Metabolically active Ecc15 triggers dispersal behavior in Drosophila larvae

To characterize larval behavior when exposed to a source of food contaminated with Ecc15, baker’s yeast (Saccharomyces cerevisiae), a nutritive substrate supporting optimal larval growth [27], was mixed with increasing concentrations of stationary-phase Ecc15 cultured overnight (S1A Fig). This yeast–bacteria mixture was then placed on nutrient-rich LB agar plates and left to incubate at room temperature for 30 minutes, a duration corresponding to the early phase of bacterial regrowth in static liquid culture (S1B Fig). Wild-type (w¹¹¹⁸) second-instar larvae were subsequently transferred directly onto the contaminated food for behavioral assays (Fig 1A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Metabolically active Ecc15 cells trigger rapid and dose-dependent escape of Drosophila larvae.

(A) Schematic of the assay used to evaluate larval responses to Ecc15-contaminated food. (B) Quantification of larval dispersal from uncontaminated yeast (N.C.) and yeast contaminated with increasing concentrations of Ecc15 (OD₆₀₀ = 5, 10, 20, and 40) over time. (C) Comparison of larval dispersal on uncontaminated yeast versus yeast contaminated with Ecc15, E. coli, P. entomophila and L. brevis at identical bacterial densities (OD₆₀₀ 40). (D) Escape behavior of fed larvae and larvae starved for 5 h on uncontaminated or Ecc15-contaminated yeast. (E) Dispersal of larvae exposed to live versus heat-killed Ecc15. (F) Effect of nutrient-rich LB agar, carbohydrate-rich but nitrogen-limited apple agar, low-nutrient agar, and nutrient-free agarose substrates on Ecc15-induced escape. For this and all subsequent figures, quantified dispersal assay results are shown as mean ± SD of technical triplicates from a representative biological experiment. Each experiment was independently repeated at least three times, using groups of ≥40 larvae per replicate. Statistical analyses were performed using a mixed-effects model (REML) with time and condition as fixed factors, (ns = not significant; *p < 0.05; **p < 0.01; ****p < 0.0001). Comparisons between selected conditions are shown. For all experiments, raw data and full statistical outputs are provided in S1 Data.

https://doi.org/10.1371/journal.ppat.1014506.g001

Exposure to Ecc15-contaminated food caused larvae to disperse within minutes, and this response increased both over time and with bacterial load (Fig 1B; corresponding CFU values in S1C Fig). At bacterial densities equivalent to OD₆₀₀ 20–40, more than 80% of the larvae had dispersed within 2.5 hours, whereas lower concentrations (OD₆₀₀ 5–10) elicited a weaker initial response. Importantly, CFU counts remained stable during the 30 min pre-incubation period on LB agar across all tested concentrations, indicating that differences in larval behavior are unlikely to result from differential bacterial growth prior to larval exposure (S1D Fig). Once larvae moved away from the contaminated food, they remained off the substrate for the rest of the observation period, suggesting active avoidance (Figs 1B, S2). After 24 hours, nearly all larvae had abandoned the Ecc15-contaminated yeast at every concentration tested, whereas larvae consistently remained on uncontaminated food (S2 Fig). Based on the rapidity and reproducibility of the response at OD₆₀₀ 40, this concentration was used in subsequent experiments unless otherwise indicated.

The robustness of this dispersal behavior was verified across multiple Drosophila melanogaster genetic backgrounds including Canton-S, Oregon-R, and yellow white larvae, all of which displayed rapid escape from Ecc15-contaminated food (S3 Fig). To examine the temporal stability of this behavior, yeast–Ecc15 mixtures stored on LB agar plates at room temperature for 2, 4, or 6 days were tested. In every case, larvae continued to disperse robustly and extensively, demonstrating that this escape response is stable and independent of the age of the bacterial preparation (S4 Fig).

To evaluate the specificity of the escape response, larvae were exposed to yeast mixed with different bacterial species, including the facultative anaerobic Escherichia coli (E. coli), a non-pathogenic member of the Enterobacteriaceae family, Pseudomonas entomophila (P. entomophila), an established entomopathogen [7], and Levilactobacillus brevis (L. brevis), a common member of the Drosophila microbiota generally considered a commensal, but which can exert deleterious effects depending on host immune status and microbiota composition [16,28]. All bacterial suspensions were prepared from stationary-phase cultures and applied to LB agar plates using the same procedure as for Ecc15. While larvae rapidly abandoned food contaminated with Ecc15, they remained on yeast containing E. coli, P. entomophila, L. brevis, as well as on uncontaminated yeast (Figs 1C, S5 Fig and S1–S3 Movies). These results show that the dispersal behavior is not a general response to bacterial presence or pathogenicity, but is instead likely triggered by cues specific to Ecc15.

Since starvation has been shown to influence pathogen-induced behavioral decisions in larvae exposed to P. entomophila [26], the impact of larval nutritional state on escape from Ecc15-contaminated food was examined. No significant difference in dispersal was observed between fed larvae and larvae starved for 5 h on PBS-soaked filter paper. Both groups rapidly abandoned Ecc15-contaminated food, indicating that this escape response is largely insensitive to larval nutritional state (Fig 1D).

To determine whether bacterial viability is required for larval dispersal, larvae were exposed to either live or heat-killed Ecc15. Only live Ecc15 bacteria induced dispersal, whereas heat-killed bacteria produced no response (Fig 1E). Larvae also moved away from food containing live Ecc15 mixed with either metabolically active or inactive yeast, showing that the behavior is driven by bacterial rather than yeast-derived signals (S6 Fig). Consistently, larvae withdrew from Ecc15-contaminated ripe mashed banana but remained on uncontaminated food, demonstrating that this behavior occurs across distinct nutritive substrates (S7 Fig). Together, these findings support the conclusion that larval dispersal is triggered specifically by bacterial factors associated with live Ecc15 and is not restricted to a particular food substrate.

