Transcriptomic Alteration in FUS-ALS Points Towards Apoptosis-Rather than Ferroptosis-Related Cell Death Pathway

Amyotrophic lateral sclerosis (ALS) is a progressive and irreversible neurodegenerative disease characterized by the selective loss of motor neurons (MNs) in the motor cortex, brain stem and spinal cord, leading to progressive muscle weakness and death from respiratory failure typically within 2–5 years of symptom onset. While the majority of ALS cases are sporadic (sALS), the remaining 5–10% of cases are familial (fALS) with mutations found in over 40 Mendelian inherited genes [1]. Among those, mutations in C9orf72, SOD1, FUS, TARDBP and TBK1 are the most prevalent in European populations and have the highest penetrance [2].These causative genes encode proteins with different functions, and the pathogenesis has been implicated with defects in stress response, mitochondrial dysfunction, hyperexcitability, impaired protein homeostasis, aberrant RNA metabolism, and impaired DNA repair [3,4]. In ALS, MN death is recognized to be both cell-autonomous, driven by intrinsic cellular defects or genetic mutations, and non-cell autonomous, caused by the external factors or detrimental signals from other cells (e.g., astrocytes, microglia) [2].

Despite extensive research studies having been performed, the etiology of ALS MN degeneration, including the primary cell death mechanisms responsible for MN death, remains poorly understood. This prevents the development of effective interventions and precise therapy stratification. Cell death is a vital biological process that contributes to development, homeostasis maintenance and disease prevention in multicellular organisms. Cell death is broadly classified into two groups: programmed/regulated cell death (energy-dependent) and non-programmed/necrotic cell death (energy-independent) [5,6]. Programmed cell death is defined by strictly regulated mechanisms and encompasses orchestrated signaling cascades operated at the molecular level and further divided into apoptotic or non-apoptotic programmed cell death. In contrast, non-programmed cell death represents an uncontrolled biological process and occurs under the influence of accidental cell damage [6,7]. MN death was initially implicated to be apoptotic, the most extensively studied form of cell death, and triggered by the intrinsic pathway (internal stress such as DNA damage, mitochondrial dysfunction leading to caspase activation), and the extrinsic pathway (activation of death receptors, tumor necrosis factor or p75 neurotrophin receptors) [8]. However, recent scientific investigations have reported an emerging repertoire of non-apoptotic modalities of MN cell death, leading to the implication of ferroptosis in the pathogenesis of various neurodegenerative diseases including ALS [9,10,11,12,13].

Ferroptosis, an iron- and lipid-peroxidation (LPO)-dependent form (and caspase-independent form) of non-apoptotic cell death [14], is typically induced by the excessive accumulation of iron and lipid reactive oxygen species (ROS) and/or inactivation of cellular antioxidant systems, contributing to neuronal damage. More recently, abnormality of iron homeostasis and anomalous accumulation of iron have been implicated in ALS patients (motor cortex, spinal cord and cerebral regions) [14,15,16,17,18] and animal models [19,20,21,22,23], respectively. Furthermore, a research study using induced pluripotent stem cell (iPSC)-derived MNs from sALS patients showed the involvement of lipid peroxidation and ferroptosis in MN cell death [24]. In addition, previous studies in ALS patients have demonstrated that cell death triggered by lipid peroxides (iron-dependent regulated necrosis) resulted in significant downregulation of endogenous mechanisms involved in protecting cells against ferroptosis, such as the glutathione peroxidase 4 anti-oxidant defense checkpoint (GSH/GPX4), and has been associated with degeneration of MNs and disease progression in ALS (sALS and fALS), suggesting a potential link between ALS and ferroptosis [11,16,25]. Consequently, the overexpression of GPX4 in transgenic SOD1 mice significantly mitigates symptoms and improved motor neuron function [23,26]. Additionally, recent evidence explored ferroptotic cell death in FUS-ALS, uncovering mitochondrial disturbances and heightened vulnerability to ferroptosis in FUS P525L MNs [18].

