Dissection of artifactual and confounding glial signatures by single-cell sequencing of mouse and human brain

In the paper

“Previously, we optimized a cold mechanical dissociation protocol to isolate microglia for scRNA-seq with minimal ex vivo transcriptional alterations15.”

What the sources say

• Supported

  The Results section, “Isolation of microglia to minimize ex vivo activation,” says the authors generated single-cell suspensions under consistently cold conditions to minimize ex vivo activation and transcriptional activity. The STAR Methods describe cold mincing and Dounce homogenization, followed by purification of microglia for sequencing. No page number is provided in our copy. (LLM, conf=0.90)

  cites [15] Hammond, T. R. et al. Single-cell RNA sequencing of microglia throughout the mouse lifespan and in the injured brain reveals complex cell-state changes. Immunity 50, 253–271.e6 (2019). source

In the paper

“As the tissue-resident macrophages of the brain, microglia are highly sensitive to perturbations in their environment13,14.”

What the sources say

• Supported

  The Abstract reports rapid and extensive changes in microglia-specific gene expression after transfer from brain tissue to culture. The Discussion likewise describes rapid, profound changes within six hours, supporting the claim that microglia are highly sensitive to changes in their environment. (LLM, conf=0.93)

  cites [13] Gosselin, D. et al. An environment-dependent transcriptional network specifies human microglia identity. Science 356, (2017). source
• Supported

  The Discussion (“Microglial fate is continuously instructed by CNS-specific cues”) says microglial signature genes change extremely rapidly after isolation and that microglia quickly accommodate different tissue contexts. The Results (“Microglial specification requires sustained CNS instruction”) report a marked decline in signature-gene expression within four hours of isolation, supporting sensitivity to environmental change. Page numbers are not provided in our copy. (LLM, conf=0.93)

  cites [14] Bohlen, C. J. et al. Diverse requirements for microglial survival, specification, and function revealed by defined-medium cultures. Neuron 94, 759–773.e8 (2017). source

In the paper

“In line with our previous work15, we find that adult microglia are largely homogenous.”

What the sources say

• Supported

  In the Results section on microglial states across development (page not provided), the paper reports considerably less diversity in adult (P100) microglia than in early development. It also says adult microglia were largely assigned to three central clusters without unique defining genes, supporting the characterization “largely homogenous,” rather than completely uniform. (LLM, conf=0.90)

  cites [15] Hammond, T. R. et al. Single-cell RNA sequencing of microglia throughout the mouse lifespan and in the injured brain reveals complex cell-state changes. Immunity 50, 253–271.e6 (2019). source

In the paper

“Of note, we observed substantial loss of some cell-surface receptors following enzymatic digestion, in line with previous work in peripheral immunology (Supplementary Fig. 1d–f and Supplementary Note 1)3,5–9.”

What the sources say

• ?Not checked — no full text of the sourcecites [3] Abuzakouk, M., Feighery, C. & O’Farrelly, C. Collagenase and Dispase enzymes disrupt lymphocyte surface molecules. J. Immunol. Methods 194, 211–216 (1996). source
• ?Not checked — no full text of the sourcecites [5] Grange, C. et al. Phenotypic characterization and functional analysis of human tumor immune infiltration after mechanical and enzymatic disaggregation. J. Immunol. Methods 372, 119–126 (2011). source
• Supported

  In “Enzymatic digestion affects expression level of surface molecules on immune cells” (page not provided), the paper reports dramatic decreases in several surface markers after dispase digestion of splenocytes. It also notes that collagenase had only minor effects on most markers, so the finding applies particularly to dispase. (LLM, conf=0.93)

  cites [6] Autengruber, A., Gereke, M., Hansen, G., Hennig, C. & Bruder, D. Impact of enzymatic tissue disintegration on the level of surface molecule expression and immune cell function. Eur. J. Microbiol. Immunol. (Bp) 2, 112–120 (2012). source
• Supported

  The Results section, “Effect of collagenase and skin culture on expression of surface markers and pathogen-binding receptors,” reports that trypsin cleaved several surface markers and pathogen-binding receptors, while collagenase blends cleaved other receptors. The Discussion describes cleavage of many CLRs and Siglecs by type IV collagenase as dramatic, supporting the claim that enzymatic digestion can cause substantial loss of some cell-surface receptors. (LLM, conf=0.94)

  cites [7] Botting, R. A. et al. Phenotypic and functional consequences of different isolation protocols on skin mononuclear phagocytes. J. Leukoc. Biol. 101, 1393–1403 (2017). source
• Supported

