Diverse states and stimuli tune olfactory receptor expression levels to modulate food-seeking behavior

  1. Ian G McLachlan
  2. Talya S Kramer
  3. Malvika Dua
  4. Elizabeth M DiLoreto
  5. Matthew A Gomes
  6. Ugur Dag
  7. Jagan Srinivasan
  8. Steven W Flavell

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This study asks how diverse signals are integrated at the cellular level to generate adaptive behaviors. The authors show that prolonged food deprivation (i.e. fasting) of C. elegans broadly alters gene expression in food sensing neurons, thereby altering foraging behavior and chemosensory neuron responses to food. The fasting-induced genes include many chemoreceptors, one of which mediates responses to specific volatile components of food. Finally, they show that food controls the expression of a fasting-induced chemoreceptor via multiple external (i.e. sensory) and internal (potentially metabolic) cues. The paper is of importance to scientists with an interest in adaptive behaviour as well as the cellular and molecular mechanisms underlying integration of stimuli.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Animals must weigh competing needs and states to generate adaptive behavioral responses to the environment. Sensorimotor circuits are thus tasked with integrating diverse external and internal cues relevant to these needs to generate context-appropriate behaviors. However, the mechanisms that underlie this integration are largely unknown. Here, we show that a wide range of states and stimuli converge upon a single Caenorhabditis elegans olfactory neuron to modulate food-seeking behavior. Using an unbiased ribotagging approach, we find that the expression of olfactory receptor genes in the AWA olfactory neuron is influenced by a wide array of states and stimuli, including feeding state, physiological stress, and recent sensory cues. We identify odorants that activate these state-dependent olfactory receptors and show that altered expression of these receptors influences food-seeking and foraging. Further, we dissect the molecular and neural circuit pathways through which external sensory information and internal nutritional state are integrated by AWA. This reveals a modular organization in which sensory and state-related signals arising from different cell types in the body converge on AWA and independently control chemoreceptor expression. The synthesis of these signals by AWA allows animals to generate sensorimotor responses that reflect the animal’s overall state. Our findings suggest a general model in which sensory- and state-dependent transcriptional changes at the sensory periphery modulate animals’ sensorimotor responses to meet their ongoing needs and states.

Article activity feed

  1. Version published to 10.7554/elife.79557 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    McLachlan and colleagues find surprisingly widespread transcriptional changes occurring in C. elegans neurons when worms are prevented from smelling food for 3 hours. Focusing most of the paper on the transcription of a single olfactory receptor, the authors demonstrate many molecular pathways across a variety of neurons that can cause many-fold changes in this receptor. There is some evidence that the levels of this single receptor can adjust behavior. I believe that the wealth of mostly very convincing data in this paper will be of interest to researchers who think about sensory habituation, but I think the authors' framing of the paper in terms of hunger is misleading.

    There is a lot to like about this paper, but I just cannot get over how off the framing is. Unless I am severely misunderstanding, the paper is about sensory habituation, but the word habituation is not used in the paper. Instead, we hear very often about hunger (6x), state (92x), and sensorimotor things (23x). This makes little sense to me. The worms are "fasted" (111x) for 3 hours, but most of the expression changes are reversed if the worms can smell, but not eat, the food. And I've heard about the fasted state, noting that worms don't eat more food after this type of "fasting". So what is with all of this hunger/state discussion?

    We think that the most straightforward interpretation of our data is that both sensory experience and internal nutritional state modulate str-44 expression. However, we agree that in the previous manuscript draft there was a disproportionate emphasis on state (as compared to sensory experience). The revised manuscript corrects this. However, several results in the manuscript do suggest that state is important, so we have not removed this from the manuscript. The lines of evidence that suggest this are:

    (1) Animals exposed to inedible aztreonam-treated food show an increase in str-44 expression compared to animals exposed to untreated, ingestable food. Thus, food ingestion acts to suppress str-44 expression (Figure 1E).

    (2) Animals exposed to food odor in the absence of food show an intermediate level of str-44 expression between “on bacteria” and “off bacteria” controls (Figure 1E). This incomplete suppression suggests that food odors alone can not explain the suppression of str-44 expression in well-fed animals.

    (3) Animals that lack intestinal rict-1, a component of the TOR2 nutrient-sensing complex, show an increase in str-44 expression, which suggests that nutrient sensing in the intestine impacts str-44 expression (Figure 5).

    (4) When animals are off food, osmotic stress inhibits the upregulation of str-44 (Figure 1G), reduces the enhanced behavioral sensitivity to butyl acetate (Figure 2G), and reduces the enhanced AWA activity in response to food (Figure 3). This physiological stressor provides a competing state that also impacts str44 expression.

