Next Article in Journal
Hydrophobic Natural Deep Eutectic Solvents for Extraction of Bioactive Compounds: Multiscale Characterization, Quantum Simulations, and Molecular Interaction Studies with Cry j 1 and Amb a 1 Allergens
Previous Article in Journal
Enhancing Copper Leaching from Refractory Copper Oxide Ore Using Organic Cationic Surfactant
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Separations
Volume 12
Issue 8
10.3390/separations12080213
separations-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Ultrafiltration of Water Has a Temporary Effect on Cell Numbers, but Profoundly Changes the Composition of Bacterial Populations—The ‘Reset’ Phenomenon

by
Benjamin H. Meyer
1,*,
Bernd Bendinger
1,
Martin Hippelein
2 and
Andreas Nocker
1,*
1
IWW Institute for Water Research, Moritzstrasse 26, 45476 Mülheim an der Ruhr, Germany
2
Institute for Infection Medicine, Christian-Albrecht University of Kiel, University Medical Center Schleswig-Holstein, 24105 Kiel, Germany
*
Authors to whom correspondence should be addressed.
Separations 2025, 12(8), 213; https://doi.org/10.3390/separations12080213
Submission received: 8 July 2025 / Revised: 11 August 2025 / Accepted: 12 August 2025 / Published: 15 August 2025
(This article belongs to the Special Issue Sustainable Biological Removal and Separation Processes in Wastewater Treatment)
Browse Figures
Versions Notes

Abstract

Ultrafiltration strips water of bacteria. The common misconception is that the filtrate is thus free of bacteria. This only applies, however, in the case that the filtrate compartment is sterile. In real-world applications, the filtrate is rapidly re-colonized, followed by regrowth. In extreme cases of low water usage, the cell numbers in the filtrate can even exceed those in the feed water, probably due to a combination of the microbial enrichment of the bulk water from surfaces, regrowth in the water body itself, and nutrient enrichment on the filter membrane. Regrowth is made possible because dissolved nutrients can freely pass through the membranes. This explains why the initial decrease in cell numbers in drinking water installation systems with ultrafiltration is often followed by an increase in the periphery of the plumbing system. The extent of actual regrowth hereby depends mostly on water usage behaviours. A shorter frequency of membrane wash cycles is beneficial for reducing cell numbers. Neither frequent wash cycles nor cleaning in place (CIP) in filtration units, however, seem to modulate the maximal regrowth potential. Although the effect of ultrafiltration on cell numbers is not sustainable, it causes profound changes in the bacterial communities, with highly distinct populations in the feed water and the filtrate. The microbiological “reset” is demonstrated using examples both from the fields of drinking water and water reuse. Overall, our results suggest that ultrafiltration has a profound impact on the microbiome, but the cell numbers in filtrates depend mostly on the water usage and operational conditions.

Graphical Abstract

1. Introduction

Ultrafiltration (UF) is an established water treatment process used to retain macromolecular substances, suspended solids and particles. The technology is widely used in many industries, including food and beverage, biotechnology, pharmaceutics and healthcare, but also by mining, chemical or petrochemical companies. In the case of water treatment, its major applications in water treatment are the removal of turbidity, natural organic matter, flocs or microplastic from water [1,2]. Especially in water reuse applications, UF is often integrated into the water treatment process [3]. From a microbiological point of view, ultrafiltration with a typical pore size of 20–30 nm effectively retains bacteria and larger cells [4], but also most viruses [5]. The retention rates for bacteria are often stated with at least 6-log units [6]. As a result, UF can physically disinfect water, thereby providing an important contribution to the hygienization of the corresponding water. A common misconception, however, is that the filtered water is free of microorganisms. While this can apply to fresh filtrates, the treated water typically does not stay free of microbes [7]. The main reason is that the receiving pipes or water storage compartments are not sterile. Microorganisms are thus reintroduced into the filtered water.
While the majority of UF units serve to treat water on a large scale, small-scale UF systems have been increasingly used in the drinking water installation (DWI) of large buildings to remove unwanted bacteria and particles [8,9]. The primary intention when using UF is to increase water quality via particle retention, including the removal of unwanted bacteria like Legionella. The latter goes along with the hope to be able to reduce the hot water temperature [8], which is intended as a thermal barrier for Legionella. In drinking water installations with central warm water production systems, the temperature is typically set to 60 °C at the exit of hot water boilers [10]. Being able to reduce the hot water temperature would help reduce the overall energy demand of buildings.
Currently, we lack understanding of the extent of the bacterial regrowth potential after the UF treatment, the sustainability of ultrafiltration and the factors influencing regrowth. Studies have shown that UF membranes efficiently retain particulates (including bacteria) and high-molecular-weight (MW) organic compounds, whereas this does not apply to dissolved low-MW organic compounds [11]. While the high-MW organic compounds are typically degraded by microorganisms more slowly, the low-MW compounds tend to be easily biodegradable with faster kinetics and constitute the major part of the assimilable organic carbon (AOC) [11]. It has been shown in laboratory experiments that the extent of the reduction in bacterial regrowth by membrane filtration is thus limited [7].
In this study, we examined the sustainability of ultrafiltration in real-world applications. The first example comprises small-scale ultrafiltration units placed at the point of entry (POE) of drinking water in drinking water installations (DWI). The questions associated with the ultrafiltration of cold drinking water were about the reductions in cell numbers and regrowth potentials, as well as the sustainability of the effects towards the periphery of the corresponding plumbing systems and over time. Another specific aspect was the effect of UF backflush cycles and cleaning in place (CIP). The second example is from the field of industrial water reuse, addressing the effect of ultrafiltration as part of a treatment process converting whey condensates in a dairy company into high-quality water. The analyses were mostly performed using flow cytometry and were rounded up by high-resolution next-generation sequencing to assess the effect of UF on bacterial communities.

2. Materials and Methods

2.1. Sampling Locations and UF-System

In the case of drinking water installations, UF systems were installed at the POE within the framework of the research project UltraF (BMWK, FKZ 03ET1617A). The corresponding buildings, 1–4, comprised multi-family houses (building 1–3) and a student dormitory (building 4) in different German cities (Table 1). In all buildings, UF units equipped with multibore® membranes (inge GmbH, Greifenberg, Germany) with a membrane pore size of 20 nanometers were installed at the point of entry (POE) of the drinking water into the DWI. The drinking water did not contain disinfectant residual.
Building 3 contained a small UF system with a crossflow filtration unit consisting of two filter modules (5 m3/h). It used backwash with filtered water to clean the filter modules. Building 4 contained a large single-filter UF system with dead-end filtration (9.7 m3/h). The cleaning mechanism relied on forward flush with unfiltered water for membrane cleaning.
Reuse samples were collected from a pilot plant located on the production site of a dairy plant in Edewecht (Lower Saxony, Germany), belonging to the dairy company Deutsches Milchkontor (DMK, Edewecht, Germany), within the framework of the EU research project B-WaterSmart (Horizon 2020 grant agreement nr. 869171). The pilot plant was built by Envirochemie GmbH (Rossdorf, Germany) to treat whey vapour condensates resulting from the production of dairy products. The hybrid treatment included both biological and physical treatment. Biological treatment was based on a moving bed bioreactor, a fluidized bed reactor and a multilayer filter [12]. The subsequent physical treatment included ultrafiltration and reverse osmosis. The hollow fiber UF module (Toray HFUG-2020AN, surface area of 90 m2) had a pore size of 10 nm and was run in dead-end mode for intervals of 20 to 30 min between backflushes. Samples were taken from the runoff of the multilayer filter (UF feed), the fresh UF filtrate and the effluent of the UF filtrate tank. Samples were taken at four time points over a time period of 13 months. Due to variations in the absolute cell numbers over the optimization of the pilot plant, only the cell numbers for one sampling date (11 December 2023) are shown; however, the overall trends remained the same. Samples for nanopore sequencing were taken on the same sampling date.

2.2. Preparation of AOC-Free Glassware for Regrowth Measurements

An alkaline solution of potassium permanganate was made by separately dissolving (A) 30 g of permanganate (cat.nr. 105082; Merck KGaA, Darmstadt, Germany) and (B) 100 g sodium hydroxide pellets (cat.nr. 106498; Merck KGaA, Darmstadt, Germany) in 500 mL of deionized water each and equilibrated to room temperature. The two solutions were subsequently combined and stored in a 1 L bottle in the dark prior to use. Clean glass vials (50 mL, cat.nr. 7612150; Th. Geyer GmbH & Co.KG, Renningen, Germany) were first washed with deionized water and then filled with this alkaline solution and sealed with appropriate caps (cat. nr. 292401305, DWK Life Sciences, Wertheim, Germany). The entire surface area of the glassware was in contact with the solution. The permanganate solution was decanted (for recycling purposes) and glassware and caps were rinsed three times with tap water and twice with deionized water. Dried glassware and caps were wrapped in aluminum foil. Glassware was heated at 280 °C for at least 8 h or overnight, and caps were heated at 180 °C during that time. Caps were provided with a Teflon seal (special fabrication by Max Werth GmbH Co. KG, Mülheim an der Ruhr, Germany) prior to use. Other glassware was treated accordingly.

