Urine was UV irradiated in a cylindrical stainless steel photoreactor (45 cm length and 3.1 cm OD). The photoreactor was equipped with a 65 W low pressure high output mercury lamp (GPHHVA357VH, LightTech, Hungary) emitting UV light at wavelengths of 254 nm and 185 nm. The photoreactor was then placed in a chamber with running water to maintain a temperature of 21 ± 2 °C inside the reactor. The fluence rate of the lamp was measured by iodine/iodide actinometry (0.184 ± 0.005 mW cm⁻²) following the procedure as described before (Rahn, 1997). Quantum yields were calculated assuming all the light reaching iodide-iodate solution had a wavelength of 254 nm.

Urine donations (n = 37) were collected from both male and female volunteers (aged 20–65 years) using high-density polyethylene bottles with lids. The collected urine was pooled, dosed with 2.35 g KOH L⁻¹ and mixed. The alkalized urine was then kept at room temperature (20 ± 2 °C) until further use (15–20 days).

Urine which had been alkalinized (pH 10.8) was inoculated with Escherichia coli and Enterococcus faecium to study the inactivation of bacteria by plate count and fate of their resistance genes by qPCR when subjected to treatment of UV, H₂O₂ or in combination (UV/H₂O₂). The inoculated urine was subjected to the treatments for time periods of 5, 10, 20, 40, and 80 min (Table 1) and analyzed by destructive sampling design with a single replicate per time point. As such, no technical or biological replicates were performed at individual time points. In addition to the main experiment using urine at pH 10.8, urine alkalinized to 12.5 and neutralized to pH 7.0 was studied for some time periods. Treatments at pH 7 and 12.5 were conducted as controls to test the efficacy of the treatment; fresh urine were accounted by treatment at pH 7 while treatment at pH 12.5 represents urine alkalized by strong bases such as KOH.

The strains used for this study were Escherichia coli (CCUG 62975) with bla_{CTX − M} gene and Enterococcus faecium (NCTC 12202) with van-A gene. Inactivation of bacteria was investigated by plate count whereas the fate of their resistance genes (bla_{CTX − M} and van-A respectively) was investigated by qPCR. The bacteria were cultivated in two steps in nutrient broth and overnight cultures (37 °C, 12h) and aliquoted into 35 ml portions for E. coli and 40 ml portions for E. faecium, respectively. The portions was then centrifuged (4,500 rpm for 10 min), and the supernatant was discarded while bacterial cells were retained in the pellet. The bacteria pellets were resuspended in 0.5 ml saline solution and refrigerated at 4 °C and used that same day. At the start of an experiment the prepared bacteria were used to inoculate 600 ml urine, resulting in a start concentration of 8 log₁₀ cfu ml⁻¹ and 6 log₁₀ cfu ml⁻¹ urine for E. coli and E. faecium respectively.

For enumeration of bacteria, 1 ml of urine was sampled before and after each treatment (Table 1). One (1) ml urine was serially diluted in Buffered saline solution with peptone and Tween 80. Cultivation of E. faecium strain was carried out by growing it on CHROMagar (Chromagar TM) with and without vancomycin (6 mg L⁻¹), while the E. coli strain was grown on Tryptone Bile X-Glucuronide (TBX) agar with and without cefotaxime (6 mg L⁻¹). All agars were incubated at 37±2 °C for 24 ± 2h after which distinct colonies were counted using image processing software (OpenCFU) (Geissmann, 2013).

Treatment procedure: resuspended E. coli and E. faecium pellets were spiked in 600 ml of urine and mixed for 30 s on a magnetic stirrer (Section 2.3). The urine was then poured down into the photoreactor. Once the photoreactor was placed in the cooling bath, the UV lamp was turned on to start the treatment. At start and end of each treatment period a 1 ml sample was taken for bacteria enumeration and 50 ml of urine was collected, mixed with 20% (v/v) of Tris-EDTA, and stored at −20 °C until use for ARG analysis. For the treatments involving hydrogen peroxide, 1.25 g H₂O₂ L⁻¹ (36 mM) was added to the inoculated urine before pouring it down the photoreactor. Controls, were performed following the same procedure using the same photoreactor but without UV irradiation or hydrogen peroxide, to mimic the experimental condition, but studied for fewer time intervals (Table 1). Control samples were measured at selected intervals during the main experiment, based on the expectation of minimal variation under the tested conditions as indicated by pre-trial experiments. Additionally, post-trial measurements conducted at the final treatment time points (20 and 80 min) confirmed the stability of the control samples.

