The two-step chemosensory system underlying the oligophagy of silkworm larvae

Oligophagous insect herbivores specifically identify host-plant leaves using their keen sense of taste1. Plant secondary metabolites and sugars are key chemical cues for insects to identify host plants and evaluate their nutritional value, respectively2. However, it is poorly understood how the insect chemosensory system integrates the information from various gustatory inputs. Here we report that a two-step chemosensory system is responsible for host acceptance by larvae of the silkworm Bombyx mori, a specialist for several mulberry species. The first step controlled by a gustatory organ, the maxillary palp (MP), is host-plant recognition during palpation at the leaf edge. Surprisingly, MP detects chlorogenic acid, quercetin glycosides, and β-sitosterol, which stimulate feeding by the silkworm3–6, with ultra-sensitivity (thresholds of aM to fM). Detecting a mixture of these compounds triggers test biting. The second step is evaluation of the sugar content in the leaf sap exuded by test biting. Low-sensitivity chemosensory neurons in another gustatory organ, the maxillary galea (MG), mainly detect sucrose in the leaf sap exuded by test biting, allowing larvae to accept the leaf and proceed to persistent biting. Our present work shows the behavioral and neuronal basis of host acceptance in the silkworm, mainly driven by six phytochemicals. It also reveals that the ultra-sensitive gustation via MP strictly limits initiation of feeding in the silkworm unless it detects a certain combination of host compounds, suggesting an essential role of MP in host-plant selection. The two-step chemosensory system reported herein may commonly underlie stereotyped feeding behavior in phytophagous insects and determine their feeding habits.

certain combination of host compounds, suggesting an essential role of MP in hostplant selection. The two-step chemosensory system reported herein may commonly underlie stereotyped feeding behavior in phytophagous insects and determine their feeding habits.

