13-(2-Methylbenzyl) Berberine Is a More Potent Inhibitor of MexXY-Dependent Aminoglycoside Resistance than Berberine

We previously showed that berberine attenuates MexXY efflux-dependent aminoglycoside resistance in Pseudomonas aeruginosa. Here, we aimed to synthesize berberine derivatives with higher MexXY inhibitory activities. We synthesized 11 berberine derivatives, of which 13-(2-methylbenzyl) berberine (13-o-MBB) but not its regiomers showed the most promising MexXY inhibitory activity. 13-o-MBB reduced the minimum inhibitory concentrations (MICs) of various aminoglycosides 4- to 128 fold for a highly multidrug resistant P. aeruginosa strain. Moreover, 13-o-MBB significantly reduced the MICs of gentamicin and amikacin in Achromobacter xylosoxidans and Burkholderia cepacia. The fractional inhibitory concentration indices indicated that 13-o-MBB acted synergistically with aminoglycosides in only MexXY-positive P. aeruginosa strains. Time-kill curves showed that 13-o-MBB or higher concentrations of berberine increased the bactericidal activity of gentamicin by inhibiting MexXY in P. aeruginosa. Our findings indicate that 13-o-MBB inhibits MexXY-dependent aminoglycoside drug resistance more strongly than berberine and that 13-o-MBB is a useful inhibitor of aminoglycoside drug resistance due to MexXY.


Introduction
Pseudomonas aeruginosa is a major cause of nosocomial infections. Treatment of P. aeruginosa infections with antimicrobial concentrations insufficient to inhibit P. aeruginosa growth results in the emergence of new multidrug resistant P. aeruginosa strains [1] that are difficult to eradicate and may increase mortality [2].
Wild-type P. aeruginosa expresses low MexXY levels but elevated MexXY has been detected in aminoglycoside-resistant P. aeruginosa strains [11,12]. Therefore, the development of MexXY inhibitors would allow the use of lower concentrations of aminoglycoside drugs that can cause severe side effects such as kidney damage [13].
There have been various reports of inhibitors of RND-type multidrug efflux pumps, but no clinical applications have been published to date [14]. Phenyl-arginine-β-naphthylamide (PAβN, MC-207,110), a well-known efflux pump inhibitor, does not inhibit aminoglycoside resistance due to MexXY [15]. We previously reported that berberine attenuates MexXY-dependent aminoglycoside resistance in P. aeruginosa [15], consistent with a recent report that berberine has high affinity to a MexXY model protein in silico [16].
Berberine is an isoquinoline quaternary alkaloid isolated from many kinds of medicinal plants such as Coptis chinensis, Coptis rhizome, Coptis japonica and Phellondendron amurense [17] and has weak antibacterial activity against Gram-negative bacteria such as P. aeruginosa [18]. Various derivatives of berberine have been developed and studied for their anti-hyperglycemic, anti-cancer, anti-inflammatory, anti-Alzheimer's disease and anti-microbial activities [19]. Derivatives with multidrug resistance pump inhibitory activity against Staphylococcus aureus [20] and that reduce fluconazole resistance against Candida albicans [21] have been reported. In addition, quaternary ammonium compounds inhibit the biofilm formation in P. aeruginosa and C. albicans have been reported [22].
The optimum concentration of berberine to inhibit MexXY in P. aeruginosa cells is more than 512 µg/mL [15], which is too high for clinical application. In this study, we aimed to synthesize berberine derivatives with higher MexXY inhibitory activities.

