Chronic wounds, generally comprising diabetic foot ulcers, pressure ulcers, venous leg ulcers, burn wounds, and wounds “older than 3 months of age” (Kirketerp-Møller et al., 2011) are a worldwide healthcare issue. They are causing a cycle of pain, anxiety and reduced quality of life for the individual patient, and present a considerable cost to the health care providers and patient (Dowsett, 2015). It is estimated that chronic wound therapies account for 2–4% of the total health system expenses, including material cost, nurse time, and hospitalization, that estimate is excluding the indirect costs such as loss of productivity and out-of-pocket expenses for patients (Sen et al., 2009). Over 90% of chronic wounds contain bacteria and fungi acquired from the skin, oral mucosa, enteric tract, or the environment (Kucera et al., 2014), which prevent tissue remodeling and healing. These bacteria can form a multispecies biofilm, which is often held responsible for the further development of an infection (Kucera et al., 2014). Infections commonly benefit from a priori compromised immune functions due to other diseases and/or their treatments, resulting in the lack of particular immune system components such as B cells, T cells, antibodies, neutrophils, damage to immune organs etc. Secondary immunodeficiency is often exploited by various microorganisms leading to grievous infections. Immunodeficiency is a well-established risk factor for cancer patients receiving chemotherapy (Pardoll, 2012; Khan et al., 2015).

Any break in the barrier function of the skin and other epithelial layers, predisposes an individual to infections (Schreier and Chatterjee, 2015). One of the major issues in controlling the healing initiation versus infection is sustaining the moisture balance in the tissue. Swollen and suppurative wounds provide perfect growth conditions for anaerobic and aerobic pathogens colonization and expansion thereof (Kirketerp-Møller et al., 2011), whereas a dehydrated wound surface can delay the healing process. Hence, while preserving a moist wound environment is advisable (Lilge et al., 2000) concurrent eradication and suppression of pathogenic microorganisms are highly desirable. A delay or failure in the treatment of moist wounds can lead to bacterial colonization progression and can cause the development of systemic infection.

Currently, there are some therapies to control infected wounds. The most classical – antibiotic therapy – starts to be inefficient as multidrug-resistant strains keep spreading and continue to expand unchecked. The increase in multidrug resistant strains to conventional antibacterial therapies has prompted the development of alternatives antimicrobial therapies, particularly for hard-to-heal wounds (Troxler et al., 2006). Negative pressure wound therapy promotes wound healing by applying a vacuum through a special dressing. It increases the granulation tissue formation and the local blood flow and enhances the bacterial clearance (Ene et al., 2015). Wound dressing saturated with silver, iodine or other antimicrobial agents can help the body achieve the ideal moist, warm healing conditions and simultaneously protect the wound from environmental exposure during this process (Murphy and Evans, 2012). Numerous plants are used in folk medicine against various diseases, albeit the exact mechanism of their action remains largely unknown, although antimicrobial properties of this phytotherapy have been investigated (Halcon and Milkus, 2004; Joubert et al., 2008; Kumar et al., 2008) and many of them are considered biofilm disruption agents. Combination of ethnopharmacology and antibiotic therapy could provide an effective bactericidal tool for the treatment of various bacterial and yeast infections (Taraszkiewicz et al., 2013). Low-Level Laser Therapy (LLLT) seems to exert possibly exploitable antimicrobial effects, especially against Pseudomonas aeruginosa caused infections, whereas it appears to be less or ineffective against other bacterial strains (Nussbaum et al., 2003, 2009).

To address the local nature of chronically infected wounds Antimicrobial Photodynamic Therapy (APDT) provides appe- aling properties. In APDT, a chemical compound (photo- sensitizer) absorbs light photons with sufficient quantum energy resulting in its electronic excited state, from which energy can be transferred to biomolecules or to molecular oxygen. The latter leads to the generation of Reactive Oxygen Species (ROS) or singlet oxygen formation, which can cause cell damage and microorganism’s death as recently reported by Taraszkiewicz et al. (2013) in in vitro studies of APDT. Localization of the therapy is achieved by two different mechanisms; first the confinement of the activation light to the clinical target area and second the difference in temporal photosensitizer association and uptake by bacteria versus host cells.

While efficacy of APDT has been demonstrated in various planktonic in vitro studies, transition to in vivo studies showed variable results, with some pre-clinical models showing good efficacy, whereas the majority of the studies demonstrated a poor translation of the APDT efficacy in vivo.

