Trans-generational desensitization and within-generational resensitization of a sucrose-best neuron in the polyphagous herbivore Helicoverpa armigera (Lepidoptera: Noctuidae)

Dietary exposure of insects to a feeding deterrent substance for hours to days can induce habituation and concomitant desensitization of the response of peripheral gustatory neurons to such a substance. In the present study, larvae of the herbivore Helicoverpa armigera were fed on diets containing either a high, medium or low concentration of sucrose, a major feeding stimulant. The responsiveness of the sucrose-best neuron in the lateral sensilla styloconica on the galea was quantified. Results showed the response of the sucrose-best neuron exposed to high-sucrose diets decreased gradually over successive generations, resulting in complete desensitization in the 5th and subsequent generations. However, the sensitivity was completely restored in the ninth generation after neonate larvae were exposed to low-sucrose diet. These findings demonstrate phenotypic plasticity and exclude inadvertent artificial selection for low sensitivity to sucrose. No significant changes were found in the sensitivity of caterpillars which experienced low- or medium-sucrose diets over the same generations. Such desensitization versus re-sensitization did not generalise to the phagosimulant myo-inositol-sensitive neuron or the feeding deterrent-sensitive neuron. Our results demonstrate that under conditions of high sucrose availability trans-generational desensitization of a neuron sensitive to this feeding stimulant becomes more pronounced whereas re-sensitization occurs within one generation.

by dietary experience that also affected food selection behaviour 13,31 . In most cases, after a period of exposure to diets containing a chemical stimulus, in later larval instars the gustatory neurons exhibited reduced sensitivity to the stimulus [27][28][29][32][33][34] .
The process of gustatory desensitization of sensilla styloconica to feeding deterrent substances depends on the insect species and the deterrent itself. Some studies documented that the sensitivity of the deterrent neuron in the sensilla styloconica decreased significantly but not completely after a chronic dietary exposure to the deterrent substance from neonates to the final instar 27,28 . In other cases, several days of dietary exposure to a deterrent resulted in the complete desensitization of the deterrent neuron 29,34,35 . Such desensitization of peripheral gustatory neurons to feeding deterrent substances could be mediated by gustatory transduction pathways, centrifugal control by the central gustatory system or post-ingestive mechanisms 21,29,36 .
It is known that the maxillary sensilla styloconica of caterpillars of all lepidopteran species studied contain a "sucrose-best"neuron, the activation of which leads to stimulation of feeding behavior [37][38][39][40][41][42] . Compared to the number of reports on the plasticity of gustatory neurons in response to exposure to feeding deterrents, the effects of exposure to varying levels of phagostimulants on the gustatory sensitivity of plant-feeding insects have received little attention. After exposure for a few hours to food containing a high level of carbohydrates but a low level of protein, the responsiveness to sucrose in the blowfly Phormia regina, and the caterpillars of the lepidopterans Spodoptera littoralis and Grammia geneura exhibited desensitization to sucrose 38,43,44 . Little is known, however, about desensitization and resensitization of sucrose-best neurons in sensilla styloconica of caterpillars elicited by varying levels of dietary sucrose.
The cotton bollworm, Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae), is a typical polyphagous species feeding on at least 160 plant species [45][46][47] . Taste neurons sensitive to sucrose and the deterrent azadirachtin are located in the lateral sensillum, whereas neurons responding to the sugar alcohol myo-inositol and the deterrents strychnine and strophanthin-K reside in the medial sensillum 27,37,[48][49][50] . Deterrent-sensitive neurons in the maxillary sensilla of H. armigera caterpillars reared on artificial diets containing either strychnine or strophanthin-K from neonate to the 5 th instar exhibited reduced sensitivity to the two chemicals compared with the caterpillars reared on normal diets 27 . Sinigrin was found to be deterrent to H. armigera caterpillars in dual-choice leaf disk assays and it excites the same deterrent neuron in the medial sensillum as strychnine and strophantin-K. (Tang et al., unpubl. results). In the present study, the electrophysiological activity of the sucrose neuron in the lateral sensillum of H. armigera larvae that experienced diets differing in the content of the major feeding stimulant sucrose, high-sucrose (HS), medium-sucrose (MS) and low-sucrose (LS) diets were investigated.
