Mutant bacteria that do not release uracil are no longer able to prime larvae

We previously reported that wild-type flies from larvae reared in the presence of Levilactobacillus brevis (L. brevis) are attracted to a 1 mM sucrose solution [37]. However, this reflex is considerably attenuated when the sucrose solution contains bacterial peptidoglycan at a concentration of 200 µg/ml, suggesting that PGN has a bitter valence for adult flies (Figs 1A, S1A, and S1B and [37]). We have previously demonstrated that PGN derived from bacteria is perceived by ppk23+ neurons located in the proboscis of adult flies and triggers aversion. We also showed that adults derived from larvae raised under axenic conditions lose this capacity while continuing to react to caffeine, an aversive molecule, demonstrating that their gustatory system remains functional. Furthermore, the induced cohabitation of axenic larvae with L. brevis, but not with Lactiplantibacillus plantarum, is sufficient to restore the aversive response to PGN in adults [37]. This suggests that certain specific bacterial species can prime larvae to give rise to PGN-sensitive adults (S1A Fig). To identify the nature of the bacterial priming signal(s), we used a mono-association protocol in which axenic larvae are reared in the presence of the desired bacterial strains or compounds. Emerged adults are tested for their ability to respond to PGN using the proboscis extension reflex (PER) as a read out for the gustatory response (S1B Fig). While larval cultivation with L. brevis primes the animals, neither heat-killed L. brevis nor the supernatant of a L. brevis. solution were able to reproduce the priming effects of live L. brevis (Fig 1A). In order to search for a putative priming inducer, we looked for molecular characteristics present in bacterial species capable of priming but absent in others. While, as mentioned above, the opportunistic pathobionts L. brevis is capable of priming axenic larvae, L. plantarum, considered a symbiont for Drosophila, is not [37]. As previous reports demonstrate that the nucleoside catabolism pathway controlling bacterial uracil is an essential trigger for the transition from commensal to pathogen [39,40], we tested whether it could also be involved in the priming of axenic larvae. To this end, and for the remainder of the study, we used Erwinia carotorova (also known as Pectobacterium carotovorum, strain Ecc¹⁵) for which valuable mutants with impact on uracil metabolism had previously been generated [41,42]. Unlike wild-type bacterial strains that were capable of larval priming, two mutants (Ecc¹⁵Δ4 and Ecc¹⁵pyrE::Tn5) whose ability to release uracil is impaired, lost this property (Fig 1B). This result, which suggests that uracil is involved in priming, was confirmed by experiments showing that simply adding purified uracil at 1mM to the medium is sufficient to trigger priming in larvae raised under axenic conditions (Fig 1B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Larval exposure to uracil is required for adult PGN-mediated inhibition of PER.

(A) Live L. brevis are necessary and sufficient to prime the larvae. Sterilized eggs were exposed to live or dead bacteria as well as supernatant. The resulting adults were raised on antibiotics-containing media (ATB). PER index of w⁻ flies to control solutions of sucrose (Sucrose 1 mM) and sucrose + PGN from E. coli K12 (Sucrose 1mM + PGN200) were obtained. (B) Uracil is necessary and sufficient to prime the larvae. Sterilized eggs were exposed to the different treatments and the resulting adults raised on antibiotics-containing media (ATB). Specifically, eggs were either monoassociated with bacterial strains, or exposed to LB media or to LB media containing uracil or cytosine. Addition of uracil or Ecc¹⁵ 24h after sterilized egg deposition was tested. The resulting adults were raised on antibiotics-containing media (ATB). PER index of w⁻ flies to control solutions of sucrose and sucrose + PGN from E. coli K12 at 200 µg/mL (PGN200) were obtained. (C) The Duox enzyme is necessary in enterocytes for the priming. PER index to solutions of sucrose and sucrose + PGN (PGN200) of GAL4 (mex-GAL4/+) and UAS (+/UAS-Duox_IR) control flies as well as animals with the Duox transcript impaired via RNAi in enterocytes (mex-GAL4/UAS-Duox_IR) were calculated. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (D) Graphical representation of the life periods during which flies are shifted from 18 °C to 29 °C for a stage-dependent RNAi assay. The ubiquitously expressed Tub-G80^ts, that inhibits the activity of GAL4, is temperature-sensitive: the GAL4 inhibitor is active at 18 °C (green) and inactivated at 29 °C (black), its inactivation allows the expression of any UAS. In the case of mex-GAL4; TubG80^ts/UAS-Duox_IR, at 18 °C the Duox transcripts will not be impaired (Duox) while it will be at 29 °C. (E) Duox activity in enterocytes is necessary during the larval life for priming. PER index to solutions of sucrose and sucrose + PGN (PGN200) of mex-GAL4; TubG80^ts/UAS-Duox_IR flies with the Duox transcript impaired all-life long, only during the larval stages or only during the adult stage were calculated. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (F) H₂O₂ is detectable within the larvae 24 h following exposure to uracil 1 mM. Sterilized eggs were exposed to LB media or to LB media containing uracil. Twenty-four h later, the concentration of H₂O₂ within the animals was quantified. To confirm that H₂O₂ is the quantified product, catalase (0.01 U/mL and 0.001 U/mL) was applied to the samples from uracil-exposed larvae or overexpressed in enterocytes (mex-GAL4/UAS-Catalase). (G) Reducing the amount of ROS impairs the priming. PER index to solutions of sucrose and sucrose + PGN (PGN200) of UAS (+/UAS-Catalase) control flies as well as animals with the Catalase over-expressed in enterocytes (mex-GAL4/UAS-Catalase) were obtained. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (H) H₂O₂ supplementation to larvae is not sufficient to prime while vitamin C (VitC) addition during the larval life prevents it. PER index to solutions of sucrose + PGN (PGN200) of w⁻ flies was calculated. Larvae were raised on conventional media with or without vitamin C (0.2 mg/mL) or on monoassociation media with or without H₂O₂ (1% on eggs or 1% on eggs then 1% on 24h-larvae (2X) or 2% on eggs). All the resulting adults were raised on antibiotics-containing media. PER index of w⁻ flies to solutions of sucrose + PGN from E. coli K12 at 200 µg/mL (PGN200) was obtained. The PER index is calculated as the percentage of flies tested that responded with a PER to the stimulation ± 95% confidence interval (CI). A PER value of 1 means that 100% of the tested flies extended their proboscis following contact with the mixture, a value of 0.2 means that 20% of the animals extended their proboscis. The number of tested flies (n) is indicated on top of each bar. For each condition, at least 3 groups with a minimum of 10 flies per group were used. Each independent group is represented as an open circle. ns indicates p > 0.05, * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001 Fisher’s exact test. Further details can be found in the detailed lines, conditions and, statistics for the figure section. The data underlying this Figure can be found in S1 Data. Created in BioRender. Kurz, L. (2026). https://core.local.biorender.dev/api/short-link/9ojribp.

https://doi.org/10.1371/journal.pbio.3004042.g001

Loss of priming in uracil-release mutants and sufficiency of exogenous uracil implicate uracil as a key determinant under our conditions. To delineate whether uracil-related molecules could prime larvae, we used the same protocol as the one involving uracil to expose sterilized eggs to cytosine or thymine. None of these molecules were able to phenocopy the priming effect of uracil (Figs 1B and S2).

