Dietary antioxidants play an important role in preventing oxidative stress. Whether animals in search of food or brood sites are able to judge the antioxidant content, and if so actively seek out resources with enriched antioxidant content, remains unclear. This study shows that the vinegar fly Drosophila melanogaster detects the presence of hydroxycinnamic acids (HCAs)-potent dietary antioxidants abundant in fruit-via olfactory cues. Flies were unable to smell HCAs directly but were found to be equipped with dedicated olfactory sensory neurons detecting yeast-produced ethylphenols that are exclusively derived from HCAs. These neurons were housed on the maxillary palps, expressed the odorant receptor Or71a, and were necessary and sufficient for proxy detection of HCAs. Activation of these neurons in adult flies induced positive chemotaxis, oviposition, and increased feeding. Further, fly larvae also sought yeast enriched with HCAs and used the same ethylphenol cues as the adults but relied for detection upon a larval unique odorant receptor (Or94b), which was co-expressed with a receptor (Or94a) detecting a general yeast volatile. Also, the ethylphenols acted as reliable cues for the presence of dietary antioxidants, as these volatiles were produced-upon supplementation of HCAs-by a wide range of yeasts known to be consumed by flies. For flies, dietary antioxidants are presumably important to counteract acute oxidative stress induced by consumption or by infection by entomopathogenic microorganisms. The ethylphenol pathway described in this study adds another layer to the fly's defensive arsenal against toxic microbes (Dweck, 2015).

Dietary antioxidants play a fundamental role in preventing oxidative stress by regulating levels of free radicals and other reactive oxygen species. Dietary antioxidants thus constitute a significant nutritional reward. Indeed, for example, frugivorous birds actively seek out fruit with a high content of antioxidants and, furthermore, are able to judge the fruit's antioxidant content by relying on visual cues alone. Whether feeding partiality toward food enriched with dietary antioxidants, as well as the ability to judge antioxidant content, is widespread remains, however, an open question (Dweck, 2015).

Oxidative stress is of importance not only to long-lived organisms, but also to animals with shorter lifespan, such as insects, in which, apart from aging, oxidative stress has also been shown to accrue from, e.g., cold exposure and through ingestion of environmental toxins. This study examined how Drosophila reacts to the presence of two polyphenolic dietary antioxidants, the hydroxycinnamic acids (HCAs) p-coumaric acid and ferrulic acid. These two HCAs are particularly abundant in fruit, the primary breeding substrate of flies, and therefore are presumably important antioxidants in wild fly populations. In flies, polyphenol antioxidants have been shown to offer protection against induced oxidative stress, and also to prolong lifespan (Dweck, 2015).

This study demonstrates that flies are able to detect the presence of HCAs via olfactory cues. Flies are, however, unable to smell HCAs directly, but they are equipped with a dedicated olfactory sensory neuron (OSN) class -- localized on the maxillary palps -- that detects volatile ethylphenols, which are exclusively derived from HCAs. Larval flies also do the proxy detection of HCAs via the same ethylphenols, albeit with a different, but similarly tuned, larval unique odorant receptor (OR). These results provide the first indication that animals are able to use olfactory cues to judge content of dietary antioxidants (Dweck, 2015).

Attempts were made to confirm that a diet supplanted with HCAs remedies the negative effects of induced oxidative stress. Flies with 20 mM paraquat (a pesticide that induces oxidative stress) dissolved either in yeast medium or in HCA-inoculated yeast medium. Flies fed with paraquat dissolved in HCA-inoculated yeast showed a significant enhancement in both survival and locomotor activity compared to flies treated with paraquat dissolved in the yeast medium. Can flies smell HCAs? Three different olfactory assays were used to monitor chemotaxis, oviposition, and feeding, respectively. In none of these assays did flies show any reaction to p-coumaric acid or ferulic acid. A lack of behavior does not, however, mean that flies are unable to smell these substances. Hence, electrophysiology was used, more specifically to single-sensillum recordings (SSRs), to investigate whether stimulation with HCAs induce alterations in spike firing rate. Using the two HCAs as a stimulus (10^-2), a system-wide screen was performed across all 48 olfactory sensory neuron (OSN) classes present on the flies' antennae and maxillary palps. Neither HCA yielded any activity from any of the contacted OSNs. It is thus concluded that the olfactory system is unable to detect these two chemicals (Dweck, 2015).