  Section 3.3 reports reduced CD4 expression after Liberase TL digestion, to the point that the CD4+ T-cell subset could no longer be identified. It also reports reduced CD28 expression on PBMC CD4+ T cells after Liberase TL treatment. The effect was marker- and enzyme-specific; many other markers were unaffected. (LLM, conf=0.88)

  cites [8] Schreurs, R. R. C. E. et al. Quantitative comparison of human intestinal mononuclear leukocyte isolation techniques for flow cytometric analyses. J. Immunol. Methods 445, 45–52 (2017). source
• Partly supported

  The Tissue Disaggregation > Enzymes section says dispase can cleave specific cell-surface molecules or antigens relevant to T-cell analysis. The Enzymatic and Mechanical Dissociation section also warns that strong or concentrated enzymes may compromise surface markers. These passages support loss of some markers, but do not establish that the loss is substantial or occurs with enzymatic digestion generally. (LLM, conf=0.84)

  cites [9] Reichard, A. & Asosingh, K. Best practices for preparing a single cell suspension from solid tissues for flow cytometry. Cytometry A 95, 219–226 (2019). sourceExperimental — accuracy not yet measured

In the paper

“We also found a high concordance between genes that were upregulated in ‘digested’ PBMCs and those that were observed as upregulated in tissue-resident cells following tissue digestion in the Pasciuto et al.42 and Crinier et al.46 studies (Supplementary Fig. 13f).”

What the sources say

• Partly supported

  The STAR★Methods (Method Details; page unavailable) describe digestion of brain tissue to prepare single-cell suspensions, and the Results report differentially expressed genes in isolated microglia. The copy does not show that those genes were upregulated because of digestion, or compare digested with undigested tissue. The claimed high concordance is the citing authors’ own finding, not a result attributed to Pasciuto et al. (LLM, conf=0.76)

  cites Pasciuto et al.42 Pasciuto, E. et al. Microglia require CD4 T cells to complete the fetal-to-adult transition. Cell 182, 625–640.e24 (2020). sourceExperimental — accuracy not yet measured
• Partly supported

  The Results report genes expressed more strongly in splenic than blood NK cells, supporting the use of Crinier et al. as a source for a tissue-cell gene signature. But the STAR Methods say the spleen specimens were mechanically disrupted, not digested; the paper therefore does not support describing those genes as upregulated following tissue digestion. The high-concordance comparison is the citing authors’ own finding, not one attributed to Crinier et al. (LLM, conf=0.86)

  cites Crinier et al.46 Crinier, A. et al. High-dimensional single-cell analysis identifies organ-specific signatures and conserved NK cell subsets in humans and mice. Immunity 49, 971–986.e5 (2018). sourceExperimental — accuracy not yet measured

In the paper

“revealed that ex vivo activation resulted in mis-annotation of exAMs as extra microglial clusters (Supplementary Note 3)37,39.”

What the sources say

• ?Could not verify

  The STAR★Methods section discusses over-split clusters in general, but does not connect them to ex vivo activation or microglia. Searches for “exAMs,” “ex vivo,” and activated microglia found no passage in our copy addressing the claimed mis-annotation. (LLM, conf=0.72)

  cites [37] Zeisel, A. et al. Molecular architecture of the mouse nervous system. Cell 174, 999–1014.e22 (2018). sourceExperimental — accuracy not yet measured
• Partly supported

  The Results section, “Major Cell Types in the Adult V-SVZ” (page unavailable), identifies two microglial subtypes, one likely representing an activated state. It does not say that ex vivo activation caused that state or that exAMs were mis-annotated as extra clusters. (LLM, conf=0.82)

  cites [39] Mizrak, D. et al. Single-cell analysis of regional differences in adult V-SVZ neural stem cell lineages. Cell Rep. 26, 394–406.e5 (2019). sourceExperimental — accuracy not yet measured

In the paper

“Previous reports comparing cell isolation protocols or cells postisolation have often focused on aspects such as cell viability and/or yield4,62 over whether the isolated cells are truly reflective of their in vivo state13,14.”

