Journal of Hazardous Materials 438 (2022) 129396

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat

Research Paper

Advanced wastewater treatment with ozonation and granular activated

carbon filtration: Inactivation of antibiotic resistance targets in a long-term
pilot study
K. Slipko a, *, D. Reif a, H. Schaar a, E. Saracevic a, A. Klinger a, L. Wallmann a, J. Krampe a,
M. Woegerbauer b, P. Hufnagl c, N. Kreuzinger a
a
TU Wien, Institute for Water Quality and Resource Management, Karlsplatz 13/226, 1040 Vienna, Austria
b
Department for Integrative Risk Assessment, Austrian Agency for Health and Food Safety, Spargelfeldstraße 191, 1220 Vienna, Austria
c
Institute for Medical Microbiology and Hygiene - Center for Anthropogenic Infections, Austrian Agency for Health and Food Safety, Währingerstrasse 25a, 1090 Vienna,
Austria

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Ozonation and GAC were evaluated for

ARB and ARGs abatement in a long-term
study.
• Ozone doses ≥ 0.6 g O3/g DOC were
required to affect tested DNA targets.
• Ampicillin resistant E. coli was less sus­
ceptible to ozonation effects than other
ARB.
• A relative abundance of blaTEM-1 gene
increased after GAC treatment.
• ΔUV254 was tested as a proxy parameter
to indicate DNA damage by ozone.

A R T I C L E I N F O A B S T R A C T

Editor: Jianhua Guo The inactivation of antibiotic resistant bacteria (ARB) and genes (ARGs) in an advanced plant combining
ozonation and granular activated carbon (GAC) filtration applied for effluent after conventional activated sludge
Keywords: treatment at a full-scale urban wastewater treatment plant was investigated for over 13 consecutive months. The
Urban wastewater nitrite compensated specific ozone dose ranged between 0.4 and 0.7 g O3/g DOC with short-time sampling
Advanced wastewater treatment
campaigns (0.2–0.9 g O3/g DOC). Samples were analysed with culture-dependent methods for bacterial targets
Ozone
and with qPCR for genes. The log removal values were correlated with a decrease of the matrix UV absorption at
GAC
Antibiotic resistance 254 nm (ΔUV254) and indicated a range of ΔUV254 that corresponds to a sufficient membrane damage to affect
DNA. For trimethoprim/sulfamethoxazole resistant E. coli, sul1, ermB and tetW, this phase was observed at
ΔUV254 of ~30 % (~0.5 g O3/g DOC). For ampicillin resistant E. coli and blaTEM-1, it was observed around 35–40
% (~0.7 g O3/g DOC), which can be linked to mechanisms related to oxidative damages in bacteria resistant to
bactericidal antibiotics. GAC treatment resulted in a further abatement for trimethoprim/sulfamethoxazole
E. coli, sul1 and tetW, and in increase in absolute and relative abundance of ermB and blaTEM-1.

* Corresponding author.
E-mail address: [email protected] (K. Slipko).

https://doi.org/10.1016/j.jhazmat.2022.129396
Received 4 January 2022; Received in revised form 13 June 2022; Accepted 14 June 2022
Available online 17 June 2022
0304-3894/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

1. Introduction Nitrite compensated specific ozone doses between 0.4 and 0.6 g O3/g
DOC were found to be suitable for efficient chemical CECs abatement
Biological wastewater treatment, including conventional activated (Bourgin et al., 2018; Hollender et al., 2009; Rizzo et al., 2020). Inac­
sludge (CAS) treatment, is only partially effective in the removal of tivation of ARB and removal of ARGs under similar ozone doses as
chemical and microbial contaminants of emerging concern (CECs) chemical CECs would be highly beneficial for the application of ozon­
(Alexander et al., 2016; Hiller et al., 2019; Krzeminski et al., 2019; ation as a wastewater treatment step (Czekalski et al., 2016). The effect
Kümmerer, 2009). Indeed, residual chemical CECs, e.g., pharmaceutical of ozonation on ARB and ARGs has not been studied in detail so far
active compounds like antibiotics and microbial CECs as e.g., antibiotic (Rizzo et al., 2020), with most of the studies carried out in lab-scale size
resistant bacteria (ARB) and antibiotic resistance genes (ARGs) were only (Czekalski et al., 2016; Iakovides et al., 2019; Lamba and Aham­
reported to be disseminated into the environment with effluent and mad, 2017; Michael-Kordatou et al., 2017; Moreira et al., 2016; Oh
excess sludge from municipal wastewater treatment plants (WWTPs) et al., 2014; Sousa et al., 2017; Zheng et al., 2017; Zhuang et al., 2015).
(Berendonk et al., 2015; Hiller et al., 2019; Kümmerer, 2009; Pazda Studies treating municipal wastewater in a pilot- or even full-scale are
et al., 2019; Vaz-Moreira et al., 2014). scarce (Table 1) and vary in sampling procedures, operation parameters,
WWTPs act as collection points for ARB, ARGs and antibiotics from a examined ARB and ARGs, and methods applied for ARB and ARGs
variety of sources like hospitals, households, and veterinary husbandries detection/quantification. Though investigations under full-scale and
(Krzeminski et al., 2019). A co-existence of human commensals and “real life” conditions are of core importance since intrinsic temporal
pathogens with environmental bacteria under various selection pres­ fluctuations in quantitative and qualitative wastewater matrix parame­
sures from antibiotics, heavy metals and biocides, and stress factors ters, varying operation conditions (ozone dose and contact time) and
(nutrient, redox conditions, and temperature) during biological waste­ changes in abundance of ARB and ARGs occur under “real-life” condi­
water treatment may facilitate the potential of transfer and proliferation tions and affect the process performance.
of resistant strains via horizontal or vertical gene transfer (Michael-­ In our study, the demonstrator plant for chemical CECs removal from
Kordatou et al., 2018; Poole, 2017). Therefore, biological WWTPs are CAS effluent by ozonation followed by GAC filtration was monitored and
identified as one of the most important routes for the propagation of assessed for ARB and ARGs abatement for 13 consecutive months. The
antibiotic resistance from humans to the environment (Berendonk et al., study focused on:
2015; Krzeminski et al., 2019; Pärnänen et al., 2019; Pruden et al., 2013;
Rizzo et al., 2013). Moreover, the presence of residual amounts of i. Investigating, if the range of ozone doses suggested to be applied
chemical and microbiological CECs in biologically treated effluents may for chemical CECs abatement (0.4–0.6 g O3/g DOC) is suitable for
compromise the reuse and safety of treated wastewater, a practice ARB and ARGs abatement under “real-life” conditions as well
increasingly applied to address water scarcity globally (Krzeminski ii. Correlating ARB and ARGs abatement with ΔUV254 to assess if
et al., 2019; Michael-Kordatou et al., 2018; Rizzo et al., 2020). this parameter could be applied as a surrogate for the elimination
Ozonation proved to be an effective and economically feasible option of these targets similar to its application for organic CEC removal
to achieve a further reduction of many chemical CECs by more than 80%
(Bourgin et al., 2018; Gerrity et al., 2012; Lee et al., 2013; Rizzo et al.,
2019). This technology is also suitable to reduce or inactivate pathogens
(Czekalski et al., 2016; Dodd, 2012; Von Gunten, 2003). The mecha­
nisms of bacteria disinfection or inactivation by ozone include the Table 1
Comparison of pilot-scale ozonation studies targeting ARB and/or ARGs abate­
disruption of bacterial cell walls and release of intracellular constitu­
ment. *The ranges of specific ozone doses were calculated based on the applied
ents, the damage of nucleic acids, and the cleavage of carbon-nitrogen
flow proportional ozone dose [4 mg O3/L and 6 mg O3/L] and range of measured
bonds in proteins (Alexander et al., 2016; Iakovides et al., 2019; Rizzo DOC in wastewater reported by the authors of cited study.
et al., 2020). Although the ozone dose applied for effluent ozonation can
Treatment Ozone dose Target ARB Target ARGs Reference
ensure efficient removal of chemical CECs and inactivation of microor­
scale
ganisms, it may lead to a formation of undesirable organic and inorganic
Pilot-scale 0.9 ± 0.1 g Enterococcus, vanA, blaVIM, (Alexander
by-products, especially bromate (Wu et al., 2018). Therefore, biological
O3/g DOC P. aeruginosa, ermB, ampC et al., 2016)
post-treatment is recommended after ozonation (e.g., biological acti­ Staphylococcus,
vated carbon filter, BAC) to ensure the degradation of by-products Enterobacteria
(Rizzo et al., 2020). However, regrowth of ARB in technical filters like Lab-scale 0.45–0.55 g E. coli, sul1 (Czekalski
BAC may compromise the inactivation efficiency of these biological and full- O3/g DOC Cultivable et al., 2016)
scale heterotrophic
CECs by biological repair mechanisms and regrowth (Czekalski et al.,
bacteria
2016). Therefore, treatment in BAC filters should be thoroughly inves­ Pilot-scale 1 g O3/g E. coli, sul1, blaTEM, (Hembach
tigated for its application for ARB and ARGs abatement from DOC A. baumannii, tetM, CTX-M, et al., 2019)
wastewater. Intestinal CTX-M-32,
enterococci blaOXA-48,
The full-scale application of ozonation for ARB and ARGs inactiva­
blaVIM,
tion requires technical optimization of the ozonation process consid­ CMY-2, vanA,
ering both the bacterial species and associated ARGs, and the mcr-1,
physicochemical properties of the wastewater matrix (Michael-Korda­ blaNDM
tou et al., 2018). Furthermore, suitable strategies for accurate ozone Pilot-scale 0.54–0.72 g E. coli, – (Kirchner
and Enterococcus et al., 2020)
dosing are needed for achieving desirable abatement of CECs while
0.79–1.26 g
minimizing the risk of effluent toxicity. To evaluate the ozone treatment O3/g DOC*
efficacy for the reduction of chemical CECs, the relative decrease in UV Pilot-scale 0.5–2.5 mg Cultivable – (Li et al.,
absorption at 254 nm (ΔUV254) as a difference between the matrix UV254 O3/L heterotrophic 2018)
bacteria
before and after treatment was suggested as an easily accessible surro­
Pilot-scale 0.5–11 mg E. coli, – (Luczkiewicz
gate parameter (Wu et al., 2018) suitable for continuous and online O3/L Enterococcus et al., 2011)
ozonation process control (Bahr et al., 2007; Stapf et al., 2016). In this Pilot-scale 0.73 g O3/g E. coli, – (Lüddeke
paper, for the first time we were applying the ΔUV254 approach on the DOC Enterococcus, et al., 2015)
assessment of the inactivation of E. coli and wastewater autochthonous and
Staphylococcus
bacteria as well as the abatement of ARB and ARGs.

