Bioaugmentation of a structural extracellular polymeric substances (EPS) producer to improve activated sludge bioflocculation: lessons learned

Bioaugmentation or the addition of indigenous or allochthonous wild-type or genetically modified organisms is used to solve practical problems in several biotechnological domains. This technique is often aimed at improving the catabolism of specific compounds in wastewater treatment plants. To a lesser extent, it has been exploited to enhance the bioflocculation performance of microorganisms. An evident example is the seeding of granules or biofilms, as summarized in a review within the context of accelerating aerobic granule formation or improving granule stability (Kent et al. 2018). A more targeted approach is described by Xin et al. (2017), who augmented a Rhodocyclaceae-related Acinetobacter sp. named TN-14 with heterotrophic nitrification and aerobic denitrification function, which promoted aerobic granule development. This strain promoted the EPS production of sludge, by increasing the hydrophobicity. Similarly, the harvesting efficiency of microalgae has been improved by mixing them with cultures of bacteria, fungi, or flocculating strains of microalgae (Demir et al. 2020) producing extracellular polymers. The latter review also describes cases where the extracellular polymers are extracted and added as bioflocculants to the microalgae culture. In a similar context, Sarang & Nerurkar (2020) use a bioflocculant extracted from Bacillus cereus CR4 for application in microalgae harvesting. Congo red and Thioflavin T (ThT) staining of the bacterial culture and other assays, including transmission electron microscopy and Fourier-transform infrared spectroscopy, on the purified extract identified the bioflocculant as an amyloid adhesin. Amyloid adhesins are highly stable proteinaceous molecules that are part of the structural EPS and are previously investigated by our research group (Christiaens et al. 2022) and other groups.

These studies suggest a potential for augmentation of pure cultures, specifically structural EPS producers, as targeted bioflocculation remediation approaches within the field of wastewater treatment.

This potential will be investigated in this work using Azoarcus, which was identified as a promising candidate for this augmentation experiment based on the following characteristics. Azoarcus are chemoorganoheterotrophs that can use O₂ or as an electron acceptor. Azoarcus species are believed to be one of the dominant denitrifiers in intermittently aerated nitrifying and denitrifying sludge from plants treating industrial and coking wastewater (Juretschko et al. 2002; Ma et al. 2015) as well as in communal wastewater treatment plants with biological nitrogen and phosphorus removal (Thomsen et al. 2007). Recent data from 368 plants with relevant process types (carbon and nitrogen removal with or without enhanced biological phosphorus removal) reports a global maximum of 1.4% relative abundance (Dueholm et al. 2022). Azoarcus belong to the phylogenetic group of Betaproteobacteria that are known to be strong microcolony formers with high resistance to shear stress (Klausen et al. 2004). Azoarcus belong to the family of Rhodocyclaceae in which many EPS-producing denitrifiers are classified (Rosenberg et al. 2014). More specifically, an extracellular polysaccharide biosynthesis gene cluster similar to the one in Zoogloea and linked to floc formation in Azoarcus has been identified (An et al. 2016). Moreover, Azoarcus has been identified as a potential amyloid producer based on sequential ThT staining and fluorescence in situ hybridization, but these results were not confirmed with conformational antibodies (Larsen et al. 2008).

This study aims to further explore the impact of structural EPS producers on bioflocculation in general and on bioaugmentation of Azoarcus communis in particular. Three research hypotheses were formulated: (i) A. communis is a bioflocculation-stimulating activated sludge organism, (ii) more amyloid-like substances can be introduced to activated sludge via augmentation of A. communis, and (iii) bioflocculation of activated sludge can be improved via augmentation of A. communis. In order to enable targeted monitoring of the survival, persistence, and incorporation of the augmented structural EPS producer, this study constructed an antibiotic-resistant and fluorescent mutant strain.