What the sources say

• ?Not checked — no full text of the sourcecites [4] Cardona, A. E., Huang, D., Sasse, M. E. & Ransohoff, R. M. Isolation of murine microglial cells for RNA analysis or flow cytometry. Nat. Protoc. 1, 1947–1951 (2006). source
• ?Not checked — no full text of the sourcecites [62] de Haas, A. H., Boddeke, H. W., Brouwer, N. & Biber, K. Optimized isolation enables ex vivo analysis of microglia from various central nervous system regions. Glia 55, 1374–1384 (2007). source
• Supported

  This source is cited for the question of whether isolated cells reflect their in vivo state, not for the preceding claim about other reports’ focus on viability or yield. The Discussion reports rapid, profound gene-expression changes in microglia after transfer to culture and identifies this as a limitation of studying microglia outside the brain environment. (LLM, conf=0.89)

  cites [13] Gosselin, D. et al. An environment-dependent transcriptional network specifies human microglia identity. Science 356, (2017). source
• Supported

  The citation to this paper is attached to whether isolated cells reflect their in vivo state, not to the preceding claim about viability and yield. The Results section, “Microglia change gene expression profiles in culture” (page unavailable), reports extensive differences between cultured and freshly isolated microglia; the Discussion, “Microglial fate is continuously instructed by CNS-specific cues” (page unavailable), adds that even acutely isolated cells have shortcomings as models of their native state. (LLM, conf=0.90)

  cites [14] Bohlen, C. J. et al. Diverse requirements for microglial survival, specification, and function revealed by defined-medium cultures. Neuron 94, 759–773.e8 (2017). source

In the paper

“Our inhibitor-based protocol offers much greater experimental and tissue-/cell-type flexibility than previously developed experimental and computational solutions for this pervasive issue (see additional discussion of some alternative methods in Supplementary Note 7)33,34,45,65,67,68.”

What the sources say

• Partly supported

  STAR Methods (Method Details; page unavailable) describes a prior tissue-dissociation protocol with digestion times optimized by brain region. STAR Methods (Quantification and Statistical Analysis; page unavailable) also describes identifying computational components associated with digest-time artifacts. These passages support citing the paper as an example of earlier approaches, but they do not compare those approaches with the citing authors’ inhibitor-based protocol or establish its greater flexibility. (LLM, conf=0.82)

  cites [33] Saunders, A. et al. Molecular diversity and specializations among the cells of the adult mouse brain. Cell 174, 1015–1030.e16 (2018). sourceExperimental — accuracy not yet measured
• Partly supported

  The Discussion (page unavailable) describes limitations of earlier approaches: single-nucleus sequencing is unsuitable for some fresh-cell studies, and transcription/translation blockers do not prevent degradation of existing RNAs and proteins. Those passages provide context for comparing methods, but this paper does not evaluate the citing authors’ inhibitor-based protocol or establish that it offers greater experimental and tissue-/cell-type flexibility than previous experimental and computational solutions. (LLM, conf=0.78)

  cites [34] Mattei, D. et al. Enzymatic dissociation induces transcriptional and proteotype bias in brain cell populations. Int. J. Mol. Sci. 21 (2020). sourceExperimental — accuracy not yet measured
• Supported

  The Results (Comparison of tissue dissociation protocols) identify cold-active protease dissociation as a previously developed experimental approach and report that it is less damaging than warm dissociation. This supports citing the paper as an example of an alternative method; the statement about the citing authors’ inhibitor protocol being more flexible is their own comparison, not a finding attributed to this source. (LLM, conf=0.84)

  cites [45] Denisenko, E. et al. Systematic assessment of tissue dissociation and storage biases in single-cell and single-nucleus RNA-seq workflows. Genome Biol. 21, 130 (2020). source
• ?Not checked — no full text of the sourcecites [65] Adam, M., Potter, A. S. & Potter, S. S. Psychrophilic proteases dramatically reduce single-cell RNA-seq artifacts: a molecular atlas of kidney development. Development 144, 3625–3632 (2017). source
• ?Not checked — no full text of the sourcecites [67] Sankowski, R. et al. Mapping microglia states in the human brain through the integration of high-dimensional techniques. Nat. Neurosci. 22, 2098–2110 (2019). source
• ?Not checked — no full text of the sourcecites [68] Pombo Antunes, A. R. et al. Single-cell profiling of myeloid cells in glioblastoma across species and disease stage reveals macrophage competition and specialization. Nat. Neurosci. 24, 595-610 (2021). source

In the paper

“Consistent with the reports from other tissues in mice, we found that typical enzymatic digestion of mouse brain tissue induced a host of transcriptional alterations34,43,44,65.”

