The combined antibacterial effects of sodium new houttuyfonate and berberine chloride against growing and persistent methicillin-resistant and vancomycin-intermediate Staphylococcus aureus

Infections caused by drug-resistant Staphylococcus aureus, especially vancomycin-intermediate Staphylococcus aureus (VISA), leave clinicians with limited therapeutic options for treatment. Persister cells is a leading cause of recalcitrant infection and antibiotic treatment failure, and there is no drug in clinical use that specifically targets persister cells currently. Here, we report a promising combination therapy of sodium new houttuyfonate (SNH) and berberine chloride (BBR) which is able to eradicate both growing and persistent drug-resistant Staphylococcus aureus. The susceptibility test showed SNH exhibited anti-MRSA activity with MIC90 at 64 μg/mL, while BBR showed weak anti-MRSA activity with MIC90 at 512 μg/mL. MICs of BBR in combination with 1/2 MIC SNH decreased by 4 to 64 folds compared with MICs of BBR alone. The results of time-killing assays revealed that the combined use of sub-MIC SNH and BBR offered an in vitro synergistic action against growing MRSA (including pathogenic MRSA) and VISA strains. More importantly, the combination of SNH and BBR was able to eradicate VISA Mu50 and pathogenic MRSA persister cells. The synergistic effect is likely related to the interruption of the cell membrane caused by SNH, which is confirmed by scanning electron microscope and membrane potential and permeability analysis. Our study provide a promising clinical curative strategy for combating drug-resistant S. aureus infections, especially for recalcitrant infections caused by persister cells.


Background
Staphylococcus aureus is one of the most common Gram-positive human pathogens that cause both superficial and invasive infections [1,2]. Over the past two decades, methicillin-resistant S. aureus (MRSA) with multidrug resistance has become increasingly prevalent in both health-care and community settings, and glycopeptide antibiotics such as vancomycin have been used as the last line of defense against MRSA infections [3][4][5]. However, with the widespread use of glycopeptide antibiotics, MRSA strains with reduced susceptibility to vancomycin such as vancomycin-intermediate S. aureus (VISA) have also emerged [6,7]. In addition to the development of antibiotic resistance, there is increasing concern on persister cells of S. aureus as they are suspected to be a leading cause of chronic and refractory infections [8,9]. Persister cells are a subpopulation that exhibits transiently high antibiotic tolerance while the majority of the bacterial population are killed [9]. They are often slow-growing or growth-arrested and able to resume growth when the stress is removed, which can lead to recalcitrant infection and antibiotic treatment failure [10][11][12][13]. Due to the burden arising from antibiotic resistance and persister formation in S. aureus, new antibiotics and therapeutic options are urgently needed.
Berberine, an isoquinoline alkaloid originally obtained from plants such as Phellodendri chinensis Cortex and Coptis chinensis Franch and fully chemically synthesized now [14], has been widely used to treat bacterial diarrhea and gastroenteritis for many years in China [15,16]. Several studies have demonstrated the antibacterial activities of berberine alone or as a synergist of antibiotics [17][18][19][20][21][22]. However, the antibacterial activity of berberine used alone is weak, and none of these studies have revealed its ability to inhibit persister cells. This motivates us to investigate on BBR-based combination therapy against drug-resistant and/or persistent S. aureus.
Previously we discovered that sodium new houttuyfonate (SNH) showed synergistic effect against S. aureus when combined with oxacillin or netilmicin [23], which attracted our interest in investigating it in the BBRbased combination therapy. SNH (C 14 H 27 NaO 5 S, MW = 330.4) is a chemically stable derivative of houttuynin which is an active ingredient from a traditional Chinese herb Houttuynia cordata Thunb [24]. Studies have shown that houttuynin exhibits a variety of medicinal properties including activities on anti-inflammatory, anti-viral and antibacterial [25][26][27]. However, the bioactivities of SNH such as the antibacterial effect are poorly investigated.
Here we report a promising synergistic combination of SNH and berberine chloride (BBR) which is likely in use for the treatment of MRSA and VISA infections, especially recalcitrant infections caused by persister cells. In this study, we systematically evaluated the in vitro activity of SNH and BBR, alone and in combination, against MRSA and VISA and for the first time found their strong synergistic effect on both growing and persister cells. The action of SNH with BBR on the cell membrane was examined to investigate the underlying mechanism of synergism.