In the monoassociation protocol, bacteria or chemical compounds are deposited onto sterilized eggs, and the resulting adults are subsequently tested for PER. In our previous study investigating the PER response to a sucrose + PGN mixture and the requirement for larval exposure to bacteria, we identified a critical developmental window between oviposition and 48 h post-oviposition [37]. We therefore asked whether the effect of uracil depended on the same critical period. To address this question, we exposed 24h-old larvae derived from sterilized eggs to uracil rather than exposing the eggs themselves, and subsequently assayed PER in the resulting adults. This treatment failed to prime the larvae (Fig 1B). We obtained the same results with 24 h-old larvae derived from sterilized eggs and exposed to Ecc¹⁵. These results confirm the existence of a critical developmental window and further narrow it to a period occurring before 24 h after larval hatching.

Our previous report, together with the present study, establishes a link between larval experience and the ability of adult flies to discriminate between a food mixture containing sucrose alone and one containing sucrose supplemented with PGN. We demonstrated that this discrimination relies on a subset of ppk23+ neurons located in the adult proboscis. Exposure to PGN induced a measurable increase in intracellular calcium levels in these neurons, indicating their activation by bacterial cues. In contrast, ppk23+ neurons from adults derived from larvae not exposed to bacteria failed to respond to PGN stimulation [37].

To investigate whether larval priming with uracil could modulate the responsiveness of adult ppk23+ neurons, we performed GCaMP calcium imaging experiments. Specifically, we compared the neuronal responses to PGN in adults originating from larvae exposed to uracil with those in adults derived from non-exposed larvae. This approach allowed us to examine, at the cellular level, how early-life exposure to microbial-associated signals influences the development of sensory discrimination mechanisms in adulthood. While ppk23+ proboscis’s neurons of adults from larvae exposed or not to uracil responded to salt (positive controls), only the adults originating from uracil-exposed larvae responded to PGN (S3 Fig), supporting the idea that larval experience influences neuronal responsiveness later in adulthood.

The production of reactive oxygen species by enterocytes is necessary for larval priming

Previous studies demonstrate that uracil acts as a microbe-derived factor that modulates intestinal immunity in Drosophila [39]. Uracil-dependent activation of Duox leads to the production in the intestinal tract of reactive oxygen species (ROS), which combat invading pathobionts but also induce damage to host tissues [42–44]. It has been shown that the inactivation of a single bacterial gene involved in uracil production is sufficient to induce a phenotypic change from a colitogenic bacterium to a commensal bacterium [45]. These results, which demonstrate the importance of Duox-dependent ROS production in mediating the effects of uracil, prompted us to test whether ROS were also involved in larval priming. To do this, we induced RNA interference-mediated reduction of Duox transcripts specifically in enterocytes using the mex-GAL4 driver (S4 Fig). While control adult strains showed the described aversion to PGN in the PER test, this was no longer the case for mex-GAL4/UAS-Duox_IR adult flies (Fig 1C). Taking advantage of the GAL4/GAL80^ts binary system, we reduced ROS levels in either the larval or adult stage (Fig 1D and 1E). While reducing ROS levels in larval guts suppressed priming, this was not the case when the reduction was induced in the adult stage (Fig 1E). Based on the genetic data and the identified critical period, we tested whether ROS, and H₂O₂ in particular, could be detected in 24-h-old larvae following exposure of sterilized eggs to uracil. To confirm that the measured signal corresponded to H₂O₂, we either treated uracil-exposed samples with catalase or genetically overexpressed catalase (an enzyme that degrades hydrogen peroxide) in enterocytes (mex-GAL4/UAS-Catalase). Our results demonstrate that H₂O₂ levels are increased following uracil exposure (Fig 1F). Interestingly, priming in larvae was suppressed by GAL4/UAS-mediated overexpression of catalase, in the gut (Fig 1G). Similarly, the addition of the antioxidant vitamin C (VitC) to the medium at a non-bactericidal concentration (S5A Fig) suppressed larval priming (Fig 1H). However, using either the monoassociation protocol (S1B Fig) or axenic media (ATB), neither a single application of 1% H₂O₂ (on eggs only) nor two applications (on eggs and 24-h-old larvae; 2×), nor 2% H₂O₂, were sufficient to induce priming (Figs 1H, S5B, and S5C). Overall, these results support a model in which uracil triggers in larvae a Duox-dependent ROS production in larval enterocytes required for priming.

The B, C, and E isoforms of TrpA1, but not the A and D, are involved in larval priming

Our previous data demonstrate that expression of the TrpA1 ion channel in larval Gr66a+ neurons is necessary for larval priming by bacteria [37]. Drosophila TrpA1, which shares conserved sensory functions with its mammalian ortholog and is activated by heat, ROS, UV light, and irritant chemicals (Fig 2A), is alternatively spliced to produce five isoforms (Fig 2B) [46–51]. By generating and phenotypically analyzing knock-in (KI) flies expressing a single TrpA1 isoform and knock-out flies (KO) lacking selected TrpA1 isoforms, Gu and colleagues conclude that a given sensory stimulus preferentially activates a specific TrpA1 isoform in vivo [46]. Using this genetic toolkit, we sought to identify the number and nature of TrpA1 isoform (s) involved in larval priming to bacteria. Although these TrpA1 mutants all responded similarly to sucrose, their PER to a sucrose + PGN solution differs. Among the KI lines, B-KI, C-KI, and E-KI flies reacted like the wild type. In contrast, flies expressing only isoform A or D showed a response similar to that of TrpA1-KO mutant flies, demonstrating that they lack the ability to mediate larval priming. These results, which suggest that isoforms B, C, and E, but not A and D, are capable of priming, were confirmed by analysis of the different KO mutant flies (Fig 2C). Adults from lines in which at least one of the three competent isoforms was present (E-KO, BC-KO, and AD-KO) all responded to PGN like the controls. This result was consistent with the fact that BCE-KO flies, which lack all three competent isoforms (B, C and E), were the only flies whose ability to respond to PGN was impaired (Fig 2C). These results demonstrate that expression of any one of the three isoforms, B, C, or E is sufficient to ensure larval priming. The simplest explanation would be that B, C, and E are functionally redundant and share a functional property that is absent in isoforms A and D. Careful examination of the TrpA1 peptides suggests that this is not the case. Indeed, no protein domain that is specifically present in proteins TrpA1 B, C, and E but absent in A and D could be identified. An alternative hypothesis would be that the type of cells(s) in which the TrpA1 isoforms are expressed is the important criterion to consider and that isoforms B, C and E are co-expressed in one or more cells that would be necessary to ensure TrpA1-dependent priming. It should be noted that these two hypotheses (the nature of the peptides or the domain of expression) are not mutually exclusive (see below).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. TrpA1 and specifically the B, C and E isoforms are involved in the priming.

