Multi-dimensional decoding of the TSPO signal in animal models of neuroinflammation using fluorescence-conjugated DPA714 Authors: Huanhuan Song, Zhibao Zhu, Limei Xiao, Qiuping Zhao, Chun Li, Aowei Lv, Xintong Yu, Zhuyu Gao, Jie Lin, Youliang Wang, Yuanyuan Zhang, Lijie Zhang, Runyun Li, Shaobo Yao, Mengchao Cui, Yi Lin, Ying Fu, V. Wee Yong, Rui Li Published: 2026 ABSTRACT Neuroinflammation is a core pathological process in multiple central nervous system diseases. Although positron emission tomography (PET) targeting the translocator protein (TSPO) (e.g., ¹⁸F-DPA-714) is widely used for in vivo inflammatory imaging, the cellular heterogeneity and functional states underlying its imaging signals remain unclear, severely limiting its mechanistic research and clinical translational value. This study used the multifunctional fluorescent TSPO probe (Cy5-PEG3-DPA714) to establish a multidimensional decoding framework that enables systematic analysis of the TSPO signal in neuroinflammation across in vivo, tissue, and single-cell levels. Multiple animal models demonstrated the probe’s ability to dynamically monitor acute and chronic neuroinflammation noninvasively in vivo. Histopathology confirmed that the TSPO signal mainly comes from activated Iba1+ microglia. Using the probe’s fluorescence properties, we effectively isolated specific microglial subpopulations with distinct TSPO signaling profiles via flow cytometry. Single-cell analysis revealed that the TSPO signal is specifically enriched in a microglia subpopulation with TSPO probehigh. Importantly, this subpopulation shares features with disease-associated microglia (DAM), exhibiting a robust pro-inflammatory phenotype and significant metabolic reprogramming. Furthermore, we found that TSPO probehigh microglia are highly dependent on the Colony Stimulating Factor 1 Receptor (CSF1R) signal pathway, and selective elimination using the CSF1R inhibitor (PLX3397) reduces neuroinflammation. This study systematically decodes the TSPO signal in neuroinflammation through a multimodal approach, providing novel insights and tools for precision therapeutic strategies targeting specific pathological microglia subpopulations.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12974-026-03829-9. Introduction Neuroinflammation is a common pathological pathway in multiple central nervous system (CNS) diseases, including stroke, multiple sclerosis (MS), and Alzheimer’s disease (AD), and is characterized by continuous activation of glial cells, release of inflammatory mediators, and impaired neural function [1–3]. As innate immune cells of the CNS, microglia play a pivotal regulatory role in disease onset and progression by transitioning between homeostatic and pathogenic functional states [2, 4–8]. Recent advancements in single-cell transcriptomics have been established that pathological stress drives microglia into profound phenotypic shifts. These activated states, known as disease-associated microglia (DAM), are characterized by a unique transcriptional signature coupling pro-inflammatory activity with significant metabolic reprogramming [9]. Dynamically analyzing the spatial distribution and molecular characteristics of DAM is crucial for understanding disease mechanisms and developing targeted therapies. The 18 kDa transporter protein (TSPO), localized to the mitochondrial outer membrane [10], is significantly upregulated in activated microglia or astrocytes, making it a core target for monitoring neuroinflammation [11–13]. TSPO-positron emission tomography (PET) imaging (e.g., ¹⁸F-DPA-714) is a major tool for clinically assessing inflammation and has been widely used to study various neuroinflammatory diseases [14–19], with signal intensity closely correlating with disease severity [20–23] and therapeutic response [21, 24, 25]. Notably, the clinical application of ¹⁸F-DPA-714 in humans requires prior genotyping of the TSPO rs6971 single nucleotide polymorphism (SNP), which directly affects tracer binding [26]. Despite the significant clinical and preclinical value of TSPO-PET, the unknown of biological interpretation limits its translational application : Given TSPO-PET cannot simultaneously identify, isolate, and perform in-depth molecular characterization of the specific cells that take up the tracer [13, 27–29], the precise cellular origin and functional state represented by the TSPO signal remain unclear. To overcome this limitation, we require a strategy that bridges macroscopic imaging with cellular-level biology. Accordingly, this study employs the multifunctional near-infrared (NIR) fluorescent probe (Cy5-PEG3-DPA714), which retains the clinically validated TSPO-targeted capability of the DPA-714 ligand [30]. By establishing an integrated analytical platform that combines in vivo imaging, histopathology, flow cytometry sorting, and single-cell transcriptomics, we systematically decode the TSPO signal in neuroinflammation across spatiotemporal, cellular, and functional dimensions. We found that the TSPO signal mainly comes from activated Iba1+ microglia. Importantly, TSPO probe-labeled (TSPO probehigh) microglia shares features with DAM, exhibiting a robust pro-inflammatory phenotype