Requirements for bacterial metabolism were further tested by placing yeast–Ecc15 mixtures on media with varying nutrient compositions, including nutrient-rich LB agar, carbohydrate-rich but nitrogen-limited apple agar, low-nutrient agar, and nutrient-free agarose. Larvae dispersed rapidly on both LB and apple-agar, whereas no dispersal was observed on agar or agarose, where bacterial metabolic activity is expected to be minimal or absent (Figs 1F and S8). These findings indicate that larval escape requires metabolically active bacteria and occurs only under conditions that support bacterial metabolic activity, independent of substrate nutrient composition.

Taken together, these results demonstrate that Drosophila melanogaster larvae exhibit a rapid, robust dispersal response to Ecc15-contaminated food, triggered by signals from metabolically active bacteria, and dependent on substrate-driven bacterial metabolic activity.

Larval avoidance of Ecc15 emerges after initial non-selective food approach

To determine whether Drosophila larvae require direct contact with Ecc15 contaminated yeast to initiate dispersal, an avoidance assay was performed. Larvae were positioned at a distance from the food source under three conditions: uncontaminated yeast, yeast inoculated with E. coli, or yeast inoculated with Ecc15 (Fig 2A). In all conditions, larvae rapidly oriented toward the food and entered it, confirming an attraction to nutritional cues (Fig 2B). However, upon contacting yeast contaminated with Ecc15, larvae displayed a rapid escape response and left the substrate shortly thereafter (Fig 2B). By contrast, larvae that reached uncontaminated yeast or yeast containing E. coli remained on the food throughout the assay (Fig 2B). These observations indicate that larvae do not detect Ecc15 at a distance but instead respond to cues perceived upon direct contact with the contaminated food.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Initial non-selective approach precedes larval escape from Ecc15-contaminated food.

(A) Schematic of the assay monitoring larval behavior toward a distant food source. (B) Quantification of larvae located outside the food over time when placed at a distance from uncontaminated yeast (N.C.) or yeast contaminated with Ecc15. or E. coli. Data are presented as means ± SD of technical triplicates using groups of 30 Larvae. Statistical analyses were performed using a mixed-effects model (REML) with time and condition as fixed factors, followed by Tukey’s post-hoc tests comparing uncontaminated food with either Ecc15-contaminated or E. coli-contaminated food conditions (ns = not significant; *p < 0.05; **p < 0.01; ***p < 0.001). (C) Schematic of the binary food-choice assay. (D-F) Distribution of larvae at 0, 20, and 120 min in food-choice assays when presented with paired patches: uncontaminated versus Ecc15-contaminated yeast (D); E. coli versus Ecc15-contaminated yeast (E); two Ecc15-contaminated patches (F). Data are presented as means ± SD. Individual points represent experimental replicates (each consisting of groups of ≥ 30 larvae). Statistical analyses were performed using Fisher’s exact test (ns = not significant; **p < 0.01; ****p < 0.0001).

https://doi.org/10.1371/journal.ppat.1014506.g002

To examine larval behavior in the presence of multiple food options, a two-choice assay was conducted. Larvae were placed equidistant between two food patches, and three pairwise comparisons were tested: uncontaminated yeast versus yeast inoculated with Ecc15; E. coli versus Ecc15 contaminated yeast; and two patches both containing Ecc15 (Fig 2C). Larval positions were recorded at 20 and 120 minutes to capture their initial orientation and later decision-making. After 20 minutes, larvae reached one of the food patches in all conditions without showing any preference, indicating that early orientation is not influenced by bacterial contamination (Fig 2D–2F). By 120 minutes, however, clear differences emerged: larvae moved away from Ecc15-contaminated food and preferentially occupied uncontaminated or E. coli-associated food (Fig 2D–2E). When both patches contained Ecc15, larvae abandoned the food entirely and accumulated on the surrounding agar, consistent with a broad avoidance response in the absence of an alternative food source (Fig 2F). These findings demonstrate that Drosophila larvae initially approach food without discrimination but retreat upon contact with Ecc15-contaminated substrates, subsequently redirecting their movement toward innocuous food sources. Combined with previous evidence that dispersal depends on bacterial metabolic activity rather than mere presence, these results suggest that contact-based detection of Ecc15 involves specific bacterial metabolites or virulence-associated factors.

Major Ecc15-derived factors or canonical immune and nociceptive signaling do not mediate larval dispersal behavior

Metabolically active Ecc15 produces multiple factors, including virulence proteins, metabolites, and structural components, that influence Drosophila physiology and potentially cue larval dispersal. The contribution of the secreted virulence factor Evf, which promotes gut colonization and triggers systemic immune activation [17,29], was first assessed. Larvae exposed to yeast contaminated with either wild-type or evf-deficient Ecc15 strains (Ecc15 Δevf) [17] dispersed equally, indicating that Evf is not required (Fig 3A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Ecc15-induced larval dispersal occurs independently of major bacterial factors, IMD immune signaling, and TRPA1 nociception.

(A) Larval response to yeast contaminated with wild-type Ecc15 or the evf-deficient mutant (Ecc15 Δevf). (B) Escape behavior induced by wild-type or uracil-deficient Ecc15 (Ecc15 Δ4). (C) Larval dispersal on food supplemented with increasing concentrations of purified PGN (10, 20, 40 μg/mL). (D–F) Dispersal of IMD pathway mutants: PGRP-LC^E12 (D), PGRP-LE¹¹² (E), and Dredd^D55 (F) on uncontaminated yeast and yeast mixed with Ecc15. (G) Escape response to yeast supplemented with purified LPS (125, 250, 500 µg/mL). (F) TrpA1 mutant larvae behavior on uncontaminated and Ecc15-contaminated food. Statistical analyses were performed using a mixed-effects model (REML) with time and condition as fixed factors, (ns = not significant; ****p < 0.0001). Comparisons between selected conditions are shown.

https://doi.org/10.1371/journal.ppat.1014506.g003

The role of uracil, a microbe-derived metabolite known to activate DUOX-dependent epithelial defenses in adults [16], was then examined. Comparable levels of dispersal were observed when larvae were exposed to wild-type or uracil-deficient Ecc15 strains (Ecc15 Δ4) [30], showing that uracil is also dispensable (Fig 3B).

