Astrocytic Reactivity Triggered by defective Mitophagy Activates NF-kB Signaling and Causes Neurotoxicity in Frontotemporal Dementia Type 3


 Background: Frontotemporal dementia type 3 (FTD3) caused by a point mutation in the charged multivesicular body protein 2B (CHMP2B), affects mitochondrial ultrastructure and function as well as endosomal-lysosomal fusion in neurons. However, there is a critical knowledge gap in understanding how mutations in CHMP2B affect astrocytes. Hence, we investigated the disease mechanisms in astrocytes derived from hiPSC with mutations in CHMP2B and their impact on neurons.Methods: To dissect the astrocyte-specific impact of mutant CHMP2B expression, we generated astrocytes from human induced pluripotent stem cells (hiPSCs) from FTD3 patients and their CRISPR/Cas 9 gene edited isogenic controls and produced heterozygous and homozygous CHMP2B-mutant hiPSC via CRISPR/Cas 9 knock-in gene editing. Additionally, we confirmed our findings in CHMP2B mutant mice. The hiPSC were subjected to astrocyte differentiation and the mutation dependent effects were investigated using immunocytochemistry, western blot, cytokine assays, transmission electron microscopy, RNA-sequencing and gas chromatography-mass spectrometry. Finally, neurons were exposed to conditioned media of mutant astrocytes and viability, growth and motility were measured.Results: To dissect the astrocyte-specific impact of mutant CHMP2B expression, we generated astrocytes from human induced pluripotent stem cells (hiPSCs) and confirmed our findings in CHMP2B mutant mice. Our findings include perturbed mitochondrial dynamics with impaired glycolysis, increased reactive oxygen species and elongated mitochondrial morphology, indicating increased mitochondrial fusion in FTD3 astrocytes. Furthermore, we identified a shift in astrocyte homeostasis triggering a reactive astrocyte phenotype and increased release of toxic cytokines. This cumulates in NF-kB pathway activation with increased production of CHF, LCN2 and C3, which cause neurodegeneration. The neurotoxic effect was investigated by exposing hiPSC-derived neurons to astrocyte-conditioned media, which severely reduced neurite outgrowth capacities. Rescue experiments targeting ROS could restore ROS levels back to normal levels, indicating that the impaired removal of abnormal mitochondria triggers the pathological cascade in CHMP2B mutant astrocytes culminating in the formation of neurotoxic reactive astrocytes.Conclusion :Our data provide mechanistic insights into how defective mitophagy causes impaired mitochondrial fission, leading to the adoption of reactive astrocyte properties with increased cytokine release, NFkB activation and elevated expression of neurotoxic proteins in FTD3.


22
To dissect the astrocyte-specific impact of mutant CHMP2B expression, we generated     Co-culture with Astrocyte Condition Media 16 To assess the effects of neurite outgrowth, condition media obtained from FTD3 related 17 astrocytes and controls were cultured on wild type neurons for 5 days, then cells were 18 fixed using 4% PFA for immunocytochemistry. Wild type neurons were kept in neuronal 19 differentiation media as reference. Neurite length was assessed using neurite tracer 20 software of Image J. The analysis was carried out blinded by an independent 21 investigator.  For all experiments, data are presented as mean ± SEM (standard errors of the mean). 1 Statistical analysis was made in GraphPad Prism 7.03 and determined using Student's 2 t test; by one-way ANOVA with a Tukey's post-test or by or two-way ANOVA with 3 Bonferroni correction for differences of mean between each group, as indicated. Puncta 4 quantifications were determined using one-way ANOVA with a Tukey's post-test and 5 qPCR were determined using student t.test. Metabolic labeling were tested using two-6 way ANOVA with Tukey multiple comparisons test. Statistic significance was labelled in 7 figures as (*p < 0.05, **p<0.01 and ***p<0.001).

11
Astrocytes were differentiated from hiPSC lines derived from two related patients with 12 CHMP2B mutation (referred to throught the paper as FTD3 Patient 1 and FTD3 Patient 13 2), and their corresponding isogenic control, in which the mutation was corrected via 14 CRISPR/Cas9 gene editing (referred to throught the paper as Isogenic control 1 and 15 Isogenic Control 2, respectively). Additionally, two CRISPR/Cas9 knock-in cell lines 16 were generated carrying the same CHMP2B mutation (referred to throught the paper as 17 a homozygous and a heterozygous). All FTD3-hiPSCs and control-hiPSCs were 18 successfully differentiated into mature astrocytes with comparable efficiencies and 19 duration of differentiations following modified, previously established protocols [22,23] 20 ( Figure 1a). All astrocyte cultures expressed characteristic markers: transcription factor 21 SOX-9 (SOX9), protein S100-B (S100β) and Aquaporin-4 (AQP4) validated by average 85% of astrocytes were S100β, AQP4 and SOX9 positive in all investigated 2 hiPSC-derived astrocyte differentiations ( Figure. 1c-e, Table S1a). Furthermore,  Figure S1a). Analysis of differential gene expression between patient and control 7 astrocytes, revealed 1,133 genes with a log2 fold change (LFC) of gene expression ≥ 1; 8 and a significant P-value (<=0.05) after adjusting for multiple testing (Additional Figure   9 S1b).  Table S1f) and P62 ( Figure 2a, Table S1b, f). Our results, uncovered high expression  Table S1f) nor abundance of 1 accumulated endosomes and autophagosome (Additional Figure S5).  Figure S4). Consequently, we investigated if this impairment of autophagy mediated 9 clearance is caused by the inability to recruit lysosomes to the amphisomes. Co-  Table S1b).