2.3. Sample Collection

Samples of the drinking water (potable water cold, PWC) prior and after ultrafiltration were collected directly in 1000 mL sample flasks at the point of (water) entry (POE) in the building, after the UF unit, and additionally at the entry of the drinking water heater in buildings 3 and 4. Furthermore, the 2nd liter of PWC at 2 to 6 different point of use (POU) locations distributed over the entirety of each building (building 1–4) was collected in 1000 mL sample flasks. The sampling procedure included the removal of tap attachments, disinfection and the flushing of the first liter to avoid the sampling of stagnated water. Sampling was performed over a time period of 12 months in buildings 3 and 4, over 21 months in building 1 and over 39 months in building 2.
For regrowth measurements, samples were collected in 250 mL AOC-free glass bottles (Schott AG, Mainz, Germany). Unfiltered drinking water prior to the UF was used as a control. Samples of filtered water were taken directly after the UF and before the entry of the drinking water heater. Samples were taken on three to four different sampling dates spread over approximately three months. The drinking water did not contain any disinfectant. Samples were sent cold over night to the laboratory.
In the case of samples from the dairy company, water volumes of approximately 400 mL were collected in AOC-free borosilicate bottles for flow cytometric analysis. To conserve the microbiological status of the samples and to avoid the loss of cellular viability, samples were transported cold (4 ± 2 °C) to the laboratory and analyzed within 24 h after sampling. For nanopore sequencing, 2 L samples volumes were collected in sterile glassware and transported cold to the laboratory.

2.4. Flow Cytometry

Total cell numbers were quantified using flow cytometry. For this purpose, cells were stained with the fluorescent dye SYBR Green I (10,000 × stock; cat.nr. S-7567; Life Technologies Ltd., Darmstadt, Germany). The dye was diluted to a working stock concentration of 100 × using dimethylsulfoxide (DMSO; Sigma Aldrich, Steinheim am Albuch, Germany) and stored at −20 °C until use. For analysis, aliquots (250 µL) of water samples were transferred into a 96-well plate (cat. nr. 601808, HJ-Bioanalytik GmbH, Erkelenz, Germany). Each sample was analyzed three times (three technical replicates). Water sample aliquots of 200 µL were transferred from this plate into the wells of a second 96-well plate with pre-aliquoted 2 µL of the 100 × SG working stock solution in each well using a 12-channel pipette. The samples and dye solution were thoroughly mixed by pipetting up and down several times. Staining was performed at 37 °C for 13 min in an air incubator. Data was collected using a Novocyte benchtop instrument (ACEA Biosciences Inc., CA, USA) equipped with a 488 nm laser and an autosampler to enable analysis on a 96-well plate basis. Data were analyzed using the instrument-specific NovoExpress® software (version 1.6.2) and a gating procedure similar to the one described by Gatza et al. [13]. The settings included an analysis speed of 66 µL/min and the trigger was set to green fluorescence (FITC channel) with a detection threshold of 600 fluorescence units. Ultrapure water supplemented with a bacteriostatic solution (BD Biosciences, NJ, USA) was used as sheath fluid. In the case that the total signals exceeded approximately 5000 signals/sec, water samples were diluted with 0.1 µm of filtered mineral water (Evian, Evian-les-Bains, France). Prior to the measurement of samples, a 0.5% chlorine solution was used to clean the flow cytometric system. The oxidant was removed by subsequently flushing the system for two minutes with filtered Evian water. The cleaning procedure allowed a detection threshold of around 200–2000 cells/mL.

2.5. Assessment of Regrowth Potential

After the quantification of day 0 cell numbers, water samples from the 250 mL sample bottles were split into three 50 mL AOC-free borosilicate glass vials sealed with a cap with Teflon sealant. Vials were filled up to a 20 mL mark. These vials were incubated at 22 °C (±1 °C) for seven days to allow bacterial growth, followed by the repeated flow cytometric assessment of cell numbers. The resulting cell numbers are referred to as day 7 values. In the case of ultrafiltrates, regrowth was caused by autochthonous bacteria that were contained in the filtrates and that presumably originated from the non-sterile surfaces of the receiving filtrate compartments.

2.6. Genomic DNA Extraction and Sequencing

For microbial community analysis of the drinking water, 0.5 L of each sample was filtered through 0.2 µm (0.2 µm polycarbonate filters; Pieper Filter GmbH, Bad Zwischenahn, Germany) using a vacuum manifold. Genomic DNA was extracted using the Aquadien™ DNA Extraction and Purification Kit (cat. nr. 3578121, Bio-Rad Laboratories GmbH, Feldkirchen, Germany). For the drinking water samples, Illumina MiSeq sequencing was performed by Eurofins Genomics (Ebersberg, Germany) after amplifying the V3/4 region of the 16S rRNA gene (467 bp ± 20 bp) using universal primers (forward: 5′-TCGTCGGCAGCGTCAGATGT GTATAAGAGACAGCCTACGGGNGGCWGCAG-3′ and reverse: 5′-GTCTCGTGG GCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′). An insufficient PCR product for sequence analysis was obtained for the sample taken directly after ultrafiltration; therefore, samples from before entry into the drinking water heater (DWH) were used (distance to the UF: approx. 2 m). A sufficient PCR product for sequence analysis could only be obtained in the case of drinking water from building 4 as the drinking water in the other buildings did not contain sufficient bacterial biomass to provide enough PCR product for sequence analysis.
For the treatment of whey vapour condensates, 2 L sample volumes were collected in sterile glassware, transported cold to the laboratory, and filtered using 0.2 µm polycarbonate filters (Pieper Filter GmbH, Bad Zwischenahn, Germany). As in the case of drinking water, insufficient PCR product for sequence analysis was obtained from the sample taken directly after UF filtration. The effluent of the multilayer filter (i.e., UF inflow) was therefore compared with effluent of the UF filtrate tank, which contained higher cell numbers than the fresh filtrate. Genomic DNA was extracted using the FAST DNA SPIN kit from MP Biomedicals. Nanopore sequencing was performed by Biofidus AG (Bielefeld, Germany). Library preparation and sequencing was performed according to the manufacturer’s instructions (SQK-16S114.24, Oxford Nanopore Technologies, Oxford, UK). Sequencing was performed using an R10.4.1 flow cell (FLO-MIN114, Oxford Nanopore Technologies) applying the super-accurate basecalling mode.

2.7. Sequence Analysis

Sequence analysis was performed on the basis of 16S rRNA genes. The sequence analyses of the drinking water samples were performed by the sequencing company Eurofins Genomics (Ebersberg, Germany). In detail, paired-end reads were initially processed using cutadapt [14] (version 2.7) to remove primer and adapter sequences. As a first step of the microbiome analysis, all reads with ambiguous bases (‘N’) were removed. Chimeric reads were identified and removed based on the de novo algorithm of UCHIME [15], as implemented in the VSEARCH package [16]. The remaining set of high-quality reads was processed using minimum entropy decomposition [17,18]. Minimum Entropy Decomposition (MED) provides a computationally efficient means of partitioning marker gene datasets into OTUs (Operational Taxonomic Units). Each OTU represents a distinct cluster with significant sequence divergence to any other cluster. By employing Shannon entropy, MED only uses the information-rich nucleotide positions across reads and iteratively partitions large datasets while omitting stochastic variation. The MED procedure outperforms classical, identity-based clustering algorithms. Sequences can be partitioned based on relevant single-nucleotide differences without being susceptible to random sequencing errors. This allows the decomposition of sequence datasets with a single nucleotide resolution. Furthermore, the MED procedure identifies and filters random “noise” in the dataset, i.e., sequences with a very low abundance (less than ≈0.02% of the average sample size). To assign taxonomic information to each OTU, DC-MEGABLAST alignments of cluster-representative sequences to the sequence database were performed. A specific taxonomic assignment for each OTU was then transferred from the set of best-matching reference sequences (lowest common taxonomic unit of all best hits). Hereby, a sequence identity of 70% across at least 80% of the representative sequence was a minimal requirement for considering the reference sequences. The further processing of OTUs and taxonomic assignments was performed using the QIIME software package [19] version 1.9.1, http://qiime.org/ (accessed on 14 September 2022). The abundances of bacterial taxonomic units were normalized using the lineage-specific copy numbers of the relevant marker genes to improve estimates [20]. Shannon–Wiener indices were created for the description of diversity [21].
For the nanopore sequencing of the whey condensate samples, full-length 16S rRNA gene amplicons were quality-controlled using Filtlong (https://github.com/rrwick/Filtlong, accessed on 27 May 2024) to a minimum mean quality score of 25 and filtered for size (1200–1650 bp) using a custom awk script. VSEARCH (v2.14.2, https://github.com/torognes/vsearch, accessed on 27 May 2024) was used to dereplicate quality-controlled reads, which were then aligned to the SILVA 138 reference database (https://academic.oup.com/nar/article/41/D1/D590/1069277, accessed on 27 May 2024) using minimap2 (v2.17-r941, https://academic.oup.com/bioinformatics/article/34/18/3094/4994778, accessed on 27 May 2024). The resulting alignments were filtered to retain only those with ≤150 mismatches using a custom awk script and to remove low-quality alignments (mapping quality < 20) with SAMtools (v1.10, https://academic.oup.com/bioinformatics/article/25/16/2078/204688, accessed on 27 May 2024). Sorted and indexed alignments were then used to extract read clusters, which were matched against the SILVA 138 reference database, generating a sample-specific OTU table with taxonomic annotations. Sample-specific OTU tables were finally concatenated into an abundance matrix, which was then imported into R for processing and visualization.