The potential involvement of two conserved Gram-negative cell wall components was next evaluated. Although Ecc15-derived PGN is a strong activator of IMD signaling [13,15] and can induce aversive feeding responses in adult flies [31], supplementation of yeast with increasing concentrations of purified PGN did not induce dispersal (Fig 3C). In agreement, larvae lacking the PGN receptors PGRP-LC or PGRP-LE [32,33], or the downstream IMD effector Dredd [34], escaped from Ecc15-contaminated food similarly to wild-type larvae (Fig 3D–3F). The same approach was then used to assess the role of LPS, which activates TRPA1-dependent avoidance pathways in adult flies [22]. Purified LPS similarly failed to trigger larval dispersal at any of the concentrations tested (Fig 3G). Consistently, loss of TrpA1 function did not impair dispersal on Ecc15-contaminated yeast (Fig 3H).

Altogether, these findings demonstrate that larval escape behavior from Ecc15-contaminated food is not triggered by individual bacterial components or metabolites such as Evf, uracil, PGN, or LPS, nor does it depend on canonical IMD signaling or TRPA1-mediated sensing of noxious cues. Instead, the behavior is likely driven by the detection of Ecc15-derived signals through alternative sensory pathways.

Aversive gustatory receptor Gr33a mediates larval dispersal from Ecc15-contaminated food

In Drosophila, non-volatile aversive cues are detected by gustatory receptor neurons (GRNs) expressing specific gustatory receptors (GRs) [35–37]. In larvae, the gustatory receptors Gr33a and Gr66a are co-expressed in bilateral pairs of GRNs located primarily in the terminal organ (TO) and in pharyngeal sensilla, which together constitute the core aversion-sensitive GRN population [38]. To evaluate their functional requirement in Ecc15-induced dispersal, Gr66a-positive neurons were silenced by overexpressing the inward-rectifier potassium channel Kir2.1 (Gr66a-GAL4 > UAS-Kir2.1), thereby hyperpolarizing these cells and preventing synaptic transmission [39]. Larvae with inactive Gr66a-positive neurons showed markedly reduced dispersal from Ecc15-contaminated food compared to controls, while their behavior on uncontaminated yeast remained unchanged (Fig 4A). A similar reduction in dispersal was observed following silencing of Gr33a-positive neurons (Gr33a-GAL4 > UAS-Kir2.1) (S9A Fig). Together, these findings indicate that aversion-sensitive GRNs are essential for the escape behavior triggered by Ecc15 contamination.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. The gustatory receptor Gr33a mediates larval escape from Ecc15-contaminated food.

(A) Larval dispersal following silencing of Gr66a-expressing neurons (Gr66a-GAL4 > UAS-Kir2.1) on uncontaminated (N.C.) or Ecc15-contaminated food. (B-E) Escape response of wild-type larvae and loss-of-function mutants GR66a^ex83 (B), GR32a^KO (C), GR93a³ (D) and Gr33a¹ (E) exposed to uncontaminated and Ecc15-contaminated food. (F) Comparison of dispersal on control and Ecc15-containing food in wild-type, Gr33a¹; UAS-Gr33a, and Gr33a¹; Gr33a-GAL4 > UAS-Gr33a genotypes. Statistical analyses were performed using a mixed-effects model (REML) with time and condition as fixed factors, (ns = not significant; *p < 0.05; ****p < 0.0001). Comparisons between selected conditions are shown. For all Figures, baseline behavioral characterization of parental GAL4 and UAS lines on Ecc15-contaminated food are provided in S17 Fig.

https://doi.org/10.1371/journal.ppat.1014506.g004

In addition to Gr66a and Gr33a, Gr32a and Gr93a, two receptors involved in aversive contact behaviors, are expressed in partially overlapping subsets of aversion-sensitive GRNs [38,40–42]. To determine the specific contribution of each GR, loss-of-function mutant alleles for Gr33a (Gr33a¹), Gr66a (Gr66a^ex83), Gr32a (Gr32a^KO), and Gr93a (Gr93a³) were analyzed. Gr66a^ex83 and Gr32a^KO larvae dispersed at wild-type levels from contaminated food, indicating that Gr66a and Gr32a are not required for Ecc15-induced dispersal (Fig 4B–4C). Gr93a³ mutants showed heightened dispersal on contaminated food and mild nonspecific dispersal on uncontaminated yeast, suggesting that Gr93a modulates gustatory sensitivity rather than specifically mediating bacterial avoidance (Fig 4D). By contrast, Gr33a¹ larvae no longer escape from Ecc15-contaminated yeast, and remain on the substrate throughout the assay (Fig 4E). This phenotype was not attributable to defects in attraction or locomotion: when positioned at a distance from a food source, Gr33a¹ larvae readily oriented toward and reached both clean and contaminated substrates, but, unlike controls, failed to withdraw after contacting Ecc15-contaminated yeast (S9B Fig). Reintroducing Gr33a expression in Gr33 positive neurons (Gr33a¹; Gr33a-GAL4 > UAS-Gr33a) restored dispersal to wild-type levels (Fig 4F). Together, these results demonstrate that Gr33a plays a critical role in Ecc15-induced dispersal, highlighting its contribution to the contact-dependent detection of Ecc15-derived cues within the larval gustatory circuit.

orco–Or49a olfactory and Gr33a gustatory circuits cooperate to mediate larval escape from Ecc15-contaminated food

Olfaction provides a critical sensory input that complements gustation in detecting aversive environmental cues [43]. To determine whether volatile compounds contribute to the dispersal behavior triggered by bacterial contamination, larvae carrying null mutations in the olfactory co-receptor orco (orco¹ and orco²) were tested. While wild-type and heterozygous orco¹/ + larvae rapidly escaped from Ecc15 contaminated substrate, anosmic orco¹ mutants displayed a complete loss of this avoidance behavior (Fig 5A). Similar results were obtained with orco² mutants and orco¹/orco² transheterozygotes (S10A-S10B Fig). Consistently, blocking synaptic transmission in orco-expressing olfactory receptor neurons (ORNs) using the temperature-sensitive blocker shibire^ts [39] (orco-GAL4 > UAS-shi^ts) markedly reduced dispersal from contaminated food (S10C–S10D Fig). These findings show that orco-dependent ORNs are required for detecting volatile cues that trigger larval escape.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. The orco–Or49a olfactory pathway mediates larval escape response to Ecc15-contaminated food.

