Insects are the most diverse group of animals on earth with over a million documented species, many being essential for ecosystem functionality (Grimaldi and Engel, 2005). Their ecological significance spans from nutrient cycling and plant pollination to serving as food sources for many taxa (Scudder, 2017). Moreover, certain insect species have profound economic impacts given their pest status and the interactions that ensue when completing their life cycles in commercially valuable plants/fruits (Lemelin et al., 2016). One such interaction is the resistance/tolerance to the toxic compounds found in certain fruits, which partially protect plants from herbivores (Weng et al., 2012). Understanding the resistance mechanisms to these toxic compounds can provide insights into insects' evolutionary adaptations and survival strategies, especially when considering closely related species (Aluja and Mangan, 2008; Ali and Agrawal, 2012; Giron et al., 2018).

A particularly important group of insects is represented by the true fruit flies (Diptera: Tephritidae) which includes more than 5,000 species (Norrbom, 2010; Liquido et al., 2019). As part of their life history, fruit fly females lay their eggs inside their host’s pulp or seeds, where they find favorable conditions for the growth and development of their larvae (Aluja and Mangan, 2008). This behavior causes certain fruit fly species to affect a wide range of commercially valuable fruit crops, limiting international trade of agricultural products (Aluja and Mangan, 2008). In fact, some of these fruit flies are listed among the fruit-tree pest insect species with the greatest economic impact worldwide (White and Elson-Harris, 1992). Among these species, those belonging to the Anastrepha genus are one of the most relevant in Mexico and the Neotropics (Hernández-Ortiz et al., 2010). Such is the case of Anastrepha ludens (Loew) (Diptera: Tephritidae), commonly known as the Mexican Fruit Fly.

Anastrepha ludens is classified as a highly polyphagous pest, due to its feeding habits which confer it the ability to attack many plant species of several families, including many economically important fruit species such as mango (Mangifera indica L.; Anacardiaceae), citrus (Citrus aurantium L. or Citrus x sinensis L.; Rutaceae), peach (Prunus persica L.; Rosaceae), and pepper (Capsicum pubescens Ruiz y Pav.; Solanaceae) (Birke et al., 2015; Birke and Aluja, 2018); or wild fruit species, including Casimiroa edulis La Llave and Lex (also known as white sapote or matasano) and C. greggii (S. Watson) F. Chiang (commonly known as yellow chapote), both within the Rutaceae family. These last two species are the purported ancestral hosts of A. ludens (Birke et al., 2015).

In contrast to highly polyphagous tephritid species (i.e., A. ludens, the Oriental fruit fly - Bactrocera dorsalis Hendel, or the Mediterranean fruit fly - Ceratitis capitata Wiedemann), there are other fruit fly species with strict monophagous feeding habits. A good example of a monophagous fruit fly is represented by B. oleae Gmelin, commonly known as the Olive Fly. This fly specializes in infesting the fruit of Olea europaeae L (olive tree), which contains high levels of phenolic compounds (up to 14% dry weight-based), many of them being active against insects (Ben-Yosef et al., 2015). Among these compounds, Oleuropein, a glucid that is lethal to Bactrocera oleae larvae, stands out. Despite this, the larvae of B. oleae can develop optimally in the fruit of the olive tree, because it hosts the bacterium Candidatus Erwinia dacicola (Enterobacteriaceae), a symbiont found in its digestive tract that assists the larvae to catabolize oleuropein (Ben-Yosef et al., 2015). A similar association between a key bacterium degrading toxic polyphenols (i.e., Komagataeibacter) and a fruit fly (the stenophagous A. striata) was recently reported (Ochoa-Sánchez et al., 2022). Here we chose as study model another fruit fly species adapted to monophagy: Anastrepha acris Stone (Diptera: Tephritidae), which is phylogenetically very close to A. ludens (Mengual et al., 2017), but exclusively infests fruits of Hippomane mancinella L (Malpighiales: Euphorbiaceae) (Aluja and Norrbom, 2000; Aluja et al., 2020). Hippomane mancinella is a tree that produces a fruit commonly known as “Apple of Death", which is highly toxic to many animals, including humans and insects (Rao, 1974; Adolf and Hecker, 1984).

