2.1.

Lipid Materials

The lipids 1,2-dibehenoyl-sn-glycero-3-phosphocholine, 1,2-dipalmitoyl-sn-glycero-3-phosphate, and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine were obtained from Corden Pharma (Switzerland), and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(poly(ethylene glycol))-2000] (ammonium salt) was obtained from Laysan Lipids (Arab, Alabama). All materials were used as received. When not in use, lipids were stored at $-20°\mathrm{C}$.

2.2.

Lipid Solution Preparation

NB samples were produced in-house. The base protocol for the production of shell-stabilized NBs used for the current work was reported elsewhere.³⁵ The protocol with the addition of SB differs minimally from the published work. For the synthesis of the shell solution the lipids (see Supplementary Material) were dissolved in glycerol (Acros Organics, Morris Plains, New Jersey) under heat at 80°C for 30 min. SB dye (Sigma Aldrich, Cleveland, Ohio) was measured in volumes of 0, 1, 2, 3, 4, and 5 mg and added to dissolved volumes for lipids measured to produce a 10-ml batch. Once SB was fully dissolved in lipids under heat without visible precipitates, phosphate buffered saline (PBS) and glycerol was added to the solution and sonicated for a final 10 ml yield of SB + lipid solution. The 10 ml yield was aliquoted into ten 1-ml aliquots in 1.5-ml vials and sealed by a rubber cap. The final SB concentrations in the vials were 0, 0.1, 0.2, 0.3, 0.4, and 0.5 mg SB per 1-ml lipid solution.

2.3.

NB Preparation

To prepare the vials for activation, air was removed from each vial manually by needle and syringe. 10 ml of octafluoropropane (C3F8) (Synquest Laboratories, Alachua, Florida) was then injected into the vial while a second needle was used for venting during the process. The NBs were formed by mechanical agitation for 45 s using a Vialmix mechanical shaker [Bristol Myers Squibb (BMS), New York City, New York]. The vial was then inverted, and differential centrifugation was used to separate the bubbles by size.³⁵ After centrifugation, the vial was kept inverted during the sample draw to limit mixing of activated bubbles and control for bubble size distribution. The penetration depth of the 18 G needle into the vial was limited to 5 mm for a consistent draw. $300\mu \mathrm{L}$ of activated bubbles were drawn from the bottom of each vial and was used as activated, stock bubble solution. To further narrow the agent size distribution, the resulting isolated NB solutions were filtered. Here, activated stock bubble solution was diluted in PBS in a 1:9 dilution and passed through a single-use polyethersulfone filter unit (FroggaBio, Toronto, Ontario, Canada) with 450-nm pore size. The syringe was depressed by hand at a rate of one droplet per 5 s and the refuse was collected for immediate use.

2.4.

Measurement of Bubble Concentration and Size

A 1:500 dilution with PBS (pH 7.4) of activated bubbles from stock bubble solution was prepared for each sample type in PBS and sized using the Archimedes RMM system (Malvern Panalytical, Inc). RMM has been established as a technique for counting and sizing of suspended buoyant and nonbuoyant particles³⁶ including NBs.³⁸ Using RMM, buoyant particle concentration and diameter were determined. A nanosensor that provides measurements from 100 nm to $2\mu \mathrm{m}$ was used to characterize the NBs. The nanosensor was previously calibrated with National Institute of Standards and Technology (NIST) traceable 565-nm polystyrene bead standards, (ThermoFisher 4010S, Waltham, Massachusetts). About 500 particles were measured for each trial performed ($n=3$). The sensor and microfluidic tubing were cleaned with deionized (DDI) water in between each run. Data were exported from the Archimedes software (version 1.2) and analyzed for positive and negative counts, which corresponded to buoyant (bubble) and nonbuoyant particles, respectively. A density of $0.008{\mathrm{g}·\mathrm{ml}}^{-1}$ for positively buoyant particles (density of $0.008\mathrm{g}/\mathrm{cc}$) and $1.3{\mathrm{g}·\mathrm{ml}}^{-1}$ for negatively buoyant particles (density of $1.34\mathrm{g}/\mathrm{cc}$) were used to convert the measured mass to a particle diameter. A sample distribution for the Archimedes system is shown in Fig. 2. Filtered bubbles were prepared in 1 in 10 dilutions in PBS due to lower bubble yield from the filtration process. This process was repeated three times for each sample type.

