Gold nanoclusters augment a fluoroquinolone lethality against biofilm associated persister cells

Persister cells are an important medical concern, leading to the overuse of antibiotics and ultimately contributing to antimicrobial resistance. The use of an adjuvant that augments the antibiotic efficacy has yet to be explored for combating persister cells within biofilm infections. Here we demonstrate a paradigm shift in targeting bacteria in chronic infections by coadministration of a conventional antibiotic with a novel engineered gold nanocluster (AuNC@CPP). AuNC@CPP was found to reduce the biofilm concentration (MBEC) of ofloxacin up to 300-fold (from > 3000 μg/mL to 10 μg/mL ). Compared to ofloxacin alone (FLOXIN®Otic), coadministration of ofloxacin with AuNC@CPP induces up to a 10,000-fold reduction in bacterial burden in a validated mouse model of chronic P. aeruginosa of ear infection mimicking chronic suppurative otitis media. The biocompatibility of AuNC@CPP encourages efforts for development as an antibiotic adjunct for combating bacterial biofilm infections. The treatments were administrated by oral gavage at a dose of 10 mg/kg (i.e., 1000 mg/mL) daily for 14 days. During the 35-day study, the body weights of animals were measured every two days. At 35 days of post-treatment, mice were sacrificed. Blood and organs were collected. All organs were preserved, fixed in 10 % neutral formalin buffer, processed into paraffin embedding, and stained with hematoxylin and eosin for pathology analysis using a light microscope. The microscopic analysis of tissue (histopathology), including esophagus, stomach, heart, kidney, spleen, pancreas, thymus, small intestine, colon, bladder, testis and ovary were carried out for potential histological alteration. Results of the organ-to-body weight ratio (relative organ weight, ROW) were analyzed as the ratio of the organ weight (mg) to the body weight of the animal at necropsy day (g). The following organs were examined for ROW: thymus, spleen, heart, kidneys, liver and testis. Blood samples were subjected to toxicity analysis. An inferior vena cava blood collection was performed at sacrifice. 150 µl of blood was placed in a K2EDTA tube for hematology analysis and the remaining blood sample was placed in a 1.5 mL Eppendorf tube for serum extraction. Serum was separated by centrifuging the blood to remove the cellular fraction for liver and renal function testing.

Biofilm formation, a tactic used by bacterial pathogens to evade drug treatment and the human immune response, is a growing threat to human health with a total annual cost in the USA at around $94 billion 1, 2 3 . Persister, non metabolic cells that remain after treatment in chronic bacterial infections are responsible for antibiotic tolerance within biofilms and lead to recurrence of infections when the antibiotic is ceased [4][5][6] . Biofilms shield persister cells from the immune system by providing a physical diffusion barrier. Over time, with repeated failed antibiotic attack on persister cells, development of antibiotic resistance occurs 7,8 . Clinically, biofilm-related infections including chronic suppurative otitis media (CSOM), catheter-associated urinary tract infections, gastrointestinal infections, chronic cutaneueos wounds and cystic fibrosis lung disease 9, 10 are often attributable to Pseudomonas aeruginosa (P. aeruginosa) and can be difficult to cure with current antibiotic strategies.
Antibiotic resistance of P. aeruginosa is linked with recurrent or recalcitrant infections and failed eradication 11 . To address this, it has been proposed to target both growing bacteria and non-growing persister cells for more effective treatment of persistent infections 12 . Huge efforts have been taken to use antibiotics in combination with adjuvants targeting important metabolic pathways contributing to drug resistance (i.e., permeablisers, lactamase inhibitors, efflux pump inhibitors, quorum sensing inhibitors, toxin inhibitors etc.) [13][14][15] . These antibiotic adjuvants require an active metabolism to exert their action, thus resulting in effective killing of growing bacteria but not against non-growing persister cells. Therefore, there is need for a new antibiotic adjuvant specifically aimed at eradication of biofilm associated persister cells.
Induction of reactive oxygen species (ROS) production, especially hydroxyl radicals (HO • ) is a common mechanism by which bactericidal antibiotics kill their target cells 16,17 .
Here, we designed gold nanoclusters that can potentiate a standard of care fluoroquinolone (ofloxacin) against biofilm associated persister cells. As the result of this potentiation, the bacterial burden in the middle ear with effusion of a validated mouse model of CSOM was reduced up to 10,000-fold below the level of bacterial colonies obtained upon topical administration of ofloxacin alone.
Ofloxacin fails to eradicate in vitro biofilm from otopathogenic P. aeruginosa.