Although few research studies have been carried out on complex mechanisms of ALS and ferroptosis, their specific relationship on a more system biology level such as transcriptomics remains unclear. Since the FUS core function is to regulate transcription, we wanted to systematically investigate how much mutations in FUS affect ferroptosis-related gene (FRGs) expression. In the present study, using iPSC-derived MNs expressing FUS-ALS mutations, we performed a comprehensive bioinformatics analysis by systematically analyzing five RNA-sequencing (RNA-seq) datasets, including our dataset from the publicly available gene expression omnibus (GEO) database. We employed integrated differential expression gene (DEG) analysis to identify target genes and altered pathways related to ferroptosis in MNs between FUS mutant and healthy controls (WT), which contributes to further explore the pathomechanisms and selective MN vulnerability in FUS-ALS.

Although multiple mechanisms have been implicated in the pathogenesis of ALS, including glutamate excitotoxicity, mitochondrial dysfunction, oxidative stress, immune dysregulation and impaired axonal transport [44], underlying pathomechanisms causing MN degeneration and cell death pathways are poorly understood. Significant advancements have been made in microarray/RNA-sequencing analysis to identify gene expression profiles in different cells/tissues and contexts, thus helping to reveal important biological pathways under different conditions [28,45,46]. However, there have been few transcriptomic studies focused specifically on ferroptosis-associated genes and pathways in ALS.

For doing so, we compared our own and different available datasets to increase the number of biological replicates and generalizability. With that, we systematically explored the role of the ferroptosis-related gene signature in FUS-ALS. Briefly, we collected five comparable RNA-seq datasets from the GEO public database and performed integrated bioinformatic analysis to specifically look for mRNAs associated with ferroptosis and their related pathways that are differentially expressed in MNs from iPSC-derived FUS mutant-ALS patients and WT controls. A total of 672 common DEGs that are combined in five RNA-seq datasets were identified between FUS mutant and WT control MNs. The enrichment analysis revealed that DEGs were primarily misregulated in gene expression (transcription), organelle homeostasis, ER-protein folding and cell adhesion pathways. The accumulation of ROS is an important mechanism that can be the cause or consequence of mitochondrial dysfunction of MNs and can promote damage to the cell/organelle membrane, which leads to ferroptosis and other programmed cell deaths, such as apoptosis. Indeed, in GSEA analysis of RNA-seq data, we found that key genes involved in DNA repair, p53 signaling, apoptosis, antioxidant activity, cellular response to ROS and stress-activated protein kinase signaling cascade were mostly enriched with higher enrichment scores, whereas gene sets related to mitochondrial function, cellular metabolism and ubiquitin/proteasome-mediated protein catabolic process were downregulated in FUS mutant MNs. These results suggest that increased cellular stress, caused by an imbalance between ROS and antioxidant defense systems, may induce apoptosis, and that played an important role in the pathogenesis of FUS-ALS.

Several studies have reported that FUS plays a crucial role in various cellular processes such as proteostasis, DNA repair (nuclear and mitochondrial), mitochondrial functions and RNA splicing or RNA metabolic processes [47,48,49]. Abnormal expression of FUS has been reported to exacerbate the accumulation of DNA damage, and p53 is involved in activating DNA repair pathways. However, in case of prolonged DNA damage that is beyond repair, the p53 promotes neuronal apoptosis by increasing the transcription of pro-apoptotic genes [47,49,50,51], suggesting a crucial role of FUS in the DNA repair-induced apoptosis mechanism in ALS pathology. Besides that, in a GSEA analysis, the degree of ferroptosis and iron ion homeostasis/response to iron of FUS MNs was lower, indicating that the genes related to ferroptosis were transcribed at lower levels compared to a WT control group. This downregulation of ferroptosis genes (a gene set) in the gene expression dataset suggests a decrease in the activity of ferroptosis (or less prevalent) in the FUS MNs, and this depletion could be due to multiple factors, including altered cellular metabolism or signaling pathways, various types of external stimuli, increasing antioxidant defenses and reducing iron availability, and even from changes in gene expression regulation that are altered in response to the FUS mutation, which was clearly evident from our GO/GSEA enrichment analysis of DEGs.

