Pharmaceuticals and personal care products have been recognized as a growing environmental concern since the late 1990s 1 for a number of reasons. They are designed to perform biological functions 2 and have been detected in various environmental compartments, including surface waters 3, groundwater 4, drinking water 5, and municipal wastewater effluent 6. The aquatic environment receives continuous inputs of pharmaceuticals and personal care products via municipal/industrial wastewater effluents and urban/agricultural runoff, exposing aquatic ecosystems to long-term, low-level contamination and potentially causing toxicity to nontarget organisms and the development of antibiotic-resistant microorganisms 7. Despite the potential for pharmaceuticals and personal care products to cause effects and their presence and persistence in aquatic systems, limited toxicological data are available for nontarget organisms 7.
Sulfonamide antibiotics (SAs) are one of the most commonly prescribed groups of antibiotics globally in both human and veterinary medicine. They are routinely detected in municipal wastewater effluent and surface waters in the low microgram per liter range (for individual SAs), and it is common to find multiple SAs co-occurring in environmental samples 8–10. Despite the widespread use and environmental occurrence of SAs, little toxicological information exists with which to conduct environmental risk assessments for these compounds 11. In particular, freshwater invertebrate species are underrepresented in the published literature. Several studies have been published on the toxicity of a variety of SAs to freshwater invertebrates, but only two species have been tested in these studies: Daphnia magna and Ceriodaphnia dubia12–20.
Therefore, the objective of the present study was to assess the aquatic toxicity of four SAs to the survival and growth of Hyalella azteca, an epibenthic amphipod ubiquitous in North American freshwater ecosystems, sensitive to a wide variety of compounds, and extensively used in toxicity tests 21, 22. The test compounds selected were sulfaguanidine (SG), sulfathiazole (ST), sulfamerazine (SM), and sulfasalazine (SM), which were chosen to span a range of chemical structures and properties (Table 1) with the expectation that such information would inform environmental risk assessments for this class of compounds. To our knowledge, this is the first research examining the toxicity of SAs to freshwater amphipods.
Table 1. Physicochemical properties of four sulfonamide antibiotics used in chronic (four-week) toxicity tests with Hyalella azteca
Hyalella azteca were cultured in 2-L polypropylene containers according to the methods described in Borgmann et al. 21. Cultures and experiments were maintained at 25°C with a photoperiod of 16 h light to 8 h dark, and amphipods were fed finely ground Tetra-Min (Ulrich Baensch) fish food flakes sifted through a 500-µm screen. Cultures and experiments were maintained in dechlorinated city tap water from Burlington, Ontario, Canada, (originating from Lake Ontario, Canada; hardness 130 mg/L, alkalinity 90 mg/L, pH 7.9–8.6). Juvenile amphipods were removed from the breeding containers weekly and used in toxicity tests initiated 2 to 3 d later (i.e., amphipods were 2–10 d old at test initiation).
The four SAs (SG, ST, SM, and SS) were purchased from Sigma-Aldrich Chemicals (purity of each compound was >98%); chemical structures are shown in Table 1. Stock solutions of each SA for use in toxicity testing were prepared in high-performance liquid chromatography–grade methanol (MeOH), purchased from Caledon Laboratories and stored at −20°C when not in use.
Four-week static-renewal, water-only toxicity tests were conducted to assess the effects of the four SAs on survival and growth of Hyalella. Nominal SA test concentration ranges were based on the results of one-week range-finding tests (data not shown) as follows: 0.14 to 19 µM for SG, 0.31 to 39 µM for ST, 0.30 to 38 µM for SM, and 0.10 to 13 µM for SS. All tests were conducted in a geometric series with a factor of 2. Each test also included a negative control and a solvent (MeOH) control; total MeOH levels in each experiment were maintained at a constant level in all test containers except the negative controls (<0.5% for all tests). Tests were initiated by adding 2.5 mg of Tetra-Min, appropriate volumes of the SA stock solution and MeOH, one square of 5 × 5 cm cotton gauze presoaked in ultrapure (Thermo Scientific Barnstead NANOpure Water Purification System) water, 400 ml of water, and 20 juvenile amphipods (2–10 d old) to 500-ml high-density polyethylene wide-mouthed containers. Hyalella were removed from test containers weekly, counted, and transferred to fresh test containers to minimize accumulation of uneaten food and waste products (e.g., ammonia) and to allow for weekly monitoring of survival during the test. Amphipods were fed Tetra-min as follows: 2.5 mg twice in weeks 1 and 2, 2.5 mg three times in week 3, and 5 mg three times in week 4. At the end of the four-week exposure, amphipods from each replicate were removed, counted, and weighed as a group to obtain wet weights. A minimum of two tests were conducted for each compound, with each test consisting of three replicates per control and two replicates per SA concentration; an additional test was conducted for ST due to high control mortality in week 4 of one test, and four additional tests were conducted for SM in an attempt to reduce the variability among tests.
