CKI, Sophora flavescens Ait. (SF)-free CKI, Heterosmilax yunnanensis Gagnep. (HY)-free CKI, and macrozamin (CAS^#6327-93-1) were prepared by Beijing Zhendong Guangming Pharmaceutical Research Institute (Beijing, China) and authenticated by the quality controller Qin Li. Compound 48/80 (C48/80, Cat^#C2313), propranolol, triprolidine and Evans blue were from Sigma-Aldrich (St Louis, MO, United States). Cimetidine and ginkgolide B were from Tokyo Chemical Industry (Tokyo, Japan). Metergoline, icatibant, rupatadine fumarate and meclofenoxate (Mecl) were from MedChemExpress (Monmouth Junction, NJ, United States). TSI-01 was from GLPBIO Co. (Montclair, CA, United States). SB290157 and PMX53 were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, United States) and Tocris Bioscience (Bristol, United Kingdom), respectively. MoTP was from Abcam (Cambridge, United Kingdom). ELISA kits for mouse total IgE (tIgE) and mouse mast cell protease 1 (MMCP1) were from Biolegend (San Diego, CA, United States) and Invitrogen (San Diego, CA, United States), respectively. PAF ELISA kit was from BioVision Incorporated (Milpitas, CA, United States). Imject™ aluminium adjuvant was obtained from Thermo Scientific (Waltham, MA, United States). Matrine (CAS^#519-02-8) and oxymatrine (CAS^#16837-52-8) were from National Institutes for Food and Drug Control (Beijing, China). Sophocarpine (CAS^#145572-44-7), sophoridine (CAS^#6882-68-4) and oxysophocarpine (CAS^#26904-64-3) were from Chengdu Herbpurify Co., LTD (Chengdu, China). Piscidic acid (CAS^#469-65-8) and trifolirhizin (CAS^#6807-83-6) were obtained from Institute of Chinese Materia Medica China Academy of Chinese Medical Sciences (Beijing, China) and Guangzhou Langou biotech Co. LTD (Guangzhou, Guangdong, China), respectively. Shrimp tropomyosin (ST) was extracted from Penaeus japonicus and purified using an isoelectric precipitation method as we previously described (Gao et al., 2017). All other reagents were of analytical grade.

Rat basophilic leukemia cell line (RBL-2H3) and human monocytic cell line (THP-1) were obtained from Center for Excellence in Molecular Cell Science of Chinese Academy of Sciences (Shanghai, China). Human mast cell line LAD2 (Columbia University, United States) was presented by Prof. Renshan Sun (the Third Military Medical University, Chongqing, China). Whole blood cells were obtained from BALB/c mice (female, 18–20 g) which were from Vital River Experimental Animal Services (Beijing, China) and housed in a vivarium under standard conditions of temperature and humidity and with a 12 h light/dark cycle.

All animal care and experimental protocols and procedures were approved by the Committee for Care and Welfare of Laboratory Animals in Institute of Medicinal Plant Development of Chinese Academy of Medical Sciences and Peking Union Medical College. Animal experiments were reported in compliance with the ARRIVE guidelines. Anesthetic drug (isoflurane) and all other necessary measures were used to reduce animals suffering during experimental procedures.

The constituents in CKI, SF-free CKI and HY-free CKI were assayed by a Waters HPLC system. A high resolution HPLC column (Waters XSelect CSHTM C18, 5 μm, 4.6 mm × 250 mm) was used. Column temperature was set at 30°C. Mobile phases were 0.2% (v/v) KH₂PO₄ in water (pH 3.0) and CH₃OH with a gradient elution as listed in Table 1. The flow rate was 0.6 ml/min and the detection wavelength was 211 nm.

To amplify Th2 response, aluminium adjuvant was used during mouse immunization. The mice were intraperitoneally injected with ST (60 µg/mouse) or CKI (50 µl/mouse) or equivoluminal normal saline (NS) in the presence of aluminum adjuvant (100 µl/mouse) once a week. Seven days after the fifth immunization, serum tIgE and MMCP1 were determined using the commercial ELISA kits. ST was used as a positive control.