Results
Antibacterial spectrum of SNH and BBR against grampositive and -negative pathogens The results of antibacterial spectrum assay (Table 1) showed that SNH was active against Gram-positive bacteria including staphylococci and enterococci with MICs ranging from 16 to 64 μg/mL, while BBR exhibited poor antibacterial activity for all tested Gram-positive bacteria with MIC ranging from 8 to > 128 μg/mL. Both SNH and BBR showed little antibacterial activity against the tested Gram-negative bacteria with MIC higher than 128 μg/mL.

MICs of BBR and SNH against S. aureus strains
The MICs of SNH and BBR against 124 strains of MRSA are shown in Table 2 and the cumulative inhibition of the bacteria at different MIC levels of the two compounds was presented in Fig. 1. SNH inhibited all the MRSA strains at concentrations ranging from 16 to 256 μg/mL, with MIC 50 and MIC 90 values at 32 and 64 μg/mL, respectively, which are essentially consistent with previous reports [23]. BBR showed moderate activity against S. aureus and the MIC range of BBR for the tested MRSA strains was 32-512 μg/mL, with MIC 50 and MIC 90 at 128 and 512 μg/mL. The MICs of SNH and BBR for the quality control strain ATCC 29213 were 32 and 128 μg/mL, respectively, all within the expected ranges as reported previously [23,28,29].

MICs of BBR against S. aureus in combination with sub-MICs of SNH
To examine the antibacterial effect of BBR combined with sub-MIC level of SNH, 15 MRSA clinical isolates were randomly selected with three ATCC strains being used as controls. The MICs of BBR alone and in combination with 1/2 MIC of SNH were presented in Table 3 and Fig. 2. MICs of BBR decreased by 4 to 64 folds in combination with 1/2 MIC of SNH when compared with MICs of BBR alone.
Time-killing assay of SNH-BBR combination on growing MRSA and VISA cells Time-killing assays were performed with MRSA ATCC33591, ATCC43300, VISA Mu50 and three pathogenic MRSA clinical isolates CCPM(A)-P-0116173, CCPM(A)-P-010850 and CCPM(A)-P-011012. As shown in Fig. 3, SNH alone inhibited the growth of the tested bacteria within 8-24 h despite the subsequent regrowth of the bacteria, while BBR alone had little effect on bacterial growth. In contrast, the combination of SNH and BBR resulted in significant synergistic effect on all the tested strains and exhibited a typical bactericidal mode of action. The SNH-BBR combinations eradicated the viable count by ≥4log10 CFU/mL when compared with the starting inoculum or with any of the most active agent used alone, with the viable counts below the detection limit (10 CFU/mL) for ATCC33591, ATCC43300, CCPM(A)-P-0116173, CCPM(A)-P-010850 and CCPM(A)-P-011012. The synergistic effect was observed as early as 2 h after SNH-BBR treatment and lasted for 72 h.
The combination of SNH and BBR eradicated VISA and pathogenic MRSA persister cells Keren et al. has shown that almost all stationary phase S. aureus cells are persistent and exhibited tolerance to antibiotics [30]. To validate that the stationary phase cells used in our study are persistent, MIC determination and killing assay of ciprofloxacin and linezolid against strains used in the persister assays were performed. The stationary phase cells of ciprofloxacin-susceptible MRSA strain ATCC33591 exhibited tolerance to ciprofloxacin (10 μg/mL) (Table S1, Figure S1A). The stationary phase cells of all the strains were tolerant to linezolid ( Figure S1B) although they are susceptible to linezolid based on the MIC values.
To evaluate the effectiveness of SNH and BBR against persister cells, ATCC33591 (MRSA), Mu50 (VISA) and three pathogenic MRSA CCPM(A)-P-0116173, CCPM(A)-P-010850 and CCPM(A)-P-011012 were cultured with shaking for 24 h to reach stationary phase. As shown in Fig. 4, compared with the untreated control, treatment with SNH or BBR alone significantly decreased the number of persistent S. aureus, with a greater reduction of persister cells in the BBR treatment group (except for CCPM(A)-P-010850). When compared with any agent used alone, the SNH-BBR combination reduced the number of S. aureus persisters by at least 3log10, and persister cells of ATCC33591 and pathogenic MRSA CCPM(A)-P-0116173 and CCPM(A)-P-011012 were totally eradicated by the combination (Fig. 4a, c and e).

