Network pharmacology to investigate the pharmacological mechanisms of muscone in Xingnaojing injections for the treatment of severe traumatic brain injury

Reagents

XNJI was obtained from Jiminkexin Pharmaceutical Company (Wuxi, China). The number of China Food and Drug Administration was Z32020562. Muscone and LY294002 (PI3K inhibitor) were obtained from Sigma (St. Louis, MO, USA). We obtained antibodies against p-65, p65, p-PI3K, PI3K, p-Akt1 (Thr308), Akt1, Bax, Bcl-2, cleaved caspase-3, p53, and β-actin from Abcam Biotechnology (Abcam, Cambridge, MA, USA).

Acquisition of chemical ingredients and evaluation of pharmacokinetics

The XNJI ingredients are collected based on previous studies and the Traditional Chinese Medicine System Pharmacology Database (TCMSP) (https://tcmspw.com/tcmsp.php) (Ru et al., 2014), the Bioinformatics Analysis Tool for Molecular mechanism of Traditional Chinese Medicine (BATMAN-TCM) (Liu et al., 2016b), the Encyclopedia of Traditional Chinese Medicine (ETCM) (Xu et al., 2019), PubChem (https://pubchem.ncbi.nlm.nih.gov/), and SwissADME (http://www.swissadme.ch/) to determine the components of XNJI. These resources provided the names or structure of specific compounds. We used the ADME parameter-based virtual algorithm to identify bioactive components with a drug-likeness threshold ≥ 0.18 (Liu et al., 2016a) and barrier permeability of ≥0.30 (Li et al., 2015). Besides, some molecules in the previous studies that have been proven to pass through the blood–brain barrier will be included.

Screening bioactive component protein targets

The active ingredient targets in XNJI were obtained from the TCMSP target section. We used Swisstarget Prediction for target prediction if our search did not return results (http://www.swisstargetprediction.ch/). TBI-related target genes were obtained from DisGeNET (http://www.disgenet.org/), Online Mendelian Inheritance in Man (OMIM, https://omim.org/), and the GeneCards database (https://www.genecards.org/). We removed any duplicates from the previous search results. The genes overlapping the compounds and target genes of TBI were included in a Venn diagram using a bioinformatics tool (http://www.bioinformatics.com.cn/).

Constructing the compound-target network

We used Cytoscape ver. 3.7.2 (https://cytoscape.org/) to construct and visualize the network. Nodes were used to show target proteins or molecules, and edges indicated the correlations between targets and compounds. Larger nodes indicated more links.

Construction of the protein interaction network

We used STRING 11.0 (http://string-db.org/cgi/input.pl) to establish the network of TBI-related PPI (Szklarczyk et al., 2017). The top 30 hub proteins were screened for analysis.

Gene ontology and pathway enrichment analysis

GO and KEGG functions were analyzed using the DAVID Bioinformatics Resources database (ver. 6.8) to illustrate the overlap (https://david.ncifcrf.gov/) (Huang da, Sherman & Lempicki, 2009). We created bar and bubble charts of GO and KEGG using R (version 4.0.2).

TBI model

Sprague-Dawley (SD) rats, 250–300 g, were provided by the Cancer Institute of the Chinese Academy of Medical Science (Beijing, China). All animals were handled following the ethical guidelines for animal-related experiments (ethical approval number: 2013022177). The experimental procedures were approved by the Animal Ethics Committee of the Chinese PLA General Hospital. Rats fasted 6–8 h before the operation. Our model was constructed following the controlled cortical impact (CCI) method (Romine, Gao & Chen, 2014; Schwulst & Islam, 2019) and SD rats were anesthetized by aspiration using isoflurane (velocity of flow 7-8L/min, concentration 3%). The craniotomy was performed midway between Bregma (x = 0 mm, y = 0 mm, z = 0 mm, after calibration) and lambda (−0.3 mm <z <0.3 mm) on the left side; the location of the craniotomy (x = 3 mm, y = −3.5mm) was lateral to the midline. Skull slices were then removed with the dura undisturbed. The CCI was applied perpendicular to the surface of the brain via PinPoint™ Precision Cortical Impactor (RWD, Shenzhen, China) with an impact tip measuring five mm in diameter, 4 m/s impact velocity, 500 ms impact duration time, and six mm depth. The craniotomy was sealed using bone wax and then the scalp incision was disinfected and sutured.

