key: cord-0425470-3fjak47m
authors: Bajracharya, Rinie; Cruz, Esteban; Götz, Jürgen; Nisbet, Rebecca M.
title: Ultrasound-mediated delivery of novel tau-specific monoclonal antibody enhances brain uptake but not therapeutic efficacy
date: 2021-10-19
journal: bioRxiv
DOI: 10.1101/2021.10.18.464732
sha: 1fa069d6b28ea63dabc7e089ef954f8149a8de0e
doc_id: 425470
cord_uid: 3fjak47m

Tau-specific immunotherapy is an attractive therapeutic strategy for the treatment of Alzheimer’s disease and other tauopathies. However, targeting tau effectively remains a considerable challenge due to the restrictive nature of the blood-brain barrier (BBB), which excludes 99.9% of peripherally administered antibodies. We have previously shown that the delivery of tau-specific monoclonal antibody (mAb) with low-intensity scanning ultrasound in combination with intravenously injected microbubbles (SUS+MB) increases the passage of IgG antibodies into the brain. SUS+MB transiently opens tight junctions to allow paracellular transport, but also facilitates transcellular transport, particularly for larger cargoes. However, therapeutic efficacy after enhanced brain delivery has not been explored. To assess whether ultrasound-mediated delivery of tau-specific mAbs leads to an enhanced therapeutic response, K369I tau transgenic K3 mice were passively immunised once weekly for 12 weeks with a novel mAb, RNF5, in combination with SUS+MB. While none of the treatment arms improved behaviour or motor functions in these mice, we found that both RNF5 and SUS+MB treatments on their own reduced tau pathology, but, surprisingly, the combination of both (RNF5+SUS+MB) did not achieve an additive reduction in tau pathology. This was despite observing increased antibody penetration in the brain. Interestingly, a significant fraction of the antibody in the combination treatment was visualized in brain endothelial cells, suggesting that paracellular transport may not be the preferred uptake mechanism for RNF5. Taken altogether, more research is warranted to develop SUS+MB as a delivery modality for anti-tau antibodies.

Tauopathies, such as Alzheimer's disease (AD), are a group of neurodegenerative diseases in which abnormally phosphorylated tau aggregates into intracellular tau inclusions characteristic of any given tauopathy. In pre-clinical trials, immunotherapy has emerged as a valuable technique to reduce tau pathogenesis. However, in human clinical trials, it turned out to be challenging to achieve the primary endpoint outcomes. While there may be several reasons for this, we argue that achieving sufficient antibody concentration in the brain is a large determinant of therapeutic efficacy. Here, the blood-brain barrier (BBB) is a significant impediment to the passage of therapeutics into the central nervous system (CNS), with only 2% of peripherally administered small molecule drugs entering the brain [1] . For large molecule drugs, such as IgG antibodies, this diminishes to 0.1% and thus highlights the demand for novel approaches to increase the antibody concentration in the brain to facilitate increased target engagement [2, 3] . An emerging technology to enhance the passage of drugs across the BBB is the use of low intensity focused ultrasound to transiently open the BBB [4, 5] . In previous work, we have optimised a technique whereby repeated low-intensity ultrasound is applied in a scanning mode together with intravenous injection of microbubbles (scanning ultrasound with microbubbles, SUS +MB ), allowing for safe and transient BBB opening across the whole brain [6] [7] [8] . Whilst pre-clinical studies performed by our lab and others have demonstrated that molecules of various sizes, as well as antibodies, are able to be delivered to the brain using ultrasound [9] [10] [11] [12] [13] , therapeutic efficacy after enhanced brain delivery of a tauspecific mAb has not yet been assessed. Therefore, we aimed to investigate the therapeutic efficacy of our novel anti-tau mAb, RNF5, and whether its ability to reduce pathogenic tau could be enhanced by combining its delivery with SUS +MB in K396I tau transgenic mice, a model of tauopathy that develops motor deficits from an early age [14] . To examine this, we performed a preclinical study with multiple arms in which mice were treated with either RNF5 alone, SUS +MB alone, or the combination of both (RNF5+SUS +MB ) once weekly for 12 weeks, assessing behavioural and motor functions together with biochemical and histological readouts as a proxy for therapeutic efficacy. 

Surface plasmon resonance measurements were performed and analysed with assistance from the National Biologics Facility (NBF) at the University of Queensland. RNF5 was tested for binding to human 2N4R tau (hTau) and mouse 2N4R tau (mTau) using the Biacore T200 instrument. Briefly, hTau and mTau were immobilised onto an amine-coupling activated (NHS/EDC) CM5 Series S Sensor Chip (GE Healthcare) with a contact time of 180 s at 5 µL/min flow rate and at 25°C. For immobilisation, hTau was diluted to 42 nM and mTau to 455 nM in acetate buffer at pH 5.5 and pH 4, respectively. Binding of RNF5 to tau was tested using a Multi-Cycle High Performance Kinetics assay. RNF5 injections were performed using HPS-EP+ buffer (GE Healthcare) as a running buffer at a flow rate of 30 µL/min for 180 s followed by a dissociation time of 1,200 s in 3-fold increasing concentration steps (0.5-44.3 nM). For a given concentration, injections were done in duplicate, and regeneration of the chip surface was achieved with glycine pH 1.7. The data was blank-subtracted and fitted to a Two-State reaction model, using BIAevaluation software (GE Healthcare).

