Acoustically activatable liposomes as a translational nanotechnology for site-targeted drug delivery and noninvasive neuromodulation

Ultrasound-gated drug delivery leverages clinical therapeutic ultrasound systems in clinical use for myriad applications¹ to provide on-demand pharmacology targeting millimetre-sized brain and bodily regions of interest. For ultrasonic drug uncaging, following intravenous infusion of ultrasound-sensitive drug-loaded nanocarriers, focused ultrasound (FUS) applied to a region of interest is sufficient to induce drug uncaging from the nanocarriers while they circulate in the blood or are resident in the parenchyma² (Fig. 1a). The freed drug then enters the otherwise unperturbed parenchyma.

Several ultrasonic drug uncaging nanocarriers have been described, including perfluorocarbon-based systems in which the perfluorocarbon material may inertially cavitate under ultrasound, risking sonoporation or other forms of tissue injury^3,4. In addition, since perfluorocarbons are poor solvents generally, the drug is loaded in the particle shell limiting drug loading and risking nonspecific release in the body. While perfluorocarbon double emulsions and gas-containing liposomes load the drug in an aqueous or solvent core, perfluorocarbon double emulsions usually have large (>10 µm) particle sizes preventing safe intravenous administration⁵ and echogenic liposomes entrap gas yielding instability and risking damaging inertial cavitation under sonication⁶. Alternatively, heat-sensitive liposomes release drug with medium heating⁷. While effective, this heating may induce tissue heat shock and presents technical challenges in highly vascularized tissue, in organs that move with cardiorespiratory motion, and near bony interfaces, with corresponding high actuating device performance requirements⁷. Other liposomes incorporate a porphyrin or related molecule, producing reactive oxygen species under sonodynamic ultrasound inducing therapeutic release⁸. This requires continuous ultrasound for minutes necessitating similar technical constraints as heat-activated liposomes, and was validated using direct intraparenchymal injection. Another ultrasound-responsive liposome uses supercritical carbon dioxide incorporation⁹ with bubble generation under ultrasound, also risking inertial cavitation and instability, as with microbubbles.

We aimed for an ultrasonic drug uncaging vehicle that (1) enables high drug loading of varied drugs of interest, (2) can be activated following intravenous (not intraparenchymal) administration, (3) is formulated with validated pharmaceutical excipients and (4) can be activated with a lower duty cycle, shorter ultrasound protocol than heat or sonodynamic activation. We met this target by designing ultrasound-responsive drug nanocarriers that incorporate excipients that shift the acoustic characteristics of the nanocarrier liquid core to maximize its ultrasound responsiveness while maintaining stability without ultrasound. We have formulated these acoustically activatable liposomes (AALs) with validated pharmaceutical excipients and production processes and confirmed their ultrasonic drug uncaging performance in vivo in both the brain and body, confirming their high clinical translational potential.

Towards a next-generation ultrasonic drug uncaging nanocarrier, since our previous liquid perfluorocarbon-based formulation demonstrated no vaporization during effective uncaging^10,11, we hypothesized that we could maintain ultrasound sensitivity with improved stability and drug loading by replacing the core perfluorocarbon with an aqueous droplet that has sufficiently different acoustic characteristics versus its surroundings to enable an ultrasound interaction. In addition, if the internal buffer has a sufficiently different osmolarity versus its surroundings, then following the membrane permeability induced by the ultrasound interaction, this osmotic gradient would induce a fluid shift that accelerates drug release (Fig. 1a). Finally, by replacing perfluorocarbons with validated pharmaceutical excipients, we would increase clinical translatability.

To implement this design, we started with liposomes as they are common drug delivery vehicles formulated with validated pharmaceutical excipients¹². We incorporated sugars into the liposome internal core since they are GRAS (generally recognized as safe) excipients¹³ that may shift acoustic properties such as the speed of sound (c) and medium density (ρ), which determine the acoustic impedance (Z = ρc) and bulk modulus (K = ρc²)^14,15. Indeed, with moderate added sugar, the acoustic impedances of standard liposomal internal media shifted between 1.55 MPa s m⁻¹ and 1.65 MPa s m⁻¹ (Extended Data Table 2) compared with 0.0004 MPa s m⁻¹ for perfluorocarbon gas and 0.7–0.9 MPa s m⁻¹ for perfluorocarbon liquids¹⁶.

Loading the anaesthetic and antidepressant ketamine (Extended Data Table 1), compared with no added excipient, liposomes incorporating varied percentages of sucrose showed increased ultrasonic drug uncaging, with 5% added sucrose yielding ~40% uncaging and 10% added sucrose yielding ~60% uncaging, compared with a non-significant trend towards ~20% drug uncaging with no added sucrose (Fig. 1b). The lack of 100% drug release likely reflects our limiting of ultrasound parameters to those most relevant for in vivo application, floor and saturation effects of this assay, inhomogeneity of the applied ultrasound field, reloading of the freed drug before separation for quantification and heterogeneity of the liposomal particles’ response to ultrasound. Sucrose incorporation yielded no differences in physicochemical properties such as size, polydispersity, zeta potential, drug loading or unencapsulated drug (Extended Data Table 1). However, while 10% additional sucrose increased uncaging efficacy relative to 5% additional sucrose, it also increased drug release without ultrasound during 37 °C plasma incubation, indicating a trade-off of ultrasound responsiveness and stability with this scheme (Fig. 1c).

As sucrose incorporation shifts both acoustic parameters and osmolarity, to discern which contributed more to uncaging, we prepared liposomes with equiosmolar internal buffers but varying acoustic parameters (Extended Data Table 2). Drug-loaded liposomes with equiosmolar additional sodium chloride, glucose or sucrose had similar physicochemical characteristics with each other (Extended Data Table 1) yet showed differing drug uncaging under the same ultrasound protocol, in correlation to their acoustic impedance (Fig. 1d,e and Extended Data Table 2), indicating that differential acoustic characteristics, rather than another physical parameter, most contribute to ultrasound responsiveness. Hyperosmolarity of the internal medium may contribute to a baseline leak or instability (Fig. 1c) or accelerates drug release by driving water influx following ultrasound-increased membrane permeability. Going forward, 5% added sucrose was chosen for its balance of ultrasound responsiveness and formulation stability.

With 5% added sucrose, in vitro dose–response relationships were noted of drug uncaging with ultrasound parameters such as in situ peak negative pressure and mechanical index (MI; Fig. 1f). There was increased uncaging at 37 °C versus room temperature (RT) (~22 °C; Fig. 1f), although the response at RT indicates that medium heating is not necessary for uncaging performance. In addition, at these intensities and duty cycles, ultrasound-induced medium heating is at most +0.1–0.2 °C per the bio-heat equation¹⁷, arguing against a thermal mechanism. Indeed, when varying ultrasound frequency, we observed that uncaging scaled with MI more than the pressure (Extended Data Fig. 1a,b), indicating that radiation force and heating do not substantially contribute to the uncaging mechanism, as they require lower or similar intensities at higher frequencies. Instead, a mechanical (oscillatory or cavitation) mechanism appears more likely—albeit with low risk of damaging inertial cavitation as the liposomes do not contain gas. When recording acoustic backscatter and emissions during sonication, we readily observed broadband signal offset, ultraharmonics and band widening with microbubbles and did not see these signals with the liposome internal buffer or liposomes versus saline across MI = 0.6–3.0 (Extended Data Fig. 1c). Instead, with higher sonication pressures, the liposome internal medium and liposomes showed differential scatter compared with saline, consistent with a reflection from the flow tubing containing these media of different acoustic impedances (Extended Data Table 2) and a high concentration of scattering particles for the liposomes. Aside from this nonspecifically increased scatter, we observed no specific signs of inertial cavitation (ultraharmonics, subharmonics, band widening and so on) over saline of these liposomes or the liposome internal buffer for sonication up to 1.5 MPa, well above the pressure needed for effective drug release and what would be used in vivo. To investigate potential cavitation-related bioeffects in vivo, commercial microbubbles or AALs were infused intravenously in rats with co-administered Evans Blue dye, and sonication was applied transcranially to the rodent brain. For liposomes, sonication that is more than sufficient for drug uncaging was used. For microbubbles, a lower intensity and duty cycle protocol typical for blood–brain barrier disruption experiments was used. While Evans Blue extravasation was readily observed with microbubble sonication, indicating blood–brain barrier disruption, no blood–brain barrier disruption or other cavitation-related tissue change was seen with liposomes (Extended Data Fig. 1d). Regarding heating, simulations indicate that no substantial tissue parenchymal heating would be observed and experimental observations demonstrated at most minimal heating at the skull bony interface with an ultrasound protocol sufficient for uncaging (Extended Data Fig. 3d), further supporting that a thermal uncaging mechanism is not necessary.

