Changes in the spinal cord proteome of an amyotrophic lateral sclerosis murine model determined by differential in-gel electrophoresis

What Can Machine Learning Approaches in Genomics Tell Us about the Molecular Basis of Amyotrophic Lateral Sclerosis?

Amyotrophic Lateral Sclerosis (ALS) is the most common late-onset motor neuron disorder, but our current knowledge of the molecular mechanisms and pathways underlying this disease remain elusive. This review (1) systematically identifies machine learning studies aimed at the understanding of the genetic architecture of ALS, (2) outlines the main challenges faced and compares the different approaches that have been used to confront them, and (3) compares the experimental designs and results produced by those approaches and describes their reproducibility in terms of biological results and the performances of the machine learning models. The majority of the collected studies incorporated prior knowledge of ALS into their feature selection approaches, and trained their machine learning models using genomic data combined with other types of mined knowledge including functional associations, protein-protein interactions, disease/tissue-specific information, epigenetic data, and known ALS phenotype-genotype associations. The importance of incorporating gene-gene interactions and cis-regulatory elements into the experimental design of future ALS machine learning studies is highlighted. Lastly, it is suggested that future advances in the genomic and machine learning fields will bring about a better understanding of ALS genetic architecture, and enable improved personalized approaches to this and other devastating and complex diseases.

Immiscible inclusion bodies formed by polyglutamine and poly(glycine-alanine) are enriched with distinct proteomes but converge in proteins that are risk factors for disease and involved in protein degradation

Poly(glycine-alanine) (polyGA) is one of the polydipeptides expressed in Frontotemporal Dementia and/or Amyotrophic Lateral Sclerosis 1 caused by C9ORF72 mutations and accumulates as inclusion bodies in the brain of patients. Superficially these inclusions are similar to those formed by polyglutamine (polyQ)-expanded Huntingtin exon 1 (Httex1) in Huntington’s disease. Both have been reported to form an amyloid-like structure suggesting they might aggregate via similar mechanisms and therefore recruit the same repertoire of endogenous proteins. When co-expressed in the same cell, polyGA₁₀₁ and Httex1(Q₉₇) inclusions adopted immiscible phases suggesting different endogenous proteins would be enriched. Proteomic analyses identified 822 proteins in the inclusions. Only 7 were specific to polyGA and 4 specific to Httex1(Q₉₇). Quantitation demonstrated distinct enrichment patterns for the proteins not specific to each inclusion type (up to ~8-fold normalized to total mass). The proteasome, microtubules, TriC chaperones, and translational machinery were enriched in polyGA aggregates, whereas Dnaj chaperones, nuclear envelope and RNA splicing proteins were enriched in Httex1(Q₉₇) aggregates. Both structures revealed a collection of folding and degradation machinery including proteins in the Httex1(Q₉₇) aggregates that are risk factors for other neurodegenerative diseases involving protein aggregation when mutated, which suggests a convergence point in the pathomechanisms of these diseases.

Mantis-ml: Disease-Agnostic Gene Prioritization from High-Throughput Genomic Screens by Stochastic Semi-supervised Learning

Access to large-scale genomics datasets has increased the utility of hypothesis-free genome-wide analyses. However, gene signals are often insufficiently powered to reach experiment-wide significance, triggering a process of laborious triaging of genomic-association-study results. We introduce mantis-ml, a multi-dimensional, multi-step machine-learning framework that allows objective assessment of the biological relevance of genes to disease studies. Mantis-ml is an automated machine-learning framework that follows a multi-model approach of stochastic semi-supervised learning to rank disease-associated genes through iterative learning sessions on random balanced datasets across the protein-coding exome. When applied to a range of human diseases, including chronic kidney disease (CKD), epilepsy, and amyotrophic lateral sclerosis (ALS), mantis-ml achieved an average area under curve (AUC) prediction performance of 0.81-0.89. Critically, to prove its value as a tool that can be used to interpret exome-wide association studies, we overlapped mantis-ml predictions with data from published cohort-level association studies. We found a statistically significant enrichment of high mantis-ml predictions among the highest-ranked genes from hypothesis-free cohort-level statistics, indicating a substantial improvement over the performance of current state-of-the-art methods and pointing to the capture of true prioritization signals for disease-associated genes. Finally, we introduce a generic mantis-ml score (GMS) trained with over 1,200 features as a generic-disease-likelihood estimator, outperforming published gene-level scores. In addition to our tool, we provide a gene prioritization atlas that includes mantis-ml's predictions across ten disease areas and empowers researchers to interactively navigate through the gene-triaging framework. Mantis-ml is an intuitive tool that supports the objective triaging of large-scale genomic discovery studies and enhances our understanding of complex genotype-phenotype associations.