We addressed the following inter-related questions: (1) Can chronic exposure to HS diets during larval development induce desensitization of the sucrose-best neuron in the lateral sensillum styloconicum of H. armigera? (2) Can the chronic exposure to HS diets result in a complete desensitization of the sucrose-best neuron? If so, how long is needed to achieve this? (3) Can the sensitivity of the sucrose-best neuron be fully restored upon exposure to a normal dietary sucrose content and how long is needed for full restoration? (4) Does exposure of caterpillars to LS diets induce plasticity of the sucrose-best neuron?

Results
With the aid of Autospike software plus visual inspection, spikes in the responses from lateral sensilla of the 5 th instar caterpillars of H. armigera with different feeding experiences were sorted out and assigned to different types of gustatory receptor neurons known to be present in each chemosensillum. By comparing to the responses to sucrose with those obtained to KCl, at least three types receptor neurons were identified as the "salt", "sucrose" and "water" best responding unites (see exemplary traces in Fig. 1). In most traces obtained from the responses to sucrose, the dominating "medium-sized" spikes were from the sucrose-best neuron, the "largest-sized" spikes were from the "salt" best responding neuron, and few "small-sized" spikes were from the "water" best responding neurons (see exemplary traces in Fig. 1).
Desensitization to sucrose after HS-exposure over generations. The sensitivity of the sucrose-best neuron in the lateral sensillum of H. armigera caterpillars exposed to HS diet was significantly affected by both generation and sucrose concentration (univariate ANOVA: generation, df = 7, F = 42.707, P < 0.001; sucrose concentration, df = 4, F = 20.598, P < 0.001). There was also a significant interaction between generation and sucrose concentration affecting response frequency (univariate ANOVA: generation × concentration, df = 28, F = 3.066, P < 0.001). The 5 th instar caterpillars exposed to HS diet exhibited decreased sensitivity of the sucrose-best neuron in the lateral sensillum of Wild caterpillars (see exemplary trace "F1" in Fig. 1) (Fig. 2A). In subsequent generations sensitivity to sucrose decreased graduallyto a slight response in F5 generation. The response frequency of the sucrose-best neuron to sucrose from F5 to F8 generation were not significantly different from those to the control solvent KCl (Post-hoc SNK-test after univariate ANOVA: all P = 0.143) (see exemplary traces in Fig. 1) (Fig. 2A).
The sucrose-best neuron in Wild caterpillars of H. armigera displayed obvious dose-dependent responses to a concentration series of sucrose from 0.001 mM to 10 mM (Wild in Fig. 2A). Similarly, caterpillars exposure to HS diet in F1, F2 and F4 generations also exhibited dose-dependent response patterns to sucrose but the sensitivity was significantly lower than that in Wild (Post-hoc SNK-test after univariate ANOVA: all P < 0.01) (F1, F2 and F4 in Fig. 2A). However, caterpillars exposed to HS diet from the 5 th generation to the 8 th generation did no longer display a dose-dependent response pattern to sucrose and the response intensity was significantly lower than that in the previous generations (Post-hoc SNK-test after univariate ANOVA: all P < 0.01) ( Fig. 2A). So, it is apparent that the sucrose-best neuron of 5 th instar larvae had become completely desensitized in the 5 th generation since exposure to high-sucrose diets. Responsiveness to LS/MS dietary exposure over generations. To establish that it was exposure to a HS diet across successive generations that produced the desensitization of the sucrose-best neuron, the responsiveness of the sucrose-best neuron in the lateral sensillum of H. armigera caterpillars exposed to either LS diet or MS diet over generations was also investigated (Fig. 2B,C). Firstly, the sucrose-best neuron of caterpillars exposed to LS diet displayed dose-dependent responses in each generation from Wild to F8 (Fig. 2B and see exemplary traces in Fig. 3). The sensitivity of the sucrose-best neuron in caterpillars reared on LS diet was significantly affected by both generation and sucrose concentration (univariate ANOVA: generation, df = 8, F = 2.855, P = 0.004; concentration, df = 4, F = 325.953, P < 0.001; generation × concentration, df = 32, F = 0. 468, P = 0.995). Compared to Wild, the response intensity of the sucrose-best neuron of caterpillars in F1-LS decreased significantly (Post-hoc SNK-test after univariate ANOVA: P < 0.01), whereas the sensitivity of the sucrose-best neuron in F2-LS was significantly higher than that of F1-LS (Post-hoc SNK-test after univariate ANOVA: P < 0.05) and was similar to that of the Wild and the F3~F8 generations (Post-hoc SNK-test after univariate ANOVA: P = 0.136) (Fig. 2B).
