Safflower yellow in Carthami Flos is responsible for Xuebijing Injection-induced immediate hypersensitivity reaction through activating complement C3

Abstract

Objective

Xuebijing Injection (XBJI) is mainly used for treating sepsis in China, and even COVID-19 recently. This study aimed to clarify the molecular mechanism(s) and identify the potential “common culprit(s)” for XBJI-caused immediate hypersensitivity reaction (IHR) which is the main type of its adverse reactions.

Methods

Antiserum against XBJI was prepared by intraperitoneal immunization in combination with aluminum adjuvant for five weeks. Antagonistic experiments were performed by using several antagonists against different mediators in Evans Blue leakage model. Propranolol-pretreated mice were used to determine the capacity of XBJI to trigger systemic IHR. Serum total IgE (tIgE) and mouse mast cell protease 1 (MCPT-1) levels, complement activation, and the levels of supernatant inflammatory mediators were determined by ELISAs. Lipopolysaccharide (LPS)-activated RAW264.7 macrophages were used for evaluating the anti-inflammatory activity of XBJI, while human mast cells (LAD2) were used for assessing the effect of XBJI on mast cell degranulation.

Results

Continuous treatment (i.p.) with XBJI along with aluminum adjuvant did not elevate the levels of serum tIgE and MCPT-1. In vitro, XBJI could not directly cause the degranulation of LAD2 cells. It induced a robust Evans Blue leakage after the first injection in mouse paw. Mechanism study demonstrated that antagonists for histamine H1/H2 receptors and complement C3a receptor counteracted XBJI-induced IHR. XBJI also directly activated complement C3 in human serum. Through screening five herbs of XBJI and the constituents, only safflower yellow (SY) in Carthami Flos was able to induce IHR. The discolored-XBJI not only did not induce IHR locally and systemically, but also could suppressing the production of proinflammatory mediators in LPS-activated RAW264.7 macrophages.

Conclusion

XBJI failed to induce immune IHR, but potently triggered non-immune IHR through direct activating complement C3 to provoke histamine release. SY in Carthami Flos was the underlying “common culprit” responsible for XBJI-caused IHR. The anti-inflammatory action of XBJI can be retained after decolorization. Our study provides a scientific basis for not only preventing and treating XBJI-caused IHR clinically, but also improving its production process.

1. Introduction

Xuebijing Injection (XBJI) is an herbal prescription consisting of five traditional Chinese herbs, including Carthami Flos (Honghua; Carthamus tinctorius L.), Paeoniae Radix Rubra (Chishao; Paeonia lactiflora Pall.), Chuanxiong Rhizoma (Chuanxiong; Ligusticum chuanxiong Hort.), Salviae Miltiorrhizae Radix et Rhizoma (Danshen; Salvia miltiorrhiza Bge.) and Angelicae Sinensis Radix [Danggui; Angelica sinensis (Oliv.) Diels]. In 2004, XBJI had been licensed by the National Medical Products Administration (NMPA, China) for the treatment of sepsis and multiple organ dysfunction syndrome. In 2020, it was further approved for the treatment of severe/critical systemic inflammatory response syndrome or/and multiple organ failure of COVID-19.

Recently, a large double-blind, placebo-controlled, multicenter, randomized clinical trial was conducted to determine the efficacy of XBJI for 1 817 patients with sepsis. The data indicated that treatment with XBJI resulted in a significantly lower 28-day mortality rate (18.8%), compared with the placebo group (26.1%) (Liu et al., 2023). Along with its benefits, a deep concern about serious adverse drug reactions (ADRs) was also raised because there were numerous serious adverse events occurred during the study (about 22% of the patients died) but the causal analysis was missing (Unger & Clissold, 2023). In fact, the first case of XBJI-caused allergy was reported as early as 2007 (Wang & Feng, 2007). Subsequently, with the increased application, there were quite a few clinical case reports about the ADRs of XBJI (Wang et al., 2018, Zheng et al., 2019). The manifestations of XBJI-caused ADRs often involve multiple systems (e.g., skin pruritus, erythra, chest tightness, fever, labored breathing, erubescence, diarrhea, shiver, dizziness), of which approximately 49.13%−78.57% of ADRs occurred within 1 h after administration (Zheng et al., 2019; Wei, Liu, Wu, & Cheng, 2022), which belongs to typical immediate hypersensitivity reaction (IHR).

IHR can be triggered by a series of vasoactive mediators, such as histamine, complement-derived anaphylatoxin, platelet-activating factor (PAF), leukotrienes, and bradykinin (BK) (Gao et al., 2021). Drugs are one of the most common inducers of IHR. Usually, signs and symptoms of drug-induced IHR occur within 6 h after drug exposure and may include cutaneous, respiratory, cardiovascular, gastrointestinal symptoms, or anaphylaxis (Dykewicz & Lam, 2020). According to the pathogenesis, IHR can be classified as either immune (IN-IHR) or non-immune reactions (NIN-IHR). Traditional IN-IHR is an IgE-mediated immunologic reaction (type I allergy) based on the cross-link of IgE and its high-affinity receptor Fc epsilon receptor I (FcεRI) expressed on mast cells. In contrast to IN-IHR, NIN-IHR can occur independent of antigen–antibody reaction through the direct stimulation to induce histamine release (Quan, Sabaté-Brescó, Guo, Martín, & Gastaminza, 2021) or PAF production (Gao et al., 2021), activate complement (Gao et al., 2018, Gao et al., 2019) or contact system (Gao et al., 2020) and so on (Muñoz-Cano, Picado, Valero, & Bartra, 2016), thus leading to anaphylactoid symptoms. In the present study, we clarified the molecular mechanism(s) and identified the potential “common culprit(s)” for XBJI-induced IHR. Our finding can provide a scientific basis for the safe use of XBJI clinically.