Proteomics Approaches for Biomarker and Drug Target Discovery in ALS and FTD

Neurodegenerative disorders such as amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are increasing in prevalence but lack targeted therapeutics. Although the pathological mechanisms behind these diseases remain unclear, both ALS and FTD are characterized pathologically by aberrant protein aggregation and inclusion formation within neurons, which correlates with neurodegeneration. Notably, aggregation of several key proteins, including TAR DNA binding protein of 43 kDa (TDP-43), superoxide dismutase 1 (SOD1), and tau, have been implicated in these diseases. Proteomics methods are being increasingly applied to better understand disease-related mechanisms and to identify biomarkers of disease, using model systems as well as human samples. Proteomics-based approaches offer unbiased, high-throughput, and quantitative results with numerous applications for investigating proteins of interest. Here, we review recent advances in the understanding of ALS and FTD pathophysiology obtained using proteomics approaches, and we assess technical and experimental limitations. We compare findings from various mass spectrometry (MS) approaches including quantitative proteomics methods such as stable isotope labeling by amino acids in cell culture (SILAC) and tandem mass tagging (TMT) to approaches such as label-free quantitation (LFQ) and sequential windowed acquisition of all theoretical fragment ion mass spectra (SWATH-MS) in studies of ALS and FTD. Similarly, we describe disease-related protein-protein interaction (PPI) studies using approaches including immunoprecipitation mass spectrometry (IP-MS) and proximity-dependent biotin identification (BioID) and discuss future application of new techniques including proximity-dependent ascorbic acid peroxidase labeling (APEX), and biotinylation by antibody recognition (BAR). Furthermore, we explore the use of MS to detect post-translational modifications (PTMs), such as ubiquitination and phosphorylation, of disease-relevant proteins in ALS and FTD. We also discuss upstream technologies that enable enrichment of proteins of interest, highlighting the contributions of new techniques to isolate disease-relevant protein inclusions including flow cytometric analysis of inclusions and trafficking (FloIT). These recently developed approaches, as well as related advances yet to be applied to studies of these neurodegenerative diseases, offer numerous opportunities for discovery of potential therapeutic targets and biomarkers for ALS and FTD.

Multi-Study Proteomic and Bioinformatic Identification of Molecular Overlap between Amyotrophic Lateral Sclerosis (ALS) and Spinal Muscular Atrophy (SMA)

Unravelling the complex molecular pathways responsible for motor neuron degeneration in amyotrophic lateral sclerosis (ALS) and spinal muscular atrophy (SMA) remains a persistent challenge. Interest is growing in the potential molecular similarities between these two diseases, with the hope of better understanding disease pathology for the guidance of therapeutic development. The aim of this study was to conduct a comparative analysis of published proteomic studies of ALS and SMA, seeking commonly dysregulated molecules to be prioritized as future therapeutic targets. Fifteen proteins were found to be differentially expressed in two or more proteomic studies of both ALS and SMA, and bioinformatics analysis identified over-representation of proteins known to associate in vesicles and molecular pathways, including metabolism of proteins and vesicle-mediated transport—both of which converge on endoplasmic reticulum (ER)-Golgi trafficking processes. Calreticulin, a calcium-binding chaperone found in the ER, was associated with both pathways and we independently confirm that its expression was decreased in spinal cords from SMA and increased in spinal cords from ALS mice. Together, these findings offer significant insights into potential common targets that may help to guide the development of new therapies for both diseases.

Identification of Apo B48 and other novel biomarkers in amyotrophic lateral sclerosis patient fibroblasts

AIM

Amyotrophic lateral sclerosis (ALS) is a debilitating fatal neurodegenerative disorder. 90-95% of ALS cases are sporadic with no clear risk factors associated with the disease. Identification of biomarkers associated with ALS may lead to early detection and make it more amenable to therapeutic intervention.

MATERIALS & METHODS

SILAC was used to quantitatively analyze the proteomes of ALS and control human fibroblasts.

RESULTS

Out of total of 861 proteins identified, 33 were found to be differentially regulated. ApoB48 and Hsp20 were downregulated, while Fibulin-1 was upregulated.

CONCLUSION

We report the differential regulation of these proteins in ALS fibroblasts, and their potential as novel biomarkers and possible drug targets for ALS.