Fig. 2

Sample size distribution of NBs and nonbuoyant particles prepared from lipid solution without SB before pore filtration as measured using the Archimedes RMM system.

2.5.

Transmission Electron Microscopy (TEM)

TEM images of the bubbles were obtained using a FEI Tecnai G2 Spirit BioTWIN TEM operated at 120 kV based on a previously reported method.^39,40 A dilute suspension of the sample ($10\mu \mathrm{L}$) was placed upside down on a 400 mesh Formvar-coated copper grid. The sample was then stained with 2% uranyl acetate, after which the sample was allowed to dry for 30 min.

2.6.

Surface Tension

A CAM 200 optical goniometer by KSV Instrument, Ltd. with a pendant drop tensiometry setup was used to measure the surface tension of the various sample types at a lipid to air interface. A flat tipped needle (0.48-mm O.D.) was used. To obtain the most accurate results,⁴¹ samples were extruded from the needle to hold the largest possible, stable droplet. Water was used for calibration of the system. Measurements of water and lipid droplets were collected at 22.3 +/−0.3°C.

2.7.

Acoustic Stability

To characterize changes in bubble echogenicity over time under constant insonation and in a more clinically relevant setting, grayscale intensity changes generated by the SBNBs were measured in vitro using a linear transducer (Toshiba, Tochigi-Ken, Japan) and a clinical US scanner (Toshiba Aplio) at 6-MHz transmit frequency, 12 MHz receiving frequency and a peak negative pressure of 240 kPa (mechanical index, 0.1). This data were collected using nonlinear contrast mode imaging. Phantoms were custom designed from agarose mold (1% agarose, $99\%{\mathrm{H}}_2\mathrm{O}$). Each phantom had three narrow channels (see Fig. 3). Bubbles were imaged using US for 8 consecutive minutes. The phantom channels were aligned to the center of the transducer element placed atop such that the transducer was in an inverted orientation.

Fig. 3

(a) Inverse bubble stability setup featuring a linear transducer (Toshiba, Tochigi-Ken, Japan, 6-MHz transmit, 12-MHz receive, 240 kPa peak negative pressure, M.I, 0.1), and 1% agarose mold with three narrow channels ($3\times 5\times 6\mathrm{mm}$) separated 4 mm apart; (b) US images from the inverted setup; Image of 1 in 100 diluted $0.5\mathrm{mg}/\mathrm{ml}$ SB B NBs in PBS; top of image corresponds to the top of the transducer, bottom of image corresponds to air-exposed top of channel; sample ROI indicated by yellow box; figure created with BioRender.com.

Activated bubble solution was prepared at constant dilution of stock solution (1 in 100). Six images were acquired for each sample. A ROI was selected within each channel region. The average US power of all non-zero elements in a selected ROI was tracked with respect to time for the duration of the 8 min excitation. This data was averaged for all six measurements and normalized to data from a noise region within the image.

2.8.

Single Bubble Ultrasound Scattering

The acoustic response of single Sudan Black nanobubble (SBNB) agents was studied using the Vevo 770 US system. The high frequency of the transducer paired with the adjustable pulse length was important for localization of signals from single agents. A focused RMV-710B transducer ($f$-number 2.1; aperture 7.14 mm; focal length 15 mm) with center frequency of 25 MHz and 100% bandwidth was used. This transducer has a mechanical scan head that oscillates laterally as it scans a fixed region of interest. Bubbles were stimulated with 30 cycle pulses at 25 MHz at varying pressures (0.2, 0.6, 1, 2, and 3 MPa). For each power setting, 25 data sets were collected. Each data set contained 100 RF lines where one RF line represents the measured backscatter data associated with the material along the axial path of the transducer.^42–44 This experiment was repeated for each of three different vials of the 0, 0.2, and 0.4 mg SB samples three times each (nine total).

Stock bubble solutions were filtered by pore filtration to minimize size-dependent effects in SBNB stimulation. After filtration, the bubble concentration was determined and used to prepare a dilution with a consistent number of NBs per volume in DI for each measurement ($\sim 2000\mathrm{NBs}/\mathrm{ml}$). The diluted bubble solution was gently stirred to disperse the bubbles. The transducer head was submerged into the beaker and held in place 3 cm below the surface of the water. A new dilution was prepared for each measurement to ensure high signal detection that was concentration independent. A sample image of the raw data from the system is available in the Supplementary Material.

