Supporting Information

Bavi et al. 10.1073/pnas.1409011111

SI Materials and Methods for the excised patch configuration (Fig. S1 B and C). All models
Liposome Preparation and Patch Fluorometry. Liposomes made of had a thickness of 3.5 nm (unless otherwise specified), having
azolectin [99.9% (wt/wt)] (P5638; Sigma) and rhodamine-PE isotropic and homogeneous material properties, for which the
[0.1% (wt/wt)] were prepared using a Nanion Vesicle Prep Pro. bending deformations are important. Hence, internal pressure
Briefly, 20 μL of 5 mM lipid dissolved in chloroform and 350 μL causes negligible stretching and shell permeability is less important
of 520 mM D-sorbitol were placed on indium tin oxide (ITO) to the deformation (7) (Fig. S1). We used shell theory for our
slides. An alternating electrical field of 5 Hz and 3 V was applied computational model, because a shell element can sustain bending
for 120 min to produce liposomes, which were stored at 4 °C. moments and maintain irregular geometry. Membrane theory
Images of creeping patch membranes were taken using a confocal failed to explain the deformation of the lipid bilayer under as-
microscope (LSM 700; Carl Zeiss) equipped with a long working pirating pressure for either cell-attached or excised patch con-
distance water immersion objective (×63; NA 1.15; Carl Zeiss). A figurations. This is because, by definition, a membrane cannot
555-nm laser was used to excite the fluorophore and creep of the sustain a bending moment. Thus, irregular undulations on the
patch membrane was detected using a long-pass 560-nm filter. membrane surface cannot remain stable without proper con-
Borosilicate glass pipettes (Drummond Scientific) were pulled straints. In fact, when membrane elements were used for typical
using a Flaming/Brown pipette puller (P-87; Sutter Instruments) patch geometries, we could not converge to a solution under any
and the tip of each pipette was cut with a microforge (Narishige; reasonable load. Further, even in very low pressures, the patch
MF-900) to a diameter of ∼2 μm. For visualizing the creep, the deforms like a bell rather than having a parabolic or hemispheric
tip was bent ∼30°, using the microforge to make it parallel to the shape (typical shapes that can be found in typical experiments).
bottom of the chamber when the pipette was mounted on a mi- The two-node SAX1 element was used, which is two-node stress/
cromanipulator (1). Negative pressure steps of −5 mmHg were displacement element that uses one point integration of the
generated by a High Speed Pressure Clamp-1 apparatus (HSPC-1; linear interpolation function for the distribution of loads (ABAQUS
ALA Scientific Instruments) and monitored by a piezoelectric 6.11-2). The FE model consisted of 297 and 112 linear axisym-
pressure transducer (PM015R; World Precision Instruments). To metric elements (SAX1) for the cell-attached and excised patch
remove adhesion tension, the pipette was filled with 0.1% BSA. configurations, respectively. Sensitivity to mesh density (total
After incubation for 30 min at room temperature, the pipette was number of elements and nodes in the computational model) was
washed several times with distilled water. also studied, meaning the original model was remeshed to obtain
meshes of different density. The results were seen to be in-
Electrophysiological Experiments. Liposomes were prepared by the dependent of mesh size, beyond the number of elements stated
dehydration/rehydration (D/R) method (1). Briefly, 2 mg of for each FE model. Our results were also independent of the type
azolectin lipids (P5638; Sigma) was dissolved in CHCl3, and ni- of element. For instance, the quadratic three-node element,
trogen gas was applied to form lipid films. After suspension in SAX2, could also be used. Although compared with SAX1 ele-
200 μL D/R buffer [200 mM KCl, 5 mM Hepes (pH 7.2, adjusted ments, a lower number of SAX2 elements are needed to converge
with KOH)], the solution was subjected to sonication for 5 min to the results, the chance of simulation abortion (due to severe
to make lipid clouds, and purified MscL protein was added in the distortion of elements) was seen to be much higher for SAX2. A
ratio of protein to lipid of 1:1,000 (wt/wt). D/R buffer was added finer mesh (maximum aspect ratio of ∼1:7) was used in the cur-
up to 3 mL and the mixture was incubated for 1 h, after which rent study to accommodate the highly curved geometry near the
biobeads (Bio-Rad) were added and incubated for a further 3 h to pipette tip and to avoid distortion of elements. Due to the axi-
remove the detergent. The solution was centrifuged at 250,000 × g symmetric feature of the problem, we modeled an axisymmetric
and the pellet was resuspended in 60 μL D/R buffer, spotted onto wire in our FE model. A fillet radius was considered at the opening
a glass slide, and dehydrated under vacuum overnight at 4 °C. The of the micropipette to mimic the experimentally used micropipettes
dried film was rehydrated with D/R buffer at 4 °C for 3 h and used and to reduce element distortion that would prevent termination of
for the patch-clamp experiment. The channel currents were am- the FE computation (Fig. S1A). Moreover, the fillet radius ap-
plified with an AxoPatch 1D amplifier (Axon Instruments) and peared to have no significant effect on the results as long as the
data were acquired at a sampling rate of 5 Hz with 2-kHz filtration pipette was large enough (3). Symmetrical boundary conditions
in the patch solution [200 mM KCl, 40 mM MgCl 2, and 5 mM were used on the liposome, restricting its horizontal movement in
Hepes (pH 7.2, adjusted with KOH)]. Pressure was applied man- the axis of symmetry. As illustrated in Fig. S1, because the mi-
ually with a syringe for flare-up experiments; ramp pressures were cropipette was significantly stiffer than the liposomes, the mi-
generated by a High Speed Pressure Clamp-1 apparatus (HSPC-1; cropipette was assumed to be rigid and fixed at its reference point
ALA Scientific Instruments). (restricted from moving in all translational and rotational direc-
tions). Frictionless, hard contact (penalty method), finite sliding,
Finite-Element Simulation. Finite-element (FE) simulation has surface-to-surface contact was implemented between the micro-
been used widely to model the micropipette aspiration technique pipette and the liposome surface. For patch fluorometry experi-
to study the mechanical behavior of several different cell types ments in all FE simulations, suction was increased from 0 to its
(2–6). Given the geometric nonlinearities that had to be taken into maximum value within 0.05 s for each step. ABAQUS requires
consideration, we used commercial finite-element analysis (FEA) Young’s modulus, E, and the Poisson ratio, ν, for elastic models,
software (Abaqus/Standard; Dassault Systems Simulia) for simu- and neo-Hookean material parameters can be expressed in terms of
lations as well as for prediction of the stress and strain distribution C10 and D1. E and C10 could be obtained experimentally and ν and
in azolectin liposomes and excised membrane patches exposed to D1 were assumed to be 0.5 and 0, respectively, as lipid bilayers can
pipette aspiration. As the inertial forces were negligible during be considered as almost incompressible materials (8–12). All these
suction, the procedure was considered quasistatic. The lipid vesi- values are stated in the relevant legends for each FE simulation.
cles were assumed to be deformable thin-walled spheres for cell- For viscoelastic materials, ABAQUS uses a Prony series expan-
attached configuration (Fig. S1A). Thin L-shaped shells were used sion of the dimensionless relaxation modulus. ABAQUS inputs for

