Study on the Main Components Interaction from Flos Lonicerae and Fructus Forsythiae and Their Dissolution In Vitro and Intestinal Absorption in Rats

The Flos Lonicerae-Fructus Forsythiae herb couple is the basic components of Chinese herbal preparations (Shuang-Huang-Lian tablet, Yin-Qiao-Jie-Du tablet and Fufang Qin-Lan oral liquid), and its pharmacological effects were significantly higher than that in Flos Lonicerae or Fructus Forsythiae, but the reasons remained unknown. In the present study, pattern recognition analysis (hierarchical cluster analysis (HCA) and principal component analysis (PCA)) combined with UHPLC-ESI/LTQ-Orbitrap MS system were performed to study the chemical constitution difference between co-decoction and mixed decoction in the term of chemistry. Besides, the pharmacokinetics in vivo and intestinal absorption in vitro combined with pattern recognition analysis were used to reveal the discrepancy between herb couple and single herbs in the view of biology. The observation from the chemical view in vitro showed that there was significant difference in quantity between co-decoction and mixed decoction by HCA, and the exposure level of isoforsythoside and 3, 5-dicaffeoylquinic acid in co-decoction, higher than that in mixed decoction, directly resulted in the discrepancy between co-decoction and mixed decoction using both PCA and HCA. The observation from the pharmacokinetics displayed that the exposure level in vivo of neochlorogenic acid, 3, 4-dicaffeoylquinic acid, isoforsythoside and forsythoside A, higher than that in single herbs, was the main factor contributing to the difference by both PCA and HCA, interestingly consistent with the results obtained from Caco-2 cells in vitro, which indicated that it was because of intestinal absorption improvement of neochlorogenic acid, 3, 4-dicaffeoylquinic acid, isoforsythoside and forsythoside A that resulted in a better efficacy of herb couple than that of single herbs from the perspective of biology. The results above illustrated that caffeic acid derivatives in Flos Lonicerae-Fructus Forsythiae herb couple could be considered as chemical markers for quality control of its preparations.


Introduction
Herbs used together in couples (Yaodui in Chinese) are the basic composition units of Chinese herbal formulas and have special clinical significance in Traditional Chinese Medicine (TCM). The herb couples are much simpler than other complicated formulas yet retain the basic therapeutic features. It is important to elucidate the compatibility foundation of TCM formulas by investigating the law and effective substance of herb couples, which are the major areas of research supported by Chinese government [1]. There are magnitudes of methods to study the compatibility rule and principles of herb couples to some extent including literature research [2], extraction and separation [3], pharmacological effects [4][5][6], pharmacokinetics in vivo of active components [7] and serum pharmaco-chemistry methods [8], etc. However, up to now, few studies were performed to explain the reasonability of herb couples based on the systemic biopharmaceutics both in vivo and in vitro.
Flos Lonicerae (FLJ) possesses wide pharmacological actions, such as antibacteria, anti-inflammation, antivirus, antiendotoxin, blood fat reduction, etc [9], and Fructus Forsythiae (FF) has antibacterial, antiviral, antioxidant, anti-inflammatory, anti-obesic effects, etc [10]. The two herbs are the basic components of Chinese herbal preparations such as Shuang-Huang-Lian oral liquid, Yin-Qiao-Jie-Du tablet, Qin-Re-Jie-Du oral liquid and Fufang Qin-Lan oral liquid, which are extensively used in clinical practice [11], and it was shown by Lin (2008) [12] that the pharmacological effects such as anti-inflammatory and antipyretic effects in FLJ-FF herb couples were significantly stronger than that in FLJ or FF, but the reasons were unknown. Thus, it was presumed that the dissolution in vitro of main ingredients was significantly higher than that of FLJ and FF or the intestinal absorptions of active components were improved combined with FLJ or FF.
In short, the current study aims to demonstrate the optimal efficacy of FLJ-FF herb couple based on thoughts we provided above analyzed by both hierarchical cluster analysis (HCA) and principal component analysis (PCA) for the further development of herb couples preparations. The specific objectives of the current study include: (1) To study the difference between co-decoction and mixed decoction based on the dissolution in vitro. (2) To research the discrepancy between herb couple and single herbs based on the pharmacokinetics in vivo. (3) To illustrate the difference via in vitro Caco-2 cells between herb couple and single herbs.