Effects of the SNH and BBR on cellular morphology
The structure of SNH contains a hydrophilic sulfinyl head and a hydrophobic alkyl tail with 12 carbon atoms, indicating amphiphilic properties [23]. Our hypothesis for the synergistic effect between SNH and BBR is that SNH may disrupt the cell membrane and facilitated the entrance of BBR into the bacterial cell. To address this, we examined the cell morphology of S. aureus  ATCC33591 treated with sub-MIC concentration of SNH or BBR by scanning electron microscope (Fig. 5). Cells of ATCC33591 or BBR-treated ATCC33591 exhibited a relatively "clean" surface. However, bacterial cells treated with SNH or SNH-BBR combination showed a "rough" cell surface with granules on it, which may be the result of leakage of cell contents.

SNH increased the permeability and decreased the membrane potential
If SNH acts as a surfactant and causes damage on S. aureus cell membrane, changes in the permeability and membrane potential of the bacteria should be observed after treatment with the compound. To determine the effect of SNH and BBR on the membrane potential and   The red/green fluorescence ratio of DiOC 2 (3) provides a cell size-independent measure of bacterial membrane potential. TO-PRO-3 is believed to be membrane impermeant and the far-red fluorescence of the dye (~695 nm) can be detected with increased cell permeability. CCCP reduces cell membrane potential to zero but does not affect permeability, while Nisin reduces membrane potential and increases membrane permeability as well.
They were selected as a positive control for cell membrane potential and permeability detection, respectively. The concentrations used for SNH (1/4 MIC) and BBR (1/8 MIC) were based on the time-killing assay. As shown in Fig. 6, sub-MIC of SNH reduced the membrane potential (Fig. 6a) and increased the membrane permeability of the bacterial cell (Fig. 6b) to the same extent as the positive controls (CCCP and nisin, respectively). However, BBR had no effect on the membrane potential and permeability of the bacteria. The cells treated with sub-MIC BBR exhibited the same properties as the untreated cells. The combination of  sub-MIC SNH and BBR led to a decreased membrane potential and an increased membrane permeability of the MRSA strain as the same as SNH alone. The results firmly supported our hypothesis that SNH, not BBR, interrupts the cell membrane integrity.

