Background
C9ORF72 hexanucleotide repeat expansion (
C9ORF72RE) is the most common mutation found within the ALS-FTD spectrum, giving rise to incurable, rapidly progressive and fatal disease pathologically characterised by degeneration of cortical neurons and upper and spinal motor neurons. Cortical circuit dysfunction is a consistent and prominent finding in
C9ORF72RE patients [
5,
42,
50,
62]. Altered cortical network excitability is considered to be an early pathogenic driver of ALS and FTD contributing directly to excitotoxicity-mediated neurodegeneration of upper motor neurons and cortical neurons [
21,
22,
36,
37,
58‐
60,
62]. Furthermore, clinical neurophysiological studies of
C9ORF72RE patients have demonstrated notable impairments in cortical network plasticity at the pre-symptomatic stage [
5]. For many progressive neurodegenerative diseases, including ALS-FTD, functional impairments in plasticity are thought to manifest early in disease progression, being representative of altered synaptic homeostasis that precede and potentially cause neuronal dysfunction and/or loss, and lead to cognitive impairments [
38,
55,
56].
Our current mechanistic understanding of potential sources of altered cortical network excitability in ALS-FTD is derived largely from mutant murine models (SOD1 and TDP-43 mutations) of ALS and ALS-FTD [
17,
30,
45,
48,
65], but does not extend to provide a physiological basis for altered network excitability. Similarly, our understanding of the potential synaptic plasticity dysregulation that may occur in ALS-FTD has come from studies that use ex vivo brain slice preparations from rodent models of rare genetic mutations [
24,
54]. Functional impairments in synaptic plasticity in ALS-FTD have yet to be examined in a human-based model system. Identified physiological disturbances in ALS motor neurons [
55] also may provide insights into cortical neuron pathophysiology though this must remain highly tentative given that diverging potential pathophysiological mechanisms in
C9ORF72RE cortical and spinal neurons are established [
52]. Despite its proposed pathogenicity and early prominence, cortical dysfunction in ALS-FTD remains poorly defined at both the synaptic and network level.
To address this, we have used human induced pluripotent stem cell (iPSC) derived cortical neurons from patients harbouring
C9ORF72RE mutations, combined with gene-edited isogenic paired lines [
52], to interrogate the consequence of
C9ORF72RE on cortical neuronal physiology. In view of dysregulation of glutamate homeostasis being a major hypothesis underlying ALS-FTD [
9,
55], we have examined physiological perturbations in iPSC-derived glutamatergic cortical neurons. We determine that
C9ORF72RE neurons display altered network properties that are underpinned by synaptic dysfunction, but not altered intrinsic cellular excitability, and display impairments in synaptic plasticity. Our transcriptomic analysis highlights dysregulated molecular pathways in accordance with physiological observations. Our observations are notably different from those previously reported for
C9ORF72RE motor neurons and provide evidence of cortical-specific pathophysiology that may contribute to cortical dysfunction in ALS-FTD.
Methods
iPSCs
Dermal fibroblasts from patient and control individuals were obtained under full Ethical/Institutional Review Board approval at the University of Edinburgh. Fibroblasts were reprogrammed to iPSCs by either Sendai virus or retrovirus expressing OCT4, SOX2, C-MYC, and KLF4. iPSCs were maintained in Matrigel (BD Biosciences)-coated plastic dishes in E8 medium (Life Technologies) at 37 °C and 5% CO
2. Lines were derived from three patients harbouring repeat expansions in the
C9ORF72 gene [
52] and a healthy individual with no known association with neurodegenerative disease.
Anterior precursor (aNPC) derivation
Human iPSCs were maintained on Matrigel (Corning), with Advanced DMEM/F12, 20% Knockout Serum Replacement, FGF-2 (10 ng/mL), L-glutamine (1 mM), 2-mercaptoethanol (100 mM) and 1% penicillin/streptomycin (P/S). All media were obtained from Life Technologies. Human iPSCs were neurally converted in suspension in chemically defined medium as described [
6]. The media was changed to base media (Advanced-DMEM/F12, 1% P/S, 1% Glutamax, 1% N-2), 0.5% B-27, FGF-2 (2.5 ng/mL) upon observation of radially organised structures in neurospheres (10–21 days) and plated on Laminin (Sigma) coated tissue culture plates (Nunc) a week later. Neural rosettes were mechanically isolated, dissociated with Accutase (Sigma) and 20-40 k cells were plated in one Laminin-coated well of a 6-well plate in proliferation media (Base media, 0.2% B-27, 10 ng/mL FGF-2). aNPCs were grown to high density before passaging 1:2 with Accutase on laminin coated plates until passage 5–6 and maintained on 1:100 Reduced-growth factor Matrigel-coated plates thereafter or cryopreserved as described [
6].