Importantly, this formulation shows stability for both drug loading and ultrasonic uncaging across months of refrigerated storage (Fig. 1g). By electron microscopy, the formulation particles show the expected spherical morphology (Fig. 1h) and decreased in size with sonication, cohering with ultrasound-induced internal content leak (Fig. 1h). Importantly, no gas bubbles or voids were noted within the liposomes with or without sonication. Effective uncaging was seen at body temperature with 250 kHz ultrasound and an in situ peak negative pressure of 0.9 MPa, MI = 1.8, within FDA and iTRUSST guidelines for safe ultrasound application^18,19.

To explore this platform’s generalizability, we prepared liposomes with or without 5% additional sucrose and loaded with lidocaine, bupivacaine and ropivacaine: local anaesthetics that would enable noninvasive peripheral neuromodulation (complementing the central neuromodulation offered by ketamine uncaging) and provide a range of chemical features (logP and pKa) that may differentially affect loading. The ammonium sulfate-based method that we used allows high drug loading of amphiphilic weak basic drugs, particularly those of appropriate size, polarity and logP of their nonionized species to permit passive transit through a lipid bilayer, and of pKa so that an acidic environment yields protonation of the drug to trap it internally¹². Notably, these features are similar to those that permit passive transport across the blood–brain/nerve barrier. This ultrasonic uncaging strategy may work for any therapeutic compound that may be stably loaded into liposomes, whether loading is active with this or another chemical battery, or if the loading is passive. These local anaesthetic-loaded liposomes showed similar size, polydispersity and zeta potential as the ketamine-loaded liposomes (Extended Data Table 1). They showed differential drug loading reflecting their chemical characteristics (logP and pKa) as they relate to the ammonium sulfate active loading mechanism (Extended Data Table 1). Increased uncaging was observed with added sucrose internally (Fig. 1i–k), confirming the generalizability of this design for ultrasound-responsive nanocarriers. Each liposome showed the expected spherical morphology, with a liquid core with no observable gas or voids (Extended Data Fig. 1e).

For characterizing AAL performance in vivo, we normalized dose by drug mg kg⁻¹ between unencapsulated and liposomal formulations as the safety profile, pharmacokinetics and pharmacodistribution of drugs are usually reported as functions of total dose rather than a target site concentration. Following a dose-matched intravenous bolus, there were differential blood pharmacokinetics between AAL and unencapsulated drug formulations (Extended Data Fig. 4a and Extended Data Table 3) and higher drug metabolite blood levels with the AAL formulation versus unencapsulated drug. At 1 h post administration, solid organs showed increased accumulation of the drug with the unencapsulated formulation compared with the AAL (Extended Data Fig. 4b). The combination of higher blood drug metabolite concentration yet lower solid organ unmetabolized drug accumulation with the AAL formulation suggests that some clearance of the AAL formulation is likely due to sequestration and metabolism of the whole particle by organs such as the liver. Further, this end organ sequestration and metabolism could contribute to nonspecific leak of the unmetabolized drug and may be a less present mechanism in humans versus rodents given the lower human weight-normalized hepatic blood flow and mass.

To assess the pharmacodistribution achieved with AAL ultrasonic drug uncaging, in adult rats, we used solid-phase microextraction (SPME) following unencapsulated drug or liposome infusion, with or without ultrasound application to the brain²⁰. Ultrasound (or sham) was applied to a frontal region unilaterally during intravenous infusion of either unencapsulated ketamine HCl (KetHCl) or ketamine-loaded liposomes (Fig. 2a). After treatment, we stereotactically sampled both the sonicated brain and a contralateral control region 5 mm apart. Craniectomies were performed to enable brain SPME sampling; for non-SPME experiments (Figs. 4 and 5) sonication was completed fully transcranially. The transducer lateral FWHM (full-width at half-maximum) is 7.4 mm (Extended Data Fig. 2) and the axial focal length (50%, FWHM) is 18.6 mm, which covers the full dorsal–ventral axis of the rat brain allowing control sampling only lateral to the focus and not axially. Varying ultrasound parameters or total AAL dose demonstrated dose–response relationships of the amount of drug delivered (Fig. 2b,c). No similar dose–response relationship was observed for control/sham site drug concentration, reflecting a floor effect of SPME (Supplementary Fig. 1) that likely overestimates the true in vivo nonspecific drug release rate.

Blood samples showed higher blood ketamine levels with each liposomal formulation compared with unencapsulated drug, confirming our bolus pharmacokinetics (Fig. 2d and Extended Data Fig. 4a). Despite this higher blood ketamine concentration with liposomes, the ketamine brain levels with sham or in the control region were less than half that of dose-matched free ketamine (Fig. 2e). While ultrasound did not alter the pharmacodistribution following free ketamine infusion, ultrasound yielded higher brain ketamine levels with the liposomes compared with sham or the contralateral non-sonicated region, confirming the performance of this technique and a <5 mm spatial resolution (Fig. 2a,e), just over half the FWHM of this 250 kHz transducer (Extended Data Fig. 2), confirming that AAL-mediated drug delivery is limited spatially by the applied ultrasound field. Furthermore, higher brain ketamine levels were seen with sonication with liposomes with additional sucrose compared with liposomes without added sucrose, confirming that this simple manipulation increases the ultrasound sensitivity both in vitro and in vivo. Despite higher uncaging in vitro, no difference was seen in vivo between liposomes made with 10% versus 5% added sucrose, likely reflecting the instability of the 10% added sucrose formulation (Fig. 1b,c). Histology showed no brain parenchymal injury with AAL-mediated ultrasonic drug uncaging (Fig. 2f–j), arguing against an unsafe degree of inertial cavitation (Extended Data Fig. 1) or tissue heating (Extended Data Fig. 3c,d) with these AAL and ultrasound doses that are effective for spatially localized ultrasound targeted drug delivery in vivo.

We next enabled site-specific pharmacologic noninvasive neuromodulation with ketamine uncaging²¹. We selected the 5% sucrose AAL (S5-KL) as an optimum for uncaging performance in vitro and in vivo (Figs. 1 and 2) and denoted it SonoKet. We targeted ketamine delivery to brain regions important for mediating its affective (medial prefrontal cortex (mPFC))²² and dissociative (retrosplenial cortex (RsC))²³ actions (Fig. 3a and Extended Data Fig. 3a,b). During 1.5 mg kg⁻¹ SonoKet or free KetHCl infusion for 5 min, ultrasound or sham was applied to either the mPFC or RsC during the last 2.5 min of infusion.

While ultrasound did not affect the pharmacodistribution of unencapsulated KetHCl (Fig. 3b,c), with SonoKet there was selectively increased delivery in the sonicated brain regions (mPFC, 2.1×; RsC, 2.0×) with no difference in the non-sonicated brain region versus sham, and no difference of delivery or clearance between mPFC and RsC (Fig. 3b,d). With SonoKet uncaging, there was higher ketamine delivery versus dose-matched free KetHCl (mPFC, 1.2×; RsC, 1.5×; Fig. 3b). Sonicated regions consistently had twofold or higher delivery than non-sonicated regions, providing a suitable therapeutic window, noting that this delivery rate without ultrasound is likely an overestimate given an SPME floor effect (Supplementary Fig. 1). Importantly, each experiment showed a similar clearance profile, with ketamine and its metabolites clearing by 15–20 min post-treatment (Fig. 3c–e, Extended Data Fig. 4 and Extended Data Tables 3 and 4). While ketamine’s metabolites have been proposed as pivotal to its actions, their brain levels were negligible compared with unmetabolized ketamine, with no differences of metabolite distribution or clearance with free or SonoKet-encapsulated ketamine (Fig. 3c–e, Extended Data Fig. 4 and Extended Data Tables 3 and 4). Importantly, the biodistribution analyses demonstrate lower solid organ ketamine uptake with SonoKet administration versus dose-matched unencapsulated ketamine (Extended Data Fig. 4b), confirming that ultrasonic ketamine uncaging should lower systemic side effects while selectively delivering ketamine to these functionally significant brain targets, with otherwise similar pharmacokinetics compared with unencapsulated ketamine.