Typically, to quantify a therapy’s efficacy, based on the eradicated bacterial number, one group of hosts need to be sacrificed immediately after treatment and a second group need to be kept up to 50 h to determine the microbiota survival fraction. Blood samples can also be taken to analyze possible bacteremia development. Hamblin et al. (2003) and Wong et al. (2005) created full thickness excisional wounds, by using surgical scissors and forceps. Small wounds (8–9 mm × 12.5 mm) were infected with bioluminescent bacterial strains P. aeruginosa (10⁶) or Vibrio vulnificus (10⁶), allowing observation of wound healing and infection development within individual animals. As wounds remain uncovered, mice had to be kept separate to prevent opportunistic contaminations between animals. Zolfaghari et al. (2009) used two wound models: excisional and superficial. In the first case, shaved and depilated skin was pinched by sterile forceps and a 6 mm circular area (28 mm²) was cut down till the subcutaneous areolar tissue using scissors. In the second model a 25 mm² square shaped wound was created by scarification using a 27G needle. Both skin preparations were inoculated with 10⁸ CFUml^-1 MRSA cells on the wounds. To assess the effectiveness of the therapy mice were sacrificed upon treatment and the number of surviving bacteria was quantified following incubation on agar plates.

The ideal pre-clinical model of hard-to-heal wounds should be based on an immunocompromised host, thus permitting systematic infection. However, that model requires weeks to be established and carries significant risk of sepsis and survival failure. Also the use of immunocompromised animals can dramatically increase the cost of the research. Minimal animal number, wound drying, dressing changes causing wound oxygenation variations, unnecessary animal pain, or large surface area damage are some of the important factors to be considered during the in vivo experimental design stage.

Here we demonstrate a Tegaderm^TM based topical infection model for antimicrobial PDT as local antimicrobial therapy, and demonstrate its utility by testing Rose Bengal (RB), porphyrin TMPyP New Methylene Blue (NMB), or TLD1411 mediated APDT.

By the use of strong skin adhering, vapor and light transmitting dressings (Lilge et al., 2000), a moist wound environment is maintained providing perfect condition for wound healing as well as bacterial growth and the development of infection. Dressings can be kept for several days, isolating wounds from environmental- and cross-contaminations, permitting co-housing of animals thus reducing their stress level. Of additional advantage is the use of bioluminescent bacterial strains permitting non-invasive longitudinal monitoring of the therapeutic efficacy in each animal, reducing the number of animals required to achieve statistical significance. Moreover, from an animal care point, the possibility to decrease the wound size as well as provide opportunistic infection prevention will promote better tissue reconstruction.

To demonstrate the utility of the murine model, we validated the efficacy of 4 different photosensitizers RB, porphyrin TMPyP, NMB, and TLD1411 as APDT mediators in planktonic solution, demonstrated the need for a covered wound for maintaining an ongoing infection and finally demonstrated the ability to quantify APDT outcome in this murine model, in vivo as time delay until pathogen regrowth is noted.

While various infected wound models have been published, there is still a need for an applicable model mimicking the clinical features of the disease, with short implementation time and efficient translation of the study results into clinical practice. Lee et al. (2011) created a large wound (1.5 by 2.5 cm) by tape-stripping and sandpaper based removing of stratum corneum and inoculating a bacterial suspension containing ~10¹⁰ CFUml^-1 S. aureus cells. Biopsies were collected at specific time points to monitor wound healing and host cell growth. Mice were kept in individual cages to prevent cross contamination, affecting overall well-being of the animal by causing stress. Saito et al. (2012) exposed dorsal skin to 90°C to induce burn injury up to 20% of the total body surface area and then inoculated with 10⁸ CFU of P. aeruginosa. The model mimics large, festering wounds and can cause high pain and stress levels. Application of an occlusive dressing to avoid cross contamination is favorable in this model. Similar wounds were established by Hashimoto et al. (2012), where preheated steel was applied to the dorsal surface. Large burned areas (6–6.5 cm²) were immediately infected using 10⁸ P. aeruginosa cells.

The murine model of an infected wound presented here is robust and spontaneous bacterial clearing was not observed in the Tegaderm^TM covered wounds, thus it presents a favorable environment for testing APDT efficacy in vivo. The wound cover allows for oxygen and water vapor across the barrier and does not impede light penetration either for BLI, APDT, or LLLT (Lilge et al., 2000). It provides constant growth environment for both PAK and XEN, as demonstrated in Figures 2 and 3, but prevents opportunistic and environmental wound cross-contamination. Several mice can be housed in the same cage reducing stress levels during the experiment. Wounds, which were not covered or intentionally uncovered, dried out rapidly and infections were controlled without intervention either by the murine immune system or simple dying of the bacterium in an unfavorable environment possibly leading to an overestimation of the APDT efficacy.

