In vitro and in vivo evaluation of diethyldithiocarbamate with copper ions and its liposomal formulation for the treatment of Staphylococcus aureus and Staphylococcus epidermidis biofilms

Surgical site infections (SSIs) are mainly caused by Staphylococcus aureus (S. aureus) and Staphylococcus epidermidis (S. epidermidis) biofilms. Biofilms are aggregates of bacteria embedded in a self-produced matrix that offers protection against antibiotics and promotes the spread of antibiotic-resistance in bacteria. Consequently, antibiotic treatment frequently fails, resulting in the need for alternative therapies. The present study describes the in vitro efficacy of the Cu(DDC)2 complex (2:1 M ratio of diethyldithiocarbamate (DDC−) and Cu2+) with additional Cu2+ against S. aureus and S. epidermidis biofilms in models mimicking SSIs and in vitro antibacterial activity of a liposomal Cu(DDC)2 + Cu2+ formulation. The in vitro activity on S. aureus and S. epidermidis biofilms grown on two hernia mesh materials and in a wound model was determined by colony forming unit (CFU) counting. Cu2+-liposomes and Cu(DDC)2-liposomes were prepared, and their antibacterial activity was assessed in vitro using the alamarBlue assay and CFU counting and in vivo using a Galleria mellonella infection model. The combination of 35 μM DDC− and 128 μM Cu2+ inhibited S. aureus and S. epidermidis biofilms on meshes and in a wound infection model. Cu(DDC)2-liposomes + free Cu2+ displayed similar antibiofilm activity to free Cu(DDC)2 + Cu2+, and significantly increased the survival of S. epidermidis-infected larvae. Whilst Cu(DDC)2 + Cu2+ showed substantial antibiofilm activity in vitro against clinically relevant biofilms, its application in mammalian in vivo models is limited by solubility. The liposomal Cu(DDC)2 + Cu2+ formulation showed antibiofilm activity in vitro and antibacterial activity and low toxicity in G. mellonella, making it a suitable water-soluble formulation for future application on infected wounds in animal trials.


Introduction
Surgical site infections (SSI) are amongst the most common surgeryassociated infections and occur in 1.5-20% of surgeries, depending on the nature of the surgery and country in which it is performed [1]. SSIs develop at the organ/tissue site of surgery [2] and can range from wound or implant infections to organ infections [3]. Following a surgical procedure, such as hernia mesh repair [4], infections can affect the incision site (from superficial to deep tissue), implanted material and any part of the anatomy that was exposed or manipulated during surgery [5][6][7]. Consequently, SSIs represent a significant burden, by increasing patient morbidity and mortality, and adding additional cost to health systems [2,3,5].
The most common pathogens associated with SSIs are Staphylococcus aureus (S. aureus) and coagulase negative staphylococci, including Staphylococcus epidermidis (S. epidermidis), which are natural components of the respiratory tract and skin microbiota, respectively [8]. Therefore, prevention of SSIs requires pre-operative preparations of the surgical site and antibiotic prophylaxis [2]. If an infection is detected, the routine treatment relies on additional antibiotic therapy [9,10]. However, over the last two decades, the antibiotic missuse and overuse has promoted the emergence of resistant strains, such as methicillin resistant S. aureus (MRSA). The situation is exacerbated by biofilm infections, which are frequently staphylococcal, that offer antibiotic tolerance [11,12]. Biofilms are aggregates of bacteria embedded in a protective matrix, which enables bacteria to persist in hostile conditions, communicate with each other and become highly tolerant to antibiotics [13]. In comparison to planktonic forms of bacteria, biofilm bacteria require 10 to 1000-fold higher concentrations of antibiotics to be eradicated [14]. This is a major concern, as biofilms are present in over 80% of SSIs and are a major cause of delayed wound healing [10]. In addition, patient mortality is increased by 2 to 11-fold in MRSA-associated SSIs, compared to susceptible S. aureus associated SSIs and surgeries without infections [15]. Therefore, there is an unmet need for new antimicrobial agents targeting MRSA and S. epidermidis biofilms to prevent and treat SSIs.