All animal experiments were conducted following the guidelines of the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and approved by the University of Queensland Animal Ethics Committee (QBI/554/17/NHMRC). For treatment studies, 5-week-old male K3 mice expressing human 1N4R tau with the K369I mutation under the control of mThy.1.2 promoter were utilized [14] . Age-matched wild-type (WT) male littermates were used as a control in behavioural experiments. For labelled antibody delivery experiments, 10 -30 week-old K3 mice of mixed gender were used.

Microbubbles comprising a phospholipid shell and octafluoropropane gas core were generated in-house as previously described [13] . Microbubble preparations yielded a mixture of microbubbles that were under 10 µm in size, with a mean diameter of 1 µm and a concentration of 1 -5 x 10 9 microbubbles/ml, similar to what has been reported for commercial microbubble preparations such as Definity ® .

An integrated focused ultrasound system was used (TIPS, Philips Research) [6] . This system consisted of an annular array transducer with a natural focus of 80 mm, a radius of curvature of 80 mm, a spherical shell of 80 mm with a central opening of 31 mm diameter, a 3D positioning system, and a programmable motorized system to move the ultrasound focus in the x and y planes to cover the entire brain area. A coupler mounted to the transducer was filled with degassed water and placed on the head of the mouse with ultrasound gel for coupling, to ensure unobstructed propagation of the ultrasound to the brain.

5-week-old K3 male and female mice were randomly assigned to one of the following groups:

RNF5 only (RNF5), SUS +MB only (SUS +MB ), or SUS +MB and RNF5 combined (RNF5+SUS +MB ), with the inclusion of K3 sham and WT littermate controls for behavioural studies. All animals receiving injections and/or SUS +MB treatment were anaesthetised and prepared as previously described [8] . Eight mice were used per experimental group and mice were treated once per week for 12 weeks. The sham K3 cohort received anaesthesia, but no retro-orbital injection or SUS +MB . For the antibody only and antibody plus SUS +MB groups, 15 mg/kg of RNF5 was injected on its own via retro-orbital injection or as a mixture of 15 mg/kg of RNF5 and microbubbles (1 µL per gram body weight), respectively. Animals receiving SUS +MB were placed under the ultrasound transducer, immobilised by a head frame. The parameters for the ultrasound delivery were 1 MHz centre frequency, 0.5 MPa peak rarefactional pressure (reduced from 0.7 MPa used in previous studies due to the young age and low weight of K3 mice [13] ), 10 Hz pulse repetition frequency, 10% duty cycle, and a 6s sonication time per spot. The focus of the transducer was 1.5 mm and 12 mm in the transverse and axial planes, respectively. The motorized position system moved the focus of the transducer array in a grid with 1.5 mm spacing between individual sites of sonication such that ultrasound was delivered in a zig-zag pattern from the back of the brain moving towards the front of the brain. Mice received a total of 48 spots of sonication in a 6x8 raster grid pattern for the first 9 weeks of treatment, then received a total of 72 spots of sonication in a 9x8 raster grid pattern in the last 3 weeks of treatment to include sonication of the cerebellum.

The weight of all animals was monitored weekly in both experiments involving K3 mice. Mice were tested weekly on the Rotarod to track motor function before, during and after treatment.

Briefly, mice were habituated in the behavioural room for at least 30 minutes prior to testing.

Light levels were kept constant at 60% (~70 lux) throughout the test. Animals were placed on the Rotarod using the linear acceleration mode (4 to 40 rpm) over 90 s, with a total test time of 180 s (90 s at 40 rpm). Latency to fall was recorded as the longest time a mouse spent on the rod out of 5 attempts. Grip strength of forelimbs and hindlimbs was tested once before and once after antibody treatment by allowing mice to grasp a metal T-bar, and the average peak force to release the metal bar out of 10 trials was recorded. The activity monitor was used as an additional test to measure general animal behaviour before and after antibody treatment.

The activity monitor is an automated animal tracking behavioural test whereby an individual animal is placed in a square-shaped open field box in which animal movement, position and speed are monitored by infrared beam breaks that project across the open field along the X, Y and Z axes. Mice were placed in the activity monitor box and allowed to explore for 30 minutes.

The software was set up according to the General Open Field test from the Activity Monitor version 7 manual (MED Associates, Inc.). All data was transmitted to a PC and analysed using Activity Monitor 7 software, SOF-812 (MED Associates, Inc.). Untreated wild-type littermates were used as controls. Given that K3 mice exhibit a pronounced motor deficit, Jump time and Ambulatory time were included as read-outs.

24 hours after completion of the behavioural testing, treated mice and age-matched WT littermate controls were anaesthetised with an overdose of sodium pentabarbitone and transcardially perfused with PBS, their brains dissected, and the hemispheres separated. The right hemisphere was immersion-fixed in 4% PFA (Sigma) for 24 hours, and then embedded in paraffin using a benchtop tissue processor (Leica). Sagittal paraffin-embedded brain sections at 7 μm and 14 μm thickness were obtained using a microtome. The left hemisphere was snapfrozen in liquid nitrogen and stored at -80 o C until further processing.

