Content uploaded by Jerome Wenger
Author content
All content in this area was uploaded by Jerome Wenger on Nov 20, 2017
Content may be subject to copyright.
Content uploaded by Jerome Wenger
Author content
All content in this area was uploaded by Jerome Wenger on Nov 20, 2017
Content may be subject to copyright.
1
Planar Optical Nano-Antennas Resolve
Cholesterol-Dependent Nanoscale Heterogeneities in
the Plasma Membrane of Living Cells
Raju Regmi,1,2
Pamina
M. Winkler,1 Valentin
Flauraud,
3
Kyra
J.
E.
Borgman,
1
Carlo Manzo,1
Jürgen Brugger,
3
Hervé
Rigneault,
2
Jérôme
W
enger,
∗
,
2
and
María
F.
García-P
ar
ajo
∗
,
1
,
4
1 ICFO-Institut de Ciencies Fotoniques, The Barcelona Institute of Science and Technology,
08860
Barcelona, Spain,
2 Aix Marseille Univ, CNRS, Centrale Marseille, Institut Fresnel, UMR
7249,
Marseille, France,
3 Microsystems Laboratory, Institute of Microengineering, Ecole
Polytechnique
Fédérale de
Lausanne, 1015 Lausanne, Switzerland,
4 ICREA, Pg. Lluís Companys 23,
08010
Barcelona,
Spain
E-mail:
jerome.wenger@fresnel.fr;
maria.garcia-
par
ajo@icf
o
.eu
Abstract
Optical
nano-antennas
can
efficiently confine
light into
nanoscopic
hotspots enabling single-
molecule detection sensitivity at biological relevant conditions. This innovative approach to
breach the diffraction limit offers a versatile platform to investigate the dynamics of individual
biomolecules in living cell membranes and their partitioning into cholesterol-dependent lipid
nanodomains. Here, we present optical
nano-antenna
arrays with accessible surface hotspots to
study the characteristic diffusion dynamics of phosphoethanolamine (PE) and sphingomyelin
(SM) in the plasma membrane of living cells at the nanoscale. Fluorescence burst analysis
2
and fluorescence correlation spectroscopy performed on nano-antennas of different gap sizes
show that unlike PE, SM is transiently trapped in cholesterol-enriched nanodomains of 10 nm
diameter with short characteristic times around 100
µ
s. Removal of cholesterol led to free dif-
fusion of SM, consistent with the dispersion of nanodomains. Our results are consistent with
the existence of highly transient and fluctuating nanoscale assemblies enriched by cholesterol
and sphingolipids in living cell membranes, also known as lipid rafts. Quantitative data on
sphingolipids partitioning into lipid rafts is crucial to understand the spatiotemporal heteroge-
neous organization of transient molecular complexes on the membrane of living cells at the
nanoscale. The proposed technique is fully bio-compatible and thus provides various oppor-
tunities for biophysics and live cell research to reveal details that remain hidden in confocal
diffraction-limited measurements.
Keywords: optical nano-antenna, nanophotonics, fluorescence correlation spectroscopy (FCS),
live cell membrane, lipid rafts
The plasma membrane plays a major role in cell physiology and is thus of fundamental im-
portance to living systems. The spatial organization and diffusion dynamics of its constituents
(lipids and proteins) occurring at the nanoscale largely influence cellular processes such as trans-
membrane signaling, intracellular
trafficking
and cell
adhesion.
1,2 Recent advances in cell biology
have shown that the plasma membrane is
significantly
more complex than just a continuous fluidic
system.
3–5 It has been postulated that sphingolipids, cholesterol and certain types of proteins can
be enriched into dynamic nanoscale assemblies or nanodomains, also termed lipid
rafts.
6–8 Lipid
rafts have been defined as highly dynamic and fluctuating nanoscale assemblies of cholesterol and
sphingolipids that in the presence of lipid- or protein-mediated activation events become stabi-
3
lized to compartmentalize cellular
processes.
2,5,9 However, the true nature of these nanodomains
remains debated with many conflicting evidences and predicated domain sizes in the broad range
of 10-200 nm, primarily because of their transient nature and nanoscopic
sizes.
8–14
Early investigations on membrane organization were mostly based on fluorescence recovery
after photobleaching
(FRAP)
15 and single particle tracking
(SPT).
3,16 Both techniques are limited
either in space (with
µ
m2 probe area in FRAP) or in time (with millisecond temporal resolution
in SPT). Fluorescence correlation spectroscopy (FCS) is a widely adopted alternative for studying
dynamics and biomolecular
interactions.
17 FCS determines the average transit time from statistical
averaging over many individual molecule diffusion
events.
18 Its high temporal resolution together
with its rather straightforward data analysis makes FCS an attractive tool to probe the spatiotempo-
ral organization of cell
membranes.
10,11,19 However, conventional FCS on confocal microscopes is
unable to resolve the nanoscale organization of lipids due to the limited 200 nm spatial resolution
set by diffraction.
Various approaches have been implemented over the past decade to breach the diffraction limit
in FCS, but membrane studies have so far remained above a 40-50 nm detection size. Stimulated
emission depletion microscopy (STED) constrains the excitation spot down to
∼
30
nm
20 and
has been combined with FCS to explore the nanoscale dynamics occurring in lipid membranes
on living
cells.
21–24 An alternative strategy takes advantage of nanophotonic structures to engi-
neer the light intensity distribution at the
nanoscale.
25 Some notable designs include zero-mode
waveguides, 26–31 bowtie
structures,
32–34 gold
nanorods
35 and sub-wavelength tip based NSOM
probes.
