Metabolism of Rutin and Poncirin by Human Intestinal Microbiota and Cloning of Their Metabolizing α-L-Rhamnosidase from Bifidobacterium dentium

Abstract

To understand the metabolism of flavonoid rhamnoglycosides by human intestinal microbiota, we measured the metabolic activity of rutin and poncirin (distributed in many functional foods and herbal medicine) by 100 human stool specimens. The average α-L-rhamnosidase activities on the p-nitrophenyl-α-L-rhamnopyranoside, rutin, and poncirin subtrates were 0.10 ± 0.07, 0.25 ± 0.08, and 0.15 ± 0.09 pmol/min/mg, respectively. To investigate the enzymatic properties, α-L-rhamnosidase-producing bacteria were isolated from the specimens, and the α-L-rhamnosidase gene was cloned from a selected organism, Bifidobacterium dentium, and expressed in E. coli. The cloned α-L-rhamnosidase gene contained a 2,673 bp sequcence encoding 890 amino acid residues. The cloned gene was expressed using the pET 26b(+) vector in E. coli BL21, and the expressed enzyme was purified using Ni²⁺-NTA and Q-HP column chromatography. The specific activity of the purified α-L-rhamnosidase was 23.3 µmol/min/mg. Of the tested natural product constituents, the cloned α-L-rhamnosidase hydrolyzed rutin most potently, followed by poncirin, naringin, and ginsenoside Re. However, it was unable to hydrolyze quercitrin. This is the first report describing the cloning, expression, and characterization of α-L-rhamnosidase, a flavonoid rhamnoglycosidemetabolizing enzyme, from bifidobacteria. Based on these findings, the α-L-rhamnosidase of intestinal bacteria such as B. dentium seem to be more effective in hydrolyzing (1 →6) bonds than (1 →2) bonds of rhamnoglycosides, and may play an important role in the metabolism and pharmacological effect of rhamnoglycosides.

Introduction

Most herbal medicines and functional foods are orally administered to humans. Their constituents are therefore inevitably brought into contact with gastrointestinal microbiota in the intestine [3,25] and metabolized to bioactive, ineffective, or toxic compounds before they get absorbed from the gastrointestinal tract [10,20,25]. The exact composition of this highly complex gut ecosystem, known as the microbiome, varies among individuals [4,27]. All individuals have their own indigenous intestinal bacterial strains, which consist of trillions of individual microbes in the intestine.

Flavonoid rhamnoglycosides such as rutin, poncirin, naringin, and hesperidin are distributed in many vegetables and herbal medicines [15,17]. In humans, orally administered rutin was not detectable in blood, but the aglycone form quercetin was observed in blood [7]. The biotransformed quercetin exhibits an antitumor effect more potently than rutin [5]. Similar to rutin, orally administered poncirin is also metabolized to ponciretin, which has antiinflammatory and anti-H. pylori effects [12]. Likewise, other flavonoid rhamnoglycosides such as hesperidin and naringin also share similar metabolic fates in the human body [12,29]. Therefore, to absorb the metabolites of these rhamnoglycosides into the blood, these rhamnolgycosides should be metabolized to their aglycones and phenolic acids in the intestine by intestinal microbiota, which produces α-L-rhamnosidase(s) and β-D-glucosidase(s). Of these rhamnoglycosides, orally administered rutin was not absorbed into the blood, whereas their corresponding glucosides, attached to bulky ligands, were transported by the Na+-glucose cotransporter [21]. In fact, the quercetin glucoside or quercetin has been reported to be absorbed directly in the small intestine [7]. Therefore, α-L-rhamnosidase(s) may be essential for the absorption of rhamnoglycosides.

Since the purification of bacterial α-L-rhamnosidases from Bacteroides JY-6 [9], other α-L-rhamnosidases produced by human intestinal microflora such as Clostridium stercorarium [30], Lactobacillus acidophilus [2], Lactobacillus plantarum [1], and Pediococcus acidilactici [19] have also been characterized. Nevertheless, the role of human intestinal bacterial α-L-rhamnosidase(s) in the metabolism of rhamnoglycosides has not been thoroughly investigated.