However, the antibacterial activity of 35 μM DDC − and 128 μM Cu 2+ was only observed on biofilms grown in a microtiter well plate over 24 h [31] and can alter when exposed to biofilms grown over multiple days or in conditions similar to SSIs [32]. In addition, the Cu(DDC) 2 complex is insoluble (<0.1 mg/ml) in water, limiting its practicality in the clinical setting [33]. This necessitates the development of a pharmaceutical formulation for optimal drug delivery to infection sites and improved antibacterial efficacy. To improve the solubility of Cu(DDC) 2 , nanoparticles including liposomal formulations of Cu(DDC) 2 have been developed and successfully used as therapeutically active agents against cancer cells [33][34][35][36][37], with enhanced activity against breast cancer cells [38], glioblastoma [39] and neuroblastoma cells [40].
Inspired by this, our aim was to evaluate the antibacterial properties

Biofilm formation on hernia meshes
Round coupons (1.5 cm diameter) of polyester and polypropylene meshes were placed in a 12-well plate and immersed in 2 ml of a bacterial suspension (2 × 10 6 CFU/ml) of S. aureus ATCC 6538, MRSA Mu50 or S. epidermidis ATCC 35984 in TSB and incubated at 37 • C on a rotating platform at 70 rpm (3D Gyratory Mixer; Ratek Instruments, Boronia, Australia). After 24 h incubation, meshes with attached bacteria were washed by immersing the meshes into 3 ml 0.9% (w/v) saline for 30 s at room temperature, three times consecutively, and placed into fresh TSB. Following another 72 h incubation, the meshes were washed, as previously described with 0.9% saline, to remove loosely attached cells and placed into TSB solutions containing 35 μM DDC − + 128 μM Cu 2+ .
Control wells contained TSB alone (untreated control). Following 24 h treatment incubation at 37 • C on a rotating platform (70 rpm), a third washing step was performed prior to CFU counting or imaging of the coupons.
For CFU counting, meshes were collected in 10 ml 0.9% saline and biofilms were extracted from the mesh and disrupted by a series of vortexing (5 min, maximum speed, VM1 Vortex Mixer, Ratek Instruments Pty Ltd, Victoria, Australia) and sonication (15 min, Soniclean 80TD, Pulse swept power 60 W, Soniclean Pty Ltd, South Australia, Australia), prior to serial dilution and plating on TSA. CFU were counted following 24 h incubation at 37 • C. For imaging, the last washing step was performed with phosphate buffered saline. Meshes were covered and incubated with a 1:500 dilution of LIVE/DEAD BacLight staining (1:1 mix of SYTO 9/propidium iodide; Life Technologies, Scoresby, Australia) in TSB for 20 min in the dark and imaged using the Olympus FV1000 Live cell imaging system (Olympus, Shinjuku, Japan) and a 20 × /0.5 W objective. Quantitation of live/dead cells was performed using ImageJ software (NIH, Bethesda, MA, USA). Briefly, the contrast/ brightness was adjusted globally to images to minimize background before setting a threshold to highlight cells for automated counting.

In vitro wound model
An artificial dermis made of collagen (Corning, NY, USA) and hyaluronic acid (1.20-1.80 MDa; Lifecore Biomedical, MN, USA) was prepared as previously described by Brackman et al. [41]. According to established protocols [42], freeze-dried bovine plasma was rehydrated in 10 ml 0.9% saline, 19 ml Bolton broth (LabM, Lancashire, UK), 1 ml freeze-thaw laked horse blood (Biotrading, Mijdrecht, Netherlands) and 20 μl heparin 100 IU. An artificial dermis was placed in each well of a 24-well plate and soaked with 1 ml of this mixture. Then, an overnight culture of MRSA Mu50 or S. epidermidis ATCC 35984 in TSB adjusted to an OD 600 0.1, was diluted 1:100 in 0.9% saline, 10 μl were added on top of each dermis (equal to 10 4 CFU/well) and incubated statically at 37 • C for 24 h. Following biofilm formation, 1 ml of 35 μM DDC − + 128 μM Cu 2+ in TSB was added. Controls included biofilms exposed to TSB (untreated control). After 24 h treatment exposure, each dermis was placed in 10 ml of 0.9% saline, and biofilms were extracted from the dermis and disrupted by three consecutive vortexing and sonication cycles for 30 s each. After serial dilution, plating on TSA and incubation at 37 • C for 24 h, CFU were counted to determine antibiofilm activity.
