Ten-week-old C57bl6-J mice were housed in an enriched environment at the INMED animal facility, maintained on a 12 h light / 12 h dark cycle with controlled temperature (23 ± 2°C), and food and water were provided ad libitum. A stereotaxic procedure was performed using an aseptic technique. Briefly, 30 min before the surgery, buprenorphine (0.03 mg/kg) was administered intraperitoneally (i.p), after which the mice were anesthetized with 4% isoflurane vaporized in air, along with an additional 0.3% oxygen enrichment. Surgical anesthesia was maintained using 2% isoflurane with 0.3% oxygen, while the mice were positioned in a stereotaxic frame (David Kopf Instruments, Tujunga, CA). The body temperature was maintained at 37 ± 2°C using a heating pad (Harvard Apparatus). The stereotaxic coordinates were selected to allow injection above CA1 (reference to bregma, anteroposterior axis: −1, laterality: −1.2, verticality: −1.5). The skull was drilled, leaving the dura intact. Using a microinjector (micropump 4, WPI), 1.5 microliters were injected at a rate of 100 nL/min with a NanoFil® D needle (WPI). After the injection, the needle was left in place for an additional 15 min to prevent backflow of the liquid. Altogether, the total duration of the surgery did not exceed 40 min. The skin was sutured, and the animals were placed in the post-operative room of the facility. The brains were then either perfused and fixed for immunochemistry or freshly harvested for western blot analysis 1 week after the injection.

A controlled cortical impact (CCI) procedure was performed using an aseptic technique. Buprenorphine (0.03 mg/kg) was injected intraperitoneally (i.p) 30 min before the surgery. The mice were then anesthetized using 4% isoflurane vaporized in air, along with an additional 0.3% oxygen enrichment. During the surgical procedure, anesthesia was maintained with 2 to 2.5% isoflurane in air, along with 0.3% oxygen. The mice were positioned in a stereotaxic frame (David Kopf Instruments). The body temperature was monitored throughout the procedure using a rectal probe and was maintained at 37 ± 2°C with a heating pad (Harvard Apparatus). A unilateral craniotomy was performed over the right parietal cortex, within the boundaries of the bregma and lambda while leaving the dura intact, using a high-speed drill. The CCI procedure was performed using a Leica impactor with the following parameters: tip diameter 3 mm, speed 6 m/s, depth 1.5 mm, and duration 200 msec. The impact perpendicularly compressed the curvature of the sensorimotor cortex. The animals were allowed to recover on the heating pad before their transfer to the post-surgical room. Before the start of the experiments, the animals were randomly assigned to subgroups, namely sham-vehicle +/− BrainFectIN, sham-bumetanide +/− BrainFectIN, CCI-vehicle +/− BrainFectIN, and CCI-bumetanide +/− BrainFectIN. Bumetanide injections were performed i.p twice daily for a one-week period at a 2 mg/kg concentration.

Each animal was checked daily after the CCI procedure for weight loss and general condition, according to the protocol validated by our local committee (Apafis #2797). For blood collection, we used the mandibular vein to collect blood from the sham animals and animals with TBI, and a heart puncture was performed when a larger volume was needed from the transfected animals. After applying a local sanitizer to the cheek, the mandibular vein was pierced perpendicularly with a lancet. Blood drops were collected in a blood tube, not exceeding the maximum volume. A small pressure was then applied to stop the bleeding. We used an appropriate lancet (blood lancet Nahita FM024/60425012), and the maximum blood volume allowed for collection was adhered to Table 1. For the heart puncture, after deep anesthesia using 4% isoflurane, the animal was placed in the supine position. After applying a local sanitizer to visualize the depression of the xiphoid appendix, we used a 25-gauge needle. The needle was inserted gently into the xiphoid hollow at an angle of 30°–45° toward the top. A larger volume of blood was withdrawn; if the blood did not flow, the needle location was slightly adjusted.

The exosomes were isolated from 100 microliters of the blood serum, after the addition of 5 μL protease and phosphatase inhibitor (Pierce Protease and Phosphatase Inhibitor Mini Tablets, EDTA-FREE, Invitrogen A32961). The total exosomes were first isolated from the blood serum using the “Total Exosome Isolation Reagent (from serum)” (Thermo Fisher Scientific, Invitrogen 4478360) according to the manufacturer’s instruction. After the collection, the exosomes were stored at 2°C to 8°C for up to 1 week or at −20°C for longer-term storage. The neuronal fraction was enriched by immunoprecipitation using the L1CAM antibody (eBio5G3 (5G3), Thermo Fisher Scientific, #14-1719-82) but was probed with CD63 to avoid interfering with the KCC2 staining.

