pubs.acs.org/NanoLett
Insights from a Nanoparticle Minuet:
Two-Dimensional Membrane Profiling
through Silver Plasmon Ruler Tracking
Guoxin Rong,† Hongyun Wang,† and Björn M. Reinhard*
Department of Chemistry and The Photonics Center, Boston University, Boston, Massachusetts 02215
ABSTRACT Individual pairs of polymer-tethered silver nanoparticles, so-called silver plasmon rulers, enable distance and orientation
measurements on the nanoscale. The reduced linear dichroism and the spectrum of the light scattered from individual plasmon rulers
encode information about their orientation and average interparticle separation, respectively. We took advantage of the gain in
information silver plasmon rulers offer as probes in optical tracking and analyzed the translational and rotational motions as well as
the extension of individual silver plasmon rulers diffusing on the plasma membrane of lysed HeLa cells. Consistent with a
compartmentalization of the cell surface on the length scales of the plasmon rulers, most rulers were either immobilized or performed
a confined lateral diffusion. Structural details of a plasmon ruler’s confinement region became accessible utilizing the orientation and
interparticle separation dependent optical response of the plasmon rulers. This approach, which we refer to as polarization-resolved
plasmon coupling microscopy, enabled a detailed structural characterization of individual membrane compartments and provided a
quantitative metrics to characterize the structural lateral heterogeneity of cell membranes on submicrometer length scales. In
combination with adequate tracking methods, the “dance” performed by membrane confined dimers of flexibly linked noble metal
nanoparticles revealed deep insight into the underlying membrane morphology.
KEYWORDS Plasmon coupling microscopy, molecular rulers, nanoplasmonics, cell surface heterogeneity, active nanostructures
T
lipids in the membrane.6-9 There is evidence that the
coexistence of liquid ordered (lo) and liquid-disordered (ld)
domains provides additional spatial organization of biological membranes.10 A partitioning of the membrane in lo and
ld domains and a selective recruitment of membrane proteins into these domains could play an active role in signaling
processes by facilitating direct interactions between the
receptors in areas of high local concentration.11 The size,
lifetime, and mechanism of the formation of the lo domains
remain, however, matters of intense debate. The latter is
partly due to a current lack of adequate tools for probing the
dynamic membrane structure on nanometer length scales
and thus motivates the development of dynamic high resolution cell membrane profiling approaches.12
Important insights into the organization of cellular surfaces has been deduced from particle tracking experiments.13-17 In fact, it was only due to the fast temporal
resolution accessible in single particle tracking that it became
possible to detect the hop motion of individual lipid molecules
and membrane proteins in the plasma membrane of mammalian cells. These observations eventually led to the formulation of the anchored membrane-protein picket model.3
One limitation of conventional high-speed particle tracking is that individual nanoparticle labeled surface species can
no longer be resolved once they approach each other to
within the diffraction limit of ∼400 nm in the visible. We
have recently shown that this limitation can be overcome
by utilizing the distance dependence of the plasmon coupling between individual noble metal nanoparticles.18 If two
he plasma membrane of mammalian cells is a complex hybrid material that contains approximately
equal contributions in weight from both proteins and
lipids. It represents a two-dimensional diffusion system, in
which the spatial organization and dynamics of both lipids
and proteins plays an important role in regulating signal
transduction and membrane traffic.1 Some of the biological
functionality of the membrane is enabled by its spatial
organization and different elements have been identified to
play a role in structuring the membrane.2 Single particle
tracking experiments have shown that the plasma membrane is compartmentalized; phospholipids3 and membrane
proteins4 perform a “hop” diffusion between different compartments in the membrane. The observed compartmentalization of the membrane on submicrometer length scales
was attributed to the underlying cortical actin network that
defines fences or corrals in the membrane and that can serve
as scaffold for anchored transmembrane proteins in an
anchored membrane-protein picket model.3 Specialized
scaffolding molecules associated with the membrane have
also been identified to organize various components on the
plasma membrane involved in important cellular activities
such as exocytosis and endocytosis.5
Another potential cause for lateral heterogeneity in biological membranes arises from the self-organization of the
* To whom correspondence should be addressed. E-mail: [email protected].
†
These authors contributed equally.
Received for review: 10/7/2009
Published on Web: 12/17/2009
© 2010 American Chemical Society
230
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
noble metal nanoparticles approach each other to within ∼2
particle diameters the individual particle plasmons start to
couple, and the resonance wavelength red shifts with decreasing interparticle separation.19,20 Spectral shifts that
result from direct near-field interactions between individual
diffusing particles at constant refractive index can be detected as a ratiometric shift, provided that the particles are
simultaneously tracked on two color channels. We have used
this approach, which we called plasmon coupling microscopy (PCM), to resolve close contacts between individual
nanoparticle labeled fibronection-integrin complexes.18
Since the optical signal of individual nanoparticles in PCM
is based on light scattering, the particles do not blink or
bleach. In contrast to fluorescence-based optical superresolution methods,21-24 PCM is not limited by the photophysical instabilities of fluorescent dyes and can, in principle,
monitor nanometer length scale distances between individual nanoparticle labeled surface groups with high temporal resolution for an unlimited time.25-27
PCM is not limited to probe interactions between individual nanoparticle probes. Very recently Aaron et al. used
a plasmon coupling based imaging approach to monitor
global receptor regulation states in living cells.28 The ability
to study interactions between individual nanoparticlelabeled surface groups with PCM is, however, extremely
useful for many mechanistic studies, especially if one considers that plasmon coupling does not only affect the
spectrum of the scattered light but also its polarization.29
Unlike for individual spherical particles, which scatter light
of all polarizations with equal probability, the light scattered
from strongly coupled noble metal nanoparticles is polarized
with the polarization of the scattered light pointing into the
direction of the long dimer axis.