and significant metabolic reprogramming. Furthermore, we found that TSPO probehigh microglia are highly dependent on the Colony Stimulating Factor 1 Receptor (CSF1R) signal pathway, and selective elimination using the CSF1R inhibitor (PLX3397) reduces neuroinflammation. This research framework aims to reveal the biological basis of elevated TSPO signal in neuroinflammation and to establish translational potential by linking imaging signals to pathogenic cell subpopulations and intervention therapy. Results Specific binding of the TSPO-targeted probe to inflammatory cells We selected the near-infrared (NIR) fluorescent TSPO-targeted probe (Cy5-PEG3-DPA714) (Fig. 1A) (fig. S1A), synthesized based on the second-generation TSPO-PET ligand DPA714. The purity of the TSPO probe was verified by high-performance liquid chromatography (HPLC) and ion-trap mass spectrometry (ITMS) (fig. S1, B and C). To decode the cellular basis of the TSPO signal in neuroinflammation, we performed in situ and pathological validation of TSPO binding under inflammatory conditions using Cy5-PEG3-DPA714. Immunofluorescence analysis showed that the probe signal exhibited a high degree of spatial overlap (co-localization) with the TSPO-specific antibody signal in the BV2 cells, with or without lipopolysaccharide (LPS) stimulation (Fig. 1, B and C), confirming its molecular targeting at the cellular level. Furthermore, we performed pathological colocalization validation in acute intracerebral hemorrhage (ICH) and cerebral ischemia models and found that the probe signal was precisely localized to regions with high TSPO expression (Fig. 1, D and E) (fig. S2). We also found that activated microglia showed significantly higher uptake of the TSPO probe than resting microglia (fig. S2A), indicating that the probe can effectively detect TSPO upregulation in the inflammatory microenvironment. Fig. 1Validation of the Near-infrared TSPO-targeted probe. A The TSPO probe is based on the second-generation TSPO-PET (18F-DPA714) and is prepared by retaining the ligand DPA714 and replacing the radionuclide with the near-infrared fluorescent dye Cy5 for conjugation. B BV2 cells were cultured under standard conditions for 24 hours, with or without LPS (10 ng/ml). Representative immunofluorescence images of TSPO probe (red) and TSPO antibody (green) with DAPI counterstain (blue). Scale bars are 30 μm and 10 μm in the enlarged images. C TSPO probe and TSPO antibody fluorescence intensities were quantified along the white lines in the enlarged images shown in panel B. D The ICH mice were sacrificed after anesthesia at 24 hours postoperatively. Brain tissues were obtained and immunostained with the TSPO probe (red) and the TSPO antibody (green) to assess whether the TSPO probe and antibody could recognize the TSPO signal in activated microglia. The lower-right panel shows a higher-magnification image of the signal indicated by the white arrow. Scale bars are 60 μm and 4 μm at the indicated magnifications. E TSPO probe and TSPO antibody fluorescence intensities were quantified along the white lines in the enlarged images in panel D. F-G Male C57BL/6J mice were used to establish ICH models. Neuroinflammation was visualized and quantified by in vivo chemiluminescence imaging at 24 hours after ICH induction in WT mice, using the IVIS Lumina X5 Spectrum System. Results are expressed as region of interest (ROI) measurements of the brain, analyzed by Loess regression. n = 5 mice per group. Error bars indicate means ± SEM. ****P<0.0001. Two-tailed paired Student’s t-test Validation of the Near-infrared TSPO-targeted probe. A The TSPO probe is based on the second-generation TSPO-PET (18F-DPA714) and is prepared by retaining the ligand DPA714 and replacing the radionuclide with the near-infrared fluorescent dye Cy5 for conjugation. B BV2 cells were cultured under standard conditions for 24 hours, with or without LPS (10 ng/ml). Representative immunofluorescence images of TSPO probe (red) and TSPO antibody (green) with DAPI counterstain (blue). Scale bars are 30 μm and 10 μm in the enlarged images. C TSPO probe and TSPO antibody fluorescence intensities were quantified along the white lines in the enlarged images shown in panel B. D The ICH mice were sacrificed after anesthesia at 24 hours postoperatively. Brain tissues were obtained and immunostained with the TSPO probe (red) and the TSPO antibody (green) to assess whether the TSPO probe and antibody could recognize the TSPO signal in activated microglia. The lower-right panel shows a higher-magnification image of the signal indicated by the white arrow. Scale bars are 60 μm and 4 μm at the indicated magnifications. E TSPO probe and TSPO antibody fluorescence intensities were quantified along the white lines in the enlarged images in panel D. F-G Male C57BL/6J mice were used to establish ICH models. Neuroinflammation was visualized and quantified by in vivo chemiluminescence imaging at 24 hours after ICH induction in WT mice, using the IVIS Lumina X5 Spectrum System. Results are expressed as region of interest (ROI) measurements of the brain, analyzed by Loess regression. n = 5 mice per group. Error bars indicate means ± SEM. ****P<0.0001. Two-tailed