As Alzheimer’s tangle pathology progresses, neurodegeneration sweeps through the brain in a stereotypical fashion. But why do some neurons perish early on, while their neighbors persist until the bitter end? A study published January 11 in Nature Neuroscience addressed this question by tracing the gene-expression profiles of neurons in the brains of people who died in the early or late stages of the disease. Among myriad subpopulations of cells, the researchers zeroed in on subsets of excitatory neurons that express the transcription factor RORB as the first to succumb. Initially in the entorhinal cortex, and then later in the outer neocortex, excitatory neurons bearing this particular marker were selectively vulnerable to tau accumulation, and to death. The study, led by Lea Grinberg and Martin Kampmann, both at the University of California, San Francisco, also pegged a type of reactive astrocyte that may shirk its duties of protecting and nourishing neurons. In all, the findings hint at common mechanisms underlying selective vulnerability to AD pathogenesis in different regions of the brain.

  • Single-nuclei transcriptomics identifies selectively vulnerable neurons.
  • RORB-positive cells are prone to accumulate tau and die.
  • A subset of reactive astrocytes were spotted on the scene, too.

“The identification and molecular characterization of selectively vulnerable neurons in AD is a very important contribution,” commented Hansruedi Mathys of the Massachusetts Institute of Technology. “This work will undoubtedly facilitate future studies to understand why these neurons are so vulnerable.”​

For decades in Alzheimer’s research, selective vulnerability has loomed as the big question, without much progress toward an answer. “In Alzheimer’s disease, which is so complex, selective cellular vulnerability is a key anchoring point for mechanistic discovery,” wrote Jessica Rexach of the University of California, Los Angeles, to Alzforum. “We must understand what differs between vulnerable and spared cells as disease progresses. By using highly curated neuropathological specimens, and validation across independent studies, this important work sets an exciting foundation for continued discovery.”

As documented by Braak staging, neurofibrillary tangles overtake the brain in a defined sequence, appearing in the entorhinal cortex in early disease and moving to the outer reaches of the cortex in the final stages. Neuronal death tracks closely behind. Grinberg, a neuropathologist who has examined 3,000 human brains over the years, was struck by the selectivity with which neurons succumb to the disease. “Why is it that one neuron dies when tau tangles appear in the region, while its neighbor survives?” she wondered. Past studies have attempted to define regional and morphological characteristics of vulnerable cells, but Grinberg wanted to dive deeper and define vulnerability at the molecular level (Aug 2016 news). To probe the underpinnings of neuronal weakness to AD pathology, Grinberg joined forces with Kampmann, a cell biologist who uses genetic tools to unearth disease mechanisms.

Co-first authors Kun Leng, Emmy Li, and colleagues started by procuring brain samples from people who had died at different stages AD. Their cohort included 10 men, all ApoE3 homozygotes, who lived to between 50 and 91 years of age. Three had neuropathology in Braak stage 0, four in stage 2, and three in stage 6. From each brain, the researchers sampled cells from the entorhinal cortex (EC) and the superior frontal gyrus (SFG). These two regions are typically inundated with tau tangles by Braak stage II and V, respectively.

Across all the samples, the researchers harvested more than 42,000 cells from the EC, and 63,000 from the SFG. Using single-nucleus RNA sequencing, they analyzed the gene-expression profiles of each cell. While previous studies already used single-nuclei approaches to compare the transcriptomes of brain cells from people with AD to those of healthy controls, the current study is unique in its analysis along disease progression (May 2019 news; Nov 2019 news). 

Who Dies First? Researchers sampled the gene-expression profiles of neurons in the entorhinal cortex (EC) and superior frontal gyrus (SFG), regions invaded by tangles at early and late stages, respectively. [Courtesy of Leng et al., Nature Neuroscience, 2020.]

The first order of business was to identify selectively vulnerable cell types—i.e., those that decreased in number as disease progressed. Using broad cell type markers at first, the researchers saw a drop in excitatory neurons in the EC starting at Braak stage 2, and later a dip in excitatory neurons in the SFG at stage 6. In contrast, numbers of inhibitory neurons did not drop as disease got worse.

Next, the researchers zoomed in on transcriptomic differences between excitatory neurons within the EC, grouping nine distinct subpopulations based on their gene-expression profiles. Of these, three subpopulations halved in number in Braak stage 2, pegging them as a selectively vulnerable population.

Rooting Out the Weak. In the EC, researchers detected nine subpopulations of excitatory neurons (left). Of those, s1/amber, s2/olive, and s4/turquoise shrank in Braak stage 2 (right). On right, cells are color-coded by donor. [Courtesy of Leng et al., Nature Neuroscience, 2020.]