To evaluate the functional efficacy of localized ketamine delivery in the context of acute stress, we measured electrocorticography (ECoG) over the mPFC and RsC while awake subjects were in a restraint that induces acute stress in rodents²⁴ (Fig. 6a). Different ECoG oscillation patterns have been associated with different ketamine actions, with frontal gamma band activity being a biomarker of affective action²⁵ and retrosplenial slower band activity correlated to dissociation^23,26. ECoG potentials were processed to determine the power spectral density across 1–55 Hz before, during and after treatment following similar analyses of intracranial recordings from humans receiving ketamine²⁷.

We first characterized the dose–response relationship for unencapsulated KetHCl and found that there was dose-dependent increased gamma (25–55 Hz) power for both the mPFC and RsC (Extended Data Fig. 5 and Supplementary Fig. 1) and decreased slower band power for the RsC (Supplementary Fig. 1). The sustained gamma response for the 45 min recordings despite the relatively rapid ketamine clearance (Fig. 3c,e and Extended Data Fig. 4) was expected given previous characterizations²⁸.

During antidepressant ketamine treatment, 0.5 mg kg⁻¹ of ketamine infused over 40 min results in plasma ketamine levels of ~200 ng ml⁻¹ in adult humans²⁹, similar to rat plasma ketamine levels seen with 0.75 mg kg⁻¹ of SonoKet infused over 5 min (Figs. 2 and 3, Extended Data Fig. 4 and Extended Data Tables 3 and 4). We therefore selected 0.75 mg kg⁻¹ as the dose going forward. In the awake restraint that can act as an acute stressor²⁴ (Fig. 4a, Extended Data Fig. 5 and Supplementary Fig. 1), rats were administered SonoKet with and without sonication, dose-matched KetHCl or saline with sonication in a random ordering with at least 1 week for washout between experiments. No significant ECoG change was seen with SonoKet infusion without ultrasound application versus saline. However, with mPFC ketamine uncaging, gamma power increases were observed to a trend-wise greater degree and similar spectral pattern compared with dose-matched free ketamine (Fig. 4b–d and Extended Data Fig. 5). By contrast, with RsC ketamine uncaging, acute increases in principally slower theta (4–8 Hz), alpha (8–15 Hz) and beta (15–25 Hz) bands were observed with then sustained activities in beta and low gamma bands, in a distinct spectral pattern to both dose-matched unencapsulated ketamine and mPFC uncaging (Fig. 4b–d, Extended Data Fig. 5 and Supplementary Figs. 3 and 4). While RsC SonoKet uncaging showed apparent delta power increases (Fig. 4b,c and Supplementary Fig. 3), cohering with previous characterizations^23,27, this may have been driven by changes in the saline condition, suggesting a nonspecific effect of this experimental paradigm. Site specificity was supported by sonication directed to the mPFC while recording from the RsC and vice versa sonication directed to the RsC while recording from the mPFC (Extended Data Fig. 6), as each yielded a different band pattern compared with recording and sonicating both at mPFC or recording and sonicating both at RsC. Sonication at mPFC while recording from RsC showed less gamma activity that was less sustained than recording and sonicating both at mPFC, and no appreciable low-frequency induction and less prominent beta induction compared with recording and sonicating both at RsC. Meanwhile, sonicating at RsC and recording from mPFC showed less high-frequency beta/gamma induction compared with sonicating and recording both from mPFC, and showed less prominent low-frequency induction, with a longer-term decrease in lower band power compared with sonicating and recording both from RsC. Similarity of these patterns may reflect volume transmission of EEG/ECoG signals³⁰ or the known functional connectivity between these regions³¹. These data confirm the functional site specificity of AAL uncaging, in correlation to its site-specific pharmacodistribution (Figs. 2 and 3). Overall, this differential functional electrophysiologic response to ketamine uncaging in the mPFC versus RsC, despite similar levels of ketamine delivery to each brain region, and to greater effect at either site compared with dose-matched unencapsulated ketamine (Fig. 3), confirms the site-specific physiologic action underlying ketamine’s multiple functional effects.

For evaluating the behavioural effects of site-targeted ketamine uncaging, despite variability of rodent behavioural responses to subanaesthetic ketamine that may not have human parallels^32,33, certain rodent behaviours are indicative of relevant circuit activities. In rodents, ketamine induces locomotion correlated to mesolimbic dopaminergic circuit modulation³³ and stress and affect-related effects correlated to limbic circuit activity²². To evaluate both functional domains, we used the open-field test and quantified parameters related to locomotion and stress/affective processing. Rats were placed in a restraint while awake to induce a brief, acute stress (Fig. 5a), received either saline, unencapsulated KetHCl or SonoKet with sonication directed to either the mPFC or RsC, and then placed in an open field for 20 min. There were no effects on locomotion (Fig. 5b–f and Supplementary Figs. 5 and 6), cohering with previous results demonstrating limited contribution of frontal circuits to ketamine’s locomotor effects³⁴. By contrast, mPFC ketamine uncaging yielded more time in the field centre versus dose-matched unencapsulated ketamine or saline (Fig. 5b,g,h; multiple comparison corrected P values, saline versus SonoKet P = 0.037, Hedges g = 1.24; KetHCl versus SonoKet P = 0.028, Hedges g = 1.35), with no differences seen with RsC uncaging. The lack of locomotion induction coheres with previous similar dose studies and indicates no mesolimbic dopaminergic modulation³⁵. Meanwhile, increased centre time—interpreted as a lower stress, anxiolytic-like phenotype—specifically with mPFC ketamine uncaging coheres with the identification of the mPFC as critical to the affective efficacy of ketamine^22,36.

After confirming that ketamine uncaging yields central neuromodulation (Figs. 3–5), we implemented uncaging-mediated peripheral neuromodulation by targeting local anaesthetics to a peripheral nerve as a noninvasive version of nerve block procedures completed routinely for pain therapy and anaesthesia³⁷ (Fig. 1a). We uncaged an AAL loaded with the local anaesthetic ropivacaine to induce anaesthesia of the rat sciatic nerve. Adult rats were administered a 5 min intravenous infusion of either saline or a ropivacaine-loaded AAL (SonoRopi, previously denoted S5-RL; Fig. 1i). With sonication then applied to the sciatic nerve of one limb, anaesthesia to a mechanical (von Frey fibre) stimulus was observed only in the sonicated limb of the SonoRopi-administered animals, with this anaesthesia lasting >1 h following treatment (Fig. 6a–c). No such change was seen in the saline-treated animals or in the contralateral non-sonicated limb of the SonoRopi-treated animals (Fig. 6b,c). No electrocardiographic abnormalities or visible intolerances were noted with SonoRopi uncaging (Fig. 6d–f). Histology confirmed no parenchymal injury (Fig. 6g–k).

We have validated a vehicle for ultrasonic drug uncaging with high drug loading, minimal drug release without ultrasound and effective drug uncaging with pulsed low-intensity ultrasound both in vitro and in vivo for multiple drugs and bodily regions of interest. We have therefore proved a design scheme for ultrasound-responsive drug delivery nanotechnologies in which the acoustic properties of the drug-loaded nanocarrier core are shifted by adding excipients that increase interactions with applied ultrasound. In this implementation, the acoustic properties of the core most contribute to ultrasound responsiveness, as liposomes made with cores that were equiosmolar but had different acoustic properties showed differential uncaging performance in correlation to their acoustic properties (Fig. 1d,e). While we have used the GRAS excipient sucrose to maximize clinical translatability, an even further ultrasound-responsive vehicle could be achieved with a different excipient that further shifts the core acoustic characteristics. However, care must be taken that this does not destabilize the liposome, as we observed with 10% added sucrose (Fig. 1c). While an ammonium sulfate-based active loading battery was utilized, this strategy should be applicable to any liposome and loading scheme or to polymeric liquid-core nanocarriers.

