您当前的位置: 首页 > 网页快照
Acoustically triggered mechanotherapy using genetically encoded gas vesicles | Nature Nanotechnology
Abstract . Recent advances in molecular engineering and synthetic biology provide biomolecular and cell-based therapies with a high degree of molecular specificity, but limited spatiotemporal control. Here we show that biomolecules and cells can be engineered to deliver potent mechanical effects at specific locations inside the body through ultrasound-induced inertial cavitation. This capability is enabled by gas vesicles, a unique class of genetically encodable air-filled protein nanostructures. We show that low-frequency ultrasound can convert these biomolecules into micrometre-scale cavitating bubbles, unleashing strong local mechanical effects. This enables engineered gas vesicles to serve as remotely actuated cell-killing and tissue-disrupting agents, and allows genetically engineered cells to lyse, release molecular payloads and produce local mechanical damage on command. We demonstrate the capabilities of biomolecular inertial cavitation in vitro, in cellulo and in vivo, including in a mouse model of tumour-homing probiotic therapy. You have full access to this article via your institution. Download PDF Download PDF Main . Most existing approaches to disease treatment are either biochemical or physical. On the one hand, molecular and cellular therapies typically target diseased tissues through biomolecular recognition, and act via altered signalling or chemical cytotoxicity, but lack the ability to exert physical force or focus on specific anatomical locations 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 . On the other hand, surgery and non-invasive techniques, such as focused ultrasound (FUS), treat diseases by producing strong mechanical or thermal effects at precise anatomical locations, but usually lack the ability to target these effects on a molecular or cellular basis. If the molecular precision of synthetic biology could be combined with the spatial targeting and physical potency of non-invasive surgery, this would open a new realm of therapeutic possibilities. Here we enable this combination by showing that targeted biomolecules and engineered cells can act as seeds for inertial cavitation—a powerful physical phenomenon that involves the formation, growth and violent collapse of gas bubbles. This activity is triggered by FUS, which enables remote activation with high spatiotemporal precision deep within the tissue. Our approach takes advantage of gas vesicles (GVs)—a unique class of genetically encodable air-filled protein nanostructures that evolved in photosynthetic bacteria and archaea as a means to achieve cellular buoyancy 11 . GVs comprise amphiphilic protein shells with typical cylindrical widths of 45–250?nm and lengths of 100–600?nm that are permeable to gas but exclude liquid water due to their hydrophobic interior surface 12 , 13 (Fig. 1a,b ). GVs are thermodynamically stable, and their gas contents are in dynamic exchange with the surrounding media. Recently it was discovered that GVs’ gas-filled interiors allow them to serve as contrast agents for high-frequency diagnostic ultrasound, magnetic resonance imaging and optical coherence tomography 14 , 15 , 16 , 17 , 18 , 19 , 20 . In addition, it was shown that engineered multigene clusters that encode GVs can be heterologously expressed in genetically modified bacteria 21 and mammalian cells 22 , which enables non-invasive imaging of gene expression. However, the use of GVs and GV-expressing cells as therapeutic agents has not been investigated. Fig. 1: Purified GVs act as seeds for stable and inertial cavitation. a , Schematic drawing of a GV. The GV’s amphiphilic protein shell encloses a stable, gas-filled structure. b , Representative transmission electron microscopy images of intact (left) and collapsed (right) Ana GVs. c , Proposed mechanism of GV-seeded cavitation. An ultrasound (US) pulse with a positive pressure higher than the critical collapse pressure, P col , collapses the GV, which results in the release of a nanoscale air bubble. The released nanobubble undergoes cavitation if the PNP of the ultrasound pulse reaches below the critical cavitation pressure, P cav . Over several cycles, the nanobubble coalesces with neighbouring bubbles to convert into a micrometre-scale bubble, which can eventually undergo violent inertial cavitation. d , Diagram of the in vitro PCD set-up used to measure the acoustic signatures of cavitation activity in response to FUS. e , Representative frequency spectra of backscattered signals from purified GVs (0.3?nM) insonated by a single ultrasound pulse at varying PNPs, 30 cycles and 670?kHz. f , Mean harmonic signal from GVs (0.3?nM), BSA (concentration (mg?ml –1 ) matched to the GV concentration (mg?ml –1 )) and PBS as a function of PNP ( n ?=?16 for GVs and n ?=?8 for BSA and PBS). g , Mean broadband signal from GVs, BSA and PBS as a function of PNP ( n ?=?16 for GVs, n ?=?8 for BSA and PBS). Statistical analysis: orange , GVs versus PBS; orange +, GVs versus BSA; purple , BSA versus PBS. h , Average broadband measurements from GVs insonated with varying ultrasound pulse lengths (PNP?=?0.6?MPa, n ?=?12). i , Broadband signal from different concentrations of GVs insonated with a single 1.0?MPa pulse ( n ?=?5). Error bars, mean?±?s.e.m. n represents the independent samples for all the experiments. Null hypothesis testing was performed using two-sided t -tests: p ?Scale bars, 200?nm. Full size image In this study, we tested the hypothesis that at lower ultrasound frequencies GVs can serve as nuclei for the formation and cavitation of free bubbles, and thus turn targeted and cell-expressed GVs into mechanical warheads that can be activated with millimetre precision using FUS (Fig. 1c ). In previous work, seeded cavitation was provided by synthetic agents 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , which are entering clinical use due to their relative simplicity and reliability. However, synthetic agents are challenging to connect to cell-based therapeutics or specific extravascular biomolecular targeting 31 , 32 . Genetically encoded cavitation nuclei could build on the understanding of the physical processes at the core of ultrasound cavitation and combine it with the unique molecular specificity and active targeting abilities of disease-homing cells. We demonstrate the nucleation and cavitation of bubbles from GVs using acoustic spectroscopy and ultrafast optical microscopy, then examine the functionality of acoustically detonated biomolecules in several biological contexts. GVs act as seeds for bubble formation and cavitation . Our hypothesis that GVs can nucleate bubbles for inertial cavitation arises from the fact that GVs collapse under applied acoustic pressure above a genetically defined threshold (Fig. 1b ), which releases the air inside them to the surrounding media 14 , 15 , 17 , 21 , 22 . After collapse, the gas molecules released from the GVs are expected to form nanoscale bubbles, which should dissolve within milliseconds due to Laplace pressure 26 . However, we hypothesized that under sustained ultrasound pulses at frequencies in the sub-MHz range, the liberated nanobubbles could serve as seeds for cavitation—a process in which bubbles expand and contract during the negative and positive phases of sound waves, and can grow in size through coalescence 33 (Fig. 1c ). These processes are favoured at lower ultrasound frequencies and higher negative pressures, and can be facilitated by the presence of hydrophobic surfaces 24 , such as the exposed interior of collapsed GVs (Fig. 1b ). We postulated that low-frequency ultrasound pulses could be used to break open the GVs, release their gas contents as nanobubbles and drive cavitation. At sufficiently high amplitudes, these bubbles would undergo rapid growth and violent collapse in a process known as inertial cavitation, to unleash powerful mechanical effects 34 (Fig. 1c ). To test this basic hypothesis, we measured the acoustic emissions of GVs purified from Anabaena flos-aquae (Ana) during exposure to sub-MHz ultrasound. FUS was applied to GV suspensions in a custom-built chamber with acoustically transparent walls, and the emitted signals were recorded with an orthogonally positioned imaging transducer that acted as a passive cavitation detector (PCD) (Fig. 1d ). Unless stated otherwise, we insonated samples throughout this study at 670?kHz—within the range of frequencies commonly used in therapeutic ultrasound. The PCD transducer was a 128-element linear array with a centre frequency of 18?MHz. Throughout this work, we focused on low-to-moderate pressure levels as defined by mechanical index values (peak negative pressure (PNP) divided by the square root of the frequency) of less than 1.9 MPa?MHz –1/2 , which is below the safety limit of ultrasound imaging 35 . Bubbles that undergo stable cavitation emit sound waves at harmonic multiples of the transmitted frequency, whereas those that undergo inertial cavitation produce emissions with broad spectral content. We observed harmonic signals from GV suspensions during insonation at 0.2?MPa PNP (Fig. 1e,f ), and broadband emissions at pressures of 0.4?MPa or higher (Fig. 1e,g ). The latter threshold pressure and corresponding mechanical index (0.49?MPa?MHz –1/2 ) represent acoustic parameters substantially below the thresholds typically needed for spontaneous cavitation in tissue 36 , 37 , 38 . Similar signals were not recorded from the buffer or from solutions of the control protein bovine serum albumin (BSA). Inertial cavitation increased with the PNP up to 1?MPa (Fig. 1g ) and grew moderately with a pulse length above three cycles (Fig. 1h ). The broadband signal increased with GV concentration until it reached a peak at 0.3?nM (Fig. 1i ), above which acoustic shadowing interfered with the measurement (Extended Data Fig. 1 ). These results confirm the basic hypothesis that GVs can serve as seeds for inertial cavitation. As GVs are also used for ultrasound imaging 21 , 22 , typically at frequencies of several MHz, we also measured cavitation responses to insonation at 3?MHz. We found that inertial cavitation at this frequency required much higher pressures (Extended Data Fig. 2a,b ), consistent with the lower efficiency of bubble cavitation at higher frequencies and the increased pressure required to collapse GVs at frequencies above the gas permeation rate of their protein shell 39 . This result affirms the ability of GVs to be imaged safely using typical diagnostic parameters 19 , 20 while seeding inertial cavitation at lower frequencies. To more directly visualize GV-nucleated bubble formation and cavitation, we imaged this process optically using a five-million-frames-per-second camera (Fig. 2a ). We attached GVs to acoustically transparent Mylar-bottomed dishes using biotin and streptavidin chemistry. Before insonation, we observed a dark pattern indicative of intact GV clusters (Fig. 2b ), whose gas interiors scatter light 11 , 40 . After ultrasound was applied and reached a sufficient amplitude, this dark pattern suddenly disappeared (1.4 to 1.8?