Bacterial strains were collected from the sinonasal cavities of chronic rhinosinusitis patients. Ethics approval was obtained from The Queen Elizabeth Hospital Human Research Ethics Committee, Woodville, SA, Australia. Strains included one S. aureus clinical isolate (parent strain P1), which was prolonged subcultured at MIC or higher concentrations of gentamicin (≥2 μg/ml). This induced small colony variants (SCV1) that featured elevated gentamicin tolerance. Another strain was a clinically isolated S. aureus small colony variant (SCV2).

Hydrogels were prepared by adequate mixing of dextran-aldehyde, succinyl-chitosan, and a buffer solution, as previously described (Richter et al., 2017b). Def (20 mM) and/or GaPP (100 or 500 μg/ml) were incorporated in the gel and compared to blank gel and antibiotic loaded gel including 5 μg/ml ciprofloxacin (Cip) and 100 μg/ml gentamicin (Gent). Cip and Gent were chosen due to their clinical relevance as antibiotic therapies for respiratory tract, skin, blood, bone, and soft tissue infections. Furthermore, fluoroquinolone antibiotics (such as Cip) were described as being highly effective against SCVs in vitro and in vivo (Garcia L. et al., 2012; Garcia et al., 2013).

A bacterial suspension was prepared in 0.9% saline and adjusted to 1.0 McFarland units (~3 × 10⁸ CFU/ml). After diluting bacteria 1:1,000 in tryptone soya broth, 2 μl were spotted on UV-sterilized Whatman polycarbonate membrane filters (0.2 μm pore size, GE Healthcare, Little Chalfont, UK) and placed on tryptone soya agar (Merritt et al., 2005). Following biofilm formation after 24 h incubation at 37°C, the membrane filters were transferred onto AB trace agar (minimal growth agar including 0.5% glucose and 0.5% peptone). One hundred microliters of freshly prepared hydrogel was placed on biofilms and incubated for 2.5 days at 37°C. The filter membranes were then aseptically transferred to new AB trace agar plates and incubated for 2.5 days at 37°C. Finally, the filters were collected in PBS to extract bacteria by vortexing (1 min) and sonication (15 min), prior to serial dilutions and plating on tryptone soya agar for CFU counting and colony morphology determination after 3 days incubation at 37°C.

The antibiofilm effect of drug loaded gels was rated using the Bliss Independence Model (Morones-Ramirez et al., 2013; Thangamani et al., 2015). The synergy of treatment combinations (Def-GaPP100, Def-GaPP100-Cip, Def-GaPP100-Gent, Def-GaPP500) was calculated according to the following equation, with values above zero corresponding to synergistic effects. (1)S=(aMG)*(bMG)-(abMG)

S = synergistic effect, a = Log₁₀ of biofilm after exposure to compound a, b = Log₁₀ of biofilm after exposure to compound b, ab = Log₁₀ of biofilm after exposure to treatment combination ab, MG = Log₁₀ of untreated biofilm (maximum growth).

Hyaluronic acid (1.20–1.80 MDa, Lifecore Biomedical, Chaska, MN) and collagen (Corning Incorporated, Corning, NY) were used to prepare an artificial dermis (Brackman et al., 2016), which was immersed in lyophilized bovine plasma, 19 ml Bolton broth (Oxoid), 1 ml horse blood and 10 IU of heparin. Subsequently, 10 μl of an overnight culture adjusted to 1 × 10⁶ CFU/ml was spotted on top of the dermis and incubated for 24 h at 37°C. The formed biofilms were exposed to 150 μl gel (blank gel, Def, GaPP 500, Def-GaPP500, and Cip gels) for 24 h at 37°C. Following a washing step, the dermis was immersed in 10 ml of 0.9% saline to recover the bacteria by vortexing and sonication (alternating cycles of 3 × 30 s), prior to serial dilutions and plating for CFU counting. In addition, the colony morphology was determined.

Synchronized nematodes, Caenorhabditis elegans AU37 (glp-4; sek-1), were grown to L4 stage, suspended in OGM medium (95% M9 buffer, 5% brain heart infusion broth, 10 μg/ml cholesterol) and added into 96-well plates with at least 20 worms per well (Brackman et al., 2011). Nematodes were infected with 25 μl of an overnight culture adjusted to 2 × 10⁹ CFU/ml in OGM medium and exposed to 25 μl of treatment (Def 20 mM, GaPP 500 μg/ml or a combination of both). Uninfected nematodes in OGM medium as well as infected but untreated nematodes were used as controls. The number of viable and dead nematodes was assessed every 24 h over 3 days incubation at 25°C. Subsequently, nematodes were first washed in M9 buffer containing 1 mM sodium azide, then washed in PBS prior to counting. The nematodes were mechanically disrupted by vortexing the worms in microtubes with 1.0 mm silicon carbide beads for 10 min (BioSpec Products, Bartlesville, OK). Serial dilutions of the supernatants were plated (tryptone soya agar with 7.5% NaCl) for CFU counting and colony morphology determination.

