Microglia Depletion Reduces Human Neuronal APOE4-Related Pathologies in a Chimeric Alzheimer’s Disease Model

SUMMARY

Despite strong evidence supporting the important roles of both apolipoprotein E4 (APOE4) and microglia in Alzheimer’s Disease (AD) pathogenesis, the effects of microglia on neuronal APOE4-related AD pathogenesis remain elusive. To examine such effects, we utilized microglial depletion in a chimeric model with induced pluripotent stem cell (iPSC)-derived human neurons in mouse hippocampus. Specifically, we transplanted homozygous APOE4, isogenic APOE3, and APOE-knockout (APOE-KO) iPSC-derived human neurons into the hippocampus of human APOE3 or APOE4 knock-in mice, then depleted microglia in half of the chimeric mice. We found that both neuronal APOE and microglial presence were important for the formation of Aβ and tau pathologies in an APOE isoform-dependent manner (APOE4 > APOE3). Single-cell RNA-sequencing analysis identified two pro-inflammatory microglial subtypes with elevated MHC-II gene expression enriched in chimeric mice with human APOE4 neuron transplants. These findings highlight the concerted roles of neuronal APOE, especially APOE4, and microglia in AD pathogenesis.

Graphical Abstract

eTOC blurb

Huang and colleagues examine the effects of APOE4 and microglial depletion on Alzheimer’s Disease (AD) pathogenesis in a chimeric mouse model with transplanted human induced pluripotent stem cell-derived neurons. They demonstrate the importance of neuronal APOE and the role of both APOE4 and microglia in promoting AD pathologies.

INTRODUCTION

Apolipoprotein E4 (APOE4) is the strongest genetic risk factor for Alzheimer’s Disease (AD). Of the three main APOE isoforms—APOE2, APOE3, and APOE4—APOE4 increases AD risk, reduces age of disease onset, and promotes classical AD pathologies, including deposition of amyloid-beta (Aβ) peptides and hyperphosphorylated tau (p-tau) protein^1-7. Although APOE4 is strongly linked to AD^1,2,8,9, its roles in AD pathogenesis are complex and merit continued exploration. While astrocytes are the main producers of APOE in the central nervous system, neurons and microglia can also produce APOE in response to stress, injury, or aging^10-16. Studies indicate that the effects of APOE4 in AD may depend on the cell type in which it is produced^4,6,17,18.

Several studies have documented the detrimental effects of APOE4 in AD, using either mouse models or human induced pluripotent stem cell (hiPSC)-derived models. Mouse models have traditionally provided a convenient system for studying APOE4’s multifarious AD-related effects in a diverse, mature in vivo environment. In human APOE3 knock-in (E3KI) or APOE4 knock-in (E4KI) mice expressing mutant human amyloid precursor protein (APP) or microtubule associate protein tau (MAPT), several experiments have demonstrated that APOE4 exacerbates the formation of Aβ plaques and tau tangles and promotes neurodegeneration^18-21. In particular, neuronal APOE4 has proved to significantly affect AD pathology^16,22.

However, these mouse model studies have limited therapeutic translatability, often relying on early-onset AD or other tauopathy-related mutations to produce pathologies and lacking key human-specific cellular hallmarks of late-onset AD^6,23. To study human-specific aspects of late-onset AD, many scientists have turned to in vitro hiPSC-derived cell models. Several labs using hiPSC-derived neurons have identified cellular features of APOE4-driven AD pathologies, including increased levels of secreted Aβ species and intracellular phosphorylated tau species^12,24,25. However, these in vitro hiPSC-derived cellular models have restricted ability to model mature cell behavior, produce Aβ or tau aggregates, and recapitulate multicellular interactions found in vivo.

To address these limitations, our lab developed an in vivo chimeric late-onset AD mouse model in which hiPSC-derived neurons with different APOE genotypes are transplanted into APOE knock-in (EKI) mouse hippocampi²⁶. This hybrid model facilitates human neuronal maturation and AD pathology far beyond what is capable in vitro, even producing Aβ aggregates reminiscent of pathology found in late-onset AD²⁶. Furthermore, this model enables investigation of human neuronal interactions with other cell types in vivo throughout APOE4-related AD pathogenesis. We specifically focused on the effect of microglia, which are consistently implicated in AD pathogenesis^27,28. Microglia maintain brain health through surveillance of and activated response to cellular debris as well as phagocytosis of pathological protein aggregates, like Aβ and tau assemblies^29,30. Recent studies have found that microglia in AD may interact with neurons to contribute to AD pathogenesis, either by reducing clearance of Aβ and tau^7,23,31,32 or by active seeding/spreading of Aβ and tau aggregates^28,33-35. The advent of effective microglial depletion tools allowed us to differentiate between these two possibilities in the context of different APOE isoforms. In particular, PLX3397 (PLX) is a potent inhibitor of vital microglial protein colony-stimulating factor 1 receptor (Csf1r), enabling depletion of microglia without significantly affecting other brain cell types^36-38. In our chimeric mice, we used PLX depletion of microglia to investigate microglial influence on neuronal APOE4-related AD pathologies.

RESULTS

Transplanted hiPSC-derived neurons survive in APOE-KI mouse hippocampus

Our lab previously generated and characterized three different hiPSC lines: 1) homozygous human APOE4 (hE4) hiPSCs from an AD patient, 2) isogenic homozygous human APOE3 (hE3) hiPSCs generated from the hE4 hiPSC line by gene editing, and 3) homozygous human APOE-knockout (hEKO) hiPSCs from a patient homozygous for an ablative APOE frameshift mutation (c.291del, p.E97fs)^12,39. To generate a chimeric AD model, we differentiated these three hiPSC lines into mixed neuronal cultures (including both excitatory and inhibitory neurons) using an established neuronal differentiation protocol^12,26, and transplanted them into the hippocampi of 4-month-old E4KI and E3KI mice. As previous studies have indicated that APOE4 may have effects on early brain development^40-42, mice were transplanted at 4 months of age to better isolate the effects of APOE isoforms on biological processes pertinent to aging and late-onset AD. hE4 and hE3 neurons were transplanted into E4KI and E3KI mice of matching APOE genotype (hE4 neurons into E4KI mice and hE3 neurons into E3KI mice), while hEKO neurons were transplanted into E4KI mice to examine the interactive effects of human neurons lacking APOE with APOE4-expressing microglia. To improve cell transplant survival, mouse host immune response was blocked with an immunosuppressant cocktail administered on days 0, 2, 4, and 6 after transplantation^26,43. The chimeric mice were then aged for 8 months and either fed an AIN-76A control chow for the entire duration or fed a control chow for the first 4 months and then an AIN-76A-PLX chow containing 400 mg/kg of PLX for the remaining 4 months (Figure 1A). Conditions were labeled according to the APOE genotype of the transplanted human cells (hE3-, hE4-, hEKO-), the APOE genotype of the host mice (E3KI, E4KI), and the type of chow the chimeric mice received (control chow, PLX chow) (Figure 1A). In total, six groups of mice were used in this study: hE4-E4KI, hE4-E4KI-PLX, hE3-E3KI, hE3-E3KI-PLX, hEKO-E4KI, and hEKO-E4KI-PLX (Figure 1A).

Figure 1. Transplanted human neuronal progenitors survive and develop into neurons in chimeric mouse hippocampus.

(A) Experimental design: iPSCs with different APOE genotypes were differentiated into neuronal progenitors and transplanted into human APOE-KI mice. Chimeric mice were aged for 8 months, with half the mice receiving PLX3397 (PLX) for the latter 4 months. All mice were used for histological or transcriptomic analysis.

(B) Representative images of human cell transplants in the hippocampus of 12-month-old chimeric mice (8 months post transplantation). Top row: Human Nuclear Antigen (HNA, red) marking human cell transplants; second row: human-preferential MAP2 (gray) marking neurons; third row: composite of HNA and MAP2 images. Scale bar, 100 μm.

(C) Quantification of number of HNA+ cells per human transplant. Each dot represents one mouse per condition (hE3-E3KI, n=5; hE3-E3KI-PLX, n=5; hE4-E4KI, n=9; hE4-E4KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6).

(D) Quantification of size of human transplants per condition, as percent of hippocampus. Each dot represents one mouse (hE3-E3KI, n=5; hE3-E3KI-PLX, n=6; hE4-E4KI, n=9; hE4-E4KI-PLX, n=8; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6).

(E–F) Representative immunohistochemical images of transplanted human cells expressing mature neuronal (NEUN and MAP2) and synaptic (synaptophysin) markers. Scale bar, 25 μm.

(G,H) Whole-cell current-clamp recordings of transplanted E4/4 (top) and iE3/3 (bottom) iPSC-derived neurons in ex vivo slices demonstrating the ability to fire action potentials (G, the membrane potential responses to a 1-s, 50-pA depolarizing current injection) and the ability to receive spontaneous excitatory post-synaptic currents (H, the membrane current traces recorded at a holding potential of −70mV).

For quantifications, all data are expressed as mean ± S.E.M. Differences between groups were determined by two-way ANOVA with Benjamini’s post hoc test for multiple comparisons

At 8 months post-transplantation, the chimeric mice were transcardially perfused, with one brain hemisphere fixed for immunohistochemical analyses and the other hemisphere harvested for single cell RNA-sequencing (scRNA-seq) of isolated hippocampal microglia. Transplanted cells stained positive for human nuclear antigen (HNA) (Figure 1B) and for mature neuronal marker MAP2 (Figure 1B and Figure S1D). The vast majority of HNA⁺ cells did not stain positive for astrocyte marker GFAP (Figure S1A) or oligodendrocyte marker OLIG2 (Figure S1B), indicating that mostly neurons, not astrocytes or oligodendrocytes, developed from the transplanted human cells (Figure S1C). Through immunohistochemical analysis, we confirmed that all human cell transplants survived and found that transplants across conditions did not differ significantly in cell number or size of transplants (Figures 1C and 1D). Similar to previously published studies in chimeric mice^26,44-46, the transplanted cells did not migrate far from the original transplant site. Furthermore, the transplanted human neurons stained positive for neuronal nuclear marker NeuN (Figure 1E) and human synaptic marker Synaptophysin (Figure 1F). Whole-cell patch-clamp recordings of the transplanted human neurons in ex vivo slices of chimeric mice also demonstrated their ability to generate action potentials (Figure 1G) and receive synaptic input in the form of spontaneous post-synaptic currents (Figure 1H). Together, these data signify successful survival and maturation as well as functional integration of the transplanted human neurons in E3KI and E4KI mouse hippocampi. The MAP2 antibody used for these experiments shows preferential recognition of human versus mouse neurons, resulting in the transplanted human neurons appearing substantially brighter than the surrounding mouse neurons (Figure 1B). As a result, for all following stains, we were able to use MAP2 staining with this antibody to identify the transplanted human neurons.

