A multi-input optic glomerulus mediates opposing behavioral responses to visual objects

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife Assessment

    This study provides important findings on the neural mechanisms underlying visually guided avoidance in Drosophila. The results identify LC10d visual projection neurons and their downstream partners DNa10 as drivers of object avoidance, in contrast to the courtship-promoting LC10a neurons. The evidence supporting a role for LC10d is solid, although some analyses would benefit from improved statistical quantification and presentation. In contrast, the evidence that DNa10 specifically mediates LC10d-dependent avoidance is incomplete, due to potential DNa04 contamination and other experimental confounds.

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

Prey, predators or conspecifics are first detected as visual objects in many seeing animals. Vision guides behavioral actions towards or away from these objects. An error in this visual perception could prove fatal. How object information is untangled to avoid errors remains unclear. Here we show that LC10d visual projection neurons in Drosophila melanogaster mediate avoidance of visual objects in the absence of a chemosensory profile. LC10d neurons are broadly tuned to objects and project to the same retinorecipient brain region that receives inputs from LC10a neurons, which are required for tracking. The descending neurons DNa10 are directly downstream of the anterior-facing LC10d sub-population and mediate LC10d-dependent avoidance. Our work demonstrates the use of two similar neuron types and chunking of the visual field into zones as strategies to disentangle similar sets of visual cues requiring nearly opposite behavioral responses.

Article activity feed

  1. Author response:

    We thank the reviewers and eLife editors for the constructive and important comments on our manuscript. We will address them all in the next revision.

  2. eLife Assessment

    This study provides important findings on the neural mechanisms underlying visually guided avoidance in Drosophila. The results identify LC10d visual projection neurons and their downstream partners DNa10 as drivers of object avoidance, in contrast to the courtship-promoting LC10a neurons. The evidence supporting a role for LC10d is solid, although some analyses would benefit from improved statistical quantification and presentation. In contrast, the evidence that DNa10 specifically mediates LC10d-dependent avoidance is incomplete, due to potential DNa04 contamination and other experimental confounds.

  3. Reviewer #1 (Public review):

    Summary:

    In this paper, Ribeiro and colleagues investigate male orienting behavior during courtship to identify the neural pathways involved in visually-guided avoidance. The major claims are:

    (1) Before courtship has begun, the male generally avoids other flies. This avoidance depends on LC10d neurons (Figures 1 and 2).

    (2) LC10d neurons are broadly tuned to visual objects, and they mediate visual avoidance behavior (Figure 3).

    (3) Unilateral activation of LC10d neurons drives contralateral turns (Figure 4).

    (4) Anterior-facing LC10d neurons preferentially connect to a descending neuron called DNa10, and these neurons mediate avoidance behaviors (Figure 5).

    Overall the initial observations are supported with solid behavioral evidence, although there are issues with clarity of anayses and writing that make it difficult to evaluate at times. I thought that the latter two claims are not well supported by the data and analyses as presented. I think that improving the analysis, statistics, and, in some cases, the claims being made with the data could overcome these weaknesses.

    Strengths:

    (1) Figures 1 and 2 show solid behavioral evidence with some interesting and clever analyses, including the change detection and event-triggered analyses.

    (2) The visual tuning characterization (Figure 3) shows strong evidence that these neurons are more broadly tuned to visual objects than LC10a.

    Weaknesses:

    I found this manuscript to be a very difficult read, and I think it needs a major overhaul of the writing throughout. At multiple points, the writing is very unclear. I also ran into some points where the claims in the text are not consistent with the data. Some representative examples include:

    (1) In lines 209-210, it is stated that "Both forward accelerations and decelerations contributed to an increase in female frontal positions (Figure 2D)". However, Figure 2D seems to show that there is a decrease in female frontal positions for accelerations (all genotypes) and perhaps a weak increase in frontal position for the two silencing genotypes.

    (2) Related to the plots showing the y-axis projection of the heatmaps (e.g. Figure 1 G, I):

    a) I believe what the authors want is the orientation of the male relative to the location of the female; to quantify whether he is facing her. These plots show the distance, but then the orientation is not clear beyond anterior vs. posterior.

    b) How were these projections normalized? Are the different lines comparable in amplitude? Shouldn't these have units of % rather than a.u.?

    c) The quantification (AUC in the anterior field) is just mentioned in the figure legend with p-values. It would be useful to show the AUC values to get a sense of fly-to-fly variability and effect sizes.

    (3) There is analysis of the LOV-LexA experiments (Figure 4). With the way the data are currently presented, I don't feel confident in the claim that LC10d drives contralateral turning.