2
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

iii. The differences in the behaviour of tested ARB and ARGs during within a day (data not shown).
ozone treatment: may antibiotic resistance mechanisms link to Samples from the raw influent wastewater were also collected to
the differences in response to ozonation between bacteria? monitor the quality of the inflow to the WWTP (data not shown). All
iv. Investigating the risk of potential ARGs selection by ozone samples were transported cooled and processed within 24 h after
treatment collection.
v. Assessing a biological activated GAC filter (BAC) for its effects on
ARB and ARGs abundance. 2.3. Chemical analyses

2. Materials and methods Collected wastewater samples were analysed for COD (ISO 15705),
DOC (EN 1484), PO3-4 -P (ISO 6878), NH4 -N (ISO 11732), NOx-N (ISO
+
-
A comprehensive one–year monitoring of a full–scale ozonation 13395), NO2-N (ISO 13395) and suspended solids (DIN 38409–2). The
plant coupled with GAC filtration was carried out by applying cultur­ UV absorbance at 254 nm (UV254) was measured with a lab-photometer
e–based and PCR–based methods as suggested in Michael–Kordatou (Lamda 35) from Perkin Elmer (USA).
et al. (2018) and Von Sperling et al. (2018). Over 13 consecutive
months, samples were analysed to determine changes in ARB&ARGs 2.4. Selection of investigated ARB and ARGs
abundance before and after ozonation, and GAC treatment as described
in the following subchapters. The selection of Escherichia coli as a target microorganism was based
on its broad application as an indicator for fecal contamination, its
2.1. Technical operation omnipresence in WWTPs and the availability of studies focusing on the
inactivation of this organism by ozonation.
The investigated WWTP is located in the southeast of Austria, has a The occurrence of four targeted ARGs (sul1, ermB, tetW, blaTEM-1) was
capacity of 7250 population equivalents and treats municipal waste­ frequently reported in both influent and effluent of WWTPs (Pärnänen
water. The WWTP consists of mechanical treatment and a conventional et al., 2019; Pazda et al., 2019). The choice of ARGs was based on their
activated sludge (CAS) stage with nitrification and denitrification. The following characteristics: (i) conferring resistance to different types of
CAS effluent after the secondary clarifier (CAS-OUT) was applied as the antibiotics with various modes of action, and (ii) differences in major
feed for the multibarrier demonstrator plant. Over the investigated bacterial hosts in wastewater. Tested genes confer resistance to sulfon­
period, in the CAS-OUT, the average chemical oxygen demand (COD) amides (sul1), macrolide–lincosamide–streptogramin B (MLSB) antibi­
and dissolved organic carbon (DOC) concentrations were 14.26 ± 2.45 otics like erythromycin, azithromycin, and clarithromycin (ermB),
mg/L and 4.26 ± 0.49 mg/L, respectively. The one-year average NH+ 4 -N, tetracyclines (tetW), and type A beta-lactams, like ampicillin (blaTEM-1)
NOx-N, PO3- 4 -P and suspended solids concentrations in CAS-OUT (n = (Pärnänen et al., 2019). Furthermore, sul1 and ermB genes were reported
13) can be found in Table S1. The three ozone reactors operated in series to be linked mostly with environment-associated taxa, whereas tetW and
had a total volume of 12 m3 and the hydraulic retention time varied blaTEM-1 genes with human-associated taxa in WWTPs (Yin et al., 2019).
between 9 and 40 min, depending on the inflow dynamics (a detailed
scheme can be found in Mišík et al., 2020). A nitrite compensated spe­ 2.5. Culture-dependent analysis
cific ozone dose of 0.55 g O3/g DOC was targeted in the automated
process control system for routine operation. Since nitrite reacts fast and Chromogenic coliform agar (CCA) (VWR Chemicals, USA) was used
stoichiometrically with ozone, the ozone consumed for its oxidation to enumerate total and resistant Escherichia coli and coliforms in
(3.43 g O3/g NO2-N) was subtracted from the ozone dose in order to wastewater. Reasoner’s 2 A agar (R2A), conventionally used for
determine the so-called nitrite compensated specific ozone dose. The microbiological testing of potable waters, was chosen to investigate
specific ozone dose control was based on a site-specific UV-DOC corre­ slow-growing heterotrophs in the effluents (Reasoner and Geldreich,
lation model and continuous measurement with an online UV spec­ 1985), especially targeting ozonated samples, in which bacteria cells are
trometer probe (s∷can, Vienna, Austria). Since nitrite was not measured expected to be damaged and require time to repair and grow (Czekalski
online, the nitrite compensated specific ozone dose was calculated a et al., 2016). Media were prepared according to the manufacturer’s in­
posteriori based on laboratory analysis. Due to occasional nitrite peaks structions and spiked with the appropriate concentration of antibiotic
in the WWTP effluent, the nitrite compensated specific ozone doses (to reach final concentrations of 32 mg/L ampicillin (AMP) or 4 mg/L
during routine operation ranged between 0.4 and 0.7 g O3/g DOC. trimethoprim (TMP) together with 76 mg/L sulfamethoxazole (SMX)).
Additionally, the wider range of ozone doses was investigated during Concentrations of antibiotics were chosen to enumerate resistant mi­
additional experiments at two of the sampling campaigns (in September croorganisms based on information provided by the Clinical and Labo­
and in March). Within these campaigns, 3 additional ozone doses were ratory Standards Institute CLSI (2017). Serial dilutions of samples were
tested (besides the standard ozone dose) from the wider range of ozone prepared in sterile 0.85% (w/v) NaCl solution in four replicates target­
doses between 0.2 and 0.9 g O3/g DOC. Finally, the activated carbon ing 10–80 colonies per plate. Prepared dilutions were vacuum filtered
filter was filled with 1.8 m3 of granular activated carbon (GAC), type through cellulose acetate filters (47 mm Ø, pore size 0.45 µm, Pall Life
Epibon A (Donau Carbon, Frankfurt, Germany) and treated a side stream Sciences, USA). The filter was transferred onto the medium and incu­
of 8 m3 /h, which resulted in an empty bed contact time of 13.5 min bated for 24 h at 37 ◦ C (CCA medium) or for 5 d at 37 ◦ C (R2A medium).
Blue (E. coli) and pink (coliforms) colonies were counted on CCA me­
2.2. Sampling scheme dium, and all colonies were counted on R2A plates. The lower and upper
quantification limits (LOQ) were estimated as 5 and 150 colonies per
Sampling campaigns were performed between May 2018 and May plate, respectively. Results outside this range were disregarded, giving a
2019. Grab samples of CAS effluent (CAS-OUT), effluent after ozonation total dataset of n = 13 for total heterotrophs, n = 17 for total Escherichia
(O3-OUT), and final effluent after GAC filter (GAC-OUT) were collected coli, n = 10 for Escherichia coli resistant to 32 mg/L ampicillin, n = 11 for
on a monthly basis for 13 consecutive months. Therefore, 13 samples Escherichia coli resistant to 4 mg/L trimethoprim and 76 mg/L
were collected from each sampling site (5 L of CAS-OUT, 10 L of O3-OUT sulfamethoxazole.
and 10 L of GAC-OUT). During September and March sampling, the
three extra ozone doses were investigated (besides the standard dose), 2.6. DNA extraction
resulting in additional three O3-OUT samples. The quality of the effluent
after CAS treatment regarding tested ARB and ARGs did not fluctuate Samples were shaken vigorously and vacuum filtered through