Differentiation of aNPCs into cortical neuronal cultures
aNPCs were plated in default media on poly-ornithine (Sigma), laminin (Sigma), fibronectin (Sigma) and Matrigel-coated coverslips in which primary mouse astrocytes have been propagated. Primary mouse astrocytes were prepared as previously described [
26]. Density of astrocytes was 100,000 per 13 mm coverslip at least 48 h prior to plating aNPCs. Cultures were fed twice a week. Default medium was supplemented with forskolin (10 μM, Tocris) from days 7–21 after aNPC platedown (200,000 per coverslip) and with BDNF and GDNF (both 5 ng/mL) from day 28 onwards. Coverslips were then processed fixed and stained as previously described [
6]. Multi-electrode arrays were first coated with poly-D-lysine then the laminin, fibronectin and Matrigel-coating was applied to the region containing the electrode arrays (60MEA200/30iR-Ti, Multi Channel Systems). aNPCs were plated to the coating spot and left for 2 h to adhere. Array wells were then flooded with default media containing suspended DIV14 mouse astrocytes.
Immunohistochemistry
Five–six weeks old cultures on glass coverslips were fixed in 4% PFA at room temperature (RT) for 20 min. They were permeabilised with 0.1% tritonX-100, blocked with 6% goat serum and stained with primary antibodies against βIII-tubulin (dilution 1:500, Sigma), human nuclei (dilution 1:200, Millipore), nestin (dilution 1:200, Millipore), GFAP (1:400, Sigma), synapsin-1 (dilution 1:500, Sigma) and PSD-95 (dilution 1:250; Neuromab) sequentially for 2 h at RT. These were then probed with appropriate secondary antibodies and mounted with FluorSave and imaged in Zeiss LSM Z10 confocal microscope using 63X objective. For synaptic density analysis, 5 fields of 20 μm region across 3 coverslips were analysed for the co-localised puncta of synapsin-1 and PSD-95 using colocalization plugin in ImageJ.
Total RNA was extracted from cortical neurons from 2 independent isogenic corrected paired cell lines at day 35 post platedown using RNeasy Mini kit (Qiagen), according to the manufacturer’s instructions. RNA samples were assessed for concentration (NanoDrop ND-100 Spectrometer, NanoDrop Technologies) and quality (Agilent 2200 Tapestation, Agilent Technologies) before library preparation. Library preparation and sequencing were carried out by Edinburgh Genomics (Edinburgh, UK). For each sample, cDNA was converted to a sequencing library using the TruSeq stranded mRNA-seq library. Barcoded libraries were pooled and sequenced on an Illumina HiSeq 4000 using 75 base paired-end reads to generate at least 111 million raw reads per sample. The reads were mapped to the primary assembly of the human (hg38) reference genome contained in Ensembl release 90 [
12]. Alignment was performed with STAR, version 2.5.3a [
16]. Tables of per-gene read counts were generated from the mapped reads with featureCounts version 1.5.2 [
33]. Differential gene expression analysis, using DESeq2 version 1.18.1, specifically examined the intersection in commonly and concordantly differentially expressed genes between the two mutant-isogene pairs, using a false discovery rate of 20%, achieved by setting a Benjamini-Hochberg corrected
p-value threshold of 0.2 (genes with an average FPKM < 1 were disregarded). Gene ontology (GO) analysis was performed on all the differentially expressed genes to identify putatively altered pathways or processes using topGO version 2.30.1 [
1]. RNA-seq data are available upon request to the corresponding authors.
Morphology
Cortical NPCs were sparsely transduced with lentivirus expressing GFP in order to label individual cells for analysis (ca. 1 viral particle to 10 cells). Following labelling with GFP, NPCs were differentiated as mentioned above and immunohistochemistry was performed against GFP and β3-tubulin. These were then probed with appropriate secondary antibodies and mounted with FluorSave and imaged in Zeiss LSM Z10 confocal microscope using 20X objective. Total neurite length (sum of all the processes) in the GFP channel was manually traced using ImageJ.