μs, Fig. 2b and Extended Data Fig. 3 ), consistent with GV collapse. After 2.4??s, we observed dark bubbles forming and cavitating inside the field of view (Fig. 2b and Supplementary Video 1 ), and continuing to grow over the subsequent cycles. Meanwhile, the control samples failed to show significant cavitation (Fig. 2c ). The GV-nucleated bubbles grew and shrank at the frequency of the ultrasound waves (Fig. 2d–f ). The phase of the wave that corresponded to GV collapse was π apart from the phase of the maximal bubble growth rate. As the latter corresponds to the negative peak of the ultrasound cycle, this confirms that GVs collapse at the peak positive pressure (Fig. 2e ). Consistent with this relationship, bubble size peaked at 3/2π, which corresponds to the conclusion of rarefaction (Fig. 2f ). Similar results were seen across bubbles (Fig. 2g ), which corroborates our acoustic measurements and provides direct support for the physical mechanism depicted in in Fig. 1c . Experiments performed in the presence of surfactant and the modelling of bubble dissolution kinetics suggested that bubble growth occurs primarily through coalescence (Fig. 2h,i and Extended Data Fig. 4 ). Fig. 2: Ultrafast optical imaging and acoustic recording of GV-seeded bubble formation and cavitation. a , Schematic drawing of the high frame rate (HFR) camera set-up that enables GV cavitation imaging at a frame rate of 5?MHz. Mfps, million frames per second. b , HFR camera images immediately before GV collapse (1.4??s, left), immediately after GV collapse (1.8??s, middle) and after the formation of bubbles (4.2??s, right). c , Number of unique cavitation loci in streptavidinated dishes with and without biotinylated GVs, on insonation with a single 1.4?MPa burst ( p ?=?0.0411, using a Wilcoxon rank sum test, n ?=?6 independent samples, field of view 200?×?125??m). FUS pulses with 30 cycles at 670?kHz were used unless otherwise stated. d , Representative high-speed camera frames showing every other maximum and minimum of bubble cavitation, preceded by GV collapse. e , Bubble growth rate, quantified as the temporal derivative of the normalized average inverted pixel intensity in d (left). The plot on the right shows each maximum in the growth rate aligned to the phase offset from the time of GV collapse. f , Bubble size and phase offset from GV collapse analysed from HFR images, as for e . g , Average phase offset for peak size and peak growth rate for three different regions of interest ( n ?=?3). h , Passive cavitation recordings measured with low and high concentrations of SDS. i , The time it takes for the PCD signal to reach 25% or its final amplitude as a function of SDS concentration ( n ?=?4 independent samples). Error bars, mean?±?s.e.m. p ?Scale bars, 20??m. The representative result in b belongs to one of the six repeats presented in c . a.u., arbitrary units. Full size image Receptor-targeted GVs serve as acoustically actuated cellular disruptors . After establishing that GVs can physically nucleate inertial cavitation, we investigated the feasibility of using this mechanical phenomenon in several model biological applications. In the first model, we tested whether GVs targeted to specific cell-surface receptors on tumour cells could be used as a warhead for acoustically actuated cell killing. To test this possibility, we prepared GVs genetically engineered to display an RGD peptide on the C terminus of their outer shell protein, GvpC, which gives them affinity for αVβ 3 integrin receptors commonly overexpressed in tumours 41 . We incubated these functionalized agents with U87 glioblastoma cells cultured on Mylar film (Fig. 3a,b ). For visualization, the GVs were also chemically labelled with Alexa Fluor 488. To the monitor cellular disruption, we supplemented the media with propidium iodide (PI), a dye that becomes fluorescent when it enters disrupted cells and interacts with the nucleic acids. Prior to ultrasound exposure, there was a negligible PI signal. However, after insonation for ten seconds, we observed PI uptake in many cells throughout the field of view (Fig. 3c–e ). Ultrasound alone in the absence of GVs did not result in noticeable PI uptake. Fig. 3: Molecularly targeted GVs serve as ultrasound-triggered disruptors of mammalian cells. a , Schematic drawing of the fluorescent microscopy set-up used to image the GV-mediated cell disruption. RGD-functionalized GVs were attached to U87 cells grown on Mylar-bottomed dishes. b , Bright field (BF) image of U87 cells and fluorescence images of GVs (green) and PI (red) 0.5?s before the application of ultrasound. c , PI fluorescence 5?s (left), 60?s (middle) and 300?s (right) after ultrasound exposure. d , Change in PI signal measured from individual cells (grey) and the average (red) before, during and after FUS application (represented by blue shading). e , Percentage of PI-positive cells after ultrasound exposure with and without GV attachment ( p ?=?0.0002 using a two-sided t -test, n ?=?4 independent samples). f , HFR camera image after 9.2??s showing the formation of bubbles during the ultrasound application to cells treated with GVs. g , Number of unique cavitation loci observed in dishes that contained U87 cells with and without GVs ( p ?=?4.1300?×?10 –12 , n ?=?8 independent samples). Null hypothesis testing was performed using two-sided t -tests. p ?Scale bars, 20??m. The representative results in b – d belong to one of the four repeats presented in e , whereas the representative result in f belongs to one of the eight repeats presented in g . LED, light-emitting diode. Full size image To directly visualize the effect of ultrasound on GVs attached to U87 cells, we imaged the cells with our ultrafast optical set-up. After the collapse of cell-attached GVs (Extended Data Fig. 5a,b ), we observed bubble formation and cavitation (Fig. 3f , Extended Data Fig. 5c and Supplementary Video 2 ). The number of cavitation events in the cells treated with GVs was significantly greater than the number of random cavitation events in the untreated controls (Fig. 3g ). Taken together, these results suggest that GVs can be used as targeted, acoustically triggered mechanical warheads for cellular disruption. Targeted GVs are frequently clustered, which could contribute to their function as cavitation nuclei (Extended Data Fig. 6a ). Genetically expressed GVs enable cells to undergo inertial cavitation . In addition to their use as purified biomolecules, GVs can be expressed inside engineered bacteria and mammalian cells 21 , 22 . We hypothesized that cells that express GVs could be triggered to undergo intracellular bubble formation and cavitation under low-frequency ultrasound, and result in cellular lysis and the release of a co-expressed intracellular payload (Fig. 4a ). To test this concept in bacteria, we engineered a 14-gene operon that combined the GV-encoding genes GvpA–GvpU from the ARG1 gene cluster 21 with a gene that encoded the luminescent NanoLuc protein as a model releasable payload (Fig. 4b ). This construct was transformed into Salmonella typhimurium SL1344, which has been used in multiple experimental bacterial therapies 2 , 3 . Cells transformed with this gene cluster produced abundant cytoplasmic GVs (Fig. 4c ). These cytoplasmic GVs are known to assemble into bundles within engineered cells (Extended Data Fig. 6b ) Fig. 4: GVs as genetically encoded seeds for cellular inertial cavitation and payload release. a , Proposed mechanism of intracellular GV-seeded cavitation and cell disruption. The ultrasound-trigged collapse of GVs inside the cell creates a nanobubble, which cavitates, disrupts the cell membrane and releases the cell contents, which includes a genetically encoded payload. b , Genetic construct combining a hybrid GV gene cluster from Ana and Bacillus megaterium with the NanoLuc luciferase (Lux) payload. c , Phase-contrast microscopy image of S. typhimurium cells expressing the construct in b . GVs appear inside each cell as white inclusions, while the rest of the cell appears black. d , Mean broadband emissions from NanoLuc-expressing cells (Lux, negative control) and cells that co-express GVs and NanoLuc, at various pressure levels ( p values for each pressure level P (MPa): P ?=?0.76, p ?=?0.0270; P ?=?0.88, p ?=?0.0006; P ?=?1, p ?=?0.0141, n ?=?8). e , Representative agar plates (left and middle) and average colony counts (right) for Lux and Lux?+?GV cells exposed to ultrasound at 300?kHz, PNP?=?1?MPa, normalized by the number of colonies from non-exposed control samples ( p ?=?0.0002, n ?