EBF-2 was generated by purification of FAHF-2 with a safe solvents consisting of butanol and ethyl acetate (52). 8 herb constituents (Prunus mume, Zanthoxylum schinifolium, Angelica sinensis, Zingiber officinalis, Cinnamomum cassia, Phellodendron chinense, Coptis chinensis, and Panax ginseng) were extracted using butanol, while the Ganoderma lucidum was extracted using ethyl acetate and the dried extracts were combined to generate the EBF-2 powder substance. Three batches of FAHF-2 and EBF-2 were tested in this study-manufacturing date and shelf life are listed in Supplemental Table 1 . Botanical information for individual herbs, including geographical location, harvest season, pre-processing, heavy metal and pesticide residues, and quality control methods, have been published previously (54).

The IgE-producing human plasma cell line U266 (ATCC, MD) was grown in complete media containing RPMI 1640 medium supplemented with 10% FBS, 1 mM sodium pyruvate, 1×10^-5 M β-ME and 0.5% penicillin-streptomycin (55, 56) at 2×10⁴ cells/mL in 48 well plates. Three batches of FAHF-2 and 3 batches of EBF-2 at serial dilution concentrations starting at 500 μg/mL and 120 μg/mL, respectively, were used in culture for 6 days. Berberine (purity>98%, Sigma Aldrich, St Louis, MO) at serial concentrations starting at 5 μg/mL were also tested on U266 cells. Supernatants were harvested, IgE levels were determined by ELISA (Mabtech Inc, OH) and cell viability was evaluated by trypan blue exclusion (55).

Six-week-old female C3H/HeJ mice purchased from the Jackson Laboratory (Bar Harbor, ME) were maintained in pathogen-free facilities at the Mount Sinai vivarium according to standard guidelines (57). Mice were intragastrically (i.g.) sensitized with 10 mg of homogenized peanut (PN) in 0.5 mL PBS containing 75 mg sodium bicarbonate, 10 µg of the mucosal adjuvant cholera toxin (CT) (List Laboratories, Campbell, CA), and 16.5 µL (1.1 µL/g body weight) of 80 proof Stolichnaya Vodka^® (a source of food grade ethanol) to neutralize stomach pH and increase gastrointestinal permeability, three times during week 0 (50). Thereafter, sensitization was done weekly as above except that the CT dose given was 20 µg. The boosting dose of 50 mg PN was given at weeks 6 and 8 using the same gavage solution as in weeks 1 through 5. These mice were defined as peanut allergic (PNA) mice. One day following the last boost, at which hypersensitivity was developed (58), PNA mice received EBF-2 treatment at 3.84 mg in 0.5 mL drinking water, twice a day for 4 weeks. One week after completing the treatment, mice were challenged with ground peanut (200 mg) i.g. and again 4 weeks later. Anaphylactic reactions were accessed ( Figure 1A ). Sham treated PNA mice and naïve mice (unsensitized/untreated) were used as disease and normal controls, respectively. In a separate experiment, to determine the persistent impact of EBF-2 on IgE producing cells by flowcytometry analysis, EBF-2 treated, and sham treated PNA mice received periodic oral exposure of either boiled (10 mg/mouse) or roasted peanut (200 mg/mouse) approximately every 10-15 weeks. Mice were terminated using ketamine/xylazine euthanasia protocol. Briefly, mice were given over-dose (15µL/g body weight) of ketamine-xylazine mixture (100mg/mL and 10 mg/mL respectively) intraperitoneally. After the mice were in deep anesthetics, blood samples were collected, and mice were sacrificed by cervical dislocation after which tissue samples were collected. All animal experiments were approved and performed according to the instruction and guidelines of the Institutional Animal Care and Use Committee (IACUC) of Icahn School of Medicine at Mount Sinai.”

Symptoms were evaluated 30-40 minutes following oral PN challenge as described previously (51, 59), and symptoms were scored utilizing the scoring system ranging from 0 (no reaction) to 5 (fatal reactions), described previously (59). Rectal temperatures were also measured immediately after scoring using a rectal probe (Harvard Apparatus, Holliston, MA).

