Prenatally-Derived Macrophages Support Choroidal Health and Decline in Age-Related Macular Degeneration

Hallmark findings in age-related macular degeneration (AMD) include the accumulation of extracellular lipid and vasodegeneration of the choriocapillaris. Choroidal inflammation has long been linked to AMD, but little is known about the immune cells of the human choroid. Using 3D multiplex immunofluorescence, single-cell RNA-sequencing, and flow cytometry, we unravel the cellular composition and spatial organization of the human choroid and the immune cells within it. We identify two populations of choroidal macrophages with distinct FOLR2 expression that account for the majority of myeloid cells. FOLR2+ macrophages predominate in the nondiseased eye, express lipid-handling machinery, uptake lipoprotein particles, and contain high amounts of lipid. In AMD, FOLR2+ macrophages are decreased and functionally attenuated with respect to lipoprotein metabolism. In mice, FOLR2+ macrophages are negative for the postnatal fate-reporter Ms4a3, and their depletion causes an accelerated AMD-like phenotype. Our results show that prenatally-derived resident macrophages decline in AMD and are implicated in multiple hallmark functions known to be compromised in the disease. Citation: Fortmann, Seth D. et al. 2025. Prenatally-Derived Macrophages Support Choroidal Health and Decline in Age-Related Macular Degeneration. Journal of Experimental Medicine. In Press. This repository contains reanalyzed single-cell RNA-sequencing (scRNA-seq) data from RPE/Choroid of nondiseased and age-related macular degeneration human donors. There are 99 scRNA-seq samples derived from 7 previously deposited GEO Accessions (GEO Accessions: GSE203499, GSE230348, GSE210543, GSE135922, GSE149100, GSE183320, GSE202735). Each sample is referenced below with its original BioSample identifier per the depositing authors. All scRNA-seq data are 3' 10X Genomics and raw sequencing data are available per the previously deposited BioSample FASTQs. SCRNA-SEQ PROCESSING Raw FASTQ files were downloaded from the GEO database (GEO Accessions: GSE203499, GSE230348, GSE210543, GSE135922, GSE149100, GSE183320, GSE202735). Samples were demultiplexed and aligned using CellRanger Count (Version 7.0.2). The prebuilt human reference transcriptome from 10X Genomics, GRCh38, was used for alignment. CellRanger output files were loaded into Scanpy (Wolf et al., 2018), and the scAR package (Sheng et al., 2022) from scvi-tools (Gayoso et al., 2022) was used to identify and remove ambient RNA. To filter out low quality cells, we used the following thresholds: >300 for total genes, >300 for total counts, and <0.35 for mitochondrial proportion. To identify potential doublets we used 2 algorithms, Solo (Bernstein et al., 2020) and DoubletDetection (Gayoso and Shor, 2022), and we removed cells that were identified as probable doublets by both. For size factor normalization of the denoised raw counts, we used scranPY (Fortmann, 2023), a python implementation of r-scran::computeSumFactors (Lun et al., 2016). For dimensionality reduction, we used 3,000 highly variable genes, principal component analysis (PCA) using 20 components, Harmony for batch correction using individual sample identifiers (Korsunsky et al., 2019), nearest neighbor analysis using 80 neighbors, and UMAP with default parameters. For clustering, we used the Leiden algorithm with a resolution of 0.25 (Traag et al., 2019). Lastly, we removed a cluster containing contaminating rods and 3 other small clusters of unknown origin. For subclustering of myeloid cells, we subset the data on the CD14+ cluster and then used the single-cell Variational Inference (scVI) model from scvi-tools to compute a latent space representation for dimensionality reduction. We removed 1 sample from the dataset that contained a large population of unknown cells that did not overlap with the other 98 samples. Cell cycle scores were computed with the Scanpy function score_genes_cell_cycle() using the cell cycle gene list from (Tirosh et al., 2016). The denoised expression matrix was used along with 3 categorical covariates (the individual sample identifier, the donor identifier, and the study identifier) and 3 continuous covariates (the mitochondrial proportion, the G2M score, and the S score). Nearest neighbors were recomputed using the scVI latent space and then UMAP was recomputed using default parameters. PROCESSED FILES Scanpy (.H5AD) and Seurat (.RDS) versions of [i] complete RPE/choroid merged dataset and [ii] subclustered myeloid dataset. The processed files contain 3 different gene expression matrices: [1] scranPY normalized expression ("scranPY"; see GitHub sfortma2/scranPY), [2] denoised counts ("denoised"; see GitHub Novartis/scar), [3] raw cell ranger counts ("CR_counts"). These files also contain a variety of metadata: "GEO_SampleID" = the per sample ID used in this repository, "GEO_Sample_Title" = the sample title used by the original depositing authors, "GEO_Donor" = the donor identifier used in this repository; some samples come from the same donor, "BioSample" = the BioSample identifier of each sample, "location" = macular vs peripheral, "diseasegroup" = nondiseased vs AMD, "diseasetype" = nondiseased vs nvAMD vs atrophic AMD vs early AMD vs intermediate AMD, "age" = donor age if available, "sex" = donor sex if available, "location_diseasetype" = combined location and diseasetype, "location_diseasegroup" = combined location and diseasegroup, "barcode" = original barcode, "clusters" = cell cluster, "clusters_less" = generalized cell cluster; only in complete RPE/choroid merged dataset, "clusters_more" = cell clusters plus immune specific clusters; only in complete RPE/choroid merged dataset, "n_counts" = number of transcripts in cell, "log_counts" = log number of transcripts in cell, "n_genes" = number of genes in cell, "log_genes" = log number of genes in cell, "size_factors" = scranPY size factor used for normalization, "percent_mito" = ratio of mitochondrial transcripts in cell, "percent_ribo" = ratio of ribosomal transcripts in cell, "count_cutoff" = transcript cutoff used for QC, "gene_cutoff" = gene cutoff used for QC, "mito_cutoff" = mitochondrial ratio used for cutoff, "doublet_score_SOLO" = doublet score calculated using SOLO, "singlet_score_SOLO" = singlet score calculated using SOLO, "doublet_labels_SOLO" = doublet vs singlet label per SOLO, "doublet_score_DoubletDetection" = doublet score calculated using DoubletDetection, "doublet_labels_DoubletDetection" = doublet vs singlet label per DoubletDetection, "CR__reads_per_cell" = CellRanger reads per cell for sample, "CR__total_reads" = CellRanger total reads for sample, "CR__seq_sat" = CellRanger sequencing saturation for sample, "CR__Q30_RNA" = CellRanger RNA quality for sample, "CR__intronic_reads" = CellRanger intronic reads for sample, "CR__chemistry" = detected chemistry per CellRanger.