After approval by the ethical committee of Shandong Provincial Hospital, an observational study of CSF from 73 patients with SAH (confirmed by a head computed tomography angiography) between May 2019 and May 2020 was performed. Inclusion criteria: (1) aneurysm treated endovascularly (aneurysm coiled and aneurysm clipped) <24-h post-rupture and (2) external ventricular drainage placed <48-h post-rupture. Exclusion criteria were as follows: (1) CNS disease history, (2) CNS infection, and (3) systemic disease (diabetes mellitus, malignancy, and cirrhosis). On admission, clinical and hemorrhage severity were assessed by the Hunt and Hess (HH) grade, World Federation of Neurological Surgeons (WFNS) grade, and Fisher score (Dong et al., 2019). Patients received intravenous infusion of nimodipine for at least 7 days. Euvolemia was maintained, and hypotension was avoided with vasopressors (Dong et al., 2019). The end point was assessed at day 8. CSF samples were collected in sterile tubes or catheter and stored at −80°C. CSF samples of patients with normal pressure hydrocephalus (NPH) were as the experimental control because CSF is difficult to obtain from healthy individuals (Kwan et al., 2019).

After approval by ethics committee of Shandong First Medical University, SAH models were produced in male Sprague–Dawley rats (12 weeks old, 320–350 g, Jinan Pengyue Laboratory Animal Center) by single blood injection model (Wu et al., 2017). Briefly, rats were deeply anesthetized (5% isoflurane) and then maintained (2% isoflurane) using a rodent ventilator (MatrixVMR). Non-heparinized autologous blood (0.3 ml) was injected into cisterna magna for 3 min using a 1-ml syringe with a 25-G needle under a stereotaxic apparatus. Sham-operated group underwent the same procedures except for injection of blood.

Three behavioral activity test of scoring system (Table 1) was performed at day 2 after SAH as previously described (Wu et al., 2017). Sequence of testing was randomized. Scoring was evaluated to record appetite, activity, and deficits blindly. About 50 μl of CSF was extracted from the cisterna magna (0.5 cm depth) with 1-ml syringe under a stereotaxic apparatus and then stored at −80°C (Zhang et al., 2020).

The HSP27 of CSF was assayed using an enzyme linked immunosorbent assay (ELISA) Kit (ab108862, Abcam) according to the manufacturer's instruction and expressed in ng/ml. Western blot was conducted as previously described (Zhao et al., 2020). Briefly, total protein was extracted with a protein extraction kit (BC3710, Solarbio), which was supplemented with a protease inhibitor cocktail (P8340, Sigma), and analyzed by bicinchoninic acid (BCA) protein concentration kit (PC0020, Solarbio). About 20 μg of total protein was separated using sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis and then electrotransferred onto nitrocellulose membrane. After blocking with 5% (w/v) non-fat milk, the membranes were incubated with anti-HSP27 (1:1,000, ab2790, Abcam), anti-active caspase-3 (1:1,000, ab2302, Abcam), anti-FLAG (1:1,000, F1804, Sigma), anti-phospho-MKK4 (Ser257/Thr261; 1:1,000, #9165, CST), anti-phospho-c-Jun N-terminal kinase (JNK) (Thr183/Tyr185; 1:1,000, #4688, CST), anti-phospho-c-Jun (Ser63; 1:1,000, #2361, CST), anti-MKK4 (1:1,000, #9152, CST), and β-actin (1:1,000, #4970, CST) at 4°C overnight and then incubated with anti-mouse immunoglobulin G-horseradish peroxidase (IgG-HRP; 1:3,000, #7076, CST) or anti-rabbit IgG-HRP (1:3,000, #7074, CST) linked antibody for 2 h at the room temperature. After washing with Tris-buffered saline with Tween 20 (TBST) buffer (T1081, Solarbio), the protein bands were visualized by using chemiluminescent substrate (#34577, Thermo Scientific) in a ChemiDoc^TM MP Imaging System (Bio-Rad) and quantified by Image J software.

