Mechanisms contributing to differential genetic risks for TREM2 R47H and R62H variants in Alzheimer’s Disease

Nurun Fancy, Nanet Willumsen, Stergios Tsartsalis, Combiz Khozoie, Aisling McGarry, Robert C Muirhead, Eleonore Schneegans, Karen Davey, Vicky MN Chau, Amy M Smith, William Scotton, John Hardy, Dann Huh, Paul M Matthews, Johanna S Jackson
doi: https://doi.org/10.1101/2022.07.12.22277509

SUMMARY

Coding variants in the microglial TREM2 ectodomain differentially (R47H> R62H) increase the risk of Alzheimer’s disease (AD). To define mechanisms responsible, we characterised neuropathology and transcriptomic responses in heterozygotes for these TREM2 variant alleles (TREM2var) and for common allele homozygotes (CV) in non-diseased and AD brain cortical tissue from 58 donors. Increased neurodegeneration in the TREM2var AD cortex was associated with genotype-dependent reductions in expression of Disease Associated Microglia (DAM) genes and increased expression of complement and Type I and II interferon pathways in microglia, phagocytosis and amyloid binding pathways and Disease Associated Astrocyte (DAA) genes in astrocytes, and growth factor, ubiquitination and apoptotic pathways in neurons. The microglial phenotypes and secondary differences in tissue β-amyloid deposition and in astrocyte and neuronal responses describe a variably partial loss of TREM2 function with variant alleles (R47H>R62H) relative to CV and suggest mechanisms that could account for differences in genetic risks conferred.

INTRODUCTION

APOE polymorphisms and TREM2 coding variants confer the greatest independent genetic risks for late onset AD1. TREM2 is expressed primarily in microglia, in which functional interactions between APOE and TREM2 regulate β-amyloid clearance, lipid homeostasis and inflammatory responses2. Higher soluble TREM2 CSF levels are associated with slower hippocampal atrophy and cognitive decline in APOE4 carriers3. Expression of rare TREM2 variant (TREM2var) alleles affecting the coding sequence is associated with increased AD risk4,5. The R47H (rs75932628, minor allele frequency [MAF]=0.2% [ALFA]) variant confers the greatest relative AD risk (∼3.9 times that of homozygous common alleles [CV]). Carriers show earlier symptoms, faster cognitive decline and greater CSF hyperphosphorylated-tau (pTau) levels6. The R62H (rs143332484) variant is more common (MAF=0.97%) and has a smaller effect on disease risk (∼1.4 times CV)7,8. These data suggest that TREM2 has a protective effect for AD that is reduced with risk variants.

Functional changes with TREM2 KO or the R47H TREM2 allelic variant have been characterised in microglia and in animal models. β-amyloid clearance is reduced in TREM2 KO mice9 with increased expression of stress-associated autophagocytic vesicles, impaired microglial lipid metabolism and decreased survival10–12. Fewer studies have described the secondary consequences of TREM2 KO or TREM2var expression on astrocytes and neurons. Initial studies of astrocytes in mouse TREM2 KO models suggested few effects on their transcriptomic signatures13, although both TREM2 KO or R47H human β-amyloid models show an increased neuritic plaque density9. TREM2 KO in a humanised microglia mouse tauopathy model led to greater tau hyperphosphorylation and dysregulation of neuronal stress kinase pathways14. Fewer studies of the impact of these variants in human tissue have been reported. These have either not shown differences or have reported increased amyloid load and glial inflammatory activation with TREM2var relative to CV15–18. Neuronal tau pathology with TREM2 R47H and R62H was reported to show no change15,16 or a decrease17 relative to CV. Studies of human AD post-mortem tissue with both variants have described greater neuronal loss11,17,19.

Mechanisms by which the differences in AD risk are conferred by TREM2var have not been characterised. Both TREM2var have decreased affinity for β-amyloid, which has been related to reduced microglial β-amyloid phagocytosis and degradation in vitro for R47H20. Responses to lipid ligands in a reporter cell line were less impaired with R62H than with R47H21. We reasoned that contrasts between cellular responses in human post-mortem brain with and without AD pathology would provide insights into the differential risks between TREM2var and the CV. Using immunohistochemistry (IHC) and imaging mass cytometry (IMC), we characterised neocortical β-amyloid and AT8/PHF1 pTau pathology with CV and the R47H or R62H TREM2var. We performed single nuclear RNA sequencing to test for differences in cell-type proportions and for the differential gene expression (DGE) in TREM2var and CV cortical tissues with greater 4G8+ β-amyloid- or PHF1+ pTau-immunostaining. This analysis identified novel disease-associated transcriptomic differences in microglia, associated astrocytic responses and secondary neuronal responses that can explain differences in the genetic AD risk conferred by these two TREM2 risk variants.

RESULTS

We performed histopathology and snRNAseq on post-mortem neocortical tissue from both the mid temporal gyrus (MTG) and the somato-sensorimotor (SOM) cortices of non-diseased control (NDC) donors with clinical and neuropathological diagnoses of AD. Tissues carried one of three TREM2 genotypes: CV (30 total: n=9, NDC; n=21, AD), R47H (11 total: n=2, NDC; n=9, AD) or R62H (17 total: n=5, NDC; n=12, AD) TREM2var (Fig. 1a, Table 1).

View this table:
Table 1: Demographic characteristics and neuropathological findings across genotypes and disease groups.

Post-mortem interval (PMI). RNA integrity value (RIN). Mean ± SD. *unknown for 3 cases, **unknown for 1 case.

Figure 1:
Figure 1: TREM2var leads to increases in amyloid pathology.

58 post-mortem tissue blocks from donors with CV or TREM2 variant (R47H, R62H) from two brain regions (MTG and SOM) were studied. Formalin fixed, paraffin embedded sections (FFPE) were used for histopathology and paired frozen tissue from the homologous regions in the contralateral hemisphere of the same brain was used for transcriptomics in each case (a, b; scale bars = 100 µm). TREM2var carriers showed an increase in the percentage β-amyloid (4G8+) immunostained area (b, c). Imaging mass cytometry (IMC) (d; scale bar = 100 µm) revealed an increase in the proportion of total plaques which were APOE+ (white arrows) with both variants in AD cases (e). IMC also showed increased numbers of dense plaques (relative to total plaques; f) and AT8+ dense plaques (relative to total plaques, g) in R62H. Insert shows a neuritic amyloid plaque distinguished as 4G8+ (purple) with associated AT8+ pTau (green) by IMC (scale bar = 50 µm). Box plots denote median value with 1st and 3rd quartiles. *p<0.05, **p<0.01, ***p<0.001.

Both TREM2 variant alleles are associated with increased amyloid pathology

Immunohistochemical analysis revealed a median ∼2-fold increase in cortical 4G8 β-amyloid burden with AD in TREM2var relative to CV across the combined neocortical regions (Fig. 1b, c). There was a mean ∼3-fold increase of APOE+ amyloid plaques in AD cases with R47H or R62H relative to CV (Fig. 1d, e). We found a ∼2-fold increase in the proportion of dense-cored plaques (Fig. 1d, f) and 3-fold more dense-core AT8+ neuritic plaques in R62H relative to CV, and with similar numbers in APOE4-positive or -negative genotypes (Fig. 1g,). We found a similar proportion of dense-cored and neuritic plaques in CV and R47H cases (Fig. 1f, g), even though the total β-amyloid load was higher with R47H (Fig. 1c).

Differences in neuronal tau hyperphosphorylation with TREM2var

AT8+ pTau immunoreactivity22 was increased in CV AD versus CV NDC (AD, 3.57±0.47%; NDC, 0.07±0.06% area [mean±SEM], p<0.0001). We found a ∼3-fold decrease in the relative AT8+ immunostained area for combined SOM and MTG from TREM2var carriers relative to CV (CV, 3.57±0.47%; TREM2var, 1.15±0.17% area, both mean±SEM, p<0.05) (Fig. 2a, b).

Figure 2:
Figure 2: TREM2var leads to changes in neuronal tau phosphorylation.

Histopathology revealed a significant reduction in the percentage section area immunostained with AT8+ pTau for both TREM2var in AD (a,b). No differences in staining area for the non-diseased control tissues were observed. Adjacent sections were stained for PHF1+ pTau (c). No differences in PHF1+ immunostained areas between common allele (CV) AD cases and those from TREM2var heterozygote donors were found (d). Scale bars = 100 µm. Box plots denote median value with 1st and 3rd quartiles. *p<0.05.

We tested whether the reduced AT8 binding at Ser202/Thr205/Ser20823 epitopes for TREM2var tissues reflected differences in the amount of pTau pathology by immunostaining adjacent sections with PHF1, which recognises tau phosphorylated at Ser396/Ser40424, and immunostaining a subset of the sections with Tau5 for assessment of total (phosphorylated and non-phosphorylated) tau. There were no significant differences in mean total Tau5 pTau or PHF1 pTau immunostaining between CV and either variant (Fig. 2c,d). The genotype-dependent differences in tau hyperphosphorylation suggests differences in neuronal stress-activation between TREM2var and CV25.