2. Materials and methods

2.1. Materials and reagents

Commercial XBJI (Cat^#Z20040033; Lot^#2108051) was from Tianjin Chase Sun Pharmaceutical Co., Ltd. (Tianjin, China). Carthami Flos (Lot^#BJ-20221103-HH31), Paeoniae Radix Rubra (Lot^#BJ-20221103-SC126), Chuanxiong Rhizoma (Lot^#BJ-20221103-XC77), Salviae Miltiorrhizae Radix et Rhizoma (Lot^#BJ-20221103-D5), and Angelicae Sinensis Radix (Lot^#BJ-20221103-G36) were obtained from Anguo Chinese Medicinal Herbs Market (Baoding, Hebei, China) and authenticated by Prof. Yulin Lin (Institute of Medicinal Plant Development of Chinese Academy of Medical Sciences). The voucher specimens (XBJI-2022-001) were deposited in our herbarium. Compound 48/80 (C48/80; Cat^#C2313-250MG; Lot^#102M4086V), Evans Blue (Cat^#E2129; Lot^#20150611024), lipopolysaccharide (LPS; Cat^#L4130-100MG; Lot^#0000130726) and N^G-nitro-L-arginine methyl ester (L-NAME; Cat^#N5751-1G; Lot^#107K1055) were from Sigma-Aldrich (St Louis, MO, USA). Cimetidine (Cat^#C1252; Lot^#EWE7J-DI) and ginkgolide B (Cat^#G0553; Lot^#TTLDK-FO) were from Tokyo Chemical Industry (Tokyo, Japan). Metergoline (Cat^#HY-B1033; Lot^#34235) and icatibant (Cat^#HY-17446; Lot^#13421) were from MedChemExpress (Monmouth Junction, NJ, USA). SB290157 (Cat^#GC11800; Lot^#2) and cell counting kit-8 (CCK-8; Cat^#GK10001; Lot^#42) were purchased from Good Laboratory Practice Bioscience (GLPBIO; Montclair, CA, USA). PMX53 (Cat^#5473; Lot^#6A) was obtained from Tocris Bioscience (Bristol, UK). Dulbecco’s modified Eagle Medium (DMEM; Cat^#12100-04; Lot^#2290467) and StemPro-34 serum-free medium (SFM; Cat^#10640-019; Lot^#2514996) were produced by Gibco (Grand Island, NY, USA). Enzyme-linked immunosorbent assay (ELISA) kits for mouse total IgE (tIgE; Cat^#432401; Lot^#B344000), interleukin-6 (IL-6; Cat^#431301; Lot^#B371164), and tumor necrosis factor-alpha (TNF-α; Cat^#430901; Lot^#B357292) were from Biolegend (San Diego, CA, USA). Mouse mast cell protease 1 (MCPT-1; Cat^#88-7503-22; Lot^#343824-003) ELISA kits were from Invitrogen (San Diego, CA, USA). Griess reagent kit (Cat^#S0021S; Lot^#061219200601) was from Beyotime Biotechnology (Shanghai, China). Diphenhydramine hydrochloride (Cat^#S61775; Lot^#L12J7X8944) and dexamethasone (Cat^#S17003; Lot^#Y07D6C7163) were from Yuanye Biotechnology (Shanghai, China). Human C3a ELISA kit (Cat^#550499; Lot^#2264716) and normal human serum (Cat^#SL010; Lot^#20230605) were obtained from BD Biosciences (San Diego, CA, USA) and Solarbio Life Sciences Co., Ltd. (Beijing, China), respectively. Shrimp tropomyosin (ST) from Litopenaeus vannamei was prepared as we previously described (Gao et al., 2021).

2.2. Cells and animals

The human mast cell line LAD2 (from Michael D. Gershon, MD, Columbia University, USA) was the generous gift of Prof. Renshan Sun (the Third Military Medical University, Chongqing, China) and cultured in StemPro-34 SFM supplemented with L-glutamine (2 mmol/L) and recombinant human stem cell factor (rhSCF) (100 ng/mL) in a humidified incubator (5.0% CO₂, 37 °C). The RAW264.7 cells were obtained from American Type Culture Collection (ATCC, Rockville, MD, USA) and cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) in a humidified incubator (5.0% CO₂, 37 °C). Cell passage number was kept below ten. All animals were purchased from Vital River Experimental Animal Services [Beijing, China; license number: SCXK (Beijing) 2021-0006] and were housed in a specific pathogen free (SPF) environment under standard condition with a 12 h light/dark cycle and suitable temperature and humidity.

2.3. Ethics statements

Animal study was reviewed and approved by the Institutional Care and Use Committee of the Institute of Medicinal Plant Development (IMPLAD) of Chinese Academy of Medical Sciences (ethical clearance number^#SLXD-20230420015). And all animal procedures were conducted in accordance with internationally accepted principles for laboratory animal use and care: National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978). During the experimental procedures, the animals were anesthetized with isoflurane (3%) inhalation and euthanized by cervical dislocation.