Evans blue extravasation in mouse hind paw was measured as previously described (McNeil et al., 2015). Briefly, the mice were injected (i.v.) with Evans blue (0.65 μmol). Five minutes later, one paw was intraplantarly injected with CKI (10 μl) or C48/80 (0.3 μg/paw), and the other paw was injected with 10 μl of NS. Thirty minutes later, the mice were euthanized. Paw tissues were collected and Evans blue in the paw tissues was extracted by DMF (CAS^#68-12-2) at 50°C overnight (>20 h). Optical density (OD) values were read at 620 nm. C48/80 was used as a positive control. For local antagonist experiment, 10 μl of antagonist or NS was intraplantarly injected into the paw 10 min before Evans blue injection. For systemic antagonist experiment, the mice were intravenously injected with PAF receptor antagonist or NS 30 min before Evans blue injection.

To increase the severity of anaphylaxis, the mice were pretreated (i.v.) with propranolol (0.118 μmol/mouse) (TenBrook et al., 2004; Khodoun et al., 2009) 20 min before CKI (1.25–5 ml/kg) or C48/80 (4 mg/kg) challenge (i.p.). The mice in the negative control group were received equivoluminal NS. Thirty minutes later, the rectal temperature was measured (Gao et al., 2021).

Cells were incubated with test substance at 37°C for 1 h. The cell culture supernatant was collected and centrifuged (1,000 × g) at 4°C for 20 min. The supernatant was used for PAF assay by using commercial PAF ELISA kit. PAF concentration was calculated according to the standard curve. In the cell-free experiment, RBL-2H3 cells were lysed on ice by ultrasonication and the obtained cell lysates were centrifuged (5,000 × g) at 4°C for 5 min. The supernatant (85 μg protein, 80 μl/well) was incubated with CKI or matrine at 37°C for 1 h in the presence or absence of blocker. PAF level was determined according to the above description.

The data were reported as the mean ± SD from a representative experiment. All of the experiments reported in this work were repeated at least three times with the same pattern of results. All data were analyzed by GraphPad Prism 8.0 using one-way ANOVA followed by the Tukey posttest. A student’s t-test was used when only two groups were compared. p < 0.05 was considered significant.

The main constituent profiles of CKI, HY-free CKI and SF-free CKI were analyzed by HPLC-UV, respectively. The retention time values of the identified constituents were compared with that of the reference substances. Eight constituents (macrozamin, matrine, sophocarpine, sophoridine, oxysophocarpine, oxymatrine, piscidic acid, trifolirhizin) were identified (Figure 1), and six of them were quantified (Table 2).

Clinical data (January 1996 - February 2020) showed that approximately 70% of CKI-induced ADRs occurred within 30 min after injection, belonging to typical IHRs (Dian et al., 2020). In contrast to IgG/FcγRIII-mediated IHR, IgE/FcεRI-mediated IHR is more common because it is easier to be triggered (MacGlashan, 2012; Finkelman et al., 2016; Montañez et al., 2017). If CKI-induced IHR is through IgE/FcεRI signaling, serum tIgE and MMCP1 (another specific marker for IgE-mediated mast cell activation (Khodoun et al., 2011)) levels should be significantly increased. We immunized the mice for 5 weeks by using a mixture of CKI and aluminum adjuvant (Gao et al., 2018). As a result, the positive control ST markedly elevated serum tIgE and MMCP1 levels after continuous intraperitoneal immunization, while CKI could not increase these two markers (Figure 2), demonstrating that CKI-induced IHR was independent of IgE/FcεRI-mediated signaling.

In addition to IgE/FcεRI-mediated IHR, NIN-IHR is another important type of IHRs. In contrast to IN-IHRs, NIN-IHRs can be triggered after a single exposure to stimulus without the participation of antibody. To evaluate whether CKI could cause NIN-IHRs, local and systemic in vivo models were used. As shown in Figures 3A,B, whether positive control C48/80 (0.3 μg/paw) or CKI (10 μl/paw) significantly induced Evans blue extravasation of mouse paw after a single intraplantar injection. We next evaluated whether CKI could cause anaphylactoid shock (detected as hypothermia). To increase the sensitivity, mice were pretreated with propranolol which does not induce anaphylaxis by itself (TenBrook et al., 2004; Khodoun et al., 2009). As shown in Figure 3C, C48/80 led to a marked decrease of rectal temperature, and CKI (2.5 ml/kg - 5 ml/kg) also caused obvious hypothermia in the propranolol-pretreated mice. These results demonstrate that CKI can induce NIN-IHRs.