The advantage of topical therapy is the ability to deliver higher concentration of antibiotics to the treatment site when compared with oral or parenteral antibiotics. Thus, topical antibiotic administration is a therapeutic strategy proposed to treat problematic bacterial pathogens in a biofilm 21 . To challenge this hypothesis, we used two strains of P. aeruginosa, including the antibiotic sensitive wild-type reference strain PA01 and a clinical otopathogenic strain isolated from a CSOM patient (PA CSOM). Biofilms were challenged with a fluoroquinolone as they are commonly used to treat infections caused by P. aeruginosa and the current standard of care in CSOM [22][23][24] , the only available antibiotic effective at killing both growing and nongrowing cells 25 , and they have no restriction to diffuse into P. aeruginosa biofilms 26 . To demonstrate that the diffusion barrier of bacterial biofilm (dense cell cluster and sticky biofilm matrix) plays a minor role in defense against fluoroquinolone attacks, we first compared the biofilm formation capacity of PA CSOM and PA01. We quantify the biofilm formation capacity by the normalized biofilm matrix value (crystal violet stained biofilm absorbance at 595 nm, A595/ total cell growth, optical density at 600 nm, OD600). We found that PA01 is the higher biofilm producer (Fig. 1a). We also found that the minimum inhibitory concentration (MIC) of ofloxacin against PA01 and PA CSOM were 3.3 μg/mL and > 52.8 μg/mL, respectively (supplementary Fig. 1). Resistance to ofloxacin was defined as a MIC ≥ 8 µg/ mL, suggesting that PA CSOM is a fluoroquinolone-resistant strain 27,28 .
After confirming that PA01 produces more biofilm than does PA CSOM, we used these strains to determine the minimum biofilm eradication concentration (MBEC) of FLOXIN®Otic against 48 h old biofilms using the commercially available MBEC Assay® (Innovotech Inc. Edmonton, Canada). This new technology is useful for predicting clinical failure and clinical success of antimicrobial therapy against biofilm bacteria 29.30 In this assay, the MBEC is identified when incubated recovery media has an optical density at 650 nm (OD650) ≤ 0.1 or no regrowth of bacteria when spotted on Luria broth (LB) agar plates (supplementary Fig. 2). Our results show that ofloxacin has a MBEC of 750 μg/mL (about 227 x MIC) against PA01 biofilm ( Fig. 1b). This finding is consistent with that of Masadeh and collegues who reported a MBEC of 640 μg/mL against 24 h old strain P. aeruginosa (ATCC 27853) biofilm 31 . Contrary to the results for PA01, no eradication of the PA CSOM biofilm was obtained after treatment with ofloxacin at 3000 μg/mL, the concentration in the commercial preparation (FLOXIN®Otic). The incubated recovery media from all concentrations tested shown an OD650 > 0.1 (Fig. 1c). As shown in the photograph (Fig. 1d), the treatment of PA01 biofilm with ofloxacin at 750 μg/mL leaves no viable cells compared to bacterial growth in the recovery media of PA CSOM after treated with ofloxacin at 3000 μg/mL. Given that PAO1 produces more biofilm thanPA CSOM, yet has a much higher susceptibility to ofloxacin, we concluded that the diffusion barrier of biofilm likely plays a minor role in the defense against fluoroquinolones. Of note, the maximum concentration of ofloxacin in otorrhea of CSOM pateints with a persistent purulent discharge ranged from 405 to 653 μg/mL at 8 h after topical administration of FLOXIN®Otic 32 . The lack of in vitro succeptibility for FLOXIN®Otic of PA CSOM biofilms suggest that the drug will not eradicate persister cells in ears infected with this clinical otopathogenic strain of P. aeruginosa, leaving the patients at continuing risk of recurrence. Our results highlight that delivering a high concentration of ofloxacin by topical administration does not always lead to eradication of P. aeruginosa biofilms.
The resistance of PA CSOM to ofloxacin does not correlate with high catalase activity.
To establish whether the survival of PA CSOM biofilms was due to tolerance or strains that had acquired high level resistance to 3000 μg/mL of ofloxacin, the planktonic logarithmic-phase culture of surviving persister cells was challenged with 100 μg/mL of ofloxacin (Fig. 2). To be considered effective, the antimicrobial drug needs to kill ≥ 99.9% equivalent to 3 log10 CFU/ml reduction compared to the initial inoculum (no antimicrobial treatment) 33 . When re-grown in the absence of ofloxacin, persister cells of PA CSOM became sensitive to 100 μg/mL of ofloxacinmediated killing (i.e., 3 log10 CFU/ml reduction), providing conclusive proof that the survival cells were not resistant to 3000 μg/mL of ofloxacin.
