Ultrasound induces similar temporal endothelial expression patterns of eNOS and KLF2 as normal flow

Abstract

Objective:

To determine the sensitivity of vascular endothelial cells to long durations of low-intensity pulsed ultrasound (LIPUS) compared to normal flow and identify the duration that maximizes expression of two mechanosensitive genes related to healthy endothelial function, endothelial nitric oxide synthase (eNOS) and Krüppel-like factor 2 (KLF2).

Methods:

Custom ultrasound exposure tanks were developed and the acoustic field was characterized. Human umbilical vein endothelial cells were seeded into culture plates and exposed to LIPUS at a frequency of 1 MHz and acoustic pressure of 217 kPa for 20 min, 1 h, 6 h, 9 h, or 24 h. As a comparator, other cells were exposed to normal flow. RT-qPCR was used to assess mRNA expression of eNOS and KLF2.

Results:

Maximum eNOS and KLF2 expression occurred at 6 h and was localized to the beam path. Both genes exhibited qualitatively similar patterns of expression under LIPUS compared to normal flow. LIPUS induced a more rapid beneficial response compared to normal flow, but flow induced higher expression of both genes. eNOS expression after 6 h of LIPUS was dependent on RNA yield and culture duration prior to experiments.

Conclusion:

Endothelial cells exposed to longer durations of LIPUS than typically employed exhibited greater expression of beneficial genes. The temporal gene expression patterns resulting from LIPUS and normal flow suggest activation of similar signaling pathways. However, LIPUS also caused increased RNA yield that may be linked to proliferation, which would suggest more of a wound healing than atheroprotective phenotype.

Introduction

The arterial mechanical environment is sensed by vascular endothelial cells and is a key driver of most endothelial functions. The mechanical effects of blood flow are particularly important. Unidirectional flow at a normal magnitude (i.e., normal flow) promotes anticoagulation properties, the barrier function, regulation of vascular tone, and anti-inflammatory properties [1]. It also protects the artery from atherosclerosis [2]—a chronic inflammatory disease characterized by the accumulation of lipids and immune cells in the inner lining of arteries to form plaques that underlie heart attack and stroke. In contrast, so-called disturbed blood flow disrupts these primary endothelial functions and promotes the development of atherosclerosis [3-5]. Our recent study further demonstrated that restoring blood flow from disturbed to normal in the atherosclerotic arteries of mice promoted plaque stabilization [6], which strongly suggests that activating the same beneficial mechanosensitive signaling within the endothelium as normal flow can be therapeutic. A remaining challenge is identifying a mechanism to deliver beneficial mechanical stimuli to atherosclerotic arteries.

Ultrasound has been used to noninvasively deliver tunable mechanical stimuli into the body in a targeted manner for many therapeutic applications, including drug delivery, thrombolysis in treating ischemic stroke, and lithotripsy to disintegrate kidney stones [7]. In addition, the direct mechanical effects of low-intensity pulsed ultrasound (LIPUS) on cells have been shown to induce beneficial biological effects. Much of this work has focused on bone cells to accelerate fracture healing [8]. There are only a few studies on endothelial cells and most of these have investigated the use of LIPUS to induce angiogenesis. Towards this end, in vitro studies have shown that LIPUS promotes endothelial cell migration [9], expression of proangiogenic mediators such as vascular endothelial growth factor [10, 11], and formation of capillary-like structures [9, 12]. In vivo studies have demonstrated that treating the post-infarcted myocardium with LIPUS improves outcomes such as survival rate, left ventricular ejection fraction, and infarct size in mice [10] and pigs [13]. An interesting feature of these studies is that endothelial cells are only exposed to LIPUS for 2-30 min (sometimes once per day over multiple days). This duration is much shorter than the 4-24+ h typically used in studies of endothelial cells under flow [14]. In fact, short durations of flow have been shown to promote dysfunctional endothelial behaviors [15, 16]. Thus, the duration of a mechanical stimulus is a central determinant of the endothelial response. The purpose of this study was to investigate whether endothelial cells exposed to LIPUS were similarly sensitive to duration as cells exposed to normal flow and whether longer durations promote greater expression of two well-established mechanosensitive and atheroprotective genes related to normal endothelial function, endothelial nitric oxide synthase (eNOS) and Krüppel-like factor 2 (KLF2).

