Visible Light as an Antimicrobial Strategy for Inactivation of Pseudomonas fluorescens and Staphylococcus epidermidis Biofilms.

The increase of antimicrobial resistance is challenging the scientific community to find solutions to eradicate bacteria, specifically biofilms. Light-Emitting Diodes (LED) represent an alternative way to tackle this problem in the presence of endogenous or exogenous photosensitizers. This work adds to a growing body of research on photodynamic inactivation using visible light against biofilms. Violet (400 nm), blue (420 nm), green (570 nm), yellow (584 nm) and red (698 nm) LEDs were used against Pseudomonas fluorescens and Staphylococcus epidermidis. Biofilms, grown on a polystyrene surface, were irradiated for 4 h. Different irradiance levels were investigated (2.5%, 25%, 50% and 100% of the maximum irradiance). Surviving cells were quantified and the inactivation kinetic parameters were estimated. Violet light could successfully inactivate P. fluorescens and S. epidermidis (up to 6.80 and 3.69 log10 reduction, respectively), while blue light was effective only against P. fluorescens (100% of maximum irradiance). Green, yellow and red irradiation neither increased nor reduced the biofilm cell density. This is the first research to test five different wavelengths (each with three intensities) in the visible spectrum against Gram-positive and Gram-negative biofilms. It provides a detailed study of the potential of visible light against biofilms of a different Gram-nature.


Introduction
Biofilms are bacterial communities, mono-or multi-species, embedded in a self-produced matrix mainly composed of proteins, polysaccharides, extracellular DNA and lipids, which constitute the extracellular polymeric substance (EPS) [1][2][3][4][5][6]. The EPS allows bacteria to adhere to many surfaces and it protects the community from hazardous compounds. This crucial layer of protection decreases the bacterial susceptibility to disinfectants and antibiotics [7,8] and promotes cell communication via quorum sensing [9]. Biofilms living on surfaces have evolved to be more resistant to antibiotics, compared to their planktonic form [10,11]. They are identified as the major source of microbial contamination within the food industries, households and clinical environment [2]. Moreover, the development of antibiotic-resistant bacteria is raising a public health concern, and, at the same time, this underlines the urge for the scientific community to develop alternative antibacterial methods and strategies to fight infections [12]. Biofilms of different strains have been found in factories [13],

Results
Biofilms of P. fluorescens and S. epidermidis, grown on polystyrene, were exposed to colored LED arrays (violet, blue, green, yellow and red). Different irradiances were tested (2.5%, 25%, 75% and 100% of I max ). The Geeraerd et al. model was fitted to experimental data [62]. Specifically, for P. fluorescens (for all the light colors and conditions) and S. epidermidis for the violet light, the Geeraerd et al. model without shoulder was used, while for S. epidermidis for the blue, green, yellow and red light, the model without shoulder and tail, which reduced to a log-linear regression model, was employed. The model allowed for the estimation of the inactivation parameters, i.e., initial population N 0 , residual population N res (if applicable depending on the model) and inactivation rate k max . Sublethal injury kinetics, expressing the sub-lethally injured cell percentages within the biofilm due to the treatment, were calculated using Equation (6) for both bacterial biofilms.

Light Characterization
The LED arrays were optically characterized. The emission spectra were measured, and they were found to have an emission peak (λ max ) at 400, 420, 570, 584 and 698 nm for violet, blue, green, yellow and red, respectively. The values of full width at half maximum, ∆λ, ranged from 20 to 95 nm (details are reported in Table 1). The power densities emitted by the LED when the percentage of I max was set at 2.5%, 25%, 75% and 100% are listed in Table 1 and ranged from 0.043 to 29.2 mW cm −2 . After 4 h of light exposure, the total dose received by the biofilms was calculated using Equation (1). It ranged from 10.1 to 420.5 J cm −2 for violet, from 2.5 to 6.9 J cm −2 for blue, from 1.6 to 3.7 J cm −2 for green, from 0.6 to 2.0 J cm −2 for yellow and from 1.7 to 5.0 J cm −2 for red LED array. The violet LED array emitted Antibiotics 2020, 9,171 4 of 23 the highest irradiance and 2.5% I max was adopted to have irradiances comparable to the remaining LED arrays. The detailed irradiances and doses for each condition are also reported in Table 1.