Bavi et al. www.pnas.org/cgi/content/short/1409011111 1 of 16

viscosity are shear relaxation modulus ratio, bulk modulus, and ΔT
relaxation time. A parametric set of simulations has been per- KA = : [S3]
α
formed. Assuming lipid bilayers as incompressible materials, wide
ranges of shear relaxation modulus ratio, g_i (0.1–0.9), and re- ΔT is the change in tension due to the change of negative pres-
laxation time, tau_i (10 μs to 1 s), have been considered in our sure at each pressure step. Depending on the thickness of the
simulations to cover the rheometry of different lipids with dif- lipid bilayer, we can relate the area stretch elasticity to Young’s
ferent characteristics. When we increase g_i, it means that the modulus, using the following expression (Fig. S2A):
long-term shear modulus decreases (more fluid behavior) and if
we decrease tau_i, we reduce the rate of transition from the 2KA ð1 − νÞ
E= : [S4]
short-term to the long-term modulus. The ABAQUS Analysis t
User’s Manual covers viscoelasticity in detail.
In cases of uniform stretching and bending, the bilayer behaves as
Micropipette Aspiration Technique (Constitutive Model Based on an an incompressible elastic body (13). Thus, ν is the Poisson ratio
Equibiaxial Tension Assumption for Liposome Elongation in the Pipette, that can be assumed to be near 0.5 (8–11) and t is the thickness
Model 1). A popular structural model for liposomes assumes that of the unstressed lipid bilayer. For lipid bilayers (assuming uni-
they have mostly elastic behavior. They cannot be simply modeled form lateral pressure distribution with depth in the uncoupled
as a thin liquid film because the hydrocarbon-chain interior of the monolayers), the elastic modulus is related to the bending ri-
membrane exhibits elastic behavior when its thickness is varied gidity, kb , through kb = Et3 =24, where t is the bilayer thickness
(13). The use of static analysis was justified here because the (10, 22, 23).
timescale of relaxation from viscoelastic effects is on the order of Constitutive Model Based on Uniaxial Linear Elastic Assumption
tens of microseconds (10, 14). (Model 2). Although the patch fluorometry experiment did not
During deformation, lipid bilayers bear external loads and exhibit all ideal uniaxial test conditions, one could consider this
resist bending deformation. Membrane mechanical properties experiment as a uniaxial test. If we look at the liposome behavior on
have been extensively studied by application of pressure across an the patch scale (>1 μm, range of pipette radius) rather than on the
aspirated liposome in a glass pipette (micropipette aspiration lipid raft scale (10–200 nm) (24), the cylindrical pipette precludes
technique). In those experiments, however, the lipid glass ad- expansion of lipid in the radial direction. Hence, a uniaxial as-
hesion and seal formation were overlooked (10, 15, 16). Thus, the sumption was more appropriate than an equibiaxial assumption
traditional analytical model used to estimate membrane tension for obtaining a stress–strain curve of the lipid during patch fluo-
in stretched membrane patches based on Laplace’s law had to be rometry experiments. In both the traditional and the alternative
modified with regard to the radius of the liposome patch to models, it was assumed that there was no substantial slippage of
calculate accurately the membrane tension, T. The membrane the lipid molecules from outside the pipette into the pipette
tension in the presence of the adhesion tension is expressed as during suction (after the initial equilibrium position). Moreover,
(17, 18) the effect of the normal force that the pipette exerted on the lipid
inside the pipette was disregarded. Although the azolectin lipid
PRd showed almost linear elastic behavior, to improve the accuracy
T= : [S1] of our calculations, the nominal longitudinal strain was line-
2ð1 − Rd =Rv Þ
arized as
This equation was applied to both the portion of the vesicle  
ΔLi
inside the micropipette with the inner radius of Rp and that «i = «i−1 + «0 = 0 and i = 1; 2; 3; 4 ; [S5]
outside the pipette with the radius of Rv . P is the pressure Li−1
difference between the outside and the inside of the patch;
Rd is the local radius of curvature of the patch area, which is where L0 is the initial projection length and ΔLi is the change of
equal to ðR2p + h2 Þ=2h, and h is the height of the patch dome corresponding projection lengths at each pressure step. Using
(Fig. S1 D–F). In the absence of the adhesion tension, a sim- Eq. S1, the nominal tensile stress was also calculated as
pler form of Eq. S1 can be derived for micropipette analysis,
where Rd = Rp (7, 15). Obviously the stretching modulus re- T
σ= : [S6]
sulting from Eq. S1 will be higher than the resulting stretching t
modulus when we use T = PRp =2ð1 − Rp =Rv Þ to calculate the
tension. This is because Rd is always greater than Rp . Thus, for From the slopes of the plots of the applied tension vs. the nominal
the same strain, higher values of T were obtained at the cor- longitudinal strain of all of the experimental data obtained in this
responding tensions. study, Young’s modulus of azolectin lipid could be calculated
The resulting membrane strain, α, which is the area change, ΔA, (Fig. S2B).
normalized by the initial area, A0 , was calculated as follows:
Large-Strain Isotropic Hyperelastic Constitutive Model (Model 3). It is
   3 ! important to mention that a hyperelastic material is still an elastic
ΔA 1 Rp 2 Rp ΔL material, which means that it returns to its original form after
α= ∼ − : [S2]
A0 2 R R Rp deformation forces have been removed. The linear elastic coef-
ficients of azolectin liposomes were discussed in the previous
ΔL is the change in projection length produced by an increase in section. Given that elastic material models are intended for elastic
applied pressure (Fig. S1). Eq. S2 was deduced with the viable strains that usually remain small (<5%) and hyperelastic material
assumption that the internal volume of the vesicle during micro- models are more appropriate for most biological materials, par-
pipette aspiration remains constant due to incompressibility of ticularly at large strain magnitudes (>5%) (25), we fitted a hy-
aqueous solution inside the vesicle (15, 19). Following a typical perelastic model to our experimental data and introduced the
micropipette aspiration protocol, the area stretch elasticity mod- corresponding coefficients. In fact, herein we show that liposomes
ulus under a high membrane tension regime [when T > 0.5 mN/m could be modeled as a large-strain isotropic hyperelastic material.
(20, 21)], KA , was calculated as follows: Hyperelastic materials also are Cauchy elastic, which means that

