Nucleoredoxin-like 2 Metabolic Signaling Impairs Its Potential Contribution to Neurodegenerative Diseases

Alzheimer’s disease (AD) threatens the foundations of humanity and our society. AD is a neurological disorder primarily affecting the elderly through memory disorders, cognitive decline, and loss of autonomy. The dramatic consequences of this late-onset disease were illustrated sensitively in Michael Haneke’s masterpiece, Amour. Researchers and governments have invested colossal efforts to develop a treatment for this terrible disease. Currently and in the past decade the amyloid cascade has dominated the therapeutic research on AD, but the absence of benefit for patients treated with drugs that reduce brain amyloid deposit questions the role of β-amyloid as a causative agent (Herrup, 2021). The recent approval of aducanumab by the United States Food and Drug Administration, a drug that targets β-amyloid, is at the center of a scandal among clinicians and researchers, as it does not provide therapeutic benefits to patients and is listed among the breakdowns of the year 2021 by Science magazine (Voosen et al., 2021). The lack of progress in curing AD and recent therapeutic failures calls for further exploratory research. While there are competing theories on disease causation, one constant is aging. Nevertheless, the presence of the anatomical biomarkers of the sporadic form of the Alzheimer’s disease, the late-onset AD, the neurofibrillary tangles made of TAU protein, as well as β-amyloid plaques originally described by Alois Alzheimer in healthy aged brain, shows that aging is only a biological clue. The oxidative stress theory of aging postulates that age-associated physiologic dysfunctions arose from a slow steady accumulation of oxidative damage to macromolecules. This is most relevant to the central nervous system, which consumes 20% of the body’s oxygen and is highly vulnerable to oxidative stress. The appearance of such damages could theoretically increase with age because of a reduction of the activity of repairing systems. Most free radicals are generated as byproducts of the mitochondrial electron transport chain which hypothesis that mitochondria play a critical role in aging and associated neurodegenerative diseases. Most organs, such as the brain, have two common NADPH-dependent redox systems powered by glucose through the pentose phosphate pathway: the glutathione system and the thioredoxin system (Ren et al., 2017). We have shown in the retina that the thioredoxin-like protein encoded by the nucleoredoxin-like 1 (NXNL1) gene binds and reduces oxidized cysteines of the microtubule-binding domain of TAU (Fridlich et al., 2009; Cronin et al., 2010). The Nxnl1 gene that is specifically expressed by photoreceptors prevents TAU aggregation, and since its paralogue Nxnl2 is also expressed in other parts of the brain, RdCVF2L could prevent the formation of neurofibrillary tangles found in the brain of AD and other tauopathies such as frontotemporal dementia. In the retina rod photoreceptors produce by alternative splicing, and secrete the other product of the NXNL1 gene, the rod-derived cone viability factor (RdCVF), a truncated thioredoxin that upon binding to its receptor, Basigin-1 at the surface of the cones, increases the update of glucose via GLUT1 (Aït-Ali et al., 2015). The metabolism of glucose by aerobic glycolysis provides the hydrophilic head of the phospholipids used for the daily renewal of the cone outer segments that carry the light-sensitive opsin molecule (Camacho et al., 2019). The combination of the two products of the NXNL1 gene, RdCVF and RdCVFL is a therapy developed for inherited retinal degenerations. Given that the redox power of RdCVFL, the thioredoxin encoded by the NXNL1 gene is powered by glucose, RdCVF and RdCVFL was conceptualized as a metabolic and redox signaling between rods and cones (Leveillard and Aït-Ali, 2017; Leveillard and Sahel, 2017). The combination of the two products of the NXNL1 gene is under clinical development to treat inherited retinal degenerations (Clerin et al., 2020). In the retina, we have found that NXNL1, and its paralogue NXNL2 which also encodes for a trophic factor RdCVF2 and a thioredoxin-like protein RdCVF2L has partially redundant functions (Jaillard et al., 2012). The main difference between these two genes is that NXNL2 expression is not restricted to the retina. There is growing evidence for a close link between altered glucose metabolism and Alzheimer’s disease pathogenesis. Recently, we revealed that the metabolic signaling of the NXNL2 gene supports brain function (Jaillard et al., 2021). In the mouse, the inactivation of the Nxnl2 gene triggers a complex syndrome in which fear, pain sensitivity, coordination, learning, and memory are deficient as early as 2 months of age. These deficits of brain functions are correlated to the loss of expression of both products of the NXNL2 gene, the trophic factor RdCVF2 and the thioredoxin-like protein RdCVF2L by a cell subset of the area postrema. The area postrema lacks a normal blood-brain barrier allowing the neurons to be exposed to blood-borne agents, such as satiety hormones. By its juxtaposition of the 4th ventricle, a signal generated in the area postrema could theoretically circulate in the cerebrospinal fluid to reach the brain areas, such as the hippocampus. Since Nxnl2–/– syndrome encompasses deficits in hippocampal-dependent learning and memory, we measured ex vivo field excitatory postsynaptic potentials in the CA1 region while stimulating Schaffer collateral originating from CA3 to record long term potentiation (LTP), one the biological mechanisms supporting learning of goal-directed spatial task. For two distinct genetic forms of Nxnl2 inactivation, constructed on distinct genetic background, these electrophysiological recordings are reduced showing that Nxnl2 is essential for hippocampal plasticity and spatial memory in the mouse. Since the absence of antibody did not allow us to demonstrate the presence of the trophic factor RdCVF2 in the cerebrospinal fluid, we tested the consequence of the absence of this NXNL2 gene product on the hippocampal pyramidal neurons glycolysis. A metabolomic analysis of standardized specimens of the hippocampus reveals that glycolysis is abnormal for the Nxnl2–/– mice