(A) Behavioral response of wild-type, orco¹ mutant and orco¹/ + heterozygotes larvae on uncontaminated (N.C) and Ecc15-supplemented yeast. (B) Response of GH146-GAL4 > UAS-Shi^ts larvae to Ecc15-contaminated food at restrictive temperature (31°C). (C) Escape induced by Ecc15 at 60 and 120 min after silencing individual larval olfactory neuron classes (OrX-GAL4 > UAS-Kir2.1). Data are presented as means ± SD. Individual squares represent biological replicates (each consisting of groups of ≥ 40 larvae. (D) Larval dispersal of Or49a-GAL4 > UAS-Kir2.1 larvae on uncontaminated or Ecc15-contaminated food. (E) Escape behavior of Or49a-GAL4 > UAS-orco^RNAi larvae on Ecc15-contaminated food. Statistical analyses were performed using a mixed-effects model (REML) with time and condition as fixed factors for panels A, B, D, and E (ns = not significant; **p < 0.01; ****p < 0.0001). Comparisons between selected conditions are shown. Panel C was analyzed using the Mann–Whitney test (ns = not significant; *p < 0.05; **p < 0.01).

https://doi.org/10.1371/journal.ppat.1014506.g005

To further delineate the olfactory circuitry, the role of projection neurons (PNs) transmitting sensory input from peripheral ORNs was examined. orco-positive ORNs project their axons to specific glomeruli within the larval antennal lobe, where they synapse onto GH146-GAL4-labeled PNs [44]. Silencing these neurons with shibires^ts strongly reduced dispersal on contaminated yeast, without affecting behavior on uncontaminated yeast, demonstrating that orco-dependent PNs are necessary for olfactory signal transmission driving dispersal (Fig 5B).

GH146-positive PNs send axons to two higher-order processing centers: the lateral horn (LH), which mediates innate olfactory responses, and the mushroom body (MB), involved in learning and multisensory integration [45,46]. Blocking synaptic transmission in Kenyon cells, the intrinsic MB neurons, using OK107-GAL4 > UAS-shibire^ts [47] had no effect on larval escape from contaminated food, suggesting that the MB is dispensable and that the rapid dispersal response relies on innate olfactory processing (S11 Fig).

Next, the role of individual odorant receptors (Ors) was investigated. The larval olfactory system contains 21 ORNs in the dorsal organ (DO), each expressing a specific Or together with Orco [48,49]. Selective inactivation of Orco-dependent ORN subsets via Kir2.1 overexpression (OrX-GAL4 > UAS-Kir2.1) showed that inhibition of Or49a-positive neurons specifically and reproducibly reduced dispersal, without affecting behavior on uncontaminated yeast (Fig 5C–5D). To rule out locomotor or orientation deficits, Or49a-GAL4 > UAS-Kir2.1 larvae were placed away from the food source; they oriented toward both uncontaminated and contaminated yeast but exhibited reduced escape frequency after contact with contaminated food, consistent with impaired escape initiation (S12 Fig).

Or49a expression was previously reported in one DO neuron co-expressing Orco and one TO neuron [48,49] (S13A Fig). Targeted knockdown of orco in Or49a-expressing neurons (Or49a-GAL4 > UAS-orco^RNAi) significantly decreased dispersal, indicating that Or49a-dependent olfactory signaling contributes to the detection of volatile bacterial cues required for escape (Fig 5E). Given that Or49a is also expressed in the TO, overlap with known aversive gustatory neuron populations was examined. Double-driver labeling (Or49a-GAL4; Gr33a-GAL4 > UAS-GFP) revealed one additional TO neuron compared to Gr33a-GAL4 > UAS-GFP controls alone, demonstrating that Or49a-GAL4 and Gr33a-GAL4 label distinct neuronal populations (S13A Fig). This observation was further supported by dual-reporter analysis using Or49a-GAL4 > UAS-GFP and Gr66a-LexA > LexAOP-mCherry, which showed no co-localization between reporter signals (S13B Fig). Together, these findings indicate that Or49a-GAL4-positive neurons are distinct from known aversive gustatory populations, while the contribution of the TO Or49a neuron to escape behavior remains unresolved.

To gain further insight into the functional relationship between the olfactory (Or49a–Orco) and gustatory (Gr33a) pathways, genetic interaction analysis was performed in Gr33a¹/ + ; orco¹/ + larvae. These double heterozygotes displayed escape behavior comparable to that of wild‑type and single heterozygous controls (S14 Fig). Accordingly, the results do not support a simple combinatorial threshold model in which the two pathways contribute additively.

Altogether, these findings reveal that larval escape from Ecc15-contaminated food depends on a cooperative multisensory network combining gustatory and olfactory inputs: Gr33a-dependent neurons detect contact-dependent bacterial cues, while Or49a–Orco signaling mediates perception of volatile compounds, jointly driving rapid dispersal.

