Spatial Transcriptomics Uncovers Coordinated Cell Responses to Amyloid

With single-cell resolution, scientists now have a good idea of what amyloid plaques do to nearby cells in the brains of mice. Scientists led by Bart De Strooper, KU Leuven, Belgium, used a combination of spatial transcriptomics and in situ hybridization to build a detailed map of the gene-expression changes that occur around β-amyloid deposits. Coordinated changes of astrocyte and microglial transcriptomes indicated crosstalk between the cells. Oligodendrocytes did their own thing, ramping up expression of myelin-related genes soon after plaques deposited, only to temper that as plaque load increased. In their paper posted August 12 on bioRχiv, the authors conclude that plaques, rather than being innocuous sequela of pathology, induce strong and coordinated cellular responses. The work builds on the concept of a “cellular phase” of AD originally proposed by De Strooper and Eric Karran, then at Alzheimer Research U.K., Cambridge (see Alzforum Mar 2016 webinar). 

  • In mice, amyloid plaques prompt nearby cells to shift transcription.
  • Astrocytes and microglia communicate intensely.
  • Oligodendrocytes overexpress, and later underexpress, genes needed for myelin repair.

“It’s a really fascinating study because it combines a whole-brain approach across the disease course with cell-specific analysis,” said Pieter Jelle Visser, VU University Medical Center, the Netherlands. He was not involved in the research. “Most previous studies use a more focused approach on cell type or region, but with these combined techniques, it’s possible to capture the larger picture.”

In recent years, researchers have found that microglia adopt a unique transcriptional profile in response to plaques (Jun 2017 news on Keren-Shaul et al., 2017; Srinivasan et al., 2019). How about other cell types? Researchers have tried to study astrocytes, oligodendrocytes, and neurons adjacent to plaques by isolating the cells, but such procedures both disrupt the cells and lack spatial resolution that might be critical to understand cellular responses to plaques.

Spatial Transcriptomics. From three adjacent brain slices, the outer two are stained for plaques, neurons, and astrocytes, while the inner yields a corresponding map of the transcriptome. [Courtesy of Chen et al., 2019.]

Co-first authors Wei-Ting Chen and Ashley Lu and colleagues combined a type of tissue-slice transcriptomics with in situ hybridization to try to preserve spatial information while still measuring RNA at the single-cell level. First, they used trios of adjacent coronal slices, each 10 μm thick, in a spatial-transcriptomics approach (Ståhl et al., 2016). The outer slices were used to identify cells and pathological changes, such as amyloid plaques, and to serve as a spatial reference for the inner slice used for transcriptomics (see image above). This slice was placed on a glass slide containing 1,000 reverse-transcriptase primer arrays, each capable of measuring complete transcriptomes. By referencing the transcriptomes with anatomical slides, the researchers were able to map expression changes in the vicinity of plaques.

Next, the researchers used in-situ hybridization to narrow down expression of individual genes to the specific cells that expressed them, tracking how they changed as plaques deposited. “The strength of this approach is clearly that spatial information is preserved and that transcriptomic changes can be comprehensively studied in defined brain regions with high resolution in an unbiased manner,” noted Annett Halle, German Center for Neurodegenerative Diseases, Bonn.

In this way, Chen and colleagues studied slices from 10 wild-type mice and 10 APPNL-G-F knock-ins, which carry the human APP gene with the Swedish, Iberian, and Artic mutations. Knock-in mice start accumulating diffuse plaques around three months of age, and they became positive for Thioflavin S around six months. By 18 months, they had extensive plaque.

The authors first compared transcriptomes of brain slices taken from young and old APPNL-G-F mice. In slices from 3-month-old animals, diffuse plaques were surrounded by transcripts related to myelination, which suggested to the authors that oligodendrocytes had begun to fix myelin damage. In slices from 18-month-old mice, however, cells near plaques barely expressed those genes. This fits with a recent report suggesting that oligodendrocyte precursors around plaques become senescent, stop differentiating into myelin-repairing oligodendrocytes, and instead release inflammatory molecules (Apr 2019 news). 

What about other cell types? In slices from 18-month-old mice, the researchers identified a co-expression network of 57 genes that was upregulated around plaques. Most of these plaque-induced genes are known to be expressed by astrocytes or microglia and some, such as Trem2, Tyrobp, and ApoE, are well-known AD genes. Thirty-six, however, were not previously associated with plaques. These included genes involved in the classical-complement cascade, endocytosis, lysosomal degradation, antigen processing, immune responses, and oxidation-reduction responses.