Part 1:36 SD rats were categorized into six arbitrary groups, which were as follows: TBI at 24, 72, and 168 h after injury, sham, TBI + XNJI, and TBI + muscone for RT-PCR. Part 2:48 SD rats were categorized into four groups randomly (sham, TBI + muscone, TBI, and TBI + muscone + LY294002 (PI3K inhibitor)) at 168 h for Western blot and apoptosis detection (Fig. S1).

Drug administration

XNJI was injected intravenously 10 ml/kg daily (Yang et al., 2015), and muscone was administered at 0.54 mg/kg/d equivalently according to the manufacturer’s instructions. The muscone injection was administrated with LY294002 (10 uL, 100 nmol in 25% DMSO of PBS) in the TBI + muscone + LY294002 group (Ge et al., 2017). The other groups received the same volume (Fig. S1).

Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labelling (TUNEL) assay

Cellular apoptosis was identified based on a TUNEL assay according to the manufacturer’s protocol (Meilunbio, Dalian, China). Slides were prepared by dewaxing, hydration, washing, and penetration. A 50 ul Tunel detection solution was added and the slides were washed with PBS and incubated for 60 mins. Finally, glass substrates were sealed with an anti-fluorescence attenuation solution. Fluorescence quantitative analysis was completed using Image J (National Institutes of Health, Bethesda, MD, USA).

Immunohistochemical (IHC) staining

IHC analysis was performed as described in a previous study (Zhang et al., 2018).

Western blot

Ten-thousand gram The TBI rat Cerebral cortex was lysed in RIPA buffer for 30 min and subsequently centrifuged for 5 min at 4 °C 10,000 g, to get proteins. The generated protein was separated using 10% SDS gel. The resulting proteins were transferred electrophoretically from SDS–PAGE to PVDF membranes with a wet-blot transfer apparatus (Bio-Rad, California, USA). The membranes were incubated overnight at 4 °C with p53, cleaved caspase3, p-p65, p65, PI3K, Akt1, and p-Akt1 (Thr308), and p-PI3K antibodies, followed by a second antibody marked with horseradish-peroxidase for 90 min at room temperature. Bands were visualized with an improved chemiluminescence kit following the manufacturer’s protocol and were detected with a Bio-Rad-based imaging system (Bio-Rad, Hercules, CA, USA). We analyzed densitometry quantitation using Image J.

RT-PCR

RNA was extracted from the affected brain cortex using TRIzol^® reagent (Invitrogen Life Technologies, Paisley, Scotland). All RNA (4 ul) was changed into complementary DNA (cDNA) in a reverse manner using M-MLV Reverse Transcriptase (Invitrogen, Paisley, Scotland). The cDNA was incubated using the Fast SYBR^®Green Master Mix Bulk Pack (Invitrogen, Paisley, Scotland) for quantitative fluorescence analysis. The primer sequence was as follows: BCL-2, forward primer, 5′ -CCGGGAGAT CGATGAAGTA-3′, reverse primer, 5′-CATATTTGTTTGGGGCA TGTCT-3′; Bax, forward primer, GTGAGCGGCTGCTTGTCT, reverse primer, GTGGGGGT CCCGAAGTAG; Caspase 3, forward primer, AATTCAAGGGACGG GTCATG, reverse primer, GCTTGTGCGCGTACAGTTTC; GAPDH, forward primer, ACAGCAACAGGGTGGTGGAC, reverse primer, TTTGAGGGTGCAGCGAACTT.

Forty amplification cycles (pre-denaturation at 94 °C for 2 min, denaturation for 30 s at 94 °C, and annealing for 30 s at 62 °C) were applied in the amplification reaction analysis using Rotor-Gene Q PCR’s detection system (Qiagen Biosystems, CA, USA).

Statistical analysis

All data are shown as mean ± standard deviation (SD). The statistical significance was analyzed using one- or two-way ANOVA in Graphpad Prism 8; statistical significance was P < 0.05.