Larval behavioral response to Ecc15-contaminated food modulates survival, development, and pathogen dissemination

The preceding results show that larval escape from Ecc15-contaminated substrates is triggered by the combined action of gustatory and olfactory cues upon contact with the food source. This behavior likely reduces the duration of exposure to contaminated substrates, raising the question of its consequences for larval survival and development. To address this, second-instar larvae were placed on uncontaminated yeast or yeast contaminated with Ecc15 or E. coli, and their growth and developmental progression were monitored.

After 24 h, larvae maintained on uncontaminated or E. coli–contaminated food exhibited comparable growth (Fig 6A–6B) and progressed normally through development, with similar pupation and adult emergence rates, consistent with their ability to thrive on substrates enriched with non-pathogenic bacteria (Figs 6C–6E and S15). In contrast, larvae exposed to Ecc15 left the contaminated substrate, showed no significant growth (Fig 6A–6B), and rarely pupated, with almost no adult emergence (Figs 6C–6E and S15). These observations indicate that exposure to Ecc15 triggers dispersal and persistent avoidance, while restrictive conditions that prevent relocation lead to developmental arrest and high mortality.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Exposure to Ecc15-contaminated food impairs larval development and promotes bacterial dissemination.

(A) Representative images of larvae at 0 h and 24 h after exposure to control food (N.C.) or food supplemented with E. coli or Ecc15. (B) Quantification of larval body lengths at 0 h and 24 h following exposure to control or contaminated food in arbitrary units (A.U.). Data are presented as means ± SD. Individual points represent individual larvae (n ≥ 20). (C) Representative images illustrating pupal formation four days after larval exposure to control or contaminated food; boxed regions are shown at higher magnification below. (D–E) Proportion of larvae that reached the pupal stage after four days (D) or eclosed as adults after seven days (E) following exposure to control or contaminated food. Data are presented as means ± SD. Individual points represent biological replicates (each consisting of 25 larvae). (F) Schematic of the assay used to evaluate developmental outcomes of larvae that relocate to uncontaminated food after an initial exposure to Ecc15-contaminated substrate. (G) Proportion of larvae exposed to increasing concentrations of Ecc15 (OD₆₀₀ = 5, 10, 20, and 40) that reached the pupal stage or eclosed following relocation to an uncontaminated food source (N.C.). Data are presented as means ± SD. Individual points represent biological replicates (each consisting of groups of ≥ 35 larvae). (H) Schematic of the bacterial dissemination assay. (I) Ecc15 retrieved (CFU) from initially uncontaminated food (N.C) two hours after larvae were exposed to contaminated substrate. Data are presented as means ± SD. Individual squares represent biological replicates. Statistical analyses were performed using the Mann–Whitney test for panel B and Fisher’s exact test for panels D and E (ns = not significant, *p < 0.05; ****p < 0.0001). Panel G was analyzed using Chi-square tests that demonstrate that Ecc15 concentration impacts on the pupation of larvae and on emergence of adult flies (****p < 0.0001). Panel I was analyzed using unpaired t test (*p < 0.05; **p < 0.01; ***p < 0.001).

https://doi.org/10.1371/journal.ppat.1014506.g006

To determine whether access to alternative resources mitigates the developmental consequences of Ecc15 exposure, larvae were placed on uncontaminated yeast or yeast contaminated at increasing Ecc15 bacterial loads, with each yeast patch positioned at a distance from a second, uncontaminated food source (Fig 6F). After 24 hours, larvae initially placed on uncontaminated yeast largely remained on the original patch, indicating limited dispersal in the absence of pathogen cues (S16A–S16B Fig). In comparison, larvae exposed to Ecc15 consistently relocated to the uncontaminated patch at all bacterial concentrations (S16B Fig). Despite this relocation, developmental analysis revealed a dose-dependent impairment, with pupation and eclosion rates significantly lower than those of larvae initially placed on uncontaminated food (Fig 6G). Thus, access to an uncontaminated refuge partially alleviates, but does not fully prevent, the detrimental effects of initial Ecc15 exposure.

Given the robust dispersal behavior triggered by Ecc15, including relocation to uncontaminated substrates, the possibility that larvae contribute to bacterial dissemination was next examined. Larvae were placed on contaminated yeast at increasing bacterial loads in the same two-patch assay configuration (Fig 6H). Within two hours, larvae were consistently detected on the clean substrate, with the number of relocated individuals increasing with the initial bacterial load (S16C Fig). Viable Ecc15 cells were recovered from these previously uncontaminated patches, with abundance correlating with both the number of migrating larvae and the initial bacterial load (Fig 6I).

Together, these findings show that the escape response elicited by Ecc15-contaminated food, characterized by initial contact followed by rapid withdrawal and relocation, reduces direct exposure to contaminated substrates yet fails to fully prevent detrimental developmental consequences. At the same time, this behavior promotes the transfer of viable bacteria to new food sources, indicating that larval Drosophila can contribute to the environmental dissemination of Ecc15.

Larval escape is triggered under specific ecological and bacterial metabolic conditions

Behavioral responses to pathogens have been far less explored in Drosophila larvae than in adults. Previous work showed that larvae can abandon food contaminated with P. entomophila, but this response emerges over several hours and is strongly influenced by the physiological state of the animal. In particular, starvation reduces avoidance behavior, indicating that internal nutritional cues modulate pathogen-induced responses. In that study, no avoidance response to Ecc15 was detected under the conditions tested [26].

The escape behavior described here differs in several key respects. It is rapidly initiated, within minutes after contact with Ecc15-contaminated food, and comparable dispersal occurs in both fed and starved larvae. Together, these differences indicate that the response observed in this study represents a distinct form of behavioral avoidance.

In the present work, larval escape was found to depend on the physiological state of Ecc15. It occurred only when stationary-phase cells were provided with access to nutrients permitting metabolic reactivation, while bacterial numbers remain stable, ruling out proliferation as the trigger. Thus, escape appears to rely on bacterial metabolic activity rather than abundance, highlighting the importance of bacterial physiological state in host detectability and behavioral impact. The activity was retained for several days on the substrate, suggesting that bacterial metabolic activation leads to the production of behaviorally active, as yet unidentified cues that remain stable in the food environment and can drive behavior independently of ongoing growth.

