A novel synaptopathy- defective synaptic vesicle protein trafficking in the mutant CHMP2B mouse model of frontotemporal dementia

Mutations in the ESCRT- III subunit CHMP2B cause frontotemporal dementia (FTD) and lead to impaired endolysosomal trafficking and lysosomal storage pathology in neurons. We investigated the effect of mutant CHMP2B on synaptic pathology, as ESCRT function was recently implicated in the degradation of synaptic vesicle (SV) proteins. We report here that expression of C- terminally truncated mutant CHMP2B results in a novel synaptopathy. This unique synaptic pathology is characterised by selective retention of presynaptic SV trafficking proteins in aged mutant CHMP2B transgenic mice, despite significant loss of postsynaptic proteins. Furthermore, ultrastructural analysis of primary cortical cultures from transgenic CHMP2B mice revealed a significant increase in the number of presynaptic endosomes, while neurons expressing mutant CHMP2B display defective SV recycling and alterations to functional SV pools. Therefore, we reveal how mutations in CHMP2B affect specific presynaptic proteins and SV recycling, identifying CHMP2B FTD as a novel synaptopathy. This novel synaptopathic mechanism of impaired SV physiology may be a key early event in multiple forms of FTD, since proteins that mediate the most common genetic forms of FTD all localise at the presynapse.


| INTRODUC TI ON
Synaptopathies are disorders resulting from dysfunction of synapses and are associated with the earliest stages of multiple neuronal diseases. Synapse loss is a key feature in dementia, with synaptic dysfunction preceding neuronal death (Sheng et al., 2012;Terry, 2000).
In Alzheimer's disease (AD), synapse degradation occurs before formation of amyloid deposits, and this degradation is the best correlate of cognitive decline in both animal models and AD patients (Dekosky & Scheff, 1990;Mucke et al., 2000;Serrano-Pozo et al., 2011;Terry et al., 1991). In frontotemporal dementia (FTD) patients have lower synaptic density in the superficial layers of the frontal cortex (Ferrer, 1999;Liu et al., 1996). However, it is not clear what is causing the damage to synapses in FTD, nor how this damage contributes to the clinical outcome for the disorder. The link is important because FTD is the second most common form of young-onset dementia (Harvey et al., 2003;Ratnavalli et al., 2002). FTD is characterised by atrophy of the frontal and temporal lobes that results in personality, behaviour and language changes (McKhann et al., 2001;Neary et al., 1998). Several genetic mutations cause FTD. Mutations in the genes that encode tau (MAPT), progranulin (GRN) and C9orf72 are the most common, while additional rare mutations have been identified in valosin-containing protein (VCP), TDP-43 (TARDBP), fused in sarcoma (FUS) (Rohrer & Warren, 2011), TANK-binding kinase 1 (TBK1) (Gijselinck et al., 2015;Le Ber et al., 2015;Pottier et al., 2015;van der Zee et al., 2017) and charged multivesicular body protein 2B (CHMP2B) (Lindquist et al., 2008;Skibinski et al., 2005).
A mutation in CHMP2B, found in a Danish cohort, causes an autosomal dominant form of FTD (Lindquist et al., 2008;Skibinski et al., 2005). The mutation disrupts a splice acceptor site and generates a C-terminally truncated variant of the protein. Physiological levels of this mutant CHMP2B are sufficient to recapitulate the patient phenotype in mice, producing axonal degeneration, gliosis and progressive neurodegeneration (Clayton et al., 2015(Clayton et al., , 2017Gascon et al., 2014;Ghazi-Noori et al., 2012). The mechanism by which CHMP2B, a subunit of the endosomal sorting complex required for transport-III (ESCRT-III), causes these neurological deficits is not known. ESCRTs 0-III are highly conserved multi-subunit protein complexes that mediate numerous cellular processes involving scaffolding membrane deformation and budding (Vietri et al., 2020). Impaired endolysosomal trafficking is seen in primary cortical cultures from the mutant CHMP2B mouse model (Clayton et al., 2018), but the link between impaired endolysosomal trafficking and neuronal cell death is not well characterised.
Communication between neurons is reliant on the release of neurotransmitters from the presynaptic terminal through the fusion of synaptic vesicles (SVs) with the presynaptic plasma membrane.
Maintaining the fidelity of neuronal communication depends on the efficient endocytosis of SV membrane and cargo components to sustain further rounds of exocytosis (Südhof, 2013). This finely tuned cycle is dependent on the action of a number of endocytosis proteins, and defects in this process have been implicated in numerous neurodevelopmental and neurodegenerative conditions (Bonnycastle et al., 2020;Melland et al., 2020). Different modes of SV retrieval with characteristic molecular requirements have been described, and are known to be activated by different stimulation paradigms; clathrin-mediated endocytosis, kiss-and-run, activity dependent bulk endocytosis (ADBE) and ultrafast endocytosis have all been reported at central nerve terminals (Chanaday et al., 2019).
While molecular mechanisms that control SV endocytosis have been described, little is known about the process by which SV proteins are trafficked for degradation. Interestingly, the ESCRT complex was recently implicated in the degradation of a subset of SV proteins (Sheehan et al., 2016). However, the impact of mutation in the ESCRT subunit CHMP2B on SV physiology is unknown.
We show that mutant CHMP2B leads to synaptic loss, and a concurrent retention of SV-associated proteins in aged CHMP2B mice. Presynaptic endosomes are significantly increased in mutant CHMP2B primary cortical neurons. In neurons expressing mutant CHMP2B the total recycling pool of SVs is increased, fewer SVs fuse during a defined stimulus and SV endocytosis starts to fail when the system is stressed. Thus, we report here that defects in SV protein physiology caused by mutant CHMP2B affect presynaptic SV trafficking, leading to synaptopathy in CHMP2B FTD. followed by destruction of the brain. Mice were housed in conventional caging.