3. Results

3.1. Cell Numbers and Regrowth Potential of the Microbiome After UF

The effect of UF on cell numbers was assessed using flow cytometry and is shown for two very different applications applying UF to (A) drinking water to protect the drinking water installation (DWI) and (B) to biologically treated whey condensates (B). In the case of the drinking water, UF systems were installed in three different buildings at the point of entry (POE) between the public distribution system and the plumbing system (Figure 1A). In the case of whey condensates, ultrafiltration was the first membrane treatment step after biological treatment, which consisted of a moving bed bioreactor, a fluidized bed reactor and a multilayer filter (Figure 1B). Water samples were collected in AOC-free glass bottles and transported cold to the laboratory to conserve the microbiological status at the time point of sampling (day 0).
In the drinking water installation, the total cell counts (TCCs) measured in each building at 3–4 different time points declined, on average, from 75,000 to 110,000 cells/mL at the POE to 800–2700 total cells per mL after the UF treatment, corresponding to a decrease of 97–99% (Figure 1A). In the case of the biologically treated whey condensate, the total cell counts (TCCday 0) dropped from 1600.000 to 18,000 cells/mL, corresponding to >99% (Figure 1B).
After the quantification of day 0 cell numbers, the samples were incubated for seven days at 22 °C to allow the regrowth of the given microbiome on the basis of the assimilable nutrients contained in the corresponding samples. In the case of ultrafiltered drinking water, the TCC increased to 0.7–1.9 × 106 cells/mL, representing a 684- to 3600-fold increase (Figure 1A). The cell numbers in the unfiltered control increased to 1.2–2.0 × 106 cells/mL, corresponding to a 11- to 27-fold increase. Therefore, the TCC in the sample after UF and after regrowth reached levels comparable to those of the regrown sample that was not subjected to UF.
In the case of biologically pre-treated whey condensates, the 7-day incubation of the unfiltered sample resulted in a reduction in the cell count from 1.6 × 106 to 4.1 × 105 TCC/mL (Figure 1B), probably due to the establishment of a different microbiome. The decrease in cell counts was also observed on other sampling dates. The fresh UF filtrate, on the other hand, showed an increase in cell numbers from 1.8 × 104 to 4.9 × 105 TCC/mL, corresponding to a 27-fold increase. The subsequent passage of the ultrafiltrate through the filtrate tank typically resulted in a substantial increase in TCCday 0 (3.3 × 105 TCC/mL) and TCCday 7 values (2.0 × 106 TCC/mL). The reason for the increase in TCCday 0 values lies probably in the migration of surface-bound bacteria into the filtrate, which would suggest that the TCCday 0 greatly depended on the cleanliness of the filtrate tank. Cleaning prior to sampling typically resulted in lower cell numbers. The increase in the regrowth potential in the filtrate tank, on the other hand, was probably due to the addition of a biodegradable corrosion inhibitor.
Overall, the results corroborate the sufficient availability of dissolved nutrients in the filtrates to support substantial regrowth. The removal of particulates does not have a substantial effect on the regrowth potential. In the following, the focus was on the sustainability of the filtration effect and the influence of the operational conditions of the UF system.

3.2. Effect of UF Dimension and Backwash Frequency on Actual Cell Numbers in Filtrates

To assess the impact of membrane wash cycles and the performance of UF units in real-world applications, two different DWIs equipped with different UF units were compared. The wash frequencies in buildings 3 and 4 were set to 4 h and 24 h, respectively. The UF units in the two buildings furthermore differed in their dimensions. While the UF unit in building 3 consisted of two small filter units and a filtration capacity of 5 m3/h, the single UF unit in building 4 was substantially larger, with a filtration capacity of 9.7 m3/h (Table 1). The two UF systems also differed in their cleaning mechanisms to prevent membranes from clogging and to prevent excessive pressure loss across the membranes. The units in building 3 were cleaned by classical backwash (water flowing in the opposite direction), whereas the unit in building 4 was cleaned by forward flush (water flowing parallel to the membrane in crossflow to loosen the filter cake).
Cell numbers at day 0 were assessed in cold water (potable water cold, PWC) samples from the POE prior to ultrafiltration, directly after the UF unit and at the entrance into a drinking water heater (DWH). The latter served for the production of hot potable water (PWH). Samples were taken at 13 different time points over one full year. The wash frequencies were kept constant during this time period.
In building 3, the cell numbers in PWC samples at the POE varied and ranged between 32,000 and 122,000 TCC/mL (Figure 2A, left). This variation in cell numbers at the POE occurred over time and corresponds to changes in the cell count of the supplied drinking water and were not affected by the installation of the UF unit. The cell numbers at the sampling point after the UF unit were low, with 6500 TCC/mL at the first time of sampling, and further decreased in the remaining study period to 800–1200 TCC/mL. Compared to the cell numbers at the POE, ultrafiltration resulted in a reduction in cell numbers by 96–99% (Figure 2A, middle). At the sampling point at the entry of the DWH, the installation of the UF unit resulted in a strong reduction in cell numbers (Figure 2A, right). While the cell numbers were still slightly elevated seven days after the installation of the UF unit, they became smaller in the following sampling events. Not considering the first sampling event after UF installation, the average reduction in cell numbers at the entry of the DWH accounted for 96% (92–99%). The installation of the UF unit therefore led to a continued decrease in cell numbers over the study time period (Figure 2A).
The effect of the UF system in the DWI of building 4 was different. This building, a newly constructed dormitory, was equipped with a UF unit prior to the commissioning of the DWI with drinking water. Furthermore, the wash frequency of the large UF system with one membrane filter unit was set to 24 h. The cell numbers in the drinking water at the POE ranged between 6000 and 18,000 cells/mL over the study period (Figure 2B, left). The highest cell numbers coincided with public holidays, which typically correlated with lower water consumption in the dormitory and therefore increased water age prior to consumption (Figure 2B, green background). When comparing the cell numbers at the POE with the ones after the UF and prior to the DWH, cell numbers were surprisingly higher after UF than at the POE before filtration in the first sampling rounds (Figure 2B, middle). The UF in this case did not result in reduced cell numbers. As the membrane integrity was confirmed by pressure tests, it was likely that the higher cell counts after the UF were caused by biofouling on the membrane filtrate site, where the microbiome can proliferate to high cell numbers and release cells into the drinking water. The daily physical wash events by forward flushes can only remove the microorganisms enriched on the feed side, and not the microorganisms growing on the filtrate side of the membrane. Consequently, the UF unit was subjected to a chemical cleaning-in-place (CIP) procedure to remove any biomass on the filtrate side. Only after the CIP did the TCCday0 levels drop below the ones prior to UF (Figure 2B, middle). Six days after the CIP, the TCCday0 reduction corresponded to 63% compared to pre-UF concentrations at the POE. Half a year later, the TCC reduction still accounted for 57–89%; however, the cell reductions did not reach levels of 96–99% after the UF system, as seen in buildings 1–3. A similar tendency could be observed at the entry of the DWH. The measured cell counts at the DWH were, however, slightly higher than directly after UF, indicating regrowth in the DWI (Figure 2B, right).
Overall, the results indicate that shorter wash frequencies are beneficial for reducing cell numbers in the filtrate. The data also shows the efficiency of a CIP to reduce elevated TCCday 0 values in the filtrate.