Ethics statement
All procedures had the approval of the Animal Ethics Committee of the Nanjing University of Chinese Medicine.
Trifluoroacetic acid, Lucifer yellow (LY) and DMSO were purchased from Sigma Chemical Co. (St. Louis, MO). Phosphoric acid, acetic acid, formic acid, methanol and acetonitrile (HPLC grade) were purchased from Merck (Merck, Germany), and water was purified by a Milli-Q water purification system (Millipore, Bedford, MA, USA). All other chemicals and reagents were of analytical grade.
Male Sprague-Dawley (SD) rats (,250 g) were supplied by the Experimental Animal Center of Nanjing University of Chinese Medicine (Certificate No. SCXK2008-0033). The experimental procedures were in compliance with the animal ethics committee of the Nanjing University of Chinese Medicine.
Preparation of mixed decoction and co-decoction FLJ (30.0 g, 40 meshes) and FF powders (30.0 g, 40 meshes) were decocted with boiling water (1:10, w/v) for 60 min by heating reflux, respectively. The two extracted solutions were filtered through 5 layer gauzes with the concentrations of 100 mg raw medicine per milliliter, respectively. The mixed decoction was prepared by mixing the two extracted solutions (1:1, v/v). FLJ (15.0 g, 40 meshes) combined with FF powders (15.0 g, 40 meshes) (1:1(w/w)) as co-decoction were decocted with boiling water (1:10, w/v) for 60 min by heating reflux. The extracted solution was filtered through 5 layer gauzes with the concentrations of 100 mg raw medicine per milliliter.

Rat in vivo pharmacokinetics study
Product A (extract mixed with FLJ extract and water, (1:1, v/v)), product B (extract mixed with FF extract and water, (1:1, v/v)) and product C (extract mixed with FLJ and FF extracts, (1:1, v/v)) were given to oral administration to rats, respectively.
In order to study the pharmacokinetics in vivo based on drugdrug interaction of main ingredients between FLJ and FF, oral administration to rats at the same concentration was necessary, and we found that there was no significant difference of ingredients in product A or product B except caffeic acid, quinic acid, rutin, hyperoside, isoquercitrin and quercetin in quantity ( Fig. 2), compared with that in product C group.
Male SD rats (,250 g) were kept in an environmentally controlled breeding room (temperature: 2062uC, relative humidity: 6065%) for 1 week. The animals were fasted for 12 h prior to drug administration. The rats were randomly divided into three groups with no less than six rats in each group to receive various administrations at a single oral dose (10 mL?kg 21 ) by gastric gavage. After dosing for 0, 5, 10, 20, 30, 40, 55, 70, 100, 160, 250, 600 and 1440 min, blood was collected from the pre-intubated catheter and put into tubes with heparin sodium injection (10 mL) at predetermined time points. Subsequently, plasma was prepared by centrifugation at 1,8166g, 4uC for 7 min, and immediately analyzed or stored at 270uC for further analysis.

