IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 64, NO. 5, MAY 2017 785
Vibrational Responses of Bound and Nonbound
Targeted Lipid-Coated Single Microbubbles
Tom van Rooij, Member, IEEE, Inés Beekers, Student Member, IEEE, Kirby R. Lattwein,
Antonius F. W. van der Steen, Fellow, IEEE, Nico de Jong, Member, IEEE,
and Klazina Kooiman, Member, IEEE
Abstract—One of the main challenges for ultrasound
molecular imaging is acoustically distinguishing nonbound
microbubbles from those bound to their molecular target.
In this in vitro study, we compared two types of in-house
produced targeted lipid-coated microbubbles, either consisting of
1,2-dipalmitoyl-sn-glycero-3-phosphocholine, C16:0 (DPPC) or
1,2-distearoyl-sn-glycero-3-phosphocholine, C18:0 (DSPC) as the
main lipid, using the Brandaris 128 ultrahigh-speed camera to
determine vibrational response differences between bound and
nonbound biotinylated microbubbles. In contrast to previous
studies that studied vibrational differences upon binding, we
used a covalently bound model biomarker (i.e., streptavidin)
rather than physisorption, to ensure binding of the biomarker to
the membrane. The microbubbles were insonified at frequencies
between 1 and 4 MHz at pressures of 50 and 150 kPa. This
paper shows lower acoustic stability of bound microbubbles,
of which DPPC-based microbubbles deflated most. For DPPC
microbubbles with diameters between 2 and 4 µmdrivenat
50 kPa, resonance frequencies of bound microbubbles were all
higher than 1.8 MHz, whereas those of nonbound microbubbles
were significantly lower. In addition, the relative radial excursions
at resonance were also higher for bound DPPC microbubbles.
These differences did not persist when the pressure was increased
to 150 kPa, except for the acoustic stability which further
decreased. No differences in resonance frequencies were observed
between bound and nonbound DSPC microbubbles. Nonlinear
responses in terms of emissions at the subharmonic and second
harmonic frequencies were similar for bound and nonbound
microbubbles at both pressures. In conclusion, we identified
differences in vibrational responses of bound DPPC microbubbles
with diameters between 2 and 4 µm that distinguish them from
nonbound ones.
Index Terms— Biotin-streptavidin, lipid-coating, molecular
imaging, nonlinear behavior, targeted microbubbles, ultrahigh-
speed optical imaging, ultrasound contrast agents.
Manuscript received November 3, 2016; accepted March 3, 2017. Date of
publication March 7, 2017; date of current version May 1, 2017. This work
was supported in part by NanoNextNL, a microtechnology and nanotechnol-
ogy consortium of the Government of The Netherlands, in part by the Center
for Translational Molecular Medicine, and in part by the Netherlands Heart
Foundation (PARISk).
T. van Rooij, I. Beekers, K. R. Lattwein, and K. Kooiman are with
the Department of Biomedical Engineering, Thorax Center, Erasmus MC,
3000 Rotterdam, The Netherlands (e-mail: tvrooij
.
@.
gmail.com).
A. F. W. van der Steen and N. de Jong are with the Department of
Biomedical Engineering, Thorax Center, Erasmus MC, 3000 Rotterdam,
The Netherlands, and also with the Laboratory of Acoustical Wavefield
Imaging, Faculty of Applied Sciences, Delft University of Technology,
2628 Delft, The Netherlands and also with the Netherlands Heart Institute,
3511 Utrecht, The Netherlands.
This paper has supplementary downloadable material available at
http://ieeexplore.ieee.org., provided by the author. Ultrahigh-speed recordings
of a bound DPPC and DSPC microbubble insonified at a pressure of 50 and
150 kPa and a frequency of 1 and 4 MHz.
Digital Object Identifier 10.1109/TUFFC.2017.2679160
I. INTRODUCTION
U
LTRASOUND contrast agents that consist of targeted
microbubbles are emerging in their applications for
ultrasound molecular imaging [1]–[3]. These microbubbles
have a ligand attached to their shell by which they can be
targeted to a specific biomarker, for example, α
v
β
3
that is
expressed on the cellular membrane of endothelial cells in
neovasculature [4], [5]. For successful translation of ultrasound
molecular imaging to the clinic, two major problems still need
to be tackled: 1) producing microbubbles of the same size that
also behave identical in an ultrasound field and 2) distinguish-
ing the response of a single targeted microbubble bound to
a specific biomarker from a nonbound targeted microbubble.
