Added Value of MBR and Ozonation for Advanced Wastewater Treatment Based on Antibiotic Resistance Genes and Bacteroidales as a Marker for Fecal Gene Load

Abstract

Large wastewater treatment plants (WWTP) are increasingly supplemented with quaternary treatment. Classical monitoring hereby relies mostly on the measurement of oxygen demand, micropollutants and the nutrients phosphorus and nitrogen. From a microbiological perspective, relevant parameters to assess treatment performance include the removal efficacies of the fecal gene load as a proxy of pathogenic risk, antibiotic resistance genes and the bacterial regrowth potential. For this purpose, a combination of flow cytometry and quantitative PCR, together with a viability assessment, was applied to characterize a full-scale pilot plant. The pilot plant comprised conventional treatment and MBR and ozonation for advanced treatment. The assessment of fecal gene load was based on the quantification of Bacteroidales of human origin, as these obligate anaerobic bacteria cannot replicate within wastewater treatment plants. Whereas conventional treatment resulted in only moderate removal of these parameters, quaternary treatment typically led to a much stronger decrease. MBR treatment contributed most strongly to the removal with an appr. 6 log reduction compared to the primary clarification effluent, corroborating its microbiological merit for wastewater treatment. In addition to removing microorganisms and their genetic content, data also suggested a 95% reduction in extracellular DNA. Ozonation further enhanced microbiological removal. From an analytical perspective, the study shows the added value of using a long amplicon qPCR approach together with sample treatment with a viability dye to minimize false-positive signals and to avoid underestimation of treatment performance. The chosen diagnostic approach shows promise in assessing the microbiological treatment efficacy of WWTPs and as a basis to decide on the microbiological necessity of treatment upgrades.

1. Introduction

WWTPs are recognized as critical barriers that limit the spread of contaminants [1]. Traditionally, municipal wastewater treatment plants (WWTPs) have, however, only been optimized for the removal of chemical and biological oxygen demand (COD and BOD) and the nutrients phosphorus and nitrogen. Recent years have seen an increasing focus on the release of organic pollutants, toxicity, antibiotic resistance, and microbiological parameters [2,3]. Within the European Union, advanced wastewater treatment and improved surveillance have been strengthened by the EU Urban Wastewater Treatment Directive (UWWTD 2024/3019). Foremost organic pollutants have gained much attention in light of the quaternary treatment stage that is becoming mandatory for WWTPs with ≥150,000 p.e. in EU member states. For WWTPs with ≥10,000 p.e., the necessity depends on a mandatory risk assessment.

From a microbiological “One Health” point of view, the removal of hygienically relevant parameters and antibiotic resistances is of most relevance, both when upgrading existing treatment plants or designing new ones. Minimizing the microbiological effluent load of these parameters, however, is not an easy task, given the lack of knowledge about the treatment efficacy of individual treatment stages [4,5]. Different treatment steps can hereby not be assigned a specific efficacy per se, as the effectiveness of treatment not only depends on the applied treatment technology, but also largely on operational conditions and the characteristics of the wastewater [6]. As a consequence, the microbiological treatment efficacy is typically unknown. This applies both to conventional treatment (including mechanical and biological stages along with precipitation) and advanced treatment technologies comprising, e.g., ultrafiltration, ozonation, UV, and, for high-end applications requiring high water qualities, also reverse osmosis.

The goal of the study was to assess the added microbiological value of a combination of a membrane bioreactor (MBR) and ozonation as advanced treatment processes following conventional treatment. For this purpose, a suite of non-conventional parameters was applied as microbiological performance indicators. The analysis included the assessment of cell numbers and bacterial regrowth potential using flow cytometry and the qPCR-based quantification of antibiotic resistance genes (ARGs) and human Bacteroidales bacteria. ARGs comprised the commonly found marker genes sul1, tetM, ermB, blaTEM and the class 1 integron-integrase gene intl1, which correlates with many resistance genes [7].

Bacteroidales were used as an indicator of fecal pollution and as a process parameter for assessing treatment efficiency. Bacteria of the order Bacteroidales are typically used to determine the source of fecal material as part of microbial source tracking (MST; [8]). Particularly, the bacteria belonging to the genus Bacteroides co-evolve with the host [9], resulting in the presence of characteristic Bacteroidales populations in different hosts such as humans, ruminants, pigs, horses, and birds. In this study, human-associated Bacteroidales were quantified. In the human colon, Bacteroidales are present in higher concentrations than Escherichia coli in fecal material with up to 10¹¹ Bacteroidales per gram of stool [10] compared to 10⁷–10⁸ E. coli per g of stool [11]. The dominant bacterial genus within the order Bacteroidales in the human gut is Bacteroides. This genus accounts for a major fraction of the gut bacterial community, making up 20–52% of the human fecal flora [12,13]. As a consequence, human-associated Bacteroidales are present in high numbers in municipal wastewater [14] with a global distribution [15]. Since these bacteria are obligate anaerobes, they cannot multiply outside the digestive tract [16,17], unlike E. coli or Enterococci that can also be of environmental origin [18,19,20].

Assessment of treatment performance was complemented by the quantification of antibiotic resistance genes (ARGs) and overall removal of bacterial gene load based on universal 16S rRNA genes. As for the quantification of Bacteroidales, molecular assessment of overall bacteria considered the effects of PCR amplicon length. Amplification of longer sequences makes results more specific for microorganisms with intact DNA, or in other words, selects against signals from microorganisms with degraded DNA [21]. The long amplicon qPCR (LA-qPCR) approach also bears the advantage of improving the efficiency of sample treatment with viability dyes compared to short amplicon qPCR (SA-qPCR). Samples in this study were in part treated with propidium monoazide (PMA), which allows to selectively detect microorganisms with intact cell membranes. The efficiency of exclusion of membrane-compromised cells increases with amplicon size [22].