Cu 2+ -liposomes and Cu(DDC) 2 -liposomes were stored at 4-6 • C for up to 3 months and were sterile filtered under aseptic conditions through a 0.2 μm cellulose acetate filter (VWR International) before use.
As previously described by Hartwig et al. [40], the hydrodynamic diameter (d h ) and the polydispersity index (PDI) were measured via dynamic light scattering (ZetaPals, Brookhaven Instruments Corporation, Holtsville, NY, USA) and encapsulated Cu 2+ concentrations were determined by measuring absorbance of complexed Cu 2+ with DDC − in methanol at a wavelength of λ max = 435 nm with a GENESYS 10S UV-Vis spectrophotometer (Thermo Fisher Scientific). Liposomes were used in biofilm challenge experiments to provide the equivalent of 35 μM DDC − and/or 128 μM Cu 2+ .

Antibacterial activity of liposomes
Overnight cultures of MRSA Mu50 and S. epidermidis 35984 in nutrient broth were adjusted to an OD 600 0.5 and further 1:15 (v/v) diluted in nutrient broth. Black-walled 96-well microtiter plates (Greiner Bio-one, Frickenhausen, Germany) were inoculated with 100 μl bacterial suspension and incubated for 24 h at 37 • C on a rotating platform at 70 rpm. The biofilm was rinsed with 0.9% saline, exposed to tions. The treatments were removed, and the biofilm rinsed with 0.9% saline, before viability was detected by either measurement of metabolic activity with the alamarBlue assay or CFU counting.
The alamarBlue assay was performed according to Richter et al. [43] and rinsed biofilms were incubated with a 10% (v/v) alamarBlue™ Cell Viability Reagent (Thermo Fisher Scientific) solution in nutrient broth.
The fluorescence was measured hourly on a TECAN Spark plate reader (Männedorf, Switzerland) at λ excitation = 530 nm/λ emission = 590 nm until maximum fluorescence was reached, then viability was calculated using Equation (1). Antibiofilm activity of the different treatments was determined as percentage of biofilm viability, where the fluorescence intensity of treated and untreated biofilms is represented by I treated and I untreated , respectively and I blank represents the background fluorescence of the 10% v/v alamarBlue solution [43].
CFU counting was performed according to Van den Driessche et al. [44] and 100 μl of 0.9% saline were added to each rinsed biofilm. To disrupt the biofilm, the plates were shaken at 150 rpm and sonicated (5 min each), and the content of each well was collected separately. This process was repeated twice to extract all biofilms cells and serial dilutions of these suspensions were plated on TSA and incubated at 37 • C for 24 h, prior to CFU counting.

In vivo cytotoxicity and antibacterial activity
Galleria mellonella (G. mellonella) larvae (Angel-Zentrum, Freiburg, Germany) were used on the day of receipt and 30 larvae were assigned to each treatment group. Larvae were injected in the last left proleg with micro-fine (30 gauge) needle insulin syringes (BD, Franklin Lakes, NJ, USA). Four control groups were included, (i) not-injected larvae (uninfected, untreated control), (ii) larvae injected with 0.9% saline (uninfected, vehicle control), (iii) larvae injected with treatment (uninfected, treated control to determine toxicity) and (iv) larvae injected with a bacterial suspension and 0.9% saline (infected, vehicle control). To determine treatment efficacy, larvae were injected with a S. epidermidis ATCC 35984 suspension (OD 600 0.05) in nutrient broth and with Cu (DDC) 2 -liposomes, Cu 2+ -liposomes, [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes] or [Cu(DDC) 2 -liposomes + free Cu 2+ ]. Considering the dilution factor within the larvae, the concentrations of the liposomal formulations were increased 10-fold compared to the concentrations used in vitro. A total volume of 20 μl was injected comprising treatment or 0.9% saline in a 1:1 mix with a bacterial suspension in nutrient broth. Larvae were housed in petri dishes in the dark at 37 • C and the larvae survival was monitored daily over 4 days.

Statistical analysis
Results were statistically analyzed using GraphPad Prism version 9.00 for Windows (GraphPad Software, CA, USA) and statistical significance was determined with an α = 0.05. All experiments were carried out at least in triplicate. Parametric data are represented by the mean ± standard deviation (SD), which was analyzed using paired 2-tailed ttests, one-way analysis of variance (ANOVA) with Dunnett's multiple comparison test for finding differences between treatment groups and untreated controls and two-way ANOVA with Š idák's multiple comparison tests, as described in the figure legends. G. mellonella survival data was analyzed using Kaplan-Meier survival curves with significant differences between groups determined by log-rank test, significance was Bonferroni-Holm-corrected for multiple comparisons.