Analysis of the RF data was done using MATLAB R2019b. RF data were converted from time to frequency domain to yield the power spectra of the data. The data was then subject to a set of criteria for identification as single bubble signals. RF lines lacking bubbles, truncated RF lines, or lines with signals from more than one bubble were removed. The remaining accepted signals were finally classified as either linear, nonlinear, or transient. A linear oscillation has a minimum and maximum radius that is consistent throughout the US excitation. Such signals were labeled p1. In some cases, linear signals occurred such that a steady increase or decrease in amplitude with cycle number was detected. Such signals were labeled as ${\mathrm{p}1}_{\text{growth}}$ or ${\mathrm{p}1}_{\text{diss}}$ respectively.

Nonlinear signals were identified through consistent periodicity, which was identified through patterns that occurred in the time domain signal as a function of cycle number (every two peaks; every three peaks; and every four peaks) and thus were labeled p2, p3, p4, p5, p6, or p7 depending on the pattern observed (see Supplementary Material).^35,44 The most commonly occurring repetitive sequence in the RF line was used to classify the signal behavior. For some signals, no clear pattern in oscillation was visible, however the oscillations were nonlinear. These signals have been denoted ${\mathrm{p}}_{\mathrm{x}}$. To account for the low numbers of nonlinear signals detected overall, the nonlinear signals were all grouped together under the common classification, ${\mathrm{p}}_{\mathrm{n}}$

Destructive signals or cavitation signals were those such that the signal demonstrated a bubble that does not complete all 30 oscillations and power spectra with visibly broadband shapes in comparison to other signals.

Classified signals were examined to track the total number of signals that are classified into the five groups: (1) linear, (2) linear and increasing in amplitude (growth), (3) linear and decreasing in amplitude (dissolution), (4) nonlinear, and (5) cavitation signals.

2.9.

Multimodal PA-US Imaging

PA and US multimodal imaging was done using the Vevo LAZR 2100 system with LZ250 transducer operating at 21 MHz central frequency (13-24 MHz bandwidth). The transducer is a 256-element linear array transducer coupled with a laser ($ƛ=680–970\mathrm{nm}$) system. The transducer and optical illumination share a common focus at 11-mm imaging depth.⁴⁵ US and PA RF data were acquired simultaneously when acquiring PA images. US images were collected at 1% transducer power (peak negative pressure 0.922 MPa) and PA images at 100% laser power ($\text{max fluence}=20[{\mathrm{mJ}/\mathrm{cm}}^2]$) at an excitation wavelength of 700 nm. About 25 frames of US and PA images were acquired and were subsequently analyzed using MATLAB.

About 10 wt%, 10 kPa polyacrylamide phantoms containing six 1-mm diameter vessels were prepared the day of imaging using degassed, DDI water. The DDI water was degassed using a SRDS-1000 (FUS Instruments, Toronto) water degassing system. For each phantom, the polyacrylamide solution was prepared and poured into a $2\mathrm{cm}\times 2\mathrm{cm}$ holder with six parallel fire-polished borosilicate vessels (1-mm O.D). The borosilicate vessels were carefully removed after polymerization leaving 6 hollow channels and the phantom was removed from the holder for temporary storage. In storage, phantoms were hydrated in PBS in a beaker at room temperature.

Phantom channels were filled with activated solutions of SBNBs each at a 1 in 30 dilution. The phantom was replaced in the phantom holder and the channels were covered at the ends using glass slides to retain the contents. The sealed phantom-in-holder was placed in a DDI water bath for imaging. The centers of the vessels were approximately aligned at the laser-transducer focus (schematic in Fig. 4). Each phantom was imaged at three different cross sections along the channel length. This process was repeated three times, yielding a total of nine cross sections.

Fig. 4

Submerged multimodal US/PA experimental setup; a 10% polyacrylamide phantom with 6 hollow channels (1-mm diam; 1-mm separation) housed in a clear holder to block vessel ends; Vevo LAZR 2100 system with 256-element linear array LZ250 transducer (21 MHz central frequency; 13 to 24 MHz bandwidth) and common US/PA focus at 11-mm imaging depth. Figure created with BioRender.com.