The overall objective of this study was to use this non-conventional diagnostic approach to assess the added value of quaternary wastewater treatment employing MBR and ozonation in comparison with conventional treatment. Another objective was to build up experience regarding the performance of the different diagnostic methods. Apart from assessing the retention of microorganisms and their intracellular genes, a side aspect of this study addressed the question of whether MBR treatment is also effective in retaining free extracellular DNA (exDNA). Free DNA typically emerges upon cell lysis, which leads to the release of nucleic acids that were previously intracellular. Given the pore size of the UF membrane of approximately 20–50 nm, the membranes are effective in removing microorganisms, whereas this is not expected for molecules like DNA from the mere standpoint of pore size.

2. Materials and Methods

2.1. Wastewater Treatment

Wastewater treatment was performed in a pilot plant located in Dinslaken on the site of a WWTP operated by Emschergenossenschaft und Lippeverband (EGLV). An overview of the treatment train is shown in Figure 1. The WWTP primarily received domestic wastewater from a nearby urban catchment area. For this purpose, the wastewater is pumped via two pumping stations to a collection shaft, from which the pilot plant is supplied with wastewater. The inflow to the plant is regulated to a fixed inflow volume. The 1000 PE pilot wastewater treatment plant consists of a screen, a combined grit chamber with primary treatment, and a conventional activated sludge process consisting of two biological lines (500 PE each) with upstream denitrification. Thanks to the separate secondary clarifiers for each line and their own sludge circuits, the lines can be operated independently of each other. The conventionally treated effluent was further subjected to MBR and downstream ozonation (Figure 1). The MBR plant was equipped with a submerged flat-sheet membrane module of the type BIO-CEL^® L-2 (480 m²) with a polyethersulfone (PES) membrane (Pall Corporation, Port Washington, NY, USA). The pore size of the membrane was 45 nm. The downstream ozone plant was equipped with a Wedeco ozone generator of the Modular HC series (Xylem Wedeco, Herford, Germany). The ozone was introduced by a pump-injector system (PIS). A specific ozone concentration of 0.2–0.8 mg O₃/mg DOC was applied, representing the typical range of conventional ozone plants to further treat municipal wastewater [23,24,25].

Figure 1. Schematic overview of the wastewater treatment of the studied pilot plant. Conventional treatment consisted of primary and secondary clarification, whereas MBR and ozonation served for advanced treatment.

2.2. Sampling and Overview of Sample Processing

In total, samples from 12 sampling events were analyzed; the investigation period was from 7 November 2023 to 9 October 2024. This time allowed sufficient time to overcome intermittent operational problems and to make adjustments to ozone concentrations. To minimize diurnal variations, 24 h composite samples were collected using automatic sampling devices (Hach Lange, Düsseldorf, Germany) equipped with glass containers (2.9 L). Samples were cooled to 4 °C ± 2 °C during storage. The samples were subsequently transferred from the collection containers into sterile glass bottles and transported cold to the laboratory for same day-analysis. Samples for flow cytometric analysis were contained in glass vessels that were free of assimilable organic carbon (AOC). All samples were stored cold (4 °C ± 2 °C) prior to analysis. Samples for qPCR analysis were filtered and either treated with PMA or not prior to DNA extraction to assess the effect of PMA treatment (for an overview of sample processing, see Figure 2).

Figure 2. Sample volumes and overview of sample processing.

2.3. Flow Cytometry

Flow cytometry was used to quantify total cell counts (TCC, comprising cells with intact and damaged cell membranes), intact cell counts (ICC, cells with intact cell membranes) and regrowth potentials. For flow cytometric measurements, sample volumes of 250 µL were placed in the wells of an uncoated 96-well microtiter plate (cat. no. 601808; HJ-Bioanalytik GmbH, Erkelenz, Germany). Using a multichannel pipette, defined sample volumes were transferred from these plates to additional microtiter plates containing pre-aliquoted dyes. For flow cytometric quantification of TCC, 200 µL of the sample was added to 2 µL of SYBR™ Green I (100× in DMSO) and mixed by repeated pipetting up and down. To determine the ICC values, the cells were stained with a dye mixture of SYBR™ Green I (cat. no. S7567, Fisher Scientific Inc., Waltham, MA, USA) and propidium iodide (cat. no. P3566, Fisher Scientific Inc., Waltham, MA, USA). For this purpose, propidium iodide (1.5 mM) was mixed with SYBR™ Green I (100× in DMSO) in a ratio of 1:6, and 200 µL of the sample was added to 2.4 µL of the dye mixture. After mixing, the samples were incubated for 13 min at 37 °C in a heating cabinet.

The samples were then analyzed in a NovoCyte ACEA (ACEA Biosciences Inc., San Diego, CA, USA) equipped with a 488 nm laser. The measurements were performed using the following instrument parameters: fast settings (66 µL/min), analysis volume: 50 µL, FL-1 threshold: 600 (trigger on FL-1). Green fluorescence was measured on the FL-1 channel (530 nm), red fluorescence on the FL-3 channel (675 nm). The gate used for the analysis was based on that of Gatza et al. [26]. The resulting cell numbers are stated as day 0 cell counts reflecting the microbiological status quo at the time point of sampling.