Although flies are unable to smell the HCAs directly, they could still be able to detect the presence of these chemicals via proxies. Many yeast species, including those consumed by flies, are known to be able to metabolize HCAs into ethylphenols, specifically 4-ethylphenol and 4-ethylguaiacol. Attempts were made to verify that fruits utilized by flies contain HCAs. Indeed, high-performance liquid chromatography (HPLC) analysis of banana pulp revealed the presence of both p-coumaric acid and ferulic acid. Next, whether the HCA amounts present in banana were sufficient to induce production of ethylphenols by yeasts was investigated. Banana-based medium was innoculated with Brettanomyces bruxellensis, a yeast species isolated from wild flies and known for its potent ability to convert HCAs into ethylphenols. Indeed, in yeasts grown on medium mixed with banana pulp, ethylphenols were identifed in the headspace. Similarly, growth of Brettanomyces on medium supplanted with HCAs resulted in the production of ethylphenols, but not when Brettanomyces was grown on standard medium (Dweck, 2015).

Do flies react to the HCA induced changes in the yeast's volatile headspace? It was first verified that flies reacted to the smell of Brettanomyces yeast, which they did, with flies displaying strong preference for this yeast in the three previously mentioned assays. Next, flies were confronted with a choice between Brettanomyces grown with or without HCAs (henceforth referred as HCA+ and HCA-). In all assays, flies clearly preferred HCA+ yeasts. To verify that this preference is mediated via olfaction, this experiment was repeated with flies lacking Orco, a co-receptor necessary for function in the majority of all OSNs. Indeed, Orco-/- flies did not differentiate between the two treatments in any of the three assays, demonstrating that OSNs expressing ORs are necessary for this behavior. It was next asked whether the preference for HCAs is mediated via ethylphenols. To address this issue, flies were provided with a binary choice of Brettanomyces (grown on standard medium) spiked with either 4-ethylguaiacol and 4-ethylphenol (10^-4 dilution) or solvent (mineral oil). Flies preferred the Brettanomyces with added ethylphenols in all three assays. Similarly, flies that were given a choice between HCA+ Brettanomyces and yeasts grown on standard medium, but spiked with ethylphenols, showed no preference either way in all assays. Finally, the behavioral valence of the ethylphenols themselves was examined, and as expected, flies in all three assays showed a strong preference for these yeast metabolites. It is hence concluded that although flies are unable to smell HCAs directly, they are able to detect volatiles derived from HCAs (Dweck, 2015).

How do flies detect the ethylphenols? A system-wide SSR screen was performed stimulating with the two ethylphenols. Strong responses to these two chemicals (at 10^-4 dilution) were exclusively observed from a single OSN class, namely palp basiconic type 1B (pb1B). To determine the specificity of these neurons, a battery of 154 compounds (screened at a higher dose [10^-2] was tested to obtain the upper limit of the receptive range). The chosen stimulus included representatives of all relevant chemical classes but focused on substances of structural similarity to the HCA derived ethylphenols. Out of the screened chemicals, none produced a stronger response than 4-ethylguaiacol, and only nine of the compound -- all structurally similar to 4-ethylguaiacol -- yielded a response of >100 spikes/s. Dose-response relationships were examined for the six most efficient agonists using gas chromatography (GC) for controlled stimulus delivery. As suspected, 4-ethylguaiacol was indeed the most efficient ligand, triggering responses already at 10^-7dilution. To determine whether the additional ligands for pb1B also activate other OSN classes, an exhaustive SSR screen was performed, this time stimulating with the seven primary agonists for pb1B (at 10^-4 dilution) across all 48 OSN classes. With the exception of guaiacol, which also strongly activated antennal basiconic type 6B (ab6B, expressing Or49b), none of the other volatiles triggered significant activity from OSN classes other than pb1B. It is hence concluded that at ecologically relevant concentrations, the ethylphenols and structurally similar phenolic compounds exclusively activate the pb1B pathway (Dweck, 2015).

The presence of HCAs might also lead to other changes in the yeast's volatile profile, which in turn could activate other subpopulations of OSNs. To control for this eventuality, repeated the system-wide SSR screen was repeated, but employed GC was employed to screen headspace collections from HCA+ and HCA- Brettanomyces. Stimulation with the former activated 12 OSN classes, whereas nine were activated with the latter. The additional OSN classes activated by the HCA+ Brettanomyces headspace were pb1B, ab5B, and ab9A. The pb1B neurons were, as expected, triggered by 4-ethylguaiacol and 4-ethylphenol (as identified via GC-linked mass spectroscopy). The large amount of 4-ethylguaiacol in the HCA+ sample was also sufficient to trigger weak activity from ab9A, whereas the response from ab5B in the HCA+ sample stemmed from greatly increased levels of phenylethanol compared to the HCA- treatment (Dweck, 2015).

