Adult male Sprague–Dawley rats weighing 180–220 g were obtained from the Experimental Animal Center of Central South University (CSU), Changsha, China. All rats were maintained under specific pathogen-free (SPF) conditions with a 12-hour light/dark cycle, a controlled temperature (25°C), a relative humidity (45–55%), and ad libitum had access to water and food pellets throughout the study. All procedures were conducted following the guidelines and approved by the Institutional Animal Care and Use Committee of CSU (No: 2018sydw0159).

The collagenase-induced ICH rat model was an important way for studying ICH and had been used by various researchers. All experimental rats were deeply anesthetized intraperitoneally with 3% pentobarbital sodium (50 mg/kg) and placed in a stereotaxic frame (Stoelting Co., Chicago, IL, United States). After disinfection and incision, a hole was drilled in the skull and collagenase type VII (Sigma-Aldrich, United States, 0.5 U in 2.5 μL of 0.9% sterile saline) was injected slowly (over 2 min) via microliter syringe into the right globus pallidus, according to the following coordinates relative to bregma: 1.4 mm posterior, 3.2 mm lateral, and 6 mm ventral to the cortical surface. The needle stayed for an additional 10 min to prevent reflux then slowly removed. Sham group rats underwent the same procedure, except that the syringe contained 2.5 μL of 0.9% saline solution without collagenase. Due to the practical operation of the experiment, NIK siRNAs group rats underwent the same procedure but syringed into the left globus pallidus.

BYHWD was prepared and subjected to quality control as previously described (Cui et al., 2015). Astragalus mongholicus Bunge [Fabaceae; Astragali Radix], Angelica sinensis (Oliv.) Diels [Apiaceae; Angelica Sinensis Radix], Paeonia lactiflora Pall. [Paeoniaceae; Paeonia Radix Rubra], Ligusticum Chuanxiong Hort [Apiaceae; Chuanxiong Rhizoma], Carthamus tinctorius L. [Asteraceae; Carthami Flos], Prunus persica (L.) Batsch [Rosaceae; Persicae Semen], and Pheretima aspergillum (E. Perrier) [Megascolecidae; Pheretima] at a ratio of 60:6:4.5:3:3:3:3 (dry weight) were used (Table 1). All botanical drugs were available from the Chinese Medicinal Pharmacy of Xiangya Hospital, Central South University (Changsha, China). Briefly, all botanical drugs were immersed in distilled water for 1-h, and boiled for 30 min at 100°C. Finally, the powder (yield: 11.9%) was dissolved in distilled water at a concentration of 0.13 g/ml for intragastric administration.

We purchased the standard reference materials of amygdalin, hydroxysafflor yellow A, calycosin, and digoxin from Yuanye Bio-Technology Co., Ltd. (Shanghai, China). Digoxin is not the endogenous compound of BYHWD and plasma, and it does not obviously interfere with the retention times of all three analytes. Qualitative analysis was carried out using an LC-MS system (Shimadzu 8050, Kyoto, Japan) in negative ion mode. The plasma samples were added with digoxin, vortex mixed for 10 s, and then vortex mixed for 1 min and centrifuged for 15 min (13,000 rpm, 4°C) after being added 800 μL acetonitrile. The obtained supernatants were dried in a nitrogen dryer, diluted with 10% acetonitrile–water, repeated the extraction above, and injected into the LC-MS for analysis (Supplementary Figure S1).

There were two parts to the entire experiment. In the first part, rats were randomly divided into three groups: the sham group, the ICH group, and the BYHWD group. For the BYHWD group, rats were administered by gastrogavage with BYHWD once daily for the duration of the experimental observations after ICH modeling. According to our previous study (Cui et al., 2015), we chose 4.36 g/kg as the dosage of BYHWD in the present study. The distilled water with equal volumes was administered to the sham and ICH groups. Rats were sacrificed on days 3, 7, and 14, post-ICH after neural tests. In the second part, intact rats first randomly received a microinjection of NIK targeting (siNIK) or non-targeting siRNA (siControl) into the right lateral ventricles at concentrations of 10 mM. The siNIK group (NIK siRNAs + ICH) received NIK siRNAs (NIK siRNAs, Cyagen Biosciences, aCSF, 10 μL, i.c.v.) to inhibit the noncanonical NF-κB. The control group (siRNAs control + Sham and siRNAs control + ICH) received scramble siRNAs (siRNAs control, Cyagen Biosciences, aCSF, 10 μL, i.c.v.). Both were injected immediately with a cannula implantation system (RWD Life Science) once per day for 5 days. ICH surgery was conducted after 5 days of recovery. ICH induction provided a detailed description in the “collagenase model” previously. The BYHWD was administered to the ICH BYHWD group as before, and distilled water with equal volumes was administered to the siRNAs control + Sham, the siRNAs control + ICH, and the NIK siRNAs + ICH. Rats were sacrificed at days 7 and 14 after ICH, respectively.

The modified neurological severity score (mNSS), the corner turn test, and the foot-fault test were used to assess the neurological injury at various time points after ICH by two observers who blinded to the group assignments independently and their scores were averaged. These tests were detailedly performed in our previous studies (Narantuya et al., 2010; Cecatto et al., 2014; Krafft et al., 2014).

The modified neurological severity score (mNSS) examined motor ability, walking ability, sensory ability, balance, reflexes, and the presence of abnormal movements. And a higher score meant a more severe injury (normal score: 0; maximal deficit score: 18).

Rats would proceed into a corner with an angle of 30°degrees. To exit the corner, individual rats could turn either left or right, and the direction taken was then recorded. This was repeated 10 times per animal, with at least 30 s between trials. The percentage of right turns was calculated. The percentage of right turn in normal rats was about 50%, while the injured ones would have a higher percentage.

Rats were tested for placement dysfunction of forelimbs with the modified foot-fault test. Rats were set on an elevated grid surface (85.8 × 2.5 cm², with grids of different sizes) and placed their paws on the wire while moving along the grid. With each weight-bearing step, the paw may fall or slip between the wire, which was recorded as a foot fault. The total number of steps (movement of each forelimb) that the rats used to cross the grid was counted, and the total number of foot faults for the left forelimb was recorded. Data are presented as the percentage of foot faults per the total number of steps (normal score: 0%; maximal deficit score: 100%).

Rats were deeply anesthetized with 3% pentobarbital sodium (50 mg/kg, i.p.). For molecular biology experiments, including Western blot and quantitative real-time polymerase chain reaction (RT-qPCR), rats were perfused with 0.9% saline, and brain samples around the hematoma were collected in ice-cold saline. They were immediately stored in liquid nitrogen to ensure brain tissue staying fresh then stored at −80°C until analysis.

Rats were sacrificed at days 3, 7, and 14 after ICH induction, and the intact brain tissues were removed immediately. Brain tissues were divided into two hemispheres along the midline and the ipsilateral hemisphere divided into three parts by stainless steel brain matrices. The hematoma area brain tissue was weighed using an electronic analytical balance. After drying in an oven at 95°C for 72 h, until the sample weights were nearly consistent, the dry weight was obtained and calculated as follows: water content of brain tissues (%) = (wet weight–dry weight)/(wet weight) × 100%.

Collected fresh samples were homogenized in RIPA lysis buffer with a protease inhibitor. The homogenate was centrifuged at 12,000 rpm for 30 min at 4 °C. The supernatants were collected immediately. Protein samples were separated by SDS-PAGE gels and transferred onto PVDF membranes (Millipore, United States). The membranes were blocked with 5% BSA for 2 h at room temperature and then probed with primary antibodies overnight with gentle shaking at 4°C. After three washes in PBS-T, the membranes were subsequently incubated with horseradish peroxidase–conjugated anti-mouse IgG (1:5000, Proteintech, United States) or anti-rabbit IgG (1:6000, Proteintech, United States) secondary antibodies for 2 h at room temperature. Membranes were washed again, and the protein bands were visualized using enhanced chemiluminescence (ECL). The exposed films were scanned and analyzed by Quantity One analysis software.

The primary antibodies used were the following: rabbit anti-NIK (1:200, Abcam, United Kingdom); rabbit anti-IKKα (1:1000, Cell Signaling Technology, United States); rabbit anti-NF-κB p100/52 (1:1000, Cell Signaling Technology, United States); mouse anti-β-actin (1:5000, proteintech, United States); rabbit anti-Histone H3 (1:1000, Cell Signaling Technology, United States); rabbit anti-TNF-α (1 ug/ml, Abcam, United Kingdom); and rabbit anti-IL-β (0.2 ug/ml, Abcam, United Kingdom).

