Efficacy and Mechanism of Antibiotic Resistance Gene Degradation and Cell Membrane Damage during Ultraviolet Advanced Oxidation Processes

Combinations of UV with oxidants can initiate advanced oxidation processes (AOPs) and enhance bacterial inactivation. However, the effectiveness and mechanisms of UV-AOPs in damaging nucleic acids (e.g., antibiotic resistance genes (ARGs)) and cell integrity represent a knowledge gap. This study comprehensively compared ARG degradation and cell membrane damage under three different UV-AOPs. The extracellular ARG (eARG) removal efficiency followed the order of UV/chlorine > UV/H2O2 > UV/peracetic acid (PAA). Hydroxyl radical (•OH) and reactive chlorine species (RCS) largely contributed to eARG removal, while organic radicals made a minor contribution. For intracellular ARGs (iARGs), UV/H2O2 did not remove better than UV alone due to the scavenging of •OH by cell components, whereas UV/PAA provided a modest synergism, likely due to diffusion of PAA into cells and intracellular •OH generation. Comparatively, UV/chlorine achieved significant synergistic iARG removal, suggesting the critical role of the RCS in resisting cellular scavenging and inactivating ARGs. Additionally, flow cytometry analysis demonstrated that membrane damage was mainly attributed to chlorine oxidation, while the impacts of radicals, H2O2, and PAA were negligible. These results provide mechanistic insights into bacterial inactivation and fate of ARGs during UV-AOPs, and shed light on the suitability of quantitative polymerase chain reaction (qPCR) and flow cytometry in assessing disinfection performance.


Text S3. DNA Extraction Methods
In this study, we adopted vacuum filtration of oxidant-, UV-, and AOP-treated cell suspensions through polycarbonate membranes (reported to have the lowest affinity to eDNA 4 ), followed by the extraction by DNeasy ® PowerSoil ® Pro Kits of cells retained on the membrane, for iARG recovery and purification prior to analyses.Five-mL samples were diluted and the cells (containing iARGs) were collected by vacuum filtration onto 0.22 µm membranes.The cells on the membranes were subsequently extracted into 0.1-mL concentrate by the DNeasy ® PowerSoil ® Pro Kits (Qiagen) according to instructions.
The iARG recovery of this method could be calculated by control extraction of E. coli HB101-harbored pWH1266 plasmid (carrying blaTEM-1) before disinfection.According to plating on ampicillin-containing agar, we found the cell concentration of the sample was ~5×10 7 CFU/mL.Note that we added 5-mL sample for membrane filtration and collected only 0.1 mL of extract for iARG quantification, according to instructions of PowerSoil ® .Thus, the cells in our sample were concentrated 50 times (i.e., final E. coli concentration in the extract should be ~2.5×10 9CFU/mL).Analyses by qPCR showed the ARG concentration in the 0.1-mL control extract was ~2×10 9 copies/mL, indicating good recovery.Nonetheless, each E. coli HB101 may contain multiple pWH1266 plasmids, hence the absolute recovery (i.e, recovered ARG copies/ARG copies in the samples) cannot be calculated with certainty.As an additional check, the recovery of E. coli genome was estimated by comparing the extracted total DNA concentration versus theoretical concentration of E. coli genome in the raw sample to be ~33%. 5However, this result may not represent the plasmid recovery efficiency due to the different DNA size and conformation.

Text S4. qPCR Analysis
DNA samples were analyzed by qPCR on a StepOnePlus real-time PCR instrument (Applied Biosystems, Foster City, CA) with the termparuture profile summarized as Table S2.
We used a signal threshold at 0.007 for determination of threshold cycle number (Cq) and resulted in a standard curve with a slope of -0.29 (representing qPCR efficiency) and a R 2 of 0.998 (Figure S2).Autocalved phosphate buffer (prepared with DI water) was used as the notemplate control (NTC) and we found the signal was below the threshold after 40 amplification cycles.
As the NTC did not lead to amplification, the limit of detection (LOD) was arbitrarily determined as 3 copies per reaction -the lowest possible LOD for a qPCR reaction. 6,7Therefore, considering 10 L of samples was added per reaction, the LOD was calculated to be 300 copies/mL for this study.The lowest concentration used in the standard curve is 2.23×10 5 copies/mL and still resulted in reproducible measurement, hence limit of quantification (LOQ) for the assay should be lower than 2.23×10 5 copies/mL.All the samples collected from degradation experiments had a concentration higher than 2.23×10 5 copies/mL (Figures 3, S2).

Text S5. Photochemcial Calculations
The first-order rate constants for photolysis of oxidants (Table S3) was calculated by eq S1.
c is the concentration of the photophore (in M); Φ is the quantum yield (in mol/Einstein, or unitiless if we consider ensitein as moles of photons); ε is the molar absorption coefficient at the specific wavelength (in M -1 cm -1 ); α is the absorption coefficient (in cm -1 ) of water matrix; d is the path length, which equals the water depth in our set-up (Figrue S1) (in cm); I is the fluence rate received by the reactor at the current set-up (in Einstein/(Ls)).When the experiments were conducted in clean phosphate buffer,  = 0 and the total absorbance was less than 0.02, hence eq S1 could be simplified to eq S2.
When the experiments were conducted in matrix simulating source water,  was determined to be 0.12 cm -1 by a UV-visible spectrophotometer, which is much higher than the absorbance by oxidants (c), and the total absorbance was less than 0.02, hence eq S1 could be simplified to eq S3.
In addition, unit conversion can be conducted for fluence rate I by eq S4.
where h is the Planck constant (6.626×10 -34 Js, c is the speed of light in vacuum (3×10 8 m/s), λ is the photon wavelength (254 nm in this study).

Figure S1 . 21 Figure S2 .
Figure S1.Set-up of the UV reactor in this study.

Figure S6 .
Figure S6.Flow cytometry plots for UV alone treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S7 .
Figure S7.Flow cytometry plots for H2O2 alone treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S8 .
Figure S8.Flow cytometry plots for UV/H2O2 treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S9 .
Figure S9.Flow cytometry plots for free chlorine treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S10 .
Figure S10.Flow cytometry plots for UV/free chlorine treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S11 .
Figure S11.Flow cytometry plots for PAA treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S12 .
Figure S12.Flow cytometry plots for UV/PAA treatment of E. coli (triplicate results are shown for each time point, conditions see Figure S5).

Figure S14 .
Figure S14.Flow cytometry plots for oxidants or UV alone treatments of BAC bacteria community (duplicate results are shown, conditions see Figure S13).

Figure S15 .
Figure S15.Flow cytometry plots for UV-AOP treatments of BAC bacteria community (duplicate results are shown, conditions see Figure S13).

Figure S16 .
Figure S16.Flow cytometry plots of blank samples (DI water) for determination of the noise gating in BAC community samples.

Table S1 :
Composition of the synthetic source water

Table S2 :
Temperature profile of qPCR analyses

Table S3 :
Principal reactions in the kinetic model.