A freeze-dried culture of A. communis Swub 3 (DSM12120) was acquired from the Deutsche Sammlung von Mikroorganismen (DSMZ) databank. An Escherichia coli K-12 curli-deficient mutant (SM2257) (Prigent-Combaret et al. 2001) and a mutant with upregulated curli production (SM5578) (Vidal et al. 1998) were used as model organisms. All strains were cultured aerobically at 28 °C. A. communis was grown overnight after inoculation at an initial optical density (OD₆₀₀) of 0.05 in shake flasks in tryptic soy broth (TSB) consisting of 17 g L⁻¹ pancreatic digest of casein, 3 g L⁻¹ papaic digest of soybean 3 g L⁻¹, 2.5 g L⁻¹ glucose, 5 g L⁻¹ NaCl, 2.5 g L⁻¹ K₂HPO₄. E. coli SM2257 and SM2258 were grown statically for 48 h after inoculation at an initial OD₆₀₀ of 0.05 in the M63 medium (Larsen et al. 2007).

Rifampicin powder (Sigma–Aldrich) was dissolved in DMSO and filter sterilized through a 0.2-μm syringe filter. The concentration of the stock solution was determined spectrophotometrically by absorbance measurement at 475 nm. From this absorbance, the concentration was calculated via the Lambert–Beer law with the molar extinction coefficient equal to 15,400 M⁻¹ cm⁻¹ for rifampicin in phosphate buffer (Florey 1976). A gentamicin sulfate sterile filtered 50 mg mL⁻¹ solution in water was purchased from Merck Life Science.

Rifampicin-resistant (Rif^r) mutants were constructed by inoculation of the A. communis bacterial stock onto tryptic soy agar (TSA) plates supplemented with 50 mg L⁻¹ rifampicin. Spontaneous rifampicin-resistant mutants were replated three times to TSA containing 100 mg L⁻¹ rifampicin. Finally, colonies from these plates were grown overnight at 28 °C in TSB supplemented with 100 mg L⁻¹ rifampicin, harvested by centrifugation at 3,220 g for 10 min, and stored in 20% w/v glycerol at −80 °C in sterile cryovials until further use.

Fluorescent-labeled bacteria were constructed using the mini-Tn5 random transposon insertion system as described first by De Lorenzo et al. (1990). Briefly, a minitransposon delivers the plasmid, containing a fluorescent protein gene and an antibiotic resistance gene as a selection marker, from a donor strain to a host strain. The E. coli S17-I λpir::pMRE-Tn5-145 donor strain, conferring gentamicin resistance and mScarlet-I encoding for red-fluorescent proteins, was grown for 6 h at 37 °C until the exponential growth phase in lysogeny broth (LB) consisting of 10 g L⁻¹ tryptone, 5 g L⁻¹ yeast extract, and 10 g L⁻¹ NaCl, supplemented with 20 mg L⁻¹ gentamicin. The A. communis Rif^r host strain was grown overnight at 28 °C until the early stationary phase in TSB supplemented with 100 mg L⁻¹ rifampicin. Cells were harvested by centrifuging 5 mL of culture for 10 min at 3,220 g, washed three times using sterile 0.9% NaCl to remove all antibiotics, and resuspended in 500 μL 0.9% NaCl to obtain a dense suspension. Bi-parental mating was performed by spotting 20 μL of a 1:1 donor:host mix on a TSA plate. After 24 h of incubation at 28 °C, cells were scraped off, suspended in 0.9% NaCl, and plated on TSA supplemented with 100 mg L⁻¹ rifampicin and 20 mg L⁻¹ gentamicin to select for transconjugants. Fluorescent mutants were identified using a Safe Imager™ 2.0 Blue-Light Transilluminator, grown in TSB supplemented with 100 mg L⁻¹ rifampicin and 20 mg L⁻¹ gentamicin, harvested, and stored in 20% w/v glycerol at −80 °C in sterile cryovials until further use. The A. communis mutant carrying the mScarlet-I gene is hereafter referred to as A. communis Rif^r-mSc.