Discussion
The current limited extent of new drug development is being massively outpaced by emerging antibiotic resistance, pushing humans to the edge of a "post-antibiotic" age. Despite many efforts on drug discovery, very limited new classes of antibiotics have been developed for clinical uses during the past decades. The growing need to combat antimicrobial resistance calls for developing new effective strategies. Combinations of antibiotics or antimicrobial agents offer a productive strategy to deal with the global threat of increasing resistance [31].
Berberine has been part of traditional Chinese medicine for thousands of years with extensive bioactivities such as antimicrobial activities [32]. In this study, the combination of BBR and SNH showed a synergistic effect against antibiotic-resistant S. aureus. The in vitro antibacterial activities of the two compounds were evaluated against 121 clinical MRSA isolates, as well as ATCC strains. To maximize the relevance of our research to clinical situations, the clinical isolates covered almost all types of clinical infections, including samples from nasal colonization, superficial skin and soft tissue infection, and blood infection (Table S1). BBR showed poor antibacterial activity against S. aureus with MIC 90 at 512 μg/ mL, which is consistent with previous studies and might be attributed to the strong hydrophilicity and low permeability through cell membrane due to the quaternary ammonium group of BBR [18,19]. Therefore, the combined use of compounds that increase membrane permeability (such as SNH) with BBR may increase the amount of the latter into bacterial cells and promote antibacterial activity. As predicted, the combination of sub-MIC SNH and BBR in this study eradicated growing MRSA cells to the limit of detection as shown in the time-killing assay. Notably, the combination therapy is not only effective on MRSA strains, but on VISA strain Mu50. This is of great significance since there are few options for clinicians on treating VISA infections.
It is well recognized now that recalcitrant infections are related to the survival of dormant persister cells. However, there is no drug in clinical use that specifically targets persister cells currently. In this study, we found that sub-MIC levels of SNH or BBR can significantly reduce the number of S. aureus persister cells and the combination of the two compounds effectively eradicated the persisters including MRSA and VISA strains (Fig. 4). Surprisingly, the ability of killing pathogenic MRSA persister cells was exhibited at sub-MIC of SNH and BBR ( Figure S2). Treatment with 1/2 MIC of SNH in combination with 1/2 MIC BBR decreased 3log10 of CFU/mL of pathogenic S. aureus persister cells compared with any other groups. Kim et al. [33] demonstrated that antimicrobial agents inducing rapid permeabilization of cell membrane should be effective against MRSA persisters. In our study, we found that SNH decreased the membrane potential to the same extent of CCCP, and increased cell permeability considerably (Fig. 6). Considering that membrane-active agents may exhibit toxicity in mammalian cells [34], hemolytic activities of SNH and BBR were tested against human Fig. 6 SNH led to a decreased potential and an increased permeability of cell membrane in S. aureus ATCC33591. Bacterial membrane potential (a) was represented by the ratio of red/green fluorescence of DiOC 2 (3). CCCP (5 μM) was used as the positive control in the DiOC 2 (3)-based membrane potential assay. Bacterial membrane permeability (b) was measured by TO-PRO-3. Nisin (25 μg/mL, green) was used as a positive control in the TO-PRO-3-based membrane permeability assay. Untreated ATCC33591 cells were regarded as a negative control (black) in both assays. Cells treated with 1/4 MIC SNH, 1/8 MIC BBR and the SNH-BBR combination were presented as blue, orange and red, respectively. Asterisks denote statistical significance as determined by one-way ANOVA followed by Tukey's multiple-comparison analysis ****, P < 0.0001 erythrocytes ( Figure S3). The hemolysis assay showed that SNH caused approximately 6% hemolysis at MIC 90 and did not cause hemolysis at MIC 50 , and BBR had no hemolytic effect.
In view of the morphological observations and the analysis of membrane potential and permeability in S. aureus in the present study, the synergistic effect between BBR and SNH could be caused by SNH interrupting the cell membrane thus promoting BBR into microbial cells. Previous studies proposed a strategy of the combination of BBR with multidrug-resistance pump inhibitors against S. aureus [35,36]. Our work raises the possibility that combination of BBR with antimicrobial agents that targeting the cell membrane is also an effective strategy against drug-resistant S. aureus. Notably, the concentration of BBR used in the combination therapy is relatively high compared with commonly used antibiotics, which might limit its application on systematic administration. However, reduction of concentration of BBR can be achieved by structural modification, and several berberine derivates with better antibacterial activity have been reported recently [37,38]. Moreover, several studies had put efforts into developing transdermal products of BBR to deliver the drug safely and efficiently [39,40]. Besides, BBR has been used topically for the treatment of burns, acne and periodontal inflammation in clinical trials [41][42][43]. Considering that S. aureus is the leading cause of skin and soft tissue infections and topical creams are more acceptable for patients with skin infections, the topical products of SNH-BBR combination are of great clinical significance in the future. Further studies will focus on determining the in vivo efficacy of this combination therapy in rodent models and screening more antimicrobial agents targeting bacterial cell membrane to expand the list of BBR-based combination therapy.

Conclusions
A small portion of microbial persisters can cause recurrent or intractable infections, eventually increasing resistance to antibiotics. Infections caused by drugresistant S. aureus, especially VISA, leave clinicians with few therapeutic options for treatment. In the present study, we demonstrated the synergistic antibacterial effect of SNH and BBR in killing growing and persistent MRSA and VISA cells. Our study also provides a promising combination strategy for combating drug-resistant S. aureus infections, especially for recalcitrant infections caused by persister cells.