Importantly, under the conditions tested here, escape is observed with Ecc15 but not with other bacterial species tested, including E. coli, P. entomophila, and L. brevis. This suggests that larval escape is not triggered by bacterial presence in a generic manner, but rather by specific bacterial signals associated with Ecc15 under conditions permissive for metabolic activity. Whether other bacteria relevant to Drosophila ecology, including additional Erwinia species [6], can also trigger rapid escape remains an open question. More broadly, the widespread coupling between bacterial metabolic activity and production of extracellular factors during growth suggests that similar mechanisms may underlie cue generation beyond Ecc15.

Larval escape relies on complementary gustatory and olfactory cues

Our findings reveal that larval escape from Ecc15 relies on the cooperation of two distinct chemosensory modalities: gustation and olfaction. To our knowledge, this is the first demonstration of a rapid, multisensory escape behavior triggered by a natural pathogen in Drosophila larvae. The gustatory receptor Gr33a, a broad aversion receptor associated with bitter and toxic compound detection [36,40–42], is essential for the response. Although Gr33a often functions together with Gr66a in adult taste circuits, our data indicate that Gr33a acts here independently of Gr66a, pointing to a Gr33a-dependent mechanism of danger detection. The requirement for contact-dependent gustatory input suggests that bacterial metabolic activity produces at least one non-volatile cue capable of triggering Gr33a-dependent larval escape.

In parallel, an olfactory component is required. The odorant receptor Or49a, functioning together with the co-receptor Orco and previously implicated in the detection of the parasitoid wasp pheromone iridomyrmecin [52], is also critical for dispersal. This finding indicates that the Or49a‑Orco olfactory receptor complex is required for detecting a volatile cue produced by Ecc15, distinct from iridomyrmecin, and highlights its functional versatility. The involvement of Or49a aligns with growing evidence that the larval olfactory system, though anatomically simple, contains dedicated circuits for detecting ecologically salient threats [52,53].

Strikingly, disruption of either the gustatory or the olfactory pathway abolishes escape, indicating that both sensory modalities are required for behavioral output. Volatile and contact-dependent cues therefore likely provide complementary information about bacterial contamination. Such multisensory processing may improve substrate evaluation by relying on both volatile and contact-dependent information associated with contaminated substrates. Whether these cues correspond to different chemical forms of the same compound or to distinct metabolites remains unknown.

Two models can account for this dual requirement. In a convergence model, gustatory and olfactory pathways interact within shared downstream circuits to jointly drive the response. In a threshold model, each pathway contributes independently to a cumulative signal, such that loss of either input reduces that signal below the threshold needed to trigger escape. Consistent with the involvement of distinct sensory populations, no overlap was detected between Gr33a- and Or49a-expressing neurons. Moreover, in Gr33a¹/ + ; orco¹/ + double heterozygous larvae, no enhanced behavioral deficit was found relative to single heterozygotes or wild-type controls. Together, these results are more consistent with circuit-level interdependence or partial functional coupling than with a strict additive threshold mechanism. However, the genetic perturbations used here likely reduce rather than abolish sensory activity, precluding a definitive mechanistic distinction between circuit-level convergence and broader functional cooperation.

Although the circuit-level relationship between gustatory and olfactory inputs remains unresolved, neuronal manipulations provide insight into the broader circuit organization underlying escape behavior. Silencing GH146-positive projection neurons, which innervate the larval antennal lobe and relay olfactory information to higher brain centers, indicated that escape depends on olfactory projection pathways [45,53]. These neurons project to both the LH and MB [54–56], yet silencing Kenyon cells had no effect, suggesting that associative MB circuits are not required. Taste information from Gr33a-expressing neurons is relayed to the subesophageal zone (SEZ), the primary gustatory processing center in larvae [38,40,42]. While SEZ output pathways remain incompletely mapped, they clearly connect gustatory processing to higher-order protocerebral regions.

The combined requirement for GH146-positive projection neurons and Gr33a-expressing neurons, together with the rapid, stereotyped, and decision-like nature of the behavior, supports the view that escape is likely routed through non-associative circuits consistent with innate valence pathways. However, the precise circuit locus underlying the joint contribution of these pathways remains to be determined. At a broader anatomical level, this interpretation aligns with the established role of the LH in mediating hardwired responses to ecologically salient stimuli, whereas the MB is generally required for learned aversions [46,57].

These findings further establish this behavioral paradigm as a tractable model for dissecting how gustatory and olfactory cues cooperate to guide ecologically relevant avoidance responses. The genetic accessibility of the larva, together with the ability to manipulate both sensory modalities independently, makes this system well suited for mapping the cellular and circuit mechanisms underlying the translation of bacterial-derived cues into behavior.

Larval escape reduces local exposure while promoting pathogen dissemination

Larval escape from Ecc15-contaminated food promotes relocation toward uncontaminated resources, thereby reducing the duration of contact with contaminated substrates. Developmental analyses nevertheless indicate that this response provides only partial mitigation of the detrimental effects associated with Ecc15 exposure. Although access to an uncontaminated refuge improved pupation and eclosion relative to larvae maintained under restrictive conditions without alternative food sources, developmental success remains substantially impaired compared to uncontaminated controls. These findings indicate that escape alters exposure dynamics without fully preventing the deleterious consequences associated with Ecc15 exposure. More generally, they suggest that the outcome of avoidance behavior depends critically on environmental spatial organization and resource heterogeneity.

Importantly, these experiments were performed under simplified laboratory conditions that likely differ from the ecological complexity of fermenting substrates. In natural environments, bacterial density, microbial community composition, substrate turnover, and food distribution would all be expected to influence both exposure dynamics and the effectiveness of larval relocation. Under more heterogeneous ecological conditions, rapid withdrawal from contaminated substrates could therefore contribute more substantially to reducing exposure to harmful microbes.