    We apologize for not adequately describing how three hours of fasting impacts C. elegans behavior in the initial submission. This is obviously a key piece of information and we have corrected this in the revised manuscript. [lines 68-70; 123-126] Regarding pharyngeal pumping rates, C. elegans typically exhibits pharyngeal pumping at a near-maximal rate on the OP50 laboratory diet even when well-fed.

    Consequently, even much longer starvation times will fail to induce more feeding under these conditions. However, many other feeding-related behaviors do change with three hours of fasting, such as velocity on and off food, turning rates, roaming/dwelling behavior on OP50 food, and sensitivity to odorants. Thus, three hours of fasting is sufficient to impact several food search behaviors.

    To more directly address whether sensory habituation in AWA alters str-44 expression, we performed an additional experiment. We exposed wild-type animals to the str-44 odorants butyl acetate or propyl acetate and measured str-44 expression. If habituation explains this effect (e.g. repeated exposure of an odorant reduces transcription/translation of the receptor), we would expect that exposure to these odorants would reduce str-44 expression in “off bacteria” animals. However, we observed no differences between odor-exposed animals and controls. [Figure 4-figure supplement 2B; lines 414-421]

    And the discussion of internal states is often naïve. In the second paragraph of the introduction, we are told that "Recent work has identified specific cell populations that can induce internal states", beginning with AgRP neurons, which have been known to control the hunger state in mammals for nearly 40 years |||(Clark J. T., Kalra P. S., Crowley W. R., Kalra S. P. (1984). Neuropeptide Y and human pancreatic polypeptide stimulate feeding behavior in rats. Endocrinology 115 427-429. Hahn T. M., Breininger J. F., Baskin D. G., Schwartz M. W. (1998). Coexpression of Agrp and NPY in fasting-activated hypothalamic neurons. Nat. Neurosci. 1 271-272). Instead, the authors cite three papers from 2015, whose major contribution was to show that AgRP activity surprisingly decreases when animals encounter food. These papers absolutely did not identify AgRP neurons as inducing internal states or driving behavioral changes typical of hunger (Aponte, Y., Atasoy, D., and Sternson, S. M. (2011). AGRP neurons are sufficient to orchestrate feeding behavior rapidly and without training. Nat. Neurosci. 14, 351-355. doi: 10.1038/nn.2739; Krashes, M. J., Koda, S., Ye, C., Rogan, S. C., Adams, A. C., Cusher, D. S., et al. (2011). Rapid, reversible activation of AgRP neurons drives feeding behavior in mice. J. Clin. Invest. 121, 1424-1428. Doi: 10.1172/jci46229). Nor did Will Allen's work in Karl Deisseroth's lab discover neurons that drive thirst behaviors.

    We agree that this introductory paragraph did not do justice to the literature and improperly cited only relatively recent work. We have addressed this oversight. [lines 48-53]

    Later in the same paragraph, we hear that: "However, animals can exhibit more than one state at a time, like hunger, stress, or aggression. Therefore, the sensorimotor pathways that implement specific motivated behaviors, such as approach or avoidance of a sensory cue, must integrate information about multiple states to adaptively control behavior." This is undoubtedly true, but it's not clear what it has to do with any of the data in this paper - I don't even think this is really about hunger, much less the interaction between hunger and other drives.

    To summarize: I think the authors could give the writing of the paper a serious rethink. I want to stay far away from telling people how to write their papers, so if the authors insist on framing this obviously sensory paper as being about hunger and sensorimotor circuitry I think they should at least explain to their readers why they are doing that in light of the evidence against it (and I think they should state clearly that worms don't actually eat more in this fasted state).

    Please see the comments above that address these concerns.

    I was also surprised by how unsurprised the authors seemed by the incredibly widespread changes they observed after 3 hours away from food. Over 1400 genes change at least 4-fold? That seems like a lot to me. But the authors, maybe for narrative reasons, only comment on how many of them are GPCRs (16.5%, which isn't that much of an overrepresentation compared to 8.5% in the whole genome). For me, these widespread and strong changes are much of the takeaway from this paper. But it does make you wonder how important the activity of one particular GPCR (selected more or less randomly) could be to the changes the worm undergoes when it can't smell food.