Treatment of biofilms on hernia mesh materials
When we consider the antibacterial properties of Cu(DDC) 2 + Cu 2+ observed in microtiter plates possibly not correlating with complex biofilms present in SSIs [30], we used two biofilm models mimicking SSIs to further investigate the antibiofilm activity of 35 μM DDC − + 128 μM Cu 2+ in vitro. These models are based on an implant infection and a wound infection.
As an example of SSI on an implant, we investigated the biofilm formation of S. aureus and S. epidermidis on two commonly used, commercially available, hernia mesh materials and the ability of Cu (DDC) 2 + Cu 2+ to reduce the bacterial load on these meshes. S. aureus ATCC 6538, MRSA Mu50 and S. epidermidis ATCC 35984 formed extensive biofilms during 96 h batch incubations on polyester and polypropylene mesh material with log (CFU/mesh) values ranging from 7.21 to 8.91 (Fig. 1). The imaging of S. aureus ATCC 6538 biofilms on polyester meshes showed a multifilament mesh structure, exhibiting niches for bacteria to attach (Fig. 1D, top left). In contrast, the mono filaments of the polypropylene mesh were surrounded by S. aureus ATCC 6538 biofilms (Fig. 1D, top right). Studies suggest that staphylococci biofilms on hernia meshes may be associated with hernia repair failure and contribute to mesh shrinkage, chronic pain or hernia recurrence [45], and there may be an association between mesh porosity and the formation of biofilms [46].
When treated with 35 μM DDC − + 128 μM Cu 2+ , viability of S. aureus ATCC 6538 in biofilms was reduced on polyester and polypropylene meshes (Fig. 1A). Similar results were observed in MRSA Mu50 (Fig. 1B) and S. epidermidis ATCC 35984 (Fig. 1C) biofilms log 10 reduction on polyester meshes and polypropylene meshes. The S. aureus ATCC 6538 and MRSA Mu50 log 10 reduction was higher on polypropylene meshes compared to polyester meshes. This could be due to multifilament meshes forming denser biofilms than monofilament meshes because of the increased surface and presence of niches [47]. In addition, the highly hydrophobic Cu(DDC) 2 complex that is formed instantly when DDC − and Cu 2+ are mixed, might not reach the bacteria embedded in the niches of the multifilament mesh.
The imaging of Cu(DDC) 2 + Cu 2+ treated S. aureus ATCC 6538 (Fig. 1D, bottom left) confirmed a substantial number of bacteria in the niches formed by the intertwined filaments but showed mostly dead bacteria (red) on the polyester mesh and was associated with CFU reduction. In contrast, the S. aureus ATCC 6538 biofilm that previously surrounded the polypropylene filaments was in parts removed during washing steps, resulting in only few dead bacteria (red) imaged (Fig. 1D  bottom right). We quantified the viability based on the percentage of green and red fluorescent cells, which showed the viability was reduced when treated with Cu(DDC) 2 + Cu 2+ compared to the untreated control on polyester and polypropylene meshes (Supplementary file 1). However, significant background was present due to autofluorescence of the polyester and polypropylene that compose the meshes, which significantly affected automated counting of live and dead cells. This was unavoidable since further background removal would eliminate valid signal from the analysis. Therefore, the microscopy images visually complement the quantitative assessment of log 10 reduction of bacteria due to Cu(DDC) 2 + Cu 2+ . As the overall successful salvage rate of infected meshes can be as low as 10% and be inferior for infected polyester mesh compared to polypropylene mesh [4], the substantial log 10 reduction of Cu(DDC) 2 + Cu 2+ on both mesh material highlights the combination as a promising treatment approach for infected hernia meshes.