23
To further substantiate the autophagy abnormalities in human FTD3 astrocytes, we  Next we investigated if the autophagic defects in our FTD3 astrocytes, leads to 7 accumulation of dysfunctional mitochondria and enhanced production of reactive 8 oxidative species (ROS). Assessment of mitochondrial ROS (mROS) revealed 9 significantly increased levels in homozygous mutant astrocytes compared to control 10 astrocytes, substantiating that mitochondria function and turnover is affected. (Figure   11 3a). Furthermore, our RNA-seq analyses revealed altered expression levels of genes 12 related to oxidative stress and damage in FTD3 mutant astrocytes vs. controls (   Table S1c). 16 Strikingly, RNA-seq analysis revealed differential expression of genes functionally 17 linked to mitochondrial fission/fusion processes (Table S2) Table S2).

17
Collectively, our results suggest that the impaired mitochondrial function and antioxidant 18 capacity of CHMP2B astrocytes contribute to both metabolic and oxidative stress.    (Table S1d). However, the difference in labeled alanine found in 2 FTD3 patient 2 did not reach statistical significance compared to the respective isogenic 3 control 2 (Figure 4e). Interestingly, the 13 C labeling (%) of most of the amino acids and 4 metabolites except for glutamate (FTD3 patient 1, Figure 4g), obtained from a first turn 5 of the TCA cycle in FTD3 astrocytes was significantly lower than their controls,  Table S1d). It can be speculated that the increase in labeled glutamate in FTD3 patient 8 1 compared to the isogenic control 1 may be derived from a small population of 9 metabolically active neurons within the culture that takes up the 13 C-labeled glutamine 10 derived from the astrocytes and synthesizes 13 C -glutamate via glutaminase. In line with 11 our observation of hampered mitochondrial respiration, differential expression analysis 12 based on RNA-seq revealed that numerous genes related to energy production 13 processes such as TCA and ETC are down-regulated in FTD3 patients compared to its  Table S1d). All metabolites evaluated were 4 decreased in FTD3 patients, heterozygous and homozygous astrocytes compared to 5 respective controls, which supports the observed energy hypometabolism. Reduced   Table S1e). In addition, we stained for glial fibrillary acidic protein (GFAP), extensively used as an astrocyte marker, and additionally used as an indicator of 1 astrocyte reactivity [74,75]. Based on qualitative ICC analysis we observed an increase 2 in GFAP expression in FTD3 patients, heterozygous and homozygous astrocytes 3 compared to their respective control, which further supporting that mutation in CHMP2B 4 trigger astrocyte reactivity (Figure 5a, right-column). Consistent with this, the differential 5 expression analysis based on RNA-seq revealed up-regulation of mitogen-activated 6 protein kinase 1(MAP3K1) in FTD3 patient astrocytes (LFC = 1.32 adj. P-value = 3.79E-7 13, Table S2). MAP3K1 is a serine/threonine kinase which activates conserved helix-  Table S1e). Even more intriguing we found a significant upregulation of complement C3 13 (C3), fibulin 5 (FBLN5) and serpin family G member 1 (SERPING1) in FTD3 patients 14 (Table S2)   Collectively our findings indicate that conversion of resting astrocytes to reactive astrocytes via mitochondrial deficiencies and NF-kB activation leads to increased pro-23 inflammatory cytokine release, which triggers an autoregulatory loop reinforcing the toxic reactive astrocyte phenotype and further enhances cytokine production that poses 1 a toxic threat to neurons.  It has previously been reported that excess accumulation of autophagosomes has a 24 negative impact on neuronal survival, and dysfunctional ESCRT-III appears to cause neurodegeneration through numerous mechanisms [93]. However, the exact 1 contribution of autophagy to neurodegeneration is largely unknown. In this study, we 2 report that astrocytes generated from FTD3 patient hiPSC or with introduced CHMP2B 3 mutation, display interrupted amphisomes-lysosomal fusion degradation. In relation to 4 this, we found an increased P62 expression in our CHMP2B mutant astrocytes, as 5 previously observed in our glial population, which is a key sign of increased/impaired [65]. We were able to show that UCDA reduced the mitochondrial ROS levels but did 20 not rescue mitochondrial phenotypes in CHMP2B mutant astrocytes. 21 Intriguingly, we observed throughout our study a more profound phenotype in the    Ideally, patient data would be added to further support our findings, but due to the 13 limitations in accessing those we can only indicate that overall gliosis is present in FTD3 14 patients. More complex co-culture systems such as brain organoids and co-cultures with 15 microglia will be implemented in the future to validate our findings.

Lead Contact
Further information and requests for resources and reagents should be directed to 1 Kristine Freude (kkf@sund.ku.dk).  The CRISPR/Cas9 introduced iPSC CHMP2B lines generated in this study will be made 10 available on request through Material Transfer Agreement. The patient derived hiPSC 11 cannot be shared due to specific regional restrictions implemented on patient material 12 by Region Hovestaden, Denmark.

15
Quantifications demonstrate a FTD3 dependent increase in P62 labeling, P62 mean 16 area of distribution, RAB7 markers and P62-RAB7 co-labelling, which cannot be 17 rescued with rapamycin treatment. k-n) LAMP1 and colocalization puncta number 18 quantification of FTD3 astrocytes and controls with 500 nM rapamycin treatment.