2.4. Assessment of Regrowth Potential

After flow cytometric quantification of day 0 cell numbers, water samples contained in AOC-free vials were stored at 22 °C (±1 °C) for seven days to assess the regrowth potential. After seven days, the cell numbers were once more quantified by flow cytometry. Only ICC data is shown after regrowth. Intact cell numbers obtained after seven days of incubation (ICC_{day 7}) represent the cell concentrations that are supported by the nutrients contained in the original samples.

2.5. PMA Treatment and DNA Extraction

For molecular biological analyses, the samples were filtered on the day of sampling using 0.2 µm polycarbonate filters (prod. no. PC02CP04700, Pieper Filter GmbH, Bad Zwischenahn, Germany) and a vacuum filtration bar. The filterable sample volume was highly dependent on the type of sample. Typical sample volumes were 25 mL (effluent primary clarification, 50–100 mL (effluent secondary clarification and 500–1000 mL for the effluents of MBR and ozonation. Two identical filtration preparations were made from each sample. One filter was treated with the integrity dye propidium monoazide (PMA) to limit signals to membrane-intact bacteria; the other was not. The PMA treatment was carried out directly on the vacuum filtration bar.

For filter treatment with PMA (prod. no. 40013; Biotium, Fremont, CA, USA), 1 mg aliquots of the dye were dissolved in 786 µL of 25% dimethyl sulfoxide (DMSO) to produce a stock solution of 2.5 mM. A total of 40 µL of this stock solution was added to 2 mL of sterile water (preheated to 37 °C in a water bath). This solution with a final PMA concentration of 50 µM was used to coat the filters. To protect the light-sensitive dye, the filtration bars were covered with aluminum foil during treatment. The exposure time was 10 min, during which the preparations were shaken occasionally to ensure even coating with the dye solution. After incubation, the remaining dye solution was suck off.

Following PMA treatment, the filters were transferred to a sterile Petri dish with the filter cake facing upwards then irradiated without the plastic lid with blue LED light (465 nm) for 15 min in a photoactivation device (PhotoActivation Universal Light, PAUL, GenIUL, Terrassa, Spain) for 15 min with blue LED light (465 nm). The filters were then covered with a lid and stored in the Petri dishes at −20 °C until DNA extraction.

DNA extraction from the filtered biomass was performed using the FastDNA™ SPIN Kit for Soil (prod. no. 116560200-CF; MP Biomedicals, Solon, OH, USA). The cell material was broken down using a bead beater (BeadBug™ 6 Homogenizer, Benchmark Scientific, Sayreville, NJ, USA) and processed as described in the section ‘case study wastewater’ of the manufacturer’s instructions. The elution volume was 200 µL.

2.6. Isolation of Extracellular DNA

Extracellular DNA (exDNA) was obtained from samples taken from the MBR tank and the MBR filtrate. The sample from the MBR tank was taken using a ladle, while the MBR filtrate sample was taken from a ball valve in the drain pipe. The ball valve was flamed before sampling, and sampling took place after one liter had drained. Three independent samples were taken at intervals of 15 min. The sample volume was approximately 250 mL. After cold transport to the laboratory, the samples were left to stand for approximately 45 min to allow particles in the pre-MBR sample to settle.

The supernatants of the sedimented samples from the MBR tank were transferred to 25 mL Falcon tubes using a pipette. A total of 25 mL of each sample was then filtered (0.1 µm Acrodisc filter, prod. no. 4611, Pall Corporation, Port Washington, NY, USA) to remove cells. The filtrates were collected in sterile 50 mL Falcon tubes. These cell-free filtrates (25 mL) were then mixed with 2.5 mL of sodium acetate solution (3 M, pH 5.2) and 17.5 mL of isopropanol to precipitate nucleic acids. The mixtures were incubated for 30 min at room temperature, followed by centrifugation at 12,000× g for 30 min at 4 °C. The supernatants were carefully decanted and the pellets were washed with 5 mL of cold 70% ethanol. This washing step was repeated once. After decanting, any remaining alcohol was carefully removed with a pipette and the pellets were air-dried with the lid open. The dry pellets were each resuspended in 100 µL TE buffer and examined using quantitative PCR.

2.7. Quantitative PCR

The extracted DNA was examined for various parameters (Table 1) using quantitative PCR (qPCR). PCR detection was performed using a CFX96 Real-time System C1000 Touch (BioRad Laboratories, Hercules, CA, USA). PCR reactions had a total volume of 20 µL comprising 2 µL template, 0.5 µL recombinant albumin (20 mg/mL, cat.nr. B9200S, New England Biolabs GmbH, Frankfurt am Main, Germany) and 10 µL master mix. Primer concentrations varied depending on the specific assay and were 200 nM (BacUni), 250 nM (human Bacteroidales) or 400 nM (ARGs). For the Taqman-based Bacteroidales qPCR detection, the final probe concentration was 250 nM using the KAPA PROBE Fast master mix (prod. no. KK4703; Kapa Biosystems, Wilmington, MA, USA) PowerTrack SYBR Green Master Mix (prod. no. A46109; Thermo Fisher Scientific, Waltham, MA, USA) was used for SYBR Green-based qPCR reactions. Albumin was added to the preparations to minimize inhibition. For the same reasons, the extracted DNA solutions were diluted prior to use as a template in qPCR. A dilution series of 1:4, 1:16 and 1:64 was quantified for each sample using a universal 16S rRNA gene-based qPCR assay and compared with the undiluted sample. Inhibition was indicated when concentrations of genomic units (GU) of samples increased with the dilution factor. Optimal dilutions were 1:64 or 1:16 for primary and secondary clarification samples, respectively, and 1:4 for MBR filtrates and ozonated water.

