Abstract
Purpose
Age-related macular degeneration (AMD) is a leading cause of irreversible vision loss in the elderly, characterized by chronic retinal inflammation and immune dysregulation. While myeloid cells have been increasingly implicated in AMD pathogenesis, the specific immune subsets responsible remain poorly defined. This study aimed to identify causal immune cell populations and elucidate their functional roles in AMD progression.
Methods
We employed an integrative multiomics strategy encompassing Mendelian randomization (MR) analysis using genome-wide association study summary statistics, single-cell RNA sequencing (scRNA-seq) analysis of retinal pigment epithelium (RPE)/choroid tissues from patients with AMD and healthy controls (GSE230348), and flow cytometric (FCM) validation in a sodium iodate–induced dry AMD mouse model.
Results
MR analysis identified a significant causal association between CD64 expression on CD14⁻CD16⁻ monocytes and increased AMD risk (odds ratio, 1.179; P < 0.001). scRNA-seq profiling revealed a pronounced enrichment of CD14⁻CD16⁻ monocytes in AMD tissues, with FCGR1A (CD64) expression specifically localized within this subset. Pseudotime trajectory analysis demonstrated dynamic activation and differentiation states among monocyte populations in AMD. Ligand–receptor interaction modeling identified three major signaling pathways, MIF–CD74–CXCR4, IGF1–IGF1R, and SEMA3C–PLXND1, mediating interactions between CD14⁻CD16⁻ monocytes and RPE cells. FCM analysis of retinal single-cell suspensions in AMD mice confirmed a significantly higher proportion of CD64⁺ myeloid cells compared to controls.
Conclusions
This study identifies CD64⁺CD14⁻CD16⁻ monocytes as potential contributors to AMD and reveals their putative immunomodulatory crosstalk with RPE cells. These findings highlight CD64 as a promising biomarker and therapeutic target for mitigating myeloid-driven inflammation in AMD.
Keywords: age-related macular degeneration, CD64+ monocytes, mendelian randomization, single-cell RNA sequencing, immune microenvironment
Age-related macular degeneration (AMD) is a progressive retinal degenerative disorder and the leading cause of irreversible visual loss among the elderly population worldwide.1 Clinically, AMD is characterized by retinal pigment epithelium (RPE) degeneration, photoreceptor cell loss, and, in advanced cases, choroidal neovascularization or geographic atrophy.2 With the global acceleration of population aging, the public health burden of AMD is expected to increase significantly. Despite extensive research, the pathogenic mechanisms underlying AMD remain incompletely understood, and effective early diagnostic markers and therapeutic targets are still lacking.3
Emerging evidence suggests that the innate immune system, particularly the mononuclear phagocyte system, plays a pivotal role in the development and progression of AMD.4,5 In physiological conditions, microglia are the predominant immune cells residing in the inner retina, where they contribute to immune homeostasis.6 Under pathological conditions such as AMD, microglia become activated and migrate toward the outer retina. Simultaneously, disruption of the blood–retinal barrier facilitates the infiltration of circulating monocytes into the RPE and choroidal compartments.6 The accumulation and activation of these myeloid-derived cells have been implicated in the amplification of local inflammation, oxidative stress, and tissue injury.7
Among myeloid subsets, monocytes expressing Fc gamma receptors (FcγRs), including CD16, CD32, and CD64, have been shown to regulate local immune responses in several chronic inflammatory diseases.8,9 These receptors are involved in antigen presentation, phagocytosis, and proinflammatory signaling. However, the phenotypic heterogeneity, spatial distribution, and functional relevance of these monocyte subsets in the context of AMD remain largely undefined.
To systematically investigate these unresolved issues, we implemented a multitiered research approach incorporating genomic inference, transcriptomic profiling, and experimental verification. Mendelian randomization (MR) analyses were performed using large-scale genome-wide association study (GWAS) summary statistics to evaluate potential causal associations between peripheral immune cell subsets and AMD risk. Second, single-cell RNA sequencing (scRNA-seq) data derived from human RPE/choroid tissues were analyzed to investigate the abundance, phenotypic diversity, and transcriptional activity of relevant immune populations. Third, flow cytometric (FCM) analysis was conducted in a sodium iodate–induced dry AMD mouse model to provide in vivo validation of key findings. Through the integration of genetic, transcriptomic, and experimental data, this study aims to delineate the immunological landscape of AMD and identify candidate cellular targets with potential diagnostic and therapeutic value.
Materials and Methods
Mendelian Randomization Analysis
Data Sources for Exposure and Outcome
We obtained 731 immune-related traits from the GWAS Catalog (accession numbers GCST0001391 to GCST0002121), including 192 relative cell counts, 32 morphological parameters, 118 absolute cell counts, and 389 median fluorescence intensities, based on a cohort of 3757 individuals of European ancestry. Summary statistics for AMD were derived from the FinnGen R11 release (https://www.finngen.fi/en), including 430,221 individuals (11,023 cases and 419,198 controls). The AMD phenotype was defined based on national hospital discharge and mortality records using International Classification of Diseases, 10th Revision code H35.3 (degeneration of macula and posterior pole), with the inclusion criterion of first diagnosis at or above 50 years of age. As the phenotype was not further stratified into dry/wet or early/late AMD, it reflects the overall genetic susceptibility to AMD in the population rather than specific clinical subtypes. Since the exposure and outcome data sets were from different sources, there was no sample overlap.