Total RNA was extracted with a TRIzol reagent (Invitrogen, Carlsbad, CA, United States) following the manufacturer’s instructions. Amplification was performed using SYBR Green All-in-one TM qPCR Mix (GeneCopoeia) on a ViiA^TM7 RT-qPCR system (Applied Biosystems). The following thermocycling protocol was used: 95°C for 10 min, 40 cycles of 15 s at 95°C, 50 s at 60°C, and melting was done at 60°C. The primers for IL-1β, TNF-α, and β-actin were designed with Premier 5.0 software for rats. The sequences of the primers are shown in Table 2. Melting curves of all the samples were generated as controls for specificity. The expression data were normalized to the expression of β-actin with the 2^−ΔΔCt method.

NIK targeting (NIK siRNAs) or non-targeting siRNA (siRNAs control) was injected into the right lateral ventricles of rats according to the following coordinates relative to bregma: 1 mm posterior, 1.5 mm lateral, and 4 mm ventral to the cortical surface at concentrations of 10 mM. 10 µL of either the NIK siRNAs or siRNAs control were injected into the lateral ventricle at 1 μL/min. The siNIK group (NIK siRNAs + ICH) received NIK siRNAs (NIK siRNAs, Cyagen Biosciences, aCSF, 10 μL, i.c.v.) to inhibit the noncanonical NF-κB. The control group (siRNAs control + sham and siRNAs control + ICH) received scramble siRNAs (siRNAs control, Cyagen Biosciences, aCSF, 10 μL, i.c.v.). Both were injected immediately with a cannula implantation system (RWD Life Science) once per day for 5 days. The inhibitory effect of NIK siRNA is shown in Supplementary Figure S2. The rat Map3k14 shRNA design and siRNAs control target sequences (NIK siRNAs) were: CCT​TGG​AAA​GGA​GAA​TAT​AAA and scramble-shRNA-PGK (siRNA control): CCT​AAG​GTT​AAG​TCG​CCC​TCG.

Results were presented as the mean ± SEM. One-way analysis of variance (ANOVA) was used for comparing differences. A Levene test was used to test the variance congruence of the data. Post hoc tests for multiple comparisons, using the LSD or the Bonferroni test if the variance is equal, or Tamhane's test if the variance is not equal. A p-value of <0.05 by SPSS 24 software was considered as statistically significant.

To define the effects of BYHWD in neurological behavioral impairment after ICH, we recorded the body weight and performed behavioral testing 1 h before sacrifice at each time point. Recovery of neurological function after ICH was achieved by BYHWD treatment; all results are shown in Figure 1. All rats were observed to be healthy and with normal body functions before modeling. Day 3 after ICH, there was little weight gain in collagenase-induced rats (ICH and BYHWD groups), while the sham group grew normally (Figure 1D). Meanwhile, the ICH and BYHWD group rats had severe neurological behavioral impairment compared to the sham group, which also showed the success of ICH modeling. Substantially higher scores of mNSS showed from the ICH (p < 0.01) and BYHWD groups, respectively (Figure 1A). Similarly, the percentages of the ICH and BYHWD groups were notably higher than the sham in the corner turn test (Figure 1B) and the foot-fault test (Figure 1C). Day 7 after ICH, the test scores of ICH rats were still higher than sham in mNSS (p < 0.001), the corner turn test (p < 0.01), and the foot-fault test (p < 0.001), while the scores of BYHWD rats were gradually decreasing. Day 14 after ICH, the test scores of ICH rats had remained unchanged as before and were still higher than sham in mNSS (p < 0.001, Figure 1A), the corner turn test (p < 0.01, Figure 1B), and the foot-fault test (p < 0.001, Figure 1C). However, after 14-day treatment with BYHWD, the neurological function impairment of the BYHWD group was ameliorated at varying degrees in mNSS (Figure 1A), the corner turn test (Figure 1B), and the foot-fault test (Figure 1C). Additionally, for weight change (Figure 1D), the ICH group grow slowly and the weight of the BYHWD group gained about over 20%.

As shown in Figure 2, to explore the effects of BYHWD treatment on ICH-induced inflammation in brain tissues, we examined brain edema using a wet/dry method on days 3, 7, and 14, evaluating the impact of BYHWD treatment on ICH-induced brain edema with its dynamic variations (Figure 2A). In the ipsilateral hematoma area of these three groups, the ICH group showed significant increases in water content compared with the Sham group, while BYHWD treatment considerably alleviated the brain water content on day 14 after ICH. Inflammatory cytokine activation contributed to the evolution of secondary degeneration after ICH, both mRNA and protein levels of TNF-α and IL-1β, which were the most important indicators of inflammation response and were detected by RT-qPCR and Western blot analysis on days 3, 7, and 14 after ICH (Figures 2B,C). All the observation time points, the results showed that these two indicators were increased in the ICH group compared with the sham group. However, the BYHWD group significantly decreased both expression levels compared with the ICH group as treatment time progressed. Treatment with BYHWD remarkably decreased the IL-1β and TNF-α secretion after ICH, compared with the ICH group. Above all, inflammatory cytokine could be well inhibited by BYHWD. The results demonstrated that the use of BYHWD may contribute to the control of ICH-induced inflammation.

We investigated the specific indicators of the noncanonical NF-κB pathway, including NIK, Ikkα, NF-κB p100, and NF-κB p52, dynamic changes of protein expression after ICH using Western blot (Figures 3A–D).

The results showed that several specific indicators of the noncanonical NF-κB pathway, including NIK, Ikkα, NF-κB p100, and NF-κB p52, were increased in brain tissue after ICH. In the cytoplasm, day 3 after modeling, the expression levels of NIK, IKKα, NF-κB p100, and NF-κB p52 were markedly increased in the ICH group compared with the sham group. And these indicators were relatively decreased in the BYHWD treatment group compared with the ICH group. After long-term treatment, NIK, Ikkα, NF-κB p100, and NF-κB p52 in the ICH group were still in high expression on days 7 and 14 after ICH, but the BYHWD group significantly improved. In the nucleus, NF-κB p100 was not detected because it was turned into p52 before entering the nucleus. The results of NIK, Ikkα, and NF-κB p52 were consistent with that of the cytoplasm. Day 3 after modeling, compared with the sham group, the expression of NIK, IKKα, and NF-κB p52 was remarkably increased in the ICH group. The expression of these three indicators was reduced in the BYHWD group compared with the ICH group, and the decrease in IKKα turned to be statistically different. On days 7 and 14, the results were similar to those in the cytoplasm. These findings suggested that specific indicators of the noncanonical NF-κB pathway were highly expressed in ICH rats’ brain tissues and the BYHWD treatment can inhibit NIK, the key promoters of noncanonical NF-κB pathways, in brain tissues after ICH.

Following the determination that NIK is highly expressed in brain tissue after ICH, we further investigated whether targeting of NIK in ICH alone could suppress the inflammatory responses in the ICH model. To answer this, we pretreated rats with either NIK targeting (NIK siRNAs) or non-targeting siRNA (siRNAs control) before ICH surgery. Then using RT-qPCR and the Western blot analysis measured the mRNA and the protein level of NIK (Figure 4) and the inflammation indicators (Figure 5) in brain tissues on days 7 and 14 after ICH.

Compared with the siRNAs control + sham group, the relative mRNA level of NIK in each ICH group was increased at different degree (Figure 4). The results showed that the relative mRNA and protein expression of NIK in the NIK siRNAs + ICH group was downregulated than that in the siRNAs control + ICH, which represented the success of pretreatment. Compared with the NIK siRNAs + ICH group, the ICH + BYHWD group showed no statistical difference in the relative mRNA expression of NIK levels and protein levels of NIK on days 7 and 14 after ICH. But mRNA and protein levels of NIK both in the NIK siRNAs + ICH group and the ICH + BYHWD group were much lower than those in the siRNAs control + ICH group showed on days 7 and 14 after ICH. This suggests that the activation of key regulators of noncanonical NF-κB pathway was inhibited in siRNA-pretreated rats after ICH, while NIK was also inhibited by BYHWD treatment.

The results showed that inflammatory factors were notably in the brain tissue after ICH. The relative mRNA and protein expression of TNF-α/IL-1β in the NIK siRNAs group was significantly decreased than those in the control siRNA group after ICH of both observation time (Figure 5). These observations indicated that after downregulating NIK, the key regulator of the noncanonical pathway, the activation of the noncanonical pathway was inhibited. However, experiments of first part have shown that BYHWD can effectively inhibit NIK.

Relative mRNA and protein levels of inflammatory indicators were comparable between the NIK siRNAs + ICH group and the ICH + BYHWD group on days 7 and 14 after ICH, suggesting that BYHWD alleviates the inflammatory response after the recovery of ICH and may be involved in NIK-mediated repression of the noncanonical NF-κB pathway.