PLX treatment effectively depletes microglia, but not astrocytes, in the hippocampus of chimeric mice

We measured the effects of PLX treatment on microglia counts in the hippocampus. Upon staining the chimeric mouse brain sections with microglial marker Iba1, we found significant microglial depletion in all three groups treated with PLX (Figure 2A and 2B). Unexpectedly, microglial depletion was more efficacious in hE4-E4KI-PLX mouse hippocampus (93%) than in hEKO-E4KI-PLX (84%) and hE3-E3KI-PLX (67%) mouse hippocampus. While this attenuated microglial depletion effect in hE3-E3KI-PLX mice may be enhanced by the increased baseline number of microglia in hE3-E3KI mice compared to hE4-E4KI mice (Figure 2B), this difference merited further investigation.

Figure 2. PLX depletes microglia, but does not affect astrocytes, in chimeric mouse hippocampus.

(A) Representative images of microglia (Iba1, red) in the hippocampus of chimeric mice of each condition. Scale bar, 500 μm.

(B) Quantification of number/mm² of Iba1⁺ microglia in the hippocampus. Each dot represents one mouse (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=8; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6).

(C) Representative images of astrocytes (GFAP, red) in the hippocampus of chimeric mice of each condition. Scale bar, 500 μm.

(D) Quantification of number/mm² of GFAP⁺ astrocytes in the hippocampus. Each dot represents one mouse (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=8; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6).

All data are expressed as mean ± S.E.M. Differences between groups were determined by two-way ANOVA with Benjamini’s post hoc test for multiple comparisons. Adjusted p-values (q-values) are displayed.

To this end, we sought to measure baseline levels of Csf1r expression in microglia of E3KI and E4KI mouse hippocampus. To match the age at which PLX was first administered, we took 8-month-old untransplanted E3KI and E4KI mice, isolated hippocampal CD11b⁺/CD45^int microglia using fluorescence-activated cell sorting (FACS), and measured Csf1r levels on a BD LSR Fortessa^™ X-20 cell analyzer (Figure S2A). Comparing each of the six mice per condition (Figure S2B), we found that E4KI microglia showed significantly increased (38%) median fluorescence intensity (MFI) of Csf1r than E3KI microglia (Figures S2C and S2D). This suggests that higher baseline expression of Csf1r, which may be an APOE4-driven effect in microglia, renders microglia more susceptible to the effects of Csf1r antagonist PLX in E4KI mice.

To test whether the human neuron transplants may have influenced the increased PLX sensitivity of E4KI mice, we ran a small cohort study treating untransplanted E3KI and E4KI mice with PLX for approximately 2 months (Figure S2E). Similar to chimeric mice, untransplanted E4KI mice exhibited lower baseline microglial counts than untransplanted E3KI mice, and PLX-treated E4KI mice showed significantly reduced microglia counts compared to PLX-treated E3KI mice (Figure S2F). These data suggest that the baseline number and the particular PLX susceptibility of microglia in E4KI mice are not due to effects from the transplanted human neurons.

We also tested whether PLX treatment led to a depletion of astrocytes, the other main glial cell type in the hippocampus. Staining with astrocyte marker GFAP revealed no significant differences in hippocampal astrocyte number across all chimeric mouse conditions (Figures 2C and 2D). This result aligns with other studies reporting no changes in overall astrocyte viability upon PLX dosage both in vitro and in vivo^38,47,48.

Microglial depletion decreases Aβ aggregates in the presence of neuronal APOE4

We previously reported that transplanted human neurons produced Aβ aggregates and Thioflavin-S⁺ small plaques in EKI mice, with APOE4 exacerbating this phenotype²⁶. The ability of transplanted human neurons to generate Aβ aggregates, which do not form in human in vitro neuronal models^12,49 nor in mouse in vivo models without familial AD mutations of APP^50,51, provided a unique opportunity to study microglial impact with different APOE genotypes on human neuronal APOE-driven Aβ pathologies, which better mimic late-onset AD. Upon staining chimeric mouse hippocampal sections with human Aβ-specific monoclonal antibody 3D6, we found 3D6⁺ Aβ aggregates formed within or immediately surrounding the MAP2⁺ transplanted human neurons in all conditions (Figure 3A). Similarly to our earlier study with a chimeric mouse model²⁶, a significant majority of Aβ aggregates formed within 100 μm of the human neuron transplants (Figures S3A, S3C, and S3E) and very few in hippocampal regions outside that boundary (Figure S3F). These replicated results reiterate the conclusion that the human neuronal transplants are the source of these Aβ aggregates.

Figure 3. Human neuronal APOE isoform and microglial depletion affect Aβ pathology in chimeric mouse hippocampus.

(A) Representative images of Aβ aggregates within and immediately surrounding human neuronal transplants. Top row: 3D6⁺ Aβ aggregates (green); second row: composite of 3D6⁺ Aβ aggregates (green) and MAP2⁺ (gray) human neuron transplants. Scale bar, 50 μm.

(B) Quantification of number of 3D6⁺ Aβ aggregates/μm² within a 100 μm perimeter per transplant area (hE3-E3KI, n=5; hE3-E3KI-PLX, n=6; hE4-E4KI, n=9; hE4-E4-KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6). Unadjusted p-value for comparison between hE4-E4KI and hE4-E4KI-PLX (p = 0.0426) is significant.

(C) Representative images of Thioflavin-S⁺ dense-core deposits within and immediately surrounding human neuronal transplants. Top row: Thioflavin-S⁺ dense-core deposits (green); second row: composite of Thioflavin-S⁺ dense-core deposits (green) and MAP2⁺ (gray) human neuron transplants. Scale bar, 50 μm.

(D) Quantification of number of Thioflavin-S⁺ dense-core deposits/μm² within a 100 μm perimeter per transplant area (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6). Unadjusted p-value for comparison between hE4-E4KI and hE4-E4KI-PLX (p = 0.0318) is significant.

(E) Representative images of 3D6⁺ Aβ aggregates (magenta) and Thioflavin-S⁺ dense-core deposits (green) within human neuron transplants. Diffuse Aβ deposits were defined as 3D6⁺/Thioflavin-S⁻ deposits. Scale bar, 100 μm.

(F) Representative magnified images of 3D6⁺ Aβ aggregates (magenta, first column), Thioflavin-S⁺ dense-core deposits (green, second column), and composite of Aβ and Thioflavin-S images (third column) showing both dense-core and diffuse Aβ deposits within human neuron transplants. Top row: hE4-E4KI; bottom row: hEKO-E4KI. Scale bar, 20 μm.

(G) Quantification of number of diffuse (3D6⁺/Thioflavin-S⁻) Aβ deposits/μm² within a 100 μm perimeter per transplant area (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6). Unadjusted p-value for comparison between hEKO-E4KI-PLX and hE4-E4KI-PLX (p = 0.0238) is significant.

For all quantifications, values are normalized to the hE4-E4KI condition, and each dot represents one mouse. All data are expressed as mean ± S.E.M. Differences between groups were determined by two-way ANOVA with Benjamini’s post hoc test for multiple comparisons. Adjusted p-values (q-values) are displayed.

For all following quantifications, we compared: 1) hE4-E4KI mice versus hE3-E3KI mice and hE4-E4KI-PLX mice versus hE3-E3KI-PLX mice to measure the effects of APOE4 versus APOE3, 2) hEKO-E4KI mice versus hE4-E4KI mice and hEKO-E4KI-PLX mice versus hE4-E4KI-PLX mice to specifically measure the effects of neuronal APOE-KO versus APOE4, and 3) mice fed control chow versus PLX chow within each APOE genotype group to measure the effects of microglial depletion. All conditions were normalized to results from hE4-E4KI mice, since they served as the common factor in comparisons with both hE3-E3KI and hEKO-E4KI mice.

When we quantified the number of Aβ aggregates per square micrometer within 100 μm of the transplants, we found that APOE isoform had a significant effect on Aβ aggregate counts. Of the chimeric mice fed with control chow, the hE3-E3KI mice displayed significantly fewer Aβ aggregates than hE4-E4KI mice (Figure 3B), confirming previous findings that APOE4 exacerbates amyloid pathology^12,20,26,52. Upon microglial depletion, the average Aβ aggregate number in hE4-E4KI mice reduced by half, with an adjusted p-value (q-value) just shy of significance (p = 0.0559) (Figure 3B). These results suggest that the presence of microglia promotes neuronal APOE4-related Aβ pathology, aligning with several studies showing that microglia contribute to seeding amyloid plaques^29,34-36. In contrast, depleting microglia had no significant effect on Aβ aggregate number in hE3-E3KI-PLX mice. These data support the conclusion that the presence of both human APOE4 neurons and APOE4 microglia promote Aβ aggregate formation.

Neuronal APOE deficiency increases diffuse Aβ deposits independent of microglia

Furthermore, we initially found that hEKO-E4KI mice had a higher number of 3D6⁺ Aβ aggregate-like deposits than hE4-E4KI mice (Figure 3B). Unlike in hE4-E4KI-PLX mouse hippocampus, microglial depletion had no effect on the number of Aβ deposits in the hEKO-E4KI-PLX mouse hippocampus (Figure 3B). To investigate further, we compared the Aβ staining pattern in hEKO-E4KI mice, both with and without PLX treatment, to hE4-E4KI mice with neuronal APOE4, and noticed distinct morphological differences in the hEKO-E4KI Aβ aggregate-like deposits, including increased size (Figure S3I) and a diffuse feathered structure (Figure 3A). To better understand the nature of these deposits, we stained the chimeric mouse hippocampus with dense-core plaque marker Thioflavin-S (Figure 3C). As with Aβ, Thioflavin-S staining showed a significant increase in number of dense-core deposits in hE4-E4KI mice compared to hE3-E3KI mice (Figure 3D). However, hEKO-E4KI and hE4-E4KI conditions showed no differences in size or number of dense-core deposits (Figure 3D and Figure S3J), nor was there a significant reduction in dense-core deposits (although there appeared to be a trend in hE4-E4KI-PLX mice) upon removal of microglia. Importantly, while many Aβ aggregates in hE3-E3KI and hE4-E4KI mice colocalized with Thioflavin-S in a manner reminiscent of dense-core amyloid plaques, the majority of the Aβ aggregates in hEKO-E4KI mice were negative for Thioflavin-S (Figures 3E and 3F). We quantified these 3D6⁺/Thioflavin-S⁻ Aβ deposits, referred to as “diffuse” Aβ deposits, and saw a significant increase in diffuse Aβ deposits in hEKO-E4KI compared to hE4-E4KI mice (Figure 3G). These results indicate that removal of APOE from human neurons in vivo specifically promotes diffuse Aβ deposit formation, which is not significantly influenced by microglial depletion. Interestingly, this phenotype occurred in the presence of high levels of astrocyte-secreted APOE4 in hEKO-E4KI mouse hippocampus, indicating that the transplanted human neuron-produced APOE4 is a key driving factor for formation of dense-core Aβ plaques.