In this manuscript we take advantage of the distance and
orientation dependent optical properties of preassembled
dimers of flexibly linked silver nanoparticles, so-called silver
plasmon rulers,30,31 as probes in particle tracking. Our aim
is to apply these active nanostructures to map the twodimensional organization of mammalian plasma membranes on submicrometer length scales. Plasma membranes
are compartmentalized and the individual compartments of
a cell membrane represent potential traps for the silver
plasmon ruler probes. We use spherical silver nanoparticles
with an average diameter of ∼30 nm and an equilibrium
separation in solution of ∼14 nm in this work. The average
compartment size in HeLa cells has been reported to be on
the order of ∼70 nm.32 Since the average interparticle
separation and rotational freedom of the plasmon rulers in
compartments of this size is expected to depend on structural details of the confinement (see Figure 1), the probes
represent promising tools for a two-dimensional profiling of
individual membrane compartments.
II. Results and Discussion. Noble metal nanoparticles
with diameters >20 nm are efficient light scatters25,26 and
can be localized with high spatial precision in very short
© 2010 American Chemical Society
FIGURE 1. Pairs of individual polymer-tethered nanoparticles serve
as probes in polarization resolved plasmon coupling microscopy
(PRPCM) to probe the spatial organization of the plasma membrane.
Analysis of the motion of diffusing nanoparticles on a cell membrane
through conventional particle tracking provides information about
the membrane compartmentalization. For dimers that are trapped
in attractive sites on the membrane surface, the wavelength and
the polarization of the scattered light provide additional information
about the interparticle separation and the rotational freedom of the
dimers in these sites. The average interparticle separation and
mobility of the plasmon rulers is expected to decrease with increasing confinement as illustrated for examples (a) and (b).
FIGURE 2. (a) Fluorescence and (b) darkfield images of a cell
membrane after acoustic rupturing and staining with fluorescently
tagged phalloidin. The phalloidin binds specifically to F-actin
subunits and stains the actin network supporting the membrane in
the fluorescent image.
integration times,27,33,34 provided that the scattering background is low. This requirement is not necessarily fulfilled
for individual particles bound to living cells. Different cellular
components, in particular the nucleus, create a high scattering background that makes high-speed tracking (in this
study with frame rates up to 500 Hz) of small nanoparticle
probes challenging. To maximize the signal-to-noise in single
plasmon ruler tracking, all experiments in this work were
therefore performed on membranes of lysed cells. Removal
of the cell nucleus led to a significant reduction of the
scattering background.
We prepared plasma membranes of cervical cancer
(HeLa) cells immobilized on a glass support through acoustic
rupturing35 of cells grown on No1 coverglass slides. In Figure
2 we show fluorescence and darkfield scattering images of
a resulting membrane after treatment with fluorescently
tagged phalloidin. Phalloidin binds selectively at the interface
of F-actin subunits and is commonly used as stain for actin
231
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
FIGURE 3. Experimental setup for polarization resolved plasmon coupling microscopy (PRPCM). (a) In a microscope with darkfield configuration
the samples are illuminated with unpolarized light with alternating excitation wavelengths. The light scattered from individual nanoparticle
dimers is collected with a 100× objective and then split into two othogonal polarization channels that are reimaged on two translated regions
of a electron multiplying charge coupled device (EMCCD). The intensities I1 and I2 on the two polarization channels in each frame n are used
to calculate the reduced linear dichroism P and the total intensities of two subsequent frames are used to calculate the intensity ratio R. (b)
Image of silver plasmon rulers bound to a HeLa membrane on two orthogonal polarization channels. (c) Trajectory of an individual plasmon
ruler marked in (b). The figure shows the scattering image at t ) 0 and the fitted position of the maximum as function of time as blue trace.
filaments.36 The cortical actin, which plays a prominent role
for the spatial organization of the membrane, is clearly
visible in Figure 2a as dense network of branched filaments
below the inner leaflet of the plasma membrane.
Our cell membrane in vitro model has two important
advantages over conventional membrane models. (i) It
includes the underlying actin network that supports and
structures the membrane and (ii) the membranes have the
same composition as intact cellular membranes. However,
like for any in vitro system, our model system does not have
the full functionality of the living cell. Despite this limitation,
the lysed cells remain a useful model for the development
of the proposed cell surface profiling approach and provide
valuable insight into the compartmentalization of cell membranes on submicrometer length scales.
The signal-to-noise in particle tracking depends on the
brightness and contrast of the individual particles. Ideal
probes for PCM have large scattering cross sections in
spectral ranges with low background from the cellular
membrane, sharp resonances, and steep resonance wavelength (λres) versus interparticle separation (S) relationships.