Here, we report the fecal activities metabolizing flavonoid rhamnoglycosides, rutin and poncirin, in 100 human stool specimens. We also cloned the α-L-rhamnosidase gene from the selected microorganism B. dentium, which was isolated from human fecal microbiota as an α-L-rhamnosidaseproducing lactic acid bacterium, and expressed it in E. coli, and investigated the enzyme’s properties.

 

Materials and Methods

Materials

p-Nitrophenyl-α-L-rhamnopyranoside, rutin, naringin, quercetin, quercitrin, naringenin, hesperidin, and hesperetin were purchased from Sigma (St. Louis, MO, USA). E. coli strain TH2 and pKF3 were purchased from TaKaRa Co. (Tokyo, Japan). E. coli DH5α and BL21(DE3) were purchased from RBC Bioscience Co. (New Taipei, Taiwan). pGEM-T easy vector was purchased from Promega (Madison, WI, USA). pET26b(+) was purchased from Novagen (Darmstadt, Germany).

Poncirin was isolated according to the method of Kim et al. [11], and ginsenoside Re was isolated according to the method of Shin et al. [26].

Subjects

A total number of 100 healthy subjects were chosen from the Korean population (average age, 40.74 ± 13.87 years; age range, 20-72 years; 54 males, 46 females). Exclusion criteria included smoking and medication, including regular or current use of antibiotics. The recruitment of subjects and the collection of their stools were approved by the Committee for the Care and Use of Clinical Study in the Medical School of Kyung Hee University (KMC IRB 0922-08-A1).

Specimen Preparation

The human fecal specimens (about 1 g), prepared according to a previous method [13], were collected in plastic cups 9 h after fasting, and then carefully mixed with a spatula and suspended in cold saline (9 ml). The fecal bacterial suspension was centrifuged at 500 ×g for 5 min. The resulting supernatant was sonicated for 10 min and then centrifuged at 10,000 ×g for 20 min. The resulting supernatant was used to assay the enzyme activity.

α-L-Rhamnosidase Activity Assay

For the assay of α-L-rhamnosidase activity, the reaction mixture (total volume of 0.5 ml) contained 0.1 ml of 1 mmol/l p-Nitrophenylα-L-rhamnopyranoside (pNR), 0.3 ml of 0.1 mol/l phosphate buffer, pH 7.0, and 0.1 ml of the fecal suspension or enzyme solution and was incubated at 37℃ for 30 min [9]. The reaction was stopped by the addition of 0.5 ml of 0.5 mol/l NaOH, centrifuged at 3,000 ×g for 10 min, and the absorbance measured at 405 nm (UV-vis spectrophotometer, Shimadzu UV-1201).

One unit of enzyme activity was defined as the amount required to catalyze the formation of 1.0 µmole of p-nitrophenol per minute under the standard assay conditions. Specific activity was defined in terms of units per milligram of protein. Specific rhamnosidase activity was measured using the substrate pNR.

Assay of Intestinal Bacterial Enzyme Activity Metabolizing Rutin and Poncirin

For the fecal enzyme activity for rutin and poncirin, the reaction mixture (0.5 ml) containing 0.125 ml of the human fecal suspension and 0.1 mmol/l rutin or poncirin was incubated at 37℃ for 4 h, and 1.5 ml of MeOH was added to stop the reaction. The reaction mixture was centrifuged at 3,000 ×g for 10 min and the levels of rutin and poncirin in the resulting supernatant were analyzed by HPLC (Hewlett Packard Series 1050 HPLC system). The instrument was controlled and the data were processed by a HP Chemstation (Rev. A. 09.03). The analytical column was an Agilent Hypersil ODS (100 × 4.6 mm i.d., 5 µm; Agilent Technologies, Santa Clara, CA, USA) protected by a C18 Security Guard Cartridge (Phenomenex, Torrance, CA, USA). The elution solvent was acetonitrile (ACN) and distilled and deionized water (DDW). Separations were performed with a linear gradient of 0~70% ACN in DDW, including 0.05% formic acid, for 15 min and an isocratic elution for 5 min in 70% ACN at a flow rate of 1.0 ml/min, and detection at 280 nm. A sample volume of 20 µl was used for injection.

Isolation of B. dentium and Gene Cloning of α-L-Rhamnosidase

To isolate α-L-rhamnosidase-producing bifidobacteria from human feces, we inoculated the human fecal suspension in BL agar plates and anaerobically cultured for 2 days according to the previous method of Park et al. [23]. The grown clones were cultured in 5 ml of GAM broth and collected (5,000 ×g). Among these collected bacteria, the most potently α-L-rhamnosidase-producing bacterium K13 was selected. K13 was identified to be B. dentium by analyzing its sugar utilization and 16S rRNA sequencing.