Fluoroquinolones potentially induce formation of reactive oxygen species (O2 •and H2O2), which damage DNA, lipids and proteins causing cell death 34,35 . O2 •is not very reactive with biomolecules but it does react rapidly with another molecule of O2 •to form H2O2, which is stable and could diffuse through the bacterial cell membrane to form HO • radicals via the Fenton reaction (H2O2 + Fe 2+ → HO • + HO − + Fe 3+ ). To cope with the destructive nature of this oxidative process, catalases are needed in P. aeruginosa persister cells to protect them from fluoroquinolones eradication 36 . Given the protective role of catalases in P. aeruginosa biofilm resistance to H2O2, we hypothesized that strong catalase activity could protect PA CSOM against ofloxacin-induced HO • radicals via the Fenton reaction. Comparing PAO1 and PA CSOM planktonic cells, it can be seen that catalase activity was similar in both P. aeruginosa strains (supplementary Fig. 3). Moreover, 3% H2O2 fails to eradicate planktonic stationary phase cultures of PA CSOM and PA01 (supplementary Fig. 4). It can therefore be concluded that protection against oxidative stress is not the main reason why ofloxacin at 3000 μg/mL fails to kill PA CSOM, but has a significantly high antimicrobial activity against PA01.

Gold nanoclusters augment the lethal action of ofloxacin against biofilms and persister cells
The maximum achievable physiological concentration of ofloxacin in the middle ear mucosa of CSOM patients after topical administration of FLOXIN®Otic is 602 μg/mL 32 . The PA CSOM biofilm was able to survive up to 3000 μg/mL ofloxacin, approximately five times the therapeutically achievable concentration. Because novel antibiotic development takes a decade or longer, a way to effectively use currently available ofloxacin by lowering its MBEC into a range clinically achievable, provides a potentially shorter path to development. We engineered a peptide-gold hybrid nanocluster (AuNC@CPP) that comprises a cell-penetrating peptides (CPP) Ac-YGRKKRRQRRR-(β-Ala)-(β-Ala)-(β-Ala)-Cys-CONH2 and thiolated polyethylene glycol with a carboxyl termination (an efficient protecting ligand that confers good stability to AuNC@CPP in solution as well as in biological systems). UV-Vis spectrum of AuNC@CPP showed a monotonous decrease from UV into the visible but no surface plasmon resonance peak at 520 nm indicative of the formation of ultrasmall particles (core diameter ≤ 2 nm) 37 (supplementary Fig. 5). The absence of surface plasmon resonance peak demonstrates that AuNC@CPP are ultrasmall with a core diameter ≤ 2 nm. AuNC@CPP exhibits an emission peak at 438 nm (emission energy 2.8 eV). According to the correlation number of gold atoms, N, per cluster with emission energy, AuNC@CPP is an Au8 nanoclusters 38 . Estimating the AuNC@CPP diameter (D in Å) using Equation 1 39,40 , revealed that the gold core is approximately 0.63 nm. Characterization of AuNC@CPP by agarose gel electrophoresis shown that AuNC@CPP are negatively charged, so they move toward the positive electrode (supplementary Fig. 5).
AuNC@CPP alone exhibits a MBEC of 1600 μg/mL against 48 h old PA01 biofilms (supplementary Fig. 6). However, both AuNC@CPP and CPP alone were not able to eradicate PA CSOM biofilm at the concentrations tested (supplementary Fig. 7). When combined with AuNC@CPP (800 μg/mL), the MBEC of ofloxacin against PA CSOM was >300-fold lower than the MBEC of ofloxacin on its own (> 3000 μg/mL). As shown in the photograph ( (Table 1). Together, these findings highlight that there is much to be gained with existing drugs by taking advantage of synthetic nanotechnology.