Similarly, the responsiveness of the sucrose-best neuron in caterpillars exposed to MS diet also displayed dose-dependent responses from Wild to F8 generation (Fig. 2C). Generation of exposure to MS-diet had no significant effect on the sensitivity of the sucrose-best neuron (univariate ANOVA: generation, df = 8, F = 1.644, P = 0.110; concentration, df = 4, F = 373.503, P < 0.001; generation × concentration, df = 32, F = 0.490, P = 0.992). Representative recordings of electrophysiological activity of the gustatory receptor neurons in the lateral sensillum of larvae exposed to high-sucrose (HS) diet in successive generations. Traces represent electrophysiological responses to 1 mM sucrose. Sucrose was dissolved in 2 mM KCl. The duration of each trace is 1 s (except those in rectangles). Spikes in each big rectangle show the expansions of part of the corresponding trace (the small rectangle), suggesting that different gustatory receptor neurons were activated. K: the "largestsized"spikes from the salt responding neuron. S: the medium-sized spikes from the "sucrose-best" neuron; W: the small-sized spikes from the "water" responding neuron; Note that the medium-sized spikes from the sucrose-best neuron had the dominating activities in most traces.

Responsiveness of caterpillars reared on HS diet and switched to LS or MS diets. The sensitivity
of the sucrose-best neuron of caterpillars that experienced one diet over eight generations and then were switched to other diets in the ninth generation were tested to investigate whether the changing of diets could result in a taste change. Three groups of caterpillars experienced different artificial diets differing in sugar content were investigated as described in materials and methods. In Group I, the responsiveness of the sucrose-best neuron was significantly higher after caterpillars of H. armigera exposed to HS diet for eight generations were exposed to LS or MS diet (univariate ANOVA: experience, df = 3, F = 49.112, P < 0.001; concentration, df = 4, F = 23.262, P < 0.001; experience × concentration, df = 12, F = 12.742, P < 0.001) (see exemplary traces in Figs 4 and 5A). When neonate F9-caterpillars were exposed to LS diet (F8-HS + F9-LS), the response of the sucrose-best neuron of 5 th instar caterpillars was significantly higher than that of F8-HS, F8-HS + F9-MS and F9-HS (Post-hoc SNK-test after univariate ANOVA: all P < 0.01) (Fig. 5A) and similar to that of Wild and F1-HS (Post-Hoc SNK Test after ANOVA: P = 0.075) (Fig. 5B).

Responsiveness of caterpillars reared on LS/MS diet and then switched to other diets. In
Group II, after the F9-neonate caterpillars derived from generation F8-LS were exposed to HS diet or MS diet, the responsiveness of the sucrose-best neuron of the 5 th instar caterpillars in F8-LS + F9-MS decreased but not significantly (Post-hoc SNK-test after univariate ANOVA: P = 0.292), while that of F8-LS + F9-HS caterpillars decreased significantly compared to that of F8-LS and F9-LS caterpillars (Post-hoc SNK-test after univariate ANOVA: all P < 0.05) (Fig. 6A). For example, 1 mM sucrose elicited 65.1 ± 9.34 spk s −1 in F8-LS + F9-HS caterpillars, which was significantly lower than 101.9 ± 8.66 spk s −1 recorded from F8-LS and F2-LS to F7-LS caterpillars (Post-Hoc SNK Tests after ANOVA: all P < 0.05) (Fig. 6A′).

Effects of diets with different sucrose levels on the deterrent and myo-inositol neurons.
The effects of HS diet or LS diet on the responsiveness of the myo-inositol-sensitive neuron and the sinigrin-sensitive neuron in the medial sensillum of H. armigera caterpillars were investigated. Results showed that the sensitivity of the myo-inositol-sensitive neuron in the medial sensillum of caterpillars in Wild was significantly higher than that of caterpillars in F1-HS, F9-HS and F9-LS in the response to 1 mM myo-inositol (Post-Hoc SNK Test after ANOVA: P < 0.05), while the response frequencies among the latter three groups were similar (Post-Hoc SNK Test, P = 0.840) (Fig. 7A,A′). On the other hand, 1 mM sinigrin elicited a similar response from the deterrent neuron in the medial sensillum of caterpillars from Wild, F1-HS, F9-HS and F9-LS (Post-Hoc SNK Test after ANOVA: P = 0.915) (Fig. 7B,B′).