To further investigate the role of APOE in Aβ pathogenesis, we examined the percentage of dense-core and diffuse Aβ deposits that stained positive for APOE protein (Figure S4A). Many previous studies have demonstrated that APOE can interact directly with Aβ to stabilize aggregation, can be found deposited in Aβ plaques in vivo, and can direct Aβ-induced microglial chemotaxis^3,53-56. We saw that APOE was primarily found in dense-core, not diffuse, Aβ deposits (Figures S4B and S4C). Interestingly, the percentage of APOE⁺ dense-core Aβ deposits was also significantly higher in chimeric mice with human neuronal APOE (hE3-E3KI and hE4-E4KI) than in hEKO-E4KI mice, highlighting the potential role of human neuronal APOE in dense-core Aβ deposit etiology. Upon microglial depletion, hE4-E4KI-PLX mice showed a significantly increased percentage of APOE⁺ dense-core Aβ deposits compared to hE4-E4KI mice (Figure S4B), possibly due to reduced APOE clearance by microglia. In contrast, hEKO-E4KI-PLX mice, like hEKO-E4KI mice, had a low percentage of APOE⁺ dense-core Aβ deposits, even without microglial clearance of astrocytic APOE. These data indicate that neuronal APOE plays a key role in the composition and development of dense-core Aβ deposits.

Microglial depletion reduces p-tau deposits in the presence of neuronal APOE4

We next examined the chimeric mice for formation of p-tau deposits. Upon staining with p-tau-specific monoclonal antibodies AT8 and PHF1, we found both AT8⁺ and PHF1⁺ p-tau deposits in these chimeric mice (Figures 4A and 4D), reminiscent of p-tau pathology previously described in a chimeric mouse model of transplanted human neurons with a pathogenic tau mutation⁴⁴. A significant majority of these p-tau deposits, as with Aβ aggregates, were located within 100 μm of the human neuronal transplants (Figures S3B, S3D, and S3G). In hippocampal regions further than 100 μm from the transplant, p-tau deposits were rarely found (Figure S3H). Detailed examination of p-tau deposits indicated an intracellular localization, with aggregates immediately surrounded by soma-like MAP2⁺ staining (Figures 4B and 4E). These data strongly suggest that the p-tau deposits originate from the human neuron transplants.

Figure 4. Human neuronal APOE isoform and microglial depletion affect p-tau deposits in chimeric mouse hippocampus.

(A) Representative images of p-tau deposits within and immediately surrounding human neuronal transplants. Top row: AT8⁺ p-tau deposits (green); second row: composite of AT8⁺ p-tau deposits (green) and MAP2⁺ (gray) human neuron transplants. Scale bar, 50 μm.

(B) Many AT8⁺ p-tau deposits are present within transplanted MAP2⁺ human neurons. Each image represents the same field of view, stained for p-tau (AT8, green, first column), human neuronal transplants (MAP2, gray, second column), and a composite of AT8 and MAP2 (third column). Scale bar, 20 μm.

(C) Quantification of number of AT8⁺ p-tau deposits/μm² within a 100 μm perimeter per transplant area (hE3-E3KI, n=4; hE3-E3KI-PLX, n=7; hE4-E4KI, n=8; hE4-E4KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=5). Identical adjusted p-values (q-values) displayed in this graph represent true data, with the following unadjusted p-values: hE4-E4KI versus hE3-E3KI (p = 0.0150), hE4-E4KI versus hEKO-E4KI (p = 0.0176), hE4-E4KI versus hE4-E4KI-PLX (p = 0.0068).

(D) Representative images of p-tau deposits within and immediately surrounding human neuronal transplants. Top row: PHF1⁺ p-tau deposits (green); second row: composite of PHF1⁺ p-tau deposits (green) and MAP2⁺ (gray) human neuron transplants. Scale bar, 25 μm.

(E) Representative magnified image of p-tau deposits (PHF1, green, first column) within human neuronal. Transplants (MAP2, gray, second column), with a composite of PHF1 and MAP2 (third column). Scale bar, 20 μm.

(F) Quantification of number of PHF1⁺ p-tau deposits/μm² within a 100 μm perimeter per transplant area (hE3-E3KI, n=5; hE3-E3KI-PLX, n=6; hE4-E4KI, n=9; hE4-E4KI-PLX, n=9; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6). Unadjusted p-value for comparison between hE4-E4KI and hE3-E3KI (p = 0.0243) and between hE4-E4KI and hEKO-E4KI (p = 0.0278) are significant.

For all quantifications, values are normalized to the hE4-E4KI condition, and each dot represents one mouse. All data are expressed as mean ± S.E.M. Differences between groups were determined by two-way ANOVA with Benjamini’s post hoc test for multiple comparisons. Adjusted p-values (q-values) are displayed.

Quantification of the number of p-tau deposits per square micrometer within 100 μm of the transplants revealed that the average number of deposits was higher in the hippocampus of hE4-E4KI mice than in hE3-E3KI and hEKO-E4KI mice (Figures 4C and 4F). The reduced p-tau deposit counts in hE3-E3KIand hEKO-E4KI mice match a recently published study showing that, compared to APOE4, APOE3 and neuronal APOE4-KO dramatically reduce tau pathology in PS19/EKI mice²². Upon microglial depletion, hE4-E4KI-PLX mouse hippocampus showed significantly reduced levels of p-tau deposits compared to control hE4-E4KI mouse hippocampus (Figures 4C and 4F), although deposit size was unaffected across all conditions (Figure S3K). Microglial depletion did not significantly affect p-tau deposit numbers in hE3-E3KI-PLX or hEKO-E4KI-PLX mouse hippocampus. These results further suggest that E4KI microglia promote formation of amyloid and p-tau deposits in the presence of human neuronal APOE4.

Microglial depletion increases APOE levels within human neuron transplants

We then examined how microglial depletion affects APOE levels in human neuron transplants with different APOE genotypes. APOE protein analysis by immunohistochemistry within the transplants of the control chimeric mice revealed relatively low APOE expression, most of which colocalized with GFAP⁺ astrocytes (Figure 5A). Upon microglial depletion, APOE staining within the transplants increased significantly in hE3-E3KI-PLX and hE4-E4KI-PLX mice (Figure 5B), suggesting that microglia may contribute to APOE clearance from the vicinity of the transplanted human neurons. Interestingly, hEKO-E4KI-PLX mice had around half of the APOE staining within the transplants compared to hE4-E4KI-PLX mice, suggesting that at least ~50% of the transplant-localized APOE4 in hE4-E4KI mice is from the transplanted human neurons. Previous studies have identified neurons with high APOE levels in PS19/E4KI mice upon microglial depletion³⁸. Similarly, we found that APOE staining in the transplants of hE3-E3KI-PLX and hE4-E4KI-PLX mouse hippocampus was colocalized with both GFAP⁺ astrocytes and MAP2⁺ human neurons (Figure 5A), and APOE staining correlated with human neuron transplant area significantly in hE4-E4KI-PLX mice (Figure 5G) and moderately in hE3-E3KI-PLX mice (Figure 5F), but not in hE3-E3KI, hE4-E4KI, hEKO-E4KI, and hEKO-E4KI-PLX mice (Figures 5C-5E and 5H). Of note, microglia numbers were negatively correlated with intra-transplant APOE staining in hE4-E4KI-PLX mouse hippocampus (Figure S4H), but not other mouse groups (Figures S4D-S4G and S4I). Together, these data indicate that microglia may clear APOE generated from transplanted human neurons and mouse astrocytes.

Figure 5. Microglial depletion increases APOE levels within human neuronal transplants.

(A) Representative images of APOE staining within human neuronal transplants. Each column represents the same field of view, with a composite of APOE (green) with Iba1⁺ microglia (magenta, top row), MAP2⁺ human neuron transplants (magenta, second row), and GFAP⁺ astrocytes (magenta, third row, with magnified image insets showing APOE and GFAP overlap). Scale bar, 50 μm. Scale bar for magnified insets, 25 μm.

(B) Quantification of average APOE fluorescence intensity within the human neuron transplants (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=8; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6). All data are expressed as mean ± S.E.M. Differences between groups were determined by two-way ANOVA with Benjamini’s post hoc test for multiple comparisons.

(C–H) Correlations for each condition between % ApoE⁺ transplant area and size of transplants (hE3-E3KI, n=5; hE3-E3KI-PLX, n=7; hE4-E4KI, n=9; hE4-E4KI-PLX, n=8; hEKO-E4KI, n=4; hEKO-E4KI-PLX, n=6).

Pearson’s correlation analyses (two-sided).

For quantifications, each dot represents one mouse.

Transcriptional characterization of hippocampal microglia in chimeric mice

To investigate microglial reactivity to human neuron transplants in each condition, we performed scRNA-seq on microglia from the hippocampi of chimeric mice. We dissected and dissociated hippocampi from mice of chimeric conditions (hE3-E3KI, hE4-E4KI, hEKO-E4KI) and untransplanted control genotypes (E3KI and E4KI). Microglia were then purified via FACS (Figure S5). Flow cytometry gating removed debris, nonsingle cells, and dead cells, before sorting specifically for CD11b⁺/CD45^int microglia (Figures S5A and S5B). After scRNA-seq (Figure S6A), we clustered the sequenced cells (Figure S6B), tested for quality control (Figure S6C-S6H and Table S1), and assessed the purity of the sorted microglia. Cells were clustered by the Louvain algorithm⁵⁷ and visualized by Uniform Manifold Approximation and Projection (UMAP), revealing 16 distinct cell clusters (Figure S6B). Nearly all sequenced cells expressed microglia-specific markers Cx3cr1 and Csf1r (Figures S6I and S6J), confirming their microglial identity. Interestingly, expression of Cd74, a pro-inflammatory MHC-II marker for reactive microglia^58,59, highlighted select microglial clusters, including clusters 4, 7, 11, and 12 (Figure S6K), that warranted further investigation (see below). When tested for markers of non-microglial cell types, the sequenced cells had very sparse or no expression of Syn1 (neurons), Gfap (astrocytes), Slc17a7 (excitatory neurons), Gad1/Gad2 (inhibitory neurons), and Mbp (oligodendrocytes) (Figure S6L-S6P), further supporting the high purity of the sorted microglia.