The HSP27-overexpressing adeno-associated virus (AAV) and HSP27 shRNA (Table 2) were constructed by OBiO Technology (Shanghai, China). Full-length rat cDNA of HSP27 was cloned into the pAAV-CAG-3FLAG vector. AAV-HSP27-RNAi was constructed using the pAKD-CMV-EGFP-H1 vector, which contains a CMV promoter (driving EGFP) and a H1 promoter (driving shRNA expression). The sequence of HSP27 shRNA is 5′-GCTACATCTCTCGGTGCTTCA-3′ and 5′-GCCCAAAGCAGTCACACAATC-3′ as previously described (Stetler et al., 2008). AAV was produced by co-transfection (one AAV vector and two helper vectors) of HEK293T cells. At 72 h after transfection, cells were harvested and lysed using a freeze–thaw procedure. AAV of cell lysate was purified using a heparin-agarose column and concentrated using an ultrafiltration device. The virus titer was measured and shown in Table 2. The HSP27 peptides (Table 3) were synthesized by ChinaPeptides (Shanghai, China). For intracerebroventricular (i.c.v.) injection, rats received a single injection (0.8 mm posterior, 1.2 mm lateral, and 3.8 mm depth) of AAV (5 μl), TAT (0.3 mg, 10 μl), or TAT-HSP27₆₅₋₉₀ (0.3 mg, 10 μl) with a 30-G needle of a 10-μl Hamilton syringe under a stereotaxic apparatus.

Staining of slices was performed as previously described (Wang et al., 2019). Briefly, rats were deeply anesthetized and perfused transcardially with ice-cold 4% paraformaldehyde. Rat's brain was post-fixed with 4% paraformaldehyde for 12 h, dehydrated with 30% sucrose/phosphate-buffered saline (PBS) at 4°C for 3 days, and then cut into 10-μm thickness coronal slices (−2.5 to −5 mm from bregma) using a Leica cryostat microtome. Coronal slices were permeabilized with 0.5% Triton X-100 and blocked with 5% goat serum, incubated with anti-NeuN rabbit antibody (1:200, #12943, CST), anti-NeuN mouse antibody (1:200, #94403, CST), anti-HSP27 (1:200, ab2790, Abcam), anti-Iba-1 (1:200, #19741, Wako) primary antibody at 4°C overnight and then incubated with the Anti-Mouse IgG-TRITC (1:100, T5393, Sigma) or Anti-Rabbit IgG-FITC (1:100, F9887, Sigma) at room temperature for 2 h. Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining was assayed using in situ Cell Death Detection Kit with Fluorescein (11684795910, Roche) according to the manufacturer's instruction. Images were captured using a fluorescence microscope (BX51, Olympus) and analyzed with Image J software.

Primary cortical neurons were obtained from E18 rat embryos as previously described (Zhang et al., 2018). Briefly, the dissociated neurons (10⁶ cells) were plated in 60 mm dish with neurobasal medium [2% B27 (Gibco), 1% L-glutamine (Gibco), 0.3% D-glucose (Sigma), and 1% fetal bovine serum (Gibco)] and fed with fresh medium every 3 days. Hemolysate-induced cortical neuron death model was produced as described previously (Li et al., 2017). Briefly, hemolysate was prepared from mouse arterial whole blood by freezing and stored at −80°C. In order to induce neuronal death, hemolysate in medium (1:50) was used to stimulate for 24 h. Neurons were treated with hemolysate in medium (1:50) or plus indicated HSP27 peptides (0.03 mg/ml, Table 3) for 24 h. Cell viability was measured with Cell Counting Kit-8 (CCK-8, Dojindo) according to the manufacturer's instruction. Cell apoptosis was assayed with TUNEL bright-red apoptosis kit (A113, Vazyme) according to the manufacturer's instruction. Images of cultured neurons were captured under a phase contrast microscope (phase contrast, ×200). Pictures of TUNEL staining were captured under a fluorescence microscope (red, ×200). In addition, the total protein of neurons was extracted and then analyzed by Western blot.