(A) Graphical representation of the TrpA1 channel responding to various stimuli through the entry of ions into the cell. Heat, specific chemicals, ROS and mechanical stimuli trigger the TrpA1 proteins participating to the generation of an action potential. (B) Graphical representation of the TrpA1 gene (top) with colored boxes representing exons and colors related to functional domains of the translated protein. Five spliced variant isoforms have been reported so far, they all share exons 4–11 and exons 13–18. (C) TrpA1-B, -C and -E isoforms are involved in priming. PER index to solutions of sucrose and sucrose + PGN (PGN200) of control flies as well as KO and KI animals for the TrpA1 isoforms was calculated. Depending on the genotype of the tested animal, the TrpA1 isoforms that are present are listed under the x-axis. (D–D‴) The TrpA1-GAL4 driver (pan-isoforms) is widely expressed. Confocal images of larvae expressing gfp under the control of the pan-isoforms TrpA1-GAL4 driver with a whole L3 larvae (D), the anterior part of a larvae with anteriormost on the left (D′), the larval brain (D″) and the gut (D‴). In (D), the larger rectangle is a magnification of the most anterior part of the animal housing two gfp+ symmetrical neurons whose dendrites extend toward the external environment (t). (b) is for brain, (c) is for C4da neurons, (d) is for dorsal pharyngeal sensilla ganglion, (d′) is for dorsal pharyngeal organ ganglion, (v) is for ventral pharyngeal sensilla ganglion and (e) is for entero-endocrinal cells. In D′ and D″, the scale bar represents 100 µm. In D′ and D″, the asterisk indicates the anterior part of the organ. A schematic representation of the whole larva indicates the depicted area. The PER index is calculated as the percentage of flies tested that responded with a PER to the stimulation ± 95% confidence interval (CI). A PER value of 1 means that 100% of the tested flies extended their proboscis following contact with the mixture, a value of 0.2 means that 20% of the animals extended their proboscis. The number of tested flies (n) is indicated on top of each bar. For each condition, at least 3 groups with a minimum of 10 flies per group were used. Each independent group is represented as an open circle. ns indicates p > 0.05, * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001 Fisher’s exact test. Further details can be found in the detailed lines, conditions and, statistics for the figure section. The data underlying this Figure can be found in S1 Data. Created in BioRender. Kurz, L. (2026). https://core.local.biorender.dev/api/short-link/9ojribp.

https://doi.org/10.1371/journal.pbio.3004042.g002

TrpA1 B, C and E isoforms are co expressed in some anterior neurons

As expected from a functionally pleiotropic protein, TrpA1 is expressed in many cell types in both larvae and adults. In addition, Gu and colleagues, have previously shown that each TrpA1 isoform has a unique expression pattern and that multiple isoforms are often co-expressed in the same cells. To investigate whether the shared ability to prime of B, C, and E isoforms can be attributed to their co-expression in some functionally important cells, we looked for cells that would express all 3 isoforms. For that, we crossed the TrpA1-isoform-GAL4 flies with UAS-gfp and compared their expression domain. Interestingly, all 3 isoforms were expressed in a single and isolated anteriormost neuron probably belonging to one of the bitter neurons of the sensory terminal organ (TO) and in few neurons of the Dorsal and Ventral organs (Figs 2D and 3A–3C–3C). Additionally, all three isoforms were also expressed in the horseshoe-shaped BLP larval brain neurons that have been shown to play a role in thermal nociception [53], while only C and E were expressed in the larval VNC (Figs 3D–3F). Finally, all three isoforms and the pan-isoforms were expressed in gut cells, which are likely to be entero-endocrinal cells (EEC) (Figs 2D and 3G–3I–3I). However, while as reported earlier, TrpA1-C and TrpA1-E were expressed in C4da larval nociceptors neurons [46], it was not the case for TrpA1-B (S6 Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. The expression pattern of the B isoform contains all the elements common to isoforms B, C and E.

(A–I) Confocal images of larvae expressing gfp under the control of the TrpA1-B-GAL4 driver (A, D and G), the TrpA1-C-GAL4 driver (B, E and H) or the TrpA1-E-GAL4 driver (C, F and I). Shown are the anterior parts (A–C), the brains (D–F) and the guts (G–I). In (A–C), the anterior is on the left and the larger square is a magnification of the most anterior part of the animal housing a neuron whose dendrite extends toward the external environment (t). In (D–I), the asterisk indicates the anterior part of the organ. (b) is for brain, (c) is for C4da neurons, (d) is for dorsal pharyngeal sensilla ganglion, (d′) is for dorsal pharyngeal organ ganglion, (v) is for ventral pharyngeal sensilla ganglion and (e) is for entero-endocrinal cells. In (B), strong gfp expression in C4da neurons makes it difficult to distinguish other transgene-expressing cells, but the neuron extending its dendrite toward the external environment is still detectable and outlined in the larger square. A schematic representation of the whole larva indicates the depicted area. Created in BioRender. Kurz, L. (2026). https://core.local.biorender.dev/api/short-link/9ojribp.

https://doi.org/10.1371/journal.pbio.3004042.g003

Since the TrpA1-B KI line has the most restricted expression pattern of the 3 reporter lines with no expression in the VNC and in the C4da larval nociceptors neurons, we further focused on this line. We first confirmed, via RNAi-mediated downregulation, that TrpA1 inactivation in the larval but not the adult TrpA1-B-GAL4 expression domain was sufficient to block priming (Fig 4A and 4B). This result, together with those previously reported, demonstrate that inactivating TrpA1 in either Gr66a-GAL4 [37] or in TrpA1-B-GAL4 larval cells prevent larval priming. This suggests that TrpA1-B-GAL4 and Gr66a-GAL4 are co-expressed in one or more cells in which TrpA1 expression would be required for priming. TrpA1 expression in bitter-sensing GRNs, including Gr66a+ neurons, has been well established in adults [49,54,55].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. TrpA1-B+ cells are Gr66a+ as well as necessary for priming with the TrpA1-B isoform sufficient.