Main
Host-plant selection by phytophagous insects is dependent on their acceptance or rejection of plants. To clarify the mechanism of host-plant acceptance by silkworm larvae, we observed larval feeding behavior towards a host leaf from white mulberry Morus alba. When a silkworm encounters a leaf, it first palpates the leaf edge using a peripheral gustatory organ known as the maxilla, intermittently bites the edge several times, and finally engages in continuous biting (2-3 times per second) with its head shaking in the dorso-ventral direction along the leaf edge (Fig. 1a, Supplementary Video 1). The intermittent biting with palpation and the continuous biting with head-moving are termed test biting and persistent biting, respectively 7,8 . We hypothesized that sensing of chemical cues from a M. alba leaf via the maxilla induces test biting because test biting always occurs after palpation with the maxilla. The maxilla consists of the maxillary palp (MP) and maxillary galea (MG) (Fig. 1b). To assess the roles of MP and MG in induction of test biting, we used MP-or MG-ablated larvae. MP-ablated larvae showed palpation, but no test or persistent biting (Fig. 1c, d, Supplementary Video 2). MG-ablated larvae showed palpation and test biting, and stopped biting within 1 minute, and did not progress to persistent biting (Fig. 1c, d, Supplementary Videos 3). When we ablated an olfactory organ antenna (AN), the larvae showed palpation, test biting, and persistent biting similar to intact larvae (Fig. 1c, d, Supplementary Video 4). These results indicate that MP and MG are essential for induction of test and persistent biting, respectively.
To assess whether such MP-and MG-controlled biting contributes to the oligophagy of the silkworm, we observed feeding behavior towards leaves of various plant species. The silkworm is a specialist for some Morus species, including M. alba. In addition, the leaves of Cichorioideae plants of the Asteraceae (e.g., dandelion [Taraxacum officinale] and Indian lettuce [Lactuca indica]) are consumed a relatively small amount by silkworm larvae 9,10 . Of the larvae, 88.9 ± 5.3% showed test biting of M. alba within 1 minute after reaching the leaf edge, compared to 70.0 ± 10.0 and 63.3 ± 13.3% for Sonchus oleraceus and T. officinale, respectively (Fig. 2a). Of the larvae, 80 ± 11.5, 53.3 ± 14.5, and 30 ± 10% proceeded to persistent biting of M. alba, S. oleraceus, and T. officinale, respectively. In contrast, the larvae had a lower probability of test biting (3-33%) of 12 inedible leaves, which were finally rejected by most of them ( Fig. 2a, b). Thus, the probability of test biting was higher for edible than for inedible leaves and persistent biting was induced only by edible leaves. These results suggest that the host plant is recognized prior to biting.
Insects sense leaf-surface compounds during palpation 1 . To elucidate whether leafsurface compounds induce test biting, we wiped M. alba leaves with water or methanol, which markedly decreased the proportion of larvae showing test biting (Fig. 2c). Conversely, 76.7 ± 8.8%, 37.5 ± 10.3%, and 36.0 ± 10.8% of the larvae showed test biting of filter paper treated with methanol and leafsurface extracts of M. alba, S. oleraceus, and T. officinale, respectively (Fig. 2d, Extended Data Fig. 1a-c). MP ablation diminished test biting towards filter paper treated with an extract of M. alba leaf-surface (Fig. 2d). Furthermore, tip recording of the sensilla in the MP 12 revealed that MP neurons responded to the leaf-surface extract (Fig. 2e). These findings indicate that compounds in edible leaf-surface extracts stimulate the MP and trigger test biting.
To identify inducers of test biting, we searched for secondary metabolites in edible leaves of M. alba, Lactuca indica, and T. officinale in the plant-metabolite database KNApSAcK 11 because secondary metabolites are thought to be key chemical cues for host-plant recognition 2 . The search yielded chlorogenic acid (CGA) and quercetin-3-Orhamnoside (Q3R). CGA reportedly increases the amount of food intake by the silkworm 6 : Q3R is an analog of isoquercitrin (ISQ), which reportedly induces biting by the silkworm 5 . In addition, we focused on β-sitosterol (βsito) because it also reportedly induces biting by the silkworm 4 . We first recorded the responses of MP and MG towards these compounds. Surprisingly, MP responded to the four compounds at the attomolar and femtomolar levels (Fig. 3a). Based on the shape, amplitude, and frequency of the spikes, CGA, ISQ, Q3R, and βsito were estimated to stimulate at least one, three, two, and one neuron(s), respectively. It is plausible that this ultra-sensitivity of the MP enables detection of trace amounts of CGA, Q3R, ISQ, and βsito at the leaf-surface. In contrast, the MG did not respond to these compounds (Extended Data Fig. 2b, 2c). Next, we assessed whether the four compounds induce test biting. Filter paper, which were treated with each single compound, mixtures of two compounds, and a mixture of ISQ, Q3R, and βsito induced test biting by 20-40% of larvae. In contrast, mixtures of three compounds (CGA+ISQ+βsito, and CGA+Q3R+βsito) and the mixture of all four compounds resulted in a high probability of test biting comparable to the M. alba leaf-surface extract (Fig. 2d and  3b), but did not induce persistent biting (Extended Data Fig. 1d). Filter papers treated with extremely dilute mixtures of CGA, ISQ, and βsito still induced test biting to some extent (Fig. 3c). Meanwhile, a mixture of fructose, sucrose, glucose, and myo-inositol did not induce biting (Fig. 3b). These results suggest that a trace amount of set of CGA+ISQ/Q3R+βsito contribute to host recognition and induction of test biting.
Next, we investigated the role of MG in inducing persistent biting. The lateral sensillum (LS) in the MG is involved in recognition of feeding stimulants. The LS has three neurons specifically tuned to glucose, sucrose, and myo-inositol at around the millimolar level 14 . Therefore, we hypothesized that sugars in the leaf sap exuded by test biting stimulate the MG and induce persistent biting. As feeding initiation is strictly regulated by test biting, we conducted an agar-based food intake assay using starved larvae 16 to simply evaluate persistent biting. In this assay, starved larvae no longer exert strict oligophagy and sometimes randomly bite; this biting substitutes for test biting, and consequently persistent biting occurred in the absence of the inducers of test biting. Alternatively, unlike when feeding on leaves, the MG directly detected high concentrations of compounds at the agar surface during palpation, resulting in induction of persistent biting. Larvae fed agar diet containing sucrose at > 10 mM with showing persistent biting (Fig.  4a). The amount of food intake seemed to correlate with the length of persistent biting.
Indeed, a sucrose dose-dependent increase in the body weight was observed in intact and MP-ablated larvae (Extended Data Fig. 3b, c). The magnitude of the sucrose-induced increase in larval weight was significantly smaller in MG-ablated larvae than in intact larvae (Fig. 4a), suggesting an important role for MG in modulating the amount of food intake. Meanwhile, myo-inositol and glucose themselves did not induce larval weight increase ( Fig. 4b and Extended Data Fig. 3d), whereas myo-inositol showed a supplemental effect in the presence of sucrose (Fig. 4b) in consistent with a previous study 16 . These results suggest that sucrose and myo-inositol contribute to induction of persistent biting by stimulating MG.
Finally, we assessed whether that the inducers of test biting (CGA, ISQ, and βsito) and sugars (sucrose, myo-inositol, and glucose) resulted in an increase in larval weight similar to that induced by M. alba leaf extract and intact leaf. A mixture of 10 mM sucrose, 5 mM myo-inositol and 5 mM glucose, similar to the concentrations in M. alba leaves 17 , resulted in an increase in larval weight in the presence, but not the absence, of a mixture of sugars, similar to a M. alba leaf and a leaf-extract-containing agar-based diet (Fig. 4c). The test biting induced by CGA+ISQ+βsito accelerated the first bite (Extended Data Fig. 4), which might cause persistent biting and consequently increased the total food intake. Therefore, we concluded that these six compounds are major phytochemical drivers of silkworm feeding (Fig.  4d).
Since the 1970s, entomologists have noticed that insects identify their host plants at the leaf-surface, but the mechanism remains unknown. We report here the ultra-sensitive gustation by MP underlies host-plant recognition at leaf-surface in the silkworm. CGA, ISQ/Q3R, and βsito are not specific to edible leaves and other phytochemicals may be also involved in host-plant recognition. Nevertheless, it is plausible that the presence of all the three compounds is a marker of mulberry leaves at least in the ecosystem in which Bombyx madarina, an ancestor of the domesticated silkworm, resides. Additionally, rejection of inedible leaves upon detection of feeding deterrents 13,19 seems to facilitate strict host plant selection. Host-plant recognition by MP may be conserved among phytophagous insects. For example, insect oligophagy and polyphagy are likely consequences of strict and loose restriction, respectively, of feeding initiation by MP. Tuning of neurons in MP for unique combinations of phytochemicals must reflect insect-plant relationships.