What made these three subsets so susceptible? Two of them uniquely expressed three transcripts—RORB, CTC-340A15.2, and CTC-535M15.2. The latter two are noncoding transcripts of unknown function. RORB (RAR-related Orphan Receptor b) encodes a transcription factor that drives the development of layer IV cortical neurons, although it is also expressed in other cortical layers (Jabaudon et al., 2012; Oishi et al., 2016). Its role in the adult brain is not understood.

Compared to other excitatory neurons in the EC, the vulnerable cells were relatively flush with transcripts encoding axonal proteins and voltage-gated potassium channels, while they had few transcripts encoding synaptic signaling proteins.

What happened in the SFG, where tau tangles don’t appear until Braak stage 5? There, the researchers identified 11 subpopulations of excitatory neurons. Of those, two shrank toward Braak stage 6. Notably, these two vulnerable SFGs populations also expressed the same three transcripts that stood out in vulnerable excitatory neurons in the EC. At a broad transcriptome level, the vulnerable SFG neurons had more in common with the vulnerable neurons in the EC than they did with any other EC subpopulation. Kampmann emphasized that this does not mean that the vulnerable neurons in the EC are the same exact type as those in the SFG. In fact, there were many gene-expression differences between them. However, it does suggest that similar mechanisms might underlie their tendency to crumble once tangles come to town.

The researchers spotted similar subpopulations of neurons within other gene-expression datasets from postmortem brain samples reported from Li Huei Tsai’s lab at MIT and Rick Livesey’s group at University College London (Mathys et al., 2019; Marinaro et al., 2020). 

To validate their snRNA-Seq findings, the researchers used immunofluorescence to evaluate 26 separate postmortem brain samples with AD pathology spanning Braak stages 0 to 6. In contrast to the all-male, ApoE3-only cohort used for the snRNA-Seq analysis, this cohort included 10 women and four ApoE4 carriers. The researchers reported that in this sample, too, the proportion of RORB-expressing excitatory neurons in the EC waned with increasing Braak stages, while excitatory neurons devoid of RORB did not decline. Strikingly, the researchers detected accumulation of phospho-tau specifically in RORB+ neurons. Morphologically, RORB+ excitatory neurons came in different shapes and sizes, including both pyramidal neurons and large, multipolar neurons (see image below). Together, the findings suggested that RORB+ neurons are selectively vulnerable to tauopathy in AD.

Shapes and Sizes. Large multipolar (m1 and m3-m5) and a pyramidal (p1) neuron express RORB (green). Some also accumulate tangles (red). From layer II EC of a person with AD. [Courtesy of Leng et al., Nature Neuroscience, 2020.]

While they were at it, the researchers also investigated what genes glial cells were expressing. For astrocytes, they found four subpopulations in the EC, six in the SFG. In each region, at least one such group practically screamed with GFAP, a marker of reactive astrocytes. These cells expressed few markers of homeostatic astrocytes, suggesting they had ditched their neuroprotective persona for reactive ones. The paper does not include a detailed analysis of how these reactive populations changed with worsening disease.

For microglia, the researchers detected four subpopulations in the EC, and five in the SFG. However, they were unable to find the transcripts that have been associated with homeostatic or disease-associated microglia in other studies (Jun 2017 news; Sep 2017 news; May 2019 news). The researchers suggest that low-abundance transcripts may have been lost because they resided in the cytoplasm, and the study used only nuclei. A recent study found that snRNA-Seq misses transcripts that have been linked to AD and might be abundant in the cytosol (Oct 2020 news). 

In a comment to Alzforum, Arizona-based researchers Thomas Beach and Geidy Serrano of Banner Sun Health Research Institute in Sun City; Diego Mastroeni of Arizona State University in Tempe; and Matthew Huentelman of the Translational Genomics Research Institute in Phoenix, discussed the trade-offs of using single-cell versus nuclei for transcriptomics. On one hand, plucking intact, whole cells from postmortem brain is fraught with difficulty, and single-nuclei approaches may more closely reflect the proportion of cells types found in vivo. On the other hand, nuclear transcriptomes cannot account for the many transcripts that are exported to the cytoplasm, especially those awaiting translation in synaptic terminals. “All methods have strengths and weaknesses, and so we believe it is important to constantly cross-check results across methodologies,” they wrote.

Kampmann believes the study provides a detailed molecular handle on which types of cells are most vulnerable. It also tightens the correlation between tau and cell death, albeit without clinching causality. “What our study does say is that the same types of neurons most likely to accumulate tau are also those that die first,” he said.

In regard to RORB, Kampmann said that cell-culture studies will reveal whether this transcription factor causes vulnerability, or merely marks cells that are vulnerable for a different, as-yet-unknown reason. Using CRISPR and iPSC-based cell models, he plans to investigate the role of RORB and other genes in rendering neurons susceptible to death. Grinberg and Kampmann are also expanding their snRNA-Seq studies to samples from women and people with different ApoE genotypes, as well as different regions of the brain, including subcortical areas gripped by tau pathology even before the entorhinal cortex.—Jessica Shugart

Comments

  1. In Alzheimer’s disease, which is so complex, selective cellular vulnerability is a key anchoring point for mechanistic discovery. To leverage it, we must understand what differs between vulnerable and spared cells as disease progresses.