Greater neurodegeneration in AD with TREM2var

snRNAseq of the same neocortical regions in the paired, cryopreserved contralateral hemisphere for each brain allowed discrimination of transcriptomes for 22 cell types or sub-types (Fig. 3a). As both TREM2var had similar levels of β-amyloid and pTau pathology, we tested for differences in the relative abundance of nuclei extracted from NDC and AD tissues for CV and combined TREM2var, controlling for APOE genotype, brain region, post-mortem delay, sex and age in the regression model. We did not find significant differences in the relative abundance between CV and TREM2var in tissue from NDC, but numbers of excitatory (EN, mean relative to NDC, −13%) and inhibitory neuronal nuclei (IN, −22%) were reduced with AD in the combined TREM2var group versus CV (Fig. 3b). Numbers of nuclei were relatively reduced with AD for 4/6 EN and 3/5 IN sub-types with TREM2var (Fig. 3c). Similar analysis to determine the effects of APOE4 on cell type proportions showed similar changes in relative proportions of EN and IN numbers (Fig. 3d).

Figure 3:
Figure 3: Loss of neurons with TREM2var.

snRNAseq allowed discrimination of the transcriptomes of 22 different cell types (a) and Dirichlet analysis revealed significant decreases in the proportions of excitatory (EN) and inhibitory (IN) neuronal nuclei in the combined TREM2var nuclei relative to CV in AD but not in non-diseased control tissues. Most neuronal nuclei sub-type clusters were smaller in the TREM2 tissue (c). Stratification by APOE genotype revealed smaller reductions in neuronal proportions in AD cases for APOE4 heterozygotes (d) than was found in those who did not carry this allele (controlling for APOE4 genotype) (b). Box plots denote median value with 1st and 3rd quartiles. *p<0.05, **p<0.01, ***p<0.001

TREM2 genotype determines microglial transcriptional responses to β-amyloid

To understand how pathology affected DGE, we tested for different transcript expression correlated to greater neocortical β-amyloid or PHF1 immunostaining in paired FFPE sections. We found reduced numbers of DGE (normalized to total number expressed) with increasing β-amyloid for both TREM2var (3.3%, CV; 0.9%, R62H; 0.7% R47H) (Fig. 4a). Preserved differential expression of APOE and B2M with a reduced total DAM response suggested inhibition beyond the TREM2-independent stage26.

Figure 4:
Figure 4: Reduced microglial and increased astrocytic differential gene expression with greater tissue β-amyloid for TREM2var relative to common allele (CV) neocortical tissue.

TREM2var microglia showed genotype-dependent reductions in numbers of genes differentially expressed (logFC>0.25, padj<0.05) with greater 4G8+ β-amyloid immunostained area. This included reductions in numbers of up-regulated (red) and down-regulated (blue) DAM genes. Illustrative genes are highlighted below each volcano plot (a). In contrast, numbers of differentially expressed genes were increased in astrocytes for TREM2var. This included increases in up-regulated (red) and down-regulated (blue) Disease Associated Astrocyte (DAA) genes (b). Percentages denote differentially expressed genes as a proportion of total genes expressed.

TREM2var microglia had inflammatory transcriptional phenotypes distinct from those with the CV. CV microglia up-regulated ‘ERBB signalling’, ‘tyrosine kinase signalling’ and transcription and apoptosis pathways but pathways associated with adaptive immunity, cytokine production and signalling and the TYROBP network were down-regulated (Fig 5a). However, TREM2var microglia expressed pro-inflammatory transcriptional gene signatures. Microglial transcripts most highly differentially expressed with β-amyloid for both TREM2var, but not for CV, included MHC Class I and II molecules (HLA-A, HLA-DRA, HLA-DPA1)27 and pathways such as ‘MHC class II receptor activity’. Type II interferon pathways (‘interferon gamma signalling’) were up-regulated for both variants considered individually (Fig. 4c). ‘Cellular response to interferon-gamma’ ‘and ‘type I interferon signalling’ pathways were up-regulated in combined TREM2var microglial nuclei. Cytokine production and signalling pathways were up-regulated in R47H but not in R62H or CV. Interferon pathways were not significantly differentially expressed in CV. Expression of complement signalling genes (C1QA, C1QB, C1QC) and pathways (‘complement activation’, ‘initial triggering of complement’) were increased in TREM2var and individual variants with greater β-amyloid (R62H, C1QA and C1QB; R47H, C1QB) but not with CV.

Figure 5:
Figure 5: Most significantly differentially expressed pathways with greater tissue β-amyloid in microglia (a) or astrocytes (b) from R47H or R62H TREM2var or CV cortex.

Pathways differentially modulated by TREM2var suggested a pro-inflammatory microglial phenotype (a) and increased astrocytic β-amyloid binding and decreased growth factor signalling pathways in astrocytes (b). Pathways were selected from all identified with the full DEG list after filtering for pathways with FDR>0.05 or those associated with only single genes, expressed primarily in cells outside the brain, redundancy within or across databases or biologically non-specific

The two TREM2var differentially expressed different gene sets with greater pathology. ABCA1, the protein product of which mediates APOE lipidation required for microglial amyloid processing28, was up-regulated with greater β-amyloid in R62H microglia, unchanged in CV and down-regulated in R47H whereas ‘sterol transporter activity’ involved in lipid processing was up-regulated in R62H. R47H microglia up-regulated endocytosis pathways (‘early endosome’, ‘clathrin-coated endocytotic vesical membrane’) but R62H microglia down-regulated ‘phagocytosis’ pathway genes. ‘Cellular senescence’ and ‘regulation of endothelial cell proliferation’ pathways were up-regulated in R62H, but not R47H or CV microglia.

Increased expression of pathways associated with ‘regulation of neuron death’ and ‘neuron apoptotic process’ with TREM2var microglia versus CV included responses to β-amyloid that may contribute to neuronal stress and degeneration. There were relatively few DGE with increasing PHF1 for any genotype, consistent with preclinical literature29 and our previous work in CV using AT830.

TREM2 genotype-dependent increases in astrocyte responses to amyloid pathology

TREM2var astrocytes expressed more DGE (3.2% of total expressed genes) with increases in β-amyloid than CV (0.1%) (Fig. 4b). For example, ‘brain-derived neurotrophic factor (BDNF) signalling pathway’ was up-regulated in R62H but unaffected in R47H and CV. Other growth factor pathways were affected; we found increased VEGF or PDGF signalling with R47H or R62H, respectively, but reduced FGF signalling in R62H. ‘Amyloid beta binding’ was amongst the most highly up-regulated pathways for R47H but not significantly expressed with CV or R62H. Occasionally the two TREM2var were associated with genes showing differences in directions of associations such as ABCA1, which was decreased in R47H astrocytes but increased with R62H. APOE and CLU were uniquely up-regulated in R47H, and are associated with astrocyte-mediated neurotoxicity32. Some commonalities were present between TREM2var such as up-regulation of cytoskeletal pathways (‘regulation of actin cytoskeleton’) and down-regulation of axonal pathways (‘axon guidance’). CV astrocytes up-regulated genes involved in the adaptive immune response (CST3, IGF2R) and ‘neutrophil mediated immunity’ (Fig. 5b).

DGE differences suggested TREM2var astrocyte functional compensation for impaired microglial DAM responses and amyloid phagocytosis. Transcripts encoding for endocytosis (BMP2K) in R62H and phagocytosis (MARCKS) for combined TREM2var or R47H individually, and PLPP1 for combined TREM2var or R62H individually were up-regulated with greater β-amyloid density33, but the ‘clathrin coat of endocytic vesicle’ pathway was reduced for R47H. Genes associated with ubiquitin protein ligase activity (WFS1, TMBIM6) were up-regulated with greater β-amyloid with combined or individual TREM2var. CLU, an AD risk gene that limits Aβ accumulation, was up-regulated with R47H, but not R62H34. FBXO2, which regulates APP levels and processing35, was up-regulated for R62H, but not for CV and down-regulated for R47H. Reactive astrocytes can function as phagocytes via the ABCA1 pathway36. Both expression of ABCA1 and MEGF10 were increased for R62H but unchanged in CV whilst ABCA1 expression was decreased for R47H. Taken together, these results describe distinct astroglial transcriptional phenotypes with greater pathology for the two TREM2var and suggest partial potential functional compensation for TREM2var related impairments of microglial amyloid processing by astrocytes.

TREM2 genotype-dependent neuronal transcriptional differences suggest mechanisms contributing to greater neurodegeneration associated with AD risk variants

Relative reductions in numbers of EN and IN neuronal nuclei (Fig. 3b) support evidence11,17,19 for increased neurodegeneration with TREM2var. To identify responsible mechanisms, we regressed expressed neuronal genes with increasing PHF1+ pTau or 4G8+ β-amyloid immunostaining. DGE with greater PHF1 was dependent on genotype (Fig. 6 a, b): 2.8% of total expressed genes were differentially expressed with greater PHF1 for R47H EN, more than R62H EN (0.02%) or CV (0%) EN. IN showed greater DGE for R47H (3.9%) versus R62H (0.2%) or CV (0%). However, we did not find significant EN or IN DGE with greater β-amyloid in the combined TREM2var, suggesting differential responses between R47H and R62H. R62H had fewer DGE with increasing β-amyloid than CV in both EN and IN. Relative DGE numbers were comparable between CV and R47H in EN and increased for R47H IN (Fig. 6a, b). With the low numbers of DGE, CV neurons had relatively few impacted pathways with increasing β-amyloid and none with PHF1. Those associated with greater β-amyloid defined up-regulated chaperone transcriptional responses and energy metabolism (mitochondrial oxidative phosphorylation and glycolysis), interferon signalling, negative regulation of NOTCH, TNF signalling, EGF signalling, senescence pathway induction and regulated cell death by apoptosis/necroptosis. Down-regulated pathways included those for epigenetic modifier mechanisms (‘chromatin remodelling’, ‘lysine-acetylated histone binding’) and neurite processes (‘dendrite’). These pathways were differentially expressed in R47H EN and IN in response to increased PHF1 (Fig. 7a).

