Ultraviolet (Spot)light on Water Treatment: Targeting Inactivation Efficiency and Stress Responses of Gram-Positive and Gram-Negative Bacteria Using UV-B and UV-C LEDs

Abstract

This study examines the inactivation kinetics and stress responses of Gram-positive and Gram-negative waterborne bacteria using ultraviolet (UV)-B and UV-C LEDs at varying fluences. Our findings show that UV-light-emitting diodes (LED) treatment effectively inactivates both bacterial types, achieving over 4-log reductions at 255 nm and 285 nm wavelengths. Notably, inactivation rates at 285 nm, especially at higher fluences, are comparable to or exceed those at 255 nm. Additionally, UV-B treatment at 285 nm requires shorter exposure times for the same UV dose due to its deeper penetration into water and higher radiant flux. Stress responses varied between species: 255 nm exposure caused more direct DNA damage, triggering the SOS response with recA upregulation, particularly in Gram-positive L. innocua; while 285 nm exposure primarily induced oxidative stress, leading to soxS upregulation, especially in Gram-negative bacteria. These results suggest that UV-B complements UV-C effects by causing reactive oxygen species (ROS) formation in addition to DNA damage, challenging DNA repair. Given the higher cost of UV-C LEDs, our results support the optimization of water treatment systems using UV-B LEDs, which is a promising approach for improving bacterial inactivation while reducing exposure time and energy use.

1. Introduction

According to the World Health Organization (WHO 2023), at least 1.7 billion people lack access to safe drinking water worldwide, with waterborne diseases like cholera, diarrhea, or hepatitis A causing millions of deaths annually. Inactivation of microorganisms using ultraviolet light has gained increasing prominence as a non-thermal technology for microbial inactivation across diverse settings like water, air, surfaces, and food products without compromising their quality or organoleptic properties [1,2]. Ultraviolet-light-emitting diodes are increasingly preferred over conventional mercury lamps due to their eco-friendly profile, presenting a sustainable alternative to chemical disinfection methods such as chlorination or ozonation, as well as to costly and complex membrane filtration [3]. Major advantages of UV-LEDs include their adjustable design (offering different wavelengths and tunable intensity), extended lifespan, rapid start-up, minimal disposal costs, and on-demand operation without chemical additives or harmful genotoxic by-products [4,5,6]. Additionally, UV disinfection devices can be effectively utilized on a small scale for point-of-use applications. The substantial differences between traditional mercury lamps and UV-LEDs regarding emission spectra and radiant flux, however, necessitate accurate characterization of their disinfection performance to evaluate the applicability of UV-LEDs for broad industrial implementation [2,3].

The UV region covers the wavelength range of 100–400 nm and is divided into three distinct bands: UV-A (315–400 nm), UV-B (280–315 nm), and UV-C (100–280 nm). Although the germicidal effect of UV light has been widely studied, the UV resistance of different bacterial species and their intra- and interspecific variability have been largely neglected [7]. Additional critical factors affecting the efficacy of UV treatment include the physicochemical characteristics of the treatment medium, treatment temperature, and the physiological state of cells such as growth phase and environmental stress history prior to irradiation [7]. The spectral sensitivity might not necessarily follow the DNA absorbance spectrum and is further dependent on the applied wavelength and UV dose, causing UV response and stress mechanisms to vary among different microorganisms [8,9]. Previous studies have predominantly focused on Gram-negative Escherichia coli strains, either as prevalent fecal indicators for water pollution [4] or common foodborne pathogens. However, limited research has investigated potential differences in the physiological responses, inactivation kinetics, and UV resistances of Gram-negative and Gram-positive organisms, which may react very differently. In this context, another Gram-negative coliform species, Citrobacter freundii, commonly found in water, sewage, soil, and food, as well as in biofilms that develop in pipelines, cooling towers, and heat exchangers [10,11], has been far less investigated. More importantly, C. freundii exhibits a growing incidence of multidrug resistance, highlighting the need for effective alternative inactivation approaches [12,13,14]. Among Gram-positive Bacteria, Listeria spp. are fairly ubiquitous in nature [15], with L. innocua being the most frequently isolated and encountered non-pathogenic Listeria species, found in aquatic environments like rivers, brackish water, and urban wastewaters [16,17]. More importantly, Listeria spp. can form biofilms in tap water, leading to potential cross-contamination of food products and increased resistance due to reduced UV penetration [18,19]. Listeria spp. are often used as indicator organisms for environmental monitoring as they are able to withstand low temperatures, high salt concentrations, and high pH values [20]. L. innocua serves as a surrogate organism for the foodborne pathogen L. monocytogenes as they share similar genetic and physiological traits [21], and it has recently gained scientific interest due to its developing tolerance toward various abiotic stresses (e.g., dehydration) and hypervirulence genetically identical to that of L. monocytogenes [22]. Using whole-genome sequencing, Mufana et al. [22] demonstrated the changing characteristics of L. innocua, particularly its growing antimicrobial resistance to quaternary ammonium compounds found in many common disinfectant products. In this context, UV light has progressively gained scientific acclaim as an advanced disinfection technology enabling the inactivation of specific resistance genes and eliminating horizontal gene transfer activities [6].

