ALS-FTD-linked CCNFS621G drives increased hippocampal astrocyte ramification and mitochondrial dysfunction and impairs motor neuron excitability Authors: Liam Robinson, Dzung Do-Ha, Flora Cheng, Claire H. Stevens, Rossana Rosa Porto, Madilyn Coles, Jamesha Subachandran, Predrag Kalajdzic, Joanna Lui, Rachelle Balez, Sonia Sanz Muñoz, Mauricio Castro Cabral-da-Silva, Tracey Berg, Marco Morsch, Leszek Lisowski, Rachel H. Tan, Gaétan Burgio, Tim Karl, Albert Lee, Roger S. Chung, Ian Blair, Lezanne Ooi Published: 2026 ABSTRACT Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are neurodegenerative diseases with overlapping pathology. Mutations in CCNF, encoding the E3 ubiquitin ligase, Cyclin F, can cause ALS, FTD, or both, even within the same family. Most prior studies of CCNFS621G have relied on overexpression systems, potentially confounding outcomes through disruption of endogenous Cyclin F. Here, we generated the first knock-in mouse model of endogenous CcnfS621G using CRISPR/Cas9. Heterozygous and homozygous CcnfS621G mice showed no motor decline or neuronal loss after 18 months, however immunohistochemistry revealed increased hippocampal astrocyte ramification, with sex-, age, and subfield-dependent effects. These data indicate that endogenous CcnfS621G may prime early astrocyte alterations in the absence of overt neurodegeneration. Similar astrocyte morphological changes were observed in canonically affected regions of sporadic ALS and FTD-ALS patients post mortem, as well as in CCNFS621G iPSC-derived astrocytes following inflammatory stimulation. Proteomics on Ccnf mice identified early dysregulation of pathways related to translation, mitochondrial function, cytoskeletal remodelling, synaptic transmission and neuroinflammation. Correspondingly, CCNFS621G iPSC-derived astrocytes displayed impaired mitochondrial membrane potential and altered network morphology under both basal and inflammatory stimuli. As altered neuronal excitability is a hallmark of ALS, we examined astrocyte-driven changes to neuronal excitability. CCNFS621G iPSC-derived motor neurons cultured alone were hyperexcitable, firing more action potentials than isogenic controls. Remarkably, co-culture with CCNFS621G astrocytes, but not isogenic control astrocytes, abolished repetitive firing, increased the proportion of neurons unable to generate action potentials, and reduced voltage-gated sodium currents in CCNFS621G and isogenic control neurons. Together, these findings identify astrocyte alterations as an early feature of CCNFS621G-mediated disease, in the absence of neuronal loss. Moreover, the combination of astrocytic mitochondrial dysfunction and the ability of CCNFS621G astrocytes to suppress repetitive neuronal firing suggests a critical astrocyte-driven non-cell autonomous mechanism that may contribute to an oligogenic role for CCNF in ALS/FTD pathogenesis.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12974-026-03827-x. Background Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are patho-mechanistically related neurodegenerative diseases with the capacity to occur along a spectrum. Causative, heritable mutations in a set of risk genes have been identified that cause either ALS or FTD, or both (ALS-FTD). Despite a complex array of disease aetiologies associated with the ALS-FTD spectrum, deficits in cellular proteostasis remain a focal point of neurodegeneration [1, 2]. Proteostasis-altering mutations linked to the development of both ALS and FTD, such as CCNFS621G [3], therefore offer a platform to identify disease mechanisms and drug targets for the treatment of ALS and FTD. ALS is a neurodegenerative disease characterised by the progressive loss of upper and lower motor neurons in the motor cortex and spinal cord. Together, this results in progressive paralysis, and death within an average of 3–5 years, often due to respiratory failure [4]. The typical age of onset for this disease is between the ages of ~ 58–60, and currently carries a 1 in 300 overall lifetime risk, though due to ageing populations on a global scale, ALS incidence is projected to increase by 69% between 2015 and 2040 [5, 6]. Sporadic ALS (sALS) accounts for ~ 90% of all ALS cases and can develop through the combined influence of genetic vulnerabilities and environmental risk factors [7]. The remaining ~ 10% of cases are termed familial ALS (fALS), which have near identical clinical presentations to sALS, however can be attributed to the inheritance of an autosomal dominant genetic mutation in one of > 40 genes, including CCNF, SOD1, C9ORF72, TARDBP, VCP and FUS [3, 8]. Whilst primarily a neuromuscular disease, cognitive deficits are recorded in up to 65% of ALS patients [9], suggesting that pathology commonly extends beyond the motor system, highlighting the need for broader perspectives on disease mechanisms. FTD shares pathogenic mechanisms with ALS, although is characterised by neuronal death and dysfunction within the frontal and temporal lobes, as well as in the hippocampus [10]. As a result, people with FTD develop a variety of cognitive, behavioural and language deficits [11]. Familial FTD accounts for 20–50% of cases, with 60% of these bearing mutations in either MAPT, GRN or C9ORF72, and a smaller proportion of cases in genes, including CCNF, TARDBP and FUS [3, 12]. Concomitant development of ALS and FTD can significantly reduce life expectancy to ~ 2–2.5 years [13]. Overlapping pathophysiology, such as cytoplasmic aggregates of TDP-43, inflammation and mitochondrial dysfunction, suggest conserved disease mechanisms between ALS and FTD. Astrocytes play critical roles in ALS- and FTD-driving disease mechanisms, impacting disease onset, severity and progression [14]. Astrocytes regulate neuronal function through maintenance of the brain microenvironment, supporting neuronal health and neurotransmission [15]. Astrocytes also have critical functions related to neuroinflammation, with dual neuroprotective and neuroinflammatory responses, the intricate balance of which can be perturbed by ALS or FTD-linked mutations, leading to neurodegeneration. For example, mutations in the primary ALS-linked genes, SOD1, C9ORF72, TARDBP and FUS, all affect cellular proteostasis, and have all been demonstrated to increase astrocyte-mediated inflammation, with detrimental effects on the function of co-cultured motor neurons [16–19]. Recently, mutations in the CCNF gene were identified in ALS and FTD patients [3]. CCNF encodes the Cyclin F protein, an E3 ligase of the ubiquitin proteasome system (UPS) with critical roles in cellular proteostasis. Cyclin F is also responsible for regulating key proteins involved in cell cycle regulation, including the Early Region 2 Binding Factor (E2F) family of transcription factors, and DNA repair through Exonuclease 1 (EXO1) [20, 21]. Cyclin F is also involved in ubiquitination of neurodegeneration-relevant proteins, including TDP-43 [22], HSP90AB1 [23], VCP [24], SFPQ [25] and Sequestome-1 (SQSTM1/p62) [26]. The function of Cyclin F is regulated in part through phosphorylation of Ser621 – this phosphorylation site and the associated regulation are lost in the ALS-FTD-linked variant CCNFS621G [3, 27, 28]. CCNFS621G thus results in elevated and aberrant E3 ligase activity in CCNFS621G bearing cells, leading to the accumulation of ALS-linked Lys48 ubiquitinated proteins [27], deficits to autophagy [26, 27], systemic ubiquitome dyshomeostasis [29], dysfunctional ER-Golgi trafficking, Golgi fragmentation [30], increased apoptotic signalling, and motor neuron branching deficits [31, 32]. The vast majority of research on CCNFS621G has thus far centred on overexpression of exogenous Cyclin FS621G (and other mutations) to drive pathogenic phenotypes. In animal models, overexpression and thus dysregulation of wild-type Cyclin F has been adequate to promote motor, cognitive, behavioural or cellular pathology in both zebrafish and mice [31, 33], suggesting that regulation of native Cyclin F activity is crucial to proteostasis. To date, no animal models have assessed the impact of endogenous expression of native Cyclin F mutations on ALS-FTD disease phenotypes, and no published studies to date have considered the effects of the mutation in non-neuronal cells, such as astrocytes. Herein, we present a CRISPR/Cas9-generated mouse model of CcnfS621G, along with examination of identified pathological phenotypes in human sALS and FTD-ALS post mortem tissue and iPSC-derived CCNFS621G models. In the absence of motor phenotypes across Ccnf mutant mice, we observed behavioural phenotypes associated with impaired fear-memory formation and reduced exploratory drive. Proteomic analyses of heterozygous and homozygous CcnfS621G mice identified significant alterations in pathways associated with translational reprogramming, mitochondrial dysfunction, neuroinflammation and cytoarchitectural remodelling, providing molecular pathways to target in downstream mechanistic experiments. Consistent with this, we identified a distinct increase in the ramification of hippocampal astrocytes in CcnfS621G mice, a phenotype which displayed both time, subregion and sex dependency. We confirmed the broader presence of this morphological phenotype in canonically affected regions from both pure sALS and FTD-ALS human post mortem tissue representing late-stage disease. Consistent with the Ccnf mouse proteomics, in iPSC-derived CCNFS621G astrocytes, we identified significant impairments to mitochondrial network morphology and membrane depolarisation under both basal and inflammatory conditions. Furthermore, we demonstrated that CCNFS621G astrocytes significantly reduced action potential firing of CCNFS621G, or isogenic control motor neurons in co-culture. Together, this research highlights an early astrocyte phenotype driven by CCNFS621G that occurs in the absence of neuronal loss. Materials and methods CcnfS621G mouse model generation The generation of CcnfS621G mice using CRISPR/Cas9 was approved from the Institutional animal Ethics committees from the Australian National University in agreement with the code of practice from the National Health and Medical Research Council (NHMRC) of Australia (AEEC 2014/58). CcnfS621G/wt founders in a C57BL6/NCrl background strain (Charles Rivers) were generated using CRISPR/Cas9 gene editing technology according to our previously published protocols [34] using the following single guide RNA (sgRNA) guide 1: 5’- CCTTCTCGCCTTCGCTCTCC TGG – 3 and a single stranded oligonucleotide as a donor template: ssOdn1 5’-CACCCGCATGCCACATCCCTTCCCCAGCACGAGTGGCACTCACCATCACCCTCCTTCTCGCCTTCGCCCTCCTGGTCTCCCTCATAGCCAGAGCAGCACTCCAGGCTCCAGTCCAGAAGGCTGCCTAGCAGTGTCT-3’. The procedure for pronuclear injection and mouse genotyping was previously described [35]. The founder retained for downstream analysis was confirmed by Sanger sequencing. The mice were backcrossed over ten times to a C57BL6/NCrl background to remove any potential off target effects. Whole mouse brains were perfused with paraformaldehyde (PFA) for 24 h before being stored at 4 °C in PBS until further use. Human post mortem brain tissue Cases with a pathological diagnosis of sALS (n = 8), fALS (CCNFS621G) (n = 1), sporadic FTD-ALS (n = 6) and healthy controls (n = 5) were acquired from the New South Wales Brain Banks. This study used 10 µm thick slide mounted formalin fixed paraffin embedded post mortem brain tissue sections from the motor cortex (relevant to ALS) and the superior frontal cortex (relevant to FTD-ALS). This research project was approved by the University of Wollongong Human Research Ethics Committee (approval number 17/035). Cohort information is outlined in Supplementary Table 3. Chromogenic immunohistochemistry Whole mouse brains were cut along the sagittal plane to separate hemispheres. One hemisphere from the 10 and 18-month cohorts were homogenised for proteomics. Remaining mouse brain hemispheres were processed in a Lecia ASP200S tissue processor, incubated in 70% ethanol (1 × 1 h), 90% ethanol (2 × 1 h), 100% ethanol (2 × 1 h, followed by 1 × 1 h 20 min), xylene (3 × 1 h 20 min) and paraffin wax (3 × 1 h). Processed tissues were immediately embedded in paraffin wax blocks using a Leica EG1150 H and Leica EG1150 C embedding station. 