While there was a nonzero rate of drug leak without ultrasound both in vitro (Fig. 1) and in vivo (Figs. 2 and 3), the observed in vivo ultrasonic uncaging effect sizes (Figs. 2–6) offer a sufficient therapeutic index for each drug tested and are commensurate with or exceed the demonstrated on- versus off-target delivery of benchmark systems. This baseline nonspecific leak potentially reflects the lack of internal precipitation seen with these liposomes (Figs. 1 and 2) or the metabolism of the AAL contributing to drug leak (Extended Data Fig. 4), and may be overestimated owing to a floor effect of the SPME assay that we used (Supplementary Fig. 1). Notably, this potential metabolic effect may be less present in humans given the lower hepatic blood flow and metabolism of humans compared with rodents. Future efforts will reduce this baseline leak while further maximizing ultrasound responsiveness by including materials that increase internal drug binding and further shift the liposome core acoustic parameters.

The pharmacologic central noninvasive neuromodulation demonstrated here underscores the recent interest in repurposing recreational psychotropic drugs such as ketamine for mental health applications³⁸. However, tempering that excitement are concerns regarding the abuse liability, ethics and safety of clinically utilizing potent hallucinogenic recreational drugs^39,40. To address these concerns, as an alternative to synthesizing new chemical entities³⁸, this study supports and provides the technological means for an alternative strategy to achieve selective activation of the therapeutic actions of drugs such as ketamine by targeting unmodified drug delivery to the brain regions most critical for the desired action. In principle, a similar approach could be used for myriad neuropsychiatric drugs that have potent therapeutic effects but dose-limiting side effects owing to action at off-target brain or bodily regions. As current approved protocols for drugs such as ketamine in psychiatric care require close clinical monitoring of the patient⁴¹, applying ultrasonic drug uncaging only minimally complicates workflow: adding ultrasound device operation but potentially saving on patient monitoring if dissociative, sedative and sympathomimetic properties are indeed minimized. In addition, as demonstrated for peripheral neuromodulation via ropivacaine uncaging, this approach more generally maximizes a drug’s therapeutic effect over its side effects. These results will be validated in future clinical studies of honing the action of potent neurally active drugs through regionally targeted drug delivery.

Materials

All chemicals and reagents were of the highest purity grade. Lipoid: hydrogenated soy phosphatidylcholine (HSPC) (catalogue number Lipoid S PC-3) and [N-(methoxypolyethyleneglycol-2000)-1,2-distearoyl-sn-glycero-3-phosphoethanolamine, sodium salt] (DSPE-PEG 2000) (catalogue number PE18:0/18:0, PEG 2000). Evonik: cholesterol (PhytoChol Inject HU). KetHCl injectable solution (100 mg ml⁻¹; Dechra Pharmaceuticals, procured through Stanford University Environmental Health & Safety). Hiemdia: ammonium sulfate (catalogue number PCT0003). Fisher Scientific: absolute ethanol (200 proof) (catalogue number BP2818-100), sucrose (catalogue number S5-500), HPLC and LC/MS grade water, methanol, acetonitrile, 2-propanol and formic acid. Sigma-Aldrich: HEPES buffer solution (catalogue number 83264-500ML-F) and l-histidine monohydrochloride monohydrate (catalogue number H8125). Repligen: TFF filters (C02-S05U-05-N (SN 20020493-03/21-057). Lampire Biological Laboratories: male canine plasma (catalogue number 7302009). Cytiva: PD-10 column Sephadex G-25 M (catalogue number 17085101) and Sephacryl S-500 (catalogue number 1706130). Milli-Q water was used to prepare all buffers. Millipore Sigma: Cerilliant certified standard solutions of ketamine hydrochloride, ketamine-D₄ hydrochloride, norketamine hydrochloride, norketamine-D₄ hydrochloride and hydroxynorketamine hydrochloride; Supel BioSPME 96-Pin Devices (product number 59680-U).

AAL synthesis and characterization

Liposome production

Initially, large multilamellar vesicles were prepared by dissolving lipid components (HSPC/DSPE-PEG 2000/cholesterol 52.8:42.3:4.8 molar ratio) in heated ethanol and then diluting the mixture to 10% ethanol with 250 mM ammonium sulfate alone or with 5–10% (by weight) sucrose, 5% glucose or 73 mM NaCl depending on the experiment. The solution was extruded (Avestin LF-50 with 200 nm pore polycarbonate Whatman filter at 65–70 °C) ten times to generate unilamellar liposomes. Samples were processed with tangential flow filtration (5 × 5-fold dilution/reconcentration) against a buffer of 10 mM HEPES, 145 mM NaCl (pH 7.4) with 0%, 5% or 10% sucrose depending on the internal buffer osmolarity, to generate a transmembrane ammonium gradient. For loading, the drugs were added to 10-fold diluted liposome at final 1 mg ml⁻¹ concentration and heated to 55 °C for 1.5 h. To remove the unencapsulated drug, repeated TFF (4 × 5-fold dilution/reconcentration) was performed against a buffer of 10 mM histidine (pH 7.4) with 10% or 15% sucrose. Finally, the samples were sterilized using a 220 nm PVDF filter and stored at 4 °C.

Physiochemical characterization

The Z-average diameter, polydispersity index (PDI) and zeta potential of liposomes were measured by dynamic light scattering (DLS) with a Malvern Zetasizer Nano ZS90 (Malvern). Cryo-electron microscopy (cryo-EM) was used for structural analysis. Drug loading efficiency was measured by destructing the liposome using methanol followed by HPLC to quantify the total drug presented in the sample and reported as DL in mg ml⁻¹ concentration. For the free drug measurement, initially unencapsulated drug was separated from liposome using PD10 column followed by HPLC to quantify the percent free drug using the following formula: FD(%) = (drug in free fraction/sum of drug in free and liposome fractions) × 100 (method details in Supplementary Information).

In vitro ultrasonic drug uncaging

PCR tube containing liposomes (1:4 diluted in canine plasma) was placed in a custom 3D-printed holder held at the focus of a 250 kHz or 650 kHz hydrophone-calibrated FUS transducer and degassed water of either 25 °C or 37 °C was used for coupling. For the flow uncaging, the focal zone of a 250 kHz FUS transducer was aligned with tubing through which liposomes, diluted 30-fold with canine plasma, were flowed at a rate of 130 μl min⁻¹. An ultrasound (60 s, 25% duty cycle, 5 Hz pulse repetition frequency (PRF), varying peak pressure) was applied and a total volume of 1 ml was collected after sonication for each condition. Liposomes after uncaging were separated from unencapsulated free drug by a homemade Sephacryl S-500 column, with PBS as the elution buffer, collecting the first 5.5 ml of elute as the liposome fraction and the next 8 ml as the unencapsulated drug fraction. Drug concentration in each elute was quantified by HPLC. The % drug uncaging was calculated using the following formula:

In vitro acoustic emissions recordings during sonication of an AAL and its internal buffer

To characterize acoustic emissions during ultrasonic uncaging, the magnitude of the received echo spectrum was measured using a flow chamber set-up. In brief, the focal zone of a 250 kHz FUS transducer was aligned with a tubing segment through which liposomes flowed at a constant rate of 130 μl min⁻¹. Ultrasound was delivered with a 0.1% duty cycle and a 1 Hz pulse repetition frequency, with peak pressures varying across experimental conditions. A hydrophone (ONDA) was positioned 2.5 cm from the tubing at a 90° angle relative to the transducer axis to capture acoustic backscatter during sonication. The echo signals were collected by PicoScope and subjected to fast Fourier transform analysis. The resulting spectra were plotted to assess the frequency-dependent magnitude of the acoustic emissions.