What the sources say

• Supported

  The section “Enzymatic tissue digestion induces transcriptional biases in brain cells” reports significant transcriptional changes across multiple cell types after enzymatic digestion of mouse hippocampal tissue, compared with mechanical dissociation. It lists hundreds of deregulated genes in neurons, glia, and other brain cells, supporting the claim of a host of alterations. (LLM, conf=0.94)

  cites [34] Mattei, D. et al. Enzymatic dissociation induces transcriptional and proteotype bias in brain cell populations. Int. J. Mol. Sci. 21 (2020). source
• ?Not checked — no full text of the sourcecites [43] Van Hove, H. et al. A single-cell atlas of mouse brain macrophages reveals unique transcriptional identities shaped by ontogeny and tissue environment. Nat. Neurosci. 22, 1021–1035 (2019). source
• ?Not checked — no full text of the sourcecites [44] van den Brink, S. C. et al. Single-cell sequencing reveals dissociation-induced gene expression in tissue subpopulations. Nat. Methods 14, 935–936 (2017). source
• ?Not checked — no full text of the sourcecites [65] Adam, M., Potter, A. S. & Potter, S. S. Psychrophilic proteases dramatically reduce single-cell RNA-seq artifacts: a molecular atlas of kidney development. Development 144, 3625–3632 (2017). source

In the paper

“However, few studies have examined the human brain for cell-type-specific differences in transcriptional states as a result of tissue handling or acute pre/postmortem processes12, and to our knowledge none have done so in a cell-type-specific manner using snRNA-seq.”

What the sources say

• Partly supported

  The ‘Functional genomics in the human brain’ section says agonal state and the time between death and tissue processing can affect observed gene expression. ‘Moving towards interrogating individual cell populations’ describes cell-type-specific analysis of postmortem human brain as a challenge. Neither passage establishes that few studies have examined cell-type-specific transcriptional responses to tissue handling or acute pre/postmortem processes. (LLM, conf=0.82)

  cites [12] Jaffe, A. E. Postmortem human brain genomics in neuropsychiatric disorders—how far can we go? Curr. Opin. Neurobiol. 36, 107–111 (2016). sourceExperimental — accuracy not yet measured

In the paper

“However, this correlation appears to be strongly driven by a small number of samples from younger donors from a single study50 (Fig. 4m, red and blue squares).”

What the sources say

• Partly supported

  The Results (Cohort selection and cross-sample alignment; page unavailable) identify this as a study of postmortem samples from 10 donors, supporting the citation’s identification of a single study with a small cohort. The available text does not establish that its donors were younger than those in the citing paper’s comparison. The claim that these samples drive the age correlation is the citing authors’ interpretation of their Fig. 4m, not a finding attributed to this source. (LLM, conf=0.76)

  cites [50] Leng, K. et al. Molecular characterization of selectively vulnerable neurons in Alzheimer’s disease. Nat. Neurosci. 24, 276–287 (2021). sourceExperimental — accuracy not yet measured

In the paper

“Second, in case–control frameworks, we demonstrate with our reanalysis of two examples from the literature that, without knowledge of the presence of the exAM signature, it can result in spurious identification of the exAM signature as differentially expressed (Supplementary Note 4)41,42,69,70.”

What the sources say

• Supported

  The citation identifies a study used as an example from the literature; the spurious-identification finding is explicitly the citing authors’ own reanalysis, not a finding attributed to this source. In Results (page unavailable), Keren-Shaul et al. describe comparing immune cells from Alzheimer’s-model mice with matched wild-type controls, so the paper fits that identification role. (LLM, conf=0.85)

  cites [41] Keren-Shaul, H. et al. A unique microglia type associated with restricting development of Alzheimer’s disease. Cell 169, 1276–1290.e17 (2017). source
• Supported