3.3. Effect of UF Dimension and Backwash Frequency on Regrowth Potentials in Filtrates

Next, the regrowth potentials in the two buildings were analyzed for DWIs in buildings 3 and 4 in more detail. Whereas the installation of the UF unit in building 3 led to a TCCday 0 reduction of 97–99% in the filtrate (compared to the unfiltered drinking water, Figure 2A), the regrowth potentials remained comparable to that of unfiltered drinking water at the POE not only in the fresh filtrate, but also at the entry of the DWH, with an average of 1.9 × 106 TCCday 7/mL (Figure 3A left). This finding suggested no or only a small effect of UF on regrowth potentials.
In building 4, where the CIP was successful in reducing TCCday 0 values, the regrowth potential of the ultrafiltered sample was higher than the unfiltered water at the POE (Figure 3A right). The TCCday 7 values were thereby subject to some variation over time. At the POE, the regrowth potentials increased over time, averaging 0.3 × 106 TCCday 7/mL, while directly after the UF and at the entry of the DHW, the TCCday 7 values remained relatively constant, with an average of 0.83 and 0.87 × 106 TCCday 7/mL, respectively. The chemical cleaning in place (CIP) therefore did not change the regrowth potential after the UF (Figure 3B).
Overall, the regrowth potentials of the drinking water did not undergo a substantial change in any of the two buildings because of ultrafiltration. This indicates that the wash frequency and the CIP influence the actual cell numbers (reflected by TCCday 0 values), but not the regrowth potential of the drinking water.

3.4. Sustainability of the UF Effect Within a DWI

To obtain an insight into the extent of microbial regrowth in drinking water on its way from the ultrafiltration unit into the periphery, we looked at the relative changes in TCCday 0 values within the DWI of buildings 1–4 over one year. Specifically, the total cell numbers directly after UF and at different points-of-use (POU) sampling sites were compared with the ones in the drinking water at the POE. All buildings were equipped with UF units at the POE. Cell numbers in the POU sampling sites tended to be above those directly after ultrafiltration. In buildings 1–3, the total cell numbers directly after the UF units showed an average reduction of 95% (92–97%) compared to unfiltered drinking water at the POE (Figure 4). In building 4, after the CIP, an average reduction after the UF of 73% was apparent over the remaining year. The low reduction in TCC observed after the UF is probably due to the persistence of conditions that initially led to membrane biofouling. These conditions remained throughout the study, causing a renewed increase in cell concentrations on the filtrate side of the membrane, particularly during periods of low water consumption (Figure 4).
For POU samples, taken in the periphery of the DWIs in building 1–3, the reduction in cell numbers amounted, on average, to only 76% compared to the POE. This finding therefore suggested that the cell numbers in drinking water had undergone an average increase of 19% after ultrafiltration during passage through the DWI. In building 4, the reduction in cell numbers at the periphery was only 10% compared to the POE. This indicates that the cell concentrations in the DWI increased by 63% after ultrafiltration. The increase is primarily attributed to extended stagnation periods during holiday breaks, during which higher TCC values were measured compared to periods of regular water usage.
The reason for this increase in the cells concentration is the microbial regrowth within the installation, reducing the effect of ultrafiltration at the POU. In all buildings, the TCC values in the periphery of the DWI showed higher variations than the ones in the proximity of the POE, probably due to varying stagnation times. Overall, the ultrafiltration of drinking water at the house entrance led to substantially lower cell numbers in the drinking water in the periphery, but the UF effect was diminished by regrowth. The extent of regrowth could, however, strongly vary in different buildings (Figure 4).

3.5. Sustainability of the UF Effect over Time

The observed effect of ultrafiltration on the reduction in the total cell numbers in drinking water was maintained over time at the entrance of the DWH (Figure 2). The DWH was located in the middle of the DWI. To examine whether this also held true in the periphery of a DWI, we examined the TCCday 0 values at different POU sampling points over approx. one year. The multifamily building 3 served as an example. After the installation of the UF unit, cell concentrations dropped most significant for the samples taken directly after UF and at the DWH (Figure 5). The TCCday 0 values in drinking water sampled at the three different faucets in different bathrooms showed less pronounced reductions, but were significantly below those of the unfiltered drinking water at the POE. The higher cell numbers at the peripheric sampling sites were in line with the increase in cell numbers during passage through the DWI.
Although the effect of ultrafiltration was diminished by regrowth and the UF effect varied between the different POU sampling points, the reduction in cell numbers was sustained over the study time of one year.

3.6. Effect of Ultrafiltration on Bacterial Community Composition

In the next step, we investigated the impact of UF on the composition of bacterial communities. Sequence analysis was conducted for both drinking water (A) and whey vapor condensates (B). A comparison of samples collected before and after UF revealed substantial shifts in bacterial community structures.
For drinking water, the number of detected bacterial families decreased markedly from 72 to 29 following UF treatment. Correspondingly, the Shannon–Wiener diversity index declined from 1.4 at the point of entry to 1.1 after UF (Figure 6 and Supplementary Figure S1). At the point of entry, the cold water was dominated by two families: Nitrospiraceae (17.6%) and Thioprofundaceae (14.1%); together, these accounted for 32% of the total bacterial abundance. Other highly abundant families included an unclassified Alphaproteobacteria family (7.4%), Gallionellaceae (5.9%), Caulobacteraceae (3.6%), and Rhodocyclaceae (3.4%). Collectively, these six most abundant families represented 53% of the total community (Figure 6). After UF, the unclassified Alphaproteobacteria family increased in relative abundance from 7.4% to 19.9%, becoming the most dominant family. Other prominent families included Bdellovibrionaceae (15.1%), Rhodobacteraceae (10.2%), Comamonadaceae (8.6%), Haematococcaceae (6.3%), and Chitinophagaceae (5.8%). Together, these six families accounted for 66% of the total bacterial abundance (Figure 6).
UF also had a substantial impact on the bacterial community composition in the case of biologically treated whey vapour condensate. Here, the number of detected bacterial families also decreased markedly from 65 to 43 following UF treatment (Figure 6, Supplementary Figure S1). The bacterial community in the unfiltered whey condensate was dominated by the two families Oxalobacteraceae (54.3%) and Comamonadaceae (28.1%), which together comprised 82% of the total bacterial abundance. Other abundant families included Spirosomaceae (6.2%), Rhodocyclaceae (2.5%), Chitinophagaceae (2.4%), and Sphingobacteriaceae (1.6%). These six families collectively represented 95% of the total community (Figure 6). Following UF treatment, the relative abundance of Sphingobacteriaceae increased dramatically from 1.6% to 49.2%, becoming the most dominant family. Additional abundant families included Weeksellaceae (15.5%), Sphingomonadaceae (6.7%), Comamonadaceae (which decreased from 28.1% to 5.9%), Saccharimonadaceae (4.5%), and Oxalobacteraceae (which decreased from 54.3% to 4.3%). These six families together accounted for 86% of the total bacterial abundance (Figure 6).
These results demonstrate that ultrafiltration caused a profound change in the composition of the bacterial communities. The bacterial populations in the feed and the filtrate were highly distinct. In the case of drinking water, the impact of UF could also be observed in more peripheral sampling sites, meaning that the effect on the microbiome was maintained within the DWI.