Figure 6:
Figure 6: Neuronal differential gene expression for R62H or R47H TREM2var or CV tissue with either greater 4G8+ β-amyloid or PHF1+ pTau immunostained area.

EN differentially expressed more genes with increasing 4G8+ β-amyloid area for CV than for either TREM2 variant, but R47H tissue differentially expressed substantially more genes than either CV or R62H with greater PHF1 (a). R47H also differentially expressed substantially more genes in IN when regressed against either 4G8+ β-amyloid or PHF1+ area (b). Percentages denote differentially expressed genes as a proportion of total genes expressed. logFC>0.25, padj<0.05.

Figure 7:
Figure 7: Most significantly differentially expressed pathways with greater tissue β-amyloid in EN (a) or IN (b) from R47H or R62H TREM2var or CV tissue.

EN, in response to β-amyloid, had several impacted pathways in the CV and R47H genotypes including an up-regulation of ubiquitination in CV and a down-regulation of neuronal plasticity pathways in R47H (a). In response to PHF1, only pathways in R47H were impacted including those involved in neuronal and synaptic plasticity (b). IN pathways were only affected by the R47H genotype and included those involved in ERBB signalling in response to β-amyloid (c) and growth factor signalling in response to PHF1 (d). Pathways were selected based on the removal of pathways with FDR<0.05, single genes, pathways identified with other cells, over-lapping descriptions and non-specific descriptions.

Differential expression of potentially protective neurotrophic signalling pathways with greater PHF1 varied by TREM2var genotype (Fig. 7b, d). Pathways associated with NF-κB, such as ‘NF-κB is activated and signals survival’, were up-regulated in CV EN and IN in response to increasing β-amyloid but unchanged for R47H and R62H. Increased FGF signalling in CV neurons was associated with greater β-amyloid but down-regulated for R47H EN. ‘Signalling by NGF’ was down-regulated in R47H EN and IN and BDNF signalling was down-regulated in R47H IN. Finally, we found that the ‘MAP3K8 (TPL2)-dependent MAPK1/3 activation’ and related MAPK pathways were up-regulated in CV EN (with increasing β-amyloid) and IN (with increasing PHF1), while MAPK pathway expression was not affected by pathology in TREM2var EN, although the ‘MAPK cascade’ and ‘MAPK1/3 signalling’ were down-regulated in combined TREM2var and R47H IN with greater PHF1.

We found evidence for genotype-dependent transcriptional heterogeneity in DGE related to potentially compensatory neuronal plasticity. For example, pathways involved in neurogenesis (‘positive regulation of axon extension’, ‘axonal growth cone’) were up-regulated with increasing β-amyloid in combined TREM2var EN, while those for neuronal processes (‘axon development’, ‘dendrite’) were down-regulated in R47H EN and IN in response to β-amyloid and PHF1. None of these were differentially expressed in R62H or CV with greater β-amyloid. Genes associated with synaptic plasticity were differentially expressed with greater PHF1 in TREM2var and particularly in R47H EN. Expression of genes associated with neurodevelopment or plasticity (SCN3B, CRMP1, PALM, SEMA4D) and synaptic transmission (BSN, SV2A, NPTXR) were down-regulated in R47H EN with greater PHF1. We found related genes were up-regulated in R47H EN, including those for cell adhesion (CNTN5, CDH20, SDK1), glutamate receptors (GRM3, GRM5-AS), the potassium channel (KCNH5), an AMPAR scaffolding protein (GRIP1) and calcium signalling (GCA, ITPR2). Synaptic plasticity pathways down-regulated in R47H IN with increasing β-amyloid and PHF1 included ‘transmission across chemical synapses’ and ‘GABA receptor activation’ while increasing PHF1 in EN was associated with down-regulation of ‘regulation of synaptic plasticity’ and ‘GABAergic synapse’. These pathway genes were not differentially expressed with increasing PHF1 in R62H or CV EN. Taken together, these data highlight greater transcriptional differences with increased β-amyloid or PHF1 in the TREM2var that are characterised by prominent down-regulation of synaptic assembly and signalling pathways, particularly in IN. Those pathways down-regulated in R47H suggest both reduced neurotrophin signalling and plasticity as molecular correlates for the greater AD risk conferred by this allelic variant.

Ubiquitin-mediated protein degradation contributes to protection of cells from ageing and degeneration37 and could contribute to greater neuronal resilience to increasing pathology for CV. CV EN increased gene expression (UBB, UBC, RPS27A, ANKUB1, CKB and TUBB) and ubiquitination pathways with greater β-amyloid, including those for ‘ubiquitin ligase binding’ as well as ‘proteasome binding’. While some ubiquitination genes were up-regulated in R47H IN (UBB, UBC, CKB), none were found for R62H IN or with greater PHF1 for any of the genotypes.

We found genotype-dependent evidence for activation of neuronal stress and apoptosis pathways primarily with greater β-amyloid. EN showed an up-regulation in stress pathways such as ‘cellular response to stress’ for CV with greater 4G8+ β-amyloid but was not found for either of the TREM2var groups. CV EN up-regulated the ‘regulation of apoptotic process’ pathway. The lack of similar differences in gene expression with TREM2var suggests that this could be related to the greater CV microglial responses to β-amyloid. However, consistent with their relative reduction in numbers of neuronal nuclei in AD (Fig. 2b), the combined TREM2var down-regulated ‘negative regulation of apoptotic process’ pathways in IN and R47H EN down-regulated genes involved in cell survival such as TYRO338, RAC1 and LRP139 with greater tissue PHF1+. The X-linked inhibitor of apoptosis, XIAP, was down-regulated in R47H both EN and IN in response to greater 4G8+ β-amyloid. ‘Apoptosis cell clearance’ was down-regulated in R47H EN with increasing PHF1+ as was the ‘DAP12 signalling’ pathway which activates apoptosis in neurons. We did not find either pathway was differentially expressed for CV or R62H.

Altered microglial APP and complement and neuronal growth factor signalling and impaired cell adhesion interactions with TREM2 AD risk variants

We transcriptomically characterised differences in cortical cell-cell signalling for TREM2 R47H and R62H relative to CV using CellChat40 (Fig. 8a). Differences in numbers of cell-cell signalling interactions identified suggested TREM2 genotype-dependent interactions between microglia, astrocytes and neuronal sub-types in combined AD/NDC samples. Numbers of interactions between microglia and astrocytes identified were lower for R47H than for CV, but similar for R62H and CV. Consistent with this, bi-directional signalling between microglia and astrocytes was lower for R47H than for R62H. Greater numbers of signalling pathways between astrocytes and EN were identified for R62H than for R47H. Numbers of interaction pathways between EN and IN were generally reduced relative to CV for both TREM2var.

Figure 8.
Figure 8. Contrasting glial activation and neuronal cell adhesion and neurotrophic cellular interactions by TREM2 genotype.

Signalling pathways identified by CellChat showed greater differences with R47H than with R62H when compared to CV. Differences in numbers of signalling pathways identified are shown as a percentage of the total number of interactions identified the reference genotype in (a) (red = increased interactions, blue = fewer interactions, thickness of line shows magnitude of change). Scatterplots describing differences in relative probabilities of signalling via named pathways are shown for each TREM2 genotype contrast and cell type (b, microglia; c, astrocytes; d, excitatory neurons; e, inhibitory neurons). Scatterplot axes quantify inferred outgoing (ligand) or incoming (receptor) signalling probabilities as scaled differences for association with the selected cell type relative to all cells together on the abscissa or ordinate, respectively. Target signalling pathways represented in black are found present all genotypes, whereas a coloured symbol denotes a target significantly identified only in one of the TREM2var genotypes. Symbol shapes describe the specificity of the significant interaction (circle, identified with both TREM2var genotypes; square, incoming specific to one genotype; triangle, outgoing specific to one genotype; diamond, incoming and outgoing signalling specific to one genotype).