36,37 These various approaches allow to confine the illumination light in the range of 50 to
100 nm. Resonant optical nanogap antennas have shown great potential to further constrain the
laser light on a sub-20 nm
scale
38 and greatly enhance the light-matter
interactions.
39–42 However,
so far the applications of such resonant nanogap antennas have been mostly employed to probe
fluorescent
molecules in solutions at high micromolar concentration. Recently, we have developed
a new class of nano-antennas that maximizes access to the antenna hotspot region together with
extreme planarity and
biocompatibility.
43 This methodology has been validated using model lipid
4
membranes,
44 underscoring its high potential to investigate the nanoscale architecture of living
cell membranes.
In this work, we combine FCS with these planar optical nanogap antennas to investigate for
the first time the nanoscopic organization of lipid rafts in the plasma membrane of living cells
at
a
spatial resolution of 10 nm. The antenna design has been specifically developed for FCS with
sub-diffraction spatial
resolution.
41 It combines a central nanogap antenna to create the highly
confined electromagnetic hotspot (of dimensions
∼
10 and 35 nm) surrounded by a rectangular
cladding to prevent direct excitation of background molecules diffusing away from the central
nanogap. By applying planarization, etch-back and template stripping methods, we have improved
our initial design to produce arrays of nano-antennas with controlled gap sizes, sharp edges and
planar hotspots facing the upper surface of the
sample.
43 Using these planar nano-antennas with
gap sizes down to 10 nm, we investigate here the diffusion dynamics of phosphoethanolamine
(PE) and sphingomyelin (SM) on the plasma membrane of living Chinese hamster ovary (CHO)
cells. Compared to earlier works using confocal
FCS,
10,11,19 nanoaperture
FCS,
26–30 or STED-
FCS, 21–24 our study is the first to breach into the sub-30 nm spatial scale on living cell membranes.
Together with cholesterol depletion experiments, we provide compelling evidence of short-lived
cholesterol-induced
∼
10 nm nanodomain partitioning in plasma membranes and discuss the im-
pact of these results in the context of lipid rafts.
The planar antenna platform contains multiple gold
nano-antenna
arrays with nominal gap sizes
of 10 nm and 35 nm on which a circular cell culture well is mounted for live CHO cell culturing.
Figure 1a,b depicts the strategy chosen for the fluorescence live cell experiment conducted on the
nanogap antenna platform. A 640 nm laser light illuminates a single nano-antenna in the sample
plane of an inverted microscope with a high-NA water immersion objective. Throughout this study,
the linear polarization of the laser beam is set parallel to the antenna main axis so as to excite the
nanogap
mode.
43 A highly confined nanometric hotspot of illumination light is created on the
surface of the nanogap region which is in direct contact with the adhered plasma membranes of
living CHO cells. Importantly, the planarization strategy avoids possible curvature induced effects
5
on the cell membrane and thus provides an ideal platform for live cell membrane
research
29,30
(Supporting Information Fig. S1 shows AFM images indicating a planarity better than 3 nm for the
top surface).
The cells were incubated on the nano-antennas at 37o C for nearly 48 hours prior to the experi-
ments to allow them to freely grow and adhere onto the antenna platform. Lipid analogs (either PE-
or SM-BSA complexes) labeled with the lipophilic organic dye Atto647N were incorporated into
the plasma membrane of the living cells just before the fluorescence measurements (see Methods
for details on staining protocol). The choice of Atto647N as fluorescent dye allows an excellent
overlap with the antenna’s main plasmonic resonance (Fig. S2), maximizing the fluorescence en-
hancement in the nanogap. Figure 1c shows a representative confocal image of the morphology
of the CHO cells adhered on a glass coverslip taken after the incorporation of the fluorescent lipid
analogs.
Figure 2a,b shows representative single-molecule fluorescence time traces for PE and SM in
the confocal and in the nano-antenna configuration. The resolution given by the diffraction limited
spot in the confocal scheme does not allow to resolve heterogeneities that may occur at the sub-
200 nm spatial scale, and as a result, the time traces for both PE and SM appear indistinguishable.
In contrast, the highly confined surface hotspot originating from the 10 nm gap antenna clearly
reveals differences in the characteristic diffusion dynamics for PE and SM. As shown in Fig. 2b,
PE displays sharp peaks in the fluorescence time trace as a result of the sub-diffraction excitation
hotspot created by the planar nanogap antenna. Unlike PE, the signature of SM is discernibly dif-
ferent at the nanoscale: the short bursts (a hallmark of free diffusion in ultra-small detection areas)
are accompanied by high intensity bursts of significantly longer durations. This is a direct indi-
cation that the nanoscopic diffusion of SM on the cell membrane is deviating from free Brownian
diffusion as compared to larger macroscopic scales.
To provide more quantitative information about the fluorescence time traces, we performed a
fluorescence burst analysis to represent the distributions of burst duration versus burst intensity
(see
Methods).
34 Figure 2c shows the results for both PE and SM for the 10 nm nanogap antenna
6
compared to the confocal configuration. The scatter plots for PE and SM in the confocal configura-
tion show no visible differences with burst durations in the range 1-100 ms and intensities around
20-30 counts/ms. However, in stark contrast, the distributions obtained on the nano-antennas show
clear differences between PE and SM. Diffusion events in sub-ms time scales are notably observed
with the nano-antennas exhibiting burst durations as short as 10
µ
s. Such short events are more
than two orders of magnitude faster than in the case of the confocal reference. Regarding the diffu-
sion dynamics for PE (red dots) probed with the nanogap antennas a general trend can be deduced,
namely, brighter events arise at shorter timescales. These can be understood as the detection of a
“best burst event” directly resulting as a consequence of an individual molecule diffusing through
the hotspot in the optimal position and orientation for maximum enhancement. The tighter the
excitation beam confinement, the higher is the local intensity which leads to higher fluorescence
intensity and shorter burst duration (see Supplementary Information Fig. S3 and S4 for additional
fluorescent
time traces and analysis on different antennas and cells). We thus relate the events with
burst durations below 1 ms to the trajectories occurring within the nanogap
region.