2.4. HPLC analysis

Commercial XBJI was assayed by an Essentia LC-15C HPLC system (Shimadzu, Kyoto, Japan) equipped with a UV/Vis detector (Shimadzu, Kyoto, Japan). A reverse-phase XBridge C₁₈ column (250 mm × 4.6 mm, 5 μm; Waters, Milford, MA, USA) was used. Gradient elution (program in Table 1) was performed with a mixture of mobile phase A [acetonitrile: methanol = 60: 40 (volume percentage), containing 0.5% glacial acetic acid] and mobile phase B (0.5% glacial acetic acid in H₂O). The flow rate was 1.0 mL/min, and the column temperature was kept at room temperature. The detection wavelength was 280 nm.

Table 1.

Gradient elution program of chromatographic separation.

Time (min)	Mobile phases
	A (%)	B (%)
0−5	5	95
5−90	5−45	95−55
90−92	45−100	55−0

A: acetonitrile/methanol = 60/40 (volume percentage) with 0.5% glacial acetic acid;

B: H₂O with 0.5% glacial acetic acid.

2.5. Preparation of XBJI and its five intermediate fractions

Self-prepared XBJI and its five intermediate fractions were made according to the National Drug Standard (YBZ01242004-2010Z-2012). The extract of Carthami Flos (F1) was extracted using the percolation method with 30% ethanol, and subsequently purified through an alcohol precipitation method. And the other four intermediate fractions (F2, the extract of Paeoniae Radix Rubra; F3, the extract of Chuanxiong Rhizome; F4, the extract of Salviae Miltiorrhizae Radix et Rhizome; F5, the extract of Angelicae Sinensis Radix) were prepared by water extraction and purified by alcohol precipitation, followed by extraction with water-saturated n-butanol. For the preparation of self-prepared XBJI, the five intermediate fractions (2.33 g F1; 290 mg F2; 90.4 mg F3; 164.2 mg F4; 95.6 mg F5) were proportionally mixed and dissolved in saline for injection to make 100 mL. All self-prepared injections were sterilized by filtration using 0.22 μm filter.

2.6. Preparation of safflower yellow (SY) and safflower red (SR)

SY was extracted from Carthami Flos using a water extraction method, while SR was extracted and purified through alkali extraction followed by acid precipitation, as previously described (Lech et al., 2021, Ma and Terahara, 2008).

2.7. Preparation of antiserum against XBJI or ST

The animals were divided into three groups: vehicle group (the negative control), ST group (the positive control) and XBJI group. Aluminum adjuvant was used during mouse immunization to enhance the Th2 response. BALB/c mice (female, 18−20 g) received intraperitoneal injection of ST (3 mg/kg) or XBJI (10 mL/kg) or equivoluminal sterile saline along with aluminum adjuvant (5 mL/kg) once a week according to our previous study (Gao et al., 2018). Seven days after the fifth immunization, the mice were sacrificed, and the antisera were collected. Serum tIgE and MCPT-1 levels were measured using the commercial ELISA kits.

2.8. Passive sensitization and challenge

Passive systemic anaphylaxis (PSA) and passive cutaneous anaphylaxis (PCA) were performed according to a previously reported study with slight modifications (Gao et al., 2018). For the PSA, BALB/c mice (female, 18−20 g) were sensitized (i.v.) with XBJI antiserum (undiluted; 10 mL/kg) or ST antiserum (1: 20 dilution; 10 mL/kg). Twenty-four hours later, the mice were challenged (i.v.) with XBJI (10 mL/kg) or ST (10 mg/kg). The negative control (NC) mice were treated with an equal volume of vehicle. Four hours later, the whole blood samples were collected, and the serum MCPT-1 concentration was measured using an ELISA kit. For the PCA, SD rats (female, 160−180 g) received intracutaneous injections of 100 μL of XBJI antiserum (undiluted) or ST antiserum (1: 50 dilution). Forty-eight hours later, the rats were challenged (i.v.) with XBJI (10.5 mL/kg) or ST (3 mg/kg) containing Evans Blue (30 mg/kg). One hour after challenge, the rats were euthanized and the blue spots in the dorsal inboard skin were recorded.

2.9. Evans Blue leakage assay

The manifestations (generalized erythema, urticaria or angioedema) of IHR are a direct result of the actions of the vasoactive mediators (Wilkerson, 2022), which are released from mast cells or other effector cells and can lead to vasodilation and increasing capillary permeability. The current method for evaluating the changes in vascular permeability is based on the leakage of plasma protein-bound Evans Blue in the skin tissue. Briefly, BALB/c mice (male, 18−20 g) were injected (i.v.) with Evans Blue solution (47 mg/kg). Five minutes later, the left paw of mice was injected intraplantarly with XBJI (0.75 mL/kg) and the right paw was injected with equivoluminal vehicle. C48/80 (7.5 mg/kg) was used as a positive control. Thirty minutes later, the mice were euthanized and photographed. In the antagonist experiment, 10 μL of antagonist (2 mmol/L ginkgolide B, 12.4 mmol/L metergoline, 1 mmol/L icatibant, 6.9 mmol/L diphenhydramine, 40 mmol/L cimetidine, 2 mmol/L PMX53 or 3.8 mmol/L SB290157) or vehicle was intraplantarly injected into the paw 10 min before Evans Blue injection. The paw tissues were collected, and Evans Blue was extracted using N,N-dimethylformamide (DMF) overnight at 50 °C. The concentration of dye in paw tissue was determined by measuring the optical density (OD) values at 620 nm. The concentrations of the dye in the paw tissues were calculated by the standard curve of Evans Blue (Gao et al., 2021, McNeil et al., 2015).