=?8 for Lux?+?GV and n ?=?6 for Lux). f , Bioluminescent signal in the media that surrounds Lux or Lux?+?GV cells with and without exposure to ultrasound at 334?kHz, PNP?=?0.85?MPa. The signal from each sample is normalized by the signal measured from the same cells after chemical lysis (two-way ANOVA—the percentage of variation attributed to the interaction is 20.46% with p ?=?3.3825?×?10 –5 , n ?=?6). g , Structure of the three plasmids that contain the mammalian GV operon. h , Selective bursting of mammalian HEK293 cells using ultrasound. The percentage of dead GV-expressing cells, stained positive by the Zombie NIR fluorescent dye after ultrasound exposure at 334?kHz, PNP?=?1?MPa, was significantly higher than that of cells that were not exposed to ultrasound ( p ?=?0.0087, n ?=?6 and n ?=?5, respectively, using a Wilcoxon rank sum test). The readouts of uninduced cells with and without ultrasound exposure ( n ?=?6) serve as additional controls. Null hypothesis testing was performed using two-sided t -tests. p ?Scale bar, 15??m. n represents biological replicates ( d – f ) or independent samples ( h ). The representative GV?+?LUX results in e belong to one of eight repeats. Full size image To test whether the GV-expressing bacteria could serve as sources of inertial cavitation, we measured the broadband acoustic emissions from cell suspensions exposed to FUS. As it was previously shown that ARG1 GVs have a relatively high collapse pressure 21 and collapse is easier at lower frequencies 39 , we lowered our applied ultrasound frequency to 300?kHz in these experiments to ensure efficient collapse and subsequent cavitation at achievable pressure levels. The mechanical index remained below 1.9?MPa?MHz –1/2 in all the experiments. In response to ultrasound pulses, the GV-expressing S. typhimurium cells emitted a high level of broadband signals, which increased with the PNP (Fig. 4d ). These signals were well above the non-specific cavitation signals measured in control cells that expressed only NanoLuc. Next, we examined the ability of acoustic detonation to cause cell lysis and release a co-expressed protein payload from the engineered bacteria. The FUS treatment of GV-expressing S. typhimurium cells resulted in a 42% drop in the number of colony-forming units (Fig. 4e ), whereas no such reduction was seen in the NanoLuc-only controls, which confirms the GV- and ultrasound-specific cell lysis of a large fraction of the cells. Ultrasound exposure also caused the engineered cells to release their intracellular NanoLuc payload, as shown by the bioluminescence of the media (Fig. 4f ). Taken together, these results suggest that GV-expressing engineered cells can serve as ultrasound-activated cellular ‘explosives’, which release proteins into their surroundings in response to a remote trigger. To demonstrate acoustic detonation in mammalian cells, we applied FUS to the human cancer cell line HEK293 genetically engineered to express GVs in response to chemical induction 22 (Fig. 4g ), and compared the induced cells to uninduced controls. Ultrasound exposure resulted in cell lysis, as observed by counting cells labelled with a positive cell-death marker using flow cytometry. This lysis was highly specific to GV-expressing cells (Fig. 4h ). The fractional lysis of acoustically actuated cells can be attributed to the relatively small quantity of GVs produced in mammalian cells compared to bacteria using existing genetic constructs (Extended Data Fig. 7 ) 22 , and is expected to be improved in the future with additional genetic engineering. GVs seed inertial cavitation and disrupt targeted tissue in vivo . After establishing the GVs as biomolecular and genetically encoded cavitation nuclei in vitro and in cellulo, we tested the ability of GVs to seed cavitation and mediate tissue disruption in vivo. First, we looked at the acoustic signature of GV cavitation in a mouse tumour xenograft. For this purpose, a three-dimensional (3D)-printed holder was used to co-align the foci of the FUS transducer and an imaging probe (Fig. 5a ), with a needle-guide incorporated to facilitate precise injections into the tumour core. We used this set-up to perform ultrasound-guided acoustic detonations, in which we acquired images of the GVs injected into the tumour, performed FUS treatment with PCD monitoring and subsequently used imaging to confirm the destruction of intact GVs. Adult BALB/c mice were injected in their flanks with MC26 cells to form bilateral subcutaneous tumours. After the tumours reached dimensions of 6–10?mm (Fig. 5b , left), we performed ultrasound-guided GV injections and FUS treatment. We injected one tumour in each mouse with purified GVs and the contralateral tumour with saline as a vehicle control. To facilitate GV imaging against the tissue background, we used modified Ana GVs engineered to produce non-linear contrast 20 (Fig. 5b , middle). We applied FUS to the tumours before and after the GV or saline injection as we recorded the PCD signals. Strong broadband emissions, which represented inertial cavitation, were only detectable in GV-injected tumours (Fig. 5c,d ). After FUS exposure, the ultrasound image contrast from the injected GVs disappeared, which demonstrated their collapse (Fig. 5b , right). These results confirm that GVs can serve as biomolecular cavitation nuclei in the context of a disease-relevant living tissue. Fig. 5: GV-seeded cavitation and tissue disruption in vivo. a , Schematic drawing of in vivo FUS and imaging/PCD set-up. A 670?kHz FUS transducer and an 18?MHz imaging linear-array transducer were aligned using a 3D-printed cone holder. BALB/c mice harbour bilateral MC26 subcutaneous hind-limb tumour xenografts. b , Ultrasound images of tumours injected with engineered non-linear GVs or saline (sham control). The greyscale B-mode images present the anatomy of the tumour (left), and the amplitude modulation (AM) imaging (middle) shows GV-selective non-linear contrast, which disappears after FUS application (right). Scale bar, 3?mm. c , Representative PCD spectra measured before and after the injection of purified GVs (2.7?nM) or saline into tumours. Single 1.0?MPa, 30-cycle-long pulses were transmitted from a 670?kHz FUS transducer. d , Relative broadband signal measured during FUS application before and after GV or saline injections into tumours ( p ?=?0.002, n ?=?4 animals). e , AM images showing GV accumulation inside the liver 2?min after systemic injection (top). The signal was erased after insonation of the upper abdomen (bottom). Scale bar, 1?mm. f , Colour-unmixed histological microscopy image of a H&E-stained liver after the systemic administration of GVs and a targeted ultrasound application. The dimensions of the ultrasound focus are represented by the dashed circle. Signs of tissue disruption were clustered around the targeted liver region. Scale bar, 1?mm. g , Increase in the number of damage-related pixels in the medial lobe of mice injected with GVs versus saline control ( p ?=?0.008, n ?=?5 animals). h , Number of haemorrhagic foci surrounded by necrotic regions in livers of mice injected with GVs versus saline control ( p ?=?0.009, n ?=?5 animals). Null hypothesis testing was performed using two-sided t -tests. p ?cancer immunotherapy . Finally, we endeavoured to demonstrate the use of genetically encoded cavitation nuclei in the context of in vivo cell-based therapy. In particular, we focused on tumour-homing bacteria, which are rapidly emerging as a novel class of cancer therapeutics due to their ability to infiltrate the core of solid tumours from systemic circulation and be engineered to deliver a variety of local molecular therapies 2 , 3 , 10 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 . We hypothesized that GV-expressing bacteria could home to the cores of solid tumours and be activated with ultrasound to produce local cavitation, which mechanically disrupts the tumour tissue and thereby improves the efficacy of antitumour molecular immunotherapy. To test this concept, we expressed GVs in Escherichia coli Nissle 1917 (EcN) cells (Fig. 6a ). This cell strain is approved for human probiotic use and is commonly employed in microbial tumour therapies 47 , 51 . We engineered EcN cells to form a recently developed second-generation GV construct with improved expression and ultrasound contrast 52 . We injected immunocompetent mice with A20 lymphoma cells to cause the formation of subcutaneous solid tumours. After the tumours reached a substantial size, we injected the mice intravenously with the engineered EcN cells and gave them three days to infiltrate and populate the tumours before inducing them to express GVs with systemic l -arabinose administration. Fig. 6: Tumour mechanotherapy seeded by tumour-homing probiotics. a , Diagram illustrating the experiment performed to assess anticancer treatment that combined image-guided microbial mechanotherapy and immunotherapy in a mouse tumour model. Created with BioRender.com. b , xAM images showing in situ GV production inside an A20 tumour (top) compared with scans from control mice (bottom) injected with mScarlet (mSc)-expressing bacteria, before (left) and after (right) FUS exposure. The red square indicates the centre of the FUS focus. Scale bar, 2?mm. c , Tumour-homing cells colonize deep tumour regions (green, non-linear xAM imaging), distinct from highly perfused regions observed using intravascular microbubble ultrasound contrast agents (red, ultrasound super localization). Scale bar, 1?mm. d , Tumour growth in animals treated with bacteria induced to express mSc or GVs in situ, with or without FUS. All the mice were co-treated with αCTLA-4 and αPD-L1. FUS resulted in a significantly reduced tumour growth in the presence of GV-expressing bacteria compared with those of both control conditions (two-way ANOVA interaction: p ?