Multi-electrode array (MEA) electrophysiology
Extracellular recordings from 59 channels per array were acquired at 37 °C in the culture media using a Multi Channel Systems MEA system at a sampling rate of 20 kHz. Data was analysed using the Multichannel Systems software and in-house custom Matlab scripts.
Patch-clamp electrophysiology
For other electrophysiological experiments, whole-cell patch-clamp recordings were performed as described [
6,
34] using electrodes filled with (in mM): 155 K-gluconate, 2 MgCl
2, 10 Na-HEPES, 10 Na-PiCreatine, 2 Mg
2-ATP, and 0.3 Na
3-GTP, pH 7.3, 300 mOsm. For spontaneous action potential activity, cells were typically bathed in an extracellular recording comprising (in mM): 152 NaCl, 2.8 KCl, 10 HEPES, 2 CaCl
2, 10 glucose, pH 7.3, 320–330 mOsm. For mEPSC recordings, the extracellular solution was supplemented with TTX (1 nM), picrotoxin (50 μM) and MgCl
2 (1.5 mM). For intrinsic membrane and excitability properties, the extracellular solution was supplemented with CNQX (5 μM) and D-APV (50 μM). Recordings were performed at room temperature (20-23 °C). Current and voltage measurements were typically low-pass filtered online at 2 kHz, digitized at 10 kHz and recorded to computer using the WinEDR V2 7.6 Electrophysiology Data Recorder (J. Dempster, Department of Physiology and Pharmacology, University of Strathclyde, UK;
www.strath.ac.uk/Departments/PhysPharm/). Series resistance compensation was applied up to 75%. Recordings were omitted from analysis if the series resistance changed by more than 20% during the experiment, or if they exceeded 20 MΩ.
Burst analysis
Burst detection for both single cells and MEAs were performed using custom-written Matlab scripts. For patch-clamp recordings action potentials were identified using threshold detection (routinely set at − 10 mV) and bursts were defined as groups of action potentials with a minimum inter burst period set as log10 of the intra spike interval. For each MEA, 4–10 active channels were selected for further analysis. Bursts were identified as activity 2–5 times the standard deviation of the baseline (as determined by the signal-to-noise ratio) with a minimum quiet period set to define separate bursts. This was routinely set to 5 s given the lowest observed inter burst period was 9.6 s. The spike threshold was variable, but consistent between each of our isogenic and C9 pairs for each experiment and optimal to detect as many, but variable, number of channels with robust activity. On all MEAs there were both active and inactive channels (likely due to some electrodes not having active cells close enough) but all the active channels showed the same pattern of activity with low standard deviation in the burst start times across channels ranging from 0.10 to 0.78 s, indicating a synchronous network across the area of the MEA electrodes.
mEPSC analysis
mEPSC recordings were analysed offline using the WinEDR software stated above. A dead time window of 10 ms was set and individual mEPSCs were detected using an algorithm that selected for mEPSCs below a − 4 to – 6 pA amplitude threshold and greater than 1 ms in duration. mEPSCs that had a monotonic rising phase with a 10–90 rise time of lower than 6 ms and a Ƭ-decay with a decay time constant of lower than 25 ms were selected for analysis. Recordings were then visually inspected for validity. For mEPSC analysis (Fig.
2c), data were obtained from at least 2-min recordings and neurons that displayed mEPSC frequencies under 0.05 Hz were omitted from the analysis. For sucrose experiments, baseline mEPSC properties were determined from a 2-min stretch of mEPSC activity of at least 0.05 Hz. The transient phase was determined from the onset of sucrose application to the transition of mEPSC activity to steady-state activity. Steady-state data were determined from at least a 30 s stretch of mEPSC activity in continued presence of sucrose.
Statistical analysis
Statistical analysis was performed using GraphPad Prism software. Data are represented as mean ± s.e.m. *p < 0.05, **p < 0.01, ***p < 0.001. The number of experimental replicates (for MEA recordings, this indicates number of plates; for patch-clamp recordings this indicates number of cells) is denoted as n and N represents the number of independent de novo preparations of batches from which n is obtained. Data were initially determined to be parametric or non-parametric before applying either one-way ANOVA with Bonferroni’s multiple comparisons test or unpaired t-tests or Welch’s t-test, as appropriate.
