Attenuation of antigen-specific T helper 1 immunity by Neolitsea hiiranensis and its derived terpenoids

Extraction and isolation from the Taiwanese N. hiiranensis

The crude extracts and the secondary metabolites were prepared and isolated from the leaves of N. hiiranensis according to the previous report (Liou et al., 2011). Briefly, Taiwanese N. hiiranensis were collected at Mudan (Pingtung County, Taiwan) and identified by Dr. Ih-Sheng Chen, one of the authors. The dried leaves were extracted with three times cold MeOH, and then the different partition of crude extracts was prepared with the differential proportion solvents system, including EtOAc:H₂O, n-hexane: EtOAc, acetone: H₂O, and n-hexane: acetone for further isolating the secondary metabolites. Seven sesquiterpenoids, (e.g., (-)-ent-6α-methoxyeudesm-4(15)-en-1β-ol, hiiranlactones A–D, (+)-villosine, hiiranepoxide), one triterpenoid (hiiranterpenone), and 22 known compounds were identified and elucidated by spectroscopic analysis and single crystal X-ray diffraction (Liou et al., 2011). An established QSAR model for drug-induced liver injury (DILI) was utilized for prediction of non/less toxic pure compounds for further functionality tests (Huang et al., 2015).

Reagents and chemicals

All reagents were purchased from Sigma (St Louis, MO) unless otherwise stated. Fetal bovine serum (FBS) and cell culture medium RPMI 1640 were used from Hyclone (Logan, UT). Enzyme-linked immunosorbent assay (ELISA) sets for cytokine and antibody measurement were purchased from BD Biosciences (San Diego, CA). Isol-RNA lysis reagent were purchased from 5-Prime (Gaithersburg, MD). RevertAid RT kit was purchased from Thermo for Reverse transcription-polymerase chain reactions (RT-PCR).

Animals

Male BALB/c mice (5 weeks old) were purchased from BioLasco (Ilan, Taiwan). On arrival, mice were randomly transferred to plastic cages containing aspen bedding (five mice per cage) and acclimatized for at least one week before initiating experiments. Mice were housed in a temperature (22 ± 2 °C), humidity (50 ± 20%) and light (12-hour light/dark cycle)-controlled environment. Food and water were supplied ad libitum.

Animal model for antigen-specific T-cell function

The experimental protocol was approved by the Kaohsiung Medical University Institutional Animal Care and Use Committee (IACUC number 101132). Mice were administered daily by intraperitoneal injection of crude extracts (5 and 20 mg/kg) for three doses before antigen sensitization. The protocol was shown in Fig. 1. Mice were randomly divided into the following groups: naïve control (NA), vehicle (4% DMSO)-treated group (VH), and N. hiiranensis-treated groups (5 and 20 mg/kg in 4% DMSO). Vehicle and/or N. hiiranensis were administered to mice daily by intraperitoneal injection for three consecutive days (day 1–3). Except for the NA group, mice were sensitized with OVA 12 h after the third dose of VH or N. hiiranensis on day 3 by an intraperitoneal injection with 0.1 mL per mouse of sensitization solution containing 100 µg OVA and 1 mg alum (as adjuvant) in saline. The mice and then challenged with OVA/alum at day 9. After OVA challenge, the mice were sacrificed at day 10 and their spleens were prepared and made into single-cell suspensions. The splenocytes were re-stimulated with OVA (100 µg/mL) in culture for 72 h to induce cell proliferation and cytokine production.

Cell proliferation assay

Splenocytes from the mice were aseptically cultured in RPMI 1640 medium supplemented with 5% heat-inactivated FBS, 100 µg/mL streptomycin, and 100 U/mL penicillin at 37 °C in 5% CO₂. Splenocytes (7 × 10⁶ cells/mL) were seeded into 96-well plates. The cells were either left unstimulated or stimulated with OVA for 72 h. The viability of splenocytes was determined by the 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl-tetrazolium bromide (methylthiazol tetrazolium) assay. A methylthiazol tetrazolium stock solution (5 mg/mL in phosphate buffered saline) was then added to each well (10 µL/well) and incubated for 4 h. The formed formazan was dissolved with a lysis buffer (10% SDS in N, N-demethylformamide) overnight in the dark. The optical density was measured at 570 nm (and at 630 nm as a background reference) using a microplate reader (Dynatech Laboratories Inc, Chantilly, VA, USA).