Biocompatibility assessments
We used an industry standard MTT viability assay to test the cytotoxicity of AuNC@CPP against adenocarcinomic human alveolar basal epithelial cells (A549 cells). We found that cells exhibited more than 90% viability upon direct exposure to AuNC@CPP at 3200 μg/mL for 24 h (supplementary Fig. 8). Furthermore, we showed that administration of AuNC@CPP is unlikely to be of concern for systemic toxicity or induction of gastrointestinal illnesses if the entire dose pass the Eustachian tube (i.e., canal that connects the middle ear to the nasopharynx) and is ingested. Indeed, 35 days after administration by oral gavage at a dose of 10 mg/kg (or 1000 μg/mL) daily for 14 days, no statistical significance in body weight loss was seen between placebo control (PBS) and AuNC@CPP or between sexes (supplementary Fig. 9). In addition, there was no noticeable change in fecal form during the observations. AuNC@CPP does not prompt significant change in hematologic,liver and kidney function (supplementary Table 1 and 2). There was no change in almost all organ-to-body weight ratio (Figure 4c-h). The heart-tobody weight ratio was significantly increased in male treated AuNC@CPP group ( Figure 4e). As it did not corroborate the histopathological data, it is not considered toxicity 41 . Light microscopic examination of sections of organs of PBS (control) and treated AuNC@CPP group showed a normal histology and absence of any gross pathological lesions (Fig. 4). Further studies of increasing time exposure and doses are necessary to determine if toxicity occurs at higher doses or if there is a no adverse effects limit.

Reduction of bacterial burden in CSOM mouse model of chronic P. aeruginosa infection
To determine the clinical relevance of this therapy adjuvant, the in vivo efficacy of Floxin®Otic was compared with that of Floxin®Otic plus AuNC@CPP. We have chosen CSOM as an in vivo model of P aeruginosa biofilm-related infections given its significant global burden as the leading cause of hearing loss in children in developing countries, with an incidence of 31 million cases and prevalence of >300 million 42 43, 44 . CSOM currently has no cure with end stage disease resulting in surgery, not available where resources are often limited or non-existent 45 Fig. 11), indicating that the formation of surface-bound HO • radicals on gold nanocluster contributes to the killing of biofilms and associated persister cells. These two observations provides support for the hypothesis that stimulating HO • radicals production could eradicate persister cells 48 . To determine whether the eradication in persister cells was specific to AuNC@CPP or more generally associated to peroxidase-like activity of metal nanoparticles, a comparative study using iron oxide nanoparticles (Fe3O4; 797146 Sigma-Aldrich) with a peroxidase-like function was conducted. Combination of Fe3O4 and ofloxacin was unable to eradicate PA CSOM biofilms (supplementary Fig. 12), confirming that the eradication in persister cells was specific to AuNC@CPP.

Discussion
The fraction of persister cells in biofilms is usually low (~0.01%), and they should be distinguished from the metabolically inactive bacteria, which constitute a large fraction of the biofilm 49 . Nanotechnology approaches for treatment of biofilm-associated infections fall into two categories, antibiotic loaded nano-systems 50 and nano-systems for disassembling bacterial extracellular matrix 51 . The goal of the former strategy is to improve the bioavailability of conventional antibiotics. This study, as well as prior work suggests that increasing the dose of the drug has little effect on persister cells death 52 . This may explain why many antibiotic loaded nano-systems fail to eradicate established biofilms 53 . The latter nanotechnology strategy is derived from the idea that alteration of the biofilm matrix makes persister cells more vulnerable to conventional drugs. In fact, antibiotic treatments often fail to eradicate planktonic persister cells of P. aeruginosa 54,55 . As a pertinent example, the use of PLGA nanoparticles loaded with ciprofloxacin, a fluoroquinolone, was able to eradicate 99.8% of established P. aeruginosa biofilms. However, 0.2% of the cell population remained intact and constitutes a main source of persister cells 56 . Futhermore, dispersed biofilm cells represent a distinct stage in the transition from biofilm to planktonic lifestyles and are highly virulent against macrophages compared with planktonic, non biofilm cells 57 . A reminder of the clinical risk in this approach is that in vivo dispersion of mobile biofilm bacteria has been shown to cause fatal sepsis in the absence of antibiotic therapy in a mouse wound model 58 . If the persister cells, which retain their phenotype for days or weeks after withdrawal from biofilm, are not directly addressed the chronic infection cycle will continue 54,59 .
Synergistic combination of two antimicrobial agents appear particularly attractive in the case of P. aeruginosa biofilms. Disadvantages of combination therapy is that if tolerance has already emerged to one drug, the combination may end up promoting the transmission of resistance to a partner drug 60 . Therefore, tolerance is an important factor to consider in designing combination treatments that prevent the evolution of resistance. The ability of citrate-capped silver nanoparticles (AgNPs) in combination with the aminoglycoside antibiotic tobramycin or aztreonam, to prevent the recovery of PAO1 biofilms in vitro has become an attractive nanotechnology strategy 61,62 . However, emergence of P. aeruginosa resistance to AgNPs motivated the use of ultrasmall AuNC as an antibiotic adjuvant 63 . The antibiotic-adjuvant combination has the advantage of preserving existing antibiotics. The current work advocates that antibiotic-adjuvant combination could be a potential therapeutic option to target persister cells. Several lines of evidence suggest that persister cells accumulate at least some fluoroquinolones-induced damage probably through the contribution of HO • radicals generated by bactericidal antibiotics 64,65 . We reasoned that the fluoroquinolones treatment may be clinically significant against persister cells if a way is found to boost intracellular production of HO • radicals. Under acidic pathological conditions, the surface-bound HO • radicals generated from decomposition of H2O2 on AuNC@CPP mediated persister cell death. We also found that Fe3O4 with a peroxidase-like function could not augment the ofloxacin efficacy aginst P.