Since microbubbles of the same size can still have dif-
ferent acoustic properties [6]–[10], producing monodisperse
microbubbles may not necessarily result in microbubbles
that have, for example, the same resonance frequency. But
if it is possible to determine the acoustic parameters that
are specific for bound targeted microbubbles, they may be
distinguished from nonbound targeted microbubbles based on
their acoustic signal. Several studies investigated the difference
in acoustic properties of bound and nonbound microbubbles,
but these studies reported conflicting results. In the low-
frequency range (2–4 MHz) a shift in resonance frequency was
found for microbubbles after binding [11], [12], whereas at
11 and 25 MHz no shift was observed [13]. For the responses
at the subharmonic frequency either a change in frequency [13]
or no change in amplitude and frequency [14] was reported
upon binding. In contrast, for the response at the second
harmonic frequency, the results reported in [14] and [15]
were in agreement with each other: the amplitude increased
for bound microbubbles. Finally, Overvelde et al. [12] and
Zhao et al. [14] found a decrease in the vibrational response
at the fundamental frequency for bound microbubbles.
All acoustic studies on bound versus nonbound targeted
microbubbles used either physisorption as a method to attach
a model biomarker to an artificial surface (membrane or capil-
lary) [11], [12], [14], [15] or had the model biomarker embed-
ded in agarose [13]. Physisorption or physical adsorption relies
on electrostatic binding through van der Waals forces between
the biomarker and the membrane, but is in fact a very weak
bond [16], which can result in detachment of the biomarker
from the membrane or capillary. As a result, the biomarker
can cover the whole targeted microbubble, including the area
that is not directly in contact with the membrane. This was
0885-3010 © 2017 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted,
but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
786 IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 64, NO. 5, MAY 2017
reported in [17] for the model biomarker streptavidin that was
physisorbed to an OptiCell membrane. Functionalization of
lipid-coated microbubbles with streptavidin changes the prop-
erties, such as elasticity [18]–[20] and acoustic stability [20].
Consequently, the comparisons made in previous studies
between bound microbubbles and nonbound microbubbles are
in fact a comparison between bound lipid-coated microbubbles
covered by streptavidin and nonbound lipid-coated microbub-
bles, which did not have streptavidin on their shell. In addition,
both physisorption and embedding a model biomarker in
agarose are far from the in vivo situation, where biomarkers
are incorporated into the cellular membrane.
We covalently linked a model biomarker to an artificial
surface to study the vibrational responses of single bound
targeted microbubbles and nonbound targeted microbubbles
aiming to find parameters to discriminate them acoustically.
Super-resolution confocal laser scanning fluorescence
microscopy showed that covalent coupling of the model
biomarker streptavidin to a hydrogel prevented the streptavidin
to bind to the biotinylated lipid-shell of the microbubble
outside the binding area [21]. That study compared the
lipid distribution and binding area of two types of targeted
lipid-coated microbubbles that were either coated with mainly
1,2-dipalmitoyl-sn-glycero-3-phosphocholine, C16:0 (DPPC)
which is the main shell component of Definity (Lantheus
Medical Imaging, North Billerica, MA, USA) or mainly
1,2-distearoyl-sn-glycero-3-phosphocholine, C18:0 (DSPC)
which is the main lipid component of SonoVue, Lumason, and
BR14 (Bracco Imaging S.p.A., Milan, Italy) [10], [21]–[25].
