Unequal rewarding of three metabolizable sugars – sucrose, fructose and glucose – in olfactory learning and memory in Bactrocera dorsalis

For animals in starvation, it is crucial to learn and evaluate the true nutritive value or the toxicity of a food source after absorption (Burke and Waddell, 2011; de Araujo et al., 2008; Fujita and Tanimura, 2011; Sclafani and Ackroff, 2004; Simões et al., 2012; Wright et al., 2010). During appetitive conditioning, association with the taste of the food reward is enough to form short-term memories, whereas only nutritional food rewards during conditioning can lead to the formation of long-term memories. This phenomenon is true in fruit flies (Burke and Waddell, 2011; Musso et al., 2015), honeybees (Wright et al., 2007) and oriental fruit flies (Yu et al., 2019). Subjects that were conditioned to associate an odor with sucrose, a metabolizable sugar, were able to form long-lasting memories of the conditioned odor. If sucrose was replaced with arabinose or L-glucose, a nonmetabolizable sugar with a sweet taste, although the subjects showed a high learning score immediately after conditioning, they could not form robust 24-h memory (Burke and Waddell, 2011; Musso et al., 2015). In honeybees, when the food reward (sucrose) was delivered only to the antennae during conditioning without consumption, stimulation of the antenna produced an association with an odor, but the memory decayed fast. Interestingly, feeding with a tasteless but nutritious sugar, such as sorbitol or xylose, following antennal stimulation could rescue the deficit in 24-h memory formation (Mustard et al., 2018). Both pre- and postingestive processes are involved in appetitive olfactory conditioning with food reward, and postingestive signaling is considered to be the mechanism for assessing food value and forming robust memories of sensory cues (Huetteroth et al., 2015).

The feature of postingestive signaling in food source evaluation during appetitive conditioning has been well documented. Consuming a metabolizable food is necessary to trigger the postingestive signal (Wright, 2011). However, few studies have tested whether all metabolizable sugars are equally rewarding to animals, and most studies have focused on model animals. Studies in Drosophila melanogaster confirm that flies form robust long-term memories for several metabolizable sugars, including sucrose, fructose and glucose (Burke and Waddell, 2011; Musso et al., 2015; Perisse et al., 2013). By contrast, honeybees and mice have robust long-term memories when they are fed sucrose and glucose, but not fructose (Müller, 2002; Sclafani and Ackroff, 2016; Simcock et al., 2018). These results indicate that the postingestive mechanisms for assessing food value are different among animals. Thus, it is necessary to study how reward quality affects learning and memory in other insect species.

The oriental fruit fly Bactrocera dorsalis is a major pest of vegetables and fruits in the Asia-Pacific region, with a broad range of host species (Drew and Hancock, 1994; Wan et al., 2012). Our previous work indicated that B. dorsalis shows excellent learning capability. In the laboratory, restrained flies could learn to associate an odor with food rewards and punishments (Liu et al., 2017, 2015). When flies are subjected to the olfactory conditioning paradigm, an odor is presented to their antennae, and then a sugar solution is presented first to their tarsi to elicit the proboscis extension reflex (PER) and then to their proboscis to allow them to consume the sugar solution, thus allowing for formation of the association of an odor with taste, and also with the postingestive consequences of consuming sugar. Moreover, the taste information was detected through two different gustatory receptor neurons: those in the tarsi and proboscis.

The purpose of the present study was to further investigate whether metabolizable sugar leads to equal rewards for B. dorsalis in terms of learning and memory, and to examine whether different pathways of sugar signals affect learning and memory. To explore these issues, in our experiments, we conditioned flies to associate an orange odor with three different sugars: sucrose, glucose and fructose. Furthermore, we conditioned flies by stimulation of the tarsi alone, proboscis alone, or both tarsi and proboscis without feeding, and then tested them for their responses to the conditioned orange odor 10 min and 24 h after conditioning. Finally, by feeding flies various sugars when stimulating tarsi with a given sugar, we examined how taste and postingestive signals affect learning and memory.

Adult Bactrocera dorsalis (Hendel 1912) males were obtained from a laboratory colony established at South China Agricultural University (Guangzhou, China) as described by Liu et al. (2015). The males were separated from the colony within 3 days of emergence to avoid mating, housed in a cage (30×30×30 cm) under laboratory conditions (25-27°C, 60-80% relative humidity, 14 h light:10 h dark photocycle, with light starting at 07:00 h) and provided with an artificial diet (sugar:hydrolyzed yeast 3:1, the main component of the sugar is sucrose) and water. The flies used in our experiments were 10 days old. Before the experiments, each fly was first fed to satiety with 1.0 mol l⁻¹ sucrose and then left for 12 h in a rearing cage with water.

