Reagents.

Minocycline was purchased from Sigma-Aldrich (St. Louis, MO, catalog no. M9511). Several doses were studied previously in rats after stroke, and 45 mg/kg via an intraperitoneal injection (ip) was proven to be neuroprotective. The dose used in the current study was previously reported (Zhao et al. 2011). N-[2-(5-hydroxy-1H-indol-3-yl) ethyl]-2-oxopiperidine-3-carboxamide (HIOC) is a derivative of N-acetyl serotonin that selectively activates the TrkB receptor. [N2–2-2-oxoazepan-3-yl amino] carbonyl phenyl benzo (b) thiophene-2-carboxamide (ANA 12) is a selective TrkB antagonist. Both drugs were purchased from Tocris Bioscience (catalog nos. 5961 and 4781) and injected intracerebroventricularly. Other reagents are mentioned in the text where appropriate.

Experimental ICH in rats and study protocol.

A total of 150 Sprague-Dawley (S-D) rats (250–350 g, from the Third Military Medical University) were used in the study. All methods were performed in accordance with animal care guidelines at the Animal Ethics Committee of the Third Military Medical University, and the Southwest Hospital of the Third Military Medical University approved all experimental protocols [SYXK(Yu)2012-0012]. ICH in rats was induced via an injection of autologous blood from the femoral arteria as previously described (Jiang et al. 2017). Briefly, animals were placed on a feedback-controlled heating pad after anesthetization using pentobarbital (40 mg/kg ip). Animals were stereotaxically positioned, and a 1.0-mm-diameter cranial burr hole was drilled. A 29-gauge needle was inserted into the right caudate nucleus (coordinates: 0.2 mm anterior, 5.5 mm ventral and 3.5 mm lateral to the bregma). Each rat was injected with 100 μl of autologous arterial blood using a microinfusion pump at a constant rate of 10 μl/min. Arterial blood pressure, glucose level, $\mathrm{P}{\mathrm{a}}_{\mathrm{O}}{}_{{}_2}$ and $\mathrm{P}{\mathrm{a}}_{\mathrm{C}\mathrm{O}}{}_{{}_2}$ were monitored during the procedure. Minocycline was injected ip 15 min after surgery and 12 h later. For long-term investigations, minocycline was injected twice daily for 7 days (ip). ANA 12 and HIOC were injected intracerebroventricularly 3 h after ICH.

Forty-six S-D rats were randomly divided into four groups (Sham, Sham + minocycline, ICH + vehicle, and ICH + minocycline) to detect the long-term (28–32 days) effects of minocycline using behavior testing and magnetic resonance imaging (MRI) image changes. Six rats died during the ICH and MRI anesthesia procedure.

Twenty-four rats were used for immunohistochemistry detecting the effects of minocycline on microglial polarization, neurogenesis and TrkB/BDNF expression after ICH.

We randomly divided 80 rats into four groups: ICH + vehicle, ICH + minocycline, ICH + minocycline + ANA 12, and ICH + HIOC. Among them, in each group six rats were used for PCR detection, six for Western blot assays, and six for immunofluorescence. We lost eight rats during the procedure.

Intracerebroventricular injection.

Intracerebroventricular injections were used For ANA 12 and HIOC administration as reported previously (Tang et al. 2015). Briefly, a small burr hole was drilled into the left skull relative to bregma: 1.5 mm posterior; 1.0 mm lateral. A 10-μl Hamilton syringe filled with ANA 12 (2 μg in 5 μl of DMSO) or HIOC (6 μg in 5 μl of DMSO) was stereotaxically inserted into the left lateral ventricle through the burr hole to a 4.0-mm depth. The microinjection rate was 0.5 μl/min.

Morris water maze test.

All animals were subjected to a modified Morris water maze (MWM) test 28 days after ICH and drug administration (Ahn et al. 2013). Briefly, rats were placed in a metal pool (50 cm in depth, 200 cm in diameter) filled with water to find the submerged platform within 90 s. Each rat performed eight trials daily for 4 days, and the latency time was monitored and averaged across the trials. Acquisition training was performed over 5 days using a random set of start locations in the four quadrants. The animals were tested for spatial memory retention 24 h after the final trial. Visible platform trials were performed on day 6 to assess visual acuity. The swim distance in the target quadrant and percentage of time spend in the quadrant where the platform used to be were calculated in the study (n = 10).

5-Bromo-2-deoxyuridine labeling.

5-Bromo-2-deoxyuridine (BrdU, Sigma) in 0.9% saline was administered (100 mg/kg ip twice daily for 3 days) as previously described with slight modifications (Lei et al. 2015).