Efficacy in an in vitro wound model
As second in vitro SSI model, the artificial dermis model was chosen, as it closely resembles a chronic wound infection with similar nutritional conditions found in wound exudate and a dermis-like scaffold based on hyaluronic acid and collagen on which bacteria can attach and form biofilms [41,48]. Here, MRSA Mu50 and S. epidermidis ATCC 35984 biofilms were grown on an artificial dermis and exposed to 35 μM DDC − + 128 μM Cu 2+ (Fig. 2). The combination of Cu(DDC) 2 + Cu 2+ demonstrated a significant biofilm reduction in MRSA Mu50 and in S. epidermidis ATCC 35984 biofilms ( Fig. 2A). While the log 10 reduction was smaller compared to the mesh attachment model for both MRSA and S. epidermidis biofilms, Cu(DDC) 2 + Cu 2+ exposure still visually reduced the biofilms on the artificial dermis (Fig. 2B) and resulted in 97.2% and 81.5% MRSA Mu50 and S. epidermidis ATCC 35984 reduction, respectively, despite nutrient rich in vivo-like conditions. We propose three explanations for a reduced exposure of Cu(DDC) 2 + Cu 2+ with the biofilm on the artificial dermis.
Firstly, when DDC − and Cu 2+ solutions are mixed, the water insoluble Cu(DDC) 2 complex precipitates and sediments to the bottom of the well [49]. In previous biofilm experiments, including the biofilm on mesh material, biofilms were grown or placed at the bottom of wells, allowing for precipitated Cu(DDC) 2 to sediment onto and interact with the biofilms, while excess Cu 2+ was available in solution. In the wound model, biofilms are formed on top of the artificial dermis at the air-liquid interface (Fig. 2B). Therefore, when exposed to Cu(DDC) 2 + Cu 2+ , limited amount of Cu(DDC) 2 would precipitate onto the biofilm on the artificial dermis, while the remaining Cu(DDC) 2 might interact with the hydrophobic collagen or simply sediment to the bottom of the well. Secondly, Cu 2+ was shown to increase cross-linking of collagen in a concentration dependent matter [50], which can result in a reduced availability of Cu 2+ for the antibiofilm activity. Lastly, DDC − can be degraded to diethylamine and carbon sulfide in the presence of blood, due to the presence of plasma proteins and may therefore not be available to form the Cu(DDC) 2 complex [51]. Similar effects of the microenvironmental conditions in the artificial dermis model on the antibiofilm activity of antimicrobial agents were reported [42,48,52]. For example, Grassi et al. [48] observed inferior biofilm inhibition by antimicrobial peptides in the artificial dermis model compared to a 3D lung epithelial model due to the presence of blood and proposed the development of nanocarriers as drug delivery vehicle [53]. Consequently, to increase water solubility of Cu(DDC) 2 , prevent Cu(DDC) 2 sedimentation and protect DDC − from degradation, Cu 2+ and Cu(DDC) 2 were incorporated into PEGylated liposomes.

Characterization of Cu 2+ -liposomes and Cu(DDC) 2 -liposomes
PEGylated Cu 2+ -liposomes and Cu(DDC) 2 -liposomes were prepared and characterized according to Hartwig et al. [40]. The size, expressed as the d h , and the PDI were determined for Cu 2+ -liposomes and Cu (DDC) 2 -liposomes (Fig. 3) and were similar to previously reported values [40]. The size of both the Cu 2+ -liposomes and the Cu(DDC) 2 -liposomes were below 200 nm, allowing for sterile filtration and excluding the presence of large aggregates and extra-liposomal Cu (DDC) 2 [40]. In addition, the PDI of Cu 2+ -liposomes and Cu(DDC) 2 -liposomes was below 0.2, indicating a homogenous population of liposomes [54,55], which has previously been confirmed by imaging of mostly unilamellar vesicles in cryo-electron microscopy images [33,40]. The production of Cu(DDC) 2 -liposomes is based on DDC − diffusing through the membrane of Cu 2+ -liposomes and forming the insoluble Cu (DDC) 2 complex within the liposomes, which is characterized by the color change [49]. In addition, Wehbe et al. [33] showed that the amount of Cu(DDC) 2 in liposomes correlates with the amount of Cu 2+ in liposomes by comparing Cu 2+ to lipid ratio to Cu(DDC) 2 to lipid ratio. Therefore, it can be assumed that both liposomes have the same lipid constitution and consequently a similar amount of PEG polymers per liposome. Based on this assumption, the different sizes of the liposomes and the homogenous vesicle population, the PEGylation of Cu 2+ -liposomes would be denser compared to Cu(DDC) 2 -liposomes (Fig. 3).