Blood samples were collected via sub-mandibular bleeding 30 minutes after scoring and measurement of body temperature following oral peanut challenge. Plasma was harvested within 20 minutes after blood collection and stored at -80°C until analyzed. Histamine was measured using a commercial enzyme immunoassay kit (Fisher Scientific, NJ) as described by the manufacturer (58).

PN-specific-IgE, IgG1 and IgG2a in serum was measured as reported previously (50, 51, 59). Briefly, microtiter plates were coated with peanut extract (sample wells), anti-mouse IgE (BD Biosciences, San Jose, CA, for IgE reference wells), or DNP-HSA (Sigma-Aldrich for IgG2a and IgG1 reference wells) and incubated overnight at 4° C. Subsequently, the plates were blocked with 2% BSA-PBS after washing. Washed plates were incubated with diluted serum samples, mouse IgE (BD Biosciences), anti-DNP-IgG2a, or anti-DNP-IgG1 (Accurate Antibodies, Westbury, NY) overnight at 4°C and later developed by using biotinylated anti-IgE, IgG2a or IgG1 detection antibodies (BD Biosciences), avidin-peroxide, and ABTS substrate (KPL, St Paul, MN).

Mice were sacrificed at week 78 of the protocol and single-cell suspensions of splenocytes were prepared in ice cold staining buffer (PBS including 0.5 mM EDTA, 0.05 mM sodium azide, 0.5% BSA). First, surface staining with unlabeled anti-IgE (to block membrane IgE), APC anti-CD138, BV711-anti-CD3, and anti-CD16/32 (Fc-block) (all from BD Biosciences), CA was performed. Live-dead discriminating dye (Live-Dead Aqua, Invitrogen, CA) was included. Cells were washed and incubated with fixation/permeabilization buffer (BD Biosciences, CA) for 15 mins, washed with permeabilization buffer (BD Biosciences, CA), and then incubated with FITC-anti-IgE, in permeabilization buffer. After washing, cells were treated with Cytofix buffer (BD Biosciences, CA) for 15 mins for post-fixation, washed, and then data were acquired on an LSRII flow cytometer (Becton Dickinson, CA). Flow cytometry analysis was performed using Flow Jo (Tree Star, CA) as follows. Live singlet cells were then analyzed for IgE⁺ plasma cells (FITC-IgE +; APC-CD138+ cells).

To evaluate the safety of EBF-2, the sub-chronic toxicity assay was performed on C3H/HeJ mice as in our previous studies (52). Naïve C3H/HeJ mice were fed 40 mg/mL of EBF-2, which is 5 times the daily therapeutic dose, for 14 days. Sham (water) fed mice served as controls (sham). Blood samples were collected at the end of the experiment. Liver and kidney function and complete blood count (CBC) were performed by ALX laboratories, NY.

HPLC analysis was performed on a Waters 2690 HPLC system coupled with a 2996 PDA detector (Waters, Milford, MA) for each batch of FAHF-2 and EBF-2 using the method described previously (52). Each sample was first dissolved in 2 mL of the mixture of mobile phases at a 1:1 ratio and centrifuged at 10,000 rpm for 10 mins. The sample amount injected for HPLC fingerprint analysis was based on the human daily dose used (1:200 of human daily dose). FAHF-2’s human daily dosage is 19.8g, therefore the concentration of FAHF-2 injected was 99mg/mL. For EBF-2, the human daily dose is 4.58g, therefore the concentration of EBF-2 injected to HPLC was 22.9mg/mL. 10 µL of the supernatant was injected into the HPLC system and separated on a ZORBAX SB-C18 (5 µm, 150 mm x 4.6 mm, column (Agilent, Santa Clara, CA). Aqueous formic acid (0.1%) was used as mobile phase A, while acetonitrile (Fisher Scientific, NJ) served as mobile phase B. The separation was performed using a linear gradient elution of 2% to 25% mobile phase B in 45 min, 25% to 35% in the following 25 mins, 35% to 55% in the next 15 mins, and 55% to 75% in the final 10 mins. The flowrate was maintained at 1 mL/min. Data was collected and processed using Waters Empower software.

EBF-2 active compound in tissue samples were analyzed using LC-MS system. Briefly, tissue samples were cut into small pieces and soaked in methanol. The extracts were analyzed on an Agilent 1200 HPLC instrument with an Agilent Eclipse Plus-C18 column (5 μm, 250 mm × 4.6 mm), coupled with an Agilent 6130 Single Quad Mass Spectrometry. The samples were eluted with acetonitrile-water (5-95%) containing 0.1% formic acid (v/v) over 100 mins.

