Abstract

   Neovascular age‐related macular degeneration (NVAMD) is a common
   retinal disease causing vision loss in the elderly. Neuroinflammation
   significantly contributes to NVAMD's etiology. This study explores the
   role of Programmed cell death ligand 1 (PD‐L1), an immune checkpoint
   (ICP) in microglia, known for limiting neuroinflammation in
   neurodegenerative diseases, and its potential function in NVAMD. This
   work finds increased PD‐L1 expression in retinal microglia following
   laser injury. PD‐L1 knockout (KO) or inhibitory PD‐L1 antibody
   treatment worsens vascular leakage and neoangiogenesis in a
   laser‐induced NVAMD mouse model, effects reversible by microglia
   depletion with PLX5622. This study underscores that choroidal
   neovascularization (CNV) may be regulated by multiple mechanisms, with
   PD‐L1 modulation representing one of these pathways. Blocking PD‐L1
   elevated proinflammatory factors and p‐ERK levels, indicating
   microglial overactivation in NVAMD. Conversely, enhancing PD‐L1
   signaling reduced neuroinflammation and neovascularization via ERK.
   These findings highlight PD‐L1's role in neoangiogenesis and
   neuroinflammation in NVAMD, suggesting its potential as a target for
   immunomodulatory treatment in NVAMD.

   Keywords: microglia/macrophage; neovascular age‐related macular
   degeneration; neuroinflammation, PD‐L1
     __________________________________________________________________

   PD‐L1 limits neuroinflammation in neovascular AMD by modulating
   microglial activation. Its deficiency exacerbates vascular leakage and
   choroidal neovascularization (CNV) via ERK signaling, while enhancing
   PD‐L1 reduces inflammation and neovascularization. Targeting PD‐L1 may
   offer a novel immunomodulatory strategy for NVAMD.

   graphic file with name ADVS-12-2400747-g009.jpg

1. Introduction

   Age‐related macular degeneration (AMD) is a chronic, progressive ocular
   disorder primarily affecting the central vision, making it a leading
   cause of irreversible blindness in the elderly population worldwide.^[
   [38]^1 , [39]^2 , [40]^3 ^] It was estimated that by 2020,
   approximately 196 million people worldwide were affected by AMD, and
   this number is projected to reach 288 million by 2040.^[ [41]^1 ,
   [42]^4 , [43]^5 ^] AMD predominantly affects individuals over the age
   of 50 and significantly impairs activities such as reading, driving,
   and facial recognition, thus diminishing patients' quality of life.^[
   [44]^6 ^] Furthermore, AMD increases the risk of falls, depression, and
   loss of independence, making it a growing public health concern.^[
   [45]^7 , [46]^8 , [47]^9 ^]

   Currently, anti‐vascular endothelial growth factor (anti‐VEGF) therapy
   is the primary therapeutic approach for the neovascular age‐related
   macular degeneration (NVAMD).^[ [48]^3 , [49]^10 ^] Nevertheless,
   anti‐VEGF therapy has some limitations, such as the economic burden,
   complications arising from frequent intravitreal injections and the
   treatment resistance.^[ [50]^11 ^] Significantly, approximately
   one‐third of patients fail to benefit from anti‐VEGF therapy owing to
   the macular fibrosis or atrophy.^[7] Furthermore, prolonged
   administration of anti‐VEGF agents may potentially result in neuronal
   damage.^[ [51]^12 ^] Given the limitation of anti‐VEGF therapies, there
   is a pressing need to explore the pathogenesis of NVAMD and the novel
   alternative therapies.

   Genetic aberrations associated with innate immunity have been
   empirically established as contributory factors in increasing
   susceptibility to NVAMD.^[ [52]^13 , [53]^14 , [54]^15 ^] Studies from
   both NVAMD patients and animal models further provide evidence of
   perturbed innate immune homeostasis.^[ [55]^16 , [56]^17 , [57]^18 ^]
   Retinal microglia, the resident immune cells, regulate the tissue
   integrity in NVAMD.^[ [58]^19 ^] During the early stage of NVAMD,
   degeneration of the retinal pigment epithelium (RPE) and disruption of
   Bruch's membrane trigger the proliferation and migration of microglia
   to the injury site, which will facilitate tissue repair by releasing
   neuromodulators.^[ [59]^20 ^] However, neuroprotective role of the
   activated microglia is transient, and hyperactive microglia may leading
   to vascular leakage, which may contribute to choroidal
   neovascularization (CNV) progression.^[ [60]^21 , [61]^22 ^] Targeting
   retinal immune homeostasis and modulating microglial reactivity may be
   a potential strategy to mitigate and alleviate the progression of
   NVAMD.

   Immune checkpoints (ICPs) play a pivotal role in regulating immune
   homeostasis, particularly in the neuroimmune and neuroinflammatory
   responses.^[ [62]^23 ^] While the bulk of research efforts have been
   dedicated to elucidate the functional intricacies of ICP molecules in
   cancer and peripheral immunity,^[ [63]^24 , [64]^25 , [65]^26 , [66]^27
   , [67]^28 ^] some ICPs are also expressed in the central nervous system
   (CNS) cells, encompassing neurons,^[ [68]^29 , [69]^30 ^] astrocytes,^[
   [70]^23 , [71]^31 ^] and microglia.^[ [72]^32 , [73]^33 , [74]^34 ^]
   ICPs are linked with the maintenance of CNS immune homeostasis, as well
   as the progression of neuroinflammatory and neurodegenerative
   diseases.^[ [75]^33 , [76]^35 ^] One of the prominent ICPs expressed in
   CNS is Programmed Cell Death Protein 1 (PD‐1),^[ [77]^36 , [78]^37 ^]
   accompanied by its ligands Programmed Cell Death lig 1 (PD‐L1)^[
   [79]^38 ^] and PD‐L2.^[ [80]^39 ^] PD‐L2 is predominantly expressed in
   antigen‐presenting cells, whereas PD‐L1 has a broad expression.^[
   [81]^39 ^] Clinical investigations reported the elevation of PD‐1 level
   in NVAMD patients^[ [82]^40 ^] and after block PD‐1/PD‐L1, NVAMD
   patients undergo unresponsive to anti‐VEGF treatments,^[ [83]^41 ^]
   which suggests the potential alterations of PD‐1/PD‐L1 pathway during
   the pathogenesis of NVAMD.

   To investigate the role of retinal PD‐L1 in the progression of NVAMD,
   we established a laser‐induced CNV mouse model.^[ [84]^42 ^] Our
   results demonstrated a significant upregulation of PD‐L1 expression in
   retinal microglia within this model. PD‐L1 blockade led to a pronounced
   exacerbation of retinal neuroinflammation and pathological
   angiogenesis, highlighting its regulatory role in immune modulation
   through intrinsic microglial signaling pathways. These findings provide
   critical insights into the potential therapeutic application of ICP
   regulators for improving NVAMD treatment.

2. Results

2.1. Retinal PD‐L1 Expression Increased in Laser‐Induced CNV

   In the CNS, neuroinflammation upregulates PD‐L1 (encoded by
   the Cd274 gene) to maintain immune homeostasis.^[ [85]^43 ^] However,
   it remains unclear whether retinal PD‐L1 expression is altered during
   CNV formation. To investigate this, we measured retinal PD‐L1 mRNA and
   protein levels in mice following laser injury. Compared to controls,
   PD‐L1 mRNA level increased approximately 22‐fold at 1‐day post‐laser
   photocoagulation, peaking at 3 days and gradually declining, yet
   remaining significantly elevated (Figure [86]1A). Correspondingly,
   retinal PD‐L1 protein expression also rose as early as 3 h post‐laser
   treatment and persisted for over 14 days (Figure [87]1B). Since PD‐L1
   is also expressed in RPE,^[ [88]^44 , [89]^45 ^] we investigated the
   change of PD‐L1 expression within the RPE‐choroid tissue. Our results
   showed no significant changes in PD‐L1 mRNA (Figure [90]1C) and protein
   (Figure [91]1D) after laser photocoagulation. These findings indicate
   that PD‐L1 expression in the retina is markedly elevated during
   laser‐induced CNV formation.

Figure 1.

   Figure 1
   [92]Open in a new tab

   PD‐L1 expression in retina and retinal pigment epithelium (RPE) choroid
   in the laser‐induced choroidal neovascularization (CNV) mouse model. A)
   Statistical result of PD‐L1mRNA expression change in mouse retina at
   various time points: Control, 3 h post‐laser (3 h), 1‐, 3‐, 7‐, and
   14‐days post‐laser (1, 3, 7, and 14 days). B) Western blot example
   images (top) and statistical result (bottom) of PD‐L1 protein level in
   the mouse retina at different time points. C) Same as (A), but for the
   RPE‐choroid tissue. D) Same as (B), but for the RPE‐choroid tissue (n =
   4 mice for each group). Data are presented as mean ± SEM. ^* p < 0.05,
   ^** p < 0.01, ^*** p < 0.001, compared to the Control group. One‐way
   ANOVA with Tukey's multiple comparisons post‐hoc test.

2.2. PD‐L1 Knockout or Blockade Increase Vascular Permeability and
Neovascularization in CNV

   Given the significant increase of retinal PD‐L1 level following the
   laser injury, we explored the correlation between PD‐L1 expression and
   the initiation of CNV by changing PD‐L1. Firstly, we used a PD‐L1
   knockout (KO) mouse line, and the successful deletion of PD‐L1 was
   confirmed by genomic PCR of mouse tail DNA samples and WB analyses of
   retina‐RPE‐choroidal lysates (Figure [93]S1A,B, Supporting
   Information).