3
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

cellulose acetate filters (47 mm Ø, pore size 0.45 µm, Pall Life Sciences, abundance after ozonation at higher and lower ambient temperatures.
USA) in at least 3 replicates. This type of filter was required to collect The used p-value was 0.05.
sufficient biomass for DNA extraction from samples of GAC-OUT and
raw wastewater - influent to the WWTP. Filters were cut into pieces, 3. Results
transferred into Eppendorf tube, and stored at − 20 ◦ C until further
processing. Volumes for filtration were chosen individually, depending 3.1. The correlation of relative decrease of UV254 absorption and nitrite
on the sample’s predicted biomass concentration and turbidity to allow compensated specific ozone dose
maximal DNA yield (0.4–0.8 L of CAS-OUT, 0.8–1.2 L of O3-OUT, 1–2.5
L of GAC-OUT). Total DNA from the filters was extracted with the As expected, the observed relative decrease of UV254 absorption
DNeasy PowerWater Kit (Qiagen, Germany) according to the provided (ΔUV254) was found to strongly correlate with the nitrite compensated
protocol, except for the following modifications: (i) a 2 mL bead beating specific ozone dose that was compensated for the ozone consumption by
tube was used instead of a 5 mL tube, (ii) a cell lysis step was performed nitrite during the monitoring campaign (Fig. 1), thus providing a solid
in a MP Biomedicals™ FastPrep 4™ 5 G Instrument for 2 × 40 s at 6 m/s, base for further data analysis and correlations. The obtained linear
(iii) the columns were air-dried for at least 5 min before addition of regression was of R2 = 0.89, and p-value = 3.95E-09, and Pearson
elution buffer, (iv) the columns were incubated for at least 2 min with correlation: R2 = 0.94, p-value = 4.61E-09.
elution buffer. The quality and concentration of extracted DNA were
measured with an Eppendorf BioSpectrometer® and extracts were 3.2. Inactivation of ARB and ARGs
stored at − 80 ◦ C until further processing.
The data for ARB&ARGs abundance collected during the one-year
2.7. Quantitative PCR analysis monitoring campaign and two one-day sampling campaigns were
correlated with ozone dose and ΔUV254. Regression models were
Targeted genes: 16 S rRNA gene and ARGs (sul1, ermB, blaTEM-1 and computed in R and the best fit was chosen for each dataset (Table S4).
tetW) were quantified by TaqMan assays, designed and optimized by For most of the targets, the adjusted R2 and p-value for linear regression
Ingenetix GmbH (Vienna, Austria). All reactions were performed in a were higher when correlated with ΔUV254 than with nitrite compen­
Roche Light-Cycler 480 (Roche Applied Science, Germany) in a 10 μL sated specific ozone dose (except for ermB gene). The values were above
reaction mixture, containing 1 x LightCycler® 480 Probes Master 0.5 (R2) and below 0.01 (p) for targets except for total heterotrophs,
(Roche, Germany), 1 x TaqMan assay and 2 μL of a sample, according to blaTEM-1 and ermB genes. We suggest that this is a promising result in the
the protocols in Table S2. All samples and standards were assayed in direction of application of ΔUV254 as a surrogate parameter in the
triplicates. Standard curves were prepared for each run by 10-fold monitoring of ARB and ARGs during ozonation.
dilution of the standard, ranging 107–101 copies for the investigated An increase in ozone dose did not result in changes in the log removal
ARGs and 108–103 copies for the 16 S rRNA gene. The amplification of total heterotrophic bacteria, which was between − 0.81–0.73 with no
efficiency ranged 90–105%. LOD and LOQ of qPCR assays are described found correlation. The log removal of total Escherichia coli and Escher­
in Table S3. ichia coli resistant to 32 mg/L AMP, and 4 mg/L TMP with 76 mg/L SMX
revealed a correlation with both nitrite compensated specific ozone dose
2.8. Data analysis and ΔUV254 in treated effluent (Fig. 2). The log removal value for bac­
terial targets was increasing until reaching a plateau around 2.50 logs
The absolute abundance of total and antibiotic resistant bacteria for total E. coli (corresponding to ΔUV254 of 30% and nitrite compen­
revealed by culture-based methods was presented as colony forming sated specific ozone dose of 0.5 g O3/g DOC) or 1.99 logs for TMP and
units per mL of the sample (CFU/mL) according to Novo and Manaia SMX resistant E. coli (corresponding to ΔUV254 of 25% or nitrite
(2010). The absolute abundance of 16 S rRNA gene and ARGs was given compensated specific ozone dose of 0.4 g O3/g DOC). For AMP resistant
as copies/mL of the sample. The relative abundance of ARGs was E. coli, the plateau phase was not as evident, and 2.50 log inactivation
calculated as a ratio of ARGs abundance to 16 S rRNA gene abundance. was reached with 45% ΔUV254 and 0.9 g O3/g DOC.
The log removal values (LRV) of total and antibiotic resistant bacteria, The ANOVA test revealed significant differences between log re­
and targeted genes were calculated according to Von Sperling et al. movals of total and antibiotic resistant E. coli at the tested ozone doses
(2018). For ozone treatment, CAS-OUT and O3-OUT values were taken, (p-value = 4.43E-07, Table S5). The Tukey test showed that results for
and for GAC treatment, O3-OUT and GAC-OUT were taken to calculate 0.02, 0.18, 0.25 and 0.92 g O3/g DOC (from both monitoring and one-
LRV. Positive LRV represents a reduction of the target, and negative LRV day sampling campaign data) differed significantly from results at
an increase of its abundance after treatment. other ozone doses (Table S6).
R software (version 4.0.4., R Core Team, 2021) was used for statis­ We did not observe fluctuations in the absolute abundance of bac­
tical analyses and visualization of the outcomes (main packages: corr­ terial targets in CAS-OUT samples; the reported abundance values var­
plot, pheatmap, drc, stats, qpcR, devtools). Regression models for ozone ied within the same order of magnitude (Table S7). For total E. coli, the
doses, the ΔUV254, bacteria and genes were computed (linear, expo­ average absolute abundance in CAS-OUT during the sampling campaign
nential, logarithmic, logistic and log-logistic with varied numbers of was 187 ± 248 CFU/mL (median = 120 CFU/mL).
parameters). In addition, Pearson correlations between datasets and Log removal values of targeted genes increased with increasing
heatmaps were also computed and visualized. Targets were considered ΔUV254 (Fig. 3). Moreover, higher ΔUV254 values were required to
correlated for R2 > 0.5, and p < 0.05. Relatedness with values of 0.5 < achieve similar log removal values as for the inactivation of E. coli. Data
R2 < 0.7 was considered as a moderate correlation and R2 > 0.7 as a did not show a plateau phase and did not support the estimation of
strong correlation. higher asymptotes with statistical relevance. The qPCR signals for ermB
Data normality was tested with Shapiro-Wilk test and the equality of and blaTEM-1 gene were in some cases close to the LOQ value (although
the variances was evaluated with Leven’s test. To test statistical signif­ above LOQ). The results for observed ΔUV254 significantly varied from
icance between the datasets, ANOVA was used for normally distributed each other (p-value = 6.04E-05) Table S8.
data and Kruskal-Wallis test was used for not normally distributed data.
To reveal significantly different groups within datasets, the following 3.3. Changes in ARGs relative abundance
tests were applied: Tukey test for normally distributed data and Dunn
test for not normally distributed data. T-test was used to determine the The absolute abundance of ARGs was related to the absolute abun­
significance of the difference between the change in ARGs relative dance of the 16 S rRNA gene to calculate the relative abundance of ARGs

4
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 1. Correlation between nitrite compensated specific ozone dose and ΔUV254 (linear regression R2 = 0.89, p-value = 3.95E-09) for data collected during one-year
monitoring including two daily campaigns, n = 18.