    Single-nuclear RNA sequencing of human disease tissue brings incredible new opportunities here—but specimen selection and validation are paramount. By using highly curated neuropathological specimens, and validation across independent studies, this important work sets an exciting foundation for continued discovery.

  2. This report is intriguing and reinforces the power of single-nucleus studies of human postmortem brain tissue for understanding neurodegenerative disease. The identification of RORB as a possibly critical protein associated with selective vulnerability may lead to new mechanistic insights into neurofibrillary tangle formation, if confirmed in other single-nucleus studies and hopefully as well by studies using other approaches to single-cell or defined-cell-type analysis.

    The authors used CP13, marker of early to middle stages of neurofibrillary tangle development (Vingtdeux et al., 2011). Comparison with a late-stage tangle marker such as PHF-1 might be of assistance in sorting out the stage at which RORB becomes involved.

    One of the major deficiencies of single-nucleus studies, however, is that there is the assumption that the nuclear transcriptome will be a useful reflection of the cytoplasmic transcriptome. The nuclear transcriptome differs in that it contains many pre-mRNAs that are not represented in the cytoplasm. Nuclear transcripts may pass out of the nucleus relatively quickly into the cytoplasm, where they may then accumulate and persist over relatively long time periods. Complex regulatory processes exist to ensure that nuclear transcript synthesis maintains a steady-state cytoplasmic abundance, and generally greater cytoplasmic transcript half-lives mean that nuclear synthesis rates and steady-state concentrations are likely to be much lower (Timmers and Tora, 2018). Especially of interest amongst cytoplasmic transcripts are those with roles in presynaptic terminals. Nuclear transcripts may not give much or any information regarding these as both their abundance and their rate of translation to protein could be greater within synaptic terminals than elsewhere in the cytoplasm or nucleus (Jung et al., 2014; Jung and Holt, 2011; Kim and Jung, 2020). 

    All single-cell methods need to take into account the likelihood that some cells or nuclei or transcripts may be lost in processing or will have processing-related abundance changes and these losses and changes are likely to be selective rather than general. We have tried to approach this by comparing deconvoluted transcriptomes from isolated whole cells with those from adjacent bulk tissue homogenates. One of the more remarkable results, reported by ourselves and at least one other group, is that the cell isolation and/or sorting protocols can cause significant artifactual upregulation of microglial transcripts and that these may mask in vivo group differences. (Serrano et al., 2021; Kang et al., 2018; Kang et al., 2017). It is possible that this effect may explain the lack of microglial transcript changes in this interesting report by Leng et al.

    All methods have strengths and weaknesses and so we believe it is important to constantly cross-check results across methodologies. We encourage readers to compare the results from Leng et al. with our group’s published results on laser-captured tangle-bearing and non-tangle-bearing neurons (Dunckley et al., 2006), as well the cited reference (Liang et al., 2008) and several others (Liang et al., 2007; Liang et al., 2010; Liang et al., 2008; Liang et al., 2007; Mastroeni et al., 2018; Stamper et al., 2008). 

    The human brain material was obtained from autopsies with a very wide range in postmortem intervals (PMI), from 12 to 50 hours. Longer PMIs will result in progressive loss of selective transcripts that may be tissue, region, gene, and even genotype-dependent (Zhu et al., 2017; Birdsill et al., 2011; Walker et al., 2016). Did the authors attempt to determine, within their cases, whether PMI influenced the abundance of any transcripts, and, in particular, any of the transcripts identified as marking selectively vulnerable neurons?

    Although the authors examined brain tissue from several Braak stages of AD, the subjects had various combinations of pathology, including absence of tau and Aβ, presence of tau but not Aβ, presence of Aβ but not tau, and presence of both. To isolate the changes relating to these two major types of AD pathology, future studies should look for expression changes associated with disease progression involving only Aβ or only tau, or in a group with both changes present from the start.

    Additionally, it is well known that AD is most commonly complicated by one or more other major age-related molecular or vascular pathologies (Beach and Malek-Ahmadi, 2020) that have significant clinical effects. These would presumably have significant effects on gene expression as well, but so far there have not been concerted efforts to incorporate the full complexity inherent to the aging brain.

    References:

    . AMPK is abnormally activated in tangle- and pre-tangle-bearing neurons in Alzheimer's disease and other tauopathies. Acta Neuropathol. 2011 Mar;121(3):337-49. PubMed.

    . Transcript Buffering: A Balancing Act between mRNA Synthesis and mRNA Degradation. Mol Cell. 2018 Oct 4;72(1):10-17. PubMed.