   The formation of CNV in mice typically peaks on day 7 after laser
   injury.^[ [94]^42 , [95]^46 ^] Therefore, we selected this time point
   to assess the severity of CNV in each group of mice, primarily
   evaluating two parameters: vascular permeability and the extent of
   neovascularization.^[ [96]^46 ^] In this study, we used fluorescein
   fundus angiography (FFA) and optical coherence tomography (OCT) to
   observe vascular leakage and subretinal fluid accumulation, which were
   used to assess vascular permeability.^[ [97]^47 ^] We quantified CNV
   thickness using OCT and determined its surface area through IB4
   staining of RPE‐choroid flat mounts to comprehensively evaluate the
   extent of CNV.^[ [98]^42 ^]

   The experimental results showed that, on day 7 after laser modeling,
   PD‐L1 KO mice exhibited more extensive and severe vascular leakage
   during both early and late phases of FFA, with leakage scores
   significantly higher than those in WT mice (Figure [99]2A,B).
   Consistent with the FFA results, OCT detection revealed that the
   proportion of subretinal fluid at the laser injury sites in PD‐L1 KO
   mice was significantly higher than in WT mice (38.5% versus 21.8%,
   respectively) (Figure [100]2C). Further OCT evaluation of CNV thickness
   showed that PD‐L1 KO mice had significantly thicker CNV compared to WT
   mice (Figure [101]2D). The IB4 staining of RPE‐choroid flat mounts also
   confirmed that the CNV surface area in PD‐L1 KO mice was markedly
   larger than in WT mice (Figure [102]2C,F). These results suggest that
   PD‐L1 gene KO leads to increased vascular permeability and
   neovascularization in mice after laser injury, resulting in more severe
   CNV.

Figure 2.

   Figure 2
   [103]Open in a new tab

   PD‐L1 knockout (KO) or blockade increase vascular permeability and
   neovascularization in choroidal neovascularization (CNV). A) Left
   panels display the fundus images of WT and KO mice at 7 days post‐laser
   injury. Middle and right panels present fluorescein fundus angiography
   (FFA) images of WT and KO mice at two distinct time points (early and
   late phases) at 7 days after laser injury, respectively. B) Summaries
   of the FFA grade scores of each laser spot (n = 42 and 37 laser spots
   from WT and KO mice, respectively). C) Left panels show the examples of
   optical coherence tomography (OCT) scan, and right panels show the
   laser‐induced CNV stained with IB4 in retinal pigment epithelium
   (RPE)/choroidal flat mounts at 7 d after laser injury from WT and KO
   mice. D) Quantification of the laser‐induced CNV thickness and E)
   percentages of subretinal fluids after laser injury (n = 20 and 27
   laser spots from WT and KO mice, respectively). F) Quantification of
   the laser‐induced CNV area in RPE/choroidal flat mounts (n = 7 WT mice
   and 10 KO mice). G) Left panels show the examples of OCT scan, middle
   panels show FFA images and right panels show images of IB4 staining in
   RPE/choroidal flat mounts at 7 days after laser injury with anti‐PD‐L1
   antibody or vehicle intravitreal injections. H–J) Quantifications of
   the laser‐induced CNV thickness (J), percentages of subretinal fluids
   (K), and laser‐induced CNV area (L) (n = 4 mice per group). Data are
   presented as mean ± SEM. Unpaired t‐test for CNV thickness and CNV area
   in (D,F,H,J). ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001 compared to the
   WT‐CNV group or vehicle group. Mann–Whitney test for average FFA grade
   scores in (I), ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001 compared to
   the vehicle group. Chi‐square test for FFA grade scores in (B) and
   percentages of subretinal fluids in (E). ^* p < 0.05, ^** p < 0.01,
   ^*** p < 0.001 compared to the WT‐CNV group. Source data are provided
   as a Source Data file.

   To exclude the potential interference of systemic PD‐L1 KO on ocular
   development and overall systemic effects, we injected a PD‐L1 blocking
   antibody (anti‐PD‐L1) into the vitreous of WT mice to assess the impact
   of locally blocking intraocular PD‐L1 signaling on CNV formation. The
   results showed that, similar to the PD‐L1 KO group, mice that received
   intravitreal injections of anti‐PD‐L1 exhibited significantly higher
   fluorescence leakage scores compared to the control vehicle group on
   day 7 after laser‐induced CNV formation (Figure [104]2G,I).
   Additionally, both the thickness and area of neovascularization were
   significantly increased following anti‐PD‐L1 injection
   (Figure [105]2G,H,J). These findings suggest that local blockade of
   PD‐L1 within the eye has effects similar to systemic PD‐L1 KO, leading
   to increased vascular permeability and neovascularization area,
   resulting in more severe CNV in the model mice.

2.3. Activation of PD‐L1 Ameliorates CNV in Mice

   Previous experimental results demonstrated that blocking PD‐L1
   exacerbates CNV formation in mice. This suggests that the upregulation
   of PD‐L1 observed during CNV might represent a protective
   anti‐inflammatory response, similar to its role in other nervous system
   diseases.^[ [106]^43 ^] However, whether further activation of PD‐L1
   could improve CNV pathology remains unclear. Studies have shown that
   soluble PD‐1 can bind to PD‐L1, triggering downstream signaling
   cascades.^[ [107]^48 ^] In neuroinflammatory conditions such as
   multiple sclerosis, PD‐L1 expressed on immune cells interacts with PD‐1
   to transmit inhibitory signals, playing a crucial role in maintaining
   immune homeostasis and suppressing inflammation.^[ [108]^49 ^] We
   hypothesized that exogenous PD‐1 protein, by binding to PD‐L1, might
   activate downstream PD‐L1 signaling and further inhibit CNV formation
   in the CNV mouse model.

   To test this hypothesis, we injected PD‐1 protein or an equal volume of
   control vehicle into the vitreous of WT and PD‐L1 KO mice following
   laser photocoagulation. The FFA results showed that intravitreal
   injection of PD‐1 significantly suppressed vascular leakage in WT mice
   compared to control vehicle (Figure [109]3A,C). OCT examinations and
   IB4 staining of RPE‐choroid flat mounts further demonstrated that PD‐1
   injection significantly reduced CNV thickness and area in WT mice
   (Figure [110]3A,B,D). However, in PD‐L1 KO mice, intravitreal injection
   of PD‐1 protein did not significantly reduce vascular leakage, nor did
   it decrease CNV area or thickness compared to vehicle injection
   (Figure [111]3E–H). These findings indicate that in the CNV mouse
   model, intravitreal administration of exogenous PD‐1 targets PD‐L1 and
   enhances its protective effects, thereby reducing vascular leakage and
   pathological neovascularization.

Figure 3.

   Figure 3
   [112]Open in a new tab

   Activation of PD‐L1 ameliorates choroidal neovascularization (CNV) in
   mice. A) Left panels show the examples of optical coherence tomography
   (OCT) scan, middle panels show the fundus fluorescein angiography (FFA)
   example images, and right panels show the IB4 staining in retinal
   pigment epithelium (RPE)/choroidal flat mounts in WT mice after
   intravitreal injection of PD‐1 or vehicle and 7 days after laser
   injury. B–D) Quantifications of the laser‐induced CNV thickness (B),
   percentage of subretinal fluids (C), and laser‐induced CNV area (D) (n
   = 4 mice per group). E) Same as (A), but for the PD‐L1 KO mice. F–G)
   Same as (B–D), but for the PD‐L1 KO mice (n = 4 mice per group). Data
   are presented as mean ± SEM. Unpaired t‐test for CNV thickness in (B)
   and (F), average leakage degree in (C) and (G), CNV area in (D) and
   (H). ^^ p < 0.05, ^^^ p < 0.01, and ^^^^ p < 0.001, compared to the
   control group. ^* p < 0.05, ^** p < 0.01, and ^*** p < 0.001, compared
   to the CNV + vehicle group. Source data are provided as a Source Data
   file.

2.4. PD‐L1 Deficiency Exacerbates CNV by Enhancing Microglial Activation and
Neuroinflammation

   Prior studies suggested a strong correlation between vascular
   permeability and the extent of inflammation.^[ [113]^46 , [114]^50 ^]
   The pronounced vascular leakage we observed in PD‐L1 KO mice indicates
   the elevation of retinal neuroinflammation after CNV. Inflammation in
   CNV is largely mediated by microglia and macrophages,^[ [115]^51 ^]
   which also express PD‐L1.^[ [116]^34 ^] In conditions like multiple
   sclerosis^[ [117]^49 ^] and spinal cord injury,^[ [118]^43 ^] blocking
   PD‐L1 in these cells worsens their activation and neuroinflammation.
   Our immunofluorescence (IF) and 3D reconstruction analyses further
   revealed a marked increase in IBA1+ cells, forming distinct clusters at
   the laser‐induced lesion site (Figure [119]4A). Therefore, we
   hypothesized that PD‐L1 blockade may aggravate CNV by enhancing
   microglia/macrophage‐mediated inflammation. To test this hypothesis, we
   designed an RNA sequencing (RNA‐seq) experiment on retinal tissues from
   PD‐L1 KO and WT mice at the peak of microglial activation, which
   previous studies have identified as day 3 post‐laser.^[ [120]^46 ,
   [121]^52 ^] Our results also confirmed this timing, showing substantial
   microglia accumulation at the laser sites on day 3 post‐laser (Figure
   [122]S1C,D, Supporting Information). RNA‐seq analysis revealed
   significant gene expression differences between PD‐L1 KO and WT mice.
   Specifically, 10 017 genes were upregulated and 10 610 were
   downregulated in PD‐L1 KO mice compared to WT. Notably, genes linked to
   microglial activation, such as Csf1 and Cx3cr1,^[ [123]^53 , [124]^54
   ^] were significantly upregulated in PD‐L1 KO mice (Figure [125]4B).
   Gene Ontology (GO) enrichment analyses (Figure [126]4C) revealed a
   similar enrichment pattern, with differentially expressed genes
   involved in innate immunity and inflammation pathways. Top 5 Key terms
   included “response to bacterium” “immune system process,” “innate
   immune response,” “cellular response to interferon‐beta,” and
   “extracellular space.” These RNA‐seq results suggest that PD‐L1
   deficiency may influence microglial activation, potentially driving the
   exacerbation of CNV.