Specific differences in ligand-receptor pairs showing significant genotype-dependent interaction differences were characterised based on the average interaction probabilities (Fig. 8b). These highlighted reduced microglial interactions of APP, a receptor associated with microglial activation by β-amyloid41, for both TREM2var. Reduced signalling interactions for complement, GAS and TGF? were found for both TREM2var relative to CV. Interactions involving PTPRM, an AD risk gene involved in synapse formation42, were lower for both TREM2var. Contrasting TREM2 genotypes, microglial inflammatory response interactions involving SPP1 and C3 complement, and the growth arrest protein GAS, were lower in R47H than for R62H. Greater microglial CD45-mediated signalling interactions were associated with R47H. Strengths for most of the astrocyte interactions identified were reduced for both TREM2var relative to CV, particularly for cell adhesion (involving, CNTN, CADM and NRG) and PTPRM, although R47H NCAM ligand signalling was increased relative to both CV and R62H and NRG ligand interaction strengths were greater for R47H than for R62H. Prominent reductions in signalling through EN laminin and neuronal adhesion factor signalling pathways (involving EPHA, CNTN, NGL and NRXN) for TREM2var suggest associated dysfunction in mechanisms for synapse organisation and stabilisation43. A similar pattern was found in IN for R62H, while R47H IN were associated with increased receptor interactions for NCAM and greater IGF ligand signalling strength. Together, these results thus describe further TREM2 genotype-dependent differences in glial activation related signalling and highlight dysfunctional cell adhesion and growth factor interactions between neurons and astrocytes and neuronal mechanisms that could contribute to greater neuronal vulnerability to neurodegeneration with TREM2var in AD, particularly for the selectively more vulnerable EN44.

DISCUSSION

We explored differences in cellular neuropathology associated with TREM2 R47H and R62H alleles to discover mechanisms by which these variants potentiate neurodegeneration in AD and to understand why the disease risk of the two variants differ. We studied a unique post-mortem tissue collection from 58 donors homozygous for the CV or heterozygous for one of the TREM2var. β-amyloid was increased in AD with both TREM2var, but, while R62H showed proportionally increased dense-core plaques, these were not found for R47H, suggesting greater loss-of-function. DGE with increasing β-amyloid showed a progressively decreasing relative numbers of microglial DGE by genotype (CV>R62H>R47H). Differential DAM signature expression with greater β-amyloid was reduced for TREM2var (particularly in R47H), but the microglia nonetheless displayed pro-inflammatory phenotypes marked by increased interferon and complement pathway expression. This was accompanied by genotype-dependent relative increases in astrocyte DAA gene expression, as well as increases in astrocyte phagocytosis, ‘amyloid beta binding’ and BDNF pathway expression, suggesting partial astroglial functional compensation for impaired in microglial amyloid clearance33. Both TREM2var genotypes had a loss of AT8 immunostaining (Ser202/Thr205/Ser208 epitope binding) and relative reductions in numbers of EN and IN neurons. The latter were associated with decreased synaptic plasticity-related gene expression in EN and IN and increased expression of synaptic, neuronal stress and apoptosis pathways with both greater β-amyloid- and PHF1 pTau, particularly with R47H. Considered together, these results describing increased β-amyloid deposition, pro-inflammatory microglial and astrocyte activation and impaired neuronal plasticity responses could account mechanistically for increased TREM2var-associated AD risk. The lower R62H risk may arise from less impairment of TREM2-dependent microglial functions, reduced astrocyte inflammatory activation and greater neuronal resilience with increased expression of neurotrophic pathways.

We did not find differences in neuronal nuclei relative numbers in TREM2var NDC tissue relative to CV implying that TREM2var neurotoxicity arises with increased amyloid/pTau, consistent with the view that TREM2var are genetic susceptibility factors rather than independently causal factors for disease. Although our neuropathological and transcriptomic observations of amyloid and microglial activation differences with TREM2var are similar to those reported in previous studies9,15–17, our comparison of R47H and R62H is novel. We have related microglial activation differences with AD pathology to those in astrocytes for the first time. Our description of differences in tau phosphorylation45 relative to CV and genotype-dependent differences in expression of neuronal plasticity, stress response and apoptosis pathways with TREM2var provide insights into mechanisms of neuronal vulnerability in AD.

Multiple observations here highlight the less severe AD phenotypes associated with R62H relative to R47H. Dense-core plaques, which are associated with microglial activation by β-amyloid46, were found to increase with increased total β-amyloid only for R62H. Consistent with this, a reduced number of DAM transcripts were found for R62H, but very few DAM genes were differentially expressed in R47H microglia. The increased plaque APOE expression is consistent with a model in which microglia and astrocytes up-regulate plaque APOE in a TREM2-independent manner, but progression to the TREM2-dependent DAM stage26 is impaired. Identification of increased expression of ‘sterol transporter activity’ and vascular endothelial proliferation pathways with greater β-amyloid selectively argues for caution in attributing differences between the genotype-associated pathologies solely to β-amyloid processing47 and inflammation or even to primary effects in microglia48. Neuritic dense-core plaques were increased with R62H, but not in R47H. We interpret the AT8 binding in neurites associated with dense plaques with R62H as further evidence of a partial secondary TREM2 response as pTau AT8 phosphorylation is believed to first accumulate in distal dendrites49.

There has been little exploration of roles for astrocyte in TREM2var-associated neurodegeneration. Astrocytes interact functionally with microglia and neurons and may contribute to β-amyloid clearance50. TREM2 KO mice showed few astrocytic transcriptomic differences in response to amyloid and pTau13, although reactive astrocytes in the TREM2 KO51 may engulf synapses52, further exacerbating synapse loss. Here we described marked TREM2-genotype dependent astrocytic transcriptional differences with increasing pathology. DAA gene set enrichment highlighted an inflammatory and neurotoxic transcriptional profile31. However, we found transcriptomic evidence for increased amyloid phagocytosis that could partially compensate for impaired microglial amyloid clearance53. Enhancement of astrocytic phagocytosis could provide a therapeutic alternative to modulation of TREM2 activity for increasing β-amyloid clearance54.

TREM2 appears to be neuroprotective in a number of neurodegeneration models11,17,19,55,56 and increased AD susceptibility is conferred clinically by allelic variants leading to a partial loss-of-function6. Increasing CV TREM2 activity could be therapeutic57, although inferences based on the loss-of-function here are complicated by multiple TREM2 roles47 and a novel inflammatory microglial phenotype with TREM2var expression. DGE with increasing pathology loads for TREM2var EN and IN suggested similar sub-type vulnerabilities involving plasticity, growth factor production and stress response mechanisms. For example, with the less severe R62H variant, neurons up-regulated NGF and FGF signalling as β-amyloid increased but growth factor pathways were down-regulated in R47H neurons. R62H neurons increased expression of neuronal plasticity and synaptic genes with greater β-amyloid, but similar genes supporting synapse formation and neurotransmission were down-regulated in R47H. Neuronal transcriptomes in R47H were differentially enriched for apoptotic pathways, reduced neurotrophic signalling and reduced ubiquitination with greater tissue β-amyloid.

A major strength of our study is the glial and neuronal transcriptional pathology characterisation in healthy and diseased human cortical tissue from donors with TREM2 risk genotypes or CV. However, we recognise that it also has several limitations. Availability of tissue from the rare TREM2var donors limited the numbers of brains and regions available. Tissue with earlier pathology (Braak stages III-IV) and NDC (Braak 0-II)58 tissue were especially limited. We sought to mitigate bias towards late-stage AD by sampling regions with lower (SOM) and moderate (MTG) pathology and through regression against β-amyloid and PHF1 in the same region (contralateral hemisphere) as used for transcriptomics, allowing the DGE to be in the context of local pathology. As we were limited to relatively small amounts of frozen tissue, we employed snRNAseq which is sparse and biased compared to whole cell transcriptomics and with a lower representation of glial inflammatory responses59,60. Contralateral homologous fixed sections were used for histopathology, rather than adjacent sections from the transcriptomic tissue. However, while this may reduce the analytical power by increasing variance, we do not believe it introduces major bias because of the bilateral symmetry of AD pathology61. CellChat inferred TREM2var effects on cell-cell signalling, but will identify only a proportion of interactions, while other interactions may be over-represented62. Complementary analyses with additional interaction mapping tools, analyses of larger datasets or multiple datasets together could address these limitations.

Our study emphasises the importance of the microglial contribution to β-amyloid clearance in AD and its sensitivity to even a partial loss of TREM2 function with the R62H variant. Results provoke the hypothesis that TREM2var increase the AD risk through a cascade of cellular pathology initiated by reduced microglial responses to β-amyloid and enhanced astrocyte activation leading to greater neuronal stress and activation of neuronal apoptotic pathways. We defined specific pathways engaged for the two TREM2var alleles that may explain differential risks that they confer. Together, they support therapeutic strategies for TREM2 activation in AD while highlighting the need to consider consequences related to the full range of TREM2 functions and cell types expressing TREM2.

Data Availability

The full snRNAseq data will be made available for download from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) upon publication.

AUTHOR CONTRIBUTIONS

NF led the snRNAseq analysis and shinyapp development. NW led the IHC and snRNAseq data generation. ST conducted the pathway and CellChat analysis. CK conducted the DGE analysis. AMG, KD and AS conducted snRNAseq. RM and VC conducted the IMC and IHC. ES conducted IMC analysis. WS and JH conducted genotyping. DH provided analytical support. JSJ and PMM conducted study design, staff supervision, analysis and interpretation of data and funding. JSJ wrote the first draft of the manuscript. PMM edited the manuscript. All authors contributed, reviewed and approved the manuscript.