Table 1. Oligonucleotides are used for amplification of different target genes together with cycling programs. Nucleotide symbols refer to the IUPAC nomenclature codes. Amplicon sizes are approximate and can vary for different bacterial species and strains.

Oligonucleotide	Sequence (5′-3′)	Amplicon Size (Approximate Value)	Assay and Cycling Conditions	Reference
BacUni short (gene target: 16S rRNA genes)
BacUni-1387F BacUni-1492R	GCCTTGTACACWCCGCCC GGYTACCTTGTTACGACTT	105 bp	SYBR Green assay 95 °C: 15 s 55 °C: 20 s 72 °C: 15 s	[21]
BacUni long (gene target: 16S rRNA genes)
BacUni-9F BacUni-1492R	GAGTTTGATCCTGGCTCAG GGYTACCTTGTTACGACTT	1483 bp	SYBR Green assay 95 °C: 15 s 55 °C: 20 s 72 °C: 80 s	[21]
Human Bacteroidales short (gene target: 16S rRNA genes)
HF183 BFDrev BFD-FAM	ATCATGAGTTCACATGTCCG CGTAGGAGTTTGGACCGTGT CTGAGAGGAAGGTCCCCCACATTGGA-FAM	167 bp	TaqMan assay 95 °C: 15 s 58 °C: 50 s	[27,28,29]
Human Bacteroidales long (gene target: 16S rRNA genes)
HF183 Bac708R BFD-FAM	ATCATGAGTTCACATGTCCG CAATCGGAGTTCTTCGTG CTGAGAGGAAGGTCCCCCACATTGGA-FAM	520 bp	TaqMan assay 95 °C: 15 s 58 °C: 50 s	[27,28,29]
Antibiotic resistance gene tetM
S_tetM-F S_tetM-R	GGTTTCTCTTGGATACTTAAATCAATCR CCAACCATAYAATCCTTGTTCRC	88 bp	SYBR Green assay 95 °C: 15 s 60 °C: 1 min	[30]
Antibiotic resistance gene ermB
ermB-F ermB-R	TGAATCGAGACTTGAGTGTGCAA GGATTCTACAAGCGTACCTT	71 bp	SYBR Green assay 95 °C: 15 s 60 °C: 1 min	[31]
Antibiotic resistance gene blaTEM
qblaTEM-F qblaTEM-R	TTCCTGTTTTTGCTCACCCAG CTCAAGGATCTTACCGCTGTTG	112 bp	SYBR Green assay 95 °C: 15 s 60 °C: 1 min	[30]
Antibiotic resistance gene sul1
sul1-F sul1-R	CGCACCGGAAACATCGCTGCAC TGAAGTTCCGCCGCAAGGCTCG	161 bp	SYBR Green assay 95 °C: 15 s 60 °C: 1 min	[32]
class 1 integron-integrase gene intl1
intl-F intl-R	GCCTTGATGTTACCCGAGAG GATCGGTCGAATGCGTGT	196 bp	SYBR Green assay 95 °C: 15 s 60 °C: 1 min	[32]

Individual assays with gene targets, amplicon sizes and cycling conditions are listed in Table 1. General bacterial elimination performance was determined on the basis of 16S rRNA genes using the BacUni assay [21]. The qPCR was performed in two variants: the amplification of short (approx. 105 bp) and long (approx. 1483 bp) sequences. The advantage of detecting longer sequences is the preferential detection of bacteria with intact DNA. The amplification of longer sequences furthermore increases the efficiency of the PMA treatment, as the probability that at least one PMA molecule has bound to the target sequence region increases with increasing sequence length [22].

Two different amplicon sizes (approx. 167 bp and 520 bp) were also generated when quantifying human-associated Bacteroidales, whereas only one amplicon size was assessed for antibiotic resistance genes.

2.8. Limit of Detection

The limit of detection (LoD) of the different qPCR assays was determined from the regression analysis of the corresponding standard curves. The calculation was on the basis of the standard deviation (SD) and the slope of the regression lines using the equation:

$$\mathrm{L}\mathrm{o}\mathrm{D}={\displaystyle \frac{3.3 \times {\mathrm{S}\mathrm{D}}_{\mathrm{r}\mathrm{e}\mathrm{g}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}}}{{\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}}_{\mathrm{r}\mathrm{e}\mathrm{g}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}}}}$$

The resulting theoretically detectable copy number per qPCR well with 2 µL template was used to calculate the LoD per liter of sample, considering a correction factor of 1.47 for the loss during DNA extraction, the template dilution factor of 4 (applying to MBR filtrates and ozonated water samples) and the sample volume. LoDs were calculated as follows: 1.32 × 10³ GU/L (BacUni short), 1.22 × 10³ GU/L (BacUni long), 1.46 × 10³ GU/L (Bacteroidales short), 2.07 × 10³ GU/L (Bacteroidales long), 1.20 × 10³ GU/L (tetM), 1.41 × 10³ GU/L (ermB), 1.45 × 10³ GU/L (blaTEM), 1.17 × 10³ GU/L (sul1) and 1.50 × 10³ GU/L (intl1).

3. Results

The studied pilot plant treated mainly domestic wastewater. Conventional treatment included primary clarification, an activated sludge process and secondary clarification (Figure 1). Effluents of the aeration tank and the secondary clarifier were mixed in a ratio of 75:25 and further subjected to advanced treatment consisting of MBR and ozonation.

