An analysis of lethal and sublethal interactions among type I and type II pyrethroid pesticide mixtures using standard Hyalella azteca water column toxicity tests

INTRODUCTION

Pyrethroid insecticides are commonly detected at toxic concentrations in the surface water and sediment of water bodies worldwide 1-3. These pesticides are applied to urban and agricultural landscapes, and their use has been progressively increasing over the past 20 yr 4. Many pyrethroids are classified as contaminants of emerging concern because of their broad-spectrum toxicities, and their persistence and prevalence in the environment 5. These chemicals are lethal to multiple nontarget aquatic invertebrates at low parts per billion concentrations, and cause chronic and sublethal effects at fractions of those levels 6. Multiple field studies have implicated pyrethroids in environmental toxicity and ecological degradation 7, 8, although sediment concentrations of pyrethroids are low or moderate in some areas, and not always toxic to Hyalella 9, 10, possibly because of different regional use patterns. Where toxicity does occur, bifenthrin is often implicated as the major contributor in urban areas 11-14, and in other areas multiple pyrethroids (i.e., mixtures) are often the culprit 7, 11, 12. Regulations on maximum allowable concentrations in the environment are important for the ecological health of water bodies that receive urban or agricultural discharge. Fish populations worldwide are in decline 15-18, and pesticides are one of the many stressors to which fish are exposed. Although few studies have implicated direct effects of pesticides on fish mortality in nature, effects on prey abundance may have a significant impact on fish populations.

The present study investigates the interactive or additive effects of 4 type I and type II pyrethroid insecticides on an ecologically important epibenthic amphipod, H. azteca. Resident to streams and estuaries throughout the United States, the genus Hyalella includes many species that are important prey to native fish, birds, and amphibians. Hyalella azteca is certified by the US Environmental Protection Agency (USEPA) for sediment (10-d) and water reference (4-d) toxicity testing 19. Although it is particularly sensitive to pyrethroid pesticides, H. azteca is a species complex that appears to develop resistance to pyrethroids in the field 20. Using H. azteca, the present study investigated the lethal and sublethal interactions between chemicals commonly found in the aquatic environment. Concentrations tested were within environmentally detected ranges 1, 21, 22.

The primary targets of pyrethroid insecticides are voltage-gated sodium channels in the nervous system. Pyrethroids disrupt sodium channel activity by binding to channel receptors in the activated state and inhibiting their inactivation, causing nerve hyperactivity, neuromuscular dysfunction, and paralysis. The 2 types of pyrethroids are divided based on their chemical structures and consequent effects on sodium channels. Type II pyrethroids (i.e., cyfluthrin, lambda-cyhalothrin) contain a cyano group at the α-position of the 3-phenoxybenzyl alcohol moiety and cause longer lasting delays in channel inactivation, whereas type I pyrethroids (i.e., bifenthrin, permethrin) do not contain a cyano group and produce shorter acting delays 23. These classes are not absolute, and some pyrethroids exhibit intermediate properties 24.

In vitro studies have found both concentration-additive and less-than-additive interactions between pyrethroid pesticides 25-27, whereas in vivo studies have shown less than concentration-additive interactions 28, 29. The present study examines the interactions of all possible mixtures of 2 type I and 2 type II pyrethroids on mortality, swimming motility, and growth, and thus provides the most comprehensive in vivo analysis of pyrethroid mixture toxicity available to date.

MATERIALS AND METHODS

Toxicity testing

Water column toxicity testing methods (4 d and 10 d) for H. azteca were adapted from USEPA protocols for sediment toxicity testing 30, 31. Pesticides were purchased from Sigma-Aldrich and were used within 3 mo of purchase. Pesticides were prepared in methanol, and all treatments were spiked with a combination of stock and methanol as needed to attain 0.01% methanol, including a solvent control. Amphipods were purchased from Aquatic Research Organisms. All batches were acclimated to test conditions for a period of 72 to 96 h, and exposures were initiated with 10-d-old to 14-d-old H. azteca.

Each month, reference tests were conducted to evaluate the sensitivity of test organisms over time using sodium chloride (NaCl) as the reference toxicant. All control survival fell within the USEPA test acceptability criteria for water column toxicity testing (90% survival). Effect concentrations also fell within the 95% confidence interval of results across the present study period, indicating that organism sensitivity was consistent.