Discussion
Sugars serve as universal sources of metabolic energy to organisms and most animals have the ability to taste sugars that in many species constitute primary stimulatory signals for feeding 51 . The sugar concentrations investigated here are within the range of reported levels in different plants and plant organs on which H. armigera is known to feed. The sugar content varies greatly in the principal host plants of the polyphagous herbivore H. armigera, e.g. 1.73~5.37% of dry weight in the green leaves of tobacco, ca. 8% in tomato fruits in green ripening stage, 16% in corn stalks, and up to 20.5% in floral buds of cotton [52][53][54][55][56][57] . We found that dietary exposure to the HS diet desensitized the sucrose-best gustatory neuron. The desensitization developed to stronger degrees over successive generations, reaching its maximum after five generations of exposure and remaining at this level in the next three generations. Once the caterpillars were exposed to the LS diet, the sucrose-best neuron was completely restored to its wildtype condition, showing that no selection for low sensitivity had occurred after eight generations. Taste plasticity induced by early exposure to stimuli within one generation was also reported in other insect species and animals. For example, honeybees Apis mellifera decreased the gustatory responsiveness to sucrose after switching to an sucrose-enriched diet for 6 h to 24 h 18 , and adult rats exposure to a sodium-deficient diet for 10 days selectively decreased NaCl responses in the chorda tympani nerve 26 . Mice that experienced unlimited exposure to sucrose diets in early life reduces motivation to acquire sucrose in adulthood 58 , but no similar findings have yet been reported on trans-generational taste plasticity in insects and other animals except Dias and Ressler reported that the fear odor could be inherited transgenerationally at behavioral, neuroanatomical and epigenetic levels from parental olfactory experiences via parental gametes 59 .
The exposure of caterpillars to HS diet for successive generations resulted in a gradual desensitization of the sucrose-best neuron, while the exposure to either LS diet or MS diet for successive generations failed to elicit significant change in the sensitivity of the sucrose-best neuron of H. armigera. On the other hand, after the neonate caterpillars exposed to the same diet for eight generations were exposed to another diet, the sensitivity of the sucrose-best neuron in the 5 th instar caterpillars demonstrated significant plasticity in most situations. For example, the exposure to LS diet elicited stronger responses of F8-HS + F9-LS or F8-MS + F9-LS caterpillars compared to that of F8-HS or in F8-MS caterpillars (Figs 5A and 6B), while the exposure to HS diet desensitized the sucrose-best neuron of caterpillars in F8-LS + F9-HS or in F8-MS + F9-HS compared to that in F8-LS or in F8-MS (Fig. 6A,B). This plasticity of the sucrose-best neuron of H. armigera caterpillars was at least in part consistent with plasticity in the chemosensitivity to sucrose in the locust Locusta migratoria, the blowfly P. regina, and the caterpillars S. littoralis and G. geneura exposed to high-sucrose/low-protein diets or to low-sucrose/ high-protein diets 38,43,44,60 .