Instrumental Variable Selection
Single-nucleotide polymorphisms (SNPs) were selected using a genome-wide significance threshold of P < 1 × 10⁻⁶ (and P < 1 × 10⁻⁸ for AMD). Linkage disequilibrium was pruned by setting r² < 0.001 within a 10,000-kb window. F-statistics (F = β²/SE²) were calculated to evaluate instrument strength, with SNPs with F < 10 excluded. To minimize potential pleiotropic effects, we cross-referenced all instrumental SNPs with the GWAS Catalog and removed those with genome-wide significant associations (P < 5 × 10⁻8) with major AMD confounders such as smoking, alcohol use, diabetes, hypertension, or body mass index (BMI). Palindromic SNPs were excluded, and directionality was confirmed using Steiger filtering.
MR Analysis and Sensitivity Tests
MR analyses were conducted using the R packages TwoSampleMR and MR-PRESSO (version 4.4.0). The primary method was inverse-variance weighted (IVW), with complementary approaches including weighted median, MR-Egger, simple mode, and weighted mode. Heterogeneity was assessed using Cochran's Q statistic (P < 0.05 considered significant), while MR-Egger intercept was used to test for horizontal pleiotropy. MR-PRESSO global and outlier tests were applied to identify and correct for influential SNPs. Multiple testing was controlled using false discovery rate (FDR) correction, with adjusted P < 0.05 considered statistically significant and 0.05 < P < 0.1 interpreted as suggestive.
Single-Cell Transcriptomic Analysis
Data Acquisition and Preprocessing
scRNA-seq data were obtained from the GEO data set GSE230348, which was submitted by the Stambolian Lab at the University of Pennsylvania. This data set provides high-quality single-cell expression matrices and metadata, including samples from 13 patients with AMD and 7 healthy donors. Donor tissues include both retinal and RPE/choroid regions, covering macular and peripheral areas. AMD samples are further classified based on clinical presentation into early to intermediate AMD, intermediate AMD, and atrophic AMD. Quality control and downstream analysis were performed using the Seurat package (v4.0). Cells were retained if the number of detected genes ranged from 200 to 8000 and the proportion of mitochondrial genes was below 10%. After normalization, principal component analysis and t-distributed stochastic neighbor embedding were used for dimensionality reduction and visualization. Cell types were annotated based on canonical marker genes for immune and nonimmune populations (Supplementary Table S1).10
Monocyte Subtype Annotation and Proportion Analysis
Based on MR results, monocyte populations were further stratified into four subsets according to CD14 and CD16 expression: CD14⁺CD16⁺, CD14⁺CD16⁻, CD14⁻CD16⁺, and CD14⁻CD16⁻. The relative abundance of each subset was calculated per donor sample. To compare the proportions of these subsets between AMD and control groups, statistical analyses were performed using the two-tailed Mann–Whitney U test, as the data were nonparametric and derived from independent individuals. All statistical tests were conducted using R, and P < 0.05 was considered significant.
CD64 Expression Analysis
Expression levels of CD64 (encoded by FCGR1A) were examined across annotated monocyte subsets, with a focus on spatial expression patterns and disease-specific enrichment, particularly in CD14⁻CD16⁻ monocytes.
Pseudotime Trajectory Inference
Using the Monocle2 package, pseudotime trajectories of monocytes were constructed based on differentially expressed genes. Dimensionality reduction was performed via the DDRTree algorithm, and cellular states were inferred to visualize differentiation trajectories. Differences in the distribution of CD14⁻CD16⁻ monocytes along the pseudotime trajectory between AMD and control groups were analyzed to infer dynamic activation patterns.
Cell–Cell Communication Analysis
Cell–cell communication analysis was performed using the CellChat R package to infer ligand–receptor interactions between annotated monocyte subsets and RPE cells. Signaling networks were constructed for both AMD and control groups, and bubble plots were used to visualize communication probabilities and interaction strengths across different signaling pathways.
Flow Cytometric Validation
Tissue Processing and Single-Cell Suspension
A dry AMD mouse model was established using 6- to 8-week-old C57BL/6J mice by intraperitoneal injection of sodium iodate (NaIO₃, 35 mg/kg; Aladdin, Shanghai, China; cat. S108351).11 Control animals received an equivalent volume of phosphate-buffered saline (PBS). A total of 18 mice were used across three independent experiments (n = 3 per group per experiment) to ensure reproducibility. On day 3 postinjection, mice were euthanized, and retinas were carefully dissected. Following vitreous removal, the retinal tissue was enzymatically dissociated using collagenase I (1 mg/mL) and DNase I (0.1 mg/mL) at 37°C for 30 minutes with gentle trituration. The digestion was terminated by adding 10% fetal bovine serum. Cell suspensions were filtered through a 70- µm strainer to remove debris and centrifuged at 350 × g for 5 minutes at 4°C. The resulting cell pellets were resuspended in PBS and subjected to flow cytometric staining.