=?0.0013 between the GVs with FUS and mSc with FUS and p ?=?0.0031 between the GVs with FUS and those without FUS; n ?=?11 animals for each GV group and n ?=?9 animals for the mSc group.) One of the GV control mice reached the humane endpoint at day 9. e , The combined treatment with the checkpoint inhibitors and FUS results in a significant increase in survival time in mice with GV-expressing tumour-homing cells ( n ?=?11 animals for the GV groups and n ?=?9 for the animals in the mSc group. p ?=?0.0114 between GVs with FUS and mSc with FUS using the Gehan–Breslow–Wilcoxon test. p ?=?0.0453 between GVs with FUS and GVs without FUS). p ?synthetic biology, FUS and bubble mechanics. In one class of potential future applications, purified GVs targeted to cells such as tumours via specific surface markers could serve as ultrasound-triggered disruptors of the plasma membrane, and so cause cell death and make the interior of the target cells more accessible to synergistic drugs and to discovery by the immune system. The development of these applications will benefit from GVs’ fundamental physical stability 14 , the tunability of their size and shape, and the ability of their surfaces to be functionalized via genetic fusion or chemical methods 17 , 41 . These properties will give GVs a unique profile relative to that of synthetic cavitation nuclei 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 . In a second class of potential uses, GVs expressed by engineered cells could provide a new capability for therapeutic synthetic biology, leveraging the ability of cellular agents to migrate to diseased tissues, recognize molecular signals, proliferate and turn on the expression of therapeutic transgenes 4 , 9 , 10 . Giving such agents the ability to express GVs as intracellular cavitation nuclei could turn them into ultrasound-triggered cellular disruptors, which release intracellular drugs and damage target tissues, such as tumours, to facilitate the entry of drugs and effective immunotherapy 2 , 44 , 57 . The ability of FUS to specify where and when in the body these effects take place could help overcome the off-target toxicity that arises from cell migration to untargeted tissues 49 , 58 . Such targeting could be aided by the ability of GVs and GV-expressing cells to be visualized with ultrasound and magnetic resonance imaging 14 , 15 , 16 , 17 , 18 , 19 , 20 . These and other potential applications should be evaluated in dedicated disease model studies. Additional work is also needed to further understand and advance the fundamental capabilities of biomolecular cavitation. The efficiency of the steps that lead from GV collapse to bubble formation, growth, coalescence and cavitation may be affected by the size, shape, functionalization and clustering of GVs, and by their interactions with cellular structures, such as the cytoskeleton, lipid membrane or extracellular matrix. Understanding these dependencies would allow GVs and GV-expressing cells to be engineered for optimal cavitation behaviour. Finally, previous work showed that repeated systemic injections of large doses of GVs are well tolerated by animals 42 , 59 . Similarly, GV expression by tumour-homing cells was well tolerated in our longitudinal in vivo experiments. However, clinical translation requires formal studies of immune responses as well as of dose-limiting and long-term toxicity. With these advances, biomolecular and cellular inertial cavitation promises an explosion of future applications. Methods . Chemicals . All the chemicals were purchased from Sigma Aldrich unless otherwise noted. Cell lines . All the cell lines were ordered from American Type Culture Collection. Gas vesicle purification, modification and quantification . Ana GVs were produced and purified as described previously 41 . As a quality assurance step, optical density (OD) measurements were performed for each sample at 0–12?bar using an echoVis Vis-NIR light source coupled with an STS-VIS spectrometer (Ocean Optics) and a 176.700-QS sample chamber (Hellma Analytics). GvpC was removed from the GVs used in the in vivo experiment to facilitate their ultrasound detection. GvpC was removed and replaced by an engineered, recombinantly expressed, protein, GvpC-RGD, for experiments in which the GVs were attached to U87 cells 17 . This GvpC variant (GvpC-R3) was also shorter than the wild-type protein, which resulted in GV collapse at lower pressure levels 17 . Prior to the experiments, GV concentrations were measured using a spectrophotometer (NanoDrop ND-1000, Thermo Scientific) at 500?nm. In vitro passive cavitation detection . The in vitro set-up (Fig. 1d ) was aligned in a three-step process. First, an L22-14v 128-element Verasonics imaging probe was positioned such that its focus, set to a depth of 8?mm, was at the centre of a 1?cm?×?1?cm 3D-printed chamber. Then, an optic fibre hydrophone (Precision Acoustics) was positioned at the centre of the chamber using B-mode imaging. Finally, a 670?kHz (Precision Acoustics) or 3?MHz FUS transducer (Ultrasonic S-Lab) was mounted on a computer-controlled 3D translatable stage (Velmex) orthogonally to the Verasonics probe. A MATLAB program automatically scanned and aligned the transducer’s focus at the centre of the chamber, according to the feedback from the hydrophone. Microcentrifuge tubes (2?ml, Thermo Scientific) were similarly used for the experiments that include cavitation of the GVs with different sodium dodecyl sulfate (SDS) concentrations. A solution of purified GVs (OD 500 ?=?0.5, or 0.3?nM, unless stated differently) or GV-expressing bacteria (OD 600 ?=?1) was gently pipetted into the 3D-printed chamber to minimize the introduction of bubbles. Purified GV solutions were prepared several days in advance to allow for natural degassing into atmospheric conditions. The Verasonics scanner was programmed to function with zero-amplitude transmit so that it could be used as a PCD. A MATLAB script was written to synchronize the acquisition of GV signals by triggering the FUS burst and accounting for the propagation time of the insonating wave to the focus. The GV solution was stirred gently during acquisition using a micro stir bar and magnetic stirrer (Thermo Scientific). To compensate for the slow rise time of the Precision Acoustics transducer, a custom waveform with 33 cycles was created such that the first 4 cycles had 1.5 times the amplitude of the last 29 cycles; this waveform induced our FUS to reach the desired PNP output at the 4th cycle. For the experiments in which the effect of the number of cycles on cavitation was investigated, a waveform with three cycles at twice the amplitude of the remaining cycles was used so that the desired PNP was reached at the third negative peak. In vitro cancer cell sonoporation experiments . Glass-bottomed 35?mm Petri dishes (Matsunami) were modified to enable the U87 cell culture and subsequent ultrasound application. The glass was removed using a glass cutter and Mylar film (2.5?μm thickness, Chemplex) was fixed over the hole via a polydimethylsiloxane elastomer (Sylgard 184 silicone, Dow Corning). After curing for 24?h at 40?°C, these dishes were sterilized using ultraviolet light. U87 cells were plated on the Mylar film dish and incubated for 2 days at 37?°C in 2?ml of DMEM media. To facilitate the attachment of RGD-functionalized GVs to the cell membranes, the GVs were resuspended in fresh DMEM (final concentration of 0.5 OD 500 ) and added to the centre of the dishes. The centre of each dish was sealed using an 18?mm round cover glass and kept inverted for 2?h at room temperature. Then, the cells were washed and recovered with fresh DMEM. Finally, 10?μg?ml –1 PI (Invitrogen) was added to the medium just before the ultrasound experiment. In this experiment, the cells were insonated using a 670?kHz transducer (Precision Acoustics) positioned in a water tank at an angle of 20° relative to the water surface, to minimize standing waves (Fig. 3a ). The ultrasound transducer was aligned to the microscope using a hydrophone. The cells were insonated for 10?s with PNP?=?1.5?MPa ultrasound pulses at a 2?ms pulse repetition rate. Fluorescence recording began 1?min before insonation, continued throughout the ultrasound exposure and ended 10?min after the insonation. Fluorescence signals were collected using a 10× immersion objective (Olympus, NA (numerical aperture) 0.3), and a sCMOS camera (Zyla 5.5, Andor) at a 10?Hz frame rate. After this acquisition, we used saponin (100?μg?ml –1 , Sigma) to perforate all the cell membranes, and the resulting image was used as a mask for cell-body detection. The fluorescence images were processed using NeuroCa 60 to extract fluorescent signals from individual cells. Cells were defined as PI positive if the signal intensity increased by more than 2% after the ultrasound application. High frame rate camera imaging experiments . We assembled a high-speed microscopy set-up capable of directly capturing GV collapse and bubble cavitation events (Fig. 2a ). Our set-up used a 2?W 532?nm laser (MLL-F-532-2W, CNI) controlled by an optical beam shutter (Thorlabs SH05, KSC101). Right-angle prism mirrors directed the laser light through a water bath and into a sample dish that contained the imaged samples. To image purified GVs, the GVs were biotinylated by incubating them for 1?h with a 10,000-fold molar excess of sulfo-NHS-biotin (Thermo Fisher Scientific). Then, the non-attached biotin was removed by dialysis in PBS. The Mylar dishes were prepared by first treating them with ultraviolet light, and then incubating them with 180??l of 0.1?mg?ml –1 poly- d -lysine hydrobromide. The dishes were then incubated with sulfo-NHS-biotin (180??l, 2?mM, Thermo Fisher Scientific) for 1?h. After washing away the free biotin with PBS, the dishes were incubated with streptavidin (180??l, 7.35??M, G-Biosciences) for 1?h and washed again to remove unbound streptavidin. Finally, the biotinylated GVs were attached to the dishes. After adding 180??l of GVs at OD 500 ?=?2 to each dish, the centre of each dish was sealed using an 18?mm round cover glass and kept inverted for 2?h at room temperature. Then, the excess GVs were washed away with PBS. All the wash steps were repeated at least three times. For imaging cells, the same dishes and cell culture processes described in the previous section were used, except with a final GV concentration of OD 500 ?=?2. Dishes that contained GVs or cells were positioned above a water tank and aligned with the transducer focus, as described above. A 10× water immersion Plan Fluor objective (Olympus, NA 0.3) was lowered into the solution in the dish. A series of prism mirrors and converging lenses with focal lengths of 200 and 50?mm delivered the image into a Shimadzu HPV-X2 camera, which acquired 256 images (Extended Data Fig. 3 ) over 51.2??s, at a sampling rate of five million frames per second. To account for acoustic propagation through the water, the camera was externally triggered to begin acquisition 40??s after the start of the ultrasound pulse. In these experiments, a single 30-cycle pulse with PNP of 1.4 MPa and a central frequency of 670 kHz was used to insonate the sample. Bacterial expression and experiments . GV-expressing S. typhimurium were produced by transforming cells with a plasmid that encoded an engineered genetic construct that comprised a GV operon 21 , 41 and a NanoLuc luciferase. Cells transformed with a NanoLuc-only plasmid were used as the controls. Constructs were assembled using Gibson cloning. The genetic constructs were cloned into the pTD103 plasmid (gift from J. Hasty), with expression driven by a luxI promoter (BBa_R0062) on induction with 3?nM N -(β-ketocaproyl)- l -homoserine lactone. The cells were cultured for 24?h at 30?°C after induction, then centrifugated for 4?h at 150 × g and 4?°C to enrich for buoyant cells. Samples of the cells used in the PCD experiments were stored for 2 days at 4?°C before these experiments and were always gently pipetted so as to minimize media gassing and bubble formation. Cells used for the NanoLuc release experiment were washed four times by 2?h of centrifugation at 150x g and 4?°C to remove any NanoLuc molecules that may be present in the media prior to the experiment. PCD recordings from GV-expressing cells were performed using the same set-up and protocol used for the PCD recordings from purified GVs, with cells at a concentration of OD 600 ?=?1. The same experimental set-up was also used in the bacteria lysis and payload release experiments; however, in these experiments, samples were insonated for 30?s at 300?kHz, PNP?=?1?MPa, with a pulse repetition interval of 2?ms. In addition, to place the cell samples at the main lobe of the ultrasound beam, the sample chamber was filled with 1% agar and a 3D printed insert was used for leaving a 4?mm diameter empty well at its centre. Cell samples were loaded into this well at a concentration of OD 600 ?=?2 and a volume of 50?ml, and covered with paraffin oil. For colony counts, the cells were plated on agar plates with kanamycin. Plates were imaged with a ChemiDoc Gel Imaging System (Bio-Rad) using a white epi-illumination protocol. Then, the colonies were counted to determine the total colony forming units. In the payload-release experiments, the samples were insonated for 30?s at 334?kHz, PNP?=?0.85?MPa, with a pulse repetition interval of 2?ms. The solution was aspirated from the chamber after exposure to ultrasound, pipetted into 100?kDa Spin-X UF concentrators (Corning) and centrifuged at 300 × g for 30?min to separate the supernatant fluid from the pellet and the buoyant cells. Then, the NanoLuc signal was measured using a Nano-Glo Luciferase assay kit (Promega) and a plate reader system (Molecular Devices). Full chemical lysis of the cells using SoluLyse Protein Extraction Reagent (Genlantis) was used as a positive control. Mammalian expression and experiments . mARG-HEK cells 22 were grown in 10?cm dishes. Once they reached 60–70% confluency, they were induced with 1??g/ml doxycycline and 5?mM sodium butyrate for 6 days. The cells were then trypsinized and resuspended in fresh DMEM before they were moved to the cavitation chamber. The samples were positioned in agar wells at the centre of the acoustically transparent cuvette, as discussed above. In these experiments, samples were insonated for 30?s at 334?kHz, PNP?=?1?MPa, with a pulse repetition interval of 2?ms. Cells were stained with Zombie NIR viability dye (BioLegend) following the manufacturer’s protocol, and cell lysis was measured using cell cytometry. Relative cell death was measured using the Beckman Coutler Cytoflex Flow Cytometer (Beckman Coutler Inc.) based on Zombie near-infrared fluorescence (Extended Data Fig. 9 ). In vitro ultrasound imaging of GV-expressing cells . To create phantoms for in vitro ultrasound imaging, wells were cast with molten 1% w/v agarose in PBS using a custom 3D-printed template. The bacteria and mammalian cell samples were loaded into agarose phantoms with the same concentration as that used in cavitation experiments. Next, the cells were mixed at a 1:1 ratio with 42?°C low-melt agarose and pipetted into the wells before solidification. The volume of each well was 25??l and it contained either 2?×?10 5 trypsinized mARG-HEK cells or OD 600 ?=?1 (about 2?×?10 6 ) GV-expressing S. typhimurium cells. The phantoms were submerged in PBS, and ultrasound images were acquired using a Verasonics Vantage system and L22-14v transducer (Verasonics). Each frame was formed from 89 focused beam-ray lines, each with a 40-element aperture and 8?mm focus 20 . A three half-cycle transmit waveform at 17.9?MHz was applied to each active array element. Each image captured a circular cross-section of a well with a 4?mm diameter and centre positioned at a depth of 8?mm. The signal was acquired at 0.27?MPa (2?V) for ten frames and the acoustic pressure was increased to 1.57?MPa (10?V) to collect 46 additional frames. Subtracted ultrasound images were constructed by subtracting the collapsing frame (frame 11) from frame four post-collapse (frame 15). In vivo passive cavitation detection . All in vivo experiments were performed on BALB/c female mice under a protocol approved by the Institutional Animal Care and Use Committee of the California Institute of Technology. No randomization or blinding were necessary in this study. Mice were anaesthetized using 1–2.5% isoflurane during all the injection and imaging procedures. The MC26 colorectal cancer cell line was maintained as per standard cell culture techniques. Four female BALB/c mice, aged 8 weeks, were given subcutaneous inoculations of 1?×?10 6 MC26 cells into both the right and left hind flanks. Tumours were monitored and permitted to grow to a diameter of 6–10?mm over 10–20 days. None of these tumours reached the maximal tumour size of 1,500 mm 3 permitted by the ethics committee and all of their dimensions were well below 15?mm. A 3D-printed theranostic holder co-aligned a 670?kHz FUS transducer (Precision Acoustics) and an L22-14v 128-element imaging probe (Verasonics) by fixing the FUS focus at 12?mm along the imaging plane (Fig. 5a ). The holder places the FUS cone facing downwards and the imaging probe at approximately 30° from the vertical. A 3D-printed needle-guide was mounted to the side of the cone such that the tip of a 1?inch (25.4?mm) 30-gauge injection needle also intersected the focus. The holder was mounted on a manually controlled 3D positioner. Mice were anaesthetized, maintained at 37?°C on a heating pad, depilated over the tumour region and positioned with the tumour facing directly upwards. Prior to the experiment, ultrasound gel was centrifugated at 2,000 × g for 10?min to remove bubbles, heated to 37?°C and then carefully applied to couple the cone and probe to the tumour. B-mode anatomical imaging was used to confirm the absence of bubbles in the gel application and to position the centre of the tumour at an axial depth of 12?mm. B-mode and AM 20 images of the tumour were saved pre- and post-insonation. Insonation comprised a single 670?kHz, 30-cycle, PNP?=?1?MPa burst. The same PCD script was used as that for in vitro PCD acquisitions. As part of the experiment, 20??l of GVs (OD 500 ?=?4.5) with GvpC removed or saline were infused directly into the tumour at a flow rate of 10??l?min ?1 via a Genie Touch syringe pump through PE10 catheter tubing and a 30-gauge needle (BD). Injection at the focal zone was confirmed via B-mode imaging by locating the needle tip, and AM and B-mode images were recorded pre- and post-insonation. PCD measurements were performed during each insonation. Cavitation effect on liver tissue in vivo . Five female BALB/c mice aged 8 weeks were included in each group. The 3D-printed in vivo theranostic holder was aligned and used, as discussed above. Mice were anaesthetized, maintained at 37?°C on a heating pad, depilated over the abdomen and lower torso region, and placed in a supine position. The therapeutic and imaging transducer we coupled to the mouse with a centrifuged ultrasound gel prepared as described above. B-mode anatomical imaging was used to position the FUS beam at the region of interest in the liver tissue. Saline or 200??l of Ana GVs (OD 500 ?=?37) with GvpC removed were infused systemically using a tail vein catheter at a flow rate of 100??l?min ?1 via a Genie Touch syringe pump through PE10 catheter tubing and a 30-gauge needle (BD). To confirm the liver uptake and post-treatment GV collapse, B-mode and AM 20 scans of the liver were saved pre- and post-injection, and post-insonation. Here, insonation comprised a 30-cycle ultrasound pulse train at 670?kHz, PNP?