To zero in on which cells in the slice expressed these genes, the authors used in situ hybridization (Ke et al., 2013). In a nutshell, they generated probes for each of the 57 genes, as well as for RNAs specific to microglia, astrocytes, neurons, and glia, and incubated them on a slide that was later stained for plaques. The patterns of co-localization revealed which cells expressed which transcripts (see image below).

Cell-Specific Expression. In a coronal section of the mouse brain (left), binding of probes for each of 57 plaque-induced genes, and for 27 cell-type-specific transcripts (bottom right), identify transcripts expressed by microglia, oligodendrocytes, astrocytes, and neurons, and which are up- or downregulated in the vicinity of plaques (white, top right). [Courtesy of Chen et al., 2019.]

Most plaque-induced genes were expressed by either microglia or astrocytes. Oligodendrocytes expressed only a few, including the complement-cascade protein C4 and some lysosomal genes. Interestingly, in the wild-type mice, the expression of these plaque-induced genes was not co-regulated. However, in the APPNL-G-F mice, their expression became coordinated at six months of age, and grew stronger as plaques accumulated, suggesting strong crosstalk between astrocytes and microglia around amyloid plaques.

The most intense RNA changes occurred within 10 μm from the edge of plaques. These were weaker in cells located farther away.

“The transcriptomic changes that are occurring in different cells surrounding plaques at various distances is intriguing,” wrote Shane Liddelow, New York University, to Alzforum. “Traditional sequencing efforts have largely missed these nuanced responses.” He was particularly interested in the localized response of astrocytes, which become reactive around plaques and secrete proinflammatory factors. “It suggests astrocytes’ negative effects on neurons are secondary to a positive response trying to destroy or remove plaque, or perhaps cells that the plaque has already damaged.”

The authors noted that while APPNL-G-F knock-in mice model Aβ deposition and inflammation, they offer little information about tau tangles, neurodegeneration, or memory deficits. The next steps will be to apply these techniques to human postmortem tissue to see if human cells respond similarly, De Strooper said. He also wants to see if transcriptomes differ between people whose cognition declines and those who remain healthy despite the presence of plaques. That could help explain why some people are resilient in the face of ongoing amyloid pathology, and might point to new therapeutic approaches. Liddelow added that spatial transcriptomics of human tissue will help identify the most appropriate rodent models for studying plaque-induced responses.

Michael Heneka, University of Bonn, Germany, cautioned that the gene-transcription data should be backed up by protein-level analysis, because only a small percentage of transcripts for inflammatory proteins are translated.—Gwyneth Dickey Zakaib

Comments

  1. Wei-Ting and collaborators present, to the best of our knowledge, the first application of spatial transcriptomics and in situ sequencing to a murine model of Alzheimer’s disease (AD). This study, motivated by the empirical study of a complex “cellular phase of AD” (De Strooper and Karran, 2016) in response to amyloid plaques identified 57 responsive genes with correlated expression. This network module involves amyloid Plaque Induced Genes (PIGs) whose correlation is driven by patterns of expression that correlate with plaque load, occur in proximity to plaques, and involve multicellular populations of glial cell types. Consistent with the hypothesis of inflammatory-like cellular alterations potentially triggering brain dyshomeostasis in response to pathology, the responsive gene network involves alteration of complement, oxidative stress, and inflammation pathways, as well as brain-wide dysregulated microglia and astrocyte crosstalk.

    Interestingly, the authors also identified a separate gene network that suggests abnormal myelination. In particular, oligodendrocyte-expressing genes were altered in response to pathology, showing a qualitatively distinct response conditional on the stage of pathology, being overexpressed in early stages and downregulated in late. Gene co-expression analysis further revealed a gene module consistent with the observed differential expression, and involving both oligodendrocytes and oligodendrocyte precursor cells (the OL module). Notably, a pathology-responsive expression signature suggestive of general alterations in myelination is consistent with what we found while analyzing the single-cell heterogeneity of postmortem cortical samples (Mathys et al. 2019). The authors’ observations also suggest that, in contrast to the inflammatory PIG module, which showed a regional homogeneous response, the particular response dynamics of the OL module might reflect differential brain region vulnerability. Further single-cell studies involving multiple brain regions will likely help exploring these observations and spatial heterogeneity in the context of the human brain.

    Although spatial information is lost during single-cell or nuclei profiling of human postmortem samples, the identification of robust gene in situ co-expression signatures in the vicinity of pathogenic hallmarks of neurodegeneration—such as the one provided here by the authors—might potentially aid the computational inference of spatial information in human data. Can we guess whether clear subpopulations of cells in human brain AD atlases were originally localized to the vicinity of plaques? Is all this information really lost? These are interesting questions we can now start exploring thanks to both the increasingly available single-cell profiles of human brain pathology and, as coherently shown in this study, to the application of new technologies of spatial and in situ transcriptomics to multiple brain regions of animal models.