Surface Staining and Flow Cytometric Analysis
Fc receptors were blocked using TruStain FcX (anti-mouse CD16/32, 2 µL/100 µL suspension; BioLegend, San Diego, CA, USA) on ice for 10 minutes. Cells were then aliquoted (10⁶ cells/50 µL per tube) and incubated with anti–CD11b-PE, anti–CD64-APC, anti–Ly6C-APC, and anti–CX3CR1-PC7 antibodies (BioLegend, San Diego, CA, USA) for 20 minutes in the dark on ice. FCM acquisition was performed using a BD LSRFortessa cytometer, and data were analyzed with FlowJo v10 software (FlowJo LLC, Ashland, OR, USA). The proportion of CD64⁺ cells within this population was quantified and statistically compared between the AMD and control groups using an unpaired two-tailed Student's t-test, and P < 0.05 was considered significant.
Results
CD64⁺CD14⁻CD16⁻ Monocytes Exhibit a Causal Association With AMD Risk
The overall study design is illustrated in Figure 1, comprising three major components: MR analysis, single-cell transcriptomic profiling, and FCM validation in a mouse model.
Figure 1.
Overview of the study workflow. The study consisted of three major components: (1) MR analysis of immune traits and AMD risk, (2) single-cell transcriptomic analysis of human RPE/choroid to examine CD64⁺ monocytes, and (3) in vivo validation in a sodium iodate-induced AMD mouse model.
To validate the reliability of the AMD phenotype in the FinnGen R11 data set, we compared its top genome-wide significant loci with canonical AMD risk loci reported by the International AMD Genomics Consortium and other large-scale studies.12,13 We found that the FinnGen data set showed strong signals at well-established loci such as ARMS2/HTRA1 (chr10q26), CFH (chr1q31), C3 (chr19), and APOE (chr19), supporting the genetic consistency and suitability of this data set for AMD-related Mendelian randomization analysis.
We first performed a comprehensive MR analysis to assess the potential causal relationship between 731 immune cell traits and AMD risk, using the IVW method as the primary analytical strategy. Prior to multiple testing correction, 13 immune cell subtypes showed a positive association with AMD (odds ratio [OR] > 1, P < 0.05), while 16 subtypes showed a negative association (OR < 1, P < 0.05) (Fig. 2A).
Figure 2.
MR analysis identifies a significant causal relationship between CD64⁺CD14⁻CD16⁻ monocytes and AMD. (A) Forest plot showing immune cell traits with nominal associations (P < 0.05) with AMD. (B) Scatter plot and funnel plot demonstrate the consistency and absence of directional bias in the MR analysis for CD64⁺CD14⁻CD16⁻ monocytes.
After FDR correction, only CD64⁺CD14⁻CD16⁻ monocytes remained significantly associated with AMD risk (OR, 1.179; 95% confidence interval, 1.095–1.269; P_FDR < 0.001). The scatter and funnel plots confirmed the robustness of the results (Fig. 2B). The F-statistic for the instrumental variables was 18.98, indicating strong instrument strength. Cochran's Q test revealed no significant heterogeneity across instrumental variables. MR-Egger regression and MR-PRESSO analysis did not detect evidence of horizontal pleiotropy or outlier bias (Supplementary Table S2). Reverse MR analysis showed no evidence for a causal effect of AMD on immune cell traits (all P > 0.05).
CD64⁺CD14⁻CD16⁻ Monocytes Are Enriched in AMD and Exhibit Functional Dynamics and Interaction With RPE Cells
To explore the functional characteristics of CD64⁺CD14⁻CD16⁻ monocytes in AMD, we performed an in-depth analysis of single-cell RNA sequencing data (GSE230348) from the human RPE/choroid. Using Seurat clustering and canonical marker annotation, we identified four immune cell clusters and eight nonimmune cell clusters (Fig. 3A). Compared to healthy controls, the AMD group, particularly the macular region, showed a markedly increased proportion of immune cells, indicating enhanced local immune activation (Fig. 3B).
Figure 3.
CD14⁻CD16⁻ monocytes are enriched in AMD choroid and show distinct FCGR1A (CD64) expression. (A) Annotation of eight nonimmune cell clusters in human RPE/choroid based on canonical markers. (B) Annotation of four immune cell clusters, including monocytes, dendritic cells, T/NK cells, and B cells. (C) Subtype annotation of monocytes into CD14⁺CD16⁻, CD14⁻CD16⁺, CD14⁺CD16⁺, and CD14⁻CD16⁻ based on marker expression. (D) Relative proportion of monocyte subtypes in each group (AMD-Central, AMD-Peripheral, Healthy-Central, Healthy-Peripheral) visualized by a stacked bar plot. (E) t-Distributed stochastic neighbor embedding plot showing spatial expression of FCGR1A (CD64) in monocytes. (F) Expression levels of FCGR1A across monocyte subsets, visualized by feature plots.