=?1.5?MPa and a 2?ms pulse repetition interval for 30?s. Following the ultrasound exposure, we perfused the mice and collected their livers. Tissue was fixed for 48?h in 10% formalin and then moved to 70% EtOH for storage. Next, the fixed tissue was embedded in paraffin, sectioned and stained with H&E. The ImageJ 61 colour unmixing function was applied to the H&E images, which resulted in separate images of the tissue, the blood and the background 62 according to their distinct characteristic colours (Extended Data Fig. 8 ). The colour unmixing parameters are given in Supplementary Table 1 . Bleeding was quantified by summing over the blood-positive pixels (greyscale?>?220) and by counting the haemorrhagic foci. The haemorrhagic foci were defined according to the H&E images as areas that showed bleeding, in red, surrounded by necrotic tissue in pink. Cavitation seeded by tumour homing cells in vivo . To establish A20 tumours in mice, 5?×?10 6 A20 murine lymphoma cells were collected, suspended in PBS and mixed with Matrigel (overall volume of 100??l), before injecting them subcutaneously into the right flank of each 7–9-week-old BALB/c mouse. When the tumour volumes reached approximately 150?mm 3 , EcN cells were injected into each of the A20 tumour-bearing mice via a tail vein. Plasmids that contained engineered genetic circuits 52 were transformed into Nissle 1917 E. coli (Mutaflor). Nissle cells were cultured in LB broth (Sigma) that contained the appropriate antibiotics, supplemented with 1% (w/v) glucose for overnight cultures and 0.4% (w/v) glucose for shake flask cultures. Cells were grown from glycerol stocks overnight in a shaking incubator (37?°C, 250?r.p.m.). The next day, OD 600 measurements were taken, and the saturated cultures were diluted 100×?. Diluted cultures were then allowed to grow to the exponential phase until they reached 0.3–0.6 OD 600 and then were collected by centrifugation (3,500 × g for 5?min), washed with phosphate buffer saline (PBS) four times and suspended in PBS at 0.625?OD 600 . A 100??l aliquot of the resulting solution was injected into each mouse. Starting three days post-injection of the EcN cells, they were induced with 120?mg of l -arabinose IP daily for two days. On the third and sixth day post-EcN injection, all the mice were also injected with a combination of αCTLA-4 (200??g per mouse per time point) and αPD-L1 (100??g per mouse per time point) checkpoint inhibitors intraperitoneally. These murine checkpoint inhibitors (αCTLA-4 clone 9D9 and αPD-L1 clone 10F.9G2) were obtained from a commercially available source (BioXCell). On the fifth day post bacteria injections, mice with GV-expressing bacteria and control mice were exposed to FUS under ultrasound guidance. As in previous experiments with tumour-homing cells, the tumours were insonated for 30?s at 334?kHz, PNP?=?1?MPa and a pulse repetition interval of 2?ms. Tumour sizes were recorded regularly for 40 days or until they reached the humane endpoint criteria. When tumours reached 1,500?mm 3 , or one of their dimensions reached 15?mm, the mice were euthanized. Passive cavitation detection data processing . The acoustic emissions acquired by the PCD were sampled by a Verasonics scanner at 62.5?MHz and processed using a MATLAB (2017b, Mathworks) script. Single channel, 8,192-point fast Fourier transform frequency spectrum estimations of the radio-frequency recordings from the 128 transducer elements were averaged to produce each PCD frequency spectrum estimation. To calculate the average amount of stable cavitation, an acquisition with a clear harmonic response was used to manually select and save the peak harmonic frequencies (Supplementary Table 2 ). Then, a trend curve was fitted to each spectral estimation using the MATLAB Curve Fitting tool. The smoothing parameter was chosen such that the resulting curves included only the broadband signal and no harmonic peaks. This smoothened curve was subtracted from the original frequency spectrum, and the resulting flattened spectrum with harmonic peaks was integrated over the peaks, each with a bandwidth of 191?kHz for the 670?kHz measurements, or 610?kHz for 3?MHz measurements (so as to include the full harmonic peak). Integrals were performed using trapezoidal sums, and then the integral was divided by the product of the number of peaks and the peak bandwidth. To calculate the average amount of broadband emission, the baseline noise at PNP?=?0?MPa was subtracted from each reading. Then, the flattened spectrum with harmonic peaks was subtracted. The remaining spectrum was integrated between 7.4 and 28.6?MHz, which marked the beginning and end of baseline noise, and then divided by 21.2?MHz. Statistical analysis . For statistical significance testing, we used two-sided heteroscedastic t -tests with a significance level of type I error set at 0.05 to reject the null hypothesis, unless mentioned otherwise. A Wilcoxon rank sum test was used in high-speed camera experiments, which included a control group that was found to be non-Gaussian using a Lilliefors test ( p ?=?0.001). Similarly, the Wilcoxon rank sum test was used in the experiment with the GV-expressing HEK293 cells in which the ultrasound-treated group was found to be non-Gaussian. A two-way analysis of variance (ANOVA) test was used in the payload release experiment and for tumour size measurements. Sample sizes for all the experiments, including the animal experiments, were chosen on the basis of preliminary experiments to be adequate for statistical analysis. Reporting Summary . Further information on research design is available in the Nature Research Reporting Summary linked to this article. Data availability . Plasmids sequences are given in Supplementary Table 3 , and will be made available through Addgene. All the raw data related to the plots and graphs are available at https://github.com/shapiro-lab/GV_cavitation.git . All the other materials and data are available from the corresponding author upon reasonable request. Code availability . MATLAB codes are available from the corresponding author upon reasonable request. References . 1. Milenic, D. E., Brady, E. D. & Brechbiel, M. W. Antibody-targeted radiation cancer therapy. Nat. Rev. Drug Discov. 3 , 488–498 (2004). CAS ? Article ? Google Scholar ? 2. Din, M. O. et al. Synchronized cycles of bacterial lysis for in vivo delivery. Nature 536 , 81–85 (2016). CAS ? Article ? Google Scholar ? 3. Danino, T. et al. Programmable probiotics for non-invasive urinary detection of cancer. Sci. Transl. Med. 7 , 289ra84 (2015). Article ? CAS ? Google Scholar ? 4. Claesen, J. & Fischbach, M. A. Synthetic microbes as drug delivery systems. ACS Synth. Biol. 4 , 358–364 (2015). CAS ? Article ? Google Scholar ? 5. Daniel, C., Roussel, Y., Kleerebezem, M. & Pot, B. Recombinant lactic acid bacteria as mucosal biotherapeutic agents. Trends Biotechnol. 29 , 499–508 (2011). CAS ? Article ? Google Scholar ? 6. Steidler, L. et al. Treatment of murine colitis by Lactococcus lactus secreting interleukin-10. Adv. Sci. 289 , 1352–1355 (2011). Google Scholar ? 7. Davili, M. L. et al. Efficacy and toxicity management of 19-28z CAR T cell therapy in B cell acute lymphoblastic leukemia. Sci. Transl. Med. 6 , 224ra25 (2014). Article ? CAS ? Google Scholar ? 8. Jackson, H. J., Rafiq, S. & Brentjens, R. J. Driving CAR T-cells forward. Nat. Rev. Clin. Oncol. 13 , 370–383 (2016). CAS ? Article ? Google Scholar ? 9. Scheller, L. & Fussenegger, M. From synthetic biology to human therapy: engineered mammalian cells. Curr. Opin. Biotechnol. 58 , 108–116 (2019). CAS ? Article ? Google Scholar ? 10. Riglar, D. T. & Silver, P. A. Engineering bacteria for diagnostic and therapeutic applications. Nat. Rev. Microbiol. 16 , 214–225 (2018). CAS ? Article ? Google Scholar ? 11. Walsby, A. E. Gas vesicles. Annu. Rev. Plant Physiol. 26 , 427–439 (1975). CAS ? Article ? Google Scholar ? 12. Pfeifer, F. Distribution, formation and regulation of gas vesicles. Nat. Rev. Microbiol. 10 , 705–715 (2012). CAS ? Article ? Google Scholar ? 13. Dutka, P. et al. Measuring gas vesicle dimensions by electron microscopy. Protein Sci. 30 , 1081–1086 (2021). CAS ? Article ? Google Scholar ? 14. Shapiro, M. G. et al. Biogenic gas nanostructures as ultrasonic molecular reporters. Nat. Nanotechnol. 9 , 311–316 (2014). CAS ? Article ? Google Scholar ? 15. Lu, G. J. et al. Imaging of gas-filled protein nanostructures. Nat. Mater. 17 , 456–463 (2018). CAS ? Article ? Google Scholar ? 16. Shapiro, M. G. et al. Genetically encoded reporters for hyperpolarized xenon magnetic resonance imaging. Nat. Chem. 6 , 629–634 (2014). CAS ? Article ? Google Scholar ? 17. Lakshmanan, A. et al. Molecular engineering of acoustic protein nanostructures. ACS Nano 10 , 7314–7322 (2016). CAS ? Article ? Google Scholar ? 18. Kunth, M., Lu, G. J., Witte, C., Shapiro, M. G. & Schro, L. Protein nanostructures produce self-adjusting hyperpolarized magnetic resonance imaging contrast through physical gas partitioning. ACS Nano 11 , 10939–10948 (2018). Article ? CAS ? Google Scholar ? 19. Maresca, D., Sawyer, D. P., Renaud, G., Lee-Gosselin, A. & Shapiro, M. G. Nonlinear X-wave ultrasound imaging of acoustic biomolecules. Phys. Rev. X 8 , 41002 (2018). CAS ? Google Scholar ? 20. Maresca, D. et al. Nonlinear ultrasound imaging of nanoscale acoustic biomolecules. Appl. Phys. Lett. 110 , 0073704 (2017). Article ? CAS ? Google Scholar ? 21. Bourdeau, R. W. et al. Acoustic reporter genes for noninvasive imaging of microorganisms in mammalian hosts. Nature 553 , 86–90 (2018). CAS ? Article ? Google Scholar ? 22. Farhadi, A., Ho, G. H., Sawyer, D. P., Bourdeau, R. W. & Shapiro, M. G. Ultrasound imaging of gene expression in mammalian cells. Science 365 , 1469–1475 (2019). CAS ? Article ? Google Scholar ? 23. Airan, R. D. et al. Noninvasive targeted transcranial neuromodulation via focused ultrasound gated drug release from nanoemulsions. Nano Lett. 17 , 652–659 (2017). CAS ? Article ? Google Scholar ? 