Materials and Methods

Experimental setup

The entire experimental setup used to expose endothelial cells to LIPUS is depicted in Fig. 1. Two custom ultrasonic exposure tanks were designed and fabricated out of acrylic, each with internal dimensions of 128 x 146 x 176 mm. The acoustic sources were 1 MHz unfocused immersion transducers of 13 mm diameter (Evident, Waltham, MA, USA) mounted to the bottom of the tanks via through-holes and a water-tight seal created with gaskets. The transducers were driven by a function generator (SGD 1032X, Siglent, Solon, OH, USA) and radio-frequency (RF) power amplifier (A150, E&I, Rochester, NY, USA) at 55 dB gain. The amplified signal was split into four channels, each channel connected to an independent 1000-Ohm ceramic rheostat to allow tuning of the signal before reaching each of four transducers (two transducers per tank). Polystyrene 6-well cell culture plates were held within the tank by a mount that positioned the center of each culture well directly above the center of a transducer. The distance between the plate and the membrane of the transducers was adjustable via a series of mount peg holes placed along the sides of the tank. The cavity of the bioreactor was filled with sterile degassed distilled water up to the sides of the culture plate.

Figure 1.

Ultrasound bioreactor setup. (A) Complete setup that includes the function generator, amplifier, and rheostats for delivery of tunable power to the transducers and two ultrasound exposure tanks within the incubator. (B) Close up of the ultrasound exposure tank with transducers submerged in degassed water and a 6-well cell culture plate held by supports on the side of the tank at the surface of the water (a weight ensures the plate does not drift during the experiment). (C) An illustration of ultrasound waves propagating through the center portion of an endothelial cell layer within a culture well.

Acoustic field measurements

We used an acoustic field characterization tank with a motorized three-axis gantry system to perform spatial mapping of the peak acoustic pressure field generated by the 1 MHz unfocused transducers in water (Fig. 2A). A capsule hydrophone with sensor aperture of 400 μm (HGL-400, Onda, Sunnyvale, CA, USA) was secured in a custom mount at the top of the tank above the transducer. The transducers were driven by a pulser/receiver (JSR Ultrasonics DPR300, BYK, Pittsford, NY, USA). Using the X and Y positioning of the gantry, the center of the transducer pressure field was determined by finding the maximum signal amplitude. Axial mapping was conducted by varying the Z-position of the transducer relative to the hydrophone from 20 mm to 120 mm (0.5 mm resolution) to determine and characterize the near-field-to-far-field transition (Fig. 2B). From this map, we set the culture plate position from the membrane of the transducers to 73 mm. In-plane spatial scans of the pressure field at this axial distance from the transducer were taken in a 35 mm X 35 mm grid in 0.25 mm intervals (Fig. 2C).

Figure 2.

Acoustic field characterization tank. (A) A tank with 3-axis gantry system and controller that allows characterization of the 3-D acoustic field generated by an ultrasound transducer in water. Insert shows a close up of the hydrophone and a 1 MHz transducer within the tank. (B) Experimental profile of the acoustic pressure as a function of hydrophone distance from the transducer (i.e., z-position). The near field-to-far field transition and culture plate positon within the bioreactor are indicated. (C) Representative experimental spatial scan of the acoustic pressure using the 1 MHz transducer. The image shows that the bulk of the acoustic pressure is concentrated within a central area with a diameter of about 10 mm.

We also conducted acoustic field measurements within the wells of culture plates to directly evaluate the relationship between peak acoustic pressure and voltage input to the transducers. To do this, we placed the same capsule hydrophone within a custom 3D-printed holder made of VeroClear (Objet 500 Connex3 3D printer, Stratasys, Eden Prairie, MN) that was mounted onto a 6-well culture plate to position the hydrophone at the center of the well it was placed in (Fig. 3A). This assembly was then placed in each exposure tank at the same distance from the transducers as the cell-seeded plates (73 mm). Acoustic pressure versus input voltage to the transducers (after 55 dB gain from the amplifier) was measured over a relevant pressure range both with and without the bottom of the culture plate to evaluate the transmission of the ultrasound wave through the plate (Fig. 3B). The plastic culture plates caused a mean drop in signal amplitude of 11%.

Figure 3.

Setup for acoustic field characterization within the wells of a culture plate. (A) Complete setup including a custom 3-D printed holder that attaches to a culture plate and positions a capsule hydrophone at the center of one of four of the wells. Insert 1 (left) shows the sensing tip of the hydrophone in a well with the bottom removed to allow direct measure of the acoustic pressure. Insert 2 (right) shows a pinducer. (B) Acoustic pressure output as a function of input voltage to the transducer (after 55 dB gain from the amplifier) for the cases of (1) the bottom of the culture plate is removed to allow direct measurement of the acoustic pressure within the tank and (2) the plastic culture plate is intact. The intact culture plates cause a mean drop in acoustic pressure of 11%.

Temperature measurements

The temperature within the ultrasound exposure tank was monitored during experiments using a data acquisition system (Keithley DAQ6510, Tektronix, Beaverton, OR, USA) and four platinum resistance thermosensitive temperature detectors (Omega, Norwalk, CT, USA). One detector was placed within the water bath of the exposure tank to monitor the temperature continuously during each experiment. Additional temperature measurements were also taken within the wells of culture plates (without cells) at the target acoustic pressure to identify any differences from the exposure tank.

Cell culture

HUVECs pooled from 10 donors (ATCC, Manassas, Virginia, USA) were used for all experiments at passages 4 to 7. Cells were maintained per the supplier’s protocol with vascular cell basal medium, supplemented with endothelial cell growth kit-BBE (ATCC) and 0.1% Penicillin-Streptomycin-Amphotericin B Solution (ATCC) in a humidified incubator at 37°C and 5% CO₂. For experiments, cells were seeded at 30,000 cells/cm² into polystyrene 6-well culture plates coated with rat-tail type I collagen 18-24 h prior to the start. Each culture well was provided with 2.41 ml of medium to give a fluid height of 2.5 mm. Cells were visually inspected with a light microscope immediately before and after each experiment to ensure a healthy appearance.

LIPUS experimental protocol

Two hours prior to experiments, the ultrasound exposure tank was sterilized, assembled, filled with degassed distilled water (stored in a container within the incubator to ensure physiologic temperature), and placed in the incubator. Immediately prior to the start, a 1MHz ultrasound transducer was used to detect the acoustic pressure (calibrated using the capsule hydrophone) within each well of a culture plate (without cells) by placing it within a similar custom 3D-printed holder as the one used for the hydrophone calibrations. The holder-plate assembly was then placed in each exposure tank at the same distance from the transducers as the cell-seeded plate (73 mm), the output voltage of the function generator was set, and the resistance of the rheostat for each transducer was adjusted until the target acoustic pressure for that experiment was reached. Once set, the endothelial cell-seeded culture plates were placed within the exposure tank and visually inspected to ensure no air bubbles were present at the plate-water interface. A small weight was placed on top of the culture plate lid to prevent floating or other movement during the experiment. Cell layers in these plates were exposed to a 1 MHz sine wave that was pulsed at 1 Hz (500,000 cycles per burst period) to simulate the cardiac cycle. Control cell layers were seeded into wells of the same culture plate that did not receive direct ultrasound stimulation. Preliminary experiments were performed to identify the maximum acoustic pressure that could be applied to endothelial cells in the basal-to-apical direction without causing detachment (Fig. 4). A target pressure of 217 kPa was found. Experiments at this target pressure were conducted over five durations: 20 min, 1 h, 6 h, 9 h, and 24 h.

Figure 4.

Representative phase images of fixed endothelial cell layers at the center and mid-periphery of the wells of culture plates after exposure to low-intensity pulse ultrasound (LIPUS). Images were taken of cell layers after 6 h of no LIPUS (control; top), LIPUS at 217 kPa (middle), and LIPUS at 265 kPa (bottom). Cells exposed to LIPUS at 265 kPa exhibited detachment.

Flow experimental protocol

As a comparator, endothelial cells were exposed to normal flow within the wells of culture plates by using an orbital shaker (Grant instruments PSU-10i, Fisher, Waltham, MA, USA) that moved the plates in a horizontal circular orbit of 5 mm radius at 250 revolutions per minute (RPM) within an incubator. The outer periphery was subjected to unidirectional flow at a normal shear stress (mean of 1.4 Pa) and gene expression was evaluated in this region, following our previous work [1]. For this experiment, control cell layers were exposed to static conditions by seeding into a separate culture plate at the same time as the plate exposed to orbital flow. Flow experiments were conducted over five durations: 20 min, 1 h, 6 h, 24 h, 48 h, and 72 h.

RT-qPCR

At the end of each experiment, the well plates were removed from the incubator and cells were immediately lysed for total ribonucleic acid (RNA). We used a custom 3D-printed extractor to isolate RNA from the cell layers [1]. For the LIPUS-exposed cell layers, the spatial profile of the acoustic pressure field was used to guide the cell analysis. Therefore, the central 0-6 mm and, in some cases, peripheral 7-17.4 mm (radial distances) were isolated separately. For flow-exposed cell layers, the peripheral 11.0-17.4 mm was isolated.

RT-qPCR was performed to evaluate two genes that are mechanosensitive and atheroprotective, eNOS and KLF2 [1]. Briefly, cells were lysed with RLT lysis buffer (QIAGEN, MD, USA) and RNA was isolated with the RNeasy Mini Kit (QIAGEN). mRNA purity was assessed by measuring the 260/280 absorbance with a Take3 plate on a plate reader (BioTek Cytation 5, Agilent, Santa Clara, CA); ratios between 1.95 and 2.10 indicated an acceptable purity. Reverse transcription (RT) was done with an iScript^™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA) converting 200 ng of RNA with a MiniAmp Thermal Cycler (Applied Biosystems, Waltham, MA, USA). Primers were designed (Table 1) and validated in our previous work [1]. PCR was run following the manufacturer's protocol, using 200 pg/μl in triplicate with PowerUp SYBR Green Master Mix (ThermoFisher Scientific, Waltham, MA, USA) and a QuantStudio 3 real-time PCR system (Applied Biosystems, Waltham, MA, USA). Data were analyzed with the delta-delta Ct method. This method expresses a null response as 1 (i.e., 2⁰ = 1).

Table 1.

RT-qPCR primers for each gene.

Gene	Primer Sequence	Accession Number
eNOS	F: GTGAAGGCGACAATCCTGTA R: GGACACCACGTCATACTCATC	NM_000603.5
KLF2	F: AGAGTTCGCATCTGAAGGCG R: GTCTGAGCGCGCAAACTTC	NM_016270.4
HPRT1	F: CGCCCAAAGGGAACTGATA R: CTGTGGCCATCTGCTTAGT	NM_000194.3

Immunocytochemistry

Cell layers used for immunostaining to evaluate eNOS protein expression were maintained in an incubator for an additional 6 h following ultrasound exposure (12 h total from the start of the experiment). After, samples were fixed in 4% paraformaldehyde (Thermo Fisher Scientific, Waltham, MA, USA) under agitation for 10 minutes. Samples were then permeabilized in 0.1% Triton X-100 (Sigma, St. Louis, MO, USA) in PBS for 5 minutes and blocked for one hour in 5% bovine serum albumin (Sigma) with 0.1 M glycine (Sigma) and 0.1% Tween-20 (Sigma). Primary antibodies against eNOS (Abcam ab76198, Waltham, MA, USA) were applied overnight at 4°C in blocking solution at 1:100. Secondary antibodies tagged with AlexaFluor-488 (Abcam ab150125) diluted to 1:1000 in blocking buffer were applied under agitation for 1 hour. Nuclei were counterstained using DAPI at 0.1% for 5 minutes. Imaging was performed on a Zeiss LSM 800 Airyscan confocal microscope using a tile scan from the center of each well to a radial position of 5.5 mm (the approximate diameter of the ultrasound beam). Identical imaging parameters were used for control and experimental samples. Images were processed using a custom ImageJ macro accounting for cell density in each image. Data are reported as fold change in eNOS relative to static control where 1 is no difference from control.

Statistics

All statistical tests were performed in Minitab (version 21.4.0, State College, PA, USA). Quantities are reported as mean ± standard deviation (SD). If the data were normally distributed (determined by a Shapiro-Wilk test), group comparisons were performed using a Welch’s one-way analysis of variance (ANOVA) and pairwise comparisons, including those post-hoc of the ANOVA, were performed using an unpaired t-test with equal or unequal variances based on an F-test. Otherwise, a Kruskal-Wallis test (group comparisons) and Mann Whitney U test (pairwise comparisons) with Levene’s test for variances were used. Pairwise comparisons were done using a two-tailed approach. Multiple comparisons were corrected using the Holm-Bonferroni method. Correlation analysis was performed by calculating a Pearson’s correlation coefficient. An adjusted p-value of less than 0.05 was considered statistically significant.

Results

Ultrasound transducers exhibit stability and minimal heating over 24 h

We evaluated the stability of our transducers over longer duration experiments by performing mock 24 h experiments (i.e., no cells) at a target peak pressure of 217 kPa (1 Hz pulsing, 500,000 cycles per pulse) with a 1 MHz transducer in the culture wells that read the acoustic pressure every 2 min. The mean peak pressure over six wells (three different experiments) was 218±3 kPa (Fig. 5A). Thus, the signal deviated from the target by an average of only 0.5%. In-tank temperature measurements showed an increase in temperature from 33.5°C at the start of the experiment to a steady-state value of 38.2°C, which was achieved in approximately 12 h (Fig. 5B). Because setup immediately prior to an experiment requires opening the incubator, resulting in heat loss, we performed another 24 h experiment without turning on the transducers and found that the temperature increased from 33.4°C to a steady state value of 36.7°C. Thus, 24 h of pulsed ultrasound (pulsing at 1 Hz) at 217 kPa causes a temperature rise of only 1.5°C.

Figure 5.

Pressure and temperature measured within the wells of culture plates over 24 h of exposure to low-intensity pulsed ultrasound at a target pressure of 217 kPa. (A) Acoustic pressure demonstrated high stability over the long experiment time with a mean deviation from the target pressure of only 0.5%. The target pressure is indicated by a horizontal line at 217 kPa. (B) The in-well temperature increased over the experiment time from a mean of 33.5°C to 38.2°C. However, null experiments (i.e., experiments setup in an identical manner, but without powering the transducers) exhibited an increase from a mean of 33.2°C to 36.7°C (caused by warming of the system back to the ambient incubator temperature), demonstrating that the temperature increase due to heating from the transducers was only 1.5°C. Normal physiologic temperature of 37°C is indicated with a horizontal line.

Endothelial expression of eNOS under LIPUS is localized to the beam path

The boundary conditions of the culture wells, including the medium-air interface, may affect the pressure field experienced by the cell layer. Thus, we measured the pressure field at different radial positions within culture wells of a plate placed in our ultrasound exposure tank using a needle hydrophone (NM-0.5, ndtXducer) and found a nearly uniform signal around the target pressure of 217 kPa over most of the beam path that decreased nonlinearly in the periphery to a minimum value of 37 kPa at the peripheral edge of the well (Fig. 6A). This trend aligned with our results from the spatial scan (Fig. 2C). To assess potential differences in the endothelial biological response between the center and periphery of the cell layers, we evaluated eNOS expression after 6 h of exposure to 217 kPa. Cells within the beam path exhibited a substantial mean expression level of 2.74±0.58, whereas cells at the periphery had a significantly lower mean expression level of only 1.10±0.08 (p=0.0009) (Fig. 6B). However, this low value at the periphery was significantly higher than null (one-sample t-test, p=0.03). Importantly, these findings demonstrate that reflections resulting from the medium-air interface do not substantially impact the endothelial response away from the beam path when exposed to ultrasound in the basal-to-apical direction.

Figure 6.

Expression of endothelial nitric oxide synthase (eNOS) after 6 h of exposure to low intensity pulsed ultrasound at 217 kPa is localized to the beam path. (A) Acoustic pressure measured as a function of radial position within the wells of culture plates showed a dramatic nonlinear reduction from 217 kPa at the center (0 mm radius) to 37 kPa at the edge of the well (17.4 mm) (a reduction of 83%). Solid line shows polynomial piecewise fit to pinducer acoustic pressure measurements. Dashed line shows idealized pressure field. The center and periphery regions of the wells where RNA was extracted for evaluation of gene expression are indicated. (B) Endothelial expression of eNOS is significantly greater in the center (C) versus periphery (P) regions of the wells (n = 6 cell layers). Bars represent mean±SD. *Indicates statistically significant difference, wherein ***p<0.001. A value of 1 indicates no change in expression and is denoted by a horizontal line. (C) The in-tank temperature increased over the experiment time from a mean of 33.8°C to 37.1°C. However, null experiments setup in an identical manner but without powering the transducers exhibited an increase from a mean of 33.2°C to 36.1°C (caused by warming of the system back to the ambient incubator temperature). Normal physiologic temperature of 37°C is denoted by a horizontal line.

