3.1. Removal of BPs from water-sediment microcosms
In this study, the effectiveness of the removal of BPs present at a concentration of 10 mg L− 1 in the created river water and sediment microcosms was examined. Statistical analysis of obtained data showed significant differences (p < 0.05) both between the time of sampling points and microcosms. Since time has a significant impact on removing pollutants from the environment (Zaborowska et al. 2020), in this study, we analyzed the differences between microcosms at each sampling point. Figure 1 shows the percentage of residual content of BPA and BPS in the non-bioaugmented microcosms (II) and in the bioaugmented microcosms with different treatments (IV-VIII) during a 10-week experiment. As indicated by the data, the removal of BPA proceeded much faster in comparison with BPS removal. BPA disappeared from microcosms II, IV, and VI within 30 days. However, in bioaugmented microcosms without autochthonous bacteria (VII and VIII) BPA was still detected on the last day of the experiment. Regarding BPS, this compound was not completely removed in tested microcosms during the experimental period. Nevertheless, in bioaugmented microcosms (III, IV, V, and VI) BPS disappearance was significantly higher (p < 0.05) as compared to microcosms without autochthonous bacteria (VII and VIII). Surpassingly, there was no significant difference in BPS removal between bioaugmented microcosms and microcosm II consisting of river water and sediment. The obtained results showed that the introduction of BPs-removing bacteria BG12 and K1MN into the water-sediment microcosms did not significantly increase the removal efficiency of either BPs compared with the non-bioaugmented microcosm II. However, a significant decrease of BPs was observed in bioaugmented microcosms as compared to microcosms VII and VIII, indicating that BPs removal was primarily due to the activity of indigenous microorganisms. The presence of pollutants mixture may influence the removal of individual substances, which was probably the reason for the low efficiency of BPs removal from these microcosms by the introduced consortium (González-González et al. 2022). Especially that, BG12 was effective in BPA removal, while K1MN was in BPS elimination, but from BSM medium where each of them was a single carbon source (section 2.2). The high ability of microorganisms inhabiting water and sediments to BPs removal was observed in the previous studies. For instance, Zhou et al., (2020) have demonstrated that BPA is rapidly removed in the river, suggesting that autochthonous BPA-removing bacteria are widely distributed in the water. The vital role of microorganisms in BPA biodegradation was demonstrated by Tong et al. (2021), who observed the complete inhibition of BPA removal in sterilized sediment contaminated with BPA. Similarly, in microcosms consisting of river water or activated sludges from municipal or village WWTPs, BPS was biodegraded in a range of 40%-55% in 52 days (Frankowski et al. 2020). In another study, an activated sludge bioreactor completely removed BPS at a concentration of 50 mg L− 1 in 10 days (Huang et al., 2019).
Besides the metabolic activity of microorganisms, adsorption is another significant BPs removal way (Ferrer-Polonio et al., 2021). Therefore, BPs-contaminated microcosms with autoclaved river water and sediment (VII) and BPs-contaminated microcosms composed of BSM and consortium (VIII) were included in the experiment to analyze which of these two strategies contributes to BPs loss in this study. Due to the non-significant differences in BPs content between microcosms VII and VIII, it was reported that bacterial activity played a key function in BPs removal, which was in accordance with previous results (Sun et al., 2017).
The data presented in Fig. 1 were supported by the values of the removal rate (k) and the half-life (t1/2) constants (Table 2). The obtained k and t1/2 values confirmed that BPs removal was more effective in microcosms II, IV, V, and VI as compared to microcosms VII-VIII, and BPA removal proceeded much faster than BPS removal. Many studies on BPA removal process kinetics assumed a simple first-order (Rajendran et al., 2017) or even a pseudo-first-order kinetic models (Kohtani et al. 2003). However, Abargues et al. (2018) demonstrated that the second-order kinetic model is suitable for predicting BPA removal, which was confirmed in this study. The second-order kinetic model gave reasonable good fits using a linear least-squares analysis (0.912–0.972). The calculated k and t1/2 constants showed that the removal rate of BPA decreased in the following order: microcosm II > microcosms IV and VI > microcosms VII and VIII. The BPS removal constants were calculated based on a pseudo-zero kinetic model, which also gave good fits using a linear least-squares analysis (0.919–0.998). The of k and t1/2 constants were the highest for microcosms IV and V followed by microcosm II. Moreover, the constants were at least one order of magnitude smaller than those calculated for BPA. Such values probably result from BPS recalcitrance to microbial removal (Ogata et al. 2013). The first-order kinetic model revealed that BPS was removed faster (k = 0.04–0.16 day− 1, t1/2= 5.8–17.3 days) during biological wastewater treatment with activated sludge compared to our study (Kovačič et al. 2021). On the other hand, the higher persistence of BPS to removal in the surface water of Taihu Lake was observed by Zhou et al. (2020), who studied the removal of eight bisphenols in the natural environment. They found that the amount of BPS at the end of 49-day incubation was almost unchanged in comparison with day 1. The results of our and the studies mentioned above suggest that the efficiency of BPS removal in the natural environment might strongly depend on the structure and richness of microbial communities, which comprise various bacteria with different abilities to degrade pollutants.
