Cell senescence, loss of splicing, and lipid metabolism in TDP-43-related neurodegenerative processes

In recent work, we have shown that cell senescence of mouse fibroblasts in vitro associates with a build-up of cryptic exons in selected mRNAs, whose level is usually controlled by the activity of TAR DNA binding protein of 43 kDa (Tdp-43) (Torres et al., 2022). In vivo, we also found traits of cell senescence in the motor neuron disease model achieved by overexpressing SOD-G93A, the SOD1 gene (harboring a single amino acid substitution of glycine to alanine at codon 93). These mice express an age-related increase in the p21 and p16 mRNA levels, with enhanced protein levels of codified proteins in the cytosol of several cells present in the lumbar spinal cords of the model. Most cells showing increased p16 immunoreactivity were identified as astrocytes and microglia, with neuronal cells relatively spared from this senescence biomarker’s build-up. In addition to increased signs of replicative senescence (increased p16 and p21 expression), these mice also exhibit some characteristics related to senescence-associated secretory profile, such as increased levels of interleukin-6, and interleukin-1a, in a sex-specific manner. We qualify the cellular senescence in this model as atypical, as we were not able to significantly change the motor phenotype by treatment of anti-BCL2-BCLx antagonist, Navitoclax®, whose use has been beneficial in other preclinical models of age-related neurodegenerative diseases, such as Alzheimer’s disease. Of note, reinforcing the complexity of senolytical approaches, in vitro treatments of fibroblasts employing a quercetin-dasatinib® (but not Navitoclax®) combinations were able to abolish the senescence-associated build-up of p16 and p21 (Torres et al., 2022). These results suggest the importance of glial cell senescence in the association between aging and selective neuron demise. Disease-specific reactive astrocytosis secondary to neuron loss and astrocytopathy due to intrinsic alterations of astrocytes occur in neurodegenerative diseases, overlap each other, and, together with astrocyte senescence, contribute to disease-specific astrogliopathy in aging and neurodegenerative diseases with abnormal protein aggregates in old age, including amyotrophic lateral sclerosis (ALS). In addition to the well-known increase in glial fibrillary acidic protein and other proteins in reactive astrocytes, astrocytopathy is evidenced by the deposition of abnormal proteins and could contribute to neuronal loss, either by losing homeostatic support functions or by acting as sources of noxious agents towards neurons (Ferrer, 2017).

We found a correlation between p16 mRNA and tdp-43 dysfunction in the G93A mice. Also, the overexpression of functional TDP-43 in mice diminished the levels of mice-specific cryptic exons found in adiponectin receptor 2 (adipor2) mRNA, leading to diminished AdipoR2 levels (Torres et al., 2022), whose function is mainly related to membrane homeostasis. ADIPOR1 and ADIPOR2 are essential factors required for the maintenance of desaturase activity and adequate levels of unsaturated fatty acids in the membrane phospholipids, as well as for preventing membrane rigidification in cells exposed to exogenous cell stressors (Ruiz et al., 2019). In addition to ADIPOR2, recent data show that another factor, RNF145, an E3 ubiquitin ligase, is persistent in unsaturated lipid membranes, enabling ADIPOR2 ubiquitination and degradation through its lipid-sensitive interaction (Volkmar et al., 2022). In the presence of stimuli leading to the enhanced rigidification of membranes, such as enrichment of saturated fatty acids, RNF145 is quickly auto-ubiquitinated and eliminated, stabilizing ADIPOR2, whose hydrolase activity restores lipid homeostasis and avoiding lipotoxicity (Volkmar et al., 2022). The loss of AdipoR2 secondary to tdp-43 dysfunction in the G93A mice could explain part of the severe alteration in lipid composition seen in this model (Cacabelos et al., 2016). It is well known that TDP-43 is affected by similar ring finger proteins, such as PRAJA1, ZNF179, SCF^{cyclin F}, CUL2 or PARKIN. Whether RNF145 activity could affect TDP-43 is yet unknown. We should remember that the loss in AdipoR2 secondary to tdp-43 dysfunction might only be found in mice, as ADIPOR2 mRNA in humans does not exhibit known TDP-43 interaction. In addition to ADIPOR2, other lipid-relevant genes might be influenced by TDP-43. Mechanistically, TDP-43 binds with the UG-rich sequence1 of ABHD2 3′UTR to enhance the mRNA stability of ABHD2, an acyl hydrolase implicated in fatty acid remodeling of the membrane. Afterward, TDP-43 promotes the production of free fatty acid and fatty acid oxidation-originated reactive oxygen species in an ABHD2-dependent manner to suppress apoptosis in cell lines (Liu et al., 2022). Thus, TDP-43 dysfunction in cell stress may contribute to cell demise by losing ABHD2-dependent lipids interacting with oxidative stress.