GraphPad Prism 6 was used to perform statistical analyses. All results were expressed as means ± standard deviation (SD). Data of figure were analyzed using one-way analysis of variance (ANOVA) followed by the Bonferroni's multiple comparisons test and the two-tailed t-test. p < 0.05 considered statistically significant.

Table 4 shows the clinical characteristics of patients with aSAH and NPH. There was no significant difference in age and gender between two groups (p > 0.05). In patients with aSAH, HH grade, WFNS grade, and modified Fisher score (on admission) were 2.6 ± 1.0, 2.4 ± 1.5, and 2.9 ± 0.7, respectively. Next, the temporal course of HSP27 level (day 1 to 8) following aSAH is depicted in Figure 1. HSP27 level of CSF on day 1 was significantly increased in comparison to that in patients with NPH. HSP27 level of day 2 to 4 was gradually decreased in comparison to day 1 (Figure 1A). Furthermore, there was a significant difference in HSP27 level (day 1) between patients with aSAH with HH grade of I–III vs. IV–V (Figure 1B), WFNS grades of I–III vs. IV–V (Figure 1C), and Fisher scores of 1–2 vs. 3–4 (Figure 1D).

Next, we evaluated the expression of HSP27 in brain using the rat SAH model. The calculated mortality rate at 72 h is given in Table 5. Blood clots were observed on the circle of Willis in SAH groups (Figure 2A). ELISA showed that HSP27 level of CSF was obviously increased at 12 h in comparison to that in sham group and then declined significantly (Figure 2B). In immunofluorescence staining, HSP27 can be co-located with NeuN (a marker for neuron) (Figure 2C), whereas rarely co-located with Iba-1 (a marker for macrophages/microglia) (Figure 2D). Moreover, HSP27 staining was significantly increased at 12 h compared with that of the sham group, whereas obviously declined at 72 h (Figures 2C,D). These results indicate that expression of HSP27 is first increased and then declined in rat SAH.

To confirm whether endogenous HSP27 has effect on neurological function, rat SAH was subjected to HSP27 knockdown using AAV-eGFP-shRNA (Figure 3A). The calculated mortality rate at 48 h is given in Table 5. After microinjection into the lateral ventricle of rat (Figure 3A), AAV-eGFP-shRNA effectively infected the basal cortex, produced considerable expression of eGFP (Figure 3B), and significantly decreased expression of HSP27 (Figures 3C, 4C), suggesting an effective knockdown effect. Statistical results of TUNEL staining showed that numerous TUNEL positive cortical cells significantly increased in SAH + HSP27 shRNA group as compared with that of SAH + negative control of shRNA group on day 2 after SAH (Figure 3D). Moreover, the behavior scores in SAH + HSP27 shRNA group were significantly increased when compared to SAH + negative control of shRNA group (Figures 4A,B), suggesting that HSP27 shRNA worsens neurological function after SAH. Western blot analysis showed that the activation of caspase-3 was significantly increased in SAH + HSP27 shRNA group as compared to SAH + negative control of shRNA group (Figure 4C). These results indicate that knockdown of endogenous HSP27 increases cell apoptosis and deteriorates neurological deficit in rat SAH.

To test whether HSP27 overexpression has effect on neurological function, rat SAH was infected with AAV-HSP27-3FLAG (Figure 5A), which expresses full-length HSP27. The calculated mortality rate at 48 h is given in Table 5. The behavior score analysis showed that behavior scores in SAH + AAV vector encoding HSP27 group were significantly decreased when compared to SAH + AAV vector group (Figure 5B), suggesting an improvement effect. After microinjection into the lateral ventricle of rat, immunofluorescence staining showed that AAV-HSP27-3FLAG obviously increased expression of HSP27 in the basal cortex (Figure 5C). Western blot confirmed the increased HSP27 protein expression by detecting FLAG expression (Figure 6). Moreover, statistical results of TUNEL staining showed that numerous TUNEL positive cortical cells significantly decreased in SAH + AAV vector encoding HSP27 group as compared with that of SAH + AAV vector group on day 2 after SAH (Figure 5D). Previous review showed that HSP27 exerts an anti-apoptosis role by inhibiting MKK/JNK cell death signal, and or mitochondria-related pro-apoptotic factors (Shan et al., 2021). Thus, we further examined the activity of MKK/JNK and caspase-3 by Western blot. Data showed that the phosphorylation of MKK4, JNK, and c-Jun and the activation of caspase-3 were obviously increased in SAH + vehicle and SAH + AAV vector group, whereas significantly decreased in SAH + AAV vector encoding HSP27 group (Figure 6). These results indicate that HSP27 overexpression attenuated cortical cell apoptosis and neurological deficit in rat SAH.

