Astrocyte CD9-tGFP reporter mouse shows astrocyte EV cargo preferentially enriched at synaptic mitochondria

The work establishes the first astrocyte-specific CD9-tGFP EV reporter mouse (Aldh1l1-Cre; CD9-tGFPfl/fl), enabling selective labeling of astrocyte-derived CD9-positive EVs in vitro and in vivo. The TIGER (CD9-tGFP) reporter mouse was a Cre-dependent general EV-labeling tool, but astrocyte-specific applications had not been reported; CD63 and CD81 reporters mainly label EV subsets enriched in those two tetraspanins. Immunoblotting detected a 50 kDa CD9-tGFP fusion protein recognized by both anti-CD9 and anti-tGFP antibodies only with Cre; brain sections without Cre showed only weak background fluorescence; 13.2% ± 1.6% of brain EVs and 89.3% ± 2.2% of primary astrocyte EVs were CD9-tGFP positive.

Astrocyte-derived CD9-tGFP EV cargo reaches vascular and neuronal compartments in vivo and is internalized by cerebellar Purkinje cells. Prior tracking of astrocyte EVs largely relied on lipophilic dyes, which are prone to non-specific transfer and diffusion and cannot resolve cellular origin or subcellular location in vivo. 65% ± 7% of CD9-tGFP puncta lay within 5 μm of GFAP-positive processes versus 10% ± 3% near Iba1-positive microglia (paired t-test, p < 0.0001, six independent brain sections); 67% ± 14% of Purkinje neurons contained internalized CD9-tGFP puncta, mean 4.2 ± 0.7 puncta/cell, with puncta number positively correlated with soma area (r = 0.579, p = 0.0005) while density was not (r = −0.040, p = 0.83).

In neuronal (N2a) cells, astrocyte EV cargo is preferentially enriched at mitochondria, whereas endothelial cells show a bimodal distribution, indicating cell-type-specific uptake and transport routes. Previous studies largely stopped at observing EV uptake by neurons and did not characterize organelle-level fate or compare recipient cell types. Mean edge-to-edge distance between internalized CD9-tGFP and nearest mitochondria was 93 nm in N2a cells versus 1128 nm in endothelial cells (Mann–Whitney U, p < 0.0001); 60% of N2a EVs were in direct contact (0 nm) versus 22% in endothelial cells; Monte Carlo randomization (10,000 iterations) showed 77% of N2a EVs within 100 nm versus 8.5% expected, and 22% in endothelial cells versus 1.4% expected; the endothelial distance distribution had a bimodality coefficient of 0.764 and a 2-component Gaussian mixture fit best (ΔBIC = −17.2).

In vivo, astrocyte-derived CD9-tGFP cargo preferentially co-localizes with synaptic mitochondria. The association of astrocyte EV cargo with recipient neuronal mitochondria had not previously been documented in vivo. Non-synaptic mitochondria were 4.5-fold more abundant than synaptic mitochondria, yet CD9-tGFP-labeled synaptic mitochondria density was 3-fold higher than labeled non-synaptic mitochondria; immunoblotting showed stronger Tom20 in the non-synaptic fraction but stronger CD9-tGFP in the synaptic fraction; endogenous CD9 was readily detected in whole-brain lysate but only at very low levels in mitochondrial fractions.

The model is intended for researchers who need to track astrocyte-derived CD9-positive EVs in mice, applicable to imaging and biochemical analysis in cortex, hippocampus, and cerebellum under baseline physiological conditions, and to comparisons across disease stages, brain regions, or therapeutic interventions. Because the reporter is built on CD9-tGFP, it covers CD9-carrying EV populations, including EVs released predominantly via plasma membrane budding; the authors note it is well suited to visualizing stress-responsive and ceramide-mediated EV release. Beyond in vivo imaging, the authors propose enriching CD9-tGFP-positive EVs by anti-tGFP immunoaffinity capture or fluorescence-based particle sorting for proteomic, lipidomic, or transcriptomic profiling.

The authors state that current imaging cannot distinguish physical association from fusion of EV cargo with the outer mitochondrial membrane, nor whether rapid transport occurs directly from the plasma membrane or via endocytosis; whether CD9-tGFP-positive EV cargo also interacts with other compartments such as ER, endosomes, lysosomes, or Golgi remains to be determined. At the vasculature, imaging cannot resolve whether EVs were taken up by endothelial cells or entered the vascular lumen. About 29% of endothelial particles lay distant from mitochondria, and their destination is undefined. In addition, the functional coupling between Purkinje neurons and Bergmann glia cannot be inferred causally from correlative data, which the authors describe as "correlative." This was a full-text read, but specific values in figures and supplementary figures should be checked against the original figures.