Here the model was only utilized in the acute stage of infection, so sufficiently long to open the possibility of resilience against APDT, but insufficient to generate a deep infection, particularly because the pO₂ in chronically infected wounds has been reported to be 30% to 90% lower than in healthy tissue (Schreml et al., 2010). The latter may require photosensitizers exploiting both type I and II APDT mechanisms (Fong et al., 2015; Shi et al., 2015).

However, the combination of BLI together with consistent growth environment allows testing various APDT protocols including the number of retreatment frequencies and extending the infection control.

As the APDT activation wavelength impacts the penetration depth in the in vivo model, the most common wavelength of 525 nm was selected. However, the varying molar extinction coefficient of these 4 photosensitizers led to self-shielding and possibly with that reason limited efficacy in planktonic solution for RB and NMB was observed. Selecting different concentrations to match the molar extinction coefficient would impact on the number of photosensitizers per bacterium, whereas selecting different wavelength to match the molar extinction coefficient at constant concentration would affect the light distribution in the tissue and hence by diffuse reflected light also the photon density at the tissue surface.

At time of APDT light treatment in vivo the target must be considered 2D with minimum full-thickness wounds and no observable presence of blood in an infected area. Hence, comparing the efficacy using this animal model system is valid purely on tissue optics consideration.

Nevertheless, the added attenuation of PS in in vivo condition needs to be considered, particularly when comparing PS efficacy in APDT. The worst-case impact on APDT radiance exposure on the target can be estimated based on the solution volume (30 μl) over the 6 mm diameter wound, which would result in an ~1 mm layer large compared to the diffusion distance of the ROS. As the PS acts also as a 525 nm light filter prior to the pathogens on the tissue surface, the resulting optical densities are RB = 1.83, NMB = 0.235, TMPyP = 0.653, and TLD1411 = 0.4155, respectively. In particular, for RB the high concentration employed in conjunction with the high molar extinction coefficient eventually caused a reduction of the radiant exposure at the microbes to 1.48 Jcm^-2. Similarly, the worst-case radiant exposures for the other photosensitizers would be NMB = 58.2 Jcm^-2, TMPyP = 22.2 Jcm^-2, and TLD1411 = 38.4 Jcm^-2, respectively. However, the solution was rapidly absorbed into the tissue and the filter effect by the photosensitizer had a lesser impact on the sensitizer specific APDT efficacy differences.

Conversely, trying to achieve a comparable OD in the tissue to eliminate effect of self-shielding as a confounding factor. There is a need for either adjustment of the photosensitizer’s concentration affecting the ratio of photosensitizing molecules per bacterium or selecting particular wavelength for each photosensitizer so that the molar extinction coefficients are the same, which in turn would impact the photon distribution in the tissue. Hence, our approach of using constant wavelength and radiant exposures in the sensitizer was compared.

To maintain reasonable penetration of the activated light into the tissue, the presence of oxy- or deoxy-hemoglobin needs to be considered as their absorption coefficients for 525 nm are > 10² cm^-1 and the penetration depth would be limited to < 100 μm (Jacques, 2013). Hence, the model can only be valid for wounds free of blood or scab. When treating burns a longer absorption wavelength is required for NMB or mixture of Ru(II) containing photosensitizer with carrying molecules, such as transferrin, which notably increases their molar extinction coefficient in the 600 to 800 nm range (Kaspler et al., 2016). Care must be taken regarding possible LLLT effects at these longer wavelength (Nussbaum et al., 2003).

A radiant exposure of 150 Jcm^-2 can be achieved in 12.5 min according to the guidelines for safe irradiation of the skin, and the duration is also clinically acceptable. However, the experiments showed that for optimal efficacy of the therapy, a maximum irradiance of 50 mWcm^-2 should not be exceeded allowing sufficient oxygen diffusion throughout the Tegaderm^TM wound cover. So care must be taken as possible photosensitizer reuptake into mammalian host cells, particular cells associated with the immune system needs to be considered when the irradiation times need to be extended significantly.

The efficacy of various photosensitizers against bacteria in planktonic solutions have been previously published for RB (Melo et al., 2011) and TMPyP (Cassidy et al., 2010) reporting concentration of 10 μmol/L (6 log of killing) and 250 μg/L (4 log of killing), respectively.