    . Remote control of gene function by local translation. Cell. 2014 Mar 27;157(1):26-40. PubMed.

    . Local translation of mRNAs in neural development. Wiley Interdiscip Rev RNA. 2011 Jan-Feb;2(1):153-65. Epub 2010 Oct 19 PubMed.

    . Local mRNA translation in long-term maintenance of axon health and function. Curr Opin Neurobiol. 2020 Feb 19;63:15-22. PubMed.

    . Whole-Cell Dissociated Suspension Analysis in Human Brain Neurodegenerative Disease: A Pilot Study. medRxiv. January 9, 2021. medRxiv.

    . Microglial translational profiling reveals a convergent APOE pathway from aging, amyloid, and tau. J Exp Med. 2018 Sep 3;215(9):2235-2245. Epub 2018 Aug 6 PubMed.

    . Translational profiling of microglia reveals artifacts of cell sorting. bioRxiv, May 9, 2017.

    . Gene expression correlates of neurofibrillary tangles in Alzheimer's disease. Neurobiol Aging. 2006 Oct;27(10):1359-71. PubMed.

    . Altered neuronal gene expression in brain regions differentially affected by Alzheimer's disease: a reference data set. Physiol Genomics. 2008 Apr 22;33(2):240-56. PubMed.

    . Gene expression profiles in anatomically and functionally distinct regions of the normal aged human brain. Physiol Genomics. 2007 Feb 12;28(3):311-22. Epub 2006 Oct 31 PubMed.

    . Neuronal gene expression in non-demented individuals with intermediate Alzheimer's Disease neuropathology. Neurobiol Aging. 2010 Apr;31(4):549-66. PubMed.

    . Alzheimer's disease is associated with reduced expression of energy metabolism genes in posterior cingulate neurons. Proc Natl Acad Sci U S A. 2008 Mar 18;105(11):4441-6. Epub 2008 Mar 10 PubMed.

    . Laser-captured microglia in the Alzheimer's and Parkinson's brain reveal unique regional expression profiles and suggest a potential role for hepatitis B in the Alzheimer's brain. Neurobiol Aging. 2018 Mar;63:12-21. Epub 2017 Nov 1 PubMed.

    . Neuronal gene expression correlates of Parkinson's disease with dementia. Mov Disord. 2008 Aug 15;23(11):1588-95. PubMed.

    . Systematic analysis of gene expression patterns associated with postmortem interval in human tissues. Sci Rep. 2017 Jul 14;7(1):5435. PubMed.

    . Postmortem interval effect on RNA and gene expression in human brain tissue. Cell Tissue Bank. 2011 Nov;12(4):311-8. PubMed.

    . Characterization of RNA isolated from eighteen different human tissues: results from a rapid human autopsy program. Cell Tissue Bank. 2016 Sep;17(3):361-75. Epub 2016 Apr 16 PubMed.

    . Alzheimer's Disease Neuropathological Comorbidities are Common in the Younger-Old. J Alzheimers Dis. 2021;79(1):389-400. PubMed.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Aggregation-Prone Gene Expression Signature Mapped in Brain
  2. When It Comes to Alzheimer’s Disease, Do Human Microglia Even Give a DAM?
  3. Single-Cell Expression Atlas Charts Changes in Alzheimer’s Entorhinal Cortex
  4. Hot DAM: Specific Microglia Engulf Plaques
  5. ApoE and Trem2 Flip a Microglial Switch in Neurodegenerative Disease
  6. Single-nucleus RNA Sequencing Misses Activation of Human Microglia

Paper Citations

  1. . RORβ induces barrel-like neuronal clusters in the developing neocortex. Cereb Cortex. 2012 May;22(5):996-1006. Epub 2011 Jul 28 PubMed.
  2. . Mutually repressive interaction between Brn1/2 and Rorb contributes to the establishment of neocortical layer 2/3 and layer 4. Proc Natl Acad Sci U S A. 2016 Mar 22;113(12):3371-6. Epub 2016 Mar 7 PubMed.
  3. . Single-cell transcriptomic analysis of Alzheimer's disease. Nature. 2019 Jun;570(7761):332-337. Epub 2019 May 1 PubMed.
  4. . Molecular and cellular pathology of monogenic Alzheimer’s disease at single cell resolution. bioRxiv. July 14, 2020. BioRxiv.

Further Reading

Papers

  1. . Neuronal subtypes and diversity revealed by single-nucleus RNA sequencing of the human brain. Science. 2016 Jun 24;352(6293):1586-90. PubMed.

Primary Papers

  1. . Molecular characterization of selectively vulnerable neurons in Alzheimer's disease. Nat Neurosci. 2021 Feb;24(2):276-287. Epub 2021 Jan 11 PubMed.