Human neuronal APOE reduces homeostatic microglia in chimeric mouse hippocampus

Further analysis of the 16 cell clusters revealed several distinct clusters of interest, specifically clusters 1, 4, and 12. The percentage of cluster 1 microglia (Figures 6A-6E, pink) in hEKO-E4KI mice (37.6%) was nearly double that found in hE4-E4KI mice (21.9%) or hE3-E3KI mice (23.6%) (Figure 6F and Table S2). Analysis of the differentially expressed genes (DEGs) revealed upregulation of homeostatic microglial markers, including P2ry12, Cx3cr1, Fcrls, Cst3, Crybb1, and Sparc^60-63, relative to other microglial clusters (Figures 6G and 6J and Table S2). Additionally, cluster 1 showed strongly downregulated expression of genes associated with pro-inflammatory or activated microglia, like Cd52, APOE, Lyz2, and Cd74⁶⁴ (Figure 6G). Based on these transcriptomic features, cluster 1 representing a homeostatic microglial subpopulation reduced in hE4-E4KI and hE3-E3KI mice compared to hEKO-E4KI mice, suggesting human neuronal APOE prompts reduction of homeostatic microglia.

Figure 6. Pro-inflammatory microglia subpopulations enriched in hE4-E4KI mouse hippocampus.

(A–E) Feature plots highlighting cells in microglial clusters 1 (pink), 4 (gold), and 12 (blue).

(F) Quantification of fraction of cells per condition for clusters 1, 4, and 12.

(G–I) Volcano plot of the DEGs between cluster 1 (G), cluster 4 (H), or cluster 12 (I) and all other clusters. Dashed lines represent log₂ fold change threshold of 0.5 and p-value threshold of 10 x 10⁻²⁰. NS, not significant.

(J) Dot plot of normalized average expression of selected marker genes for clusters 1, 4, and 12. Dot size is proportional to percentage of cells expressing a given gene.

To measure homeostatic microglia levels in the chimeric mouse hippocampi, we stained brain sections with homeostatic microglia marker P2ry12 (Figure 7A). Consistent with the transcriptomic data, hEKO-E4KI mice showed significantly higher hippocampal P2ry12 coverage compared to both hE3-E3KI and hE4-E4KI mice (Figure 7B). As hEKO-E4KI and hE4-E4KI mice had microglia and astrocytes producing APOE4, this result further supports the conclusion that the presence of neuronal APOE4 prompts downregulation of homeostatic markers like P2ry12 in microglia.

Figure 7. Increased homeostatic microglia in hEKO-E4KI mouse hippocampus and increased pro-inflammatory microglia in hE4-E4KI mouse hippocampus.

(A) Representative images of homeostatic microglial marker P2ry12 staining within the hippocampus of chimeric mice. First column: P2ry12 (green); second column: microglial marker Iba1 (magenta); third column: composite of P2RY12 and Iba1 images. Scale bar, 100 μm.

(B) Quantification of P2ry12, as percent area of hippocampus (hE3-E3KI, n=5; hE4-E4KI, n=9; hEKO-E4KI, n=4). Identical adjusted p-values (q-values) displayed in this graph represent true data, with the following unadjusted p-values: hE3-E3KI versus hEKO-E4KI (p = 0.0128) and hE4-E4KI versus hEKO-E4KI (p = 0.0213).

(C) Representative images of pro-inflammatory microglial marker Cd68 staining within the hippocampus of chimeric mice. First column: Cd68 (green); second column: microglial marker Iba1 (magenta); third column: composite of Cd68 and Iba1 images. Scale bar, 100 μm.

(D) Quantification of Cd68, as percent area of hippocampus (hE3-E3KI, n=5; hE4-E4KI, n=9; hEKO-E4KI, n=4). Identical adjusted p-values (q-values) displayed in this graph represent true data, with the following unadjusted p-values: hE3-E3KI versus hE4-E4KI (p = 0.0166) and hE4-E4KI versus hEKO-E4KI (p = 0.0226).

For quantifications, each dot represents one mouse. All data are expressed as mean ± S.E.M. Differences between groups were determined by one-way ANOVA with Benjamini’s post hoc test for multiple comparisons. Adjusted p-values (q-values) are displayed.

Further exploring APOE isoform effects on this homeostatic microglia cluster, DEG analysis in hE4-E4KI microglial cluster 1 versus hE3-E3KI microglial cluster 1 revealed that hE4-E4KI microglia cluster 1 expressed increased levels of pro-inflammatory and AD-associated genes like Cdk8 (which promotes NF-κB signaling and chemokine expression)^65,66, Cmss1 (upregulated in APOE4 male mice)⁶⁷, Marcks (associated with cell motility and phagocytosis and highly expressed by senile amyloid plaque-associated microglia)^68,69, and Tmem176b (upregulated by amyloid plaque-adjacent microglia and associated with lysosomal degradation)⁷⁰ (Figure S7A and Table S3). Although we cannot determine whether these effects are attributed to microglial or neuronal APOE4, it is clear that APOE4 induces pro-inflammatory transcriptomic changes even in homeostatic microglia, compared to APOE3.

To isolate the effects of neuronal APOE4 on E4KI microglia, we examined DEGs in hE4-E4KI microglial cluster 1 versus hEKO-E4KI microglial cluster 1. Notably, hE4-E4KI microglia cluster 1 showed increased expression of antigen-presentation markers (Cd74 and H2-K1), vesicle/endosomal trafficking markers (Dynll1 and Washc4), and microglial chemotaxis marker P4ha1^71-73 (Figure S7B and Table S3). Taken together, these findings indicate that the presence of neuronal APOE4 may prime homeostatic E4KI microglia for a relatively pro-inflammatory, reactive state. Gene Ontology (GO) analysis of differential pathway regulation in hE4-E4KI microglial cluster 1 versus hEKO-E4KI microglial cluster 1 revealed enrichment of cytoplasmic ribosomal and protein folding pathways (Figure S7C and Table S3), suggesting neuronal APOE4 may alter protein translation and folding in homeostatic microglia.

Pro-inflammatory microglia subpopulations are more abundant in hE4-E4KI chimeric mice

In contrast to cluster 1, the percentage of microglia in cluster 4 (Figure 6A-6E, gold) was much higher in hE4-E4KI mice (30.2%) and hE3-E3KI mice (19.9%) compared to hEKO-E4KI mice (7.6%) (Figure 6F and Table S2). The difference in cluster 4 microglia representation was even more stark between transplanted and untransplanted mice, with untransplanted E3KI and E4KI mice having only 0.4% and 1.3% of microglia attributed to cluster 4 (Figure 6F and Table S2). These data imply that cluster 4 may represent a microglia subpopulation responding specifically to transplanted human neurons expressing APOE, especially APOE4. Closer DEG examination in this cluster revealed significant upregulation of MHC-II antigen presentation genes (Cd74, H2-Ab1, Tap1)^74,75 and chemokine/interferon response signaling genes (Stat1, Irf1, B2m, Ccl5, Ccl12)^76-78 (Figure 6H and Table S3). Each gene subset represents distinct markers for MHC-II expressing (MHC-II) microglia and interferon-responsive (IFN-R) microglia, two activated microglia subpopulations previously described in AD mouse models⁷⁴. Interestingly, these cluster 4 microglia do not fully match either MHC-II or IFN-R microglia, instead displaying a gene expression profile incorporating elements of both. Cluster 4 microglia also showed significant downregulation of homeostatic microglia markers (Fcrls, Mertk, Elmo1, Nav2) (Figure 6H and Table S3), indicative of activated microglia. Additionally, several downregulated genes from cluster 4 matched those downregulated specifically in reactive hippocampal microglia (Plxna4, Maml3, Fchsd2, Slc8a1)⁷⁹.

Compared to hE3-E3KI microglia, hE4-E4KI microglia from cluster 4 upregulated many pro-inflammatory markers (Figure S7D and Table S3), including genes from MHC-II microglia (H2-Ab1, H2-Aa, Ly6e) and IFN-R microglia (Stat1). Cluster 4 microglia from hE4-E4KI mice also showed increased levels of Cdk8 and Malat1, potent microglial inflammasome activators^{65,66,80,81,82}. Comparison of hE4-E4KI versus hEKO-E4KI microglia for cluster 4 yielded increased levels of MHC-II microglia genes (H2-Aa, H2-Eb1) and additional pro-inflammatory markers (Cd74, Stat1, Irf1) (Figure S7E and Table S3). Lamp1 and Lyz2, lysosomal genes and activated phagocytic microglia markers, were also upregulated in hE4-E4KI microglia, indicating that human neuronal APOE4 may promote microglial uptake/engulfment. GO analysis of differential pathway regulation in hE4-E4KI versus hEKO-E4KI microglia revealed that hE4-E4KI cluster 4 microglia showed alterations in pathways related to MHC protein complex, antigen processing/presentation, cellular response to interferons, and cytokine response (Figure S7F and Table S3). Interestingly, cytokine response and other pro-inflammatory genes were also upregulated in in vivo human microglia from a chimeric mouse model in response to amyloid⁸³, suggesting that the pro-inflammatory response in our chimeric mice is also relevant to the response of human microglia in AD. Together, these data indicate that human neuronal APOE, especially APOE4, induces pro-inflammatory microglial subpopulations, such as MHC-II microglia.

Microglial cluster 12 (Figures 6A-6E, blue), though small, was highly represented in hE4-E4KI mice. The percentage of cluster 12 microglia in hE4-E4KI mice was 3.2% (Figure 6F and Table S2), over 3-fold higher than the percentage in hE3-E3KI mice (0.9%) and over 30-fold higher than the percentage in hEKO-E4KI mice (0.09%) (Table S2). In fact, hEKO-E4KI mice have a similarly low percentage of microglia in cluster 12 as untransplanted mice (E3KI: 0.1%, E4KI: 0.01%) (Figure 6F and Table S2). This marked difference in the proportion of cluster 12 microglia between hE4-E4KI and hEKO-E4KI mice may indicate that cluster 12 represents a microglial subpopulation responding more strongly to transplanted human neurons with APOE, particularly APOE4. Closer examination of DEGs in microglial cluster 12 showed more pro-inflammatory genes upregulated than in microglial cluster 4. Although cluster 12 microglia displayed upregulated levels of some disease-associated microglia (DAM) markers, including Trem2, Ctsl, and Cd9, their expression profile does not fully match the DAM profile⁸⁴. Instead, the cluster 12 microglia gene expression profile most closely resembles MHC-II microglia, with a very high proportion of upregulated DEGs in the MHC antigen presentation pathway: Cd74, H2-Ab1, H2-Aa, H2-Eb1, H2-K1, and B2m^75,85 (Figure 6I and 6J and Table S2). Another study showed that MHC-II microglia exhibit increased Aβ phagocytic capacity relative to other microglia in APP/PS1 AD model mice⁸⁶. In line with these findings, cluster 12 microglia also exhibit a number of upregulated lysosomal genes (Lgals3bp, Ctsb, Ctsc, Ctss, Ctsd) associated with increased microglial phagocytosis and AD risk^87,88. Additionally, microglial exosome genes Cd9 and Cd81 were highly upregulated in cluster 12 microglia. As exosomes were previously identified as a potential mechanism for pathological amyloid and tau deposition^89,90, it is interesting to find significant upregulation of exosome markers in a microglial cluster particularly abundant in the hE4-E4KI mice, which had high levels of amyloid and tau pathologies. GO analysis of DEG-enriched pathways revealed that microglial clusters 4 and 12 had shared pathways (Figure S7G and Table S3) such as cytosolic ribosomal pathways, antigen processing and presentation pathways, and immune response pathways, in addition to each having unique pathways (Figure S7G and Table S3).