All photosensitizers have displayed good to excellent efficacy in bacterial inactivation for planktonic solutions particularly for PAK where 3 of 4 reached 6 log₁₀ inactivation which is considered as sterilization state and the other photosensitizer (NMB) achieved at least 3 log₁₀ of inactivation for 150 Jcm^-2. The efficacy limits for RB and TMPyP are probably an artifact due to their high molar extinction coefficient and the high concentration used, which prevented effective illumination of the planktonic solution over the 1 cm sample depth. Radiant exposure with a much lower absorption coefficient of NMB compared to RB, its limited efficacy cannot be attributed to self-shielding. Additionally, as the photosensitizers have comparable singlet oxygen quantum yield one needs to conclude that NMB does not associate well with PAK. TLD1411 appears to present higher efficacy compared to other photosensitizers with a concentration less than 10 μM being sufficient for transient PAK sterilization at 100 Jcm^-2. This may be explained by the ability of these ruthenium complexes to lead to DNA cleavage (Mari et al., 2014; Shi et al., 2015).

The high efficacy of these PSs in vitro is poorly translated into the in vivo applications, as previously noted (Amin et al., 2016). The reason for this efficacy loss can be explained by host factors as well as bacterial burden and the virulence of individual strains as previously noted (Hamblin et al., 2003) and reflects a more challenging environment also for the treatment of human superficial infections.

One caveat to these experiments is the small number (n = 4–5) employed for the different study groups. While this sample size is too small to allow statistical analysis of the results between photosensitizers, it is large enough to demonstrate an APDT effect and the report time line for BLI signal delay and time to signal saturation.

While only up to a 1.5 log₁₀ reduction in BLI associated radiance was detected by PS mediated APDT in vivo, it may still be of clinical use. Based on the BLI radiance expression of the PAK and XEN growth curve, pre and post light exposure, it becomes evident that for certain protocols and conditions APDT treated wounds do not reach the BLI radiance intensity of the unexposed control for at least 24 h in case of PAK mediated infection and more than 4 days for XEN mediated ones. Despite the lack of complete response in wound sterilization this additional time increases the period to early onset of risk for local or systemic infection, thus providing the possibility to perform wound culture and antibiotic sensitivity testing, as about 90% of blood and other cultures are detected and identified within 48 h (Cockerill et al., 2004). Peptide nucleic acid fluorescence in situ hybridization (PNA FISH) methods are currently used for direct identification of selected Staphylococcus, Enterococcus, Klebsiella, and Candida species in less than 2 h (Shepard et al., 2008). While a 1 to 2 h delay for PCA determination of loss or the presence of genes encoding antibiotic resistance are available and appear reasonable, PCA will not describe functional protein and may be misleading, hence, generally antibiotic susceptibility testing commonly requires 1 to 3 days and hence a general infection control for 24 to 72 h is desirable.

Our murine model of consistent local wound infection with Staphylococcus aureus (MRSA) and P. aeruginosa was similar to chronic wounds, without spontaneous clearing due to wound drying. The progress of the infection can be monitored real-time for bioluminescent bacteria, thus permitting the rapid evaluation of APDT agents and treatment protocols in vivo under physiologically relevant conditions for bacterial growth.

The in vivo evaluation of antimicrobial effect of photo- sensitizers responded less effectively compared to what were seen in vitro, an observation also reported by others; however, the potentially significant clinical reduction of the bacterial load remains apparently feasible, causing delay in full activation of inflammatory response and hence reducing the risk of developing sepsis.

The presented murine model is a suitable platform to evaluate the feasibility and safety of repeated cycles of PDT, based on clearance time of photosensitizer from host mammalian cells.

GF has performed in vitro experiments with the use of RB, TMPyP, and NMB against P. aeruginosa, as well as all in vivo procedures. He wrote the first draft of the manuscript. KK assisted and supported the in vivo experiments. YA performed the in vitro experiments. JN: participated in the conception and design of the work; advised on the experimental design and writing of the manuscript. MG participated in the conception and design of the work; drafted the manuscript and revised it critically for important intellectual content. KB participated in designing of the experiments and helped drafting manuscript. LL designed the experiments together with Drs. MG, JN and KB, supervised GF, KK, and YA on a daily basis and is the senior responsible author for the manuscript together with GF. He also acts as the corresponding author on this manuscript.

The 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.