The microbiological characterization of treatment efficacy included cell counts and bacterial growth potentials determined by flow cytometry, as well as concentrations of bacteria and resistance genes determined by molecular biological methods. The difference between the effluents of primary and secondary clarification provided information about the microbiological efficiency of conventional wastewater treatment. On the other hand, the additional reduction after MBR and ozonation provided information about the efficiency of these advanced treatment processes and the added value of the quaternary treatment stage.

3.1. Intact Cell Concentrations and Regrowth Potentials Assessed by Flow Cytometry

Conventional wastewater treatment led to a moderate reduction in concentrations of bacteria with intact cell membranes (Figure 3A). The effluent of primary clarification contained 4.4 × 10⁷ intact cells/mL. Compared to this value, median ICC values were only 0.4 log units lower after secondary clarification. Subsequent MBR treatment, on the other hand, resulted in a strong reduction in ICC to 2.0 × 10⁵ intact cells/mL corresponding to a decrease of 2.3 log units relative to the effluent of the primary clarification. The fact that signals were detectable in the filtrate can be explained by the enrichment of the filtrate by previously surface-attached bacteria and regrowth due to the presence of nutrients. Their presence, therefore, does not allow conclusions on the integrity of the UF membrane. Flow cytometric fingerprints of water in the MBR tank and in the 24 h MBR filtrate were profoundly distinct, corroborating that the bacteria in the filtrate did not originate from the MBR tank (Figure 3B). The MBR tank sample was dominated by a single, narrow low-nucleic-acid (LNA) population (LNA fraction: 88.7%), whereas the MBR effluent exhibited a markedly more heterogeneous distribution with multiple high-fluorescence subpopulations (LNA fraction: 19.4%).

Figure 3. Change in intact cell counts (ICCs) along the treatment train and of flow cytometric fingerprints prior to and after ultrafiltration. (A) ICC values (bacteria/mL) on day 0 (reflecting the microbiological status at the time point of sampling) and after incubation of seven days at 22 °C for the respective wastewater treatment stages. Data represent averages from 12 sampling rounds between 7 November 2023 and 9 October 2024. During ozonation, the specific concentration was between 0.20 and 0.62 mg ozone/mg DOC. (B) Representative flow cytometric fingerprints of water in the MBR tank and in the 24 h-MBR filtrate after staining with SG and PI.

Ozonation led to a further decrease in ICC to 9.3 × 10⁴ intact cells/mL, corresponding to an overall decrease of 2.7 log units along the entire treatment train. Repeated-measures ANOVA and pairwise p-values confirmed that the differences in ICC between the different treatment stages were statistically significant.

The effluents of primary clarification and secondary clarification processes each exhibited negative regrowth potential, i.e., the cell counts decreased during the seven-day incubation at 22 °C. In the effluent of primary clarification, the day 7 values were about three times lower than the day 0 values, and in the effluent of secondary clarification, they were four times lower. This is often observed in wastewater. Possible causes include the die-off of bacteria, an increase in the relative proportion of bacteria with a greater cell size and overall changes in the composition of the microbial communities. Overall, the conventional treatment reduced the regrowth potential by 0.6 log units.

Water after both the MBR and ozonation processes showed positive regrowth, i.e., cell counts increased during the seven-day incubation period. In comparison with the effluent of secondary clarification, MBR treatment led to a significant reduction in ICC_{day 7} values, while the ozone treatment caused a slight increase. The latter can be explained by the oxidative conversion of DOC to AOC. The quaternary treatment reduced the regrowth potential by another 0.9 log units, adding up to a decrease of approx. 1.5 log units along the entire treatment train. As for day 0 values, it could be shown that the differences in ICC_day7 values between the different treatment stages were statistically significant, with the exception between MBR and ozonation.

3.2. Bacterial Concentrations Assessed by Quantitative PCR

Bacterial concentrations were also assessed by quantitative PCR. Quantification of 16S rRNA genes was based on two different amplicon sizes of appr. 100 bp (Figure 4A) and 1500 bp (Figure 4B) from samples that were either treated with PMA or not. As seen for intact cell numbers, the concentration of 16S rRNA gene copies continuously decreased along the treatment train. A particularly strong reduction was caused by the MBR treatment. Sample treatment with PMA led to a further reduction in bacterial signals in all cases, suggesting the presence of membrane-damaged cells. Using the LA-qPCR approach, the median 16S rRNA gene copy numbers were lower compared with the SA-qPCR approach. In the absence of PMA treatment, the reason for the discrepancy can be seen in the higher selectivity of LA-qPCR for bacteria with intact DNA. When applying PMA treatment, in addition, signals from bacteria with damaged cell membranes are more efficiently excluded by LA-qPCR compared with SA-qPCR. The PMA effect was particularly pronounced after ozonation, corroborating the effect of this oxidative treatment on cell membrane integrity.

Figure 4. Bacterial concentrations (bacterial GU/L) quantified by 16S rRNA gene qPCR-amplification of (A) short sequences (appr. 100 bp) or (B) long sequences (approx. 1500 bp) for the respective wastewater treatment stages over 11 months (n = 12). Samples were either left untreated or treated with PMA. For ozonation, the specific ozone doses ranged between 0.20 and 0.62 mg ozone/mg DOC.

Log reductions in bacterial 16S rRNA gene concentrations were very consistent between SA-qPCR and LA-qPCR, with differences mainly caused by PMA treatment. In the absence of PMA treatment, bacterial gene concentrations were reduced by conventional treatment by approx. 1.2 log units and by 4 log units over the entire treatment train. With PMA treatment, conventional treatment resulted in a reduction of approx. 1.2–1.4 log units and by 4.7 log units over the entire treatment train.