NIN-IHR can be triggered by many mediators, such as histamine, complement-derived anaphylatoxin 3 or 5 (C3 or C5), bradykinin, 5-hydroxytryptamine and PAF, etc. (Vliagoftis et al., 1990; Muñoz-Cano et al., 2016). To unveil the underlying mechanism for CKI-induced NIN-IHR, a series of antagonists were used one by one. The obtained results showed that CKI-caused Evans blue leakage was not blocked by histamine H1/2 receptor antagonists (triprolidine and cimetidine), C3 or C5 antagonist (SB290157 and PMX53), bradykinin B2 receptor antagonist (icatibant) and 5-hydroxytryptamine receptor antagonist (metergoline) (Figure 4A). Finally, intraplantar injection of three PAF receptor antagonists (rupatadine, ginkgolide B and MoTP) significantly countered CKI-caused Evans blue leakage (Figure 4B). To further confirm the above results, the effect of systemic administration of PAF receptor antagonist was observed. As expectedly, except for the lowest dosage (2 μmol/kg), the other two dosages (10–20 μmol/kg) of rupatadine completely antagonized CKI-caused Evans blue leakage (Figure 4C). These results indicate that CKI-caused NIN-IHR is mediated by PAF.

PAF is a short half-life but highly potent phospholipid which can be synthesized by a variety of cells (e.g., platelets, monocytes, neutrophils, basophils, and mast cells, etc.) in response to various stimuli (Chao and Olson, 1993; Gill et al., 2015). Given that CKI is an intravenous preparation, we first evaluated the effect of CKI on whole blood cells. As shown in Figure 5A, CKI was able to stimulate whole blood cells to release PAF in a concentration-dependent manner. Next, monocytes (THP-1), basophils (RBL-2H3) and mast cells (LAD2) were also subjected to CKI stimulation. Similarly, CKI could also promote PAF production in these three cells (Figures 5B–D), showing that its effect was non-cell-selective. Since RBL-2H3 cell line is more convenient to be cultured, we chose it in the subsequent active constituent and mechanism studies.

CKI is a bis-herbal formulation. To identify which one is the prime culprit for CKI-induced NIN-IHRs, we prepared HY-free CKI and SF-free CKI. The obtained data showed that HY-free CKI, rather than SF-free CKI, markedly caused Evans blue leakage in the paw after the first intraplantar injection (Figures 6A,B). Consistently, in the anaphylactoid shock model, HY-free CKI was able to significantly lower mouse rectal temperature (Figure 6C). These findings indicate that SF but not HY contributes to NIN-IHRs.

The above results have demonstrated that SF was responsible for CKI-induced IHRs (Figure 6) and SF-alkaloids were the main constituents in CKI (Table 2). Thus, we prepared SF main alkaloids solution (SFMAS) through mixing five major alkaloids (matrine, oxymatrine, sophocarpine, sophoridine and oxysophocarpine) according to their respective proportions in CKI, and then compared its effect with CKI at equivalent concentrations. Unexpectedly, SFMAS exerted a stronger effect on PAF production (Figure 7A). To evaluate the contribution of six quantified constituents (Table 2) to IHR, we next compared their effects (at their respective concentrations equivalent to 10% CKI) with CKI. As a result, matrine and oxymatrine could significantly promote PAF production (Figure 7B). More unexpectedly, the capacity of matrine, the second abundant alkaloid in CKI, to promote PAF production even exceeded that of CKI (Figure 7B). These results strongly suggest that counteractive substances may exist in CKI, and SF-alkaloids are the prime culprits for CKI-induced NIN-IHRs.

PAF is produced through rapid synthesis in response to stimuli and is not stored intracellularly (Gill et al., 2015). Our previous finding also demonstrated that there was undetectable PAF inside RBL-2H3 cells (data not shown). An important question that arose was whether CKI-caused PAF production depended on the transmembrane signal. To address this issue, we determined the effect of CKI on PAF production in RBL-2H3 cell lysates. As a result, when the transmembrane signal pathway was cut off, the effect of CKI or matrine was still present, or rather stronger (Figure 8A), suggesting that CKI or matrine may directly activate certain synthetase(s) of PAF. To our knowledge, PAF can be synthesized via two pathways, remodeling pathway or de novo pathway (Gill et al., 2015). We next investigated which one could be activated by CKI or matrine. As shown in Figure 8B, the blocker of de novo pathway (Mecl), rather than remodeling pathway’s (TSI-01), could counteract the effects of CKI and matrine, indicating that they were through activating de novo pathway to promote PAF production.