Neurodegenerative diseases (NDDs) are multifactorial disorders that are known to progress gradually with perceptible loss of neuronal function (Angelini et al., 2009). Although each disease has its own molecular pathological mechanism and clinical manifestations, they share a range of pathways that are thought to overlap and contribute to neuronal death (Hussain et al., 2018). The complexity of the NDDs makes it very challenging to pinpoint specific defects associated with the diseases’ development. Therefore, a lot of research efforts are still needed to reveal the genetic determinants of NDDs to improve their diagnosis and treatment. Even though recent studies conclude that mitochondrial dysfunction is a common overlapping feature between all neurodegenerative diseases, clear links between genes playing a role in mitochondrial function and NDDs are not well established (Huang et al., 2020).

The human mitochondrial proteome includes more than 1500 proteins, only 13 of which are mitochondrially encoded. The remaining mitochondrial proteins are nuclear encoded (Neupert, 2015; Wiedemann and Pfanner, 2017). In this study, we focused on Friedreich’s ataxia (FRDA), which is an autosomal recessive disorder and the most common inherited ataxia (Hart and Schapira, 2013). We compared the transcriptomic changes in nuclear-encoded mitochondrial (NEM) genes between FRDA and other seven NDDs in an attempt to identify common dysregulated genes and connect what is known about the diseases to FRDA. The diseases analyzed included Alzheimer’s disease (AD), amyotrophic lateral sclerosis (ALS), Creutzfeldt–Jakob disease (CJD), frontotemporal dementia (FTD), Huntington’s disease (HD), multiple sclerosis (MS), and Parkinson’s disease (PD). We discuss below some known links between the eight NDDs and mitochondrial dysfunction.

FRDA is triggered by an inadequate amount of an NEM protein known as frataxin, which has a vital role in iron metabolism and proper mitochondrial function (Llorens et al., 2019). About 90% of FRDA patients have an unstable GAA trinucleotide repeat in the first intron of both alleles of the FXN gene that encodes the frataxin protein (Campuzano et al., 1996). FRDA is a disease that cannot be treated effectively, which encouraged us to analyze the transcriptomic changes in NEM genes in FRDA patients and also other NDDs as a step towards identifying common genetic associations and potential drug targets.

We were keen to include diseases that are previously reported to be associated with mitochondria dysfunction. The dysfunction of mitochondria is considered to be an early feature of AD, as the Aβ protein is imported into the mitochondria via the TOM machinery (Picone et al., 2014). Aβ-induced mitochondrial dysfunction is associated with neuronal damage and cognitive decline (Swerdlow et al., 2010). In addition, it causes defects in key respiratory enzymes, accumulation of mitochondrial reactive oxygen species (ROS), and altered mitochondrial biogenesis (Moreira et al., 2010). As for ALS, even though the most common genetic cause of the disease is the expansion of the hexanucleotide intronic repeat GGGGCC in the C9orf72 gene (Ghasemi and Brown, 2018), mutations in the NEM gene SOD1 are considered to be the second most frequent cause of ALS (Bernard et al., 2020). Disruption of the mitochondrial structure and network are reported in the vast majority of ALS patients by multiple studies (reviewed in Smith et al., 2019). Moreover, these mitochondrial structural and functional impairments are reported in early ALS stages, suggesting that mitochondrial dysfunction could be an upstream cause of ALS degeneration rather than a consequence (Wang et al., 2013; Magrané et al., 2014).

Regarding FTD, the neurons are reported to exhibit a higher rate of mitochondrial ROS production and lipid peroxidation compared to controls, which leads to exacerbated damage due to oxidative stress. This ultimately leads to neuronal death by either apoptosis or necrosis (Bourens et al., 2013). Multiple studies note abnormalities in structure and function of mitochondria in FTD patients, but the exact mechanisms of mitochondrial dysfunction in FTD are yet to be elucidated.

In HD, several studies provide evidence for a link between the mutant Huntingtin (mHTT) protein and mitochondrial abnormalities that could ultimately lead to neuronal damage and degeneration in affected brain regions (Li et al., 2020). Mitochondrial dysfunction in HD may occur through aberrant transcriptional regulation as mHTT binds to several transcriptional regulators and interferes with their function (Dong and Cong, 2018). In particular, expression of PGC-1α, a master transcriptional co-regulator of mitochondrial biogenesis and antioxidant enzymes, is reduced in HD, which may contribute to mitochondrial impairment (Scarpulla, 2011; Johri et al., 2013).