4. Discussion

The study addressed the effect of ultrafiltration on cell numbers, regrowth potentials and the microbiome. Examples were presented from two very different applications: drinking water in plumbing systems and water production from whey vapour condensates as a side product of dairy production.
The primary motivation to employ ultrafiltration from a microbiological point of view is to reduce cell numbers. UF membranes typically have bacterial retention efficiencies of at least 6 log-units (i.e., ≥99.9999% [22,23]). These are determined based on standardized laboratory protocols. In practice, the quantification of cell numbers by flow cytometry, on the other hand, revealed typical reductions in cell signals that ranged between 91 and 99% (Figure 1 and Figure 4); in extreme cases, as in building 4, these were even less. This is in good agreement with another study on a multi-family building in Germany that reported a UF-induced ‘cleaning effect’ of 91% of the TCC [8]. The discrepancy does, however, not reflect the efficiency of the actual bacterial retention or the integrity of the membranes. The detection of cells in the filtrates rather owes itself to the fact that the filtrate compartments are not sterile. The microorganisms populating the surfaces of these compartments are in equilibrium with the bulk water [24].
The data indicated that the cell numbers strongly depend on the water usage profile, which determines the time of stagnation. In plumbing systems, the drinking water stagnates on average 23 h per day [25]. Given time, the bacteria in the drinking water proliferate [26,27]. The regrowth kinetics are strongly dependent on the ambient temperature, but in the case of drinking water, weak regrowth can already be detected after overnight stagnation [26] with peak regrowth occurring from day 4 to day 7 [7,27,28]. This is made possible by the existence of a nutrient basis. Part of those nutrients originate from some of the materials used in plumbing systems. Elastomeric materials like EPDM and other plastic materials are known to promote bacterial growth [29]. These nutrients add to the regrowth potential that is intrinsic to the corresponding drinking water. In case of drinking waters that originate from treated surface waters, the regrowth potential is primarily caused by oxidative treatment steps like ozonation or chlorine-based disinfection [28]. Oxidative treatments result in the release of dissolved nutrients that were formerly biologically fixed in form of biomass. Furthermore, oxidative processes lead to a conversion of DOC into AOC by breakdown of complex organic molecules that become more bioavailable [30]. Regrowth is hereby not limited to drinking water without a disinfection residual (as is the case in this study), but also occurs in plumbing systems receiving drinking water with a secondary disinfectant [31,32]. In the case of whey, the contained organic constituents are known to be biodegradable [33,34]. These assimilable nutrients are typically dissolved low MW molecules, which are not efficiently retained by ultrafiltration and that can pass membranes unhindered [11,35].
The extent of regrowth could strongly vary in different buildings (Figure 5). Apart from water usage and different materials, regrowth is known to depend on the provided drinking water itself [28], the size and complexity of the DWI [36], characteristics of the plumbing system like different pipe diameters (causing different surface to volume ratios [37] and other conditions specific to the DWI (including temperatures [38,39]. Differences in the extent of regrowth within the same building, on the other hand, indicate differences in usage behaviours. In building 3, samples from different peripheral faucets contained different cell numbers at different time points.
Usage behaviour is closely linked to the size of UF units. Over-dimensioned units whose filtration capacity and membrane surface area are not in proportion to water consumption (as is the case in building 4) can lead to increased microbial regrowth and therefore high cell numbers. This is supported by the observed differences in bacterial regrowth between buildings 1 and 4 (Figure 4). Both buildings significantly varied in size (90 living units in building 1 vs. 30 in building 4), but were equipped with the identical UF unit with a filtration capacity of 9.7 m3/h. While the total annual PWC consumption in building 1 exceeded 3500 m3, consumption in building 4 was more than four times lower with approx. 800 m3. Due to the large dimension of the UF system in regard to the water usage and the lack of continuous daily operation during holiday periods, cell numbers in building 4 increased significantly during those times (Figure 2B). In contrast, cell numbers after UF in building 1 remained consistently low with minimal variation over time, owing to its permanent operation and the consistently high water consumption in this building (Figure 4).
Our findings furthermore corroborate that wash frequencies of UF membranes and CIP events have an effect on actual cell numbers in the water. Shorter backwash intervals have been reported to be more effective in mitigating fouling although at increased operational costs [40]. The accumulation of biomass on the feed side of the filter membranes probably also causes the release of dissolved intracellular nutrients from cells that disintegrate over time. The effect of such ‘nutrient leaching’ across membranes by disintegrating biomass is not well understood, but studies support that the turnover of microorganisms by death (often caused by viral lysis and protozoan grazing) and growth of new microorganisms and the resulting microbial population dynamics is typically greatly underestimated [41,42,43,44]. Overall, our findings corroborate the importance of the selection of a suitable UF-system dimension with appropriate operational settings to minimize cell numbers in the filtrate. This is not limited to drinking water, but also applies to other areas of use [45,46,47].
Whereas ultrafiltration of water was shown to have only a temporary effect on cell numbers, it profoundly changed the composition of bacterial populations (Figure 6). This finding supports its effectiveness as a microbial barrier and agrees with reports that regrowth can lead to the establishment of a different microbiome including the risk of hygienically relevant microorganisms [48]. Differences between the microbiomes in the feed water and the filtrate are not thought to be caused by selective membrane rejection, but by the regrowth of a different bacterial community in the filtrate due to different physical, chemical and biological conditions in the post-UF pipes. Especially the “last meters before the tap” were considered to be most critical for water quality [49]. Distal ends of building water supplies both harbour the highest cell counts and show most deviations from the municipal water supply microbiome [31]. The reason could lie in the fact that regrowth is most prominent in the periphery of plumbing systems as it is the end of plumbing lines where water tends to stagnate longest and sees the highest temperatures allowing bacteria to multiply [50]. Strong regrowth means a greater number of replication cycles and therefore a higher chance that microbial communities evolve in a distinct direction. Ultrafiltration could be seen a means to reinforce this effect due to initially low cell numbers with the concomitant presence of dissolved nutrients. As ultrafiltration results in a strong log-decrease of cell numbers without a substantial effect on the regrowth potential, the resulting high regrowth factors might therefore enhance more dramatic changes of the microbiome over a short distance compared to ‘normal’ regrowth.
The findings underline that the water that was stripped of microorganisms cannot be considered biologically stable [36,51]. The barrier function of ultrafiltration leaves the ecological niches unoccupied. In the absence of disinfection residuals, the recolonization with new microorganisms and their subsequent regrowth is thus strongly dependent on the receiving pipes or storage compartments where new microorganisms are introduced. From a hygienic focus, it is important to assure after the ‘microbiological reset’ that the subsequent biological stabilization is based on hygienically safe microbes. The possibility to direct the establishment of a microbiome after the ‘reset’ can be both seen as a chance and a risk. The chance of ultrafiltration lies in the replacement of an undesired microbiome with a favourable one, whereas the risk lies in the establishment of unwanted microorganisms in the biologically unstable water. Biological stabilization of water after ultrafiltration should therefore be considered in critical applications including water reuse. In case of drinking water, biological stabilization can occur through a hygienically safe microbiome associated with the pipe surfaces, whereas in case of industrial applications, passage of UF-filtrates over biological filters might be an option to achieve controlled biological stabilization to minimize the risk of uncontrolled occupation of ecological niches.

5. Conclusions

Ultrafiltration efficiently removes bacteria from water but does not have a lasting effect on cell numbers. The reason is that the membranes are not a barrier for dissolved nutrients that support the regrowth of microbes that are present in the post-UF compartments. The seeding of the filtrate probably occurs through bacteria located on the surfaces of the compartments receiving the filtrate and on the filtrate side of the UF membranes. These bacteria continue to replicate in the bulk water giving rise to increasing cell concentrations over time. The extent of regrowth greatly depends on the residence time of the bulk water in the water systems. The actual cell numbers are further determined by the dimension of the UF unit and the frequency of the membrane wash cycles. Data indicate that an UF unit that is oversized in relation to the water consumption, in combination with long wash frequencies, can result in high cell numbers on the filtrate side. In extreme cases the cell numbers in the filtrate can exceed the ones on the feed side. Actual cell numbers could be reduced by a CIP. Neither the wash frequencies nor the CIP, however, seemed to influence the maximal regrowth potential. Despite the temporary effect of UF on cell numbers, it does induce a ‘microbiological reset‘ of the microbial community. The exchange of the microbiome is enhanced by the removal of microbial competition and the availability of free ecological niches concomitant with the availability of dissolved low MW nutrients that can permeate the UF membrane. As the filtrate is biologically instable, an important aspect in this context is to assure that the right microbiome establishes in the filtrate that meets the requirements of the specific application.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/separations12080213/s1, Figure S1: Taxonomic diversity of bacterial communities in cold water samples before (POE) and after UF treatment in a drinking water distribution system of a large dormitory building (left) and in whey condensates before and after UF treatment (right). Schematic presentation separating the pipes by a UF-system shown above the diagram. Relative abundances of bacterial orders based on 16S rRNA gene.

Author Contributions

Conceptualization, B.H.M., A.N., M.H. and B.B.; methodology, B.H.M., A.N., M.H. and B.B.; investigation, B.H.M., A.N., M.H. and B.B.; data curation, B.H.M. and A.N.; writing—original draft preparation, B.H.M., and A.N.; writing—review and editing, B.H.M., A.N., M.H. and B.B.; project administration, B.B., A.N. and M.H.; funding acquisition, A.N., M.H. and B.B. All authors have read and agreed to the published version of the manuscript.