To further investigate a putative transcriptional activation of ferroptosis, we intersected the above-mentioned DEGs with FRGs collected from FerrDb and GeneCards. By this, we identified 31 potential DEGs (16 upregulated and 15 downregulated) related to ferroptosis (DEFRGs) between FUS mutant and WT controls. To understand the role of these DEFRGs in FUS-ALS, we further carried out GO/pathway enrichment and PPI analysis. The results of this analysis showed that the regulation of multiple biological functions and pathways, including increased ROS levels, regulation of gene expression (RNA polymerase II-specific) and stress response, were altered, which was in accordance with previous studies on ALS [52,53,54]. Taken together, the potential biological functions of DEFRGs in FUS-ALS involved the regulation of various signaling pathways and processes. The identified pathways might be triggered by upstream pathways, rather than being the primary pathway affected (e.g., ferroptosis) activated by a unique stimulus or response to a specific stress signal. Notably, the GSEA analysis of gene expression data showed that the ferroptosis pathway, including response to cellular iron/iron homeostasis, was downregulated in FUS MNs. It should be noted, DEFRGs also pointed towards a downregulation of ferroptosis (upregulated suppressor of the ferroptosis; downregulated driver of the ferroptosis). We assume that the collective effect of misregulated suppressors and drivers of ferroptosis may induce the decreased expression of ferroptosis at an early time point, while this might be different at later time points, which needs future investigations.

While a considerable amount of work has focused upon the metabolic adaptations that are found in FUS MNs and which enable specific adaptations, including apoptosis, metabolites, iron homeostasis and the need for protein homeostasis, the precise vulnerabilities of neuronal populations arising from these adaptations or alterations under physiological conditions to cell death remain poorly understood. In this context, through our gene expression studies of FUS-ALS datasets, we provided preliminary clues about the underlying cell death pathways to uncover the sensitivities of MNs to ferroptosis. On one hand, our enrichment analysis has revealed that DEGs in FUS MNs were primarily enriched in apoptosis or apoptosis-mediated functions. On the other hand, ferroptosis (or DEFRGs) showed reduced expression levels without significant enrichment of classical ferroptosis markers (e.g., GPX4) or ferroptosis-related processes, which suggest that FUS MNs are less prone to ferroptosis, at least at the time point of analysis. Despite the differences between the two cell death pathways, accumulating evidence suggests that ferroptosis and apoptosis can be induced by similar stress signals (e.g., ER stress or oxidative stress) and shared common regulators, such as p53. They may occur in the same damaged cells either sequentially or simultaneously and can even through their combined actions induce cell death [55,56]. As an essential gene, p53 is required not only for transcription of the pro-apoptotic genes to various stress signals but also for the suppression of the anti-ferroptosis protein solute carrier family 7 member 11 (SLC7A11, a subunit of the cysteine–glutamate antiporter), which plays key role in cystine uptake and GSH metabolism [57]. Fitting to this, in GSEA analysis we identified upregulation of p53-signaling pathway and also found that the SLC7A11 gene was downregulated in the ferroptosis gene set. Regulation of ferroptosis by p53 is context-dependent (e.g., gene mutation and cell-type), can promote the presence of ferroptosis by inhibiting SLC7A11 transcription, and can also reduce cell sensitivity to ferroptosis by acting on transcription-independent mechanisms [58]. A recent study showed that in cancer cells p53 binds to and sequesters pro-ferroptotic enzyme dipeptidyl-peptidase-4 (DPP4) within the nucleus, forming an inactive complex, thus preventing its interaction with NOX1 (NADPH oxidase 1) and decreasing lipid peroxidation/ferroptosis [59]. In other contexts, p53 activation can lead to the expression of its downstream genes involving the cell cycle inhibitor CDKN1A/p21 pathway, which in turn increases intracellular GSH/GPX4 levels to prevent lipid peroxidation and reduce ferroptosis sensitivity [60]. Further, in tumor cells p53 also regulates ferroptosis sensitivity by upregulating the expression of a polyamine metabolism-related enzyme, spermidine/spermine N-acetyltransferase 1 (SAT1) [61]. However, the exact underlying mechanism of p53 mediating reduced ferroptosis sensitivity in the above pathways is unknown and needs to be further elucidated. Further, in our differential enrichment analysis, gene expression and the stress response pathways were found in DEFRGs, while DEGs associated with the DNA repair/p53 and with ubiquitin proteasomal response were connected to response to apoptosis and are consistent with the role of these specific pathways in each respective death program [16]. Although, the cellular consequences of ferroptosis regulation by p53 could be complex, cell type-specific, bidirectional (positive or negative regulation), and context-dependent with distinct mechanisms, p53 was proposed to play a critical regulatory role in the crosstalk between ferroptosis and apoptosis [56,58]. Thus, p53 might be involved in activating DNA repair pathways (e.g., upregulation of DNA damage response genes or during extensive DNA damage) and may either trigger ferroptosis to help remove dead/damaged MNs or reduce the sensitivity of MNs to ferroptosis and promote normal cell survival, especially during the mild stress or injury. Our study's main focus is to systematically investigate cell death activation pathways as well as cell adaptability in FUS mutant neurons to resist early forms of stress before MN loss is more obvious. Moreover, our own FUS-ALS datasets are from early time points where cells are viable before any visible signs or symptoms of cell damage occur [43,47,62]. Similarly, no early phase of cell stress or death has been observed in other FUS-ALS datasets reported in this study (Table 1), implying early time points are indeed crucial for capturing dynamic changes in signaling/survival cascades and gene expression events that forecast a cell's fate [39,40,41,42]. Additional studies are required to explore how the compensatory mechanisms at harvest time relate to the long-term progression of the disease, where initial adaptations might eventually fail to counteract the damage.