Although the exact etiology of MS is still unknown, previous studies report increasing evidence that mitochondrial dysfunction is a common underlying theme of MS; ultimately leading to axonal degeneration (reviewed in Su et al., 2009). This mitochondrial dysfunction is manifested in dysregulation of Ca²⁺ homeostasis, ROS levels, and opening of the permeability transition pore. The characteristic chronic neuro-inflammatory stimuli in MS impair the neuro-axonal homeostasis and cause increased ROS-caused oxidative stress, which further damages the mitochondria and initiates a vicious cycle (Barcelos et al., 2019). In CJD, mitochondrial degeneration and electron-dense bodies of lysosomes within degenerating axons are observed (Liberski et al., 2005, 2010). Recent proteomics and gene expression studies suggest an altered gene expression pattern in proteins linked to oxidative phosphorylation (Ansoleaga et al., 2016). Mitochondrial dysfunction is also linked to PD diagnosis and progression (Gaweda-Walerych and Zekanowski, 2014).

In this study, we report transcriptional changes in multiple NEM genes for the eight NDDs of interest and highlight the common dysregulated genes and pathways with a focus on FRDA.

In this study, we aimed at linking transcriptional changes in NEM genes to FRDA and seven other NDDs. This was done through weighted gene co-expression network analysis (WGCNA) and differential expression analysis (DEA) of microarray and RNA-sequencing (RNA-seq). A detailed description of the workflow is represented in Figure 1. The data sets used for each disease and the number of patients and control samples are indicated in Table 1.

First, WGCNA was employed to identify the gene co-expression modules in FRDA. The analysis was performed on RNA-seq data of FRDA patients and normal samples (GSE104288) (Napierala et al., 2017). We analyzed only the NEM genes to link dysregulation of mitochondrial dysfunction to the disease. Around 1650 NEM genes retrieved from Eslamieh et al. (2017) were analyzed (full list provided in Supplementary Table S1). We also checked the confidence (conf.) score for mitochondrial localization for all proteins encoded by the genes (Supplementary Table S1) (Binder et al., 2014).

The WGCNA analysis identified seven gene co-expression modules (Figures 2A,B). We performed a module stability test using the same data set and confirmed that the modules identified in Figure 2A are persevered (Figure 2C) (Langfelder et al., 2011; Li et al., 2015). We further focused on four modules that had the highest correlation coefficients: the brown, turquoise, yellow, and red modules (Figure 2A). The full list of genes for each of the four modules is provided in Supplementary Table S2.

Mitochondria are pivotal in ROS regulation, apoptosis, energy generation, and metabolic pathways, including oxidation of fatty acids, steroid, and heme biosynthesis. Consequently, it is not surprising that mitochondrial function is compromised in many diseases, particularly those associated with high-energy-demand organs, such as brain, heart, and skeletal muscles (Sangar et al., 2012; Watson et al., 2020). Our main interest was to identify NEM genes that are dysregulated in FRDA, which is still incurable and is challenged by the lack of treatment that can at least delay neurodegeneration despite numerous research efforts (Ocana-Santero et al., 2021). Here, we compare the transcriptomic changes between healthy controls and eight NDDs, including FRDA. We then selected the NEM genes to focus the analysis on them. We additionally narrowed down the selection to include genes coding for proteins that are imported to the mitochondria with high confidence (≥3) (Binder et al., 2014). Finally, we checked the pathways enriched in FRDA and investigated their enrichment in the seven diseases. Besides focusing on the transcription levels of specific genes, employing a pathway-centric approach enabled us to tackle different players that contribute to the robust pathway function.