Interestingly, microglial clusters 7 and 11 also showed high Cd74 expression (Figure S6J). DEG analysis revealed that, like clusters 4 and 12, cluster 7 had increased expression of MHC proteins for antigen presentation, such as H2-Eb1, H2-Ab1, and H2-Aa (Figure S7H and Table S2). Cluster 11 had very small cell number and increased expression of Apo17c and Cd300e, both related to antigen processing and presentation (Figure S7I and Table S2). While these clusters were not particularly enriched in any of the chimeric mouse conditions with different APOE genotypes, they warrant further future study.

Equally interesting is the expression pattern of Cd68, a lysosomal protein highly expressed in activated phagocytic microglia in AD models^91-93. Several studies have demonstrated that expression of Cd68, unlike other microglial activation markers, was consistently increased in post-mortem brains with AD and APOE4 relative to controls⁹⁴ and was associated with neurodegeneration and cognitive decline in aging^21,95. In our scRNA-seq dataset, clusters 4 and 12 of hE4-E4KI microglia exhibited uniquely high Cd68 expression compared to other hE4-E4KI clusters (Figure S7J), a pattern absent in microglial clusters from other conditions. Fittingly, clusters 4 and 12 display upregulated markers of MHC-II microglia, which also highly express Cd68 in inflammatory AD model mice⁸⁶. These data support the conclusion that microglial clusters 4 and 12, both enriched in hE4-E4KI mice, are highly reactive and pro-inflammatory microglia subpopulations. Immunohistochemistry confirmed (Figure 7C) that hE4-E4KI mice displayed significantly higher hippocampal Cd68⁺ microglial coverage compared to hE3-E3KI and hEKO-E4KI mice (Figure 7D). With the transcriptomic data, this result indicates that neuronal APOE4 promotes a pro-inflammatory hippocampal microglia response.

Many publications have identified microglia-mediated T cell infiltration as contributors to neuronal dysfunction and neurodegeneration in AD^96-100. We detected CD3⁺ T cells near the human transplants, but found no significant difference in hippocampal T cell counts across conditions. However, hE4-E4KI-PLX mice exhibited a trend toward reduced T cell number, with only one-quarter the number of T cells found in hE4-E4KI mice (Figures S7K and S7L). This trend aligns with previously published results demonstrating microglia’s pivotal role in T cell recruitment and activation in an APOE4 context in tauopathy mouse hippocampi⁹⁶.

DISCUSSION

In this study, we examined the effects of human neuronal APOE isoforms on AD pathology utilizing an in vivo chimeric disease model. We demonstrated that the APOE4 isoform increased Aβ and p-tau deposits in hE4-E4KI mice compared to hE3-E3KI mice. These results match previous studies, including our previous chimeric AD model study, showing that APOE4 exacerbates amyloid and tau pathology^{12,20,22,26,38,52,101,102}. We also explored the effect of microglia depletion on human neuronal APOE4-related AD pathologies. Past microglial depletion studies in AD mouse models produced mixed results: microglia depletion reduced Aβ plaque burden in 5XFAD mice^{35,36,103,104}, reduced tau pathology in PS19/P301S tauopathy mice^38,89, had no effect on pathology in aged APP/PS1 or 5XFAD or 3xTg-AD tau mice^47,105-107, and even increased tau spread in tau-injected 5XFAD mice¹⁰⁸. Our results demonstrate that, in an AD model with human neurons in mice with no early-onset AD mutations, microglial depletion reduced Aβ and tau aggregates in hE4-E4KI mice. This indicates that microglia promote human neuronal APOE4-related Aβ and tau pathologies. As discussed in our previous paper establishing this model, the formation of aggregates in our chimeric disease model has a particular relevance to human late-onset AD²⁶, signifying strong potential for translatability of these findings.

Similar to our results in hEKO-E4KI mice, two recent studies in APP/PS1 AD model mice demonstrated that general removal of APOE induced a more diffuse morphology of amyloid aggregates^109,110. Importantly, our results identify neurons as a relevant source of APOE for compaction of amyloid aggregates. Interestingly, pathological analyses in post-mortem human brains showed that diffuse amyloid plaques appeared more in the brains of older cognitively normal patients than AD patients, who displayed more dense-core plaques¹¹¹. These results suggest that diffuse amyloid deposits formed in hEKO-E4KI mice may not be pathologically detrimental, in alignment with studies showing that removal of APOE, especially neuronal APOE4, is associated with ameliorated AD-related neurodegeneration and cognitive deficits^{22,38,112,113}.

The significant increase in transplant-localized APOE upon microglial depletion in our chimeric model resembles a previous finding that microglial depletion in PS19-E4 tauopathy mice led to increased APOE in hippocampal neurons³⁸. Both studies suggest that microglia clear APOE produced by neurons and astrocytes; thus, microglial depletion prevents this clearance and leaves behind neuronal APOE. Since both Aβ and p-tau pathologies were reduced upon microglia depletion in hE4-E4KI mice, these data support the hypothesis that neuronal APOE4-related Aβ and p-tau deposits require microglial participation.

Examination of the gene expression profile of hippocampal microglia in the chimeric mice revealed three main microglial clusters of interest: homeostatic cluster 1 and pro-inflammatory clusters 4 and 12. The gene expression profiles of clusters 4 and 12 partially resembled AD-related microglial subpopulations IFN-R and DAM, but the highest profile alignment for both clusters (particularly cluster 12) was with MHC-II microglia. Although less well-studied than DAM, microglia that highly express MHC-II markers are enriched in human AD brains^114,115. Past studies indicate that MHC-II microglia, which often express subsets of DAM markers¹¹⁶, are specifically associated with increased neuron loss and neuropathology in neurodegeneration model mice^{85,86,117,118}. As these MHC-II microglia in clusters 4 and 12 are particularly found in hE4-E4KI mice, these cells may contribute to the higher AD pathology found in hE4-E4KI mice, though the relevant mechanisms require further study.

Neuronal APOE appears to play a key role in modulating microglial status. Chimeric mice with human neuronal APOE showed reduction of P2ry12, one of several homeostatic microglial markers whose collective downregulation has been associated with neurodegeneration in AD¹¹⁹. Neuronal APOE4 in particular upregulated pathways related to activation, antigen presentation, and interferon response in microglial clusters 4 and 12, including dramatically increased expression of AD pathology-associated lysosomal gene Cd68 compared to other clusters. Interestingly, recent studies found that APOE4 microglia had a reduced pro-inflammatory response (compared to APOE3 microglia) in an ApoE-KO amyloid mouse model¹²⁰ and deleting APOE4 from microglia in APP/PS1:APOE4-KI mice restored microglial reactivity¹²¹. Our results indicate that upregulation of pro-inflammatory markers, like Cd68, in hE4-E4KI microglia is potentially a response specific to human neurons, particularly those expressing APOE4, or to Aβ and p-tau protein without pathological mutations. Depletion of specific pro-inflammatory clusters may serve to reduce AD-related pathologies in hE4-E4KI-PLX mice.

Limitations of the Study

There are limitations to consider in this study. To improve human neuron transplant survival, an immunosuppressant cocktail was administered in 3 doses over 6 days post-transplantation. These immunosuppressants, targeting mouse LFA-1α, CD40L, and CTLA-4, may have temporarily affected microglial number or activation states. However, it is unlikely that these effects persisted long enough to influence these results, as all samples were collected approximately 8 months after the final cocktail dose.

Unlike the isogenic APOE3 human cell line, the APOE-KO human cell line used in this study is not derived from the APOE4 human cell line or an AD patient background. However, our lab has demonstrated that this APOE-KO line, when transfected in vitro with APOE4 or APOE3 cDNA, produces similar phenotypes to the AD patient-derived APOE4 and APOE3 lines, including the neuronal production of Aβ peptides and accumulation of p-tau species¹². These results establish a baseline for comparison among the APOE-KO line and the isogenic APOE4 and APOE3 lines, particularly supporting that diffuse amyloid deposit formation in the hEKO-E4KI condition arise from a lack of neuronal APOE rather than absence of an AD patient background.

Although microglia were significantly depleted in all conditions, depletion was less effective in hE3-E3KI-PLX than hE4-E4KI-PLX mice. Accordingly, the lack of microglial effect on Aβ and p-tau deposits in hE3-E3KI-PLX mice may indicate either that APOE3 microglia play a limited role in pathology formation or that depletion of those microglial populations was insufficient to produce an effect. Additionally, validating the translatability of these findings with human microglia is crucial. In particular, a previously published chimeric model engrafting human stem cell-derived microglia into the mouse brain⁸³ presents a powerful tool to explore human-specific AD pathogenesis. Finally, it remains unclear whether formation of Aβ and p-tau deposits worsens neurotoxicity^{104,122,123,124,125} or protects against pathological proteins^126-128. One recent study in an APOE4 tauopathy mouse model demonstrated that microglial presence promotes both tau pathology and neurodegeneration³⁸, suggesting that microglial removal may be therapeutically beneficial. However, confirmation of whether depleting microglia and reducing aggregates will ameliorate or exacerbate neurodegeneration and cognitive deficits in AD is essential before pursuing microglial depletion as a possible therapeutic avenue.

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Mice with human APOE3 or APOE4 knocked-in at the mouse Apoe locus on a C57BL/6 background were originally obtained from Taconic¹²⁹. All animals were bred in-house using trio breeding producing 8–10 pups per litter on average, which were weaned at 28 days. Male littermates at 3-5 months of age were randomly assigned to experimental groups. Animals were housed in a pathogen-free barrier facility on a 12 hr light cycle (lights on at 7 am and off at 7 pm) at 19–23°C and 30%–70% humidity. Animals were identified by ear punch under brief isofluorane anesthesia and genotyped by polymerase chain reaction (PCR) of a tail clipping at weaning. All animals otherwise received no procedures except those reported in this study. All animal experiments were conducted in accordance with the guidelines and regulations of the National Institutes of Health, the University of California, and the Gladstone Institutes under IACUC protocol AN117112.