U266 cells (1.0 × 10⁶ cells/mL) were incubated with or without berberine for 3 days. Cells were harvested and total RNA was isolated using Trizol (Gibco BRL, Rockville, MD). The RNA concentrations were quantified by triplicate optical density (OD) readings (Bio-Rad SmartSpect 3000; Bio-Rad, Hercules, CA). Reverse transcription was performed to yield cDNA using ImProm-II™ Reverse Transcriptase (Promega Corporation, Madison, WI) as per the manufacturer’s instructions. The RT-PCR amplification was performed using Maxima™ SYBR Green qPCR Master Mix (2X) kit (Fisher Scientific, Pittsburgh, PA). Primer sequences of XBP1, BLIMP1, STAT-6, BCL-6 and GAPDH were from previously published literature and listed in Supplemental Table 2 (55, 60–62).

To determine the effect of EBF-2’s active compound berberine on mitochondrial metabolism of an IgE producing plasma cell line, we used the Seahorse mitochondrial stress assay. XF cell Mito stress test kits were obtained from Agilent (Santa Clara, CA). Each well of the XF24 cell culture plates was coated with 50 µL of Corning Cell-Tak cell and tissue adhesive at a density of 3.5 µg/cm² for 20 mins followed by washing with 200 µL water and 20 min of drying. Fresh assay medium was prepared by supplementing 2 mM glutamine into XF cell base medium, DMEM with an adjusted pH of 7.4. Next, 1 × 10⁵ of U266 cells resuspended in 100 µL of assay medium were seeded into each well of the coated plate by centrifugation in a swing-bucket rotor at 450 rpm for 1 min without braking. After reversing the orientation of the plates, they were centrifuged again at 650 rpm for 1 min without braking. Plates were transferred to a 37°C incubator not supplemented with CO₂ and incubated for 25-30 mins. Then, 500 µL of warm assay medium, containing DMSO or various concentrations of BBR was slowly and gently added into the wells. After a 15 mins CO₂-free incubation, the cells were ready for the assay on a Seahorse XFe24 Analyzer. Oligomycin (3 µM final concentration), carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP, 3 µM final concentration) and rotenone and antimycin (3 µM and 1 µM final concentration, respectively) were diluted in the assay medium and loaded into ports A, B, and C of the XF24 assay plate. The machine was calibrated, and the assay was performed using the Mito stress test assay protocol per the manufacturer’s recommendations. The extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) were measured under basal conditions after sequential addition of the above-mentioned drugs.

All statistical analyses were performed using GraphPad Prism 9 (San Diego, CA). One-way ANOVA (analysis of variance) was performed followed by Bonferroni correction for all pairwise comparisons. For skewed data, differences between groups were analyzed by one-way ANOVA on ranks followed by Dunn’s method for all pairwise comparisons. Data for symptom score correlation with IgE were analyzed using Spearman correlation. Pearson correlation was used for all other correlation analyses. p-value calculations were two-tailed and a p value < 0.05 was considered as statistically significant.

As compared to the untreated cells, EBF-2 significantly decreased IgE production beginning at 1.9 μg/mL (p<0.05) with complete inhibition of IgE production at 60 μg/mL (p<0.001) ( Figure 1A ). The IC₅₀ value was 4.70 ± 1.16 μg/mL ( Figure 1B ). There was no observed cytotoxicity at any tested concentrations ( Figure 1C ). The parent formula FAHF-2 also significantly decreased IgE production at 62.5 μg/mL (p<0.01) with complete inhibition of IgE production at 500 μg/mL(p<0.001) ( Figure S1A ). The IC₅₀ value of the parent formula FAHF-2 was 79.7 ± 17.39 μg/mL ( Figure S1B ), with no cytotoxic effect across all concentration ( Figure S1C ). Furthermore, we analyzed the effect of three different batches of FAHF-2 (F2-1106, F2-0202, F2-0909) ( Figures S2A–C ) and EBF-2 (EBF-2-0303, EBF-2-0808, EBF-2-0130) ( Figures S2D–F ) on IgE production by U266 cells respectively and found consistent results between different batches. Taken together, EBF-2 is markedly more inhibitory on IgE production than its parent formula while retaining high cellular safety and batch to batch consistency