  The citation identifies a study used as an example for the citing authors’ reanalysis, rather than attributing the exAM-artifact conclusion to Pasciuto et al. The Results (Figure 6; page unavailable) compare differential gene expression in wild-type and MHC II KO microglia, making this paper a case–control example. The spurious-identification conclusion is stated as the citing authors’ own reanalysis. (LLM, conf=0.85)

  cites [42] Pasciuto, E. et al. Microglia require CD4 T cells to complete the fetal-to-adult transition. Cell 182, 625–640.e24 (2020). source
• Supported

  The citation identifies a literature study available for the citing authors’ reanalysis; the sentence attributes the reanalysis and its conclusion about spurious exAM differential expression to the citing authors, not to Krasemann et al. In STAR*METHODS (Method Details; page unavailable), Krasemann et al. describe disease-versus-control microglial expression comparisons, and Data and Software Availability reports deposited RNA-seq data. The source does not itself make the exAM claim. (LLM, conf=0.82)

  cites [69] Krasemann, S. et al. The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity 47, 566–581.e9 (2017). source
• Partly supported

  In Results and discussion (page unavailable), Kang et al. show that cellular isolation can bias microglial expression profiles and exaggerate some responses to LPS. This supports citing the paper as a relevant literature example. The source does not identify an exAM signature or report the two-example case–control reanalysis; the citing sentence describes that reanalysis as the citing authors’ own work. (LLM, conf=0.78)

  cites [70] Kang, S. S. et al. Microglial translational profiling reveals a convergent APOE pathway from aging, amyloid, and tau. J. Exp. Med. 215, 2235–2245 (2018). sourceExperimental — accuracy not yet measured

In the paper

“This pseudobulk approach also overcomes the inherent limitations of DE analysis comparisons between clusters, which suffer from using the same data to both select and test the null hypothesis24,25.”

What the sources say

• ?Not checked — no full text of the sourcecites [24] Gao, L. L., Bien, J. & Witten, D. Selective Inference for Hierarchical Clustering. Preprint at https://arxiv.org/abs/2012.02936 (2020). source
• Partly supported

  Challenge II (Status; page not supplied) identifies the double use of data in clustering followed by differential testing between clusters. It describes pseudobulk as a method for cross-sample comparisons, but does not say pseudobulk overcomes that double-use problem; the Open problems section instead identifies integrative and selective-inference approaches as possible solutions. (LLM, conf=0.88)

  cites [25] Lähnemann, D. et al. Eleven grand challenges in single-cell data science. Genome Biol. 21, 31 (2020). sourceExperimental — accuracy not yet measured

In the paper

“Additionally, very few transcriptomic studies provide information on acute premortem variables or cause of death, which are particularly important as they may potentially have a much larger effect on RNA quality and gene expression71–76,78–80.”

What the sources say

• ?Not checked — no full text of the sourcecites [71] Hardy, J. A. et al. The patients dying after long terminal phase have acidotic brains; implications for biochemical measurements on autopsy tissue. J. Neural Transm. 61, 253–264 (1985). source
• ?Not checked — no full text of the sourcecites [72] Harrison, P. J. et al. The relative importance of premortem acidosis and postmortem interval for human brain gene expression studies: selective mRNA vulnerability and comparison with their encoded proteins. Neurosci. Lett. 200, 151–154 (1995). source
• ?Not checked — no full text of the sourcecites [73] Li, J. Z. et al. Systematic changes in gene expression in postmortem human brains associated with tissue pH and terminal medical conditions. Hum. Mol. Genet. 13, 609–616 (2004). source
• Partly supported

  The Abstract and Conclusion report that agonal conditions around death substantially affect RNA integrity and gene-expression profiles, more than postmortem factors do. The paper does not establish that very few transcriptomic studies provide information on acute premortem variables or cause of death; searches for reporting practices found no such assessment. The indexed copy provides no page numbers for these passages. (LLM, conf=0.85)

  cites [74] Tomita, H. et al. Effect of agonal and postmortem factors on gene expression profile: quality control in microarray analyses of postmortem human brain. Biol. Psychiatry 55, 346–352 (2004). sourceExperimental — accuracy not yet measured
• Partly supported