Analysis of monocyte subtype distributions revealed a pronounced enrichment of the CD14⁻CD16⁻ subtype in the AMD-central group (Figs. 3C, 3D). Statistical comparison across individual donors using the Mann–Whitney U test confirmed that this increase was significant (P < 0.05), supporting a subtype-specific role for CD14⁻CD16⁻ monocytes in AMD pathogenesis (Supplementary Fig. S1). Notably, FCGR1A (CD64) expression was spatially aggregated within the CD14⁻CD16⁻ monocyte cluster in AMD samples (Figs. 3E, 3F), further supporting its subtype specificity and functional relevance.
Pseudotime trajectory reconstruction using Monocle2 demonstrated distinct state transitions between AMD and control groups. In controls, monocyte subtypes exhibited compact and stable distributions along a single trajectory branch (Fig. 4A). In contrast, the AMD group displayed a more heterogeneous and branched trajectory architecture, with CD14⁻CD16⁻ monocytes predominantly located at the root of the trajectory, suggesting an early activation state and potential for functional plasticity (Fig. 4B).
Figure 4.
CD14⁻CD16⁻ monocytes in AMD exhibit early pseudotime positioning and divergent activation trajectory. (A) Pseudotime trajectory of monocytes from control samples. (B) Pseudotime trajectory of monocytes from AMD samples. Cell states are visualized along developmental paths inferred using single-cell trajectory analysis; coloring indicates pseudotime progression and subset identity based on CD14/CD16 expression.
To further clarify the cellular identity of the CD64⁺CD14⁻CD16⁻ myeloid population observed in AMD tissues, we investigated whether this cluster might represent retinal microglia. We examined the expression of six microglia-associated genes, including four canonical microglial markers (TREM2, CX3CR1, TMEM119, and P2RY12) and two immune-related reference genes (HLA-DRA and AIF1), across CD14/CD16-defined monocyte subsets. The CD14⁻CD16⁻ population exhibited minimal or undetectable expression of these markers, with levels significantly lower than those observed in microglia-enriched clusters (Supplementary Fig. S2). These results suggest that the CD64⁺CD14⁻CD16⁻ subset does not correspond to tissue-resident microglia but rather constitutes a distinct, non–central nervous system–derived, inflammation-associated monocyte population.
To investigate potential intercellular communication between CD14⁻CD16⁻ monocytes and RPE cells, we employed CellChat to model ligand–receptor interactions based on single-cell transcriptomic data. The top three signaling pathways identified were macrophage migration inhibitory factor (MIF)–CD74–CXCR4, insulin-like growth factor 1 (IGF1)–IGF1 receptor (IGF1R), and semaphorin-3C (SEMA3C)–plexin D1 (PLXND1) (Fig. 5A). CD14⁻CD16⁻ monocytes demonstrated relatively high signaling probabilities with RPE cells in all three pathways (Figs. 5B–D), indicating a multifaceted role in modulating the local immune microenvironment.
Figure 5.
CellChat analysis reveals intercellular communication between CD14⁻CD16⁻ monocytes and RPE cells in AMD. (A) Bubble plot showing predicted signaling interactions between annotated monocyte subsets and RPE cells based on CellChat analysis. (B–D) Heatmaps of ligand–receptor interactions for three selected signaling pathways: (B) MIF–CD74–CXCR4 axis, (C) IGF1–IGF1R axis, and (D) SEMA3C–PLXND1 axis. Interaction strength and participating genes are indicated by color intensity and dot size.
Collectively, the single-cell data reveal the enrichment, dynamic progression, and signaling engagement of CD64⁺CD14⁻CD16⁻ monocytes in AMD, reinforcing their functional significance as a key immune effector population.
Flow Cytometry Validates the Enrichment of CD64⁺ Monocytes in an AMD Mouse Model
To experimentally validate the enrichment of CD64⁺ cells in AMD, we utilized a sodium iodate–induced dry AMD mouse model and performed FCM analysis of retinal single-cell suspensions. Recognizing that murine monocyte classification is typically defined by Ly6C and CX3CR1 expression patterns, we employed multicolor flow cytometry with CD11b, Ly6C, and CX3CR1 to characterize monocyte subsets.14,15
The gating strategy is shown in Figures 6A–C, and total cells were first gated based on forward and side scatter, followed by exclusion of doublets cells. CD11b⁺ myeloid cells were then identified for downstream subset analysis. Monocyte subsets were then classified based on Ly6C and CX3CR1 expression: classical monocytes (Ly6C⁺CX3CR1⁻) and nonclassical monocytes (Ly6C⁻CX3CR1⁺). The proportion of classical monocytes was significantly increased in the model group compared to controls (Figs. 6D–F; P < 0.0001). Furthermore, CD64 expression within the Ly6C⁺ population was markedly upregulated in the AMD group (Figs. 6G–I; P < 0.0001), suggesting an expansion of proinflammatory CD64⁺ monocytes under degenerative retinal stress.