Differences in Microorganisms' Biofilms
Two model biofilms were developed from two different bacteria: P. fluorescens, which are Gram-negative bacteria, and S. epidermidis, which are Gram-positive ones. Although the mono-species biofilms were grown using the same methodology, after 24 h of growth, the resulting biofilms were slightly different in (i) Colony Forming Units (CFU) per area and (ii) resistance to desiccation, perhaps due to the difference in Gram classification. In fact, (i) 24 h-grown P. fluorescens biofilms reached a population of approximately 10 8 CFU cm −2 , while S. epidermidis biofilms had a population of 10 7 CFU cm −2 (these population densities were then exposed to light treatment). Moreover, (ii) P. fluorescens biofilms were more sensitive to evaporation of water that caused some cells to die because of drying (black line and dots in Figure 1). On the contrary, S. epidermidis biofilms showed no reductions due to drying (black line and dots in Figure 2). The population of the S. epidermidis biofilm also remained constant after 4 h.   (o for general medium,GM, and x for selective medium, SM) and lines (solid line for GM, dashed line for SM) represent the control (no light exposure). Likewise, the colored symbols and lines represent the fitting after 2.5% (only for violet LED array, light blue), 25% (yellow), 75% (green) and 100% of Imax (red). Sublethal injury kinetics: the colored lines represent the SI after 2.5% (only for violet LED array, light blue), 25% (yellow), 75% (green) and 100% of Imax (red).
(a)    Figure 1. Different irradiation conditions were tested for each color (2.5% for violet only, 25%, 75% and 100% of Imax). N0, Nres and kmax were estimated and their respective standard errors (SE) and the root mean squared error (RMSE) values were calculated. These values were summarized in Table 2, along with the statistical analysis. The overall log10 reductions (N0 − Nres) were also calculated and are listed in the same table.
The inactivation kinetics are characterized by a log-linear inactivation phase, from time 0 to approximately 90-100 min, followed by a tailing phase ( Figure 1). The log-linear inactivation is present in the control as well, due to a desiccation process occurring to the biofilms within the first 100 min.
Regardless of the treatment, the rate of inactivation kmax was in general very similar to the control (no significant differences), as can be noticed from the kmax values displayed in Table 2. This means that the inactivation rate was not influenced by the light exposure, while the residual population Nres was strongly influenced in some conditions. Different percentages of Imax (2.5% for violet only, 25%, 75% and 100% of Imax) per LED color resulted in higher or equivalent log10 reductions compared to the control (no light exposure). The log10 reductions indicated that the highest reductions were obtained with violet light (in all conditions) and blue light when Imax equaled 100%. Indeed, the most effective inactivation was observed when the violet array was tested on biofilms. This resulted in a log10 reduction ≤ 6.80 CFU cm −2 , while the reduction of the control was only 3.87 CFU cm −2 (due to natural desiccation). A log10 reduction ≤ 5.37 CFU cm −2 was induced for biofilms exposed to blue LED irradiation. Furthermore, experiments were carried out with green, yellow and red irradiation. These revealed no additional reduction compared to the control.

Sublethal Injury
The model was also fitted to the data resulting from the selective plating media. The inactivation kinetics (dashed lines) and the parameters, related to selective media and estimated from the Geeraerd et al. [62] model, are displayed in Figure 1 and Table 2. The sublethal injury of P. fluorescens was determined by plating the bacterial suspension on general (GM) and selective media (SM) to . Likewise, the colored symbols and lines represent the fitting after 2.5% (only for violet LED array, light blue), 25% (yellow), 75% (green) and 100% of I max (red). Sublethal injury kinetics: the colored lines represent the SI after 2.5% (only for violet LED array, light blue), 25% (yellow), 75% (green) and 100% of I max (red).  Figure 1. Different irradiation conditions were tested for each color (2.5% for violet only, 25%, 75% and 100% of I max ). N 0 , N res and k max were estimated and their respective standard errors (SE) and the root mean squared error (RMSE) values were calculated. These values were summarized in Table 2, along with the statistical analysis. The overall log 10 reductions (N 0 − N res ) were also calculated and are listed in the same table.

P. Fluorescens
The inactivation kinetics are characterized by a log-linear inactivation phase, from time 0 to approximately 90-100 min, followed by a tailing phase ( Figure 1). The log-linear inactivation is present in the control as well, due to a desiccation process occurring to the biofilms within the first 100 min.
Regardless of the treatment, the rate of inactivation k max was in general very similar to the control (no significant differences), as can be noticed from the k max values displayed in Table 2. This means that the inactivation rate was not influenced by the light exposure, while the residual population N res was strongly influenced in some conditions. Different percentages of I max (2.5% for violet only, 25%, 75% and 100% of I max ) per LED color resulted in higher or equivalent log 10 reductions compared to the control (no light exposure). The log 10 reductions indicated that the highest reductions were obtained with violet light (in all conditions) and blue light when I max equaled 100%. Indeed, the most effective inactivation was observed when the violet array was tested on biofilms. This resulted in a log 10 reduction ≤ 6.80 CFU cm −2 , while the reduction of the control was only 3.87 CFU cm −2 (due to natural desiccation). A log 10 reduction ≤ 5.37 CFU cm −2 was induced for biofilms exposed to blue LED irradiation. Furthermore, experiments were carried out with green, yellow and red irradiation. These revealed no additional reduction compared to the control. Table 2. P. fluorescens parameters. Estimated parameters with standard errors of P. fluorescens inactivation using the Geeraerd et al. model are listed, i.e., N 0 , N res , k max , log 10 reduction and root mean squared error (RMSE) for general and selective media [62].