3.3. Concentrations of Bacteroidales Assessed by Quantitative PCR

Changes in human Bacteroidales concentrations along the treatment train served as a proxy for the removal of the human fecal gene load. As for the overall bacterial load, qPCR-based quantification was based on two different amplicon sizes (Figure 5A,B). Conventional wastewater treatment resulted in an approx. 1.5–2 log reduction in the fecal indicator, with the values after PMA treatment being lower than those without PMA treatment (Figure 5). Advanced treatment resulted in a substantially better removal of Bacteroidales with an additional approx. 4 log decrease compared to the effluent of secondary clarification. MBR treatment had the greatest effect, with values after ozonation only being slightly below those after MBR effluent. The fact that ozonation did not result in a more pronounced reduction was probably due to the qPCR detection limit of approx. 10³ GU/L for this parameter (LoD of SA-qPCR: 1.46 × 10³ GU/L, LoD of LA-qPCR: 2.07 × 10³ GU/L).

Figure 5. Concentrations of Bacteroidales (GU/L) quantified by qPCR-amplification of (A) short sequences (appr. 167 bp) or (B) long sequences (approx. 520 bp) for the respective wastewater treatment stages over 11 months (n = 12). Samples were either left untreated or treated with PMA. For ozonation, the specific ozone doses ranged between 0.20 and 0.62 mg ozone/mg DOC.

Whereas qPCR values based on LA-qPCR were not significantly different from those of SA-qPCR for the conventional treatment stages, the amplification of longer sequences increased the probability of complete signal suppression after advanced treatment. In combination with PMA treatment, no Bacteroidales signals were detectable after ozonation, indicating efficient reduction in this human fecal marker below the LoD of the assay.

3.4. Concentrations of ARGs

The study involved qPCR quantification of the four frequently found antibiotic resistance genes sul1, tetM, ermB, blaTEM and the class 1 integron-integrase gene intl1, which correlates with many resistance genes. All genes were detectable in the specific wastewater. As with the previous microbiological parameters, detection was carried out with and without PMA treatment of the samples.

For all ARGs, conventional treatment led to moderate reductions between 1 and 2 log units (Figure 6A–E). Removal by conventional treatment was most efficient for tetM, ermB and blaTEM and somewhat less for sul1 and intl1, with PMA treatment leading to a considerable signal reduction. As for the other studied parameters, the strongest signal reduction in ARGs was brought about by MBR treatment. Ozonation led to further reduction in these genes. Especially after PMA treatment, the effect of ozonation was particularly evident in the case of sul1 and intl1, where signal reductions of ≥5 log units were achieved. For the other ARGs, the ozonation effect might not have been so strong due to signals being close to the detection threshold already after MBR treatment.

Figure 6. Concentrations of the antibiotic genes (genome copies/L) (A) sul1, (B) tetM, (C) ermB, (D) blaTEM and (E) the intron intl1 for the respective wastewater treatment stages over 11 months (n = 12). Samples were either left untreated or treated with PMA. For the samples after ozonation, the specific ozone doses ranged between 0.20 and 0.62 mg ozone/mg DOC.

3.5. Relationship Between Specific Ozone Doses and Microbiological Treatment Efficacies

Ozonation was performed in a specific dose range between 0.20 and 0.62 mg ozone/mg DOC. A dose-effect relationship between the described analytical parameters and the ozone concentration could not be found for any parameter in the ozone concentration range investigated here. The correlation between the Bacteroidales and the intl1 concentrations on the one hand and the specific ozone doses on the other hand serves as an example (Figure 7).

Figure 7. Relationship between the log reduction of (A) human Bacteroidales (long amplicons) and (B) intl1 signals (with and without PMA treatment) and the specific ozone concentration (0.20 and 0.62 mg ozone/mg DOC) over the study period of 11 months (n = 11). Log reductions refer to the difference between concentrations in the primary effluent and in the effluent after ozonation.

The lack of a dose–response relationship might, in part, be explained by the signals being near the detection thresholds after MBR treatment. The qPCR detection thresholds for Bacteroidales and the ARGs were calculated to be in the range between 1.17 × 10³ GU/L (sul1) and 2.07 × 10³ GU/L (Bacteroidales long). It should also be noted that ozone concentrations used in this study were in a range adjusted for the removal of organic pollutants. Higher ozone concentrations might be necessary for the efficient destruction of genetic material.

3.6. Retention of Extracellular DNA

Nucleic acids and genes are not only present intracellularly within microorganisms but can also be found outside cells (i.e., extracellularly). While MBR was shown to provide an effective barrier against bacteria and their intracellular DNA, we addressed the question of whether and how efficiently extracellular DNA (exDNA) is retained. To answer this question, samples were taken from the MBR tank and the MBR filtrate. After filtration of the samples to remove bacteria and cells, the filtrates were examined for the presence of free exDNA. For this purpose, the DNA in the cell-free filtrates was concentrated by precipitation and analyzed for the presence of 16S rRNA genes by qPCR. The samples from the MBR tank contained an average of 1.0 × 10⁶ (±3.6 × 10⁵) bacterial genome units (GU)/L, while the samples from the MBR filtrate contained an average of 4.8 × 10⁴ (±8.1 × 10³) bacterial genome units (GU)/L. This difference in free exDNA concentrations between feed and filtrate corresponded to a reduction of 95.3%.

4. Discussion

A consistent understanding of the efficacies of different treatment stages for the removal of relevant pollutants is essential for designing effective mitigation strategies and for informing regulatory frameworks aiming at reducing environmental dissemination. This study characterized a wastewater treatment pilot plant with respect to the removal efficiencies of a suite of microbiological parameters. Results corroborated a strong microbiological benefit of quaternary treatment compared to conventional wastewater treatment. The finding is in line with previous studies reporting only a limited reduction in hygienically relevant bacteria and ARGs by conventional wastewater treatment [30,32]. This is not surprising as WWTPs have historically not been developed for efficient microbiological treatment [5], but for the removal of oxygen demand and nutrients.

The latter was indirectly measured by the assessment of the regrowth potential, which represents the maximal bacterial concentrations supported by the nutrient content of the particular water. Nutrients were reported to have a more substantial impact on bacterial community composition than physicochemical factors [33]. To determine the levels of assimilable nutrients, cell numbers were quantified in a 2-point measurement at the beginning and the end of an incubation at 22 °C. Looking at the entire treatment process, the resulting ICC_day7 values dropped on average from 2 × 10⁷ cells/mL in the samples after primary clarification to 5 × 10⁵ cells/mL after ozonation. The latter is below the typical cell counts of ≥10⁶ intact cells/mL for surface waters in populated areas (e.g., lakes or rivers, [34,35]). The discharge of wastewater, treated in this way, would therefore not lead to an increase in the load of assimilable nutrients and would not add growth potential to the receiving water. Biological post-treatment after ozonation does not appear necessary. The maximal cell numbers even compare favorably with tap water, considering that stagnated, surface water-derived drinking waters have been reported to support maximal cell numbers of up to approx. 10⁶ intact cells/mL in regrowth tests [36,37,38]. It has to be acknowledged, however, that specific cell numbers always depend on the bacterial microbiome that establishes during regrowth.

Actual intact cell concentrations (ICC_day0) dropped from 4 × 10⁷ cells/mL to 9 × 10⁴ cells/mL (after ozonation), corresponding to a log reduction of 2.7. The greatest reduction, not surprisingly, was caused by MBR treatment. As the filtrate is enriched after membrane passage with microorganisms previously attached to the surfaces of the post-filtration compartments, the data do not allow conclusions on the filtration performance or the integrity of the ultrafiltration membrane. Comparing flow cytometric fingerprints of the MBR tank and the MBR filtrate showed distinct differences, indicating that cells found in the filtrate were due to enrichment from surface-attached bacteria or regrowth rather than bacterial breakthrough of the membrane. Data suggested that ultrafiltration resulted in a reset of the bacterial community composition. Similar findings were presented for wastewater treatment processes using reverse osmosis membranes [39] and other ultrafiltration membranes [40]. Longer water stagnation would entail an increase in day 0 cell numbers to the upper limit of the seven-day regrowth potential.

Comparable log-reductions were obtained for MBR-treatment when applying qPCR for quantifying the overall bacterial load, although absolute values for genomic units (qPCR) cannot be directly compared with the cells/mL used in flow cytometry. Without PMA treatment, MBR-removal rates of 3.5 and 3.4 log units were obtained on the basis of short and long amplicons, respectively (Figure 4). PMA treatment increased the MBR removal rate to 3.9 (short amplicons) and 3.7 (long amplicons) log units. Overall, qPCR data based on 16S rRNA gene copy numbers confirmed that MBR had the strongest effect on the bacterial reduction within the treatment train. Ozonation resulted in a less pronounced overall signal reduction, although the effect of PMA treatment was strongest for this treatment step. The latter is not surprising, as an oxidative step is expected to produce the highest proportion of membrane damage, especially when particle numbers have reached low concentrations and ozone demand is low. Ozone is known to inflict both membrane damage and nucleic acid damage [41,42]. The small impact of amplicon length (which, in the absence of PMA treatment, is a sensor for DNA integrity) compared with the pronounced effect of PMA treatment (as a sensor of membrane integrity) suggested that ozonation in the applied Ct range inflicted mainly membrane damage and little DNA damage. This observation is in line with the findings of Jäger et al. [43], who also applied a qPCR approach with different amplicon lengths. The authors reported DNA lesions and other DNA alterations affecting PCR efficiency, after UV-treatment (400 J/m²) as part of advanced wastewater treatment, but not after ozone treatment with even higher ozone doses (1 mg ozone/g DOC) than used in this study [43].

Compared to overall bacterial concentrations, greater log reductions were obtained for the specific bacterial group of Bacteroidales or antibiotic resistance genes. In case of Bacteroidales, conventional treatment achieved reductions in only approx. 2 log units, whereas advanced treatment resulted in reductions close to or below the detection threshold. The greater reduction in comparison to the overall bacterial population might be explained by the inability of this bacterial group to replicate in the WWTP. Bacteroidales, as abundant intestinal obligate anaerobes, are becoming more popular for the evaluation of water treatment and wastewater surveillance [44]. Whereas typically used for microbial source tracking (MST), Demeter et al. [44] emphasized an increasing number of articles reporting the use of genetic MST markers for assessing treatment performance. The high abundance and consistent detection of Bacteroidales result in a higher mean predictive potential than cultivation-based fecal indicator organisms and provide an advantage for measuring high-range log reductions across different treatment stages [44,45]. The previously reported concentration range of Bacteroidales genetic markers between 8 and 10 log₁₀ copies/L [46,47] in raw wastewater aligns very well with the concentrations found in our study in samples after primary clarification. Also, the approx. 2 log reduction in Bacteroidales genetic markers in the conventional treatment correlates well with the reductions reported for different human MST qPCR markers (Bacteroidales-specific crAssphages or PMMoV) in WWTPs with activated sludge (for a review see [44]).