The genomic DNA of B. dentium was isolated using a Wizard Genomic DNA purification kit (Promega). The B. dentium α-L-rhamnosidase (BdRham) gene was amplified using the oligonucleotides containing restriction enzyme sites (BamHI, HindIII) at each 5’ end (forward, GGATCCGATGAGAATCATGGACAC; reverse, AAGCTTCAGCGCCACGACCAAGTG). The sequence was confirmed by analyses in both orientations. The amplified fragment was ligated with TA cloning vector, pGEM T easy vector, and then the generated vector was transformed into E. coli strain DH5α. White-positive colonies were picked, and purified plasmids from them were sequenced. The positive clone was digested with two restriction enzymes and then ligated with the linearized pET26b(+) vector by treatment with the same enzymes. The pET26_Bdrham was introduced into the expression host, E. coli strain BL21(DE3).

Protein Purification and Electrophoresis

One positive colony was inoculated into 6 ml of LB broth containing 30 µg/ml kanamycin and incubated overnight at 37℃ and 200 rpm. The fresh overnight culture was seeded in 600 ml of sterile LB broth containing the same antibiotic. The culture was grown at 37℃ and 200 rpm to an optical density at 600 nm of 0.4. At that point, 0.5 mmol/l isopropyl β-D-thiogalactopyranoside (IPTG) was added and the culture was incubated for 24 h at 20℃ and 200 rpm.

The induced cells were harvested by centrifugation at 6,000 ×g for 20 min. The harvested cells were resuspended in Bind buffer (20 mmol/l sodium phosphate buffer (pH 7.0) containing 500 mmol/l NaCl and 20 mmol/l imidazole]), placed on ice, and disrupted by sonication. The supernatant, obtained by centrifugation at 15,000 ×g for 25 min at 4℃, was filtered through a 0.4 µm syringe filter and loaded onto a 5 ml Ni2+-NTA column that was pre-equilibrated with the Bind buffer. The protein was eluted with a linear gradient: 20 to 800 mmol/l imidazole for 20 column volumes. The enzyme fraction was desalted by dialysis in 20 mmol/l sodium phosphate buffer (pH 7.5), and loaded onto a Q-HP ion column pre-equilibrated with the same buffer. rBdRham was eluted using a linear gradient of 0 to 1 mol/l NaCl in sodium phosphate buffer (pH 7.5). The enzyme fraction was dialyzed against 20 mmol/l sodium phosphate buffer (pH 7.0). The pure enzyme was stored at 4℃ until it was used.

The Bradford assay was used to calculate the protein concentration with bovine serum albumin as a standard. Electrophoresis was performed using a polyacrylamide gel (10% separating gel and 3% stacking gel) with or without sodium dodecyl sulfate.

Characterization of Recombinant BdRham

In order to establish the standard reaction conditions for rBdRham, several enzymatic properties were measured using pNR: optimal pH and temperature, effect of salts and divalent metal cations. The reaction mixture containing 100 µl of 1 mmol/l pNR, 100 µl of the enzyme, and 300 µl of specific buffer solution was incubated for 30 min at a specific temperature. The reaction was stopped by adding 500 µl of 0.5 mol/l NaOH, and its absorbance at 405 nm was measured with a UV spectrophotometer (Shimadzu UV-120-02; Tokyo, Japan).

The optimal pH for the purified enzyme was examined in 50 mmol/l sodium citrate buffer (pH 3-6), 50 mmol/l sodium phosphate buffer (pH 6-8), 50 mmol/l MOPS buffer (pH 6.5-7), and 50 mmol/l Tris-HCl buffer (pH 7-10) at 37℃. For the determination of optimal temperature, 50 mmol/l MOPS buffer (pH 7.0) containing the substrate was preincubated at a specific temperature (5℃ to 60℃) for 5 min, and then the enzyme was added. The activities at 37℃ were measured with various concentrations (0 to 500 mmol/l) of sodium chloride in 50 mmol/l MOPS buffer (pH 7.0). The effects of several metals (BaCl2, CaCl2, CoCl2, CuCl2, MgCl2, NiCl2, PbCl2, and ZnCl2, each at 1 mM) on the activity of the purified enzyme were also determined. The relative activities were expressed as a percent of the activity versus the control reaction.

