. Neuronal ApoE upregulates MHC-I expression to drive selective neurodegeneration in Alzheimer's disease. Nat Neurosci. 2021 Jun;24(6):786-798. Epub 2021 May 6 PubMed.


Please login to recommend the paper.


  1. This fascinating paper from the Huang lab describes a novel role for neuronal ApoE expression in Alzheimer’s disease.

    Zalocusky and colleagues find that neuronal ApoE is linked to immune response activation and selective neurodegeneration. Specifically, they show that increased ApoE drives the expression of MHC-I, which leads to tau pathology as well as synapse and neuronal loss. These compelling data provide mechanistic insight to cell-type vulnerability in Alzheimer’s disease. Because the relationship between neuronal ApoE and immune gene expression is also found in individuals without pathology or neurodegeneration, it will be useful to define how ApoE interacts with disease-associated biological pathways to drive degeneration phenotypes. It will also be interesting to see how microglia and other glial cells respond to aberrant MHC-I expression.

    Overall, this is an important and thought-provoking paper that raises numerous questions regarding the mechanisms of cell type vulnerability and neuroinflammation.

    View all comments by Li-Huei Tsai
  2. Selectivity of neuronal loss under disease conditions is still a mystery. In the present study, Zalocusky and colleagues attribute a key role to neuronal expression of APOE in regulating MHC-I expression, and thereby, selective neuronal vulnerability. This new concept regarding neuronal vulnerability is interesting in light of the emerging understanding of the role displayed by the immune system in supporting life-long brain plasticity and repair (Moalem et al., 1999; Rapalino et al., 1998; Ziv et al., 2006; Baruch et al., 2015; Filiano et al., 2016), including its effects in neurodegenerative diseases.

    Neuronal expression of APOE (Wang et al., 2018; Xu et al., 1999) and MHC-I (Lee et al., 2014) are new observations and are still far from being fully understood. Specifically, ApoE4 expression has been identified as a risk factor in human patients suffering from AD; the severity of the disease is greater in patients carrying APOE4 allele (Tao et al., 2018). In a mouse model of tauopathy, human ApoE expression—especially of the ApoE4—led to increased tau pathology, neuroinflammation, and neuronal loss.

    Regarding MHC-I, it has generally been presumed that it is not expressed by neurons under physiological conditions (Cebrián  et al., 2014). The initial study that showed MHC-I gene expression by neurons in response to IFN-γ was reported in 1995 (Neumann et al., 1995). Other studies suggested that neuronal expression of this molecule is involved in neuroinflammatory processes, and participates in immune-mediated neurodegeneration. However, accumulating data have subsequently demonstrated MHC-I expression by subsets of neurons in both adult and developing mammalian brains, even under physiological conditions (Lindå  et al., 1999; Huh et al., 2000; Letellier et al., 2008). Neuronal MHC-I expression was further implicated in models of axonal regeneration, where it was shown to  regulate the ability of neurons to regenerate axons (Sabha et al., 2008). Moreover, it was suggested that neuronal expression of MHC-I during the first week after sterile injury enhances axonal regeneration, and these findings support the observation that elevated neuronal MHC-I expression promotes the recovery of locomotor abilities after spinal cord injury (Joseph et al., 2011). In light of these results, it is clear that expression of MHC-I by neurons it is not all or none, good or bad, but rather, is dependent on context, timing, and persistence.

    The proposed mechanism in the present study, that neuronal ApoE expression might be an important factor driving within-neuron-type variability under both normal physiological and pathophysiological conditions, is intriguing. The authors further showed that reducing functional MHC-I by knocking down or knocking out B2M, a protein required for functional expression of all MHC-I genes, is sufficient to significantly reduce tau pathologies in vitro in cultured primary neurons, or in vivo in a pathological tau-P301S overexpression mouse model.