Cell Lines

All hiPSC lines were derived from human skin fibroblasts from donors and reprogrammed as previously described¹² and were maintained at 37°C with 5% humidity. All hiPSC lines were characterized for normal pluripotency gene expression, apoE genotypes, karyotypes, and capability of differentiating into neural stem cells as well as different types of neurons in culture. All hiPSC lines were tested negative for mycoplasma.

METHOD DETAILS

hiPSC Culture

The E4/4 hiPSC line was generated as described^130,131 from skin fibroblasts of a subject with an APOE4/4 (E4) genotype¹². The isogenic E3/3 (E3) hiPSC line was generated from this parental E4/4 hiPSC line as previously described¹². hiPSCs were maintained in mTeSR medium (85850, StemCell Tech) on 6-well plates precoated with hESQ, LDEV-free Matrigel (354277, Corning). The medium was changed daily, and cells were routinely passaged 1:10–1:15 using Accutase (NC9464543, Fishersci) for dissociation. Rho kinase (ROCK) inhibitor (1254, Tocris) was added to medium at 10μM on day of passaging.

Neuronal Differentiation of hiPSCs

hiPSCs were differentiated into neurons as previously described^12,26, with slight modifications to increase yield. Briefly: hiPSCs were dissociated with Accutase for 5–8 minutes before being quenched with warm (37°C) N2B27 medium made of 1:1 DMEM/F12 (11330032, Thermo Fisher) and Neurobasal Media (21103049, Thermo Fisher), 1% N2 Supplement (21103049, Thermo Fisher), 1% B27 (17504044, Thermo Fisher), 1% MEM Non-essential Amino Acids (11140050, Thermo Fisher), 1% Glutamax (35050061, Thermo Fisher), and 0.5% penicillin–streptomycin (15140122, Thermo Fisher). Dissociated hiPSCs were pelleted by centrifugation, resuspended in embryoid body media [10μM SB431542 (1614, Tocris) and 0.25μM LDN (04-0074, Stemgent) in N2B27] with 10μM ROCK inhibitor (1254, Tocris), and grown in suspension in a T-75 flask (12-565-349, Fisher Scientific). Flasks were shaken manually once per hour for the first 3 hrs of incubation. On day 2, embryoid bodies had formed, and fresh embryoid body medium was replaced (embryoid bodies pelleted, old medium aspirated, cells resuspended in fresh medium) to remove ROCK inhibitor. Embryoid body medium was replaced similarly on days 4 and 6. On day 8, spheres were plated as neural progenitors onto a 10cm dish precoated with growth factor reduced (GFR) Matrigel (CB-40230A, Fisher Scientific). Neural progenitors were allowed to form neuronal rosettes and sustained in N2B27 media alone for days 8–15. Half of the media was replaced every 48–72 hrs depending on confluency and media consumption. Neuronal rosettes were lifted on day 16 using STEMdiff^™ Neural Rosette Selection Reagent (05832, StemCell Tech) as directed by manufacturer and plated into 3 wells of a 6-well plated precoated with GFR Matrigel in N2B27 with 100ng/ml FGFb (100-18B, Peprotech) and 100ng/ml EGF (AF-100-15, Peprotech). This N2B27 medium with FGFb and EGF was replaced daily. On day 20, neural progenitors were dissociated with Accutase, quenched with N2B27, and resuspended in STEMdiff^™ Neural Progenitor Medium (05833, StemCell Tech) at 1.2x10⁶ cells/2ml for 1 well of 6-well plate, precoated with GFR Matrigel. Neural progenitor cells were fed with fresh Neural Progenitor Medium daily. On day 28, medium was switched to complete neuronal medium (10ng/ml BDNF (450-02, Peprotech) and 10ng/ml GDNF (450-10, Peprotech) in N2B27) with 10nM DAPT (2634, Tocris). Cells were fed with fresh complete neuronal medium daily for 7 days and then harvested for cell transplantation.

Cell Transplantation Preparation

hiPSC-derived neurons (D35-42) were washed in 1X PBS then incubated in warm Accutase (Millipore) for 15 min or until neurons dissociated with gentle tapping. Accutase (Millipore) was neutralized with N2B27 medium to bring total volume to around 30 mL and then cells were filtered through a 40 μm strainer (Fisher) to ensure a single cell suspension. Single cells were then centrifuged and resuspended to concentration of 500 cells/nL in 1X HBSS (GIBCO) supplemented with 10 ng/mL BDNF (Peprotech), 10 ng/mL GDNF (Peprotech) and 100 ng/mL DNaseI (Roche) and kept at 4°C until transplantation.

Stereotaxic Surgery for Cell Transplantation

Mice were anesthetized with an intraperitoneal injection of ketamine (60 mg/kg) and xylazine (30 mg/kg) and maintained on 0.8%-1.0% isofluorane (Henry Schein). Mice were secured in a stereotaxic alignment system model 940 using earbars and a tooth bar (Kopf Instruments). The scalp was prepared by removing hair with scissors and sterilizing with 70% ethanol. The scalp was then cut open using a scalpel and sterilized with 70% ethanol. The cranial sutures were better visualized using 3% hydrogen peroxide. Following identification of Bregma, bilateral rostral and caudal stereotaxic sites were drilled with a 0.5 mm microburr (Fine Science Tools) using coordinates X = ± 1.65, Y = 2.00, Z = 1.7 and X = ± 2.90, Y = 3.20, Z = 2.2, with Z measured from the surface of the brain. Cell suspensions (500 cells/nL) were loaded into 60 μm tip diameter, 30° beveled glass micropipette needles (Nanoject, Drummond Scientific Company). Approximately 30,000 cells per site were injected at 4 sites (2 per hemibrain) at a rate of 25 nL/sec and allowed to diffuse for 1 min. Each mouse received four total cell transplants, two per hemibrain hippocampus. Following surgery, mice were sutured with nylon monofilament non-absorbable 6-0 sutures (Henry Schein), and administered analgesics buprenorphine (0.0375 mg/kg intraperitoneally), ketofen (5 mg/kg subcutaneously), and saline (500μL intraperitoneally). Mice were monitored on a heating pad until ambulatory and provided Hydrogel for hydration. Immunosuppressants were administered immediately after transplantation (day 0) via intraperitoneal injection followed by injections on day 2, 4, and 6 post transplantation. Immunosuppressants were comprised of a cocktail of anti-mouse CD40L (CD154) (BioXCell), anti-mouse CTLA-4 (CD152) (BioXCell), and anti-mouse LFA-1a (CD11a) (BioXcell), and all were used at a concentration of 20 mg/kg.

PLX3397 and Control Chow

PLX3397 (Pexidartinib, HY-16749) was provided by MedChemExpress. For PLX chow, PLX3397 was incorporated into AIN-76A chow (Research Diets) at a concentration of 400mg/kg. Chimeric mice were fed either a control AIN-76A chow for 8 months, or fed control AIN-76A chow for 4 months and then PLX chow for 4 months.

Collection of Mouse Tissue

Mice were deeply anesthetized with intraperitoneal injections of avertin (Henry Schein) and transcardially perfused for 1 min with 0.9% saline. Right hemi-brains were drop-fixed in 4% paraformaldehyde (15710, Electron Microscopy Sciences), rinsed in PBS (Corning) for 24 hrs, and cryoprotected in 30% sucrose (Sigma) for 48 hrs at 4°C. The fixed right hemi-brains were sliced into 30μm coronal sections spanning the hippocampus on a freezing sliding microtome (Leica) and stored in cryoprotectant (30% Ethylene Glycol, 30% Glycerol, 40% 1X PBS) at −20°C. For the left hemi-brains, the hippocampus was dissected out and further prepared for microglia isolation and single-cell RNA-sequencing.

Immunohistochemistry

Multiple sections from each mouse (30μm thick, 300μm apart) were transferred to a 24-well plate and washed 2x5min with PBS to remove cryoprotectant. Sections were treated with UV radiation overnight in PBS. The next day, sections were washed 2x5min in PBS-T (PBS + 0.1% Tween-20), then incubated in PBS-TX (PBS + 0.5% Triton-X) for 2x15min. Sections were then blocked for non-specific binding in a solution of 10% Normal Donkey Serum (017000121, Jackson Immuno) in PBS-TX for 1 hr at RT. After blocking, sections underwent an additional incubation in M.O.M. blocking buffer (1 drop M.O.M. IgG (MKB-2213-1, Vector Labs) per 4ml PBS) for 1 hr at RT, and then incubated in primary antibody overnight at 4°C in a solution of M.O.M. protein concentrate (BMK-2202, Vector Labs) in PBS. The next day, sections were washed 3x10min with PBS-T and incubated in relevant fluorescently-labeled secondary antibodies with M.O.M. protein concentrate (BMK-2202, Vector Labs) in PBS for 1 hr at RT. Sections were then washed 3x10min in PBS-T, mounted, dried, coverslipped with Gold Prolong Antifade Mounting Medium (Thermofisher P36930), and sealed with clear nail polish. For Thioflavin-S staining, mounted sections were stained with 0.015% thioflavin-S in 50% ethanol diluted in PBS for 10 min and washed three times for 5 min/wash with 1X PBS before coverslipping. Slides were imaged with an Aperio VERSA slide scanning microscope (Leica) at 10X magnification or a FV3000 confocal laser scanning microscope (Olympus) at 20X, 40X, or 60X.

Immunohistochemistry included the following antibodies: [Primary antibodies] Human Nuclear Antigen (ab215755, Abcam, 1:100); 3D6 (Elan Pharmaceuticals, 1:1000); AT8 (MN1020, Thermofisher, 1:100); MAP2 (PA1-10005, Thermofisher, 1:200); APOE (178479, Millipore, 1:1000); GFAP (Z0334, Dako, 1:500); GFAP (MAB3402, Sigma, 1:800); OLIG2 (AB9610, Millipore Sigma, 1:1000); Synaptophysin (50-134-87, Fisher Scientific, 1:100); NeuN (ABN90, Millipore Sigma, 1:1000); Iba1 (019-19741, Wako, 1:200); Iba1 (ab5076, Abcam, 1:100); Thioflavin-S (sc391005, Santa Cruz Biotech, 0.015% in 50% Ethanol/PBS); PHF1 (Peter Davies, 1:100); CD3 (MCA2690, Bio-Rad, 1:100); P2ry12 (HPA014518, Sigma, 1:200); Cd68 (14-0681-82, Thermofisher, 1:200); DAPI (62248, Thermofisher, 1:20000). [Secondary antibodies] Alexa Fluor; Jackson Immuno Research, 1:1000.