indicating their share in bacterial communities of collected samples. For between 0 % and 30 % (nitrite compensated specific ozone doses be­
genes sul1 and tetW, a higher relative abundance was observed indi­ tween 0 and 0.5 g O3/g DOC), (ii) ΔUV254 between 30 % and 50 %
cating that these genes were more abundant within the wastewater (0.5–0.9 g O3/g DOC), and (iii) the highest ΔUV254 (equal 50.34 %,
bacterial community (in a range of 0 – 0.15 and 0 – 0.20, respectively) corresponding to 0.92 g O3/g DOC) The first cluster (i.e., 0–30 %),
than ermB and blaTEM-1 genes (ranging from 0 to 0.015 and 0–0.08, with corresponds to the majority of nitrite compensated specific ozone doses
one signal at 0.24, respectively). below 0.5 g O3/g DOC, and the second cluster – above 0.5 g O3/g DOC.
To visualize how ozone treatment affected ARGs relative abundance, The patterns of targeted bacteria and genes removal differed visibly
a change in the relative abundance was calculated for each data point between these two clusters. For the cluster < 0.5 g O3/g DOC, log
(monitoring campaign and one-day sampling campaigns) as a percent­ removal values of bacteria and genes were lower than for the second
age increase or decrease (taking CAS-OUT sample as a starting point for cluster, which was especially visible for the genes. Within the second
calculation). The average decrease in the relative abundance of ARGs cluster, the LRV of bacterial targets were at rather similar levels even
was at a similar level for all tested genes and ranged from 51% (sul1) to with increasing ozone dose. However, the LRV for genes increased with
68% (blaTEM-1), Tab. S9. For ermB and blaTEM-1 genes, the increase in increasing ozone dose, which was especially visible for 16 S rRNA gene
relative abundance after ozonation was higher than for sul1 and tetW. and tetW. The highest ΔUV254 (corresponding to 0.92 g O3/g DOC) was
The differences between the relative abundance of ARGs at different situated on a simplicifolious clade and showed the highest log removal
UV254 were not statistically significant (p-value = 0.30, Tab. S10). values for all tested genes.
Furthermore, the changes in relative abundance were related to Similarities of response patterns between different genes, as well as
sampling dates and average changes in air temperature at the sampling total and antibiotic resistant E. coli, suggested that there could be cor­
site. We hypothesized that the changes in ARGs relative abundance after relations between the behaviour of targeted bacteria and/or genes,
ozonation may be related to seasonal changes in bacterial communities. which was further investigated. Log removals of the monitored targets,
The T-test aimed to estimate the significance of differences between as well as the relative abundance of ARGs, were correlated with each
changes in ARGs relative abundance when air temperatures were below other by Pearson correlations (Fig. 6). The log removal of AMP and
or equal to 14 ◦ C (n = 6) versus above 14 ◦ C (n = 5). The 14 ◦ C was the TMP/SMX resistant E. coli correlated negatively with the relative
median value for the temperature dataset. The data points from the one- abundance of sul1 gene. This observation suggests a significant rela­
day sampling campaigns for ozone doses outside the standard operating tionship between the increase in log removal of resistant E. coli and the
range were excluded to not affect the number of data points at the same decrease in the relative abundance of these genes, thus, their share in the
air temperature. Obtained p-values were < 0.05 for each gene (0.003 for bacterial community.
sul1, 0.038 for ermB, 0.035 for tetW) except for blaTEM-1 (0.406). Plotting Total and resistant E. coli as well as tested genes (except blaTEM-1)
sul1 and ermB relative abundance changes against sampling dates sug­ correlated positively with changes in nitrite compensated specific ozone
gested a seasonality in change of these ARGs abundance after ozonation. dose and ΔUV254. Among bacteria, the strongest correlation was found
Correlation between these changes and air temperature resulted in for AMP E. coli (R2 = 0.93) and among genes for tetW (R2 = 0.9).
asymmetrical bell-shape curves with similar centre locations around
7 ◦ C (Fig. 4A-B). For tetW and blaTEM-1 genes, this correlation could not
be observed (Fig. 4C-D). 3.5. GAC filter

In this study, a GAC filter applied in a multibarrier system for CECs

3.4. Correlations between response patterns abatement as a final step was investigated for its efficiency in reducing
ARB and ARGs abundance in the final effluent. The residual abundance
Log removals of targeted bacteria and genes for each ΔUV254 were of the total, AMP and TMP/SMX resistant E. coli after ozone treatment
clustered into dendrograms and visualized with a heatmap (Fig. 5). The was slightly reduced after GAC filtration (Fig. 7). However, the total
response patterns clustered into 3 major distinct groups: (i) ΔUV254 number of heterotrophs was up to 1 log higher in the final effluent of the

5
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 2. Calculated log removal values for (A) total heterotrophs, n = 13, (B) total Escherichia coli, n = 17, (C) Escherichia coli resistant to 32 mg/L ampicillin,
n = 10, and (D) to 4 mg/L trimethoprim and 76 mg/L sulfamethoxazole, n = 11 with regression models (log-logistic). The results outside LOQ range were
not considered.

multibarrier system (after GAC) than in ozonated effluent. In general, a times.

reduction in 16 S rRNA gene, sul1, and tetW ARGs abundance could be
observed in GAC-OUT (up to 1 log). Changes in LRV of ermB gene varied 4. Discussion
strongly from close to log 1 reduction and close to 1 log increase. In the
case of blaTEM-1 gene, the majority of LRV was negative, suggesting an 4.1. Application of ΔUV254 as an indicator for ARB&ARGs inactivation
increase in its abundance in GAC-OUT. by ozonation
For most sampling points, a reduction in sul1 and tetW relative
abundance was observed in GAC-OUT samples (Tab. S11). However, in Monitoring of ARB and ARGs inactivation in wastewater by ozona­
the case of ermB gene, close to 50% of sampling points showed a tion involves costly and time-consuming lab procedures that do not
reduction in relative abundance at a similar level as for sul1 and tetW. provide an immediate result (e. g. due to bacteria incubation). There­
For the rest of the data points, the average ermB relative abundance was fore, the application of a surrogate parameter is of great benefit in the
approximately 70 times higher than before GAC. This trend was even monitoring of ARB and ARGs abatement by ozonation (Dickenson et al.,
stronger for blaTEM-1 gene, with only one data point reporting a reduc­ 2009). Prediction models and correlations of chemical CECs abatement
tion in its relative abundance and an average increase close to 3000 by ozonation with ΔUV254 in wastewater have been previously well

6
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 3. Calculated log removal values for (A) 16 S rRNA gene, n = 17, (B) sul1 gene, n = 17, (C) ermB gene, n = 17, (D) blaTEM-1 gene, n = 15, (E) tetW gene, n = 17
with regression models (log-logistic). The results outside LOQ range were not considered.

developed and indicated that ΔUV254 may be used as a surrogate compensated specific ozone dose and ΔUV254 (linear regression: R2 =
parameter for the elimination of these targets (Gerrity et al., 2012). This 0.88, and p-value = 3.95E-09), which was reported before as well (Bahr
parameter was also tested with E. coli and other bacteria but not for ARB et al., 2007; Nöthe et al., 2009). The observed correlation between
and ARGs (Gerrity et al., 2012; Lee et al., 2016; Wu et al., 2018). ΔUV254 and total E. coli inactivation was similar to one observed by
Moreover, ΔUV254 is suggested as a surrogate parameter for estimating Gerrity et al. (2012) with an R2 value for linear regression comparable to
ozone exposure in real wastewater systems for chemical CECs (Buffle our findings (0.50 in our study compared to 0.47). The plateau phases
et al., 2006) and even for some bacteria and viruses (Gerrity et al., 2012; for bacteria and lag phases for genes caused difficulties with fitting the
Wu et al., 2018). Since the ARB and ARGs inactivation by ozone depends regression models and resulted in weaker linear regression fits. In gen­
on ozone exposure (Michael-Kordatou et al., 2018; Rizzo et al., 2020), eral, for tested ARB and ARGs, better fits of regression models and
linking the ARB and ARGs abatement by ozone with ΔUV254 could stronger linear and Pearson correlations were obtained with ΔUV254
provide additional information on their response to ozonation. There­ than nitrite compensated specific ozone dose. This suggests that corre­
fore, we decided to link responses of various target ARB and ARGs at lating ARB and ARGs log removal data with the ΔUV254 could better
chosen nitrite specific ozone doses with ΔUV254. represent these biological CECs exposure to ozone in wastewater, which
In our study, we have observed a strong correlation between nitrite was also suggested by Wu et al. (2018).

7
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 4. Changes in relative abundance of ARGs during ozone treatment related to average air temperature at sampling location for: (A) sul1 gene with fitted
polynomial regression (adjusted R2 = 0.71, p-value = 0.008, 3rd degree), n = 11, (B) ermB gene with fitted polynomial regression (adjusted R2 = 0.38, p-value =
0.10, 3rd degree), n = 11, and (C) tetW gene, n = 11, (D) blaTEM-1 gene, n = 8. Negative change represents reduction of the gene abundance by a shown percent.

These observations are promising results in the direction of appli­ was reported by Gerrity et al. (2012) in their lab-scale ozone disinfection
cation of ΔUV254 as a surrogate parameter in the monitoring of ARB and of E. coli providing a regression model described by a sigmoidal curve
ARGs during ozonation. Further studies in this direction are required with a plateau phase (approximately 40 % vs. 30% of ΔUV254 in our
and should either involve targeting the range of ozone doses that result study). Wu et al. (2018) also observed a steady decrease of E. coli
in correlations as linear as possible or in multiple long-term studies that cultivability (CFU/mL) up to a specific ozone dose of 0.4 g O3/g DOC
would help in generating models predicting the behaviour of target ARB (which would correspond to approximately ΔUV254 of 25 % in our
and ARGs at chosen ΔUV254. study) and no significant further decrease for higher ozone doses.
Even though regression models for total and resistant E. coli showed
4.2. Inactivation of ARB and ARGs by ozone treatment strong similarities in our results, the ΔUV254 required to reach the
plateau phase varied between these two targets. For total E. coli, the
For total and antibiotic resistant E. coli and tested ARGs, the log plateau was reached with ΔUV254 of approximately 30 % (0.51 g O3/g
inactivation increased with increasing ΔUV254 (and therefore nitrite DOC), for TMP/SMX resistant E. coli with ΔUV254 of 25 % (0.39 g O3/g
compensated specific ozone dose). However, for bacterial targets, above DOC), and for AMP resistant E. coli with 45% ΔUV254 (0.87 g O3/g
a certain ΔUV254, a further increase in ozone dose did not result in a DOC). Alexander et al. (2016) suggested that these differences may
higher abatement, reaching a plateau phase above which no further result from varying levels of susceptibility against oxidative stress be­
reduction of E. coli cultivability could be achieved. A similar observation tween the bacteria resistant or susceptible to different types of

8
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 5. Heatmap with dendrogram for data collected for

each ΔUV254. Log removal values for total and resistant E.
coli as well as for ARGs and 16 S rRNA gene were clustered
against ΔUV254 (presented in the Figure as dUV). Each
ΔUV254 corresponds to the nitrite compensated specific
ozone dose applied in the study. The applied ozone doses
were divided into three ranges (dO3_range: 0–0.5, 0.5–0.9,
and 0.9), and corresponding samples were marked with
colours to visualize the differences in responses between
the ranges, especially for doses below and above 0.5 g O3/
g DOC. Analysed targets were total Escherichia coli, n = 17
(“Total E. coli”), Escherichia coli resistant to 32 mg/L
ampicillin, n = 10 (“AMP res. E.coli”), and to 4 mg/L
trimethoprim and 76 mg/L sulfamethoxazole, n = 11
(“TMP/SMX res. E.coli”), 16 S rRNA gene, n = 17, sul1
gene, n = 17, ermB gene, n = 17, blaTEM-1 gene, n = 13,
and tetW gene, n = 17.