Antibiofilm activity of liposomal Cu(DDC) 2 + Cu 2+
The liposomes were assessed for their activity against MRSA Mu50 and S. epidermidis ATCC 35984 biofilms (Fig. 4). As a fast and high throughput method [44], the alamarBlue assay was first performed to determine antibiofilm activity of the liposomal formulations (Fig. 4A). Treatment with Cu 2+ -liposomes or Cu(DDC) 2 -liposomes showed no activity against MRSA Mu50 and S. epidermidis ATCC 35984 biofilms. Similar to the effects of free Cu 2+ and Cu(DDC) 2 on MRSA and S. epidermidis biofilms [31], Cu 2+ -liposomes and Cu(DDC) 2 -liposomes concentrations up to a 4-fold increase did not inhibit biofilm viability (data not shown). The combination of [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes] also showed no antibiofilm activity against MRSA Mu50 and S. epidermidis ATCC 35984. This could be a result of the Cu (DDC) 2 -liposomes, the Cu 2+ -liposomes or both liposomes not releasing their content extracellularly or, following bacterial uptake, intracellularly. However, cellular uptake of PEGylated Cu(DDC) 2 -liposomes were observed in LS cells after 6 h incubation [40], which suggest bacterial uptake of the Cu(DDC) 2   viability was similar to the activity of free Cu(DDC) 2 + Cu 2+ against MRSA Mu50 and S. epidermidis ATCC 35984 biofilms. To further confirm these results, CFU counting was performed for treatments showing a reduction in biofilm viability with the alamarBlue assay (Fig. 4B). Treatment with [Cu(DDC) 2 -liposomes + free Cu 2+ ] and Cu(DDC) 2 + Cu 2+ resulted in a significant MRSA Mu50 log 10 reduction and a significant S. epidermidis ATCC 35984 log 10 reduction. As the antibiofilm activity of [Cu(DDC) 2 -liposomes + free Cu 2+ ] against MRSA and S. epidermidis was similar to free Cu(DDC) 2 + Cu 2+ and treatment with free Cu 2+ alone previously showed no antibiofilm activity against MRSA Mu50 and S. epidermidis ATCC 35984 at the tested concentration [31], we concluded that Cu(DDC) 2 was released from the Cu(DDC) 2 -liposomes, either intracellularly following bacterial uptake or extracellularly, but not the uncomplexed Cu 2+ from the Cu 2+ -liposomes.
Liposomes can penetrate the biofilm and release their content by fusing with the bacterial phospholipid membrane [56,57]. This interaction is dependent on biofilm properties, including bacterial species and matrix composition, and by the liposomal physicochemical properties [56]. Liposomes vary in surface charge, lipid composition, bilayer rigidity, surface modification, size and the incorporation of PEG polymers in the liposomal membrane [58,59]. As Cu(DDC) 2 -liposomes are produced by DDC − diffusion into Cu 2+ -liposomes, it can be expected that Cu(DDC) 2 -liposomes and Cu 2+ -liposomes have the same lipid constitution [33] and are only different in size and membrane PEGylation density. The denser PEGylation of the Cu 2+ -liposomes compared to the Cu(DDC) 2 -liposomes (Fig. 3) can present a physical barrier for Cu 2+ -liposome interaction with bacterial membranes or biofilm matrix, and therefore, prevent the intracellular uptake of the liposomal content [58]. PEGylated liposomes were previously shown to reduce interaction with target cells [60] and limit interactions with bacterial biofilms [61]. Liposomes with a PEGylated surface showed improved penetration of Pseudomonas aeruginosa biofilms but reduced the affinity of liposomes to bacteria compared to non-PEGylated liposomes. The PEG modifications on the liposome surface increase hydrophilicity of liposomes which increased the affinity to biofilm matrix components, such as extracellular polymeric substance [59]. In addition, PEGylated DSPC-containing liposomes with a low surface charge and rigid bilayer reduce adsorption of the DSPC-liposomes on S. aureus biofilms compared to non-PEGylated liposomes [61]. To investigate if the PEG polymers are hindering adsorption of Cu 2+ -liposomes on MRSA and S. epidermidis biofilms and consequently result in reduced antibiofilm activity of [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes], the penetration of fluorescently-labelled liposomes into the biofilm should be determined using microscopical analysis [61,62] and the antibiofilm activity of non-PEGylated [Cu (DDC) 2 -liposomes + Cu 2+ -liposomes] should be investigated. As hydrophilic PEG polymers integration on the surface of Cu(DDC) 2 -liposomes is necessary for superior drug to lipid ratio and improvement of colloidal stability during storage compared to non-PEGylated Cu (DDC) 2 -liposomes [40] and [Cu(DDC) 2 -liposomes + free Cu 2+ ] showed high antibiofilm activity against MRSA and S. epidermidis, incorporating Cu(DDC) 2 into PEGylated liposomes is a water-soluble alternative for a potential application on surgical site infections.