DECLARATION OF INTERESTS

This study also was supported by an investigator-initiated grant from Biogen IDEC to PMM and JSJ. PMM has received consultancy fees from Roche, Celgene, and Neurodiem. He has received honoraria or speakers’ fees from Novartis and Biogen and has received research or educational funds from Biogen and Novartis.

METHODS

Brain tissue

This study was carried out in accordance with the Regional Ethics Committee and Imperial College Use of Human Tissue guidelines. Cases were selected based first on neuropathological diagnosis (NDC or AD) from UK brain banks (London Neurodegeneration [King’s College London], Newcastle Brain Tissue Resource, Queen’s Square Brain Bank [UCL], Manchester Brain Bank, Oxford Brain Bank and Parkinson’s UK [Imperial College London] Brain Bank). We then excluded cases with clinical or pathological evidence for small vessel disease, stroke, cerebral amyloid angiopathy, diabetes, Lewy body pathology (TDP-43), or other neurological diseases. Where the information was available, cases were selected with a post-mortem delay of less than 49 h (Table 1. MTG and SOM cortex from a final set of 58 cases including 16 non-diseased control (NDC) cases (Braak stage 0–II) and 42 AD cases (Braak stage III–VI) were used (total of 107 brain samples) (Fig. 1a, Table 1). Cortical samples from two regions were prepared from each brain to characterise pathology and transcript expression with both higher (mid temporal gyrus) and lower (sensorimotor) tissue densities of pTau (AT8-, PHF1-antibody positive) in neurofibrillary tangles and β-amyloid plaques (4G8-antibody positive)(Table 2). A subset of FFPE sections were immunostained with a Tau5 antibody (n=13 CV, n=20 TREM2var, all AD).

View this table:
Table 2: Antibodies used for Immunohistochemistry.

CB=citrate buffer. EDTA= ethylenediaminetetraacetic acid.

Immunohistochemistry (IHC)

IHC was performed on FFPE sections (n=107) from the MTG and SOM of each brain studied and paired with material from the cryopreserved contralateral hemisphere used for nuclear preparations for snRNA sequencing. Standard immunostaining procedures as recommended by the manufacturer were followed using the ImmPRESS Polymer (Vector Laboratories) and Super Sensitive Polymer-HRP (Biogenex) kits. Briefly, after dewaxing and rehydration of slides, endogenous peroxidase activity was blocked with 0.3% H2O2, followed by antigen retrieval (Table 2). For immunostaining using ImmPRESS kits, non-specific binding was blocked using 10% normal horse serum. Primary antibodies were incubated overnight at 4 °C. Species-specific ImmPRESS or Super Sensitive kits and DAB were used for antibody visualisation. Tissue was counter-stained by incubation in Mayer’s haematoxylin (TCS Biosciences) for 2 min, followed by dehydration, clearing and mounting.

Image Analysis

Digital images were generated from IHC stained slides scanned using a Leica Aperio AT2 Brightfield Scanner (Leica Biosystems). Images were analysed using Halo software (Indica Labs) after optimisation of Indica Labs macros for each antibody (β-Amyloid, PHF-1 and TAU5 = Area Quantification v1, P-tau = Multiplex IHC v1.2). Data from both cortical regions in each brain were combined. Data was analysed using R package with Wilcoxon signed-rank test.

Imaging Mass Cytometry (IMC)

FFPE 5-10μm sections from MTG were immunostained using lanthanide tagged antibodies (Table 3) before ablation. The slides underwent routine dewaxing and rehydration before undergoing antigen retrieval, in a pH8 Ethylenediaminetetraacetic acid (EDTA) buffer. The slides were then treated with a 10% normal horse serum (Vector Laboratories) blocking solution before incubation with an antibody cocktail (Table 3) at 4°C overnight. The slides were washed in 0.02% Triton X-100 (Sigma-Aldrich) before incubation with the Iridium-intercalator (Fluidigm) then washed and air-dried. All antibody conjugation was performed using the Maxpar X8 protocol (Fluidigm).

View this table:
Table 3: Antibodies used for imaging mass cytometry.

IMC was performed using a Hyperion Tissue Imager (Fluidigm) coupled to a Helios mass cytometer. The instrument was first tuned using the manufacturer’s 3-Element Full Coverage Tuning Slide before the slides were loaded into the device. Four 500×500μm2 regions of interest within the grey matter were selected and ablated using a laser operating at a frequency of 200Hz with 1μm resolution. The data was stored as .mcd files compatible with MCD Viewer software (Fluidigm) and exported as TIFF files. Post-acquisition image processing using ImageJ software allowed threshold correction and the despeckle function to reduce background noise. The image was then masked using CellProfiler and the data was opened with HistoCAT63and exported as .csv files.

Colocalisation of IMC markers

IMC images generated using the 4G8, AT8 and APOE were used to determine the colocalisation of AT8 and APOE with ?-amyloid plaques in MTG region from AD cases only. These included 4 technical replicates for 31 cases. Out of these, 16 were had the CV, 6 had R47H, and 9 had R62H. We used the Fiji (1.0) image analysis software to manually label the plaques in a manner blinded to genotype. For each composite image, plaques were classified as the following: diffuse 4G8+ β-amyloid plaques; dense 4G8+ β-amyloid plaques; APOE+ dense plaques; AT8+ diffuse plaque; AT8+ dense plaque; AT8+ and APOE+ diffuse plaques; and AT8+ and APOE+ dense plaques. Manual counts results were loaded into R (v. 4.0.2). The aggregated dataset was then reconciliated with the genotype information to carry out the statistical analysis. A Wilcoxon signed-rank test was used for testing the significance of differences between the CV and TREM2var groups.

Single nucleus isolation

Single nucleus isolation was conducted on 90 tissue blocks from CV (48 total: n=18, NDC [9 MTG, 9 SOM]; n=21, AD [19 MTG, 11 SOM]) or either the R47H (17 total: n=2 [2 MTG], NDC; n=15, AD [8 MTG, 7 SOM]) or R62H (25 total: n=7 [5 MTG, 2 SOM], NDC; n=18, AD [10 MTG, 8 SOM]) TREM2var. Homologous fresh frozen brain tissue blocks from the MTG and SSC were cryosectioned at 80um and 200mg of grey matter was collected in RNAse-free Eppendorf tube, as previously described30. Nuclei were isolated as previously described30 using a protocol based on 64. All steps were carried out on ice or at 4°C. Tissue was homogenised in homogenisation buffer (1% Triton X-100, 0.4 U/μl RNAseIn + 0.2 U/μl SUPERaseIn, 1ul 1mg/ml DAPI) using a 2ml glass douncer. Homogenate was centrifuged at 4°C for 8mins, 500g and supernatant removed. Homogenate was resuspended in homogenisation buffer and filtered through a 70um filter followed by density gradient centrifugation at 13,000g for 40mins. Supernatant was removed and nuclei were washed and filtered in PBS buffer (PBS + 0.5mg/ml BSA + 0.4 U/μl RNAseIn + 0.2 U/μl SUPERaseIn). Nuclei then underwent primary antibody staining (NeuN 1:500, Merck, MAB377, SOX10 1:250, Bio-Techne, AF2864) for 1hr on ice, washed in PBS buffer, centrifuged at 500g and resuspended in secondary antibody solution (488 donkey anti goat,1:1000, A32814, 647 donkey anti mouse, 1:1000, A32787) for 30m mins on ice. Nuclei were pelleted, washed twice in PBS buffer and resuspended in 1ml PBS buffer. 100ul of nuclei solution was set aside on ice for single nuclear processing.

Single nucleus processing and sequencing

Isolated nuclei stained with Acridine Orange dye were counted on a LUNA-FL Dual Fluorescence Cell Counter (Logos Biosystems, L20001). Sufficient nuclei for recovery of 7000 cells were used for 10x Genomics Chromium Single Cell 3’ sprocessing and library generation. All steps were conducted according to the 10x Genomics Chromium Single Cell 3’ Reagent Kits v3 User Guide, with 8 cycles of cDNA amplification until fragmentation, where 25ng of amplified cDNA per sample was taken through for fragmentation. The final index PCR was conducted at 14 cycles. cDNA and library prep concentration were measured using Qubit dsDNA HS Assay Kit (ThermoFisher, Q32851) and DNA and library preparations were assessed using the Bioanalyzer High-Sensitivity DNA Kit (Agilent, 5067-4627). Pooled samples at equimolar concentrations were sequenced on an Illumina HiSeq 4000 according to the standard 10X Genomics protocol.

The full snRNAseq data will be made available for download from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) upon publication.

Pre-processing and quality-control of snRNA sequencing data

Alignment and demultiplexing of raw sequencing data was performed using 10X Genomics Cell Ranger v3.1, with a pre-mRNA GRCh38 genome reference including both introns and exons. Downstream primary analyses of gene-cell matrices were performed using our scFlow pipeline65. Ambient RNA profiling was performed using emptyDrops with a lower parameter of <100 counts, an alpha cutoff of ≤0.001, and with 70,000 Monte-Carlo iterations66. Cells were filtered for ≥400 and ≤40000 total counts and ≥200 and ≤20000 total expressive features, where expressivity was defined as a minimum of 2 counts in at least 3 cells. The maximum proportion of counts mapping to mitochondrial genes was set to 5%. Doublets were identified using the DoubletFinder algorithm, with a doublets-per-thousand-cells increment of 8 cells (recommended by 10X Genomics), a pK value of 0.005, and embeddings were generated using the first ten principal components calculated from the top 2000 most highly variable genes (HVGs)67.