Speed of sound measurement

A clean 20-gallon fish tank was filled with deionized water and degassed overnight. A 650 kHz 30 mm aperture f1.0 FUS transducer (Sonic Concepts), a 35.6 cm long PVC cylinder with a 1.5 inch diameter, and a hydrophone were placed in a row underwater. Both devices were linked to an oscilloscope (Keysight Technologies) to view the time of flight between the transducer and the hydrophone. The PVC pipe was wrapped in an ultrasound-compatible plastic probe cover and sealed with O-rings to create a separate internal fluid compartment. First, the pipe was loaded with 37 °C deionized water, and the ultrasound pulse arrival time was used as a reference, along with the known speed of sound in deionized water at 37 °C, for subsequent measurements. Each buffer was sequentially loaded after heating, and the difference in pulse arrival time was recorded. A temperature measurement after each run confirmed minimal heat loss. The differences in pulse arrival time owing to different speeds within the length of the pipe were translated into speeds of sound of the various buffers. Finally, measurements of density at 37 °C were performed by weighing 10 ml of each buffer using a balance.

Animals

All animal experiments were carried out in accordance with the Stanford IACUC and Administrative Panel on Laboratory Animal Care (APLAC). Male Long–Evans rats (Charles River Laboratories; Envigo) and male Sprague–Dawley rats (Charles River Laboratories) were used in all in vivo studies with ketamine and ropivacaine, respectively. All rats were between 7 and 10 weeks old with body weight 250–450 g. Isoflurane was used to anaesthetized animals for surgical and terminal procedures and briefly for awake animal experimental set-up.

Animal treatment preparation

Ketamine hydrochloride (Dechra Veterinary Products) was diluted in 0.9% sterile saline to obtain 1 mg ml⁻¹ solutions. All treatments were administered via intravenous tail vein infusion continuously over 5 min with an infusion pump (World Precision Instruments).

In vivo ultrasound protocol and blood–brain barrier (BBB) opening verification

A custom 250 kHz FUS transducer (designed and constructed by R. Watkins, Stanford University) powered by an amplifier (240 L, E&I) was utilized in all in vivo experiments. Calibration of voltages was conducted using a hydrophone (ONDA). Skull attenuation was accounted for and calculated based on weight⁴⁴ to achieve the desired in situ pressure. In all experiments where ultrasound was applied, continuous FUS was applied to either the mPFC or RsC after 2.5 min of drug infusion for 2.5 min (250 kHz, 25% duty cycle, 1.1 MPa estimated peak in situ pressure, 50 ms pulse width). For BBB opening verification, 4 ml kg⁻¹ of 2% Evans Blue dye was administered via the tail vein immediately after FUS treatment and ketamine-loaded liposome infusion. The rat was then anaesthetized and perfused with PBS before the brain was extracted. As a positive control for BBB opening, Definity microbubble infusion was used instead of ketamine-loaded liposomes, with ultrasound applied at 0.5 MPa with 1% duty cycle and 1 Hz PRF for 3 min.

Ultrasound simulations

To simulate the pressure-field distributions during FUS treatment, a male Long–Evans rat (weight 453 g) was imaged on a Quantum GX micro-CT. The obtained micro-CT image was 1,024 × 1,024 × 553 with a cubic voxel size of 0.086 mm. The images were resampled linearly to a cubic voxel size of 0.34 mm. The bone, soft tissue and water were isolated based on their Hounsfield units (1,200 HU for bone/soft tissue and 930 HU for soft tissue/water threshold). Density and sound speed were linearly interpolated in each region using hounsfield2density, a predefined function through the k-Wave MATLAB toolbox⁴⁵. The region surrounding the animal was defined as water. The transducer was defined as a bowl with a diameter and radius of curvature of 100 mm. At 250 kHz centre frequency, the points per wavelength was 17.44 in water and Courant–Friedrichs–Lewy stability criterion of 0.1 leading to a time step of 22.8 ns. The simulation was run for 85 µs, allowing the initial wave to travel to the length of the simulation grid (125 mm).

In vivo measurements of temperature change with ultrasound

To assess temperature changes during in vivo ultrasound-mediated drug uncaging, thermal measurements were taken adjacent to the skull at both the sonicated and contralateral brain sites. After dorsal scalp exposure, 2 mm burr holes were drilled into the skull to allow for insertion of a thermocouple probe. Animals underwent the in vivo ultrasound protocol listed above. Thermocouple probes were inserted 3 mm vertically into the brain parenchyma through the burr holes to measure temperature at three time points: immediately before ultrasound exposure, immediately after and 1 min post-exposure.

SPME

Rats were fixed into a stereotaxic frame, administered 2 ml of saline subcutaneously and kept on a heating pad at 37 °C. After dorsal scalp exposure, 2 mm burr holes were drilled into the skull for SPME pin insertion (relative to bregma, −5 mm A/P and ±2.5 mm M/L for Fig. 4; +3.2 mm A/P for mPFC, or −4 mm for RsC, both at +1 mm M/L to the right). A durotomy was performed with a 32 g needle. Burr holes were used only in the SPME and thermal probe experiments to permit passage of the probes for direct assessment of the brain immediately after treatment. For rat subjects that received ultrasound, the ultrasound transducer was positioned directly above the desired burr hole via a three-axis positioning system (ThorLabs), with a coupling cone and ultrasound gel for coupling. Rats received 1.5 mg kg⁻¹ of SonoKet, KetHCl or saline vehicle. Pins were loaded into a custom-designed, 3D-printed stereotaxic holder for precise positioning and sampling, inserted 3 mm ventrally into the brain via the burr holes (SPME absorptive medium centred at ~2.5 mm ventral) and left in contact with the rat brain tissue for 5 min (refs. ^46,47,48,49). Post-sampling, SPME pins were washed to remove residual blood, desorbed into 50 µl of MeOH/H₂O (9:1 v/v) solvent containing 1% formic acid and quantified by liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS; method details in Supplementary Information).

Blood pharmacokinetics

Blood samples from rats (n = 3 per group) that received 1.5 mg kg⁻¹ intravenous bolus or infused SonoKet or dose-matched KetHCl were collected via tail snipping into EDTA at different time points, and 3–4× cold acetonitrile was mixed to precipitate the plasma protein, which was then removed using a centrifuge. The collected supernatant was dried, reconstituted with 100 μl of MeOH/H₂O (9:1 v/v) solvent containing 1% formic acid and quantified by LC-MS/MS.

Biodistribution

Adult rats (n = 4 per group) received either free or liposomal ketamine as an intravenous bolus dose of 1.5 mg kg⁻¹. Rats were killed at 1 h from the time of administration and perfused with 1× PBS via transcardial perfusion to remove blood from the systemic circulation. The collected brain, liver, kidney, spleen, lung, heart and spinal cord were homogenized in equal-weight volume of 1× PBS. Similar to blood processing, 3–4× cold acetonitrile was mixed to precipitate the protein, which was then removed using a centrifuge. The collected supernatant was dried, reconstituted with 100 μl of MeOH/H₂O (9:1 v/v) solvent containing 1% formic acid and quantified by LC-MS/MS.

Awake-restraint electrophysiology recording and analysis

Surgical set-up

Animals were anaesthetized and positioned stereotaxically for implantation of custom ECoG electrodes. A midline incision was made over the scalp and five holes were drilled through the skull with stereotaxic guidance⁴³. All electrodes were positioned relative to bregma. mPFC electrodes were positioned at +1 mm A/P and +2 mm M/L to the right (positive electrode), and −6 mm A/P and ±2 mm M/L (ground and reference electrodes, respectively). To provide enough ultrasound clearance without sonicating the electrodes in experiments where the RsC was sonicated while recording from the mPFC, the ground and reference electrodes were both moved to −11 mm A/P and respectively ±2 mm M/L. Two stabilizing screws were implanted at −2 mm A/P and ±3 mm M/L for structural integrity. RsC electrodes for all recordings from the RsC were positioned at −5 mm A/P and +3 mm M/L to the right (positive electrode), and −11 mm A/P and ±3 mm M/L (ground and reference electrodes, respectively). Two stabilizing screws were implanted at −8 mm A/P and ±4 mm M/L for structural integrity. Electrodes were screwed into the skull without breaching the dura and dental cement applied to fix the electrodes in place. Animals were housed in separate cages afterwards and allowed for at least 7 days of recovery before recordings.

