Paula Isabel Moreira

We propose five recommendations to make AI-based research studies more suitable for a clinical readership. First, authors should justify the added value of complex and potentially more opaque AI approaches. Second, rigorous description of input data, diagnostic criteria, and preprocessing is essential to avoid biased or clinically irrelevant outcomes. Third, benchmarking against clinically relevant performance thresholds should be established a priori. Fourth, method sections should combine an accessible lay summary with detailed technical supplement. Fifth, model explainability is encouraged to mitigate opacity. These recommendations aim to support AI research that is methodologically robust and interpretable for AD researchers.

Alzheimer’s disease is the most prevalent neurodegenerative disorder and affects the lives not only of those who are diagnosed but also of their caregivers. Despite the enormous social, economic and political burden, AD remains a disease without an effective treatment and with several failed attempts to modify the disease course. The fact that AD clinical diagnosis is most often performed at a stage at which the underlying pathological events are in an advanced and conceivably irremediable state strongly hampers treatment attempts. This raises the awareness of the need to identify and characterize the early brain changes in AD, in order to identify possible novel therapeutic targets to circumvent AD’s cascade of events. One of the most auspicious targets is mitochondria, powerful organelles found in nearly all cells of the body. A vast body of literature has shown that mitochondria from AD patients and model organisms of the disease differ from their non-AD counterparts. In view of this evidence, preserving and/or restoring mitochondria’s health and function can represent the primary means to achieve advances to tackle AD. In this review, we will briefly assess and summarize the previous and latest evidence of mitochondria dysfunction in AD. A particular focus will be given to the recent updates and advances in the strategy options aimed to target faulty mitochondria in AD.

Mitochondria interact with the endoplasmic reticulum (ER) through contacts called mitochondria-associated membranes (MAMs), which control several processes, such as the ER stress response, mitochondrial and ER dynamics, inflammation, apoptosis, and autophagy. MAMs represent an important platform for transport of non-vesicular phospholipids and cholesterol. Therefore, this region is highly enriched in proteins involved in lipid metabolism, including the enzymes that catalyze esterification of cholesterol into cholesteryl esters (CE) and synthesis of triacylglycerols (TAG) from fatty acids (FAs), which are then stored in lipid droplets (LDs). LDs, through contact with other organelles, prevent the toxic consequences of accumulation of unesterified (free) lipids, including lipotoxicity and oxidative stress, and serve as lipid reservoirs that can be used under multiple metabolic and physiological conditions. The LDs break down by autophagy releases of stored lipids for energy production and synthesis of membrane components and other macromolecules. Pathological lipid deposition and autophagy disruption have both been reported to occur in several neurodegenerative diseases, supporting that lipid metabolism alterations are major players in neurodegeneration. In this review, we discuss the current understanding of MAMs structure and function, focusing on their roles in lipid metabolism and the importance of autophagy in LDs metabolism, as well as the changes that occur in neurogenerative diseases.

Diabetes has been associated with an increased risk of cognitive decline and dementia. However, the mechanisms underlying this association remain unclear and no effective therapeutic interventions exist. Accumulating evidence demonstrates that mitochondrial defects are a key feature of diabetes contributing to neurodegenerative events. It has also been demonstrated that the putative tumor suppressor WW domain-containing oxidoreductase 1 (WWOX) can interact with mitochondria in several pathological conditions. However, its role in diabetes-associated neurodegeneration remains unknown. So, this study aimed to evaluate the role of WWOX activation in high glucose-induced neuronal damage and death. Our experiments were mainly performed in differentiated SH-SY5Y neuroblastoma cells exposed to high glucose and treated (or not) with Zfra1-31, the specific inhibitor of WWOX. Several parameters were analyzed namely cell viability, WWOX activation (tyrosine 33 residue phosphorylation), mitochondrial function, reactive oxygen species (ROS) production, biogenesis, and dynamics, autophagy and oxidative stress/damage. The levels of the neurotoxic proteins amyloid β (Aβ) and phosphorylated Tau (pTau) and of synaptic integrity markers were also evaluated. We observed that high glucose increased the levels of activated WWOX. Interestingly, brain cortical and hippocampal homogenates from young (6-month old) diabetic GK rats showed increased levels of activated WWOX compared to older GK rats (12-month old) suggesting that WWOX plays an early role in the diabetic brain. In neuronal cells, high glucose impaired mitochondrial respiration, dynamics and biogenesis, increased mitochondrial ROS production and decreased mitochondrial membrane potential and ATP production. More, high glucose augmented oxidative stress/damage and the levels of Aβ and pTau proteins and affected autophagy, contributing to the loss of synaptic integrity and cell death. Of note, the activation of WWOX preceded mitochondrial dysfunction and cell death. Importantly, the inhibition of WWOX with Zfra1-31 reversed, totally or partially, the alterations promoted by high glucose. Altogether our observations demonstrate that under high glucose conditions WWOX activation contributes to mitochondrial anomalies and neuronal damage and death, which suggests that WWOX is a potential therapeutic target for early interventions. Our findings also support the efficacy of Zfra1-31 in treating hyperglycemia/diabetes-associated neurodegeneration.