The current findings also show that different sucrose contents in artificial diets elicited different degrees of desensitization or resensitization of the sucrose-best neuron in H. armigera. For example, LS diet elicited a stronger response in caterpillars of F8-HS + F9-LS than the response elicited by MS diet in F8-HS + F9-MS (Fig. 5A). On the contrary, the HS diet significantly desensitized the sucrose-best neuron of caterpillars in F8-HS + F9-HS (F9-HS) compared to the response of caterpillars in F8-HS + F9-MS elicited by MS diet (Fig. 5A). Similarly, HS diet elicited a significantly lower response in the caterpillars of F8-LS + F9-HS than MS did in F8-LS + F9-MS (Fig. 6A). Similar findings on the plasticity of the sensitivity to sucrose were also found in S. littoralis exposed to diets differing in sucrose and protein content for hours 61 . The shifts in carbohydrate/protein ratio in the HS/LS/MS artificial diet in the current experiment might have had effects on the development of H. armigera caterpillars as reported for Heliothis virescens 62 and Helicoverpa zea 63 . However, we found that caterpillars of H. armigera exposed to HS and LS diets in our rearing colonies had similar pupal development time, pupal mass, and pupal survival (% eclosing) as well as larval development time, while larval survival (% pupating) and the number of eggs produced by F9-HS individuals (larval survival: 69.53%; number of eggs/female: 344.84) was lower than those of the F9-LS individuals (larval survival: 85.41%; number of eggs/ female: 460.47) and the F9-MS individuals (larval survival: 93.26%; number of eggs/female: 622.83). On the contrary, F9-MS individuals had higher pupal survival (% eclosing) (92.08%) than that of F9-HS individuals (74.55%) as well as F9-LS individuals (84.10%). The current data suggest that the plasticity of the sucrose-best neuron in H. armigera depends on dietary sucrose content rather than on protein/sucrose ratio or altered development. We consider this explanation plausible for two reasons. First, the absolute content of protein in LS, HS and MS diets was similar and the protein/carbohydrate ratio in the current study was not so unbalanced as that in the studies on H. virescens, H. zea and G. geneura 38,62,63 . Bernays et al. 38 regarded the sensitivity changes of taste neurons to sucrose and amino acids in G. geneura to reflect carbohydrate but not protein imbalance. Second, the HS diet or the LS artificial diet in the current experiment did decrease the sensitivity of the neuron sensitive to myo-inositol in F1 generation, but did not change further on the basis of sugar concentration or trans-generationally, indicating artificial diets differing in sugar concentrations and generational exposure had no bearing on overall plasticity. Third, the sensitivity of the deterrent neuron to sinigrin in H. armigera did not exhibit any plasticity to any sugary diets in any tested generations, suggesting that there was no general physiological effect on H. armigera caterpillars experiencing LS diet or HS diets for eight generations. (B′ ) comparison of response intensity from neuron to sinigrin. Columns represent the mean response frequency + /− SE in the response to 1 mM myo-inositol (A′ ) and to 1 mM sinigrin (B′ ). Numbers below the columns show the numbers of insects tested. SNK post hoc test for one-way ANOVA was used to compare mean spike frequencies of the sensilla across caterpillars with different feeding experiences (P < 0.05). Means with different lower case letters are differ significantly (P < 0.05).
Scientific RepoRts | 6:39358 | DOI: 10.1038/srep39358 It is known that the myo-inositol serves as a phagostimulant and sinigrin as a deterrent to non-adapted lepidopteran caterpillars [64][65][66] . Our data demonstrated that neither exposure of F1 caterpillars nor trans-generational exposure to HS/LS diets desensitized the sinigrin-sensitive neuron in the medial sensillum of H. armigera. On the other hand, the sensitivity of the myo-inositol-sensitive neuron in the medial sensillum was lower in F1-HS caterpillars compared to that in Wild caterpillars, but did not exhibit significant plasticity in successive generations regardless of the diets experienced. We interprete this as evidence for specificity of desensitization of the sucrose-best neuron caused by exposure to the HS diet across successive generations. Specificity in the desensitization of gustatory neurons was also reported for other caterpillars, e.g. a gustatory neuron in the medial sensillum of caterpillars of G. geneura exposed to a high-sucrose diet was less responsive to sucrose while the response to fructose in the lateral sensillum was constant 38 . Similarly, dietary exposure to caffeine in Manduca sexta caterpillars desensitized a caffein-sensitive neuron but neither the sucrose neuron nor the myo-inositol-sensitive neuron were affected 29 .
It was unexpected that not only the sensitivity of the sucrose-best neuron in F1-HS caterpillars decreased compared to that in Wild caterpillars but also that of F1-LS and F1-MS caterpillars. Decreased sensitivity was also observed in the myo-inositol-sensitive neuron of caterpillars in generation F1-HS. We postulate that F1 caterpillars had not yet completely adapted to the artificial rearing diet but F2 caterpillars had, as judged by the recovered sensitivity of the sucrose-best neuron in F2-LS (Fig. 2B) or F2-MS caterpillars (Fig. 2C). Another sign of ongoing adaptation was the observation that 4 th or the 5 th instar caterpillars of H. armigera captured directly from the natural fields had a higher mortality than that of second or third instar caterpillars if reared on normal artificial diets (unpublished data).