Water-quality parameters, including dissolved oxygen, pH, specific conductivity, and total ammonia (NH3/NH), were measured at the start and end of every turnover. Mean dissolved oxygen concentration was 7.7 mg/L (standard deviation [SD] = 1.0, range 4.7–10 mg/L), mean pH was 8.3 (SD = 0.29, range 7.1–9.4), mean specific conductivity was 340 µS/cm (SD = 25, range 260–480 µS/cm), and total ammonia levels (NH3/NH) were always below 0.03 mM.
Water samples (10 ml) were collected from each concentration at the beginning and end of every weekly turnover, and all samples were stored at 4°C in the dark pending chemical analysis. Concentrations of SG and ST were determined in all water samples collected from each experiment. Because of the number of tests conducted for SM (six in total), a subset of water samples was analyzed to reduce the time and cost required. All samples were analyzed from one experiment, and for the remaining five SM experiments all week 1 samples were analyzed, but only one tracking concentration (4.7 µM nominal, chosen to approximate the four-week lethal concentration resulting in 50% mortality [LC50]) was analyzed from weeks 2, 3, and 4 to monitor weekly changes in concentration. For SS, one tracking concentration (13 µM nominal) was analyzed for each experiment; this was the highest concentration tested and was chosen because no effects were observed in SS toxicity tests (as described in Results).
Acetonitrile (CH3CN, high-performance liquid chromatography grade) and ammonium acetate (NH4[OAc]) were purchased from Caledon Laboratories. A stock solution containing all four SAs, each at a concentration of 10 mg/L, was prepared in 1:1 CH3CN:MeOH. Standard solutions were prepared by diluting the stock solution with 20 mM NH4(OAc). Isotopically labeled 13C6-sulfamethazine was purchased from Cambridge Isotope Laboratories. Deionized water was obtained using a Milli-Q (Millipore) system.
As described in Balakrishnan et al. 23, all samples were analyzed using a Quattro Ultima tandem liquid chromatograph triple quadrupole mass spectrometer (MS; Micromass) equipped with a Z-Spray electrospray ionization source and operated in the positive-ion mode. Nitrogen was used as the drying and nebulizing gas at flow rates of 500 and 70 L/h, respectively. Collision-induced dissociations were conducted using 2.5 × 10−3 mbar argon in a hexapole collision cell. MassLynx software (Version 4.1; Waters) was utilized for both data acquisition and processing. The MS was operated in multiple reaction monitoring mode, with a dwell time of 100 ms per ion pair and an interchannel delay of 50 ms. The optimized electrospray ionization MS/MS conditions for analysis of SAs are summarized in Table 2. The MS apparatus was attached to an Acquity UPLC system (Waters).
Table 2. Optimized electrospray ionization-mass spectrometer/mass spectrometer conditions for the analysis of sulfonamide antibioticsa
Parent ion, [M + H]+ (m/z)
Daughter ion (m/z)
Cone voltage (kV)
Collision energy (eV)
Optimal capillary voltage in each case was 3.5 kV.
Prior to sample injection, 13C6-sulfamethazine (10 mg/L, prepared in 20 mM NH4OAc) was added as an internal standard, such that the final concentration of 13C6-sulfamethazine in all samples was 1.67 mg/L. Ionization parameters for 13C6-sulfamethazine are given in Table 2.