DNA extraction and strain confirmation: to quantify the abundance of bla_{CTX − M} and van-A gene, DNA was extracted from a 22 ml mixture of tris-EDTA and urine using DNeasy blood and tissue test kit (cat.no 69504, Qiagen, Germany) after equilibration of samples to room temperature. The DNA concentration was measured using Qubit^® 3.0 Fluorometer (life technologies, Malaysia). Confirmatory 16S-rRNA –based Sanger sequencing was run on the extracted DNA to check E. coli and E. faecum and rule out contamination (Supplementary Table 1).

Resistant genes amplification, cloning and Plasmid DNA isolation: The bla_{CTX − M} and van-A genes were PCR amplified, sequenced for confirmation and cloned in preparation for quantitative Real-Time PCR (qPCR)-based gene quantification. bla_{CTX − M} and van-A genes from the extracted DNA were amplified using the primers and reaction conditions in Table 2. Each PCR reaction (25 μl) contained 5 μl template DNA, 12.5 μl Taq polymerase, 1 μl of both forward and reverse primer and 5.5 μl water. Optimized PCR conditions listed in Table 2 were followed and the PCR product was quantified using Qubit^®. The PCR product size was confirmed using gel electrophoresis (1% agarose in tris-EDTA).

Following the manufacturer's instructions, the PCR product was cloned to plasmid using pGEM^®-T vectors (Promega GeneJET plasmid miniprep kit, Thermo Scientific, K0502) and transformed to JM 109 high efficiency competent E. coli cells overnight at 37 °C. Transformation was confirmed by plating the cells in 50 μg ml⁻¹ ampicillin-containing agar plates. Plasmid DNA extraction was performed on transfromants grown overnight in LB broth as per the instructions in GeneJet purification kit (Thermo Scientific, K0702). Furthermore, the PCR product and plasmid DNA from transformed cells were sanger sequenced by Macrogen Europe (Netherlands) to confirm the amplification was from the intended gene of interest. The obtained sequence was nucleotide blasted on CARD database (Alcock et al., 2023) for confirmational purposes (Supplementary Table 2).

Standard curve preparation and qPCR reaction: The extracted Plasmid DNA were successively diluted to prepare a standard qPCR curve in the range of 10 to 10⁸. A qPCR reaction was performed using 96 well qPCR machine (QuantStudioTM 5 Real-Time PCR, applied biosystems, Thermo Fisher Scientific, USA). The built-in design and Analysis software 2.0 was used to design the plates, set the reaction condition, and collect the data. A triplicate of negative control was run in each qPCR as quality control. PCR conditions were optimized, following the protocols outlined in Table 2. The primers and probes used in this study were taken from previous publications (listed in Table 2), with modification. Each qPCR reaction (25 μl) contained 3 μl template DNA (1:100 diluted), 12.5 μl Maxima^TM probe qPCR master mix (Thermo scientific, K0261, USA), 1 μl of both forward and reverse primer, 0.6 μl probe and 6.9 μl nuclease free water. Quantitative PCR for bla_{CTX − M} and van-A genes were performed using the primers and probes listed in Table 2. The expected amplicon size was confirmed through agarose gel electrophoresis (Supplementary Figure 1).

All the chemicals and reagents used in the study were of an analytical grade. For measurements of pH and EC, a pH electrode with an integrated Pt1000 temperature sensor (6.0258.010, Herisau, Switzerland) and an EC cell (6.0917.080, Metrohm, Herisau, Switzerland) connected to pH/EC meter (Metrohm, CH-9100 Herisau, Switzerland) were used. To adjust the pH of the urine, 5M KOH and 1 M H₂SO₄ were used. The UV absorbance of the urine was measured in the wavelength range of 190–400 nm using a Lambda 365 UV-vis spectrophotometer (Perkin-Elmer, United States) with 1 cm optical path length, prior to which urine samples were diluted 100-fold with Milli-Q water.

The concentration of total nitrogen (N_tot), total ammonia nitrogen, and chemical oxygen demand (COD) was determined colorimetrically using Spectroquant^® test kits (Merck KGaA, Darmstadt, Germany) and a spectrophotometer (NOVA 60 A, Merck KgaA, Germany). COD measurements were adjusted following the method described by (Kang et al. 1999) to account for the potential interference of residual peroxide. The concentration of residual peroxide in urine was determined following the procedure described by (Arve and Popat 2021).