While our results may provide new insights into an early transcriptomic regulation of FRGs, identifying potential pathways by using the data generated by different laboratories, they possess certain limitations and warrant further consideration. Firstly, this study relied mainly on the GEO database, including analysis of previously published datasets. Thus, the selection of datasets, lack of relevant clinical data/severity of the patients and batch-to-batch variation may differ from the interpretations of previous experiments, most likely due to potential biases caused by the small sample size, and more datasets are needed to confirm our findings. Secondly, the FRGs are sourced from the manually updated website FerrDb, and more relevant genes remain to be explored. Finally, our results are based on bioinformatic analysis, without experimental validation of DEFRGs. Therefore, future experimental studies are required to verify the reliability and significance of the DEFRG results to explore the complex regulatory network of ferroptosis underlying FUS-ALS pathogenesis.

In the present study, we performed a comprehensive bioinformatic analysis of currently available RNAseq datasets of FUS-ALS human MN datasets. By this, we identified 31 ferroptosis-related DEGs and identified their participating gene functions and pathways. We also detected key hub genes that were closely associated with ferroptosis and were mainly involved in signal transduction pathways and cell–cell adhesion functions. However, results point more towards ferroptosis being not the main cell death pathway activated—at least on transcriptional level—at the time point investigated. This does fit to the recent failure of the CardinALS trial, which is a phase 2 study on a lipoxygenase inhibitor in ALS (NCT05349721). In contrast, DEGs and GSEA enrichment analysis of gene expression data in FUS MNs highlighted the importance of the apoptosis pathway, which would also fit previous reports of increased apoptosis rates in FUS-ALS MNs [62]. Further studies—including longitudinal ones—are needed to finally unravel the sequence of cell death forms mainly contributing to MN death in (FUS-) ALS. Finally, future studies are warranted to investigate cell death pathways on a protein level.