Figure 6.
Flow cytometric analysis reveals increased CD64⁺ inflammatory monocytes in AMD model mice. (A–B) Gating strategy for singlet and viable cells and exclusion of aggregates and debris. (C) Identification of CD11b⁺ myeloid cells. (D–E) Representative plots showing Ly6C and CX3CR1 expression in CD11b⁺ cells from control (PBS-injected, D) and NaIO₃-treated (E) mouse retinas. (F) Proportions of classical (Ly6C⁺CX3CR1⁻) and nonclassical (Ly6C⁻CX3CR1⁺) monocyte subsets in control and AMD groups. (G–H) Representative plots of CD64 expression within Ly6C⁺ classical monocytes from control (G) and AMD (H) retinas. (I) Proportion of CD64⁺ cells among Ly6C⁺ monocytes. Each group included three mice; three technical replicates per mouse were analyzed. Data are shown as mean ± SEM. Statistical comparisons were performed using an unpaired two-tailed Student's t-test (****P < 0.0001).
Discussion
AMD is a complex degenerative retinal disorder characterized by chronic inflammation, RPE damage, and remodeling of the local immune microenvironment. Recent evidence highlights the critical role of innate immunity, particularly the mononuclear phagocyte system, in AMD pathogenesis. In this study, we employed a multimodal strategy incorporating MR, single-cell transcriptomics, and in vivo FCM to systematically characterize the involvement of CD64⁺ monocytes, especially the CD14⁻CD16⁻ subset in the immunopathological landscape of AMD.
MR analysis revealed a significant positive causal association between CD64 expression on CD14⁻CD16⁻ monocytes and AMD risk, providing genetic support for a potential causal contribution of this immune subset. Reverse MR analysis showed no significant feedback effect of AMD on CD64 expression, further strengthening the unidirectional inference.
Single-cell transcriptomic analysis of human RPE/choroid tissues demonstrated a marked enrichment of CD14⁻CD16⁻ monocytes in AMD samples, with CD64 expression specifically concentrated within this subset. Pseudotime trajectory analysis provided insights into the dynamic functional states of these cells. In the control group, monocyte subtypes exhibited a compact and linear trajectory, suggesting a quiescent or homeostatic state. In contrast, monocytes from AMD samples displayed a highly branched and heterogeneous trajectory. CD14⁻CD16⁻ monocytes were positioned predominantly at the root of this trajectory, indicating an early activation state with strong potential for differentiation. Notably, previous studies have shown that CD64 is highly expressed on M1-type macrophages and enhances phagocytic and antigen-presenting functions.16 Our findings support the hypothesis that CD64⁺CD14⁻CD16⁻ monocytes may serve as an inflammation-primed subset with the potential to differentiate into activated effector cells within the retinal microenvironment.
Building upon the spatial enrichment of CD64⁺CD14⁻CD16⁻ monocytes, CellChat analysis identified three prominent ligand–receptor signaling axes mediating their interaction with RPE cells: MIF–CD74–CXCR4, IGF1–IGF1R, and SEMA3C–PLXND1. Recent studies have shown that CD64 can activate the nuclear factor kappa B (NF-κB) signaling pathway, thereby promoting the assembly of the NLRP3 inflammasome and the subsequent release of proinflammatory cytokines, including IL-1β and IL-18.17 MIF is a classical pro-inflammatory cytokine that activates downstream NF-κB and extracellular signal–regulated kinase signaling cascades via its receptor CD74, thereby promoting chemokine production and immune cell recruitment.18 Its blockade has been shown to mitigate tissue injury across multiple disease models,19,20 implicating it as a possible contributor to immune dysregulation in AMD. CD64 may synergistically enhance this effect, contributing to a self-perpetuating inflammatory amplification loop in AMD. The IGF1–IGF1R axis was found to be selectively enriched in CD14⁻CD16⁻ monocytes, suggesting subtype-specific engagement with RPE cells. As IGF1 signaling is known to regulate oxidative stress resilience and metabolic homeostasis in RPE, as well as shares downstream convergence with NF-κB activation, it raises the possibility that CD64⁺ monocytes mediate immunometabolic coupling in AMD.21 In addition, the SEMA3C–PLXND1 axis, typically involved in maintaining epithelial integrity and cell adhesion, was also notably active. Although Sema3C has been shown to exert protective and antiangiogenic effects in the retina, its functional integrity may be impaired under inflammatory conditions, particularly in the presence of elevated IL-1β. This dysregulation can disrupt the RPE barrier and exacerbate microenvironmental instability.22 These signaling interactions collectively suggest that CD64⁺ monocytes not only accumulate in AMD tissues but actively participate in immune–structural crosstalk with RPE cells. This may contribute to a self-sustaining cycle of inflammation, tissue remodeling, and immune dysregulation in AMD.