    We agree with the reviewer that given the widespread gene expression changes in fasted animals, the changes in AWA are only a small part of the picture. We have added a discussion of this to the revised manuscript. In addition, we provide some discussion of how our gene expression profiling results relate to others in the field. For example, animals that lack the fasting-responsive transcription factor DAF-16 have been shown to have >3,000 genes differentially expressed relative to controls (Kaletsky, Lakhina et al., 2016). Given the large number of genes changing in those data and in our data, it is possible that transcriptional changes are extremely widespread during fasting. [lines 588-593]

    str-44 is very convincingly upregulated when worms can't smell food, but it's clear from the data that this upregulation has very little to do with the actual lack of eating, and more with the lack of being able to sense bacteria for 3 hours. In Figure 1E, when worms are fasted, but in the presence of bacteria, receptor levels are largely unchanged (there are 5 outliers, out of ~50 samples). Since receptor expression doesn't change in this case even though the worms are in the fasted state, it cannot be "state-dependent" - unless the state is not having smelled food for the last 3 hours. And, in my opinion, that would divorce the word "state" from its ordinary meaning.

    We have more closely examined that dataset, but we don’t feel that it would be accurate to say that the aztreonam (inedible) condition matches the fed. The highest points in the aztreonam-treated condition are most visible on the plot, but the effect is driven by the bulk of the data. Even if we remove the top 5 datapoints from the aztreonam condition, the effect is still statistically significant. Moreover, we performed this experiment over multiple days and the effect was present on each day. However, the reviewer’s point is well taken that sensory experience is equally (if not more) important for str-44 regulation and the text of the initial manuscript did not properly reflect this. As described above, we have modified the revised manuscript so that it is more balanced.

    The authors argue that str-44 expression modulates food-seeking behavior in fasted worms by causing them to preferentially seek out butyl and propyl acetate. However, the behavioral data to back this up has me a little worried. For example, take Figures 2F and 2G. They are the exact same experiment: comparing how many worms choose 1:10,000 butyl acetate compared to ethanol when the worms are either fasted or fed. In the first experiment (2F), ~70% chose butyl acetate for fasted worms and ~60% for fed worms. But in the replicate, ~60% choose butyl acetate for fasted worms and ~50% for fed worms. A 10% variability in baseline behavior is fine (but not what I would call a huge state change), but when the difference between conditions is the same size as baseline variability I start to disbelieve. Can the authors explain this variability? Or am I misunderstanding?

    We and others often observe large variance in C. elegans chemotaxis behavior over time because of small changes in environmental variables such as temperature, humidity, and pressure, so it is standard to always run wild-type controls together with all experimental groups and compare within day. The experiment in Figure 2F was conducted before the others in Figure 2G and Figure 4F. However, we remain highly confident in this result – we observed a difference in fed vs starved every time that we ran this experiment, which (in sum total for wild-type) was on 6 different days, with at least 3 plates per day (40-200 worms per plate).

    And I'll say it just one last time, I think the authors are overselling their results...or at least the str-44 and AWA results (they are dramatically underselling the results that show the widespread changes in the expression level of 10% of the genome in response to not smelling food for 3 hours):

    "Our results reveal how diverse external and internal cues... converge at a single node in the C. elegans nervous system to allow for an adaptive sensorimotor response that reflects a complete integration of the animal's states."

    This implies that str-44 expression AWA is the determinant of whether a worm will act fasted or fed. I have already expressed why I don't believe this is the case (inedible bacteria experiment, Figure 1E), but just because things like osmotic stress suppress the upregulation of str-44, that doesn't mean that it is the site of convergence. It could be any of the other 1400 genes that changed 4+ fold with bacterial deprivation. And even in terms of the actual AWA neuron, it was chosen because it showed modest upregulation of chemoreceptors (1.8 fold compared to ~1.5 fold in ASE and ASG), even though chemoreceptors were highly upregulated in other neurons as well.

    We agree that AWA chemoreceptors alone are unlikely to explain all of the behavioral changes observed in an animal that has been removed from food, and we certainly did not intend to imply that str-44 expression in AWA is the central determinant of whether the animal acts as though it is fasted or fed. Rather, we have shown that str-44 expression can explain some of these behavioral changes. We have added language throughout the manuscript to indicate that we expect other fasting-regulated genes to be of importance. See also: response to Essential Revision #1.

    Overall, and despite my critiques (and possibly tone), I really like this paper and think there really is a lot of interesting data in there.