Stable isotope labeling for proteomic analysis of tissues in mouse

Since the first metabolic labeling experiments with stable isotopes beginning of the last century, several approaches were pursued to monitor protein dynamics in living animals. Today, almost all model organisms from bacteria to rodents can be fully labeled with SILAC (stable isotope labeling of amino acids in cell culture) amino acids. The development of special media and diets containing the labeled amino acids provides an efficient way to metabolically label prokaryotic and eukaryotic organisms. Preferentially, the essential amino acid lysine ((13)C6-lysine) is used to label mice (Mus musculus) and after one generation the natural isotope is fully replaced by the stable (13)C6-lysine isotope. So far, the SILAC mouse approach has been used to analyze several transgenic and knockout mouse models. Spike-in of labeled proteins into non-labeled samples provides an accurate relative protein quantification method without any chemical modification. Here we describe how to establish a SILAC mouse colony and describe the analysis of skeletal muscle tissue with different metabolic and contractile profiles.

Gene expression profile of SOD1-G93A mouse spinal cord, blood and muscle

The exact pathway leading to neuron death and muscle atrophy in amyotrophic lateral sclerosis (ALS) has not yet been elucidated. Gene expression profile of spinal cord, blood and muscle could provide signalling pathways and systemic alterations useful for future biomarker development. In our study we compared whole genome expression profiles of lumbar spinal cord with peripheral blood and tibialis anterior muscle in 16 mutant SOD1-G93A mice and 15 wild-type littermates. In SOD1-G93A mice, 11 genes were significantly differentially expressed in spinal cord, and 16 genes in blood, while much larger transcriptional changes were noted in muscle (1745 genes significant; six overlapping with spinal cord (0.3%)) probably due to muscle atrophy. Overlap with spinal cord was enriched for significant genes in blood (six of 16 overlapping with spinal cord (37.5%)). Three genes were significantly down-regulated in all three tissues, and were closely related to mitochondrial function. Furthermore, clustering the significant genes in spinal cord and in blood, but not in muscle, could identify the SOD1-G93A mice. We conclude that blood gene expression profile overlapped with profile of spinal cord, allowing differentiation of SOD1-G93A mice from wild-type littermates. Blood gene expression profiling may be a promising biomarker for ALS patients.

Molecular signatures of amyotrophic lateral sclerosis disease progression in hind and forelimb muscles of an SOD1(G93A) mouse model

AIMS

This study utilized proteomics, biochemical and enzymatic assays, and bioinformatics tools that characterize protein alterations in hindlimb (gastrocnemius) and forelimb (triceps) muscles in an amyotrophic lateral sclerosis (ALS) (SOD1(G93A)) mouse model. The aim of this study was to identify the key molecular signatures involved in disease progression.

RESULTS

Both muscle types have in common an early down-regulation of complex I. In the hindlimb, early increases in oxidative metabolism are associated with uncoupling of the respiratory chain, an imbalance of NADH/NAD(+), and an increase in reactive oxygen species (ROS) production. The NADH overflow due to complex I inactivation induces TCA flux perturbations, leading to citrate production, triggering fatty acid synthase (FAS), and lipid peroxidation. These early metabolic changes in the hindlimb followed by sustained and comparatively higher metabolic and cytoskeletal derangements over time precede and may catalyze the progressive muscle wasting in this muscle at the late stage. By contrast, in the forelimb, there is an early down-regulation of complexes I and II that is associated with the reduction of oxidative metabolism, which promotes metabolic homeostasis that is accompanied by a greater cytoskeletal stabilization response. However, these early compensatory systems diminish by a later time point.

INNOVATION

The identification of potential early- and late-stage disease molecular signatures in an ALS model: muscle albumin, complex I, complex II, citrate synthase, FAS, and phosphoinositide 3-kinase functions as diagnostic markers and peroxisome proliferator-activated receptor γ co-activator 1α (PGC1α), Sema-3A, and Rho-associated protein kinase 1 (ROCK1) play the role of disease progression markers.

CONCLUSION

The differing pattern of cellular metabolism and cytoskeletal derangements in the hind and forelimb identifies the potential dysmetabolism/hypermetabolism molecular signatures associated with disease progression, which may serve as diagnostic/disease progression markers in ALS patients.

Amyotrophic lateral sclerosis: new insights into underlying molecular mechanisms and opportunities for therapeutic intervention

Recent years have witnessed a renewed interest in the pathogenic mechanisms of amyotrophic lateral sclerosis (ALS), a late-onset progressive degeneration of motor neurons. The discovery of new genes associated with the familial form of the disease, along with a deeper insight into pathways already described for this disease, has led scientists to reconsider previous postulates. While protein misfolding, mitochondrial dysfunction, oxidative damage, defective axonal transport, and excitotoxicity have not been dismissed, they need to be re-examined as contributors to the onset or progression of ALS in the light of the current knowledge that the mutations of proteins involved in RNA processing, apparently unrelated to the previous "old partners," are causative of the same phenotype. Thus, newly envisaged models and tools may offer unforeseen clues on the etiology of this disease and hopefully provide the key to treatment.