(A) TrpA1 is necessary for priming in TrpA1-B+ cells. PER index to solutions of sucrose and sucrose + PGN (PGN200) of control flies (+/UAS-TrpA1_IR) as well as animals with the TrpA1 transcripts impaired via RNAi in TrpA1-B+ cells (TrpA1-B-GAL4/UAS-TrpA1_IR) was calculated. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (B) TrpA1 is necessary for priming in TrpA1-B+ cells only during larval stage. PER index to solutions of sucrose and sucrose + PGN (PGN200) of TrpA1-B-GAL4; TubG80^ts/UAS-TrpA1_IR animals with the TrpA1 transcripts impaired via RNAi in TrpA1-B+ cells only during the larval stages (left) or only during the adult stage (right) was calculated. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (C–C″) Confocal images of the anterior part of a representative larva expressing mCherry under the control of the Gr66a-LexA driver as well as gfp under the control of TrpA1-B-GAL4 driver. In C, dendrites from three Gr66a+ neurons are detectable, but only one cellular body is clearly visible. (D) TrpA1-B isoform is sufficient for priming in Gr66a+ cells while TrpA1-A is not. PER index to solutions of sucrose and sucrose + PGN (PGN200) of control flies impaired for priming (UAS-TrpA1-B; TrpA1⁻ and UAS-TrpA1-A; TrpA1⁻) as well as animals with the TrpA1 transcripts over-expressed only in Gr66a+ cells in the TrpA1⁻ mutant background (Gr66a-GAL4; TrpA1⁻/UAS-TrpA1-B; TrpA1⁻ and Gr66a-GAL4; TrpA1⁻; UAS-TrpA1-A; TrpA1⁻) was calculated. Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. (E) Impairing TrpA1 expression in Gr39a.b+ cells prevent priming. PER index to solutions of sucrose and sucrose + PGN (PGN200) of control flies (+/UAS-TrpA1_IR). Larvae were raised on conventional media and the resulting adults raised on antibiotics-containing media. The +/UAS-TrpA1_IR data from Fig 4E are those present on Fig 4A. The PER index is calculated as the percentage of flies tested that responded with a PER to the stimulation ± 95% confidence interval (CI). A PER value of 1 means that 100% of the tested flies extended their proboscis following contact with the mixture, a value of 0.2 means that 20% of the animals extended their proboscis. The number of tested flies (n) is indicated on top of each bar. For each condition, at least 3 groups with a minimum of 10 flies per group were used. Each independent group is represented as an open circle. ns indicates p > 0.05, * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, **** indicates p < 0.0001 Fisher’s exact test. Further details can be found in the detailed lines, conditions and, statistics for the figure section. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3004042.g004

Consistently, using the binary system GAL4/UAS and LexA/LexAop, we showed that one of the three Gr66a-positive neurons is TrpA1-B-GAL4 (Figs 4C, 4C′ and 4C″). We also excluded the gut part as being essential for priming since Gr66a-GAL4 is not expressed in the gut ([37] and S7 Fig). To further identify the neuron(s) and the TrpA1 isoform mediating priming, we try to rescue the pan-isoform TrpA1 mutant by providing different TrpA1 isoforms in Gr66a+ neurons. While the ability to prime that is lost in TrpA1 mutant was fully rescued by expressing TrpA1-B in Gr66a+ cells, this was not the case for TrpA1-A (Figs 4D and S8). Altogether, these results have several implications. (i) they demonstrate that the larval neurons and not the enteroendocrine cells are required for priming (ii) they demonstrate that the expression of TrpA1-B in Gr66a+ cells is sufficient for priming (iii) they strongly suggest that the TrpA1-B+/Gr66a+ anterior neuron is the cell in which TrpA1-B (or probably either C and E) need to be expressed to allow priming (iv) they demonstrate that all the TrpA1 isoforms are not functionally substitutable as far as priming is concerned. All of these results were confirmed using another GAL4 line (Gr39a.b), whose expression pattern in the larvae resembles that of TrpA1-B (S9A Fig) and which is not expressed in the EEC (S9C Fig). Co-expression confirmed that one of the Gr39a.b-GAL4 TO neurons is TrpA1-B positive (S9B Fig). Moreover, impairing TrpA1 expression in the Gr39a.b-GAL4 cells that do not include EEC was sufficient to abrogate the adult PGN avoidance phenotype observed in controls (Fig 4E).

The TrpA1-B positive TO neuron is activated by H₂O₂

After identifying, on the one hand, that TrpA1-B expression in the TrpA1-B+/Gr66a+ anterior neuron is necessary for priming and, on the other hand, that ROS are also involved, we hypothesized that the latter could be the signal activating the former. To test this model, we used calcium imaging to measure the ability of Gr66a+ and TrpA1+ anterior neurons to respond to H₂O₂. When larvae expressing Gr66a-GAL4/UAS-GCaMP were stimulated with a 1% H₂O₂ solution, only one of the three anterior neurons responded strongly to the stimuli (Fig 5A–5D). Interestingly, the neuron that responded to H₂O₂ was the only one to respond to quinine (Figs 5D and S10A–S10F). Using the same approach, we were able to show that the TO TrpA1-B-GAL4 neuron was activatable by H₂O₂ and that this response was abrogated if TrpA1 transcripts were cell-specifically downregulated (Fig 5E–5H). Having demonstrated that this larval neuron was required for priming and can be activated by quinine, we tested whether larval exposure to quinine using the monoassociation protocol was sufficient to prime larvae, as previously shown for uracil, and found that it was not (S10G Fig). These results support that a single TO TrpA1+/Gr66a+ neuron contributes to larval priming and is directly activated by ROS in a TrpA1-dependent manner.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. The TrpA1+ neuron at the tip of the larval anterior part responds to H₂O₂ in a TrpA1-dependent manner.

(A and E) Graphical representation of larval region observed during the GCaMP assay with 3 neurons detectable without stimulation in Gr66a-GAL4/UAS-GCaMP7s animals (A–D) and 1 neuron detectable without stimulation in TrpA1-GAL4/UAS-GCaMP7s larvae as well as in TrpA1-B-GAL4/UAS-GCaMP7s animals (E–H). (B–D) Addition of H₂O₂ triggers a calcium concentration increase in one out of the three Gr66a+ cells detectable in the anteriormost area of the larvae. (B) Representative images showing the GCaMP intensity before and after addition of either the control water or the H₂O₂ 1%. (C) Averaged ± SEM time course of the GCaMP intensity variations (ΔF/F0%) for Gr66a+ neurons. The addition of water (n = 9 flies) or H₂O₂ 1% (n = 6 flies) at a specific time is indicated by the arrow. (D) Averaged fluorescence intensity of negative peaks ± SEM in response to water (n = 9), H₂O₂ 1% (n = 6), or quinine (10 mM) (n = 7). (E–H) Addition of H₂O₂ triggers a calcium concentration increase in the unique TrpA1+ cell detectable in the anteriormost area of the larvae. (F) Representative images showing the GCaMP intensity before and after addition of either the control water or the H₂O₂ 1%. (G) Averaged ± SEM time course of the GCaMP intensity variations (ΔF/F0%) for TrpA1+ as well as TrpA1-B+ neurons and TrpA1-B cells expressing UAS-TrpA1_IR (TrpA1-B-GAL4/UAS-GCaMP7s; TrpA1_IR). The addition of water (n = 5–7 flies) or H₂O₂ 1% (n = 6–7 flies) at a specific time is indicated by the arrow. (H) Averaged fluorescence intensity of negative peaks ± SEM in response to water (n = 5–7 flies), H₂O₂ 1% (n = 6–7 flies), or quinine (10 mM) (n = 5–8 flies). Scale bar is 20 µm. * Indicates p = 0.0132, ** indicates p ≤ 0.0098, *** indicates p ≤ 0.009, Kruskal–Wallis H test was used and the reference for a given assay is the same genotype exposed to water. Further details can be found in the detailed lines, conditions and, statistics for the figure section. The data underlying this Figure can be found in S1 Data. Created in BioRender. Kurz, L. (2026). https://core.local.biorender.dev/api/short-link/9ojribp.

