Effects of pharmaceuticals compounds and calcium on granulation, microbiology, and performance of anaerobic granular sludge systems

Applying anaerobic and aerobic granular sludge in wastewater treatment is among the most useful and promising biotechnologies. The self-immobilization of microorganisms into anaerobic granules significantly increases the biomass's ability to withstand high loads and shocks, improves its retention in the reactor, and increases its tolerance to toxicity since the compact structure of the granule protects the bacteria from inhibitors and toxic pollutants (Sponza 2001). Due to these unique properties, anaerobic granulation technologies have been exploited for the biological treatment of municipal and industrial wastewater (Liu & Tay 2004).

The most accepted theory of sludge granulation is that rod-shaped and filamentous bacteria initiate the formation of granules and generate structures that support the further growth of the granules (Doloman et al. 2017). Granulation is a complex process, depending on many physicochemical and microbiological factors, including substrate characteristics, extracellular polymeric substances (EPS), positive ions, pH, temperature, and reactor operation (Dignac et al. 1998).

Some studies have reported that the addition of some positive ions, such as Ca²⁺, Mg²⁺, Fe²⁺, and Fe³⁺, may stimulate the formation of granules by attracting negatively charged bacteria or agglomerates, which decreases the repulsive electrostatic forces between bacteria or bacterial agglomerates, allowing the self-aggregation of cells (Cao et al. 2014).

The EPS are also related to sludge granulation. These substances are composed of a mixture of macromolecules, including proteins, polysaccharides, humic substances, and nucleic acids, which trap the microbial cells in a three-dimensional matrix (Ding et al. 2015). Miksch & Konczak (2012) proposed that the crucial role of protein in granular stability was based on its high correlation with hydrophobicity (0.97) and that the protein concentration increased fivefold, while polysaccharide remained unchanged during sludge granulation.

The presence of pharmaceuticals in the aquatic environment has received increasing attention in the last decade. A wide range of pharmaceuticals and their metabolites are continuously introduced into the aquatic environment through human consumption, excretion, and incomplete removal during sewage treatment. It leads to constant exposure of the aquatic environment to a large cocktail of bioactive molecules that eventually affect water quality and can impact the ecosystem, drinking water supply, and human health (van Nuijs et al. 2015).

Pharmaceutical products are being detected in water matrices at levels ranging from micrograms per liter (μg/L) to nanograms per liter (ng/L). Although these concentrations of pharmaceuticals in water samples are relatively low, their potential impact on the aquatic environment is a major concern. Effects such as endocrine disruption, the incidence of cancer, and reproductive problems in humans have been associated with these substances (USEPA 1997). In this way, technologies that aim to completely remove these microcontaminants must be studied. Anaerobic technology is a viable option for treating wastewater containing pharmaceutical compounds due to its advantages of supporting a high organic load, lower sludge production, and lower operating cost compared to the conventional activated sludge process (Shi et al. 2017). However, in the case of anaerobic operating systems that require granular biomass, it is interesting to understand how microorganisms react to the presence of pharmaceutical compounds, mainly in the formation of the granule.

As anaerobic granular technologies have been widely exploited for the biological treatment of municipal and industrial wastewater, this study aims to evaluate the potential of Ca²⁺ as a granulation precursor for anaerobic sludge and how pharmaceuticals compounds affect granulation. For this purpose, continuous and intermittent dosages of Ca²⁺ in the presence and absence of pharmaceuticals compounds (loratadine, prednisone, fluconazole, fenofibrate, betamethasone, 17α-ethinyl estradiol, and ketoprofen) were assessed. Granules size and time for granulation were compared. Moreover, reactors’ performances concerning volatile fatty acids, organic matter, and alkalinity were compared. Samples from the microbial community were analyzed by high-throughput sequencing using the Illumina Platform. Finally, pharmaceuticals removal was investigated in a granular sludge bioreactor (EGSB).

The composition of the synthetic sewage used is in Table S1. Calcium chloride was used as a precursor of sludge granulation, and its concentration was increased to 250 mg L⁻¹ in some reactors.