Integration, clustering, and visualization of data

The linked inference of genomic experimental relationships (LIGER) package was used to calculate integrative factors across samples68. LIGER parameters used included: k: 50, lambda: 5.0, thresh: 0.0001, max_iters: 100, knn_k: 20, min_cells: 2, quantiles: 50, nstart: 10, resolution: 1, num_genes: 3000, center: false. Two-dimensional embeddings of the LIGER integrated factors were calculated using the uniform-manifold approximation and projection (UMAP) algorithm with the following parameters: pca_dims: 30, n_neighbours: 50, init: spectral, metric: euclidean, n_epochs: 200, learning_rate: 1, min_dist: 0.4, spread: 0.85, set_op_mix_ratio: 1, local connectivity: 1, repulsion_strength: 1, negative_sample_rate: 5, fast_sgd: false (McInnes et al., 2018). The Leiden community detection algorithm was used to detect clusters of cells from the 2D UMAP (LIGER) embeddings; a resolution parameter of 0.0001 and a k value of 50 was used69.

Assigning cell type labels to snRNAseq cells

Automated cell-typing was performed essentially as previously described using the Expression Weighted Celltype Enrichment (EWCE) algorithm in scFlow against a previously generated cell-type data reference from the Allan Human Brain Atlas65,70. The top five marker genes for each automatically annotated cell-type were determined using Monocle 3 and validated against canonical cell-type markers71.

Dirichlet modelling of relative cell-type composition

To identify changes in relative cell-type composition across categorical variables (e.g., TREM2var vs CV), we used a Dirichlet-multinomial regression72 which accounts for dependencies in cell-type proportions within samples while accounting for additional co-variates (e.g., sex + age + PMD + brain_region + apoe_group). Threshold for significance was set at an adjusted p-value <0.05.

Differential gene expression analysis

We used model-based analysis of single-cell transcriptomics (MAST) to identify genes differentially expressed (associated) with histopathological features (using 4G8 amyloid, and pTau–PHF1), using each feature as a dependent variable in a zero-inflated regression analysis using a mixed-model 73. Data from both regions was combined. Additionally, diagnosis (control, AD) was used as a dependent variable to identify DGE between experimental groups. Models were fit separately for each cell-type, with and without stratification by TREM2 genotype (none, TREM2var, R47H, or R62H). The model specification was zlm(∼dependent_variable + (1|sample) + cngeneson + pc_mito + sex + APOE + brain_region, method = “glmer”, ebayes = F). The fixed-effect term cngeneson is the cellular detection rate as previously described, and pc_mito accounts for the relative proportion of counts mapping to mitochondrial genes. Each model was fit with and without the dependent variable and compared using a likelihood ratio test. Genes expressed in at least 10% of cells (minimum of 2 counts per cell) were evaluated for gene expression. The threshold for significant differential gene expression was a log2 fold-change of at least 0.25 and an adjusted p-value < 0.05. Additionally, the percentage of inter-individual variance in expression between the control subjects was calculated for each gene, and genes with unusually high variance (>2 S.D.) were omitted from the results. DAM response was generated by taking significantly expressed genes identified in 26 with logFC>0 and p<0.05. DAA response was generated by taking significantly expressed genes in 31. Both responses were compared to our DGE for each cell type.

Impacted pathway analysis

Impacted pathway analysis (IPA) was performed essentially as previously described using the WebGestaltR and enrichR packages in scFlow74,75. Statistically significant differentially expressed genes were submitted for IPA with the over-representation analysis (ORA) enrichment method against the ‘GO_Biological_Process’, ‘Reactome’, and ‘Wikipathway’ databases. The false-discovery rate (FDR) was calculated using the Benjamini-Hochberg method and filtering was applied at a significance threshold of ≤0. Results were ordered by the enrichment score, defined as the proportion of genes overlapping with a pathway relative to the expected overlap. Pathways were selected based on the removal of pathways with FDR<0.05, single genes, pathways identified with other cells, over-lapping descriptions and non-specific descriptions.

Gene set enrichment analyses (GSEA) to detect TREM2-dependent DAM response

TREM2-dependent DAM response gene set was generated by taking the top 1% highly expressed genes unique to stage2 TREM2 dependent cluster identified in 26 and converting them to human orthologs. We then used AUCell (R package v1.6.1) to quantify the expression of the gene set signature in each nucleus. Normalised data were processed in AUCell using the AUCell_build. Rankings function to generate ranking of each gene. The resulting rankings, along with the gene lists of interest, were then run by the function AUCell_calcAUC (aucMaxRank set to 5% of the number of input genes) to generate AUC scores of the gene set which was then scaled across nuclei. We used the dream function from variancePartition package in R76 to compare the expression changes across the categorical variables. We used a linear mixed-model where individual samples were set as the random effect to avoid the pseudoreplication bias and the following covariates (total_features_by_counts, pc_mito, sex, brain_region and age) were set as fixed effects. LogFC and p-value was calculated between TREM2 variants (R47H and R62H) and CV while using diagnosis as an interaction term.

Cell-cell communication analysis

Cell-cell communication analysis was performed using CellChat v1.0.040. CellChat employs a curated database of potential signalling ligand-receptor pairs from the literature. Among all these potential ligand-receptor pairs, cell-cell interactions are identified based on mass action models, along with differential expression analysis and statistical tests on cell groups. The CellChat algorithm was applied with default parameters (unless otherwise specified) was applied to the subset of the dataset that corresponded to the astrocytes, microglia, EN and IN, separately on the samples corresponding to each genotype. The human CellChat database was used for the ligand-receptor pairs. The normalized count matrix was employed for the analysis without any projection onto a protein interaction (PPI) network. The output of CellChat comprises a set of ligand-receptor pairs that are considered ‘active’ and the corresponding probability value. The cell-cell communication results may also be represented at the signalling pathway level, i.e. highlighting signalling pathways that include genes identified as ligands and/receptors from the CellChat analysis.

ACKNOWLEDGEMENTS

We thank the donors and their families for the use of human brain tissue in this study and the UK brain bank staff for making it available. Tissue samples were provided by the London Neurodegenerative Diseases Brain Bank at King’s College London. The brain bank receives funding from the UK Medical Research Council and as part of the Brains for Dementia Research programme, jointly funded by Alzheimer’s Research UK and the Alzheimer’s Society. Tissue for this study was provided by the Newcastle Brain Tissue Resource which is founded in party by a grant from the UK Medical Research Council (G0400074), by NIHR Newcastle Biomedical Research Centre and Unit awarded to the Newcastle upon Tyne NHS Foundation Trust and Newcastle University, and as part of the Brains for Dementia Research Programme jointly funded by Alzheimer’s Research UK and Alzheimer’s Society. In addition, tissue was provided by the Queen’s Square Brain Bank, UCL. Tissue samples were supplied by The Manchester Brain Bank, which is part of the Brains for Dementia Research programme, jointly funded by Alzheimer’s Research UK and Alzheimer’s Society. We acknowledge the Oxford Brain Bank, supported by the Medical Research Council (MRC), the NIHR Oxford Biomedical Research Centre and the Brains for Dementia Research programme, jointly funded by Alzheimer’s Research UK and Alzheimer’s Society. Tissue samples and associated clinical and neuropathological data were supplied by Parkinson’s UK Brain Bank at Imperial, funded by Parkinson’s UK, a charity registered in England and Wales (258197) and in Scotland (SC037554). We are grateful to Diana P. Benitez for her support in the human tissue management. Figure 1 was created with BioRender.com.

Infrastructure, including particularly the LMS/NIHR Imperial Biomedical Research Centre Flow Cytometry Facility and the Imperial BRC Genomics Facility, was supported by the National Institute for Health Research (NIHR) Biomedical Research Centre (BRC). ST was supported by an “Early Postdoc Mobility” scholarship (P2GEP3_191446) from the Swiss National Science Foundation and a “Clinical Medicine Plus” scholarship from the Prof Dr Max Cloëtta Foundation (Zurich, Switzerland). JH acknowledges support from the Dolby Foundation and the UK DRI. PMM acknowledges generous personal support from the Edmond J Safra Foundation and Lily Safra and an NIHR Senior Investigator Award. This work was supported by a grant to PMM from the UK Dementia Research Institute, which receives its funding from UK DRI Ltd., funded by the UK Medical Research Council, Alzheimer’s Society, and Alzheimer’s Research UK. The study is an output from the UK Dementia Institute Multi-omics Atlas Project for Alzheimer’s Disease (MAP-AD; map-ad.org).