Analysis of the Biotransformed Natural Products by HPLC

Several natural products, rutin, poncirin, naringin, quercitrin, and ginsenoside Re, were added to the enzyme. The reaction mixtures containing 50 µl of substrate (natural products), 50 µl of enzyme, and 400 µl of sodium phosphate buffer (pH 7.0) were incubated at 37℃ for 14 h. In order to stop the reactions, 500 µl of MeOH was added to the reaction mixtures.

The reaction mixtures were analyzed using HPLC (Agilent 1100 series). The analytical column was Capcell PAK UG80 (150 mm × 4.6 mm, 0.5 µm; Shiseido, Japan). DDW containing 0.3% phosphoric acid and ACN were used for mobile phase to analyze rutin, poncirin, naringin, and quercitrin. In order to analyze ginsenoside Re, we used DDW and ACN for the mobile phase. Detection of flavonoid glycosides and ginsenoside Re was carried out at 280 nm and 203 nm, respectively. The retention times of rutin, poncirin, naringin, and ginsenoside Re were 11.3, 17.3, 13.0, and 20.6 min, respectively.

Kinetic Constants

Kinetic constants of pNR were determined by supplying 1.25 µg of enzyme with various concentrations of pNR (from 0.125 to 1 mmol/l). Kinetic constants of several natural products containing a rhamnose moiety in their structure (rutin, poncirin, naringin, quercitrin, and ginsenoside Re) were also measured by supplying 1.87 µg of enzyme with various concentrations of natural products (from 0.0625 to 1 mmol/l). Kinetic constants were derived from Michaelis-Menten equations.

 

Results

Metabolism of Rutin and Poncirin by Human Intestinal Microbiota

To evaluate the metabolic activity of human intestinal microbiota on rutin and poncirin, we first measured the α-L-rhamnosidase activity on pNR for 100 fecal specimens from 100 healthy Korean individuals (Fig. 1A). The α-L-rhamnosidase activity was 0~0.42 µmol/min/mg. The average activities (mean ± SD) of total, female, and male derived specimens were 0.10 ± 0.07, 0.10 ± 0.06, and 0.11 ± 0.09 µmol/min/mg, respectively. There were no significant differences in the activities of α-L-rhamnosidase between males and females, or among different age groups.

Fig. 1.Fecal p-nitrophenyl-α-L-rhamnopyranoside, rutin, and poncirin metabolizing activities. (A) Average α-L-rhamnosidase activity. (B) Rutin- and poncirin-metabolizing activities (black bar, rutin; white bar, poncirin). (C) Profile of the relationship between α-L-rhamnosidase and poncirin-metabolizing activity. (D) Profile of the relationship between α-L-rhamnosidase and rutinmetabolizing activity. (E) Profile of the relationship between rutin-metabolizing and poncirin-metabolizing activity. These enzyme activities were measured for 100 human fecal specimens prepared according to Materials and Methods.

Next, we measured the metabolic activities of 100 Korean fecal specimens on rutin and poncirin (Fig. 1B). The metabolic activities were 0.01~0.42 and 0.01~0.38 pmol/min/mg, respectively. The average metabolic activities for the total, female, and male specimens were 0.25 ± 0.08, 0.27 ± 0.07, and 0.24 ± 0.09 pmol/min/mg for rutin and 0.15 ± 0.09, 0.16 ± 0.09, and 0.13 ± 0.09 pmol/min/mg for poncirin, respectively. These activities were also not different between males and females, or among different age groups. However, the metabolizing activity of the sample on rutin was proportional to that on poncirin (Figs. 1C-1E).