Aliquots of 10 µl were injected onto a 5-µm-pore size XTerra MS-C18 column (2.1 × 250 mm; Waters) at 35°C. Aqueous ammonium acetate (20 mM, 0.1% formic acid, pH 3; A) and a 2:1 acetonitrile:methanol solution (containing 20 mM ammonium acetate, B) were used as mobile-phase solvents for gradient elution at a flow rate of 200 µL/min. The gradient was increased from 2 to 20% B within 2 min and then increased again to 70% B within 15 min before being ramped to 100% B by 18 min. The gradient was held at 100% B for 5 min, whereupon the initial conditions (2% B) were reestablished and held for 12 min to ensure minimal carryover between injections. The source temperature was set to 120°C, while that for desolvation was set to 350°C.
All statistical analyses for toxicity tests were performed using SYSTAT 12 for Windows (SYSTAT Software). Statistical analyses for SG and ST were conducted using measured concentrations averaged (using arithmetic means) over the total duration of the exposure (e.g., week 1 concentrations are averaged from those measured at the start and end of week 1, week 2 concentrations are averaged from those measured at the start and end of weeks 1 and 2, etc.). Data for SM were also reported as measured concentrations, but because only a subset of samples was analyzed (as described above), a linear regression on log-transformed data was used to determine the relationship between nominal and measured concentrations for SM (Fig. 1) to obtain estimates of SM toxicity based on measured concentrations. Data were pooled for the regression analysis because there were no detectable differences among individual experiments: 95% confidence intervals (CIs) overlapped for y intercepts and slopes of linear regressions performed on week 1 data, and tracking concentrations were not different (analysis of variance, p = 0.11). Nominal concentrations were used to express SS results graphically as only the highest test concentration was measured.
Statistical analysis of survival and growth data followed procedures described in Bartlett et al. 24. In total, two experiments each were conducted for SG and SS, three experiments for ST, and six for SM. Data were excluded from statistical analysis if control survival did not meet the performance criteria (65% survival in a four-week test or <10% mortality per week) described by Borgmann 25. One experiment each from ST and SM were eliminated from week 4 calculations due to poor control survival.
Survival data from chronic toxicity tests were used to calculate mortality rates (m) for each week using Equation 1
where t is time, Nfinal is the number of animals surviving at t, and Ninitial is the number of animals at t = 0. If mortality was 100% in one of the replicates at the end of each week, 0.5 animal was assumed to have survived for the purpose of calculating m. This was only done for the lowest concentration resulting in 100% mortality. Mortality rates were fourth-root–transformed and then fitted to the nonlinear regression model described by Equation 2
where m′ is control mortality, CSA is the measured concentration of SA in water (micromoles per liter), and a and n are constants. The LC25 (lethal concentration resulting in 25% mortality) for each week was calculated using Equation 3
Model parameter estimates, 95% CIs, and r2 values were provided by SYSTAT. Control mortality rates and values of n are summarized in Supplemental Data (Table S1).
Growth of amphipods in chronic toxicity tests was measured as individual growth at the end of week 4 (wet weight/Nfinal), and data were logarithmically transformed prior to statistical analysis. Growth (G) was then fitted to the nonlinear regression model described by Equation 4
where max is the maximum mass at CSA = 0 and effective concentration associated with 50% reduction in growth (EC50) is the concentration at G = 0.5 max. The values of max and n are summarized in Supplemental Data (Supplemental Data, Table S1). The effective concentration causing a 25% reduction in G (EC25) for growth was calculated using Equation 5
Model parameter estimates, 95% CIs, and r2 values were provided by SYSTAT.
Initial estimates of toxicity were obtained from individual experiments after which the data for each SA were pooled for statistical analysis if the 95% CIs around each corresponding variable in the models overlapped between experiments and if the data were visually indistinguishable between experiments. Individual SM experiments were excluded from statistical analysis in cases where LC50s/EC50s could not be calculated (LC50s/EC50s for these experiments were reported as greater than the highest test concentration), and toxicity end points for week 3 were reported for pooled data (three experiments) as well as two experiments individually because the 95% CIs around the model variables failed to overlap.
Measured SA concentrations in water ranged from 88 (ST) to 110% (SS) of nominal concentrations in t = 0 samples. Levels decreased over one-week static-renewal periods for SG, ST, and SM, dropping approximately 33, 39, and 17%, respectively, whereas SS concentrations remained stable (Table 3). Concentrations of all SAs were, as expected, below method detection limits for all control and solvent control samples.