Snap-frozen brains from K3, WT and tau knockout mice were extracted using the sarkosyl extraction method with some modifications [15] . Briefly, snap-frozen hemispheres were homogenized in 3 volumes of RIPA-A buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1%

Nonidet P-40, 5 mM EDTA, pH 8.0, 0.5% sodium deoxycholate, 50 mM sodium fluoride, 200 mM phenylmethyl sulfonyl fluoride and 100 mM sodium vanadate, supplemented with cOmplete TM mini protease inhibitor cocktail tablets (Roche)). Homogenates were centrifuged at 20,000 x g for 20 minutes at 4 o C and the supernatant was transferred to a new tube and an equal volume of RIPA-B buffer (same as RIPA-A with the addition of 2% sarkosyl) was added to each. This was incubated for 1 hour at room temperature while rotating before centrifugation at 100,000 x g for 90 minutes at 4 o C using the Optima TM MAX-XP Ultracentrifuge (Beckman Coulter). The supernatant was then collected (sarkosyl-soluble fraction). The remaining pellet was re-suspended in 250 μL RIPA-C buffer (same as RIPA-A with the addition of 1% SDS), sonicated and spun at 1000 x g for 1 minute to remove bubbles. Samples were then collected (sarkosyl-insoluble fraction). For both extraction methods, total protein concentration was estimated by performing a BCA assay (Thermo Fisher Scientific), and absorbance was measured at 562 nm using a POLARstar OPTIMA plate reader (BMG Labtech).

5 μg of total protein was electrophoresed on a 10% Tris-glycine SDS-PAGE gel. Proteins were transferred to Immun-Blot Low Fluorescence PVDF membrane (Bio-Rad) using the Transblot Turbo Transfer System (Bio-Rad) and then stained with REVERT 700 Total Protein Stain according to the manufacturer's instructions (LI-COR). Total protein was imaged in the 700 nm channel with the Odyssey FC Imaging System (LI-COR) then membranes were washed and blocked for 1 hour in Odyssey  Blocking Buffer (LI-COR). Membranes were incubated in primary antibody solution overnight at 4°C by rocking. Membranes were washed with TBS-T and incubated with the secondary antibody for 1 hour at room temperature. Membranes were imaged in the 800 nm channel and fluorescence was quantified using the Image Studio™ Lite software (LI-COR).

RNF5 was covalently conjugated to Alexa Fluor 647 dye (Thermo Fisher Scientific) in PBS with 0.1M sodium bicarbonate, separated from free-dye and concentrated to volumes appropriate for injections, as described previously [9] .

To compare antibody localisation of the RNF5 with and without SUS +MB , K3 mice of 

To assess brain uptake of Alexa Fluor 647-conjugated RNF5, mouse brains were imaged using the Odyssey ® Fc Imaging System (700 nm laser, 2 minute scan time). Following this scan, whole brains were embedded in paraffin and sectioned coronally at 7 µm thickness using a microtome, followed by immunohistochemistry.

Mounted brain tissue was first dehydrated in a series of xylene and ethanol washes. Antigen retrieval was conducted in a domestic microwave (850W) in citrate buffer pH 5.8 for 15 min and was cooled to room temperature. Sections were analysed by immunohistochemistry as described [16] , using 2-3 sections per mouse. Briefly, sections were incubated with primary antibodies overnight. For immunofluorescence staining, sections were washed and incubated with respective secondary antibodies, then counterstained with DAPI nuclear stain (4', 6diamidino-2-phenylindole) (1:5,000). Alexa Fluor 647-conjugated RNF5 was detected in situ without additional amplification. For immunohistochemical staining, sections were stained using the nickel-diaminobenzidine (Nickel-DAB) method with no counterstain applied. All images were obtained on an automated slide scanner (Axio Imager Z2, Zeiss) using a Metafer Vslide Scanner program (Metasystems) at 20x magnification. Confocal Z-stack images were obtained on a Zeiss 510 confocal microscope at 63x magnification.

Image quantification and analysis were performed blinded using ImageJ software. For all analyses, a control with no primary antibody was used for thresholding. More specifically, for immunofluorescence staining quantification, the 'percentage (%) area positivity' for RN235-, AT8-and AT180-tau was obtained on thresholded images of the cortex and hippocampus using To assess neuronal health of mice that were administered RNF5-647 in the presence of ultrasound, '% area positivity' for MAP2 and Iba1 was quantified in areas that had RNF5 uptake versus areas that did not have RNF5 uptake using the 'area fraction' method on thresholded images. ROIs were manually drawn around areas that had RNF5 uptake and the same ROI was then used to quantify MAP2 and Iba1 in areas that did not have RNF5 uptake.

For the per neuron correlative analysis between MAP2 and RNF5, the median fluorescence intensity of 63x confocal Z-stack images was analysed. Background levels of RNF5 fluorescence from a 'no primary antibody control' was measured and subtracted from the experimental data. MAP2 was used to draw ROIs around individual neurons. Neurons that were positive for RNF5, however, had indistinguishable MAP2 staining, therefore, for these neurons, RNF5-647 was used to draw ROIs around individual neurons.