https://doi.org/10.1371/journal.pbio.3004042.g005

Uracil induces a Duox-dependent signal activating TrpA1-B+ neurons in naïve larvae

To validate our model that uracil released by bacteria activates intestinal Duox resulting in the production of ROS that are then detected by the anteriormost TrpA1-B+ neuron, we tested whether larvae exposed to uracil could release a Duox-dependent signal capable of activating TrpA1-B+ neurons in naïve individuals, i.e., larvae that had not been exposed to uracil or any prior stimulating cue (TrpA1-B-GAL4/UAS-GCaMP). To do so, larvae reared on antibiotic-containing food were transferred to liquid medium and incubated with 1 mM uracil or water for 3 hours. The resulting conditioned medium was assayed for calcium responses in naïve, non–uracil-exposed animals (Fig 6A). Strikingly, while neither uracil alone nor water from larvae not exposed to uracil induced any response, the medium conditioned by uracil-treated larvae triggered a calcium rise in the anteriormost TrpA1-B+ neuron (Figs 6B, 6C, and S11). Importantly, when Duox expression was specifically reduced in enterocytes of the uracil-treated larvae (mex-Gal4/UAS-Duox_IR), the conditioned medium no longer activated the TrpA1-B+ neuron (Fig 6B and 6C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. The TrpA1-B+ neuron responds to medium conditioned by larvae exposed to uracil.

(A) Graphical representation of the experimental design for the larva-conditioned medium. Fifteen larvae previously raised on antibiotic-containing medium are transferred at the L3 stage in a well containing either 100 µL of water or 100 µL of 1mM uracil diluted in water for 3 hours. After 3 hours in the dark at 25 °C, 50 µL of the medium in which the larvae were incubated is tested for its ability to trigger a calcium rise in the anterior part of TrpA1-B-GAL4/UAS-GCaMP7s larvae, which we consider naïve since they were raised on antibiotics and were not previously exposed to the larva-conditioned medium or to any of the chemicals tested. (B) Representative images showing the GCaMP intensity in the anteriormost TrpA1-B+ neuron before and after addition of 50 µL of the medium: uracil alone or media form w⁻ larvae exposed 3h to uracil or media form mex-GAL4/UAS-Duox_IR larvae exposed 3h to uracil. (C) Averaged fluorescence intensity of negative peaks ± SEM int the TrpA1-B+ anteriormost neuron in response to either the chemicals used in the conditioned media assay (water, uracil) or the medium in which the larvae were incubated in (media form w⁻ larvae+water, media form w⁻ larvae+uracil 1mM, media form mex-GAL4/+ larvae+uracil 1mM, media form +/UAS-Duox_IR+uracil 1mM, media form mex-GAL4/UAS-Duox_IR+uracil 1mM). Each symbol represents a tested larva. (D) ROS are produced by larvae following exposure to uracil. L3 larvae were exposed to uracil and the presence of ROS was assayed within the wells over time. To confirm that ROS are the detected chemicals, Vitamin C (VitC) was applied to the samples. Scale bar is 20 µm. *p < 0.05, **p ≤ 0.0012, ***p ≤ 0.0002. Statistical analyses were performed using the Kruskal–Wallis test for assays involving w⁻ animals (non-normal distributions) and one-way ANOVA followed by Tukey’s multiple-comparisons test for assays involving GAL4/UAS lines (after confirmation of normality). Further details can be found in the detailed lines, conditions and, statistics for the figure section. The data underlying this Figure can be found in S1 Data. Created in BioRender. Kurz, L. (2026). https://core.local.biorender.dev/api/short-link/9ojribp.

https://doi.org/10.1371/journal.pbio.3004042.g006

To determine whether ROS is produced by larvae under these conditions and to characterize its time course, we measured ROS levels over time using a chemiluminescent probe in wells containing exposed larvae. Larvae exposed to buffer exhibited a basal level of ROS that remained stable over the course of the experiment (Fig 6D). This signal was abolished upon addition of vitamin C (VitC). In contrast, addition of uracil triggered an increase in ROS compared to buffer alone; this signal then gradually decreased over time. At 3 hours, ROS levels following uracil exposure remained higher than in the buffer condition. These results indicate that uracil exposure generates a Duox-dependent signal in donor larvae—likely a ROS such as H₂O₂—that can activate TrpA1-B+ neurons in naïve recipients.

Free-feeding assays confirm PGN avoidance and reveal its dependence on larval microbial exposure and TrpA1

Using the PER method, we showed that adult avoidance of PGN relies on larval exposure to bacteria that produce uracil and is mediated by ROS and specific TrpA1 isoforms in host Gr66a+ neurons. However, since proboscis extension reflects only one stage of a complex feeding sequence involving smell, tarsal detection, ingestion, and pharyngeal evaluation, we asked whether this aversion also manifests in a more physiological and ecologically relevant context. We therefore turned to a two-choice free-feeding paradigm in which flies can move freely, explore the environment, and voluntarily ingest the substrate of their choice. Using a 96-well assay with colored agar [56], we assessed preference by scoring abdominal coloration after ingestion (Fig 7A and 7B). Control (w⁻) females do not discriminate between food sources when 1 mM sucrose is associated with a blue or red dye (preference index: P.I. = 0.5), flies with purple abdomen were observed (Fig 7B and 7C). However, animals robustly preferred 1 mM sucrose over a sucrose + PGN mixture, displaying a strong preference index for sucrose alone (P.I. > 0.8) (Fig 7C). To determine whether larval microbial experience was required for the emergence of this adult PGN avoidance, we tested flies raised on antibiotics throughout the larval stage. These individuals showed a markedly reduced avoidance of sucrose+PGN (P.I. < 0.55), with most animals ingesting both solutions. Finally, to evaluate the contribution of TrpA1—as anticipated from our PER results—we tested TrpA1-KO flies. These mutants consumed sucrose and sucrose+PGN equally, resulting in a strongly diminished preference index toward sucrose alone (P.I. = 0.51) (Fig 7C). Finally, the TrpA1-BCE-KO animals did not distinguish sucrose alone from the sucrose mixed with PGN (P.I. = 0.51) while AD-KO did make a discrimination, indicating that -A and -D isoforms are dispensable and suggesting −B and/or −C and/or −E as necessary.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. PGN is avoided by free-moving animals in a 2-choices assay.