34 In the case of
PE, the bursts with durations above 1 ms feature a lower intensity in the range of 20-70 counts/ms,
which is only slightly increased as compared to the confocal level. We assign these longer burst
duration events to the residual excitation of diffusing molecules within the larger 300
×
140 nm2
box aperture region where the electromagnetic field intensity enhancement is negligible and thus
comparable to the confocal reference.
In contrast to PE, SM probed with the nano-antenna arrays shows a significantly broader dis-
tribution of burst lengths against peak burst intensities (Fig. 2c). High intensities are observed for
burst durations below and above 1 ms. Since these events were not observed for PE, we relate
their occurrence to nanoscopic heterogeneities such as transient molecular complexes on the cell
membrane hindering the diffusion of SM. To support this conclusion, we perturbed the cholesterol
composition in the cell membrane with
methyl-
β
-cyclodextrin (MCD) as cholesterol is expected
to play a significant role in the formation and stability of the lipid nanodomains. The result of the
burst analysis for SM after MCD treatment recovers a distribution which closely resembles the one
7
for PE (Supporting Information Fig. S5). In other words, the intense bursts of duration between
0.1 and 10 ms disappear after cholesterol depletion, consistent with the loss of nanodomains. Alto-
gether, the results from the fluorescence burst analysis demonstrate the benefits of planar nanogap
antennas to explore the nanoscopic organization of lipids in live cell membranes. Clear differ-
ences between PE and SM diffusion dynamics are unveiled that otherwise would remain hidden in
confocal measurements.
To further support these results, we performed fluorescence correlation spectroscopy (FCS)
analysis. FCS records the fluorescence intensity fluctuations as the fluorophores transit through
the detection spot. These fluctuations are analyzed by computing the temporal autocorrelation
function, averaging over thousands of single-molecule diffusion events. We used two different gap
sizes (10 and 35 nm) to quantify the lipid dynamics for increasing detection areas in cell mem-
branes. Figure 3a,b shows the normalized correlation traces for PE and SM in case of the nano-
antennas and the confocal reference. Each of these traces is taken on an individual nano-antenna
(more traces are shown in Fig. S3 and S4 to demonstrate the consistency of our results). Similar
to the burst analysis, we find no significant differences between the FCS curves for PE and SM for
the confocal reference (gray circles in Fig. 3a,b the overlay of the confocal FCS data is shown in
Fig. S6), yielding comparable diffusion times of 25
±
4 ms (PE) and 30
±
4 ms (SM), respectively.
In the case of the nano-antennas, we observe that decreasing the gap size leads to a faster diffu-
sion,
confirming
that the
fluorescence
signal stems from the nanogap region. We use a two-species
model to fit the FCS data in order to account for the
fluorescence
contributions stemming from the
nanogap and from the surrounding aperture area (see details in Methods section). A key feature
in FCS is that the molecules contribute to the correlation amplitude in proportion to the square of
their fluorescence brightness, hence the signal from molecules in the nanogap experiencing maxi-
mum enhancement will have a dominating contribution to the FCS
curves.
45 The complete results
and values for the FCS fits are detailed in the Supporting Information Tables S1-S3.
The differences between PE and SM diffusion dynamics are highlighted in Fig. 3c where a
direct comparison of the FCS data for the 10 nm gap antenna is shown for different fluorescent
8
lipid analogs. Contrarily to the confocal case (Fig. S6), the difference in diffusion times between
the two lipids becomes more prominent at the nanoscale, with PE exhibiting diffusion times of
0
.
25
±
0.06 ms and SM of
0
.
35
±
0.04 ms. Moreover, after MCD treatment, the diffusion dynamics
for cholesterol-depleted SM closely resembles that of PE with a diffusion time of 0.19
±
0.03 ms
(Fig. 3d). These FCS results confirm the presence of cholesterol-enriched nanodomains hindering
the diffusion of SM, in agreement with the results found for the fluorescence burst analysis. In
addition, we retrieved an anomaly value alpha for SM that depended on the probed area, deviating
from unity as the illumination area reduced, from α
∼
0.85 (for the 35 nm gap antenna) to α
∼
0
.
65
(for the 10 nm gap antenna), which is fully consistent with hindered diffusion (see Table S1 to S3).
In contrast, the α values were significantly larger and closer to unity for the cases of PE and SM
after MCD treatment (α
∼
0.85) and did not depend on the probe area, as expected for Brownian,
unhindered diffusion.
To further analyze and exploit the FCS data we take advantage of the large number of planar
nano-antennas with controlled gaps to carry out a FCS analysis over 60 different antennas and
cells. This approach follows the so-called FCS diffusion
law,
19,28 which is a representation of the
diffusion time versus the detection area. Extrapolation of the experimental curve to the intercept
with the time axis provides information on the type of diffusion exhibited by the molecule, i.e.,
free diffusion is characterized by a linear curve crossing the origin (0,0), while hindered diffusion
due to the occurrence of nanodomains leads to a positive intercept on the time
axis.