The influence of seven pharmaceutical compounds on the granulation of anaerobic sludge was investigated. The pharmaceutical compounds evaluated were: loratadine (CAS no. 79794-75-5), prednisone (CAS no. 53-03-2), fluconazole (CAS no. 86386-73-4), fenofibrate (CAS no. 49562-28-9), betamethasone (CAS no. 378-44-9), 17α-ethinyl estradiol (CAS no. 57-63-6), and ketoprofen (CAS no. 22071-15-4). Each pharmaceutical product was spiked in a concentration of 2 μg L⁻¹. This concentration is consistent with the range in which pharmaceuticals can be found in real domestic sewage (Bisognin et al. 2019). The therapeutic classes of pharmaceutic products are in Table 1.

Five batch reactors of 1.0 L were used to evaluate the effects of pharmaceuticals compounds on the anaerobic sludge granulation. All reactors were sealed to maintain anaerobic conditions. They were operated as follows: 8 hours after the beginning of the operation, half of the reaction medium was removed, and the volume was completed with fresh feed (8 hours cycle). Then, after 16 hours of operation, the procedure was repeated – half of the reaction medium was removed, and fresh feed completed the volume (16 hours cycle). Thus, the reactors were operated on alternating cycles of 8 and 16 hours successively.

R1, R3, and R5 reactors were control reactors. R1 was fed with synthetic sewage, and 250 mg L⁻¹ of CaCl₂ was spiked into this synthetic sewage. R3 was fed with synthetic sewage, and 250 mg L⁻¹ of CaCl₂ was spiked fortnightly into the synthetic sewage. R5 was fed just with synthetic sewage.

The R2 reactor was fed with synthetic sewage and continuous addition of CaCl₂ and pharmaceuticals. The continuous addition means that each time the reactor was fed, CaCl₂ and pharmaceuticals were added at concentrations of 250 mg L⁻¹ of CaCl₂ and 2 μg L⁻¹ of each pharmaceutical compound.

The R4 reactor was fed with synthetic sewage and intermittent addiction of CaCl₂ and pharmaceuticals. The intermittent addition of CaCl₂ and pharmaceuticals can be understood as: the reactor was fed with sewage with low CaCl₂ concentration (7 mg L⁻¹) and no pharmaceuticals for one week. The following week, the reactor was fed with a high concentration of CaCl₂ (250 mg L⁻¹) besides the pharmaceutical compounds’ solution (2 μg L⁻¹ of each pharmaceutical compound).

For inoculation, it was used sludge from a UASB (upflow anaerobic sludge blanket) reactor situated in the Onça wastewater treatment plant in Belo Horizonte city (Copasa – Minas Gerais/BR). In each reactor, 500 mL of sludge with 9.65 g L⁻¹ total suspended solids (TSS), 4.65 g L⁻¹ fixed suspended solids (FSS), and 5.0 g L⁻¹ volatile suspended solids (VSS) was added.

The size of the granules was monitored once a week with the assistance of a laser dispersion particle size distribution analyzer (Model LA-950, Horiba).

Concerning reactors monitoring, the supernatant liquid removed from reactors (effluent) and each of the feed solutions (influent) were characterized. The parameters pH, dissolved organic carbon (DOC) (Method 5310 A), and VSS (Method 2440) were evaluated twice a week according to Standard Methods for the Examination of Water and Wastewater (2005). The volatile fatty acids (VFAs) were monitored according to Dillalo & Albertson (1961), and alkalinity according to Dilallo and Albertson, modified by Ripley et al. (1986). The influent and effluent samples were analyzed periodically for reactor monitoring. COD concentration (Method 5520), ammoniacal nitrogen (N-NH₄⁺) (Method 4500-B and C), and phosphorus (P-PO₄³⁻) (Method 4500 B and C) were also determined according to Standard Methods for the Examination of Water and Wastewater (2005). The EPS and the soluble microbial products (SMP) extraction are in the supplementary material.