Escape behavior also promotes bacterial dissemination. Dispersing larvae transferred viable Ecc15 cells to previously uncontaminated food patches, indicating that larval movement contributes directly to bacterial spread under the conditions examined here. Thus, the same behavior that reduces local exposure to contaminated food also enhances the spatial redistribution of viable bacteria across food resources. Comparable forms of insect-mediated bacterial dissemination are well documented in the ecology of plant-associated microbes, where host mobility contributes substantially to bacterial transmission between substrates [58]. In this context, the ability of Drosophila adults to disseminate Erwinia species in natural environments raises the possibility that both larval and adult stages participate in the environmental circulation of these bacteria [11,12].

Together, these observations suggest that larval escape generates dual ecological consequences for both host and bacterium. While relocation reduces contact with contaminated food, larval movement concurrently redistributes viable Ecc15 cells across food patches. Rather than reflecting an outcome that is exclusively beneficial to either partner, this behavioral response may therefore represent an ecological interaction in which host avoidance indirectly contributes to bacterial dissemination. Escape, triggered under conditions that support bacterial metabolic activity, likely reflects the sufficiency of growth-associated bacterial cues to induce larval relocation.

Conclusions and perspectives

This study identifies a rapid escape behavior in Drosophila larvae triggered by the natural pathogen Ecc15. This response is selective, requiring metabolically active bacteria and engaging complementary gustatory (Gr33a-dependent) and olfactory (Orco–Or49a-dependent) pathways. Together, these sensory inputs enable larvae to detect contaminated food sources and rapidly relocate to alternative resources when available.

Beyond its sensory basis, this behavior also has important ecological consequences. While dispersal can reduce prolonged exposure to contaminated substrates under spatially heterogeneous conditions, it also promotes the transfer of viable bacteria to new food patches. These findings therefore highlight how a single host avoidance behavior can reshape both host exposure and microbial spatial distribution.

More broadly, this work establishes larval escape from Ecc15 as a tractable model for understanding how microbial physiological states shape multisensory behavioral responses and ecological interactions. This system provides a framework for investigating how neural processing and environmental context jointly influence host–microbe interactions.

Fly stocks and husbandry

The following Drosophila melanogaster strains were used in this study. The w¹¹¹⁸ strain (no. 5905, Bloomington Drosophila Stock Center, BDSC) served as the reference wild-type strain in all main experiments unless otherwise indicated. Additional lines obtained from the BDSC included: Canton-S (no. 64349), Oregon-R (no. 25211), PGRP-LE¹¹² (no. 33055), orco¹ (no. 23129), orco² (no. 23130), GH146-GAL4 (no. 91812), OK107-GAL4 (no. 854), UAS-Shibire^ts (no. 44222), UAS-Kir2.1 (no. 6595), Gr66a-GAL4 (no. 28801), Gr33a-GAL4 (no. 57624), Orco-GAL4 (no. 23292), UAS-mCD4-Tomato (no. 35841), Gr66a^ex83 (no. 25027), Gr33a¹ (no. 31427), Gr93a³ (no. 27592), Gr33a¹; Gr33a-GAL4 (no. 31425), Gr33a¹; UAS-Gr33a-GAL4 (no. 31424), Or1a-GAL4 (no. 9949), Or7a-GAL4 (no. 23907), Or13a-GAL4 (no. 9945), Or22c-GAL4 (no. 9953), Or24a-GAL4 (no. 9957), Or30a-GAL4 (no. 9960), Or33b-GAL4 (no. 9964), Or35a-GAL4 (no. 9968), Or42a-GAL4 (no. 9972), Or42b-GAL4 (no. 9976), Or47a-GAL4 (no. 9982), Or49a-GAL4 (no. 9985), Or59a-GAL4 (no. 9989), Or63a-GAL4 (no. 9991), Or67b-GAL4 (no. 9996), Or74a-GAL4 (no. 23124), Or82a-GAL4 (no. 23125), Or83a-GAL4 (no. 23127), Or85c-GAL4 (no. 23914), Or94b-GAL4 (no. 23145), TRPA1 (no. 26504) and UAS-6xGFP (no. 52262). The yellow white reference strain was kindly provided by Dr. B. Charroux (Aix-Marseille University, France). The Dredd^D55 and Gr32a^KO mutant strains were generous gifts from Dr. B. Lemaire (EPFL, Switzerland) and Dr. J.R. Carlson (Yale University, USA), respectively. The UAS-Orco^RNAi line (KK100825) was obtained from the Vienna Drosophila Resource Center (VDRC). The PGRP-LC^E12 mutant was described previously [59]. The Gr66a-LexA; LexAop-mCherry line was kindly provided by Dr. Yali V. Zhang (University of Pennsylvania, USA).

Flies were maintained at 25 °C on standard yeast/cornmeal medium under a 12 h light/12 h dark cycle. For 1 L of food, 8.2 g agar (VWR), 80 g cornmeal (Westhove Farigel maize H1), and 80 g yeast extract (VWR) were boiled for 10 min, cooled, and supplemented with 5.2 g methylparaben sodium salt (MERCK) and 4 mL 99% propionic acid (CARLO ERBA).

Preparation and inactivation of bacterial and yeast cultures

The bacterial strains used included rifampicin-resistant Erwinia carotovora carotovora 15 (Ecc15) [6], Ecc15 Δevf [29], Ecc15 Δ4 [30], as well as Escherichia coli K-12 M4100, Pseudomonas entomophila, and Levilactobacillus brevis, kindly provided by Dr. I. Gomperts Boneca (Institut Pasteur, France), Dr. A. Gallet (CNRS, INRAE, France), and Dr. F. Leulier (CNRS, France), respectively.