The PER test was performed similarly to the test described in Fujita and Tanimura (2011). The flies were fixed in a small tube (35 mm long) with their forelegs and heads extending from the narrow cut end (4 mm outer diameter) of the tube (Liu et al., 2015). Before the PER test, the flies were allowed to rest for 1 h to acclimatize to the experimental conditions. In the proboscis test, each fly was first satiated with water until it stopped responding to water, and then stimulated with sugar by touching its labellum to elicit the PER. Proboscis extension was observed within 3 s. Each fly was used only once and exposed to only one concentration of each test sugar. Ten concentrations (0.003, 0.01, 0.03, 0.1, 0.3, 0.6, 1.0, 1.3, 1.6 and 2.0 mol l⁻¹) of three sugars (sucrose, fructose and glucose) were tested. Each test included 60 individuals. In the tarsi test, the experimental protocol was similar to that in the proboscis test; the only difference was that PER was elicited by touching the forelegs with sugar.

The nutritional value of each sugar for B. dorsalis was verified by housing flies in a container (52 mm inner diameter×45 mm) with various sugars as their sole source of food, and the sugar solution was changed every 2 days. Each group included 10 male flies. The number of living flies was counted every 12 h. A control group was provided water alone. Six repetitions were conducted for each condition.

The olfactory appetitive paradigm was performed as described in Liu et al. (2015). The flies received stimulation with an odor as the conditioned stimulus (CS), which was paired with a closely timed appetitive reward as the unconditioned stimulus (US). In this experiment, an orange odor (20 µl, CS) (Pretty Valley, Guangzhou, China) was delivered for 6 s, and, 3 s after the onset of odor presentation, 1 mol l⁻¹ sugar solution (US) was presented for 3 s so that presentation of the CS and US overlapped for 3 s. Odor and sugar delivery are described in detail in Liu et al. (2015). Five different conditioning protocols were used, in which the presentation of the US varied: tarsi-only conditioning (TC), proboscis-only conditioning (PC), tarsi-plus-proboscis conditioning (TPC), tarsi-plus-water-feeding conditioning (TCW) and tarsi-plus-proboscis-feeding conditioning (TPFC). In the TC protocol, only tarsi were stimulated using a sugar-solution-moistened toothpick, and the flies were not allowed to touch the sugar solution with their proboscis or to feed during conditioning. In the PC protocol, a filter paper (0.4×10 mm) soaked with 5 μl sugar solution was used to touch the proboscis directly, and the flies were not allowed to feed during conditioning. The flies were very sensitive to the test sugar; thus, when the filter paper was close to them, they extended their proboscis. Because the amount of sugar solution dropped on the filter paper was quite small, it was difficult for flies to feed on it during the 3 s presentation. In the TPC protocol, both tarsi and proboscis were stimulated with the sugar-soaked filter paper used in the PC protocol and the flies were not allowed to feed during conditioning. In the TCW protocol, the tarsi were stimulated with the sugar-solution-moistened toothpick to elicit proboscis extension, and a toothpick soaked with water was applied to the proboscis instead of the sugar solution. The water was consumed by the flies for 3 s. The TPFC protocol was designed to investigate how sugar quality affects learning and memory. The tarsi were stimulated with one sugar to elicit proboscis extension, although another sugar was applied to the proboscis and the fly was allowed to feed on it for 3 s. For example, when the tarsi were stimulated with a sucrose solution, a glucose solution was applied to the proboscis instead of the sucrose solution (T_sPF_gC). For each type of conditioning, the flies underwent five conditioning trials with a 10-min inter-trial interval. During conditioning, if flies associated the CS with the US successfully, the presentation of the CS would elicit the PER. Thus, the PER to the CS association was recorded during the first 3 s of the odorant presentation. The flies that extended their proboscis on the first trial were discarded to avoid innate sensitivity to the CS. Olfactory memory tests were administered 10 min and 24 h after conditioning. Each fly was tested for memory retention at only one time point. In the memory test, the odor stimulus was presented alone, and the PER to the odor was recorded in the first 3 s. Each experiment was tested on three different days.