Brain slice preparation and immunofluorescence.

Rats were euthanized 3 days after ICH (n = 6). Frozen brain slices (18 μm) and immunofluorescence were performed as described previously (Tang et al. 2016). Slices were blocked with 5% bovine serum albumin and incubated overnight at 4°C with primary antibodies against BrdU (goat, Abcam, 1:500), Iba-1 (rabbit, WAKO, 1:1,000), MAP2 (mouse, Boster, 1:500), CD68 (mouse, AbD Serotec, 1:500), CD206 (mouse, Santa Cruz Biotechnology, 1:200), CD16 (mouse, Abcam, 1:500), and arginase 1 (rabbit, Santa Cruz Biotechnology, 1:200). Sections were washed and incubated with Alexa Fluor 488 (1:400, Abcam) or Cy3-labeled secondary antibodies (1:400, Jackson Laboratories) for 1 h at room temperature. DAPI (1:1,000, Beyotime Biotechnology) was used to stain cell nuclei. Images were captured using a ×20 or ×40 objective of a Zeiss confocal microscope (LSM780, Zeiss). Image J software was used to analyze the intensity of the regions of interest.

Western blot analysis.

Western blotting was performed to measure the protein levels of MAP2, Tuj-1, DCX, nestin, NeuN, and TrkB in the brains of the animal groups (n = 6 for each group) as previously described (Chen et al. 2015). Rats were transcardially perfused with saline before decapitation, and ipsilateral coronary brain tissues (0–4 mm from the bregma) were collected on ice. Bicinchoninic acid protein concentration detection and SDS-PAGE were performed as previously reported (Jiang et al. 2017). The primary antibodies used were against TrkB (rabbit, Abcam, 1:1,000), MAP2 (mouse, Boster China, 1:500), Tuj-1 (mouse, Santa Cruz Biotechnology, 1:100), NeuN (rabbit, Abcam, 1:1,000), and nestin (mouse, Abcam, 1:1,000). GAPDH (diluted 1:1,000, Santa Cruz Biotechnology) was used as the loading control. Images were acquired using a ChemiDOCTMxrs+ system (Bio-Rad) and processed using Image Laboratory (Microsoft). Corresponding histograms were generated using GraphPad Prism 6 software.

Real-time PCR.

Real-time PCR was used to detect changes in the mRNA levels of the TrkB pathway and NPCs (neural precursors cells), such as Tuj-1, DCX, and nestin, as described previously (Tao et al. 2016). Total RNA from the hematoma region was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA). The primer sequences used were listed in Table 1. Expression levels were quantified with standard samples (n = 6).

Magnetic resonance imaging and cortical thickness measurement.

A 7.0-T/200-mm Varian scanner (Bruker BioSpin) was used to calculate the thickness of the cortex after ICH and minocycline administration. MRI and the method for calculating cortical thickness were described previously (Tang et al. 2016). The T2 fast spin-echo sequence (TR/TE = 3,000/45 ms) was used for all animals after behavioral testing. The field of view was 3.5 × 3.5 cm², and the matrix was 256 × 256. Seventeen coronal slices were obtained to cover the entire axis of the lateral ventricles (0.7 mm thick). Image J was used to measure the distance from the cingulum to layer II of the cortex in eight sections that were interspaced 0.7 mm and ranged from +3.2 to −3.8 from the bregma.

Golgi-Cox staining.

Golgi-Cox staining was performed on brain tissue according to the protocol of the FD Rapid GolgiStain Kit (FD NeuroTechnologies).

Statistical analysis.

All data are reported as means ± SE. The SPSS 13.0 software package (SPSS, Chicago, IL) was used for data analysis. One-way analysis of variance (ANOVA) was used as appropriate, followed by Student-Newman-Keuls tests. Two-way ANOVA with the Bonferroni post hoc test was used when necessary. A Welch's test was used in two-group comparisons. A P value less than 0.05 was considered statistically significant.

Minocycline improved long-term cognitive deficits after ICH.

We designed an experimental protocol to measure the long-term cognitive effects (MWM test) 28 days after ICH and minocycline administration (Fig. 1A). ICH animals swam a significantly greater distance than those in the sham group in the MWM with the platform removed (P < 0.01, Fig. 1B). Minocycline-treated animals swam a significantly shorter distance than vehicle-injected animals. By contrast, minocycline administration to the sham group did not significantly impact learning memories in ICH rats. Rats in the ICH + vehicle group spent a shorter time in the targeted quadrant compared with rats in the ICH + minocycline group (Fig. 1C; n = 12).