We decided to use silver and not gold plasmon rulers in our
studies since silver nanoparticles of ∼30 nm diameter have
their resonances in the blue region of the electromagnetic
spectrum where the scattering background from cellular
membranes is lower (see Supporting Information, Figure S1)
than in the green where the resonance of gold nanoparticles
with comparable size lie. Silver plasmon rulers also have a
larger dynamic range in the distance dependent spectral
response than gold plasmon rulers and are therefore more
sensitive for detecting small distance changes.30
Silver nanoparticles are in general less stable in salt
buffers than gold nanoparticles and undergo oxidative
© 2010 American Chemical Society
corrosion.37,38 Chemical self-assembly procedures for silver
nanoparticles require therefore sufficient stabilization of the
particles through appropriate surface ligands. We used a
monolayer of thiolated alkyl-[polyethylene glycol(PEG)]acetate (HS-(CH2)5-(OCH2CH2)6-OCH2-COOH) to render
the silver nanoparticles stable in Hank’s buffered solution
(137.93 mM NaCl, 5.33 mM KCl, 4.13 mM NaHCO3, 0.441
mM KH2PO4, 0.338 mM Na2HPO4, 5.56 mM D-glucose). The
stabilized silver particles were then assembled into dimers
using a DNA-programmed self-assembly strategy. Details
regarding the assembly and characterization of silver plasmon rulers containing nanoparticles with an average diameter of d ) 30 ( 4 nm (as obtained by transmission electron
microscopy (TEM)) are summarized in the Methods of the
Supporting Information. The resulting preparations contained typically ∼75% plasmon rulers.
The PEG stabilized silver plasmon rulers attached efficiently and nonreversibly to freshly prepared plasma membranes and could be tracked with frame rates of up to 500
Hz with signal-to-noise ratios >3. All experiments were
performed in home-built flowchambers at 25 °C.
Polarization Resolved and Ratiometric Tracking of
Silver Plasmon Rulers. Our aim was to implement a tracking scheme that yields information not only about the
plasmon rulers’ spatial coordinates but also about the polarization as well as the resonance wavelength of the light
scattered from individual silver plasmon rulers. Our experimental setup is based on a conventional inverted darkfield
microscope (Olympus IX71) and is illustrated in Figure 3a.
Unpolarized whitelight from a 100 W tungsten lamp passes
a filter wheel (Lambda 10-3, Sutter Instrument), which can
toggle between two different monochromatic filters. The
light is then injected into the specimen plane at oblique
232
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
angles using an oil darkfield condenser (numerical aperture
(NA) ) 1.2). This illumination geometry ensures that only
light that is scattered in the specimen plane is collected by
the 100× oil immersion objective (NA ) 0.6). The collected
light passes another 1.6× magnification lens and is then split
into two images with orthogonal light polarizations using a
polarizing beam splitter and finally reimaged on two translated areas on the same electron multiplying charge coupled
device (EMCCD, Andor iXon+). The effective pixel size in this
imaging detector is 160 nm.
Figure 3b shows an image of silver dimers on a cell
membrane simultaneously recorded on two orthogonal
polarization channels. Individual dimers were simultaneously tracked on both channels by fitting39 the dimer
image on two polarization channels I1 and I2. Each channel
was background corrected by the scattering intensity of the
membrane in that channel, and we used the maxima of the
curve-fitted39 images, or point-spread-functions, of the individual dimers to determine their coordinates. Under typical experimental conditions we achieve a localization precision of σ ) 60 nm.
The trajectories of the individual dimers were obtained
by linking the coordinates of the maxima over time using a
custom written MATLAB program that was based on the
nearest-neighbor-tracking algorithm developed by Wieser
and Schuetz.40 The intensities, I1(n) and I2(n) of the plasmon
rulers were then obtained by integrating over the pointspread-functions on the two polarization channels in each
frame n.
The blow-up in Figure 3c displays the trajectory of a single
plasmon ruler diffusing on the cell membrane recorded on
one of the polarization channels. A second trajectory of
the same plasmon ruler was simultaneously recorded on the
second polarization channel. The relative intensities of the
plasmon rulers on the two channels encode information
about the orientation of the long dimers axis as a function
of space and time. The integrated intensities I1(n) and I2(n)
of a dimer on the two polarization channels are used to
calculate the reduced linear dichroism41 P(n) ) (I1(n) - I2(n))/
(I1(n) + I2(n)) for a dimer in each frame n. In anisotropic
noble metal nanostructures, such as dimers29 and nanorods42 P depends on the orientation of the long dimer axis
and can therefore be used to monitor the rotation motion
of those nanostructures on the membrane surface.
The possibility of alternating the excitation wavelength
between subsequent frames through toggling between different bandpass filters allows a ratiometric tracking of
individual plasmon rulers. Ratiometric spectral information
are obtained by computing the intensity ratios R(n′) ) Itot(n)/
Itot(n - 1) from the total intensities Itot(n) ) I1(n) + I2(n) of
individual plasmon rulers in two subsequent frames n - 1
and n for the entire movie. The new index n′ is defined as n′
) n/2; n was chosen to refer to the frame recorded with the
longer wavelength. This approach generates R values for all
plasmon rulers tracked on the cell surface as a function of
© 2010 American Chemical Society
time. The sensitivity of the ratiometric tracking approach for
detecting distance changes in plasmon rulers depends on
the interparticle separation and the choice of the filter
combination. We used combinations of bandpass filters that
covered the spectral range between 430 nm (corresponding
to ∼λres of an individual particle) and 530 nm (corresponding
to ∼λres of a strongly coupled dimer).