paired Student’s t-test In vivo imaging with the TSPO-targeted probe visualizes neuroinflammation dynamics To evaluate the probe’s potential for in vivo monitoring of neuroinflammation, we used NIR imaging to dynamically visualize neuroinflammation across multiple animal models. Firstly, the Cy5-PEG3-DPA714 probe was administered intravenously via the tail vein in healthy adult mice. The results showed that the probe presented stable in the mouse brain throughout the observation period, indicating that the TSPO-targeting NIR probe is suitable for in vivo imaging applications (fig. S3). In the ICH model, after tail vein injection of the probe, TSPO fluorescence intensity in the intracerebral (lesion) hemisphere was significantly higher than in the contralateral (control) side, demonstrating excellent contrast and visualization (Fig. 1, F and G). In the experimental autoimmune encephalomyelitis (EAE) model, the probe rapidly accumulated and reached peak signal intensity within 10 min after tail vein injection, then maintained a relatively stable high-intensity signal throughout the monitoring window (Fig. 2, A and B). In a cuprizone (CPZ)-induced demyelination model, longitudinal imaging revealed dynamic changes in the TSPO signal with disease progression. In vivo probe signal peaked during the acute demyelination phase (week 3) and subsequently attenuated during remyelination phase (week 5), and showed a further significant reduction during the chronic stage (week 12) (Fig. 2, E and F). This rapid pharmacokinetic profile and stable signal-to-noise ratio demonstrate the probe’s potential to precisely capture the spatiotemporal evolution of neuroinflammation in vivo, enabling dynamic decoding of inflammatory activity. Fig. 2In vivo assessment of neuroinflammation utilizing the TSPO-targeted probe. A-B C57BL/6J mice, aged 6 to 8 weeks, were immunized with MOG35–55 combined with CFA and PT. At day 23 post-immunization, normal and EAE mice underwent in vivo imaging with the TSPO probe (0.4 mg/kg) after tail-vein administration. Visualization and quantification of neuroinflammation after TSPO probe administration using the IVIS Lumina X5 Spectrum System. Results are expressed as region-of-interest (ROI) measurements of the brain, analyzed by Loess regression. The number of each group is as follows: n = 7 mice in the Normal group and n = 5 mice in the EAE group. Error bars indicate means ± SEM. ****P<0.0001. two-way ANOVA followed by Bonferroni post hoc test. C-D The EAE mice were sacrificed at day 23 after immunization. The lumbar spinal cord was harvested and analyzed by immunohistochemistry. Representative immunohistochemistry images and quantification of TSPO probe combined with IBA1 or GFAP on the white matter area of the lumbar spinal cord of EAE mice. The lower-right panel shows a higher-magnification image of the signal indicated by the white rectangle. The number of mice per group is as follows: n = 7 mice per group. Error bars indicate means ± SEM. ***P=0.0006. Mann-Whitney U test. The scale bars are 150 μm and 10 μm at the magnifications shown. E-F Healthy male WT mice, aged 6 to 8 weeks, were fed a diet containing 0.3% cuprizone (CPZ) mixed with powdered standard rodent chow. The cuprizone diet was administered for 3, 5, or 12 weeks to induce acute demyelination, remyelination, or chronic demyelination. Visualization and quantification of neuroinflammation by in vivo chemiluminescence imaging. The number of each group is as follows: n = 5 mice per group. Error bars indicate means ± SEM. Week 3 versus Week 5: ****P<0.0001, Week 3 versus Week 12: ****P<0.0001, Week 5 versus Week 12: ****P<0.0001. Two-way ANOVA followed by Bonferroni post hoc test. G-H Representative immunohistochemistry images and quantification of TSPO probe combined with IBA1 or GFAP on corpus callosum sections of the brain of 3-week CPZ mice. The lower-right panel shows a higher-magnification image of the TSPO signal indicated by the white rectangle. The number of each group is as follows: n = 7 mice per group. Error bars indicate means ± SEM. ***P=0.0006. Mann-Whitney U test. The scale bars are 150 μm and 10 μm at the magnifications shown In vivo assessment of neuroinflammation utilizing the TSPO-targeted probe. A-B C57BL/6J mice, aged 6 to 8 weeks, were immunized with MOG35–55 combined with CFA and PT. At day 23 post-immunization, normal and EAE mice underwent in vivo imaging with the TSPO probe (0.4 mg/kg) after tail-vein administration. Visualization and quantification of neuroinflammation after TSPO probe administration using the IVIS Lumina X5 Spectrum System. Results are expressed as region-of-interest (ROI) measurements of the brain, analyzed by Loess regression. The number of each group is as follows: n = 7 mice in the Normal group and n = 5 mice in the EAE group. Error bars indicate means ± SEM. ****P<0.0001. two-way ANOVA followed by Bonferroni post hoc test. C-D The EAE mice were sacrificed at day 23 after immunization. The lumbar spinal cord was harvested and analyzed by immunohistochemistry. Representative immunohistochemistry images and quantification of TSPO probe combined with IBA1 or GFAP on the white matter area of the lumbar