Recently, Aluja et al. (2020) analyzed the interaction between H. mancinella, A. acris, and A. ludens, and found that diets enriched with different concentrations of dried fruit pulp of H. mancinella have a differential effect on the development of the larvae of both species. Not surprisingly, in A. ludens they found that H. mancinella produces a more pronounced, negative impact on larvae development as there is no natural association between the fly and this toxic fruit in nature. For example, only 0.08% of A. ludens larvae developed into pupae in a diet supplemented with 12.5% H. mancinella compared with A. acris larvae, which fully metamorphosed into pupae and viable adults. However, the number and weight of the exposed pupae decreased as concentration increased. This study nicely complemented the classical literature on the topic of specialized insects dealing with toxic plants (e.g., Berenbaum et al., 1989; Adler et al., 1995; Zalucki et al., 2001; Agrawal and Kurashige, 2003; Harvey et al., 2007). It becomes clear that monophagous species (i.e., highly specialized ones) are not totally immune to plant defense, but they have developed physiological adaptations which allow individuals to cope with deleterious allelochemicals contained in their hosts, exhibiting greater tolerance to these chemicals than polyphagous species. The active compounds reported in H. mancinella include the ellagitannins Hippomanin A and B, phenylpropanoids, coumarins, and flavonoids (e.g., naringenin, kaempferol, hesperidin, and quercetin) (Rao, 1974; Aluja et al., 2020). Hippomanin A (a hydrolyzable tannin) is the main active, toxic compound of H. mancinella fruit. When it undergoes acid hydrolysis, it yields glucose, ellagic acid, and gallic acid (Rao, 1974; Rao, 1977). However, Hippomanin A was not detected in any of the developmental stages of A. acris and neither in its parasitoid; Doryctobracon areolatus (Szépligeti) (Hymenoptera: Braconidae), which suggests that A. acris developed detoxification mechanisms to metabolize it (Aluja et al., 2020).

Despite their economic importance, there is still scant work aimed at understanding the mechanisms used by fruit flies to defend themselves from the toxic specialized metabolites of their plant hosts. In this sense, gene expression studies can be helpful in our quest to elucidate the metabolic impact of plant compounds on herbivorous insects and their effects on their feeding behavior and development (Thitz et al., 2020). For example, Wang et al. (2020) and Zhao et al. (2022) tested the reaction of Spodoptera litura (Fabricius) (Lepidoptera: Noctuidae) fed on a diet enriched with tannins. They found that the genes encoding for proteins involved in xenobiotic metabolism (e.g., cytochrome, and glutathione-related proteins) and digestive enzymes (e.g., lipases and carbohydrases), were highly regulated in the midgut upon diet intake. In Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) moth, Zheng et al. (2022) evaluated the transcriptomic changes in response to gossypol and tannic acid and identified several transcript-encoding-enzymes involved in immunity, digestion, and detoxification metabolism, including glutathione S-transferases (GSTs), UDP-glucosyl-transferases (UGTs), hydrolases, serine-proteases, lipases, Hsp20, or aldehyde dehydrogenases, explaining how tannins are degraded within the digestive tract of the insect (Zheng et al., 2022). Celorio-Mancera et al. (2011) evaluated the defensive mechanisms of the generalist insect H. armigera against the toxic compounds produced by plants within the genus Gossypium L (Malvales: Malvaceae). These authors determined that some phytoregulators (i.e., jasmonic acid, salicylic acid, and ethylene) and phenolic compounds modified the transcriptional responses of the larvae at the primary metabolism level and activated its stress response (Celorio-Mancera et al., 2011).

Based on all the above, but particularly following the emphatic call by Ali and Agrawal (2012) recommending the study of related insects with contrasting feeding habits to foster a deeper understanding of the many mechanisms behind herbivory, here we applied a comparative transcriptomic approach to identify the response of A. acris and A. ludens larvae when fed a lyophilized extract of H. mancinella. The latter, to gain insight into the resistance and detoxification mechanisms used by these sister species to overcome the effects of toxic phytochemicals. Based on the “phytochemical coevolution theory/hypothesis” (Cornell and Hawkins, 2003), we predicted that albeit their close phylogenetic relationship, A. acris, the strict specialist, would have evolved resistance/degrading mechanisms that the congener, the broad generalist, either lost or never developed during evolution.