The implemented ratiometric tracking scheme using
alternating excitation wavelengths is intended to augment
a simultaneous polarization resolved tracking of silver plasmon rulers and is referred to as polarization resolved plasmon coupling microscopy (PRPCM). Polarization-resolved
tracking was performed with temporal resolutions of up to
500 Hz. This temporal resolution was accessible for measurements of P only without acquiring additional spectral
information. The maximum temporal resolution accessible
in PRPCM for measurements of R was significantly lower
and limited by the mechanics of the filterwheel to ∼5 Hz.
This is, however, not an intrinsic limitation of the technology;
replacement of the filterwheel, for instance, with two shuttered monochromatic lasers would allow significantly higher
frame rates. In this work we limited ourselves, however, to
the temporal resolution available with the instrumentation
readily available in our laboratory. We emphasize that the
current implementation of PRPCM is not well suited to
monitor rapid fluctuations in the interparticle separation.
Instead, it provides reliable estimates of the time-averaged
interparticle separation in individual diffusing plasmon rulers.
Diffusion Modes of Silver Plasmon Rulers on HeLa
Cell Membranes. In “normal” Brownian lateral diffusion the
mean square displacement (MSD) of an object grows linearly
as a function of time (t), MSD ∝ t. A cell membrane is,
however, heterogeneous and contains traps for the nanoparticles with varying potential depths spatially and temporally distributed over its entire surface. These traps influence
the dimer motion and can result in an anomalous diffusion
in which the diffusion coefficient becomes a function of time.
In this case the mean square displacement (MSD) grows
nonlinearly and is described by a power law in t.43,44
To characterize the translational mobility of silver nanoparticle dimers on different time scales we apply here the
moment scaling spectrum introduced by Ferrari et al.45 This
analysis is based on the calculated moments of displacements,45,46 where a moment µ of order ν for a specific frame
shift ∆n corresponding to a time shift δt ) ∆n∆t (∆t is the
fixed time interval between two frames), is defined as
µν,l(∆n) )
1
Ml - ∆n
Ml-∆n-1
∑
|x
bl(n + ∆n) - b
x l(n)| ν
n)0
where b
xl(n) is the position vector (xl(n),yl(n)) on trajectory l
at t ) n∆t. Assuming a power law of the form µν(δt) ∝ δtγv,45
the scaling coefficients γν are determined through linear fits
to double-logarithmic plots of µν versus δt and the twodimensional diffusion coefficients (Dν) of order ν > 0 are
obtained from the ordinate intercepts It as Dv ) (2v)-1exp (It).46
233
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
silver plasmon rulers on the cell membrane shows that most
of the probes are either immobilized or perform a confined
diffusion.
The performed polarization resolved tracking studies
enabled us to evaluate the rotational mobility of the plasmon
rulers as a function of D2, as well. To that end we fitted a
single exponential to the autocorrelation of P and obtained
a characteristic rotational correlation time τ from the fit (see
Supporting Information, Figure S2). The resulting τ values
are plotted against D2 in Figure 4b. While for D2 < 2 × 10-3
µm2/s most of the measured diffusion times are below 2.5
× 10-2 s, for D2 > 2 × 10-3 µm2/s we find a broader
distribution of τ values with τ > 2.5 × 10-2 s. The essential
absence of autocorrelation in P for D2 < 2 × 10-3 µm2/s
suggests that these plasmon rulers are immobilized and that
the long dimer axis is fixed in space. The higher τ values for
plasmon rulers with D2 > 2 × 10-3 µm2/s, on the other hand,
indicate two-dimensional rotational motions that are characterized by nonzero rotational correlation times.
The broad spread of the τ values for D2 > 2 × 10-3 µm2/s
shows that the rotational dynamics differs substantially
between individual plasmon rulers. Our analysis of the
plasmon rulers’ lateral diffusion revealed that SMSS increases
continuously with growing D2 for D2 > 2 × 10-3 µm2/s
indicating a steady increase in the lateral mobility with
growing linear diffusion coefficient. There is no comparable
strong correlation between τ and D2 for the rotational
dynamics. Instead, the τ values are broadly distributed for
D2 > 2 × 10-3 µm2/s. In principle, this distribution could arise
from impurities of our plasmon ruler preparations, which
always contain some monomers. However, since the light
scattered from an individual spherical particle is not polarized, a contamination with monomers cannot account for
the differences in the rotational dynamics in the range τ )
0.025-0.38 s for D2 > 2 × 10-3 µm2/s. Instead, it is more
probable that the distribution of the τ values arises from the
heterogeneity of the cell membrane due to local differences
in the chemical or structural nanoenvironment of the individual plasmon rulers. Both of these factors would directly
affect the rulers’ rotational dynamics.
Monitoring Orientation and Interparticle Separation
of Two-dimensional Confined Silver Plasmon Rulers:
Watching a Nanoparticle Minuet in Real Time. Deeper
insight into the lateral heterogeneity of the plasma membrane is potentially available through PRPCM. For those
plasmon rulers that were tracked by polarization resolved
ratiometric tracking (red filled circles in Figure 4a), the
measured P and R values offer additional information about
the interparticle separation and the orientation of the dimer
axis on the membrane. The R values can be converted into
approximate interparticle separations S, provided that an
S(R) calibration relationship is available. We derived such
relationships for different filter combinations from scattering
spectra of silver dimers with known interparticle separation
S (see Supporting Information, Figure S3). Our experimental
FIGURE 4. (a) Slope of the moment scaling spectrum (SMSS) versus
linear diffusion coefficient (D2). Black filled circles were obtained
from tracks of individual plasmon rulers recorded in a polarization
resolved fashion with high temporal resolution (500 Hz). The red
filled circles belong to trajectories obtained through polarization
resolved ratiometric tracking with a temporal resolution of 5 Hz.