Statistical analyses were performed using GraphPad Prism 9.0 software. All data was analysed using one-way ANOVA followed by Tukey's multiple comparisons test or t-test where appropriate, except for the Rotarod test where a two-way ANOVA was used, followed by Tukey's multiple comparisons test. Correlations were assessed using the Spearman's correlation test. All values are reported as mean ± standard error of the mean (SEM). Outliers were removed using the ROUT method (Q=1).

We generated a pan-tau antibody, RNF5, against a region within the N-terminus of tau spanning amino acids 35-44 ( Figure 1A) . To characterise the RNF5 IgG, we first established the specificity of RNF5 by western blotting with mouse brain lysates. This demonstrated that RNF5 binds total tau in the brain lysates of both wild-type and K3 tau transgenic mice that express 0N4R tau with the K369I mutation, thus indicating that RNF5 binds both mouse and human tau ( Figure 1B) . Western blot analysis of cortical brain homogenate from AD brains and healthy controls demonstrates that RNF5 can bind multiple forms of tau, further confirming that RNF5 is a pan-tau antibody ( Figure 1C ). Furthermore, in cortical brain tissue, RNF5 binds to the characteristic neurofibrillary tangles in AD brains (indicated by yellow arrow heads) which is absent from healthy controls ( Figure 1D ). Surface plasmon resonance (SPR) revealed that RNF5 has an affinity of 159 nM to 2N4R mouse tau and 25.3 nM to human 2N4R tau ( Figure 1E and F). Thus, in vitro, RNF5 binds tau with 6-fold higher affinity than mouse tau ( Figure 1G ). Importantly, this data was fitted to a two-state reaction model as both the association and dissociation phases appear to have two components -a fast association followed by a slower association, and a fast dissociation followed by a slower dissociation.

To investigate the therapeutic efficacy of RNF5 and whether its ability to reduce tau pathology could be enhanced by increasing its concentration in the brain, we combined the intravenous delivery of RNF5 in mice with SUS +MB (Figure 2A ). Mice treated with either RNF5 alone or SUS +MB alone were used as controls, while litter-mate wild-type (WT) mice were used as additional control to the behavioural analysis. As reported previously, the weight of the K3 mice in this study was reduced compared to WT mice [14] . There was no significant difference, however, between the weight of treated mice compared to sham mice ( Figure 2B ). Weekly assessment of motor behaviour on the Rotarod revealed no significant improvements in latency to fall off the Rotarod for any of the treated groups compared to sham mice ( Figure 2C ). All treated mice, as well as the sham controls, improved at similar rates between 8 to 12 weeks of age. While the Rotarod can be used to test motor ability, coordination and limbic strength, we decided to also test the K3 mice in the activity monitor, to assess rearing ability (vertical time) and ambulation (ambulatory time), as well the grip strength test, to assess forelimb and hindlimb strength. K3 mice performed significantly worse in all tests compared to WT mice ( Figure 2D to G). End-point behavioural testing after completion of the treatment revealed no significant improvements in rearing ability, ambulation or grip strength between any of treatment groups compared to sham ( Figure 2D to G).

We next sought to investigate whether the treatment had any effect on total and phosphorylated tau species in K3 mice. The strain develops robust tau pathology particularly in the cortex and hippocampus, and to a lesser extent in the cerebellum [14] . As pathology is most robust in the former two areas, the cerebellum was not analysed. Firstly, using immunofluorescent analyses, we assessed 3 different phospho-tau antibodies, RN235, AT8 and AT180, all of which stain tau inclusions to varying degrees [17] and bind to epitopes of tau that are elevated in AD [18, 19] . Labelling with RN235, which detects Pick body-like inclusions in K3 mice and is specific for phospho-S235, revealed significant reductions (by approximately 2-fold) in the number of RN235-positive inclusions in both the cortex and hippocampus of treated mice, in comparison to sham mice and no significant difference between each of the treated groups ( Figure 3A and 4A). Labelling with AT8, which detects both Pick body-like inclusions and diffuse species of tau in K3 mice and is specific for phospho-S202/T205, revealed a significant reduction in the AT8-positive area in the cortex of the SUS +MB only (by 19-fold) and RNF5 only (by 5-fold) treated groups but surprisingly, not in the RNF5+SUS +MB treated group, compared to sham ( Figure 3B ). In the hippocampus, however, the AT8 positive area was significantly reduced only in the SUS +MB -treated mice (by 2.5-fold) compared to sham mice, as well as compared to RNF5-and combination-treated mice ( Figure 4B ). Finally, labelling with AT180, which predominantly detects pre-NFT-like inclusions and diffuse species of tau in K3 mice and is specific for phospho-T231, revealed significant reductions in the AT180-positive area in the cortex for all treatment groups (RNF5, SUS +MB and RNF5+SUS +MB ) compared to sham mice (by approximately 2.5-fold), with no significant difference between RNF5 and RNF5+SUS +MB treated mice ( Figure 3C ). In the hippocampus, there was a trend towards a reduction in AT180positive area across all treatment groups in comparison to sham mice ( Figure 4C ), however this did not reach significance. Taken altogether, these results indicate that while RNF5, SUS +MB and the combination of RNF5+SUS +MB are all successful at reducing tau pathology, the combination treatment provided no added benefit.