In vivo toxicity and antimicrobial activity of liposomal DDC − + Cu 2+
G. mellonella is an invertebrate infection model that is cost-and timeefficient, can mimic physiological conditions of mammals, such as temperature of 37 • C, and expresses a cellular and humoral innate immune system [63]. This immune system is capable of recognizing  pathogens and recruiting hemocytes to engulf pathogens and produce reactive oxygen species and antimicrobial peptides [64][65][66]. This model is in use for investigating pathogen virulence, for determining pharmacokinetic properties of antimicrobial agents and in vivo screening for antimicrobial activity and toxicity [66][67][68][69]. Efficacy and toxicity of antibiotics in G. mellonella infection models were reported to empirically support the observed effects of antibiotics in murine infection models and antibiotic susceptibility in humans [70].
To investigate potential toxic effects of the liposomes in vivo, G. mellonella larvae were exposed to liposomes and the survival was monitored over 4 days. Injection with Cu(DDC) 2 -liposomes, Cu 2+ -liposomes, the combination of [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes] and the combination of [Cu(DDC) 2 -liposomes + free Cu 2+ ] showed similar survival rates as the vehicle control (0.9% saline) and the untreated larvae, indicating no treatment toxicity in G. mellonella (Fig. 5A). Likewise, injection of free Cu 2+ (concentration within larvae 128 μM) has been previously shown to be not toxic to G. mellonella larvae [31].
Interestingly, the [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes] combination significantly increased the survival rate of S. epidermidis-infected G. mellonella larvae, despite showing no antibiofilm activity in vitro. This increase in S. epidermidis-infected larvae survival was not significantly different to the [Cu(DDC) 2 -liposomes + free Cu 2+ ] combination (p > 0.05). Consequently, the Cu 2+ -liposomes released their content in vivo, rendered excess Cu 2+ available and resulted in antibacterial activity. However, G. mellonella larvae were injected with bacteria and liposomes simultaneously, not allowing for in vivo formation of biofilms before treatment. Therefore, the in vivo activity of [Cu(DDC) 2 -liposomes + Cu 2+ -liposomes] might be limited to planktonic bacteria. In addition, survival of S. epidermidis-infected larvae, treated with Cu(DDC) 2 -liposomes alone was not significantly different to the survival rate of S. epidermidis-infected, untreated larvae, validating previously determined effects of free Cu(DDC) 2 + Cu 2+ in S. epidermidis-infected larvae, where excess of Cu 2+ was crucial for antibacterial activity. Moreover, absence of toxicity of Cu(DDC) 2 -liposomes and Cu 2+ -liposomes in G. mellonella larvae are in line with previous toxicity results of free Cu (DDC) 2 + Cu 2+ in G. mellonella and cell culture studies [31]. Consequently, the lack of toxicity and high efficacy of liposomal Cu(DDC) 2 + Cu 2+ observed in the G. mellonella model justify progressing to a mammalian in vivo infection model for pharmacological testing.