US/PA data from the six contrast agents were analyzed using MATLAB R2019a. The laser energy was recorded for each frame and used to normalize the PA RF-data before further analysis. To analyze the intensity differences for each vessel, a mask was created of circles with centers corresponding to the centers of the vessels and diameters matching vessel diameter (as determined from the US image). The mask was multiplied by the log-compressed RF-matrix to isolate the signal from each vessel. The average signal from a vessel was determined by taking the mean signal within a single vessel region. This was then averaged for all 25 frames and all nine cross sections of the same vessel type. The standard deviation of the mean vessel signal of the nine cross sections was determined.

4.1.

Physical Property Characterization

TEM images of 0 [Fig. 5(a)], 0.3 mg [Fig. 5(b)], and 0.5 mg [Fig. 5(c)]. The dark ridges on the bubble surface may be shell buckling which is thought to play an important role in bubble dynamics.⁴⁰ Also present in the images are flecks (indicated by arrows) which are attributed to nonbuoyant lipid particulates.

Fig. 5

Representative transmission electron microscopy images for lipid-shelled NBs with differing SB B content (a) 0 mg; (b) 0.3; and (c) $0.5\mathrm{mg}/\mathrm{ml}$ in the lipid shell solution.

The average buoyant particle concentration and diameter of three measurements are shown for each activated SBNB formulation in Figs. 6(a) and 6(b). From a one-way ANOVA test differences between bubble yields for various SBNB formulations were determined to be not significant (ns). However, from two-tailed t-tests comparing $0.3\mathrm{mg}/\mathrm{ml}$ and $0.4\mathrm{mg}/\mathrm{ml}$ SB to control the concentration was reduced with statistical significance. Differences between the average bubble diameter for varying bubble formulations were determined to be not significant. Overall, we produced high agent yield ($\sim {10}^{11}$) and consistent agent size (range: $\sim 120$ to 700 nm; mean $\sim 200$ to 300 nm).

Fig. 6

(a) SBNB concentration and (b) diameter from RMM ($n=3$) differences between the average bubble diameter for varying bubble formulations were also determined to be not significant.

Successful formulation of safe, stable and, requires the repeatable preparation of homogenous (monodisperse) populations.⁴⁶ The term polydispersity index (PDI) is used to describe the degree of non-uniformity of a size distribution of particles. The numerical value of PDI ranges from 0.0 (a perfectly uniform sample) to 1 (a highly polydisperse sample). This statistic is provided upon measurement from the RMM system and compared here. The PDI was averaged for all samples (See Fig. S4 in the Supplementary Material). The results show the addition of SB B die to the stabilizing lipid shell does not significantly alter the PDI of the population, although there is a slight increasing trend. This was statistically validated by a one-way analysis of variance test ($n=3$).

4.2.

Surface Tension

Samples of the raw data from the pendant drop tensiometry measurements are shown in Fig. 7. The results of the surface tension measurements determined by pendant drop tensiometry are shown in Fig. 8. A multiple comparison test of the mean surface tension of the droplets was performed to identify groups of significance. As shown by the horizontal bars in Fig. 8, all lipid formulations with SB dye were determined to be statistically different (*) from the 0 mg SB formulation. Each formulation type had statistical significance when compared to the formulation with 0.1-mg difference in SB solution concentration. However, the difference between the 0.3-mg lipid and 0.5-mg lipid was not significant (ns). Water droplets were not included in the significance tests. Compared to water droplets, the lipid solutions have a reduced surface tension.

Fig. 7

Raw data for pendant drop tensiometry. Profiles of pendant drops of (a) PGG solution and PGG solution with (b) 0.1; (c) 0.2; (d) 0.3; (e) 0.4; (f) $0.5\mathrm{mg}/\mathrm{ml}$ added SB B, respectively, for lipid-shell stabilized NBs; ($\text{scale bar}=0.5251\mathrm{mm}$).

Fig. 8

Surface tension as measured from bulk droplets of lipid solution containing varying contributions of SB B dissolved in the lipid. ($n=20$; bars represent standard deviation).

4.3.