At UCL mouse work was performed under UK government Home
Office project licence 7009014, approved by local Animal Welfare and Ethical Review Body. Mice were housed in a category 3 SPF facility in individually ventilated cages under negative pressure in groups of 3-5 animals with environmental condition targets of temperature 20 ± 2°C, relative humidity 55% ± 10%, 12:12 hour photo period. Mice were provided with water and pelleted diet ad lib. All cages are provided with environmental enrichment in the form of nesting material, chew blocks and mouse houses.
The previously described mutant CHMP2B Intron5 expressing mouse line Tg153 (Ghazi-Noori et al., 2012) was backcrossed over 10 generations to C57Bl6J and was maintained as a homozygous line.

| Whole brain homogenates
10% (w/v) brain homogenates (minus the cerebellum and olfactory bulb) were prepared in phosphate-buffered saline containing cOmplete EDTA-free protease inhibitors (Roche) using a TissueRuptor (Qiagen) to make a 10% w/v solution and combined 1:1 with 2% sarkosyl (N-lauroylsarcosine) in D-PBS. Benzonase (50 U/ml; Novagen) was added to remove DNA and the homogenates incubated with constant agitation at 37°C for 1 h, followed by ultracentrifugation at 100 000 g for 30 min at 4°C. 2× Laemmli sample buffer was added to the supernatant and heated at 100°C for 10 min prior to sodium dodecyl sulphate-polyacrylamide gel electrophoresis.
Membranes were blocked for 1 h at room temperature with 5% milk/ PBS-T and probed with primary antibodies (see Table 1) overnight at 4°C in 1% milk/PBS-T with constant agitation. Detection was performed with anti-mouse and anti-rabbit-IRDye 680 and 800CW (LiCor), 1:10 000 fluorescent secondary antibodies for visualisation using a LiCor Odessey scanner. Quantification was performed using Licor software, bands were normalised to total protein loading, and averages were taken of three mice per genotype.

| Primary cortical cultures
Primary cortical cultures were prepared from mixed sex mice (postnatal day 0 or day 1), with 3 mice used per culture. Mice were euthanised by decapitation. Briefly, the cortices were dissected, pooled, digested in trypsin (Sigma) and triturated with a fine fire polished Pasteur pipette to achieve a single cell suspension. Cells were plated in a minimal volume of Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin and 1% Glutamax (all Invitrogen), at a density of 1000 cells/ mm 2 on coverslips coated with polyd-lysine (Sigma). 1-2 h after plating, maintenance medium of Neurobasal A containing 2% B27, 0.25% penicillin/streptomycin and 0.25% Glutamax (all Invitrogen) was added to the cells. Neurons were cultured at 37°C and 5% CO 2 .

| Statistics
Statistical analyses were performed in Graphpad. Statistical tests and N numbers are indicated in the figure legends and were used to calculate significance values with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Western blot data were not assessed for normality because of low sample size. Electron microscopy data were assessed for normality using the Shapiro-Wilk normality test. Sample size calculations were not performed, and N numbers were based on previous studies of a similar nature (Clayton et al., 2015(Clayton et al., , 2018Nicholson-Fish et al., 2015). Comparisons between non-transgenic and mutant CHMP2B cultures were performed by a researcher blinded to sample origin.

| Study design
This study was not pre-registered, randomisation was not performed to allocate treatments to different experimental groups, no exclusion criteria were pre-determined, no test for outliers was performed, and the study was exploratory.