ECoG recording

At the beginning of each recording session, rats were briefly anaesthetized to be catheterized via the tail vein, placed in a thin, flexible plastic restraint cone (Amazon.com), and positioned in a custom head-restraining apparatus⁵⁰. Rats received oxygen via the nose cone to prevent hypoxia. Recordings were performed with an 8-Channel Cyton Biosensing Board^51,52 at a sampling frequency of 1,000 Hz. For subjects that received ultrasound, the transducer was positioned directly above either the mPFC (+3.2 mm A/P, −0.5 mm M/L) or the RsC (−4 mm A/P, 0 mm M/L) via the 3-axis positioning system and coupled with ultrasound gel. Implanted screws were utilized for precise targeting according to the rat brain atlas⁴³. Data acquisition began 25–30 min after the animal is in the restraint to allow for complete isoflurane clearance. A baseline acquisition (5 min) was recorded before starting the treatment. Rats received 0.75 mg kg⁻¹ of SonoKet, KetHCl or saline vehicle. Data were acquired continuously for 35 min following a 5 min baseline and a 5 min treatment infusion protocol, for a total recording of 45 min.

Data analysis

Electrophysiological data were filtered using MNE-Python with a band-pass filter with a low cut-off of 1 Hz and a high cut-off of 200 Hz (ref. ⁵³). The data were then denoised by decomposing into 5 levels of Daubechies 8 wavelets, zeroing outlier coefficients and then reconstructing the modified data⁵⁴. The first 45 min of the recording was defined as a single epoch and time–frequency representation was computed with a multitaper technique and adjusted to baseline (initial 0–5 min of recording) with percent change from baseline using MNE-Python. For band power trace plots, the time–frequency representation was averaged within frequency band cut-offs (delta, 1–4 Hz; theta, 4–8 Hz; alpha, 8–15 Hz; beta, 15–25 Hz; gamma, 25–55 Hz) and a moving average of the band power was computed with a 2-min-long convolving window²⁷. One recording within the RsC 1 mg kg⁻¹ KetHCl condition was excluded owing to excessive artefacts that could not be corrected by denoising. The area under the curve (AUC) was calculated by integrating the percent of power change within each frequency band as indicated above in time bins indicated as time of sonication (7.5–10 min), time immediately after treatment (10–25 min) and time after clearance (25–45 min) and depicted in arbitrary units.

Behavioural open-field analysis

Rats were surgically implanted with a targeting screw and given at least 7 days to recover before behavioural tests. To acclimate to the experimenter and reduce stress, rats were handled 3 days before recording. All behavioural tests were performed in an environmentally controlled room. Open-field locomotor activity was recorded from above in a custom-built white Plexiglas apparatus (90 cm × 90 cm × 40 cm). Animals were placed at the centre of the field and randomized for the treatment group.

Rats were placed under isoflurane briefly to be catheterized via the tail vein and loaded into the awake restraint⁵⁰. They were administered oxygen via the nose cone within the restraint to prevent hypoxia. After ensuring that the rats are fully awake before starting treatment, rats received an intravenous infusion of 0.75 mg kg⁻¹ SonoKet, KetHCl or saline vehicle for 5 min, followed by the in vivo ultrasound protocol. After treatment, rats were immediately placed at the centre of the arena, where they were allowed to freely explore for 20 min while locomotor activity was recorded.

Before analysis, videos were first reencoded for format compatibility and clipped to 20 min starting at 5 s after the initialization of the recording with FFmpeg. ToxTrac was then used to track animal position with tracking settings matching the ToxId algorithm and detection settings adjusted ad hoc^55,56. Custom Python scripts were then used to quantify and plot cumulative frame-to-frame distance travelled and time spent in the centre of the open field, which was defined as a concentric square field with half the width of the full square field.

Ropivacaine in vivo experiments

Mechanical sensitivity

Rats were acclimated to a raised stainless steel mesh table for 30 min. Baseline paw withdrawal responses were obtained using monofilaments and the von Frey percent response method (10 pokes per filament 1–26 g)^57,58. On the basis of baseline responses, the 26 g monofilament was selected to evaluate the effects of local uncaging ropivacaine. On the procedure day, rats were anaesthetized during the experiment. Saline vehicle or ropivacaine-loaded liposomes were administered, followed by application of ultrasound using a dorsal approach to target the sciatic nerve. Paw withdrawal response was then evaluated using the 26 g monofilament and the von Frey percent response method 30 min, 1 h and 4 h post-treatment.

Electrocardiographic analysis

Under anaesthesia, three alligator electrodes were placed on the rats in limb lead II position: the negative electrode was placed on the front paws and the positive electrode on the left hind paw. ECG recordings were taken at baseline, immediately and 5 min after injection of either free ropivacaine-HCl or ropivacaine-loaded AAL using a Sy-W002 Vet 3 Channel ECG Machine (Sunny Medical Equipment Limited). The following changes in the ECG pattern were assayed: heart rate (bpm), duration of QRS complex and QT interval (corrected for given heart rate in each animal), and periodic repolarization dynamics (PRD).

Histological safety analysis

Drug administration and tissue collection

Rats were administered an intravenous infusion of 0.75 mg kg⁻¹ SonoKet, dose-matched KetHCl or vehicle over 5 min (n = 3 per group), followed by ultrasound application (−5 mm A/P and +2.5 mm M/L to the right relative to bregma). After 72 h post-treatment, the animals were anaesthetized and transcardially perfused with 1× PBS followed by 4% paraformaldehyde (PFA) diluted in PBS. Brains were extracted and stored in 4% PFA for 24 h, 15% sucrose for the next 48 h and 30% sucrose for the last 48 h. They were washed in PBS and frozen in embedding medium. Coronal brain sections (30 µm thick) were cut using a CM1800 cryostat (Leica Microsystems) and were stored in 30% sucrose and 30% ethylene glycol in 0.1 M PB at −20 °C until processed for immunohistochemistry.

Microscopy and image analysis

Every 12th section (360 μm apart) was stained with haematoxylin and eosin (H&E; Vector Laboratories), Fluoro-Jade C (Biosensis), recombinant anti-GFAP antibody (ab33922, Abcam) and recombinant IBA-1 (ab178846, Abcam) to evaluate for parenchymal damage, neuronal degeneration, and astrocytic or microglial activation, respectively. Alexa Fluor 488 secondary antibody (Thermo Fisher Scientific) was used after primary anti-GFAP incubation. Cy5 secondary antibody (Thermo Fisher Scientific) was used after primary anti-IBA1 incubation. Tissue sections were free-float mounted on microscope glass slides (Fisher). All histology images were collected with a fluorescence microscope (BZ-X800, Keyence). For quantifiable histological markers (Fluoro-Jade C, IBA-1⁺ and GFAP⁺ cells), signals above thresholded background were used for manual region of interest segmentation to calculate the total mean fluorescent area of cells using BZ-X Advanced Analysis Software (Keyence). For the ropivacaine uncaging safety analysis, a similar process was complete except for targeting and collecting of the sciatic nerve and surrounding thigh tissues following ropivacaine-HCl (n = 3) or ropivacaine-loaded AAL (n = 3) administration. Muscle and nerve tissues beneath the targeted site were extracted and stored in 4% PFA for 24 h. Tissues were frozen in optimal cutting medium compound and stored at –80 °C until serially sectioned with a cryostat at 7 μm. Every 20th section (140 μm apart) was stained with H&E for gross examination of tissue damage.

General statistical analysis

Rats were randomly assigned to treatment conditions. Data where n ≤ 4 were plotted as bar plots and n ≥ 5 were plotted as box plots. For comparisons between two groups, a two-tailed, two-sample Student’s t-test was conducted. For comparisons between multiple groups, a one-way, two-sided, analysis of variance (ANOVA) was conducted, followed by a post hoc Tukey’s honestly significant difference (HSD) test for pairwise comparisons. Effect size was calculated using Hedges g. All comparisons were two-tailed. Statistical tests, sample sizes N, corrected P values and effect sizes g are reported for each analysis in the text and figure captions. Pharmacokinetic parameters were estimated using the NonCompart package⁵⁹ and comparisons were made with two-tailed Student’s t-tests in R version 4.1.3. The remaining statistical analyses were performed using GraphPad Prism 10 (GraphPad Software) and custom scripts in Python.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Lipsman, N., Mainprize, T. G., Schwartz, M. L., Hynynen, K. & Lozano, A. M. Intracranial applications of magnetic resonance-guided focused ultrasound. Neurother. J. Am. Soc. Exp. Neurother. 11, 593–605 (2014).