at 2-month in support of the memory and LTP deficits, suggesting Nxnl2 supports metabolic signaling in the brain. Nevertheless, since pyramidal neurons of the hippocampus that generate LTP represent a subset of the overall hippocampal cell population, and we measured the steady-state concentration of glycolytic metabolites and not their fluxes, we could not ascertain that pyramidal neurons themselves have a deficit in glycolysis. Nevertheless, we correlated our findings with a reduction in the hippocampus specimens of the expression of GLUT4, the glucose transporter whose activity is regulated by insulin signaling. Metabolic impairments are not specific to AD since they were observed in other neurodegenerative diseases as Parkinson’s disease, Huntington’s disease, frontotemporal dementia, and amyotrophic lateral sclerosis (Cunnane et al., 2020). Genome-wide association studies have identified genetic susceptibility loci for sporadic AD, so that the disease cannot simply result from an increase of exposure to oxidative agents and has a genetic component (Lambert et al., 2009). The highest statistical significance susceptibility locus for late-onset AD is the gene encoding apolipoprotein E (APOE). The risk allele, APOEε4 is genetically associated with neuronal insulin deficiency, altered lipid metabolism, and decreased glucose uptake. Taking in consideration that a reduction in glucose metabolism, due to the lack of RdCVF2, would also trigger a reduction of the redox power of the thioredoxin-like enzyme RdCVF2L, we examined post-translational and structural status of TAU in the whole brain of the Nxnl2–/– mice. Interestingly, the phosphorylation of TAU using two phospho-TAU antibodies (AT8 and AT100) that characterize neurofibrillary tangle in Alzheimer’s brain was unaffected at 2 months but increased in mouse brain at 18 months, as if aging was an additional parameter acting to exacerbate metabolic deficit to lead to the observed tauopathy. In the aged brains of the Nxnl2–/– mice, we also observed an increase of TAU oligomerization and aggregation, the latter representing the fundamental step in the formation of neurofibrillary tangle. Altogether, these observations can be recapitulated in a working model (Figure 1). In non-pathological conditions blood-borne satiety hormones and glucose, whose concentration increases after feeding, transfer directly to a subsets of neurons of the area postrema. Somehow, this signal triggers the production and the release of RdCVF2 in the cerebrospinal fluid, resulting in its circulation toward the pyramidal cells of the hippocampus. By binding to its receptor, still unidentified, at the surface of pyramidal neurons, RdCVF2 stimulates glucose uptake, which results in increased glycolysis used to sustain the extension of lipid membrane surface at the dendritic spines during memory acquisition or consolidation. In pathological conditions, such as those created by removing the Nxnl2 gene in the mouse, the shortage of glucose in pyramidal neurons reduces the repair activity of the two NADPH-dependent redox systems, and probably that of RdCVF2L. The metabolic plasticity of pyramidal neurons allows them to deal with this redox power dysfunction for some time, but aging adds to the burden, which results in a tauopathy. What is quite interesting is the parallel between the progression in two stages of the phenotype of the Nxnl2–/– mouse and mild cognitive impairment that predispose later to dementia in human. The NXNL2 gene does not carry susceptibility alleles for AD so far, much as the MAPT gene encoding TAU, suggesting that while a genetic association in genome-wide association studies is informative, its absence cannot be a biological valid rejection criterion.Figure 1: Hypothetical working model.Under non pathological conditions, a satiety hormone is triggering the release of rod-derived cone viability factor 2 (RdCVF2) from the area postrema to the 4th ventricle. RdCVF2 circulates in the cerebrospinal fluid and reach its cell surface receptor on hippocampal pyramidal neurons increasing glucose uptake via GLUT4. Aerobic glycolysis participates in the membrane surface increase to form new dendritic spines. Metabolism of this glucose by the pentose phosphate pathway (PPP) increases the redox power of thioredoxin, such as RdCVF2L that reduces TAU aggregation resulting from increase oxidative stress during aging.Since the products of the NXNL2 gene have, by analogy, the same therapeutic potential that those of the NXNL1 gene (Clerin et al., 2020), we delivered to newborn Nxnl2–/– mice, RdCVF2 and RdCVF2L using an adeno-associated viral vector to correct the phenotype. The recombinant adeno-associated viral vector was injected in the heart and is then transferred to the brain in the young animals as they have not yet formed a functional blood-brain barrier. We found that both RdCVF2 and RdCVF2L partially prevents the LTP deficit, and interestingly, the combination of the two transgenes completely erases this deficit in a synergistic manner. This demonstrates that the synergistic action of RdCVF2 on metabolism and RdCVF2L on redox power is coupled to the benefit of the function of hippocampal neurons. This system is similar to our findings in the retina between the actions of RdCVF and RdCVFL and we propose that glucose is the link between the two splice forms of the gene. The failure of all clinical trials carried out to date on AD with devastating consequences, leads the scientific community to look at new avenues. Glucose, being the major energy source for neurons, the regulation of its metabolism is central in this reevaluation. The biological activity of the two products of the NXNL2 gene merits a special interest toward this goal. We thank John Han (Thomas Jefferson University, Phadelphia, USA) for his help. This work was funded by the Agence Nationale pour la Recherche, Fondation Vaincre Alzheimer and Sorbonne Université. Availability of data and materials:All data generated or analyzed during this study are included in this published article and its supplementary information files. Open peer reviewer:Silvia Zampar, Georg-August-Universitat Gottingen Universitatsmedizin, Germany. Additional file:Open peer review report 1.P-Reviewer: Zampar S; C-Editors: Zhao M, Liu WJ, Wang Lu; T-Editor: Jia Y