In this section, we discuss the application of US to AD mouse models. Histopathologically, AD is characterised by extracellular Aβ deposits (senile plaques), intraneuronal Tau aggregates (neurofibrillary tangles and neuropil threads), as well as neuroinflammation [27]. The currently available AD therapies (except for the recently approved—and highly disputed—anti-Aβ antibody aducanumab) do not delay the neurodegenerative process and only provide symptomatic relief. Approved AD drugs are the acetylcholine esterase inhibitors donepezil (Aricept), rivastigmine (Exelon) and galantamine (Razadyne), and the NMDA receptor antagonist memantine (Namenda). The mainstay of the clinical trials that have been performed over the recent years are vaccines targeting Aβ and Tau, based on the assumption that these two molecules, which form insoluble toxic protein aggregates that constitute the hallmark lesions of AD, both initiate and drive the disease. The leading hypothesis in the field is the amyloid cascade hypothesis that places Aβ upstream of Tau [28]; however, there is also a crucial role for Tau given that Aβ toxicity is Tau-dependent, as formulated in the Tau axis hypothesis [29,30]. In transgenic animal models with either Aβ or Tau pathology, transgenic expression of mutant forms of the genes encoding APP (from which Aβ is derived by proteolytic cleavage) and Tau leads to the formation of pathological aggregates and ensuing cognitive impairments, and removing these aggregates restores cognitive functions [31].

Regarding protein aggregates, there are two fundamental treatment options, either preventing their formation in the first place or when they have formed removing them. It seemed therefore reasonable to trial the US technology and attempt to clear Aβ and Tau aggregates and thereby restore cognitive functions. It came as a surprise that Aβ could be significantly cleared by using US^+MB without a therapeutic agent [32,33]. By using US in a scanning mode (SUS) and 5–8 weekly treatment sessions, not only was Aβ effectively cleared, but memory functions were also improved or even restored as shown in three complementary behavioural tests, the Y-maze, novel objection recognition, and active place avoidance tests [33]. Aβ clearance by dormant microglia was identified as an underlying therapeutic mechanism. The microglia had taken up Aβ into their lysosomes, a process potentially mediated by unknown blood-borne factors that had entered the brain in response to US^+MB-mediated BBB opening [32,33]. In a recent RNAseq transcriptomic analysis of microglia isolated from US^+MB-treated APP23 mice followed by incubation with the amyloid dye methoxy-XO4, ‘cell cycle’ and ‘phagocytosis’ genes were specifically activated, hinting at the underlying microglial activation mechanism [34] (see also a recent study aggregating 75 murine transcriptomes in response to US^+MB [35]). Importantly, US^+MB was even shown to be safe in aged APP23 mice that display pronounced cerebral amyloid angiopathy (CAA). Again, Aβ was found to be cleared effectively by microglia and while not reducing the total plaque area, US^+MB reduced the fraction of large plaques. An obvious question is whether the clearance effects are long-lasting. Poon and colleagues performed a time-course experiment using two-photon microscopy to examine this. The study revealed that one single US^+MB session (dose) reduced the size of plaques for two weeks, advocating fortnightly treatment sessions in human AD [36] (for comparison, vaccinations are often spaced out by four weeks). Longer time frames have not been investigated [36]. The team also performed two-photon microscopy to assess leukocyte infiltration in response to US^+MB treatment and found significantly more neutrophils in the sonicated compared to the contralateral hemisphere, suggesting a role not only for brain-resident microglia but also peripheral neutrophils, the first line of defence of the innate immune system [37]. Of note, US^+MB has assisted the delivery of other agents, including the GSK inhibitor AR-A014418 [38], a TrkA agonist [39], and exosomes [40] in Aβ-depositing mice, and Aβ-targeting metal complexes in wild-type mice [41]. More recently, stem cells have been delivered together with anti-Aβ antibodies [42]. Of note, while in APP23 mice it has been shown that BBB opening is required to clear Aβ [43], US^only applied in another AD mouse model (using a transducer at a higher (1.875 MHz) frequency) activated endothelial nitric oxide synthase (eNOS), an enzyme with a role in angiogenesis and vasodilation, stimulated microglia and removed Aβ [44]. In the same publication, US^only-induced cognitive improvements in a model of vascular dementia were found to be abrogated on an eNOS knockout background [44].