Funding

Experiments in relation to the treatment of whey vapour condensates were performed within the EU research network B-WaterSmart (EU funding program Horizon 2020; grant agreement nr. 869171; sub-project “Living Lab East Frisia”). The regrowth aspects of drinking water were covered by the ULTRA-F project (ultrafiltration as an element to achieve energetic efficiency in drinking water hygiene) project financed by the German Ministry for Economic Affairs and Climate Action (BMWK, FKZ 03ET1617).

Data Availability Statement

Data is contained within the article or Supplementary Material.

Acknowledgments

Denise Windrich and Dietmar Pütz from IWW are gratefully acknowledged for sample preparation, flow cytometric measurements and other technical analyses. We would like to thank our B-WaterSmart project partners IWW Analytik und Service GmbH, Deutsches Milchkontor (DMK), EnviroChemie GmbH and OOWV (Oldenburgisch-Ostfriesischer Wasserverband). A special thanks goes to Barbara Zimmermann (IWW) for collaboration in the field of process engineering and Mark Pannekens (IWW) for scientific support. Our gratitude also goes to the ULTRA-F consortium (Projekt “ULTRA-F”), including the Institute of Energy Technology and Institute of Medical Microbiology and Virology at the Technische Universität Dresden, the DVGW-German Water Centre Dresden and the Institute for Hygiene and Public Health at the University Hospital Bonn.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Al Aani, S.; Mustafa, T.N.; Hilal, N. Ultrafiltration membranes for wastewater and water process engineering: A comprehensive statistical review over the past decade. J. Water Process Eng. 2020, 35, 101241. [Google Scholar] [CrossRef]
  2. Arnal, J.M.; Garcia-Fayos, B.; Verdu, G.; Lora, J. Ultrafiltration as an alternative membrane technology to obtain safe drinking water from surface water: 10 years of experience on the scope of the AQUAPOT project. Desalination 2009, 248, 34–41. [Google Scholar] [CrossRef]
  3. Devaisy, S.; Kandasamy, J.; Nguyen, T.V.; Ratnaweera, H.; Vigneswaran, S. Membranes in Water Reclamation: Treatment, Reuse and Concentrate Management. Membranes 2023, 13, 605. [Google Scholar] [CrossRef]
  4. Davey, J.; Schäfer, A.I. Ultrafiltration to Supply Drinking Water in International Development: A Review of Opportunities. In Appropriate Technologies for Environmental Protection in the Developing World: Selected Papers from ERTEP 2007, Ghana, Africa, 17–19 July 2007; Yanful, E.K., Ed.; Springer: Dordrecht, The Netherlands, 2009; pp. 151–168. [Google Scholar]
  5. Jacquet, N.; Wurtzer, S.; Darracq, G.; Wyart, Y.; Moulin, L.; Moulin, P. Effect of concentration on virus removal for ultrafiltration membrane in drinking water production. J. Membr. Sci. 2021, 634, 119417. [Google Scholar] [CrossRef]
  6. Arkhangelsky, E.; Gitis, V. Effect of transmembrane pressure on rejection of viruses by ultrafiltration membranes. Sep. Purif. Technol. 2008, 62, 619–628. [Google Scholar] [CrossRef]
  7. Nocker, A.; Kaspareit, Y.; Windrich, D.; and Bendinger, B. Reduction of bacterial regrowth potential by membrane filtration for different water types and different pore sizes. Environ. Technol. 2025, 46, 3492–3503. [Google Scholar] [CrossRef]
  8. Genuardi, M.D.; Wiegand, M.; Opel, O. Installation of an ultrafiltration plant in a multi-family house to reduce the hot-water temperature and energy demand: A case study in Germany. J. Build. Eng. 2023, 66, 105898. [Google Scholar] [CrossRef]
  9. Arnal, J.M.; García-Fayos, B.; Sancho, M. Application of ultrafiltration for drinking water production in decentralised systems: 20 years of the AQUAPOT project. Desalination Water Treat. 2018, 103, 296–306. [Google Scholar] [CrossRef]
  10. Whiley, H. Legionella Risk Management and Control in Potable Water Systems: Argument for the Abolishment of Routine Testing. Int. J. Environ. Res. Public Health 2017, 14, 12. [Google Scholar] [CrossRef] [PubMed]
  11. Schurer, R.; Schippers, J.C.; Kennedy, M.D.; Cornelissen, E.R.; Salinas-Rodriguez, S.G.; Hijnen, W.A.M.; van der Wal, A. Enhancing biological stability of disinfectant-free drinking water by reducing high molecular weight organic compounds with ultrafiltration posttreatment. Water Res. 2019, 164, 114927. [Google Scholar] [CrossRef] [PubMed]
  12. Nocker, A.; Zimmermann, B. Brüdenaufbereitung in der Milchwirtschaft aus mikrobiologischer Sicht. gwf-Wasser|Abwasser 2025, 4, 35–38. [Google Scholar]
  13. Gatza, E.; Hammes, F.; Prest, E. Assessing Water Quality with the BD Accuri™ C6 Flow Cytometer. White Paper. BD Biosciences. 2013. Available online: https://www.umces.edu/sites/default/files/accuri-wp-assessing-water-quality.pdf (accessed on 22 March 2022).
  14. Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. 2011, 17, 10. [Google Scholar] [CrossRef]
  15. Edgar, R.C.; Haas, B.J.; Clemente, J.C.; Quince, C.; Knight, R. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 2011, 27, 2194–2200. [Google Scholar] [CrossRef]
  16. Rognes, T.; Flouri, T.; Nichols, B.; Quince, C.; Mahé, F. VSEARCH: A versatile open source tool for metagenomics. PeerJ 2016, 4, e2584. [Google Scholar] [CrossRef] [PubMed]
  17. Eren, A.M.; Maignien, L.; Sul, W.J.; Murphy, L.G.; Grim, S.L.; Morrison, H.G.; Sogin, M.L. Oligotyping: Differentiating between closely related microbial taxa using 16S rRNA gene data. Methods Ecol. Evol. 2013, 4, 1111–1119. [Google Scholar] [CrossRef]
  18. Eren, A.M.; Morrison, H.G.; Lescault, P.J.; Reveillaud, J.; Vineis, J.H.; Sogin, M.L. Minimum entropy decomposition: Unsupervised oligotyping for sensitive partitioning of high-throughput marker gene sequences. ISME J. 2015, 9, 968–979. [Google Scholar] [CrossRef]
  19. Caporaso, J.G.; Kuczynski, J.; Stombaugh, J.; Bittinger, K.; Bushman, F.D.; Costello, E.K.; Fierer, N.; Peña, A.G.; Goodrich, J.K.; Gordon, J.I.; et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 2010, 7, 335–336. [Google Scholar] [CrossRef]
  20. Angly, F.E.; Dennis, P.G.; Skarshewski, A.; Vanwonterghem, I.; Hugenholtz, P.; Tyson, G.W. CopyRighter: A rapid tool for improving the accuracy of microbial community profiles through lineage-specific gene copy number correction. Microbiome 2014, 2, 11. [Google Scholar] [CrossRef]
  21. Shannon, C.E. A Mathematical Theory of Communication. Bell Syst. Tech. J. 1948, 27, 379–423. [Google Scholar] [CrossRef]
  22. Lovins, W.A.; Taylor, J.S.; Hong, S.K. Micro-Organism Rejection by Membrane Systems. Environ. Eng. Sci. 2002, 19, 453–465. [Google Scholar] [CrossRef]
  23. US-EPA. Environmental Protection Agency. Low Pressure Membrane Filtration for Pathogen Removal: Application, Implementation, and Regulatory Issues; National Service Center for Environmental Publications: Cincinnati, OH, USA, 2001. [Google Scholar]
  24. Wingender, J.; Flemming, H.-C. Biofilms in drinking water and their role as reservoir for pathogens. Int. J. Hyg. Environ. Health 2011, 214, 417–423. [Google Scholar] [CrossRef]
  25. Proctor, C.R.; Hammes, F. Drinking water microbiology—From measurement to management. Curr. Opin. Biotechnol. 2015, 33, 87–94. [Google Scholar] [CrossRef] [PubMed]
  26. Lautenschlager, K.; Boon, N.; Wang, Y.; Egli, T.; Hammes, F. Overnight stagnation of drinking water in household taps induces microbial growth and changes in community composition. Water Res. 2010, 44, 4868–4877. [Google Scholar] [CrossRef]
  27. Gillespie, S.; Lipphaus, P.; Green, J.; Parsons, S.; Weir, P.; Juskowiak, K.; Jefferson, B.; Jarvis, P.; Nocker, A. Assessing microbiological water quality in drinking water distribution systems with disinfectant residual using flow cytometry. Water Res. 2014, 65, 224–234. [Google Scholar] [CrossRef]
  28. Nocker, A.; Meyer, B.; Bendinger, B. Effects of Surface Water Treatment for Drinking Water Production, Distribution and Heating on Biological Stability. Appl. Sci. 2025, 15, 5834. [Google Scholar] [CrossRef]
  29. Bucheli-Witschel, M.; Kötzsch, S.; Darr, S.; Widler, R.; Egli, T. A new method to assess the influence of migration from polymeric materials on the biostability of drinking water. Water Res. 2012, 46, 4246–4260. [Google Scholar] [CrossRef]
