summarizes the final Δf and ΔD values. Variations in the frequency and dissipation change values measured for Table 1. Final Values of Frequency and Energy Dissipation Signals for the Lipid Mixtures Used in This Study substrate oxidized gold lipid mixture (DPPC/DMPC)

T (°C)

0−100 100−0 90−10 70−30 50−50 30−70 10−90

15 25 25 25 25 25 25

MPA-modified gold

Δf (Hz)

ΔD (arb. unit)

−29 −56 −39.1 −120.7 −199.4 −68.9 −18

8.4 9.5 11.2 38.5 33.6 13.56 7.1

T (°C)

Δf (Hz)

ΔD (arb. unit)

25 42 42 42 42 42 42

−16 −11.2 −8.5 −9.2 −9.5 −10.1 −10.8

2.7 1.8 1.55 1.65 1.8 1.9 1.9

Figure 3. (A) Differential scanning calorimetry measurements for DPPC: a, neat DPPC; b, DPPC/DMPC (90:10); c, DPPC/DMPC (70:30); d, DPPC/DMPC (50:50); e, DPPC/DMPC (70:30); f, DPPC/DMPC (10:90); h, neat DMPC. (B) The latent heat of the main transition measured from the DSC peaks in (A).

continuous membrane coverage of different compositions are caused by the difference in their viscoelastic properties.59,62 When osmotic stress was used to promote liposome rupture,59 QCM results showed that the liposomes first swell, as indicated by a slight increase in the dissipation, followed by a sharp increase in frequency parallel with a sharp decrease in the dissipation signal, indicating mass loss as water is released from the ruptured liposomes. The two signals reach stable values within a few minutes. During this process, the deposit changes from highly dissipative to semirigid characteristics (Figure S1). Fluorescence microscopy was used to confirm the morphology of the deposits. Examples are shown in Figure 2. Intact

peaks: a very sharp peak at 41.37 °C and a small broad peak at 35.4 °C that are consistent with the main and pretransition temperatures of DPPC. Adding 10% of DMPC shifts the main and pretransition peaks to 40.2 and 32.3 °C, respectively; the peak broadened but the integral area of the peak (latent heat) did not change substantially (Figure 3B). By increasing the mole percentage of DMPC, the main transition peak shifted to lower temperatures and broadened further. The DPPC pretransition peak disappeared at 50% mixing ratio (Figure 3A, trace d). However, when the mole percentage of DMPC increased to more than 50%, the main transition peak again became sharper, shifting further toward lower temperatures, eventually reaching 25.18 °C for 90% DMPC and 23.6 °C for the neat DMPC lipid (Figure 3A, trace h). There is a liner relation between the latent heat of the main transition and the percentage of DPPC lipid: neat DMPC requires 21.66 kJ/mol to “melt” the hydrocarbon chains while neat DPPC needs 37.44 kJ/mol. These values are consistent with literature reports.14,23 Phase Transitions of Highly Vesicular Deposits on Oxidized Gold. Δf and ΔD sensograms of the effect of temperature sweeping (after subtraction of the reference channels) are shown in Figure 4A (data for all lipid compositions are shown in Figure S4). It can be seen that the steep changes take place in a narrow temperature range, consistent with the “chain melting” of the lipids as described before.46,49 The negative temperature sweep mirrored the trend of the positive sweep for both parameters, and the trends are highly reproducible. According to the previously established method,46,49 the temperature at which the frequency and dissipation changes are the steepest corresponds to the phase transition temperature, identified by plotting the first derivative of the sensor signals against time (that corresponds to temperature), as shown in Figure 4B. Viscoelastic changes are primarily seen in the dissipation signal; however, due to the interdependence of the frequency and dissipation channels in a viscoelastic system, changes are also observed in the frequency signal that is shown here for comparison. For this mixture, welldefined peaks are observed at ∼31.6 °C. It was also observed that the size distribution of the liposomes has a profound effect

Figure 2. Examples of lipid deposits imaged with fluorescence microscopy. Liposomes were labeled with a lipid conjugated atto-596 (red) dye and loaded with carboxyfluorescein. (A) Liposomes deposited on oxidized gold at 25 °C; (B) membrane deposited onto MPA at 42 °C (residual liposomes are visible); (C) membrane deposited onto MPA at 25 °C and washed with water (residual liposomes mostly removed).

