Radix Paeoniae Rubra were purchased from Beijing Tonrentang Traditional Chinese Herbs Co. Ltd. The standard chemicals albiflorin, benzoylpaeoniflorin, oxypaeoniflorin, paeonol, and paeoniflorin were purchased from J&K Chemical Co. Ltd. Radix Paeoniae Rubra were soaked in DI water for 30 min before use. Then the slices were heated at 100°C and boiled until boiling and boiled for another 30 min. This process was repeated to obtain the decoction of Paeoniae Radix aqueous extract. The powder of the extract was obtained by vacuum drying as previously reported (Liu et al., 2020a). The main chemical constituents of Radix Paeoniae Rubra, including albiflorin, benzoylpaeoniflorin, oxypaeoniflorin, paeonol, and paeoniflorin, were all purchased from Beijing J&K Scientific Ltd.

S. aureus α-toxin (Sigma-Aldrich, 50 μg/ml) or the supernatants of S. aureus ATCC29213 (1 × 10⁹ CFU/ml, collection from 6 h growth bacteria) were incubated with compounds for about 1 h. Then mixtures of samples were added into 5% fresh rabbit blood cells for 30 min to detect the hemolytic activity. The positive control of anti-hemolytic activity of α-toxin used was MED14983^* (Sigma-Aldrich, 5 μg/ml). The hemolysis rates of compounds were detected by a plate reader (Tecan Infinite 200pro) and calculated based on previous protocol (Liu et al., 2017, 2020d) as below,

Hemolysis rate (%) = [OD_{450 sample} – OD_{630 nm blank}] / [OD_{450 0.1% Triton X-100} – OD_{630 nm blank}] × 100%

C18 column, Inertsil ODS-3 (4.6 × 250 mm, Particle size 5 μm); Mobile phase:water (A)-acetonitrile (B), Gradient elution procedure: 0–10 min, 95% A, 10–20 min, 95-90% A, 20–30 min, 90-80% A, 30–35 min, 80–100% A. Flow rate: 1 ml/min, UV detector, Detection wavelength: 230 nm; Injection volume 50 μ; Column temperature 25°C.

High performance liquid chromatography mobile phase methanol (97%), mobile phase flow rate: 0.8 ml/min; Sample dosage: 10 L; Detection wavelength: 270 nm; Column temperature: 30°C. Electrospray mass spectrometry (ESMS) was performed using a positive and negative ion mode. Mass charge ratio (M/Z) detection range: 50~1000; Spray voltage: 4500 V, sheath gas and auxiliary gas are nitrogen, metal capillary temperature: 260°C, metal capillary voltage 20 V.

The analysis of FTIR were carried out by Shanghai WEIPU Chemical Technology Service.¹

Structure Analysis involved drying the extracts and sending them to Shanghai Microspectral Analysis and Testing Center for Fourier Transform Infrared Spectrometer (FTIR; United States Thermo Scientific Nicolet IS5) detection and Nuclear Magnetic Resonance (NMR) analysis (German Bruker Avance III400MHZ).

S. aureus α-toxin (50 μg/ml) was treated with PRAE at the concentrations of 0.06, 0.15, 0.25, 0.5, 1, and 2 mg/ml for 30 min. Sodium deoxycholate (NaDC, 5 mM) was used as the control to prevent α-toxin dimerization. PRAE (0.15, 0.5, and 2 mg/ml) interacted with α-toxin and was treated with or without NaDC for 30 min. Then the proteins were separated by native polyacrylamide gel electrophoresis with 8% polyacrylamide gels without β-mercaptoethanol and DDT. Finally, the bands were identified by silver staining method.

Mice (five-weeks-old) were purchased from the Academy of Military Medical Sciences, Beijing, China (Certificate Number: SCXK-PLA 2012-0004). All animal experiments were approved by the Institutional Animal Care and Use Committee at the Academy of Military Medical Sciences Institute (Beijing, China; approval no. SYXK2014-0002).

4-week-old female ICR mice (~30 g) were randomly divided into four groups (n ≥ 10 every group). Intranasal infection of S. aureus (10⁹ CFU/g) were the infected groups. No infected mice served as the negative control. The mice treated by MED14893* and α-toxin were used as positive control. PRAE-a (1 mg/g) was treated with S. aureus infected mice. Mice were also treated with S. aureus α-toxin (50 ng/g). After infection and treatment, the mouse lung area was determined and recorded and the lung tissues of the mice were collected to detect the concentrations of α-toxin by ELISA. The numbers of S. aureus were counted by colony count technique (CFUs), according to previously published methods (Liu et al., 2020b,c). All mice both dead and alive were used for analysis.

Statistical analysis was performed using GraphPad Prism 8.0 software. Results were expressed as means ± SD and p-values were calculated using unpaired t-test between two groups or one-way ANOVA among multiple groups. All mice were used for analysis. All experiments were performed at least in triplicate.