Sublethal Injury
The model was also fitted to the data resulting from the selective plating media. The inactivation kinetics (dashed lines) and the parameters, related to selective media and estimated from the Geeraerd et al. [62] model, are displayed in Figure 1 and Table 2. The sublethal injury of P. fluorescens was determined by plating the bacterial suspension on general (GM) and selective media (SM) to quantify the sublethal cell damage after irradiation. Preliminary experiments were carried out to check that SM did not affect the growth of the cells of the biofilms after mechanical detachment of the biofilm using a scraper. The SI curve related to the control experiments (no light exposure) showed a peak at about 50%. This corresponds to the phase-change between the log-linear phase and the tail in the inactivation curve (column 1, Figure 1). As violet exposure is tested, SI curves have a higher peak corresponding to 25% and 100% I max and a lower one corresponding to 2.5% I max in the same time frame. 75% I max is an exception, showing a plateau corresponding to the tail of the inactivation curve. Generally, the other SI curves have a similar trend compared to the control, except for 75% I max with blue and green exposure.

Inactivation Kinetics, Parameters and log 10 Reduction
For S. epidermidis biofilms, similar experiments were carried out. Figure Table 3, for both models. As in the previous P. fluorescens experiment, different irradiation conditions (2.5% for violet only, 25%, 75% and 100% of I max ) were tested for each color. For biofilms exposed to violet light, the inactivation kinetics were characterized by a log-linear inactivation phase, lasting approximately 80 to 90 min, after which a tailing phase followed ( Figure 2). The control did not undergo a desiccation process as seen with P. fluorescens. The initial population was approximately 6.84 CFU cm −2 and the residual population was almost the same (6.27 CFU cm −2 ). The standard errors of the inactivation rate, k max , demonstrate that this parameter could not be estimated accurately. Based on the parameters' estimation, the log 10 reduction was determined and had a much better accuracy. When blue, green, yellow and red LED arrays were used (in all conditions), the cells within the biofilm after 4 h of treatment were almost the same compared to the control, and the model without shoulder and without tail (log-linear regression) was used to fit the data and represent the kinetics. The rates of inactivation, k max , did not show any increase for these LED array exposures (Table 3).
A log 10 reduction of approximately 3.69 CFU cm −2 was induced by violet exposure (harshest condition) compared to approximately 0.57 CFU cm −2 of the control. Regarding biofilms exposed to other colors (blue, green, yellow and red), no substantial reduction was observed in terms of biofilm populations (Figure 2). The estimated reductions for each condition are reported in detail for both general and selective media in Table 3.

Sublethal Injury
The different evolution in SI for controls of S. epidermidis for violet light ( Figure 2a)  . With regards to SI kinetics related to violet light, the maximum SI value decreases (100%, 95% and 75% related to the plateau values) as the irradiance increases from 2.5% to 75% I max (Figure 2). This means that the stronger the irradiance, the stronger the killing effect. Indeed, in the harshest condition, the majority of the cells are dead rather than damaged. No evidence of this trend is present when I max equals 100%. Instead, SI curves lay underneath the control for blue, green, yellow and red exposures. Negligible SI values were observed for those treatments.

Inactivation by Light and Dose for 1-log 10 Reduction
To understand the mere effect of the irradiation on biofilms, without considering the effect of the desiccation that results in a reduction of the population of P. fluorescens (Section 3.2), the inactivation by light itself was calculated (N res control -N res light). This data analysis concerns the fits obtained with violet-exposed biofilms, for which the residual population was estimated through the Geeraerd et al. model [62]. In Figure 3, the population of bacteria killed employing violet light exposure (N res control -N res light) was reported as a function of the percentage of I max . Figure 3 demonstrates that light treatments with 100% and 75% I max were found to cause a reduction of approximately 3.0 log 10 and 1.5 log 10 CFU cm −2 in the population of both P. fluorescens and S. epidermidis biofilms. Whereas 25% and 2.5% I max caused a reduction of approximately 1.39 and 0.63 log 10 CFU cm −2 for P. fluorescens respectively, and for S. epidermidis, 0.59 and 0.02 log 10 CFU cm −2 , respectively.  Moreover, the dose for 1-log 10 reduction was calculated following Equation (6). It represents the required dose to produce 1-log 10 inactivation in the population of the specific bacterial strain. It was estimated for each violet-investigated condition. It is displayed for both P. fluorescens and S. epidermidis as a function of the irradiance (Figure 4). The most effective treatments (violet treatments and 100% I max with blue light on P. fluorescens) were considered in this analysis. Figure 4 reveals that the dose required for 1-log 10 reduction was ≥ 31 J cm −2 for P. fluorescens and ≥ 133 J cm −2 for S. epidermidis, with 400 nm peaking light (violet). However, P. fluorescens required only 4 J cm −2 when blue light was used. Figure 4 also displays the specific bacterial trends. Regarding P. fluorescens inactivation (Figure 4, round legend symbols), a little increase in dose for 1-log 10 reduction was required as irradiance increased. However, an opposite trend was evident in S. epidermidis data (Figure 4, squared legend symbols), where a higher irradiance resulted in a decrease of the dose for 1-log 10 reduction.