| Reduction of postsynaptic, but not presynaptic proteins in aged CHMP2B mutant mice
The ESCRT complex was recently implicated in the degradation of presynaptic SV proteins (Sheehan et al., 2016); however, it is not known if mutation in CHMP2B disrupts this highly specialised membrane trafficking pathway. Characteristic FTD-like pathology is well established at 18 months in mutant CHMP2B mice, with significant accumulation of p62, ubiquitin and lysosomal storage related aggregates (Clayton et al., 2015;Ghazi-Noori et al., 2012). We have reported significant cortical volume loss, despite established gliosis, in aged mutant CHMP2B mouse brains, which is because of significant neuronal loss (Clayton et al., 2017).
In order to determine whether mutant CHMP2B has an effect on presynapses, we first investigated the levels of pre-and post-synaptic proteins. As expected, there was a significant reduction in the postsynaptic markers PSD-95 (control 1.00 ± 0.07,  Table 2).

| Aged CHMP2B mutant mice specifically retain a subset of SV trafficking proteins
To test whether the observed retention of SV-associated proteins was because of the selective preservation of entire presynaptic units, despite observed loss of markers of the postsynaptic compartment, we blotted for additional presynaptic components. MUNC-18 and Syntaxin 1A localise to the plasma membrane of the presynaptic terminal, and mediate tethering of SVs for fusion events, but do not participate in SV recycling (Kavanagh et al., 2014). These presynaptic components were significantly decreased in mutant CHMP2B mouse brain (MUNC-18 control 1.00 ± 0.04, CHMP2B Intron5 0.76 ± 0.02, p = 0.004, Syntain 1A control 1.00 ± 0.05, CHMP2B Intron5 0.86 ± 0.03, p = 0.036) Figure 1c), suggesting that the observed F I G U R E 1 Reduction of postsynaptic and select presynaptic proteins in aged CHMP2B mutant mice. Brain homogenates from 18-monthold CHMP2B Intron 5 and non-transgenic female mice were extracted in 1% sarkosyl and immunoblotted for markers of the postsynaptic and presynaptic compartments. The amount of PSD-95 and Homer (a) in CHMP2B Intron5 brain homogenates is significantly reduced when compared to non-transgenic controls. Brain homogenates were additionally immunoblotted for Amphiphysin, Synapsin, Endophilin and Synuclein (b). No significant difference was seen between non-transgenic and CHMP2B Intron5 mice for Amphiphysin, Synapsin or Endophilin, while Synuclein was significantly increased in aged mutant CHMP2B homogenates. Brain homogenates were also immunoblotted for MUNC-18 and Syntaxin 1A (c), which were significantly decreased in aged mutant CHMP2B mouse brain. Mean ± SEM is shown. Unpaired t-test, *p < 0.05, **p < 0.01, ns = not significant. N = 3 mice of each genotype TA B L E 2 Summary of statistics

SVs rest
Mann-Whitney test 0.9985

SVs recovery
Mann-Whitney test 0.4684

Figure 3
SypHy ΔF/F 0 Two-way ANOVA ANOVA retention of SV-associated presynaptic proteins in mutant CHMP2B mouse brain may be restricted to components of SV endocytosis, and not extend to all parts of the presynaptic terminal.
No alteration to postsynaptic or presynaptic components was observed at 6 months of age in mutant CHMP2B mouse brains ( Figure S1). This shows that these changes are a result of an ongoing degenerative process and not a developmental defect.
These data show that SV trafficking components are selectively retained in aged mutant CHMP2B mouse brain. In conjunction with the recent observation that the ESCRT complex mediates degradation of SV components (Sheehan et al., 2016), this led us to investigate the physiology of SV trafficking in mutant CHMP2B neurons.

| Mutant CHMP2B synapses are characterised by increased number of presynaptic endosomes
We reasoned that this increase in a subset of SV proteins may indicate an alteration to SV trafficking pathways. To assess this, we used electron microscopy to examine the ultrastructure of the presynaptic SV pool in primary cortical cultures to determine whether evidence of dysfunction was apparent.
We analysed synapses of mature primary cortical mutant  (Figure 2i). In addition, no significant difference in the diameter of the presynaptic endosomes was seen in any of the conditions assessed ( Figure   S3). Therefore, the increased number of endosomes was not because of stalled SV generation from these structures. Although ESCRTs are important for formation of multivesicular bodies (MVBs), no difference was seen in the number of MVBs seen per synapse ( Figure S4).
Therefore, our ultrastructural analysis of mutant CHMP2B presynaptic anatomy reveals an increased number of endosomes in resting CHMP2B nerve terminals. This increase was not because of ADBE, since generation of endosomes directly from the presynaptic membrane is unaffected by the presence of mutant CHMP2B.
Instead, there appears to be a permanent increase in presynaptic endosomes, as evidenced by their sustained increase above baseline levels following stimulation.