In vitro Caco-2 monolayer model
Caco-2 cells were cultured in high glucose DMEM with 10% fetal bovine serum, 1% nonessential amino acids. Cells were cultured in a humidified atmosphere of 5% CO 2 at 37uC. After reaching 80% confluens, Caco-2 cells were harvested with 0.05% trypsin-EDTA solution and seeded on top of CC inserts in 6-well plates, which has a surface area of 4.2 cm 2 , at a density of 1.0610 5 cells/cm 2 . The protocols for cell culture in Transwell inserts were similar to those described previously [11].
Hank's balanced salt solution (HBSS) was used as the transport buffer for the transport study in Caco-2 cells monolayer model. It was prepared by dissolving 9.5 grams of commercial available HBSS powder in 1000 mL water. The pH value of the buffer was adjusted to pH 6.0 by 85% of phosphoric acid.
MTT test was used to estimate the potential cytotoxicities of the studied product A, product B and product C toward Caco-2 cells. The Caco-2 cells were seeded onto a 96-well plate at a seeding density of 5610 4 cells/well in DMEM culture medium and cultured at 37uC for 24 h. Subsequently, the culture medium was replaced with 100 mL of product A, product B and product C dissolved in HBSS (pH 6.0) at different studied concentrations. Blank HBSS (pH 6.0) was employed as a negative control. Then the 96-well plate was incubated at 37uC for 24 h. Thereafter, 20 mL of 5 mg/mL MTT solution in HBSS was added to each well and the plate was incubated for another 4 h. The solutions in each well were then removed followed by dissolving the remained formazan crystals in the cells with 200 mL of DMSO. The absorbance of the mixture in the 96-well plate was then measured with a Kinetic microplate reader (Molecular Devices) at 570 nm. The cytotoxicity of each compound was calculated as the percentage of the absorbance relative to that of the negative control.
Cell culture experiments were described previously [11]. Briefly, after culture medium was aspirated, the cell monolayers were washed three times with blank HBSS. The transepithelial electrical resistance (TEER) values of cell monolayers were measured, which were more than 250 V6cm 2 . The monolayers were incubated with the blank HBSS for 1 h with 37uC. Thereafter the incubation medium was aspirated. Afterwards, a solution containing the compound was loaded onto the apical side. The amounts of transported compound were measured as a function of time. Donor samples (400 mL) (Apical side) and receiver samples (400 mL) (Basolateral side) were taken at different times (typically 1 h), followed by the addition of 400 mL drug donor solution to the donor side (AP) and 400 mL of blank buffer to the receiver side (BL). The samples were taken at 0, 1, 2, 3 and 4 h after incubation. At the end of the transport experiment, integrity of the monolayer was monitored by TEER value.
Chemical sample (mixed decoction and co-decoction) analyzed by UHPLC-ESI/LTQ-Orbitrap-MS. Mixed decoction and co-decoction were diluted five times with methanol, respectively, and then filtered through 0.22 mm membrane before injection into the UHPLC-ESI/LTQ-Orbitrap-MS system for analysis. The UHPLC analysis was performed on the Dionex UltiMate 3000 analytical system acquired from Fisher Scientific (Thermo Fisher scientific, Waltham, MA, USA) that consisted of an autosampler equipped with a column oven, a tray compartment cooler, and a quaternary pump with a built in solvent degasser, all piloted by Xcalibur software. The chromatographic separation was achieved using Syncronis C 18 column (10062.1 mm, 3 mm) (Thermo Scientific, Waltham, MA, USA). The injection volume used was 5 mL. The mobile phase was composed of A (acetonitrile) and solvent B (0.05% formic acid, v/v) with a linear gradient elution: 0-3 min, 5-5% A; 3-25 min, 5-22% A; 25-30 min, 22-60%; 30-35 min, 60-95% A; 35-36 min, 95-5% A, and hold for 4 min, at a flow rate of 400 mL/min, resulting chromatographic full-width-at-half-maximum (FWHM) of 3-5 s. The column oven and tray cooler temperatures were set to 30 and 4uC, respectively.
The Dionex UltiMate 3000 UHPLC system was hyphenated with a LTQ-Orbitrap XL mass spectrometer (Thermo Scientific, Waltham, MA, USA). The system was equipped with an ESI source, operated in positive ionization mode using the following parameters: capillary temperature 350uC, capillary voltage 47 V, sheath gas flow 30 arbitrary units, auxiliary gas 8 arbitrary units, source voltage 4.5 kV and tube lens voltage 130 V. Spectra was recorded in the range of m/z 50-1500 Da with a resolution of 30,000. For identification purposes, two scan events were applied for the MS experiments. The first was in a full scan MS mode and the second was a data dependent scan that selected the most intense ion or specified ions in another setting from the first scan event for the acquisition of MS/MS spectra. The collision energy for collision-induced dissociation or high-collision energy dissociation mode was adjusted to 35% of maximum, and the isolation width of precursor ions was m/z 2.0 Da. In the qualitative study, the compounds were identified by comparison with reference compounds when available and with literature data, from retention times, MS, MS/MS and MS n analysis. In the quantitative study, full scan MS mode was applied, and then extracted ion chromatograms were generated from their theoretical exact masses using a mass tolerance of 5 ppm. After integration of the peaks, their respective areas were used for the quantifications.
Chemical sample (product A, product B and product C) quantified by UPLC-ESI-MS/MS. Product A, product B and product C were diluted five thousand times with 10% acetonitrile/ methanol (4:1, v/v) containing 0.4% formic acid and 0.5 mM sodium formate, respectively, and then filtered through 0.22 mm membrane before injection into the UPLC-ESI-MS/MS system for analysis. The methodology validations have been studied and to be published elsewhere.