The main active ingredients of the Paeoniae Radix extract include albiflorin, benzoylpaeoniflorin, oxypaeoniflorin, paeonol, and paeoniflorin (Figure 1). We found that Paeoniae Radix aqueous extract (PRAE) could inhibit the hemolytic function of bacterial toxins. However, the specific antitoxic mechanisms of PRAE and the structure of active components in the extract are not clear. Thus, to analyze the compounds in PRAE, HPLC was employed to separate the extract. Results showed that there are three main peaks in the sample HPLC spectrum and we named them PRAE-a (24.773 min), PRAE-b (36.832 min), and PRAE-c (45.423 min; Figure 2A). The retention time of PRAE-a was similar to that of paeoniflorin (23. 5 min), while the retention time of PRAE-b and PRAE-c was different from that of other chemical components as demonstrated by the fact that the peak times of hydroxypaeoniflorin, paeonol, albiflorin, and benzoylpaeoniflorin were at 14.0, 51.5, 21.3, and 49.9 min, respectively, (Figure 2A). We further discovered that PRAE-a might be paeoniflorin by comparing the standard drugs of active ingredients in the extract, which include albiflorin, benzoylpaeoniflorin, oxypaeoniflorin, paeonol, and paeoniflorin, because PRAE-a and paeoniflorin had the closest peak times at 24.773 and 23.485, respectively (Figures 2A,B). And we further discovered that PRAE-a had the best antihemolytic activity of α-toxin compared to these standard drugs (Figure 2B). To further investigate the antihemolytic function of PRAE-a, we prepared plenty of PRAE-a from original drugs (Figure 2C). We employed specific neutralizing antibodies of α-toxin (MED14893^*) as the positive control and used supernatants of S. aureus to detect the antihemolytic activity of PRAE-a. Results showed that PRAE-a have similar antihemolytic activity effects as MED14893^* (Figure 2D).

To further characterize the structure of PRAE-a, mass spectrometry (MS) analysis was used to detect the functional groups of the component. MS images showed that PRAE-a is similar to the structure of paeoniflorin (Figure 3A), as shown in the peaks of [M + Na]⁺ at m/z = 503 and [M + K]⁺ at m/z = 519 from MS⁺ image (upper) and the peak of [M+HCOOH-H]⁻ at m/z = 525 from MS⁻ image (bottom). Meanwhile, Fourier transform infrared spectrometer (FTIR) analysis showed that PRAE-a had a paeoniflorin-like structure through detecting the hydroxy, carbonyl group, benzene ring, and ether bond at the wavelength of 3428, 1712, 1602.32, and 1279 cm⁻¹, respectively (Figure 3B). Interestingly, we found that PRAE might contain a sugar structure due to the similar waveform to brown sugar methanol extract (Figure 3C). Moreover, gas chromatography mass spectrometry (GC-MS) analysis indicated that the structure of benzoate and the structural fragment of paeoniflorin might be derived from PRAE-a, while the carbohydrate might be derived from sucrose and glucose (Figure 4). The analysis of PRAE-a by ¹H- and ¹³C-nuclear magnetic resonance (NMR) showed that there existed sugar components in PRAE-a (Figures 5A,B). Subsequently, the comparison of the ¹H-NMR spectrum between glucose and saccharose or PRAE-a showed that the proportion of glucose, saccharose, paeoniflorin, and paeoniflorin derivatives were approximately 72-74%, 9-11%, 12-14%, and 4-6%, respectively (Figures 5C,D and Table 1). It was exhibited that PRAE-a had a paeoniflorin-like structure and had a different carbon sequence (red) from paeoniflorin (Figure 5E). Altogether, the results illustrated that PRAE-a had a paeoniflorin-like structure and it could be deduced that the carbohydrate from PRAE-a was necessary.

Previous studies revealed that the hemolytic toxin monomer needed to assemble into heptamer on the target cell membrane to achieve perforation damage (von Hoven et al., 2019). To further investigate how PRAE-a abolished the hemolytic activity of α-toxin, we used native polyacrylamide gel electrophoresis to detect the assembly of α-toxin under PRAE-a treatment at a protein level. The protective agent of endotoxin, sodium deoxycholate (NaDC), was employed as positive control to promote the self-assembly of α-toxin monomer into oligomers. We found that administration of NaDC alone promoted that polymerization of hemolytic toxin monomers to form 2, 4, 6, and 7-oligomer. However, when induced by additional PRAE-a, it was observed that with increasing PRAE-a concentration, the formation of dimers was significantly increased, but the formation of high oligomers was significantly decreased with a significant dose dependence (1 or 2 mg/ml, Figure 6A). In addition, the hemolytic toxin had a tendency to self-assemble without the addition of NaDC, but the ability to self-assemble into high oligomers was markedly decreased with the addition of PRAE (Figure 6B). These results illustrated that the mechanism of PRAE-a abolished the hemolytic activity of α-toxin by destroying α-toxin to lead to accumulation of α-toxin dimer, which is useless for pore forming of α-toxin on the cellular membrane (Figure 6C).

We next sought to evaluate the therapy of PRAE-a for S. aureus infection. A pulmonary infection model in mice demonstrated that α-toxin facilitated the death of S. aureus-infected mice (Figure 7A), whereas PRAE-a treatment increased the survival of the infected mice (Figure 7B). In addition, PRAE-a could effectively ameliorate lung enlargement in the infected mice (Figure 7C). The concentration of α-toxin in the lung decreased after PRAE-a treatment (Figure 7D), which was consistent with the result that PRAE-a directly inhibited α-toxin (Figure 3). In addition, PRAE-a prevented mice from developing tissue necrosis, inflammatory cell infiltration, and alveolar rupture in lung caused by α-toxin of S. aureus in the infected mice (Figure 7E). And PRAE-a might have an anti-inflammatory effect by decreasing α-toxin-induced dust cell increase (Figure 7F). We also found that PRAE-a effectively reduced the numbers of S. aureus in the lung (Figure 7G). Results suggested that the bacterial loading in the lung were controlled by PRAE-a treatment and it was tightly associated with the direct inhibition of α-toxin by PRAE-a, because α-toxin could promote bacterial lung infection (Cohen et al., 2016). To further investigate the injury of the lung caused by α-toxin, mice were pretreated with α-toxin orally and the therapeutic effect of PRAE-a was evaluated; we found that both PRAE-a and positive control MED14893^* could inhibit the levels of α-toxin in the lung, but only PRAE-a decreased the lung injury of α-toxin (Figure 7D). Taken together, PRAE-a could be used as a drug candidate for the therapy of the α-toxin-caused pulmonary S. aureus infection.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.