The identity of the fluorescent mutant strain was checked via Sanger sequencing. Cells were picked from the agar plate and suspended in UV-sterilized nuclease-free water and lysed at 95 °C for 5 min (Horecka & Chu 2017). The partial 16S rRNA genes were amplified using the universal primer pair 27F (5′-AGA–GTT–TGA–TCM–TGG–CTC–AG-3′) and 1492R (5′-GGT–TAC–CTT–GTT–ACG–ACT–T-3′) (Frank et al. 2008) using the PCR conditions described in Supplementary material S1. The amplicon was purified from primers, primer dimers, and deoxynucleotide triphosphates (dNTPs) using the QIAquick PCR Purification kit (QIAGEN^®) according to the manufacturer's instructions. Sanger sequencing was performed by GATC Biotech AG, Germany. The resulting partial 16S rRNA gene sequences were aligned with NCBIs refseq_rna databased for Reference RNA sequences using the MegaBLAST program optimized for highly similar sequences in BLASTN version 2.13.0+ (Altschul et al. 1990; Zhang et al. 2000).

To check whether the inserted mutations affected growth, the growth rate of the wild-type and mutant strains was determined. Replicate cultures were inoculated at an initial OD₆₀₀ of 0.05, and their OD₆₀₀ was measured at specific time intervals as a proxy for cell density. Based on these data, the specific growth rate μ was estimated by multi-phase linear regression using MATLAB^® R2022b (MathWorks^®) as detailed in Supplementary material S2.

To check the stability of the insertion of the antibiotic and fluorescent markers, the mutant was grown overnight at 28 °C and 150 rpm in TSB containing 100 mg L⁻¹ rifampicin and 20 mg L⁻¹ gentamicin. The culture was washed 3 times in 0.9% NaCl to remove all antibiotics prior to dilution to an OD₆₀₀ of 0.05. A subculture was inoculated by transferring 50 μL of this culture into 25 mL fresh TSB without antibiotics and grown until the late exponential phase at 28 °C and 150 rpm for around 24 h. This procedure was followed by 10 successive similar transfers. Finally, the colony-forming units CFU method was used to quantify the stability of the antibiotic markers: a series of tenfold dilutions was prepared in a 96-well plate. Six 5 μL replicates were spotted on rectangular TSA plates with and without the antibiotics rifampicin and gentamicin, and the antifungal additive nystatin (Thermo Scientific™). A 5 mg mL⁻¹ stock solution of nystatin in DMSO was prepared. Plates were incubated for a minimum of 48 h at 28 °C before counting. Experimental data were analyzed for statistical significance using a Student's t-test using MS Excel. Additionally, 20 colonies were picked and dissolved in PBS, and their fluorescence was assessed under the Olympus IX83 fluorescence microscope.

Amyloid-like substances were detected and the ThT emission signal was quantified as described before (Christiaens et al. 2022). Prior to ThT staining, cultures were grown in TSB were washed twice using PBS to remove the autofluorescent growth medium.

The number of augmented bacteria was monitored by counting their CFU on selective media as a proxy for the number of viable mutants in the reactor. First, sludge samples were deflocculated using a manual tissue grinder (VWR) or ultrasonication. Then, CFU enumeration was performed using TSA plates with antibiotics and 50 μg mL⁻¹ of the antifungal additive nystatin. In addition, the location and morphology of the augmented bacteria were visualized by phase contrast and fluorescence microscopy using an Olympus IX83 inverted microscope equipped with a U-FGWA and U-FYW mirror unit. All recipients containing the fluorescent mutant, including the Erlenmeyers and reactor vessel, were shielded from light to avoid bleaching.

The identity of some bacteria in sludge samples was investigated using FISH as described by Amann (1995) using various oligonucleotide probes targeting Azoarcus, Zoogloea, and most bacteria (Supplementary material S3). The fluorescence signal was visualized on an Olympus FluoView™ FV1000 confocal microscope.

To enable in-depth monitoring of the augmented strain in the activated sludge with respect to survival and persistence and initial attachment and integration, an antibiotic-resistant and fluorescent mutant strain was constructed.

Rifampicin-resistant mutants developed after 6 days on TSA plates containing 50 mg L⁻¹ rifampicin. One mutant was selected as the host strain for fluorescent protein gene insertion. After bi-parental mating of this host and the E. coli donor, growth of transconjugants was observed within 4–7 days on LB and TSA plates containing 20 mg L⁻¹ gentamycin and 100 mg L⁻¹ rifampicin. Roughly 50% of these transconjugants effectively expressed the mScarlet-I fluorescent protein. Alignment revealed that the Rif^r-mSc strain showed 99.63% sequence similarity with A. communis Swub3. Therefore, this mutant was selected for future experiments.