liposomes are seen on the oxidized gold surface (in yellow color due to the colocalization of the red membrane label and green CF dye loaded into the liposomes, Figure 2A), whereas on MPA-modified gold a continuous bilayer membrane is formed as proven by the continuous red background (Figure 2B), albeit with a significant number of intact liposomes still observable. Upon washing with deionized water, these liposomes either rupture or are washed away and a continuous membrane is formed (Figure 2C). In mixed membranes, occasionally labelenriched domains have been observed, likely due to colocalization of the label with the lipid of the same acyl chain length (Figure S3). D

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. Effect of temperature ramping on the f and D signals in the case of DPPC/DMPC (50:50) intact liposomes. The gray line indicates the temperature profile; blue line is Δf; red line refers to ΔD. (A) Δf and ΔD sensograms for DPPC/DMPC intact liposomes after subtraction of the reference; (B) first-order time derivatives of Δf and ΔD; (C−F) Gaussian fitting of dF/dT and dD/dT results to identify the phase transition temperature.

on the observed peak, and hence, in these experiments, uniform sized liposomes were used (Figures S5 and S6). Gaussian peak fitting was applied to analyze dF/dT and dD/ dT results. Figure 4 shows an example of the phase transition of DPPC/DMPC (50:50) intact liposomes. Both dD/dT and dF/ dT give sharp peaks at 31.6 °C that have a clear Gaussian profile (Figure 4C,E). These peaks are consistent with the phase transition temperature of 50% DPPC/50% DMPC from the gel (Lβ) to the liquid-crystalline (Lα) phase. The cooling peaks were observed with a hysteresis at 29.6 °C (Figure 4D,F). Figure 5 (left panels) shows the first-order time derivatives for the dissipation change upon temperature sweeping for unilamellar liposomes of DPPC/DMPC mixtures of different ratios. Focusing on the main transition peaks, neat DPPC exhibits a well-defined peak at 39.8 °C (the average of the values from the positive and negative temperature sweep to cancel the hysteresis effect); this value is somewhat lower than the gel−liquid phase transition of neat DPPC recorded by DSC which is 41.3 °C according to our measurements and the literature.23 The difference is consistent with the different “state” of ULVs in QCM and MLVs in the DSC measurements, the latter forming a relatively broad (0.4 PDI) population of ∼150 nm diameter vesicles as shown by DLS (Table 2). The absence of pretransition in unilamellar vesicles is explained with the high membrane tension that inhibits membrane undulations.44 However, an equally feasible explanation is that the “ripple” phase transition is the result of breaking interbilayer interactions in bilayer stacks of MLVs and bulk phases that is absent in ULVs. When introducing DMPC, the main phase transition temperature shifts to lower temperatures, reaching 38.1, 35.8, 31.7, 28.6, and 24.8 °C for 10, 30, 50, 70, and 90 mol percentages of DMPC, respectively. The main transition of neat

Figure 5. First-order time derivatives for the dissipation change (red solid line) upon temperature sweeping (blue dashed line) of (left) vesicular deposits and (right) supported bilayers of the same DPPC/ DMPC mixtures.

DMPC is seen at 22.6 °C. These results show that the two lipids mix mostly homogeneously in vesicles; however, we observed double transition peaks in DPPC/DMPC (90:10) and (70:30) (Figure 5 asterisk) that likely correspond to two liposome populations of different sizes, the presence of which was confirmed by DLS (Table 2). It has been reported before that the size of the liposomes has a direct influence on the value of gel−liquid phase transition in the case of high surface curvature (i.e., in small unilamellar liposomes).14,44 However, the effect has not been observed for equilibrium size liposomes (often described as large unilamellar liposomes, LUVs), likely due to the very small phase transition temperature variations that size changes introduce in this range. E

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry Table 2. Chain Melting Transition Temperatures and DLS Results for DPPC/DMPC Mixtures main phase transition temperatures, °C lipid mixture (DPPC/DMPC)

suspended liposomes

highly vesicular layer

100−0 90−10 70−30 50−50 30−70 10−90 0−100

41.3 40.2 36.7 32.5 29.4 25.2 23.6

39.8 37.4 and 38.1 34.9 and 35.1 31.7 28.6 24.8 22.6

DLS measurements

supported lipid bilayer 39.4 23.1, 23.5, 24.7, 23.0, 20.3, 22.5

34.8, 30.5, 31.7, 28.9, 24.5,

and 41.2 34.8, and 41.5 and 41.5 33.5, and 42.3 31.0, 37.9, and 40.9

size (nm)