To complement the above data, we next assessed the levels of total and phosphorylated tau in sarkosyl-soluble and insoluble fractions of whole brain lysates (excluding the cerebellum).

Sarkosyl extraction is an established method to investigate soluble and insoluble (aggregated) tau in brain. In the sarkosyl-soluble fractions, which contain soluble tau, no differences were found for total mouse and human tau as detected with antibody Tau-5; also, no differences were evident in AT8 phosphorylation between any groups ( Figure 5A ). Detection of AT180 positive tau, however, revealed a trend towards a reduction in RNF5-treated mice compared to sham mice ( Figure 5A ). Most interestingly, levels of sarkosyl-soluble AT180-tau were significantly increased in the combination RNF5+SUS +MB group, compared to mice treated with RNF5 only ( Figure 5A ). In the sarkosyl-insoluble fraction, which contains aggregated species of tau, no differences were found for total tau (Tau-5) or phospho-tau (AT8 and AT180) between any groups ( Figure 5B ).

Several in vitro and in vivo studies have demonstrated that antibody-tau complexes can be phagocytosed by microglia through the interaction of the antibody Fc region with Fc receptors on microglia [20, 21] . We have also previously reported that SUS +MB enhances microglial activation in APP23 mice and this in turn aided plaques to be cleared through microglial phagocytosis [6] . Taken together, we speculated that the lack of enhanced efficacy in the combination group may be due to overactivation of microglia. Therefore, we sought to investigate whether microglial activation was enhanced in the combination treatment cohort compared to mice administered with the single treatments, by staining for Iba1, a cytoplasmic microglial marker ( Figure 6A ). We found that 'microglial count' and 'microglial percentage area', a proxy of microglial activation, were elevated in mice that had been treated with the combination compared to WT, Sham and RNF5-treated mice, but not compared to SUS +MB mice ( Figure 6C-E) , suggesting that the SUS +MB component induces microglial activation. In addition, the microglia in mice treated with RNF5 were comparable to sham treated mice despite observing significant reductions in p-tau in RNF5-treated mice compared to sham. We speculated that this elevation of Iba1-positive microglia in the groups treated with SUS +MB and RNF5+SUS +MB may indicate increased phagocytic activity; however, no significant differences were seen in the area of immunoreactivity for the microglial lysosomal marker CD68 between any groups ( Figure 6B and F) . Additionally, we observed no histological damage in any of the treated mice as shown by hematoxylin and eosin (H&E) staining (not shown). Overall, in the absence of a difference in the microglia between the SUS +MB only and RNF5+SUS +MB treated mice, the effects on microglia seen in the combination group are most likely mediated by the SUS +MB treatment. Furthermore, these results may suggest a microgliaindependent mechanism of RNF5-mediated tau pathology reduction or the possibility of an immediate short response following delivery that has resolved itself over time, which could explain the inability to detect this effect. However, since these mice were sacrificed one week following their final treatment, what we have observed in these mice so far may not be reflective of the mechanisms taking place in the hours immediately following treatment.

Therefore, to gain a deeper understanding of the events immediately after delivery, mice were treated with fluorescently labelled RNF5, with or without SUS +MB , and sacrificed at earlier time-points.

We first sought to assess the localisation of fluorescently labelled RNF5 in the brain immediately after antibody delivery. To explore this, we administered fluorescently labelled RNF5 with or without SUS +MB and sacrificed the mice 2 and 8 hours after treatment. At 2 hours, using an endothelial marker for detecting blood vessels (lectin) [22] , RNF5 fluorescence is lower in the blood vessels when delivered without SUS +MB , compared to RNF5 delivered with SUS +MB (Figure 7A ). This is further evidenced by the fact that the brightness and contrast of the images had to be enhanced for RNF5 to be detectable in the RNF5 only group ( Figure 7B ). By 8 hours, RNF5 is no longer detectable when delivered without SUS +MB suggesting that RNF5 may be distributed diffusely throughout the brain parenchyma at levels below detection ( Figure 7B ). On the other hand, when RNF5 is delivered with SUS +MB , while still localised to blood vessels at 8 hours, a portion of RNF5 is detectable in the brain parenchyma ( Figure 7C ) and in cells ( Figure 7D ). As RNF5 is detectable in the brain parenchyma at 8 hours following SUS +MB treatment suggests increased brain penetration of RNF5 compared to mice treated with RNF5 without SUS +MB , as expected. However, as RNF5 is detected in the blood vessels at this time-point, it might get trapped in the vessels following treatment with SUS +MB , which does not occur without ultrasound. In fact, we were able to detect RNF5 in blood vessels at 24 hours post-SUS +MB (not shown). Furthermore, as RNF5 was only detectable in the brain parenchyma at the 8hour, but not the 2-hour time-point following SUS +MB , it suggests that there is a sequential process by which IgG traverses the BBB, rather than an immediate transfer between the brain endothelial cells in a paracellular fashion.