Table 3. Measured concentrations of sulfonamides during four-week static-renewal water-only toxicity tests with Hyalella azteca at the beginning (t = 0) and end (t = 7) of each weekly turnover
Measured concentration (% nominal percentage)
SD = standard deviation; Min = minimum; Max = maximum; n = number of measurements.
t = 0
t = 7
t = 0
t = 7
t = 0
t = 7
t = 0
t = 7
Sulfaguanidine had a strong effect on amphipod survival, and LC50s decreased almost threefold from week 1 (2.6 µM) to week 4 (0.90 µM) (Table 4). Mortality rate models described the data well: r2 values for weeks 1, 2, and 3 were 0.74, 0.76, and 0.72, respectively (data not shown); increased variability in controls and in lower test concentrations resulted in a lower r2 value for week 4 (0.53) (Fig. 2A). Chronic (four-week) survival remained stable until SG concentrations reached 0.47 µM, then dropped rapidly to 0% at 1.9 µM (Fig. 2A). Sulfaguanidine had a negligible effect on growth (r2 = 0.21) (Fig. 3A).
Table 4. Chronic toxicity of sulfonamide antibiotics to Hyalella azteca in four-week static-renewal water-only exposures
Two experiments conducted in total. Data pooled from two experiments for all toxicity calculations.
Toxicity estimates for survival (lethal concentration associated with 25% [LC25] and 50% [LC50] mortality) and growth (effective concentrations associated with 25% [EC25] and 50% [EC50] reduction in growth) are expressed as measured sulfonamide concentrations (micromoles per litre); numbers in parentheses are 95% confidence intervals.
Three experiments conducted in total. Data pooled from three experiments for weeks 1 to 3 and from two experiments for week 4; one experiment was eliminated from week 4 calculations due to high control mortality.
Model was able to fit data even though <50% mortality occurred; therefore, the calculated LC50 was greater than the highest test concentration.
Six experiments conducted in total.
Data pooled from two experiments.
LC25s and LC50s could not be calculated from four experiments because insufficient mortality occurred.
Data pooled from three experiments.
Individual LC25s and LC50s reported for two experiments because 95% confidence intervals for model parameters did not overlap with all experiments.
LC25s and LC50s could not be calculated from one experiment because insufficient mortality occurred.
Survival data pooled from five experiments; one experiment eliminated from calculations due to high control mortality.
Growth data pooled from two experiments.
EC25s and EC50s could not be calculated from three experiments because insufficient growth inhibition occurred; one experiment eliminated from calculations due to high control mortality.
No effects on survival or growth at any tested concentration
Sulfathiazole effects occurred more slowly and at higher concentrations than those of SG. Survival in ST exposures was greater than 50% during week 1; therefore, the LC50 exceeded the highest test concentration of 28 µM (extrapolated LC50 = 49 µM); LC50s decreased from 4.3 to 1.6 µM from weeks 2 to 4 (Table 4). The fit of the mortality rate model was low for week 1 (r2 = 0.37) but described the data well for weeks 2 to 4 when effects on survival were more pronounced: r2 values were 0.60 and 0.62 for weeks 2 and 3, respectively (data not shown), and 0.61 for week 4 (Fig. 2B). Toxic effects were observed over a much wider range for ST than SG, with survival at four weeks remaining stable until 0.80 µM and then decreasing gradually to almost complete mortality at 28 µM. Differences between SG and ST were most pronounced at week 1, when the ST LC50 was 19-fold higher than that for SG. These differences became less pronounced over time, and the four-week LC50 for ST was just twofold higher than that for SG (Table 4). Sulfathiazole was more strongly associated with growth effects than was SG (r2 = 0.56) (Fig. 3B), but growth was a less sensitive measure of toxicity than survival: the EC50 was eightfold higher than the LC50 (Table 4).
Sulfamerazine toxicity was similar to that of ST but generally lower than that of SG (Table 4). As observed with SG and ST, LC50s for SM decreased over time from ≥22 µM at week 1 to 3.9 µM at week 4 (Table 4); a more definitive estimate of the one-week LC50 was not possible since four out of the six tests did not yield sufficient mortality to calculate LC50s. The r2 values for the mortality rate models were 0.56, 0.62, 0.53 to 0.79, and 0.44 for weeks 1, 2, 3, and 4, respectively. Survival data were quite variable among experiments, as demonstrated by week 3 LC50s, which ranged 10-fold from 1.9 to 19 µM (Table 4), and the amount of scatter in the data by week 4, which resulted in a low r2 value of 0.44 (Fig. 2C). Variability in SM data was not improved by conducting additional experiments (six in total), for reasons that remain unclear. No consistent effects of SM on growth were observed. Growth data were highly variable, EC50s could be calculated for only two experiments, and even in these experiments there was a weak relationship of SM with growth (r2 = 0.33) (Fig. 3C and Table 4).