Discussion
Increased synaptic glutamate transmission within the cortex presents an attractive hypothesis to potentially explain cortical network hyperexcitability present in early symptomatic
C9ORF72RE patients [
42,
50,
62] and ALS in general [
22]. Our data provide a human in vitro mechanistic exploration of physiological impairments in
C9ORF72RE patient-derived excitatory cortical neurons that reveal that perturbed network activity is underpinned by functional synaptic alterations that impact upon excitability. Furthermore, noting that ALS-FTD patients exhibit impairments in network plasticity, we have determined that
C9ORF72RE cortical neurons exhibit impairments in synaptic plasticity. Importantly, the physiological alterations observed in iPSC-derived
C9ORF72RE cortical neurons are disparate from that previously observed in iPSC-derived
C9ORF72RE motor neurons where intrinsic excitability appears to be primarily affected [
55]. Our data reveal that intrinsic excitability is not affected in
C9ORF72RE cortical neurons.
An increase in network burst frequency in
C9ORF72RE patient-derived excitatory cortical neurons is highly consistent with a mechanism requiring increased excitatory input. Our data demonstrate that
C9ORF72RE cortical neurons display an increased synaptic input as a result of an increased synaptic density. Such findings are broadly consistent with murine models of ALS, where increased synaptic input of excitatory cortical neurons are observed in pre-symptomatic mutant TDP-43 mice ([
17]; but see [
25]) and SOD
G93A mice [
17,
48,
57]. Cortical neurophysiological impairments were not found in a
C9ORF72RE murine model though this model does not display classical ALS-FTD pathology or neurodegeneration [
44]. Our transcriptomic approach has revealed potential causes to this increase in synaptic density. PCDHGC4, a γ-protocadherin, negatively regulates the function of neuroligin-1, a post-synaptic molecule that interacts with pre-synaptic neurexin to maintain and promote synapse structures in forebrain neurons [
39]. Reduced expression of PCDHGC4 in our
C9ORF72RE cortical neurons is therefore compatible with increased neuroligin-1 function and increased synaptic density. Overexpression of neuroligin-1 has previously been shown to increase excitatory synaptic activity in in vitro cortical neurons [
8]. Equivalently, CBLN1, is a pre-synaptically expressed molecule that interacts with neurexins and promotes synaptogenesis [
51] and is upregulated in
C9ORF72RE cortical neurons. Contrastingly, CBLN1 has been reported to be downregulated in
C9ORF72RE iPSC-derived motor neurons [
49]. Together, these studies indicate that increased cortical glutamate-mediated synaptic input is an early feature of ALS. Future work will require to determine when increased synaptic density alongside altered network excitability presents in ALS progression in patients.
Many ALS-focused studies describing altered glutamatergic input have examined synaptic function without assessing presynaptic function in detail, nor have they examined the consequences for network activity. An increase in excitatory synaptic input might be expected to increase network burst duration in addition to burst frequency [
32]. However, our assessment of network activity revealed a decrease in network burst duration and appears to be consistent with a decrease in glutamatergic synaptic transmission. Consistent with this, our evaluation of pre-synaptic function revealed an estimated reduced size and replenishment of vesicular RRP. Importantly, a reduced RRP size and replenishment rate has been previously shown to generate early burst termination to shorten burst duration [
10,
32]. This provides the most parsimonious explanation of the observed shorter network burst duration in
C9ORF72RE cortical neurons. A disruption in the vesicular RRP suggests mechanisms in which synaptic vesicular trafficking are impaired. Noting that C9ORF72 is detected in pre-synaptic terminals, our data resonate with previous studies highlighting the role of C9ORF72 protein in vesicular trafficking within the trans-Golgi network and endosomal signalling and suggest that
C9ORF72 haploinsufficiency may result in a reduced RRP [
2,
19,
53]. Our transcriptomic data provide further evidence of dysregulated genes associated with impaired vesicular trans-Golgi network and endosomal signalling in ALS, consistent with a growing body of evidence of impaired vesicular trafficking in ALS that may impact on the RRP size [
11,
15,
46]. For example, our data set highlights an upregulation of the
GOPC gene, a chaperone protein that is expressed across the trans-Golgi network and endosomes. Amongst many interactions, GOPC is associated with syntaxin-6 that regulates endosomal vesicular transport [
7]. Furthermore, TDP-43 protein appears to bind GOPC RNA [
41]. Collectively, our data show a reduction in the RRP that is consistent with impairments in vesicular trafficking.