Fig. 6. Pearson correlations between monitored targets (only significant ones are shown, p-value > 0.05) for data collected during the one-year monitoring campaign
(13 consecutive months) and two one-day sampling campaigns. ΔUV254 was presented in the Figure as dUV and applied in the study nitrite compensated specific
ozone doses as O3-dose. AMP E. coli represents AMP resistant E. coli and TMP/SFX E. coli - TMP/SFX resistant E. coli.

9
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

Fig. 7. Box plot with whiskers from minimum

to maximum, with the displayed median value
for log removal values calculated for GAC
treatment (taking O3-OUT and GAC-OUT as
before and after treatment). Analysed targets
were total heterotrophs, n = 9, total Escher­
ichia coli, n = 11 (“Total E. coli”), Escherichia
coli resistant to 32 mg/L ampicillin, n = 9
(“AMP res. E. coli”), and to 4 mg/L trimetho­
prim and 76 mg/L sulfamethoxazole, n = 4
(“TMP/SMX res. E. coli”), 16 S rRNA gene,
n = 11, sul1 gene, n = 11, ermB gene, n = 11,
blaTEM-1 gene, n = 9, and tetW gene, n = 11.
Positive log removal represents the reduction of
the gene abundance by a shown log.

antibiotics. The oxidative stress may be induced by reactive oxygen against ozone.
species (ROS) generated by the interaction of bactericidal antibiotics The abundance of total heterotrophic bacteria (based on CFU results)
with their classical targets and activated ROS protective responses oscillated between 1 log reduction and 1 log increase after treatment.
(Acker and Coenye, 2017; Kohanski et al., 2007). The trace abundance The applied medium would be suitable for bacteria thriving in both bulk
of antibiotics (and potentially their transformation products formed and flocs. The abundance of flocs, which are three-dimensional high
during ozonation (Szabó et al., 2016)) could induce the antibiotic cell-density structures, result in difficulties in a reliable enumeration of
resistance adaptation response as well as sub-lethal stress response. total heterotrophic bacteria directly from wastewater samples on agar
These mechanisms would activate anti-oxidative and oxidative plates, as one CFU can represent several bacteria within an activated
damage-associated responses against produced ROS (e.g., recA sludge floc that may be split up to several smaller units and thus CFUs in
response) in ARB (Alexander et al., 2016). Therefore, ARB that can ozonation. Therefore, total heterotrophs seem to not be suitable for an
effectively induce these responses have an advantage over non-resistant indicator of the efficiency of total bacteria disinfection by ozone treat­
bacteria in dealing with the damages caused by ozonation. Our obser­ ment, and changes in 16 S rRNA gene abundance should be used instead.
vation of reaching the plateau phase for cultivability of AMP resistant Since E. coli is rather associated with bulk (not flocs), the relationship
E. coli at ΔUV254 higher than for TMP/SMX resistant E. coli would sup­ between ΔUV254 and changes in CFU could be observed. It is worth
port this hypothesis (AMP is a bactericidal antibiotic in contrast to TMP noting that the presence of particles can protect microorganisms: bac­
and SMX, which are bacteriostatic). teria abundant at > 12 µm into the floc were reported to be protected
The log removal values of targeted genes increased with increasing from ozone treatment (Czekalski et al., 2016).
ozone dose and did not reach a plateau phase. Moreover, higher ozone
doses were required to achieve similar LRV of ARGs as compared to ARB
4.3. Recommended ozone doses for ARB&ARGs abatement in wastewater
(e.g., ΔUV254 of 30 % resulted in 2 logs inactivation of AMP resistant
E. coli and approximately 1 log abatement of sul1, tetW and ermB gene).
Specific ozone doses between 0.4 and 0.6 g O3/g DOC (nitrite
The need for a higher ozone dose to achieve DNA damage or leakage
compensated) were reported to be efficient for chemical CECs abate­
from attacked bacteria cells was also observed by other authors (Cze­
ment from wastewater (Rizzo et al., 2020). Doses above 0.5 g O3/g DOC
kalski et al., 2016; Lee et al., 2016; Wu et al., 2018). At lower ozone
were suggested as feasible to achieve significant levels of microbial
doses, a certain lag phase could be observed, except for ermB (up to
inactivation (Von Sonntag and Von Gunten, 2012). Inactivation of ARB
20–25% ΔUV254, thus, 0.4 g O3/g DOC for 16 S rRNA gene, sul1 and
and ARGs at similar ozone doses as required for chemical CECs abate­
tetW, and approximately 35 % ΔUV254 for blaTEM-1, thus 0.6 g O3/g
ment would be highly beneficial for the operation of wastewater ozon­
DOC). Furthermore, the observed lag phases seem to correspond to the
ation and its application in real-life conditions. Since wastewater is a
plateau phases for bacterial targets. Cho et al. (2010) and Wu et al.
complex matrix, which composition fluctuates over time, the control
(2018) reported that the lag phase was finished when sufficient mem­
and adaptation of the effective applied ozone dose are vital for the
brane damage was reached to cause DNA leakage or for ozone to
efficient abatement of targets.
penetrate into cell plasma and cause further DNA damage. This would
In our study, a significant difference between the pattern in bacteria
suggest that ΔUV254 of 30% (0.5 g O3/g DOC) at the link between the
and gene responses to ozone treatment at ΔUV254 below and above 30%
start of plateau phases for bacteria and the end of lag phases for genes
(0.5 g O3/g DOC) was observed. The loss of cultivability of bacterial
could correspond to reaching sufficient membrane impairment to cause
targets was already reached below 0.5 g O3/g DOC while a significant
DNA damage. Previous works (Czekalski et al., 2016; Wu et al., 2018)
log removal of genes could be observed above 0.5 g O3/g DOC. This
reported significant lag phases for DNA damage until approx. 0.4 g O3/g
corresponds to the overlap of plateau phases for bacterial targets and lag
DOC. The higher lag phase observed for blaTEM-1 gene may be correlated
phases for genes around ΔUV254 of 30 %. This would implicate that for
with higher ΔUV254 required to reach a plateau phase for AMP resistant
tested targets (except for AMP resistant E. coli and blaTEM-1) a sufficient
E. coli, suggesting higher robustness of ampicillin resistant bacteria
membrane impairment resulting in significant DNA damage or leakage