aeruginosa biofilms. This data suggests that peroxidase-like activity is an important, but probably not the only factor that increases the effectiveness of ofloxacin when combined with AuNC@CPP. The planktonic P. aeruginosa persister cells were not eradicated upon exposure to 3% H2O2-induced oxidative stress. Paradoxically, several authors have speculated that oxidative stress mediated by induction of ROS generation is the dominant antibacterial mechanism of action for ultrasmall AuNCs 66,67 . Therefore, in this study, a line of research was highlighted for further investigations if the correlation between ROS production and antibacterial capacity is the cause of bacterial cell death. The exact biomolecular mechanisms of how AuNC@CPP interact with bacterial cells have not been studied in detail. We plan on investigating these mechanisms in future studies.
We are aware of several limitations of this study. In particular, the use of unique murine infection models to demonstrate the antibacterial capability in vivo. It will be essential to establish whether AuNC@CPP-ofloxacin combination is more effective than ofloxacin alone in mouse model of chronic P. aeruginosa of lung infection mimicking Cystic Fibrosis 68 . Since gold nanoparticles did not cause ototoxicity 69 , the possibility that AuNC@CPP could induce ototoxic effects was not assessed in the present study and therefore requires further investigation.
In conclusion, we have established a synergistic antimicrobial agent through the combination of AuNC@CPP and a fluoroquinolone. We have demonstrated that the combination of AuNC@CPP and ofloxacin was highly effective in a mouse model of chronic P. aeruginosa of ear infection mimicking CSOM. Taken together, the results demonstrate the opportunity that that ultrasmall AuNC offer for rescuing antibiotics from tolerance in biofilms. The treatments were administrated by oral gavage at a dose of 10 mg/kg (i.e., 1000 mg/mL) daily for 14 days. During the 35-day study, the body weights of animals were measured every two days. At 35 days of post-treatment, mice were sacrificed. Blood and organs were collected. All organs were preserved, fixed in 10 % neutral formalin buffer, processed into paraffin embedding, and stained with hematoxylin and eosin for pathology analysis using a light microscope. The microscopic analysis of tissue (histopathology), including esophagus, stomach, heart, kidney, spleen, pancreas, thymus, small intestine, colon, bladder, testis and ovary were carried out for potential histological alteration. Results of the organ-to-body weight ratio (relative organ weight, ROW) were analyzed as the ratio of the organ weight (mg) to the body weight of the animal at necropsy day (g). The following organs were examined for ROW: thymus, spleen, heart, kidneys, liver and testis. Blood samples were subjected to toxicity analysis. An inferior vena cava blood collection was performed at sacrifice. 150 µl of blood was placed in a K2EDTA tube for hematology analysis and the remaining blood sample was placed in a 1.5 mL Eppendorf tube for serum extraction. Serum was separated by centrifuging the blood to remove the cellular fraction for liver and renal function testing.      Representative histological photomicrograph of organs by H&E staining. There was no obvious morphologic change on the histological structure of tissues after daily oral gavage at a dose of 10 mg/kg daily for 14 days with AuNC@CPP and PBS. Fig. 5 In vivo efficacy in mouse model of chronic P. aeruginosa ear infection mimicking CSOM. Comparison of the number of bacteria per milliliter (CFU/mL) from middle ear effusion 14 days after the end of the following treatments: placebo control (phosphate-buffered saline, PBS), FLOXIN®Otic (24 µg of ofloxacin) and combination (24 µg of ofloxacin + 296 µg of AuNC@CPP). The CFU/mL from each mouse are plotted as individual points and error bars represent the deviation in CFU/mL within an experimental group. Φ indicates that no middle ear effusion was able to be sampled due to technical constraints. There was no difference between the PBS control group and the ofloxacin group. AuNC@CPP plus FLOXIN®Otic combination led to a 5 log reduction in bacteria 14 days after treatment. *p≤0.05 and Not significant (N.S).