Contrary to the results from other SAs, SS had no effects on survival or growth compared to controls at any tested concentration (Figs. 2D and 3D); the maximum concentration tested was 13 µM (measured). We were unable to test higher concentrations due to limits of the solubility of SS in MeOH.
The relative toxicity of SAs determined from effects on survival of Hyalella in chronic toxicity tests, in order of most to least toxic, was SG > ST ∼ SM > SS; and effects on survival were observed earlier (week 1) and at lower concentrations in SG exposures than with other SAs. However, it should be noted that differences among SG, ST, and SM were more pronounced from weeks 1 to 3; by week 4, LC50s differed by fourfold or less, although LC50s were lowest for SG and highest for SM (Table 4). Sulfasalazine was the only SA tested that demonstrated no effect on survival at concentrations up to 13 µM (Fig. 3D) and clearly differed from the other SAs tested.
No effects of SAs on growth were observed at or below the LC50 for any of the compounds tested (Table 4), indicating that growth was a less sensitive measure of toxicity than survival. The strongest effect on growth occurred with ST (Fig. 3B), negligible growth effects were seen for SG and SM (Fig. 3A and C), and no growth effects were observed for SS (Fig. 3D). Similar results have been observed for Hyalella exposed to inorganic (cadmium, copper, lead, mercury, zinc) and organic (pentachlorophenol, polychlorinated biphenyls, tributyltin) compounds, where effects on growth occurred at concentrations similar to, or higher than, survival 21, 24, 26, 27.
As noted previously, the toxicological data from the present study were quite variable, particularly for SM. Additional tests were conducted in an attempt to provide more conclusive results regarding the effects of SM on survival and growth, but this attempt was unsuccessful (Fig. 2C). Some variability in Hyalella toxicity tests is not unusual. Borgmann et al. 21 measured a continuous mortality rate of approximately 12% per week in Hyalella cultures; therefore, it is not surprising to observe 40% mortality in controls and low-effect test concentrations by the end of a four-week exposure. Although the variability of some of the data was not ideal, mean control survival in each SM test did exceed the recommended performance criteria of 65% 25. The authors can offer no explanation for the variability in the SM data. Water-quality parameters measured during the tests were within the desired range, and no obvious sources of experimental error were recorded. While the variability of the data precludes making definitive conclusions regarding the toxicity of SAs to Hyalella, the present study does provide important baseline data for an initial assessment of the effects of SAs on survival and growth of Hyalella, as well as their relative toxicity.
Although physicochemical differences among SAs, such as molecular size and structure as well as water solubility, may have contributed to observed differences in toxicity to Hyalella (Table 1), the major factor influencing bioavailability, bioaccumulation, and toxicity of ionizable compounds, such as SAs, is ionization state 28, 29. Neutral compounds are generally more bioavailable because they are more lipophilic and can diffuse across cell membranes more quickly and easily than their corresponding ionized forms 30, 31. Because the SAs in the present study have different pKa values (Table 1), they will also have different ionization states at experimental pH (mean = 8.3, range 7.1–9.4), which will ultimately influence their bioavailability.
The Henderson–Hasselbalch equation was used to calculate the percentage of neutral and ionized SA that should occur within the range of experimental pH values and the range of pKa values obtained from the literature (Table 1). For the experimental pH range measured in the present study, the neutral form should be 98 to 100% of total SG, 0.47 to 72% of total ST (5.6–14% at pH 8.3), and 0.23 to 53% of total SM (2.9–6.6% at pH 8.3); SS should always be anionic, primarily (67–100%) with a charge of −1, but up to 33% of total SS should have a charge of −2 (3.8% at pH 8.3).
The ranking of SAs, in terms of their relative neutrality as a proportion of total SA present at experimental pH, was SG > ST ∼ SM > SS. This corresponds directly with the observed toxicity of SAs, with both the onset and magnitude of toxicity increasing with proportion of neutral SA present. Toxicity occurred most rapidly and at the lowest concentration for SG, which was present primarily in its neutral form. In contrast, SS did not exhibit any toxic effects within the concentration range and exposure duration tested and was present solely in ionic form (charge of –1 to –2 depending on pH).