Bavi et al. www.pnas.org/cgi/content/short/1409011111 2 of 16

the stress is determined by the current state of deformation, not perform enough different standard experiments on the liposome
by the path or history of deformation. The Cauchy stress can be vesicles (because of their sensitive properties, size, and form).
derived from the strain energy function, which is given by (26) Here we use the simplest constitutive model, the neo-Hookean
model. The neo-Hookean model is the first-order polynomial form
X
N  i  j XN
1  el 2i of the general hyperelastic model with N = 1. It uses only linear
U= Cij I 1 − 3 I 2 − 3 + J −1 [S7] functions of the invariants. In this model, the strain energy density
i + j=1
Di
i=1 is a linear function of deviatoric strain invariants, I 1 and I 2 , and
can be derived from Eq. S7 as follows:
ðDeviatoric partÞ ðVolumetric partÞ;  
U = C10 I 1 − 3 : [S14]
where U is the strain energy per unit of reference volume and i and j
are integer numbers. As shown in Eq. S7, U contains a deviatoric In the neo-Hookean model, shear modulus, G, is
part and a volumetric part. N is the polynomial order, Cij and Di are
temperature-dependent material parameters, I i are deviatoric strain G = 2C10 : [S15]
invariants of the left Cauchy–Green deformation tensor, and J el is
the elastic volume ratio. In this section we summarize briefly the As a result, we may simplify the boundary and loading conditions
equations of incompressible isotropic nonlinear elasticity that are inside the pipette as illustrated in Fig. S3. Also the geometry and
required for comparing the theory with patch fluorometry data. We membrane stresses of the lipid bilayer during a micropipette as-
assume homogeneous deformations that can be classified as piration experiment are indicated. RP and σ represent the inner
pure homogeneous strain, i.e., deformations of the form radius of the pipette and the longitudinal stress of the mem-
brane, respectively. L is the length of projection of the lipid
x1 = λ1 X1 ; x2 = λ2 X2 ; x3 = λ3 X3 ; [S8] inside the micropipette. λ1 and λ2 are the stretching ratios in
directions 1 (horizontal) and 2 (vertical), respectively. Fig. S3,
where X1 , X2 , and X3 are rectangular Cartesian coordinates that Right depicts the planar form of a liposome with the associated
identify material particles in some unstressed reference configu- boundary conditions caused by the rigid micropipette. The rigid
ration. x1 , x2 , and x3 are the corresponding coordinates after cylindrical pipette around the patch area prevents any growth of
deformation with respect to the same axes. λi ði = 1; 2; 3Þ are the the radius in the portion of the liposome within the pipette,
stretch ratios in the principal directions. The first and second λ1 = 1 (Fig. S3). Also, due to the boundary conditions of the
deviatoric strain invariants in Eq. S7 are defined as liposome bilayer within the pipette and the incompressibility of
azolectin lipid, λ3 = 1=λ2 = 1=λ. Hence, Eq. S14 can be expressed
I 1 ¼ λ21 þ λ22 þ λ23 ; I 2 ¼ λ21 λ22 þ λ22 λ23 þ λ21 λ23 : [S9] in terms of λ as
 
The principal stretch ratios λi ði = 1; 2; 3Þ are related to the prin- G 2 1
U= λ + 2−2 : [S16]
cipal nominal strain «i through «i = λi − 1. The principal stretch λi 2 λ
can be linearized, as previously explained in Eq. S5, to achieve
more accuracy. With the assumption of full incompressibility for Thus, using Eq. S13, the nominal stress in the main direction 2
lipid membranes (7), can be expressed as
 
J el = λ1 λ2 λ3 = 1: [S10] 1
σ=G λ− 3 ; [S17]
λ
Thus, the volumetric part of strain energy (U) becomes equal to
zero. The principal Cauchy stresses σ i ði = 1; 2; 3Þ [defined as force where σ = T/t (t = 3.5 nm). T can be calculated from Eq. S1. The
per unit deformed cross-sectional area normal to the xi ði = 1; 2; 3Þ slope of the nominal stress and λ − 1=λ3 indicate the shear mod-
axis in the deformed configuration] are related to the stretches ulus (Fig. 1C).
through U according to the equations
Equations for Excised Patch Configuration. To suppress unknown
∂U thermodynamic effects such as membrane pretension (27) and
σ i = λi − P i = 1; 2; 3: [S11]
∂λi thermal undulations (19–21) involved in the mechanical behav-
ior of liposomes and, more importantly, to study the rheological
However, for calculation of the Cauchy stresses, we need to know behavior of the lipid in the excised patch configuration, micro-
the exact thickness at different parts of the patch area during as- pipette aspiration was performed on three different excised
piration of the liposomes. To avoid this at this juncture, the cor- patches (Fig. 3). This method has several advantages over the
responding nominal stresses (defined as per unit undeformed cell-attached configuration, including simplicity and accuracy.
cross-sectional area) are the stresses that are often measured di- For tension in the excised patch membrane, T, in Eq. S1, based
rectly in experiments and thus are given by (26) on a principle of surface chemistry (Laplace’s law) is changed as
follows (12, 27–31):
∂U
σi = − Pλ−1 i = 1; 2; 3: [S12]
∂λi i
PRd
T= : [S18]
2
For an incompressible material it is always possible to superimpose
a hydrostatic stress without producing strain and Eq. S12 changes to Rd is the radius of curvature of the patch (Fig. S1 D–F). Note
that there is no attached liposome part in the excised patch
∂U configuration. Therefore, the fundamental assumption of a con-
σi = i = 1; 2; 3: [S13]
∂λi served internal volume for the vesicle due to the incompressibility
of the water inside the liposome during micropipette aspiration is
As presented in the following sections, the results show linear be- no longer viable. Membrane deformations in excised patch ex-
havior for azolectin lipid bilayer (Fig. S2), and we were unable to periments are conventionally characterized by the relative change