Single colonies were used to inoculate 250 mL of culture medium. Ecc15 and P. entomophila were grown in LB broth (Lennox formulation, Invitrogen, cat. L22897) at 30 °C under agitation (250 rpm), whereas E. coli was cultured in LB at 37 °C under the same conditions. L. brevis was grown in MRS broth (Sigma-Aldrich, cat. 69966) at 37 °C under static, anaerobic-like conditions. Bacterial growth was monitored by OD₆₀₀. Cultures were centrifuged at 3800 rpm for 15 min, and pellets were resuspended in PBS to OD₆₀₀ = 200 before mixing with baker’s yeast to obtain the desired final OD₆₀₀.

To establish the correspondence between OD₆₀₀ values and bacterial load in yeast–bacteria mixtures, yeast–Ecc15 samples were prepared at defined final OD₆₀₀ values (5, 10, 20, and 40). Aliquots (100 µL) of these mixtures were homogenized in 1 mL PBS, and colony-forming units (CFUs) were determined by serial dilution and plating on LB agar supplemented with rifampicin, followed by overnight incubation at 30 °C.

Heat-killed Ecc15 was obtained by incubating pellets at 95 °C for 10 min followed by rapid cooling at −20 °C. Baker’s yeast (S. cerevisiae) was inactivated at 80 °C for 10–15 min and cooled to room temperature. Heat-killed bacteria and inactivated yeast were processed identically to live cells in subsequent assays.

Behavioral assays

A 100 µL bacterial suspension was mixed with 0.5 g commercial baker’s yeast (~0.5 mL) or, when indicated, with 0.5 g mashed ripe banana (~0.5 mL). Final OD₆₀₀ was calculated as: Final OD₆₀₀ = 100 μL × Initial OD₆₀₀/ 500 μL. A 100 µL drop of food–bacteria mixture was placed on LB agar plates and incubated for 30 min, unless otherwise specified. Second-instar larvae (48 h at 25°C after egg laying) were washed in water and tested in groups of n ≥ 40 per replicate, unless otherwise stated. Each experiment included at least three technical replicates and was independently repeated three times or more.

In dispersal assays, larvae were placed directly on food; in avoidance and binary choice assays, larvae were positioned 4 cm from the food source(s). At each time point, larvae outside the food were counted under a ZEISS Stemi 508 stereomicroscope. At the end of each assay, the food was suspended in water and larvae counted after liquid removal. Escape percentages were calculated relative to the initial number of larvae.

Behavioral assays were performed at room temperature, unless otherwise specified, in darkness; illumination was applied only briefly during counting to minimize phototactic bias.

Imaging of larval behavior

For time-point imaging, groups of >100 larvae were placed directly on the food source. Experiments were conducted in darkness, and white light was applied only briefly for image capture. Images were acquired using a Logitech HD Pro Webcam C920.

For movie generation, time-lapse imaging was performed using the same setup, controlled by Yawcam v0.8.0, capturing one frame every 2 s over 3 h 30 min. During acquisition, white-light intensity and direction were minimized to reduce phototactic responses. Image sequences were compiled into movies using Fiji (ImageJ v1.54p) and rendered at 24 fps. All images, including those used for time-point photography and movie generation, were processed with Photoshop CS6.

Larval growth, development, and survival

Second-instar larvae were placed on control yeast or yeast mixed with Ecc15 or E. coli. After 24 h, larvae were washed in phosphate-buffered saline (PBS), fixed in 70% ethanol, mounted, and imaged using a Zeiss Discovery Lumar V12 microscope. The size of ≥20 larvae per condition was quantified in arbitrary units using Fiji (ImageJ v1.54g) and compared to unexposed second instar larvae.

For development and survival assays, groups of 25 second instar larvae per condition were monitored for pupation (day 4) and adult eclosion (day 7) following exposure to the food source. Counting was performed using a ZEISS Stemi 508 stereomicroscope, and images were acquired with a Logitech HD Pro Webcam C920.

Assessment of bacterial dissemination by Drosophila larvae

To evaluate bacterial transfer from contaminated to uncontaminated food sources, groups of 25 larvae were placed on yeast mixed with Ecc15 (OD₆₀₀ = 5–40), located 4 cm from uncontaminated yeast. All conditions were performed in four technical replicates. After 2 h, the proportion of larvae that reached the uncontaminated yeast was quantified. The uncontaminated food was collected and homogenized in 1 mL PBS. Colony-forming units (CFUs) were determined by serial dilution of the homogenate and plating on LB agar supplemented with rifampicin, followed by overnight incubation at 30 °C.

Bacterial components and culture media

Ultrapure peptidoglycan from E. coli (PGN-ECndi; InvivoGen, USA, cat. Tlrl-kipgn) and lipopolysaccharides (LPS) from E. coli O111: B4 (Sigma-Aldrich, cat. L2630) were used as bacterial components in experimental assays.

For the preparation of solid media, Luria–Bertani (LB) broth (Lennox formulation, Invitrogen, cat. L22897), agar (BD Bacto, Becton Dickinson, cat. 214010), high-purity agarose (NEEO Ultra, Euromedex, cat. 2267.4), and commercial apple juice were used. Apple agar consisted of apple juice supplemented with agar and 0.005% Tween-20 to improve surface wetting and promote uniform contact between the yeast–bacteria mixture and the substrate. All reagents were handled according to the manufacturers’ instructions.

Confocal imaging

Second instar larvae of the specified genotypes were heat-killed (60 °C, 10 s), mounted in Vectashield (Vector Labs, Cat. No. H-1200), and imaged using a Zeiss LSM 780 confocal microscope with a 20 × air objective. Images were processed using Adobe Photoshop CS6.

Data representation and statistical analyses

All graphical representations and statistical analyses were performed using GraphPad Prism 8. Larval escape dynamics were analyzed using a mixed-effects model (REML) with Time and Condition as fixed factors, including their interaction. The T0 time point was excluded because all conditions are identical at this stage. Post-hoc multiple comparisons were carried out using Tukey’s test. For all experiments, full raw datasets and complete statistical outputs are provided in S1 and S2 Data.