All experiments were conducted in triplicate and are presented as mean ± standard deviation (SD) or mean ± standard error of the mean (SEM). Results were analyzed using two-way analysis of variance with Dunnett's test (GraphPad Prism version 7.02, GraphPad Software, La Jolla, CAL). Statistical significance was assessed at the 95% confidence level.

The colony morphology of bacterial strains is shown in Figure 1 and the growth curves are displayed in Figure 2. A summary of bacterial characteristics is shown in Table 1.

SCV1 was observed to be a catalase positive, coagulase positive, and hemolysis positive strain with hemin/menadione auxotrophy. SCV2 was determined as a catalase negative, coagulase positive, and hemolysis negative strain with thymidine auxotrophy. P1 was identified to be catalase positive, coagulase negative, and hemolysis positive with hemin/menadione auxotrophy. As depicted in Figure 2, during 24 h SCV1 and SCV2 showed a slower growth rate and a lower OD 600 value than the parent strain P1. While SCV1 reached stationary and decline phase after 24 h, SCV2 continued to grow reaching an OD 600 of 0.95 after 40 h. Loops taken after 24 h revealed a small colony morphology for both SCV1 and SCV2, while after 40 h SCV2 presented as a mix of small colonies and very few normal sized colonies. SCV1 showed a small morphology after 40 h.

SCV1 and SCV2 were 4- and 16-fold more susceptible to Def and 4- and 8-fold less susceptible to Cip and Gent compared to the parent strain P1. SCV1 and P1 had low MICs for GaPP (6.25 and 12.5 μg/ml, respectively), while SCV2 was observed to have a MIC above 50 μg/ml for GaPP. The MICs for the Def-GaPP combination were typically lower than the MICs for individual compounds, however, the extent of this difference was strain-dependent. As displayed in Table 1, SCV1 was susceptible to Mup and Van, and showed increasing MIC values for Ceph, Doxy, and Chlor, and was not susceptible to Amoxi and Strep. SCV2 was susceptible to Doxy, less susceptible to Chlor and not susceptible to Mup, Ceph, Van, Amoxi, and Strep. P1 showed low MIC values for Doxy and Mup, and increasing MICs for Van, Ceph, Amoxi, and Chlor, and was not susceptible to Strep.

The blank gel and the Def gel showed no antibiofilm activity (data not shown), while the GaPP gel demonstrated a concentration- and strain-dependent effect. Gel loaded with a low concentration of GaPP (100 μg/ml) showed a log₁₀ reduction of 1.7 and 1.8 against biofilms of SCV1 and its parent strain P1, respectively, but no antibiofilm activity against SCV2 (data not shown), while at 500 μg/ml GaPP showed a log₁₀ reduction of 4.3, 1.4, and 2.0 in SCV1, SCV2, and P1 biofilms (Figure 3). Cip and Gent loaded gels were observed to have only a minor effect against SCV1, SCV2, and P1 (log₁₀ reduction of 0.4, 0.1, and 1.1, respectively for Cip and 0.2, 0.8, and 1.1, respectively for Gent). A small colony morphology of both SCV1 and SCV2 has been observed when analyzing CFUs after treatment exposure.

The hemin auxotroph SCV1 (Figure 3A) showed increased susceptibility to gel loaded with Def-GaPP100-Cip, Def-GaPP100-Gent, GaPP 500, and Def-GaPP500 (log₁₀ reduction of 3.4, 5.4, 4.3, and 4.4, respectively), and decreased susceptibility to monotherapy with Cip or Gent compared to its parent strain P1. Interestingly, Def-GaPP100 combined with Gent showed a high degree of synergy (Figure 4) and significant activity against the highly Gent-tolerant SCV1 compared to Gent alone and Def-GaPP100 (log₁₀ reduction of 5.4 for Def-GaPP100-Gent vs. 0.2 for Gent and 1.3 for Def-GaPP100, p < 0.0001).