Ultrasound Stability

Power decay curves from 8 minutes of constant US stimulation are shown in Fig. 9. These power decay curves show a decrease in %-maximum US power with respect to time until 2 min is reached. After the 2-min time point, the 0.3, 0.4, and 0.5 mg SBNB power curves increase. It is possible that this increase is a result of bubble growth and bubble joining. If only the first 2 min of the decay curve are considered, an exponential fit can be used to determine the rate of the population decay in that time window. Exponential fits to un-averaged curves were determined to obtain decay rates for the first 2 min. These decay rates were obtained for all six measurements and subsequently averaged (Fig. 9). With increasing SB contribution an increase in power decay rate is expressed with a sharp increase detected between the 0.2 and $0.3\mathrm{mg}/\mathrm{ml}$ SBNBs and the maximum decay rate detected from the $0.5\mathrm{mg}/\mathrm{ml}$ SBNBs.

Fig. 9

(a) US power decay normalized to maximum power of time series (Toshiba, Tochigi-Ken, Japan, 6-MHz transmit, 12 MHz receive, 240 kPa peak negative pressure, M.I, 0.1) for 1 in 100 dilution NBs in PBS with differing concentrations of SB B in the lipid shell. (b) Bubble decay rates after 8 min continuous US stimulation for lipid-stabilized with varying concentrations of SB B dye in the lipid shell.

4.4.

Single-Bubble Excitations

The total counts of recorded signals above threshold from a single bubble with respect to acoustic pressure were recorded [Figs. 10(a)–10(c)]. The linear and nonlinear signal counts were compared. To identify the types of signals that most significantly contributed to the total number of signals detected by a particular formulation, the average percent of signals above threshold for each signal type was plotted and compared [Figs. 10(d)–10(f)].

Fig. 10

(a)–(c) Single bubble signal count as a function of acoustic pressure. ($n=3$) and SB contribution in lipid-shell. (d)–(f) Average contribution of signal types to total signal count summed over all pressures ($n=3$); average total number of signals = 511 ($0\mathrm{mg}/\mathrm{ml}$), 261 ($0.2\mathrm{mg}/\mathrm{ml}$), and 246 ($0.4\mathrm{mg}/\mathrm{ml}$).

The lowest bubble activity was recorded at the minimum pressure used in the experiments (0.2 MPa) for all three formulations. Here, we use bubble activity to refer to the number of counts registered above threshold. At this lower pressure, the 0-mg SBNB formulation generated the most counts. The bubble behavior by all formulations at 0.2 MPa was dominated by p1 signals. No nonlinear signals were detected at 0.2 MPa for any of the NB formulations. As pressure of the incident acoustic wave increased, an increase in the number of linear and nonlinear counts was recorded.

At 0.6 MPa, the number of ${\mathrm{p}1}_{\text{growth}}$ signals increased, reaching a maximum at 0.6 MPa for all NB formulations, and decreased with increasing pressure beyond 0.6 MPa. Additionally, as the SB concentration increased, the number of ${\mathrm{p}1}_{\text{growth}}$ signals detected increased (red line) and dominated the number of signals detected. For this pressure, more p1 signals (blue line) were recorded on average than ${\mathrm{p}1}_{\text{growth}}$ signals (red line) for the $0\mathrm{mg}/\mathrm{ml}$ NBs. In the case of the $0.2\mathrm{mg}/\mathrm{ml}$ formulation, p1 and ${\mathrm{p}1}_{\text{growth}}$ signal counts were approximately equal. For the $0.4\mathrm{mg}/\mathrm{ml}$ formulation, however, the counts of ${\mathrm{p}1}_{\text{growth}}$ signals doubled p1 counts.

Beyond 0.6 MPa, the counts of linear signals (p1, p1growth, and p1diss) decreased. A peak in nonlinear signals occurred at 1 MPa for all NB formulations (purple line). A steady increase in bubble cavitation started at 1 MPa and peaked at 3 MPa (green line). No bubble cavitation events were detected before 1 MPa for any of the NB formulations. Due to the low number of signals registered above the threshold that were nonlinear for all NB types, all nonlinear signals were grouped together. Differences in the different types of nonlinear oscillations (p2 and p3) may have gone undetected due to the low overall detection of nonlinear signals.