5 µm sagittal sections were taken from the mouse brains using a Leica RM2255 microtome and collected on glass slides. To prepare for immunohistochemical staining of both mouse and human tissue, sections were deparaffinised in xylene (2 washes, 5 min) and rehydrated in an ethanol gradient (2 washes, 3 min – 100%, 96%, 70%) before a final wash in MilliQ water (3 min). Antigen retrieval was performed by boiling tissue sections in 0.1 M sodium citrate for 9 min. Sections were left to cool in the citrate buffer for 45 min. Following this, sections were washed in MilliQ water (1 × 3 min), before endogenous peroxidase and alkaline phosphatase activity was inhibited by a 10 min incubation in BLOXALL® (Vector Laboratories SP-6000-100). Slides were then rinsed twice in MilliQ followed by Tris-buffered saline (TBS) for 3 min, before non-specific binding was blocked in a 40-min incubation with 5% normal horse serum (Vector Laboratories, S-2000-20) or 5% goat serum (Gibco 16210–064) diluted in TBS. After blocking, sections were incubated overnight at 4 °C in a humidified chamber with primary antibodies diluted in blocking solution. Primary antibodies against GFAP (Abcam Ab53554, 1:200 for mouse tissue, DAKO M0761, 1:200 for human tissue), NeuN (Abcam Ab177487, 1:500) pan TDP-43 (ProteinTech 10782-2-AP, 1:800), IBA1 (Abcam ab5076, 1:100), lysine 48 ubiquitin (K48-ubiquitin) (Abcam Ab140601, 1:150), p62 (Sigma-Aldrich p0067, 1:200 for mouse tissue, BD Transduction Labs 610833, 1:500 for human tissue) and phosphorylated TDP-43 (Cosmo Bio CAC-TIP-PTD-M01,1:1000) were used in this study. Negative control sections were incubated without primary antibodies overnight in the blocking buffer. The next day unbound primary antibodies were removed in TBS (3 washes, 5 min). Sections were then incubated at 37 °C with biotinylated secondary antibodies (Vector Laboratories, BA-9500 or BA-1000, 1:200) for 30 min. Following incubation, unbound secondary antibodies were removed in TBS (3 washes, 5 min). Alkaline phosphatase was added using the ABC-AP Vector Red kit (Vector Laboratories, AK5000); Reagents A and B were diluted 1:500 in blocking buffer and allowed to pre-incubate for 30 min prior to addition to sections. After 1 h, excess ABC-AP kit solution was removed from sections in TBS (3 washes, 5 min). Chromogenic staining was visualised using the Immpact Vector Red substrate kit (Vector Laboratories SK-5105); the staining solution was assembled immediately prior to addition, with 16 µL Reagent 1 and 12 µL Reagent 2 for every 1000 µL Diluent. Incubation times on sections were optimised for each antibody. Staining was halted and excess Immpact Vector Red removed by washing in MilliQ (5 min). Sections were counterstained with Haematoxylin and dehydrated through the reverse ethanol and xylene gradient above, prior to mounting with Dibutylphthalate Polystyrene Xylene (DPX) (Sigma-Aldrich 06522) and coverslipping. Sections were imaged on a Leica DM750 light microscope with a K3C colour camera attached. Proteomic analyses Approximately 5–10 mg of weighted frozen mouse brain tissue was resuspended in 50 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 25 mM NEM, 1% NP40 containing complete protease inhibitor cocktail (4693159001, Sigma-Aldrich) and phosSTOP (4906837001, Sigma-Aldrich), and homogenised for 30 s using an Omni TH220 tissue homogenizer. The tissue homogenate was probe-sonicated (10 s, setting 3, Sonic Ruptor 250, Omni International) and centrifuged at 14,000 xg for 30 min at 4 °C. The clarified supernatants were transferred to new tubes and protein quantification was carried out using the Pierce BCA kit (23225, Thermo Fisher Scientific). Approximately 20 μg protein were denatured by boiling at 95 °C for 5 min in Laemmli (1610747, Bio-Rad) loading buffer containing 50 mM dithiothreitol (DTT, NP0009, Thermo Fisher Scientific). Denatured proteins were separated by gel electrophoresis using 4–15% Tris–glycine gels (4565083, Bio-Rad), according to the manufacturer's instructions for approximately 3–5 min (1–3 cm into the gel) and stained with Coomassie Brilliant Blue. Stained proteins were excised into 3–5 fractions for each sample and prepared for in-gel trypsin digestion, as described previously [36]. In-gel trypsin digestion Each gel fraction was destained with 50 mM ammonium bicarbonate, 50% methanol, followed by 50 mM ammonium bicarbonate, 50% acetonitrile (ACN) before being dehydrated with 100% ACN. The gel pieces were dried under vacuum centrifugation. Proteins were reduced with 10 mM DTT for 30 min at 55 °C, alkylated with 55 mM IAA for 60 min at room temperature in the dark. Trypsin digestion (1:50 enzyme:protein) was added to the gel pieces and incubated overnight at 37 °C (V5111, Promega). Extracted peptides were lyophilised and then resuspended in 0.1% formic acid (FA) for desalting using C18 OMIX tips (A57003100K, Agilent). Samples were lyophilised again and resuspended in 0.1% FA, bath-sonicated for 20 min, and then centrifuged at 14,000 xg for 15 min to remove insoluble debris, and analysed by LC–MS/MS. Reverse phase C18 liquid chromatography mass spectrometry (RP-LC–MS/MS) and data processing The Ultimate 3000 nanoLC (Thermo Fisher Scientific) fitted with the Acclaim PepMap RSLC column particle size of 2 μm, diameter of 0.075 mm, and length of 150 mm (Thermo Fisher Scientific) was used, employing a 60 min gradient (2–80% v/v ACN, 0.1% v/v FA) running at a flow rate of 300 nL/min to separate peptides. Subsequently, eluted peptides were ionised into the Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) that had an electrospray source fitted with an emitter tip 10 μm in diameter (New Objective) and maintained at 1.6 kV electrospray voltage. The capillary temperature was set to 250 °C. A data-dependent “Top 10” method operating in FT acquisition mode with HCD fragmentation was used for MS/MS fragmentation to select precursor ions. On the Q Exactive Plus, FT-MS analysis was carried out at 70,000-times resolution with an AGC target of 1 × 106 ions in full MS and a maximum injection time of 30 ms. Additionally, MS/MS scans were carried out at 17,500-times resolution with an AGC target of 2 × 104 ions with the maximum injection time set to 50 ms. The ion selection threshold was set to 25,000 counts to trigger MS/MS fragmentation. HCD fragmentation was performed using an isolation width of 2.0 Da with a normalized collision energy of 27. Raw files were searched with Proteome Discoverer 2.4 (Thermo Fisher Scientific) against Uniprot Mus musculus FASTA database (Release 2018_03, 28th March 2018, 16,929 sequences and Release 2024_07, 24th July 2024, 17196 sequences) incorporating the Sequest search algorithm. Search parameters accounted for 20 ppm precursor ion tolerance and 0.1 Da MS/MS fragment ion tolerance. The search allowed for static modifications of cysteine carbamidomethylation and variable modifications of methionine oxidation, asparagine, and glutamine deamidation on acetylated N-terminal residues. Two missed cleavages were allowed for. The data were processed through Percolator for estimation of false discovery rates. Protein identifications were validated employing a q-value of 0.05. Protein identification required at least one unique peptide per protein. Statistically significant differences in protein, peptide sequence, and peptide groups identified were calculated using Proteome Discoverer 2.4. High-confidence proteins with abundance ratios having p < 0.05 were selected and included for pathway analysis. An absolute expression fold-change cutoff of 2 was used to identify analysis-ready molecules in Ingenuity Pathway Analysis (IPA) to provide insights into specific pathways affected based on the proteomic analysis. Data were analysed with the use of QIAGEN IPA. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository [37] with the dataset identifier PXD064648 and 10.6019/PXD064648. Interaction networks of proteins with expression profiles significantly different (p < 0.05, fold-change cutoff of 2) to those of wild-type mice were created using STRING. Gene ontology and reactome pathway enrichment were performed using in-built STRING analysis tools. Interaction networks were exported from STRING to Cytoscape (version 3.10.2), where MCL clustering was performed to identify functionally related local protein clusters; an inflation parameter of 4 was used for all analyses. Functional enrichment and annotation were manually performed on these protein clusters using in-built Cytoscape tools. Proteins were colour-coded depending upon their regulation status using abundance ratio values comparing expression to wild-type mice; proteins with abundance ratio values < 1 were coloured blue and were downregulated, proteins with abundance ratio values > 1 were coloured red and were upregulated. iPSC-derived astrocyte generation All experiments involving the use of human induced pluripotent stem cells (iPSCs) were approved by the University of Wollongong Human Research Ethics Committee (approval number 13/272). The fALS iPSCs used in this study were previously generated from dermal fibroblasts donated by a 59-year-old male bearing a monoallelic ALS-FTD-linked CCNFS621G mutation [38]. An isogenic control iPSC line was generated from this parental line using CRISPR-Cas9 gene editing to correct the S621G mutation to wild-type; sequencing and karyotyping confirmed successful correction of the mutation, with characterisation confirming retained pluripotency (Supplementary Fig. 13). Given that the ALS/FTD associated S621G mutation lies in the PEST domain of CCNF and has been shown to alter Cyclin F regulation, an additional ‘bi-allelic’ iPSC line was generated from the patient line through the deletion of the PEST region on the wild-type allele (CCNFS621G/ΔPEST) (Supplementary Fig. 14). This line was included to assess Cyclin F regulation in iPSC-derived models. We hypothesised that this biallelic CCNF mutation would further exacerbate any phenotypes driven by alterations in Cyclin F function. The iPSC-derived astrocytes used in morphological and mitochondrial analyses were generated as previously described [39], as adapted from [40]. Immunocytochemistry for astrocyte morphology analysis Day 14 iPSC-derived astrocytes were plated out at 10,000 cells/well in black optical bottom 96 well plates (Greiner Bio-One, 655090). The following day, astrocytes were co-stimulated with 50 ng/ml TNF-α (ThermoFisher Scientific, PHC3015L) and 10 ng/ml IL-1β (ThermoFisher Scientific, PHC0815). After 48 h, cells were fixed in 4% PFA for 10 min, prior to permeabilisation with 0.1% TritonX-100 (Sigma-Aldrich T9284-100ML) in PBS for 10 min. Nonspecific binding was blocked with 10% goat serum (Gibco, 16210–064) diluted in PBS for 1 h at room temperature. Cells were then incubated overnight at 4 °C with