Enzyme-linked immunosorbent assay (ELISA) for serum antibodies

ELISA plates were coated with 0.05% OVA in coating buffer (0.1 M NaHCO₃) and blocked with 1% bovine serum albumin in phosphate-buffered saline containing 0.05% Tween 20 (PBST). After washing with PBST, the serum samples were added into wells (50 µL/well) and incubated for 1 h. After another washing, horseradish peroxidase-conjugated anti-mouse IgG₁, IgG_2a or IgM was added (50 µL/well) and incubated for 1 h. Finally, wells were washed and a tetramethylbenzidine solution (50 µL/well) was added for colorimetric detection of bound peroxidase conjugate. The reaction was terminated by adding 150 µL of 3 N H₂SO₄ per well. The optical density (OD) was measured at 450 nm using a microplate reader (Dynatech Laboratories, Chantilly, VA, USA). Total IgE was measured according to manufacture’s instruction (BD Pharmingen).

Cytokine measurement by ELISA

To examine the effects of N. hiiranensis on specific subsets of T cells, splenocytes (7 × 10⁶ cells/mL) were cultured in 48-well plates (300 µL/well) followed by OVA re-stimulated (100 µg/mL) for 72 h. The supernatants were harvested and quantified for IL-2, IL-4, IL-12p70, IL-10, IL-13 and IFN-γ by ELISA kits according to manufacture’s instruction (BD Pharmingen).

In silico prediction of hepatotoxicity and genotoxicity

Quantitative structure-activity relationship (QSAR) models are useful tools for in silico estimating toxicity properties of chemicals according to toxicity-related descriptors of physicochemical properties and fingerprints (Perkins et al., 2003). QSAR models have been extensively applied to prioritize chemicals for potential toxicity (Cronin et al., 2003; Gramatica, Cassani & Sangion, 2016). In this study, toxicity properties of tested compounds were predicted by both the admetSAR server (Cheng et al., 2012) and our hepatotoxicity prediction model (Huang et al., 2015). The genotoxicity, carcinogenicity and acute oral toxicity of tested compounds are predicted by admetSAR with probabilities representing the confidence of prediction. The hepatotoxicity model is a special QSAR model utilizing toxicity information in human that no cross-species extrapolation is required. Similar to admetSAR, a confidence score is given by the hepatotoxicity prediction model. Generally, a score/probability close to 1 indicates a higher probability that a toxicity is associated with a given chemical. In contrast, a score closed to 0 indicates that a toxicity is unlikely associated with a given chemical.

RNA isolation and real-time reverse transcription-polymerase chain reactions (RT-PCR)

Total RNA from whole splenocytes (stimulated with OVA for 48 h) was isolated using an isol-RNA lysis reagent (5-Prime). The RNA samples (5 µg) were then treated with an RQ1 RNase-free DNase kit (Promega, Southampton, UK) according to the manufacturer’s instructions to remove contaminated DNA and the quality of total RNA was confirmed by agarose gel electrophoresis. The RNA concentration of each sample were quantified using determination of optical density at 260 nm (OD260) by a microplate reader (Thermo varioskan flash; Thermo Fisher Scientific, Waltham, MA, USA). One µg of total RNA of each sample was reverse-transcribed by RevertAid RT Kit (Thermo) into cDNA products using oligo (dT) as primer. Real-time PCR was performed in a 96-well optic tray by an ABI PRISM^® 7900HT Sequence Detection System (Applied Biosystems, UK). During real-time RT-PCR process, we used Luminaris Color HiGreen High ROX qPCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA) which provides a highly specific and sensitive method to quantify mRNA expression. The HPRT gene was used as an endogenous control to normalize the expression of target genes. The primers are: 5^′-GCCAGGGAACCGCTTATATG-3^′ and 5^′-GACGATCATCTGGGTCACATTCT-3^′ for T-bet, 5^′-TACCCTCCGGCTT- CATCCT-3^′ and 5^′-TGCACCTGATACTTGAGGCAC-3^′ for GATA-3, 5^′-GCC- AAGTTTGAGGTCAACAAC-3^′ and 5^′-CCGAATCAGCAGCGACTC-3^′ for IFN-γ, 5^′-CCATATCCACGGATGCGACA-3^′ and 5^′-AAGCCCGAAAGAGTCTCTGC-3^′ for IL-4, 5^′-TCAGTCAACGGGGGACATAAA-3^′ and 5^′-GGGGCTGTA- CTGCTTAACCAG-3^′ for HPRT (Shi et al., 2010), and 5^′-CCTCAATGGTATAGCAGAAC and 5^′ -TAGCCTTGG AATCCTTGG for IL-12Rβ2 (Chognard et al., 2014).