  The Introduction says agonal conditions can affect brain-tissue biochemistry and that ischaemia can alter mRNA preservation; the Discussion reports lower brain pH after a prolonged terminal phase. The paper does not establish that very few transcriptomic studies report acute premortem variables or cause of death, nor does it measure a larger effect on RNA quality or gene expression. Page numbers were not available in our copy. (LLM, conf=0.84)

  cites [75] Monoranu, C. M. et al. pH measurement as quality control on human post mortem brain tissue: a study of the BrainNet Europe consortium. Neuropathol. Appl. Neurobiol. 35, 329–337 (2009). sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [76] Durrenberger, P. F. et al. Effects of antemortem and postmortem variables on human brain mRNA quality: a BrainNet Europe study. J. Neuropathol. Exp. Neurol. 69, 70–81 (2010). source
• ?Not checked — no full text of the sourcecites [78] Yates, C. M., Butterworth, J., Tennant, M. C. & Gordon, A. Enzyme activities in relation to pH and lactate in postmortem brain in Alzheimer-type and other dementias. J. Neurochem. 55, 1624–1630 (1990). source
• ?Not checked — no full text of the sourcecites [79] Webster, M. J. Tissue preparation and banking. Prog. Brain Res. 158, 3–14 (2006). source
• Partly supported

  The Discussion (page unavailable) says pre- and post-mortem factors affect the microglial transcriptome, and Methods > Sources of transcriptomic variation (page unavailable) includes cause of death among the covariates examined. The paper does not establish that very few transcriptomic studies report acute premortem variables or cause of death, or that these variables have a much larger effect on RNA quality and gene expression. (LLM, conf=0.84)

  cites [80] de Paiva Lopes, K. et al. Genetic analysis of the human microglial transcriptome across brain regions, aging and disease pathologies. Nat. Genet. 54, 4-17 (2022). sourceExperimental — accuracy not yet measured

In the paper

“which is prevalent in published literature and can substantially confound downstream analyses”

In the paper

“Here, we demonstrate that use of enzymatic dissociation on brain tissue induces an aberrant ex vivo gene expression signature, most prominently in microglia, which is prevalent in published literature and can substantially confound downstream analyses.”

In the paper

“We demonstrate that this protocol effectively eliminates the artifactual ex vivo transcriptional signature in mouse CNS tissue, as well as in non-CNS cell types and across species.”

In the paper

“although it is highly variable between subjects”

In the paper

“Through the use of acutely resected neurosurgical tissue, we reveal that a similar signature can be detected in microglia following prolonged exposure to room temperature.”

In the paper

“In the mouse CNS, we find microglia are preferentially sensitive to ex vivo artifacts and characterize this ex vivo response through in-depth comparative analyses of multiple microglial isolation protocols.”

In the paper

“We also identify a similar signature in postmortem human brain single-nucleus RNA-sequencing datasets”

In the paper

“and show that this signature is induced in freshly isolated human tissue by exposure to elevated temperatures ex vivo”

In the paper

“We find a similar gene signature is present in postmortem microglia and astrocytes, across all snRNA-seq datasets analyzed, although it is highly variable between subjects.”

In the paper

“The microglial factor (Fig. 4a) was strikingly similar to the mouse signature, with 79% (30 of 38) of the genes directly overlapping with mouse signature or that were genes from similar families/pathways.”

In the paper

“The astrocyte postmortem factor (Fig. 4b) exhibited modest overlap with the mouse signature, as 44% (12 of 27) directly overlapped or were genes in similar families.”

In the paper

“We identified a consensus upregulated DE signature shared between the comparisons of each experimental group with the ENZ-NONE group (Supplementary Tables 4–8).”

In the paper

“These differences in clustering were driven by shared gene expression changes across multiple lymphoid and myeloid cell types (Supplementary Fig. 13d–f and Methods).”

In the paper

“demonstrated that this activation signature was almost exclusively enriched in cells from the exAM cluster.”

In the paper

“Subclustering analysis of our dataset revealed that microglia/myeloid cells were the only major cell class that exhibited clustering driven by the presence/absence of the inhibitor cocktail (Supplementary Fig. 8a–f).”

In the paper

“When we performed the same module scoring using the mouse signature, we found that microglia and some astrocytes were again the most enriched for the signature (Supplementary Fig. 14, rightmost column).”

In the paper

“Analysis of the proportion of cells from each experimental group per cluster revealed that only the microglial clusters differed significantly (*FDR < 0.003) (Fig. 2e and Supplementary Table 18).”