For the rat cutaneous Evans Blue leakage assay (Gao et al., 2018), the rats (male, 160−180 g) received intracutaneous injections of 100 μL of test substances (1 mg/mL), followed by an immediate injection (i.v.) of Evans Blue solution (30 mg/kg). After 15 min, the rats were euthanized and the blue spots in the dorsal inboard skin were recorded. The diameter > 5 mm was considered as positive.

2.10. Anaphylactic shock assay

The capacity of XBJI to cause anaphylactic shock can be detected as hypothermia. To increase the severity of anaphylaxis, the BALB/c mice (half male and half female, 18−20 g) were pretreated (i.v.) with propranolol (1.53 mg/kg) which does not induce anaphylaxis by itself. Twenty minutes later, the mice were challenged with XBJI (45 mL/kg, i.p.) or received equivoluminal saline (negative control group). Thirty minutes later, the rectal temperature was measured (Gao et al., 2021).

2.11. β-Hexosaminidase release assay

To assess the direct impact of XBJI on cell degranulation, LAD2 cells (5 × 10⁴ cells/well) were cultured in 96-well plates and randomly divided into six groups: the negative control group (vehicle), the positive control group (10 μg/mL C48/80) and the experimental groups (3.13%, 6.25%, 12.50% and 25.00% XBJI). The cells were treated with XBJI at 37 °C for 1.5 h. (Gao et al., 2019). And the supernatant (30 μL) was transferred to a 96-well black flat-bottom plate with 50 μL of substrate solution (0.571 mg/mL 4-methylumbelliferyl N-acetyl-b-D-glucosaminide in the buffer contained 133 mmol/L sodium citrate and 133 mmol/L NaCl, pH 4.3). The reaction was stopped after 1.5 h at 37 °C by adding 180 μL/well stop buffer [50 mmol/L glycine and 5 mmol/L ethylenediaminetetraacetic acid disodium salt (EDTA-Na₂), pH 10.5; 200 μL/well]. Fluorescence was determined by a fluorescence microplate reader at λ_ex355nm/λ_em460nm (Gao et al., 2017).

2.12. Complement activation assay in vitro

The experiment was divided into five groups, namely, the negative control group (vehicle), the positive control group (1.00% Tween-80) (Weiszhár et al., 2012), and three treatment groups (1.00%, 2.00%, and 4.00% XBJI or F1 or SY or AHSYB). Normal human serum (1: 1000 dilution; 100 μL) was mixed with 4 μL of calcium chloride (CaCl₂; 1 mol/L), 4 μL of magnesium chloride (MgCl₂; 1 mol/L), and 10 μL of test substances, followed by incubation for 20 min at 37 °C. The reaction was stopped by adding 22 μL of 0.5 mol/L EDTA (pH 8.0) and cooling to 0 °C (Gao et al., 2019). Next, the reaction solution was transferred to the ELISA plate (100 μL/well) and determined according to the manufacturer’s instruction. The level of C3a was indirectly quantified by measuring the level of C3a-desArg, a stable derivative of C3a.

2.13. Cell viability assay

Cell viability of RAW264.7 cells was assayed by CCK-8 method. The cells were seeded in 96-well plates at a density of 4 × 10⁵ cells per well, and then randomly divided into five groups: the negative control group (vehicle), XBJI group, discolored-XBJI group, self-prepared XBJI group and SPs-free XBJI group. The cells were incubated with different XBJIs (0.63%−5.00%). Twenty-four hours later, 10 μL of CCK-8 reagent was added into each well for further 1 h. Absorbance was measured at 450 nm. The relative cell viability was calculated in contrast to the control group.

2.14. Measurement of inflammatory mediators

RAW264.7 cells (4 × 10⁵ cells per well) were pretreated with XBJIs (0.63%, 1.25%, and 2.50%) for 1 h. Subsequently, the cells were stimulated with LPS (10 ng/mL) for an additional 24 h. Supernatant nitrite (NO₂^–) was measured to indirectly represent NO level by Griess reaction, while other inflammatory mediators (IL-6 and TNF-α) were measured by ELISA kits according to the manufacturers’ instructions. L-NAME was used as the positive control for the measurement of NO levels, while dexamethasone was used as the positive control for determining other inflammatory mediators. Concentrations of the mediators were calculated based on their standard curves.

2.15. Statistical analysis

The data were displayed as mean ± SD and analyzed by GraphPad Prism version 9.3.0 (GraphPad Software, Inc., La Jolla, CA, USA). The investigators were blinded to experimental settings during data analysis. Differences among multiple groups were compared using One-way analysis of variance (ANOVA) with the Tukey’s post hoc analysis, while differences between only two groups were compared using Student’s t-test. A value of P < 0.05 was considered statistically significant.