It was shown that the lipid distribution was more
homogeneous for DPPC-based microbubbles than for DSPC-
based microbubbles and that the binding area for DPPC-based
microbubbles was significantly larger than for DSPC-based
microbubbles [21]. We previously determined the acoustic
properties of these DPPC and DSPC-based microbubbles in a
setup where the microbubbles were floating against an Opti-
Cell wall (nonbound) [10] and hypothesized that the difference
in ligand distribution and binding area could alter the acoustic
response after adherence of the microbubble to its molecular
target. In this paper, we investigated the vibrational response
of in-house produced targeted DPPC-based and DSPC-based
microbubbles using the Brandaris 128 ultrahigh-speed optical
camera [26] when they had bound to a streptavidin-coated
hydrogel and compared their responses to those of nonbound
microbubbles floating against the hydrogel. We aimed to
identify differences in vibrational responses that may be used
to discriminate bound from nonbound microbubbles.
II. M
ATERIALS AND METHODS
A. Microbubble Preparation
Biotinylated lipid-coated microbubbles with a C
4
F
10
gas core (F2 Chemicals Ltd, Preston, UK) were made
as previously described [21], [27] by sonication for
1 min. The coating was composed of 59.4 mol% DSPC
(P6517, Sigma-Aldrich, Zwijndrecht, The Netherlands) or
DPPC (850355, Avanti Polar Lipids, Alabaster, AL, USA),
35.7 mol% polyoxyethylene-40-stearate (PEG-40 stearate,
P3440, Sigma-Aldrich), 4.1 mol% 1,2-distearoyl-sn-glycero-3-
phosphoethanolamine-N-[carboxy(polyethylene glycol)-2000]
(DSPE-PEG(2000), 880125, Avanti Polar Lipids); and
0.8 mol% 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-
N-[biotinyl(polyethylene glycol)-2000] [DSPE-PEG(2000)-
biotin, 880129, Avanti Polar Lipids].
A 25-μm-thick polyester membrane was mounted on a
custom-made rectangular polyvinylchloride holder (same size
as a microscope objective glass) and was custom-coated with a
1–2-μm-thick polycarboxylate hydrogel (XanTec bioanalytics
GmbH, Düsseldorf, Germany). For the bound targeted
microbubbles, the hydrogel was activated and strepta-
vidin (S4762, Sigma-Aldrich) was subsequently covalently
attached to the hydrogel, using the amine coupling kit
(K AN-50, XanTec bioanalytics GmbH) according to the
instructions of the manufacturer, as previously described [21].
Briefly, streptavidin was dissolved in acetate buffer (2 mM,
pH 5.4) (1 mg/mL). After desalting the streptavidin by use
of a PD-10 desalting column (GE Healthcare Bio-Sciences),
the concentration was determined spectrophotometrically at
570 nm using a Pierce BCA Protein Assay kit (Thermo
Scientific) and Thermo Multiskan EX. Three polyester mem-
branes were placed in a 5-Slidemailer (Heathrow Scientific,
Northgate, U.K.) with 18 mL of 1 M NaCl + 0.1 M NaB
(pH 10) elution buffer (K AN-50, XanTec bioanalytics
GmbH), followed by an incubation with 18 mL of 1.6% (w/v)
EDC · HCL (K AN-50, XanTec bioanalytics GmbH) in
activation NHS/MES buffer (K AN-50, XanTec bioanalytics
GmbH),and18mLof33μg/mL desalted streptavidin
in 2 mM acetate buffer at pH 5.2–5.4. Finally, 18 mL
of 1 M ethanolamine hydrochloride (pH 8.5) quenching
buffer (K AN-50, XanTec bioanalytics GmbH) was used
to terminate the reaction. The targeted microbubbles were
allowed to adhere to the streptavidin-coated membrane in
air-equilibrated phosphate buffered saline (PBS) containing
calcium and magnesium (DPBS, 14080, Invitrogen, Thermo
Fischer Scientific, Landsmeer, The Netherlands) by flotation
for 5 min. Then, the membrane was gently washed three times
with air-equilibrated PBS containing calcium and magnesium
using a 3 mL plastic Pasteur pipette. For the nonbound targeted
microbubbles the hydrogel was treated in the same way, except
for the addition of streptavidin. The targeted microbubbles
were added below the hydrogel-coated polyester membrane of
the custom-made holder and floated up due to buoyancy. The
hydrogels with the nonbound targeted microbubbles and bound
targeted microbubbles were orientated in the setup as shown
in Fig. 1.
B. Microbubble Spectroscopy
The vibrational responses of the bound and
nonbound targeted microbubbles were captured using
the Brandaris 128 ultrahigh-speed camera operated at
∼15 million frames/s [26]. Single microbubbles were
investigated in Region Of Interest (ROI) mode [28] using
the microbubble spectroscopy technique [6] in combination
with the exact same setup as in [10], except for a higher
magnification microscope objective (60×,NA = 0.9,
VAN ROOIJ et al.: VIBRATIONAL RESPONSES OF BOUND AND NONBOUND TARGETED LIPID-COATED SINGLE MICROBUBBLES 787
Fig. 1. Configuration and composition of nonbound targeted microbub-
bles (top) floating against a hydrogel and targeted microbubbles bound to this
hydrogel via streptavidin (bottom) in the experimental setup (not to scale).
Olympus, Tokyo, Japan). Briefly, a broadband single element
polyvinyl difluoride (PVDF) transducer (25-mm focal
distance, f-number 1.1, center frequency 5 MHz, PA275,
Precision Acoustics Ltd, Dorchester, UK) transmitted a
Gaussian tapered eight-cycle sine wave burst at transmit
frequencies swept from 1 to 4 MHz (increment steps
of 200 kHz) at a peak-negative pressure (P
A
) of 50 or
150 kPa at the focus. The pressures were calibrated with
two calibrated PVDF needle hydrophones in a separate
measurement beforehand (0.2-mm diameter PA2030 and
1-mm diameter PA1875, Precision Acoustics). The optic
focus was aligned with the acoustic focus, to ensure that the
microbubble received the intended pressure. The ultrasound
was triggered on the second recording of each microbubble
to obtain the initial resting diameter and the noise level with
our contour tracking algorithm in the first recording. The
experiments were conducted at room temperature and the
sample was submersed in air-equilibrated PBS containing
calcium and magnesium. All microbubbles were exposed
to ultrasound within 2 h after addition to the custom-made
holder.
C. Data Analysis
Diameter-time (D-t) curves were obtained using custom-
designed image analysis software [6] that determines the
vibrational responses as described elsewhere [10]. Briefly,
the acoustic stability of the microbubbles was quantified as
the difference between the mean diameter of the microbubble
in the initial D-t curve (D
0
) and the final D-t curve (D
end
).
Next, the asymmetry of the D-t curves was measured as
the ratio E/C between the relative expansion E,definedas
(D
max
–D
0
)/D
0
, and the relative compression C,defined
as (D
0
–D
min
)/D
0
, of the microbubble. Where D
max
is the
maximum diameter, D
min
the minimum diameter in the
D-t curve, and D
0
the resting diameter before vibration.
The E/C ratios were used to classify the asymmetry as:
1) compression-only behavior (E/C < 0.5); 2) normal
excursion (0.5 ≤ E/C ≤ 2); or 3) expansion-only behavior
(E/C > 2) [29].
Using the fast Fourier transformation (FFT) the frequency
content of the D-t curves was analyzed in terms of the
amplitude at the transmit frequency ( f
T
). These amplitudes
were fit to a resonance curve of a linear oscillator by a least-
mean-squares method [6], [10] to determine the resonance
frequency ( f
res
) of the microbubble. F
res
was usually located
in between two insonifications. The microbubble diameters at
these insonifications are known, and the diameter at resonance
D
res
was determined from an interpolation between these
two insonifications. The maximum relative radial excursions
(i.e., at f
res
) were defined as the maximum amplitude of the
FFT divided by the corresponding resting diameter of the
microbubble [10]. The same approach was used to determine
the subharmonic resonance frequencies ( f
sub
) and the sec-
ond harmonic resonance frequencies, and the corresponding
maximum relative radial excursions. Next, the maximum
relative radial excursions were transformed into pressures
using [9], [10]
P
S
=−
ρω
2
res
R
2
res
ε
d
(1)
where P
S
is the scattered pressure at a distance d from
the microbubble, ρ = 1 · 10
3
kg/m
3
is the density of the
surrounding fluid (PBS), ω
res
= 2π f
res
the angular resonance
frequency, R
res
the corresponding radius, and ε is the maxi-
mum relative radial excursion amplitude. All calculations were
performed in MATLAB (The MathWorks Inc., Natick, MA,
USA).
D. Statistics
Shapiro–Wilk tests for normality showed that the data was
not normally distributed, so we used nonparametric testing. For
comparing the acoustic stability of the microbubbles we used
Wilcoxon signed-rank tests. When comparing groups, e.g.,
bound DSPC and nonbound DSPC, we used Mann-Whitney
U tests. Medians and interquartile ranges (IQRs) are reported
and were calculated using Tukey’s Hinges method. Statistical
analyses were performed using SPSS (Statistics 21, IBM
Corporation, Armonk, NY, USA) and a p-value <0.05 was
regarded as significant.
III. R
ESULTS
In total, 143 single microbubbles having a D
0
between
1.5 and 10 μm were analyzed. At 50 kPa, 46 bound DPPC
microbubbles were insonified; 18 of which were also insoni-
fied at 150 kPa. For bound DSPC microbubbles, 43 were
788 IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 64, NO. 5, MAY 2017
Fig. 2. Still frames of the Brandaris ultrahigh-speed recordings of a bound DPPC-based microbubble insonified at pressures of 50 and 150 kPa
(see Supplemental Material for the movies). The initial state is indicated with “no US” and D
0
is determined from this recording. At a frequency of 1 MHz,
two frames have been selected: one in the expansion phase and one in the compression phase of the oscillation, as also indicated by the black dots in the
D-t curves. The curve at a pressure of 150 kPa shows inertial cavitation and thus asymmetric behavior. Acoustic deflation is clearly visible when comparing
D
end
, determined from the 4-MHz recordings, with D
0
.
insonified at 50 kPa; 15 of which were also insonified at
150 kPa. None of the bound microbubbles detached during
the experiments since every microbubble remained within
the optic focus. For the nonbound microbubbles we included
26 DPPC and 28 DSPC microbubbles, which were all insoni-
fied at both 50 and 150 kPa.
A. Acoustic Stability
Fig. 2 shows an example of a bound DPPC-based microbub-
ble insonified at a pressure of 50 kPa and subsequently
at 150 kPa. Comparing D
0
with D
end
at both pressures shows
a clear decrease in diameter. The corresponding ultrahigh-
speed Brandaris recordings can be found in the Supplementary
Material, as well those of a bound DSPC-based microbubble.
Overall, at P
A
= 50 kPa, both bound DPPC and DSPC-
based microbubbles deflated significantly more than when
they were nonbound (p = 0.0001), as shown in Fig. 3.
The median size for the bound DPPC microbubbles after
insonification was 83% of D
0
, while this was 98% for the non-
bound DPPC microbubbles. The median size of bound DSPC
microbubbles was 93% of D
0
after insonification, whereas
nonbound DSPC microbubbles maintained their original size
(100% of D
0
). At a pressure of 150 kPa, the size difference
between bound DPPC and nonbound DPPC microbubbles
was not significant. The median diameter after insonification
decreased to 53% of D
0
for bound DPPC microbubbles and
to 56% for nonbound DPPC microbubbles. In case of DSPC
microbubbles, those that had bound deflated more than those
that had not (p = 0.004). For the DSPC microbubbles this
was 76% for the bound ones and hardly any shrinkage (98%
of their initial size) for the nonbound ones. In addition, for
both bound and nonbound microbubbles, those based on DPPC
deflated more than those based on DSPC at 50 kPa (bound:
p = 0.001, nonbound: p = 0.031) and also at 150 kPa
(both p = 0.0001).