Finally, the stability of the introduced antibiotic and fluorescent markers was examined to assess if the bioaugmented A. communis could be correctly monitored during the long-term reactor experiment.

A total of 11 consecutive transfers, approximately corresponding to 131 generations, were grown in a medium not containing antibiotics. The final culture was plated on TSA and various selective media. The number of colony-forming units per liter on TSA supplemented with 50 mg L⁻¹ rifampicin (5.20E + 11 ± 5.77E + 10) or 20 mg L⁻¹ gentamycin (5.83E + 11 ± 5.34E + 10) was equal to the number on TSA (5.57E + 11 ± 6.05E + 10) (p = 0.11106708 and p = 0.476912649, respectively, both > 0.05). Less growth, but not significantly less, was observed on TSA supplemented with 100 mg L⁻¹ rifampicin (4.87E + 11 ± 4.42E + 10) or 100 mg L⁻¹ rifampicin and 20 mg L⁻¹ gentamicin (4.70E + 11 ± 7.00E + 10) (p = 0.063173788 and p = 0.075087307, respectively, both > 0.05). These results suggest that all antibiotic resistance markers were stably inherited by later generations, but too high concentrations of rifampicin inhibited complete recovery. Moreover, slower growth was observed in plates containing 100 mg L⁻¹ rifampicin. Additionally, the impact of the antifungal additive nystatin was assessed indicating no effect on recovery. The number of colony-forming units on TSA supplemented with 50 mg L⁻¹ nystatin (5.10E + 11 ± 2.52E + 10) was equal to the number on TSA (p = 0.155103406 > 0.05).

The stable inheritance of the mScarlet-I gene after 131 generations was assessed by fluorescence microscopy on 20 colonies, randomly picked from TSA plates, and TSA plates supplemented with 20 mg L⁻¹ gentamicin or 100 mg L⁻¹ rifampicin and 20 mg L⁻¹ gentamicin. All colonies appeared fluorescent using the U-FGWA and U-FYW filter set indicating the persistent presence of the mScarlet-I fluorescent protein gene.

At the end of the 40-day startup period, communities were mixed once to ensure equal conditions for both bioreactors. One MBR was augmented with A. communis Rif^r-mSc and another served as a nonaugmented reference reactor. Two series of augmentation events took place: by day 14, a total of 1.20 g A. communis had been added to the reactor followed by an additional 2.15 g by day 31. To compare, the average total sludge content in the augmented reactor during the augmentation phase was 13.38 g. Note that all following results have been represented at ‘days since the start of augmentation’.

This experiment aims to assess the impact of structural EPS on activated sludge flocculation by actively adding a structural EPS producer. Traditionally, inorganic and synthetic flocculants are used for bioflocculation remediation. However, these flocculants are associated with a lack of biodegradability and with health hazards (Lee et al. 2014). Bioflocculants, i.e., natural polymers extracted from bacteria or other sources, are being investigated as an alternative but their commercialization is hampered due to their higher production cost (Liu et al. 2010). Moreover, (bio)flocculants add to the waste sludge volume and have to be reapplied regularly. If the abundance of flocculants in the activated sludge system can be increased by adding (or stimulating) a bioflocculant producer, these disadvantages are avoided. In fact, the bioflocculant producer can play an active role in the biodegradation of pollutants and maintain its presence in the sludge through growth.

A case study is presented involving A. communis, a denitrifier with amyloid-like substances production and flocculation potential. In view of monitoring the augmented strain, the introduction of biological markers and their stability are discussed. Since these markers indicated that only a low fraction of augmented bacteria was sustained in the system after augmentation, adaptations to the approach are suggested.

ThT staining suggested that A. communis was capable of producing amyloids when grown in monoculture in TSB or in synthetic wastewater. To our knowledge, this research provides the first evidence of Thioflavinophilic EPS production of a monoculture of a specific species of the genus Azoarcus. Additional research using other analytical methods (e.g., FT-IR) should be considered to (i) check the specificity of the ThT dye, which is prone to unspecific binding to, e.g., cellulose fibers (Retna Raj & Ramaraj 2001), and (ii) identify EPS-components besides amyloids. Such components are expected to be present based on visual observation of a prominent slimy or gel-like fraction when A. communis is harvested during the exponential growth phase.