3. Results

3.1. Fingerprint of XBJI and its constituent identification

Main composition profiles of XBJI were analyzed via HPLC-UV. Eight principal constituents from XBJI [salvianic acid A (1), 3,4-dihydroxybenzaldehyde (2), chlorogenic acid (3), hydroxysafflor yellow A (4), caffeic acid (5), paeoniflorin (6), ferulic acid (7), and senkyunolide I (8)] were identified by comparing the retention times (t_R) with their chemical standards (Fig. 1), of which five constituents were quantified (Table 2).

Fig. 1.

Chromatogram fingerprint of XBJI and chemical structures of eight principal constituents.

Table 2.

Concentrations of five constituents in XBJI.

Peak no.	Constituents	Concentration (μg/mL)
2	3,4-Dihydroxybenzaldehyde	11.40
4	Hydroxysafflor yellow A	391.93
6	Paeoniflorin	1 045.20
7	Ferulic acid	162.16
8	Senkyunolide I	37.28

3.2. XBJI fails to induce IgE-mediated IHR but can cause NIN-IHR

As the most common IN-IHR, IgE-mediated IHR is triggered by the binding of an allergen (antigen) to its IgE antibodies that bound to FcεRI on the pre-sensitized mast cells (Bradding and Cruse, 2008, Shamji et al., 2021, Sutton et al., 2008). To evaluate whether XBJI can induce IgE-mediated IHR, the mice were immunized with XBJI in combination with aluminum adjuvant to amplify IgE production. For the positive control, we chose ST that had been successfully prepared in our lab as reported previously (Gao et al., 2018). Five weeks after intraperitoneal immunization, ST remarkedly increased the levels of serum tIgE (398.5-fold increase) and MCPT-1 (4.6-fold increase; a specific marker for IgE-mediated mast cell activation) (Khodoun, Strait, Armstrong, Yanase, & Finkelman, 2011), while XBJI could not (Fig. 2A−B). To further confirm this finding, we next determined serum MCPT-1 level of the mice passively sensitized by the XBJI antiserum (undiluted) or ST antiserum (1: 20 dilution) and challenged with corresponding “allergen”. As shown in Fig. 2C, in contrast to ST (with a 3.5-fold increase of serum MCPT-1), XBJI challenge could not increase serum MCPT-1 level at all. Furthermore, in PCA assay (data not shown), the rats in ST group (ST antiserum with 1: 50 dilution + ST challenge) showed markedly blue spots in the dorsal inboard skin (diameter ≥ 9 mm), while the rats in XBJI group (undiluted XBJI antiserum + XBJI challenge) could not do. These results demonstrate that XBJI does not induce IgE-mediated IHR.

Fig. 2.

XBJI fails to induce IgE-mediated IHR but can cause NIN-IHR. (A−B) Effects of XBJI on serum tIgE and MCPT-1. Mean ± SD, n = 10; ^**P < 0.01 vs vehicle. (C) Changes of serum MCPT-1 levels in mice passively sensitized by antiserum and challenged by corresponding antigen. Mean ± SD, n = 6; ^**P < 0.01 vs vehicle. (D) Representative images of Evans Blue leakage. (E) Quantification of Evans Blue leakage into the paw. Mean ± SD, n = 8; ^**P < 0.01 vs vehicle.

Unlike IN-IHR, NIN-IHR can be triggered by a single exposure to a stimulus independent of antigen-antibody reaction. We next assessed whether XBJI could induce NIN-IHR using Evans Blue leakage assay (Gao et al., 2021). As shown in Fig. 2D−E, similarly to C48/80, XBJI could also induce a significant Evans Blue leakage after the first intraplantar injection, showing that XBJI indeed can trigger NIN-IHR.

3.3. Histamine H1/H2 receptor antagonists counter XBJI-induced NIN-IHR

There are many mediators that can trigger NIN-IHR, such as histamine (Gao et al., 2019), PAF (Gao et al., 2021), BK (Gao et al., 2020), and 5-hydroxytryptamine (5-HT) (Bruhns & Chollet-Martin, 2021). To explore the underlying mechanisms of XBJI-induced NIN-IHR, a series of antagonists for these mediators were used. The results, as shown in Fig. 3A, indicated that XBJI-caused Evans blue leakage was not blocked by ginkgolide B (PAF receptor antagonist), metergoline (5-HT receptor antagonist), and icatibant (BK-B2 receptor antagonist), but could be markedly countered by the antagonists for both histamine H1 and H2 receptors (diphenhydramine and cimetidine) (Fig. 3B−C), showing that histamine was responsible for XBJI-induced NIN-IHR.

Fig. 3.

Effects of different antagonists on XBJI-induced Evans Blue leakage. (A) Representative images of effects of three antagonists on XBJI-induced Evans Blue leakage. (B−C) Effects of histamine H1 and H2 receptor antagonists on XBJI-induced Evans Blue leakage. Mean ± SD, n = 8; ^**P < 0.01 vs XBJI alone.

