In this study, we showed that sublethal doses of a neonicotinoid (thiacloprid) and of a phenylpyrazole (fipronil) highly increased mortality of honeybees previously infected by the microsporidian parasite N. ceranae. Although the exact mechanism involved in this synergistic effect remains unclear, our data suggest that the sensitization process is not strongly linked to a decrease of detoxification capacity in infected bees or necessarily by an enhancement of N. ceranae proliferation after exposure to insecticides.
During our experiments, no mortality was observed at 10 days p.i. in infected honeybees. Numerous foci were visible in their epithelial cells and a mean of 18.4.106
) spores/honeybee was measured in their digestive tract. Our results contrast with previous data by Higes et al. 
who described 100% of honeybee mortality at 8 days p.i. with N. ceranae
, but are comparable to mortality rate and spore production observed by Paxton et al. 
. In addition, the very highly significant enhanced sucrose consumption by infected honeybees is consistent with the energetic stress recently described by Mayack and Naugh 
. Thus, at 10 days p.i., we can consider that the infected honeybees of our study displayed a level of N. ceranae
invasion seen in forager honeybees 
The first evidence of a synergistic interaction between Nosema
infection and insecticide exposure in honeybees was described by Alaux et al. 
. These authors demonstrated that Nosema
spp treatment combined with exposure to imidacloprid, another neonicotinoid, resulted in a higher mortality of honeybees. Based on these results, we hypothesized that N. ceranae
infection could alter the functioning of detoxification system. We assessed the ECOD and GST activities to test this hypothesis. In insects, ECOD and GST activities have often served as convenient measures of overall phases I and II metabolizing enzyme activities. In addition, levels of ECOD and GST activities have been associated with sensitivity to insecticides 
. Our results showed that ECOD activity remained unchanged at 10 days p.i, in fat body and midgut whereas GST activity increased significantly in both tissues. Therefore, these data indicated that the higher mortality observed after insecticide exposure in N. ceranae
-infected honeybees was not strongly linked to a decrease in detoxification capacity. However, we cannot exclude that infection by N. ceranae
could modify other enzymes involved in detoxification of these insecticides. Despite this observation, exposure to sublethal doses of fipronil and thiacloprid increased the mortality rate in N. ceranae
-infected honeybees. Indeed, in our experimental procedure, at 10 day p.i. honeybees were chronically exposed to very low doses of insecticides during 10 days. Fipronil consumption was not different in infected and uninfected honeybees. The mean of daily fipronil consumption corresponded to an exposure equivalent to the LD50
/158 (25.3±4.8 pg/bee) for uninfected honeybees and LD50
/148 (26.9±0.8 pg/bee) for infected honeybees. A comparable exposure level has been previously considered as sublethal by Aliouane et al. 
. In the same way, honeybees exposed to thiacloprid consumed a similar daily quantity of insecticides of LD50
/151 (112.1±4.4 ng/bee) and LD50
/112 (152.8±8.7 ng/bee) for uninfected and infected honeybees, respectively. As suspected, these levels of fipronil and thiacloprid exposure had no effect on the mortality of uninfected honeybees and on their behavior. Surprisingly, the same levels of exposure caused symptoms of poisoning in infected honeybees and influenced the mortality rate.
Because this metabolic hypothesis failed to explain the sensitization process observed with mortality data, we assessed the effect of exposure to insecticides on spore production. Our results indicated that exposure to fipronil and thiacloprid had antagonist effects on spore production. Indeed, in comparison to infected honeybees not exposed to insecticides, the spore production decreased by about 33% during exposure to fipronil whereas the spore production increased by 40% with thiacloprid exposure. These results then, do not explain the mortality increase observed in the presence of insecticides. First, exposures to fipronil and thiacloprid induced an increase in mortality among infected honeybees but had opposite effects on spore production. Second, in the case of thiacloprid, the spore overproduction did not seem sufficient to explain the enhancement of honeybees' mortality.
The interactive effect seen between N. ceranae
and insecticides on honeybee mortality was consistent with the observations in honeybees infected with Nosema
sp and exposed to imidacloprid 
. While the synergistic effect observed by Alaux et al. 
seemed to be linked to an increased consumption of imidacloprid by infected honeybees, the synergistic effect observed in our study, however, was not due to increased food intake following infection. These new data on the synergistic action of Nosema
and insecticides highlight that such interactions are not restricted to neonicotinoids (imidacloprid, thiacloprid) but extend to other classes of insecticides including phenylpyrazoles.