The short-term impact on flocculation was assessed against two types of samples and using either A. communis or one of two reference strains: E. coli with upregulated curli production and a curli-deficient E. coli mutant. These three strains are referred to as ‘bioflocculant producers’ in this paragraph. In these experiments, the bioflocculants are added together with their producing strain, without prior extraction and purification steps. Removal of growth medium, washing, and resuspension in PBS are the only pretreatment steps. This approach differs from much other work regarding bioflocculants but omits extra preparation time. The turbidity of the clarifying suspension of kaolin and bioflocculant producer was more strongly decreased in case of A. communis and E. coli with upregulated curli production compared to the E. coli curli-deficient mutant. This can be the result of several flocculating mechanisms dependent on the nature of the bioflocculant such as (i) bridging mediated by cations and charge neutralization in case of cation-dependent bioflocculants or (ii) direct attachment and bridging in case of cation-independent bioflocculants (Lian et al. 2008; Li et al. 2009; Liu et al. 2010). Lower flocculation activity was observed against deflocculated sludge but the trends remained similar. It should be noted that the interpretation of these results is more complex due to (i) potential sludge reflocculation, (ii) potential incomplete deflocculation providing less sludge surface area to bind to, and (iii) general heterogeneity of the sludge sample. Moreover, the ultrasonication of the sludge might result in partial extraction of its EPS. Therefore, the FAI measures the combined flocculation effect of sludge on A. communis and of A. communis on sludge. Overall, these experiments suggest that amyloids benefit the flocculation activity of a strain, making A. communis an interesting candidate for the augmentation experiment. As another potential application, A. communis might be utilized for short-term activated sludge recovery during a deflocculation event, yet this application was not investigated. Repetitions using different sludge samples and different amyloid-producing and -deficient strains should be considered for further research to evaluate the generalizability of these results.

Based on these results, it is recommended to consider the structural EPS-producing capacity and, more specifically, amyloid-producing capacity to evaluate candidate strains for bioaugmentation aiming to improve bioflocculation, among other factors. The proposed ThT staining procedure might be used in this context. This recommendation corresponds to the use of Congo Red as a reagent to select microorganisms for flocculant production as an alternative to synthetic polymers (Rebah et al. 2018).

The startup phase of MBR operation coincided with a quick increase of ThT fluorescence in the sludge compared to the inoculum. This observation suggests that the amyloid-like content has increased. The startup phase also coincides with an increased floc size. Thus, after 40 days of MBR operation, A. communis was introduced into well-flocculating and amyloid-like substances-rich sludge as a baseline.

Two series of augmentation events were performed and the presence of the augmented strain was monitored qualitatively via fluorescence microscopy and quantitatively via the enumeration of colony-forming units. After the second augmentation series, fluorescent clusters were observed, seemingly grown from single adsorbed cells, and at the end of reactor operation, the number of colony-forming units stabilized.

These two observations suggest that some of the augmented A. communis stably established themselves in the sludge and were actively growing. However, only a minor fraction of the augmented bacteria survived. Three potential explanations are presented here. Firstly, the presence of indigenous microorganisms that fill the same metabolic ecological niche might have led to competition for A. communis. 16S rRNA gene sequencing indeed revealed the presence of several denitrifiers, including Azoarcus, in a sample from the same origin as the reactor inoculum, and Zoogloea has been detected in the MBR using FISH. Moreover, Wilderer et al. (1991) have suggested that it may be more difficult for an inoculated bacterial strain to outcompete the indigenous bacteria if the latter are fully adapted to the environment. Furthermore, the mScarlet-I mutant has a significantly lower growth rate compared to the wild-type Azoarcus. It should be noted that A. communis, like other plant-associated Azoarcus species, grows well on salts of organic acids and aromatic substrates but cannot metabolize carbohydrates (Reinhold-Hurek et al. 1993). Glucose thus remains available in the synthetic wastewater to be consumed by, e.g., other denitrifiers. Secondly, since bacteria were augmented in suspension, and were observed to be adsorbed onto the outside of the flocs after 24 h, metazoan and protozoan grazing might have contributed to the decline of the augmented Azoarcus. In this case, mScarlet-I fluorescence was detected within several rotifers and to a lesser extent in stalked ciliates that are both able to consume whole bacteria (Cybis & Horan 1997). The effect of grazing should not be underestimated: the work by Bouchez et al. (2000) points toward grazing by ciliates and other protozoa as the main reason for the failure of bioaugmentation of an aerobic denitrifying bacterium augmented in a nitrifying acetate-fed SBR. Thirdly, several abiotic factors might have adversely impacted the survival of the augmented A. communis. While growth on synthetic wastewater in aerobic conditions was confirmed in preliminary experiments, these batch experiments did not account for the cyclic nutrient and pH variations in the reactor, the hydrodynamics caused by bubble aeration, or the operating temperature of around 24 °C instead of 28 °C.