Cloning and Sequence Analysis

To understand the properties of the enzymes metabolizing rutin and poncirin, we selected α-L-rhamnosidase-producing Bifidobacterium dentium from human feces and cloned the α-L-rhamnosidase gene from select B. dentium species for characterization studies. The cloned BdRham nucleotide (GenBank gene Accession No. KF147170) consisted of a 2,673 bp sequence encoding a protein containing 890 amino acid residues. The cloned BdRham gene and amino acid sequence were identical to those of B. dentium Bd1 previously reported in GenBank (CP001750) (Fig. 2). However, the cloned BdRham gene and amino acid sequence shared a very weak homology with the genes and the corresponding proteins from C. stercorarium (GenBank gene Accession No. CP003992), L. acidophilus (GenBank gene Accession No. NC_006814), and L. plantarum (GenBank gene Accession No. NC_020229). The amino acid sequence of the cloned BdRham shared only 32%, 32%, and 20% sequence identity with that of the α-L-rhamnosidases of C. stercorarium, L. acidophilus, and L. plantarum, respectively (sequences were aligned using Clustal Omega).

Fig. 2.DNA sequence alignment of α-L-rhamnosidase from B. dentium (K-13) with α-L-rhamnosidase from B. dentium Bd1. The asterisks below the aligned sequence indicate the same DNA sequence between the two sequences.

Production of Recombinant α-L-Rhamnosidase of B. dentium

To express the cloned α-L-rhamnosidase gene, it was inserted into the pET26b(+) vector, which has a 6×His tag codon, and the expression vector was transformed using E. coli BL21. The addition of the His-tag facilitated the purification of recombinant α-L-rhamnosidase using Ni-HiTrap IMAC HP, and rBdRham was further purified by Q-HP column chromatography. The molecular mass of the purified enzyme was determined to be approximately 100 kDa by SDS-PAGE (Fig. 3A). After the two purification steps, the specific activity of rBdRham was determined to be 23.3 U/mg (Table 1).

Fig. 3.SDS-PAGE of BdRhaI at various purification steps (A) and its optimal pH (B) and temperature stability (C). Samples were analyzed on a 10% SDS–PAGE gel: M, protein size marker (kDa); 1, soluble crude extract; 2, preparation after Ni2+-affinity column; 3, BdRhaI after purification on Q-HP column.

Table 1.One unit of enzyme activity was defined as the amount required to catalyze the formation of 1.0 µmol of p-nitrophenol per minute under the standard assay conditions.

Properties of rBdRham

The enzymatic activity of rBdRham in MOPS buffer (pH 7.0) at various temperatures ranging from 5℃ to 60℃ was measured. The optimal pH for rBdRham activity was 6.0 (Fig. 3B). The optimal temperature for rBdRham activity was about 35℃ and enzymatic activity was stable between 30℃ and 40℃ (Fig. 3C). The activity of rBdRham decreased rapidly above 40℃. However, 61% of the residual activity was observed at 25℃. The activity of rBdRham was not affected by 0-500 mmol/l concentration of salts such as NaCl. However Ca2+ and Mg2+ (0.1 mmol/l) increased the rBdRham activity by 125% and 156%, respectively; but Cu2+ (0.1 mmol/l) decreased the enzyme activity by 62%. Other divalent cations (Ba2+, Co2+, Ni2+, Pb2+, Zn2+) did not influence the enzymatic activity (data not shown).

Substrate Specificity

Measurement of the substrate specificity of rBdRham using synthetic p-nitrophenyl glycosides revealed that pNR acts as a substrate for rBdRham, but not the other p-nitrophenyl glycosides. The only exception was the p-nitrophenyl arabinoside substrate, and when the activity of pNR was set as 100%, rBdRham showed less than 0.2% relative activity on p-nitrophenyl arabinoside. We also measured the substrate specificity of rBdRham for natural products. The rBdRham potently hydrolyzed naturally occurring rhamnoglycosides. Of those tested (rutin, poncirin, naringin, and ginsenoside Re), rutin was the best substrate. However, quercitrin, which has the α-L-rhamnose moiety attached directly to an aglycone, was not hydrolyzed by rBdRham. The Km, Vmax, kcat, and kcat/Km for pNR and some natural substrates (rutin, poncirin, naringin, ginsenoside Re) were determined using the Michaelis-Menten equation (Table 2). The Km and Vmax of rBdRham on pNR, rutin, and poncirin substrates were 1.06 mmol/l and 41.59 U/mg, 2.19 mmol/l and 1.44 U/mg, and 0.37 mmol/l and 0.14 U/mg, respectively.

Table 2.Kinetic values of rBdRham.