Assuming that the mode of toxic action is similar among SAs, differences in their ionization state could also explain why LC50s for these compounds converged by week 4. In theory, the bioaccumulation (and, hence, toxicity) of ST and SM would be delayed because a lower proportion of these compounds would be present in neutral form, resulting in slower rates of bioaccumulation, but they would have similar toxicity to SG once sufficient uptake had occurred. Unfortunately, bioaccumulation of SAs was not measured in the present study because the analytical methodology was not yet sufficiently advanced to measure SAs in such small tissue samples (i.e., on the order of milligrams dry weight) at low concentrations (i.e., on the order of micrograms per gram). The fact that SA toxicity in the present study was based on environmental (i.e., water) concentrations instead of bioaccumulation may have contributed, in part, to the variability of the data, as it has been established that bioaccumulation of contaminants is a better predictor of toxicity than environmental concentrations 24. Bioaccumulation measures only the bioavailable fraction of the contaminant and takes into account the external factors that control chemical speciation and uptake, whereas basing toxicity on environmental concentrations includes these external sources of variability, such as variations in pH, which are important for the bioavailability of SAs.
Numerous studies have reported that the toxicity of ionizable compounds is pH-dependent and directly related to the proportion of the compound present in its un-ionized form. Valenti et al. 31 examined the toxicity of sertraline (pKa = 9.47) to the fathead minnow (Pimephales promelas) at pH 6.5, 7.5, and 8.5 (i.e., 5, 12, and 28% of total sertraline in nonionic form, respectively) and found that toxicity increased (48-h LC50s were 647, 205, and 72 µg/L, respectively) and that manifestation of toxicity occurred more quickly (LT50s at 500 µg/L were > 48, 32, and 5 h, respectively) as pH (and the concentration of the nonionic form of sertraline) increased. Nakamura et al. 32 reported similar pH-dependent toxicological relationships during experiments in which Japanese medaka (Oryzias latipes) were exposed to fluoxetine (pKa = 10.1) at pH 7, 8, and 9: 96-h LC50s were 5.5, 1.3, and 0.2 mg/L, respectively, and toxicity showed a positive correlation with the un-ionized form of fluoxetine. Fent and Looser 33 reported that uptake rates, bioaccumulation, and mortality of Daphnia magna exposed to tributyltin chloride (pKa = 6.51) are significantly higher at pH 8, where the undissociated neutral species predominates, than at pH 6, where the dissociated cationic species predominates, and concluded that the neutral form can penetrate biomembranes much more easily than the charged, hydrophilic cation.
The observation of pH-dependent toxicity for monovalent ions also applies to divalent ions. Rendal et al. 34 observed increasing toxicity of chloroquine (pKa = 6.33, 10.47) to the water flea (D. magna) with increasing pH as the proportion of the neutral form of chloroquine increased. In the same study, the effects of pH on the toxicity (measured as transpiration rate) and bioavailability (measured as root concentration factors) of chloroquine to the basket willow (Salix viminalis) were examined. Toxicity was 10-fold higher, and bioavailability was fourfold to sevenfold higher at pH 9 than at pH 6 34.
The sensitivity of Hyalella to SAs is comparable to or greater than that of other freshwater crustaceans. With the exception of SS, the acute toxicity (one-week exposures) of SAs to Hyalella in the present study occurred between 2.6 and 49 µM, and chronic toxicity (four-week exposures) occurred above 0.36 µM, with more severe effects observed between 0.9 and 12 µM (Table 4 and Fig. 2). To our knowledge, this is the first time the toxicological effects of SAs have been studied in Hyalella. Previous research on the toxicity of SAs to freshwater invertebrates has been conducted primarily on two crustaceans, Ceriodaphnia dubia and D. magna (Table 5); acute toxicity (24–96 h) ranges between 18 and 2,060 µM (nominal), and chronic toxicity (7–21 d) ranges between 0.83 and 55 µM (nominal) 12–20. When comparing the toxicity of the specific SAs tested in our study, the acute effects of SG, ST, and SM occurred at concentrations sevenfold, more than sixfold, and up to 39-fold lower, respectively, in Hyalella than in Daphnia, while the chronic effects of SG occurred at fourfold lower concentrations (Tables 4 and 5). We were unable to find any publications that examined the aquatic toxicity of SS. The difference in sensitivity among these three crustaceans illustrates the importance of obtaining toxicity data on a variety of species representing a range of ecological niches to ensure that susceptible species or groups are adequately protected.
Table 5. Acute and chronic effects of sulfonamide antibiotics to crustaceans
Concentrations of SAs measured in environmental waters are generally much lower than those causing toxicity to Hyalella in the present study. In a survey of the published literature (Supplemental Data, Table S2, and references therein), maximum SA concentrations ranged from 0.0090 to 12 µg/L (0.032–48 nM) in surface waters, 0.0060 to 1,600 µg/L (0.023–7,500 nM) in groundwater, 22 to 400 µg/L (82–1,400 nM) in agricultural waters, and 0.16 to 3.0 µg/L (0.61–12 nM) in wastewater samples. Environmental SA concentrations exceeded those expected to cause chronic toxicity to Hyalella (i.e., 360 nM) in only a small number of localized areas, such as immediately downstream of a landfill containing waste from pharmaceutical production 35 or in wastewater from large swine operations 36, 37.
However, SAs remain an environmental concern because their potential impacts on the aquatic environment may be underestimated by laboratory exposures. The majority of toxicity tests conducted with SAs are short in duration (i.e., on the order of hours, days, or weeks) and measure acute end points (i.e., lethality, immobility); these tests may not accurately assess environmental impacts because aquatic ecosystems receive continuous inputs of these compounds 7 and, therefore, environmental exposures occur over an extended period of time. Chronic effects in crustaceans also occur at lower concentrations than acute effects: LC50s for Hyalella decreased from weeks 1 to 4 in the present study, and EC50s for reproduction occurred at concentrations an order of magnitude lower than EC50s for immobility in Daphnia exposed to SAs 12, 13. Therefore, long-term, low-level exposures such as those that occur in aquatic ecosystems could affect sensitive chronic end points at environmentally relevant concentrations. Another concern is that toxicity tests conducted with individual SAs do not account for mixture interactions. Sulfonamide antibiotics are generally found in environmental samples as chemical mixtures with other SAs and pharmaceuticals and personal care products, which could result in synergistic, additive, or antagonistic effects 12, 13, 38. Additional considerations in the assessment of environmental impacts of SAs include environmental conditions affecting toxicity, such as ultraviolet light 16, 18 and pH 30, fate and toxicity of SA metabolites 8, 39, and aquatic sediments as sources and/or sinks of SAs 40, 41.
Despite the limitations of laboratory tests in predicting environmental impacts, the toxicity data collected from the present study are an important first step in understanding the potential effects of SAs in the aquatic environment and form the basis for designing more environmentally relevant investigations to further assess the risk of SAs to aquatic ecosystems.
The relative toxicity of SAs to Hyalella, from most to least toxic, was SG > ST ∼ SM > SS. Differences in the onset and magnitude of toxicity among SAs were likely due to differences in pKa values and the proportion of each SA present in its neutral versus ionized form. Survival was affected by SG, ST, and SM, but only ST had an observable effect on growth. Sulfasalazine caused no detectable toxicity up to the highest concentration tested. Sulfonamide toxicity to Hyalella in the present study occurred at lower concentrations than those affecting two freshwater crustacean species but at concentrations well above those typically found in the environment. Further research on SAs is required to provide information on the effects on other invertebrate species, the effects of low-level, long-term exposures on sensitive chronic end points (e.g., reproduction), the effects of chemical interactions within mixtures on toxicity, and the effects of environmental conditions (e.g., ultraviolet light, pH, sediment) on toxicity. However, the present study provides important baseline data for risk assessment of SAs, and the fact that LC50s decreased over time indicates that long-term tests may be required to provide an environmentally relevant assessment for these compounds.
Tables S1 and S2. (201 KB DOC).
The authors gratefully acknowledge the contributions of the following people: S. Dupré, Y. Gu, A. Hesch, T. Kay, and J. Wong. Funding was provided by the New Substances Assessment and Control Bureau of Health Canada. Two anonymous reviewers provided valuable contributions to an earlier version of the manuscript.