ThT fluorescence intensity was relatively low in the regions where A. communis was embedded suggesting a lower amyloid-like content in these regions. With this assumption in mind, two possible explanations are proposed. Firstly, several abiotic factors might have hampered the amyloid-like substances production of A. communis in the reactor. As elaborated before, the preliminary experiments showing that amyloid-like substance production occurred in synthetic wastewater in aerobic and anoxic conditions did not perfectly mimic the reactor conditions. Secondly, the identified A. communis might be predominantly dead and thus unable to produce amyloid-like EPS. This hypothesis can be tested using cell-impermeable nucleic acid stains indicating compromised membranes. However, the in-house available SYTOX Orange and Propidium Iodide stains were not suitable for this purpose because of partial spectral overlap with mScarlet-I.

However, augmentation of this strain did not lead to measurable effects on the reactor sludge morphology after 24 h and longer. This discrepancy might be caused by the already excellent flocculation status of the reactor sludge at the time of augmentation providing few single cells or light small flocs to bind to. Again, a higher fraction of augmented bacteria in the sludge might have led to clearer effects. In total 3.33 g of A. communis was added gradually to 13.38 g of reactor biomass while Xin et al. (2017) added as much as 13 g of TN-14 to 3.3 g of conventional sludge to facilitate aerobic granulation.

Antibiotic resistance. Antibiotic resistance is an effective tool for monitoring the survival and persistence of an augmented strain in activated sludge but is clearly a monitoring tool limited to the laboratory. Given that the presence of antibiotic-resistant organisms in the environment can contribute to a public health problem (e.g., Zaman et al. 2017), it is not appropriate to use such an approach to monitor augmentation in full-scale systems. In these full-scale systems, the augmentation should be performed with the wild-type strain. Evidently, in laboratory-scale experiments, it is necessary to handle antibiotic-containing growth media and antibiotic-resistant organisms properly, i.e., by autoclaving prior to disposal.

Flocculation activity index. Results for the flocculation activity index (FAI) against kaolin cannot be compared to the literature because of the adapted equation for FAI accounting for the contribution of the bioflocculant producer to the absorbance. Moreover, the absorbance was measured after 5 min of settling instead of 1. With this method, the bioflocculant producers investigated in this study might often appear more effective, while in reality a prolonged settling time was required to even obtain a measurable effect. Finally, the pipetting of the supernatant appeared very prone to human error as shown in the large standard deviation for three separate measurements for of the same deflocculated activated sludge sample (0.660 ± 0.092, N = 3).