10
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

would be reached at around 0.5 g O3/g DOC. Czekalski et al. (2016), Lee genes.
et al. (2016) and Wu et al. (2018) suggested a similar scheme of bacteria
cell inactivation: the effects on cultivability (CFU) were followed by the 4.5. Multibarrier approach: GAC filter as a post-treatment after ozonation
effects on membrane permeability leading to significant DNA damage.
The chosen ozone dose should ensure not only membrane but also Activated carbon (AC) treatment has been widely investigated for the
DNA damage to (i) minimize the abundance of intact ARGs that may removal of chemical CECs from wastewater (Rizzo et al., 2020) and was
remain within the resulting cell debris even after the loss of ARB suggested as a post-treatment after ozonation to minimize potentially
viability (Dodd, 2012), and (ii) maximize the negative effects on cell formed by-products and maximize the further removal of chemical CECs
components and thus minimize the effectiveness of possible repair (Ravasi et al., 2019). The impact of AC application on ARB and ARGs
mechanisms induced by oxidative stress. Moreover, the bacteria resis­ abatement from the effluent is not yet clear with different studies
tant to bactericidal antibiotics may exhibit higher robustness towards reporting limited removal or even increased abundance after treatment
ozone since the overlap between the plateau and lag phases for ampi­ by various adsorption/filtration technologies (Hiller et al., 2019). In our
cillin resistant E. coli and blaTEM-1 resistance gene occurred at higher study, the GAC filtration resulted in a further abatement for TMP/SMX
ΔUV254 than for other targets (> 0.6 g O3/g DOC). Our findings suggest E. coli, sul1 and tetW genes, and a slight decrease or no effects on 16 S
that the ozone doses required for inactivation of tested ARB and ARGs rRNA gene, AMP E. coli, and total E. coli abundance. Czekalski et al.
correspond to more than ~35% ΔUV254 or ~0.6 gO3/gDOC. These (2016) observed that the application of sand filtration as a
values are similar to the findings of Wu et al. (2018). post-treatment resulted in the increase of sul1 and 16 S rRNA genes
abundance after passing the sand filter. In our study, an increase in the
4.4. ARGs selection during ozone treatment both absolute and relative abundance of ermB and blaTEM-1 genes could
be observed after passing the GAC filter, and it was significantly higher
A few studies have reported an increase in abundance after ozone for the latter gene. Lüddeke et al. (2015) also reported that filter pas­
treatment for various ARGs (e.g., of vanA and blaVIM by Alexander et al., sages of ozonated effluent partially caused an increase in the resistance
2016, of sul1 and tetG by Zhuang et al., 2015) suggesting a possible level. The significant increase in blaTEM-1 gene relative abundance fol­
selection during ozone treatment for bacteria possessing these genes. lows the previous observation of higher robustness of AMP resistant
The selection of bacteria carrying ARGs alongside with the general E. coli and this corresponding ARG against ozone. This may suggest that
reduction of bacterial community diversity may lead to a proliferation of bacteria less susceptible to stress caused by ozone treatment may prevail
ARB in emptied niches in treated wastewater and their further dissem­ in the GAC filter and be abundant in the final effluent.
ination into receiving environments. Therefore, we monitored changes
in relative abundance after ozonation at tested nitrite compensated 5. Conclusions
specific ozone doses to evaluate if (i) selection of specific ARGs occurs,
and (ii) if there may be a correlation between ozone dose and selection The multibarrier demonstrator plant combining ozonation and GAC
of specific genes. filtration for chemical CECs removal from CAS effluent was monitored
Changes in ARGs relative abundance after ozonation varied strongly and assessed for its suitability for ARB and ARGs abatement during 13
between the genes. The increase in the relative abundance of tested consecutive months of routine operation.
ARGs after ozone treatment was observed for up to 45% of datapoints
(depending on the gene) and was higher for blaTEM-1 and ermB than for • For total E. coli, TMP/SMX resistant E. coli and ARGs: sul1, ermB, and
sul1 and tetW genes. These observations suggest that factors favouring tetW, sufficient membrane damage leading to significant DNA
especially the first two ARGs’ bacterial hosts could be of importance. impairment or leakage from affected cells was observed at around
The observed increases and decreases of relative abundance did not ΔUV254 of 30% (corresponding to 0.5 g O3/g DOC). For AMP resis­
significantly correlate with ΔUV254 or suspended solids. Therefore, we tant E. coli and blaTEM-1 gene, the overlap was observed at a slightly
further investigated if these changes may be connected with seasonal higher ΔUV254 of around 35–40% (corresponding to around
shifts in bacteria type and abundance resulting from changes in bacterial 0.7 g O3/g DOC). After Alexander et al. (2016), we hypothesized that
community structure at WWTPs. Such changes were reported to occur for these targets, the lower susceptibility towards ozone may be
twice a year and repeatedly assemble into seasonal steady states connected with sub-lethal stress response mechanisms in ARB resis­
(LaMartina et al., 2021). For genes: sul1 and ermB, we could observe tant to bactericidal antibiotics. This hypothesis could be investigated
seasonal shifts in response to ozone treatment: a decrease in relative in future studies bringing more insight into the mechanisms of ARB
abundance clustered from May to December and an increase from reaction to stress conditions.
January to May next year. These shifts correlated with changes in air • Reaching the ozone dose resulting in DNA damage is vital for
temperature with a maximum increase in relative abundance occurring maximizing the efficiency of ARB and ARGs abatement by ozone. The
around 7 ◦ C and no changes or decrease at extrema: temperature around results suggest that the ozone doses required for inactivation of
0 and 20 ◦ C. Those trends could not be observed for blaTEM-1 and tetW tested ARB and ARGs correspond to more than ~35% ΔUV254 or
genes, which would suggest that these genes are carried by different ~0.6 g O3/g DOC, to ensure reaching sufficient DNA damage.
groups of hosts in the WWTPs bacterial community. According to Yin • The increase of ARGs relative abundance after ozonation varied be­
et al. (2019), the abundance of sul1 gene in activated sludge was tween genes and was higher for blaTEM-1 and ermB than for sul1 and
strongly correlated with α-Proteobacteria and γ-Proteobacteria like tetW. The changes did not significantly correlate with ΔUV254 or
Pseudomonas aeruginosa, and ermB gene was correlated with δ- and suspended solids. However, for sul1 and ermB, these changes
γ-Proteobacteria. Beta-lactam and tetracycline resistance genes were appeared to be seasonal with a decrease in relative abundance from
linked to Actinobacteria, the former also with Enterobacteriaceae and the May to December and an increase from January to May, which we
latter with Firmicutes. Furthermore, seasonal shifts in WWTPs bacterial hypothesize to be strongly dependent on major hosts of targeted
community were reported to be driven mostly by genes, their origin, and seasonal abundance fluctuations.
environment-associated taxa, whereas human-associated taxa seemed to • The GAC filtration resulted in a further abatement of TMP/SMX
not follow seasonal trends (LaMartina et al., 2021). Thus, since blaTEM-1 E. coli, sul1 and tetW genes, and a slight decrease or no effects on the
and tetW genes are rather linked to human-associated taxa (Firmicutes, 16 S rRNA gene, AMP E. coli, and total E. coli abundance. The sig­
Enterobacteriaceae and other γ-Proteobacteria), the effects of hosts nificant increase in blaTEM-1 gene relative abundance follows the
abundance seasonality on changes in their relative abundance after previous observation of higher robustness of AMP resistant E. coli
ozone treatment could not be observed, on the contrary to sul1 and ermB and this associated ARG against ozone.