3.4. XBJI triggers NIN-IHR by activating complement C3

Considering that histamine is released by mast cells (Riley & West, 1953), we firstly determined whether the XBJI could directly induce mast cell degranulation. However, it failed to evoke the degranulation of LAD2 cells (Fig. 4A), suggesting that XBJI-caused histamine release might result from an indirect mechanism. It is known that anaphylatoxin C3a and C5a can activate mast cells to release histamine (Johnson et al., 1975, Schäfer et al., 2013). Thus, we determined the effects of C3a or C5a antagonist on XBJI-induced vascular leakage. As a result, SB290157 (a C3a receptor antagonist), but not PMX53 (a C5a receptor antagonist), significantly attenuated Evans Blue leakage induced by XBJI (Fig. 4B−C). Moreover, it could directly activate C3 in vitro in a concentration-dependent manner (Fig. 4D). These results highly suggested that XBJI-caused NIN-IHR was attributed to the direct activation of complement C3.

Fig. 4.

XBJI-induced NIN-IHR is mediated by C3a. (A) Effect of XBJI on mast cells degranulation. Mean ± SD, n = 3; ^##P < 0.01 vs vehicle. (B−C) Effects of PMX53 and SB290157 on XBJI-induced Evans Blue leakage. Mean ± SD, n = 8; ^**P < 0.01 vs XBJI alone. (D) Effect of XBJI directly activated C3 in normal human serum in vitro. Mean ± SD, n = 3; *P < 0.05 and ^**P < 0.01 vs vehicle.

3.5. Carthami Flos is responsible for XBJI-induced NIN-IHR

To find out the common culprit(s) for XBJI-caused NIN-IHR, we prepared XBJI (self-prepared XBJI) and the extract of each herb (F1−F5) according to the preparation procedure of commercial XBJI (National Drug Standard^#YBZ01242004-2010Z-2012). Similarly to commercial XBJI (Fig. 2D), self-prepared XBJI also induced a significant Evans Blue leakage (Fig. 5A); of which F2−F5 did not have this capacity at their equivalent concentrations in XBJI (data not shown), whereas F1 showed potent capacity to induce NIN-IHR (Fig. 5B) that could be blocked by SB290157 (Fig. 5C). Moreover, F1 could also directly activate C3 in vitro (Fig. 5D), strongly suggesting that Carthami Flos is the common crime of XBJI-induced NIN-IHR. To further confirm these finding, we deliberately prepared F1-free XBJI. Unsurprisingly, it failed to induce NIN-IHR (Fig. 5E). These results demonstrate that Carthami Flos is responsible for XBJI-induced NIN-IHR.

Fig. 5.

Carthami Flos is common crime of XBJI-induced NIN-IHR. (A−B) Effects of self-prepared XBJI (A) and F1 (B) on Evans Blue leakage model. Mean ± SD, n = 8; ^**P < 0.01 vs vehicle. (C) Effect of SB290157 on F1-induced Evans Blue leakage model. Mean ± SD, n = 8; ^**P < 0.01 vs F1 alone. (D) Effect of F1 directly activated C3 in normal human serum in vitro. Mean ± SD, n = 3; *P < 0.05 and ^**P < 0.01 vs vehicle. (E) Effect of F1-free XBJI on Evans Blue leakage model. Mean ± SD, n = 8; ^nsP > 0.05 vs vehicle.

3.6. SY but not SR can induce NIN-IHR

3.6.1. NIN-IHR disappears when F1 is subjected to decoloration

Carthami Flos, which triggers NIN-IHR of XBJI via the parenteral route, also serves as a main material of Danhong Injection (DHI), another injection evaluated by us several years ago. Confusingly, DHI was incapable of inducing NIN-IHR (data not shown). By comparing their preparation procedures, we found that DHI had an additional step to remove pigments by using activated carbon which is frequently used for preparing traditional Chinese medicine injections (TCMIs). These raised our doubt about safflower pigments (SPs). To clarify this speculation, XBJI and F1 were discolored by using activated carbon. Excitingly, both discolored-derivatives indeed did not induce Evans Blue leakage anymore (Fig. 6A−B). To eliminate the possibility that Carthami Flos-induced IHR results from the compatibility effect with the other four herbs, we next prepared SPs-free XBJI which was made up of F2−F5 and discolored-F1. The result indicated that SPs-free XBJI also could not trigger NIN-IHR (Fig. 6C).

Fig. 6.

NIN-IHR disappears when F1 is subjected to decoloration (Mean ± SD, n = 8). (A−C) Effects of discolored-XBJI (A), discolored-F1 (B) and SPs-free XBJI (C) on Evans Blue leakage model. ^nsP > 0.05 vs vehicle. (D−E) Effect of XBJI and discolored-XBJI and self-prepared XBJI and SPs-free XBJI on rectal temperature of propranolol-pretreated mice. ^##P < 0.01 vs vehicle; ^**P < 0.01 vs XBJI; ^&&P < 0.01 vs self-prepared XBJI.

Given that XBJI is an intravenous preparation, we also determined its capacity to cause systemic NIN-IHR. Consistently, intraperitoneal injection of both commercial and self-prepared XBJI (45 mL/kg, three times of the clinical dose) led to obvious hypothermia, while this reaction was significantly weakened after activated carbon treatment (Fig. 6D−E). These findings demonstrate that SPs are responsible for XBJI-caused NIN-IHR.