The mice were deeply anesthetized with an intraperitoneal injection of ketamine/xylazine and then transcardially perfused with cold phosphate-buffered saline (PBS 0.01 M), followed by a 3% paraformaldehyde solution (AntigenFix, Diapath). The brains were post-fixed overnight in 3% paraformaldehyde at 4°C and then coronally sliced using a Leica VT1200S Vibratome. A total of 60 μm-thick sections were permeabilized and blocked in PBS with 0.3% Triton X-100 and 5% normal goat serum (NGS) for 1 h at room temperature. Then, the sections were stained with primary antibodies diluted in PBS with 5% NGS and 0.1% Triton X-100 at 4°C overnight using anti-Dsred (Takara Bio Clontech, living colors polyclonal antibody, 632496), anti-NeuN (Merck Millipore; MAB377), and anti-Gad67 (Merck Millipore, MAB5406). After washing with PBS, the slices were incubated with the corresponding Alexa Fluor 488 and 555-conjugated secondary antibodies, diluted in PBS (1/500, Thermo Fisher Scientific, Invitrogen A11001), for 2 h at room temperature and finally counterstained for 1 min with Hoechst 33258 (10 μg/mL in PBS, Sigma-Aldrich, 94403). The sections were mounted onto Superfrost Plus glass slides in Fluoromount-G Mounting Medium. For each section, serial images were taken using a fluorescence microscope equipped with an apotome module and 20× or 40× objectives.

The animals were euthanized by decapitation after deep isoflurane anesthesia. The hippocampi were quickly dissected out, flash-frozen in liquid nitrogen, and stored at −80°C. The brain tissues were homogenized in RIPA buffer (50 mM Tris–HCl pH 8, 150 mM NaCl; SDS 0.1%; Deoxycholic Acid 0.5%; 1% Triton X-100) supplemented with a Protease/Phosphatase Inhibitor Tablet (Thermo Fisher Scientific). The proteins were run on a polyacrylamide gel (Bolt 4–12% Bis-Tris plus, Invitrogen by Thermo Fisher Scientific) and transferred to a nitrocellulose membrane (GE Healthcare Life Science). Following the application of a blocking solution containing Tris-buffered saline, 0.1% Tween and 5% bovine serum albumin (BSA), the membranes were incubated overnight at 4°C with primary antibodies diluted in a blocking solution (Tris-buffered saline/ 0.1% tween/ 2.5% BSA): anti-Dsred (Clontech, Living Color Ds-Red Polyclonal Antibody, 632,496), anti-KCC2 [home-made antibody, (Ludwig et al., 2003)], and Tubuline-β3 (Biolegend, 802001). Horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG (Agilent Dako) was used as secondary antibodies, diluted in 5% bovine albumin at room temperature for 2 h. Bands were then detected using SuperSignal West Pico (Thermo Fisher Scientific, #34080) and analyzed with the G:Box imaging system (Syngene). Quantifications were performed using the Gel Plot Analyzer plugin in ImageJ.

A 20 mM stock solution of bumetanide (Sigma-Aldrich, B3023) was prepared by dissolving 36.4 mg of the powder in 1 mL of absolute ethanol. The injected solution was created by mixing 40 μL of the stock solution with 4 mL of 1X PBS. A dose of 26.7 μL per gram of the animal body weight was then injected intraperitoneally (2 mg/kg), twice daily (9 AM and 5 PM). A vehicle solution was prepared using the same procedure but without the bumetanide powder to maintain the volume and diluent.

The KCC2-mCherry construct was created by inserting a ubiquitin promoter in place of the CMV promoter in the pmCherry vector (Clontech) to enhance transgene expression under in vivo conditions. The vector and transgene were sequenced before cloning, and the full plasmid was also sequenced.