Desensitization of the deterrent-sensitive neurons in H. armigera and other caterpillar species in general took 2 h to 24 h, however, it was not complete 27,28,34,36 . Our results show that it took five generations of exposure to a high dietary sucrose concentration to completely desensitize the sucrose-best neuron. Transgenerational experiments on desensitization to deterrents have not been performed to our knowledge. Several mechanisms may explain exposure-elicited desensitization, e.g. a reduced number of membrane-bound gustatory receptor proteins 10,67-69 and/or down-regulated secondary messenger signal-transduction pathways [70][71][72] . The sensitivity of the sucrose-best neuron in the last instar caterpillars of H. armigera declined gradually over successive generations, suggesting that the reduced sensitivity could be transferred from one generation to the next and may involve an epigenetic phenomenon. Based on the full recovery of electrophysiological sensitivity within one larval generation, genetic selection on low sensitivity to sucrose can be excluded.
The regulation of gustatory plasticity in insects is clearly very complex. Notable in these responses is the observation that, after exposure to HS diet over generations, caterpillars of H. armigera showed lower responsiveness to sucrose, while after switching after eight generations to a LS diet higher responsiveness to sucrose was found. It seems likely that this restoration of the sensitivity of the sucrose-best neuron is driven, at least in part, by low a carbohydrate level in the haemolymph. Therefore, we consider the operation of a post-ingestive centrifugal nutritional feedback mechanism involving the central nervous system as the most parsimonious explanation for the sucrose exposure-elicited gustatory de-and resensitization. The exposure to a high dietary level of sucrose may result in a reduced sensitivity of the sucrose-best neuron in H. armigera since abundant dietary sucrose may not require gustatory sensitivity to sucrose and may save investments in maintaining sucrose chemoreception. Moreover, it was reported that the biogenic amines, c.a. the octopamine and serotonin in extracellular sensillum lymph, could adjust the sensitivity of olfactory receptor neurons of M. sexta by modulating the transepithelial potential of the accessory cells in the sensillum 73 . Therefore, it is possible to explore whether such neurotransmitters or hormone in extracellular sensillum a affected by artificial diets, resulting in direct or indirect plasticity of the sucrose-best neuron.
A nutritional feedback mechanism has been shown to operate within 5 min in response to injection of a mixture of amino acids into the haemocoel in the orthopteran L. migratoria and resulted in a reduced response of taste receptors to several amino acids 74 . Similarly, the gustatory response of labellar taste sensilla in the blowfly P. regina to glucose decreased significantly two hours after the injection of 1 M trehalose into the haemolymph 75 . Therefore, we hypothesize that the plasticity of the sucrose-best neuron in caterpillars of H. armigera induced by diets differing in sucrose content is related to variations in the levels of sugars in the haemolymph. In a next study, the levels of sugars in the haemolymph of H. armigera caterpillars after exposure to different levels of sucrose in the artificial diet will be quantified to test this hypothesis.

Materials and Methods
Insects. Helicoverpa armigera larvae were collected from a tomato field in Zhengzhou, Henan province, China, and were divided into different experimental groups. Based on the diet composition described by Wu & Gong 76 , three diets differing in sucrose concentration, high-sucrose (HS), medium-sucrose (MS) and low-sucrose (LS), were prepared to rear the experimental larvae. Including the sucrose in wheat bran, soybean powder and other dietary contents, the percentages of sucrose in the HS, MS and LS diets were 17.15%, 5.69% and 2.65% of dry weight, respectively which were within the range of sugar contents reported for natural host plants of H. armigera caterpillars (see Discussion). All colonies of H. armigera were maintained in the laboratory under controlled photoperiod (L16:D8) and temperature (27 ± 1 °C). Adults were supplied with a 10% v/v solution of sucrose in water.
Three groups of H. armigera caterpillars exposed to the three dietary sucrose levels were used in this experiment. Every group consisted of caterpillars captured directly from the field (Wild) as described above, caterpillars exposed to each diet for eight successive generations and caterpillars that were switched to the other two diets in the ninth generation as shown in Fig. 8 Chemicals. Sucrose, myo-inositol and sinigrin were obtained from Sigma Chemical Co. (purity > 99.5%). For electrophysiological tests, sucrose was presented in a series of concentrations from 0.001 to 10 mM dissolved in 2 mM KCl, an appropriate electrolyte for the electrophysiology of Helicoverpa caterpillars 27,77 . As a control for the specificity of effects of diets on the sensitivity of the sucrose-best neuron, we also tested whether diets containing different contents of sucrose have effects on the sensitivity of another phagostimulant neuron and a deterrent neuron, i.e. the independent phagostimulant neuron to myo-inositol and the deterrent neuron to sinigrin in the medial sensillum of H. armigera caterpillars 37 . Concentration of both chemicals was 1 mM, a concentration often used in lepidopteran taste studies 64,78 . Electrophysiological recording. The tip recording technique 79,80 was used to investigate the electrophysiological activity of neurons in the lateral sensilla styloconica on the galea of H. armigera larvae. In brief, 5 th instar larvae between 24 h and 36 h (during the photophase) since the penultimate moult were transected between the first and second pair of thoracic legs and the excised head capsule was mounted on a silver wire electrode that was connected to the input of a pre-amplifier (Syntech Taste Probe DTP-1, Hilversum, The Netherlands). The maxillary lateral sensilla styloconica were investigated for the sensitivity to sucrose of caterpillars that had fed on diets differing in sucrose content, while the medial sensilla were stimulated to assess the sensitivity of the myo-inositol-sensitive neuron or the deterrent neuron. Recordings of electrophysiological activity were obtained from the sensillum styloconicum of one side of the mouthparts of each caterpillar to different concentrations of sucrose or myo-inositol or sinigrin using a glass microelectrode (tip diameter ca. 30 μ m) filled with stimulus solution. Each caterpillar was only tested by one of the three stimuli with series concentration and then discarded. The order in which the series of sucrose concentrations were applied to a single sensillum was from low to high concentration of sucrose. The duration of each stimulation was 10 s and an interval of at least 3 min between two stimulations was observed. At least 10 larvae in each dietary group were tested for their sensitivity of a gustatory receptor neuron to the sucrose concentration series, myo-inositol or sinigrin. Amplified signals were digitized by an A/D-interface (IDAC-4, Syntech) and sampled and stored on a personal computer. Electrophysiological responses were quantified by counting the number of spikes in the first second after the start of stimulation.
Spikes were analyzed and counted visually by the experimenter with the aid of Autospike version 3.7 software (Syntech), running the sub-routine "Spike Conversion", by which all spikes in a recording can be classified according to their amplitude but also their waveform (shape). We then, in general, applied an amplitude threshold and the scale bar to select the target spike groups originating from the "sucrose-best neuron" or other neurons tested. In general, this medium-sized spike type but not the "largest" spike type was identified from the "sucrose-best" neuron because these spikes had two distinct features: (1) it always exhibited the most regular waveform; (2) it had an obvious phasic temporal response pattern. Applying these criteria, amplitude, uniform waveform and phasic temporal pattern resulted into the counts we reported. We then verified the counts arrived at by visual inspection and adjusted the count when the software seemed to have unduly skipped or included a waveform. Such adjustments were within 5% of the count reported by the Autospike software.
Data analysis. The mean response frequency, i.e. the number of spikes fired by the sucrose-best neuron in the lateral sensillum in the first second (spk.s −1 ) was calculated. For statistical analysis, the original number of spikes per second in response to each stimulus was square-root transformed. The univariate ANOVA with Student-Newman-Keuls (SNK) Post-hoc test was used to compare: (1) the mean response frequency of the sucrose-best neuron in caterpillars exposed to one type of artificial diet over eight successive generations in response to different concentrations of sucrose; (2) the mean response frequency of the sucrose-best neuron to concentrations of sucrose in caterpillars exposed to one type of artificial diet in the F8 and F9 generation, and Figure 8. Timelines of exposure regimes and diet-switching of three groups of caterpillars exposed to diets differing in sucrose levels used in the experiments. HS diet: high-sucrose diet; LS diet: low-sucrose diet; MS diet: medium-sucrose diet. caterpillars exposed to other two artificial diets in F9 generation, i.e. caterpillars in F8-HS, F9-HS, F8-HS + F9-LS  and F8-HS + F9-MS. An one-way ANOVA followed by Post-hoc SNK test was used to compare: (1) the mean response frequency of the sucrose-best neuron in caterpillars with different feeding experiences in response to 1 mM sucrose; (2) the mean response frequency of the myo-inositol-sensitive neuron or the sinigrin-sensitive neuron in the medial sensillum of caterpillars from Wild, F1-HS, F9-HS, F9-LS in response to 1 mM myo-inositol or 1 mM sinigrin. The significance level was set at P < 0.05 or P < 0.01. All statistical analyses were conducted using SPSS version 10.0 (SPSS Inc., Chicago, IL, USA).