Antibacterial Activity of Berberine Derivatives toward P. aeruginosa
We first measured the minimum inhibitory concentrations (MICs) of 11 berberine derivatives ( Figure 1) synthesized against P. aeruginosa mutants PAGU g 1927, which expresses MexXY, and PAGU g 1931, which does not express MexXY. A difference in the activity of a derivative toward the two strains indicates that the MexXY activity is not masked by the other four pumps (MexAB, MexCD, MexEF and MexVW) [15]. The MIC values of the berberine derivatives were lower in both strains compared to berberine ( Table 1), suggesting that these berberine derivatives had higher anti-pseudomonas activity compared with berberine. These berberine derivatives showed similar MIC values that differed no more than, 4-fold. Their MIC values against PAGU g 1927 were 2-fold greater than against PAGU g 1931, indicating that the derivatives are MexXY substrates. resistant P. aeruginosa strains [11,12]. Therefore, the development of MexXY inhibitors would allow the use of lower concentrations of aminoglycoside drugs that can cause severe side effects such as kidney damage [13].
There have been various reports of inhibitors of RND-type multidrug efflux pumps, but no clinical applications have been published to date [14]. Phenyl-arginine-β-naphthylamide (PAβN, MC-207,110), a well-known efflux pump inhibitor, does not inhibit aminoglycoside resistance due to MexXY [15]. We previously reported that berberine attenuates MexXY-dependent aminoglycoside resistance in P. aeruginosa [15], consistent with a recent report that berberine has high affinity to a MexXY model protein in silico [16].
Berberine is an isoquinoline quaternary alkaloid isolated from many kinds of medicinal plants such as Coptis chinensis, Coptis rhizome, Coptis japonica and Phellondendron amurense [17] and has weak antibacterial activity against Gram-negative bacteria such as P. aeruginosa [18]. Various derivatives of berberine have been developed and studied for their anti-hyperglycemic, anti-cancer, antiinflammatory, anti-Alzheimer's disease and anti-microbial activities [19]. Derivatives with multidrug resistance pump inhibitory activity against Staphylococcus aureus [20] and that reduce fluconazole resistance against Candida albicans [21] have been reported. In addition, quaternary ammonium compounds inhibit the biofilm formation in P. aeruginosa and C. albicans have been reported [22].
The optimum concentration of berberine to inhibit MexXY in P. aeruginosa cells is more than 512 µg/mL [15], which is too high for clinical application. In this study, we aimed to synthesize berberine derivatives with higher MexXY inhibitory activities.

Antibacterial Activity of Berberine Derivatives toward P. aeruginosa
We first measured the minimum inhibitory concentrations (MICs) of 11 berberine derivatives ( Figure 1) synthesized against P. aeruginosa mutants PAGU g 1927, which expresses MexXY, and PAGU g 1931, which does not express MexXY. A difference in the activity of a derivative toward the two strains indicates that the MexXY activity is not masked by the other four pumps (MexAB, MexCD, MexEF and MexVW) [15]. The MIC values of the berberine derivatives were lower in both strains compared to berberine ( Table 1), suggesting that these berberine derivatives had higher antipseudomonas activity compared with berberine. These berberine derivatives showed similar MIC values that differed no more than, 4-fold. Their MIC values against PAGU g 1927 were 2-fold greater than against PAGU g 1931, indicating that the derivatives are MexXY substrates.