  30. Ramseier, M.K.; Peter, A.; Traber, J.; von Gunten, U. Formation of assimilable organic carbon during oxidation of natural waters with ozone, chlorine dioxide, chlorine, permanganate, and ferrate. Water Res. 2011, 45, 2002–2010. [Google Scholar] [CrossRef] [PubMed]
  31. Ling, F.; Whitaker, R.; LeChevallier, M.W.; Liu, W.-T. Drinking water microbiome assembly induced by water stagnation. ISME J. 2018, 12, 1520–1531. [Google Scholar] [CrossRef]
  32. Grimard-Conea, M.; Bédard, E.; Prévost, M. Can free chlorine residuals entering building plumbing systems really be maintained to prevent microbial growth? Sci. Total Environ. 2024, 939, 173651. [Google Scholar] [CrossRef] [PubMed]
  33. Mamani Condori, M.A.; Montesinos Pachapuma, K.A.; Gomez Chana, M.P.; Quispe Huillca, O.; Veliz Llayqui, N.E.; López-Rosales, L.; García-Camacho, F. An Environmentally Sustainable Approach for Raw Whey Treatment through Sequential Cultivation of Macrophytes and Microalgae. Appl. Sci. 2024, 14, 8139. [Google Scholar] [CrossRef]
  34. Zandona, E.; Blažić, M.; Režek Jambrak, A. Whey Utilization: Sustainable Uses and Environmental Approach. Food Technol. Biotechnol. 2021, 59, 147–161. [Google Scholar] [CrossRef]
  35. Peters, C.D.; Rantissi, T.; Gitis, V.; Hankins, N.P. Retention of natural organic matter by ultrafiltration and the mitigation of membrane fouling through pre-treatment, membrane enhancement, and cleaning—A review. J. Water Process Eng. 2021, 44, 102374. [Google Scholar] [CrossRef]
  36. Prest, E.I.; Hammes, F.; van Loosdrecht, M.C.M.; Vrouwenvelder, J.S. Biological Stability of Drinking Water: Controlling Factors, Methods, and Challenges. Front. Microbiol. 2016, 7, 45. [Google Scholar] [CrossRef]
  37. Tsvetanova, Z.; Hoekstra, E. Impact of Surface to Water Volume Contact Ratio on the Biomass Production Potential of Products in Contact with Drinking Water; JRC49181; OPOCE: Lisbon, Portugal, 2008. [Google Scholar]
  38. Proctor, C.R.; Rhoads, W.J.; Keane, T.; Salehi, M.; Hamilton, K.; Pieper, K.J.; Cwiertny, D.M.; Prévost, M.; Whelton, A.J. Considerations for large building water quality after extended stagnation. AWWA Water Sci. 2020, 2, e1186. [Google Scholar] [CrossRef]
  39. Aloraini, S.; Alum, A.; Abbaszadegan, M. Impact of Pipe Material and Temperature on Drinking Water Microbiome and Prevalence of Legionella, Mycobacterium, and Pseudomonas Species. Microorganisms 2023, 11, 352. [Google Scholar] [CrossRef]
  40. Alhussaini, M.A.; Binger, Z.M.; Souza-Chaves, B.M.; Amusat, O.O.; Park, J.; Bartholomew, T.V.; Gunter, D.; Achilli, A. Analysis of backwash settings to maximize net water production in an engineering-scale ultrafiltration system for water reuse. J. Water Process Eng. 2023, 53, 103761. [Google Scholar] [CrossRef]
  41. Pherribo, G.J.; Taga, M.E. Bacteriophage-mediated lysis supports robust growth of amino acid auxotrophs. ISME J. 2023, 17, 1785–1788. [Google Scholar] [CrossRef]
  42. Wu, H.; Wan, S.; Ruan, C.; Niu, X.; Chen, G.; Liu, Y.; Zhu, K.; Schulin, R.; Wang, G. Phage-bacterium interactions and nutrient availability can shape C and N retention in microbial biomass. Eur. J. Soil Sci. 2022, 73, e13296. [Google Scholar] [CrossRef]
  43. Kuikman, P.J.; Jansen, A.G.; van Veen, J.A.; Zehnder, A.J.B. Protozoan predation and the turnover of soil organic carbon and nitrogen in the presence of plants. Biol. Fertil. Soils 1990, 10, 22–28. [Google Scholar] [CrossRef]
  44. Finlay, B.J.; Esteban, G.F.; Clarke, K.J.; Olmo, J.L. Biodiversity of Terrestrial Protozoa Appears Homogeneous across Local and Global Spatial Scales. Protist 2001, 152, 355–366. [Google Scholar] [CrossRef]
  45. Vincent-Vela, M.-C.; Cuartas-Uribe, B.; Álvarez-Blanco, S.; Lora-García, J. Analysis of an ultrafiltration model: Influence of operational conditions. Desalination 2012, 284, 14–21. [Google Scholar] [CrossRef]
  46. Panteleev, A.A.; Panteleev, A.A.; Bobinkin, V.V.; Lysenko, S.E.; Smirnov, V.B. Optimization of Operating Conditions for Ultrafiltration Units. Therm. Eng. 2021, 68, 777–784. [Google Scholar] [CrossRef]
  47. Pervov, A.G.; Tikhonov, K.V.; Makisha, N.A. Determination of Optimal Operation Pressure Values for Ultrafiltration Wastewater Treatment. Membr. Membr. Technol. 2020, 2, 159–168. [Google Scholar] [CrossRef]
  48. Rahmatika, I.; Kurisu, F.; Furumai, H.; Kasuga, I. Dynamics of the Microbial Community and Opportunistic Pathogens after Water Stagnation in the Premise Plumbing of a Building. Microbes Environ. 2022, 37, ME21065. [Google Scholar] [CrossRef] [PubMed]
  49. Flemming, H.-C.; Bendinger, B.; Exner, M.; Gebel, J.; Kistemann, T.; Schaule, D.G.; Szewzyk, U.; Wingender, J. The last meters before the tap: Where drinking water quality is at risk. In Microbial Growth in Drinking-Water Supplies Problems, Causes, Prevention and Research Needs; IWA Publishing: London, UK, 2014; pp. 207–238. [Google Scholar]
  50. Ji, P.; Parks, J.; Edwards, M.A.; Pruden, A. Impact of Water Chemistry, Pipe Material and Stagnation on the Building Plumbing Microbiome. PLoS ONE 2015, 10, e0141087. [Google Scholar] [CrossRef]
  51. Proksch, P.; Kerschbaumer, D.J.; Schönher, C.; Zunabovic-Pichler, M.; Mayr, E.; Perfler, R. Pilotierung einer Ultrafiltrationsanlage an einem Uferfiltratstandort in Österreich—Untersuchungen der bakteriellen Zellzahlen und der biologischen Stabilität im Filtrat. Osterr. Wasser- Und Abfallwirtsch. 2020, 72, 484–489. [Google Scholar] [CrossRef]
Separations 12 00213 g001
Figure 1. Effect of ultrafiltration on total cell numbers directly after filtration (day 0, black) and regrowth potential (day 7, grey). The upper limit of the grey boxes indicates the TCC obtained after 7 days at 22 °C. Drinking water before and after ultrafiltration in three different DWIs (A) and biologically treated whey condensates before and after ultrafiltration as part of a water treatment train in a reuse plant (B). Unfiltered water at the POE or the unfiltered, biologically treated whey vapour condensate (UF feed) served as references. In the case of water reuse, the data from one exemplary sampling event is shown. Error bars indicate the standard deviations of three independent measurements.
Figure 1. Effect of ultrafiltration on total cell numbers directly after filtration (day 0, black) and regrowth potential (day 7, grey). The upper limit of the grey boxes indicates the TCC obtained after 7 days at 22 °C. Drinking water before and after ultrafiltration in three different DWIs (A) and biologically treated whey condensates before and after ultrafiltration as part of a water treatment train in a reuse plant (B). Unfiltered water at the POE or the unfiltered, biologically treated whey vapour condensate (UF feed) served as references. In the case of water reuse, the data from one exemplary sampling event is shown. Error bars indicate the standard deviations of three independent measurements.
Separations 12 00213 g001
Separations 12 00213 g002
Figure 2. Total cell counts in cold drinking water at the POE (before UF), after ultrafiltration (after UF) and at the entry of the drinking water heater (entry DWH) in two different DWIs equipped with a small UF with an 4 h membrane wash frequency in building 3 (A) or a large UF in building 4 with a 24 h membrane wash frequency (B). Samples were taken at 13 different time points (Sampling (S)1 to 13). Over a period of 12 to 14 months, green shaded areas indicate periods with low water consumption due to holiday breaks. Blue dashed lines indicate the start of ultrafiltration. Numbers above the grey arrows show the TCC reductions for the different sampling points compared to the TCC in the corresponding POE sample at the given time point.
