Antimicrobial efficacy and inactivation kinetics of a novel LED-based UV-irradiation technology

Background Ultraviolet (UV)-light-emitting diodes (UV-LEDs) are energy efficient and of special interest for the inactivation of micro-organisms. In the context of the coronavirus disease 2019 pandemic, novel UV technologies can offer a powerful alternative for effective infection prevention and control. Methods This study assessed the antimicrobial efficacy of UV-C LEDs on Escherichia coli, Pseudomonas fluorescens and Listeria innocua, as well as severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), human immunodeficiency virus-1 (HIV-1) and murine norovirus (MNV), dried on inanimate surfaces, based on European Standard EN 17272. Results This study found 90% inactivation rates for the tested bacteria at mean UV-C doses, averaged over all three investigated UV-C wavelengths, of 1.7 mJ/cm2 for E. coli, 1.9 mJ/cm2 for P. fluorescens and 1.5 mJ/cm2 for L. innocua. For the tested viruses, UV doses <15 mJ/cm2 resulted in 90% inactivation at wavelengths of 255 and 265 nm. Exposure of viruses to longer UV wavelengths, such as 275 and 285 nm, required much higher doses (up to 120 mJ/cm2) for inactivation. Regarding inactivation, non-enveloped MNV required much higher UV doses for all tested wavelengths compared with SARS-CoV-2 or HIV-1. Conclusion Overall, the results support the use of LEDs emitting at shorter wavelengths of the UV-C spectrum to inactivate bacteria as well as enveloped and non-enveloped viruses by exposure to the appropriate UV dose. However, low availability and excessive production costs of shortwave UV-C LEDs restricts implementation at present, and supports the use of longwave UV-C LEDs in combination with higher irradiation doses.


Introduction
Light-emitting diodes (LEDs) are a well-established light source based on semiconductors used for many technical and scientific purposes [1]. Due to their intrinsic properties regarding optical features, energy efficiency and lifetime, LEDs have many advantages compared with conventional light sources [2]. LEDs are easy to control; they provide a narrow, well-defined spectral power distribution; and they have a mechanically compact structure [3]. Therefore, LED technology enables the design and implementation of tailor-made lighting set-ups that provide not only optimized irradiation conditions, but are also energy efficient and sustainable [4]. The utilization of ultraviolet (UV) light, mainly UV-C light in the range between 100 and 280 nm, for disinfection is a well-known approach, and is used for water treatment, food processing and in clinical areas [5e7]. With rapid advances in the semiconductor industry, the development of LEDs emitting in the UV-C range was just a matter of time [8]. LEDs with a peak wavelength close to the absorption maxima of DNA and RNA, which is approximately 265 nmKlicken oder tippen Sie hier, um Text einzugeben. [9], are of particular interest for the inactivation of micro-organisms. Due to the ongoing coronavirus disease 2019  pandemic, the potential of UV light, especially UV-C LED light, to inactivate microorganisms on surfaces or in aerosols was discovered [10,11]. During a crisis, raw materials become particularly scarce, and disinfectants and hand sanitizers were in short supply at the beginning of the COVID-19 pandemic [12]. Under these challenging conditions, novel UV technologies can offer a powerful alternative for effective infection prevention and control [13]. Substantial progress in semiconductor research has enabled the development of economic and environmentally sustainable disinfection systems based on mercury-free UV-LED technology [14]. If infections are caused by surface contact, the micro-organisms involved are usually found in a dried or semi-dried state. Although contact infection is a crucial pathway, many studies have focused on the impact of UV radiation on micro-organisms in solution. Therefore, information about the kinetics of dried micro-organisms on surfaces is limited, and mainly focuses on traditional UV-C light sources with an emission wavelength at 254 nm [15,16]. Furthermore, it is unclear whether inactivation data from liquids can be applied directly to surface inactivation [17].
This study investigated the inactivation kinetics of the bacterial test organisms Escherichia coli, Pseudomonas fluorescens and Listeria innocua, and viruses such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), human immunodeficiency virus 1 (HIV-1) and murine norovirus (MNV), dried on inanimate surfaces, by UV-C LED light based on European Standard EN 17272. Furthermore, the effects of various wavelengths in the UV-C (255, 265 and 275 nm) and UV-B (285 nm) ranges on inactivation were studied. The study findings contribute to the development of energy efficient, economically feasible and eco-friendly UV-LED-based disinfection systems.