We next determined EBF-2’s inhibitory effect on IgE levels and its protective effect against peanut anaphylaxis in a murine model ( Figure 2A ). Four weeks after discontinuation of EBF-2 treatment (at week 14 following initial PN sensitization), peanut (PN)-specific IgE levels were significantly reduced by approximately 70% in EBF-2 treated mice compared to sham treated mice ( Figure 2B , p<0.05 vs. Sham). Following intragastric challenge, all sham-treated mice developed anaphylactic symptoms, with symptom severity scores ranging from 2-3. In sharp contrast, EBF-2 treated mice were completely protected from anaphylaxis ( Figure 2C , EBF-2 vs. Sham: median score 0 vs. 2, p<0.05). Hypothermia, a decrease in core body temperature, is a symptom of anaphylaxis in mice. We measured rectal temperatures every 30 minutes after the intragastric challenge, EBF-2 prevented hypothermia, the mean post-challenge body temperature in EBF-2 group was significantly higher than the Sham group and not different from naïve mice ( Figure 2D , sham vs. EBF-2 vs. naïve: 35.18±0.3°C vs. 37.33±0.2°C vs. 37.33±0.2°C, p<0.001 vs. sham). Anaphylaxis is associated with an increase in plasma histamine levels and plasma histamine levels in EBF-2 mice were markedly and significantly lower than in sham mice ( Figure 2E Sham vs. EBF-2 mean±SEM: 14,134±1004 nM vs. 1689±340 nM). The EBF-2 treated group’s plasma histamine level was not significantly different from the normal range of histamine levels in the naïve group (1392±213 nM). PN-specific IgG1 and PN-specific IgG2a production were not affected ( Figures 2F, G ). In this model, symptom severity and plasma histamine levels strongly correlated with IgE levels ( Figures 2H, I , r=0.74, p <0.01; r=0.81, p <0.001, respectively), whereas body temperatures at challenge were inversely correlated with IgE ( Figure 2J , r=-0.73, p <0.01).

We investigated the long-term protection by EBF-2 and its effect on long-lived IgE producing plasma cells in a peanut anaphylaxis murine model ( Figure 3A ). EBF-2 significantly reduced PN-specific IgE during therapy (weeks 8-26) (p<0.001) and maintained consistent levels even after completion of the treatment, up to week 78 (p<0.01, p<0.001) ( Figure 3B ). Post-treatment challenges conducted at week 30, week 40 and week 70, respectively, showed that EBF-2 treated mice were completely protected from anaphylaxis ( Figure 3C , EBF-2 vs Sham, p<0.05), prevented hypothermia ( Figure 3D , p<0.001), and reduced plasma histamine levels ( Figure 3E , p<0.001). Furthermore, EBF-2 treated mice showed significantly lower numbers of the IgE⁺/CD138+ plasma cells from the spleen compared to Sham treated mice ( Figures 3F, G , p<0.001). There was a significant correlation between IgE+/CD138+ plasma cells and PN-specific IgE levels ( Figure 3H , r=0.84, p< 0.0001). These data highlight the persistent protection of EBF-2 from anaphylaxis following peanut exposure.

EBF-2’s safety was evaluated with a sub-chronic toxicity protocol. C3H/HeJ mice were fed 5 times the normal daily dose of EBF-2 for 14 consecutive days and observed for 2 weeks. No morbidity or mortality was observed. Serum ALT and BUN levels of EBF2 and sham-treated mice were all within normal range ( Table 1 ). CBC results in the EBF-2 treated group were similar to those of sham treated mice and were all within the normal range. Thus, the EBF-2 formula has a high safety profile.