The bacteria inactivation, i.e., the reduction in bacteria concentrations, for each treatment period was given as -log₁₀ cfu by normalizing end concentrations to start concentrations as log₁₀ (C_t/C₀). Models for E. coli and E. faecium inactivation over time were fitted for each treatment by combining the normalized inactivation data for the different treatment times which were studied independently. Changes in log₁₀ reduction over time was tested against two inactivation models, a log-linear model (Equation 1), and a model for shouldered inactivation curves (Equation 2) suggested for UV inactivation (Harm, 1980) and also used for chemical inactivation. Inactivation kinetics along with prediction interval (95% confidence limit), the latter when the number of data points allowed, (Minitab 15; Minitab Ltd., United Kingdom), was derived using Equations 1–3.

where C₀, and C_t are concentrations of colony forming units at time zero and time t, k is first-order inactivation rate (min⁻¹), t is time (min), and n is an empirical value which is used to calculate the lag period l (min). Both Equations 1, 2 were used to model reduction kinetics of ARGs.

Results from qPCR (cycles) were converted to gene copies using Equation 4.

where X_o is gene copy number, E_amp is efficiency of amplification, C_q is the cycle number, m and b are the slope and the constant of the regression equation of the standard curve, respectively.

Treatments employing UV are also expressed in terms of incident fluence/UV dose and conversion of treatment time into UV dose is calculated according to Equation 5.

Plating E. coli on a cefotaxime containing plate after 80 min treatment revealed the inactivation of E. coli with 4 log₁₀ for UV, and more than 6.5 and 7.5 log₁₀ for H₂O₂ and UV/H₂O₂ treatment, respectively (Figure 1A). All three treatments resulted in a higher inactivation of E. coli compared to the control (without UV and H₂O₂), which exhibited only a 0.5 log₁₀ inactivation over 80 min at the initial pH 10.8 (Figure 2A). E. coli was inactivated with treatment by UV following a log-linear inactivation with an inactivation rate constant (k) of −0.06 log₁₀ cfu min⁻¹ (Table 3). At 5^th min, treatments of H₂O₂ and UV/H₂O₂ had E. coli concentrations below the detection limit, indicating an inactivation of more than 1 log₁₀ cfu min⁻¹ (Figures 1A,B).

Inactivation of E. faecium was observed for treatments of UV and UV+H₂O₂ with a 2.7 log₁₀ and a 6 log₁₀ inactivation in 80 minutes, respectively (Figure 1B). However, treatment of H₂O₂ alone had no effect on the inactivation of E. faecium. Unlike UV treatment, inactivation of E. faecium under treatments of UV/H₂O₂ had a lag phase of 15 min (Table 3). For E. faecium, inactivation rate constants of −0.033 and, −0.093 log₁₀ cfu min⁻¹ were observed for treatments of UV and UV/H₂O₂, respectively (Table 3). Inactivation for E. faecium was faster when UV treatment was combined with H₂O₂ (Figures 1A,B).

For treatment at pH 7.0, both UV and H₂O₂ achieved 3 log₁₀ inactivation of E. coli within the 80 min treatment, while no inactivation was observed for the control measurements (without UV and H₂O₂). However, for E. faecium, inactivation at pH 7.0 was observed only with UV treatment, with a 1.5 log₁₀ inactivation in 80 min. For UV and H₂O₂ treatments at pH 12.5, inactivation beyond the detection limit was observed within 20 min of treatment; >7 log₁₀ and >3 log₁₀ inactivation for E. coli and E. faecium, respectively. The same inactivation was observed after 20 min in control (without UV and H₂O₂) at pH 12.5. A post-trial experiment conducted using the same KOH alkalized urine showed more than 7 log₁₀ inactivation for E. coli within 1 min of exposure to a pH of 12.5 (data not shown). Bacterial inactivation results indicate a clear trend over time, it is important to note that the data were generated using a single-replicate design without repeated measurements at each time point. Consequently, the precision of individual data points is limited, hence observed trends require careful interpretation.