primary antibodies against GFAP (Abcam ab68428, 1:250) diluted in blocking buffer. The following day primary antibodies were removed in three washes of 0.01% TritonX-100 in PBS, before cells were incubated with Alexa Fluor 488 secondary antibodies at 1:1000 (Life Technologies, A11008) for 1 h at room temperature. Secondary antibodies were removed in three washes of 0.01% TritonX-100 in PBS, and cell nuclei were stained with 1:1000 Hoechst33342 (Life Technologies, H3570) in PBS for 15 min. Excess Hoechst was removed in two PBS washes and cells were stored at 4 °C in PBS until imaging. Images were collected at 20X magnification on an SP8 LIGHTNING confocal microscope (Leica Microsystems, Germany). The proportion of cells positive for the astrocyte markers, SOX-9 and GFAP, was determined through manual counting, using the ‘Cell Counter’ ImageJ plugin. Tetramethylrhodamine Ethyl Ester (TMRE) analysis of mitochondrial membrane potential and network morphology Day 14 astrocytes were plated out at 40,000 cells per chamber of an 8-chamber glass bottom coverslip (IBIDI, 80827). The following day, astrocytes were co-stimulated with 50 ng/ml TNF-α and 10 ng/ml IL-1β. After 48 h, astrocytes were stained with 50 nM TMRE (Abcam, ab113852), 250 nM CellTracker Green (ThermoFisher Scientific, C2925), and 1:5000 Hoechst33342 in Hanks Buffered Saline Solution (HBSS) for 30 min. The staining solution was washed off using HBSS, before astrocytes were imaged on an SP8 LIGHTNING confocal microscope at 40X magnification with an additional digital zoom of 2.25, and a 2048 × 2048 pixel resolution. All cells were maintained at 37 °C and 5% CO2 in HBSS during image acquisition. Pre-incubation with 20 µM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) for 20 min was used as a negative control to collapse membrane potential and normalise fluorescence intensity readings within each cell line and replicate. Mouse and human tissue image analysis 20X magnification images were analysed using custom CellProfiler pipelines (Version 4.2.5). Astrocyte morphology was measured from GFAP staining; briefly, colour deconvolution was performed to broadlyisolate the Vector Red staining from the haematoxylin counterstain, further background was removed by processing modules. Whole astrocytes were isolated using advanced settings within the‘IdentifyPrimaryObjects’ module and skeletonised (GFAP-stained samples). The resulting object skeletons were analysed and measured using the ‘MeasureObjectSizeShape’ module, overlays of thesemeasurements on the original image were saved and manually examined to ensure successful and representative analyses. Hippocampal granule cell layer thickness was measured manually in ImageJ by drawingand measuring lines across the granule cell layers. Three measurements were taken and averaged from each mouse in both the CA1 and dentate gyrus regions. iPSC-derived astrocyte image analysis Fluorescent channels were separated from 20X confocal images of GFAP-stained astrocytes using ImageJ. These images were processed and analysed in CellProfiler, in a similar manner to mouse and human tissue staining. Firstly, the GFAP channel image intensity was rescaled, smoothed and subtracted from the original GFAP channel image to remove background. Whole astrocytes and nuclei were identified using IdentifyPrimaryObjects. Following this, nuclei and astrocyte primary objects were related to refine the identification of astrocyte cell bodies as secondary objects. Similarly to the tissue analysis, these objects were measured, and overlays saved for manual quality control checks to ensure accuracy in image processing and analysis. For TMRE analysis, cell identification was performed as previously described using CellTracker Green and Hoechst33342. Networked mitochondria were segregated from isolated (non-networked) mitochondria by a minimum object diameter of 100 pixel units to enable differential analyses. Mitochondrial network skeletonisation was performed using the ‘MorphologicalSkeleton’ module. All mitochondrial objects (children) were related to their corresponding (parent) cell objects to generate average mitochondrial measurements per cell. Motor neuron differentiation The iPSCs were differentiated into electrophysiologically mature motor neurons using a previously established protocol [41]. Briefly, anterior neural precursors (aNPCs) were transitioned into motor neuron precursor medium (BrainPhys with 1 × N2, 1 × B27, 1 × non-essential amino acids) and supplemented with 0.1 µM retinoic acid, 0.5 µM purmorphamine, 2 µM SB431542, 1 µM CHIR99021 and 2 µM DMH-1). After 5 days, motor neuron precursors were dissociated with 1 × Accutase and replated at a density of 25,000 cells/cm2 into the final maturation motor neuron maturation medium (motor neuron medium; BrainPhys with 1 × N2, 1 × B27, 1 × non-essential amino acids). For the first 10 days motor neuron medium was supplemented with 0.5 µM retinoic acid, 5 µM purmorphamine, 10 µM DAPT, BDNF (10 ng/mL), GDNF (10 ng/mL), IGF (10 ng/mL), db-cAMP (500 ng/mL), L-ascorbic acid (200 ng/mL). After 10 days, retinoic acid, purmorphamine and DAPT were removed from the medium, and the cells were maintained with the motor neuron medium and the remaining supplements. Astrocyte differentiation for co-cultures Neurospheres were fully dissociated into single cells by pre-incubating spheres in PBS/EDTA for 3 min followed by 1 × Accutase. During the dissociation process cells were kept at 37ºC and every 3–5 min cell clumps were titrated to aid dissociation. Depending on neurosphere size, the dissociation process varied between 15 to 30 min. Once cells were fully dissociated, the cell suspension was diluted at least 6-fold with DMEM/F12 to inactivate the dissociation enzyme. The cell suspension was subsequently passed through a 40 μm sieve (Miltenyi Biotec, Germany) to filter out remaining cell clumps. The single cell suspension was spun at 600 g for 5 min and the cell pellet was resuspended in Astrocyte Maturation (AM) medium (DMEM/F12 with 2% (v/v) B27, 1% (v/v) N2, 1% non-essential amino acids and 1% (v/v) GlutaMax). Single cells were plated out on Matrigel coated tissue culture plates or ⌀13 mm coverslips at a density of 14,000 cells/cm2. The medium was supplemented with ciliary neurotrophic factor (CNTF; Miltenyi Biotec, Germany) and neuregulin-1 (NRG-1; Peprotech, USA), both at 10 ng/mL. Partial media changes were performed every 2nd day. During the first week of maturation, medium was slowly transitioned from DMEM/F12 to BrainPhys-based AM medium with identical supplements. Astrocytes were matured for at least 4 weeks until cells were processed for subsequent analysis. Motor neuron and astrocyte co-culture For the generation of a fully humanised co-culture system using patient derived cells, iPSC-derived motor neurons and astrocytes were firstly differentiated separately from ALS donors and matched controls. After 2 weeks in the final motor neuron maturation stage, dissociated astrocytes were added to the motor neuron cultures. Whole-cell patch clamping and electrophysiology Whole-cell patch clamping was performed on iPSC-derived motor neurons with and without astrocytes after 4–5 weeks of maturation. All patch clamping experiments were performed at 22–24 °C after 15 min equilibration in artificial cerebral spinal fluid (aCSF: 135 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 10 mM HEPES and 10 mM D-glucose, adjusted to pH 7.4 (with NaOH) and ~ 305 mOsm). The patching electrodes were filled with intracellular recording solution (150 mM KCl, 2 mM MgCl2, 10 mM HEPES, 4 mM Mg-ATP, 0.3 mM Na-GTP, 10 mM Na2PCr and 1 mM EGTA; adjusted to pH 7.4 (with KOH) and ~ 295 mOsm). Signal recordings were obtained at 10 Hz using Digidata 1550 and amplified using a MultiClamp 700B amplifier (Axon Instruments, Molecular Devices Electrophysiology). Cells were visualized using an inverted Leica DM IRB microscope and chosen based on neuronal morphology, which included large cell bodies and long neurite extension. Once cells were identified, whole-cell patch clamp configuration was achieved after establishing a GΩ seal (> 1 GΩ). Pipette capacitance was compensated on-line during cell-attached configuration prior to breaking into the cell. A test pulse of −10 mV was used to measure basic membrane properties including cell capacitance, membrane resistance and series resistance. Series resistance was monitored throughout the experiment to ensure the integrity of the patch. Cells with a series resistance (Rs) above 20 MΩ were excluded from subsequent analysis. For both voltage and current clamp recordings cells were held at a holding potential of −60 mV. Currents were recorded by injecting 20 ms voltage steps at 10 mV intervals from −90 to 100 mV. On-line P/N leak subtraction protocol was applied during recordings of voltage-sensitive currents to account for passive leak currents. Action potentials were evoked by increasing current steps until cells fired action potentials or 600 pA. Cells were categorized as firing repetitively if they produced ≥ 3 APs without significant adaptation. All action potential properties were measured from the first evoked action potential. All measurements and analysis of recorded traces was performed using Clampfit 10.7. Results CcnfS621G mice display time and sex-dependent increases in hippocampal astrocyte ramification in the absence of neuronal loss As both ALS and FTD phenotypes are associated with aging, we assessed heterozygous and homozygous mouse cohorts at 10, 15 and 18 months of age, compared to their littermate wild-type controls. No significant motor deficits were observed in CRISPR/Cas9-generated CcnfS621G heterozygous or homozygous mice by 10 or even 18 months (Supplementary Fig. 1). Consistent with the lack of motor phenotype, neither significant motor neuron loss nor TDP-43 pathology were detected in the motor cortices of the mouse cohort, even by 18 months of age (Supplementary Fig. 2A-D). Additionally, lysine 48 (K-48) ubiquitin staining revealed the presence of inclusions across the motor cortex, frontal cortex and hippocampus in all 18-month mice (Supplementary Fig. 3). A significant increase in inclusion density was observed between wildtype mice and homozygous mice in the motor cortex (p = 0.043). No other significant differences in K48 inclusion density were observed (Supplementary Fig. 3). Similarly, analysis of p62 foci at 18 months revealed no significant differences, compared to wildtype mice, although foci density was significantly increased in homozygous mice, compared to heterozygous mice in the motor cortex only (Supplementary Fig. 4). Behavioural profiling of the Ccnf mice at 10-months revealed early alterations to fear-memory formation and exploratory drive (Supplementary Fig. 5–9). Such phenotypes have previously been associated with degeneration or dysfunction within the hippocampus [42]. To determine whether these behavioural alterations were accompanied by hippocampal degeneration we measured the thickness of the granule cell layer of the CA1 and dentate gyrus regions, but no significant differences were identified in 18-month mice (Supplementary Fig. 2B-C). Despite the absence of advanced neurodegenerative phenotypes within the lifespan of CRISPR-Cas9 CcnfS621G mice, examination of human post mortem CCNFS621G ALS donor tissue (n = 1) revealed characteristic p62 and phosphorylated TDP-43 (pTDP-43) pathology in both neuronal and glial cells of the motor cortex and spinal cord (Supplementary Fig. 10), similar to that observed in sporadic ALS cases [43]. These findings confirm the presence of hallmark TDP-43 pathology in the end-stage human disease context of CCNFS621G carriers and highlight the involvement of non-neuronal cell types in disease pathogenesis of CCNFS621G ALS, which has previously not been characterised in depth. Given the role of astrocytes and astrocyte activation in fear memory encoding and retrieval [44, 45], we assessed astrocyte morphology as an indicator of astrocyte activation in the hippocampus. 20X magnification images of GFAP-stained dentate gyrus and CA1 images were captured from 10, 15 and 18-month Ccnf mice. Representative images can be viewed in Supplementary Fig. 11. Custom CellProfiler image analysis pipelines were used to perform colour deconvolution and single-cell morphological characterisation (Fig. 1A). For readability, only sexes and timepoints with statistical significance are shown (Fig. 1B). No statistically significant alterations in astrocyte skeleton morphology were detected at 10-months in either male or female mice. However, at 15-months, single-cell analyses revealed significant increases in astrocyte arborisation in the dentate gyrus of homozygous male mice (Fig. 1B). Meanwhile, in 15-month female mice, these differences were instead observed in the CA1 subfield specifically in heterozygous mice (Fig. 1B). At 18-months, male mice continued to present significant alterations to dentate gyrus astrocyte morphology, with significant increases in branching and skeleton length observed in both homozygotes and heterozygotes (Fig. 1C). Notably, homozygous males from this 18-month cohort also displayed significant increases in astrocyte ramification in the CA1 subfield; a region previously only affected in female mice (Fig. 1C). Correspondingly, hippocampal IBA1-positive microglial density was elevated in the CA1 region, whilst the motor cortex displayed no significant shifts in microglial density (Supplementary Fig. 12). Analysis of female 18-month mice was not possible due to breeding issues. Overall, these data demonstrate a significant shift in the morphology of hippocampal CcnfS621G astrocytes towards a more ramified phenotype that is dependent upon age, sex, region and allele dosage.Fig. 1CcnfS621G mice exhibit time, sex and region-specific alterations in hippocampal astrocyte ramification. A Representative hippocampal GFAP immunostaining pre- and post-processing, with an annotated astrocyte skeleton. B Analysis of the dentate gyrus from 15- and 18-month Ccnf mice, with total skeleton length measured in pixels. Data points represent mean measurements per mouse (4–5 mice per genotype and sex), single-cell data from each mouse are plotted as overlaid violins for visualisation. Mouse sex is depicted alongside corresponding graphs. For readability, only sexes and timepoints with statistical significance are shown. C Analysis of the CA1 region from 15- and 18-month Ccnf mice. Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2) or gamma distributions and mouse origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method. All images were captured at 20X magnification on a Leica DM750 microscope with attached K3C colour camera, and identical imaging settings. Scale bar represents 100 µm and is equivalent across B-C. Figure created with the aid of Biorender CcnfS621G mice exhibit time, sex and region-specific alterations in hippocampal astrocyte ramification. A Representative hippocampal GFAP immunostaining pre- and post-processing, with an annotated astrocyte skeleton. B Analysis of the dentate gyrus from 15- and 18-month Ccnf mice, with total skeleton length measured in pixels. Data points represent mean measurements per mouse (4–5 mice per genotype and sex), single-cell data from each mouse are plotted as overlaid violins for visualisation. Mouse sex is depicted alongside corresponding graphs. For readability, only sexes and timepoints with statistical significance are shown. C Analysis of the CA1 region from 15- and 18-month Ccnf mice. Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2) or gamma distributions and mouse origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method. All images were captured at 20X magnification on a Leica DM750 microscope with attached K3C colour camera, and identical imaging settings. Scale bar represents 100 µm and is equivalent across B-C. Figure created with the aid of Biorender Proteomic analyses identified alterations in pathways related to mitochondrial function, translation, neuroinflammatory signaling and synaptic transmission in CcnfS621G Mice To explore the molecular pathways impacted by CcnfS621G at the early (10-month) and late (18-month) stages, we analysed the brain proteomes of heterozygous and homozygous CcnfS621G mice, in comparison to their wild-type littermate controls. For the mouse studies, the cohorts aged to 18 months resulted in unavoidable loss of mice, which prevented sex-matched cohorts. Consequently, the proteomic analyses were performed primarily on grouped homozygous, heterozygous, and wild-type cohorts using the samples available at the same end-stage time point. In 10-month heterozygous CcnfS621G mice, 272 proteins were downregulated while 329 were upregulated (> 2-fold expression, p-value < 0.05), compared to homozygous 10-month mice with 204 downregulated and 415 upregulated proteins (>2-fold expression, p-value < 0.05). In 18-month heterozygous mice 76 proteins were downregulated and 79 proteins were upregulated (> 2-fold expression, p-value < 0.05); meanwhile in 18-month homozygous mice 210 proteins were downregulated and 86 were upregulated (> 2-fold expression, p-value < 0.05). Gene ontology and reactome pathway enrichment analyses identified significant groupings of differentially expressed proteins in processes associated with translation and mitochondrial function, across all tested genotypes and timepoints (Figs. 2 and 3). Markov cluster algorithm (MCL) clustering and functional enrichment revealed local interacting protein clusters that reflected these gene ontology results (Figs. 2 and 3).Fig. 2Proteomic analyses of 10-month CcnfS621G mouse brains revealed widespread changes to translation and mitochondrial pathways. Protein interaction networks were generated in STRING from proteins with significantly different expression from corresponding wildtype mice. Reactome pathway enrichment analyses were performed with in-built STRING tools. Networks were exported to Cytoscape v3.10.2 where MCL clustering (inflation parameter 4) and functional enrichment on local protein clusters were performed. For the 10-month cohort: 2 × female, 1 × male wild-type; 3 × female heterozygotes; 2 × female, 1 × male homozygote were used for analysis. A Reactome pathway enrichment gene ontology analyses of up- and downregulated proteins, generated in STRING. B MCL clusters of interest with functional enrichment annotation, generated in CytoscapeFig. 3Proteomic analyses of 18-month CcnfS621G mouse brains revealed continued perturbations to translation and mitochondria. Protein interaction networks were generated in STRING from proteins with significantly different expression from corresponding wildtype mice. Gene ontology and reactome pathway enrichment analyses were performed with in-built STRING tools. Networks were exported to Cytoscape v3.10.2 where MCL clustering (inflation parameter 3) and functional enrichment on local protein clusters were performed. For the 18-month cohort: 5 × female, 3 × male wild-type; 5 × female, 5 × male heterozygotes; 4 × male homozygotes were used for analysis. A Gene ontology and reactome pathway enrichment analysis from 18-month mice. Due to lower numbers of differentially expressed proteins, pooled results of up- and downregulated proteins are displayed. Reactome pathways enrichment was not available for 18-month heterozygote data. B MCL clusters of interest with functional enrichment annotations, generated in Cytoscape Proteomic analyses of 10-month CcnfS621G mouse brains revealed widespread changes to translation and mitochondrial pathways. Protein interaction networks were generated in STRING from proteins with significantly different expression from corresponding wildtype mice. Reactome pathway enrichment analyses were performed with in-built STRING tools. Networks were exported to Cytoscape v3.10.2 where MCL clustering (inflation parameter 4) and functional enrichment on local protein clusters were performed. For the 10-month cohort: 2 × female, 1 × male wild-type; 3 × female heterozygotes; 2 × female, 1 × male homozygote were used for analysis. A Reactome pathway enrichment gene ontology analyses of up- and downregulated proteins, generated in STRING. B MCL clusters of interest with functional enrichment annotation, generated in Cytoscape Proteomic analyses of 18-month CcnfS621G mouse brains revealed continued perturbations to translation and mitochondria. Protein interaction networks were generated in STRING from proteins with significantly different expression from corresponding wildtype mice. Gene ontology and reactome pathway enrichment analyses were performed with in-built STRING tools. Networks were exported to Cytoscape v3.10.2 where MCL clustering (inflation parameter 3) and functional enrichment on local protein clusters were performed. For the 18-month cohort: 5 × female, 3 × male wild-type; 5 × female, 5 × male heterozygotes; 4 × male homozygotes were used for analysis. A Gene ontology and reactome pathway enrichment analysis from 18-month mice. Due to lower numbers of differentially expressed proteins, pooled results of up- and downregulated proteins are displayed. Reactome pathways enrichment was not available for 18-month heterozygote data. B MCL clusters of interest with functional enrichment annotations, generated in Cytoscape Altered mitochondria and metabolic proteins emerged as an early phenotype within 10-month CcnfS621G mice (Fig. 2) and persisted in the aged 18-month cohort (Fig. 3). Notably, Rtn4ip1, Fundc1 and Hddc2 were dysregulated in both 10-month and 18-month homozygous. Gdpgp1, Slc25a19 and Rdh13 were significantly downregulated in both 10-month and 18-month heterozygous mice. Strikingly, 10-month mice and 18-month homozygous mice presented divergent expression of numerous electron transport chain complex subunits (Figs. 2B and 3B). In 18-month mice, alterations in the mitochondrial-related proteome also extended to proteins involved in mitophagy, as well as mitochondrial fission, fusion and trafficking dynamics, including Rhot2, Mfn1 and Fundc1 (Fig. 3B). Reflecting a state of altered global and mitochondrial translation in CcnfS621G mice, MCL clusters also revealed intricate patterns of regulation in the expression of numerous small (Rps) and large (Rpl) ribosomal subunits, including nuclear-encoded mitochondrial ribosome subunits (Mrps, Mrpl) (Figs. 2 and 3). Ingenuity Pathway analysis (IPA) revealed other common pathways dysregulated between 10- and 18-months CcnfS621G mice, including signalling by Rho Family GTPases, contributed by the upregulation of proteins, such as Gng10, Pikfyve, Gng5, Slc9a1, Itgb2 and Rhot2 and the downregulation of Mylk, Ppp1r12a, Rhobtb2 in the 10-month cohort. Downregulation of Rhot2, Gnb4, Pik3r1, Gnb3, Arggef17, Septin1 were found to contribute to the same pathway in the 18-month homozygous CcnfS621G mice. MCL clusters also reflected these alterations in Rho GTPase signalling in the 10-month cohort (Fig. 2B), and centred around upregulation of Cdc42, a regulator of cell polarity