24. Kwan, J. J. et al. Ultrasound-propelled nanocups for drug delivery. Small 11 , 5305–5314 (2015). CAS ? Article ? Google Scholar ? 25. Kwan, J. J. et al. Ultrasound-induced inertial cavitation from gas-stabilizing nanoparticles. Phys. Rev. E 92 , 023019 (2015). CAS ? Article ? Google Scholar ? 26. Ferrara, K., Pollard, R. & Borden, M. Ultrasound microbubble contrast agents: fundamentals and application to gene and drug delivery. Annu. Rev. Biomed. Eng. 9 , 415–447 (2007). CAS ? Article ? Google Scholar ? 27. Kooiman, K., Vos, H. J., Versluis, M. & De Jong, N. Acoustic behavior of microbubbles and implications for drug delivery. Adv. Drug Deliv. Rev. 72 , 28–48 (2014). CAS ? Article ? Google Scholar ? 28. Kotopoulis, S., Dimcevski, G., Gilja, O. H., Hoem, D. & Postema, M. Treatment of human pancreatic cancer using combined ultrasound, microbubbles, and gemcitabine: a clinical case study. Med. Phys. 40 , 072902 (2013). Article ? CAS ? Google Scholar ? 29. Sheeran, P. S. & Dayton, P. A. Phase-change contrast agents for imaging and therapy. Curr. Pharm. Des. 18 , 2152–2165 (2012). CAS ? Article ? Google Scholar ? 30. Kripfgans, O. D., Fowlkes, J. B., Miller, D. L., Eldevik, O. P. & Carson, P. L. Acoustic droplet vaporization for therapeutic and diagnostic applications. Ultrasound Med. Biol. Biol. 26 , 1177–1189 (2000). CAS ? Article ? Google Scholar ? 31. K?se, G., Darguzyte, M. & Kiessling, F. Molecular ultrasound imaging. Nanomaterials 10 , 1935 (2020). Article ? CAS ? Google Scholar ? 32. Woudstra, L. et al. Development of a new therapeutic technique to direct stem cells to the infarcted heart using targeted microbubbles: StemBells. Stem Cell Res. 17 , 6–15 (2016). CAS ? Article ? Google Scholar ? 33. Ashokkumar, M., Lee, J., Kentish, S. & Grieser, F. Bubbles in an acoustic field: an overview. Ultrason. Sonochem. 14 , 470–475 (2007). CAS ? Article ? Google Scholar ? 34. Coussios, C. C. & Roy, R. A. Applications of acoustics and cavitation to noninvasive therapy and drug delivery. Annu. Rev Fluid Mech. 40 , 395–420 (2008). Article ? Google Scholar ? 35. Szabo, T. L. Diagnostic Ultrasound Imaging: Inside Out (Academic, 2004). 36. Tran, B. C., Seo, J., Hall, T. L., Fowlkes, J. B. & Cain, C. A. Microbubble-enhanced cavitation for noninvasive ultrasound surgery. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 50 , 1296–1304 (2003). Article ? Google Scholar ? 37. Church, C. C. Frequency, pulse length, and the mechanical index. Acoust. Res. Lett. 6 , 162 (2007). Article ? Google Scholar ? 38. Holland, C. K. & Apfel, R. E. An improved theory for the prediction of microcavitation thresholds. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 36 , 204–208 (1989). CAS ? Article ? Google Scholar ? 39. Cherin, E. et al. Acoustic behavior of Halobacterium salinarum gas vesicles in the high-frequency range: experiments and modeling. Ultrasound Med. Biol. 43 , 1016–1030 (2017). Article ? Google Scholar ? 40. Walsby, A. E. The pressure relationships of gas vacuoles. Proc. R. Soc. Lond. B 178 , 301–326 (1971). Article ? Google Scholar ? 41. Lakshmanan, A. et al. Preparation of biogenic gas vesicle nanostructures for use as contrast agents for ultrasound and MRI. Nat. Protocols 12 , 2050–2080 (2017). CAS ? Article ? Google Scholar ? 42. Le Floc’h, J. et al. In vivo biodistribution of radiolabeled acoustic protein nanostructures. Mol. Imaging Biol. 20 , 230–239 (2018). Article ? CAS ? Google Scholar ? 43. Duong, M. T. Q., Qin, Y., You, S. H. & Min, J. J. Bacteria-cancer interactions: bacteria-based cancer therapy. Exp. Mol. Med. 51 , 1–15 (2019). CAS ? Article ? Google Scholar ? 44. Dang, L. H., Bettegowda, C., Huso, D. L., Kinzler, K. W. & Vogelstein, B. Combination bacteriolytic therapy for the treatment of experimental tumors. Proc. Natl Acad. Sci. USA 98 , 15155–15160 (2001). CAS ? Article ? Google Scholar ? 45. Leschner, S. et al. Tumor invasion of Salmonella enterica serovar Typhimurium is accompanied by strong hemorrhage promoted by TNF-α. PLoS ONE 4 , e6692 (2009). Article ? CAS ? Google Scholar ? 46. Kang, S.-R. et al. Imaging of tumor colonization by Escherichia coli using 18 F-FDS PET. Theranostics 10 , 4958–4966 (2020). CAS ? Article ? Google Scholar ? 47. Gurbatri, C. R. et al. Engineered probiotics for local tumor delivery of checkpoint blockade nanobodies. Sci. Transl. Med. 12 , eaax0876 (2020). CAS ? Article ? Google Scholar ? 48. Jiang, S.-N. et al. Inhibition of tumor growth and metastasis by a combination of Escherichia coli- mediated cytolytic therapy and radiotherapy. Mol. Ther. 18 , 635–642 (2010). CAS ? Article ? Google Scholar ? 49. Ryan, R. M. et al. Bacterial delivery of a novel cytolysin to hypoxic areas of solid tumors. Gene Ther. 16 , 329–339 (2009). CAS ? Article ? Google Scholar ? 50. Groot, A. J. et al. Functional antibodies produced by oncolytic clostridia. Biochem. Biophys. Res. Commun. 364 , 985–989 (2007). CAS ? Article ? Google Scholar ? 51. Leventhal, D. S. et al. Immunotherapy with engineered bacteria by targeting the STING pathway for anti-tumor immunity. Nat. Commun. 11 , 2739 (2020). CAS ? Article ? Google Scholar ? 52. Hurt, R. C. et al. Genomically mined acoustic reporter genes enable on-demand in vivo monitoring of tumor-homing bacteria. Preprint at bioRxiv https://doi.org/10.1101/2021.04.26.441537 (2021). 53. Errico, C. et al. Ultrafast ultrasound localization microscopy for deep super-resolution vascular imaging. Nature 527 , 499–502 (2015). CAS ? Article ? Google Scholar ? 54. Eranki, A. et al. High-intensity focused ultrasound (hIFU) triggers immune sensitization of refractory murine neuroblastoma to checkpoint inhibitor therapy. Clin. Cancer Res. 26 , 1152–1161 (2020). CAS ? Article ? Google Scholar ? 55. Qu, S. et al. Non-thermal histotripsy tumor ablation promotes abscopal immune responses that enhance cancer immunotherapy. J. Immunother. Cancer 8 , 1–12 (2020). Article ? Google Scholar ? 56. Chavez, M. et al. Distinct immune signatures in directly treated and distant tumors result from TLR adjuvants and focal ablation. Theranostics 8 , 3611–3628 (2018). CAS ? Article ? Google Scholar ? 57. H?cker, G., Redecke, V. & H?cker, H. Activation of the immune system by bacterial CpG-DNA. Immunology 105 , 245–251 (2002). Article ? Google Scholar ? 58. Forbes, N. S., Munn, L. L., Fukumura, D. & Jain, R. K. Sparse initial entrapment of systemically injected Salmonella typhimurium leads to heterogeneous accumulation within tumors. Cancer Res. 63 , 5188–5193 (2003). CAS ? Google Scholar ? 59. Ling, B. et al. Biomolecular ultrasound imaging of phagolysosomal function. ACS Nano 14 , 12210–12221 (2020). CAS ? Article ? Google Scholar ? 60. Jang, M. J. & Nam, Y. NeuroCa: integrated framework for systematic analysis of spatiotemporal neuronal activity patterns from large-scale optical recording data. Neurophotonics 2 , 035003 (2015). Article ? Google Scholar ? 61. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9 , 671–675 (2012). CAS ? Article ? Google Scholar ? 62. Bar-Zion, A., Yin, M., Adam, D. & Foster, F. S. Functional flow patterns and static blood pooling in tumors revealed by combined contrast-enhanced ultrasound and photoacoustic imaging. Cancer Res. 76 , 4320–4331 (2016). CAS ? Article ? Google Scholar ? Download references Acknowledgements . The authors thank D. Piraner, A. Lakshmanan, A. Farhadi and P. Ramesh for helpful discussions. In addition, we thank A. Farhadi for his help with the GvpC-RGD variant and H. Davis for his inputs on the optical design of the high-speed set-up. We thank M. Harel ( www.maayanillustration.com ) for the illustrations in this paper. We also thank A. McDowall for help with electron microscopy and C. Rabut for help with the animal experiments. This project was supported by the David and Lucile Packard Fellowship for Science and Engineering (M.G.S.) and the Heritage Medical Research Institute (M.G.S.). In addition, this project received funding from the European Union’s Horizon 2020 research and innovation programme under Marie Sk?odowska-Curie grant agreement no. 792866 (A.B.-Z.). A.B.-Z. was also supported by the Lester Deutsch Fellowship. A.N. was supported by the Amgen scholars programme. S.S. is supported by the NSF Graduate Research Fellowship. M.H.A. is supported by the NSF Graduate Research Fellowship and the P.D. Soros Fellowship. M.T.B. is supported by the NSF Graduate Research Fellowship. D. Maresca is supported by the Human Frontiers Science Program Cross-Disciplinary Fellowship. Author information . Affiliations . Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA, USA Avinoam Bar-Zion,?Atousa Nourmahnad,?Sangjin Yoo,?Marjorie T. Buss,?Dina Malounda,?Audrey Lee-Gosselin,?Margaret B. Swift,?David Maresca?&?Mikhail G. Shapiro Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA David R. Mittelstein Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA, USA Shirin Shivaei,?Robert C. Hurt?&?Mohamad H. Abedi Authors Avinoam Bar-Zion View author publications You can also search for this author in PubMed ? Google Scholar Atousa Nourmahnad View author publications You can also search for this author in PubMed ? Google Scholar David R. Mittelstein View author publications You can also search for this author in PubMed ? Google Scholar Shirin Shivaei View author publications You can also search for this author in PubMed ? Google Scholar Sangjin Yoo View author publications You can also search for this author in PubMed ? Google Scholar Marjorie T. Buss View author publications You can also search for this author in PubMed ? Google Scholar Robert C. Hurt View author publications You can also search for this author in PubMed ? Google Scholar Dina Malounda View author publications You can also search for this author in PubMed ? Google Scholar Mohamad H. Abedi View author publications You can also search for this author in PubMed ? Google Scholar Audrey Lee-Gosselin View author publications You can also search for this author in PubMed ? Google Scholar Margaret B. Swift View author publications You can also search for this author in PubMed ? Google Scholar David Maresca View author publications You can also search for this author in PubMed ? Google Scholar Mikhail G. Shapiro View author publications You can also search for this author in PubMed ? Google Scholar Contributions . A.B.