US^+MB has also been applied to Tau transgenic mouse models. Tau is a much harder target than Aβ, given that its pathology is largely intracellular and therefore less accessible to therapeutic agents. In Tau transgenic pR5 mice (that carry the P301L mutation found in frontotemporal dementia), US^+MB (which constituted one of four treatment arms) reduced Tau pathology as shown for the amygdala [24]. Reduced Tau pathology was also reported in another Tau transgenic mouse model, rTg4510, using unilateral US^+MB treatments [45]. In Tau transgenic K3 mice (carrying the frontotemporal dementia mutation K369I) with a memory and motor deficit, US^+MB was shown to partly ameliorate these behavioural phenotypes, and Tau was cleared by the activation of neuronal autophagy (rather than microglial activation) [46]. Given that Tau is a protein that has been demonstrated to be degraded in lysosomes via autophagy [47], US^+MB therefore appears to boost an intrinsic mechanism that neurons use to clear Tau. Finally, US^+MB has also been applied to a mouse model with both Aβ and Tau pathology in the 3xTg-AD strain (although in these mice the Tau pathology is very modest) [48]. The study confirmed microglial engulfment of amyloid and improvements in learning and memory. Moreover, a proteomic analysis was performed to gain additional insight into potential mechanisms.

US^+MB has further been combined with anti-Tau and anti-Aβ antibodies, exploring antibody formats ranging from single-chain variable fragments (ScFvs) with a molecular weight of 29 kDa to full-sized immunoglobulins with a molecular weight of 150 kDa. Nisbet and colleagues explored US^+MB together with an ScFv antibody fragment targeting 2N Tau (RN2N), with the epitope being present in 2N4R Tau isoform over-expressing pR5 mice [24]. The study found that the combination treatment was superior to the single treatment arms of either using US^+MB or the RN2N ScFv alone. Interestingly, in the combination treatment arm, the antibody fragment was not only shown to be taken up by the brain, but also to effectively distribute into neuronal cell bodies and dendrites, demonstrating that US^+MB not only overcomes the BBB but also the plasma membrane as the second barrier [24]. In a follow-up study, different RN2N antibody formats were explored (ScFv, Fab and IgG), revealing an up to 30-fold increased uptake of the therapeutic antibody mediated by US^+MB+mAb [49].

Jordao and colleagues evaluated the anti-Aβ antibody BAM-10 and showed US^+MB-mediated Aβ reductions in the TgCRND8 mouse model [50]. The clinically approved anti-Aβ antibody aducanumab has been tested in APP23 mice in a combination trial, revealing improved cognitive outcomes compared to the single treatment arms of either antibody alone or US^+MB alone [51]. A similar approach was pursued by Lemere and colleagues, who delivered an anti-pyroglutamylated Aβ antibody to aged APP/PS1dE9 mice [52]. They showed that mice administered with the combination treatment had reduced hippocampal plaque burden compared to controls. In contrast, in the aducanumab study, all three treatment arms (antibody alone, US^+MB alone, and the combination treatment US^+MB+mAb) reduced plaque burden in the hippocampus, whereas in the cortex, only the combination treatment was effective [51]. It was argued that hippocampal plaques develop later than cortical plaques, and that the differences observed in clearance for the two brain areas could be related to the timing of the treatments.

Collectively, these promising preclinical data have spurred several small cohort clinical trials that have shown safety and feasibility of US^+MB-mediated BBB opening in AD (NCT04118764, NCT04526262, NCT03119961, NCT03739905, NCT03671889, and NCT02986932). The trials differ in that they either use the MRI-guided ExAblate Neuro system multi-element array operated at 220 kHz (Insightec) [53,54], a 1 MHz transducer implanted in the skull bone thickness and connected to an external power supply via a transdermal needle connection during activation and facing the brain (Carthera: Sonocloud [55]), or a single 0.25 MHz element transducer with guidance provided by the Brainsight neuronavigation system (Columbia University) as outlined in a macaque study [56]. Whether US^+MB-medicated BBB opening will achieve statistically significant Aβ reductions and cognitive improvements in human patients, and whether these improvements are long-lasting, needs to be determined. In support of combining US^+MB with an anti-Aβ antibody, in a recent article, Karran and De Strooper argue that ‘the speed of amyloid removal from the brain by a potential therapy will be important in demonstrating clinical benefit in the context of a clinical trial’ [1], clearly making a case for US^+MB-mediated enhanced delivery of antibodies such as aducanumab into the brain of AD patients.