    In this regard, it is interesting to note that B2M expression was found to be dependent on Type I interferon (Baruch et al., 2014; Deczkowska et al., 2017), the level of which is chronically elevated in aging and neurodegenerative conditions; it is therefore quite possible that MHC-I expression is regulated by Type I inflammation. While the functional connection between APOE and MHC-I expression by neurons is interesting, further studies are needed to fully dissect out the causal relationships connecting APOE expression with adaptive immunity, and the implications to all forms of APOE.


    . Autoimmune T cells protect neurons from secondary degeneration after central nervous system axotomy. Nat Med. 1999 Jan;5(1):49-55. PubMed.

    . Implantation of stimulated homologous macrophages results in partial recovery of paraplegic rats. Nat Med. 1998 Jul;4(7):814-21. PubMed.

    . Immune cells contribute to the maintenance of neurogenesis and spatial learning abilities in adulthood. Nat Neurosci. 2006 Feb;9(2):268-75. PubMed.

    . Breaking immune tolerance by targeting Foxp3(+) regulatory T cells mitigates Alzheimer's disease pathology. Nat Commun. 2015 Aug 18;6:7967. PubMed.

    . Unexpected role of interferon-γ in regulating neuronal connectivity and social behaviour. Nature. 2016 Jul 21;535(7612):425-9. Epub 2016 Jul 13 PubMed.

    . Gain of toxic apolipoprotein E4 effects in human iPSC-derived neurons is ameliorated by a small-molecule structure corrector. Nat Med. 2018 May;24(5):647-657. Epub 2018 Apr 9 PubMed.

    . Specific regional transcription of apolipoprotein E in human brain neurons. Am J Pathol. 1999 Feb;154(2):601-11. PubMed.

    . Synapse elimination and learning rules co-regulated by MHC class I H2-Db. Nature. 2014 May 8;509(7499):195-200. Epub 2014 Mar 30 PubMed.

    . Association of Chronic Low-grade Inflammation With Risk of Alzheimer Disease in ApoE4 Carriers. JAMA Netw Open. 2018 Oct 5;1(6):e183597. PubMed.

    . Neuronal MHC-I expression and its implications in synaptic function, axonal regeneration and Parkinson's and other brain diseases. Front Neuroanat. 2014;8:114. Epub 2014 Oct 13 PubMed.

    . Induction of MHC class I genes in neurons. Science. 1995 Jul 28;269(5223):549-52. PubMed.

    . Expression of MHC class I heavy chain and beta2-microglobulin in rat brainstem motoneurons and nigral dopaminergic neurons. J Neuroimmunol. 1999 Nov 1;101(1):76-86. PubMed.

    . Functional requirement for class I MHC in CNS development and plasticity. Science. 2000 Dec 15;290(5499):2155-9. PubMed.

    . Normal adult climbing fiber monoinnervation of cerebellar Purkinje cells in mice lacking MHC class I molecules. Dev Neurobiol. 2008 Jul;68(8):997-1006. PubMed.

    . MHC I expression and synaptic plasticity in different mice strains after axotomy. Synapse. 2008 Feb;62(2):137-48. PubMed.

    . Transgenic mice with enhanced neuronal major histocompatibility complex class I expression recover locomotor function better after spinal cord injury. J Neurosci Res. 2011 Mar;89(3):365-72. Epub 2010 Dec 22 PubMed.

    . Aging-induced type I interferon response at the choroid plexus negatively affects brain function. Science. 2014 Aug 21; PubMed.

    . Mef2C restrains microglial inflammatory response and is lost in brain ageing in an IFN-I-dependent manner. Nat Commun. 2017 Sep 28;8(1):717. PubMed.