Footnotes

  • ↵# joint senior authors

REFERENCES

  1. 1.↵
    Jonsson, T. et al. Variant of TREM2 Associated with the Risk of Alzheimer’s Disease. N. Engl. J. Med. 368, 107–116 (2013).
    OpenUrlCrossRefPubMedWeb of Science
  2. 2.↵
    Yeh, F. L., Hansen, D. V. & Sheng, M. TREM2, Microglia, and Neurodegenerative Diseases. Trends Mol. Med. 23, 512–533 (2017).
    OpenUrlCrossRefPubMed
  3. 3.↵
    Franzmeier, N. et al. Higher CSF sTREM2 attenuates ApoE4-related risk for cognitive decline and neurodegeneration. Mol. Neurodegener. 15, 1–10 (2020).
    OpenUrlCrossRef
  4. 4.↵
    Guerreiro, R. et al. TREM2 Variants in Alzheimer’s Disease. N. Engl. J. Med. 368, 117–127 (2012).
    OpenUrlPubMedWeb of Science
  5. 5.↵
    Sims, R. et al. Rare coding variants in PLCG2, ABI3, and TREM2 implicate microglial-mediated innate immunity in Alzheimer’s disease. Nat. Genet. 2017 499 49, 1373–1384 (2017).
    OpenUrlCrossRefPubMed
  6. 6.↵
    Gratuze, M., Leyns, C. E. G. & Holtzman, D. M. New insights into the role of TREM2 in Alzheimer ’ s disease. 9, 1–16 (2018).
    OpenUrl
  7. 7.↵
    Li, R., Wang, X. & He, P. The most prevalent rare coding variants of TREM2 conferring risk of Alzheimer’s disease: A systematic review and meta‐analysis. Exp. Ther. Med. 21, 1–1 (2021).
    OpenUrlCrossRef
  8. 8.↵
    Jin, S. C. et al. Coding variants in TREM2 increase risk for Alzheimer’s disease. Hum. Mol. Genet. 23, 5838–5846 (2014).
    OpenUrlCrossRefPubMed
  9. 9.↵
    Leyns, C. E. G. et al. TREM2 function impedes tau seeding in neuritic plaques. Nat. Neurosci. 2019 228 22, 1217–1222 (2019).
    OpenUrlCrossRefPubMed
  10. 10.↵
    Ulland, T. K. et al. TREM2 Maintains Microglial Metabolic Fitness in Alzheimer’s Disease. Cell 170, 649-663.e13 (2017).
    OpenUrlCrossRefPubMed
  11. 11.↵
    Sayed, F. A. et al. AD-linked R47H-TREM2 mutation induces disease-enhancing microglial states via AKT hyperactivation. Sci. Transl. Med. 13, 3947 (2021).
    OpenUrl
  12. 12.↵
    Nugent, A. A. et al. TREM2 Regulates Microglial Cholesterol Metabolism upon Chronic Phagocytic Challenge. Neuron 105, 837-854.e9 (2020).
    OpenUrlCrossRefPubMed
  13. 13.↵
    Lee, S. H. et al. TREM2-independent oligodendrocyte, astrocyte, and T cell responses to tau and amyloid pathology in mouse models of Alzheimer disease. Cell Rep. 37, 110158 (2021).
    OpenUrl
  14. 14.↵
    Bemiller, S. M. et al. TREM2 deficiency exacerbates tau pathology through dysregulated kinase signaling in a mouse model of tauopathy. Mol. Neurodegener. 12, 1–12 (2017).
    OpenUrlCrossRef
  15. 15.↵
    Toomey, C. E. et al. Investigation of pathology, expression and proteomic profiles in human TREM2 variant postmortem brains with and without Alzheimer’s disease. Brain Pathol. 30, 794–810 (2020).
    OpenUrl
  16. 16.↵
    Prokop, S. et al. Impact of TREM2 risk variants on brain region-specific immune activation and plaque microenvironment in Alzheimer’s disease patient brain samples HHS Public Access. Acta Neuropathol 138, 613–630 (2019).
    OpenUrl
  17. 17.↵
    Zhou, Y. et al. Human and mouse single-nucleus transcriptomics reveal TREM2-dependent and-independent cellular responses in Alzheimer’s disease HHS Public Access Author manuscript. (2020). doi:10.1038/s41591-019-0695-9
    OpenUrlCrossRefPubMed
  18. 18.↵
    Joshi, P. et al. TREM2 modulates differential deposition of modified and non-modified Aβ species in extracellular plaques and intraneuronal deposits. Acta Neuropathol. Commun. 9, (2021).
  19. 19.↵
    Nguyen, A. T. et al. APOE and TREM2 regulate amyloid-responsive microglia in Alzheimer’s disease. Acta Neuropathol. 140, 477–493 (2020).
    OpenUrlCrossRef
  20. 20.↵
    Qin, Q. et al. TREM2, microglia, and Alzheimer’s disease. Mech. Ageing Dev. 195, 111438 (2021).
    OpenUrl
  21. 21.↵
    Song, W. et al. Alzheimer’s disease-associated TREM2 variants exhibit either decreased or increased ligand-dependent activation. Alzheimer’s Dement. 13, 381–387 (2017).
    OpenUrlCrossRefPubMed
  22. 22.↵
    Braak, H., Alafuzoff, I., Arzberger, T., Kretzschmar, H. & Tredici, K. Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol. 112, 389 (2006).
    OpenUrlCrossRefPubMedWeb of Science
  23. 23.↵
    Malia, T. J. et al. Epitope mapping and structural basis for the recognition of phosphorylated tau by the anti-tau antibody AT8. Proteins 84, 427–434 (2016).
    OpenUrlCrossRefPubMed
  24. 24.↵
    Otvos, L. et al. Monoclonal antibody PHF-1 recognizes tau protein phosphorylated at serine residues 396 and 404. J. Neurosci. Res. 39, 669–673 (1994).
    OpenUrlCrossRefPubMedWeb of Science
  25. 25.↵
    Lopes, S. et al. Tau protein is essential for stress-induced brain pathology. Proc. Natl. Acad. Sci. U. S. A. 113, E3755–E3763 (2016).
    OpenUrlAbstract/FREE Full Text
  26. 26.↵
    Keren-Shaul, H. et al. A Unique Microglia Type Associated with Restricting Development of Alzheimer’s Disease. Cell 169, 1276-1290.e17 (2017).
    OpenUrlCrossRefPubMed
  27. 27.↵
    Tooyama, I., Kimura, H., Akiyama, H. & McGeer, P. L. Reactive microglia express class I and class II major histocompatibility complex antigens in Alzheimer’s disease. Brain Res. 523, 273– 280 (1990).
    OpenUrlCrossRefPubMedWeb of Science
  28. 28.↵
    Wahrle, S. E. et al. ABCA1 Is Required for Normal Central Nervous System ApoE Levels and for Lipidation of Astrocyte-secreted apoE *. J. Biol. Chem. 279, 40987–40993 (2004).
    OpenUrlAbstract/FREE Full Text
  29. 29.↵
    Sierksma, A. et al. Novel Alzheimer risk genes determine the microglia response to amyloid-β but not to TAU pathology. EMBO Mol. Med. 12, (2020).
  30. 30.↵
    Smith, A. M. et al. Diverse human astrocyte and microglial transcriptional responses to Alzheimer’s pathology. Acta Neuropathol. 143, 75–91 (2022).
    OpenUrl
  31. 31.↵
    Habib, N. et al. Disease-associated astrocytes in Alzheimer’s disease and aging. Nat. Neurosci. 23, 701–706 (2020).
    OpenUrl
  32. 32.↵
    Guttenplan, K. A. et al. Neurotoxic reactive astrocytes induce cell death via saturated lipids. Nat. 2021 5997883 599, 102–107 (2021).
    OpenUrl
  33. 33.↵
    Konishi, H. et al. Astrocytic phagocytosis is a compensatory mechanism for microglial dysfunction. EMBO J. 39, (2020).
  34. 34.↵
    Wojtas, A. M. et al. Astrocyte-derived clusterin suppresses amyloid formation in vivo. Mol. Neurodegener. 15, 1–14 (2020).
    OpenUrlCrossRef
  35. 35.↵
    Atkin, G. et al. F-box Only Protein 2 (Fbxo2) Regulates Amyloid Precursor Protein Levels and Processing *. J. Biol. Chem. 289, 7038–7048 (2014).
    OpenUrlAbstract/FREE Full Text
  36. 36.↵
    Morizawa, Y. M. et al. Reactive astrocytes function as phagocytes after brain ischemia via ABCA1-mediated pathway. Nat. Commun. 2017 81 8, 1–15 (2017).
    OpenUrlCrossRefPubMed
  37. 37.↵
    Schmidt, M. F., Gan, Z. Y., Komander, D. & Dewson, G. Ubiquitin signalling in neurodegeneration: mechanisms and therapeutic opportunities. Cell Death Differ. 2021 282 28, 570–590 (2021).
    OpenUrlCrossRef
  38. 38.↵
    Zhong, Z. et al. Protein S Protects Neurons from Excitotoxic Injury by Activating the TAM Receptor Tyro3–Phosphatidylinositol 3-Kinase–Akt Pathway through Its Sex Hormone-Binding Globulin-Like Region. J. Neurosci. 30, 15521 (2010).
    OpenUrlAbstract/FREE Full Text
  39. 39.↵
    Fuentealba, R. A., Liu, Q., Kanekiyo, T., Zhang, J. & Bu, G. Low density lipoprotein receptor-related protein 1 promotes anti-apoptotic signaling in neurons by activating Akt survival pathway. J. Biol. Chem. 284, 34045–34053 (2009).