19,28 The nano-
antenna detection area was estimated as the product of the gap size (measured by transmission
electron microscopy TEM) times the full width at half maximum for the intensity profile along
the direction perpendicular to the antenna main axis (simulated by finite difference time domain
(FDTD) method, see Supporting Information Fig. S7 for simulations results). Moreover, the area
sizes were further
confirmed
by calibration measurements both on freely diffusing dyes in solution
and on pure PE lipid bilayers using antennas of different gap
sizes.
44 The 10 and 35 nm gap
antennas correspond respectively to 300 nm2 and 1250 nm2 illumination areas. As the diffusion
time proportionally scales with the detection area, the diffusion coefficient D is retrieved from
9
the slope of the linear fit matching the measured transient diffusion times obtained from the FCS
curves versus the effective detection areas according to the relation D
=
probe
area
/
4
×
τ
diff
.
18
Figure 4a-c summarizes the characteristic diffusion times for PE, SM and SM after cholesterol
depletion for two antenna gap areas. The extension to include the confocal data is shown in Fig. S8.
From these graphs we derive the following three values plotted in Fig. 4d-f: the diffusion coeffi-
cient (from the slope), the time axis intercept (by extrapolating the linear fit for vanishing probe
area) and the normalized spread in the data points
(defined
as the width of upper and lower quartiles
divided by the median value). The diffusion
coefficients
derived from nano-antenna measurements
are DPE = 0.44
±
0.07
µ
m2/s, DSM = 0.38
±
0.19
µ
m2/s and DMCD-SM = 0.46
±
0.07
µ
m2/s
(Fig. 4d) and they are consistent with the confocal measurements and values reported indepen-
dently using
STED-FCS.
21 These coefficients represent the diffusion speed in the lipidic region
between the nanodomains, with an additional contribution from diffusion within the nanodomains
and diffusion of the domains themselves.
Extrapolating the fits in Fig. 4a-c towards diminishing probe area leads to the intercepts with
the time axis as summarized in Fig. 4e. The almost zero intercept hitting the origin observed for
PE confirms the expected free Brownian motion diffusion mode. In stark contrast, SM features a
positive y-intercept of about 100
µ
s, which highlights a significant deviation from free Brownian
diffusion and the occurrence of nanoscopic domains hindering SM diffusion. Depletion of choles-
terol results on SM diffusion with a close-to-zero time intercept, demonstrating the crucial role
of cholesterol establishing the nanodomains and hindering SM diffusion. Such small nanoscale
heterogeneities have never been detected so far with confocal microscopy, although STED-FCS
down to 1000 nm2 detection area could infer their
occurrence.
21 Our results are fully aligned with
these previous findings and importantly, we further reduce the detection areas down to 300 nm2 .
Lastly, we take a closer look at the statistical dispersion of the diffusion times for each gap
area, and introduce the normalized data spread as the width from upper to lower quartiles divided
by the median value (Fig. 4f). The spread in diffusion times for PE and SM after MCD treatment
remains under 25% and can be partially assigned to nanometer variations of the gap size between
10
nanoantennas.
44 These variations stem from the nanofabrication process as a consequence of the
finite
grain size of gold and/or the scattering of electrons used during the electron beam lithography.
In contrast to PE and MCD-SM, the data for SM features a
significantly
higher statistical dispersion
around 50%, which cannot be related solely to dispersion in the nanoantenna sample, but instead
it results from large variations in the SM diffusion behavior, as already noted for the fluorescence
burst analysis (Fig. 2c). These results are fully consistent with the presence of cholesterol-enriched
nanodomains affecting SM diffusion.
Altogether, our results provide compelling evidence for the existence of highly transient and
fluctuating
nanoscale
assemblies
of sterol and
sphingolipids
in living cell membranes. These exper-
imental observations stand in excellent agreement with the notion that without stabilizing proteins,
lipid rafts can be viewed as intrinsic nanoscale membrane heterogeneities that are small and highly
transient. 2,6–8 We estimate the characteristic residence time of the fluorescent SM lipid analogs in
the nanodomain from the y-intercept in Fig. 4b,e, and find a value around 100
µ
s . The typical size
of the nanodomains could in principle also be deduced from the FCS diffusion laws which should
feature a characteristic transition from confined to normal
diffusion.
19,28 As we do not observe this
characteristic transition in our data, we conclude that the typical size of the nanodomain is smaller
than the smallest gap size of our nanoantenna, that is 10 nm. Both the typical nanodomain size
about 10 nm and the transient time about 100
µ
s stand in good agreement with the predictions
from stochastic
models
46 and recent high-speed interferometric scattering (iSCAT) measurements
on mimetic lipid bilayers containing
cholesterol.
12 We believe that this shorter characteristic time
as compared to earlier experimental works using
STED-FCS
21–23 is related to the smaller 10 nm
resolution achieved in our case. The nanoantenna approach is straightforward to implement on
any confocal microscope equipped for FCS as contrarily to STED, it does not require adding any
supplementary illumination beam. As additional advantage, the excellent planarity of the surface
rules out any artefact potentially induced by the curvature of the cell
membrane.
29 We believe that
these advantages and the excellent spatiotemporal resolution largely compensate for the need for
nanofabrication and the more complex FCS fitting procedure.