Figure 4.

   Figure 4
   [127]Open in a new tab

   PD‐L1 knockout (KO) promotes inflammatory activation of microglia in
   choroidal neovascularization (CNV) mice. A) IB4 (green) and IBA1(red)
   staining and 3D reconstruction of retinal flat mounts from control
   group (Left) and CNV model mice (Right). IBA1+ microglia were clustered
   at the laser lesioned location. B) Volcano plot obtained from DESeq2
   analysis of the retina in the laser‐induced CNV mouse model, comparing
   wild‐type CNV (WT‐CNV) and KO (KO‐CNV) mice at 3 days after laser
   injury (n = 4 mice per group). C) Gene Ontology (GO) functional
   enrichment analysis of the differential expression genes (DEGs). D)
   Expression levels of CD68 mRNA in the retinas of WT and PD‐L1 KO mice
   from control and laser‐treated groups, assessed 3 days after laser
   injury. E) Representative images of IBA1+ cells at the laser lesion
   sites in the retina. F) Quantitative analysis of the IBA1+ area at the
   laser lesion sites in the retina. G–I) Corresponding analyses as
   described in (D–F), performed on retinal pigment epithelium (RPE)
   choroid tissue samples. J) Expression levels of IL‐1β, IL‐6, TNF‐α, and
   CCL‐2 mRNA in the retinas of WT and PD‐L1 KO mice from control and CNV
   groups, evaluated 3 days after laser‐induced injury. K) Corresponding
   analyses for IL‐1β, IL‐6, TNF‐α, and CCL‐2 mRNA expression in
   RPE‐choroid tissue samples. Data are presented as mean ± SEM, with n =
   4 mice per group. Statistical significance is denoted as follows: ^^ p
   < 0.05, ^^^ p < 0.01, ^^^^ p < 0.001 (compared to the WT‐control
   group); ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001 (compared to the
   WT‐CNV group); and ^# p < 0.05, ^## p < 0.01, ^### p < 0.001 (compared
   to the KO‐control group). Statistical analysis was performed using
   one‐way ANOVA followed by Tukey's multiple comparisons test.

   To further validate this hypothesis, we assessed microglial/macrophage
   activation in the retinas of PD‐L1 KO and WT mice by measuring CD68
   expression and IBA1+ cell numbers.^[ [128]^55 , [129]^56 ^] The results
   showed a significant increase in retinal CD68 mRNA levels in WT mice on
   day 3 post‐laser, with an even greater elevation in PD‐L1 KO mice
   (Figure [130]4D). Immunostaining revealed substantial IBA1+ cell
   accumulation at laser sites in both groups, with significantly more
   IBA1+ cells in PD‐L1 KO mice (Figure [131]4E,F). A similar trend was
   observed in the RPE‐choroid tissue (Figure [132]4G–I).

   Given that microglia and macrophages are major sources of
   proinflammatory cytokines in the retina and RPE‐choroid,^[ [133]^57 ,
   [134]^58 ^] we compared cytokine expression levels in these tissues. On
   day 3 post‐laser, retinal and RPE‐choroid tissues showed significantly
   increased mRNA levels of IL‐1β, IL‐6, TNF‐α, and CCL‐2. In PD‐L1 KO
   mice, the upregulation of these cytokines was more pronounced
   (Figure [135]4J,K). These findings suggest that PD‐L1 plays a critical
   role in limiting inflammation in CNV, and its deficiency exacerbates
   CNV by enhancing microglial activation and neuroinflammation. In
   summary, these data suggest the expression of PD‐L1 in retinal
   microglial cells is increased after laser injury, and KO of PD‐L1 will
   promote retinal microglial cell activation following laser injury,
   which may exacerbate vascular leakage and neovascular lesions by
   elevating the release of proinflammatory neurotoxic factors.

   An intriguing observation pertains to the mRNA expression levels of
   VEGF, a pivotal factor in promoting neovascularization.^[ [136]^59 ,
   [137]^60 ^] VEGF mRNA levels in the RPE‐choroid of PD‐L1 KO mice were
   significantly elevated compared to those in WT mice, whereas no
   significant differences were observed between WT and PD‐L1 KO groups in
   the retina (Figure [138]S2A,C, Supporting Information), consistent with
   RNA‐seq results (Figure [139]4B). These findings indicate that PD‐L1
   deficiency exacerbates inflammation in the retina and RPE‐choroid
   following laser‐induced CNV and is associated with increased VEGF mRNA
   expression in the choroid. However, the absence of PD‐L1 does not
   appear to significantly influence VEGF mRNA levels in the retina.

   To further explore the underlying mechanism, we examined whether
   PD‐1‐induced PD‐L1 activation attenuates microglia/macrophage‐mediated
   inflammation. On the third day post‐laser injury, intravitreal PD‐1
   injection in WT mice significantly reduced CD68 mRNA expression in the
   retina and RPE‐choroid (Figure [140]S3A,G, Supporting Information),
   indicating effective suppression of microglia/macrophage activation.
   Additionally, in the CNV model, PD‐1 injection decreased the mRNA
   levels of inflammatory cytokines and chemokines, including IL‐1β,
   TNF‐α, IL‐6, and CCL‐2, in the retina and RPE‐choroid of WT mice
   (Figure [141]3B–E,H–K, Supporting Information). However, the influence
   of PD‐1 on VEGF mRNA levels in both the retina and RPE‐choroid is not
   statistically significant (Figure [142]S2B,D, Supporting Information).

   In conclusion, these results highlight the crucial role of PD‐L1 in
   moderating neuroinflammation and vascular responses in CNV. Pronounced
   vascular leakage in PD‐L1 KO mice, accompanied by increased microglial
   activation and elevated inflammatory cytokines, underscores PD‐L1's
   potential as a therapeutic target for controlling inflammation and
   preventing the progression of CNV.

2.5. Upregulation of PD‐L1 in Microglia/Macrophages Directly Modulates
Microglial Inflammatory Responses

   Studies have shown that PD‐L1 is expressed in microglia/macrophages and
   upregulated in response to inflammatory stimuli, transmitting
   inhibitory signals that prevent excessive activation of these cells and
   helping to regulate neuroinflammation.^[ [143]^49 , [144]^61 ^]
   However, it remains unclear whether PD‐L1 influences microglial
   function through a similar mechanism in CNV. Additionally, the
   expression of PD‐L1 in retinal and RPE‐choroid microglia/macrophages
   has not been thoroughly investigated. To address this, we first used IF
   staining to assess PD‐L1 expression in microglia/macrophages in both
   retinal and RPE‐choroid tissues.

   The results of multiplex IF staining demonstrated that microglia in the
   retina express PD‐L1 (Figure [145]5A). Flow cytometry analysis further
   revealed a significant increase in PD‐L1 expression on retinal
   microglia/macrophages in CNV model mice (Figure [146]5B). Given the
   important role of microglia/macrophages in the RPE‐choroid during CNV
   progression,^[ [147]^46 , [148]^62 , [149]^63 ^] we also assessed PD‐L1
   expression in these cells within the RPE‐choroid. Multiplex IF staining
   showed that microglia near the laser‐induced lesions in the RPE‐choroid
   express PD‐L1 (Figure [150]5C). Subsequently, we isolated
   microglia/macrophages from the RPE‐choroid using magnetic bead sorting
   and performed qPCR analysis. The qPCR results indicated a significant
   upregulation of PD‐L1 mRNA in RPE‐choroid microglia/macrophages
   following CNV induction (Figure [151]5D).

Figure 5.

   Figure 5
   [152]Open in a new tab

   Upregulation of PD‐L1 in microglia/macrophages directly modulates
   microglial inflammatory responses. A) Immunofluorescence (IF) staining
   of the laser injured retina from wild‐type (WT) mouse at 3 days after
   laser injury with anti‐PD‐L1 (cyan) and anti‐IBA1 (red) antibodies. B)
   Left panels show the gating strategy for flow cytometric analysis of
   microglia cells in the retina. Microglia cells were identified by CD11b
   expression. Middle panels show the flow plots of PD‐L1+ microglia in
   retinas from control group and choroidal neovascularization (CNV) model
   mice. Right panels show percentages of PD‐L1+ microglia in control
   group and CNV model mice (n = 4 samples from control group and CNV
   model group). C) IF staining of the laser injured retinal pigment
   epithelium (RPE) choroid from WT mouse at 3 days after laser injury
   with anti‐PD‐L1 (cyan) and anti‐IBA1 (red) antibodies. D) Quantitative
   analysis of PD‐L1 mRNA in CD11b+ cells derived from the retinal pigment
   epithelium (RPE) choroid of WT mice, comparing the control group and
   laser‐treated group 3 days after laser injury (n = 4 samples from
   control group and CNV model group). E) IF staining of the primary
   microglia from WT mice with anti‐PD‐L1 (green) and anti‐CD11b (red)
   antibodies. F) Same as (E), but for the primary microglia from PD‐L1 KO
   mice. G–J) PBS control group, LPS group, and LPS + PD‐1 group. Levels
   of proinflammatory cytokines and VEGF secretion in primary microglia.
   The cells were derived from either WT or PD‐L1 KO mice (n = 4 per
   group). Data are presented as mean ± SEM, ^* p < 0.05, ^** p < 0.01,
   ^*** p < 0.001 compared with the control group. Statistical analysis in
   (B–D) was performed using an unpaired t‐test with Tukey's multiple
   comparisons test. G–J) ^^ p < 0.05, ^^^ p < 0.01, ^^^^ p < 0.001
   compared with the WT‐control group. ^* p < 0.05, ^** p < 0.01, ^*** p <
   0.001 compared with the WT‐LPS group. ^# p < 0.05, ^## p < 0.01, ^### p
   < 0.001 compared with the KO‐control group. All data in (G–J) were
   analyzed using one‐way ANOVA with Tukey's multiple comparisons test.