A further generalization of this phenomenon in honeybees exposed to other insecticides would not be surprising. Several classes of chemical insecticides have already shown potency to interact synergistically with entomopathogenic fungi in other insect species. These kinds of combination are commonly used in integrated pest management because they counteract resistance to insecticides of many insects 
and allow reducing insecticide doses spread in the environment 
. For instance, organophosphorus compounds (oxydemeton methyl) and pyrethrinoïd (permethrin) insecticides used in combination with Beauvaria bassiana
, induced a higher impact on Spilarctia obliqua 
and Anopheles gambiae 
survival, respectively, than the use of these control agents alone. In general, the synergistic effect of these combinations appears at insecticide doses considered sublethal to the target insect 
. As is our study, the major pattern observed with these combinations included an increase in insecticide toxicity (decrease in LD50 or LC50) and a decreased time to onset of insect mortality. This suggests that susceptibility of insects to pesticides is a more complex phenomenon than previously thought. The influence of parasitism in the ecosystem must be considered in toxicological studies. As shown in our study, the use of the LD50
as an indicator of systemic insecticide toxicity leads to an underestimation of the deleterious effects induced in infected honeybees. Indeed, we demonstrated that sublethal doses of insecticides highly impacted Nosema
-infected honeybee mortality. This precaution is important since N. ceranae
spreads rapidly and can affect more than 80% of honeybee colonies 
Numerous examples of interactions between chemicals and pathogens that affect the insect lifespan have been described 
. Unfortunately, physiological mechanisms involved in these interactions remain poorly understood and may even appear to be contradictory. One of the current hypotheses explaining the synergistic effect of such combinations suggests that pathogen metabolites may interfere with the detoxification process 
. Reallocation of insecticide-detoxifying enzymes to counteract parasitic infections possibly reduces the quantity of enzymes available to target insecticides resulting in changes of insecticide toxicokinetics. Thus, it is possible that the synergistic effect results in an effective increase in sensitivity to insecticides in the presence of a proliferating parasite infection. Ironically, the few published data about the effect of parasitism on metabolizing enzymes of insects showed that a large set of parasitic infections could activate several proteins implicated in insect detoxification (e.g.
CYP's, GST, esterases) 
. Consistently, we showed in our study, that 10 days after infection by N. ceranae
, the GST activity was enhanced in midgut and fat body of honeybees, in agreement with the increase of the antioxidant activity recently described in Nosema
-infected queens 
. This result contrasts with the enhancement of infected-honeybees susceptibility to insecticides, suggesting that GST would not be involved in detoxification process of both fipronil and thiacloprid. However, the production of microsomal monooxygenases is an inducible process 
and it remains possible that induction of detoxification genes in response to exposure to insecticides was prevented by Nosema
infection. Thus, uninfected honeybees would be able to respond to insecticides by enhancing detoxification process whereas infected honeybees may not. This could explain the symptoms of intoxication observed in infected honeybees.
Curiously, in most studies reporting synergistic effect of fungus/insecticide combination, the impact of exposure to insecticides on parasite virulence was not investigated 
. In the rare studies addressing this aspect, the parasite virulence was not enhanced by the insecticides. Instead, despite the synergistic effect on insect mortality, it appears that exposure to insecticides tends to decrease germination or proliferation of the fungus 
. Indeed, insecticides have potential to affect the various developmental stages of entomopathogenic fungi 
to further justify why studies of compatibility between parasites and insecticides are important for developing IPM applications. In our study, fipronil and thiacloprid have antagonist effect on N. ceranae
proliferation whereby fipronil decreases slightly spore production in honeybees. This effect can be attributed either to the cytotoxic effect of fipronil on the intestinal epithelium 
or to its pro-oxidant action 
that may affect the reproduction cycle of N. ceranae
, but this assertion should be confirmed by other experiments. In contrast, thiacloprid increased spore production in our study. This result was not consistent with the observations done by Alaux et al. 
who showed that imidacloprid decreases slightly spore production in honeybees. Thus, in our studies, the synergistic effect of N. ceranae
infection and exposure to insecticide did not appear to be linked to enhancement of N. ceranae
virulence by insecticides.
To conclude, our study confirms that interactions between N. ceranae and insecticides constitute a significant risk for honeybee health. The increasing prevalence of N. ceranae in European apiary combined with the constant toxic pressure undergone by honeybees, appears to contribute to the honeybee colony depopulation. A better understanding of physiological effects induced both by low doses of pesticides and Nosema infection seems essential to elucidate the synergistic effects observed on honeybee mortality. The discovery of molecular and cellular mechanisms involved in the adverse effects induced by pathogens and pesticides would confirm the influence of these stressors on honeybee health. In addition, these data provide additional information that will allow a better assessment of risk associated with these stressors and highlight the urgent need of veterinary products for treating nosemosis.