spinal cord of EAE mice. The lower-right panel shows a higher-magnification image of the signal indicated by the white rectangle. The number of mice per group is as follows: n = 7 mice per group. Error bars indicate means ± SEM. ***P=0.0006. Mann-Whitney U test. The scale bars are 150 μm and 10 μm at the magnifications shown. E-F Healthy male WT mice, aged 6 to 8 weeks, were fed a diet containing 0.3% cuprizone (CPZ) mixed with powdered standard rodent chow. The cuprizone diet was administered for 3, 5, or 12 weeks to induce acute demyelination, remyelination, or chronic demyelination. Visualization and quantification of neuroinflammation by in vivo chemiluminescence imaging. The number of each group is as follows: n = 5 mice per group. Error bars indicate means ± SEM. Week 3 versus Week 5: ****P<0.0001, Week 3 versus Week 12: ****P<0.0001, Week 5 versus Week 12: ****P<0.0001. Two-way ANOVA followed by Bonferroni post hoc test. G-H Representative immunohistochemistry images and quantification of TSPO probe combined with IBA1 or GFAP on corpus callosum sections of the brain of 3-week CPZ mice. The lower-right panel shows a higher-magnification image of the TSPO signal indicated by the white rectangle. The number of each group is as follows: n = 7 mice per group. Error bars indicate means ± SEM. ***P=0.0006. Mann-Whitney U test. The scale bars are 150 μm and 10 μm at the magnifications shown TSPO-targeted probe reveals microglia as the major contributors to the TSPO signal in neuroinflammation Through multiplex fluorescence co-staining analysis in EAE and CPZ inflammatory models, we identified the cellular source of the TSPO signal. High-magnification immunofluorescence imaging revealed that the TSPO signal of Cy5-PEG3-DPA714 probe predominantly localized to amoeboid Iba1⁺ microglia, with minimal co-localization to GFAP⁺ astrocytes (Fig. 2, C and G). Quantitative analysis further confirmed that the probe signal primarily originated from Iba1⁺ microglia (Fig. 2, D and H), indicating that these cells are the major contributors to the elevated TSPO signal during inflammation rather than GFAP⁺ astrocytes. These findings collectively indicate that the enhanced TSPO signal detected in vivo during neuroinflammation is primarily driven by inflammatory responses within activated microglia. TSPO-targeted probe sorting identifies the specific pathogenic phenotype of microglia subpopulation with TSPO probehigh signal Utilizing the TSPO probe’s compatibility with flow cytometry, we injected Cy5-PEG3-DPA714 into CPZ model and directly sorted TSPO probehigh and probelow cells from brain microglia based on the characteristic of mean fluorescence intensity of the probe (fig. S4). Combined with single-cell RNA sequencing analysis (Fig. 3A), we further decoded the TSPO signal at the cellular functional level. We identified three cell clusters, including Mrc1+Lyz2+ macrophages, Apoe+Fth1+ activated microglia, and P2ry12+Tmem119+ homeostatic microglia (Fig. 3B and fig. S5A), we found that the TSPO probehigh microglia highly expressed a series of hallmark genes (e.g., Jun, Egr1, Ccl3, and Ier5) linked to inflammation (Fig. 3C). Gene set enrichment analysis (GSEA) also revealed that TSPO probehigh microglia exhibited pronounced neurodegenerative and pro-inflammatory transcriptional features, with significant activation of the oxidative phosphorylation (Fig. 3D). Further analysis revealed impairment in lysosomal function, antigen presentation, phagocytic capacity, and lipid metabolism in TSPO probehigh microglia (fig. S5B). The above results show that TSPO probehigh microglia have distinct transcriptional features like disease-associated microglia (DAM) [9]. Fig. 3Single-Cell transcriptomic profiling of TSPO probe-labeled microglia. A We extracted brain microglia of CPZ model after TSPO probe injection. Single-cell suspensions were subjected to fluorescence-activated cell sorting to isolate TSPO probelow and probehigh cell populations for single-cell RNA-seq analysis. B UMAP plot of myeloid cell populations from the mouse brain clustered into microglia (activated and homeostatic states) and macrophages. C Volcano plot showing differentially expressed genes between TSPO probelow and probehigh microglia. A total of 1201 upregulated genes and 2319 downregulated genes were identified (adj. P < 0.05). D The bubble chart shows the KEGG pathway enrichment of upregulated genes in TSPO probehigh microglia. E UMAP plots show the cellular intensity in TSPO probelow and probehigh myeloid cells. F The paired line plot shows the proportions of homeostatic and activated microglia in TSPO probelow and probehigh microglia. The number of each group is as follows: n = 4 mice per group. Error bars indicate means ± SEM. ****P<0.0001 in homeostatic microglia and **P=0.0015 in activated microglia. Two-tailed paired Student’s t-test. G Bubble plot showing the upregulated gene related to inflammatory signal pathways, including TLR, S1P, BTK, as well as metabolism signal pathways (e.g., mTOR, AMPK, HIF1α) in TSPO probehigh microglia Single-Cell transcriptomic profiling of TSPO probe-labeled microglia. A We extracted brain microglia of CPZ model after TSPO probe injection. Single-cell