From average contribution of signal types to total signal count [Figs. 10(d)–10(f)] no apparent trend as a function of SB dye concentration is observed for p1 signals (blue wedge). However, upon the addition of SB dye, the relative number of ${\mathrm{p}1}_{\text{growth}}$ signals increases (red wedge). On average, the ${\mathrm{p}1}_{\text{growth}}$ signals contribute approximately twice as much to the total signal count of $0.2\mathrm{mg}/\mathrm{ml}$ SBNBs compared to the NBs without dye. Similarly, the ${\mathrm{p}1}_{\text{growth}}$ signals contributed three times more to the total $0.4\mathrm{mg}/\mathrm{ml}$ SBNBs than the $0\mathrm{mg}/\mathrm{ml}$ NBs. No apparent trend with increasing dye concentration is observed for ${\mathrm{p}1}_{\text{diss}}$ oscillations (yellow wedge).

The number of the various $p_n$ signals were low (see Supplementary Material). However, the contribution of all ${\mathrm{p}}_{\mathrm{n}}$ signals to the total number of signals detected from the NB formulation decreased with increasing SB concentration tending toward statistical significance ($p<0.1$) (purple wedge). Additionally, the total contribution of cavitation signals decreased as a function of SB concentration (green wedge). Cavitation signal counts and total signal counts varied greatly across samples and therefore this trend was not established with statistical significance.

4.5.

Multimodal Imaging

Multimodal US/PA images were collected using the Vevo LAZR system. A representative image of the six-channel filled SBNBs-PBS dilutions, in order of increasing SB concentration from left to right, is shown in Fig. 11. The top and bottom images are from PA and US imaging, respectively, collected from the same cross section of the phantom channel.

Fig. 11

Representative PA(top)/US (bottom) images with channels containing 1:29 dilution of 0 mg, 0.1 mg, 0.2 mg, 0.3 mg, 0.4 mg, 0.5 mg Sudan Black from left to right; average PA (center; $n=9$) and US (right; $n=9$) signal from SBNBs.

In the PA image, six vessel cross sections are identifiable with increasing PA signal amplitude from left to right. The increase in signal corresponds to the increase in dye contribution. The first vessel (left-most), containing undyed NBs generated little detectable signal due to the low absorption profile of lipids/PFC. The five right-most channels containing SB are most clearly delineated. In the US image, by comparison, all six channels are clearly delineated. High amplitude signal at the top of each vessel cross section in the US images corresponds to differences in acoustic impedance at the phantom-PBS interface at the channel walls. Differences in US backscatter between the bubble populations in the channels were minimal (Fig. 11–bottom left).

An analysis of the average PA and US power is summarized in Fig. 11 (center and right). Horizontal lines marked by a star (*) are used to indicate groups that are statistically different from one another, as determined from one-way ANOVA and multiple comparisons tests. The computed average PA power from the vessel cross sections (Fig. 11–center) shows that an increase in average PA power occurs with increasing SB solution concentration. The average PA power of the undyed bubble formulation was lower compared to all dyed NBs. Additionally, the average PA power of the 0.5-mg bubble formulation was determined to be different from the other NB formulations. The groups with SB dye ranging from 0.1 to 0.4 mg SB integration were determined not to be statistically different from one another.

For the concentration of $0.1\text{-}\mathrm{mg}/\mathrm{ml}$ SB, there was a 4 dB increase in power in comparison to the control ($0\mathrm{mg}/\mathrm{ml}$). The SBNBs with the highest contrast ($0.5\mathrm{mg}/\mathrm{ml}$) was on average 8-dB higher than the control. Undyed agents produced low PA signals. For dyed SBNBs the PA signal increase with SB concentration was not linear. A plateau in signal is seen in average PA signal plots. The computed average US power from the vessel cross sections (Fig. 11- right) showed a trend of increasing power with increasing SB. A test of significance indicated a statistical difference between 0.1, 0.2, and $0.3\mathrm{mg}/\mathrm{ml}$ SBNBs in a pairwise comparison with undyed NBs. Statistical difference between the average US power of 0.5 mg SBNBs and 0.4-mg SBNBs was also established. However, the groups with SB dye ranging from 0.1 to 0.4 mg SB integration did not show statistically significant differences.