We also measured the temperature in the exposure tank during the 6 h experiments and found an increase from 33.8°C to 37.1°C, but the final mean temperature from the null experiments was 36.1°C, demonstrating a change in temperature due to ultrasound heating of only 1°C (Fig. 6C). The dramatic difference in eNOS expression at the center versus periphery of culture wells exposed to LIPUS, which closely aligns with changes in acoustic pressure, strongly suggests that such modest heating has negligible effects on the endothelial response. Note, there was no statistical difference between temperature measured in the exposure tank versus culture wells (p=0.11).

LIPUS induces a similar temporal expression of eNOS and KLF2 as normal flow

We next evaluated the time-dependent response of endothelial cells to a peak acoustic pressure of 217 kPa after 20 min, 1 h, 6 h, 9 h, and 24 h of exposure. At 20 min, eNOS expression was slightly reduced to 0.77±0.10 (Fig. 7A), in line with the known atherogenic effects of physiologic mechanical stimuli, such as normal flow, over short durations. At 1 h, expression increased to 1.14±0.07 and, at 6 h, it peaked with a maximum value of 2.74±0.58. After, expression dropped off slightly at 9 h (2.24±0.53) and then declined considerably to 1.06±0.30 at 24 h. Similarly, KLF2 exhibited a negligible response at 20 min (1.13±0.11) that then increased to a peak at 6 h of 2.00±0.37, followed by a decline to 0.93±0.14 at 24 h (Fig. 7B).

Figure 7.

Expression levels of endothelial nitric oxide synthase (eNOS) and Krüppel-like factor 2 (KLF2) show qualitatively similar trends over exposure time between low-intensity pulsed ultrasound and normal flow. (A, B) Endothelial expression of (A) endothelial nitric oxide synthase (eNOS) and (B) Krüppel-like factor 2 (KLF2) as a function of exposure time to low-intensity pulsed ultrasound at 217 kPa. A value of 1 indicates no change in expression and is denoted by a horizontal line. n = 4-6 cell layers per time point. (C, D) Expression of (C) eNOS and (D) KLF2 normalized to the maximum value for cell layers exposed to LIPUS and normal flow. (E, F) Absolute values of expression of (E) eNOS and (F) KLF2. Points represent the mean expression at each time, bars represent mean±SD, and lines are polynomial fits (in some cases, piecewise) to illustrate the trend. ^#p<0.1, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. For panels (E,F), *(normal font) indicates LIPUS-induced expression is greater than normal flow, whereas *(script font) indicates the inverse.

We then compared the LIPUS-induced eNOS and KLF2 expression trends over time to those obtained with normal flow. Remarkably, the trends were qualitatively very similar, but LIPUS induced peak beneficial signaling more rapidly than flow (Fig. 7C-D). Comparing individual time points, at 20 min eNOS was reduced under both LIPUS and flow (0.77±0.10 versus 0.64±0.24, respectively; p=0.34), but at 1 h endothelial cell layers exposed to LIPUS exhibited a significantly higher expression (1.14±0.07 versus 0.56±0.03, respectively; p<0.0001) (Fig. 7E). At 6 h, eNOS increased in both groups, but was significantly higher in the flow group (2.74±0.58 versus 4.84±0.42, respectively; p=0.0003), whereas at 24 h expression levels declined in the LIPUS group down to null but continued to increase in the flow group (1.06±0.30 versus 6.67±1.78, respectively; p=0.0072). Interestingly, trends in KLF2 expression over time exhibited uniformly lower levels in endothelial cells exposed to LIPUS compared to those exposed to flow (Fig. 7F).

LIPUS induces radially-dependent eNOS protein expression within the beam path

Endothelial cell layers exposed to LIPUS at 217 kPa for 6 h followed by 6 h of continued culture exhibited slightly increased eNOS protein expression over the entire beam path of 1.05±0.11 relative to controls (Fig. 8A). Interestingly, we found a highly significant positive correlation between eNOS expression and radial position within the beam path (ρ=0.94, p<0.0001) (Fig. 8B). At the edge of the beam path, mean eNOS expression exhibited its maximum value of 1.17±0.27, although due to large variations between cell layers, this value was not statistically different from null (one-sample t-test, p=0.11) (Fig. 8C).

