RESOURCE https://doi.org/10.1038/s41593-020-00764-7 Molecular characterization of selectively vulnerable neurons in Alzheimer’s disease Kun Leng1,2,3,4,13, Emmy Li1,2,3,13, Rana Eser 5, Antonia Piergies5, Rene Sit2, Michelle Tan2, Norma Neff 2, Song Hua Li5, Roberta Diehl Rodriguez6, Claudia Kimie Suemoto7,8, Renata Elaine Paraizo Leite7, Alexander J. Ehrenberg 5, Carlos A. Pasqualucci7, William W. Seeley5,9, Salvatore Spina5, Helmut Heinsen7,10, Lea T. Grinberg 5,7,11 ✉ and Martin Kampmann 1,2,12 ✉ Alzheimer’s disease (AD) is characterized by the selective vulnerability of specific neuronal populations, the molecular sig- natures of which are largely unknown. To identify and characterize selectively vulnerable neuronal populations, we used single-nucleus RNA sequencing to profile the caudal entorhinal cortex and the superior frontal gyrus—brain regions where neurofibrillary inclusions and neuronal loss occur early and late in AD, respectively—from postmortem brains spanning the progression of AD-type tau neurofibrillary pathology. We identified RORB as a marker of selectively vulnerable excitatory neu- rons in the entorhinal cortex and subsequently validated their depletion and selective susceptibility to neurofibrillary inclusions during disease progression using quantitative neuropathological methods. We also discovered an astrocyte subpopulation, likely representing reactive astrocytes, characterized by decreased expression of genes involved in homeostatic functions. Our characterization of selectively vulnerable neurons in AD paves the way for future mechanistic studies of selective vulnerability and potential therapeutic strategies for enhancing neuronal resilience. elective vulnerability is a fundamental feature of neurodegen- Here, we performed snRNA-seq on postmortem brain tissue erative diseases, in which different neuronal populations show from a cohort of individuals spanning the progression of AD-type a gradient of susceptibility to degeneration. Selective vulner- tau neurofibrillary pathology to characterize changes in the rela- S 1–3 ability at the network level has been extensively explored in AD . tive abundance of cell types and cell-type subpopulations. We However, little is known about the mechanisms underlying selective discovered a selectively vulnerable subpopulation of excitatory vulnerability at the cellular level in AD, which could provide insight neurons in the EC and validated the selective depletion of this into disease mechanisms and lead to therapeutic strategies. subpopulation during AD progression with quantitative histo- The entorhinal cortex (EC) is one of the first cortical brain pathology, using multiplex immunofluorescence in EC regions regions to exhibit neuronal loss in AD4. Neurons in the external EC delineated by rigorous cytoarchitectonic criteria. Furthermore, layers, especially in layer II, accumulate tau-positive neurofibril- we uncovered an astrocyte subpopulation likely corresponding to lary inclusions and die early in the course of AD5–10. However, these reactive astrocytes that showed downregulation of genes involved selectively vulnerable neurons have yet to be characterized at the in homeostatic function. molecular level. Furthermore, it is unknown whether there are dif- ferences in vulnerability among subpopulations of these neurons. Results Although rodent models of AD have offered important insights, the Cohort selection and cross-sample alignment. We performed available models fail to simultaneously capture some critical disease snRNA-seq on cell nuclei extracted from postmortem brain tissue processes, such as the accumulation of neurofibrillary inclusions (Methods) from the EC at the level of the mid-uncus area and from and neuronal loss11, limiting the extrapolation of findings from the superior frontal gyrus (SFG) at the level of the anterior com- rodent models to address selective vulnerability. missure (Brodmann area 8), from ten male individuals carrying Recently, single-nucleus RNA sequencing (snRNA-seq) has the apolipoprotein E gene (APOE) ε3/ε3 genotype representing the enabled large-scale characterization of transcriptomic profiles of cortical-free, early and late stages of AD-type tau neurofibrillary individual cells from postmortem human brain tissue12,13. However, pathology (Braak stages1 0, 2 and 6; Fig. 1a and Table 1). The neuro- snRNA-seq studies of AD published to date have focused on pathological hallmarks of AD are amyloid plaques and neurofibril- cell-type-specific differential gene expression between individu- lary inclusions. Since the accumulation of neurofibrillary inclusions als with AD and healthy controls14,15, without explicitly addressing measured by the Braak staging system is the best correlate of clinical selective vulnerability. cognitive decline, after neuronal loss16, we reasoned that profiling 1Institute for Neurodegenerative Disease, University of California, San Francisco, San Francisco, CA, USA. 2Chan Zuckerberg Biohub, San Francisco, CA, USA. 3Biomedical Sciences Graduate Program, University of California, San Francisco, San Francisco, CA, USA. 4Medical Scientist Training Program, University of California, San Francisco, San Francisco, CA, USA. 5Memory and Aging Center, Weill Institute for Neurosciences, Department of Neurology, University of California, San Francisco, San Francisco, CA, USA. 6Department of Neurology, Universidade de São Paulo, Faculdade de Medicina, São Paulo, Brazil. 7Department of Pathology, Universidade de São Paulo, Faculdade de Medicina, São Paulo, Brazil. 8Division of Geriatrics, Department of Clinical Medicine, Universidade de São Paulo, Faculdade de Medicina, São Paulo, Brazil. 9Department of Pathology, University of California, San Francisco, San Francisco, CA, USA. 10Department of Psychiatry, University of Würzburg, Würzburg, Germany. 11Global Brain Health Institute, University of California, San Francisco, San Francisco, CA, USA. 12Department of Biochemistry and Biophysics, University of California, San Francisco, San Francisco, CA, USA. 13These authors contributed equally: Kun Leng, Emmy Li. ✉e-mail: [email protected]; [email protected] 276 NatURE NEUROscIENCE | VOL 24 | FEBRuaRY 2021 | 276–287 | www.nature.com/natureneuroscience NATURE NEUROSCIENCE RESOURCE a AD-tau neurofibrillary pathology Isolation of Generation of Next- Reverse Library Data analysis SFG cell nuclei droplets with generation transcription generation barcoded beads sequencing EC (n = 3) (n = 4) (n = 3) Stage 0 Stage 2 Stage 6 AD-tau neuropathological progression (Braak stage) b Participant number c EC 3 Braak SFG 1 EC:Exc.2 stage SFG:Oligo.2 2 0 EC:OPC EC:Exc.3 EC:Exc.4 SFG:Astro.1 5 SFG:Inh.4 SFG:Oligo.1 EC:Exc.5 SFG:Inh.1 EC:Endo EC:Inh.1 6 Braak stage SFG:Astro.2 SFG:Exc.1 SFG:Exc.2 EC:Exc.1 EC:Inh.3 7 SFG:Endo EC:Astro EC:Inh.2 2 SFG:Exc.7 SFG:Inh.2 SFG:Exc.5 4 SFG:Exc.6 EC:Oligo SFG:OPC SFG:Exc.3 EC:Micro 8 SFG:Exc.4 Braak SFG:Micro 9 stage SFG:Inh.3 10 6 d e Relative expression High Low SLC17A7 SLC17A7 GAD1 GAD1 MBP MBP CLDN5 CLDN5 PDGFRA PDGFRA AQP4 AQP4 CD74 CD74 log-scaled normalized counts log-scaled normalized counts 2.5 4 Braak stage (× 1,000) (× 1,000) No. of cells 0 0 No. of cells 0 4 2 6 6 (× 1,000) 0 (× 1,000) 0 No. of genes No. of genes EC:OPC EC:Inh.1 EC:Inh.4 EC:Inh.3 EC:Inh.2 EC:OPC EC:Inh.2 EC:Inh.1 EC:Inh.3 EC:Endo EC:Oligo EC:Endo EC:Astro EC:Micro EC:Micro EC:Exc.2 EC:Exc.6 EC:Exc.4 EC:Exc.5 EC:Exc.1 EC:Exc.7 EC:Exc.3 EC:Exc.3 EC:Exc.4 EC:Exc.1 EC:Exc.5 EC:Exc.2 EC:Oligo.2 EC:Oligo.1 EC:Astro.1 EC:Astro.2 f g Braak Braak Comparison P stage adjusted 0.8 stage 3 0.75 0 0 1 0.08 0.6 0.50 1 2 2 0.4 2 0.25 2 5 × 10–10 0.2 Relative 6 Relative 6 abundance 0 abundance 0 3 0.03 Exc Inh Exc Inh Oligo Astro OPC Micro Endo Oligo Astro OPC Micro Endo Cell type Cell type Fig. 1 | AD progression differentially affects the cell-type composition of the EC and SFG. a, Schematic of experimental design and sample processing. Darker shades of red in brain cartoons reflect more severe AD-tau neurofibrillary pathology. b,c, t-SNE projection of cells from the EC (b) and SFG (c) in their respective alignment spaces, colored by individual of origin (center) or cluster assignment (outer). d,e, Heat maps and hierarchical clustering of clusters and cluster marker expression (top); ‘high’ and ‘low’ relative expression reflect above- and below-average expression, respectively (Methods). Expression of cell-type markers in each cluster (second from top). Bar graphs show the average number of cells and average number of genes detected per cell in each cluster. f,g, Relative abundance of major cell types across Braak stages. For each brain region, statistical significance of differences in relative abundance across Braak stages (Braak 0: n = 3, Braak 2: n = 4, Braak 6: n = 3; n, number of participants sampled) was determined by beta regression and adjusted for multiple comparisons (Methods). Cell-type abbreviations: exc, excitatory neurons; oligo, oligodendrocytes; astro, astrocytes; inh, inhibitory neurons; OPC, oligodendrocyte precursor cells; micro, microglia; endo, endothelial cells. NatURE NEUROscIENCE | VOL 24 | FEBRuaRY 2021 | 276–287 | www.nature.com/natureneuroscience 277 RESOURCE NATURE NEUROSCIENCE Table 1 | Description of postmortem cohort Participants used for snRNA-seq Participant no. Braak stage Sex Age at death Postmortem interval ADNC score CDR before APOE genotype Source (years) (h) death 1 0 M 50 13 A0, B0, C0 0 ε3/ε3 BBAS 2 0 M 60 12 A0, B0, C0 0.5 ε3/ε3 BBAS 3 0 M 71 12 A1, B0, C0 0 ε3/ε3 BBAS 4 2 M 72 15 A1, B1, C0 0 ε3/ε3 BBAS 5* 2 M 77 4.9 A2, B1, C1 0.5 ε3/ε3 UCSF 6* 2 M 87 30 A2, B1, C2 2 ε3/ε3 UCSF 7* 2 M 91 50 A1, B1, C1 0 ε3/ε3 UCSF 8* 6 M 72 6.9 A3, B3, C3 3 ε3/ε3 UCSF 9* 6 M 82 6.7 A3, B3, C3 3 ε3/ε3 UCSF 10 6 M 82 9 A3, B3, C3 3 ε3/ε3 UCSF Participants used for immunofuorescence validation Participant no.