Biophysical Journal Volume 106 March 2014 10331043 1033

Local Viscoelastic Properties of Live Cells Investigated Using Dynamic and

Quasi-Static Atomic Force Microscopy Methods

Alexander Cartagena and Arvind Raman*

School of Mechanical Engineering and Birck Nanotechnology Center, Purdue University, West Lafayette, Indiana

ABSTRACT The measurement of viscoelasticity of cells in physiological environments with high spatio-temporal resolution is a
key goal in cell mechanobiology. Traditionally only the elastic properties have been measured from quasi-static force-distance
curves using the atomic force microscope (AFM). Recently, dynamic AFM-based methods have been proposed to map the local
in vitro viscoelastic properties of living cells with nanoscale resolution. However, the differences in viscoelastic properties esti-
mated from such dynamic and traditional quasi-static techniques are poorly understood. In this work we quantitatively recon-
struct the local force and dissipation gradients (viscoelasticity) on live fibroblast cells in buffer solutions using Lorentz force
excited cantilevers and present a careful comparison between mechanical properties (local stiffness and damping) extracted
using dynamic and quasi-static force spectroscopy methods. The results highlight the dependence of measured viscoelastic
properties on both the frequency at which the chosen technique operates as well as the interactions with subcellular components
beyond certain indentation depth, both of which are responsible for differences between the viscoelasticity property maps
acquired using the dynamic AFM method against the quasi-static measurements.

INTRODUCTION
Many recent efforts in cell mechanobiology (14) aim to
quantitatively measure the mechanical properties of living
cells and relate them to cell structure and function. Cells
engage in complex processes changing the viscoelastic
response (59) of the cell membrane, cytoskeleton, and
cytoplasm. These changes are often heterogeneous in spatial
extent within the cell, and change with time. The measure-
ment of the progressive spatio-temporal variations in visco-
elastic properties within living cells in their native
physiological liquid environments could shed important
insight into cellular processes such as morphogenesis
(10,11), mechanotransduction (1214), migration/locomo-
tion (1518), metastasis (1925), apoptosis (26,27), aging
(27), focal adhesion (18,2832), disease progression (33),
and drug-cell interactions (3436).
The atomic-force microscope (AFM) is unique among
other cell mechanical measurement techniques (1,6,37,38)
in its ability to measure the local force and dissipative gra-
dients as well as map them across the cell surface with sub-
10-nm resolution. As a result, the AFM allows researchers
to develop quantitative methods to map the local mechanical
properties of living cells. The standard force-volume
method is the most widely used imaging method to extract,
simultaneously, the topography and mechanical properties
of a live cell. It is based on the acquisition of slow-speed,
quasi-static force-distance curve measurements on a grid
of points defined by the user. The mechanical properties
in force-volume maps are extracted offline for each recorded
force curve by fitting to an analytical tip-sample contact
Submitted November 12, 2013, and accepted for publication December 31,
2013.
*Correspondence: raman@purdue.edu
Editor: Daniel Muller.
2014 by the Biophysical Society
0006-3495/14/03/1033/11 $2.00
model. Another conventional way to evaluate the visco-
elastic properties of live cells is through the acquisition of
force-distance curves by applying a rectangular load-relax-
ation (stress-relaxation) pattern (21,39). Several groups
have estimated the local mechanical properties using these
methods on live bacteria (4044), and eukaryotic cells
(15,21,34,39,4450) in their native liquid environment;
however, the methods are low-speed and low-resolution,
limiting their potential for high throughput biomechanical
assay of cells.
To address some of the speed and spatial resolution chal-
lenges in conventional AFM techniques for mapping cell
mechanical properties, a method called multi-harmonic
AFM was recently reported (4) that employed amplitude-
modulation atomic force microscopy (AM-AFM), a tech-
nique allowing achievement of sub-10-nm resolution
high-speed mapping of local nanomechanical properties of
live cells in physiological conditions. However, the effective
local properties (loss and storage modulus) over the nuclear
region were found to be generally 35 times larger
compared to values acquired from quasi-static force-dis-
tance curves. Understanding the basic reasons behind the
differences in measured properties using the two methods
is key to developing quantitative tools to measure cell visco-
elasticity using dynamic AFM.
To address this fundamental issue, in this article we study
the local force gradient and damping on live rat fibroblast
cells in buffer solution, measured as a function of indenta-
tion using a soft AFM microcantilever excited by Lorentz
force near its natural frequency (~78 kHz). We find that,
at small indentations, the dynamically measured force
gradient is ~3 times that of the statically measured one,
showing the frequency-dependence. However, the ratio of
the dynamic to static force gradients begins to increase
http://dx.doi.org/10.1016/j.bpj.2013.12.037
1034
when the AFM tip indents the cell sufficiently to interact
with the nuclear region of the cell and its associated cyto-
skeletal structure, nuclear envelope, and additional subcel-
lular elements. On the peripheral region, the indentation is
much smaller and there is less difference between the
measured static and dynamic force gradients. On the other
hand, the effective contact viscosity is less near the nuclear
region than the peripheral regions.
We conclude that, when using AM-AFM with directly
excited probes, the mapped repulsive force gradient appears
larger on the nuclear region compared to the peripheral re-
gion for two different reasons:
1. Because larger forces need to be applied over the nuclear
region to achieve the same oscillation amplitude as on the
peripheral regions, this requires the AFM tip to be pressed
sufficiently into the cell membrane to interact primarily
with the nuclear complex and cytoskeleton in the nuclear
region.
2. The inherent viscoelasticity of the cell leads to a different
mechanical response at the higher frequency of oscilla-
tion in the dynamic AFM method compared to the
quasi-static method.
The findings suggest that, at least on live fibroblast cells,
maps of material properties that were created using both the
force-volume method and the dynamic AFM method are
not directly comparable because of the indentation- and
frequency-dependence.
See the Supporting Material for additional details
on the text.
MATERIALS AND METHODS
Sample preparation
Rat fibroblast cells (ATCC CRL-1213; American Type Culture Collection,
Manassas, VA) were cultured in Dulbeccos modified Eagles medium
(Invitrogen, Carlsbad, CA) containing low D-glucose (1000 mg l
1), 10%
Fetal Bovine Serum (Invitrogen), 1% Penicillin-Streptomycin (Invitrogen),
and 0.1% Amphotericin B (Sigma-Aldrich, St. Louis, MO). The fibroblast
cells were grown to ~75% confluence in 75 cm2 flasks. The medium was
removed every three days, and cells were subcultured every seven days at a ra-
tio of 1:8. For the experiments, the cells grown on plastic flasks were trypsi-
nized with 0.5% Trypsin/EDTA solution (Invitrogen) and the cell
suspension was deposited on polystyrene-plasma-treated 60  15 mm petri
dishes (BD Falcon, Franklin Lakes, NJ) precoated with 0.1% gelatin in
water (STEMCELL Technologies, Vancouver, British Columbia, Canada).
The cells were planted/grown on the dish 12 days before experiments and
kept in an incubator at 37 C in a 5% CO2 atmosphere to ensure complete
spreading.
Theory of dynamic AFM material property
mapping and spectroscopy in amplitude
modulation AFM on live cells
It has been shown in a previous work (4) that harmonic signals (multi-har-
monic observables) of cantilever vibration can be combined to extract local
nanomechanical properties once an analytical tip-sample interaction model
Biophysical Journal 106(5) 10331043
Cartagena and Raman
is prescribed. In this work, we do not focus on extracting constitutive
material properties like storage or loss modulus from the amplitude and
phase-shift data of the oscillating cantilever. This would require us to use
well-validated elastic contact mechanics models derived from the literature
such as that of Hertz (4,15,20,34,46), Sneddon (51), or bottom effect cone
correction (BECC) (from Gavara and Chadwick (52)). However, these
models have only been validated on live cells using quasi-static indentation
curves, and it remains an open question as to which models to choose for
dynamic AFM in which the cantilever oscillates in the range of 5
10 kHz. Rheological models for live eukaryotic cells (51) generally do
not take into account the heterogeneity of nuclear and peripheral regions.
Although three-element models have been used for cell viscoelasticity mea-
surements (9,21,53,54), we use here the two-element Kelvin-Voigt model
for the following reasons:
1. The dynamic AFM method we discuss here uses two observables, the
cantilever harmonic amplitude and phase to determine the sample visco-
elasticity. As a consequence, only two unknown viscoelastic parameters
can be determined at each Z location (during approach curves) or at any
pixel (during a scan). 2. Thus, the single frequency method presented
here can only treat a viscoelastic model with two unknown parameters
such as the Kelvin-Voigt element (spring and dashpot in parallel) and the
Maxwell element (spring and dashpot in series). 3. Of these two, the
Kelvin-Voigt model is physically relevant because the oscillation time of
the experiments (8 kHz 125 ms) is faster than the live fibroblast cell relax-
ation time (~0.1 s) (55).
In light of this, we convert the amplitude and phase-shift data of the oscil-
dynamic
lating cantilever into local dynamic repulsive-force gradient ksample and
dynamic
damping csample values. Together these are the parameters of an equiva-
lent Kelvin-Voigt element and can be regarded as the local viscoelastic
properties of the cell.
The theory behind this is described below. Assuming that the Lorentz
force cantilever driving frequency is near the first resonance frequency,
the single degree-of-freedom equation of motion governing the tip motion
q(t) when it oscillates far from the sample surface is:
q 1 Fmag sinudr t
q_ q ; (1)
u2far ufar Qfar kcant
where ufar is the cantilever frequency (rad/s), Qfar is the quality factor far
from the sample surface, udr is the cantilever drive frequency (rad/s),
Fmag is the magnitude of the magnetic excitation force, kcant is the calibrated
cantilever spring constant, and q_ is the tip velocity. Solving the steady-state
vibration response q(t) A1 sin(udrt f1), it can be easily shown that when
udr is tuned to the peak amplitude of the resonance curve far from the sur-
face, the following expressions hold:
q
udr=u 1
1=2Q2 ;
far far
Fmag Qfar
A1far q (2)
kcant 1
1= 2 ;
4Qfar
p
tan f1far 4Q2far
2 :
Here, A1far is the oscillation amplitude and f1far is the phase lag far from the
sample surface. Note that by Eq. 2, when tuning the cantilever far from the
sample, the phase lag at the frequency of peak amplitude with low Q, say 2,
is not to be set to 90 ; instead, it should be set to f1far 75 . Thus, when the
drive is tuned at the peak amplitude of resonance far from the sample we
must have
kcant A1far q
Fmag 1
1=4Q2 :
Qfar far
Viscoelasticity of Live Cells Investigated Using AFM 1035