All mean values were presented with the standard deviation (SD). Normality was tested for each distribution, with a significance level set at 5%. Two-tailed Student’s t-test, the Mann–Whitney U test, or one-way ANOVA was used accordingly, using Prism software (GraphPad Software, Inc., La Jolla, CA, United States). Box plots reported the median, interquartile range, and total range of the data, with statistical significance represented as follows: ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Our initial investigation aimed to determine whether KCC2 can be detected in blood serum. A previous study successfully identified it in cerebrospinal fluid (Duarte et al., 2013), making it essential to first ascertain its presence in blood serum. We collected 0.2 mL of blood from the mandibular vein of the mice and isolated the serum by centrifugation at 2,000 g for 30 min at 4°C. Subsequently, we conducted a western blot under denaturing conditions using the clarified serum. As shown in Figure 1A, we clearly observed the full-length KCC2 band at 140 kDa in all tested samples (n = 30). Utilizing our custom-made antibody specific to the N-terminus of KCC2 (Ludwig et al., 2003), it was challenging to definitively identify degraded forms of KCC2. Nonetheless, after analyzing the gel band profiles with red Ponceau, we found no discernible difference compared to the brain tissue samples (Figure 1B, data not shown). This intriguing result prompted us to investigate whether the secretion of KCC2 into the blood was constitutive or dependent on specific conditions.

In the subsequent step, we aimed to investigate the presence of KCC2 in the circulating exosomes, originating either from the neuronal cells or from the entire cell population, as these vesicles are known to travel long distances. We utilized a specific extraction kit to purify the total exosomes from the peripheral serum through immunoprecipitation. The isolation of the total exosomes was confirmed by detecting the pan-exosomal marker CD63 (Hashkavayi et al., 2020) in each sample using western blot analysis (Figure 1B). The figure illustrates both the glycosylated form (63 kDa) and non-glycosylated form (27 kDa) of the CD63 protein, with a predominance of the glycosylated form at 63kd.

To confirm the neuronal origin of KCC2, we proceeded with the purification of the neuronal exosomes from the total exosome samples once again, this time using immunoprecipitation. To verify the neuronal origin of KCC2, we conducted western blotting with the neuronal exosome marker L1CAM (CD171) (Pulliam et al., 2019; Figure 1C).

KCC2 is primarily expressed in the central nervous system (Blaesse et al., 2009), making the discovery of its full-length form in circulating blood particularly intriguing. To verify the CNS origin of the molecule, we chose to genetically modify the brain neuronal cells through in vivo transfection using a BrainFectIN® agent. Using a stereotaxic approach, we injected DNA encoding C-terminal tagged mCherry-KCC2. One week after the injection, we harvested the brains and performed co-immunoprecipitation. Our previous work with BrainFectIN® had already demonstrated potent and sustained modification of CNS cells (Scala et al., 2019). The post hoc analysis 1 week after the injection revealed transgene expression through immunohistochemistry, using a specific antibody against the mCherry tag at the injection site, CA1 (Figures 2Aa), and in the dentate gyrus (DG) (Figures 2Ab). High magnification of the CA1 region of the hippocampus (Figure 2B) confirmed the robust expression of the mCherry-tagged KCC2. We further confirmed the neuronal expression of the transgene in the principal neurons using the neuronal marker NeuN (Figure 2C) and in the interneurons using the Gad67 antibody (Figure 2D).

At the protein level, we verified the presence of the mCherry-tagged KCC2 in the ipsilateral hippocampal extract. This region was selected due to its targeting by the injection (1.76 ± 0.20, n = 5 vs. 1.0 ± 0.15, n = 5, p = 0.03, Figures 2E,F). As previously demonstrated by immunohistochemistry, there was no diffusion of the transgene to the contralateral side (0.83 ± 0.17, n = 5 vs. 1.00 ± 0.13, n = 5, p = 0.84, Figures 2G,H).

In this context, we were able to identify the 35 kDa-shifted KCC2 band corresponding to the tagged KCC2 in the pan-exosomes CD63-containing exosomes (Figure 3A). The CNS origin was further reinforced and confirmed using L1CAM immunoprecipitation of the neuronal exosomes, confirming the neuronal enrichment of the mCherry-KCC2 (Figure 3B). Furthermore, by combining the KCC2 antibody, which recognizes the N-terminus part of the protein, and the mCherry antibody, which recognizes the mCherry molecule at the C-terminal part of KCC2, we conclusively confirmed that full-length KCC2 was embedded in the exosomes, particularly the neuronal exosomes. The combination of these two antibodies targeted both the N- and C-terminus regions of full-length KCC2, as indicated by the bands at 140 and 175 kDa in Figure 2.