  3. eLife

    Evaluation Summary:

    This study asks how diverse signals are integrated at the cellular level to generate adaptive behaviors. The authors show that prolonged food deprivation (i.e. fasting) of C. elegans broadly alters gene expression in food sensing neurons, thereby altering foraging behavior and chemosensory neuron responses to food. The fasting-induced genes include many chemoreceptors, one of which mediates responses to specific volatile components of food. Finally, they show that food controls the expression of a fasting-induced chemoreceptor via multiple external (i.e. sensory) and internal (potentially metabolic) cues. The paper is of importance to scientists with an interest in adaptive behaviour as well as the cellular and molecular mechanisms underlying integration of stimuli.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    McLachlan and colleagues find surprisingly widespread transcriptional changes occurring in C. elegans neurons when worms are prevented from smelling food for 3 hours. Focusing most of the paper on the transcription of a single olfactory receptor, the authors demonstrate many molecular pathways across a variety of neurons that can cause many-fold changes in this receptor. There is some evidence that the levels of this single receptor can adjust behavior. I believe that the wealth of mostly very convincing data in this paper will be of interest to researchers who think about sensory habituation, but I think the authors' framing of the paper in terms of hunger is misleading.

    There is a lot to like about this paper, but I just cannot get over how off the framing is. Unless I am severely misunderstanding, the paper is about sensory habituation, but the word habituation is not used in the paper. Instead, we hear very often about hunger (6x), state (92x), and sensorimotor things (23x). This makes little sense to me. The worms are "fasted" (111x) for 3 hours, but most of the expression changes are reversed if the worms can smell, but not eat, the food. And I've heard about the fasted state, noting that worms don't eat more food after this type of "fasting". So what is with all of this hunger/state discussion?

    And the discussion of internal states is often naïve. In the second paragraph of the introduction, we are told that "Recent work has identified specific cell populations that can induce internal states", beginning with AgRP neurons, which have been known to control the hunger state in mammals for nearly 40 years |||(Clark J. T., Kalra P. S., Crowley W. R., Kalra S. P. (1984). Neuropeptide Y and human pancreatic polypeptide stimulate feeding behavior in rats. Endocrinology 115 427-429. Hahn T. M., Breininger J. F., Baskin D. G., Schwartz M. W. (1998). Coexpression of Agrp and NPY in fasting-activated hypothalamic neurons. Nat. Neurosci. 1 271-272). Instead, the authors cite three papers from 2015, whose major contribution was to show that AgRP activity surprisingly decreases when animals encounter food. These papers absolutely did not identify AgRP neurons as inducing internal states or driving behavioral changes typical of hunger (Aponte, Y., Atasoy, D., and Sternson, S. M. (2011). AGRP neurons are sufficient to orchestrate feeding behavior rapidly and without training. Nat. Neurosci. 14, 351-355. doi: 10.1038/nn.2739; Krashes, M. J., Koda, S., Ye, C., Rogan, S. C., Adams, A. C., Cusher, D. S., et al. (2011). Rapid, reversible activation of AgRP neurons drives feeding behavior in mice. J. Clin. Invest. 121, 1424-1428. Doi: 10.1172/jci46229). Nor did Will Allen's work in Karl Deisseroth's lab discover neurons that drive thirst behaviors. Later in the same paragraph, we hear that: "However, animals can exhibit more than one state at a time, like hunger, stress, or aggression. Therefore, the sensorimotor pathways that implement specific motivated behaviors, such as approach or avoidance of a sensory cue, must integrate information about multiple states to adaptively control behavior." This is undoubtedly true, but it's not clear what it has to do with any of the data in this paper - I don't even think this is really about hunger, much less the interaction between hunger and other drives.

    To summarize: I think the authors could give the writing of the paper a serious rethink. I want to stay far away from telling people how to write their papers, so if the authors insist on framing this obviously sensory paper as being about hunger and sensorimotor circuitry I think they should at least explain to their readers why they are doing that in light of the evidence against it (and I think they should state clearly that worms don't actually eat more in this fasted state).

    I was also surprised by how unsurprised the authors seemed by the incredibly widespread changes they observed after 3 hours away from food. Over 1400 genes change at least 4-fold? That seems like a lot to me. But the authors, maybe for narrative reasons, only comment on how many of them are GPCRs (16.5%, which isn't that much of an overrepresentation compared to 8.5% in the whole genome). For me, these widespread and strong changes are much of the takeaway from this paper. But it does make you wonder how important the activity of one particular GPCR (selected more or less randomly) could be to the changes the worm undergoes when it can't smell food.

    str-44 is very convincingly upregulated when worms can't smell food, but it's clear from the data that this upregulation has very little to do with the actual lack of eating, and more with the lack of being able to sense bacteria for 3 hours. In Figure 1E, when worms are fasted, but in the presence of bacteria, receptor levels are largely unchanged (there are 5 outliers, out of ~50 samples). Since receptor expression doesn't change in this case even though the worms are in the fasted state, it cannot be "state-dependent" - unless the state is not having smelled food for the last 3 hours. And, in my opinion, that would divorce the word "state" from its ordinary meaning.

