A Priming Circuit Controls the Olfactory Response and Memory in Drosophila
Curation statements for this article:-
Curated by eLife
eLife Assessment
This work, combining behavioural genetics and calcium imaging, provides evidence for a form of learning in Drosophila that derives solely from direct or (optogenetically induced) phantom experience of punishment or reward. Flies that experience foot-shock alone show a subsequent decrease in avoidance to all odorants, together with increased odor-evoked activation of reward-encoding dopaminergic neurons that innervate the mushroom body. Phantom reward, delivered via optogenetic activation of reward-encoding dopaminergic neurons, increases subsequent odour-avoidance. While the findings are valuable to the field, there are aspects of the work that are incomplete, and some of the conclusions and terminology are also not completely justified; three major issues include : (a) the use of the term "priming" to describe this form of learning seems inappropriate and inconsistent with the accepted definition of this term; (b) a key 1998 publication with an initial description of this behavioural phenomenon needs to be cited and presented as context; and (c) the work on reward induced increase in odor-aversion seems relatively preliminary.
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (eLife)
Abstract
Abstract
Priming is a process by which exposure to a stimulus affects the response to a subsequent stimulus in humans. In this study we found that in Drosophila, a prior encounter with an aversive stimulus results in enhanced preference for a following novel odor, while an appetitive stimulus leads to reduced preference for a new odor. This priming behavior of flies relies on the well-studied olfactory memory circuits including Kenyon cells (KC), dopaminergic neurons (DANs), and mushroom body output neurons (MBONs). Aversive stimulus results in increased odor responses in reward DANs that innervate the γ4 lobe of the mushroom body (MB) and decreased odor responses in a γ4γ5-innvervating repulsive MBON. We concluded that these neurons are required for the priming effects in flies. These results characterized the newly found priming behavior in flies and demonstrated the sheer influence of unconditioned stimulus on odor perception during associative learning.
Article activity feed
-
Author response:
We thank both reviewers for their valuable comments. We have prepared a point-by-point response below.
Reviewer #1 (Public review):
Weaknesses:
(1) The conclusions regarding the links between neural and behavioral mechanisms are mostly well supported by the data. However, what is less convincing is the authors' argument that their study offers evidence of 'priming'. An important hallmark of priming, at least as is commonly understood by cognitive scientists, is that it is stimulus specific: i.e., a repeated stimulus facilitates response times (repetition priming), or a repeated but previously ignored stimulus increases response times (negative priming). That is, it is an effect on a subsequent repeated stimulus, not ANY subsequent stimulus. Because (prime or target) stimuli are not repeated in the current experiments, …
Author response:
We thank both reviewers for their valuable comments. We have prepared a point-by-point response below.
Reviewer #1 (Public review):
Weaknesses:
(1) The conclusions regarding the links between neural and behavioral mechanisms are mostly well supported by the data. However, what is less convincing is the authors' argument that their study offers evidence of 'priming'. An important hallmark of priming, at least as is commonly understood by cognitive scientists, is that it is stimulus specific: i.e., a repeated stimulus facilitates response times (repetition priming), or a repeated but previously ignored stimulus increases response times (negative priming). That is, it is an effect on a subsequent repeated stimulus, not ANY subsequent stimulus. Because (prime or target) stimuli are not repeated in the current experiments, the conditions necessary for demonstrating priming effects are not present. Instead, a different phenomenon seems to be demonstrated here, and one that might be more akin to approach/avoidance behavior to a novel or salient stimulus following an appetitive/aversive stimulus, respectively.
(2) On a similar note, the authors' claim that 'priming' per se has not been well studied in non-human animals is not quite correct and would need to be revised. Priming effects have been demonstrated in several animal types, although perhaps not always described as such. For example, the neural underpinnings of priming effects on behavior have been very well characterized in human and non-human primates, in studies more commonly described as investigations of 'response suppression'.
We thank the reviewer for these critical comments. After careful consideration of both reviews, we agree that “priming” may not be the most accurate term to describe the behavioral phenomenon. We plan to revise our terminology throughout the manuscript accordingly to better capture the generalized nature of the effect we observe.