Our data show transcriptional changes in PYCARD and CASP1 that function in the NLPR3 inflammasome pathway in several NDDs. CASP1 codes for caspase-1 that is involved in the activation of interleukin 1β (IL-1β), a key inflammatory mediator (Denes et al., 2012). PYCARD is a multiprotein complex that induces pyroptosis and neuroinflammation (Broz and Dixit, 2016; Lang et al., 2018). Caspase-1 usually becomes proteolytically active only after its dimerization with inflammasomes, such as PYCARD (Codolo et al., 2013). We show the overexpression of PYCARD in FRDA-children, FTLD-TDP, HD, and PD. On the other hand, FTLD-TDP, PD, and HD showed significant upregulation in CASP1, unlike FRDA-adults, as it was downregulated. Our results align with recent studies that show that NLRP3 inflammasomes can trigger neuroinflammation through a wide range of stimuli (Yang et al., 2019). Inflammasomes are also activated by oxidative stress and excessively activated microglia, both of which contribute to the pathogenesis and progression of FTLD-TDP, HD, and PD (Lefkowitz and Lefkowitz, 2008; Hernandez et al., 2017; Carmo et al., 2018; Voet et al., 2019). Furthermore, studies performed on HD brain samples detected an elevation and excessive activation of caspase-1, and its inhibition was shown to slow down disease progression (Caron et al., 2018). This is also reported for other NDDs (Yang et al., 2019). Even though CASP1 was not overexpressed in the FRDA data set we analyzed, a previous study investigating frataxin-deficient lymphoblasts from FRDA patients shows its upregulation (Tan et al., 2003). However, we think we did not detect an overexpression in CASP1 in FRDA as its overexpression is usually associated with diseases linked to toxic aggregation accumulating in FTD, HD, and PD patients’ cells, which are not reported for FRDA (Wellington et al., 1998).

We interestingly found transcriptional changes in several genes involved in mitochondrial protein synthesis pathways, including translation initiation, elongation, and termination (Mai et al., 2017; Kummer and Ban, 2021). The small 28S and large 39S ribosomal subunits are coded for by the human mitochondrial ribosomal protein (MRP) gene family (Gopisetty and Thangarajan, 2016). Changes in the expression of MRPs destroy ribosomal composition, disrupt mitochondrial metabolic functions, and result in a multitude of mitochondrial diseases (Huang et al., 2020). MRP genes are also associated with neuronal differentiation and development (Gopisetty and Thangarajan, 2016). For example, MRPL2, MRPL14, MRPS10, MRPS18A, and MRPS18B are candidate genes for spinocerebellar ataxia with blindness and deafness (Papapetropoulos et al., 2006). Our results show that many of the MRPs are differentially expressed in MS and FRDA. In MS, MRPS35, MRPS30, MRPL1, and MRPL15 were downregulated, and MRPS28, MRPS35, and MRPL19 were downregulated in FRDA-adults. Interestingly, MRPL52 was upregulated in FRDA-adults. In FRDA-children, MRPL1 and MRPS10 were also downregulated. In agreement with this finding, MRPL family members were previously reported to be significantly underexpressed in FRDA fibroblasts (Napierala et al., 2017).

Changes in expression levels of genes functioning in mitochondrial protein import were also observed in several diseases. SLC25A12 and DNAJC19 were downregulated in MS. The latter was also downregulated in FRDA-children. TIMM8A on the contrary was upregulated in FRDA-adults in our analysis. Finally, CHCHD7 was also downregulated in MS. Mutations in DNAJC19 are previously associated with dilated cardiomyopathy in ataxia syndrome (Davey et al., 2006). Defects in TIMM8A are also linked to the deafness dystonia NDD (Koehler et al., 1999; Hoogenraad et al., 2002).

The enrichment analysis revealed several genes involved in the citric acid (TCA) cycle, respiratory electron transport, and Complex I Biogenesis to be differentially expressed in several NDDs. Alterations in glucose metabolism can affect the maintenance of neurotransmission and neuronal function and impact the ability to learn and memorize (Yan et al., 2020). It is also reported that the mitochondrial citric acid cycle can regulate the pathogenesis of neuroinflammation and neurodegeneration (Garabadu et al., 2019). The identified genes that play a role in one or more of these pathways included COX6C, UQCRB, and NDUFA5. COX6C was downregulated in FRDA-adults, and UQCRB was downregulated in FRDA-adults, FRDA-children, and MS. NDUFA5 was also downregulated in FRDA-adults, FRDA-children, and AD. NUBPL was, on the other hand, upregulated in FRDA-adults. Furthermore, NDUFA5 was reported to be differentially expressed in FRDA patients (Télot et al., 2018). UQCRB was also previously reported to be downregulated in FRDA (Télot et al., 2018). We did not detect changes in COX6C in AD; however, downregulation of the gene was previously linked to AD (Bi et al., 2018).