Recombinant plasmid DNA carrying either bla_{CTX − M} or van-A gene was amplified and visualized by gel-electrophoresis and the results showed DNA fragments in accordance with the expected PCR product length (Supplementary Figure 1). Confirmatory sequencing of the plasmid insertions revealed a 100% identity match for van-A gene in E. faecium (ARO:3000010) and 100% identity match for bla_{CTX − M} gene in E. coli (ARO:3005661) when blasted against reference sequences on CARD database (Alcock et al., 2023). The efficiency of the qPCR reaction for plasmid standard curves of each respective gene were in the range of 95 and 105%, and R² of 0.99.

With the 80 min treatment of urine at an initial pH of 10.8, a 1 and 3 log₁₀ reduction of the bla_{CTX − M} gene copy was observed for UV and UV/H₂O₂ treatment, respectively (Figure 2A). Treatment with H₂O₂ alone and control without UV and H₂O₂, revealed no reduction of both bla_{CTX − M} and van-A gene over 80 min treatment (Figure 2). The bla_{CTX − M} gene showed a log-linear reduction with a k value of −0.009 ± 0.001 and −0.055 ± 0.003 log₁₀ gene copies min⁻¹ for treatments of UV and UV/H₂O₂, respectively (Table 3). Degradation of van-A gene was also observed for treatments of UV and UV/H₂O₂ (Figure 2C). Treatment of UV/H₂O₂ for 80 min resulted in 2 log₁₀ gene copy reduction with the k value of −0.04 log₁₀ gene copies min⁻¹ with a lag time of 31 and 80 min treatment with UV resulted in a 0.5 log₁₀ gene copy reduction with a k value of −0.0052 log₁₀ gene copies min⁻¹ (Figure 2C, Table 3).

For 20 min treatment at pH 12.5, the bla_{CTX − M} gene was reduced with 2.2, 3.5, and 4.5 log₁₀ gene copies for control, and treatments of UV and H₂O₂, respectively (Figure 2B). However, among the 80 min treatments at pH 7.0, gene copy reduction was observed only for H₂O₂ with 0.5 log₁₀ gene copies (Figure 2B). Reduction of van-A gene occurred only at pH 12.5 with a comparable gene reduction of 1 log₁₀ for control and treatments of UV and H₂O₂ (Figure 2D).

Treatment of alkalized urine (pH 10.8) for 80 min with H₂O₂ and UV/H₂O₂ resulted in more than 6 log₁₀ inactivation for both E. coli and E. faecium. However, inactivation of the two organisms differed for the H₂O₂ treatment, i.e., E. coli was inactivated with more than 6 log₁₀ while E. faecium showed persistence toward treatments of H₂O₂ (1.25 g L⁻¹) with only 0.01 ± 0.08 log₁₀ inactivation. The reduction of E. coli is in line with earlier studies, e.g., 30 mM H₂O₂ gave >6 log₁₀ E. coli reduction in citric acid-Na₂HPO₄ buffer solution under 5 min (Raffellini et al., 2011) while 0.3 mM H₂O₂ resulted in no reduction of E. coli in phosphate buffer solution within 30 min (Sun et al., 2016). The high dose of H₂O₂ (37 mM) in this study could be one of the reasons for the higher inactivation of E. coli as compared to previous studies.

For treatments involving reactive oxygen species (H₂O₂, OH^*) to inactivate microorganisms, the first step in the process is the damage of cell walls. E. faecium, a gram positive bacterium, has a thicker cell wall (25 nm; Mishra et al., 2012) compared to E. coli (4 nm; Gan et al., 2008). Therefore, the underlying difference in cell structure could explain the fast inactivation of E. coli while E. faecium concentrations were not decreased by H₂O₂ treatment alone (Figure 1) (Zhang et al., 2023; Rodríguez-Chueca et al., 2015). Moreover, exposure to H₂O₂ has the ability to modify cell surface charge of Gram-positive bacteria which leads to an aggregation of bacteria, thus retarding treatment efficacy of microbial inactivation (Zhang et al., 2023). In contrast to Gram-positive bacteria, Gram-negative bacteria, such as E. coli, does not undergo such a change in surface charge upon exposure to oxidants, such as H₂O₂, thus restricting cell aggregation. This allows a free interaction of the oxidant with the cell (Zhang et al., 2023). Such a phenomenon could explain the observed lag phase for E. faecium when exposed to the combination of UV + H₂O₂ (Figure 1B), which resulted in an immediate inactivation of E. coli (Figure 1A).