   These findings demonstrate that both retinal and RPE‐choroid
   microglia/macrophages express PD‐L1 and that PD‐L1 expression is
   significantly increased under CNV pathological conditions. This
   suggests that the retinal PD‐L1 upregulation may be similar to previous
   study, which linked to response to inflammatory stimuli, transmitting
   negative regulatory signals that limit excessive activation of these
   cells, thus helping to control neuroinflammation.^[ [153]^49 , [154]^61
   ^]

   Previous studies have demonstrated that PD‐L1 expressed on macrophages
   in vitro continuously transmits negative regulatory signals, thereby
   suppressing macrophage proliferation and the secretion of inflammatory
   cytokines. Blocking PD‐L1 or knocking out the PD‐L1 gene abolishes this
   regulatory mechanism, leading to heightened macrophage activation.^[
   [155]^64 , [156]^65 ^] To further investigate whether PD‐L1 directly
   acts on microglia/macrophages, we conducted in vitro experiments using
   primary microglia derived from WT and PD‐L1 KO mice. The identity of
   primary microglia was confirmed by CD68 IF staining (Figure [157]S4A,
   Supporting Information), and PD‐L1 expression was validated using
   Western blot analysis (Figure [158]S4B, Supporting Information). LPS
   was used to simulate an inflammatory environment.

   IF staining showed that PD‐L1 is expressed in WT primary microglia and
   is significantly upregulated following LPS stimulation
   (Figure [159]5E). This finding aligns with the observed upregulation of
   PD‐L1 in microglia/macrophages in the retinas and RPE‐choroid of CNV
   model mice in this study (Figure [160]5A–D). In contrast, primary
   microglia derived from PD‐L1 KO mice did not express PD‐L1 either
   before or after LPS stimulation (Figure [161]5F). These results were
   further corroborated by Western blot analysis (Figure [162]S4C,D,
   Supporting Information).

   To explore whether PD‐L1 directly regulates inflammatory responses, we
   conducted in vitro experiments using primary microglia from WT and
   PD‐L1 KO mice. PD‐L1 expression in WT microglia was significantly
   upregulated following LPS stimulation, consistent with ex vivo CNV
   findings. ELISA revealed that LPS‐induced secretion of proinflammatory
   cytokines (IL‐1β, IL‐6, TNF‐α) was further enhanced in PD‐L1 KO
   microglia, confirming the inhibitory role of PD‐L1. Interestingly,
   while activated microglia exhibited increased VEGF secretion, blocking
   PD‐L1 did not affect VEGF levels, indicating that PD‐L1 primarily
   modulates proinflammatory activity rather than proangiogenic functions.
   Additionally, exogenous PD‐1 significantly suppressed cytokine
   secretion in WT microglia but not in PD‐L1 KO cells, demonstrating that
   PD‐1 reinforces PD‐L1's inhibitory signals to mitigate inflammation.

   These results demonstrate that disrupting PD‐L1 signaling significantly
   influences microglia and macrophages, causing increased activation and
   intensifying inflammatory responses. Furthermore, while activated
   microglia and macrophages secrete VEGF, this activity seems unrelated
   to PD‐L1 signaling. This suggests that the inflammatory and angiogenic
   roles of these cells are controlled by separate pathways,^[ [163]^46 ^]
   with PD‐L1 specifically regulating their inflammatory functions.

2.6. Microglia Depletion Abolishes the Effect of PD‐L1 on CNV in Mice

   To investigate whether PD‐L1 regulates CNV in mice directly through
   microglia, we utilized a PLX5622 diet to deplete microglia.^[ [164]^66
   ^] PLX5622 is an inhibitor specifically designed to target the
   colony‐stimulating factor‐1 receptor (CSF‐1R).^[ [165]^67 , [166]^68 ^]
   In WT mice, IF staining further confirmed that 1 week of PLX5622
   treatment effectively depletes retinal microglia (Figure [167]S5A,B,
   Supporting Information). qPCR analysis confirmed that 1 week of
   treatment with PLX5622 resulted in a significant downregulation of
   retinal microglia‐associated genes (Figure [168]S5C–F, Supporting
   Information).

   After depleting microglia using PLX5622, PD‐L1 mRNA and protein levels
   in the retina were significantly reduced compared to those in mice on a
   normal diet (including WT‐control and WT‐CNV groups) and did not
   increase notably following laser‐induced CNV (Figure [169]6A–C). In the
   RPE‐choroid, PD‐L1 mRNA and protein levels in PLX5622‐control mice
   showed no significant differences compared to normal diet groups.
   However, in the PLX5622‐CNV group, PD‐L1 expression in the RPE‐choroid
   exhibited a decreasing trend, with PD‐L1 protein levels in the choroid
   significantly lower than those in PLX5622‐control mice
   (Figure [170]6D–F). These findings suggest that PD‐L1 in the retina is
   predominantly expressed by microglia/macrophages, while in the
   RPE‐choroid, its expression is only partially attributed to these
   cells.

Figure 6.

   Figure 6
   [171]Open in a new tab

   Microglia depletion abolishes the effect of PD‐L1 on choroidal
   neovascularization (CNV) in mice. A) Quantitative analysis of PD‐L1
   mRNA expression in the retina of mice on two different diets, 3 days
   after laser injury. B) Representative Western blot images showing
   retinal PD‐L1 protein levels in mice on a normal diet or PLX5622 diet,
   3 days post‐laser injury. C) Quantitative analysis of retinal PD‐L1
   protein levels in mice fed a normal diet or PLX5622 diet. D) PD‐L1 mRNA
   expression in retinal pigment epithelium (RPE) choroid tissues,
   analyzed using the same method as in panel (A). E) Representative
   Western blot images showing PD‐L1 protein levels in the RPE‐choroid of
   mice on two different diets, 3 days after laser injury. F) Quantitative
   analysis of PD‐L1 protein levels in RPE‐choroid tissues, analyzed using
   the same method as in panel C (n = 4 mice per group). G) Top panels
   show representative fundus fluorescein angiography (FFA) images from
   wild‐type (WT) and PD‐L1 knockout (KO) mice fed with PLX5622 diet at 7
   days post‐laser injury. Bottom panels show the example images of
   optical coherence tomography (OCT) scan. H) Summaries of the FFA grade
   scores of each laser spot from WT and PD‐L1 KO mice fed with PLX5622
   diet at 7 days post‐laser injury (n = 23 and 25 laser spots from WT and
   KO mice, respectively). I) Summaries of the average leakage grade from
   WT and PD‐L1 KO mice fed with normal or PLX5622 diet at 7 days
   post‐laser injury. For normal diet group, n = 11 WT mice and 10 KO
   mice. For PLX5622 diet group, n = 6 WT mice and 6 KO mice. J)
   Immunofluorescence (IF) staining of IB4 and IBA1 in RPE/choroidal flat
   mounts from laser‐induced CNV mice. The left two panels display
   RPE/choroidal images of WT and PD‐L1 KO mice from the normal diet
   group, and the right two panels exhibit RPE/choroidal images of WT and
   PD‐L1 KO mice from the PLX5622 diet group. K,L) Quantifications of CNV
   thickness and area in RPE/choroidal flat mounts at 7 days after laser
   injury. For normal diet, n = 7 WT mice and 10 KO mice. For PLX5622 diet
   group, n = 6 WT mice and 6 KO mice. Data are presented as mean ± SEM.
   Chi‐square test for FFA grade scores in (H). Nonparametric test for
   leakage average degree in (I). One‐way ANOVA with Tukey's multiple
   comparisons test was used for analyzing PD‐L1 expression levels in
   (A,C,D,F), as well as CNV thickness in (K) and CNV area in (L). ^* p <
   0.05; ^** p < 0.01; and ^*** p < 0.001 compared to the normal food WT
   group. ^^ p < 0.05; ^^^ p < 0.01; and ^^^^ p < 0.001, compared to the
   normal food KO group. Source data are provided as a Source Data file.

   To further investigate the effects of microglial depletion on
   inflammation, we assessed inflammatory markers in the retina and
   RPE‐choroid of WT and PD‐L1 KO mice. The mRNA levels of IL‐1β, IL‐6,
   and TNF‐α were significantly reduced in PLX5622‐treated mice compared
   to those on a normal diet, with no significant differences between WT
   and PD‐L1 KO mice (Figure [172]S5G,H, Supporting Information).

   Consistent with these findings, FFA analysis revealed that after
   PLX5622 treatment, only ≈4% of WT and PD‐L1 KO mice exhibited grade 4
   leakage^[ [173]^53 ^] (Figure [174]6G,H), compared to 9.5% and 29.7%,
   respectively, in normal diet groups (Figure [175]2A,B). Furthermore,
   the average leakage grades of CNV were significantly reduced in both WT
   and PD‐L1 KO mice (Figure [176]6G,I). OCT analysis confirmed a
   significant reduction in CNV thickness in PLX5622‐treated WT and KO
   mice compared to normal diet groups (Figure [177]6G,K). IB4 staining of
   RPE‐choroid flat mounts further demonstrated a marked decrease in CNV
   lesion areas in PLX5622‐treated mice (Figure [178]6J,L). Importantly,
   no significant differences in CNV leakage or lesion size were observed
   between WT and PD‐L1 KO mice after PLX5622 treatment
   (Figure [179]6H,I,K,L).