    The authors argue that str-44 expression modulates food-seeking behavior in fasted worms by causing them to preferentially seek out butyl and propyl acetate. However, the behavioral data to back this up has me a little worried. For example, take Figures 2F and 2G. They are the exact same experiment: comparing how many worms choose 1:10,000 butyl acetate compared to ethanol when the worms are either fasted or fed. In the first experiment (2F), ~70% chose butyl acetate for fasted worms and ~60% for fed worms. But in the replicate, ~60% choose butyl acetate for fasted worms and ~50% for fed worms. A 10% variability in baseline behavior is fine (but not what I would call a huge state change), but when the difference between conditions is the same size as baseline variability I start to disbelieve. Can the authors explain this variability? Or am I misunderstanding?

    And I'll say it just one last time, I think the authors are overselling their results...or at least the str-44 and AWA results (they are dramatically underselling the results that show the widespread changes in the expression level of 10% of the genome in response to not smelling food for 3 hours):

    "Our results reveal how diverse external and internal cues... converge at a single node in the C. elegans nervous system to allow for an adaptive sensorimotor response that reflects a complete integration of the animal's states."

    This implies that str-44 expression AWA is the determinant of whether a worm will act fasted or fed. I have already expressed why I don't believe this is the case (inedible bacteria experiment, Figure 1E), but just because things like osmotic stress suppress the upregulation of str-44, that doesn't mean that it is the site of convergence. It could be any of the other 1400 genes that changed 4+ fold with bacterial deprivation. And even in terms of the actual AWA neuron, it was chosen because it showed modest upregulation of chemoreceptors (1.8 fold compared to ~1.5 fold in ASE and ASG), even though chemoreceptors were highly upregulated in other neurons as well.

    Overall, and despite my critiques (and possibly tone), I really like this paper and think there really is a lot of interesting data in there.

  5. eLife

    Reviewer #2 (Public Review):

    This study addressed a broad question of the neuroscience of how diverse signals are integrated at the cellular level to generate adaptive behaviors. The authors systematically dissected the mechanisms of integration of diverse physiological cues in an olfactory neuron to change olfactory receptor expression.

    Major strengths of the manuscript include (i) the broad investigations of many candidate pathways providing a big picture of the signalling working up-stream of str-44 expression regulation, (ii) the identification of compounds that activate str-44 and srd-28 receptors, and (iii) the connection that is made with cellular physiology and behavior to demonstrate the relevance of the mechanism into play. The pan-neuronal ribotagging data in fed and starved animals will furthermore constitute a valuable resource for neuroscience, particularly for the C. elegans community.

    Aspects that could be improved relate to the representation of statistical test results and the validity/interpretation of experiments attempting to tease apart the "sensory" versus "metabolic" contribution of food deprivation. Whereas the main conclusions are largely supported by the data, and the proposed model is plausible, several aspects/results should be taken into consideration for a deeper discussion of the results.

  6. eLife

    Reviewer #3 (Public Review):

    McLachlan and colleagues investigated the molecular mechanisms that lead to an adaptive response of a single pair of chemosensory neurons. Taking advantage of the single cell resolution of the C. elegans nervous system as well as genetic and behavioural tools, they observe that after fasting, animals show an altered profile in the expression of chemosensory GPCRs. They focused on two GPCR genes, str-44 and srd-28, both highly upregulated in AWA neurons after fasting which correlated with altered chemosensory behaviour. They further showed that upregulation and behavioural changes depend on both, external (food cues and osmotic stress) as well as internal signals from food-sensing neurons and the intestine. Artificially increasing str-44 and srd-28 by overexpression in AWA mimics the fasted state. They provide evidence that STR-44 is a chemoreceptor of the attractants propyl acetate and butyl acetate by ectopically expressing str-44 and srd-28 in ASH. They provide evidence for a model in which a combination of pathways together modulates str-44 expression in AWA. These include other food-sensing neurons modulating the activity of AWA neurons, intestinally expressed factors that are involved in metabolism as well as pathways detecting environmental stress. A chemosensory role for the proposed phenotypes for STR-44 could be strengthened by providing AWA calcium imaging and behavioural evidence of chemosensory defects of mutants lacking str-44.

  7. Version published to 10.1101/2022.04.27.489714v1 on bioRxiv