Since different biomolecules (e.g., DNA, proteins, and lipids) absorb UV radiation at specific wavelengths, the variability in cellular targets affects the efficiency and required dose of each wavelength. In a preliminary study, we assessed the microbial inactivation efficiency of UV-LEDs at 255 nm, 265 nm, 275 nm, and 285 nm. We observed distinct differences in inactivation kinetics between UV-C and UV-B irradiation, suggesting different mechanisms of inactivation, such as DNA damage versus reactive oxygen species formation [23]. For enhanced understanding, we investigated UV-LED inactivation efficiencies and induced stress responses under the influence of both UV-B and UV-C irradiation (targeting different cellular components), along with increasing UV doses (also known as fluence), using three representative Gram-positive and Gram-negative (varying cell wall structure) indicator microorganisms: E. coli, C. freundii, and L. innocua. The two central wavelengths were specifically selected for their distinct effects on cellular components: 255 nm (UV-C), which is close to the DNA absorbance peak and highly effective at inactivating microorganisms primarily through direct DNA damage; and 285 nm (UV-B), which is close to the microbial protein absorbance spectrum and complements UV-C’s effects through additional DNA damage, protein denaturation, and ROS formation. Generated ROS include superoxide anion (O₂•⁻), hydrogen peroxide (H₂O₂), and hydroxyl radical (•OH) that can oxidize amino acid side chains, leading to structural changes and loss of function, and further induce the formation of carbonyl groups on protein side chains, resulting in protein carbonylation. Alogside ROS, other oxidative and nitrosative species like hypochlorous acid (HOCl) and nitrogen dioxide (NO₂) contribute to microbial inactivation and cellular damage by affecting protein stability, activity, and interactions [24]. Therefore, we combined traditional culture-dependent methods (CFU) with reverse transcriptase quantitative PCR (RT-qPCR), targeting selected genes such as recA (DNA repair) and soxS (oxidative stress) to broaden our understanding of targeted disinfection mechanisms as a function of UV wavelength and UV dose. This approach is essential for designing efficient and tailored UV-LED-based inactivation strategies for broad industrial applications such as water treatment.

2. Materials and Methods

The experimental protocol consisted of four main phases illustrated in Figure 1: (1) inoculating glycerol stock of challenge microorganisms onto solid medium; (2) culturing overnight and adjusting bacterial concentration using a McFarland standard [25] before transferring to new petri dishes; (3) subjecting samples to UV irradiation at various wavelengths and doses, also including positive (inoculated, non-radiated) and negative (non-inoculated, sterilized) controls; and (4a) assessing inactivation kinetics by spreading treated samples on agar plates; followed by (4b) analyzing gene expression via RT-qPCR to evaluate selected genes involved in DNA repair and oxidative stress response.

2.1. Bacterial Strains and Cultivation

Escherichia coli (DSM 6367, DSMZ, Braunschweig, Germany), Citrobacter freundii (ATCC 8090, ATCC, Manassas, VA, USA), and Listeria innocua (DSM 20649, DSMZ, Germany) were used as representative Gram-negative and Gram-positive bacteria in the present investigation. E. coli and C. freundii were both cultivated in Tryptic Soy broth (Sigma Aldrich, St. Louis, MO, USA) and incubated overnight at 30 °C and 110 rpm. L. innocua was cultivated in Tryptic Soy broth supplemented with 0.6% [w/v] yeast extract (Sigma Aldrich, USA) and incubated overnight at 30 °C and 110 rpm. Subsequently, the cells were washed three times with 0.9% NaCl (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) via centrifugation at 4600 rpm for 10 min. To achieve initial concentrations of each microorganism of approximately 1.9 × 10⁸ CFU mL⁻¹, the optical density was determined at 625 nm with a spectrophotometer (Jenway7315, Cole-Parmer Ltd., Vernon Hills, IL, USA). Therefore, a McFarland turbidity standard curve was generated to adjust the initial concentrations.

2.2. UV-LED Device

The UV-irradiation system was built in-house and has a modular design based on LED technology. Due to its modular design, UV-LED panels could be easily exchanged, and the nominal wavelengths of $\lambda =255 \mathrm{n}\mathrm{m}$ (CUD5GF1A, SEOUL VIOSYS, Seoul, Republic of Korea) and $\lambda =285 \mathrm{n}\mathrm{m}$ (S-S35B-F2-285-01-2-110, Sensor Electronic Technology Inc., Columbia, SC, USA) were applied for irradiation experiments. To apply the corresponding UV dose, the average UV irradiance at the sample plane was determined with the UV radiometer X1-Optometer equipped with the calibrated UV detector UV-3726-5 (Gigahertz-Optik, Türkenfeld, Germany) along a $20 \mathrm{m}\mathrm{m}\times 20 \mathrm{m}\mathrm{m}$ measurement grid. To fully characterize the irradiation conditions, the spectral power distribution was measured using the spectrometer MAYA 2000 Pro, including diffraction grating #HC-1, an entrance slit of $5 \mathrm{\mu }\mathrm{m}$, optical fiber QP600-1-SR-BX with a core diameter of $600 \mathrm{\mu }\mathrm{m}$, and a cosine corrector CC-3-UV-S (Ocean Insights, Orlando, FL, USA), with a resulting spectral resolution of $0.66 \mathrm{n}\mathrm{m}$ FWHM. The UV-C panel emitting at a nominal wavelength of $255 \mathrm{n}\mathrm{m}$ was equipped with $4\times 3$ LED chips and achieved an irradiance (fluence rate) of $\left(90\pm 12\right) \mathsf{\mu }{\mathrm{W}\mathrm{c}\mathrm{m}}^{-2}$, and the UV-B panel at $285 \mathrm{n}\mathrm{m}$ was equipped with $4\times 6$ LED chips and achieved an irradiance of $\left(234\pm 12\right) \mathsf{\mu }{\mathrm{W}\mathrm{c}\mathrm{m}}^{-2}$. This irradiance corresponds to 100% intensity. From spectral measurements, a peak wavelength of $(260\pm 26) \mathrm{n}\mathrm{m}$ (${\lambda }_{peak}\pm FWHM$) and $282\pm 25 \mathrm{n}\mathrm{m}$ (${\lambda }_{peak}\pm FWHM$) were determined for UV-C panels and UV-B panels, respectively. Details about the UV irradiation apparatus were previously described in Schöbel et al. [23]. Measured UV conditions present during irradiation experiments are summarized in Supplementary Table S1; the spectra and the irradiation system are shown in Supplementary Figure S1 and Supplementary Figure S2, respectively.