PDI

150 82 (36%) and 400 (46%) 75 (32%) and 250 (68%) 120 94 108 150

0.44 0.25 0.29 0.28 0.28 0.26 0.4

Phase Transitions of Supported Single Bilayer Membranes. After forming supported single bilayer membranes on MPA-modified gold, temperature sweeping experiments were performed by applying the same temperature profile as in the previous section. Complex peak structures are seen in the dD/ dT plots, whereas the dF/dT plots yielded unclear results. An example of the primary data and the analysis are shown in Figure 6 for DPPC/DMPC (50:50). Gaussian fitting was Figure 7. Phase diagram of the DPPC/DMPC mixture for (A) suspended and supported liposomes where phase transition temperatures were measured by DSC (triangles) and QCM (circles); (B) supported bilayer membranes of the same DPPC/DMPC mixtures based on QCM measurements.

However, when the temperature exceeds the main transition temperature for DMPC, chain melting is observed in the DMPC rich domains that depends very slightly on the mixing ratio (squares). A viscoelastic change is also observed at ∼41 °C in all DPPC containing mixtures which indicates the presence of DPPC rich domains (triangles). Between these two phase transition temperatures, a third peak is seen, the position of which depends on the mixing ratio of the two lipids (circles). Hence, the results reveal the coexistence of three domains: DMPC rich, DPPC rich, and a mixed domain. For example, in the case of DPPC/DMPC (50:50), the middle peak is observed at 31.7 °C which is the same temperature that is observed as the only phase transition temperature of liposomes of the same composition. Domain Separation Is Inhibited by Curvature Tension. In the three cases examined, domain separation has only been detected in planar membranes. However, domain formation was reported previously in giant unilamellar vesicles (GUVs) upon osmotic stress.41 In GUVs, the curvature tension is negligibly small, and thus, in their equilibrium state, the properties of GUV membranes are closer to the planar membranes used in the present study. To establish whether osmotic pressure will lead to domain separation in the MLVs as well, DSC experiments were also performed in hyper- and hypotonic glucose solutions. The results indicate the absence of any domain separation under either of the conditions (Figure S7). Thus, domain separation in MLVs is inhibited by membrane curvature. Main Transition of Phospholipid Membranes. The phase transition temperatures show a linear dependence on the DPPC/DMPC lipid ratio in the homogeneously mixed liposome samples (Figure 8). Noticeably, the DPPC/DMPC (50:50) mixture exhibits a phase transition temperature that is equal to the average of neat DMPC and DPPC values (31.3 °C). Incidentally, it is also close to the phase transition of 15:0 PC lipid,23 and the latent heat of the main transition equals 28.87 kJ/mol which is almost the same as the latent heat of the

Figure 6. Effect of temperature ramping on the frequency (blue line) and dissipation (red line) signals in the case of DPPC/DMPC (50:50) supported bilayer. The gray line indicates the temperature profile. (lower panels) Gaussian fitting of dD/dT results for increasing temperature (15 to 45 °C) and decreasing temperature (45 to 15 °C), respectively.

applied to analyze dD/dT results. When the temperature starts increasing, dD/dT shows a sharp peak at 16.4 °C (DMPC subtransition); another peak is observed at 18.8 °C that is consistent with the subtransition temperature for DPPC (the literature reports DPPC subtransition at ∼18 °C63). dD/dT shows further peaks at 26.5 and 31.7 °C; these peaks are consistent with the phase transition temperatures of two different domains of the mixture: one rich in DMPC lipid and the other containing an equal mole ratio of DMPC and DPPC. A small peak is observed at 41.5 °C that is interpreted as the phase transition temperature of DPPC rich domains. With a hysteresis, the same peak positions were observed for the negative temperature sweep. Results of the analysis for all DMPC/DPPC mixtures are summarized in Table 2. From these results, it is apparent that the lipids form nanodomains of different compositions. The results allow the construction of a phase diagram for supported DMPC/DPPC lipid bilayers (Figure 7 B). When the temperature is lower than the main transition temperature for DMPC, all domains are in the gel state (I). F