   Google Scholar 

2. Wang, J. B. et al. Focused ultrasound for noninvasive, focal pharmacologic neurointervention. Front. Neurosci. 14, 675 (2020).

   PubMed  PubMed Central  Google Scholar 

3. Sirsi, S. R. & Borden, M. A. Advances in ultrasound mediated gene therapy using microbubble contrast agents. Theranostics 2, 1208–1222 (2012).

   PubMed  PubMed Central  CAS  Google Scholar 

4. Pitt, W. G., Husseini, G. A. & Staples, B. J. Ultrasonic drug delivery—a general review. Expert Opin. Drug Deliv. 1, 37–56 (2004).

   PubMed  PubMed Central  CAS  Google Scholar 

5. Fabiilli, M. L., Lee, J. A., Kripfgans, O. D., Carson, P. L. & Fowlkes, J. B. Delivery of water-soluble drugs using acoustically triggered perfluorocarbon double emulsions. Pharm. Res. 27, 2753–2765 (2010).

   PubMed  PubMed Central  CAS  Google Scholar 

6. Huang, S.-L. & MacDonald, R. C. Acoustically active liposomes for drug encapsulation and ultrasound-triggered release. Biochim. Biophys. Acta 1665, 134–141 (2004).

   PubMed  CAS  Google Scholar 

7. Dou, Y., Hynynen, K. & Allen, C. To heat or not to heat: challenges with clinical translation of thermosensitive liposomes. J. Control. Release 249, 63–73 (2017).

   PubMed  CAS  Google Scholar 

8. Rwei, A. Y. et al. Ultrasound-triggered local anaesthesia. Nat. Biomed. Eng. 1, 644–653 (2017).

   PubMed  PubMed Central  CAS  Google Scholar 

9. Orita, Y. et al. Acoustic-responsive carbon dioxide-loaded liposomes for efficient drug release. Ultrason. Sonochem. 94, 106326 (2023).

   PubMed  PubMed Central  CAS  Google Scholar 

10. Zhong, Q. et al. Polymeric perfluorocarbon nanoemulsions are ultrasound-activated wireless drug infusion catheters. Biomaterials 206, 73–86 (2019).

    PubMed  PubMed Central  CAS  Google Scholar 

11. Wang, J. B., Aryal, M., Zhong, Q., Vyas, D. B. & Airan, R. D. Noninvasive ultrasonic drug uncaging maps whole-brain functional networks. Neuron 100, 728–738.e7 (2018).

    PubMed  PubMed Central  CAS  Google Scholar 

12. Zucker, D., Marcus, D., Barenholz, Y. & Goldblum, A. Liposome drugs’ loading efficiency: a working model based on loading conditions and drug’s physicochemical properties. J. Control. Release 139, 73–80 (2009).

    PubMed  CAS  Google Scholar 

13. Inactive Ingredients in Approved Drug Products Search: Frequently Asked Questions (FDA, 2025); https://www.accessdata.fda.gov/scripts/cder/iig/index.cfm

14. Research C. for D. E. and Inactive Ingredients in Approved Drug Products Search: Frequently Asked Questions (FDA, 2022); https://www.accessdata.fda.gov/scripts/cder/iig/index.cfm

15. Nithiyanantham, S. & Palaniappan, L. Ultrasonic study on some monosaccharides in aqueous media at 298.15K. Arab. J. Chem. 5, 25–30 (2012).

    CAS  Google Scholar 

16. Strohm, E. M. & Kolios, M. C. Sound velocity and attenuation measurements of perfluorocarbon liquids usingphotoacoustic methods. In 2011 IEEE International Ultrasonics Symposium 2368–2371 (IEEE, 2011).

17. Nyborg, W. L. Solutions of the bio-heat transfer equation. Phys. Med. Biol. 33, 785–792 (1988).

    PubMed  CAS  Google Scholar 

18. Martin, E. et al. ITRUSST consensus on standardised reporting for transcranial ultrasound stimulation. Brain Stimul. 17, 607–615 (2024).

    PubMed  Google Scholar 

19. Marketing Clearance of Diagnostic Ultrasound Systems and Transducers—Guidance for Industry and Food and Drug Administration Staff (FDA, 2025); https://www.fda.gov/media/71100/download

20. Lendor, S. et al. Solid phase microextraction-based miniaturized probe and protocol for extraction of neurotransmitters from brains in vivo. Anal. Chem. 91, 4896–4905 (2019).

    PubMed  CAS  Google Scholar 

21. Airan, R. Neuromodulation with nanoparticles. Science 357, 465–465 (2017).

    PubMed  Google Scholar 

22. Moda-Sava, R. N. et al. Sustained rescue of prefrontal circuit dysfunction by antidepressant-induced spine formation. Science 364, eaat8078 (2019).

    PubMed  PubMed Central  CAS  Google Scholar 

23. Vesuna, S. et al. Deep posteromedial cortical rhythm in dissociation. Nature 586, 87–94 (2020).

    PubMed  PubMed Central  CAS  Google Scholar 

24. Gamaro, G. D. et al. The effects of acute and repeated restraint stress on the nociceptive response in rats. Physiol. Behav. 63, 693–697 (1998).

    PubMed  CAS  Google Scholar 

25. Farmer, C. A. et al. Ketamine metabolites, clinical response, and gamma power in a randomized, placebo-controlled, crossover trial for treatment-resistant major depression. Neuropsychopharmacology 45, 1398–1404 (2020).

    PubMed  PubMed Central  CAS  Google Scholar 

26. Li, D. & Mashour, G. A. Cortical dynamics during psychedelic and anesthetized states induced by ketamine. NeuroImage 196, 32–40 (2019).

    PubMed  CAS  Google Scholar 

27. Tian, F. et al. Characterizing brain dynamics during ketamine-induced dissociation and subsequent interactions with propofol using human intracranial neurophysiology. Nat. Commun. 14, 1748 (2023).

    PubMed  PubMed Central  CAS  Google Scholar 

28. Pinault, D. N-Methyl d-aspartate receptor antagonists ketamine and MK-801 induce wake-related aberrant γ oscillations in the rat neocortex. Biol. Psychiatry 63, 730–735 (2008).

    PubMed  CAS  Google Scholar 

29. Shafer, S. StevenLShafer/stanpumpR (2024).

30. Van Den Broek, S. P., Reinders, F., Donderwinkel, M. & Peters, M. J. Volume conduction effects in EEG and MEG. Electroencephalogr. Clin. Neurophysiol. 106, 522–534 (1998).

    PubMed  Google Scholar 

31. Barthas, F. & Kwan, A. C. Secondary motor cortex: where ‘sensory’ meets ‘motor’ in the rodent frontal cortex. Trends Neurosci. 40, 181–193 (2017).

    PubMed  CAS  Google Scholar 

32. Polis, A. J., Fitzgerald, P. J., Hale, P. J. & Watson, B. O. Rodent ketamine depression-related research: finding patterns in a literature of variability. Behav. Brain Res. 376, 112153 (2019).

    PubMed  PubMed Central  Google Scholar 

33. Di Ianni, T. et al. Sex dependence of opioid-mediated responses to subanesthetic ketamine in rats. Nat. Commun. 15, 893 (2024).

    PubMed  PubMed Central  Google Scholar 

34. Ali, F. et al. Ketamine disinhibits dendrites and enhances calcium signals in prefrontal dendritic spines. Nat. Commun. 11, 72 (2020).

    PubMed  PubMed Central  CAS  Google Scholar 

35. Hetzler, B. E. & Swain Wautlet, B. Ketamine-induced locomotion in rats in an open-field. Pharmacol. Biochem. Behav. 22, 653–655 (1985).