Bavi et al. www.pnas.org/cgi/content/short/1409011111 3 of 16

in the visible area measured with respect to a somewhat arbi- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
trarily chosen initial state area, A0. The areal strain (fractional ðS1 − S2 Þ2 + ðS1 − S3 Þ2 + ðS1 − S2 Þ2
area change) is defined by Sv = ; [S21]
2
A − A0
α= ; [S19] where Si (i = 1, 2, 3) are the stress components in the principal
A0 directions. Von Mises stress is used for fracture analysis of duc-
tile materials. The stress distributions were nonuniform, with
where A is the total deformed area of the patch. Using very basic continuous regions of high and low stress. Moreover, to distin-
geometric relations, the deformed area can be calculated from guish the role of the rigid micropipette in the movement of lipid
the cylindrical length, L; the pipette radius, Rp; and the height of molecules during micropipette aspiration, the vertical displace-
the dome of the patch, h (Fig. 3) by ment field and in-plane maximum principal strain field in the
  azolectin vesicle were calculated and are shown in Fig. S5. By
A = 2πRp L + π R2p + h2 : [S20] comparing the vertical displacement field and the in-plane max-
imum principal strain field it can be observed that, although the
apex of the patch has the largest vertical displacement, the ele-
Because the other models discussed in this paper (uniaxial and
ments within the liposome–pipette normal contact region have
hyperelastic models) do not deal with the attached part of the ves-
the greatest strains (Fig. S5). This implies that when a local stress
icle in the cell-attached configuration (and thus the constant vol-
is produced in this area, the lipid membrane is unable to recon-
ume assumption), they are all still valid also for the excised patch
figure itself and reduce the strain. Consequently, movement of
configuration. In those models, however, for the calculation of
the elements (in an FE simulation) or the phospholipids of the
membrane tension, Eq. S18 should be used instead of Eq. S1.
membrane (in reality) (10) is restricted and this facilitates mem-
The material properties obtained from different models (models
brane rupture (Figs. S3 and S5). The computational results show
1–3) for excised patch configuration are demonstrated (Fig. S4
that the response is mostly dominated by local stretching of the
and Fig. 3C).
liposome rather than its shear and/or bending effects near this
Supporting Patch Fluorometry Data Without Adhesion Tension. To normal contact area. No substantial result sensitivity to the con-
calculate the bilayer material properties (i.e., KA) we used ΔT, tact conditions between the liposome and the pipette is ob-
which is the change in tension due to the change of negative served, as the stress and the resulting deformation are mainly
pressure at each pressure step (Eq. S3). We believe this reduces dominated by force regime and longitudinal movement of mem-
brane within the pipette rather than by normal and tangential
the potential influence of adhesion tension, assuming this value
effects near the pipette opening.
stays constant during pressure application. Furthermore, we
A set of FE computations was performed to study the effects of
carried out additional patch-fluorometry experiments, using BSA
the radius of the vesicle on tension, thickness variation, and in-
to reduce adhesion tension. As mentioned in the literature, we plane stress field of the patch in the cell-attached configuration
used 0.02% BSA (32) and the result was similar to what we (Fig. S6). The vesicle size ranged from 3.1 μm (small vesicles that
previously measured. However, this concentration failed to com- could be found abundantly in each sample) to 12.4 μm (rare
pletely remove adhesion tension. Using only 0.1% BSA allowed typical sizes encountered experimentally). We showed that the
complete removal of adhesion tension (Movie S5). The corre- radius of the vesicle outside liposome had no substantial effect
sponding value calculated for KA using model 1 is ∼91 mN/m, on tension distribution, thickness variation, and stress field in the
which is similar to that calculated in the absence of BSA (∼112 patch area within the pipette (Fig. S6). This is consistent with
mN/m; Table 1). The corresponding value of KA using model 3 is what we had assumed for deducing our mathematical elastic and
10 mN/m (∼14 mN/m without BSA; Table 1), given that 0.1% hyperelastic equations. The tensions estimated from the Laplace
might affect the membrane properties and increase the irre- equation for the cell-attached configuration (Eq. S1) and the
producibility of KA (32–35). excised patch configuration (Eq. S18) were compared with the
Thus, a small amount of adhesion tension aids experimental FE results. On the basis of our computational results, for rela-
simplicity and is likely present in a large number of the published tively small vesicles (i.e., Rv < 9 μm) the estimated tension ob-
micropipette aspiration (MA) reports, which is signified clearly by tained from Eq. S1 is always an overestimation. On the other
the radius of curvature being larger than the radius of the pipette, hand, for larger vesicles, like the tension estimated from Eq. S18,
which can be seen in the initial equilibrium state of the patch in the results were always lower than the (FE) values (Fig. 2D).
previous studies (7, 21, 36). Importantly, this adhesion tension This also allowed us to choose the more appropriate equation
does not seem to affect our calculated values for KA. (between Eqs. S1 and S18) for tension estimation in the process
of calculating the mechanical properties of lipid bilayer.
Supporting Computational Data. As mentioned, shell theory was Previously, we showed that a micropipette aspiration protocol
used in our computations, which is a more advanced theory for based on an equibiaxial tension assumption results in over-
estimation of tension in thin-walled shells compared with Lap- estimation of lipid elastic moduli. Moreover, the mechanical
lace’s law. The spatial profiles of the aspirated liposome calcu- properties of larger liposomes are shown to be stiffer than those of
lated by the FE simulation are presented in Fig. 2 A and B and smaller vesicles. However, unlike the cell-attached configuration,
Fig. S5. The vesicle has a diameter of 6.2 μm with the inner the mechanical properties obtained from an isolated patch in
diameter of the micropipette being 2.8 μm (both are typical sizes micropipette aspiration (excised patch configuration) are very
encountered experimentally). The suction begins from 0 and similar using all three different material models (Fig. S4 A and B,
reaches a value of −30 mmHg (∼4 kPa), instantaneously (in 0.01 s), Fig. 3B, and Table S2). In a similar manner to that performed for
and is then kept constant for 0.01 s. The computations were the cell-attached configuration, we also modeled the excised
performed for the material parameters of C10 = 0.5 MPa and patch experiment (Fig. S1 B and C). The mechanical properties
kb = 5.36 × 10−21 (neo-Hookean model). The in-plane maximum obtained were used as input data for our computational model to
and effective (von Mises) stress fields are represented in Fig. 2 A see whether we could observe the same rheological behavior for
and B. Von Mises stress, Sv , is an equivalent stress of distortion our lipid bilayer model under experimental conditions (i.e.,
energy of a material, which in principal directions can be cal- typical patch and pipette geometry in conjunction with similar
culated from load and boundary conditions). Thus, the length of lipid membrane

Bavi et al. www.pnas.org/cgi/content/short/1409011111 4 of 16

within the pipette at four different pressures was compared with Effect of Intermonolayer Dissipation: Modeling the Bilayer as Two Sliding
experimentally derived values and very good agreement was Slabs. Previous models (models 1–3) were conceptualized on the
found between the two approaches (Fig. S4 C and D and Fig. basis of the mechanical behavior for a unit membrane structure
3C). We believe that the relatively small difference between the for simplicity. This assumption is quite viable when there is a rapid
two is mainly attributable to the adhesion tension between lipid displacement between layers (i.e., applying instantaneous suction).
and micropipette, which was not taken into account in our com- In this case, their relative lateral motion will be opposed by a
putational model. considerable viscous drag at the bilayer midplane, which will lead
Both typical pipette shapes, cylindrical and tapering, were to “dynamic coupling” of the monolayers, causing them to behave
simulated to see whether there was any difference in rheological as a single slab (43). However, the monolayers are able to slide
behavior between the two during aspiration. Furthermore, a set of one relative to another in the case of regular displacements. This
calculations was performed, considering similar computational is because monolayers are tied together by a weak van der Waals
conditions to indicate the effect of initial membrane thickness on (vdw) attraction at the midplane, stemming from the aqueous half
spaces surrounding the bilayer (43).
the maximum membrane stress (in the apex of the patch), up to
Given that the boundary and loading conditions are different
suction pressure 40 mmHg (Figs. S7 and S8). The initial thickness
between the monolayers during micropipette aspiration, herein,
of the lipid bilayer did not affect the tension distribution but it had
we examine to what extent this affects the stress distribution
a considerable effect on the membrane stress distribution within between the monolayers in both cell-attached and excised con-
the patch area. The thinner the membrane, the higher was the figurations. The interlayer drag has been assumed to be velocity
stress at any given pressure. dependent and it follows the postulated constitutive relation for
interlayer coupling τ = bΔvs. This assumption has been used in
Effect of Membrane Fluidity (Internal Dissipation) on the Stress–Strain
many other continuum mechanics (coarse grain) models of lipid
Distribution of Patched Lipid Bilayers. Visco-hyperelasticity is the
bilayers, where b represents the magnitude of coupling em-
property of materials that exhibit both viscous and hyperelastic
bodied in a constant drag coefficient (of order 108 N− s/m3).
characteristics when undergoing large deformation. Given that Δvs represents the relative rate that molecules (nodes in our
fluid–lipid biomembranes show both liquid-like and solid-like be- continuum model) in opposite monolayers slip past each other
havior, numerous researchers have adopted viscoelastic models for as the bilayer deforms (43).
describing the behavior of different lipid bilayers and cell mem- As mentioned before, the monolayers are in contact with each
branes (3, 6, 14, 37–42). other via a weak force stemming from vdw interaction between
Although for azolectin we and others (1) have seen negligible the two hydrophobic surfaces. The interlayer pressure as a func-
stiffness nonlinearity (Figs. 1C and 3B and Figs. S2 and S4 A and tion of the distance can be computed, adopting the vdw stress
B) and hysteresis (Movie S5), we considered a low viscous (flu- between two parallel bilayers,
idic) behavior for lipids in addition to the lipid hyperelasticity to
see how this changes the stress distribution in the patch area. In −AH
other words, a hyperelasticity model was used to study the result S= : [S22]
6πD3
of the reversible bond stretching along the crystallographic planes
of the lipid bilayer. Herein, viscosity was added to our previous Here S is the stress, AH is the Hamaker constant, and D is the
(hyperelastic) model to capture the influence of the fluidity (in- distance between the two monolayers. For lipid bilayers in water
ternal damping) and creep of lipid molecules inside the pipette the Hamaker constant is ∼5 ( ± 2) × 10−21 J (7, 44). Let us consider
during micropipette aspiration. This enabled us to investigate the this number for hydrocarbons in the monolayers. The normal trac-
influence of lipid internal viscosity on the stress–strain distribution tions can be positive, indicating an attractive interaction between
within the patch area during micropipette aspiration of the bilayer. the surfaces, or negative, in the case of repulsive forces.
In the previous hyperplastic model (Figs. 2 and 3), the results The resulting expression of the interlayer contact interactions
from finite-element simulations predict that there is a persistent has been implemented in an ABAQUS user subroutine for sim-
heterogeneity in tension with the maximum at the top of the patch ulating the vdw force. In ABAQUS/Standard, user subroutine
dome of highly viscous membranes. In our introduced alternative UINTER can be used to define the constitutive interaction be-
visco-hyperelastic model, we assume a surface viscosity as the tween two deforming surfaces. Monolayers are defined as the
internal dissipative mechanisms of the bilayer. Our results in- master and slave surfaces, and the UINTER is called for each
dicate that for the membranes with low viscosity (high fluidity), slave node at each time increment of the analysis. Inputs to this
such nonuniform distribution of tension is still feasible for dif- subroutine are the initial and incremental relative positions of
ferent loading conditions (step or ramp), various fluidities, and each slave node with respect to its closest point on the master
surface and the material properties defined for the monolayers.
typical experimental time courses. In a time course of 5 s (a
The constitutive calculation thus involves computing the tractions
typical experimental time) the maximum tension in the patch
based on the increments in relative position of the slave node with
starts to expand over the patch area due to the relaxation along
respect to the master surface.
the tension gradient (Movies S6 and S7). This happens in both Interestingly, our results indicate that higher membrane stress
cell-attached and excised bilayer models. As shown, after 5 s the values are developed in the outer leaflet compared with the inner
difference between the stress in the dome (maximum) and the one by considering the lipid bilayer as two sliding surfaces. The
stress near the wall (minimum) is about 50% (Movie S6). Hence, asymmetry in the stress profiles of the two leaflets exists in both
comparing results from the visco-hyperelastic model with our excised and cell-attached systems. However, we showed that for
previous results, not only is the heterogeneous distribution of the similar characteristics and conditions, the dissimilarity between
stress valid with the existence of membrane fluidity but also it in- the stress profiles of two monolayers is much more noticeable in
troduces a new time-dependent axisymmetric growth of the high- the excised configuration compared with the cell-attached one.
stress region (at the patch center) toward the low-stress area (near In the excised patch system, the maximum stress that arises in the
the pipette wall). This tense area in the center of the patch de- upper monolayer (the one that is in contact with the pipette) is
velops to the sides of the patch area (adjacent to the pipette wall) about 30% larger than the maximum stress in the inner monolayer
as the lipid creeps inside the pipette during the simulated experi- (Fig. 3 D and E), whereas in the cell-attached configuration, the
mental time. Thus, stress heterogeneity is valid for a wide range of difference in the monolayers’ maximum stress is less than 2%
instantaneous and long-term shear modulus and relaxation times. (Fig. 2 E and F). The difference between the monolayers stems

Bavi et al. www.pnas.org/cgi/content/short/1409011111 5 of 16

from the fact that the inner leaflet in the excised patch has a higher pore in different MS channels (close to or away from the mid-
degree of freedom for lateral movement and relaxation than the plane), the dissimilarity between the distributed stresses in the
inner leaflet in the cell-attached conformation. Based on these monolayers can affect activation of MS channels reconstituted
results, we can suggest that depending on the location of the and investigated in an excised patch system (45, 46).

1. Nomura T, et al. (2012) Differential effects of lipids and lyso-lipids on the 22. Allen KB, Layton BE (2009) Determination of the forces imposed by micro and
mechanosensitivity of the mechanosensitive channels MscL and MscS. Proc Natl Acad nanopipettes during DOPC: DOPS liposome manipulation. Chem Phys Lipids 162(1–2):
Sci USA 109(22):8770–8775. 34–52.
2. Sato M, Theret DP, Wheeler LT, Ohshima N, Nerem RM (1990) Application of the 23. Bloom M, Evans E, Mouritsen OG (1991) Physical properties of the fluid lipid-bilayer
micropipette technique to the measurement of cultured porcine aortic endothelial component of cell membranes: A perspective. Q Rev Biophys 24(3):293–397.
cell viscoelastic properties. J Biomech Eng 112(3):263–268. 24. Lingwood D, Simons K (2010) Lipid rafts as a membrane-organizing principle. Science
3. Vaziri A, Mofrad MR (2007) Mechanics and deformation of the nucleus in micropipette 327(5961):46–50.
aspiration experiment. J Biomech 40(9):2053–2062. 25. Holzapfel GA, Ogden RW (2006) Mechanics of Biological Tissue (Springer, Berlin).
4. Jafari Bidhendi A, Korhonen RK (2012) A finite element study of micropipette 26. Ogden R, Saccomandi G, Sgura I (2004) Fitting hyperelastic models to experimental
aspiration of single cells: Effect of compressibility. Comput Math Methods Med 2012, data. Comput Mech 34(6):484–502.
10.1155/2012/192618. 27. White SH (1980) Small phospholipid vesicles: Internal pressure, surface tension, and
5. Zhou E, Lim C, Quek S (2005) Finite element simulation of the micropipette aspiration surface free energy. Proc Natl Acad Sci USA 77(7):4048–4050.
of a living cell undergoing large viscoelastic deformation. Mech Adv Mater Structures 28. Tanford C (1979) Hydrostatic pressure in small phospholipid vesicles. Proc Natl Acad
12(6):501–512. Sci USA 76(7):3318–3319.
6. Trickey WR, Baaijens FP, Laursen TA, Alexopoulos LG, Guilak F (2006) Determination 29. Opsahl LR, Webb WW (1994) Lipid-glass adhesion in giga-sealed patch-clamped
of the Poisson’s ratio of the cell: Recovery properties of chondrocytes after release membranes. Biophys J 66(1):75–79.
from complete micropipette aspiration. J Biomech 39(1):78–87. 30. Suchyna TM, Markin VS, Sachs F (2009) Biophysics and structure of the patch and the
7. Evans E, Needham D (1987) Physical properties of surfactant bilayer membranes: gigaseal. Biophys J 97(3):738–747.
Thermal transitions, elasticity, rigidity, cohesion and colloidal interactions. J Phys 31. Ursell T, Agrawal A, Phillips R (2011) Lipid bilayer mechanics in a pipette with glass-
Chem 91(16):4219–4228. bilayer adhesion. Biophys J 101(8):1913–1920.
8. Kirk GL, Gruner SM, Stein D (1984) A thermodynamic model of the lamellar to inverse 32. Zhou Y, Raphael RM (2005) Effect of salicylate on the elasticity, bending stiffness, and
hexagonal phase transition of lipid membrane-water systems. Biochemistry 23(6): strength of SOPC membranes. Biophys J 89(3):1789–1801.
1093–1102. 33. Shoemaker SD, Vanderlick TK (2002) Intramembrane electrostatic interactions
9. Kuzmin PI, Akimov SA, Chizmadzhev YA, Zimmerberg J, Cohen FS (2005) Line tension destabilize lipid vesicles. Biophys J 83(4):2007–2014.
and interaction energies of membrane rafts calculated from lipid splay and tilt. 34. Wu Y, Fletcher GL (2000) Efficacy of antifreeze protein types in protecting liposome
Biophys J 88(2):1120–1133. membrane integrity depends on phospholipid class. Biochim Biophys Acta 1524(1):
10. Rodowicz KA, Francisco H, Layton B (2010) Determination of the mechanical 11–16.
properties of DOPC:DOPS liposomes using an image procession algorithm and 35. Yokouchi Y, et al. (2001) Effect of adsorption of bovine serum albumin on liposomal
micropipette-aspiration techniques. Chem Phys Lipids 163(8):787–793. membrane characteristics. Colloids Surf B Biointerfaces 20(2):95–103.
11. Akimov SA, Kuzmin PI, Zimmerberg J, Cohen FS (2007) Lateral tension increases the 36. Evans E, Rawicz W, Smith BA (2013) Back to the future: Mechanics and thermodynamics
line tension between two domains in a lipid bilayer membrane. Phys Rev E Stat Nonlin of lipid biomembranes. Faraday Discuss 161:591–611.
Soft Matter Phys 75(1 Pt 1):011919. 37. Waugh R, Evans EA (1976) Viscoelastic properties of erythrocyte membranes of
12. Hamill OP, Martinac B (2001) Molecular basis of mechanotransduction in living cells. different vertebrate animals. Microvasc Res 12(3):291–304.
Physiol Rev 81(2):685–740. 38. Jamali Y, Azimi M, Mofrad MR (2010) A sub-cellular viscoelastic model for cell
13. Kralchevsky PA, Paunov VN, Denkov ND, Nagayama K (1995) Stresses in lipid population mechanics. PLoS ONE 5(8):pii:e12097.
membranes and interactions between inclusions. J Chem Soc Faraday Trans 91(19): 39. Canham PB (1970) The minimum energy of bending as a possible explanation of the
3415–3432. biconcave shape of the human red blood cell. J Theor Biol 26(1):61–81.
14. Earnshaw JC, Crawford GE (1989) Viscoelastic relaxation of bilayer lipid membranes: 40. Mills JP, Qie L, Dao M, Lim CT, Suresh S (2004) Nonlinear elastic and viscoelastic
II. Temperature dependence of relaxation time. Biophys J 55(5):1017–1021. deformation of the human red blood cell with optical tweezers. Mech Chem Biosyst
15. Evans E, Rawicz W (1997) Elasticity of “fuzzy’” biomembranes. Phys Rev Lett 79(12): 1(3):169–180.
2379–2382. 41. Lubarda V, Marzani A (2009) Viscoelastic response of thin membranes with application
16. Kwok R, Evans E (1981) Thermoelasticity of large lecithin bilayer vesicles. Biophys J to red blood cells. Acta Mech 202(1–4):1–16.
35(3):637–652. 42. Smeulders JB, Blom C, Mellema J (1990) Linear viscoelastic study of lipid vesicle dispersions:
17. Mitchison J, Swann M (1954) The mechanical properties of the cell surface I. The cell Hard-sphere behavior and bilayer surface dynamics. Phys Rev A 42(6):3483–3498.
elastimeter. J Exp Biol 31(3):443–460. 43. Evans E, Yeung A (1994) Hidden dynamics in rapid changes of bilayer shape. Chem
18. Bowman CL, Gottlieb PA, Suchyna TM, Murphy YK, Sachs F (2007) Mechanosensitive Phys Lipids 73(1):39–56.
ion channels and the peptide inhibitor GsMTx-4: History, properties, mechanisms and 44. De Haas K, et al. (1997) Rheological behavior of a dispersion of small lipid bilayer
pharmacology. Toxicon 49(2):249–270. vesicles. Langmuir 13(25):6658–6668.
19. Henriksen JR, Ipsen JH (2004) Measurement of membrane elasticity by micro-pipette 45. Belyy V, Kamaraju K, Akitake B, Anishkin A, Sukharev S (2010) Adaptive behavior of
aspiration. Eur Phys J E Soft Matter 14(2):149–167. bacterial mechanosensitive channels is coupled to membrane mechanics. J Gen
20. Needham D, Nunn RS (1990) Elastic deformation and failure of lipid bilayer Physiol 135(6):641–652.
membranes containing cholesterol. Biophys J 58(4):997–1009. 46. Häse CC, Le Dain AC, Martinac B (1995) Purification and functional reconstitution of
21. Rawicz W, Olbrich KC, McIntosh T, Needham D, Evans E (2000) Effect of chain length the recombinant large mechanosensitive ion channel (MscL) of Escherichia coli. J Biol
and unsaturation on elasticity of lipid bilayers. Biophys J 79(1):328–339. Chem 270(31):18329–18334.

Bavi et al. www.pnas.org/cgi/content/short/1409011111 6 of 16

 a b

d e f

Sym. axis
Micropipette
Sym. axis

Sym. axis
Micropipette

Micropipette

Rp Rp Rp

h=Rp=Rd
h
x h=0 x 2
Rp + h
2
x
Rd = Rd
Rd = ∞ 2h
Rd

Rv Rv Rv

Fig. S1. Finite-element (FE) model of a lipid bilayer and a micropipette including schematic diagrams of the geometry of cell-attached micropipette aspiration.
(A) For the FE model of cell-attached configuration, a liposome is modeled as a 2D axisymmetric semicircle with the radius of Rv . Rp is the inner radius of the
micropipette. An edge fillet at the opening of the micropipette is used to mimic the experimentally used micropipettes and to avoid element distortion in this
region. (B and C) FE model of excised patch membrane, where the lipid bilayer is modeled as a thin shell in flat form according to the realistic prestressed shape
of the lipid bilayer at resting state, within a cylindrical pipette with the radius Rp , and a tapering pipette with the normal angle of θ, respectively. (D) Flat state
of the patch area when the adhesion tension is much higher than the tension caused by the applied pressure within the pipette. (E) A general state of the
patch area. Rd is the radius of the dome (radius of curvature), which can be calculated from the geometry of the patch using ððR2p + h2 Þ=2hÞ; h is the height of
the patch dome. (F) A special case where the adhesion tension is negligible and thus the patch area has a hemispherical shape.

Bavi et al. www.pnas.org/cgi/content/short/1409011111 7 of 16

Fig. S2. Patch fluorometry results for three azolectin lipid vesicles of similar diameter (6–8 μm). (A) Variation of the membrane tension plotted against
corresponding areal strain. Shown is how the membrane tension changes almost linearly with respect to areal strain; plots are fitted with linear regression
lines. (B) Variation of the nominal longitudinal in-plane membrane stress vs. the linearized longitudinal strain fitted with linear regression lines. The diagram
demonstrates the almost linear change of the nominal stress–strain field in the longitudinal direction.

Fig. S3. Membrane stresses and boundary conditions of the lipid bilayer during a micropipette aspiration experiment. RP and σ represent the inner radius of
the pipette and the longitudinal stress of the membrane, respectively. λ1 and λ2 are the stretching ratios in directions 1 (horizontal) and 2 (vertical), re-
spectively. L is the length of projection of the lipid within the micropipette. (Right) The planar form of a liposome with the associated boundary conditions
caused by the rigid micropipette. (Left) The normal contact area between the liposome and the micropipette as well as one of the potential regions for rupture
initiation in the liposome during micropipette aspiration.

Bavi et al. www.pnas.org/cgi/content/short/1409011111 8 of 16

Fig. S4. Patch fluorometry results for three excised membrane patches and validation of simulations with observations from patch fluorometry experiments.
(A) Variation of the membrane tension plotted against corresponding areal strain. Shown is how the membrane tension changes almost linearly with respect to
the areal strain in the excised configuration; plots are fitted with linear regression lines. (B) Alteration of the nominal longitudinal in-plane membrane stress vs.
the linearized longitudinal strain fitted with linear regression lines. The diagram demonstrates the almost linear change of the nominal stress–strain field in the
longitudinal direction. (C and D) Comparisons between the measured aspiration lengths of azolectin lipid within the pipette and those simulated using the
neo-Hookean hyperplastic model. The inner diameters of the micropipette are typical sizes encountered experimentally. The suction begins from 0 and reaches
a value of −20 mmHg (−5 mmHg increments) instantaneously (in 0.01 s) and is then kept constant for 0.01 s. The computations are performed for the material
parameters of C10 = 0.71 MPa and kb = 5.36 × 10−21.

Fig. S5. Membrane stresses and displacement fields of lipid bilayer during a micropipette aspiration experiment. (A) Spatial profiles of the aspirated liposome
calculated by FE simulation. The vesicle has a diameter of 6.2 μm. The inner diameter of the micropipette is 2.8 μm (both are typical sizes encountered ex-
perimentally). The suction begins from 0 and reaches a value of −30 mmHg (∼4 kPa) instantaneously (in 0.01 s) and is then kept constant for 0.01 s. The
computations are performed for the material parameters of C10 = 0.5 MPa and kb = 5:36 × 10−21 J (neo-Hookean model). Shown are (A) the vertical dis-
placement field (μm) and (B) in-plane maximum principal strain field in the azolectin vesicle.

Bavi et al. www.pnas.org/cgi/content/short/1409011111 9 of 16

Fig. S6. FE results (tension, bilayer thickness, and in-plane stress) of a patch in the cell-attached configuration. The inner diameter of the micropipette is
2.8 μm (a typical size encountered experimentally). The suction begins from 0 and reaches a value of −30 mmHg (∼4 kPa), instantaneously (in 0.01 s), and is then
kept constant for 0.01 s. The computations are performed for the material parameters of C10 = 0.5 MPa and kb = 5:36 × 10−21 J (model 3: neo-Hookean model).
Each row shows the bilayer tension variation, the bilayer thickness change, and the membrane stress distribution within the patch area for different radii of
vesicle, Rv = 3.1 μm (First Row), Rv = 6.2 μm (Second Row), Rv = 9.3 μm (Third Row), and Rv = 12.4 μm (Fourth Row). Solid line in Left column represents tension
estimated using Eq. S1 (cell-attached), and the dashed line represents tension estimated using Eq. S18 (excised patch).

Bavi et al. www.pnas.org/cgi/content/short/1409011111 10 of 16

Fig. S7. Effect of initial thickness of bilayer on tension, thickness variations, and in-plane stress within the patch area in the excised patch configuration (using
a cylindrical pipette). In these FE models, the inner diameter of the cylindrical micropipette is 2.8 μm (a typical size encountered experimentally). The suction
begins from 0 and reaches a value of −40 mmHg (∼5.3 kPa) instantaneously (in 0.01 s) and is then kept constant for 0.01 s. The computations are performed for
the material parameters of C10 = 0.5 MPa and kb = 5:36 × 10−21 J (neo-Hookean model). Each row shows the bilayer tension variation, the thickness change, and
the membrane stress distribution within the patch area for different bilayer thicknesses, t = 3.5 nm (First Row), t = 4 nm (Second Row), t = 4.5 nm (Third Row),
and t = 5 nm (Fourth Row). Solid lines in Left column represent tension estimated using Eq. S18 (excised patch).

Bavi et al. www.pnas.org/cgi/content/short/1409011111 11 of 16

Fig. S8. Effect of initial thickness of bilayer on tension, thickness variations, and in-plane stress within the patch domain in excised patch configuration (using
a conical pipette). All of the model properties are the same as those of the patch model with a cylindrical pipette (Fig. S7), except that the initial inner diameter
of the conical micropipette is 1.2 μm before applying suction and the angle of the pipette is 10° (these are typical values encountered experimentally). Thus, the
suction begins from 0 and reaches a value of −40 mmHg (∼5.3 kPa) instantaneously (in 0.01 s) and is then kept constant for 0.01 s. The computations are
performed for the material parameters of C10 = 0.5 MPa and kb = 5:36 × 10−21 J (neo-Hookean model). Each row shows the bilayer tension variation, the
thickness change, and the membrane stress distribution within the patch area for the bilayer thickness, t = 3.5 nm (First Row), t = 4 nm (Second Row), t = 4.5 nm
(Third Row), and t = 5 nm (Fourth Row). Solid lines in Left column represent tension estimated using Eq. S18 (excised patch).

Table S1. Validation of the proposed material properties

of azolectin liposomes by comparing computational and
experimentally derived values of the change in projection
length, ΔL, at two different pressures (μm)
Different approaches ΔL, in −10 mmHg ΔL, in −20 mmHg

FE: linear elastic, model 1 0.4 0.6

FE: linear elastic, model 2 0.7 1.2
FE: neo-Hookean, model 3 0.8 1.6
Patch fluorometry experiment 0.8 1.7

Bavi et al. www.pnas.org/cgi/content/short/1409011111 12 of 16

 Table S2. Pretension in the lipid within the pipette before pressurizing
Experiment no. Excised patch Cell-attached, small vesicles Cell-attached, large vesicles

1 0.1 8.2 5.3

2 0.4 3.4 13.2
3 0.8 8.2 —

Extrapolations back to zero area dilation gave a value for the tension in the resting membrane of up to 13 ± 3
mN/m for very large vesicles, in close agreement with earlier measurements. Pretension values for the excised
patch configuration may indicate the adhesion tension between the glass and azolectin lipid bilayer, whereas
those of the vesicles are not just attributable to the adhesion tension. As can be seen, the pretensions in excised
patches are much lower than those found in vesicles. All values are in mN/m.

Movie S1. Movement of fluorescent-labeled azolectin liposomes (cell-attached configuration) during micropipette aspiration using confocal microscopy.
Liposomes were aspirated by applying a negative pressure of −20 mmHg (5-mmHg increments).

Movie S1

Bavi et al. www.pnas.org/cgi/content/short/1409011111 13 of 16

Movie S2. Simulation of the movement of a liposome patch (cell-attached configuration) and stress distribution during micropipette aspiration, using the
finite-element method. A hyperelastic material model has been adopted here.

Movie S2

Movie S3. Movement of fluorescent-labeled azolectin lipid (excised patch configuration) during micropipette aspiration, using confocal microscopy. The
membrane patches were aspirated by applying a negative pressure of −20 mmHg (5-mmHg increments).

Movie S3

Bavi et al. www.pnas.org/cgi/content/short/1409011111 14 of 16

Movie S4. Simulation of the movement of azolectin lipid (excised patch configuration) and stress distribution during patch clamping, using the finite-element
method. A hyperelastic material model has been adopted here.

Movie S4

Movie S5. Movement of fluorescent-labeled azolectin liposome (cell-attached configuration) during aspiration, using a micropipette coated with 0.1% BSA.
Liposomes were aspirated by applying a negative pressure of −15 mmHg (ramp) and then the pressure was released stepwise.

Movie S5

Bavi et al. www.pnas.org/cgi/content/short/1409011111 15 of 16

Movie S6. Simulation of the movement of a liposome patch (cell-attached configuration) and stress distribution during micropipette aspiration, using the
finite-element method. The material model is visco-hyperelastic.

Movie S6

Movie S7. Simulation of the movement of a liposome patch (excised patch configuration) and stress distribution during patch clamping, using the finite-
element method. The material model is visco-hyperelastic.

Movie S7

Bavi et al. www.pnas.org/cgi/content/short/1409011111 16 of 16