Among the pathways identified to be enriched in FRDA is “TP53 regulates metabolic genes.” Many recent studies show that p53 plays a role in neuropathogenesis and have found that increased levels of p53 are linked to neuronal cell death (Eun et al., 2010; Proctor and Gray, 2010). The DNA damage inducible transcript 4 “DDIT4” protein, also known as REDD1, regulates mTOR signaling (Cam et al., 2014). The DDIT4 gene was found to be upregulated in AD, CJZ, FTLD, HD, MS, and PD. DDIT4 was the highest elevated transcript (98-fold), and its encoded protein, RTP801, was considerably increased in cellular models of PD (Ryu et al., 2005). DDIT4 is also associated with AD and HD (Kim et al., 2003; Martín-Flores et al., 2020; Pérez-Sisqués et al., 2021).

Other DEGs playing a role in enriched pathways were identified. For example, SCP2 and AMACR functioning in peroxisomal protein import were downregulated in MS. CYBA functioning in detoxification of ROS was upregulated in HD, PD, and CJD. TRIT1, which modifies both cytosolic and mitochondrial tRNAs (Suzuki et al., 2011; Khalique et al., 2020), was also downregulated in FRDA-adults.

HSPA1A, which codes for heat shock protein 70 (HSP70), was upregulated in MS, FTD, HD, and PD. On the contrary, it was downregulated in FRDA-children. Expression levels of HSPs can change according to the response and function of the heat shock protein, the disease, the affected cell, and even the brain region (Leak, 2014). It is speculated that the rise in the expression of HSP70 in neurodegenerative conditions is to help slow the disease progression and delay aging as shown in PD, HD, and FTD (Lazarev et al., 2013; Calderwood and Murshid, 2017). In FRDA, on the other hand, the reason for the significant decline of HSP70 expression is because HSP70 is one of the essential components that facilitates the mitochondrial FeS cluster assembly, which is disrupted as a result of frataxin mutation (Stemmler et al., 2010). Finally, BCAT-1 was downregulated in MS. Previous reports show that knockdown of BCAT-1 demonstrate PD-like features, including progressive motor deficits and neurodegeneration developing with age (Mor et al., 2020).

ALS surprisingly showed very few similarities in the transcriptome profile of NEM genes when compared with the other NDDs. The total NEM DEGs in ALS were relatively fewer than the average of the other diseases. However, it still shares one NEM DEG with FRDA-adults, CJD, and MS, three with each of AD and HD as well as five NEM DEGs with each of FTD and PD. However, the genes were not part of the turquoise or brown modules and, therefore, were not focused on in the analysis. The full list of ALS DEGs is available in Supplementary Table S8.

The findings support an association of transcriptional changes in NEM genes with NDDs (Figure 7). The identified DEGs could possibly act as novel biomarkers, and the proteins for which they code can act as drug targets for the NDDs we discussed. Indeed, regulation of several players is already reported. For example, capase-1 is shown to be inhibited by VX-765 (Belnacasan). Belnacasan was able to stop the accumulation of amyloid β protein, prevent axonal neuronal degeneration, reduce release of inflammasome-associated cytokines, and ameliorate memory and cognitive function when tested on AD mice model (Flores et al., 2018). The same drug also inhibited proinflammatory caspase activation in MS mice models and reduced neurodegeneration (McKenzie et al., 2018). NDUFA5 is also shown to be upregulated by Pioglitazone (Coletta et al., 2009). Moreover, synthetic small molecules that specifically target UQCRB were also designed and showed an anti-angiogenic effect, causing repression of tumor growth in mouse xenograft models (Jung et al., 2014). Recent data also show that dexamethasone induces a novel epigenetic function for HDAC4, which involves switching on DDIT4 expression in ataxia telangiectasia and, consequently, increasing the protein levels (Ricci et al., 2020). Overall, our findings open doors for investigating the drugability of the players and further testing of known inhibitors on the NDDs discussed.

Despite the complex nature of NDDs, continuous research efforts are directed toward finding a cure to halt the escalation of symptoms and provide a definitive treatment. In this study, we investigate the transcriptomic changes in NEM genes in FRDA and seven other NDDs. Our results provide an expanded overview showing association of transcriptional dysregulation in NEM genes with all NDDs analyzed. In addition, pathways crucial for proper mitochondrial function were also associated with the DEGs. The NEM genes pinpointed in this study can serve as potential markers and drug targets, which can be further investigated in animal models and also in individual patients to confirm their dysregulation in a personalized manner.