In this study, we sought to establish the parameters to focally disrupt the BBB across a cohort of marmoset monkeys. By integrating our MRI-based marmoset atlases21,22 with a motorized stereotactic positioning system (RK-50 Marmoset; FUS Instruments Incorporated, Toronto, ON, Canada, Fig. 1) we were able to focally sonicate sites across the dorsal surface of the marmoset cortex with high accuracy (Figs. 1–8) by way of spherically focused single-element transducers. Of the two transducers tested – at 515 kHz and 1.46 MHz – we found that the higher frequency 1.46 MHz transducer (with a 24.5 mm focal length) allowed for disruptions that could be limited to the ~ 2.5–3 mm cortical thickness of the marmoset cortex. We optimized parameters (minimum acoustic pressure, minimum microbubble dosage, burst duration, number of bursts) from which BBB disruption occurred without hemorrhage or edema and to the extent that Evans blue and/or GBCA extravasation occurred. From these experiments, we establish parameters for safe BBB disruption (as reported by H & E staining and immunohistochemistry) in the marmoset at 1.46 MHz. At these parameters, we found that spatial bounding of Evans blue staining and GBCA reporting (Fig. 7) were similar, such that the determination of a BBB disruption can be conducted in vivo with MRI-based contrast agents. Taken together, the experiments described here provide an account from which in vivo, minimally invasive substance delivery experiments can be designed around.
As with the rodent brain, marmosets have a lissencephalic cortex, making this species an ideal candidate for tFUS-based BBB disruption, allowing for sonications along the columnar organization of the cortex, unencumbered by cortical folds present in most other primate species. As we show here, however, sonication parameters from rodent species (e.g., rats, mice)8 cannot be simply ported for use in the marmoset, nor can those used in for other larger nonhuman primate species such as macaques (with a much thicker skull, and folded cortex)4. Indeed, in addition to the parameters driving the transducer, the physical properties of transducers also complicate comparisons (e.g., focal length, frequency). The experiments presented here demonstrate that a 35 mm single element spherically focused 1.46 MHz transducer can be used for cortical disruptions in the marmoset with minimal deleterious effects of the marmoset head morphology (e.g., skull thickness or the presence of temporalis muscles; Fig. 5), particularly when the center of focus is at or near the surface of cortex (Figs. 1–8), allowing for further spatial minimization of cortical disruption. The use of a single-element transducer rather than an array of transducers simplifies the means necessary to conduct tFUS in the marmoset. We made use of a motorized positioning system and automated atlas targeting, but these experiments could also be conducted with a transducer mounted to a stereotactic manipulator arm, further simplifying the equipment needed to use focused ultrasound to sonicate the marmoset brain.
Although it should be noted that microbubble experiments (dosage, clearance time) may be parameter and transducer-dependent31, we found that the minimum microbubble dosage (Definity, Lantheus Medical Imaging, Billerica, MA, USA), via the tail or saphenous vein was greater than 20 µL/kg when injected as a bolus (Fig. 3). To target a smaller distribution of microbubbles (with larger bubbles being more buoyant), we drew from as close to the bottom of the microbubble vial as possible (after activation and slowly inverting the vial) with a 21-gauge needle, and another 21-gauge needle to vent the vial. We also chose to use a 26-gauge catheter to avoid premature destruction of the microbubbles during injection of our saline-diluted microbubble solution. During initial experiments, we found that the use of a spring-loaded extension led to inconsistent BBB perturbation, likely due to premature bursting of the microbubbles. As such, all injections were made directly into the catheter hub. In terms of microbubble clearance time, we found it to be relatively fast (< 2 minutes) in the marmoset, even with a high dose of 200–400 µL/kg – at least to the extent that the circulating microbubble concentration resulted in BBB disruption (Fig. 4). We found that the acoustic emissions indicated cavitation of the microbubbles at 30 seconds. In particular, consistent with previous reports in rats8 using a similar transducer and hydrophone hardware, subharmonic broadband noise was evident at 30 and 120 seconds post microbubble bolus injection, as well as an increased amplitude at the second harmonic (Supplementary Fig. 1). This likely corresponded to microbubble cavitation, although it did not correspond to opening at 120 seconds post-microbubble injection. This subharmonic effect was not apparent at 240- or 480-seconds post-microbubble injection.
Our original intent to determine BBB disruption as a function of burst duration and number of bursts was not to assess damage (visible to the naked eye, Fig. 5), but as shown by the H & E staining (supplementary Fig. 2), the pressure used for marmoset SK was above of what can be considered safe (Fig. 8). These data, however, are particularly useful for demonstrating the extent of damage that can occur at the extremes of acoustic pressure, burst durations, and at high microbubble dosages (400 µL/kg). We found that these high-energy sonications resulted in diffuse tissue damage usually accompanied by microbleeds. Of particular interest, we found that the subdural space presented hemorrhage. This area is particularly sensitive for hemorrhages due to the numerous presence of perforating arteries in a compact space and thus the microbubble concentration may have been higher.
Given the results of the H & E staining of the animals that showed reliably successful disruptions of the BBB (Fig. 2–6), we applied what we found to be safe parameters (acoustic pressure = 1.8 MPa, burst duration = 20 ms, burst period = 1,000 ms, number of bursts = 60, microbubble dose = 100 µL/kg) to a frontal site (area 8a) in marmoset G to determine the length of time that the BBB remained open, as indexed by permeability to a bolus of GBCA at 2, 4, and 8 hours after sonication. The BBB was clearly open at 8 hours post-sonication, as indicated by increased intensity and distribution resulting from the GBCA and at the start of the 8-hour post-sonication scan (Fig. 7). Although this long-duration disruption may present some risk (e.g., blood-borne bacteria) that the BBB would normally protect against circulating toxins or pathogens in the bloodstream32, it is an experimentally advantageous treatment window for injecting substances that may be dangerous if injected as a bolus, rather than slowly infused. The histology and immunohistochemistry in marmoset G supported the previous results that the parameters were safe, such that no readily apparent damage was observed in the H & E staining (Fig. 8). There is visible microglial activation (Fig. 7) due to tissue perturbation (Iba1), but not tissue damage (DAPI, NeuN, H & E) compared to a contralateral non-sonicated 8a site indicating that these are safe and reliable parameters to open the BBB for an extended period. Microglia in cortical regions showed signs of activation through increased Iba1 expression and changes in cell bodies and processes, without significant changes in cell numbers. FUS-induced BBB disruption has been shown to trigger transient glial activation33. Depending on the type and severity of brain injury, activated microglia as well as infiltrating macrophages can exacerbate neuroinflammation and neurodegeneration. It has been demonstrated, however, that microglia activation resolved by 15 d after FUS with no progression to a glial scar, suggesting that FUS does not cause lesion-like microgliosis34.
In summary, we demonstrate safe and effective disruption of the BBB in the marmoset with a spatial specificity of approximately 1 mm radially and 2.5 mm axially (in cortex) using a 1.46 MHz transducer. We were able to reliably perturb the BBB across the dorsal surface of the marmoset cortex with a minimum acoustic pressure to be between 1.8–2.2 MPa and a minimum microbubble dosage of 20 µL/kg via tail-vein injection. We demonstrate that these parameters (paired with 60 20 ms bursts, spaced at 1000 ms) led to the BBB being open for greater than 8 hours and did not lead to cortical tissue damage, as reported by H & E staining. Higher acoustic pressures and/or excessive microbubble dosage (> 200 µL/kg) led to tissue damage (Fig. 8). The series of experiments presented here establish methods for safely, reproducibly, and focally perturbing the BBB using tFUS in the common marmoset monkey.