3.6.2. SY but not SR triggers NIN-IHR

Based on the chemical compositions and the colours, SPs are divided into SY and SR (Watanabe, Hasegawa, Yamamoto, Nagai, & Terabe, 1997). To elucidate their contribution to SPs-induced NIN-IHR, SY and SR were separated from Carthami Flos, respectively. As a result, only SY could cause significant Evans Blue leakage (Fig. 7A−B), which could be countered by SB290157 (Fig. 7C). It also could directly activate C3 in vitro at the equivalent concentration of 2%−4% XBJI (Fig. 7D). These results demonstrate that SY but not SR is responsible for SPs-induced NIN-IHR.

Fig. 7.

SY but not SR causes NIN-IHR. (A−B) Effects of SY (A) and SR (B) on Evans Blue leakage model. Mean ± SD, n = 8; ^**P < 0.01 and ^nsP > 0.05 vs vehicle. (C) Effect SB290157 on SY-induced Evans Blue leakage. Mean ± SD, n = 8; ^**P < 0.01 vs SY alone. (D) Effect of SY directly activated C3 in normal human serum in vitro. Mean ± SD, n = 3; ^**P < 0.01 vs vehicle.

3.6.3. Five constituents of SY can cause Evans Blue leakage

Next, we evaluated the capacities of ten available constituents from SY (Table 3) to cause Evans Blue leakage in rats. As shown in Fig. 8A, five of them [isoquercitrin, anhydrosafflor yellow B (AHSYB), 6-hydroxykaempferol-3,6,7-triglucoside, 6-hydroxykaempferol 3,6-diglucoside, 6-hydroxykaempferol 3-beta-rutinoside] notably increased vasopermeability after the first intracutaneous injection at the concentration of 1 mg/mL. We also determined whether these constituents could directly activate C3 in vitro. As a result, AHSYB could directly activate C3 in vitro with the half-effective concentration (EC₅₀) value of 0.77 mg/mL (Fig. 8B). Regrettably, the EC₅₀ values of the other four constituents could not be obtained because they were incompatible with the assay system.

Table 3.

Ten available constituents from SY.

No.	Constituents	CAS No.
1	Isoquercitrin	482-35-9
2	Nicotiflorin	17650-84-9
3	Rutin	153-18-4
4	Anhydrosafflor yellow B (AHSYB)	184840-84-4
5	6-Hydroxykaempferol-3,6,7-triglucoside	145134-62-9
6	6-Hydroxykaempferol 3-rutinoside-6-glucoside	145134-63-0
7	6-Hydroxykaempferol 3-o-beta-d-glucoside	145134-61-8
8	6-Hydroxykaempferol 3,6-diglucoside	142674-16-6
9	6-Hydroxykaempferol 3-beta-rutinoside	205527-00-0
10	Hydroxysafflor yellow A	78281-02-4

Fig. 8.

Effects of ten constituents of SY on Evans Blue leakage. (A) Representative images of Evans Blue leakage of rat dorsal skin induced by ten constituents (1 mg/mL) in SY. (B) Effect of AHSYB directly activated C3 in normal human serum in vitro. Mean ± SD, n = 3; ^**P < 0.01 vs vehicle.

3.7. Impact of decolorization on anti-inflammatory action of XBJI in vitro

Considering that XBJI is primarily indicated for sepsis, we compared the anti-inflammatory effect of XBJI before and after decolorization by using a classical sepsis model in vitro. As shown in Fig. 9, in contrast to non-discolored XBJIs (commercial XBJI and self-prepared XBJI), both discolored-XBJI and SPs-free XBJI still could lower the production of NO, IL-6, and TNF-α in LPS-stimulated RAW264.7 cells at their non-cytotoxic concentrations (≤ 2.5%). Of course, we also noticed that decolorization indeed impaired the anti-inflammatory action of XBJI to a certain extent. For instance, at the concentration of 0.63%, two discolored XBJIs could not reduce NO production anymore (P > 0.05), while the two non-discolored XBJIs could do (P < 0.01).

Fig. 9.

Effects of decolorization on anti-inflammatory action of XBJI in vitro (n = 3). (A−B) Effects of XBJI and discolored-XBJI (A), self-prepared XBJI and SPs-free XBJI (B) on viability of RAW264.7 cells. (C) Effects of XBJI and discolored-XBJI on supernatant NO, IL-6, and TNF-α production in LPS-activated RAW264.7 cells. ^##P < 0.01 vs normal control; *P < 0.05 and ^**P < 0.01 vs LPS alone; ^ΔP < 0.05 vs 1.25% XBJI group; ^&P < 0.05 and ^&&P < 0.01 vs 2.50% XBJI group. (D) Effects of self-prepared XBJI and SPs-free XBJI on supernatant NO, IL-6, and TNF-α production in LPS-activated RAW264.7 cells. ^##P < 0.01 vs normal control; *P < 0.05 and ^**P < 0.01 vs LPS alone; ^$$P < 0.01 vs 0.63% self-prepared XBJI group; ^ΔΔP < 0.01 vs 1.25% self-prepared XBJI group; ^&P < 0.05 and ^&&P < 0.01 vs 2.50% self-prepared XBJI group.