In vitro screening of cytokine production by antigen-specific T cells

OVA-primed splenocytes were generated according to previous protocol. Briefly, 6–8 week mice were sensitized by intraperitoneal injection of OVA (10 mg OVA absorbed to 100 mg alum as adjuvant) twice on day 1 and 14. On day 15, the mice were sacrificed and their spleens were harvested and made into single-cell suspensions. The OVA-primed splenocytes (7 × 10⁶ cells/mL) were either left untreated (control), 0.05% DMSO (VH) and/or secondary metabolites (1–10 µM) followed by re-stimulation with OVA (100 µg/mL) for 72 h. The cell proliferation activity, cytokine productions, and mRNA expression of target genes were measured as described above.

Flow cytometry analysis of intracellular cytokine and transcription factor staining in CD4⁺ cells

OVA-primed splenocytes were cultured in a 12-well plate and stimulated with OVA and β-caryophyllene oxide for 36 h. For analysis of intracellular cytokine production, the cells then treated with GolgiStop (0.6 µL/mL; BD Biosciences) for 10 h prior to being harvested for antibody staining. The cells were next stained with FITC-conjugated anti-mouse CD4 mAb (clone GK1.5; Biolegend, CA, USA) for 30 min on ice. The splenocytes then were fixed and permeabilized using Fixation and Perm/Wash buffers (BD Biosciences) before staining for intracellular IFN-γ by PE-conjugated anti-mouse IFN-γ mAb (clone XMG1.2; Biolegend) for 30 min on ice. Ten thousand CD4⁺ cells were acquired on a BD LSR II flow cytometer (BD Biosciences). The mean fluorescence intensity (MFI) of IFN-γ in total CD4⁺ cells was quantified by gating CD4⁺ cells and then analyzed using FlowJo software (Treestar, Inc., CA). For detection of T-bet and GATA-3, splenocytes treated with β-caryophyllene oxide for 36 h and then harvested for anti-CD4 mAb staining. Next the cells were fixed and permeabilized using True-Nuclear Transcription Factor Buffer Set (Biolegend) according to the manufacturer’s instructions. PerCP-Cy5.5-conjugated anti-mouse T-bet mAb (clone 4B10; Biolegend) and PerCP-Cy5.5 conjugated anti-mouse GATA-3 mAb (clone 16E10A23; Biolegend) were applied to detect protein level of T-bet and GATA-3 in CD4⁺ cells. Ten thousand CD4⁺ cells were acquired on a BD LSR II flow cytometer (BD Biosciences). The mean fluorescence intensity (MFI) of T-bet or GATA-3 in total CD4⁺ cells was quantified by gating CD4⁺ cells and then analyzed using FlowJo software (Treestar, Inc., Ashland, OR, USA).

Statistical analysis

Homogeneous data were evaluated by a parametric analysis of variance (ANOVA) with Dunnett’s test to assess the statistical differences between the treatment groups and the VH control group by software Prism 5.0 (GraphPad Software Inc., San Diego, CA, USA). P < 0.05 was defined as statistical significance. The mean ± standard error was presented for individual experiments.

Crude extracts of leaves of N. hiiranensis did not affect body weight, spleen index, and cellularity in vivo

To investigate the potential of N. hiiranensis for in vivo use, we first investigated the direct immunotoxicity of N. hiiranensis in vivo. As shown in Table 1, intraperitoneal injection of N. hiiranensis extracts (5 and 20 mg/kg) didn’t affect the body weight, the spleen index, and the population of CD4⁺, CD8⁺, B220⁺, and CD11b⁺ in spleens of the mice received treatment. We then examined the effect of the N. hiiranensis extracts on T-cell mediated humoral and cell-mediated immune responses in vivo. The N. hiiranensis extracts decreased the serum level of OVA-specific IgM and IgG_2a. The serum level of OVA-specific IgG_2a was dose-dependently suppressed by N. hiiranensis extracts. At the dose of 20 mg/kg, the level of OVA-specific IgM was significantly decreased. In contrast, N. hiiranensis didn’t affect OVA-specific IgG₁ and total-IgE production. The result indicated that repeating administration of N. hiiranensis significantly suppressed Th1-associated antibody production (Fig. 2)

Crude extracts of leaves of N. hiiranensis attenuated antigen-specific IL-2, IL-12, and IFN-γ cytokine production in vivo