   These results indicate that microglia/macrophage depletion via PLX5622
   significantly reduces inflammation in the retina and RPE‐choroid,
   suppressing CNV formation and leakage. Moreover, microglial/macrophage
   depletion reverses the increased inflammation and exacerbated CNV
   observed with PD‐L1 blockade.

   Collectively, these findings suggest that the regulatory effects of
   PD‐L1 on CNV in mice are mediated through microglia/macrophages.

2.7. PD‐L1 Modulates CNV Progression by Regulating the MAPK/ERK Pathway in
Microglia

   To elucidate the specific molecular mechanisms underlying the
   laser‐induced CNV and investigate the influence of PD‐L1 on retinal
   microglial cells in the CNV model, we conducted RNA‐seq analysis on
   retinal tissues. Firstly, we compared the differential gene expression
   profiles between two distinct groups: WT mice at 3 days post‐laser
   injury (WT‐CNV) and control mice without laser intervention
   (WT‐Control) (Figure [180]7A). PD‐L1 mRNA (coded by Cd274 gene) was
   upregulated in the WT‐CNV group (Figure [181]7A), consistent with our
   previous findings from WB and qPCR (Figure [182]1A,B). KEGG pathway
   analyses unveiled a substantial enrichment of differentially expressed
   genes, primarily associated with inflammation and neovascularization
   (Figure [183]7B). We conducted a comprehensive pathway enrichment
   analysis using gene set enrichment analysis (GSEA) between WT (WT‐CNV)
   and KO (KO‐CNV) mice at 3 days after laser injury. Notably, our
   analysis revealed a significant upregulation of MAPK/ERK pathway in the
   retinas of KO mice compared to WT mice (Figure [184]7C).

Figure 7.

   Figure 7
   [185]Open in a new tab

   PD‐L1 modulates choroidal neovascularization (CNV) progression by
   regulating the MAPK/ERK pathway in microglia. A) Volcano plot obtained
   from DESeq2 analysis of wild‐type (WT) mice retina at 3 days after
   laser injury compared with WT mice without laser intervention. B) KEGG
   pathway analysis of the differential expression genes (DEGs). C) Gene
   set enrichment analysis (GSEA) pathway analysis of the DEGs from PD‐L1
   knockout (KO) mice retina and WT mice retina at 3 days after laser
   injury. D) Western blot images and E) statistical results of the
   expression levels of p‐ERK in the retinas of control mice and mice with
   3 days after laser treatment, for both WT and PD‐L1 KO mice (n = 4 mice
   per group). F) Western blot images and G) statistical result of the
   p‐ERK expression in the retinas from WT mice of the control, CNV +
   vehicle, CNV + PD‐1 groups at 3 days after laser injury (n = 4 mice per
   group). H–K) Same as (D–G), but for primary microglia. Data are
   presented as mean ± SEM. One‐way ANOVA with Tukey's multiple
   comparisons test was used for statistical analyses, ^^ p < 0.05, ^^^ p
   < 0.01, and ^^^^ p < 0.001, compared to the WT‐control group. ^* p <
   0.05, ^** p < 0.01, and ^*** p < 0.001 compared to the WT‐CNV or WT‐LPS
   group. ^# p < 0.05, ^## p < 0.01, and ^### p < 0.001 compared to the
   KO‐control group. Source data are provided as a Source Data file.

   The MAPK signaling pathway comprises a network of critical components
   including p38, extracellular signal‐regulated kinases (ERK), and c‐Jun
   N‐terminal kinase (JNK).^[ [186]^69 , [187]^70 ^] Activation of MAPK
   pathway assumes a central role in regulating inflammatory responses and
   innate immunity.^[ [188]^69 , [189]^71 ^] Previous studies showed that
   inhibiting PD‐L1 leads to an upregulation of phosphorylated ERK^[
   [190]^72 , [191]^73 ^] and influences the functions of microglial
   cells.^[ [192]^43 ^] Based on these studies and our RNA‐seq analysis,
   we inferred that PD‐L1 exerts an influence on CNV via the ERK
   pathway‐based microglial activation. We employed WB to evaluate the
   p‐ERK level in the retinas following laser‐induced injury and observed
   a significant upregulation of retinal p‐ERK level in both WT and PD‐L1
   KO mice following laser photocoagulation (Figure [193]7D,E). Notably,
   KO mice had a higher elevation of p‐ERK in comparison to WT mice
   (Figure [194]7D,E).

   Previous results demonstrated that PD‐1 suppresses microglial
   activation and inflammation via PD‐L1. To investigate whether PD‐1
   exerts its effects through the downstream ERK pathway after binding to
   PD‐L1, Western blot analysis was performed. In the laser‐induced CNV
   model, PD‐1 significantly reduced ERK phosphorylation in the retina
   (Figure [195]7F,G), consistent with the observation that PD‐L1 KO
   further increased p‐ERK levels in the retina of CNV model mice
   (Figure [196]7D,E).

   In retinal homogenates from CNV‐induced PD‐L1 KO mice, we observed
   further activation of the p‐ERK pathway. Given the diverse cell types
   present in the retina, we designed in vitro experiments to specifically
   assess the effect of PD‐L1 on the ERK pathway in microglia. Consistent
   with previous studies,^[ [197]^74 ^] LPS stimulation significantly
   increased p‐ERK protein levels in primary microglia (isolated from WT
   and PD‐L1 KO mice) (Figure [198]7H,I). Furthermore, in line with our in
   vivo findings, LPS induced a more pronounced increase in p‐ERK
   expression when PD‐L1 signaling was blocked (Figure [199]7H,I). To
   determine whether PD‐1 directly modulates the ERK signaling pathway in
   microglia, primary microglia derived from WT and PD‐L1 KO mice were
   pretreated with PD‐1 protein or control vehicle 2 h before LPS
   stimulation. PD‐1 significantly inhibited LPS‐induced activation of the
   ERK pathway in WT primary microglia (Figure [200]7J,K).

   In conclusion, our findings demonstrate that the MAPK/ERK pathway is
   activated in both the retinas of CNV mice and microglia following
   inflammatory stimulation. Reactively upregulated PD‐L1 provides
   sustained negative regulatory signals that partially suppress ERK
   activation. Disruption of PD‐L1 signaling removes this negative
   feedback, exacerbating neuroinflammation mediated by microglia and
   macrophages. Additionally, activation of PD‐L1 by PD‐1 further
   attenuates MAPK/ERK pathway activation in the retinas of CNV mice and
   in microglia.

2.8. ERK Inhibitor Suppresses Microglial Activation and Reduces CNV Lesion
Area

   To investigate whether PD‐L1 specifically regulates microglial
   activation through the MAPK pathway, we employed the ERK pathway
   inhibitor ASTX029.^[ [201]^75 ^] ASTX029 was intravitreally injected
   into WT and PD‐L1 KO mice immediately following CNV induction. On the
   third day post‐laser injury, Western blot analysis revealed a
   significant reduction in retinal p‐ERK levels in both WT and KO mice
   treated with ASTX029 (Figure [202]8A). IB4 staining of RPE‐choroid flat
   mounts demonstrated a substantial decrease in CNV lesion areas in
   ASTX029‐treated mice compared to PBS‐treated controls
   (Figure [203]8B,C). Furthermore, the differences in retinal p‐ERK
   levels and CNV lesion sizes between WT and KO mice were abolished by
   ASTX029 treatment (Figure [204]8B,C, Supporting Information).

Figure 8.

   Figure 8
   [205]Open in a new tab

   ERK inhibitor suppresses microglial activation and reduces choroidal
   neovascularization (CNV) lesion area. A) Representative Western blot
   images (top) and statistical analysis (bottom) of p‐ERK expression in
   the retinas of wild‐type (WT) and PD‐L1 knockout (KO) mice, 3 days
   post‐laser treatment, following intravitreal injection of ASTX029 or
   PBS. B) IB4 staining of retinal pigment epithelium (RPE)/choroidal flat
   mounts from WT and KO mice, 7 days after laser injury, following
   intravitreal injection of ASTX029 or PBS (n = 4 mice per group). C)
   Quantification of the laser‐induced CNV area. D) Effects of different
   concentrations of ASTX029 on primary microglia derived from WT mice. E)
   Representative Western blot images and (F) statistical analysis of
   p‐ERK expression levels. G–I) Corresponding analyses of IL‐1β, IL‐6,
   and TNF‐α mRNA expression in LPS‐stimulated primary microglia samples
   derived from WT or KO mice. Data are presented as mean ± SEM (n = 4
   mice or four samples of primary microglia per group). Statistical
   significance is denoted as follows: ^^ p < 0.05, ^^^ p < 0.01, ^^^^ p <
   0.001 (compared to the WT‐control group); ^* p < 0.05, ^** p < 0.01,
   ^*** p < 0.001 (compared to the WT‐CNV or WT‐LPS group); ^# p < 0.05,
   ^## p < 0.01, ^### p < 0.001 (compared to the KO‐control group).
   Statistical analyses were performed using one‐way ANOVA followed by
   Tukey's multiple comparisons test.

   To further confirm whether ERK pathway inhibition directly suppresses
   microglial secretion of inflammatory cytokines, we treated primary
   microglia derived from WT mice with ASTX029 in vitro. CCK‐8 assays
   indicated that 0.5 nmol L^−1 of ASTX029 had no significant impact on
   microglial viability, and this concentration was used for subsequent
   experiments (Figure [206]8D). Western blot analysis showed that ASTX029
   significantly reduced p‐ERK levels in LPS‐stimulated primary microglia
   and suppressed the secretion of IL‐1β, IL‐6, and TNF‐α
   (Figure [207]8E–I). Notably, ASTX029 also reversed the differences in
   ERK activation between microglia derived from WT and KO mice,
   consistent with the ex vivo findings (Figure [208]8A).