11
In conclusion, we have demonstrated the promising approach of exploiting planar optical nano-
antennas with accessible surface nanogaps to investigate the nanoscale architecture of live cell
membranes. The key strengths of our approach rely on the 10 nm spatial resolution combined
with a microsecond time resolution on a nearly perfectly flat substrate compatible with live cell
culturing. The single-molecule data on nanoantennas reveal striking differences between PE and
SM diffusion dynamics that remain hidden in confocal measurements. Fluorescence burst and
correlation spectroscopy analysis for PE are consistent with a free Brownian diffusion model. In
contrast, the diffusion dynamics of SM at the nanoscale show heterogeneities in both time and
space which are cholesterol dependent. Indeed, removal of cholesterol leads to a recovery of free
Brownian diffusion for SM, consistent with the loss of nanodomains. Our results are consistent
with the existence of dynamic nanodomains on the plasma membranes of living cells of
∼
10 nm
diameter which is comparable to our measurement gap size. The corresponding transient trap-
ping times are short of about
∼
100
µ
s. We believe that the combination of optical nano-antennas
with fluorescence microscopy has a high potential to investigate the dynamics and interactions of
raft associated proteins and their recruitment into molecular complexes on the plasma membrane
of living cells. The proposed technique is fully bio-compatible and thus provides ample oppor-
tunities for biophysics and live cell research with single-molecule sensitivity at nanometric and
(sub)microsecond spatiotemporal resolution, far beyond the diffraction limit of light.
Methods
Planar
Nanogap Antenna Array
Fabrication
Large scale nano-antenna arrays with surface
nanogaps were fabricated by combining electron beam lithography with planarization, etch back
and template
stripping.
43 First, EBL was used to pattern features with negative tone hydrogen
silsesquioxane (HSQ) resist on top of 100 mm silicon wafer. A 50 nm thick gold film was then
deposited by electron beam evaporation at low temperature to reduce the gold grain size. Flowable
oxide (Dow Corning FOX-16) was spun to planarize the overall structure and was followed by an
12
etch back step to selectively remove the sacrificial top metal layer clearing the aperture geometry.
A 30 s etch with hydrofluoric acid diluted 1:10 in deionized water was then used to clear out the
residual HSQ in the antennas. Finally, the antenna structures were embedded in an UV curable ad-
hesive polymer (OrmoComp, Microresiste Technology GmbH) and stripped away from the silicon
wafers. The narrowest gap region lying at the bottom of the structure due to the metal diffusion
during the evaporation, was now flipped over to maximize the contact with the sample providing
direct accessibility to the plasmonic antenna hotspot. This method is fully scalable and shows ex-
cellent geometry control and planarity (see Supporting Information Fig. S1).
Cell culture, Atto647N-labeling and Cholesterol depletion of CHO cells CHO cells were
seeded on a coverslip containing planar nano-antennas with surface nanogaps and were allowed
to grow and spontaneously attach at 37o C in a controlled atmosphere with 5% of CO2 for nearly
48 hours. Lipid conjugates were separately prepared by labeling 1,2-Dipalmitoyl-sn-glycero-3-
phosphoethanolamine (DPPE) and Sphingomyelin (SM) with the organic dye Atto647N (from
Invitrogen) as described in
Ref.
21 Prior to the fluorescence experiments, the lipid analogues were
incorporated in the cell membrane during a 3 mins incubation period at room temperature dis-
solved in the corresponding medium for CHO cells (Ham’s F12 nutrient mixture). Stained cell
cultures were rinsed and washed to remove residual dye molecules before placing the sample cov-
erslip on the piezo-stage of an inverted microscope to carry out the measurements. For cholesterol
depletion experiments, the CHO cells were incubated in serum free buffer with 10 mM methyl-
β -cyclodextrin (MCD) for 30 mins at 37oC, and then the fluorescent labeling was carried out as
previously described. All fluorescence stainings were performed at a
∼
300 nM concentration of
Atto647N and the measurements were completed within 30 mins after the incorporation of the
fluorescent analogs. From the number of detected fluorescence bursts (Fig. 2c) and the FCS am-
plitude, we estimate that the density of fluorescent lipids for the antenna experiments is on the
order of 20 to 80 probes per
µ
m2.
13
Experimental
Setup The experiments were performed with a commercial MicroTime 200
setup equipped with an inverted confocal microscope (Olympus 60
×
, 1.2 NA water-immersion
objective) and a three-axis piezoelectric stage (PhysikInstrumente, Germany) allowing to select
individual nano-antennas. A linearly polarized 640 nm picosecond laser diode (Pico-Quant LDH-
D-C-640) in continuous wave mode was used to resonantly excite individual nano-antennas with
the laser linear polarization aligned with the main axis of the antenna dimer. The emitted fluo-
rescence signal was collected in epi-detection mode through a dichroic mirror and the signal was
finally split into two avalanche photodiodes (PicoQuant MPD-50CT). An emission filter and a
band pass 650-690 nm filter just before each detector was used to eliminate the scattered light by
the excitation laser. A 30
µ
m pinhole in the detection arm yielded 0.5 fl confocal detection volume
at the sample plane. The fluorescence time traces were recorded on a fast time-correlated sin-
gle photon counting module in the time-tagged time-resolved mode (PicoQuant MPD-50CT). All
the fluorescence measurements were performed by illuminating the sample at an excitation power
density of
∼
2-3 kW/cm2. The measurements were acquired for a typical run time of 50 s and the
correlation amplitudes were computed for
∼
20 s windows with the commercial software package
SymPhoTime 64. Cells were cultured on different antenna samples, each sample containing dif-
ferent gap sizes.