(3) The outcome measure - i.e., difference scores between the two odors or odor and non-odor (i.e., the number of flies choosing to approach the novel odor versus the number approaching the non-odor (air)) - appears to be reasonable to account for a natural preference for odors in the mock-trained group. However, it does not provide sufficient clarification of the results. The findings would be more convincing if these relative scores were unpacked - that is, instead of analyzing difference scores, the results of the interaction between group and odor preference (e.g., novel or air) (or even within the pre- and post-training conditions with the same animals) would provide greater clarity. This more detailed account may also better support the argument that the results are not due to conditioning of the US with pure air.
We use the PI score as a standard metric to quantify all the odor preference in behavioral assays because it allows for robust comparison across different genetic or treatment groups under the same experimental setting. In T-maze, real time tracking of fly trajectories is technically difficult. With olfactory arenas, we showed some examples of fly distribution in quadrants over the entire odor choice test period (Figure 2—figure supplement 2) for both pre-trained and post-trained groups and discussed the trajectories in Discussion. We will ensure this point is clarified in the revised text.
Reviewer #2 (Public review):
[…] They finally recorded from different mushroom body output neurons, including the one (MBON-γ4γ5) likely affected by the increased activity of the corresponding γ4 reward dopaminergic neurons after shock preexposure. They recorded odour-evoked responses from these neurons before and after shock preexposure, but did not find any plasticity, while they found a logical effect during spaced cycles of aversive training.
We thank the reviewer for the summary. We would like to clarify that we did, in fact, observe plasticity in MBON-γ4γ5 following shock exposure, as shown in Figure 4B.
Overall, the study is very interesting with a substantial amount of behavioural analysis and in vivo 2-photon calcium imaging data, but some major (and some minor) issues have to be resolved to strengthen their conclusions.
(1) According to neuropsychological work (Henson, Encyclopedia of Neuroscience (2009), vol. 7, pp. 1055-1063), « Priming refers to a change in behavioral response to a stimulus, following prior exposure to the same, or a related, stimulus. Examples include faster reaction times to make a decision about the stimulus, a bias to produce that stimulus when generating responses, or the more accurate identification of a degraded version of the stimulus". Or "Repetition priming refers to a change in behavioural response to a stimulus following re-exposure" (PMID: 18328508). I therefore do not think that the effects observed by the authors are really the investigation of the neural mechanisms of priming. To me, the effect they observed seems more related to sensitisation, especially for the activation of sweet-sensing neurons. For the shock effect, it could be a safety phenomenon, as in Jacob and Waddell, 2020, involving (as for sugar reward) different subsets for short-term and long-term safety.
As noted in our response to Reviewer #1, we plan to revise our use of the term “priming” in the manuscript to more accurately interpret the behavioral phenomenon.
(2) The author missed the paper from Thomas Preat, The Journal of Neuroscience, October 15, 1998, 18(20):8534-8538 (Decreased Odor Avoidance after Electric Shock in Drosophila Mutants Biases Learning and Memory Tests). In this paper, one of the effects observed by the authors has already been described, and the molecular requirement of memory-related genes is investigated. This paper should be mentioned and discussed.
We thank the reviewer for bringing this important reference to our attention. We will cite the Preat (1998) paper and discuss its relevant findings in relation to our own in the revised manuscript.
(3) Overall, the bidirectional effect they observed is interesting; however, their results are not always clear, and the use of a delta PI is sometimes misleading. The authors have mentioned that shocks induced attraction to the novel odour, while they should stick to the increase or decrease in preference/avoidance.
The ΔPI is calculated either as (trained PI – mock PI) for different animals or as (post PI – pre PI) for the same animals, with the specific calculation clarified in each figure legend. A positive ΔPI signifies an increase in preference for the odor, which is equivalent to a relative attraction or a decrease in avoidance.
As not all experiments are done in parallel logic, it is not always easy to understand which protocol the authors are using. For example, only optogenetics is used in the appetitive preexposure. Does exposing flies to sugar or activating reward dopaminergic neurons also increase odour avoidance? The observed increased odour avoidance after optogenetic activation of sweet-sensing neurons involve reward (e.g., decreased response) and/or punishment (e.g., increased response) to increase odour avoidance?
We used different behavioral assays (T-maze or arena), stimuli (real shock or optogenetics), and protocols (different or same animal groups) to robustly demonstrate the phenomenon across platforms. We explained each protocol in the figures or texts, and we’ll make them clearer to follow in the revised version. We focused on activating a clean set of sugar sensing neurons because this optogenetic stimulus is an effective and efficient substitute to real sugar. We agree that testing reward dopaminergic neuron activation is a logical extension and will consider adding these experiments in the revised work.