Only a small fraction of the augmented bacteria survived in the bioreactor such that their hypothesized impact on amyloid-like content and bioflocculation could not be assessed. Competition, grazing, and abiotic factors might have hampered the survival and persistence of A. communis. Three modifications to the experimental design and bioaugmentation approach are formulated here to assess these concerns. Firstly, future bioreactor experiments might operate the reactor without a membrane but with a short settling phase followed by effluent removal from the top. This method of hydraulic selective pressure and washout is often used to promote granulation from floccular sludge (Qin et al. 2004a, 2004b). This adaptation to the reactor mechanical design and operation will thus select for the best settling flocs and is hypothesized to give a selective advantage to the EPS-producing augmented strain. Indeed, the potential presence of indigenous denitrifiers did not provide A. communis with enough sufficient selective advantage to establish itself in the reactor based on its denitrifying function. Still, we suggest using a membrane during the first 24 h after augmentation to give suspended cells the time to adhere and avoid their washout. Secondly, while the membrane cell retention system avoided the washout of suspended cells including A. communis, it did not protect the bacteria from metazoan and protozoan grazing. Future work might therefore use the embedding of bacteria within an alginate matrix as another cell retention strategy. This embedding has proven to be successful against protozoan grazing in a nitrifying laboratory-scale reactor augmented with an aerobic denitrifying bacterium (Bouchez et al. 2009). Thirdly, while A. communis could not thrive under the abiotic conditions present in the reactor, other EPS-producing denitrifiers might. Future research might rely on 16S rRNA sequencing of samples stored from this experiment to select dominant potential amyloid-producing genera for augmentation.

In terms of cost and efficiency, this direct augmentation of EPS producers (presumably with alginate) will be more suitable for other higher added-value biotechnology applications, where competition is lower and influent conditions are more controlled. Potential applications include sand filters for drinking water production (aimed at removing specific micropollutants or other emerging contaminants), the retention of non-flocculating bacteria (that will then form an aggregate through the bioflocculant producing microorganism), and the harvesting of microalgae (Jiang et al. 2006; Lee et al. 2009; Wang et al. 2021). For full-scale activated sludge systems, the promotion of native structural EPS producers and their structural EPS production by adjusting the operating conditions is still preferred over bioaugmentation.

This study aimed to explore the impact of structural EPS on bioflocculation of activated sludge by augmentation of a structural EPS producer. The selected A. communis strain was shown to produce gel-like material with thioflavinophilic properties, suggesting the presence of amyloid adhesins. A. communis was thus identified as a proper candidate for the proposed augmentation experiment, thereby confirming the first research hypothesis. Augmentation of Azoarcus was performed gradually over 31 days. Enumeration of colony-forming units on selective agar plates supplemented with antibiotics suggested that only a fraction of augmented Azoarcus survived. Three explanations were proposed: competition with indigenous organisms filling the same metabolic function (denitrifiers), grazing by rotifers and ciliates, and unfavorable abiotic conditions were proposed as explanations. Grazing was visually confirmed through the detection of the Azoarcus-produced mScarlet-I fluorescent protein emission inside these organisms. As a result of the low abundance of augmented Azoarcus in the bioreactor, the hypothesized beneficial impact on amyloid-like content (hypothesis 2) and bioflocculation performance (hypothesis 3) could not be assessed.

A framework with two recommendations for future bioaugmentation experiments aiming to improve bioflocculation was established. Firstly, the construction of a fluorescent protein-producing mutant is recommended. This modification enables microscopic monitoring of the location of the augmented bacteria in the sludge floc, which might bring valuable insights into the initial attachment and further integration processes. The stable inheritance of this fluorescent marker was confirmed in the studied Gram-negative strain. Secondly, structural EPS-producing capacities should be included as a criterium for strain selection. Short-term experiments demonstrate the beneficial impact of Azoarcus and amyloid-producing E. coli strains on the flocculation of clay and deflocculated sludge.

An-Sofie Christiaens holds a PhD grant for Strategic Basic Research from the Research Foundation-Flanders (FWO-1S41422N). Aquafin (Belgium) is acknowledged for providing sludge. We thank Daniel Otzen (Aarhus University) for the E. coli K-12 mutants, Ivo Vankelekom and Ayesha Ilyas (KU Leuven) for the fabrication of the membranes, Dirk Springael, Tran Quoc Tran, and Tinh Nguyen Van (KU Leuven) for enabling the transposon mutagenesis and help with Sanger sequencing, Tom Van Gerven (KU Leuven) for providing access to the ultrasonic processor, Johan Martens (KU Leuven) for providing access to the microplate reader, and Johan Hofkens and Rik Nuyts (KU Leuven) for providing access to the confocal microscope.

All relevant data are available from an online repository. The repository can be accessed via: https://doi.org/10.48804/ROBCTI.

The authors declare there is no conflict.