In the PER assay, the presence or absence of a PER was registered for each individual. The proportions of the PERs triggered at a certain sugar concentration among the three sugar groups were analyzed using the Chi-square test for homogeneity. If the results were heterogeneous, then the analyses were complemented using multiple comparison Z-tests. In a certain conditioning experiment, the effect of successive training on the PER was analyzed using the Cochran test (Q). In the conditioning experiments, the PER acquisition during five conditioning trials of each fly was recorded (from 0 to 4; the first trial was excluded from analysis because it was designed to have a null response; thus, the maximum acquisition number was 4) and then used for analyzing learning performance. Kruskal–Wallis (KW) and Dunn's post hoc tests were used for analyzing individual learning performance among groups, and the significance level was P<0.05. In the retention tests, data recording and analysis were the same as those used in the PER assay. All data were analyzed using SPSS v20. In the figures, data are reported as probabilities of the PER.

To identify the phagostimulatory strength of sucrose, fructose and glucose, the tarsi and proboscis of individual flies were stimulated with a concentration series of the three types of sugars, and the probabilities of the PER were measured. In the proboscis assay, the proportions of PERs did not differ for the three sugars at all tested concentrations (0.003 mol l⁻¹: χ²=0.325, P=0.85; 0.01 mol l⁻¹: χ²=1.238, P=0.538; 0.03 mol l⁻¹: χ²=0.728, P=0.698; 0.1 mol l⁻¹: χ²=1.449, P=0.485; 0.3 mol l⁻¹: χ²=0.44, P=0.804; 0.6 mol l⁻¹: χ²=0.30, P=0.862; 1.0 mol l⁻¹: χ²=0.41, P=0.814; 1.3 mol l⁻¹: χ²=4.44, P=0.109; 1.6 mol l⁻¹: χ²=4.37, P=0.112; 2.0 mol l⁻¹: χ²=2.81, P=0.246; all d.f.=2; Fig. 1A). In the tarsi assay, the flies were much more sensitive to sucrose than to fructose or glucose when the sugar concentration was higher than 1 mol l⁻¹ (1.3 mol l⁻¹: χ²=20.24, P<0.001; 1.6 mol l⁻¹: χ²=10.27, P<0.05; 2.0 mol l⁻¹: χ²=13.31, P<0.01; all d.f.=2; Fig. 1B). No difference was observed at other concentrations (0.003 mol l⁻¹: χ²=1.114, P=0.573; 0.01 mol l⁻¹: χ²=1.129, P=0.544; 0.03 mol l⁻¹: χ²=1.646, P=0.439; 0.1 mol l⁻¹: χ²=0.808, P=0.668; 0.3 mol l⁻¹: χ²=4.14, P=0.126; 0.6 mol l⁻¹: χ²=0.96, P=0.619; 1.0 mol l⁻¹: χ²=4.89, P=0.087; all d.f.=2; Fig. 1B).

Three concentrations (0.3 mol l⁻¹, 1 mol l⁻¹ and 2 mol l⁻¹) of each sugar (sucrose, fructose and glucose) were supplied as food sources, and, after the first 7 days, no flies died in any of the treatments. With water, dead flies were observed after 2 days and all flies died in the subsequent 2 days. These results indicated that sucrose, fructose and glucose provided similar nutritional value for the oriental fruit flies (Fig. S1).

We used each of the three sugars as reinforcement in olfactory conditioning, and measured appetitive learning performance and memory persistence. For each conditioning, the PER frequency increased significantly over the four training trials (sucrose: Q=47.793; fructose: Q=40.146; glucose: Q=28.789; all d.f.=3, P<0.01; Fig. 2A). Although each sugar produced an association, learning revealed clear differences in individual performance among the flies reinforced with the different sugars (KW test, KW=27.26, P<0.05; Fig. 2A). The learning performance in the sucrose group was significantly higher than that in the fructose and glucose groups (Dunn's test, P<0.05). No significant differences in learning performance were found between the fructose and glucose groups (Dunn's test, P>0.05).

In the retention tests, memories were tested 10 min and 24 h after conditioning, and the long-lasting memory for the CS was evaluated by comparing the responses at 10 min with the responses at 24 h. The PER frequency at 24 h did not differ from that at 10 min in flies conditioned with sucrose (χ²=1.93, P=0.165) or glucose (χ²=0.02, P=0.881). By contrast, the flies conditioned with fructose showed a significant decrease in PER frequency at 24 h compared with the responses at 10 min (χ²=5.02, P<0.05) (Fig. 2B).

To determine whether the gustatory information elicited by sugars was the main reason causing the difference in learning for the sugars, the sugar was presented by touching the tarsi or proboscis without feeding. In the four conditioning protocols (TC, PC, TPC and TCW), the PER frequency increased significantly over the four training trials regardless of the sugar used for conditioning (TC: sucrose: Q=17.613, fructose: Q=24, glucose: Q=34.5; PC: sucrose: Q=12.6, fructose: Q=22.138, glucose: Q=14.571; TPC: sucrose: Q=35.958, fructose: Q=31, glucose: Q=16.091; TCW: sucrose: Q=29.847, fructose: Q=23.28, glucose: Q=26.069; all d.f.=3, P<0.01; Fig. 3). However, the level of learning performance in the three sugar groups depended on the conditioning protocol. The flies conditioned via the TC and PC protocols, in which they experienced stimulation of the tarsi or proboscis, showed no difference in learning performance among the three sugar groups (TC: KW=0.60, P=0.740; PC: KW=0.03, P=0.985; Fig. 3A,B). When both tarsi and proboscis were stimulated with sugar (TPC protocol) and the flies were not allowed to feed on the sugar, the levels of learning performance among the three sugar groups were different (KW=10.49, P<0.01). The flies trained with sucrose had a higher learning performance than those trained with fructose (Dunn's test, P<0.05) or glucose (Dunn's test, P<0.05). There was no difference in learning performance between the fructose and glucose groups (Dunn's test, P>0.05; Fig. 3C). To determine whether the behavior of feeding on something, even water, following stimulation of the tarsi could affect the discrimination of the sugars, a TCW protocol was conducted, in which the tarsi were stimulated with sugar to elicit proboscis extension and then water was applied to the proboscis for 3 s. A difference in learning among the three sugar groups was observed (KW=10.01, P<0.01). The learning performance of the sucrose-conditioned flies was significantly greater than that of the flies conditioned with fructose (Dunn's test, P<0.05) or glucose (Dunn's test, P<0.05; Fig. 3D).

Short-term memory was tested 10 min after conditioning, and similar results were observed to those in learning performance. For the four conditioning protocols, the difference in 10-min memory among the three sugar groups was observed in the TPC (χ²=11.20, P<0.01) and TCW (χ²=8.54, P<0.05) protocols, but not in the TC (χ²=0.36, P=0.837) and PC (χ²=0.27, P=0.875) protocols (Fig. 4). In the TPC and TCW conditioning protocols, the responses to the conditioned odor in flies conditioned with sucrose were significantly higher than those in flies conditioned with fructose (P<0.05) or glucose (P<0.05).

To further test the effect of fructose on memory persistence, the flies were conditioned using tarsi stimulation with one sugar and proboscis feeding on another sugar, and memory persistence was evaluated by comparing the responses to the CS at 10 min with the responses at 24 h after conditioning. The type of sugar fed during conditioning had a significant effect on the 24-h memory. When flies were fed sucrose, the level of responses to the CS at 24 h did not differ from that at 10 min regardless of whether the tarsi were stimulated with fructose (χ²=0.14, P=0.707) or glucose (χ²=0.03, P=0.854) (Fig. 5A). Similarly, when flies were fed glucose, the responses at 24 h also did not differ from those at 10 min regardless of whether the tarsi were stimulated with sucrose (χ²=0.14, P=0.714) or fructose (χ²=1.21, P=0.271) (Fig. 5C). However, when flies were fed fructose, the responses at 24 h were significantly lower than those at 10 min, regardless of whether the tarsi were stimulated with sucrose (χ²=7.18, P<0.01) or glucose (χ²=4.60, P<0.05) (Fig. 5B).

Our previous studies proved that B. dorsalis can learn to evaluate the nutritional value of sugar through postingestive mechanisms, and that feeding metabolizable sugar is the main factor in forming robust long-lasting memory (Yu et al., 2019). In this study, although B. dorsalis succeeded in associating an odor with three metabolizable sugars (sucrose, fructose and glucose), the flies were more likely to learn when they were rewarded with sucrose but not when rewarded with fructose or glucose. However, studies in honeybees indicate that bees learn when they are trained with sucrose or glucose but not with fructose (Simcock et al., 2018). Most notably, we found that fructose failed to support a high level of long-lasting memory. This finding is consistent with that previously reported in honeybees (Simcock et al., 2018). By contrast, studies in D. melanogaster showed that fructose could support long-lasting memories (Burke and Waddell, 2011). Taken together, all these findings indicate that the three metabolizable sugars are unequally rewarding to B. dorsalis, and that the role of metabolizable sugars in reinforcement may be species specific.

One process of appetitive olfactory conditioning is the association of an odor with a positive taste (preingestive process) (Wright et al., 2007). In B. dorsalis olfactory conditioning, the flies receive sensory information about the taste of the sugar reward through gustatory receptors on the proboscis and tarsi, and produce an association with an odor. The results of the TPC experiment revealed that flies rewarded with sucrose were more likely to learn to associate an odor with the reward, suggesting that flies could learn to discriminate three sweet sugars through preingestive signals triggered by stimulating the proboscis and tarsi. To further understand which signal is the main factor in discrimination, e.g. the signal from the proboscis or tarsi, the proboscis and tarsi were stimulated with sugar separately. In the PC experiment, it is not surprising that the three sugars equally rewarded B. dorsalis because the three sugars were equally phagostimulatory when the flies tasted sugars only with the proboscis. When the flies tasted sugars only with the tarsi (TC), sucrose was the most phagostimulatory sugar to elicit the PER. However, the flies did not have better learning for sucrose than for the other sugars. The loss of discrimination among the sugars as reinforcers might be due to necessary stimulation of the proboscis. Thus, a TCW experiment was conducted. The results from the TCW experiment indicated that feeding on water following the stimulation of the tarsi with sugar affected the preingestive perception of the quality of the three sugars and therefore affected the learning performances for the sugars. It is still unclear whether the mechano- and hygrosensory stimulation of the proboscis after stimulation of the tarsi with sugar might modulate the reward signal, causing the unequal rewarding for the three sugars during learning formation. To better explain the correlation between the preingestive signal and learning capability, an electrophysiological recording from labellar and tarsi taste sensilla, which are used to determine the taste sensitivity of external chemosensilla to sugar (Fujita and Tanimura, 2011), will be necessary in future study.

The other process of appetitive olfactory conditioning is the association of an odor with the postingestive consequences of ingesting a metabolizable sugar (Burke and Waddell, 2011; Gruber et al., 2013). Our previous research in B. dorsalis confirmed that the formation of long-lasting memories depended upon feeding on a sweet-tasting and metabolizable sugar (Yu et al., 2019). The present experiments revealed that this memory formation is not true for all sweet-tasting and metabolizable sugars (Fig. 1, Fig. S1). The flies fed fructose during conditioning did not produce robust long-lasting memories of a conditioned odor (Fig. 2B, Fig. S2). Although the levels of learning performance between the sucrose-fed flies and glucose-fed flies were different, the long-lasting memories formed with these two sugars did not extinct. By testing a novel odor (Fig. S2), the specific memories to the trained orange odor were confirmed, suggesting that the difference was not caused by contextual cues. To further confirm the deficit of long-lasting memory in fructose-fed flies, the flies were conditioned using tarsi stimulation with one sugar and feeding on another sugar (TPFC protocol). The flies fed fructose during conditioning did not have a high level of long-lasting memories, even though their tarsi were stimulated with sucrose or glucose. By contrast, flies fed sucrose or glucose during conditioning had long-lasting memories of the conditioned odor, no matter what sugar was used to stimulate the tarsi. Moreover, in all TPFC conditioning groups, the flies showed relatively good learning performance (Fig. S3). These results suggest that postingestive feedback after consuming fructose is not sufficient to support a high level of long-lasting memory. Our results are consistent with those previously reported in honeybees (Simcock et al., 2018). Moreover, work in honeybees (Simcock et al., 2018) suggests that glucose flux in the hemolymph is the postingestive signal of metabolizable reward that the brain uses to detect that a metabolizable reward was eaten. A greater glucose flux was observed in bees fed sucrose and glucose, although not in bees fed fructose, which might be the reason why bees fed fructose did not have a high level of long-lasting memory. Interestingly, studies in D. melanogaster found that fructose is an effective reinforcer of olfactory CS (Burke and Waddell, 2011; Musso et al., 2015). Miyamoto et al. (2012) reported a fructose receptor, GR43a, in the D. melanogaster brain that functions as a nutrient sensor for hemolymph fructose. It is possible that these receptors detect fructose and produce effective postingestive signals of food reward, which induces a long-lasting memory with the conditioned odor.

In conclusion, three sweet-tasting and metabolizable sugars – sucrose, fructose and glucose – unequally rewarded B. dorsalis during appetitive olfactory conditioning. Better learning and persistent memory were observed following olfactory conditioning with sucrose than with fructose or glucose. B. dorsalis could learn to discriminate the three sugars through both pre- and postingestive signals, which might be the mechanism flies use for assessing the quality of the food and then making decisions for host selection. Moreover, it will be interesting to explore the flies’ responsiveness to sugar mixtures. In the field, B. dorsalis searches for nectar as a food resource. As nectars are mixtures of sugars rather than single compounds, and have various sugar ratios in different types of flowers, the responsiveness of flies to sugar mixtures might reflect their foraging behavior for nectar. Studies evaluating similar responsiveness have been performed in insects, such as Agrotis ipsilon (Hostachy et al., 2019).

We would like to thank the anonymous referees whose comments improved the quality of the manuscript.