Plasmon rulers PR1-3 are analyzed in more detail in the text. (b)
Rotational correlation time (τ) versus D2 for the data set recorded
with 500 Hz.
The linear diffusion coefficient is a special case in this
treatment and obtained for ν ) 2.
The γν versus ν relationship is the moment scaling
spectrum and its slope (SMSS) characterizes the mobility of
the silver nanoparticle dimers on the membrane. SMSS values
of 0, 0.5, and 1 correspond to immobilized, free, and
directed (ballistic) diffusion, respectively.46 SMSS values between 0 and 0.5 indicate confined motions, whereas slopes
in the range 0.5-1 indicate motions that are in the superdiffusive regime.46
The moment scaling approach provides a quantitative
measure that enables a systematic comparison of a large
number of single dimer trajectories with each other.47 In
Figure 4a we plot the fitted SMSS values against the respective
linear diffusion coefficients D2 for a total of 70 trajectories.
The plot summarizes the results obtained for polarizationresolved tracking (temporal resolution 500 Hz, black filled
circles) as well as for polarization resolved ratiometric tracking (temporal resolution 5 Hz, red filled circles).
The silver plasmon rulers on the cellular membrane show
a range of different lateral diffusion modes. For D2 < 2 ×
10-3 µm2/s the measured SMSS values are low and independent of the diffusion coefficient but for D2 > 2 × 10-3 µm2/s,
SMSS grows with increasing D2. Silver plasmon rulers with
diffusion coefficients in the range D2 ) 1 × 10-5 - 2 × 10-3
µm2/s have very low SMSS values (SMSS < 0.05) and are
considered to be immobilized. Plasmon rulers with D2 ) 2
× 10-3 - 7 × 10-2 µm2/s and SMSS ) 0.05-0.40 perform
confined diffusion and plasmon rulers with D2 ) 7 × 10-2
- 2 × 10-1 µm2/s and SMSS ) 0.40-0.52 perform quasi-free
diffusion. Overall, our analysis of the lateral mobility of the
© 2010 American Chemical Society
234
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
approach of correlating single dimer Rayleigh scattering
spectra of individual silver plasmon rulers with their interparticle separation as obtained by TEM followed essentially
the experimental procedures described in ref.48 In this work
we only apply the S(R) calibrations shown in Supporting
Information, Figure S3, a detailed analysis of the distance
dependent plasmon coupling will be published elsewhere.49
The S(R) calibration was performed with silver nanoparticles that were immobilized on Formvar (nr ≈ 1.45) and
immersed in a refractive index of nr ) 1.47, whereas the
membranes (nr ≈ 1.40) in the tracking experiments are
sandwiched between glass (nr ≈ 1.50) and buffer solution
(nr ≈ 1.34). The difference in the refractive index between
these two experimental conditions will affect the S(R) calibration and lead to a systematic error in the derived distances.50 Another source of error for the conversion of R into
S values are particle-to-particle variations in particle size and
shape, which also directly influence the resonance wavelength.51 The above potential error sources currently limit
the accuracy of PRPCM. Nevertheless, even the ability to
estimate interparticle separations and to correlate these
approximates with the silver plasmon ruler orientation
provides valuable insight into the organization of the membrane as a whole as we demonstrate in the following.
Figure 5 illustrates the information content accessible in
a typical PRPCM experiment. The figure contains the diffusion tracks and the corresponding P, R, and S values as a
function of time for the plasmon rulers labeled as PR1-3 in
Figure 4a. We chose PR1-3 for the following analyses
because they cover a broad range of translational diffusion
modes observed in our experiments. PR1 was recorded with
a bandpass (BP) filter combination comprising 430BP10
(center wavelength, 430 nm; spectral width, 10 nm) and
470BP10, PR2 and 3 were recorded using a combination of
450BP10 and 490BP10. The plotted P values refer to the
reduced linear dichroism obtained in the long wavelength
channel.
PR1 (Figure 5a) is confined but exhibits the highest degree
of translational mobility among PR1-3. The approximate
average interparticle separation for PR1 is obtained as the
average of all individual S measurements in the trajectory
as Sav ) 13.4 ( 2.8 nm, which is in agreement with the
expected end-to-end distance of the DNA tether (S0 ) 13.7
nm, see Methods in the Supporting Information). On shorter
time scales we observe larger systematic changes in the
5-point sliding average of S resulting from variations in the
interparticle separation during the plasmon ruler’s diffusion
in the confinement region.
The recorded P values for PR1 show a large relative
spread and fluctuate seemingly randomly in the range
between -0.4 and +0.5. PR1-3 were tracked with temporal
resolutions of 5 Hz, but even at this relatively low temporal
resolution the orientation dependence of the polarization is
not “averaged out”. The observation of a large dynamic
range in P indicates a rotational motion in which longer static
© 2010 American Chemical Society
phases with fixed orientation of the long dimer axis alternate
with rotational phases in which the plasmon rulers rapidly
find a new orientation on the surface. We refer to this
rotational motion as “hop” rotation in the following.