   These results collectively suggest that PD‐L1 regulates
   microglia‐mediated neuroinflammation through the MAPK pathway,
   highlighting its role as a key modulator of the MAPK/ERK signaling
   cascade.

3. Discussion

   Neuroinflammation and dysregulation of neuroimmune processes are
   prominent features in various retinal degenerative disorders, including
   AMD.^[ [209]^20 ^] ICPs have emerged as crucial regulators of
   inflammation and immune responses, exhibiting significant therapeutic
   potential in diverse fields, such as oncology^[ [210]^76 ^] and
   inflammatory disorders.^[ [211]^77 ^] Among these, the PD‐1/PD‐L1 axis
   stands out as a pivotal molecular component.^[ [212]^38 ^] This study
   investigated the role and possible mechanisms of PD‐L1 in the
   laser‐induced CNV model. We observed that retinal PD‐L1 expression,
   particularly on microglia/macrophages, was increased following the
   laser‐induced CNV. Blockage of PD‐L1 by PD‐L1 genetic ablation or
   antibody exacerbated the pathological neovascularization in the CNV
   model, which may due to retinal microglia/macrophages activation and
   the elevation of neuroinflammation via ERK‐related pathway. In
   addition, intravitreal delivery of PD‐1, the ligand of PD‐L1, reduced
   retinal p‐ERK activation and neuroinflammation to mitigate vascular
   leakage and neovascularization in the CNV model. These evidences
   highlight the pivotal role of ICPs in regulating innate neuroimmune
   responses and neuroinflammatory processes in the laser‐induced CNV
   mouse model, and suggest that targeting the PD‐L1 pathway may be a
   potent strategy to treat the NVAMD.

   In the central nervous system, PD‐L1 exhibits a typical upregulation in
   response to inflammatory stimuli, which will reduce neuroinflammation
   and mitigate disease severity.^[ [213]^34 ^] Previous investigations
   have substantiated an augmented expression of PD‐L1 in inflamed
   retinas,^[ [214]^78 ^] although retinal PD‐L1 level was reported to be
   diminished in the proliferative diabetic retinopathy.^[ [215]^79 ^] Our
   retinal RNA‐seq, PCR, and WB analyses disclosed a noteworthy elevation
   of retinal PD‐L1 subsequent to the laser‐induced CNV
   (Figures [216]1A,B, and [217]7A).

   PD‐L1 is expressed across multiple cell types within CNV lesions.^[
   [218]^80 ^] In conditions such as postoperative brain injury^[ [219]^61
   ^] and spinal cord injury,^[ [220]^43 ^] PD‐L1 on microglia can
   suppress excessive activation and mitigate neuroinflammation.
   Consistent with these findings, this study demonstrates that PD‐L1 is
   expressed on microglia/macrophages in the retina and RPE‐choroid of
   mice, with expression levels increasing following CNV induction
   (Figure [221]5A–D). Additionally, primary microglia from WT mice
   exhibited PD‐L1 expression, which was enhanced post‐inflammatory
   stimulation (Figure [222]5E, Figure [223]S4C,D, Supporting
   Information). Blocking PD‐L1 promoted the secretion of inflammatory
   cytokines by microglia/macrophages (Figure [224]5G–I), whereas PD‐1
   activation of PD‐L1 inhibited these cells and reduced cytokine release
   (Figure [225]5G–I). Furthermore, the overall retinal PD‐L1 expression
   patterns mirrored those observed in microglia/macrophages, with
   significant upregulation following either CNV induction or inflammatory
   stimuli. Removal of microglia/macrophages led to a marked decrease in
   retinal PD‐L1 levels, indicating that PD‐L1 expression in the retina
   predominantly originates from these cells.

   Despite an increase in PD‐L1 expression among activated
   microglia/macrophages within the RPE‐choroid in CNV model mice
   (Figure [226]5C,D), overall PD‐L1 levels in the RPE‐choroid tissue did
   not significantly alter (Figure [227]1C,D). After depleting
   microglia/macrophages, a downward trend in PD‐L1 levels in RPE‐choroid
   tissue was observed (Figure [228]6D–F). This discrepancy may stem from
   other cell types within the RPE‐choroid, such as RPE cells, which are
   also capable of expressing PD‐L1.^[ [229]^44 , [230]^81 ^] RPE damage
   constitutes a pivotal factor in the pathogenesis of NVAMD^[ [231]^82 ^]
   potentially leading to inconsistent PD‐L1 expression patterns between
   microglia/macrophages and the overall tissue. Future research should
   delve deeper into PD‐L1 expression across various cell types within the
   RPE‐choroid and specifically quantify PD‐L1 in RPE‐choroid
   microglia/macrophages to elucidate the role of PD‐L1 in modulating
   inflammation in the CNV model more precisely.

   During activation, microglia/macrophages release a significant amount
   of inflammatory cytokines, major contributors to tissue damage and
   vascular leakage.^[ [232]^46 ^] Reactively upregulated PD‐L1 can
   inhibit the activation levels and cytokine secretion of
   microglia/macrophages.^[ [233]^61 , [234]^83 ^] In alignment with prior
   research, our study shows that blocking PD‐L1 signaling leads to
   enhanced activation of microglia/macrophages in the retina and
   RPE‐choroid of CNV mice, and consequently higher levels of inflammation
   in these tissues (Figure [235]4). Further, in vitro studies illustrate
   that inhibiting PD‐L1 signaling results in increased secretion of
   proinflammatory cytokines from microglia following LPS stimulation
   (Figure [236]5G–I), whereas activating PD‐L1 markedly reduces cytokine
   output (Figure [237]5G–I). These observations suggest that PD‐L1's
   protective mechanism in CNV likely involves suppressing inflammatory
   cytokine secretion by microglia/macrophages, thereby mitigating
   inflammation‐associated vascular leakage and pathological angiogenesis.

   Moreover, to assess whether PD‐L1's protective effects specifically
   rely on microglia/macrophages, we utilized PLX5622 to deplete these
   cells in mice. Following microglial/macrophage depletion, we observed
   decreased levels of inflammatory cytokines, reduced vascular leakage,
   and a diminished extent of CNV (Figure [238]6G–L, Figure [239]S5G,H,
   Supporting Information), reinforcing that moderating inflammation in
   these tissues is an effective strategy for treating CNV. Additionally,
   once microglia/macrophages were depleted, the differences in local
   inflammation, vascular leakage, and neovascular area between PD‐L1 KO
   and WT mice were eliminated, indicating that PD‐L1's role in modulating
   CNV predominantly involves these cells.

   Despite the predominant use of anti‐VEGF therapy in the management of
   CNV,^[ [240]^84 ^] activated microglia/macrophages not only produce
   increased inflammatory cytokines but also secrete VEGF.^[ [241]^85 ^]
   According to Anne Wolf et al., the proinflammatory and VEGF‐secreting
   functions of microglia may be regulated through distinct mechanisms.^[
   [242]^46 ^] Our findings confirm that PD‐L1 signaling specifically
   inhibits cytokine production without affecting VEGF levels in vitro in
   microglia and in the retina of CNV model mice (Figure [243]5J, Figure
   [244]S3, Supporting Information), indicating that PD‐L1's inhibitory
   effect on CNV does not operate via anti‐VEGF pathways. Given the
   efficacy of anti‐VEGF in suppressing CNV, future clinical strategies
   might explore the potential synergistic effects of combining PD‐L1 and
   anti‐VEGF therapies, potentially mirroring dual‐specificity antibody
   treatments used in cancer therapy.^[ [245]^86 ^] Additionally, the
   cellular distribution of VEGF signaling warrants further exploration to
   fully understand its role in the CNV.

   It is important to note that extensive research has demonstrated that
   CNV in mice often presents with significant local inflammation
   immediately after model induction, with microglia/macrophage activation
   peaking on the third day post‐laser treatment and CNV formation peaking
   on the seventh day.^[ [246]^52 ^] This suggests that the inflammatory
   processes in CNV may precede and continually influence pathological
   neovascularization. Our results indicate that immediate post‐laser PD‐1
   injection can suppress CNV by inhibiting inflammation, highlighting the
   potential of early anti‐inflammatory treatments to improve clinical
   outcomes.

   Through RNA sequencing, we observed that laser‐induced CNV
   significantly influenced the MAPK pathway, a common target for
   mitigating neuroinflammation (Figure [247]7A–C). The ERK subgroup, a
   crucial component of the MAPK family, holds particular significance.
   Activation of the ERK pathway has also been observed in retinal
   degenerative diseases, including AMD, and inhibiting ERK has been shown
   to alleviate retinal neuroinflammation.^[ [248]^87 ^,154] Consistent
   with previous studies, our research revealed a significant upregulation
   of the ERK pathway in the retinas of CNV mice and in microglia
   activated by LPS. Inhibition of the ERK pathway effectively attenuates
   CNV and reduces proinflammatory cytokine secretion by microglia
   (Figure [249]8). These findings suggest that activation of the ERK
   pathway plays a crucial role in the activation of microglia/macrophages
   and the inflammatory response in CNV, making it a potential therapeutic
   target.