(A–C) Larval cohabitation with bacteria and the TrpA1 channel are necessary for a PGN-triggered aversion in a 2-choices assay. (A) The plates with the dye and (B) the flies that ingested the colored media leading to (C) the preference indexes for different conditions and genotypes as well as the fraction of flies that ingested the proposed mixtures. For (C), the averaged preference index is in red with the standard error of the mean obtained from 6 to 8 independent experiments. The total amount of females tested is indicated on top of each assay. A preference index (PI) over 0.5 indicates a preference toward sucrose while a PI of 0.5 indicates either that two opposite behaviors occurred in the population or that a majority of animals consumed both solutions. The fraction of flies that ingested one, the other or both solutions allow to discriminate. In light colors are the fractions of flies that ingested one solution or both solutions. The animals from all the experiments were pooled for statistical analyses. ns indicates p > 0.05, **** indicates p < 0.0001 Chi-squared Test. Further details can be found in the material and methods section and in the source data file. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3004042.g007

Together, these results demonstrate that adult female flies avoid ingesting sucrose + PGN mixtures even under free-feeding conditions and that this avoidance requires prior larval cohabitation with bacteria. Moreover, TrpA1 appears to be crucial in limiting PGN ingestion, probably due to its role in the sensory priming that occurs during larval microbial exposure.

Fly stocks

The reference strain in this study corresponds to w1118 (w⁻) BL#5905, mex-GAL4 BL#91367, UAS-Duox_IR BL#38907, mex-GAL4; TubG80ts (this study and [63]; TubG80ts is BL#7016), UAS-Catalase BL#24621, TrpA1-KO/ BCE-KO/ AD-KO/ BC-KO/ E-KO/ A-KI/ B-KI/ C-KI/ D-KI/ E-KI, TrpA1-GAL4/ A-GAL4/ B-GAL4/ C-GAL4/ E-GAL4 (all TrpA1 genetic tools were very kind gift from the Xiang lab with a lot of support and advices [46]), UAS-GFP6x BL#52262, TrpA1-B-GAL4, UAS-GFP6x (this study), Gr66a-LexA; LexAop-mCherry (gift from Zhang lab [64]), UAS-TrpA1-IR BL#36780, Gr39a.b-GAL4 BL#57632, Gr66a-GAL4 BL#28801, TubG80ts; UAS-TrpA1_IR (this study), UAS-TrpA1-B; TrpA1⁻ (this study), UAS-TrpA1-A; TrpA1⁻ (this study), Gr66-GAL4; TrpA1⁻ (this study), UAS-GCaMP7s (gift from Matthieu Cavey), UAS-GCaMP7s; UAS-TrpA1_IR (this study) and Gr66a-LexA, lexAop-GAL80; ppk23-GAL4 (gift form the Gordon lab).

Fly culture

Flies were grown at 25 °C on a yeast/cornmeal medium in 12 h/12 h light/dark cycle-controlled incubators. For 1 L of food, 8.2 g of agar (VWR, cat. #20768.361), 80 g of cornmeal flour (Westhove, Farigel maize H1), and 80 g of yeast extract (VWR, cat. #24979.413) were cooked for 10 min in boiling water. 5.2 g of Methylparaben sodium salt (MERCK, cat. #106756) and 4 mL of 99% propionic acid (CARLOERBA, cat. #409553) were added when the food had cooled down. It is important to mention that our conventional media allows the presence of several bacterial species [37] and is a protein-rich and sugar-poor culture medium. For germ-free condition, we raised flies on antibiotic (ATB) media. A mix of four ATB is added to the conventional media: Kanamycin + Tetracycline + Ampicillin + Erythromycin. For the experiments with Vitamin C and H₂O₂, we add 200 µL of Vitamin C at 0. 2 mg/mL on eggs laid for 6 hours on conventional media and 200 µL of H₂O₂ 1% on eggs laid for 6 hours on ATB media. We use a protein-rich rearing media, in contrast to the sugar-rich media used in some other laboratories. Consequently, our animals may be more sensitive to low sucrose concentrations (1 mM) compared to flies raised on sugar-rich media. Therefore, assays using 1 mM sucrose may not yield reproducible results when performed with flies reared on sugar-rich media.

Bacterial stocks

Levilactobacillus brevis (L. brevis) is a gift form François Leulier’s Lab.

Erwinia carotovora subsp. carotovora 15 (Ecc¹⁵) [65], Ecc¹⁵pyrE::tn5 [45] and Ecc¹⁵ ∆4 [66] are gifts from Won-Jae Lee Lab.

Bacterial culture

Ecc strains were grown on standard LB agar plates at 30 °C at least 18 hours and L. brevis was grown in MRS agar in anaerobic-like conditions at 37 °C for at least 48 hours. A single colony was used to prepare liquid cultures. Ecc strains were grown in 200mL LB media (Lennox, Sigma-Aldrich ref L3022 or L2897) at 30 °C with agitation and L. brevis was grown in 50 mL MRS liquid media (Sigma-Aldrich ref 69964 agar and 69966 broth) static at 37 °C in sealed 50 mL tubes for anaerobic conditions. After overnight growth, the cultures were centrifuged for 15 min at 4,200 rpm. Bacterial doses were adjusted by measuring culture turbidity at an optical density of 600 nm to set the culture to OD 1. For the heat kill experiment, we set the OD to 1 from an overnight-incubated culture, then incubate this sample at 95 °C for 5 min. We confirm the heat-kill experiment using an MRS agar plate containing the sample before and after heat kill. For the supernatant experiment, we set the OD to 1 from an overnight-incubated culture, then centrifuged and the supernatant is then filtered with a 0.45 µm filter.