The thymidine auxotroph SCV2 (Figure 3B) showed elevated susceptibility to gel incorporating Def-GaPP100-Cip, Def-GaPP100-Gent, GaPP 500, and Def-GaPP500 (log₁₀ reduction of 3.1, 1.7, 1.4, and 2.5, respectively), and no susceptibility to Cip or Def-GaPP100. Notably, Def-GaPP100 combined with Cip showed high synergy (Figure 4) and significant activity against SCV2, manifestly exceeding the effect of the individual compounds and Def-GaPP100 (log₁₀ reduction of 3.1 for Def-GaPP100-Cip vs. 0.1 for Cip and 0.1 for Def-GaPP100, p < 0.0001).

Against the hemin auxotroph P1 (Figure 3C) the combination of Def-GaPP100 gel with Cip or Gent showed significantly higher activity than the antibiotics alone (log₁₀ reduction of 1.8 for Def-GaPP100-Cip vs. 1.1 for Cip, p < 0.001; and 1.9 for Def-GaPP100-Gent vs. 1.1 for Gent, p < 0.0001). However, no synergistic effect was observed compared to the Def-GaPP100 gel (1.8 log₁₀ reduction). The highest activity against P1 was achieved with Def-GaPP500 gel (2.4 log₁₀ reduction).

The antibiofilm activity of loaded hydrogels was macroscopically analyzed over 5 days of treatment exposure. All biofilms grew extensively when exposed to blank gel and gel incorporating Def (not shown), Cip and Gent (Figure 5). A concentration- and strain-dependent antibiofilm effect was apparent after exposure to gels incorporating GaPP (not shown), Def-GaPP, Def-GaPP-Cip, and Def-GaPP-Gent. While Def-GaPP100 gel moderately inhibited bacterial growth, the combination of Def-GaPP100 with either Cip or Gent resulted in a substantial antibiofilm effect (Figure 5).

Biofilms grown on an artificial dermis were exposed to drug loaded hydrogels to determine the antibiofilm activity in an in vitro wound model (Figure 6). The visual analysis of CFUs after treatment exposure confirmed a small colony morphology for the majority of both SCV1 and SCV2. The untreated control and the blank gel showed similar growth of all biofilms, indicating no antibiofilm effect of the blank gel. The Def gel demonstrated substantial antibiofilm activity against SCV1 (log₁₀ reduction of 0.9), but failed to be effective against P1 and SCV2. The GaPP 500 gel showed minor antibiofilm activity with a 0.2–0.4 log₁₀ reduction. In contrast, Def-GaPP500 gel showed significant antibiofilm effects against SCV1, SCV2, and P1 with a 1.4, 1.0, and 0.9 log₁₀ reduction, respectively, thereby demonstrating significantly higher activity than the individual compounds (p < 0.05–0.0001) and slightly higher activity than Cip gel (Figure 6).

Nematodes were infected with bacteria and their survival rate was determined with and without Def, GaPP 500, or Def-GaPP500 treatment (Figure 7). The worm killing was strain-dependent with 75, 25, and 57% survival in SCV1, SCV2, and P1 infected worms, respectively, while uninfected controls showed 88% survival over 3 days. When worms were exposed to Def, 45, 73, and 71% of SCV1, SCV2, and P1-infected nematodes survived, while 73, 48, and 81% infected worms survived when treated with GaPP 500. The combination of Def-GaPP500 showed a similar survival rate in SCV1 and P1 infected worms as GaPP 500 alone (71 and 87% survival, respectively). In contrast, the survival rate of SCV2 infected worms was substantially higher (86% survival) when treated with Def-GaPP500 compared to the individual treatments. Compared to the uninfected control the Def-GaPP500 treatment achieved similar survival rates in SCV2 and P1 infected worms. Furthermore, the survival of SCV1 and P1 infected worms after GaPP 500 treatment was not different to uninfected controls (Figure 7).

Following 3 days of infection the bacterial load per worm was quantified (Figure 8) by enumeration of CFU. Small colony phenotypes of SCV1 and SCV2 were observed. Consistent with results in the colony biofilm model and wound model, the treatment with Def alone showed no significant effect and failed to reduce the CFU per worm. In contrast, both GaPP 500 and Def-GaPP500 showed a significant (p < 0.05) reduction of the bacterial load, resulting in a log₁₀ of 2.0 for GaPP and 2.3 for Def-GaPP in SCV1 infected worms (SCV1 infection control: log₁₀ of 3.6 CFU/worm), 2.9 and 2.5 in P1 infected worms (P1 infection control: log₁₀ of 3.6 CFU/worm), and 3.0 and 2.8 in SCV2 infected worms (SCV2 infection control: log₁₀ of 4.5 CFU/worm).

Competing financial interests: PW holds a patent on the chitosan-dextran hydrogel. PW and SV hold a patent application on the treatment combination of deferiprone and gallium-protoporphyrin. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.