Importantly, vesicular release is typically stimulated via Ca
2+-dependent mechanisms. Despite a reduced RRP, our evaluation of depolarisation/Ca
2+-dependent vesicular release appears to be equivalent in
C9ORF72RE excitatory cortical neurons versus isogenic controls. Indeed, our afterhyperpolarisation (AHP) data suggest that calcium mediated influx is not impacted to influence intrinsic excitability properties. However, we must remain cautious that AHP and exocytosis could be independently calcium regulated processes in our cells, subject to localised intracellular calcium regulation. One potential explanation could be that localised Ca
2+-dependent mechanisms controlling vesicular release in
C9ORF72RE excitatory cortical neurons are enhanced over control lines to generate the higher release probability required to elevate the fold increase in mEPSC frequency to comparable levels to the control lines. Dysregulated cytoplasmic Ca
2+ levels in
C9ORF72RE-derived motor neurons have been previously reported [
13,
28] and this elevation in Ca
2+ levels may contribute to an increased release probability. However, our findings contrast with those of Jensen et al. [
28] who suggest that KCl-stimulated release is impaired, due to a GA-driven loss of the protein SV2, in
C9ORF72RE patient-derived cortical neurons.
Our data show that synaptic potentiation in
C9ORF72RE excitatory cortical neurons is impaired. Notably, transcranical magnetic stimulation-based studies show both presymptomatic and post-symptomatic
C9ORF72RE patients exhibit an abolishment of activity-dependent cortical network plasticity [
5]. Together, these data suggest impairments in functional synaptic plasticity may emerge as an early pathophysiological event in
C9ORF72RE–mediated disease progression to impair network plasticity. The pathological determinants of the impairments in synaptic plasticity remain unknown in
C9ORF72RE cortical neurons, though a very recent study has shown synaptic plasticity impairments in murine C9orf72 knock out animals [
27], suggesting a role for the C9orf72 protein in plasticity mechanisms. In addition to our own data set indicating impact upon synaptic physiology, data sets highlight that altered gene expression in
C9ORF72RE patient tissue [
46] and disrupted cell signalling pathways in iPSC-derived
C9ORF72RE neurons [
13] are implicated in synaptic plasticity. Furthermore, recent transcriptomic work has associated the expression of di-peptide repeat proteins with a reduction in expression of a mediator of synaptic plasticity [
15]. Our data therefore firmly determines synaptic plasticity impairments are present in human
C9ORF72RE cortical neurons.
Crucially, C9ORF72
RE mutations are causal to both ALS and FTD. In this regard, we note that the vast majority of clinical pathophysiological measurements describing hyperexcitability are made from the motor cortex, which is primarily affected in ALS [
22]. Nonetheless cortical hypexcitability is evidenced in rodent models of FTD [
20]. We acknowledge that our data may have preferential relevance for FTD over ALS, or vice versa. This is likely to become more apparent with increased pathophysiological characterisation of
C9ORF72RE ALS-FTD patients. Further, early identification of pre-symptomatic individuals and longitudinal stratification of these observations will allow us to place further confidence upon the pathophysiological staging that our observations are most likely to mirror. Our data represent one in vitro time point, but as discussed, closely align with aspects of pre-symptomatic cortical neuron impairments evidenced in rodent models of ALS and ALS-FTD. Importantly, the network excitability is not investigated in these models. Our data suggest that increases in synaptic transmission may lead to increased network burst frequency, however at the same time, indicate that other concurrent processes reduce burst activity, reflective of multiple changes in the molecular landscape of synaptic function, potentially in an attempt to homeostatically compensate against other network impairments. Indeed, work performed more extensively in the context of Alzheimer’s disease has established that homeostatic network activity adaptation, including impaired synaptic plasticity, is an early feature and precedes that of pathophysiological network failure likely resulting in the manifestation of the clinically observed network hyperexcitability [
18]. In this regard, there are intriguing similarities in our data set to cortical neurons derived from Alzheimer’s patient iPSCs that display increased synaptic activity at the same time point [
23].
.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.