UV-LED irradiation set-up
All UV-C and UV-B irradiation experiments were conducted using modular apparatus ( Figure S1A, see online supplementary material). Due to the modular design, the different wavelengths could be applied by exchanging irradiation modules equipped with the corresponding UV-LEDs. A detailed description of the irradiation set-up can be found elsewhere [18]. UV-LEDs with central wavelengths of approximately 255, 265, 275 and 285 nm were applied in this study, and the spectral power distributions of all four UV-LEDs used, including peak wavelength and full width at half intensity, were measured ( Figure

Inactivation experiments
Inactivation experiments were designed based on European Standard EN 17272 [19]. Gram-negative E. coli (DSM 6367) and P. fluorescens (DSM 50090), and Gram-positive L. innocua (DSM 20649) were used as bacterial test organisms. Virucidal efficacy testing was performed with enveloped SARS-CoV-2 (2019-nCoV/Italy-INMI1 NR-52284) and HIV-1 (R9-BaL and Yu2B), as well as non-enveloped MNV (MNV-1.CW3). A defined volume of 50 mL of microbial or viral culture at an initial concentration of 10 9 colony-forming units (CFU) per mL (bacteria), 10 7 plaqueforming units (PFU) per mL (SARS-CoV-2 and MNV), or 1 mg of viral envelope protein p24 per mL (HIV-1) was applied to the centre of the stainless steel (1.4301) test specimen, and distributed evenly over an area of 2.0AE0.2 cm 2 of the carrier (40mm diameter) using a sterile inoculation loop, resulting in an inoculum of 5,10 6 CFU per mm 2 (bacteria), 5,10 4 PFU per mm 2 (SARS-CoV-2 and MNV) and 5,10 À3 mg per mm 2 . The loaded carriers were dried in a laminar flow cabinet until completely evaporated, but not >2 h, followed by exposure to the designated UV dose. After UV irradiation, bacteria/viruses were recovered from the carriers by repeated addition of respective culture medium without supplements. For bacteria, irradiated samples were serially diluted (10-fold), and spread on the respective agar plates before incubation for 24 h and CFU determination. In the case of SARS-CoV-2 and MNV, serial dilutions (10-fold) were used as the inoculum for a viral plaque assay, which was incubated for 72 h before staining and plaque counting [19,20]. UV-mediated inactivation of HIV-1 was quantified by incubating the recovered virus dilutions with human peripheral blood mononucleated cells for 7 days, followed by p24 enzyme-linked immunosorbent assay [21]. Next to UV-exposed samples, a non-irradiated control was prepared and evaluated with each experimental run (Figure S1B, Step 3, see online supplementary material). The resulting CFU for bacteria, PFU for SARS-CoV-2 and MNV, or p24 concentration for HIV-1 of the control N 0 , and the resulting CFU for bacteria, PFU for SARS-CoV-2 and MNV, or p24 concentration for HIV-1 of the irradiated N D were used to calculate corresponding survival rates N D /N 0 for bacteria or normalized activity N D /N 0 for viruses. The experimental approach is depicted in Figure S1B (see online supplementary material), and a more detailed description of the inactivation experiments is provided in the online supplementary material.