Proboscis Extension Reflex (PER) behavior

All flies used for the test were females between 5 and 7 days old. Unless experimental conditions require it, the flies are kept and staged at 25 °C to avoid any temperature changes once they are put into starvation. The day before, the tested flies are starved in an empty tube with water-soaked plug for 24 h at 25 °C. Eighteen flies are tested in one assay, 6 flies are mounted on one slide and in pairs under each coverslip. To prepare the slide, three pieces of double-sided tape are regularly spaced on a slide. Two spacers are created on the sides of each piece of tape by shaping two thin cylinders of UHT paste. To avoid the use of carbon dioxide flies are anesthetized on ice. Under the microscope, two flies are stuck on their backs, side by side, on same piece of tape so that their wings adhere to the tape. A coverslip is then placed on top of the two flies and pressed onto the UHT paste, blocking their front legs and immobilizing them. Once all slides are prepared, they are transferred to a humid chamber and kept at 25 °C for 1.5 h to allow the flies to recover before the assay. Flies are tested in pairs, the test is carried out until completion on a pair of flies (under the same coverslip), and then move on to the next pair. Before the test, water is given to each pair of flies to ensure that the flies are not thirsty and do not respond with a PER to the water in which the solutions are prepared. Stimulation with the test solution is always preceded and followed by a control stimulation with a sweet solution, to assess the fly’s condition and its suitability for the test. During the test small strips of filter paper are soaked in the test solution and used to contact the fly’s labellum (three consecutive times per control and test phase). Contact with the fly’s proboscis should be as gentle as possible. Ideally, the head should not move. A stronger touch may prevent the fly from responding to subsequent stimulation. Based on the protocol needed the test is done following the sequence and the timing in the Table 1 below.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Aversion protocol.

https://doi.org/10.1371/journal.pbio.3004042.t001

All solutions to be tested are prepared the test day and stored at room temperature. In the aversion protocol, the control stimulation is performed with 1 mM sucrose (D(+)-sucrose ≥99.5%, p.a. Carl Roth GmbH + Co. KG). This concentration is sufficient to elicit a PER but is not so high as to influence the response to the subsequent test stimulation. PGN is E. coli K12: Invivogen, catalog code #tlrl-kipgn. After each control or test stimulation, a water-soaked strip is used to tap the proboscis and clean it. The response of each fly to each stimulation is recorded. We distinguish between two types of flies: those able to respond and those unable to respond. To make this distinction, we use three control stimulations before the test and three after the test. Flies that do not respond during the control stimulations (1 mM sucrose) are classified as “unable to respond” and are excluded from the analysis, regardless of their responses during the tests. Flies that respond during the control stimulations are classified as “able to respond,” and only their responses during the tests are included in the statistical analyses. Consequently, each fly ‘able to respond’ is attributed a 0 when it does not extend the proboscis toward the tested solution while it is attributed a 1 when it extends the proboscis toward the tested solution. Each fly represents an independent observation and the PER index is calculated as the percentage of flies tested that responded with a PER to the TEST stimulation and represented with the ± 95% confidence Interval (CI). For instance, if out of 30 tested animals, 21 extended their proboscis toward the tested solution, the PER index will be of 0.7. As the values obtained from one fly are categorical data with a Yes or No value, we used the Fisher’s exact test and the 95% CI to test the statistical significance of a possible difference between a test sample and the related control. This test takes into account the response of each fly and does not simply compare the distribution of several values without taking into account the size of the cohort used to generate this value. For PER assays, at least 3 independent experiments involving up to 18 flies each were performed. The results from all the flies were gathered and the total amount of animals tested is indicated in the graph. In addition, we do not show the average response from one experiment representative of the different biological replicates, but the PER index from all the tested animals in one graph. While several independent batches of flies are tested, the final PER index includes all the flies and the PER index of each independent batch is also shown as open dots in the graphs to illustrate the variability. In case of stage-dependent experiments, flies are shifted from one condition to another upon hatching.

Monoassociation

An oviposition of w⁻ germ-free flies is set up on an apple agar plate with yeast at 25 °C for 6 hours. Three petri dishes are filled with 2.6% bleach, 70% ethanol (Ethanol 96° RPE Carlo Erba Ref 414638) and autoclaved purified distilled water, respectively. The embryos are collected by filling the plate in which oviposition occurred with purified distilled water and using a small brush to gently detach them from the flies’ food. A 40 µm cell strainer is used to collect the embryos. The cell strainer with the embryos is then dipped into: bleach 2.6% for 5 min, ethanol 70% for 1 min, purified water for 1 min, ethanol 70% for 1 min, purified water for 1 min. The brush used to collect the embryos is sterilized in 2.6% bleach for 10 min, rinsed and then used to transfer the sterile embryos onto the desired media.

We then deposit bleach eggs on ‘steril’ conventional media (medium is deposited in sterilized Falcon tubes in sterile conditions and sealed with a cap) and add 200 µL of the monoassociation we want (MRS broth, MRS broth + L. brevis or supernatant or heat-killed bacteria, LB broth, LB broth + bacteria, LB broth + uracil, LB broth + thymine, LB broth + cytosine, LB brothy + H₂O₂).

RNAi

All the tested animals were F1 obtained from a cross between parents possessing the GAL4 transgene and parents possessing the UAS-RNAi construction or from crosses between w⁻ animals (genetic background of the GAL4 and UAS lines) and a transgenic line to serve as ctrl.

For stage-dependent RNAi experiments, the ubiquitously expressed Tub-GAL80^ts, that inhibits the activity of GAL4, is temperature sensitive: it’s active at 18 °C and inactivated at 29 °C, allowing the expression of UAS when animals are raised at 29 °C and preventing it at 18 °C. For RNAi in larvae only, animals were raised from eggs to early pupae at 29 °C and then shifted at 18 °C. for RNAi in adults only, eggs, larvae and pupae were raised at 18 °C and the virgin adults shifted to 29 °C.

Larval-conditioned medium

Conditioned medium was used to highlight signals such as ROS produced by enterocytes in response to 1 mM uracil to potentially activate the TrpA1-B+ neuron. In a 96-well plate (NUNC), conditioned medium was obtained by placing 15 larvae of adequate genotype, reared on antibiotic-containing medium, in a well containing either 100 µL of water or 100 µL of 1mM uracil diluted in water for 3 hours. Fifty microliters of conditioned medium was then tested on the anterior part of TrpA1-GAL4/UAS-GCaMP7s larva reared on normal food in calcium imaging experiments. 100 µL of 1mM uracil was also placed in a well without any larvae during 3 hours to test the activity of 1mM uracil on the TrpA1-B+ neuron localized in the terminal organ of TrpA1-GAL4/UAS-GCaMP7s larva.

Reactive oxygen species measurement in first-stage larvae

Thirty L1 larvae were collected from 24-hours monoassociation tubes containing either LB or LB supplemented with 1 mM uracil. Samples were homogenized and centrifuged. ROS levels were quantified using the Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen, #A12222), as previously described [67]. Briefly, 20 µL of sample was mixed with 100 µL of reaction buffer containing 50 µM Amplex Red reagent and 0.2 U/mL horseradish peroxidase. Fluorescence intensity was measured using a CLARIOstar Plus microplate reader (BMG Labtech) with excitation at 545 nm (±15 nm) and emission at 590 nm (±8 nm). A standard curve was generated using serial dilutions of H₂O₂ and used to calculate H₂O₂ concentrations in the samples.