Fluorescence Burst Analysis Single-molecule fluorescence time traces were acquired in the
Tagged Time-Resolved (TTTR) mode (recording each event at its arrival time) with 4 ps temporal
resolution. Fluorescence bursts analysis was then carried out with a likelihood-based algorithm
to test the null hypothesis (no burst, recording compatible with background noise) against the hy-
pothesis that a single-molecule burst arises as a consequence of a molecule crossing the excitation
area. Probabilities associated to false positive and missing event errors were both set to
10
−
3
.
49
Fluorescence
Correlation
Spectroscopy The temporal fluctuations of the fluorescence inten-
sity
F
(
t
) around the average value were analyzed to compute the temporal correlation G(τ )
=
14
2
⟨
δ
F
(
t
).δ
F
(t
+
τ
)
⟩
/
⟨
F
(
t
)
⟩
, where
δ
F
(
t
)
=
F
(
t
)
−
⟨
F
(
t
)
⟩
is the fluctuation of the fluorescence
signal arising due to the molecules crossing the detection volume mediated by Brownian diffusion,
τ
is the delay (lag) time, and
⟨
⟩
indicates time averaging. The mobility of molecules shows strong
dependence on the local environment and thus in living systems the concept of ideal Brownian
diffusion may not always hold true. Considering possible anomalous diffusion in living cells, the
temporal correlation of the
fluorescence
intensity F can be written
as:
18
where τdi f f (i) the average residence time of the
i
t
h diffusing modality,
ρ
(i) denotes the respective
amplitude contribution and α (i) being anomaly parameter of the
same.
21 We find that the FCS
curves recorded with a nanoscopic illumination can only be fitted with a model assuming two
different diffusion modalities (i.e. ndiff
=
2).
To define the probe areas used in the FCS diffusion laws (Fig. 4a-c), we use the product of the
gap size (measured by TEM) by the full width at half maximum for the intensity profile along the
direction perpendicular to the antenna main axis (computed by FDTD), following a calibration for
model lipid
membranes.
44 Therefore, 10 nm and 35 nm gap sizes are associated respectively to
300 and 1250 nm2 probe areas.
Supporting Information
Supporting Information available: AFM image of antenna array, overlap between antenna’s res-
onance and fluorescence spectra, fluorescence data for PE on different nano-antennas, overlay of
FCS curves from different nanoantennas, representative time trace of cholesterol depleted SM,
overlay of the confocal FCS data for PE and SM, FDTD simulations of intensity distributions,
FCS diffusion laws with confocal data, fitting parameters for FCS curves.
15
Additional
inf
ormation
CM present address: Universitat de Vic, Universitat Central de Catalunya (UVic-UCC), C. de la
Laura 13, 08500 Vic, Spain
The authors declare no competing financial interests.
Acknowledgement
The authors thank Merche Rivas Jiménez, Felix Campelo and Erik Garbacik for technical support
and fruitful discussions. The research leading to these results has received funding from the Eu-
ropean Commission’s Seventh Framework Programme (FP7-ICT-2011-7) under grant agreements
ERC StG 278242 (ExtendFRET) and 288263 (NanoVista). Financial support by the Spanish Min-
istry of Economy and Competitiveness (“Severo Ochoa" Programme for Centres of Excellence in
R&D (SEV-2015-0522) and FIS2014-56107-R grants) and Fundacion Privada Cellex is gratefully
acknowledged. RR is supported by the Erasmus Mundus Doctorate Program Europhotonics (Grant
159224-1-2009-1-FR-ERA MUNDUS-EMJD). PMW is supported by the ICFOstepstone Fellow-
ship, a COFUND Doctoral Program of the Marie Skłodowska-Curie Action from the European
Commission. CM acknowledges funding from the Spanish Ministry of Economy and Competi-
tiveness and the European Social Fund (ESF) through the Ramón y Cajal program 2015 (RYC-
2015-17896).
Refer
ences
(1) Brown, D. A.; London, E. J. Biol. Chem., 2000, 275, 17221-17224.
(2) Lingwood, D.; Simons, K. Science, 2010, 327, 46-50.
(3) Kusumi, A.; Nakada, C.; Ritchie, K.; Murase, K.; Suzuki, K.; Murakoshi, H.; Kasai, R. S.;
Kondo, J.; Fujiwara, T.; Annu. Rev. Biophys. Biomol. Struct., 2005, 34, 351-378.
(4) Gowrishankar, K.; Ghosh, S.; Saha, S. C. R.; Mayor, S.; Rao, M. Cell 2012, 149, 1353-1367.
16
(5) Sezgin, E.; Levental, I.; Mayor, S.; Eggeling, C. Nat. Rev. Mol. Cell Biol., 2017, 18, 361-374.
(6) Mayor, S.; Rao, M. Traffic, 2004, 5, 231-240.
(7) Simons, K.; Gerl, M. J. Nat. Rev. Mol. Cell Biol., 2010, 11, 688-699.
(8) Hancock, J. F. Nat. Rev. Mol. Cell Biol., 2006, 7 456-62.
(9) Pike, L. J. J. Lipid Res. 2006, 47, 1597-1598.
(10) Lenne, P. F.; Wawrezinieck, L.; Conchonaud, F.; Wurtz, O.; Boned, A.; Guo, X. J.; Rigneault,
H.; He, H. T.; Marguet, D. EMBO J. 2006, 25, 3245-3256.
(11) Marguet, D.; Lenne, P. F.; Rigneault, H.; He, H. T. EMBO J. 2006, 25, 3446-3457.
(12) Wu, H.-M.; Lin, Y.-H.; Yen, T.-C.; Hsieh, C.-L. Sci. Rep. 2016, 6, 20542.