Bacterial strains
A total of 121 MRSA strains from hospitals in China between 2008 and 2017 were employed in this study. All bacterial strains used throughout this study were obtained from the Collection Center of Pathogen Microorganism of Chinese Academy of Medical Sciences (CAMS-CCPM) in China. All clinical isolates identified as MRSA were confirmed with the VITEK 2-COMPACT system (bioMerieux, Marcy I'Etoile, France), the standard oxacillin agar dilution method recommended by the Clinical and Laboratory Standards Institute (CLSI) [44] and detection of mecA gene by PCR (primers are Forward, 5′-TCCAGATTACAACTTCACCAGG-3′, and Reverse, 5′-CCACTTCATATCTTGTAACG-3′). Relevant characteristics of the clinical isolates of MRSA utilized in this study are presented in Table S1. S. aureus strains from the American Type Culture Collection (ATCC), ATCC29213 (MSSA), ATCC33591 (MRSA), ATCC43300 (MRSA) and/or Mu50 (ATCC700699, VISA) were used as quality control strains in this study according to the needs of specific assays. All isolates were stored at − 80°C and streaked on tryptic soy agar (TSA) plates to make overnight cultures.

Antimicrobial agents
Sodium new houttuyfonate (SNH) was provided by Beijing Standard Herbs Medical Science & Technology Development Co. LTD (Beijing, China). Berberine chloride (BBR) was commercially purchased from Haoyang Biotech Co. LTD. (Shaanxi, China). SNH was dissolved in warmed distilled water as described previously [23]. BBR was dissolved in DMSO as a stock solution at a concentration of 10.24 mg/mL and stored at − 20°C after sterilizing through 0.22 μm filters.

Susceptibility test
The antibacterial spectrum of SNH and BBR was tested with the agar dilution method according to CLSI [44]. Basically, serial two-fold dilutions of SNH or BBR ranging from 128 to 0.03 μg/mL were made in 15 mL sterile Mueller-Hinton (MH) agar. All isolates including Gram-positive and -negative bacterial strains were tested at a final inoculum of 10 4 CFU per spot on MH agar with serial dilutions of SNH or BBR using a multipoint inoculator (Denley Instruments) and incubated at 35°C for 24 h. The minimum concentration with no bacterial growth was considered to be the MIC of the compound.
The MICs of SNH and BBR against S. aureus were further determined by the broth microdilution method according to CLSI [44]. Briefly, 100 μL of serially diluted compounds (starting concentrations were 1024 μg/mL) in cation-adjusted MH (CAMH) broth were added to wells of 96-well plates. Ten microliter of 5 × 10 6 CFU/ mL log phase bacterial culture was added to each well. The plates were incubated at 35°C for 24 h and the MIC values were determined by the minimal concentration of wells that with the absence of visual turbidity as read with the naked eye.
Combination antimicrobial susceptibility test was performed based on MIC assay with only small modification. Briefly, serial-diluted BBR in combination with SNH at a final concentration of 1/2 MIC was filled in the wells of a 96-well plate and the bacteria were inoculated at a final turbidity of 5 × 10 5 CFU/mL in each well. After incubation of the plate at 35°C for 24 h, the MIC values of BBR combined with 1/2 MIC SNH were read.
All the MIC values (single or in combination) were determined in triplicate on different days.

Time-killing assays
Kinetics of bactericidal activity of SNH and BBR alone and in combination were determined by time-killing assay according to the protocol published previously [23]. The experiment was performed against ATCC33591, ATCC43300, Mu50 and three randomly selected MRSA clinical isolate CCPM(A)-P-0116173, CCPM(A)-P-010850 and CCPM(A)-P-011012. Briefly, overnight cultures were diluted 1:1000 in fresh medium and grown to a turbidity of OD 600nm = 0.8, then diluted to 10 6 CFU/mL with CAMH broth in a 250 mL sterile flask. Sub-MIC concentrations of SNH and BBR alone or in combination were added. At specified time intervals (0, 2, 4, 6, 8, 24, 48 and 72 h), 100 μL aliquots were serially diluted with 10-fold in 0.9% saline, plated on TSA plates in triplicates and incubated at 35°C for 24 h. Then the viable colonies were counted after the incubation. The combination of SNH and BBR was considered synergistic if the bacterial killing showed a ≥ 2log10 decrease in colony count as compared to the most active monotherapy.
Persister assay MRSA ATCC33591, VISA Mu50 and three pathogenic MRSA isolates CCPM(A)-P-0116173, CCPM(A)-P-010850 and CCPM(A)-P-011012 were tested for killing of persisters by antimicrobials. Previous work has shown that almost all stationary phase S. aureus cells are tolerant to conventional antibiotics and persistently behaving, thus being used in persister assay [30,45]. To validate the persister cells used in our study are tolerant to antibiotics, an overnight culture of S. aureus strains was diluted 1:1000 and cultured at 35°C, 220 rpm for 3 h (ATCC33591) or 24 h (all the strains) to reach log phase or stationary phase respectively. One milliliter of log phase or stationary phase cells were challenged with 10 μg/mL ciprofloxacin (for ATCC33591) or 20 μg/mL linezolid (all the strains) for 3 h at 35°C. Before and after the antibiotic challenge a sample was serial diluted and viable cells were counted.
Colonies from a fresh plate were inoculated into 10 mL of tryptic soy broth (TSB) and grown at 35°C, 220 rpm for 24 h to form a persister phenotype in stationary phase. The starting inoculum was about 10 9 CFU/mL for each strain. Appropriate concentrations of SNH and BBR, alone or in combination, were added to each tube with S. aureus persister cells. At specified time points, 10 μL aliquots of samples were removed from the tubes, serially diluted and spot-plated onto TSA plates to determine CFU/mL.