Pharmaceutical identification and quantification were performed on the samples of influent and effluent streams of the reactor. First, the samples were filtrated with white band quantitative filter paper (Quanty, JP40. JProlab, São José dos Pinhais, PR, Brazil), with the most pores of 25 μm size. Next, they were concentrated by solid-phase extraction. For concentration, cartridges (Strata C18-E, 55 μm, 70 A, Ref. 8B-S001-HCH) were conditioned with 5 mL of methanol (Baker Analyzed^® HPLC Solvent) followed by 5 mL of ultrapure water (Gehaka) (Gros et al. 2013). After conditioning the cartridge, the sample percolated it at a velocity of around 4 mL min⁻¹. After percolation, the cartridge was rinsed with 10 mL of ultrapure water. The cartridges were stored at −20 °C. Finally, analytes were eluted with 4 mL of methanol. Then, the sample was filtered on Syringe-driven filters (Jetbiofil) before injection on high-performance liquid chromatography–mass spectrometry (HPLC/MS).

Pharmaceuticals quantification was performed with the Shimadzu HPLC model Prominence DGU/20A3. The chromatographic column used was a 2 mm diameter, 50 mm length, and 2 μm particle size Shimadzu C18 model Shim-pack XR-ODS. The micrOTOF-QII mass spectrometer (Bruker) with an electrospray ionization source (ESI) was used. The method used is detailed in Faria et al. (2020).

With Statistica 10.0 software, the differences between the average size of the granules at all reactors were tested using Kruskal–Wallis nonparametric statistical tests followed by the multiple comparisons by the level of significance (α) of 5%.

For the molecular biology assay, biomass samples were collected from each of the five reactors. The methodological steps followed to perform DNA extraction and polymerase chain reaction (PCR), assembly of the amplicon library, and sequencing analysis are described in Arcanjo et al. (2021).

The removal of pharmaceutical compounds by anaerobic granular sludge was assessed in an expanded granular sludge bed (EGSB) bioreactor. The sludge used in EGSB was also obtained from the UASB of the Onça wastewater treatment plant. The reactor had a volume of 2.72 L, 1.5 m high, and an inner diameter of 0.06 m. The operation was conducted under mesophilic conditions (25±4 °C), and the upflow velocity was 3.8 m h⁻¹. The hydraulic retention time (HRT) was 13.51 h. More information about the reactor can be found in Faria et al. (2020). The next 50 days after inoculation, the reactor was fed with synthetic sewage to promote the microorganisms’ adaptation. Next, the pharmaceutical compounds were introduced continuously to the reactor feed for 130 days. Pharmaceuticals concentration and the wastewater composition were the same as for batch experiments.

The results of the granulation tests (Figure 1(a)) indicate that granulation occurred in all reactors. Nonetheless, according to the results of the statistical tests, there was no significant difference between the average sizes of the reactors granules that were operated continuously or intermittently spiked with CaCl₂ and pharmaceuticals within 120 days of operation. Only the R5 reactor showed granules with significantly smaller mean diameters than the other reactors.

Yu et al. (2001) defined that sludge is considered granular when 10% of the granules exhibited a diameter over 2.0 mm. Performing an analysis of the cumulative percentages of the granules’ average sizes of all the reactors, it was verified that the R4 reactor (intermittent pharmaceuticals and CaCl₂) presented 10% of the granules with mean diameters greater than 2.11 mm in 93 days, while the R3 reactor (intermittent CaCl₂) required 121 days for 10% of the granules to have mean diameters greater than 2.25 mm. The reactors R1 and R2 did not achieve cumulative percentages higher than 2.0 mm during the entire operating time (Figure 1(b)). It suggests that the continuous addition of CaCl₂ precursor does not promote sludge granulation during monitoring time (130 days).

These results evidence that the intermittent addition of the precursor is more interesting for the faster granulation of the anaerobic sludge. In addition, the presence of pharmaceuticals did not impair the sludge granulation. The toxicity associated with the compounds may have created a stressful condition, contributing to more EPS production and, consequently, a greater potential for granulation. Sreekanth et al. (2009) used a hybrid upflow anaerobic sludge blanket reactor to treat an effluent with drugs in thermophilic conditions. The average size of the sludge granule increased from 0.20 mm to 1.8 mm in diameter after 200 days of operation.

Table 2 presents monitoring data for the reactors. It is verified that the presence of the pharmaceuticals in R2 and R4 reactors did not cause a decline in reactors’ performance. Comparing R2 and R1 reactors, an increase in DOC removal was observed, attributed to microorganisms present in this reactor. It will be explained in item 3.3.

Comparing the performance of R3 and R4 reactors, where the addition of the CaCl₂ precursor and the pharmaceuticals products occurred intermittently (Table 2), it was possible to observe that the presence of the drugs did not have a great influence on alkalinity, pH values, volatile fatty acids concentration, and DOC removal.

The EPS quantification results showed higher concentrations of proteins (Figure 2). Hence, under the conditions of this study, higher protein/carbohydrate ratios in EPS are related to anaerobic sludge granulation. Punal et al. (2003) also observed high concentrations of proteins in the characterization of EPS, and these authors related the high protein concentration with the effective formation of the anaerobic granules. Furthermore, these results support the hypothesis that there was a greater EPS production in reactors fed with sewage containing pharmaceuticals (R2 and R4).

The good estimator was above 92%, indicating a suitable coverage by the sequencing of all samples. To assess the diversity in categorical data, Shannon and Simpson's indexes were evaluated (Table 3).

Considering relative abundance values above 0.2%, 9 bacterial phyla were identified: OP8, Actinobacteria, Caldiserica, Bacteroidetes, Synergistetes, Proteobacteria, Firmicutes, Chloroflexi, and Euryarchaeota (Figure 3).

Bacteria from the phyla Chloroflexi, Bacteroidetes, Firmicutes, Proteobacteria, Synergistete, and Caldiserica have already been reported in studies that used anaerobic reactors to treat sewage (Watanabe et al. 2017). Members of the Firmicutes are common in human feces, and Proteobacteria are present in activated sludge wastewater treatment plants and are microorganisms capable of removing organic pollutants (Antwi et al. 2017). In addition, performing microbiota analysis of an anaerobic digester that treated food waste, Li et al. (2015) found the main phyla Bacteroidetes, Firmicutes, Chloroflexi, Spirochaetae, and Synergistetes. These data corroborate the data obtained in this work.

The data analysis evidenced that the addition of pharmaceuticals and CaCl₂ in the reactor feed interfered with the abundance of phyla. Regarding the relative abundances of phyla in reactors R5 (control) and R1 (CaCl₂ continuous), there is an increase in abundance in Chloroflexi, Synergistetes, Actinobacteria, OP8, decreased abundance in Caldiserica, Bacteroidetes, Euryarchaeota, Proteobacteria, and no effect on the abundance of the phylum Firmicutes.

With the addition of pharmaceuticals compounds to the reactor (R2 reactor), the phyla Proteobacteria, Firmicutes, Bacteroidetes, and Caldiserica decreased their abundances, while the phyla Euryarchaeota, Chloroflexi, Synergistetes, Actinobacteria increased their abundances. The phylum OP8 showed no change in relative abundance with the addition of drugs. It suggests that the organisms of this phylum can adapt to new environmental conditions with the constant presence of pharmaceutical compounds.

The greater DOC removal obtained in the R2 reactor may be related to the increase in the abundance of the phyla Chloroflexi and Synergistetes since the members of these phyla are related to the higher removal of DOC observed in the R2 reactor (Table 2).

Comparing the relative abundances of the phyla in the control reactor (R5) with the reactor that was intermittently fed with CaCl₂ (R3), it is possible to verify that the intermittent addition of CaCl₂ increased the abundance of some phyla: Chloroflexi, Synergistetes, Actinobacteria, OP8, while the abundance of other phyla decreased: Caldiserica, Proteobacteria, Firmicutes, Euryarchaeota, Bacteroidetes. Members of the phylum Chloroflexi are related to the granulation in UASB reactors (Watanabe et al. 2017). Thus, the increased abundance of Chloroflexi may be the reason why granulation occurs in R3 (10% of the granules with a diameter greater than 2.0 mm). In contrast, the lower abundance in R5 may be related to this reactor having its granules significantly smaller in diameter.

Comparing the relative abundances of the R3 reactor with the R4 reactor, it is possible to verify the abundance of phyla Actinobacteria, OP8, Caldiserica, Bacteroidetes, Euryarchaeota decreased with the addition of pharmaceutical products. However, some phyla suffered a positive influence with the addition of pharmaceuticals – they increased their relative abundances: Proteobacteria, Firmicutes, Chloroflexi, while the phylum Synergistetes was not influenced by pharmaceuticals.