4. Discussion

TCMIs are the combination of modern pharmaceutical technology and traditional Chinese prescription, which were born in 1941 and played a great role in the backward medical conditions at that time (Zheng, Wu, Gao, & Ouyang, 2022). Until now, there is still a considerable amount of clinical application in China. After the outbreak of COVID-19 epidemic, XBJI, whose main indication is sepsis, was listed as a recommended drug in the “Chinese Diagnosis and Treatment Plan for SARS-CoV-2 Infection (Trial 10th edition)” (National Health Commission & State Administration of Traditional Chinese Medicine, 2023). With the increased application, its ADRs have attracted considerable attention. According to its manifestations, the majority of them should be IHR. In this study, we systemically investigated XBJI-induced IHR using in vitro and in vivo models. After continuous five weeks intraperitoneal immunization with XBJI, serum tIgE and MCPT-1 were not increased (Fig. 2A−B), indicating that XBJI did not cause IgE-mediated IHR. Nevertheless, it could potently induce NIN-IHR, showing the significant Evans Blue leakage of mouse paw (increased vasopermeability) after the first intraplantar injection (Fig. 2D−E), which was consistent with previous finding of other researchers who also observed XBJI-induced NIN-IHR through intravenously injecting XBJI along with Evans Blue (Yi et al., 2021).

Given that NIN-IHR can be triggered by many mediators, including histamine, PAF, or BK, etc. (Muñoz-Cano, Picado, Valero, & Bartra, 2016), we deliberately chose their antagonists to investigate the molecular mechanism of XBJI-induced IHR. As a result, the histamine H1/H2 receptor antagonists potently inhibited XBJI-caused Evans Blue leakage (Fig. 3), showing that histamine was the primary effective molecular. Finally, we confirmed that XBJI was through directly activating complement C3 to induce histamine release of mast cells (Fig. 4), thus leading to NIN-IHR.

To find out the potential “common culprit(s)” for XBJI-caused IHR, we separately prepared the extract from each herb (F1−F5) according to the preparation procedure of XBJI and evaluated their ability to induce IHR. At the equivalent concentration in XBJI, only the extract of Carthami Flos (F1) was capable of causing Evans Blue leakage (Fig. 5B) which could also be blocked by C3a receptor antagonist (Fig. 5C). Intriguingly, DHI, another TCMI containing Carthami Flos, did not show this effect based on our previous finding (data not shown). The diverse actions drove us to compare their preparation procedures. We found that there was an extra step of activated carbon absorption for preparing DHI. Indeed, F1 treated by activated carbon (discolored-F1) could not induce NIN-IHR anymore. Moreover, whether the commercial XBJI treated by activated carbon (discolored-XBJI) or the self-prepared SPs-free XBJI also failed to cause IHR (Fig. 6). Given that activated carbon can mainly adsorb pigments (Sun, Jiang, & Xu, 2009), we next focused on SPs which are mainly divided into SY and SR. Our results demonstrated that SY, rather than SR, potently caused IHR at its concentration in XBJI (Fig. 7A−B). After screening ten available constituents from SY, five of them markedly increased vasopermeability (Fig. 8A). Nevertheless, the final concentration of these constituents (1 mg/mL) is much higher than their actual concentrations in XBJI, indicating that there might be a series of active constituents similar to these five constituents that collectively activate C3 to induce IHR. Our finding clearly demonstrates that SY in Carthami Flos should be the “common culprit” for XBJI-caused IHR.

SY is a class of water-soluble flavonoid and the major active ingredients in Carthami Flos, a well-known “blood circulation and silt” herb. Pharmacological studies showed that it possessed various activities, particularly improving cardiovascular and cerebrovascular diseases, such as anti-atherosclerosis, anti-thrombotic and cardioprotective effect (Chen et al., 2022, Wu et al., 2013). The main indication of XBJI is sepsis, an inflammation (cytokine storm)-related disease. Does the activated carbon treatment have an impact on the efficacy of XBJI? To address this issue, a classical in vitro model of sepsis was used (Lin & Yeh, 2005). The result showed that discolored-XBJI still retained the anti-inflammatory activity (Fig. 9), demonstrating that the non-adsorbed constituents in XBJI still exert anti-inflammatory activity.

5. Conclusion

In summary, the present study systemically evaluated the potential of XBJI to cause IHR for the first time. We found that XBJI failed to induce IN-IHR, but potently triggered NIN-IHR through direct activating complement C3 to provoke histamine release. After screening five herbs of XBJI, we identified that SY in Carthami Flos was responsible for XBJI-caused IHR. And given that the impact of decolorization on XBJI’s anti-inflammatory activity is acceptable, it is necessary and feasible to improve the production process of XBJI in the future.

6. Limitation

Although five constituents of SY can increase the vascular permeability at the concentration of 1 mg/mL which is much higher than their actual concentrations in XBJI, XBJI-induced IHR still cannot be attributed to them. It remains unclear how the intercomponent interaction triggers IHR. Moreover, although activated carbon mainly adsorbs pigments, its adsorption is non-specific. Therefore, the impact of activated carbon treatment on the compositions of XBJI needs to be further elucidated.

CRediT authorship contribution statement

Wenjing Li: Methodology, Investigation, Visualization, Writing – original draft, Writing – review & editing. Yuan Gao: Methodology, Investigation, Visualization, Writing – original draft, Writing – review & editing, Funding acquisition, Supervision. Jingjing Yan: Methodology, Investigation. Min Cai: Writing – review & editing. Chenchen Zang: Writing – review & editing. Zhuangzhuang Liu: Methodology. Ximeng Li: Methodology. Runlan Cai: Validation. Yun Qi: Conceptualization, Writing – review & editing, Funding acquisition, Supervision.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.