To determine whether RNF5 delivered with SUS +MB is localised to neurons as a target celltype with tau deposition, neuronal co-staining was conducted with MAP2. Surprisingly, we observed that the cells in the hippocampus that had internalised RNF5 at 8 hours post-treatment were MAP2-negative ( Figure 8A ). They were also Iba1-and GFAP-negative indicating that RNF5 was not being internalised by microglia or astrocytes in these areas either (not shown).

Given that MAP2 loss is associated with neuronal degeneration as observed in animal models of stroke [23] , we determined whether the lack of MAP2 in these cells may be due to poor neuronal health. Therefore, we performed H&E staining in mice sacrificed at 2 and 8 hours following RNF5+SUS +MB treatment and eosinophilic neurons or 'red' neurons in the hippocampus were quantified. This region was analysed as a proxy for overall brain health because we had previously observed that the pyramidal cell layer in the hippocampus was particularly vulnerable to neuronal damage compared to other brain regions. Surprisingly, we in microglial infiltration in the areas in which RNF5 had extravasated/been internalised into neurons compared to areas in which RNF5 was not present, which indicates that inflammation and possible microgliosis were occurring in areas where RNF5 was present ( Figure 8C) . We also observed a 2.5-fold reduction in the percentage area of MAP2 in these areas compared to areas without RNF5 ( Figure 8D ). To investigate this in more depth, we performed a per neuron correlative analysis and observed a negative correlation between MAP2 and RNF5 (r = -0.543, p<0.0001) indicating that neuronal degeneration is associated with RNF5 uptake ( Figure 8E ).

Developing effective treatments for CNS disorders continues to present a major challenge due to the role of the BBB in preventing the passage of drugs, including biologics such as antibodies, across the brain. Therefore, in this study, we not only characterised the efficacy of a novel anti-tau mAb, RNF5, but also when combined with SUS +MB to increase brain penetration of the antibody, anticipating an increased therapeutic effect. Our major finding was that treatment with either RNF5 or SUS +MB alone, broadly speaking, reduced phosphorylated tau levels in K3 mice to a similar extent, whereas the combination treatment (RNF5+SUS +MB ) had no additive effects, despite an increase in antibody uptake. Given that K3 mice display both tau-induced cognitive and motor impairments, we used the latter as a proxy for a potential therapeutic effect. The significant reductions in pathology however were not reflected by an overall improvement in motor impairment. We speculate that given the pronounced, progressive pathology, the achieved reductions in pathological tau were not sufficient to improve the behavioural deficits. Furthermore, given the reported degeneration of cerebellar basket cells as well as of the sciatic nerve in the K3 mice [14, 24] , it is possible that SUS +MB sonication did not target all of this tissue, and similarly, the tissue may not have been equally accessible to the antibody.

In the absence of SUS +MB , we demonstrate that intravenous injections of RNF5 alone was able to reduce tau pathology in K3 mice. This highlights that RNF5 is an effective treatment by itself. Importantly, the RNF5 signal was not observed in the brain parenchyma of mice in the absence of SUS +MB . This is consistent with what has been reported for other tau IgGs in clinical development [25, 26] and suggests that IgG levels may be below the detection limit but still sufficient to reduce phosphorylated tau.

It is not possible to conclusively determine the mechanism of action of RNF5 and its role in tau reduction in this study. When delivered with SUS +MB , RNF5 was only internalised into degenerating neurons and this is in line with other publications that have shown that antibodies are not readily internalised into neurons [25] or are only detected in dying neurons [27] which presents the possibility that RNF5 is acting in the extracellular space. Since we did not detect any increase in the phagocytic activity of the microglia in the RNF5-treated mice, we speculate that microglial phagocytosis may not be playing a significant role in tau clearance, despite being one of the well-documented mechanisms of action following passive immunotherapy with tau antibodies [20, 21, 28, 29] . Alternatively, it may be possible that RNF5 is acting in the extracellular space to prevent the internalisation of tau seeds in the K3 mice, thus preventing the formation of insoluble tau species. This is additionally supported by other reports that N-terminal tau antibodies prevented tau seeding in cell-based assays [30] as well as in passive immunisation studies in tau transgenic mice [25, [31] [32] [33] .

Focused ultrasound is known to open the BBB by disrupting tight junction integrity [34] as well as increasing vesicular formation [35, 36] , thus allowing the transport of large molecules across the BBB through paracellular and transcellular routes, respectively [37, 38] .

While many factors dictate uptake, including size, charge and shape, a recent study has shown a role for transcytoplasmic, caveolin-mediated mechanisms in the transport of large cargoes [38] . Given that paracellular transport is likely an instant process [37] but we did not observe parenchymal localisation of RNF5 at 2 hours post-treatment, this suggests that RNF5 traversed the BBB through vesicle-mediated transcytosis. The specifics of IgG transport across the brain following ultrasound, however, has not been determined and therefore requires further investigation.

Whilst our study and others suggest that having increased IgG concentrations in the brain is beneficial, our results seem to also suggest that attempting to deliver a high localised concentration of IgG to the brain may be detrimental using the experimental conditions applied.