It has been previously proposed and demonstrated that there is a transient loss of KCC2 expression after trauma (Medina et al., 2014). Utilizing the controlled-cortical impact model, we observed a robust decrease in the KCC2 levels in the early days after the trauma, followed by complete recovery after 1 week (Goubert et al., 2019). Considering these findings, we also investigated the impact of TBI on the exosomes, both total and neuronal, in the blood serum using the same paradigm.

At 7 days post-trauma (dpt), we clearly observed a significant decrease in the normalized total expression of KCC2 in the blood serum (sham 100% ± 14 vs. CCI 59.8% ± 7, *p = 0.015, n = 4, Figure 4A), as well as in the total exosomes (sham 100% ± 16 vs. CCI 59% ± 8, *p = 0.015, n = 4, Figure 4B) and the neuronal-enriched fraction of the exosomes (sham 100% ± 18 vs. CCI 60% ± 8, *p = 0.016, n = 4, Figure 4C).

Considering the possibility that this loss of expression might persist for a longer duration in the blood compared to the CNS expression, we conducted the same analysis 1 month after the trauma. This timeframe was chosen because we had previously demonstrated that KCC2 expression in the hippocampus recovered to sham levels in the same animal model, while the subjects exhibited cognitive dysfunction, particularly depression-like behavior (Goubert et al., 2019).

Once again, we observed a substantial loss of KCC2 expression in the blood serum (sham 100% ± 8 vs. CCI 64% ± 11, *p = 0.03, n = 4, Figure 4D), in the total exosomes (sham 100% ± 8 vs. CCI 70% ± 13, *p = 0.016, n = 4, Figure 4E), and in the neuronal exosomes (sham 100% ± 7 vs. CCI 70% ± 7, *p = 0.016, n = 4, Figure 4F). These findings support the hypothesis of a sustained and permanent decrease in KCC2 expression in exosomes following TBI.

Previous research, including studies by Goubert et al. (2019) and Pellegrino et al. (2011), has shown that bumetanide, a sodium–potassium–chloride transporter antagonist, exerts a powerful effect in preventing cell death in interneurons and principal cells. In addition, it has been found to prevent rapid and transient degradation of KCC2 in various trauma models (Hu et al., 2017; Medina et al., 2014) and epilepsy (Kourdougli et al., 2017). Given its ability to efficiently restore KCC2 expression in neuronal cells, we sought to investigate its potential impact on KCC2 modulation in exosome content.

To explore this, we administered intraperitoneal injections of 2 mg/kg bumetanide twice daily for 1 week. Subsequently, we collected blood serum to assess KCC2 expression. One week after the trauma (7dpt), the KCC2 expression levels were significantly improved compared to the CCI condition, resembling the levels seen in the sham group. This improvement was observed both in the blood serum (CCI 59% ± 7 vs. CCIbum 95% ± 10, *p = 0.028, Figure 5A), total exosomes (CCI 59% ± 8 vs. CCIbum 90% ± 9, *p = 0.028, Figure 5B), and neuronal exosomes (CCI 60% ± 8 vs. CCIbum 89% ± 4, *p = 0.028, n = 4, Figure 5C).

Later, our focus shifted to the long-term effects of bumetanide on the persistence of KCC2 expression and its correlation with the prevention of depressive-like behavior, which has been previously observed. To investigate whether the treatment could also modify KCC2 expression in the exosomes, we assessed the KCC2 levels in the blood 1 month after the treatment. Notably, we observed differences in KCC2 expression in the blood compared to the CCI condition in the blood serum (CCI 64% ± 11 vs. CCIbum 90% ± 3, *p = 0.028, n = 4, Figure 5D), total exosomes (CCI 70% ± 13 vs. CCIbum 1.01% ± 3, *p = 0.028, n = 4, Figure 5E), and neuronal exosomes (CCI 70% ± 7 vs. CCIbum 96% ± 8, *p = 0.028, n = 4, Figure 5F). These findings suggest that KCC2 expression in exosomes could be a relevant marker for detecting depressive-like behavior, consistent with the functional changes observed in our previous study (Goubert et al., 2019).