Data analysis and statistics
From the resulting survival rates and normalized activities, N D /N 0 , the inactivation rate in percent was calculated by (1-N D /N 0 ),100%. To determine the UV dose, D 90 in mJ/cm 2 , at which 90% of the micro-organisms were inactivated, relative inactivation rate curves were calculated with CFU (bacteria), PFU or p24 concentrations (viruses) by a normalized fourparameter non-linear regression model in GraphPad Prism v.9. Statistical analyses were performed using analysis of variance for normally distributed datasets, or KruskaleWallis test with Dunn's multiple correction for non-normally distributed datasets. 8

Bacterial inactivation via UV radiation
Inactivation of E. coli, P. fluorescens and L. innocua using 255, 265, 275 and 285 nm at various UV doses (ranging from 0 to 50 mJ/cm 2 ) is depicted in Figure 1. The greatest reductions in E. coli survival rates were observed at 255 and 265 nm at low UV doses, while increased UV doses were required for similar results at 275 and 285 nm ( Figure 1A). Calculation of D 90 values demonstrated that a UV dose <1.0 mJ/cm 2 for 255 and 265 nm, 4.0 mJ/cm 2 for 275 nm, and <8.5 mJ/cm 2 for 285 nm resulted in a 10% bacterial survival rate ( Figure 1B). For P. fluorescens, similar survival rates were observed for all tested wavelengths, and analyses of D 90 values further demonstrated a reduction in bacterial growth at UV doses of 1.7 mJ/cm 2 , 1.8 mJ/cm 2 , 2.3 mJ/cm 2 and 1.5 mJ/cm 2 for 255, 265, 275 and 285 nm, respectively ( Figure 1C, D). In the case of L. innocua, the strongest decrease in survival rate was found for 255 nm. This observed bactericidal efficacy weakened with increasing wavelength ( Figure 1E). D 90 analyses ranged from 0.34 mJ/cm 2 at 255 nm up to 5.7 mJ/cm 2 at 285 nm ( Figure 1F). UV-dosedependent normalized inactivation for E. coli, P. fluorescens and L. innocua, including non-linear regression curves and calculated D 90 values, is shown in Figure S2 (see online supplementary material).

Viral inactivation via UV radiation
UV exposure of SARS-CoV-2 at 255 and 265 nm led to a greater reduction in normalized activity at lower UV doses compared with UV treatment at 275 nm or 285 nm (Figure 2A). In addition, calculation of D 90 values demonstrated a 90% inactivation rate at UV doses <15 mJ/cm 2 for 255 and 265 nm, while much higher doses were needed for 275 and 285 nm ( Figure 2B). For 275 nm, the dose required to achieve 90% inactivation was almost five-fold higher compared with 265 nm ( Figure 2B). The UV dose required to achieve 90% inactivation at 255 nm was two-fold higher for HIV-1 compared with SARS-CoV-2 ( Figure 2B, D). Although similar D 90 doses were detected for both enveloped viruses using 265 nm, 90% inactivation of HIV-1 was achieved using 13.9 mJ/cm 2 and 15.2 mJ/cm 2 for 275 and 285 nm, respectively ( Figure 2C,D). Much higher UV doses were needed for all tested wavelengths to reduce the normalized activity of MNV compared with SARS-CoV-2 or HIV-1 ( Figure 2E). D 90 analyses revealed that best inactivation was obtained at 70.0 mJ/cm 2 at 255mm , while inactivation values reached up to 124 mJ/cm 2 for the higher wave lenghts ( Figure 2F).

Comparison of D 90 values within bacterial or viral strains or wavelengths
Analyses of D 90 values, irrespective of wavelength, within all tested bacteria revealed no significant differences, while significant differences were found between the two tested enveloped viruses (SARS-CoV-2 and HIV-1) and the nonenveloped MNV ( Figure 3A,C). In contrast, comparison of normalized inactivation values by wavelength clearly demonstrated an advantage of the use of shorter UV wavelengths for inactivation of bacterial growth ( Figure 3B). This advantage of using shorter UV wavelengths was also observed for the inactivation of non-enveloped, but not enveloped, viruses ( Figure 3D,E).

Discussion
Although many UV inactivation kinetic studies of bacteria and viruses can be found in the literature, most have focused on micro-organisms in solution (e.g. water or culture medium), so reliable data on UV dose dependence, such as D 90 values, are rare for surface UV disinfection [16,22].
For the bacteria in the present study, 90% inactivation occurred at a mean UV-C dose, averaged over all three investigated UV-C  [26] for P. fluorescens, and from 0.8 mJ/cm 2 to 3.5 mJ/cm 2 for L. innocua [23] depending on experimental design and the strain used. These findings differed with the UV source and the bacteria environment during irradiation, but the results are similar to the present findings. A study in liquids with low-pressure mercury lamps (254 nm) demonstrated that viruses such as SARS-CoV-2 can be inactivated at 550 mW/cm 2 for 30 s using 254 nm [27]. In addition, the authors confirmed that UV-C irradiation induced genome damage but did not affect viral morphology and proteins [27].
Comparing the results within the bacteria strains, no significant differences in D 90 values were found between the Gram-positive and Gram-negative bacteria, which is initially contradictory to the literature [26]. Due to differences in cell wall composition, Gram-positive bacteria should be less sensitive to UV irradiation than Gram-negative bacteria [28]. However, these previous studies were performed with microorganisms in solution, and considered higher log-reductions [28,29]. In the present study, the difference in D 90 values between Gram-positive and Gram-negative bacteria may not be prominent due to the lower UV dose regime. Furthermore, the present study was conducted using dried bacteria on stainless steel surfaces, which does not reflect the in-vivo situation. These conditions may also have reduced bacterial sensitivity to UV irradiation [23,29].
Regarding the UV wavelength, a significant difference in D 90 values was found between UV-B and the three tested UV-C wavelengths, while no significant difference was found between the three UV-C wavelengths. Previous studies with E. coli in solution showed D 90 levels of approximately 8 mJ/cm 2 at a UV-B wavelength of 285 nm [30] and 3 mJ/cm 2 at a UV-C wavelength of 265 nm [31], which is in agreement with the present study. In addition, similar findings were reported for virus inactivation in liquid suspension [13]. UV irradiation of SARS-CoV-2 using UV-B was also less effective for virus inactivation compared with UV-C [13]. One explanation for this wavelength-dependent difference is the absorption spectra of DNA/RNA, which have their maximum values in the UV-C range at 260 nm [29], and a main inactivation mechanism is direct DNA/RNA damage [32]. Recently, Sobotka et al. showed inactivation of human coronaviruses by UV-C LED technology at 275 nm, which is consistent with the present data [33]. A study in liquids with low-pressure mercury lamps (254 nm) demonstrated that non-enveloped viruses required higher doses for inactivation compared with enveloped viruses [34]. The present study also compared enveloped and non-enveloped viruses to test inactivation efficacy using UV-LEDs, and the results are in accordance with these findings. A possible explanation for the higher sensitivity of enveloped viruses may be reactive oxygen species (ROS) induced by UV-C irradiation causing lipid peroxidation in the external lipid bilayer membrane [34]. Although this inactivation by ROS production was also observed in bacteria using UV-B [35], the D 90 values do not reflect increased inactivation at 285 nm. Lower UV dose exposure and testing on stainless steel surfaces could be causative factors. Of note, the effects of material-based inhibition could be excluded in this study, as UV-C treatments on stainless steel discs lasted for <1 h [36,37]. Overall, the study data support the use of shorter wavelengths of the UV-C spectrum to inactivate bacteria, enveloped viruses and non-enveloped viruses. However, LEDs emitting at wavelengths of 255 and 265 nm are currently not commercially available due to excessive production costs. Therefore, the present study suggests the use of higher wavelengths in the UV-C spectrum (e.g. 275 nm for clinical applications), which requires higher irradiation doses to achieve sufficient microbial inactivation. Potential clinical applications of UV-C LED technology could be the decontamination of medical tools or surfaces in critical areas, such as hospital beds in intensive care units [38].The presented modular system could serve as a design template for possible commercial systems as the UV-LED modules are easily exchangeable. In addition, a scale-up is feasible due to the detailed measurements of the irradiance in combination with optical simulations [39]. Nevertheless, further studies and adjustments are needed to optimize the UV-C-mediated inactivation of all pathogens, as UV-C disinfection was shown to be more efficient against viruses than bacteria in this study.