and cytoskeletal organisation, indicative of cytoskeletal remodelling (Fig. 2A-B). Upregulation of Il6st, a key proponent of astrocyte inflammatory responses and mediator of cytokine signalling, was detected in both heterozygous and homozygous 10-month mice, alluding to significant changes in inflammatory signalling and sensitivity. IPA also showed dysregulation in synaptogenesis signalling pathways in the homozygotic CcnfS621G mice. MCL clustering confirmed further alterations in glutamatergic synapse regulation, as well as synaptic structure and vesicle release, potentially alluding to alterations in neuronal signalling and function (Figs. 2B and 3B). Finally, only a single protein, RPL17, a central component of the 60S ribosomal subunit, was significantly downregulated in all mice, regardless of genotype or age. RPL17 also contributes to the predicted dysregulation of the nonsense-mediated decay pathway in all mice, via other proteins, such as Rlp39, Rpl32, Rps14, Eif4a3, Rpl29 and Magohb. Overall, the proteomic analysis of CcnfS621G mice revealed significant alterations in major pathways associated with translation and mitochondrial function, with IPA analysis and MCL clusters also identifying alterations in cytoskeletal remodelling, inflammatory signalling, nonsense-mediated decay and synaptic transmission. Because these changes occurred in the absence of neuronal loss at the examined timepoints, the data provide insight into pathway alterations that arise early in disease progression. Moreover, these findings identified phenotypes for downstream validation in the context of human disease under endogenous expression conditions. Increased ramification of GFAP+ve astrocytes in human post mortem sporadic pure ALS and FTD-ALS donor brain tissue To examine whether the morphological phenotype identified in CcnfS621G mouse astrocytes is present in the broader population of sporadic ALS and FTD-ALS human donors, we analysed post mortem tissue from sALS and FTD-ALS donors and compared astrocyte morphologies to healthy control donors. Brain regions were selected based on availability and demographics to represent major regions impacted in ALS (motor cortex) and FTD-ALS (superior temporal gyrus). Consistent with the phenotypes identified in CcnfS621G mouse tissue, we identified significant differences in astrocyte branching across both sALS and FTD-ALS cases. In the motor cortex, significant increases in astrocyte branching were identified in both pure sporadic ALS and FTD-ALS cases (Fig. 4A). Whereas, in the superior frontal cortex, a brain region particularly relevant to FTD, significant increases in astrocyte branching were only detected in FTD-ALS donors, compared to healthy controls (Fig. 4B). Qualitative assessment of motor cortex GFAP+ve astrocytes revealed a similar morphology in a CCNFS621G fALS donor (n = 1) and sALS cases (Supplementary Fig. 10). Due to the availability of only a single CCNFS621G human post mortem tissue case, statistical analyses were not performed.Fig. 4Human post mortem sALS and FTD-ALS donor GFAP+ astrocytes exhibit region-specific increases in arborisation. Sections from the Motor Cortex and Superior Frontal Cortex of healthy control, sALS and FTD-ALS were immunohistochemically stained for GFAP. A-B Representative images and single-cell morphological analysis of grey matter GFAP+ve astrocytes from the motor cortex (A—11533 cells analysed across 4 × Healthy Control, 8 × sALS, 3 × FTD-ALS donors) and superior frontal cortex (B—10045 cells analysed across 3 × Healthy Control, 5 × sALS, 4 × FTD-ALS donors). Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2) or gamma distributions and donor origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**). N = 3–8 cases per group. All images were captured at 20X magnification on a Leica DM750 microscope with attached K3C colour camera, and identical imaging settings. Scale bar is equal to 100 µm and is equivalent across all images Human post mortem sALS and FTD-ALS donor GFAP+ astrocytes exhibit region-specific increases in arborisation. Sections from the Motor Cortex and Superior Frontal Cortex of healthy control, sALS and FTD-ALS were immunohistochemically stained for GFAP. A-B Representative images and single-cell morphological analysis of grey matter GFAP+ve astrocytes from the motor cortex (A—11533 cells analysed across 4 × Healthy Control, 8 × sALS, 3 × FTD-ALS donors) and superior frontal cortex (B—10045 cells analysed across 3 × Healthy Control, 5 × sALS, 4 × FTD-ALS donors). Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2) or gamma distributions and donor origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**). N = 3–8 cases per group. All images were captured at 20X magnification on a Leica DM750 microscope with attached K3C colour camera, and identical imaging settings. Scale bar is equal to 100 µm and is equivalent across all images Together, these data provide evidence that the alterations in astrocyte morphology identified in the CcnfS621G mice are also present in human sALS and FTD-ALS post mortem tissue in disease-specific regions. CCNFS621G iPSC-derived astrocytes display modest increases in ramification following inflammatory cytokine stimulation We previously showed that CCNFS621G iPSC-derived motor neurons display alterations in the ubiquitinated proteome, as well as activation of apoptotic pathways [29, 31, 32]. However, the phenotype of CCNFS621G iPSC-derived astrocytes has yet to be investigated. Given that the ramified astrocyte phenotype identified in human sporadic ALS and ALS-FTD tissue post mortem was similar to the hippocampal alterations in CcnfS621G mice, we assessed the same phenotype in human CCNFS621G iPSC-derived astrocytes. Human fALS CCNF donor CCNFS621G/wt and isogenic control (CCNFIC) astrocytes were differentiated from iPSCs, producing a highly enriched population of GFAP-positive astrocytes (Supplementary Fig. 15). Examination of Cyclin F fluorescence intensity across CCNFS621G and isogenic control astrocytes revealed no significant perturbation of Cyclin F expression or localisation (Supplementary Fig. 16). Morphological analysis of 14,193 astrocytes revealed a modest increase in ramification in CCNFS621G astrocytes, compared to CCNFIC astrocytes under inflammatory conditions (Fig. 5A-B). Whilst detected as statistically significant, mutation-specific differences produced minimal effect sizes (6–8%), compared to inflammation-induced shifts (20–30%). Overall, these data demonstrate that iPSC-derived CCNF astrocytes respond morphologically to inflammatory stimuli. Furthermore, the ramified astrocyte phenotype detected in the CRISPR/Cas9 CcnfS621G mouse model and in human post mortem sALS and FTD-ALS tissue is recapitulated in iPSC-derived CCNFS621G astrocytes, with larger alterations in astrocyte morphology under inflammatory conditions.Fig. 5CCNFS621G iPSC-derived astrocytes display modest alterations in cellular morphology under inflammatory conditions. A Representative processing of GFAP-stained iPSC-derived astrocytes and skeleton overlay output from a custom CellProfiler image analysis pipeline. B Single-cell skeleton analysis of iPSC-derived astrocytes, comparing major branches (trunks), secondary and tertiary branches (non-trunk branches) and branch ends, cell skeleton length (measured in pixel units) (under both basal (-) and inflammatory (+, 48 h co-stimulation with 50 ng/ml TNF-α and 10 ng/ml IL-1β) conditions. A total of 14,193 GFAP-positive astrocyte skeletons were analysed across three biological replicates, the means of which are denoted with data points. Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2), gamma, or tweedie distributions and differentiation origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Statistically significant comparisons with an effect size < 10% are denoted with ### CCNFS621G iPSC-derived astrocytes display modest alterations in cellular morphology under inflammatory conditions. A Representative processing of GFAP-stained iPSC-derived astrocytes and skeleton overlay output from a custom CellProfiler image analysis pipeline. B Single-cell skeleton analysis of iPSC-derived astrocytes, comparing major branches (trunks), secondary and tertiary branches (non-trunk branches) and branch ends, cell skeleton length (measured in pixel units) (under both basal (-) and inflammatory (+, 48 h co-stimulation with 50 ng/ml TNF-α and 10 ng/ml IL-1β) conditions. A total of 14,193 GFAP-positive astrocyte skeletons were analysed across three biological replicates, the means of which are denoted with data points. Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2), gamma, or tweedie distributions and differentiation origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Statistically significant comparisons with an effect size < 10% are denoted with ### CCNFS621G iPSC-derived astrocytes present significant deficits in mitochondrial network morphology and membrane potential under basal and inflammatory conditions Mitochondrial-associated proteins were among the most consistently dysregulated in the Ccnf mouse proteomics across ageing (Figs. 2 and 3) and we previously identified oxidative phosphorylation pathways to be depleted in the iPSC-derived CCNFS621G motor neuron ubiquitome [29]. Consequently, we assessed changes in mitochondrial network morphology and mitochondrial membrane potential in iPSC-derived astrocytes under basal and inflammatory conditions. TMRE fluorescence was used to quantify mitochondrial membrane potential at the single-cell level, revealing mitochondria present either as an extensive network or as single, isolated organelles. These data revealed significant depolarisation of networked mitochondrial membranes under basal conditions, which was further compounded by inflammatory stimulation in monoallelic CCNFS621G/wt and biallelic mutation CCNFS621G/ΔPEST astrocytes (Fig. 6A-B). In parallel, significant reductions in mitochondrial network branches, branch ends, and skeleton length were identified. These data demonstrate a genetic dose-dependent phenotype when comparing CCNFS621G/wt and CCNFS621G/ΔPEST astrocytes under both basal and inflammatory conditions (Fig. 6B). Meanwhile, isolated mitochondria displayed significant shifts in membrane potential under basal conditions, suggesting hyperpolarisation in CCNFS621G/ΔPEST astrocytes, with no significant differences identified in inflammatory stimulated cells (Fig. 6B). Measurements of isolated mitochondrial unit ‘major axis length’ were generated from CellProfiler outputs and represent the longest diameter of the isolated mitochondrial unit, providing a basic quantitation of mitochondrial truncation, which may broadly be suggestive of fragmentation or heightened trafficking. Following inflammatory cytokine exposure, a reduction in the major axis length of isolated mitochondria was detected across all genotypes. Notably, this phenotype was exacerbated in stimulated CCNFS621G/ΔPEST astrocytes, compared to stimulated CCNFIC astrocytes (Fig. 6B), suggesting truncation or increased trafficking of mitochondrial units under inflammatory and disease contexts. Representative TMRE staining images from all CCNF astrocytes are provided in Supplementary Fig. 17.Fig. 6CCNFS621G iPSC-derived astrocytes display pronounced alterations in mitochondrial membrane potential and network morphology under both basal and inflammatory conditions. A Example images of inflammatory cytokine-treated iPSC-derived astrocytes stained with 50 nM TMRE, demonstrating pre- and post-processing in CellProfiler, segregating and skeletonising mitochondrial networks utilising CellTracker Green and Hoechst for single-cell segmentation. B Single-cell analysis of mitochondrial network membrane potential and network skeleton morphology under both basal (-) and inflammatory (+, 48 h co-stimulation with 50 ng/ml TNF-α and 10 ng/ml IL-1β) condition, 2,219 mitochondrial networks analysed across 1,248 cells. All intensity measurements were normalised to the average residual intensity remaining following a 20 min pre-incubation with 20 µM FCCP. C Example segmentation of ‘non-networked’ isolated mitochondrial units. D Analysis of isolated mitochondrial unit membrane potential and major axis length. A total of 79,265 isolated mitochondrial units were analysed across 1,248 cells. All images were collected at 40X magnification with a digital zoom factor of 2.25 on a Leica SP8 LIGHTNING confocal microscope. Cells were maintained at 37 °C and 5% CO2 during imaging. Data points represent the total mean per replicate, with shapes representing paired biological replicates (N = 3). Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2, for integer skeleton counts), gamma (for skeleton length) or beta (TMRE Mean Intensity) distributions and differentiation origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Scale bars represent 50 µm CCNFS621G iPSC-derived astrocytes display pronounced alterations in mitochondrial membrane potential and network morphology under both basal and inflammatory conditions. A Example images of inflammatory cytokine-treated iPSC-derived astrocytes stained with 50 nM TMRE, demonstrating pre- and post-processing in CellProfiler, segregating and skeletonising mitochondrial networks utilising CellTracker Green and Hoechst for single-cell segmentation. B Single-cell analysis of mitochondrial network membrane potential and network skeleton morphology under both basal (-) and inflammatory (+, 48 h co-stimulation with 50 ng/ml TNF-α and 10 ng/ml IL-1β) condition, 2,219 mitochondrial networks analysed across 1,248 cells. All intensity measurements were normalised to the average residual intensity remaining following a 20 min pre-incubation with 20 µM FCCP. C Example segmentation of ‘non-networked’ isolated mitochondrial units. D Analysis of isolated mitochondrial unit membrane potential and major axis length. A total of 79,265 isolated mitochondrial units were analysed across 1,248 cells. All images were collected at 40X magnification with a digital zoom factor of 2.25 on a Leica SP8 LIGHTNING confocal microscope. Cells were maintained at 37 °C and 5% CO2 during imaging. Data points represent the total mean per replicate, with shapes representing paired biological replicates (N = 3). Statistical analysis was performed on single-cell data in RStudio using a generalised linear mixed effects model (glmmTMB), fitted with either negative binomial (nbinom2, for integer skeleton counts), gamma (for skeleton length) or beta (TMRE Mean Intensity) distributions and differentiation origin as a random effect. Estimated marginal means were obtained from this model using the emmeans package, and pairwise comparisons were performed. Raw p-values from these contrasts were adjusted for multiple comparisons using the false discovery rate (FDR) method; p < 0.05 (*), p < 0.01 (**), p < 0.001 (***). Scale bars represent 50 µm Collectively, these data demonstrate that CCNFS621G/wt and CCNFS621G/ΔPEST iPSC-derived astrocytes present significant perturbations to mitochondrial membrane potential, along with deficits in mitochondrial network structure and remodelling following inflammatory cytokine stimulation. Given the functional impact on CCNFS621G astrocyte mitochondria, coupled with the proteomic evidence indicating alterations in proteins related to synaptic transmission, we pursued further investigations into CCNFS621G astrocyte-driven neuronal dysfunction. CCNFS621G iPSC-derived astrocytes impact co-cultured motor neuron function Given that proteomic analysis in CcnfS621G mice revealed alterations in protein translation, mitochondrial pathways, cytoskeletal remodelling, inflammatory signalling, and synaptic function—yet without overt neuronal loss within their lifespan—we sought to determine whether these molecular and astrocytic phenotypes translate into early electrophysiological changes indicative of neurodegenerative processes. The CcnfS621G mice exhibited a ramified astrocyte morphological phenotype, which was also present in pure sporadic sALS and FTD-ALS human post mortem tissue and in human iPSC-derived astrocytes following inflammatory stimulation, demonstrating a consistent phenotype, regardless of the model or brain region assessed. The iPSCs provide a simple reductionist system with which to analyse disease-related alterations. An early phenotype of ALS identified in sporadic and familial ALS patients, animal models and human cell models is alterations in neuronal excitability. Multiple studies suggest the neurodegenerative pathway involves increased excitability, followed by decreased excitability, observed in humans with fALS and sALS and in cell and animal models of fALS [45, 46]. Furthermore, we previously showed that SOD1G93A mice exhibit a reduced proportion of repetitively firing motor neurons and an increased proportion of single spiking neurons in pre-symptomatic stages, with these single spiking neurons subsequently degenerating during symptomatic stages [46]. This positions neuronal firing properties as a readout of motor neuron vulnerability. We thus reasoned that the CCNFS621G mutation driven astrocyte alterations could alter neuronal signaling and function. To assess this, we differentiated motor neurons and astrocytes from iPSCs to model the astrocytic-motor neuron interaction in ALS and measure the electrophysiological properties of motor neurons in a “Mix and Match” system (Fig. 7A). As such, CCNFIC (isogenic control) or CCNFS621G/wt motor neurons were each cultured: (i) without astrocytes (mono-cultures) or (ii) with CCNFIC astrocytes, (iii) monoallelic CCNFS621G/wt astrocytes or (iv) biallelic CCNFS621G/ΔPEST (co-cultures).Fig. 7iPSC-derived CCNFS621G astrocytes abolish repetitive motor neuron firing and reduce sodium channel function of CCNFS621G and isogenic control motor neurons. A Astrocytes were added to motor neuron cultures after 2 weeks of maturation and the cultures were further matured for 2 weeks before electrophysiological analysis. B Action potential firing was assessed in current clamp; 10 pA step currents were injected from 0 to 100 pA until the cell fired an action potential. The neurons were stratified based on their spiking pattern. Purple bars represent repetitive firing neurons; green bars represent non repetitive firing neurons; black bars represent non-spiking motor neurons. The spiking pattern was analysed using pClamp10 software, with CCNFIC motor neurons (left), CCNFS621G/WT motor neurons (right). C Na+ currents were recorded in voltage clamp. If needed, a current was injected to maintain the resting membrane potential of the cell at −60 mV. 10 mV voltage steps were injected from −90 to 100 mV and the current response was recorded from motor neurons of A CCNFS621G/WT, B CCNFIC. Symbols in orange or dark magenta indicate that motor neurons were co-cultured with astrocytes differentiated from ALS-patient derived iPSCs (CCNFS621G/wt or CCNFS621G/ΔPEST). Blue symbols indicate motor neurons co-cultured with iPSC-derived astrocytes from isogenic control lines where the CCNF mutation was reversed (CCNFIC). N = 3 biological replicates, with a total of, at minimum, 15 cells analysed per group. The plotted data represents the mean peak Na+ current at each voltage step ± SEM, as calculated using pClamp10. A two-way-ANOVA with Tukey’s post-hoc analysis was performed. Statistical significance is shown for the peak Na + current with *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 iPSC-derived CCNFS621G astrocytes abolish repetitive motor neuron firing and reduce sodium channel function of CCNFS621G and isogenic control motor neurons. A Astrocytes were added to motor neuron cultures after 2 weeks of maturation and the cultures were further matured for 2 weeks before electrophysiological analysis. B Action potential firing was assessed in current clamp; 10 pA step currents were injected from 0 to 100 pA until the cell fired an action potential. The neurons were stratified based on their spiking pattern. Purple bars represent repetitive firing neurons; green bars represent non repetitive firing neurons; black bars represent non-spiking motor neurons. The spiking pattern was analysed using pClamp10 software, with CCNFIC motor neurons (left), CCNFS621G/WT motor neurons (right). C Na+ currents were recorded in voltage clamp. If needed, a current was injected to maintain the resting membrane potential of the cell at −60 mV. 10 mV voltage steps were injected from −90 to 100 mV and the current response was recorded from motor neurons of A CCNFS621G/WT, B CCNFIC. Symbols in orange or dark magenta indicate that motor neurons were co-cultured with astrocytes differentiated from ALS-patient derived iPSCs (CCNFS621G/wt or CCNFS621G/ΔPEST). Blue symbols indicate motor neurons co-cultured with iPSC-derived astrocytes from isogenic control lines where the CCNF mutation was reversed (CCNFIC). N = 3 biological replicates, with a total of, at minimum, 15 cells analysed per group. The plotted data represents the mean peak Na+ current at each voltage step ± SEM, as calculated using pClamp10. A two-way-ANOVA with Tukey’s post-hoc analysis was performed. Statistical significance is shown for the peak Na + current with *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 Since astrocytes can influence neuronal firing properties [47, 48], investigations were conducted into the active membrane properties of iPSC-derived motor neurons co-cultured with iPSC-derived astrocytes. To examine whether the addition of astrocytes would alter this phenotype, action potential firing was quantified in the ‘Mix and Match’ co-cultures (Fig. 7B). Neurons were stratified into three categories: (1) repetitive firing neurons; (2) non repetitive firing neurons (i.e. single action potential; single spiking); (3) non-spiking neurons (no action potential (AP)). Consistent with an increased neuronal excitability phenotype, an increase in repetitive action potential firing of CCNFS621G/wt and CCNFS621G/ΔPEST iPSC-derived motor neurons was observed, compared to CCNFIC, when neurons were cultured without astrocytes (Fig. 7B). The most noticeable changes that occurred with the addition of astrocytes to motor neuron cultures were in the population of repetitively firing neurons and non-spiking neurons. In CCNFS621G/wt motor neurons, the addition of astrocytes reduced the percentage of repetitive firing neurons from 25% in mono-cultures to approximately 10% in CCNFS621G/wt motor neurons + CCNFIC astrocyte co-cultures. Similarly, the addition of CCNFIC astrocytes to CCNFS621G/ΔPEST motor neurons caused a 14% reduction in repetitively firing neurons. CCNFS621G/wt astrocytes abolished the proportion of motor neurons that fired repetitively in both CCNFIC and CCNFS621G/wt motor neurons (Fig. 7B). CCNFS621G/wt motor neurons + CCNFS621G/wt astrocyte co-cultures lost the capacity for repetitive firing and had the highest proportion of non-firing neurons (36%) (Fig. 7B). For co-cultures established with CCNFIC motor neurons, the addition of CCNFIC astrocytes increased repetitive