11
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

• ΔUV254 is considered a surrogate parameter for chemical CECs and induce strong population shifts. Sci. Total Environ. 559, 103–112. https://doi.
org/10.1016/j.scitotenv.2016.03.154.
abatement and ozone exposure in real wastewater systems. Our ob­
Bahr, C., Schumacher, J., Ernst, M., Luck, F., Heinzmann, B., Jekel, M., 2007. SUVA as
servations indicate that correlating ARB and ARGs log removal data control parameter for the effective ozonation of organic pollutants in secondary
with the ΔUV254 gives a better representation of these biological effluent. Water Sci. Technol. A J. Int. Assoc. Water Pollut. Res. 55 (12), 267–274.
CECs exposure to ozone in wastewater. https://doi.org/10.2166/WST.2007.418.
Berendonk, T.U., Manaia, C.M., Merlin, C., Fatta-Kassinos, D., Cytryn, E., Walsh, F.,
Bürgmann, H., Sørum, H., Norström, M., Pons, M.-N., Kreuzinger, N., Huovinen, P.,
Environmental Implication Stefani, S., Schwartz, T., Kisand, V., Baquero, F., Martinez, J.L., 2015. Tackling
antibiotic resistance: the environmental framework. Nat. Rev. Microbiol. 13 (5),
310–317. https://doi.org/10.1038/nrmicro3439.
The worldwide increase of antibiotic resistance (AR) levels is one of Bourgin, M., Beck, B., Boehler, M., Borowska, E., Fleiner, J., Salhi, E., Teichler, R., von
the biggest threats to global health and food security. Wastewater Gunten, U., Siegrist, H., McArdell, C.S., 2018. Evaluation of a full-scale wastewater
treatment plant upgraded with ozonation and biological post-treatments: abatement
treatment plants (WWTPs) are identified as one of the most important
of micropollutants, formation of transformation products and oxidation by-products.
AR propagation routes from humans to the environment. AR is only Water Res. 129, 486–498. https://doi.org/10.1016/j.watres.2017.10.036.
partially removed in biological wastewater treatment, which indicates a Buffle, M., Schumacher, J., Meylan, S., Jekel, M., 2006. Ozonation and advanced
need for advanced treatment. We investigate the removal of AR targets oxidation of wastewater: Effect of O3 Dose, pH, DOM and HO•-scavengers on ozone
decomposition and HO• generation. Ozone. Sci. Eng. 28, 247–259. https://doi.org/
from WWTPs effluent by ozonation followed by granular activated 10.1080/01919510600718825.
carbon filtration in a long-term pilot-scale study. Our results are of a Cho, M., Kim, J., Yeon, J., Yoon, J., Kim, J., 2010. Mechanisms of Escherichia coli
practical relevance to WWTPs operators, for future legislations, and help inactivation by several disinfectants. Water Res. 44 (11), 3410–3418. https://doi.
org/10.1016/j.watres.2010.03.017.
to address AR dissemination problems. CLSI, 2017. Performance standards for antimicrobial susceptibility testing. 27th ed. CLSI
supplement M100. Wayne, PA: Clinical and Laboratory Standards Institute. In
Performance standards for antimicrobial susceptibility testing. 27th ed. CLSI
CRediT authorship contribution statement
supplement M100. Wayne, PA: Clinical and Laboratory Standards Institute. 〈https://
doi.org/10.1039/C4DT01694G〉.
K. Slipko: Conceptualization, Methodology, Formal analysis, Inves­ Czekalski, N., Imminger, S., Salhi, E., Veljkovic, M., Kleffel, K., Drissner, D., Hammes, F.,
tigation, Software, Writing – original draft, Writing - review & editing, Bürgmann, H., Von Gunten, U., 2016. Inactivation of antibiotic resistant bacteria and
resistance genes by ozone: from laboratory experiments to full-scale wastewater
Visualization; D. Reif: Conceptualization, Methodology, Investigation, treatment. In: Environ. Sci. Technol., 50, pp. 11862–11871. https://doi.org/
Writing – review & editing; H. Schaar: Conceptualization, Methodol­ 10.1021/acs.est.6b02640.
ogy, Investigation, Writing – review & editing; E. Saracevic: Method­ Dickenson, E.R.V., Wert, E.C., Snyder, S.A., 2009. Appl. Surrog. Indic. Assess. Remov.
Effic. Trace Org. Chem. Chem. Oxid. Wastewaters 43 (16), 6242–6247.
ology, Investigation, Resources; A. Klinger: Methodology, Dodd, M.C., 2012. Potential impacts of disinfection processes on elimination and
Investigation; L. Wallmann: Methodology, Investigation; J. Krampe: deactivation of antibiotic resistance genes during water and wastewater treatment.
Resources, Writing – review & editing, Supervision; M. Woegerbauer: J. Environ. Monit. 14 (7), 1754–1771. https://doi.org/10.1039/c2em00006g.
Gerrity, D., Gamage, S., Jones, D., Korshin, G.V., Lee, Y., Pisarenko, A., Trenholm, R.A.,
Resources, Supervision, Project administration; P. Hufnagl: Resources, von Gunten, U., Wert, E.C., Snyder, S.A., 2012. Development of surrogate correlation
Supervision; N. Kreuzinger: Conceptualization, Resources, Writing – models to predict trace organic contaminant oxidation and microbial inactivation
review & Editing, supervision, Project administration, Funding during ozonation. Water Res. 46 (19), 6257–6272. https://doi.org/10.1016/j.
watres.2012.08.037.
acquisition.
Hembach, N., Alexander, J., Hiller, C., Wieland, A., Schwartz, T., 2019. Dissemination
prevention of antibiotic resistant and facultative pathogenic bacteria by
ultrafiltration and ozone treatment at an urban wastewater treatment plant. Sci. Rep.
Declaration of Competing Interest 9 (1), 1–12. https://doi.org/10.1038/s41598-019-49263-1.
Hiller, C.X., Hübner, U., Fajnorova, S., Schwartz, T., Drewes, J.E., 2019. Antibiotic
microbial resistance (AMR) removal efficiencies by conventional and advanced
The authors declare that they have no known competing financial
wastewater treatment processes: a review. Sci. Total Environ. 685, 596–608. https://
interests or personal relationships that could have appeared to influence doi.org/10.1016/j.scitotenv.2019.05.315.
the work reported in this paper. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., Mcardell, C.S., Ort, C.,
Singer, H., Von Gunten, U., Siegrist, H., 2009. Elimination of organic
micropollutants in a municipal wastewater treatment plant upgraded with a full-
Data Availability scale post-ozonation followed by sand filtration. Environ. Sci. Technol. 43 (20),
7862–7869. https://doi.org/10.1021/es9014629.
Data will be made available on request. Iakovides, I.C., Michael-Kordatou, I., Moreira, N.F.F., Ribeiro, A.R., Fernandes, T.,
Pereira, M.F.R., Nunes, O.C., Manaia, C.M., Silva, A.M.T., Fatta-Kassinos, D., 2019.
Continuous ozonation of urban wastewater: Removal of antibiotics, antibiotic-
Acknowledgments and disclaimers resistant Escherichia coli and antibiotic resistance genes and phytotoxicity. Water
Res. 159, 333–347. https://doi.org/10.1016/j.watres.2019.05.025.
Kirchner, K., Brückner, I., Klaer, K., Hammers-Wirtz, M., Pinnekamp, J., Rosenbaum, M.
This project has received funding from the European Union’s Hori­ A., 2020. Microbial counts and antibiotic resistances during conventional
zon 2020 research and innovation programme under the Marie Skło­ wastewater treatment and wastewater ozonation. Ozone. Sci. Eng. 42 (2), 108–119.
https://doi.org/10.1080/01919512.2019.1645641.
dowska-Curie grant agreement No 675530. The content of this article
Kohanski, M.A., Dwyer, D.J., Hayete, B., Lawrence, C.A., 2007. A Common Mech. Cell.
reflects only the authors’ views and the Research Executive Agency is Death Induc. Bactericidal Antibiot. 797–810. https://doi.org/10.1016/j.
not responsible for any use that may be made of the information it cell.2007.06.049.
contains. The authors acknowledge TU Wien University Library for Krzeminski, P., Tomei, M.C., Karaolia, P., Langenhoff, A., Almeida, C.M.R., Felis, E.,
Gritten, F., Andersen, H.R., Fernandes, T., Manaia, C.M., Rizzo, L., Fatta-
financial support through its Open Access Funding Program. Kassinos, D., 2019. Performance of secondary wastewater treatment methods for the
removal of contaminants of emerging concern implicated in crop uptake and
antibiotic resistance spread: a review. Sci. Total Environ. 648, 1052–1081. https://
Appendix A. Supporting information
doi.org/10.1016/j.scitotenv.2018.08.130.
Kümmerer, K., 2009. Antibiotics in the aquatic environment - a review - Part I. In.
Supplementary data associated with this article can be found in the Chemosphere. https://doi.org/10.1016/j.chemosphere.2008.11.086.
LaMartina, E.Lou, Mohaimani, A.A., Newton, R.J., 2021. Urban wastewater bacterial
online version at doi:10.1016/j.jhazmat.2022.129396.
communities assemble into seasonal steady states. Microbiome 9 (1), 1–13. https://
doi.org/10.1186/S40168-021-01038-5.
References Lamba, M., Ahammad, S.Z., 2017. Performance comparison of secondary and tertiary
treatment systems for treating antibiotic resistance. Water Res. 127, 172–182.
https://doi.org/10.1016/j.watres.2017.10.025.
Acker, H. van, Coenye, T., 2017. The role of reactive oxygen species in antibiotic-
Lee, Y., Gerrity, D., Lee, M., Bogeat, A.E., Salhi, E., Gamage, S., Trenholm, R.A., Wert, E.
mediated killing of bacteria. Trends Microbiol. 25 (6), 456–466. https://doi.org/
C., Snyder, S.A., Von Gunten, U., 2013. Prediction of micropollutant elimination
10.1016/j.tim.2016.12.008.
during ozonation of municipal wastewater effluents: Use of kinetic and water
Alexander, J., Knopp, G., Dötsch, A., Wieland, A., Schwartz, T., 2016. Ozone treatment of
conditioned wastewater selects antibiotic resistance genes, opportunistic bacteria,