The study of the molecular responses of herbivorous insects to plant chemical defenses can be useful in our quest to better understand the mechanisms/metabolic routes that are activated to cope with toxic allelochemicals. Using a transcriptomic approach, here we sought to identify the regulated genes in the larvae of two closely phylogenetically related fruit flies with contrasting feeding habits (i.e., monophagous vs. polyphagous), A. acris and A. ludens, when they are exposed to the pulp of H. mancinella, a plant that produces fruit that is highly toxic. In both species, an alteration was observed in the levels of transcript expression coding for enzymes related to cuticle biosynthesis, hypoxia, protease activity (trypsin and serine proteases) and metabolism of xenobiotics (i.e., responses to abiotic stress) (Table 2). However, a higher number of these transcripts were only recorded in A. ludens compared to A. acris, the species that thrives in H. mancinella (Table 2). In addition to these transcripts found, we were able to map the phosphoinositol biosynthesis pathway (PI3K) in both species (Figure 4). On the other hand, activation in routes related to fermentative metabolism (energy production) and repression of pathways involved in the process of ubiquitination and biosynthesis and degradation of amino acids was only identified in A. ludens (Figure 4). The only pathway exclusively affected in A. acris was the degradation of anandamides (Figure 4). As will be discussed in what follows, it is likely that the lower regulation observed in A. acris, the strict specialist, is likely due to the fact that: 1) larvae of this species have the resistance mechanisms to effectively counteract the negative effects of the toxic allelochemicals present in H. mancinella fruit; 2) the low concentration of H. mancinella lyophilizate mixed in the diet in which the larvae were reared did not stress the larvae of A. acris enough as in nature they are exposed to higher concentrations (an aspect we are currently investigating); 3) these larvae might possibly have a symbiotic intestinal microbiota that facilitates the metabolization of the toxic compounds present in the pulp of H. mancinella (something we will research in a future study). Contrary to our prediction, it appears that A. ludens, the sister species that evolved into a broad generalist, has maintained most of the broad resistance mechanisms that it shares with A. acris which were expressed when exposed to the toxic diet it does not encounter in nature. But A. ludens appears to not have developed the specialized biochemical pathways or structural defensive mechanisms A. acris possesses that enable it to thrive in the highly toxic H. mancinella. A comparative genomics study would be necessary to shed definitive light on this.

The fruit of H. mancinella is rich in bioactive polyphenols, such as the ellagitannin Hippomannin A, considered the most toxic compound present in this fruit (Schaeffer et al., 1954; Lauter and Foote, 1955; Rao, 1974; Rao, 1977; Barbehenn et al., 2006; Aluja et al., 2020). The first defense line of insects to face these secondary metabolites is the cuticle. The larval cuticle is a physical barrier that can limit the passage of toxic compounds from the environment to the hemolymph (Pryor et al., 1947; Barrett, 1983; Ashida and Brey, 1995; 1997). We found upregulation of transcripts that encode for cuticle proteins (chitin-binding proteins) in both species (Table 2) and could be related to a response leading to the reinforcement of the cuticle of the larvae. In insects, an increase in the thickness of the cuticle has been related to resistance against secondary metabolites present in plants and insecticides (Balabanidou et al., 2018). For example, when larvae of Spodoptera frugiperda were exposed to camptothecin (a pentacyclic quinoline alkaloid isolated from Camptotheca acuminata Decne.), histopathological changes were observed in the midgut structure, followed by upregulation of several genes encoding for cuticular proteins (Shu et al., 2021). Consistent with this, we found that of the upregulated cuticle-proteins, 28 were not orthologous in A. ludens, and four were not orthologous in A. acris. We note, however, that many of these Unigenes were partial, and this could have biased the resolution of the orthology detection procedure followed. If the above is true, then the damage generated by the pulp of lyophilized H. mancinella could first occur through cuticular disruption. Considering this, A. acris can likely reinforce the resistance against H. mancinella by adapting its cuticle through structural proteins in the cuticle that are more efficient than those found in A. ludens, a species whose larvae are not naturally exposed to the toxicants in the fruit of this plant.

We also identified many transcripts encoding proteases, including trypsin (Table 2). Trypsin is a non-specific serine endopeptidase responsible for the maturation of various zymogens such as prophenoloxidase (important for immune responses) (Ashida and Brey, 1997; Perdomo-Morales et al., 2007; Soares et al., 2011; Lu et al., 2014; Wu et al., 2015). Prophenoloxidase is synthesized in insect hemocytes, actively transported into the cuticle through the epidermis, and randomly distributed in the endocuticle of the body wall (Ashida and Brey, 1997). Since phenoloxidase (mature version of phenoloxidase) is central to the immune response, as it can oxidize phenolic compounds to induce melanization of the cuticle, it is likely that the phenoloxidase system actively participates as a defense mechanism in response to H. mancinella compounds (Lu et al., 2014; Wu et al., 2015).