Table 2. Rate constant (k), values of half-life (t1/2), and correlation coefficient (R2) of BPA second-order removal kinetic (A) and BPS pseudo-zero order removal kinetic (B) in the microcosms.
Microcosms: II – water and sediment, BPs, (control); IV - consortium, water and sediment, BPs; V - consortium, water and sediment, BPS; VI - consortium, water and sediment, BPA; VII - consortium, autoclaved water and sediment, BPs, (control); VIII – consortium, BPs, (control).
3.2. Survival of introduced strains in water-sediment microcosms
The effectiveness of bioaugmentation is strongly connected with the survival rate of introduced cells; hence, an important aspect of this study was monitoring the survival of the inoculants. To achieve this goal BG12 and K1MN were tagged with FP protein and counted under a fluorescent microscope. Using this technique, it was found that neither introduced strain was detected in any microcosms from day 40 (Table S1). The compilation of the results listed in Fig. 1 and Table S1 is presented as a 3D response surface plot in Fig. 2 (a-c). The red region indicates the region with the highest number of bacteria which includes the sum of both introduced strains, while the blue area represents the region where bacteria were not present. The distribution of the red region was found to be dependent on time and the amount of either BPA or BPS or both of them in microcosms. In microcosm IV (Fig. 2, A), it was confirmed that after 35 days of the experiment and below 40% of the initial concentration of both bisphenols, no bacteria were detected. In microcosm V (Fig. 2, B), introduced bacteria have not been detected at about 50% of the initial BPS concentration and after 35 days of incubation. The number of introduced FP-tagged cells decreased fastest in microcosms amended only with BPA (Fig. 2, C). After 20 days of the experiment, bacteria were not detected. So far, only a few studies have focused on monitoring the survival of introduced strains in aquatic environments bioaugmented with selected strains (Zhou and Gough, 2016). Such studies allowed us to evaluate whether the higher effectiveness of pollutants removal resulted from the activity of the introduced strains or rather from biostimulation that occurs when dead cells of inoculants serve as an additional source of energy and carbon, leading to increased activity and the abundance of autochthonous microorganisms (Płociniczak et al., 2020). Such an effect was observed after the bioaugmentation of hydrocarbon-contaminated soil with Rhodococcus erythropolis CD 106 (Pacwa-Płociniczak et al., 2019). The decline in the number of introduced BG12 and K1MN strains until 30 day and their total lack in the successive sampling days observed in our study might be explained by the loss of planktonic cells due to competing indigenous microbes, protozoan grazing, and suboptimal environmental conditions. Taking these into account, the successful application of bioaugmentation is dependent on the use of appropriate bacterial strains featuring resistance to external, abiotic, and biotic factors (Xiong et al., 2017).
3.3. Bacterial community compositions in water-sediment microcosms
To answer the question if inoculants and BPs changed the diversity and/or activity of the autochthonous bacterial communities in microniches, thes tested microcosms were subjected to high-throughput sequencing of the 16S rRNA gene.
Analysis of samples from treatments I-VI on days 0, 35, and 70 yielded a total of 3,480,480 valid reads. The number of reads varied from 37,104 (sum of the reads for the sample IV day 35) to 147,6501 (sum of reads for the sample I day 0) and were grouped into OTUs (using 97% minimum similarity), and these ranged from 498 (V day 35) to 4615 (I day 0).
High-throughput sequencing analysis and alpha diversity indices are shown in Table S2. Statistically significant (p < 0.05) differences in the values of these indices were observed for all tested indices, indicating a different species richness between samples from microcosms IV-VI on days 35 and 70. For samples from microcosm IV, different significant values were obtained for Simpson, for microcosm V for Shannon and Simpson, and for microcosm VI for Chao1 indices. These results were confirmed by principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity matrix (Fig. 3). For all analyzed microcosms, the microbial diversity was the same at the beginning of the experiment (day 0). Furthermore, a clear relative dispersion between microcosms I-III and IV-VI was observed on days 35 and 70, indicating that both bioaugmentation and BPs changed the bacterial community structure. PERMANOVA analysis confirmed the significance of clustering (p = 0.001); however, with three replicates for each sample, the clustering was not significant with pair-wise comparisons of the types.
The taxonomic composition of the bacterial communities in the analyzed microcosms at the phylum level showed the predominance of Proteobacteria (including Alphaproteobacteria, Betaproteobacteria Gammaproteobacteria, and Deltaproteobacteria) Acidobacteria, Actinobacteria, and Bacteroidetes were abundant in all the samples at the beginning of the experiment (Fig. 4). This finding is in accordance with outcomes reported previously, which have shown that Proteobacteria members have been found to be responsible for the removal of diverse phenolic compounds, including BPA (Czarny et al., 2020). In turn, Cydzik-Kwiatkowska et al. (2017) revealed that the abundances of Proteobacteria and Bacteroidetes were over 0.5% of all identified sequences, which was the highest percentage of bacteria in aerobic granules used to treat wastewater containing BPA in concentration up to 12 mg L− 1. This suggests that the main controlling factor affecting the relative abundance of the above-mentioned phyla is likely to be the concentration of bisphenols in the treated microcosms.
During the removal, a remarkable difference in the bacterial community between microcosms I-III and microcosms IV-VI was found. In those first groups of samples, the bacterial community was comparable and composed primarily of four different phylogenetic groups at the family taxonomic rank: Xanthomonadaceae, Hydrogenophilaceae, Chitinophagaceae, and Nitrospiraceae. In microcosms IV-VI, the most dominant bacterial families in all samples were Hyphomicrobiaceae (including Hyphomicrobium genera with an abundance of 12.88% for microcosm IV, 13.64% for microcosm V, and 6.66% for microcosm VI; data not presented), Chitinophagaceae, and Microbacteriaceae. Additionally, the Isosphaeraceae family was also present in high abundance in microcosms IV, Cryomorphaceae in microcosm V, and Caulobacteraceae in microcosm VI. No reads belonging to the Pseudomonadaceae and Moraxellaceae families were detected.
On day 70, the re-organization of the bacterial community compared to days 0 and 35 was observed. Bacterial community structures in microcosms I-III were comparable and composed mainly of Ellin 329, Chitinophagaceae, Xanthomonadaceae, and Rhodospirillaceae. In microcosms IV Acidimicrobiales followed by Chitinophagaceae, Caulobacteraceae, and Hyphomicrobiaceae were the main dominant group. These two latter families also dominated in microcosm V, while Xanthomonadales followed by Chitinophagaceae dominated in microcosm VI.
To better understand the BPs removal process, a heat map clustering analysis of the top 18 most abundant taxa in the tested microcosms was used (Fig. 5). It illustrates how the bacterial population changed depending on the duration of the incubation period and the type of BPs analog. Comparison of the microbial community composition of microcosms from different treatments and times of the experiment performed using the OTUs with the highest abundance distinguished two clusters. The first cluster included microcosms I-III of all tested days, and microcosms VI of day 70, while the second cluster included microcosms IV and V of days 35 and 70 and microcosms VI of day 35. As Fig. 4 shows, during the removal, the proportions of Acidimicrobiales and Caulobacteriaceae in microcosm IV and V on day 70 exceeded Hyphomicrobiaceae and Cryomorphaceae, which were dominant on day 35. In bioaugmented microcosm VI amended only with BPA, Cryomorphaceae, Hyphomicrobiaceae, and Caulobacteriaceae prevalent on day 35.
These basic findings are consistent with research showing that members of Bacteroidetes, Chloroflexi, Gemmatimonadetes, Actinobacteria, and Planctomycetes were the major bacterial groups in BPA-removing anaerobic sediments (Yang et al. 2015). In another study Cyanobacteria, Chloroflexi, Planctomycetes, and Acidobacteria were the second predominant phyla present in mangrove sediment contaminated with endocrine-disrupting chemicals, including BPA (Yuan et al. 2017). Hyphomicrobiaceae, especially Hyphomicrobium, Caulobacteraceae, Chitinophagaceae, and Cryomorphaceae were previously found to metabolize different hardly degradable xenobiotics (Wang et al., 2019). This may explain their high abundance in microcosms IV-VI.
To further identify the microbial biomarkers in each treatment, a LEfSe analysis was applied. Figure 6 shows the bacterial groups between each of the six microcosms on days 35 and 70 with significant differences with log10 (LDA scores) > 4.0, while Figure S1 with log10 (LDA scores) > 2.5. No statistical differences for tested microcosms on day 0 were obtained. LEfSe analysis revealed that different taxa were significantly enriched in the different treatments, thus indicating that the various niches in different microcosms could select for specific microbial community members. Bacteria taxa that differ significantly in microcosms II-VI, besides Chitinophagaceae, were not considered core community members in the microcosm I composed only of river water and sediment. Remarkably, in the same microcosms sampled on different days, a significantly high abundance of various bacterial taxa were detected. Only the Hyphomicrobiaceae family and Parvibaculum genus were present in significantly high abundance in microcosms V on both tested days. Interestingly, the Sphingomonas genus was enriched considerably in microcosm VI on day 35. Therefore, we speculate that these bacteria could use BPS as a carbon substrate. Moreover, the high abundance of Sphingomonas sp. was found in microcosm VI amended only with BPA. Previously, Sphingomonas strains were demonstrated to be good BPA-degraders (Spivacks et al., 1994).
To sum up, no reads belonging to the Pseudomonas and Acinetobacter in all microcosms on days 35 and 70 were found. This confirms that the introduced bacteria did not survive in the microcosms after day 30 of the experiment. Moreover, the composition of the active BPs-removing communities during the 10-week incubation period was shaped by the bioaugmentation and the addition of BPs and was altered to achieve accelerated BPs removal. From this standpoint, reconstruction of the bacterial community may be due to the selection and maintenance of BPs-removing strains in treated microcosms until BPs have been converted and/or the appearance of strains that utilize intermediates of BPs removal. These results revealed that both bioaugmentation and BPs changed the composition of major bacterial groups. Most of the dominant bacterial groups which were present on day 35 can remove BPA, while those present during the whole incubation period can degrade/transform BPS. These findings agree well with previous studies wherein the microbial community has also been shifted during the removal of various organic compounds (Huang et al., 2019).
3.4 Predicted metabolic functions of bacterial communities in BPs-amended microcosms
To complete the information about genes responsible for metabolic pathways, a more detailed analysis of differences in genes and pathways abundances potential between the microcosms without BPs addition (I, III) and BPs-amended microcosms (II, IV, V, VI) was performed. The functional prediction of the metagenome of the microcosms from days 35 and 70 revealed the presence of 432 pathways and 2339 proteins. PICRUSt2 detected in the analyzed microcosms pathways and proteins, which were annotated to the following categories: genetic information processing, metabolism, cellular processes, and environmental information processing. The most significant differences were detected in the abundance of genes of the protocatechuate removal pathway, which were predominant on day 35 in microcosm II and day 70 in treatments II and IV (P23-PWY; Figure S2), thus suggesting that autochthonous microflora may utilize BPs as carbon sources. Previously, it was shown that BPA and BPS are removed to many intermediates, including catechol (Ogata et al., 2013). These results, as well as the findings presented in this study tie well with a previous analysis demonstrating that protocatechuate removal might be involved in BPA transformation (Zhou et al. 2015).
In addition, Cydzik-Kwiatkowska et al., (2021) have found a high abundance of genes coding enzymes responsible for benzoate removal in BPA-exposed aerobic granular sludge. This is in accordance with the results presented in this study, as 4-carboxymuconolactone decarboxylase (EC.4.1.144) was the most abundant protein in BPA-amended microcosms. This may be related to the structure of BPA, which contains acetate. Moreover, a high abundance of glutathione transferase (EC.2.5.1.18) engaged in the metabolism of xenobiotic by cytochrome P450 was observed in microcosms amended with BPs. It was shown that this enzyme is responsible for the removal of a large variety of compounds, including chlorinated biphenyls and creosote (Brennan et al. 2009; Smułek et al. 2020).