For soft cantilevers in liquids, their resonance frequencies (measured in damping (viscoelastic properties) at that particular indentation value. More-
dynamic 0 00
rad/s) and Q factors near the sample (i.e., unear and Qnear) are different from over, ksample and cdynamic
sample used here are equivalent to K and K of Mahaffy
the values measured far from the surface (i.e., ufar and Qfar) because of the et al. (58). Substituting Eq. 6 into Eq. 7, substituting the resulting expres-
hydrodynamic coupling between the cantilever and the sample (56). As a sion for Fts into Eq. 4, and taking into account Eqs. 2, 3, and 5, one can
result, the oscillation amplitude and the phase lag near the surface (i.e., match the harmonic terms of each side of the equation of motion (Eq. 4)
A1near and f1near) are different from their values far from the sample (i.e., to yield the following:
A1far and f1far). As shown by Raman et al. (4), the quantities near and far
!
from the sample are related by
kcant A1far kcant A1far
!2 q dynamic
ksample cosf1
cosf1near
A1far 1
1=4Q2 Qfar A1 A1near Qfar
udr far
1
cosf1near ; q
unear A1near Qfar  1
1=4Q2 ;
q (3) far

A1far 1
1=4Q2
udr far
sinf1near :
Qnear unear A1near Qfar
It is important to note that the hydrodynamic coupling to the surface is
different on live fibroblast cells that are tall (~24 mm in height) and soft
(11000 kPa), as compared to when the cantilever is located at the same
height above a rigid surface. For example, when the cantilever is excited
at resonance far from the sample so that f1far 74 and with a free
amplitude A1far ~ 9 nm, we find that when brought close to the sample,
just before the tip-sample interactions begin, A1near /f1near values on
the gelatin-coated dish and live cell are typically 2.5 nm/95 and
3.5 nm/89 , respectively. This clearly demonstrates that the squeeze-film
hydrodynamic effect is much stronger on the substrate than on the live
cell. This hydrodynamic correction is essential to account for while
measuring the viscoelastic properties of live cells. This correction is rela-
tively straightforward to perform for the case when the cantilever is
excited at very low frequencies (57), where the cantilever inertia is not
important. Here we have presented the correction for the case when the
cantilever is excited near its resonance frequency, and accounts for both
the added damping and the added mass of the fluid in the near vicinity
of the sample surface.
When the cantilever is brought closer to the sample so that it interacts
with the soft cell surface, the equation of motion becomes:
q 1 _
Fmag sinudr t Fts Z q; q
q q_ ;
2
unear unear Qnear kcant
(4)
where Fts is the tip-sample interaction force, and Z is the distance between
the tip and sample assuming that the cantilever is rigid or unbent. As
observed in Raman et al. (4) and also demonstrated in Results and Discus-
sion, the cantilever motion while interacting with live cells is dominated by
the zeroth and first harmonics, leading to the following form of the tip
motion:
qt A0 A1 sinudr t
f1 : (5)
To convert zeroth and first harmonics observables into local viscoelastic
properties, we first annotate the tip indentation d(t) into the sample as
dt
Z q
Z
A0
A1 sinudr t
f1 : (6)
As described in Results and Discussion and in Raman et al. (4), the tip
oscillation amplitude A1 is much smaller compared to the net average
indentation d0
(Z A0) while imaging live cells in AM-AFM. Next,
we describe the interaction forces as a Taylor series expansion in (d d0)
and discard higher-order terms,
Fts Fts d0 ksample
dynamic
d
d0 cdynamic _
sample d;
dynamic
where ksample and cdynamic
sample represent the parameters of a Kelvin-Voigt
element, specifically the conservative local force gradient (stiffness) and
!
kcant A1far kcant A1far
cdynamic
sample sinf1
sinf1near
Qfar A1 udr A1near Qfar udr
q
 1
1=4Q2 : (8)
far
These equations allow the conversion of the harmonic observables (A0, A1,
and f1) into the local viscoelastic properties, i.e., the dynamic repulsive
dynamic
force gradient ksample and damping cdynamic
sample , while taking into account
the near-surface hydrodynamic corrections.
It is important to reiterate the different assumptions under which the
above equations are accurate:
Assumption 1
The cantilever is driven directly, not acoustically or via sample excitation,
so that the cantilever has a well-defined transfer function (59,60) with the
Lorentz force being the only source of excitation without interference
from fluid-borne excitation (61) that arises when the either the cantilever
or the sample are excited using piezoelectric transducers.
Assumption 2
The tip is in continuous contact with the sample and that the oscillation
amplitude is small relative to the net indentation of the tip into the sample.
Assumption 3
The eigenmode of the cantilever oscillating away from the sample remains
unchanged when compared to being pressed against the sample. This
assumption in fact enables us to state that the kcant does not change when
coupled to the sample. This assumption is known to be correct only when
the contact stiffness/repulsive gradient is much smaller than the cantilever
stiffness (nominally 0.09 N/m) (62).
Assumption 4
We need to recognize, as stated before, that the hydrodynamic correction is
different on the live cell compared to that on the gelatin-coated glass
surface.
Typically, A1near and f1near on a gelatin-coated dish are ~1.54 nm/95
98 , whereas on live cells the values are ~25 nm/8589 on nucleus and
~1.83.5 nm/9295 on periphery. However, Eq. 8 uses a single value of
A1near and f1near. Because we are more interested in the properties of the
cell than the gelatin surface, we choose to process all the data using
A1near and f1near on top of the cell. This implies that the predicted repulsive
gradients on a gelatin-coated dish are systematically larger than the actual
value due to the hydrodynamic effect. However, we expect the correct
values to be mapped on the live cell. All computations and data processing
(7)
were performed using the software MATLAB (The MathWorks, Natick,
MA). For each set of multi-harmonic observable (A0, A1, and f1) curves
dynamic
the viscoelastic properties maps (ksample and cdynamic
sample ) were extracted by us-
ing Eq. 8.

Biophysical Journal 106(5) 10331043