    View all comments by Michal Schwartz
  3. This is another pearl from the Yadong Huang lab at the Gladstone Institutes/UCSF cementing the hypothesis that neurons express ApoE, that this expression is genotype-dependent (APOE4 more expressed than APOE3), and that it is detrimental and linked to pathology. Most in the field initially dismissed a (patho)physiological role for neuronal ApoE in neurodegeneration, but the Huang lab provided ever-increasing experimental evidence employing, in this current study, single cell/nucleus RNA sequencing of individual neurons. The results leave little doubt that neurons can express substantial levels of APOE transcripts in humanized mice and human postmortem brains and that this expression is linked with pathology. The study also opens a glimpse in what is likely to emerge from single-cell molecular profiling of brain cells in the years to come, and how it may finally provide answers to the perennial question of “selective cellular vulnerability.”

    Intriguingly, ApoE expression at the single cell/nucleus level strongly correlates with expression of immune pathway genes, most notably MHC class I. Genetic deletion of human APOE in neurons of APOE humanized mice protects them from ApoE4 induced neurodegeneration and lowered MHC-I levels as well as related genes including Tap2, which is involved in peptide loading on the MHC-I/b2-microglobulin (B2M) complex.

    Zalocusky and colleagues further show that ApoE expression correlates with tau pathology—an observation they made earlier—and that this is MHC-I dependent. Because B2M is necessary for stable cell surface expression of MHC-I they were able to lower expression of MHC-I by genetically reducing B2M expression in neurons and show that this too, leads to reduced tau pathology in cell culture and in mice. Together, the authors propose a cascade whereby APOE4 and/or stress increase neuronal ApoE expression, which then leads to increased MHC-I surface expression and subsequent abnormal tau phosphorylation and neurodegeneration.

    This exciting new pathway offers potential new therapeutic targets for neurodegenerative diseases. It is interesting that B2M rises prominently with normal aging, impairs neurogenesis, and promotes cognitive impairment in young mice, while B2M deletion improves cognition in aged mice (Smith et al., 2015). Our recent observation that virus-specific, CD8-positive, T effector cells are clonally expanded in the cerebrospinal fluid of Alzheimer's disease patients, and can be localized to degenerating neurons in brain parenchyma, opens the possibility that ApoE may promote neuronal susceptibility to potential T cell killing by upregulating the cognate receptor (i.e., MHC-I-B2M-peptide) for virus-specific T cells or T cells of yet unknown specificity for neuronal antigens (Gate et al., 2020). 

    ApoE-driven MHC-I expression, as the authors discussed, could also interfere with synaptic plasticity as laid out in the elegant cascade uncovered by Carla Shatz and colleagues at Harvard Medical School (Huh et al., 2000). Future studies need to disentangle these possibilities and assign pathological relevance to them; equally important, we need to understand how ApoE regulates MHC-I expression and how this promotes tau phosphorylation at a mechanistic molecular level.


    . β2-microglobulin is a systemic pro-aging factor that impairs cognitive function and neurogenesis. Nat Med. 2015 Aug;21(8):932-7. Epub 2015 Jul 6 PubMed.

    . Clonally expanded CD8 T cells patrol the cerebrospinal fluid in Alzheimer's disease. Nature. 2020 Jan;577(7790):399-404. Epub 2020 Jan 8 PubMed.

    . Functional requirement for class I MHC in CNS development and plasticity. Science. 2000 Dec 15;290(5499):2155-9. PubMed.

    View all comments by Tony Wyss-Coray
  4. An important finding in this study is the demonstration of clear correlations between neuronal ApoE expression and neuronal phenotype in humanized ApoE-KI mouse models, involving shifts in cellular metabolism, immune response, and pathways linked to neurodegenerative diseases. Neuronal pathways involved in cellular metabolism and immune response also correlated with ApoE expression in human patients with MCI or AD. Another key observation is that in humanized ApoE-KI mice expressing ApoE3 or ApoE4, the subpopulation of high-ApoE-expressing neurons increases with age at first, and then sharply declines. Interestingly, they show that both the increase and the decline occur sooner in ApoE4-KI mice compared to ApoE3-KI mice. In addition, results show a correlation in neurons between ApoE and MHC-I expression, suggestive of neuroimmune regulatory pathway interactions.