We then proceed to examine the effects of the crude extracts on the functionality of antigen-specific T cells by measuring the production of IL-2 (T-cell growth factor for clonal expansion), IFN-γ and IL-4 (the signature Th1 and Th2 differential cytokines), IL-12 (induction of IFN-γ production and Th1 differentiation), and IL-13 (Th2 cytokine closely related IL-4 to induce allergic Th2 responses). To check whether or not the induction of OVA was successful, the splenocytes were isolated and divided into non-stimulated (Non-OVA) and OVA 100 µg/mL re-stimulated (OVA) groups and culture for 72 h to induce antigenspecific cytokine production. The result showed induced cell proliferation in OVA-restimulated groups compared to their corresponding non-stimulated groups (Fig. 3A). In addition, the leaves extractions of N. hiiranensis did not affect the proliferation of OVA-specific splenocytes. Despite not affecting cell proliferation, the extracts significantly decrease the production of antigen-specific IL-2 and IL-12 in the N. hiiranensis 20 mg/kg group (Figs. 3B and 3D). The production of IFN-γ by the cells was also attenuated by the treatment. IFN-γ was significantly lowered at both 5 mg/kg and 20 mg/kg treatment groups (P < 0.05 for each group compared to the vehicle control group; Fig. 3C). By contrast, IL-4 and IL-13 were not affected (Figs. 3E and 3F). The results indicated that the extracts of the leaves of N. hiiranensis have differential effects on Th1 responses.

In silico selection of potential secondary metabolites of N. hiiranensis without undesired toxicity

Quantitative structure–activity relationship (QSAR) models of admetSAR (Cheng et al., 2012) and our hepatotoxicity prediction model (Huang et al., 2015) were applied to predict genotoxicity, carcinogenicity, and hepatotoxicity of 14 secondary metabolites from N. hiiranensis (Fig. 4). We firstly converted our tested compounds into SMILES (Simplified Molecular Input Line Entry System) representations, a line notation for representing the structure of molecules and reactions. The SMILES is subsequently submitted to admetSAR and our hepatotoxicity model to predict their genotoxic, carcinogenic, and hepatotoxic potentials. Table 2 showed the predicted toxicity of 14 compounds. The probability of genotoxicity (Ames Test), carcinogenicity, and hepatotoxicity among the 14 selected compounds were 0.05–0.49, 0.07–0.24, and 0.45–0.60, respectively, indicating that these selected compounds were classified as AMES-negative, non-carcinogenic and less hepatotoxic compounds. Moreover, the estimated acute oral toxicity of these compounds was predicted as class III (US EPA category system; LD₅₀ value is between 500–5,000 mg/kg) with the probability of 0.45–0.83.

Secondary metabolites of N. hiiranensis attenuated IFN-γ production in vitro

OVA-prime splenocytes were generated for screening the immunomodulatory effects of the compounds from N. hiiranensis. The OVA-primed splenocytes were re-stimulated with OVA (100 µg/mL) in the presence of vehicle and/or 10 µM of selected compounds for 72 h in vitro. The selected pure compounds didn’t affect the proliferation activity nor induce the direct cytotoxicity (Fig. 5A). IL-2 and IL-4 were not significantly affected by the 14 selected compounds compared to VH. Interestingly, the antigen specific IFN-γ cytokine production were significantly suppressed by β-caryophyllene oxide (2), hiiranlactone D (8), and trans-phytol (13) with the inhibition rate of 44%, 32%, and 35%, respectively, comparing to VH (referred as 100%) (P < 0.05). Spathulenol (12) slightly inhibited IFN-γ and IL-4 productions without statistical significance (Fig. 5).

We further investigated the concentration-dependent effects of these compounds (2, 8, 12, and 13) on cytokine production. The effects of β-caryophyllene oxide (2), hiiranlactone D (8), spathulenol (12), and trans-phytol (13) on the cell viability and Th1/Th2 cytokine secretions are shown in Figs. 6–9. β-caryophyllene oxide (1–50 µM) didn’t alter the proliferation activity as well as IL-2 cytokine production of OVA-specific cells. By contrast, β-caryophyllene oxide inhibited IFN-γ production in a dependent manner and inhibited IL-4 production with an approximately 40% of inhibition rate at the concentrations higher than 25 µM (Fig. 6). Hiiranlactone D (8) didn’t affect cell viability at the concentration of 50 µM. Hiiranlactone D attenuated IL-2 production of OVA-specific cells at 50 µM and inhibited IFN-γ production in a concentration dependent manner starting from concentrations above 10 µM (P < 0.05). No effect on IL-4 was observed for hiiranlactone D (Fig. 7). Spathulenol also didn’t affect cell viability at the concentration of 50 µM. Spathulenol inhibited IL-4 production at 50 µM, while neither IL-2 nor IFN-γ were significantly affected (Fig. 8). trans-Phytol didn’t affect cell viability at the concentration of 50 µM. Interestingly, trans-phytol enhanced antigen-specific IL-2, and IL-4 production in an concentration-dependent manner with significant inhibitions started from 25 and 50 µM for IL-2 and IL-4, respectively. trans-Phytol significantly inhibited IFN-γ production in a concentration-dependent manner at the concentrations between 10 and 50 µM (Fig. 9). The results demonstrated a differential immunomodulatory effects of trans-phytol on the Th1/Th2 cytokine expression in antigen-specific T cells.