Inhibition of Drug Resistance in P. aeruginosa Using Combined Berberine Derivatives
We investigated the MexXY inhibitory activities of the berberine derivatives by measuring the MICs of gentamicin in the presence of the derivatives against P. aeruginosa mutants PAGU g 1927 and PAGU g 1931 ( Table 2). The concentrations of the berberine derivatives were 1/2, 1/4, or 1/8 that of the MICs for PAGU g 1931.
The MIC of gentamicin for PAGU g 1927 in the presence of 256 µg/mL berberine was 128 µg/mL ( Table 2), which is one-eighth that of gentamicin alone (Table 1). Compounds 1-5 and 7 exhibited apparently increased MexXY inhibitory activity, with compound 7 reducing the MIC of gentamicin 64-fold. Compound 7 was named 13-o-MBB. Compounds 8 and 9, which are regioisomers of 13-o-MBB, increased sensitivity to gentamicin by up to 4-fold and were the weakest MexXY inhibitors.
We also examined changes in sensitivity to drugs other than gentamicin by combination with the berberine derivatives ( Table 3). The combined use of 13-o-MBB 128 µg/mL reduced the MIC values of various substrate drugs (amikacin, tobramycin, kanamycin, gentamicin, spectinomycin, norfloxacin, ciprofloxacin, erythromycin, carbenicillin, ethidium bromide, tetracycline, chloramphenicol, azithromycin and cefepime) targeting MexXY by 2-fold to 16-fold ( Table 3). The regiosomer 13-(3-methylbenzyl) berberine bromide (13-m-MBB) increased the spectinomycin sensitivity of PAGU g 1927 8-fold and that of cefepime 4-fold at 64 µg/mL, whereas the other derivatives did not change the sensitivity to spectinomycin more than 2-fold. In addition, the combined use of 64 µg/mL of 13-(4-methyl-benzyl)-berberine bromide (13-p-MBB) increased sensitivity t cefepime 4-fold for PAGU g 1927. Moreover, the sensitizing action of 13-(3-methyl-benzyl)-berberine bromide and 13-(4-methyl-benzyl)-berberine bromide did not exceed that of 13-o-MBB for PAGU g 1927. Taken together, these results suggest that the o-methyl group of 13-o-MBB increases antimicrobial sensitivity in a MexXY-dependent manner. We investigated whether the inhibitory action of 13-o-MBB against MexXY-dependent drug resistance can be observed in PAGU 1606, a multidrug resistant P. aeruginosa clinical strain, and its MexXY-deficient strain PAGU g 1659 ( Table 4). The MIC of amikacin alone against PAGU 1606 was 256 µg/mL and 64 µg/mL when combined with berberine. In contrast, the combined use of 13-o-MBB and amikacin decreased the MIC to 16 µg/mL. Thus, 13-o-MBB inhibits amikacin resistance 4-fold more effectively than berberine in a MexXY-dependent drug resistant strain. Another aminoglycoside drug, 13-o-MBB, inhibited drug resistance two to four times stronger than berberine but had no greater effect on the drug resistance of PAGU 1606 than the other aminoglycosides. However, the MICs of norfloxacin, erythromycin and azithromycin were increased towards PAGU g 1659, a pump-deficient strain. The sensitizing action of 13-o-MBB for various aminoglycosides was compared with that of berberine at the same concentrations as tested against P. aeruginosa clinical strains but using Burkholderia cepacia PAGU 0013 and Achromobacter xylosoxidans PAGU 0002 ( Table 5). The two non-P. aeruginosa strains are naturally resistant to aminoglycosides due to the presence of MexXY orthologs [7,23]. 13-o-MBB at the same concentration as berberine increased the sensitivity to the aminoglycosides more than 4-fold over that of berberine. In addition, comparison of the MICs of the aminoglycosides in combination with 13-o-MBB towards a clinical strain of P. aeruginosa and its mexXY-deficient strain provided similar MIC values. 13-o-MBB greatly increased the sensitivity to aminoglycoside drugs for P. aeruginosa, B. cepacia, and A. xylosoxidans, increasing the sensitivity to amikacin more than 128-fold and to gentamycin more than 512-fold for A. xylosoxidans.

Interaction between 13-o-MBB and Aminoglycoside Drugs
The fractional inhibitory concentration (FIC) values were determined using 13-o-MBB or berberine and gentamicin or amikacin in combination with P. aeruginosa strains PAGU 1606 and PAGU g 1927 and their MexXY-defective mutants PAGU g 1659 and PAGU g 1931 ( Table 6)

Time-Killing Assay
The bactericidal activity of gentamicin together with berberine and 13-o-MBB against P. aeruginosa was investigated using PAGU g 1933 and PAGU g 1929. PAGU g 1933 was killed after 4 h treatment with gentamicin at 2 µg/mL whereas the growth of PAGU g 1929 was suppressed but no bactericidal action was observed (Figure 2). Treatment of PAGU g 1929 for 4 h with 2 µg/mL gentamicin in combination with 256 µg/mL berberine reduced the number of colonies about 100-fold. In addition, treatment of PAGU g 1929 with a combination of 2 µg/mL gentamicin and 64 µg/mL 13-o-MBB enhanced the bactericidal action of gentamicin more than 10-fold over that of 256 µg/mL berberine.

Interaction between 13-o-MBB and Aminoglycoside Drugs
The fractional inhibitory concentration (FIC) values were determined using 13-o-MBB or berberine and gentamicin or amikacin in combination with P. aeruginosa strains PAGU 1606 and PAGU g 1927 and their MexXY-defective mutants PAGU g 1659 and PAGU g 1931 ( Table 6)

Time-Killing Assay
The bactericidal activity of gentamicin together with berberine and 13-o-MBB against P. aeruginosa was investigated using PAGU g 1933 and PAGU g 1929. PAGU g 1933 was killed after 4 h treatment with gentamicin at 2 µg/mL whereas the growth of PAGU g 1929 was suppressed but no bactericidal action was observed (Figure 2). Treatment of PAGU g 1929 for 4 h with 2 µg/mL gentamicin in combination with 256 µg/mL berberine reduced the number of colonies about 100-fold. In addition, treatment of PAGU g 1929 with a combination of 2 µg/mL gentamicin and 64 µg/mL 13-o-MBB enhanced the bactericidal action of gentamicin more than 10-fold over that of 256 µg/mL berberine. (a)