2.3. Survival Experiments

One mL of each bacterial suspension was dispensed into three petri dishes (diameter 37 mm, height 5 mm), resulting in a sample depth of 2.8 mm. UV treatment involved exposure to two different wavelengths ($255 \mathrm{n}\mathrm{m}$ and $285 \mathrm{n}\mathrm{m}$) across varying dose ranges: 2 to $25 \mathrm{m}\mathrm{J}{ \mathrm{c}\mathrm{m}}^{-2}$ for C. freundii (with high doses exceeding $5 \mathrm{m}\mathrm{J}{ \mathrm{c}\mathrm{m}}^{-2}$); and 5 to 150 $\mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$ for E. coli and L. innocua (with high doses exceeding $20 \mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$). Both positive (inoculated but not exposed to UV treatment) and negative controls (sterilized but left uninoculated) were included in the experimental design, with sterile working practices adhered to in order to prevent potential cross-contamination during analysis. As anticipated, no colony-forming units were detected in the negative controls. Prior to measurements, each module was pre-warmed for 15 min in a temperature-controlled environment at room temperature. Cell suspensions were vortexed before filling the petri dishes. Irradiation times were calculated from measured intensities (see Supplementary Table S1) using the UV dose Equation (1):

$$t={\displaystyle \frac{D}{I}}·1000{\displaystyle \frac{\mathsf{\mu }\mathrm{W}}{\mathrm{m}\mathrm{W}}},$$

(1)

where $t$ is the irradiation time in s, $D$ is the irradiation dose in $\mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$, and $I$ the intensity measured in $\mathrm{\mu }\mathrm{W} {\mathrm{c}\mathrm{m}}^{-2}$.

Generally, intensity is defined as incident energy per area and per time, and as such, the UV dose corresponds to the total energy per area exposed to the sample. Post UV treatment, each sample was vortexed and diluted. Subsequently, $50 \mathsf{\mu }\mathrm{L}$ of both treated and untreated diluted samples were plated onto Tryptic soy agar plates and incubated at appropriate temperatures for 24 h. Alterations in bacterial membrane integrity post UV treatment were assessed using scanning electron microscopy (SEM) (JSM-IT200, JEOL, Tokyo, Japan), following the methodology outlined by Xu et al. [26].

2.4. Inactivation Kinetics, Model Fitting and Statistical Analysis

To compare the inactivation kinetics for different UV wavelengths with respect to the UV dose, the survival rate $SR$ was calculated from the inactivation experiments as the ratio of CFU from treated and untreated samples. Depending on the present microorganism and prevailing environmental conditions, inactivation kinetics can be described using various prediction models [27,28]. For this study, a two-phase-decay model, shown in Equation (2), was chosen [29].

$$SR\left(D\right)=\left(1-f\right)·{10}^{-k_1·D}+f·{10}^{-k_2·D}$$

(2)

Here, the parameters $k_1$ and $k_2$ are the inactivation rate constants for the initially stress-sensitive (phase 1) and stress-resistant (phase 2) subpopulations, respectively. The parameter $f$ represents the fraction of the stress-resistant subpopulation. Non-linear regression was applied to experimental data using Equation (2) to determine $k_1$ and $k_2$. Model fit acceptance criteria included a coefficient of determination $R^2 \ge 0.99$, and root mean squared error $RMSE<0.01$. MATLAB 2022b [30] was used for non-linear regression. From the determined inactivation rates $k_1$ and $k_2$, UV doses required for 1-log and 4-log reductions were estimated using Equation (2). Statistical analysis was performed using a $t$-test ($n=3, * p<0.05, ** p<0.01$) with GraphPad Prism 10 (Dotmatics, Boston, MA, USA).

2.5. Extraction of RNA, Synthesis of Complementary DNA (cDNA) and RT-qPCR

After UV treatment, samples were incubated for 30 min at 37 °C in the dark, followed by centrifugation at 15,500 rpm for 5 min at 4 °C. The supernatants were carefully removed, and the bacterial pellet was processed for RNA extraction using the TRIzol Max Bacterial RNA Isolation Kit (Invitrogen, Carlsbad, CA, USA), following the manufacturer’s protocol. Subsequently, $500 \mathrm{n}\mathrm{g}$ of RNA was reverse-transcribed into complementary DNA (cDNA) using the iScript™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA). The synthesized cDNA served as a template for RT-qPCR analyses. Intracellular damage mechanisms indicated via induction of gene expression was quantified via RT-qPCR. Relative quantification, which evaluates the expression of a target gene relative to that of a reference gene, was chosen as the most commonly used method for gene expression analyses. Data normalization using housekeeping genes—which are involved in cellular basal metabolism and thus assumed to be constitutively expressed—was performed to correct variations between cells or culture conditions, thus allowing for gene expression analyses to be compared between the different samples [31]. According to Nicot et al. [32] and Linardić and Braybrook [33], the use of at least two carefully selected reference genes is necessary for reliable quantitative gene expression. In the present investigation, the following reference genes were used: rpoB and secA for C. freundii; hcaT and idnT for E. coli; and rpoB and rplD for L. innocua (Supplementary Table S2). Relative quantification ($RQ=2^{-∆∆Ct}$) was evaluated by using the average of both reference genes. The final result of this method is presented as the fold change of target gene expression in a target sample relative to a reference sample (untreated control), using reference genes for normalization. Each individual target gene was analyzed separately with its specific set of housekeeping genes to adjust for differences in amplification efficiencies during data analysis, ensuring that observed differences in expression levels are not artifacts arising from technical variations.

