Comparative brain metabolomics reveals shared and distinct metabolic alterations in Alzheimer’s disease and progressive supranuclear palsy

Richa Batra, Jan Krumsiek, Xue Wang, Mariet Allen, Colette Blach, Gabi Kastenmüller, Matthias Arnold, Nilüfer Ertekin-Taner, Rima F. Kaddurah-Daouk, Alzheimer’s Disease Metabolomics Consortium (ADMC)
doi: https://doi.org/10.1101/2023.07.25.23293055

Abstract

Metabolic dysregulation is a hallmark of neurodegenerative diseases, including Alzheimer’s disease (AD) and progressive supranuclear palsy (PSP). While metabolic dysregulation is a common link between these two tauopathies, a comprehensive brain metabolic comparison of the diseases has not yet been performed. We analyzed 342 postmortem brain samples from the Mayo Clinic Brain Bank and examined 658 metabolites in the cerebellar cortex and the temporal cortex between the two tauopathies. Our findings indicate that both diseases display oxidative stress associated with lipid metabolism, mitochondrial dysfunction linked to lysine metabolism, and an indication of tau-induced polyamine stress response. However, specific to AD, we detected glutathione-related neuroinflammation, deregulations of enzymes tied to purines, and cognitive deficits associated with vitamin B. Taken together, our findings underscore vast alterations in the brain’s metabolome, illuminating shared neurodegenerative pathways and disease-specific traits in AD and PSP.

1 Introduction

Tau protein hyperphosphorylation and its abnormal accumulation in the brain is a hallmark of several neurodegenerative diseases, including Alzheimer’s disease (AD) and progressive supranuclear palsy (PSP)1. The major clinical symptom of AD is dementia, which predominantly impairs memory2, while PSP is a Parkinsonian movement disorder3. Despite both representing tauopathies, these diseases have several key differences: First, AD is a dual proteinopathy, displaying deposition of both β-amyloid and tau tangles, while PSP is considered to be a pure tauopathy1. Second, the specific tau proteins in the two diseases differ in their protein isoforms4. While AD presents neuronal tangles composed of tau of both three (3R) and four (4R) microtubule-binding repeat isoforms, shifting from 3R to 4R with disease progression5, PSP demonstrates different tau-inclusion patterns predominantly composed of 4R in neurons, astrocytes, oligodendrocytes, and white matter. Further, although both diseases are associated with tau-encoding MAPT haplotypes6, 7, distinct genetic risk factors have been discovered for AD8 and PSP7, signifying divergent genetic components contributing to these two diseases. These differences in the genetic risk, involved proteins, and tau-isoforms might be the basis of the divergent molecular and clinical presentations of these two pathologies.

Comparing the two diseases at the molecular level, we previously reported a high overlap in transcriptomic signatures of these diseases9, 10, including dysregulation of biological processes like myelination. As a particularly affected cellular process, metabolic dysregulation involving mitochondrial dysfunction is a common disease pathway in neurodegenerative diseases11. Especially in AD pathology, such metabolic modulation is an established key component, affecting cellular processes both in the brain and in the periphery12–18. Moreover, our recent study on postmortem human brain tissue suggested tau as the potential driver of AD-associated metabolic alterations in the central nervous system15. In the context of PSP, several small-scale studies have indicated the presence of metabolic dysregulation in CSF and in blood19, 20.

As AD and PSP share the above-mentioned pathological features, i.e., tauopathy and metabolic alterations, comparing their metabolomic profiles can help identify shared biological pathways that could be targeted for therapeutic interventions. Conducting a thorough brain metabolic comparison of the two diseases is crucial to determine the precise alterations in metabolic homeostasis, which is a hallmark of neurodegenerative diseases15, 21. To date, there have been no brain metabolic profiling studies of PSP and no comparative metabolomic studies for these tauopathies.

To address this need, we present a human brain-based metabolic comparison derived from a sizable cohort of neuropathologically diagnosed AD, PSP, and control donors. We analyzed 342 brain samples, comprising 181 from the cerebellar cortex (CER) and 161 from the temporal cortex (TCX). These two brain regions exhibit varying degrees of proteinopathy across the two diseases, with CER being largely unaffected in both22, 23 and TCX being affected predominantly in AD23, 24. While proteinopathy in the CER of AD patients seems to emerge in later stages25, CER has been reported to exhibit disease-driven changes, including atrophy, diffused amyloid presence, and increased microglial activity26. From the perspective of PSP, proteinopathy has traditionally only been reported in the anterior CER27; however, a recent whole-brain magnetic resonance spectroscopy-based study has revealed metabolic changes across various brain regions including both CER and TCX28. Moreover, our previous research has identified transcriptomic alterations in both diseases and brain regions9, 10. Hence, we anticipated our analysis will not only identify the metabolomic changes linked with proteinopathies but also the metabolomic dysregulation instigated by neurodegeneration even in the absence of substantial proteinopathy. An overview of the study is depicted in Figure 1.

Figure 1:
Figure 1: Study overview

342 Mayo Clinic Brain Bank samples were included in this analysis. For each sample, diagnosis and demographic information were available. Metabolomic profiling was performed on samples from the cerebellar cortex (CER) and temporal cortex (TCX) regions. Metabolic signatures of each diagnosis, i.e., Alzheimer’s disease (AD) and progressive supranuclear palsy (PSP) in comparison to metabolic profiles of controls without neurodegenerative disease. A comparison of the metabolic signatures of the two tauopathies was performed, highlighting the overlap in altered metabolites and metabolic pathways commonly dysregulated in the two diseases.

2 Results

2.1 Metabolic signatures of the two diseases

We analyzed 342 brain samples from the Mayo Clinic Brain Bank, from Alzheimer’s disease (AD) and progressive supranuclear palsy (PSP) patients as well as controls without neurodegenerative pathologic diagnoses, across two brain regions, cerebellar cortex (CER) and temporal cortex (TCX). Demographic characteristics of the samples across the diagnoses and brain regions are shown in Table 1.

View this table:
Table 1: Cohort overview

AD = Alzheimer’s disease. PSP = Progressive supranuclear palsy. IQR = interquartile range, i.e., middle 50% of the value range.

Samples were profiled using an untargeted LC-MS-based metabolomics platform. This resulted in measurements of 658 metabolites from various metabolic ‘super-pathways’ (Figure 2a, Supplementary Table 1), covering lipids (44.4%), amino acids (23.3%), nucleotide (7.0%), carbohydrates (5.3%), cofactors and vitamins (3.8%), peptides (2.7%), xenobiotics (2.7%), energy-related metabolites (1.5%), and a series of uncharacterized metabolites (9.3%).

Figure 2:
Figure 2: Overview of metabolites and their associations with AD and PSP

a. Distribution of metabolites detected by metabolomics platform across metabolic classes. b. Number of metabolic associations for each brain region and each disease. c. Overlap of metabolic associations between the two diseases in cerebellar cortex samples. d. Overlap of metabolic associations between the two diseases in temporal cortex samples.

To determine the metabolic imprint of the two diseases in the brain, we performed a brain region-wise statistical comparison between disease and control samples. All models were corrected for the number of APOE ε4 alleles, age at death, and sex. In the CER region, a total of 153 metabolites were associated with AD, 35 were associated with PSP (Figure 2b), of which 27 were commonly associated with both diseases (Figure 2c). In the TCX region, 63 metabolites were associated with AD, only 2 were associated with PSP (Figure 2b) and none were shared between the two diseases (Figure 2d). Detailed statistical results from the two brain regions can be found in Supplementary Tables 2-5, respectively.

Overall, AD-related metabolic alterations were observed in both brain regions, while PSP-related alterations had a regional preference for CER and were less pronounced compared to AD-related alterations. Of note, the sample sizes were comparable for both brain regions as well as diseases (Table 1), suggesting that statistical power was unlikely to cause this discrepancy. In the following, we focused on the CER region for a detailed comparison of AD and PSP, as this region had commonly perturbed metabolites between these two diseases, whereas TCX had none.

2.2 Comparison of AD and PSP metabolic signatures in CER

To compare the metabolic signatures of the two diseases in CER, each metabolite was categorized using a ‘sub-pathway’ annotation. There were 153 metabolites significantly associated with AD, 136 of these 153 were annotated with 53 pathways, and the remaining 17 metabolites were uncharacterized and thus were excluded from further analysis. Similarly, there were 35 metabolites significantly associated with PSP, 30 of which were annotated with 19 pathways, and the remaining 5 metabolites were uncharacterized and thus were excluded from further analysis. Overall, 35 pathways were exclusively dysregulated in AD, one pathway was exclusive to PSP, and 18 pathways were shared across the two diseases. These pathways were distributed across all eight super-pathways, with amino acids and lipids representing the groups with the most changes (Figure 3).

Figure 3:
Figure 3: Functional annotation of metabolic signatures in CER region

Within the cerebellar cortex region, for each disease and pathway, the number of positively or negatively associated metabolites is shown in orange an green boxes respectively. These pathways are further annotated by super-pathways. The significance of these pathways in one or both diseases is shown per pathway. Venn diagram depicts the overlap of the number of significant pathways in the two diseases. FA – Fatty acid; LC – Long chain.

Pathways exclusively dysregulated in AD

35 pathways were dysregulated in AD but not in PSP. Among these, various amino acid pathways were negatively associated with AD including the urea cycle, alanine & aspartate, tyrosine, and branched-chain amino acid (BCAA) metabolism, while glutathione metabolism was positively associated. Further, the metabolism of vitamins B1, B2, and B6 was negatively associated with AD, while vitamin C metabolism was positively associated. Moreover, several pathways belonging to the metabolic groups of peptides, carbohydrates, energy metabolites, and nucleotide metabolism were dysregulated only in AD.

Pathways exclusively dysregulated in PSP

Dicarboxylate metabolism was the only pathway uniquely dysregulated in PSP. However, within this pathway, only one of the seven metabolites, the fatty acid 3-hydroxyadipate, was significantly lower in PSP as compared to controls, while changes in the remaining six metabolites were insignificant. Thus, there is limited evidence to support the dysregulation of this pathway in PSP.

Pathways commonly dysregulated in both diseases

18 pathways were commonly perturbed in AD and PSP. These pathways were mainly from the amino acid and lipid groups. Within the amino acid group, creatine, glutamate, and lysine metabolism were positively associated with both diseases, while polyamine, glycine, and histidine metabolism were negatively associated. Within the lipid group, ceramides, diacylglycerol, and phospholipids were positively associated with both diseases, while sphingomyelin, lysophospholipid, phosphotidylcholine, phosphoethanolamine, and FA (Acyl Carnitine, Monounsaturated) were negatively associated.

Overall, in the two diseases, common metabolic perturbations were better captured at the functional level of pathways (18/54 changed, 33.34%) than at the individual metabolite level (27/161 changed, 16.77%).

2.3 Functional insights into unique and shared metabolic dysregulation in AD and PSP in CER

As outlined above, in the CER region, 35 metabolic pathways were exclusively dysregulated in AD, only one pathway was exclusively dysregulated in PSP, and 18 pathways were shared across the diseases (Figure 4). In the following, we discuss functional insights into the commonalities and differences of metabolic dysregulation in the two diseases based on selected pathways.

Figure 4:
Figure 4: Comparison of deregulated pathways in the two diseases

Within the cerebellar cortex region, 3 pathways were exclusively dysregulated in AD, 18 pathways were shared between the two diseases, and on pathway was exclusively dysregulated in PSP. The labeled processes indicate selected pathways discussed in th text. Included for illustrative purposes are stylistic representations of the tau isoform associated with each tauopathy. Within these protein structures, residues in R1–R4 and in the C-terminal domain are colored purple, blue, green, gold, and orange, respectively24.

2.3.1 Metabolic pathways altered exclusively in AD in CER

Within the CER region, 35 metabolic pathways were dysregulated exclusively in AD. Based on the number of metabolites altered, we highlight the three most perturbed pathways from three different metabolic groups in the following.

Glutathione metabolism and neuroinflammation

Neuroinflammation is central to AD pathogenesis and is marked by the release of pro-inflammatory molecules as well as reactive oxygen species29. Antioxidants such as glutathione play a key role in combating oxidative stress resulting from neuroinflammation30. Neuroinflammation and oxidative stress in the brain are believed to mediate neuronal dysfunction and death29. In our data, 7 out of 13 metabolites from glutathione metabolism were higher in AD as compared to controls. Of these seven, glutathione oxidized (GSSG), ophthalmate, and 4-hydroxy-nonenal-glutathione, are known biomarkers of oxidative stress31–33. Of note, based on genomic and transcriptomic data, we have previously demonstrated that this pathway is disrupted in AD34.

Purine metabolism and enzymatic reactions

Purine bases serve as cofactors for various biochemical reactions and are thus crucial for many metabolic processes30. Moreover, adenosine has been implicated in the modulation of cognition and memory, as well as the permeability of the blood-brain barrier35. In our data, 5 metabolites from adenine metabolism and one from guanine metabolism were higher in AD as compared to controls. Purine metabolism has been shown to be perturbed in AD; however, effect directions were inconsistent between cerebrospinal fluid and brain regions13, 35, 36. Overall, our findings indicate a dysregulation of purines in AD brains which has the potential to alter various brain functions.

Vitamin B metabolism and cognitive deficit

Multiple B vitamins are considered protective against cognitive decline37. In our data, thiamine (vitamin B1), riboflavin (vitamin B2), and pyridoxine (vitamin B6) were found to be lower in AD as compared to controls. Thiamine-dependent enzymes play central roles in glucose metabolism38. In AD brains, a reduction in thiamine levels has been associated with the diminished activity of thiamine-dependent enzymes, including transketolase from the pentose phosphate pathway and 2-ketoglutarate dehydrogenase complex from the tricarboxylic acid (TCA) cycle38, 39. Riboflavin and pyridoxine in their co-enzymatic forms are involved in one-carbon metabolism40. Riboflavin acts as a co-factor in the folate cycle, while pyridoxine is involved in the transsulphuration pathway40. Overall, these vitamins are essential for a well-functioning metabolism, and reduction in their levels has been associated with a decline in cognitive functions37, 41.

Of note, all of the above-mentioned AD-related metabolic dysregulations identified in the CER region were also detected in our previous findings in the dorsolateral prefrontal cortex region of the ROS/MAP cohort15.

2.3.2 Metabolic pathways altered in both diseases

Within the CER region, 18 metabolic pathways were dysregulated in both diseases. In the following, we highlight pathways from the two biggest metabolic classes, lipids and amino acids.

Lipid metabolism and oxidative stress

In neurodegenerative diseases, lipid metabolism has been widely implicated in disease pathogenesis, including β-amyloid-associated lipid rafts, increased lipid peroxidation42, altered cholesterol and lipid homeostasis43, as well as phospholipase 2 mediated lipid cleavage44. In our data, lipids from various lipid classes were dysregulated, with 15 lipids dysregulated in AD and 8 of those 15 also dysregulated in PSP.

Oxidative stress, a hallmark of neurodegenerative diseases, induces lipid peroxidation, which leads to increased phospholipase activity45, 46. The breakdown of phospholipids by phospholipases leads to compromised membrane integrity47, which affects neuronal excitability45. In our data, several phosphatidylcholines (PCs), the major membrane phospholipids, were lower in both diseases compared to controls. Phospholipases degrade PCs into lyso-PCs, which get further broken down into glycerophoshorylcholine (GPC) and fatty acids48. GPC was high in both diseases as compared to controls in our study.

Moreover, oxidative stress is also known to increase ceramide production, which in turn may trigger synaptic dysfunction and neuronal death49. Ceramides play an important role in many cellular processes, including lipid signaling, metabolic homeostasis, and mitochondria-linked apoptosis50. In our data, ceramide levels were positively associated with both diseases with varying degrees. N-stearoyl-sphingadienine (d18:2/18:0)* was high in both diseases compared to controls, while lactosyl-N-stearoyl-sphingosine (d18:1/18:0)*, a lactosylceramide, and N-stearoyl-sphinganine (d18:0/18:0)*, a dihydroceramide, were high only in AD compared to controls. Moreover, the myelin sheath which assists in signal transduction across neurons is composed of ceramides50. Notably, in our previous transcriptome-based studies, we observed myelination dysregulation in both AD and PSP9, 51.

Lysine metabolism and mitochondrial dysfunction

Lysine is an essential amino acid that can be catabolized by two pathways: the saccharopine pathway in mitochondria and the pipecolate pathway in cytosol52. In our data, 10 metabolites from lysine metabolism were dysregulated. Of these, all 10 were perturbed in AD, while only 3 metabolites were altered in PSP. One of the ten metabolites was saccharopine, which was high in both AD and PSP as compared to controls. Previous studies in C.elegans and mice have argued that saccharopine accumulation leads to mitochondrial dysfunction53. Our results are thus another potential piece of evidence in support of mitochondrial dysregulation in the pathogenesis of tauopathies54.

Polyamine metabolism and tau-mediated stress

In our study, two out of eight metabolites from polyamines metabolism, spermidine and spermine, were lower in both tauopathies as compared to controls. Polyamines are known for their neuroprotective effects, such as the clearance of apoptotic cells via efferocytosis55 and autophagy56, both mechanisms are affected in AD and other neurodegenerative diseases57. Moreover, mouse-based studies have speculated a complex interaction between polyamines and tau neuropathology58, 59. Specifically, it has been hypothesized that spermidine and spermine reduce tau fibrillization, oligomerization, and tau seeding/propagation, while tau can lead to polyamine stress response58. The polyamine stress response is defined as the alteration in the brain polyamine metabolism that is stimulated by a stressor leading to an increase in putrescine level, while levels of spermine and spermidine decrease or remain unchanged60, 61. Notably, in our data putrescine levels were unchanged. Previous metabolomic studies have reported that metabolite ratios associate better with outcomes than individual metabolites62. Therefore, we performed a ratio analysis for the metabolites in the polyamine metabolite pathway (Supplementary Tables 6-7). In our analysis, ratios of putrescine/spermidine and putrescine/spermine were higher in both tauopathies as compared to controls indicating relatively higher levels of putrescine compared to the other polyamines.

3 Discussion

The two tauopathies AD and PSP exhibit several shared and unique characteristics. They differ in the genetic risk factors involved, tau-isoforms, main symptoms, and type of proteinopathy, with PSP being a pure tauopathy and AD exhibiting deposition of β-amyloid along with tau1. At the molecular level, the two diseases display similarities, particularly in the alterations of biological processes including myelination9 and mitochondrial metabolism10. However, the comparative analysis of alterations in bioenergetic homeostasis, a key feature of neurodegeneration21, between AD and PSP was an aspect not previously explored until this study. We provide a comparative account of metabolic alterations in AD and PSP. In a sizable cohort of 342 samples of AD, PSP or control donors from the Mayo Clinic Brain Bank, we profiled 658 metabolites in CER and TCX brain regions. 200 (30.4%) of these metabolites were significantly associated with at least one of the diseases in at least one of the regions. In the following, we discuss the key findings of our study.

Our results show that AD-related metabolic alterations encompass both regions, i.e., CER and TCX. We identified 153 AD-associated metabolic alterations within the CER region and 63 alterations within the TCX region. Remarkably, we found a larger number of metabolic alterations in CER compared to TCX. This was unexpected as CER is relatively devoid of significant gross neuropathology in AD23, although some pathology has been observed in the late phases of the disease25. The metabolic alterations observed in CER may be due to the presence of either neuropathology in late-phase AD or toxic β-amyloid/tau oligomer species. Alternatively, the metabolic changes seen in CER could reflect the modulation of CER because of the pathological changes in other regions of the brain, for example, due to altered neural network connectivity63. The greater number of metabolic alterations in the CER compared to the TCX could indicate the varying metabolic needs of these regions and how these needs shift with age or during neurodegeneration64. Further, our results indicate that PSP-related metabolic alterations were scarce in TCX (2 metabolites), while some metabolic alterations were observed in CER (35 metabolites). Since TCX has limited neuropathology in PSP24, our results were in line with what may be expected from this relatively unaffected region. While the CER generally shows minimal significant neuropathology in PSP22, it may still be susceptible to metabolic alterations due to the above-mentioned reasons.

Within the CER region, AD and PSP significantly differed in the number of metabolic alterations (153 in AD and 35 in PSP). As opposed to our previous speculation that tau is the potential driver of AD-associated metabolic alterations in the dorsolateral prefrontal cortex15, the present results stipulate that tau may not be the sole driver of metabolic alterations in the cerebellar cortex. Based on these results, it can be inferred that various factors differentiating these diseases may have contributed to the differences in their metabolic imprint. Furthermore, although our analysis adjusted for age as a covariate, the observed differences in age could potentially contribute to the variance in the number of metabolic alterations identified within AD and PSP brains. The median age at death of PSP donors was lower than AD donors, and controls were older than both AD and PSP cases.

Despite the differences in the two tauopathies, common metabolic processes were altered across the two diseases. Pathway analysis of the altered metabolites in the CER region indicated a higher overlap in affected biological processes (33.34%) across the pathologies as compared to individual metabolites (16.77%). We found metabolic alterations shared between the two tauopathies in various pathways and highlighted the specifics of several selected pathways. We speculate that these similarities in the two diseases can result from either common metabolic rewiring to combat the neurodegenerative processes or common perturbation of the metabolism due to the neurodegenerative process. AD showed more unique alterations, while PSP shared most of its altered processes with AD. We found that lipid metabolism-mediated oxidative stress, lysine metabolism-associated mitochondrial dysfunction, and tau-mediated polyamine stress were shared processes between the two diseases. Meanwhile, glutathione metabolism-mediated oxidative stress and neuroinflammation, purines-linked enzymatic deregulations, and vitamin B-associated cognitive deficits were unique to AD.

Our study had several limitations. First, biological and technical variations due to the postmortem interval, i.e., the time interval between death and preservation of the brain tissue, cannot be avoided in studies on postmortem brain tissue. One way to partially account for this effect is by using this interval as a confounder in the statistical models15. However, this was not possible in the present study due to missing data on postmortem interval. Thus, the potential degradation of certain metabolites until sample preservation could not be adjusted for. Second, bulk tissue profiling as used in this study lacks resolution at the single-cell level that might be necessary to distinguish the contribution of various cell fractions. Differential proportions of cell types or differences in microenvironment could affect bulk levels of metabolites in various ways: (1) Significant differences in cell proportion could alter the levels of metabolites in the bulk sample65. (2) Neuropathology could reprogram certain cell types to produce different levels of metabolites. For instance, tau accumulation within neurons could reprogram their metabolism leading to aberrations at both the metabolome and transcriptome level9. (3) The cellular microenvironment could influence the metabolic profiles of certain cell types. For example, pathology leads to the activation of microglial cells67, which could have a different metabolic profile than a microglial cell in a steady state. In summary, one or more of these mechanisms may drive the changes in metabolic profiles at the cell type-specific level. Future studies need to focus on single-cell or spatial metabolic profiling approaches to delineate these mechanisms.

In summary, we presented the first large-scale metabolic comparison of two tauopathies - AD and PSP - focusing on two different brain regions. Our study is a step toward cataloging the common and divergent metabolic changes observed in these two diseases. Follow-up studies are needed to investigate the specific metabolic pathways commonly implicated in these diseases. Moreover, the contribution of specific cell types toward the observed metabolic changes remains to be determined. Future studies with specific cell populations, using either spatial metabolomics or iPSC-derived cell type populations, will enable a precise comparison of the metabolic changes in the two diseases at the cell type specific level.

4 Methods

4.1 Cohorts, clinical data, and neuropathological data

The Mayo Clinic cohort consisted of 162 temporal cortex (TCX) and 182 cerebellar cortex (CER) samples. Among those samples, 138 TCX and CER samples were from the same brain donor. All samples received diagnoses at autopsy following neuropathologic evaluation. Briefly, sample donors with Braak neurofibrillary tangles (NFT) stage of IV or greater received a neuropathologic diagnosis of Alzheimer’s disease (AD)23; progressive supranuclear palsy (PSP) cases were diagnosed according to NINDS neuropathologic criteria68; control samples were defined by a Braak NFT stage of III or less, CERAD neuritic and cortical plaque densities of 0 (none) or 1 (sparse), and lack of any of the following pathologic diagnoses: AD, Parkinson’s disease (PD), dementia with Lewy bodies (DLB), vascular dementia (VaD), PSP, motor neuron disease (MND), corticobasal degeneration (CBD), Pick’s disease (PiD), Huntington’s disease (HD), FTLD, hippocampal sclerosis (HipScl) or dementia lacking distinctive histology (DLDH). All 378 sample donors were of North Americans of European descent. Details on this cohort have been provided in previous studies9, 10, 69.

4.2 Metabolomics profiling

The untargeted HD4 metabolomics platform from Metabolon Inc. was used to measure brain metabolic profiles. Fractions of the tissue samples were used for two ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS; positive ionization), a UPLC-MS/MS (negative ionization), and a UPLC-MS/MS polar platform (negative ionization). Based on the spectra, area under the curve was used to quantify the peak intensities. An internal spectral database was used for compound identification. Further details can be found in a previous publication15.

4.3 Data preprocessing

A total of 827 metabolites were identified by the metabolomics platform. Of these, 658 metabolites with <25% missing values were used for the analysis, the remaining 169 were excluded. To correct for sample-wise variation, probabilistic quotient normalization was performed70, followed by log2 transformation. To impute the remaining missing values, a k-nearest-neighbor-based algorithm was applied as described previously71. The local outlier factor method was used for outlier detection72, identifying no samples to be excluded. 278 irregularly high or low single concentrations with absolute z-score above q = abs(qnorm(0.0125/n)), with n representing the number of samples, were set to missing. This formula finds the cutoff for values with less than 2.5% two-tailed probability to originate from the same normal distribution as the rest of the measurement values, after applying a Bonferroni-inspired correction factor (division by sample size). These values were then imputed using a k-nearest-neighbor-based algorithm15.

4.4 Differential analysis of metabolites

342 out of 344 samples were used for the analysis; 2 with missing APOE ε4 status were excluded. To identify metabolites associated with diagnosis, two subsequent logistic regressions were used as described previously73. The first model included diagnosis as the outcome, metabolites as the predictors, but no confounding factors. P-values of this model were corrected using the Benjamini-Hochberg (BH) method74 for multiple hypothesis testing. The second logistic regression model included diagnosis as the outcome, metabolites with adjusted p-values < 0.25 from the first model as predictors, and sex, number of APOE ε4 alleles, and age at death as confounders. Metabolites with p-value < 0.05 in the second model were used to define the metabolic signatures of the diseases. All analyses were performed using the maplet R package75.

For a selected set of metabolites i.e., from polyamine metabolism, pairwise ratios were computed (Supplementary Tables 6-7). To this end, exponentials of the log2 transformed metabolic profiles were computed, ratios were generated, and then log2 transformed. The differential analysis of these ratios was performed in the same way as individual metabolites.

Ethics approval and consent to participate

This study was approved by the appropriate Mayo Clinic Institutional Review Board. All participants or next-of-kin provided consent.

Data availability

Metabolomics as well as clinical data for the Mayo Clinic cohorts are available via the AD Knowledge Portal. The AD Knowledge Portal is a platform for accessing data, analyses, and tools generated by the Accelerating Medicines Partnership (AMP-AD) Target Discovery Program and other National Institute on Aging (NIA)-supported programs to enable open-science practices and accelerate translational learning. The data, analyses, and tools are shared early in the research cycle without a publication embargo on a secondary use. Data is available for general research use according to the following requirements for data access and data attribution (https://adknowledgeportal.org/DataAccess/Instructions). For access to the metabolomics data used in this manuscript see: https://www.synapse.org/#!Synapse:syn26446587

An interactive view of AD associations from this study can be found at https://krumsieklab.shinyapps.io/tauopathies/

All R scripts to generate the tables and figures of this paper are available at https://github.com/krumsieklab/ad-mayo-tauopathies

Author contributions

RB, JK, XW, NET, and RKD designed the study and analytic approaches. RB and JK designed the computational and statistical methods, performed the analysis, interpreted the results, and drafted the manuscript. MAllen, MArnold, and GK contributed to scientific discussions. NET, XW, MAllen provided samples and phenotypic data on the Mayo Clinic cohort, interpreted the results, edited the manuscript. CB curated and managed data. NET acquired funding, supervision and overall direction for the Mayo Clinic cohort. RKD acquired funding and is the overall PI of the Alzheimer’s disease metabolomics consortium. All authors read and reviewed the manuscript.

Competing interests

R.K-D., MArnold, GK are (through their institutions) inventors on key patents in the field of metabolomics, including applications for Alzheimer’s disease. R.K-D. holds equity in Metabolon Inc., a metabolomics technologies company. This platform was used in the current analyses. R.K-D. formed Chymia LLC and PsyProtix, a Duke University biotechnology spinout aiming to transform the treatment of mental health disorders. MArnold and GK hold equity in Chymia LLC and IP in PsyProtix. JK holds equity in Chymia LLC and IP in PsyProtix and is cofounder of iollo.

Acknowledgments

We thank the patients and families for their participation, without whom these studies would not have been possible.

This work was done as part of the National Institute of Aging’s Accelerating Medicines Partnership for AD (AMP-AD) and was supported by NIH grants 1U19AG063744, 1R01AG069901-01A1, U01AG061357, P30AG10161, P30AG72975, R01AG15819, R01AG17917, U01AG46152, U01AG61356, RF1AG058942, RF1AG059093, and U01AG061359. The results published here are in whole or in part based on data obtained from the AD Knowledge Portal (https://adknowledgeportal.org).

The Mayo Clinic samples are part of the RNAseq study data led by Dr. Nilüfer Ertekin-Taner, Mayo Clinic, Jacksonville, FL as part of the multi-PI U01 AG046139 (MPIs Golde, Ertekin-Taner, Younkin, Price). Samples were provided from the following sources: The Mayo Clinic Brain Bank. Data collection was supported through funding by NIA grants P50 AG016574, R01 AG032990, U01 AG046139, R01 AG018023, U01 AG006576, U01 AG006786, R01 AG025711, R01 AG017216, R01 AG003949, NINDS grant R01 NS080820, CurePSP Foundation, and support from Mayo Foundation. Study data includes samples collected through the Sun Health Research Institute Brain and Body Donation Program of Sun City, Arizona. The Brain and Body Donation Program is supported by the National Institute of Neurological Disorders and Stroke (U24 NS072026 National Brain and Tissue Resource for Parkinson’s Disease and Related Disorders), the National Institute on Aging (P30 AG19610 Arizona Alzheimers Disease Core Center), the Arizona Department of Health Services (contract 211002, Arizona Alzheimers Research Center), the Arizona Biomedical Research Commission (contracts 4001, 0011, 05-901 and 1001 to the Arizona Parkinson’s Disease Consortium) and the Michael J. Fox Foundation for Parkinson’s Research.

RB is also supported by Alzheimer’s association award AARFD-22-974775. RB thanks her colleagues from the Krumsiek lab for fruitful discussions and support in this work. NET, XW and MA are supported by R01AG061796, U01AG046139 and U19AG074879. NET is also supported by the Alzheimer’s Association Zenith Award.

Footnotes

References

  1. 1.↵
    Uversky VN. Intrinsically disordered proteins and their (disordered) proteomes in neurodegenerative disorders. Front Aging Neurosci. 2015;7(MAR). doi:10.3389/fnagi.2015.00018
    OpenUrlCrossRefPubMed
  2. 2.↵
    Knopman DS, Amieva H, Petersen RC, et al. Alzheimer disease. Nat Rev Dis Primer. 2021;7(1):33. doi:10.1038/s41572-021-00269-y
    OpenUrlCrossRefPubMed
  3. 3.↵
    Kurz C, Ebersbach G, Respondek G, Giese A, Arzberger T, Höglinger GU. An autopsy-confirmed case of progressive supranuclear palsy with predominant postural instability. Acta Neuropathol Commun. 2016;4:120. doi:10.1186/s40478-016-0391-7
    OpenUrlCrossRef
  4. 4.↵
    Wagshal D, Sankaranarayanan S, Guss V, et al. Divergent CSF τ alterations in two common tauopathies: Alzheimer’s disease and progressive supranuclear palsy. J Neurol Neurosurg Psychiatry. 2015;86(3):244–250. doi:10.1136/jnnp-2014-308004
    OpenUrlAbstract/FREE Full Text
  5. 5.↵
    Uchihara T. Neurofibrillary changes undergoing morphological and biochemical changes - How does tau with the profile shift of from four repeat to three repeat spread in Alzheimer brain? Neuropathol Off J Jpn Soc Neuropathol. 2020;40(5):450–459. doi:10.1111/neup.12669
    OpenUrlCrossRef
  6. 6.↵
    Allen M, Kachadoorian M, Quicksall Z, et al. Association of MAPT haplotypes with Alzheimer’s disease risk and MAPT brain gene expression levels. Alzheimers Res Ther. 2014;6(4):39. doi:10.1186/alzrt268
    OpenUrlCrossRefPubMed
  7. 7.↵
    Chen JA, Chen Z, Won H, et al. Joint genome-wide association study of progressive supranuclear palsy identifies novel susceptibility loci and genetic correlation to neurodegenerative diseases. Mol Neurodegener. 2018;13(1):41. doi:10.1186/s13024-018-0270-8
    OpenUrlCrossRef
  8. 8.↵
    Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat Genet. 2022;54(4):412–436. doi:10.1038/s41588-022-01024-z
    OpenUrlCrossRefPubMed
  9. 9.↵
    Allen M, Wang X, Burgess JD, et al. Conserved brain myelination networks are altered in Alzheimer’s and other neurodegenerative diseases. Alzheimers Dement. 2018;14(3):352–366. doi:10.1016/j.jalz.2017.09.012
    OpenUrlCrossRefPubMed
  10. 10.↵
    Wang X, Allen M, İş Ö, et al. Alzheimer’s disease and progressive supranuclear palsy share similar transcriptomic changes in distinct brain regions. J Clin Invest. 2022;132(2). doi:10.1172/JCI149904
    OpenUrlCrossRef
  11. 11.↵
    Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases | Nature. Accessed August 18, 2022. https://www.nature.com/articles/nature05292
  12. 12.↵
    Nho K, Kueider-Paisley A, MahmoudianDehkordi S, et al. Altered bile acid profile in mild cognitive impairment and Alzheimer’s disease: Relationship to neuroimaging and CSF biomarkers. Alzheimers Dement. 2019;15(2):232–244. doi:10.1016/j.jalz.2018.08.012
    OpenUrlCrossRef
  13. 13.↵
    Kaddurah-Daouk R, Zhu H, Sharma S, et al. Alterations in metabolic pathways and networks in Alzheimer’s disease. Transl Psychiatry. 2013;3:e244. doi:10.1038/tp.2013.18
    OpenUrlCrossRef
  14. 14.
    Arnold M, Nho K, Kueider-Paisley A, et al. Sex and APOE ε4 genotype modify the Alzheimer’s disease serum metabolome. Nat Commun. 2020;11(1). doi:10.1038/s41467-020-14959-w
    OpenUrlCrossRefPubMed
  15. 15.↵
    Batra R, Arnold M, Wörheide MA, et al. The landscape of metabolic brain alterations in Alzheimer’s disease. Alzheimers Dement. Published online July 13, 2022. doi:10.1002/ALZ.12714
    OpenUrlCrossRef
  16. 16.
    Toledo JB, Arnold M, Kastenmüller G, et al. Metabolic network failures in Alzheimer’s disease: A biochemical road map. Alzheimers Dement. 2017;13(9):965–984. doi:10.1016/j.jalz.2017.01.020
    OpenUrlCrossRefPubMed
  17. 17.
    Mahajan UV, Varma VR, Griswold ME, et al. Dysregulation of multiple metabolic networks related to brain transmethylation and polyamine pathways in Alzheimer disease: A targeted metabolomic and transcriptomic study. PLoS Med. 2020;17(1):e1003012. doi:10.1371/JOURNAL.PMED.1003012
    OpenUrlCrossRef
  18. 18.↵
    MahmoudianDehkordi S, Arnold M, Nho K, et al. Altered bile acid profile associates with cognitive impairment in Alzheimer’s disease-An emerging role for gut microbiome. Alzheimers Dement. 2019;15(1):76–92. doi:10.1016/j.jalz.2018.07.217
    OpenUrlCrossRefPubMed
  19. 19.↵
    Mori A, Ishikawa KI, Saiki S, et al. Plasma metabolite biomarkers for multiple system atrophy and progressive supranuclear palsy. PloS One. 2019;14(9):e0223113. doi:10.1371/journal.pone.0223113
    OpenUrlCrossRefPubMed
  20. 20.↵
    Takahashi K, Iwaoka K, Takahashi K, et al. Cerebrospinal fluid levels of oxidative stress measured using diacron-reactive oxygen metabolites and biological antioxidant potential in patients with Parkinson’s disease and progressive supranuclear palsy. Neurosci Lett. 2021;757:135975. doi:10.1016/j.neulet.2021.135975
    OpenUrlCrossRef
  21. 21.↵
    Wilson DM, Cookson MR, Van Den Bosch L, Zetterberg H, Holtzman DM, Dewachter I. Hallmarks of neurodegenerative diseases. Cell. 2023;186(4):693–714. doi:10.1016/j.cell.2022.12.032
    OpenUrlCrossRef
  22. 22.↵
    Roemer SF, Grinberg LT, Crary JF, et al. Rainwater Charitable Foundation criteria for the neuropathologic diagnosis of progressive supranuclear palsy. Acta Neuropathol (Berl). 2022;144(4):603–614. doi:10.1007/s00401-022-02479-4
    OpenUrlCrossRef
  23. 23.↵
    Braak H, Braak E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol (Berl). 1991;82(4):239–259. doi:10.1007/BF00308809
    OpenUrlCrossRefPubMedWeb of Science
  24. 24.↵
    Kovacs GG, Lukic MJ, Irwin DJ, et al. Distribution patterns of tau pathology in progressive supranuclear palsy. Acta Neuropathol (Berl). 2020;140(2):99–119. doi:10.1007/s00401-020-02158-2
    OpenUrlCrossRef
  25. 25.↵
    Hampel H, Hardy J, Blennow K, et al. The Amyloid-β Pathway in Alzheimer’s Disease. Mol Psychiatry. 2021;26(10):5481–5503. doi:10.1038/s41380-021-01249-0
    OpenUrlCrossRef
  26. 26.↵
    Larner AJ. The cerebellum in Alzheimer’s disease. Dement Geriatr Cogn Disord. 1997;8(4):203–209. doi:10.1159/000106632
    OpenUrlCrossRefPubMedWeb of Science
  27. 27.↵
    Burciu RG, Ofori E, Shukla P, et al. Distinct patterns of brain activity in progressive supranuclear palsy and Parkinson’s disease. Mov Disord Off J Mov Disord Soc. 2015;30(9):1248–1258. doi:10.1002/mds.26294
    OpenUrlCrossRef
  28. 28.↵
    Klietz M, Mahmoudi N, Maudsley AA, et al. Whole-Brain Magnetic Resonance Spectroscopy Reveals Distinct Alterations in Neurometabolic Profile in Progressive Supranuclear Palsy. Mov Disord. n/a(n/a). doi:10.1002/mds.29456
    OpenUrlCrossRef
  29. 29.↵
    Leng F, Edison P. Neuroinflammation and microglial activation in Alzheimer disease: where do we go from here? Nat Rev Neurol. 2021;17(3):157–172. doi:10.1038/s41582-020-00435-y
    OpenUrlCrossRefPubMed
  30. 30.↵
    Rose S, Melnyk S, Pavliv O, et al. Evidence of oxidative damage and inflammation associated with low glutathione redox status in the autism brain. Transl Psychiatry. 2012;2(7):e134. doi:10.1038/tp.2012.61
    OpenUrlCrossRef
  31. 31.↵
    Dello SAWG, Neis EPJG, de Jong MC, et al. Systematic review of ophthalmate as a novel biomarker of hepatic glutathione depletion. Clin Nutr. 2013;32(3):325–330. doi:10.1016/j.clnu.2012.10.008
    OpenUrlCrossRefPubMed
  32. 32.
    Chen JJ, Thiyagarajah M, Song J, et al. Altered central and blood glutathione in Alzheimer’s disease and mild cognitive impairment: a meta-analysis. Alzheimers Res Ther. 2022;14(1):1–17. doi:10.1186/S13195-022-00961-5/TABLES/4
    OpenUrlCrossRef
  33. 33.↵
    Dalleau S, Baradat M, Guéraud F, Huc L. Cell death and diseases related to oxidative stress:4-hydroxynonenal (HNE) in the balance. Cell Death Differ. 2013;20(12):1615–1630. doi:10.1038/cdd.2013.138
    OpenUrlCrossRefPubMed
  34. 34.↵
    Allen M, Zou F, Chai HS, et al. Glutathione S-transferase omega genes in Alzheimer and Parkinson disease risk, age-at-diagnosis and brain gene expression: An association study with mechanistic implications. Mol Neurodegener. 2012;7(1). doi:10.1186/1750-1326-7-13
    OpenUrlCrossRefPubMed
  35. 35.↵
    Ansoleaga B, Jové M, Schlüter A, et al. Deregulation of purine metabolism in Alzheimer’s disease. Neurobiol Aging. 2015;36(1):68–80. doi:10.1016/j.neurobiolaging.2014.08.004
    OpenUrlCrossRefPubMed
  36. 36.↵
    Kaddurah-Daouk R, Rozen S, Matson W, et al. Metabolomic changes in autopsy-confirmed Alzheimer’s disease. Alzheimers Dement. 2011;7(3):309–317. doi:10.1016/j.jalz.2010.06.001
    OpenUrlCrossRefPubMed
  37. 37.↵
    Morris MC, Schneider JA, Tangney CC. Thoughts on B-vitamins and dementia. J Alzheimers Dis JAD. 2006;9(4):429–433.
    OpenUrl
  38. 38.↵
    Gibson GE, Hirsch JA, Fonzetti P, Jordon BD, Cirio RT, Elder J. Vitamin B1 (thiamine) and dementia. Ann N Y Acad Sci. 2016;1367(1):21–30. doi:10.1111/nyas.13031
    OpenUrlCrossRefPubMed
  39. 39.↵
    Gibson GE, Sheu KF, Blass JP, et al. Reduced activities of thiamine-dependent enzymes in the brains and peripheral tissues of patients with Alzheimer’s disease. Arch Neurol. 1988;45(8):836–840. doi:10.1001/archneur.1988.00520320022009
    OpenUrlCrossRefPubMedWeb of Science
  40. 40.↵
    Porter K, Hoey L, Hughes CF, Ward M, McNulty H. Causes, Consequences and Public Health Implications of Low B-Vitamin Status in Ageing. Nutrients. 2016;8(11):725. doi:10.3390/nu8110725
    OpenUrlCrossRef
  41. 41.↵
    An Y, Feng L, Zhang X, et al. Dietary intakes and biomarker patterns of folate, vitamin B6, and vitamin B12 can be associated with cognitive impairment by hypermethylation of redox-related genes NUDT15 and TXNRD1. Clin Epigenetics. 2019;11(1):139. doi:10.1186/s13148-019-0741-y
    OpenUrlCrossRef
  42. 42.↵
    Gómez-Ramos A, Díaz-Nido J, Smith MA, Perry G, Avila J. Effect of the lipid peroxidation product acrolein on tau phosphorylation in neural cells. J Neurosci Res. 2003;71(6):863–870. doi:10.1002/jnr.10525
    OpenUrlCrossRefPubMed
  43. 43.↵
    Adibhatla RM, Hatcher JF. Role Of Lipids In Brain Injury And Diseases. Future Lipidol. 2007;2(4):403–422. doi:10.2217/17460875.2.4.403
    OpenUrlCrossRefPubMedWeb of Science
  44. 44.↵
  45. 45.↵
    Hermann PM, Watson SN, Wildering WC. Phospholipase A2 – nexus of aging, oxidative stress, neuronal excitability, and functional decline of the aging nervous system? Insights from a snail model system of neuronal aging and age-associated memory impairment. Front Genet. 2014;5. Accessed September 1, 2022. https://www.frontiersin.org/articles/10.3389/fgene.2014.00419
  46. 46.↵
    Niedzielska E, Smaga I, Gawlik M, et al. Oxidative Stress in Neurodegenerative Diseases. Mol Neurobiol. 2016;53(6):4094–4125. doi:10.1007/s12035-015-9337-5
    OpenUrlCrossRef
  47. 47.↵
    Dias C, Nylandsted J. Plasma membrane integrity in health and disease: significance and therapeutic potential. Cell Discov. 2021;7(1). doi:10.1038/s41421-020-00233-2
    OpenUrlCrossRefPubMed
  48. 48.↵
    Klein J. Membrane breakdown in acute and chronic neurodegeneration: Focus on choline-containing phospholipids. J Neural Transm. 2000;107(8-9):1027–1063. doi:10.1007/s007020070051
    OpenUrlCrossRefPubMedWeb of Science
  49. 49.↵
    Cutler RG, Kelly J, Storie K, et al. Involvement of oxidative stress-induced abnormalities in ceramide and cholesterol metabolism in brain aging and Alzheimer’s disease. Proc Natl Acad Sci U S A. 2004;101(7):2070–2075. doi:10.1073/pnas.0305799101
    OpenUrlAbstract/FREE Full Text
  50. 50.↵
    Pant DC, Aguilera-Albesa S, Pujol A. Ceramide signalling in inherited and multifactorial brain metabolic diseases. Neurobiol Dis. 2020;143:105014. doi:10.1016/j.nbd.2020.105014
    OpenUrlCrossRef
  51. 51.↵
    Wang X, Allen M, Li S, et al. Deciphering cellular transcriptional alterations in Alzheimer’s disease brains. Mol Neurodegener. 2020;15(1):38. doi:10.1186/s13024-020-00392-6
    OpenUrlCrossRef
  52. 52.↵
    Hallen A, Jamie JF, Cooper AJL. Lysine metabolism in mammalian brain: an update on the importance of recent discoveries. Amino Acids. 2013;45(6):10.1007/s00726-013-1590-1591. doi:10.1007/s00726-013-1590-1
    OpenUrlCrossRefPubMed
  53. 53.↵
    Zhou J, Wang X, Wang M, et al. The lysine catabolite saccharopine impairs development by disrupting mitochondrial homeostasis. J Cell Biol. 2019;218(2):580–597. doi:10.1083/jcb.201807204
    OpenUrlAbstract/FREE Full Text
  54. 54.↵
    Pérez MJ, Jara C, Quintanilla RA. Contribution of Tau Pathology to Mitochondrial Impairment in Neurodegeneration. Front Neurosci. 2018;12:441. doi:10.3389/fnins.2018.00441
    OpenUrlCrossRefPubMed
  55. 55.↵
    Yurdagul A, Subramanian M, Wang X, et al. Macrophage Metabolism of Apoptotic Cell-Derived Arginine Promotes Continual Efferocytosis and Resolution of Injury. Cell Metab. 2020;31(3):518–533.e10. doi:10.1016/j.cmet.2020.01.001
    OpenUrlCrossRef
  56. 56.↵
    Freitag K, Sterczyk N, Wendlinger S, et al. Spermidine reduces neuroinflammation and soluble amyloid beta in an Alzheimer’s disease mouse model. J Neuroinflammation. 2022;19(1):172. doi:10.1186/s12974-022-02534-7
    OpenUrlCrossRef
  57. 57.↵
    Boada-Romero E, Martinez J, Heckmann BL, Green DR. The clearance of dead cells by efferocytosis. Nat Rev Mol Cell Biol. 2020;21(7):398–414. doi:10.1038/s41580-020-0232-1
    OpenUrlCrossRef
  58. 58.↵
    Sandusky-Beltran LA, Kovalenko A, Placides DS, et al. Aberrant AZIN2 and polyamine metabolism precipitates tau neuropathology. J Clin Invest. 2021;131(4). doi:10.1172/JCI126299
    OpenUrlCrossRef
  59. 59.↵
    Sandusky-Beltran LA, Kovalenko A, Ma C, et al. Spermidine/spermine-N1-acetyltransferase ablation impacts tauopathy-induced polyamine stress response. Alzheimers Res Ther. 2019;11:58. doi:10.1186/s13195-019-0507-y
    OpenUrlCrossRef
  60. 60.↵
    Gilad GM, Gilad VH. Overview of the Brain Polyamine-Stress-Response: Regulation, Development, and Modulation by Lithium and Role in Cell Survival. Cell Mol Neurobiol. Published online 2003:13.
  61. 61.↵
    Hayashi Y, Tanaka J, Morizumi Y, Kitamura Y, Hattori Y. Polyamine levels in brain and plasma after acute restraint or water-immersion restraint stress in mice. Neurosci Lett. 2004;355(1):57–60. doi:10.1016/j.neulet.2003.10.027
    OpenUrlCrossRefPubMed
  62. 62.↵
    Altmaier E, Ramsay SL, Graber A, Mewes HW, Weinberger KM, Suhre K. Bioinformatics Analysis of Targeted Metabolomics—Uncovering Old and New Tales of Diabetic Mice under Medication. Endocrinology. 2008;149(7):3478–3489. doi:10.1210/en.2007-1747
    OpenUrlCrossRefPubMedWeb of Science
  63. 63.↵
    Zhou J, Seeley WW. Network dysfunction in Alzheimer’s disease and frontotemporal dementia: implications for psychiatry. Biol Psychiatry. 2014;75(7):565–573. doi:10.1016/j.biopsych.2014.01.020
    OpenUrlCrossRefPubMedWeb of Science
  64. 64.↵
    Tomasi D, Wang GJ, Volkow ND. Energetic cost of brain functional connectivity. Proc Natl Acad Sci U S A. 2013;110(33):13642–13647. doi:10.1073/pnas.1303346110
    OpenUrlAbstract/FREE Full Text
  65. 65.↵
    Kim T, Lim CS, Kaang BK. Cell type-specific gene expression profiling in brain tissue: comparison between TRAP, LCM and RNA-seq. BMB Rep. 2015;48(7):388–394. doi:10.5483/BMBRep.2015.48.7.218
    OpenUrlCrossRef
  66. 66.
    Chung D eun C, Roemer S, Petrucelli L, Dickson DW. Cellular and pathological heterogeneity of primary tauopathies. Mol Neurodegener. 2021;16(1):57. doi:10.1186/s13024-021-00476-x
    OpenUrlCrossRef
  67. 67.↵
    Streit WJ, Xue QS, Tischer J, Bechmann I. Microglial pathology. Acta Neuropathol Commun. 2014;2(1):142. doi:10.1186/s40478-014-0142-6
    OpenUrlCrossRefPubMed
  68. 68.↵
    Hauw JJ, Daniel SE, Dickson D, et al. Preliminary NINDS neuropathologic criteria for Steele-Richardson-Olszewski syndrome (progressive supranuclear palsy). Neurology. 1994;44(11):2015–2015. doi:10.1212/WNL.44.11.2015
    OpenUrlCrossRefPubMed
  69. 69.↵
    Allen M, Carrasquillo MM, Funk C, et al. Human whole genome genotype and transcriptome data for Alzheimer’s and other neurodegenerative diseases. Sci Data. 2016;3:160089. doi:10.1038/sdata.2016.89
    OpenUrlCrossRefPubMed
  70. 70.↵
    Dieterle F, Ross A, Schlotterbeck G, Senn H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in1H NMR metabonomics. Anal Chem. 2006;78(13):4281-4290. doi:10.1021/ac051632c
    OpenUrlCrossRefPubMed
  71. 71.↵
    Do KT, Wahl S, Raffler J, et al. Characterization of missing values in untargeted MS-based metabolomics data and evaluation of missing data handling strategies. Metabolomics. 2018;14(10):128. doi:10.1007/s11306-018-1420-2
    OpenUrlCrossRef
  72. 72.↵
    Breunig MM, Kriegel HP, Ng RT, Sander J. LOF. Proc 2000 ACM SIGMOD Int Conf Manag Data - SIGMOD 00. Published online 2000:93–104. doi:10.1145/342009.335388
    OpenUrlCrossRef
  73. 73.↵
    Zehetmayer S, Posch M. False discovery rate control in two-stage designs. BMC Bioinformatics. 2012;13:81. doi:10.1186/1471-2105-13-81
    OpenUrlCrossRef
  74. 74.↵
    Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Ser B Methodol. 1995;57(1):289–300. doi:10.1111/J.2517-6161.1995.TB02031.X
    OpenUrlCrossRef
  75. 75.↵
    Chetnik K, Benedetti E, Gomari DP, et al. mapletL: an extensible R toolbox for modular and reproducible metabolomics pipelines. Bioinformatics. 2022;38(4):1168–1170. doi:10.1093/bioinformatics/btab741
    OpenUrlCrossRef
Back to top
Posted July 27, 2023.
Download PDF

Supplementary Material

Data/Code
Email
Comparative brain metabolomics reveals shared and distinct metabolic alterations in Alzheimer’s disease and progressive supranuclear palsy
Richa Batra, Jan Krumsiek, Xue Wang, Mariet Allen, Colette Blach, Gabi Kastenmüller, Matthias Arnold, Nilüfer Ertekin-Taner, Rima F. Kaddurah-Daouk, Alzheimer’s Disease Metabolomics Consortium (ADMC)
medRxiv 2023.07.25.23293055; doi: https://doi.org/10.1101/2023.07.25.23293055
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area