-Z. and M.G.S. conceived the study. A.B.-Z., A.N., D. Maresca, D.R.M., S.Y. and S.S. designed, planned and conducted the in vitro experiments. A.B.-Z., A.N., M.T.B., R.C.H., A.L.-G. and M.B.S. designed, planned and conducted in vivo experiments. A.B.-Z. edited the gene circuits with the guidance of M.H.A. A.B.-Z., A.N., D.R.M., S.Y. and D. Maresca analysed the data. D. Malounda prepared the purified GVs. All the authors discussed the results. A.B.-Z., A.N. and M.G.S wrote the manuscript with input from all the authors. All the authors have given their approval for the final version of the manuscript. M.G.S. supervised the research. Corresponding author . Correspondence to Mikhail G. Shapiro . Ethics declarations . Competing interests . The California Institute of Technology has filed a patent application related to this manuscript. The authors have no other competing interests. Additional information . Peer review information Nature Nanotechnology thanks Mark Borden and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Extended data . Extended Data Fig. 1 GVs attenuate ultrasound at high concentrations. . ( a ) Illustration of the sample chamber and setup as seen in the images. ( b ) B-mode images of purified Ana GVs in different concentrations showing acoustic shadowing at high concentrations. Scale bar, 3?mm. ( c ) Average broadband emissions measured as a function of GV concentration. GVs were insonated with a single 30-cycle pulse with PNP?=?1.0?MPa (n?=?5). Extended Data Fig. 2 GV-seeded cavitation at 3?MHz requires higher pressure levels. . ( a ) Broadband signals recorded from GVs (0.3?nM) and BSA (matched in mg/mL to GVs concentration) insonated at 3?MHz. Broadband signal increased with pressure and was significantly higher for GV samples at PNP≥0.5?MPa (p?higher pressure levels, n?=?8 independent samples). ( b ) Comparison between broadband signals from GVs insonated with 670?kHz and 3?MHz pulses (n?=?16 and n?=?8 independent samples, respectively). Error bars, mean ± s.e.m (a-b). Extended Data Fig. 3 High frame rate optical imaging of GV collapse and bubble cavitation. . High-speed camera frames (left to right then top to bottom) of GV collapse and cavitation (200?ns between each frame, 31×31?μm field of view), focusing on a single bubble. Initial black spots, which correspond to intact GVs, first disappear due to collapse, liberating gas that coalesces into a cavitating bubble in this region. GVs were insonated with a single 30-cycle pulse with PNP?of?1.4?MPa and a central frequency of 670?kHz. Extended Data Fig. 4 Simulation of free bubble dissolution. . The kinetics of gas bubble dissolution were calculated based on the modified-EP (Epstein and Plesset) equation, following the analysis in [25]: \(- \frac{{{{{\mathbf{dr}}}}}}{{{{{\mathbf{dt}}}}}} = \frac{{{{\mathbf{L}}}}}{{{{{\mathbf{r}}}}/{{{\mathbf{D}}}}_{{{\mathbf{w}}}}}}\left( {\frac{{1 + 2\sigma /{{{\mathbf{P}}}}_a{{{\mathbf{r}}}} - {{{\mathbf{f}}}}}}{{1 + 3\sigma /4{{{\mathbf{P}}}}_a{{{\mathbf{r}}}}}}} \right)\) , where P a = 101.3?kPa is the hydrostatic pressure outside the bubble and r is the bubble radius. Here, L = 0.02 is Ostwald’s coefficient, D w = 2 × 10 ?5 cm 2 s ?1 is the gas diffusivity in water, σ = 72?mNm ?1 is the surface tension, and f?=?1 is the ratio between the gas concentration in the medium versus that at saturation. This model assumes a perfectly spherical geometry and neglects the potentially stabilizing effects of the nearby collapsed GV shell. However, it provides useful simulations that illustrate the time constants relevant to the process of GV cavitation. ( a ) Radius-time curves of free air-filled bubbles of different initial sizes. The gas liberated from a collapsed GV occupied the volume of a sphere with a radius of 89?nm under atmospheric conditions and no surface tension, and is expected to have an initial radius slightly larger than 20?nm when surface tension between air and water is assumed across its surface. The actual initial radius is expected to be somewhere between these two values, depending on the degree of stabilization by collapsed GV shells or other solution components. ( b ) Time before 50% volume reduction for free air-filled bubbles of different sizes. These time constants support the ability of nanobubble to survive the half-cycle between GV collapse (peak pressure) and peak rarefaction. In addition, they can guide the selection of the pulse repetition interval after the initial bubble growth. Extended Data Fig. 5 High frame rate optical imaging of GVs attached to tumour cells. . ( a ) GVs attached to U87 cells (0.4??s) are collapsed by the ultrasound wave (0.8??s). ( b ) Differential map comparing pre- and post-collapse images ( c ) Only after the collapse of the GVs are cavitation events seen (3.4??s and 9.2??s). The samples were insonated with a single 30-cycle pulse with PNP?=?1.4?MPa a central frequency of 670?kHz. The representative result in panels a-c belongs to one of the 8 repeats presented in Fig. 3g . Scale bars represent 20??m (a-c). Extended Data Fig. 6 Targeted and expressed GVs are frequently grouped in close proximity. . The close proximity between expressed or targeted GVs could play an important role in GV cavitation ( a ) SEM image of Ana GVs attached to U87 cell, forming large patches. ( b ) TEM image of GVs expressed in S. typhimurium showing a large cluster. Scale bar is 5?μm ( a ), 1?μm ( b ). SEM scans of GVs attached to cells and TEM scans of GV expressing S. typhimurium were repeated more than 10 times, all with similar results showing patches or clusters of GVs. Extended Data Fig. 7 Ultrasound images comparing GV expression in bacteria and mammalian cells. . ( a ) Ultrasound images of agarose phantoms containing S. typhimurium cells expressing GVs . The initial frame shows the echo from collapsing GVs (left, Peak US,), and the second one presents the residual signal from the cells after bubble dissolution (middle, Collapsed). The GV-specific signal, calculated as the difference between these two images, reveals high GV content in bacteria (right, Difference). ( b ) Ultrasound images of agarose phantoms containing GV-expressing HEK293 cells. The bacteria and mammalian cell samples were loaded into agarose phantoms at the same concentration as used in cavitation experiments. The volume of each well was 25??l and it contained either 2 × 10 5 trypsinized mARG-HEK cells or OD 600 ?=?1 (about 2 × 10 6 ) GV-expressing S. typhimurium cells. The combined volume of the mARG-HEK cells greatly exceeds the combined volume of the bacterial cells. However, the partial volume occupied by GVs in mammalian cell is much lower than in bacteria, resulting in lower GV-specific signal. Extended Data Fig. 8 Color deconvolution of H&E stains reveals effects of GV cavitation on surrounding tissue. . Histologic stains of liver samples were collected after systemic saline injection followed by FUS exposure (negative control, a-d ) or GV injection followed by FUS exposure ( e-h ). Color deconvolution was applied to H&E stains of liver sections ( a, e ) to obtain separate images of red blood cells ( b, f ) and tissue ( c, g ). The residual unmixed images are presented in ( d , h ). Necrotic regions in the H&E images ( e , zoom-in in i ) were found around hemorrhagic foci ( f , zoom-in in j ) in the livers of mice injected with GVs following FUS exposure. Scale bar is 2?mm ( d , h ), 200?μm ( j ). The representative results in this figure belong to one of the 5 repeats presented in Fig. 5g, h . Extended Data Fig. 9 Flow Cytometry Quantification. . Gating strategy for quantifying cell death in mArg-HEK cells, including the SSC-FSC gating of one sample from each population and its Zombie NIR fluorescence histogram. Cell death was quantified by gating the fraction of cells that emitted Zombie NIR fluorescence. The cutoff was the same for all samples. Supplementary information . Supplementary Information . Supplementary Tables 1–3. Reporting Summary . 41565_2021_971_MOESM3_ESM.avi . Supplementary Video 1 Representative high frame rate movie of GV attached to a Mylar plate. A series of 256 images showing cavitation nucleated by GVs attached to a Mylar plate were collected over 51.2??s at 5 million frames per second (fps) The movie is displayed at 5 fps, 1 million times slower than the real time. 41565_2021_971_MOESM4_ESM.avi . Supplementary Video 2 Representative high frame rate movie of GVs attached to tumour cells. A series of 256 images showing cavitation nucleated by GVs attached to U87 tumour cells were collected over 51.2??s at 5 million frames per second (fps) The movie is displayed at 5 fps, 1 million times slower than the real time. The swaying background is a result of the movement of the Mylar membrane at the bottom of the dish during of the ultrasound pulse. Rights and permissions . Reprints and Permissions About this article . Cite this article . Bar-Zion, A., Nourmahnad, A., Mittelstein, D.R. et al. Acoustically triggered mechanotherapy using genetically encoded gas vesicles. Nat. Nanotechnol. (2021). https://doi.org/10.1038/s41565-021-00971-8 Download citation Received : 12 July 2020 Accepted : 03 August 2021 Published : 27 September 2021 DOI : https://doi.org/10.1038/s41565-021-00971-8 Share this article . Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative .
From:
系统抽取主题     
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)  
(1)