Discussion
We previously reported antibacterial and cytotoxic results of Cu (DDC) 2 + Cu 2+ against S. aureus and S. epidermidis in vitro and in G. mellonella larvae [31]. While the antibiofilm activity of Cu(DDC) 2 + Cu 2+ was determined in an in vitro biofilm model that is sufficient for an initial high throughput screening of novel antimicrobial drugs [31], this model is limited by the lack of resemblance to the microenvironment present in a human wound. Specific factors, such as wound exudate, host tissue, access to nutrients, formation of a biofilm gradient, presence of multiple bacterial species, inflammatory responses, and the immune system, influence the progression of a biofilm infection and the wound healing process [32]. By investigating the efficacy of antimicrobial compounds in physiologically relevant in vitro biofilm models of surgical site infections, instabilities of the drug or interactions with wound components can be detected and addressed to increase animal study validity before progressing to costly animal studies [48]. Although Cu (DDC) 2 + Cu 2+ preserved significant antibiofilm activity in challenging host-mimicking conditions, many factors present in an infected surgical wound, such as multiple bacterial species, the inflammatory response and the immune system were not incorporated in these in vitro models and can alter the outcome of future in vivo studies. Here, the effects of Cu (DDC) 2 + Cu 2+ on biofilms of the artificial dermis assay were diminished by the low water solubility of Cu(DDC) 2 and by possible interactions with matrix components, which significantly limits the clinical application of the free compounds and shows the importance of an appropriate drug delivery system. By narrowing the gap between in vitro results and in vivo translation, we comply with the 3Rs principles by Russell et al. [71] to improve the welfare of animals used for research.
While Cu(DDC) 2 showed in vitro activity against Mycobacteria [25], Streptococci [30,72], and Mycoplasma [73], the antibacterial effects have yet to be confirmed in animal models. In contrast, the research on Cu(DDC) 2 as cancer treatment has progressed to in vivo experiments and first clinical trials. The application of Cu(DDC) 2 in clinical trials is based on the separate oral administration of disulfiram and copper ions and the in-situ formation of Cu(DDC) 2 [74]. However, poor biostability and solubility of disulfiram and Cu(DDC) 2 often limit the treatment efficacy [51]. Alternative strategies are based on the encapsulation of Cu(DDC) 2 into nanocarrier, such as micelles [75], cyclodextrins [76] and liposomes [39,77,78]. Here, Cu 2+ -liposomes and Cu(DDC) 2 -liposomes composed of DSPC, cholesterol and DSPE-mPEG 2000 were investigated, as characteristics, including size, PDI, imaging, drug-to-lipid ratio and stability were described by Hartwig et al. [40] and Wehbe et al. [39] and freeze-drying of the liposomes enabled prolonged storage [79]. In addition, intravenous administration of 12.5 mg/kg modified PEGylated Cu(DDC) 2 -liposomes (without cholesterol) and 8 mg/kg of Cu (DDC) 2 -liposomes composed of DSPC and cholesterol were well tolerated in mice [39]. However, Wehbe et al. [39] only investigated the safety of Cu(DDC) 2 -liposomes and not the combination of [Cu (DDC) 2 -liposomes + Cu 2+ -liposomes] or [Cu(DDC) 2 -liposomes + free Cu 2+ ], which is necessary for the antibiofilm activity. Furthermore, the outcome of in vivo safety experiments could be altered by the different lipid composition of the PEGylated liposomes and the non-PEGylated liposomes, due to changes in circulation lifetime after intravenous administration [39]. While the non-PEGylated liposomes were not investigated because of instabilities during storage [40], the PEGylated Cu(DDC) 2 -liposomes with cholesterol were stable and showed no toxicity in G. mellonella at 6.4 mg/kg. G. mellonella larvae are a good indicator for toxicity and efficacy before progressing to mammalian studies, but the mechanisms of toxicity of the tested compounds can be altered by lack of mammal-specific metabolization processes. Therefore, the combined results of G. mellonella and cell assay studies are a predictor of low toxicity of antimicrobial agents but do not replace safety experiments in mammals [70,80].

Conclusion
The Cu(DDC) 2 + Cu 2+ combination at concentrations of 35 μM DDC − + 128 μM Cu 2+ reduced the bacterial load of MRSA and S. epidermidis biofilms in an implant and wound model in vitro. In addition, the low water solubility of Cu(DDC) 2 was overcome by incorporating the agents into liposomal carriers. Liposomal Cu(DDC) 2 + Cu 2+ showed antibiofilm activity in vitro against MRSA and S. epidermidis and in vivo efficacy against S. epidermidis, while being nontoxic. Therefore, the Cu(DDC) 2 + Cu 2+ combination represents a promising treatment strategy against S. aureus and S. epidermidis biofilm infections. Future studies will investigate the safety and efficacy of liposomal Cu(DDC) 2 + Cu 2+ in a mammalian model of wound infection.