Reactive oxygen species measurement over time in third-stage larvae

Reactive oxygen species (ROS) levels in the larval medium were detected using the L-012 chemiluminescent probe (Merck-Sigma-Aldrich). Before assays, 15 third-instar larvae were placed in 100 µL of PBS per well in a 96-well Microfluor 1 black plate (Thermo Scientific) and 8 wells of 15 larvae were used per condition. 8 additional wells containing PBS without larvae were used as background control. The experiment was repeated twice. To initiate the experiment, the PBS was replaced with either 100 µL of PBS alone (control), 100 µL of PBS supplemented with ascorbic acid (20 µg/mL), 100 µL of PBS supplemented with uracil (1 mM), or 100 µL of PBS supplemented with uracil (1 mM) and ascorbic acid (20 µg/mL). L-012 was added at a final concentration of 200 µM in all wells.

Chemiluminescence was measured every 5 min during 3 h using an Infinite M200Pro microplate reader (TECAN) with an integration time of 3 s per well at room temperature in the dark. ROS levels were expressed as relative luminescence units (RLU). At the end of the experiment, H₂O₂ was added to each well at final concentration of 1 mM to verify that L-012 had not been depleted during the experiment. Background luminescence values were subtracted to luminescence values detected in wells containing larvae.

Choice assay

Two-way choice assays were performed essentially as described [68]. Briefly, 5- to 7-day-old female flies (30 ± 10 flies per experiment) were starved for ∼18 h at 25 °C and placed into a 96-well plate with 1% agarose in each well. Alternating wells contained either red (sulforhodamine B, 0.2 mg/mL; Sigma-Aldrich) or blue dye (bleu E133 meilleur du chef, 0.3%). Then we deposit 10 µL of solution containing sucrose 1mM or Sucrose1mM + PGN 200 µg/ml on top of the well. Each experiment is carried out in duplicate by placing the components on two different colors (e.g., sucrose 1mM on blue wells for plate A and on red wells for plate B) to verify whether there is an effect of color. The flies were allowed to feed in the dark for 90 min at ∼23 °C. We transferred the plates to −20 °C for 15 min in order to rapidly freeze the flies in order to count the flies based on their abdominal color under a binocular microscope. The numbers of flies with blue, red, or purple (mixed red and blue) abdomens were tabulated (Fig 7A and 7B), and the P.I. was determined: (N_suc + 0.5N_mix)/(N_suc + N_test + N_mix). The flies without colors in their abdomen were discarded. The number of flies indicated corresponds to animals with colored abdomens.

Microscopy

No immunostaining was performed. To image larvae sensory organs and body as well as gut and brains, full larvae were killed with heat and mounted whole on slides using Vectashield fluorescent mounting medium. Organs were dissected in PBS, rinsed with PBS and directly mounted on slides using Vectashield fluorescent mounting medium. The tissues were visualized directly after. Images were captured with LSM 780 Zeiss confocal microscope (20× air objective was used). Images were processed using Adobe Photoshop and FiJi softwares.

Statistics

GraphPad Prism 8 software was used for statistical analyses. For in vivo calcium imaging, the D’Agostino–Pearson test to assay whether the values are distributed normally was applied. As not all the data sets were considered normal, non-parametric statistical analysis such as Kruskal–Wallis H test was used for all the data presented. For PER datasets, each fly ‘able to respond’ is attributed a 0 when it does not extend the proboscis toward the tested solution while it is attributed a 1 when it extends the proboscis toward the tested solution. As the values obtained from one fly are categorical data with a Yes or No value, we used the Fisher’s exact test and the 95% CI to test the statistical significance of a possible difference between a test sample and the related control. This test takes into account the response of each fly and does not simply compare the distribution of several values without taking into account the size of the cohort used to generate this value. For PER assays, at least 3 independent experiments involving up to 18 flies each were performed. The results from all the flies were gathered and the total amount of animals tested is indicated in the graph. In addition, we do not show the average response from one experiment representative of the different biological replicates, but the PER index from all the tested animals in one graph. Each fly represents an independent observation and the PER index is calculated as the percentage of flies tested that responded with a PER to the TEST stimulation and represented with the ± 95% CI. For instance, if out of 30 tested animals, 21 extended their proboscis toward the tested solution, the PER index will be of 0.7. Several independent batches of flies are tested, the final PER index includes all the flies and the PER index of each independent batch is also shown as open dots in the graphs to illustrate the variability.

Chemicals

PGN-ECndi ultrapure peptidoglycan catalog code #tlrl-kipgn InvivoGen USA, D (+)-Saccharose ≥99.5%.ref 4621.1 Carl Roth, H₂O₂ solution ref. H3410 (Figs 1 and 5) and H1009 (S5 Fig) Sigma-Aldrich, Acid L-ascorbic (Vitamin C) ref. A92902 Sigma-Aldrich, Uracil ≥99.0%. refU0750 Sigmal-Aldrich, Cytosine ≥99.0%. refC3506 Sigmal-Aldrich, Thymine ≥99.0%. refT0376 Sigmal-Aldrich Quinine ref. Q1125 Sigma-Aldrich and Catalase ref.C40 Sigmal-Aldrich.

Calcium imaging experiments

In vivo larval calcium imaging experiments were performed on beginning of third-instar larvae. Larvae were immobilized in a small drop of distilled water placed between two plastic coverslips (22 mm × 22 mm, Agar Scientific). The two coverslips were held together to prevent movement of the anterior part of the larva using an alligator clip attached to a support. Stimulation was performed manually using a pipette with gel-loading tip by applying water, Uracil (1 mM) diluted in water, H₂O₂ (1%), quinine (10 mM, Sigma-Aldrich #Q11125) diluted in water or 50 µL conditioned medium through a hole made in the upper coverslip allowing the solutions to come into contact with the larval TO. For larval calcium imaging experiments, GCaMP7s was excited using a Lumencor diode light source at 482 nm ± 25. Emitted light was collected through a 505–530 nm band-pass filter. Images were collected every 500 ms using a Hamamatsu/HPF-ORCA Flash 4.0 camera and processed using Leica MM AF 2.2.9. Each experiment consisted of a recording of 70–100 images before stimulation and 160 images after stimulation. Data were analyzed as previously described [31] by using FIJI (https://fiji.sc/). For larvae fluorescence quantifications, a background fluorescence variation was calculated and subtracted to the fluorescence variation signal.

Detailed lines, conditions and statistics for the figure section

All these data are downloadable with the source data file.

Graphical abstract and some icons were created in BioRender (https://www.biorender.com/) under an institutional license (Aix-Marseille Université).