Natural sediment and associated communities were sampled on February 2018 at the upstream section of the Morcille river (Saint Joseph, 46°10′39.1″N 4°38′10.1″E), located in a forested non-urban area, upstream of vineyards. About 150 kg of wet surface sediment (0–3 cm) was collected using an Ekman grab, sieved at 2 mm, and brought to the laboratory. After sieving, sediment particle-size classes were distributed as follows: 48.8% (500 μm–2 mm), 30.7% (200–500 μm), 12.5% (50–200 μm), 8.0% (<50 μm), corresponding to a relatively coarse sediment, with a low organic matter content (3.1%) and a relatively high water content (36.1%).

The microcosms used for the experiment consisted of 15 glass indoor channels (length×width×height = 83cm×11cm×10cm), as previously described in Mahamoud Ahmed et al. (2018). After sediment homogenization, the bottom of each channel was spread with 7.2 kg wet weight (ww) of sediment and filled with 10 L of circulating water (2:1 v/v demineralized water:groundwater). The experimental design was similar to that used by Kergoat et al. (2021). Briefly, five experimental treatments (with independent triplicate channels) were tested: (i) control condition without SMX or SMZ, (ii) low-SMX: sediment immersed in water contaminated with SMX, at a nominal concentration of 500 ng L^–1; (iii) high-SMX: sediment immersed in water contaminated with SMX, at a nominal concentration of 5,000 ng L^–1, (iv) low-SMZ: sediment immersed in water contaminated with SMZ, at a nominal concentration of 500 ng L^–1, and (v) high-SMZ: sediment immersed in water contaminated with SMZ, at a nominal concentration of 5,000 ng L^–1. The experiment lasted 4 weeks and water was replenished weekly (for more details of the experimental procedure, see Kergoat et al., 2021).

Sediments were sampled for antibiotic and microbial analyses at the beginning of the experiment (Day 0, D0) and after 7 (D7), 14 (D14), 21 (D21), and 28 (D28) days. At each sampling time (i.e., before water replenishment), about 250 g of sediment was collected uniformly in each channel (using subsamples randomly distributed over the length of the channel), homogenized, and subsampled for further analyses.

SMX and SMZ concentrations in the sediment were measured at each sampling time (except for the REF treatment, for which only D0 and D28 samples were analyzed). After sampling, sediments were freeze-dried and crushed; 6.0000 ± 0.0005 g of sediment sample was then added to Nalgene PPCO 50 mL centrifuge tubes (previously checked for lack of SMX and SMZ sorption capacity) together with 50 μL of a 5,000 μg L^–1 SMX-d4 (Toronto Research Chemicals) and SMZ-d4 (LCG standards) mixture and 20 mL of MeOH/H₂O 4/1 (v/v) extractant. The tubes were stirred for 18 h using an orbital shaker (Heidolph Reax) at 50 rpm and room temperature. After centrifuging (12,500 × g for 15 min), the supernatant was concentrated in a Speed-Vac vacuum concentrator heated at 35°C. The residue was then dissolved in 10 mL of distilled water. After adjusting the pH to 3.8 ± 0.2, samples were concentrated 20-fold on Oasis HLB-500 mg cartridges (WatersTM) according to the manufacturer’s recommendations (elution: 10 mL methanol). The methanol fraction was concentrated in a Speed-Vac vacuum concentrator at room temperature, and the residue was dissolved in 500 μL HPLC-grade water. Antibiotics were assayed by LC/ESI-MS on a Thermo Scientific UHPLC Ultimate 3000 RSLC coupled to an Orbitrap Q-Exactive analyzer. The analyses were carried out in positive mode. The UHPLC was equipped with a Luna Omega Polar C18 column, 100 × 2.1 mm, 1.6 μm (Phenomenex) at 30°C with acetonitrile gradient + 0.1% formic acid (Solvent A) and water + 0.1% formic acid (Solvent B) 0–2.5 min, 30–64.5% A (linear) 2.5–2.6 min, 64.5–99% A (linear) 2.6–5 min, 99% A 5–5.1 min, 99–30% A 5.1–8 min, and 30% A. Flow rate was 0.45 mL min^–1. For the mass spectrometer, gaseous nitrogen was used as nebulizer gas (50 AU). The spray voltage was 3.0 kV. The limit of quantification for SMX and SMZ was 0.08 μg kg^–1 of dry sediment.