    PubMed  CAS  Google Scholar 

36. Alexander, L., Jelen, L. A., Mehta, M. A. & Young, A. H. The anterior cingulate cortex as a key locus of ketamine’s antidepressant action. Neurosci. Biobehav. Rev. 127, 531–554 (2021).

    PubMed  CAS  Google Scholar 

37. Cuvillon, P. et al. A comparison of the pharmacodynamics and pharmacokinetics of bupivacaine, ropivacaine (with epinephrine) and their equal volume mixtures with lidocaine used for femoral and sciatic nerve blocks: a double-blind randomized study. Anesth. Analg. 108, 641–649 (2009).

    PubMed  CAS  Google Scholar 

38. Cameron, L. P. et al. A non-hallucinogenic psychedelic analogue with therapeutic potential. Nature 589, 474–479 (2021).

    PubMed  CAS  Google Scholar 

39. Reardon, S. MDMA therapy for PTSD rejected by FDA panel. Nature https://doi.org/10.1038/d41586-024-01622-3 (2024).

40. Bonaventura, J. et al. Pharmacological and behavioral divergence of ketamine enantiomers: implications for abuse liability. Mol. Psychiatry 26, 6704–6722 (2021).

    PubMed  PubMed Central  CAS  Google Scholar 

41. SPRAVATO® (esketamine) | Healthcare Professional Website. SPRAVATO® (esketamine): A Prescription Nasal Spray | SPRAVATO® HCP (2025); https://www.spravatohcp.com/trd-efficacy-safety/

42. Swanson, L. W. Brain Maps III: Structure of the Rat Brain (Gulf Professional Publishing, 2004).

43. Paxinos, G. & Watson, C. The Rat Brain in Stereotaxic Coordinates: Hard Cover Edition (Elsevier, 2006).

44. O’Reilly, M. A., Muller, A. & Hynynen, K. Ultrasound insertion loss of rat parietal bone appears to be proportional to animal mass at submegahertz frequencies. Ultrasound Med. Biol. 37, 1930–1937 (2011).

    PubMed  PubMed Central  Google Scholar 

45. Treeby, B. E. & Cox, B. T. k-Wave: MATLAB toolbox for the simulation and reconstruction of photoacoustic wave fields. J. Biomed. Opt. 15, 021314 (2010).

    PubMed  Google Scholar 

46. Roy, K. S. et al. Optimizing a high-throughput solid-phase microextraction system to determine the plasma protein binding of drugs in human plasma. Anal. Chem. 93, 11061–11065 (2021).

    PubMed  CAS  Google Scholar 

47. Huq, M., Tascon, M., Nazdrajic, E., Roszkowska, A. & Pawliszyn, J. Measurement of free drug concentration from biological tissue by solid-phase microextraction: in silico and experimental study. Anal. Chem. 91, 7719–7728 (2019).

    PubMed  CAS  Google Scholar 

48. Hassani, S. A. et al. Dose-dependent dissociation of pro-cognitive effects of donepezil on attention and cognitive flexibility in rhesus monkeys. Biol. Psychiatry Glob. Open Sci. 3, 68–77 (2023).

    PubMed  Google Scholar 

49. Boyacı, E. et al. High-throughput analysis using non-depletive SPME: challenges and applications to the determination of free and total concentrations in small sample volumes. Sci. Rep. 8, 1167 (2018).

    PubMed  PubMed Central  Google Scholar 

50. Stenroos, P. et al. Awake rat brain functional magnetic resonance imaging using standard radio frequency coils and a 3D printed restraint kit. Front. Neurosci. 12, 548 (2018).

51. OpenBCI Software. OpenBCI. https://openbci.com/ (2024).

52. OpenBCI_Python. OpenBCI Archive (2024); https://github.com/openbci-archive/OpenBCI_Python

53. Gramfort, A. et al. MEG and EEG data analysis with MNE-Python. Front. Neurosci. 7, 267 (2013).

54. Welch, P. The use of fast Fourier transform for the estimation of power spectra: a method based on time averaging over short, modified periodograms. IEEE Trans. Audio Electroacoustics 15, 70–73 (1967).

    Google Scholar 

55. Rodriguez, A., Zhang, H., Klaminder, J., Brodin, T. & Andersson, M. ToxId: an efficient algorithm to solve occlusions when tracking multiple animals. Sci. Rep. 7, 14774 (2017).

    PubMed  PubMed Central  Google Scholar 

56. Rodriguez, A. et al. ToxTrac: a fast and robust software for tracking organisms. Methods Ecol. Evol. 9, 460–464 (2018).

    Google Scholar 

57. Deuis, J. R., Dvorakova, L. S. & Vetter, I. Methods used to evaluate pain behaviors in rodents. Front. Mol. Neurosci. 10, 284 (2017).

    PubMed  PubMed Central  Google Scholar 

58. Kolarczyk, L. M. & Williams, B. A. Transient heat hyperalgesia during resolution of ropivacaine sciatic nerve block in the rat. Reg. Anesth. Pain Med. 36, 220–224 (2011).

    PubMed  PubMed Central  Google Scholar 

59. Kim, H., Han, S., Cho, Y.-S., Yoon, S.-K. & Bae, K.-S. Development of R packages: ‘NonCompart’ and ‘ncar’ for noncompartmental analysis (NCA). Transl. Clin. Pharmacol. 26, 10 (2018).

    PubMed  PubMed Central  Google Scholar 

60. Airan, R. D. et al. Purohitetal. GitHub https://github.com/Airan-Lab (2025).

Download references

Author notes

1. These authors contributed equally: Mahaveer P. Purohit, Brenda J. Yu, Kanchan Sinha Roy, Yun Xiang.

Authors and Affiliations

1. Department of Radiology, Stanford University, Stanford, CA, USA

   Mahaveer P. Purohit, Brenda J. Yu, Kanchan Sinha Roy, Yun Xiang, Sedona N. Ewbank, Matine M. Azadian, Alex R. Hart, Gabriella P. B. Muwanga, Payton J. Martinez, Jeffrey B. Wang, Ali K. Taoube, Eric Markarian, Nicholas Macedo, Audrey K. Kwan, Diego Gomez Lopez & Raag D. Airan

2. Biophysics Program, Stanford University, Stanford, CA, USA

   Brenda J. Yu

3. Neurosciences Program, Stanford University, Stanford, CA, USA

   Sedona N. Ewbank, Matine M. Azadian & Gabriella P. B. Muwanga

4. Department of Chemistry, Stanford University, Stanford, CA, USA

   Alex R. Hart

5. Department of Anesthesiology and Critical Care Medicine, John Hopkins Hospital, Baltimore, MD, USA

   Jeffrey B. Wang

6. Department of Medicine, Health, and Society, Vanderbilt University, Nashville, TN, USA

   Diego Gomez Lopez

7. Department of Psychiatry and Behavioral Sciences, Stanford University, Stanford, CA, USA

   Raag D. Airan

8. Department of Materials Science and Engineering, Stanford University, Stanford, CA, USA

   Raag D. Airan

Contributions

Conceptualization: R.D.A., Y.X., M.P.P., B.J.Y. and K.S.R. Methodology: M.P.P., B.J.Y., K.S.R., Y.X., S.N.E., M.M.A., J.B.W., A.R.H., P.J.M. and R.D.A. Investigation: M.P.P., B.J.Y., K.S.R., Y.X., S.N.E., M.M.A., A.R.H., G.P.B.M, P.J.M., A.K.T., E.M., N.M., A.K.K. and D.G.L. Analysis: K.S.R., B.J.Y., Y.X., S.N.E., P.J.M., M.P.P., M.M.A., A.R.H., G.P.B.M. and R.D.A. Visualization: M.P.P., B.J.Y. and R.D.A. Funding acquisition: R.D.A. Project administration: R.D.A. Supervision: R.D.A. Writing (initial draft): B.J.Y., M.P.P. and R.D.A. Review and editing: all authors.

Corresponding author

Correspondence to Raag D. Airan.

Competing interests

R.D.A. has equity or has received consulting fees from Cordance Medical, NaviFUS and Lumos Labs and grant funding from AbbVie Inc. All other authors declare no competing interests.

Peer review information

Nature Nanotechnology thanks the anonymous reviewers for their contribution to the peer review of this work.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions