Transcriptomic Evidence for Alterations in Astrocytes and Parvalbumin Interneurons in Subjects With Bipolar Disorder and Schizophrenia

Toker, Lilah; Mancarci, B. Ogan; Tripathy, Shreejoy J.; Pavlidis, Paul

doi:10.1016/j.biopsych.2018.07.010

Cited by 99 publications

(101 citation statements)

References 87 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Second, if cellular composition is of interest, it would be reasonable to analyze composition more directly by inspecting the expression of known markers rather than by using indirect means via clustering and enrichment analysis. This parallels the situation for analysis of differential expression, where changes in measured expression levels can be due to changes in composition (Mancarci et al, 2017;Toker et al, 2018). On the other hand, machine learning applications of coexpression to tasks such as gene function prediction are not directly affected by our findings, as success in prediction does not necessarily depend on the biological meaning of the features used.…”

Section: Discussionsupporting

confidence: 55%

Untangling the effects of cellular composition on coexpression analysis

Farahbod

Pavlidis

2019

Preprint

Self Cite

View full text Add to dashboard Cite

Background: Coexpression analysis is one of the most widely used methods in genomics, with applications to inferring regulatory networks, predicting gene function, and interpretation of transcriptome profiling studies. Most studies use data collected from bulk tissue, where the effects of cellular composition present a potential confound. However, the impact of composition on coexpression analysis have not been studied in detail. Here we examine this issue for the case of human brain RNA analysis.Results: We found that for most genes, differences in expression levels across cell types account for a large fraction of the variance of their measured RNA levels in brain (median R 2 = 0.64). We then show that genes that have similar expression patterns across cell types will have correlated RNA levels in bulk tissue, due to the effect of variation in cellular composition. We demonstrate that much of the coexpression in the bulk tissue can be attributed to this effect. We further show how this compositioninduced coexpression masks underlying intra-cell-type coexpression observed in single-cell data.Attempt to correct for composition yielded mixed results. Conclusions:The dominant coexpression signal in brain can be attributed to cellular compositional effects, rather than intra-cell-type regulatory relationships, and this is likely to be true for other tissues.These results have important implications for the relevance and interpretation of coexpression in many applications.

show abstract

Section: Discussionsupporting

confidence: 55%

Untangling the effects of cellular composition on coexpression analysis

Farahbod

Pavlidis

2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Positively-weighted SRM genes were enriched for neuronal affiliation (45) (permutation test, P < 10 -4 ) and, more specifically, for genes differentially expressed in fast-spiking parvalbumin-positive inhibitory interneurons (46) ( Table S2; permutation test, FDR-corrected P < 0.01). Negatively-weighted SRM genes were enriched for astrocytes, microglia and neuronal affiliation (45,46) (permutation tests, all P < 10 -4 ) (Figure 2B).…”

Section: Functional and Schizophrenia-related Enrichment Of Schizotypmentioning

confidence: 97%

“…We assigned a cellular affiliation score to each gene in the SRM gene list according to prior criteria for four cell types: neuron, astrocyte, microglia or oligodendroglia (45); and for a more fine-grained set of cell types (46) (Table S2). We used a data resampling procedure to test the null hypothesis that SRM genes were randomly assigned to different cell types.…”

Section: Enrichment Analysismentioning

confidence: 99%

“…The SRM gene list was tested for enrichment by genes characteristic of specific cell types using two sets of prior criteria (45,46). Positively-weighted SRM genes were enriched for neuronal affiliation (45) (permutation test, P < 10 -4 ) and, more specifically, for genes differentially expressed in fast-spiking parvalbumin-positive inhibitory interneurons (46) ( Table S2; permutation test, FDR-corrected P < 0.01).…”

Section: Functional and Schizophrenia-related Enrichment Of Schizotypmentioning

confidence: 99%

See 1 more Smart Citation

Schizotypy-related magnetization of cortex in healthy adolescence is co-located with expression of schizophrenia-related genes

Romero-García

Seidlitz

Whitaker

et al. 2018

Preprint

View full text Add to dashboard Cite

= 249 words (max 250 words) Main text = 3997 words (max 4000 words). References = 72 Figures = 4 Table = 1 Intra-cortical magnetization and patterning of schizophrenia-related genes 2 Abstract Background Genetic risk is thought to drive clinical variation on a spectrum of schizophrenia-like traits but the underlying changes in brain structure that mechanistically link genomic variation to schizotypal experience and behaviour are unclear. MethodsWe assessed schizotypy using a self-reported questionnaire, and measured magnetization transfer (MT), as a putative micro-structural MRI marker of intra-cortical myelination, in 68 brain regions, in 248 healthy young people (aged 14-25 years). We used normative adult brain gene expression data, and partial least squares (PLS) analysis, to find the weighted gene expression pattern that was most co-located with the cortical map of schizotypy-related magnetization (SRM). ResultsMagnetization was significantly correlated with schizotypy in bilateral posterior cingulate cortex and precuneus (and for disorganized schizotypy also in medial prefrontal cortex; all FDR-corrected P < 0.05), which are regions of the default mode network specialized for social and memory functions. The genes most positively weighted on the whole genome expression map co-located with SRM were enriched for genes that were significantly downregulated in two prior case-control histological studies of brain gene expression in schizophrenia. Conversely, the most negatively weighted genes were enriched for genes that were transcriptionally up-regulated in schizophrenia. Positively weighted (downregulated) genes were enriched for neuronal, specifically inter-neuronal, affiliations and coded a network of proteins comprising a few highly interactive "hubs" such as parvalbumin and calmodulin. ConclusionsMicrostructural MRI maps of intracortical magnetization can be linked to both the behavioural traits of schizotypy and to prior histological data on dysregulated gene expression in schizophrenia. previous version of this paper was posted on bioRxiv: https://www.biorxiv.org/content/10.1101/487108v1 DATA and CODE SHARING DataRegional magnetisation transfer (MT) for 68 cortical regions, schizotypy scores, age, gender, socio-economic status, scanning site, total brain volume and Euler values, for N=248, are available at: https://github.com/RafaelRomeroGarcia/Schizotypy_MT_geneExp. CodeCortical parcellation of gene expression maps to estimate regional mean gene expression (62): (https://github.com/RafaelRomeroGarcia/geneExpression_Repository) Generate null-models that preserve the spatial contiguity of cortical regions for permutation testing (31): https://github.com/frantisekvasa/rotate_parcellation PLS analysis and bootstrapping to estimate PLS weights (15): https://github.com/KirstieJane/NSPN_WhitakerVertes_PNAS2016/tree/master/SCRIPTS Generate Figure S1 from raw Gandal (5) and Fromer (49) datasets: https://github.com/RafaelRomeroGarcia/Schizotypy_MT_geneExpMapping regional values to the cortical surface for visuali...

show abstract

“…The observed gene expression profiles in bulk brain tissue can be dramatically influenced by differences in cellular composition. Such differences can be a result of variation in gray/white matter ratios introduced during tissue extraction, inter-subject variability or represent disease related alterations [14][15][16]. To study the contribution of various technical and biological sources of variation in our dataset we first estimated marker gene profiles (MGPs) for the major classes of cortical cell types (astrocytes, microglia, oligodendrocytes, and neurons) in our samples by summarizing the expression of the cell type-specific marker genes as previously described [15].…”

Section: Cell Composition Is a Major Confounder Of Gene Expression Inmentioning

confidence: 99%

Common gene expression signatures in Parkinson’s disease are driven by changes in cell composition

Nido

Dick

Toker

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

AbstractBackgroundThe etiology of Parkinson’s disease (PD) is largely unknown. Genome-wide transcriptomic studies in bulk brain tissue have identified several molecular signatures associated with the disease. While these studies have the potential to shed light into the pathogenesis of PD, they are also limited by two major confounders: RNA post mortem degradation and heterogeneous cell type composition of bulk tissue samples. We performed RNA sequencing following ribosomal RNA depletion in the prefrontal cortex of 49 individuals from two independent case-control cohorts. Using cell-type specific markers, we estimated the cell-type composition for each sample and included this in our analysis models to compensate for the variation in cell-type proportions.ResultsRibosomal RNA depletion results in substantially more even transcript coverage, compared to poly(A) capture, in post mortem tissue. Moreover, we show that cell-type composition is a major confounder of differential gene expression analysis in the PD brain. Correcting for cell-type proportions attenuates numerous transcriptomic signatures that have been previously associated with PD, including vesicle trafficking, synaptic transmission, immune and mitochondrial function. Conversely, pathways related to endoplasmic reticulum, lipid oxidation and unfolded protein response are strengthened and surface as the top differential gene expression signatures in the PD prefrontal cortex.ConclusionsDifferential gene expression signatures in PD bulk brain tissue are significantly confounded by underlying differences in cell-type composition. Modeling cell-type heterogeneity is crucial in order to unveil transcriptomic signatures that represent regulatory changes in the PD brain and are, therefore, more likely to be associated with underlying disease mechanisms.

show abstract

Transcriptomic Evidence for Alterations in Astrocytes and Parvalbumin Interneurons in Subjects With Bipolar Disorder and Schizophrenia

Cited by 99 publications

References 87 publications

Untangling the effects of cellular composition on coexpression analysis

Untangling the effects of cellular composition on coexpression analysis

Schizotypy-related magnetization of cortex in healthy adolescence is co-located with expression of schizophrenia-related genes

Common gene expression signatures in Parkinson’s disease are driven by changes in cell composition

Contact Info

Product

Resources

About