All RT-qPCR analyses were performed in triplicates using nine technical replicates. RT-qPCR was performed on a CFX96 Touch Deep Well Real-Time PCR System (Bio-Rad, USA) targeting recA, involved in DNA damage repair, and soxS, involved in sensing oxidative stress. Details on the used primers are summarized in Supplementary Table S2. Each qPCR reaction mix contained 5 μL iQ™ SYBR^® Green Supermix (Bio-Rad, USA), 250 nM of forward and reverse primer, 100 ng of cDNA, and UltraPure DNase/RNase-Free Distilled Water (Invitrogen, USA) to a final volume of 10 µL. Each run included negative and non-template controls (UltraPure DNase/RNase-Free Distilled Water, Invitrogen, USA). Cycling conditions (40 cycles) included an initial denaturation step at 95 °C for 30 s, followed by 95 °C for 15 s. Specific annealing temperatures were used for each primer pair for 30 s: C. freundii recA and soxS 55 °C; rpoB and secA 52.5 °C; E. coli recA, soxS, hcaT, and idnT, all 50 °C; and L. innocua recA, rpoB, and rplD 52 °C. A final elongation step at 72 °C for 30 s completed the protocol. Primers were initially designed using NCBI’s Primer BLAST tool [33], which integrates primer design with BLAST search functionality. This integration enables researchers to design primers specific to their target gene or sequence of interest, considering factors such as primer size, melting temperature (Tm), GC content, and potential secondary structures. This approach minimizes the risk of non-specific amplification and enhances the reliability of PCR results, establishing it as a standard practice for validating primers in gene-specific studies. Following primer design, primer specificity was assessed through in silico PCR using the same platform [34], as well as melt curve analysis, ensuring robust validation against potential off-target sequences.

3. Results

UV-LED systems were utilized to study inactivation kinetics, followed by RT-qPCR to investigate the underlying stress mechanisms. Survival curves (Figure 2) and inactivation rates (Figure 3) for each target microorganism exposed to varying UV doses at two central wavelengths (UV-C at $255 \mathrm{n}\mathrm{m}$ and UV-B at $285 \mathrm{n}\mathrm{m}$) were determined. Data on controls and individual measurements of inactivation kinetics across different UV doses and wavelengths are provided in Supplementary Figure S3. The required UV doses for 1-log and 4-log reductions were estimated using a two-stage model and summarized in Supplementary Table S3. Additionally, gene expression profiles of recA and soxS for each microorganism, depending on the applied UV dose and wavelength, were analyzed to understand the underlying stress mechanisms in Gram-positive and Gram-negative bacteria (Figure 4).

3.1. Wavelength Dependent Inactivation Kinetics

3.1.1. Disinfection Performance

In general, UV doses of approximately $10 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ were sufficient to inactivate 90% of all tested microorganisms at wavelengths of $255 \mathrm{n}\mathrm{m}$ and $285 \mathrm{n}\mathrm{m}$ (Figure 2 and Supplementary Table S3). Between these two wavelengths, the efficacy for 1-log reductions was greater at $255 \mathrm{n}\mathrm{m}$, requiring lower radiation doses (Supplementary Table S3 and Figure 2). For instance, a $23 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ UV dose resulted in a 4-log reduction of L. innocua at $255 \mathrm{n}\mathrm{m}$, whereas a higher dose of $50 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ was necessary to achieve the same reduction level at $285 \mathrm{n}\mathrm{m}$ (Supplementary Table S3). In contrast, E. coli exhibited higher 4-log reductions at $285 \mathrm{n}\mathrm{m}$ compared to $255 \mathrm{n}\mathrm{m}$, needing lower UV doses for the same inactivation effect (approximately $86 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ at 255 nm and $79 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ at $285 \mathrm{n}\mathrm{m}$, Supplementary Table S3). For C. freundii, very low UV intensities were effective for inactivation, and similar doses were needed to achieve 1-log ($1-2 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$) and 4-log ($11-12 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$) reductions regardless of the wavelength (Supplementary Table S3). The bactericidal efficacy weakened with the increasing dose, resulting in the tailing phenomenon at high UV doses, indicated by a smaller inactivation rate $k_2$ (phase 2) compared to $k_1$ (phase 1) according to the model.

3.1.2. Inactivation Rates

To compare the efficacy of UV irradiation at different wavelengths ($255 \mathrm{n}\mathrm{m}$ vs. $285 \mathrm{n}\mathrm{m}$), the rate constants for each target microorganism were calculated for the lower and higher doses ($k_1$ and $k_2$, respectively). The results, shown in Figure 3 and

Table 1. Summary of the determined inactivation rates $k_1$ and $k_2$ from the two-phase decay model (Equation (2)) for the investigated microorganisms at two central UV wavelengths. Values represent means $\pm 95\%$ confidence levels ($n=3$).