Inactivation studies employing H₂O₂ treatment coupled with pH have demonstrated that both high pH (pH 9; Batterman et al., 2001) and low pH (pH 3; Raffellini et al., 2011) results in enhanced inactivation (>2 log₁₀ higher inactivation) compared to the neutral pH. This may explain the observed difference in inactivation of both bacteria to pH controls between pH 10.8 and 12.5 (Figure 1). However, the inactivation of both bacteria in urine alkalized to a pH 12.5 was so rapid that any added effect of H₂O₂ or UV could not be observed (Figure 1).

In this study an inactivation of 2.7 and 1.5 log₁₀ was observed for UV treatment at pH 7.0 for E. coli and E. faecium, respectively. (Hokanson et al. 2016) showed E. coli to have a higher susceptibility to photolysis, with a photolysis coefficient of 238,593 L Einstein⁻¹ cm⁻¹, compared to E. faecalis, with a coefficient of 147,116 L Einstein⁻¹ cm,-1 which may explain the higher inactivation rate of E. coli compared to E. faecium in this study. (Mckinney and Pruden 2012) studied the inactivation of antibiotic resistant E. faecium (van-A) and E. coli (tet(A)) using UV in filtered wastewater. The authors reported that for a 3 log₁₀ inactivation E. faecium required an at least 2-fold higher UV dose compared to E. coli, similar to what was observed in this study. Conversely, a study by (He et al. 2021) claimed that G+ bacteria, such as E. faecalis, are more susceptible for photocatalysis than G- bacteria, such as E. coli, indicating that susceptibility toward UV treatment cannot be generalized by the gram features of the cell. A review by (Hijnen et al. 2006) also stated that the sensitivity toward UV treatment differs between different strains of the same species. Additionally, E. faecium occurs in pairs or chains and enterococci in general are prone to clustering leading to less exposure to UV, which can explain the demoted inactivation compared to E. coli due to UV and H₂O₂ treatment in the present study.

Studies conducted using monochromatic low pressure mercury lamp (254 nm) reported that up to 15 mJ cm⁻² UV dose and 20 mJ cm⁻² is required for >5 log₁₀ inactivation of E. coli in deionized water (Harris et al., 1987) and secondary treatment effluent, respectively (Nasser et al., 2006). However, in this study UV doses of 880 mJ cm⁻² and < 55 mJ cm⁻² were required for 4 log₁₀ inactivation of E. coli with UV alone and UV/H₂O₂ treatment at pH 10.8, respectively (Figure 1). Additionally, (Hokanson et al. 2016) reported an inactivation k value of −0.506 cm² mJ⁻¹ for E. coli under UV treatment (254 nm) in water. In this study E. coli was inactivated in KOH alkalized urine at pH 10.8 with a k value of −0.0053 cm² mJ⁻¹ and −0.0824 cm² mJ⁻¹ for treatments with UV and UV/H₂O₂, respectively (Supplementary Table 3). However, the exposure to a lower wavelength UV light has an increased effectiveness toward microbial inactivation as it can cause comparatively high damage compared to UV 254 nm, which can explain the observed difference (Clauß, 2006). Indeed, (Clauß 2006) studied microbial inactivation using a krypton-chloride excimer lamp emitting photon at 222 nm and a LP mercury lamp at 254 nm and reported UV treatment at 222 nm resulting in comparatively higher inactivation compared to treatment at 254 nm. Additionally, (Moussavi et al. 2019) reported that vacuum UV (185 + 254 nm) treatment resulted in 2-fold inactivation of E. coli compared to UV 254 nm when given the same treatment conditions. Furthermore, (Giannakis et al. 2018) reported a >4 log₁₀ E. coli inactivation in real urine for a treatment time of 45 min by UV (254 nm) at pH 7.0, contrasting to the present study in which only 3 log₁₀ E. coli inactivation was achieved with 80 min of UV treatment at pH 7.0, even when the lamp we used emitted light at both 185 and 254 nm. A possible explanation for these results is that the effect of a lower wavelength (185 nm) is overshadowed by the presence of organic matters in urine that has high UV absorbance at this wavelength (Demissie et al., 2024).

Inactivation studies conducted in wastewater effluent requires a higher UV dose when compared with pure water or phosphate solution (Hijnen et al., 2006). The study by (Giannakis et al. 2018) revealed that >4-fold higher treatment time was required for comparable E. coli inactivation in real urine (>45 min) compared to activated sludge effluent water ( ≤ 10 min). Depending on the organisms intended to be removed, it may require even higher doses. For example, a UV dose of 400 mJ cm⁻² was required for 2.5 log₁₀ inactivation of Ascaris suum eggs in phosphate buffer saline solution and 560 mJ cm⁻² for >4 log₁₀ inactivation of Aspergillus niger spores in demineralized water (Brownell and Nelson, 2006; Clauß, 2006; Masjoudi et al., 2021). Microbial particle association, light interference, and scavenging properties of the matrix are possible reasons for the requirement of high treatment time/UV dose for inactivation in wastewater and urine solutions (Giannakis et al., 2018; Örmeci and Linden, 2002). On the contrary, inactivation studies employing a combination of UV/H₂O₂ show that treatment time significantly decreases with an increase in H₂O₂ dose. UV activates H₂O₂ by resulting in two OH^* which are non-selective oxidants that enhance microbial inactivation (Vilhunen et al., 2011; Giannakis et al., 2018), which likely explains the higher observed inactivation of E. faecium during treatment UV/H₂O₂ compared to treatments of UV and H₂O₂ alone (Figure 1B).

Aside from pH 12.5, controls without UV and H₂O₂ at pH 7.0 as well as 10.8 did not result in any inactivation of both test organisms (Figure 2). E. faecalis, a close relative of E. faecium, is reported to survive high pH (pH 11), however, microbial growth was highly affected for pH >11.5 (Mchugh et al., 2004; Starliper and Watten, 2013). High alkaline pH inactivated bacteria through the action of hydroxyl anions. Hydroxyl anions have the ability to (i) damaging cytoplasmic membrane, (ii) denature enzymes and (iii) damage DNA (Siqueira Jr and Lopes, 1999). This therefore explains the fast inactivation of both test organisms at pH 12.5.

In this study, up to 3 log₁₀ gene reduction was observed for bla_{CTX − M} with treatments of UV/H₂O₂ whereas only 2 log₁₀ was noted for van-A gene with the same treatment and time (80 min ≈ 880 mJ cm⁻²; Figure 2). Amplicon size is reported as one of the factors for the indifference in degradation rate of genes as there are fewer pyrimidine dimer (TT,CT,TC or CC) targets as the gene size gets shorter (He et al., 2022; Mckinney and Pruden, 2012). (He et al. 2019) studied degradation of extracellular blt gene with amplicon sizes ranging from 266 bp to 1017 bp and reported that there is a 4-fold increase in gene degradation during UV treatment at pH 7.0 for the large amplicon size (−0.12 cm² mJ⁻¹, 1017 bp) compared to the smaller amplicon size (−0.025 cm² mJ⁻¹, 266 bp). The relative difference in amplicon size could explain the relatively higher degradation of bla_CTX−M gene (336 bp) and van-A gene (96 bp) in the present study (Table 2).

According to a review by (Han et al. 2023), a comparatively higher UV dose is required to degrade ARGs rather than inactivate ARBs. This is in line with the results in the present study which shows a difference in inactivation of ARBs (>6 log₁₀) with degradation of ARGs ( ≤ 4 log₁₀) (Figures 1, 2). Interferences of matrix, formation of cell clusters, and scavenging of both photons and oxidants formed during UV irradiation by lysed cell matters may contribute to the demoted degradation of ARGs compared to the inactivation of ARBs. In this study, a UV treatment time of 80 min (which is equivalent to 880 mJ cm⁻²) was required to degrade a 96 bp van-A gene by 0.5 log₁₀ gene copies min⁻¹. Notably, (Mckinney and Pruden 2012) reported a UV dose (254 nm) of 200 mJ cm⁻² to be required for 4 log₁₀ reduction of both intracellular and extracellular van-A (1,030 bp) gene in phosphate buffer solution.

During treatments of urine with UV, sulfate, phosphate, and carbonate-radicals are formed in addition to the hydroxyl radicals (Zhang et al., 2016, 2015). These radicals react with DNA bases with a rate constant of 5 × 10⁷ −9 × 10⁹ L mol⁻¹ s⁻¹. For example, thiamine reacts with 2.1 and 1.1 × 10⁹ L mol⁻¹ s⁻¹ with SO4-* and PO42-*, respectively (Ma et al., 2018). Therefore, such radicals might also be involved in the reduction of bla_{CTX − M} and van–A gene during UV irradiation of urine.