Screening active molecules in XNJI

XNJI compounds were obtained from earlier studies and public databases. The BBB permeability and drug likeness’s threshold values were set greater than 30% and 0.18, respectively, to screen for active ingredients in the databases. In Moschus, cholesterol, n-nornuciferine, 17-beta-estradiol, cholesteryl ferulate, and muscone supported the established pharmacokinetic standard. For Radix-Curcumae β-sitosterol, sitosterol, Oxy-curcumenol, 1,7-Diphenyl-3-acetoxy-6(E)-hepten, 4,5-Epoxy-12-acetoxy-7a,11a-dihydrogermacradien-8-one, difurocumenone, caryophyllene, Germacron, curcumin, curdione, and curzerenone were active compounds. Caryophyllene was shared with Curcumae-Radix and Borneolum-Syntheticum. There were seven active ingredients in Borneolum-Syntheticum with the exception of caryopyllene. Radix Curcumae and Fructus Gardeniae both have beta-sitosterol. Fructus Gardeniae possesses 13 active compounds including beta-sitosterol. We were able to obtain 35 active molecules based on the ADME parameters in the TCMSP database and previous studies. We then used PubChem to produce 2D structures for the next analysis.

Active ingredients and TBI target predictions

The targets of the active ingredients in XNJI were acquired using TCMSP. Swisstarget Prediction identified other potential targets using 2D structures. The GeneCards, Disgenet, and Omit databases were used to retrieve disease-related targets. Duplicate targets were removed from our results. The gene name was transferred from the protein name using the UniProt database (http://www.UniProt.org/). Data in these databases are taken from the literature and are constantly updated to provide relatively comprehensive references.

Constructing compound-target networks

We retrieved 173 genes linked to TBI from the intersection of active ingredients and disease targets and used these to construct a Venn diagram (Fig. 1A). The network of compound-targets indicated that there were 173 target nodes, 35 compound nodes, and 404 edges (Fig. 1B). Larger nodes were used to indicate a greater degree of intersection to improve visualization. A larger degree in the network indicated the importance of the target or compound. The node with the largest degree value was muscone (degree = 42). It was much larger than the average value (3.81), indicating that it has potential therapeutic use in the treatment of TBI. Both AR (degree = 22) and ESR1 (degree = 20) played crucial roles in the network. Our analysis revealed that CASP3 (degree = 6), RELA (p65, a subunit of the NF-kB family, degree = 6), and BCL-2 (degree = 6) were larger than the average.

We performed GO and KEGG pathways enrichment analyses to verify the association between the predicted target and TBI. As shown in Figs. 2A and 2B, the GO terms (which include NAD-dependent histone deacetylase activity, steroid binding, membrane raft, membrane microdomain, response to molecule of bacterial origin, and response to antibiotics) were enriched in bar and bubble plots. We used an identical method to analyze potential targets in the critical pathway. The top 20 pathways were filtered and probably implicated in the signaling pathways of apoptosis, PI3K-Akt, p53 and the cell cycles, among others (Figs. 2C and 2D).

PPI network and top 30 targets from the PPI network

We explored the hub proteins that were potential therapeutic targets in the PPI network to determine the actions of the molecular mechanisms in TBI. We removed the unrelated targets which left a total of 169 targets and 1,279 interactions in the PPI network (Fig. 3A). The linked targets were given a different color and size to represent the meaning of their interaction.

Our results showed that the five hub proteins AKT1, CASP3, RELA, BCL2L1, CASP8 (Degree: AKT1 = 88, CASP3 = 60, RELA = 37, BCL2L1 = 35, and CASP8 = 21; Betweenness: AKT1 = 6272.24553, CASP3 = 1679.86775, RELA = 486.18204, BCL2L1 = 537.55344, and CASP8 = 75.49232) were located at the core position. These were included in the top 30 key proteins that were visualized in a bar plot according to their degree in the PPI network (Fig. 3B). The nodes gradually become redder, bigger, and more centered according to their importance. We retrieved an interactive pathway from the KEGG Pathway database to determine the molecular mechanisms of the five proteins in TBI (Kanehisa et al., 2017). We found that these hub proteins may correlate with PI3K/Akt, p53 apoptosis, cell cycle, NF-kB, and glycolysis/gluconeogenesis (Fig. 2D). AKT1 is a critical regulator of apoptosis and cell survival in the central nervous system (Chen et al., 2018; Mehrafza et al., 2019); CASP3, BCL2L1 and CASP8 participate in the cell anti-apoptosis and apoptosis process (Ali et al., 2020; Loo et al., 2020; Sun & Pei, 2016); RELA is a crucial mediator in the NF-kB pathway and activates the survival and progression of cell (Cahill, Morshed & Yamini, 2016; Du et al., 2019; Gugliandolo et al., 2018). Muscone plays a critical role in the component-target network and animal experiments have shown that it can attenuate neuronal apoptosis (Wei et al., 2012). We found that AKT1 was targeted by muscone in PPI network. Meanwhile, muscone exhibited a high degree of matching with Akt1 (Fig. S4). Therefore, muscone may modulate neuronal apoptosis through the PI3K/Akt, NF-kB signaling pathway and p53 inhibition.

Experimental verification

Temporal profiles of TBI-induced and XNJI-treated or muscone-treated changes in bax, BCL-2 and caspase 3, and p53 mRNA levels

The expression levels of BCL-2, bax, p53 and caspase 3 mRNA in the damaged cortex were investigated at a various range of times in TBI. As shown in Fig. 4, mRNA gradually increased 24 h after TBI up until 168 h after the damage occurred (Fig. 4) (P < 0.05). We detected mRNA expression levels at the 168 h mark in the TBI + XNJI, TBI + muscone, and sham groups. The results showed that these mRNA were remarkably down-regulated 168 h after treatment with XNJI and muscone, respectively (Fig. S3).

Muscone affects neuronal apoptosis in the injured cerebral cortex 168 h after TBI

We used the TUNEL assay, IHC, and Western blot to explore the function of muscone in neuronal apoptosis. According to representative TUNEL staining, the red fluorescent spots indicated cellular apoptosis (Fig. 5A). The TUNEL-positive cells in the TBI group (11.4%–15.3%) exceeded those in the sham group (1.3%−2.5%) according to our quantitative results (Fig. 5B) (P < 0.05). Muscone treatment attenuated neuronal apoptosis 168 h post-TBI (4.3%−5.9%), and an injection of LY294002 (a specific inhibitor of PI3K) reversed the effect (7.9%−9.2%). This trend was observed in immunohistochemistry as well (Fig. S2).

Muscone decreased Akt1 and PI3K phosphorylation in TBI

It has been shown that PI3K may activate and phosphorylate Akt1 (Yang et al., 2015). We used the Western blot to explore protein levels and determine how muscone affected the changes of Akt1/PI3K pathway.

The P-Akt1 level increased in the TBI group compared with the sham group; however, adding LY294002 counteracted the muscone’s protective effect. The phosphorylated PI3K(p-PI3K), Akt1, and phosphorylated Akt1 (p-Akt1, at the Thr308 site) protein expression were similar (Fig. 6).

Muscone inhibited the activation of the NF-kB pathway

LY294002 was used to block the Akt1 activity involved in regulating NF-kb in order to investigate the underlying molecular mechanism. According to quantitative results, both the phosphorylated NF-kB/p65 (p-p65) and NF-kB/p65 were significantly inhibited in the muscone-treated groups compared to those in the TBI group. We observed lower p-65 and p65 protein levels in the TBI + muscone + LY294002 group. The results indicated that LY294002 probably increased phosphorylated NF-kB/p65 and NF-kB/p65 when compared with the muscone + TBI groups through the PI3K/Akt1 pathway (Figs. 7A, 7E and 7F) (P < 0.05).

Muscone reduced p53 expression 168 h after TBI

P53 affects the NF-kB pathway and acts as a key node to down-regulate neuroapoptosis (Li et al., 2018). Our results showed that the apoptosis-related levels (BCL-2, bax and cleaved caspase 3) were significantly altered 168 h after TBI compared to the sham group. Moreover, expression of these proteins were improved by muscone and offset by LY294002 (Figs. 7A–7D).

To further examine the effects of muscone on apoptosis, we conducted a Western blot analysis of the p53 expression. Quantitative analysis determined that there was a significant change between the TBI and sham groups. The p53 levels were significantly inhibited upon the injection of muscone compared to the TBI group. After TBI, higher levels of p53 were found in rats that were injected with LY294002 + muscone,compared to those in the TBI + muscone group (Fig. 7A and 7G) (P < 0.05). This suggests that muscone may ameliorate p53-mediated apoptosis through the PI3K/Akt pathway.