Fig 2: Proposed function of erucamide in the retina.a, In healthy retinas, a certain level of erucamide is essential to nourish the neural layers and retinal vessels by maintaining the secretion of neurotrophic and angiogenic factors from microglia. IS/OS, photoreceptor inner segment/outer segment junction; GCL, ganglion cell layer. b, In retinas with retinitis pigmentosa (RP), erucamide is diminished, photoreceptors are degenerated, and retinal vessels are atrophic in the deep plexus. c, In retinitis pigmentosa retinas rescued with erucamide injections, erucamide regulated the secretion of neurotrophic and angiogenic factors from microglia. These microglia were activated through all layers, especially in the ONL at P32, leading to an increase in retinal vessel density in the deep plexus and a slight increase in the thickness of both ONL and INL. d, In retinitis pigmentosa retinas injected simultaneously with erucamide and AAV-TMEM19 or downstream neutralizing antibodies, these rescue effects are blocked, leading to continued photoreceptor degeneration and deep plexus atrophy, similarly to untreated retinitis pigmentosa retinas in b. Figure created in BioRender; Wei, G. https://biorender.com/8d5lglz (2026).

Fig 3: Identifying the binding protein of erucamide in microglia by PAL.a, ‘Fully functionalized’ lipid probes of erucamide, oleamide and palmitamide possessing photoreactive diazirine and alkyne handles. b, Chemoproteomics experiment workflow of PAL for in situ profiling in HMC3 cells. Cells were treated with vehicle (ethanol or EtOH) or a lipid probe for 3 h. Cells were then UV irradiated and lysed, and lipid probe-labeled proteins were conjugated to biotin azide by CuAAC reaction. The biotin-labeled proteins were then enriched with streptavidin beads and trypsin digested, and the digested peptides were labeled with TMT, followed by combination, high-performance liquid chromatography (HPLC) fractionation and LC–MS/MS analysis. c, A plot highlighting proteins selectively enriched by erucamide diazirine probe (Eru.da) over oleamide-diazirine probe (Ole.da) and palmitamide-diazirine probe (Pal.da). x and y axes represent ratios of relative protein enrichment for corresponding probes (x axis: Eru.da/Pal.da, y axis: Eru.da/Ole.da). d, Confirmation of selective engagement of TMEM19 by the Eru.da performed in HEK293T cells transiently expressing TMEM19. Data are representative of three independent experiments with similar results.Source data

Fig 4: TMEM19 is a target protein of erucamide in MPCs.a, Predicted protein structures of mouse and human TMEM19 showed six transmembrane domains according to the AlphaFold Protein Structure Database51. pLDDT, AlphaFold per-residue confidence measure (0–100), corresponding to the model’s predicted local Distance Difference Test computed on Cα atoms (lDDT-Cα) score; higher values indicate greater confidence. b, The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of genes whose regulation by erucamide is lost or attenuated upon TMEM19 knockdown. This is in practice the difference between the effect of erucamide within scrambled cells (Extended Data Fig. 5a) and TMEM19-knockdown cells (Extended Data Fig. 5c) in RNA-seq. ‘Cytokine–cytokine receptor pathways’ is the most significantly deregulated one (adjusted P value = 0.0066). JAK–STAT, Janus kinase-signal transducer and activator of transcription. c, Normalized gene counts of cytokine–cytokine receptor interaction by RNA-seq indicate changes of TNFRSF8, TNFSF13B, TNFSF18 and PDGFRB in angiogenesis-related pathways. Representative genes are colored. d, Results of qPCR analysis of direct effective cytokines and growth factors in MPCs used for RNA-seq (n = 4, two-sided multiple unpaired t-test; for each gene, two comparisons are shown: Scr-C18-pSi versus Scr-Eru-C18-pSi and Scr-Eru-C18-pSi versus siTMEM19-Eru-C18-pSi. TMEM19: P = 0.000290, P < 0.0001; VEGF: NS, P = 0.0306; FGF2: P = 0.0377, P = 0.0374; PDGFA: P = 0.0026, P = 0.0478; IGF1: P = 0.0241, P = 0.0022; BDNF: P = 0.0055, P = 0.0290; TNF: NS, P = 0.000845). Data are representative of three independent biological replicates. e, Schematic of AAV2 viruses and Eru-C18-pSi (214 ng) intravitreal injection in rd10 mice. f, Vascular density quantification of retinal vessels of superficial + intermediate, and deep plexus (SP + IP, and DP). Two-sided multiple unpaired t-test, n = 6 for AAV2-Scr+vehicle, n = 7 for AAV2-Scr+Eru, n = 6 for AAV2-shTMEM19+vehicle, n = 7 for AAv2-shTMEM19+Eru, DP: P = 0.03 for Scr-C18-pSi versus Scr-Eru-C18-pSi, P = 0.01 for Scr-C18-pSi versus AAV2-shTMEM19-Eru-C18-pSi. g, ERG measurements of all scotopic waves and photopic responses to a flash in dark-adapted rd10 mice injected with AAV2-Scr or AAV2-shTMEM19 viruses, then with Eru or vehicle at P12 and P22 (two-sided multiple unpaired t-test, n = 6 per group, P = 0.042624 Scr-Eru-C18-pSi versus shTMEM19-Eru-C18-pSi). h, Representative traces of scotopic B-wave at 50 cd s/m2 intensity by ERG in mice treated with vehicle (left) and erucamide (right), for the same mouse. The right eye was injected with AAV-shTMEM19 viruses, while the left eye was injected with AAV2-Scr viruses. For d, f and g, box plots show the median (center line), 25–75th percentile (box bounds) and minimum to maximum values (whiskers). All individual data points are shown. Asterisks denote statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.Source data