In this study, RNF5 was delivered at a dose of 15 mg/kg (375 µg for a 25 g mouse). In comparison to other studies that have delivered antibodies retro-orbitally, administered doses were lower without ultrasound (10 mg/kg, [39] ) and even lower in combination with ultrasound (40 µg [12] and 5 mg/kg [13] ). This suggests the dose of RNF5 delivered into the brain may have been too high and that the benefit of SUS +MB might have reached a threshold in that attempting to deliver too much antibody into the brain may then become toxic. This is especially the case because RNF5 is of a murine IgG2b isotype, and it is important to note that high concentrations of a high effector isotype antibody may ultimately be deleterious, as we have demonstrated previously [40] . By further exploring the possibility of toxicity, on top of observing acute neuronal degeneration, we also found that microglia had infiltrated the areas in the brain parenchyma into which RNF5 had extravasated. This may explain the lack of a combinatorial effect in the RNF5+SUS +MB group which is supported by studies which have shown that inflammation drives tau pathology [41, 42] .

Interestingly, when RNF5 was delivered with SUS +MB , we observed that as RNF5 internalisation into neurons increased over time, so did the extent of neuronal degeneration.

FcR-mediated endocytosis [28, [43] [44] [45] [46] , internalisation mediated by electrostatic interactions with the plasma membrane [47] and through lipid rafts on the surface of the plasma membrane [48] . However, neurons are not specialised to express nor do they cope well with high intracellular concentrations of IgG, as has been shown by Elmer and colleagues, where viralmediated expression of IgG antibodies resulted in glycoprotein accumulation and subsequent neuronal degeneration [49] . While RNF5 was not expressed intracellularly in our study, it is possible that forcing large amounts of IgG into cells can be detrimental to neuronal health due to neurons being incapable of processing large quantities of glycoprotein.

Since we also observed RNF5 in blood vessels up to 24 hours following SUS +MB , we also speculate that neuronal degeneration and inflammation could have been induced through hypoxia-induced ischemia due to the accumulation and subsequent trapping of IgG within blood vessels. Since tau accumulation in cerebral microvasculature has been shown in human patients with AD, progressive supranuclear palsy and Lewy body dementia, as well as in the Tg2576 tau transgenic mouse model [50] , it is possible that RNF5 binds to tau within brain endothelial cells (although it is not normally detected at high levels), and the resulting overaccumulation of IgG within the microvasculature is occluding the vessels, preventing the transfer of oxygen, and causing neuronal degeneration as a result of ischemia. The loss of the somatodendritic marker MAP2 seen in our study following IgG delivery with SUS +MB is a phenomenon that is characteristically observed in models of stroke where a middle cerebral artery occlusion has been performed [23] , and in humans who have undergone various modes of hypoxia-ischemia-related deaths [51] . In the latter study, it has been shown that the hippocampus is selectively vulnerable to oxygen deprivation compared to other brain regions, which is in line with the current study given that we saw selective neuronal damage in the hippocampus. On the other hand, this also implies that the hippocampus may be more susceptible to antibody uptake and/or the effects of SUS +MB . In addition, the presence of the Fc domain may play a role in IgG entrapment within microvessels. This is supported by several studies that have shown that IgG efflux from the brain to blood is mediated through neonatal Fc receptors (FcRn) expressed on both luminal (blood-side) and abluminal (brain-side) membranes of the brain endothelial cells that line cerebral blood vessels [52] , which, in our study, may contribute to the presence of RNF5 in blood vessels for a long period of time following SUS +MB [53] [54] [55] .

In conclusion, further studies are warranted to better understand what dictates safety and efficacy of anti-tau antibody therapeutics and how this is impacted by the delivery technique. Importantly, it will also be crucial to assess whether these mechanisms are taudependent and the role a pre-existing pathology has. As we have previously shown that delivery of a tau antibody in a single chain format with SUS +MB increases its efficacy [10] Overall, our findings suggest that while ultrasound-mediated delivery of RNF5 IgG markedly enhances brain uptake, the localised increase in IgG concentration in the brain may be facilitating some opposing mechanisms that counteract the combinatorial therapeutic effects one would expect following combination treatment. However, given the plethora of variables involved in this study (which includes antibody concentration, antibody isotype, ultrasound parameters, microbubble distribution, and differences in blood flow and tissue oxygenation), it is difficult to pinpoint just one cause for the lack of an additive effect. While more research is required to elucidate the feasibility of this technique, this work provides an important initial understanding for the optimisation of this delivery modality. Hindlimb grip strength, All data represented as mean ± SEM (WT, n=8; Sham, n=8; RNF5, n=8; SUS +MB , n=8; RNF5+SUS +MB , n=7; ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05, one-way ANOVA with Tukey's multiple comparison's test). 