Five replicates of 10 amphipods were exposed in 100 mL of treatment water contained in 250-mL beakers. Individual pesticide toxicity tests were conducted within each of the mixture tests. Experimental treatments included 7 concentrations of each individual chemical, 6 equipotent concentrations of their mixture, and a 3 × 3 factorial design of treatments encompassing nonequipotent mixtures (Figure 1). Mortality was scored daily, and all mixtures were analyzed for acute 4-d lethality. Organisms were considered dead if all appendages, including pleopods, were immobile after repeated stimulation using a water stream. Two mixtures, permethrin–bifenthrin and permethrin–cyfluthrin, were also tested for 10-d (subchronic) lethality, as well as the sublethal responses swimming motility and growth after test termination.

Statistical analysis

The objective of this experiment was to evaluate the impact of multiple doses of selected pyrethroids on mortality, swimming motility, and growth in H. azteca. For analysis of mortality, the measured outcome of this trial was dead versus alive at the end of the time period under study. This binomial measure was evaluated through the application of generalized linear models and the use of a logit link function. Specifically, we modeled the number of animals found dead (y) among beakers containing n individuals exposed to dosages of pyrethroid 1 (X₁) and pyrethroid 2 (X₂). Accordingly, we considered y to be binomially distributed over n trials with probability θ (i.e., y ∼ Bin(n, θ)). The probability of dying (0 ≤ θ ≤ 1) was then modeled as a function of the dose of each pyrethroid, X₁ and X₂, and their interaction, X₁X₂. The regression model is illustrated in Equation 1.

(1)

Mixture interactions that test significant in a generalized linear model represent either negative or positive interactions, distinguishable by the sign of the interaction parameter estimate (β₃X₁X₂), whereas no significant interaction indicates that the pyrethroid effects are concentration-additive, as the null model in a generalized linear model is additive. Significant negative interactions in a generalized linear model could represent 1 of 3 levels of a less-than-additive response. The antagonism could be 1) less-than-additive, but follow the model of independent action; 2) less-than-additive by the independent action model but more toxic in combination than either chemical individually; or 3) a truly antagonistic response such that the mixture of chemicals is less toxic than the component chemicals individually (Figure 2). To substantiate and/or further characterize the generalized linear model results, data were subsequently plotted against the theoretical models of concentration addition and independent action. These comparisons confirmed the presence or absence of interactions (e.g., a fit to concentration addition would agree with no generalized linear model interaction), and in cases of negative interactions, would define the interaction as 1 of the above 3 results (e.g., do the data follow the independent action model?).

The concentration addition model calculates the mixture effect concentration (ECx_mix) required to elicit a given response (x_mix) based on the proportional contribution from each pyrethroid (p_i) in the mixture and their individual effect concentrations (ECx_i). The independent action model estimates the effect expected in a mixture (E(c_Mix)) based on the product of the individual proportions of nonresponse (1 – E(c_i)) from each pyrethroid. Models are provided below in Equations 2 (concentration addition) and 3 (independent action).

(2)

(3)

The concentration addition and independent action models require individual toxicant dose–response curves for the calculation of the theoretical mixture response models. To control for variability between experiments, these individual exposures were incorporated as a part of each of the mixture tests (Figure 1 depicts the experimental design); therefore each mixture experiment contained a range of 6 or 7 treatments of each individual chemical, in addition to the mixture treatments tested. Lethal concentrations and confidence intervals were calculated separately for each chemical from each experiment using USEPA standard statistical protocols and the CETIS program (Ver 1.1.2, Tidepool Scientific Software). These values are provided in Table 1.

Table 1. Acute and chronic lethal effect concentrations (LC50s) and confidence intervals (in parentheses) for individual toxicants tested within each mixture experiment

Experiment/chemical	4-d LC50 (ng/L)	10-d LC50 (ng/L)
Bifenthrin–permethrin
Bifenthrin	1.89 (1.74–2.05)	1.48 (1.37–1.59)
Permethrin	49.06 (43.98–54.14)	35.84 (33.59–38.09)
Permethrin–cyfluthrin
Permethrin	18.2 (15.33–21.08)	15.69 (13.83–17.54)
Cyfluthrin	2.80 (2.38–3.22)	2.16 (1.93–2.38)
Bifenthrin–cyfluthrin
Bifenthrin	2.70 (2.52–2.87)	–
Cyfluthrin	1.99 (1.72–2.27)	–
Permethrin–cyhalothrin, lambda
Permethrin	40.76 (37.20–44.32)	–
Cyhalothrin, lambda	2.53 (2.37–2.69)	–
Bifenthrin–cyhalothrin, lambda
Bifenthrin	2.23 (2.09–2.36)	–
Cyhalothrin, lambda	1.00 (0.86–1.14)	–
Cyfluthrin–cyhalothrin, lambda
Cyfluthrin	3.22 (2.92–3.52)	–
Cyhalothrin, lambda	2.03 (1.86–2.22)	–

Because of the high level of mortality by the end of each subchronic exposure, few treatments and replicates were available for sublethal testing, and were not sufficient for model comparisons. As a result, sublethal data were analyzed only using logistic regression statistics.

RESULTS

Most mixtures of pyrethroids were concentration-additive, the only exception being the acute 4-d mixtures with cyfluthrin, a type II pyrethroid, with either of the 2 type I pyrethroids bifenthrin and permethrin. All pyrethroids significantly affected motility and growth individually at sublethal concentrations, and were concentration-additive in mixtures.

Mixture effects: mortality

Two of the 6 acute 4-d mixture exposures (permethrin-cyfluthrin and bifenthrin-cyfluthrin) negatively interacted and were additive by the independent action model, whereas neither of the subchronic 10-d mixtures interacted (Table 2 and Figures 3-6). All other pyrethroid mixtures, including combinations of the type I and type II pyrethroid pesticides, did not interact and were concentration-additive.

Table 2. Logistic regression results for chemical interactions on mortality after acute 4-d and chronic 10-d exposures

Mixture	Exposure time (d)	p value	Parameter estimate	Standard error
Bifenthrin–permethrin	4	0.108	–0.008545	0.005288
Bifenthrin–permethrin	10	0.402	–0.01031	0.01225
Permethrin–cyfluthrin	4	0.00594*	–0.02019	0.007225
Permethrin–cyfluthrin	10	0.118	–0.01910	0.01216
Bifenthrin–cyfluthrin	4	<0.001*	–0.2623	0.06025
Cyfluthrin–cyhalothrin, lambda	4	0.174	0.1897	0.1389
Permethrin–cyhalothrin, lambda	4	0.156	–0.008694	0.006097
Bifenthrin–cyhalothrin, lambda	4	0.219	–0.3589	0.2908

• ^* Statistically significant at p < 0.05.

DISCUSSION

The toxicity of pyrethroid mixtures to H. azteca was predominantly concentration-additive across lethal and sublethal endpoints. Significant interactions were detected in just 2 of the 8 mixtures tested (4-d cyfluthrin–bifenthrin and 4-d cyfluthrin–permethrin). Both mixtures interacted antagonistically on 4-d lethality and followed the independent action model; however, the mixture of cyfluthrin–permethrin was also tested for effects on 10-d lethality, swimming motility, and growth, and was additive by concentration addition for all 3 of these endpoints. Although antagonistic interactions were the exception in the present study, they have been the norm for all studies testing type I and type II pyrethroid mixtures on lethal endpoints to date 28, 29. Brander et al. 28 calculated an antagonistic interaction between cyfluthrin and permethrin after a 10-d exposure to H. azteca. Although the authors did not compare mixture data with the independent action model, the antagonism is described as “slight,” and appears to concur with the 4-d results from the present study. Schleier and Peterson 29 found an interaction between permethrin and cypermethrin in Drosophila melanogaster survival, a mixture not tested in the present study; therefore a direct comparison cannot be made.

Mechanistically, site-specific competition could explain the interactions and additive responses of pyrethroid mixtures found in the present and previous studies. Toxicants with similar mechanisms of action can be competitive agonists if they either compete for the same binding site in a cell or tissue, or bind at different sites and inhibit toxicant affinity or efficacy allosterically 32. An antagonistic effect could occur if 2 toxicants targeted the same site, but had different efficacies, particularly if the toxicant with the lower efficacy possessed a higher binding affinity. In this case, at concentrations at which binding sites in a cell or tissue are limiting, the toxicant with the higher binding affinity (and lower efficacy) would displace the other, thereby reducing the toxicological effect, and an antagonistic interaction would result. Allosterically, toxicants can compete by causing a conformational change at the binding site, thereby reducing the binding affinity or hindering the target effect. Also in this case, one would expect greater competition at higher concentrations, where binding is more likely to overlap.