firing neurons to 17%, compared to CCNFIC motor neuron mono-cultures (10%). As observed in the CCNFIC motor neurons + CCNFS621G/wt astrocyte co-cultures, CCNFIC motor neurons + CCNFS621G/wt astrocytes completely abolished the population of repetitive firing neurons. Action potential firing and shape are largely influenced by the influx and efflux of Na+ and K+ ions. Previous research using measurements in living patients suggested that voltage-gated sodium channel activity changes drive the alterations in excitability in ALS during degeneration [49]. The CCNFS621G/wt motor neurons alone had a peak Na+ current of 1.2 ± 0.3 nA. In CCNFS621G/wt motor neurons + CCNFIC astrocyte co-cultures, no significant differences in Na+ currents were observed, with the peak current being 0.9 ± 0.1 nA and 1.2 ± 0.1 nA, respectively. However, the peak Na+ current in neurons from CCNFS621G/wt motor neurons + CCNFS621G/wt astrocyte co-cultures was significantly reduced to 0.7 ± 0.1 nA, compared to CCNFS621G/wt motor neurons cultured alone (P ≤ 0.05) (Fig. 7C). A similar effect was observed in CCNFIC motor neurons, whereby CCNFIC motor neurons + CCNFIC astrocytes exhibited an increase in Na+ currents from 0.9 ± 0.1 nA (CCNFIC MNs) to 1.3 ± 0.1 nA (P ≤ 0.001). Conversely, CCNFIC motor neurons + CCNFS621G/wt astrocytes showed a significant reduction in Na+ currents compared to CCNFIC motor neurons + CCNFIC astrocyte co-cultures (0.7 ± 0.1 nA, P ≤ 0.0001; 1.0 ± 0.1 nA, P ≤ 0.05, respectively) (Fig. 7C). Together these data show that action potential firing properties of motor neurons are impacted by CCNFS621G astrocytes. CCNFS621G/wt and CCNFS621G/ΔPEST iPSC-derived motor neurons cultured without astrocytes exhibited increased action potential firing and increased Na+ currents, compared to isogenic control neurons. However, the addition of CCNFS621G/wt iPSC-derived astrocytes drove a loss of repetitive action potential firing in both CCNFS621G and CCNFIC neurons. CCNFS621G astrocytes also increased the proportion of neurons that failed to fire an action potential and reduced Na+ currents. Together these data suggest that CCNFS621G astrocytes reduce neuronal firing. Discussion This study presents the first CRISPR/Cas9-generated mouse model of the ALS-FTD-linked CCNFS621G mutation, enabling investigation of the molecular effects of a disease-causing mutation under physiological expression levels. Despite an absence of motor or neuronal death phenotypes within the lifespan of these mice, we identified early shifts in behaviour and hippocampal astrocyte morphology, along with widespread proteomic alterations in pathways related to ubiquitin signaling, mitochondrial function, translation, cytoskeletal remodelling, inflammatory signalling, and synaptic transmission in CcnfS621G mice. By integrating these findings with analysis from human post mortem tissue and donor-derived iPSC models, we reveal convergent astrocytic alterations and demonstrate the ability of CCNFS621G astrocytes to impair motor neuron excitability in a non-cell autonomous manner. Together, our results suggest that under endogenous conditions in mice, CCNFS621G does not directly drive neurodegeneration in isolation, but instead promotes early cellular dysfunction and glial-neuronal crosstalk alterations, which may prime the CNS for neurodegeneration. Our data thus correlates with and supports emerging multistep models of ALS and FTD pathogenesis [50]. There is a growing appreciation that the onset of ALS requires multiple steps, some mediated by genetic factors and others influenced by environmental exposures and ageing. In the controlled housing environment of the CcnfS621G mice, additional triggers may be required to drive ALS onset, including inflammatory activation, protein aggregation, TDP-43 mislocalisation or other genetic or epigenetic susceptibilities. In this way, it is plausible that CCNFS621G acts as an oligogenic contributor towards ALS and FTD pathogenesis, lowering the threshold for neurodegeneration in the presence of additional environmental insults or co-occurring genetic vulnerabilities. In the original multi-generational family described in Williams et al. 2016, carriers of CCNFS621G exhibited marked heterogeneity in phenotypic presentation, with a mix of pure ALS, FTD-ALS or FTD, supporting the presence of additional disease-modifying factors that determine disease manifestation. Further support for an oligogenic role of Cyclin F in ALS arises from the presence of CCNF variants identified in the wider sALS population [3], suggesting that CCNF variants may act in concert with other known oligogenic contributors, such as NEK1, ANXA11 and TBK1 [51, 52]. The lack of motor or late-stage hallmark cellular pathology in CRISPR/Cas9-generated CcnfS621G mice is not unexpected, given the endogenous expression of Cyclin F and the shorter lifespan of mice compared to the typical age of FTD/ALS onset in humans. Previous studies featuring endogenous levels of ALS-FTD-linked mutations often have not shown severe motor phenotypes or overt signatures of neurodegeneration within the lifespan of mice [53]. For example, mouse models endogenously expressing Tdp43Q331K [54] or Tdp43M337V [55] displayed no significant motor decline or overt neurodegeneration after 18 months (TdpQ331K) or 2 years (TdpM337V). However, the mice instead presented significant cognitive decline (TdpQ331K) and astrogliosis prior to neuronal loss phenotypes (TdpM337V, prior to minor neuronal loss at 2.5 years) [54, 55]. Despite a frequent lack of advanced disease phenotypes, endogenous models still hold relevance as they provide valuable insight into the pre-motor disease state, which is vital for the identification and understanding of early disease biomarkers and treatment options to prevent symptom onset. With regards to CCNF, other mouse models have employed AAV-mediated overexpression of human Cyclin FS621G [33]. These mice present with FTD-relevant cognitive decline and behavioural abnormalities, along with accumulation of ubiquitinated proteins and cytoplasmic inclusions of TDP-43 [33]. Interestingly, GFAP fluorescence intensity was reduced in the cortices of only mice overexpressing Cyclin FS621G [33]. As protein levels of another astrocyte marker S100β were unchanged, this was interpreted as a decrease in astroglial activation, relative to EGFP and Cyclin Fwt mice [33]. Given that accumulation of ubiquitinated proteins and TDP-43 aggregates are typically potent neuroinflammatory stimuli, this may signal a pathological impairment of astrocyte activation and implicates Cyclin F in astrocytic activation. Notably, overexpression of wild-type Cyclin F in these mice was also sufficient to drive similar pathogenic phenotypes, highlighting the importance of CCNF levels in neuronal homeostasis [33]. In comparison, the endogenous expression of native mouse Cyclin FS621G in this study produced milder behavioural and cognitive alterations in the absence of TDP-43 pathology, or direct neuronal loss. Conversely, hallmark pTDP-43 and p62 pathology were observed in human CCNFS621G post mortem ALS donor tissue, with significant glial cell involvement. Thus, this endogenous Ccnf mouse model may be recapitulating an earlier phase of disease progression. The lack of other genetic and environmental risk factors, such as ageing, may limit the scope of CCNF pathology that can be modelled under endogenous physiological conditions without compounding oligogenic or environmental stressors. Consistent with the likelihood that CRISPR/Cas9 CcnfS621G mice recapitulate early disease alterations, the comprehensive motor and behavioural phenotyping did not reveal any major motor impairments for heterozygote or homozygote CcnfS621G mice of either sex. In the accelerod, homozygote females showed more loops compared to heterozygotes but not wildtype females, suggesting that homozygote females slipped more often than heterozygote females; however latency to slip was similar across genotypes. Female Ccnf transgenic mice also displayed reduced locomotion across tests and lower levels of exploratory behaviour in the open field. Anxiety-related behaviours of Ccnf transgenic mice were not different from control mice, and whilst sociability of male mice was intact, social recognition memory was impaired in homozygote males. Female mice regardless of genotype did not display any preferences across the two test trials. The fear-associated memory to a context was not affected in Ccnf transgenic mice of either sex. However, the conditioned freezing response to a cue was impaired in homozygote females compared to both wildtype, as well as heterozygote females. Finally, the acoustic startle response and sensorimotor gating of Ccnf transgenic mice were similar to control animals. Collectively, these subtle behavioural alterations, in the absence of neuronal loss, indicate the presence of innate cellular dysfunction and motivate further examination of non-neuronal cell types, such as astrocytes, which can influence neuronal activity and ALS-FTD disease progression. Across ALS and FTD, astrocytes are key modulators of disease onset and progression, where mutations in major ALS-FTD-linked genes disrupt proteostasis and promote maladaptive inflammatory astrocyte responses that impair neuronal support and function. Alterations in astrocyte morphology are known to underpin the functional changes associated with inflammation, with increased branching or increased ‘ramification’ commonly associated with ‘reactive’ astrocyte states [56]. Such morphological parameters have previously been used to successfully discern these reactive astrocyte phenotypes across various astrocyte subpopulations [57]. In CcnfS621G mice, a significant increase in hippocampal astrocyte ramification was detected, a phenotype that displayed both time and dose-dependent (i.e., hetero/homozygous) characteristics. As similar shifts in astrocyte ramification were detected in the highly neuroinflammatory end-stage brain environment of human sALS and FTD-ALS donors and in iPSC-derived astrocytes under inflammatory stimulation, this alteration in morphology likely correlates with an increase in astrocyte reactivity. Whilst not encompassing the entire astrocyte population, GFAP-positivity has long been used to phenotypically discern reactive astrocytes [58]. Morphological alterations occurring in these potentially reactive GFAP+ populations, paired with a simultaneous increase in microglial density observed in the same region and timepoint in CcnfS621G mice, may suggest a ‘priming’ of glial cells towards heightened responses to neuroinflammatory stimuli. This ‘priming’, evoked by CcnfS621G, may increase susceptibility to subsequent neuroinflammatory insults, including peripheral and systemic immune stressors, which are increasingly implicated in ALS pathogenesis [59–61]. Consistent with the astrocyte remodelling observed in CcnfS621G mice, astrocytes from human pure sporadic ALS and FTD-ALS donors displayed region-specific morphological profiles. In the motor cortex—a central region of ALS pathology, both sALS and FTD-ALS donor astrocytes exhibited significant increases in arborisation. However, in the superior frontal cortex, a region commonly affected in FTD and typically spared in pure sALS [62], FTD-ALS astrocytes were significantly more ramified. This region-dependent morphological shift may link with the spread of ALS/FTD-linked neuropathology across brain regions during disease progression [63–65], potentially underlying the selective