Activity of the novel T137A SOD1 mutation in amyotrophic lateral sclerosis patients

Aim: More than 160 mutations in the Cu/Zn superoxide dismutase 1 (SOD1) gene are known and identifiable in approximately 20% of patients with familial history of amyotrophic lateral sclerosis (ALS) and in the 1–2% of the sporadic cases. A new SOD1 variant, the T137A missense mutation, was recently discovered. The aim of this study was to better clarify the activity of T137A-SOD1 in ALS patients. Methods: The activity of erythrocyte SOD enzyme and the plasma total antioxidant capacity were measured in 20 ALS patients with wild-type SOD1, in five ALS-patients harboring mutated SOD1 and in seven healthy controls. Results: Erythrocyte Cu/Zn-SOD activity was significantly lower in mutated patients than in wild-type participants and in controls, regardless of the type of mutation. Our data demonstrate that the SOD1 enzyme harboring the novel T137A mutation presents typical features of other known SOD1 variants. Conclusion: These results provide further evidence of the possible pathogenic role of the T137A mutation in ALS.

Altered gene expression, mitochondrial damage and oxidative stress: converging routes in motor neuron degeneration

Motor neuron diseases (MNDs) are a rather heterogeneous group of diseases, with either sporadic or genetic origin or both, all characterized by the progressive degeneration of motor neurons. At the cellular level, MNDs share features such as protein misfolding and aggregation, mitochondrial damage and energy deficit, and excitotoxicity and calcium mishandling. This is particularly well demonstrated in ALS, where both sporadic and familial forms share the same symptoms and pathological phenotype, with a prominent role for mitochondrial damage and resulting oxidative stress. Based on recent data, however, altered control of gene expression seems to be a most relevant, and previously overlooked, player in MNDs. Here we discuss which may be the links that make pathways apparently as different as altered gene expression, mitochondrial damage, and oxidative stress converge to generate a similar motoneuron-toxic phenotype.

SOD1 and TDP-43 animal models of amyotrophic lateral sclerosis: recent advances in understanding disease toward the development of clinical treatments

Amyotrophic lateral sclerosis (ALS) is a fatal motor neuron disease with no cure. Breakthroughs in understanding ALS pathogenesis came with the discovery of dominant mutations in the superoxide dismutase 1 gene (SOD1) and other genes, including the gene encoding transactivating response element DNA binding protein-43 (TDP-43). This has led to the creation of animal models to further our understanding of the disease and identify a number of ALS-causing mechanisms, including mitochondrial dysfunction, protein misfolding and aggregation, oxidative damage, neuronal excitotoxicity, non-cell autonomous effects and neuroinflammation, axonal transport defects, neurotrophin depletion, effects from extracellular mutant SOD1, and aberrant RNA processing. Here we summarise the SOD1 and TDP-43 animal models created to date, report on recent findings supporting the potential mechanisms of ALS pathogenesis, and correlate this understanding with current developments in the clinic.

Proteins that bind to misfolded mutant superoxide dismutase-1 in spinal cords from transgenic amyotrophic lateral sclerosis (ALS) model mice

Mutant superoxide dismutase-1 (SOD1) has an unidentified toxic property that provokes ALS. Several ALS-linked SOD1 mutations cause long C-terminal truncations, which suggests that common cytotoxic SOD1 conformational species should be misfolded and that the C-terminal end cannot be involved. The cytotoxicity may arise from interaction of cellular proteins with misfolded SOD1 species. Here we specifically immunocaptured misfolded SOD1 by the C-terminal end, from extracts of spinal cords from transgenic ALS model mice. Associated proteins were identified with proteomic techniques. Two transgenic models expressing SOD1s with contrasting molecular properties were examined: the stable G93A mutant, which is abundant in the spinal cord with only a tiny subfraction misfolded, and the scarce disordered truncation mutant G127insTGGG. For comparison, proteins in spinal cord extracts with affinity for immobilized apo G93A mutant SOD1 were determined. Two-dimensional gel patterns with a limited number of bound proteins were found, which were similar for the two SOD1 mutants. Apart from neurofilament light, the proteins identified were all chaperones and by far most abundant was Hsc70. The immobilized apo G93A SOD1, which would populate a variety of conformations, was found to bind to a considerable number of additional proteins. A substantial proportion of the misfolded SOD1 in the spinal cord extracts appeared to be chaperone-associated. Still, only about 1% of the Hsc70 appeared to be associated with misfolded SOD1. The results argue against the notion that chaperone depletion is involved in ALS pathogenesis in the transgenic models and in humans carrying SOD1 mutations.