Calculation
For Caco-2 monolayer model, the apparent permeability coefficient (P app ) was calculated as P app = [(dQ/dt)]/[A6C], dQ/dt (mg/S) was the flux rate, A was the effective surface area of the cell monolayer (4.2 cm 2 ), and C 0 (mg/mL) was the initial drug concentration in the donor chamber.

Pharmacokinetic analysis
The main pharmacokinetic parameters including the peak plasma concentration (C max ), the time to C max (T max ), the AUC

Statistical analysis
Statistical significance in the P app values and pharmacokinetic parameters were estimated by the analysis of variance (Student ttest) or one-way ANOVA. A p value of less than 0.05 was considered to be significantly different. All data were expressed as mean6SD.

Pattern recognition analysis
The data for chemical difference between co-decoction and mixed decoction, and intestinal absorption both in vitro and in vivo were all visualized by applying pattern recognition methods, such as HCA and PCA, which has been extensively applied to the biomolecules analysis [13][14][15][16]. Both HCA and PCA were done by SPSS 20.0 software. Between-group linkage method was applied, and squared Euclidean distance was selected as measurement.

Chemical constitution difference between co-decoction and mixed decoction by pattern recognition analysis
As illustrated in Fig. 3A, B and Table 1, we found that 32 peaks and 31 peaks were detected in FLJ extract and FF extract in the positive ion model by UHPLC-ESI/LTQ-Orbitrap MS system, respectively and the influence of the co-decoction on the quality was little, compared with the mixed decoction (Fig. 3C, D). However, it was shown (Fig. 4a) that there was significant difference in quantity between mixed decoction and co-decoction analyzed by HCA, and we found (Fig. 4b, c) that the most possible ingredients resulted in the discrepancy between co-decoction and mixed decoction were isoforsythoside and 3, 5-dicaffeoylquinic acid analyzed by PCA combined with HCA. Besides, the contents of isoforsythoside and 3, 5-dicaffeoylquinic acid in co-decoction were significantly higher than that in mixed decoction, which indicated that isoforsythoside and 3, 5-dicaffeoylquinic acid might be the main chemical markers in herb couples.
The bioavailability interaction of main components from FLJ and FF using in vivo pharmacokinetics study As shown in Fig. 5 and Table 2, C max , MRT, T 1/2z and AUC 0R' of caffeic acid derivatives, flavonoids glucoside, lignins glucoside and iridoids glucoside were improved, but that of flavonoids aglycone and lignins aglycone were decreased in product C group compared with that in product A or B group (most of them had significant differences), which indicated that the intestinal absorption of caffeic acid derivatives, flavonoids glucoside, lignins glucoside and iridoids glucoside might be improved or the transformation from glucoside to aglycone by bacterial metabolism might be inhibited as FLJ combined with FF. In addition, the AUC 0Rt (Fig. 6) as variates analyzed by both HCA and PCA of neochlorogenic acid, chlorogenic acid, cryptochlorogenic acid, 3, 5-dicaffeoylquinic acid, 3, 4-dicaffeoylquinic acid, luteoloside, genistin, astragalin, loganin, pillyrin, pinoresinol-b-Dglucoside, isoforsythoside, forsythoside A and forsythoside B in product C group were increased significantly to 286% (1554263640.1) ng?min/mL, 215% (76148614584) ng?min/ mL, 235% (1924864037.2) ng?min/mL, 162% (4944.76595.54) ng?min/mL, 318% (1464.66309.02) ng?min/mL, 137% (285.71676.804) ng?min/mL, 131% (32.88062.0100) ng?min/ mL, 145% (250.30628.344) ng?min/mL, 121% (6450.56484.60) ng?min/mL, 213% (3374.06265.32) ng?min/mL, 299% (1025.26171.44) ng?min/mL, 203% (1232061429.1) ng?min/ mL, 230% (252296426.79) ng?min/mL and 175% (5642.56122.33) ng?min/mL, respectively, compared to that in product A or B groups, but the AUC 0Rt of genistein, luteolin and arctigenin in product C group were lower significantly when the same dosages were administrated to rats. Besides, the bioavailability of caffeic acid in product C group had no significant difference, compared with that in product B group, and surprisingly significantly lower than that in product A group, and the bioavailability of quinic acid in product C group had no significant difference, compared with that in product A group, and higher significantly than that in product B group although the contents of caffeic acid and quinic acid (Fig. 3) in product C group were significantly higher than that in product A or B group. And the bioavailability of hyperoside in product C group was significantly higher than that in product A or B group, which corresponded to their administration dosages to rats, but that of quercetin was opposite. In addition, the bioavailability of rutin in product C group was significantly higher than that in product A group, but no significant difference compared to that in product B group, and that of isoquercitrin was no significantly different among product A, B and C groups, though the dosage of product C group was higher than that of product A or B group. In short, it was found surprisingly that the bioavailability of some ingredients, such as caffeic acid and quercetin, etc. were significantly higher than that in product C, which was contrary to their administration dosages to rats. Therefore, it is necessary for us to consider caffeic