B. Linear Oscillation Behavior
The resonance frequencies in relation to D
res
are shown in
Fig. 4. First of all, at a pressure of 50 kPa the resonance
frequencies of bound DSPC microbubbles were similar to
those of nonbound DSPC microbubbles. For DPPC-based
microbubbles, the resonance frequencies of bound microbub-
bles were significantly higher than for nonbound DPPC
microbubbles (p = 0.045). To further highlight the differences
in resonance frequencies between bound and nonbound DPPC
microbubbles, we compared the resonance frequencies of those
having D
res
< 4 μm. For larger microbubbles all reso-
nance frequencies were similar, but for microbubbles having
a D
res
< 4 μm, the resonance frequencies of bound DPPC
microbubbles were significantly higher than for nonbound
DPPC microbubbles (p = 0.002). In addition, no overlap was
found between the median (IQR) resonance frequencies of
bound DPPC microbubbles and nonbound DPPC microbub-
bles (Table I). In contrast, the resonance frequencies of bound
and nonbound DSPC microbubbles were similar for all studied
sizes (p = 0.494). The resonance frequencies of bound DSPC
microbubbles were significantly higher than those of bound
DPPC-based microbubbles at P
A
= 50 kPa (p = 0.001), for
the nonbound DSPC and DPPC microbubbles no difference
was found. All resonance frequencies at a pressure of 150 kPa
were similar. The number of microbubbles included in Figs.
4 and 5 is lower than the total number of studied microbub-
bles, since some resonance peaks were below or above the
measuring range (<1or>4 MHz); the resonance frequency
could therefore not be determined.
For bound DPPC microbubbles, the maximum relative radial
excursions at a pressure of 50 kPa were significantly higher
than for the nonbound DPPC microbubbles (p = 0.002, Fig. 5,
Table I). Although the maximum relative radial excursions
of bound DSPC microbubbles were not significantly different
from nonbound DSPC microbubbles (p = 0.157) over the
VAN ROOIJ et al.: VIBRATIONAL RESPONSES OF BOUND AND NONBOUND TARGETED LIPID-COATED SINGLE MICROBUBBLES 789
Fig. 3. Diameter change during ultrasound exposure expressed as D
0
/D
end
for bound DPPC (50 kPa: n = 46, 150 kPa: n = 18), nonbound DPPC (50 kPa:
n = 28, 150 kPa: n = 28), bound DSPC (50 kPa: n = 43, 150 kPa: n = 15),
and nonbound DSPC microbubbles (50 kPa: n = 26, 150 kPa: n = 26). The
filled black circles are outliers.
whole resonance frequency range, the maximum relative radial
excursions for bound DSPC microbubbles were significantly
higher for resonance frequencies >2MHz(p = 0.001).
In addition, the maximum relative radial excursions of bound
DSPC microbubbles were significantly lower than of bound
DPPC microbubbles (p = 0.0001), but similar for the non-
bound DSPC and DPPC microbubbles. At a driving pres-
sure of 150 kPa the maximum relative radial excursions
of bound and nonbound DPPC microbubbles were similar,
but significantly higher for bound DSPC than nonbound
DSPC microbubbles (p = 0.001). The maximum relative
radial excursions for bound DPPC and bound DSPC-based
microbubbles were similar (Fig. 5, Table I). For nonbound
Fig. 4. Resonance frequencies ( f
res
) of bound DPPC (filled red circles),
nonbound DPPC (red crosses), bound DSPC (blue open circles), and non-
bound DSPC (blue crosses) microbubbles plotted versus the diameter at reso-
nance (D
res
) at P
A
= 50 kPa (top panel) and P
A
= 150 kPa (bottom panel).
microbubbles the maximum relative radial excursions were
significantly higher (p = 0.03) for DPPC microbubbles than
for DSPC microbubbles.
C. Nonlinear Oscillation Behavior
The asymmetry of the radial excursions at each transmit
frequency was expressed as the ratio between the relative
expansion E and relative compression C.At50kPa,the
median of the radial excursions was compression-dominated
with 0.5 < E/C < 1 for bound targeted microbubbles
of both types (Fig. 6) at all frequencies. For the nonbound
microbubbles the oscillations were mostly symmetric, except
for the frequencies between 1 and 1.6 MHz for which the
radial excursions of DPPC microbubbles were compression-
dominated. At P
A
= 150 kPa the excursion behavior of
bound microbubbles at frequencies between 1 and 1.8 MHz
ranged from symmetric to expansion-dominated, whereas at
higher frequencies the behavior of both microbubble types was
compression-dominated. An example of a bound DPPC-based
microbubble showing asymmetric oscillations at 150 kPa due