(13) van Zanten, T. S.; Cambi, A.; Koopman, M.; Joosten, B.; Figdor, C. G.; Garcia-Parajo, M. F.
Proc. Natl. Acad. Sci. USA., 2009, 106, 18557-18562.
(14) van Zanten, T. S.; Gómez, J.; Manzo, C.; Cambi, A.; Buceta, J.; Reigada, R.; Garcia-Parajo,
M. F. Proc. Natl. Acad. Sci. USA., 2010, 107, 15437-15442.
(15) Meder, D. M.; Moreno, J.; Verkade, P.; Vaz, W. L. C.; Simons, K. Proc. Natl. Acad. Sci.
USA., 2006, 103, 329-334.
(16) Dietrich, C.; Yang, B.; Fujiwara, T.; Kusumi, A.; Jacobson, K. Biophys. J., 2002, 82, 274-
284.
(17) Maiti, S.; Haupts, U.; Webb, W. W. Proc. Natl. Acad. Sci. USA., 1997, 94, 11753-11757.
(18) Bacia, K.; Kim, S. A.; Schwille, P. Nat. Methods, 2006, 3, 83-89.
(19)
Wawrezinieck,
L.; Rigneault, H.; Marguet, D.; Lenne, P.-F. Biophys. J., 2005, 89, 4029-4042.
(20) Kastrup, L.; Blom, H.; Eggeling, C.; Hell, S.W. Phys. Rev. Lett., 2005, 94(17), 1-4.
17
(21) Eggeling, C.; Ringemann, C.; Medda, R.; Schwarzmann, G.; Sandhoff, K.; Polyakova, S.;
Belov, V. N.; Hein, B.; von Middendorff, C.; Schönle, A.; Hell, S. W. Nature, 2009, 457,
1159-1162.
(22) Mueller, V.; Ringemann, C.; Honigmann, A. ; Schwarzmann, G.; Medda, R.; Leutenegger,
M.; Polyakova, S.; Belov, V. N.; Hell, S. W.; Eggeling, C. Biophys. J., 2011, 101, 1651-1660.
(23) Honigmann, A.; Mueller, V.; Ta, H.; Schoenle, A.; Sezgin, E.; Hell, S. W.; Eggeling, C. Nat.
Commun. 2014, 5, 5412.
(24) Vicidomini, G.; Ta, H.; Honigmann, A. ; Mueller, V.; Clausen, M. P.; Waithe, D.; Galiani, S.;
Sezgin, E.; Diaspro, A.; Hell, S. W.; Eggeling, C. Nano Lett., 2015, 15, 5912-5918.
(25) Punj, D.; Ghenuche, P.; Moparthi, S. B.; de Torres, J.; Grigoriev, V.; Rigneault, H.; Wenger,
J. WIREs Nanomed. Nanobiotechnol., 2014, 6, 268-282.
(26) Edel, J. B.; Wu, M.; Baird, B.; Craighead, H. G. Biophys. J. 2005, 88, L43-L45.
(27) Moran-Mirabal, J. M.; Torres, A. J.; Samiee, K. T.; Baird, B. A.; Craighead, H. G. Nanotech-
nology 2007, 18, 195101.
(28) Wenger, J.; Conchonaud, F.; Dintinger, J. Wawrezinieck, L.; Ebbesen, T. W.; Rigneault, H.;
Marguet, D.; Lenne, P.-F. Biophys. J., 2007, 92, 913-919.
(29) Kelly, C. V.; Baird, B. A.; Craighead, H. G. Biophys. J. 2011, 100, 34-36.
(30) Kelly, C. V.; Wakefield, D. L.; Holowka, D. A.; Craighead, H. G.; Baird, B. A. ACS Nano
2014, 8, 7392-7404.
(31) Richards, C. I.; Luong, K.; Srinivasan, R.; Turner, S. W.; Dougherty, D. A.; Korlach, J.;
Lester, H. A. Nano Lett. 2012, 12, 3690-3694.
(32) Lohmüller, T.; Iversen, L.; Schmidt, M.; Rhodes, C.; Tu, H. L.; Lin, W. C.; Groves, J. T.
Nano Lett. 2012, 12, 1717-1721.
18
(33) Mivelle, M.;, Van Zanten, T. S; Neumann, L.; van Hulst, N.; García-Parajo, M. F. Nano Lett.,
2012, 12 (11), 5972-5978.
(34) Flauraud, V.; van Zanten, T.S.; Mivelle, M.; Manzo, C.; García-Parajo, M.F.; Brugger, J.
Nano Lett. 2015, 15, 4176-4182.
(35) Pradhan, B.; Khatua, S.; Gupta, A.; Aartsma, T.; Canters, G.; Orrit, M. J. Phys. Chem. C
2016, 120, 25996-26003.
(36) García-Parajo, M. F. Nat. Photonics, 2008, 2, 201-203.
(37) Manzo, C.; Van Zanten, T. S; García-Parajo, M. F. Biophys. J., 2011, 1000, 08-10.
(38) Novotny, L.; van Hulst, N. Nat. Photonics 2011, 5, 83-90.
(39) Kinkhabwala, A.; Yu, Z. F.; Fan, S. H.; Avlasevich, Y.; Mullen, K.; Moerner, W. E. Nat.
Photonics 2009, 3, 654-657.
(40) Acuna, G. P.; Möller, F. M.; Holzmeister, P.; Beater, S.; Lalkens, B.; Tinnefeld, P. Science
2012, 338, 506-510.