The CD64⁺CD14⁻CD16⁻ monocyte population identified in this study does not conform to the classical CD14/CD16-based monocyte classification system, which was originally developed for peripheral blood.23–25 According to the classification proposed by Ziegler-Heitbrock et al. in 2010,23 circulating monocytes can be divided into three subsets based on CD14 and CD16 expression: classical (CD14⁺⁺CD16⁻), intermediate (CD14⁺⁺CD16⁺), and nonclassical (CD14⁺CD16⁺⁺). However, previous studies have highlighted limitations of the CD14/CD16-based system, particularly under inflammatory conditions or within tissue environments, where the expression of CD14 and CD16 can be rapidly altered by stimuli such as lipopolysaccharide (LPS), leading to unreliable subtype discrimination.26
Emerging evidence suggests that monocyte differentiation follows a dynamic continuum, and intermediate states such as CD14⁻CD16⁻ may represent transient or reprogrammed phenotypes rather than residual populations. In local tissue environments, monocyte–macrophage lineages often undergo region-specific adaptation. In the retina, direct interactions between RPE cells and immune cells can drive myeloid cells toward a highly activated, tissue-adapted phenotype. Our data support this interpretation: CD64⁺CD14⁻CD16⁻ monocytes were markedly enriched in AMD tissues, exhibited distinct transcriptional signatures, and engaged in extensive ligand–receptor interactions with RPE cells, suggesting active adaptation to the retinal microenvironment.
Supporting the biological relevance of this subset, Ong et al.27 proposed an alternative marker panel, including CD64, CD86, CD33, CCR2, and HLA-DR, that enables more stable classification of monocyte subsets under inflammatory conditions. These findings reinforce the validity of using CD64 as an anchor marker to define pathologically relevant monocyte subsets in AMD.
To further interrogate the pathogenic relevance of CD64⁺ monocytes in AMD, we performed additional flow cytometric analysis in the sodium iodate–induced mouse model. In mice, monocyte subsets are primarily delineated by Ly6C and CX3CR1 expression, as opposed to the CD14/CD16-based classification widely used in humans. Classical monocytes are characterized as Ly6C⁺CX3CR1⁻ and are associated with inflammatory functions, whereas nonclassical monocytes (Ly6C⁻CX3CR1⁺) exhibit patrolling and reparative roles.14,15
Our data revealed a marked expansion of classical Ly6C⁺ monocytes in AMD retinas, accompanied by a significant increase in CD64 expression within this subset. These results suggest that CD64 marks a proinflammatory and functionally reprogrammed monocyte population under conditions of retinal stress. Given CD64’s roles in antigen uptake, cytokine production, and immune effector activation, its elevated expression likely reflects enhanced immunopathological activity during RPE degeneration.
CD64 is a high-affinity receptor for the Fc region of IgG and is constitutively expressed on macrophages, monocytes, and eosinophils.9 Functionally, it mediates antibody-dependent cellular cytotoxicity, phagocytosis, receptor-mediated endocytosis, and proinflammatory responses, positioning it as a pivotal effector molecule in innate immunity.8
The immunological significance of CD64 has been well documented across multiple inflammatory conditions and myeloid malignancies. In rheumatoid arthritis, CD64 is highly expressed on M1 macrophages and serves as a key inflammatory marker.28 Anti-CD64 antibodies can suppress anticyclic citrullinated peptide antibody–induced proinflammatory cytokine production and osteoclastogenesis.29 Similarly, in adult-onset Still's disease, the expression level of CD64 on circulating monocytes has been recognized as a sensitive biomarker for monitoring disease activity.30 Beyond autoimmune diseases, in acute myeloid leukemia, CD64 serves as a stable marker of myeloid cells and has been exploited in various targeted strategies, including RNA interference systems and antibody–drug conjugates.31,32
In the context of AMD, our findings suggest that CD64 may not only reflect immune activation but also participate in immunopathogenic signaling. Its coenrichment with MIF, IGF1, and SEMA3C signaling pathways supports a role in amplifying inflammatory and structural dysregulation within the retinal microenvironment. These results highlight the potential of CD64⁺ monocytes as a tractable target for therapeutic intervention in AMD.
Despite the integrative multiomics strategy employed in this study, there are several limitations to consider. First, the sodium iodate–induced mouse model used for in vivo validation mimics certain aspects of dry AMD, particularly geographic atrophy.33 NaIO3 selectively damages RPE cells through necrotic or necroptotic pathways following systemic administration, resulting in patchy RPE loss and subsequent photoreceptor degeneration.34 It also triggers innate immune responses, including complement activation and microglial inflammation.11 These features make it suitable for studying acute retinal inflammation and immune activation. However, as an acute injury model, NaIO3 does not fully replicate the chronic progression, pathological heterogeneity, or individual variability of human AMD. Thus, the increased presence of CD64⁺ myeloid cells observed in this model should be interpreted as evidence of immune involvement rather than proof of direct pathogenicity over time. Future studies utilizing chronic or genetically predisposed AMD models (such as APOE4 knock-in or CFH-deficient mice) will be necessary to assess long-term contributions of this population. Second, the functional characteristics of CD64⁺ monocytes remain incompletely understood. Future investigations integrating single-cell multiomics and in vitro functional assays will be essential to further elucidate their roles in immune regulation and disease progression. Lastly, we did not adopt the five-marker gating panel proposed by Ong et al.27 for a more refined immune lineage classification. This decision was primarily due to the limited number of immune cells obtainable from ocular tissues and the technical constraints of our flow cytometry platform. Future studies incorporating higher-dimensional cytometry (e.g., spectral flow or cytometry by time of flight) and spatial transcriptomics may allow for a more systematic dissection of the phenotypic heterogeneity within the CD64⁺ monocyte population.