Hydrogen peroxide, however, does not damage DNA directly but rather through the production of hydroxyl radicals reacting with iron containing molecules (Mendoza-Chamizo et al., 2018). Hydroxyl radicals react with nucleotide bases to a create single lesion on DNA (Cadet and Wagner, 2013). For instance, thiamine reacts with OH^* with a rate 7.4 × 10⁹ L mol⁻¹ s⁻¹ (Ma et al., 2018). This explains the higher reduction of ARGs by UV/H₂O₂ treatment compared to UV alone (Figures 2A,C). Further, with a UV dose of 600 mJ cm⁻² by the low pressure UV lamp (254 nm) around 10¹⁴ M OH^* are formed in a solution containing 0.3 mM H₂O₂ (Rosenfeldt et al., 2006). Thus, the increased formation of OH^* for experiments involving a combination of H₂O₂ and UV explains the higher reduction of ARG compared to UV alone (Figures 2A,C). However, treatment of H₂O₂ resulted in higher reduction of bla_{CTX − M} gene compared to UV treatment at pH 12.5 (Figure 2D). As explained in section 4.1, gram-negative bacteria is more susceptible to H₂O₂ treatment compared to gram-positive ones (Zhang et al., 2023). Consequently, DNA of E. coli will be more exposed to H₂O₂ action when compared to DNA of E. faecium, which is protected by cell aggregates. This explains the higher reduction of bla_{CTX − M} gene compared to van-A gene in treatments at pH 12.5 (Figures 2B,D).

High pH (pH > 11) inactivates pathogens through disruption of cytoplasmic membrane which leads to cell lysis and the release of DNA to the solution (Mendonca et al., 1994). At pH > 11, DNA is denatured and becomes single stranded by abstraction of hydrogen by OH ion (England et al., 2021; Bivehed et al., 2023). Gram-negative microorganisms are more prone to cell lysis at pH > 11 compared to gram-positive microorganisms. This also explains the comparative ARG reduction difference between bla_{CTX − M} and van-A gene with control treatment at pH 12.5. (Goetsch et al. 2020) reported 2 log₁₀ reduction in the transformation efficiency of extracellular plasmid DNA harboring ampicillin and tetracycline resistant gene incubated in hydrolyzed urine for 24 h. Therefore, DNA damage for treatments at pH 12.5 could be greater when considering transformation efficiency of ARGs downstream urine processing steps. For instance, urine dehydration, one of the nutrient concentration process steps following urine stabilization, could be done with a temperature reaching 60 °C (Simha et al., 2020) and enzymes or proteins responsible for DNA repair could be denatured at this temperature which could further decrease the chance of ARG transfer (Boulon et al., 2010).

The results of inactivation of ARBs and degradation of ARGs indicate that it takes a comparably longer treatment time, or UV dose, for degradation of ARGs compared to ARBs, which is in line with previous studies conducted in water and wastewater matrices (Hokanson et al., 2016; He et al., 2022). However, comparisons between bacterial inactivation and ARG degradation results should be interpreted with caution, as the inactivation data were obtained from a single-replicate design without repeated measurements at each time point. Increasing the pH to 12.5 resulted in increased inactivation and degradation of ARGs to a level in which the use of UV or H₂O₂ treatment was not necessary. Thus, nutrient recovery technologies aimed at recovering N in the form urea-N could employ either a combination of UV/H₂O₂ treatment at pH 10.8 or prolonged storage time (>3 h) at pH 12.5 for enhanced 6 log₁₀ inactivation of pathogenic bacteria and degradation of ARGs (>4 log₁₀ gene copy), thereby keeping the nutrient potential intact (Supplementary Figure 2). Furthermore, considering UV/H₂O₂ treatment as pre-treatment for the recovery and use of nutrients from source separated urine, an 80 min treatment at pH 10.8 was sufficient to meet the performance target of 6 log₁₀ microbial reduction set for unrestricted use of excreta for agricultural purposes (WHO, 2006). However, since the results presented here are based on laboratory conditions, further investigation is required for its applicability in decentralized source separation systems. (Demissie 2023) discussed the limitations of such technology for source separated urine and it's recommended for future studies to address such limitations.