Immunohistochemical Analyses

Immunohistochemical analyses were conducted in Fiji (ImageJ)¹³² via automated ImageJ macros to the extent possible. For all cell counts, images were set to a standardized threshold value across all conditions for each stain. For all aggregate counts, only particles within the hippocampus sized 10um^2 and above were analyzed. Aggregate densities were calculated and then normalized to transplant area to account for differing sizes of human cell transplants amongst conditions. For most quantifications, the results for each mouse represent the average of two hippocampal sections per mouse per stain. N for each condition is as follows: hE3-E3KI, 5 mice; hE3-E3KI-PLX, 6 mice; hE4-E4KI, 9 mice; hE4-E4KI-PLX, 9 mice; hEKO-E4KI, 4 mice; hEKO-E4KI-PLX, 6 mice. All compared confocal images were taken as z-stacks of similar depths and collapsed via z-project in Fiji (ImageJ). Analysts drawing regions of interest and setting standard threshold values were blinded to exclude possibility of bias. For all figures, unless otherwise specified, the following conditions were compared: hE3-E3KI vs hE3-E3KI-PLX, hE4-E4KI vs hE4-E4KI-PLX, hEKO-E4KI vs hEKO-E4KI-PLX, hE3-E3KI vs hE4-E4KI, hE4-E4KI vs hEKO-E4KI, hE3-E3KI-PLX vs hE4-E4KI-PLX, hE4-E4KI-PLX vs hEKO-E4KI-PLX.

Electrophysiology

Transplanted mice were deeply anesthetized with isoflurane and decapitated. The brain was rapidly removed from the skull and placed in the 4°C slicing solution comprised of 110 mM Choline Chloride, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 2 mM CaCl₂, 1.3 mM Na Pyruvate, 1 mM L-Ascorbic Acid, and 10 mM dextrose. 300mm sagittal sections were cut using a vibratome (VT 1200s, Leica). Following slicing, the slices were transferred into a vapor interface chamber aerated with 95% O₂/5% CO₂ gas mixture and allowed to recover at 34°C for 1 h before recording. Brain slices were placed into a submerged dual-side recording chamber (RC-27D, Warner Scientific) and perfused at 5 mL/min with oxygenated ACSF at 34°C. Recording and holding ACSF solution was comprised of 124 mM NaCl, 26 mM NaHCO₃, 10 mM Glucose, 1.25 mM NaH₂PO₄, 2.5 mM KCl, 1.25 mM MgCl₂, and 1.5 mM CaCl₂. Patch pipettes were filled with potassium-gluconate based solution containing 122.5 mM K-gluconate, 8 mM KCl, 10 mM HEPES, 2 mM MgCl₂, 0.2 mM EGTA, 2 mM ATPNa, and 0.3 mM GTPNa, pH set to 7.2–7.3 with KOH and with osmolarity of 270-280 mOsm. Neurons were imaged using a modified Olympus BX51 WI microscope with a 60x objective (Scientifica Inc, Great Britain). Patch-clamp recordings were performed using a Multiclamp 700B amplifier. The signals were sampled at 10 kHz and digitized using an Axon Digidata 1550B with Axon pCLAMP software (all from Molecular Devices, CA, USA). Data analysis was performed using custom software in IgorPRO (Wavemetrics Inc, USA). Spontaneous post-synaptic currents (PSCs) were recorded in voltage-clamp configuration at a holding potential of −70mV.

Microglia Isolation

Microglia were isolated using an adaptation of a previously described protocol¹³³. The hippocampus of each mouse’s left hemi-brain was dissected on ice, rinsed with cold Dissection Buffer (1X HBSS, 5ug/mL Actinomycin, 10uM Triptolide, 27.1ug/mL Anisomycin), and placed into a pre-chilled 12-well plate with 2 mL/well of Hibernate A Buffer (BrainBits Hibernate A Buffer, 5ug/mL Actinomycin, 10uM Triptolide, 27.1ug/mL Anisomycin). Hippocampi from 4-6 mice per condition were pooled into one well for further processing. Tissue from each pooled sample was moved to a 100mm dish, minced with a razor blade (Personna double-blade, 0.004” thickness), then placed in a 5mL Eppendorf tube with 1mL pre-warmed Enzyme Buffer (1X DPBS with Ca²⁺and Mg²⁺, 1.5mg/ml Collagenase D, 50ug/ml DNaseI, 5ug/ml Actinomycin, 10uM Triptolide, 27.1ug/ml Anisomycin). Samples in Eppendorf tubes were then incubated in a 35°C water bath for 30 minutes, with gentle trituration at the 15-, 25-, and 30-minute timepoints. The homogenate was then filtered (70um MACS SmartStrainers) into a 50mL conical tube and centrifuged at 4°C at 450g for 7 minutes. The supernatant was carefully aspirated, and cells were resuspended in three parts 1X DPBS and one part Debris Removal Solution (Miltenyi Biotec). The homogenate was moved to a 15mL conical tube, and 4mL 1X DPBS was carefully overlaid atop the cell solution. After centrifuging at 3000g for 10 minutes at 4°C, the supernatant solution and myelin debris layer were carefully aspirated, and the cells were washed with PBS and spun down one last time at 1000g for 10 minutes before staining.

For staining, cells were resuspended in Blocking Buffer (1X DPBS, 0.2% BSA (Sigma), 2.5ug/mL CD16/CD32 (BD Biosciences)) and a portion of each sample was distributed into different tubes for staining controls. After a 5-minute incubation at 4°C, cells were incubated in Antibody Solution (1X DPBS, 0.2% BSA, 1.89ug/mL CD45 PE-Cyanine7 (Thermofisher), 3.53ug/mL CD11b Brilliant Violet 421 (BioLegend)) for 40 minutes at 4°C. Cells were then washed with 1X DPBS with 0.2% BSA, filtered through a 40um Flowmi Cell Strainer (Millipore Sigma) into a pre-chilled 5mL FACS tubes (Stem Cell Technologies), and kept on ice until sorting. DRAQ7 APC dye was added to each sample just before cell sorting to determine live/dead cells. Using a BD FACS Aria Fusion I or BD Aria II at 4°C with a 100um nozzle (20 psi), live single cells were identified and gated per standard sorting protocols, then further sorted to isolate a CD11b⁺/CD45^int microglia population (Supp Fig 6). Sorted cells were collected in pre-chilled 2mL LoBind tubes (Fisher Scientific) with Collection Buffer (RPMI 1640 HEPES Modified with 2% FBS (Fisher Scientific)). Cells were centrifuged at 500g at 4°C for 5 minutes, then loaded onto 10x Genomics Next GEM chip G at cell counts of approximately 3,000 cells (PLX-depleted conditions) to 25,000 cells (control and untransplanted conditions). The scRNA-seq libraries were prepared using the Chromium Next GEM Single Cell 3’ Library and Gel Bead kit v.3.1 (10x Genomics) according to the manufacturer’s instructions. Libraries were sequenced on an Illumina NovaSeq 6000 sequencer at the UCSF CAT Core.

To measure Csf1r levels, microglia from untransplanted E3KI and E4KI mouse hippocampi (6 mice per condition) were isolated and sorted as described above. Mice aged around 8 months were selected for this experiment to best measure Csf1r levels at an age comparable to that of chimeric mice during PLX3397 treatment. Hippocampi from each mouse were treated as separate samples, and all samples were sorted for CD11⁺/CD45^int microglia and measured for CSF1R levels using a BD LSR Fortessa^™ X-20 cell analyzer.

Cell sorting and analyzing involved the following antibodies/dyes: Brilliant Violet 421^™ anti-mouse/human CD11b (Biolegend, 101236, 3.53ug/mL); CD45 monoclonal antibody (30-F11) PE-Cyanine7 (Thermofisher 25-0451-82, 1/89ug/mL); rat anti-mouse CD16/CD32 Mouse BD Fc Block (BD Biosciences 553141, 2.5ug/mL). DRAQ7^™ dye (Novus NBP2-81126).

Pre-Processing and Clustering of Mouse scRNA-seq Samples

The scRNA-seq samples included a total of five samples, one from each of the different conditions (E3KI, E4KI, hE3-E3KI, hE4-E4KI, hEKO-E4KI). Each sample combined 4-6 hippocampi, one from each male mouse. The demultiplexed fastq files for these samples were aligned to the standard mouse reference genome (2020 version, refdata-gex-mm10-2020-A)¹³⁴ separately using the 10x Genomics Cell Ranger v7.0.0 count pipeline, as described in the Cell Ranger documentation, and merged through CellRanger aggr. The include-introns flag for the count pipeline was set to true to count the reads mapping to intronic regions. For quality control assessment, cells were filtered to keep only cells with greater than or equal to 250 UMI and 200 unique genes, and less than 5% mitochondrial genes. Normalization was performed using the NormalizeData and ScaleData functions of the R package for single-cell analysis Seurat v4.3.0.1^135-137.

Graph-based clustering was performed using the Seurat v4.3.0.1 functions FindNeighbors and FindClusters. First, the cells were embedded in a k-nearest neighbor (KNN) graph (with k=20) based on the Euclidean distance in the PCA space. The edge weights between two cells were further modified using Jaccard similarity. Next, clustering was performed using the Louvain algorithm implementation in the FindClusters Seurat function. Clustering with 15 PCs and 0.5 resolution resulted in 16 distinct biologically relevant clusters, which was used for further analyses. Differentially Expressed Genes (DEG) were found using Seurat’s FindMarkers function. Volcano plots were created with EnhancedVolcano (Bioconductor package version 1.18.4) using unadjusted p-values, and the GO pathway analysis work was done using all significantly differentially regulated genes, both up- and down-regulated, with a combination of Enrichplot 1.18.4 and Clusterprofiler 4.6.2 (Bioconductor packages)¹³⁸.

QUANTIFICATION AND STATISTICAL ANALYSIS

Most immunohistochemical statistics were conducted as a two-way ANOVA in Graphpad Prism 10 (Graphpad Software Inc.). Figure 7 was analyzed using a one-way ANOVA. All data are shown as mean ± S.E.M. For all quantifications, unless specified, n represents the number of mice from which analysis images were gathered. No data were excluded based on statistical tests. Adjusted p-values (q-values) displayed are from the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli post hoc test for multiple comparisons (abbreviated in figures as Benjamini’s post hoc test for multiple comparisons). Adjusted P <0.05 was considered significant, and all significant p-values were included in Figures or noted in Figure legends.