    References:

    De Strooper B, Karran E. The Cellular Phase of Alzheimer's Disease. Cell. 2016 Feb 11;164(4):603-15. PubMed.

    Mathys H, Davila-Velderrain J, Peng Z, Gao F, Mohammadi S, Young JZ, Menon M, He L, Abdurrob F, Jiang X, Martorell AJ, Ransohoff RM, Hafler BP, Bennett DA, Kellis M, Tsai LH. Single-cell transcriptomic analysis of Alzheimer's disease. Nature. 2019 Jun;570(7761):332-337. Epub 2019 May 1 PubMed.

    View all comments by Jose Davila-Velderrain View all comments by Li-Huei Tsai

  2. This work by De Strooper and colleagues adds very interesting new aspects to previous work by several groups on transcriptomic changes in glia that have shaped our understanding of the diversity of glial phenotypes in neurodegeneration and the pathways involved in disease development and progression.

    One important aspect concerns the methods that are used here: This is the first time that techniques such as spatial transcriptomics and in situ sequencing have been applied to brain tissue with Aβ pathology, thus adding an additional layer of information to our understanding of the transcriptomic response to Aβ plaques. The strength of this approach is clearly that spatial information is preserved and that transcriptomic changes can be comprehensively studied in defined brain regions with high resolution in an unbiased manner. Another advantage is that the transcriptomic analysis in not confined to one cell type only (such as microglia) but that reaction of several cell types can synchronously be deducted, thus potentially uncovering inter-cell-dependent responses to pathological insults.

    Furthermore, there are several very interesting biological/pathophysiological conclusions that can be drawn from this study.

    For example, the authors characterize a number of genes from a co-expression network that are spatially correlated with Aβ plaques and, hence, called Plaque-Induced Genes, or PIGs. Although some of them have been described in bulk RNA-sequencing data in mouse models with Aβ pathology either in whole tissue (e.g., complement-related genes such as C1qa, C1qb, C1qc, and C4a; Srinivasan et al., 2016) or in isolated microglia (Orre et al., 2014), here the authors spatially link the upregulation of these genes to the plaque microenvironment, a clearly novel and important finding, which needs further investigation.

    Another very interesting aspect is the response of oligodendrocytes to Aβ pathology. The authors find an upregulation of oligodendrocyte genes very early during disease progression and a depletion of oligodendrocyte genes during late stages of disease. This finding is new and especially interesting in relation to the recent single-nucleus RNA-Seq study in human AD tissue by Mathys et al., in which increased AD pathology correlated with a global transcriptional activation in oligodendrocytes in male patients (Mathys et al., 2019). 

    One can easily envision questions that would be very interesting to investigate in follow-up studies: What does the transcriptomic response look like in human tissue in different brain regions at different neuropathological stages and in response to different Aβ plaque types? Is the transcriptomic response to other types of pathology of neurodegeneration (e.g., tau pathology) similar, and does it show specific spatial distribution? Is the plaque-dependent response in microglia, astrocytes and oligodendrocytes caused by inter-cellular crosstalk or by a stereotypical cellular reaction to the plaque microenvironment? Are there sex-related differences (given that this study focused on male mice)?

    Finally, and perhaps most importantly: What is the impact of this plaque-dependent gene expression program on cellular function and how can gene expression or cellular response be targeted for future therapy?

    References:

    Srinivasan K, Friedman BA, Larson JL, Lauffer BE, Goldstein LD, Appling LL, Borneo J, Poon C, Ho T, Cai F, Steiner P, van der Brug MP, Modrusan Z, Kaminker JS, Hansen DV. Untangling the brain's neuroinflammatory and neurodegenerative transcriptional responses. Nat Commun. 2016 Apr 21;7:11295. PubMed.

    Orre M, Kamphuis W, Osborn LM, Jansen AH, Kooijman L, Bossers K, Hol EM. Isolation of glia from Alzheimer's mice reveals inflammation and dysfunction. Neurobiol Aging. 2014 Dec;35(12):2746-60. Epub 2014 Jun 14 PubMed.

    Mathys H, Davila-Velderrain J, Peng Z, Gao F, Mohammadi S, Young JZ, Menon M, He L, Abdurrob F, Jiang X, Martorell AJ, Ransohoff RM, Hafler BP, Bennett DA, Kellis M, Tsai LH. Single-cell transcriptomic analysis of Alzheimer's disease. Nature. 2019 Jun;570(7761):332-337. Epub 2019 May 1 PubMed.