 

Discussion

The indigenous intestinal bacterial microbiota in different individuals varies and shows a unique composition [4,27]. The composition of healthy intestinal microbiota is often affected by exogenous and endogenous factors such as diet, antibiotics, and stress. It is relatively stable throughout adulthood in the absence of disease and antimicrobial therapy. Even if these fecal enzyme activities are affected by diet [6,22,24], it could be recovered if the diet or supplements were stopped for a short period [6,14,24]. Ikeda et al. [8] reported that some intestinal bacterial enzyme activities did not appear to be associated with specific populations [8]. These results suggest that all individuals have their indigenous enzymatic activities attributed to the intestinal microbiota. These findings indicate that the activities of the bioactive components present in functional foods and herbal medicines containing hydrophilic constituents may be dependent on the metabolic activity of the intestinal microbiota.

To understand the variations in the metabolic activities of intestinal bacteria from different individuals on herbal constituents, we measured the metabolic activities of 100 human fecal specimens for rutin and poncirin, which are widely distributed in functional foods and herbal medicines, and a synthetic α-L-rhamnosidase substrate, pNR. Their metabolic activities showed significant differences among the tested samples. However, the metabolic activity was not influenced by age or sex. Moreover, we evaluated the relationship between the rutin-metabolic activities and α-L-rhamnosidase activities. The potencies of rutin-metabolic activities in individuals were significantly in proportion to those of poncirin-metabolic ones, but out of proportion to those of p-nitrophenyl-α-L-rhamnopyranoside-metabolic activities. Thus, it is evident that intestinal microflora produce many kinds of α-L-rhamnosidase(s), and among them, only a few select α-L-rhamnosidase(s) may transform rutin and poncirin to their corresponding aglycones, which exhibit potent biological activities compared with their parental constituents. Based on these findings, we believe that the intestinal bacterial metabolic activities of rutin and poncirin are significantly variable in individuals.

In addition, we cloned and expressed the α-L-rhamnosidase gene from B. dentium, isolated from human fecal specimens. The B. dentium α-L-rhamnosidase gene consisted of 2,673 nucleotides, and the molecular mass of the enzyme was 100 kDa. The molecular mass of rBdRham was similar to that of Aspergillus nidulans [16] and Pichia angusta X349 [28]. The molecular mass of rBdRham was higher than that of other intestinal bacteria, such as Lactobacillus plantarum NCC245 [1], Clostridium stercorarium [30], and Pediococcus acidilactici [19]. The optimal temperature for rBdRham was similar to that for the α-L-rhamnosidase from Pseudomonas paucimobilis FP2001[18] and P. angusta X349 [28]. Whereas the optimal temperature for most α-L-rhamnosidases from intestinal bacteria is relatively high (50-70℃), rBdRham is heat-labile. The Vmax of rBdRham on pNR was 41.59 U/mg. This activity was much lower than that of the α-L-rhamnosidase from P. paucimobilis FP2001 [18]. In addition, rBdRham was capable of acting on some flavonoids and saponin glycosides. In contrast to the enzyme from other lactic acid bacteria such as L. plantarum [1] and P. acidilactici [19], the α-L-rhamnosidase from B. dentium could metabolize (1→2) O-glycosides such as poncirin, naringin, and ginsenoside Re, as well as (1→6) O-glycosides such as rutin. However, quercitrin, in which the rhamnose moiety is attached to the aglycone directly, was not hydrolyzed by rBdRham. These findings suggest that rBdRham is an exo-type enzyme, which removes α-L-rhamnose attached to the other sugar moiety of glycosides. The Vmax of rBdRham on rutin, poncirin, naringin and ginsenoside Re was 1.44, 0.14, 0.14, and 0.05 U/mg, respectively. rBdRham was more effective on flavonoid glycosides than on saponin glycosides. rBdRham hydrolyzed the (1→6) bond found in rutin more efficiently than the (1→2) cleavage of the poncirin and naringin bonds. Similar to rBdRham, the enzymes from the human fecal samples metabolized rutin three times more efficiently than poncirin. This is the first report describing the cloning, expression, and characterization of α-L-rhamnosidase, a flavonoid rhamnoglycoside-metabolizing enzyme, from bifidobacteria.

Based on these findings, the α-L-rhamnosidase of intestinal bacteria such as B. dentium seems to be more effective in hydrolyzing (1→6) bonds than (1→2) bonds of rhamnoglycosides and may play an important role in the metabolism and pharmacological effect of rhamnoglycosides.