Microorganism	$\mathbf{UV} \mathbf{Wavelength} \mathbf{in} \mathbf{n}\mathbf{m}$	$\mathbf{Inactivation} \mathbf{Rate} \mathbf{in} \mathbf{c}{\mathbf{m}}^2\mathbf{m}{\mathbf{J}}^{-1}$
		${\mathit{k}}_{\mathbf{1}}$	${\mathit{k}}_{\mathbf{2}}$
C. freundii	$255$	$1.04 \pm 0.03$	$0.10 \pm 0.05$
	$285$	$0.51 \pm 0.18$	$0.19 \pm 0.08$
E. coli	$255$	$0.22 \pm 0.09$	$0.10 \pm 0.04$
	$285$	$0.01 \pm 0.003$	$0.03 \pm 0.026$
L. innocua	$255$	$0.28 \pm 0.01$	$0.01 \pm 0.007$
	$285$	$0.15 \pm 0.05$	$0.04 \pm 0.02$

, indicate that at low doses, the inactivation constants were significantly higher at $255 \mathrm{n}\mathrm{m}$ (p < 0.01) compared to $285 \mathrm{n}\mathrm{m}$.

At higher UV doses, the trend reversed, with inactivation rates at $285 \mathrm{n}\mathrm{m}$ being consistently higher regardless of the target microorganism. For example, E. coli exhibited comparable inactivation rates to L. innocua, with rates of $0.2 \mathrm{c}{\mathrm{m}}^2 \mathrm{m}{\mathrm{J}}^{-1}$ at $255 \mathrm{n}\mathrm{m}$ and $0.1 \mathrm{c}{\mathrm{m}}^2 \mathrm{m}{\mathrm{J}}^{-1}$ at $285 \mathrm{n}\mathrm{m}$ versus $0.3 \mathrm{c}{\mathrm{m}}^2 \mathrm{m}{\mathrm{J}}^{-1}$ at $255 \mathrm{n}\mathrm{m}$ and $0.2 \mathrm{c}{\mathrm{m}}^2 \mathrm{m}{\mathrm{J}}^{-1}$ at $285 \mathrm{n}\mathrm{m}$, respectively. The obtained inactivation rates clearly demonstrated that UV treatment at $255 \mathrm{n}\mathrm{m}$ was significantly more efficient (p < 0.01) at inactivating all tested bacteria at lower UV doses. Conversely, UV treatment at $285 \mathrm{n}\mathrm{m}$ exhibited higher efficacy at higher doses (Figure 3).

3.2. Wavelength Dependent Stress Mechanisms

This study also examined wavelength- and UV dose-dependent stress mechanisms, specifically focusing on the gene expression profiles of recA and soxS, indicative of DNA damage and oxidative stress, respectively. The induction of stress-related genes upon exposure to $255 \mathrm{n}\mathrm{m}$ and $285 \mathrm{n}\mathrm{m}$ was assessed at varying doses: $2 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ and $15 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ for C. freundii; and $10 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ and $75 \mathrm{m}\mathrm{J} \mathrm{c}{\mathrm{m}}^{-2}$ for both E. coli and L. innocua, respectively. The analysis was conducted under sublethal conditions (>1-log reductions), facilitated by the stable expression of reference genes and adequate target template amplification, likely due to a resilient sub-population. Scanning electron microscopy revealed structurally intact cells post UV-B and UV-C exposure, exemplified in Supplementary Figure S4 for E. coli.

Our RT-qPCR results demonstrated high recA activation at $255 \mathrm{n}\mathrm{m}$ and low fluence. Conversely, high fluence at $285 \mathrm{n}\mathrm{m}$ induced high soxS activation, aligning with increased ROS formation (Figure 4). Regarding the tested microorganisms, we found a significant upregulation of recA expression across both Gram-negative and Gram-positive microorganisms in response to $255 \mathrm{n}\mathrm{m}$ UV irradiation at low doses (Figure 4). This upregulation was particularly pronounced in E. coli, followed by L. innocua. In Gram-negative bacteria, higher UV doses corresponded to reduced recA activation, whereas Gram-positive L. innocua exhibited increased recA response with higher doses. Notably, UV-B exposure did not significantly upregulate recA in C. freundii or E. coli. In contrast, L. innocua consistently showed strong recA upregulation across all conditions. Overall, irradiation with shorter wavelengths, particularly $255 \mathrm{n}\mathrm{m}$, induced the highest recA activation, implying more severe DNA damage compared to $285 \mathrm{n}\mathrm{m}$ (Figure 4).

Targeting soxS involved in the oxidative stress response, particularly in response to reactive oxygen species, exposure to longer wavelengths induced increased expression in the two Gram-negative bacteria, E. coli and C. freundii. Furthermore, UV-C treatment significantly reduced soxS activation compared to UV-B (Figure 4). This reduction in soxS response was more pronounced in C. freundii than in E. coli. Notably, soxS is absent in non-enteric bacteria; thus, data for L. innocua are not available. In contrast to recA, irradiation with $285 \mathrm{n}\mathrm{m}$ resulted in higher soxS activation, suggesting heightened oxidative stress likely due to increased ROS production (Figure 4).