Discussion
This study investigated the antibacterial effect of a high-tech optical set up, composed of an exchangeable LED array emitting in the visible range, on P. fluorescens and S. epidermidis biofilms. The impact of several LED arrays, characterized by different wavelengths and irradiances, was examined. In this research, biofilms were grown on a plastic (polystyrene) surface to facilitate the extension of this study to real contaminated surfaces within the food, healthcare, households and hospital settings [58]. Future studies should account for different (bio)materials and assess the photodynamic inactivation impact on non-flat surfaces as well.

P. fluorescens and S. epidermidis Inactivation by LEDs
The most evident outcome, resulting from the experiments carried out, was the inactivation of the Gram-negative P. fluorescens and Gram-positive S. epidermidis biofilms, utilizing violet LED light. In contrast, the performed experiments proved neither promotion nor reduction of biofilm inactivation when green, yellow and red light exposure was employed. Blue light had an exceptional impact depending on the bacterial biofilms which were tested: it had a killing effect on P. fluorescens (with 100% of I max ) while it was harmless for S. epidermidis.

Violet Light
A previous study, conducted with violet light (405 nm LED) on planktonic bacteria, found a reduction of 1.0 log 10 within the population [52]. The bacteria undergoing the treatment were the food pathogen Staphylococcus aureus, while refrigerated (4 • C), for a treatment time of 7.5 h and a total dose of 486 J cm −2 . The same authors attributed the antibacterial effect induced by the LED light to physical damage of the bacterial wall rather than to DNA destruction: DNA was found not to be damaged [52]. Ferrer-Espada et al. who focused their work on biofilms grown using a CDC reactor, found a reduction Antibiotics 2020, 9, 171 13 of 23 of 1.88, 2.78 and 3.18 log 10 for Acinetobacter baumannii, Escherichia coli and Pseudomonas aeruginosa irradiated with 405 nm [32]. They exposed biofilms to doses of 162, 576 and 500 J cm −2 respectively, compared to 420 J cm −2 used in the current research. Inactivation was claimed by Halstead et al. as well, who investigated 400 nm light against 34 isolates and found violet light effective against P. aeruginosa and S. aureus [53]. Based on these findings and the literature, there is no doubt about the antibacterial activity of the violet light [32,40,41,43,49,51,54,56,[63][64][65][66][67]. However, the evaluation of its effectiveness and a direct comparison of the results, obtained by different studies, are more complicated. The complexity is due to the usage of planktonic suspensions rather than biofilms, different protocols used to grow biofilms, different initial populations, models used for the inactivation, exposure times, supplied doses, irradiances and, finally, specific bacterial properties and features (PSs contents and types) under investigation. In this work, the possibility to change the irradiance emitted by the set up allowed for the investigation of several irradiances. It was found that the stronger the violet irradiance became, the lower the residual population, N res , was. This was valid for both examined bacteria and it is in complete agreement with the literature focused on the UV range, where powerful sources promoted complete eradication [23]. The findings of Section 2.5 also demonstrated that inactivation (light-killed population) of P. fluorescens biofilms was stronger than S. epidermidis biofilms when the total dose irradiating the samples was lower (2.5% and 25% I max , Figure 3 in Section 2.5). It can be conceivably hypothesized that the difference between P. fluorescens and S. epidermidis inactivation may be due to the presence of different endogenous PSs within each bacterial species having different absorption spectra [38][39][40][41]47]. Another reason may be the different quantity of PSs absorbing light per bacterial cell: a higher concentration leads to increased reactive oxygen species production and, therefore, a stronger killing effect under lower irradiation conditions. The level of endogenous PSs is indeed an intrinsic characteristic of the strain. Moreover, the probability of photoexcitation of a PS is driven by the direct interaction, photons-PS. Because the two species are characterized by a different cell wall as a result of the Gram-nature, the physiological and morphological structure of the bacteria played a role as well. This aspect is discussed in Section 3.3.