Attempts were then made to determine which of these three OSN classes are necessary for the proxy detection of HCAs. The temperature-sensitive mutant dynamin Shibire^ts was used to shut down synaptic transmission in the OSN classes specifically activated in the HCA+ sample. At the restrictive temperature (32°C), flies expressing Shibire^ts from the promoter of the OR expressed in pb1B OSNs--Or71a--displayed no preference toward HCA-inoculated yeasts in any of the three employed assays. The preference of flies with ab9A and ab5A silenced (via Shibire^ts expression from the promoters of Or69a and Or47a, respectively was, however, not different from that of flies tested at a permissive temperature (25°C) or from parental control lines at restrictive temperatur. It is hence concluded that Or71a alone is necessary for the substitute detection of HCAs. Is activation of pb1B then sufficient to induce the observed preference? Next, expression of the temperature-sensitive cation channel dTRPA1 was driven in the pb1B OSNs, which enabled the conditional activation of this specific OSN population at temperatures above 26°C. Specific activation of pb1B neurons indeed triggered attraction, egg laying, and feeding. In short, the Or71a pathway is both necessary and sufficient for the detection of the HCA derived yeast volatiles (Dweck, 2015).

In nature, flies are not only confronted with Brettanomyces, but also encounter a wide range of yeast species. If the ethylphenols indeed serve as a general signal enabling identification of HCA enriched substrates, other yeast growing on HCA-containing sources would be expected to also produce these volatiles. To investigate this issue, HCA-induced production of volatile phenols was examined in a range of additional yeast species, namely Wickerhamomyces anomalus, Torulaspora delbrueckii, Hanseniaspora uvarum, Metschnikowia pulcherrima, and Saccharomyces cerevisiae. All of these yeasts have previously been isolated from the surface or guts of drosophilid flies. The conversion of HCAs into volatile phenols involves two steps: first a hydroxycinnamate decarboxylase enzyme converts the HCAs into vinyl derivatives, which are subsequently reduced by a vinyl phenol reductase into the corresponding ethyl derivatives (4-ethylphenol and 4-ethylguaiacol). The examined yeasts ability to complete these synthesis steps differed, with none of the yeasts being able to synthesize 4-ethylphenol. Nevertheless, when stimulated with the HCA+ yeast headspace, the amounts and types of volatile phenols present in were sufficient to activate pb1B OSNs in GC-SSR measurements. Moreover, flies confronted with the same binary choice between HCA+ and HCA- yeasts as before (Dweck, 2015).

It is, however, not inconceivable that HCAs in combination with other yeast might cause other changes in the volatile profile than does the combination of Brettanomyces and HCAs. To examine this issue, a system-wide GC-SSR screen was again performed, now stimulating with the headspace from the five above mentioned yeasts. Although the other yeast headspace activated a slightly different subset of OSNs than did Brettanomyces, only ab9A and pb1B were additionally recruited by stimulation with the HCA+ yeast headspace compared to HCA-. Hence, it is concluded that ethylphenols serve as a consistent and reliable signal for the presence of HCAs (Dweck, 2015).

Being able to detect HCA-enriched patches and favorable food yeasts should be important not only for adult flies, but also for larvae. Although essentially confined to their food, the microhabitat of larvae is not uniform, and thus being able to navigate toward suitable pockets within the fruit home should be an important ability. Although Or71a is not expressed in the larval stage, it is possible that among the larval unique OR genes, there are receptors that are able to make the same proxy detection of HCAs as adults do, or, alternatively, to detect HCAs directly. Whether larvae respond behaviorally to HCAs was examined. Larvae confronted with HCAs in a binary-choice larval olfactory preference assay showed no reaction to the HCAs. Although displaying no overt behavior in response to the presence of HCAs, larvae could still be able to smell HCAs. To examine whether larvae can smell HCAs, SSR was performed from the dorsal organ (DO)-the larval nose. The DO is innervated by 21 OSNs, and by gently inserting the recording electrode into this structure, it was possible to simultaneously record the activity of (presumably) all OSNs residing within the DO. Stimulation with HCAs yielded no activity from any of the discernable neurons in multiple recordings. It is thus concluded that larvae, like adults, are unable to detect the presence of HCAs directly (Dweck, 2015).