In insects there is an antagonistic interaction between the ingestion of phenols and the assimilation of nutrients, which can cause an inhibition of the digestion of carbohydrates or proteins, affecting their nutrition, growth, and development (Schopf, 1986; Felton et al., 1992; Haukioja et al., 2002; Pascacio-Villafán et al., 2016; Punia et al., 2020; Thitz et al., 2020). This antagonistic interaction is produced due to the inhibition of digestive enzymes by polyphenols. In the case of ellagitannins, it has been documented that they can inhibit alfa-amylase and trypsin activity (McDougall and Stewart, 2005). Many phytophagous insects counteract the polyphenols' effect by a high protease synthesis rate (Brioschi et al., 2007; Spit et al., 2014; Zhu-Salzman and Zeng, 2015), a mechanism we identified in both fly species in this study. We identified the upregulation of trypsin and serine proteases, 21 in A. ludens and four in A. acris. However, our results suggest differences in these enzymes in both flies, with two peptidases in A. acris not being orthologous with A. ludens. On the other hand, 13 peptidases present in A. ludens were not orthologous in A. acris.

Interestingly, one non-orthologous serine-type endopeptidase in A. acris is a rhomboid-like1 protein. Rhomboid serine proteases, together with ADAM (a disintegrin and metalloproteinase) proteins, are involved in the Epidermal Growth Factor Receptor (EGFR) pathway, catalyzing the cleavage of membrane-tethered epidermal growth factor (EGF)-like ligands, being a regulator and activator of EGFR (Lee et al., 2001; Urban, 2002; Blobel, 2005). In A. acris, a homologous of the ADAM metalloprotease was upregulated (Unigen: acris005570). In Drosophila, EGFR regulates the penetration resistance to the insecticide avermectin, regulating the chitin synthesis and promoting the increment of peritrophic membrane thickness in larvae resistant to avermectin (Chen et al., 2016). Interestingly in human cells, EGFR activation can be inhibited by epigallocatechin-gallate, triggering the internalization of the EGFR (Adachi et al., 2007; Adachi et al., 2008). Even though the EGFR pathway seems conserved in all eukaryotes, differences could exist between A. ludens and A. acris, based on the difference in the rhomboid protein, that hypothetically could be related to the maturation of the EGFR-ligand.

When A. ludens larvae are reared in diets or fruit containing high concentrations of phenolic compounds (e.g., Malus domestica Borkh), a reduction in egg production, larval weight, and an increase in larval development and pupation times has been recorded (Aluja et al., 2014; Pascacio-Villafán et al., 2014). The induction of deformities was also observed in pupae, as well as an increase in mortality and the duration of the biological cycle, which lasts up to three times more compared to larvae reared on natural hosts such as grapefruit, with a lower content of phenolic compounds (Aluja et al., 2014; Pascacio-Villafán et al., 2014). At a molecular level, phenolic compounds bind to the amino acid-building proteins in the diet and/or to digestive tract enzymes (Schopf, 1986; Felton and Duffey, 1991). This happens because the aromatic rings of amino acids and polyphenols interact through π-type interactions, destabilizing the tridimensional structure of proteins, which induces protein denaturation and coalescence with other proteins (Adamczyk et al., 2012; Chen et al., 2019). These effects inactivate the function of the proteins affected, and the biological processes in which they participate (Haslam, 2007; Spalinger et al., 2010; Adamczyk et al., 2012). Tannins, a particular class of polyphenols, can produce high levels of semiquinones and quinones, which affect the redox status of the cell contributing to poor absorption of nutrients and generation of cytotoxic effects in the affected tissues (Appel, 1993; Salminen et al., 2004; Barbehenn et al., 2005; Barbehenn and Constabel, 2011).

In addition, and in contrast to what was observed in A. acris, in A. ludens a high number of transcripts that encode for proteins involved in energy processes and in the biosynthesis and degradation of amino acids were found (Figure 4). So, not surprisingly, the larvae of A. ludens suffered more metabolic changes than A. acris. For example, we observed the activation of fermentative pathways, inferring that energy metabolism was one of the most affected aspects in A. ludens. This effect was not observed in A. acris, suggesting that the basal defense mechanisms of A. acris were sufficient to prevent the metabolic effects observed in A. ludens.