The above-related facts may support cell senescence-inducing loss in TDP-43 function. It is unknown whether the inverse, i.e., loss of TDP-43 function, contributes to cell senescence. TDP-43 aggregates are present in several age-related neurodegenerative processes, and we hypothesize that the resulting loss might impact age-related changes in lipid homeostasis. Age causes profound changes in the lipidome of the central nervous system. Indeed, lipid metabolism is dysregulated in aging. There are several signs of senescent cells potentially caused by alterations in lipid metabolism, including increased cellular size, loss of nuclear membrane integrity, increased lysosome and mitochondrial mass, and components of the senescence-associated secretory profile that can be “transported” by exosomes. Since changes in the lipid component of membranes may be necessary for each of these “senescence traits,” including exosomes, and lipid-enriched extracellular vesicles, we could infer that lipids are intimately related to the process of cellular senescence in neuronal aging. In addition to cellular membranes, other lipid-rich cellular components, such as lipid droplets (LDs), can also be involved in cell senescence. LDs are implicated in energy production, the synthesis of membranes and signaling molecules, and the defense of cells against lipotoxicity and free radicals by controlling the storage of extra fatty acids, cholesterol, and ceramides as well as their subsequent release in response to cell needs and environmental stressors. Senescent cells have more LDs than growing cells. Increased lipid uptake and elevated lipid production pathways, such as increased diacylglycerol acyltransferase 1, or downregulated lipid degradation can all contribute to deregulated LD build-up in senescent cells. It is interesting to note that senescent cells absorb more lipids (Flor et al., 2017), though it is unclear whether inhibiting lipid uptake will stop or delay cellular senescence. Lipid processing and metabolism genes, including those involved in lipid binding, storage, biosynthesis, and breakdown, are significantly changed in senescence caused by DNA-damaging chemicals (Flor et al., 2017). In this line, it has been hypothesized that LDs accumulation could be implicated in motor neuron diseases. ALS is associated with the enrichment of LDs in nuclei-enriched fractions isolated from cortex samples of patients in comparison to age-matched individuals (Ramírez-Núñez et al., 2021). Of note, LDs may also be derived from the senescence-related loss of lipophagy. Regarding its tissue distribution, LDs are preferentially found in neuroglial cells, which comprise an array of diverse cells essential for the support of neurons. Glial senescence has been described in the central nervous system, which may contribute to the pathogenesis of diverse neurodegenerative conditions (reviewed in Ferrer (2017)). Previous data reported the existence of (glial) cell senescence in the G93A transgenic rat, in line with cell senescence in other models of age-related neurodegenerative pathologies. Altered autophagy has been implicated in the build-up of LDs enriched in a microglial population associated with cerebral aging (Marschallinger et al., 2020). These cells are incapable of phagocytosis, generate a high amount of reactive oxygen species, and release cytokines that promote inflammation. These traits are also present in senescent cells, reinforcing lipid metabolism’s role in senescence-associated cellular changes. In the G93A mice, a relevant percentage of the Iba1⁺ cells in the lumbar spinal cord also showed signs of senescence, such as increased p16 expression (Torres et al., 2022). In mice, the LD enriched microglia associated with brain aging, denominated lipid-laden cells (LLCs), coexpressing LC3 and beclin-1, components of autophagic vesicles, with LDs (Shimabukuro et al., 2016). This association suggests a potential loss of lipophagy in TDP-43 deficient cells due to the relationship between loss of TDP-43 and autophagy defects (Torres et al., 2018). Of note, LLC exhibits traits compatible with a senescence-associated secretory phenotype, such as pro-inflammatory cytokine TNF-α (Shimabukuro et al., 2016). Interestingly, some of the LLCs exhibit enhanced β galactosidase activity, a classical property of senescent cells. Gene ontology analyses of LLCs reveal increased values in two pathways related to lipid metabolism: fatty acid beta-oxidation (P < 0.001) and phagosome maturation (P < 0.00001) (Marschallinger et al., 2020), potentially related to autophagy. In the case of fatty acid beta-oxidation, recent data show that, in the context of TDP-43 proteinopathy, components of the carnitine shuttle are misexpressed, and genetic modification of the expression of CPT1 or CPT2, two essential parts of the carnitine shuttle, alleviates TDP-43-dependent locomotor dysfunction in a variant-dependent manner. Additionally, beta-hydroxybutyrate or medium-chain fatty acids improve locomotor activity, supporting the idea that overcoming the carnitine shuttle deficit is neuroprotective. These data show how the carnitine shuttle and lipid beta-oxidation may be involved in ALS. Regarding phagosome maturation, we have previously shown that the loss of TDP-43 is associated with the loss of autophagy related 4B cysteine peptidase (ATG4b), a key player in autophagic vesicles (Torres et al., 2018). Also, progranulin expression in microglia strongly prevents LDs accumulation (Marschallinger et al., 2020). Previous data showed that TDP-43 depletion from mouse adult brains affected nearly 600 mRNA, including fus and progranulin, with special involvement of transcripts coding proteins implicated in synaptic activity.