Figure 2. Total cell counts in cold drinking water at the POE (before UF), after ultrafiltration (after UF) and at the entry of the drinking water heater (entry DWH) in two different DWIs equipped with a small UF with an 4 h membrane wash frequency in building 3 (A) or a large UF in building 4 with a 24 h membrane wash frequency (B). Samples were taken at 13 different time points (Sampling (S)1 to 13). Over a period of 12 to 14 months, green shaded areas indicate periods with low water consumption due to holiday breaks. Blue dashed lines indicate the start of ultrafiltration. Numbers above the grey arrows show the TCC reductions for the different sampling points compared to the TCC in the corresponding POE sample at the given time point.
Separations 12 00213 g002
Separations 12 00213 g003
Figure 3. Total cell counts at the time point of sampling (day 0) and the regrowth potentials (day 7) of drinking water before and after ultrafiltration in building 3 with a backwash frequency of 4 h (A) and building 4 with a forward flush frequency of 24 h (B). aUF = after ultrafiltration. The blue boxes below the bars indicate unfiltered water, while the shaded boxes indicate filtered water. Samples were taken at different time points. Error bars indicate the standard deviations of three independent samples. A schematic depiction of the DWI, including the POE (blue triangle), the sampling point located after the UF unit (shaded box) and before entry into the DWH (blue box), is shown above the diagram.
Figure 3. Total cell counts at the time point of sampling (day 0) and the regrowth potentials (day 7) of drinking water before and after ultrafiltration in building 3 with a backwash frequency of 4 h (A) and building 4 with a forward flush frequency of 24 h (B). aUF = after ultrafiltration. The blue boxes below the bars indicate unfiltered water, while the shaded boxes indicate filtered water. Samples were taken at different time points. Error bars indicate the standard deviations of three independent samples. A schematic depiction of the DWI, including the POE (blue triangle), the sampling point located after the UF unit (shaded box) and before entry into the DWH (blue box), is shown above the diagram.
Separations 12 00213 g003
Separations 12 00213 g004
Figure 4. Effect of ultrafiltration on TCC values in buildings 1–4 directly after UF and in the periphery. The TCCday 0 values of the drinking water entering the buildings at the POE are defined as 100% and serve as a reference for cell counts directly after UF (aUF) and at the peripheral points of use (POU). The POU sampling sites comprise sink faucets in the bathrooms of the different buildings. Values represent averages over the total monitoring time of each building. Error bars indicate standard deviations between independent samples and the number of samples for building 1: POE and aUF: 12, POU: 61; building 2: POE and aUF: 20, POU: 32, building 3: POE und aUF: 10, POU: 47; and building 4: POE and aUF: 9, POU: 37.
Figure 4. Effect of ultrafiltration on TCC values in buildings 1–4 directly after UF and in the periphery. The TCCday 0 values of the drinking water entering the buildings at the POE are defined as 100% and serve as a reference for cell counts directly after UF (aUF) and at the peripheral points of use (POU). The POU sampling sites comprise sink faucets in the bathrooms of the different buildings. Values represent averages over the total monitoring time of each building. Error bars indicate standard deviations between independent samples and the number of samples for building 1: POE and aUF: 12, POU: 61; building 2: POE and aUF: 20, POU: 32, building 3: POE und aUF: 10, POU: 47; and building 4: POE and aUF: 9, POU: 37.
Separations 12 00213 g004
Separations 12 00213 g005
Figure 5. Total cell counts per ml at different PWC sampling sites within the DWI of building 3 over a period of 14 months. The first sampling time point was prior to installation of the UF. The start of the ultrafiltration after the POE is indicated by a dashed line (17 August 2022). TCCday 0 values are shown for the following sampling sites: POE (blue triangle), directly after UF (crossed square), the entry of the drinking water heater (blue square) and three peripheral POU sites (light to dark purple circles). The POU sites were sink faucets in the bathrooms (PWC-POE-1 to 03).
Figure 5. Total cell counts per ml at different PWC sampling sites within the DWI of building 3 over a period of 14 months. The first sampling time point was prior to installation of the UF. The start of the ultrafiltration after the POE is indicated by a dashed line (17 August 2022). TCCday 0 values are shown for the following sampling sites: POE (blue triangle), directly after UF (crossed square), the entry of the drinking water heater (blue square) and three peripheral POU sites (light to dark purple circles). The POU sites were sink faucets in the bathrooms (PWC-POE-1 to 03).
Separations 12 00213 g005
Separations 12 00213 g006
Figure 6. Taxonomic diversity of bacterial families in cold water samples before and after UF treatment in the drinking water distribution system of the large dormitory building 4 (A) and for biologically treated whey vapour condensates (B). Relative abundances of bacterial families are based on 16S rRNA genes. The most abundant families (top ten) of each sample are listed with the relative abundances on the right. For easy comparison, the samples before UF are directly compared with samples after UF and vice versa.
Figure 6. Taxonomic diversity of bacterial families in cold water samples before and after UF treatment in the drinking water distribution system of the large dormitory building 4 (A) and for biologically treated whey vapour condensates (B). Relative abundances of bacterial families are based on 16S rRNA genes. The most abundant families (top ten) of each sample are listed with the relative abundances on the right. For easy comparison, the samples before UF are directly compared with samples after UF and vice versa.
Separations 12 00213 g006
Table 1. Overview of the sampling sites, locations and specifications. MFH = multi-family house. Risers = rising main pipes in plumbing system. n.a. = not applicable.
Table 1. Overview of the sampling sites, locations and specifications. MFH = multi-family house. Risers = rising main pipes in plumbing system. n.a. = not applicable.
Sampling SiteDrinking Water InstallationReuse Plant
12345
LocationNeuruppinKielHamburgMunichEdewecht
Building-TypeMFHMFHMFHDormitorypilot plant
living units90101630n.a.
UF locationPOEPOEPOEPOEafter biolog. treatment
Membrane materialPolyethersulfonPolyethersulfonPolyethersulfonPolyethersulfonPVDF
Membrane pore size20 nm20 nm20 nm20 nm10 nm
Type of filtrationdead enddead endcrossflowdead enddead end
Filtration capacity9.7 m3/h3 to 20 m3/h5 m3/h9.7 m3/hn.a.
Number of risers18289n.a.
Times of sampling (TCCday 0)131913134
Times of sampling (TCCday 7)44334
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Meyer, B.H.; Bendinger, B.; Hippelein, M.; Nocker, A. Ultrafiltration of Water Has a Temporary Effect on Cell Numbers, but Profoundly Changes the Composition of Bacterial Populations—The ‘Reset’ Phenomenon. Separations 2025, 12, 213. https://doi.org/10.3390/separations12080213

AMA Style

Meyer BH, Bendinger B, Hippelein M, Nocker A. Ultrafiltration of Water Has a Temporary Effect on Cell Numbers, but Profoundly Changes the Composition of Bacterial Populations—The ‘Reset’ Phenomenon. Separations. 2025; 12(8):213. https://doi.org/10.3390/separations12080213

Chicago/Turabian Style

Meyer, Benjamin H., Bernd Bendinger, Martin Hippelein, and Andreas Nocker. 2025. "Ultrafiltration of Water Has a Temporary Effect on Cell Numbers, but Profoundly Changes the Composition of Bacterial Populations—The ‘Reset’ Phenomenon" Separations 12, no. 8: 213. https://doi.org/10.3390/separations12080213

APA Style

Meyer, B. H., Bendinger, B., Hippelein, M., & Nocker, A. (2025). Ultrafiltration of Water Has a Temporary Effect on Cell Numbers, but Profoundly Changes the Composition of Bacterial Populations—The ‘Reset’ Phenomenon. Separations, 12(8), 213. https://doi.org/10.3390/separations12080213

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop