Parallel reinforcement pathways for conditioned food aversions in the honeybee

Abstract

Avoiding toxins in food is as important as obtaining nutrition. Conditioned food aversions have been studied in animals as diverse as nematodes and humans [1, 2], but the neural signaling mechanisms underlying this form of learning have been difficult to pinpoint. Honeybees quickly learn to associate floral cues with food [3], a trait that makes them an excellent model organism for studying the neural mechanisms of learning and memory. Here we show that honeybees not only detect toxins but can also learn to associate odors with both the taste of toxins and the postingestive consequences of consuming them. We found that two distinct monoaminergic pathways mediate learned food aversions in the honeybee. As for other insect species conditioned with salt or electric shock reinforcers [4-7], learned avoidances of odors paired with bad-tasting toxins are mediated by dopamine. Our experiments are the first to identify a second, postingestive pathway for learned olfactory aversions that involves serotonin. This second pathway may represent an ancient mechanism for food aversion learning conserved across animal lineages.

Figure 1

Honeybees Are More Sensitive to Quinine Than to Amygdalin in Sucrose Solution (A) The structure of the honeybee's proboscis revealed by scanning electron microscopy: the galea (GL) of the two maxillae, and the labium comprised of the two labial palps (LP) attached to the glossa (GS). The arrow indicates the end of the galea from which (B) was photographed. Scale bar represents 500 μm. (B) The dorsal view, looking down to the tip of the galea. The arrow indicates the first of the ten sensilla chaetica from which tip recordings were made. Scale bar represents 50 μm. (C) Honeybees were more likely to reject solutions containing quinine (logistic regression: χ₁² = 49.9, p < 0.001), and honeybees fed prior to testing were more sensitive to toxins in solution. n_quin = 30, n_amy = 30. Asterisks indicate where the response was significantly different to the 1.0 M sucrose control (least-squares multiple comparison tests, p < 0.05). The Δ value on the y axis is the deviation from the mean probability of drinking 1.0 M sucrose alone. (D) The rate of response of the neurons (GRN) in each sensillum depended on the stimulating solution. Each solution produced a distinct ratio of activity in the galeal GRN population. Asterisks indicate the stimuli with responses in GRN classes 1 and 2 that were significantly different to the sucrose control (t test, p < 0.05). n_S = 71, n_S+A = 43, n_S+Q = 35, n_A = 42, n_Q = 39, n_KCl = 23. Error bars represent ±standard error of the mean (SEM). (E–J) Two-second tip recordings were made from the galeal sensilla. Each voltage trace represents the following stimuli: 300 mM sucrose (E), 300 mM sucrose with 10 mM amygdalin (F), 300 mM sucrose with 10 mM quinine (G), 10 mM amygdalin (H), 10 mM quinine (I; see also Figure S1C), 1 mM KCl (J; the electrolyte used as the baseline conducting solution).

Figure 2

Honeybees Use Both Pre- and Postingestive Mechanisms to Learn to Avoid Toxins in Sucrose Solution (A) Low levels of PER indicate that honeybees learn to avoid extending their proboscis toward an odor associated with quinine in 1.0 M sucrose. (C) The presence of amygdalin, on the other hand, did not substantially affect acquisition during the first 3–4 trials, indicating that bees did not readily detect the toxin in the reward and instead associated the odor with sucrose. However, after the fourth trial, they began to cease exhibiting PER in response to odor at a rate that depended on the toxin dose in the reward (logistic regression: χ₁² = 77.9, p < 0.001). Note: the 0.01 mM dose was not tested for quinine; the “control” acquisition curve (1.0 M sucrose) is the same in both (A) and (C) (n = 95). n_quin: 0.1 mM = 48, 1 mM = 27, 10 mM = 37, 100 mM = 61. n_amyg: 0.01 mM = 80, 0.1 mM = 58, 1 mM = 57, 10 mM = 69, 100 mM = 51. (B and D) Olfactory memory consolidation was a decreasing function of toxin dose for both quinine (B) and amygdalin (D) within 10 min (colored bars) and again at 18–24 hr after conditioning (gray bars) (logistic regression: quinine: χ₄² = 167, p < 0.001; amygdalin: χ₅² = 310, p < 0.001). At the 24 hr test, the response to the odor dropped for subjects conditioned with 1.0 M sucrose (control) (χ₁² = 7.87, p = 0.005) but did not change for solutions containing quinine or amygdalin (logistic regression: quinine: χ₄² = 7.59, p = 0.108; amygdalin: χ₅² = 7.34, p = 0.194). n_quinine: 0.1 mM = 48, 1 mM = 27, 10 mM = 37, 100 mM = 61. n_amygdalin: 0.01 mM = 80, 0.1 mM = 58, 1 mM = 57, 10 mM = 69, 100 mM = 51. (E) Honeybees quickly learned to recognize an odor paired with sucrose and to avoid another odor paired with 1.0 M sucrose containing 10 mM quinine during a differential learning task. (F) Bees conditioned with a 1.0 M sucrose solution containing 100 mM amygdalin, however, did not readily make this distinction and stopped responding to both odors. n_quin = 38, n_amyg = 71. Error bars represent ±SEM.

Figure 3

Dopamine Is Involved in Learning to Associate an Odor with a Toxin before Ingestion, but Serotonin Is Involved in Learning to Avoid a Toxin after Ingestion (A) When honeybees were injected with 0.1 mM of the DA receptor antagonist, flupenthixol (FX), they had greater difficulty learning to avoid the odor paired with a solution containing 1.0 M sucrose with 10 mM quinine during a differential learning task (n_saline = 52, n_FX = 54). The sucrose-reinforced acquisition curves for the saline and FX groups were not significantly different (logistic regression: χ₁² = 1.22, p = 0.269). (B) If injected with a 0.1 mM dose of the 5HT receptor antagonist cocktail composed of methiothepin and ketanserin (MK), honeybees performed differential learning more rapidly than the control, as indicated by the greater divergence in the purple curves on trials 3 and 4. n_saline = 72, n_MK = 40. (C) Injection with a 0.1 mM dose of the DA receptor antagonist, FX, did not impair olfactory learning toward either 1.0 M sucrose or a 1.0 M sucrose solution containing amygdalin in a simple learning task. n_saline,suc = 35, n_saline,amy = 37, n_FX,suc = 36, n_FX,amy = 34. (D) On the other hand, injection of a 1 mM dose of the cocktail of 5HT receptor antagonists (MK) abolished the ability of honeybees to learn to avoid an odor associated with an amygdalin-sucrose reinforcer. The acquisition curve produced by conditioning with the toxin after injection with the antagonist cocktail was significantly different from that for honeybees subjected to the same conditioning but injected with saline (logistic regression: χ₁² = 12.1, p = 0.001). n_saline,amy = 43, n_MKamy = 43, n_saline,amy = 27, n_MK,amy = 40. Note: the reinforcer was 10 mM amgydalin in (C) and 100 mM amgydalin in (D); both were presented in 1.0 M sucrose. See also Figure S3D. Error bars represent ±SEM.