Diabetes mellitus type 2 (T2DM) has been associated with alterations in the male reproductive tract, especially in the epididymis. Although it is known that T2DM alters epididymal physiology, disturbing mitochondrial function and favoring oxidative stress, the mechanisms remain unknown. Sirtuin 1 (SIRT1), peroxisome proliferators-activated receptor γ coactivator 1α (PGC-1α), and sirtuin 3 (SIRT3) are key regulators of mitochondrial function and inducers of antioxidant defenses. In this study, we hypothesized that the epididymal SIRT1/PGC-1α/SIRT3 axis mediates T2DM-induced epididymis dysfunction by controlling the oxidative profile. Using 7 Goto-Kakizaki (GK) rats (a non-obese model that spontaneously develops T2DM early in life), and 7 age-matched Wistar control rats, we evaluated the protein levels of SIRT1, PGC-1α, and SIRT3, as well as the expression of mitochondrial respiratory complexes. The activities of epididymal glutathione peroxidase (GPx), glutathione reductase (GR), superoxide dismutase (SOD), and catalase (CAT) were determined, as well as the epididymal antioxidant capacity. We also evaluated protein nitration, carbonylation, and lipid peroxidation in the epididymis. The T2DM rats presented with hyperglycemia and glucose intolerance. Epididymal levels of SIRT1, PGC-1α, and SIRT3 were decreased, as well as the expression of the mitochondrial complexes II, III, and V, in the T2DM rats. We found a significant decrease in the activities of SOD, CAT, and GPx, consistent with the lower antioxidant capacity and higher protein nitration and lipid peroxidation detected in the epididymis of the T2DM rats. In sum, T2DM disrupted the epididymal SIRT1/PGC-1α/SIRT3 pathway, which is associated with a compromised mitochondrial function. This resulted in a decline of the antioxidant defenses and an increased oxidative damage in that tissue, which may be responsible for the impaired male reproductive function observed in diabetic men.

The retina, as part of the central nervous system (CNS), can be the perfect target for in vivo, in situ, and noninvasive neuropathology diagnosis and assessment of therapeutic efficacy. It has long been established that several age-related brain changes are more pronounced in Alzheimer's disease (AD). Nevertheless, in the retina such link is still under-explored. This study investigates the differences in the aging of the CNS through the retina of 3× Tg-AD and wild-type mice. A dedicated optical coherence tomograph imaged mice's retinas for 16 months. Two neural networks were developed to model independently each group's ages and were then applied to an independent set containing images from both groups. Our analysis shows a mean absolute error of 0.875±1.1 × 10⁻² and 1.112±1.4 × 10⁻² months, depending on training group. Our deep learning approach appears to be a reliable retinal OCT aging marker. We show that retina aging is distinct in the two classes: the presence of the three mutated human genes in the mouse genome has an impact on the aging of the retina. For mice over 4 months-old, transgenic mice consistently present a negative retina age-gap when compared to wild-type mice, regardless of training set. This appears to contradict AD observations in the brain. However, the ‘black-box” nature of deep-learning implies that one cannot infer reasoning. We can only speculate that some healthy age-dependent neural adaptations may be altered in transgenic animals.

Endoplasmic reticulum (ER) stress and mitochondrial dysfunction, which are key events in the initiation and/or progression of several diseases, are correlated with alterations at ER-mitochondria contact sites, the so-called "Mitochondria-Associated Membranes" (MAMs). These intracellular structures are also implicated in NLRP3 inflammasome activation which is an important driver of sterile inflammation, however, the underlying molecular basis remains unclear. This work aimed to investigate the role of ER-mitochondria communication during ER stress-induced NLRP3 inflammasome activation in both peripheral and central innate immune systems, by using THP-1 human monocytes and BV2 microglia cells, respectively, as in vitro models. Markers of ER stress, mitochondrial dynamics and mass, as well as NLRP3 inflammasome activation were evaluated by Western Blot, IL-1β secretion was measured by ELISA, and ER-mitochondria contacts were quantified by transmission electron microscopy. Mitochondrial Ca²⁺ uptake and polarization were analyzed with fluorescent probes, and measurement of aconitase and SOD2 activities monitored mitochondrial ROS accumulation. ER stress was demonstrated to activate the NLRP3 inflammasome in both peripheral and central immune cells. Studies in monocytes indicate that ER stress-induced NLRP3 inflammasome activation occurs by a Ca²⁺-dependent and ROS-independent mechanism, which is coupled with upregulation of MAMs-resident chaperones, closer ER-mitochondria contacts, as well as mitochondrial depolarization and impaired dynamics. Moreover, enhanced ER stress-induced NLRP3 inflammasome activation in the immune system was found associated with pathological conditions since it was observed in monocytes derived from bipolar disorder (BD) patients, supporting a pro-inflammatory status in BD. In conclusion, by demonstrating that ER-mitochondria communication plays a key role in the response of the innate immune cells to ER stress, this work contributes to elucidate the molecular mechanisms underlying NLRP3 inflammasome activation under stress conditions, and to disclose novel potential therapeutic targets for diseases associated with sterile inflammation.