Figure 8.

Endothelial nitric oxide synthase (eNOS) protein expression as a function of radial position in the low-intensity pulsed ultrasound (LIPUS) beam path. (A) Representative immunostained images from control and LIPUS-exposed endothelial cells at beam center and periphery (blue is DAPI and green is eNOS). (B) Linear correlation of eNOS protein expression with radial position in the beam path. (C) Peak eNOS expression at 5.5 mm from the center of LIPUS-exposed cell layers relative to control cell layers (one sample t-test). Data are mean±SD. ****p<0.0001.

LIPUS-induced eNOS expression is dependent on higher relative RNA yield

We observed that total RNA yield from endothelial cell layers exposed to LIPUS for 6 h at 217 kPa was significantly higher than controls (17.1±2.12 versus 12.8±1.07, p=0.0025) (Fig. 9A). Further, eNOS expression exhibited a very high positive correlation with relative RNA yield of ρ=0.92 (p=0.0097, Fig. 9B). Interestingly, when the culture period prior to introduction of LIPUS was increased from 24 h to 48 h, no difference in RNA yield was found between experimental and control cell layers (33.6±8.33 versus 31.4±7.83, p=0.67, Fig. 9C) and there was no increase in eNOS expression. As a result, eNOS expression after 6 h of LIPUS was significantly higher for cell layers seeded 24 h versus 48 h prior to the start of experiments (2.74±0.58 versus 1.04±0.23, p<0.0001, Fig. 9D).

Figure 9.

Endothelial nitric oxide synthase (eNOS) expression after exposure to low-intensity pulsed ultrasound (LIPUS) is dependent on RNA yield. (A) Comparison of RNA yield between cells exposed to 6 h of LIPUS after 24 h in culture and control. (B) Linear correlation of eNOS to relative RNA yield. (C) Comparison of RNA yield between cells exposed to 6 h of LIPUS after 48 h in culture and control. (D) Comparison of eNOS expression after 6 h of LIPUS for cells seeded 24 h versus 48 h prior to the start of experiments. Data are mean±SD. **p<0.01 and ****p<0.0001.

Discussion

Previous mechanobiology studies have shown that it takes about 24 h for endothelial cells to adapt to the mechanical effects of flow in terms of morphology [17], permeability [18], and elimination of the initial transient inflammatory response to normal flow from static conditions [16]. As such, in this study, we sought to determine the response of endothelial cells to the mechanical effects of LIPUS over time up to this maximum duration to evaluate potentially higher beneficial effects. We found that endothelial expression of eNOS and KLF2 were highly dependent on the duration of exposure to LIPUS. At the shortest duration of 20 min, endothelial cells exhibited more of an atherogenic phenotype with reduced eNOS and a negligible change in KLF2. Under longer durations of LIPUS, the expression of these genes rapidly increased to a peak at 6 h followed by a decline back to baseline within 24 h. Remarkably, the trends exhibited by these genes over time are qualitatively similar to those found for endothelial cells under normal flow, suggesting activation of the same signaling pathways for mechanotransduction.

To our knowledge, no other study has evaluated the effects of LIPUS on endothelial cells over long durations of exposure (>1 h). However, our results for short durations of exposure align with one study of endothelial cells exposed to LIPUS for 20 min that demonstrated increased nuclear translocation of yes-associated protein (YAP)/transcriptional co-activator with PDZ-binding motif (TAZ) [19], which is associated with inflammation and an atherogenic phenotype [20]. On the other hand, Altland et al. [21] demonstrated a 2-fold increase in nitrate/nitrite (a marker of nitric oxide production) in the media of endothelial cell layers exposed to low-intensity continuous ultrasound for only 10 min, indicating improved endothelial function. This discrepancy from our data may point to differences in timing of expression of the different markers (nitrate/nitrite assay versus mRNA) or it may result from differences in the chosen ultrasound parameters, wherein the authors used continuous ultrasound at a frequency of 27 kHz, peak acoustic pressure of about 86 kPa, and transducers were placed above the cell layers so the beam was traveling in the apical-to-basal direction.

