The classification of neurodegenerative disease from acoustic speech data

Neurodegenerative diseases often affect speech. Speech acoustics can be used as objective clinical markers of pathology. Previous investigations of pathological speech have primarily compared controls with one specific condition and excluded comorbidities. We broaden the utility of speech markers by examining how multiple acoustic features can delineate diseases. We used supervised machine learning with gradient boosting (CatBoost) to differentiate healthy speech and speech from people with multiple sclerosis or Friedreich ataxia. Participants performed a diadochokinetic task where they repeated alternating syllables. We extracted 74 spectral and temporal prosodic features from the speech recordings, which were subjected to machine learning. Results showed that Friedreich ataxia, multiple sclerosis and healthy controls were all identified with high accuracy (over 82%). Twenty-one acoustic features were strong markers of neurodegenerative diseases, falling under the categories of spectral qualia, spectral power, and speech rate. We demonstrated that speech markers can delineate neurodegenerative diseases and distinguish healthy speech from pathological speech with high accuracy. Findings emphasize the importance of examining speech outcomes when assessing indicators of neurodegenerative disease. We propose large-scale initiatives to broaden the scope for differentiating other neurological diseases and affective disorders.

Epigenetic Heterogeneity in Friedreich Ataxia Underlies Variable FXN Reactivation

Friedreich ataxia (FRDA) is typically caused by homozygosity for an expanded GAA triplet-repeat in intron 1 of the FXN gene. The expanded repeat induces repressive histone changes and DNA hypermethylation, which result in epigenetic silencing and FXN transcriptional deficiency. A class I histone deacetylase inhibitor (HDACi-109) reactivates the silenced FXN gene, although with considerable inter-individual variability, which remains etiologically unexplained. Because HDAC inhibitors work by reversing epigenetic silencing, we reasoned that epigenetic heterogeneity among patients may help to explain this inter-individual variability. As a surrogate measure for epigenetic heterogeneity, a highly quantitative measurement of DNA hypermethylation via bisulfite deep sequencing, with single molecule resolution, was used to assess the prevalence of unmethylated, partially methylated, and fully methylated somatic FXN molecules in PBMCs from a prospective cohort of 50 FRDA patients. Treatment of the same PBMCs from this cohort with HDACi-109 significantly increased FXN transcript to levels seen in asymptomatic heterozygous carriers, albeit with the expected inter-individual variability. Response to HDACi-109 correlated significantly with the prevalence of unmethylated and partially methylated FXN molecules, supporting the model that FXN reactivation involves a proportion of genes that are amenable to correction in non-dividing somatic cells, and that heavily methylated FXN molecules are relatively resistant to reactivation. FXN reactivation is a promising therapeutic strategy in FRDA, and inter-individual variability is explained, at least in part, by somatic epigenetic heterogeneity.

The Mutational Dynamics of Short Tandem Repeats in Large, Multigenerational Families

Short tandem repeats (STRs) are tandemly repeated sequences of 1-6 bp motifs. STRs compose approximately 3% of the genome, and mutations at STR loci have been linked to dozens of human diseases including amyotrophic lateral sclerosis, Friedreich ataxia, Huntington disease, and fragile X syndrome. Improving our understanding of these mutations would increase our knowledge of the mutational dynamics of the genome and may uncover additional loci that contribute to disease. Here, to estimate the genome-wide pattern of mutations at STR loci, we analyzed blood-derived whole-genome sequencing data for 544 individuals from 29 three-generation CEPH pedigrees. These pedigrees contain both sets of grandparents, the parents, and an average of 9 grandchildren per family. Using HipSTR we identified de novo STR mutations in the 2nd generation of these pedigrees. Analyzing ~1.6 million STR loci, we estimate the empircal de novo STR mutation rate to be 5.24*10-5 mutations per locus per generation. We find that perfect repeats mutate ~2x more often than imperfect repeats. De novo STRs are significantly enriched in Alu elements (p < 2.2e-16). Approximately 30% of STR mutations occur within Alu elements, which compose only ~11% of the genome, and ~10% are found in LINE-1 insertions, which compose ~17% of the genome. Phasing these de novo mutations to the parent of origin shows that parental transmission biases vary among families. We estimate the average number of de novo genome-wide STR mutations per individual to be ~85, which is similar to the average number of observed de novo single nucleotide variants.

Body Mass Index and Height in the Friedreich Ataxia Clinical Outcome Measures Study

Background and ObjectivesBody mass index (BMI) and height are important indices of health. We tested the association between these outcomes and clinical characteristics in Friedreich ataxia (FRDA), a progressive neuromuscular disorder.MethodsParticipants (N = 961) were enrolled in a prospective natural history study (Friedreich Ataxia Clinical Outcome Measure Study). Age- and sex-specific BMI and height Z-scores were calculated using CDC 2000 references for participants younger than 18 years. For adults aged 18 years or older, height Z-scores were also calculated, and absolute BMI was reported. Univariate and multivariate linear regression analyses tested the associations between exposures, covariates, and BMI or height measured at the baseline visit. In children, the superimposition by translation and rotation analysis method was used to compare linear growth trajectories between FRDA and a healthy reference cohort, the Bone Mineral Density in Childhood Study (n = 1,535 used for analysis).ResultsMedian age at the baseline was 20 years (IQR, 13–33 years); 49% (n = 475) were women. A substantial proportion of children (17%) were underweight (BMI-Z < fifth percentile), and female sex was associated with lower BMI-Z (β = −0.34, p < 0.05). In adults, older age was associated with higher BMI (β = 0.09, p < 0.05). Regarding height, in children, older age (β −0.06, p < 0.05) and worse modified Friedreich Ataxia Rating Scale (mFARS) scores (β = −1.05 for fourth quartile vs first quartile, p < 0.01) were associated with shorter stature. In girls, the magnitude of the pubertal growth spurt was less, and in boys, the pubertal growth spurt occurred later (p < 0.001 for both) than in a healthy reference cohort. In adults, in unadjusted analyses, both earlier age of FRDA symptom onset (=0.09, p < 0.05) and longer guanine-adenine-adenine repeat length (shorter of the 2 GAA repeats, β = −0.12, p < 0.01) were associated with shorter stature. Both adults and children with higher mFARS scores and/or who were nonambulatory were less likely to have height and weight measurements recorded at clinical visits.DiscussionFRDA affects both weight gain and linear growth. These insights will inform assessments of affected individuals in both research and clinical settings.

Ferroptosis and NRF2: an emerging battlefield in the neurodegeneration of Alzheimer’s disease

Ferroptosis is an iron- and lipid peroxidation-dependent cell death modality and emerging evidence indicates that ferroptosis has great explanatory potential for neuronal loss and associated CNS dysfunction in a range of neurodegenerative diseases (e.g., Alzheimer’s, Parkinson’s and Huntington’s diseases, Motor neuron disease, Friedreich ataxia (FRDA)). Ferroptotic death results from lethal levels of phospholipid hydroperoxides that are generated by iron-dependent peroxidation of polyunsaturated fatty acids (PUFAs), such as arachidonic and adrenic acids, which are conjugated to specific phospholipids (e.g., phosphatidylethanolamines (PEs)). The major cellular protector against ferroptosis is glutathione peroxidase 4 (GPX4), a membrane-associated selenoenzyme that reduces deleterious phospholipid hydroperoxides to their corresponding benign phospholipid alcohols in a glutathione-dependent manner. Other complementary protective systems have also been identified that act to bolster cellular defences against ferroptosis. Many pharmacological modulators of the ferroptosis pathway have been identified, targeting proteins involved in iron homoeostasis and autophagy; the production and detoxification of lipid peroxides, and cyst(e)ine/glutathione metabolism. While a growing number of cell signalling pathways converge to regulate the ferroptosis cascade, an emerging understanding of ferroptosis regulation suggests that the ferroptotic ‘tone’ of cells can be set by the transcription factor, nuclear factor erythroid 2-related factor 2 (NRF2), which transcriptionally controls many key components of the ferroptosis pathway. In this review, we provide a critical overview of the relationship between ferroptosis and NRF2 signalling. With a focus on the role of ferroptosis in Alzheimer’s disease (AD), we discuss how therapeutic modulation of the NRF2 pathway is a viable strategy to explore in the treatment of ferroptosis-driven neurodegeneration.

Steroidogenic electron-transfer factors and their diseases

Most steroidogenesis disorders are caused by mutations in genes encoding the steroidogenic enzymes, but work in the past 20 years has identified related disorders caused by mutations in the genes encoding the cofactors that transport electrons from NADPH to P450 enzymes. Most P450s are microsomal and require electron donation by P450 oxidoreductase (POR); by contrast, mitochondrial P450s require electron donation via ferredoxin reductase (FdxR) and ferredoxin (Fdx). POR deficiency is the most common and best-described of these new forms of congenital adrenal hyperplasia. Severe POR deficiency is characterized by the Antley-Bixler skeletal malformation syndrome and genital ambiguity in both sexes, and hence is easily recognized, but mild forms may present only with infertility and subtle disorders of steroidogenesis. The common POR polymorphism A503V reduces catalysis by P450c17 (17-hydroxylase/17,20-lyase) and the principal drugmetabolizing P450 enzymes. The 17,20-lyase activity of P450c17 requires the allosteric action of cytochrome b5, which promotes interaction of P450c17 with POR, with consequent electron transfer. Rare b5 mutations are one of several causes of 17,20-lyase deficiency. In addition to their roles with steroidogenic mitochondrial P450s, Fdx and FdxR participate in the synthesis of iron-sulfur clusters used by many enzymes. Disruptions in the assembly of Fe-S clusters is associated with Friedreich ataxia and Parkinson disease. Recent work has identified mutations in FdxR in patients with neuropathic hearing loss and visual impairment, somewhat resembling the global neurologic disorders seen with mitochondrial diseases. Impaired steroidogenesis is to be expected in such individuals, but this has not yet been studied.