The critical question when applying a new set of indicators is about the relevance of the measured removal rates. In some decentralized wastewater treatment systems, removal rates of Bacteroidales and fecal indicator bacteria were found to be similar [48]. Typically, however, when comparing qPCR with culture-based removal of fecal indicators, the latter are reduced more strongly, leading to an overestimation of treatment efficiency [49]. Removal rates can typically be ranked in the following order: cultivation-based indicators > viability qPCR > qPCR. qPCR in combination with PMA-based live-dead discrimination was deemed suitable to evaluate the elimination processes [49]. The consequence of the faster removal rate of cultivation-based indicators (such as E. coli or Enterococci) is that the resulting low concentrations of culturable bacterial indicators can make them more difficult to accurately quantify and to calculate reliable reductions [50]. In case of membrane processes, typically no culturable indicators or only very low concentrations are found in the filtrates [51,52], making sporadic findings more an indicator of membrane integrity breaches. Overall, a useful fecal indicator is characterized by the following properties: high abundance in feces, easy detection, no growth in the environment or the WWTP and removal rates comparable to relevant pathogens [53]. All these prerequisites seem fulfilled for Bacteroidales, with the only question being about the removal rates in comparison with pathogens. A comparison of Bacteroidales removal with that of traditional fecal markers like E. coli or Enterococci was not part of this study. Typically, genetic markers are, however, more persistent than culturable fecal indicators in wastewater treatment [54]. Discrepancies are mainly obtained for disinfection processes that have a strong impact on culturability, whereas the DNA persists longer [50,55]. This means lower removal rates for Bacteroidales DNA than for colony-forming units of E. coli, making DNA markers harder to eliminate entirely during treatment. Given that DNA markers are therefore conservative indicators that are hard to remove during treatment, it is remarkable that Bacteroidales PMA-qPCR signals were below the detection threshold after MBR and ozonation despite their high persistence and abundance. Efficiency assessment on the basis of genetic markers enables the microbiological evaluation of processes where culturable findings are low or not present. Given the high daily discharges of free genetic material into the environment [56,57], the removal of genetic material might gain relevance in the future in comparison to mere removal of colony-forming units.

The genetic approach is already fully accepted for monitoring antibiotic resistance. Reasons lie in the straightforward detection protocols, the versatility of gene targets once the DNA is available, and the rapid availability of results. Another benefit might be the consideration of the possibly underestimated role of free DNA in the dissemination of antibiotic resistance [58]. Our data corroborates previous reports that conventional treatment is not adequate for the removal of ARGs into the aquatic environment [59]. On average, MBR and ozonation resulted in an additional ARG removal of 3.8 log units after conventional treatment. The strongest reduction was seen for ermB with a removal of 4.3 log units. Like for the other microbiological parameters studied, MBR treatment had the strongest impact on log removal of ARGs. This is in line with previous findings reporting that ultrafiltration was most effective in removing antibiotic-resistant bacteria, whereas ozone was less effective compared to ultrafiltration [32]. Authors report that ultrafiltration removed an average of 5 log units of both ARGs and facultative pathogenic bacteria, while ozonation adjusted to micro-pollutant treatment removed about 2 log units on average. An overall reduction of 4–5 log units (min. 99.99%), including ARGs and facultative pathogenic bacteria through the entire treatment train, was seen as a reasonable reference value for treatment efficiency [32]. This was the case of the treatment combination in our study.

Last but not least, it could be confirmed that MBR membranes contribute to the removal of extracellular DNA. Free-floating DNA that passes a 0.1 µm filter was reduced by 95.3%. The technical procedure did not account for free DNA attached to larger particles or colloids, which have also been reported to be removed by ultrafiltration membranes [60]. The reduction found here is in good agreement with reductions in free DNA containing ARGs between 95 and 99.8% reported for different ultrafiltration membrane types [61]. The same authors discussed retention mechanisms mainly including size exclusion, membrane-surface charge and adsorption [61]. Overall, the retention of extracellular DNA aligns well with previous reports that advanced treatment processes like ultrafiltration or ozonation showed the lowest concentrations of exDNA [62]. Despite the low concentrations, exDNA accounted for up to 80% of the total DNA in the treated wastewater effluents [62].

5. Conclusions

The study corroborates that the combination of MBR and ozone treatment offers significant microbiological added value compared to conventional treatment. While conventional treatment only achieved a moderate reduction in the selected microbiological parameters (cell count, bacterial regrowth potential, bacterial gene load and antibiotic resistance genes), the fourth treatment stage had a significantly higher elimination performance. This was primarily attributed to the MBR process, resulting in the largest LOG reductions. In addition to the removal of cells, MBR was also shown to lead to a reset of the bacterial community composition and to contribute to the retention of free extracellular DNA with a reduction of 95.3%. The low signals after MBR might have contributed to the observation that ozonation did not exhibit a measurable dose–response relationship. Ozonation, however, led to a further reduction, especially for parameters that were not close to the detection limit of the corresponding qPCR reaction. The microbiological mechanism of ozone in the studied Ct range seemed mainly attributable to bacterial membrane damage rather than to DNA damage. The use of the highly abundant genetic marker of human-associated Bacteroidales allowed for the quantification of the removal efficiency after the MBR ultrafiltration, where typically no culturable fecal indicators are present. Despite the high persistence of this genetic marker during treatment, the combination of long amplicon qPCR and live-dead distinction showed Bacteroidales signals to be below the detection threshold, corroborating efficient treatment performance.