One challenge with in vitro studies is the many factors that may influence the mechanical effects of LIPUS on cells. The individual effects of acoustic pressure, frequency, transducer orientation relative to the cell layer, and boundary conditions specific to the experimental setup on the cell response are under-explored questions. In our exposure tank, we chose to place the cell layers at a distance from the transducers in the far field to generate a stable acoustic pressure field that tapers uniformly over the culture well and allows better evaluation of the response of endothelial cells to acoustic pressure magnitude (when cells are isolated from specific regions of the culture well). A complexity of this approach is the medium-air interface above the cells that causes reflections. In a pilot study, we found that placement of an acoustic absorber in the beam path within the culture well lowered the acoustic pressure and eliminated the eNOS response (data not shown). We accounted for the effect of the absorber by increasing the incident pressure amplitude from 217 to 300 kPa, but this also increased the temperature within the well and, in turn, lowered cell viability (this was at least partly due to the close proximity of the absorber to the cells). Additional studies are needed to determine whether reflections from the air interface contribute to the cell response merely by increasing the magnitude of pressure within the well or through changes in wave directionality. Regardless, we expect that the LIPUS parameters identified through our in vitro studies will be close to what is required to achieve the same endothelial response in vivo, but future in vivo studies will be needed to precisely characterize the acoustic field within the body and potentially tune these parameters.

It is interesting that we found a high positive linear correlation between eNOS protein expression and radial position within the beam path that peaked at the periphery (~5.5 mm from the center). This phenomenon might be due to radial heterogeneities in the acoustic pressure field, although our measurements suggest a fairly uniform field until the very periphery of the beam path, where the pressure drops off precipitously (Fig. 6A). Another possible explanation is that there is a subtle increase in cell density from center to periphery of the culture well due to the meniscus of the cell suspension during seeding. This slightly higher density at the periphery of the beam path may improve the endothelial response to LIPUS—for the chosen seeding density. While more work is needed to better understand the cause of this phenomenon, it is clear that, in general, spatial heterogeneities in both the acoustic field and cell response should be taken into consideration when performing in vitro experiments.

Another interesting finding was that endothelial cell layers exposed to LIPUS exhibited higher mRNA yields compared to controls that strongly associated with eNOS mRNA expression. Because experimental and control cell layers were seeded at the same density, lysed over the same area (post experiment), and eluted at identical volumes, this finding suggests that LIPUS with our chosen parameters induced proliferation. This aligns with one in vitro study that showed LIPUS causes proliferation of HUVECs [22] and several other in vivo studies that showed LIPUS applied to the post-infarcted myocardium induces angiogenesis [10, 13, 23], suggesting a proliferative effect. To our knowledge, our finding is the first to demonstrate a positive correlation between mRNA yield and eNOS expression that is dependent on culture time prior to the experiment, which suggests a relationship to cell density. However, more work is needed to fully establish this relationship.

Conclusion

The few studies that have examined the endothelial response to LIPUS collectively demonstrate that there are many parameters important in determining this response, but little rationale is given for the choice of these parameters. In this study, we reasoned that endothelial cells exposed to longer continuous durations of LIPUS may exhibit stronger beneficial biological responses similar to endothelial cells exposed to normal flow. Indeed, our results demonstrated maximum expression of eNOS and KLF2 after 6 h of LIPUS. Moreover, we found similar temporal patterns of eNOS and KLF2 between LIPUS and normal flow, suggesting that both mechanical stimuli are regulated by the same signaling pathways. Since our recent study demonstrated that normal flow can be therapeutic in atherosclerosis [6], our findings herein also suggest that LIPUS could be therapeutic in the context of atherosclerosis. However, flow promoted far greater expression of both atheroprotective genes compared to LIPUS, at least for the regimens used herein. In addition, the beneficial endothelial response to LIPUS was dependent on RNA yield that may be related to increases in cell density, which would suggest that these regimens promote more of a wound healing than atheroprotective phenotype. Overall, this points to the need for more studies that characterize the endothelial response to the many LIPUS parameters over a broad range of values to determine whether other parameters are more beneficial for eliciting an atheroprotective phenotype similar to normal flow.

Data Availability Statement

Research data used in this manuscript are available from the corresponding author on reasonable request.