suspensions were subjected to fluorescence-activated cell sorting to isolate TSPO probelow and probehigh cell populations for single-cell RNA-seq analysis. B UMAP plot of myeloid cell populations from the mouse brain clustered into microglia (activated and homeostatic states) and macrophages. C Volcano plot showing differentially expressed genes between TSPO probelow and probehigh microglia. A total of 1201 upregulated genes and 2319 downregulated genes were identified (adj. P < 0.05). D The bubble chart shows the KEGG pathway enrichment of upregulated genes in TSPO probehigh microglia. E UMAP plots show the cellular intensity in TSPO probelow and probehigh myeloid cells. F The paired line plot shows the proportions of homeostatic and activated microglia in TSPO probelow and probehigh microglia. The number of each group is as follows: n = 4 mice per group. Error bars indicate means ± SEM. ****P<0.0001 in homeostatic microglia and **P=0.0015 in activated microglia. Two-tailed paired Student’s t-test. G Bubble plot showing the upregulated gene related to inflammatory signal pathways, including TLR, S1P, BTK, as well as metabolism signal pathways (e.g., mTOR, AMPK, HIF1α) in TSPO probehigh microglia Although the TSPO probehigh microglia predominantly correspond to activated microglia, a small fraction of “non-activated” microglia is also present among TSPO probehigh (Fig. 3, E and F). Interestingly, the TSPO probehigh “non-activated” microglia also exhibited upregulation of inflammatory and metabolic gene sets at higher levels than TSPO probelow “activated” microglia (fig. S5C), which may represent a homeostatic state of specific pathogenic microglia. Notably, further analysis of MS-modifying therapeutics (DMTs)-associated signal pathways and cellular metabolism showed that the TSPO probehigh microglia concurrently exhibited activation of Toll-like receptor (TLR), sphingosine-1-phosphate (S1P), and Bruton’s tyrosine kinase (BTK) pathways, alongside marked upregulation of the mammalian target of rapamycin (mTOR) and hypoxia-inducible factor 1α (HIF-1α) signal pathways (Fig. 3G) In summary, these findings suggest that the microglial subpopulation with TSPO probehigh may possess unique survival dependencies and therapeutic targeting potential. TSPO probehigh microglia are sensitive to CSF1R inhibition Transcriptomic pathway analysis revealed a significantly enhanced Colony Stimulating Factor 1 Receptor (CSF1R) signal in the microglia subpopulation with TSPO probehigh (Fig. 4, A and B), suggesting that CSF1R is a potential therapeutic target. Based on the above result, we hypothesized that the microglia subpopulation with TSPO probehigh are sensitive to CSF1R inhibition. Treatment with the CSF1R inhibitor (PLX3397) in the EAE model not only effectively improved neurological function (Fig. 4C) but also substantially reduced whole-brain TSPO signal uptake (Fig. 4, D and E). Flow cytometry quantification further confirmed that the CSF1R inhibitor selectively eliminated the pathogenic microglia with TSPO probehigh while minimally affecting low-expressing homeostatic microglia, thereby markedly alleviating neuroinflammation (Fig. 4, F to I). These findings confirm that the translational value of the TSPO signal extends beyond serving as a diagnostic marker for neuroinflammation to being a pivotal tool for monitoring the efficacy of precision therapies targeting specific pathogenic microglia. Fig. 4PLX3397 preferentially targets disease-associated microglia with TSPO probehigh. A UMAP plots show the CSF1R pathway score in the TSPO probelow and probehigh microglial cells. B Quantification of CSF1R pathway score in the TSPO probelow and probehigh microglia. The number of each group is as follows: n = 4 mice per group. ****P<0.0001. Two-tailed paired Student’s t-test. C Healthy WT mice, aged 6 to 8 weeks, were treated daily with the CSF1R inhibitor (PLX3397) or vehicle control for 14 days before EAE induction and continued through the end of the experiments. Mice were immunized with MOG/CFA/PT on day 14 and sacrificed on day 23. Single cells were isolated from the brain and spinal cord. Behavioral evaluation was scored daily on a 0-5 scale. The number of each group is as follows: n = 8 mice in the Normal group, n = 8 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Error bars indicate means ± SEM. **P<0.0001. Two-way ANOVA followed by Bonferroni post hoc test. D-E Visualization and quantification of neuroinflammation by in vivo chemiluminescence imaging on day 23 after immunization using the IVIS Lumina X5 Spectrum System. Results are expressed as region-of-interest (ROI) measurements of the brain, analyzed by Loess regression. The number of each group is as follows: n = 7 mice in the Normal group, n = 5 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Error bars indicate means ± SEM. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: ****P<0.0001, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001. two-way ANOVA followed by Bonferroni post hoc test. F Flow cytometry plots show gating of living cells and CD11b+CD45int microglia that stain the TSPO probe. G-I The number of total microglia, TSPO probelow and probehigh microglia were recorded and analyzed by flow cytometry. The number of each group is as follows: n = 8 mice in the Normal group, n = 8 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: P = 0.0696, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001 in panel G. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: P = 0.1230, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001 in panel H. Normal versus Vehicle-EAE: P = 0.0005, Normal versus PLX3397-EAE: P = 0.4001, Vehicle-EAE versus PLX3397-EAE: P = 0.0149 in panel I. one-way ANOVA with Tukey’s multiple comparisons post hoc test PLX3397 preferentially targets disease-associated microglia with TSPO probehigh. A UMAP plots show the CSF1R pathway score in the TSPO probelow and probehigh microglial cells. B Quantification of CSF1R pathway score in the TSPO probelow and probehigh microglia. The number of each group is as follows: n = 4 mice per group. ****P<0.0001. Two-tailed paired Student’s t-test. C Healthy WT mice, aged 6 to 8 weeks, were treated daily with the CSF1R inhibitor (PLX3397) or vehicle control for 14 days before EAE induction and continued through the end of the experiments. Mice were immunized with MOG/CFA/PT on day 14 and sacrificed on day 23. Single cells were isolated from the brain and spinal cord. Behavioral evaluation was scored daily on a 0-5 scale. The number of each group is as follows: n = 8 mice in the Normal group, n = 8 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Error bars indicate means ± SEM. **P<0.0001. Two-way ANOVA followed by Bonferroni post hoc test. D-E Visualization and quantification of neuroinflammation by in vivo chemiluminescence imaging on day 23 after immunization using the IVIS Lumina X5 Spectrum System. Results are expressed as region-of-interest (ROI) measurements of the brain, analyzed by Loess regression. The number of each group is as follows: n = 7 mice in the Normal group, n = 5 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Error bars indicate means ± SEM. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: ****P<0.0001, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001. two-way ANOVA followed by Bonferroni post hoc test. F Flow cytometry plots show gating of living cells and CD11b+CD45int microglia that stain the TSPO probe. G-I The number of total microglia, TSPO probelow and probehigh microglia were recorded and analyzed by flow cytometry. The number of each group is as follows: n = 8 mice in the Normal group, n = 8 mice in the Vehicle-EAE group, and n = 7 mice in the PLX3397-EAE group. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: P = 0.0696, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001 in panel G. Normal versus Vehicle-EAE: ****P<0.0001, Normal versus PLX3397-EAE: P = 0.1230, Vehicle-EAE versus PLX3397-EAE: ****P<0.0001 in panel H. Normal versus Vehicle-EAE: P = 0.0005, Normal versus PLX3397-EAE: P = 0.4001, Vehicle-EAE versus PLX3397-EAE: P = 0.0149 in panel I. one-way ANOVA with Tukey’s multiple comparisons post hoc test Discussion This study established a multidimensional decoding strategy for the transporter protein (TSPO) signal. By integrating longitudinal in vivo imaging, histopathology with specific probe labeling, and single-cell transcriptomics analysis, we systematically revealed the spatial, cellular, and functional implications of the TSPO signal in neuroinflammation. The TSPO signal reflects functional shifts in specific pathogenic microglial subpopulations rather than increased cell density alone. The microglial subpopulation with TSPO probehigh is characterized by pro-inflammatory, hypermetabolic features and sensitivity to the Colony Stimulating Factor 1 Receptor (CSF1R) inhibition. These findings elevate TSPO from a broad inflammatory imaging marker to a key intervention target for neuroinflammatory evolution and metabolic state, filling a long-standing gap in TSPO-positron emission tomography (PET) research. The cellular source of the TSPO signal, particularly the relative contributions of microglia and astrocytes, has been a focal point of discussion in the field [31–33]. Inflammatory stimulation of BV2 cells and pathological co-localization analysis revealed significant enrichment of the TSPO signal in Iba1+ microglia. Consistent with several recent findings [34–36], our results also support the view that the TSPO signal in disease states is highly correlated with microglial burden rather than reflecting cell density or increased TSPO expression. This cell-specific confirmation addresses resolution gaps in PET imaging. While traditional TSPO-PET radiotracers, such as 18F-DPA-714, lack cellular-level resolution, the Cy5-PEG3-DPA714 fluorescent probe developed in this study enables the precise isolation and molecular decoding of phenotypes at the single-cell level through integrated flow cytometry and single-cell RNA sequencing. Furthermore, whereas traditional immunohistochemistry offers detailed cellular images, but it can’t track inflammation over time. In contrast, the Cy5-PEG3-DPA714 probe provides superior near-infrared (NIR) imaging for real-time, high-quality tracking of neuroinflammation dynamics in vivo [37, 38]. This longitudinal monitoring capability