In summary, our study identifies CD64⁺CD14⁻CD16⁻ monocytes as a functionally distinct and immunologically engaged subset enriched in AMD retinal tissue. Through integrative analyses spanning genetic association, single-cell profiling, and animal model validation, we provide converging evidence for their potential role in mediating local inflammation and RPE immune crosstalk. Beyond serving as a sensitive marker of myeloid activation, CD64 may represent a modulatory node linking immune signaling and tissue remodeling pathways. These findings position CD64⁺ monocytes as promising candidates for future biomarker development and immunomodulatory interventions. Potential strategies, such as anti-CD64 antibodies, epigenetic or metabolic reprogramming, and engineered immune cell therapies, may enable precision-targeted approaches to disrupt chronic inflammation in AMD.
Supplementary Material
Acknowledgments
Supported by the Shenzhen Science and Technology Program (No. JCYJ20240813104223030 & No. JCYJ20220818102603007) and the Shenzhen Medical Research Fund (No. A2402010).
Disclosure: C. Li, None; L. Zhou, None; T. Luo, None; H. Sun, None; R. Ma, None; J. Wang, None; M.-M. Yang, None
References
- 1. Wong WL, Su X, Li X, et al.. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Global Health. 2014; 2(2): e106–e116. [DOI] [PubMed] [Google Scholar]
- 2. Fleckenstein M, Schmitz-Valckenberg S, Chakravarthy U.. Age-related macular degeneration: a review. JAMA. 2024; 331(2): 147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Wang Z, Zhang Y, Xu C, Peng A, Qin H, Yao K.. Advancements in age-related macular degeneration treatment: from traditional anti-VEGF to emerging therapies in gene, stem cell, and nanotechnology. Biochem Pharmacol. 2025; 236: 116902. [DOI] [PubMed] [Google Scholar]
- 4. Zhao N, Hao XN, Huang JM, Song ZM, Tao Y. Crosstalk between microglia and Müller glia in the age-related macular degeneration: role and therapeutic value of neuroinflammation. Aging Dis. 2024; 15(3): 1132–1154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Zhao Q, Lai K.. Role of immune inflammation regulated by macrophage in the pathogenesis of age-related macular degeneration. Exp Eye Res. 2024; 239: 109770. [DOI] [PubMed] [Google Scholar]
- 6. Li C, Zhou L, Sun H, Yang M.. Age-related macular degeneration: a disease of cellular senescence and dysregulated immune homeostasis. CIA. 2024; 19: 939–951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Roubeix C, Nous C, Augustin S, et al.. Splenic monocytes drive pathogenic subretinal inflammation in age-related macular degeneration. J Neuroinflammation. 2024; 21(1): 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Akinrinmade OA, Chetty S, Daramola AK, Islam M-U, Thepen T, Barth S. CD64: an attractive immunotherapeutic target for M1-type macrophage mediated chronic inflammatory diseases. Biomedicines. 2017; 5(3): 56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Guilliams M, Bruhns P, Saeys Y, Hammad H, Lambrecht BN.. The function of Fcγ receptors in dendritic cells and macrophages. Nat Rev Immunol. 2014; 14(2): 94–108. [DOI] [PubMed] [Google Scholar]
- 10. Menon M, Mohammadi S, Davila-Velderrain J, et al.. Single-cell transcriptomic atlas of the human retina identifies cell types associated with age-related macular degeneration. Nat Commun. 2019; 10(1): 4902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Enzbrenner A, Zulliger R, Biber J, et al.. Sodium iodate-induced degeneration results in local complement changes and inflammatory processes in murine retina. Int J Mol Sci. 2021; 22(17): 9218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Fritsche LG, Igl W, Bailey JN, et al.. A large genome-wide association study of age-related macular degeneration highlights contributions of rare and common variants. Nat Genet. 2016; 48(2): 134–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Gorman BR, Voloudakis G, Igo RP Jr, et al.. Genome-wide association analyses identify distinct genetic architectures for age-related macular degeneration across ancestries. Nat Genet. 2024; 56(12): 2659–2671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fang P, Li X, Dai J, et al.. Immune cell subset differentiation and tissue inflammation. J Hematol Oncol. 2018; 11(1): 97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Yang J, Zhang L, Yu C, et al.. Monocyte and macrophage differentiation: circulation inflammatory monocyte as biomarker for inflammatory diseases. Biomark Res. 2014; 2(1): 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Lei Q, Li L, Huang W, Qin B, Zhang S.. HEV ORF3 downregulates CD14 and CD64 to impair macrophages phagocytosis through inhibiting JAK/STAT pathway. J Med Virol. 2019; 91(6): 1112–1119. [DOI] [PubMed] [Google Scholar]