1036 Cartagena and Raman

Quasi-static force spectroscopy

In quasi-static force spectroscopy, an unexcited AFM probe approaches and
is retracted from the sample surface and the deflection of the microcantile-
ver tip q is recorded as a function of Z-piezo extension. The resulting
force-distance (F-Z) curve is converted to a force-indentation (F-d) curve
by d Z q. The slope of the F-d curve is the local repulsive force gradient
static
ksample and can be calculated as a function of the indentation d at each point.

AFM experimental protocol

The experimental setup is as follows: Before beginning the AFM imaging,
cells were rinsed thoroughly with 2 mL PBS twice and then 2 mL of fresh ster-
ile PBS was added to simulate near-physiological conditions when imaging.
All live cell AFM imaging and viscoelastic measurements were performed
with a model No. MFP-3D-Bio AFM (Asylum Research, Santa Barbara,
CA) mounted on a model No. IX-71 inverted optical microscope (Olympus,
Melville, NY) which was itself placed on a vibration table inside an acoustic
isolation enclosure. This allows easy positioning and monitoring of cells. We
used soft microcantilevers (model No. BL-TR400PB microcantilever;
Olympus) with a nominal spring constant of 0.09 N/m, and nominal tip radius
of 42 nm (512 nm). The iDrive (Asylum Research) AC mode was used for
the experiment that uses Lorentz force excitation to apply an oscillating
driving force directly to the microcantilever. The importance of using such
directly excited probes for quantitative measurements in liquids has been dis-
cussed before (61,63). During each experimental measurement, we first
localized the cells using the inverted optical microscope, and checked for
viability by means of cell morphology and extracellular matrix spreading/
anchorage.
Before doing experiments on live fibroblast cells to measure quantitative
local viscoelastic properties, certain calibrations need to be performed. The
FIGURE 1 (a) Schematic of a Lorentz force excited microcantilever in-
AFM probe must be engaged on a stiff substrate (mica) and a force-distance
teracting with a live fibroblast cell by dynamic spectroscopy. In conven-
curve obtained to calculate the optical photodiode deflection sensitivity.
tional quasi-static spectroscopy a static cantilever is ramped toward the
Then, the probe is withdrawn from the stiff sample surface and the canti-
soft cell and retracted to record an F-Z curve. However, in dynamic spec-
lever spring constant was calibrated by the thermal noise method (64,65).
troscopy a microcantilever is excited at near the resonant frequency and
Typical estimated values for effective cantilever spring constants were in
moved toward the sample and then retracted to record three channels of in-
the range of 0.030.1 N/m and Q-factors were in the range of 1.52.
formation: mean deflection (A0), first harmonic amplitude, and phase lag
The AFM cantilever probe was directly driven (Lorentz force excitation) at
(A1 and f1). (b) Tune curves performed at 4 mm away and in contact
the peak of the resonance curve of the fundamental flexural mode (typically
with the cell with a force of ~1 nN show that the transfer functions of a fully
7.59 kHz) and then engaged to the live fibroblast cell in its buffer solution.
vibrated cantilever changes when interacting with the cell, resulting in
Phase contrast optical imaging was used to identify a viable cell on which to
changes in amplitude and phase that are closely related to the cell visco-
perform AFM imaging and spectroscopy. We then image the topography of
elastic properties. Therefore, an equivalent spring-dashpot model for the
the cell using AM-AFM while simultaneously acquiring the multi-harmonic
cantilever in contact with the cell can be used to extract the viscoelastic
data. The free amplitudes far from the sample (A1far) are typically in the range
properties. To see this figure in color, go online.
of 316 nm. The phase lag (f1far), from the low quality factor Qfar ~ 1.8, is set
to 74 using the third expression in Eq. 2. When brought closer to the cell, the
amplitude reduces to A1near ~ 25 nm and the phase lag changes to f1near ~85 multi-harmonic theory described above to reconstruct frequency- and inden-
100 just before the probe tip begins to interact with the cell surface. This tation-dependent local force and dissipation gradients on live fibroblast cells.
change in amplitude and phase from far to near the sample is a consequence Here we present data acquired on three fibroblast cells from different cell
of the strong squeeze-film hydrodynamic effect between the cantilever and cultures on different days using different AFM probes (10 repeats). The re-
the cell surface (66). The AM-AFM images are typically acquired at setpoints sults are consistent and repeatable. Many additional cells have been studied,
A1/A1near ~ 7080% at scan speeds of 0.25 Hz. The data were filtered using but we present those for which comprehensive quasi-static and dynamic
low-pass filters with cutoff frequencies of 500 Hz for mean deflection (0th data sets were acquired.
harmonic), and 1 kHz for 1st harmonic data. The images were rendered
with the software IGOR PRO 6.2 (WaveMetrics, Lake Oswego, OR).
After obtaining the AM-AFM topography and multi-harmonic variables
images, three locations of interest (nucleus, periphery, and gelatin-coated RESULTS AND DISCUSSION
dish) are identified. Subsequent approach curves were performed at these Cantilever response while interacting with a live
points to measure their static and dynamic viscoelastic properties. Fig. 1
a shows the schematic diagram of the experiment. Ten quasi-static and dy-
cell
namic approach curves with Z-piezo frequency of 0.25 Hz (linear speed of
Fig. 1 b shows the cantilevers frequency response to Lorentz
750 nm/s) were performed on the cell locations of interest. For quasi-static
curves, the local repulsive force gradient is simply the local slope on the F-d
force excitation off and in contact with the live cell nucleus
curve. For dynamic approach curves, the acquired multi-harmonic and periphery as well as the gelatin-coated dish. This clearly
observables curves (A0, A1, and f1) against Z were used together with the shows that the cell adds stiffness and damping to the