Our results strongly suggested that wavelength-dependent effects are significant: shorter wavelengths ($255 \mathrm{n}\mathrm{m}$) caused more direct DNA damage, triggering the SOS response, as indicated by recA upregulation, particularly in Gram-positive L. innocua. In contrast, longer wavelengths ($285 \mathrm{n}\mathrm{m}$) primarily induced oxidative stress, leading to soxS upregulation, especially in Gram-negative bacteria (Figure 4). For Gram-negative and Gram-positive bacteria, E. coli demonstrated a distinct response pattern between the DNA damage response (recA) and oxidative stress response (soxS), with significant variations depending on the wavelength of UV radiation (Figure 4). Similarly, C. freundii showed a response pattern similar to E. coli but with a more pronounced reduction in soxS activation under UV-C, indicating heightened susceptibility to direct DNA damage. In contrast, L. innocua consistently exhibits a robust recA response (Figure 4), indicative of strong DNA repair mechanisms.

4. Discussion

In the present investigation, the survival and inactivation rates of C. freundii, E. coli, and L. innocua were determined after UV-B and UV-C exposure at varying doses. Stress responses post UV treatment were elucidated via RT-qPCR targeting the genes recA and soxS, highlighting the distinct strategies that Gram-negative and Gram-positive bacteria employ to manage UV-induced stress depending on the wavelength. To enhance readability, the results are discussed separately in two sections.

4.1. Wavelength Dependent Inactivation Kinetics

Concerning the efficacy of UV treatment in reducing the number of viable cells, our results of $4.5 \mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$ for a 1-log reduction of E. coli (Supplementary Table S3) compare well with those previously reported for various E. coli strains in 15 different studies using UV-C light emitting at $255 \mathrm{n}\mathrm{m}$, where a 1-log reduction was achieved at $3.9\pm 1.9 \mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$ [35]. Regarding the survival curves (Figure 2), the initial rapid decline in viable organisms indicates that UV exposure to low doses effectively reduces the microbial population; however, as the UV dose increases beyond a certain threshold, the rate of reduction slows down or “tails off”. These results are consistent with the findings observed by Rattanakul and Oguma [4], who noted tailing phenomena where microorganisms exhibited a residual population after being inactivated at approximately a 4.5-log reduction across all wavelengths. The commonly observed tailing effect implies that despite initial effective reduction of microbial populations under UV treatment, a residual population persists as UV doses increase. This phenomenon suggests that some microorganisms possess mechanisms such as resistance or efficient repair capabilities, or they may be shielded from UV exposure. For instance, compounds that scatter light, insoluble particles, and color pigments can hinder light penetration, thereby reducing the effectiveness of log-inactivation [36,37,38]. Bernbom et al. [39] demonstrated that biofilm formation in Listeria sp. protected bacterial cells during UV-C exposure, likely through physiological changes involving the development of extracellular compounds or physical shielding. Moreover, the efficacy of inactivation may decrease when the wavelength used for treatment approaches the peak absorbance wavelength of light-absorbing compounds [40]. The consistent sharper decline in viable organisms when exposed to UV-C at lower doses compared to UV-B treatment (Figure 2) suggests that UV-B may be less effective initially in reducing the microbial population. However, as the UV-B dose increases, there is a stronger decline in viable organisms, which might imply that higher UV-B doses are more effective at reducing the microbial population, including resistant sub-populations (Figure 2). This differential effectiveness underscores the importance of selecting appropriate UV wavelengths and doses based on specific disinfection requirements and conditions. UV-LEDs emitting light in the range of $255-285 \mathrm{n}\mathrm{m}$ induce varying inactivation rates for E. coli, with reported rate constants ranging from $0.15$ to $0.81 \mathrm{c}{\mathrm{m}}^2 \mathrm{m}{\mathrm{J}}^{-1}$ based on previous studies [41], which align well with the results in the present investigation (Table 1). Among the organisms tested, C. freundii showed the highest inactivation rate constants (Table 1). This finding suggests that C. freundii is more sensitive to UV exposure, requiring the lowest UV dose for inactivation, and that among enteric bacteria, certain species may be more susceptible to UV exposure than others. Consistent with our findings, Soro et al. [2] reported that enteric bacteria, including E. coli and presumably C. freundii, typically require lower UV doses for effective inactivation compared to cocci, like Listeria spp., as well as bacterial spores, viruses, and fungi. This result is further in line with the findings of Ramsay et al. [42], who noted that in water, E. coli exhibited slightly greater tolerance to $254 \mathrm{n}\mathrm{m}$ UV-C irradiation compared to L. innocua, achieving reductions of 2.2 and 2.9 logs, respectively.

The varying resistance among bacteria to UV radiation is mostly attributed to structural differences, particularly the thicker peptidoglycan layer in Gram-positive bacteria. This layer serves as a physical barrier that absorbs and scatters UV radiation, thereby reducing its penetration into the cell and its ability to cause damage to DNA and other cellular components [43]. Furthermore, Gram-positive bacteria produce pigments that act as UV protectants by absorbing and dissipating UV radiation before it reaches sensitive cellular components [44]. Additionally, Gram-positive bacteria often possess more efficient DNA repair mechanisms compared to Gram-negative ones. Although considered less effective than those in Gram-positive bacteria, Gram-negative bacteria are able to produce higher levels of porphyrins that might mitigate the impact of photoexcitation, as reported by McSharry et al. [45] and Endarko et al. [46]. UV radiation, along with other environmental stresses, can further induce bacteria to enter the viable but nonculturable (VBNC) state. Currently, 85 bacterial species have been identified as capable of entering a VBNC state, including several emerging waterborne pathogens found in drinking water such as Helicobacter, Yersinia, Legionella, Pseudomonas, Escherichia, and Listeria spp. [47,48]. In the VBNC state, bacteria remain viable but do not grow on standard culture media, allowing them to potentially evade detection while still being capable of regrowth under favorable conditions. When bacteria are in a viable state, they can repair UV-induced damage, which includes mechanisms like upregulating genes involved in DNA repair. Trevors [49] reported that all viable but non-culturable cells of E. coli in river water, deionized water, and even chlorinated water retained gene expression. This fact is also particularly important regarding spore-forming bacterial species that may still be viable [50]. In this context, previous investigations explored the role of the resuscitation-promoting factor (Rpf), which contributes to the revival of VBNC cells [51] and exhibits distinct differences among Gram-negative and Gram-positive bacteria.