The blood-brain barrier: Physiology and strategies for drug delivery

Insights into the mechanisms of action of anti-Abeta antibodies in Alzheimer's disease mouse models

Open questions for Alzheimer's disease immunotherapy

Noninvasive localized delivery of Herceptin to the mouse brain by MRI-guided focused ultrasound-induced blood-brain barrier disruption

Noninvasive and targeted delivery of therapeutics to the brain using focused ultrasound

Scanning ultrasound removes amyloid-β and restores memory in an Alzheimer's disease mouse model

Safety and Efficacy of Scanning Ultrasound Treatment of Aged APP23 Mice

Repeated ultrasound treatment of tau transgenic mice clears neuronal tau by autophagy and improves behavioral functions

Ultrasound-mediated blood-brain barrier opening enhances delivery of therapeutically relevant formats of a tau-specific antibody

Combined effects of scanning ultrasound and a tau-specific single chain antibody in a tau transgenic mouse model

Noninvasive and localized neuronal delivery using short ultrasonic pulses and microbubbles

Antibodies targeted to the brain with image-guided focused ultrasound reduces amyloid-beta plaque load in the TgCRND8 mouse model of Alzheimer's disease

A comparative study of the effects of Aducanumab and scanning ultrasound on amyloid plaques and behavior in the APP23 mouse model of Alzheimer disease

Parkinsonism and impaired axonal transport in a mouse model of frontotemporal dementia

A preparation of Alzheimer paired helical filaments that displays distinct tau proteins by polyacrylamide gel electrophoresis

Parkinsonism and impaired axonal transport in a mouse model of frontotemporal dementia

A Novel Antibody Targeting Tau Phosphorylated at Serine 235 Detects Neurofibrillary Tangles

Tau phosphorylation: the therapeutic challenge for neurodegenerative disease

Specific tau phosphorylation sites correlate with severity of neuronal cytopathology in Alzheimer's disease

Distinct Therapeutic Mechanisms of Tau Antibodies: Promoting Microglial Clearance Versus Blocking Neuronal Uptake

Microglial internalization and degradation of pathological tau is enhanced by an anti-tau monoclonal antibody

Use of labeled tomato lectin for imaging vasculature structures

Spatio-temporal overview of neuroinflammation in an experimental mouse stroke model

Sodium selenate mitigates tau pathology, neurodegeneration, and functional deficits in Alzheimer's disease models

Anti-tau antibodies that block tau aggregate seeding in vitro markedly decrease pathology and improve cognition in vivo

Prevention of tau seeding and propagation by immunotherapy with a central tau epitope antibody

Therapeutic antibody targeting microtubule-binding domain prevents neuronal internalization of extracellular tau via masking neuron surface proteoglycans

Antibody-mediated clearance of tau in primary mouse microglial cultures requires Fcγ-receptor binding and functional lysosomes

Humanized tau antibodies promote tau uptake by human microglia without any increase of inflammation

A common antigenic motif recognized by naturally occurring human VH5-51/VL4-1 anti-tau antibodies with distinct functionalities

Passive immunization targeting the N-terminal projection domain of tau decreases tau pathology and improves cognition in a transgenic mouse model of Alzheimer disease and tauopathies

Tau passive immunization inhibits not only tau but also Aβ pathology

Tau passive immunization blocks seeding and spread of Alzheimer hyperphosphorylated Tau-induced pathology in 3 × Tg-AD mice

Effect of focused ultrasound applied with an ultrasound contrast agent on the tight junctional integrity of the brain microvascular endothelium

Cellular mechanisms of the blood-brain barrier opening induced by ultrasound in presence of microbubbles

The role of caveolin-1 in blood-brain barrier disruption induced by focused ultrasound combined with microbubbles

The size of blood-brain barrier opening induced by focused ultrasound is dictated by the acoustic pressure

Role for caveolin-mediated transcytosis in facilitating transport of large cargoes into the brain via ultrasound

Comparison of the lateral tail vein and the retro-orbital venous sinus routes of antibody administration in pharmacokinetic studies

Tau antibody isotype induces differential effects following passive immunisation of tau transgenic mice

NLRP3 inflammasome activation drives tau pathology

Mechanistic Studies of Antibody-Mediated Clearance of Tau Aggregates Using an ex vivo Brain Slice Model

Antibody uptake into neurons occurs primarily via clathrin-dependent Fcgamma receptor endocytosis and is a prerequisite for acute tau protein clearance

Two novel Tau antibodies targeting the 396/404 region are primarily taken up by neurons and reduce Tau protein pathology

Dynamics of Internalization and Intracellular Interaction of Tau Antibodies and Human Pathological Tau Protein in a Human Neuron-Like Model

Tau antibody chimerization alters its charge and binding, thereby reducing its cellular uptake and efficacy

Neuronal uptake of tau/pS422 antibody and reduced progression of tau pathology in a mouse model of Alzheimer's disease

Gene delivery of a modified antibody to Aβ reduces progression of murine Alzheimer's disease

Cerebral Microvascular Accumulation of Tau Oligomers in Alzheimer's Disease and Related Tauopathies

Microtubule-associated protein 2 (MAP2)--a promising approach to diagnosis of forensic types of hypoxia-ischemia

Blood-brain barrier transport machineries and targeted therapy of brain diseases

Mediated efflux of IgG molecules from brain to blood across the blood-brain barrier

IgG-assisted age-dependent clearance of Alzheimer's amyloid beta peptide by the blood-brain barrier neonatal Fc receptor

Efflux of monoclonal antibodies from rat brain by neonatal Fc receptor, FcRn

The authors would like to thank Dr Martina Jones from the Queensland node of the National Biologics Facility for assistance with performing the SPR experiment. We would also like to

The authors declare that they have no competing interests.