Biophysical Journal 106(5) 10331043

Viscoelasticity of Live Cells Investigated Using AFM 1037

cantilever oscillation, confirming the validity of modeling the
interaction as a Kelvin-Voigt viscoelastic element consisting
of a repulsive force gradient and viscous dashpot model.
Before describing the detailed results, it is worth high-
lighting repeatable key features one can observe from
dynamic approach curves on the cell:
Our first important finding is that the AFM tip requires
large net sample indentation to reduce the cantilever ampli-
tude to a setpoint value A1 not only on the cell nuclear region
but also on the periphery. To demonstrate this we measured
dynamic approach curves at different locations in the image
and observe the mean indentation, oscillation amplitude
reduction, and phase-lag shift (Fig. 2). In Fig. 2 it can be
clearly seen that as the cantilever tip is pressed down against
the top of the cell, the first harmonic amplitude A1 decreases
very slowly whereas the zeroth harmonic amplitude A0 in-
creases, rapidly becoming >A1. From Fig. 2 b it can be
deduced that on top of the cell nucleus it takes ~400 nm
of Z-piezo extension beyond initial interaction with the
cell surface to reduce the AFM probe oscillation amplitude
to A1 ~ 0.75 A1near. Also, from Fig. 2, c and d, on the cell
periphery region and gelatin-coated dish it takes ~50 and
25 nm of Z-piezo extension, respectively, to reduce the
amplitude to a value of A1 ~ 0.75 A1near. Thus, at the imaging
setpoint A1 ~ 0.75 A1near, indentation/amplitude ratio (d/A1)
is >~100 times over that of the nucleus, to ~510 times over
the peripheral region and ~25 times over the gelatin-coated
dish. In other words, for the microcantilever to reach a spe-
cific amplitude reduction, it requires a large net indentation
into the living cell. This implies that when imaging, the
AFM tip is in permanent contact with a large net indentation
dynamic
allowing the use of linear theory to extract ksample and
dynamic
csample , as described in Materials and Methods.
The second important finding is that the resonance fre-
quency of the cantilever shifts significantly due to hydrody-
namic squeezing effects from initial measurements taken
while tuning to when it begins to interact with the sample.
This can be observed in Fig. 2, bd, by noting that the phase
lag before interaction begins is ~85100 even though while
tuning (at ~4 mm far from the cell surface) the phase lag was
set to 75 whereas the drive frequency was chosen to match
the resonance peak. As mentioned, we use Lorentz force exci-
tation to drive the microcantilever near resonance; therefore,
this phase shift is due to near-surface hydrodynamic effects
that have been considered in the theory presented before.
Depth-dependent viscoelastic properties
We then measured the viscoelastic properties of live
fibroblasts cells using the dynamic and quasi-static AFM

FIGURE 2 Dynamic approach curves performed on different locations across the sample. (a) Three-dimensionally rendered topography image of a live
fibroblasts. (Crosses) Measurement locations. (bd) Dynamic-approach curves for multi-harmonic observables (A0, A1, and f1) acquired in the nucleus and
periphery of a live fibroblast cell and the stiffer gelatin-coated petri dish, respectively. (Green vertical lines) Typical imaging setpoint ratios of 85 and 75%.
(Curves) Behavior of the multi-harmonic amplitudes and phase (A0, A1, and f1) as the microcantilever moves toward and interacts with different regions of a
live fibroblast cell in a soft repulsive regime. It is clear that the strongest and most easily accessible harmonic signals that reflect local viscoelastic properties
are those from the 0th and 1st harmonics for live cells. Topography image was acquired using a Lorentz force excited Olympus (Melville, NY) microcanti-
lever (BL-TR400PB: kcant 0.083 N/m, Q 1.9, udr 7.79 kHz, and A1far 8.5 nm). To see this figure in color, go online.

Biophysical Journal 106(5) 10331043

1038
spectroscopy methods as described in Materials and
Methods, and obtained the following key results:
When comparing the static and dynamic force gradients
(at 0*udr and udr, respectively) on the cell nucleus as a func-
tion of indentation (Fig. 3 a), the dynamic values are three
times larger than those measured statically until a critical
depth of ~100 nm, beyond which the ratio of dynamic force
dynamic static
gradient ksample to ksample becomes larger. However, in the
dynamic
case of the periphery shown in Fig. 3 b, ksample and
static
ksample are comparable and vary in a similar manner with
average indentation. These two plots are the averages of
10 repeats at the same location on the cell. The data are
highly repeatable with a typical variance of 10% over the
10 repeats for all the cells discussed in this article. From
dynamic static
Fig. 3, a and b, we note the ratio of ksample to ksample strongly
depends on indentation depth on the nuclear region but not
significantly on the cell periphery.
FIGURE 3 Frequency- and indentation-depth dependence of the visco-
elastic properties measured on a live fibroblast cell. The dynamic visco-
elastic properties have been measured at a fixed high excitation
dynamic
frequency of udr 7.79 kHz. (a and b) Dynamic ksample (blue) and static
static
ksample (red) force gradients at different mean indentations d0 on the nuclear
dynamic static
and peripheral region of a live fibroblast cell. Increase of ksample /ksample
force gradients after a critical indentation depth is observed in the cell nu-
cleus but not in the periphery. (c) Damping cdynamic
sample at different mean in-
dentations d0 on the nucleus and periphery region of the cell. The plots
are the averages of 10 dynamic and static approach curves performed in
each location on the cell. (Insets) Images are the topography of the cell
probed. (Crosses) Area where the measurements were made. To see this
figure in color, go online.
Biophysical Journal 106(5) 10331043
Cartagena and Raman
Sample dynamic damping cdynamic sample also strongly depends
on indentation depth becoming larger as the tip is pressed
down on the cell. Dynamic damping was also extracted for
the two regions (Fig. 3 c) showing that there is a variation
in damping across the cell. The plots in Fig. 3 c are the aver-
ages of 10 repeats at the same location on the cell. The data
are highly repeatable with a typical variance of 10% over the
10 repeats for all the cells discussed in this article. The damp-
ing behavior for the two regions shows an intrinsic critical
indentation depth where it deviates from zero value and in-
creases monotonically. Finally, we note that the damping is
larger on the periphery than the nucleus.
The viscoelastic properties of live fibroblast cells
exhibited a weak dependence on the operating oscillation
amplitude. Dynamic force-indentation curves at different
cantilever free oscillation amplitudes from 5 to 25 nm for
a maximum loading force of ~2.5 nN were performed and
no significant difference in dynamic viscoelastic properties
was observed. This suggests that dynamic viscoelastic
properties measurements are independent of the free oscilla-
tion amplitude.
Fig. 4 shows the extracted viscoelasticity properties of
three different fibroblasts cells from different cell cultures.
In the nuclear region of all cells the indentation depth
dynamic
dependency is observed, where the ratio of ksample
static
to ksample increases beyond a certain indentation while on
dynamic static
the periphery ksample and ksample follow a similar depth
dynamic
dependence. Also, csample on the nucleus and periphery
of all cells demonstrates the indentation-depth dependency.
Our observation that the ratio of dynamic and static stiff-
ness increases beyond a certain indentation level on the
nuclear region, but not on the periphery, could arise from
the following possibilities:
As the AFM probe approaches and is being pressed
down into the cell, the AFM tip initially interacts with
brushes and the brush-type structures layer (combination
of glycocalyx, microvilli, and microridges) on the cell
membrane surface (67,68), and it is possible that the visco-
elastic properties of these extramembrane components are
very different from that of the underlying subcellular
components leading to the observed depth-dependent
divergence of the static and dynamic stiffness. However,
if this were the reason for the observations, the depth-
dependent divergence of static and dynamic stiffness
should be detected not only in the nuclear region but across
the entire cell, which is not the case.
Another explanation for the observed divergence between
the dynamic and static force gradients on the nuclear region
could be that the increase in dynamic and static force
gradient ratio begins when the tip indents sufficiently to
interact with the nuclear complex. The nuclear complex is
expected to have quite different viscoelastic properties
(69) compared to the cytoskeleton that lays just beneath
the cell membrane. This hypothesis is consistent with
the observation that the indentation-depth-dependent
Viscoelasticity of Live Cells Investigated Using AFM 1039

dynamic static
FIGURE 4 Viscoelastic properties response for different fibroblast cells. (a) Dynamic ksample (solid lines) and static ksample (dashed lines) force gradients
and (b) damping cdynamic
sample as a function of mean indentations d 0 on the nuclear region. (c and d) Same measurements are collected on the peripheral region. The
dynamic static
plots represent the averages of 10 dynamic and static approach curves performed in each specified location on each cell. Increase of ksample /ksample force
gradients in the nucleus is observed for every cell, which confirm our major finding that there are difference in viscoelastic properties in the nucleus and
is initially frequency-dependent, but after a critical indentation depth is frequency- and indentation-dependent. The dynamic viscoelastic properties of
the cells have been measured using three different cantilevers at a fixed high excitation frequency udr and with kcant and Q. Cell No. 1: 7.79 kHz,
0.083 N/m, and 1.9. Cell No. 2: 7.69 kHz, 0.076 N/m, and 1.8. Cell No. 3: 8.18 kHz, 0.076 N/m, and 1.8. (Insets) Topography images of the cells. (Crosses)
Locations where the measurements were made. To see this figure in color, go online.

divergence of static and dynamic stiffness is observed only
over the nuclear region.
Implications on mapping in vitro local
viscoelastic properties
Now that we have studied the viscoelastic properties as a
function of indentation depth at high excitation frequency
on different cell locations, we turn our attention to visco-
elastic property maps that can be easily extracted in AM-
AFM scan over live cells. In Fig. 5 ,we show a series of
AM-AFM images acquired over the live fibroblast cell in
physiological conditions. The multi-harmonic observables
(A0 and f1) are converted to detailed local material prop-
erty maps of effective dynamic repulsive force gradient
dynamic
ksample and damping cdynamic as described earlier. We
sample
also convert these maps into loss tangent tand (Fig. 5 h).
The loss tangent is the ratio between the energy dissipated
and the energy stored in one cycle of oscillation in contact
with the sample (70). In this case, the loss tangent is
defined as
 
tan d cdynamic  ksample
sample u dr =k dynamic :
sample
dynamic
The map in Fig. 5 f suggests that ksample over the
nuclear region is greater than in the peripheral part of
the cell and the gelatin-coated dish. Moreover, from
Fig. 5 g, the cdynamic
sample over the nuclear region is lower in
magnitude than the periphery and dish. This result was
also observed in Raman et al. (4). On the other hand,
many prior works have shown that the gelatin-coated
dish has the largest elastic modulus, followed by the pe-
ripheral region, and lastly the nucleus (45,47,48). Also,
it was expected that the cell nucleus should be the surface
with higher damping compared to the peripheral region
and the dish. This apparent contradiction between the
maps in Fig. 5 and known properties can be explained
in terms of the force spectroscopy results presented earlier.
Fig. 6 shows a repulsive gradient with respect to applied
force plot on the nucleus of a live fibroblast cell. It
dynamic
can be clearly seen that, at forces >200 pN, ksample in-
creases rapidly. We actually apply forces >750 pN when
imaging in AM-AFM over the nuclear region, and as a
dynamic
result we observe large ksample . However, on the periph-
eral region (Fig. 6), we apply very small forces <150
pN while imaging in AM-AFM yielding relatively small
dynamic
, lesser than the nucleus. At such low forces, the
force gradient on the gelatin is also very small in

Biophysical Journal 106(5) 10331043

1040 Cartagena and Raman

FIGURE 5 Multi-harmonic AFM images acquired on a live fibroblast cell in physiological conditions. (a) Optical phase contrast micrograph using a 10
objective; (b) topographic image; (c) 0th harmonic amplitude map A0; (d) first harmonic phase map f1; (e) applied mean normal force map F0ts; and (fh)
dynamic
maps of local dynamic force gradient ksample , damping cdynamic th
sample , and loss tangent tand, respectively, extracted from the measured 0 and first harmonic data
using the theory described in the text. Surprisingly, the loss tangent values of the cell periphery and dish are larger than the cell nucleus. Images were acquired
using a Lorentz force excited Olympus (Melville, NY) microcantilever (BL-TR400PB: kcant 0.083 N/m, Q 1.9, and udr 7.79 kHz). The scale bar on
images represents 10 mm (size; 60  60 mm2, pixels; 256  256). To see this figure in color, go online.

comparison to the force gradient on the cell at much
higher forces.
Comparison with microrheological models of live
cells
It is interesting to compare the results in this work to prior
works on the rheology and viscoelasticity of live cells. Pre-
vious work by Alcaraz et al. (51) has used a power-law fre-
quency-dependent structural damping model to estimate the
complex shear modulus G*(u) of a live cell. This model has
been validated systematically at low frequencies (~0.1
100 Hz), although not at the high frequencies (510 kHz)
as presented in this work. A benchmark calculation using
values reported by Alcaraz et al. (51) for A549 human
lung epithelial cells, suggests Gstorage (8 kHz)/Gstorage
dynamic
(0.25 Hz) ~ 9.8. However, the ratio of ksample udr to
static
ksample is smaller than the one predicted by the rheological
power-law model ~3.56.5. There could be several reasons
for this:
1. The rheological models do not take into account the
indentation depth dependence;
2. Cell microrheological models have not been validated for
high frequencies; and
3. A549 cells lack the cytoskeletal element concentration
and organization that characterize fibroblasts.
The viscoelastic properties behavior dependence on inden-
tation depth over more limited frequency ranges (20
400 Hz) of polymer gels and NIH3T3 fibroblast cells has
been shown before by Mahaffy et al. (57,58). However,
the divergence between low and high frequency as a func-
tion of sample indentation has not been presented before.
dynamic
Moreover, to the best of our knowledge, ksample data on
the nucleus and periphery regions reported in this work
are the first AFM high-frequency- and indentation-depen-
dent (~8 kHz) nanorheological measurements in live cells.
This finding has a major impact when comparing quantita-
tive results obtained from quasi-static and high-frequency
dynamic methods.
CONCLUSIONS
We have demonstrated the ability of dynamic AFM to
quantify the nanorheological properties of live cells and

Biophysical Journal 106(5) 10331043

Viscoelasticity of Live Cells Investigated Using AFM
FIGURE 6 Plots are the averages of 10 dynamic force gradients of (a)
dynamic
ksample and (b) damping cdynamic
sample against the applied force performed on
the nucleus (blue), periphery (red) of the live fibroblast cell, and the
gelatin-coated petri dish (green). At forces higher than ~800 pN on the
cell nucleus the dynamic force gradient and damping becomes higher
than the periphery and dish, increasing rapidly as the cell indentation pro-
gressively increases. This explains the stiffness and damping contrast
between the peripheral and nuclear regions observed in Fig. 5s maps. To
see this figure in color, go online.
compared the dynamic and quasi-static methods to under-
stand the viscoelastic response of live fibroblast cells at
very low frequencies (quasi-static) and at the cantilever
resonance frequency (dynamic).
We have also found interesting differences between
viscoelastic measurements made using quasi-static and
dynamic AFM modes. On the fibroblasts nuclear region
the local dynamic force gradient is larger than the statically
measured one for small indentations (0250 nm), however,
beyond a certain critical indentation depth (say >250 nm)
the ratio of force gradients derived from the dynamic and
quasi-static methods increases rapidly due to interactions
with subcellular components such as the nuclear complex.
This depth dependency is also seen in the viscous damping.
However, on the peripheral parts of the fibroblasts the ratio
of dynamic and static force gradients does not change appre-
ciably with indentation.
Consequently, it is difficult to compare maps of visco-
elastic properties acquired using the quasi-static and
dynamic AFM modes. Specifically, when these properties
are mapped over live cells in buffer media using AM-
AFM, the indentation required to maintain constant oscilla-
tion amplitude changes from pixel to pixel because the
nanomechanical properties on the cell are heterogeneous.
Specifically, the tip indents the cell much more (300
1041
500 nm) over the nuclear region and much lesser on the pe-
riphery (~50 nm). The combination of two effects (i.e., high
frequency vibration and variation of indentation depth) per-
formed while imaging the cell leads to the dynamic repul-
sive force gradient on the nucleus to be generally greater
than those mapped using standard quasi-static methods.
These results confirm that dynamic AFM methods can in
fact be used for the quantitative mapping of viscoelastic
properties of subcellular components of nuclear and periph-
eral regions of live cells. However, the interpretation of
these properties and comparison with quasi-static AFM
measurements requires careful consideration of the fre-
quency- and indentation-dependence. We have demon-
strated the frequency- and indentation-dependence of local
viscoelastic properties of living cells by comparing the force
gradients determined from dynamic and quasi-static force
spectroscopy methods, reporting, for the first time to our
knowledge, measurements for high frequencies with an
elucidation of the biomechanical role of subcellular compo-
nents. This has significant relevance not only for cell mecha-
nobiology but also for AFM-based imaging and force
spectroscopy of live cells.
SUPPORTING MATERIAL
One table, seven figures, References (7173), and Supplemental informa-
tion are available at http://www.biophysj.org/biophysj/supplemental/
S0006-3495(14)00011-3.
The authors thank Prof. E. Nauman (School of Mechanical Engineering and
Weldon School of Biomedical Engineering, Purdue University) for
providing the rat fibroblast cells.
This research was supported in part by the National Science Foundation
Materials World Network under grant No. DMR 1008189.
REFERENCES
1. Suresh, S. 2007. Biomechanics and biophysics of cancer cells. Acta
Biomater. 3:413438.
2. Guck, J., F. Lautenschlager, ., M. Beil. 2010. Critical review: cellular
mechanobiology and amoeboid migration. Integr. Biol. (Camb). 2:
575583.
3. Azeloglu, E. U., and K. D. Costa. 2011. Atomic force microscopy in
mechanobiology: measuring microelastic heterogeneity of living cells.
Methods Mol. Biol. 736:303329.
4. Raman, A., S. Trigueros, ., S. A. Contera. 2011. Mapping nanome-
chanical properties of live cells using multi-harmonic atomic force
microscopy. Nat. Nanotechnol. 6:809814.
5. Putman, C. A. J., K. O. van der Werf, ., J. Greve. 1994. Viscoelasticity
of living cells allows high resolution imaging by tapping mode atomic
force microscopy. Biophys. J. 67:17491753.
6. Bausch, A. R., W. Moller, and E. Sackmann. 1999. Measurement of
local viscoelasticity and forces in living cells by magnetic tweezers.
Biophys. J. 76:573579.
7. Fabry, B., G. N. Maksym, ., J. J. Fredberg. 2001. Scaling the micro-
rheology of living cells. Phys. Rev. Lett. 87:148102.
8. Trepat, X., L. Deng, ., J. J. Fredberg. 2007. Universal physical
responses to stretch in the living cell. Nature. 447:592595.
Biophysical Journal 106(5) 10331043
1042
9. Vadillo-Rodriguez, V., and J. R. Dutcher. 2009. Dynamic viscoelastic
behavior of individual Gram-negative bacterial cells. Soft Matter.
5:50125019.
10. Ingber, D. E. 1993. The riddle of morphogenesis: a question of solution
chemistry or molecular cell engineering? Cell. 75:12491252.
11. Montell, D. J. 2008. Morphogenetic cell movements: diversity from
modular mechanical properties. Science. 322:15021505.
12. Wang, N., J. P. Butler, and D. E. Ingber. 1993. Mechanotransduction
across the cell surface and through the cytoskeleton. Science. 260:
11241127.
13. Ingber, D. E. 1997. Tensegrity: the architectural basis of cellular me-
chanotransduction. Annu. Rev. Physiol. 59:575599.
14. Ingber, D. E. 2003. Tensegrity I. Cell structure and hierarchical systems
biology. J. Cell Sci. 116:11571173.
15. Rotsch, C., K. Jacobson, and M. Radmacher. 1999. Dimensional and
mechanical dynamics of active and stable edges in motile fibroblasts
investigated by using atomic force microscopy. Proc. Natl. Acad. Sci.
USA. 96:921926.
16. Yamazaki, D., S. Kurisu, and T. Takenawa. 2005. Regulation of cancer
cell motility through actin reorganization. Cancer Sci. 96:379386.
17. Discher, D. E., P. Janmey, and Y. L. Wang. 2005. Tissue cells feel and
respond to the stiffness of their substrate. Science. 310:11391143.
18. Lo, C. M., H. B. Wang, ., Y. L. Wang. 2000. Cell movement is guided
by the rigidity of the substrate. Biophys. J. 79:144152.
19. Guck, J., S. Schinkinger, ., C. Bilby. 2005. Optical deformability as
an inherent cell marker for testing malignant transformation and met-
astatic competence. Biophys. J. 88:36893698.
20. Cross, S. E., Y. S. Jin, ., J. K. Gimzewski. 2007. Nanomechanical
analysis of cells from cancer patients. Nat. Nanotechnol. 2:780783.
21. Darling, E. M., S. Zauscher, ., F. Guilak. 2007. A thin-layer model for
viscoelastic, stress-relaxation testing of cells using atomic force micro-
scopy: do cell properties reflect metastatic potential? Biophys. J. 92:
17841791.
22. Cross, S. E., Y. S. Jin, ., J. K. Gimzewski. 2008. AFM-based analysis
of human metastatic cancer cells. Nanotechnology. 19:384003.
23. Faria, E. C., N. Ma, ., R. D. Snook. 2008. Measurements of elastic
properties of prostate cancer cells using AFM. Analyst (Lond.).
133:14981500.
24. Kumar, S., and V. M. Weaver. 2009. Mechanics, malignancy, and
metastasis: the force journey of a tumor cell. Cancer Metastasis Rev.
28:113127.
25. Goetz, J. G., S. Minguet, ., M. A. Del Pozo. 2011. Biomechanical re-
modeling of the microenvironment by stromal caveolin-1 favors tumor
invasion and metastasis. Cell. 146:148163.
26. Pelling, A. E., F. S. Veraitch, ., M. A. Horton. 2009. Mechanical dy-
namics of single cells during early apoptosis. Cell Motil. Cytoskeleton.
66:409422.
27. Gourlay, C. W., and K. R. Ayscough. 2005. The actin cytoskeleton: a
key regulator of apoptosis and aging? Nat. Rev. Mol. Cell Biol.
6:583589.
28. Ilic, D., Y. Furuta, ., T. Yamamoto. 1995. Reduced cell motility and
enhanced focal adhesion contact formation in cells from FAK-deficient
mice. Nature. 377:539544.
29. Pelham, Jr., R. J., and Y. Wang. 1997. Cell locomotion and focal adhe-
sions are regulated by substrate flexibility. Proc. Natl. Acad. Sci. USA.
94:1366113665.
30. Balaban, N. Q., U. S. Schwarz, ., B. Geiger. 2001. Force and focal
adhesion assembly: a close relationship studied using elastic micropat-
terned substrates. Nat. Cell Biol. 3:466472.
31. Riveline, D., E. Zamir, ., A. D. Bershadsky. 2001. Focal contacts as
mechanosensors: externally applied local mechanical force induces
growth of focal contacts by an mDia1-dependent and ROCK-indepen-
dent mechanism. J. Cell Biol. 153:11751186.
Biophysical Journal 106(5) 10331043
Cartagena and Raman
32. Matthews, B. D., D. R. Overby, ., D. E. Ingber. 2004. Mechanical
properties of individual focal adhesions probed with a magnetic micro-
needle. Biochem. Biophys. Res. Commun. 313:758764.
33. Maciaszek, J. L., and G. Lykotrafitis. 2011. Sickle cell trait human
erythrocytes are significantly stiffer than normal. J. Biomech. 44:
657661.
34. Rotsch, C., and M. Radmacher. 2000. Drug-induced changes of cyto-
skeletal structure and mechanics in fibroblasts: an atomic force micro-
scopy study. Biophys. J. 78:520535.
35. Bhadriraju, K., and L. K. Hansen. 2002. Extracellular matrix- and cyto-
skeleton-dependent changes in cell shape and stiffness. Exp. Cell Res.
278:92100.
36. Brandt, R. 2001. Cytoskeletal mechanisms of neuronal degeneration.
Cell Tissue Res. 305:255265.
37. van Vliet, K. J., G. Bao, and S. Suresh. 2003. The biomechanics
toolbox: experimental approaches for living cells and biomolecules.
Acta Biomater. 51:58815905.
38. Bao, G., and S. Suresh. 2003. Cell and molecular mechanics of biolog-
ical materials. Nat. Mater. 2:715725.
39. Benoit, M., and H. E. Gaub. 2002. Measuring cell adhesion forces with
the atomic force microscope at the molecular level. Cells Tissues
Organs (Print). 172:174189.
40. Arnoldi, M., M. Fritz, ., A. Boulbitch. 2000. Bacterial turgor pressure
can be measured by atomic force microscopy. Phys. Rev. E Stat. Phys.
Plasmas Fluids Relat. Interdiscip. Topics. 62 (1 Pt B):10341044.
41. Touhami, A., B. Nysten, and Y. F. Dufrene. 2003. Nanoscale mapping
of the elasticity of microbial cells by atomic force microcopy. Lang-
muir. 19:45394543.
42. Dupres, V., F. D. Menozzi, ., Y. F. Dufrene. 2005. Nanoscale mapping
and functional analysis of individual adhesins on living bacteria. Nat.
Methods. 2:515520.
43. Longo, G., L. M. Rio, ., S. Kasas. 2012. Force volume and stiffness
tomography investigation on the dynamics of stiff material under bac-
terial membranes. J. Mol. Recognit. 25:278284.
44. Dufrene, Y. F., D. Martnez-Martn, ., D. J. Muller. 2013. Multipara-
metric imaging of biological systems by force-distance curve-based
AFM. Nat. Methods. 10:847854.
45. Radmacher, M., M. Fritz, ., P. K. Hansma. 1996. Measuring the visco-
elastic properties of human platelets with the atomic force microscope.
Biophys. J. 70:556567.
46. A-Hassan, E., W. F. Heinz, ., J. H. Hoh. 1998. Relative microelastic
mapping of living cells by atomic force microscopy. Biophys. J. 74:
15641578.
47. Haga, H., S. Sasaki, ., T. Sambongi. 2000. Elasticity mapping of
living fibroblasts by AFM and immunofluorescence observation of
the cytoskeleton. Ultramicroscopy. 82:253258.
48. Solon, J., I. Levental, ., P. A. Janmey. 2007. Fibroblast adaptation and
stiffness matching to soft elastic substrates. Biophys. J. 93:44534461.
49. Darling, E. M. 2011. Force scanning: a rapid, high-resolution approach
for spatial mechanical property mapping. Nanotechnology. 22:175707.
50. Prabhune, M., G. Belge, ., M. Radmacher. 2012. Comparison of
mechanical properties of normal and malignant thyroid cells. Micron.
43:12671272.
51. Alcaraz, J., L. Buscemi, ., D. Navajas. 2003. Microrheology of
human lung epithelial cells measured by atomic force microscopy.
Biophys. J. 84:20712079.
52. Gavara, N., and R. S. Chadwick. 2012. Determination of the elastic
moduli of thin samples and adherent cells using conical atomic force
microscope tips. Nat. Nanotechnol. 7:733736.
53. Darling, E. M., S. Zauscher, and F. Guilak. 2006. Viscoelastic proper-
ties of zonal articular chondrocytes measured by atomic force micro-
scopy. Osteoarthritis Cartilage. 14:571579.
54. Vadillo-Rodriguez, V., T. J. Beveridge, and J. R. Dutcher. 2008. Surface
viscoelasticity of individual Gram-negative bacterial cells measured
using atomic force microscopy. J. Bacteriol. 190:42254232.
Viscoelasticity of Live Cells Investigated Using AFM
55. Bausch, A. R., F. Ziemann, ., E. Sackmann. 1998. Local measure-
ments of viscoelastic parameters of adherent cell surfaces by magnetic
bead microrheometry. Biophys. J. 75:20382049.
56. Tung, R. C., A. Jana, and A. Raman. 2008. Hydrodynamic loading of
microcantilevers oscillating near rigid walls. J. Appl. Phys. 104:
114905.
57. Mahaffy, R. E., C. K. Shih, ., J. Kas. 2000. Scanning probe-based fre-
quency-dependent microrheology of polymer gels and biological cells.
Phys. Rev. Lett. 85:880883.
58. Mahaffy, R. E., S. Park, ., C. K. Shih. 2004. Quantitative analysis of
the viscoelastic properties of thin regions of fibroblasts using atomic
force microscopy. Biophys. J. 86:17771793.
59. Xu, X., and A. Raman. 2007. Comparative dynamics of magnetically,
acoustically, and Brownian motion driven microcantilevers in liquids.
J. Appl. Phys. 102:034303.
60. Xu, X., M. Koslowski, and A. Raman. 2012. Dynamics of surface-
coupled microcantilevers in force modulation atomic force micro-
scopymagnetic vs. dither piezo excitation. J. Appl. Phys. 111:
054303.
61. Kiracofe, D., and A. Raman. 2011. Quantitative force and dissipation
measurements in liquids using piezo-excited atomic force microscopy:
a unifying theory. Nanotechnology. 22:485502.
62. Rabe, U., K. Janser, and W. Arnold. 1996. Vibrations of free and sur-
face-coupled atomic force microscope cantilevers: theory and experi-
ment. Rev. Sci. Instrum. 67:32813293.
63. de Beer, S., D. van den Ende, and F. Mugele. 2008. Atomic force mi-
croscopy cantilever dynamics in liquid in the presence of tip sample
interaction. Appl. Phys. Lett. 93:253106.
1043
64. Hutter, J. L., and J. Bechhoefer. 1993. Calibration of atomic-force
microscope tips. Rev. Sci. Instrum. 64:18681873.
65. Butt, H.-J., and M. Jaschke. 1995. Calculation of thermal noise in
atomic force microscopy. Nanotechnology. 6:17.
66. Xu, X., C. Carrasco, ., A. Raman. 2008. Unmasking imaging forces
on soft biological samples in liquids when using dynamic atomic force
microscopy: a case study on viral capsids. Biophys. J. 95:25202528.
67. Sokolov, I., S. Lyer, ., C. D. Woodworth. 2007. Detection of surface
brush on biological cells in vitro with atomic force microscopy. Appl.
Phys. Lett. 91:023902.
68. Iyer, S., R. M. Gaikwad, ., I. Sokolov. 2009. Atomic force micro-
scopy detects differences in the surface brush of normal and cancerous
cells. Nat. Nanotechnol. 4:389393.
69. Jamali, Y., M. Azimi, and M. R. K. Mofrad. 2010. A sub-cellular visco-
elastic model for cell population mechanics. PLoS ONE. 5:e12097.
70. Proksch, R., and D. G. Yablon. 2012. Loss tangent imaging: theory and
simulations of repulsive-mode tapping mode atomic force microscopy.
Appl. Phys. Lett. 100:073106.
71. Cartagena, A., M. Hernando-Perez, ., A. Raman. 2013. Mapping
in vitro local material properties of intact and disrupted virions at
high resolution using multi-harmonic atomic force microscopy. Nano-
scale. 5:47294736.
72. Dulebo, A., J. Preiner, ., P. Hinterdorfer. 2009. Second harmonic
atomic force microscopy imaging of live and fixed mammalian cells.
Ultramicroscopy. 109:10561060.
73. Kiracofe, D., J. Melcher, ., S. Hu. 2013. VEDA: virtual environment
for dynamic AFM. https://nanohub.org/resources/veda. http://dx.doi.
org/10.4231/D3KW57J1T.
Biophysical Journal 106(5) 10331043