The author should always statistically test the fly behavioural performances against 0 to have an idea of random choice or a clear preference toward an odour.
Our primary focus is on the change in preference induced by training, rather than the innate odor preference itself, which can be highly variable due to physiological and environmental factors. Statistical testing against 0 for innate preference scores is not standard practice in this specific paradigm, as the critical question is whether a treatment alters behavior relative to a control.
On the appetitive side, the internal hunger state would play an important role. The author should test it or at least discuss it.
For appetitive experiments, we always starve the flies on 1% agar for two days prior to behavioral tests to standardize their hunger state. We will consider adding fed flies as control groups in the revised work.
(4) The authors found a discrepancy between genetic backgrounds; sometimes the same odour can be attractive or aversive.
We observed minor discrepancies in innate odor preferences across genetic backgrounds, which is a known and common occurrence. Different genotypes and temperatures can result in different baseline PI scores. However, the key finding is that the relative change in odor preference following an aversive stimulus is consistent: it increases the relative preference for an odor compared to air. This sometimes reverses valence (aversion to attraction) and other times simply reduces aversion. Our analysis focuses on this consistent, relative change.
Different effects between the T-maze and the olfactory arena are found. The authors proposed that: "Punishment priming effect was still not detected, probably due to the insensitivity of the optogenetic arena". This is unclear to me, considering all prior work using this arena. The author should discuss it more clearly.
The punishment effect with CS+ present was reliably detected in the T-maze (Figure 1A) but was not significant in the olfactory arena (Figure 2—figure supplement 1B-C). We hypothesize that the olfactory arena assay is less sensitive than the T-maze for detecting such subtle behavioral changes. This is evidenced by the fact that even classical odor-shock conditioning yields lower PI in the arena (typically ~0.4) than in the T-maze (~0.8), likely due to the greater distance flies must explore and travel. The higher variance in the arena may therefore mask more modest effects. Here the effect under investigation was induced by optogenetically activating only a small subset of aversive dopaminergic neurons, a stimulus that is likely weaker than full electric shock. This reduced stimulus strength may have contributed to the challenge of detecting a significant effect in the less sensitive arena paradigm.
They mentioned that flies could not be conditioned with air and electric shock. However, flies could be conditioned with the context + shock, which is changing in the T-maze and not in the optogenetic area.
While flies can be conditioned to context, during the optogenetic stimulation period in the arena, the light is delivered uniformly across all four quadrants. Therefore, any potential context conditioning would be equivalent across the entire chamber and should not bias the final distribution of flies between the odor and air quadrants during the test, nor affect the calculated PI score.
-
-
-
eLife Assessment
This work, combining behavioural genetics and calcium imaging, provides evidence for a form of learning in Drosophila that derives solely from direct or (optogenetically induced) phantom experience of punishment or reward. Flies that experience foot-shock alone show a subsequent decrease in avoidance to all odorants, together with increased odor-evoked activation of reward-encoding dopaminergic neurons that innervate the mushroom body. Phantom reward, delivered via optogenetic activation of reward-encoding dopaminergic neurons, increases subsequent odour-avoidance. While the findings are valuable to the field, there are aspects of the work that are incomplete, and some of the conclusions and terminology are also not completely justified; three major issues include : (a) the use of the term "priming" to describe this …
eLife Assessment
This work, combining behavioural genetics and calcium imaging, provides evidence for a form of learning in Drosophila that derives solely from direct or (optogenetically induced) phantom experience of punishment or reward. Flies that experience foot-shock alone show a subsequent decrease in avoidance to all odorants, together with increased odor-evoked activation of reward-encoding dopaminergic neurons that innervate the mushroom body. Phantom reward, delivered via optogenetic activation of reward-encoding dopaminergic neurons, increases subsequent odour-avoidance. While the findings are valuable to the field, there are aspects of the work that are incomplete, and some of the conclusions and terminology are also not completely justified; three major issues include : (a) the use of the term "priming" to describe this form of learning seems inappropriate and inconsistent with the accepted definition of this term; (b) a key 1998 publication with an initial description of this behavioural phenomenon needs to be cited and presented as context; and (c) the work on reward induced increase in odor-aversion seems relatively preliminary.
-
Reviewer #1 (Public review):
Summary:
The authors present an investigation of associative learning in Drosophila in which a previous exposure to an aversive stimulus leads to an increase in approach behaviors to a novel odor relative to a previously paired odor or no odor (air). Moreover, this relative increase is larger compared to that of a control group - i.e., presented with a (different) odor only. Evidence for the opposite effect with an appetitive stimulus, delivered indirectly by optogenetically activating sugar sensory neurons, which leads to a reduction in approach behavior to a novel odor, was also presented. The olfactory memory circuits underpinning these responses, which the authors refer to as 'priming', are revealed and include a feedback loop mediated by dopaminergic neurons to the mushroom body.
Strengths:
(1) The …
Reviewer #1 (Public review):
Summary:
The authors present an investigation of associative learning in Drosophila in which a previous exposure to an aversive stimulus leads to an increase in approach behaviors to a novel odor relative to a previously paired odor or no odor (air). Moreover, this relative increase is larger compared to that of a control group - i.e., presented with a (different) odor only. Evidence for the opposite effect with an appetitive stimulus, delivered indirectly by optogenetically activating sugar sensory neurons, which leads to a reduction in approach behavior to a novel odor, was also presented. The olfactory memory circuits underpinning these responses, which the authors refer to as 'priming', are revealed and include a feedback loop mediated by dopaminergic neurons to the mushroom body.
Strengths:
(1) The study includes a solid demonstration of the effect of the valence of a previous stimulus on sensory preferences, with an increase or decrease in preference to novel over no odor following an aversive or appetitive stimulus, respectively.
(2) The demonstration of bidirectional effects on odor preferences following aversive or rewarding stimuli is compelling.
(3) The evidence for distinct neural circuits underpinning the odor preferences in each context appears to be robust.
Weaknesses:
(1) The conclusions regarding the links between neural and behavioral mechanisms are mostly well supported by the data. However, what is less convincing is the authors' argument that their study offers evidence of 'priming'. An important hallmark of priming, at least as is commonly understood by cognitive scientists, is that it is stimulus specific: i.e., a repeated stimulus facilitates response times (repetition priming), or a repeated but previously ignored stimulus increases response times (negative priming). That is, it is an effect on a subsequent repeated stimulus, not ANY subsequent stimulus. Because (prime or target) stimuli are not repeated in the current experiments, the conditions necessary for demonstrating priming effects are not present. Instead, a different phenomenon seems to be demonstrated here, and one that might be more akin to approach/avoidance behavior to a novel or salient stimulus following an appetitive/aversive stimulus, respectively.
(2) On a similar note, the authors' claim that 'priming' per se has not been well studied in non-human animals is not quite correct and would need to be revised. Priming effects have been demonstrated in several animal types, although perhaps not always described as such. For example, the neural underpinnings of priming effects on behavior have been very well characterized in human and non-human primates, in studies more commonly described as investigations of 'response suppression'.
(3) The outcome measure - i.e., difference scores between the two odors or odor and non-odor (i.e., the number of flies choosing to approach the novel odor versus the number approaching the non-odor (air)) - appears to be reasonable to account for a natural preference for odors in the mock-trained group. However, it does not provide sufficient clarification of the results. The findings would be more convincing if these relative scores were unpacked - that is, instead of analyzing difference scores, the results of the interaction between group and odor preference (e.g., novel or air) (or even within the pre- and post-training conditions with the same animals) would provide greater clarity. This more detailed account may also better support the argument that the results are not due to conditioning of the US with pure air.
-
Reviewer #2 (Public review):
The manuscript by Yang et al. investigates how a prior experience (notably by the activation of sensory/reinforcing dopaminergic neurons) alters olfactory response and memory expression in Drosophila. They refer to a priming effect with the definition: "Priming is a process by which exposure to a stimulus affects the response to a subsequent stimulus in Humans". The authors observed that exposing flies to a series of shocks (or the optogenetic activation of aversively reinforcing dopaminergic neurons) decreases ensuing odour avoidance. Conversely, optogenetic activation of sweet-sensing neurons increases following odour avoidance. They proposed that the reduced odour avoidance was due to the involvement of reward dopaminergic neurons involved during shock (or the optogenetic activation of aversively …
Reviewer #2 (Public review):
The manuscript by Yang et al. investigates how a prior experience (notably by the activation of sensory/reinforcing dopaminergic neurons) alters olfactory response and memory expression in Drosophila. They refer to a priming effect with the definition: "Priming is a process by which exposure to a stimulus affects the response to a subsequent stimulus in Humans". The authors observed that exposing flies to a series of shocks (or the optogenetic activation of aversively reinforcing dopaminergic neurons) decreases ensuing odour avoidance. Conversely, optogenetic activation of sweet-sensing neurons increases following odour avoidance. They proposed that the reduced odour avoidance was due to the involvement of reward dopaminergic neurons involved during shock (or the optogenetic activation of aversively reinforcing dopaminergic neurons). They indeed show the involvement of reward dopaminergic neurons innervating the mushroom body (the fly learning and memory centre) during shock preexposure. Recording (calcium activity) from reward dopaminergic neurons before and after shock preexposure shows that only a small subset of dopaminergic neurons innervating the mushroom body γ4 compartment increases their response to odour after shock. They then showed the requirement of the γ4 reward dopaminergic neurons during shock preexposure on ensuing odour avoidance. They also tested the role of the dopamine receptor in the mushroom body. They finally recorded from different mushroom body output neurons, including the one (MBON-γ4γ5) likely affected by the increased activity of the corresponding γ4 reward dopaminergic neurons after shock preexposure. They recorded odour-evoked responses from these neurons before and after shock preexposure, but did not find any plasticity, while they found a logical effect during spaced cycles of aversive training.
Overall, the study is very interesting with a substantial amount of behavioural analysis and in vivo 2-photon calcium imaging data, but some major (and some minor) issues have to be resolved to strengthen their conclusions.
(1) According to neuropsychological work (Henson, Encyclopedia of Neuroscience (2009), vol. 7, pp. 1055-1063), « Priming refers to a change in behavioral response to a stimulus, following prior exposure to the same, or a related, stimulus. Examples include faster reaction times to make a decision about the stimulus, a bias to produce that stimulus when generating responses, or the more accurate identification of a degraded version of the stimulus". Or "Repetition priming refers to a change in behavioural response to a stimulus following re-exposure" (PMID: 18328508). I therefore do not think that the effects observed by the authors are really the investigation of the neural mechanisms of priming. To me, the effect they observed seems more related to sensitisation, especially for the activation of sweet-sensing neurons. For the shock effect, it could be a safety phenomenon, as in Jacob and Waddell, 2020, involving (as for sugar reward) different subsets for short-term and long-term safety.
(2) The author missed the paper from Thomas Preat, The Journal of Neuroscience, October 15, 1998, 18(20):8534-8538 (Decreased Odor Avoidance after Electric Shock in Drosophila Mutants Biases Learning and Memory Tests). In this paper, one of the effects observed by the authors has already been described, and the molecular requirement of memory-related genes is investigated. This paper should be mentioned and discussed.
(3) Overall, the bidirectional effect they observed is interesting; however, their results are not always clear, and the use of a delta PI is sometimes misleading. The authors have mentioned that shocks induced attraction to the novel odour, while they should stick to the increase or decrease in preference/avoidance. As not all experiments are done in parallel logic, it is not always easy to understand which protocol the authors are using. For example, only optogenetics is used in the appetitive preexposure. Does exposing flies to sugar or activating reward dopaminergic neurons also increase odour avoidance? The observed increased odour avoidance after optogenetic activation of sweet-sensing neurons involve reward (e.g., decreased response) and/or punishment (e.g., increased response) to increase odour avoidance? The author should always statistically test the fly behavioural performances against 0 to have an idea of random choice or a clear preference toward an odour. On the appetitive side, the internal hunger state would play an important role. The author should test it or at least discuss it.
(4) The authors found a discrepancy between genetic backgrounds; sometimes the same odour can be attractive or aversive. Different effects between the T-maze and the olfactory arena are found. The authors proposed that: "Punishment priming effect was still not detected, probably due to the insensitivity of the optogenetic arena". This is unclear to me, considering all prior work using this arena. The author should discuss it more clearly. They mentioned that flies could not be conditioned with air and electric shock. However, flies could be conditioned with the context + shock, which is changing in the T-maze and not in the optogenetic area.
-