The increase in the abundance of the Chloroflexi phylum may also explain the more quick granulation in the R4 reactor and the non-decreased in the DOC removal (Table 2) since bacteria from this phylum, besides have been related to granules formation, are involved with the removal of organic material (Antwi et al. 2017). Moreover, some studies have already reported that representatives of phylum Firmicutes are carriers of drug-resistant genes (Zhao et al. 2011).

Corroborating with what was observed, some authors highlighted that the most abundant phyla associated with biomass granulation were Proteobacteria, Firmicutes, Bacteroidetes, Chloroflexi, and Actinobacteria (Wilén et al. 2018).

Concerning the genera, according to Figure 4, it is possible to verify that in all reactors in which the feed was modified, either with calcium chloride or with pharmaceuticals, the genera E6 and T78 were enriched. The presence of genus E6 bacteria has been reported in environments with recalcitrant compounds, such as petroleum and surfactants (Faria et al. 2017). Ji et al. (2020) highlighted the presence of the T78 genus in anaerobic membrane bioreactors and related these bacteria to carbohydrate degradation.

Genera belonging to both the Clostridiales order and the Clostridiaceae family may play an important role in anaerobic digesters, being correlated with biogas production and COD removal (Nguyen et al. 2018). Wéry et al. (2010) mention that bacteria of this family can persist after wastewater treatment and could be used as indicators of human fecal contamination. According to Figure 4, the relative abundance of these genera decreased in reactors R2, R3, and R4, demonstrating that the anaerobic digestion process is impacted by the addition of calcium chloride intermittently and by the presence of pharmaceutical compounds.

VadinCA02 is related to fermentative metabolism in anaerobic digestion processes (Wéry et al. 2010). Reactors R2 and R3 were the ones that showed the highest relative abundances of these genera (Figure 4) and also the reactors that showed greater DOC average removal values (Table 2), which is closely related to microorganisms in activity in the reactors.

The genus Syntrophus usually participates in the metabolism involved in the degradation of fatty acids, and the Porphyromonadaceae family is part of the microbiota of the human and animal gastrointestinal tract (Ji et al. 2020). A reduction in the abundance of these genera was observed in all reactors except in the control reactor (R5) (Figure 4). It demonstrated once again that anaerobic digestion was disturbed by the insertion of pharmaceutical compounds and CaCl₂ in the fed.

Ye et al. (2012) studied wastewater treatment in bioreactors and detected Actinomycetales among the ten most abundant orders. For the genus of this order, it was observed that the addition of CaCl₂ continuously in the fed increased its relative abundance and that the intermittent addition of this salt reduced its relative abundance (Figure 4).

When comparing the relative abundance of the Helicobacteraceae family in the reactors with the R5 reactor (10%), it can be seen that the highest relative abundance occurred in R1 (7%) and R2 (8%) reactors. Intermittent regime reactors (R3 and R4) may have created a stress condition, not favoring these organisms. On the other hand, the continuous addition of CaCl2 and pharmaceuticals must have enabled the adaptation of these organisms and favored the maintenance of their greater abundance.

When R1 and R2 are compared, it is clear that the presence of drugs did not impact the relative abundance of these organisms. Zarei-Baygi et al. (2020) reported an increase in the abundance of this taxon after the addition of antibiotics in an anaerobic membrane bioreactor. These authors state that the selection of the Helicobacteraceae family after the adding antibiotics suggests a high probability of gender resistance to multiple drugs. This is in line with what was observed in this study.

According to Figure 4, a decrease in the c_WCHB1-03 genus relative abundance in reactors with pharmaceutical compounds (R2 and R4) was verified. It indicates that this genus is not resistant to these pharmaceutical products.

Some authors claim that Methanosaeta members are particularly important in the formation and stability of granules (Gonzales-Gil et al. 2001). Corroborating this, a reduction in the relative abundance of this genus was observed in the reactors that needed more time for sludge granulation (R1 and R2). Methanobacterium members are reported as responsible for hydrogenotrophic methanogenesis in anaerobic reactors (Gonzales-Gil et al. 2001), and its presence was pronounced in all reactors (Figure 4).

Because of the presented results, it was possible to observe that the presence of pharmaceutical compounds in the reactors increased the relative abundance of taxa related to pharmaceutical compounds resistance and recalcitrant compounds degradation, such as E6 and Helicobacteraceae. In general, the insertion of calcium chloride and the presence of pharmaceutical compounds in the feed of the batch's reactors (R1–R4) interfered in the anaerobic digestion process, with different values of DOC removal. Besides, high standard deviations associated with values shown in Table 2 were obtained, which characterizes a high distribution of values. Genera related to the success of the anaerobic digestion process had their relative abundances decreased in these reactors: Syntrophus and Clostridiales.

Thus, it can be verified that the microbiota responsible for the anaerobic digestion of wastewater is sensitive to variations in feed compositions. However, even observing reductions in DOC removals in most reactors, no accumulation of fatty acids was observed in any reactor studied, which indicates that methanogenic archaea are consuming acids and producing methane. This highlights the versatility of the processes that involve anaerobic digestion and its range of applications.

Finally, it was verified that the strong influence of the genus Methanosaeta on the anaerobic sludge granulation since the reactors that had the lowest and late granulation (R1 and R2) were the reactors that had the lowest abundance of Methanosaeta.

The granular sludge was used to inoculate the granular sludge bioreactor (EGSB type) to evaluate the efficiency of removing pharmaceutical compounds. Among the pharmaceuticals evaluated, betamethasone, fenofibrate, and prednisone were not detected in the reactor effluent, being biodegraded or adsorbed (Faria et al. 2020) and, therefore, effectively removed.

The system was unable to remove ketoprofen, while fluconazole was only 1% removed. Ketoprofen log Dow of 0.39 may be a reason why there was low sludge adsorption. Similarly, fluconazole has a low log Dow value of 0.56. Besides, the fluconazole has an electron withdrawing group on the aromatic ring, making the molecule resistant to anaerobic degradation. Removal efficiencies of 17α-ethinylestradiol and loratadine were, respectively, 45 and 61%.

In addition to the removal of pharmaceutical compounds, the removal of chemical oxygen demand (COD), phosphorus (P), and ammoniacal nitrogen (N-NH₄⁺) were monitored (Table 4). While COD removal efficiency was 91%, P concentration was practically unchanged, and N-NH₄⁺ concentration increased.

The high COD removal and the non-removal (or even increase) of the P concentration have been observed in anaerobic conditions due to phosphorus accumulating organisms' (PAOs) presence (Show et al. 2020). Although a specific search for PAOs was not carried out, it is known that there are PAOs belonging to the phylum Actinobacteria (Stokholm-Bjerregaard et al. 2017,) and organisms of this phylum were identified in all reactors.

Regarding N-NH₄⁺, it is usually found at high concentrations in anaerobic bioreactor effluents since it is a product of the anaerobic decomposition of organic nitrogen (Babson et al. 2013). Thus, it explains the higher ammoniacal nitrogen concentrations in the reactor effluent.

The intermittent dosage of CaCl₂ proved to be a promising strategy for improving the granulation process since it increased granules size and reduced the time for sludge granulation. Through the microbial community analysis, it was noticed that the Methanosaeta genus was related to granulation. Besides, it was not observed that pharmaceutical compounds’ presence impairs the sludge granulation or anaerobic degradation. On the contrary, it was noticed that the intermittent presence of these compounds could stress the microbial community, increasing the production of EPS and favoring granulation. Otherwise, the 16S rRNA gene analysis revealed taxa resistance to recalcitrant compounds (E6 and Helicobacteraceae) in the presence of pharmaceuticals. Finally, the granular sludge bioreactor effectively removed Betamethasone, Fenofibrate, and Prednisone. Since micropollutants are increasingly identified in sewage, these discoveries are valuable for operating systems requiring granular sludge. Accordingly, this study provides significant upgrades to the advancement of anaerobic granular sludge technology, promoting the evolution of this technology.

The authors gratefully acknowledge the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – CAPES (Coordination of Improvement of Higher Level Personnel) for granting financial resources.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.