It is especially noticeable that the permethrin–cyfluthrin mixture interaction was concentration-dependent. At the low concentrations, the mixture followed the concentration addition model most closely, but was less than additive as the concentrations increased. To a small degree this trend is also noticeable in the bifenthrin–cyfluthrin mixture, the only other mixture in which an antagonistic effect was significant. Given a limited number of sodium channel binding sites for pyrethroids to interact with, one could postulate that at low concentrations, at which binding sites are not a limiting factor, pyrethroids would be additive, whereas at higher concentrations competition for binding sites would be greater and an antagonistic effect could result. This effect was not found in the other type I and type II pyrethroid mixtures containing lambda-cyhalothrin, possibly because of a lower relative binding affinity than cyfluthrin.

Studies conducted on isolated mammalian neurons indicate that competitive interactions at specific target sites do occur between type I and type II pyrethroids. Song et al. 27 tested the interactions of tetramethrin and fenvalerate on rat dorsal root ganglion neurons, and found that tetramethrin competitively inhibited the effects of fenvalerate. The effect was reversible after tetramethrin washout, suggesting either competitive binding at the same site or allosteric activity at different sites. In a similar study, Motomura and Narahashi 26 tested the interactions of tetramethrin (type I) and deltamethrin (type II) on voltage-gated sodium channels in rat hippocampal neurons. Cells were treated subsequently with each chemical, and neuron pulses exhibited effects of tetramethrin only in the mixture, again indicating that tetramethrin competitively displaced or allosterically inhibited the effects of deltamethrin. Additional research in cell cultures from aquatic species is warranted. Testing the competitive or allosteric interactions of type I and type II pyrethroids, particularly mixtures between cyfluthrin and bifenthrin or permethrin, would help to confirm the molecular mechanism of the interactions detected.

Two of the pyrethroid mixtures, bifenthrin–permethrin and permethrin–cyfluthirn, were tested beyond 4-d lethality for subchronic and sublethal effects: 10-d lethality, swimming motility, and growth. Both mixtures were additive across all subchronic and sublethal effects, including the permethrin–cyfluthrin mixture in which an interaction was significant in the 4-d exposure. Although the present study is, to our knowledge, the first to investigate the sublethal interactions of pyrethroid mixtures in H. azteca, previous studies in other species have found similar results. Hasenbein et al. 33 measured the effects of ternary mixtures of permethrin, lambda-cyhalothrin, and the organophosphate pesticide chlorpyrifos on chironomid (Chironomus dilutus) motility, and found significant effects at sublethal concentrations, but did not specifically test whether the pesticides were additive or interactive. Barata et al. tested the combined effects of pyrethroid–metal binary mixtures on Daphnia magna feeding 34 and reproduction 35, and both studies found the mixtures to be additive by concentration addition. In mammalian species models, Wolansky et al. 36 found dose additivity in equipotent mixtures of 11 pyrethroids (including type I and type II) in terms of rat motor activity after oral gavage administration, whereas Yang 37 found that individual pyrethroids and their mixtures significantly affected embryogenesis and morphology, although data were not statistically analyzed for interactions, nor were they compared with the additive concentration addition or independent action models.

CONCLUSIONS

Multiple studies have confirmed that mixtures of pyrethroids occur at toxic concentrations in the environment and cause ecosystem effects 7, 8, 21. This is not surprising because of the potent effects of pyrethroids on nontarget organisms and their increasing use for urban and agricultural pest control. The present study confirms that pyrethroids are highly toxic to nontarget species and are additive or close to additive regardless of type. In addition, pyrethroids inhibit growth and swimming motility in H. azteca at sublethal concentrations and are concentration-additive for these endpoints. As a result, to protect native species, environmental water quality regulations should assume dose additivity among pyrethroid pesticides, and safety factors incorporated into risk assessment should account for mixture effects.

Data availability

Data are available on request. Please contact [email protected]