12
K. Slipko et al. Journal of Hazardous Materials 438 (2022) 129396

specific information. Environ. Sci. Technol. 47 (11), 5872–5881. https://doi.org/ Ravasi, D., König, R., Principi, P., Perale, G., Demarta, A., 2019. Effect of powdered
10.1021/es400781r. activated carbon as advanced step in wastewater treatments on antibiotic resistant
Lee, Y., Imminger, S., Czekalski, N., Gunten, U. von, Hammes, F., 2016. Inactivation microorganisms. Curr. Pharm. Biotechnol. 20 (1), 63–75. https://doi.org/10.2174/
efficiency of Escherichia coli and autochthonous bacteria during ozonation of 1389201020666190207095556.
municipal wastewater effluents quantified with flow cytometry and adenosine tri- Reasoner, D.J., Geldreich, E.E., 1985. A new medium for the enumeration and subculture
phosphate analyses. Water Res. 101, 617–627. https://doi.org/10.1016/j. of bacteria from potable water. Appl. Environ. Microbiol. https://doi.org/10.1128/
watres.2016.05.089. aem.49.1.1-7.1985.
Li, G., Ben, W., Ye, H., Zhang, D., Qiang, Z., 2018. Performance of ozonation and Rizzo, L., Manaia, C., Merlin, C., Schwartz, T., Dagot, C., Ploy, M.C., Michael, I., Fatta-
biological activated carbon in eliminating sulfonamides and sulfonamide-resistant Kassinos, D., 2013. Urban wastewater treatment plants as hotspots for antibiotic
bacteria: a pilot-scale study. Chem. Eng. J. 341, 327–334. https://doi.org/10.1016/j. resistant bacteria and genes spread into the environment: a review. Sci. Total
cej.2018.02.035. Environ. 447, 345–360. https://doi.org/10.1016/j.scitotenv.2013.01.032.
Luczkiewicz, A., Jankowska, K., Bray, R., Kulbat, E., Quant, B., Sokolowska, A., Rizzo, L., Malato, S., Antakyali, D., Beretsou, V.G., Đolić, M.B., Gernjak, W., Heath, E.,
Olanczuk-Neyman, K., 2011. Antimicrobial resistance of fecal indicators in Ivancev-Tumbas, I., Karaolia, P., Lado Ribeiro, A.R., Mascolo, G., McArdell, C.S.,
disinfected wastewater. Water Sci. Technol. 64 (12), 2352–2361. https://doi.org/ Schaar, H., Silva, A.M.T., Fatta-Kassinos, D., 2019. Consolidated vs new advanced
10.2166/wst.2011.769. treatment methods for the removal of contaminants of emerging concern from urban
Lüddeke, F., Heß, S., Gallert, C., Winter, J., Güde, H., Löffler, H., 2015. Removal of total wastewater. Sci. Total Environ. 655, 986–1008. https://doi.org/10.1016/j.
and antibiotic resistant bacteria in advanced wastewater treatment by ozonation in scitotenv.2018.11.265.
combination with different filtering techniques. Water Res. 69, 243–251. https:// Rizzo, L., Gernjak, W., Krzeminski, P., Malato, S., McArdell, C.S., Perez, J.A.S.,
doi.org/10.1016/j.watres.2014.11.018. Schaar, H., Fatta-Kassinos, D., 2020. Best available technologies and treatment trains
Michael-Kordatou, I., Andreou, R., Iacovou, M., Frontistis, Z., Hapeshi, E., Michael, C., to address current challenges in urban wastewater reuse for irrigation of crops in EU
Fatta-Kassinos, D., 2017. On the capacity of ozonation to remove antimicrobial countries. Sci. Total Environ. 710, 136312 https://doi.org/10.1016/j.
compounds, resistant bacteria and toxicity from urban wastewater effluents. scitotenv.2019.136312.
J. Hazard. Mater. 323, 414–425. https://doi.org/10.1016/j.jhazmat.2016.02.023. Sousa, J.M., Macedo, G., Pedrosa, M., Becerra-Castro, C., Castro-Silva, S., Pereira, M.F.R.,
Michael-Kordatou, I., Karaolia, P., Fatta-Kassinos, D., 2018. The role of operating Silva, A.M.T., Nunes, O.C., Manaia, C.M., 2017. Ozonation and UV254nm radiation
parameters and oxidative damage mechanisms of advanced chemical oxidation for the removal of microorganisms and antibiotic resistance genes from urban
processes in the combat against antibiotic-resistant bacteria and resistance genes wastewater. J. Hazard. Mater. 323, 434–441. https://doi.org/10.1016/j.
present in urban wastewater. Water Res. 129, 208–230. https://doi.org/10.1016/j. jhazmat.2016.03.096.
watres.2017.10.007. Stapf, M., Miehe, U., Jekel, M., 2016. Application of online UV absorption measurements
Mišík, M., Ferk, F., Schaar, H., Yamada, M., Jaeger, W., Knasmueller, S., Kreuzinger, N., for ozone process control in secondary effluent with variable nitrite concentration.
2020. Genotoxic activities of wastewater after ozonation and activated carbon Water Res. 104, 111–118. https://doi.org/10.1016/J.WATRES.2016.08.010.
filtration: Different effects in liver-derived cells and bacterial indicators. Water Res. Szabó, L., Tóth, T., Engelhardt, T., Rácz, G., Mohácsi-farkas, C., 2016. Science of the total
186. https://doi.org/10.1016/j.watres.2020.116328. environment change in hydrophilicity of penicillins during advanced oxidation by
Moreira, N.F.F., Sousa, J.M., Macedo, G., Ribeiro, A.R., Barreiros, L., Pedrosa, M., radiolytically generated • OH compromises the elimination of selective pressure on
Faria, J.L., Pereira, M.F.R., Castro-Silva, S., Segundo, M.A., Manaia, C.M., Nunes, O. bacterial strains. Sci. Total Environ., 551–552, 393–403. https://doi.org/10.1016/j.
C., Silva, A.M.T., 2016. Photocatalytic ozonation of urban wastewater and surface scitotenv.2016.02.002.
water using immobilized TiO2 with LEDs: Micropollutants, antibiotic resistance Vaz-Moreira, I., Nunes, O.C., Manaia, C.M., 2014. Bacterial diversity and antibiotic
genes and estrogenic activity. Water Res. 94, 10–22. https://doi.org/10.1016/j. resistance in water habitats: Searching the links with the human microbiome. FEMS
watres.2016.02.003. Microbiol. Rev. 38 (4), 761–778. https://doi.org/10.1111/1574-6976.12062.
Nöthe, T., Fahlenkamp, H., Sonntag, C., 2009. Ozonation of Wastewater: Rate of Ozone Von Gunten, U., 2003. Ozonation of drinking water: Part II. Disinfection and by-product
Consumption and Hydroxyl Radical Yield. Environmental science & technology 43 formation in presence of bromide, iodide or chlorine. Water Res. 37 (7), 1469–1487.
(5990–5). https://doi.org/10.1021/es900825f. https://doi.org/10.1016/S0043-1354(02)00458-X.
Novo, A., Manaia, C.M., 2010. Factors influencing antibiotic resistance burden in Von Sonntag, C., Von Gunten, U., 2012. Chemistry of Ozone in Water and Wastewater
municipal wastewater treatment plants. Appl. Microbiol. Biotechnol. 87 (3), Treatment From Basic Principles to Applications. IWA Publishing.
1157–1166. https://doi.org/10.1007/s00253-010-2583-6. Von Sperling, M., Verbyla, M.E., Mihelcic, J.R., 2018. Part Four. Management of Risk
Oh, J., Salcedo, D.E., Medriano, C.A., Kim, S., 2014. Comparison of different disinfection From Excreta and Wastewater Understanding Pathogen Reduction in Sanitation
processes in the effective removal of antibiotic-resistant bacteria and genes. Systems: Units of Measurement, Expressing Changes in Concentrations, and Kinetics.
J. Environ. Sci. (China) 26 (6), 1238–1242. https://doi.org/10.1016/S1001-0742 In J. B. Rose & B. Jiménez-Cisneros (Eds.), Water and Sanitation for the 21st Century:
(13)60594-X. Health and Microbiological Aspects of Excreta and Wastewater Management (Global
Pärnänen, K.M.M., Narciso-Da-Rocha, C., Kneis, D., Berendonk, T.U., Cacace, D., Do, T. Water Pathogen Project). Michigan State University, E. Lansing, MI, UNESCO. https://
T., Elpers, C., Fatta-Kassinos, D., Henriques, I., Jaeger, T., Karkman, A., Martinez, J. doi.org/https://doi.org/10.14321/waterpathogens.54.
L., Michael, S.G., Michael-Kordatou, I., O’Sullivan, K., Rodriguez-Mozaz, S., Wu, J., Cheng, S., Cai, M.H., Wu, Y.P., Li, Y., Wu, J.C., Li, A.M., Li, W.T., 2018. Applying
Schwartz, T., Sheng, H., Sørum, H., Manaia, C.M., 2019. Antibiotic resistance in UV absorbance and fluorescence indices to estimate inactivation of bacteria and
European wastewater treatment plants mirrors the pattern of clinical antibiotic formation of bromate during ozonation of water and wastewater effluent. Water Res.
resistance prevalence. Sci. Adv. 5 (3) https://doi.org/10.1126/sciadv.aau9124. 145, 354–364. https://doi.org/10.1016/j.watres.2018.08.030.
Pazda, M., Kumirska, J., Stepnowski, P., Mulkiewicz, E., 2019. Antibiotic resistance Yin, X., Deng, Y., Ma, L., Wang, Y., Chan, L.Y.L., Zhang, T., 2019. Exploration of the
genes identified in wastewater treatment plant systems – a review. In: Science of the antibiotic resistome in a wastewater treatment plant by a nine-year longitudinal
Total Environment, Vol. 697. Elsevier B.V. https://doi.org/10.1016/j. metagenomic study. Environ. Int. 133, 105270 https://doi.org/10.1016/J.
scitotenv.2019.134023. ENVINT.2019.105270.
Poole, K., 2017. At the nexus of antibiotics and metals: the impact of Cu and Zn on Zheng, J., Su, C., Zhou, J., Xu, L., Qian, Y., Chen, H., 2017. Effects and mechanisms of
antibiotic activity and resistance. Trends Microbiol. 25 (10), 820–832. https://doi. ultraviolet, chlorination, and ozone disinfection on antibiotic resistance genes in
org/10.1016/j.tim.2017.04.010. secondary effluents of municipal wastewater treatment plants. Chem. Eng. J. 317,
Pruden, A., Larsson, D.G.J., Amézquita, A., Collignon, P., Brandt, K.K., Graham, D.W., 309–316. https://doi.org/10.1016/j.cej.2017.02.076.
Lazorchak, J.M., Suzuki, S., Silley, P., Snape, J.R., Topp, E., Zhang, T., Zhu, Y.-G., Zhuang, Y., Ren, H., Geng, J., Zhang, Y., Zhang, Y., Ding, L., Xu, K., 2015. Inactivation of
2013. Management of options for reducing the release of antibiotics. Environ. Health antibiotic resistance genes in municipal wastewater by chlorination, ultraviolet, and
Perspect. 121 (8), 878–885. ozonation disinfection. Environ. Sci. Pollut. Res. 22 (9), 7037–7044. https://doi.org/
R Core Team, 2021. R: A language and environment for statistical computing. (4.0.4.). R 10.1007/s11356-014-3919-z.
Foundation for Statistical Computing.

13