   The relationship between PD‐L1 and the ERK pathway is noteworthy.
   Previous studies have documented that PD‐L1 activation in spinal cord
   injury can inhibit the MAPK pathway, specifically ERK.^[ [250]^43 ^]
   However, additional research has demonstrated that the PD‐1 active
   PD‐L1 signaling, leading to the increased phosphorylation of ERK in T
   cells.^[ [251]^88 ^] These suggest that the signaling mechanisms of
   PD‐L1 may vary across diverse tissues and diseases. (Figure [252]7C–E).
   In our study, we found that blocking PD‐L1 with antibodies or knocking
   out the PD‐L1 gene further enhanced the activation of the MAPK/ERK
   pathway in the retinas of CNV mice and in LPS‐stimulated microglia
   (Figure [253]7D,E,H,I). Conversely, intervention with PD‐1 resulted in
   a decrease in ERK phosphorylation levels (Figure [254]7F,G,J,K). These
   findings align with those of Hartley et al,^[ [255]^89 ^] suggesting
   that PD‐L1, as a negative co‐stimulatory ICP molecule, can continuously
   suppress the activation of certain intracellular signaling pathways,
   thus regulating cellular functions. In our experiments, we observed
   that the upregulated PD‐L1 could only partially limit ERK pathway
   activation, and even enhanced PD‐L1 signaling via PD‐1 could not fully
   suppress ERK activation. This may be due to the complex regulatory
   mechanisms of the ERK pathway,^[ [256]^90 ^] which prevent PD‐L1
   signaling from completely reversing ERK pathway activation in
   proinflammatory microglia.

   This study is subject to several limitations. NVAMD is a multifaceted
   condition influenced by a variety of factors, including age,
   metabolism, genetics, and environmental elements.^[ [257]^3 , [258]^13
   ^] However, our employed model primarily replicates one crucial facet
   of NVAMD observed in humans, the development of CNV, which is induced
   through laser‐induced injury. This model initiates a wound‐healing
   response subsequent to an insult at the level of Bruch's membrane,
   heavily relying on an inflammatory process.^[ [259]^42 ^] Consequently,
   it underscores the predominant role of ICP regulation. When
   contemplating potential therapeutic targets, it is crucial to consider
   additional variables such as genetics and age. In addition, due to the
   expression of PD‐L1 in various cell types,^[ [260]^33 , [261]^43 ,
   [262]^44 ^] additional experiments employing retinal microglia
   conditional KO mice are imperative to ascertain the precise role of
   PD‐L1 in CNV. In our experiments, we used global PD‐L1 KO mice to
   demonstrate that PD‐L1 deficiency exacerbates laser CNV injury and
   angiogenesis. However, this global KO approach does not elucidate the
   specific role of microglial PD‐L1. Further experiments employing
   conditional KO models, such as microglia‐specific PD‐L1 KO mice, are
   needed to convincingly establish the microglial‐specific effects of
   PD‐L1 on angiogenesis.

   In summary, our research provides compelling evidence for the
   involvement of ICPs in the regulation of CNV. We propose that PD‐L1
   serves as a regulatory node, modulating retinal microglia functions via
   ERK. Specifically, PD‐1, through the PD‐L1/ERK signaling pathway,
   mitigates retinal neuroinflammation, thus providing an effective method
   for suppressing CNV. The manipulation of ICPs presents a promising and
   indispensable role in the NVAMD treatment.

4. Experimental Section

Animals

   The experiment utilized adult mice aged 8–10 weeks. PD‐L1 KO mice with
   a C57BL/6 background were obtained from Cyagen Biosciences Inc.
   (Suzhou, China), while PD‐L1 WT mice with a genetically equivalent
   background served as the control group. To minimize potential
   experimental biases resulting from gender differences, an equal number
   of male and female mice were employed unless otherwise specified. Mice
   were housed in a controlled environment at a temperature of 22 ± 2 °C
   with a 12:12‐h light–dark cycle, and they had free access to ample
   provisions of food and water. Genotyping was performed using PCR
   analysis on tail DNA samples.^[ [263]^91 ^] All experimental procedures
   adhered to the guidelines established by the Animal Care and Use
   Committee of Fudan University (Shanghai, China; approval No.
   2023DW008).

Laser Photocoagulation

   Mice were anesthetized by intraperitoneal injection of 1% pentobarbital
   sodium (40 mg kg^−1). Pupil dilation was achieved using 0.5%
   tropicamide eye drops. The laser photocoagulation procedure followed
   the guidelines described in reference.^[ [264]^42 ^] A diode laser
   system, designed in a slit‐lamp style (Quantel Medical Vitra),
   employing a green laser at 532 nm with a power of 100 mW, a duration of
   100 ms, and a spot size of 100 µm, was utilized. To facilitate fundus
   observation, a glass coverslip was positioned in front of each eye.
   Four laser spots were administered around the optic nerve. Successful
   laser photocoagulation was determined by the minimal presence of
   bubbles preceding the laser spot, indicating the rupture of Bruch's
   membrane.^[ [265]^42 ^]

Intravitreal Delivery

   For intravitreal delivery, mice were anesthetized, and their pupils
   were dilated using the aforementioned methods. A 34‐gauge needle was
   inserted into the vitreous space approximately 1.5 mm below the limbus,
   and 1 µL solution was administered bilaterally with a Nanofil
   syringe.^[ [266]^92 ^] The following groups were designated: anti‐PD‐L1
   (5 ng PD‐L1 antibody, clone 10F.9G2, Biolegend), PD‐1 (5 ng PD‐1
   protein, RP01170, Abclone), and vehicle (saline only), ASTX029 (50 µm,
   S20663, MedMol.).^[ [267]^93 ^]

Fundus Photography and FFA

   The fundus photography and FFA examination were conducted at 7 days
   post‐laser photocoagulation.^[ [268]^46 ^] Mice were appropriately
   immobilized and subjected to anesthesia. Pupil dilation was achieved
   using a combination of tropicamide eye drops. Mice were positioned
   correctly, carbachol eye drops were applied to the cornea.
   Subsequently, the cornea was brought into contact with the microscope
   lens of the Optoprobe Science LTD small animal fundus imaging system
   (OPTO‐RIS). The experimental table and lens focus were adjusted
   according to facilitate optimal imaging conditions, following which
   photographs were taken.

   We performed FFA to evaluate vascular leakage at 7 days after laser
   treatment.^[ [269]^46 ^] After anesthesia and pupil dilation, the mice
   received an intraperitoneal injection of 1.7 mL kg^−1 of a 2% sodium
   fluorescein solution. Subsequent photographs were taken to document the
   FFA images. Both eyes were photographed to ensure comprehensive data
   collection. Following the experiment, eyes were rinsed with saline
   solution, and levofloxacin eye drops were administered as a preventive
   measure against potential infections. Retinal arterial filling
   represents the early phase of angiography ((1–2 min after fluorescein
   injection)), whereas complete filling of both retinal arteries and
   veins marks the late phase of angiography (4–5 min after fluorescein
   injection).^[ [270]^94 , [271]^95 ^]

   The assessment of fluorescein leakage intensity in the study followed
   the established Takehana grading criteria.^[ [272]^42 ^] The criteria
   provided a systematic framework for classifying the observed patterns:
   Grade 1, denoted as “no leakage,” described the presence of faint high
   fluorescence or speckled fluorescence; Grade 2, referred to as
   “questionable leakage,” indicated the absence of size or intensity
   increase in late‐phase high fluorescence; Grade 3, termed “leaky,”
   described the situation where the fluorescence intensity within the
   high fluorescence area increased while the size remained unchanged;
   Grade 4, known as “pathologically significant leakage,” characterized
   both an increase in fluorescence intensity and size. The average grade
   represents the mean value of leakage grades for each laser spot on the
   retina. Adhering to this standardized classification scheme allowed for
   a more precise and consistent evaluation of fluorescein leakage
   intensity in the experimental analysis.

Optical Coherence Tomography (OCT)

   We employed OCT to quantify the CNV thickness and detect the presence
   of sub‐retinal fluid.^[ [273]^96 ^] After administering anesthesia and
   pupil dilation, surface anesthesia was further achieved using
   proparacaine hydrochloride eye drops. Subsequently, mice were
   positioned on an elevated table. The eye's position was carefully
   adjusted, and medical carbomer eye drop gel was applied to the cornea
   under examination. The Ultramicro Ophthalmol Imaging System (ISOCT,
   OPTOPROBE) light source was finely tuned to focus, and the lens was
   adjusted to focus on the retina. OCT of the retina was then performed.
   Both eyes of each animal were photographed for comprehensive data
   collection. CNV thickness was measured by manually drawing line
   segments perpendicular to the RPE at the site of CNV in OCT images.

Flat Mounts, Immunohistochemistry, and Image Analysis

   Mice were euthanized using cervical dislocation, and their eyeballs
   were removed and fixed in 4% paraformaldehyde (PFA) for 2 h. The retina
   and RPE‐choroid complex were isolated and subjected to permeabilization
   and blocking. Following this, the primary antibody, anti‐iba1 (Wako,
   019–19741), was applied to the samples and allowed to incubate for 24 h
   at 4 °C. Subsequent to the primary antibody incubation, the samples
   were subsequently incubated with a secondary antibody, goat anti‐rabbit
   AlexaFluorTM 647 (A21244, Invitrogen), for 2 h. In the case of
   RPE/choroid samples, staining was conducted using IB4 (Sigma‐Aldrich).
   Following three PBS washes, the retina and RPE/choroid samples were
   carefully positioned flat on glass microscope slides for imaging
   purposes. The area of laser lesions was quantitatively evaluated using
   ImageJ in a blinded manner.

Flow Cytometry Staining and Analysis

   Tissues were rinsed in 1% three‐antibody PBS buffer, followed by
   fragmentation using ophthalmic scissors. After digestion with trypsin
   and collagenase, the tissue fragments were washed and resuspended in
   high‐glucose DMEM culture medium with FBS. The resulting cell
   suspension was filtered and prepared for analysis. Cells were incubated
   with specific antibodies, followed by centrifugation and resuspension
   in flow cytometry staining buffer. Flow cytometry analysis was
   performed to measure the expression levels of CD11B (ab8878, abcam) and
   PD‐L1 (ab205921, abcam).