4.2. Wavelength-Dependent Stress Mechanisms

For efficient disinfection purposes, particularly in water treatment, allowing DNA repair to occur is undesirable and might be prevented by specifically targeting these enzymes. RT-qPCR has emerged as a dependable method for accurately assessing gene expression dynamics in response to UV radiation or other environmental stressors. By utilizing RT-qPCR to target the recA and soxS genes, the current investigation aimed to uncover the molecular responses of bacteria to UV-induced stress, contributing to a more comprehensive understanding of bacterial survival strategies under varying UV wavelengths and doses.

RecA expression is significantly influenced by UV treatment due to its critical role in DNA repair and the activation of the SOS response system in bacteria. UV radiation causes damage to bacterial DNA, leading to the formation of lesions such as thymine dimers. In response to this DNA damage, bacterial cells activate the SOS response pathway. RecA protein plays a central role in this pathway by facilitating the repair of damaged DNA and regulating the expression of genes involved in DNA repair and mutagenesis. Studies have demonstrated that exposure to UV radiation leads to a rapid increase in recA gene expression. For instance, research by Ojha et al. [52] and Kidambi et al. [53] has shown that recA expression can increase more than two-fold following UV exposure, aligning well with the high activation observed in our investigation (Figure 4). This upregulation is crucial for the repair of UV-induced DNA damage and for maintaining bacterial survival under stress conditions. Our results clearly indicate that exposure to $255 \mathrm{n}\mathrm{m}$, corresponding to the absorption peak of DNA, has a direct germicidal effect by acting directly on the DNA molecule itself, thereby triggering the SOS response with recA upregulation, especially notable in Gram-positive L. innocua (Figure 4). In contrast, with exposure to longer wavelengths, more indirect mechanisms of damage occur, inducing greater oxidative stress. Here, ROS generated by UV exposure induce oxidative damage not only to DNA but also to other cellular components like proteins, resulting in soxS upregulation, particularly observed in Gram-negative bacteria (Figure 4). The wavelength-dependent differences observed align with findings from Pousty et al. [54], although their study exclusively used E. coli as the test organism. Drawing definitive conclusions about how Gram-positive and Gram-negative bacteria differ in their responses to different UV wavelengths is challenging due to the limited number of microorganisms tested. Additionally, the DNA damage incurred depends on various factors, including the wavelength of UV light, leading to stress mechanisms that can stem from direct damage to cellular components or indirect effects on biochemical processes [3]. Matallana-Surget et al. [55] reported that species with a high genomic GC content are more susceptible to UV-induced mutagenesis due to cytosine-containing photoproducts being highly mutagenic. Low and high GC contents correlated well with the tolerance or resistance of the different species, respectively, with Listeria sp. and their comparatively low GC content of about 38% on average being the least sensitive (on average, E. coli: ~51%; and C. freundii: ~52%).

Regarding Gram-positive and Gram-negative bacteria, L. innocua seems to rely extensively on the SOS response to manage UV-induced stress, demonstrating strong activation of recA across various conditions, suggesting a robust capacity for DNA repair and effective upregulation of DNA repair mechanisms under elevated UV stress. In contrast, Gram-negative bacteria seem to exhibit a dual response mechanism, characterized by distinct pathways for DNA damage and oxidative stress. More precisely, in E. coli, significant upregulation of recA at low UV doses suggests substantial DNA damage at $255 \mathrm{n}\mathrm{m}$, prompting initiation of the SOS response, while reduced recA activation at higher doses may indicate severe damage that overwhelms repair mechanisms or triggers pathways leading to cell death. Across all organisms investigated, a consistent pattern in stress-related responses emerged: (i) when the UV-C inactivation constant $k_1$ is lower (indicating higher UV tolerance), there is generally higher expression of recA, suggesting that organisms with higher UV tolerance activate more robust DNA repair mechanisms to counteract direct DNA damage; and (ii) when the UV-B inactivation constant $k_2$ is higher, there tends to be higher expression of soxS due to oxidative stress. Moreover, $285 \mathrm{n}\mathrm{m}$ UV seems to induce less DNA damage or a different type of stress that does not primarily activate the SOS response. In this context, exposure to $280 \mathrm{n}\mathrm{m}$ UV LEDs was reported to notably suppress photoreactivation in E. coli. Additionally, while both dark and light repair mechanisms were observed in L. innocua, exposure to $280 \mathrm{n}\mathrm{m}$ LEDs resulted in irreversible damage to the bacterial cells [56]. This damage was evident when targeting soxS as an indicator for oxidative stress (Figure 4).