Pattern recognition analysis of pharmacokinetic profiling in vivo
Hierarchical clustering of the pharmacokinetics was presented in Fig. 7a1 and a2. The dendrogram showed the cluster relationships among product A, product B and product C groups in vivo. We found that the samples were divided into two main clusters, respectively. As shown in Fig. 7a1, cluster I represented product A1-A7, and cluster II grouped product C1-C7, which was illustrated by PCA (Fig. 7b1) by which samples were also clearly separated two domains, consistent with HCA analysis. Also in Fig. 7a2, product B1-B7 was marked in cluster I, and product C1-C6 was divided into cluster II, which was proved by PCA (Fig. 7b2) that samples were clearly grouped into two domains, correspondence to HCA analysis. In addition, it was displayed (Fig. 7c1) that the exposure level in vivo of one group of ingredients (neochlorogenic acid and 3, 4-dicaffeoylquinic acid) were the main influencing factor contributing to the difference between product C and product A group by both PCA and HCA. Similar results were obtained in the case of another group of ingredients (isoforsythoside and forsythoside A) between product C and product B group (Fig. 7c2).

The absorption interaction of main component from FLJ and FF using in vitro Caco-2 cells model
Caco-2 cells were exposed to various concentrations of product A, B and C (0.039, 0.078, 0.16, 0.31, 0.63, 1.3, 2.5, 5.0, 10 and 20 mg raw medicine per milliliter. It was shown that product A, B and C at different concentrations were all safe for the Caco-2 cells from MTT test. As shown in Fig. 8A, the P app for caffeic acid, genistein, luteolin, quercetin, neochlorogenic acid, chlorogenic acid, cryptochlorogenic acid, 3, 5-dicaffeoylquinic acid, 3, 4-dicaffeoylquinic acid, genistin, luteoloside, astragalin, hyperoside, isoquercitrin, rutin and loganin in product C group were increased significantly to 275%, 293%, 240%, 125%, 136%, 140%, 133%, 164%, 137%, 165%, 174%, 131%, 129%, 169%, 213% and 211% respectively compared to that in product A group, and the P app (Fig. 8B) was increased to 247%, 256%, 195%, 234%, 210%, 110%, 187%, 165% and 210% for caffeic acid, quercetin, hyperoside, isoquercitrin, rutin, arctigenin, isoforsythoside, forsythoside A and forsythoside B, respectively compared with that in product B group, though some of them had different concentrations. Besides, the P app for quinic acid in product C group was increased to 301% compared to that in product A group, but decreased to 24.0% compared to that in product B group. Meanwhile, there were no significant difference in P app for pillyrin and pinoresinol-b-Dglucoside between product C group and B group. In addition, we found that dipsacoside B and macranthoidin B in product C or A group and arctiin in product C or B group might not be absorbed well into Caco-2 cells. The result above showed that the P app for most of ingredients such as caffeic acid derivatives were improved significantly by FLJ combined with FF, which might be attributed to the influence of components in FLJ or FF on the efflux transporters.
Pattern recognition analysis of absorption profiling in vitro Fig. 9a1 and a2 showed the hierarchical clustering of the absorption in vitro, and we found the samples were divided into two main clusters, respectively. As shown in Fig. 9a1, cluster I represented product A1-A4, and cluster II grouped product C1-C4, which was illustrated by PCA (Fig. 9b1) that samples were also clearly separated two domains, consistent with HCA analysis. Besides, it was shown (Fig. 9a2) that product B1-B4 was marked in cluster I, and product C1-C4 was divided into cluster II, which was also proved by PCA (Fig. 9b2) that samples were clearly grouped into two domains, correspondence to HCA analysis. Meanwhile, it was displayed (Fig. 9c1) that the intestinal absorption level in vitro for one group of ingredients (neochlorogenic acid and 3, 4-dicaffeoylquinic acid) were the main influencing factors leading to the difference between product C and product A group. Similar results were obtained in the case of another group of ingredients (isoforsythoside and forsythoside A) between product C and product B group (Fig. 9c2).