    While the cellular gene-signature approaches used as a basis for the authors’ explorations are very useful, there are several weaknesses in the representation of the data and the conceptual over-simplifications that follow. The conclusion that ApoE expression increases in neurons that are pathological may be too simplistic, since the increases more likely reflect a change from physiological control of lipid transfer from astrocytes to neurons—a pathophysiological adaptation to the failure of glial coping mechanisms against other causes of neuronal stress. High-ApoE-expressing neurons do not mean ApoE causes vulnerability.

    The authors describe populations of high-ApoE-expressing neurons in MCI and AD patients and interpret this as evidence that increased ApoE expression causes selective vulnerability. While these high-ApoE-expressing neurons may be selectively vulnerable, they do not consider that increased ApoE expression may itself be a stress response, rather than pathological.

    Conceptually critical, there is no evidence presented in the data that the known aging and AD regional brain and individual neuronal vulnerability (relative vulnerability) differences are reflected in the measured ApoE expression profiles. Instead, the authors treat all “neurons” as equal in the brain. Nuclear neuronal gene signatures are all lumped together, while we know that MCI and AD have regional, cellular, and staged pathological progression that depends on the individual neuronal cell types. Such neuronal vulnerability is not necessarily “selective,” but relative. Not all brain neurons degenerate in aging, MCI, and AD. Brain neurons are heterogenous and can be categorized by neurotransmitter, anatomy, and gene-expression profiles. The relative neuronal cell type degeneration risks vary for each neuronal cell type and brain region depending on the underlying and real vulnerability to neuronal aging and AD risk factors. Therefore, the data shown, that neurons in general increase (no regional differences/clustering were observed) their ApoE expression may simply reflect initial neuronal adaptive responses to the underlying neuronal and glial causes of pathology, independent of ApoE levels.

    The association found in the report between ApoE levels and MHC-I is interesting due to the well-known role of ApoE in resolution of inflammatory processes and metabolic control under stress. Numerous previous studies have demonstrated that ApoE can cross-talk with components of the immune system, including C1q. These prior relevant pathophysiological results are not discussed or referenced. For example, Yin et al demonstrated that ApoE can functionally attenuate C1q activation under physiological lipid stress in vivo (Yin et al., 2019). Also not mentioned are other important studies (by Maria Ioannou in particular) showing that there is a necessary functional cross-talk between astrocytes and neurons in lipid transfer, where the astrocytes play the normal key physiological role (Ioannou et al., 2019Qi et al., 2021; Sienski et al., 2021). Given that the authors themselves show that the majority of ApoE expression (~ fivefold higher) occurs in astrocytes/glia cell clusters, also with time, the changes observed in neurons are likely the tip of the iceberg in terms of the actual ApoE lipid and lipid transfer in a pathophysiological context.

    In conclusion, while this article has numerous interesting observations, key questions remain about the pathophysiological roles of the lipid binding protein ApoE and its human allele forms ApoE 2, 3, and 4, and how to mitigate the inherent risk of each ApoE form or their loss of function.


    . Neuron-Astrocyte Metabolic Coupling Protects against Activity-Induced Fatty Acid Toxicity. Cell. 2019 May 30;177(6):1522-1535.e14. Epub 2019 May 23 PubMed.

    . ApoE4 Impairs Neuron-Astrocyte Coupling of Fatty Acid Metabolism. Cell Rep. 2021 Jan 5;34(1):108572. PubMed.

    . APOE4 disrupts intracellular lipid homeostasis in human iPSC-derived glia. Sci Transl Med. 2021 Mar 3;13(583) PubMed.

    . ApoE attenuates unresolvable inflammation by complex formation with activated C1q. Nat Med. 2019 Mar;25(3):496-506. Epub 2019 Jan 28 PubMed.

    View all comments by Ole Isacson

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. Does ApoE in Neurons Drive Selective Vulnerability in Alzheimer’s?