Blue Light
In some works, blue light (420-470 nm) has been tested as well, showing a strong effect against both Gram-positive (Bacillus cereus, Listeria monocytogenes and methicillin-resistant S. aureus) and Gram-negative pathogens (P. aeruginosa, Salmonella Typhimurium and E. coli) in their planktonic form [39][40][41]49,50,54,56,[65][66][67][68]. However, Abana et al. claimed that the effect of 455 nm is not completely bactericidal on E. coli but depends on the microorganism (pathogenic or non-pathogenic) and growth phase. They demonstrated that the stationary phase was more resistant compared to the exponential phase [45]. Based on our findings, blue light only had an impact on P. fluorescens biofilms under the harshest condition (100% I max ). No evidence of inactivation was observed on S. epidermidis biofilms. The inactivation obtained with the blue light on P. fluorescens biofilms correlates favorably with Amin et al. and further supports the idea that P. fluorescens likely contains porphyrins absorbing specific wavelengths of the blue emission spectrum (420 nm for instance) that might promote cell death [37]. Hyun and Lee observed that P. fluorescens compared to L. monocytogenes was found to be the most sensitive strain to blue light treatment [50]. However, these authors were using a blue light peaking at 460 nm and investigating the effect on bacteria inoculated and left dry on packaging materials (no biofilms). The light provoked disruption of the cell membrane, and release of intracellular component post-treatment [50].

Green, Yellow and Red Light
The poor performances (unaltered population density) obtained with green, yellow and red light exposure in terms of inactivation were in contrast with some works that, instead, evidenced the influence of this visible range on bacteria. Some studies performed on planktonic form showed reduction or promotion on bacterial cells depending on the temperature at which the experiments were performed. Ghate et al. investigated the impact of 521 and 642 nm on four different pathogens at different temperatures (20, 15 and 10 • C) for a total exposure time of 7.5 h [49]. At 20 • C, they found an increase of the population densities (approximately 1.5 log 10 ) when the four bacteria were exposed to one of these two wavelengths. At 15 and 10 • C, instead, a reduction was observed when bacteria were irradiated at 521 nm. Specifically, 1.7, 1.7, 1.0 and 0.9 log 10 at 15 • C and 1.8, 1.7, 1.5 and 1.5 log 10 10 • C for S. Typhimurium, S. aureus, E.coli and L. monocytogenes, respectively. No variation within the population densities was found at 642 nm.
Other studies showed that the bacterial species under investigation played a crucial role on the inactivation. In fact, Kim et al. found a reduction up to 60% and 40% on S. aureus and E. coli, respectively [48]. On the contrary, they found an increase of the population of 2.5-fold after 8 h for Pseudomonas gingivalis. These experiments were performed with irradiation at 525 nm on planktonic cells. Yu and Lee, on one hand, found that green light promoted the antifungal activity of Bacillus amyloliquefaciens but, on the other hand, claimed that it reduced mycelial growth on Colletotrichum acutatum [57,58].
Then, there are other studies where a reduction is observed regardless of the two above-mentioned temperature and species. Ghate et al. studied the effect of 461 and 521 nm against the pathogens E. coli, S. Typhimurium and L. monocytogenes [46]. Treatments of 7.5 h were applied with a total dose ranging from 100 to 600 J cm −2 and led to the conclusion that 521 nm decreased the bacteria in their planktonic form. The efficiency of the green light treatment was lower than blue light but still, it caused a reduction of 1-2 log 10 CFU mL −1 at both acidic and alkaline conditions. Kumar et al. also demonstrated at 520 nm a reduction of 1.2 log 10 at 25 • C of S. aureus and 0.7 log 10 at 25, 10 and 4 • C of L. monocytogenes. Some studies performed on fungi showed similar impacts [30].
Brito Aragão et al. focused, instead, on biofilms. They tested two different red peaking LEDs with an emission range of 620-660 nm and found that irradiation alone was causing no reduction within the population [55], while Yu and Lee found enhanced biofilms' formation and motility on B. amyloliquefaciens [57]. Even though these wavelengths were chosen because they showed growth enhancing or bactericidal effects, none of these effects was observed against biofilms of P. fluorescens and S. epidermidis under different conditions. This research demonstrated the ineffectiveness of these wavelengths in terms of inactivation of two specific bacterial biofilms. Future research should be focused on the use of the examined ineffective wavelengths to promote antibacterial activity in synergy with exogenous PSs [27,44,69,70].

Sublethal Injury
Although the sublethal injury (SI) kinetics have completely different trends and percentages of injury, confirming the response to the treatments depends on the microorganism, a generalization on the light impact is possible. The SI curves for violet LEDs demonstrate that for both microorganisms, the percentages of SI increase when the irradiance increases. This can be easily explained: the number of photons reaching the biofilms is higher and consequently, so is the sublethal injury. However, the tendency of the SI is slightly different compared to what was found in other research. Ghate et al. determined that the SI only increased with the exposure time [46]. In contrast, in this study, it reaches a plateau or a peak. This is probably due to the difference in population growth forms: planktonic compared to biofilms.

Factors Influencing the Inactivation Kinetics
Several factors influenced the kinetics of the photoinactivation: (i) the Gram-nature and the intrinsic difference within the developed biofilms belonging to the two species, (ii) the endogenous PSs within microorganisms, (iii) the direct impact on enzymes and (iv) the production of specific molecules by microorganisms.
First of all, regarding the Gram-nature, an important result that emerged from this study is that Gram-negative P. fluorescens and Gram-positive S. epidermidis biofilms, responded with the same photoinactivation to a treatment with violet light at the two highest irradiances (100% and 75% of I max ): for both species, populations were inactivated with 3 log 10 and 1.5 log 10 without any significant differences between the two microorganisms (Section 2.5). These findings are in contradiction with previous studies reported in the literature, where some authors confirmed a stronger inactivation effect on Gram-positive microorganisms, which have a different cell wall lacking both outer membrane and lipopolysaccharide layer but have a thicker peptidoglycan layer [51]. They are also in contrast with others claiming that the most sensible bacteria are the Gram-negative ones, even though they are characterized by a more complex cell wall [53]. Halstead et al. found that among 34 isolates, the most susceptible were the Gram-negative bacteria A. baumannii ACI_19606, for which a 93.5% reduction was observed after 15 min of light exposure [53]. The intrinsic differences between the two types of bacteria (Gram-nature) and the biofilm they formed is also an interesting aspect. A comparison between the curves of the controls for P. fluorescens and S. epidermidis showed how the biofilms, developed using the two strains, reacted differently under drying conditions. P. fluorescens underwent a process of desiccation that can justify the reduction in the biofilm population (no light-exposure), while the S. epidermidis biofilms did not show any reduction. Gram-negative bacteria seemed more sensitive to drying than Gram-positive, as previous authors confirmed [15]. The comparison between the kinetics of P. fluorescens and S. epidermidis requires further comments. Although the inactivation kinetics are alike in shape, they are slightly different regarding the time frame of the log-linear phase, which was shorter for S. epidermidis compared to P. fluorescens.
Then, based on the literature that focused on the investigation of the endogenous PSs responsible for the photoinactivation and considering the absorption spectra of the porphyrins and flavins, it is possible to claim that the wavelengths with the strongest antibacterial effect are the ones belonging to the main absorption peak: violet-blue range [38][39][40][41]47]. Although the other wavelengths of the visible spectra followed in correspondence with the secondary absorption peaks of the same chromophores, they were harmless [38][39][40][41]47].
Another factor that can be involved is the direct damage caused by photons to the enzymes succinate dehydrogenase (SDH) and lactate dehydrogenase (LDH). This process does not involve reactive oxygen species production [71]. The damage mechanism is due to the tail of the violet LED spectrum continuing in the UV-A spectral region, which is responsible for their destruction [71]. Finally, the last factor that may play a role in the detrimental effect of light on P. fluorescens biofilms is its effect on the production of pyocyanin or pyoverdine (siderophores), blue extracellular products that may contribute to light absorption [67,72].

Doses and Bacterial Trends
The analyses of the dose required for 1 log 10 reduction highlighted the impact of violet or blue exposure on P. fluorescens and S. epidermidis biofilms as antibacterial treatment, compared to the other ineffective wavelengths (blue, green, yellow and red). Concerning P. fluorescens, even better inactivation was achieved with blue light rather than violet (Figure 4). A remarkable difference in the dose required for 1 log 10 reduction between blue and violet light was observed. The blue dose required to reduce the bacteria population in the biofilm was found to be 8 times higher compared to the violet dose (Section 2.5). A lower dose means lower energy requirement per area and translates into lower energy consumption and lower costs, thus representing an advantage as new technology.
The examination of the bacterial trends (dose for 1 log 10 reduction as a function of irradiance) demonstrated interesting features. It was found that the dose for 1 log 10 reduction increases asymptotically as the irradiance increases for P. fluorescens biofilms, whereas it decreases asymptotically as the irradiance increases for S. epidermidis biofilms. Tomb et al. observed the opposite trend for P. fluorescens (required dose decreases as a function of irradiance), although S. epidermidis' trend is in line and practically the same as the one proposed by the same authors [67]. It was hypothesized that the opposite trend might be due to a saturation effect. The excess of irradiance does not enhance the number of excited endogenous PSs within P. fluorescens cells that might be previously promoted to an excited molecular state [73].
The presence of a tail in the kinetics, obtained using the Geeraerd et al. model, made clear that the inactivation with violet light after 180 min can no more and no longer increase the number of dead cells [62]. An increase in the exposure time after a certain cut off value is useless and does not promote any additional activation. This is an important aspect that is first different in planktonic form where cells are free to float [46], compared to the situation where the cells are trapped within the matrix (biofilm). Moreover, the exposure time in this work is an average between the long treatments, such as the 4 days of exposure tested by Hyun et al. against P. fluorescens and L. monocytogenes [50] and the short exposures, like Rupel et al. who used treatment times of a few minutes. The latter used a powerful laser source at 455 nm, assuring an effect on a short time but using a light source characterized by high energy consumption and cost compared to LEDs [74].

Biofilm Growth
Bacterial biofilms were grown on the polystyrene surface of petri dishes (50 mm diameter, 8 mm height, Simport, Canada). 400 µL of inoculum suspension, of approximately 10 7 CFU mL −1 , was spread on a circular area (7.0 cm 2 ) with a plastic loop. After inoculation, the petri dishes were closed and left to grow for 24 h at a constant temperature of 25 • C (P. fluorescens) and 20 • C (S. epidermidis), under sterile conditions.

Rinsing Procedure
After 24 h, the mature biofilms had grown at the solid-liquid interface. The excess of bacterial suspension was removed using a pipette. 3 mL of Phosphate Buffer Saline (PBS, Sigma Aldrich, Saint Louis, MO, USA) solution was poured into the petri dish and then gently shaken. Finally, the PBS was carefully removed. The rinsing procedure was repeated twice. The biofilm was then ready to be treated using light or for quantification.

Light Set up and Optical Conditions
An array composed of 120 LEDs was used as the light source for the investigation. The violet LEDs (UVC5TZ-400-30) were purchased from Bivar (California) while the blue (L-53MBD), green (L-1503CGDK), yellow (L-53YD) and red (L-53HD) were purchased from Kingbright (Taiwan). All of them had an emission in the visible region of the light spectrum. They were optically characterized in terms of wavelength, irradiance and emission in time. The setup was provided with a mechanism that allowed for the replacement of the LED array. A liquid crystal display allowed the operator to change the electric current flowing through the LEDs (0-30 mA), and thus, the power density emitted by the light source. Different conditions were investigated for each color: 100%, 75% and 25% of the maximum irradiances (I max ) were tested on the bacterial biofilms. For the violet LED array, 2.5% was explored as well, to obtain an irradiance comparable to the other LED sources. The violet-emitted irradiance was measured using a radiometer RM-12 and sensor (UVA+ 339-440 nm) from Opsytec Dr.
Grobel (Germany). The spectrometer MF250N from UPRtek Corp. (Taiwan) was used for the other LED colors. The total dose for each condition was calculated as follows [51]: (1) where t TOT represents the total exposure time equal to 4 h. It represents the total energy delivered for the irradiated area. Figure 5 displays a scheme of the set up (Figure 5a), two pictures of the set up (top, Figure 5c, and gross section, Figure 5d) and the LEDs matrices (Figure 5b) used during the experiment.
in terms of wavelength, irradiance and emission in time. The setup was provided with a mechanism that allowed for the replacement of the LED array. A liquid crystal display allowed the operator to change the electric current flowing through the LEDs (0-30 mA), and thus, the power density emitted by the light source. Different conditions were investigated for each color: 100%, 75% and 25% of the maximum irradiances (Imax) were tested on the bacterial biofilms. For the violet LED array, 2.5% was explored as well, to obtain an irradiance comparable to the other LED sources. The violet-emitted irradiance was measured using a radiometer RM-12 and sensor (UVA+ 339-440 nm) from Opsytec Dr. Grobel (Germany). The spectrometer MF250N from UPRtek Corp. (Taiwan) was used for the other LED colors. The total dose for each condition was calculated as follows [51]: Total dose = Irradiance · tTOT (1) where tTOT represents the total exposure time equal to 4 h. It represents the total energy delivered for the irradiated area. Figure 5 displays a scheme of the set up (Figure 5a), two pictures of the set up (top, Figure 5c, and gross section, Figure 5d) and the LEDs matrices (Figure 5b) used during the experiment.

Biofilm Treatments
Mature biofilms were treated for different light exposure times: 30, 60, 90, 120, 150, 210 and 240 min. The distance array-samples equaled 6.5 cm and the temperature during the whole experiment was kept constant at 20 • C by placing the set up inside an incubator. The biofilms were placed open underneath the LED array. The control referred to biofilms in the dark that did not undergo the light treatment. The control biofilms were placed (open) inside the incubator and could be dried by the air.

Biofilm Quantification
After the treatment, 2 mL of PBS solution was added to each biofilm. The biofilm was detached from the surface using a cell scraper (20 mm blade, Carl Roth, Karlsruhe, Germany). Afterwards, the suspension was homogenized with a pipette, it became overall turbid and no pellicles were floating in it. Serial dilution suspensions were prepared in saline solution (0.85% NaCl) and these suspensions were plated both on general medium (GM) and on selective medium (SM). Tryptic Soy Agar (TSA) and CFC-supplemented-Pseudomonas Agar Base (Cetrimide-Fucidin-Cephalosporin-PAB, Thermo Scientific, Waltham, Massachusetts, United States) were used as GM and SM for P. fluorescens. While for S. epidermidis, brain heart infusion (BHI) supplemented with 13 g L −1 of bacteriological agar (Merck, Darmstadt, Germany) and Staphylococcus agar containing 75 g L −1 NaCl were used as GM and SM, respectively.

Inactivation Model
The Geeraerd et al. inactivation model was used to fit the light-response data [62]. This non-linear model is based on a sigmoidal-like curve with a shoulder (lag phase), a log-linear inactivation and a tail (indicating the existence of a residual population surviving the inactivation treatment). The data collected in this work did not show any evidence of a shoulder, so the model was simplified to the model without shoulder expressed by the following (static) equation: or in some cases, to the model without shoulder and without tail, which was reduced to the classical log-linear regression model: In these equations, N(t) (CFU cm −2 ) represents the biofilm microbial population at time t (min), N 0 (CFU cm −2 ) represents the initial cell density, k max (min −1 ) is the maximum inactivation rate and N res (CFU cm −2 ) is the biofilm residual population. The difference between the initial population (N 0 ) and the residual population (N res ) is referred to as log 10 reduction. During parameters estimation, logarithmically transformed data are used. The root mean squared error (RMSE) between the observed (N exp ) and predicted (N mod ) log 10 values was used to evaluate the goodness of the fit, where n is the number of measurements and p is the number of estimated parameters.

Sublethal Injury
The sublethal injury (SI), consequent to light exposure, was calculated using the following equation [75]: Sublethal Injury(%) = Cell Counts on GM − Cell Counts on SM Cell Counts on GM × 100%.
SI is calculated based on the continuous fit of the Geeraerd et al. model [62]. It gives a percentage related to the damage caused by light exposure.

Dose for 1 log 10 Reduction
The estimation of the parameter N res of the control (no light exposure) and of the light-exposed biofilms allowed for the calculation of the light dose (J cm −2 ) required to reduce the bacterial population with 1 log 10 (90%). The required dose gave an idea of the strength of the killing potential of the light treatment. The dose for 1 log 10 reduction has been calculated as: Dose for 1 log 10 red = Total dose N res control − N res light (6) where N res was estimated by the fitting of the Geeraerd et al. model [62].

Statistical Analysis
To statistically compare the estimated parameters evaluated using the Geeraerd et al. model, a one-way analysis of variance (ANOVA) was carried out using a multiple comparison tests based on the multcompare function implemented in MATLAB R2016a (The Math Works, Inc.) [62]. A confidence level of 95% was used to assess the significant difference. A p-value less than 0.05 was considered significantly different.
Previous research on the antimicrobial potential of visible light has focused mainly on bacteria growing in a planktonic form and to a lesser extent, on biofilms, which have been identified as a crucial problem for hygiene control in households, industries and clinical environments in the last years. Therefore, this research focused on the effect of five different wavelengths of visible LED light on the inactivation dynamics of mature biofilms. Moreover, this research included both a Gram-negative and Gram-positive bacterial strain, as the Gram-nature is seen as an important factor in the resistance against antimicrobial treatments. This work adds to a growing body of research on the effect of photodynamic inactivation against biofilms. The broad study explores the effect of several wavelengths and irradiance levels on common biofilm-formers with a different Gram-nature. As such, it provides insights into (the limits of) the potential of LED light as an antibiofilm strategy.
The results demonstrated that the high-energy violet light was able to inactivate biofilms of both the Gram-negative P. fluorescens and Gram-positive S. epidermidis. Blue light also had an inactivating effect on P. fluorescens but not on S. epidermidis. This demonstrates the higher resistance of the Gram-positive biofilms. Green, yellow and red light had no effect on the biofilm population compared to untreated control samples. Even though biofilms are more antimicrobial-resistant than planktonic bacteria, this research demonstrated the potential of simple visible light as an anti-biofilm strategy. However, the research also demonstrated that this potential is limited to violet light when considering a treatment that should be effective against a broad spectrum of bacteria.
Bacterial inactivation by photons is hypothesized to be mediated by reactive oxygen species. Future work should focus on improving the understanding of the inactivation mechanism of visible light. Such an improved understanding will allow further finetuning of this promising technology.