N-terminal region of HSP27 is composed of 1–120 amino acids and proved to be a necessary domain for neuroprotection (Stetler et al., 2008). So, we designed and synthesized peptides from the N-terminal 1–120 amino acids of HSP27 (Figure 7A), which had no effect on cell activity in cultured cortical neuron (Figure 7B), aiming to find the key peptide that affects cell apoptosis. Using a hemolysate-induced cortical neuron death model, Western blot and TUNEL staining showed that HSP27₆₁₋₉₀ and HSP27₆₅₋₉₀ peptides, but not HSP27₁₋₃₀, HSP27₃₁₋₆₀, and HSP27₉₁₋₁₂₀ peptides, can effectively inhibit hemolysate-induced the increased the activation of caspase-3 (Figure 7C) and reduce hemolysate-elevated the number of TUNEL positive cells (Figure 7D). These results suggest that the N-terminal 65–90 amino acids of HSP27 are the key area for affecting cell apoptosis.

Furthermore, we investigated whether TAT-HSP27₆₅₋₉₀ peptide could improve neurological deficit in rat SAH. The calculated mortality rate at 48 h is given in Table 5. As expected, after microinjection into the lateral ventricle of rat SAH (Figure 8A), TAT-HSP27₆₅₋₉₀ significantly decreased the behavior score as compared with the TAT-treated SAH group (Figure 8B), suggesting that TAT-HSP27₆₅₋₉₀ treatment attenuates SAH-induced neurological deficits. In addition, Western blot analysis suggested that the vehicle or TAT-treated SAH group showed upregulation of active caspase-3, which was attenuated by TAT-HSP27₆₅₋₉₀ treatment (Figure 8C). Furthermore, TUNEL staining suggested that TUNEL-positive cells of the basal cortex were significantly decreased at 48 h in the TAT-HSP27₆₅₋₉₀-treated SAH group when compared to the TAT-treated SAH group (Figure 8D), indicating TAT-HSP27₆₅₋₉₀ reduced cortical cell apoptosis after SAH. There are two commonly used SAH models: blood injection model and endovascular puncture model (Kooijman et al., 2014). Later, we explored the effect of TAT-HSP27₆₅₋₉₀ on neurological deficit and cell apoptosis in the endovascular puncture rat SAH model. There was no statistical difference in SAH grading score among the vehicle, TAT, or TAT-HSP27₆₅₋₉₀-treated SAH group (Supplementary Figure S1A). TAT-HSP27₆₅₋₉₀ treatment significantly increased the average modified Garcia score as compared with the TAT-treated SAH group (Supplementary Figure S1B), suggesting that TAT-HSP27₆₅₋₉₀ treatment improved neurological deficits after SAH. Moreover, NeuN staining suggested that NeuN-positive neurons of the hippocampal CA1 region were significantly increased at 24 h in the TAT-HSP27₆₅₋₉₀-treated SAH group in comparison with that of the TAT-treated SAH group (Supplementary Figure S1C). TUNEL staining suggested that TUNEL-positive cells of the basal cortex were significantly decreased at 24 h in the TAT-HSP27₆₅₋₉₀-treated SAH group when compared to the TAT-treated SAH group (Supplementary Figure S1C), indicating TAT-HSP27₆₅₋₉₀ reduced cortical cell death after SAH in the endovascular puncture model.