Potential microbial activities for aerobic respiration, denitrification, β-glucosidase, phosphatase, and leucine aminopeptidase were measured using the protocols previously described by Mahamoud Ahmed et al. (2018) and adapted from Foulquier et al. (2013).

Briefly, to measure aerobic respiration and denitrification rates, 10 g of wet sediment was immersed in 10 mL of distilled water under aerobic conditions (aerobic respiration) or 10 mL of a KNO₃ (2.16 g L^–1) solution under anaerobic conditions (denitrification) in 150 mL glass flasks. After 2 h and 5 h, headspace gases were sampled and analyzed by gas chromatography on an MTI 200 microcatharometer (MTI Analytical Instruments). Aerobic respiration and denitrification were expressed as ng of CO₂ or N₂O per g of sediment dw^–1 h^–1, respectively. To measure β-glucosidase, phosphatase, and leucine aminopeptidase enzymatic activities, 1.2 g of wet sediment was incubated for 30 min with pre-determined optimal concentrations of fluorogenic substrates, namely 4-methylumbelliferyl-β-D-glucopyranoside (MUF-Glu, CAS No. 18997-57-4) for β-glucosidase, 4-methylumbelliferyl phosphatase (MUF-P, CAS No. 3368-04-5) for phosphatase, and L-leucine-7-amido-4-methylcoumarin hydrochloride (Leu-AMC, CAS No. 62480-44-8) for leucine aminopeptidase. The activities were stopped with glycine buffer (0.05 M glycine, 0.2 M NH₄OH, pH 10.4), the samples were centrifuged, and fluorescence was measured (excitation wavelength 360 nm, emission wavelength 460 nm) using a microplate reader (Synergy HT BioTek Instruments). Standard curves of MUF (CAS No. 90-33-5) and AMC (CAS No. 26093-31-2) were plotted to convert fluorescence values into enzymatic activities. Results were then expressed as nmol of hydrolyzed compound per g of sediment dw^–1 h^–1.

Data (microbial parameters and sulfonamide concentrations) were checked for normality [Shapiro-Wilk test on ANOVA (analysis of variance) residuals, (Royston, 1982)] and homoscedasticity (Fligner-Killeen test, Conover et al., 1981). For each treatment, differences in SMX or SMZ concentrations across time were assessed by a repeated measure ANOVA followed by a pairwise t-test with Bonferroni adjustment. Maximal SMX and SMZ concentrations were compared using a t-test. For each sampling time, differences in microbial parameters between treatments (low-SMX, high-SMX, low-SMZ, or high-SMZ) and reference were assessed by ANOVA followed by a Dunnett post hoc test. To explore correlations between microbial parameters and sulfonamide concentrations in sediments, Pearson correlations were estimated for certain treatments. For all analyses, results were considered as significant if p < 0.05. Statistical analyses were performed using R software version 3.4.4 (R Core Team, 2020). Most figures were plotted using R package ggplot2 (Wickham, 2016).

Despite the lack of identified potential sources of contamination by sulfonamides in the sampled river section, a low concentration of SMZ was detected in the collected sediment (0.06 ± 0.10 μg kg^–1 dw at D0; Figures 1B,D), while no SMX was detected (limit of detection 0.08 μg kg^–1 dw). The SMZ and SMX analysis performed at D28 in the REF sediment confirmed the lack of SMX contamination and showed that SMZ concentrations in the non-spiked sediment were relatively stable over the experiment (0.10 ± 0.10 μg kg^–1 dw at D28; data not shown). Unexpectedly, sediments in the SMX treatments also contained low concentrations of SMZ.

The sediment sorption capacity of organic contaminants (such as sulfonamide antibiotics) depends on several variables, including the intrinsic properties of the contaminant and the physical and chemical properties of the water-sediment system (Luo et al., 2019; Huang et al., 2020). In our experimental conditions, SMZ accumulated more in sediment than SMX (Figure 1). The maximum concentrations of SMZ accumulated in sediments were significantly higher than the maximum concentrations of SMX (t-test, p < 0.05). Indeed in “low” conditions, the maximal concentration in SMZ (1.03 ± 0.11 μg kg^–1 dw at D21; Figure 1B) was 7-fold higher than in SMX (0.14 ± 0.05 μg kg^–1 dw at D21; Figure 1A), in “high” conditions, the maximal SMZ concentration (6.20 ± 0.32 μg kg^–1 dw at D21; Figure 1D) was 2-fold higher than the maximal SMX concentration (2.89 ± 0.16 μg kg^–1 dw at D28; Figure 1C). This result is in line with those of Zhong et al. (2013), who with two types of lake sediments reported a higher sorption capacity of SMZ than of SMX. However, at both low and high treatments, SMZ concentrations increased exponentially in sediments from D0 to D21 before decreasing from about 18% (high-SMZ, Figure 1D) and 36% (low-SMZ, Figure 1B) between D21 and D28. Two non-exclusive hypotheses can be proposed to explain the observed decrease in the sediment concentration of SMZ at the end of the experiment. The first one is release of the tested contaminant from the sediment to the overlying water (i.e., desorption process; Chen et al., 2014). The second is photodegradation and/or biodegradation processes leading to the dissipation and/or transformation of the studied antibiotics (Carstens et al., 2013; Li and Cui, 2020). These processes could be more pronounced at the end of the experiment owing to the adaptation of some sediment microorganisms after chronic exposure to the tested sulfonamide, which could enhance their ability to co-metabolize and biodegrade this antibiotic (Ricken et al., 2013; Topp et al., 2013; Woappi et al., 2016; Martin-Laurent et al., 2019).

Based on the concentrations measured weekly from D7 to D28, the mean levels of exposure of sediment microbial communities to SMX and SMZ were respectively, 2.10 ± 0.86 μg SMX kg^–1 dw and 4.53 ± 1.95 SMZ μg kg^–1 dw in the high treatment and 0.09 ± 0.05 μg SMX kg^–1 dw and 0.79 ± 0.19 SMZ μg kg^–1 dw in the low treatment. Based on the existing literature, the high-treatment conditions can be considered as relatively high, but were relevant environmental concentrations, significantly lower than maximum concentrations recently reported in stream and river sediments (8 μg kg^–1 for SMX and 248 μg kg^–1 for SMZ; De Vries and Zhang, 2016, and references therein) or estuarine sediments (100 μg kg^–1 for SMX and 20 μg kg^–1 for SMZ; Zheng et al., 2021). Similarly, the low-treatment conditions were representative of low environmental concentrations, the measured values being significantly lower than the mean and median values reported in the studies cited above (e.g., 0.52 μg SMX kg^–1 and 2.87 μg SMZ kg^–1 in stream and river sediments; De Vries and Zhang, 2016, and references therein).

The time course of the functional response of sediment microbial communities to chronic exposure to environmental concentrations of SMX or SMZ was assessed by weekly measurement of five potential activities respectively, contributing to C (β-glucosidase and aerobic respiration), N (leucine-aminopeptidase and denitrification), and P (phosphatase) cycles (Figure 2). Most of the microbial activities increased over time, reflecting microbial growth and activity. As previously reported in several studies assessing the effects of SMX and SMZ on aquatic microbial biofilms (Yergeau et al., 2010, 2012; Kergoat et al., 2021; Paumelle et al., 2021), the functional responses observed in the present experiment varied markedly according to the antibiotic tested.

The SMX treatments significantly affected the two activities involved in the C cycle: β-glucosidase (Figure 2A) and aerobic respiration (Figure 2E). In sediments exposed to the high-SMX treatment, β-glucosidase activity was significantly higher than in the control throughout the experiment (p < 0.05 from D7 to D28, Figure 2A) whereas this activity was similar to controls in sediments exposed to low concentrations of SMX. In addition, a positive correlation (Pearson correlation coefficient r = 0.733, p < 0.05) was found between glucosidase activity and SMX concentrations in sediments. A similar increase in β-glucosidase activity was already reported by Paumelle et al. (2021) in leaf litter-associated microbial communities exposed for 16 days to a nominal concentration of 5 μg L^–1 of SMX or SMZ with a difference in the magnitude of the observed response according to the presence or absence of periphytic communities. They hypothesized that the β-glucosidase increase was due to the use of an external C supply in the antibiotic treatments and/or to an adaptative response to resist antibiotics (Paumelle et al., 2021). By contrast, Kergoat et al. (2021) observed no change in the β-glucosidase activity of periphytic biofilms exposed for 28 days to SMX concentrations between 0.2 and 4 μg L^–1. Moreover, Liu et al. (2016) observed a negative effect of SMX on the β-glucosidase activity of soil microbial communities, but the concentrations they used (5 and 50 mg SMX kg^–1 dw soil) were far higher than the environmental concentrations tested here.

On contrary to SMX, SMZ exposure only had slight effect on β-glucosidase activity and respiration (Figures 2A,E). Thus β-glucosidase activity remained similar to controls in sediments exposed to low concentrations of SMZ whereas exposure to high concentration of SMZ led to β-glucosidase activities lower than controls after 21 days of exposure (Figure 2A). To our knowledge, this is the 1st report of β-glucosidase inhibition by exposure to environmentally relevant concentrations of SMZ. This inhibition may indicate a higher toxicity of SMZ than SMX or a higher exposure of sediment microbial communities to SMZ than SMX. Indeed, the higher concentration of SMZ than SMX indicated that microbial communities attached to sediments are likely to be more exposed to SMZ than to SMX. Nevertheless, further investigations are needed to determine whether these differences in toxicity are linked to different modes of action or to different behavior within the aquatic ecosystem.

In the present experiment, the chronic exposure to SMX also tended to stimulate the aerobic respiration of sediment communities (Figure 2E). It was thus positively correlated to SMX concentrations in sediment exposed to the high-SMX condition (r = 0.726, p < 0.05). In addition, after 28 days of exposure to the low-SMX condition, aerobic respiration was higher than in controls. Given the marked decrease in SMX concentrations (about 49%) observed in the low-SMX treatment between D21 and D28 (Figure 1A), it can be hypothesized that the increased production of CO₂ observed at D28 could be at least partly due to the mineralization of the antibiotic. However, this hypothesis must be taken with caution, since no significant increase in aerobic respiration was observed at D28 in the high-SMZ and low-SMZ treatments (Figure 2E), although sediment SMZ concentrations significantly decreased in both treatments between D21 and D28 (Figures 1B,D).

The contamination of sediment by SMZ affected all the measured activities positively or negatively (Figure 2), but the effects were only transient and varied according to the type of activity and the antibiotic concentration, as previously reported by Kergoat et al. (2021) with periphytic biofilms. Surprisingly, the effects of the high-SMZ treatment on β-glucosidase were opposite to those observed here in the high-SMX treatment (Figure 2A) and to those previously reported by Paumelle et al. (2021) with leaf-litter microbial communities chronically exposed to SMZ. In the present study (Figure 2A), β-glucosidase activity from sediments exposed to high concentration of SMZ was about 20% lower than control levels at D21 (p < 0.05) and D28 (not significant). In sediments under high-SMZ exposure, the phosphatase activity (Figure 2C) was also lower than the control level, especially at D7 (about 20% lower, p < 0.05) and D28 (about 30% lower, p < 0.05). This last observation is in line with Kergoat et al. (2021), who found that the phosphatase activity of periphytic biofilms could be sensitive to chronic exposure to SMZ. Yang et al. (2020) also observed a significant inhibition of this activity in soil exposed for 28 days to SMZ concentrations ranging from about 14 to 223 mg SMZ kg^–1 dw soil. Whereas β-glucosidase and phosphatase activities (Figures 2A,C) were not sensitive to low levels of SMZ (<4 μg SMZ kg^–1 dw), leucine aminopeptidase activity (Figure 2B) was found to be transiently stimulated at both low and high levels of SMZ exposure after 14 days of exposure (p < 0.05). This transient microbial response had also been observed previously by Kergoat et al. (2021) in periphytic biofilms exposed to mean SMZ concentrations ranging from about 106 to 157 ng SMZ L^–1. To our best knowledge, there is no other published study investigating the chronic effect of SMZ on the leucine aminopeptidase activity. However, there is evidence that this antibiotic can impact the N-cycle in different environmental compartments including soil (D’Alessio et al., 2019) and sediment (Hou et al., 2015), through changes in denitrification processes. This was also the case here (Figure 2D), where we observed a slight but significant transient inhibition of denitrification in the high-SMZ treatment at D7 (p < 0.05) and a significant transient stimulation of this metabolic activity in the low-SMZ treatment at D14 and D21 (p < 0.05).