PR2 in Figure 5b explores a smaller area of the cell
membrane than PR1. The stronger spatial confinement in
case of PR2 coincides with a decrease in the average
interpartice separation to Sav ) 11.7 ( 2.3 nm. The fluctuations in P observed in the first 80 s of the trajectory of PR2
in Figure 5b are again characteristic of a rotational motion
in which the long dimer axis hops between discrete angles
on the cell surface. This behavior changes during the interval
t ≈ 80-92 s in which the P values remain close to zero. Since
plasmon rulers have nonzero scattering cross sections along
both the long and the perpendicular short dimer axis, the
observation of P ≈ 0 is attributed to a rotation of the long
dimer axis that is fast on the time scale of the measurement.
In this case all plasmon ruler orientations contribute with
equal probability to the measurement resulting in no net
polarization. The onset of the rapid rotational motion at t )
80 s is accompanied by a systematic increase in S by
approximately 2 nm and a decrease in the point-to-point
fluctuations of S. These observations indicate an increase in
the translational and rotational mobility of the plasmon
rulers which could result from heterogeneities within the
confinement region or changes of its structure as a function
of time.
The diffusion track of PR3 in Figure 5c is spatially more
confined than those of PR1 and 2 and the average interparticle separation obtained from the measured R values is
further decreased to Sav ) 9.7 ( 1.3 nm. PR3 still exhibits
some rotational dynamics, most prominently in the interval
t ) 58-115 s, during which plasmon ruler rotation is
correlated with a change in the interparticle separation.
Additional more transient orientation changes can be identified at other times throughout the trajectory. All of the
recorded P values lie, however, in the range of P ≈ 0-0.6.
The lower magnitude and frequency of the P fluctuations for
PR3 when compared with PR1 and 2 indicates a reduction
of the rotational freedom for PR3.
Overall, we conclude that for PR1-3 in Figure 5 an
increasing spatial confinement is correlated with a decreasing rotational mobility and decreasing average interparticle
separation.
Next, we want to analyze the confinement of PR3 in more
detail. We already know from Figure 5c that PR3 explores a
total area on the order of 300 × 300 nm2 during our
observation time and that the orientation of the long plasmon ruler axis is constrained to P ≈ 0-0.6 during the entire
observation time. Since S and P were obtained from tracking
experiments, they can be analyzed as a function of time as
shown in Figure 5 and as a function of their location on the
cell surface. In Figure 6, we performed such a spatial analysis
and created S and P maps for two different definition ranges.
Whereas Figure 6a1 includes all S values of the trajectory,
235
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
FIGURE 5. Polarization resolved ratiometric tracking of three different silver plasmon rulers. (a) PR1, (b) PR2, and (c) PR3. The panel on the
left shows the track of the plasmon ruler on the cell membrane. The right panels contain the intensity ratio of the monitored wavelength
channels (R, red), the derived approximate interparticle separation (S, blue), and the reduced linear dichroism (P, black) as function of time
during the plasmon ruler’s diffusion on the cell membrane. For S, we included a 5-point sliding average as solid blue line.
Figure 6a2 contains only the first S quartile, corresponding
to the shortest interparticle separations. Similarly, Figure 6b1
includes all P values, whereas Figure 6b2 contains only the
fourth P quartile.
An inspection of the S and P maps in Figure 6a1,b1 shows
that the data points are not equally distributed across the
© 2010 American Chemical Society
entire confinement region but clustered in three subregions,
which are labeled as C1-C3 in Figure 6a. C1-C3 are not
strictly associated with distinct times in the trajectory but
get populated throughout the observation, and we conclude
that PR3 shuttles between these subregions. Despite this
translational movement, the reduced linear dichroism re236
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
FIGURE 6. (a) S and (b) P maps for one plasmon ruler (PR3) that is performing a confined diffusion on a HeLa membrane. The maps were
created from a PRPCM trajectory of 120 s length. The S map in (a1) includes all data points whereas (a2) only shows the shortest interparticle
separation (first quartile). The P map in (b1) includes all data, the P map in (b2) contains only the fourth quartile (P ) 0.3-0.6). The P and
S distributions show that PR3 preferentially resides in three subregions marked as C1-C3.
mains between P ≈ 0-0.6 due to a constrained rotational
mobility in the entire confinement region. We point out that,
although the drawn circumferences around C1-C3 in Figure
6 are somewhat arbitrary, our tracking experiments indicate
that each of the subregions C1-C3 has dimensions on the
order of ∼70 nm, which is consistent with previous findings
of the compartmentalization of HeLa cell surfaces.32
Interestingly, the less populated region of the green
shaded confinement region between x ) -150 nm and x )
-50 nm is enriched in data points with longer interparticle
separations suggesting less spatial constraints. It is unclear
why this region is much less populated than C1-C3. One
possible explanation could be a chemical heterogeneity of
the membrane that favors a localization of PR3 in C1-C3.
Correlation of the S and P maps in Figure 6 offers further
insight into the subregions C1-C3. Close inspection of
Figure 6a2,b2 reveals that nearly two-thirds of the data
points in C1 in Figure 6a2 superimpose with the data points
in the same area in Figure 6b2 (for a magnification of C1,
please see Supporting Information, Figure S4). In C1 the
shortest interparticle separations are correlated with the highest P values indicative of a preferential alignment of the long
plasmon ruler axis in the compressed plasmon ruler. One
structural model that could account for the more stringent
orientation of the compressed plasmon ruler configuration
is an anisotropic shape of C1. Assuming that the extension
of C1 is shorter along one specific direction so that it can
only accommodate the long plasmon ruler axis in this
orientation if the ruler adjusts (i.e., decreases) its interparticle
separation, the geometry of C1 would induce the observed
preferential orientation of PR3 with short S values. Independent of the exact nature of the preferential orientation, the
© 2010 American Chemical Society
absence of a comparable correlation in C2 and C3 underlines
the heterogeneity of the membrane on deeply subdiffraction
limit length scales.
Overall, our analyses of Figures 5 and 6 have demonstrated that PRPCM of silver plasmon rulers is highly complementary to conventional single particle tracking in cell
membrane profiling. PRPCM provides additional information about the orientation and interparticle separation of the
plasmon rulers and the resulting S and P maps contribute
to a systematic characterization of the cell membrane
morphology on submicrometer length scales. Our studies
have also indicated a rich dynamics in S and P for cell surface
confined silver plasmon rulers which motivates further
studies with improved temporal resolution in the future.
III. Conclusions. Using a polarization resolved tracking
approach we monitored the translational and rotational
mobility of individual silver plasmon rulers on lysed HeLa
cells with a temporal resolution of 500 Hz. A moment scaling
spectrum analysis of the translational motion of the individual plasmon rulers revealed that most of the plasmon
rulers are either immobilized or undergo a confined diffusion
on the cell surface. Those plasmon rulers that undergo
confined diffusion show large differences in the rotational
dynamics which was attributed to the lateral heterogeneity
of the membrane. To obtain additional information about
the compartmentalization of the membrane, we implemented a polarization resolved plasmon coupling microscopy (PRPCM) that enabled us to track individual silver
plasmon rulers diffusing on the surface of lysed HeLa cells
and simultaneously monitor the plasmon rulers’ orientation
and interparticle separation. PRPCM provides a quantitative
237
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238
metrics (D2, τ, P, R, S) for a detailed characterization of the
mobility, orientation and extension of plasmon rulers at
different locations on the cell surface as a function of time.
Consistent with a compartmentalization of the membrane
on length scales of the plasmon rulers (dimers of 30 nm
particles with center to center separation of ∼44 nm) we find
that confinement of the plasmon rulers on the cell surface
affects their rotational freedom and average interparticle
separation. We demonstrated that by mapping S and P on
the cell surface, detailed insight into the cell membrane
organization with nanometer scale spatial resolution becomes accessible. The gain in information in polarization
resolved ratiometric tracking when compared with conventional particle tracking makes PRPCM a promising tool for
analyzing the lateral heterogeneity of complex cellular
surfaces.
(19) Su, K. H.; Wei, Q. H.; Zhang, X.; Mock, J. J.; Smith, D. R.; Schultz,
S. Nano Lett. 2003, 3, 1087–1090.
(20) Reinhard, B. M.; Siu, M.; Agarwal, H.; Alivisatos, A. P.; Liphardt,
J. Nano Lett. 2005, 5, 2246–2252.
(21) Hell, S. W.; Wichmann, J. Opt. Lett. 1994, 19, 780–782.
(22) Betzig, E.; Patterson, G. H.; Sougrat, R.; Lindwasser, O. W.;
Olenych, S.; Bonifacino, J. S.; Davidson, M. W.; LippincottSchwartz, J.; Hess, H. F. Science 2006, 313, 1642–1645.
(23) Rust, M. J.; Bates, M.; Zhuang, X. W. Nat. Methods 2006, 3, 793–
795.
(24) Hell, S. W. Science 2007, 316, 1153–1158.
(25) Yguerabide, J.; Yguerabide, E. E. Anal. Biochem. 1998, 262, 137–
156.
(26) Yguerabide, J.; Yguerabide, E. E. Anal. Biochem. 1998, 262, 157–
176.
(27) Nan, X. L.; Sims, P. A.; Xie, X. S. ChemPhysChem 2008, 9, 707–
712.
(28) Aaron, J.; Travis, K.; Harrison, N.; Sokolov, K. Nano Lett. 2009, 9,
3612-3618.
(29) Wang, H.; Reinhard, B. M. J. Phys. Chem. C 2009, 113, 11215–
11222.
(30) Sonnichsen, C.; Reinhard, B. M.; Liphardt, J.; Alivisatos, A. P. Nat.
Biotechnol. 2005, 23, 741–745.
(31) Gunnarsson, L.; Rindzevicius, T.; Prikulis, J.; Kasemo, B.; Kall, M.;
Zou, S. L.; Schatz, G. C. J. Phys. Chem. B 2005, 109, 1079–1087.
(32) Murase, K.; Fujiwara, T.; Umemura, Y.; Suzuki, K.; Iino, K.;
Yamashita, H.; Saito, M.; Murakoshi, H.; Ritchie, K.; Kusumi, A.
Biophys. J. 2004, 86, 4075–4093.
(33) Gelles, J.; Schnapp, B. J.; Sheetz, M. P. Nature 1988, 33, 450–453.
(34) Thompson, R. E.; Larson, D. R.; Webb, W. W. Biophys. J. 2002,
82, 2775–2783.
(35) Brown, R. B.; Audet, J. J. R. Soc. Interface 2008, 5, S131–S138.
(36) Wulf, E.; Deboden, E.; Bautz, F. A.; Faulstich, H.; Wieland, T. H.
Proc. Nat. Acad. Sci. U.S.A. 1979, 76, 4498–4502.
(37) Doty, R. C.; Tshikhudo, T. R.; Brust, M.; Fernig, D. G. Chem. Mater.
2005, 17, 4630–4635.
(38) Lee, J.-S.; Lytton-Jean, A. K. R.; Hurst, S. J.; Mirkin, C. A. Nano
Lett. 2007, 7, 2112–2115.
(39) D’Errico, J. Surface Fitting Using Gridfit. Matlab Central; The
Mathworks, Inc.: Natick, MA; http://www.mathworks.com/
matlabcentral/fileexchange/8998-surface-fitting-gridfit. Accessed
December 1, 2009.
(40) Wieser, S.; Schuetz, G. J. Methods 2008, 46, 131–140.
(41) Wei, C. Y.; Lu, C. Y.; Kim, Y.; Vanden Bout, D. J. Fluoresc. 2007,
17, 797–804.
(42) Pierrat, S.; Hartinger, E.; Faiss, S.; Janshoff, A.; Soennichsen, C.
J. Phys. Chem. C 2009, 113, 11179–11183.
(43) Bouchaud, J. P.; Georges, A. Phys. Rep. 1990, 195, 127–293.
(44) Feder, T. J.; Brust-Mascher, I.; Slattery, J. P.; Baird, B.; Webb,
W. W. Biophys. J. 1996, 70, 2767–2773.
(45) Ferrari, R.; Manfroi, A. J.; Young, W. R. Phys. D 2001, 154, 111–
137.
(46) Sbalzarini, I. F.; Koumoutsakos, P. J. Struct. Biol. 2005, 151, 182–
195.
(47) Ewers, H.; Smith, A. E.; Sbalzarini, I. F.; Lilie, H.; Koumoutsakos,
P.; Helenius, A. Proc. Nat. Acad. Sci. U.S.A. 2005, 102, 15110–
15115.
(48) Yang, L.; Yan, B.; Reinhard, B. M. J. Phys. Chem. C 2008, 112,
15989–15996.
(49) Yang, L.: Wang, H.; Reinhard, B. M. To be submitted for
publication.
(50) Jain, P. K.; el-Sayed, M. A. Nano Lett. 2008, 8, 4347–4352.
(51) Kelly, K. L.; Coronado, E.; Zhao, L. L.; Schatz, G. C. J. Phys. Chem.
B 2003, 107, 668–677.
Acknowledgment. This work was financially supported
by the National Institute of Health through Grants 5 R21
EB008822-02 and 1 R01 CA138509-01.
Supporting Information Available. Figures S1-S4 and
Methods. This material is available free of charge via the
Internet at http://pubs.acs.org.
REFERENCES AND NOTES
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
Whitty, A. Nat. Chem. Biol. 2008, 4, 435–439.
Edidin, M. Trends Cell Biol. 2001, 11, 492–496.
Fujiwara, T.; Ritchie, K.; Murakoshi, H.; Jacobson, K.; Kusumi, A.
J. Cell Biol. 2002, 157, 1071–1081.
Suzuki, K.; Ritchie, K.; Kajikawa, E.; Fujiwara, T.; Kusumi, A.
Biophys. J. 2005, 88, 3659–3680.
Gundelfinger, E. D.; Kessel, M. M.; Qualmann, B. Nat. Rev. Mol.
Cell Biol. 2003, 4, 127–139.
Simons, K.; Vaz, W. L. C. Ann. Rev. Biophys. Biomol. Struct. 2004,
33, 269–295.
London, E. Biochim. Biophys. Acta 2005, 1746, 203–220.
McIntosh, T. J.; Soimons, S. A. Ann. Rev. Biophys. Biomol. Struct.
2006, 35, 177–198.
Pike, L. Biochem. J. 2004, 378, 281–292.
Hancock, J. F. Nat. Rev. Mol. Cell Biol. 2006, 7, 456–462.
Hammond, A. T.; Heberle, F. A.; Baumgart, T.; Holowka, D.; Baird,
B.; Feigenson, G. W. Proc. Nat. Acad. Sci. U.S.A. 2005, 102, 6320–
6325.
Shaw, A. S. Nat. Immunol. 2006, 11, 1139–1142.
Kwik, J.; Boyle, S.; Fooksman, D.; Margolis, L.; Sheetz, M. P.;
Edidin, M. Proc. Nat. Acad. Sci. U.S.A. 2003, 100, 13964–13969.
Sheetz, M. P.; Turney, S.; Qian, H.; Elson, E. L. Nature 1989, 340,
284–288.
Qian, H.; Sheetz, M. P.; Elson, E. L. Biophys. J. 1991, 60, 910–
921.
Kusumi, A.; Nakada, C.; Ritchie, K.; Murase, K.; Suzuki, K.;
Murakoshi, H.; Kasai, R. S.; Kondo, J.; Fujiwara, T. Annu. Rev.
Biophys. Biomol. Struct. 2005, 34, 351–378.
Edidin, M.; Kuo, S. C.; Sheetz, M. P. Science 1991, 254, 1379–
1382.
Rong, G.; Wang, H.; Skewis, L. R.; Reinhard, B. M. Nano Lett.
2008, 8, 3386–3393.
© 2010 American Chemical Society
238
DOI: 10.1021/nl903350f | Nano Lett. 2010, 10, 230-238