    View all comments by Annett Halle View all comments by Stanislav Kozlov View all comments by Marc Beyer

  3. A central question in AD research is the relationship between amyloid plaques and neurodegeneration. In this highly innovative study, the authors focus on the spatial localization of changes in cell types and their signatures, specifically with respect to amyloid plaques in mouse models of Alzheimer’s disease. The use of high-throughput spatial arrays, which allow for genome-wide measurements in space, combined with targeted investigation of key “plaque-induced genes” at the single-cell level allows the authors to investigate cell-type-specific changes and interactions that are likely to be set off by the deposition of these plaques. In support of the “cellular model” of AD progression espoused by the authors, they identify rings of altered cell-type signatures among microglia and astrocytes with distance from amyloid plaques, suggesting that these changes may lead to the sustained alterations in downstream cell types that are hallmarks of AD. Overall, the use of novel spatial transcriptomics techniques, combined with validation studies looking at spatial changes at the single-cell level, lead to a new way of characterizing the cascade of molecular changes in AD.

    View all comments by Vilas Menon

Make a Comment

To make a comment you must login or register.

References

Webinar Citations

  1. Webinar: Can ‘Cellular Phase’ Unite Disparate Data on Alzheimer’s Pathogenesis?

News Citations

  1. Hot DAM: Specific Microglia Engulf Plaques
  2. Plaques Age Glial Precursors, Stoking Inflammation

Research Models Citations

  1. APP NL-G-F Knock-in

Paper Citations

  1. Keren-Shaul H, Spinrad A, Weiner A, Matcovitch-Natan O, Dvir-Szternfeld R, Ulland TK, David E, Baruch K, Lara-Astaiso D, Toth B, Itzkovitz S, Colonna M, Schwartz M, Amit I. A Unique Microglia Type Associated with Restricting Development of Alzheimer's Disease. Cell. 2017 Jun 15;169(7):1276-1290.e17. Epub 2017 Jun 8 PubMed.
  2. Srinivasan K, Friedman BA, Etxeberria A, Huntley MA, van der Brug MP, Foreman O, Paw JS, Modrusan Z, Beach TG, Serrano GE, Hansen DV. Alzheimer's Patient Microglia Exhibit Enhanced Aging and Unique Transcriptional Activation. Cell Rep. 2020 Jun 30;31(13):107843. PubMed.
  3. Ståhl PL, Salmén F, Vickovic S, Lundmark A, Navarro JF, Magnusson J, Giacomello S, Asp M, Westholm JO, Huss M, Mollbrink A, Linnarsson S, Codeluppi S, Borg Å, Pontén F, Costea PI, Sahlén P, Mulder J, Bergmann O, Lundeberg J, Frisén J. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science. 2016 Jul 1;353(6294):78-82. PubMed.
  4. Ke R, Mignardi M, Pacureanu A, Svedlund J, Botling J, Wählby C, Nilsson M. In situ sequencing for RNA analysis in preserved tissue and cells. Nat Methods. 2013 Sep;10(9):857-60. Epub 2013 Jul 14 PubMed.

Further Reading

Papers

  1. Rodriques SG, Stickels RR, Goeva A, Martin CA, Murray E, Vanderburg CR, Welch J, Chen LM, Chen F, Macosko EZ. Slide-seq: A scalable technology for measuring genome-wide expression at high spatial resolution. Science. 2019 Mar 29;363(6434):1463-1467. Epub 2019 Mar 28 PubMed.
  2. Maniatis S, Äijö T, Vickovic S, Braine C, Kang K, Mollbrink A, Fagegaltier D, Andrusivová Ž, Saarenpää S, Saiz-Castro G, Cuevas M, Watters A, Lundeberg J, Bonneau R, Phatnani H. Spatiotemporal dynamics of molecular pathology in amyotrophic lateral sclerosis. Science. 2019 Apr 5;364(6435):89-93. PubMed.

Primary Papers

  1. Chen WT, Lu A, Craessaerts K, Pavie B, Frigerio CS, Mancuso R, Qian X, Lalakova J, Kühnemund M, Voytyuk I, Wolfs L, Snellinx A, Munck S, Jurek A, Navarro JF, Saido TC, Lundeberg J, Fiers M, De Strooper B. Spatial and temporal transcriptomics reveal microglia-astroglia crosstalk in the amyloid-β plaque cell niche of Alzheimer’s disease. bioRxiv August 12, 2019. BioRxiv.

Annotate

To make an annotation you must Login or Register.