Larvae could still, however, make the same proxy detection of HCAs as adults. First whether larvae respond behaviorally to the odor of Brettanomyces grown with or without HCAs-was examined. Both HCA+ and HCA-Brettanomyces triggered positive chemotaxis from the larvae in the olfactory preference assay. Larvae confronted with a binary choice between HCA+ and HCA- cultured Brettanomyces clearly preferred the odor of the former. Orco-/- larvae presented with the same choice did not show any preference, verifying that ORs indeed mediate this preference. Which volatiles do the larvae rely on? Larval GC-SSR measurements were performed, stimulating with HCA+ and HCA- Brettanomyces headspace collections. Compared with HCA-, stimulation with HCA+ samples yielded additional responses toward 4-ethylguaiacol and phenethyl alcohol, the latter again most likely due to the increased amounts in the HCA+ samples. Larvae also displayed increased spike firing rate in response to stimulation with the other primary ligands for Or71a, and, similarly to the situation in the adults, 4-ethylguaiacol elicited the strongest response. In GC-SSR dose-response trials, larvae were, however, less sensitive to 4-ethylguaiacol than were adults, with discernable responses to 4-ethylguaiacol requiring a 3-fold larger dose in larvae than in adults. How do larvae react behaviorally to 4-ethylguaiacol? Application of 4-ethylguaiacol in the larval olfactory choice assay resulted in positive chemotaxis. Moreover, larvae given a choice between HCA+ Brettanomyces and HCA- Brettanomyces spiked with 4-ethylguaiacol showed no preference either way, suggesting that the presence of 4-ethylguaiacol in the HCA+ samples indeed confers the attraction. It is thus concluded that the larvae perform the same proxy detection of HCAs as adults, relying on the presence of ethylphenols to identify antioxidant-enriched patches (Dweck, 2015).

Attempts were made to determine which OR(s) in the larva detect the ethylphenols. In a recent study, 19 out of the 21 expressed larval ORs were deorphaned using a panel of ~500 chemicals. Although the ethylphenols were not included in the test panel, chemicals of structural proximity were. To identify candidate OR(s) detecting the ethylphenols, a chemometric approach was undertaken. The ethylphenols were plotted in a 32-dimensional odorant space together with the primary larval OR ligands. A principal component analysis (PCA) plot revealed that the primary Or71a ligands clustered closest with the aromatic ligand for Or94a and Or94b, namely guaiacol acetate (or 2-methoxyphenyl acetate). Thermogenetic silencing of the OSNs expressing Or94a and Or94b by expression of Shibire^ts from the promoter of the latter (the two ORs are co-expressed in the same OSN) indeed abolished preference in a binary-choice test between HCA+ and HCA-Brettanomyces. Furthermore, optogenetic activation of the Or94a/Or94b pathway induced attraction in larvae expressing Channelrhodopsin-2 (ChR-2) from the Or94b promoter, with larvae preferring the side illuminated with blue light (470 nm, activating the ChR-2 molecules, in contrast to parental lines and wild-type (WT) larvae, which are all repelled by blue light. Similarly, larvae confronted with a choice of HCA+ and HCA- Brettanomyces--the latter illuminated with blue light--showed no preference either way (Dweck, 2015).

To verify that Or94a/Or94b respond to the ethylphenols, the 'empty-neuron' system was used to determine the response properties of these two receptors. Heterologous expression of Or94a and Or94b, respectively, in ab3A OSNs conferred responsiveness toward the ethylphenols. Out of the nine primary ligands of Or71a, Or94b responded most strongly to 4-ethylguaiacol. This compound, however, only elicited minor responses from Or94a, which instead was strongly activated by guaiacol. Moreover, GC dose-response trials showed that these ligands induced responses already at very low concentrations from the respective ORs. Both Or94a and Or94b were also activated by stimulation with the Brettanomyces headspace in GC-SSR experiments. It is noted with interest that guaiacol--similar to 4-ethylguaiacol--activates a different receptor than in the adults, although with similar tuning. Guaiacol is a common microbial volatile (produced, e.g., by all the yeasts examined in this study), and its presence in nature would reliably indicate the occurrence of microbes, to larvae as well as adults (Dweck, 2015).