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

analysis of the calorimetry results and literature data, the main transition was identified as the breaking of van der Waals interactions between the acyl chains.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.6b04256. Liposome deposition (technical details); fluorescence microscopy (technical details); data for fluorescence microscopy; raw data for QCM temperature ramping experiments; dynamic light scattering; effect of liposome size on phase transition temperatures; effect of tonicity on phase transition temperatures (PDF)

Figure 8. Relationship between the number of carbon atoms per headgroup pair in the hydrocarbon core and the main transition temperature. ■ for symmetric saturated diacyl phosphatidylcholines that have 13, 14, 15, 16, 17, and 18 carbon atoms in their acyl chains;64 ● measured in this work by DSC for DMPC/DPPC mixtures; ▲ measured in this work by QCM for DMPC/DPPC mixtures; ▼ for asymmetric mixed acyl phosphatidylcholines in which the chain at 1 position is longer than in 2 position; ∗ for asymmetric mixed acyl phosphatidylcholines in which the chain at 2 position is longer than in 1 position.54,65



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Adam Mechler: 0000-0002-6428-6760 Author Contributions

main transition of 15:0 PC lipids.66 From this observation, we hypothesize that the main lipid phase transition temperature of saturated phospholipids depends on the number of carbon atoms in the hydrocarbon core and only that. This presupposes a linear relationship between the number of carbon atoms in the hydrophobic core (i.e., 4 acyl chains between two headgroups) and the phase transition temperature, which was confirmed by comparing literature data on chain melting temperatures for a range of saturated lipids as shown in Figure 8. The dependence of the phase transition temperature on the alkyl chain length suggests that the main transition temperature is a function of the van der Waals bond energies between the chains. This allows the formulation of a simple model in which the “chain melting” is in fact the point at which the Boltzmann energy overcomes the attraction between the lipid molecules. In a van der Waals crystal, this is the point of vaporization; lipids however still remain in a bilayer due to hydrophobic forces in the aqueous environment. According to our DSC measurements on free liposomes and literature data,14 the latent heats of the main transitions of DMPC and DPPC are 37.44 and ∼21.66 kJ/mol, respectively. The difference in carbon atoms in the bilayer between the headgroups is 8 (two per acyl chain in two opposing molecules), and hence, according to the proposed model, the effective molar interaction energy per methylene groups is ∼2 kJ/mol. This value is very close to the calculated molar interaction energy of ∼2.9 kJ/mol per methylene groups in alkyl-aromatic compounds.67 Hence, the energetic analysis supports the model.

The manuscript was written through contributions of both authors. Both authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS I.Y.H. expresses his special thanks to the Higher Committee for Education Development in Iraq (HCED) for supporting this work.



REFERENCES

(1) Petty, H. R. Molecular biology of membranes: structure and function; Springer Science & Business Media: New York, 2013. (2) Singer, S. J.; Nicolson, G. L. Science 1972, 175, 720−731. (3) Kusumi, A.; Nakada, C.; Ritchie, K.; Murase, K.; Suzuki, K.; Murakoshi, H.; Kasai, R. S.; Kondo, J.; Fujiwara, T. Annu. Rev. Biophys. Biomol. Struct. 2005, 34, 351−378. (4) Pralle, A.; Keller, P.; Florin, E.-L.; Simons, K.; Hörber, J. J. Cell Biol. 2000, 148, 997−1008. (5) Veatch, S. L.; Keller, S. L. Biophys. J. 2003, 85, 3074−3083. (6) de Almeida, R. F. M.; Borst, J.; Fedorov, A.; Prieto, M.; Visser, A. Biophys. J. 2007, 93, 539−553. (7) Johnston, L. J. Langmuir 2007, 23, 5886−5895. (8) Kraft, M. L.; Weber, P. K.; Longo, M. L.; Hutcheon, I. D.; Boxer, S. G. Science 2006, 313, 1948−1951. (9) Howland, M. C.; Szmodis, A. W.; Sanii, B.; Parikh, A. N. Biophys. J. 2007, 92, 1306−1317. (10) Stottrup, B. L.; Heussler, A. M.; Bibelnieks, T. A. J. Phys. Chem. B 2007, 111, 11091−11094. (11) Gandhavadi, M.; Allende, D.; Vidal, A.; Simon, S.; McIntosh, T. Biophys. J. 2002, 82, 1469−1482. (12) Simons, K.; Ikonen, E. Nature 1997, 387, 569−572. (13) Biltonen, R. L. J. Chem. Thermodyn. 1990, 22, 1−19. (14) Drazenovic, J.; Wang, H.; Roth, K.; Zhang, J.; Ahmed, S.; Chen, Y.; Bothun, G.; Wunder, S. L. Biochim. Biophys. Acta, Biomembr. 2015, 1848, 532−543. (15) Heimburg, T. Biochim. Biophys. Acta, Biomembr. 1998, 1415, 147−162. (16) Suurkuusk, J.; Lentz, B. R.; Barenholz, Y.; Biltonen, R.; Thompson, T. Biochemistry 1976, 15, 1393−1401.