Discussion
In the present study, the systemic biopharmaceutics combined with HCA and PCA were firstly performed to rapidly elucidate the compatibility foundation of FLJ-FF herb couple, and it was found consistently (Fig. 4 & 7) that the caffeic acid derivatives including isoforsythoside, 3, 5-dicaffeoylquinic acid, neochlorogenic acid, 3, 4-dicaffeoylquinic acid and forsythoside A were the most important ingredients resulting in difference between herb couple and single herbs, and interestingly, their exposure levels in the views of not only chemistry in vitro but also biology in vivo were all significantly higher than that in single herbs, highly consistent with the data that the pharmacological activities such as antiinflammatory and antipyretic effects in FLJ-FF herb couple were significantly higher than that in FLJ or FF [17]. Besides, it was reported that caffeic acid derivatives such as isochlorogenic acids and forsythoside A had strong antioxidant, antibacterial and antiviral activities [18][19][20], and chito-oligosaccharide (COS) at dosage of 25 mg/kg could improve their pharmacological effects such as antiviral activity via enhancing the intestinal permeabilities and the in vivo bioavailabilities of caffeic acid derivatives significantly in FF-FLJ herb couple preparations [11]. The studies above indicated that the caffeic acid derivatives might be the most significant components contributing to the pharmacological effects.
It was shown (Fig. 9) from Caco-2 cells in vitro combined with HCA and PCA that neochlorogenic acid, 3, 4-dicaffeoylquinic acid, isoforsythoside and forsythoside A were the main ingredients resulting in the difference between single herbs and herb couple, and the values of intestinal absorption in vitro were significantly higher than that in single herbs, surprisingly consistent with the results obtained from pharmacokinetics in vivo (Fig. 7), which indicated that it was because of intestinal absorption improvement of neochlorogenic acid, 3, 4-dicaffeoylquinic acid, isoforsythoside and forsythoside A that the efficacies of herb couple were better than that of single herbs.
In the previous study, we found that the poor intestinal absorption of the caffeic acid derivatives was one of the most important factors resulting in the low oral bioavailability, and they permeated mainly via the paracellular pathways in the intestine, and the intestinal absorption of phenolic acid such as chlorogenic acid, neochlorogenic acid, cryptochlorogenic acid, 3, 4-dicaffeoylquinic acid and 3, 5-dicaffeoylquinic acid were influenced by P-gp, MRP2 and BCRP, but that of phenylethanoid glycosides such as forsythoside A, isoforsythoside and forsythoside B were affected by P-gp and MRP2, not BCRP [21][22]. Although the bioavailability of active ingredients (Fig. 6) (neochlorogenic acid, 3, 4-dicaffeoylquinic acid, isoforsythoside and forsythoside A) in herb couple were improved compared with that in single herbs, their intestinal absorptions (Fig. 8) were still unsatisfactory. Thus, the studies on how to improve the bioavailability of the caffeic acid derivatives in herb couple by pharmaceutical methods such as absorption enhancers based on tight junctions need to be further investigated.

Conclusion
Current findings from both the chemical and biological aspects consistently demonstrated that the biopharmaceutics characteristics (dissolution in vitro and intestinal absorptions both in vivo and Figure 9. The absorption differences among product A, B and C groups analyzed by both HCA and PCA. (a1, a2: HCA; b1, b2: PCA; The most important ingredients influencing the difference between product C and product A group (c1) or between product C and product B group (c2)). doi:10.1371/journal.pone.0109619.g009 in vitro) of caffeic acid derivatives in FLJ-FF herb couple, higher than that in FLJ or FF, contributed to the optimal efficacy of herb couples analyzed by both HCA and PCA, which indicated that caffeic acid derivatives should be considered as chemical markers to control the quality of its preparations.