- 17. Zhang H, Li L, Liu L.. FcγRI (CD64) contributes to the severity of immune inflammation through regulating NF-κB/NLRP3 inflammasome pathway. Life Sci. 2018; 207: 296–303. [DOI] [PubMed] [Google Scholar]
- 18. Osipyan A, Chen D, Dekker FJ.. Epigenetic regulation in macrophage migration inhibitory factor (MIF)-mediated signaling in cancer and inflammation. Drug Discov Today. 2021; 26(7): 1728–1734. [DOI] [PubMed] [Google Scholar]
- 19. Tanese K, Ogata D.. The role of macrophage migration inhibitory factor family and CD74 in the pathogenesis of melanoma. Exp Dermatol. 2024; 33(6): e15122. [DOI] [PubMed] [Google Scholar]
- 20. Kong YZ, Chen Q, Lan HY.. Macrophage migration inhibitory factor (MIF) as a stress molecule in renal inflammation. Int J Mol Sci. 2022; 23(9): 4908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Ravera S, Puddu A, Bertola N, et al.. IGF-1 signaling modulates oxidative metabolism and stress resistance in ARPE-19 cells through PKM2 function. Int J Mol Sci. 2024; 25(24): 13640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Yang W, Hu J, Uemura A, Tetzlaff F, Augustin HG, Fischer A.. Semaphorin-3C signals through Neuropilin-1 and PlexinD1 receptors to inhibit pathological angiogenesis. EMBO Mol Med. 2015; 7(10): 1267–1284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Ziegler-Heitbrock L, Ancuta P, Crowe S, et al.. Nomenclature of monocytes and dendritic cells in blood. Blood. 2010; 116(16): e74–e80. [DOI] [PubMed] [Google Scholar]
- 24. Ożańska A, Szymczak D, Rybka J.. Pattern of human monocyte subpopulations in health and disease. Scand J Immunol. 2020; 92(1): e12883. [DOI] [PubMed] [Google Scholar]
- 25. Wong KL, Yeap WH, Tai JJ, et al.. The three human monocyte subsets: implications for health and disease. Immunol Res. 2012; 53(1–3): 41–57. [DOI] [PubMed] [Google Scholar]
- 26. Ahrazoglu T, Kluczny JI, Kleimann P, et al.. Design of a robust flow cytometric approach for phenotypical and functional analysis of human monocyte subsets in health and disease. Biomolecules. 2024; 14(10): 1251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ong SM, Teng K, Newell E, et al.. A novel, five-marker alternative to CD16-CD14 gating to identify the three human monocyte subsets. Front Immunol. 2019; 10: 1761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Theeuwes WF, Di Ceglie I, Dorst DN, et al.. CD64 as novel molecular imaging marker for the characterization of synovitis in rheumatoid arthritis. Arthritis Res Ther. 2023; 25(1): 158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Min HK, Lee JY, Lee SH, Ju JH, Kim HR.. Suppressing anti-citrullinated protein antibody-induced osteoclastogenesis in rheumatoid arthritis using anti-CD64 and PAD-2 inhibitors. Clin Exp Rheumatol. 2025; 43(1): 79–86. [DOI] [PubMed] [Google Scholar]
- 30. Shimizu T, Kikuchi-Taura A, Tsuji S, Matsushita M, Ohshima S, Saeki Y.. Up-regulation of CD64 expression on monocytes in patients with active adult-onset Still disease: a possible biomarker of disease activity. J Clin Rheumatol. 2020; 26(2): 67–72. [DOI] [PubMed] [Google Scholar]
- 31. Yong SB, Chung JY, Kim SS, Choi HS, Kim YH.. CD64-targeted HO-1 RNA interference enhances chemosensitivity in orthotopic model of acute myeloid leukemia and patient-derived bone marrow cells. Biomaterials. 2020; 230: 119651. [DOI] [PubMed] [Google Scholar]
- 32. Akinrinmade OA, Fajemisin EA, Daramola AK, et al.. Anti-CD64(scFv)-SNAP-Auristatin F: an in vitro proof of concept study for a recombinant antibody conjugate targeting CD64+ acute monocytic leukemia. Eur J Med Chem. 2025; 290: 117520. [DOI] [PubMed] [Google Scholar]
- 33. Machalińska A, Lubiński W, Kłos P, et al.. Sodium iodate selectively injuries the posterior pole of the retina in a dose-dependent manner: morphological and electrophysiological study. Neurochem Res. 2010; 35(11): 1819–1827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Upadhyay M, Bonilha VL.. Regulated cell death pathways in the sodium iodate model: insights and implications for AMD. Exp Eye Res. 2024; 238: 109728. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.