Discussion
The addition of 128 µg/mL 13-o-MBB increased the sensitivity to aminoglycosides by 2-fold to 8-fold in comparison with 256 µg/mL berberine in the MexXY-positive P. aeruginosa strain PAGU g 1927 (Tables 2 and 3). The antimicrobial activity of 13-o-MBB was not significantly different from that of the 13-o-MBB regioisomers 13-(3-methylbenzyl) berberine bromide and 13-(4methylbenzyl) berberine bromide, although the drug resistance inhibitory action of 13-o-MBB on the MexXY system is greater than that of these two regioisomers. This indicates that 13-o-MBB has greater inhibitory action against MexXY-dependent drug resistance than berberine and the other berberine derivatives we synthesized.
The deletion of mexXY from PAGU 1606 strain generated the PAGU g 1659 strain. The addition of 13-o-MBB increased PAGU g 1659 resistance towards norfloxacin, erythromycin and azithromycin 2fold to 4-fold. Norfloxacin, erythromycin and azithromycin are substrates for MexCD-OprM and increased resistance towards norfloxacin, erythromycin and azithromycin may be due to the induction of MexCD-OprJ [24].
The addition of 13-o-MBB 256 µg/mL increased the efficacies of azithromycin and gentamicin to a Clinical and Laboratory Standards Institute (CLSI) breakpoint (amikacin is 64 µg/mL, gentamycin is 16 µg/mL) in a clinical strain of P. aeruginosa highly resistant to aminoglycosides. Amino acid residue Y613 within the loop of the drug binding pocket of MexY is directly involved in the recognition of aminoglycoside drugs, based on a decrease in sensitivity to aminoglycoside drugs upon mutation of Y613 have been reported [25]. Tobramycin and berberine have been reported to compete for Y613 on the docking simulations of tobramycin or berberine on MexY [16]. Furthermore, they claimed that the results of a combined berberine/tobramycin assay on different clinical isolates of P. aeruginosa were consistent with the in silico findings [16]. The results of our combination assay using berberine and 13-o-MBB with aminoglycosides are consistent with this report [16] and substantiate that the main mechanism of action of berberine and 13-o-MBB is competition for MexY inhibition. Another possible mechanism is suppression of MexY expression. However, Berberine decreased MexY mRNA only 0.8 to 0.9-fold have been reported [26]. Another reported that the MIC of amikacin and gentamicin was increased only up to 4-fold even in a strain P. aeruginosa that expresses 10-21 times more MexY mRNA than the PAO1 strain [12]. Our study of inhibited resistance

Discussion
The addition of 128 µg/mL 13-o-MBB increased the sensitivity to aminoglycosides by 2-fold to 8-fold in comparison with 256 µg/mL berberine in the MexXY-positive P. aeruginosa strain PAGU g 1927 (Tables 2 and 3). The antimicrobial activity of 13-o-MBB was not significantly different from that of the 13-o-MBB regioisomers 13-(3-methylbenzyl) berberine bromide and 13-(4-methylbenzyl) berberine bromide, although the drug resistance inhibitory action of 13-o-MBB on the MexXY system is greater than that of these two regioisomers. This indicates that 13-o-MBB has greater inhibitory action against MexXY-dependent drug resistance than berberine and the other berberine derivatives we synthesized.
The deletion of mexXY from PAGU 1606 strain generated the PAGU g 1659 strain. The addition of 13-o-MBB increased PAGU g 1659 resistance towards norfloxacin, erythromycin and azithromycin 2-fold to 4-fold. Norfloxacin, erythromycin and azithromycin are substrates for MexCD-OprM and increased resistance towards norfloxacin, erythromycin and azithromycin may be due to the induction of MexCD-OprJ [24].
The addition of 13-o-MBB 256 µg/mL increased the efficacies of azithromycin and gentamicin to a Clinical and Laboratory Standards Institute (CLSI) breakpoint (amikacin is 64 µg/mL, gentamycin is 16 µg/mL) in a clinical strain of P. aeruginosa highly resistant to aminoglycosides. Amino acid residue Y613 within the loop of the drug binding pocket of MexY is directly involved in the recognition of aminoglycoside drugs, based on a decrease in sensitivity to aminoglycoside drugs upon mutation of Y613 have been reported [25]. Tobramycin and berberine have been reported to compete for Y613 on the docking simulations of tobramycin or berberine on MexY [16]. Furthermore, they claimed that the results of a combined berberine/tobramycin assay on different clinical isolates of P. aeruginosa were consistent with the in silico findings [16]. The results of our combination assay using berberine and 13-o-MBB with aminoglycosides are consistent with this report [16] and substantiate that the main mechanism of action of berberine and 13-o-MBB is competition for MexY inhibition. Another possible mechanism is suppression of MexY expression. However, Berberine decreased MexY mRNA only 0.8 to 0.9-fold have been reported [26]. Another reported that the MIC of amikacin and gentamicin was increased only up to 4-fold even in a strain P. aeruginosa that expresses 10-21 times more MexY mRNA than the PAO1 strain [12]. Our study of inhibited resistance by berberine showed that the gentamicin MIC for PAGU1606 was reduced 4-fold to 16-fold by berberine ( Table 4), suggesting that the inhibition of MexY expression is not the main mechanism of action of berberine and 13-o-MBB.
13-o-MBB showed cytotoxicity against Caco-2 cells, a human epithelial colorectal adenocarcinoma cell line, at 30 µg/mL (data not shown). Thus, a concentration of 256 µg/mL 13-o-MBB could be toxic to human cells. There is thus a need to synthesize a compound that exhibits inhibitory action against MexXY system-dependent drug resistance at a lower concentration than 13-o-MBB and that is non-toxic to human cells.