CONCLUSIONS Domain separation in binary membranes has been successfully ascribed to equilibrium of thermodynamic phases. It was demonstrated that domain separation requires tensionless membranes, and thus, it is absent in liposomes and not detectable in calorimetric measurements on liposome suspensions. A phase diagram was constructed for single bilayer DMPC/DPPC membranes using quartz crystal microbalancebased nanoviscosity measurements. On the basis of energetic G

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry (17) Biltonen, R. L.; Lichtenberg, D. Chem. Phys. Lipids 1993, 64, 129−142. (18) Lichtenberg, D.; Freire, E.; Schmidt, C.; Barenholz, Y.; Felgner, P.; Thompson, T. Biochemistry 1981, 20, 3462−3467. (19) Uline, M. J.; Schick, M.; Szleifer, I. Biophys. J. 2012, 102, 517− 522. (20) Kong, X.; Qin, S.; Lu, D.; Liu, Z. Phys. Chem. Chem. Phys. 2014, 16, 8434−8440. (21) Huang, C. Biochemistry 1991, 30, 26−30. (22) Lewis, R. N.; McElhaney, R. N. Biochemistry 1985, 24, 2431− 2439. (23) Koynova, R.; Caffrey, M. Biochim. Biophys. Acta, Rev. Biomembr. 1998, 1376, 91−145. (24) Lewis, R.; Pohle, W.; McElhaney, R. N. Biophys. J. 1996, 70, 2736−2746. (25) Sunamoto, J.; Goto, M.; Iwamoto, K.; Kondo, H.; Sato, T. Biochim. Biophys. Acta, Biomembr. 1990, 1024, 209−219. (26) Ulrich, A. S.; Watts, A. Biophys. J. 1994, 66, 1441−1449. (27) Garidel, P.; Blume, A. Biochim. Biophys. Acta, Biomembr. 1998, 1371, 83−95. (28) Boggs, J. M.; Rangaraj, G.; Koshy, K. M. Chem. Phys. Lipids 1986, 40, 23−34. (29) De Almeida, R. F.; Fedorov, A.; Prieto, M. Biophys. J. 2003, 85, 2406−2416. (30) Elliott, R.; Szleifer, I.; Schick, M. Phys. Rev. Lett. 2006, 96, 098101. (31) Feigenson, G. W.; Buboltz, J. T. Biophys. J. 2001, 80, 2775− 2788. (32) Heberle, F. A.; Wu, J.; Goh, S. L.; Petruzielo, R. S.; Feigenson, G. W. Biophys. J. 2010, 99, 3309−3318. (33) Hsueh, Y. W.; Zuckermann, M.; Thewalt, J. Concepts Magn. Reson., Part A 2005, 26A, 35−46. (34) Ionova, I. V.; Livshits, V. A.; Marsh, D. Biophys. J. 2012, 102, 1856−1865. (35) Konyakhina, T. M.; Feigenson, G. W. Biochim. Biophys. Acta, Biomembr. 2016, 1858, 153−161. (36) Konyakhina, T. M.; Wu, J.; Mastroianni, J. D.; Heberle, F. A.; Feigenson, G. W. Biochim. Biophys. Acta, Biomembr. 2013, 1828, 2204−2214. (37) Marsh, D. Biochim. Biophys. Acta, Biomembr. 2010, 1798, 688− 699. (38) Baumgart, T.; Hess, S. T.; Webb, W. W. Nature 2003, 425, 821−824. (39) Dietrich, C.; Bagatolli, L.; Volovyk, Z.; Thompson, N.; Levi, M.; Jacobson, K.; Gratton, E. Biophys. J. 2001, 80, 1417−1428. (40) Chen, D.; Santore, M. M. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 179−184. (41) Oglęcka, K.; Rangamani, P.; Liedberg, B.; Kraut, R. S.; Parikh, A. N. eLife 2014, 3, e03695. (42) Goñi, F. M.; Alonso, A.; Bagatolli, L. A.; Brown, R. E.; Marsh, D.; Prieto, M.; Thewalt, J. L. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids 2008, 1781, 665−684. (43) Marsh, D. Biochim. Biophys. Acta, Biomembr. 2009, 1788, 2114− 2123. (44) Ahmed, S.; Wunder, S. L. Langmuir 2009, 25, 3682−3691. (45) Chen, D.; Santore, M. M. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 179−184. (46) Hasan, I. Y.; Mechler, A. Anal. Chem. 2016, 88, 5037−5041. (47) Mechler, A.; Praporski, S.; Piantavigna, S.; Heaton, S. M.; Hall, K. N.; Aguilar, M.-I.; Martin, L. L. Biomaterials 2009, 30, 682−689. (48) Claessens, M. M. A. E.; Leermakers, F. A. M.; Hoekstra, F. A.; Stuart, M. A. C. Langmuir 2007, 23, 6315−6320. (49) Hasan, I. Y.; Mechler, A. Soft Matter 2015, 11, 5571−5579. (50) Ipsen, J. H.; Karlstrom, G.; Mourtisen, O. G.; Wennerstrom, H.; Zuckermann, M. J. Biochim. Biophys. Acta, Biomembr. 1987, 905, 162− 172. (51) Marquês, J. T.; de Almeida, R. F.; Viana, A. S. Soft Matter 2012, 8, 2007−2016.