is crucial for understanding the biological implications. Collectively, these findings further support prioritizing the TSPO signal as a reliable biomarker for microglia-mediated immune responses, providing a theoretical foundation for subsequent interventions that precisely target microglia [39]. This study reports a key shift in the TSPO signal from cellular localization to functional characterization. Despite long being considered a “marker of glial activation,” recent studies show that this interpretation isn’t universal [13, 32]. Using probe-based sorting and single-cell RNA sequencing, we identified a microglia subpopulation with TSPO probehigh, providing a functional perspective beyond cell density. This subpopulation aligns with disease-associated microglia (DAM) [9] features and temporally and spatially correlates with microglia becoming pathogenic. Pathway analysis shows that TSPO probehigh microglia not only express pro-inflammatory genes but also undergo metabolic reprogramming and functional changes [9, 35, 39–41], explaining why increased TSPO signal often causes severe tissue damage and a transition from defense to damage. Consequently, TSPO imaging not only localizes inflammatory lesions but also serves as an “in vivo metabolic monitor” to evaluate the bioenergetic status of immune cells in the brain. The CSF1R signaling pathway is critical for microglial homeostasis and proliferation. In patients with MS, CSF1R expression is strongly upregulated within microglia surrounding demyelinating cortical white matter lesions, and the level of its ligand, M-CSF, is also elevated [42]. These findings suggest that the CSF1R signaling pathway may be involved in shaping the pathogenic inflammatory environment in MS, and its inhibition is therefore being explored as a therapeutic approach in MS [42, 43]. However, the therapeutic efficacy of CSF1R inhibition in neuroinflammation is strictly stage-dependent. While prophylactic intervention effectively suppresses the initial inflammatory cascade [44–46], late-stage depletion often exacerbates disease severity by disrupting immune homeostasis [47]. Our findings further demonstrate that the TSPO probe specifically labels disease-associated microglia (DAM), with signal intensity increasing at the peak of disease and decreasing during late-stage, thereby effectively indicating the time window for microglia-targeted CSF1R intervention. Furthermore, we found that the CSF1R signaling pathway is particularly active in TSPO probehigh microglia. PLX3397 exhibits potent subpopulation selectivity, preferentially eliminating pathogenic microglia subpopulation with TSPO probehigh while relatively sparing homeostatic microglia (TSPO probelow). This suggests therapeutic efficacy may originate from precise clearance of driver cell populations rather than global microglial ablation, potentially mitigating side effects associated with widespread elimination [48]. Although this study offers detailed insights via multidimensional analysis, some limitations remain. First, animal models cannot fully replicate the complexity and heterogeneity of human neurological disorders, affecting the direct link between imaging and cellular function. Validation of these subpopulations’ presence and clinical relevance in humans is needed. Second, Cyanine 5 (Cy5) dye’s tissue penetration limits its use in large animals or deep human brain structures. As summarized in Supplemental Table 1, traditional TSPO radiopharmaceuticals (e.g., 18F-DPA-714) remain the benchmark for clinical neuroinflammation imaging due to their unrestricted 3D localization and deep-tissue penetration. In contrast, the substantial molecular dimensions of our Cy5-conjugated probe inherently limit its reach into deep-seated neuroanatomical structures, which may lead to the under-representation of activated microglia in those regions. To address this, future research should develop second near-infrared region (NIR-II) probes with improved tissue penetration, combined with PET/MRI, to provide deeper clinical insights. Lastly, the unclear dynamic change in microglia between high and low TSPO expression across disease stages is key for future validation and expansion. From a clinical translation perspective, this study decoded the biological significance of the TSPO signal in neuroinflammation across imaging, cellular, and molecular dimensions using a multifunctional probe. This advances TSPO from a coarse inflammatory marker to a precise indicator of cellular states. Enlarged TSPO expression correlates with metabolic reprogramming and pathogenic transformation of microglia, demonstrating the feasibility of targeted intervention. Materials and methods Animals C57BL/6J mice (aged 6–8 weeks) were obtained from the Beijing SPF Biotechnology Co., Ltd. Mice were maintained under pathogen-free conditions with a 12-hour light/dark cycle, ad libitum access to food and water, and controlled temperature and humidity. All procedures were approved by the Institutional Animal Care and Use Committee of Fujian Medical University and conducted in accordance with the US Public Health Service’s Policy on the Humane Care and Use of Laboratory Animals. Investigators