The transcriptional responses of both phylogenetically related species involve the activation of transcripts that encode for similar enzymes, such as enzymes with heme-binding sites, proteases, or chitin-binding enzymes (Table 2). It is likely that the high concentration of tannins present in the pulp of H. mancinella, together with the metal chelating activity of the o-dihydroxyphenyl groups of tannins, causes a decrease in the bioavailability of iron, copper, and other metallic macro or micronutrients (McDonald et al., 1996; Afsana et al., 2004; Fiesel et al., 2015; Delimont et al., 2017). So, if tannins were chelating metals, proteins that have an active site coordinated with metal ions, such as those with heme groups, would become dysfunctional. For example, the activation of genes that encode for hemocyanin (a heme-containing protein) could cause a decrease in the oxygen transport activity through the hemolymph (Engel and Brouwer, 1991; Martin and Hose, 1995). Therefore, if the tannins are interfering with the oxygen transport activity, and favoring hypoxic conditions, a plausible countermeasure to compensate hypoxia could be the transcriptional activation of hemocyanins (Engel and Brouwer, 1991; Martin and Hose, 1995; Salminen and Karonen, 2011). Also, the decrease in the concentration of oxygen available in the hemolymph would explain the activation of the fermentative pathways observed in A. ludens. Usually, the activation of fermentative pathways leads to the rapid production of energy, but this process is insufficient when these conditions are maintained for a long time, affecting the metabolic performance of the organism (Hoback and Stanley, 2001).

In both species, the activation of genes coding for proteins related to the metabolism of xenobiotics and in oxidation-reduction processes was also found (Table 2). As was the case with other defensive responses, in A. ludens the number of these transcripts was much higher compared to A. acris. In this sense, the activation of transcripts related to the first stage of xenobiotic metabolism is relevant, since these enzymes represent one of the first lines of defense against toxic compounds (Li et al., 2007). So, the activation of transcripts encoding for P450 monooxygenase enzymes, glutathione-S-transferases, and transporters could be related to the coordinated response of detoxification pathways and toxin excretion (Chahine and O'Donnell, 2012; Zheng et al., 2022). In future studies, it would be of interest to identify the anatomical site where these transcripts are altered.

Repression of signaling via phosphoinositides was also identified in both species. For example, in A. ludens the transcript of phosphatidylinositol 4, 5-bisphosphate kinase (PI3K) was specifically repressed (Figure 4). PI3K is an important regulator of cell growth in insects, as it inhibits the signaling that leads to a reduction in body size, tissue growth, and starvation (Britton et al., 2002; Lin and Smagghe, 2019). This happens because PI3K activity is regulated by the availability of protein in the diet and coordinates cellular metabolism based on the nutritional conditions of the individual (Britton et al., 2002).

Notably, in A. acris, the anandamide degradation pathway was repressed (Figure 4). Anandamide is a bioactive fatty acid amide widely studied in mammals that functions as a messenger molecule regulating feeding behavior (e.g., hunger, and satiety perception), among other functions (Sharkey et al., 2014). However, in insects, there are few studies on the metabolism of anandamide and its functions (Ban et al., 2003; Farrell and Merkler, 2008; Tortoriello et al., 2013; Jeffries et al., 2014).

In other study models such as the plant Arabidopsis thaliana, it has been reported that fungi toxin elimination occurs via a vesicle-mediated transport which is translocated from the plasma membrane to vacuoles (Wang et al., 2020). This mechanism, which is not exclusive to plants, could occur also in insects in which the detoxification strategy includes catabolism and the sequestration of plant-defense secondary metabolites. The above can explain, at least in part, why proteins like SNARE are differentially expressed. SNARE proteins represent a superfamily of small, mostly membrane-anchored proteins that catalyze the fusion of membranes in all eukaryotic cells (Margiotta, 2022).

The larvae of A. acris and A. ludens responded at a transcriptional level to counteract the negative effects caused by the ingestion of H. mancinella pulp, expressing overall similar mechanisms. However, in A. ludens the number of transcripts altered was much greater than in A. acris. These transcriptional responses of both species were mainly related to hypoxia, cuticle biosynthesis, and xenobiotic metabolism. Our results indicate that A. acris can easily handle a low level of toxicity in H. mancinella lyophilizates added to an artificial diet (6%), but additional experiments with higher concentrations of the toxicants in the fruit are needed to fully understand the biochemical/molecular mechanisms larvae of this species use to degrade or block these deleterious allelochemicals. From our results it becomes clear that during the divergence of these phylogenetically close species, A. acris developed adaptations to deal with the toxicants of its only known host, whereas A. ludens, following another evolutionary path, maintained certain common biochemical routes, expressed here when exposed to the toxicants of a fruit that is not a host in nature, and apparently lost others that inhibit its development in H. mancinella. But further research is needed to clarify if A. acris indeed evolved novel toxicant degradation/blocking routes, or if its gut contains symbiotic bacteria that are the ones that metabolize the toxicants. This is needed as in this study we studied the transcriptome of entire larvae, including the gut. In future studies we will concentrate on the gut and the bacteria contained in it. Further research is also needed on possible morphological adaptations in the midgut of the larvae (e.g., thickness of gut walls) through which the entrance of toxicants via the gut into the hemolymph could be blocked.