Bacterial Strains and Growth Conditions
The bacterial strains used in this study are described in Table 7. Bacterial cells were grown in Luria (L) broth and on L agar (1.5%) under aerobic conditions at 37 • C, as previously described [27].

Antibiotic Susceptibility Assay
MICs were assessed in cation-adjusted Mueller-Hinton (MH) broth after about 18-22 h of incubation at 37 • C (for P. aeruginosa) or after about 20-24 h of incubation at 35 • C (for A. xylosoxidans and B. cepacia) using the two-fold serial micro-titer broth dilution method described previously [15]. The categorization as susceptible, intermediate, and resistant was performed according to the interpretive standards of the CLSI.
The FIC index was calculated as described previously [15]. The effects of the drugs were interpreted to be indicative of synergy when the index was ≤0.5.

Time-Killing Assay
We examined the bactericidal activity of gentamicin monotherapy or combination therapy with berberine or berberine derivatives towards PAGU g 1929 and PAGU g 1933. Each measurement was started by inoculating between 5 × 10 6 to 2 × 10 7 CFU/mL in cation-adjusted MH broth and incubating at 150 rpm at 37 • C on a shaker. Samples were withdrawn to measure the survival counts on MH agar plates at 0, 1, 2, 3 and 4 h. The MH agar plates were incubated at 37 • C for 16-18 h. The concentrations of drugs tested were gentamicin 2 µg/mL, berberine 256 µg/mL, and 13-o-MBB 64 µg/mL. The fraction surviving vs. the control for each sample was determined by taking the average CFU/mL values of the treated samples and dividing by the value for the same sample at 0 h. Each experiment was repeated at least three times, and a representative experiment is shown.

13-Benzylberberine Derivatives; General Procedure
Each benzyl bromide (1.0 mmol) was added in a dropwise manner to a stirred solution of KI (310 mg, 1.86 mmol, 1.86 equiv) and dihydroberberine (337 mg, 1.0 mmol, 1 equiv) in CH 3 CN (40 mL), and the resulting mixture was held at reflux for 4 h. The reaction mixture was then filtered, and the filtrate was collected and evaporated to dryness in vacuo to give the crude residue. The residue was purified by column chromatography over neutral alumina using CHCl 3 /CH 3 OH (50:1 to 20:1) as eluent and recrystallization to give the final compounds 1-11. Compounds 1-10 were known compounds and their characterisation data were identical to those given in the literature. Their

Conclusions
Eleven berberine derivatives were synthesized and tested for MexXY-dependent inhibition of gentamicin resistance using a Pseudomonas aeruginosa positive-MexXY strain and a negative-MexXY strain. 13-o-MBB showed the greatest inhibitory effect on MexXY-dependent gentamicin resistance. Regioisomers of 13-o-MBB exhibited no greater MexXY-dependent inhibition of gentamicin resistance than berberine. 13-o-MBB inhibited resistance to aminoglycosides 4-fold to 16-fold compared with berberine against the four tested P. aeruginosa clinical strains, and Achromobacter xylosoxidans and Burkholderia cepacia. These results indicate that 13-o-MBB inhibits the resistance to aminoglycosides in a MexXY-dependent manner more strongly than berberine. 13-o-MBB is thus a useful inhibitor of aminoglycoside drug resistance due to MexXY.