Transcription Analysis by RT‐qPCR

   Tissue samples obtained from ex vivo mouse experiments were promptly
   collected and immediately snap‐frozen. Total RNA extraction was carried
   out using the Tissue RNA Purification Kit from EZBioscience (USA).
   Subsequently, cDNA was synthesized using the EZscript Reverse
   Transcription Mix II with gDNA Remover (EZBioscience). Gene expression
   analysis was conducted employing the Bio‐Rad RT‐PCR System (Hercules,
   USA) with 2× SYBR Green qPCR Master Mix. β‐actin served as the internal
   control for normalization purposes. The relative changes in gene
   expression were quantified using the 2^−ΔΔCt method. Each sample was
   assayed in triplicate, and the presented data represents the mean of
   three independent assessments. Primer information is provided in Table
   [274]S1 (Supporting Information).

Western Blot Analysis

   Fresh retinal/RPE‐choroid tissue was homogenized using a lysis buffer
   (P0013B, Beyotime) for protein extraction. The protein concentration of
   each sample was determined with a BCA protein assay kit (P0012S,
   Beyotime). Equal amounts of protein were loaded onto 10% SDS‐PAGE gels
   and subsequently transferred to PVDF membranes. After blocking with 5%
   milk for 2 h at room temperature, the membranes were incubated
   overnight at 4 °C with primary antibodies. Following this, they were
   incubated with horseradish peroxidase (HRP) conjugated secondary
   antibodies (A0216 or A0208, Beyotime) for 2 h at room temperature. The
   primary antibodies used in Western blotting included: ERK1/ERK2 Rabbit
   pAb (ABclone, A16686); Phospho‐ERK1‐T202/Y204 + ERK2‐T185/Y187 Rabbit
   mAb (ABclone, AP0974); Anti‐PD‐L1 antibody (abcam, ab213480); and mouse
   anti‐GAPDH (60004‐1‐Ig, Proteintech). Immunoblots were visualized using
   an ECL Kit Chemiluminescence (P0018S, Beyotime), and Image‐J software
   1.80 (National Institutes of Health, USA) was employed for data
   analysis.

3D Image Reconstruction and Analysis

   Z‐stack confocal images were obtained using either a Zeiss LSM 900
   confocal microscope with a Plan‐Apochromat ×20/1.4 NA objective.
   Three‐dimensional retinal images were constructed using the Surface
   module within Imaris 9.5.0 software (Bitplane, Switzerland).

RNA Sequencing (RNA‑seq)

   Total RNA was extracted using the TRIzol reagent (Invitrogen, CA, USA)
   following the manufacturer's protocol. RNA purity and quantification
   were assessed using the NanoDrop 2000 spectrophotometer (Thermo
   Scientific, USA). RNA integrity was determined using the Agilent 2100
   Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Subsequently,
   libraries were prepared using the VAHTS Universal V6 RNA‐seq Library
   Prep Kit in accordance with the manufacturer's instructions.
   Transcriptome sequencing and analysis were conducted by OE Biotech Co.,
   Ltd. (Shanghai, China). The libraries were sequenced on an Illumina
   NovaSeq 6000 platform, generating 150 bp paired‐end reads. Differential
   expression analysis was performed using DESeq2.^[ [275]^97 ^] A
   threshold for significantly DEGs was set at Q value < 0.05 and fold
   change > 2 or fold change < 0.5. A heatmap was generated using an
   online platform for data analysis and visualization available at
   [276]https://www.bioinformatics.com.cn (last accessed on 10 July 2023).
   For further analysis, GO enrichment analysis^[ [277]^98 ^] and KEGG
   pathway enrichment analysis^[ [278]^99 ^] of DEGs were conducted to
   identify significantly enriched terms based on the hypergeometric
   distribution. GSEA was carried out using GSEA software.^[ [279]^100 ^]

Primary Microglia Isolated and Culture

   Primary microglia were isolated from 1–10‐day‐old neonatal mice
   (C57BL/6J WT or PD‐L1 KO). The mice were disinfected with 75% ethanol
   for 5–10 min, then transferred to pre‐chilled PBS for cerebral cortex
   dissection. Cortical tissue was washed three times with PBS, minced
   into ≈1 × 1 mm pieces, and digested in 0.25% trypsin with collagenase
   at 37 °C in a shaking water bath for 30 min. Digestion was terminated
   by adding FBS, and the tissue was gently triturated to form a
   single‐cell suspension. The suspension was filtered through a 100 µm
   strainer, centrifuged at 1000 rpm for 5 min, and the cell pellet was
   resuspended in DMEM/F12 medium supplemented with 10% FBS and 50 ng
   mL^−1 recombinant mouse CSF‐1. The cells were plated in T25 flasks,
   evenly distributed by gentle shaking, and cultured at 37 °C in a 5% CO₂
   incubator. After 24 h, the medium was replaced, and cell growth was
   monitored. Half‐medium changes were performed every 2 days. By days
   7–9, cells exhibited stratified growth, and at day 14, flasks were
   placed on a shaker at 220 rpm for 2–4 h to detach microglia. The cell
   suspension was centrifuged, the pellet resuspended in DMEM/F12 medium
   with 10% FBS and 50 ng mL^−1 CSF‐1, and the suspension was replated
   into T25 flasks. After 1–2 h of incubation, nonadherent cells were
   removed, leaving purified microglia for subsequent experiments.

CCK‐8 Assay

   Cell viability was assessed following the protocols provided by the
   Cell Counting Kit‐8 (Dojindo Molecular Technologies, Inc., Kumamoto,
   CK04, Japan). Primary microglia cells were plated in 96‐well plates at
   a density of 2000 cells per well and incubated in FBS‐free medium for
   12 h prior to treatment. Treatments included ASTX029 at concentrations
   of 1 × 10^−9, 0.1 × 10^−9, 0.5 × 10^−9, 1 × 10^−9, 5 × 10^−9 m; cells
   were incubated with the CCK8 solution for 1.5 h at 37 °C. Absorbance
   was measured at 450 nm using a BioRad xMark microplate reader. Cell
   viability was calculated as follows: Cell viability (%) = (absorbance
   of treated sample/absorbance of control) × 100

ELISA

   Cell culture supernatants were thawed on ice and centrifuged at
   1000 rpm for 10 min at 4 °C using a pre‐cooled centrifuge. ELISA kits,
   including IL‐1β (RK00006), IL‐6 (RK00008), and TNF‐α (RK00027) from
   Abclon, were brought to room temperature for approximately 30 min to
   equilibrate reagents and antibodies. The assays were conducted
   following the manufacturer's instructions. Absorbance at 450 nm was
   measured using a microplate reader, and the data were recorded for
   analysis.

Magnetic‐Activated Cell Sorting (MACS)

   RPE‐choroid cells were isolated using magnetic bead sorting. Mice were
   euthanized, and their eyeballs were enucleated to collect RPE‐choroid
   tissues, with four choroids pooled per group. The tissues were placed
   in pre‐chilled RPMI 1640 medium, washed with PBS, and minced into small
   pieces in digestion buffer containing 20 IU mL^−1 papain and 200 IU
   mL^−1 DNase prepared in Hanks’ balanced salt solution (HBSS). The
   samples were digested at 37 °C for 30 min, then centrifuged at 1000 rpm
   for 5 min at room temperature. The supernatant was discarded, and the
   cell pellet was resuspended in HBSS containing DNase and protease
   inhibitors to terminate digestion. The cells were further resuspended
   in 180 µL of pre‐chilled PBS containing 0.5% BSA, followed by the
   addition of 20 µL of anti‐CD11b microbeads (Miltenyi Biotec, Germany)
   and incubated at 4 °C for 15 min. The labeled cells were then loaded
   onto magnetic LS columns placed in a MACS separator. After washing,
   CD11b‐positive cells were eluted by pushing the plunger through the
   column, yielding purified cells suitable for downstream PCR analysis.

Statistical Analysis

   Figure legends specify the sample size (n) for each analysis.
   Statistical results were reported as Mean ± standard error of the mean
   (SEM). To assess data normality, we employed the Shapiro–Wilk test. For
   normally distributed data, homoscedasticity was assessed using the
   Brown–Forsythe test within GraphPad Prism 9.4. Multiple comparisons
   were conducted utilizing either one‐way ANOVA followed by Tukey's
   post‐hoc analysis or two‐way ANOVA followed by Tukey's post‐hoc
   analysis, contingent upon the homogeneity of variances. Differences
   between two groups were assessed using an unpaired two‐tailed Student's
   t‐test. For non‐normally distributed data, unpaired comparisons were
   performed via the Mann–Whitney test. In the context of the laser‐CNV
   model, we employed a linear mixed model to simultaneously consider
   correlations between measurements from the same mouse, under the
   assumption of exchangeable eyes and correlations for repeated
   measurements within the same eye (in cases of repeated laser burns).

Conflict of Interest

   The authors declare no conflict of interest.

Author contributions

   Y.Z., J.J., and Y.L. contributed equally to this work. Y.Z., J.J., and
   Y.L. worked on conceptualization, data curation, formal analysis,
   methodology, and wrote the original draft. X.D., Y.S., and Q.T. worked
   on data curation and formal analysis. L.X. and L.C. worked on
   conceptualization, supervision, funding acquisition, and reviewed and
   edited the writing. All authors revised and agreed on the final version
   of the manuscript.

Supporting information

   Supporting Information
   [280]ADVS-12-2400747-s001.docx^ (2.9MB, docx)

Acknowledgements