    (4) Figure 4, D and E: It should be clearer (in the text, labels on plots, and/or figure legend) what the "unilateral expression factor" is. The text (line 319) refers to a count of cells expressing on each side, but I can't figure out how it was calculated. Do negative values correspond to bias on one half of the brain while positive values correspond to a bias on the other side?

    (5) Figure 5: A straightforward reading of the DN blocking experiments would suggest that silencing DNa04, but not DNa10, changes male orientation behavior. The data in Figure 5 either only show results from the combined DNa04/10 driver or show negative results for the DNa10-only driver.

  4. Reviewer #2 (Public review):

    Summary:

    This work by Ribeiro and colleagues examines circuit mechanisms by which the fruit fly regulates its behavioral response to moving objects. Previous studies, including one by the authors, identified that LC10a visual projection neurons promote tracking of female flies during male courtship behavior. In the present paper, the authors focus on LC10d visual projection neurons and show that these neurons mediate avoidance of visual objects, thereby revealing opposing functions performed by closely related visual projection neurons. In contrast to LC10a's role in promoting courtship pursuit, LC10d appears to promote object avoidance before courtship begins, as silencing LC10d results in increased tracking of visual objects. In vivo functional imaging experiments show that LC10d is broadly tuned to moving objects, and analyses of the connectome and subsequent behavioral experiments identify a class of descending neurons, DNa10, as a downstream behavioral effector of a subset of LC10d with frontal receptive fields. Together, this study suggests an important role of LC10d in regulating behavioral responses to moving visual objects.

    Strengths:

    (1) Specific genetic drivers were used to perform behavioral manipulation and calcium imaging, revealing behavioral roles and stimulus tuning of previously uncharacterized LC10d neurons.

    (2) Many experiments were performed in parallel for LC10d and LC10a, allowing for direct comparisons of behavioral phenotypes and stimulus responses between these two neuron types.

    (3) In addition to the role of LC10d neurons in male object avoidance, the paper also investigates their role in females and shows that this avoidance maneuver (i.e., displacing the object out of the frontal visual field) in females facilitates male courtship behavior, suggesting how a sexually isomorphic behavioral response can lead to sexually dimorphic consequences during courtship.

    Weaknesses:

    (1) The study would benefit from more statistical rigor. Many behavioral observations central to the paper's conclusion are not accompanied by statistical testing; for example, the apparent pre-courtship avoidance of the visual object and its suppression upon LC10d silencing.

    (2) Many important panels report only center values without error bars or confidence intervals; thus, it is not possible to assess the variance of data (e.g., fly-to-fly variability), making it difficult to judge the robustness of the effects.

  5. Reviewer #3 (Public review):

    At a distance, animals have to make decisions on whether to approach or flee. The sensory information at such distances is often ambiguous and contains only sparse visual information. Primarily using the pre-courtship behavior of D. melanogaster as a model, Ribeiro et al. investigated how animals respond to such sparse visual information. The authors identified that male flies predominantly avoid females before courtship, placing them in their posterior field of view. Unlike when blocking other visual projection neurons, this avoidance behavior was not found upon blocking LC10d transmission. This work is the first to identify LC10d involvement in such a behavior as well as its broad tuning to objects. Further behavioral experiments identified the involvement of DNa10, a descending neuron downstream of select anterior-facing LC10d neurons. In addition to demonstrating the use of this circuit across sexes, the motivation and context for the work are clearly described.

    Beyond its implications for Drosophila researchers, this work is important for the broader circuit neuroscience community by generating evidence for the use of visual chunking during complex social interactions. The strength of evidence supports the majority of their conclusions, but there are aspects of the manipulations to DNa10 that need to be clarified and further described.

    In Figure 5, the authors use two lines for manipulating DNa10 activity, one of which contains a contaminating neuron (DNa04). For Figure 5E and 5I (lines 344 - 355), the authors state the effects of DNa10/04, but the results from the DNa10 alone data are either not shown or not performed. In the male CNS dataset, the DNa04 second top major target is also LC10d, suggesting a confound for these experiments that is not discussed. Additionally, the authors do not discuss the number of DNa10 cells or the expression level across these two lines that could account for any differences found. The authors should discuss these potential effects in the text.

  6. Authors published a paper in Biorxiv, they used python-bx in the study. Including #RRIDs will make this less ambiguous. SciScore made a table with this resource, see “Automated Services” module (download as csv, xml or #jats) #RRID #OpenScience

  7. A multi-input optic glomerulus mediates opposing behavioral responses to visual objects https://www.biorxiv.org/content/10.64898/2026.01.26.701771v1

  8. A multi-input optic glomerulus mediates opposing behavioral responses to visual objects https://www.biorxiv.org/content/10.64898/2026.01.26.701771v1