| SV exocytosis is significantly impaired in mutant CHMP2B primary cultures
The increased number of endosomes within CHMP2B mutant nerve terminals, in conjunction with retention of select presynaptic SV proteins, suggests dysfunction in the clearance of SV proteins by the endolysosomal system. To determine whether this impacted the dynamics of SV turnover, we co-transfected primary cultures with CHMP2B constructs and a genetically encoded fluorescent reporter of SV trafficking called pHluorins. These reporters are SV cargoes linked to a lumenal pH-sensitive EGFP, allowing them to fluoresce when at the plasma membrane but their fluorescence is quenched when SVs reacidify following endocytosis (Kavalali & Jorgensen, 2014). Fluorescence quenching can be used as an estimate of the speed of SV cargo retrieval, as the endocytosis of SVs is rate limiting compared to SV acidification (Atluri & Ryan, 2006;Egashira et al., 2015).
Mutant and wildtype CHMP2B expressing neurons that co- Taken together, these results show that fewer SVs fuse during a defined stimulus and SV endocytosis starts to fail when the system is stressed. Additionally, the total recycling pool of SVs is increased in mutant CHMP2B expressing neurons-this may occur as a compensatory mechanism to the endolysosomal block.

| Subneuronal region-specific endolysosomal defects in FTD
Recent studies have shown that FTD causative genes converge on dysfunction of the endolysosomal system. We have recently shown that mutant CHMP2B causes defective neuronal endolysosomal trafficking, leading to a lysosomal storage pathology (Clayton et al., 2015(Clayton et al., , 2018. However, the molecular mechanism by which defective endolysosomal trafficking specifically causes neuronal dysfunction, rather than global cell death, remains to be described.    Granseth & Lagnado, 2008) (Milovanovic et al., 2018), and in fact, all of these particular SV-associated proteins have been postulated to be possible phase separation competent proteins (Milovanovic & De, 2017

| SV trafficking dysfunction in dementia
Numerous protein networks and signalling cascades contribute to ensure the tight regulation of SV trafficking. Several proteins associated with neurodegenerative disease have been implicated in the SV cycle, suggesting that disruption of this highly tuned SV cycle precedes synapse loss and eventual neurodegeneration in several dementias. For example, α-synuclein over-expression or mutation is associated with several dementias, including Parkinson's disease, Lewy body dementia, multiple system atrophy and some variants of AD. Although the specific physiological role of α-synuclein is not yet well understood, disruption of α-synuclein is associated with defects in SV cycling (Cabin et al., 2002;Fusco et al., 2016;Scott & Roy, 2012;Wang et al., 2014).
The recently published observations that the FTD associated proteins progranulin, tau and C9orf72 have all been found at the presynaptic terminal raises the possibility that presynaptic endolysosomal defects impact SV trafficking in numerous genetic forms of FTD (Frick et al., 2018;Petoukhov et al., 2013;Zhou et al., 2017).
Supporting this possibility, pathogenic tau has recently been shown to alter presynaptic functions through binding with SVs and reducing their mobility and release rate (Zhou et al., 2017). Additionally, glycine-alanine dipeptides associated with C9orf72 mutation have recently been shown to alter SV fusion in neurons expressing GFP tagged glycine-alanine dipeptides (Jensen et al., 2020).

| CHMP2B FTD is a novel synaptopathy with a novel mechanism
Defective SV protein degradation may lead to several knock-on effects at both the cellular and the circuit level in mutant CHMP2B mice. Of particular interest would be the downstream effect on neurotransmission and ultimately on neuronal circuits in areas of the brain affected. Behavioural deficits are not seen in this mouse model until 18 months of age, when neuron loss is well established (Clayton et al., 2017), indicating that homeostatic mechanisms may be able to compensate for initial synaptopathy. In agreement with this, brain array tomography in the Tg4510 mouse model of tauopathy has shown that tau induced loss of a subset of synapses may be initially compensated for by increase in other synapse subtypes (Kopeikina et al., 2013), which may explain why synaptic protein levels were not found to be altered at 6 months in mutant CHMP2B mice.
Our data showing that FTD causative mutant CHMP2B causes defective SV trafficking, and the convergence of several distinct neurodegenerative proteins on the SV pathway suggests that defects in SV trafficking are an important and early event in neurodegenerative pathogenesis. This warrants further mechanistic investigation in terms of potential therapeutic targets, in particular, the pathways responsible for sustained presynaptic performance in mutant CHMP2B provide promise as a route to increase synaptic resilience.

E TH I C S A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
All mouse work was performed in accordance with the ethical standards laid down in the Animals (Scientific Procedures) Act 1986, UK.

ACK N OWLED G EM ENTS
We thank Kerri Venner for technical support with electron microscopy.

DATA AVA I L A B I L I T Y S TAT E M E N T
Raw data is available from the corresponding author upon request.