Given that Or94a and Or94b are co-expressed in the same neurons, how do larvae distinguish HCA- from HCA+Brettanomyces when the headspace activates the same neural pathway? A possible explanation could be that the dual activation of Or94a and Or94b by the HCA+Brettanomyces sample would lead to a stronger signal into the central nervous system, in turn causing the behavioral preference. To test this notion, the larvae were challenged with a mixture of 4-ethylguaiacol and guaiacol (10^-4 dilution, total volume 10 μl) against guaiacol (10^-4 dilution, 10 μl volume), a situation chemically mimicking the HCA-/HCA+Brettanomyces choice. Indeed, larvae displayed a significant preference for the mixture over the single component Preference for the mixture remained even when double amounts (i.e., 20 μl) of guaiacol were tested against 10 μl of the mixture, a treatment that would presumably compensate for any effects stemming from an increased volatility of the mix. Next, an available Or94b null mutant (no expression of Or94b was detected in RT-PCR experiments with larval cDNA was tested. As expected, Or94b^-/- larvae showed no response to stimulation with 4-ethylguaiacol in SSR experiments nor did these larvae show any reaction to 4-ethylguaiacol in behavioral tests, whereas the response to guaiacol was no different from that of WT larvae. Furthermore, Or94b^-/- larvae confronted with a choice between HCA+ and HCA-Brettanomyces displayed no preference either way. In summary, larvae, like adults, identify the presence of HCAs via ethylphenols. Curiously, detection is done via a separate receptor from adults, albeit with similar tuning, which moreover is co-expressed with a receptor detecting a general yeast signal. The larval Or94a/Or94b OSNs thus offers coincidence detection of two distinct, but ecologically related, volatiles (Dweck, 2015).

This study has shown that flies are able to recognize substrates enriched with HCAs. Flies--adults as well as larvae--do so by relying on specific volatile ethylphenols (4-ethylphenol and 4-ethylguaiacol), which are exclusively derived from HCAs. In adult flies, the ethylphenols are detected by maxillary palp OSNs that express Or71a. This neuron population is both necessary and sufficient for the proxy detection of HCAs. It was demonstrated that the ethylphenols are generated by a wide range of yeasts consumed by flies and thus act as a consistent and reliable signal for the presence of HCAs. It was further shown that larvae perform the same proxy detection of HCAs via the ethylphenols as the adults, but do so via a different OR (Or94b) only expressed in the larval stage (Dweck, 2015).

In humans, oxidative stress has been implicated in triggering or enhancing a range of diseases typically associated with aging, inter alia cancer and neurodegenerative disorders. For a short-lived species like the fly, the need to prevent the onset of aging related diseases would appear to be an unlikely reason for having a dedicated proxy detection system for dietary antioxidants. For flies, antioxidants could play an important role in counteracting acute oxidative stress induced by immune defense responses and detoxification processes upon consumption or infection by entomopathogenic microorganisms, which co-occur with beneficial food yeasts in the flies' habitat. The importance played by toxic microbes in the fly's ecology is also illustrated by the remarkably sensitive and selective detection system for geosmin, a volatile indicating the presence of harmful microorganisms. The ethylphenol pathway described here thus adds another layer to the fly's defensive arsenal against toxic microbes (Dweck, 2015).

This study proposes that the ecological significance of the pb1B circuit is to alert flies to the presence of dietary antioxidants. Proxy detection of non-volatile nutrients and health-promoting compounds is most likely an important function of the olfactory system. Many volatiles that humans perceive as having a positive impact on flavor are in fact derived from essential nutrients or from other compounds having direct health benefits. These volatiles are accordingly attractive to humans precisely because they reliably signal the presence of their health-promoting precursors. For a generalist species such as the fly, having dedicated OSNs tuned to volatiles indicating the presence of essential nutrients would make sense. Further research will surely reveal more instances of proxy detection of nutrients in the fly's olfactory system, as well as in other organisms (Dweck, 2015).

The pb1B pathway joins a growing number of non-pheromonal OSN classes for which dedicated and non-redundant functions has been assigned. Functionally segregated pathways identified so far include the above-mentioned geosmin circuit fed by Or56a, CO2 avoidance mediated via Gr21a and Gr63a, aversion toward select acids via Ir64a, oviposition preference for citrus-like fruits via Or19a, attraction toward farnesol (exact ecological function unclear) via Or83c, attraction toward vinegar via Or42b and Or92a, preference for the yeast metabolites phenylacetic acid and phenylacetaldehyde via Ir84a, and attraction to ammonia and select amines through Ir92a. It is thought that precise and non-redundant functions, linked to ecologically relevant behaviors can be assigned to most, if not all, of the flies' (known) 48 classes of OSNs. Thus, in contrast to the widespread notion that individual odorants are predominantly decoded via combinatorial patterns of glomerular activation, the fly's olfactory system appears to mainly extract information from its chemical surrounding via dedicated olfactory pathways. Although functionally segregated, the respective pathways would still function in concert, with behavioral decisions arising based on the relative input-or lack thereof-into combinations of dedicated circuits, each carrying a distinct ecological message (Dweck, 2015).