We previously showed that berberine (BBR) isolated from FAHF-2 and B-FAHF-2 reduced IgE by plasma cell lines and human PBMCs from food allergic patients (55), and demonstrated that the concentration of BBR can be a pharmacological marker of FAHF-2 and EBF-2. We therefore determined the concentration of BBR in EBF-2 and compared with parent formula FAHF-2 ( Figure 4A ). A total of 29 peaks (P) were detected in FAHF-2 and EBF-2 batches. The major peak 13 (P13) was identified as BBR ( Figure 4B ). The peak area of BBR (Mean ± SED) in EBF-2 was significantly higher than that in FAHF-2 (62.83 ± 3.53% vs 33.26 ± 6.40% overall total peaks p<0.05 ( Supplemental Table 3 ). We have identified other peaks such as Magnoflorine (P6), Phellodendrine (P8), Jatrorrhizine (P12), Ganolucidic acid D (P18), and Ganoderic acid H (P27), but the differences between these peaks were not statistically significant. After the purification process, the constituents (less polar small molecules) in EBF-2 were more concentrated. We also calculated the BBR concentration by using the equation collected from the peak area of BBR standard versus the concentration (BBR concentration (µg/mL) = BBR pear area/1000/72.23). The BBR in FAHF-2 was calculated as 0.36%, while the EBF-2 contains 4.4% of BBR. The BBR concentration was approximately 12 times higher in EBF2 than in FAHF-2 (data not shown). We detected BBR in liver and fat tissue 5 days after oral administration of EBF-2 by LC-MS and demonstrated that BBR is a major bioavailable compound within EBF-2 when compared to naïve mice ( Figure 4C ).

We evaluated the active compound BBR identified from EBF-2 and determined its effect on the regulation of IgE plasma cells at the transcriptional level, in vitro using U266 cells. BBR dose-dependently inhibited IgE production approaching 100% inhibition at 5 µg/mL ( Figure 5A ), without any cytotoxicity across the doses (0.625 – 5 μg/mL) ( Figure 5B ) with an IC₅₀ value 1.946 μg/mL ( Figure 5C ). We evaluated BBR’s effects on the gene expression of XBP1, BLIMP1 and STAT-6, which are genes that have been shown to be upregulated during plasma cell activation. BBR significantly inhibits the gene expression of XBP1, BLIMP1 and STAT6 compared to untreated cells (p <0.01, Figure 5D ). BCL-6 reportedly inhibits long-lived plasma cell survival, and it has been shown to be upregulated in plasma cell activation. Therefore, we measured the effect of BBR on the gene expression of BCL-6, however our results showed that the increase expression of BCL-6 gene was not statistically significant. Taken together, we showed that the suppression of IgE in IgE⁺ plasma B cells by BBR is mediated by down-regulation of XBP1 and BLIMP1.

Plasma cells have a nutrient uptake and energy demand for the production and secretion of antibodies compared to their counterparts of B cells and plasmablasts (44). Emerging evidence suggests that glucose availability and energy metabolism are important for regulating plasma cell antibody production, secretion and survival at post-transcriptional levels (44, 63). Traditionally, BBR has been used as a glucose lowering agent in Type II diabetes through inhibition of mitochondrial respiratory complex I (64–67). We next asked whether BBR disrupts energy metabolism and changes glucose utilization in IgE plasma cell line U266 by a seahorse Mito stress assay. The assay measures mitochondrial respiration and function glycolysis by directly measuring OCR (Oxygen Consumption Rate) and ECAR (Extracellular Acidification Rate). Vehicle treated U266 cells were used to establish some key parameters of mitochondrial function ( Figure 6A ) including basal respiration, proton leak (after oligomycin injection, which inhibits ATP production through complex V), maximal respiration capacity (after injecting FCCP, which uncouples ATP production from electron transport), and non-mitochondrial respiration (after injecting complex I and III inhibitors rotenone and antimycin A). Interestingly, BBR pretreatment significantly suppressed basal OCR (p<0.05) and FCCP-induced maximal OCR (p<0.001) in a dose-dependent manners, with almost complete inhibition of mitochondrial respiration at 3 µg/mL and 10 µg/mL, respectively. To compensate for energy crisis resulted from mitochondrial respiratory inhibition, BBR-treated U266 cells increased the rate of glycolysis to increase ATP production from glycolysis pathway, as indicated by significantly elevated ECAR ( Figure 6B , BBR at 10 µg/mL vs. untreated cells, p<0.05). These results suggest BBR inhibits mitochondrial respiration in IgE⁺PCs, which lead to cellular energy crisis and decreased the availability of glucose molecules for other pathways such as IgE Ab glycosylation.