performing staining and analysis were blinded to experimental groups. EAE induction Female mice, aged 6–8 weeks, were injected subcutaneously to actively induce EAE, following the methods described in previous publications [37]. In short, all animals received a vaccination in two areas on their backs with an emulsion containing 200 µg of myelin oligodendrocyte glycoprotein peptide 35–55 (MOG35-55) (Genescript, Piscataway, NJ) and 5 mg/ml of heat-killed Mycobacterium tuberculosis H37Ra (Difco, Detroit, MI). On days 0 and 2 post-immunization, 200 ng of pertussis toxin (List Biologic, Campbell, CA) dissolved in phosphate-buffered saline (PBS) was administered intraperitoneally (i.p.). Daily evaluations of clinical EAE symptoms ranged from 0 to 5, with 0 indicating no clinical signs; 0.5, partial tail paralysis; 1, fully limp tail; 1.5, limp tail and waddling gait; 2, paralysis of one hind limb; 2.5, paralysis of one limb and weakness of another limb; 3, paralysis of both hind limbs; 3.5, paralysis of one forelimb and both hind limbs; 4, paralysis of the trunk; and 5, death. CPZ-induced demyelination model Male mice were fed a diet containing 0.3% CPZ (Sigma) mixed with powdered standard rodent chow. The Cuprizone diet was administered for 3, 5, or 12 weeks to induce demyelination [49]. Administration of PLX3397 PLX3397 (Abmole, CA, USA) was dissolved in PBS containing 5% dimethyl sulfoxide (DMSO) and 25% polyethylene glycol 300 (PEG300). Mice were given daily doses of either vehicle PBS or PLX3397 (40 mg/kg) by oral administration for 14 days before the induction of EAE. Treatment was continued until the completion of the experiments. Synthesis and administration of the TSPO-targeted probe The TSPO-targeted probe (Cy5-PEG3-DPA714) [37] was synthesized via a classic carbodiimide-mediated amide coupling of the commercially available TSPO ligand DPA714 (MedChemExpress, HY-122607) with the fluorescent dye Cy5-PEG3-NH2 (Aladdin, B596095). Briefly, DPA714 was activated with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride and hydroxybenzotriazole in anhydrous N, N-dimethylformamide under a nitrogen atmosphere. After stirring for 3 h at room temperature, a solution of Cy5-PEG3-NH2 and N, N-diisopropylethylamine in N, N-dimethylformamide was added. The reaction mixture was stirred in the dark for 18 h. The crude product was purified and lyophilized to obtain the final probe as a dark blue solid. The purity and identity of the TSPO-targeted probe were confirmed by analytical high-performance liquid chromatography (HPLC) and ion-trap mass spectrometry (ITMS). The TSPO-targeted probe was prepared as a 4 mg/mL stock in dimethyl sulfoxide (DMSO) and subsequently diluted in 0.9% sterile saline to a working concentration of 0.08 mg/mL. Mice were administered a single 100 µL bolus via tail vein injection, achieving a dose of 0.4 mg/kg in a final vehicle consisting of 2% (v/v) DMSO and 98% (v/v) saline. IVIS optical imaging To enable live tracking of inflammation in the brain tissue, we used an NIR fluorescence probe (Cy5-PEG3-DPA714) that binds to TSPO. Briefly, mice were deeply anesthetized with inhaled isoflurane, and their head hair was carefully removed. The fluorescence imaging mode was selected, and the fundamental parameters for autofluorescence imaging were obtained using the IVIS Lumina X5 Spectrum System (PerkinElmer, Waltham, Mass). Anesthesia (isoflurane) was administered via a nasal cone, followed by an injection of 0.4 mg/kg of the TSPO probe into the lateral tail vein. The camera was exposed at 10-minute intervals for a total of 60 min following the intervention. Quantification of bioluminescence emitted from a specific region of interest (ROI) was performed by measuring the radiance efficiency in units of (p/s/cm2/sr)/(µW/cm2) using the Living Image software (Caliper Life Sciences, Hopkinton, MA). Statistical analysis Statistical analysis was performed using GraphPad Prism 9.0 software (La Jolla, CA, USA). Investigators performing pathological staining and analysis were blinded to the experimental groups during the experiments. Sample sizes were deemed adequate based on the magnitude and consistency of measurable differences between groups. All data are expressed as mean ± standard error of the mean (SEM). Normality was assessed using the Kolmogorov-Smirnov or Shapiro-Wilk tests. For multiple-group analysis, one-way analysis of variance (ANOVA) with Tukey’s post hoc multiple comparisons test was used. For experiments involving two independent variables, two-way ANOVA with Bonferroni’s multiple comparisons test was used. For analysis with 2 groups, the unpaired t-test, paired t-test, or the Mann-Whitney U-test was used depending on the results of normality testing. All results represented two-sided tests. P < 0.05 was considered statistically significant for all measures. The n values represent the number of cases, ROIs, or cells quantified in each group. The n values are indicated in the figure legends for easy association with their respective test. See the figure legends for detailed statistical information corresponding to each experiment. Supplementary Information Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2