    OpenUrlAbstract/FREE Full Text
  40. 40.↵
    Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12, (2021).
  41. 41.↵
    Manocha, G. D. et al. APP Regulates Microglial Phenotype in a Mouse Model of Alzheimer’s Disease. J. Neurosci. 36, 8471–8486 (2016).
    OpenUrlAbstract/FREE Full Text
  42. 42.↵
    Mo, X. et al. PTPRM Is Critical for Synapse Formation Regulated by Zinc Ion. Front. Mol. Neurosci. 15, (2022).
  43. 43.↵
    Jiang, X., Sando, R. & Südhof, T. C. Multiple signaling pathways are essential for synapse formation induced by synaptic adhesion molecules. Proc. Natl. Acad. Sci. U. S. A. 118, (2021).
  44. 44.↵
    Fu, H. et al. A tau homeostasis signature is linked with the cellular and regional vulnerability of excitatory neurons to tau pathology. Nat. Neurosci. 22, 47–56 (2019).
    OpenUrlCrossRefPubMed
  45. 45.↵
    Malia, T. J. et al. Epitope mapping and structural basis for the recognition of phosphorylated tau by the anti-tau antibody AT8. Proteins Struct. Funct. Bioinforma. 84, 427–434 (2016).
    OpenUrl
  46. 46.↵
    Wang, Y. et al. TREM2-mediated early microglial response limits diffusion and toxicity of amyloid plaques. J Exp Med 213, 667–675 (2016).
    OpenUrlAbstract/FREE Full Text
  47. 47.↵
    Deczkowska, A., Weiner, A. & Amit, I. The Physiology, Pathology, and Potential Therapeutic Applications of the TREM2 Signaling Pathway. Cell 181, 1207–1217 (2020).
    OpenUrlCrossRefPubMed
  48. 48.↵
    Yang, A. C. et al. A human brain vascular atlas reveals diverse mediators of Alzheimer’s risk. Nat. 2022 6037903 603, 885–892 (2022).
    OpenUrl
  49. 49.↵
    Hall, G. F., Lee, V. M. Y., Lee, G. & Yao, J. Staging of Neurofibrillary Degeneration Caused by Human Tau Overexpression in a Unique Cellular Model of Human Tauopathy. Am. J. Pathol. 158, 235 (2001).
    OpenUrlCrossRefPubMedWeb of Science
  50. 50.↵
    Raha, S., Ghosh, A., Dutta, D., Patel, D. R. & Pahan, K. Activation of PPARα enhances astroglial uptake and degradation of β;-amyloid. Sci. Signal. 14, 4747 (2021).
    OpenUrl
  51. 51.↵
    Taylor, X. et al. A1 reactive astrocytes and a loss of TREM2 are associated with an early stage of pathology in a mouse model of cerebral amyloid angiopathy. J. Neuroinflammation 17, 1– 18 (2020).
    OpenUrlCrossRef
  52. 52.↵
    Jay, T. R. et al. TREM2 is required for microglial instruction of astrocytic synaptic engulfment in neurodevelopment. Glia 67, 1873–1892 (2019).
    OpenUrlCrossRefPubMed
  53. 53.↵
    Jones, R. S., Minogue, A. M., Connor, T. J. & Lynch, M. A. Amyloid-β-induced astrocytic phagocytosis is mediated by CD36, CD47 and RAGE. J. Neuroimmune Pharmacol. 8, 301–311 (2013).
    OpenUrlCrossRefPubMed
  54. 54.↵
    Park, J. S. et al. Blocking microglial activation of reactive astrocytes is neuroprotective in models of Alzheimer’s disease. Acta Neuropathol. Commun. 9, 1–15 (2021).
    OpenUrl
  55. 55.↵
    Xie, M. et al. TREM2 interacts with TDP-43 and mediates microglial neuroprotection against TDP-43-related neurodegeneration. Nat. Neurosci. 25, 26–38 (2022).
    OpenUrl
  56. 56.↵
    Ren, M. et al. TREM2 overexpression attenuates neuroinflammation and protects dopaminergic neurons in experimental models of Parkinson’s disease. Exp. Neurol. 302, 205– 213 (2018).
    OpenUrlCrossRefPubMed
  57. 57.↵
    Schlepckow, K. et al. Enhancing protective microglial activities with a dual function TREM2 antibody to the stalk region. EMBO Mol. Med. 12, e11227 (2020).
    OpenUrlCrossRefPubMed
  58. 58.↵
    Braak, H. & Braak, E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 82, 239–259 (1991).
    OpenUrlCrossRefPubMedWeb of Science
  59. 59.↵
    Thrupp, N. et al. Single-Nucleus RNA-Seq Is Not Suitable for Detection of Microglial Activation Genes in Humans. Cell Rep. 32, (2020).
  60. 60.↵
    Bakken, T. E. et al. Single-nucleus and single-cell transcriptomes compared in matched cortical cell types. PLoS One 13, e0209648 (2018).
    OpenUrl
  61. 61.↵
    Moossy, J., Zubenko, G. S., Martinez, A. J. & Rao, G. R. Bilateral Symmetry of Morphologic Lesions in Alzheimer’s Disease. Arch. Neurol. 45, 251–254 (1988).
    OpenUrlCrossRefPubMedWeb of Science
  62. 62.↵
    Dimitrov, D. et al. Comparison of methods and resources for cell-cell communication inference from single-cell RNA-Seq data. Nat. Commun. 2022 131 13, 1–13 (2022).
    OpenUrlCrossRef
  63. 63.↵
    Schapiro, D. et al. histoCAT: analysis of cell phenotypes and interactions in multiplex image cytometry data. Nat. Methods 2017 149 14, 873–876 (2017).
    OpenUrlCrossRefPubMed
  64. 64.↵
    Krishnaswami, S. R. et al. Using single nuclei for RNA-seq to capture the transcriptome of postmortem neurons. Nat. Protoc. (2016). doi:10.1038/nprot.2016.015
    OpenUrlCrossRefPubMed
  65. 65.↵
    Khozoie, C. et al. scFlow: A Scalable and Reproducible Analysis Pipeline for Single-Cell RNA Sequencing Data. bioRxiv 2021.08.16.456499 (2021). doi:10.1101/2021.08.16.456499
    OpenUrlAbstract/FREE Full Text
  66. 66.↵
    Lun, A. T. L. et al. EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data. Genome Biol. 20, (2019).
  67. 67.↵
    McInnes, L., Healy, J., Saul, N. & Großberger, L. UMAP: Uniform Manifold Approximation and Projection. J. Open Source Softw. 3, 861 (2018).
    OpenUrl
  68. 68.↵
    Welch, J. D. et al. Single-Cell Multi-omic Integration Compares and Contrasts Features of Brain Cell Identity. Cell 177, 1873-1887.e17 (2019).
    OpenUrlCrossRefPubMed
  69. 69.↵
    Traag, V. A., Waltman, L. & van Eck, N. J. From Louvain to Leiden: guaranteeing well-connected communities. Sci. Reports 2019 91 9, 1–12 (2019).
    OpenUrl
  70. 70.↵
    Skene, N. G. & Grant, S. G. N. Identification of vulnerable cell types in major brain disorders using single cell transcriptomes and expression weighted cell type enrichment. Front. Neurosci. 10, 16 (2016).
    OpenUrlPubMed
  71. 71.↵
    Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 2014 324 32, 381–386 (2014).
    OpenUrlCrossRefPubMed
  72. 72.↵
    Smillie, C. S. et al. Intra-and Inter-cellular Rewiring of the Human Colon during Ulcerative Colitis. Cell 178, 714-730.e22 (2019).
    OpenUrlCrossRefPubMed
  73. 73.↵
    Finak, G. et al. MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol. 16, (2015).
  74. 74.↵
    Liao, Y., Wang, J., Jaehnig, E. J., Shi, Z. & Zhang, B. WebGestalt 2019: gene set analysis toolkit with revamped UIs and APIs. Nucleic Acids Res. 47, W199–W205 (2019).
    OpenUrlCrossRefPubMed
  75. 75.↵
    Chen, E. Y. et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics 14, 128 (2013).
    OpenUrlCrossRefPubMed
  76. 76.↵
    Hoffman, G. E. & Schadt, E. E. variancePartition: Interpreting drivers of variation in complex gene expression studies. BMC Bioinformatics 17, 1–13 (2016).
    OpenUrlCrossRefPubMed
Back to top
Posted July 13, 2022.
Download PDF
Data/Code
Email
Mechanisms contributing to differential genetic risks for TREM2 R47H and R62H variants in Alzheimer’s Disease
Nurun Fancy, Nanet Willumsen, Stergios Tsartsalis, Combiz Khozoie, Aisling McGarry, Robert C Muirhead, Eleonore Schneegans, Karen Davey, Vicky MN Chau, Amy M Smith, William Scotton, John Hardy, Dann Huh, Paul M Matthews, Johanna S Jackson
medRxiv 2022.07.12.22277509; doi: https://doi.org/10.1101/2022.07.12.22277509
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area