Supplementary Material

1

Table S1. ScRNA-seq data related to Figure S6, including cells per condition, cell per cell cluster, nUMI per cell by cluster, genes per cell by cluster, and percent mitochondria per cell by cluster.

2

Table S2. ScRNA-seq data related to Figure 6 and Figure S7, including DEGs in clusters 1, 4, 7, 11, and 12 versus other clusters and percent of cells per condition assigned to each cluster.

3

Table S3. ScRNA-seq data related to Figure S7, including DEGs and pathways in hE4-E4KI vs hE3-E3KI and hE4-E4KI vs hEKO-E4KI for select clusters.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-Human Nuclear Antigen	Abcam	Cat# ab84680; RRID: AB_2154610
Mouse anti-Human Nuclear Antigen	Millipore Sigma	Cat# MAB1281; RRID:AB_94090
Rabbit anti-IBA1	Wako	Cat# 019-19741 RRID:AB_839504
Goat anti-IBA1	Abcam	Cat# ab5076; RRID: AB_2224402
Rabbit anti-IBA1	Abcam	Cat# ab178846; RRID: AB_2636859
Rabbit anti-GFAP	Agilent Dako	Cat# Z0334; RRID: AB_10013382
Mouse anti-GFAP	Sigma	Cat# MAB3402; RRID: AB_94844
Chicken anti-MAP2	Thermo Fisher Scientific	Cat# PA1-10005; RRID: AB_1076848
Mouse anti-3D6	Elan Pharmaceuticals	N/A
Mouse anti-AT8	Thermo Fisher Scientific	Cat# MN1020; RRID: AB_223647
Rabbit anti-pTau(Ser202/Thr205)	ABClonal	Cat# AP0894; RRID: AB_2771603
Goat anti-APOE	Millipore Sigma	Cat# 178479; RRID: AB_10682965
Rabbit anti-OLIG2	Millipore Sigma	Cat# AB9610; RRID: AB_570666
Mouse anti-Synaptophysin	Fisher Scientific	Cat# 50-134-87; RRID: Not registered
Guinea pig anti-NeuN	Millipore Sigma	Cat# ABN90; RRID: AB_11205592
Mouse anti-PHF1	Peter Davies Lab	N/A
Hamster anti-CD3	Bio-Rad	Cat# MCA2690; RRID: AB_905951
Rabbit anti-P2RY12	Millipore Sigma	Cat# HPA014518; RRID: AB_2669027
Mouse anti-CD68	Thermofisher Scientific	Cat# 14-0681-82; RRID: AB_2572857
Rat Anti-Mouse CD16/CD32 (Mouse BD Fc Block)	BD Biosciences	Cat# 553141; RRID: AB_394656
CD45 Monoclonal Antibody (30-F11), PE-Cyanine7 (100ug)	ThermoFisher Scientific	Cat# 25-0451-82; RRID: AB_2734986
Brilliant Violet 421™ anti-mouse/human CD11b Antibody (125uL)	BioLegend	Cat# 101236; RRID: AB_11203704
PE anti-mouse CD115 (CSF-1R) Antibody	Biolegend	Cat# 135505; RRID: AB_1937254
Donkey anti-mouse 488nm	Thermo Fisher Scientific	Cat# A-21202; RRID:AB_141607
Donkey anti-rabbit 488nm	Thermo Fisher Scientific	Cat# A-21206; RRID:AB_2535792
Donkey anti-goat 488nm	Abcam	Cat# ab150129; RRID: AB_2687506
Donkey anti-mouse 594nm	Thermo Fisher Scientific	Cat# R37115; RRID:AB_2556543
Donkey anti-rabbit 594nm	Thermo Fisher Scientific	Cat#A-21207; RRID:AB_141637
Chemicals, Peptides, Disposable Materials, and Recombinant Proteins
Mouse-on-Mouse Blocking Reagent	Vector Labs	Cat# MKB-2213;
Normal Donkey Serum	Jackson Labs	Cat# 017-000-121
Triton-X	Millipore Sigma	Cat# T8787-250mL
Tween-20	Millipore Sigma	Cat# P2287-500mL
Accutase	Millipore Sigma	Cat# SCR005
NEAA	GIBCO	Cat# 11140-050
Glutamax	GIBCO	Cat# 35050-061
Pen/Strep	GIBCO	Cat# 15140-122
SB	Stemgent	Cat# 04-0010-10
LDN	Stemgent	Cat# 04-0074
Rock Inhibitor	Tocris	Cat# 1254
bFGF	Peprotech	Cat# 100-18C
EGF	Peprotech	Cat# AF-100-15
Poly-L-Lysine	Sigma	Cat# 4707
PBS	Thermo Fisher Scientific	Cat# 14190250
Laminin	GIBCO	Cat# 23017-015
DMEM/F12	GIBCO	Cat# 11330-032
Neurobasal	GIBCO	Cat# 21103-049
DAPT	Tocris	Cat# 2634
BDNF	Peprotech	Cat# 450-02
GDNF	Peprotech	Cat# 450-10
Matrigel hESC Qualified	Corning	Cat# 354277
Matrigel Growth Factor Reduced	Corning	Cat# 354230
N2 Supplement	Thermo Fisher Scientific	Cat# 17502048
mTESR	Stemcell technologies	Cat# 85850
B27 Supplement	Thermo Fisher Scientific	Cat# 17504044
STEMdiff Neural Progenitor Medium	Stem Cell Technologies	Cat# 05833
Ketamine	Henry Schein	Cat# 1049007
Xylazine (Anased)	Henry Schein	Cat# 1311139
Isofluorane	Henry Schein	Cat# 029405
Buprenorphine	Henry Schein	Cat# 055175
Ketofen	Henry Schein	Cat# 005487
Avertin (2,2,2-Tribromoethanol)	Millipore Sigma	Cat# T48402
ProLong Gold Antifade Mountant	Thermo Fisher Scientific	Cat# P36930
InVivoMab anti-mouse CD40L (CD154) Clone MR-1	BioXCell	Cat# BE0017-1
InVivoMab anti-mouse CTLA-4 (CD152) Clone 9D9	BioXCell	Cat# BE0164
InVivoMab anti-mouse LFA-1a (CD11a) Clone M17/4	BioXcell	Cat# BE0006
Collagenase D	Sigma-Aldrich	Cat# 11088866001
Debris Removal Solution	Miltenyi Biotec	Cat# 130-109-398
Critical Commercial Assays
Chromium Next GEM Single Cell 3' Kit v3.1	10x Genomics	Cat#1000268
Deposited data
scRNA-seq of microglia from E3KI, E4KI, hE3-E3KI, hE4-E4KI, hEKO-E4KI mice	This paper	GEO: GSE248020
Experimental Models: Cell Lines
Human: ApoE4/4 iPSC line	Wang et al.12	N/A
Human: Isogenic ApoE3/3 iPSC line	Wang et al.12	N/A
Human: ApoE-KO iPSC line	Wang et al.12	N/A
Experimental Models: Organisms/Strains
Mouse: ApoE4-KI: B6.129P2-Apoetm3(APOE*4)Mae N8	Taconic	Cat#1549-M
Mouse: ApoE3-KI: B6.129P2-Apoetm2(APOE*3)Mae N8	Taconic	Cat#1548-M
Software and Algorithms
FlowJo v10.9.0	BD Biosciences	https://www.flowjo.com/solutions/flowjo
Fiji v2.14	Schindelin et al.132	https://imagej.net/software/fiji/downloads
Seurat v4.3.01	Stuart et al.135 ; Hao et al.136; Butler et al.137	https://cran.r-project.org/web/packages/Seurat/index.html
R v.4.2.2	R Core Team, 2019	http://www.R-project.org/
RStudio 2022.07.2+576	R Core Team, 2019	https://cran.rstudio.com/
CellRanger v7.0.0	10x Genomics	https://github.com/10XGenomics/cellranger
patchwork: 1.1.2	R Core Team, 2019	https://cran.r-project.org/package=patchwork
dplyr: 1.1.3	R Core Team, 2019	https://cran.r-project.org/package=dplyr
ggplot2: 3.4.4	R Core Team, 2019	https://cran.r-project.org/package=ggplot2
plotrix: 3.8.2	R Core Team, 2019	https://cran.r-project.org/package=plotrix
colorspace: 2.1.0	R Core Team, 2019	https://cran.r-project.org/package=colorspace
EnhancedVolcano: 1.16.0	Bioconductor	https://bioconductor.org/packages/release/bioc/html/EnhancedVolcano.html
clusterProfiler: 4.6.2	Bioconductor; Wu et al.138	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html
Axon pCLAMP 11	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clamp-system/acquisition-and-analysis-software/pclamp-software-suite
IgorPRO	Wavemetrics Inc	https://www.wavemetrics.com/products/igorpro
Other
BD FACS Aria Fusion I	BD Biosciences	https://www.bdbiosciences.com/content/dam/bdb/marketing-documents/BD_FACSAria_fusion_brochure.pdf
BD FACS Aria II	BD Biosciences	http://www.bdbiosciences.com/documents/BD_FACSAria_II_cell_sorter_brochure.pdf
BD LSRFortessa X-20 Cell Analyzer	BD Biosciences	https://go.bd.com/fortessax20-cellanalyzer.html
Chromium Controller & Next GEM Accessory Kit	10x Genomics	Cat# 1000204
Illumina NovaSeq 6000	Illumina	https://www.illumina.com/systems/sequencing-platforms/novaseq.html
BD FACS Aria Fusion I	BD Biosciences	https://www.bdbiosciences.com/content/dam/bdb/marketing-documents/BD_FACSAria_fusion_brochure.pdf
Olympus BX51 WI	Scientifica Inc	https://www.scientifica.uk.com/products/olympus-bx51-wi
Axon Digidata 1550B	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clampsystem/digitizers/axondigidata-1550b-plushumsilencer
Ultra-quiet Imaging Chambers for Slice Studies	Warner Scientific	Cat# RC-27D
Fully automated vibrating blade microtome	Leica	Cat# VT 1200s
Cell Scraper	Fisher Scientific	Cat #08-100-241
24-well Coverslips	Corning	Cat# 354085
Vector Laboratories ImmEDGE™ Hydrophobic Barrier Pen	VWR International Inc	Cat# 101098-065

Highlights.

• In vivo chimeric modeling of human APOE4-related Alzheimer’s Disease pathogenesis

• Neuronal APOE4 promotes Aβ and p-tau pathologies in the chimeric mouse model

• Microglial depletion reduces neuronal APOE4-induced Aβ and p-tau pathologies

• Neuronal APOE4 promotes inflammatory response of microglia in the chimeric model