Minocycline decreased the proliferation of M1 microglia and promoted M2 microglia polarization after ICH.

We detected the proliferation and percentage of M1 microglia in the different groups to investigate the effects of minocycline on microglia polarization. We found a significant increase in CD68- and BrdU-positive cells around the hematoma in ICH brain tissue compared with sham animals (P < 0.01, Fig. 2, A and C). The number of CD68+ and BrdU+ cells decreased significantly after minocycline administration (P < 0.05, Fig. 2, A and C). ICH induced a higher percentage of CD68+ microglia than in the sham group (P < 0.01, Fig. 2, B and D). Fewer CD68-positive microglia around the hematoma were observed in minocycline-treated ICH brain slices compared with the vehicle-only group (P < 0.05, Fig. 2, B and D). No significant impact of minocycline on microglia was observed in sham animals.

We investigated the relationship between M1 and M2 microglia around the hematoma using immunohistochemistry and PCR after ICH and minocycline administration. Notably, a significant decrease in CD16-positive cells (M1 microglia) (Fig. 3A) and increase in CD16+ and arginase1+ cells (Fig. 3, B and C) was observed, which reflected the transformation of M1 microglia to M2 microglia after minocycline administration. Minocycline downregulated CD16 mRNA expression (Fig. 3D) after ICH and increased arginase 1 mRNA expression (Fig. 3E).

Minocycline enhanced microglia-synapse contacts in the cortex and induced neurogenesis.

We measured the direct contacts of microglia and neuronal cells in the cortex (Fig. 4A) and examined the relationship between M2 microglia and nestin expression in the subventricular zone. More ramified microglia were found to be colabeled with nestin protein in minocycline-treated brain tissue than in vehicle-only brains (Fig. 4, B and D). ICH decreased the number of processes of neuronal cells and reduced MAP2 and Tuj-1 expression. Minocycline significantly increased the expression of these two proteins in neuronal processes (Fig. 4C). This self-restoration increased the expression of some NPCs, such as nestin, and minocycline effectively enhanced ICH-induced self-restoration of neuronal cells.

We examined the effects of minocycline on the cerebral cortex and neuronal spinal dendrites and found a significant increase in the cerebral cortex thickness after minocycline treatment compared with vehicle-treated animals (Fig. 5B). Minocycline restored neuronal spinal dendrites after ICH and created more synapses (Fig. 5C).

Minocycline increased TrkB and BDNF expression in M2 microglia.

Many ameboid microglia clustered around the hematoma after ICH. Minocycline transformed the ameboid microglia into ramified microglia with increased BDNF expression (Fig. 6A). PCR and Western blotting were used to measure TrkB expression. TrkB mRNA expression decreased 12 h after ICH and continued to decrease over the next 12 h to a stable level. Minocycline increased TrkB mRNA expression 6 h after ICH (P < 0.05, Fig. 6B). A significant decrease in TrkB protein expression was observed 24 h after ICH, and minocycline administration reversed this loss of TrkB (P < 0.01, Fig. 6C).

The TrkB/BDNF pathway played a pivotal role in minocycline-induced neurogenesis after ICH.

Immunofluorescence and quantitative analyses revealed numerous changes: 1) Minocycline reduced the ICH-induced increase in M1 polarized microglia (CD68+). A TrkB-selective antagonist reversed the effect of minocycline, and a TrkB-selective agonist produced similar effects on the number of CD68-positive cells (Fig. 7A). 2) Minocycline and HIOC induced greater expression of M2-polarized microglia (CD206+) around the hematoma compared with vehicle-treated animals and animals treated with minocycline and ANA 12 (Fig. 7B). 3) Minocycline increased the number of mature neurons (NeuN+) in ICH animals, and ANA 12 reversed this neuroprotective effect (Fig. 7C). Integrated density analysis in the region of interest generated the same results (Fig. 7D).

We detected protein and mRNA expression of mature neurons (NeuN) and NPCs, such as Tuj-1, DCX, and nestin. Minocycline increased the expression of NeuN protein, which was reversed by ANA 12, and HIOC increased mature neuron expression to the same degree as minocycline (Fig. 8A). Minocycline increased the expression of Tuj-1, DCX, and nestin (Fig. 8, B and C) protein and mRNA. Selective blockade of the TrkB pathway automatically reversed the increase in NPC expression, and a TrkB agonist produced the same neuroprotective effects as minocycline.