Scanning electron microscopy
An overnight culture of S. aureus ATCC33591 was diluted 1:1000 in TSB and grown to log phase (OD 600nm = 0.8) at 35°C. Bacterial cells were then incubated with or without sub-MIC level of SNH, BBR or their combination for 4 h, and collected and fixed with 2.5% glutaraldehyde overnight. The fixed bacterial cells were washed and resuspended in phosphate buffer. The samples were dehydrated, gold-coated and observed by scanning electron microscope (Hitachi SU8020, Japan).

Bacterial membrane potential and permeability determinations
The membrane potential and permeability of bacteria were determined using dyes of diethyloxacarbocyanine iodide (DiOC 2 (3)) and TO-PRO-3, respectively, as previously reported [45,46]. Overnight culture of S. aureus ATCC33591 was diluted to approximately 10 6 CFU/mL and treated with sub-MIC of SNH, BBR or their combination for 4 h at 37°C. Parallelly, carbonyl cyanide 3-chlorophenylhydrazone (CCCP, 5 μM) treated cells were used as a positive control to depolarize bacterial membrane potential by eradicating the proton gradient, while untreated cells were set as the negative control. Nisin (25 μg/mL) reduces membrane potential and renders bacterial cells permeable to TO-PRO-3, and was selected as the positive control for detecting cell permeability.
The cells were then stained with 30 μM DiOC 2 (3) and 100 nM TO-PRO-3 (Molecular Probes, USA) for 15 min at room temperature. After centrifugation, washing and resuspension, 1 mL of stained cells were pipetted for analysis on a BD FACSCalibur flow cytometer and 100, 000 counts were recorded. DiOC 2 (3) was excited at 488 nm, and its green and red fluorescence were detected through a 525-and 610-nm bandpass filter, respectively. Red fluorescence is emitted from DiOC 2 (3) aggregates, which depends on the size and membrane potential of bacteria cells, while green fluorescence generated from a single dye molecule varies with the size of cells or clumps, and is largely independent of bacterial membrane potential. Thus, the red/green fluorescence ratio provides a cell size-independent measure of bacterial membrane potential. TO-PRO-3 was excited at 630 nm, and its far-red fluorescence was detected through a 695nm long-pass emission filter. TO-PRO-3 with positive charges can enter and stain nucleic acids to produce substantially increased fluorescence in bacteria with damaged membranes, but excluded by bacteria with intact cell membranes. The dye is therefore used to indicate the permeability of bacterial cell membranes.

Statistical analysis
Statistical analysis was performed with GraphPad Prism 8 software. P-values were calculated by one-way ANOVA followed by Tukey's multiple comparisons test to analyze the differences between each pair of groups. P-values of < 0.05 were considered significant. Each experiment was performed at least 3 times and presented as mean ± SD.
Additional file 1: Figure S1. Stationary phase cells of S. aureus were tolerant to antibiotics, Figure S2. MRSA persisters were killed by sub-MIC level of SNH-BBR combination, Figure S3. Hemolytic activity of SNH and BBR against human erythrocytes, Table S1. MIC of ciprofloxacin and linezolid against S. aureus strains used in persister assay, Table S2. Information of clinical strains used in this study.