In the paper

“Microglia were the most prominent cell type in the dataset using this digestion protocol (Fig. 2b,c,e), likely indicating that the Miltenyi dissociation system utilized is not appropriate for general unbiased profiling of brain tissue when more accurate cell-type proportions are desired.”

In the paper

“Only the factors from the microglial and astrocyte analyses exhibited a 25% or greater overlap with the mouse gene list (Supplementary Table 22).”

In the paper

“we confirm that, while extremely rare, the chemokine cluster Ccl4+ cells are a true in vivo state and not merely an additional dissociation-induced response (Fig. 1i).”

In the paper

“we found that non-microglial CNS-associated macrophages present in our dataset exhibited enrichment of the artifactual exAM signature when tissue was enzymatically digested without inhibitors (Supplementary Fig. 11a–d).”

In the paper

“Analysis of this scRNA-seq dataset revealed significant differences in clustering in most cell types represented in the dataset (including both lymphoid and myeloid populations; Supplementary Fig. 13a,b), dependent on inhibitor condition (Supplementary Fig. 13c, circled).”

In the paper

“analysis of data from a separate mouse experiment revealed that artifacts are possible even when using ideal isolation methods when there are small differences in sample processing before cell capture (Supplementary Fig. 4 and Supplementary Note 2).”

In the paper

“However, in other cell types, including oligodendrocytes/oligodendrocyte precursors, we did identify slightly increased IEG expression in cells digested without inhibitors (Supplementary Fig. 8g–i).”

In the paper

“we analyzed several standard scRNA-seq quality control metrics and demonstrate that the exAM cluster was of equal or significantly better quality compared with either the homeostatic cluster or all non-exAM cells in all of the metrics examined (P < 0.05) (Supplementary Fig. 3a).”

In the paper

“We observed enrichment of the exAM signature via gene module scoring in the myeloid populations, consistent with ex vivo activation (Supplementary Fig. 9a–e).”

In the paper

“We also confirmed that the exAM signature was not sensitive to differences in single-cell capture/library preparation (Supplementary Figs. 5 and 6).”

In the paper

“In total, we found an iNMF factor in each cell type that shared at least one gene with the combined mouse signature gene lists (Supplementary Table 22).”

In the paper

“Our consensus exAM signature does overlap with the dissociation signature in the study by Van Hove et al.[43] (19 of 25 genes in consensus exAM list; Supplementary Tables 4, 5 and 21a).”

What the sources say

• ?Not checked — no full text of the sourcecites Van Hove et al.[43] Van Hove, H. et al. A single-cell atlas of mouse brain macrophages reveals unique transcriptional identities shaped by ontogeny and tissue environment. Nat. Neurosci. 22, 1021–1035 (2019). source

In the paper

“An additional cluster was nearly exclusively composed of cells from samples enzymatically digested without inhibitors (ENZ-NONE) (Fig. 1c,e), characterized by expression of several immediate early genes (IEGs) (for example, Fos, Jun), stress-induced (Hspa1a, Dusp1) and immune-signaling genes (Ccl3, Ccl4) (Fig. 1d and Supplementary Fig. 2c,d).”

In the paper

“However, we do find some genes that appear tissue-/cell-type-specific to myeloid cells, such as Ccl3 and Ccl4 (Supplementary Table 21d).”

In the paper

“Our results demonstrate that some signatures previously characterized as tissue-resident signatures are often significantly enriched for dissociation-related signatures, underscoring the need for recognizing these artifactual signatures to disentangle them from true in vivo biologically driven transcriptional signatures of tissue residency (Supplementary Note 5 and Supplementary Fig. 12).”

In the paper

“The mouse signature showed enrichment preferentially in monocytes, again suggesting that some of the signature we originally identified in microglia may be common throughout myeloid lineage cells (Supplementary Fig. 13e).”

In the paper

“confirming that cold Dounce homogenization is sufficient to prevent ex vivo activation.”

In the paper

“We found that an issue resulting in increased time at room temperature for one sample during cell sorting induced the exAM signature selectively in that replicate but not the other three (Supplementary Fig. 4), highlighting the importance of the inclusion of true biological replicates in scRNA-seq experiments.”

In the paper

“Similar analyses across the postmortem astrocytes found nearly identical results (Supplementary Fig. 16a–c).”