In addition, SREBP2, a master regulator of cholesterol homeostasis, and the downstream genes it controls were considerably changed in cell lines expressing high levels of TDP-43. Increased TDP-43 reduced SREBP2 transcriptional activity, which prevented the production of cholesterol. Cholesterol levels were markedly reduced in the CSF fluids of ALS patients and the spinal cords of TDP-43 overexpressing mice (Egawa et al., 2022). Indeed, oligodendrocytes from FTD patients with TDP-43 pathology have decreased HMGCR and HMGCS1 levels and co-aggregation of LDLR and TDP-43. These findings suggest that TDP-43 regulates cholesterol homeostasis in oligodendrocytes and that dysmetabolic disorders associated with TDP-43 proteinopathies are related to HMGCR (Ho et al., 2021). Interestingly, HMGCR is also one of the targets of RNF145, the E3 ubiquitin ligase related to ADIPOR2. Therefore, a key enzyme in lipid metabolism, HMGCR, is controlled both by TDP-43 and by the RNF145 E ubiquitin ligase, thereby contributing to the control of lipid metabolism. Globally, we propose that TDP-43 dysfunction is related to cell senescence with the implication of disturbances in lipid metabolism and autophagy (Figure 1).

Figure 1.

TDP-43 dysfunction, through modification of key genes in lipid metabolism and autophagy (HGMCR, AdipoR, and ATG4b), could contribute to cell senescence, leading to neuronal and glial cell stress.

This cellular stress can be characterized by lipid droplet build-up, with autophagy disturbances, all linked by mitochondrial demise. The physiological response to this change, characterized by astrogliosis with astrocytopathy, through disarranged cell proliferation, could contribute again to cell senescence, which would fuel TDP-43 dysfunction, thereby closing the vicious circle. AdipoR: Adiponectin receptor; ATG4b: autophagy related 4B cysteine peptidase; HGMCR: 3-hydroxy-3-methylglutaryl-CoA reductase; TDP-43: TAR DNA binding protein of 43 kDa. Created with Microsoft Office Power Point 2016.

Nonetheless, the following questions are open for research. Is TDP-43 function required for controlling lipid metabolism at other nodes, such as those dependent on mitochondria (i.e., fatty acid oxidation and phospholipid synthesis at the mitochondria-endoplasmic reticulum interfaces)? Is TDP-43 dysfunction relevant in cell senescence during brain aging? Might genetically encoded differences in these pathways contribute to the phenotypic diversity of motor neuron diseases? The answers to these questions should clarify the role of the changes in neuronal and glial lipid metabolism in the pathophysiology of aging and aging-related neurodegenerative diseases.

Grants were received from the Instituto de Salud Carlos III (PI 17-000134, PI 20-0155) to MPO, and from the Generalitat de Catalunya 2017SGR696 to RP. PT is a “Margarita Salas” fellow from the Spanish Ministry of Universities [Financed by European Union-NextGenerationEI funds]. Support was also received in the form of a FUNDELA Grant, RedELA-Plataforma Investigación and the Fundació Miquel Valls (Jack Van den Hoek donation) (to MPO). FEDER funds are acknowledged (“A way to make Europe”) (to MPO).