ALS-linked mutant superoxide dismutase 1 (SOD1) alters mitochondrial protein composition and decreases protein import

Mutations in superoxide dismutase 1 (SOD1) cause familial ALS. Mutant SOD1 preferentially associates with the cytoplasmic face of mitochondria from spinal cords of rats and mice expressing SOD1 mutations. Two-dimensional gels and multidimensional liquid chromatography, in combination with tandem mass spectrometry, revealed 33 proteins that were increased and 21 proteins that were decreased in SOD1(G93A) rat spinal cord mitochondria compared with SOD1(WT) spinal cord mitochondria. Analysis of this group of proteins revealed a higher-than-expected proportion involved in complex I and protein import pathways. Direct import assays revealed a 30% decrease in protein import only in spinal cord mitochondria, despite an increase in the mitochondrial import components TOM20, TOM22, and TOM40. Recombinant SOD1(G93A) or SOD1(G85R), but not SOD1(WT) or a Parkinson's disease-causing, misfolded α-synuclein(E46K) mutant, decreased protein import by >50% in nontransgenic mitochondria from spinal cord, but not from liver. Thus, altered mitochondrial protein content accompanied by selective decreases in protein import into spinal cord mitochondria comprises part of the mitochondrial damage arising from mutant SOD1.

Galectin-3 is a candidate biomarker for amyotrophic lateral sclerosis: discovery by a proteomics approach

The discovery of biomarkers for neurodegenerative diseases will have a major impact on the efficiency of therapeutic clinical trials and may be important for understanding basic pathogenic mechanisms. We have approached the discovery of protein biomarkers for amyotrophic lateral sclerosis (ALS) by profiling affected tissues in a relevant animal model and then validating the findings in human tissues. Ventral roots from SOD1(G93A) "ALS" mice were analyzed by label-free quantitative mass spectrometry, and the resulting data were compared with data for matched samples from nontransgenic littermates and transgenic mice carrying wild-type human SOD1 (SOD1(WT)). Of 1299 proteins, statistical inference of the data in the three groups identified 14 proteins that were dramatically altered in the ALS mice compared with the two control groups. The protein galectin-3 emerged as a lead biomarker candidate on the basis of its differential expression as assessed by immunoblot and immunocytochemistry in SOD1(G93A) mice as compared to controls and because it is a secreted protein that could potentially be measured in human biofluids. Spinal cord tissue from ALS patients also exhibited increased levels of galectin-3 when compared to controls. Further measurement of galectin-3 in cerebrospinal fluid samples showed that ALS patients had approximately twice as much galectin-3 as normal and disease controls. These results provide the proof of principle that biomarker identification in relevant and well-controlled animal models can be translated to human disease. The challenge is to validate our biomarker candidate proteins as true biomarkers for ALS that will be useful for diagnosis and/or monitoring disease activity in future clinical trials.

Early excitability changes in lumbar motoneurons of transgenic SOD1G85R and SOD1G(93A-Low) mice

This work characterizes the properties of wild-type (WT) mouse motoneurons in the second postnatal week and compares these at the same age and in the same conditions to those of two different SOD1 mutant lines used as models of human amyotrophic lateral sclerosis (ALS), the SOD1(G93A) low expressor line and SOD1(G85R) line, to describe any changes in the functional properties of mutant motoneurons (Mns) that may be related to the pathogenesis of human ALS. We show that very early changes in excitability occur in SOD1 mutant Mns that have different properties from those of WT animals. The SOD1(G93A-Low) low expressor line displays specific differences that are not found in other mutant lines including a more depolarized membrane potential, larger spike width, and slower spike rise slope. With current pulses SOD1(G93A-Low) were hyperexcitable, but both mutants had a lower gain with current ramps stimulation. Changes in the threshold and intensities of Na(+) and Ca(2+) persistent inward currents were also observed. Low expressor mutants show reduced total persistant inward currents compared with WT motoneurons in the same recording conditions and give arguments toward modifications of the balance between Na(+) and Ca(2+) persistent inward currents. During the second week postnatal, SOD1(G93A-Low) lumbar motoneurons appear more immature than those of SOD1(G85R) compared with WT and we propose that different time course of the disease, possibly linked with different toxic properties of the mutated protein in each model, may explain the discrepancies between excitability changes described in the different models.