In the paper

“Through the use of a transcriptional and translational inhibitor cocktail at multiple steps of the protocol, we were successfully able to eliminate the ex vivo activation during enzymatic digestion, both in FACS-sorted myeloid cells and in an unsorted preparation of all CNS cell types.”

In the paper

“Interestingly, we found this induction preferentially occurred in microglia and other brain myeloid cells.”

In the paper

“The only statistically significant correlation was found with age of the donor (Fig. 4l–n) (Spearman: r = 0.3208, *P = 0.0001, n = 48).”

In the paper

“Analysis of correlation in only samples of ages 75+ found no significant correlation (age 75+ Spearman: r = 0.1879, P = 0.2653, n = 37).”

In the paper

“Taken together, our results demonstrate that computational solutions, such as regression of a digestion signature or cell removal, can bias the interpretations of true biological states (Supplementary Notes 4, 5, and 7).”

In the paper

“while the proliferative cluster was statistically significant between the DNC-NONE and enzymatic digestion with inhibitors (ENZ-INHIB) groups (#FDR ≤ 0.02) (Fig. 1e and Supplementary Table 3)”

In the paper

“Analysis of our own and previous human postmortem snRNA-seq datasets revealed a signature similar to the mouse exAM signature present in both microglia and astrocytes.”

In the paper

“Using a combination of our own snRNA-seq and a reanalysis of published literature, we find that a similar signature is also expressed by microglia and astrocytes in postmortem brain.”

In the paper

“We also demonstrate that this signature can be induced by technical variation in sample processing (freezing delay) in human brain tissue”

In the paper

“To determine whether this signature could be upregulated in human brain, we designed a study using a technical variable (freezing delay) in acutely resected neurosurgical tissue and demonstrated that microglia upregulated the signature in the timeframe examined.”

In the paper

“These results suggest that the presence of this signature in postmortem brain samples may be the result of a combination of acute premortem (agonal state, cause of death, comorbidities and so on) and postmortem (postmortem interval (PMI), storage time, RNA quality and so on) variables.”

In the paper

“which is prevalent in published literature and can substantially confound downstream analyses”

• Also stated in: § Presence of artifactual microglia signature is common across multiple experimental parameters · hedged

  “Our reanalyses demonstrate that induction of the exAM signature can occur differentially across experimental conditions and may impact downstream analyses.”

In the paper

“We also perform reanalysis of several published datasets to demonstrate how the use of enzymatic and/or room-temperature processing can impact downstream results and conclusions, underscoring its prevalence and the need for new, flexible protocols that address this issue.”

In the paper

“likely indicating that the Miltenyi dissociation system utilized is not appropriate for general unbiased profiling of brain tissue when more accurate cell-type proportions are desired.”

In the paper

“we find the majority of genes are shared across all tissue/cell types, indicative of a common cellular response to dissociation (Supplementary Table 21d).”

In the paper

“These results suggest that the gene expression signature we identified in postmortem human microglia can be induced by a number of cell-autonomous molecular processes likely including, among others, hypoxia, cell death and temperature stress.”

In the paper

“suggestive of differential time scales of activation for these two brain populations, as has been previously described35.”

What the sources say

• ?Not checked — no full text of the sourcecites [35] Satoh, J. & Kim, S. U. HSP72 induction by heat stress in human neurons and glial cells in culture. Brain Res. 653, 243–250 (1994). source

In the paper

“indicating that its presence in postmortem brains may not be reflective of the in vivo variable of interest in a particular study.”

In the paper

“Given the prevalence of the signature in current literature, our results have substantial implications for many single-cell datasets: two examples include annotation of cell types and subtypes and DE in case–control studies.”

In the paper

“this increase should be interpreted with caution, given that this population makes up only 0.1% of total cells (27 of 19,563) across all 12 samples.”

In the paper

“The presence of dissociation-induced changes in atlas or survey studies and experiments analyzing digested versus undigested tissues (or only digested tissue) has the potential to obscure biologically meaningful differences that would otherwise contribute more substantially to clustering and downstream analysis (Supplementary Note 3).”

In the paper

“The potential for meta variables unrelated to the particular human condition/disease being studied to impact the signatures observed and potentially confound downstream analysis highlights the need for more complete metadata to be made public with studies of postmortem brain.”