vulnerability and heightened involvement of particular brain regions throughout disease progression. These morphological shifts in the sporadic human cohort, whilst not driven by CCNF variants, to the best of our knowledge, enable us to contextualise the morphological shifts observed in the CCNF mouse astrocytes, and support an increase in astrocyte reactivity. To explore the molecular mechanisms underpinning these astrocyte morphological shifts, and the cellular pathways affected by endogenous Cyclin FS621G expression, proteomic analyses were performed on CcnfS621G mice. The pathways altered in Ccnf mouse proteomics likely represent early changes in the neurodegenerative disease pathway, preceding potential neuronal loss. Potentially relevant for astrocytic remodelling, clusters of proteins related to actin polymerisation and cytoarchitectural remodelling were identified in the CcnfS621G mouse proteomic analysis. These clusters in 10-month mice centred around upregulation of Cdc42 and altered Rho GTPase signalling. Prior studies manipulating Cdc42 in astrocytes have demonstrated significant alterations in astrocyte polarity, morphology and migration to sites of injury, which may suggest that these broader processes are altered in CcnfS621G mice [66–68]. Beyond astrocytes, significant perturbations to axonal outgrowth have previously been detected in the spinal cords of zebrafish transiently overexpressing exogenous Cyclin FS621G [31]. The neuronal branching deficits observed in this model may potentially be underpinned by Cyclin FS621G-driven alterations in synaptogenesis pathways, along with Cdc42 and Rho GTPase signalling, which are heavily involved in neurite growth cone dynamics [69]. Additionally, 10-month CcnfS621G mouse proteomics revealed increased expression of the interleukin-6 signal transducer, Il6st, also known as gp130. This protein complexes with the interleukin-6 receptor upon cytokine binding and acts to mediate inflammatory signalling. Il6st upregulation may thus indicate increased neuroinflammation or increased sensitivity to inflammatory stimuli [70], providing further support for a potential ‘priming’ of enhanced neuroinflammatory responses in Ccnf mice. One of the largest and most conserved pathways altered in the Ccnf mouse proteome across 10 and 18-months was related to translation. These alterations are indicative of translational reprogramming and were likely mediated through the regulation of eukaryotic translation initiation factors, such as Eif2, Eif3 and Eif5, which may represent an effort to maintain existing cellular protein levels and reduce ER burden. In line with these findings, cells over-expressing CCNFS621G have previously been shown to interrupt transport along the ER-Golgi axis, promoting ER-stress, ER associated protein degradation (ERAD) and Golgi fragmentation [30]. Cyclin F has also previously been identified to interact with canonical EIF2 signalling pathways [25], and it is therefore plausible that Cyclin FS621G could influence the regulation of these proteins, thereby driving widespread changes in cellular protein translation. In addition to translation, proteomic analyses in Ccnf mice also consistently highlighted mitochondrial function as a highly altered pathway, encompassing mitochondrial translation, mitophagy, fission–fusion dynamics and oxidative phosphorylation. In 10-month mice, mitochondrial electron transport chain complex subunits were upregulated, signalling increased mitochondrial oxidative phosphorylation activity, potentially as an early compensatory response against emerging proteostasis stress. Consistent with this, presymptomatic Alzheimer’s disease mouse models also exhibit early mitochondrial hyperactivity, which has been proposed to drive increased ROS production and thus mitochondrial-linked neuroinflammation [71]. Importantly, as disease advances, these increases in mitochondrial activity were observed to decline, leading to impaired bioenergetic capacity [71]. In line with this trajectory, numerous ETC components were downregulated in homozygous 18-month mice. Comparable reductions in ETC components have also been observed in sALS iPSC-derived motor neurons and attributed to significant increases in ROS production and energetic deficits [72]. Together, the data suggest a transition from early mitochondrial hyperactivity to potential later-stage hypoactivity in CcnfS621G mice. The 18-month mice also exhibited altered expression of proteins involved in mitochondrial fission/fusion dynamics, including Mfn1 upregulation and downregulation of Mff, suggesting enhanced mitochondrial fusion and reductions in mitochondrial fission. Downregulation of Rhot2 in 18-month CcnfS621G mice may also indicate a reduction in mitochondrial trafficking, and alterations to mitochondrial autophagy (mitophagy) [73, 74]. Mitophagy is essential in the maintenance of healthy mitochondria and facilitates the removal of damaged mitochondria. The primary mechanism involves the kinase PINK1 and the E3 ligase Parkin, and is heavily dependent upon ubiquitin as a signalling mediator [75]. We previously showed that CCNFS621G impacts the cellular ubiquitome (ubiquitin-bound proteome), leading to a depletion in the pool of available ubiquitin and widespread disruption of UPS regulation in iPSC-derived motor neurons [29], which is likely to affect mitophagy. Impairments to mitophagy, along with altered fission–fusion dynamics is likely detrimental to mitochondrial network maintenance [76], leading to the accumulation of damaged mitochondria, further compounding mitochondrial dysfunction and oxidative stress. In accordance with the mouse proteomic data, examination of human iPSC-derived monoallelic CCNFS621G/wt and biallelic mutation CCNFS621G/ΔPEST astrocyte mitochondria revealed significant membrane depolarisation and reduced mitochondrial network complexity, both in basal and inflammatory contexts. Notably, the severity of mitochondrial network disruption scaled with increasing mutation burden across CCNFS621G/wt and CCNFS621G/ΔPEST astrocytes under basal conditions, potentially signalling innate mitochondrial dysfunction caused by CCNFS621G. Previous studies have demonstrated that mitochondrial dysfunction sensitises astrocytes to heightened neuroinflammatory responses, with detrimental impacts upon the physiology and function of co-cultured neurons, following inflammatory cytokine stimulation [77]. Treatment with inflammatory cytokines elicited partial mobilisation of mitochondrial networks in CCNFS621G/wt astrocytes, increasing mitochondrial network complexity from basal conditions. However, this response appeared to be muted in stimulated CCNFS621G/ΔPEST astrocytes, which presented less complex mitochondrial networks than both stimulated CCNFIC and CCNFS621G/wt astrocytes. Mitochondrial network remodelling and mitochondrial dynamics facilitate bioenergetic flexibility, which is critical in cellular responses to stress [78, 79]. A muted mitochondrial network remodelling response in CCNFS621G/ΔPEST astrocytes, along with depolarisation of mitochondria in both mono- and bi-allelic CCNFS621G astrocytes during inflammatory stimulation, may therefore indicate an impaired ability to adapt their metabolic state to inflammatory stress. Impairments to astrocyte metabolic support can disrupt neurotransmitter homeostasis, alter Ca2+ homeostasis, impair glutamate uptake and reduce the availability of astrocyte-derived metabolic substrates for neurons [80, 81]. These deficits can alter astrocyte-mediated regulation of synaptic transmission [82]. Aside from potential deficits in mitophagy, fission and fusion, a potential mechanism underpinning these mitochondrial deficits in iPSC-derived CCNFS621G astrocytes may also involve SQSTM1 (p62), an autophagy adaptor protein responsible for directing ubiquitinated cargo to autophagy machinery [83]. Previous studies have highlighted a distinct role of p62 in mitochondrial dynamics, bioenergetics and mitochondrial DNA biogenesis [84, 85]. In the context of inflammation, studies have demonstrated that autophagy is crucial in the maintenance of mitochondrial network integrity in astrocytes [86]. CCNFS621G has previously been demonstrated to alter p62 solubility, likely affecting its ability to undergo liquid–liquid phase separation, a key feature of p62 function [26, 87]. Thus, the mitochondrial dysfunction observed in CCNFS621G astrocytes may result, at least in part, from Cyclin FS621G-mediated alterations to p62 function, compromising mitochondrial network maintenance, membrane potential and function. Finally, both CCNFS621G/wt and CCNFS621G/ΔPEST iPSC-derived astrocytes drove significant neuronal excitability changes in a co-culture system, with either ‘healthy’ (isogenic control), CCNFS621G/wt or CCNFS621G/ΔPEST iPSC-derived motor neurons. CCNFS621G motor neurons cultured alone exhibited increased action potential firing and Na+ currents. However, in co-cultures, the CCNFS621G astrocytes reduced motor neuron firing and Na+ currents of either isogenic control or CCNFS621G motor neurons. This suggests that CCNFS621G astrocytes exert a dominant non-cell autonomous modulatory effect on neuronal excitability that may be driven in part by astrocytic mitochondrial dysfunction. Similarly, Zhao and colleagues have also reported that C9ORF72 astrocytes caused a progressive loss of motor neuron action potential firing and a reduction in voltage-gated currents [88]. This further supports the notion that astrocyte dysfunction influences neuronal excitability adaptation in ALS. More broadly, astrocytes are increasingly recognised as active metabolic partners of neurons that couple neuronal activity to local energy supply [89, 90]. Astrocyte-derived lactate, for example, has been shown to directly regulate neuronal excitability in mouse models of neuropsychiatric disorders, highlighting the importance of astrocyte-mediated metabolic support for neuronal function. These findings raise the possibility that mitochondrial dysfunction in CCNFS621G astrocytes impairs astrocytic-neuronal metabolic and homeostatic support, thereby reshaping motor neuron excitability in co-culture [91, 92]. These studies underpin previously unexplored elements of non-cell autonomous CCNFS621G pathology in astrocytes and neurons, providing further evidence for an oligogenic role of CCNF in ALS/FTD pathogenesis. Conclusions This study presents the first CRISPR/Cas9-generated mouse model of CcnfS621G, demonstrating that endogenous expression in mice is insufficient to drive advanced neurodegenerative cellular or motor pathology, but instead produces subtle behavioural and astrocytic morphological alterations in the presence of widespread proteomic remodelling. The CcnfS621G mice thus likely recapitulate pathways altered early in the neurodegenerative cascade. Integration with human post mortem and human iPSC data revealed convergent astrocyte dysfunction and non-cell autonomous impacts of CCNFS621G astrocytes on motor neurons, including reduced action potential firing and Na+ currents. Together these findings support a model in which CCNFS621G acts as an oligogenic contributor to ALS and FTD pathogenesis, promoting early cellular dysfunction that lowers the threshold for neurodegeneration and primes the CNS for maladaptive inflammatory responses. Supplementary Information Supplementary Material 1. Supplementary methods [93]. Supplementary results [94, 95]. Supplementary Material 1. Supplementary methods [93]. Supplementary results [94, 95].