In summary, UV inactivation provides several notable advantages over other inactivation methods for water and food treatment. It offers broad-spectrum efficacy against various microorganisms without promoting resistance and operates rapidly even at low concentrations. Unlike chlorination, it avoids producing harmful byproducts like trihalomethanes (THMs), and, unlike ozonation, UV systems are simpler and less costly to maintain [57]. Additionally, UV treatment does not introduce chemical residues, preserving the taste and quality of food while avoiding heat damage to sensitive products. Its compact size and targeted wavelengths make it versatile for integration into different systems and applications [58]. However, while UV-B and UV-C light are highly effective for disinfection and other applications, it is crucial that personnel working with UV sources are properly trained on the associated hazards and safety protocols. To safely harness the benefits of UV technology, proper safety measures must be implemented, including engineering controls such as interlocks, personal protective equipment like skin and eye protection, and procedural controls related to distance and duration of exposure.

UV-B treatment is increasingly advocated due to its economic advantages and ability to achieve comparable or superior inactivation rates and log reductions at higher UV doses. UV-LEDs operating at 285 nm, for example, demonstrate inactivation rate constants similar to those of mercury lamps. Consistent with our findings, Martino et al. [59] demonstrated increasing inactivation kinetics with decreasing wavelength (255 > 265 > 285 nm). However, energy efficiency trends inversely with wavelength (255 < 265 < 285 nm) due to the currently lower wall plug efficiency for lower UV LED wavelengths emissions. Furthermore, the external quantum efficiency of UV-B LEDs is higher compared to UV-C LEDs [60], resulting in higher fluence rates (same treatment time) or shorter treatment times (same fluence rate). Despite this, UV-B LEDs exhibit higher optical power outputs than UV-C LEDs at the same current [61]. UV-B irradiation also suppresses photoactivation and dark repair more effectively than low-pressure (LP) UV lamps, 265 nm LEDs, or simultaneous exposure to both, likely due to greater damage to photolyase enzymes at higher wavelengths [62]. However, the commercial expansion of germicidal UV-LEDs is currently hindered by their low external quantum efficiency (~20% for 275 nm, ~15% for 285 nm, and ~5% for 265 nm LEDs) [63]. A technoeconomic evaluation by Maclsaac et al. 2023 [64] determined a reduction in power consumption of about 25% and 43% for UV LEDs at 280 nm compared to conventional low-pressure UV lamps for relevant UV doses of $30 \mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$ and $40 \mathrm{m}\mathrm{J} {\mathrm{c}\mathrm{m}}^{-2}$, respectively. In addition, UV LEDs are promising in terms of life cycle analysis compared to mercury lamps, as shown in the study by Pizzichetti et al. (2024) [65]. As UV-B light can penetrate certain media or materials more effectively than UV-C, its use is advantageous for applications where UV-C might be less effective. UV-B systems typically consume less energy than UV-C systems, resulting in lower operational costs, particularly for large-scale applications. Additionally, UV-B induces oxidative stress alongside direct DNA damage, generating ROS, which further results in oxidative modifications of DNA bases, single-strand breaks, and cross-linking of DNA and proteins. This dual action can enhance the overall inactivation process and potentially reduce the effectiveness of DNA repair mechanisms, which can be beneficial for certain disinfection contexts [66]. Dual approaches combining UV-A at 365 nm—which primarily induces indirect DNA damage through ROS—with UV-C have further been proven more effective in inactivating microorganisms, creating a more challenging environment for DNA repair and increasing the levels of residual DNA damage across cellular components [67].

5. Conclusions

Our results clearly demonstrate that UV-LED treatment efficiently inactivates both Gram-negative and Gram-positive bacteria, achieving over 4-log reductions at both $255 \mathrm{n}\mathrm{m}$ and $285 \mathrm{n}\mathrm{m}$. The inactivation rates at $285 \mathrm{n}\mathrm{m}$ and high fluence are comparable to, or even higher than, those achieved at $255 \mathrm{n}\mathrm{m}$. Additionally, UV-B treatment at $285 \mathrm{n}\mathrm{m}$ required shorter exposure times to achieve the same UV dose, attributed to its deeper penetration depth in water and higher optical output. Stress responses were both wavelength- and dose-specific and varied across Gram-positive and Gram-negative bacteria. The notable increase in recA expression at lower UV doses suggests considerable DNA damage at 255 nm, which triggers the SOS response. In contrast, diminished recA activation at higher doses may reflect overwhelming damage that exceeds the repair capacity or induces pathways that lead to cell death. At 255 nm, soxS expression decreased with increasing UV dose, suggesting a diminished response to oxidative stress at higher doses. In contrast, at 285 nm, soxS expression increased with higher doses, indicating a heightened oxidative stress response as the UV dose rises. Upregulation of soxS indicates increased ROS formation associated with oxidative stress, which not only complements the effects of UV-C but also enhances the overall inactivation process by diminishing the effectiveness of DNA repair mechanisms. Our findings suggest that achieving efficient water treatment through UV irradiation can be enhanced by employing UV-B LEDs or a combination of UV-C and UV-B LED systems, rather than relying solely on UV-C. Based on the insights gained in the present investigation, further investigations should focus on potential dual wavelengths synergies (simultaneous or sequential); different target microorganisms, including viruses; comparisons between environmental vs. lab-seeded samples (which may exhibit different susceptibilities to inactivation); VBNC bacteria (noted for their increasing resistance and ability to regain growth); reactivation experiments in the dark to confirm regrowth; and the use of propidium monoazide quantitative polymerase chain reaction (PMA-qPCR) to assess VBNC bacteria in environmental samples. Addressing these aspects, along with proper reactor design for large-scale assessments, will ultimately contribute to the development of tailored UV-LED treatment methods indispensable for broad industrial application.