In the present data, β-caryophyllene oxide and trans-phytol are the most effective secondary metabolites from N. hiiranensis to suppress IFN-γ production. We next determined the effects of β-caryophyllene oxide and trans-phytol on the production of other Th1 and Th2 cytokines, including IL-12, IL-13, and IL-10. In Fig. 10, trans-phytol significantly decreased IL-12 production, while both IL-13 and IL-10 were not altered (Figs. 10E and 10F). Interestingly, although β-caryophyllene oxide significantly suppressed IFN-γ production (Fig. 6C), the IL-12 production was not significantly altered by β-caryophyllene oxide (Fig. 10A). In order to further confirm β-caryophyllene oxide directly suppressed IFN-γ production by CD4⁺ cells, the intracellular cytokine staining approach was applied. In the supplemental data (Fig. S1), β-caryophyllene oxide significantly suppressed the cellular level of IFN-γ in the CD4⁺ cells.

β-caryophyllene oxide differentially modulated the development of Th1 and Th2 cells at transcription level

T-bet is a Th1-specific T-box transcription factor which controls the expression of Th1 cytokines and directs Th1 lineage commitment, while GATA-binding protein 3 (GATA-3) is a Th2-specific transcription factor which augments Th2-specific cytokines and Th2 differentiation to suppress Th1 immune responses. These transcription factors play crucial roles to regulate the homeostasis of Th cells (Agnello et al., 2003; Koch et al., 2009; Laurence et al., 2007). To understand the effect of N. hiiranensis at transcription level of T-cell function, RT-PCR was performed. As β-caryophyllene oxide was the most effective compound selected from N. hiiranensis which dramatically inhibited IFN-γ production. We next determined how β-caryophyllene oxide regulates Th1/Th2 associated gene expression. IFN-γ, IL-4, and Th1/Th2 differential transcription factors, T-bet and GATA-3 were determined. The mRNA expression of T-bet was significantly down-regulated by approximately 2-3-fold at 1 and 10 µM (Fig. 11A), whereas GATA-3 wasn’t significantly altered after treatment of β-caryophyllene oxide (Fig. 11D). IFN-γ was also down-regulated by approximately 1.5-2-fold (Fig. 11B); however, IL-4 was not changed (Fig. 11E). As T-bet response to IFN-γ may lead to the up-regulation of IL-12Rβ2 expression on Th1 cell surface for Th1 cell responsiveness to IL-12 stimulation (Hamza, Barnett & Li, 2010), we next determined whether β-caryophyllene oxide attenuated the expression of IL-12Rβ2. In consist with IFN-γ expression, the IL-12Rβ2 expression was down-regulated by approximately 1.5-2-fold (Fig. 11C). These results showed that the differentiation and functionality of Th1 cells were more sensitive to be attenuated by β-caryophyllene oxide.

To confirm the protein level of T-bet and GATA-3 in β-caryophyllene oxide-treated cells, intracellular staining of transcription factors in CD4⁺ cells were analyzed by flow cytometry. The proportion of T-bet⁺CD4⁺ and GATA-3⁺CD4⁺in total CD4⁺cells was quantified. Figure 12 showed that β-caryophyllene oxide decreased the total percentage of T-bet⁺CD4⁺ cells in CD4⁺ cells from 13% (VH) to 8% (β-caryophyllene oxide, 50 µM) and the level of mean fluorescence intensity of T-bet was significantly decreased (Fig. 12B) By contrast, the proportion of GATA-3⁺CD4⁺ and protein level of GATA-3 in CD4⁺ cells were not changed by β-caryophyllene oxide (Figs. 12C–12D).