(41) Punj, D.; Mivelle, M.; Moparthi, S.B.; van Zanten, T.S.; Rigneault, H.; van Hulst, N.F.;
García-Parajo, M.F.; Wenger, J. Nat. Nanotechnol. 2013, 8, 512-516.
(42) Puchkova, A.; Vietz, C.; Pibiri, E.; Wünsch, B.; Sanz Paz, M.; Acuna, G. P.; Tinnefeld, P.
Nano Lett. 2015, 15, 8354-8359.
(43) Flauraud, V.; Regmi, R.; Winkler, P. M.; Alexander, D. T. L.; Rigneault, H.; van Hulst, N. F.;
García-Parajo, M. F.; Wenger, J.; Brugger, J. Nano Lett. 2017, 17, 1703-1710.
(44) Winkler, P. M.; Regmi, R.; Flauraud, V.; Brugger, J.; Rigneault, H.; Wenger, J.; García-
Parajo, M. F. ACS Nano 2017, 11, 7241-7250.
(45) Langguth, L.; Koenderink, A. F. Opt. Express 2014, 22, 15397-15409.
19
(46) Nicolau, D. V.; Burrage, K.; Parton, R. G.; Hancock, J. F. Mol. Cell. Biol. 2006, 26, 313-323.
(47) Spillane, K. M.; Ortega-Arroyo, J.; de Wit, G.; Eggeling, C.; Ewers, H.; Wallace, M. I.;
Kukura, P. Nano Lett. 2014, 14, 5390-5397.
(48) Spindler, S.; Ehrig, J.; König, K.; Nowak, T.; Piliarik, M.; Stein, H. E.; Taylor, R. W.;
Garanger, E.; Lecommandoux, S.; Alves, I. D.; Sandoghdar, V. J. Phys. Appl. Phys. 2016, 49,
274002.
(49) Zhang, K.; Yang, H. J. Phys. Chem. B, 2005, 109, 21930-21937.
20
Figure 1:
Planar
gold
nano-antenna arrays
for probing single-molecule dynamics in
the
plasma
membrane
of living cells. (a) CHO cells are seeded onto a microscopic coverslip con-
taining multiple planar nano-antennas with 10 and 35 nm gap sizes. The inset shows the cross
section of the antenna-in-box stripped and embedded into a polymer, bringing the region of max-
imum electromagnetic field intensity onto the surface in direct contact with the plasma membrane
of living cells. (b) From left to right: macro-photograph of a coverslip with a stripped Au film with
large-scale planar antenna arrays; dark field optical micrograph of a small portion of the antenna
arrays showing here 625 antennas with 10 nm nominal gap size; transmission electron microscope
(TEM) images of antennas with 10 and 35 nm gap size. (c) Confocal image of CHO cells showing
the morphology after incorporating the fluorescent SM lipid analog labeled with Atto647N.
21
Figure 2: Single-molecule fluorescence time traces in living CHO cells. (a,b) Fluorescence time
traces for phosphoethanolamine (PE, left) and sphingomyelin (SM, right) labeled with Atto647N
recorded with confocal (a) and with a 10 nm gap planar nano-antenna (b). The binning time is
0.1 ms for all traces. The diffraction-limited spot in the confocal configuration cannot resolve the
nanoscopic and heterogeneous membrane organization, thus results in indistinguishable fluores-
cence time traces for both PE and SM. However, the highly confined nano-antenna hotspot reveals
clear differences in the diffusion dynamics of PE and SM. (c) The fluorescence time traces are
analyzed to produce scatter plots showing the distribution of fluorescence burst intensity versus
burst duration. Single-molecule events in sub-ms time scales are observed with nano-antennas
(color dots) as the confined electromagnetic hotspots allow to probe the dynamics occurring be-
yond the diffraction limit. Single molecule events obtained with confocal illumination are shown
for comparison (black dots).
22
Figure 3:
Nano-antenna
FCS on living cell
membranes.
(a,b) Normalized fluorescence corre-
lation curves for Atto647N labeled PE (a) and SM (b) lipid analogs probed with nano-antennas
of varying gap size. The color lines are experimental data and the black curves are numerical
fits. Each FCS trace is a representative example taken on an individual nano-antenna. FCS curves
recorded on different nano-antennas and different cells are shown in Fig. S4. The diffraction-
limited confocal measurements are shown in gray for direct comparison. (c) Comparison of FCS
curves for PE and SM for a 10 nm gap antenna. Unlike the confocal reference, the nano-antenna re-
veals clear differences between the dynamics of PE and SM at the nanoscale. (d) After cholesterol
depletion, the SM diffusion dynamics are significantly faster and resemble the PE case.
23
Figure 4:
Characteristic
diffusion dynamics of
membrane
lipids probed with
ultra-confined
nano-antenna
hotspots. The diffusion time measured by FCS (for 60 different nano-antennas) is
plotted as a function of the probe area for PE (a), SM (b), and SM after MCD treatment(c). The
solid lines are linear fits through the median values. In the case of free diffusion, the origin (0,
0
)
is aligned with the expected line, while a positive intercept at the y-axis denotes hindered diffusion
due to nanodomains. (d) The diffusion
coefficients
computed from the slopes in a-c are compared
with confocal results. (e) y-axis intercept deduced from the linear fits in a-c. PE and MCD-treated
SM show near-zero y-intercept consistent with free diffusion, while the significant y-intercept for
SM indicates that the diffusion is constrained by nanodomains. (f) Normalized spread in diffusion
time (width of upper and lower quartiles / median) in each case. The large dispersion observed is
SM is another indication that sphingolipids are preferentially recruited into transient nanoscopic
domains.