(52) Reviakine, I.; Simon, A.; Brisson, A. Langmuir 2000, 16, 1473− 1477. (53) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397−1402. (54) Reimhult, E.; Hook, F.; Kasemo, B. Langmuir 2003, 19, 1681− 1691. (55) Serro, A. P.; Carapeto, A.; Paiva, G.; Farinha, J. P. S.; Colaco, R.; Saramago, B. Surf. Interface Anal. 2012, 44, 426−433. (56) Steinem, C.; Janshoff, A.; Ulrich, W. P.; Sieber, M.; Galla, H. J. Biochim. Biophys. Acta, Biomembr. 1996, 1279, 169−180. (57) Burridge, K. A.; Figa, M. A.; Wong, J. Y. Langmuir 2004, 20, 10252−10259. (58) Cohen, F. S.; Akabas, M. H.; Finkelstein, A. Science 1982, 217, 458−460. (59) Hasan, I. Y.; Mechler, A. Biointerphases 2016, 11, 031017. (60) Korlach, J.; Schwille, P.; Webb, W. W.; Feigenson, G. W. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 8461−8466. (61) Hamada, T.; Kishimoto, Y.; Nagasaki, T.; Takagi, M. Soft Matter 2011, 7, 9061−9068. (62) Voinova, M. V.; Jonson, M.; Kasemo, B. Biosens. Bioelectron. 2002, 17, 835−841. (63) Zhao, F.; Cheng, X.; Liu, G.; Zhang, G. J. Phys. Chem. B 2010, 114, 1271−1276. (64) Lerebours, B.; Wehrli, E.; Hauser, H. Biochim. Biophys. Acta, Biomembr. 1993, 1152, 49−60. (65) Seitz, M.; Ter-Ovanesyan, E.; Hausch, M.; Park, C. K.; Zasadzinski, J. A.; Zentel, R.; Israelachvili, J. N. Langmuir 2000, 16, 6067−6070. (66) Lewis, R.; Mak, N.; McElhaney, R. N. Biochemistry 1987, 26, 6118−6126. (67) Mirgorod, Y. A. Russ. J. Phys. Chem. 2004, 78, 574−576.

H

DOI: 10.1021/acs.analchem.6b04256 Anal. Chem. XXXX, XXX, XXX−XXX